{
  "best_metric": 1.06758451461792,
  "best_model_checkpoint": "outputs/checkpoint-55214",
  "epoch": 5.0,
  "eval_steps": 2906,
  "global_step": 58115,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.603630732168976e-05,
      "grad_norm": 82.0,
      "learning_rate": 4e-05,
      "loss": 3.0424,
      "step": 1
    },
    {
      "epoch": 0.00017207261464337952,
      "grad_norm": 1.2265625,
      "learning_rate": 8e-05,
      "loss": 1.4657,
      "step": 2
    },
    {
      "epoch": 0.00025810892196506923,
      "grad_norm": 13.125,
      "learning_rate": 0.00012,
      "loss": 3.1349,
      "step": 3
    },
    {
      "epoch": 0.00034414522928675903,
      "grad_norm": 54.75,
      "learning_rate": 0.00016,
      "loss": 2.4939,
      "step": 4
    },
    {
      "epoch": 0.0004301815366084488,
      "grad_norm": 1.890625,
      "learning_rate": 0.0002,
      "loss": 1.5247,
      "step": 5
    },
    {
      "epoch": 0.0005162178439301385,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00024,
      "loss": 2.4668,
      "step": 6
    },
    {
      "epoch": 0.0006022541512518283,
      "grad_norm": 23.25,
      "learning_rate": 0.00028000000000000003,
      "loss": 2.0472,
      "step": 7
    },
    {
      "epoch": 0.0006882904585735181,
      "grad_norm": 2.21875,
      "learning_rate": 0.00032,
      "loss": 2.1526,
      "step": 8
    },
    {
      "epoch": 0.0007743267658952078,
      "grad_norm": 20.375,
      "learning_rate": 0.00035999999999999997,
      "loss": 1.585,
      "step": 9
    },
    {
      "epoch": 0.0008603630732168976,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004,
      "loss": 1.6784,
      "step": 10
    },
    {
      "epoch": 0.0009463993805385873,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00044,
      "loss": 2.183,
      "step": 11
    },
    {
      "epoch": 0.001032435687860277,
      "grad_norm": 12.5,
      "learning_rate": 0.00048,
      "loss": 1.6778,
      "step": 12
    },
    {
      "epoch": 0.0011184719951819668,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005200000000000001,
      "loss": 1.8029,
      "step": 13
    },
    {
      "epoch": 0.0012045083025036566,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005600000000000001,
      "loss": 1.7968,
      "step": 14
    },
    {
      "epoch": 0.0012905446098253463,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006,
      "loss": 1.5311,
      "step": 15
    },
    {
      "epoch": 0.0013765809171470361,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00064,
      "loss": 1.9783,
      "step": 16
    },
    {
      "epoch": 0.0014626172244687258,
      "grad_norm": 6.375,
      "learning_rate": 0.00068,
      "loss": 2.0602,
      "step": 17
    },
    {
      "epoch": 0.0015486535317904156,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007199999999999999,
      "loss": 1.1631,
      "step": 18
    },
    {
      "epoch": 0.0016346898391121053,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00076,
      "loss": 2.408,
      "step": 19
    },
    {
      "epoch": 0.001720726146433795,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008,
      "loss": 1.9395,
      "step": 20
    },
    {
      "epoch": 0.0018067624537554847,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00084,
      "loss": 1.3689,
      "step": 21
    },
    {
      "epoch": 0.0018927987610771746,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00088,
      "loss": 1.3424,
      "step": 22
    },
    {
      "epoch": 0.0019788350683988642,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00092,
      "loss": 1.7665,
      "step": 23
    },
    {
      "epoch": 0.002064871375720554,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00096,
      "loss": 1.5091,
      "step": 24
    },
    {
      "epoch": 0.002150907683042244,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001,
      "loss": 1.6828,
      "step": 25
    },
    {
      "epoch": 0.0022369439903639336,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009999999992687987,
      "loss": 2.1183,
      "step": 26
    },
    {
      "epoch": 0.002322980297685623,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009999999970751944,
      "loss": 1.731,
      "step": 27
    },
    {
      "epoch": 0.0024090166050073133,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009999999934191877,
      "loss": 2.6165,
      "step": 28
    },
    {
      "epoch": 0.002495052912329003,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000999999988300778,
      "loss": 1.4077,
      "step": 29
    },
    {
      "epoch": 0.0025810892196506926,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999999817199656,
      "loss": 1.5875,
      "step": 30
    },
    {
      "epoch": 0.002667125526972382,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0009999999736767507,
      "loss": 1.1149,
      "step": 31
    },
    {
      "epoch": 0.0027531618342940723,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000999999964171133,
      "loss": 1.7714,
      "step": 32
    },
    {
      "epoch": 0.002839198141615762,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999999532031125,
      "loss": 1.5152,
      "step": 33
    },
    {
      "epoch": 0.0029252344489374515,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009999999407726897,
      "loss": 1.5683,
      "step": 34
    },
    {
      "epoch": 0.003011270756259141,
      "grad_norm": 1.078125,
      "learning_rate": 0.000999999926879864,
      "loss": 1.2014,
      "step": 35
    },
    {
      "epoch": 0.0030973070635808312,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0009999999115246358,
      "loss": 1.7832,
      "step": 36
    },
    {
      "epoch": 0.003183343370902521,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0009999998947070052,
      "loss": 1.1225,
      "step": 37
    },
    {
      "epoch": 0.0032693796782242105,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0009999998764269721,
      "loss": 0.7642,
      "step": 38
    },
    {
      "epoch": 0.0033554159855459006,
      "grad_norm": 3.203125,
      "learning_rate": 0.0009999998566845368,
      "loss": 1.3293,
      "step": 39
    },
    {
      "epoch": 0.00344145229286759,
      "grad_norm": 13.1875,
      "learning_rate": 0.000999999835479699,
      "loss": 2.1398,
      "step": 40
    },
    {
      "epoch": 0.00352748860018928,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000999999812812459,
      "loss": 1.4079,
      "step": 41
    },
    {
      "epoch": 0.0036135249075109695,
      "grad_norm": 2.90625,
      "learning_rate": 0.0009999997886828167,
      "loss": 1.5417,
      "step": 42
    },
    {
      "epoch": 0.0036995612148326596,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009999997630907723,
      "loss": 1.4048,
      "step": 43
    },
    {
      "epoch": 0.003785597522154349,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999997360363258,
      "loss": 1.4445,
      "step": 44
    },
    {
      "epoch": 0.003871633829476039,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009999997075194774,
      "loss": 1.1612,
      "step": 45
    },
    {
      "epoch": 0.0039576701367977285,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000999999677540227,
      "loss": 1.9122,
      "step": 46
    },
    {
      "epoch": 0.0040437064441194185,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000999999646098575,
      "loss": 1.8301,
      "step": 47
    },
    {
      "epoch": 0.004129742751441108,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000999999613194521,
      "loss": 1.4711,
      "step": 48
    },
    {
      "epoch": 0.004215779058762798,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0009999995788280655,
      "loss": 1.0847,
      "step": 49
    },
    {
      "epoch": 0.004301815366084488,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009999995429992084,
      "loss": 1.1031,
      "step": 50
    },
    {
      "epoch": 0.004387851673406177,
      "grad_norm": 1.125,
      "learning_rate": 0.00099999950570795,
      "loss": 1.834,
      "step": 51
    },
    {
      "epoch": 0.004473887980727867,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009999994669542902,
      "loss": 1.4281,
      "step": 52
    },
    {
      "epoch": 0.004559924288049557,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999994267382293,
      "loss": 1.41,
      "step": 53
    },
    {
      "epoch": 0.004645960595371246,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999993850597674,
      "loss": 2.1383,
      "step": 54
    },
    {
      "epoch": 0.0047319969026929365,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009999993419189043,
      "loss": 1.26,
      "step": 55
    },
    {
      "epoch": 0.0048180332100146266,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0009999992973156405,
      "loss": 1.1556,
      "step": 56
    },
    {
      "epoch": 0.004904069517336316,
      "grad_norm": 1.0625,
      "learning_rate": 0.000999999251249976,
      "loss": 2.0926,
      "step": 57
    },
    {
      "epoch": 0.004990105824658006,
      "grad_norm": 1.0625,
      "learning_rate": 0.000999999203721911,
      "loss": 2.2266,
      "step": 58
    },
    {
      "epoch": 0.005076142131979695,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009999991547314452,
      "loss": 1.5117,
      "step": 59
    },
    {
      "epoch": 0.005162178439301385,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999991042785796,
      "loss": 1.3287,
      "step": 60
    },
    {
      "epoch": 0.005248214746623075,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009999990523633137,
      "loss": 1.6356,
      "step": 61
    },
    {
      "epoch": 0.005334251053944764,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009999989989856477,
      "loss": 1.127,
      "step": 62
    },
    {
      "epoch": 0.0054202873612664544,
      "grad_norm": 1.109375,
      "learning_rate": 0.000999998944145582,
      "loss": 2.0196,
      "step": 63
    },
    {
      "epoch": 0.0055063236685881445,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009999988878431165,
      "loss": 1.1096,
      "step": 64
    },
    {
      "epoch": 0.005592359975909834,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009999988300782516,
      "loss": 1.8488,
      "step": 65
    },
    {
      "epoch": 0.005678396283231524,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009999987708509874,
      "loss": 2.2341,
      "step": 66
    },
    {
      "epoch": 0.005764432590553214,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000999998710161324,
      "loss": 2.2145,
      "step": 67
    },
    {
      "epoch": 0.005850468897874903,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009999986480092617,
      "loss": 1.3247,
      "step": 68
    },
    {
      "epoch": 0.005936505205196593,
      "grad_norm": 0.828125,
      "learning_rate": 0.0009999985843948006,
      "loss": 1.1213,
      "step": 69
    },
    {
      "epoch": 0.006022541512518282,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009999985193179406,
      "loss": 1.6354,
      "step": 70
    },
    {
      "epoch": 0.006108577819839972,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009999984527786824,
      "loss": 1.6949,
      "step": 71
    },
    {
      "epoch": 0.0061946141271616625,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000999998384777026,
      "loss": 1.8509,
      "step": 72
    },
    {
      "epoch": 0.006280650434483352,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009999983153129715,
      "loss": 1.4287,
      "step": 73
    },
    {
      "epoch": 0.006366686741805042,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999982443865193,
      "loss": 1.399,
      "step": 74
    },
    {
      "epoch": 0.006452723049126732,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009999981719976691,
      "loss": 1.7428,
      "step": 75
    },
    {
      "epoch": 0.006538759356448421,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009999980981464218,
      "loss": 1.0223,
      "step": 76
    },
    {
      "epoch": 0.006624795663770111,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0009999980228327773,
      "loss": 1.4118,
      "step": 77
    },
    {
      "epoch": 0.006710831971091801,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009999979460567359,
      "loss": 1.0072,
      "step": 78
    },
    {
      "epoch": 0.00679686827841349,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009999978678182976,
      "loss": 2.0012,
      "step": 79
    },
    {
      "epoch": 0.00688290458573518,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009999977881174629,
      "loss": 1.625,
      "step": 80
    },
    {
      "epoch": 0.00696894089305687,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009999977069542316,
      "loss": 1.1717,
      "step": 81
    },
    {
      "epoch": 0.00705497720037856,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009999976243286047,
      "loss": 1.1254,
      "step": 82
    },
    {
      "epoch": 0.00714101350770025,
      "grad_norm": 1.5,
      "learning_rate": 0.0009999975402405817,
      "loss": 1.6095,
      "step": 83
    },
    {
      "epoch": 0.007227049815021939,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009999974546901632,
      "loss": 1.6898,
      "step": 84
    },
    {
      "epoch": 0.007313086122343629,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009999973676773494,
      "loss": 1.024,
      "step": 85
    },
    {
      "epoch": 0.007399122429665319,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0009999972792021407,
      "loss": 1.107,
      "step": 86
    },
    {
      "epoch": 0.007485158736987008,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000999997189264537,
      "loss": 1.2092,
      "step": 87
    },
    {
      "epoch": 0.007571195044308698,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009999970978645388,
      "loss": 1.5148,
      "step": 88
    },
    {
      "epoch": 0.0076572313516303885,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0009999970050021463,
      "loss": 1.0621,
      "step": 89
    },
    {
      "epoch": 0.007743267658952078,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009999969106773599,
      "loss": 1.512,
      "step": 90
    },
    {
      "epoch": 0.007829303966273767,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009999968148901796,
      "loss": 1.8613,
      "step": 91
    },
    {
      "epoch": 0.007915340273595457,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000999996717640606,
      "loss": 1.2644,
      "step": 92
    },
    {
      "epoch": 0.008001376580917147,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009999966189286394,
      "loss": 1.7194,
      "step": 93
    },
    {
      "epoch": 0.008087412888238837,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009999965187542797,
      "loss": 1.1842,
      "step": 94
    },
    {
      "epoch": 0.008173449195560527,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009999964171175275,
      "loss": 0.8879,
      "step": 95
    },
    {
      "epoch": 0.008259485502882215,
      "grad_norm": 3.265625,
      "learning_rate": 0.000999996314018383,
      "loss": 1.682,
      "step": 96
    },
    {
      "epoch": 0.008345521810203906,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009999962094568466,
      "loss": 1.8095,
      "step": 97
    },
    {
      "epoch": 0.008431558117525596,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009999961034329184,
      "loss": 1.3529,
      "step": 98
    },
    {
      "epoch": 0.008517594424847286,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000999995995946599,
      "loss": 1.2034,
      "step": 99
    },
    {
      "epoch": 0.008603630732168976,
      "grad_norm": 1.25,
      "learning_rate": 0.0009999958869978884,
      "loss": 1.8542,
      "step": 100
    },
    {
      "epoch": 0.008689667039490666,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009999957765867871,
      "loss": 1.6692,
      "step": 101
    },
    {
      "epoch": 0.008775703346812354,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009999956647132955,
      "loss": 1.311,
      "step": 102
    },
    {
      "epoch": 0.008861739654134044,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009999955513774138,
      "loss": 1.5235,
      "step": 103
    },
    {
      "epoch": 0.008947775961455734,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009999954365791425,
      "loss": 1.3141,
      "step": 104
    },
    {
      "epoch": 0.009033812268777424,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999953203184814,
      "loss": 1.8853,
      "step": 105
    },
    {
      "epoch": 0.009119848576099114,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999952025954316,
      "loss": 1.6229,
      "step": 106
    },
    {
      "epoch": 0.009205884883420803,
      "grad_norm": 1.046875,
      "learning_rate": 0.000999995083409993,
      "loss": 1.7419,
      "step": 107
    },
    {
      "epoch": 0.009291921190742493,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000999994962762166,
      "loss": 1.7717,
      "step": 108
    },
    {
      "epoch": 0.009377957498064183,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999948406519508,
      "loss": 1.4227,
      "step": 109
    },
    {
      "epoch": 0.009463993805385873,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0009999947170793479,
      "loss": 1.1323,
      "step": 110
    },
    {
      "epoch": 0.009550030112707563,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0009999945920443581,
      "loss": 1.4326,
      "step": 111
    },
    {
      "epoch": 0.009636066420029253,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000999994465546981,
      "loss": 1.2262,
      "step": 112
    },
    {
      "epoch": 0.009722102727350941,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009999943375872174,
      "loss": 1.3021,
      "step": 113
    },
    {
      "epoch": 0.009808139034672632,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009999942081650678,
      "loss": 1.8899,
      "step": 114
    },
    {
      "epoch": 0.009894175341994322,
      "grad_norm": 0.875,
      "learning_rate": 0.0009999940772805322,
      "loss": 0.9828,
      "step": 115
    },
    {
      "epoch": 0.009980211649316012,
      "grad_norm": 1.0,
      "learning_rate": 0.0009999939449336112,
      "loss": 1.4986,
      "step": 116
    },
    {
      "epoch": 0.010066247956637702,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009999938111243051,
      "loss": 2.2208,
      "step": 117
    },
    {
      "epoch": 0.01015228426395939,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009999936758526146,
      "loss": 0.8248,
      "step": 118
    },
    {
      "epoch": 0.01023832057128108,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009999935391185395,
      "loss": 1.3149,
      "step": 119
    },
    {
      "epoch": 0.01032435687860277,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009999934009220807,
      "loss": 1.3326,
      "step": 120
    },
    {
      "epoch": 0.01041039318592446,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999932612632385,
      "loss": 1.9149,
      "step": 121
    },
    {
      "epoch": 0.01049642949324615,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009999931201420131,
      "loss": 0.8017,
      "step": 122
    },
    {
      "epoch": 0.01058246580056784,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999929775584052,
      "loss": 1.4837,
      "step": 123
    },
    {
      "epoch": 0.010668502107889529,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000999992833512415,
      "loss": 1.0005,
      "step": 124
    },
    {
      "epoch": 0.010754538415211219,
      "grad_norm": 1.0,
      "learning_rate": 0.0009999926880040432,
      "loss": 0.9563,
      "step": 125
    },
    {
      "epoch": 0.010840574722532909,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00099999254103329,
      "loss": 1.38,
      "step": 126
    },
    {
      "epoch": 0.010926611029854599,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009999923926001554,
      "loss": 1.5715,
      "step": 127
    },
    {
      "epoch": 0.011012647337176289,
      "grad_norm": 1.125,
      "learning_rate": 0.0009999922427046408,
      "loss": 1.7349,
      "step": 128
    },
    {
      "epoch": 0.011098683644497977,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000999992091346746,
      "loss": 1.2795,
      "step": 129
    },
    {
      "epoch": 0.011184719951819667,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009999919385264716,
      "loss": 1.3612,
      "step": 130
    },
    {
      "epoch": 0.011270756259141358,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0009999917842438179,
      "loss": 1.1378,
      "step": 131
    },
    {
      "epoch": 0.011356792566463048,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0009999916284987856,
      "loss": 0.5512,
      "step": 132
    },
    {
      "epoch": 0.011442828873784738,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009999914712913748,
      "loss": 1.9296,
      "step": 133
    },
    {
      "epoch": 0.011528865181106428,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009999913126215866,
      "loss": 1.6033,
      "step": 134
    },
    {
      "epoch": 0.011614901488428116,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009999911524894208,
      "loss": 2.1332,
      "step": 135
    },
    {
      "epoch": 0.011700937795749806,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009999909908948782,
      "loss": 1.6479,
      "step": 136
    },
    {
      "epoch": 0.011786974103071496,
      "grad_norm": 1.25,
      "learning_rate": 0.000999990827837959,
      "loss": 1.3522,
      "step": 137
    },
    {
      "epoch": 0.011873010410393186,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000999990663318664,
      "loss": 1.7049,
      "step": 138
    },
    {
      "epoch": 0.011959046717714876,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009999904973369935,
      "loss": 1.7349,
      "step": 139
    },
    {
      "epoch": 0.012045083025036565,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009999903298929483,
      "loss": 1.9349,
      "step": 140
    },
    {
      "epoch": 0.012131119332358255,
      "grad_norm": 0.71875,
      "learning_rate": 0.0009999901609865282,
      "loss": 0.8542,
      "step": 141
    },
    {
      "epoch": 0.012217155639679945,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009999899906177345,
      "loss": 2.1662,
      "step": 142
    },
    {
      "epoch": 0.012303191947001635,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000999989818786567,
      "loss": 1.0578,
      "step": 143
    },
    {
      "epoch": 0.012389228254323325,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009999896454930268,
      "loss": 0.9955,
      "step": 144
    },
    {
      "epoch": 0.012475264561645015,
      "grad_norm": 0.90625,
      "learning_rate": 0.000999989470737114,
      "loss": 1.1925,
      "step": 145
    },
    {
      "epoch": 0.012561300868966703,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009999892945188294,
      "loss": 1.1293,
      "step": 146
    },
    {
      "epoch": 0.012647337176288393,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009999891168381732,
      "loss": 1.3017,
      "step": 147
    },
    {
      "epoch": 0.012733373483610083,
      "grad_norm": 1.15625,
      "learning_rate": 0.000999988937695146,
      "loss": 1.0397,
      "step": 148
    },
    {
      "epoch": 0.012819409790931774,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009999887570897484,
      "loss": 2.1754,
      "step": 149
    },
    {
      "epoch": 0.012905446098253464,
      "grad_norm": 1.609375,
      "learning_rate": 0.000999988575021981,
      "loss": 1.6687,
      "step": 150
    },
    {
      "epoch": 0.012991482405575152,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009999883914918444,
      "loss": 1.4707,
      "step": 151
    },
    {
      "epoch": 0.013077518712896842,
      "grad_norm": 0.8125,
      "learning_rate": 0.000999988206499339,
      "loss": 1.3881,
      "step": 152
    },
    {
      "epoch": 0.013163555020218532,
      "grad_norm": 1.375,
      "learning_rate": 0.000999988020044465,
      "loss": 1.4797,
      "step": 153
    },
    {
      "epoch": 0.013249591327540222,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009999878321272236,
      "loss": 1.8261,
      "step": 154
    },
    {
      "epoch": 0.013335627634861912,
      "grad_norm": 1.0,
      "learning_rate": 0.0009999876427476148,
      "loss": 1.3016,
      "step": 155
    },
    {
      "epoch": 0.013421663942183602,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0009999874519056396,
      "loss": 1.7018,
      "step": 156
    },
    {
      "epoch": 0.01350770024950529,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009999872596012983,
      "loss": 1.3062,
      "step": 157
    },
    {
      "epoch": 0.01359373655682698,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009999870658345916,
      "loss": 1.1372,
      "step": 158
    },
    {
      "epoch": 0.01367977286414867,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009999868706055198,
      "loss": 1.7547,
      "step": 159
    },
    {
      "epoch": 0.01376580917147036,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009999866739140836,
      "loss": 1.8466,
      "step": 160
    },
    {
      "epoch": 0.013851845478792051,
      "grad_norm": 0.734375,
      "learning_rate": 0.0009999864757602836,
      "loss": 0.7122,
      "step": 161
    },
    {
      "epoch": 0.01393788178611374,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009999862761441206,
      "loss": 2.1028,
      "step": 162
    },
    {
      "epoch": 0.01402391809343543,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009999860750655948,
      "loss": 1.8576,
      "step": 163
    },
    {
      "epoch": 0.01410995440075712,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009999858725247071,
      "loss": 1.4676,
      "step": 164
    },
    {
      "epoch": 0.01419599070807881,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009999856685214581,
      "loss": 1.7736,
      "step": 165
    },
    {
      "epoch": 0.0142820270154005,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999854630558481,
      "loss": 1.7641,
      "step": 166
    },
    {
      "epoch": 0.01436806332272219,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0009999852561278777,
      "loss": 0.8895,
      "step": 167
    },
    {
      "epoch": 0.014454099630043878,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000999985047737548,
      "loss": 2.0469,
      "step": 168
    },
    {
      "epoch": 0.014540135937365568,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0009999848378848591,
      "loss": 0.775,
      "step": 169
    },
    {
      "epoch": 0.014626172244687258,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009999846265698116,
      "loss": 1.322,
      "step": 170
    },
    {
      "epoch": 0.014712208552008948,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0009999844137924066,
      "loss": 1.7798,
      "step": 171
    },
    {
      "epoch": 0.014798244859330638,
      "grad_norm": 1.125,
      "learning_rate": 0.0009999841995526444,
      "loss": 1.3844,
      "step": 172
    },
    {
      "epoch": 0.014884281166652327,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009999839838505256,
      "loss": 1.4287,
      "step": 173
    },
    {
      "epoch": 0.014970317473974017,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009999837666860509,
      "loss": 1.5782,
      "step": 174
    },
    {
      "epoch": 0.015056353781295707,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000999983548059221,
      "loss": 0.6317,
      "step": 175
    },
    {
      "epoch": 0.015142390088617397,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009999833279700362,
      "loss": 0.9988,
      "step": 176
    },
    {
      "epoch": 0.015228426395939087,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009999831064184976,
      "loss": 1.1988,
      "step": 177
    },
    {
      "epoch": 0.015314462703260777,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009999828834046057,
      "loss": 1.0354,
      "step": 178
    },
    {
      "epoch": 0.015400499010582465,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009999826589283608,
      "loss": 1.7036,
      "step": 179
    },
    {
      "epoch": 0.015486535317904155,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009999824329897642,
      "loss": 1.4339,
      "step": 180
    },
    {
      "epoch": 0.015572571625225845,
      "grad_norm": 0.921875,
      "learning_rate": 0.000999982205588816,
      "loss": 1.5206,
      "step": 181
    },
    {
      "epoch": 0.015658607932547534,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999819767255174,
      "loss": 1.9152,
      "step": 182
    },
    {
      "epoch": 0.015744644239869224,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999817463998686,
      "loss": 0.8349,
      "step": 183
    },
    {
      "epoch": 0.015830680547190914,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009999815146118703,
      "loss": 0.9765,
      "step": 184
    },
    {
      "epoch": 0.015916716854512604,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0009999812813615235,
      "loss": 1.2722,
      "step": 185
    },
    {
      "epoch": 0.016002753161834294,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009999810466488285,
      "loss": 2.0215,
      "step": 186
    },
    {
      "epoch": 0.016088789469155984,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009999808104737863,
      "loss": 1.027,
      "step": 187
    },
    {
      "epoch": 0.016174825776477674,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009999805728363976,
      "loss": 1.4086,
      "step": 188
    },
    {
      "epoch": 0.016260862083799364,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009999803337366628,
      "loss": 1.3974,
      "step": 189
    },
    {
      "epoch": 0.016346898391121054,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009999800931745828,
      "loss": 1.0755,
      "step": 190
    },
    {
      "epoch": 0.016432934698442744,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009999798511501584,
      "loss": 1.7511,
      "step": 191
    },
    {
      "epoch": 0.01651897100576443,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009999796076633899,
      "loss": 2.0851,
      "step": 192
    },
    {
      "epoch": 0.01660500731308612,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009999793627142783,
      "loss": 1.6152,
      "step": 193
    },
    {
      "epoch": 0.01669104362040781,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009999791163028246,
      "loss": 1.3434,
      "step": 194
    },
    {
      "epoch": 0.0167770799277295,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009999788684290289,
      "loss": 1.3102,
      "step": 195
    },
    {
      "epoch": 0.01686311623505119,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009999786190928925,
      "loss": 1.2071,
      "step": 196
    },
    {
      "epoch": 0.01694915254237288,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009999783682944157,
      "loss": 0.9442,
      "step": 197
    },
    {
      "epoch": 0.01703518884969457,
      "grad_norm": 0.6875,
      "learning_rate": 0.0009999781160335997,
      "loss": 0.596,
      "step": 198
    },
    {
      "epoch": 0.01712122515701626,
      "grad_norm": 0.875,
      "learning_rate": 0.0009999778623104446,
      "loss": 0.7755,
      "step": 199
    },
    {
      "epoch": 0.01720726146433795,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009999776071249516,
      "loss": 0.8682,
      "step": 200
    },
    {
      "epoch": 0.01729329777165964,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009999773504771215,
      "loss": 1.8846,
      "step": 201
    },
    {
      "epoch": 0.01737933407898133,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009999770923669546,
      "loss": 1.7062,
      "step": 202
    },
    {
      "epoch": 0.017465370386303018,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009999768327944522,
      "loss": 1.3455,
      "step": 203
    },
    {
      "epoch": 0.01755140669362471,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009999765717596148,
      "loss": 1.5104,
      "step": 204
    },
    {
      "epoch": 0.0176374430009464,
      "grad_norm": 1.140625,
      "learning_rate": 0.000999976309262443,
      "loss": 2.0205,
      "step": 205
    },
    {
      "epoch": 0.01772347930826809,
      "grad_norm": 1.0625,
      "learning_rate": 0.000999976045302938,
      "loss": 1.5206,
      "step": 206
    },
    {
      "epoch": 0.01780951561558978,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009999757798811,
      "loss": 1.031,
      "step": 207
    },
    {
      "epoch": 0.01789555192291147,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009999755129969303,
      "loss": 1.0921,
      "step": 208
    },
    {
      "epoch": 0.01798158823023316,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009999752446504294,
      "loss": 1.7542,
      "step": 209
    },
    {
      "epoch": 0.01806762453755485,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009999749748415981,
      "loss": 1.2017,
      "step": 210
    },
    {
      "epoch": 0.01815366084487654,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0009999747035704372,
      "loss": 1.5366,
      "step": 211
    },
    {
      "epoch": 0.01823969715219823,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009999744308369475,
      "loss": 1.9634,
      "step": 212
    },
    {
      "epoch": 0.01832573345951992,
      "grad_norm": 1.09375,
      "learning_rate": 0.00099997415664113,
      "loss": 1.3052,
      "step": 213
    },
    {
      "epoch": 0.018411769766841606,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009999738809829853,
      "loss": 1.2587,
      "step": 214
    },
    {
      "epoch": 0.018497806074163296,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999736038625142,
      "loss": 0.9864,
      "step": 215
    },
    {
      "epoch": 0.018583842381484986,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009999733252797176,
      "loss": 1.6881,
      "step": 216
    },
    {
      "epoch": 0.018669878688806676,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009999730452345962,
      "loss": 1.428,
      "step": 217
    },
    {
      "epoch": 0.018755914996128366,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009999727637271508,
      "loss": 1.637,
      "step": 218
    },
    {
      "epoch": 0.018841951303450056,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009999724807573825,
      "loss": 1.2136,
      "step": 219
    },
    {
      "epoch": 0.018927987610771746,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009999721963252918,
      "loss": 1.8377,
      "step": 220
    },
    {
      "epoch": 0.019014023918093436,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009999719104308797,
      "loss": 1.2775,
      "step": 221
    },
    {
      "epoch": 0.019100060225415126,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999716230741469,
      "loss": 1.3269,
      "step": 222
    },
    {
      "epoch": 0.019186096532736816,
      "grad_norm": 1.125,
      "learning_rate": 0.0009999713342550944,
      "loss": 0.7134,
      "step": 223
    },
    {
      "epoch": 0.019272132840058506,
      "grad_norm": 1.28125,
      "learning_rate": 0.000999971043973723,
      "loss": 1.1016,
      "step": 224
    },
    {
      "epoch": 0.019358169147380193,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009999707522300337,
      "loss": 1.5509,
      "step": 225
    },
    {
      "epoch": 0.019444205454701883,
      "grad_norm": 1.140625,
      "learning_rate": 0.000999970459024027,
      "loss": 0.8488,
      "step": 226
    },
    {
      "epoch": 0.019530241762023573,
      "grad_norm": 1.0625,
      "learning_rate": 0.000999970164355704,
      "loss": 1.4885,
      "step": 227
    },
    {
      "epoch": 0.019616278069345263,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009999698682250655,
      "loss": 1.3996,
      "step": 228
    },
    {
      "epoch": 0.019702314376666953,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009999695706321124,
      "loss": 1.0619,
      "step": 229
    },
    {
      "epoch": 0.019788350683988643,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009999692715768455,
      "loss": 1.6555,
      "step": 230
    },
    {
      "epoch": 0.019874386991310333,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0009999689710592657,
      "loss": 1.0889,
      "step": 231
    },
    {
      "epoch": 0.019960423298632023,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000999968669079374,
      "loss": 0.6851,
      "step": 232
    },
    {
      "epoch": 0.020046459605953713,
      "grad_norm": 1.078125,
      "learning_rate": 0.000999968365637171,
      "loss": 1.5584,
      "step": 233
    },
    {
      "epoch": 0.020132495913275403,
      "grad_norm": 1.296875,
      "learning_rate": 0.000999968060732658,
      "loss": 1.7551,
      "step": 234
    },
    {
      "epoch": 0.020218532220597094,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009999677543658355,
      "loss": 1.0813,
      "step": 235
    },
    {
      "epoch": 0.02030456852791878,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0009999674465367046,
      "loss": 0.8315,
      "step": 236
    },
    {
      "epoch": 0.02039060483524047,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0009999671372452662,
      "loss": 1.4632,
      "step": 237
    },
    {
      "epoch": 0.02047664114256216,
      "grad_norm": 1.375,
      "learning_rate": 0.0009999668264915212,
      "loss": 1.3855,
      "step": 238
    },
    {
      "epoch": 0.02056267744988385,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009999665142754704,
      "loss": 0.9653,
      "step": 239
    },
    {
      "epoch": 0.02064871375720554,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009999662005971146,
      "loss": 1.2018,
      "step": 240
    },
    {
      "epoch": 0.02073475006452723,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009999658854564551,
      "loss": 1.7052,
      "step": 241
    },
    {
      "epoch": 0.02082078637184892,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009999655688534928,
      "loss": 1.3454,
      "step": 242
    },
    {
      "epoch": 0.02090682267917061,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009999652507882283,
      "loss": 1.1441,
      "step": 243
    },
    {
      "epoch": 0.0209928589864923,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009999649312606624,
      "loss": 1.2651,
      "step": 244
    },
    {
      "epoch": 0.02107889529381399,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009999646102707967,
      "loss": 1.9789,
      "step": 245
    },
    {
      "epoch": 0.02116493160113568,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009999642878186314,
      "loss": 1.0946,
      "step": 246
    },
    {
      "epoch": 0.021250967908457367,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000999963963904168,
      "loss": 1.3104,
      "step": 247
    },
    {
      "epoch": 0.021337004215779058,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009999636385274074,
      "loss": 1.3012,
      "step": 248
    },
    {
      "epoch": 0.021423040523100748,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009999633116883502,
      "loss": 1.661,
      "step": 249
    },
    {
      "epoch": 0.021509076830422438,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009999629833869976,
      "loss": 1.7563,
      "step": 250
    },
    {
      "epoch": 0.021595113137744128,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009999626536233504,
      "loss": 1.6642,
      "step": 251
    },
    {
      "epoch": 0.021681149445065818,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999623223974098,
      "loss": 1.7497,
      "step": 252
    },
    {
      "epoch": 0.021767185752387508,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009999619897091765,
      "loss": 0.7691,
      "step": 253
    },
    {
      "epoch": 0.021853222059709198,
      "grad_norm": 1.046875,
      "learning_rate": 0.000999961655558652,
      "loss": 1.4918,
      "step": 254
    },
    {
      "epoch": 0.021939258367030888,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009999613199458365,
      "loss": 1.0666,
      "step": 255
    },
    {
      "epoch": 0.022025294674352578,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009999609828707315,
      "loss": 1.8481,
      "step": 256
    },
    {
      "epoch": 0.022111330981674268,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009999606443333378,
      "loss": 1.1794,
      "step": 257
    },
    {
      "epoch": 0.022197367288995955,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009999603043336566,
      "loss": 1.1271,
      "step": 258
    },
    {
      "epoch": 0.022283403596317645,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009999599628716888,
      "loss": 1.647,
      "step": 259
    },
    {
      "epoch": 0.022369439903639335,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000999959619947435,
      "loss": 1.0059,
      "step": 260
    },
    {
      "epoch": 0.022455476210961025,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000999959275560897,
      "loss": 1.0804,
      "step": 261
    },
    {
      "epoch": 0.022541512518282715,
      "grad_norm": 1.3125,
      "learning_rate": 0.000999958929712075,
      "loss": 1.1958,
      "step": 262
    },
    {
      "epoch": 0.022627548825604405,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009999585824009703,
      "loss": 1.7705,
      "step": 263
    },
    {
      "epoch": 0.022713585132926095,
      "grad_norm": 1.0,
      "learning_rate": 0.0009999582336275843,
      "loss": 0.9266,
      "step": 264
    },
    {
      "epoch": 0.022799621440247785,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009999578833919174,
      "loss": 1.3033,
      "step": 265
    },
    {
      "epoch": 0.022885657747569475,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009999575316939713,
      "loss": 1.5945,
      "step": 266
    },
    {
      "epoch": 0.022971694054891165,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009999571785337463,
      "loss": 1.8388,
      "step": 267
    },
    {
      "epoch": 0.023057730362212855,
      "grad_norm": 1.109375,
      "learning_rate": 0.000999956823911244,
      "loss": 1.4843,
      "step": 268
    },
    {
      "epoch": 0.023143766669534542,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999564678264653,
      "loss": 1.1793,
      "step": 269
    },
    {
      "epoch": 0.023229802976856232,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000999956110279411,
      "loss": 0.7998,
      "step": 270
    },
    {
      "epoch": 0.023315839284177922,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009999557512700822,
      "loss": 1.6635,
      "step": 271
    },
    {
      "epoch": 0.023401875591499612,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009999553907984802,
      "loss": 1.6003,
      "step": 272
    },
    {
      "epoch": 0.023487911898821302,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000999955028864606,
      "loss": 0.9741,
      "step": 273
    },
    {
      "epoch": 0.023573948206142992,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009999546654684604,
      "loss": 1.9399,
      "step": 274
    },
    {
      "epoch": 0.023659984513464682,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009999543006100448,
      "loss": 1.2514,
      "step": 275
    },
    {
      "epoch": 0.023746020820786373,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009999539342893601,
      "loss": 1.295,
      "step": 276
    },
    {
      "epoch": 0.023832057128108063,
      "grad_norm": 0.8125,
      "learning_rate": 0.0009999535665064071,
      "loss": 1.1176,
      "step": 277
    },
    {
      "epoch": 0.023918093435429753,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009999531972611877,
      "loss": 1.7192,
      "step": 278
    },
    {
      "epoch": 0.024004129742751443,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000999952826553702,
      "loss": 1.0158,
      "step": 279
    },
    {
      "epoch": 0.02409016605007313,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009999524543839516,
      "loss": 1.4645,
      "step": 280
    },
    {
      "epoch": 0.02417620235739482,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009999520807519376,
      "loss": 1.1748,
      "step": 281
    },
    {
      "epoch": 0.02426223866471651,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000999951705657661,
      "loss": 1.3584,
      "step": 282
    },
    {
      "epoch": 0.0243482749720382,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009999513291011227,
      "loss": 0.907,
      "step": 283
    },
    {
      "epoch": 0.02443431127935989,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009999509510823241,
      "loss": 1.2408,
      "step": 284
    },
    {
      "epoch": 0.02452034758668158,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009999505716012662,
      "loss": 0.9284,
      "step": 285
    },
    {
      "epoch": 0.02460638389400327,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009999501906579502,
      "loss": 1.8068,
      "step": 286
    },
    {
      "epoch": 0.02469242020132496,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009999498082523771,
      "loss": 1.6773,
      "step": 287
    },
    {
      "epoch": 0.02477845650864665,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000999949424384548,
      "loss": 1.6724,
      "step": 288
    },
    {
      "epoch": 0.02486449281596834,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000999949039054464,
      "loss": 1.6939,
      "step": 289
    },
    {
      "epoch": 0.02495052912329003,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999486522621265,
      "loss": 0.7765,
      "step": 290
    },
    {
      "epoch": 0.025036565430611717,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009999482640075364,
      "loss": 1.1679,
      "step": 291
    },
    {
      "epoch": 0.025122601737933407,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999478742906947,
      "loss": 1.203,
      "step": 292
    },
    {
      "epoch": 0.025208638045255097,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009999474831116028,
      "loss": 1.8525,
      "step": 293
    },
    {
      "epoch": 0.025294674352576787,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999470904702617,
      "loss": 1.7013,
      "step": 294
    },
    {
      "epoch": 0.025380710659898477,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009999466963666725,
      "loss": 2.4592,
      "step": 295
    },
    {
      "epoch": 0.025466746967220167,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999463008008368,
      "loss": 1.1464,
      "step": 296
    },
    {
      "epoch": 0.025552783274541857,
      "grad_norm": 0.875,
      "learning_rate": 0.0009999459037727551,
      "loss": 1.0979,
      "step": 297
    },
    {
      "epoch": 0.025638819581863547,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009999455052824289,
      "loss": 1.6259,
      "step": 298
    },
    {
      "epoch": 0.025724855889185237,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009999451053298593,
      "loss": 1.4003,
      "step": 299
    },
    {
      "epoch": 0.025810892196506927,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999447039150476,
      "loss": 1.6582,
      "step": 300
    },
    {
      "epoch": 0.025896928503828617,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009999443010379949,
      "loss": 1.8285,
      "step": 301
    },
    {
      "epoch": 0.025982964811150304,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009999438966987022,
      "loss": 1.6116,
      "step": 302
    },
    {
      "epoch": 0.026069001118471994,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009999434908971709,
      "loss": 1.9283,
      "step": 303
    },
    {
      "epoch": 0.026155037425793684,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009999430836334022,
      "loss": 2.0978,
      "step": 304
    },
    {
      "epoch": 0.026241073733115374,
      "grad_norm": 0.828125,
      "learning_rate": 0.0009999426749073973,
      "loss": 1.5183,
      "step": 305
    },
    {
      "epoch": 0.026327110040437064,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009999422647191572,
      "loss": 2.172,
      "step": 306
    },
    {
      "epoch": 0.026413146347758754,
      "grad_norm": 1.515625,
      "learning_rate": 0.000999941853068683,
      "loss": 2.0693,
      "step": 307
    },
    {
      "epoch": 0.026499182655080444,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009999414399559762,
      "loss": 1.1869,
      "step": 308
    },
    {
      "epoch": 0.026585218962402134,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000999941025381038,
      "loss": 1.2077,
      "step": 309
    },
    {
      "epoch": 0.026671255269723824,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009999406093438698,
      "loss": 1.1796,
      "step": 310
    },
    {
      "epoch": 0.026757291577045515,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999401918444723,
      "loss": 1.3989,
      "step": 311
    },
    {
      "epoch": 0.026843327884367205,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000999939772882847,
      "loss": 1.6521,
      "step": 312
    },
    {
      "epoch": 0.02692936419168889,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000999939352458995,
      "loss": 1.717,
      "step": 313
    },
    {
      "epoch": 0.02701540049901058,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009999389305729177,
      "loss": 1.944,
      "step": 314
    },
    {
      "epoch": 0.02710143680633227,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009999385072246164,
      "loss": 1.5468,
      "step": 315
    },
    {
      "epoch": 0.02718747311365396,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000999938082414092,
      "loss": 1.3585,
      "step": 316
    },
    {
      "epoch": 0.02727350942097565,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000999937656141346,
      "loss": 0.9906,
      "step": 317
    },
    {
      "epoch": 0.02735954572829734,
      "grad_norm": 0.796875,
      "learning_rate": 0.0009999372284063795,
      "loss": 1.5008,
      "step": 318
    },
    {
      "epoch": 0.02744558203561903,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000999936799209194,
      "loss": 1.7013,
      "step": 319
    },
    {
      "epoch": 0.02753161834294072,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009999363685497904,
      "loss": 2.0696,
      "step": 320
    },
    {
      "epoch": 0.027617654650262412,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009999359364281704,
      "loss": 1.7832,
      "step": 321
    },
    {
      "epoch": 0.027703690957584102,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000999935502844335,
      "loss": 1.424,
      "step": 322
    },
    {
      "epoch": 0.027789727264905792,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999350677982852,
      "loss": 1.1182,
      "step": 323
    },
    {
      "epoch": 0.02787576357222748,
      "grad_norm": 1.046875,
      "learning_rate": 0.000999934631290023,
      "loss": 1.4948,
      "step": 324
    },
    {
      "epoch": 0.02796179987954917,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009999341933195488,
      "loss": 0.8513,
      "step": 325
    },
    {
      "epoch": 0.02804783618687086,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0009999337538868646,
      "loss": 1.2194,
      "step": 326
    },
    {
      "epoch": 0.02813387249419255,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009999333129919713,
      "loss": 1.2324,
      "step": 327
    },
    {
      "epoch": 0.02821990880151424,
      "grad_norm": 1.0625,
      "learning_rate": 0.00099993287063487,
      "loss": 1.8624,
      "step": 328
    },
    {
      "epoch": 0.02830594510883593,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999324268155627,
      "loss": 1.8406,
      "step": 329
    },
    {
      "epoch": 0.02839198141615762,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009999319815340501,
      "loss": 1.3087,
      "step": 330
    },
    {
      "epoch": 0.02847801772347931,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009999315347903337,
      "loss": 2.4103,
      "step": 331
    },
    {
      "epoch": 0.028564054030801,
      "grad_norm": 1.265625,
      "learning_rate": 0.000999931086584415,
      "loss": 1.1157,
      "step": 332
    },
    {
      "epoch": 0.02865009033812269,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009999306369162949,
      "loss": 1.0992,
      "step": 333
    },
    {
      "epoch": 0.02873612664544438,
      "grad_norm": 1.1875,
      "learning_rate": 0.000999930185785975,
      "loss": 1.1572,
      "step": 334
    },
    {
      "epoch": 0.028822162952766066,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009999297331934563,
      "loss": 1.3883,
      "step": 335
    },
    {
      "epoch": 0.028908199260087756,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009999292791387406,
      "loss": 1.4456,
      "step": 336
    },
    {
      "epoch": 0.028994235567409446,
      "grad_norm": 1.5625,
      "learning_rate": 0.000999928823621829,
      "loss": 1.5032,
      "step": 337
    },
    {
      "epoch": 0.029080271874731136,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009999283666427229,
      "loss": 1.2515,
      "step": 338
    },
    {
      "epoch": 0.029166308182052826,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0009999279082014232,
      "loss": 0.8146,
      "step": 339
    },
    {
      "epoch": 0.029252344489374516,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009999274482979318,
      "loss": 1.8193,
      "step": 340
    },
    {
      "epoch": 0.029338380796696206,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009999269869322498,
      "loss": 1.4751,
      "step": 341
    },
    {
      "epoch": 0.029424417104017896,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009999265241043788,
      "loss": 1.5818,
      "step": 342
    },
    {
      "epoch": 0.029510453411339586,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009999260598143199,
      "loss": 2.0047,
      "step": 343
    },
    {
      "epoch": 0.029596489718661276,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009999255940620743,
      "loss": 1.6837,
      "step": 344
    },
    {
      "epoch": 0.029682526025982967,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009999251268476437,
      "loss": 1.582,
      "step": 345
    },
    {
      "epoch": 0.029768562333304653,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009999246581710293,
      "loss": 1.7086,
      "step": 346
    },
    {
      "epoch": 0.029854598640626343,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009999241880322325,
      "loss": 2.2388,
      "step": 347
    },
    {
      "epoch": 0.029940634947948033,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999237164312546,
      "loss": 1.6101,
      "step": 348
    },
    {
      "epoch": 0.030026671255269723,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009999232433680972,
      "loss": 1.1651,
      "step": 349
    },
    {
      "epoch": 0.030112707562591413,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009999227688427615,
      "loss": 2.4859,
      "step": 350
    },
    {
      "epoch": 0.030198743869913103,
      "grad_norm": 1.140625,
      "learning_rate": 0.000999922292855249,
      "loss": 0.8109,
      "step": 351
    },
    {
      "epoch": 0.030284780177234794,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000999921815405561,
      "loss": 1.7344,
      "step": 352
    },
    {
      "epoch": 0.030370816484556484,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009999213364936988,
      "loss": 1.3068,
      "step": 353
    },
    {
      "epoch": 0.030456852791878174,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000999920856119664,
      "loss": 1.8955,
      "step": 354
    },
    {
      "epoch": 0.030542889099199864,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000999920374283458,
      "loss": 1.9888,
      "step": 355
    },
    {
      "epoch": 0.030628925406521554,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009999198909850819,
      "loss": 2.0341,
      "step": 356
    },
    {
      "epoch": 0.03071496171384324,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009999194062245377,
      "loss": 1.7829,
      "step": 357
    },
    {
      "epoch": 0.03080099802116493,
      "grad_norm": 1.453125,
      "learning_rate": 0.000999918920001826,
      "loss": 2.0414,
      "step": 358
    },
    {
      "epoch": 0.03088703432848662,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000999918432316949,
      "loss": 0.6283,
      "step": 359
    },
    {
      "epoch": 0.03097307063580831,
      "grad_norm": 1.078125,
      "learning_rate": 0.000999917943169908,
      "loss": 0.98,
      "step": 360
    },
    {
      "epoch": 0.03105910694313,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009999174525607039,
      "loss": 1.2706,
      "step": 361
    },
    {
      "epoch": 0.03114514325045169,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999169604893385,
      "loss": 1.4434,
      "step": 362
    },
    {
      "epoch": 0.03123117955777338,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0009999164669558134,
      "loss": 0.8519,
      "step": 363
    },
    {
      "epoch": 0.03131721586509507,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009999159719601297,
      "loss": 1.2324,
      "step": 364
    },
    {
      "epoch": 0.03140325217241676,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009999154755022891,
      "loss": 1.873,
      "step": 365
    },
    {
      "epoch": 0.03148928847973845,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000999914977582293,
      "loss": 1.216,
      "step": 366
    },
    {
      "epoch": 0.03157532478706014,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009999144782001429,
      "loss": 2.121,
      "step": 367
    },
    {
      "epoch": 0.03166136109438183,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00099991397735584,
      "loss": 1.5533,
      "step": 368
    },
    {
      "epoch": 0.03174739740170352,
      "grad_norm": 1.359375,
      "learning_rate": 0.000999913475049386,
      "loss": 1.418,
      "step": 369
    },
    {
      "epoch": 0.03183343370902521,
      "grad_norm": 1.046875,
      "learning_rate": 0.000999912971280782,
      "loss": 1.7923,
      "step": 370
    },
    {
      "epoch": 0.0319194700163469,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009999124660500303,
      "loss": 1.3065,
      "step": 371
    },
    {
      "epoch": 0.03200550632366859,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009999119593571315,
      "loss": 1.451,
      "step": 372
    },
    {
      "epoch": 0.032091542630990275,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009999114512020877,
      "loss": 1.4728,
      "step": 373
    },
    {
      "epoch": 0.03217757893831197,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999109415849,
      "loss": 1.3887,
      "step": 374
    },
    {
      "epoch": 0.032263615245633655,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009999104305055701,
      "loss": 2.1049,
      "step": 375
    },
    {
      "epoch": 0.03234965155295535,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009999099179640993,
      "loss": 1.0168,
      "step": 376
    },
    {
      "epoch": 0.032435687860277035,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009999094039604894,
      "loss": 0.6503,
      "step": 377
    },
    {
      "epoch": 0.03252172416759873,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009999088884947416,
      "loss": 1.2962,
      "step": 378
    },
    {
      "epoch": 0.032607760474920415,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009999083715668575,
      "loss": 2.2695,
      "step": 379
    },
    {
      "epoch": 0.03269379678224211,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009999078531768386,
      "loss": 1.4497,
      "step": 380
    },
    {
      "epoch": 0.032779833089563795,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009999073333246866,
      "loss": 1.0866,
      "step": 381
    },
    {
      "epoch": 0.03286586939688549,
      "grad_norm": 1.28125,
      "learning_rate": 0.000999906812010403,
      "loss": 1.7215,
      "step": 382
    },
    {
      "epoch": 0.032951905704207175,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009999062892339888,
      "loss": 0.9255,
      "step": 383
    },
    {
      "epoch": 0.03303794201152886,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009999057649954463,
      "loss": 1.1757,
      "step": 384
    },
    {
      "epoch": 0.033123978318850555,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999052392947764,
      "loss": 2.1057,
      "step": 385
    },
    {
      "epoch": 0.03321001462617224,
      "grad_norm": 0.921875,
      "learning_rate": 0.000999904712131981,
      "loss": 1.4327,
      "step": 386
    },
    {
      "epoch": 0.033296050933493936,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009999041835070616,
      "loss": 1.2006,
      "step": 387
    },
    {
      "epoch": 0.03338208724081562,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009999036534200196,
      "loss": 1.1485,
      "step": 388
    },
    {
      "epoch": 0.033468123548137316,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009999031218708566,
      "loss": 1.9005,
      "step": 389
    },
    {
      "epoch": 0.033554159855459,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009999025888595744,
      "loss": 1.4408,
      "step": 390
    },
    {
      "epoch": 0.033640196162780696,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009999020543861742,
      "loss": 1.1785,
      "step": 391
    },
    {
      "epoch": 0.03372623247010238,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009999015184506577,
      "loss": 2.466,
      "step": 392
    },
    {
      "epoch": 0.033812268777424076,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009999009810530265,
      "loss": 0.9248,
      "step": 393
    },
    {
      "epoch": 0.03389830508474576,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009999004421932822,
      "loss": 1.4565,
      "step": 394
    },
    {
      "epoch": 0.03398434139206745,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009998999018714263,
      "loss": 1.628,
      "step": 395
    },
    {
      "epoch": 0.03407037769938914,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009998993600874603,
      "loss": 1.265,
      "step": 396
    },
    {
      "epoch": 0.03415641400671083,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009998988168413861,
      "loss": 1.7926,
      "step": 397
    },
    {
      "epoch": 0.03424245031403252,
      "grad_norm": 2.0625,
      "learning_rate": 0.000999898272133205,
      "loss": 1.3841,
      "step": 398
    },
    {
      "epoch": 0.03432848662135421,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009998977259629187,
      "loss": 1.9804,
      "step": 399
    },
    {
      "epoch": 0.0344145229286759,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009998971783305288,
      "loss": 1.2225,
      "step": 400
    },
    {
      "epoch": 0.03450055923599759,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009998966292360369,
      "loss": 1.2605,
      "step": 401
    },
    {
      "epoch": 0.03458659554331928,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009998960786794444,
      "loss": 1.7625,
      "step": 402
    },
    {
      "epoch": 0.03467263185064097,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009998955266607534,
      "loss": 2.0243,
      "step": 403
    },
    {
      "epoch": 0.03475866815796266,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009998949731799649,
      "loss": 1.0203,
      "step": 404
    },
    {
      "epoch": 0.03484470446528435,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009998944182370811,
      "loss": 1.1201,
      "step": 405
    },
    {
      "epoch": 0.034930740772606036,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009998938618321033,
      "loss": 0.8424,
      "step": 406
    },
    {
      "epoch": 0.03501677707992773,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009998933039650328,
      "loss": 1.8571,
      "step": 407
    },
    {
      "epoch": 0.03510281338724942,
      "grad_norm": 0.984375,
      "learning_rate": 0.000999892744635872,
      "loss": 0.7547,
      "step": 408
    },
    {
      "epoch": 0.03518884969457111,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000999892183844622,
      "loss": 1.0126,
      "step": 409
    },
    {
      "epoch": 0.0352748860018928,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009998916215912847,
      "loss": 0.8118,
      "step": 410
    },
    {
      "epoch": 0.03536092230921449,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009998910578758616,
      "loss": 1.3055,
      "step": 411
    },
    {
      "epoch": 0.03544695861653618,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009998904926983544,
      "loss": 1.2951,
      "step": 412
    },
    {
      "epoch": 0.03553299492385787,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009998899260587645,
      "loss": 1.5887,
      "step": 413
    },
    {
      "epoch": 0.03561903123117956,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000999889357957094,
      "loss": 1.719,
      "step": 414
    },
    {
      "epoch": 0.03570506753850125,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998887883933443,
      "loss": 1.1941,
      "step": 415
    },
    {
      "epoch": 0.03579110384582294,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009998882173675172,
      "loss": 2.2298,
      "step": 416
    },
    {
      "epoch": 0.035877140153144624,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009998876448796142,
      "loss": 1.1705,
      "step": 417
    },
    {
      "epoch": 0.03596317646046632,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000999887070929637,
      "loss": 0.889,
      "step": 418
    },
    {
      "epoch": 0.036049212767788004,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009998864955175874,
      "loss": 1.2127,
      "step": 419
    },
    {
      "epoch": 0.0361352490751097,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000999885918643467,
      "loss": 1.3177,
      "step": 420
    },
    {
      "epoch": 0.036221285382431384,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009998853403072776,
      "loss": 1.3252,
      "step": 421
    },
    {
      "epoch": 0.03630732168975308,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009998847605090207,
      "loss": 1.0388,
      "step": 422
    },
    {
      "epoch": 0.036393357997074764,
      "grad_norm": 1.25,
      "learning_rate": 0.000999884179248698,
      "loss": 1.8186,
      "step": 423
    },
    {
      "epoch": 0.03647939430439646,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009998835965263117,
      "loss": 1.2202,
      "step": 424
    },
    {
      "epoch": 0.036565430611718144,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009998830123418628,
      "loss": 1.3599,
      "step": 425
    },
    {
      "epoch": 0.03665146691903984,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009998824266953534,
      "loss": 1.3032,
      "step": 426
    },
    {
      "epoch": 0.036737503226361524,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000999881839586785,
      "loss": 1.5828,
      "step": 427
    },
    {
      "epoch": 0.03682353953368321,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009998812510161597,
      "loss": 1.4544,
      "step": 428
    },
    {
      "epoch": 0.036909575841004905,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009998806609834785,
      "loss": 1.6721,
      "step": 429
    },
    {
      "epoch": 0.03699561214832659,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000999880069488744,
      "loss": 1.4869,
      "step": 430
    },
    {
      "epoch": 0.037081648455648285,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009998794765319575,
      "loss": 1.8696,
      "step": 431
    },
    {
      "epoch": 0.03716768476296997,
      "grad_norm": 1.5,
      "learning_rate": 0.0009998788821131207,
      "loss": 1.8058,
      "step": 432
    },
    {
      "epoch": 0.037253721070291665,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009998782862322352,
      "loss": 1.3351,
      "step": 433
    },
    {
      "epoch": 0.03733975737761335,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998776888893032,
      "loss": 1.1881,
      "step": 434
    },
    {
      "epoch": 0.037425793684935045,
      "grad_norm": 1.34375,
      "learning_rate": 0.000999877090084326,
      "loss": 1.6451,
      "step": 435
    },
    {
      "epoch": 0.03751182999225673,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009998764898173054,
      "loss": 1.8224,
      "step": 436
    },
    {
      "epoch": 0.037597866299578425,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009998758880882435,
      "loss": 1.1942,
      "step": 437
    },
    {
      "epoch": 0.03768390260690011,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009998752848971418,
      "loss": 1.9565,
      "step": 438
    },
    {
      "epoch": 0.0377699389142218,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009998746802440022,
      "loss": 1.6775,
      "step": 439
    },
    {
      "epoch": 0.03785597522154349,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009998740741288263,
      "loss": 1.486,
      "step": 440
    },
    {
      "epoch": 0.03794201152886518,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000999873466551616,
      "loss": 1.2321,
      "step": 441
    },
    {
      "epoch": 0.03802804783618687,
      "grad_norm": 1.109375,
      "learning_rate": 0.000999872857512373,
      "loss": 1.1897,
      "step": 442
    },
    {
      "epoch": 0.03811408414350856,
      "grad_norm": 1.515625,
      "learning_rate": 0.000999872247011099,
      "loss": 1.9862,
      "step": 443
    },
    {
      "epoch": 0.03820012045083025,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000999871635047796,
      "loss": 2.1568,
      "step": 444
    },
    {
      "epoch": 0.03828615675815194,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009998710216224658,
      "loss": 1.5089,
      "step": 445
    },
    {
      "epoch": 0.03837219306547363,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00099987040673511,
      "loss": 1.0655,
      "step": 446
    },
    {
      "epoch": 0.03845822937279532,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009998697903857304,
      "loss": 2.2668,
      "step": 447
    },
    {
      "epoch": 0.03854426568011701,
      "grad_norm": 1.21875,
      "learning_rate": 0.000999869172574329,
      "loss": 1.4408,
      "step": 448
    },
    {
      "epoch": 0.0386303019874387,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009998685533009074,
      "loss": 1.3487,
      "step": 449
    },
    {
      "epoch": 0.038716338294760386,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009998679325654675,
      "loss": 1.5339,
      "step": 450
    },
    {
      "epoch": 0.03880237460208208,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009998673103680112,
      "loss": 1.4967,
      "step": 451
    },
    {
      "epoch": 0.038888410909403766,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009998666867085402,
      "loss": 1.7513,
      "step": 452
    },
    {
      "epoch": 0.03897444721672546,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009998660615870563,
      "loss": 1.9441,
      "step": 453
    },
    {
      "epoch": 0.039060483524047146,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009998654350035613,
      "loss": 0.8262,
      "step": 454
    },
    {
      "epoch": 0.03914651983136884,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009998648069580573,
      "loss": 1.0094,
      "step": 455
    },
    {
      "epoch": 0.039232556138690526,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009998641774505458,
      "loss": 0.848,
      "step": 456
    },
    {
      "epoch": 0.03931859244601222,
      "grad_norm": 1.125,
      "learning_rate": 0.000999863546481029,
      "loss": 2.0069,
      "step": 457
    },
    {
      "epoch": 0.039404628753333906,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009998629140495084,
      "loss": 1.5384,
      "step": 458
    },
    {
      "epoch": 0.0394906650606556,
      "grad_norm": 1.234375,
      "learning_rate": 0.000999862280155986,
      "loss": 1.8629,
      "step": 459
    },
    {
      "epoch": 0.039576701367977286,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009998616448004638,
      "loss": 1.9215,
      "step": 460
    },
    {
      "epoch": 0.03966273767529897,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009998610079829434,
      "loss": 1.1686,
      "step": 461
    },
    {
      "epoch": 0.039748773982620667,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009998603697034267,
      "loss": 1.2415,
      "step": 462
    },
    {
      "epoch": 0.03983481028994235,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009998597299619156,
      "loss": 1.2121,
      "step": 463
    },
    {
      "epoch": 0.03992084659726405,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009998590887584122,
      "loss": 1.5184,
      "step": 464
    },
    {
      "epoch": 0.04000688290458573,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009998584460929182,
      "loss": 1.2507,
      "step": 465
    },
    {
      "epoch": 0.04009291921190743,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009998578019654353,
      "loss": 1.6185,
      "step": 466
    },
    {
      "epoch": 0.04017895551922911,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009998571563759656,
      "loss": 1.4179,
      "step": 467
    },
    {
      "epoch": 0.04026499182655081,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009998565093245109,
      "loss": 1.9187,
      "step": 468
    },
    {
      "epoch": 0.040351028133872494,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009998558608110733,
      "loss": 1.6498,
      "step": 469
    },
    {
      "epoch": 0.04043706444119419,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009998552108356543,
      "loss": 1.1639,
      "step": 470
    },
    {
      "epoch": 0.040523100748515874,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009998545593982561,
      "loss": 1.5649,
      "step": 471
    },
    {
      "epoch": 0.04060913705583756,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009998539064988805,
      "loss": 0.7886,
      "step": 472
    },
    {
      "epoch": 0.040695173363159254,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009998532521375294,
      "loss": 1.4231,
      "step": 473
    },
    {
      "epoch": 0.04078120967048094,
      "grad_norm": 1.078125,
      "learning_rate": 0.000999852596314205,
      "loss": 0.8964,
      "step": 474
    },
    {
      "epoch": 0.040867245977802634,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009998519390289086,
      "loss": 1.4631,
      "step": 475
    },
    {
      "epoch": 0.04095328228512432,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000999851280281643,
      "loss": 1.7266,
      "step": 476
    },
    {
      "epoch": 0.041039318592446014,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000999850620072409,
      "loss": 1.7392,
      "step": 477
    },
    {
      "epoch": 0.0411253548997677,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009998499584012097,
      "loss": 0.7894,
      "step": 478
    },
    {
      "epoch": 0.041211391207089394,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009998492952680463,
      "loss": 1.1063,
      "step": 479
    },
    {
      "epoch": 0.04129742751441108,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998486306729208,
      "loss": 2.6225,
      "step": 480
    },
    {
      "epoch": 0.041383463821732774,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009998479646158354,
      "loss": 1.9214,
      "step": 481
    },
    {
      "epoch": 0.04146950012905446,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009998472970967917,
      "loss": 1.1078,
      "step": 482
    },
    {
      "epoch": 0.04155553643637615,
      "grad_norm": 1.296875,
      "learning_rate": 0.000999846628115792,
      "loss": 1.7865,
      "step": 483
    },
    {
      "epoch": 0.04164157274369784,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009998459576728382,
      "loss": 1.9721,
      "step": 484
    },
    {
      "epoch": 0.04172760905101953,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009998452857679324,
      "loss": 1.7817,
      "step": 485
    },
    {
      "epoch": 0.04181364535834122,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000999844612401076,
      "loss": 1.1168,
      "step": 486
    },
    {
      "epoch": 0.04189968166566291,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009998439375722714,
      "loss": 1.4437,
      "step": 487
    },
    {
      "epoch": 0.0419857179729846,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998432612815206,
      "loss": 0.8741,
      "step": 488
    },
    {
      "epoch": 0.04207175428030629,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009998425835288254,
      "loss": 1.1279,
      "step": 489
    },
    {
      "epoch": 0.04215779058762798,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009998419043141878,
      "loss": 1.1016,
      "step": 490
    },
    {
      "epoch": 0.04224382689494967,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00099984122363761,
      "loss": 1.8912,
      "step": 491
    },
    {
      "epoch": 0.04232986320227136,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009998405414990937,
      "loss": 1.9,
      "step": 492
    },
    {
      "epoch": 0.04241589950959305,
      "grad_norm": 1.234375,
      "learning_rate": 0.000999839857898641,
      "loss": 1.9092,
      "step": 493
    },
    {
      "epoch": 0.042501935816914735,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009998391728362542,
      "loss": 1.2349,
      "step": 494
    },
    {
      "epoch": 0.04258797212423643,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009998384863119348,
      "loss": 1.9334,
      "step": 495
    },
    {
      "epoch": 0.042674008431558115,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000999837798325685,
      "loss": 1.2865,
      "step": 496
    },
    {
      "epoch": 0.04276004473887981,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000999837108877507,
      "loss": 1.7101,
      "step": 497
    },
    {
      "epoch": 0.042846081046201495,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009998364179674025,
      "loss": 1.5923,
      "step": 498
    },
    {
      "epoch": 0.04293211735352319,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009998357255953738,
      "loss": 1.2941,
      "step": 499
    },
    {
      "epoch": 0.043018153660844875,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009998350317614228,
      "loss": 1.513,
      "step": 500
    },
    {
      "epoch": 0.04310418996816657,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009998343364655516,
      "loss": 1.9419,
      "step": 501
    },
    {
      "epoch": 0.043190226275488255,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009998336397077621,
      "loss": 1.4357,
      "step": 502
    },
    {
      "epoch": 0.04327626258280995,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009998329414880565,
      "loss": 1.7493,
      "step": 503
    },
    {
      "epoch": 0.043362298890131636,
      "grad_norm": 1.25,
      "learning_rate": 0.0009998322418064366,
      "loss": 0.9392,
      "step": 504
    },
    {
      "epoch": 0.04344833519745332,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009998315406629047,
      "loss": 1.832,
      "step": 505
    },
    {
      "epoch": 0.043534371504775016,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009998308380574628,
      "loss": 2.2116,
      "step": 506
    },
    {
      "epoch": 0.0436204078120967,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998301339901128,
      "loss": 1.7582,
      "step": 507
    },
    {
      "epoch": 0.043706444119418396,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009998294284608571,
      "loss": 1.4191,
      "step": 508
    },
    {
      "epoch": 0.04379248042674008,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009998287214696973,
      "loss": 1.2912,
      "step": 509
    },
    {
      "epoch": 0.043878516734061776,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009998280130166356,
      "loss": 1.133,
      "step": 510
    },
    {
      "epoch": 0.04396455304138346,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009998273031016744,
      "loss": 1.9422,
      "step": 511
    },
    {
      "epoch": 0.044050589348705156,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009998265917248155,
      "loss": 1.8237,
      "step": 512
    },
    {
      "epoch": 0.04413662565602684,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009998258788860611,
      "loss": 1.4057,
      "step": 513
    },
    {
      "epoch": 0.044222661963348536,
      "grad_norm": 1.5,
      "learning_rate": 0.0009998251645854132,
      "loss": 1.8922,
      "step": 514
    },
    {
      "epoch": 0.04430869827067022,
      "grad_norm": 1.0625,
      "learning_rate": 0.000999824448822874,
      "loss": 1.0171,
      "step": 515
    },
    {
      "epoch": 0.04439473457799191,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009998237315984452,
      "loss": 1.7028,
      "step": 516
    },
    {
      "epoch": 0.0444807708853136,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009998230129121292,
      "loss": 2.0678,
      "step": 517
    },
    {
      "epoch": 0.04456680719263529,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009998222927639284,
      "loss": 1.1501,
      "step": 518
    },
    {
      "epoch": 0.04465284349995698,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009998215711538444,
      "loss": 1.6205,
      "step": 519
    },
    {
      "epoch": 0.04473887980727867,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009998208480818795,
      "loss": 1.7321,
      "step": 520
    },
    {
      "epoch": 0.04482491611460036,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000999820123548036,
      "loss": 1.4262,
      "step": 521
    },
    {
      "epoch": 0.04491095242192205,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009998193975523158,
      "loss": 1.0477,
      "step": 522
    },
    {
      "epoch": 0.04499698872924374,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000999818670094721,
      "loss": 1.4498,
      "step": 523
    },
    {
      "epoch": 0.04508302503656543,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009998179411752537,
      "loss": 1.8378,
      "step": 524
    },
    {
      "epoch": 0.045169061343887124,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009998172107939162,
      "loss": 1.4459,
      "step": 525
    },
    {
      "epoch": 0.04525509765120881,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009998164789507106,
      "loss": 1.292,
      "step": 526
    },
    {
      "epoch": 0.0453411339585305,
      "grad_norm": 1.234375,
      "learning_rate": 0.000999815745645639,
      "loss": 1.5867,
      "step": 527
    },
    {
      "epoch": 0.04542717026585219,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009998150108787036,
      "loss": 0.9621,
      "step": 528
    },
    {
      "epoch": 0.04551320657317388,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009998142746499065,
      "loss": 1.3551,
      "step": 529
    },
    {
      "epoch": 0.04559924288049557,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998135369592499,
      "loss": 1.0287,
      "step": 530
    },
    {
      "epoch": 0.04568527918781726,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009998127978067357,
      "loss": 2.1588,
      "step": 531
    },
    {
      "epoch": 0.04577131549513895,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009998120571923664,
      "loss": 1.8463,
      "step": 532
    },
    {
      "epoch": 0.04585735180246064,
      "grad_norm": 1.484375,
      "learning_rate": 0.000999811315116144,
      "loss": 1.1294,
      "step": 533
    },
    {
      "epoch": 0.04594338810978233,
      "grad_norm": 1.625,
      "learning_rate": 0.000999810571578071,
      "loss": 1.6497,
      "step": 534
    },
    {
      "epoch": 0.04602942441710402,
      "grad_norm": 1.09375,
      "learning_rate": 0.000999809826578149,
      "loss": 1.8022,
      "step": 535
    },
    {
      "epoch": 0.04611546072442571,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009998090801163805,
      "loss": 1.0044,
      "step": 536
    },
    {
      "epoch": 0.0462014970317474,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009998083321927675,
      "loss": 1.107,
      "step": 537
    },
    {
      "epoch": 0.046287533339069084,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0009998075828073125,
      "loss": 1.3136,
      "step": 538
    },
    {
      "epoch": 0.04637356964639078,
      "grad_norm": 1.125,
      "learning_rate": 0.0009998068319600176,
      "loss": 1.1851,
      "step": 539
    },
    {
      "epoch": 0.046459605953712464,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009998060796508848,
      "loss": 1.4609,
      "step": 540
    },
    {
      "epoch": 0.04654564226103416,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009998053258799165,
      "loss": 1.9917,
      "step": 541
    },
    {
      "epoch": 0.046631678568355844,
      "grad_norm": 1.375,
      "learning_rate": 0.0009998045706471148,
      "loss": 1.6961,
      "step": 542
    },
    {
      "epoch": 0.04671771487567754,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009998038139524819,
      "loss": 1.7271,
      "step": 543
    },
    {
      "epoch": 0.046803751182999224,
      "grad_norm": 1.25,
      "learning_rate": 0.0009998030557960202,
      "loss": 2.0037,
      "step": 544
    },
    {
      "epoch": 0.04688978749032092,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009998022961777316,
      "loss": 1.3414,
      "step": 545
    },
    {
      "epoch": 0.046975823797642605,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009998015350976186,
      "loss": 1.9735,
      "step": 546
    },
    {
      "epoch": 0.0470618601049643,
      "grad_norm": 1.375,
      "learning_rate": 0.0009998007725556832,
      "loss": 2.2454,
      "step": 547
    },
    {
      "epoch": 0.047147896412285985,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000999800008551928,
      "loss": 0.8096,
      "step": 548
    },
    {
      "epoch": 0.04723393271960767,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009997992430863546,
      "loss": 1.708,
      "step": 549
    },
    {
      "epoch": 0.047319969026929365,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009997984761589659,
      "loss": 2.1432,
      "step": 550
    },
    {
      "epoch": 0.04740600533425105,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009997977077697638,
      "loss": 1.0742,
      "step": 551
    },
    {
      "epoch": 0.047492041641572745,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009997969379187505,
      "loss": 1.528,
      "step": 552
    },
    {
      "epoch": 0.04757807794889443,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009997961666059283,
      "loss": 1.2839,
      "step": 553
    },
    {
      "epoch": 0.047664114256216125,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009997953938312998,
      "loss": 1.0971,
      "step": 554
    },
    {
      "epoch": 0.04775015056353781,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009997946195948668,
      "loss": 1.1312,
      "step": 555
    },
    {
      "epoch": 0.047836186870859505,
      "grad_norm": 1.234375,
      "learning_rate": 0.000999793843896632,
      "loss": 0.8609,
      "step": 556
    },
    {
      "epoch": 0.04792222317818119,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000999793066736597,
      "loss": 1.4039,
      "step": 557
    },
    {
      "epoch": 0.048008259485502885,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009997922881147647,
      "loss": 1.1499,
      "step": 558
    },
    {
      "epoch": 0.04809429579282457,
      "grad_norm": 1.4375,
      "learning_rate": 0.000999791508031137,
      "loss": 1.5576,
      "step": 559
    },
    {
      "epoch": 0.04818033210014626,
      "grad_norm": 0.84375,
      "learning_rate": 0.0009997907264857166,
      "loss": 1.0287,
      "step": 560
    },
    {
      "epoch": 0.04826636840746795,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009997899434785055,
      "loss": 2.0485,
      "step": 561
    },
    {
      "epoch": 0.04835240471478964,
      "grad_norm": 1.09375,
      "learning_rate": 0.000999789159009506,
      "loss": 0.8547,
      "step": 562
    },
    {
      "epoch": 0.04843844102211133,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009997883730787202,
      "loss": 1.0762,
      "step": 563
    },
    {
      "epoch": 0.04852447732943302,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009997875856861506,
      "loss": 1.3666,
      "step": 564
    },
    {
      "epoch": 0.04861051363675471,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009997867968317997,
      "loss": 1.3422,
      "step": 565
    },
    {
      "epoch": 0.0486965499440764,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009997860065156696,
      "loss": 1.6214,
      "step": 566
    },
    {
      "epoch": 0.04878258625139809,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009997852147377627,
      "loss": 1.639,
      "step": 567
    },
    {
      "epoch": 0.04886862255871978,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000999784421498081,
      "loss": 1.0053,
      "step": 568
    },
    {
      "epoch": 0.04895465886604147,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009997836267966274,
      "loss": 1.2187,
      "step": 569
    },
    {
      "epoch": 0.04904069517336316,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000999782830633404,
      "loss": 0.9453,
      "step": 570
    },
    {
      "epoch": 0.049126731480684846,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009997820330084125,
      "loss": 1.2812,
      "step": 571
    },
    {
      "epoch": 0.04921276778800654,
      "grad_norm": 1.28125,
      "learning_rate": 0.000999781233921656,
      "loss": 1.3817,
      "step": 572
    },
    {
      "epoch": 0.049298804095328226,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009997804333731367,
      "loss": 1.5734,
      "step": 573
    },
    {
      "epoch": 0.04938484040264992,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009997796313628566,
      "loss": 1.4311,
      "step": 574
    },
    {
      "epoch": 0.049470876709971606,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009997788278908184,
      "loss": 1.6906,
      "step": 575
    },
    {
      "epoch": 0.0495569130172933,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009997780229570245,
      "loss": 1.4967,
      "step": 576
    },
    {
      "epoch": 0.049642949324614986,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000999777216561477,
      "loss": 1.2695,
      "step": 577
    },
    {
      "epoch": 0.04972898563193668,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000999776408704178,
      "loss": 0.8627,
      "step": 578
    },
    {
      "epoch": 0.049815021939258367,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009997755993851308,
      "loss": 1.0178,
      "step": 579
    },
    {
      "epoch": 0.04990105824658006,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009997747886043367,
      "loss": 1.5012,
      "step": 580
    },
    {
      "epoch": 0.04998709455390175,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009997739763617987,
      "loss": 1.8907,
      "step": 581
    },
    {
      "epoch": 0.05007313086122343,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000999773162657519,
      "loss": 1.2472,
      "step": 582
    },
    {
      "epoch": 0.05015916716854513,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009997723474915,
      "loss": 1.0839,
      "step": 583
    },
    {
      "epoch": 0.05024520347586681,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009997715308637442,
      "loss": 1.5047,
      "step": 584
    },
    {
      "epoch": 0.05033123978318851,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009997707127742538,
      "loss": 1.8525,
      "step": 585
    },
    {
      "epoch": 0.050417276090510194,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009997698932230313,
      "loss": 1.2152,
      "step": 586
    },
    {
      "epoch": 0.05050331239783189,
      "grad_norm": 0.875,
      "learning_rate": 0.000999769072210079,
      "loss": 1.1605,
      "step": 587
    },
    {
      "epoch": 0.050589348705153574,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009997682497353993,
      "loss": 1.1228,
      "step": 588
    },
    {
      "epoch": 0.05067538501247527,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009997674257989947,
      "loss": 1.0512,
      "step": 589
    },
    {
      "epoch": 0.050761421319796954,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009997666004008677,
      "loss": 1.035,
      "step": 590
    },
    {
      "epoch": 0.05084745762711865,
      "grad_norm": 3.28125,
      "learning_rate": 0.0009997657735410204,
      "loss": 1.7301,
      "step": 591
    },
    {
      "epoch": 0.050933493934440334,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009997649452194556,
      "loss": 1.8586,
      "step": 592
    },
    {
      "epoch": 0.05101953024176202,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009997641154361754,
      "loss": 1.5138,
      "step": 593
    },
    {
      "epoch": 0.051105566549083714,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009997632841911824,
      "loss": 1.7533,
      "step": 594
    },
    {
      "epoch": 0.0511916028564054,
      "grad_norm": 1.34375,
      "learning_rate": 0.000999762451484479,
      "loss": 1.2115,
      "step": 595
    },
    {
      "epoch": 0.051277639163727094,
      "grad_norm": 1.75,
      "learning_rate": 0.0009997616173160677,
      "loss": 1.5356,
      "step": 596
    },
    {
      "epoch": 0.05136367547104878,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000999760781685951,
      "loss": 1.1043,
      "step": 597
    },
    {
      "epoch": 0.051449711778370474,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009997599445941308,
      "loss": 1.7646,
      "step": 598
    },
    {
      "epoch": 0.05153574808569216,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009997591060406101,
      "loss": 1.6843,
      "step": 599
    },
    {
      "epoch": 0.051621784393013855,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009997582660253915,
      "loss": 1.0102,
      "step": 600
    },
    {
      "epoch": 0.05170782070033554,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009997574245484769,
      "loss": 1.3301,
      "step": 601
    },
    {
      "epoch": 0.051793857007657235,
      "grad_norm": 1.171875,
      "learning_rate": 0.000999756581609869,
      "loss": 1.1007,
      "step": 602
    },
    {
      "epoch": 0.05187989331497892,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009997557372095705,
      "loss": 2.1306,
      "step": 603
    },
    {
      "epoch": 0.05196592962230061,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009997548913475835,
      "loss": 1.5189,
      "step": 604
    },
    {
      "epoch": 0.0520519659296223,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009997540440239107,
      "loss": 0.6569,
      "step": 605
    },
    {
      "epoch": 0.05213800223694399,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009997531952385546,
      "loss": 1.7679,
      "step": 606
    },
    {
      "epoch": 0.05222403854426568,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009997523449915174,
      "loss": 2.0924,
      "step": 607
    },
    {
      "epoch": 0.05231007485158737,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009997514932828019,
      "loss": 0.8031,
      "step": 608
    },
    {
      "epoch": 0.05239611115890906,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009997506401124105,
      "loss": 1.2539,
      "step": 609
    },
    {
      "epoch": 0.05248214746623075,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009997497854803458,
      "loss": 1.9589,
      "step": 610
    },
    {
      "epoch": 0.05256818377355244,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00099974892938661,
      "loss": 1.8942,
      "step": 611
    },
    {
      "epoch": 0.05265422008087413,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009997480718312057,
      "loss": 1.2307,
      "step": 612
    },
    {
      "epoch": 0.05274025638819582,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009997472128141355,
      "loss": 0.8422,
      "step": 613
    },
    {
      "epoch": 0.05282629269551751,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000999746352335402,
      "loss": 1.4275,
      "step": 614
    },
    {
      "epoch": 0.052912329002839195,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009997454903950076,
      "loss": 1.0249,
      "step": 615
    },
    {
      "epoch": 0.05299836531016089,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009997446269929548,
      "loss": 1.9786,
      "step": 616
    },
    {
      "epoch": 0.053084401617482575,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009997437621292462,
      "loss": 1.1925,
      "step": 617
    },
    {
      "epoch": 0.05317043792480427,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009997428958038843,
      "loss": 1.8465,
      "step": 618
    },
    {
      "epoch": 0.053256474232125955,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009997420280168717,
      "loss": 1.6857,
      "step": 619
    },
    {
      "epoch": 0.05334251053944765,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009997411587682108,
      "loss": 1.5509,
      "step": 620
    },
    {
      "epoch": 0.053428546846769336,
      "grad_norm": 1.21875,
      "learning_rate": 0.000999740288057904,
      "loss": 1.5842,
      "step": 621
    },
    {
      "epoch": 0.05351458315409103,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009997394158859543,
      "loss": 1.7636,
      "step": 622
    },
    {
      "epoch": 0.053600619461412716,
      "grad_norm": 0.921875,
      "learning_rate": 0.000999738542252364,
      "loss": 2.0446,
      "step": 623
    },
    {
      "epoch": 0.05368665576873441,
      "grad_norm": 6.15625,
      "learning_rate": 0.0009997376671571356,
      "loss": 0.9565,
      "step": 624
    },
    {
      "epoch": 0.053772692076056096,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009997367906002716,
      "loss": 1.5884,
      "step": 625
    },
    {
      "epoch": 0.05385872838337778,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000999735912581775,
      "loss": 2.4666,
      "step": 626
    },
    {
      "epoch": 0.053944764690699476,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009997350331016478,
      "loss": 1.6302,
      "step": 627
    },
    {
      "epoch": 0.05403080099802116,
      "grad_norm": 1.359375,
      "learning_rate": 0.000999734152159893,
      "loss": 1.291,
      "step": 628
    },
    {
      "epoch": 0.054116837305342856,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009997332697565127,
      "loss": 1.8335,
      "step": 629
    },
    {
      "epoch": 0.05420287361266454,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009997323858915099,
      "loss": 1.7679,
      "step": 630
    },
    {
      "epoch": 0.054288909919986236,
      "grad_norm": 1.3125,
      "learning_rate": 0.000999731500564887,
      "loss": 2.0914,
      "step": 631
    },
    {
      "epoch": 0.05437494622730792,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009997306137766468,
      "loss": 1.9081,
      "step": 632
    },
    {
      "epoch": 0.054460982534629616,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009997297255267916,
      "loss": 1.1827,
      "step": 633
    },
    {
      "epoch": 0.0545470188419513,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009997288358153244,
      "loss": 1.5782,
      "step": 634
    },
    {
      "epoch": 0.054633055149273,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009997279446422474,
      "loss": 1.5764,
      "step": 635
    },
    {
      "epoch": 0.05471909145659468,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009997270520075631,
      "loss": 1.4276,
      "step": 636
    },
    {
      "epoch": 0.05480512776391637,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009997261579112748,
      "loss": 1.7193,
      "step": 637
    },
    {
      "epoch": 0.05489116407123806,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009997252623533844,
      "loss": 1.2215,
      "step": 638
    },
    {
      "epoch": 0.05497720037855975,
      "grad_norm": 1.015625,
      "learning_rate": 0.000999724365333895,
      "loss": 0.9428,
      "step": 639
    },
    {
      "epoch": 0.05506323668588144,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009997234668528087,
      "loss": 1.1261,
      "step": 640
    },
    {
      "epoch": 0.05514927299320313,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009997225669101288,
      "loss": 1.6906,
      "step": 641
    },
    {
      "epoch": 0.055235309300524824,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009997216655058573,
      "loss": 2.5029,
      "step": 642
    },
    {
      "epoch": 0.05532134560784651,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009997207626399975,
      "loss": 1.6266,
      "step": 643
    },
    {
      "epoch": 0.055407381915168204,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009997198583125513,
      "loss": 1.6881,
      "step": 644
    },
    {
      "epoch": 0.05549341822248989,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000999718952523522,
      "loss": 1.2328,
      "step": 645
    },
    {
      "epoch": 0.055579454529811584,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009997180452729116,
      "loss": 1.4657,
      "step": 646
    },
    {
      "epoch": 0.05566549083713327,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009997171365607235,
      "loss": 1.0475,
      "step": 647
    },
    {
      "epoch": 0.05575152714445496,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009997162263869596,
      "loss": 1.2289,
      "step": 648
    },
    {
      "epoch": 0.05583756345177665,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009997153147516232,
      "loss": 1.6107,
      "step": 649
    },
    {
      "epoch": 0.05592359975909834,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009997144016547163,
      "loss": 0.9986,
      "step": 650
    },
    {
      "epoch": 0.05600963606642003,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009997134870962424,
      "loss": 1.5847,
      "step": 651
    },
    {
      "epoch": 0.05609567237374172,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009997125710762035,
      "loss": 1.7388,
      "step": 652
    },
    {
      "epoch": 0.05618170868106341,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009997116535946026,
      "loss": 1.031,
      "step": 653
    },
    {
      "epoch": 0.0562677449883851,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009997107346514423,
      "loss": 0.8661,
      "step": 654
    },
    {
      "epoch": 0.05635378129570679,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009997098142467253,
      "loss": 1.6856,
      "step": 655
    },
    {
      "epoch": 0.05643981760302848,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009997088923804542,
      "loss": 1.4253,
      "step": 656
    },
    {
      "epoch": 0.05652585391035017,
      "grad_norm": 1.21875,
      "learning_rate": 0.000999707969052632,
      "loss": 1.1811,
      "step": 657
    },
    {
      "epoch": 0.05661189021767186,
      "grad_norm": 1.5,
      "learning_rate": 0.0009997070442632609,
      "loss": 1.4957,
      "step": 658
    },
    {
      "epoch": 0.056697926524993544,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009997061180123442,
      "loss": 1.3158,
      "step": 659
    },
    {
      "epoch": 0.05678396283231524,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000999705190299884,
      "loss": 1.08,
      "step": 660
    },
    {
      "epoch": 0.056869999139636924,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009997042611258832,
      "loss": 1.2654,
      "step": 661
    },
    {
      "epoch": 0.05695603544695862,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000999703330490345,
      "loss": 1.1426,
      "step": 662
    },
    {
      "epoch": 0.057042071754280305,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009997023983932716,
      "loss": 2.1574,
      "step": 663
    },
    {
      "epoch": 0.057128108061602,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009997014648346657,
      "loss": 1.1992,
      "step": 664
    },
    {
      "epoch": 0.057214144368923685,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009997005298145304,
      "loss": 1.1581,
      "step": 665
    },
    {
      "epoch": 0.05730018067624538,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009996995933328682,
      "loss": 1.2767,
      "step": 666
    },
    {
      "epoch": 0.057386216983567065,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009996986553896818,
      "loss": 1.8039,
      "step": 667
    },
    {
      "epoch": 0.05747225329088876,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009996977159849743,
      "loss": 2.0601,
      "step": 668
    },
    {
      "epoch": 0.057558289598210445,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009996967751187478,
      "loss": 1.581,
      "step": 669
    },
    {
      "epoch": 0.05764432590553213,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009996958327910055,
      "loss": 1.7246,
      "step": 670
    },
    {
      "epoch": 0.057730362212853825,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009996948890017502,
      "loss": 1.2949,
      "step": 671
    },
    {
      "epoch": 0.05781639852017551,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009996939437509843,
      "loss": 1.2392,
      "step": 672
    },
    {
      "epoch": 0.057902434827497205,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009996929970387113,
      "loss": 1.1228,
      "step": 673
    },
    {
      "epoch": 0.05798847113481889,
      "grad_norm": 1.125,
      "learning_rate": 0.000999692048864933,
      "loss": 1.304,
      "step": 674
    },
    {
      "epoch": 0.058074507442140585,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009996910992296527,
      "loss": 1.6715,
      "step": 675
    },
    {
      "epoch": 0.05816054374946227,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009996901481328733,
      "loss": 1.3348,
      "step": 676
    },
    {
      "epoch": 0.058246580056783966,
      "grad_norm": 1.25,
      "learning_rate": 0.0009996891955745973,
      "loss": 1.4193,
      "step": 677
    },
    {
      "epoch": 0.05833261636410565,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009996882415548277,
      "loss": 2.0071,
      "step": 678
    },
    {
      "epoch": 0.058418652671427346,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000999687286073567,
      "loss": 1.4826,
      "step": 679
    },
    {
      "epoch": 0.05850468897874903,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009996863291308181,
      "loss": 1.4335,
      "step": 680
    },
    {
      "epoch": 0.05859072528607072,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009996853707265841,
      "loss": 1.5888,
      "step": 681
    },
    {
      "epoch": 0.05867676159339241,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009996844108608675,
      "loss": 1.1651,
      "step": 682
    },
    {
      "epoch": 0.0587627979007141,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009996834495336713,
      "loss": 1.6376,
      "step": 683
    },
    {
      "epoch": 0.05884883420803579,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009996824867449978,
      "loss": 0.9584,
      "step": 684
    },
    {
      "epoch": 0.05893487051535748,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009996815224948506,
      "loss": 1.3746,
      "step": 685
    },
    {
      "epoch": 0.05902090682267917,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000999680556783232,
      "loss": 2.1964,
      "step": 686
    },
    {
      "epoch": 0.05910694313000086,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009996795896101448,
      "loss": 0.8353,
      "step": 687
    },
    {
      "epoch": 0.05919297943732255,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009996786209755923,
      "loss": 1.5234,
      "step": 688
    },
    {
      "epoch": 0.05927901574464424,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009996776508795768,
      "loss": 1.8202,
      "step": 689
    },
    {
      "epoch": 0.05936505205196593,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009996766793221015,
      "loss": 1.2062,
      "step": 690
    },
    {
      "epoch": 0.05945108835928762,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000999675706303169,
      "loss": 0.9313,
      "step": 691
    },
    {
      "epoch": 0.059537124666609306,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009996747318227822,
      "loss": 2.1021,
      "step": 692
    },
    {
      "epoch": 0.059623160973931,
      "grad_norm": 1.078125,
      "learning_rate": 0.000999673755880944,
      "loss": 1.1986,
      "step": 693
    },
    {
      "epoch": 0.059709197281252686,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009996727784776572,
      "loss": 2.2209,
      "step": 694
    },
    {
      "epoch": 0.05979523358857438,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009996717996129249,
      "loss": 1.7215,
      "step": 695
    },
    {
      "epoch": 0.059881269895896067,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009996708192867496,
      "loss": 1.2345,
      "step": 696
    },
    {
      "epoch": 0.05996730620321776,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009996698374991344,
      "loss": 1.1947,
      "step": 697
    },
    {
      "epoch": 0.06005334251053945,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009996688542500821,
      "loss": 1.2999,
      "step": 698
    },
    {
      "epoch": 0.06013937881786114,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009996678695395956,
      "loss": 1.19,
      "step": 699
    },
    {
      "epoch": 0.06022541512518283,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009996668833676776,
      "loss": 1.0558,
      "step": 700
    },
    {
      "epoch": 0.06031145143250452,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009996658957343314,
      "loss": 1.8247,
      "step": 701
    },
    {
      "epoch": 0.06039748773982621,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009996649066395594,
      "loss": 1.1786,
      "step": 702
    },
    {
      "epoch": 0.060483524047147894,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000999663916083365,
      "loss": 1.7792,
      "step": 703
    },
    {
      "epoch": 0.06056956035446959,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009996629240657505,
      "loss": 1.4502,
      "step": 704
    },
    {
      "epoch": 0.060655596661791274,
      "grad_norm": 1.25,
      "learning_rate": 0.0009996619305867192,
      "loss": 1.5608,
      "step": 705
    },
    {
      "epoch": 0.06074163296911297,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000999660935646274,
      "loss": 1.0933,
      "step": 706
    },
    {
      "epoch": 0.060827669276434654,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009996599392444179,
      "loss": 1.6073,
      "step": 707
    },
    {
      "epoch": 0.06091370558375635,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009996589413811534,
      "loss": 1.9188,
      "step": 708
    },
    {
      "epoch": 0.060999741891078034,
      "grad_norm": 1.03125,
      "learning_rate": 0.000999657942056484,
      "loss": 1.4953,
      "step": 709
    },
    {
      "epoch": 0.06108577819839973,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000999656941270412,
      "loss": 1.6678,
      "step": 710
    },
    {
      "epoch": 0.061171814505721414,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009996559390229407,
      "loss": 0.9957,
      "step": 711
    },
    {
      "epoch": 0.06125785081304311,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000999654935314073,
      "loss": 1.2067,
      "step": 712
    },
    {
      "epoch": 0.061343887120364794,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009996539301438117,
      "loss": 0.8776,
      "step": 713
    },
    {
      "epoch": 0.06142992342768648,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00099965292351216,
      "loss": 0.5732,
      "step": 714
    },
    {
      "epoch": 0.061515959735008174,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009996519154191206,
      "loss": 1.9076,
      "step": 715
    },
    {
      "epoch": 0.06160199604232986,
      "grad_norm": 1.25,
      "learning_rate": 0.0009996509058646963,
      "loss": 1.4852,
      "step": 716
    },
    {
      "epoch": 0.061688032349651555,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009996498948488905,
      "loss": 2.3554,
      "step": 717
    },
    {
      "epoch": 0.06177406865697324,
      "grad_norm": 1.59375,
      "learning_rate": 0.000999648882371706,
      "loss": 0.7917,
      "step": 718
    },
    {
      "epoch": 0.061860104964294935,
      "grad_norm": 1.125,
      "learning_rate": 0.0009996478684331455,
      "loss": 1.8293,
      "step": 719
    },
    {
      "epoch": 0.06194614127161662,
      "grad_norm": 2.71875,
      "learning_rate": 0.0009996468530332123,
      "loss": 1.2278,
      "step": 720
    },
    {
      "epoch": 0.062032177578938315,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009996458361719092,
      "loss": 1.5867,
      "step": 721
    },
    {
      "epoch": 0.06211821388626,
      "grad_norm": 1.25,
      "learning_rate": 0.0009996448178492391,
      "loss": 1.0192,
      "step": 722
    },
    {
      "epoch": 0.062204250193581695,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009996437980652054,
      "loss": 1.4318,
      "step": 723
    },
    {
      "epoch": 0.06229028650090338,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009996427768198103,
      "loss": 1.4031,
      "step": 724
    },
    {
      "epoch": 0.06237632280822507,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009996417541130575,
      "loss": 1.4178,
      "step": 725
    },
    {
      "epoch": 0.06246235911554676,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00099964072994495,
      "loss": 2.0937,
      "step": 726
    },
    {
      "epoch": 0.06254839542286846,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009996397043154901,
      "loss": 1.5962,
      "step": 727
    },
    {
      "epoch": 0.06263443173019013,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009996386772246815,
      "loss": 0.751,
      "step": 728
    },
    {
      "epoch": 0.06272046803751183,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009996376486725268,
      "loss": 1.7615,
      "step": 729
    },
    {
      "epoch": 0.06280650434483352,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009996366186590293,
      "loss": 0.5828,
      "step": 730
    },
    {
      "epoch": 0.06289254065215522,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009996355871841917,
      "loss": 1.1073,
      "step": 731
    },
    {
      "epoch": 0.0629785769594769,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009996345542480174,
      "loss": 1.2336,
      "step": 732
    },
    {
      "epoch": 0.06306461326679859,
      "grad_norm": 1.140625,
      "learning_rate": 0.000999633519850509,
      "loss": 1.4487,
      "step": 733
    },
    {
      "epoch": 0.06315064957412028,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009996324839916697,
      "loss": 1.9888,
      "step": 734
    },
    {
      "epoch": 0.06323668588144196,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009996314466715027,
      "loss": 1.4384,
      "step": 735
    },
    {
      "epoch": 0.06332272218876366,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009996304078900109,
      "loss": 1.5913,
      "step": 736
    },
    {
      "epoch": 0.06340875849608535,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009996293676471973,
      "loss": 1.433,
      "step": 737
    },
    {
      "epoch": 0.06349479480340704,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009996283259430649,
      "loss": 1.1002,
      "step": 738
    },
    {
      "epoch": 0.06358083111072872,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009996272827776168,
      "loss": 2.0525,
      "step": 739
    },
    {
      "epoch": 0.06366686741805042,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009996262381508562,
      "loss": 1.4625,
      "step": 740
    },
    {
      "epoch": 0.06375290372537211,
      "grad_norm": 1.203125,
      "learning_rate": 0.000999625192062786,
      "loss": 1.4951,
      "step": 741
    },
    {
      "epoch": 0.0638389400326938,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000999624144513409,
      "loss": 1.3684,
      "step": 742
    },
    {
      "epoch": 0.06392497634001548,
      "grad_norm": 1.484375,
      "learning_rate": 0.000999623095502729,
      "loss": 2.0261,
      "step": 743
    },
    {
      "epoch": 0.06401101264733718,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009996220450307484,
      "loss": 1.2404,
      "step": 744
    },
    {
      "epoch": 0.06409704895465887,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009996209930974705,
      "loss": 1.4976,
      "step": 745
    },
    {
      "epoch": 0.06418308526198055,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009996199397028984,
      "loss": 1.1305,
      "step": 746
    },
    {
      "epoch": 0.06426912156930224,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009996188848470351,
      "loss": 1.5321,
      "step": 747
    },
    {
      "epoch": 0.06435515787662394,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009996178285298838,
      "loss": 1.9261,
      "step": 748
    },
    {
      "epoch": 0.06444119418394563,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009996167707514475,
      "loss": 1.9444,
      "step": 749
    },
    {
      "epoch": 0.06452723049126731,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009996157115117292,
      "loss": 0.9735,
      "step": 750
    },
    {
      "epoch": 0.064613266798589,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009996146508107321,
      "loss": 1.2614,
      "step": 751
    },
    {
      "epoch": 0.0646993031059107,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009996135886484595,
      "loss": 1.597,
      "step": 752
    },
    {
      "epoch": 0.06478533941323239,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009996125250249142,
      "loss": 1.0541,
      "step": 753
    },
    {
      "epoch": 0.06487137572055407,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009996114599400993,
      "loss": 1.4898,
      "step": 754
    },
    {
      "epoch": 0.06495741202787576,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009996103933940185,
      "loss": 0.8898,
      "step": 755
    },
    {
      "epoch": 0.06504344833519746,
      "grad_norm": 1.0625,
      "learning_rate": 0.000999609325386674,
      "loss": 0.8501,
      "step": 756
    },
    {
      "epoch": 0.06512948464251914,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009996082559180696,
      "loss": 1.3805,
      "step": 757
    },
    {
      "epoch": 0.06521552094984083,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009996071849882081,
      "loss": 1.1673,
      "step": 758
    },
    {
      "epoch": 0.06530155725716252,
      "grad_norm": 1.9375,
      "learning_rate": 0.000999606112597093,
      "loss": 1.7611,
      "step": 759
    },
    {
      "epoch": 0.06538759356448422,
      "grad_norm": 2.03125,
      "learning_rate": 0.000999605038744727,
      "loss": 1.1027,
      "step": 760
    },
    {
      "epoch": 0.0654736298718059,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009996039634311135,
      "loss": 1.571,
      "step": 761
    },
    {
      "epoch": 0.06555966617912759,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009996028866562554,
      "loss": 1.4123,
      "step": 762
    },
    {
      "epoch": 0.06564570248644928,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009996018084201563,
      "loss": 0.8427,
      "step": 763
    },
    {
      "epoch": 0.06573173879377098,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009996007287228189,
      "loss": 1.5472,
      "step": 764
    },
    {
      "epoch": 0.06581777510109266,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009995996475642467,
      "loss": 1.395,
      "step": 765
    },
    {
      "epoch": 0.06590381140841435,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009995985649444426,
      "loss": 0.8315,
      "step": 766
    },
    {
      "epoch": 0.06598984771573604,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009995974808634097,
      "loss": 1.3021,
      "step": 767
    },
    {
      "epoch": 0.06607588402305772,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009995963953211516,
      "loss": 1.6957,
      "step": 768
    },
    {
      "epoch": 0.06616192033037942,
      "grad_norm": 1.40625,
      "learning_rate": 0.000999595308317671,
      "loss": 1.7035,
      "step": 769
    },
    {
      "epoch": 0.06624795663770111,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009995942198529712,
      "loss": 1.0236,
      "step": 770
    },
    {
      "epoch": 0.0663339929450228,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009995931299270557,
      "loss": 1.7149,
      "step": 771
    },
    {
      "epoch": 0.06642002925234448,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009995920385399274,
      "loss": 1.0391,
      "step": 772
    },
    {
      "epoch": 0.06650606555966618,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009995909456915898,
      "loss": 0.8626,
      "step": 773
    },
    {
      "epoch": 0.06659210186698787,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009995898513820453,
      "loss": 1.5906,
      "step": 774
    },
    {
      "epoch": 0.06667813817430956,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009995887556112982,
      "loss": 1.678,
      "step": 775
    },
    {
      "epoch": 0.06676417448163124,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009995876583793507,
      "loss": 2.0229,
      "step": 776
    },
    {
      "epoch": 0.06685021078895294,
      "grad_norm": 1.15625,
      "learning_rate": 0.000999586559686207,
      "loss": 2.1609,
      "step": 777
    },
    {
      "epoch": 0.06693624709627463,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009995854595318693,
      "loss": 1.9789,
      "step": 778
    },
    {
      "epoch": 0.06702228340359631,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009995843579163413,
      "loss": 2.0005,
      "step": 779
    },
    {
      "epoch": 0.067108319710918,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009995832548396265,
      "loss": 1.8079,
      "step": 780
    },
    {
      "epoch": 0.0671943560182397,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009995821503017278,
      "loss": 1.9597,
      "step": 781
    },
    {
      "epoch": 0.06728039232556139,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009995810443026484,
      "loss": 1.4231,
      "step": 782
    },
    {
      "epoch": 0.06736642863288307,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009995799368423914,
      "loss": 2.1903,
      "step": 783
    },
    {
      "epoch": 0.06745246494020476,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009995788279209604,
      "loss": 0.9969,
      "step": 784
    },
    {
      "epoch": 0.06753850124752646,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009995777175383586,
      "loss": 1.6604,
      "step": 785
    },
    {
      "epoch": 0.06762453755484815,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009995766056945892,
      "loss": 1.529,
      "step": 786
    },
    {
      "epoch": 0.06771057386216983,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009995754923896553,
      "loss": 1.4753,
      "step": 787
    },
    {
      "epoch": 0.06779661016949153,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00099957437762356,
      "loss": 1.6354,
      "step": 788
    },
    {
      "epoch": 0.06788264647681322,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000999573261396307,
      "loss": 0.8292,
      "step": 789
    },
    {
      "epoch": 0.0679686827841349,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009995721437078993,
      "loss": 1.3728,
      "step": 790
    },
    {
      "epoch": 0.06805471909145659,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009995710245583405,
      "loss": 0.9201,
      "step": 791
    },
    {
      "epoch": 0.06814075539877829,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009995699039476334,
      "loss": 1.0834,
      "step": 792
    },
    {
      "epoch": 0.06822679170609998,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009995687818757815,
      "loss": 1.8728,
      "step": 793
    },
    {
      "epoch": 0.06831282801342166,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000999567658342788,
      "loss": 0.6825,
      "step": 794
    },
    {
      "epoch": 0.06839886432074335,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009995665333486566,
      "loss": 1.374,
      "step": 795
    },
    {
      "epoch": 0.06848490062806505,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009995654068933899,
      "loss": 1.4088,
      "step": 796
    },
    {
      "epoch": 0.06857093693538674,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009995642789769916,
      "loss": 1.7076,
      "step": 797
    },
    {
      "epoch": 0.06865697324270842,
      "grad_norm": 1.671875,
      "learning_rate": 0.000999563149599465,
      "loss": 1.5029,
      "step": 798
    },
    {
      "epoch": 0.06874300955003011,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009995620187608135,
      "loss": 1.755,
      "step": 799
    },
    {
      "epoch": 0.0688290458573518,
      "grad_norm": 1.078125,
      "learning_rate": 0.00099956088646104,
      "loss": 1.1767,
      "step": 800
    },
    {
      "epoch": 0.06891508216467349,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000999559752700148,
      "loss": 1.2999,
      "step": 801
    },
    {
      "epoch": 0.06900111847199518,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009995586174781413,
      "loss": 2.2059,
      "step": 802
    },
    {
      "epoch": 0.06908715477931687,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009995574807950226,
      "loss": 1.6738,
      "step": 803
    },
    {
      "epoch": 0.06917319108663857,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009995563426507953,
      "loss": 1.0,
      "step": 804
    },
    {
      "epoch": 0.06925922739396025,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009995552030454629,
      "loss": 1.4213,
      "step": 805
    },
    {
      "epoch": 0.06934526370128194,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009995540619790288,
      "loss": 1.5295,
      "step": 806
    },
    {
      "epoch": 0.06943130000860363,
      "grad_norm": 1.453125,
      "learning_rate": 0.000999552919451496,
      "loss": 1.296,
      "step": 807
    },
    {
      "epoch": 0.06951733631592533,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009995517754628682,
      "loss": 2.0282,
      "step": 808
    },
    {
      "epoch": 0.069603372623247,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009995506300131487,
      "loss": 1.7496,
      "step": 809
    },
    {
      "epoch": 0.0696894089305687,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009995494831023409,
      "loss": 2.0267,
      "step": 810
    },
    {
      "epoch": 0.0697754452378904,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009995483347304476,
      "loss": 1.053,
      "step": 811
    },
    {
      "epoch": 0.06986148154521207,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000999547184897473,
      "loss": 1.0393,
      "step": 812
    },
    {
      "epoch": 0.06994751785253377,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009995460336034198,
      "loss": 1.6189,
      "step": 813
    },
    {
      "epoch": 0.07003355415985546,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009995448808482914,
      "loss": 1.9675,
      "step": 814
    },
    {
      "epoch": 0.07011959046717715,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009995437266320918,
      "loss": 1.8698,
      "step": 815
    },
    {
      "epoch": 0.07020562677449883,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009995425709548237,
      "loss": 1.0312,
      "step": 816
    },
    {
      "epoch": 0.07029166308182053,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000999541413816491,
      "loss": 1.9368,
      "step": 817
    },
    {
      "epoch": 0.07037769938914222,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009995402552170965,
      "loss": 0.6957,
      "step": 818
    },
    {
      "epoch": 0.07046373569646391,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000999539095156644,
      "loss": 1.8824,
      "step": 819
    },
    {
      "epoch": 0.0705497720037856,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009995379336351369,
      "loss": 1.323,
      "step": 820
    },
    {
      "epoch": 0.07063580831110729,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009995367706525785,
      "loss": 1.2446,
      "step": 821
    },
    {
      "epoch": 0.07072184461842898,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000999535606208972,
      "loss": 2.0294,
      "step": 822
    },
    {
      "epoch": 0.07080788092575066,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009995344403043213,
      "loss": 2.3025,
      "step": 823
    },
    {
      "epoch": 0.07089391723307235,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009995332729386293,
      "loss": 0.9564,
      "step": 824
    },
    {
      "epoch": 0.07097995354039405,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009995321041118997,
      "loss": 1.1415,
      "step": 825
    },
    {
      "epoch": 0.07106598984771574,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0009995309338241358,
      "loss": 0.9281,
      "step": 826
    },
    {
      "epoch": 0.07115202615503742,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009995297620753413,
      "loss": 0.9061,
      "step": 827
    },
    {
      "epoch": 0.07123806246235911,
      "grad_norm": 1.359375,
      "learning_rate": 0.000999528588865519,
      "loss": 2.1232,
      "step": 828
    },
    {
      "epoch": 0.07132409876968081,
      "grad_norm": 1.625,
      "learning_rate": 0.000999527414194673,
      "loss": 1.1137,
      "step": 829
    },
    {
      "epoch": 0.0714101350770025,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009995262380628066,
      "loss": 1.0041,
      "step": 830
    },
    {
      "epoch": 0.07149617138432418,
      "grad_norm": 1.109375,
      "learning_rate": 0.000999525060469923,
      "loss": 1.324,
      "step": 831
    },
    {
      "epoch": 0.07158220769164587,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009995238814160254,
      "loss": 1.1215,
      "step": 832
    },
    {
      "epoch": 0.07166824399896757,
      "grad_norm": 1.9140625,
      "learning_rate": 0.000999522700901118,
      "loss": 1.2129,
      "step": 833
    },
    {
      "epoch": 0.07175428030628925,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009995215189252038,
      "loss": 1.7935,
      "step": 834
    },
    {
      "epoch": 0.07184031661361094,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000999520335488286,
      "loss": 1.389,
      "step": 835
    },
    {
      "epoch": 0.07192635292093263,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009995191505903687,
      "loss": 1.3805,
      "step": 836
    },
    {
      "epoch": 0.07201238922825433,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000999517964231455,
      "loss": 1.4168,
      "step": 837
    },
    {
      "epoch": 0.07209842553557601,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009995167764115483,
      "loss": 1.2999,
      "step": 838
    },
    {
      "epoch": 0.0721844618428977,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009995155871306523,
      "loss": 1.6139,
      "step": 839
    },
    {
      "epoch": 0.0722704981502194,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009995143963887703,
      "loss": 0.7936,
      "step": 840
    },
    {
      "epoch": 0.07235653445754109,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009995132041859056,
      "loss": 2.4072,
      "step": 841
    },
    {
      "epoch": 0.07244257076486277,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009995120105220623,
      "loss": 1.9033,
      "step": 842
    },
    {
      "epoch": 0.07252860707218446,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009995108153972433,
      "loss": 1.4536,
      "step": 843
    },
    {
      "epoch": 0.07261464337950616,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009995096188114526,
      "loss": 0.9597,
      "step": 844
    },
    {
      "epoch": 0.07270067968682783,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009995084207646931,
      "loss": 1.0695,
      "step": 845
    },
    {
      "epoch": 0.07278671599414953,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009995072212569686,
      "loss": 1.3454,
      "step": 846
    },
    {
      "epoch": 0.07287275230147122,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009995060202882828,
      "loss": 1.1067,
      "step": 847
    },
    {
      "epoch": 0.07295878860879292,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000999504817858639,
      "loss": 2.0296,
      "step": 848
    },
    {
      "epoch": 0.0730448249161146,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009995036139680406,
      "loss": 0.7215,
      "step": 849
    },
    {
      "epoch": 0.07313086122343629,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009995024086164915,
      "loss": 1.1828,
      "step": 850
    },
    {
      "epoch": 0.07321689753075798,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000999501201803995,
      "loss": 1.2868,
      "step": 851
    },
    {
      "epoch": 0.07330293383807968,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009994999935305545,
      "loss": 1.6452,
      "step": 852
    },
    {
      "epoch": 0.07338897014540136,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009994987837961737,
      "loss": 1.3328,
      "step": 853
    },
    {
      "epoch": 0.07347500645272305,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000999497572600856,
      "loss": 1.6789,
      "step": 854
    },
    {
      "epoch": 0.07356104276004474,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009994963599446053,
      "loss": 1.4832,
      "step": 855
    },
    {
      "epoch": 0.07364707906736642,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009994951458274248,
      "loss": 1.1184,
      "step": 856
    },
    {
      "epoch": 0.07373311537468812,
      "grad_norm": 1.21875,
      "learning_rate": 0.000999493930249318,
      "loss": 1.304,
      "step": 857
    },
    {
      "epoch": 0.07381915168200981,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009994927132102888,
      "loss": 1.126,
      "step": 858
    },
    {
      "epoch": 0.0739051879893315,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009994914947103403,
      "loss": 1.517,
      "step": 859
    },
    {
      "epoch": 0.07399122429665318,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009994902747494767,
      "loss": 0.9413,
      "step": 860
    },
    {
      "epoch": 0.07407726060397488,
      "grad_norm": 1.328125,
      "learning_rate": 0.000999489053327701,
      "loss": 1.3427,
      "step": 861
    },
    {
      "epoch": 0.07416329691129657,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009994878304450169,
      "loss": 1.6415,
      "step": 862
    },
    {
      "epoch": 0.07424933321861826,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009994866061014281,
      "loss": 1.4011,
      "step": 863
    },
    {
      "epoch": 0.07433536952593994,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000999485380296938,
      "loss": 1.0199,
      "step": 864
    },
    {
      "epoch": 0.07442140583326164,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009994841530315505,
      "loss": 1.7306,
      "step": 865
    },
    {
      "epoch": 0.07450744214058333,
      "grad_norm": 1.078125,
      "learning_rate": 0.000999482924305269,
      "loss": 1.3715,
      "step": 866
    },
    {
      "epoch": 0.07459347844790501,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000999481694118097,
      "loss": 0.9726,
      "step": 867
    },
    {
      "epoch": 0.0746795147552267,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000999480462470038,
      "loss": 1.0565,
      "step": 868
    },
    {
      "epoch": 0.0747655510625484,
      "grad_norm": 1.609375,
      "learning_rate": 0.000999479229361096,
      "loss": 2.2593,
      "step": 869
    },
    {
      "epoch": 0.07485158736987009,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009994779947912744,
      "loss": 1.5561,
      "step": 870
    },
    {
      "epoch": 0.07493762367719177,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000999476758760577,
      "loss": 2.0877,
      "step": 871
    },
    {
      "epoch": 0.07502365998451346,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009994755212690068,
      "loss": 2.5642,
      "step": 872
    },
    {
      "epoch": 0.07510969629183516,
      "grad_norm": 1.21875,
      "learning_rate": 0.000999474282316568,
      "loss": 1.7411,
      "step": 873
    },
    {
      "epoch": 0.07519573259915685,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009994730419032644,
      "loss": 1.48,
      "step": 874
    },
    {
      "epoch": 0.07528176890647853,
      "grad_norm": 1.125,
      "learning_rate": 0.0009994718000290988,
      "loss": 1.3182,
      "step": 875
    },
    {
      "epoch": 0.07536780521380022,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009994705566940755,
      "loss": 1.8813,
      "step": 876
    },
    {
      "epoch": 0.07545384152112192,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000999469311898198,
      "loss": 1.3051,
      "step": 877
    },
    {
      "epoch": 0.0755398778284436,
      "grad_norm": 1.203125,
      "learning_rate": 0.00099946806564147,
      "loss": 1.7164,
      "step": 878
    },
    {
      "epoch": 0.07562591413576529,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009994668179238951,
      "loss": 1.6506,
      "step": 879
    },
    {
      "epoch": 0.07571195044308698,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009994655687454767,
      "loss": 1.8648,
      "step": 880
    },
    {
      "epoch": 0.07579798675040868,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009994643181062187,
      "loss": 0.9388,
      "step": 881
    },
    {
      "epoch": 0.07588402305773036,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009994630660061247,
      "loss": 1.3133,
      "step": 882
    },
    {
      "epoch": 0.07597005936505205,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009994618124451985,
      "loss": 1.367,
      "step": 883
    },
    {
      "epoch": 0.07605609567237374,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009994605574234435,
      "loss": 1.7981,
      "step": 884
    },
    {
      "epoch": 0.07614213197969544,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009994593009408634,
      "loss": 1.2284,
      "step": 885
    },
    {
      "epoch": 0.07622816828701712,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009994580429974623,
      "loss": 1.8818,
      "step": 886
    },
    {
      "epoch": 0.07631420459433881,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0009994567835932434,
      "loss": 1.3167,
      "step": 887
    },
    {
      "epoch": 0.0764002409016605,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009994555227282103,
      "loss": 1.067,
      "step": 888
    },
    {
      "epoch": 0.07648627720898218,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009994542604023673,
      "loss": 1.0219,
      "step": 889
    },
    {
      "epoch": 0.07657231351630388,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009994529966157175,
      "loss": 1.4934,
      "step": 890
    },
    {
      "epoch": 0.07665834982362557,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000999451731368265,
      "loss": 0.9402,
      "step": 891
    },
    {
      "epoch": 0.07674438613094726,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009994504646600132,
      "loss": 2.2551,
      "step": 892
    },
    {
      "epoch": 0.07683042243826894,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000999449196490966,
      "loss": 1.2819,
      "step": 893
    },
    {
      "epoch": 0.07691645874559064,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000999447926861127,
      "loss": 1.3676,
      "step": 894
    },
    {
      "epoch": 0.07700249505291233,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009994466557705,
      "loss": 0.9515,
      "step": 895
    },
    {
      "epoch": 0.07708853136023402,
      "grad_norm": 1.125,
      "learning_rate": 0.0009994453832190887,
      "loss": 2.0929,
      "step": 896
    },
    {
      "epoch": 0.0771745676675557,
      "grad_norm": 1.375,
      "learning_rate": 0.0009994441092068967,
      "loss": 1.5564,
      "step": 897
    },
    {
      "epoch": 0.0772606039748774,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009994428337339277,
      "loss": 1.5977,
      "step": 898
    },
    {
      "epoch": 0.07734664028219909,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009994415568001858,
      "loss": 1.2479,
      "step": 899
    },
    {
      "epoch": 0.07743267658952077,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009994402784056743,
      "loss": 1.2382,
      "step": 900
    },
    {
      "epoch": 0.07751871289684246,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009994389985503972,
      "loss": 1.0146,
      "step": 901
    },
    {
      "epoch": 0.07760474920416416,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009994377172343582,
      "loss": 1.5602,
      "step": 902
    },
    {
      "epoch": 0.07769078551148585,
      "grad_norm": 1.5,
      "learning_rate": 0.0009994364344575609,
      "loss": 1.8592,
      "step": 903
    },
    {
      "epoch": 0.07777682181880753,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009994351502200093,
      "loss": 1.1969,
      "step": 904
    },
    {
      "epoch": 0.07786285812612923,
      "grad_norm": 1.46875,
      "learning_rate": 0.000999433864521707,
      "loss": 1.308,
      "step": 905
    },
    {
      "epoch": 0.07794889443345092,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009994325773626577,
      "loss": 1.6769,
      "step": 906
    },
    {
      "epoch": 0.07803493074077261,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009994312887428652,
      "loss": 1.1645,
      "step": 907
    },
    {
      "epoch": 0.07812096704809429,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009994299986623334,
      "loss": 1.4345,
      "step": 908
    },
    {
      "epoch": 0.07820700335541599,
      "grad_norm": 1.109375,
      "learning_rate": 0.000999428707121066,
      "loss": 1.1336,
      "step": 909
    },
    {
      "epoch": 0.07829303966273768,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009994274141190667,
      "loss": 1.2081,
      "step": 910
    },
    {
      "epoch": 0.07837907597005936,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009994261196563394,
      "loss": 1.6291,
      "step": 911
    },
    {
      "epoch": 0.07846511227738105,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009994248237328877,
      "loss": 1.9511,
      "step": 912
    },
    {
      "epoch": 0.07855114858470275,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009994235263487158,
      "loss": 1.3583,
      "step": 913
    },
    {
      "epoch": 0.07863718489202444,
      "grad_norm": 1.3125,
      "learning_rate": 0.000999422227503827,
      "loss": 1.4803,
      "step": 914
    },
    {
      "epoch": 0.07872322119934612,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009994209271982253,
      "loss": 0.9143,
      "step": 915
    },
    {
      "epoch": 0.07880925750666781,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009994196254319147,
      "loss": 1.2897,
      "step": 916
    },
    {
      "epoch": 0.0788952938139895,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009994183222048988,
      "loss": 1.7653,
      "step": 917
    },
    {
      "epoch": 0.0789813301213112,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009994170175171813,
      "loss": 1.7024,
      "step": 918
    },
    {
      "epoch": 0.07906736642863288,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009994157113687662,
      "loss": 1.4116,
      "step": 919
    },
    {
      "epoch": 0.07915340273595457,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009994144037596573,
      "loss": 1.3409,
      "step": 920
    },
    {
      "epoch": 0.07923943904327627,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009994130946898586,
      "loss": 1.2705,
      "step": 921
    },
    {
      "epoch": 0.07932547535059795,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009994117841593735,
      "loss": 1.1567,
      "step": 922
    },
    {
      "epoch": 0.07941151165791964,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000999410472168206,
      "loss": 1.2691,
      "step": 923
    },
    {
      "epoch": 0.07949754796524133,
      "grad_norm": 1.078125,
      "learning_rate": 0.00099940915871636,
      "loss": 1.2723,
      "step": 924
    },
    {
      "epoch": 0.07958358427256303,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009994078438038394,
      "loss": 2.4342,
      "step": 925
    },
    {
      "epoch": 0.0796696205798847,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000999406527430648,
      "loss": 1.5239,
      "step": 926
    },
    {
      "epoch": 0.0797556568872064,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009994052095967899,
      "loss": 1.7015,
      "step": 927
    },
    {
      "epoch": 0.0798416931945281,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009994038903022684,
      "loss": 1.3381,
      "step": 928
    },
    {
      "epoch": 0.07992772950184979,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009994025695470876,
      "loss": 2.1086,
      "step": 929
    },
    {
      "epoch": 0.08001376580917147,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009994012473312514,
      "loss": 1.6407,
      "step": 930
    },
    {
      "epoch": 0.08009980211649316,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009993999236547638,
      "loss": 2.0897,
      "step": 931
    },
    {
      "epoch": 0.08018583842381485,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009993985985176286,
      "loss": 1.3286,
      "step": 932
    },
    {
      "epoch": 0.08027187473113653,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009993972719198497,
      "loss": 1.6646,
      "step": 933
    },
    {
      "epoch": 0.08035791103845823,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009993959438614306,
      "loss": 0.9682,
      "step": 934
    },
    {
      "epoch": 0.08044394734577992,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009993946143423758,
      "loss": 1.1049,
      "step": 935
    },
    {
      "epoch": 0.08052998365310161,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009993932833626887,
      "loss": 1.772,
      "step": 936
    },
    {
      "epoch": 0.0806160199604233,
      "grad_norm": 1.125,
      "learning_rate": 0.0009993919509223733,
      "loss": 1.3328,
      "step": 937
    },
    {
      "epoch": 0.08070205626774499,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009993906170214337,
      "loss": 0.9999,
      "step": 938
    },
    {
      "epoch": 0.08078809257506668,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009993892816598736,
      "loss": 1.166,
      "step": 939
    },
    {
      "epoch": 0.08087412888238837,
      "grad_norm": 1.34375,
      "learning_rate": 0.000999387944837697,
      "loss": 1.8353,
      "step": 940
    },
    {
      "epoch": 0.08096016518971005,
      "grad_norm": 1.28125,
      "learning_rate": 0.000999386606554908,
      "loss": 1.1359,
      "step": 941
    },
    {
      "epoch": 0.08104620149703175,
      "grad_norm": 1.359375,
      "learning_rate": 0.00099938526681151,
      "loss": 1.5986,
      "step": 942
    },
    {
      "epoch": 0.08113223780435344,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009993839256075075,
      "loss": 0.9468,
      "step": 943
    },
    {
      "epoch": 0.08121827411167512,
      "grad_norm": 1.625,
      "learning_rate": 0.0009993825829429038,
      "loss": 2.2728,
      "step": 944
    },
    {
      "epoch": 0.08130431041899681,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009993812388177034,
      "loss": 1.508,
      "step": 945
    },
    {
      "epoch": 0.08139034672631851,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00099937989323191,
      "loss": 1.663,
      "step": 946
    },
    {
      "epoch": 0.0814763830336402,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009993785461855276,
      "loss": 2.0825,
      "step": 947
    },
    {
      "epoch": 0.08156241934096188,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009993771976785598,
      "loss": 1.621,
      "step": 948
    },
    {
      "epoch": 0.08164845564828357,
      "grad_norm": 1.234375,
      "learning_rate": 0.000999375847711011,
      "loss": 2.2044,
      "step": 949
    },
    {
      "epoch": 0.08173449195560527,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000999374496282885,
      "loss": 1.0157,
      "step": 950
    },
    {
      "epoch": 0.08182052826292696,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009993731433941856,
      "loss": 1.3123,
      "step": 951
    },
    {
      "epoch": 0.08190656457024864,
      "grad_norm": 1.140625,
      "learning_rate": 0.000999371789044917,
      "loss": 1.4202,
      "step": 952
    },
    {
      "epoch": 0.08199260087757033,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000999370433235083,
      "loss": 1.5596,
      "step": 953
    },
    {
      "epoch": 0.08207863718489203,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009993690759646876,
      "loss": 1.6,
      "step": 954
    },
    {
      "epoch": 0.08216467349221371,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009993677172337348,
      "loss": 1.498,
      "step": 955
    },
    {
      "epoch": 0.0822507097995354,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009993663570422285,
      "loss": 1.6523,
      "step": 956
    },
    {
      "epoch": 0.0823367461068571,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009993649953901726,
      "loss": 1.825,
      "step": 957
    },
    {
      "epoch": 0.08242278241417879,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009993636322775715,
      "loss": 1.3851,
      "step": 958
    },
    {
      "epoch": 0.08250881872150047,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009993622677044287,
      "loss": 1.0617,
      "step": 959
    },
    {
      "epoch": 0.08259485502882216,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009993609016707484,
      "loss": 1.5231,
      "step": 960
    },
    {
      "epoch": 0.08268089133614386,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009993595341765346,
      "loss": 1.9009,
      "step": 961
    },
    {
      "epoch": 0.08276692764346555,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009993581652217914,
      "loss": 1.3329,
      "step": 962
    },
    {
      "epoch": 0.08285296395078723,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009993567948065225,
      "loss": 1.7005,
      "step": 963
    },
    {
      "epoch": 0.08293900025810892,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009993554229307323,
      "loss": 1.4324,
      "step": 964
    },
    {
      "epoch": 0.08302503656543062,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009993540495944245,
      "loss": 1.5737,
      "step": 965
    },
    {
      "epoch": 0.0831110728727523,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009993526747976032,
      "loss": 0.9154,
      "step": 966
    },
    {
      "epoch": 0.08319710918007399,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009993512985402725,
      "loss": 0.9954,
      "step": 967
    },
    {
      "epoch": 0.08328314548739568,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009993499208224364,
      "loss": 1.5852,
      "step": 968
    },
    {
      "epoch": 0.08336918179471738,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009993485416440987,
      "loss": 1.9283,
      "step": 969
    },
    {
      "epoch": 0.08345521810203906,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009993471610052638,
      "loss": 1.7381,
      "step": 970
    },
    {
      "epoch": 0.08354125440936075,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009993457789059354,
      "loss": 1.4039,
      "step": 971
    },
    {
      "epoch": 0.08362729071668244,
      "grad_norm": 2.25,
      "learning_rate": 0.000999344395346118,
      "loss": 1.2116,
      "step": 972
    },
    {
      "epoch": 0.08371332702400414,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009993430103258153,
      "loss": 1.2318,
      "step": 973
    },
    {
      "epoch": 0.08379936333132582,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009993416238450314,
      "loss": 1.4863,
      "step": 974
    },
    {
      "epoch": 0.08388539963864751,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00099934023590377,
      "loss": 1.7587,
      "step": 975
    },
    {
      "epoch": 0.0839714359459692,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009993388465020358,
      "loss": 1.3179,
      "step": 976
    },
    {
      "epoch": 0.08405747225329088,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009993374556398326,
      "loss": 1.0765,
      "step": 977
    },
    {
      "epoch": 0.08414350856061258,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009993360633171646,
      "loss": 1.4496,
      "step": 978
    },
    {
      "epoch": 0.08422954486793427,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009993346695340356,
      "loss": 1.0913,
      "step": 979
    },
    {
      "epoch": 0.08431558117525596,
      "grad_norm": 2.4375,
      "learning_rate": 0.0009993332742904498,
      "loss": 2.2401,
      "step": 980
    },
    {
      "epoch": 0.08440161748257764,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009993318775864113,
      "loss": 2.3975,
      "step": 981
    },
    {
      "epoch": 0.08448765378989934,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009993304794219241,
      "loss": 1.2087,
      "step": 982
    },
    {
      "epoch": 0.08457369009722103,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009993290797969926,
      "loss": 1.2561,
      "step": 983
    },
    {
      "epoch": 0.08465972640454272,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009993276787116203,
      "loss": 1.7373,
      "step": 984
    },
    {
      "epoch": 0.0847457627118644,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000999326276165812,
      "loss": 1.5503,
      "step": 985
    },
    {
      "epoch": 0.0848317990191861,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009993248721595711,
      "loss": 1.6281,
      "step": 986
    },
    {
      "epoch": 0.08491783532650779,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009993234666929023,
      "loss": 1.1774,
      "step": 987
    },
    {
      "epoch": 0.08500387163382947,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009993220597658095,
      "loss": 2.0992,
      "step": 988
    },
    {
      "epoch": 0.08508990794115116,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009993206513782967,
      "loss": 1.5806,
      "step": 989
    },
    {
      "epoch": 0.08517594424847286,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009993192415303682,
      "loss": 1.3281,
      "step": 990
    },
    {
      "epoch": 0.08526198055579455,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009993178302220278,
      "loss": 1.6542,
      "step": 991
    },
    {
      "epoch": 0.08534801686311623,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009993164174532802,
      "loss": 1.5697,
      "step": 992
    },
    {
      "epoch": 0.08543405317043792,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009993150032241288,
      "loss": 0.8929,
      "step": 993
    },
    {
      "epoch": 0.08552008947775962,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009993135875345784,
      "loss": 1.1081,
      "step": 994
    },
    {
      "epoch": 0.08560612578508131,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009993121703846329,
      "loss": 1.6404,
      "step": 995
    },
    {
      "epoch": 0.08569216209240299,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009993107517742962,
      "loss": 1.3023,
      "step": 996
    },
    {
      "epoch": 0.08577819839972468,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009993093317035727,
      "loss": 1.4805,
      "step": 997
    },
    {
      "epoch": 0.08586423470704638,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009993079101724668,
      "loss": 2.2795,
      "step": 998
    },
    {
      "epoch": 0.08595027101436806,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009993064871809821,
      "loss": 1.2946,
      "step": 999
    },
    {
      "epoch": 0.08603630732168975,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000999305062729123,
      "loss": 0.8236,
      "step": 1000
    },
    {
      "epoch": 0.08612234362901144,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009993036368168937,
      "loss": 1.8479,
      "step": 1001
    },
    {
      "epoch": 0.08620837993633314,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009993022094442986,
      "loss": 1.5082,
      "step": 1002
    },
    {
      "epoch": 0.08629441624365482,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009993007806113415,
      "loss": 1.6009,
      "step": 1003
    },
    {
      "epoch": 0.08638045255097651,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009992993503180267,
      "loss": 1.1232,
      "step": 1004
    },
    {
      "epoch": 0.0864664888582982,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009992979185643584,
      "loss": 1.5509,
      "step": 1005
    },
    {
      "epoch": 0.0865525251656199,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009992964853503409,
      "loss": 1.0849,
      "step": 1006
    },
    {
      "epoch": 0.08663856147294158,
      "grad_norm": 1.25,
      "learning_rate": 0.000999295050675978,
      "loss": 1.5902,
      "step": 1007
    },
    {
      "epoch": 0.08672459778026327,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009992936145412747,
      "loss": 1.7226,
      "step": 1008
    },
    {
      "epoch": 0.08681063408758496,
      "grad_norm": 1.625,
      "learning_rate": 0.0009992921769462343,
      "loss": 1.5552,
      "step": 1009
    },
    {
      "epoch": 0.08689667039490664,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009992907378908616,
      "loss": 1.3966,
      "step": 1010
    },
    {
      "epoch": 0.08698270670222834,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009992892973751603,
      "loss": 2.0535,
      "step": 1011
    },
    {
      "epoch": 0.08706874300955003,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009992878553991352,
      "loss": 0.6858,
      "step": 1012
    },
    {
      "epoch": 0.08715477931687172,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00099928641196279,
      "loss": 1.6405,
      "step": 1013
    },
    {
      "epoch": 0.0872408156241934,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009992849670661295,
      "loss": 1.4679,
      "step": 1014
    },
    {
      "epoch": 0.0873268519315151,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009992835207091573,
      "loss": 1.4364,
      "step": 1015
    },
    {
      "epoch": 0.08741288823883679,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009992820728918779,
      "loss": 1.9502,
      "step": 1016
    },
    {
      "epoch": 0.08749892454615849,
      "grad_norm": 1.5,
      "learning_rate": 0.0009992806236142956,
      "loss": 1.7946,
      "step": 1017
    },
    {
      "epoch": 0.08758496085348016,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009992791728764146,
      "loss": 0.7573,
      "step": 1018
    },
    {
      "epoch": 0.08767099716080186,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009992777206782391,
      "loss": 1.3703,
      "step": 1019
    },
    {
      "epoch": 0.08775703346812355,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009992762670197734,
      "loss": 1.6601,
      "step": 1020
    },
    {
      "epoch": 0.08784306977544523,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009992748119010218,
      "loss": 0.9901,
      "step": 1021
    },
    {
      "epoch": 0.08792910608276693,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009992733553219884,
      "loss": 1.2028,
      "step": 1022
    },
    {
      "epoch": 0.08801514239008862,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009992718972826773,
      "loss": 1.2661,
      "step": 1023
    },
    {
      "epoch": 0.08810117869741031,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009992704377830934,
      "loss": 1.8332,
      "step": 1024
    },
    {
      "epoch": 0.08818721500473199,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009992689768232403,
      "loss": 1.1129,
      "step": 1025
    },
    {
      "epoch": 0.08827325131205369,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009992675144031226,
      "loss": 1.7064,
      "step": 1026
    },
    {
      "epoch": 0.08835928761937538,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009992660505227446,
      "loss": 2.3824,
      "step": 1027
    },
    {
      "epoch": 0.08844532392669707,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009992645851821106,
      "loss": 1.8275,
      "step": 1028
    },
    {
      "epoch": 0.08853136023401875,
      "grad_norm": 1.375,
      "learning_rate": 0.0009992631183812245,
      "loss": 0.7745,
      "step": 1029
    },
    {
      "epoch": 0.08861739654134045,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000999261650120091,
      "loss": 1.168,
      "step": 1030
    },
    {
      "epoch": 0.08870343284866214,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009992601803987145,
      "loss": 1.4689,
      "step": 1031
    },
    {
      "epoch": 0.08878946915598382,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009992587092170987,
      "loss": 1.651,
      "step": 1032
    },
    {
      "epoch": 0.08887550546330551,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009992572365752487,
      "loss": 1.1528,
      "step": 1033
    },
    {
      "epoch": 0.0889615417706272,
      "grad_norm": 1.4375,
      "learning_rate": 0.000999255762473168,
      "loss": 1.0296,
      "step": 1034
    },
    {
      "epoch": 0.0890475780779489,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009992542869108614,
      "loss": 0.9271,
      "step": 1035
    },
    {
      "epoch": 0.08913361438527058,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009992528098883332,
      "loss": 1.3612,
      "step": 1036
    },
    {
      "epoch": 0.08921965069259227,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009992513314055875,
      "loss": 1.6478,
      "step": 1037
    },
    {
      "epoch": 0.08930568699991397,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000999249851462629,
      "loss": 0.9571,
      "step": 1038
    },
    {
      "epoch": 0.08939172330723566,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009992483700594615,
      "loss": 0.8757,
      "step": 1039
    },
    {
      "epoch": 0.08947775961455734,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009992468871960899,
      "loss": 1.248,
      "step": 1040
    },
    {
      "epoch": 0.08956379592187903,
      "grad_norm": 1.8671875,
      "learning_rate": 0.000999245402872518,
      "loss": 1.3591,
      "step": 1041
    },
    {
      "epoch": 0.08964983222920073,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009992439170887503,
      "loss": 1.1599,
      "step": 1042
    },
    {
      "epoch": 0.0897358685365224,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009992424298447915,
      "loss": 1.1508,
      "step": 1043
    },
    {
      "epoch": 0.0898219048438441,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009992409411406456,
      "loss": 1.1719,
      "step": 1044
    },
    {
      "epoch": 0.0899079411511658,
      "grad_norm": 1.375,
      "learning_rate": 0.0009992394509763171,
      "loss": 1.8397,
      "step": 1045
    },
    {
      "epoch": 0.08999397745848749,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009992379593518103,
      "loss": 1.41,
      "step": 1046
    },
    {
      "epoch": 0.09008001376580917,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009992364662671294,
      "loss": 2.3376,
      "step": 1047
    },
    {
      "epoch": 0.09016605007313086,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009992349717222791,
      "loss": 1.7159,
      "step": 1048
    },
    {
      "epoch": 0.09025208638045255,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009992334757172637,
      "loss": 1.7063,
      "step": 1049
    },
    {
      "epoch": 0.09033812268777425,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009992319782520871,
      "loss": 1.3632,
      "step": 1050
    },
    {
      "epoch": 0.09042415899509593,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009992304793267543,
      "loss": 1.6556,
      "step": 1051
    },
    {
      "epoch": 0.09051019530241762,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009992289789412696,
      "loss": 1.4128,
      "step": 1052
    },
    {
      "epoch": 0.09059623160973931,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009992274770956372,
      "loss": 2.0802,
      "step": 1053
    },
    {
      "epoch": 0.090682267917061,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009992259737898613,
      "loss": 1.6238,
      "step": 1054
    },
    {
      "epoch": 0.09076830422438269,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009992244690239466,
      "loss": 1.6674,
      "step": 1055
    },
    {
      "epoch": 0.09085434053170438,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009992229627978973,
      "loss": 2.164,
      "step": 1056
    },
    {
      "epoch": 0.09094037683902607,
      "grad_norm": 1.921875,
      "learning_rate": 0.000999221455111718,
      "loss": 1.6079,
      "step": 1057
    },
    {
      "epoch": 0.09102641314634775,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009992199459654132,
      "loss": 1.3314,
      "step": 1058
    },
    {
      "epoch": 0.09111244945366945,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000999218435358987,
      "loss": 1.0193,
      "step": 1059
    },
    {
      "epoch": 0.09119848576099114,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009992169232924443,
      "loss": 1.5883,
      "step": 1060
    },
    {
      "epoch": 0.09128452206831283,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009992154097657887,
      "loss": 1.3834,
      "step": 1061
    },
    {
      "epoch": 0.09137055837563451,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009992138947790253,
      "loss": 1.2253,
      "step": 1062
    },
    {
      "epoch": 0.09145659468295621,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009992123783321585,
      "loss": 1.1146,
      "step": 1063
    },
    {
      "epoch": 0.0915426309902779,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009992108604251924,
      "loss": 1.484,
      "step": 1064
    },
    {
      "epoch": 0.09162866729759958,
      "grad_norm": 1.375,
      "learning_rate": 0.0009992093410581319,
      "loss": 1.2384,
      "step": 1065
    },
    {
      "epoch": 0.09171470360492127,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009992078202309808,
      "loss": 1.9646,
      "step": 1066
    },
    {
      "epoch": 0.09180073991224297,
      "grad_norm": 1.328125,
      "learning_rate": 0.000999206297943744,
      "loss": 0.8229,
      "step": 1067
    },
    {
      "epoch": 0.09188677621956466,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000999204774196426,
      "loss": 1.8167,
      "step": 1068
    },
    {
      "epoch": 0.09197281252688634,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000999203248989031,
      "loss": 1.9485,
      "step": 1069
    },
    {
      "epoch": 0.09205884883420803,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009992017223215638,
      "loss": 1.0769,
      "step": 1070
    },
    {
      "epoch": 0.09214488514152973,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009992001941940285,
      "loss": 1.1449,
      "step": 1071
    },
    {
      "epoch": 0.09223092144885142,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009991986646064298,
      "loss": 1.0845,
      "step": 1072
    },
    {
      "epoch": 0.0923169577561731,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009991971335587717,
      "loss": 2.0347,
      "step": 1073
    },
    {
      "epoch": 0.0924029940634948,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009991956010510596,
      "loss": 1.4449,
      "step": 1074
    },
    {
      "epoch": 0.09248903037081649,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009991940670832972,
      "loss": 2.1386,
      "step": 1075
    },
    {
      "epoch": 0.09257506667813817,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009991925316554893,
      "loss": 1.0571,
      "step": 1076
    },
    {
      "epoch": 0.09266110298545986,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009991909947676403,
      "loss": 1.2505,
      "step": 1077
    },
    {
      "epoch": 0.09274713929278156,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009991894564197548,
      "loss": 1.8901,
      "step": 1078
    },
    {
      "epoch": 0.09283317560010325,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009991879166118371,
      "loss": 1.1963,
      "step": 1079
    },
    {
      "epoch": 0.09291921190742493,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009991863753438918,
      "loss": 1.5278,
      "step": 1080
    },
    {
      "epoch": 0.09300524821474662,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009991848326159237,
      "loss": 1.1884,
      "step": 1081
    },
    {
      "epoch": 0.09309128452206832,
      "grad_norm": 1.21875,
      "learning_rate": 0.000999183288427937,
      "loss": 1.3667,
      "step": 1082
    },
    {
      "epoch": 0.09317732082939001,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009991817427799361,
      "loss": 2.0727,
      "step": 1083
    },
    {
      "epoch": 0.09326335713671169,
      "grad_norm": 1.5,
      "learning_rate": 0.0009991801956719258,
      "loss": 2.3128,
      "step": 1084
    },
    {
      "epoch": 0.09334939344403338,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009991786471039105,
      "loss": 1.0176,
      "step": 1085
    },
    {
      "epoch": 0.09343542975135508,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0009991770970758946,
      "loss": 0.9877,
      "step": 1086
    },
    {
      "epoch": 0.09352146605867676,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009991755455878827,
      "loss": 0.9986,
      "step": 1087
    },
    {
      "epoch": 0.09360750236599845,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009991739926398797,
      "loss": 1.0448,
      "step": 1088
    },
    {
      "epoch": 0.09369353867332014,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009991724382318898,
      "loss": 1.7697,
      "step": 1089
    },
    {
      "epoch": 0.09377957498064184,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009991708823639176,
      "loss": 1.2296,
      "step": 1090
    },
    {
      "epoch": 0.09386561128796352,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009991693250359676,
      "loss": 1.2262,
      "step": 1091
    },
    {
      "epoch": 0.09395164759528521,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009991677662480444,
      "loss": 1.3075,
      "step": 1092
    },
    {
      "epoch": 0.0940376839026069,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009991662060001526,
      "loss": 1.2992,
      "step": 1093
    },
    {
      "epoch": 0.0941237202099286,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009991646442922967,
      "loss": 1.4396,
      "step": 1094
    },
    {
      "epoch": 0.09420975651725028,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009991630811244814,
      "loss": 1.2719,
      "step": 1095
    },
    {
      "epoch": 0.09429579282457197,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000999161516496711,
      "loss": 1.258,
      "step": 1096
    },
    {
      "epoch": 0.09438182913189366,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009991599504089903,
      "loss": 1.3354,
      "step": 1097
    },
    {
      "epoch": 0.09446786543921534,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009991583828613238,
      "loss": 1.3401,
      "step": 1098
    },
    {
      "epoch": 0.09455390174653704,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009991568138537164,
      "loss": 1.7945,
      "step": 1099
    },
    {
      "epoch": 0.09463993805385873,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000999155243386172,
      "loss": 2.4754,
      "step": 1100
    },
    {
      "epoch": 0.09472597436118042,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009991536714586957,
      "loss": 2.4138,
      "step": 1101
    },
    {
      "epoch": 0.0948120106685021,
      "grad_norm": 1.7734375,
      "learning_rate": 0.000999152098071292,
      "loss": 1.4927,
      "step": 1102
    },
    {
      "epoch": 0.0948980469758238,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009991505232239658,
      "loss": 2.5169,
      "step": 1103
    },
    {
      "epoch": 0.09498408328314549,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000999148946916721,
      "loss": 1.667,
      "step": 1104
    },
    {
      "epoch": 0.09507011959046718,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009991473691495626,
      "loss": 1.0666,
      "step": 1105
    },
    {
      "epoch": 0.09515615589778886,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009991457899224956,
      "loss": 1.1265,
      "step": 1106
    },
    {
      "epoch": 0.09524219220511056,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009991442092355239,
      "loss": 0.8873,
      "step": 1107
    },
    {
      "epoch": 0.09532822851243225,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009991426270886527,
      "loss": 1.4459,
      "step": 1108
    },
    {
      "epoch": 0.09541426481975393,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009991410434818861,
      "loss": 1.8482,
      "step": 1109
    },
    {
      "epoch": 0.09550030112707562,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009991394584152294,
      "loss": 1.3566,
      "step": 1110
    },
    {
      "epoch": 0.09558633743439732,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009991378718886865,
      "loss": 1.0966,
      "step": 1111
    },
    {
      "epoch": 0.09567237374171901,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009991362839022627,
      "loss": 1.3665,
      "step": 1112
    },
    {
      "epoch": 0.09575841004904069,
      "grad_norm": 1.03125,
      "learning_rate": 0.000999134694455962,
      "loss": 1.3425,
      "step": 1113
    },
    {
      "epoch": 0.09584444635636238,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009991331035497898,
      "loss": 1.5079,
      "step": 1114
    },
    {
      "epoch": 0.09593048266368408,
      "grad_norm": 1.140625,
      "learning_rate": 0.00099913151118375,
      "loss": 1.2651,
      "step": 1115
    },
    {
      "epoch": 0.09601651897100577,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000999129917357848,
      "loss": 1.9978,
      "step": 1116
    },
    {
      "epoch": 0.09610255527832745,
      "grad_norm": 1.125,
      "learning_rate": 0.0009991283220720877,
      "loss": 0.9457,
      "step": 1117
    },
    {
      "epoch": 0.09618859158564914,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009991267253264742,
      "loss": 1.1571,
      "step": 1118
    },
    {
      "epoch": 0.09627462789297084,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009991251271210123,
      "loss": 1.333,
      "step": 1119
    },
    {
      "epoch": 0.09636066420029252,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009991235274557063,
      "loss": 1.4939,
      "step": 1120
    },
    {
      "epoch": 0.09644670050761421,
      "grad_norm": 1.109375,
      "learning_rate": 0.000999121926330561,
      "loss": 0.9125,
      "step": 1121
    },
    {
      "epoch": 0.0965327368149359,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009991203237455812,
      "loss": 1.3461,
      "step": 1122
    },
    {
      "epoch": 0.0966187731222576,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009991187197007716,
      "loss": 1.6331,
      "step": 1123
    },
    {
      "epoch": 0.09670480942957928,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009991171141961368,
      "loss": 1.5224,
      "step": 1124
    },
    {
      "epoch": 0.09679084573690097,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009991155072316816,
      "loss": 1.35,
      "step": 1125
    },
    {
      "epoch": 0.09687688204422266,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009991138988074105,
      "loss": 1.2941,
      "step": 1126
    },
    {
      "epoch": 0.09696291835154436,
      "grad_norm": 2.015625,
      "learning_rate": 0.000999112288923328,
      "loss": 1.3761,
      "step": 1127
    },
    {
      "epoch": 0.09704895465886604,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009991106775794397,
      "loss": 1.4079,
      "step": 1128
    },
    {
      "epoch": 0.09713499096618773,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009991090647757494,
      "loss": 1.5918,
      "step": 1129
    },
    {
      "epoch": 0.09722102727350942,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009991074505122624,
      "loss": 1.4042,
      "step": 1130
    },
    {
      "epoch": 0.0973070635808311,
      "grad_norm": 1.40625,
      "learning_rate": 0.000999105834788983,
      "loss": 1.0383,
      "step": 1131
    },
    {
      "epoch": 0.0973930998881528,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009991042176059161,
      "loss": 1.46,
      "step": 1132
    },
    {
      "epoch": 0.09747913619547449,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009991025989630664,
      "loss": 1.6232,
      "step": 1133
    },
    {
      "epoch": 0.09756517250279619,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009991009788604388,
      "loss": 1.0208,
      "step": 1134
    },
    {
      "epoch": 0.09765120881011786,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009990993572980378,
      "loss": 1.2955,
      "step": 1135
    },
    {
      "epoch": 0.09773724511743956,
      "grad_norm": 1.25,
      "learning_rate": 0.0009990977342758682,
      "loss": 1.5018,
      "step": 1136
    },
    {
      "epoch": 0.09782328142476125,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000999096109793935,
      "loss": 1.0581,
      "step": 1137
    },
    {
      "epoch": 0.09790931773208295,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009990944838522426,
      "loss": 0.7648,
      "step": 1138
    },
    {
      "epoch": 0.09799535403940463,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000999092856450796,
      "loss": 1.0784,
      "step": 1139
    },
    {
      "epoch": 0.09808139034672632,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009990912275895999,
      "loss": 1.9149,
      "step": 1140
    },
    {
      "epoch": 0.09816742665404801,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009990895972686587,
      "loss": 2.0462,
      "step": 1141
    },
    {
      "epoch": 0.09825346296136969,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009990879654879778,
      "loss": 1.0708,
      "step": 1142
    },
    {
      "epoch": 0.09833949926869139,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009990863322475617,
      "loss": 1.0983,
      "step": 1143
    },
    {
      "epoch": 0.09842553557601308,
      "grad_norm": 1.1875,
      "learning_rate": 0.000999084697547415,
      "loss": 1.3197,
      "step": 1144
    },
    {
      "epoch": 0.09851157188333477,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009990830613875426,
      "loss": 2.4302,
      "step": 1145
    },
    {
      "epoch": 0.09859760819065645,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009990814237679493,
      "loss": 1.9024,
      "step": 1146
    },
    {
      "epoch": 0.09868364449797815,
      "grad_norm": 1.6875,
      "learning_rate": 0.00099907978468864,
      "loss": 2.119,
      "step": 1147
    },
    {
      "epoch": 0.09876968080529984,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009990781441496194,
      "loss": 1.4538,
      "step": 1148
    },
    {
      "epoch": 0.09885571711262153,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009990765021508923,
      "loss": 1.9979,
      "step": 1149
    },
    {
      "epoch": 0.09894175341994321,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009990748586924635,
      "loss": 1.4712,
      "step": 1150
    },
    {
      "epoch": 0.0990277897272649,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009990732137743377,
      "loss": 1.8217,
      "step": 1151
    },
    {
      "epoch": 0.0991138260345866,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009990715673965198,
      "loss": 1.4132,
      "step": 1152
    },
    {
      "epoch": 0.09919986234190828,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009990699195590147,
      "loss": 1.0658,
      "step": 1153
    },
    {
      "epoch": 0.09928589864922997,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009990682702618273,
      "loss": 1.5404,
      "step": 1154
    },
    {
      "epoch": 0.09937193495655167,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000999066619504962,
      "loss": 1.2054,
      "step": 1155
    },
    {
      "epoch": 0.09945797126387336,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000999064967288424,
      "loss": 1.2988,
      "step": 1156
    },
    {
      "epoch": 0.09954400757119504,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009990633136122181,
      "loss": 1.1256,
      "step": 1157
    },
    {
      "epoch": 0.09963004387851673,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000999061658476349,
      "loss": 1.5268,
      "step": 1158
    },
    {
      "epoch": 0.09971608018583843,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009990600018808216,
      "loss": 1.5963,
      "step": 1159
    },
    {
      "epoch": 0.09980211649316012,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009990583438256409,
      "loss": 1.7106,
      "step": 1160
    },
    {
      "epoch": 0.0998881528004818,
      "grad_norm": 1.625,
      "learning_rate": 0.0009990566843108115,
      "loss": 2.3139,
      "step": 1161
    },
    {
      "epoch": 0.0999741891078035,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009990550233363383,
      "loss": 2.0033,
      "step": 1162
    },
    {
      "epoch": 0.10006022541512519,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009990533609022263,
      "loss": 0.7776,
      "step": 1163
    },
    {
      "epoch": 0.10014626172244687,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009990516970084804,
      "loss": 1.363,
      "step": 1164
    },
    {
      "epoch": 0.10023229802976856,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009990500316551052,
      "loss": 1.3189,
      "step": 1165
    },
    {
      "epoch": 0.10031833433709025,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009990483648421055,
      "loss": 1.0238,
      "step": 1166
    },
    {
      "epoch": 0.10040437064441195,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009990466965694867,
      "loss": 1.0326,
      "step": 1167
    },
    {
      "epoch": 0.10049040695173363,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009990450268372534,
      "loss": 0.672,
      "step": 1168
    },
    {
      "epoch": 0.10057644325905532,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009990433556454103,
      "loss": 1.4744,
      "step": 1169
    },
    {
      "epoch": 0.10066247956637701,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009990416829939626,
      "loss": 1.9864,
      "step": 1170
    },
    {
      "epoch": 0.10074851587369871,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009990400088829149,
      "loss": 2.1211,
      "step": 1171
    },
    {
      "epoch": 0.10083455218102039,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009990383333122721,
      "loss": 1.0696,
      "step": 1172
    },
    {
      "epoch": 0.10092058848834208,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009990366562820394,
      "loss": 1.2853,
      "step": 1173
    },
    {
      "epoch": 0.10100662479566377,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009990349777922215,
      "loss": 0.9629,
      "step": 1174
    },
    {
      "epoch": 0.10109266110298545,
      "grad_norm": 2.21875,
      "learning_rate": 0.0009990332978428234,
      "loss": 2.0513,
      "step": 1175
    },
    {
      "epoch": 0.10117869741030715,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00099903161643385,
      "loss": 1.1439,
      "step": 1176
    },
    {
      "epoch": 0.10126473371762884,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009990299335653062,
      "loss": 1.4144,
      "step": 1177
    },
    {
      "epoch": 0.10135077002495053,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009990282492371967,
      "loss": 1.6206,
      "step": 1178
    },
    {
      "epoch": 0.10143680633227221,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009990265634495267,
      "loss": 1.9103,
      "step": 1179
    },
    {
      "epoch": 0.10152284263959391,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009990248762023013,
      "loss": 1.6817,
      "step": 1180
    },
    {
      "epoch": 0.1016088789469156,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009990231874955249,
      "loss": 1.7373,
      "step": 1181
    },
    {
      "epoch": 0.1016949152542373,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009990214973292027,
      "loss": 0.9956,
      "step": 1182
    },
    {
      "epoch": 0.10178095156155897,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009990198057033398,
      "loss": 1.381,
      "step": 1183
    },
    {
      "epoch": 0.10186698786888067,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009990181126179412,
      "loss": 0.9023,
      "step": 1184
    },
    {
      "epoch": 0.10195302417620236,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009990164180730115,
      "loss": 0.947,
      "step": 1185
    },
    {
      "epoch": 0.10203906048352404,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009990147220685557,
      "loss": 1.1467,
      "step": 1186
    },
    {
      "epoch": 0.10212509679084573,
      "grad_norm": 2.171875,
      "learning_rate": 0.000999013024604579,
      "loss": 1.3592,
      "step": 1187
    },
    {
      "epoch": 0.10221113309816743,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009990113256810862,
      "loss": 1.047,
      "step": 1188
    },
    {
      "epoch": 0.10229716940548912,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009990096252980825,
      "loss": 1.9141,
      "step": 1189
    },
    {
      "epoch": 0.1023832057128108,
      "grad_norm": 1.75,
      "learning_rate": 0.0009990079234555725,
      "loss": 1.7206,
      "step": 1190
    },
    {
      "epoch": 0.1024692420201325,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009990062201535614,
      "loss": 1.3147,
      "step": 1191
    },
    {
      "epoch": 0.10255527832745419,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009990045153920543,
      "loss": 2.505,
      "step": 1192
    },
    {
      "epoch": 0.10264131463477588,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000999002809171056,
      "loss": 1.4548,
      "step": 1193
    },
    {
      "epoch": 0.10272735094209756,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009990011014905716,
      "loss": 0.6957,
      "step": 1194
    },
    {
      "epoch": 0.10281338724941926,
      "grad_norm": 1.421875,
      "learning_rate": 0.000998999392350606,
      "loss": 1.8604,
      "step": 1195
    },
    {
      "epoch": 0.10289942355674095,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000998997681751164,
      "loss": 0.9738,
      "step": 1196
    },
    {
      "epoch": 0.10298545986406263,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009989959696922511,
      "loss": 2.0421,
      "step": 1197
    },
    {
      "epoch": 0.10307149617138432,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009989942561738718,
      "loss": 1.3112,
      "step": 1198
    },
    {
      "epoch": 0.10315753247870602,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009989925411960315,
      "loss": 1.1345,
      "step": 1199
    },
    {
      "epoch": 0.10324356878602771,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000998990824758735,
      "loss": 1.2769,
      "step": 1200
    },
    {
      "epoch": 0.10332960509334939,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009989891068619874,
      "loss": 0.9316,
      "step": 1201
    },
    {
      "epoch": 0.10341564140067108,
      "grad_norm": 1.125,
      "learning_rate": 0.0009989873875057937,
      "loss": 1.1085,
      "step": 1202
    },
    {
      "epoch": 0.10350167770799278,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009989856666901592,
      "loss": 0.9969,
      "step": 1203
    },
    {
      "epoch": 0.10358771401531447,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009989839444150884,
      "loss": 1.6558,
      "step": 1204
    },
    {
      "epoch": 0.10367375032263615,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009989822206805865,
      "loss": 1.4175,
      "step": 1205
    },
    {
      "epoch": 0.10375978662995784,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009989804954866589,
      "loss": 1.2426,
      "step": 1206
    },
    {
      "epoch": 0.10384582293727954,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009989787688333104,
      "loss": 1.1214,
      "step": 1207
    },
    {
      "epoch": 0.10393185924460122,
      "grad_norm": 1.140625,
      "learning_rate": 0.000998977040720546,
      "loss": 1.937,
      "step": 1208
    },
    {
      "epoch": 0.10401789555192291,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009989753111483706,
      "loss": 1.1311,
      "step": 1209
    },
    {
      "epoch": 0.1041039318592446,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009989735801167897,
      "loss": 2.2437,
      "step": 1210
    },
    {
      "epoch": 0.1041899681665663,
      "grad_norm": 1.578125,
      "learning_rate": 0.000998971847625808,
      "loss": 1.5979,
      "step": 1211
    },
    {
      "epoch": 0.10427600447388798,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009989701136754307,
      "loss": 1.2903,
      "step": 1212
    },
    {
      "epoch": 0.10436204078120967,
      "grad_norm": 1.78125,
      "learning_rate": 0.000998968378265663,
      "loss": 0.9993,
      "step": 1213
    },
    {
      "epoch": 0.10444807708853136,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009989666413965097,
      "loss": 1.1161,
      "step": 1214
    },
    {
      "epoch": 0.10453411339585306,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009989649030679758,
      "loss": 0.7496,
      "step": 1215
    },
    {
      "epoch": 0.10462014970317474,
      "grad_norm": 1.8046875,
      "learning_rate": 0.000998963163280067,
      "loss": 1.2255,
      "step": 1216
    },
    {
      "epoch": 0.10470618601049643,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009989614220327879,
      "loss": 1.448,
      "step": 1217
    },
    {
      "epoch": 0.10479222231781812,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009989596793261435,
      "loss": 1.5971,
      "step": 1218
    },
    {
      "epoch": 0.1048782586251398,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009989579351601392,
      "loss": 1.3887,
      "step": 1219
    },
    {
      "epoch": 0.1049642949324615,
      "grad_norm": 1.53125,
      "learning_rate": 0.00099895618953478,
      "loss": 1.136,
      "step": 1220
    },
    {
      "epoch": 0.10505033123978319,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000998954442450071,
      "loss": 1.585,
      "step": 1221
    },
    {
      "epoch": 0.10513636754710488,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009989526939060173,
      "loss": 1.2977,
      "step": 1222
    },
    {
      "epoch": 0.10522240385442656,
      "grad_norm": 1.421875,
      "learning_rate": 0.000998950943902624,
      "loss": 0.9281,
      "step": 1223
    },
    {
      "epoch": 0.10530844016174826,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009989491924398961,
      "loss": 1.988,
      "step": 1224
    },
    {
      "epoch": 0.10539447646906995,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000998947439517839,
      "loss": 1.2639,
      "step": 1225
    },
    {
      "epoch": 0.10548051277639164,
      "grad_norm": 1.375,
      "learning_rate": 0.0009989456851364578,
      "loss": 1.3932,
      "step": 1226
    },
    {
      "epoch": 0.10556654908371332,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000998943929295757,
      "loss": 1.6616,
      "step": 1227
    },
    {
      "epoch": 0.10565258539103502,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009989421719957428,
      "loss": 2.0151,
      "step": 1228
    },
    {
      "epoch": 0.10573862169835671,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009989404132364195,
      "loss": 1.3938,
      "step": 1229
    },
    {
      "epoch": 0.10582465800567839,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009989386530177928,
      "loss": 1.7144,
      "step": 1230
    },
    {
      "epoch": 0.10591069431300008,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009989368913398673,
      "loss": 1.352,
      "step": 1231
    },
    {
      "epoch": 0.10599673062032178,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009989351282026486,
      "loss": 1.4455,
      "step": 1232
    },
    {
      "epoch": 0.10608276692764347,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009989333636061415,
      "loss": 0.8834,
      "step": 1233
    },
    {
      "epoch": 0.10616880323496515,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009989315975503514,
      "loss": 1.0448,
      "step": 1234
    },
    {
      "epoch": 0.10625483954228684,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009989298300352835,
      "loss": 0.834,
      "step": 1235
    },
    {
      "epoch": 0.10634087584960854,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009989280610609428,
      "loss": 2.026,
      "step": 1236
    },
    {
      "epoch": 0.10642691215693023,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009989262906273347,
      "loss": 1.3603,
      "step": 1237
    },
    {
      "epoch": 0.10651294846425191,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009989245187344642,
      "loss": 1.44,
      "step": 1238
    },
    {
      "epoch": 0.1065989847715736,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009989227453823363,
      "loss": 1.4315,
      "step": 1239
    },
    {
      "epoch": 0.1066850210788953,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009989209705709569,
      "loss": 1.5163,
      "step": 1240
    },
    {
      "epoch": 0.10677105738621698,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009989191943003302,
      "loss": 1.9968,
      "step": 1241
    },
    {
      "epoch": 0.10685709369353867,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000998917416570462,
      "loss": 1.5029,
      "step": 1242
    },
    {
      "epoch": 0.10694313000086036,
      "grad_norm": 1.0,
      "learning_rate": 0.0009989156373813575,
      "loss": 1.4177,
      "step": 1243
    },
    {
      "epoch": 0.10702916630818206,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000998913856733022,
      "loss": 1.6079,
      "step": 1244
    },
    {
      "epoch": 0.10711520261550374,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009989120746254601,
      "loss": 1.0219,
      "step": 1245
    },
    {
      "epoch": 0.10720123892282543,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009989102910586776,
      "loss": 1.6189,
      "step": 1246
    },
    {
      "epoch": 0.10728727523014712,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009989085060326793,
      "loss": 1.3467,
      "step": 1247
    },
    {
      "epoch": 0.10737331153746882,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000998906719547471,
      "loss": 1.4834,
      "step": 1248
    },
    {
      "epoch": 0.1074593478447905,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009989049316030574,
      "loss": 0.7018,
      "step": 1249
    },
    {
      "epoch": 0.10754538415211219,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009989031421994438,
      "loss": 1.0043,
      "step": 1250
    },
    {
      "epoch": 0.10763142045943389,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009989013513366357,
      "loss": 1.5263,
      "step": 1251
    },
    {
      "epoch": 0.10771745676675556,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000998899559014638,
      "loss": 1.1717,
      "step": 1252
    },
    {
      "epoch": 0.10780349307407726,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009988977652334564,
      "loss": 1.7494,
      "step": 1253
    },
    {
      "epoch": 0.10788952938139895,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009988959699930957,
      "loss": 1.4074,
      "step": 1254
    },
    {
      "epoch": 0.10797556568872065,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009988941732935615,
      "loss": 1.0547,
      "step": 1255
    },
    {
      "epoch": 0.10806160199604233,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009988923751348586,
      "loss": 1.3445,
      "step": 1256
    },
    {
      "epoch": 0.10814763830336402,
      "grad_norm": 1.25,
      "learning_rate": 0.0009988905755169927,
      "loss": 1.1986,
      "step": 1257
    },
    {
      "epoch": 0.10823367461068571,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009988887744399688,
      "loss": 1.6678,
      "step": 1258
    },
    {
      "epoch": 0.1083197109180074,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009988869719037923,
      "loss": 1.2679,
      "step": 1259
    },
    {
      "epoch": 0.10840574722532909,
      "grad_norm": 1.75,
      "learning_rate": 0.0009988851679084685,
      "loss": 1.3657,
      "step": 1260
    },
    {
      "epoch": 0.10849178353265078,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009988833624540027,
      "loss": 0.8534,
      "step": 1261
    },
    {
      "epoch": 0.10857781983997247,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009988815555403998,
      "loss": 1.0533,
      "step": 1262
    },
    {
      "epoch": 0.10866385614729415,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009988797471676657,
      "loss": 1.9428,
      "step": 1263
    },
    {
      "epoch": 0.10874989245461585,
      "grad_norm": 1.359375,
      "learning_rate": 0.000998877937335805,
      "loss": 1.2996,
      "step": 1264
    },
    {
      "epoch": 0.10883592876193754,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009988761260448236,
      "loss": 1.6704,
      "step": 1265
    },
    {
      "epoch": 0.10892196506925923,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009988743132947264,
      "loss": 1.4361,
      "step": 1266
    },
    {
      "epoch": 0.10900800137658091,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000998872499085519,
      "loss": 1.9437,
      "step": 1267
    },
    {
      "epoch": 0.1090940376839026,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009988706834172067,
      "loss": 1.7167,
      "step": 1268
    },
    {
      "epoch": 0.1091800739912243,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009988688662897944,
      "loss": 2.0906,
      "step": 1269
    },
    {
      "epoch": 0.109266110298546,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009988670477032878,
      "loss": 2.1194,
      "step": 1270
    },
    {
      "epoch": 0.10935214660586767,
      "grad_norm": 1.5,
      "learning_rate": 0.0009988652276576922,
      "loss": 1.3403,
      "step": 1271
    },
    {
      "epoch": 0.10943818291318937,
      "grad_norm": 1.875,
      "learning_rate": 0.0009988634061530127,
      "loss": 1.6783,
      "step": 1272
    },
    {
      "epoch": 0.10952421922051106,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000998861583189255,
      "loss": 1.2941,
      "step": 1273
    },
    {
      "epoch": 0.10961025552783274,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009988597587664238,
      "loss": 1.6688,
      "step": 1274
    },
    {
      "epoch": 0.10969629183515443,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009988579328845252,
      "loss": 1.1227,
      "step": 1275
    },
    {
      "epoch": 0.10978232814247613,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000998856105543564,
      "loss": 1.3882,
      "step": 1276
    },
    {
      "epoch": 0.10986836444979782,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009988542767435456,
      "loss": 1.4174,
      "step": 1277
    },
    {
      "epoch": 0.1099544007571195,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009988524464844754,
      "loss": 1.3855,
      "step": 1278
    },
    {
      "epoch": 0.1100404370644412,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009988506147663593,
      "loss": 1.2044,
      "step": 1279
    },
    {
      "epoch": 0.11012647337176289,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009988487815892019,
      "loss": 1.3838,
      "step": 1280
    },
    {
      "epoch": 0.11021250967908458,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009988469469530087,
      "loss": 1.8393,
      "step": 1281
    },
    {
      "epoch": 0.11029854598640626,
      "grad_norm": 1.125,
      "learning_rate": 0.0009988451108577851,
      "loss": 0.8965,
      "step": 1282
    },
    {
      "epoch": 0.11038458229372795,
      "grad_norm": 1.375,
      "learning_rate": 0.0009988432733035368,
      "loss": 1.8665,
      "step": 1283
    },
    {
      "epoch": 0.11047061860104965,
      "grad_norm": 1.203125,
      "learning_rate": 0.000998841434290269,
      "loss": 2.0556,
      "step": 1284
    },
    {
      "epoch": 0.11055665490837133,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009988395938179868,
      "loss": 1.9811,
      "step": 1285
    },
    {
      "epoch": 0.11064269121569302,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000998837751886696,
      "loss": 1.4432,
      "step": 1286
    },
    {
      "epoch": 0.11072872752301471,
      "grad_norm": 1.75,
      "learning_rate": 0.0009988359084964018,
      "loss": 1.1355,
      "step": 1287
    },
    {
      "epoch": 0.11081476383033641,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009988340636471096,
      "loss": 1.6216,
      "step": 1288
    },
    {
      "epoch": 0.11090080013765809,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009988322173388247,
      "loss": 0.9659,
      "step": 1289
    },
    {
      "epoch": 0.11098683644497978,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009988303695715526,
      "loss": 1.7812,
      "step": 1290
    },
    {
      "epoch": 0.11107287275230147,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009988285203452985,
      "loss": 1.7494,
      "step": 1291
    },
    {
      "epoch": 0.11115890905962317,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009988266696600683,
      "loss": 2.1088,
      "step": 1292
    },
    {
      "epoch": 0.11124494536694485,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000998824817515867,
      "loss": 1.2105,
      "step": 1293
    },
    {
      "epoch": 0.11133098167426654,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009988229639127,
      "loss": 0.9631,
      "step": 1294
    },
    {
      "epoch": 0.11141701798158823,
      "grad_norm": 1.5,
      "learning_rate": 0.000998821108850573,
      "loss": 1.6677,
      "step": 1295
    },
    {
      "epoch": 0.11150305428890991,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009988192523294913,
      "loss": 1.0924,
      "step": 1296
    },
    {
      "epoch": 0.11158909059623161,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009988173943494601,
      "loss": 2.0588,
      "step": 1297
    },
    {
      "epoch": 0.1116751269035533,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009988155349104852,
      "loss": 1.3334,
      "step": 1298
    },
    {
      "epoch": 0.111761163210875,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009988136740125718,
      "loss": 1.6178,
      "step": 1299
    },
    {
      "epoch": 0.11184719951819667,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009988118116557253,
      "loss": 1.3214,
      "step": 1300
    },
    {
      "epoch": 0.11193323582551837,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009988099478399515,
      "loss": 1.9559,
      "step": 1301
    },
    {
      "epoch": 0.11201927213284006,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009988080825652557,
      "loss": 1.4433,
      "step": 1302
    },
    {
      "epoch": 0.11210530844016176,
      "grad_norm": 1.265625,
      "learning_rate": 0.000998806215831643,
      "loss": 1.194,
      "step": 1303
    },
    {
      "epoch": 0.11219134474748343,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009988043476391194,
      "loss": 1.0738,
      "step": 1304
    },
    {
      "epoch": 0.11227738105480513,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009988024779876898,
      "loss": 1.2005,
      "step": 1305
    },
    {
      "epoch": 0.11236341736212682,
      "grad_norm": 1.46875,
      "learning_rate": 0.00099880060687736,
      "loss": 1.4197,
      "step": 1306
    },
    {
      "epoch": 0.1124494536694485,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009987987343081355,
      "loss": 1.4271,
      "step": 1307
    },
    {
      "epoch": 0.1125354899767702,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009987968602800218,
      "loss": 1.575,
      "step": 1308
    },
    {
      "epoch": 0.11262152628409189,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009987949847930242,
      "loss": 1.8114,
      "step": 1309
    },
    {
      "epoch": 0.11270756259141358,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009987931078471483,
      "loss": 1.801,
      "step": 1310
    },
    {
      "epoch": 0.11279359889873526,
      "grad_norm": 1.25,
      "learning_rate": 0.0009987912294423997,
      "loss": 1.096,
      "step": 1311
    },
    {
      "epoch": 0.11287963520605696,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009987893495787836,
      "loss": 1.4024,
      "step": 1312
    },
    {
      "epoch": 0.11296567151337865,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009987874682563058,
      "loss": 1.1509,
      "step": 1313
    },
    {
      "epoch": 0.11305170782070034,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009987855854749715,
      "loss": 1.2199,
      "step": 1314
    },
    {
      "epoch": 0.11313774412802202,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009987837012347864,
      "loss": 1.659,
      "step": 1315
    },
    {
      "epoch": 0.11322378043534372,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009987818155357561,
      "loss": 1.2265,
      "step": 1316
    },
    {
      "epoch": 0.11330981674266541,
      "grad_norm": 4.375,
      "learning_rate": 0.0009987799283778858,
      "loss": 2.3416,
      "step": 1317
    },
    {
      "epoch": 0.11339585304998709,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009987780397611814,
      "loss": 1.6711,
      "step": 1318
    },
    {
      "epoch": 0.11348188935730878,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009987761496856483,
      "loss": 1.7303,
      "step": 1319
    },
    {
      "epoch": 0.11356792566463048,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009987742581512918,
      "loss": 1.8742,
      "step": 1320
    },
    {
      "epoch": 0.11365396197195217,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009987723651581177,
      "loss": 2.2332,
      "step": 1321
    },
    {
      "epoch": 0.11373999827927385,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009987704707061314,
      "loss": 1.4626,
      "step": 1322
    },
    {
      "epoch": 0.11382603458659554,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009987685747953384,
      "loss": 1.4149,
      "step": 1323
    },
    {
      "epoch": 0.11391207089391724,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009987666774257445,
      "loss": 1.745,
      "step": 1324
    },
    {
      "epoch": 0.11399810720123893,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009987647785973549,
      "loss": 1.7693,
      "step": 1325
    },
    {
      "epoch": 0.11408414350856061,
      "grad_norm": 1.5,
      "learning_rate": 0.0009987628783101753,
      "loss": 1.6533,
      "step": 1326
    },
    {
      "epoch": 0.1141701798158823,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009987609765642115,
      "loss": 2.2973,
      "step": 1327
    },
    {
      "epoch": 0.114256216123204,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009987590733594688,
      "loss": 1.6209,
      "step": 1328
    },
    {
      "epoch": 0.11434225243052568,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009987571686959525,
      "loss": 1.2174,
      "step": 1329
    },
    {
      "epoch": 0.11442828873784737,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009987552625736688,
      "loss": 1.485,
      "step": 1330
    },
    {
      "epoch": 0.11451432504516906,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000998753354992623,
      "loss": 0.8335,
      "step": 1331
    },
    {
      "epoch": 0.11460036135249076,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009987514459528204,
      "loss": 1.2938,
      "step": 1332
    },
    {
      "epoch": 0.11468639765981244,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000998749535454267,
      "loss": 0.8634,
      "step": 1333
    },
    {
      "epoch": 0.11477243396713413,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009987476234969682,
      "loss": 1.398,
      "step": 1334
    },
    {
      "epoch": 0.11485847027445582,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009987457100809295,
      "loss": 1.5695,
      "step": 1335
    },
    {
      "epoch": 0.11494450658177752,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009987437952061566,
      "loss": 1.9716,
      "step": 1336
    },
    {
      "epoch": 0.1150305428890992,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009987418788726552,
      "loss": 1.7755,
      "step": 1337
    },
    {
      "epoch": 0.11511657919642089,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009987399610804307,
      "loss": 1.2782,
      "step": 1338
    },
    {
      "epoch": 0.11520261550374258,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000998738041829489,
      "loss": 1.842,
      "step": 1339
    },
    {
      "epoch": 0.11528865181106426,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009987361211198354,
      "loss": 1.5136,
      "step": 1340
    },
    {
      "epoch": 0.11537468811838596,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009987341989514756,
      "loss": 1.1656,
      "step": 1341
    },
    {
      "epoch": 0.11546072442570765,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009987322753244152,
      "loss": 0.885,
      "step": 1342
    },
    {
      "epoch": 0.11554676073302934,
      "grad_norm": 1.515625,
      "learning_rate": 0.00099873035023866,
      "loss": 1.4677,
      "step": 1343
    },
    {
      "epoch": 0.11563279704035102,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009987284236942155,
      "loss": 1.1573,
      "step": 1344
    },
    {
      "epoch": 0.11571883334767272,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009987264956910875,
      "loss": 1.2994,
      "step": 1345
    },
    {
      "epoch": 0.11580486965499441,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009987245662292812,
      "loss": 1.2324,
      "step": 1346
    },
    {
      "epoch": 0.1158909059623161,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009987226353088028,
      "loss": 1.0686,
      "step": 1347
    },
    {
      "epoch": 0.11597694226963778,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009987207029296577,
      "loss": 1.212,
      "step": 1348
    },
    {
      "epoch": 0.11606297857695948,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009987187690918514,
      "loss": 0.8198,
      "step": 1349
    },
    {
      "epoch": 0.11614901488428117,
      "grad_norm": 1.578125,
      "learning_rate": 0.00099871683379539,
      "loss": 1.4652,
      "step": 1350
    },
    {
      "epoch": 0.11623505119160285,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009987148970402784,
      "loss": 1.9368,
      "step": 1351
    },
    {
      "epoch": 0.11632108749892454,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009987129588265229,
      "loss": 2.0955,
      "step": 1352
    },
    {
      "epoch": 0.11640712380624624,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000998711019154129,
      "loss": 1.7545,
      "step": 1353
    },
    {
      "epoch": 0.11649316011356793,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009987090780231023,
      "loss": 1.1168,
      "step": 1354
    },
    {
      "epoch": 0.11657919642088961,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009987071354334487,
      "loss": 1.6884,
      "step": 1355
    },
    {
      "epoch": 0.1166652327282113,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009987051913851737,
      "loss": 0.854,
      "step": 1356
    },
    {
      "epoch": 0.116751269035533,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009987032458782827,
      "loss": 2.0723,
      "step": 1357
    },
    {
      "epoch": 0.11683730534285469,
      "grad_norm": 1.5,
      "learning_rate": 0.0009987012989127822,
      "loss": 1.5427,
      "step": 1358
    },
    {
      "epoch": 0.11692334165017637,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000998699350488677,
      "loss": 1.2408,
      "step": 1359
    },
    {
      "epoch": 0.11700937795749806,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009986974006059733,
      "loss": 1.4273,
      "step": 1360
    },
    {
      "epoch": 0.11709541426481976,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009986954492646766,
      "loss": 1.1959,
      "step": 1361
    },
    {
      "epoch": 0.11718145057214144,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009986934964647928,
      "loss": 1.1447,
      "step": 1362
    },
    {
      "epoch": 0.11726748687946313,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009986915422063277,
      "loss": 1.7074,
      "step": 1363
    },
    {
      "epoch": 0.11735352318678482,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009986895864892867,
      "loss": 1.2672,
      "step": 1364
    },
    {
      "epoch": 0.11743955949410652,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009986876293136757,
      "loss": 1.6009,
      "step": 1365
    },
    {
      "epoch": 0.1175255958014282,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009986856706795002,
      "loss": 1.0461,
      "step": 1366
    },
    {
      "epoch": 0.11761163210874989,
      "grad_norm": 1.9375,
      "learning_rate": 0.000998683710586766,
      "loss": 1.5139,
      "step": 1367
    },
    {
      "epoch": 0.11769766841607159,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009986817490354794,
      "loss": 1.568,
      "step": 1368
    },
    {
      "epoch": 0.11778370472339328,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009986797860256453,
      "loss": 1.6424,
      "step": 1369
    },
    {
      "epoch": 0.11786974103071496,
      "grad_norm": 1.421875,
      "learning_rate": 0.00099867782155727,
      "loss": 1.2515,
      "step": 1370
    },
    {
      "epoch": 0.11795577733803665,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000998675855630359,
      "loss": 1.3243,
      "step": 1371
    },
    {
      "epoch": 0.11804181364535835,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000998673888244918,
      "loss": 1.4861,
      "step": 1372
    },
    {
      "epoch": 0.11812784995268003,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000998671919400953,
      "loss": 1.3128,
      "step": 1373
    },
    {
      "epoch": 0.11821388626000172,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009986699490984695,
      "loss": 0.9774,
      "step": 1374
    },
    {
      "epoch": 0.11829992256732341,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009986679773374735,
      "loss": 1.366,
      "step": 1375
    },
    {
      "epoch": 0.1183859588746451,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009986660041179706,
      "loss": 1.8449,
      "step": 1376
    },
    {
      "epoch": 0.11847199518196679,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009986640294399665,
      "loss": 1.4587,
      "step": 1377
    },
    {
      "epoch": 0.11855803148928848,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009986620533034673,
      "loss": 0.903,
      "step": 1378
    },
    {
      "epoch": 0.11864406779661017,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009986600757084785,
      "loss": 1.3627,
      "step": 1379
    },
    {
      "epoch": 0.11873010410393187,
      "grad_norm": 2.109375,
      "learning_rate": 0.000998658096655006,
      "loss": 1.8525,
      "step": 1380
    },
    {
      "epoch": 0.11881614041125355,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009986561161430554,
      "loss": 1.2434,
      "step": 1381
    },
    {
      "epoch": 0.11890217671857524,
      "grad_norm": 1.75,
      "learning_rate": 0.0009986541341726327,
      "loss": 1.4051,
      "step": 1382
    },
    {
      "epoch": 0.11898821302589693,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000998652150743744,
      "loss": 1.2769,
      "step": 1383
    },
    {
      "epoch": 0.11907424933321861,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009986501658563942,
      "loss": 1.2822,
      "step": 1384
    },
    {
      "epoch": 0.1191602856405403,
      "grad_norm": 1.203125,
      "learning_rate": 0.00099864817951059,
      "loss": 0.8582,
      "step": 1385
    },
    {
      "epoch": 0.119246321947862,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009986461917063367,
      "loss": 0.9842,
      "step": 1386
    },
    {
      "epoch": 0.1193323582551837,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009986442024436403,
      "loss": 1.1799,
      "step": 1387
    },
    {
      "epoch": 0.11941839456250537,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009986422117225066,
      "loss": 1.3222,
      "step": 1388
    },
    {
      "epoch": 0.11950443086982707,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009986402195429414,
      "loss": 1.2781,
      "step": 1389
    },
    {
      "epoch": 0.11959046717714876,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009986382259049506,
      "loss": 1.2724,
      "step": 1390
    },
    {
      "epoch": 0.11967650348447045,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00099863623080854,
      "loss": 1.2283,
      "step": 1391
    },
    {
      "epoch": 0.11976253979179213,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009986342342537151,
      "loss": 1.7484,
      "step": 1392
    },
    {
      "epoch": 0.11984857609911383,
      "grad_norm": 2.671875,
      "learning_rate": 0.0009986322362404824,
      "loss": 2.3848,
      "step": 1393
    },
    {
      "epoch": 0.11993461240643552,
      "grad_norm": 2.125,
      "learning_rate": 0.0009986302367688473,
      "loss": 1.5583,
      "step": 1394
    },
    {
      "epoch": 0.1200206487137572,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009986282358388156,
      "loss": 2.2,
      "step": 1395
    },
    {
      "epoch": 0.1201066850210789,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009986262334503934,
      "loss": 1.6695,
      "step": 1396
    },
    {
      "epoch": 0.12019272132840059,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009986242296035864,
      "loss": 1.4176,
      "step": 1397
    },
    {
      "epoch": 0.12027875763572228,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009986222242984006,
      "loss": 1.0474,
      "step": 1398
    },
    {
      "epoch": 0.12036479394304396,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009986202175348417,
      "loss": 0.9128,
      "step": 1399
    },
    {
      "epoch": 0.12045083025036565,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009986182093129157,
      "loss": 1.4754,
      "step": 1400
    },
    {
      "epoch": 0.12053686655768735,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009986161996326283,
      "loss": 0.9174,
      "step": 1401
    },
    {
      "epoch": 0.12062290286500904,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009986141884939855,
      "loss": 1.9774,
      "step": 1402
    },
    {
      "epoch": 0.12070893917233072,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009986121758969932,
      "loss": 1.7348,
      "step": 1403
    },
    {
      "epoch": 0.12079497547965241,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009986101618416575,
      "loss": 1.1496,
      "step": 1404
    },
    {
      "epoch": 0.12088101178697411,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009986081463279836,
      "loss": 0.926,
      "step": 1405
    },
    {
      "epoch": 0.12096704809429579,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009986061293559782,
      "loss": 1.0732,
      "step": 1406
    },
    {
      "epoch": 0.12105308440161748,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009986041109256466,
      "loss": 1.7069,
      "step": 1407
    },
    {
      "epoch": 0.12113912070893917,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009986020910369951,
      "loss": 1.1519,
      "step": 1408
    },
    {
      "epoch": 0.12122515701626087,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009986000696900295,
      "loss": 1.6874,
      "step": 1409
    },
    {
      "epoch": 0.12131119332358255,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009985980468847555,
      "loss": 1.5906,
      "step": 1410
    },
    {
      "epoch": 0.12139722963090424,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009985960226211795,
      "loss": 2.1911,
      "step": 1411
    },
    {
      "epoch": 0.12148326593822593,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009985939968993067,
      "loss": 1.8112,
      "step": 1412
    },
    {
      "epoch": 0.12156930224554763,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009985919697191436,
      "loss": 1.4411,
      "step": 1413
    },
    {
      "epoch": 0.12165533855286931,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000998589941080696,
      "loss": 1.3051,
      "step": 1414
    },
    {
      "epoch": 0.121741374860191,
      "grad_norm": 1.21875,
      "learning_rate": 0.00099858791098397,
      "loss": 1.8188,
      "step": 1415
    },
    {
      "epoch": 0.1218274111675127,
      "grad_norm": 1.875,
      "learning_rate": 0.0009985858794289708,
      "loss": 1.3507,
      "step": 1416
    },
    {
      "epoch": 0.12191344747483437,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009985838464157053,
      "loss": 1.7591,
      "step": 1417
    },
    {
      "epoch": 0.12199948378215607,
      "grad_norm": 1.8203125,
      "learning_rate": 0.000998581811944179,
      "loss": 1.8276,
      "step": 1418
    },
    {
      "epoch": 0.12208552008947776,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009985797760143976,
      "loss": 1.1277,
      "step": 1419
    },
    {
      "epoch": 0.12217155639679946,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009985777386263674,
      "loss": 1.3214,
      "step": 1420
    },
    {
      "epoch": 0.12225759270412113,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009985756997800946,
      "loss": 1.0119,
      "step": 1421
    },
    {
      "epoch": 0.12234362901144283,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009985736594755846,
      "loss": 1.2407,
      "step": 1422
    },
    {
      "epoch": 0.12242966531876452,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009985716177128435,
      "loss": 1.8642,
      "step": 1423
    },
    {
      "epoch": 0.12251570162608622,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009985695744918774,
      "loss": 1.3307,
      "step": 1424
    },
    {
      "epoch": 0.1226017379334079,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009985675298126925,
      "loss": 1.3086,
      "step": 1425
    },
    {
      "epoch": 0.12268777424072959,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009985654836752943,
      "loss": 1.6628,
      "step": 1426
    },
    {
      "epoch": 0.12277381054805128,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009985634360796893,
      "loss": 2.0297,
      "step": 1427
    },
    {
      "epoch": 0.12285984685537296,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000998561387025883,
      "loss": 1.7096,
      "step": 1428
    },
    {
      "epoch": 0.12294588316269466,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009985593365138816,
      "loss": 1.5665,
      "step": 1429
    },
    {
      "epoch": 0.12303191947001635,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009985572845436913,
      "loss": 1.0183,
      "step": 1430
    },
    {
      "epoch": 0.12311795577733804,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009985552311153176,
      "loss": 1.2159,
      "step": 1431
    },
    {
      "epoch": 0.12320399208465972,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009985531762287672,
      "loss": 1.2157,
      "step": 1432
    },
    {
      "epoch": 0.12329002839198142,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009985511198840453,
      "loss": 1.3878,
      "step": 1433
    },
    {
      "epoch": 0.12337606469930311,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009985490620811586,
      "loss": 1.4465,
      "step": 1434
    },
    {
      "epoch": 0.1234621010066248,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009985470028201129,
      "loss": 1.7453,
      "step": 1435
    },
    {
      "epoch": 0.12354813731394648,
      "grad_norm": 1.8515625,
      "learning_rate": 0.000998544942100914,
      "loss": 1.3219,
      "step": 1436
    },
    {
      "epoch": 0.12363417362126818,
      "grad_norm": 1.25,
      "learning_rate": 0.000998542879923568,
      "loss": 0.9931,
      "step": 1437
    },
    {
      "epoch": 0.12372020992858987,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009985408162880814,
      "loss": 2.4127,
      "step": 1438
    },
    {
      "epoch": 0.12380624623591155,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009985387511944597,
      "loss": 0.7978,
      "step": 1439
    },
    {
      "epoch": 0.12389228254323324,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000998536684642709,
      "loss": 1.7927,
      "step": 1440
    },
    {
      "epoch": 0.12397831885055494,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009985346166328356,
      "loss": 0.8126,
      "step": 1441
    },
    {
      "epoch": 0.12406435515787663,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009985325471648454,
      "loss": 1.4641,
      "step": 1442
    },
    {
      "epoch": 0.12415039146519831,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009985304762387445,
      "loss": 1.6093,
      "step": 1443
    },
    {
      "epoch": 0.12423642777252,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009985284038545388,
      "loss": 0.9455,
      "step": 1444
    },
    {
      "epoch": 0.1243224640798417,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009985263300122346,
      "loss": 1.1875,
      "step": 1445
    },
    {
      "epoch": 0.12440850038716339,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009985242547118378,
      "loss": 1.6885,
      "step": 1446
    },
    {
      "epoch": 0.12449453669448507,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009985221779533545,
      "loss": 1.3446,
      "step": 1447
    },
    {
      "epoch": 0.12458057300180676,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009985200997367908,
      "loss": 0.9146,
      "step": 1448
    },
    {
      "epoch": 0.12466660930912846,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009985180200621526,
      "loss": 1.3813,
      "step": 1449
    },
    {
      "epoch": 0.12475264561645014,
      "grad_norm": 1.375,
      "learning_rate": 0.0009985159389294463,
      "loss": 1.9938,
      "step": 1450
    },
    {
      "epoch": 0.12483868192377183,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009985138563386778,
      "loss": 1.3166,
      "step": 1451
    },
    {
      "epoch": 0.12492471823109352,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009985117722898533,
      "loss": 1.1672,
      "step": 1452
    },
    {
      "epoch": 0.1250107545384152,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009985096867829786,
      "loss": 1.359,
      "step": 1453
    },
    {
      "epoch": 0.1250967908457369,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009985075998180603,
      "loss": 1.1816,
      "step": 1454
    },
    {
      "epoch": 0.1251828271530586,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000998505511395104,
      "loss": 1.5342,
      "step": 1455
    },
    {
      "epoch": 0.12526886346038027,
      "grad_norm": 1.125,
      "learning_rate": 0.0009985034215141161,
      "loss": 1.7206,
      "step": 1456
    },
    {
      "epoch": 0.12535489976770198,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009985013301751027,
      "loss": 1.7599,
      "step": 1457
    },
    {
      "epoch": 0.12544093607502366,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009984992373780698,
      "loss": 1.6481,
      "step": 1458
    },
    {
      "epoch": 0.12552697238234534,
      "grad_norm": 1.265625,
      "learning_rate": 0.000998497143123024,
      "loss": 1.5096,
      "step": 1459
    },
    {
      "epoch": 0.12561300868966704,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009984950474099703,
      "loss": 1.7685,
      "step": 1460
    },
    {
      "epoch": 0.12569904499698872,
      "grad_norm": 1.28125,
      "learning_rate": 0.000998492950238916,
      "loss": 2.1127,
      "step": 1461
    },
    {
      "epoch": 0.12578508130431043,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009984908516098667,
      "loss": 1.3259,
      "step": 1462
    },
    {
      "epoch": 0.1258711176116321,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009984887515228284,
      "loss": 1.086,
      "step": 1463
    },
    {
      "epoch": 0.1259571539189538,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009984866499778077,
      "loss": 1.5488,
      "step": 1464
    },
    {
      "epoch": 0.1260431902262755,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009984845469748105,
      "loss": 1.115,
      "step": 1465
    },
    {
      "epoch": 0.12612922653359718,
      "grad_norm": 3.484375,
      "learning_rate": 0.0009984824425138428,
      "loss": 1.7635,
      "step": 1466
    },
    {
      "epoch": 0.12621526284091886,
      "grad_norm": 2.125,
      "learning_rate": 0.000998480336594911,
      "loss": 2.2673,
      "step": 1467
    },
    {
      "epoch": 0.12630129914824056,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009984782292180212,
      "loss": 1.5692,
      "step": 1468
    },
    {
      "epoch": 0.12638733545556224,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009984761203831794,
      "loss": 1.5151,
      "step": 1469
    },
    {
      "epoch": 0.12647337176288392,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000998474010090392,
      "loss": 1.1728,
      "step": 1470
    },
    {
      "epoch": 0.12655940807020563,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000998471898339665,
      "loss": 1.5292,
      "step": 1471
    },
    {
      "epoch": 0.1266454443775273,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009984697851310046,
      "loss": 1.8279,
      "step": 1472
    },
    {
      "epoch": 0.12673148068484902,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009984676704644171,
      "loss": 1.3453,
      "step": 1473
    },
    {
      "epoch": 0.1268175169921707,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009984655543399086,
      "loss": 1.3092,
      "step": 1474
    },
    {
      "epoch": 0.12690355329949238,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009984634367574852,
      "loss": 1.5862,
      "step": 1475
    },
    {
      "epoch": 0.12698958960681409,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009984613177171533,
      "loss": 1.8482,
      "step": 1476
    },
    {
      "epoch": 0.12707562591413576,
      "grad_norm": 1.515625,
      "learning_rate": 0.000998459197218919,
      "loss": 1.3302,
      "step": 1477
    },
    {
      "epoch": 0.12716166222145744,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009984570752627885,
      "loss": 1.5513,
      "step": 1478
    },
    {
      "epoch": 0.12724769852877915,
      "grad_norm": 1.484375,
      "learning_rate": 0.000998454951848768,
      "loss": 1.7706,
      "step": 1479
    },
    {
      "epoch": 0.12733373483610083,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009984528269768639,
      "loss": 1.1077,
      "step": 1480
    },
    {
      "epoch": 0.1274197711434225,
      "grad_norm": 1.21875,
      "learning_rate": 0.000998450700647082,
      "loss": 1.1401,
      "step": 1481
    },
    {
      "epoch": 0.12750580745074422,
      "grad_norm": 1.375,
      "learning_rate": 0.0009984485728594286,
      "loss": 1.8271,
      "step": 1482
    },
    {
      "epoch": 0.1275918437580659,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009984464436139102,
      "loss": 1.6798,
      "step": 1483
    },
    {
      "epoch": 0.1276778800653876,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000998444312910533,
      "loss": 1.7755,
      "step": 1484
    },
    {
      "epoch": 0.12776391637270929,
      "grad_norm": 1.21875,
      "learning_rate": 0.000998442180749303,
      "loss": 1.4217,
      "step": 1485
    },
    {
      "epoch": 0.12784995268003096,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009984400471302265,
      "loss": 1.1005,
      "step": 1486
    },
    {
      "epoch": 0.12793598898735267,
      "grad_norm": 1.453125,
      "learning_rate": 0.00099843791205331,
      "loss": 1.8282,
      "step": 1487
    },
    {
      "epoch": 0.12802202529467435,
      "grad_norm": 1.375,
      "learning_rate": 0.0009984357755185595,
      "loss": 0.9242,
      "step": 1488
    },
    {
      "epoch": 0.12810806160199603,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009984336375259812,
      "loss": 1.6562,
      "step": 1489
    },
    {
      "epoch": 0.12819409790931774,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009984314980755816,
      "loss": 1.9729,
      "step": 1490
    },
    {
      "epoch": 0.12828013421663942,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009984293571673668,
      "loss": 1.9747,
      "step": 1491
    },
    {
      "epoch": 0.1283661705239611,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009984272148013428,
      "loss": 1.2854,
      "step": 1492
    },
    {
      "epoch": 0.1284522068312828,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009984250709775164,
      "loss": 1.355,
      "step": 1493
    },
    {
      "epoch": 0.12853824313860449,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009984229256958935,
      "loss": 0.8431,
      "step": 1494
    },
    {
      "epoch": 0.1286242794459262,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009984207789564806,
      "loss": 1.9133,
      "step": 1495
    },
    {
      "epoch": 0.12871031575324787,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009984186307592838,
      "loss": 1.5042,
      "step": 1496
    },
    {
      "epoch": 0.12879635206056955,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009984164811043092,
      "loss": 1.3711,
      "step": 1497
    },
    {
      "epoch": 0.12888238836789126,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009984143299915636,
      "loss": 1.9561,
      "step": 1498
    },
    {
      "epoch": 0.12896842467521294,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000998412177421053,
      "loss": 1.3098,
      "step": 1499
    },
    {
      "epoch": 0.12905446098253462,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009984100233927837,
      "loss": 1.206,
      "step": 1500
    },
    {
      "epoch": 0.12914049728985633,
      "grad_norm": 1.359375,
      "learning_rate": 0.000998407867906762,
      "loss": 1.9399,
      "step": 1501
    },
    {
      "epoch": 0.129226533597178,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009984057109629942,
      "loss": 1.5069,
      "step": 1502
    },
    {
      "epoch": 0.12931256990449969,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009984035525614867,
      "loss": 1.7695,
      "step": 1503
    },
    {
      "epoch": 0.1293986062118214,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009984013927022458,
      "loss": 0.7692,
      "step": 1504
    },
    {
      "epoch": 0.12948464251914307,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009983992313852773,
      "loss": 1.1001,
      "step": 1505
    },
    {
      "epoch": 0.12957067882646478,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009983970686105886,
      "loss": 0.8935,
      "step": 1506
    },
    {
      "epoch": 0.12965671513378646,
      "grad_norm": 1.25,
      "learning_rate": 0.000998394904378185,
      "loss": 1.488,
      "step": 1507
    },
    {
      "epoch": 0.12974275144110814,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009983927386880735,
      "loss": 1.4059,
      "step": 1508
    },
    {
      "epoch": 0.12982878774842985,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009983905715402598,
      "loss": 1.3809,
      "step": 1509
    },
    {
      "epoch": 0.12991482405575153,
      "grad_norm": 1.21875,
      "learning_rate": 0.000998388402934751,
      "loss": 1.6784,
      "step": 1510
    },
    {
      "epoch": 0.1300008603630732,
      "grad_norm": 1.34375,
      "learning_rate": 0.000998386232871553,
      "loss": 1.1284,
      "step": 1511
    },
    {
      "epoch": 0.1300868966703949,
      "grad_norm": 1.1875,
      "learning_rate": 0.000998384061350672,
      "loss": 1.3728,
      "step": 1512
    },
    {
      "epoch": 0.1301729329777166,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009983818883721148,
      "loss": 1.0869,
      "step": 1513
    },
    {
      "epoch": 0.13025896928503827,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009983797139358872,
      "loss": 1.4066,
      "step": 1514
    },
    {
      "epoch": 0.13034500559235998,
      "grad_norm": 1.09375,
      "learning_rate": 0.000998377538041996,
      "loss": 0.8878,
      "step": 1515
    },
    {
      "epoch": 0.13043104189968166,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009983753606904475,
      "loss": 1.0777,
      "step": 1516
    },
    {
      "epoch": 0.13051707820700337,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000998373181881248,
      "loss": 1.6138,
      "step": 1517
    },
    {
      "epoch": 0.13060311451432505,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009983710016144038,
      "loss": 1.3575,
      "step": 1518
    },
    {
      "epoch": 0.13068915082164673,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009983688198899215,
      "loss": 1.4425,
      "step": 1519
    },
    {
      "epoch": 0.13077518712896843,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009983666367078072,
      "loss": 1.7049,
      "step": 1520
    },
    {
      "epoch": 0.13086122343629011,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009983644520680673,
      "loss": 1.03,
      "step": 1521
    },
    {
      "epoch": 0.1309472597436118,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009983622659707086,
      "loss": 2.112,
      "step": 1522
    },
    {
      "epoch": 0.1310332960509335,
      "grad_norm": 1.421875,
      "learning_rate": 0.000998360078415737,
      "loss": 2.2586,
      "step": 1523
    },
    {
      "epoch": 0.13111933235825518,
      "grad_norm": 1.8984375,
      "learning_rate": 0.000998357889403159,
      "loss": 1.1986,
      "step": 1524
    },
    {
      "epoch": 0.13120536866557686,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009983556989329812,
      "loss": 1.6712,
      "step": 1525
    },
    {
      "epoch": 0.13129140497289857,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00099835350700521,
      "loss": 1.3753,
      "step": 1526
    },
    {
      "epoch": 0.13137744128022025,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009983513136198516,
      "loss": 0.8339,
      "step": 1527
    },
    {
      "epoch": 0.13146347758754195,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009983491187769127,
      "loss": 1.4011,
      "step": 1528
    },
    {
      "epoch": 0.13154951389486363,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009983469224763993,
      "loss": 0.7849,
      "step": 1529
    },
    {
      "epoch": 0.13163555020218531,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009983447247183183,
      "loss": 1.5107,
      "step": 1530
    },
    {
      "epoch": 0.13172158650950702,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009983425255026756,
      "loss": 1.7711,
      "step": 1531
    },
    {
      "epoch": 0.1318076228168287,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009983403248294783,
      "loss": 2.1709,
      "step": 1532
    },
    {
      "epoch": 0.13189365912415038,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009983381226987323,
      "loss": 1.5809,
      "step": 1533
    },
    {
      "epoch": 0.1319796954314721,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009983359191104442,
      "loss": 1.6918,
      "step": 1534
    },
    {
      "epoch": 0.13206573173879377,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009983337140646205,
      "loss": 1.2356,
      "step": 1535
    },
    {
      "epoch": 0.13215176804611545,
      "grad_norm": 1.25,
      "learning_rate": 0.0009983315075612676,
      "loss": 1.6392,
      "step": 1536
    },
    {
      "epoch": 0.13223780435343716,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009983292996003919,
      "loss": 1.7192,
      "step": 1537
    },
    {
      "epoch": 0.13232384066075883,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009983270901820001,
      "loss": 1.4876,
      "step": 1538
    },
    {
      "epoch": 0.13240987696808054,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000998324879306098,
      "loss": 0.8795,
      "step": 1539
    },
    {
      "epoch": 0.13249591327540222,
      "grad_norm": 2.90625,
      "learning_rate": 0.000998322666972693,
      "loss": 1.9217,
      "step": 1540
    },
    {
      "epoch": 0.1325819495827239,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009983204531817912,
      "loss": 1.483,
      "step": 1541
    },
    {
      "epoch": 0.1326679858900456,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009983182379333989,
      "loss": 1.3806,
      "step": 1542
    },
    {
      "epoch": 0.1327540221973673,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009983160212275223,
      "loss": 1.8877,
      "step": 1543
    },
    {
      "epoch": 0.13284005850468897,
      "grad_norm": 1.25,
      "learning_rate": 0.0009983138030641686,
      "loss": 1.6088,
      "step": 1544
    },
    {
      "epoch": 0.13292609481201068,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000998311583443344,
      "loss": 1.0264,
      "step": 1545
    },
    {
      "epoch": 0.13301213111933236,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009983093623650548,
      "loss": 1.8134,
      "step": 1546
    },
    {
      "epoch": 0.13309816742665403,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009983071398293076,
      "loss": 1.049,
      "step": 1547
    },
    {
      "epoch": 0.13318420373397574,
      "grad_norm": 1.5625,
      "learning_rate": 0.000998304915836109,
      "loss": 1.7111,
      "step": 1548
    },
    {
      "epoch": 0.13327024004129742,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009983026903854653,
      "loss": 1.3012,
      "step": 1549
    },
    {
      "epoch": 0.13335627634861913,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009983004634773834,
      "loss": 1.6236,
      "step": 1550
    },
    {
      "epoch": 0.1334423126559408,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009982982351118693,
      "loss": 1.6842,
      "step": 1551
    },
    {
      "epoch": 0.1335283489632625,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009982960052889297,
      "loss": 0.9809,
      "step": 1552
    },
    {
      "epoch": 0.1336143852705842,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009982937740085716,
      "loss": 1.3728,
      "step": 1553
    },
    {
      "epoch": 0.13370042157790588,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009982915412708009,
      "loss": 1.7576,
      "step": 1554
    },
    {
      "epoch": 0.13378645788522756,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009982893070756243,
      "loss": 1.0546,
      "step": 1555
    },
    {
      "epoch": 0.13387249419254926,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009982870714230485,
      "loss": 1.9411,
      "step": 1556
    },
    {
      "epoch": 0.13395853049987094,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009982848343130799,
      "loss": 1.6076,
      "step": 1557
    },
    {
      "epoch": 0.13404456680719262,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000998282595745725,
      "loss": 1.485,
      "step": 1558
    },
    {
      "epoch": 0.13413060311451433,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009982803557209905,
      "loss": 1.3164,
      "step": 1559
    },
    {
      "epoch": 0.134216639421836,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000998278114238883,
      "loss": 1.2335,
      "step": 1560
    },
    {
      "epoch": 0.13430267572915772,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009982758712994086,
      "loss": 1.0619,
      "step": 1561
    },
    {
      "epoch": 0.1343887120364794,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009982736269025744,
      "loss": 0.836,
      "step": 1562
    },
    {
      "epoch": 0.13447474834380108,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009982713810483866,
      "loss": 1.2029,
      "step": 1563
    },
    {
      "epoch": 0.13456078465112278,
      "grad_norm": 1.3125,
      "learning_rate": 0.000998269133736852,
      "loss": 1.1002,
      "step": 1564
    },
    {
      "epoch": 0.13464682095844446,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009982668849679772,
      "loss": 1.6315,
      "step": 1565
    },
    {
      "epoch": 0.13473285726576614,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009982646347417685,
      "loss": 1.1606,
      "step": 1566
    },
    {
      "epoch": 0.13481889357308785,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009982623830582327,
      "loss": 1.1821,
      "step": 1567
    },
    {
      "epoch": 0.13490492988040953,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009982601299173763,
      "loss": 1.3859,
      "step": 1568
    },
    {
      "epoch": 0.1349909661877312,
      "grad_norm": 1.515625,
      "learning_rate": 0.000998257875319206,
      "loss": 1.483,
      "step": 1569
    },
    {
      "epoch": 0.13507700249505292,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009982556192637284,
      "loss": 1.3809,
      "step": 1570
    },
    {
      "epoch": 0.1351630388023746,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009982533617509499,
      "loss": 1.2356,
      "step": 1571
    },
    {
      "epoch": 0.1352490751096963,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009982511027808773,
      "loss": 1.3442,
      "step": 1572
    },
    {
      "epoch": 0.13533511141701798,
      "grad_norm": 1.375,
      "learning_rate": 0.000998248842353517,
      "loss": 1.2232,
      "step": 1573
    },
    {
      "epoch": 0.13542114772433966,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009982465804688758,
      "loss": 1.694,
      "step": 1574
    },
    {
      "epoch": 0.13550718403166137,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009982443171269605,
      "loss": 1.3593,
      "step": 1575
    },
    {
      "epoch": 0.13559322033898305,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009982420523277772,
      "loss": 1.8693,
      "step": 1576
    },
    {
      "epoch": 0.13567925664630473,
      "grad_norm": 1.25,
      "learning_rate": 0.0009982397860713327,
      "loss": 1.4236,
      "step": 1577
    },
    {
      "epoch": 0.13576529295362644,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000998237518357634,
      "loss": 1.0875,
      "step": 1578
    },
    {
      "epoch": 0.13585132926094812,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009982352491866872,
      "loss": 1.1554,
      "step": 1579
    },
    {
      "epoch": 0.1359373655682698,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009982329785584994,
      "loss": 1.191,
      "step": 1580
    },
    {
      "epoch": 0.1360234018755915,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000998230706473077,
      "loss": 1.2618,
      "step": 1581
    },
    {
      "epoch": 0.13610943818291318,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009982284329304266,
      "loss": 1.6058,
      "step": 1582
    },
    {
      "epoch": 0.1361954744902349,
      "grad_norm": 1.140625,
      "learning_rate": 0.000998226157930555,
      "loss": 1.4485,
      "step": 1583
    },
    {
      "epoch": 0.13628151079755657,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009982238814734688,
      "loss": 1.0692,
      "step": 1584
    },
    {
      "epoch": 0.13636754710487825,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009982216035591747,
      "loss": 1.651,
      "step": 1585
    },
    {
      "epoch": 0.13645358341219996,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009982193241876791,
      "loss": 1.1525,
      "step": 1586
    },
    {
      "epoch": 0.13653961971952164,
      "grad_norm": 1.5,
      "learning_rate": 0.0009982170433589892,
      "loss": 1.377,
      "step": 1587
    },
    {
      "epoch": 0.13662565602684332,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000998214761073111,
      "loss": 1.1231,
      "step": 1588
    },
    {
      "epoch": 0.13671169233416502,
      "grad_norm": 1.5,
      "learning_rate": 0.0009982124773300516,
      "loss": 1.2321,
      "step": 1589
    },
    {
      "epoch": 0.1367977286414867,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009982101921298177,
      "loss": 2.3728,
      "step": 1590
    },
    {
      "epoch": 0.13688376494880838,
      "grad_norm": 1.421875,
      "learning_rate": 0.000998207905472416,
      "loss": 2.0569,
      "step": 1591
    },
    {
      "epoch": 0.1369698012561301,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009982056173578528,
      "loss": 1.7515,
      "step": 1592
    },
    {
      "epoch": 0.13705583756345177,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009982033277861353,
      "loss": 1.0448,
      "step": 1593
    },
    {
      "epoch": 0.13714187387077348,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009982010367572697,
      "loss": 1.701,
      "step": 1594
    },
    {
      "epoch": 0.13722791017809516,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009981987442712632,
      "loss": 0.772,
      "step": 1595
    },
    {
      "epoch": 0.13731394648541684,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009981964503281223,
      "loss": 2.1812,
      "step": 1596
    },
    {
      "epoch": 0.13739998279273855,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009981941549278535,
      "loss": 0.8333,
      "step": 1597
    },
    {
      "epoch": 0.13748601910006022,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000998191858070464,
      "loss": 1.5959,
      "step": 1598
    },
    {
      "epoch": 0.1375720554073819,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00099818955975596,
      "loss": 1.1409,
      "step": 1599
    },
    {
      "epoch": 0.1376580917147036,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009981872599843483,
      "loss": 1.5824,
      "step": 1600
    },
    {
      "epoch": 0.1377441280220253,
      "grad_norm": 1.125,
      "learning_rate": 0.0009981849587556357,
      "loss": 1.5802,
      "step": 1601
    },
    {
      "epoch": 0.13783016432934697,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009981826560698291,
      "loss": 1.4013,
      "step": 1602
    },
    {
      "epoch": 0.13791620063666868,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009981803519269353,
      "loss": 1.8821,
      "step": 1603
    },
    {
      "epoch": 0.13800223694399036,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009981780463269607,
      "loss": 1.6271,
      "step": 1604
    },
    {
      "epoch": 0.13808827325131207,
      "grad_norm": 1.25,
      "learning_rate": 0.0009981757392699122,
      "loss": 2.0042,
      "step": 1605
    },
    {
      "epoch": 0.13817430955863375,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009981734307557967,
      "loss": 1.0301,
      "step": 1606
    },
    {
      "epoch": 0.13826034586595543,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009981711207846205,
      "loss": 1.1308,
      "step": 1607
    },
    {
      "epoch": 0.13834638217327713,
      "grad_norm": 1.21875,
      "learning_rate": 0.000998168809356391,
      "loss": 0.9595,
      "step": 1608
    },
    {
      "epoch": 0.1384324184805988,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009981664964711144,
      "loss": 1.3446,
      "step": 1609
    },
    {
      "epoch": 0.1385184547879205,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009981641821287977,
      "loss": 1.3676,
      "step": 1610
    },
    {
      "epoch": 0.1386044910952422,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009981618663294476,
      "loss": 1.2336,
      "step": 1611
    },
    {
      "epoch": 0.13869052740256388,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000998159549073071,
      "loss": 1.9225,
      "step": 1612
    },
    {
      "epoch": 0.13877656370988556,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009981572303596749,
      "loss": 1.2351,
      "step": 1613
    },
    {
      "epoch": 0.13886260001720727,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009981549101892653,
      "loss": 1.4236,
      "step": 1614
    },
    {
      "epoch": 0.13894863632452895,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009981525885618498,
      "loss": 1.639,
      "step": 1615
    },
    {
      "epoch": 0.13903467263185065,
      "grad_norm": 1.375,
      "learning_rate": 0.000998150265477435,
      "loss": 1.1642,
      "step": 1616
    },
    {
      "epoch": 0.13912070893917233,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000998147940936027,
      "loss": 0.9545,
      "step": 1617
    },
    {
      "epoch": 0.139206745246494,
      "grad_norm": 1.625,
      "learning_rate": 0.0009981456149376336,
      "loss": 1.7154,
      "step": 1618
    },
    {
      "epoch": 0.13929278155381572,
      "grad_norm": 1.390625,
      "learning_rate": 0.000998143287482261,
      "loss": 1.1884,
      "step": 1619
    },
    {
      "epoch": 0.1393788178611374,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009981409585699162,
      "loss": 1.474,
      "step": 1620
    },
    {
      "epoch": 0.13946485416845908,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009981386282006059,
      "loss": 1.7714,
      "step": 1621
    },
    {
      "epoch": 0.1395508904757808,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009981362963743371,
      "loss": 2.1007,
      "step": 1622
    },
    {
      "epoch": 0.13963692678310247,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009981339630911166,
      "loss": 1.5145,
      "step": 1623
    },
    {
      "epoch": 0.13972296309042415,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009981316283509509,
      "loss": 1.6894,
      "step": 1624
    },
    {
      "epoch": 0.13980899939774585,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009981292921538475,
      "loss": 1.4648,
      "step": 1625
    },
    {
      "epoch": 0.13989503570506753,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009981269544998123,
      "loss": 1.6359,
      "step": 1626
    },
    {
      "epoch": 0.13998107201238924,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000998124615388853,
      "loss": 1.6081,
      "step": 1627
    },
    {
      "epoch": 0.14006710831971092,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009981222748209758,
      "loss": 1.4935,
      "step": 1628
    },
    {
      "epoch": 0.1401531446270326,
      "grad_norm": 1.015625,
      "learning_rate": 0.000998119932796188,
      "loss": 0.8754,
      "step": 1629
    },
    {
      "epoch": 0.1402391809343543,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009981175893144963,
      "loss": 1.2607,
      "step": 1630
    },
    {
      "epoch": 0.140325217241676,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009981152443759075,
      "loss": 1.7246,
      "step": 1631
    },
    {
      "epoch": 0.14041125354899767,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009981128979804283,
      "loss": 0.9665,
      "step": 1632
    },
    {
      "epoch": 0.14049728985631937,
      "grad_norm": 1.140625,
      "learning_rate": 0.000998110550128066,
      "loss": 0.879,
      "step": 1633
    },
    {
      "epoch": 0.14058332616364105,
      "grad_norm": 1.140625,
      "learning_rate": 0.000998108200818827,
      "loss": 1.0376,
      "step": 1634
    },
    {
      "epoch": 0.14066936247096273,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009981058500527185,
      "loss": 1.6462,
      "step": 1635
    },
    {
      "epoch": 0.14075539877828444,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009981034978297472,
      "loss": 1.3066,
      "step": 1636
    },
    {
      "epoch": 0.14084143508560612,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009981011441499202,
      "loss": 1.0987,
      "step": 1637
    },
    {
      "epoch": 0.14092747139292783,
      "grad_norm": 1.9375,
      "learning_rate": 0.000998098789013244,
      "loss": 1.2879,
      "step": 1638
    },
    {
      "epoch": 0.1410135077002495,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009980964324197259,
      "loss": 1.7429,
      "step": 1639
    },
    {
      "epoch": 0.1410995440075712,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009980940743693728,
      "loss": 1.2885,
      "step": 1640
    },
    {
      "epoch": 0.1411855803148929,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009980917148621911,
      "loss": 1.3778,
      "step": 1641
    },
    {
      "epoch": 0.14127161662221457,
      "grad_norm": 1.5,
      "learning_rate": 0.000998089353898188,
      "loss": 1.3035,
      "step": 1642
    },
    {
      "epoch": 0.14135765292953625,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009980869914773705,
      "loss": 1.1189,
      "step": 1643
    },
    {
      "epoch": 0.14144368923685796,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009980846275997456,
      "loss": 1.0748,
      "step": 1644
    },
    {
      "epoch": 0.14152972554417964,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009980822622653197,
      "loss": 2.1774,
      "step": 1645
    },
    {
      "epoch": 0.14161576185150132,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009980798954741,
      "loss": 1.1057,
      "step": 1646
    },
    {
      "epoch": 0.14170179815882303,
      "grad_norm": 3.46875,
      "learning_rate": 0.0009980775272260937,
      "loss": 1.3673,
      "step": 1647
    },
    {
      "epoch": 0.1417878344661447,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009980751575213076,
      "loss": 0.7612,
      "step": 1648
    },
    {
      "epoch": 0.14187387077346642,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009980727863597486,
      "loss": 1.845,
      "step": 1649
    },
    {
      "epoch": 0.1419599070807881,
      "grad_norm": 2.0,
      "learning_rate": 0.0009980704137414233,
      "loss": 1.7182,
      "step": 1650
    },
    {
      "epoch": 0.14204594338810977,
      "grad_norm": 1.25,
      "learning_rate": 0.000998068039666339,
      "loss": 2.014,
      "step": 1651
    },
    {
      "epoch": 0.14213197969543148,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009980656641345024,
      "loss": 1.0388,
      "step": 1652
    },
    {
      "epoch": 0.14221801600275316,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000998063287145921,
      "loss": 1.8793,
      "step": 1653
    },
    {
      "epoch": 0.14230405231007484,
      "grad_norm": 1.375,
      "learning_rate": 0.000998060908700601,
      "loss": 1.3153,
      "step": 1654
    },
    {
      "epoch": 0.14239008861739655,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009980585287985497,
      "loss": 0.9495,
      "step": 1655
    },
    {
      "epoch": 0.14247612492471823,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009980561474397741,
      "loss": 1.218,
      "step": 1656
    },
    {
      "epoch": 0.1425621612320399,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009980537646242815,
      "loss": 1.2289,
      "step": 1657
    },
    {
      "epoch": 0.14264819753936162,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009980513803520782,
      "loss": 1.3964,
      "step": 1658
    },
    {
      "epoch": 0.1427342338466833,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009980489946231714,
      "loss": 1.4267,
      "step": 1659
    },
    {
      "epoch": 0.142820270154005,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009980466074375682,
      "loss": 0.8117,
      "step": 1660
    },
    {
      "epoch": 0.14290630646132668,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009980442187952757,
      "loss": 1.881,
      "step": 1661
    },
    {
      "epoch": 0.14299234276864836,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009980418286963007,
      "loss": 1.1531,
      "step": 1662
    },
    {
      "epoch": 0.14307837907597007,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009980394371406501,
      "loss": 0.8845,
      "step": 1663
    },
    {
      "epoch": 0.14316441538329175,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000998037044128331,
      "loss": 0.7351,
      "step": 1664
    },
    {
      "epoch": 0.14325045169061343,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009980346496593505,
      "loss": 1.8099,
      "step": 1665
    },
    {
      "epoch": 0.14333648799793514,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009980322537337155,
      "loss": 2.2441,
      "step": 1666
    },
    {
      "epoch": 0.14342252430525682,
      "grad_norm": 1.421875,
      "learning_rate": 0.000998029856351433,
      "loss": 1.5922,
      "step": 1667
    },
    {
      "epoch": 0.1435085606125785,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00099802745751251,
      "loss": 1.7228,
      "step": 1668
    },
    {
      "epoch": 0.1435945969199002,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009980250572169537,
      "loss": 1.1559,
      "step": 1669
    },
    {
      "epoch": 0.14368063322722188,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009980226554647707,
      "loss": 1.0663,
      "step": 1670
    },
    {
      "epoch": 0.1437666695345436,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009980202522559686,
      "loss": 1.4044,
      "step": 1671
    },
    {
      "epoch": 0.14385270584186527,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0009980178475905536,
      "loss": 0.5042,
      "step": 1672
    },
    {
      "epoch": 0.14393874214918695,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000998015441468534,
      "loss": 1.133,
      "step": 1673
    },
    {
      "epoch": 0.14402477845650866,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009980130338899155,
      "loss": 2.1199,
      "step": 1674
    },
    {
      "epoch": 0.14411081476383034,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009980106248547059,
      "loss": 1.4934,
      "step": 1675
    },
    {
      "epoch": 0.14419685107115202,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000998008214362912,
      "loss": 1.2642,
      "step": 1676
    },
    {
      "epoch": 0.14428288737847372,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000998005802414541,
      "loss": 1.2448,
      "step": 1677
    },
    {
      "epoch": 0.1443689236857954,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009980033890096,
      "loss": 1.2656,
      "step": 1678
    },
    {
      "epoch": 0.14445495999311708,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000998000974148096,
      "loss": 1.2408,
      "step": 1679
    },
    {
      "epoch": 0.1445409963004388,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000997998557830036,
      "loss": 1.3917,
      "step": 1680
    },
    {
      "epoch": 0.14462703260776047,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009979961400554268,
      "loss": 1.084,
      "step": 1681
    },
    {
      "epoch": 0.14471306891508218,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009979937208242759,
      "loss": 2.4319,
      "step": 1682
    },
    {
      "epoch": 0.14479910522240386,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009979913001365902,
      "loss": 1.6233,
      "step": 1683
    },
    {
      "epoch": 0.14488514152972554,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009979888779923767,
      "loss": 1.2527,
      "step": 1684
    },
    {
      "epoch": 0.14497117783704724,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009979864543916428,
      "loss": 1.0357,
      "step": 1685
    },
    {
      "epoch": 0.14505721414436892,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009979840293343953,
      "loss": 1.7025,
      "step": 1686
    },
    {
      "epoch": 0.1451432504516906,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009979816028206415,
      "loss": 1.4771,
      "step": 1687
    },
    {
      "epoch": 0.1452292867590123,
      "grad_norm": 2.328125,
      "learning_rate": 0.000997979174850388,
      "loss": 1.1558,
      "step": 1688
    },
    {
      "epoch": 0.145315323066334,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009979767454236425,
      "loss": 1.5274,
      "step": 1689
    },
    {
      "epoch": 0.14540135937365567,
      "grad_norm": 1.375,
      "learning_rate": 0.000997974314540412,
      "loss": 1.2719,
      "step": 1690
    },
    {
      "epoch": 0.14548739568097738,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009979718822007032,
      "loss": 1.6175,
      "step": 1691
    },
    {
      "epoch": 0.14557343198829906,
      "grad_norm": 1.125,
      "learning_rate": 0.0009979694484045236,
      "loss": 1.3706,
      "step": 1692
    },
    {
      "epoch": 0.14565946829562076,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009979670131518804,
      "loss": 1.1081,
      "step": 1693
    },
    {
      "epoch": 0.14574550460294244,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009979645764427803,
      "loss": 1.9733,
      "step": 1694
    },
    {
      "epoch": 0.14583154091026412,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009979621382772306,
      "loss": 1.1232,
      "step": 1695
    },
    {
      "epoch": 0.14591757721758583,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009979596986552386,
      "loss": 1.3958,
      "step": 1696
    },
    {
      "epoch": 0.1460036135249075,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009979572575768115,
      "loss": 1.7536,
      "step": 1697
    },
    {
      "epoch": 0.1460896498322292,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009979548150419561,
      "loss": 0.9311,
      "step": 1698
    },
    {
      "epoch": 0.1461756861395509,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009979523710506796,
      "loss": 1.7778,
      "step": 1699
    },
    {
      "epoch": 0.14626172244687258,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009979499256029894,
      "loss": 1.3333,
      "step": 1700
    },
    {
      "epoch": 0.14634775875419426,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009979474786988925,
      "loss": 1.4796,
      "step": 1701
    },
    {
      "epoch": 0.14643379506151596,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009979450303383962,
      "loss": 2.1955,
      "step": 1702
    },
    {
      "epoch": 0.14651983136883764,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009979425805215075,
      "loss": 1.6429,
      "step": 1703
    },
    {
      "epoch": 0.14660586767615935,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009979401292482333,
      "loss": 1.0159,
      "step": 1704
    },
    {
      "epoch": 0.14669190398348103,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009979376765185813,
      "loss": 1.3401,
      "step": 1705
    },
    {
      "epoch": 0.1467779402908027,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009979352223325585,
      "loss": 0.8146,
      "step": 1706
    },
    {
      "epoch": 0.14686397659812442,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009979327666901718,
      "loss": 2.0695,
      "step": 1707
    },
    {
      "epoch": 0.1469500129054461,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009979303095914288,
      "loss": 1.0387,
      "step": 1708
    },
    {
      "epoch": 0.14703604921276778,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009979278510363365,
      "loss": 1.3902,
      "step": 1709
    },
    {
      "epoch": 0.14712208552008949,
      "grad_norm": 1.828125,
      "learning_rate": 0.000997925391024902,
      "loss": 1.0336,
      "step": 1710
    },
    {
      "epoch": 0.14720812182741116,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009979229295571327,
      "loss": 1.6649,
      "step": 1711
    },
    {
      "epoch": 0.14729415813473284,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009979204666330354,
      "loss": 1.1091,
      "step": 1712
    },
    {
      "epoch": 0.14738019444205455,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009979180022526178,
      "loss": 1.6779,
      "step": 1713
    },
    {
      "epoch": 0.14746623074937623,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009979155364158868,
      "loss": 1.5022,
      "step": 1714
    },
    {
      "epoch": 0.14755226705669794,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009979130691228496,
      "loss": 2.2883,
      "step": 1715
    },
    {
      "epoch": 0.14763830336401962,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009979106003735136,
      "loss": 0.8314,
      "step": 1716
    },
    {
      "epoch": 0.1477243396713413,
      "grad_norm": 2.078125,
      "learning_rate": 0.000997908130167886,
      "loss": 1.5718,
      "step": 1717
    },
    {
      "epoch": 0.147810375978663,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009979056585059738,
      "loss": 0.9713,
      "step": 1718
    },
    {
      "epoch": 0.14789641228598469,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009979031853877844,
      "loss": 1.5393,
      "step": 1719
    },
    {
      "epoch": 0.14798244859330636,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009979007108133252,
      "loss": 1.4913,
      "step": 1720
    },
    {
      "epoch": 0.14806848490062807,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009978982347826032,
      "loss": 1.7553,
      "step": 1721
    },
    {
      "epoch": 0.14815452120794975,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009978957572956255,
      "loss": 0.889,
      "step": 1722
    },
    {
      "epoch": 0.14824055751527143,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009978932783523996,
      "loss": 1.4645,
      "step": 1723
    },
    {
      "epoch": 0.14832659382259314,
      "grad_norm": 1.515625,
      "learning_rate": 0.000997890797952933,
      "loss": 1.4321,
      "step": 1724
    },
    {
      "epoch": 0.14841263012991482,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009978883160972325,
      "loss": 0.7416,
      "step": 1725
    },
    {
      "epoch": 0.14849866643723653,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009978858327853053,
      "loss": 1.3768,
      "step": 1726
    },
    {
      "epoch": 0.1485847027445582,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009978833480171591,
      "loss": 1.7149,
      "step": 1727
    },
    {
      "epoch": 0.14867073905187989,
      "grad_norm": 1.703125,
      "learning_rate": 0.000997880861792801,
      "loss": 1.4462,
      "step": 1728
    },
    {
      "epoch": 0.1487567753592016,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000997878374112238,
      "loss": 1.7252,
      "step": 1729
    },
    {
      "epoch": 0.14884281166652327,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009978758849754777,
      "loss": 2.2146,
      "step": 1730
    },
    {
      "epoch": 0.14892884797384495,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009978733943825273,
      "loss": 1.6106,
      "step": 1731
    },
    {
      "epoch": 0.14901488428116666,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009978709023333939,
      "loss": 1.9277,
      "step": 1732
    },
    {
      "epoch": 0.14910092058848834,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000997868408828085,
      "loss": 1.2802,
      "step": 1733
    },
    {
      "epoch": 0.14918695689581002,
      "grad_norm": 1.59375,
      "learning_rate": 0.000997865913866608,
      "loss": 1.6125,
      "step": 1734
    },
    {
      "epoch": 0.14927299320313173,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009978634174489698,
      "loss": 1.9661,
      "step": 1735
    },
    {
      "epoch": 0.1493590295104534,
      "grad_norm": 1.375,
      "learning_rate": 0.000997860919575178,
      "loss": 1.9978,
      "step": 1736
    },
    {
      "epoch": 0.1494450658177751,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009978584202452398,
      "loss": 1.3192,
      "step": 1737
    },
    {
      "epoch": 0.1495311021250968,
      "grad_norm": 1.625,
      "learning_rate": 0.0009978559194591629,
      "loss": 1.1453,
      "step": 1738
    },
    {
      "epoch": 0.14961713843241847,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009978534172169538,
      "loss": 1.8613,
      "step": 1739
    },
    {
      "epoch": 0.14970317473974018,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009978509135186206,
      "loss": 1.9615,
      "step": 1740
    },
    {
      "epoch": 0.14978921104706186,
      "grad_norm": 1.25,
      "learning_rate": 0.00099784840836417,
      "loss": 1.3213,
      "step": 1741
    },
    {
      "epoch": 0.14987524735438354,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009978459017536097,
      "loss": 1.0429,
      "step": 1742
    },
    {
      "epoch": 0.14996128366170525,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009978433936869472,
      "loss": 2.089,
      "step": 1743
    },
    {
      "epoch": 0.15004731996902693,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009978408841641894,
      "loss": 1.0497,
      "step": 1744
    },
    {
      "epoch": 0.1501333562763486,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000997838373185344,
      "loss": 1.4791,
      "step": 1745
    },
    {
      "epoch": 0.1502193925836703,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000997835860750418,
      "loss": 1.07,
      "step": 1746
    },
    {
      "epoch": 0.150305428890992,
      "grad_norm": 1.21875,
      "learning_rate": 0.000997833346859419,
      "loss": 1.1226,
      "step": 1747
    },
    {
      "epoch": 0.1503914651983137,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009978308315123543,
      "loss": 1.3806,
      "step": 1748
    },
    {
      "epoch": 0.15047750150563538,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009978283147092312,
      "loss": 1.5345,
      "step": 1749
    },
    {
      "epoch": 0.15056353781295706,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009978257964500573,
      "loss": 1.4883,
      "step": 1750
    },
    {
      "epoch": 0.15064957412027877,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009978232767348394,
      "loss": 1.4244,
      "step": 1751
    },
    {
      "epoch": 0.15073561042760045,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009978207555635856,
      "loss": 1.1262,
      "step": 1752
    },
    {
      "epoch": 0.15082164673492213,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009978182329363028,
      "loss": 1.2291,
      "step": 1753
    },
    {
      "epoch": 0.15090768304224383,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009978157088529984,
      "loss": 1.4008,
      "step": 1754
    },
    {
      "epoch": 0.15099371934956551,
      "grad_norm": 1.375,
      "learning_rate": 0.00099781318331368,
      "loss": 1.7183,
      "step": 1755
    },
    {
      "epoch": 0.1510797556568872,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009978106563183547,
      "loss": 1.5949,
      "step": 1756
    },
    {
      "epoch": 0.1511657919642089,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00099780812786703,
      "loss": 1.2686,
      "step": 1757
    },
    {
      "epoch": 0.15125182827153058,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009978055979597135,
      "loss": 1.7782,
      "step": 1758
    },
    {
      "epoch": 0.1513378645788523,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009978030665964124,
      "loss": 1.5718,
      "step": 1759
    },
    {
      "epoch": 0.15142390088617397,
      "grad_norm": 1.125,
      "learning_rate": 0.0009978005337771341,
      "loss": 1.1411,
      "step": 1760
    },
    {
      "epoch": 0.15150993719349565,
      "grad_norm": 1.375,
      "learning_rate": 0.0009977979995018861,
      "loss": 1.7056,
      "step": 1761
    },
    {
      "epoch": 0.15159597350081735,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009977954637706758,
      "loss": 0.982,
      "step": 1762
    },
    {
      "epoch": 0.15168200980813903,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009977929265835106,
      "loss": 1.2547,
      "step": 1763
    },
    {
      "epoch": 0.15176804611546071,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009977903879403978,
      "loss": 1.3317,
      "step": 1764
    },
    {
      "epoch": 0.15185408242278242,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000997787847841345,
      "loss": 0.9516,
      "step": 1765
    },
    {
      "epoch": 0.1519401187301041,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009977853062863597,
      "loss": 1.3662,
      "step": 1766
    },
    {
      "epoch": 0.15202615503742578,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000997782763275449,
      "loss": 0.9882,
      "step": 1767
    },
    {
      "epoch": 0.1521121913447475,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009977802188086205,
      "loss": 1.3269,
      "step": 1768
    },
    {
      "epoch": 0.15219822765206917,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009977776728858818,
      "loss": 1.2513,
      "step": 1769
    },
    {
      "epoch": 0.15228426395939088,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009977751255072402,
      "loss": 1.0797,
      "step": 1770
    },
    {
      "epoch": 0.15237030026671256,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009977725766727031,
      "loss": 1.809,
      "step": 1771
    },
    {
      "epoch": 0.15245633657403423,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009977700263822781,
      "loss": 1.1813,
      "step": 1772
    },
    {
      "epoch": 0.15254237288135594,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009977674746359726,
      "loss": 1.3378,
      "step": 1773
    },
    {
      "epoch": 0.15262840918867762,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000997764921433794,
      "loss": 1.4918,
      "step": 1774
    },
    {
      "epoch": 0.1527144454959993,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009977623667757499,
      "loss": 1.7507,
      "step": 1775
    },
    {
      "epoch": 0.152800481803321,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009977598106618475,
      "loss": 1.1967,
      "step": 1776
    },
    {
      "epoch": 0.1528865181106427,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009977572530920945,
      "loss": 1.2906,
      "step": 1777
    },
    {
      "epoch": 0.15297255441796437,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009977546940664985,
      "loss": 1.1539,
      "step": 1778
    },
    {
      "epoch": 0.15305859072528608,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009977521335850666,
      "loss": 1.3004,
      "step": 1779
    },
    {
      "epoch": 0.15314462703260776,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009977495716478066,
      "loss": 0.9832,
      "step": 1780
    },
    {
      "epoch": 0.15323066333992946,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009977470082547261,
      "loss": 2.0399,
      "step": 1781
    },
    {
      "epoch": 0.15331669964725114,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009977444434058323,
      "loss": 1.6592,
      "step": 1782
    },
    {
      "epoch": 0.15340273595457282,
      "grad_norm": 1.5,
      "learning_rate": 0.0009977418771011327,
      "loss": 1.8331,
      "step": 1783
    },
    {
      "epoch": 0.15348877226189453,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000997739309340635,
      "loss": 1.8829,
      "step": 1784
    },
    {
      "epoch": 0.1535748085692162,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009977367401243467,
      "loss": 1.0861,
      "step": 1785
    },
    {
      "epoch": 0.1536608448765379,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000997734169452275,
      "loss": 1.2167,
      "step": 1786
    },
    {
      "epoch": 0.1537468811838596,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000997731597324428,
      "loss": 1.4058,
      "step": 1787
    },
    {
      "epoch": 0.15383291749118128,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009977290237408126,
      "loss": 1.8137,
      "step": 1788
    },
    {
      "epoch": 0.15391895379850296,
      "grad_norm": 1.21875,
      "learning_rate": 0.000997726448701437,
      "loss": 1.3482,
      "step": 1789
    },
    {
      "epoch": 0.15400499010582466,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009977238722063079,
      "loss": 1.3961,
      "step": 1790
    },
    {
      "epoch": 0.15409102641314634,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009977212942554332,
      "loss": 1.181,
      "step": 1791
    },
    {
      "epoch": 0.15417706272046805,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009977187148488208,
      "loss": 1.9875,
      "step": 1792
    },
    {
      "epoch": 0.15426309902778973,
      "grad_norm": 1.921875,
      "learning_rate": 0.000997716133986478,
      "loss": 1.113,
      "step": 1793
    },
    {
      "epoch": 0.1543491353351114,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009977135516684122,
      "loss": 1.5075,
      "step": 1794
    },
    {
      "epoch": 0.15443517164243312,
      "grad_norm": 1.28125,
      "learning_rate": 0.000997710967894631,
      "loss": 1.7974,
      "step": 1795
    },
    {
      "epoch": 0.1545212079497548,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009977083826651421,
      "loss": 1.6966,
      "step": 1796
    },
    {
      "epoch": 0.15460724425707648,
      "grad_norm": 1.484375,
      "learning_rate": 0.000997705795979953,
      "loss": 1.2484,
      "step": 1797
    },
    {
      "epoch": 0.15469328056439818,
      "grad_norm": 1.21875,
      "learning_rate": 0.000997703207839071,
      "loss": 0.9449,
      "step": 1798
    },
    {
      "epoch": 0.15477931687171986,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009977006182425043,
      "loss": 1.284,
      "step": 1799
    },
    {
      "epoch": 0.15486535317904154,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009976980271902598,
      "loss": 1.8955,
      "step": 1800
    },
    {
      "epoch": 0.15495138948636325,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009976954346823455,
      "loss": 1.1196,
      "step": 1801
    },
    {
      "epoch": 0.15503742579368493,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009976928407187688,
      "loss": 1.2529,
      "step": 1802
    },
    {
      "epoch": 0.15512346210100664,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009976902452995375,
      "loss": 1.2711,
      "step": 1803
    },
    {
      "epoch": 0.15520949840832832,
      "grad_norm": 1.328125,
      "learning_rate": 0.000997687648424659,
      "loss": 1.3767,
      "step": 1804
    },
    {
      "epoch": 0.15529553471565,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009976850500941407,
      "loss": 1.6828,
      "step": 1805
    },
    {
      "epoch": 0.1553815710229717,
      "grad_norm": 1.25,
      "learning_rate": 0.0009976824503079905,
      "loss": 1.1881,
      "step": 1806
    },
    {
      "epoch": 0.15546760733029338,
      "grad_norm": 1.1875,
      "learning_rate": 0.000997679849066216,
      "loss": 1.2981,
      "step": 1807
    },
    {
      "epoch": 0.15555364363761506,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009976772463688246,
      "loss": 2.0438,
      "step": 1808
    },
    {
      "epoch": 0.15563967994493677,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009976746422158242,
      "loss": 1.0689,
      "step": 1809
    },
    {
      "epoch": 0.15572571625225845,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009976720366072223,
      "loss": 1.7139,
      "step": 1810
    },
    {
      "epoch": 0.15581175255958013,
      "grad_norm": 1.5,
      "learning_rate": 0.0009976694295430264,
      "loss": 1.2083,
      "step": 1811
    },
    {
      "epoch": 0.15589778886690184,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009976668210232443,
      "loss": 1.1901,
      "step": 1812
    },
    {
      "epoch": 0.15598382517422352,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009976642110478833,
      "loss": 1.3416,
      "step": 1813
    },
    {
      "epoch": 0.15606986148154522,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009976615996169515,
      "loss": 1.7047,
      "step": 1814
    },
    {
      "epoch": 0.1561558977888669,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009976589867304562,
      "loss": 2.2313,
      "step": 1815
    },
    {
      "epoch": 0.15624193409618858,
      "grad_norm": 1.8203125,
      "learning_rate": 0.000997656372388405,
      "loss": 1.9282,
      "step": 1816
    },
    {
      "epoch": 0.1563279704035103,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000997653756590806,
      "loss": 1.1043,
      "step": 1817
    },
    {
      "epoch": 0.15641400671083197,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009976511393376664,
      "loss": 0.8205,
      "step": 1818
    },
    {
      "epoch": 0.15650004301815365,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009976485206289942,
      "loss": 1.1783,
      "step": 1819
    },
    {
      "epoch": 0.15658607932547536,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009976459004647966,
      "loss": 1.6913,
      "step": 1820
    },
    {
      "epoch": 0.15667211563279704,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009976432788450817,
      "loss": 1.4664,
      "step": 1821
    },
    {
      "epoch": 0.15675815194011872,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000997640655769857,
      "loss": 0.9538,
      "step": 1822
    },
    {
      "epoch": 0.15684418824744042,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00099763803123913,
      "loss": 1.0227,
      "step": 1823
    },
    {
      "epoch": 0.1569302245547621,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009976354052529087,
      "loss": 1.1104,
      "step": 1824
    },
    {
      "epoch": 0.1570162608620838,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009976327778112005,
      "loss": 1.8995,
      "step": 1825
    },
    {
      "epoch": 0.1571022971694055,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009976301489140133,
      "loss": 1.0882,
      "step": 1826
    },
    {
      "epoch": 0.15718833347672717,
      "grad_norm": 1.125,
      "learning_rate": 0.0009976275185613545,
      "loss": 1.1399,
      "step": 1827
    },
    {
      "epoch": 0.15727436978404888,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009976248867532322,
      "loss": 1.3449,
      "step": 1828
    },
    {
      "epoch": 0.15736040609137056,
      "grad_norm": 1.421875,
      "learning_rate": 0.000997622253489654,
      "loss": 0.8907,
      "step": 1829
    },
    {
      "epoch": 0.15744644239869224,
      "grad_norm": 1.125,
      "learning_rate": 0.0009976196187706273,
      "loss": 1.5274,
      "step": 1830
    },
    {
      "epoch": 0.15753247870601395,
      "grad_norm": 1.125,
      "learning_rate": 0.00099761698259616,
      "loss": 1.1126,
      "step": 1831
    },
    {
      "epoch": 0.15761851501333562,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009976143449662597,
      "loss": 1.2508,
      "step": 1832
    },
    {
      "epoch": 0.1577045513206573,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009976117058809345,
      "loss": 1.4668,
      "step": 1833
    },
    {
      "epoch": 0.157790587627979,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009976090653401916,
      "loss": 1.2504,
      "step": 1834
    },
    {
      "epoch": 0.1578766239353007,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009976064233440391,
      "loss": 1.807,
      "step": 1835
    },
    {
      "epoch": 0.1579626602426224,
      "grad_norm": 1.5,
      "learning_rate": 0.0009976037798924847,
      "loss": 1.5992,
      "step": 1836
    },
    {
      "epoch": 0.15804869654994408,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009976011349855359,
      "loss": 0.6844,
      "step": 1837
    },
    {
      "epoch": 0.15813473285726576,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009975984886232006,
      "loss": 2.2172,
      "step": 1838
    },
    {
      "epoch": 0.15822076916458747,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009975958408054864,
      "loss": 1.296,
      "step": 1839
    },
    {
      "epoch": 0.15830680547190915,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009975931915324012,
      "loss": 1.2219,
      "step": 1840
    },
    {
      "epoch": 0.15839284177923083,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009975905408039526,
      "loss": 1.7088,
      "step": 1841
    },
    {
      "epoch": 0.15847887808655253,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009975878886201487,
      "loss": 2.4026,
      "step": 1842
    },
    {
      "epoch": 0.1585649143938742,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009975852349809965,
      "loss": 1.5746,
      "step": 1843
    },
    {
      "epoch": 0.1586509507011959,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009975825798865048,
      "loss": 0.8826,
      "step": 1844
    },
    {
      "epoch": 0.1587369870085176,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009975799233366805,
      "loss": 1.4039,
      "step": 1845
    },
    {
      "epoch": 0.15882302331583928,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009975772653315318,
      "loss": 1.8622,
      "step": 1846
    },
    {
      "epoch": 0.158909059623161,
      "grad_norm": 1.25,
      "learning_rate": 0.0009975746058710664,
      "loss": 1.7675,
      "step": 1847
    },
    {
      "epoch": 0.15899509593048267,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000997571944955292,
      "loss": 1.5851,
      "step": 1848
    },
    {
      "epoch": 0.15908113223780435,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009975692825842165,
      "loss": 1.1359,
      "step": 1849
    },
    {
      "epoch": 0.15916716854512605,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009975666187578476,
      "loss": 1.9455,
      "step": 1850
    },
    {
      "epoch": 0.15925320485244773,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000997563953476193,
      "loss": 1.1711,
      "step": 1851
    },
    {
      "epoch": 0.1593392411597694,
      "grad_norm": 1.125,
      "learning_rate": 0.000997561286739261,
      "loss": 0.8855,
      "step": 1852
    },
    {
      "epoch": 0.15942527746709112,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009975586185470586,
      "loss": 0.8321,
      "step": 1853
    },
    {
      "epoch": 0.1595113137744128,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000997555948899594,
      "loss": 1.4852,
      "step": 1854
    },
    {
      "epoch": 0.15959735008173448,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009975532777968751,
      "loss": 1.1418,
      "step": 1855
    },
    {
      "epoch": 0.1596833863890562,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009975506052389098,
      "loss": 0.9261,
      "step": 1856
    },
    {
      "epoch": 0.15976942269637787,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009975479312257056,
      "loss": 1.04,
      "step": 1857
    },
    {
      "epoch": 0.15985545900369957,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009975452557572705,
      "loss": 1.142,
      "step": 1858
    },
    {
      "epoch": 0.15994149531102125,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009975425788336123,
      "loss": 1.0277,
      "step": 1859
    },
    {
      "epoch": 0.16002753161834293,
      "grad_norm": 1.125,
      "learning_rate": 0.0009975399004547387,
      "loss": 0.8657,
      "step": 1860
    },
    {
      "epoch": 0.16011356792566464,
      "grad_norm": 1.21875,
      "learning_rate": 0.000997537220620658,
      "loss": 0.817,
      "step": 1861
    },
    {
      "epoch": 0.16019960423298632,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009975345393313773,
      "loss": 1.1753,
      "step": 1862
    },
    {
      "epoch": 0.160285640540308,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000997531856586905,
      "loss": 1.4191,
      "step": 1863
    },
    {
      "epoch": 0.1603716768476297,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009975291723872488,
      "loss": 1.7822,
      "step": 1864
    },
    {
      "epoch": 0.1604577131549514,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009975264867324166,
      "loss": 1.2424,
      "step": 1865
    },
    {
      "epoch": 0.16054374946227307,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000997523799622416,
      "loss": 1.3408,
      "step": 1866
    },
    {
      "epoch": 0.16062978576959477,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000997521111057255,
      "loss": 0.7497,
      "step": 1867
    },
    {
      "epoch": 0.16071582207691645,
      "grad_norm": 1.375,
      "learning_rate": 0.0009975184210369418,
      "loss": 1.0324,
      "step": 1868
    },
    {
      "epoch": 0.16080185838423816,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009975157295614837,
      "loss": 1.7337,
      "step": 1869
    },
    {
      "epoch": 0.16088789469155984,
      "grad_norm": 1.28125,
      "learning_rate": 0.000997513036630889,
      "loss": 1.3146,
      "step": 1870
    },
    {
      "epoch": 0.16097393099888152,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009975103422451656,
      "loss": 1.8074,
      "step": 1871
    },
    {
      "epoch": 0.16105996730620323,
      "grad_norm": 1.46875,
      "learning_rate": 0.000997507646404321,
      "loss": 1.5374,
      "step": 1872
    },
    {
      "epoch": 0.1611460036135249,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009975049491083635,
      "loss": 0.9994,
      "step": 1873
    },
    {
      "epoch": 0.1612320399208466,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009975022503573004,
      "loss": 1.4128,
      "step": 1874
    },
    {
      "epoch": 0.1613180762281683,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009974995501511404,
      "loss": 1.8958,
      "step": 1875
    },
    {
      "epoch": 0.16140411253548997,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009974968484898906,
      "loss": 1.2372,
      "step": 1876
    },
    {
      "epoch": 0.16149014884281165,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009974941453735596,
      "loss": 1.2504,
      "step": 1877
    },
    {
      "epoch": 0.16157618515013336,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009974914408021546,
      "loss": 1.6734,
      "step": 1878
    },
    {
      "epoch": 0.16166222145745504,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009974887347756842,
      "loss": 1.534,
      "step": 1879
    },
    {
      "epoch": 0.16174825776477675,
      "grad_norm": 1.765625,
      "learning_rate": 0.000997486027294156,
      "loss": 2.0896,
      "step": 1880
    },
    {
      "epoch": 0.16183429407209843,
      "grad_norm": 1.5,
      "learning_rate": 0.0009974833183575777,
      "loss": 1.1965,
      "step": 1881
    },
    {
      "epoch": 0.1619203303794201,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009974806079659578,
      "loss": 1.0575,
      "step": 1882
    },
    {
      "epoch": 0.16200636668674182,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009974778961193036,
      "loss": 1.9284,
      "step": 1883
    },
    {
      "epoch": 0.1620924029940635,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009974751828176233,
      "loss": 1.5933,
      "step": 1884
    },
    {
      "epoch": 0.16217843930138517,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000997472468060925,
      "loss": 1.372,
      "step": 1885
    },
    {
      "epoch": 0.16226447560870688,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009974697518492163,
      "loss": 1.2093,
      "step": 1886
    },
    {
      "epoch": 0.16235051191602856,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009974670341825057,
      "loss": 1.3701,
      "step": 1887
    },
    {
      "epoch": 0.16243654822335024,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009974643150608004,
      "loss": 1.8327,
      "step": 1888
    },
    {
      "epoch": 0.16252258453067195,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009974615944841087,
      "loss": 1.927,
      "step": 1889
    },
    {
      "epoch": 0.16260862083799363,
      "grad_norm": 1.875,
      "learning_rate": 0.0009974588724524387,
      "loss": 2.0844,
      "step": 1890
    },
    {
      "epoch": 0.16269465714531534,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009974561489657983,
      "loss": 1.6566,
      "step": 1891
    },
    {
      "epoch": 0.16278069345263702,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009974534240241955,
      "loss": 1.9297,
      "step": 1892
    },
    {
      "epoch": 0.1628667297599587,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009974506976276382,
      "loss": 2.031,
      "step": 1893
    },
    {
      "epoch": 0.1629527660672804,
      "grad_norm": 1.8125,
      "learning_rate": 0.000997447969776134,
      "loss": 1.2203,
      "step": 1894
    },
    {
      "epoch": 0.16303880237460208,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009974452404696917,
      "loss": 2.0086,
      "step": 1895
    },
    {
      "epoch": 0.16312483868192376,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009974425097083186,
      "loss": 1.7314,
      "step": 1896
    },
    {
      "epoch": 0.16321087498924547,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009974397774920227,
      "loss": 1.4794,
      "step": 1897
    },
    {
      "epoch": 0.16329691129656715,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009974370438208125,
      "loss": 1.8473,
      "step": 1898
    },
    {
      "epoch": 0.16338294760388883,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009974343086946956,
      "loss": 1.7985,
      "step": 1899
    },
    {
      "epoch": 0.16346898391121054,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00099743157211368,
      "loss": 1.8449,
      "step": 1900
    },
    {
      "epoch": 0.16355502021853222,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000997428834077774,
      "loss": 1.5796,
      "step": 1901
    },
    {
      "epoch": 0.16364105652585392,
      "grad_norm": 1.375,
      "learning_rate": 0.0009974260945869852,
      "loss": 0.9927,
      "step": 1902
    },
    {
      "epoch": 0.1637270928331756,
      "grad_norm": 1.375,
      "learning_rate": 0.000997423353641322,
      "loss": 1.6391,
      "step": 1903
    },
    {
      "epoch": 0.16381312914049728,
      "grad_norm": 1.625,
      "learning_rate": 0.000997420611240792,
      "loss": 1.0605,
      "step": 1904
    },
    {
      "epoch": 0.163899165447819,
      "grad_norm": 2.0,
      "learning_rate": 0.0009974178673854037,
      "loss": 1.4822,
      "step": 1905
    },
    {
      "epoch": 0.16398520175514067,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009974151220751646,
      "loss": 1.3962,
      "step": 1906
    },
    {
      "epoch": 0.16407123806246235,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009974123753100834,
      "loss": 1.4407,
      "step": 1907
    },
    {
      "epoch": 0.16415727436978406,
      "grad_norm": 1.25,
      "learning_rate": 0.0009974096270901674,
      "loss": 1.0845,
      "step": 1908
    },
    {
      "epoch": 0.16424331067710574,
      "grad_norm": 1.625,
      "learning_rate": 0.0009974068774154252,
      "loss": 1.6004,
      "step": 1909
    },
    {
      "epoch": 0.16432934698442742,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009974041262858645,
      "loss": 1.18,
      "step": 1910
    },
    {
      "epoch": 0.16441538329174912,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009974013737014936,
      "loss": 1.6739,
      "step": 1911
    },
    {
      "epoch": 0.1645014195990708,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009973986196623202,
      "loss": 1.3154,
      "step": 1912
    },
    {
      "epoch": 0.1645874559063925,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009973958641683526,
      "loss": 1.0646,
      "step": 1913
    },
    {
      "epoch": 0.1646734922137142,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000997393107219599,
      "loss": 1.5409,
      "step": 1914
    },
    {
      "epoch": 0.16475952852103587,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009973903488160673,
      "loss": 1.4056,
      "step": 1915
    },
    {
      "epoch": 0.16484556482835758,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009973875889577656,
      "loss": 1.9665,
      "step": 1916
    },
    {
      "epoch": 0.16493160113567926,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009973848276447017,
      "loss": 1.7892,
      "step": 1917
    },
    {
      "epoch": 0.16501763744300094,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000997382064876884,
      "loss": 1.2466,
      "step": 1918
    },
    {
      "epoch": 0.16510367375032264,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009973793006543205,
      "loss": 1.257,
      "step": 1919
    },
    {
      "epoch": 0.16518971005764432,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009973765349770194,
      "loss": 2.0047,
      "step": 1920
    },
    {
      "epoch": 0.165275746364966,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009973737678449885,
      "loss": 1.9982,
      "step": 1921
    },
    {
      "epoch": 0.1653617826722877,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009973709992582363,
      "loss": 1.2743,
      "step": 1922
    },
    {
      "epoch": 0.1654478189796094,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009973682292167706,
      "loss": 0.8718,
      "step": 1923
    },
    {
      "epoch": 0.1655338552869311,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009973654577205993,
      "loss": 1.1842,
      "step": 1924
    },
    {
      "epoch": 0.16561989159425278,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000997362684769731,
      "loss": 1.1922,
      "step": 1925
    },
    {
      "epoch": 0.16570592790157446,
      "grad_norm": 2.40625,
      "learning_rate": 0.0009973599103641735,
      "loss": 1.3146,
      "step": 1926
    },
    {
      "epoch": 0.16579196420889616,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009973571345039353,
      "loss": 0.6108,
      "step": 1927
    },
    {
      "epoch": 0.16587800051621784,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009973543571890238,
      "loss": 1.3597,
      "step": 1928
    },
    {
      "epoch": 0.16596403682353952,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009973515784194476,
      "loss": 1.9775,
      "step": 1929
    },
    {
      "epoch": 0.16605007313086123,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000997348798195215,
      "loss": 1.1931,
      "step": 1930
    },
    {
      "epoch": 0.1661361094381829,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009973460165163335,
      "loss": 2.0486,
      "step": 1931
    },
    {
      "epoch": 0.1662221457455046,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000997343233382812,
      "loss": 1.9437,
      "step": 1932
    },
    {
      "epoch": 0.1663081820528263,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009973404487946581,
      "loss": 1.3876,
      "step": 1933
    },
    {
      "epoch": 0.16639421836014798,
      "grad_norm": 2.953125,
      "learning_rate": 0.0009973376627518801,
      "loss": 1.6958,
      "step": 1934
    },
    {
      "epoch": 0.16648025466746968,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009973348752544862,
      "loss": 1.3323,
      "step": 1935
    },
    {
      "epoch": 0.16656629097479136,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009973320863024846,
      "loss": 1.6196,
      "step": 1936
    },
    {
      "epoch": 0.16665232728211304,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009973292958958832,
      "loss": 0.9348,
      "step": 1937
    },
    {
      "epoch": 0.16673836358943475,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009973265040346905,
      "loss": 1.8688,
      "step": 1938
    },
    {
      "epoch": 0.16682439989675643,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009973237107189143,
      "loss": 1.4163,
      "step": 1939
    },
    {
      "epoch": 0.1669104362040781,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009973209159485632,
      "loss": 1.4112,
      "step": 1940
    },
    {
      "epoch": 0.16699647251139982,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009973181197236451,
      "loss": 1.169,
      "step": 1941
    },
    {
      "epoch": 0.1670825088187215,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009973153220441682,
      "loss": 1.2922,
      "step": 1942
    },
    {
      "epoch": 0.16716854512604318,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009973125229101407,
      "loss": 1.1983,
      "step": 1943
    },
    {
      "epoch": 0.16725458143336489,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009973097223215707,
      "loss": 1.4103,
      "step": 1944
    },
    {
      "epoch": 0.16734061774068656,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009973069202784666,
      "loss": 1.4618,
      "step": 1945
    },
    {
      "epoch": 0.16742665404800827,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009973041167808364,
      "loss": 2.0925,
      "step": 1946
    },
    {
      "epoch": 0.16751269035532995,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009973013118286884,
      "loss": 1.8239,
      "step": 1947
    },
    {
      "epoch": 0.16759872666265163,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009972985054220308,
      "loss": 1.3528,
      "step": 1948
    },
    {
      "epoch": 0.16768476296997334,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009972956975608717,
      "loss": 1.4969,
      "step": 1949
    },
    {
      "epoch": 0.16777079927729502,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009972928882452194,
      "loss": 1.3591,
      "step": 1950
    },
    {
      "epoch": 0.1678568355846167,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009972900774750824,
      "loss": 2.0701,
      "step": 1951
    },
    {
      "epoch": 0.1679428718919384,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009972872652504682,
      "loss": 0.8908,
      "step": 1952
    },
    {
      "epoch": 0.16802890819926009,
      "grad_norm": 1.234375,
      "learning_rate": 0.000997284451571386,
      "loss": 2.0322,
      "step": 1953
    },
    {
      "epoch": 0.16811494450658176,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000997281636437843,
      "loss": 2.1233,
      "step": 1954
    },
    {
      "epoch": 0.16820098081390347,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009972788198498483,
      "loss": 1.6203,
      "step": 1955
    },
    {
      "epoch": 0.16828701712122515,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009972760018074095,
      "loss": 1.8318,
      "step": 1956
    },
    {
      "epoch": 0.16837305342854686,
      "grad_norm": 1.25,
      "learning_rate": 0.0009972731823105353,
      "loss": 1.6646,
      "step": 1957
    },
    {
      "epoch": 0.16845908973586854,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009972703613592337,
      "loss": 1.2464,
      "step": 1958
    },
    {
      "epoch": 0.16854512604319022,
      "grad_norm": 1.640625,
      "learning_rate": 0.000997267538953513,
      "loss": 1.3708,
      "step": 1959
    },
    {
      "epoch": 0.16863116235051193,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009972647150933813,
      "loss": 1.5816,
      "step": 1960
    },
    {
      "epoch": 0.1687171986578336,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009972618897788473,
      "loss": 1.7004,
      "step": 1961
    },
    {
      "epoch": 0.16880323496515529,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009972590630099188,
      "loss": 1.5866,
      "step": 1962
    },
    {
      "epoch": 0.168889271272477,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009972562347866044,
      "loss": 2.1368,
      "step": 1963
    },
    {
      "epoch": 0.16897530757979867,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000997253405108912,
      "loss": 1.0685,
      "step": 1964
    },
    {
      "epoch": 0.16906134388712035,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009972505739768503,
      "loss": 1.6051,
      "step": 1965
    },
    {
      "epoch": 0.16914738019444206,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009972477413904273,
      "loss": 1.2806,
      "step": 1966
    },
    {
      "epoch": 0.16923341650176374,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009972449073496511,
      "loss": 1.0013,
      "step": 1967
    },
    {
      "epoch": 0.16931945280908545,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009972420718545308,
      "loss": 1.8216,
      "step": 1968
    },
    {
      "epoch": 0.16940548911640713,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000997239234905074,
      "loss": 1.9077,
      "step": 1969
    },
    {
      "epoch": 0.1694915254237288,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009972363965012887,
      "loss": 1.0193,
      "step": 1970
    },
    {
      "epoch": 0.1695775617310505,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000997233556643184,
      "loss": 1.1251,
      "step": 1971
    },
    {
      "epoch": 0.1696635980383722,
      "grad_norm": 1.5625,
      "learning_rate": 0.000997230715330768,
      "loss": 1.919,
      "step": 1972
    },
    {
      "epoch": 0.16974963434569387,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009972278725640486,
      "loss": 1.2991,
      "step": 1973
    },
    {
      "epoch": 0.16983567065301558,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009972250283430343,
      "loss": 2.2858,
      "step": 1974
    },
    {
      "epoch": 0.16992170696033726,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009972221826677339,
      "loss": 1.7302,
      "step": 1975
    },
    {
      "epoch": 0.17000774326765894,
      "grad_norm": 1.390625,
      "learning_rate": 0.000997219335538155,
      "loss": 1.4773,
      "step": 1976
    },
    {
      "epoch": 0.17009377957498065,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009972164869543063,
      "loss": 0.8574,
      "step": 1977
    },
    {
      "epoch": 0.17017981588230233,
      "grad_norm": 1.484375,
      "learning_rate": 0.000997213636916196,
      "loss": 0.8498,
      "step": 1978
    },
    {
      "epoch": 0.17026585218962403,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009972107854238326,
      "loss": 1.7874,
      "step": 1979
    },
    {
      "epoch": 0.1703518884969457,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009972079324772245,
      "loss": 1.2998,
      "step": 1980
    },
    {
      "epoch": 0.1704379248042674,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009972050780763797,
      "loss": 1.0491,
      "step": 1981
    },
    {
      "epoch": 0.1705239611115891,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009972022222213069,
      "loss": 0.9135,
      "step": 1982
    },
    {
      "epoch": 0.17060999741891078,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000997199364912014,
      "loss": 1.3087,
      "step": 1983
    },
    {
      "epoch": 0.17069603372623246,
      "grad_norm": 1.671875,
      "learning_rate": 0.00099719650614851,
      "loss": 1.2676,
      "step": 1984
    },
    {
      "epoch": 0.17078207003355417,
      "grad_norm": 1.609375,
      "learning_rate": 0.000997193645930803,
      "loss": 1.2248,
      "step": 1985
    },
    {
      "epoch": 0.17086810634087585,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009971907842589009,
      "loss": 2.2145,
      "step": 1986
    },
    {
      "epoch": 0.17095414264819753,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009971879211328127,
      "loss": 1.4278,
      "step": 1987
    },
    {
      "epoch": 0.17104017895551923,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009971850565525463,
      "loss": 1.1177,
      "step": 1988
    },
    {
      "epoch": 0.17112621526284091,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009971821905181107,
      "loss": 1.3835,
      "step": 1989
    },
    {
      "epoch": 0.17121225157016262,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009971793230295135,
      "loss": 2.1967,
      "step": 1990
    },
    {
      "epoch": 0.1712982878774843,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009971764540867636,
      "loss": 2.0289,
      "step": 1991
    },
    {
      "epoch": 0.17138432418480598,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009971735836898695,
      "loss": 0.9857,
      "step": 1992
    },
    {
      "epoch": 0.1714703604921277,
      "grad_norm": 1.578125,
      "learning_rate": 0.000997170711838839,
      "loss": 2.4046,
      "step": 1993
    },
    {
      "epoch": 0.17155639679944937,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000997167838533681,
      "loss": 1.3359,
      "step": 1994
    },
    {
      "epoch": 0.17164243310677105,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009971649637744039,
      "loss": 1.4077,
      "step": 1995
    },
    {
      "epoch": 0.17172846941409275,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000997162087561016,
      "loss": 0.9214,
      "step": 1996
    },
    {
      "epoch": 0.17181450572141443,
      "grad_norm": 2.828125,
      "learning_rate": 0.0009971592098935254,
      "loss": 1.5616,
      "step": 1997
    },
    {
      "epoch": 0.17190054202873611,
      "grad_norm": 1.46875,
      "learning_rate": 0.000997156330771941,
      "loss": 1.6318,
      "step": 1998
    },
    {
      "epoch": 0.17198657833605782,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009971534501962709,
      "loss": 1.8554,
      "step": 1999
    },
    {
      "epoch": 0.1720726146433795,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009971505681665238,
      "loss": 1.2697,
      "step": 2000
    },
    {
      "epoch": 0.1721586509507012,
      "grad_norm": 1.390625,
      "learning_rate": 0.000997147684682708,
      "loss": 1.1728,
      "step": 2001
    },
    {
      "epoch": 0.1722446872580229,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009971447997448317,
      "loss": 1.3465,
      "step": 2002
    },
    {
      "epoch": 0.17233072356534457,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009971419133529035,
      "loss": 1.5237,
      "step": 2003
    },
    {
      "epoch": 0.17241675987266628,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009971390255069323,
      "loss": 1.7363,
      "step": 2004
    },
    {
      "epoch": 0.17250279617998796,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009971361362069258,
      "loss": 2.3775,
      "step": 2005
    },
    {
      "epoch": 0.17258883248730963,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000997133245452893,
      "loss": 0.9334,
      "step": 2006
    },
    {
      "epoch": 0.17267486879463134,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000997130353244842,
      "loss": 2.0919,
      "step": 2007
    },
    {
      "epoch": 0.17276090510195302,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009971274595827813,
      "loss": 0.7966,
      "step": 2008
    },
    {
      "epoch": 0.1728469414092747,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009971245644667197,
      "loss": 1.9651,
      "step": 2009
    },
    {
      "epoch": 0.1729329777165964,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009971216678966654,
      "loss": 1.5635,
      "step": 2010
    },
    {
      "epoch": 0.1730190140239181,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009971187698726268,
      "loss": 1.1317,
      "step": 2011
    },
    {
      "epoch": 0.1731050503312398,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009971158703946125,
      "loss": 2.0572,
      "step": 2012
    },
    {
      "epoch": 0.17319108663856148,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009971129694626311,
      "loss": 1.0516,
      "step": 2013
    },
    {
      "epoch": 0.17327712294588316,
      "grad_norm": 1.375,
      "learning_rate": 0.0009971100670766908,
      "loss": 1.8059,
      "step": 2014
    },
    {
      "epoch": 0.17336315925320486,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009971071632368003,
      "loss": 1.3593,
      "step": 2015
    },
    {
      "epoch": 0.17344919556052654,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000997104257942968,
      "loss": 1.9413,
      "step": 2016
    },
    {
      "epoch": 0.17353523186784822,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009971013511952023,
      "loss": 1.3117,
      "step": 2017
    },
    {
      "epoch": 0.17362126817516993,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000997098442993512,
      "loss": 2.209,
      "step": 2018
    },
    {
      "epoch": 0.1737073044824916,
      "grad_norm": 1.25,
      "learning_rate": 0.0009970955333379054,
      "loss": 1.2672,
      "step": 2019
    },
    {
      "epoch": 0.1737933407898133,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000997092622228391,
      "loss": 1.5799,
      "step": 2020
    },
    {
      "epoch": 0.173879377097135,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009970897096649774,
      "loss": 1.9728,
      "step": 2021
    },
    {
      "epoch": 0.17396541340445668,
      "grad_norm": 1.390625,
      "learning_rate": 0.000997086795647673,
      "loss": 1.2105,
      "step": 2022
    },
    {
      "epoch": 0.17405144971177838,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009970838801764866,
      "loss": 1.2669,
      "step": 2023
    },
    {
      "epoch": 0.17413748601910006,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009970809632514263,
      "loss": 1.1603,
      "step": 2024
    },
    {
      "epoch": 0.17422352232642174,
      "grad_norm": 1.421875,
      "learning_rate": 0.000997078044872501,
      "loss": 1.6967,
      "step": 2025
    },
    {
      "epoch": 0.17430955863374345,
      "grad_norm": 2.78125,
      "learning_rate": 0.0009970751250397193,
      "loss": 1.8357,
      "step": 2026
    },
    {
      "epoch": 0.17439559494106513,
      "grad_norm": 1.25,
      "learning_rate": 0.0009970722037530892,
      "loss": 1.0288,
      "step": 2027
    },
    {
      "epoch": 0.1744816312483868,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009970692810126196,
      "loss": 1.2266,
      "step": 2028
    },
    {
      "epoch": 0.17456766755570852,
      "grad_norm": 1.109375,
      "learning_rate": 0.000997066356818319,
      "loss": 0.6399,
      "step": 2029
    },
    {
      "epoch": 0.1746537038630302,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009970634311701962,
      "loss": 1.4415,
      "step": 2030
    },
    {
      "epoch": 0.17473974017035188,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009970605040682595,
      "loss": 1.6951,
      "step": 2031
    },
    {
      "epoch": 0.17482577647767358,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009970575755125174,
      "loss": 1.4621,
      "step": 2032
    },
    {
      "epoch": 0.17491181278499526,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009970546455029786,
      "loss": 2.0052,
      "step": 2033
    },
    {
      "epoch": 0.17499784909231697,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009970517140396515,
      "loss": 1.9741,
      "step": 2034
    },
    {
      "epoch": 0.17508388539963865,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009970487811225448,
      "loss": 1.3538,
      "step": 2035
    },
    {
      "epoch": 0.17516992170696033,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009970458467516675,
      "loss": 1.747,
      "step": 2036
    },
    {
      "epoch": 0.17525595801428204,
      "grad_norm": 1.25,
      "learning_rate": 0.0009970429109270275,
      "loss": 1.5067,
      "step": 2037
    },
    {
      "epoch": 0.17534199432160372,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009970399736486334,
      "loss": 1.6286,
      "step": 2038
    },
    {
      "epoch": 0.1754280306289254,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009970370349164943,
      "loss": 1.7252,
      "step": 2039
    },
    {
      "epoch": 0.1755140669362471,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009970340947306185,
      "loss": 1.4348,
      "step": 2040
    },
    {
      "epoch": 0.17560010324356878,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009970311530910147,
      "loss": 1.3863,
      "step": 2041
    },
    {
      "epoch": 0.17568613955089046,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009970282099976916,
      "loss": 1.2423,
      "step": 2042
    },
    {
      "epoch": 0.17577217585821217,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009970252654506572,
      "loss": 1.2305,
      "step": 2043
    },
    {
      "epoch": 0.17585821216553385,
      "grad_norm": 1.46875,
      "learning_rate": 0.000997022319449921,
      "loss": 1.4625,
      "step": 2044
    },
    {
      "epoch": 0.17594424847285556,
      "grad_norm": 1.796875,
      "learning_rate": 0.000997019371995491,
      "loss": 2.0695,
      "step": 2045
    },
    {
      "epoch": 0.17603028478017724,
      "grad_norm": 1.7421875,
      "learning_rate": 0.000997016423087376,
      "loss": 1.6522,
      "step": 2046
    },
    {
      "epoch": 0.17611632108749892,
      "grad_norm": 2.75,
      "learning_rate": 0.0009970134727255845,
      "loss": 1.8631,
      "step": 2047
    },
    {
      "epoch": 0.17620235739482062,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009970105209101254,
      "loss": 1.0122,
      "step": 2048
    },
    {
      "epoch": 0.1762883937021423,
      "grad_norm": 1.375,
      "learning_rate": 0.0009970075676410072,
      "loss": 1.4489,
      "step": 2049
    },
    {
      "epoch": 0.17637443000946398,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009970046129182384,
      "loss": 0.9005,
      "step": 2050
    },
    {
      "epoch": 0.1764604663167857,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000997001656741828,
      "loss": 1.5614,
      "step": 2051
    },
    {
      "epoch": 0.17654650262410737,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009969986991117843,
      "loss": 1.3204,
      "step": 2052
    },
    {
      "epoch": 0.17663253893142905,
      "grad_norm": 1.25,
      "learning_rate": 0.000996995740028116,
      "loss": 1.8519,
      "step": 2053
    },
    {
      "epoch": 0.17671857523875076,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009969927794908318,
      "loss": 1.5524,
      "step": 2054
    },
    {
      "epoch": 0.17680461154607244,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009969898174999406,
      "loss": 2.0372,
      "step": 2055
    },
    {
      "epoch": 0.17689064785339415,
      "grad_norm": 1.75,
      "learning_rate": 0.0009969868540554505,
      "loss": 1.6827,
      "step": 2056
    },
    {
      "epoch": 0.17697668416071582,
      "grad_norm": 1.328125,
      "learning_rate": 0.000996983889157371,
      "loss": 1.24,
      "step": 2057
    },
    {
      "epoch": 0.1770627204680375,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009969809228057098,
      "loss": 1.6184,
      "step": 2058
    },
    {
      "epoch": 0.1771487567753592,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009969779550004763,
      "loss": 1.2545,
      "step": 2059
    },
    {
      "epoch": 0.1772347930826809,
      "grad_norm": 1.734375,
      "learning_rate": 0.000996974985741679,
      "loss": 1.0467,
      "step": 2060
    },
    {
      "epoch": 0.17732082939000257,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009969720150293264,
      "loss": 1.9365,
      "step": 2061
    },
    {
      "epoch": 0.17740686569732428,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009969690428634273,
      "loss": 1.6936,
      "step": 2062
    },
    {
      "epoch": 0.17749290200464596,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009969660692439903,
      "loss": 1.1879,
      "step": 2063
    },
    {
      "epoch": 0.17757893831196764,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009969630941710243,
      "loss": 0.9584,
      "step": 2064
    },
    {
      "epoch": 0.17766497461928935,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000996960117644538,
      "loss": 1.4899,
      "step": 2065
    },
    {
      "epoch": 0.17775101092661102,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009969571396645398,
      "loss": 0.9094,
      "step": 2066
    },
    {
      "epoch": 0.17783704723393273,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000996954160231039,
      "loss": 1.6749,
      "step": 2067
    },
    {
      "epoch": 0.1779230835412544,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009969511793440436,
      "loss": 1.0985,
      "step": 2068
    },
    {
      "epoch": 0.1780091198485761,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009969481970035627,
      "loss": 1.4188,
      "step": 2069
    },
    {
      "epoch": 0.1780951561558978,
      "grad_norm": 1.71875,
      "learning_rate": 0.000996945213209605,
      "loss": 1.3344,
      "step": 2070
    },
    {
      "epoch": 0.17818119246321948,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009969422279621793,
      "loss": 1.5218,
      "step": 2071
    },
    {
      "epoch": 0.17826722877054116,
      "grad_norm": 1.53125,
      "learning_rate": 0.000996939241261294,
      "loss": 1.2977,
      "step": 2072
    },
    {
      "epoch": 0.17835326507786287,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009969362531069583,
      "loss": 0.9885,
      "step": 2073
    },
    {
      "epoch": 0.17843930138518455,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009969332634991808,
      "loss": 1.5676,
      "step": 2074
    },
    {
      "epoch": 0.17852533769250623,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009969302724379698,
      "loss": 1.5944,
      "step": 2075
    },
    {
      "epoch": 0.17861137399982793,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009969272799233346,
      "loss": 1.1711,
      "step": 2076
    },
    {
      "epoch": 0.1786974103071496,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000996924285955284,
      "loss": 1.3185,
      "step": 2077
    },
    {
      "epoch": 0.17878344661447132,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009969212905338262,
      "loss": 1.746,
      "step": 2078
    },
    {
      "epoch": 0.178869482921793,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009969182936589703,
      "loss": 1.7743,
      "step": 2079
    },
    {
      "epoch": 0.17895551922911468,
      "grad_norm": 1.8828125,
      "learning_rate": 0.000996915295330725,
      "loss": 1.5304,
      "step": 2080
    },
    {
      "epoch": 0.1790415555364364,
      "grad_norm": 1.734375,
      "learning_rate": 0.000996912295549099,
      "loss": 1.7105,
      "step": 2081
    },
    {
      "epoch": 0.17912759184375807,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009969092943141016,
      "loss": 0.7438,
      "step": 2082
    },
    {
      "epoch": 0.17921362815107975,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009969062916257408,
      "loss": 1.41,
      "step": 2083
    },
    {
      "epoch": 0.17929966445840145,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009969032874840259,
      "loss": 1.5628,
      "step": 2084
    },
    {
      "epoch": 0.17938570076572313,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009969002818889655,
      "loss": 1.2364,
      "step": 2085
    },
    {
      "epoch": 0.1794717370730448,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009968972748405683,
      "loss": 0.8602,
      "step": 2086
    },
    {
      "epoch": 0.17955777338036652,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009968942663388432,
      "loss": 1.6101,
      "step": 2087
    },
    {
      "epoch": 0.1796438096876882,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009968912563837991,
      "loss": 0.9547,
      "step": 2088
    },
    {
      "epoch": 0.1797298459950099,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000996888244975445,
      "loss": 1.3898,
      "step": 2089
    },
    {
      "epoch": 0.1798158823023316,
      "grad_norm": 1.75,
      "learning_rate": 0.000996885232113789,
      "loss": 1.6079,
      "step": 2090
    },
    {
      "epoch": 0.17990191860965327,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009968822177988403,
      "loss": 1.6627,
      "step": 2091
    },
    {
      "epoch": 0.17998795491697497,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009968792020306082,
      "loss": 1.2444,
      "step": 2092
    },
    {
      "epoch": 0.18007399122429665,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009968761848091006,
      "loss": 2.1688,
      "step": 2093
    },
    {
      "epoch": 0.18016002753161833,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009968731661343271,
      "loss": 1.746,
      "step": 2094
    },
    {
      "epoch": 0.18024606383894004,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009968701460062962,
      "loss": 1.9006,
      "step": 2095
    },
    {
      "epoch": 0.18033210014626172,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009968671244250165,
      "loss": 0.924,
      "step": 2096
    },
    {
      "epoch": 0.1804181364535834,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009968641013904974,
      "loss": 1.9657,
      "step": 2097
    },
    {
      "epoch": 0.1805041727609051,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009968610769027473,
      "loss": 1.2243,
      "step": 2098
    },
    {
      "epoch": 0.1805902090682268,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009968580509617752,
      "loss": 1.9282,
      "step": 2099
    },
    {
      "epoch": 0.1806762453755485,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00099685502356759,
      "loss": 1.4091,
      "step": 2100
    },
    {
      "epoch": 0.18076228168287017,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009968519947202003,
      "loss": 1.1584,
      "step": 2101
    },
    {
      "epoch": 0.18084831799019185,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009968489644196152,
      "loss": 1.7761,
      "step": 2102
    },
    {
      "epoch": 0.18093435429751356,
      "grad_norm": 1.375,
      "learning_rate": 0.0009968459326658438,
      "loss": 1.6982,
      "step": 2103
    },
    {
      "epoch": 0.18102039060483524,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009968428994588943,
      "loss": 0.9497,
      "step": 2104
    },
    {
      "epoch": 0.18110642691215692,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009968398647987762,
      "loss": 1.3439,
      "step": 2105
    },
    {
      "epoch": 0.18119246321947863,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009968368286854981,
      "loss": 1.9671,
      "step": 2106
    },
    {
      "epoch": 0.1812784995268003,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009968337911190688,
      "loss": 1.3033,
      "step": 2107
    },
    {
      "epoch": 0.181364535834122,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009968307520994972,
      "loss": 2.4012,
      "step": 2108
    },
    {
      "epoch": 0.1814505721414437,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009968277116267925,
      "loss": 2.1009,
      "step": 2109
    },
    {
      "epoch": 0.18153660844876537,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009968246697009634,
      "loss": 1.1438,
      "step": 2110
    },
    {
      "epoch": 0.18162264475608708,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009968216263220185,
      "loss": 1.2699,
      "step": 2111
    },
    {
      "epoch": 0.18170868106340876,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000996818581489967,
      "loss": 1.615,
      "step": 2112
    },
    {
      "epoch": 0.18179471737073044,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000996815535204818,
      "loss": 1.4455,
      "step": 2113
    },
    {
      "epoch": 0.18188075367805215,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009968124874665803,
      "loss": 1.3786,
      "step": 2114
    },
    {
      "epoch": 0.18196678998537383,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009968094382752624,
      "loss": 1.9995,
      "step": 2115
    },
    {
      "epoch": 0.1820528262926955,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009968063876308734,
      "loss": 1.7027,
      "step": 2116
    },
    {
      "epoch": 0.18213886260001722,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009968033355334228,
      "loss": 1.1405,
      "step": 2117
    },
    {
      "epoch": 0.1822248989073389,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009968002819829187,
      "loss": 2.4297,
      "step": 2118
    },
    {
      "epoch": 0.18231093521466057,
      "grad_norm": 1.5,
      "learning_rate": 0.0009967972269793704,
      "loss": 1.568,
      "step": 2119
    },
    {
      "epoch": 0.18239697152198228,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000996794170522787,
      "loss": 1.1281,
      "step": 2120
    },
    {
      "epoch": 0.18248300782930396,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009967911126131772,
      "loss": 1.5332,
      "step": 2121
    },
    {
      "epoch": 0.18256904413662567,
      "grad_norm": 3.390625,
      "learning_rate": 0.00099678805325055,
      "loss": 1.2189,
      "step": 2122
    },
    {
      "epoch": 0.18265508044394735,
      "grad_norm": 1.5,
      "learning_rate": 0.0009967849924349143,
      "loss": 2.0622,
      "step": 2123
    },
    {
      "epoch": 0.18274111675126903,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009967819301662793,
      "loss": 1.3531,
      "step": 2124
    },
    {
      "epoch": 0.18282715305859074,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009967788664446537,
      "loss": 1.3378,
      "step": 2125
    },
    {
      "epoch": 0.18291318936591242,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009967758012700465,
      "loss": 1.5561,
      "step": 2126
    },
    {
      "epoch": 0.1829992256732341,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009967727346424666,
      "loss": 1.1467,
      "step": 2127
    },
    {
      "epoch": 0.1830852619805558,
      "grad_norm": 1.625,
      "learning_rate": 0.0009967696665619233,
      "loss": 1.58,
      "step": 2128
    },
    {
      "epoch": 0.18317129828787748,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009967665970284252,
      "loss": 1.3705,
      "step": 2129
    },
    {
      "epoch": 0.18325733459519916,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009967635260419814,
      "loss": 1.3524,
      "step": 2130
    },
    {
      "epoch": 0.18334337090252087,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000996760453602601,
      "loss": 1.1572,
      "step": 2131
    },
    {
      "epoch": 0.18342940720984255,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009967573797102927,
      "loss": 1.4197,
      "step": 2132
    },
    {
      "epoch": 0.18351544351716426,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009967543043650658,
      "loss": 1.372,
      "step": 2133
    },
    {
      "epoch": 0.18360147982448594,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009967512275669292,
      "loss": 1.052,
      "step": 2134
    },
    {
      "epoch": 0.18368751613180762,
      "grad_norm": 1.25,
      "learning_rate": 0.0009967481493158919,
      "loss": 0.9225,
      "step": 2135
    },
    {
      "epoch": 0.18377355243912932,
      "grad_norm": 1.390625,
      "learning_rate": 0.000996745069611963,
      "loss": 1.6353,
      "step": 2136
    },
    {
      "epoch": 0.183859588746451,
      "grad_norm": 1.21875,
      "learning_rate": 0.000996741988455151,
      "loss": 1.7508,
      "step": 2137
    },
    {
      "epoch": 0.18394562505377268,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009967389058454655,
      "loss": 1.5024,
      "step": 2138
    },
    {
      "epoch": 0.1840316613610944,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009967358217829155,
      "loss": 0.6078,
      "step": 2139
    },
    {
      "epoch": 0.18411769766841607,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009967327362675096,
      "loss": 1.5397,
      "step": 2140
    },
    {
      "epoch": 0.18420373397573775,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000996729649299257,
      "loss": 1.3384,
      "step": 2141
    },
    {
      "epoch": 0.18428977028305946,
      "grad_norm": 2.015625,
      "learning_rate": 0.000996726560878167,
      "loss": 1.4721,
      "step": 2142
    },
    {
      "epoch": 0.18437580659038114,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009967234710042481,
      "loss": 1.8903,
      "step": 2143
    },
    {
      "epoch": 0.18446184289770284,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00099672037967751,
      "loss": 2.0589,
      "step": 2144
    },
    {
      "epoch": 0.18454787920502452,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009967172868979612,
      "loss": 0.9778,
      "step": 2145
    },
    {
      "epoch": 0.1846339155123462,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000996714192665611,
      "loss": 1.942,
      "step": 2146
    },
    {
      "epoch": 0.1847199518196679,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009967110969804685,
      "loss": 1.1751,
      "step": 2147
    },
    {
      "epoch": 0.1848059881269896,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009967079998425424,
      "loss": 1.6072,
      "step": 2148
    },
    {
      "epoch": 0.18489202443431127,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009967049012518422,
      "loss": 1.5021,
      "step": 2149
    },
    {
      "epoch": 0.18497806074163298,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009967018012083766,
      "loss": 1.5972,
      "step": 2150
    },
    {
      "epoch": 0.18506409704895466,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000996698699712155,
      "loss": 0.9602,
      "step": 2151
    },
    {
      "epoch": 0.18515013335627634,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009966955967631863,
      "loss": 1.8258,
      "step": 2152
    },
    {
      "epoch": 0.18523616966359804,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009966924923614795,
      "loss": 1.2667,
      "step": 2153
    },
    {
      "epoch": 0.18532220597091972,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009966893865070439,
      "loss": 1.8267,
      "step": 2154
    },
    {
      "epoch": 0.18540824227824143,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009966862791998881,
      "loss": 1.8519,
      "step": 2155
    },
    {
      "epoch": 0.1854942785855631,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009966831704400218,
      "loss": 0.7039,
      "step": 2156
    },
    {
      "epoch": 0.1855803148928848,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009966800602274539,
      "loss": 1.0765,
      "step": 2157
    },
    {
      "epoch": 0.1856663512002065,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009966769485621934,
      "loss": 1.8342,
      "step": 2158
    },
    {
      "epoch": 0.18575238750752818,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009966738354442494,
      "loss": 1.4945,
      "step": 2159
    },
    {
      "epoch": 0.18583842381484986,
      "grad_norm": 1.375,
      "learning_rate": 0.0009966707208736308,
      "loss": 1.1685,
      "step": 2160
    },
    {
      "epoch": 0.18592446012217156,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009966676048503472,
      "loss": 0.9388,
      "step": 2161
    },
    {
      "epoch": 0.18601049642949324,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009966644873744073,
      "loss": 1.3117,
      "step": 2162
    },
    {
      "epoch": 0.18609653273681492,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009966613684458203,
      "loss": 1.1425,
      "step": 2163
    },
    {
      "epoch": 0.18618256904413663,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009966582480645957,
      "loss": 1.7752,
      "step": 2164
    },
    {
      "epoch": 0.1862686053514583,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009966551262307422,
      "loss": 1.4097,
      "step": 2165
    },
    {
      "epoch": 0.18635464165878002,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009966520029442687,
      "loss": 1.2093,
      "step": 2166
    },
    {
      "epoch": 0.1864406779661017,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009966488782051852,
      "loss": 1.4372,
      "step": 2167
    },
    {
      "epoch": 0.18652671427342338,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009966457520135,
      "loss": 1.3154,
      "step": 2168
    },
    {
      "epoch": 0.18661275058074508,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009966426243692226,
      "loss": 1.3323,
      "step": 2169
    },
    {
      "epoch": 0.18669878688806676,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009966394952723623,
      "loss": 1.2433,
      "step": 2170
    },
    {
      "epoch": 0.18678482319538844,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000996636364722928,
      "loss": 1.1889,
      "step": 2171
    },
    {
      "epoch": 0.18687085950271015,
      "grad_norm": 1.296875,
      "learning_rate": 0.000996633232720929,
      "loss": 1.4789,
      "step": 2172
    },
    {
      "epoch": 0.18695689581003183,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000996630099266374,
      "loss": 0.792,
      "step": 2173
    },
    {
      "epoch": 0.1870429321173535,
      "grad_norm": 1.265625,
      "learning_rate": 0.000996626964359273,
      "loss": 0.9822,
      "step": 2174
    },
    {
      "epoch": 0.18712896842467522,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009966238279996344,
      "loss": 1.0864,
      "step": 2175
    },
    {
      "epoch": 0.1872150047319969,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009966206901874679,
      "loss": 1.5852,
      "step": 2176
    },
    {
      "epoch": 0.1873010410393186,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009966175509227822,
      "loss": 1.5443,
      "step": 2177
    },
    {
      "epoch": 0.18738707734664029,
      "grad_norm": 1.578125,
      "learning_rate": 0.000996614410205587,
      "loss": 1.3461,
      "step": 2178
    },
    {
      "epoch": 0.18747311365396196,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009966112680358911,
      "loss": 1.6132,
      "step": 2179
    },
    {
      "epoch": 0.18755914996128367,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000996608124413704,
      "loss": 1.0858,
      "step": 2180
    },
    {
      "epoch": 0.18764518626860535,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009966049793390347,
      "loss": 1.3375,
      "step": 2181
    },
    {
      "epoch": 0.18773122257592703,
      "grad_norm": 1.15625,
      "learning_rate": 0.000996601832811892,
      "loss": 1.1103,
      "step": 2182
    },
    {
      "epoch": 0.18781725888324874,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009965986848322862,
      "loss": 0.9571,
      "step": 2183
    },
    {
      "epoch": 0.18790329519057042,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009965955354002253,
      "loss": 1.302,
      "step": 2184
    },
    {
      "epoch": 0.1879893314978921,
      "grad_norm": 3.046875,
      "learning_rate": 0.0009965923845157195,
      "loss": 1.715,
      "step": 2185
    },
    {
      "epoch": 0.1880753678052138,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009965892321787773,
      "loss": 2.027,
      "step": 2186
    },
    {
      "epoch": 0.18816140411253549,
      "grad_norm": 1.796875,
      "learning_rate": 0.000996586078389408,
      "loss": 1.451,
      "step": 2187
    },
    {
      "epoch": 0.1882474404198572,
      "grad_norm": 1.09375,
      "learning_rate": 0.000996582923147621,
      "loss": 1.001,
      "step": 2188
    },
    {
      "epoch": 0.18833347672717887,
      "grad_norm": 1.703125,
      "learning_rate": 0.000996579766453426,
      "loss": 2.0092,
      "step": 2189
    },
    {
      "epoch": 0.18841951303450055,
      "grad_norm": 1.625,
      "learning_rate": 0.0009965766083068312,
      "loss": 1.5927,
      "step": 2190
    },
    {
      "epoch": 0.18850554934182226,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009965734487078469,
      "loss": 1.5733,
      "step": 2191
    },
    {
      "epoch": 0.18859158564914394,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009965702876564815,
      "loss": 1.0269,
      "step": 2192
    },
    {
      "epoch": 0.18867762195646562,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009965671251527446,
      "loss": 1.6789,
      "step": 2193
    },
    {
      "epoch": 0.18876365826378733,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009965639611966455,
      "loss": 0.8866,
      "step": 2194
    },
    {
      "epoch": 0.188849694571109,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009965607957881934,
      "loss": 1.214,
      "step": 2195
    },
    {
      "epoch": 0.18893573087843069,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009965576289273977,
      "loss": 0.7529,
      "step": 2196
    },
    {
      "epoch": 0.1890217671857524,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009965544606142674,
      "loss": 1.4765,
      "step": 2197
    },
    {
      "epoch": 0.18910780349307407,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009965512908488119,
      "loss": 1.1587,
      "step": 2198
    },
    {
      "epoch": 0.18919383980039578,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009965481196310404,
      "loss": 1.8348,
      "step": 2199
    },
    {
      "epoch": 0.18927987610771746,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009965449469609622,
      "loss": 1.1239,
      "step": 2200
    },
    {
      "epoch": 0.18936591241503914,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009965417728385866,
      "loss": 1.0214,
      "step": 2201
    },
    {
      "epoch": 0.18945194872236085,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000996538597263923,
      "loss": 1.7233,
      "step": 2202
    },
    {
      "epoch": 0.18953798502968253,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009965354202369802,
      "loss": 1.0328,
      "step": 2203
    },
    {
      "epoch": 0.1896240213370042,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009965322417577682,
      "loss": 1.2631,
      "step": 2204
    },
    {
      "epoch": 0.1897100576443259,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000996529061826296,
      "loss": 1.808,
      "step": 2205
    },
    {
      "epoch": 0.1897960939516476,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009965258804425729,
      "loss": 1.4453,
      "step": 2206
    },
    {
      "epoch": 0.18988213025896927,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000996522697606608,
      "loss": 1.3596,
      "step": 2207
    },
    {
      "epoch": 0.18996816656629098,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009965195133184107,
      "loss": 1.4115,
      "step": 2208
    },
    {
      "epoch": 0.19005420287361266,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009965163275779906,
      "loss": 1.1352,
      "step": 2209
    },
    {
      "epoch": 0.19014023918093437,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009965131403853566,
      "loss": 1.1975,
      "step": 2210
    },
    {
      "epoch": 0.19022627548825605,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009965099517405185,
      "loss": 1.3263,
      "step": 2211
    },
    {
      "epoch": 0.19031231179557773,
      "grad_norm": 1.5,
      "learning_rate": 0.0009965067616434851,
      "loss": 2.0469,
      "step": 2212
    },
    {
      "epoch": 0.19039834810289943,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009965035700942662,
      "loss": 1.6838,
      "step": 2213
    },
    {
      "epoch": 0.1904843844102211,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009965003770928708,
      "loss": 1.788,
      "step": 2214
    },
    {
      "epoch": 0.1905704207175428,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009964971826393082,
      "loss": 1.2555,
      "step": 2215
    },
    {
      "epoch": 0.1906564570248645,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009964939867335883,
      "loss": 1.5645,
      "step": 2216
    },
    {
      "epoch": 0.19074249333218618,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009964907893757198,
      "loss": 1.1391,
      "step": 2217
    },
    {
      "epoch": 0.19082852963950786,
      "grad_norm": 2.546875,
      "learning_rate": 0.0009964875905657123,
      "loss": 1.4493,
      "step": 2218
    },
    {
      "epoch": 0.19091456594682957,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009964843903035752,
      "loss": 0.6854,
      "step": 2219
    },
    {
      "epoch": 0.19100060225415125,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000996481188589318,
      "loss": 1.0477,
      "step": 2220
    },
    {
      "epoch": 0.19108663856147295,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009964779854229496,
      "loss": 1.3001,
      "step": 2221
    },
    {
      "epoch": 0.19117267486879463,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009964747808044797,
      "loss": 1.5293,
      "step": 2222
    },
    {
      "epoch": 0.19125871117611631,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009964715747339177,
      "loss": 1.3063,
      "step": 2223
    },
    {
      "epoch": 0.19134474748343802,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009964683672112728,
      "loss": 1.329,
      "step": 2224
    },
    {
      "epoch": 0.1914307837907597,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009964651582365547,
      "loss": 0.8972,
      "step": 2225
    },
    {
      "epoch": 0.19151682009808138,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009964619478097724,
      "loss": 1.2331,
      "step": 2226
    },
    {
      "epoch": 0.1916028564054031,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009964587359309355,
      "loss": 1.7714,
      "step": 2227
    },
    {
      "epoch": 0.19168889271272477,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009964555226000533,
      "loss": 1.0909,
      "step": 2228
    },
    {
      "epoch": 0.19177492902004645,
      "grad_norm": 1.25,
      "learning_rate": 0.0009964523078171352,
      "loss": 1.0515,
      "step": 2229
    },
    {
      "epoch": 0.19186096532736815,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009964490915821908,
      "loss": 1.5618,
      "step": 2230
    },
    {
      "epoch": 0.19194700163468983,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009964458738952293,
      "loss": 1.0656,
      "step": 2231
    },
    {
      "epoch": 0.19203303794201154,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009964426547562602,
      "loss": 1.5142,
      "step": 2232
    },
    {
      "epoch": 0.19211907424933322,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000996439434165293,
      "loss": 1.657,
      "step": 2233
    },
    {
      "epoch": 0.1922051105566549,
      "grad_norm": 1.9765625,
      "learning_rate": 0.000996436212122337,
      "loss": 1.1313,
      "step": 2234
    },
    {
      "epoch": 0.1922911468639766,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009964329886274013,
      "loss": 0.7904,
      "step": 2235
    },
    {
      "epoch": 0.1923771831712983,
      "grad_norm": 1.359375,
      "learning_rate": 0.000996429763680496,
      "loss": 0.5876,
      "step": 2236
    },
    {
      "epoch": 0.19246321947861997,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009964265372816301,
      "loss": 0.9357,
      "step": 2237
    },
    {
      "epoch": 0.19254925578594168,
      "grad_norm": 2.34375,
      "learning_rate": 0.000996423309430813,
      "loss": 1.6367,
      "step": 2238
    },
    {
      "epoch": 0.19263529209326336,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009964200801280544,
      "loss": 1.3126,
      "step": 2239
    },
    {
      "epoch": 0.19272132840058503,
      "grad_norm": 1.625,
      "learning_rate": 0.0009964168493733635,
      "loss": 1.7209,
      "step": 2240
    },
    {
      "epoch": 0.19280736470790674,
      "grad_norm": 1.515625,
      "learning_rate": 0.00099641361716675,
      "loss": 2.3193,
      "step": 2241
    },
    {
      "epoch": 0.19289340101522842,
      "grad_norm": 1.8515625,
      "learning_rate": 0.000996410383508223,
      "loss": 1.2703,
      "step": 2242
    },
    {
      "epoch": 0.19297943732255013,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009964071483977925,
      "loss": 1.52,
      "step": 2243
    },
    {
      "epoch": 0.1930654736298718,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009964039118354675,
      "loss": 2.3498,
      "step": 2244
    },
    {
      "epoch": 0.1931515099371935,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009964006738212574,
      "loss": 1.2443,
      "step": 2245
    },
    {
      "epoch": 0.1932375462445152,
      "grad_norm": 1.34375,
      "learning_rate": 0.000996397434355172,
      "loss": 1.072,
      "step": 2246
    },
    {
      "epoch": 0.19332358255183688,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009963941934372207,
      "loss": 0.846,
      "step": 2247
    },
    {
      "epoch": 0.19340961885915856,
      "grad_norm": 1.140625,
      "learning_rate": 0.000996390951067413,
      "loss": 1.732,
      "step": 2248
    },
    {
      "epoch": 0.19349565516648026,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009963877072457583,
      "loss": 1.1313,
      "step": 2249
    },
    {
      "epoch": 0.19358169147380194,
      "grad_norm": 1.546875,
      "learning_rate": 0.000996384461972266,
      "loss": 1.2814,
      "step": 2250
    },
    {
      "epoch": 0.19366772778112362,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009963812152469459,
      "loss": 1.2817,
      "step": 2251
    },
    {
      "epoch": 0.19375376408844533,
      "grad_norm": 1.390625,
      "learning_rate": 0.000996377967069807,
      "loss": 1.1945,
      "step": 2252
    },
    {
      "epoch": 0.193839800395767,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000996374717440859,
      "loss": 0.7891,
      "step": 2253
    },
    {
      "epoch": 0.19392583670308872,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009963714663601118,
      "loss": 1.5392,
      "step": 2254
    },
    {
      "epoch": 0.1940118730104104,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009963682138275744,
      "loss": 0.8462,
      "step": 2255
    },
    {
      "epoch": 0.19409790931773208,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009963649598432568,
      "loss": 2.0659,
      "step": 2256
    },
    {
      "epoch": 0.19418394562505378,
      "grad_norm": 1.375,
      "learning_rate": 0.0009963617044071679,
      "loss": 1.787,
      "step": 2257
    },
    {
      "epoch": 0.19426998193237546,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009963584475193177,
      "loss": 1.5874,
      "step": 2258
    },
    {
      "epoch": 0.19435601823969714,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009963551891797155,
      "loss": 1.5525,
      "step": 2259
    },
    {
      "epoch": 0.19444205454701885,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009963519293883709,
      "loss": 1.7933,
      "step": 2260
    },
    {
      "epoch": 0.19452809085434053,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009963486681452936,
      "loss": 2.1379,
      "step": 2261
    },
    {
      "epoch": 0.1946141271616622,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009963454054504929,
      "loss": 0.827,
      "step": 2262
    },
    {
      "epoch": 0.19470016346898392,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009963421413039784,
      "loss": 1.1359,
      "step": 2263
    },
    {
      "epoch": 0.1947861997763056,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009963388757057597,
      "loss": 1.2068,
      "step": 2264
    },
    {
      "epoch": 0.1948722360836273,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009963356086558463,
      "loss": 1.0174,
      "step": 2265
    },
    {
      "epoch": 0.19495827239094898,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009963323401542476,
      "loss": 0.9117,
      "step": 2266
    },
    {
      "epoch": 0.19504430869827066,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009963290702009737,
      "loss": 1.0042,
      "step": 2267
    },
    {
      "epoch": 0.19513034500559237,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009963257987960336,
      "loss": 1.0576,
      "step": 2268
    },
    {
      "epoch": 0.19521638131291405,
      "grad_norm": 1.28125,
      "learning_rate": 0.000996322525939437,
      "loss": 0.9236,
      "step": 2269
    },
    {
      "epoch": 0.19530241762023573,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009963192516311936,
      "loss": 1.3692,
      "step": 2270
    },
    {
      "epoch": 0.19538845392755744,
      "grad_norm": 1.625,
      "learning_rate": 0.000996315975871313,
      "loss": 1.4944,
      "step": 2271
    },
    {
      "epoch": 0.19547449023487912,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009963126986598046,
      "loss": 1.6201,
      "step": 2272
    },
    {
      "epoch": 0.1955605265422008,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009963094199966782,
      "loss": 1.5578,
      "step": 2273
    },
    {
      "epoch": 0.1956465628495225,
      "grad_norm": 1.3125,
      "learning_rate": 0.000996306139881943,
      "loss": 1.4665,
      "step": 2274
    },
    {
      "epoch": 0.19573259915684418,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009963028583156093,
      "loss": 1.0273,
      "step": 2275
    },
    {
      "epoch": 0.1958186354641659,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000996299575297686,
      "loss": 1.4874,
      "step": 2276
    },
    {
      "epoch": 0.19590467177148757,
      "grad_norm": 1.109375,
      "learning_rate": 0.000996296290828183,
      "loss": 1.5497,
      "step": 2277
    },
    {
      "epoch": 0.19599070807880925,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009962930049071097,
      "loss": 1.2378,
      "step": 2278
    },
    {
      "epoch": 0.19607674438613096,
      "grad_norm": 2.328125,
      "learning_rate": 0.000996289717534476,
      "loss": 1.7695,
      "step": 2279
    },
    {
      "epoch": 0.19616278069345264,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009962864287102915,
      "loss": 1.0697,
      "step": 2280
    },
    {
      "epoch": 0.19624881700077432,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009962831384345658,
      "loss": 1.4671,
      "step": 2281
    },
    {
      "epoch": 0.19633485330809602,
      "grad_norm": 1.375,
      "learning_rate": 0.0009962798467073083,
      "loss": 1.2785,
      "step": 2282
    },
    {
      "epoch": 0.1964208896154177,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009962765535285287,
      "loss": 2.0543,
      "step": 2283
    },
    {
      "epoch": 0.19650692592273938,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009962732588982368,
      "loss": 1.5185,
      "step": 2284
    },
    {
      "epoch": 0.1965929622300611,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009962699628164422,
      "loss": 1.1086,
      "step": 2285
    },
    {
      "epoch": 0.19667899853738277,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009962666652831543,
      "loss": 1.5795,
      "step": 2286
    },
    {
      "epoch": 0.19676503484470448,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009962633662983832,
      "loss": 1.4111,
      "step": 2287
    },
    {
      "epoch": 0.19685107115202616,
      "grad_norm": 1.84375,
      "learning_rate": 0.000996260065862138,
      "loss": 1.0584,
      "step": 2288
    },
    {
      "epoch": 0.19693710745934784,
      "grad_norm": 2.375,
      "learning_rate": 0.0009962567639744287,
      "loss": 1.2745,
      "step": 2289
    },
    {
      "epoch": 0.19702314376666955,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009962534606352649,
      "loss": 1.692,
      "step": 2290
    },
    {
      "epoch": 0.19710918007399122,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009962501558446561,
      "loss": 2.4219,
      "step": 2291
    },
    {
      "epoch": 0.1971952163813129,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009962468496026123,
      "loss": 1.9601,
      "step": 2292
    },
    {
      "epoch": 0.1972812526886346,
      "grad_norm": 1.484375,
      "learning_rate": 0.000996243541909143,
      "loss": 1.4303,
      "step": 2293
    },
    {
      "epoch": 0.1973672889959563,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009962402327642578,
      "loss": 1.4173,
      "step": 2294
    },
    {
      "epoch": 0.19745332530327797,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009962369221679663,
      "loss": 1.1337,
      "step": 2295
    },
    {
      "epoch": 0.19753936161059968,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009962336101202785,
      "loss": 1.87,
      "step": 2296
    },
    {
      "epoch": 0.19762539791792136,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009962302966212038,
      "loss": 1.445,
      "step": 2297
    },
    {
      "epoch": 0.19771143422524307,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000996226981670752,
      "loss": 1.3391,
      "step": 2298
    },
    {
      "epoch": 0.19779747053256475,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000996223665268933,
      "loss": 1.3577,
      "step": 2299
    },
    {
      "epoch": 0.19788350683988642,
      "grad_norm": 1.921875,
      "learning_rate": 0.000996220347415756,
      "loss": 1.8317,
      "step": 2300
    },
    {
      "epoch": 0.19796954314720813,
      "grad_norm": 1.09375,
      "learning_rate": 0.000996217028111231,
      "loss": 1.3173,
      "step": 2301
    },
    {
      "epoch": 0.1980555794545298,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009962137073553678,
      "loss": 0.9438,
      "step": 2302
    },
    {
      "epoch": 0.1981416157618515,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009962103851481761,
      "loss": 1.0311,
      "step": 2303
    },
    {
      "epoch": 0.1982276520691732,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009962070614896654,
      "loss": 1.0389,
      "step": 2304
    },
    {
      "epoch": 0.19831368837649488,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009962037363798457,
      "loss": 1.0999,
      "step": 2305
    },
    {
      "epoch": 0.19839972468381656,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0009962004098187265,
      "loss": 2.019,
      "step": 2306
    },
    {
      "epoch": 0.19848576099113827,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009961970818063175,
      "loss": 0.8971,
      "step": 2307
    },
    {
      "epoch": 0.19857179729845995,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009961937523426286,
      "loss": 1.3362,
      "step": 2308
    },
    {
      "epoch": 0.19865783360578165,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009961904214276696,
      "loss": 1.4708,
      "step": 2309
    },
    {
      "epoch": 0.19874386991310333,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00099618708906145,
      "loss": 1.6807,
      "step": 2310
    },
    {
      "epoch": 0.198829906220425,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009961837552439796,
      "loss": 1.3841,
      "step": 2311
    },
    {
      "epoch": 0.19891594252774672,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009961804199752685,
      "loss": 1.4561,
      "step": 2312
    },
    {
      "epoch": 0.1990019788350684,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009961770832553258,
      "loss": 1.3128,
      "step": 2313
    },
    {
      "epoch": 0.19908801514239008,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009961737450841618,
      "loss": 1.1692,
      "step": 2314
    },
    {
      "epoch": 0.1991740514497118,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000996170405461786,
      "loss": 2.0184,
      "step": 2315
    },
    {
      "epoch": 0.19926008775703347,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009961670643882086,
      "loss": 1.1658,
      "step": 2316
    },
    {
      "epoch": 0.19934612406435515,
      "grad_norm": 1.875,
      "learning_rate": 0.0009961637218634388,
      "loss": 2.0709,
      "step": 2317
    },
    {
      "epoch": 0.19943216037167685,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009961603778874866,
      "loss": 1.7427,
      "step": 2318
    },
    {
      "epoch": 0.19951819667899853,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009961570324603618,
      "loss": 1.621,
      "step": 2319
    },
    {
      "epoch": 0.19960423298632024,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009961536855820742,
      "loss": 1.7489,
      "step": 2320
    },
    {
      "epoch": 0.19969026929364192,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009961503372526337,
      "loss": 1.3938,
      "step": 2321
    },
    {
      "epoch": 0.1997763056009636,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009961469874720497,
      "loss": 1.2916,
      "step": 2322
    },
    {
      "epoch": 0.1998623419082853,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009961436362403325,
      "loss": 1.2572,
      "step": 2323
    },
    {
      "epoch": 0.199948378215607,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009961402835574915,
      "loss": 1.2002,
      "step": 2324
    },
    {
      "epoch": 0.20003441452292867,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009961369294235367,
      "loss": 0.8929,
      "step": 2325
    },
    {
      "epoch": 0.20012045083025037,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009961335738384778,
      "loss": 1.0565,
      "step": 2326
    },
    {
      "epoch": 0.20020648713757205,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009961302168023248,
      "loss": 2.0191,
      "step": 2327
    },
    {
      "epoch": 0.20029252344489373,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009961268583150874,
      "loss": 1.1527,
      "step": 2328
    },
    {
      "epoch": 0.20037855975221544,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009961234983767752,
      "loss": 1.205,
      "step": 2329
    },
    {
      "epoch": 0.20046459605953712,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009961201369873985,
      "loss": 1.0038,
      "step": 2330
    },
    {
      "epoch": 0.20055063236685883,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000996116774146967,
      "loss": 1.3839,
      "step": 2331
    },
    {
      "epoch": 0.2006366686741805,
      "grad_norm": 1.125,
      "learning_rate": 0.0009961134098554903,
      "loss": 1.5224,
      "step": 2332
    },
    {
      "epoch": 0.2007227049815022,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009961100441129782,
      "loss": 1.5582,
      "step": 2333
    },
    {
      "epoch": 0.2008087412888239,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000996106676919441,
      "loss": 0.9722,
      "step": 2334
    },
    {
      "epoch": 0.20089477759614557,
      "grad_norm": 1.546875,
      "learning_rate": 0.000996103308274888,
      "loss": 1.3019,
      "step": 2335
    },
    {
      "epoch": 0.20098081390346725,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009960999381793294,
      "loss": 1.4587,
      "step": 2336
    },
    {
      "epoch": 0.20106685021078896,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009960965666327751,
      "loss": 1.3966,
      "step": 2337
    },
    {
      "epoch": 0.20115288651811064,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009960931936352346,
      "loss": 1.2178,
      "step": 2338
    },
    {
      "epoch": 0.20123892282543232,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000996089819186718,
      "loss": 1.4678,
      "step": 2339
    },
    {
      "epoch": 0.20132495913275403,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009960864432872355,
      "loss": 1.2088,
      "step": 2340
    },
    {
      "epoch": 0.2014109954400757,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009960830659367964,
      "loss": 1.781,
      "step": 2341
    },
    {
      "epoch": 0.20149703174739741,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009960796871354108,
      "loss": 1.1246,
      "step": 2342
    },
    {
      "epoch": 0.2015830680547191,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009960763068830884,
      "loss": 1.6266,
      "step": 2343
    },
    {
      "epoch": 0.20166910436204077,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009960729251798396,
      "loss": 0.8574,
      "step": 2344
    },
    {
      "epoch": 0.20175514066936248,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000996069542025674,
      "loss": 1.1038,
      "step": 2345
    },
    {
      "epoch": 0.20184117697668416,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009960661574206012,
      "loss": 1.2063,
      "step": 2346
    },
    {
      "epoch": 0.20192721328400584,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009960627713646315,
      "loss": 1.1118,
      "step": 2347
    },
    {
      "epoch": 0.20201324959132755,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009960593838577747,
      "loss": 1.5859,
      "step": 2348
    },
    {
      "epoch": 0.20209928589864923,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009960559949000406,
      "loss": 1.4417,
      "step": 2349
    },
    {
      "epoch": 0.2021853222059709,
      "grad_norm": 2.5625,
      "learning_rate": 0.0009960526044914392,
      "loss": 1.2608,
      "step": 2350
    },
    {
      "epoch": 0.20227135851329262,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009960492126319808,
      "loss": 1.1587,
      "step": 2351
    },
    {
      "epoch": 0.2023573948206143,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009960458193216744,
      "loss": 1.3679,
      "step": 2352
    },
    {
      "epoch": 0.202443431127936,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009960424245605305,
      "loss": 2.1165,
      "step": 2353
    },
    {
      "epoch": 0.20252946743525768,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009960390283485594,
      "loss": 1.28,
      "step": 2354
    },
    {
      "epoch": 0.20261550374257936,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009960356306857703,
      "loss": 0.8611,
      "step": 2355
    },
    {
      "epoch": 0.20270154004990107,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009960322315721735,
      "loss": 1.3291,
      "step": 2356
    },
    {
      "epoch": 0.20278757635722275,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009960288310077788,
      "loss": 1.2757,
      "step": 2357
    },
    {
      "epoch": 0.20287361266454443,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009960254289925964,
      "loss": 1.9081,
      "step": 2358
    },
    {
      "epoch": 0.20295964897186614,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000996022025526636,
      "loss": 1.9903,
      "step": 2359
    },
    {
      "epoch": 0.20304568527918782,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009960186206099079,
      "loss": 1.532,
      "step": 2360
    },
    {
      "epoch": 0.2031317215865095,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009960152142424214,
      "loss": 1.6038,
      "step": 2361
    },
    {
      "epoch": 0.2032177578938312,
      "grad_norm": 2.328125,
      "learning_rate": 0.000996011806424187,
      "loss": 1.8807,
      "step": 2362
    },
    {
      "epoch": 0.20330379420115288,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009960083971552146,
      "loss": 0.9911,
      "step": 2363
    },
    {
      "epoch": 0.2033898305084746,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000996004986435514,
      "loss": 1.5988,
      "step": 2364
    },
    {
      "epoch": 0.20347586681579627,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009960015742650954,
      "loss": 0.981,
      "step": 2365
    },
    {
      "epoch": 0.20356190312311795,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009959981606439687,
      "loss": 1.4551,
      "step": 2366
    },
    {
      "epoch": 0.20364793943043966,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009959947455721438,
      "loss": 1.4862,
      "step": 2367
    },
    {
      "epoch": 0.20373397573776134,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009959913290496307,
      "loss": 1.7179,
      "step": 2368
    },
    {
      "epoch": 0.20382001204508302,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009959879110764396,
      "loss": 1.5869,
      "step": 2369
    },
    {
      "epoch": 0.20390604835240472,
      "grad_norm": 0.984375,
      "learning_rate": 0.00099598449165258,
      "loss": 1.8431,
      "step": 2370
    },
    {
      "epoch": 0.2039920846597264,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009959810707780624,
      "loss": 1.5458,
      "step": 2371
    },
    {
      "epoch": 0.20407812096704808,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009959776484528965,
      "loss": 2.0029,
      "step": 2372
    },
    {
      "epoch": 0.2041641572743698,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009959742246770929,
      "loss": 1.6486,
      "step": 2373
    },
    {
      "epoch": 0.20425019358169147,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009959707994506605,
      "loss": 1.5076,
      "step": 2374
    },
    {
      "epoch": 0.20433622988901318,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009959673727736105,
      "loss": 0.7813,
      "step": 2375
    },
    {
      "epoch": 0.20442226619633486,
      "grad_norm": 1.15625,
      "learning_rate": 0.000995963944645952,
      "loss": 1.333,
      "step": 2376
    },
    {
      "epoch": 0.20450830250365654,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000995960515067696,
      "loss": 0.9541,
      "step": 2377
    },
    {
      "epoch": 0.20459433881097824,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009959570840388514,
      "loss": 1.4432,
      "step": 2378
    },
    {
      "epoch": 0.20468037511829992,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000995953651559429,
      "loss": 1.0028,
      "step": 2379
    },
    {
      "epoch": 0.2047664114256216,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009959502176294383,
      "loss": 1.5182,
      "step": 2380
    },
    {
      "epoch": 0.2048524477329433,
      "grad_norm": 1.5,
      "learning_rate": 0.0009959467822488902,
      "loss": 1.6673,
      "step": 2381
    },
    {
      "epoch": 0.204938484040265,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009959433454177938,
      "loss": 2.1328,
      "step": 2382
    },
    {
      "epoch": 0.20502452034758667,
      "grad_norm": 1.625,
      "learning_rate": 0.00099593990713616,
      "loss": 1.8458,
      "step": 2383
    },
    {
      "epoch": 0.20511055665490838,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009959364674039979,
      "loss": 1.5135,
      "step": 2384
    },
    {
      "epoch": 0.20519659296223006,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009959330262213184,
      "loss": 1.47,
      "step": 2385
    },
    {
      "epoch": 0.20528262926955176,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009959295835881312,
      "loss": 1.5102,
      "step": 2386
    },
    {
      "epoch": 0.20536866557687344,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009959261395044465,
      "loss": 0.9967,
      "step": 2387
    },
    {
      "epoch": 0.20545470188419512,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000995922693970274,
      "loss": 1.3654,
      "step": 2388
    },
    {
      "epoch": 0.20554073819151683,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009959192469856245,
      "loss": 1.3448,
      "step": 2389
    },
    {
      "epoch": 0.2056267744988385,
      "grad_norm": 1.5,
      "learning_rate": 0.0009959157985505074,
      "loss": 1.0324,
      "step": 2390
    },
    {
      "epoch": 0.2057128108061602,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009959123486649331,
      "loss": 0.6071,
      "step": 2391
    },
    {
      "epoch": 0.2057988471134819,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009959088973289117,
      "loss": 1.2192,
      "step": 2392
    },
    {
      "epoch": 0.20588488342080358,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000995905444542453,
      "loss": 1.1369,
      "step": 2393
    },
    {
      "epoch": 0.20597091972812526,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009959019903055676,
      "loss": 1.3097,
      "step": 2394
    },
    {
      "epoch": 0.20605695603544696,
      "grad_norm": 1.28125,
      "learning_rate": 0.000995898534618265,
      "loss": 1.25,
      "step": 2395
    },
    {
      "epoch": 0.20614299234276864,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000995895077480556,
      "loss": 1.5689,
      "step": 2396
    },
    {
      "epoch": 0.20622902865009035,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009958916188924501,
      "loss": 1.5834,
      "step": 2397
    },
    {
      "epoch": 0.20631506495741203,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009958881588539577,
      "loss": 1.2209,
      "step": 2398
    },
    {
      "epoch": 0.2064011012647337,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000995884697365089,
      "loss": 1.3619,
      "step": 2399
    },
    {
      "epoch": 0.20648713757205542,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009958812344258538,
      "loss": 1.4839,
      "step": 2400
    },
    {
      "epoch": 0.2065731738793771,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009958777700362627,
      "loss": 1.1483,
      "step": 2401
    },
    {
      "epoch": 0.20665921018669878,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009958743041963251,
      "loss": 2.0592,
      "step": 2402
    },
    {
      "epoch": 0.20674524649402048,
      "grad_norm": 1.328125,
      "learning_rate": 0.000995870836906052,
      "loss": 1.5646,
      "step": 2403
    },
    {
      "epoch": 0.20683128280134216,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000995867368165453,
      "loss": 1.2696,
      "step": 2404
    },
    {
      "epoch": 0.20691731910866384,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009958638979745385,
      "loss": 2.0691,
      "step": 2405
    },
    {
      "epoch": 0.20700335541598555,
      "grad_norm": 1.625,
      "learning_rate": 0.0009958604263333186,
      "loss": 1.4372,
      "step": 2406
    },
    {
      "epoch": 0.20708939172330723,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009958569532418032,
      "loss": 2.0975,
      "step": 2407
    },
    {
      "epoch": 0.20717542803062894,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009958534787000027,
      "loss": 1.1607,
      "step": 2408
    },
    {
      "epoch": 0.20726146433795062,
      "grad_norm": 1.75,
      "learning_rate": 0.0009958500027079274,
      "loss": 1.0177,
      "step": 2409
    },
    {
      "epoch": 0.2073475006452723,
      "grad_norm": 1.125,
      "learning_rate": 0.000995846525265587,
      "loss": 1.1036,
      "step": 2410
    },
    {
      "epoch": 0.207433536952594,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000995843046372992,
      "loss": 1.4984,
      "step": 2411
    },
    {
      "epoch": 0.20751957325991569,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009958395660301528,
      "loss": 1.3564,
      "step": 2412
    },
    {
      "epoch": 0.20760560956723736,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009958360842370788,
      "loss": 1.4522,
      "step": 2413
    },
    {
      "epoch": 0.20769164587455907,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000995832600993781,
      "loss": 1.0303,
      "step": 2414
    },
    {
      "epoch": 0.20777768218188075,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009958291163002695,
      "loss": 1.8685,
      "step": 2415
    },
    {
      "epoch": 0.20786371848920243,
      "grad_norm": 1.53125,
      "learning_rate": 0.000995825630156554,
      "loss": 1.2032,
      "step": 2416
    },
    {
      "epoch": 0.20794975479652414,
      "grad_norm": 1.578125,
      "learning_rate": 0.000995822142562645,
      "loss": 2.0135,
      "step": 2417
    },
    {
      "epoch": 0.20803579110384582,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009958186535185525,
      "loss": 1.9187,
      "step": 2418
    },
    {
      "epoch": 0.20812182741116753,
      "grad_norm": 2.53125,
      "learning_rate": 0.0009958151630242872,
      "loss": 1.8309,
      "step": 2419
    },
    {
      "epoch": 0.2082078637184892,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009958116710798589,
      "loss": 1.3846,
      "step": 2420
    },
    {
      "epoch": 0.20829390002581089,
      "grad_norm": 3.90625,
      "learning_rate": 0.0009958081776852777,
      "loss": 1.6921,
      "step": 2421
    },
    {
      "epoch": 0.2083799363331326,
      "grad_norm": 1.40625,
      "learning_rate": 0.000995804682840554,
      "loss": 1.1902,
      "step": 2422
    },
    {
      "epoch": 0.20846597264045427,
      "grad_norm": 1.359375,
      "learning_rate": 0.000995801186545698,
      "loss": 0.7021,
      "step": 2423
    },
    {
      "epoch": 0.20855200894777595,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009957976888007203,
      "loss": 1.7891,
      "step": 2424
    },
    {
      "epoch": 0.20863804525509766,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009957941896056306,
      "loss": 1.4044,
      "step": 2425
    },
    {
      "epoch": 0.20872408156241934,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009957906889604393,
      "loss": 0.7725,
      "step": 2426
    },
    {
      "epoch": 0.20881011786974102,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009957871868651567,
      "loss": 1.2989,
      "step": 2427
    },
    {
      "epoch": 0.20889615417706273,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009957836833197928,
      "loss": 1.6882,
      "step": 2428
    },
    {
      "epoch": 0.2089821904843844,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009957801783243582,
      "loss": 1.6663,
      "step": 2429
    },
    {
      "epoch": 0.2090682267917061,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009957766718788632,
      "loss": 2.1194,
      "step": 2430
    },
    {
      "epoch": 0.2091542630990278,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009957731639833177,
      "loss": 1.3646,
      "step": 2431
    },
    {
      "epoch": 0.20924029940634947,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009957696546377321,
      "loss": 2.1116,
      "step": 2432
    },
    {
      "epoch": 0.20932633571367118,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009957661438421165,
      "loss": 1.0585,
      "step": 2433
    },
    {
      "epoch": 0.20941237202099286,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009957626315964816,
      "loss": 0.8343,
      "step": 2434
    },
    {
      "epoch": 0.20949840832831454,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009957591179008375,
      "loss": 1.7038,
      "step": 2435
    },
    {
      "epoch": 0.20958444463563625,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009957556027551941,
      "loss": 1.8485,
      "step": 2436
    },
    {
      "epoch": 0.20967048094295793,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009957520861595623,
      "loss": 1.2194,
      "step": 2437
    },
    {
      "epoch": 0.2097565172502796,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009957485681139518,
      "loss": 1.3447,
      "step": 2438
    },
    {
      "epoch": 0.2098425535576013,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009957450486183731,
      "loss": 1.2085,
      "step": 2439
    },
    {
      "epoch": 0.209928589864923,
      "grad_norm": 1.53125,
      "learning_rate": 0.000995741527672837,
      "loss": 1.2078,
      "step": 2440
    },
    {
      "epoch": 0.2100146261722447,
      "grad_norm": 1.75,
      "learning_rate": 0.0009957380052773531,
      "loss": 1.5502,
      "step": 2441
    },
    {
      "epoch": 0.21010066247956638,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009957344814319317,
      "loss": 1.9316,
      "step": 2442
    },
    {
      "epoch": 0.21018669878688806,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009957309561365838,
      "loss": 1.813,
      "step": 2443
    },
    {
      "epoch": 0.21027273509420977,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009957274293913188,
      "loss": 0.9861,
      "step": 2444
    },
    {
      "epoch": 0.21035877140153145,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009957239011961479,
      "loss": 1.9265,
      "step": 2445
    },
    {
      "epoch": 0.21044480770885313,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009957203715510808,
      "loss": 1.4387,
      "step": 2446
    },
    {
      "epoch": 0.21053084401617483,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009957168404561279,
      "loss": 1.249,
      "step": 2447
    },
    {
      "epoch": 0.2106168803234965,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009957133079112998,
      "loss": 2.2252,
      "step": 2448
    },
    {
      "epoch": 0.2107029166308182,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009957097739166068,
      "loss": 1.847,
      "step": 2449
    },
    {
      "epoch": 0.2107889529381399,
      "grad_norm": 1.140625,
      "learning_rate": 0.000995706238472059,
      "loss": 1.5146,
      "step": 2450
    },
    {
      "epoch": 0.21087498924546158,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009957027015776668,
      "loss": 1.499,
      "step": 2451
    },
    {
      "epoch": 0.2109610255527833,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009956991632334407,
      "loss": 1.6287,
      "step": 2452
    },
    {
      "epoch": 0.21104706186010497,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009956956234393908,
      "loss": 1.2856,
      "step": 2453
    },
    {
      "epoch": 0.21113309816742665,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009956920821955278,
      "loss": 1.1272,
      "step": 2454
    },
    {
      "epoch": 0.21121913447474835,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009956885395018618,
      "loss": 1.5143,
      "step": 2455
    },
    {
      "epoch": 0.21130517078207003,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009956849953584033,
      "loss": 1.8702,
      "step": 2456
    },
    {
      "epoch": 0.21139120708939171,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009956814497651626,
      "loss": 1.0896,
      "step": 2457
    },
    {
      "epoch": 0.21147724339671342,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009956779027221498,
      "loss": 1.5468,
      "step": 2458
    },
    {
      "epoch": 0.2115632797040351,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009956743542293759,
      "loss": 1.3923,
      "step": 2459
    },
    {
      "epoch": 0.21164931601135678,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009956708042868508,
      "loss": 1.3523,
      "step": 2460
    },
    {
      "epoch": 0.2117353523186785,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009956672528945848,
      "loss": 1.9572,
      "step": 2461
    },
    {
      "epoch": 0.21182138862600017,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009956637000525887,
      "loss": 0.9759,
      "step": 2462
    },
    {
      "epoch": 0.21190742493332188,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009956601457608725,
      "loss": 2.1507,
      "step": 2463
    },
    {
      "epoch": 0.21199346124064355,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000995656590019447,
      "loss": 1.5473,
      "step": 2464
    },
    {
      "epoch": 0.21207949754796523,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009956530328283223,
      "loss": 1.0342,
      "step": 2465
    },
    {
      "epoch": 0.21216553385528694,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009956494741875086,
      "loss": 0.8953,
      "step": 2466
    },
    {
      "epoch": 0.21225157016260862,
      "grad_norm": 1.453125,
      "learning_rate": 0.000995645914097017,
      "loss": 1.4696,
      "step": 2467
    },
    {
      "epoch": 0.2123376064699303,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009956423525568573,
      "loss": 1.608,
      "step": 2468
    },
    {
      "epoch": 0.212423642777252,
      "grad_norm": 1.484375,
      "learning_rate": 0.00099563878956704,
      "loss": 1.782,
      "step": 2469
    },
    {
      "epoch": 0.2125096790845737,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000995635225127576,
      "loss": 1.5263,
      "step": 2470
    },
    {
      "epoch": 0.21259571539189537,
      "grad_norm": 1.734375,
      "learning_rate": 0.000995631659238475,
      "loss": 2.055,
      "step": 2471
    },
    {
      "epoch": 0.21268175169921708,
      "grad_norm": 1.8046875,
      "learning_rate": 0.000995628091899748,
      "loss": 1.1533,
      "step": 2472
    },
    {
      "epoch": 0.21276778800653876,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009956245231114052,
      "loss": 1.9618,
      "step": 2473
    },
    {
      "epoch": 0.21285382431386046,
      "grad_norm": 1.53125,
      "learning_rate": 0.000995620952873457,
      "loss": 1.4532,
      "step": 2474
    },
    {
      "epoch": 0.21293986062118214,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009956173811859139,
      "loss": 1.7641,
      "step": 2475
    },
    {
      "epoch": 0.21302589692850382,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009956138080487865,
      "loss": 1.5326,
      "step": 2476
    },
    {
      "epoch": 0.21311193323582553,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009956102334620848,
      "loss": 1.2329,
      "step": 2477
    },
    {
      "epoch": 0.2131979695431472,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009956066574258197,
      "loss": 0.7051,
      "step": 2478
    },
    {
      "epoch": 0.2132840058504689,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009956030799400015,
      "loss": 1.3073,
      "step": 2479
    },
    {
      "epoch": 0.2133700421577906,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009955995010046408,
      "loss": 1.5807,
      "step": 2480
    },
    {
      "epoch": 0.21345607846511228,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000995595920619748,
      "loss": 1.3949,
      "step": 2481
    },
    {
      "epoch": 0.21354211477243396,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009955923387853333,
      "loss": 1.7524,
      "step": 2482
    },
    {
      "epoch": 0.21362815107975566,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009955887555014077,
      "loss": 1.3153,
      "step": 2483
    },
    {
      "epoch": 0.21371418738707734,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009955851707679811,
      "loss": 1.3091,
      "step": 2484
    },
    {
      "epoch": 0.21380022369439905,
      "grad_norm": 1.453125,
      "learning_rate": 0.000995581584585064,
      "loss": 1.6981,
      "step": 2485
    },
    {
      "epoch": 0.21388626000172073,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009955779969526676,
      "loss": 1.4065,
      "step": 2486
    },
    {
      "epoch": 0.2139722963090424,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000995574407870802,
      "loss": 1.835,
      "step": 2487
    },
    {
      "epoch": 0.21405833261636412,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009955708173394774,
      "loss": 1.6759,
      "step": 2488
    },
    {
      "epoch": 0.2141443689236858,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009955672253587047,
      "loss": 1.4245,
      "step": 2489
    },
    {
      "epoch": 0.21423040523100748,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009955636319284942,
      "loss": 1.5693,
      "step": 2490
    },
    {
      "epoch": 0.21431644153832918,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009955600370488565,
      "loss": 1.0143,
      "step": 2491
    },
    {
      "epoch": 0.21440247784565086,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009955564407198019,
      "loss": 1.5441,
      "step": 2492
    },
    {
      "epoch": 0.21448851415297254,
      "grad_norm": 1.421875,
      "learning_rate": 0.000995552842941341,
      "loss": 1.8301,
      "step": 2493
    },
    {
      "epoch": 0.21457455046029425,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009955492437134848,
      "loss": 1.7941,
      "step": 2494
    },
    {
      "epoch": 0.21466058676761593,
      "grad_norm": 1.5,
      "learning_rate": 0.0009955456430362432,
      "loss": 1.6579,
      "step": 2495
    },
    {
      "epoch": 0.21474662307493764,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000995542040909627,
      "loss": 1.2092,
      "step": 2496
    },
    {
      "epoch": 0.21483265938225932,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009955384373336467,
      "loss": 1.6946,
      "step": 2497
    },
    {
      "epoch": 0.214918695689581,
      "grad_norm": 1.625,
      "learning_rate": 0.000995534832308313,
      "loss": 1.4596,
      "step": 2498
    },
    {
      "epoch": 0.2150047319969027,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000995531225833636,
      "loss": 1.6322,
      "step": 2499
    },
    {
      "epoch": 0.21509076830422438,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009955276179096269,
      "loss": 1.1532,
      "step": 2500
    },
    {
      "epoch": 0.21517680461154606,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009955240085362956,
      "loss": 1.0677,
      "step": 2501
    },
    {
      "epoch": 0.21526284091886777,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000995520397713653,
      "loss": 0.9291,
      "step": 2502
    },
    {
      "epoch": 0.21534887722618945,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009955167854417096,
      "loss": 1.0766,
      "step": 2503
    },
    {
      "epoch": 0.21543491353351113,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009955131717204761,
      "loss": 1.5603,
      "step": 2504
    },
    {
      "epoch": 0.21552094984083284,
      "grad_norm": 2.25,
      "learning_rate": 0.000995509556549963,
      "loss": 1.7105,
      "step": 2505
    },
    {
      "epoch": 0.21560698614815452,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009955059399301806,
      "loss": 1.1561,
      "step": 2506
    },
    {
      "epoch": 0.21569302245547622,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009955023218611398,
      "loss": 1.8579,
      "step": 2507
    },
    {
      "epoch": 0.2157790587627979,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000995498702342851,
      "loss": 1.188,
      "step": 2508
    },
    {
      "epoch": 0.21586509507011958,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000995495081375325,
      "loss": 1.147,
      "step": 2509
    },
    {
      "epoch": 0.2159511313774413,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009954914589585721,
      "loss": 1.8571,
      "step": 2510
    },
    {
      "epoch": 0.21603716768476297,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009954878350926031,
      "loss": 0.7843,
      "step": 2511
    },
    {
      "epoch": 0.21612320399208465,
      "grad_norm": 1.5,
      "learning_rate": 0.0009954842097774285,
      "loss": 1.1288,
      "step": 2512
    },
    {
      "epoch": 0.21620924029940636,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009954805830130592,
      "loss": 1.0962,
      "step": 2513
    },
    {
      "epoch": 0.21629527660672804,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009954769547995054,
      "loss": 2.2962,
      "step": 2514
    },
    {
      "epoch": 0.21638131291404972,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009954733251367777,
      "loss": 1.2985,
      "step": 2515
    },
    {
      "epoch": 0.21646734922137142,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009954696940248874,
      "loss": 1.8639,
      "step": 2516
    },
    {
      "epoch": 0.2165533855286931,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009954660614638442,
      "loss": 1.4084,
      "step": 2517
    },
    {
      "epoch": 0.2166394218360148,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009954624274536592,
      "loss": 1.3538,
      "step": 2518
    },
    {
      "epoch": 0.2167254581433365,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009954587919943428,
      "loss": 1.5,
      "step": 2519
    },
    {
      "epoch": 0.21681149445065817,
      "grad_norm": 1.765625,
      "learning_rate": 0.000995455155085906,
      "loss": 1.3478,
      "step": 2520
    },
    {
      "epoch": 0.21689753075797988,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009954515167283592,
      "loss": 0.9657,
      "step": 2521
    },
    {
      "epoch": 0.21698356706530156,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000995447876921713,
      "loss": 1.319,
      "step": 2522
    },
    {
      "epoch": 0.21706960337262324,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009954442356659783,
      "loss": 1.4377,
      "step": 2523
    },
    {
      "epoch": 0.21715563967994495,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009954405929611654,
      "loss": 1.186,
      "step": 2524
    },
    {
      "epoch": 0.21724167598726662,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009954369488072853,
      "loss": 1.0344,
      "step": 2525
    },
    {
      "epoch": 0.2173277122945883,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009954333032043481,
      "loss": 1.4659,
      "step": 2526
    },
    {
      "epoch": 0.21741374860191,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009954296561523652,
      "loss": 2.0884,
      "step": 2527
    },
    {
      "epoch": 0.2174997849092317,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009954260076513468,
      "loss": 1.6638,
      "step": 2528
    },
    {
      "epoch": 0.2175858212165534,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009954223577013038,
      "loss": 1.214,
      "step": 2529
    },
    {
      "epoch": 0.21767185752387508,
      "grad_norm": 4.3125,
      "learning_rate": 0.0009954187063022466,
      "loss": 1.6771,
      "step": 2530
    },
    {
      "epoch": 0.21775789383119676,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000995415053454186,
      "loss": 0.9163,
      "step": 2531
    },
    {
      "epoch": 0.21784393013851847,
      "grad_norm": 1.265625,
      "learning_rate": 0.000995411399157133,
      "loss": 0.9736,
      "step": 2532
    },
    {
      "epoch": 0.21792996644584015,
      "grad_norm": 1.125,
      "learning_rate": 0.0009954077434110977,
      "loss": 1.5473,
      "step": 2533
    },
    {
      "epoch": 0.21801600275316182,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009954040862160912,
      "loss": 1.3958,
      "step": 2534
    },
    {
      "epoch": 0.21810203906048353,
      "grad_norm": 1.625,
      "learning_rate": 0.0009954004275721242,
      "loss": 1.3307,
      "step": 2535
    },
    {
      "epoch": 0.2181880753678052,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009953967674792073,
      "loss": 1.8436,
      "step": 2536
    },
    {
      "epoch": 0.2182741116751269,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009953931059373511,
      "loss": 1.4022,
      "step": 2537
    },
    {
      "epoch": 0.2183601479824486,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009953894429465663,
      "loss": 1.3761,
      "step": 2538
    },
    {
      "epoch": 0.21844618428977028,
      "grad_norm": 2.296875,
      "learning_rate": 0.000995385778506864,
      "loss": 2.4497,
      "step": 2539
    },
    {
      "epoch": 0.218532220597092,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009953821126182545,
      "loss": 1.1148,
      "step": 2540
    },
    {
      "epoch": 0.21861825690441367,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009953784452807486,
      "loss": 1.4784,
      "step": 2541
    },
    {
      "epoch": 0.21870429321173535,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009953747764943574,
      "loss": 1.287,
      "step": 2542
    },
    {
      "epoch": 0.21879032951905705,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009953711062590908,
      "loss": 1.8788,
      "step": 2543
    },
    {
      "epoch": 0.21887636582637873,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009953674345749605,
      "loss": 1.5742,
      "step": 2544
    },
    {
      "epoch": 0.2189624021337004,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009953637614419765,
      "loss": 1.0064,
      "step": 2545
    },
    {
      "epoch": 0.21904843844102212,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009953600868601502,
      "loss": 0.8689,
      "step": 2546
    },
    {
      "epoch": 0.2191344747483438,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009953564108294917,
      "loss": 1.5687,
      "step": 2547
    },
    {
      "epoch": 0.21922051105566548,
      "grad_norm": 1.25,
      "learning_rate": 0.0009953527333500121,
      "loss": 1.1842,
      "step": 2548
    },
    {
      "epoch": 0.2193065473629872,
      "grad_norm": 1.640625,
      "learning_rate": 0.000995349054421722,
      "loss": 1.7477,
      "step": 2549
    },
    {
      "epoch": 0.21939258367030887,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009953453740446326,
      "loss": 1.4563,
      "step": 2550
    },
    {
      "epoch": 0.21947861997763057,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000995341692218754,
      "loss": 1.216,
      "step": 2551
    },
    {
      "epoch": 0.21956465628495225,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009953380089440973,
      "loss": 1.6152,
      "step": 2552
    },
    {
      "epoch": 0.21965069259227393,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009953343242206734,
      "loss": 1.5435,
      "step": 2553
    },
    {
      "epoch": 0.21973672889959564,
      "grad_norm": 1.671875,
      "learning_rate": 0.000995330638048493,
      "loss": 1.6911,
      "step": 2554
    },
    {
      "epoch": 0.21982276520691732,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009953269504275667,
      "loss": 1.7413,
      "step": 2555
    },
    {
      "epoch": 0.219908801514239,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009953232613579054,
      "loss": 1.5139,
      "step": 2556
    },
    {
      "epoch": 0.2199948378215607,
      "grad_norm": 1.484375,
      "learning_rate": 0.00099531957083952,
      "loss": 0.9882,
      "step": 2557
    },
    {
      "epoch": 0.2200808741288824,
      "grad_norm": 1.6875,
      "learning_rate": 0.000995315878872421,
      "loss": 1.5144,
      "step": 2558
    },
    {
      "epoch": 0.22016691043620407,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009953121854566194,
      "loss": 2.1671,
      "step": 2559
    },
    {
      "epoch": 0.22025294674352577,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000995308490592126,
      "loss": 1.6311,
      "step": 2560
    },
    {
      "epoch": 0.22033898305084745,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009953047942789517,
      "loss": 0.9593,
      "step": 2561
    },
    {
      "epoch": 0.22042501935816916,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009953010965171071,
      "loss": 0.8574,
      "step": 2562
    },
    {
      "epoch": 0.22051105566549084,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009952973973066032,
      "loss": 1.3653,
      "step": 2563
    },
    {
      "epoch": 0.22059709197281252,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009952936966474508,
      "loss": 1.988,
      "step": 2564
    },
    {
      "epoch": 0.22068312828013423,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009952899945396608,
      "loss": 1.7887,
      "step": 2565
    },
    {
      "epoch": 0.2207691645874559,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009952862909832435,
      "loss": 0.8162,
      "step": 2566
    },
    {
      "epoch": 0.2208552008947776,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009952825859782105,
      "loss": 1.5201,
      "step": 2567
    },
    {
      "epoch": 0.2209412372020993,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000995278879524572,
      "loss": 1.6459,
      "step": 2568
    },
    {
      "epoch": 0.22102727350942097,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009952751716223393,
      "loss": 1.6023,
      "step": 2569
    },
    {
      "epoch": 0.22111330981674265,
      "grad_norm": 2.65625,
      "learning_rate": 0.000995271462271523,
      "loss": 1.3017,
      "step": 2570
    },
    {
      "epoch": 0.22119934612406436,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009952677514721339,
      "loss": 1.4961,
      "step": 2571
    },
    {
      "epoch": 0.22128538243138604,
      "grad_norm": 1.53125,
      "learning_rate": 0.000995264039224183,
      "loss": 1.2912,
      "step": 2572
    },
    {
      "epoch": 0.22137141873870775,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009952603255276812,
      "loss": 0.9609,
      "step": 2573
    },
    {
      "epoch": 0.22145745504602943,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000995256610382639,
      "loss": 2.1029,
      "step": 2574
    },
    {
      "epoch": 0.2215434913533511,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009952528937890678,
      "loss": 1.0749,
      "step": 2575
    },
    {
      "epoch": 0.22162952766067281,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000995249175746978,
      "loss": 1.2172,
      "step": 2576
    },
    {
      "epoch": 0.2217155639679945,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009952454562563808,
      "loss": 1.2727,
      "step": 2577
    },
    {
      "epoch": 0.22180160027531617,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000995241735317287,
      "loss": 1.6968,
      "step": 2578
    },
    {
      "epoch": 0.22188763658263788,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009952380129297073,
      "loss": 1.6872,
      "step": 2579
    },
    {
      "epoch": 0.22197367288995956,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009952342890936529,
      "loss": 2.1491,
      "step": 2580
    },
    {
      "epoch": 0.22205970919728124,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009952305638091345,
      "loss": 1.1965,
      "step": 2581
    },
    {
      "epoch": 0.22214574550460295,
      "grad_norm": 0.984375,
      "learning_rate": 0.000995226837076163,
      "loss": 1.0916,
      "step": 2582
    },
    {
      "epoch": 0.22223178181192463,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009952231088947493,
      "loss": 1.4783,
      "step": 2583
    },
    {
      "epoch": 0.22231781811924634,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009952193792649043,
      "loss": 0.9796,
      "step": 2584
    },
    {
      "epoch": 0.22240385442656802,
      "grad_norm": 1.3125,
      "learning_rate": 0.000995215648186639,
      "loss": 1.0821,
      "step": 2585
    },
    {
      "epoch": 0.2224898907338897,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009952119156599643,
      "loss": 1.8489,
      "step": 2586
    },
    {
      "epoch": 0.2225759270412114,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000995208181684891,
      "loss": 1.4206,
      "step": 2587
    },
    {
      "epoch": 0.22266196334853308,
      "grad_norm": 1.734375,
      "learning_rate": 0.00099520444626143,
      "loss": 1.515,
      "step": 2588
    },
    {
      "epoch": 0.22274799965585476,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009952007093895921,
      "loss": 0.9108,
      "step": 2589
    },
    {
      "epoch": 0.22283403596317647,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009951969710693886,
      "loss": 2.2452,
      "step": 2590
    },
    {
      "epoch": 0.22292007227049815,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009951932313008306,
      "loss": 1.1248,
      "step": 2591
    },
    {
      "epoch": 0.22300610857781983,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009951894900839284,
      "loss": 1.0174,
      "step": 2592
    },
    {
      "epoch": 0.22309214488514154,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009951857474186931,
      "loss": 1.3124,
      "step": 2593
    },
    {
      "epoch": 0.22317818119246322,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009951820033051362,
      "loss": 1.6577,
      "step": 2594
    },
    {
      "epoch": 0.22326421749978492,
      "grad_norm": 1.859375,
      "learning_rate": 0.000995178257743268,
      "loss": 1.6335,
      "step": 2595
    },
    {
      "epoch": 0.2233502538071066,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009951745107330996,
      "loss": 0.9055,
      "step": 2596
    },
    {
      "epoch": 0.22343629011442828,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009951707622746424,
      "loss": 1.1978,
      "step": 2597
    },
    {
      "epoch": 0.22352232642175,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009951670123679068,
      "loss": 1.3685,
      "step": 2598
    },
    {
      "epoch": 0.22360836272907167,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000995163261012904,
      "loss": 1.5263,
      "step": 2599
    },
    {
      "epoch": 0.22369439903639335,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000995159508209645,
      "loss": 1.0061,
      "step": 2600
    },
    {
      "epoch": 0.22378043534371506,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009951557539581409,
      "loss": 0.7975,
      "step": 2601
    },
    {
      "epoch": 0.22386647165103674,
      "grad_norm": 0.828125,
      "learning_rate": 0.0009951519982584023,
      "loss": 0.6246,
      "step": 2602
    },
    {
      "epoch": 0.22395250795835842,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009951482411104404,
      "loss": 1.2573,
      "step": 2603
    },
    {
      "epoch": 0.22403854426568012,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009951444825142663,
      "loss": 1.7924,
      "step": 2604
    },
    {
      "epoch": 0.2241245805730018,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009951407224698908,
      "loss": 1.8444,
      "step": 2605
    },
    {
      "epoch": 0.2242106168803235,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000995136960977325,
      "loss": 1.3692,
      "step": 2606
    },
    {
      "epoch": 0.2242966531876452,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00099513319803658,
      "loss": 0.863,
      "step": 2607
    },
    {
      "epoch": 0.22438268949496687,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009951294336476666,
      "loss": 1.7101,
      "step": 2608
    },
    {
      "epoch": 0.22446872580228858,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000995125667810596,
      "loss": 1.3334,
      "step": 2609
    },
    {
      "epoch": 0.22455476210961026,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000995121900525379,
      "loss": 1.3525,
      "step": 2610
    },
    {
      "epoch": 0.22464079841693194,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009951181317920269,
      "loss": 1.1231,
      "step": 2611
    },
    {
      "epoch": 0.22472683472425364,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009951143616105502,
      "loss": 1.2215,
      "step": 2612
    },
    {
      "epoch": 0.22481287103157532,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009951105899809606,
      "loss": 1.5576,
      "step": 2613
    },
    {
      "epoch": 0.224898907338897,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009951068169032687,
      "loss": 1.5527,
      "step": 2614
    },
    {
      "epoch": 0.2249849436462187,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009951030423774859,
      "loss": 1.3674,
      "step": 2615
    },
    {
      "epoch": 0.2250709799535404,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009950992664036226,
      "loss": 1.4924,
      "step": 2616
    },
    {
      "epoch": 0.2251570162608621,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009950954889816905,
      "loss": 1.0033,
      "step": 2617
    },
    {
      "epoch": 0.22524305256818378,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009950917101117004,
      "loss": 1.0067,
      "step": 2618
    },
    {
      "epoch": 0.22532908887550546,
      "grad_norm": 1.9453125,
      "learning_rate": 0.000995087929793663,
      "loss": 1.292,
      "step": 2619
    },
    {
      "epoch": 0.22541512518282716,
      "grad_norm": 1.078125,
      "learning_rate": 0.00099508414802759,
      "loss": 1.1357,
      "step": 2620
    },
    {
      "epoch": 0.22550116149014884,
      "grad_norm": 1.609375,
      "learning_rate": 0.000995080364813492,
      "loss": 1.2966,
      "step": 2621
    },
    {
      "epoch": 0.22558719779747052,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009950765801513806,
      "loss": 1.495,
      "step": 2622
    },
    {
      "epoch": 0.22567323410479223,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009950727940412662,
      "loss": 1.9012,
      "step": 2623
    },
    {
      "epoch": 0.2257592704121139,
      "grad_norm": 1.34375,
      "learning_rate": 0.00099506900648316,
      "loss": 1.2353,
      "step": 2624
    },
    {
      "epoch": 0.2258453067194356,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009950652174770735,
      "loss": 1.1438,
      "step": 2625
    },
    {
      "epoch": 0.2259313430267573,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009950614270230174,
      "loss": 1.1259,
      "step": 2626
    },
    {
      "epoch": 0.22601737933407898,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000995057635121003,
      "loss": 0.8381,
      "step": 2627
    },
    {
      "epoch": 0.22610341564140068,
      "grad_norm": 2.0,
      "learning_rate": 0.0009950538417710414,
      "loss": 1.8625,
      "step": 2628
    },
    {
      "epoch": 0.22618945194872236,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009950500469731434,
      "loss": 1.6957,
      "step": 2629
    },
    {
      "epoch": 0.22627548825604404,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009950462507273203,
      "loss": 1.5322,
      "step": 2630
    },
    {
      "epoch": 0.22636152456336575,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000995042453033583,
      "loss": 0.843,
      "step": 2631
    },
    {
      "epoch": 0.22644756087068743,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009950386538919432,
      "loss": 1.4718,
      "step": 2632
    },
    {
      "epoch": 0.2265335971780091,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009950348533024115,
      "loss": 1.4814,
      "step": 2633
    },
    {
      "epoch": 0.22661963348533082,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009950310512649991,
      "loss": 1.4537,
      "step": 2634
    },
    {
      "epoch": 0.2267056697926525,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000995027247779717,
      "loss": 1.5605,
      "step": 2635
    },
    {
      "epoch": 0.22679170609997418,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009950234428465767,
      "loss": 1.6846,
      "step": 2636
    },
    {
      "epoch": 0.22687774240729588,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009950196364655892,
      "loss": 1.9192,
      "step": 2637
    },
    {
      "epoch": 0.22696377871461756,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009950158286367652,
      "loss": 1.1493,
      "step": 2638
    },
    {
      "epoch": 0.22704981502193927,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009950120193601164,
      "loss": 1.8081,
      "step": 2639
    },
    {
      "epoch": 0.22713585132926095,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009950082086356538,
      "loss": 1.1995,
      "step": 2640
    },
    {
      "epoch": 0.22722188763658263,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009950043964633882,
      "loss": 1.1769,
      "step": 2641
    },
    {
      "epoch": 0.22730792394390434,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009950005828433311,
      "loss": 1.2974,
      "step": 2642
    },
    {
      "epoch": 0.22739396025122602,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009949967677754937,
      "loss": 1.3724,
      "step": 2643
    },
    {
      "epoch": 0.2274799965585477,
      "grad_norm": 1.171875,
      "learning_rate": 0.000994992951259887,
      "loss": 1.3727,
      "step": 2644
    },
    {
      "epoch": 0.2275660328658694,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000994989133296522,
      "loss": 2.0766,
      "step": 2645
    },
    {
      "epoch": 0.22765206917319109,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009949853138854102,
      "loss": 0.9956,
      "step": 2646
    },
    {
      "epoch": 0.22773810548051276,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009949814930265625,
      "loss": 1.2041,
      "step": 2647
    },
    {
      "epoch": 0.22782414178783447,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009949776707199904,
      "loss": 1.5713,
      "step": 2648
    },
    {
      "epoch": 0.22791017809515615,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009949738469657047,
      "loss": 1.37,
      "step": 2649
    },
    {
      "epoch": 0.22799621440247786,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009949700217637169,
      "loss": 1.7675,
      "step": 2650
    },
    {
      "epoch": 0.22808225070979954,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009949661951140381,
      "loss": 1.7051,
      "step": 2651
    },
    {
      "epoch": 0.22816828701712122,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009949623670166795,
      "loss": 1.5938,
      "step": 2652
    },
    {
      "epoch": 0.22825432332444293,
      "grad_norm": 1.53125,
      "learning_rate": 0.000994958537471652,
      "loss": 0.9167,
      "step": 2653
    },
    {
      "epoch": 0.2283403596317646,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000994954706478967,
      "loss": 1.2967,
      "step": 2654
    },
    {
      "epoch": 0.22842639593908629,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009949508740386358,
      "loss": 1.4083,
      "step": 2655
    },
    {
      "epoch": 0.228512432246408,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009949470401506698,
      "loss": 1.8419,
      "step": 2656
    },
    {
      "epoch": 0.22859846855372967,
      "grad_norm": 1.359375,
      "learning_rate": 0.00099494320481508,
      "loss": 1.4755,
      "step": 2657
    },
    {
      "epoch": 0.22868450486105135,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009949393680318771,
      "loss": 0.7948,
      "step": 2658
    },
    {
      "epoch": 0.22877054116837306,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009949355298010732,
      "loss": 1.1576,
      "step": 2659
    },
    {
      "epoch": 0.22885657747569474,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000994931690122679,
      "loss": 1.3769,
      "step": 2660
    },
    {
      "epoch": 0.22894261378301645,
      "grad_norm": 1.625,
      "learning_rate": 0.0009949278489967059,
      "loss": 2.4581,
      "step": 2661
    },
    {
      "epoch": 0.22902865009033813,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000994924006423165,
      "loss": 1.8687,
      "step": 2662
    },
    {
      "epoch": 0.2291146863976598,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009949201624020678,
      "loss": 1.0935,
      "step": 2663
    },
    {
      "epoch": 0.2292007227049815,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009949163169334255,
      "loss": 1.9687,
      "step": 2664
    },
    {
      "epoch": 0.2292867590123032,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000994912470017249,
      "loss": 1.6774,
      "step": 2665
    },
    {
      "epoch": 0.22937279531962487,
      "grad_norm": 1.171875,
      "learning_rate": 0.00099490862165355,
      "loss": 1.469,
      "step": 2666
    },
    {
      "epoch": 0.22945883162694658,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009949047718423392,
      "loss": 1.2842,
      "step": 2667
    },
    {
      "epoch": 0.22954486793426826,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009949009205836284,
      "loss": 1.2149,
      "step": 2668
    },
    {
      "epoch": 0.22963090424158994,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009948970678774285,
      "loss": 1.3955,
      "step": 2669
    },
    {
      "epoch": 0.22971694054891165,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000994893213723751,
      "loss": 1.3205,
      "step": 2670
    },
    {
      "epoch": 0.22980297685623333,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009948893581226072,
      "loss": 1.2058,
      "step": 2671
    },
    {
      "epoch": 0.22988901316355503,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009948855010740083,
      "loss": 1.4584,
      "step": 2672
    },
    {
      "epoch": 0.2299750494708767,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009948816425779656,
      "loss": 1.3785,
      "step": 2673
    },
    {
      "epoch": 0.2300610857781984,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009948777826344901,
      "loss": 1.1879,
      "step": 2674
    },
    {
      "epoch": 0.2301471220855201,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009948739212435936,
      "loss": 1.6349,
      "step": 2675
    },
    {
      "epoch": 0.23023315839284178,
      "grad_norm": 1.484375,
      "learning_rate": 0.000994870058405287,
      "loss": 1.3476,
      "step": 2676
    },
    {
      "epoch": 0.23031919470016346,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009948661941195816,
      "loss": 1.5996,
      "step": 2677
    },
    {
      "epoch": 0.23040523100748517,
      "grad_norm": 1.296875,
      "learning_rate": 0.000994862328386489,
      "loss": 1.6789,
      "step": 2678
    },
    {
      "epoch": 0.23049126731480685,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009948584612060204,
      "loss": 0.8837,
      "step": 2679
    },
    {
      "epoch": 0.23057730362212853,
      "grad_norm": 1.296875,
      "learning_rate": 0.000994854592578187,
      "loss": 1.808,
      "step": 2680
    },
    {
      "epoch": 0.23066333992945023,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000994850722503,
      "loss": 1.4591,
      "step": 2681
    },
    {
      "epoch": 0.2307493762367719,
      "grad_norm": 1.625,
      "learning_rate": 0.000994846850980471,
      "loss": 1.221,
      "step": 2682
    },
    {
      "epoch": 0.23083541254409362,
      "grad_norm": 1.109375,
      "learning_rate": 0.000994842978010611,
      "loss": 1.4173,
      "step": 2683
    },
    {
      "epoch": 0.2309214488514153,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009948391035934318,
      "loss": 1.5526,
      "step": 2684
    },
    {
      "epoch": 0.23100748515873698,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009948352277289443,
      "loss": 1.2824,
      "step": 2685
    },
    {
      "epoch": 0.2310935214660587,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009948313504171602,
      "loss": 1.4135,
      "step": 2686
    },
    {
      "epoch": 0.23117955777338037,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009948274716580906,
      "loss": 1.7917,
      "step": 2687
    },
    {
      "epoch": 0.23126559408070205,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009948235914517467,
      "loss": 0.9353,
      "step": 2688
    },
    {
      "epoch": 0.23135163038802375,
      "grad_norm": 1.28125,
      "learning_rate": 0.00099481970979814,
      "loss": 1.4902,
      "step": 2689
    },
    {
      "epoch": 0.23143766669534543,
      "grad_norm": 1.59375,
      "learning_rate": 0.000994815826697282,
      "loss": 1.178,
      "step": 2690
    },
    {
      "epoch": 0.23152370300266711,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000994811942149184,
      "loss": 0.8653,
      "step": 2691
    },
    {
      "epoch": 0.23160973930998882,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009948080561538572,
      "loss": 0.6565,
      "step": 2692
    },
    {
      "epoch": 0.2316957756173105,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000994804168711313,
      "loss": 1.4628,
      "step": 2693
    },
    {
      "epoch": 0.2317818119246322,
      "grad_norm": 1.75,
      "learning_rate": 0.0009948002798215632,
      "loss": 1.4992,
      "step": 2694
    },
    {
      "epoch": 0.2318678482319539,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009947963894846187,
      "loss": 2.371,
      "step": 2695
    },
    {
      "epoch": 0.23195388453927557,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009947924977004908,
      "loss": 1.6279,
      "step": 2696
    },
    {
      "epoch": 0.23203992084659728,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009947886044691912,
      "loss": 1.7296,
      "step": 2697
    },
    {
      "epoch": 0.23212595715391895,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000994784709790731,
      "loss": 2.136,
      "step": 2698
    },
    {
      "epoch": 0.23221199346124063,
      "grad_norm": 1.1875,
      "learning_rate": 0.000994780813665122,
      "loss": 1.6365,
      "step": 2699
    },
    {
      "epoch": 0.23229802976856234,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009947769160923753,
      "loss": 0.9445,
      "step": 2700
    },
    {
      "epoch": 0.23238406607588402,
      "grad_norm": 1.25,
      "learning_rate": 0.000994773017072502,
      "loss": 1.6707,
      "step": 2701
    },
    {
      "epoch": 0.2324701023832057,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009947691166055143,
      "loss": 1.5639,
      "step": 2702
    },
    {
      "epoch": 0.2325561386905274,
      "grad_norm": 1.796875,
      "learning_rate": 0.000994765214691423,
      "loss": 1.2537,
      "step": 2703
    },
    {
      "epoch": 0.2326421749978491,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009947613113302396,
      "loss": 1.2288,
      "step": 2704
    },
    {
      "epoch": 0.2327282113051708,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009947574065219756,
      "loss": 1.8079,
      "step": 2705
    },
    {
      "epoch": 0.23281424761249248,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009947535002666425,
      "loss": 0.9051,
      "step": 2706
    },
    {
      "epoch": 0.23290028391981416,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009947495925642516,
      "loss": 1.355,
      "step": 2707
    },
    {
      "epoch": 0.23298632022713586,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009947456834148143,
      "loss": 1.4042,
      "step": 2708
    },
    {
      "epoch": 0.23307235653445754,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009947417728183422,
      "loss": 1.1101,
      "step": 2709
    },
    {
      "epoch": 0.23315839284177922,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009947378607748467,
      "loss": 1.141,
      "step": 2710
    },
    {
      "epoch": 0.23324442914910093,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000994733947284339,
      "loss": 1.137,
      "step": 2711
    },
    {
      "epoch": 0.2333304654564226,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009947300323468308,
      "loss": 1.2273,
      "step": 2712
    },
    {
      "epoch": 0.2334165017637443,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009947261159623335,
      "loss": 1.6782,
      "step": 2713
    },
    {
      "epoch": 0.233502538071066,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009947221981308585,
      "loss": 1.0356,
      "step": 2714
    },
    {
      "epoch": 0.23358857437838768,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009947182788524172,
      "loss": 1.3509,
      "step": 2715
    },
    {
      "epoch": 0.23367461068570938,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009947143581270214,
      "loss": 1.6254,
      "step": 2716
    },
    {
      "epoch": 0.23376064699303106,
      "grad_norm": 1.484375,
      "learning_rate": 0.000994710435954682,
      "loss": 1.3048,
      "step": 2717
    },
    {
      "epoch": 0.23384668330035274,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000994706512335411,
      "loss": 2.1373,
      "step": 2718
    },
    {
      "epoch": 0.23393271960767445,
      "grad_norm": 1.375,
      "learning_rate": 0.0009947025872692197,
      "loss": 1.9939,
      "step": 2719
    },
    {
      "epoch": 0.23401875591499613,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009946986607561195,
      "loss": 1.4723,
      "step": 2720
    },
    {
      "epoch": 0.2341047922223178,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009946947327961217,
      "loss": 2.076,
      "step": 2721
    },
    {
      "epoch": 0.23419082852963952,
      "grad_norm": 1.5,
      "learning_rate": 0.0009946908033892383,
      "loss": 1.0492,
      "step": 2722
    },
    {
      "epoch": 0.2342768648369612,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009946868725354806,
      "loss": 1.2903,
      "step": 2723
    },
    {
      "epoch": 0.23436290114428288,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009946829402348598,
      "loss": 2.4902,
      "step": 2724
    },
    {
      "epoch": 0.23444893745160458,
      "grad_norm": 1.25,
      "learning_rate": 0.0009946790064873877,
      "loss": 1.5829,
      "step": 2725
    },
    {
      "epoch": 0.23453497375892626,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009946750712930756,
      "loss": 1.3391,
      "step": 2726
    },
    {
      "epoch": 0.23462101006624797,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009946711346519351,
      "loss": 1.5743,
      "step": 2727
    },
    {
      "epoch": 0.23470704637356965,
      "grad_norm": 1.875,
      "learning_rate": 0.0009946671965639778,
      "loss": 1.403,
      "step": 2728
    },
    {
      "epoch": 0.23479308268089133,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009946632570292152,
      "loss": 1.2735,
      "step": 2729
    },
    {
      "epoch": 0.23487911898821304,
      "grad_norm": 1.375,
      "learning_rate": 0.0009946593160476587,
      "loss": 1.5373,
      "step": 2730
    },
    {
      "epoch": 0.23496515529553472,
      "grad_norm": 1.546875,
      "learning_rate": 0.00099465537361932,
      "loss": 1.8457,
      "step": 2731
    },
    {
      "epoch": 0.2350511916028564,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009946514297442105,
      "loss": 1.672,
      "step": 2732
    },
    {
      "epoch": 0.2351372279101781,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009946474844223418,
      "loss": 1.8285,
      "step": 2733
    },
    {
      "epoch": 0.23522326421749978,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009946435376537255,
      "loss": 2.2535,
      "step": 2734
    },
    {
      "epoch": 0.23530930052482146,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009946395894383727,
      "loss": 1.1187,
      "step": 2735
    },
    {
      "epoch": 0.23539533683214317,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009946356397762958,
      "loss": 0.9283,
      "step": 2736
    },
    {
      "epoch": 0.23548137313946485,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009946316886675056,
      "loss": 1.2247,
      "step": 2737
    },
    {
      "epoch": 0.23556740944678656,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009946277361120139,
      "loss": 1.295,
      "step": 2738
    },
    {
      "epoch": 0.23565344575410824,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009946237821098322,
      "loss": 1.8241,
      "step": 2739
    },
    {
      "epoch": 0.23573948206142992,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009946198266609723,
      "loss": 0.8934,
      "step": 2740
    },
    {
      "epoch": 0.23582551836875162,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009946158697654456,
      "loss": 1.6804,
      "step": 2741
    },
    {
      "epoch": 0.2359115546760733,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009946119114232635,
      "loss": 1.9245,
      "step": 2742
    },
    {
      "epoch": 0.23599759098339498,
      "grad_norm": 1.375,
      "learning_rate": 0.000994607951634438,
      "loss": 0.7845,
      "step": 2743
    },
    {
      "epoch": 0.2360836272907167,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009946039903989802,
      "loss": 1.636,
      "step": 2744
    },
    {
      "epoch": 0.23616966359803837,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009946000277169021,
      "loss": 2.3261,
      "step": 2745
    },
    {
      "epoch": 0.23625569990536005,
      "grad_norm": 1.4375,
      "learning_rate": 0.000994596063588215,
      "loss": 1.6998,
      "step": 2746
    },
    {
      "epoch": 0.23634173621268176,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009945920980129305,
      "loss": 1.0289,
      "step": 2747
    },
    {
      "epoch": 0.23642777252000344,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009945881309910607,
      "loss": 0.6564,
      "step": 2748
    },
    {
      "epoch": 0.23651380882732514,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009945841625226164,
      "loss": 1.3515,
      "step": 2749
    },
    {
      "epoch": 0.23659984513464682,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009945801926076099,
      "loss": 2.1286,
      "step": 2750
    },
    {
      "epoch": 0.2366858814419685,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009945762212460523,
      "loss": 1.3697,
      "step": 2751
    },
    {
      "epoch": 0.2367719177492902,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009945722484379555,
      "loss": 1.4442,
      "step": 2752
    },
    {
      "epoch": 0.2368579540566119,
      "grad_norm": 2.203125,
      "learning_rate": 0.0009945682741833313,
      "loss": 2.3923,
      "step": 2753
    },
    {
      "epoch": 0.23694399036393357,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009945642984821909,
      "loss": 2.1656,
      "step": 2754
    },
    {
      "epoch": 0.23703002667125528,
      "grad_norm": 1.171875,
      "learning_rate": 0.000994560321334546,
      "loss": 1.1475,
      "step": 2755
    },
    {
      "epoch": 0.23711606297857696,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009945563427404085,
      "loss": 1.0034,
      "step": 2756
    },
    {
      "epoch": 0.23720209928589864,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009945523626997898,
      "loss": 1.1487,
      "step": 2757
    },
    {
      "epoch": 0.23728813559322035,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009945483812127016,
      "loss": 2.4379,
      "step": 2758
    },
    {
      "epoch": 0.23737417190054202,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009945443982791557,
      "loss": 1.4581,
      "step": 2759
    },
    {
      "epoch": 0.23746020820786373,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009945404138991636,
      "loss": 1.2205,
      "step": 2760
    },
    {
      "epoch": 0.2375462445151854,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009945364280727368,
      "loss": 1.2758,
      "step": 2761
    },
    {
      "epoch": 0.2376322808225071,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009945324407998875,
      "loss": 0.906,
      "step": 2762
    },
    {
      "epoch": 0.2377183171298288,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009945284520806266,
      "loss": 1.4906,
      "step": 2763
    },
    {
      "epoch": 0.23780435343715048,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009945244619149664,
      "loss": 1.6288,
      "step": 2764
    },
    {
      "epoch": 0.23789038974447216,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009945204703029182,
      "loss": 1.7523,
      "step": 2765
    },
    {
      "epoch": 0.23797642605179387,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000994516477244494,
      "loss": 1.9603,
      "step": 2766
    },
    {
      "epoch": 0.23806246235911555,
      "grad_norm": 2.078125,
      "learning_rate": 0.000994512482739705,
      "loss": 1.8365,
      "step": 2767
    },
    {
      "epoch": 0.23814849866643722,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009945084867885633,
      "loss": 1.2591,
      "step": 2768
    },
    {
      "epoch": 0.23823453497375893,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009945044893910804,
      "loss": 0.9818,
      "step": 2769
    },
    {
      "epoch": 0.2383205712810806,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009945004905472681,
      "loss": 1.0914,
      "step": 2770
    },
    {
      "epoch": 0.23840660758840232,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009944964902571381,
      "loss": 1.7015,
      "step": 2771
    },
    {
      "epoch": 0.238492643895724,
      "grad_norm": 1.125,
      "learning_rate": 0.0009944924885207018,
      "loss": 1.1558,
      "step": 2772
    },
    {
      "epoch": 0.23857868020304568,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009944884853379714,
      "loss": 1.7363,
      "step": 2773
    },
    {
      "epoch": 0.2386647165103674,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009944844807089582,
      "loss": 1.2984,
      "step": 2774
    },
    {
      "epoch": 0.23875075281768907,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009944804746336742,
      "loss": 1.1638,
      "step": 2775
    },
    {
      "epoch": 0.23883678912501075,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009944764671121309,
      "loss": 0.9563,
      "step": 2776
    },
    {
      "epoch": 0.23892282543233245,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00099447245814434,
      "loss": 1.3838,
      "step": 2777
    },
    {
      "epoch": 0.23900886173965413,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009944684477303136,
      "loss": 1.9458,
      "step": 2778
    },
    {
      "epoch": 0.2390948980469758,
      "grad_norm": 1.28125,
      "learning_rate": 0.000994464435870063,
      "loss": 0.9153,
      "step": 2779
    },
    {
      "epoch": 0.23918093435429752,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009944604225636,
      "loss": 1.444,
      "step": 2780
    },
    {
      "epoch": 0.2392669706616192,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009944564078109366,
      "loss": 1.8289,
      "step": 2781
    },
    {
      "epoch": 0.2393530069689409,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009944523916120842,
      "loss": 1.1245,
      "step": 2782
    },
    {
      "epoch": 0.2394390432762626,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009944483739670548,
      "loss": 1.2281,
      "step": 2783
    },
    {
      "epoch": 0.23952507958358427,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009944443548758602,
      "loss": 1.2678,
      "step": 2784
    },
    {
      "epoch": 0.23961111589090597,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009944403343385118,
      "loss": 1.3265,
      "step": 2785
    },
    {
      "epoch": 0.23969715219822765,
      "grad_norm": 1.375,
      "learning_rate": 0.0009944363123550216,
      "loss": 1.6724,
      "step": 2786
    },
    {
      "epoch": 0.23978318850554933,
      "grad_norm": 2.75,
      "learning_rate": 0.0009944322889254016,
      "loss": 1.4389,
      "step": 2787
    },
    {
      "epoch": 0.23986922481287104,
      "grad_norm": 1.375,
      "learning_rate": 0.000994428264049663,
      "loss": 1.3677,
      "step": 2788
    },
    {
      "epoch": 0.23995526112019272,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009944242377278184,
      "loss": 1.1056,
      "step": 2789
    },
    {
      "epoch": 0.2400412974275144,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009944202099598787,
      "loss": 1.3888,
      "step": 2790
    },
    {
      "epoch": 0.2401273337348361,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000994416180745856,
      "loss": 1.1217,
      "step": 2791
    },
    {
      "epoch": 0.2402133700421578,
      "grad_norm": 2.1875,
      "learning_rate": 0.000994412150085762,
      "loss": 2.0574,
      "step": 2792
    },
    {
      "epoch": 0.2402994063494795,
      "grad_norm": 1.375,
      "learning_rate": 0.000994408117979609,
      "loss": 1.1565,
      "step": 2793
    },
    {
      "epoch": 0.24038544265680117,
      "grad_norm": 1.25,
      "learning_rate": 0.0009944040844274082,
      "loss": 1.1405,
      "step": 2794
    },
    {
      "epoch": 0.24047147896412285,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009944000494291716,
      "loss": 1.4172,
      "step": 2795
    },
    {
      "epoch": 0.24055751527144456,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000994396012984911,
      "loss": 1.2834,
      "step": 2796
    },
    {
      "epoch": 0.24064355157876624,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009943919750946383,
      "loss": 0.9929,
      "step": 2797
    },
    {
      "epoch": 0.24072958788608792,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009943879357583651,
      "loss": 0.7882,
      "step": 2798
    },
    {
      "epoch": 0.24081562419340963,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009943838949761036,
      "loss": 1.3851,
      "step": 2799
    },
    {
      "epoch": 0.2409016605007313,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000994379852747865,
      "loss": 1.1155,
      "step": 2800
    },
    {
      "epoch": 0.240987696808053,
      "grad_norm": 1.125,
      "learning_rate": 0.0009943758090736618,
      "loss": 1.891,
      "step": 2801
    },
    {
      "epoch": 0.2410737331153747,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009943717639535055,
      "loss": 1.6277,
      "step": 2802
    },
    {
      "epoch": 0.24115976942269637,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009943677173874076,
      "loss": 1.1364,
      "step": 2803
    },
    {
      "epoch": 0.24124580573001808,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009943636693753805,
      "loss": 1.6839,
      "step": 2804
    },
    {
      "epoch": 0.24133184203733976,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000994359619917436,
      "loss": 1.125,
      "step": 2805
    },
    {
      "epoch": 0.24141787834466144,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009943555690135855,
      "loss": 0.9751,
      "step": 2806
    },
    {
      "epoch": 0.24150391465198315,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009943515166638414,
      "loss": 1.5118,
      "step": 2807
    },
    {
      "epoch": 0.24158995095930483,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000994347462868215,
      "loss": 0.9068,
      "step": 2808
    },
    {
      "epoch": 0.2416759872666265,
      "grad_norm": 3.40625,
      "learning_rate": 0.0009943434076267184,
      "loss": 1.7792,
      "step": 2809
    },
    {
      "epoch": 0.24176202357394821,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009943393509393634,
      "loss": 0.843,
      "step": 2810
    },
    {
      "epoch": 0.2418480598812699,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009943352928061622,
      "loss": 1.945,
      "step": 2811
    },
    {
      "epoch": 0.24193409618859157,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009943312332271262,
      "loss": 1.1519,
      "step": 2812
    },
    {
      "epoch": 0.24202013249591328,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009943271722022676,
      "loss": 1.5734,
      "step": 2813
    },
    {
      "epoch": 0.24210616880323496,
      "grad_norm": 1.296875,
      "learning_rate": 0.000994323109731598,
      "loss": 0.9162,
      "step": 2814
    },
    {
      "epoch": 0.24219220511055667,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009943190458151297,
      "loss": 1.4673,
      "step": 2815
    },
    {
      "epoch": 0.24227824141787835,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000994314980452874,
      "loss": 1.3451,
      "step": 2816
    },
    {
      "epoch": 0.24236427772520003,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009943109136448436,
      "loss": 1.9529,
      "step": 2817
    },
    {
      "epoch": 0.24245031403252174,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009943068453910496,
      "loss": 1.8894,
      "step": 2818
    },
    {
      "epoch": 0.24253635033984342,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009943027756915044,
      "loss": 1.9073,
      "step": 2819
    },
    {
      "epoch": 0.2426223866471651,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009942987045462196,
      "loss": 1.8521,
      "step": 2820
    },
    {
      "epoch": 0.2427084229544868,
      "grad_norm": 1.640625,
      "learning_rate": 0.000994294631955207,
      "loss": 1.2906,
      "step": 2821
    },
    {
      "epoch": 0.24279445926180848,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000994290557918479,
      "loss": 1.0297,
      "step": 2822
    },
    {
      "epoch": 0.24288049556913016,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000994286482436047,
      "loss": 1.4837,
      "step": 2823
    },
    {
      "epoch": 0.24296653187645187,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009942824055079237,
      "loss": 1.6882,
      "step": 2824
    },
    {
      "epoch": 0.24305256818377355,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00099427832713412,
      "loss": 2.1232,
      "step": 2825
    },
    {
      "epoch": 0.24313860449109526,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009942742473146486,
      "loss": 1.6501,
      "step": 2826
    },
    {
      "epoch": 0.24322464079841694,
      "grad_norm": 1.84375,
      "learning_rate": 0.000994270166049521,
      "loss": 2.0767,
      "step": 2827
    },
    {
      "epoch": 0.24331067710573862,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009942660833387493,
      "loss": 1.3977,
      "step": 2828
    },
    {
      "epoch": 0.24339671341306032,
      "grad_norm": 1.375,
      "learning_rate": 0.0009942619991823456,
      "loss": 1.4152,
      "step": 2829
    },
    {
      "epoch": 0.243482749720382,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009942579135803215,
      "loss": 1.8548,
      "step": 2830
    },
    {
      "epoch": 0.24356878602770368,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009942538265326891,
      "loss": 1.3902,
      "step": 2831
    },
    {
      "epoch": 0.2436548223350254,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009942497380394605,
      "loss": 1.9701,
      "step": 2832
    },
    {
      "epoch": 0.24374085864234707,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009942456481006475,
      "loss": 0.9289,
      "step": 2833
    },
    {
      "epoch": 0.24382689494966875,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009942415567162622,
      "loss": 1.9915,
      "step": 2834
    },
    {
      "epoch": 0.24391293125699046,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009942374638863163,
      "loss": 1.0423,
      "step": 2835
    },
    {
      "epoch": 0.24399896756431214,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009942333696108222,
      "loss": 1.5809,
      "step": 2836
    },
    {
      "epoch": 0.24408500387163384,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009942292738897914,
      "loss": 1.721,
      "step": 2837
    },
    {
      "epoch": 0.24417104017895552,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009942251767232361,
      "loss": 1.1534,
      "step": 2838
    },
    {
      "epoch": 0.2442570764862772,
      "grad_norm": 1.25,
      "learning_rate": 0.0009942210781111681,
      "loss": 1.1322,
      "step": 2839
    },
    {
      "epoch": 0.2443431127935989,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009942169780536,
      "loss": 1.3382,
      "step": 2840
    },
    {
      "epoch": 0.2444291491009206,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000994212876550543,
      "loss": 1.0036,
      "step": 2841
    },
    {
      "epoch": 0.24451518540824227,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009942087736020098,
      "loss": 1.2221,
      "step": 2842
    },
    {
      "epoch": 0.24460122171556398,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009942046692080118,
      "loss": 1.4027,
      "step": 2843
    },
    {
      "epoch": 0.24468725802288566,
      "grad_norm": 1.75,
      "learning_rate": 0.0009942005633685614,
      "loss": 1.5165,
      "step": 2844
    },
    {
      "epoch": 0.24477329433020734,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009941964560836703,
      "loss": 1.0675,
      "step": 2845
    },
    {
      "epoch": 0.24485933063752904,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009941923473533508,
      "loss": 1.3198,
      "step": 2846
    },
    {
      "epoch": 0.24494536694485072,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009941882371776148,
      "loss": 1.7928,
      "step": 2847
    },
    {
      "epoch": 0.24503140325217243,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009941841255564742,
      "loss": 1.8229,
      "step": 2848
    },
    {
      "epoch": 0.2451174395594941,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009941800124899415,
      "loss": 1.7645,
      "step": 2849
    },
    {
      "epoch": 0.2452034758668158,
      "grad_norm": 1.3125,
      "learning_rate": 0.000994175897978028,
      "loss": 1.0468,
      "step": 2850
    },
    {
      "epoch": 0.2452895121741375,
      "grad_norm": 1.0,
      "learning_rate": 0.0009941717820207462,
      "loss": 0.7355,
      "step": 2851
    },
    {
      "epoch": 0.24537554848145918,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000994167664618108,
      "loss": 1.2113,
      "step": 2852
    },
    {
      "epoch": 0.24546158478878086,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009941635457701257,
      "loss": 1.4952,
      "step": 2853
    },
    {
      "epoch": 0.24554762109610256,
      "grad_norm": 1.46875,
      "learning_rate": 0.000994159425476811,
      "loss": 1.5305,
      "step": 2854
    },
    {
      "epoch": 0.24563365740342424,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009941553037381761,
      "loss": 1.1539,
      "step": 2855
    },
    {
      "epoch": 0.24571969371074592,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000994151180554233,
      "loss": 1.2318,
      "step": 2856
    },
    {
      "epoch": 0.24580573001806763,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009941470559249937,
      "loss": 1.4717,
      "step": 2857
    },
    {
      "epoch": 0.2458917663253893,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009941429298504707,
      "loss": 1.0806,
      "step": 2858
    },
    {
      "epoch": 0.24597780263271102,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009941388023306755,
      "loss": 1.6628,
      "step": 2859
    },
    {
      "epoch": 0.2460638389400327,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009941346733656205,
      "loss": 1.0228,
      "step": 2860
    },
    {
      "epoch": 0.24614987524735438,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009941305429553176,
      "loss": 1.6818,
      "step": 2861
    },
    {
      "epoch": 0.24623591155467608,
      "grad_norm": 1.203125,
      "learning_rate": 0.000994126411099779,
      "loss": 1.3547,
      "step": 2862
    },
    {
      "epoch": 0.24632194786199776,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009941222777990167,
      "loss": 1.6014,
      "step": 2863
    },
    {
      "epoch": 0.24640798416931944,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009941181430530427,
      "loss": 1.3903,
      "step": 2864
    },
    {
      "epoch": 0.24649402047664115,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009941140068618695,
      "loss": 0.9767,
      "step": 2865
    },
    {
      "epoch": 0.24658005678396283,
      "grad_norm": 1.0625,
      "learning_rate": 0.000994109869225509,
      "loss": 0.8899,
      "step": 2866
    },
    {
      "epoch": 0.2466660930912845,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009941057301439732,
      "loss": 2.078,
      "step": 2867
    },
    {
      "epoch": 0.24675212939860622,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009941015896172741,
      "loss": 1.0656,
      "step": 2868
    },
    {
      "epoch": 0.2468381657059279,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000994097447645424,
      "loss": 0.9808,
      "step": 2869
    },
    {
      "epoch": 0.2469242020132496,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000994093304228435,
      "loss": 1.5802,
      "step": 2870
    },
    {
      "epoch": 0.24701023832057128,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009940891593663188,
      "loss": 0.6195,
      "step": 2871
    },
    {
      "epoch": 0.24709627462789296,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009940850130590883,
      "loss": 1.3639,
      "step": 2872
    },
    {
      "epoch": 0.24718231093521467,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009940808653067551,
      "loss": 1.895,
      "step": 2873
    },
    {
      "epoch": 0.24726834724253635,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009940767161093315,
      "loss": 1.1564,
      "step": 2874
    },
    {
      "epoch": 0.24735438354985803,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009940725654668298,
      "loss": 1.1174,
      "step": 2875
    },
    {
      "epoch": 0.24744041985717974,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009940684133792617,
      "loss": 0.7676,
      "step": 2876
    },
    {
      "epoch": 0.24752645616450142,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009940642598466397,
      "loss": 1.3108,
      "step": 2877
    },
    {
      "epoch": 0.2476124924718231,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009940601048689757,
      "loss": 0.8794,
      "step": 2878
    },
    {
      "epoch": 0.2476985287791448,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009940559484462818,
      "loss": 1.088,
      "step": 2879
    },
    {
      "epoch": 0.24778456508646649,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009940517905785706,
      "loss": 1.2182,
      "step": 2880
    },
    {
      "epoch": 0.2478706013937882,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009940476312658541,
      "loss": 1.2449,
      "step": 2881
    },
    {
      "epoch": 0.24795663770110987,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009940434705081442,
      "loss": 0.9609,
      "step": 2882
    },
    {
      "epoch": 0.24804267400843155,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009940393083054534,
      "loss": 1.0137,
      "step": 2883
    },
    {
      "epoch": 0.24812871031575326,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009940351446577934,
      "loss": 1.0626,
      "step": 2884
    },
    {
      "epoch": 0.24821474662307494,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000994030979565177,
      "loss": 1.2546,
      "step": 2885
    },
    {
      "epoch": 0.24830078293039662,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000994026813027616,
      "loss": 1.133,
      "step": 2886
    },
    {
      "epoch": 0.24838681923771833,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009940226450451226,
      "loss": 1.8339,
      "step": 2887
    },
    {
      "epoch": 0.24847285554504,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000994018475617709,
      "loss": 2.0445,
      "step": 2888
    },
    {
      "epoch": 0.24855889185236169,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009940143047453874,
      "loss": 1.1864,
      "step": 2889
    },
    {
      "epoch": 0.2486449281596834,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009940101324281702,
      "loss": 1.9774,
      "step": 2890
    },
    {
      "epoch": 0.24873096446700507,
      "grad_norm": 1.375,
      "learning_rate": 0.0009940059586660693,
      "loss": 1.5303,
      "step": 2891
    },
    {
      "epoch": 0.24881700077432678,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009940017834590973,
      "loss": 1.2646,
      "step": 2892
    },
    {
      "epoch": 0.24890303708164846,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009939976068072658,
      "loss": 2.3048,
      "step": 2893
    },
    {
      "epoch": 0.24898907338897014,
      "grad_norm": 1.125,
      "learning_rate": 0.0009939934287105875,
      "loss": 1.7389,
      "step": 2894
    },
    {
      "epoch": 0.24907510969629185,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009939892491690747,
      "loss": 1.0478,
      "step": 2895
    },
    {
      "epoch": 0.24916114600361353,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009939850681827391,
      "loss": 1.3092,
      "step": 2896
    },
    {
      "epoch": 0.2492471823109352,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009939808857515936,
      "loss": 1.9876,
      "step": 2897
    },
    {
      "epoch": 0.2493332186182569,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009939767018756498,
      "loss": 1.6297,
      "step": 2898
    },
    {
      "epoch": 0.2494192549255786,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009939725165549205,
      "loss": 1.1815,
      "step": 2899
    },
    {
      "epoch": 0.24950529123290027,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009939683297894174,
      "loss": 1.0673,
      "step": 2900
    },
    {
      "epoch": 0.24959132754022198,
      "grad_norm": 1.6875,
      "learning_rate": 0.000993964141579153,
      "loss": 2.0988,
      "step": 2901
    },
    {
      "epoch": 0.24967736384754366,
      "grad_norm": 1.625,
      "learning_rate": 0.0009939599519241398,
      "loss": 1.5198,
      "step": 2902
    },
    {
      "epoch": 0.24976340015486537,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009939557608243897,
      "loss": 1.7661,
      "step": 2903
    },
    {
      "epoch": 0.24984943646218705,
      "grad_norm": 1.234375,
      "learning_rate": 0.000993951568279915,
      "loss": 0.9542,
      "step": 2904
    },
    {
      "epoch": 0.24993547276950873,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000993947374290728,
      "loss": 1.8225,
      "step": 2905
    },
    {
      "epoch": 0.2500215090768304,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000993943178856841,
      "loss": 1.7382,
      "step": 2906
    },
    {
      "epoch": 0.2500215090768304,
      "eval_loss": 1.3956936597824097,
      "eval_runtime": 358.2611,
      "eval_samples_per_second": 12.977,
      "eval_steps_per_second": 1.625,
      "step": 2906
    },
    {
      "epoch": 0.2501075453841521,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009939389819782665,
      "loss": 1.7892,
      "step": 2907
    },
    {
      "epoch": 0.2501935816914738,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009939347836550162,
      "loss": 0.7357,
      "step": 2908
    },
    {
      "epoch": 0.2502796179987955,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009939305838871028,
      "loss": 1.4111,
      "step": 2909
    },
    {
      "epoch": 0.2503656543061172,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009939263826745388,
      "loss": 1.4264,
      "step": 2910
    },
    {
      "epoch": 0.25045169061343886,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009939221800173358,
      "loss": 1.134,
      "step": 2911
    },
    {
      "epoch": 0.25053772692076054,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009939179759155068,
      "loss": 1.1982,
      "step": 2912
    },
    {
      "epoch": 0.2506237632280823,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009939137703690638,
      "loss": 1.8246,
      "step": 2913
    },
    {
      "epoch": 0.25070979953540395,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009939095633780188,
      "loss": 1.296,
      "step": 2914
    },
    {
      "epoch": 0.25079583584272563,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009939053549423845,
      "loss": 1.7308,
      "step": 2915
    },
    {
      "epoch": 0.2508818721500473,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009939011450621732,
      "loss": 1.9569,
      "step": 2916
    },
    {
      "epoch": 0.250967908457369,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000993896933737397,
      "loss": 0.8935,
      "step": 2917
    },
    {
      "epoch": 0.2510539447646907,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009938927209680684,
      "loss": 1.5622,
      "step": 2918
    },
    {
      "epoch": 0.2511399810720124,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009938885067541996,
      "loss": 1.1812,
      "step": 2919
    },
    {
      "epoch": 0.2512260173793341,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000993884291095803,
      "loss": 1.2997,
      "step": 2920
    },
    {
      "epoch": 0.25131205368665577,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009938800739928907,
      "loss": 1.6956,
      "step": 2921
    },
    {
      "epoch": 0.25139808999397745,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009938758554454757,
      "loss": 1.8315,
      "step": 2922
    },
    {
      "epoch": 0.2514841263012991,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009938716354535694,
      "loss": 1.5378,
      "step": 2923
    },
    {
      "epoch": 0.25157016260862086,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000993867414017185,
      "loss": 2.0299,
      "step": 2924
    },
    {
      "epoch": 0.25165619891594254,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009938631911363342,
      "loss": 1.905,
      "step": 2925
    },
    {
      "epoch": 0.2517422352232642,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009938589668110297,
      "loss": 1.28,
      "step": 2926
    },
    {
      "epoch": 0.2518282715305859,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009938547410412839,
      "loss": 1.0551,
      "step": 2927
    },
    {
      "epoch": 0.2519143078379076,
      "grad_norm": 1.125,
      "learning_rate": 0.0009938505138271088,
      "loss": 1.8094,
      "step": 2928
    },
    {
      "epoch": 0.25200034414522926,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000993846285168517,
      "loss": 1.6699,
      "step": 2929
    },
    {
      "epoch": 0.252086380452551,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009938420550655211,
      "loss": 1.6555,
      "step": 2930
    },
    {
      "epoch": 0.2521724167598727,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000993837823518133,
      "loss": 1.0309,
      "step": 2931
    },
    {
      "epoch": 0.25225845306719435,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009938335905263656,
      "loss": 2.0573,
      "step": 2932
    },
    {
      "epoch": 0.25234448937451603,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009938293560902308,
      "loss": 1.4488,
      "step": 2933
    },
    {
      "epoch": 0.2524305256818377,
      "grad_norm": 1.359375,
      "learning_rate": 0.000993825120209741,
      "loss": 1.1151,
      "step": 2934
    },
    {
      "epoch": 0.25251656198915945,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009938208828849091,
      "loss": 1.3429,
      "step": 2935
    },
    {
      "epoch": 0.25260259829648113,
      "grad_norm": 1.328125,
      "learning_rate": 0.000993816644115747,
      "loss": 1.1297,
      "step": 2936
    },
    {
      "epoch": 0.2526886346038028,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009938124039022673,
      "loss": 1.433,
      "step": 2937
    },
    {
      "epoch": 0.2527746709111245,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009938081622444823,
      "loss": 1.5414,
      "step": 2938
    },
    {
      "epoch": 0.25286070721844617,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009938039191424045,
      "loss": 1.2297,
      "step": 2939
    },
    {
      "epoch": 0.25294674352576785,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009937996745960462,
      "loss": 1.1607,
      "step": 2940
    },
    {
      "epoch": 0.2530327798330896,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00099379542860542,
      "loss": 1.8101,
      "step": 2941
    },
    {
      "epoch": 0.25311881614041126,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009937911811705382,
      "loss": 1.4785,
      "step": 2942
    },
    {
      "epoch": 0.25320485244773294,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009937869322914133,
      "loss": 2.2223,
      "step": 2943
    },
    {
      "epoch": 0.2532908887550546,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009937826819680575,
      "loss": 1.3075,
      "step": 2944
    },
    {
      "epoch": 0.2533769250623763,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009937784302004836,
      "loss": 1.3547,
      "step": 2945
    },
    {
      "epoch": 0.25346296136969804,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009937741769887036,
      "loss": 1.7777,
      "step": 2946
    },
    {
      "epoch": 0.2535489976770197,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009937699223327302,
      "loss": 1.2451,
      "step": 2947
    },
    {
      "epoch": 0.2536350339843414,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009937656662325758,
      "loss": 1.4429,
      "step": 2948
    },
    {
      "epoch": 0.2537210702916631,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009937614086882529,
      "loss": 1.9231,
      "step": 2949
    },
    {
      "epoch": 0.25380710659898476,
      "grad_norm": 1.1875,
      "learning_rate": 0.000993757149699774,
      "loss": 0.8707,
      "step": 2950
    },
    {
      "epoch": 0.25389314290630643,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009937528892671515,
      "loss": 1.8082,
      "step": 2951
    },
    {
      "epoch": 0.25397917921362817,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009937486273903977,
      "loss": 1.0777,
      "step": 2952
    },
    {
      "epoch": 0.25406521552094985,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009937443640695251,
      "loss": 1.1873,
      "step": 2953
    },
    {
      "epoch": 0.25415125182827153,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009937400993045465,
      "loss": 0.745,
      "step": 2954
    },
    {
      "epoch": 0.2542372881355932,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000993735833095474,
      "loss": 1.2915,
      "step": 2955
    },
    {
      "epoch": 0.2543233244429149,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009937315654423202,
      "loss": 2.2284,
      "step": 2956
    },
    {
      "epoch": 0.2544093607502366,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009937272963450976,
      "loss": 0.5431,
      "step": 2957
    },
    {
      "epoch": 0.2544953970575583,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000993723025803819,
      "loss": 1.6045,
      "step": 2958
    },
    {
      "epoch": 0.25458143336488,
      "grad_norm": 1.453125,
      "learning_rate": 0.000993718753818496,
      "loss": 1.8974,
      "step": 2959
    },
    {
      "epoch": 0.25466746967220166,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000993714480389142,
      "loss": 1.4276,
      "step": 2960
    },
    {
      "epoch": 0.25475350597952334,
      "grad_norm": 2.03125,
      "learning_rate": 0.000993710205515769,
      "loss": 1.1473,
      "step": 2961
    },
    {
      "epoch": 0.254839542286845,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009937059291983898,
      "loss": 0.8539,
      "step": 2962
    },
    {
      "epoch": 0.25492557859416676,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009937016514370168,
      "loss": 1.7788,
      "step": 2963
    },
    {
      "epoch": 0.25501161490148844,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009936973722316624,
      "loss": 0.9924,
      "step": 2964
    },
    {
      "epoch": 0.2550976512088101,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009936930915823395,
      "loss": 1.2054,
      "step": 2965
    },
    {
      "epoch": 0.2551836875161318,
      "grad_norm": 2.09375,
      "learning_rate": 0.00099368880948906,
      "loss": 1.775,
      "step": 2966
    },
    {
      "epoch": 0.2552697238234535,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009936845259518369,
      "loss": 1.9898,
      "step": 2967
    },
    {
      "epoch": 0.2553557601307752,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009936802409706825,
      "loss": 1.3226,
      "step": 2968
    },
    {
      "epoch": 0.2554417964380969,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009936759545456093,
      "loss": 1.6615,
      "step": 2969
    },
    {
      "epoch": 0.25552783274541857,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009936716666766301,
      "loss": 1.79,
      "step": 2970
    },
    {
      "epoch": 0.25561386905274025,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009936673773637572,
      "loss": 0.8858,
      "step": 2971
    },
    {
      "epoch": 0.25569990536006193,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009936630866070034,
      "loss": 0.8167,
      "step": 2972
    },
    {
      "epoch": 0.2557859416673836,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009936587944063809,
      "loss": 2.1131,
      "step": 2973
    },
    {
      "epoch": 0.25587197797470534,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009936545007619025,
      "loss": 1.1311,
      "step": 2974
    },
    {
      "epoch": 0.255958014282027,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009936502056735808,
      "loss": 1.0466,
      "step": 2975
    },
    {
      "epoch": 0.2560440505893487,
      "grad_norm": 3.109375,
      "learning_rate": 0.0009936459091414282,
      "loss": 2.0775,
      "step": 2976
    },
    {
      "epoch": 0.2561300868966704,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009936416111654572,
      "loss": 1.6268,
      "step": 2977
    },
    {
      "epoch": 0.25621612320399206,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009936373117456808,
      "loss": 2.0512,
      "step": 2978
    },
    {
      "epoch": 0.2563021595113138,
      "grad_norm": 1.640625,
      "learning_rate": 0.000993633010882111,
      "loss": 1.9101,
      "step": 2979
    },
    {
      "epoch": 0.2563881958186355,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009936287085747606,
      "loss": 2.2032,
      "step": 2980
    },
    {
      "epoch": 0.25647423212595716,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009936244048236423,
      "loss": 0.9313,
      "step": 2981
    },
    {
      "epoch": 0.25656026843327884,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009936200996287687,
      "loss": 1.0721,
      "step": 2982
    },
    {
      "epoch": 0.2566463047406005,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009936157929901523,
      "loss": 1.8815,
      "step": 2983
    },
    {
      "epoch": 0.2567323410479222,
      "grad_norm": 1.875,
      "learning_rate": 0.0009936114849078057,
      "loss": 1.5091,
      "step": 2984
    },
    {
      "epoch": 0.25681837735524393,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009936071753817415,
      "loss": 0.8972,
      "step": 2985
    },
    {
      "epoch": 0.2569044136625656,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009936028644119722,
      "loss": 1.6232,
      "step": 2986
    },
    {
      "epoch": 0.2569904499698873,
      "grad_norm": 1.375,
      "learning_rate": 0.0009935985519985108,
      "loss": 1.66,
      "step": 2987
    },
    {
      "epoch": 0.25707648627720897,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009935942381413695,
      "loss": 1.9225,
      "step": 2988
    },
    {
      "epoch": 0.25716252258453065,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000993589922840561,
      "loss": 1.6376,
      "step": 2989
    },
    {
      "epoch": 0.2572485588918524,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000993585606096098,
      "loss": 1.6788,
      "step": 2990
    },
    {
      "epoch": 0.25733459519917407,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009935812879079931,
      "loss": 1.7896,
      "step": 2991
    },
    {
      "epoch": 0.25742063150649575,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009935769682762588,
      "loss": 0.6921,
      "step": 2992
    },
    {
      "epoch": 0.2575066678138174,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000993572647200908,
      "loss": 1.4039,
      "step": 2993
    },
    {
      "epoch": 0.2575927041211391,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009935683246819532,
      "loss": 1.019,
      "step": 2994
    },
    {
      "epoch": 0.2576787404284608,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000993564000719407,
      "loss": 1.0109,
      "step": 2995
    },
    {
      "epoch": 0.2577647767357825,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009935596753132823,
      "loss": 1.3191,
      "step": 2996
    },
    {
      "epoch": 0.2578508130431042,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009935553484635912,
      "loss": 1.1828,
      "step": 2997
    },
    {
      "epoch": 0.2579368493504259,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009935510201703468,
      "loss": 1.8843,
      "step": 2998
    },
    {
      "epoch": 0.25802288565774756,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009935466904335618,
      "loss": 1.6621,
      "step": 2999
    },
    {
      "epoch": 0.25810892196506924,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009935423592532487,
      "loss": 1.323,
      "step": 3000
    },
    {
      "epoch": 0.258194958272391,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00099353802662942,
      "loss": 1.7587,
      "step": 3001
    },
    {
      "epoch": 0.25828099457971265,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009935336925620886,
      "loss": 0.8585,
      "step": 3002
    },
    {
      "epoch": 0.25836703088703433,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009935293570512674,
      "loss": 0.9414,
      "step": 3003
    },
    {
      "epoch": 0.258453067194356,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009935250200969685,
      "loss": 0.8985,
      "step": 3004
    },
    {
      "epoch": 0.2585391035016777,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000993520681699205,
      "loss": 1.4005,
      "step": 3005
    },
    {
      "epoch": 0.25862513980899937,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009935163418579895,
      "loss": 1.8444,
      "step": 3006
    },
    {
      "epoch": 0.2587111761163211,
      "grad_norm": 1.234375,
      "learning_rate": 0.000993512000573335,
      "loss": 1.7297,
      "step": 3007
    },
    {
      "epoch": 0.2587972124236428,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009935076578452533,
      "loss": 1.2113,
      "step": 3008
    },
    {
      "epoch": 0.25888324873096447,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009935033136737582,
      "loss": 1.5641,
      "step": 3009
    },
    {
      "epoch": 0.25896928503828615,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009934989680588616,
      "loss": 1.9477,
      "step": 3010
    },
    {
      "epoch": 0.2590553213456078,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009934946210005766,
      "loss": 1.909,
      "step": 3011
    },
    {
      "epoch": 0.25914135765292956,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009934902724989157,
      "loss": 1.3622,
      "step": 3012
    },
    {
      "epoch": 0.25922739396025124,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000993485922553892,
      "loss": 1.6571,
      "step": 3013
    },
    {
      "epoch": 0.2593134302675729,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009934815711655177,
      "loss": 1.5788,
      "step": 3014
    },
    {
      "epoch": 0.2593994665748946,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009934772183338057,
      "loss": 1.2401,
      "step": 3015
    },
    {
      "epoch": 0.2594855028822163,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000993472864058769,
      "loss": 1.1798,
      "step": 3016
    },
    {
      "epoch": 0.25957153918953796,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00099346850834042,
      "loss": 1.4307,
      "step": 3017
    },
    {
      "epoch": 0.2596575754968597,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009934641511787717,
      "loss": 1.2168,
      "step": 3018
    },
    {
      "epoch": 0.2597436118041814,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009934597925738368,
      "loss": 1.8428,
      "step": 3019
    },
    {
      "epoch": 0.25982964811150305,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009934554325256278,
      "loss": 2.0624,
      "step": 3020
    },
    {
      "epoch": 0.25991568441882473,
      "grad_norm": 2.125,
      "learning_rate": 0.0009934510710341577,
      "loss": 1.9669,
      "step": 3021
    },
    {
      "epoch": 0.2600017207261464,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009934467080994394,
      "loss": 1.6068,
      "step": 3022
    },
    {
      "epoch": 0.26008775703346815,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009934423437214853,
      "loss": 1.2537,
      "step": 3023
    },
    {
      "epoch": 0.2601737933407898,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009934379779003083,
      "loss": 1.0186,
      "step": 3024
    },
    {
      "epoch": 0.2602598296481115,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009934336106359211,
      "loss": 1.1217,
      "step": 3025
    },
    {
      "epoch": 0.2603458659554332,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009934292419283366,
      "loss": 1.0173,
      "step": 3026
    },
    {
      "epoch": 0.26043190226275487,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009934248717775676,
      "loss": 1.9299,
      "step": 3027
    },
    {
      "epoch": 0.26051793857007655,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009934205001836268,
      "loss": 1.1975,
      "step": 3028
    },
    {
      "epoch": 0.2606039748773983,
      "grad_norm": 1.453125,
      "learning_rate": 0.000993416127146527,
      "loss": 1.5484,
      "step": 3029
    },
    {
      "epoch": 0.26069001118471996,
      "grad_norm": 1.40625,
      "learning_rate": 0.000993411752666281,
      "loss": 1.2444,
      "step": 3030
    },
    {
      "epoch": 0.26077604749204164,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009934073767429014,
      "loss": 0.9241,
      "step": 3031
    },
    {
      "epoch": 0.2608620837993633,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009934029993764013,
      "loss": 1.7467,
      "step": 3032
    },
    {
      "epoch": 0.260948120106685,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009933986205667936,
      "loss": 1.0805,
      "step": 3033
    },
    {
      "epoch": 0.26103415641400673,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009933942403140905,
      "loss": 0.8353,
      "step": 3034
    },
    {
      "epoch": 0.2611201927213284,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009933898586183055,
      "loss": 1.8512,
      "step": 3035
    },
    {
      "epoch": 0.2612062290286501,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009933854754794512,
      "loss": 1.2399,
      "step": 3036
    },
    {
      "epoch": 0.2612922653359718,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00099338109089754,
      "loss": 1.9766,
      "step": 3037
    },
    {
      "epoch": 0.26137830164329345,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009933767048725852,
      "loss": 1.5716,
      "step": 3038
    },
    {
      "epoch": 0.26146433795061513,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009933723174045997,
      "loss": 2.4429,
      "step": 3039
    },
    {
      "epoch": 0.26155037425793687,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000993367928493596,
      "loss": 1.36,
      "step": 3040
    },
    {
      "epoch": 0.26163641056525855,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000993363538139587,
      "loss": 1.1044,
      "step": 3041
    },
    {
      "epoch": 0.26172244687258023,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009933591463425857,
      "loss": 1.4162,
      "step": 3042
    },
    {
      "epoch": 0.2618084831799019,
      "grad_norm": 1.453125,
      "learning_rate": 0.000993354753102605,
      "loss": 1.5942,
      "step": 3043
    },
    {
      "epoch": 0.2618945194872236,
      "grad_norm": 1.21875,
      "learning_rate": 0.000993350358419657,
      "loss": 0.9046,
      "step": 3044
    },
    {
      "epoch": 0.2619805557945453,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009933459622937558,
      "loss": 1.128,
      "step": 3045
    },
    {
      "epoch": 0.262066592101867,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009933415647249133,
      "loss": 1.648,
      "step": 3046
    },
    {
      "epoch": 0.2621526284091887,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009933371657131426,
      "loss": 0.8786,
      "step": 3047
    },
    {
      "epoch": 0.26223866471651036,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009933327652584568,
      "loss": 1.4348,
      "step": 3048
    },
    {
      "epoch": 0.26232470102383204,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009933283633608687,
      "loss": 1.3528,
      "step": 3049
    },
    {
      "epoch": 0.2624107373311537,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000993323960020391,
      "loss": 0.8527,
      "step": 3050
    },
    {
      "epoch": 0.26249677363847546,
      "grad_norm": 1.25,
      "learning_rate": 0.0009933195552370366,
      "loss": 1.7372,
      "step": 3051
    },
    {
      "epoch": 0.26258280994579714,
      "grad_norm": 1.375,
      "learning_rate": 0.0009933151490108185,
      "loss": 0.8219,
      "step": 3052
    },
    {
      "epoch": 0.2626688462531188,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009933107413417496,
      "loss": 1.2421,
      "step": 3053
    },
    {
      "epoch": 0.2627548825604405,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009933063322298426,
      "loss": 0.9912,
      "step": 3054
    },
    {
      "epoch": 0.2628409188677622,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009933019216751104,
      "loss": 1.2665,
      "step": 3055
    },
    {
      "epoch": 0.2629269551750839,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009932975096775662,
      "loss": 2.0644,
      "step": 3056
    },
    {
      "epoch": 0.2630129914824056,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009932930962372227,
      "loss": 1.3321,
      "step": 3057
    },
    {
      "epoch": 0.26309902778972727,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009932886813540929,
      "loss": 1.7175,
      "step": 3058
    },
    {
      "epoch": 0.26318506409704895,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009932842650281896,
      "loss": 2.0023,
      "step": 3059
    },
    {
      "epoch": 0.26327110040437063,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009932798472595257,
      "loss": 0.73,
      "step": 3060
    },
    {
      "epoch": 0.2633571367116923,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009932754280481142,
      "loss": 1.5656,
      "step": 3061
    },
    {
      "epoch": 0.26344317301901404,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009932710073939682,
      "loss": 1.9273,
      "step": 3062
    },
    {
      "epoch": 0.2635292093263357,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009932665852971,
      "loss": 1.3057,
      "step": 3063
    },
    {
      "epoch": 0.2636152456336574,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009932621617575236,
      "loss": 1.4751,
      "step": 3064
    },
    {
      "epoch": 0.2637012819409791,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009932577367752508,
      "loss": 1.1851,
      "step": 3065
    },
    {
      "epoch": 0.26378731824830076,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009932533103502953,
      "loss": 1.6078,
      "step": 3066
    },
    {
      "epoch": 0.2638733545556225,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009932488824826696,
      "loss": 1.211,
      "step": 3067
    },
    {
      "epoch": 0.2639593908629442,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009932444531723869,
      "loss": 2.0694,
      "step": 3068
    },
    {
      "epoch": 0.26404542717026586,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009932400224194602,
      "loss": 1.4353,
      "step": 3069
    },
    {
      "epoch": 0.26413146347758754,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009932355902239024,
      "loss": 1.528,
      "step": 3070
    },
    {
      "epoch": 0.2642174997849092,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009932311565857264,
      "loss": 1.8226,
      "step": 3071
    },
    {
      "epoch": 0.2643035360922309,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000993226721504945,
      "loss": 0.9171,
      "step": 3072
    },
    {
      "epoch": 0.26438957239955263,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009932222849815717,
      "loss": 0.9097,
      "step": 3073
    },
    {
      "epoch": 0.2644756087068743,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009932178470156191,
      "loss": 1.8911,
      "step": 3074
    },
    {
      "epoch": 0.264561645014196,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009932134076071,
      "loss": 1.9565,
      "step": 3075
    },
    {
      "epoch": 0.26464768132151767,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009932089667560276,
      "loss": 1.085,
      "step": 3076
    },
    {
      "epoch": 0.26473371762883935,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000993204524462415,
      "loss": 0.7962,
      "step": 3077
    },
    {
      "epoch": 0.2648197539361611,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009932000807262751,
      "loss": 1.1026,
      "step": 3078
    },
    {
      "epoch": 0.26490579024348276,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000993195635547621,
      "loss": 1.995,
      "step": 3079
    },
    {
      "epoch": 0.26499182655080444,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009931911889264652,
      "loss": 1.2203,
      "step": 3080
    },
    {
      "epoch": 0.2650778628581261,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009931867408628215,
      "loss": 1.6688,
      "step": 3081
    },
    {
      "epoch": 0.2651638991654478,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009931822913567024,
      "loss": 1.8004,
      "step": 3082
    },
    {
      "epoch": 0.2652499354727695,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000993177840408121,
      "loss": 1.0246,
      "step": 3083
    },
    {
      "epoch": 0.2653359717800912,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009931733880170904,
      "loss": 0.7907,
      "step": 3084
    },
    {
      "epoch": 0.2654220080874129,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009931689341836233,
      "loss": 1.0561,
      "step": 3085
    },
    {
      "epoch": 0.2655080443947346,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009931644789077333,
      "loss": 1.6659,
      "step": 3086
    },
    {
      "epoch": 0.26559408070205626,
      "grad_norm": 1.53125,
      "learning_rate": 0.000993160022189433,
      "loss": 1.3168,
      "step": 3087
    },
    {
      "epoch": 0.26568011700937794,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009931555640287353,
      "loss": 1.6293,
      "step": 3088
    },
    {
      "epoch": 0.26576615331669967,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009931511044256537,
      "loss": 1.2523,
      "step": 3089
    },
    {
      "epoch": 0.26585218962402135,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000993146643380201,
      "loss": 1.1518,
      "step": 3090
    },
    {
      "epoch": 0.26593822593134303,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009931421808923903,
      "loss": 1.6022,
      "step": 3091
    },
    {
      "epoch": 0.2660242622386647,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009931377169622346,
      "loss": 1.6187,
      "step": 3092
    },
    {
      "epoch": 0.2661102985459864,
      "grad_norm": 2.109375,
      "learning_rate": 0.000993133251589747,
      "loss": 1.823,
      "step": 3093
    },
    {
      "epoch": 0.26619633485330807,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009931287847749408,
      "loss": 1.6575,
      "step": 3094
    },
    {
      "epoch": 0.2662823711606298,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009931243165178286,
      "loss": 1.7535,
      "step": 3095
    },
    {
      "epoch": 0.2663684074679515,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009931198468184235,
      "loss": 0.9252,
      "step": 3096
    },
    {
      "epoch": 0.26645444377527316,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000993115375676739,
      "loss": 1.3069,
      "step": 3097
    },
    {
      "epoch": 0.26654048008259484,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009931109030927878,
      "loss": 1.168,
      "step": 3098
    },
    {
      "epoch": 0.2666265163899165,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009931064290665832,
      "loss": 1.246,
      "step": 3099
    },
    {
      "epoch": 0.26671255269723826,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009931019535981383,
      "loss": 2.0963,
      "step": 3100
    },
    {
      "epoch": 0.26679858900455994,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009930974766874658,
      "loss": 0.9944,
      "step": 3101
    },
    {
      "epoch": 0.2668846253118816,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009930929983345794,
      "loss": 1.1853,
      "step": 3102
    },
    {
      "epoch": 0.2669706616192033,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009930885185394918,
      "loss": 1.5565,
      "step": 3103
    },
    {
      "epoch": 0.267056697926525,
      "grad_norm": 1.8984375,
      "learning_rate": 0.000993084037302216,
      "loss": 1.3486,
      "step": 3104
    },
    {
      "epoch": 0.26714273423384666,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009930795546227653,
      "loss": 1.6202,
      "step": 3105
    },
    {
      "epoch": 0.2672287705411684,
      "grad_norm": 1.90625,
      "learning_rate": 0.000993075070501153,
      "loss": 1.4165,
      "step": 3106
    },
    {
      "epoch": 0.26731480684849007,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000993070584937392,
      "loss": 1.2119,
      "step": 3107
    },
    {
      "epoch": 0.26740084315581175,
      "grad_norm": 1.375,
      "learning_rate": 0.000993066097931495,
      "loss": 1.1426,
      "step": 3108
    },
    {
      "epoch": 0.26748687946313343,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009930616094834762,
      "loss": 1.9321,
      "step": 3109
    },
    {
      "epoch": 0.2675729157704551,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009930571195933477,
      "loss": 1.1528,
      "step": 3110
    },
    {
      "epoch": 0.26765895207777685,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009930526282611232,
      "loss": 1.8689,
      "step": 3111
    },
    {
      "epoch": 0.2677449883850985,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009930481354868154,
      "loss": 1.6394,
      "step": 3112
    },
    {
      "epoch": 0.2678310246924202,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000993043641270438,
      "loss": 1.4489,
      "step": 3113
    },
    {
      "epoch": 0.2679170609997419,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009930391456120036,
      "loss": 1.3664,
      "step": 3114
    },
    {
      "epoch": 0.26800309730706356,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009930346485115258,
      "loss": 1.4327,
      "step": 3115
    },
    {
      "epoch": 0.26808913361438524,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009930301499690175,
      "loss": 1.389,
      "step": 3116
    },
    {
      "epoch": 0.268175169921707,
      "grad_norm": 1.375,
      "learning_rate": 0.0009930256499844918,
      "loss": 1.7286,
      "step": 3117
    },
    {
      "epoch": 0.26826120622902866,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009930211485579621,
      "loss": 0.8379,
      "step": 3118
    },
    {
      "epoch": 0.26834724253635034,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009930166456894414,
      "loss": 1.0762,
      "step": 3119
    },
    {
      "epoch": 0.268433278843672,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009930121413789428,
      "loss": 1.8432,
      "step": 3120
    },
    {
      "epoch": 0.2685193151509937,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009930076356264794,
      "loss": 1.8916,
      "step": 3121
    },
    {
      "epoch": 0.26860535145831543,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009930031284320648,
      "loss": 1.3513,
      "step": 3122
    },
    {
      "epoch": 0.2686913877656371,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009929986197957119,
      "loss": 1.7431,
      "step": 3123
    },
    {
      "epoch": 0.2687774240729588,
      "grad_norm": 1.15625,
      "learning_rate": 0.000992994109717434,
      "loss": 1.4786,
      "step": 3124
    },
    {
      "epoch": 0.2688634603802805,
      "grad_norm": 1.8125,
      "learning_rate": 0.000992989598197244,
      "loss": 1.6523,
      "step": 3125
    },
    {
      "epoch": 0.26894949668760215,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009929850852351556,
      "loss": 1.4907,
      "step": 3126
    },
    {
      "epoch": 0.26903553299492383,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009929805708311814,
      "loss": 1.1869,
      "step": 3127
    },
    {
      "epoch": 0.26912156930224557,
      "grad_norm": 1.4375,
      "learning_rate": 0.000992976054985335,
      "loss": 1.9298,
      "step": 3128
    },
    {
      "epoch": 0.26920760560956725,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009929715376976295,
      "loss": 1.5406,
      "step": 3129
    },
    {
      "epoch": 0.2692936419168889,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009929670189680784,
      "loss": 1.6863,
      "step": 3130
    },
    {
      "epoch": 0.2693796782242106,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009929624987966943,
      "loss": 0.9956,
      "step": 3131
    },
    {
      "epoch": 0.2694657145315323,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009929579771834908,
      "loss": 2.1845,
      "step": 3132
    },
    {
      "epoch": 0.269551750838854,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009929534541284813,
      "loss": 1.1908,
      "step": 3133
    },
    {
      "epoch": 0.2696377871461757,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009929489296316787,
      "loss": 1.9316,
      "step": 3134
    },
    {
      "epoch": 0.2697238234534974,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009929444036930965,
      "loss": 1.4952,
      "step": 3135
    },
    {
      "epoch": 0.26980985976081906,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009929398763127476,
      "loss": 1.1951,
      "step": 3136
    },
    {
      "epoch": 0.26989589606814074,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009929353474906455,
      "loss": 1.5675,
      "step": 3137
    },
    {
      "epoch": 0.2699819323754624,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009929308172268033,
      "loss": 1.9531,
      "step": 3138
    },
    {
      "epoch": 0.27006796868278415,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009929262855212344,
      "loss": 1.0199,
      "step": 3139
    },
    {
      "epoch": 0.27015400499010583,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000992921752373952,
      "loss": 1.4965,
      "step": 3140
    },
    {
      "epoch": 0.2702400412974275,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009929172177849694,
      "loss": 1.6299,
      "step": 3141
    },
    {
      "epoch": 0.2703260776047492,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009929126817542998,
      "loss": 1.3881,
      "step": 3142
    },
    {
      "epoch": 0.2704121139120709,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009929081442819562,
      "loss": 1.501,
      "step": 3143
    },
    {
      "epoch": 0.2704981502193926,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009929036053679524,
      "loss": 1.1933,
      "step": 3144
    },
    {
      "epoch": 0.2705841865267143,
      "grad_norm": 1.5,
      "learning_rate": 0.0009928990650123015,
      "loss": 1.0585,
      "step": 3145
    },
    {
      "epoch": 0.27067022283403597,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009928945232150165,
      "loss": 1.3123,
      "step": 3146
    },
    {
      "epoch": 0.27075625914135765,
      "grad_norm": 1.203125,
      "learning_rate": 0.000992889979976111,
      "loss": 1.0687,
      "step": 3147
    },
    {
      "epoch": 0.2708422954486793,
      "grad_norm": 1.671875,
      "learning_rate": 0.000992885435295598,
      "loss": 2.0547,
      "step": 3148
    },
    {
      "epoch": 0.270928331756001,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009928808891734912,
      "loss": 2.5468,
      "step": 3149
    },
    {
      "epoch": 0.27101436806332274,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009928763416098036,
      "loss": 1.8958,
      "step": 3150
    },
    {
      "epoch": 0.2711004043706444,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009928717926045484,
      "loss": 1.4041,
      "step": 3151
    },
    {
      "epoch": 0.2711864406779661,
      "grad_norm": 1.328125,
      "learning_rate": 0.000992867242157739,
      "loss": 1.5725,
      "step": 3152
    },
    {
      "epoch": 0.2712724769852878,
      "grad_norm": 1.375,
      "learning_rate": 0.0009928626902693891,
      "loss": 1.3285,
      "step": 3153
    },
    {
      "epoch": 0.27135851329260946,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009928581369395115,
      "loss": 1.1983,
      "step": 3154
    },
    {
      "epoch": 0.2714445495999312,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009928535821681196,
      "loss": 1.286,
      "step": 3155
    },
    {
      "epoch": 0.2715305859072529,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009928490259552272,
      "loss": 0.8944,
      "step": 3156
    },
    {
      "epoch": 0.27161662221457455,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000992844468300847,
      "loss": 1.3852,
      "step": 3157
    },
    {
      "epoch": 0.27170265852189623,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009928399092049925,
      "loss": 0.8728,
      "step": 3158
    },
    {
      "epoch": 0.2717886948292179,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009928353486676773,
      "loss": 1.7623,
      "step": 3159
    },
    {
      "epoch": 0.2718747311365396,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009928307866889144,
      "loss": 0.9734,
      "step": 3160
    },
    {
      "epoch": 0.27196076744386133,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009928262232687174,
      "loss": 1.1698,
      "step": 3161
    },
    {
      "epoch": 0.272046803751183,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009928216584070995,
      "loss": 1.4394,
      "step": 3162
    },
    {
      "epoch": 0.2721328400585047,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009928170921040742,
      "loss": 1.0106,
      "step": 3163
    },
    {
      "epoch": 0.27221887636582637,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009928125243596547,
      "loss": 1.2838,
      "step": 3164
    },
    {
      "epoch": 0.27230491267314805,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009928079551738544,
      "loss": 1.0851,
      "step": 3165
    },
    {
      "epoch": 0.2723909489804698,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009928033845466866,
      "loss": 0.9527,
      "step": 3166
    },
    {
      "epoch": 0.27247698528779146,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009927988124781646,
      "loss": 1.6663,
      "step": 3167
    },
    {
      "epoch": 0.27256302159511314,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009927942389683021,
      "loss": 1.2775,
      "step": 3168
    },
    {
      "epoch": 0.2726490579024348,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009927896640171126,
      "loss": 1.3709,
      "step": 3169
    },
    {
      "epoch": 0.2727350942097565,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009927850876246087,
      "loss": 1.8081,
      "step": 3170
    },
    {
      "epoch": 0.2728211305170782,
      "grad_norm": 1.25,
      "learning_rate": 0.0009927805097908044,
      "loss": 1.2146,
      "step": 3171
    },
    {
      "epoch": 0.2729071668243999,
      "grad_norm": 1.296875,
      "learning_rate": 0.000992775930515713,
      "loss": 1.6002,
      "step": 3172
    },
    {
      "epoch": 0.2729932031317216,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009927713497993478,
      "loss": 1.1572,
      "step": 3173
    },
    {
      "epoch": 0.2730792394390433,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009927667676417225,
      "loss": 1.8563,
      "step": 3174
    },
    {
      "epoch": 0.27316527574636496,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009927621840428499,
      "loss": 1.4025,
      "step": 3175
    },
    {
      "epoch": 0.27325131205368663,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009927575990027438,
      "loss": 1.3457,
      "step": 3176
    },
    {
      "epoch": 0.27333734836100837,
      "grad_norm": 1.75,
      "learning_rate": 0.0009927530125214178,
      "loss": 1.9687,
      "step": 3177
    },
    {
      "epoch": 0.27342338466833005,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009927484245988848,
      "loss": 1.1574,
      "step": 3178
    },
    {
      "epoch": 0.27350942097565173,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009927438352351585,
      "loss": 2.1272,
      "step": 3179
    },
    {
      "epoch": 0.2735954572829734,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009927392444302526,
      "loss": 1.0944,
      "step": 3180
    },
    {
      "epoch": 0.2736814935902951,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00099273465218418,
      "loss": 0.9052,
      "step": 3181
    },
    {
      "epoch": 0.27376752989761677,
      "grad_norm": 2.125,
      "learning_rate": 0.0009927300584969545,
      "loss": 1.7803,
      "step": 3182
    },
    {
      "epoch": 0.2738535662049385,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009927254633685892,
      "loss": 1.9297,
      "step": 3183
    },
    {
      "epoch": 0.2739396025122602,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009927208667990979,
      "loss": 1.8257,
      "step": 3184
    },
    {
      "epoch": 0.27402563881958186,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009927162687884939,
      "loss": 1.2301,
      "step": 3185
    },
    {
      "epoch": 0.27411167512690354,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009927116693367905,
      "loss": 1.8768,
      "step": 3186
    },
    {
      "epoch": 0.2741977114342252,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009927070684440016,
      "loss": 1.6536,
      "step": 3187
    },
    {
      "epoch": 0.27428374774154696,
      "grad_norm": 1.34375,
      "learning_rate": 0.00099270246611014,
      "loss": 1.0305,
      "step": 3188
    },
    {
      "epoch": 0.27436978404886864,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009926978623352198,
      "loss": 2.0361,
      "step": 3189
    },
    {
      "epoch": 0.2744558203561903,
      "grad_norm": 1.140625,
      "learning_rate": 0.000992693257119254,
      "loss": 1.068,
      "step": 3190
    },
    {
      "epoch": 0.274541856663512,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009926886504622564,
      "loss": 2.0331,
      "step": 3191
    },
    {
      "epoch": 0.2746278929708337,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009926840423642403,
      "loss": 0.9148,
      "step": 3192
    },
    {
      "epoch": 0.27471392927815536,
      "grad_norm": 1.609375,
      "learning_rate": 0.000992679432825219,
      "loss": 1.5074,
      "step": 3193
    },
    {
      "epoch": 0.2747999655854771,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009926748218452063,
      "loss": 0.6663,
      "step": 3194
    },
    {
      "epoch": 0.27488600189279877,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009926702094242156,
      "loss": 1.5744,
      "step": 3195
    },
    {
      "epoch": 0.27497203820012045,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009926655955622603,
      "loss": 1.2429,
      "step": 3196
    },
    {
      "epoch": 0.27505807450744213,
      "grad_norm": 1.59375,
      "learning_rate": 0.000992660980259354,
      "loss": 1.526,
      "step": 3197
    },
    {
      "epoch": 0.2751441108147638,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009926563635155102,
      "loss": 0.8978,
      "step": 3198
    },
    {
      "epoch": 0.27523014712208554,
      "grad_norm": 1.25,
      "learning_rate": 0.0009926517453307422,
      "loss": 1.1598,
      "step": 3199
    },
    {
      "epoch": 0.2753161834294072,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009926471257050637,
      "loss": 1.8493,
      "step": 3200
    },
    {
      "epoch": 0.2754022197367289,
      "grad_norm": 1.796875,
      "learning_rate": 0.000992642504638488,
      "loss": 2.0582,
      "step": 3201
    },
    {
      "epoch": 0.2754882560440506,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009926378821310292,
      "loss": 1.2024,
      "step": 3202
    },
    {
      "epoch": 0.27557429235137226,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009926332581827002,
      "loss": 1.031,
      "step": 3203
    },
    {
      "epoch": 0.27566032865869394,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009926286327935148,
      "loss": 1.4693,
      "step": 3204
    },
    {
      "epoch": 0.2757463649660157,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009926240059634863,
      "loss": 1.5246,
      "step": 3205
    },
    {
      "epoch": 0.27583240127333736,
      "grad_norm": 1.125,
      "learning_rate": 0.0009926193776926286,
      "loss": 1.1387,
      "step": 3206
    },
    {
      "epoch": 0.27591843758065904,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000992614747980955,
      "loss": 1.2495,
      "step": 3207
    },
    {
      "epoch": 0.2760044738879807,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009926101168284789,
      "loss": 1.484,
      "step": 3208
    },
    {
      "epoch": 0.2760905101953024,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009926054842352141,
      "loss": 1.4778,
      "step": 3209
    },
    {
      "epoch": 0.27617654650262413,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009926008502011742,
      "loss": 0.839,
      "step": 3210
    },
    {
      "epoch": 0.2762625828099458,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009925962147263722,
      "loss": 1.0475,
      "step": 3211
    },
    {
      "epoch": 0.2763486191172675,
      "grad_norm": 1.375,
      "learning_rate": 0.0009925915778108225,
      "loss": 1.3498,
      "step": 3212
    },
    {
      "epoch": 0.27643465542458917,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009925869394545382,
      "loss": 1.3266,
      "step": 3213
    },
    {
      "epoch": 0.27652069173191085,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009925822996575326,
      "loss": 0.8628,
      "step": 3214
    },
    {
      "epoch": 0.27660672803923253,
      "grad_norm": 2.84375,
      "learning_rate": 0.0009925776584198199,
      "loss": 1.5364,
      "step": 3215
    },
    {
      "epoch": 0.27669276434655427,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009925730157414132,
      "loss": 1.815,
      "step": 3216
    },
    {
      "epoch": 0.27677880065387594,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009925683716223263,
      "loss": 1.5183,
      "step": 3217
    },
    {
      "epoch": 0.2768648369611976,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009925637260625726,
      "loss": 1.3104,
      "step": 3218
    },
    {
      "epoch": 0.2769508732685193,
      "grad_norm": 2.015625,
      "learning_rate": 0.000992559079062166,
      "loss": 1.0187,
      "step": 3219
    },
    {
      "epoch": 0.277036909575841,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009925544306211198,
      "loss": 1.2504,
      "step": 3220
    },
    {
      "epoch": 0.2771229458831627,
      "grad_norm": 1.5,
      "learning_rate": 0.0009925497807394476,
      "loss": 0.7838,
      "step": 3221
    },
    {
      "epoch": 0.2772089821904844,
      "grad_norm": 1.671875,
      "learning_rate": 0.000992545129417163,
      "loss": 1.2094,
      "step": 3222
    },
    {
      "epoch": 0.2772950184978061,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00099254047665428,
      "loss": 1.5471,
      "step": 3223
    },
    {
      "epoch": 0.27738105480512776,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009925358224508116,
      "loss": 1.4122,
      "step": 3224
    },
    {
      "epoch": 0.27746709111244944,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000992531166806772,
      "loss": 1.358,
      "step": 3225
    },
    {
      "epoch": 0.2775531274197711,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009925265097221743,
      "loss": 0.6753,
      "step": 3226
    },
    {
      "epoch": 0.27763916372709285,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009925218511970325,
      "loss": 1.5935,
      "step": 3227
    },
    {
      "epoch": 0.27772520003441453,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009925171912313598,
      "loss": 1.7669,
      "step": 3228
    },
    {
      "epoch": 0.2778112363417362,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009925125298251704,
      "loss": 1.0709,
      "step": 3229
    },
    {
      "epoch": 0.2778972726490579,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009925078669784774,
      "loss": 1.5342,
      "step": 3230
    },
    {
      "epoch": 0.27798330895637957,
      "grad_norm": 1.46875,
      "learning_rate": 0.000992503202691295,
      "loss": 1.3541,
      "step": 3231
    },
    {
      "epoch": 0.2780693452637013,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009924985369636364,
      "loss": 1.0631,
      "step": 3232
    },
    {
      "epoch": 0.278155381571023,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009924938697955152,
      "loss": 0.9082,
      "step": 3233
    },
    {
      "epoch": 0.27824141787834467,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009924892011869454,
      "loss": 1.2246,
      "step": 3234
    },
    {
      "epoch": 0.27832745418566635,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009924845311379403,
      "loss": 2.2313,
      "step": 3235
    },
    {
      "epoch": 0.278413490492988,
      "grad_norm": 1.25,
      "learning_rate": 0.000992479859648514,
      "loss": 1.3858,
      "step": 3236
    },
    {
      "epoch": 0.2784995268003097,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009924751867186796,
      "loss": 1.4471,
      "step": 3237
    },
    {
      "epoch": 0.27858556310763144,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000992470512348451,
      "loss": 1.4672,
      "step": 3238
    },
    {
      "epoch": 0.2786715994149531,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009924658365378421,
      "loss": 1.1175,
      "step": 3239
    },
    {
      "epoch": 0.2787576357222748,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009924611592868665,
      "loss": 1.7689,
      "step": 3240
    },
    {
      "epoch": 0.2788436720295965,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009924564805955376,
      "loss": 2.0274,
      "step": 3241
    },
    {
      "epoch": 0.27892970833691816,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009924518004638693,
      "loss": 0.8399,
      "step": 3242
    },
    {
      "epoch": 0.2790157446442399,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009924471188918755,
      "loss": 1.1306,
      "step": 3243
    },
    {
      "epoch": 0.2791017809515616,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009924424358795695,
      "loss": 1.1638,
      "step": 3244
    },
    {
      "epoch": 0.27918781725888325,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009924377514269652,
      "loss": 0.8718,
      "step": 3245
    },
    {
      "epoch": 0.27927385356620493,
      "grad_norm": 1.8125,
      "learning_rate": 0.000992433065534076,
      "loss": 2.0162,
      "step": 3246
    },
    {
      "epoch": 0.2793598898735266,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009924283782009162,
      "loss": 1.522,
      "step": 3247
    },
    {
      "epoch": 0.2794459261808483,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000992423689427499,
      "loss": 2.1902,
      "step": 3248
    },
    {
      "epoch": 0.27953196248817,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000992418999213838,
      "loss": 1.7405,
      "step": 3249
    },
    {
      "epoch": 0.2796179987954917,
      "grad_norm": 4.75,
      "learning_rate": 0.0009924143075599476,
      "loss": 1.3202,
      "step": 3250
    },
    {
      "epoch": 0.2797040351028134,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009924096144658411,
      "loss": 1.5845,
      "step": 3251
    },
    {
      "epoch": 0.27979007141013507,
      "grad_norm": 1.234375,
      "learning_rate": 0.000992404919931532,
      "loss": 1.1376,
      "step": 3252
    },
    {
      "epoch": 0.27987610771745675,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009924002239570346,
      "loss": 1.2027,
      "step": 3253
    },
    {
      "epoch": 0.2799621440247785,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000992395526542362,
      "loss": 2.1141,
      "step": 3254
    },
    {
      "epoch": 0.28004818033210016,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009923908276875284,
      "loss": 1.6571,
      "step": 3255
    },
    {
      "epoch": 0.28013421663942184,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009923861273925475,
      "loss": 1.3796,
      "step": 3256
    },
    {
      "epoch": 0.2802202529467435,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009923814256574328,
      "loss": 1.2478,
      "step": 3257
    },
    {
      "epoch": 0.2803062892540652,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009923767224821983,
      "loss": 1.5642,
      "step": 3258
    },
    {
      "epoch": 0.2803923255613869,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009923720178668573,
      "loss": 0.8857,
      "step": 3259
    },
    {
      "epoch": 0.2804783618687086,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009923673118114242,
      "loss": 1.05,
      "step": 3260
    },
    {
      "epoch": 0.2805643981760303,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009923626043159127,
      "loss": 1.3527,
      "step": 3261
    },
    {
      "epoch": 0.280650434483352,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000992357895380336,
      "loss": 1.1033,
      "step": 3262
    },
    {
      "epoch": 0.28073647079067365,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009923531850047082,
      "loss": 1.7124,
      "step": 3263
    },
    {
      "epoch": 0.28082250709799533,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009923484731890434,
      "loss": 1.661,
      "step": 3264
    },
    {
      "epoch": 0.28090854340531707,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009923437599333547,
      "loss": 0.7515,
      "step": 3265
    },
    {
      "epoch": 0.28099457971263875,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009923390452376564,
      "loss": 1.635,
      "step": 3266
    },
    {
      "epoch": 0.2810806160199604,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009923343291019621,
      "loss": 2.0919,
      "step": 3267
    },
    {
      "epoch": 0.2811666523272821,
      "grad_norm": 1.8125,
      "learning_rate": 0.000992329611526286,
      "loss": 1.9557,
      "step": 3268
    },
    {
      "epoch": 0.2812526886346038,
      "grad_norm": 1.1875,
      "learning_rate": 0.000992324892510641,
      "loss": 1.3178,
      "step": 3269
    },
    {
      "epoch": 0.28133872494192547,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009923201720550417,
      "loss": 0.9328,
      "step": 3270
    },
    {
      "epoch": 0.2814247612492472,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009923154501595018,
      "loss": 2.062,
      "step": 3271
    },
    {
      "epoch": 0.2815107975565689,
      "grad_norm": 1.875,
      "learning_rate": 0.0009923107268240347,
      "loss": 2.0317,
      "step": 3272
    },
    {
      "epoch": 0.28159683386389056,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009923060020486545,
      "loss": 1.1534,
      "step": 3273
    },
    {
      "epoch": 0.28168287017121224,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009923012758333751,
      "loss": 1.0288,
      "step": 3274
    },
    {
      "epoch": 0.2817689064785339,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009922965481782102,
      "loss": 0.8866,
      "step": 3275
    },
    {
      "epoch": 0.28185494278585566,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009922918190831736,
      "loss": 0.9782,
      "step": 3276
    },
    {
      "epoch": 0.28194097909317734,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009922870885482793,
      "loss": 1.3843,
      "step": 3277
    },
    {
      "epoch": 0.282027015400499,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009922823565735408,
      "loss": 2.1919,
      "step": 3278
    },
    {
      "epoch": 0.2821130517078207,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009922776231589723,
      "loss": 2.2801,
      "step": 3279
    },
    {
      "epoch": 0.2821990880151424,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009922728883045874,
      "loss": 1.9816,
      "step": 3280
    },
    {
      "epoch": 0.28228512432246405,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009922681520104,
      "loss": 0.7636,
      "step": 3281
    },
    {
      "epoch": 0.2823711606297858,
      "grad_norm": 1.59375,
      "learning_rate": 0.000992263414276424,
      "loss": 1.195,
      "step": 3282
    },
    {
      "epoch": 0.28245719693710747,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009922586751026736,
      "loss": 1.676,
      "step": 3283
    },
    {
      "epoch": 0.28254323324442915,
      "grad_norm": 1.25,
      "learning_rate": 0.000992253934489162,
      "loss": 0.6482,
      "step": 3284
    },
    {
      "epoch": 0.28262926955175083,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009922491924359033,
      "loss": 2.0181,
      "step": 3285
    },
    {
      "epoch": 0.2827153058590725,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009922444489429117,
      "loss": 1.7783,
      "step": 3286
    },
    {
      "epoch": 0.28280134216639424,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009922397040102006,
      "loss": 0.8842,
      "step": 3287
    },
    {
      "epoch": 0.2828873784737159,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009922349576377842,
      "loss": 1.1988,
      "step": 3288
    },
    {
      "epoch": 0.2829734147810376,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0009922302098256763,
      "loss": 1.9075,
      "step": 3289
    },
    {
      "epoch": 0.2830594510883593,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009922254605738909,
      "loss": 1.2788,
      "step": 3290
    },
    {
      "epoch": 0.28314548739568096,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009922207098824417,
      "loss": 2.0091,
      "step": 3291
    },
    {
      "epoch": 0.28323152370300264,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009922159577513425,
      "loss": 1.4684,
      "step": 3292
    },
    {
      "epoch": 0.2833175600103244,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009922112041806075,
      "loss": 1.8421,
      "step": 3293
    },
    {
      "epoch": 0.28340359631764606,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009922064491702503,
      "loss": 1.5016,
      "step": 3294
    },
    {
      "epoch": 0.28348963262496774,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000992201692720285,
      "loss": 0.9116,
      "step": 3295
    },
    {
      "epoch": 0.2835756689322894,
      "grad_norm": 1.375,
      "learning_rate": 0.0009921969348307255,
      "loss": 1.0363,
      "step": 3296
    },
    {
      "epoch": 0.2836617052396111,
      "grad_norm": 1.34375,
      "learning_rate": 0.000992192175501586,
      "loss": 1.8627,
      "step": 3297
    },
    {
      "epoch": 0.28374774154693283,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009921874147328798,
      "loss": 1.7975,
      "step": 3298
    },
    {
      "epoch": 0.2838337778542545,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000992182652524621,
      "loss": 0.6119,
      "step": 3299
    },
    {
      "epoch": 0.2839198141615762,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009921778888768238,
      "loss": 1.7842,
      "step": 3300
    },
    {
      "epoch": 0.28400585046889787,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009921731237895021,
      "loss": 0.9719,
      "step": 3301
    },
    {
      "epoch": 0.28409188677621955,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009921683572626698,
      "loss": 1.6182,
      "step": 3302
    },
    {
      "epoch": 0.28417792308354123,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009921635892963406,
      "loss": 1.1094,
      "step": 3303
    },
    {
      "epoch": 0.28426395939086296,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009921588198905287,
      "loss": 0.878,
      "step": 3304
    },
    {
      "epoch": 0.28434999569818464,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009921540490452478,
      "loss": 1.3994,
      "step": 3305
    },
    {
      "epoch": 0.2844360320055063,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009921492767605122,
      "loss": 1.042,
      "step": 3306
    },
    {
      "epoch": 0.284522068312828,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009921445030363356,
      "loss": 1.3198,
      "step": 3307
    },
    {
      "epoch": 0.2846081046201497,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009921397278727323,
      "loss": 1.3028,
      "step": 3308
    },
    {
      "epoch": 0.2846941409274714,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009921349512697158,
      "loss": 1.5089,
      "step": 3309
    },
    {
      "epoch": 0.2847801772347931,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009921301732273003,
      "loss": 1.9722,
      "step": 3310
    },
    {
      "epoch": 0.2848662135421148,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009921253937454997,
      "loss": 1.606,
      "step": 3311
    },
    {
      "epoch": 0.28495224984943646,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009921206128243282,
      "loss": 0.8395,
      "step": 3312
    },
    {
      "epoch": 0.28503828615675814,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009921158304637993,
      "loss": 1.1378,
      "step": 3313
    },
    {
      "epoch": 0.2851243224640798,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009921110466639277,
      "loss": 0.8894,
      "step": 3314
    },
    {
      "epoch": 0.28521035877140155,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009921062614247267,
      "loss": 1.9889,
      "step": 3315
    },
    {
      "epoch": 0.28529639507872323,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009921014747462107,
      "loss": 1.4486,
      "step": 3316
    },
    {
      "epoch": 0.2853824313860449,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009920966866283935,
      "loss": 0.8462,
      "step": 3317
    },
    {
      "epoch": 0.2854684676933666,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009920918970712893,
      "loss": 1.3479,
      "step": 3318
    },
    {
      "epoch": 0.28555450400068827,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000992087106074912,
      "loss": 1.1858,
      "step": 3319
    },
    {
      "epoch": 0.28564054030801,
      "grad_norm": 1.375,
      "learning_rate": 0.0009920823136392756,
      "loss": 0.9281,
      "step": 3320
    },
    {
      "epoch": 0.2857265766153317,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009920775197643943,
      "loss": 1.0285,
      "step": 3321
    },
    {
      "epoch": 0.28581261292265336,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009920727244502818,
      "loss": 1.785,
      "step": 3322
    },
    {
      "epoch": 0.28589864922997504,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009920679276969524,
      "loss": 2.1071,
      "step": 3323
    },
    {
      "epoch": 0.2859846855372967,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009920631295044198,
      "loss": 1.3775,
      "step": 3324
    },
    {
      "epoch": 0.2860707218446184,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009920583298726985,
      "loss": 1.2737,
      "step": 3325
    },
    {
      "epoch": 0.28615675815194014,
      "grad_norm": 1.1875,
      "learning_rate": 0.000992053528801802,
      "loss": 1.0462,
      "step": 3326
    },
    {
      "epoch": 0.2862427944592618,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009920487262917448,
      "loss": 0.9242,
      "step": 3327
    },
    {
      "epoch": 0.2863288307665835,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009920439223425408,
      "loss": 1.5153,
      "step": 3328
    },
    {
      "epoch": 0.2864148670739052,
      "grad_norm": 1.375,
      "learning_rate": 0.000992039116954204,
      "loss": 1.1254,
      "step": 3329
    },
    {
      "epoch": 0.28650090338122686,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009920343101267484,
      "loss": 2.017,
      "step": 3330
    },
    {
      "epoch": 0.2865869396885486,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009920295018601883,
      "loss": 1.2166,
      "step": 3331
    },
    {
      "epoch": 0.28667297599587027,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009920246921545374,
      "loss": 1.5615,
      "step": 3332
    },
    {
      "epoch": 0.28675901230319195,
      "grad_norm": 1.546875,
      "learning_rate": 0.00099201988100981,
      "loss": 1.6268,
      "step": 3333
    },
    {
      "epoch": 0.28684504861051363,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009920150684260202,
      "loss": 1.9907,
      "step": 3334
    },
    {
      "epoch": 0.2869310849178353,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000992010254403182,
      "loss": 0.9789,
      "step": 3335
    },
    {
      "epoch": 0.287017121225157,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009920054389413095,
      "loss": 2.3717,
      "step": 3336
    },
    {
      "epoch": 0.2871031575324787,
      "grad_norm": 1.375,
      "learning_rate": 0.000992000622040417,
      "loss": 1.5724,
      "step": 3337
    },
    {
      "epoch": 0.2871891938398004,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000991995803700518,
      "loss": 1.5817,
      "step": 3338
    },
    {
      "epoch": 0.2872752301471221,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009919909839216272,
      "loss": 0.9781,
      "step": 3339
    },
    {
      "epoch": 0.28736126645444376,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009919861627037585,
      "loss": 1.9131,
      "step": 3340
    },
    {
      "epoch": 0.28744730276176544,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009919813400469258,
      "loss": 1.0232,
      "step": 3341
    },
    {
      "epoch": 0.2875333390690872,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009919765159511435,
      "loss": 1.0226,
      "step": 3342
    },
    {
      "epoch": 0.28761937537640886,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009919716904164253,
      "loss": 1.4001,
      "step": 3343
    },
    {
      "epoch": 0.28770541168373054,
      "grad_norm": 1.5625,
      "learning_rate": 0.000991966863442786,
      "loss": 1.9169,
      "step": 3344
    },
    {
      "epoch": 0.2877914479910522,
      "grad_norm": 1.9609375,
      "learning_rate": 0.000991962035030239,
      "loss": 1.6193,
      "step": 3345
    },
    {
      "epoch": 0.2878774842983739,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009919572051787988,
      "loss": 0.9366,
      "step": 3346
    },
    {
      "epoch": 0.2879635206056956,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009919523738884796,
      "loss": 1.3482,
      "step": 3347
    },
    {
      "epoch": 0.2880495569130173,
      "grad_norm": 1.265625,
      "learning_rate": 0.000991947541159295,
      "loss": 1.7308,
      "step": 3348
    },
    {
      "epoch": 0.288135593220339,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009919427069912598,
      "loss": 1.0969,
      "step": 3349
    },
    {
      "epoch": 0.28822162952766067,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009919378713843879,
      "loss": 1.3548,
      "step": 3350
    },
    {
      "epoch": 0.28830766583498235,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009919330343386933,
      "loss": 1.6476,
      "step": 3351
    },
    {
      "epoch": 0.28839370214230403,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009919281958541903,
      "loss": 1.6088,
      "step": 3352
    },
    {
      "epoch": 0.28847973844962577,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000991923355930893,
      "loss": 1.7576,
      "step": 3353
    },
    {
      "epoch": 0.28856577475694745,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009919185145688156,
      "loss": 2.2321,
      "step": 3354
    },
    {
      "epoch": 0.2886518110642691,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009919136717679723,
      "loss": 1.4606,
      "step": 3355
    },
    {
      "epoch": 0.2887378473715908,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009919088275283769,
      "loss": 1.3387,
      "step": 3356
    },
    {
      "epoch": 0.2888238836789125,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009919039818500441,
      "loss": 1.5256,
      "step": 3357
    },
    {
      "epoch": 0.28890991998623416,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009918991347329876,
      "loss": 1.153,
      "step": 3358
    },
    {
      "epoch": 0.2889959562935559,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000991894286177222,
      "loss": 1.5961,
      "step": 3359
    },
    {
      "epoch": 0.2890819926008776,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009918894361827613,
      "loss": 1.2432,
      "step": 3360
    },
    {
      "epoch": 0.28916802890819926,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009918845847496193,
      "loss": 1.9639,
      "step": 3361
    },
    {
      "epoch": 0.28925406521552094,
      "grad_norm": 1.265625,
      "learning_rate": 0.000991879731877811,
      "loss": 1.0743,
      "step": 3362
    },
    {
      "epoch": 0.2893401015228426,
      "grad_norm": 1.578125,
      "learning_rate": 0.00099187487756735,
      "loss": 1.1003,
      "step": 3363
    },
    {
      "epoch": 0.28942613783016435,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009918700218182505,
      "loss": 2.1346,
      "step": 3364
    },
    {
      "epoch": 0.28951217413748603,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000991865164630527,
      "loss": 0.715,
      "step": 3365
    },
    {
      "epoch": 0.2895982104448077,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009918603060041936,
      "loss": 1.9375,
      "step": 3366
    },
    {
      "epoch": 0.2896842467521294,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009918554459392643,
      "loss": 2.6676,
      "step": 3367
    },
    {
      "epoch": 0.2897702830594511,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009918505844357538,
      "loss": 1.7262,
      "step": 3368
    },
    {
      "epoch": 0.28985631936677275,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009918457214936757,
      "loss": 1.317,
      "step": 3369
    },
    {
      "epoch": 0.2899423556740945,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009918408571130445,
      "loss": 1.5193,
      "step": 3370
    },
    {
      "epoch": 0.29002839198141617,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009918359912938746,
      "loss": 2.3074,
      "step": 3371
    },
    {
      "epoch": 0.29011442828873785,
      "grad_norm": 1.25,
      "learning_rate": 0.00099183112403618,
      "loss": 1.0699,
      "step": 3372
    },
    {
      "epoch": 0.2902004645960595,
      "grad_norm": 1.453125,
      "learning_rate": 0.000991826255339975,
      "loss": 1.0445,
      "step": 3373
    },
    {
      "epoch": 0.2902865009033812,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000991821385205274,
      "loss": 1.8767,
      "step": 3374
    },
    {
      "epoch": 0.29037253721070294,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000991816513632091,
      "loss": 1.1567,
      "step": 3375
    },
    {
      "epoch": 0.2904585735180246,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009918116406204404,
      "loss": 1.6883,
      "step": 3376
    },
    {
      "epoch": 0.2905446098253463,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009918067661703362,
      "loss": 1.5152,
      "step": 3377
    },
    {
      "epoch": 0.290630646132668,
      "grad_norm": 1.25,
      "learning_rate": 0.0009918018902817932,
      "loss": 1.0447,
      "step": 3378
    },
    {
      "epoch": 0.29071668243998966,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009917970129548251,
      "loss": 1.0913,
      "step": 3379
    },
    {
      "epoch": 0.29080271874731134,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009917921341894465,
      "loss": 1.2062,
      "step": 3380
    },
    {
      "epoch": 0.2908887550546331,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009917872539856713,
      "loss": 1.5241,
      "step": 3381
    },
    {
      "epoch": 0.29097479136195475,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009917823723435144,
      "loss": 1.2311,
      "step": 3382
    },
    {
      "epoch": 0.29106082766927643,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009917774892629894,
      "loss": 0.8937,
      "step": 3383
    },
    {
      "epoch": 0.2911468639765981,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000991772604744111,
      "loss": 1.556,
      "step": 3384
    },
    {
      "epoch": 0.2912329002839198,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009917677187868935,
      "loss": 1.4582,
      "step": 3385
    },
    {
      "epoch": 0.29131893659124153,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000991762831391351,
      "loss": 0.9389,
      "step": 3386
    },
    {
      "epoch": 0.2914049728985632,
      "grad_norm": 1.375,
      "learning_rate": 0.0009917579425574978,
      "loss": 1.2429,
      "step": 3387
    },
    {
      "epoch": 0.2914910092058849,
      "grad_norm": 1.1875,
      "learning_rate": 0.000991753052285348,
      "loss": 1.7261,
      "step": 3388
    },
    {
      "epoch": 0.29157704551320657,
      "grad_norm": 2.390625,
      "learning_rate": 0.0009917481605749166,
      "loss": 1.0533,
      "step": 3389
    },
    {
      "epoch": 0.29166308182052825,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009917432674262175,
      "loss": 1.749,
      "step": 3390
    },
    {
      "epoch": 0.2917491181278499,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009917383728392647,
      "loss": 1.7442,
      "step": 3391
    },
    {
      "epoch": 0.29183515443517166,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009917334768140729,
      "loss": 1.1309,
      "step": 3392
    },
    {
      "epoch": 0.29192119074249334,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009917285793506563,
      "loss": 1.7101,
      "step": 3393
    },
    {
      "epoch": 0.292007227049815,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009917236804490293,
      "loss": 1.994,
      "step": 3394
    },
    {
      "epoch": 0.2920932633571367,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009917187801092061,
      "loss": 1.286,
      "step": 3395
    },
    {
      "epoch": 0.2921792996644584,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009917138783312011,
      "loss": 1.99,
      "step": 3396
    },
    {
      "epoch": 0.2922653359717801,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009917089751150287,
      "loss": 1.4222,
      "step": 3397
    },
    {
      "epoch": 0.2923513722791018,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000991704070460703,
      "loss": 0.8809,
      "step": 3398
    },
    {
      "epoch": 0.2924374085864235,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009916991643682386,
      "loss": 1.6455,
      "step": 3399
    },
    {
      "epoch": 0.29252344489374515,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009916942568376499,
      "loss": 1.7282,
      "step": 3400
    },
    {
      "epoch": 0.29260948120106683,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009916893478689511,
      "loss": 1.6815,
      "step": 3401
    },
    {
      "epoch": 0.2926955175083885,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009916844374621566,
      "loss": 0.9366,
      "step": 3402
    },
    {
      "epoch": 0.29278155381571025,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009916795256172807,
      "loss": 1.5483,
      "step": 3403
    },
    {
      "epoch": 0.29286759012303193,
      "grad_norm": 1.875,
      "learning_rate": 0.0009916746123343379,
      "loss": 2.0093,
      "step": 3404
    },
    {
      "epoch": 0.2929536264303536,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009916696976133425,
      "loss": 1.3437,
      "step": 3405
    },
    {
      "epoch": 0.2930396627376753,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009916647814543086,
      "loss": 1.1567,
      "step": 3406
    },
    {
      "epoch": 0.29312569904499697,
      "grad_norm": 1.515625,
      "learning_rate": 0.000991659863857251,
      "loss": 1.4169,
      "step": 3407
    },
    {
      "epoch": 0.2932117353523187,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009916549448221838,
      "loss": 1.4537,
      "step": 3408
    },
    {
      "epoch": 0.2932977716596404,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009916500243491218,
      "loss": 1.3956,
      "step": 3409
    },
    {
      "epoch": 0.29338380796696206,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009916451024380788,
      "loss": 1.3023,
      "step": 3410
    },
    {
      "epoch": 0.29346984427428374,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009916401790890698,
      "loss": 2.1778,
      "step": 3411
    },
    {
      "epoch": 0.2935558805816054,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009916352543021085,
      "loss": 1.8079,
      "step": 3412
    },
    {
      "epoch": 0.2936419168889271,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009916303280772098,
      "loss": 2.0423,
      "step": 3413
    },
    {
      "epoch": 0.29372795319624884,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009916254004143884,
      "loss": 1.1446,
      "step": 3414
    },
    {
      "epoch": 0.2938139895035705,
      "grad_norm": 1.8984375,
      "learning_rate": 0.000991620471313658,
      "loss": 2.0262,
      "step": 3415
    },
    {
      "epoch": 0.2939000258108922,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009916155407750334,
      "loss": 1.2231,
      "step": 3416
    },
    {
      "epoch": 0.2939860621182139,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000991610608798529,
      "loss": 2.2428,
      "step": 3417
    },
    {
      "epoch": 0.29407209842553556,
      "grad_norm": 2.171875,
      "learning_rate": 0.000991605675384159,
      "loss": 1.5375,
      "step": 3418
    },
    {
      "epoch": 0.2941581347328573,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000991600740531938,
      "loss": 1.2183,
      "step": 3419
    },
    {
      "epoch": 0.29424417104017897,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009915958042418806,
      "loss": 1.6589,
      "step": 3420
    },
    {
      "epoch": 0.29433020734750065,
      "grad_norm": 1.171875,
      "learning_rate": 0.000991590866514001,
      "loss": 0.8757,
      "step": 3421
    },
    {
      "epoch": 0.29441624365482233,
      "grad_norm": 2.078125,
      "learning_rate": 0.000991585927348314,
      "loss": 2.3288,
      "step": 3422
    },
    {
      "epoch": 0.294502279962144,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009915809867448333,
      "loss": 1.6008,
      "step": 3423
    },
    {
      "epoch": 0.2945883162694657,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009915760447035743,
      "loss": 1.447,
      "step": 3424
    },
    {
      "epoch": 0.2946743525767874,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009915711012245506,
      "loss": 1.2192,
      "step": 3425
    },
    {
      "epoch": 0.2947603888841091,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009915661563077772,
      "loss": 1.3114,
      "step": 3426
    },
    {
      "epoch": 0.2948464251914308,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009915612099532683,
      "loss": 1.7291,
      "step": 3427
    },
    {
      "epoch": 0.29493246149875246,
      "grad_norm": 1.125,
      "learning_rate": 0.0009915562621610385,
      "loss": 2.4361,
      "step": 3428
    },
    {
      "epoch": 0.29501849780607414,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009915513129311025,
      "loss": 1.5946,
      "step": 3429
    },
    {
      "epoch": 0.2951045341133959,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009915463622634743,
      "loss": 1.6799,
      "step": 3430
    },
    {
      "epoch": 0.29519057042071756,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009915414101581684,
      "loss": 1.0733,
      "step": 3431
    },
    {
      "epoch": 0.29527660672803924,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009915364566151995,
      "loss": 1.3615,
      "step": 3432
    },
    {
      "epoch": 0.2953626430353609,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009915315016345825,
      "loss": 0.9687,
      "step": 3433
    },
    {
      "epoch": 0.2954486793426826,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000991526545216331,
      "loss": 1.4306,
      "step": 3434
    },
    {
      "epoch": 0.2955347156500043,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009915215873604602,
      "loss": 0.9777,
      "step": 3435
    },
    {
      "epoch": 0.295620751957326,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009915166280669843,
      "loss": 1.0248,
      "step": 3436
    },
    {
      "epoch": 0.2957067882646477,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000991511667335918,
      "loss": 1.9346,
      "step": 3437
    },
    {
      "epoch": 0.29579282457196937,
      "grad_norm": 1.125,
      "learning_rate": 0.0009915067051672753,
      "loss": 1.2044,
      "step": 3438
    },
    {
      "epoch": 0.29587886087929105,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009915017415610713,
      "loss": 1.4271,
      "step": 3439
    },
    {
      "epoch": 0.29596489718661273,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009914967765173202,
      "loss": 0.7821,
      "step": 3440
    },
    {
      "epoch": 0.29605093349393447,
      "grad_norm": 1.5,
      "learning_rate": 0.0009914918100360369,
      "loss": 1.9644,
      "step": 3441
    },
    {
      "epoch": 0.29613696980125614,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009914868421172353,
      "loss": 1.8796,
      "step": 3442
    },
    {
      "epoch": 0.2962230061085778,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009914818727609306,
      "loss": 1.5525,
      "step": 3443
    },
    {
      "epoch": 0.2963090424158995,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009914769019671366,
      "loss": 0.991,
      "step": 3444
    },
    {
      "epoch": 0.2963950787232212,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009914719297358684,
      "loss": 1.5277,
      "step": 3445
    },
    {
      "epoch": 0.29648111503054286,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009914669560671405,
      "loss": 0.78,
      "step": 3446
    },
    {
      "epoch": 0.2965671513378646,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009914619809609673,
      "loss": 1.3129,
      "step": 3447
    },
    {
      "epoch": 0.2966531876451863,
      "grad_norm": 2.3125,
      "learning_rate": 0.0009914570044173635,
      "loss": 1.8224,
      "step": 3448
    },
    {
      "epoch": 0.29673922395250796,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009914520264363434,
      "loss": 0.8567,
      "step": 3449
    },
    {
      "epoch": 0.29682526025982964,
      "grad_norm": 1.375,
      "learning_rate": 0.0009914470470179218,
      "loss": 1.4199,
      "step": 3450
    },
    {
      "epoch": 0.2969112965671513,
      "grad_norm": 1.4375,
      "learning_rate": 0.000991442066162113,
      "loss": 0.9215,
      "step": 3451
    },
    {
      "epoch": 0.29699733287447305,
      "grad_norm": 1.71875,
      "learning_rate": 0.000991437083868932,
      "loss": 1.824,
      "step": 3452
    },
    {
      "epoch": 0.29708336918179473,
      "grad_norm": 1.390625,
      "learning_rate": 0.000991432100138393,
      "loss": 0.7352,
      "step": 3453
    },
    {
      "epoch": 0.2971694054891164,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009914271149705106,
      "loss": 1.8239,
      "step": 3454
    },
    {
      "epoch": 0.2972554417964381,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009914221283652996,
      "loss": 1.2364,
      "step": 3455
    },
    {
      "epoch": 0.29734147810375977,
      "grad_norm": 1.5,
      "learning_rate": 0.0009914171403227743,
      "loss": 1.3176,
      "step": 3456
    },
    {
      "epoch": 0.29742751441108145,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009914121508429494,
      "loss": 2.1645,
      "step": 3457
    },
    {
      "epoch": 0.2975135507184032,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009914071599258396,
      "loss": 1.4511,
      "step": 3458
    },
    {
      "epoch": 0.29759958702572487,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009914021675714595,
      "loss": 1.4423,
      "step": 3459
    },
    {
      "epoch": 0.29768562333304655,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009913971737798238,
      "loss": 1.3302,
      "step": 3460
    },
    {
      "epoch": 0.2977716596403682,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009913921785509466,
      "loss": 1.6552,
      "step": 3461
    },
    {
      "epoch": 0.2978576959476899,
      "grad_norm": 1.234375,
      "learning_rate": 0.000991387181884843,
      "loss": 0.856,
      "step": 3462
    },
    {
      "epoch": 0.29794373225501164,
      "grad_norm": 1.375,
      "learning_rate": 0.0009913821837815274,
      "loss": 1.8908,
      "step": 3463
    },
    {
      "epoch": 0.2980297685623333,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009913771842410147,
      "loss": 1.162,
      "step": 3464
    },
    {
      "epoch": 0.298115804869655,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000991372183263319,
      "loss": 1.0043,
      "step": 3465
    },
    {
      "epoch": 0.2982018411769767,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009913671808484555,
      "loss": 1.2219,
      "step": 3466
    },
    {
      "epoch": 0.29828787748429836,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009913621769964383,
      "loss": 1.9607,
      "step": 3467
    },
    {
      "epoch": 0.29837391379162004,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009913571717072825,
      "loss": 1.8086,
      "step": 3468
    },
    {
      "epoch": 0.2984599500989418,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009913521649810024,
      "loss": 1.6267,
      "step": 3469
    },
    {
      "epoch": 0.29854598640626345,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000991347156817613,
      "loss": 1.9747,
      "step": 3470
    },
    {
      "epoch": 0.29863202271358513,
      "grad_norm": 1.25,
      "learning_rate": 0.0009913421472171286,
      "loss": 1.1924,
      "step": 3471
    },
    {
      "epoch": 0.2987180590209068,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009913371361795638,
      "loss": 1.6167,
      "step": 3472
    },
    {
      "epoch": 0.2988040953282285,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009913321237049336,
      "loss": 1.0637,
      "step": 3473
    },
    {
      "epoch": 0.2988901316355502,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009913271097932524,
      "loss": 0.9906,
      "step": 3474
    },
    {
      "epoch": 0.2989761679428719,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000991322094444535,
      "loss": 1.8259,
      "step": 3475
    },
    {
      "epoch": 0.2990622042501936,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000991317077658796,
      "loss": 1.1722,
      "step": 3476
    },
    {
      "epoch": 0.29914824055751527,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009913120594360503,
      "loss": 1.808,
      "step": 3477
    },
    {
      "epoch": 0.29923427686483695,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000991307039776312,
      "loss": 1.1196,
      "step": 3478
    },
    {
      "epoch": 0.2993203131721586,
      "grad_norm": 1.25,
      "learning_rate": 0.0009913020186795966,
      "loss": 1.5682,
      "step": 3479
    },
    {
      "epoch": 0.29940634947948036,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009912969961459182,
      "loss": 0.8222,
      "step": 3480
    },
    {
      "epoch": 0.29949238578680204,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009912919721752917,
      "loss": 1.4746,
      "step": 3481
    },
    {
      "epoch": 0.2995784220941237,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009912869467677316,
      "loss": 1.3511,
      "step": 3482
    },
    {
      "epoch": 0.2996644584014454,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009912819199232527,
      "loss": 1.7262,
      "step": 3483
    },
    {
      "epoch": 0.2997504947087671,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009912768916418698,
      "loss": 1.7818,
      "step": 3484
    },
    {
      "epoch": 0.2998365310160888,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009912718619235975,
      "loss": 0.9016,
      "step": 3485
    },
    {
      "epoch": 0.2999225673234105,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009912668307684508,
      "loss": 2.5971,
      "step": 3486
    },
    {
      "epoch": 0.3000086036307322,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009912617981764438,
      "loss": 1.1539,
      "step": 3487
    },
    {
      "epoch": 0.30009463993805385,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009912567641475918,
      "loss": 1.4338,
      "step": 3488
    },
    {
      "epoch": 0.30018067624537553,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009912517286819094,
      "loss": 1.5993,
      "step": 3489
    },
    {
      "epoch": 0.3002667125526972,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000991246691779411,
      "loss": 1.2852,
      "step": 3490
    },
    {
      "epoch": 0.30035274886001895,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009912416534401116,
      "loss": 1.7698,
      "step": 3491
    },
    {
      "epoch": 0.3004387851673406,
      "grad_norm": 1.484375,
      "learning_rate": 0.000991236613664026,
      "loss": 1.5751,
      "step": 3492
    },
    {
      "epoch": 0.3005248214746623,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000991231572451169,
      "loss": 1.2956,
      "step": 3493
    },
    {
      "epoch": 0.300610857781984,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009912265298015548,
      "loss": 1.2662,
      "step": 3494
    },
    {
      "epoch": 0.30069689408930567,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009912214857151986,
      "loss": 1.5594,
      "step": 3495
    },
    {
      "epoch": 0.3007829303966274,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009912164401921152,
      "loss": 0.8491,
      "step": 3496
    },
    {
      "epoch": 0.3008689667039491,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009912113932323194,
      "loss": 1.2566,
      "step": 3497
    },
    {
      "epoch": 0.30095500301127076,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009912063448358258,
      "loss": 1.0839,
      "step": 3498
    },
    {
      "epoch": 0.30104103931859244,
      "grad_norm": 1.953125,
      "learning_rate": 0.000991201295002649,
      "loss": 1.813,
      "step": 3499
    },
    {
      "epoch": 0.3011270756259141,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009911962437328039,
      "loss": 1.4072,
      "step": 3500
    },
    {
      "epoch": 0.3012131119332358,
      "grad_norm": 1.625,
      "learning_rate": 0.0009911911910263054,
      "loss": 1.4508,
      "step": 3501
    },
    {
      "epoch": 0.30129914824055753,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009911861368831682,
      "loss": 1.8331,
      "step": 3502
    },
    {
      "epoch": 0.3013851845478792,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009911810813034072,
      "loss": 0.8238,
      "step": 3503
    },
    {
      "epoch": 0.3014712208552009,
      "grad_norm": 1.359375,
      "learning_rate": 0.000991176024287037,
      "loss": 1.5464,
      "step": 3504
    },
    {
      "epoch": 0.3015572571625226,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009911709658340724,
      "loss": 1.5995,
      "step": 3505
    },
    {
      "epoch": 0.30164329346984425,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009911659059445284,
      "loss": 1.2796,
      "step": 3506
    },
    {
      "epoch": 0.301729329777166,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009911608446184195,
      "loss": 1.3206,
      "step": 3507
    },
    {
      "epoch": 0.30181536608448767,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009911557818557605,
      "loss": 1.4584,
      "step": 3508
    },
    {
      "epoch": 0.30190140239180935,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0009911507176565667,
      "loss": 1.8957,
      "step": 3509
    },
    {
      "epoch": 0.30198743869913103,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009911456520208524,
      "loss": 0.8811,
      "step": 3510
    },
    {
      "epoch": 0.3020734750064527,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009911405849486327,
      "loss": 1.2159,
      "step": 3511
    },
    {
      "epoch": 0.3021595113137744,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009911355164399223,
      "loss": 1.5526,
      "step": 3512
    },
    {
      "epoch": 0.3022455476210961,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009911304464947357,
      "loss": 2.0328,
      "step": 3513
    },
    {
      "epoch": 0.3023315839284178,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009911253751130884,
      "loss": 0.6514,
      "step": 3514
    },
    {
      "epoch": 0.3024176202357395,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009911203022949949,
      "loss": 1.6863,
      "step": 3515
    },
    {
      "epoch": 0.30250365654306116,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009911152280404698,
      "loss": 2.0244,
      "step": 3516
    },
    {
      "epoch": 0.30258969285038284,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009911101523495285,
      "loss": 1.7156,
      "step": 3517
    },
    {
      "epoch": 0.3026757291577046,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009911050752221852,
      "loss": 2.1052,
      "step": 3518
    },
    {
      "epoch": 0.30276176546502626,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009910999966584553,
      "loss": 1.7514,
      "step": 3519
    },
    {
      "epoch": 0.30284780177234794,
      "grad_norm": 1.125,
      "learning_rate": 0.0009910949166583535,
      "loss": 0.8487,
      "step": 3520
    },
    {
      "epoch": 0.3029338380796696,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009910898352218944,
      "loss": 2.0089,
      "step": 3521
    },
    {
      "epoch": 0.3030198743869913,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009910847523490931,
      "loss": 1.8827,
      "step": 3522
    },
    {
      "epoch": 0.303105910694313,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009910796680399642,
      "loss": 1.9284,
      "step": 3523
    },
    {
      "epoch": 0.3031919470016347,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009910745822945231,
      "loss": 1.1615,
      "step": 3524
    },
    {
      "epoch": 0.3032779833089564,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009910694951127841,
      "loss": 1.1887,
      "step": 3525
    },
    {
      "epoch": 0.30336401961627807,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009910644064947624,
      "loss": 1.642,
      "step": 3526
    },
    {
      "epoch": 0.30345005592359975,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000991059316440473,
      "loss": 1.3801,
      "step": 3527
    },
    {
      "epoch": 0.30353609223092143,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009910542249499305,
      "loss": 1.8547,
      "step": 3528
    },
    {
      "epoch": 0.30362212853824316,
      "grad_norm": 2.6875,
      "learning_rate": 0.00099104913202315,
      "loss": 1.8996,
      "step": 3529
    },
    {
      "epoch": 0.30370816484556484,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000991044037660146,
      "loss": 2.006,
      "step": 3530
    },
    {
      "epoch": 0.3037942011528865,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009910389418609337,
      "loss": 1.3673,
      "step": 3531
    },
    {
      "epoch": 0.3038802374602082,
      "grad_norm": 1.5,
      "learning_rate": 0.0009910338446255283,
      "loss": 1.4485,
      "step": 3532
    },
    {
      "epoch": 0.3039662737675299,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009910287459539441,
      "loss": 1.3116,
      "step": 3533
    },
    {
      "epoch": 0.30405231007485156,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009910236458461966,
      "loss": 1.0012,
      "step": 3534
    },
    {
      "epoch": 0.3041383463821733,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009910185443023003,
      "loss": 0.9261,
      "step": 3535
    },
    {
      "epoch": 0.304224382689495,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009910134413222702,
      "loss": 2.145,
      "step": 3536
    },
    {
      "epoch": 0.30431041899681666,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009910083369061215,
      "loss": 1.0154,
      "step": 3537
    },
    {
      "epoch": 0.30439645530413834,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009910032310538686,
      "loss": 1.6517,
      "step": 3538
    },
    {
      "epoch": 0.30448249161146,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000990998123765527,
      "loss": 1.5682,
      "step": 3539
    },
    {
      "epoch": 0.30456852791878175,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009909930150411113,
      "loss": 0.9142,
      "step": 3540
    },
    {
      "epoch": 0.30465456422610343,
      "grad_norm": 1.375,
      "learning_rate": 0.0009909879048806366,
      "loss": 1.1947,
      "step": 3541
    },
    {
      "epoch": 0.3047406005334251,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009909827932841175,
      "loss": 1.1122,
      "step": 3542
    },
    {
      "epoch": 0.3048266368407468,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009909776802515696,
      "loss": 1.0431,
      "step": 3543
    },
    {
      "epoch": 0.30491267314806847,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009909725657830072,
      "loss": 0.9736,
      "step": 3544
    },
    {
      "epoch": 0.30499870945539015,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009909674498784456,
      "loss": 1.6613,
      "step": 3545
    },
    {
      "epoch": 0.3050847457627119,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009909623325378995,
      "loss": 1.7614,
      "step": 3546
    },
    {
      "epoch": 0.30517078207003356,
      "grad_norm": 1.625,
      "learning_rate": 0.0009909572137613845,
      "loss": 2.4577,
      "step": 3547
    },
    {
      "epoch": 0.30525681837735524,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009909520935489147,
      "loss": 1.4562,
      "step": 3548
    },
    {
      "epoch": 0.3053428546846769,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009909469719005057,
      "loss": 1.1839,
      "step": 3549
    },
    {
      "epoch": 0.3054288909919986,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009909418488161725,
      "loss": 2.0876,
      "step": 3550
    },
    {
      "epoch": 0.30551492729932034,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009909367242959296,
      "loss": 1.2784,
      "step": 3551
    },
    {
      "epoch": 0.305600963606642,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009909315983397924,
      "loss": 0.8576,
      "step": 3552
    },
    {
      "epoch": 0.3056869999139637,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009909264709477756,
      "loss": 1.3597,
      "step": 3553
    },
    {
      "epoch": 0.3057730362212854,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009909213421198946,
      "loss": 1.7902,
      "step": 3554
    },
    {
      "epoch": 0.30585907252860706,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009909162118561638,
      "loss": 1.1157,
      "step": 3555
    },
    {
      "epoch": 0.30594510883592874,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009909110801565987,
      "loss": 1.6445,
      "step": 3556
    },
    {
      "epoch": 0.30603114514325047,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009909059470212143,
      "loss": 1.671,
      "step": 3557
    },
    {
      "epoch": 0.30611718145057215,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009909008124500255,
      "loss": 1.0737,
      "step": 3558
    },
    {
      "epoch": 0.30620321775789383,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009908956764430473,
      "loss": 1.9144,
      "step": 3559
    },
    {
      "epoch": 0.3062892540652155,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009908905390002946,
      "loss": 1.4692,
      "step": 3560
    },
    {
      "epoch": 0.3063752903725372,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009908854001217824,
      "loss": 1.9203,
      "step": 3561
    },
    {
      "epoch": 0.3064613266798589,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009908802598075262,
      "loss": 1.4946,
      "step": 3562
    },
    {
      "epoch": 0.3065473629871806,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009908751180575405,
      "loss": 1.5265,
      "step": 3563
    },
    {
      "epoch": 0.3066333992945023,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009908699748718407,
      "loss": 1.59,
      "step": 3564
    },
    {
      "epoch": 0.30671943560182396,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009908648302504417,
      "loss": 1.2173,
      "step": 3565
    },
    {
      "epoch": 0.30680547190914564,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009908596841933584,
      "loss": 1.3884,
      "step": 3566
    },
    {
      "epoch": 0.3068915082164673,
      "grad_norm": 1.09375,
      "learning_rate": 0.000990854536700606,
      "loss": 1.6648,
      "step": 3567
    },
    {
      "epoch": 0.30697754452378906,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009908493877721996,
      "loss": 1.7435,
      "step": 3568
    },
    {
      "epoch": 0.30706358083111074,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009908442374081542,
      "loss": 1.6405,
      "step": 3569
    },
    {
      "epoch": 0.3071496171384324,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009908390856084848,
      "loss": 1.3431,
      "step": 3570
    },
    {
      "epoch": 0.3072356534457541,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009908339323732064,
      "loss": 2.0324,
      "step": 3571
    },
    {
      "epoch": 0.3073216897530758,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009908287777023346,
      "loss": 0.9377,
      "step": 3572
    },
    {
      "epoch": 0.3074077260603975,
      "grad_norm": 1.265625,
      "learning_rate": 0.000990823621595884,
      "loss": 1.8573,
      "step": 3573
    },
    {
      "epoch": 0.3074937623677192,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009908184640538696,
      "loss": 1.5374,
      "step": 3574
    },
    {
      "epoch": 0.30757979867504087,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009908133050763066,
      "loss": 2.3009,
      "step": 3575
    },
    {
      "epoch": 0.30766583498236255,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009908081446632102,
      "loss": 1.7048,
      "step": 3576
    },
    {
      "epoch": 0.30775187128968423,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009908029828145956,
      "loss": 1.6355,
      "step": 3577
    },
    {
      "epoch": 0.3078379075970059,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009907978195304776,
      "loss": 1.5574,
      "step": 3578
    },
    {
      "epoch": 0.30792394390432765,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009907926548108714,
      "loss": 1.4536,
      "step": 3579
    },
    {
      "epoch": 0.3080099802116493,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009907874886557922,
      "loss": 0.8833,
      "step": 3580
    },
    {
      "epoch": 0.308096016518971,
      "grad_norm": 1.40625,
      "learning_rate": 0.000990782321065255,
      "loss": 1.5107,
      "step": 3581
    },
    {
      "epoch": 0.3081820528262927,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000990777152039275,
      "loss": 1.4404,
      "step": 3582
    },
    {
      "epoch": 0.30826808913361436,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009907719815778673,
      "loss": 1.356,
      "step": 3583
    },
    {
      "epoch": 0.3083541254409361,
      "grad_norm": 1.265625,
      "learning_rate": 0.000990766809681047,
      "loss": 1.4484,
      "step": 3584
    },
    {
      "epoch": 0.3084401617482578,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009907616363488293,
      "loss": 1.5562,
      "step": 3585
    },
    {
      "epoch": 0.30852619805557946,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009907564615812292,
      "loss": 1.0178,
      "step": 3586
    },
    {
      "epoch": 0.30861223436290114,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009907512853782618,
      "loss": 1.4079,
      "step": 3587
    },
    {
      "epoch": 0.3086982706702228,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009907461077399424,
      "loss": 2.2894,
      "step": 3588
    },
    {
      "epoch": 0.3087843069775445,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009907409286662862,
      "loss": 0.9033,
      "step": 3589
    },
    {
      "epoch": 0.30887034328486623,
      "grad_norm": 1.21875,
      "learning_rate": 0.000990735748157308,
      "loss": 1.4058,
      "step": 3590
    },
    {
      "epoch": 0.3089563795921879,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009907305662130233,
      "loss": 1.2853,
      "step": 3591
    },
    {
      "epoch": 0.3090424158995096,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009907253828334473,
      "loss": 1.278,
      "step": 3592
    },
    {
      "epoch": 0.3091284522068313,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009907201980185946,
      "loss": 2.1764,
      "step": 3593
    },
    {
      "epoch": 0.30921448851415295,
      "grad_norm": 1.421875,
      "learning_rate": 0.000990715011768481,
      "loss": 0.9864,
      "step": 3594
    },
    {
      "epoch": 0.3093005248214747,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009907098240831214,
      "loss": 1.6492,
      "step": 3595
    },
    {
      "epoch": 0.30938656112879637,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009907046349625311,
      "loss": 1.8336,
      "step": 3596
    },
    {
      "epoch": 0.30947259743611805,
      "grad_norm": 1.234375,
      "learning_rate": 0.000990699444406725,
      "loss": 0.7384,
      "step": 3597
    },
    {
      "epoch": 0.3095586337434397,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009906942524157186,
      "loss": 1.8264,
      "step": 3598
    },
    {
      "epoch": 0.3096446700507614,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000990689058989527,
      "loss": 2.0251,
      "step": 3599
    },
    {
      "epoch": 0.3097307063580831,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009906838641281653,
      "loss": 1.53,
      "step": 3600
    },
    {
      "epoch": 0.3098167426654048,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009906786678316485,
      "loss": 2.0281,
      "step": 3601
    },
    {
      "epoch": 0.3099027789727265,
      "grad_norm": 1.375,
      "learning_rate": 0.0009906734700999922,
      "loss": 1.4661,
      "step": 3602
    },
    {
      "epoch": 0.3099888152800482,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009906682709332115,
      "loss": 1.7706,
      "step": 3603
    },
    {
      "epoch": 0.31007485158736986,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009906630703313216,
      "loss": 1.6302,
      "step": 3604
    },
    {
      "epoch": 0.31016088789469154,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009906578682943373,
      "loss": 1.7639,
      "step": 3605
    },
    {
      "epoch": 0.3102469242020133,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009906526648222746,
      "loss": 1.9269,
      "step": 3606
    },
    {
      "epoch": 0.31033296050933495,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009906474599151477,
      "loss": 1.665,
      "step": 3607
    },
    {
      "epoch": 0.31041899681665663,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009906422535729729,
      "loss": 1.103,
      "step": 3608
    },
    {
      "epoch": 0.3105050331239783,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000990637045795765,
      "loss": 1.4065,
      "step": 3609
    },
    {
      "epoch": 0.3105910694313,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009906318365835388,
      "loss": 1.1798,
      "step": 3610
    },
    {
      "epoch": 0.3106771057386217,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00099062662593631,
      "loss": 1.0332,
      "step": 3611
    },
    {
      "epoch": 0.3107631420459434,
      "grad_norm": 1.4375,
      "learning_rate": 0.000990621413854094,
      "loss": 1.7261,
      "step": 3612
    },
    {
      "epoch": 0.3108491783532651,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009906162003369054,
      "loss": 1.0335,
      "step": 3613
    },
    {
      "epoch": 0.31093521466058677,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009906109853847602,
      "loss": 1.1616,
      "step": 3614
    },
    {
      "epoch": 0.31102125096790845,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009906057689976731,
      "loss": 1.5994,
      "step": 3615
    },
    {
      "epoch": 0.3111072872752301,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009906005511756595,
      "loss": 1.746,
      "step": 3616
    },
    {
      "epoch": 0.31119332358255186,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009905953319187348,
      "loss": 1.4625,
      "step": 3617
    },
    {
      "epoch": 0.31127935988987354,
      "grad_norm": 2.015625,
      "learning_rate": 0.000990590111226914,
      "loss": 2.1964,
      "step": 3618
    },
    {
      "epoch": 0.3113653961971952,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009905848891002128,
      "loss": 0.8838,
      "step": 3619
    },
    {
      "epoch": 0.3114514325045169,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000990579665538646,
      "loss": 1.3623,
      "step": 3620
    },
    {
      "epoch": 0.3115374688118386,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009905744405422291,
      "loss": 1.3732,
      "step": 3621
    },
    {
      "epoch": 0.31162350511916026,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009905692141109774,
      "loss": 1.032,
      "step": 3622
    },
    {
      "epoch": 0.311709541426482,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009905639862449063,
      "loss": 0.91,
      "step": 3623
    },
    {
      "epoch": 0.3117955777338037,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009905587569440308,
      "loss": 1.6211,
      "step": 3624
    },
    {
      "epoch": 0.31188161404112535,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009905535262083665,
      "loss": 0.9528,
      "step": 3625
    },
    {
      "epoch": 0.31196765034844703,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009905482940379285,
      "loss": 1.2726,
      "step": 3626
    },
    {
      "epoch": 0.3120536866557687,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000990543060432732,
      "loss": 1.1728,
      "step": 3627
    },
    {
      "epoch": 0.31213972296309045,
      "grad_norm": 2.0,
      "learning_rate": 0.0009905378253927926,
      "loss": 2.3525,
      "step": 3628
    },
    {
      "epoch": 0.31222575927041213,
      "grad_norm": 1.375,
      "learning_rate": 0.0009905325889181253,
      "loss": 1.6496,
      "step": 3629
    },
    {
      "epoch": 0.3123117955777338,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009905273510087457,
      "loss": 1.0256,
      "step": 3630
    },
    {
      "epoch": 0.3123978318850555,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000990522111664669,
      "loss": 1.2238,
      "step": 3631
    },
    {
      "epoch": 0.31248386819237717,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009905168708859106,
      "loss": 1.328,
      "step": 3632
    },
    {
      "epoch": 0.31256990449969885,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009905116286724855,
      "loss": 1.3193,
      "step": 3633
    },
    {
      "epoch": 0.3126559408070206,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009905063850244096,
      "loss": 0.9224,
      "step": 3634
    },
    {
      "epoch": 0.31274197711434226,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009905011399416977,
      "loss": 1.7495,
      "step": 3635
    },
    {
      "epoch": 0.31282801342166394,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009904958934243654,
      "loss": 1.1877,
      "step": 3636
    },
    {
      "epoch": 0.3129140497289856,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000990490645472428,
      "loss": 0.6153,
      "step": 3637
    },
    {
      "epoch": 0.3130000860363073,
      "grad_norm": 1.5625,
      "learning_rate": 0.000990485396085901,
      "loss": 0.9365,
      "step": 3638
    },
    {
      "epoch": 0.31308612234362904,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009904801452647995,
      "loss": 1.5197,
      "step": 3639
    },
    {
      "epoch": 0.3131721586509507,
      "grad_norm": 1.671875,
      "learning_rate": 0.000990474893009139,
      "loss": 1.486,
      "step": 3640
    },
    {
      "epoch": 0.3132581949582724,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009904696393189347,
      "loss": 1.4331,
      "step": 3641
    },
    {
      "epoch": 0.3133442312655941,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009904643841942022,
      "loss": 0.6344,
      "step": 3642
    },
    {
      "epoch": 0.31343026757291576,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009904591276349568,
      "loss": 1.193,
      "step": 3643
    },
    {
      "epoch": 0.31351630388023743,
      "grad_norm": 1.625,
      "learning_rate": 0.0009904538696412137,
      "loss": 1.5037,
      "step": 3644
    },
    {
      "epoch": 0.31360234018755917,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009904486102129885,
      "loss": 1.1512,
      "step": 3645
    },
    {
      "epoch": 0.31368837649488085,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009904433493502967,
      "loss": 1.667,
      "step": 3646
    },
    {
      "epoch": 0.31377441280220253,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009904380870531533,
      "loss": 1.0794,
      "step": 3647
    },
    {
      "epoch": 0.3138604491095242,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000990432823321574,
      "loss": 0.8442,
      "step": 3648
    },
    {
      "epoch": 0.3139464854168459,
      "grad_norm": 1.09375,
      "learning_rate": 0.000990427558155574,
      "loss": 1.5026,
      "step": 3649
    },
    {
      "epoch": 0.3140325217241676,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009904222915551686,
      "loss": 1.1466,
      "step": 3650
    },
    {
      "epoch": 0.3141185580314893,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009904170235203735,
      "loss": 1.1279,
      "step": 3651
    },
    {
      "epoch": 0.314204594338811,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009904117540512041,
      "loss": 1.0801,
      "step": 3652
    },
    {
      "epoch": 0.31429063064613266,
      "grad_norm": 1.5,
      "learning_rate": 0.0009904064831476756,
      "loss": 1.588,
      "step": 3653
    },
    {
      "epoch": 0.31437666695345434,
      "grad_norm": 1.375,
      "learning_rate": 0.0009904012108098036,
      "loss": 1.1269,
      "step": 3654
    },
    {
      "epoch": 0.314462703260776,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009903959370376031,
      "loss": 1.3551,
      "step": 3655
    },
    {
      "epoch": 0.31454873956809776,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009903906618310902,
      "loss": 1.169,
      "step": 3656
    },
    {
      "epoch": 0.31463477587541944,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00099038538519028,
      "loss": 1.7703,
      "step": 3657
    },
    {
      "epoch": 0.3147208121827411,
      "grad_norm": 1.8125,
      "learning_rate": 0.000990380107115188,
      "loss": 1.7878,
      "step": 3658
    },
    {
      "epoch": 0.3148068484900628,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009903748276058294,
      "loss": 1.2169,
      "step": 3659
    },
    {
      "epoch": 0.3148928847973845,
      "grad_norm": 2.8125,
      "learning_rate": 0.0009903695466622199,
      "loss": 1.7047,
      "step": 3660
    },
    {
      "epoch": 0.3149789211047062,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009903642642843746,
      "loss": 1.8233,
      "step": 3661
    },
    {
      "epoch": 0.3150649574120279,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009903589804723092,
      "loss": 1.0905,
      "step": 3662
    },
    {
      "epoch": 0.31515099371934957,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009903536952260395,
      "loss": 1.4956,
      "step": 3663
    },
    {
      "epoch": 0.31523703002667125,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009903484085455802,
      "loss": 0.7877,
      "step": 3664
    },
    {
      "epoch": 0.31532306633399293,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009903431204309474,
      "loss": 1.6566,
      "step": 3665
    },
    {
      "epoch": 0.3154091026413146,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009903378308821565,
      "loss": 1.1498,
      "step": 3666
    },
    {
      "epoch": 0.31549513894863634,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009903325398992226,
      "loss": 1.6956,
      "step": 3667
    },
    {
      "epoch": 0.315581175255958,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009903272474821613,
      "loss": 1.368,
      "step": 3668
    },
    {
      "epoch": 0.3156672115632797,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009903219536309884,
      "loss": 1.4104,
      "step": 3669
    },
    {
      "epoch": 0.3157532478706014,
      "grad_norm": 1.53125,
      "learning_rate": 0.000990316658345719,
      "loss": 1.3162,
      "step": 3670
    },
    {
      "epoch": 0.31583928417792306,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009903113616263688,
      "loss": 1.754,
      "step": 3671
    },
    {
      "epoch": 0.3159253204852448,
      "grad_norm": 1.296875,
      "learning_rate": 0.000990306063472953,
      "loss": 1.0788,
      "step": 3672
    },
    {
      "epoch": 0.3160113567925665,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009903007638854878,
      "loss": 0.8333,
      "step": 3673
    },
    {
      "epoch": 0.31609739309988816,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009902954628639878,
      "loss": 2.0286,
      "step": 3674
    },
    {
      "epoch": 0.31618342940720984,
      "grad_norm": 1.46875,
      "learning_rate": 0.000990290160408469,
      "loss": 1.5768,
      "step": 3675
    },
    {
      "epoch": 0.3162694657145315,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000990284856518947,
      "loss": 1.2515,
      "step": 3676
    },
    {
      "epoch": 0.3163555020218532,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000990279551195437,
      "loss": 0.7052,
      "step": 3677
    },
    {
      "epoch": 0.31644153832917493,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009902742444379549,
      "loss": 1.2449,
      "step": 3678
    },
    {
      "epoch": 0.3165275746364966,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009902689362465157,
      "loss": 0.8255,
      "step": 3679
    },
    {
      "epoch": 0.3166136109438183,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009902636266211353,
      "loss": 1.2622,
      "step": 3680
    },
    {
      "epoch": 0.31669964725113997,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009902583155618292,
      "loss": 2.1221,
      "step": 3681
    },
    {
      "epoch": 0.31678568355846165,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009902530030686128,
      "loss": 1.6388,
      "step": 3682
    },
    {
      "epoch": 0.3168717198657834,
      "grad_norm": 1.375,
      "learning_rate": 0.0009902476891415018,
      "loss": 1.1148,
      "step": 3683
    },
    {
      "epoch": 0.31695775617310507,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009902423737805114,
      "loss": 1.687,
      "step": 3684
    },
    {
      "epoch": 0.31704379248042674,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009902370569856579,
      "loss": 1.768,
      "step": 3685
    },
    {
      "epoch": 0.3171298287877484,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009902317387569558,
      "loss": 1.7437,
      "step": 3686
    },
    {
      "epoch": 0.3172158650950701,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009902264190944216,
      "loss": 1.7872,
      "step": 3687
    },
    {
      "epoch": 0.3173019014023918,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009902210979980703,
      "loss": 1.5396,
      "step": 3688
    },
    {
      "epoch": 0.3173879377097135,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009902157754679178,
      "loss": 0.9894,
      "step": 3689
    },
    {
      "epoch": 0.3174739740170352,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009902104515039794,
      "loss": 1.0422,
      "step": 3690
    },
    {
      "epoch": 0.3175600103243569,
      "grad_norm": 1.0,
      "learning_rate": 0.0009902051261062707,
      "loss": 1.1099,
      "step": 3691
    },
    {
      "epoch": 0.31764604663167856,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009901997992748073,
      "loss": 0.9083,
      "step": 3692
    },
    {
      "epoch": 0.31773208293900024,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000990194471009605,
      "loss": 1.5726,
      "step": 3693
    },
    {
      "epoch": 0.317818119246322,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009901891413106791,
      "loss": 1.3034,
      "step": 3694
    },
    {
      "epoch": 0.31790415555364365,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009901838101780455,
      "loss": 0.9508,
      "step": 3695
    },
    {
      "epoch": 0.31799019186096533,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009901784776117192,
      "loss": 1.4165,
      "step": 3696
    },
    {
      "epoch": 0.318076228168287,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009901731436117165,
      "loss": 1.1867,
      "step": 3697
    },
    {
      "epoch": 0.3181622644756087,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009901678081780526,
      "loss": 1.7543,
      "step": 3698
    },
    {
      "epoch": 0.31824830078293037,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009901624713107432,
      "loss": 1.2973,
      "step": 3699
    },
    {
      "epoch": 0.3183343370902521,
      "grad_norm": 1.25,
      "learning_rate": 0.0009901571330098039,
      "loss": 1.0664,
      "step": 3700
    },
    {
      "epoch": 0.3184203733975738,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009901517932752505,
      "loss": 1.233,
      "step": 3701
    },
    {
      "epoch": 0.31850640970489547,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000990146452107098,
      "loss": 1.8827,
      "step": 3702
    },
    {
      "epoch": 0.31859244601221715,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009901411095053628,
      "loss": 1.7664,
      "step": 3703
    },
    {
      "epoch": 0.3186784823195388,
      "grad_norm": 1.171875,
      "learning_rate": 0.00099013576547006,
      "loss": 1.4871,
      "step": 3704
    },
    {
      "epoch": 0.31876451862686056,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009901304200012056,
      "loss": 0.7835,
      "step": 3705
    },
    {
      "epoch": 0.31885055493418224,
      "grad_norm": 1.25,
      "learning_rate": 0.0009901250730988149,
      "loss": 1.3115,
      "step": 3706
    },
    {
      "epoch": 0.3189365912415039,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009901197247629036,
      "loss": 1.0663,
      "step": 3707
    },
    {
      "epoch": 0.3190226275488256,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009901143749934877,
      "loss": 1.6968,
      "step": 3708
    },
    {
      "epoch": 0.3191086638561473,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009901090237905822,
      "loss": 1.0239,
      "step": 3709
    },
    {
      "epoch": 0.31919470016346896,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009901036711542036,
      "loss": 1.292,
      "step": 3710
    },
    {
      "epoch": 0.3192807364707907,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009900983170843668,
      "loss": 0.8769,
      "step": 3711
    },
    {
      "epoch": 0.3193667727781124,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009900929615810877,
      "loss": 1.2222,
      "step": 3712
    },
    {
      "epoch": 0.31945280908543405,
      "grad_norm": 1.109375,
      "learning_rate": 0.000990087604644382,
      "loss": 1.306,
      "step": 3713
    },
    {
      "epoch": 0.31953884539275573,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009900822462742655,
      "loss": 1.6329,
      "step": 3714
    },
    {
      "epoch": 0.3196248817000774,
      "grad_norm": 2.015625,
      "learning_rate": 0.000990076886470754,
      "loss": 1.6858,
      "step": 3715
    },
    {
      "epoch": 0.31971091800739915,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009900715252338624,
      "loss": 1.5336,
      "step": 3716
    },
    {
      "epoch": 0.3197969543147208,
      "grad_norm": 1.7734375,
      "learning_rate": 0.000990066162563607,
      "loss": 1.2491,
      "step": 3717
    },
    {
      "epoch": 0.3198829906220425,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009900607984600037,
      "loss": 1.507,
      "step": 3718
    },
    {
      "epoch": 0.3199690269293642,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009900554329230677,
      "loss": 1.7328,
      "step": 3719
    },
    {
      "epoch": 0.32005506323668587,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000990050065952815,
      "loss": 1.3472,
      "step": 3720
    },
    {
      "epoch": 0.32014109954400755,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009900446975492608,
      "loss": 1.7616,
      "step": 3721
    },
    {
      "epoch": 0.3202271358513293,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009900393277124215,
      "loss": 1.1876,
      "step": 3722
    },
    {
      "epoch": 0.32031317215865096,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009900339564423123,
      "loss": 1.8279,
      "step": 3723
    },
    {
      "epoch": 0.32039920846597264,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000990028583738949,
      "loss": 0.9887,
      "step": 3724
    },
    {
      "epoch": 0.3204852447732943,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009900232096023477,
      "loss": 0.9967,
      "step": 3725
    },
    {
      "epoch": 0.320571281080616,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009900178340325238,
      "loss": 1.6829,
      "step": 3726
    },
    {
      "epoch": 0.32065731738793773,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009900124570294928,
      "loss": 1.07,
      "step": 3727
    },
    {
      "epoch": 0.3207433536952594,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009900070785932706,
      "loss": 1.7474,
      "step": 3728
    },
    {
      "epoch": 0.3208293900025811,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009900016987238734,
      "loss": 0.9313,
      "step": 3729
    },
    {
      "epoch": 0.3209154263099028,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009899963174213162,
      "loss": 1.0954,
      "step": 3730
    },
    {
      "epoch": 0.32100146261722445,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009899909346856152,
      "loss": 1.1631,
      "step": 3731
    },
    {
      "epoch": 0.32108749892454613,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009899855505167859,
      "loss": 1.4632,
      "step": 3732
    },
    {
      "epoch": 0.32117353523186787,
      "grad_norm": 1.25,
      "learning_rate": 0.0009899801649148443,
      "loss": 1.8736,
      "step": 3733
    },
    {
      "epoch": 0.32125957153918955,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000989974777879806,
      "loss": 1.0336,
      "step": 3734
    },
    {
      "epoch": 0.3213456078465112,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009899693894116868,
      "loss": 1.4498,
      "step": 3735
    },
    {
      "epoch": 0.3214316441538329,
      "grad_norm": 1.359375,
      "learning_rate": 0.000989963999510502,
      "loss": 1.0565,
      "step": 3736
    },
    {
      "epoch": 0.3215176804611546,
      "grad_norm": 1.375,
      "learning_rate": 0.0009899586081762684,
      "loss": 0.7916,
      "step": 3737
    },
    {
      "epoch": 0.3216037167684763,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009899532154090008,
      "loss": 1.3017,
      "step": 3738
    },
    {
      "epoch": 0.321689753075798,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009899478212087155,
      "loss": 2.0268,
      "step": 3739
    },
    {
      "epoch": 0.3217757893831197,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009899424255754281,
      "loss": 1.3022,
      "step": 3740
    },
    {
      "epoch": 0.32186182569044136,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009899370285091544,
      "loss": 0.8844,
      "step": 3741
    },
    {
      "epoch": 0.32194786199776304,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00098993163000991,
      "loss": 1.8675,
      "step": 3742
    },
    {
      "epoch": 0.3220338983050847,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000989926230077711,
      "loss": 1.1658,
      "step": 3743
    },
    {
      "epoch": 0.32211993461240646,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009899208287125732,
      "loss": 1.7597,
      "step": 3744
    },
    {
      "epoch": 0.32220597091972814,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000989915425914512,
      "loss": 1.5257,
      "step": 3745
    },
    {
      "epoch": 0.3222920072270498,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009899100216835437,
      "loss": 1.4385,
      "step": 3746
    },
    {
      "epoch": 0.3223780435343715,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009899046160196836,
      "loss": 0.5842,
      "step": 3747
    },
    {
      "epoch": 0.3224640798416932,
      "grad_norm": 1.6875,
      "learning_rate": 0.000989899208922948,
      "loss": 1.8182,
      "step": 3748
    },
    {
      "epoch": 0.3225501161490149,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009898938003933524,
      "loss": 1.4598,
      "step": 3749
    },
    {
      "epoch": 0.3226361524563366,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009898883904309128,
      "loss": 1.543,
      "step": 3750
    },
    {
      "epoch": 0.32272218876365827,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000989882979035645,
      "loss": 0.8671,
      "step": 3751
    },
    {
      "epoch": 0.32280822507097995,
      "grad_norm": 1.375,
      "learning_rate": 0.0009898775662075645,
      "loss": 2.1024,
      "step": 3752
    },
    {
      "epoch": 0.32289426137830163,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009898721519466876,
      "loss": 0.9188,
      "step": 3753
    },
    {
      "epoch": 0.3229802976856233,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009898667362530297,
      "loss": 2.2225,
      "step": 3754
    },
    {
      "epoch": 0.32306633399294504,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009898613191266072,
      "loss": 1.6515,
      "step": 3755
    },
    {
      "epoch": 0.3231523703002667,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009898559005674355,
      "loss": 1.4777,
      "step": 3756
    },
    {
      "epoch": 0.3232384066075884,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009898504805755303,
      "loss": 1.0761,
      "step": 3757
    },
    {
      "epoch": 0.3233244429149101,
      "grad_norm": 1.046875,
      "learning_rate": 0.000989845059150908,
      "loss": 0.8476,
      "step": 3758
    },
    {
      "epoch": 0.32341047922223176,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009898396362935842,
      "loss": 1.0156,
      "step": 3759
    },
    {
      "epoch": 0.3234965155295535,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009898342120035746,
      "loss": 2.0651,
      "step": 3760
    },
    {
      "epoch": 0.3235825518368752,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009898287862808953,
      "loss": 1.4154,
      "step": 3761
    },
    {
      "epoch": 0.32366858814419686,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009898233591255619,
      "loss": 0.9674,
      "step": 3762
    },
    {
      "epoch": 0.32375462445151854,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009898179305375906,
      "loss": 1.444,
      "step": 3763
    },
    {
      "epoch": 0.3238406607588402,
      "grad_norm": 1.7421875,
      "learning_rate": 0.000989812500516997,
      "loss": 2.0492,
      "step": 3764
    },
    {
      "epoch": 0.3239266970661619,
      "grad_norm": 1.296875,
      "learning_rate": 0.000989807069063797,
      "loss": 1.2183,
      "step": 3765
    },
    {
      "epoch": 0.32401273337348363,
      "grad_norm": 1.296875,
      "learning_rate": 0.000989801636178007,
      "loss": 1.144,
      "step": 3766
    },
    {
      "epoch": 0.3240987696808053,
      "grad_norm": 1.5,
      "learning_rate": 0.000989796201859642,
      "loss": 0.7653,
      "step": 3767
    },
    {
      "epoch": 0.324184805988127,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009897907661087187,
      "loss": 1.1219,
      "step": 3768
    },
    {
      "epoch": 0.32427084229544867,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009897853289252525,
      "loss": 1.4485,
      "step": 3769
    },
    {
      "epoch": 0.32435687860277035,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009897798903092596,
      "loss": 1.4004,
      "step": 3770
    },
    {
      "epoch": 0.3244429149100921,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009897744502607557,
      "loss": 0.9486,
      "step": 3771
    },
    {
      "epoch": 0.32452895121741376,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009897690087797568,
      "loss": 1.1747,
      "step": 3772
    },
    {
      "epoch": 0.32461498752473544,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009897635658662786,
      "loss": 1.2675,
      "step": 3773
    },
    {
      "epoch": 0.3247010238320571,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009897581215203375,
      "loss": 1.2488,
      "step": 3774
    },
    {
      "epoch": 0.3247870601393788,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009897526757419492,
      "loss": 1.5476,
      "step": 3775
    },
    {
      "epoch": 0.3248730964467005,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009897472285311294,
      "loss": 1.9807,
      "step": 3776
    },
    {
      "epoch": 0.3249591327540222,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009897417798878942,
      "loss": 0.8513,
      "step": 3777
    },
    {
      "epoch": 0.3250451690613439,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009897363298122595,
      "loss": 2.4066,
      "step": 3778
    },
    {
      "epoch": 0.3251312053686656,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009897308783042415,
      "loss": 2.0862,
      "step": 3779
    },
    {
      "epoch": 0.32521724167598726,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009897254253638556,
      "loss": 1.3716,
      "step": 3780
    },
    {
      "epoch": 0.32530327798330894,
      "grad_norm": 2.0,
      "learning_rate": 0.0009897199709911185,
      "loss": 1.873,
      "step": 3781
    },
    {
      "epoch": 0.32538931429063067,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009897145151860453,
      "loss": 1.275,
      "step": 3782
    },
    {
      "epoch": 0.32547535059795235,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009897090579486527,
      "loss": 1.1683,
      "step": 3783
    },
    {
      "epoch": 0.32556138690527403,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009897035992789563,
      "loss": 1.1571,
      "step": 3784
    },
    {
      "epoch": 0.3256474232125957,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000989698139176972,
      "loss": 0.7434,
      "step": 3785
    },
    {
      "epoch": 0.3257334595199174,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000989692677642716,
      "loss": 1.4352,
      "step": 3786
    },
    {
      "epoch": 0.32581949582723907,
      "grad_norm": 1.34375,
      "learning_rate": 0.000989687214676204,
      "loss": 1.7449,
      "step": 3787
    },
    {
      "epoch": 0.3259055321345608,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009896817502774524,
      "loss": 1.5826,
      "step": 3788
    },
    {
      "epoch": 0.3259915684418825,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009896762844464767,
      "loss": 0.9393,
      "step": 3789
    },
    {
      "epoch": 0.32607760474920416,
      "grad_norm": 1.21875,
      "learning_rate": 0.000989670817183293,
      "loss": 1.3271,
      "step": 3790
    },
    {
      "epoch": 0.32616364105652584,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009896653484879176,
      "loss": 1.4582,
      "step": 3791
    },
    {
      "epoch": 0.3262496773638475,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009896598783603664,
      "loss": 1.4952,
      "step": 3792
    },
    {
      "epoch": 0.32633571367116926,
      "grad_norm": 1.390625,
      "learning_rate": 0.000989654406800655,
      "loss": 1.7136,
      "step": 3793
    },
    {
      "epoch": 0.32642174997849094,
      "grad_norm": 1.625,
      "learning_rate": 0.0009896489338088,
      "loss": 2.1398,
      "step": 3794
    },
    {
      "epoch": 0.3265077862858126,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009896434593848168,
      "loss": 1.456,
      "step": 3795
    },
    {
      "epoch": 0.3265938225931343,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009896379835287218,
      "loss": 1.1497,
      "step": 3796
    },
    {
      "epoch": 0.326679858900456,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000989632506240531,
      "loss": 1.0272,
      "step": 3797
    },
    {
      "epoch": 0.32676589520777766,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009896270275202602,
      "loss": 1.413,
      "step": 3798
    },
    {
      "epoch": 0.3268519315150994,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009896215473679257,
      "loss": 1.4676,
      "step": 3799
    },
    {
      "epoch": 0.32693796782242107,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009896160657835436,
      "loss": 0.8043,
      "step": 3800
    },
    {
      "epoch": 0.32702400412974275,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009896105827671294,
      "loss": 0.9989,
      "step": 3801
    },
    {
      "epoch": 0.32711004043706443,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009896050983186997,
      "loss": 1.0957,
      "step": 3802
    },
    {
      "epoch": 0.3271960767443861,
      "grad_norm": 1.5,
      "learning_rate": 0.00098959961243827,
      "loss": 2.5054,
      "step": 3803
    },
    {
      "epoch": 0.32728211305170785,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009895941251258568,
      "loss": 1.845,
      "step": 3804
    },
    {
      "epoch": 0.3273681493590295,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009895886363814762,
      "loss": 1.6159,
      "step": 3805
    },
    {
      "epoch": 0.3274541856663512,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000989583146205144,
      "loss": 0.9473,
      "step": 3806
    },
    {
      "epoch": 0.3275402219736729,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009895776545968762,
      "loss": 1.499,
      "step": 3807
    },
    {
      "epoch": 0.32762625828099456,
      "grad_norm": 1.296875,
      "learning_rate": 0.000989572161556689,
      "loss": 1.1792,
      "step": 3808
    },
    {
      "epoch": 0.32771229458831624,
      "grad_norm": 1.25,
      "learning_rate": 0.0009895666670845984,
      "loss": 1.5913,
      "step": 3809
    },
    {
      "epoch": 0.327798330895638,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009895611711806206,
      "loss": 1.0315,
      "step": 3810
    },
    {
      "epoch": 0.32788436720295966,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009895556738447715,
      "loss": 1.3554,
      "step": 3811
    },
    {
      "epoch": 0.32797040351028134,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009895501750770673,
      "loss": 1.8103,
      "step": 3812
    },
    {
      "epoch": 0.328056439817603,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000989544674877524,
      "loss": 1.1167,
      "step": 3813
    },
    {
      "epoch": 0.3281424761249247,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009895391732461579,
      "loss": 1.6173,
      "step": 3814
    },
    {
      "epoch": 0.32822851243224643,
      "grad_norm": 1.234375,
      "learning_rate": 0.000989533670182985,
      "loss": 1.6993,
      "step": 3815
    },
    {
      "epoch": 0.3283145487395681,
      "grad_norm": 1.25,
      "learning_rate": 0.000989528165688021,
      "loss": 1.4643,
      "step": 3816
    },
    {
      "epoch": 0.3284005850468898,
      "grad_norm": 1.75,
      "learning_rate": 0.0009895226597612826,
      "loss": 1.5621,
      "step": 3817
    },
    {
      "epoch": 0.32848662135421147,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009895171524027856,
      "loss": 1.8621,
      "step": 3818
    },
    {
      "epoch": 0.32857265766153315,
      "grad_norm": 1.28125,
      "learning_rate": 0.000989511643612546,
      "loss": 1.7088,
      "step": 3819
    },
    {
      "epoch": 0.32865869396885483,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009895061333905799,
      "loss": 1.0287,
      "step": 3820
    },
    {
      "epoch": 0.32874473027617657,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009895006217369039,
      "loss": 0.983,
      "step": 3821
    },
    {
      "epoch": 0.32883076658349825,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009894951086515335,
      "loss": 2.1854,
      "step": 3822
    },
    {
      "epoch": 0.3289168028908199,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009894895941344854,
      "loss": 1.6521,
      "step": 3823
    },
    {
      "epoch": 0.3290028391981416,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009894840781857752,
      "loss": 1.7057,
      "step": 3824
    },
    {
      "epoch": 0.3290888755054633,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009894785608054193,
      "loss": 1.1442,
      "step": 3825
    },
    {
      "epoch": 0.329174911812785,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000989473041993434,
      "loss": 2.0609,
      "step": 3826
    },
    {
      "epoch": 0.3292609481201067,
      "grad_norm": 1.9453125,
      "learning_rate": 0.000989467521749835,
      "loss": 1.6919,
      "step": 3827
    },
    {
      "epoch": 0.3293469844274284,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009894620000746388,
      "loss": 0.7763,
      "step": 3828
    },
    {
      "epoch": 0.32943302073475006,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009894564769678616,
      "loss": 1.2448,
      "step": 3829
    },
    {
      "epoch": 0.32951905704207174,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009894509524295192,
      "loss": 1.5118,
      "step": 3830
    },
    {
      "epoch": 0.3296050933493934,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000989445426459628,
      "loss": 1.3815,
      "step": 3831
    },
    {
      "epoch": 0.32969112965671515,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009894398990582041,
      "loss": 1.2651,
      "step": 3832
    },
    {
      "epoch": 0.32977716596403683,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009894343702252638,
      "loss": 1.6496,
      "step": 3833
    },
    {
      "epoch": 0.3298632022713585,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009894288399608232,
      "loss": 1.133,
      "step": 3834
    },
    {
      "epoch": 0.3299492385786802,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009894233082648983,
      "loss": 1.2316,
      "step": 3835
    },
    {
      "epoch": 0.3300352748860019,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009894177751375052,
      "loss": 1.9014,
      "step": 3836
    },
    {
      "epoch": 0.3301213111933236,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009894122405786606,
      "loss": 0.9352,
      "step": 3837
    },
    {
      "epoch": 0.3302073475006453,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009894067045883802,
      "loss": 1.5707,
      "step": 3838
    },
    {
      "epoch": 0.33029338380796697,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009894011671666805,
      "loss": 1.9225,
      "step": 3839
    },
    {
      "epoch": 0.33037942011528865,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009893956283135777,
      "loss": 0.9366,
      "step": 3840
    },
    {
      "epoch": 0.3304654564226103,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009893900880290876,
      "loss": 1.1725,
      "step": 3841
    },
    {
      "epoch": 0.330551492729932,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009893845463132267,
      "loss": 0.991,
      "step": 3842
    },
    {
      "epoch": 0.33063752903725374,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009893790031660112,
      "loss": 1.517,
      "step": 3843
    },
    {
      "epoch": 0.3307235653445754,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009893734585874575,
      "loss": 1.4081,
      "step": 3844
    },
    {
      "epoch": 0.3308096016518971,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009893679125775813,
      "loss": 2.0949,
      "step": 3845
    },
    {
      "epoch": 0.3308956379592188,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009893623651363993,
      "loss": 1.2481,
      "step": 3846
    },
    {
      "epoch": 0.33098167426654046,
      "grad_norm": 1.375,
      "learning_rate": 0.0009893568162639278,
      "loss": 1.8383,
      "step": 3847
    },
    {
      "epoch": 0.3310677105738622,
      "grad_norm": 1.5,
      "learning_rate": 0.0009893512659601823,
      "loss": 1.0294,
      "step": 3848
    },
    {
      "epoch": 0.3311537468811839,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009893457142251798,
      "loss": 1.5102,
      "step": 3849
    },
    {
      "epoch": 0.33123978318850555,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009893401610589364,
      "loss": 1.1066,
      "step": 3850
    },
    {
      "epoch": 0.33132581949582723,
      "grad_norm": 1.40625,
      "learning_rate": 0.000989334606461468,
      "loss": 0.949,
      "step": 3851
    },
    {
      "epoch": 0.3314118558031489,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000989329050432791,
      "loss": 1.4258,
      "step": 3852
    },
    {
      "epoch": 0.3314978921104706,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009893234929729219,
      "loss": 1.0742,
      "step": 3853
    },
    {
      "epoch": 0.33158392841779233,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009893179340818765,
      "loss": 1.1702,
      "step": 3854
    },
    {
      "epoch": 0.331669964725114,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009893123737596714,
      "loss": 1.1564,
      "step": 3855
    },
    {
      "epoch": 0.3317560010324357,
      "grad_norm": 1.6875,
      "learning_rate": 0.000989306812006323,
      "loss": 1.9846,
      "step": 3856
    },
    {
      "epoch": 0.33184203733975737,
      "grad_norm": 1.203125,
      "learning_rate": 0.000989301248821847,
      "loss": 0.8548,
      "step": 3857
    },
    {
      "epoch": 0.33192807364707905,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00098929568420626,
      "loss": 1.4492,
      "step": 3858
    },
    {
      "epoch": 0.3320141099544008,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009892901181595786,
      "loss": 1.3397,
      "step": 3859
    },
    {
      "epoch": 0.33210014626172246,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009892845506818187,
      "loss": 1.313,
      "step": 3860
    },
    {
      "epoch": 0.33218618256904414,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009892789817729966,
      "loss": 2.1148,
      "step": 3861
    },
    {
      "epoch": 0.3322722188763658,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009892734114331284,
      "loss": 1.4326,
      "step": 3862
    },
    {
      "epoch": 0.3323582551836875,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009892678396622309,
      "loss": 0.7987,
      "step": 3863
    },
    {
      "epoch": 0.3324442914910092,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00098926226646032,
      "loss": 0.9862,
      "step": 3864
    },
    {
      "epoch": 0.3325303277983309,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009892566918274122,
      "loss": 2.3054,
      "step": 3865
    },
    {
      "epoch": 0.3326163641056526,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009892511157635238,
      "loss": 1.2944,
      "step": 3866
    },
    {
      "epoch": 0.3327024004129743,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009892455382686709,
      "loss": 1.7281,
      "step": 3867
    },
    {
      "epoch": 0.33278843672029595,
      "grad_norm": 1.578125,
      "learning_rate": 0.00098923995934287,
      "loss": 1.1862,
      "step": 3868
    },
    {
      "epoch": 0.33287447302761763,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009892343789861373,
      "loss": 1.8496,
      "step": 3869
    },
    {
      "epoch": 0.33296050933493937,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009892287971984893,
      "loss": 1.1463,
      "step": 3870
    },
    {
      "epoch": 0.33304654564226105,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009892232139799423,
      "loss": 1.6112,
      "step": 3871
    },
    {
      "epoch": 0.33313258194958273,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009892176293305123,
      "loss": 2.0678,
      "step": 3872
    },
    {
      "epoch": 0.3332186182569044,
      "grad_norm": 1.3125,
      "learning_rate": 0.000989212043250216,
      "loss": 0.9307,
      "step": 3873
    },
    {
      "epoch": 0.3333046545642261,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009892064557390696,
      "loss": 1.6124,
      "step": 3874
    },
    {
      "epoch": 0.33339069087154777,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009892008667970896,
      "loss": 1.4076,
      "step": 3875
    },
    {
      "epoch": 0.3334767271788695,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009891952764242919,
      "loss": 1.7122,
      "step": 3876
    },
    {
      "epoch": 0.3335627634861912,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009891896846206936,
      "loss": 1.133,
      "step": 3877
    },
    {
      "epoch": 0.33364879979351286,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009891840913863104,
      "loss": 1.2208,
      "step": 3878
    },
    {
      "epoch": 0.33373483610083454,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009891784967211588,
      "loss": 1.3513,
      "step": 3879
    },
    {
      "epoch": 0.3338208724081562,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009891729006252553,
      "loss": 1.3621,
      "step": 3880
    },
    {
      "epoch": 0.33390690871547796,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000989167303098616,
      "loss": 1.6634,
      "step": 3881
    },
    {
      "epoch": 0.33399294502279964,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009891617041412578,
      "loss": 1.7295,
      "step": 3882
    },
    {
      "epoch": 0.3340789813301213,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009891561037531967,
      "loss": 1.4508,
      "step": 3883
    },
    {
      "epoch": 0.334165017637443,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000989150501934449,
      "loss": 0.9958,
      "step": 3884
    },
    {
      "epoch": 0.3342510539447647,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009891448986850313,
      "loss": 1.671,
      "step": 3885
    },
    {
      "epoch": 0.33433709025208636,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00098913929400496,
      "loss": 1.4894,
      "step": 3886
    },
    {
      "epoch": 0.3344231265594081,
      "grad_norm": 1.578125,
      "learning_rate": 0.000989133687894251,
      "loss": 1.9682,
      "step": 3887
    },
    {
      "epoch": 0.33450916286672977,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009891280803529215,
      "loss": 1.038,
      "step": 3888
    },
    {
      "epoch": 0.33459519917405145,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009891224713809872,
      "loss": 2.1794,
      "step": 3889
    },
    {
      "epoch": 0.33468123548137313,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009891168609784651,
      "loss": 2.3021,
      "step": 3890
    },
    {
      "epoch": 0.3347672717886948,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009891112491453712,
      "loss": 1.4529,
      "step": 3891
    },
    {
      "epoch": 0.33485330809601654,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000989105635881722,
      "loss": 0.902,
      "step": 3892
    },
    {
      "epoch": 0.3349393444033382,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009891000211875339,
      "loss": 1.437,
      "step": 3893
    },
    {
      "epoch": 0.3350253807106599,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009890944050628234,
      "loss": 1.1962,
      "step": 3894
    },
    {
      "epoch": 0.3351114170179816,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009890887875076068,
      "loss": 1.1896,
      "step": 3895
    },
    {
      "epoch": 0.33519745332530326,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009890831685219008,
      "loss": 1.15,
      "step": 3896
    },
    {
      "epoch": 0.33528348963262494,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009890775481057215,
      "loss": 1.6682,
      "step": 3897
    },
    {
      "epoch": 0.3353695259399467,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009890719262590855,
      "loss": 1.3344,
      "step": 3898
    },
    {
      "epoch": 0.33545556224726836,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009890663029820093,
      "loss": 1.3102,
      "step": 3899
    },
    {
      "epoch": 0.33554159855459004,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009890606782745092,
      "loss": 1.7508,
      "step": 3900
    },
    {
      "epoch": 0.3356276348619117,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009890550521366019,
      "loss": 1.8602,
      "step": 3901
    },
    {
      "epoch": 0.3357136711692334,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009890494245683036,
      "loss": 1.4077,
      "step": 3902
    },
    {
      "epoch": 0.33579970747655513,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009890437955696308,
      "loss": 1.8216,
      "step": 3903
    },
    {
      "epoch": 0.3358857437838768,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009890381651406,
      "loss": 1.4797,
      "step": 3904
    },
    {
      "epoch": 0.3359717800911985,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009890325332812277,
      "loss": 1.7206,
      "step": 3905
    },
    {
      "epoch": 0.33605781639852017,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009890268999915306,
      "loss": 1.1297,
      "step": 3906
    },
    {
      "epoch": 0.33614385270584185,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009890212652715247,
      "loss": 1.3285,
      "step": 3907
    },
    {
      "epoch": 0.33622988901316353,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009890156291212267,
      "loss": 2.0603,
      "step": 3908
    },
    {
      "epoch": 0.33631592532048527,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009890099915406529,
      "loss": 1.1443,
      "step": 3909
    },
    {
      "epoch": 0.33640196162780694,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009890043525298204,
      "loss": 2.2036,
      "step": 3910
    },
    {
      "epoch": 0.3364879979351286,
      "grad_norm": 1.34375,
      "learning_rate": 0.000988998712088745,
      "loss": 1.1751,
      "step": 3911
    },
    {
      "epoch": 0.3365740342424503,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009889930702174435,
      "loss": 1.0465,
      "step": 3912
    },
    {
      "epoch": 0.336660070549772,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009889874269159325,
      "loss": 1.4615,
      "step": 3913
    },
    {
      "epoch": 0.3367461068570937,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000988981782184228,
      "loss": 1.1785,
      "step": 3914
    },
    {
      "epoch": 0.3368321431644154,
      "grad_norm": 1.125,
      "learning_rate": 0.0009889761360223474,
      "loss": 1.0053,
      "step": 3915
    },
    {
      "epoch": 0.3369181794717371,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009889704884303067,
      "loss": 0.9519,
      "step": 3916
    },
    {
      "epoch": 0.33700421577905876,
      "grad_norm": 1.28125,
      "learning_rate": 0.000988964839408122,
      "loss": 1.3653,
      "step": 3917
    },
    {
      "epoch": 0.33709025208638044,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009889591889558104,
      "loss": 1.2129,
      "step": 3918
    },
    {
      "epoch": 0.3371762883937021,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009889535370733885,
      "loss": 1.2445,
      "step": 3919
    },
    {
      "epoch": 0.33726232470102385,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009889478837608722,
      "loss": 1.5117,
      "step": 3920
    },
    {
      "epoch": 0.33734836100834553,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009889422290182788,
      "loss": 1.6275,
      "step": 3921
    },
    {
      "epoch": 0.3374343973156672,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009889365728456244,
      "loss": 0.8234,
      "step": 3922
    },
    {
      "epoch": 0.3375204336229889,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009889309152429253,
      "loss": 1.7035,
      "step": 3923
    },
    {
      "epoch": 0.33760646993031057,
      "grad_norm": 1.5,
      "learning_rate": 0.0009889252562101988,
      "loss": 1.876,
      "step": 3924
    },
    {
      "epoch": 0.3376925062376323,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000988919595747461,
      "loss": 0.8304,
      "step": 3925
    },
    {
      "epoch": 0.337778542544954,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000988913933854728,
      "loss": 1.7563,
      "step": 3926
    },
    {
      "epoch": 0.33786457885227567,
      "grad_norm": 1.140625,
      "learning_rate": 0.000988908270532017,
      "loss": 1.6289,
      "step": 3927
    },
    {
      "epoch": 0.33795061515959735,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009889026057793447,
      "loss": 1.6324,
      "step": 3928
    },
    {
      "epoch": 0.338036651466919,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009888969395967272,
      "loss": 2.2199,
      "step": 3929
    },
    {
      "epoch": 0.3381226877742407,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009888912719841813,
      "loss": 1.8308,
      "step": 3930
    },
    {
      "epoch": 0.33820872408156244,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009888856029417234,
      "loss": 1.3876,
      "step": 3931
    },
    {
      "epoch": 0.3382947603888841,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009888799324693702,
      "loss": 2.2111,
      "step": 3932
    },
    {
      "epoch": 0.3383807966962058,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009888742605671384,
      "loss": 1.6447,
      "step": 3933
    },
    {
      "epoch": 0.3384668330035275,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009888685872350442,
      "loss": 1.6051,
      "step": 3934
    },
    {
      "epoch": 0.33855286931084916,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009888629124731047,
      "loss": 1.8278,
      "step": 3935
    },
    {
      "epoch": 0.3386389056181709,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009888572362813362,
      "loss": 1.2684,
      "step": 3936
    },
    {
      "epoch": 0.3387249419254926,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009888515586597554,
      "loss": 1.8206,
      "step": 3937
    },
    {
      "epoch": 0.33881097823281425,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009888458796083789,
      "loss": 2.0099,
      "step": 3938
    },
    {
      "epoch": 0.33889701454013593,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009888401991272231,
      "loss": 1.2596,
      "step": 3939
    },
    {
      "epoch": 0.3389830508474576,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000988834517216305,
      "loss": 1.0886,
      "step": 3940
    },
    {
      "epoch": 0.3390690871547793,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009888288338756408,
      "loss": 1.7377,
      "step": 3941
    },
    {
      "epoch": 0.339155123462101,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009888231491052475,
      "loss": 1.0131,
      "step": 3942
    },
    {
      "epoch": 0.3392411597694227,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009888174629051414,
      "loss": 1.0221,
      "step": 3943
    },
    {
      "epoch": 0.3393271960767444,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009888117752753394,
      "loss": 1.5075,
      "step": 3944
    },
    {
      "epoch": 0.33941323238406607,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000988806086215858,
      "loss": 1.2525,
      "step": 3945
    },
    {
      "epoch": 0.33949926869138775,
      "grad_norm": 1.046875,
      "learning_rate": 0.000988800395726714,
      "loss": 0.8033,
      "step": 3946
    },
    {
      "epoch": 0.3395853049987095,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009887947038079237,
      "loss": 1.6095,
      "step": 3947
    },
    {
      "epoch": 0.33967134130603116,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000988789010459504,
      "loss": 1.439,
      "step": 3948
    },
    {
      "epoch": 0.33975737761335284,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009887833156814717,
      "loss": 0.7166,
      "step": 3949
    },
    {
      "epoch": 0.3398434139206745,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009887776194738432,
      "loss": 0.7241,
      "step": 3950
    },
    {
      "epoch": 0.3399294502279962,
      "grad_norm": 1.140625,
      "learning_rate": 0.000988771921836635,
      "loss": 0.9918,
      "step": 3951
    },
    {
      "epoch": 0.3400154865353179,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009887662227698643,
      "loss": 2.376,
      "step": 3952
    },
    {
      "epoch": 0.3401015228426396,
      "grad_norm": 1.46875,
      "learning_rate": 0.000988760522273547,
      "loss": 1.9888,
      "step": 3953
    },
    {
      "epoch": 0.3401875591499613,
      "grad_norm": 1.53125,
      "learning_rate": 0.000988754820347701,
      "loss": 1.1223,
      "step": 3954
    },
    {
      "epoch": 0.340273595457283,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009887491169923415,
      "loss": 0.5346,
      "step": 3955
    },
    {
      "epoch": 0.34035963176460465,
      "grad_norm": 1.421875,
      "learning_rate": 0.000988743412207486,
      "loss": 1.3387,
      "step": 3956
    },
    {
      "epoch": 0.34044566807192633,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009887377059931513,
      "loss": 1.2715,
      "step": 3957
    },
    {
      "epoch": 0.34053170437924807,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009887319983493538,
      "loss": 1.1751,
      "step": 3958
    },
    {
      "epoch": 0.34061774068656975,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009887262892761101,
      "loss": 1.7036,
      "step": 3959
    },
    {
      "epoch": 0.3407037769938914,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009887205787734372,
      "loss": 1.1893,
      "step": 3960
    },
    {
      "epoch": 0.3407898133012131,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009887148668413516,
      "loss": 0.9768,
      "step": 3961
    },
    {
      "epoch": 0.3408758496085348,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009887091534798702,
      "loss": 1.3049,
      "step": 3962
    },
    {
      "epoch": 0.34096188591585647,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009887034386890094,
      "loss": 1.1061,
      "step": 3963
    },
    {
      "epoch": 0.3410479222231782,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009886977224687864,
      "loss": 2.0879,
      "step": 3964
    },
    {
      "epoch": 0.3411339585304999,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000988692004819217,
      "loss": 1.1752,
      "step": 3965
    },
    {
      "epoch": 0.34121999483782156,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009886862857403191,
      "loss": 1.0374,
      "step": 3966
    },
    {
      "epoch": 0.34130603114514324,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009886805652321088,
      "loss": 0.8346,
      "step": 3967
    },
    {
      "epoch": 0.3413920674524649,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009886748432946027,
      "loss": 1.2047,
      "step": 3968
    },
    {
      "epoch": 0.34147810375978666,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009886691199278176,
      "loss": 1.3713,
      "step": 3969
    },
    {
      "epoch": 0.34156414006710833,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009886633951317707,
      "loss": 1.1657,
      "step": 3970
    },
    {
      "epoch": 0.34165017637443,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009886576689064782,
      "loss": 0.9735,
      "step": 3971
    },
    {
      "epoch": 0.3417362126817517,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009886519412519571,
      "loss": 1.165,
      "step": 3972
    },
    {
      "epoch": 0.3418222489890734,
      "grad_norm": 1.328125,
      "learning_rate": 0.000988646212168224,
      "loss": 1.7325,
      "step": 3973
    },
    {
      "epoch": 0.34190828529639505,
      "grad_norm": 1.921875,
      "learning_rate": 0.000988640481655296,
      "loss": 1.5172,
      "step": 3974
    },
    {
      "epoch": 0.3419943216037168,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009886347497131895,
      "loss": 1.1807,
      "step": 3975
    },
    {
      "epoch": 0.34208035791103847,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009886290163419213,
      "loss": 1.4795,
      "step": 3976
    },
    {
      "epoch": 0.34216639421836015,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009886232815415085,
      "loss": 1.61,
      "step": 3977
    },
    {
      "epoch": 0.34225243052568183,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009886175453119675,
      "loss": 0.9475,
      "step": 3978
    },
    {
      "epoch": 0.3423384668330035,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009886118076533152,
      "loss": 2.057,
      "step": 3979
    },
    {
      "epoch": 0.34242450314032524,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009886060685655685,
      "loss": 1.4729,
      "step": 3980
    },
    {
      "epoch": 0.3425105394476469,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009886003280487438,
      "loss": 1.5084,
      "step": 3981
    },
    {
      "epoch": 0.3425965757549686,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009885945861028585,
      "loss": 1.7197,
      "step": 3982
    },
    {
      "epoch": 0.3426826120622903,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009885888427279288,
      "loss": 1.026,
      "step": 3983
    },
    {
      "epoch": 0.34276864836961196,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009885830979239718,
      "loss": 2.1971,
      "step": 3984
    },
    {
      "epoch": 0.34285468467693364,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009885773516910045,
      "loss": 1.2875,
      "step": 3985
    },
    {
      "epoch": 0.3429407209842554,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009885716040290433,
      "loss": 1.163,
      "step": 3986
    },
    {
      "epoch": 0.34302675729157706,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009885658549381052,
      "loss": 1.1505,
      "step": 3987
    },
    {
      "epoch": 0.34311279359889874,
      "grad_norm": 2.703125,
      "learning_rate": 0.0009885601044182068,
      "loss": 2.3003,
      "step": 3988
    },
    {
      "epoch": 0.3431988299062204,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009885543524693653,
      "loss": 1.8585,
      "step": 3989
    },
    {
      "epoch": 0.3432848662135421,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009885485990915974,
      "loss": 0.9538,
      "step": 3990
    },
    {
      "epoch": 0.34337090252086383,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009885428442849198,
      "loss": 1.9822,
      "step": 3991
    },
    {
      "epoch": 0.3434569388281855,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009885370880493493,
      "loss": 1.0965,
      "step": 3992
    },
    {
      "epoch": 0.3435429751355072,
      "grad_norm": 1.171875,
      "learning_rate": 0.000988531330384903,
      "loss": 1.0384,
      "step": 3993
    },
    {
      "epoch": 0.34362901144282887,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009885255712915974,
      "loss": 1.3643,
      "step": 3994
    },
    {
      "epoch": 0.34371504775015055,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009885198107694498,
      "loss": 1.3955,
      "step": 3995
    },
    {
      "epoch": 0.34380108405747223,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009885140488184764,
      "loss": 1.0642,
      "step": 3996
    },
    {
      "epoch": 0.34388712036479396,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009885082854386947,
      "loss": 0.8635,
      "step": 3997
    },
    {
      "epoch": 0.34397315667211564,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000988502520630121,
      "loss": 1.1183,
      "step": 3998
    },
    {
      "epoch": 0.3440591929794373,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009884967543927728,
      "loss": 1.7601,
      "step": 3999
    },
    {
      "epoch": 0.344145229286759,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009884909867266662,
      "loss": 2.4074,
      "step": 4000
    },
    {
      "epoch": 0.3442312655940807,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000988485217631819,
      "loss": 1.5012,
      "step": 4001
    },
    {
      "epoch": 0.3443173019014024,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000988479447108247,
      "loss": 1.5426,
      "step": 4002
    },
    {
      "epoch": 0.3444033382087241,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000988473675155968,
      "loss": 1.0684,
      "step": 4003
    },
    {
      "epoch": 0.3444893745160458,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009884679017749982,
      "loss": 2.2374,
      "step": 4004
    },
    {
      "epoch": 0.34457541082336746,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000988462126965355,
      "loss": 1.3639,
      "step": 4005
    },
    {
      "epoch": 0.34466144713068914,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000988456350727055,
      "loss": 1.1263,
      "step": 4006
    },
    {
      "epoch": 0.3447474834380108,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009884505730601153,
      "loss": 1.3441,
      "step": 4007
    },
    {
      "epoch": 0.34483351974533255,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009884447939645525,
      "loss": 1.5186,
      "step": 4008
    },
    {
      "epoch": 0.34491955605265423,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009884390134403836,
      "loss": 1.3735,
      "step": 4009
    },
    {
      "epoch": 0.3450055923599759,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009884332314876257,
      "loss": 1.1995,
      "step": 4010
    },
    {
      "epoch": 0.3450916286672976,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009884274481062956,
      "loss": 1.6289,
      "step": 4011
    },
    {
      "epoch": 0.34517766497461927,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009884216632964102,
      "loss": 1.1481,
      "step": 4012
    },
    {
      "epoch": 0.345263701281941,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009884158770579864,
      "loss": 1.1338,
      "step": 4013
    },
    {
      "epoch": 0.3453497375892627,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009884100893910413,
      "loss": 1.403,
      "step": 4014
    },
    {
      "epoch": 0.34543577389658436,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009884043002955916,
      "loss": 1.5195,
      "step": 4015
    },
    {
      "epoch": 0.34552181020390604,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009883985097716543,
      "loss": 1.1705,
      "step": 4016
    },
    {
      "epoch": 0.3456078465112277,
      "grad_norm": 1.375,
      "learning_rate": 0.0009883927178192461,
      "loss": 1.06,
      "step": 4017
    },
    {
      "epoch": 0.3456938828185494,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009883869244383845,
      "loss": 1.1563,
      "step": 4018
    },
    {
      "epoch": 0.34577991912587114,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000988381129629086,
      "loss": 1.7605,
      "step": 4019
    },
    {
      "epoch": 0.3458659554331928,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009883753333913678,
      "loss": 2.2299,
      "step": 4020
    },
    {
      "epoch": 0.3459519917405145,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009883695357252467,
      "loss": 0.9965,
      "step": 4021
    },
    {
      "epoch": 0.3460380280478362,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009883637366307396,
      "loss": 0.9434,
      "step": 4022
    },
    {
      "epoch": 0.34612406435515786,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009883579361078636,
      "loss": 0.8293,
      "step": 4023
    },
    {
      "epoch": 0.3462101006624796,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009883521341566357,
      "loss": 2.0922,
      "step": 4024
    },
    {
      "epoch": 0.34629613696980127,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009883463307770728,
      "loss": 1.3291,
      "step": 4025
    },
    {
      "epoch": 0.34638217327712295,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009883405259691917,
      "loss": 1.4604,
      "step": 4026
    },
    {
      "epoch": 0.34646820958444463,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009883347197330095,
      "loss": 2.0132,
      "step": 4027
    },
    {
      "epoch": 0.3465542458917663,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009883289120685434,
      "loss": 2.0517,
      "step": 4028
    },
    {
      "epoch": 0.346640282199088,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00098832310297581,
      "loss": 1.6968,
      "step": 4029
    },
    {
      "epoch": 0.3467263185064097,
      "grad_norm": 1.4375,
      "learning_rate": 0.000988317292454827,
      "loss": 1.5537,
      "step": 4030
    },
    {
      "epoch": 0.3468123548137314,
      "grad_norm": 0.84375,
      "learning_rate": 0.0009883114805056103,
      "loss": 1.2526,
      "step": 4031
    },
    {
      "epoch": 0.3468983911210531,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009883056671281779,
      "loss": 1.282,
      "step": 4032
    },
    {
      "epoch": 0.34698442742837476,
      "grad_norm": 2.0625,
      "learning_rate": 0.000988299852322546,
      "loss": 1.9276,
      "step": 4033
    },
    {
      "epoch": 0.34707046373569644,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009882940360887323,
      "loss": 1.3158,
      "step": 4034
    },
    {
      "epoch": 0.3471565000430182,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009882882184267534,
      "loss": 1.9266,
      "step": 4035
    },
    {
      "epoch": 0.34724253635033986,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009882823993366265,
      "loss": 1.5865,
      "step": 4036
    },
    {
      "epoch": 0.34732857265766154,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009882765788183686,
      "loss": 1.7859,
      "step": 4037
    },
    {
      "epoch": 0.3474146089649832,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009882707568719964,
      "loss": 1.349,
      "step": 4038
    },
    {
      "epoch": 0.3475006452723049,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009882649334975273,
      "loss": 1.8697,
      "step": 4039
    },
    {
      "epoch": 0.3475866815796266,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009882591086949784,
      "loss": 1.3299,
      "step": 4040
    },
    {
      "epoch": 0.3476727178869483,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009882532824643665,
      "loss": 1.9229,
      "step": 4041
    },
    {
      "epoch": 0.34775875419427,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009882474548057088,
      "loss": 1.0097,
      "step": 4042
    },
    {
      "epoch": 0.34784479050159167,
      "grad_norm": 1.046875,
      "learning_rate": 0.000988241625719022,
      "loss": 1.1682,
      "step": 4043
    },
    {
      "epoch": 0.34793082680891335,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009882357952043235,
      "loss": 1.8633,
      "step": 4044
    },
    {
      "epoch": 0.34801686311623503,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009882299632616304,
      "loss": 1.1316,
      "step": 4045
    },
    {
      "epoch": 0.34810289942355677,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009882241298909595,
      "loss": 0.9093,
      "step": 4046
    },
    {
      "epoch": 0.34818893573087845,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009882182950923279,
      "loss": 1.4304,
      "step": 4047
    },
    {
      "epoch": 0.3482749720382001,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000988212458865753,
      "loss": 1.4824,
      "step": 4048
    },
    {
      "epoch": 0.3483610083455218,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009882066212112512,
      "loss": 1.7348,
      "step": 4049
    },
    {
      "epoch": 0.3484470446528435,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009882007821288403,
      "loss": 1.3461,
      "step": 4050
    },
    {
      "epoch": 0.34853308096016516,
      "grad_norm": 1.484375,
      "learning_rate": 0.000988194941618537,
      "loss": 1.0608,
      "step": 4051
    },
    {
      "epoch": 0.3486191172674869,
      "grad_norm": 1.875,
      "learning_rate": 0.0009881890996803584,
      "loss": 1.4241,
      "step": 4052
    },
    {
      "epoch": 0.3487051535748086,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009881832563143217,
      "loss": 1.2045,
      "step": 4053
    },
    {
      "epoch": 0.34879118988213026,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009881774115204438,
      "loss": 1.0229,
      "step": 4054
    },
    {
      "epoch": 0.34887722618945194,
      "grad_norm": 1.25,
      "learning_rate": 0.000988171565298742,
      "loss": 1.5126,
      "step": 4055
    },
    {
      "epoch": 0.3489632624967736,
      "grad_norm": 1.375,
      "learning_rate": 0.0009881657176492333,
      "loss": 0.9457,
      "step": 4056
    },
    {
      "epoch": 0.34904929880409535,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009881598685719348,
      "loss": 1.7768,
      "step": 4057
    },
    {
      "epoch": 0.34913533511141703,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009881540180668637,
      "loss": 2.1161,
      "step": 4058
    },
    {
      "epoch": 0.3492213714187387,
      "grad_norm": 1.9296875,
      "learning_rate": 0.000988148166134037,
      "loss": 1.4319,
      "step": 4059
    },
    {
      "epoch": 0.3493074077260604,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009881423127734718,
      "loss": 1.5377,
      "step": 4060
    },
    {
      "epoch": 0.3493934440333821,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009881364579851853,
      "loss": 1.3085,
      "step": 4061
    },
    {
      "epoch": 0.34947948034070375,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009881306017691946,
      "loss": 1.826,
      "step": 4062
    },
    {
      "epoch": 0.3495655166480255,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009881247441255168,
      "loss": 1.3144,
      "step": 4063
    },
    {
      "epoch": 0.34965155295534717,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009881188850541691,
      "loss": 0.8947,
      "step": 4064
    },
    {
      "epoch": 0.34973758926266885,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009881130245551685,
      "loss": 2.0524,
      "step": 4065
    },
    {
      "epoch": 0.3498236255699905,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009881071626285326,
      "loss": 1.6887,
      "step": 4066
    },
    {
      "epoch": 0.3499096618773122,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009881012992742778,
      "loss": 1.1725,
      "step": 4067
    },
    {
      "epoch": 0.34999569818463394,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009880954344924219,
      "loss": 1.2931,
      "step": 4068
    },
    {
      "epoch": 0.3500817344919556,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009880895682829814,
      "loss": 1.1109,
      "step": 4069
    },
    {
      "epoch": 0.3501677707992773,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000988083700645974,
      "loss": 1.558,
      "step": 4070
    },
    {
      "epoch": 0.350253807106599,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009880778315814168,
      "loss": 0.7961,
      "step": 4071
    },
    {
      "epoch": 0.35033984341392066,
      "grad_norm": 1.21875,
      "learning_rate": 0.000988071961089327,
      "loss": 1.2019,
      "step": 4072
    },
    {
      "epoch": 0.35042587972124234,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009880660891697212,
      "loss": 1.5243,
      "step": 4073
    },
    {
      "epoch": 0.3505119160285641,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009880602158226176,
      "loss": 1.1648,
      "step": 4074
    },
    {
      "epoch": 0.35059795233588575,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009880543410480322,
      "loss": 1.4893,
      "step": 4075
    },
    {
      "epoch": 0.35068398864320743,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009880484648459832,
      "loss": 1.3538,
      "step": 4076
    },
    {
      "epoch": 0.3507700249505291,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009880425872164872,
      "loss": 0.7941,
      "step": 4077
    },
    {
      "epoch": 0.3508560612578508,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009880367081595617,
      "loss": 0.9712,
      "step": 4078
    },
    {
      "epoch": 0.35094209756517253,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009880308276752235,
      "loss": 1.1331,
      "step": 4079
    },
    {
      "epoch": 0.3510281338724942,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009880249457634902,
      "loss": 1.4265,
      "step": 4080
    },
    {
      "epoch": 0.3511141701798159,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009880190624243787,
      "loss": 1.5949,
      "step": 4081
    },
    {
      "epoch": 0.35120020648713757,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009880131776579064,
      "loss": 1.2473,
      "step": 4082
    },
    {
      "epoch": 0.35128624279445925,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009880072914640907,
      "loss": 1.5459,
      "step": 4083
    },
    {
      "epoch": 0.3513722791017809,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009880014038429483,
      "loss": 1.7318,
      "step": 4084
    },
    {
      "epoch": 0.35145831540910266,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009879955147944967,
      "loss": 1.8924,
      "step": 4085
    },
    {
      "epoch": 0.35154435171642434,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009879896243187535,
      "loss": 0.9873,
      "step": 4086
    },
    {
      "epoch": 0.351630388023746,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009879837324157353,
      "loss": 0.747,
      "step": 4087
    },
    {
      "epoch": 0.3517164243310677,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009879778390854594,
      "loss": 1.4973,
      "step": 4088
    },
    {
      "epoch": 0.3518024606383894,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009879719443279436,
      "loss": 1.1115,
      "step": 4089
    },
    {
      "epoch": 0.3518884969457111,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009879660481432045,
      "loss": 2.1321,
      "step": 4090
    },
    {
      "epoch": 0.3519745332530328,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009879601505312597,
      "loss": 1.3872,
      "step": 4091
    },
    {
      "epoch": 0.3520605695603545,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009879542514921263,
      "loss": 1.3866,
      "step": 4092
    },
    {
      "epoch": 0.35214660586767615,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000987948351025822,
      "loss": 1.683,
      "step": 4093
    },
    {
      "epoch": 0.35223264217499783,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009879424491323631,
      "loss": 1.9981,
      "step": 4094
    },
    {
      "epoch": 0.3523186784823195,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009879365458117678,
      "loss": 1.9366,
      "step": 4095
    },
    {
      "epoch": 0.35240471478964125,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009879306410640527,
      "loss": 1.7199,
      "step": 4096
    },
    {
      "epoch": 0.35249075109696293,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009879247348892357,
      "loss": 1.9687,
      "step": 4097
    },
    {
      "epoch": 0.3525767874042846,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009879188272873336,
      "loss": 1.2175,
      "step": 4098
    },
    {
      "epoch": 0.3526628237116063,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009879129182583637,
      "loss": 0.9573,
      "step": 4099
    },
    {
      "epoch": 0.35274886001892797,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009879070078023437,
      "loss": 1.3013,
      "step": 4100
    },
    {
      "epoch": 0.3528348963262497,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009879010959192905,
      "loss": 1.2105,
      "step": 4101
    },
    {
      "epoch": 0.3529209326335714,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009878951826092213,
      "loss": 1.1469,
      "step": 4102
    },
    {
      "epoch": 0.35300696894089306,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009878892678721538,
      "loss": 2.0975,
      "step": 4103
    },
    {
      "epoch": 0.35309300524821474,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009878833517081049,
      "loss": 1.3362,
      "step": 4104
    },
    {
      "epoch": 0.3531790415555364,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009878774341170922,
      "loss": 0.7828,
      "step": 4105
    },
    {
      "epoch": 0.3532650778628581,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009878715150991328,
      "loss": 0.8732,
      "step": 4106
    },
    {
      "epoch": 0.35335111417017984,
      "grad_norm": 1.25,
      "learning_rate": 0.0009878655946542443,
      "loss": 1.7218,
      "step": 4107
    },
    {
      "epoch": 0.3534371504775015,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009878596727824437,
      "loss": 1.8813,
      "step": 4108
    },
    {
      "epoch": 0.3535231867848232,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009878537494837482,
      "loss": 0.9156,
      "step": 4109
    },
    {
      "epoch": 0.3536092230921449,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009878478247581756,
      "loss": 1.3192,
      "step": 4110
    },
    {
      "epoch": 0.35369525939946656,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009878418986057431,
      "loss": 1.3843,
      "step": 4111
    },
    {
      "epoch": 0.3537812957067883,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009878359710264678,
      "loss": 1.0076,
      "step": 4112
    },
    {
      "epoch": 0.35386733201410997,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000987830042020367,
      "loss": 1.2654,
      "step": 4113
    },
    {
      "epoch": 0.35395336832143165,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009878241115874587,
      "loss": 1.2736,
      "step": 4114
    },
    {
      "epoch": 0.35403940462875333,
      "grad_norm": 2.390625,
      "learning_rate": 0.0009878181797277595,
      "loss": 1.7806,
      "step": 4115
    },
    {
      "epoch": 0.354125440936075,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009878122464412869,
      "loss": 1.5183,
      "step": 4116
    },
    {
      "epoch": 0.3542114772433967,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009878063117280582,
      "loss": 1.7667,
      "step": 4117
    },
    {
      "epoch": 0.3542975135507184,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009878003755880913,
      "loss": 1.8923,
      "step": 4118
    },
    {
      "epoch": 0.3543835498580401,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009877944380214027,
      "loss": 0.7438,
      "step": 4119
    },
    {
      "epoch": 0.3544695861653618,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009877884990280106,
      "loss": 2.052,
      "step": 4120
    },
    {
      "epoch": 0.35455562247268346,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000987782558607932,
      "loss": 1.0454,
      "step": 4121
    },
    {
      "epoch": 0.35464165878000514,
      "grad_norm": 1.375,
      "learning_rate": 0.0009877766167611841,
      "loss": 1.1377,
      "step": 4122
    },
    {
      "epoch": 0.3547276950873269,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009877706734877846,
      "loss": 1.2289,
      "step": 4123
    },
    {
      "epoch": 0.35481373139464856,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009877647287877508,
      "loss": 2.4191,
      "step": 4124
    },
    {
      "epoch": 0.35489976770197024,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009877587826610998,
      "loss": 1.2567,
      "step": 4125
    },
    {
      "epoch": 0.3549858040092919,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009877528351078496,
      "loss": 1.6859,
      "step": 4126
    },
    {
      "epoch": 0.3550718403166136,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000987746886128017,
      "loss": 1.5803,
      "step": 4127
    },
    {
      "epoch": 0.3551578766239353,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009877409357216195,
      "loss": 1.9856,
      "step": 4128
    },
    {
      "epoch": 0.355243912931257,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009877349838886748,
      "loss": 1.5443,
      "step": 4129
    },
    {
      "epoch": 0.3553299492385787,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009877290306292,
      "loss": 1.748,
      "step": 4130
    },
    {
      "epoch": 0.35541598554590037,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009877230759432127,
      "loss": 1.8208,
      "step": 4131
    },
    {
      "epoch": 0.35550202185322205,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009877171198307303,
      "loss": 1.6643,
      "step": 4132
    },
    {
      "epoch": 0.35558805816054373,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009877111622917701,
      "loss": 0.9751,
      "step": 4133
    },
    {
      "epoch": 0.35567409446786546,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009877052033263496,
      "loss": 1.2514,
      "step": 4134
    },
    {
      "epoch": 0.35576013077518714,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009876992429344864,
      "loss": 0.9507,
      "step": 4135
    },
    {
      "epoch": 0.3558461670825088,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009876932811161976,
      "loss": 0.8084,
      "step": 4136
    },
    {
      "epoch": 0.3559322033898305,
      "grad_norm": 1.21875,
      "learning_rate": 0.000987687317871501,
      "loss": 1.8213,
      "step": 4137
    },
    {
      "epoch": 0.3560182396971522,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009876813532004136,
      "loss": 1.397,
      "step": 4138
    },
    {
      "epoch": 0.35610427600447386,
      "grad_norm": 2.171875,
      "learning_rate": 0.0009876753871029531,
      "loss": 0.9909,
      "step": 4139
    },
    {
      "epoch": 0.3561903123117956,
      "grad_norm": 1.84375,
      "learning_rate": 0.000987669419579137,
      "loss": 2.1885,
      "step": 4140
    },
    {
      "epoch": 0.3562763486191173,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009876634506289829,
      "loss": 1.3476,
      "step": 4141
    },
    {
      "epoch": 0.35636238492643896,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009876574802525078,
      "loss": 1.0772,
      "step": 4142
    },
    {
      "epoch": 0.35644842123376064,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009876515084497296,
      "loss": 1.3776,
      "step": 4143
    },
    {
      "epoch": 0.3565344575410823,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009876455352206654,
      "loss": 1.6274,
      "step": 4144
    },
    {
      "epoch": 0.35662049384840405,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000987639560565333,
      "loss": 1.2522,
      "step": 4145
    },
    {
      "epoch": 0.35670653015572573,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009876335844837497,
      "loss": 1.6312,
      "step": 4146
    },
    {
      "epoch": 0.3567925664630474,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009876276069759329,
      "loss": 0.7235,
      "step": 4147
    },
    {
      "epoch": 0.3568786027703691,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009876216280419003,
      "loss": 1.0214,
      "step": 4148
    },
    {
      "epoch": 0.35696463907769077,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009876156476816692,
      "loss": 1.0525,
      "step": 4149
    },
    {
      "epoch": 0.35705067538501245,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000987609665895257,
      "loss": 1.5308,
      "step": 4150
    },
    {
      "epoch": 0.3571367116923342,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009876036826826817,
      "loss": 1.6224,
      "step": 4151
    },
    {
      "epoch": 0.35722274799965587,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009875976980439603,
      "loss": 1.5752,
      "step": 4152
    },
    {
      "epoch": 0.35730878430697754,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009875917119791106,
      "loss": 1.4925,
      "step": 4153
    },
    {
      "epoch": 0.3573948206142992,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009875857244881498,
      "loss": 2.4834,
      "step": 4154
    },
    {
      "epoch": 0.3574808569216209,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009875797355710955,
      "loss": 1.6198,
      "step": 4155
    },
    {
      "epoch": 0.35756689322894264,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009875737452279655,
      "loss": 1.3587,
      "step": 4156
    },
    {
      "epoch": 0.3576529295362643,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000987567753458777,
      "loss": 1.1724,
      "step": 4157
    },
    {
      "epoch": 0.357738965843586,
      "grad_norm": 12.25,
      "learning_rate": 0.0009875617602635477,
      "loss": 0.9257,
      "step": 4158
    },
    {
      "epoch": 0.3578250021509077,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009875557656422952,
      "loss": 1.8894,
      "step": 4159
    },
    {
      "epoch": 0.35791103845822936,
      "grad_norm": 3.4375,
      "learning_rate": 0.000987549769595037,
      "loss": 1.211,
      "step": 4160
    },
    {
      "epoch": 0.35799707476555104,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009875437721217901,
      "loss": 1.305,
      "step": 4161
    },
    {
      "epoch": 0.3580831110728728,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009875377732225727,
      "loss": 0.8911,
      "step": 4162
    },
    {
      "epoch": 0.35816914738019445,
      "grad_norm": 2.171875,
      "learning_rate": 0.0009875317728974022,
      "loss": 1.2954,
      "step": 4163
    },
    {
      "epoch": 0.35825518368751613,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000987525771146296,
      "loss": 0.7291,
      "step": 4164
    },
    {
      "epoch": 0.3583412199948378,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009875197679692718,
      "loss": 1.5534,
      "step": 4165
    },
    {
      "epoch": 0.3584272563021595,
      "grad_norm": 1.28125,
      "learning_rate": 0.000987513763366347,
      "loss": 1.5284,
      "step": 4166
    },
    {
      "epoch": 0.3585132926094812,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009875077573375396,
      "loss": 1.5329,
      "step": 4167
    },
    {
      "epoch": 0.3585993289168029,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009875017498828665,
      "loss": 1.9112,
      "step": 4168
    },
    {
      "epoch": 0.3586853652241246,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009874957410023455,
      "loss": 2.2008,
      "step": 4169
    },
    {
      "epoch": 0.35877140153144627,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009874897306959946,
      "loss": 0.8711,
      "step": 4170
    },
    {
      "epoch": 0.35885743783876795,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000987483718963831,
      "loss": 1.3369,
      "step": 4171
    },
    {
      "epoch": 0.3589434741460896,
      "grad_norm": 1.375,
      "learning_rate": 0.0009874777058058721,
      "loss": 1.4168,
      "step": 4172
    },
    {
      "epoch": 0.35902951045341136,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009874716912221357,
      "loss": 1.4886,
      "step": 4173
    },
    {
      "epoch": 0.35911554676073304,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009874656752126396,
      "loss": 0.8183,
      "step": 4174
    },
    {
      "epoch": 0.3592015830680547,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009874596577774011,
      "loss": 1.6414,
      "step": 4175
    },
    {
      "epoch": 0.3592876193753764,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000987453638916438,
      "loss": 2.2633,
      "step": 4176
    },
    {
      "epoch": 0.3593736556826981,
      "grad_norm": 2.9375,
      "learning_rate": 0.0009874476186297679,
      "loss": 1.1826,
      "step": 4177
    },
    {
      "epoch": 0.3594596919900198,
      "grad_norm": 2.5,
      "learning_rate": 0.0009874415969174082,
      "loss": 1.1013,
      "step": 4178
    },
    {
      "epoch": 0.3595457282973415,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009874355737793765,
      "loss": 1.7798,
      "step": 4179
    },
    {
      "epoch": 0.3596317646046632,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009874295492156907,
      "loss": 2.1634,
      "step": 4180
    },
    {
      "epoch": 0.35971780091198485,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009874235232263683,
      "loss": 1.2388,
      "step": 4181
    },
    {
      "epoch": 0.35980383721930653,
      "grad_norm": 3.5,
      "learning_rate": 0.0009874174958114268,
      "loss": 1.5167,
      "step": 4182
    },
    {
      "epoch": 0.3598898735266282,
      "grad_norm": 1.765625,
      "learning_rate": 0.000987411466970884,
      "loss": 1.9734,
      "step": 4183
    },
    {
      "epoch": 0.35997590983394995,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009874054367047574,
      "loss": 1.2524,
      "step": 4184
    },
    {
      "epoch": 0.3600619461412716,
      "grad_norm": 2.71875,
      "learning_rate": 0.0009873994050130647,
      "loss": 1.4212,
      "step": 4185
    },
    {
      "epoch": 0.3601479824485933,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009873933718958234,
      "loss": 1.8733,
      "step": 4186
    },
    {
      "epoch": 0.360234018755915,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009873873373530515,
      "loss": 1.6563,
      "step": 4187
    },
    {
      "epoch": 0.36032005506323667,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009873813013847664,
      "loss": 1.8431,
      "step": 4188
    },
    {
      "epoch": 0.3604060913705584,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009873752639909857,
      "loss": 1.3906,
      "step": 4189
    },
    {
      "epoch": 0.3604921276778801,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009873692251717273,
      "loss": 1.6862,
      "step": 4190
    },
    {
      "epoch": 0.36057816398520176,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009873631849270084,
      "loss": 1.8892,
      "step": 4191
    },
    {
      "epoch": 0.36066420029252344,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009873571432568473,
      "loss": 1.3589,
      "step": 4192
    },
    {
      "epoch": 0.3607502365998451,
      "grad_norm": 1.625,
      "learning_rate": 0.000987351100161261,
      "loss": 1.3409,
      "step": 4193
    },
    {
      "epoch": 0.3608362729071668,
      "grad_norm": 1.375,
      "learning_rate": 0.0009873450556402677,
      "loss": 1.3732,
      "step": 4194
    },
    {
      "epoch": 0.36092230921448853,
      "grad_norm": 1.609375,
      "learning_rate": 0.000987339009693885,
      "loss": 1.867,
      "step": 4195
    },
    {
      "epoch": 0.3610083455218102,
      "grad_norm": 1.875,
      "learning_rate": 0.0009873329623221302,
      "loss": 1.2205,
      "step": 4196
    },
    {
      "epoch": 0.3610943818291319,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009873269135250216,
      "loss": 1.1571,
      "step": 4197
    },
    {
      "epoch": 0.3611804181364536,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009873208633025764,
      "loss": 1.2941,
      "step": 4198
    },
    {
      "epoch": 0.36126645444377525,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009873148116548126,
      "loss": 0.8864,
      "step": 4199
    },
    {
      "epoch": 0.361352490751097,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009873087585817474,
      "loss": 2.3538,
      "step": 4200
    },
    {
      "epoch": 0.36143852705841867,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009873027040833992,
      "loss": 1.4518,
      "step": 4201
    },
    {
      "epoch": 0.36152456336574035,
      "grad_norm": 1.53125,
      "learning_rate": 0.000987296648159785,
      "loss": 1.7547,
      "step": 4202
    },
    {
      "epoch": 0.361610599673062,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009872905908109232,
      "loss": 1.4729,
      "step": 4203
    },
    {
      "epoch": 0.3616966359803837,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009872845320368312,
      "loss": 0.9575,
      "step": 4204
    },
    {
      "epoch": 0.3617826722877054,
      "grad_norm": 2.125,
      "learning_rate": 0.0009872784718375266,
      "loss": 1.2191,
      "step": 4205
    },
    {
      "epoch": 0.3618687085950271,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009872724102130274,
      "loss": 1.0222,
      "step": 4206
    },
    {
      "epoch": 0.3619547449023488,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000987266347163351,
      "loss": 0.6039,
      "step": 4207
    },
    {
      "epoch": 0.3620407812096705,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009872602826885154,
      "loss": 1.4269,
      "step": 4208
    },
    {
      "epoch": 0.36212681751699216,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009872542167885382,
      "loss": 1.2639,
      "step": 4209
    },
    {
      "epoch": 0.36221285382431384,
      "grad_norm": 1.265625,
      "learning_rate": 0.000987248149463437,
      "loss": 1.2074,
      "step": 4210
    },
    {
      "epoch": 0.3622988901316356,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009872420807132299,
      "loss": 1.3037,
      "step": 4211
    },
    {
      "epoch": 0.36238492643895726,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009872360105379345,
      "loss": 2.2797,
      "step": 4212
    },
    {
      "epoch": 0.36247096274627894,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009872299389375684,
      "loss": 1.2676,
      "step": 4213
    },
    {
      "epoch": 0.3625569990536006,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009872238659121497,
      "loss": 1.8673,
      "step": 4214
    },
    {
      "epoch": 0.3626430353609223,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009872177914616957,
      "loss": 0.8005,
      "step": 4215
    },
    {
      "epoch": 0.362729071668244,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009872117155862246,
      "loss": 1.4038,
      "step": 4216
    },
    {
      "epoch": 0.3628151079755657,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000987205638285754,
      "loss": 1.1882,
      "step": 4217
    },
    {
      "epoch": 0.3629011442828874,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009871995595603015,
      "loss": 0.7041,
      "step": 4218
    },
    {
      "epoch": 0.36298718059020907,
      "grad_norm": 1.125,
      "learning_rate": 0.0009871934794098852,
      "loss": 1.5201,
      "step": 4219
    },
    {
      "epoch": 0.36307321689753075,
      "grad_norm": 1.375,
      "learning_rate": 0.0009871873978345228,
      "loss": 1.183,
      "step": 4220
    },
    {
      "epoch": 0.36315925320485243,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009871813148342318,
      "loss": 1.5164,
      "step": 4221
    },
    {
      "epoch": 0.36324528951217416,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009871752304090303,
      "loss": 1.268,
      "step": 4222
    },
    {
      "epoch": 0.36333132581949584,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000987169144558936,
      "loss": 0.9833,
      "step": 4223
    },
    {
      "epoch": 0.3634173621268175,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009871630572839667,
      "loss": 1.8133,
      "step": 4224
    },
    {
      "epoch": 0.3635033984341392,
      "grad_norm": 1.09375,
      "learning_rate": 0.00098715696858414,
      "loss": 0.8553,
      "step": 4225
    },
    {
      "epoch": 0.3635894347414609,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009871508784594742,
      "loss": 0.7393,
      "step": 4226
    },
    {
      "epoch": 0.36367547104878256,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009871447869099867,
      "loss": 1.4083,
      "step": 4227
    },
    {
      "epoch": 0.3637615073561043,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009871386939356956,
      "loss": 1.2462,
      "step": 4228
    },
    {
      "epoch": 0.363847543663426,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009871325995366184,
      "loss": 1.4934,
      "step": 4229
    },
    {
      "epoch": 0.36393357997074766,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009871265037127732,
      "loss": 1.707,
      "step": 4230
    },
    {
      "epoch": 0.36401961627806934,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009871204064641776,
      "loss": 1.4123,
      "step": 4231
    },
    {
      "epoch": 0.364105652585391,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009871143077908498,
      "loss": 1.5327,
      "step": 4232
    },
    {
      "epoch": 0.36419168889271275,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009871082076928072,
      "loss": 1.2648,
      "step": 4233
    },
    {
      "epoch": 0.36427772520003443,
      "grad_norm": 1.375,
      "learning_rate": 0.0009871021061700676,
      "loss": 1.2525,
      "step": 4234
    },
    {
      "epoch": 0.3643637615073561,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009870960032226494,
      "loss": 1.0605,
      "step": 4235
    },
    {
      "epoch": 0.3644497978146778,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00098708989885057,
      "loss": 1.5519,
      "step": 4236
    },
    {
      "epoch": 0.36453583412199947,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009870837930538473,
      "loss": 1.974,
      "step": 4237
    },
    {
      "epoch": 0.36462187042932115,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009870776858324996,
      "loss": 1.5951,
      "step": 4238
    },
    {
      "epoch": 0.3647079067366429,
      "grad_norm": 1.328125,
      "learning_rate": 0.000987071577186544,
      "loss": 1.8931,
      "step": 4239
    },
    {
      "epoch": 0.36479394304396456,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000987065467115999,
      "loss": 1.9919,
      "step": 4240
    },
    {
      "epoch": 0.36487997935128624,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000987059355620882,
      "loss": 1.9841,
      "step": 4241
    },
    {
      "epoch": 0.3649660156586079,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009870532427012112,
      "loss": 1.2406,
      "step": 4242
    },
    {
      "epoch": 0.3650520519659296,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009870471283570045,
      "loss": 1.0304,
      "step": 4243
    },
    {
      "epoch": 0.36513808827325134,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009870410125882795,
      "loss": 0.9549,
      "step": 4244
    },
    {
      "epoch": 0.365224124580573,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009870348953950545,
      "loss": 0.8809,
      "step": 4245
    },
    {
      "epoch": 0.3653101608878947,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009870287767773471,
      "loss": 1.4569,
      "step": 4246
    },
    {
      "epoch": 0.3653961971952164,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009870226567351753,
      "loss": 2.5247,
      "step": 4247
    },
    {
      "epoch": 0.36548223350253806,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009870165352685568,
      "loss": 1.5378,
      "step": 4248
    },
    {
      "epoch": 0.36556826980985974,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009870104123775097,
      "loss": 1.1856,
      "step": 4249
    },
    {
      "epoch": 0.36565430611718147,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009870042880620517,
      "loss": 1.0595,
      "step": 4250
    },
    {
      "epoch": 0.36574034242450315,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000986998162322201,
      "loss": 2.3711,
      "step": 4251
    },
    {
      "epoch": 0.36582637873182483,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009869920351579756,
      "loss": 1.4615,
      "step": 4252
    },
    {
      "epoch": 0.3659124150391465,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009869859065693928,
      "loss": 1.716,
      "step": 4253
    },
    {
      "epoch": 0.3659984513464682,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009869797765564715,
      "loss": 1.1178,
      "step": 4254
    },
    {
      "epoch": 0.3660844876537899,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009869736451192286,
      "loss": 1.0635,
      "step": 4255
    },
    {
      "epoch": 0.3661705239611116,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009869675122576827,
      "loss": 1.4255,
      "step": 4256
    },
    {
      "epoch": 0.3662565602684333,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009869613779718515,
      "loss": 1.2716,
      "step": 4257
    },
    {
      "epoch": 0.36634259657575496,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009869552422617528,
      "loss": 1.3563,
      "step": 4258
    },
    {
      "epoch": 0.36642863288307664,
      "grad_norm": 1.125,
      "learning_rate": 0.0009869491051274048,
      "loss": 1.3735,
      "step": 4259
    },
    {
      "epoch": 0.3665146691903983,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009869429665688256,
      "loss": 1.1006,
      "step": 4260
    },
    {
      "epoch": 0.36660070549772006,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009869368265860328,
      "loss": 1.0807,
      "step": 4261
    },
    {
      "epoch": 0.36668674180504174,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009869306851790446,
      "loss": 2.0444,
      "step": 4262
    },
    {
      "epoch": 0.3667727781123634,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009869245423478785,
      "loss": 0.8119,
      "step": 4263
    },
    {
      "epoch": 0.3668588144196851,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000986918398092553,
      "loss": 1.468,
      "step": 4264
    },
    {
      "epoch": 0.3669448507270068,
      "grad_norm": 1.390625,
      "learning_rate": 0.000986912252413086,
      "loss": 0.882,
      "step": 4265
    },
    {
      "epoch": 0.3670308870343285,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009869061053094951,
      "loss": 1.5331,
      "step": 4266
    },
    {
      "epoch": 0.3671169233416502,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009868999567817988,
      "loss": 1.6425,
      "step": 4267
    },
    {
      "epoch": 0.36720295964897187,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009868938068300149,
      "loss": 1.5064,
      "step": 4268
    },
    {
      "epoch": 0.36728899595629355,
      "grad_norm": 1.625,
      "learning_rate": 0.000986887655454161,
      "loss": 1.2967,
      "step": 4269
    },
    {
      "epoch": 0.36737503226361523,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009868815026542557,
      "loss": 1.8422,
      "step": 4270
    },
    {
      "epoch": 0.3674610685709369,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009868753484303164,
      "loss": 0.8496,
      "step": 4271
    },
    {
      "epoch": 0.36754710487825865,
      "grad_norm": 1.125,
      "learning_rate": 0.0009868691927823617,
      "loss": 1.6631,
      "step": 4272
    },
    {
      "epoch": 0.3676331411855803,
      "grad_norm": 1.40625,
      "learning_rate": 0.000986863035710409,
      "loss": 1.2406,
      "step": 4273
    },
    {
      "epoch": 0.367719177492902,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009868568772144768,
      "loss": 1.0789,
      "step": 4274
    },
    {
      "epoch": 0.3678052138002237,
      "grad_norm": 0.984375,
      "learning_rate": 0.000986850717294583,
      "loss": 0.9349,
      "step": 4275
    },
    {
      "epoch": 0.36789125010754536,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009868445559507453,
      "loss": 1.005,
      "step": 4276
    },
    {
      "epoch": 0.3679772864148671,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009868383931829821,
      "loss": 1.3486,
      "step": 4277
    },
    {
      "epoch": 0.3680633227221888,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009868322289913111,
      "loss": 1.5649,
      "step": 4278
    },
    {
      "epoch": 0.36814935902951046,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000986826063375751,
      "loss": 1.2543,
      "step": 4279
    },
    {
      "epoch": 0.36823539533683214,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009868198963363189,
      "loss": 1.2857,
      "step": 4280
    },
    {
      "epoch": 0.3683214316441538,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009868137278730333,
      "loss": 1.7499,
      "step": 4281
    },
    {
      "epoch": 0.3684074679514755,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009868075579859124,
      "loss": 1.6586,
      "step": 4282
    },
    {
      "epoch": 0.36849350425879723,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009868013866749742,
      "loss": 1.42,
      "step": 4283
    },
    {
      "epoch": 0.3685795405661189,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009867952139402363,
      "loss": 2.0979,
      "step": 4284
    },
    {
      "epoch": 0.3686655768734406,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009867890397817173,
      "loss": 1.1688,
      "step": 4285
    },
    {
      "epoch": 0.36875161318076227,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009867828641994347,
      "loss": 1.6857,
      "step": 4286
    },
    {
      "epoch": 0.36883764948808395,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009867766871934072,
      "loss": 1.938,
      "step": 4287
    },
    {
      "epoch": 0.3689236857954057,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009867705087636524,
      "loss": 1.7374,
      "step": 4288
    },
    {
      "epoch": 0.36900972210272737,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009867643289101888,
      "loss": 2.2586,
      "step": 4289
    },
    {
      "epoch": 0.36909575841004905,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000986758147633034,
      "loss": 1.2922,
      "step": 4290
    },
    {
      "epoch": 0.3691817947173707,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009867519649322063,
      "loss": 1.3326,
      "step": 4291
    },
    {
      "epoch": 0.3692678310246924,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009867457808077238,
      "loss": 2.7334,
      "step": 4292
    },
    {
      "epoch": 0.3693538673320141,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009867395952596046,
      "loss": 1.658,
      "step": 4293
    },
    {
      "epoch": 0.3694399036393358,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009867334082878667,
      "loss": 2.0891,
      "step": 4294
    },
    {
      "epoch": 0.3695259399466575,
      "grad_norm": 1.375,
      "learning_rate": 0.0009867272198925282,
      "loss": 1.5565,
      "step": 4295
    },
    {
      "epoch": 0.3696119762539792,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009867210300736073,
      "loss": 1.1228,
      "step": 4296
    },
    {
      "epoch": 0.36969801256130086,
      "grad_norm": 1.34375,
      "learning_rate": 0.000986714838831122,
      "loss": 1.3203,
      "step": 4297
    },
    {
      "epoch": 0.36978404886862254,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009867086461650906,
      "loss": 1.1661,
      "step": 4298
    },
    {
      "epoch": 0.3698700851759443,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000986702452075531,
      "loss": 1.3501,
      "step": 4299
    },
    {
      "epoch": 0.36995612148326595,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009866962565624614,
      "loss": 1.1141,
      "step": 4300
    },
    {
      "epoch": 0.37004215779058763,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009866900596258998,
      "loss": 1.2755,
      "step": 4301
    },
    {
      "epoch": 0.3701281940979093,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009866838612658643,
      "loss": 1.3718,
      "step": 4302
    },
    {
      "epoch": 0.370214230405231,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009866776614823737,
      "loss": 1.8768,
      "step": 4303
    },
    {
      "epoch": 0.3703002667125527,
      "grad_norm": 1.8828125,
      "learning_rate": 0.000986671460275445,
      "loss": 1.4429,
      "step": 4304
    },
    {
      "epoch": 0.3703863030198744,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009866652576450972,
      "loss": 1.0865,
      "step": 4305
    },
    {
      "epoch": 0.3704723393271961,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009866590535913481,
      "loss": 1.2106,
      "step": 4306
    },
    {
      "epoch": 0.37055837563451777,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000986652848114216,
      "loss": 0.995,
      "step": 4307
    },
    {
      "epoch": 0.37064441194183945,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009866466412137192,
      "loss": 1.568,
      "step": 4308
    },
    {
      "epoch": 0.3707304482491611,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009866404328898752,
      "loss": 1.4687,
      "step": 4309
    },
    {
      "epoch": 0.37081648455648286,
      "grad_norm": 1.375,
      "learning_rate": 0.0009866342231427028,
      "loss": 1.3457,
      "step": 4310
    },
    {
      "epoch": 0.37090252086380454,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009866280119722197,
      "loss": 1.5308,
      "step": 4311
    },
    {
      "epoch": 0.3709885571711262,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009866217993784445,
      "loss": 0.9641,
      "step": 4312
    },
    {
      "epoch": 0.3710745934784479,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009866155853613954,
      "loss": 0.7405,
      "step": 4313
    },
    {
      "epoch": 0.3711606297857696,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009866093699210901,
      "loss": 1.0951,
      "step": 4314
    },
    {
      "epoch": 0.37124666609309126,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000986603153057547,
      "loss": 1.1956,
      "step": 4315
    },
    {
      "epoch": 0.371332702400413,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009865969347707844,
      "loss": 1.6953,
      "step": 4316
    },
    {
      "epoch": 0.3714187387077347,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009865907150608203,
      "loss": 1.1492,
      "step": 4317
    },
    {
      "epoch": 0.37150477501505635,
      "grad_norm": 1.328125,
      "learning_rate": 0.000986584493927673,
      "loss": 1.057,
      "step": 4318
    },
    {
      "epoch": 0.37159081132237803,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009865782713713608,
      "loss": 0.872,
      "step": 4319
    },
    {
      "epoch": 0.3716768476296997,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009865720473919018,
      "loss": 1.1363,
      "step": 4320
    },
    {
      "epoch": 0.37176288393702145,
      "grad_norm": 1.8125,
      "learning_rate": 0.000986565821989314,
      "loss": 1.1274,
      "step": 4321
    },
    {
      "epoch": 0.37184892024434313,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009865595951636162,
      "loss": 1.2502,
      "step": 4322
    },
    {
      "epoch": 0.3719349565516648,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009865533669148258,
      "loss": 1.1755,
      "step": 4323
    },
    {
      "epoch": 0.3720209928589865,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009865471372429617,
      "loss": 1.9194,
      "step": 4324
    },
    {
      "epoch": 0.37210702916630817,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009865409061480417,
      "loss": 1.5974,
      "step": 4325
    },
    {
      "epoch": 0.37219306547362985,
      "grad_norm": 2.21875,
      "learning_rate": 0.0009865346736300843,
      "loss": 1.4525,
      "step": 4326
    },
    {
      "epoch": 0.3722791017809516,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009865284396891074,
      "loss": 0.946,
      "step": 4327
    },
    {
      "epoch": 0.37236513808827326,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009865222043251296,
      "loss": 1.4511,
      "step": 4328
    },
    {
      "epoch": 0.37245117439559494,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000986515967538169,
      "loss": 2.3087,
      "step": 4329
    },
    {
      "epoch": 0.3725372107029166,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009865097293282438,
      "loss": 2.0081,
      "step": 4330
    },
    {
      "epoch": 0.3726232470102383,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009865034896953723,
      "loss": 1.0651,
      "step": 4331
    },
    {
      "epoch": 0.37270928331756004,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009864972486395725,
      "loss": 0.951,
      "step": 4332
    },
    {
      "epoch": 0.3727953196248817,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000986491006160863,
      "loss": 1.2619,
      "step": 4333
    },
    {
      "epoch": 0.3728813559322034,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000986484762259262,
      "loss": 1.9148,
      "step": 4334
    },
    {
      "epoch": 0.3729673922395251,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009864785169347876,
      "loss": 1.855,
      "step": 4335
    },
    {
      "epoch": 0.37305342854684675,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009864722701874583,
      "loss": 0.8462,
      "step": 4336
    },
    {
      "epoch": 0.37313946485416843,
      "grad_norm": 1.640625,
      "learning_rate": 0.000986466022017292,
      "loss": 1.028,
      "step": 4337
    },
    {
      "epoch": 0.37322550116149017,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009864597724243073,
      "loss": 1.3436,
      "step": 4338
    },
    {
      "epoch": 0.37331153746881185,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009864535214085225,
      "loss": 1.085,
      "step": 4339
    },
    {
      "epoch": 0.37339757377613353,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009864472689699556,
      "loss": 1.6106,
      "step": 4340
    },
    {
      "epoch": 0.3734836100834552,
      "grad_norm": 1.25,
      "learning_rate": 0.0009864410151086249,
      "loss": 1.526,
      "step": 4341
    },
    {
      "epoch": 0.3735696463907769,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009864347598245492,
      "loss": 2.0709,
      "step": 4342
    },
    {
      "epoch": 0.3736556826980986,
      "grad_norm": 2.640625,
      "learning_rate": 0.0009864285031177462,
      "loss": 1.6954,
      "step": 4343
    },
    {
      "epoch": 0.3737417190054203,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009864222449882344,
      "loss": 0.6249,
      "step": 4344
    },
    {
      "epoch": 0.373827755312742,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000986415985436032,
      "loss": 1.1901,
      "step": 4345
    },
    {
      "epoch": 0.37391379162006366,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009864097244611579,
      "loss": 1.4041,
      "step": 4346
    },
    {
      "epoch": 0.37399982792738534,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009864034620636298,
      "loss": 1.5277,
      "step": 4347
    },
    {
      "epoch": 0.374085864234707,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009863971982434659,
      "loss": 1.2744,
      "step": 4348
    },
    {
      "epoch": 0.37417190054202876,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009863909330006851,
      "loss": 1.6997,
      "step": 4349
    },
    {
      "epoch": 0.37425793684935044,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009863846663353054,
      "loss": 1.0832,
      "step": 4350
    },
    {
      "epoch": 0.3743439731566721,
      "grad_norm": 1.25,
      "learning_rate": 0.0009863783982473448,
      "loss": 1.0145,
      "step": 4351
    },
    {
      "epoch": 0.3744300094639938,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009863721287368223,
      "loss": 1.5959,
      "step": 4352
    },
    {
      "epoch": 0.3745160457713155,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009863658578037559,
      "loss": 1.5159,
      "step": 4353
    },
    {
      "epoch": 0.3746020820786372,
      "grad_norm": 2.1875,
      "learning_rate": 0.000986359585448164,
      "loss": 1.7185,
      "step": 4354
    },
    {
      "epoch": 0.3746881183859589,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009863533116700647,
      "loss": 1.0216,
      "step": 4355
    },
    {
      "epoch": 0.37477415469328057,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009863470364694766,
      "loss": 2.0914,
      "step": 4356
    },
    {
      "epoch": 0.37486019100060225,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009863407598464183,
      "loss": 1.4265,
      "step": 4357
    },
    {
      "epoch": 0.37494622730792393,
      "grad_norm": 1.25,
      "learning_rate": 0.0009863344818009075,
      "loss": 1.1231,
      "step": 4358
    },
    {
      "epoch": 0.3750322636152456,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009863282023329632,
      "loss": 0.8617,
      "step": 4359
    },
    {
      "epoch": 0.37511829992256734,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009863219214426033,
      "loss": 1.6732,
      "step": 4360
    },
    {
      "epoch": 0.375204336229889,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009863156391298465,
      "loss": 1.7907,
      "step": 4361
    },
    {
      "epoch": 0.3752903725372107,
      "grad_norm": 1.046875,
      "learning_rate": 0.000986309355394711,
      "loss": 1.2913,
      "step": 4362
    },
    {
      "epoch": 0.3753764088445324,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009863030702372151,
      "loss": 2.2747,
      "step": 4363
    },
    {
      "epoch": 0.37546244515185406,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009862967836573773,
      "loss": 0.7665,
      "step": 4364
    },
    {
      "epoch": 0.3755484814591758,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009862904956552162,
      "loss": 1.2604,
      "step": 4365
    },
    {
      "epoch": 0.3756345177664975,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009862842062307497,
      "loss": 1.2668,
      "step": 4366
    },
    {
      "epoch": 0.37572055407381916,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009862779153839967,
      "loss": 0.5456,
      "step": 4367
    },
    {
      "epoch": 0.37580659038114084,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009862716231149754,
      "loss": 1.226,
      "step": 4368
    },
    {
      "epoch": 0.3758926266884625,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009862653294237039,
      "loss": 1.0595,
      "step": 4369
    },
    {
      "epoch": 0.3759786629957842,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009862590343102011,
      "loss": 1.1495,
      "step": 4370
    },
    {
      "epoch": 0.37606469930310593,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009862527377744851,
      "loss": 1.3474,
      "step": 4371
    },
    {
      "epoch": 0.3761507356104276,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009862464398165745,
      "loss": 2.1145,
      "step": 4372
    },
    {
      "epoch": 0.3762367719177493,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009862401404364875,
      "loss": 1.2439,
      "step": 4373
    },
    {
      "epoch": 0.37632280822507097,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000986233839634243,
      "loss": 1.7669,
      "step": 4374
    },
    {
      "epoch": 0.37640884453239265,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009862275374098587,
      "loss": 1.4123,
      "step": 4375
    },
    {
      "epoch": 0.3764948808397144,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009862212337633535,
      "loss": 1.3357,
      "step": 4376
    },
    {
      "epoch": 0.37658091714703607,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000986214928694746,
      "loss": 1.4249,
      "step": 4377
    },
    {
      "epoch": 0.37666695345435774,
      "grad_norm": 1.125,
      "learning_rate": 0.000986208622204054,
      "loss": 1.0383,
      "step": 4378
    },
    {
      "epoch": 0.3767529897616794,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009862023142912968,
      "loss": 1.8373,
      "step": 4379
    },
    {
      "epoch": 0.3768390260690011,
      "grad_norm": 1.359375,
      "learning_rate": 0.000986196004956492,
      "loss": 1.8363,
      "step": 4380
    },
    {
      "epoch": 0.3769250623763228,
      "grad_norm": 1.5,
      "learning_rate": 0.0009861896941996586,
      "loss": 1.3502,
      "step": 4381
    },
    {
      "epoch": 0.3770110986836445,
      "grad_norm": 1.171875,
      "learning_rate": 0.000986183382020815,
      "loss": 1.5859,
      "step": 4382
    },
    {
      "epoch": 0.3770971349909662,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009861770684199793,
      "loss": 1.0999,
      "step": 4383
    },
    {
      "epoch": 0.3771831712982879,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009861707533971704,
      "loss": 1.7205,
      "step": 4384
    },
    {
      "epoch": 0.37726920760560956,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009861644369524069,
      "loss": 1.3599,
      "step": 4385
    },
    {
      "epoch": 0.37735524391293124,
      "grad_norm": 2.21875,
      "learning_rate": 0.0009861581190857066,
      "loss": 2.13,
      "step": 4386
    },
    {
      "epoch": 0.377441280220253,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009861517997970884,
      "loss": 1.4013,
      "step": 4387
    },
    {
      "epoch": 0.37752731652757465,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009861454790865707,
      "loss": 1.9465,
      "step": 4388
    },
    {
      "epoch": 0.37761335283489633,
      "grad_norm": 1.28125,
      "learning_rate": 0.000986139156954172,
      "loss": 1.6919,
      "step": 4389
    },
    {
      "epoch": 0.377699389142218,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000986132833399911,
      "loss": 0.8733,
      "step": 4390
    },
    {
      "epoch": 0.3777854254495397,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009861265084238058,
      "loss": 1.4805,
      "step": 4391
    },
    {
      "epoch": 0.37787146175686137,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009861201820258753,
      "loss": 2.1053,
      "step": 4392
    },
    {
      "epoch": 0.3779574980641831,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009861138542061377,
      "loss": 1.3433,
      "step": 4393
    },
    {
      "epoch": 0.3780435343715048,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009861075249646117,
      "loss": 0.8963,
      "step": 4394
    },
    {
      "epoch": 0.37812957067882647,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009861011943013156,
      "loss": 1.2034,
      "step": 4395
    },
    {
      "epoch": 0.37821560698614815,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009860948622162683,
      "loss": 1.2064,
      "step": 4396
    },
    {
      "epoch": 0.3783016432934698,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009860885287094878,
      "loss": 1.683,
      "step": 4397
    },
    {
      "epoch": 0.37838767960079156,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009860821937809929,
      "loss": 1.7577,
      "step": 4398
    },
    {
      "epoch": 0.37847371590811324,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009860758574308022,
      "loss": 1.7986,
      "step": 4399
    },
    {
      "epoch": 0.3785597522154349,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009860695196589344,
      "loss": 1.0757,
      "step": 4400
    },
    {
      "epoch": 0.3786457885227566,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009860631804654075,
      "loss": 1.8269,
      "step": 4401
    },
    {
      "epoch": 0.3787318248300783,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009860568398502403,
      "loss": 1.387,
      "step": 4402
    },
    {
      "epoch": 0.37881786113739996,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009860504978134515,
      "loss": 1.4234,
      "step": 4403
    },
    {
      "epoch": 0.3789038974447217,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009860441543550595,
      "loss": 1.2281,
      "step": 4404
    },
    {
      "epoch": 0.3789899337520434,
      "grad_norm": 3.0625,
      "learning_rate": 0.0009860378094750828,
      "loss": 1.3049,
      "step": 4405
    },
    {
      "epoch": 0.37907597005936505,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009860314631735402,
      "loss": 0.9156,
      "step": 4406
    },
    {
      "epoch": 0.37916200636668673,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00098602511545045,
      "loss": 1.2717,
      "step": 4407
    },
    {
      "epoch": 0.3792480426740084,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009860187663058308,
      "loss": 1.4856,
      "step": 4408
    },
    {
      "epoch": 0.37933407898133015,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009860124157397014,
      "loss": 0.9619,
      "step": 4409
    },
    {
      "epoch": 0.3794201152886518,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00098600606375208,
      "loss": 1.6535,
      "step": 4410
    },
    {
      "epoch": 0.3795061515959735,
      "grad_norm": 1.625,
      "learning_rate": 0.0009859997103429856,
      "loss": 1.6427,
      "step": 4411
    },
    {
      "epoch": 0.3795921879032952,
      "grad_norm": 2.46875,
      "learning_rate": 0.0009859933555124364,
      "loss": 1.2107,
      "step": 4412
    },
    {
      "epoch": 0.37967822421061687,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009859869992604514,
      "loss": 0.9712,
      "step": 4413
    },
    {
      "epoch": 0.37976426051793855,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009859806415870487,
      "loss": 2.097,
      "step": 4414
    },
    {
      "epoch": 0.3798502968252603,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009859742824922472,
      "loss": 1.2795,
      "step": 4415
    },
    {
      "epoch": 0.37993633313258196,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009859679219760655,
      "loss": 1.8872,
      "step": 4416
    },
    {
      "epoch": 0.38002236943990364,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000985961560038522,
      "loss": 1.1421,
      "step": 4417
    },
    {
      "epoch": 0.3801084057472253,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009859551966796356,
      "loss": 1.1013,
      "step": 4418
    },
    {
      "epoch": 0.380194442054547,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009859488318994247,
      "loss": 0.8917,
      "step": 4419
    },
    {
      "epoch": 0.38028047836186873,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009859424656979078,
      "loss": 1.897,
      "step": 4420
    },
    {
      "epoch": 0.3803665146691904,
      "grad_norm": 1.546875,
      "learning_rate": 0.000985936098075104,
      "loss": 1.7187,
      "step": 4421
    },
    {
      "epoch": 0.3804525509765121,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009859297290310317,
      "loss": 1.3899,
      "step": 4422
    },
    {
      "epoch": 0.3805385872838338,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009859233585657091,
      "loss": 0.9971,
      "step": 4423
    },
    {
      "epoch": 0.38062462359115545,
      "grad_norm": 3.03125,
      "learning_rate": 0.0009859169866791554,
      "loss": 1.9747,
      "step": 4424
    },
    {
      "epoch": 0.38071065989847713,
      "grad_norm": 12.5,
      "learning_rate": 0.0009859106133713888,
      "loss": 1.4943,
      "step": 4425
    },
    {
      "epoch": 0.38079669620579887,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009859042386424284,
      "loss": 1.518,
      "step": 4426
    },
    {
      "epoch": 0.38088273251312055,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009858978624922925,
      "loss": 2.1987,
      "step": 4427
    },
    {
      "epoch": 0.3809687688204422,
      "grad_norm": 1.5,
      "learning_rate": 0.0009858914849209999,
      "loss": 1.6402,
      "step": 4428
    },
    {
      "epoch": 0.3810548051277639,
      "grad_norm": 2.203125,
      "learning_rate": 0.0009858851059285692,
      "loss": 1.2568,
      "step": 4429
    },
    {
      "epoch": 0.3811408414350856,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000985878725515019,
      "loss": 1.4275,
      "step": 4430
    },
    {
      "epoch": 0.3812268777424073,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009858723436803683,
      "loss": 0.6818,
      "step": 4431
    },
    {
      "epoch": 0.381312914049729,
      "grad_norm": 4.3125,
      "learning_rate": 0.0009858659604246353,
      "loss": 1.7566,
      "step": 4432
    },
    {
      "epoch": 0.3813989503570507,
      "grad_norm": 3.0625,
      "learning_rate": 0.0009858595757478387,
      "loss": 1.9252,
      "step": 4433
    },
    {
      "epoch": 0.38148498666437236,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009858531896499974,
      "loss": 1.6415,
      "step": 4434
    },
    {
      "epoch": 0.38157102297169404,
      "grad_norm": 1.5,
      "learning_rate": 0.0009858468021311303,
      "loss": 2.0384,
      "step": 4435
    },
    {
      "epoch": 0.3816570592790157,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009858404131912555,
      "loss": 1.3666,
      "step": 4436
    },
    {
      "epoch": 0.38174309558633746,
      "grad_norm": 1.359375,
      "learning_rate": 0.000985834022830392,
      "loss": 1.6753,
      "step": 4437
    },
    {
      "epoch": 0.38182913189365913,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009858276310485586,
      "loss": 0.9856,
      "step": 4438
    },
    {
      "epoch": 0.3819151682009808,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009858212378457737,
      "loss": 1.455,
      "step": 4439
    },
    {
      "epoch": 0.3820012045083025,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009858148432220562,
      "loss": 1.1808,
      "step": 4440
    },
    {
      "epoch": 0.3820872408156242,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000985808447177425,
      "loss": 1.6267,
      "step": 4441
    },
    {
      "epoch": 0.3821732771229459,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009858020497118983,
      "loss": 0.9568,
      "step": 4442
    },
    {
      "epoch": 0.3822593134302676,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009857956508254953,
      "loss": 1.2057,
      "step": 4443
    },
    {
      "epoch": 0.38234534973758927,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009857892505182343,
      "loss": 1.2502,
      "step": 4444
    },
    {
      "epoch": 0.38243138604491095,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009857828487901342,
      "loss": 1.5717,
      "step": 4445
    },
    {
      "epoch": 0.38251742235223263,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000985776445641214,
      "loss": 0.9762,
      "step": 4446
    },
    {
      "epoch": 0.3826034586595543,
      "grad_norm": 1.8515625,
      "learning_rate": 0.000985770041071492,
      "loss": 1.6495,
      "step": 4447
    },
    {
      "epoch": 0.38268949496687604,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009857636350809873,
      "loss": 1.6101,
      "step": 4448
    },
    {
      "epoch": 0.3827755312741977,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009857572276697182,
      "loss": 1.2679,
      "step": 4449
    },
    {
      "epoch": 0.3828615675815194,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009857508188377039,
      "loss": 1.0962,
      "step": 4450
    },
    {
      "epoch": 0.3829476038888411,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009857444085849627,
      "loss": 1.6978,
      "step": 4451
    },
    {
      "epoch": 0.38303364019616276,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009857379969115136,
      "loss": 1.1699,
      "step": 4452
    },
    {
      "epoch": 0.3831196765034845,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009857315838173754,
      "loss": 1.3198,
      "step": 4453
    },
    {
      "epoch": 0.3832057128108062,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000985725169302567,
      "loss": 1.6003,
      "step": 4454
    },
    {
      "epoch": 0.38329174911812786,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009857187533671068,
      "loss": 1.7026,
      "step": 4455
    },
    {
      "epoch": 0.38337778542544954,
      "grad_norm": 2.203125,
      "learning_rate": 0.0009857123360110136,
      "loss": 1.5355,
      "step": 4456
    },
    {
      "epoch": 0.3834638217327712,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009857059172343064,
      "loss": 1.671,
      "step": 4457
    },
    {
      "epoch": 0.3835498580400929,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009856994970370036,
      "loss": 1.992,
      "step": 4458
    },
    {
      "epoch": 0.38363589434741463,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009856930754191247,
      "loss": 1.0721,
      "step": 4459
    },
    {
      "epoch": 0.3837219306547363,
      "grad_norm": 1.5,
      "learning_rate": 0.0009856866523806876,
      "loss": 1.4115,
      "step": 4460
    },
    {
      "epoch": 0.383807966962058,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009856802279217117,
      "loss": 1.2919,
      "step": 4461
    },
    {
      "epoch": 0.38389400326937967,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009856738020422155,
      "loss": 1.7046,
      "step": 4462
    },
    {
      "epoch": 0.38398003957670135,
      "grad_norm": 2.203125,
      "learning_rate": 0.000985667374742218,
      "loss": 1.9464,
      "step": 4463
    },
    {
      "epoch": 0.3840660758840231,
      "grad_norm": 1.5,
      "learning_rate": 0.000985660946021738,
      "loss": 1.2383,
      "step": 4464
    },
    {
      "epoch": 0.38415211219134476,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000985654515880794,
      "loss": 1.2653,
      "step": 4465
    },
    {
      "epoch": 0.38423814849866644,
      "grad_norm": 1.7421875,
      "learning_rate": 0.000985648084319405,
      "loss": 1.5063,
      "step": 4466
    },
    {
      "epoch": 0.3843241848059881,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009856416513375898,
      "loss": 1.3197,
      "step": 4467
    },
    {
      "epoch": 0.3844102211133098,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009856352169353671,
      "loss": 1.198,
      "step": 4468
    },
    {
      "epoch": 0.3844962574206315,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000985628781112756,
      "loss": 1.1306,
      "step": 4469
    },
    {
      "epoch": 0.3845822937279532,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009856223438697751,
      "loss": 0.7417,
      "step": 4470
    },
    {
      "epoch": 0.3846683300352749,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009856159052064432,
      "loss": 1.2365,
      "step": 4471
    },
    {
      "epoch": 0.3847543663425966,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009856094651227794,
      "loss": 1.7518,
      "step": 4472
    },
    {
      "epoch": 0.38484040264991826,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009856030236188025,
      "loss": 1.1674,
      "step": 4473
    },
    {
      "epoch": 0.38492643895723994,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000985596580694531,
      "loss": 1.0508,
      "step": 4474
    },
    {
      "epoch": 0.38501247526456167,
      "grad_norm": 1.03125,
      "learning_rate": 0.000985590136349984,
      "loss": 1.1649,
      "step": 4475
    },
    {
      "epoch": 0.38509851157188335,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009855836905851801,
      "loss": 1.9405,
      "step": 4476
    },
    {
      "epoch": 0.38518454787920503,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009855772434001387,
      "loss": 0.9562,
      "step": 4477
    },
    {
      "epoch": 0.3852705841865267,
      "grad_norm": 1.546875,
      "learning_rate": 0.000985570794794878,
      "loss": 1.4345,
      "step": 4478
    },
    {
      "epoch": 0.3853566204938484,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009855643447694172,
      "loss": 1.3666,
      "step": 4479
    },
    {
      "epoch": 0.38544265680117007,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009855578933237752,
      "loss": 0.8558,
      "step": 4480
    },
    {
      "epoch": 0.3855286931084918,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009855514404579708,
      "loss": 1.1409,
      "step": 4481
    },
    {
      "epoch": 0.3856147294158135,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009855449861720228,
      "loss": 1.3511,
      "step": 4482
    },
    {
      "epoch": 0.38570076572313516,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009855385304659502,
      "loss": 1.9312,
      "step": 4483
    },
    {
      "epoch": 0.38578680203045684,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000985532073339772,
      "loss": 1.1976,
      "step": 4484
    },
    {
      "epoch": 0.3858728383377785,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009855256147935068,
      "loss": 1.8857,
      "step": 4485
    },
    {
      "epoch": 0.38595887464510026,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009855191548271737,
      "loss": 1.0206,
      "step": 4486
    },
    {
      "epoch": 0.38604491095242194,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009855126934407911,
      "loss": 2.3846,
      "step": 4487
    },
    {
      "epoch": 0.3861309472597436,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009855062306343787,
      "loss": 0.8646,
      "step": 4488
    },
    {
      "epoch": 0.3862169835670653,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009854997664079548,
      "loss": 2.1005,
      "step": 4489
    },
    {
      "epoch": 0.386303019874387,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009854933007615386,
      "loss": 1.5961,
      "step": 4490
    },
    {
      "epoch": 0.38638905618170866,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000985486833695149,
      "loss": 1.4876,
      "step": 4491
    },
    {
      "epoch": 0.3864750924890304,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009854803652088048,
      "loss": 2.0095,
      "step": 4492
    },
    {
      "epoch": 0.38656112879635207,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009854738953025248,
      "loss": 1.8446,
      "step": 4493
    },
    {
      "epoch": 0.38664716510367375,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000985467423976328,
      "loss": 1.6158,
      "step": 4494
    },
    {
      "epoch": 0.38673320141099543,
      "grad_norm": 1.25,
      "learning_rate": 0.0009854609512302338,
      "loss": 1.2312,
      "step": 4495
    },
    {
      "epoch": 0.3868192377183171,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009854544770642605,
      "loss": 1.8268,
      "step": 4496
    },
    {
      "epoch": 0.38690527402563885,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000985448001478427,
      "loss": 1.2732,
      "step": 4497
    },
    {
      "epoch": 0.3869913103329605,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009854415244727529,
      "loss": 1.9354,
      "step": 4498
    },
    {
      "epoch": 0.3870773466402822,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009854350460472567,
      "loss": 1.3904,
      "step": 4499
    },
    {
      "epoch": 0.3871633829476039,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009854285662019573,
      "loss": 1.7359,
      "step": 4500
    },
    {
      "epoch": 0.38724941925492556,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009854220849368738,
      "loss": 1.2051,
      "step": 4501
    },
    {
      "epoch": 0.38733545556224724,
      "grad_norm": 1.515625,
      "learning_rate": 0.000985415602252025,
      "loss": 1.4805,
      "step": 4502
    },
    {
      "epoch": 0.387421491869569,
      "grad_norm": 1.234375,
      "learning_rate": 0.00098540911814743,
      "loss": 1.0805,
      "step": 4503
    },
    {
      "epoch": 0.38750752817689066,
      "grad_norm": 1.375,
      "learning_rate": 0.000985402632623108,
      "loss": 1.4967,
      "step": 4504
    },
    {
      "epoch": 0.38759356448421234,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009853961456790775,
      "loss": 1.1964,
      "step": 4505
    },
    {
      "epoch": 0.387679600791534,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009853896573153577,
      "loss": 1.0588,
      "step": 4506
    },
    {
      "epoch": 0.3877656370988557,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009853831675319675,
      "loss": 1.4389,
      "step": 4507
    },
    {
      "epoch": 0.38785167340617743,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000985376676328926,
      "loss": 0.8796,
      "step": 4508
    },
    {
      "epoch": 0.3879377097134991,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000985370183706252,
      "loss": 1.226,
      "step": 4509
    },
    {
      "epoch": 0.3880237460208208,
      "grad_norm": 1.25,
      "learning_rate": 0.0009853636896639648,
      "loss": 1.0927,
      "step": 4510
    },
    {
      "epoch": 0.38810978232814247,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000985357194202083,
      "loss": 1.7737,
      "step": 4511
    },
    {
      "epoch": 0.38819581863546415,
      "grad_norm": 1.515625,
      "learning_rate": 0.000985350697320626,
      "loss": 2.0529,
      "step": 4512
    },
    {
      "epoch": 0.38828185494278583,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009853441990196126,
      "loss": 1.046,
      "step": 4513
    },
    {
      "epoch": 0.38836789125010757,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009853376992990618,
      "loss": 1.3727,
      "step": 4514
    },
    {
      "epoch": 0.38845392755742925,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009853311981589926,
      "loss": 1.5618,
      "step": 4515
    },
    {
      "epoch": 0.3885399638647509,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009853246955994239,
      "loss": 0.7603,
      "step": 4516
    },
    {
      "epoch": 0.3886260001720726,
      "grad_norm": 1.4375,
      "learning_rate": 0.000985318191620375,
      "loss": 0.8614,
      "step": 4517
    },
    {
      "epoch": 0.3887120364793943,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009853116862218647,
      "loss": 1.2937,
      "step": 4518
    },
    {
      "epoch": 0.388798072786716,
      "grad_norm": 1.53125,
      "learning_rate": 0.000985305179403912,
      "loss": 1.4426,
      "step": 4519
    },
    {
      "epoch": 0.3888841090940377,
      "grad_norm": 1.625,
      "learning_rate": 0.0009852986711665363,
      "loss": 1.2244,
      "step": 4520
    },
    {
      "epoch": 0.3889701454013594,
      "grad_norm": 1.0,
      "learning_rate": 0.0009852921615097563,
      "loss": 1.5411,
      "step": 4521
    },
    {
      "epoch": 0.38905618170868106,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000985285650433591,
      "loss": 1.075,
      "step": 4522
    },
    {
      "epoch": 0.38914221801600274,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009852791379380598,
      "loss": 1.4318,
      "step": 4523
    },
    {
      "epoch": 0.3892282543233244,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009852726240231811,
      "loss": 1.5922,
      "step": 4524
    },
    {
      "epoch": 0.38931429063064615,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009852661086889748,
      "loss": 1.505,
      "step": 4525
    },
    {
      "epoch": 0.38940032693796783,
      "grad_norm": 1.421875,
      "learning_rate": 0.000985259591935459,
      "loss": 1.8136,
      "step": 4526
    },
    {
      "epoch": 0.3894863632452895,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009852530737626537,
      "loss": 1.1064,
      "step": 4527
    },
    {
      "epoch": 0.3895723995526112,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009852465541705772,
      "loss": 1.6809,
      "step": 4528
    },
    {
      "epoch": 0.3896584358599329,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009852400331592493,
      "loss": 1.2455,
      "step": 4529
    },
    {
      "epoch": 0.3897444721672546,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009852335107286886,
      "loss": 2.2121,
      "step": 4530
    },
    {
      "epoch": 0.3898305084745763,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000985226986878914,
      "loss": 1.1369,
      "step": 4531
    },
    {
      "epoch": 0.38991654478189797,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000985220461609945,
      "loss": 1.2847,
      "step": 4532
    },
    {
      "epoch": 0.39000258108921965,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009852139349218005,
      "loss": 2.0298,
      "step": 4533
    },
    {
      "epoch": 0.3900886173965413,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009852074068144995,
      "loss": 1.697,
      "step": 4534
    },
    {
      "epoch": 0.390174653703863,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009852008772880613,
      "loss": 1.7382,
      "step": 4535
    },
    {
      "epoch": 0.39026069001118474,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000985194346342505,
      "loss": 0.7595,
      "step": 4536
    },
    {
      "epoch": 0.3903467263185064,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009851878139778493,
      "loss": 1.43,
      "step": 4537
    },
    {
      "epoch": 0.3904327626258281,
      "grad_norm": 1.375,
      "learning_rate": 0.000985181280194114,
      "loss": 1.5416,
      "step": 4538
    },
    {
      "epoch": 0.3905187989331498,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009851747449913175,
      "loss": 1.2552,
      "step": 4539
    },
    {
      "epoch": 0.39060483524047146,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009851682083694795,
      "loss": 1.82,
      "step": 4540
    },
    {
      "epoch": 0.3906908715477932,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009851616703286186,
      "loss": 1.6793,
      "step": 4541
    },
    {
      "epoch": 0.3907769078551149,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009851551308687543,
      "loss": 1.8129,
      "step": 4542
    },
    {
      "epoch": 0.39086294416243655,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009851485899899056,
      "loss": 1.1392,
      "step": 4543
    },
    {
      "epoch": 0.39094898046975823,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009851420476920916,
      "loss": 1.1548,
      "step": 4544
    },
    {
      "epoch": 0.3910350167770799,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009851355039753316,
      "loss": 0.84,
      "step": 4545
    },
    {
      "epoch": 0.3911210530844016,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009851289588396444,
      "loss": 1.6459,
      "step": 4546
    },
    {
      "epoch": 0.39120708939172333,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009851224122850494,
      "loss": 0.874,
      "step": 4547
    },
    {
      "epoch": 0.391293125699045,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009851158643115656,
      "loss": 0.7778,
      "step": 4548
    },
    {
      "epoch": 0.3913791620063667,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009851093149192125,
      "loss": 1.0014,
      "step": 4549
    },
    {
      "epoch": 0.39146519831368837,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009851027641080087,
      "loss": 1.2257,
      "step": 4550
    },
    {
      "epoch": 0.39155123462101005,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009850962118779741,
      "loss": 1.6212,
      "step": 4551
    },
    {
      "epoch": 0.3916372709283318,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009850896582291272,
      "loss": 0.9031,
      "step": 4552
    },
    {
      "epoch": 0.39172330723565346,
      "grad_norm": 1.625,
      "learning_rate": 0.0009850831031614873,
      "loss": 1.0311,
      "step": 4553
    },
    {
      "epoch": 0.39180934354297514,
      "grad_norm": 1.25,
      "learning_rate": 0.0009850765466750736,
      "loss": 0.8475,
      "step": 4554
    },
    {
      "epoch": 0.3918953798502968,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009850699887699055,
      "loss": 2.188,
      "step": 4555
    },
    {
      "epoch": 0.3919814161576185,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000985063429446002,
      "loss": 1.2625,
      "step": 4556
    },
    {
      "epoch": 0.3920674524649402,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009850568687033822,
      "loss": 0.9809,
      "step": 4557
    },
    {
      "epoch": 0.3921534887722619,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009850503065420655,
      "loss": 1.0365,
      "step": 4558
    },
    {
      "epoch": 0.3922395250795836,
      "grad_norm": 1.328125,
      "learning_rate": 0.000985043742962071,
      "loss": 0.8954,
      "step": 4559
    },
    {
      "epoch": 0.3923255613869053,
      "grad_norm": 2.078125,
      "learning_rate": 0.000985037177963418,
      "loss": 1.2983,
      "step": 4560
    },
    {
      "epoch": 0.39241159769422695,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009850306115461254,
      "loss": 1.4519,
      "step": 4561
    },
    {
      "epoch": 0.39249763400154863,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009850240437102127,
      "loss": 1.3781,
      "step": 4562
    },
    {
      "epoch": 0.39258367030887037,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009850174744556988,
      "loss": 1.194,
      "step": 4563
    },
    {
      "epoch": 0.39266970661619205,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009850109037826034,
      "loss": 1.6247,
      "step": 4564
    },
    {
      "epoch": 0.39275574292351373,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009850043316909454,
      "loss": 2.1398,
      "step": 4565
    },
    {
      "epoch": 0.3928417792308354,
      "grad_norm": 1.21875,
      "learning_rate": 0.000984997758180744,
      "loss": 0.8938,
      "step": 4566
    },
    {
      "epoch": 0.3929278155381571,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009849911832520183,
      "loss": 1.236,
      "step": 4567
    },
    {
      "epoch": 0.39301385184547877,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009849846069047879,
      "loss": 2.1052,
      "step": 4568
    },
    {
      "epoch": 0.3930998881528005,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009849780291390716,
      "loss": 1.6642,
      "step": 4569
    },
    {
      "epoch": 0.3931859244601222,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009849714499548894,
      "loss": 1.614,
      "step": 4570
    },
    {
      "epoch": 0.39327196076744386,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009849648693522597,
      "loss": 1.0798,
      "step": 4571
    },
    {
      "epoch": 0.39335799707476554,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000984958287331202,
      "loss": 1.521,
      "step": 4572
    },
    {
      "epoch": 0.3934440333820872,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009849517038917359,
      "loss": 0.6914,
      "step": 4573
    },
    {
      "epoch": 0.39353006968940896,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009849451190338803,
      "loss": 1.8945,
      "step": 4574
    },
    {
      "epoch": 0.39361610599673064,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009849385327576543,
      "loss": 0.8832,
      "step": 4575
    },
    {
      "epoch": 0.3937021423040523,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009849319450630776,
      "loss": 1.4093,
      "step": 4576
    },
    {
      "epoch": 0.393788178611374,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009849253559501693,
      "loss": 1.7978,
      "step": 4577
    },
    {
      "epoch": 0.3938742149186957,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009849187654189487,
      "loss": 0.9112,
      "step": 4578
    },
    {
      "epoch": 0.39396025122601736,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000984912173469435,
      "loss": 1.0491,
      "step": 4579
    },
    {
      "epoch": 0.3940462875333391,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009849055801016472,
      "loss": 1.6358,
      "step": 4580
    },
    {
      "epoch": 0.39413232384066077,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009848989853156053,
      "loss": 1.595,
      "step": 4581
    },
    {
      "epoch": 0.39421836014798245,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009848923891113278,
      "loss": 1.3563,
      "step": 4582
    },
    {
      "epoch": 0.39430439645530413,
      "grad_norm": 4.09375,
      "learning_rate": 0.0009848857914888346,
      "loss": 2.1253,
      "step": 4583
    },
    {
      "epoch": 0.3943904327626258,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009848791924481448,
      "loss": 0.9778,
      "step": 4584
    },
    {
      "epoch": 0.39447646906994754,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009848725919892774,
      "loss": 1.1732,
      "step": 4585
    },
    {
      "epoch": 0.3945625053772692,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009848659901122523,
      "loss": 1.3988,
      "step": 4586
    },
    {
      "epoch": 0.3946485416845909,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009848593868170884,
      "loss": 1.6389,
      "step": 4587
    },
    {
      "epoch": 0.3947345779919126,
      "grad_norm": 1.625,
      "learning_rate": 0.000984852782103805,
      "loss": 1.1924,
      "step": 4588
    },
    {
      "epoch": 0.39482061429923426,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009848461759724215,
      "loss": 1.544,
      "step": 4589
    },
    {
      "epoch": 0.39490665060655594,
      "grad_norm": 1.765625,
      "learning_rate": 0.000984839568422957,
      "loss": 2.2375,
      "step": 4590
    },
    {
      "epoch": 0.3949926869138777,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009848329594554317,
      "loss": 1.3029,
      "step": 4591
    },
    {
      "epoch": 0.39507872322119936,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009848263490698638,
      "loss": 1.2088,
      "step": 4592
    },
    {
      "epoch": 0.39516475952852104,
      "grad_norm": 1.125,
      "learning_rate": 0.0009848197372662733,
      "loss": 1.573,
      "step": 4593
    },
    {
      "epoch": 0.3952507958358427,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009848131240446792,
      "loss": 1.9083,
      "step": 4594
    },
    {
      "epoch": 0.3953368321431644,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009848065094051012,
      "loss": 1.1405,
      "step": 4595
    },
    {
      "epoch": 0.39542286845048613,
      "grad_norm": 2.0,
      "learning_rate": 0.0009847998933475584,
      "loss": 2.4183,
      "step": 4596
    },
    {
      "epoch": 0.3955089047578078,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00098479327587207,
      "loss": 0.8913,
      "step": 4597
    },
    {
      "epoch": 0.3955949410651295,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009847866569786558,
      "loss": 1.8158,
      "step": 4598
    },
    {
      "epoch": 0.39568097737245117,
      "grad_norm": 1.375,
      "learning_rate": 0.0009847800366673347,
      "loss": 1.4765,
      "step": 4599
    },
    {
      "epoch": 0.39576701367977285,
      "grad_norm": 1.5,
      "learning_rate": 0.0009847734149381266,
      "loss": 1.5881,
      "step": 4600
    },
    {
      "epoch": 0.39585304998709453,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009847667917910502,
      "loss": 1.0123,
      "step": 4601
    },
    {
      "epoch": 0.39593908629441626,
      "grad_norm": 2.640625,
      "learning_rate": 0.0009847601672261254,
      "loss": 1.2143,
      "step": 4602
    },
    {
      "epoch": 0.39602512260173794,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009847535412433714,
      "loss": 1.714,
      "step": 4603
    },
    {
      "epoch": 0.3961111589090596,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009847469138428075,
      "loss": 1.4197,
      "step": 4604
    },
    {
      "epoch": 0.3961971952163813,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009847402850244533,
      "loss": 1.2942,
      "step": 4605
    },
    {
      "epoch": 0.396283231523703,
      "grad_norm": 1.203125,
      "learning_rate": 0.000984733654788328,
      "loss": 1.5468,
      "step": 4606
    },
    {
      "epoch": 0.3963692678310247,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009847270231344509,
      "loss": 2.0115,
      "step": 4607
    },
    {
      "epoch": 0.3964553041383464,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009847203900628415,
      "loss": 0.994,
      "step": 4608
    },
    {
      "epoch": 0.3965413404456681,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009847137555735192,
      "loss": 0.781,
      "step": 4609
    },
    {
      "epoch": 0.39662737675298976,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009847071196665037,
      "loss": 1.2225,
      "step": 4610
    },
    {
      "epoch": 0.39671341306031144,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009847004823418139,
      "loss": 1.9766,
      "step": 4611
    },
    {
      "epoch": 0.3967994493676331,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009846938435994696,
      "loss": 1.4414,
      "step": 4612
    },
    {
      "epoch": 0.39688548567495485,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009846872034394902,
      "loss": 2.4481,
      "step": 4613
    },
    {
      "epoch": 0.39697152198227653,
      "grad_norm": 1.125,
      "learning_rate": 0.0009846805618618948,
      "loss": 1.2673,
      "step": 4614
    },
    {
      "epoch": 0.3970575582895982,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009846739188667031,
      "loss": 1.4554,
      "step": 4615
    },
    {
      "epoch": 0.3971435945969199,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009846672744539342,
      "loss": 1.36,
      "step": 4616
    },
    {
      "epoch": 0.39722963090424157,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000984660628623608,
      "loss": 1.9772,
      "step": 4617
    },
    {
      "epoch": 0.3973156672115633,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009846539813757437,
      "loss": 1.0339,
      "step": 4618
    },
    {
      "epoch": 0.397401703518885,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000984647332710361,
      "loss": 1.7588,
      "step": 4619
    },
    {
      "epoch": 0.39748773982620667,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009846406826274788,
      "loss": 1.3293,
      "step": 4620
    },
    {
      "epoch": 0.39757377613352834,
      "grad_norm": 1.703125,
      "learning_rate": 0.000984634031127117,
      "loss": 1.2842,
      "step": 4621
    },
    {
      "epoch": 0.39765981244085,
      "grad_norm": 1.171875,
      "learning_rate": 0.000984627378209295,
      "loss": 1.776,
      "step": 4622
    },
    {
      "epoch": 0.3977458487481717,
      "grad_norm": 1.40625,
      "learning_rate": 0.000984620723874032,
      "loss": 1.2574,
      "step": 4623
    },
    {
      "epoch": 0.39783188505549344,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009846140681213475,
      "loss": 1.6128,
      "step": 4624
    },
    {
      "epoch": 0.3979179213628151,
      "grad_norm": 1.75,
      "learning_rate": 0.0009846074109512613,
      "loss": 1.545,
      "step": 4625
    },
    {
      "epoch": 0.3980039576701368,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009846007523637928,
      "loss": 1.4072,
      "step": 4626
    },
    {
      "epoch": 0.3980899939774585,
      "grad_norm": 1.4375,
      "learning_rate": 0.000984594092358961,
      "loss": 1.521,
      "step": 4627
    },
    {
      "epoch": 0.39817603028478016,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009845874309367861,
      "loss": 1.8039,
      "step": 4628
    },
    {
      "epoch": 0.3982620665921019,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009845807680972868,
      "loss": 1.368,
      "step": 4629
    },
    {
      "epoch": 0.3983481028994236,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009845741038404833,
      "loss": 0.9616,
      "step": 4630
    },
    {
      "epoch": 0.39843413920674525,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009845674381663949,
      "loss": 0.656,
      "step": 4631
    },
    {
      "epoch": 0.39852017551406693,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009845607710750408,
      "loss": 1.0105,
      "step": 4632
    },
    {
      "epoch": 0.3986062118213886,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009845541025664407,
      "loss": 1.4482,
      "step": 4633
    },
    {
      "epoch": 0.3986922481287103,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000984547432640614,
      "loss": 0.7714,
      "step": 4634
    },
    {
      "epoch": 0.398778284436032,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009845407612975804,
      "loss": 0.8518,
      "step": 4635
    },
    {
      "epoch": 0.3988643207433537,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009845340885373593,
      "loss": 1.3245,
      "step": 4636
    },
    {
      "epoch": 0.3989503570506754,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009845274143599702,
      "loss": 2.147,
      "step": 4637
    },
    {
      "epoch": 0.39903639335799707,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009845207387654327,
      "loss": 1.1637,
      "step": 4638
    },
    {
      "epoch": 0.39912242966531875,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009845140617537663,
      "loss": 1.3815,
      "step": 4639
    },
    {
      "epoch": 0.3992084659726405,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009845073833249904,
      "loss": 2.2717,
      "step": 4640
    },
    {
      "epoch": 0.39929450227996216,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009845007034791247,
      "loss": 1.0508,
      "step": 4641
    },
    {
      "epoch": 0.39938053858728384,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009844940222161888,
      "loss": 1.373,
      "step": 4642
    },
    {
      "epoch": 0.3994665748946055,
      "grad_norm": 1.171875,
      "learning_rate": 0.000984487339536202,
      "loss": 1.4558,
      "step": 4643
    },
    {
      "epoch": 0.3995526112019272,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009844806554391839,
      "loss": 1.9308,
      "step": 4644
    },
    {
      "epoch": 0.3996386475092489,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009844739699251541,
      "loss": 1.4401,
      "step": 4645
    },
    {
      "epoch": 0.3997246838165706,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009844672829941322,
      "loss": 1.1234,
      "step": 4646
    },
    {
      "epoch": 0.3998107201238923,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009844605946461378,
      "loss": 1.676,
      "step": 4647
    },
    {
      "epoch": 0.399896756431214,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009844539048811903,
      "loss": 1.262,
      "step": 4648
    },
    {
      "epoch": 0.39998279273853565,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009844472136993094,
      "loss": 1.4288,
      "step": 4649
    },
    {
      "epoch": 0.40006882904585733,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009844405211005146,
      "loss": 1.6735,
      "step": 4650
    },
    {
      "epoch": 0.40015486535317907,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009844338270848254,
      "loss": 1.3578,
      "step": 4651
    },
    {
      "epoch": 0.40024090166050075,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009844271316522615,
      "loss": 1.5717,
      "step": 4652
    },
    {
      "epoch": 0.4003269379678224,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009844204348028426,
      "loss": 1.6332,
      "step": 4653
    },
    {
      "epoch": 0.4004129742751441,
      "grad_norm": 1.421875,
      "learning_rate": 0.000984413736536588,
      "loss": 1.4539,
      "step": 4654
    },
    {
      "epoch": 0.4004990105824658,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009844070368535175,
      "loss": 1.6321,
      "step": 4655
    },
    {
      "epoch": 0.40058504688978747,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009844003357536507,
      "loss": 1.3998,
      "step": 4656
    },
    {
      "epoch": 0.4006710831971092,
      "grad_norm": 1.140625,
      "learning_rate": 0.000984393633237007,
      "loss": 1.1114,
      "step": 4657
    },
    {
      "epoch": 0.4007571195044309,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009843869293036061,
      "loss": 0.9879,
      "step": 4658
    },
    {
      "epoch": 0.40084315581175256,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009843802239534677,
      "loss": 0.8618,
      "step": 4659
    },
    {
      "epoch": 0.40092919211907424,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009843735171866114,
      "loss": 1.4132,
      "step": 4660
    },
    {
      "epoch": 0.4010152284263959,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009843668090030568,
      "loss": 1.4094,
      "step": 4661
    },
    {
      "epoch": 0.40110126473371766,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009843600994028235,
      "loss": 1.8316,
      "step": 4662
    },
    {
      "epoch": 0.40118730104103933,
      "grad_norm": 2.0,
      "learning_rate": 0.000984353388385931,
      "loss": 1.571,
      "step": 4663
    },
    {
      "epoch": 0.401273337348361,
      "grad_norm": 1.21875,
      "learning_rate": 0.000984346675952399,
      "loss": 0.8279,
      "step": 4664
    },
    {
      "epoch": 0.4013593736556827,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009843399621022474,
      "loss": 1.409,
      "step": 4665
    },
    {
      "epoch": 0.4014454099630044,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009843332468354954,
      "loss": 1.1985,
      "step": 4666
    },
    {
      "epoch": 0.40153144627032605,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009843265301521628,
      "loss": 1.233,
      "step": 4667
    },
    {
      "epoch": 0.4016174825776478,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009843198120522694,
      "loss": 0.9204,
      "step": 4668
    },
    {
      "epoch": 0.40170351888496947,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009843130925358348,
      "loss": 1.9264,
      "step": 4669
    },
    {
      "epoch": 0.40178955519229115,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009843063716028786,
      "loss": 1.7459,
      "step": 4670
    },
    {
      "epoch": 0.4018755914996128,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009842996492534202,
      "loss": 0.6434,
      "step": 4671
    },
    {
      "epoch": 0.4019616278069345,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00098429292548748,
      "loss": 0.6845,
      "step": 4672
    },
    {
      "epoch": 0.40204766411425624,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009842862003050768,
      "loss": 1.4958,
      "step": 4673
    },
    {
      "epoch": 0.4021337004215779,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009842794737062305,
      "loss": 0.9224,
      "step": 4674
    },
    {
      "epoch": 0.4022197367288996,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009842727456909615,
      "loss": 1.823,
      "step": 4675
    },
    {
      "epoch": 0.4023057730362213,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009842660162592885,
      "loss": 1.9344,
      "step": 4676
    },
    {
      "epoch": 0.40239180934354296,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009842592854112315,
      "loss": 1.4854,
      "step": 4677
    },
    {
      "epoch": 0.40247784565086464,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009842525531468106,
      "loss": 1.7692,
      "step": 4678
    },
    {
      "epoch": 0.4025638819581864,
      "grad_norm": 1.375,
      "learning_rate": 0.000984245819466045,
      "loss": 1.8188,
      "step": 4679
    },
    {
      "epoch": 0.40264991826550806,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009842390843689546,
      "loss": 1.5614,
      "step": 4680
    },
    {
      "epoch": 0.40273595457282974,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009842323478555588,
      "loss": 1.7584,
      "step": 4681
    },
    {
      "epoch": 0.4028219908801514,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009842256099258778,
      "loss": 1.0519,
      "step": 4682
    },
    {
      "epoch": 0.4029080271874731,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009842188705799313,
      "loss": 1.3421,
      "step": 4683
    },
    {
      "epoch": 0.40299406349479483,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009842121298177385,
      "loss": 2.129,
      "step": 4684
    },
    {
      "epoch": 0.4030800998021165,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009842053876393192,
      "loss": 1.682,
      "step": 4685
    },
    {
      "epoch": 0.4031661361094382,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009841986440446938,
      "loss": 1.5533,
      "step": 4686
    },
    {
      "epoch": 0.40325217241675987,
      "grad_norm": 3.078125,
      "learning_rate": 0.0009841918990338811,
      "loss": 1.3858,
      "step": 4687
    },
    {
      "epoch": 0.40333820872408155,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009841851526069013,
      "loss": 2.3391,
      "step": 4688
    },
    {
      "epoch": 0.40342424503140323,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009841784047637744,
      "loss": 1.941,
      "step": 4689
    },
    {
      "epoch": 0.40351028133872496,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009841716555045196,
      "loss": 0.6976,
      "step": 4690
    },
    {
      "epoch": 0.40359631764604664,
      "grad_norm": 1.703125,
      "learning_rate": 0.000984164904829157,
      "loss": 1.0996,
      "step": 4691
    },
    {
      "epoch": 0.4036823539533683,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009841581527377063,
      "loss": 1.5281,
      "step": 4692
    },
    {
      "epoch": 0.40376839026069,
      "grad_norm": 1.375,
      "learning_rate": 0.0009841513992301868,
      "loss": 1.6929,
      "step": 4693
    },
    {
      "epoch": 0.4038544265680117,
      "grad_norm": 1.671875,
      "learning_rate": 0.000984144644306619,
      "loss": 1.8492,
      "step": 4694
    },
    {
      "epoch": 0.4039404628753334,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009841378879670221,
      "loss": 0.7298,
      "step": 4695
    },
    {
      "epoch": 0.4040264991826551,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009841311302114162,
      "loss": 0.8229,
      "step": 4696
    },
    {
      "epoch": 0.4041125354899768,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009841243710398207,
      "loss": 1.2486,
      "step": 4697
    },
    {
      "epoch": 0.40419857179729846,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009841176104522556,
      "loss": 1.1383,
      "step": 4698
    },
    {
      "epoch": 0.40428460810462014,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009841108484487406,
      "loss": 1.1975,
      "step": 4699
    },
    {
      "epoch": 0.4043706444119418,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009841040850292956,
      "loss": 1.1652,
      "step": 4700
    },
    {
      "epoch": 0.40445668071926355,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009840973201939406,
      "loss": 2.0513,
      "step": 4701
    },
    {
      "epoch": 0.40454271702658523,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009840905539426946,
      "loss": 0.9818,
      "step": 4702
    },
    {
      "epoch": 0.4046287533339069,
      "grad_norm": 1.265625,
      "learning_rate": 0.000984083786275578,
      "loss": 1.6413,
      "step": 4703
    },
    {
      "epoch": 0.4047147896412286,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009840770171926107,
      "loss": 1.3453,
      "step": 4704
    },
    {
      "epoch": 0.40480082594855027,
      "grad_norm": 1.5,
      "learning_rate": 0.0009840702466938122,
      "loss": 0.9905,
      "step": 4705
    },
    {
      "epoch": 0.404886862255872,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009840634747792021,
      "loss": 2.3096,
      "step": 4706
    },
    {
      "epoch": 0.4049728985631937,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009840567014488008,
      "loss": 1.3424,
      "step": 4707
    },
    {
      "epoch": 0.40505893487051536,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009840499267026277,
      "loss": 1.3548,
      "step": 4708
    },
    {
      "epoch": 0.40514497117783704,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009840431505407025,
      "loss": 1.9151,
      "step": 4709
    },
    {
      "epoch": 0.4052310074851587,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009840363729630454,
      "loss": 1.2257,
      "step": 4710
    },
    {
      "epoch": 0.4053170437924804,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009840295939696762,
      "loss": 1.3338,
      "step": 4711
    },
    {
      "epoch": 0.40540308009980214,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009840228135606145,
      "loss": 1.4883,
      "step": 4712
    },
    {
      "epoch": 0.4054891164071238,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00098401603173588,
      "loss": 0.787,
      "step": 4713
    },
    {
      "epoch": 0.4055751527144455,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000984009248495493,
      "loss": 1.4142,
      "step": 4714
    },
    {
      "epoch": 0.4056611890217672,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009840024638394728,
      "loss": 1.466,
      "step": 4715
    },
    {
      "epoch": 0.40574722532908886,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009839956777678398,
      "loss": 1.3322,
      "step": 4716
    },
    {
      "epoch": 0.4058332616364106,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009839888902806136,
      "loss": 1.3365,
      "step": 4717
    },
    {
      "epoch": 0.40591929794373227,
      "grad_norm": 1.125,
      "learning_rate": 0.000983982101377814,
      "loss": 1.4461,
      "step": 4718
    },
    {
      "epoch": 0.40600533425105395,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009839753110594605,
      "loss": 1.9373,
      "step": 4719
    },
    {
      "epoch": 0.40609137055837563,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009839685193255736,
      "loss": 1.6476,
      "step": 4720
    },
    {
      "epoch": 0.4061774068656973,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000983961726176173,
      "loss": 2.0953,
      "step": 4721
    },
    {
      "epoch": 0.406263443173019,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009839549316112786,
      "loss": 1.7962,
      "step": 4722
    },
    {
      "epoch": 0.4063494794803407,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009839481356309098,
      "loss": 1.0511,
      "step": 4723
    },
    {
      "epoch": 0.4064355157876624,
      "grad_norm": 1.125,
      "learning_rate": 0.000983941338235087,
      "loss": 1.2764,
      "step": 4724
    },
    {
      "epoch": 0.4065215520949841,
      "grad_norm": 1.703125,
      "learning_rate": 0.00098393453942383,
      "loss": 2.0555,
      "step": 4725
    },
    {
      "epoch": 0.40660758840230576,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009839277391971585,
      "loss": 1.4894,
      "step": 4726
    },
    {
      "epoch": 0.40669362470962744,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009839209375550925,
      "loss": 1.9011,
      "step": 4727
    },
    {
      "epoch": 0.4067796610169492,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009839141344976517,
      "loss": 1.8037,
      "step": 4728
    },
    {
      "epoch": 0.40686569732427086,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009839073300248562,
      "loss": 1.3601,
      "step": 4729
    },
    {
      "epoch": 0.40695173363159254,
      "grad_norm": 1.8984375,
      "learning_rate": 0.000983900524136726,
      "loss": 1.974,
      "step": 4730
    },
    {
      "epoch": 0.4070377699389142,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000983893716833281,
      "loss": 1.4924,
      "step": 4731
    },
    {
      "epoch": 0.4071238062462359,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009838869081145408,
      "loss": 1.332,
      "step": 4732
    },
    {
      "epoch": 0.4072098425535576,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009838800979805255,
      "loss": 1.6417,
      "step": 4733
    },
    {
      "epoch": 0.4072958788608793,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009838732864312551,
      "loss": 0.8138,
      "step": 4734
    },
    {
      "epoch": 0.407381915168201,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009838664734667494,
      "loss": 1.3176,
      "step": 4735
    },
    {
      "epoch": 0.40746795147552267,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009838596590870286,
      "loss": 1.2367,
      "step": 4736
    },
    {
      "epoch": 0.40755398778284435,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000983852843292112,
      "loss": 1.4391,
      "step": 4737
    },
    {
      "epoch": 0.40764002409016603,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00098384602608202,
      "loss": 1.38,
      "step": 4738
    },
    {
      "epoch": 0.40772606039748777,
      "grad_norm": 1.375,
      "learning_rate": 0.0009838392074567727,
      "loss": 1.4326,
      "step": 4739
    },
    {
      "epoch": 0.40781209670480945,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00098383238741639,
      "loss": 1.9651,
      "step": 4740
    },
    {
      "epoch": 0.4078981330121311,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009838255659608912,
      "loss": 1.5694,
      "step": 4741
    },
    {
      "epoch": 0.4079841693194528,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000983818743090297,
      "loss": 1.3626,
      "step": 4742
    },
    {
      "epoch": 0.4080702056267745,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000983811918804627,
      "loss": 0.9303,
      "step": 4743
    },
    {
      "epoch": 0.40815624193409616,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009838050931039012,
      "loss": 1.4671,
      "step": 4744
    },
    {
      "epoch": 0.4082422782414179,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009837982659881396,
      "loss": 1.5692,
      "step": 4745
    },
    {
      "epoch": 0.4083283145487396,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009837914374573621,
      "loss": 2.1009,
      "step": 4746
    },
    {
      "epoch": 0.40841435085606126,
      "grad_norm": 1.71875,
      "learning_rate": 0.000983784607511589,
      "loss": 1.8867,
      "step": 4747
    },
    {
      "epoch": 0.40850038716338294,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0009837777761508397,
      "loss": 0.6814,
      "step": 4748
    },
    {
      "epoch": 0.4085864234707046,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009837709433751347,
      "loss": 1.4646,
      "step": 4749
    },
    {
      "epoch": 0.40867245977802635,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009837641091844937,
      "loss": 1.2411,
      "step": 4750
    },
    {
      "epoch": 0.40875849608534803,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009837572735789369,
      "loss": 1.1928,
      "step": 4751
    },
    {
      "epoch": 0.4088445323926697,
      "grad_norm": 1.171875,
      "learning_rate": 0.000983750436558484,
      "loss": 1.1803,
      "step": 4752
    },
    {
      "epoch": 0.4089305686999914,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009837435981231552,
      "loss": 1.0834,
      "step": 4753
    },
    {
      "epoch": 0.40901660500731307,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009837367582729705,
      "loss": 1.2804,
      "step": 4754
    },
    {
      "epoch": 0.40910264131463475,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009837299170079496,
      "loss": 1.5464,
      "step": 4755
    },
    {
      "epoch": 0.4091886776219565,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000983723074328113,
      "loss": 1.1697,
      "step": 4756
    },
    {
      "epoch": 0.40927471392927817,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009837162302334806,
      "loss": 1.4939,
      "step": 4757
    },
    {
      "epoch": 0.40936075023659985,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009837093847240722,
      "loss": 1.3863,
      "step": 4758
    },
    {
      "epoch": 0.4094467865439215,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009837025377999078,
      "loss": 1.4425,
      "step": 4759
    },
    {
      "epoch": 0.4095328228512432,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009836956894610077,
      "loss": 1.8772,
      "step": 4760
    },
    {
      "epoch": 0.40961885915856494,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009836888397073919,
      "loss": 1.1646,
      "step": 4761
    },
    {
      "epoch": 0.4097048954658866,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009836819885390802,
      "loss": 1.2899,
      "step": 4762
    },
    {
      "epoch": 0.4097909317732083,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009836751359560928,
      "loss": 1.4368,
      "step": 4763
    },
    {
      "epoch": 0.40987696808053,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009836682819584495,
      "loss": 2.0775,
      "step": 4764
    },
    {
      "epoch": 0.40996300438785166,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009836614265461706,
      "loss": 1.3213,
      "step": 4765
    },
    {
      "epoch": 0.41004904069517334,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009836545697192761,
      "loss": 1.715,
      "step": 4766
    },
    {
      "epoch": 0.4101350770024951,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009836477114777862,
      "loss": 1.7706,
      "step": 4767
    },
    {
      "epoch": 0.41022111330981675,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009836408518217206,
      "loss": 1.1716,
      "step": 4768
    },
    {
      "epoch": 0.41030714961713843,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009836339907511,
      "loss": 1.4011,
      "step": 4769
    },
    {
      "epoch": 0.4103931859244601,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009836271282659434,
      "loss": 1.1507,
      "step": 4770
    },
    {
      "epoch": 0.4104792222317818,
      "grad_norm": 8.6875,
      "learning_rate": 0.0009836202643662719,
      "loss": 2.5588,
      "step": 4771
    },
    {
      "epoch": 0.41056525853910353,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000983613399052105,
      "loss": 1.5065,
      "step": 4772
    },
    {
      "epoch": 0.4106512948464252,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009836065323234632,
      "loss": 1.4461,
      "step": 4773
    },
    {
      "epoch": 0.4107373311537469,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009835996641803662,
      "loss": 1.346,
      "step": 4774
    },
    {
      "epoch": 0.41082336746106857,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009835927946228343,
      "loss": 1.2817,
      "step": 4775
    },
    {
      "epoch": 0.41090940376839025,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009835859236508875,
      "loss": 1.2204,
      "step": 4776
    },
    {
      "epoch": 0.4109954400757119,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000983579051264546,
      "loss": 1.1876,
      "step": 4777
    },
    {
      "epoch": 0.41108147638303366,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009835721774638297,
      "loss": 1.9438,
      "step": 4778
    },
    {
      "epoch": 0.41116751269035534,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009835653022487588,
      "loss": 0.8212,
      "step": 4779
    },
    {
      "epoch": 0.411253548997677,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009835584256193534,
      "loss": 1.2796,
      "step": 4780
    },
    {
      "epoch": 0.4113395853049987,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000983551547575634,
      "loss": 1.7389,
      "step": 4781
    },
    {
      "epoch": 0.4114256216123204,
      "grad_norm": 1.125,
      "learning_rate": 0.00098354466811762,
      "loss": 1.5239,
      "step": 4782
    },
    {
      "epoch": 0.4115116579196421,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009835377872453321,
      "loss": 0.9586,
      "step": 4783
    },
    {
      "epoch": 0.4115976942269638,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009835309049587902,
      "loss": 1.0675,
      "step": 4784
    },
    {
      "epoch": 0.4116837305342855,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009835240212580144,
      "loss": 1.4627,
      "step": 4785
    },
    {
      "epoch": 0.41176976684160715,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009835171361430247,
      "loss": 1.6907,
      "step": 4786
    },
    {
      "epoch": 0.41185580314892883,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009835102496138417,
      "loss": 1.1277,
      "step": 4787
    },
    {
      "epoch": 0.4119418394562505,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009835033616704852,
      "loss": 0.9236,
      "step": 4788
    },
    {
      "epoch": 0.41202787576357225,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009834964723129753,
      "loss": 1.1547,
      "step": 4789
    },
    {
      "epoch": 0.41211391207089393,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009834895815413324,
      "loss": 1.2701,
      "step": 4790
    },
    {
      "epoch": 0.4121999483782156,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009834826893555762,
      "loss": 1.2284,
      "step": 4791
    },
    {
      "epoch": 0.4122859846855373,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009834757957557275,
      "loss": 0.9016,
      "step": 4792
    },
    {
      "epoch": 0.41237202099285897,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000983468900741806,
      "loss": 1.2636,
      "step": 4793
    },
    {
      "epoch": 0.4124580573001807,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000983462004313832,
      "loss": 1.6728,
      "step": 4794
    },
    {
      "epoch": 0.4125440936075024,
      "grad_norm": 1.125,
      "learning_rate": 0.0009834551064718255,
      "loss": 1.0802,
      "step": 4795
    },
    {
      "epoch": 0.41263012991482406,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000983448207215807,
      "loss": 1.2913,
      "step": 4796
    },
    {
      "epoch": 0.41271616622214574,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009834413065457966,
      "loss": 1.6064,
      "step": 4797
    },
    {
      "epoch": 0.4128022025294674,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009834344044618143,
      "loss": 1.4077,
      "step": 4798
    },
    {
      "epoch": 0.4128882388367891,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009834275009638804,
      "loss": 1.317,
      "step": 4799
    },
    {
      "epoch": 0.41297427514411084,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000983420596052015,
      "loss": 1.4385,
      "step": 4800
    },
    {
      "epoch": 0.4130603114514325,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009834136897262386,
      "loss": 2.1981,
      "step": 4801
    },
    {
      "epoch": 0.4131463477587542,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000983406781986571,
      "loss": 1.5346,
      "step": 4802
    },
    {
      "epoch": 0.4132323840660759,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009833998728330327,
      "loss": 0.5724,
      "step": 4803
    },
    {
      "epoch": 0.41331842037339755,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009833929622656437,
      "loss": 1.3333,
      "step": 4804
    },
    {
      "epoch": 0.4134044566807193,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009833860502844243,
      "loss": 1.3861,
      "step": 4805
    },
    {
      "epoch": 0.41349049298804097,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009833791368893947,
      "loss": 1.8729,
      "step": 4806
    },
    {
      "epoch": 0.41357652929536265,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009833722220805752,
      "loss": 2.2737,
      "step": 4807
    },
    {
      "epoch": 0.41366256560268433,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009833653058579858,
      "loss": 1.9854,
      "step": 4808
    },
    {
      "epoch": 0.413748601910006,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000983358388221647,
      "loss": 1.1079,
      "step": 4809
    },
    {
      "epoch": 0.4138346382173277,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009833514691715791,
      "loss": 1.0917,
      "step": 4810
    },
    {
      "epoch": 0.4139206745246494,
      "grad_norm": 1.875,
      "learning_rate": 0.000983344548707802,
      "loss": 1.1951,
      "step": 4811
    },
    {
      "epoch": 0.4140067108319711,
      "grad_norm": 1.421875,
      "learning_rate": 0.000983337626830336,
      "loss": 1.847,
      "step": 4812
    },
    {
      "epoch": 0.4140927471392928,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009833307035392016,
      "loss": 0.8082,
      "step": 4813
    },
    {
      "epoch": 0.41417878344661446,
      "grad_norm": 1.7890625,
      "learning_rate": 0.000983323778834419,
      "loss": 1.542,
      "step": 4814
    },
    {
      "epoch": 0.41426481975393614,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009833168527160082,
      "loss": 1.0188,
      "step": 4815
    },
    {
      "epoch": 0.4143508560612579,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009833099251839895,
      "loss": 1.7051,
      "step": 4816
    },
    {
      "epoch": 0.41443689236857956,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009833029962383832,
      "loss": 1.6435,
      "step": 4817
    },
    {
      "epoch": 0.41452292867590124,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00098329606587921,
      "loss": 1.1154,
      "step": 4818
    },
    {
      "epoch": 0.4146089649832229,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009832891341064896,
      "loss": 2.1002,
      "step": 4819
    },
    {
      "epoch": 0.4146950012905446,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009832822009202423,
      "loss": 1.4442,
      "step": 4820
    },
    {
      "epoch": 0.4147810375978663,
      "grad_norm": 1.25,
      "learning_rate": 0.0009832752663204887,
      "loss": 1.3692,
      "step": 4821
    },
    {
      "epoch": 0.414867073905188,
      "grad_norm": 1.421875,
      "learning_rate": 0.000983268330307249,
      "loss": 1.468,
      "step": 4822
    },
    {
      "epoch": 0.4149531102125097,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009832613928805435,
      "loss": 1.4689,
      "step": 4823
    },
    {
      "epoch": 0.41503914651983137,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009832544540403923,
      "loss": 1.4022,
      "step": 4824
    },
    {
      "epoch": 0.41512518282715305,
      "grad_norm": 1.625,
      "learning_rate": 0.0009832475137868159,
      "loss": 1.2487,
      "step": 4825
    },
    {
      "epoch": 0.41521121913447473,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009832405721198344,
      "loss": 1.0936,
      "step": 4826
    },
    {
      "epoch": 0.41529725544179646,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000983233629039468,
      "loss": 1.84,
      "step": 4827
    },
    {
      "epoch": 0.41538329174911814,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009832266845457376,
      "loss": 1.5375,
      "step": 4828
    },
    {
      "epoch": 0.4154693280564398,
      "grad_norm": 1.265625,
      "learning_rate": 0.000983219738638663,
      "loss": 0.8579,
      "step": 4829
    },
    {
      "epoch": 0.4155553643637615,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009832127913182646,
      "loss": 0.9621,
      "step": 4830
    },
    {
      "epoch": 0.4156414006710832,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000983205842584563,
      "loss": 1.034,
      "step": 4831
    },
    {
      "epoch": 0.41572743697840486,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009831988924375778,
      "loss": 1.7587,
      "step": 4832
    },
    {
      "epoch": 0.4158134732857266,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009831919408773303,
      "loss": 1.9982,
      "step": 4833
    },
    {
      "epoch": 0.4158995095930483,
      "grad_norm": 2.421875,
      "learning_rate": 0.0009831849879038401,
      "loss": 1.2335,
      "step": 4834
    },
    {
      "epoch": 0.41598554590036996,
      "grad_norm": 3.5,
      "learning_rate": 0.0009831780335171278,
      "loss": 2.0537,
      "step": 4835
    },
    {
      "epoch": 0.41607158220769164,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009831710777172139,
      "loss": 1.0849,
      "step": 4836
    },
    {
      "epoch": 0.4161576185150133,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009831641205041183,
      "loss": 1.5743,
      "step": 4837
    },
    {
      "epoch": 0.41624365482233505,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009831571618778617,
      "loss": 1.3901,
      "step": 4838
    },
    {
      "epoch": 0.41632969112965673,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009831502018384645,
      "loss": 1.2612,
      "step": 4839
    },
    {
      "epoch": 0.4164157274369784,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009831432403859467,
      "loss": 1.1445,
      "step": 4840
    },
    {
      "epoch": 0.4165017637443001,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000983136277520329,
      "loss": 1.772,
      "step": 4841
    },
    {
      "epoch": 0.41658780005162177,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009831293132416318,
      "loss": 1.6914,
      "step": 4842
    },
    {
      "epoch": 0.41667383635894345,
      "grad_norm": 1.3125,
      "learning_rate": 0.000983122347549875,
      "loss": 2.109,
      "step": 4843
    },
    {
      "epoch": 0.4167598726662652,
      "grad_norm": 1.25,
      "learning_rate": 0.0009831153804450796,
      "loss": 1.6534,
      "step": 4844
    },
    {
      "epoch": 0.41684590897358687,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009831084119272659,
      "loss": 1.7311,
      "step": 4845
    },
    {
      "epoch": 0.41693194528090854,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009831014419964534,
      "loss": 1.3358,
      "step": 4846
    },
    {
      "epoch": 0.4170179815882302,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009830944706526637,
      "loss": 1.8211,
      "step": 4847
    },
    {
      "epoch": 0.4171040178955519,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009830874978959164,
      "loss": 0.987,
      "step": 4848
    },
    {
      "epoch": 0.41719005420287364,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000983080523726232,
      "loss": 2.0281,
      "step": 4849
    },
    {
      "epoch": 0.4172760905101953,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000983073548143631,
      "loss": 1.8785,
      "step": 4850
    },
    {
      "epoch": 0.417362126817517,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000983066571148134,
      "loss": 1.1248,
      "step": 4851
    },
    {
      "epoch": 0.4174481631248387,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009830595927397612,
      "loss": 1.6259,
      "step": 4852
    },
    {
      "epoch": 0.41753419943216036,
      "grad_norm": 1.203125,
      "learning_rate": 0.000983052612918533,
      "loss": 1.3776,
      "step": 4853
    },
    {
      "epoch": 0.41762023573948204,
      "grad_norm": 1.0625,
      "learning_rate": 0.00098304563168447,
      "loss": 1.1792,
      "step": 4854
    },
    {
      "epoch": 0.4177062720468038,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009830386490375924,
      "loss": 1.2618,
      "step": 4855
    },
    {
      "epoch": 0.41779230835412545,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009830316649779207,
      "loss": 1.1036,
      "step": 4856
    },
    {
      "epoch": 0.41787834466144713,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009830246795054754,
      "loss": 0.7189,
      "step": 4857
    },
    {
      "epoch": 0.4179643809687688,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009830176926202768,
      "loss": 1.4972,
      "step": 4858
    },
    {
      "epoch": 0.4180504172760905,
      "grad_norm": 1.5,
      "learning_rate": 0.0009830107043223451,
      "loss": 2.107,
      "step": 4859
    },
    {
      "epoch": 0.4181364535834122,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009830037146117013,
      "loss": 1.8062,
      "step": 4860
    },
    {
      "epoch": 0.4182224898907339,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009829967234883657,
      "loss": 2.0921,
      "step": 4861
    },
    {
      "epoch": 0.4183085261980556,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009829897309523584,
      "loss": 1.2322,
      "step": 4862
    },
    {
      "epoch": 0.41839456250537727,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009829827370037001,
      "loss": 1.324,
      "step": 4863
    },
    {
      "epoch": 0.41848059881269895,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009829757416424112,
      "loss": 2.0564,
      "step": 4864
    },
    {
      "epoch": 0.4185666351200206,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009829687448685122,
      "loss": 1.6657,
      "step": 4865
    },
    {
      "epoch": 0.41865267142734236,
      "grad_norm": 4.375,
      "learning_rate": 0.0009829617466820238,
      "loss": 1.3766,
      "step": 4866
    },
    {
      "epoch": 0.41873870773466404,
      "grad_norm": 1.640625,
      "learning_rate": 0.000982954747082966,
      "loss": 1.4926,
      "step": 4867
    },
    {
      "epoch": 0.4188247440419857,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009829477460713597,
      "loss": 1.6989,
      "step": 4868
    },
    {
      "epoch": 0.4189107803493074,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009829407436472248,
      "loss": 1.9721,
      "step": 4869
    },
    {
      "epoch": 0.4189968166566291,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009829337398105824,
      "loss": 1.5403,
      "step": 4870
    },
    {
      "epoch": 0.4190828529639508,
      "grad_norm": 1.5,
      "learning_rate": 0.0009829267345614527,
      "loss": 1.3475,
      "step": 4871
    },
    {
      "epoch": 0.4191688892712725,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009829197278998561,
      "loss": 1.46,
      "step": 4872
    },
    {
      "epoch": 0.4192549255785942,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009829127198258134,
      "loss": 0.9084,
      "step": 4873
    },
    {
      "epoch": 0.41934096188591585,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000982905710339345,
      "loss": 1.1592,
      "step": 4874
    },
    {
      "epoch": 0.41942699819323753,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009828986994404709,
      "loss": 1.6433,
      "step": 4875
    },
    {
      "epoch": 0.4195130345005592,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009828916871292123,
      "loss": 1.6552,
      "step": 4876
    },
    {
      "epoch": 0.41959907080788095,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009828846734055896,
      "loss": 0.8813,
      "step": 4877
    },
    {
      "epoch": 0.4196851071152026,
      "grad_norm": 1.109375,
      "learning_rate": 0.000982877658269623,
      "loss": 1.3719,
      "step": 4878
    },
    {
      "epoch": 0.4197711434225243,
      "grad_norm": 1.234375,
      "learning_rate": 0.000982870641721333,
      "loss": 1.1605,
      "step": 4879
    },
    {
      "epoch": 0.419857179729846,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009828636237607404,
      "loss": 1.8804,
      "step": 4880
    },
    {
      "epoch": 0.41994321603716767,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009828566043878658,
      "loss": 1.0692,
      "step": 4881
    },
    {
      "epoch": 0.4200292523444894,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009828495836027293,
      "loss": 2.0064,
      "step": 4882
    },
    {
      "epoch": 0.4201152886518111,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009828425614053517,
      "loss": 1.306,
      "step": 4883
    },
    {
      "epoch": 0.42020132495913276,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009828355377957535,
      "loss": 1.9108,
      "step": 4884
    },
    {
      "epoch": 0.42028736126645444,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009828285127739553,
      "loss": 1.4086,
      "step": 4885
    },
    {
      "epoch": 0.4203733975737761,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009828214863399777,
      "loss": 1.0819,
      "step": 4886
    },
    {
      "epoch": 0.4204594338810978,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000982814458493841,
      "loss": 1.3292,
      "step": 4887
    },
    {
      "epoch": 0.42054547018841953,
      "grad_norm": 2.046875,
      "learning_rate": 0.000982807429235566,
      "loss": 2.0864,
      "step": 4888
    },
    {
      "epoch": 0.4206315064957412,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009828003985651733,
      "loss": 1.6358,
      "step": 4889
    },
    {
      "epoch": 0.4207175428030629,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009827933664826832,
      "loss": 1.3463,
      "step": 4890
    },
    {
      "epoch": 0.4208035791103846,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009827863329881163,
      "loss": 2.057,
      "step": 4891
    },
    {
      "epoch": 0.42088961541770625,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009827792980814933,
      "loss": 0.8299,
      "step": 4892
    },
    {
      "epoch": 0.420975651725028,
      "grad_norm": 1.640625,
      "learning_rate": 0.000982772261762835,
      "loss": 1.1532,
      "step": 4893
    },
    {
      "epoch": 0.42106168803234967,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009827652240321613,
      "loss": 2.0798,
      "step": 4894
    },
    {
      "epoch": 0.42114772433967135,
      "grad_norm": 1.5,
      "learning_rate": 0.0009827581848894937,
      "loss": 1.5562,
      "step": 4895
    },
    {
      "epoch": 0.421233760646993,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000982751144334852,
      "loss": 1.2272,
      "step": 4896
    },
    {
      "epoch": 0.4213197969543147,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009827441023682571,
      "loss": 1.3442,
      "step": 4897
    },
    {
      "epoch": 0.4214058332616364,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009827370589897298,
      "loss": 1.4297,
      "step": 4898
    },
    {
      "epoch": 0.4214918695689581,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009827300141992902,
      "loss": 1.3046,
      "step": 4899
    },
    {
      "epoch": 0.4215779058762798,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009827229679969594,
      "loss": 1.449,
      "step": 4900
    },
    {
      "epoch": 0.4216639421836015,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009827159203827575,
      "loss": 1.3879,
      "step": 4901
    },
    {
      "epoch": 0.42174997849092316,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009827088713567059,
      "loss": 1.0744,
      "step": 4902
    },
    {
      "epoch": 0.42183601479824484,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009827018209188243,
      "loss": 1.3112,
      "step": 4903
    },
    {
      "epoch": 0.4219220511055666,
      "grad_norm": 1.9609375,
      "learning_rate": 0.000982694769069134,
      "loss": 2.0818,
      "step": 4904
    },
    {
      "epoch": 0.42200808741288826,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009826877158076554,
      "loss": 1.2273,
      "step": 4905
    },
    {
      "epoch": 0.42209412372020993,
      "grad_norm": 1.015625,
      "learning_rate": 0.000982680661134409,
      "loss": 1.3605,
      "step": 4906
    },
    {
      "epoch": 0.4221801600275316,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009826736050494156,
      "loss": 1.2357,
      "step": 4907
    },
    {
      "epoch": 0.4222661963348533,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009826665475526958,
      "loss": 1.7087,
      "step": 4908
    },
    {
      "epoch": 0.422352232642175,
      "grad_norm": 2.03125,
      "learning_rate": 0.00098265948864427,
      "loss": 1.0422,
      "step": 4909
    },
    {
      "epoch": 0.4224382689494967,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009826524283241592,
      "loss": 2.0925,
      "step": 4910
    },
    {
      "epoch": 0.4225243052568184,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000982645366592384,
      "loss": 1.1749,
      "step": 4911
    },
    {
      "epoch": 0.42261034156414007,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000982638303448965,
      "loss": 1.6091,
      "step": 4912
    },
    {
      "epoch": 0.42269637787146175,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009826312388939227,
      "loss": 2.3773,
      "step": 4913
    },
    {
      "epoch": 0.42278241417878343,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000982624172927278,
      "loss": 1.6753,
      "step": 4914
    },
    {
      "epoch": 0.42286845048610516,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009826171055490514,
      "loss": 1.8324,
      "step": 4915
    },
    {
      "epoch": 0.42295448679342684,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009826100367592637,
      "loss": 1.1174,
      "step": 4916
    },
    {
      "epoch": 0.4230405231007485,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009826029665579352,
      "loss": 1.2144,
      "step": 4917
    },
    {
      "epoch": 0.4231265594080702,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009825958949450872,
      "loss": 1.4148,
      "step": 4918
    },
    {
      "epoch": 0.4232125957153919,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00098258882192074,
      "loss": 1.5564,
      "step": 4919
    },
    {
      "epoch": 0.42329863202271356,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009825817474849144,
      "loss": 1.0121,
      "step": 4920
    },
    {
      "epoch": 0.4233846683300353,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009825746716376309,
      "loss": 1.7719,
      "step": 4921
    },
    {
      "epoch": 0.423470704637357,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009825675943789106,
      "loss": 1.5928,
      "step": 4922
    },
    {
      "epoch": 0.42355674094467866,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009825605157087736,
      "loss": 0.8106,
      "step": 4923
    },
    {
      "epoch": 0.42364277725200034,
      "grad_norm": 1.25,
      "learning_rate": 0.0009825534356272413,
      "loss": 1.0495,
      "step": 4924
    },
    {
      "epoch": 0.423728813559322,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009825463541343338,
      "loss": 0.9893,
      "step": 4925
    },
    {
      "epoch": 0.42381484986664375,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000982539271230072,
      "loss": 1.4797,
      "step": 4926
    },
    {
      "epoch": 0.42390088617396543,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009825321869144769,
      "loss": 1.5064,
      "step": 4927
    },
    {
      "epoch": 0.4239869224812871,
      "grad_norm": 1.203125,
      "learning_rate": 0.000982525101187569,
      "loss": 0.7678,
      "step": 4928
    },
    {
      "epoch": 0.4240729587886088,
      "grad_norm": 2.3125,
      "learning_rate": 0.0009825180140493688,
      "loss": 1.0072,
      "step": 4929
    },
    {
      "epoch": 0.42415899509593047,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009825109254998974,
      "loss": 1.4826,
      "step": 4930
    },
    {
      "epoch": 0.42424503140325215,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009825038355391754,
      "loss": 1.724,
      "step": 4931
    },
    {
      "epoch": 0.4243310677105739,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009824967441672235,
      "loss": 1.2301,
      "step": 4932
    },
    {
      "epoch": 0.42441710401789556,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009824896513840624,
      "loss": 1.593,
      "step": 4933
    },
    {
      "epoch": 0.42450314032521724,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009824825571897132,
      "loss": 1.0108,
      "step": 4934
    },
    {
      "epoch": 0.4245891766325389,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000982475461584196,
      "loss": 2.046,
      "step": 4935
    },
    {
      "epoch": 0.4246752129398606,
      "grad_norm": 1.265625,
      "learning_rate": 0.000982468364567532,
      "loss": 1.318,
      "step": 4936
    },
    {
      "epoch": 0.42476124924718234,
      "grad_norm": 107.0,
      "learning_rate": 0.000982461266139742,
      "loss": 1.2814,
      "step": 4937
    },
    {
      "epoch": 0.424847285554504,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009824541663008466,
      "loss": 2.0226,
      "step": 4938
    },
    {
      "epoch": 0.4249333218618257,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009824470650508664,
      "loss": 1.1554,
      "step": 4939
    },
    {
      "epoch": 0.4250193581691474,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009824399623898227,
      "loss": 1.4376,
      "step": 4940
    },
    {
      "epoch": 0.42510539447646906,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009824328583177356,
      "loss": 1.6789,
      "step": 4941
    },
    {
      "epoch": 0.42519143078379074,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009824257528346263,
      "loss": 1.7157,
      "step": 4942
    },
    {
      "epoch": 0.42527746709111247,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009824186459405156,
      "loss": 1.3623,
      "step": 4943
    },
    {
      "epoch": 0.42536350339843415,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009824115376354241,
      "loss": 1.3473,
      "step": 4944
    },
    {
      "epoch": 0.42544953970575583,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009824044279193728,
      "loss": 0.751,
      "step": 4945
    },
    {
      "epoch": 0.4255355760130775,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009823973167923823,
      "loss": 1.072,
      "step": 4946
    },
    {
      "epoch": 0.4256216123203992,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009823902042544734,
      "loss": 1.4263,
      "step": 4947
    },
    {
      "epoch": 0.4257076486277209,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009823830903056671,
      "loss": 0.9453,
      "step": 4948
    },
    {
      "epoch": 0.4257936849350426,
      "grad_norm": 1.5,
      "learning_rate": 0.000982375974945984,
      "loss": 1.1592,
      "step": 4949
    },
    {
      "epoch": 0.4258797212423643,
      "grad_norm": 1.421875,
      "learning_rate": 0.000982368858175445,
      "loss": 1.4791,
      "step": 4950
    },
    {
      "epoch": 0.42596575754968596,
      "grad_norm": 1.125,
      "learning_rate": 0.000982361739994071,
      "loss": 1.5334,
      "step": 4951
    },
    {
      "epoch": 0.42605179385700764,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009823546204018826,
      "loss": 1.4041,
      "step": 4952
    },
    {
      "epoch": 0.4261378301643293,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009823474993989009,
      "loss": 1.254,
      "step": 4953
    },
    {
      "epoch": 0.42622386647165106,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009823403769851462,
      "loss": 1.449,
      "step": 4954
    },
    {
      "epoch": 0.42630990277897274,
      "grad_norm": 1.578125,
      "learning_rate": 0.00098233325316064,
      "loss": 1.0031,
      "step": 4955
    },
    {
      "epoch": 0.4263959390862944,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009823261279254029,
      "loss": 2.2415,
      "step": 4956
    },
    {
      "epoch": 0.4264819753936161,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009823190012794554,
      "loss": 1.8048,
      "step": 4957
    },
    {
      "epoch": 0.4265680117009378,
      "grad_norm": 1.0625,
      "learning_rate": 0.000982311873222819,
      "loss": 1.2647,
      "step": 4958
    },
    {
      "epoch": 0.4266540480082595,
      "grad_norm": 1.0625,
      "learning_rate": 0.000982304743755514,
      "loss": 1.368,
      "step": 4959
    },
    {
      "epoch": 0.4267400843155812,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009822976128775614,
      "loss": 1.6811,
      "step": 4960
    },
    {
      "epoch": 0.42682612062290287,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000982290480588982,
      "loss": 1.4056,
      "step": 4961
    },
    {
      "epoch": 0.42691215693022455,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009822833468897969,
      "loss": 1.9008,
      "step": 4962
    },
    {
      "epoch": 0.42699819323754623,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009822762117800267,
      "loss": 0.9177,
      "step": 4963
    },
    {
      "epoch": 0.4270842295448679,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009822690752596926,
      "loss": 1.1405,
      "step": 4964
    },
    {
      "epoch": 0.42717026585218965,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009822619373288152,
      "loss": 1.2184,
      "step": 4965
    },
    {
      "epoch": 0.4272563021595113,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009822547979874152,
      "loss": 1.7657,
      "step": 4966
    },
    {
      "epoch": 0.427342338466833,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009822476572355137,
      "loss": 1.059,
      "step": 4967
    },
    {
      "epoch": 0.4274283747741547,
      "grad_norm": 1.5,
      "learning_rate": 0.0009822405150731318,
      "loss": 1.7756,
      "step": 4968
    },
    {
      "epoch": 0.42751441108147636,
      "grad_norm": 1.5,
      "learning_rate": 0.00098223337150029,
      "loss": 0.9757,
      "step": 4969
    },
    {
      "epoch": 0.4276004473887981,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009822262265170095,
      "loss": 0.9941,
      "step": 4970
    },
    {
      "epoch": 0.4276864836961198,
      "grad_norm": 1.6875,
      "learning_rate": 0.000982219080123311,
      "loss": 1.5827,
      "step": 4971
    },
    {
      "epoch": 0.42777252000344146,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009822119323192155,
      "loss": 1.397,
      "step": 4972
    },
    {
      "epoch": 0.42785855631076314,
      "grad_norm": 1.296875,
      "learning_rate": 0.000982204783104744,
      "loss": 1.7288,
      "step": 4973
    },
    {
      "epoch": 0.4279445926180848,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009821976324799174,
      "loss": 1.4248,
      "step": 4974
    },
    {
      "epoch": 0.4280306289254065,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009821904804447562,
      "loss": 1.7589,
      "step": 4975
    },
    {
      "epoch": 0.42811666523272823,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009821833269992818,
      "loss": 1.4953,
      "step": 4976
    },
    {
      "epoch": 0.4282027015400499,
      "grad_norm": 1.4375,
      "learning_rate": 0.000982176172143515,
      "loss": 1.1281,
      "step": 4977
    },
    {
      "epoch": 0.4282887378473716,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009821690158774764,
      "loss": 0.5845,
      "step": 4978
    },
    {
      "epoch": 0.42837477415469327,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009821618582011873,
      "loss": 2.0314,
      "step": 4979
    },
    {
      "epoch": 0.42846081046201495,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009821546991146686,
      "loss": 1.0747,
      "step": 4980
    },
    {
      "epoch": 0.4285468467693367,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009821475386179413,
      "loss": 1.7987,
      "step": 4981
    },
    {
      "epoch": 0.42863288307665837,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009821403767110261,
      "loss": 1.3046,
      "step": 4982
    },
    {
      "epoch": 0.42871891938398005,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000982133213393944,
      "loss": 1.1229,
      "step": 4983
    },
    {
      "epoch": 0.4288049556913017,
      "grad_norm": 1.21875,
      "learning_rate": 0.000982126048666716,
      "loss": 1.5614,
      "step": 4984
    },
    {
      "epoch": 0.4288909919986234,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009821188825293633,
      "loss": 1.3658,
      "step": 4985
    },
    {
      "epoch": 0.4289770283059451,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009821117149819065,
      "loss": 0.8345,
      "step": 4986
    },
    {
      "epoch": 0.4290630646132668,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009821045460243667,
      "loss": 1.2886,
      "step": 4987
    },
    {
      "epoch": 0.4291491009205885,
      "grad_norm": 2.390625,
      "learning_rate": 0.000982097375656765,
      "loss": 1.6343,
      "step": 4988
    },
    {
      "epoch": 0.4292351372279102,
      "grad_norm": 1.46875,
      "learning_rate": 0.000982090203879122,
      "loss": 1.5476,
      "step": 4989
    },
    {
      "epoch": 0.42932117353523186,
      "grad_norm": 1.40625,
      "learning_rate": 0.000982083030691459,
      "loss": 1.5319,
      "step": 4990
    },
    {
      "epoch": 0.42940720984255354,
      "grad_norm": 1.265625,
      "learning_rate": 0.000982075856093797,
      "loss": 1.5992,
      "step": 4991
    },
    {
      "epoch": 0.4294932461498753,
      "grad_norm": 1.0,
      "learning_rate": 0.0009820686800861569,
      "loss": 1.1723,
      "step": 4992
    },
    {
      "epoch": 0.42957928245719695,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009820615026685596,
      "loss": 1.2246,
      "step": 4993
    },
    {
      "epoch": 0.42966531876451863,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000982054323841026,
      "loss": 1.865,
      "step": 4994
    },
    {
      "epoch": 0.4297513550718403,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009820471436035774,
      "loss": 1.694,
      "step": 4995
    },
    {
      "epoch": 0.429837391379162,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009820399619562346,
      "loss": 1.1672,
      "step": 4996
    },
    {
      "epoch": 0.4299234276864837,
      "grad_norm": 1.828125,
      "learning_rate": 0.000982032778899019,
      "loss": 1.9421,
      "step": 4997
    },
    {
      "epoch": 0.4300094639938054,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009820255944319508,
      "loss": 1.287,
      "step": 4998
    },
    {
      "epoch": 0.4300955003011271,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009820184085550516,
      "loss": 1.9916,
      "step": 4999
    },
    {
      "epoch": 0.43018153660844877,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009820112212683425,
      "loss": 1.4729,
      "step": 5000
    },
    {
      "epoch": 0.43026757291577045,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009820040325718442,
      "loss": 1.0936,
      "step": 5001
    },
    {
      "epoch": 0.4303536092230921,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009819968424655777,
      "loss": 1.9263,
      "step": 5002
    },
    {
      "epoch": 0.43043964553041386,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009819896509495644,
      "loss": 1.1066,
      "step": 5003
    },
    {
      "epoch": 0.43052568183773554,
      "grad_norm": 2.078125,
      "learning_rate": 0.000981982458023825,
      "loss": 1.105,
      "step": 5004
    },
    {
      "epoch": 0.4306117181450572,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009819752636883808,
      "loss": 1.6749,
      "step": 5005
    },
    {
      "epoch": 0.4306977544523789,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009819680679432525,
      "loss": 0.9748,
      "step": 5006
    },
    {
      "epoch": 0.4307837907597006,
      "grad_norm": 1.375,
      "learning_rate": 0.0009819608707884613,
      "loss": 1.6134,
      "step": 5007
    },
    {
      "epoch": 0.43086982706702226,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009819536722240284,
      "loss": 2.0298,
      "step": 5008
    },
    {
      "epoch": 0.430955863374344,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009819464722499748,
      "loss": 1.3381,
      "step": 5009
    },
    {
      "epoch": 0.4310418996816657,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009819392708663216,
      "loss": 1.8,
      "step": 5010
    },
    {
      "epoch": 0.43112793598898735,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009819320680730895,
      "loss": 0.9935,
      "step": 5011
    },
    {
      "epoch": 0.43121397229630903,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009819248638703,
      "loss": 0.7863,
      "step": 5012
    },
    {
      "epoch": 0.4313000086036307,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000981917658257974,
      "loss": 1.3513,
      "step": 5013
    },
    {
      "epoch": 0.43138604491095245,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009819104512361326,
      "loss": 1.5609,
      "step": 5014
    },
    {
      "epoch": 0.43147208121827413,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009819032428047968,
      "loss": 1.7128,
      "step": 5015
    },
    {
      "epoch": 0.4315581175255958,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009818960329639876,
      "loss": 2.2841,
      "step": 5016
    },
    {
      "epoch": 0.4316441538329175,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009818888217137263,
      "loss": 0.9106,
      "step": 5017
    },
    {
      "epoch": 0.43173019014023917,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000981881609054034,
      "loss": 1.4099,
      "step": 5018
    },
    {
      "epoch": 0.43181622644756085,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009818743949849317,
      "loss": 1.3645,
      "step": 5019
    },
    {
      "epoch": 0.4319022627548826,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009818671795064405,
      "loss": 0.9291,
      "step": 5020
    },
    {
      "epoch": 0.43198829906220426,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009818599626185815,
      "loss": 1.2236,
      "step": 5021
    },
    {
      "epoch": 0.43207433536952594,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009818527443213759,
      "loss": 1.7357,
      "step": 5022
    },
    {
      "epoch": 0.4321603716768476,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009818455246148447,
      "loss": 1.4077,
      "step": 5023
    },
    {
      "epoch": 0.4322464079841693,
      "grad_norm": 1.390625,
      "learning_rate": 0.000981838303499009,
      "loss": 1.837,
      "step": 5024
    },
    {
      "epoch": 0.43233244429149104,
      "grad_norm": 1.296875,
      "learning_rate": 0.00098183108097389,
      "loss": 0.958,
      "step": 5025
    },
    {
      "epoch": 0.4324184805988127,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009818238570395087,
      "loss": 1.1859,
      "step": 5026
    },
    {
      "epoch": 0.4325045169061344,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009818166316958865,
      "loss": 1.832,
      "step": 5027
    },
    {
      "epoch": 0.4325905532134561,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009818094049430443,
      "loss": 1.4045,
      "step": 5028
    },
    {
      "epoch": 0.43267658952077775,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009818021767810034,
      "loss": 1.5656,
      "step": 5029
    },
    {
      "epoch": 0.43276262582809943,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009817949472097846,
      "loss": 1.8997,
      "step": 5030
    },
    {
      "epoch": 0.43284866213542117,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009817877162294096,
      "loss": 1.2602,
      "step": 5031
    },
    {
      "epoch": 0.43293469844274285,
      "grad_norm": 1.578125,
      "learning_rate": 0.000981780483839899,
      "loss": 1.1618,
      "step": 5032
    },
    {
      "epoch": 0.43302073475006453,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009817732500412742,
      "loss": 1.8983,
      "step": 5033
    },
    {
      "epoch": 0.4331067710573862,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009817660148335563,
      "loss": 1.1254,
      "step": 5034
    },
    {
      "epoch": 0.4331928073647079,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009817587782167667,
      "loss": 1.2048,
      "step": 5035
    },
    {
      "epoch": 0.4332788436720296,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009817515401909262,
      "loss": 1.5271,
      "step": 5036
    },
    {
      "epoch": 0.4333648799793513,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009817443007560562,
      "loss": 1.3062,
      "step": 5037
    },
    {
      "epoch": 0.433450916286673,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009817370599121778,
      "loss": 1.5752,
      "step": 5038
    },
    {
      "epoch": 0.43353695259399466,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009817298176593122,
      "loss": 1.776,
      "step": 5039
    },
    {
      "epoch": 0.43362298890131634,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009817225739974806,
      "loss": 0.8203,
      "step": 5040
    },
    {
      "epoch": 0.433709025208638,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009817153289267042,
      "loss": 1.2688,
      "step": 5041
    },
    {
      "epoch": 0.43379506151595976,
      "grad_norm": 1.515625,
      "learning_rate": 0.000981708082447004,
      "loss": 1.9168,
      "step": 5042
    },
    {
      "epoch": 0.43388109782328144,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009817008345584016,
      "loss": 2.0758,
      "step": 5043
    },
    {
      "epoch": 0.4339671341306031,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009816935852609177,
      "loss": 2.1267,
      "step": 5044
    },
    {
      "epoch": 0.4340531704379248,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009816863345545738,
      "loss": 0.8375,
      "step": 5045
    },
    {
      "epoch": 0.4341392067452465,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000981679082439391,
      "loss": 1.9114,
      "step": 5046
    },
    {
      "epoch": 0.4342252430525682,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009816718289153907,
      "loss": 1.5259,
      "step": 5047
    },
    {
      "epoch": 0.4343112793598899,
      "grad_norm": 1.328125,
      "learning_rate": 0.000981664573982594,
      "loss": 1.4713,
      "step": 5048
    },
    {
      "epoch": 0.43439731566721157,
      "grad_norm": 1.671875,
      "learning_rate": 0.000981657317641022,
      "loss": 1.7256,
      "step": 5049
    },
    {
      "epoch": 0.43448335197453325,
      "grad_norm": 1.3125,
      "learning_rate": 0.000981650059890696,
      "loss": 1.3345,
      "step": 5050
    },
    {
      "epoch": 0.43456938828185493,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009816428007316373,
      "loss": 1.2667,
      "step": 5051
    },
    {
      "epoch": 0.4346554245891766,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000981635540163867,
      "loss": 0.9911,
      "step": 5052
    },
    {
      "epoch": 0.43474146089649834,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009816282781874067,
      "loss": 1.5287,
      "step": 5053
    },
    {
      "epoch": 0.43482749720382,
      "grad_norm": 1.140625,
      "learning_rate": 0.000981621014802277,
      "loss": 1.1192,
      "step": 5054
    },
    {
      "epoch": 0.4349135335111417,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009816137500084996,
      "loss": 1.2097,
      "step": 5055
    },
    {
      "epoch": 0.4349995698184634,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009816064838060958,
      "loss": 1.0573,
      "step": 5056
    },
    {
      "epoch": 0.43508560612578506,
      "grad_norm": 1.125,
      "learning_rate": 0.0009815992161950867,
      "loss": 0.8351,
      "step": 5057
    },
    {
      "epoch": 0.4351716424331068,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009815919471754934,
      "loss": 1.5575,
      "step": 5058
    },
    {
      "epoch": 0.4352576787404285,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009815846767473374,
      "loss": 1.4116,
      "step": 5059
    },
    {
      "epoch": 0.43534371504775016,
      "grad_norm": 1.125,
      "learning_rate": 0.00098157740491064,
      "loss": 0.9103,
      "step": 5060
    },
    {
      "epoch": 0.43542975135507184,
      "grad_norm": 1.09375,
      "learning_rate": 0.000981570131665422,
      "loss": 1.1972,
      "step": 5061
    },
    {
      "epoch": 0.4355157876623935,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009815628570117054,
      "loss": 1.0926,
      "step": 5062
    },
    {
      "epoch": 0.4356018239697152,
      "grad_norm": 1.65625,
      "learning_rate": 0.000981555580949511,
      "loss": 1.8175,
      "step": 5063
    },
    {
      "epoch": 0.43568786027703693,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00098154830347886,
      "loss": 1.7548,
      "step": 5064
    },
    {
      "epoch": 0.4357738965843586,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000981541024599774,
      "loss": 1.0466,
      "step": 5065
    },
    {
      "epoch": 0.4358599328916803,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009815337443122743,
      "loss": 1.7347,
      "step": 5066
    },
    {
      "epoch": 0.43594596919900197,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009815264626163818,
      "loss": 1.5947,
      "step": 5067
    },
    {
      "epoch": 0.43603200550632365,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009815191795121182,
      "loss": 0.9926,
      "step": 5068
    },
    {
      "epoch": 0.4361180418136454,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009815118949995046,
      "loss": 0.7476,
      "step": 5069
    },
    {
      "epoch": 0.43620407812096706,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009815046090785623,
      "loss": 0.825,
      "step": 5070
    },
    {
      "epoch": 0.43629011442828874,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009814973217493128,
      "loss": 0.9349,
      "step": 5071
    },
    {
      "epoch": 0.4363761507356104,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009814900330117771,
      "loss": 1.5937,
      "step": 5072
    },
    {
      "epoch": 0.4364621870429321,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000981482742865977,
      "loss": 1.9561,
      "step": 5073
    },
    {
      "epoch": 0.4365482233502538,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009814754513119332,
      "loss": 1.2347,
      "step": 5074
    },
    {
      "epoch": 0.4366342596575755,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009814681583496675,
      "loss": 1.4879,
      "step": 5075
    },
    {
      "epoch": 0.4367202959648972,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000981460863979201,
      "loss": 1.2349,
      "step": 5076
    },
    {
      "epoch": 0.4368063322722189,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009814535682005552,
      "loss": 1.5692,
      "step": 5077
    },
    {
      "epoch": 0.43689236857954056,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009814462710137513,
      "loss": 1.138,
      "step": 5078
    },
    {
      "epoch": 0.43697840488686224,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000981438972418811,
      "loss": 1.3437,
      "step": 5079
    },
    {
      "epoch": 0.437064441194184,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009814316724157549,
      "loss": 1.4014,
      "step": 5080
    },
    {
      "epoch": 0.43715047750150565,
      "grad_norm": 1.734375,
      "learning_rate": 0.000981424371004605,
      "loss": 2.1857,
      "step": 5081
    },
    {
      "epoch": 0.43723651380882733,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009814170681853823,
      "loss": 1.6404,
      "step": 5082
    },
    {
      "epoch": 0.437322550116149,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009814097639581083,
      "loss": 1.6034,
      "step": 5083
    },
    {
      "epoch": 0.4374085864234707,
      "grad_norm": 1.375,
      "learning_rate": 0.0009814024583228045,
      "loss": 1.5541,
      "step": 5084
    },
    {
      "epoch": 0.43749462273079237,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009813951512794921,
      "loss": 1.091,
      "step": 5085
    },
    {
      "epoch": 0.4375806590381141,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009813878428281927,
      "loss": 1.235,
      "step": 5086
    },
    {
      "epoch": 0.4376666953454358,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009813805329689273,
      "loss": 1.3258,
      "step": 5087
    },
    {
      "epoch": 0.43775273165275747,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009813732217017173,
      "loss": 1.2296,
      "step": 5088
    },
    {
      "epoch": 0.43783876796007914,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009813659090265844,
      "loss": 1.2382,
      "step": 5089
    },
    {
      "epoch": 0.4379248042674008,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009813585949435498,
      "loss": 1.1747,
      "step": 5090
    },
    {
      "epoch": 0.43801084057472256,
      "grad_norm": 0.96875,
      "learning_rate": 0.000981351279452635,
      "loss": 1.1878,
      "step": 5091
    },
    {
      "epoch": 0.43809687688204424,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009813439625538615,
      "loss": 1.2628,
      "step": 5092
    },
    {
      "epoch": 0.4381829131893659,
      "grad_norm": 1.25,
      "learning_rate": 0.0009813366442472501,
      "loss": 1.5994,
      "step": 5093
    },
    {
      "epoch": 0.4382689494966876,
      "grad_norm": 1.78125,
      "learning_rate": 0.000981329324532823,
      "loss": 1.5412,
      "step": 5094
    },
    {
      "epoch": 0.4383549858040093,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000981322003410601,
      "loss": 1.5197,
      "step": 5095
    },
    {
      "epoch": 0.43844102211133096,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009813146808806059,
      "loss": 1.6514,
      "step": 5096
    },
    {
      "epoch": 0.4385270584186527,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009813073569428588,
      "loss": 0.9213,
      "step": 5097
    },
    {
      "epoch": 0.4386130947259744,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009813000315973814,
      "loss": 1.0696,
      "step": 5098
    },
    {
      "epoch": 0.43869913103329605,
      "grad_norm": 1.640625,
      "learning_rate": 0.000981292704844195,
      "loss": 1.5513,
      "step": 5099
    },
    {
      "epoch": 0.43878516734061773,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009812853766833212,
      "loss": 1.2128,
      "step": 5100
    },
    {
      "epoch": 0.4388712036479394,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009812780471147812,
      "loss": 1.5995,
      "step": 5101
    },
    {
      "epoch": 0.43895723995526115,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009812707161385964,
      "loss": 1.1056,
      "step": 5102
    },
    {
      "epoch": 0.4390432762625828,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009812633837547884,
      "loss": 1.7833,
      "step": 5103
    },
    {
      "epoch": 0.4391293125699045,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009812560499633785,
      "loss": 1.8956,
      "step": 5104
    },
    {
      "epoch": 0.4392153488772262,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009812487147643884,
      "loss": 0.9708,
      "step": 5105
    },
    {
      "epoch": 0.43930138518454787,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009812413781578393,
      "loss": 1.9433,
      "step": 5106
    },
    {
      "epoch": 0.43938742149186955,
      "grad_norm": 1.125,
      "learning_rate": 0.000981234040143753,
      "loss": 1.7587,
      "step": 5107
    },
    {
      "epoch": 0.4394734577991913,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009812267007221506,
      "loss": 1.0532,
      "step": 5108
    },
    {
      "epoch": 0.43955949410651296,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009812193598930537,
      "loss": 1.7698,
      "step": 5109
    },
    {
      "epoch": 0.43964553041383464,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009812120176564837,
      "loss": 2.1897,
      "step": 5110
    },
    {
      "epoch": 0.4397315667211563,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009812046740124622,
      "loss": 1.7552,
      "step": 5111
    },
    {
      "epoch": 0.439817603028478,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009811973289610106,
      "loss": 1.2251,
      "step": 5112
    },
    {
      "epoch": 0.43990363933579973,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009811899825021503,
      "loss": 1.0936,
      "step": 5113
    },
    {
      "epoch": 0.4399896756431214,
      "grad_norm": 1.65625,
      "learning_rate": 0.000981182634635903,
      "loss": 1.3661,
      "step": 5114
    },
    {
      "epoch": 0.4400757119504431,
      "grad_norm": 1.3125,
      "learning_rate": 0.00098117528536229,
      "loss": 1.9335,
      "step": 5115
    },
    {
      "epoch": 0.4401617482577648,
      "grad_norm": 1.578125,
      "learning_rate": 0.000981167934681333,
      "loss": 1.5899,
      "step": 5116
    },
    {
      "epoch": 0.44024778456508645,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009811605825930532,
      "loss": 1.1902,
      "step": 5117
    },
    {
      "epoch": 0.44033382087240813,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009811532290974726,
      "loss": 1.2219,
      "step": 5118
    },
    {
      "epoch": 0.44041985717972987,
      "grad_norm": 1.171875,
      "learning_rate": 0.000981145874194612,
      "loss": 1.6202,
      "step": 5119
    },
    {
      "epoch": 0.44050589348705155,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009811385178844934,
      "loss": 1.1998,
      "step": 5120
    },
    {
      "epoch": 0.4405919297943732,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009811311601671383,
      "loss": 1.0077,
      "step": 5121
    },
    {
      "epoch": 0.4406779661016949,
      "grad_norm": 1.59375,
      "learning_rate": 0.000981123801042568,
      "loss": 0.8724,
      "step": 5122
    },
    {
      "epoch": 0.4407640024090166,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009811164405108045,
      "loss": 1.0376,
      "step": 5123
    },
    {
      "epoch": 0.4408500387163383,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009811090785718686,
      "loss": 1.3714,
      "step": 5124
    },
    {
      "epoch": 0.44093607502366,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009811017152257823,
      "loss": 1.4987,
      "step": 5125
    },
    {
      "epoch": 0.4410221113309817,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000981094350472567,
      "loss": 1.4631,
      "step": 5126
    },
    {
      "epoch": 0.44110814763830336,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009810869843122445,
      "loss": 0.8649,
      "step": 5127
    },
    {
      "epoch": 0.44119418394562504,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009810796167448361,
      "loss": 1.5071,
      "step": 5128
    },
    {
      "epoch": 0.4412802202529467,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009810722477703634,
      "loss": 1.313,
      "step": 5129
    },
    {
      "epoch": 0.44136625656026846,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009810648773888478,
      "loss": 1.4499,
      "step": 5130
    },
    {
      "epoch": 0.44145229286759013,
      "grad_norm": 1.765625,
      "learning_rate": 0.000981057505600311,
      "loss": 1.1838,
      "step": 5131
    },
    {
      "epoch": 0.4415383291749118,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009810501324047747,
      "loss": 1.7938,
      "step": 5132
    },
    {
      "epoch": 0.4416243654822335,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00098104275780226,
      "loss": 1.9698,
      "step": 5133
    },
    {
      "epoch": 0.4417104017895552,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009810353817927893,
      "loss": 2.0567,
      "step": 5134
    },
    {
      "epoch": 0.4417964380968769,
      "grad_norm": 1.125,
      "learning_rate": 0.0009810280043763833,
      "loss": 0.9292,
      "step": 5135
    },
    {
      "epoch": 0.4418824744041986,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000981020625553064,
      "loss": 1.907,
      "step": 5136
    },
    {
      "epoch": 0.44196851071152027,
      "grad_norm": 1.203125,
      "learning_rate": 0.000981013245322853,
      "loss": 1.2309,
      "step": 5137
    },
    {
      "epoch": 0.44205454701884195,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009810058636857716,
      "loss": 1.5466,
      "step": 5138
    },
    {
      "epoch": 0.4421405833261636,
      "grad_norm": 1.546875,
      "learning_rate": 0.000980998480641842,
      "loss": 1.7864,
      "step": 5139
    },
    {
      "epoch": 0.4422266196334853,
      "grad_norm": 1.5,
      "learning_rate": 0.0009809910961910851,
      "loss": 1.9589,
      "step": 5140
    },
    {
      "epoch": 0.44231265594080704,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009809837103335228,
      "loss": 1.1722,
      "step": 5141
    },
    {
      "epoch": 0.4423986922481287,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009809763230691768,
      "loss": 1.1625,
      "step": 5142
    },
    {
      "epoch": 0.4424847285554504,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009809689343980685,
      "loss": 0.9935,
      "step": 5143
    },
    {
      "epoch": 0.4425707648627721,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009809615443202196,
      "loss": 1.8918,
      "step": 5144
    },
    {
      "epoch": 0.44265680117009376,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009809541528356517,
      "loss": 1.8082,
      "step": 5145
    },
    {
      "epoch": 0.4427428374774155,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009809467599443866,
      "loss": 1.0895,
      "step": 5146
    },
    {
      "epoch": 0.4428288737847372,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009809393656464457,
      "loss": 0.9608,
      "step": 5147
    },
    {
      "epoch": 0.44291491009205886,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009809319699418506,
      "loss": 1.957,
      "step": 5148
    },
    {
      "epoch": 0.44300094639938054,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009809245728306232,
      "loss": 0.8895,
      "step": 5149
    },
    {
      "epoch": 0.4430869827067022,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000980917174312785,
      "loss": 1.3968,
      "step": 5150
    },
    {
      "epoch": 0.4431730190140239,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009809097743883573,
      "loss": 1.5245,
      "step": 5151
    },
    {
      "epoch": 0.44325905532134563,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009809023730573622,
      "loss": 1.7447,
      "step": 5152
    },
    {
      "epoch": 0.4433450916286673,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009808949703198212,
      "loss": 1.8309,
      "step": 5153
    },
    {
      "epoch": 0.443431127935989,
      "grad_norm": 1.046875,
      "learning_rate": 0.000980887566175756,
      "loss": 0.8157,
      "step": 5154
    },
    {
      "epoch": 0.44351716424331067,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000980880160625188,
      "loss": 1.3646,
      "step": 5155
    },
    {
      "epoch": 0.44360320055063235,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009808727536681392,
      "loss": 1.1871,
      "step": 5156
    },
    {
      "epoch": 0.4436892368579541,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009808653453046313,
      "loss": 1.4888,
      "step": 5157
    },
    {
      "epoch": 0.44377527316527576,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009808579355346854,
      "loss": 1.4484,
      "step": 5158
    },
    {
      "epoch": 0.44386130947259744,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009808505243583238,
      "loss": 1.3205,
      "step": 5159
    },
    {
      "epoch": 0.4439473457799191,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009808431117755679,
      "loss": 1.3,
      "step": 5160
    },
    {
      "epoch": 0.4440333820872408,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009808356977864395,
      "loss": 1.9583,
      "step": 5161
    },
    {
      "epoch": 0.4441194183945625,
      "grad_norm": 1.90625,
      "learning_rate": 0.00098082828239096,
      "loss": 1.2325,
      "step": 5162
    },
    {
      "epoch": 0.4442054547018842,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009808208655891514,
      "loss": 2.0046,
      "step": 5163
    },
    {
      "epoch": 0.4442914910092059,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009808134473810353,
      "loss": 1.4816,
      "step": 5164
    },
    {
      "epoch": 0.4443775273165276,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009808060277666333,
      "loss": 1.4133,
      "step": 5165
    },
    {
      "epoch": 0.44446356362384926,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009807986067459672,
      "loss": 1.0757,
      "step": 5166
    },
    {
      "epoch": 0.44454959993117094,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009807911843190589,
      "loss": 1.4814,
      "step": 5167
    },
    {
      "epoch": 0.44463563623849267,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009807837604859295,
      "loss": 1.1947,
      "step": 5168
    },
    {
      "epoch": 0.44472167254581435,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009807763352466016,
      "loss": 0.7839,
      "step": 5169
    },
    {
      "epoch": 0.44480770885313603,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000980768908601096,
      "loss": 1.3785,
      "step": 5170
    },
    {
      "epoch": 0.4448937451604577,
      "grad_norm": 1.8515625,
      "learning_rate": 0.000980761480549435,
      "loss": 1.3819,
      "step": 5171
    },
    {
      "epoch": 0.4449797814677794,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009807540510916402,
      "loss": 2.2714,
      "step": 5172
    },
    {
      "epoch": 0.44506581777510107,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009807466202277333,
      "loss": 1.2093,
      "step": 5173
    },
    {
      "epoch": 0.4451518540824228,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000980739187957736,
      "loss": 1.3891,
      "step": 5174
    },
    {
      "epoch": 0.4452378903897445,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00098073175428167,
      "loss": 1.4333,
      "step": 5175
    },
    {
      "epoch": 0.44532392669706616,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009807243191995573,
      "loss": 2.1817,
      "step": 5176
    },
    {
      "epoch": 0.44540996300438784,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009807168827114193,
      "loss": 1.1744,
      "step": 5177
    },
    {
      "epoch": 0.4454959993117095,
      "grad_norm": 1.390625,
      "learning_rate": 0.000980709444817278,
      "loss": 1.2968,
      "step": 5178
    },
    {
      "epoch": 0.44558203561903126,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009807020055171548,
      "loss": 1.3356,
      "step": 5179
    },
    {
      "epoch": 0.44566807192635294,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000980694564811072,
      "loss": 0.968,
      "step": 5180
    },
    {
      "epoch": 0.4457541082336746,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009806871226990513,
      "loss": 1.7784,
      "step": 5181
    },
    {
      "epoch": 0.4458401445409963,
      "grad_norm": 1.5,
      "learning_rate": 0.0009806796791811138,
      "loss": 1.672,
      "step": 5182
    },
    {
      "epoch": 0.445926180848318,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009806722342572821,
      "loss": 0.7143,
      "step": 5183
    },
    {
      "epoch": 0.44601221715563966,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009806647879275774,
      "loss": 2.3457,
      "step": 5184
    },
    {
      "epoch": 0.4460982534629614,
      "grad_norm": 1.25,
      "learning_rate": 0.0009806573401920217,
      "loss": 1.4177,
      "step": 5185
    },
    {
      "epoch": 0.44618428977028307,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009806498910506368,
      "loss": 1.3275,
      "step": 5186
    },
    {
      "epoch": 0.44627032607760475,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009806424405034444,
      "loss": 1.7678,
      "step": 5187
    },
    {
      "epoch": 0.44635636238492643,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009806349885504665,
      "loss": 1.0465,
      "step": 5188
    },
    {
      "epoch": 0.4464423986922481,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009806275351917245,
      "loss": 0.9985,
      "step": 5189
    },
    {
      "epoch": 0.44652843499956985,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009806200804272406,
      "loss": 1.0851,
      "step": 5190
    },
    {
      "epoch": 0.4466144713068915,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009806126242570363,
      "loss": 0.8753,
      "step": 5191
    },
    {
      "epoch": 0.4467005076142132,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009806051666811336,
      "loss": 1.1161,
      "step": 5192
    },
    {
      "epoch": 0.4467865439215349,
      "grad_norm": 2.671875,
      "learning_rate": 0.0009805977076995543,
      "loss": 1.6362,
      "step": 5193
    },
    {
      "epoch": 0.44687258022885656,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00098059024731232,
      "loss": 0.679,
      "step": 5194
    },
    {
      "epoch": 0.44695861653617824,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009805827855194531,
      "loss": 1.8842,
      "step": 5195
    },
    {
      "epoch": 0.4470446528435,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009805753223209746,
      "loss": 1.7804,
      "step": 5196
    },
    {
      "epoch": 0.44713068915082166,
      "grad_norm": 1.203125,
      "learning_rate": 0.000980567857716907,
      "loss": 1.2887,
      "step": 5197
    },
    {
      "epoch": 0.44721672545814334,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009805603917072718,
      "loss": 1.4381,
      "step": 5198
    },
    {
      "epoch": 0.447302761765465,
      "grad_norm": 1.84375,
      "learning_rate": 0.000980552924292091,
      "loss": 1.4568,
      "step": 5199
    },
    {
      "epoch": 0.4473887980727867,
      "grad_norm": 1.625,
      "learning_rate": 0.0009805454554713863,
      "loss": 1.1056,
      "step": 5200
    },
    {
      "epoch": 0.44747483438010843,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009805379852451796,
      "loss": 0.6882,
      "step": 5201
    },
    {
      "epoch": 0.4475608706874301,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009805305136134927,
      "loss": 0.872,
      "step": 5202
    },
    {
      "epoch": 0.4476469069947518,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009805230405763475,
      "loss": 1.7437,
      "step": 5203
    },
    {
      "epoch": 0.44773294330207347,
      "grad_norm": 1.296875,
      "learning_rate": 0.000980515566133766,
      "loss": 1.5181,
      "step": 5204
    },
    {
      "epoch": 0.44781897960939515,
      "grad_norm": 1.515625,
      "learning_rate": 0.00098050809028577,
      "loss": 0.9918,
      "step": 5205
    },
    {
      "epoch": 0.44790501591671683,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000980500613032381,
      "loss": 1.1914,
      "step": 5206
    },
    {
      "epoch": 0.44799105222403857,
      "grad_norm": 1.75,
      "learning_rate": 0.0009804931343736212,
      "loss": 1.8807,
      "step": 5207
    },
    {
      "epoch": 0.44807708853136025,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009804856543095125,
      "loss": 0.899,
      "step": 5208
    },
    {
      "epoch": 0.4481631248386819,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009804781728400768,
      "loss": 1.4563,
      "step": 5209
    },
    {
      "epoch": 0.4482491611460036,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009804706899653357,
      "loss": 1.7554,
      "step": 5210
    },
    {
      "epoch": 0.4483351974533253,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009804632056853117,
      "loss": 1.5139,
      "step": 5211
    },
    {
      "epoch": 0.448421233760647,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009804557200000256,
      "loss": 1.1635,
      "step": 5212
    },
    {
      "epoch": 0.4485072700679687,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009804482329095004,
      "loss": 1.2757,
      "step": 5213
    },
    {
      "epoch": 0.4485933063752904,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009804407444137576,
      "loss": 1.6962,
      "step": 5214
    },
    {
      "epoch": 0.44867934268261206,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000980433254512819,
      "loss": 2.0272,
      "step": 5215
    },
    {
      "epoch": 0.44876537898993374,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009804257632067064,
      "loss": 1.192,
      "step": 5216
    },
    {
      "epoch": 0.4488514152972554,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000980418270495442,
      "loss": 0.9277,
      "step": 5217
    },
    {
      "epoch": 0.44893745160457715,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009804107763790476,
      "loss": 1.236,
      "step": 5218
    },
    {
      "epoch": 0.44902348791189883,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009804032808575453,
      "loss": 1.0639,
      "step": 5219
    },
    {
      "epoch": 0.4491095242192205,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009803957839309566,
      "loss": 0.9924,
      "step": 5220
    },
    {
      "epoch": 0.4491955605265422,
      "grad_norm": 2.609375,
      "learning_rate": 0.0009803882855993035,
      "loss": 2.0662,
      "step": 5221
    },
    {
      "epoch": 0.44928159683386387,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009803807858626084,
      "loss": 1.9571,
      "step": 5222
    },
    {
      "epoch": 0.4493676331411856,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009803732847208928,
      "loss": 1.8088,
      "step": 5223
    },
    {
      "epoch": 0.4494536694485073,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000980365782174179,
      "loss": 1.4014,
      "step": 5224
    },
    {
      "epoch": 0.44953970575582897,
      "grad_norm": 3.40625,
      "learning_rate": 0.0009803582782224884,
      "loss": 1.5853,
      "step": 5225
    },
    {
      "epoch": 0.44962574206315065,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009803507728658434,
      "loss": 1.2778,
      "step": 5226
    },
    {
      "epoch": 0.4497117783704723,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009803432661042656,
      "loss": 1.1092,
      "step": 5227
    },
    {
      "epoch": 0.449797814677794,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009803357579377776,
      "loss": 1.429,
      "step": 5228
    },
    {
      "epoch": 0.44988385098511574,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009803282483664007,
      "loss": 1.6221,
      "step": 5229
    },
    {
      "epoch": 0.4499698872924374,
      "grad_norm": 1.3125,
      "learning_rate": 0.000980320737390157,
      "loss": 1.6392,
      "step": 5230
    },
    {
      "epoch": 0.4500559235997591,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009803132250090687,
      "loss": 0.9631,
      "step": 5231
    },
    {
      "epoch": 0.4501419599070808,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009803057112231575,
      "loss": 1.2243,
      "step": 5232
    },
    {
      "epoch": 0.45022799621440246,
      "grad_norm": 1.0,
      "learning_rate": 0.0009802981960324456,
      "loss": 1.4365,
      "step": 5233
    },
    {
      "epoch": 0.4503140325217242,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009802906794369548,
      "loss": 1.0029,
      "step": 5234
    },
    {
      "epoch": 0.4504000688290459,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009802831614367073,
      "loss": 1.4556,
      "step": 5235
    },
    {
      "epoch": 0.45048610513636755,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000980275642031725,
      "loss": 1.718,
      "step": 5236
    },
    {
      "epoch": 0.45057214144368923,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009802681212220297,
      "loss": 1.507,
      "step": 5237
    },
    {
      "epoch": 0.4506581777510109,
      "grad_norm": 1.25,
      "learning_rate": 0.0009802605990076436,
      "loss": 1.063,
      "step": 5238
    },
    {
      "epoch": 0.4507442140583326,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009802530753885888,
      "loss": 1.9179,
      "step": 5239
    },
    {
      "epoch": 0.45083025036565433,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000980245550364887,
      "loss": 0.8058,
      "step": 5240
    },
    {
      "epoch": 0.450916286672976,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009802380239365602,
      "loss": 1.3389,
      "step": 5241
    },
    {
      "epoch": 0.4510023229802977,
      "grad_norm": 1.71875,
      "learning_rate": 0.000980230496103631,
      "loss": 1.8149,
      "step": 5242
    },
    {
      "epoch": 0.45108835928761937,
      "grad_norm": 1.125,
      "learning_rate": 0.0009802229668661206,
      "loss": 1.8015,
      "step": 5243
    },
    {
      "epoch": 0.45117439559494105,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009802154362240517,
      "loss": 1.9778,
      "step": 5244
    },
    {
      "epoch": 0.4512604319022628,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009802079041774459,
      "loss": 1.9814,
      "step": 5245
    },
    {
      "epoch": 0.45134646820958446,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009802003707263254,
      "loss": 1.5869,
      "step": 5246
    },
    {
      "epoch": 0.45143250451690614,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009801928358707122,
      "loss": 2.4447,
      "step": 5247
    },
    {
      "epoch": 0.4515185408242278,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009801852996106284,
      "loss": 1.3765,
      "step": 5248
    },
    {
      "epoch": 0.4516045771315495,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000980177761946096,
      "loss": 1.2608,
      "step": 5249
    },
    {
      "epoch": 0.4516906134388712,
      "grad_norm": 1.109375,
      "learning_rate": 0.000980170222877137,
      "loss": 1.7579,
      "step": 5250
    },
    {
      "epoch": 0.4517766497461929,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009801626824037734,
      "loss": 1.0078,
      "step": 5251
    },
    {
      "epoch": 0.4518626860535146,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009801551405260276,
      "loss": 1.6458,
      "step": 5252
    },
    {
      "epoch": 0.4519487223608363,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009801475972439212,
      "loss": 2.1248,
      "step": 5253
    },
    {
      "epoch": 0.45203475866815795,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009801400525574764,
      "loss": 1.6276,
      "step": 5254
    },
    {
      "epoch": 0.45212079497547963,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009801325064667154,
      "loss": 1.3021,
      "step": 5255
    },
    {
      "epoch": 0.45220683128280137,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009801249589716603,
      "loss": 1.5602,
      "step": 5256
    },
    {
      "epoch": 0.45229286759012305,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009801174100723328,
      "loss": 0.537,
      "step": 5257
    },
    {
      "epoch": 0.45237890389744473,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009801098597687556,
      "loss": 1.076,
      "step": 5258
    },
    {
      "epoch": 0.4524649402047664,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009801023080609503,
      "loss": 1.8107,
      "step": 5259
    },
    {
      "epoch": 0.4525509765120881,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009800947549489392,
      "loss": 1.9007,
      "step": 5260
    },
    {
      "epoch": 0.45263701281940977,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009800872004327441,
      "loss": 1.8286,
      "step": 5261
    },
    {
      "epoch": 0.4527230491267315,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009800796445123875,
      "loss": 1.6476,
      "step": 5262
    },
    {
      "epoch": 0.4528090854340532,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009800720871878913,
      "loss": 1.4073,
      "step": 5263
    },
    {
      "epoch": 0.45289512174137486,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009800645284592775,
      "loss": 1.6345,
      "step": 5264
    },
    {
      "epoch": 0.45298115804869654,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009800569683265682,
      "loss": 1.3995,
      "step": 5265
    },
    {
      "epoch": 0.4530671943560182,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009800494067897859,
      "loss": 1.3256,
      "step": 5266
    },
    {
      "epoch": 0.45315323066333996,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009800418438489524,
      "loss": 1.2783,
      "step": 5267
    },
    {
      "epoch": 0.45323926697066164,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009800342795040896,
      "loss": 1.7694,
      "step": 5268
    },
    {
      "epoch": 0.4533253032779833,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009800267137552205,
      "loss": 1.8802,
      "step": 5269
    },
    {
      "epoch": 0.453411339585305,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000980019146602366,
      "loss": 1.1273,
      "step": 5270
    },
    {
      "epoch": 0.4534973758926267,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009800115780455492,
      "loss": 0.923,
      "step": 5271
    },
    {
      "epoch": 0.45358341219994835,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009800040080847918,
      "loss": 1.3647,
      "step": 5272
    },
    {
      "epoch": 0.4536694485072701,
      "grad_norm": 1.125,
      "learning_rate": 0.000979996436720116,
      "loss": 1.4666,
      "step": 5273
    },
    {
      "epoch": 0.45375548481459177,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000979988863951544,
      "loss": 1.4212,
      "step": 5274
    },
    {
      "epoch": 0.45384152112191345,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009799812897790978,
      "loss": 1.4799,
      "step": 5275
    },
    {
      "epoch": 0.45392755742923513,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009799737142027999,
      "loss": 1.4614,
      "step": 5276
    },
    {
      "epoch": 0.4540135937365568,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009799661372226721,
      "loss": 2.0849,
      "step": 5277
    },
    {
      "epoch": 0.45409963004387854,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009799585588387367,
      "loss": 0.7091,
      "step": 5278
    },
    {
      "epoch": 0.4541856663512002,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009799509790510159,
      "loss": 1.8111,
      "step": 5279
    },
    {
      "epoch": 0.4542717026585219,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009799433978595318,
      "loss": 0.8342,
      "step": 5280
    },
    {
      "epoch": 0.4543577389658436,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009799358152643067,
      "loss": 1.9266,
      "step": 5281
    },
    {
      "epoch": 0.45444377527316526,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009799282312653627,
      "loss": 1.3904,
      "step": 5282
    },
    {
      "epoch": 0.45452981158048694,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009799206458627218,
      "loss": 1.2393,
      "step": 5283
    },
    {
      "epoch": 0.4546158478878087,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009799130590564065,
      "loss": 1.3001,
      "step": 5284
    },
    {
      "epoch": 0.45470188419513036,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009799054708464388,
      "loss": 1.538,
      "step": 5285
    },
    {
      "epoch": 0.45478792050245204,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009798978812328408,
      "loss": 1.9167,
      "step": 5286
    },
    {
      "epoch": 0.4548739568097737,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009798902902156351,
      "loss": 1.6039,
      "step": 5287
    },
    {
      "epoch": 0.4549599931170954,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009798826977948435,
      "loss": 1.1335,
      "step": 5288
    },
    {
      "epoch": 0.45504602942441713,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000979875103970488,
      "loss": 1.0328,
      "step": 5289
    },
    {
      "epoch": 0.4551320657317388,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009798675087425916,
      "loss": 0.9313,
      "step": 5290
    },
    {
      "epoch": 0.4552181020390605,
      "grad_norm": 1.0,
      "learning_rate": 0.000979859912111176,
      "loss": 0.7501,
      "step": 5291
    },
    {
      "epoch": 0.45530413834638217,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009798523140762635,
      "loss": 1.1161,
      "step": 5292
    },
    {
      "epoch": 0.45539017465370385,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009798447146378761,
      "loss": 1.4461,
      "step": 5293
    },
    {
      "epoch": 0.45547621096102553,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009798371137960363,
      "loss": 2.4439,
      "step": 5294
    },
    {
      "epoch": 0.45556224726834726,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009798295115507664,
      "loss": 1.4261,
      "step": 5295
    },
    {
      "epoch": 0.45564828357566894,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009798219079020884,
      "loss": 1.9187,
      "step": 5296
    },
    {
      "epoch": 0.4557343198829906,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009798143028500246,
      "loss": 1.5016,
      "step": 5297
    },
    {
      "epoch": 0.4558203561903123,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009798066963945973,
      "loss": 1.2889,
      "step": 5298
    },
    {
      "epoch": 0.455906392497634,
      "grad_norm": 1.125,
      "learning_rate": 0.0009797990885358288,
      "loss": 1.3399,
      "step": 5299
    },
    {
      "epoch": 0.4559924288049557,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009797914792737414,
      "loss": 1.5997,
      "step": 5300
    },
    {
      "epoch": 0.4560784651122774,
      "grad_norm": 1.484375,
      "learning_rate": 0.000979783868608357,
      "loss": 1.9601,
      "step": 5301
    },
    {
      "epoch": 0.4561645014195991,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009797762565396982,
      "loss": 1.203,
      "step": 5302
    },
    {
      "epoch": 0.45625053772692076,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009797686430677872,
      "loss": 1.5057,
      "step": 5303
    },
    {
      "epoch": 0.45633657403424244,
      "grad_norm": 1.25,
      "learning_rate": 0.0009797610281926461,
      "loss": 1.5928,
      "step": 5304
    },
    {
      "epoch": 0.4564226103415641,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009797534119142975,
      "loss": 1.4121,
      "step": 5305
    },
    {
      "epoch": 0.45650864664888585,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009797457942327633,
      "loss": 1.6703,
      "step": 5306
    },
    {
      "epoch": 0.45659468295620753,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009797381751480662,
      "loss": 1.6644,
      "step": 5307
    },
    {
      "epoch": 0.4566807192635292,
      "grad_norm": 1.828125,
      "learning_rate": 0.000979730554660228,
      "loss": 1.3573,
      "step": 5308
    },
    {
      "epoch": 0.4567667555708509,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009797229327692713,
      "loss": 1.1212,
      "step": 5309
    },
    {
      "epoch": 0.45685279187817257,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009797153094752182,
      "loss": 0.8468,
      "step": 5310
    },
    {
      "epoch": 0.4569388281854943,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009797076847780912,
      "loss": 1.4251,
      "step": 5311
    },
    {
      "epoch": 0.457024864492816,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009797000586779126,
      "loss": 2.3409,
      "step": 5312
    },
    {
      "epoch": 0.45711090080013767,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009796924311747046,
      "loss": 1.5622,
      "step": 5313
    },
    {
      "epoch": 0.45719693710745934,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009796848022684895,
      "loss": 1.8007,
      "step": 5314
    },
    {
      "epoch": 0.457282973414781,
      "grad_norm": 21.875,
      "learning_rate": 0.0009796771719592897,
      "loss": 1.4713,
      "step": 5315
    },
    {
      "epoch": 0.4573690097221027,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009796695402471275,
      "loss": 1.496,
      "step": 5316
    },
    {
      "epoch": 0.45745504602942444,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000979661907132025,
      "loss": 1.0975,
      "step": 5317
    },
    {
      "epoch": 0.4575410823367461,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009796542726140047,
      "loss": 1.0227,
      "step": 5318
    },
    {
      "epoch": 0.4576271186440678,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009796466366930891,
      "loss": 1.5909,
      "step": 5319
    },
    {
      "epoch": 0.4577131549513895,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009796389993693003,
      "loss": 2.0157,
      "step": 5320
    },
    {
      "epoch": 0.45779919125871116,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009796313606426608,
      "loss": 1.453,
      "step": 5321
    },
    {
      "epoch": 0.4578852275660329,
      "grad_norm": 1.25,
      "learning_rate": 0.0009796237205131927,
      "loss": 1.0854,
      "step": 5322
    },
    {
      "epoch": 0.4579712638733546,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009796160789809188,
      "loss": 1.156,
      "step": 5323
    },
    {
      "epoch": 0.45805730018067625,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009796084360458607,
      "loss": 1.3479,
      "step": 5324
    },
    {
      "epoch": 0.45814333648799793,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009796007917080415,
      "loss": 1.107,
      "step": 5325
    },
    {
      "epoch": 0.4582293727953196,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009795931459674833,
      "loss": 0.9378,
      "step": 5326
    },
    {
      "epoch": 0.4583154091026413,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009795854988242083,
      "loss": 0.933,
      "step": 5327
    },
    {
      "epoch": 0.458401445409963,
      "grad_norm": 1.171875,
      "learning_rate": 0.000979577850278239,
      "loss": 1.4148,
      "step": 5328
    },
    {
      "epoch": 0.4584874817172847,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009795702003295979,
      "loss": 1.145,
      "step": 5329
    },
    {
      "epoch": 0.4585735180246064,
      "grad_norm": 1.046875,
      "learning_rate": 0.000979562548978307,
      "loss": 1.4654,
      "step": 5330
    },
    {
      "epoch": 0.45865955433192807,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000979554896224389,
      "loss": 1.2411,
      "step": 5331
    },
    {
      "epoch": 0.45874559063924975,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009795472420678663,
      "loss": 1.2219,
      "step": 5332
    },
    {
      "epoch": 0.4588316269465715,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000979539586508761,
      "loss": 1.4092,
      "step": 5333
    },
    {
      "epoch": 0.45891766325389316,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009795319295470958,
      "loss": 1.0466,
      "step": 5334
    },
    {
      "epoch": 0.45900369956121484,
      "grad_norm": 1.859375,
      "learning_rate": 0.000979524271182893,
      "loss": 1.6072,
      "step": 5335
    },
    {
      "epoch": 0.4590897358685365,
      "grad_norm": 1.5,
      "learning_rate": 0.000979516611416175,
      "loss": 1.03,
      "step": 5336
    },
    {
      "epoch": 0.4591757721758582,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000979508950246964,
      "loss": 2.0749,
      "step": 5337
    },
    {
      "epoch": 0.4592618084831799,
      "grad_norm": 1.0,
      "learning_rate": 0.0009795012876752827,
      "loss": 1.0794,
      "step": 5338
    },
    {
      "epoch": 0.4593478447905016,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009794936237011533,
      "loss": 1.2577,
      "step": 5339
    },
    {
      "epoch": 0.4594338810978233,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009794859583245985,
      "loss": 1.5979,
      "step": 5340
    },
    {
      "epoch": 0.459519917405145,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009794782915456405,
      "loss": 1.3522,
      "step": 5341
    },
    {
      "epoch": 0.45960595371246665,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009794706233643017,
      "loss": 2.1359,
      "step": 5342
    },
    {
      "epoch": 0.45969199001978833,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009794629537806044,
      "loss": 2.2582,
      "step": 5343
    },
    {
      "epoch": 0.45977802632711007,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009794552827945713,
      "loss": 1.5342,
      "step": 5344
    },
    {
      "epoch": 0.45986406263443175,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000979447610406225,
      "loss": 1.9992,
      "step": 5345
    },
    {
      "epoch": 0.4599500989417534,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009794399366155873,
      "loss": 1.5535,
      "step": 5346
    },
    {
      "epoch": 0.4600361352490751,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009794322614226812,
      "loss": 1.8418,
      "step": 5347
    },
    {
      "epoch": 0.4601221715563968,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009794245848275289,
      "loss": 1.2175,
      "step": 5348
    },
    {
      "epoch": 0.46020820786371847,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000979416906830153,
      "loss": 1.6019,
      "step": 5349
    },
    {
      "epoch": 0.4602942441710402,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000979409227430576,
      "loss": 1.9601,
      "step": 5350
    },
    {
      "epoch": 0.4603802804783619,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009794015466288203,
      "loss": 1.7336,
      "step": 5351
    },
    {
      "epoch": 0.46046631678568356,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000979393864424908,
      "loss": 1.1842,
      "step": 5352
    },
    {
      "epoch": 0.46055235309300524,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009793861808188621,
      "loss": 1.6131,
      "step": 5353
    },
    {
      "epoch": 0.4606383894003269,
      "grad_norm": 1.796875,
      "learning_rate": 0.000979378495810705,
      "loss": 1.9427,
      "step": 5354
    },
    {
      "epoch": 0.46072442570764865,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009793708094004588,
      "loss": 1.2522,
      "step": 5355
    },
    {
      "epoch": 0.46081046201497033,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009793631215881463,
      "loss": 1.5736,
      "step": 5356
    },
    {
      "epoch": 0.460896498322292,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009793554323737897,
      "loss": 1.6534,
      "step": 5357
    },
    {
      "epoch": 0.4609825346296137,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000979347741757412,
      "loss": 1.6729,
      "step": 5358
    },
    {
      "epoch": 0.4610685709369354,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009793400497390351,
      "loss": 1.1913,
      "step": 5359
    },
    {
      "epoch": 0.46115460724425705,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000979332356318682,
      "loss": 0.9499,
      "step": 5360
    },
    {
      "epoch": 0.4612406435515788,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000979324661496375,
      "loss": 1.3176,
      "step": 5361
    },
    {
      "epoch": 0.46132667985890047,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009793169652721364,
      "loss": 0.8064,
      "step": 5362
    },
    {
      "epoch": 0.46141271616622215,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000979309267645989,
      "loss": 1.1607,
      "step": 5363
    },
    {
      "epoch": 0.4614987524735438,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009793015686179552,
      "loss": 1.6891,
      "step": 5364
    },
    {
      "epoch": 0.4615847887808655,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009792938681880575,
      "loss": 2.2524,
      "step": 5365
    },
    {
      "epoch": 0.46167082508818724,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009792861663563185,
      "loss": 1.1605,
      "step": 5366
    },
    {
      "epoch": 0.4617568613955089,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009792784631227605,
      "loss": 1.1594,
      "step": 5367
    },
    {
      "epoch": 0.4618428977028306,
      "grad_norm": 1.125,
      "learning_rate": 0.0009792707584874065,
      "loss": 1.6708,
      "step": 5368
    },
    {
      "epoch": 0.4619289340101523,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009792630524502787,
      "loss": 1.1137,
      "step": 5369
    },
    {
      "epoch": 0.46201497031747396,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009792553450113995,
      "loss": 1.5451,
      "step": 5370
    },
    {
      "epoch": 0.46210100662479564,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009792476361707918,
      "loss": 0.9892,
      "step": 5371
    },
    {
      "epoch": 0.4621870429321174,
      "grad_norm": 1.0,
      "learning_rate": 0.0009792399259284778,
      "loss": 1.1973,
      "step": 5372
    },
    {
      "epoch": 0.46227307923943906,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009792322142844804,
      "loss": 0.711,
      "step": 5373
    },
    {
      "epoch": 0.46235911554676073,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009792245012388218,
      "loss": 1.6775,
      "step": 5374
    },
    {
      "epoch": 0.4624451518540824,
      "grad_norm": 2.71875,
      "learning_rate": 0.000979216786791525,
      "loss": 2.0641,
      "step": 5375
    },
    {
      "epoch": 0.4625311881614041,
      "grad_norm": 2.078125,
      "learning_rate": 0.000979209070942612,
      "loss": 2.2577,
      "step": 5376
    },
    {
      "epoch": 0.46261722446872583,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009792013536921058,
      "loss": 1.3635,
      "step": 5377
    },
    {
      "epoch": 0.4627032607760475,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009791936350400288,
      "loss": 1.2323,
      "step": 5378
    },
    {
      "epoch": 0.4627892970833692,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009791859149864037,
      "loss": 1.2722,
      "step": 5379
    },
    {
      "epoch": 0.46287533339069087,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000979178193531253,
      "loss": 2.4082,
      "step": 5380
    },
    {
      "epoch": 0.46296136969801255,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009791704706745992,
      "loss": 2.0964,
      "step": 5381
    },
    {
      "epoch": 0.46304740600533423,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009791627464164649,
      "loss": 1.3417,
      "step": 5382
    },
    {
      "epoch": 0.46313344231265596,
      "grad_norm": 1.328125,
      "learning_rate": 0.000979155020756873,
      "loss": 1.9163,
      "step": 5383
    },
    {
      "epoch": 0.46321947861997764,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009791472936958457,
      "loss": 1.1649,
      "step": 5384
    },
    {
      "epoch": 0.4633055149272993,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009791395652334058,
      "loss": 1.0971,
      "step": 5385
    },
    {
      "epoch": 0.463391551234621,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009791318353695757,
      "loss": 1.3977,
      "step": 5386
    },
    {
      "epoch": 0.4634775875419427,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009791241041043784,
      "loss": 1.6807,
      "step": 5387
    },
    {
      "epoch": 0.4635636238492644,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009791163714378361,
      "loss": 1.4039,
      "step": 5388
    },
    {
      "epoch": 0.4636496601565861,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009791086373699717,
      "loss": 1.5292,
      "step": 5389
    },
    {
      "epoch": 0.4637356964639078,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009791009019008076,
      "loss": 1.4106,
      "step": 5390
    },
    {
      "epoch": 0.46382173277122946,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009790931650303666,
      "loss": 0.9456,
      "step": 5391
    },
    {
      "epoch": 0.46390776907855114,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009790854267586716,
      "loss": 1.6952,
      "step": 5392
    },
    {
      "epoch": 0.4639938053858728,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009790776870857445,
      "loss": 1.8555,
      "step": 5393
    },
    {
      "epoch": 0.46407984169319455,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009790699460116084,
      "loss": 1.3335,
      "step": 5394
    },
    {
      "epoch": 0.46416587800051623,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009790622035362859,
      "loss": 1.6701,
      "step": 5395
    },
    {
      "epoch": 0.4642519143078379,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009790544596597996,
      "loss": 1.9443,
      "step": 5396
    },
    {
      "epoch": 0.4643379506151596,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009790467143821725,
      "loss": 0.7884,
      "step": 5397
    },
    {
      "epoch": 0.46442398692248127,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009790389677034265,
      "loss": 1.3959,
      "step": 5398
    },
    {
      "epoch": 0.464510023229803,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000979031219623585,
      "loss": 1.3289,
      "step": 5399
    },
    {
      "epoch": 0.4645960595371247,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009790234701426701,
      "loss": 1.1851,
      "step": 5400
    },
    {
      "epoch": 0.46468209584444636,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009790157192607048,
      "loss": 0.9285,
      "step": 5401
    },
    {
      "epoch": 0.46476813215176804,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009790079669777117,
      "loss": 1.3042,
      "step": 5402
    },
    {
      "epoch": 0.4648541684590897,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009790002132937135,
      "loss": 1.7459,
      "step": 5403
    },
    {
      "epoch": 0.4649402047664114,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009789924582087328,
      "loss": 1.5766,
      "step": 5404
    },
    {
      "epoch": 0.46502624107373314,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009789847017227923,
      "loss": 1.239,
      "step": 5405
    },
    {
      "epoch": 0.4651122773810548,
      "grad_norm": 1.25,
      "learning_rate": 0.0009789769438359147,
      "loss": 2.2007,
      "step": 5406
    },
    {
      "epoch": 0.4651983136883765,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009789691845481228,
      "loss": 1.2341,
      "step": 5407
    },
    {
      "epoch": 0.4652843499956982,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000978961423859439,
      "loss": 1.7059,
      "step": 5408
    },
    {
      "epoch": 0.46537038630301986,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009789536617698864,
      "loss": 1.2456,
      "step": 5409
    },
    {
      "epoch": 0.4654564226103416,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009789458982794874,
      "loss": 1.5123,
      "step": 5410
    },
    {
      "epoch": 0.46554245891766327,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009789381333882648,
      "loss": 1.0245,
      "step": 5411
    },
    {
      "epoch": 0.46562849522498495,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009789303670962412,
      "loss": 1.5881,
      "step": 5412
    },
    {
      "epoch": 0.46571453153230663,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009789225994034393,
      "loss": 1.0492,
      "step": 5413
    },
    {
      "epoch": 0.4658005678396283,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009789148303098823,
      "loss": 1.2297,
      "step": 5414
    },
    {
      "epoch": 0.46588660414695,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009789070598155924,
      "loss": 0.8854,
      "step": 5415
    },
    {
      "epoch": 0.4659726404542717,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009788992879205926,
      "loss": 1.278,
      "step": 5416
    },
    {
      "epoch": 0.4660586767615934,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009788915146249054,
      "loss": 1.9171,
      "step": 5417
    },
    {
      "epoch": 0.4661447130689151,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009788837399285538,
      "loss": 1.3233,
      "step": 5418
    },
    {
      "epoch": 0.46623074937623676,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009788759638315603,
      "loss": 2.0899,
      "step": 5419
    },
    {
      "epoch": 0.46631678568355844,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009788681863339478,
      "loss": 1.6394,
      "step": 5420
    },
    {
      "epoch": 0.4664028219908802,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009788604074357388,
      "loss": 1.2912,
      "step": 5421
    },
    {
      "epoch": 0.46648885829820186,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009788526271369564,
      "loss": 1.0009,
      "step": 5422
    },
    {
      "epoch": 0.46657489460552354,
      "grad_norm": 1.265625,
      "learning_rate": 0.000978844845437623,
      "loss": 0.9258,
      "step": 5423
    },
    {
      "epoch": 0.4666609309128452,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009788370623377619,
      "loss": 1.7237,
      "step": 5424
    },
    {
      "epoch": 0.4667469672201669,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009788292778373953,
      "loss": 1.5255,
      "step": 5425
    },
    {
      "epoch": 0.4668330035274886,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009788214919365461,
      "loss": 2.0858,
      "step": 5426
    },
    {
      "epoch": 0.4669190398348103,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009788137046352373,
      "loss": 1.653,
      "step": 5427
    },
    {
      "epoch": 0.467005076142132,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009788059159334917,
      "loss": 1.3426,
      "step": 5428
    },
    {
      "epoch": 0.46709111244945367,
      "grad_norm": 1.5,
      "learning_rate": 0.0009787981258313316,
      "loss": 1.4117,
      "step": 5429
    },
    {
      "epoch": 0.46717714875677535,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009787903343287804,
      "loss": 2.5527,
      "step": 5430
    },
    {
      "epoch": 0.46726318506409703,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009787825414258604,
      "loss": 0.9634,
      "step": 5431
    },
    {
      "epoch": 0.46734922137141877,
      "grad_norm": 1.375,
      "learning_rate": 0.0009787747471225944,
      "loss": 1.1673,
      "step": 5432
    },
    {
      "epoch": 0.46743525767874045,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000978766951419006,
      "loss": 1.1692,
      "step": 5433
    },
    {
      "epoch": 0.4675212939860621,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009787591543151167,
      "loss": 2.2077,
      "step": 5434
    },
    {
      "epoch": 0.4676073302933838,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009787513558109505,
      "loss": 1.307,
      "step": 5435
    },
    {
      "epoch": 0.4676933666007055,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009787435559065294,
      "loss": 0.8777,
      "step": 5436
    },
    {
      "epoch": 0.46777940290802716,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009787357546018765,
      "loss": 1.2379,
      "step": 5437
    },
    {
      "epoch": 0.4678654392153489,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009787279518970148,
      "loss": 1.5286,
      "step": 5438
    },
    {
      "epoch": 0.4679514755226706,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009787201477919668,
      "loss": 1.4874,
      "step": 5439
    },
    {
      "epoch": 0.46803751182999226,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009787123422867554,
      "loss": 1.6897,
      "step": 5440
    },
    {
      "epoch": 0.46812354813731394,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009787045353814037,
      "loss": 0.8194,
      "step": 5441
    },
    {
      "epoch": 0.4682095844446356,
      "grad_norm": 1.25,
      "learning_rate": 0.0009786967270759344,
      "loss": 1.1387,
      "step": 5442
    },
    {
      "epoch": 0.46829562075195735,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009786889173703701,
      "loss": 1.9623,
      "step": 5443
    },
    {
      "epoch": 0.46838165705927903,
      "grad_norm": 1.296875,
      "learning_rate": 0.000978681106264734,
      "loss": 1.2723,
      "step": 5444
    },
    {
      "epoch": 0.4684676933666007,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009786732937590486,
      "loss": 2.5003,
      "step": 5445
    },
    {
      "epoch": 0.4685537296739224,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009786654798533368,
      "loss": 1.3699,
      "step": 5446
    },
    {
      "epoch": 0.46863976598124407,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009786576645476218,
      "loss": 2.002,
      "step": 5447
    },
    {
      "epoch": 0.46872580228856575,
      "grad_norm": 1.375,
      "learning_rate": 0.000978649847841926,
      "loss": 2.2944,
      "step": 5448
    },
    {
      "epoch": 0.4688118385958875,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009786420297362725,
      "loss": 1.1268,
      "step": 5449
    },
    {
      "epoch": 0.46889787490320917,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009786342102306844,
      "loss": 1.3106,
      "step": 5450
    },
    {
      "epoch": 0.46898391121053085,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009786263893251843,
      "loss": 1.2094,
      "step": 5451
    },
    {
      "epoch": 0.4690699475178525,
      "grad_norm": 1.140625,
      "learning_rate": 0.000978618567019795,
      "loss": 0.9157,
      "step": 5452
    },
    {
      "epoch": 0.4691559838251742,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009786107433145395,
      "loss": 1.2964,
      "step": 5453
    },
    {
      "epoch": 0.46924202013249594,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009786029182094404,
      "loss": 1.8701,
      "step": 5454
    },
    {
      "epoch": 0.4693280564398176,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009785950917045212,
      "loss": 1.2315,
      "step": 5455
    },
    {
      "epoch": 0.4694140927471393,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009785872637998044,
      "loss": 1.2115,
      "step": 5456
    },
    {
      "epoch": 0.469500129054461,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009785794344953128,
      "loss": 1.1856,
      "step": 5457
    },
    {
      "epoch": 0.46958616536178266,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009785716037910696,
      "loss": 1.4292,
      "step": 5458
    },
    {
      "epoch": 0.46967220166910434,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009785637716870973,
      "loss": 0.8544,
      "step": 5459
    },
    {
      "epoch": 0.4697582379764261,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000978555938183419,
      "loss": 0.676,
      "step": 5460
    },
    {
      "epoch": 0.46984427428374775,
      "grad_norm": 1.40625,
      "learning_rate": 0.000978548103280058,
      "loss": 0.8816,
      "step": 5461
    },
    {
      "epoch": 0.46993031059106943,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009785402669770366,
      "loss": 1.3592,
      "step": 5462
    },
    {
      "epoch": 0.4700163468983911,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000978532429274378,
      "loss": 1.7549,
      "step": 5463
    },
    {
      "epoch": 0.4701023832057128,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009785245901721053,
      "loss": 1.7491,
      "step": 5464
    },
    {
      "epoch": 0.47018841951303453,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009785167496702413,
      "loss": 1.341,
      "step": 5465
    },
    {
      "epoch": 0.4702744558203562,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009785089077688086,
      "loss": 1.6184,
      "step": 5466
    },
    {
      "epoch": 0.4703604921276779,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009785010644678305,
      "loss": 1.6236,
      "step": 5467
    },
    {
      "epoch": 0.47044652843499957,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009784932197673298,
      "loss": 0.9099,
      "step": 5468
    },
    {
      "epoch": 0.47053256474232125,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009784853736673296,
      "loss": 1.3413,
      "step": 5469
    },
    {
      "epoch": 0.4706186010496429,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009784775261678527,
      "loss": 1.156,
      "step": 5470
    },
    {
      "epoch": 0.47070463735696466,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009784696772689222,
      "loss": 1.4039,
      "step": 5471
    },
    {
      "epoch": 0.47079067366428634,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009784618269705607,
      "loss": 1.4176,
      "step": 5472
    },
    {
      "epoch": 0.470876709971608,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009784539752727916,
      "loss": 1.5472,
      "step": 5473
    },
    {
      "epoch": 0.4709627462789297,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009784461221756377,
      "loss": 0.9808,
      "step": 5474
    },
    {
      "epoch": 0.4710487825862514,
      "grad_norm": 1.375,
      "learning_rate": 0.000978438267679122,
      "loss": 1.2172,
      "step": 5475
    },
    {
      "epoch": 0.4711348188935731,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009784304117832672,
      "loss": 1.4692,
      "step": 5476
    },
    {
      "epoch": 0.4712208552008948,
      "grad_norm": 1.25,
      "learning_rate": 0.0009784225544880966,
      "loss": 1.8692,
      "step": 5477
    },
    {
      "epoch": 0.4713068915082165,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009784146957936331,
      "loss": 0.7158,
      "step": 5478
    },
    {
      "epoch": 0.47139292781553815,
      "grad_norm": 1.375,
      "learning_rate": 0.0009784068356998995,
      "loss": 1.8292,
      "step": 5479
    },
    {
      "epoch": 0.47147896412285983,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009783989742069192,
      "loss": 1.159,
      "step": 5480
    },
    {
      "epoch": 0.4715650004301815,
      "grad_norm": 1.375,
      "learning_rate": 0.0009783911113147147,
      "loss": 2.2089,
      "step": 5481
    },
    {
      "epoch": 0.47165103673750325,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009783832470233093,
      "loss": 0.7317,
      "step": 5482
    },
    {
      "epoch": 0.47173707304482493,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000978375381332726,
      "loss": 2.0085,
      "step": 5483
    },
    {
      "epoch": 0.4718231093521466,
      "grad_norm": 1.0625,
      "learning_rate": 0.000978367514242988,
      "loss": 0.8829,
      "step": 5484
    },
    {
      "epoch": 0.4719091456594683,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009783596457541178,
      "loss": 1.7738,
      "step": 5485
    },
    {
      "epoch": 0.47199518196678997,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009783517758661386,
      "loss": 1.4013,
      "step": 5486
    },
    {
      "epoch": 0.4720812182741117,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009783439045790735,
      "loss": 1.2342,
      "step": 5487
    },
    {
      "epoch": 0.4721672545814334,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009783360318929458,
      "loss": 1.5647,
      "step": 5488
    },
    {
      "epoch": 0.47225329088875506,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009783281578077778,
      "loss": 1.8049,
      "step": 5489
    },
    {
      "epoch": 0.47233932719607674,
      "grad_norm": 2.40625,
      "learning_rate": 0.0009783202823235933,
      "loss": 1.2558,
      "step": 5490
    },
    {
      "epoch": 0.4724253635033984,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009783124054404148,
      "loss": 1.1137,
      "step": 5491
    },
    {
      "epoch": 0.4725113998107201,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009783045271582658,
      "loss": 0.9823,
      "step": 5492
    },
    {
      "epoch": 0.47259743611804184,
      "grad_norm": 1.5859375,
      "learning_rate": 0.000978296647477169,
      "loss": 1.2526,
      "step": 5493
    },
    {
      "epoch": 0.4726834724253635,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009782887663971475,
      "loss": 1.4723,
      "step": 5494
    },
    {
      "epoch": 0.4727695087326852,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009782808839182242,
      "loss": 1.1612,
      "step": 5495
    },
    {
      "epoch": 0.4728555450400069,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009782730000404225,
      "loss": 1.5333,
      "step": 5496
    },
    {
      "epoch": 0.47294158134732855,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009782651147637652,
      "loss": 1.3365,
      "step": 5497
    },
    {
      "epoch": 0.4730276176546503,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009782572280882757,
      "loss": 2.4084,
      "step": 5498
    },
    {
      "epoch": 0.47311365396197197,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009782493400139764,
      "loss": 1.032,
      "step": 5499
    },
    {
      "epoch": 0.47319969026929365,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009782414505408912,
      "loss": 1.5786,
      "step": 5500
    },
    {
      "epoch": 0.47328572657661533,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009782335596690425,
      "loss": 2.05,
      "step": 5501
    },
    {
      "epoch": 0.473371762883937,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009782256673984537,
      "loss": 1.1892,
      "step": 5502
    },
    {
      "epoch": 0.4734577991912587,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000978217773729148,
      "loss": 1.1397,
      "step": 5503
    },
    {
      "epoch": 0.4735438354985804,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009782098786611482,
      "loss": 0.8447,
      "step": 5504
    },
    {
      "epoch": 0.4736298718059021,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009782019821944774,
      "loss": 1.6457,
      "step": 5505
    },
    {
      "epoch": 0.4737159081132238,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000978194084329159,
      "loss": 1.5122,
      "step": 5506
    },
    {
      "epoch": 0.47380194442054546,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009781861850652158,
      "loss": 1.0081,
      "step": 5507
    },
    {
      "epoch": 0.47388798072786714,
      "grad_norm": 1.8125,
      "learning_rate": 0.000978178284402671,
      "loss": 1.237,
      "step": 5508
    },
    {
      "epoch": 0.4739740170351889,
      "grad_norm": 1.359375,
      "learning_rate": 0.000978170382341548,
      "loss": 1.0365,
      "step": 5509
    },
    {
      "epoch": 0.47406005334251056,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009781624788818694,
      "loss": 1.1302,
      "step": 5510
    },
    {
      "epoch": 0.47414608964983224,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009781545740236584,
      "loss": 1.7816,
      "step": 5511
    },
    {
      "epoch": 0.4742321259571539,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009781466677669384,
      "loss": 1.4421,
      "step": 5512
    },
    {
      "epoch": 0.4743181622644756,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009781387601117324,
      "loss": 2.0432,
      "step": 5513
    },
    {
      "epoch": 0.4744041985717973,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009781308510580634,
      "loss": 1.8544,
      "step": 5514
    },
    {
      "epoch": 0.474490234879119,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009781229406059548,
      "loss": 1.1753,
      "step": 5515
    },
    {
      "epoch": 0.4745762711864407,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009781150287554298,
      "loss": 0.8213,
      "step": 5516
    },
    {
      "epoch": 0.47466230749376237,
      "grad_norm": 1.0625,
      "learning_rate": 0.000978107115506511,
      "loss": 1.345,
      "step": 5517
    },
    {
      "epoch": 0.47474834380108405,
      "grad_norm": 1.3125,
      "learning_rate": 0.000978099200859222,
      "loss": 0.9914,
      "step": 5518
    },
    {
      "epoch": 0.47483438010840573,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009780912848135856,
      "loss": 1.797,
      "step": 5519
    },
    {
      "epoch": 0.47492041641572746,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009780833673696254,
      "loss": 1.2479,
      "step": 5520
    },
    {
      "epoch": 0.47500645272304914,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009780754485273645,
      "loss": 0.8772,
      "step": 5521
    },
    {
      "epoch": 0.4750924890303708,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009780675282868257,
      "loss": 1.1109,
      "step": 5522
    },
    {
      "epoch": 0.4751785253376925,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009780596066480322,
      "loss": 1.4266,
      "step": 5523
    },
    {
      "epoch": 0.4752645616450142,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009780516836110074,
      "loss": 1.2793,
      "step": 5524
    },
    {
      "epoch": 0.47535059795233586,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009780437591757746,
      "loss": 2.0354,
      "step": 5525
    },
    {
      "epoch": 0.4754366342596576,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009780358333423566,
      "loss": 0.9212,
      "step": 5526
    },
    {
      "epoch": 0.4755226705669793,
      "grad_norm": 1.453125,
      "learning_rate": 0.000978027906110777,
      "loss": 1.7776,
      "step": 5527
    },
    {
      "epoch": 0.47560870687430096,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009780199774810584,
      "loss": 1.2236,
      "step": 5528
    },
    {
      "epoch": 0.47569474318162264,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009780120474532244,
      "loss": 0.95,
      "step": 5529
    },
    {
      "epoch": 0.4757807794889443,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009780041160272984,
      "loss": 2.0005,
      "step": 5530
    },
    {
      "epoch": 0.47586681579626605,
      "grad_norm": 1.171875,
      "learning_rate": 0.000977996183203303,
      "loss": 1.2056,
      "step": 5531
    },
    {
      "epoch": 0.47595285210358773,
      "grad_norm": 1.375,
      "learning_rate": 0.000977988248981262,
      "loss": 1.4282,
      "step": 5532
    },
    {
      "epoch": 0.4760388884109094,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009779803133611981,
      "loss": 1.3463,
      "step": 5533
    },
    {
      "epoch": 0.4761249247182311,
      "grad_norm": 1.09375,
      "learning_rate": 0.000977972376343135,
      "loss": 1.4975,
      "step": 5534
    },
    {
      "epoch": 0.47621096102555277,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009779644379270956,
      "loss": 1.9489,
      "step": 5535
    },
    {
      "epoch": 0.47629699733287445,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009779564981131032,
      "loss": 1.8176,
      "step": 5536
    },
    {
      "epoch": 0.4763830336401962,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009779485569011807,
      "loss": 1.4711,
      "step": 5537
    },
    {
      "epoch": 0.47646906994751786,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009779406142913518,
      "loss": 0.9244,
      "step": 5538
    },
    {
      "epoch": 0.47655510625483954,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009779326702836395,
      "loss": 0.8912,
      "step": 5539
    },
    {
      "epoch": 0.4766411425621612,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009779247248780673,
      "loss": 1.7723,
      "step": 5540
    },
    {
      "epoch": 0.4767271788694829,
      "grad_norm": 1.234375,
      "learning_rate": 0.000977916778074658,
      "loss": 1.0749,
      "step": 5541
    },
    {
      "epoch": 0.47681321517680464,
      "grad_norm": 1.5,
      "learning_rate": 0.0009779088298734354,
      "loss": 1.1595,
      "step": 5542
    },
    {
      "epoch": 0.4768992514841263,
      "grad_norm": 1.9765625,
      "learning_rate": 0.000977900880274422,
      "loss": 1.8763,
      "step": 5543
    },
    {
      "epoch": 0.476985287791448,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009778929292776418,
      "loss": 1.5309,
      "step": 5544
    },
    {
      "epoch": 0.4770713240987697,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009778849768831174,
      "loss": 2.0719,
      "step": 5545
    },
    {
      "epoch": 0.47715736040609136,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009778770230908726,
      "loss": 2.1064,
      "step": 5546
    },
    {
      "epoch": 0.47724339671341304,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009778690679009305,
      "loss": 1.5951,
      "step": 5547
    },
    {
      "epoch": 0.4773294330207348,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009778611113133143,
      "loss": 1.5212,
      "step": 5548
    },
    {
      "epoch": 0.47741546932805645,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009778531533280472,
      "loss": 1.4933,
      "step": 5549
    },
    {
      "epoch": 0.47750150563537813,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009778451939451527,
      "loss": 1.578,
      "step": 5550
    },
    {
      "epoch": 0.4775875419426998,
      "grad_norm": 1.25,
      "learning_rate": 0.0009778372331646539,
      "loss": 1.4347,
      "step": 5551
    },
    {
      "epoch": 0.4776735782500215,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009778292709865738,
      "loss": 2.2474,
      "step": 5552
    },
    {
      "epoch": 0.4777596145573432,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009778213074109366,
      "loss": 1.4161,
      "step": 5553
    },
    {
      "epoch": 0.4778456508646649,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009778133424377647,
      "loss": 1.6187,
      "step": 5554
    },
    {
      "epoch": 0.4779316871719866,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009778053760670815,
      "loss": 1.1947,
      "step": 5555
    },
    {
      "epoch": 0.47801772347930827,
      "grad_norm": 1.75,
      "learning_rate": 0.0009777974082989108,
      "loss": 2.3331,
      "step": 5556
    },
    {
      "epoch": 0.47810375978662994,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009777894391332755,
      "loss": 2.1043,
      "step": 5557
    },
    {
      "epoch": 0.4781897960939516,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009777814685701992,
      "loss": 1.2219,
      "step": 5558
    },
    {
      "epoch": 0.47827583240127336,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009777734966097047,
      "loss": 1.3418,
      "step": 5559
    },
    {
      "epoch": 0.47836186870859504,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009777655232518158,
      "loss": 1.7171,
      "step": 5560
    },
    {
      "epoch": 0.4784479050159167,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009777575484965558,
      "loss": 1.58,
      "step": 5561
    },
    {
      "epoch": 0.4785339413232384,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009777495723439476,
      "loss": 1.4133,
      "step": 5562
    },
    {
      "epoch": 0.4786199776305601,
      "grad_norm": 1.40625,
      "learning_rate": 0.000977741594794015,
      "loss": 1.4152,
      "step": 5563
    },
    {
      "epoch": 0.4787060139378818,
      "grad_norm": 2.0625,
      "learning_rate": 0.000977733615846781,
      "loss": 1.9974,
      "step": 5564
    },
    {
      "epoch": 0.4787920502452035,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009777256355022693,
      "loss": 1.4084,
      "step": 5565
    },
    {
      "epoch": 0.4788780865525252,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000977717653760503,
      "loss": 1.3052,
      "step": 5566
    },
    {
      "epoch": 0.47896412285984685,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009777096706215054,
      "loss": 1.0459,
      "step": 5567
    },
    {
      "epoch": 0.47905015916716853,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009777016860852998,
      "loss": 1.4581,
      "step": 5568
    },
    {
      "epoch": 0.4791361954744902,
      "grad_norm": 1.359375,
      "learning_rate": 0.00097769370015191,
      "loss": 1.3481,
      "step": 5569
    },
    {
      "epoch": 0.47922223178181195,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009776857128213589,
      "loss": 1.7358,
      "step": 5570
    },
    {
      "epoch": 0.4793082680891336,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00097767772409367,
      "loss": 1.8084,
      "step": 5571
    },
    {
      "epoch": 0.4793943043964553,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009776697339688664,
      "loss": 1.9491,
      "step": 5572
    },
    {
      "epoch": 0.479480340703777,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009776617424469719,
      "loss": 1.668,
      "step": 5573
    },
    {
      "epoch": 0.47956637701109867,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009776537495280097,
      "loss": 1.2271,
      "step": 5574
    },
    {
      "epoch": 0.4796524133184204,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009776457552120033,
      "loss": 1.7129,
      "step": 5575
    },
    {
      "epoch": 0.4797384496257421,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009776377594989758,
      "loss": 1.1187,
      "step": 5576
    },
    {
      "epoch": 0.47982448593306376,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000977629762388951,
      "loss": 1.5023,
      "step": 5577
    },
    {
      "epoch": 0.47991052224038544,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009776217638819517,
      "loss": 1.3848,
      "step": 5578
    },
    {
      "epoch": 0.4799965585477071,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009776137639780017,
      "loss": 0.49,
      "step": 5579
    },
    {
      "epoch": 0.4800825948550288,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009776057626771247,
      "loss": 1.7039,
      "step": 5580
    },
    {
      "epoch": 0.48016863116235053,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009775977599793433,
      "loss": 1.825,
      "step": 5581
    },
    {
      "epoch": 0.4802546674696722,
      "grad_norm": 1.375,
      "learning_rate": 0.0009775897558846815,
      "loss": 1.8812,
      "step": 5582
    },
    {
      "epoch": 0.4803407037769939,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009775817503931625,
      "loss": 0.8444,
      "step": 5583
    },
    {
      "epoch": 0.4804267400843156,
      "grad_norm": 1.109375,
      "learning_rate": 0.00097757374350481,
      "loss": 1.8797,
      "step": 5584
    },
    {
      "epoch": 0.48051277639163725,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000977565735219647,
      "loss": 1.1858,
      "step": 5585
    },
    {
      "epoch": 0.480598812698959,
      "grad_norm": 1.390625,
      "learning_rate": 0.000977557725537697,
      "loss": 1.1226,
      "step": 5586
    },
    {
      "epoch": 0.48068484900628067,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009775497144589836,
      "loss": 0.9301,
      "step": 5587
    },
    {
      "epoch": 0.48077088531360235,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009775417019835303,
      "loss": 1.6215,
      "step": 5588
    },
    {
      "epoch": 0.480856921620924,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009775336881113603,
      "loss": 1.0307,
      "step": 5589
    },
    {
      "epoch": 0.4809429579282457,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009775256728424971,
      "loss": 1.0904,
      "step": 5590
    },
    {
      "epoch": 0.4810289942355674,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009775176561769643,
      "loss": 1.7741,
      "step": 5591
    },
    {
      "epoch": 0.4811150305428891,
      "grad_norm": 1.703125,
      "learning_rate": 0.000977509638114785,
      "loss": 1.5557,
      "step": 5592
    },
    {
      "epoch": 0.4812010668502108,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009775016186559832,
      "loss": 1.1539,
      "step": 5593
    },
    {
      "epoch": 0.4812871031575325,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009774935978005817,
      "loss": 0.7208,
      "step": 5594
    },
    {
      "epoch": 0.48137313946485416,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009774855755486046,
      "loss": 1.0975,
      "step": 5595
    },
    {
      "epoch": 0.48145917577217584,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009774775519000747,
      "loss": 1.7464,
      "step": 5596
    },
    {
      "epoch": 0.4815452120794976,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009774695268550161,
      "loss": 1.4302,
      "step": 5597
    },
    {
      "epoch": 0.48163124838681926,
      "grad_norm": 1.453125,
      "learning_rate": 0.000977461500413452,
      "loss": 1.8137,
      "step": 5598
    },
    {
      "epoch": 0.48171728469414093,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009774534725754058,
      "loss": 0.8588,
      "step": 5599
    },
    {
      "epoch": 0.4818033210014626,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000977445443340901,
      "loss": 1.6354,
      "step": 5600
    },
    {
      "epoch": 0.4818893573087843,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009774374127099613,
      "loss": 1.7667,
      "step": 5601
    },
    {
      "epoch": 0.481975393616106,
      "grad_norm": 1.171875,
      "learning_rate": 0.00097742938068261,
      "loss": 1.908,
      "step": 5602
    },
    {
      "epoch": 0.4820614299234277,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009774213472588705,
      "loss": 0.9568,
      "step": 5603
    },
    {
      "epoch": 0.4821474662307494,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009774133124387664,
      "loss": 2.2261,
      "step": 5604
    },
    {
      "epoch": 0.48223350253807107,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009774052762223213,
      "loss": 1.1641,
      "step": 5605
    },
    {
      "epoch": 0.48231953884539275,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009773972386095584,
      "loss": 1.6419,
      "step": 5606
    },
    {
      "epoch": 0.4824055751527144,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009773891996005019,
      "loss": 1.3694,
      "step": 5607
    },
    {
      "epoch": 0.48249161146003616,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009773811591951744,
      "loss": 1.5369,
      "step": 5608
    },
    {
      "epoch": 0.48257764776735784,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009773731173936,
      "loss": 2.0229,
      "step": 5609
    },
    {
      "epoch": 0.4826636840746795,
      "grad_norm": 1.5625,
      "learning_rate": 0.000977365074195802,
      "loss": 1.8046,
      "step": 5610
    },
    {
      "epoch": 0.4827497203820012,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000977357029601804,
      "loss": 1.3797,
      "step": 5611
    },
    {
      "epoch": 0.4828357566893229,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009773489836116297,
      "loss": 1.5954,
      "step": 5612
    },
    {
      "epoch": 0.48292179299664456,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009773409362253023,
      "loss": 1.8116,
      "step": 5613
    },
    {
      "epoch": 0.4830078293039663,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009773328874428456,
      "loss": 1.8462,
      "step": 5614
    },
    {
      "epoch": 0.483093865611288,
      "grad_norm": 1.203125,
      "learning_rate": 0.000977324837264283,
      "loss": 0.6432,
      "step": 5615
    },
    {
      "epoch": 0.48317990191860966,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009773167856896381,
      "loss": 1.8049,
      "step": 5616
    },
    {
      "epoch": 0.48326593822593134,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009773087327189344,
      "loss": 1.6769,
      "step": 5617
    },
    {
      "epoch": 0.483351974533253,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009773006783521956,
      "loss": 1.0875,
      "step": 5618
    },
    {
      "epoch": 0.48343801084057475,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009772926225894449,
      "loss": 1.0551,
      "step": 5619
    },
    {
      "epoch": 0.48352404714789643,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009772845654307063,
      "loss": 1.619,
      "step": 5620
    },
    {
      "epoch": 0.4836100834552181,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009772765068760032,
      "loss": 1.4617,
      "step": 5621
    },
    {
      "epoch": 0.4836961197625398,
      "grad_norm": 1.46875,
      "learning_rate": 0.000977268446925359,
      "loss": 1.4374,
      "step": 5622
    },
    {
      "epoch": 0.48378215606986147,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009772603855787976,
      "loss": 1.4,
      "step": 5623
    },
    {
      "epoch": 0.48386819237718315,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009772523228363421,
      "loss": 1.6005,
      "step": 5624
    },
    {
      "epoch": 0.4839542286845049,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009772442586980167,
      "loss": 1.1534,
      "step": 5625
    },
    {
      "epoch": 0.48404026499182656,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009772361931638445,
      "loss": 1.3744,
      "step": 5626
    },
    {
      "epoch": 0.48412630129914824,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009772281262338495,
      "loss": 1.2705,
      "step": 5627
    },
    {
      "epoch": 0.4842123376064699,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009772200579080547,
      "loss": 1.6324,
      "step": 5628
    },
    {
      "epoch": 0.4842983739137916,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000977211988186484,
      "loss": 1.7424,
      "step": 5629
    },
    {
      "epoch": 0.48438441022111334,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009772039170691613,
      "loss": 1.2691,
      "step": 5630
    },
    {
      "epoch": 0.484470446528435,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009771958445561099,
      "loss": 1.6562,
      "step": 5631
    },
    {
      "epoch": 0.4845564828357567,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009771877706473535,
      "loss": 1.227,
      "step": 5632
    },
    {
      "epoch": 0.4846425191430784,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009771796953429155,
      "loss": 1.0052,
      "step": 5633
    },
    {
      "epoch": 0.48472855545040006,
      "grad_norm": 1.28125,
      "learning_rate": 0.00097717161864282,
      "loss": 1.6254,
      "step": 5634
    },
    {
      "epoch": 0.48481459175772174,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00097716354054709,
      "loss": 1.2379,
      "step": 5635
    },
    {
      "epoch": 0.48490062806504347,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009771554610557495,
      "loss": 0.9347,
      "step": 5636
    },
    {
      "epoch": 0.48498666437236515,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009771473801688222,
      "loss": 1.385,
      "step": 5637
    },
    {
      "epoch": 0.48507270067968683,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009771392978863316,
      "loss": 1.0256,
      "step": 5638
    },
    {
      "epoch": 0.4851587369870085,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009771312142083013,
      "loss": 0.9721,
      "step": 5639
    },
    {
      "epoch": 0.4852447732943302,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000977123129134755,
      "loss": 1.1408,
      "step": 5640
    },
    {
      "epoch": 0.4853308096016519,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009771150426657164,
      "loss": 1.1599,
      "step": 5641
    },
    {
      "epoch": 0.4854168459089736,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000977106954801209,
      "loss": 1.3835,
      "step": 5642
    },
    {
      "epoch": 0.4855028822162953,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009770988655412568,
      "loss": 2.1249,
      "step": 5643
    },
    {
      "epoch": 0.48558891852361696,
      "grad_norm": 1.125,
      "learning_rate": 0.0009770907748858828,
      "loss": 1.4646,
      "step": 5644
    },
    {
      "epoch": 0.48567495483093864,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009770826828351114,
      "loss": 1.3766,
      "step": 5645
    },
    {
      "epoch": 0.4857609911382603,
      "grad_norm": 1.0,
      "learning_rate": 0.0009770745893889657,
      "loss": 1.056,
      "step": 5646
    },
    {
      "epoch": 0.48584702744558206,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009770664945474698,
      "loss": 0.6904,
      "step": 5647
    },
    {
      "epoch": 0.48593306375290374,
      "grad_norm": 2.34375,
      "learning_rate": 0.000977058398310647,
      "loss": 1.6006,
      "step": 5648
    },
    {
      "epoch": 0.4860191000602254,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009770503006785215,
      "loss": 0.9631,
      "step": 5649
    },
    {
      "epoch": 0.4861051363675471,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009770422016511164,
      "loss": 2.1194,
      "step": 5650
    },
    {
      "epoch": 0.4861911726748688,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009770341012284556,
      "loss": 1.3886,
      "step": 5651
    },
    {
      "epoch": 0.4862772089821905,
      "grad_norm": 3.171875,
      "learning_rate": 0.000977025999410563,
      "loss": 1.5749,
      "step": 5652
    },
    {
      "epoch": 0.4863632452895122,
      "grad_norm": 1.15625,
      "learning_rate": 0.000977017896197462,
      "loss": 1.6399,
      "step": 5653
    },
    {
      "epoch": 0.48644928159683387,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009770097915891764,
      "loss": 1.9007,
      "step": 5654
    },
    {
      "epoch": 0.48653531790415555,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00097700168558573,
      "loss": 1.4374,
      "step": 5655
    },
    {
      "epoch": 0.48662135421147723,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009769935781871466,
      "loss": 1.9773,
      "step": 5656
    },
    {
      "epoch": 0.4867073905187989,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009769854693934495,
      "loss": 1.5484,
      "step": 5657
    },
    {
      "epoch": 0.48679342682612065,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000976977359204663,
      "loss": 0.9914,
      "step": 5658
    },
    {
      "epoch": 0.4868794631334423,
      "grad_norm": 1.3125,
      "learning_rate": 0.00097696924762081,
      "loss": 1.8315,
      "step": 5659
    },
    {
      "epoch": 0.486965499440764,
      "grad_norm": 1.328125,
      "learning_rate": 0.000976961134641915,
      "loss": 1.6812,
      "step": 5660
    },
    {
      "epoch": 0.4870515357480857,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009769530202680015,
      "loss": 2.0472,
      "step": 5661
    },
    {
      "epoch": 0.48713757205540736,
      "grad_norm": 1.125,
      "learning_rate": 0.000976944904499093,
      "loss": 1.8956,
      "step": 5662
    },
    {
      "epoch": 0.4872236083627291,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009769367873352136,
      "loss": 1.4933,
      "step": 5663
    },
    {
      "epoch": 0.4873096446700508,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009769286687763867,
      "loss": 1.5881,
      "step": 5664
    },
    {
      "epoch": 0.48739568097737246,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009769205488226365,
      "loss": 0.9349,
      "step": 5665
    },
    {
      "epoch": 0.48748171728469414,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000976912427473986,
      "loss": 1.7818,
      "step": 5666
    },
    {
      "epoch": 0.4875677535920158,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009769043047304598,
      "loss": 1.8519,
      "step": 5667
    },
    {
      "epoch": 0.4876537898993375,
      "grad_norm": 1.46875,
      "learning_rate": 0.000976896180592081,
      "loss": 1.0169,
      "step": 5668
    },
    {
      "epoch": 0.48773982620665923,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009768880550588737,
      "loss": 0.8797,
      "step": 5669
    },
    {
      "epoch": 0.4878258625139809,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009768799281308616,
      "loss": 0.9413,
      "step": 5670
    },
    {
      "epoch": 0.4879118988213026,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009768717998080685,
      "loss": 2.4194,
      "step": 5671
    },
    {
      "epoch": 0.48799793512862427,
      "grad_norm": 1.484375,
      "learning_rate": 0.000976863670090518,
      "loss": 1.4467,
      "step": 5672
    },
    {
      "epoch": 0.48808397143594595,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009768555389782342,
      "loss": 1.0679,
      "step": 5673
    },
    {
      "epoch": 0.4881700077432677,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009768474064712407,
      "loss": 1.174,
      "step": 5674
    },
    {
      "epoch": 0.48825604405058937,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000976839272569561,
      "loss": 1.5014,
      "step": 5675
    },
    {
      "epoch": 0.48834208035791105,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009768311372732193,
      "loss": 1.4135,
      "step": 5676
    },
    {
      "epoch": 0.4884281166652327,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009768230005822395,
      "loss": 1.5102,
      "step": 5677
    },
    {
      "epoch": 0.4885141529725544,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000976814862496645,
      "loss": 1.1664,
      "step": 5678
    },
    {
      "epoch": 0.4886001892798761,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009768067230164596,
      "loss": 1.523,
      "step": 5679
    },
    {
      "epoch": 0.4886862255871978,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009767985821417073,
      "loss": 1.9244,
      "step": 5680
    },
    {
      "epoch": 0.4887722618945195,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000976790439872412,
      "loss": 0.863,
      "step": 5681
    },
    {
      "epoch": 0.4888582982018412,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009767822962085974,
      "loss": 1.3826,
      "step": 5682
    },
    {
      "epoch": 0.48894433450916286,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009767741511502873,
      "loss": 0.8732,
      "step": 5683
    },
    {
      "epoch": 0.48903037081648454,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009767660046975056,
      "loss": 1.3082,
      "step": 5684
    },
    {
      "epoch": 0.4891164071238063,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009767578568502758,
      "loss": 0.6304,
      "step": 5685
    },
    {
      "epoch": 0.48920244343112795,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009767497076086222,
      "loss": 1.0425,
      "step": 5686
    },
    {
      "epoch": 0.48928847973844963,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009767415569725685,
      "loss": 0.9258,
      "step": 5687
    },
    {
      "epoch": 0.4893745160457713,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009767334049421384,
      "loss": 2.5859,
      "step": 5688
    },
    {
      "epoch": 0.489460552353093,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009767252515173557,
      "loss": 1.9954,
      "step": 5689
    },
    {
      "epoch": 0.48954658866041467,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009767170966982445,
      "loss": 1.6184,
      "step": 5690
    },
    {
      "epoch": 0.4896326249677364,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009767089404848284,
      "loss": 1.4348,
      "step": 5691
    },
    {
      "epoch": 0.4897186612750581,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009767007828771314,
      "loss": 1.9337,
      "step": 5692
    },
    {
      "epoch": 0.48980469758237977,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009766926238751771,
      "loss": 1.3342,
      "step": 5693
    },
    {
      "epoch": 0.48989073388970145,
      "grad_norm": 1.40625,
      "learning_rate": 0.00097668446347899,
      "loss": 1.2955,
      "step": 5694
    },
    {
      "epoch": 0.4899767701970231,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009766763016885933,
      "loss": 1.0726,
      "step": 5695
    },
    {
      "epoch": 0.49006280650434486,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009766681385040112,
      "loss": 1.5838,
      "step": 5696
    },
    {
      "epoch": 0.49014884281166654,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009766599739252674,
      "loss": 0.8883,
      "step": 5697
    },
    {
      "epoch": 0.4902348791189882,
      "grad_norm": 1.5,
      "learning_rate": 0.000976651807952386,
      "loss": 1.6952,
      "step": 5698
    },
    {
      "epoch": 0.4903209154263099,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009766436405853909,
      "loss": 1.1566,
      "step": 5699
    },
    {
      "epoch": 0.4904069517336316,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009766354718243058,
      "loss": 1.6818,
      "step": 5700
    },
    {
      "epoch": 0.49049298804095326,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009766273016691544,
      "loss": 1.7024,
      "step": 5701
    },
    {
      "epoch": 0.490579024348275,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000976619130119961,
      "loss": 1.4242,
      "step": 5702
    },
    {
      "epoch": 0.4906650606555967,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009766109571767493,
      "loss": 1.1494,
      "step": 5703
    },
    {
      "epoch": 0.49075109696291835,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009766027828395432,
      "loss": 1.9346,
      "step": 5704
    },
    {
      "epoch": 0.49083713327024003,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009765946071083667,
      "loss": 2.202,
      "step": 5705
    },
    {
      "epoch": 0.4909231695775617,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009765864299832439,
      "loss": 1.0909,
      "step": 5706
    },
    {
      "epoch": 0.49100920588488345,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009765782514641983,
      "loss": 0.9805,
      "step": 5707
    },
    {
      "epoch": 0.49109524219220513,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009765700715512538,
      "loss": 0.7678,
      "step": 5708
    },
    {
      "epoch": 0.4911812784995268,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009765618902444347,
      "loss": 0.8495,
      "step": 5709
    },
    {
      "epoch": 0.4912673148068485,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009765537075437648,
      "loss": 1.3257,
      "step": 5710
    },
    {
      "epoch": 0.49135335111417017,
      "grad_norm": 1.75,
      "learning_rate": 0.000976545523449268,
      "loss": 0.9963,
      "step": 5711
    },
    {
      "epoch": 0.49143938742149185,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009765373379609681,
      "loss": 0.8412,
      "step": 5712
    },
    {
      "epoch": 0.4915254237288136,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009765291510788893,
      "loss": 1.4268,
      "step": 5713
    },
    {
      "epoch": 0.49161146003613526,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009765209628030553,
      "loss": 1.3856,
      "step": 5714
    },
    {
      "epoch": 0.49169749634345694,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009765127731334903,
      "loss": 1.5003,
      "step": 5715
    },
    {
      "epoch": 0.4917835326507786,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009765045820702181,
      "loss": 1.2208,
      "step": 5716
    },
    {
      "epoch": 0.4918695689581003,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009764963896132626,
      "loss": 1.8299,
      "step": 5717
    },
    {
      "epoch": 0.49195560526542204,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009764881957626478,
      "loss": 1.2053,
      "step": 5718
    },
    {
      "epoch": 0.4920416415727437,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009764800005183978,
      "loss": 1.3422,
      "step": 5719
    },
    {
      "epoch": 0.4921276778800654,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009764718038805364,
      "loss": 1.8682,
      "step": 5720
    },
    {
      "epoch": 0.4922137141873871,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0009764636058490877,
      "loss": 0.9686,
      "step": 5721
    },
    {
      "epoch": 0.49229975049470875,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009764554064240755,
      "loss": 0.9379,
      "step": 5722
    },
    {
      "epoch": 0.49238578680203043,
      "grad_norm": 1.28125,
      "learning_rate": 0.000976447205605524,
      "loss": 1.9744,
      "step": 5723
    },
    {
      "epoch": 0.49247182310935217,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000976439003393457,
      "loss": 1.1035,
      "step": 5724
    },
    {
      "epoch": 0.49255785941667385,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009764307997878987,
      "loss": 0.9841,
      "step": 5725
    },
    {
      "epoch": 0.49264389572399553,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009764225947888729,
      "loss": 1.1313,
      "step": 5726
    },
    {
      "epoch": 0.4927299320313172,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009764143883964036,
      "loss": 1.5994,
      "step": 5727
    },
    {
      "epoch": 0.4928159683386389,
      "grad_norm": 0.9375,
      "learning_rate": 0.000976406180610515,
      "loss": 1.0314,
      "step": 5728
    },
    {
      "epoch": 0.4929020046459606,
      "grad_norm": 1.09375,
      "learning_rate": 0.000976397971431231,
      "loss": 1.0475,
      "step": 5729
    },
    {
      "epoch": 0.4929880409532823,
      "grad_norm": 1.5,
      "learning_rate": 0.0009763897608585754,
      "loss": 1.938,
      "step": 5730
    },
    {
      "epoch": 0.493074077260604,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009763815488925726,
      "loss": 1.401,
      "step": 5731
    },
    {
      "epoch": 0.49316011356792566,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009763733355332463,
      "loss": 1.8408,
      "step": 5732
    },
    {
      "epoch": 0.49324614987524734,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009763651207806208,
      "loss": 1.3338,
      "step": 5733
    },
    {
      "epoch": 0.493332186182569,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009763569046347198,
      "loss": 0.7512,
      "step": 5734
    },
    {
      "epoch": 0.49341822248989076,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009763486870955675,
      "loss": 1.7575,
      "step": 5735
    },
    {
      "epoch": 0.49350425879721244,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000976340468163188,
      "loss": 1.5513,
      "step": 5736
    },
    {
      "epoch": 0.4935902951045341,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009763322478376054,
      "loss": 1.7748,
      "step": 5737
    },
    {
      "epoch": 0.4936763314118558,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009763240261188433,
      "loss": 1.6013,
      "step": 5738
    },
    {
      "epoch": 0.4937623677191775,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009763158030069263,
      "loss": 1.9969,
      "step": 5739
    },
    {
      "epoch": 0.4938484040264992,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009763075785018783,
      "loss": 0.9907,
      "step": 5740
    },
    {
      "epoch": 0.4939344403338209,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009762993526037229,
      "loss": 1.544,
      "step": 5741
    },
    {
      "epoch": 0.49402047664114257,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009762911253124849,
      "loss": 1.4543,
      "step": 5742
    },
    {
      "epoch": 0.49410651294846425,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009762828966281878,
      "loss": 1.1094,
      "step": 5743
    },
    {
      "epoch": 0.49419254925578593,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009762746665508559,
      "loss": 1.9467,
      "step": 5744
    },
    {
      "epoch": 0.4942785855631076,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009762664350805132,
      "loss": 1.7088,
      "step": 5745
    },
    {
      "epoch": 0.49436462187042934,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009762582022171839,
      "loss": 1.2382,
      "step": 5746
    },
    {
      "epoch": 0.494450658177751,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009762499679608919,
      "loss": 1.4783,
      "step": 5747
    },
    {
      "epoch": 0.4945366944850727,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009762417323116615,
      "loss": 1.5712,
      "step": 5748
    },
    {
      "epoch": 0.4946227307923944,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009762334952695164,
      "loss": 0.9078,
      "step": 5749
    },
    {
      "epoch": 0.49470876709971606,
      "grad_norm": 2.21875,
      "learning_rate": 0.0009762252568344811,
      "loss": 0.9212,
      "step": 5750
    },
    {
      "epoch": 0.4947948034070378,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009762170170065796,
      "loss": 1.7948,
      "step": 5751
    },
    {
      "epoch": 0.4948808397143595,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009762087757858359,
      "loss": 0.929,
      "step": 5752
    },
    {
      "epoch": 0.49496687602168116,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009762005331722741,
      "loss": 1.2997,
      "step": 5753
    },
    {
      "epoch": 0.49505291232900284,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009761922891659185,
      "loss": 1.5746,
      "step": 5754
    },
    {
      "epoch": 0.4951389486363245,
      "grad_norm": 1.390625,
      "learning_rate": 0.000976184043766793,
      "loss": 1.097,
      "step": 5755
    },
    {
      "epoch": 0.4952249849436462,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009761757969749216,
      "loss": 1.2342,
      "step": 5756
    },
    {
      "epoch": 0.49531102125096793,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009761675487903287,
      "loss": 1.5637,
      "step": 5757
    },
    {
      "epoch": 0.4953970575582896,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009761592992130386,
      "loss": 0.941,
      "step": 5758
    },
    {
      "epoch": 0.4954830938656113,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009761510482430749,
      "loss": 1.8654,
      "step": 5759
    },
    {
      "epoch": 0.49556913017293297,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009761427958804621,
      "loss": 1.7701,
      "step": 5760
    },
    {
      "epoch": 0.49565516648025465,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009761345421252241,
      "loss": 2.0265,
      "step": 5761
    },
    {
      "epoch": 0.4957412027875764,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009761262869773852,
      "loss": 1.1409,
      "step": 5762
    },
    {
      "epoch": 0.49582723909489806,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009761180304369695,
      "loss": 1.977,
      "step": 5763
    },
    {
      "epoch": 0.49591327540221974,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009761097725040013,
      "loss": 1.9783,
      "step": 5764
    },
    {
      "epoch": 0.4959993117095414,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009761015131785045,
      "loss": 1.3128,
      "step": 5765
    },
    {
      "epoch": 0.4960853480168631,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009760932524605033,
      "loss": 0.8992,
      "step": 5766
    },
    {
      "epoch": 0.4961713843241848,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009760849903500219,
      "loss": 0.9034,
      "step": 5767
    },
    {
      "epoch": 0.4962574206315065,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009760767268470847,
      "loss": 0.7027,
      "step": 5768
    },
    {
      "epoch": 0.4963434569388282,
      "grad_norm": 1.375,
      "learning_rate": 0.0009760684619517156,
      "loss": 1.2649,
      "step": 5769
    },
    {
      "epoch": 0.4964294932461499,
      "grad_norm": 1.25,
      "learning_rate": 0.0009760601956639388,
      "loss": 1.0743,
      "step": 5770
    },
    {
      "epoch": 0.49651552955347156,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009760519279837785,
      "loss": 2.0175,
      "step": 5771
    },
    {
      "epoch": 0.49660156586079324,
      "grad_norm": 1.28125,
      "learning_rate": 0.000976043658911259,
      "loss": 0.9745,
      "step": 5772
    },
    {
      "epoch": 0.49668760216811497,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009760353884464044,
      "loss": 2.0297,
      "step": 5773
    },
    {
      "epoch": 0.49677363847543665,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009760271165892387,
      "loss": 1.5899,
      "step": 5774
    },
    {
      "epoch": 0.49685967478275833,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009760188433397863,
      "loss": 1.174,
      "step": 5775
    },
    {
      "epoch": 0.49694571109008,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009760105686980715,
      "loss": 0.9949,
      "step": 5776
    },
    {
      "epoch": 0.4970317473974017,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009760022926641183,
      "loss": 1.633,
      "step": 5777
    },
    {
      "epoch": 0.49711778370472337,
      "grad_norm": 1.1875,
      "learning_rate": 0.000975994015237951,
      "loss": 1.3505,
      "step": 5778
    },
    {
      "epoch": 0.4972038200120451,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009759857364195937,
      "loss": 1.0642,
      "step": 5779
    },
    {
      "epoch": 0.4972898563193668,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009759774562090707,
      "loss": 1.3524,
      "step": 5780
    },
    {
      "epoch": 0.49737589262668847,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009759691746064062,
      "loss": 0.8416,
      "step": 5781
    },
    {
      "epoch": 0.49746192893401014,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009759608916116245,
      "loss": 1.3965,
      "step": 5782
    },
    {
      "epoch": 0.4975479652413318,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009759526072247497,
      "loss": 1.1792,
      "step": 5783
    },
    {
      "epoch": 0.49763400154865356,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009759443214458061,
      "loss": 1.8294,
      "step": 5784
    },
    {
      "epoch": 0.49772003785597524,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009759360342748181,
      "loss": 1.5727,
      "step": 5785
    },
    {
      "epoch": 0.4978060741632969,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009759277457118095,
      "loss": 2.0012,
      "step": 5786
    },
    {
      "epoch": 0.4978921104706186,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009759194557568049,
      "loss": 1.8653,
      "step": 5787
    },
    {
      "epoch": 0.4979781467779403,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009759111644098286,
      "loss": 2.144,
      "step": 5788
    },
    {
      "epoch": 0.49806418308526196,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009759028716709046,
      "loss": 1.1921,
      "step": 5789
    },
    {
      "epoch": 0.4981502193925837,
      "grad_norm": 2.671875,
      "learning_rate": 0.0009758945775400572,
      "loss": 2.326,
      "step": 5790
    },
    {
      "epoch": 0.4982362556999054,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009758862820173107,
      "loss": 1.1236,
      "step": 5791
    },
    {
      "epoch": 0.49832229200722705,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009758779851026895,
      "loss": 1.9269,
      "step": 5792
    },
    {
      "epoch": 0.49840832831454873,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009758696867962176,
      "loss": 1.0546,
      "step": 5793
    },
    {
      "epoch": 0.4984943646218704,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009758613870979197,
      "loss": 1.4609,
      "step": 5794
    },
    {
      "epoch": 0.49858040092919215,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009758530860078194,
      "loss": 1.5356,
      "step": 5795
    },
    {
      "epoch": 0.4986664372365138,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009758447835259416,
      "loss": 1.5424,
      "step": 5796
    },
    {
      "epoch": 0.4987524735438355,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009758364796523104,
      "loss": 1.9684,
      "step": 5797
    },
    {
      "epoch": 0.4988385098511572,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00097582817438695,
      "loss": 1.5446,
      "step": 5798
    },
    {
      "epoch": 0.49892454615847887,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009758198677298846,
      "loss": 1.3847,
      "step": 5799
    },
    {
      "epoch": 0.49901058246580055,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009758115596811388,
      "loss": 1.3082,
      "step": 5800
    },
    {
      "epoch": 0.4990966187731223,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009758032502407366,
      "loss": 1.1588,
      "step": 5801
    },
    {
      "epoch": 0.49918265508044396,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009757949394087024,
      "loss": 1.9279,
      "step": 5802
    },
    {
      "epoch": 0.49926869138776564,
      "grad_norm": 1.125,
      "learning_rate": 0.0009757866271850607,
      "loss": 1.4565,
      "step": 5803
    },
    {
      "epoch": 0.4993547276950873,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009757783135698356,
      "loss": 1.0671,
      "step": 5804
    },
    {
      "epoch": 0.499440764002409,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009757699985630515,
      "loss": 1.1891,
      "step": 5805
    },
    {
      "epoch": 0.49952680030973073,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009757616821647326,
      "loss": 1.612,
      "step": 5806
    },
    {
      "epoch": 0.4996128366170524,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009757533643749033,
      "loss": 1.3257,
      "step": 5807
    },
    {
      "epoch": 0.4996988729243741,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000975745045193588,
      "loss": 1.1756,
      "step": 5808
    },
    {
      "epoch": 0.4997849092316958,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009757367246208109,
      "loss": 1.0718,
      "step": 5809
    },
    {
      "epoch": 0.49987094553901745,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009757284026565965,
      "loss": 1.754,
      "step": 5810
    },
    {
      "epoch": 0.49995698184633913,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000975720079300969,
      "loss": 0.8492,
      "step": 5811
    },
    {
      "epoch": 0.5000430181536608,
      "grad_norm": 1.75,
      "learning_rate": 0.0009757117545539528,
      "loss": 1.8281,
      "step": 5812
    },
    {
      "epoch": 0.5000430181536608,
      "eval_loss": 1.3462311029434204,
      "eval_runtime": 355.7498,
      "eval_samples_per_second": 13.068,
      "eval_steps_per_second": 1.636,
      "step": 5812
    },
    {
      "epoch": 0.5001290544609825,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009757034284155722,
      "loss": 1.1915,
      "step": 5813
    },
    {
      "epoch": 0.5002150907683042,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009756951008858517,
      "loss": 0.9141,
      "step": 5814
    },
    {
      "epoch": 0.5003011270756259,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009756867719648154,
      "loss": 1.9165,
      "step": 5815
    },
    {
      "epoch": 0.5003871633829476,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009756784416524879,
      "loss": 1.7029,
      "step": 5816
    },
    {
      "epoch": 0.5004731996902693,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009756701099488934,
      "loss": 1.2935,
      "step": 5817
    },
    {
      "epoch": 0.500559235997591,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009756617768540564,
      "loss": 0.9223,
      "step": 5818
    },
    {
      "epoch": 0.5006452723049126,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009756534423680012,
      "loss": 1.4083,
      "step": 5819
    },
    {
      "epoch": 0.5007313086122344,
      "grad_norm": 1.375,
      "learning_rate": 0.0009756451064907522,
      "loss": 1.3591,
      "step": 5820
    },
    {
      "epoch": 0.5008173449195561,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009756367692223338,
      "loss": 1.2601,
      "step": 5821
    },
    {
      "epoch": 0.5009033812268777,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009756284305627703,
      "loss": 1.5051,
      "step": 5822
    },
    {
      "epoch": 0.5009894175341995,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009756200905120862,
      "loss": 1.7634,
      "step": 5823
    },
    {
      "epoch": 0.5010754538415211,
      "grad_norm": 14.25,
      "learning_rate": 0.0009756117490703058,
      "loss": 1.4401,
      "step": 5824
    },
    {
      "epoch": 0.5011614901488428,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009756034062374535,
      "loss": 2.0031,
      "step": 5825
    },
    {
      "epoch": 0.5012475264561645,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009755950620135539,
      "loss": 1.1215,
      "step": 5826
    },
    {
      "epoch": 0.5013335627634862,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009755867163986309,
      "loss": 0.7365,
      "step": 5827
    },
    {
      "epoch": 0.5014195990708079,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009755783693927096,
      "loss": 0.6871,
      "step": 5828
    },
    {
      "epoch": 0.5015056353781295,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000975570020995814,
      "loss": 0.9648,
      "step": 5829
    },
    {
      "epoch": 0.5015916716854513,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009755616712079684,
      "loss": 1.0008,
      "step": 5830
    },
    {
      "epoch": 0.501677707992773,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009755533200291975,
      "loss": 2.2236,
      "step": 5831
    },
    {
      "epoch": 0.5017637443000946,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009755449674595257,
      "loss": 1.4756,
      "step": 5832
    },
    {
      "epoch": 0.5018497806074164,
      "grad_norm": 1.5,
      "learning_rate": 0.0009755366134989774,
      "loss": 1.0936,
      "step": 5833
    },
    {
      "epoch": 0.501935816914738,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009755282581475768,
      "loss": 1.2734,
      "step": 5834
    },
    {
      "epoch": 0.5020218532220597,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009755199014053486,
      "loss": 1.3874,
      "step": 5835
    },
    {
      "epoch": 0.5021078895293813,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009755115432723172,
      "loss": 1.6094,
      "step": 5836
    },
    {
      "epoch": 0.5021939258367031,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009755031837485071,
      "loss": 1.0109,
      "step": 5837
    },
    {
      "epoch": 0.5022799621440248,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009754948228339425,
      "loss": 1.3739,
      "step": 5838
    },
    {
      "epoch": 0.5023659984513464,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009754864605286482,
      "loss": 1.51,
      "step": 5839
    },
    {
      "epoch": 0.5024520347586682,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009754780968326483,
      "loss": 1.2955,
      "step": 5840
    },
    {
      "epoch": 0.5025380710659898,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009754697317459675,
      "loss": 1.8054,
      "step": 5841
    },
    {
      "epoch": 0.5026241073733115,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009754613652686304,
      "loss": 1.0663,
      "step": 5842
    },
    {
      "epoch": 0.5027101436806333,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000975452997400661,
      "loss": 0.9868,
      "step": 5843
    },
    {
      "epoch": 0.5027961799879549,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009754446281420841,
      "loss": 0.6834,
      "step": 5844
    },
    {
      "epoch": 0.5028822162952766,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009754362574929243,
      "loss": 1.8733,
      "step": 5845
    },
    {
      "epoch": 0.5029682526025983,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009754278854532058,
      "loss": 0.9336,
      "step": 5846
    },
    {
      "epoch": 0.50305428890992,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009754195120229532,
      "loss": 1.506,
      "step": 5847
    },
    {
      "epoch": 0.5031403252172417,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009754111372021909,
      "loss": 0.8681,
      "step": 5848
    },
    {
      "epoch": 0.5032263615245633,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009754027609909438,
      "loss": 1.1126,
      "step": 5849
    },
    {
      "epoch": 0.5033123978318851,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009753943833892358,
      "loss": 1.5744,
      "step": 5850
    },
    {
      "epoch": 0.5033984341392067,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009753860043970918,
      "loss": 1.1927,
      "step": 5851
    },
    {
      "epoch": 0.5034844704465284,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009753776240145363,
      "loss": 1.4757,
      "step": 5852
    },
    {
      "epoch": 0.5035705067538502,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009753692422415934,
      "loss": 1.2819,
      "step": 5853
    },
    {
      "epoch": 0.5036565430611718,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009753608590782881,
      "loss": 1.133,
      "step": 5854
    },
    {
      "epoch": 0.5037425793684935,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009753524745246447,
      "loss": 1.128,
      "step": 5855
    },
    {
      "epoch": 0.5038286156758152,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009753440885806879,
      "loss": 1.6876,
      "step": 5856
    },
    {
      "epoch": 0.5039146519831369,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009753357012464419,
      "loss": 1.7344,
      "step": 5857
    },
    {
      "epoch": 0.5040006882904585,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009753273125219315,
      "loss": 1.1824,
      "step": 5858
    },
    {
      "epoch": 0.5040867245977803,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009753189224071812,
      "loss": 0.7453,
      "step": 5859
    },
    {
      "epoch": 0.504172760905102,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009753105309022155,
      "loss": 1.807,
      "step": 5860
    },
    {
      "epoch": 0.5042587972124236,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009753021380070588,
      "loss": 2.04,
      "step": 5861
    },
    {
      "epoch": 0.5043448335197454,
      "grad_norm": 1.8984375,
      "learning_rate": 0.000975293743721736,
      "loss": 1.4979,
      "step": 5862
    },
    {
      "epoch": 0.504430869827067,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009752853480462713,
      "loss": 1.0265,
      "step": 5863
    },
    {
      "epoch": 0.5045169061343887,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009752769509806893,
      "loss": 1.3205,
      "step": 5864
    },
    {
      "epoch": 0.5046029424417104,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009752685525250148,
      "loss": 1.0561,
      "step": 5865
    },
    {
      "epoch": 0.5046889787490321,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009752601526792721,
      "loss": 1.2066,
      "step": 5866
    },
    {
      "epoch": 0.5047750150563538,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009752517514434861,
      "loss": 1.0441,
      "step": 5867
    },
    {
      "epoch": 0.5048610513636754,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009752433488176811,
      "loss": 1.2244,
      "step": 5868
    },
    {
      "epoch": 0.5049470876709972,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009752349448018815,
      "loss": 1.0068,
      "step": 5869
    },
    {
      "epoch": 0.5050331239783189,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009752265393961123,
      "loss": 0.8093,
      "step": 5870
    },
    {
      "epoch": 0.5051191602856405,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009752181326003979,
      "loss": 1.3648,
      "step": 5871
    },
    {
      "epoch": 0.5052051965929623,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009752097244147629,
      "loss": 0.9266,
      "step": 5872
    },
    {
      "epoch": 0.5052912329002839,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009752013148392318,
      "loss": 1.1766,
      "step": 5873
    },
    {
      "epoch": 0.5053772692076056,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0009751929038738295,
      "loss": 1.5245,
      "step": 5874
    },
    {
      "epoch": 0.5054633055149274,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009751844915185801,
      "loss": 2.1738,
      "step": 5875
    },
    {
      "epoch": 0.505549341822249,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009751760777735086,
      "loss": 1.5423,
      "step": 5876
    },
    {
      "epoch": 0.5056353781295707,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009751676626386395,
      "loss": 1.3063,
      "step": 5877
    },
    {
      "epoch": 0.5057214144368923,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009751592461139973,
      "loss": 0.924,
      "step": 5878
    },
    {
      "epoch": 0.5058074507442141,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009751508281996069,
      "loss": 0.7257,
      "step": 5879
    },
    {
      "epoch": 0.5058934870515357,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009751424088954926,
      "loss": 2.2442,
      "step": 5880
    },
    {
      "epoch": 0.5059795233588574,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009751339882016792,
      "loss": 1.5064,
      "step": 5881
    },
    {
      "epoch": 0.5060655596661792,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009751255661181913,
      "loss": 1.2422,
      "step": 5882
    },
    {
      "epoch": 0.5061515959735008,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009751171426450534,
      "loss": 1.2732,
      "step": 5883
    },
    {
      "epoch": 0.5062376322808225,
      "grad_norm": 1.125,
      "learning_rate": 0.0009751087177822903,
      "loss": 1.4842,
      "step": 5884
    },
    {
      "epoch": 0.5063236685881441,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009751002915299267,
      "loss": 1.1323,
      "step": 5885
    },
    {
      "epoch": 0.5064097048954659,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009750918638879872,
      "loss": 1.3643,
      "step": 5886
    },
    {
      "epoch": 0.5064957412027876,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009750834348564963,
      "loss": 1.6511,
      "step": 5887
    },
    {
      "epoch": 0.5065817775101092,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009750750044354788,
      "loss": 0.9936,
      "step": 5888
    },
    {
      "epoch": 0.506667813817431,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009750665726249592,
      "loss": 0.9495,
      "step": 5889
    },
    {
      "epoch": 0.5067538501247526,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009750581394249625,
      "loss": 0.9317,
      "step": 5890
    },
    {
      "epoch": 0.5068398864320743,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000975049704835513,
      "loss": 1.5854,
      "step": 5891
    },
    {
      "epoch": 0.5069259227393961,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009750412688566355,
      "loss": 1.5821,
      "step": 5892
    },
    {
      "epoch": 0.5070119590467177,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009750328314883546,
      "loss": 1.392,
      "step": 5893
    },
    {
      "epoch": 0.5070979953540394,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009750243927306952,
      "loss": 0.9525,
      "step": 5894
    },
    {
      "epoch": 0.5071840316613611,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009750159525836817,
      "loss": 1.94,
      "step": 5895
    },
    {
      "epoch": 0.5072700679686828,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009750075110473391,
      "loss": 1.0577,
      "step": 5896
    },
    {
      "epoch": 0.5073561042760045,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009749990681216918,
      "loss": 0.8315,
      "step": 5897
    },
    {
      "epoch": 0.5074421405833262,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009749906238067645,
      "loss": 1.3089,
      "step": 5898
    },
    {
      "epoch": 0.5075281768906479,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009749821781025823,
      "loss": 2.0062,
      "step": 5899
    },
    {
      "epoch": 0.5076142131979695,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009749737310091693,
      "loss": 1.5849,
      "step": 5900
    },
    {
      "epoch": 0.5077002495052912,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009749652825265507,
      "loss": 0.9844,
      "step": 5901
    },
    {
      "epoch": 0.5077862858126129,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009749568326547508,
      "loss": 1.6494,
      "step": 5902
    },
    {
      "epoch": 0.5078723221199346,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009749483813937949,
      "loss": 1.0252,
      "step": 5903
    },
    {
      "epoch": 0.5079583584272563,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000974939928743707,
      "loss": 0.9765,
      "step": 5904
    },
    {
      "epoch": 0.508044394734578,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009749314747045125,
      "loss": 0.877,
      "step": 5905
    },
    {
      "epoch": 0.5081304310418997,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009749230192762356,
      "loss": 1.5783,
      "step": 5906
    },
    {
      "epoch": 0.5082164673492213,
      "grad_norm": 1.125,
      "learning_rate": 0.0009749145624589013,
      "loss": 1.626,
      "step": 5907
    },
    {
      "epoch": 0.5083025036565431,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009749061042525342,
      "loss": 1.5003,
      "step": 5908
    },
    {
      "epoch": 0.5083885399638648,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009748976446571592,
      "loss": 1.3225,
      "step": 5909
    },
    {
      "epoch": 0.5084745762711864,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009748891836728009,
      "loss": 1.9848,
      "step": 5910
    },
    {
      "epoch": 0.5085606125785082,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009748807212994841,
      "loss": 1.4642,
      "step": 5911
    },
    {
      "epoch": 0.5086466488858298,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009748722575372336,
      "loss": 1.8152,
      "step": 5912
    },
    {
      "epoch": 0.5087326851931515,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009748637923860741,
      "loss": 1.5589,
      "step": 5913
    },
    {
      "epoch": 0.5088187215004732,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009748553258460304,
      "loss": 1.8666,
      "step": 5914
    },
    {
      "epoch": 0.5089047578077949,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009748468579171271,
      "loss": 1.7476,
      "step": 5915
    },
    {
      "epoch": 0.5089907941151166,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009748383885993892,
      "loss": 2.1734,
      "step": 5916
    },
    {
      "epoch": 0.5090768304224382,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009748299178928412,
      "loss": 1.1124,
      "step": 5917
    },
    {
      "epoch": 0.50916286672976,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009748214457975082,
      "loss": 0.7363,
      "step": 5918
    },
    {
      "epoch": 0.5092489030370817,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009748129723134148,
      "loss": 1.6468,
      "step": 5919
    },
    {
      "epoch": 0.5093349393444033,
      "grad_norm": 1.125,
      "learning_rate": 0.0009748044974405858,
      "loss": 1.3407,
      "step": 5920
    },
    {
      "epoch": 0.5094209756517251,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009747960211790459,
      "loss": 1.6828,
      "step": 5921
    },
    {
      "epoch": 0.5095070119590467,
      "grad_norm": 1.703125,
      "learning_rate": 0.00097478754352882,
      "loss": 2.4001,
      "step": 5922
    },
    {
      "epoch": 0.5095930482663684,
      "grad_norm": 1.34375,
      "learning_rate": 0.000974779064489933,
      "loss": 1.5262,
      "step": 5923
    },
    {
      "epoch": 0.50967908457369,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009747705840624095,
      "loss": 1.9331,
      "step": 5924
    },
    {
      "epoch": 0.5097651208810118,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009747621022462745,
      "loss": 1.12,
      "step": 5925
    },
    {
      "epoch": 0.5098511571883335,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009747536190415525,
      "loss": 0.9889,
      "step": 5926
    },
    {
      "epoch": 0.5099371934956551,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009747451344482686,
      "loss": 1.2729,
      "step": 5927
    },
    {
      "epoch": 0.5100232298029769,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009747366484664474,
      "loss": 1.7025,
      "step": 5928
    },
    {
      "epoch": 0.5101092661102985,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000974728161096114,
      "loss": 0.9246,
      "step": 5929
    },
    {
      "epoch": 0.5101953024176202,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009747196723372929,
      "loss": 1.0095,
      "step": 5930
    },
    {
      "epoch": 0.510281338724942,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009747111821900092,
      "loss": 0.7585,
      "step": 5931
    },
    {
      "epoch": 0.5103673750322636,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009747026906542875,
      "loss": 1.6588,
      "step": 5932
    },
    {
      "epoch": 0.5104534113395853,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009746941977301529,
      "loss": 0.693,
      "step": 5933
    },
    {
      "epoch": 0.510539447646907,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00097468570341763,
      "loss": 0.9057,
      "step": 5934
    },
    {
      "epoch": 0.5106254839542287,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009746772077167438,
      "loss": 0.9341,
      "step": 5935
    },
    {
      "epoch": 0.5107115202615504,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000974668710627519,
      "loss": 1.631,
      "step": 5936
    },
    {
      "epoch": 0.510797556568872,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009746602121499805,
      "loss": 1.3688,
      "step": 5937
    },
    {
      "epoch": 0.5108835928761938,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009746517122841534,
      "loss": 1.0194,
      "step": 5938
    },
    {
      "epoch": 0.5109696291835154,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009746432110300622,
      "loss": 1.2952,
      "step": 5939
    },
    {
      "epoch": 0.5110556654908371,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009746347083877321,
      "loss": 2.1852,
      "step": 5940
    },
    {
      "epoch": 0.5111417017981589,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009746262043571877,
      "loss": 2.0072,
      "step": 5941
    },
    {
      "epoch": 0.5112277381054805,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009746176989384539,
      "loss": 0.8069,
      "step": 5942
    },
    {
      "epoch": 0.5113137744128022,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009746091921315557,
      "loss": 1.2335,
      "step": 5943
    },
    {
      "epoch": 0.5113998107201239,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009746006839365178,
      "loss": 2.0349,
      "step": 5944
    },
    {
      "epoch": 0.5114858470274456,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009745921743533652,
      "loss": 0.8847,
      "step": 5945
    },
    {
      "epoch": 0.5115718833347672,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009745836633821229,
      "loss": 1.3662,
      "step": 5946
    },
    {
      "epoch": 0.511657919642089,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009745751510228155,
      "loss": 0.532,
      "step": 5947
    },
    {
      "epoch": 0.5117439559494107,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009745666372754685,
      "loss": 2.0233,
      "step": 5948
    },
    {
      "epoch": 0.5118299922567323,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009745581221401061,
      "loss": 0.7424,
      "step": 5949
    },
    {
      "epoch": 0.511916028564054,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009745496056167535,
      "loss": 1.4374,
      "step": 5950
    },
    {
      "epoch": 0.5120020648713757,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009745410877054355,
      "loss": 0.9592,
      "step": 5951
    },
    {
      "epoch": 0.5120881011786974,
      "grad_norm": 1.125,
      "learning_rate": 0.0009745325684061772,
      "loss": 1.7094,
      "step": 5952
    },
    {
      "epoch": 0.5121741374860191,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009745240477190035,
      "loss": 1.8651,
      "step": 5953
    },
    {
      "epoch": 0.5122601737933408,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009745155256439392,
      "loss": 1.1151,
      "step": 5954
    },
    {
      "epoch": 0.5123462101006625,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009745070021810093,
      "loss": 1.4693,
      "step": 5955
    },
    {
      "epoch": 0.5124322464079841,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009744984773302387,
      "loss": 1.2063,
      "step": 5956
    },
    {
      "epoch": 0.5125182827153059,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009744899510916523,
      "loss": 1.3673,
      "step": 5957
    },
    {
      "epoch": 0.5126043190226276,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000974481423465275,
      "loss": 1.0103,
      "step": 5958
    },
    {
      "epoch": 0.5126903553299492,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009744728944511321,
      "loss": 1.0816,
      "step": 5959
    },
    {
      "epoch": 0.512776391637271,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000974464364049248,
      "loss": 0.9106,
      "step": 5960
    },
    {
      "epoch": 0.5128624279445926,
      "grad_norm": 1.34375,
      "learning_rate": 0.000974455832259648,
      "loss": 2.4382,
      "step": 5961
    },
    {
      "epoch": 0.5129484642519143,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000974447299082357,
      "loss": 1.2821,
      "step": 5962
    },
    {
      "epoch": 0.513034500559236,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009744387645173998,
      "loss": 1.0347,
      "step": 5963
    },
    {
      "epoch": 0.5131205368665577,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009744302285648016,
      "loss": 1.0098,
      "step": 5964
    },
    {
      "epoch": 0.5132065731738794,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009744216912245873,
      "loss": 2.1458,
      "step": 5965
    },
    {
      "epoch": 0.513292609481201,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009744131524967816,
      "loss": 1.4447,
      "step": 5966
    },
    {
      "epoch": 0.5133786457885228,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009744046123814099,
      "loss": 2.2496,
      "step": 5967
    },
    {
      "epoch": 0.5134646820958444,
      "grad_norm": 1.328125,
      "learning_rate": 0.000974396070878497,
      "loss": 1.6108,
      "step": 5968
    },
    {
      "epoch": 0.5135507184031661,
      "grad_norm": 2.828125,
      "learning_rate": 0.0009743875279880677,
      "loss": 1.4962,
      "step": 5969
    },
    {
      "epoch": 0.5136367547104879,
      "grad_norm": 1.25,
      "learning_rate": 0.0009743789837101472,
      "loss": 0.9535,
      "step": 5970
    },
    {
      "epoch": 0.5137227910178095,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009743704380447604,
      "loss": 1.3954,
      "step": 5971
    },
    {
      "epoch": 0.5138088273251312,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009743618909919326,
      "loss": 1.2348,
      "step": 5972
    },
    {
      "epoch": 0.5138948636324528,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009743533425516881,
      "loss": 1.0319,
      "step": 5973
    },
    {
      "epoch": 0.5139808999397746,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009743447927240527,
      "loss": 1.1196,
      "step": 5974
    },
    {
      "epoch": 0.5140669362470963,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009743362415090508,
      "loss": 1.1888,
      "step": 5975
    },
    {
      "epoch": 0.5141529725544179,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009743276889067077,
      "loss": 1.3921,
      "step": 5976
    },
    {
      "epoch": 0.5142390088617397,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009743191349170484,
      "loss": 1.5906,
      "step": 5977
    },
    {
      "epoch": 0.5143250451690613,
      "grad_norm": 1.25,
      "learning_rate": 0.0009743105795400979,
      "loss": 1.587,
      "step": 5978
    },
    {
      "epoch": 0.514411081476383,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009743020227758812,
      "loss": 0.8527,
      "step": 5979
    },
    {
      "epoch": 0.5144971177837048,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009742934646244232,
      "loss": 1.5186,
      "step": 5980
    },
    {
      "epoch": 0.5145831540910264,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009742849050857491,
      "loss": 1.1658,
      "step": 5981
    },
    {
      "epoch": 0.5146691903983481,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009742763441598841,
      "loss": 0.7411,
      "step": 5982
    },
    {
      "epoch": 0.5147552267056698,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009742677818468528,
      "loss": 1.6742,
      "step": 5983
    },
    {
      "epoch": 0.5148412630129915,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009742592181466807,
      "loss": 1.687,
      "step": 5984
    },
    {
      "epoch": 0.5149272993203132,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009742506530593924,
      "loss": 1.9462,
      "step": 5985
    },
    {
      "epoch": 0.5150133356276348,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009742420865850132,
      "loss": 1.1118,
      "step": 5986
    },
    {
      "epoch": 0.5150993719349566,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009742335187235682,
      "loss": 0.759,
      "step": 5987
    },
    {
      "epoch": 0.5151854082422782,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009742249494750825,
      "loss": 0.9818,
      "step": 5988
    },
    {
      "epoch": 0.5152714445495999,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000974216378839581,
      "loss": 1.2222,
      "step": 5989
    },
    {
      "epoch": 0.5153574808569216,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009742078068170888,
      "loss": 1.1506,
      "step": 5990
    },
    {
      "epoch": 0.5154435171642433,
      "grad_norm": 4.78125,
      "learning_rate": 0.0009741992334076308,
      "loss": 1.5685,
      "step": 5991
    },
    {
      "epoch": 0.515529553471565,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009741906586112326,
      "loss": 1.2867,
      "step": 5992
    },
    {
      "epoch": 0.5156155897788867,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009741820824279187,
      "loss": 0.9988,
      "step": 5993
    },
    {
      "epoch": 0.5157016260862084,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009741735048577145,
      "loss": 1.6345,
      "step": 5994
    },
    {
      "epoch": 0.51578766239353,
      "grad_norm": 1.0,
      "learning_rate": 0.0009741649259006451,
      "loss": 1.5665,
      "step": 5995
    },
    {
      "epoch": 0.5158736987008518,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009741563455567354,
      "loss": 0.9177,
      "step": 5996
    },
    {
      "epoch": 0.5159597350081735,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009741477638260107,
      "loss": 1.4712,
      "step": 5997
    },
    {
      "epoch": 0.5160457713154951,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000974139180708496,
      "loss": 1.4267,
      "step": 5998
    },
    {
      "epoch": 0.5161318076228169,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009741305962042163,
      "loss": 1.797,
      "step": 5999
    },
    {
      "epoch": 0.5162178439301385,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009741220103131969,
      "loss": 1.3348,
      "step": 6000
    },
    {
      "epoch": 0.5163038802374602,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009741134230354628,
      "loss": 1.3287,
      "step": 6001
    },
    {
      "epoch": 0.516389916544782,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009741048343710392,
      "loss": 1.1609,
      "step": 6002
    },
    {
      "epoch": 0.5164759528521036,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009740962443199511,
      "loss": 1.9652,
      "step": 6003
    },
    {
      "epoch": 0.5165619891594253,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009740876528822238,
      "loss": 1.2356,
      "step": 6004
    },
    {
      "epoch": 0.5166480254667469,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009740790600578822,
      "loss": 1.4693,
      "step": 6005
    },
    {
      "epoch": 0.5167340617740687,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009740704658469517,
      "loss": 0.7939,
      "step": 6006
    },
    {
      "epoch": 0.5168200980813904,
      "grad_norm": 1.796875,
      "learning_rate": 0.000974061870249457,
      "loss": 1.68,
      "step": 6007
    },
    {
      "epoch": 0.516906134388712,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009740532732654239,
      "loss": 1.3631,
      "step": 6008
    },
    {
      "epoch": 0.5169921706960338,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009740446748948769,
      "loss": 1.0853,
      "step": 6009
    },
    {
      "epoch": 0.5170782070033554,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009740360751378415,
      "loss": 1.8029,
      "step": 6010
    },
    {
      "epoch": 0.5171642433106771,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009740274739943428,
      "loss": 1.5773,
      "step": 6011
    },
    {
      "epoch": 0.5172502796179987,
      "grad_norm": 1.3125,
      "learning_rate": 0.000974018871464406,
      "loss": 1.3157,
      "step": 6012
    },
    {
      "epoch": 0.5173363159253205,
      "grad_norm": 1.265625,
      "learning_rate": 0.000974010267548056,
      "loss": 1.3376,
      "step": 6013
    },
    {
      "epoch": 0.5174223522326422,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009740016622453183,
      "loss": 0.6529,
      "step": 6014
    },
    {
      "epoch": 0.5175083885399638,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000973993055556218,
      "loss": 1.5741,
      "step": 6015
    },
    {
      "epoch": 0.5175944248472856,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00097398444748078,
      "loss": 1.3123,
      "step": 6016
    },
    {
      "epoch": 0.5176804611546072,
      "grad_norm": 1.625,
      "learning_rate": 0.0009739758380190298,
      "loss": 0.9289,
      "step": 6017
    },
    {
      "epoch": 0.5177664974619289,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009739672271709925,
      "loss": 1.637,
      "step": 6018
    },
    {
      "epoch": 0.5178525337692507,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009739586149366932,
      "loss": 1.5317,
      "step": 6019
    },
    {
      "epoch": 0.5179385700765723,
      "grad_norm": 1.15625,
      "learning_rate": 0.000973950001316157,
      "loss": 1.3548,
      "step": 6020
    },
    {
      "epoch": 0.518024606383894,
      "grad_norm": 1.625,
      "learning_rate": 0.0009739413863094094,
      "loss": 1.1979,
      "step": 6021
    },
    {
      "epoch": 0.5181106426912157,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009739327699164752,
      "loss": 1.8941,
      "step": 6022
    },
    {
      "epoch": 0.5181966789985374,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009739241521373801,
      "loss": 1.1495,
      "step": 6023
    },
    {
      "epoch": 0.5182827153058591,
      "grad_norm": 1.328125,
      "learning_rate": 0.000973915532972149,
      "loss": 1.0002,
      "step": 6024
    },
    {
      "epoch": 0.5183687516131807,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009739069124208068,
      "loss": 1.7295,
      "step": 6025
    },
    {
      "epoch": 0.5184547879205025,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009738982904833794,
      "loss": 1.663,
      "step": 6026
    },
    {
      "epoch": 0.5185408242278241,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009738896671598916,
      "loss": 1.1462,
      "step": 6027
    },
    {
      "epoch": 0.5186268605351458,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009738810424503685,
      "loss": 1.5115,
      "step": 6028
    },
    {
      "epoch": 0.5187128968424676,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009738724163548358,
      "loss": 1.0835,
      "step": 6029
    },
    {
      "epoch": 0.5187989331497892,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009738637888733184,
      "loss": 1.2896,
      "step": 6030
    },
    {
      "epoch": 0.5188849694571109,
      "grad_norm": 1.625,
      "learning_rate": 0.0009738551600058414,
      "loss": 1.3816,
      "step": 6031
    },
    {
      "epoch": 0.5189710057644326,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009738465297524304,
      "loss": 1.497,
      "step": 6032
    },
    {
      "epoch": 0.5190570420717543,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009738378981131103,
      "loss": 2.3436,
      "step": 6033
    },
    {
      "epoch": 0.5191430783790759,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009738292650879065,
      "loss": 0.9828,
      "step": 6034
    },
    {
      "epoch": 0.5192291146863977,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009738206306768443,
      "loss": 1.1951,
      "step": 6035
    },
    {
      "epoch": 0.5193151509937194,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000973811994879949,
      "loss": 1.0941,
      "step": 6036
    },
    {
      "epoch": 0.519401187301041,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009738033576972457,
      "loss": 1.5035,
      "step": 6037
    },
    {
      "epoch": 0.5194872236083627,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009737947191287597,
      "loss": 1.2931,
      "step": 6038
    },
    {
      "epoch": 0.5195732599156844,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009737860791745164,
      "loss": 1.7281,
      "step": 6039
    },
    {
      "epoch": 0.5196592962230061,
      "grad_norm": 1.125,
      "learning_rate": 0.0009737774378345408,
      "loss": 1.0266,
      "step": 6040
    },
    {
      "epoch": 0.5197453325303278,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009737687951088585,
      "loss": 1.273,
      "step": 6041
    },
    {
      "epoch": 0.5198313688376495,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009737601509974945,
      "loss": 1.651,
      "step": 6042
    },
    {
      "epoch": 0.5199174051449712,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009737515055004742,
      "loss": 1.9363,
      "step": 6043
    },
    {
      "epoch": 0.5200034414522928,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000973742858617823,
      "loss": 0.8044,
      "step": 6044
    },
    {
      "epoch": 0.5200894777596146,
      "grad_norm": 2.71875,
      "learning_rate": 0.000973734210349566,
      "loss": 1.512,
      "step": 6045
    },
    {
      "epoch": 0.5201755140669363,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009737255606957285,
      "loss": 1.035,
      "step": 6046
    },
    {
      "epoch": 0.5202615503742579,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009737169096563361,
      "loss": 1.2999,
      "step": 6047
    },
    {
      "epoch": 0.5203475866815797,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009737082572314136,
      "loss": 1.519,
      "step": 6048
    },
    {
      "epoch": 0.5204336229889013,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009736996034209866,
      "loss": 2.1357,
      "step": 6049
    },
    {
      "epoch": 0.520519659296223,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009736909482250806,
      "loss": 1.7977,
      "step": 6050
    },
    {
      "epoch": 0.5206056956035447,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009736822916437206,
      "loss": 1.2599,
      "step": 6051
    },
    {
      "epoch": 0.5206917319108664,
      "grad_norm": 7.96875,
      "learning_rate": 0.0009736736336769321,
      "loss": 2.1801,
      "step": 6052
    },
    {
      "epoch": 0.5207777682181881,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009736649743247401,
      "loss": 0.8905,
      "step": 6053
    },
    {
      "epoch": 0.5208638045255097,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009736563135871704,
      "loss": 1.6473,
      "step": 6054
    },
    {
      "epoch": 0.5209498408328315,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000973647651464248,
      "loss": 1.0129,
      "step": 6055
    },
    {
      "epoch": 0.5210358771401531,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009736389879559983,
      "loss": 1.2539,
      "step": 6056
    },
    {
      "epoch": 0.5211219134474748,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009736303230624468,
      "loss": 1.0755,
      "step": 6057
    },
    {
      "epoch": 0.5212079497547966,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009736216567836187,
      "loss": 1.9424,
      "step": 6058
    },
    {
      "epoch": 0.5212939860621182,
      "grad_norm": 1.375,
      "learning_rate": 0.0009736129891195393,
      "loss": 1.4088,
      "step": 6059
    },
    {
      "epoch": 0.5213800223694399,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009736043200702342,
      "loss": 1.7293,
      "step": 6060
    },
    {
      "epoch": 0.5214660586767615,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009735956496357283,
      "loss": 1.126,
      "step": 6061
    },
    {
      "epoch": 0.5215520949840833,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009735869778160475,
      "loss": 1.3063,
      "step": 6062
    },
    {
      "epoch": 0.521638131291405,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009735783046112167,
      "loss": 1.4477,
      "step": 6063
    },
    {
      "epoch": 0.5217241675987266,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009735696300212615,
      "loss": 1.0953,
      "step": 6064
    },
    {
      "epoch": 0.5218102039060484,
      "grad_norm": 1.125,
      "learning_rate": 0.0009735609540462072,
      "loss": 1.294,
      "step": 6065
    },
    {
      "epoch": 0.52189624021337,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009735522766860794,
      "loss": 1.2924,
      "step": 6066
    },
    {
      "epoch": 0.5219822765206917,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009735435979409031,
      "loss": 0.7433,
      "step": 6067
    },
    {
      "epoch": 0.5220683128280135,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000973534917810704,
      "loss": 1.0443,
      "step": 6068
    },
    {
      "epoch": 0.5221543491353351,
      "grad_norm": 1.0,
      "learning_rate": 0.0009735262362955073,
      "loss": 1.1918,
      "step": 6069
    },
    {
      "epoch": 0.5222403854426568,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009735175533953385,
      "loss": 0.9438,
      "step": 6070
    },
    {
      "epoch": 0.5223264217499785,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009735088691102228,
      "loss": 1.4896,
      "step": 6071
    },
    {
      "epoch": 0.5224124580573002,
      "grad_norm": 1.234375,
      "learning_rate": 0.000973500183440186,
      "loss": 1.5743,
      "step": 6072
    },
    {
      "epoch": 0.5224984943646219,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000973491496385253,
      "loss": 1.1244,
      "step": 6073
    },
    {
      "epoch": 0.5225845306719435,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009734828079454496,
      "loss": 1.9632,
      "step": 6074
    },
    {
      "epoch": 0.5226705669792653,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009734741181208011,
      "loss": 0.5675,
      "step": 6075
    },
    {
      "epoch": 0.5227566032865869,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009734654269113327,
      "loss": 1.5569,
      "step": 6076
    },
    {
      "epoch": 0.5228426395939086,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009734567343170703,
      "loss": 1.3442,
      "step": 6077
    },
    {
      "epoch": 0.5229286759012303,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009734480403380388,
      "loss": 1.4927,
      "step": 6078
    },
    {
      "epoch": 0.523014712208552,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000973439344974264,
      "loss": 1.3228,
      "step": 6079
    },
    {
      "epoch": 0.5231007485158737,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009734306482257711,
      "loss": 0.832,
      "step": 6080
    },
    {
      "epoch": 0.5231867848231954,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009734219500925858,
      "loss": 1.0622,
      "step": 6081
    },
    {
      "epoch": 0.5232728211305171,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009734132505747332,
      "loss": 1.8048,
      "step": 6082
    },
    {
      "epoch": 0.5233588574378387,
      "grad_norm": 1.34375,
      "learning_rate": 0.000973404549672239,
      "loss": 1.7966,
      "step": 6083
    },
    {
      "epoch": 0.5234448937451605,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009733958473851286,
      "loss": 1.3469,
      "step": 6084
    },
    {
      "epoch": 0.5235309300524822,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009733871437134272,
      "loss": 1.5221,
      "step": 6085
    },
    {
      "epoch": 0.5236169663598038,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009733784386571607,
      "loss": 0.9588,
      "step": 6086
    },
    {
      "epoch": 0.5237030026671256,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009733697322163541,
      "loss": 1.1202,
      "step": 6087
    },
    {
      "epoch": 0.5237890389744472,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009733610243910333,
      "loss": 1.4927,
      "step": 6088
    },
    {
      "epoch": 0.5238750752817689,
      "grad_norm": 0.875,
      "learning_rate": 0.0009733523151812236,
      "loss": 1.9201,
      "step": 6089
    },
    {
      "epoch": 0.5239611115890906,
      "grad_norm": 1.125,
      "learning_rate": 0.0009733436045869504,
      "loss": 1.8915,
      "step": 6090
    },
    {
      "epoch": 0.5240471478964123,
      "grad_norm": 1.140625,
      "learning_rate": 0.000973334892608239,
      "loss": 1.2724,
      "step": 6091
    },
    {
      "epoch": 0.524133184203734,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009733261792451153,
      "loss": 1.482,
      "step": 6092
    },
    {
      "epoch": 0.5242192205110556,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009733174644976046,
      "loss": 1.6087,
      "step": 6093
    },
    {
      "epoch": 0.5243052568183774,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009733087483657323,
      "loss": 1.2304,
      "step": 6094
    },
    {
      "epoch": 0.5243912931256991,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009733000308495241,
      "loss": 1.4274,
      "step": 6095
    },
    {
      "epoch": 0.5244773294330207,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009732913119490052,
      "loss": 0.9965,
      "step": 6096
    },
    {
      "epoch": 0.5245633657403425,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009732825916642014,
      "loss": 1.659,
      "step": 6097
    },
    {
      "epoch": 0.5246494020476641,
      "grad_norm": 1.578125,
      "learning_rate": 0.000973273869995138,
      "loss": 0.7388,
      "step": 6098
    },
    {
      "epoch": 0.5247354383549858,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009732651469418405,
      "loss": 1.2337,
      "step": 6099
    },
    {
      "epoch": 0.5248214746623074,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009732564225043345,
      "loss": 1.378,
      "step": 6100
    },
    {
      "epoch": 0.5249075109696292,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009732476966826456,
      "loss": 1.6191,
      "step": 6101
    },
    {
      "epoch": 0.5249935472769509,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009732389694767992,
      "loss": 2.3713,
      "step": 6102
    },
    {
      "epoch": 0.5250795835842725,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009732302408868207,
      "loss": 0.9843,
      "step": 6103
    },
    {
      "epoch": 0.5251656198915943,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009732215109127361,
      "loss": 2.0823,
      "step": 6104
    },
    {
      "epoch": 0.5252516561989159,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009732127795545704,
      "loss": 2.1984,
      "step": 6105
    },
    {
      "epoch": 0.5253376925062376,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009732040468123494,
      "loss": 1.9159,
      "step": 6106
    },
    {
      "epoch": 0.5254237288135594,
      "grad_norm": 1.25,
      "learning_rate": 0.0009731953126860986,
      "loss": 1.4793,
      "step": 6107
    },
    {
      "epoch": 0.525509765120881,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009731865771758435,
      "loss": 1.5298,
      "step": 6108
    },
    {
      "epoch": 0.5255958014282027,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009731778402816098,
      "loss": 1.1346,
      "step": 6109
    },
    {
      "epoch": 0.5256818377355243,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009731691020034228,
      "loss": 2.0729,
      "step": 6110
    },
    {
      "epoch": 0.5257678740428461,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009731603623413084,
      "loss": 2.1317,
      "step": 6111
    },
    {
      "epoch": 0.5258539103501678,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009731516212952918,
      "loss": 1.1026,
      "step": 6112
    },
    {
      "epoch": 0.5259399466574894,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009731428788653986,
      "loss": 1.3964,
      "step": 6113
    },
    {
      "epoch": 0.5260259829648112,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009731341350516547,
      "loss": 1.857,
      "step": 6114
    },
    {
      "epoch": 0.5261120192721328,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009731253898540855,
      "loss": 1.2047,
      "step": 6115
    },
    {
      "epoch": 0.5261980555794545,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009731166432727166,
      "loss": 1.4802,
      "step": 6116
    },
    {
      "epoch": 0.5262840918867763,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009731078953075734,
      "loss": 1.6211,
      "step": 6117
    },
    {
      "epoch": 0.5263701281940979,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009730991459586815,
      "loss": 1.2179,
      "step": 6118
    },
    {
      "epoch": 0.5264561645014196,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009730903952260667,
      "loss": 1.3844,
      "step": 6119
    },
    {
      "epoch": 0.5265422008087413,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009730816431097547,
      "loss": 1.456,
      "step": 6120
    },
    {
      "epoch": 0.526628237116063,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009730728896097708,
      "loss": 0.9325,
      "step": 6121
    },
    {
      "epoch": 0.5267142734233846,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009730641347261407,
      "loss": 2.1345,
      "step": 6122
    },
    {
      "epoch": 0.5268003097307064,
      "grad_norm": 1.265625,
      "learning_rate": 0.00097305537845889,
      "loss": 1.2761,
      "step": 6123
    },
    {
      "epoch": 0.5268863460380281,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009730466208080443,
      "loss": 2.0063,
      "step": 6124
    },
    {
      "epoch": 0.5269723823453497,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009730378617736293,
      "loss": 1.698,
      "step": 6125
    },
    {
      "epoch": 0.5270584186526714,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009730291013556704,
      "loss": 1.4126,
      "step": 6126
    },
    {
      "epoch": 0.5271444549599931,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009730203395541935,
      "loss": 1.4607,
      "step": 6127
    },
    {
      "epoch": 0.5272304912673148,
      "grad_norm": 1.5,
      "learning_rate": 0.0009730115763692241,
      "loss": 1.1427,
      "step": 6128
    },
    {
      "epoch": 0.5273165275746365,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009730028118007879,
      "loss": 1.1875,
      "step": 6129
    },
    {
      "epoch": 0.5274025638819582,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009729940458489104,
      "loss": 0.8317,
      "step": 6130
    },
    {
      "epoch": 0.5274886001892799,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009729852785136172,
      "loss": 1.5288,
      "step": 6131
    },
    {
      "epoch": 0.5275746364966015,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009729765097949341,
      "loss": 2.2157,
      "step": 6132
    },
    {
      "epoch": 0.5276606728039233,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009729677396928867,
      "loss": 1.2549,
      "step": 6133
    },
    {
      "epoch": 0.527746709111245,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009729589682075008,
      "loss": 1.3416,
      "step": 6134
    },
    {
      "epoch": 0.5278327454185666,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009729501953388018,
      "loss": 0.9106,
      "step": 6135
    },
    {
      "epoch": 0.5279187817258884,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009729414210868155,
      "loss": 1.7405,
      "step": 6136
    },
    {
      "epoch": 0.52800481803321,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009729326454515674,
      "loss": 1.2236,
      "step": 6137
    },
    {
      "epoch": 0.5280908543405317,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009729238684330835,
      "loss": 1.7976,
      "step": 6138
    },
    {
      "epoch": 0.5281768906478534,
      "grad_norm": 1.109375,
      "learning_rate": 0.000972915090031389,
      "loss": 1.5102,
      "step": 6139
    },
    {
      "epoch": 0.5282629269551751,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00097290631024651,
      "loss": 1.3103,
      "step": 6140
    },
    {
      "epoch": 0.5283489632624968,
      "grad_norm": 1.359375,
      "learning_rate": 0.000972897529078472,
      "loss": 1.7354,
      "step": 6141
    },
    {
      "epoch": 0.5284349995698184,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009728887465273007,
      "loss": 1.4287,
      "step": 6142
    },
    {
      "epoch": 0.5285210358771402,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009728799625930218,
      "loss": 0.9663,
      "step": 6143
    },
    {
      "epoch": 0.5286070721844618,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000972871177275661,
      "loss": 1.9334,
      "step": 6144
    },
    {
      "epoch": 0.5286931084917835,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009728623905752438,
      "loss": 1.3987,
      "step": 6145
    },
    {
      "epoch": 0.5287791447991053,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009728536024917963,
      "loss": 1.9361,
      "step": 6146
    },
    {
      "epoch": 0.5288651811064269,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009728448130253439,
      "loss": 1.5315,
      "step": 6147
    },
    {
      "epoch": 0.5289512174137486,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009728360221759124,
      "loss": 1.3737,
      "step": 6148
    },
    {
      "epoch": 0.5290372537210702,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009728272299435274,
      "loss": 1.2773,
      "step": 6149
    },
    {
      "epoch": 0.529123290028392,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009728184363282148,
      "loss": 2.0206,
      "step": 6150
    },
    {
      "epoch": 0.5292093263357137,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009728096413300003,
      "loss": 1.2828,
      "step": 6151
    },
    {
      "epoch": 0.5292953626430353,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009728008449489095,
      "loss": 1.1271,
      "step": 6152
    },
    {
      "epoch": 0.5293813989503571,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009727920471849681,
      "loss": 1.8622,
      "step": 6153
    },
    {
      "epoch": 0.5294674352576787,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009727832480382019,
      "loss": 0.7765,
      "step": 6154
    },
    {
      "epoch": 0.5295534715650004,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009727744475086368,
      "loss": 1.6348,
      "step": 6155
    },
    {
      "epoch": 0.5296395078723222,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009727656455962983,
      "loss": 0.8952,
      "step": 6156
    },
    {
      "epoch": 0.5297255441796438,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009727568423012124,
      "loss": 1.211,
      "step": 6157
    },
    {
      "epoch": 0.5298115804869655,
      "grad_norm": 1.25,
      "learning_rate": 0.0009727480376234043,
      "loss": 1.4603,
      "step": 6158
    },
    {
      "epoch": 0.5298976167942872,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009727392315629005,
      "loss": 1.4202,
      "step": 6159
    },
    {
      "epoch": 0.5299836531016089,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009727304241197262,
      "loss": 0.6965,
      "step": 6160
    },
    {
      "epoch": 0.5300696894089306,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009727216152939074,
      "loss": 0.9814,
      "step": 6161
    },
    {
      "epoch": 0.5301557257162522,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009727128050854699,
      "loss": 1.8244,
      "step": 6162
    },
    {
      "epoch": 0.530241762023574,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009727039934944393,
      "loss": 1.9827,
      "step": 6163
    },
    {
      "epoch": 0.5303277983308956,
      "grad_norm": 1.5,
      "learning_rate": 0.0009726951805208414,
      "loss": 1.3689,
      "step": 6164
    },
    {
      "epoch": 0.5304138346382173,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000972686366164702,
      "loss": 2.2797,
      "step": 6165
    },
    {
      "epoch": 0.530499870945539,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000972677550426047,
      "loss": 0.8663,
      "step": 6166
    },
    {
      "epoch": 0.5305859072528607,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009726687333049023,
      "loss": 1.2908,
      "step": 6167
    },
    {
      "epoch": 0.5306719435601824,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009726599148012932,
      "loss": 1.4006,
      "step": 6168
    },
    {
      "epoch": 0.5307579798675041,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009726510949152459,
      "loss": 0.822,
      "step": 6169
    },
    {
      "epoch": 0.5308440161748258,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009726422736467859,
      "loss": 1.3862,
      "step": 6170
    },
    {
      "epoch": 0.5309300524821474,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009726334509959394,
      "loss": 1.1117,
      "step": 6171
    },
    {
      "epoch": 0.5310160887894692,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009726246269627318,
      "loss": 1.2423,
      "step": 6172
    },
    {
      "epoch": 0.5311021250967909,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009726158015471892,
      "loss": 1.2065,
      "step": 6173
    },
    {
      "epoch": 0.5311881614041125,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009726069747493372,
      "loss": 1.0749,
      "step": 6174
    },
    {
      "epoch": 0.5312741977114342,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009725981465692017,
      "loss": 0.8851,
      "step": 6175
    },
    {
      "epoch": 0.5313602340187559,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009725893170068088,
      "loss": 1.624,
      "step": 6176
    },
    {
      "epoch": 0.5314462703260776,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009725804860621839,
      "loss": 1.1812,
      "step": 6177
    },
    {
      "epoch": 0.5315323066333993,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009725716537353528,
      "loss": 0.8308,
      "step": 6178
    },
    {
      "epoch": 0.531618342940721,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009725628200263418,
      "loss": 1.5623,
      "step": 6179
    },
    {
      "epoch": 0.5317043792480427,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009725539849351763,
      "loss": 1.2499,
      "step": 6180
    },
    {
      "epoch": 0.5317904155553643,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009725451484618823,
      "loss": 1.7998,
      "step": 6181
    },
    {
      "epoch": 0.5318764518626861,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009725363106064857,
      "loss": 1.3409,
      "step": 6182
    },
    {
      "epoch": 0.5319624881700078,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009725274713690124,
      "loss": 1.049,
      "step": 6183
    },
    {
      "epoch": 0.5320485244773294,
      "grad_norm": 1.65625,
      "learning_rate": 0.000972518630749488,
      "loss": 1.1186,
      "step": 6184
    },
    {
      "epoch": 0.5321345607846512,
      "grad_norm": 1.125,
      "learning_rate": 0.0009725097887479386,
      "loss": 1.4498,
      "step": 6185
    },
    {
      "epoch": 0.5322205970919728,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009725009453643898,
      "loss": 0.8552,
      "step": 6186
    },
    {
      "epoch": 0.5323066333992945,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009724921005988678,
      "loss": 0.7871,
      "step": 6187
    },
    {
      "epoch": 0.5323926697066161,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009724832544513983,
      "loss": 0.6405,
      "step": 6188
    },
    {
      "epoch": 0.5324787060139379,
      "grad_norm": 1.171875,
      "learning_rate": 0.000972474406922007,
      "loss": 0.8951,
      "step": 6189
    },
    {
      "epoch": 0.5325647423212596,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009724655580107203,
      "loss": 0.6768,
      "step": 6190
    },
    {
      "epoch": 0.5326507786285812,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009724567077175634,
      "loss": 1.139,
      "step": 6191
    },
    {
      "epoch": 0.532736814935903,
      "grad_norm": 1.0,
      "learning_rate": 0.0009724478560425626,
      "loss": 1.2209,
      "step": 6192
    },
    {
      "epoch": 0.5328228512432246,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009724390029857438,
      "loss": 1.2983,
      "step": 6193
    },
    {
      "epoch": 0.5329088875505463,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009724301485471326,
      "loss": 1.7242,
      "step": 6194
    },
    {
      "epoch": 0.5329949238578681,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009724212927267553,
      "loss": 1.6819,
      "step": 6195
    },
    {
      "epoch": 0.5330809601651897,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009724124355246375,
      "loss": 0.7843,
      "step": 6196
    },
    {
      "epoch": 0.5331669964725114,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009724035769408053,
      "loss": 2.0907,
      "step": 6197
    },
    {
      "epoch": 0.533253032779833,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009723947169752844,
      "loss": 1.5147,
      "step": 6198
    },
    {
      "epoch": 0.5333390690871548,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009723858556281009,
      "loss": 2.0394,
      "step": 6199
    },
    {
      "epoch": 0.5334251053944765,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009723769928992805,
      "loss": 1.9586,
      "step": 6200
    },
    {
      "epoch": 0.5335111417017981,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009723681287888494,
      "loss": 0.8122,
      "step": 6201
    },
    {
      "epoch": 0.5335971780091199,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009723592632968335,
      "loss": 1.9648,
      "step": 6202
    },
    {
      "epoch": 0.5336832143164415,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009723503964232584,
      "loss": 1.2314,
      "step": 6203
    },
    {
      "epoch": 0.5337692506237632,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009723415281681503,
      "loss": 1.6757,
      "step": 6204
    },
    {
      "epoch": 0.533855286931085,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009723326585315352,
      "loss": 1.8384,
      "step": 6205
    },
    {
      "epoch": 0.5339413232384066,
      "grad_norm": 1.796875,
      "learning_rate": 0.000972323787513439,
      "loss": 1.4619,
      "step": 6206
    },
    {
      "epoch": 0.5340273595457283,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009723149151138874,
      "loss": 1.4111,
      "step": 6207
    },
    {
      "epoch": 0.53411339585305,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009723060413329065,
      "loss": 0.9063,
      "step": 6208
    },
    {
      "epoch": 0.5341994321603717,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009722971661705224,
      "loss": 1.2105,
      "step": 6209
    },
    {
      "epoch": 0.5342854684676933,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009722882896267611,
      "loss": 1.6151,
      "step": 6210
    },
    {
      "epoch": 0.534371504775015,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009722794117016482,
      "loss": 1.0438,
      "step": 6211
    },
    {
      "epoch": 0.5344575410823368,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009722705323952099,
      "loss": 1.6511,
      "step": 6212
    },
    {
      "epoch": 0.5345435773896584,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009722616517074721,
      "loss": 1.4377,
      "step": 6213
    },
    {
      "epoch": 0.5346296136969801,
      "grad_norm": 1.234375,
      "learning_rate": 0.000972252769638461,
      "loss": 0.9441,
      "step": 6214
    },
    {
      "epoch": 0.5347156500043018,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009722438861882023,
      "loss": 1.1065,
      "step": 6215
    },
    {
      "epoch": 0.5348016863116235,
      "grad_norm": 0.984375,
      "learning_rate": 0.000972235001356722,
      "loss": 1.5199,
      "step": 6216
    },
    {
      "epoch": 0.5348877226189452,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0009722261151440463,
      "loss": 1.7812,
      "step": 6217
    },
    {
      "epoch": 0.5349737589262669,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000972217227550201,
      "loss": 1.4909,
      "step": 6218
    },
    {
      "epoch": 0.5350597952335886,
      "grad_norm": 2.625,
      "learning_rate": 0.000972208338575212,
      "loss": 1.4372,
      "step": 6219
    },
    {
      "epoch": 0.5351458315409102,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009721994482191057,
      "loss": 0.7106,
      "step": 6220
    },
    {
      "epoch": 0.535231867848232,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009721905564819077,
      "loss": 1.4819,
      "step": 6221
    },
    {
      "epoch": 0.5353179041555537,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009721816633636442,
      "loss": 1.6357,
      "step": 6222
    },
    {
      "epoch": 0.5354039404628753,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009721727688643411,
      "loss": 1.5275,
      "step": 6223
    },
    {
      "epoch": 0.535489976770197,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009721638729840245,
      "loss": 2.1502,
      "step": 6224
    },
    {
      "epoch": 0.5355760130775187,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009721549757227206,
      "loss": 1.6412,
      "step": 6225
    },
    {
      "epoch": 0.5356620493848404,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009721460770804551,
      "loss": 1.1835,
      "step": 6226
    },
    {
      "epoch": 0.5357480856921621,
      "grad_norm": 1.375,
      "learning_rate": 0.000972137177057254,
      "loss": 0.6403,
      "step": 6227
    },
    {
      "epoch": 0.5358341219994838,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009721282756531437,
      "loss": 1.1446,
      "step": 6228
    },
    {
      "epoch": 0.5359201583068055,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009721193728681499,
      "loss": 0.7591,
      "step": 6229
    },
    {
      "epoch": 0.5360061946141271,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009721104687022987,
      "loss": 1.2835,
      "step": 6230
    },
    {
      "epoch": 0.5360922309214489,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009721015631556164,
      "loss": 1.1089,
      "step": 6231
    },
    {
      "epoch": 0.5361782672287705,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009720926562281287,
      "loss": 2.0268,
      "step": 6232
    },
    {
      "epoch": 0.5362643035360922,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009720837479198619,
      "loss": 1.1361,
      "step": 6233
    },
    {
      "epoch": 0.536350339843414,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009720748382308418,
      "loss": 2.1064,
      "step": 6234
    },
    {
      "epoch": 0.5364363761507356,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009720659271610948,
      "loss": 0.9576,
      "step": 6235
    },
    {
      "epoch": 0.5365224124580573,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009720570147106466,
      "loss": 1.9494,
      "step": 6236
    },
    {
      "epoch": 0.5366084487653789,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009720481008795235,
      "loss": 1.6221,
      "step": 6237
    },
    {
      "epoch": 0.5366944850727007,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009720391856677515,
      "loss": 1.0556,
      "step": 6238
    },
    {
      "epoch": 0.5367805213800224,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009720302690753568,
      "loss": 1.4661,
      "step": 6239
    },
    {
      "epoch": 0.536866557687344,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009720213511023653,
      "loss": 1.1903,
      "step": 6240
    },
    {
      "epoch": 0.5369525939946658,
      "grad_norm": 1.546875,
      "learning_rate": 0.000972012431748803,
      "loss": 2.3246,
      "step": 6241
    },
    {
      "epoch": 0.5370386303019874,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009720035110146963,
      "loss": 1.6517,
      "step": 6242
    },
    {
      "epoch": 0.5371246666093091,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009719945889000711,
      "loss": 1.5701,
      "step": 6243
    },
    {
      "epoch": 0.5372107029166309,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009719856654049534,
      "loss": 2.1831,
      "step": 6244
    },
    {
      "epoch": 0.5372967392239525,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009719767405293696,
      "loss": 1.401,
      "step": 6245
    },
    {
      "epoch": 0.5373827755312742,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009719678142733457,
      "loss": 1.4974,
      "step": 6246
    },
    {
      "epoch": 0.5374688118385959,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009719588866369075,
      "loss": 1.412,
      "step": 6247
    },
    {
      "epoch": 0.5375548481459176,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009719499576200815,
      "loss": 1.4634,
      "step": 6248
    },
    {
      "epoch": 0.5376408844532393,
      "grad_norm": 1.75,
      "learning_rate": 0.0009719410272228937,
      "loss": 1.6892,
      "step": 6249
    },
    {
      "epoch": 0.537726920760561,
      "grad_norm": 1.15625,
      "learning_rate": 0.00097193209544537,
      "loss": 1.1947,
      "step": 6250
    },
    {
      "epoch": 0.5378129570678827,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009719231622875369,
      "loss": 0.9823,
      "step": 6251
    },
    {
      "epoch": 0.5378989933752043,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009719142277494202,
      "loss": 1.4319,
      "step": 6252
    },
    {
      "epoch": 0.537985029682526,
      "grad_norm": 1.25,
      "learning_rate": 0.0009719052918310462,
      "loss": 1.3853,
      "step": 6253
    },
    {
      "epoch": 0.5380710659898477,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000971896354532441,
      "loss": 1.9518,
      "step": 6254
    },
    {
      "epoch": 0.5381571022971694,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009718874158536309,
      "loss": 0.8122,
      "step": 6255
    },
    {
      "epoch": 0.5382431386044911,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009718784757946417,
      "loss": 1.936,
      "step": 6256
    },
    {
      "epoch": 0.5383291749118128,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009718695343554997,
      "loss": 0.8433,
      "step": 6257
    },
    {
      "epoch": 0.5384152112191345,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009718605915362312,
      "loss": 1.4178,
      "step": 6258
    },
    {
      "epoch": 0.5385012475264561,
      "grad_norm": 1.5,
      "learning_rate": 0.0009718516473368622,
      "loss": 0.8817,
      "step": 6259
    },
    {
      "epoch": 0.5385872838337779,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009718427017574189,
      "loss": 0.6945,
      "step": 6260
    },
    {
      "epoch": 0.5386733201410996,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009718337547979275,
      "loss": 1.1483,
      "step": 6261
    },
    {
      "epoch": 0.5387593564484212,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009718248064584141,
      "loss": 0.9254,
      "step": 6262
    },
    {
      "epoch": 0.538845392755743,
      "grad_norm": 1.5,
      "learning_rate": 0.0009718158567389049,
      "loss": 1.2758,
      "step": 6263
    },
    {
      "epoch": 0.5389314290630646,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000971806905639426,
      "loss": 1.3172,
      "step": 6264
    },
    {
      "epoch": 0.5390174653703863,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009717979531600039,
      "loss": 1.5509,
      "step": 6265
    },
    {
      "epoch": 0.539103501677708,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009717889993006644,
      "loss": 1.9609,
      "step": 6266
    },
    {
      "epoch": 0.5391895379850297,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009717800440614338,
      "loss": 2.1927,
      "step": 6267
    },
    {
      "epoch": 0.5392755742923514,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009717710874423383,
      "loss": 1.6638,
      "step": 6268
    },
    {
      "epoch": 0.539361610599673,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009717621294434043,
      "loss": 1.7332,
      "step": 6269
    },
    {
      "epoch": 0.5394476469069948,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009717531700646576,
      "loss": 0.8787,
      "step": 6270
    },
    {
      "epoch": 0.5395336832143165,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009717442093061247,
      "loss": 1.4077,
      "step": 6271
    },
    {
      "epoch": 0.5396197195216381,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009717352471678318,
      "loss": 1.029,
      "step": 6272
    },
    {
      "epoch": 0.5397057558289599,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000971726283649805,
      "loss": 1.0239,
      "step": 6273
    },
    {
      "epoch": 0.5397917921362815,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009717173187520707,
      "loss": 1.2357,
      "step": 6274
    },
    {
      "epoch": 0.5398778284436032,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009717083524746549,
      "loss": 1.5356,
      "step": 6275
    },
    {
      "epoch": 0.5399638647509248,
      "grad_norm": 2.921875,
      "learning_rate": 0.0009716993848175837,
      "loss": 1.6285,
      "step": 6276
    },
    {
      "epoch": 0.5400499010582466,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009716904157808838,
      "loss": 1.877,
      "step": 6277
    },
    {
      "epoch": 0.5401359373655683,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009716814453645811,
      "loss": 1.0208,
      "step": 6278
    },
    {
      "epoch": 0.5402219736728899,
      "grad_norm": 1.0625,
      "learning_rate": 0.000971672473568702,
      "loss": 0.6799,
      "step": 6279
    },
    {
      "epoch": 0.5403080099802117,
      "grad_norm": 1.25,
      "learning_rate": 0.0009716635003932726,
      "loss": 1.5202,
      "step": 6280
    },
    {
      "epoch": 0.5403940462875333,
      "grad_norm": 1.40625,
      "learning_rate": 0.000971654525838319,
      "loss": 1.6384,
      "step": 6281
    },
    {
      "epoch": 0.540480082594855,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009716455499038679,
      "loss": 1.3467,
      "step": 6282
    },
    {
      "epoch": 0.5405661189021768,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009716365725899452,
      "loss": 1.234,
      "step": 6283
    },
    {
      "epoch": 0.5406521552094984,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009716275938965772,
      "loss": 1.8343,
      "step": 6284
    },
    {
      "epoch": 0.5407381915168201,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009716186138237902,
      "loss": 1.2293,
      "step": 6285
    },
    {
      "epoch": 0.5408242278241417,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009716096323716106,
      "loss": 0.608,
      "step": 6286
    },
    {
      "epoch": 0.5409102641314635,
      "grad_norm": 1.5,
      "learning_rate": 0.0009716006495400644,
      "loss": 1.3741,
      "step": 6287
    },
    {
      "epoch": 0.5409963004387852,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009715916653291782,
      "loss": 1.0024,
      "step": 6288
    },
    {
      "epoch": 0.5410823367461068,
      "grad_norm": 1.5,
      "learning_rate": 0.0009715826797389779,
      "loss": 2.0479,
      "step": 6289
    },
    {
      "epoch": 0.5411683730534286,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009715736927694903,
      "loss": 1.482,
      "step": 6290
    },
    {
      "epoch": 0.5412544093607502,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009715647044207411,
      "loss": 1.0264,
      "step": 6291
    },
    {
      "epoch": 0.5413404456680719,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009715557146927569,
      "loss": 1.7568,
      "step": 6292
    },
    {
      "epoch": 0.5414264819753937,
      "grad_norm": 1.28125,
      "learning_rate": 0.000971546723585564,
      "loss": 1.5438,
      "step": 6293
    },
    {
      "epoch": 0.5415125182827153,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009715377310991885,
      "loss": 1.8307,
      "step": 6294
    },
    {
      "epoch": 0.541598554590037,
      "grad_norm": 1.203125,
      "learning_rate": 0.000971528737233657,
      "loss": 1.3353,
      "step": 6295
    },
    {
      "epoch": 0.5416845908973587,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009715197419889956,
      "loss": 1.5224,
      "step": 6296
    },
    {
      "epoch": 0.5417706272046804,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009715107453652307,
      "loss": 1.2451,
      "step": 6297
    },
    {
      "epoch": 0.541856663512002,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009715017473623884,
      "loss": 1.6146,
      "step": 6298
    },
    {
      "epoch": 0.5419426998193237,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009714927479804954,
      "loss": 0.8761,
      "step": 6299
    },
    {
      "epoch": 0.5420287361266455,
      "grad_norm": 1.25,
      "learning_rate": 0.0009714837472195776,
      "loss": 1.6691,
      "step": 6300
    },
    {
      "epoch": 0.5421147724339671,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009714747450796618,
      "loss": 1.2856,
      "step": 6301
    },
    {
      "epoch": 0.5422008087412888,
      "grad_norm": 1.125,
      "learning_rate": 0.000971465741560774,
      "loss": 1.8131,
      "step": 6302
    },
    {
      "epoch": 0.5422868450486105,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009714567366629405,
      "loss": 2.1242,
      "step": 6303
    },
    {
      "epoch": 0.5423728813559322,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009714477303861877,
      "loss": 2.1268,
      "step": 6304
    },
    {
      "epoch": 0.5424589176632539,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009714387227305421,
      "loss": 1.0476,
      "step": 6305
    },
    {
      "epoch": 0.5425449539705756,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009714297136960299,
      "loss": 1.1739,
      "step": 6306
    },
    {
      "epoch": 0.5426309902778973,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009714207032826774,
      "loss": 1.4399,
      "step": 6307
    },
    {
      "epoch": 0.5427170265852189,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009714116914905112,
      "loss": 1.233,
      "step": 6308
    },
    {
      "epoch": 0.5428030628925407,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009714026783195573,
      "loss": 1.943,
      "step": 6309
    },
    {
      "epoch": 0.5428890991998624,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009713936637698423,
      "loss": 1.1039,
      "step": 6310
    },
    {
      "epoch": 0.542975135507184,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009713846478413926,
      "loss": 1.7945,
      "step": 6311
    },
    {
      "epoch": 0.5430611718145057,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009713756305342344,
      "loss": 2.4143,
      "step": 6312
    },
    {
      "epoch": 0.5431472081218274,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009713666118483941,
      "loss": 0.8445,
      "step": 6313
    },
    {
      "epoch": 0.5432332444291491,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009713575917838984,
      "loss": 1.4197,
      "step": 6314
    },
    {
      "epoch": 0.5433192807364708,
      "grad_norm": 1.75,
      "learning_rate": 0.0009713485703407732,
      "loss": 1.3913,
      "step": 6315
    },
    {
      "epoch": 0.5434053170437925,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009713395475190451,
      "loss": 1.5009,
      "step": 6316
    },
    {
      "epoch": 0.5434913533511142,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009713305233187405,
      "loss": 0.9792,
      "step": 6317
    },
    {
      "epoch": 0.5435773896584358,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009713214977398859,
      "loss": 1.3022,
      "step": 6318
    },
    {
      "epoch": 0.5436634259657576,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009713124707825074,
      "loss": 0.9254,
      "step": 6319
    },
    {
      "epoch": 0.5437494622730792,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009713034424466317,
      "loss": 1.7441,
      "step": 6320
    },
    {
      "epoch": 0.5438354985804009,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009712944127322851,
      "loss": 0.9901,
      "step": 6321
    },
    {
      "epoch": 0.5439215348877227,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000971285381639494,
      "loss": 1.1462,
      "step": 6322
    },
    {
      "epoch": 0.5440075711950443,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009712763491682848,
      "loss": 1.2028,
      "step": 6323
    },
    {
      "epoch": 0.544093607502366,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000971267315318684,
      "loss": 1.7088,
      "step": 6324
    },
    {
      "epoch": 0.5441796438096876,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009712582800907179,
      "loss": 1.4233,
      "step": 6325
    },
    {
      "epoch": 0.5442656801170094,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009712492434844131,
      "loss": 1.5268,
      "step": 6326
    },
    {
      "epoch": 0.5443517164243311,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009712402054997957,
      "loss": 1.0701,
      "step": 6327
    },
    {
      "epoch": 0.5444377527316527,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009712311661368925,
      "loss": 1.2117,
      "step": 6328
    },
    {
      "epoch": 0.5445237890389745,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009712221253957297,
      "loss": 1.3445,
      "step": 6329
    },
    {
      "epoch": 0.5446098253462961,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009712130832763339,
      "loss": 1.4144,
      "step": 6330
    },
    {
      "epoch": 0.5446958616536178,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009712040397787314,
      "loss": 1.1099,
      "step": 6331
    },
    {
      "epoch": 0.5447818979609396,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009711949949029489,
      "loss": 1.1255,
      "step": 6332
    },
    {
      "epoch": 0.5448679342682612,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009711859486490125,
      "loss": 1.3854,
      "step": 6333
    },
    {
      "epoch": 0.5449539705755829,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009711769010169489,
      "loss": 1.0388,
      "step": 6334
    },
    {
      "epoch": 0.5450400068829045,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009711678520067846,
      "loss": 1.6741,
      "step": 6335
    },
    {
      "epoch": 0.5451260431902263,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009711588016185457,
      "loss": 1.5168,
      "step": 6336
    },
    {
      "epoch": 0.545212079497548,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009711497498522592,
      "loss": 1.0877,
      "step": 6337
    },
    {
      "epoch": 0.5452981158048696,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009711406967079512,
      "loss": 0.8735,
      "step": 6338
    },
    {
      "epoch": 0.5453841521121914,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009711316421856484,
      "loss": 1.6649,
      "step": 6339
    },
    {
      "epoch": 0.545470188419513,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009711225862853771,
      "loss": 1.3144,
      "step": 6340
    },
    {
      "epoch": 0.5455562247268347,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009711135290071638,
      "loss": 0.9184,
      "step": 6341
    },
    {
      "epoch": 0.5456422610341564,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009711044703510352,
      "loss": 1.889,
      "step": 6342
    },
    {
      "epoch": 0.5457282973414781,
      "grad_norm": 1.375,
      "learning_rate": 0.0009710954103170175,
      "loss": 1.0423,
      "step": 6343
    },
    {
      "epoch": 0.5458143336487998,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009710863489051374,
      "loss": 0.7644,
      "step": 6344
    },
    {
      "epoch": 0.5459003699561215,
      "grad_norm": 1.25,
      "learning_rate": 0.0009710772861154215,
      "loss": 0.9782,
      "step": 6345
    },
    {
      "epoch": 0.5459864062634432,
      "grad_norm": 1.21875,
      "learning_rate": 0.000971068221947896,
      "loss": 1.4549,
      "step": 6346
    },
    {
      "epoch": 0.5460724425707648,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009710591564025875,
      "loss": 1.9704,
      "step": 6347
    },
    {
      "epoch": 0.5461584788780866,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009710500894795227,
      "loss": 2.166,
      "step": 6348
    },
    {
      "epoch": 0.5462445151854083,
      "grad_norm": 1.9140625,
      "learning_rate": 0.000971041021178728,
      "loss": 1.7394,
      "step": 6349
    },
    {
      "epoch": 0.5463305514927299,
      "grad_norm": 1.484375,
      "learning_rate": 0.00097103195150023,
      "loss": 1.9793,
      "step": 6350
    },
    {
      "epoch": 0.5464165878000516,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009710228804440551,
      "loss": 2.0594,
      "step": 6351
    },
    {
      "epoch": 0.5465026241073733,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009710138080102298,
      "loss": 1.0734,
      "step": 6352
    },
    {
      "epoch": 0.546588660414695,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009710047341987807,
      "loss": 0.8843,
      "step": 6353
    },
    {
      "epoch": 0.5466746967220167,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009709956590097344,
      "loss": 1.6723,
      "step": 6354
    },
    {
      "epoch": 0.5467607330293384,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009709865824431175,
      "loss": 1.8001,
      "step": 6355
    },
    {
      "epoch": 0.5468467693366601,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009709775044989564,
      "loss": 1.7783,
      "step": 6356
    },
    {
      "epoch": 0.5469328056439817,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009709684251772777,
      "loss": 0.7599,
      "step": 6357
    },
    {
      "epoch": 0.5470188419513035,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009709593444781079,
      "loss": 0.897,
      "step": 6358
    },
    {
      "epoch": 0.5471048782586252,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009709502624014736,
      "loss": 1.5421,
      "step": 6359
    },
    {
      "epoch": 0.5471909145659468,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009709411789474014,
      "loss": 1.9477,
      "step": 6360
    },
    {
      "epoch": 0.5472769508732686,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009709320941159178,
      "loss": 1.6803,
      "step": 6361
    },
    {
      "epoch": 0.5473629871805902,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009709230079070495,
      "loss": 1.4661,
      "step": 6362
    },
    {
      "epoch": 0.5474490234879119,
      "grad_norm": 1.53125,
      "learning_rate": 0.000970913920320823,
      "loss": 1.7261,
      "step": 6363
    },
    {
      "epoch": 0.5475350597952335,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009709048313572648,
      "loss": 1.7313,
      "step": 6364
    },
    {
      "epoch": 0.5476210961025553,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009708957410164016,
      "loss": 1.6196,
      "step": 6365
    },
    {
      "epoch": 0.547707132409877,
      "grad_norm": 1.421875,
      "learning_rate": 0.00097088664929826,
      "loss": 1.5993,
      "step": 6366
    },
    {
      "epoch": 0.5477931687171986,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009708775562028664,
      "loss": 2.1125,
      "step": 6367
    },
    {
      "epoch": 0.5478792050245204,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009708684617302476,
      "loss": 1.3887,
      "step": 6368
    },
    {
      "epoch": 0.547965241331842,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009708593658804301,
      "loss": 1.1707,
      "step": 6369
    },
    {
      "epoch": 0.5480512776391637,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009708502686534406,
      "loss": 1.0302,
      "step": 6370
    },
    {
      "epoch": 0.5481373139464855,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009708411700493056,
      "loss": 1.8208,
      "step": 6371
    },
    {
      "epoch": 0.5482233502538071,
      "grad_norm": 2.84375,
      "learning_rate": 0.0009708320700680518,
      "loss": 1.8551,
      "step": 6372
    },
    {
      "epoch": 0.5483093865611288,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009708229687097056,
      "loss": 1.2725,
      "step": 6373
    },
    {
      "epoch": 0.5483954228684504,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000970813865974294,
      "loss": 0.7512,
      "step": 6374
    },
    {
      "epoch": 0.5484814591757722,
      "grad_norm": 6.34375,
      "learning_rate": 0.0009708047618618433,
      "loss": 1.2974,
      "step": 6375
    },
    {
      "epoch": 0.5485674954830939,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009707956563723803,
      "loss": 1.2937,
      "step": 6376
    },
    {
      "epoch": 0.5486535317904155,
      "grad_norm": 2.25,
      "learning_rate": 0.0009707865495059315,
      "loss": 1.7995,
      "step": 6377
    },
    {
      "epoch": 0.5487395680977373,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009707774412625236,
      "loss": 1.3094,
      "step": 6378
    },
    {
      "epoch": 0.5488256044050589,
      "grad_norm": 7.84375,
      "learning_rate": 0.0009707683316421832,
      "loss": 1.5241,
      "step": 6379
    },
    {
      "epoch": 0.5489116407123806,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009707592206449372,
      "loss": 1.4978,
      "step": 6380
    },
    {
      "epoch": 0.5489976770197024,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009707501082708118,
      "loss": 2.0294,
      "step": 6381
    },
    {
      "epoch": 0.549083713327024,
      "grad_norm": 1.140625,
      "learning_rate": 0.000970740994519834,
      "loss": 1.0847,
      "step": 6382
    },
    {
      "epoch": 0.5491697496343457,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009707318793920303,
      "loss": 1.3593,
      "step": 6383
    },
    {
      "epoch": 0.5492557859416674,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009707227628874275,
      "loss": 1.4198,
      "step": 6384
    },
    {
      "epoch": 0.5493418222489891,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000970713645006052,
      "loss": 0.6148,
      "step": 6385
    },
    {
      "epoch": 0.5494278585563107,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009707045257479307,
      "loss": 2.2727,
      "step": 6386
    },
    {
      "epoch": 0.5495138948636324,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009706954051130904,
      "loss": 1.9375,
      "step": 6387
    },
    {
      "epoch": 0.5495999311709542,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009706862831015575,
      "loss": 0.8528,
      "step": 6388
    },
    {
      "epoch": 0.5496859674782758,
      "grad_norm": 1.375,
      "learning_rate": 0.0009706771597133586,
      "loss": 1.4743,
      "step": 6389
    },
    {
      "epoch": 0.5497720037855975,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009706680349485206,
      "loss": 0.8598,
      "step": 6390
    },
    {
      "epoch": 0.5498580400929192,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009706589088070703,
      "loss": 1.6847,
      "step": 6391
    },
    {
      "epoch": 0.5499440764002409,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000970649781289034,
      "loss": 1.4481,
      "step": 6392
    },
    {
      "epoch": 0.5500301127075626,
      "grad_norm": 3.296875,
      "learning_rate": 0.0009706406523944389,
      "loss": 1.7701,
      "step": 6393
    },
    {
      "epoch": 0.5501161490148843,
      "grad_norm": 1.125,
      "learning_rate": 0.0009706315221233112,
      "loss": 1.6268,
      "step": 6394
    },
    {
      "epoch": 0.550202185322206,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009706223904756779,
      "loss": 1.6203,
      "step": 6395
    },
    {
      "epoch": 0.5502882216295276,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0009706132574515655,
      "loss": 1.4514,
      "step": 6396
    },
    {
      "epoch": 0.5503742579368494,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009706041230510012,
      "loss": 0.8789,
      "step": 6397
    },
    {
      "epoch": 0.5504602942441711,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009705949872740111,
      "loss": 1.0453,
      "step": 6398
    },
    {
      "epoch": 0.5505463305514927,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009705858501206222,
      "loss": 1.9031,
      "step": 6399
    },
    {
      "epoch": 0.5506323668588144,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009705767115908612,
      "loss": 0.751,
      "step": 6400
    },
    {
      "epoch": 0.5507184031661361,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009705675716847551,
      "loss": 1.1976,
      "step": 6401
    },
    {
      "epoch": 0.5508044394734578,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00097055843040233,
      "loss": 1.2687,
      "step": 6402
    },
    {
      "epoch": 0.5508904757807795,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009705492877436132,
      "loss": 1.0265,
      "step": 6403
    },
    {
      "epoch": 0.5509765120881012,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009705401437086314,
      "loss": 1.501,
      "step": 6404
    },
    {
      "epoch": 0.5510625483954229,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009705309982974108,
      "loss": 2.4224,
      "step": 6405
    },
    {
      "epoch": 0.5511485847027445,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000970521851509979,
      "loss": 1.5668,
      "step": 6406
    },
    {
      "epoch": 0.5512346210100663,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009705127033463619,
      "loss": 1.834,
      "step": 6407
    },
    {
      "epoch": 0.5513206573173879,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009705035538065869,
      "loss": 1.443,
      "step": 6408
    },
    {
      "epoch": 0.5514066936247096,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009704944028906804,
      "loss": 1.82,
      "step": 6409
    },
    {
      "epoch": 0.5514927299320314,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009704852505986693,
      "loss": 0.9327,
      "step": 6410
    },
    {
      "epoch": 0.551578766239353,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009704760969305804,
      "loss": 0.9194,
      "step": 6411
    },
    {
      "epoch": 0.5516648025466747,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009704669418864404,
      "loss": 1.1019,
      "step": 6412
    },
    {
      "epoch": 0.5517508388539963,
      "grad_norm": 1.25,
      "learning_rate": 0.000970457785466276,
      "loss": 1.3876,
      "step": 6413
    },
    {
      "epoch": 0.5518368751613181,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009704486276701141,
      "loss": 1.2328,
      "step": 6414
    },
    {
      "epoch": 0.5519229114686398,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009704394684979816,
      "loss": 0.9457,
      "step": 6415
    },
    {
      "epoch": 0.5520089477759614,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000970430307949905,
      "loss": 1.6652,
      "step": 6416
    },
    {
      "epoch": 0.5520949840832832,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009704211460259114,
      "loss": 1.8008,
      "step": 6417
    },
    {
      "epoch": 0.5521810203906048,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009704119827260273,
      "loss": 1.7155,
      "step": 6418
    },
    {
      "epoch": 0.5522670566979265,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009704028180502797,
      "loss": 1.1976,
      "step": 6419
    },
    {
      "epoch": 0.5523530930052483,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009703936519986953,
      "loss": 1.5922,
      "step": 6420
    },
    {
      "epoch": 0.5524391293125699,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000970384484571301,
      "loss": 1.652,
      "step": 6421
    },
    {
      "epoch": 0.5525251656198916,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009703753157681236,
      "loss": 1.7332,
      "step": 6422
    },
    {
      "epoch": 0.5526112019272132,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009703661455891896,
      "loss": 2.1155,
      "step": 6423
    },
    {
      "epoch": 0.552697238234535,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009703569740345264,
      "loss": 0.6381,
      "step": 6424
    },
    {
      "epoch": 0.5527832745418567,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009703478011041605,
      "loss": 1.0054,
      "step": 6425
    },
    {
      "epoch": 0.5528693108491783,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009703386267981188,
      "loss": 1.175,
      "step": 6426
    },
    {
      "epoch": 0.5529553471565001,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000970329451116428,
      "loss": 1.4182,
      "step": 6427
    },
    {
      "epoch": 0.5530413834638217,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000970320274059115,
      "loss": 2.1486,
      "step": 6428
    },
    {
      "epoch": 0.5531274197711434,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009703110956262068,
      "loss": 1.0343,
      "step": 6429
    },
    {
      "epoch": 0.5532134560784651,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009703019158177299,
      "loss": 1.5235,
      "step": 6430
    },
    {
      "epoch": 0.5532994923857868,
      "grad_norm": 2.125,
      "learning_rate": 0.0009702927346337116,
      "loss": 1.6406,
      "step": 6431
    },
    {
      "epoch": 0.5533855286931085,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009702835520741784,
      "loss": 2.0039,
      "step": 6432
    },
    {
      "epoch": 0.5534715650004302,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009702743681391572,
      "loss": 0.8995,
      "step": 6433
    },
    {
      "epoch": 0.5535576013077519,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009702651828286752,
      "loss": 0.924,
      "step": 6434
    },
    {
      "epoch": 0.5536436376150735,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009702559961427588,
      "loss": 1.5399,
      "step": 6435
    },
    {
      "epoch": 0.5537296739223952,
      "grad_norm": 1.453125,
      "learning_rate": 0.000970246808081435,
      "loss": 1.7728,
      "step": 6436
    },
    {
      "epoch": 0.553815710229717,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000970237618644731,
      "loss": 0.9161,
      "step": 6437
    },
    {
      "epoch": 0.5539017465370386,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009702284278326732,
      "loss": 1.575,
      "step": 6438
    },
    {
      "epoch": 0.5539877828443603,
      "grad_norm": 1.125,
      "learning_rate": 0.0009702192356452888,
      "loss": 1.1772,
      "step": 6439
    },
    {
      "epoch": 0.554073819151682,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009702100420826045,
      "loss": 1.7347,
      "step": 6440
    },
    {
      "epoch": 0.5541598554590037,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009702008471446475,
      "loss": 1.2146,
      "step": 6441
    },
    {
      "epoch": 0.5542458917663254,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009701916508314442,
      "loss": 1.5086,
      "step": 6442
    },
    {
      "epoch": 0.5543319280736471,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009701824531430219,
      "loss": 1.4812,
      "step": 6443
    },
    {
      "epoch": 0.5544179643809688,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009701732540794073,
      "loss": 2.1446,
      "step": 6444
    },
    {
      "epoch": 0.5545040006882904,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009701640536406276,
      "loss": 1.7165,
      "step": 6445
    },
    {
      "epoch": 0.5545900369956122,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009701548518267092,
      "loss": 1.9357,
      "step": 6446
    },
    {
      "epoch": 0.5546760733029339,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009701456486376794,
      "loss": 1.726,
      "step": 6447
    },
    {
      "epoch": 0.5547621096102555,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009701364440735649,
      "loss": 1.3147,
      "step": 6448
    },
    {
      "epoch": 0.5548481459175773,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000970127238134393,
      "loss": 1.4644,
      "step": 6449
    },
    {
      "epoch": 0.5549341822248989,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009701180308201902,
      "loss": 1.0181,
      "step": 6450
    },
    {
      "epoch": 0.5550202185322206,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009701088221309836,
      "loss": 1.0219,
      "step": 6451
    },
    {
      "epoch": 0.5551062548395422,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009700996120668002,
      "loss": 1.3933,
      "step": 6452
    },
    {
      "epoch": 0.555192291146864,
      "grad_norm": 1.375,
      "learning_rate": 0.0009700904006276667,
      "loss": 1.1931,
      "step": 6453
    },
    {
      "epoch": 0.5552783274541857,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009700811878136104,
      "loss": 1.8008,
      "step": 6454
    },
    {
      "epoch": 0.5553643637615073,
      "grad_norm": 1.15625,
      "learning_rate": 0.000970071973624658,
      "loss": 0.868,
      "step": 6455
    },
    {
      "epoch": 0.5554504000688291,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009700627580608364,
      "loss": 2.0136,
      "step": 6456
    },
    {
      "epoch": 0.5555364363761507,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009700535411221727,
      "loss": 1.9966,
      "step": 6457
    },
    {
      "epoch": 0.5556224726834724,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009700443228086939,
      "loss": 1.4347,
      "step": 6458
    },
    {
      "epoch": 0.5557085089907942,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009700351031204268,
      "loss": 1.9734,
      "step": 6459
    },
    {
      "epoch": 0.5557945452981158,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009700258820573984,
      "loss": 2.1656,
      "step": 6460
    },
    {
      "epoch": 0.5558805816054375,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009700166596196358,
      "loss": 1.4043,
      "step": 6461
    },
    {
      "epoch": 0.5559666179127591,
      "grad_norm": 1.25,
      "learning_rate": 0.0009700074358071658,
      "loss": 1.4598,
      "step": 6462
    },
    {
      "epoch": 0.5560526542200809,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009699982106200155,
      "loss": 1.8884,
      "step": 6463
    },
    {
      "epoch": 0.5561386905274026,
      "grad_norm": 1.0625,
      "learning_rate": 0.000969988984058212,
      "loss": 1.1445,
      "step": 6464
    },
    {
      "epoch": 0.5562247268347242,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009699797561217818,
      "loss": 0.8822,
      "step": 6465
    },
    {
      "epoch": 0.556310763142046,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009699705268107524,
      "loss": 1.9459,
      "step": 6466
    },
    {
      "epoch": 0.5563967994493676,
      "grad_norm": 1.75,
      "learning_rate": 0.0009699612961251506,
      "loss": 1.2649,
      "step": 6467
    },
    {
      "epoch": 0.5564828357566893,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009699520640650035,
      "loss": 1.0728,
      "step": 6468
    },
    {
      "epoch": 0.5565688720640111,
      "grad_norm": 1.546875,
      "learning_rate": 0.000969942830630338,
      "loss": 1.0593,
      "step": 6469
    },
    {
      "epoch": 0.5566549083713327,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000969933595821181,
      "loss": 1.161,
      "step": 6470
    },
    {
      "epoch": 0.5567409446786544,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009699243596375596,
      "loss": 1.8092,
      "step": 6471
    },
    {
      "epoch": 0.556826980985976,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009699151220795009,
      "loss": 1.0021,
      "step": 6472
    },
    {
      "epoch": 0.5569130172932978,
      "grad_norm": 1.390625,
      "learning_rate": 0.000969905883147032,
      "loss": 1.8419,
      "step": 6473
    },
    {
      "epoch": 0.5569990536006194,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009698966428401797,
      "loss": 1.1288,
      "step": 6474
    },
    {
      "epoch": 0.5570850899079411,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009698874011589708,
      "loss": 1.4898,
      "step": 6475
    },
    {
      "epoch": 0.5571711262152629,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009698781581034331,
      "loss": 1.5333,
      "step": 6476
    },
    {
      "epoch": 0.5572571625225845,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000969868913673593,
      "loss": 0.6619,
      "step": 6477
    },
    {
      "epoch": 0.5573431988299062,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009698596678694778,
      "loss": 1.789,
      "step": 6478
    },
    {
      "epoch": 0.5574292351372279,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009698504206911143,
      "loss": 2.1517,
      "step": 6479
    },
    {
      "epoch": 0.5575152714445496,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009698411721385297,
      "loss": 1.5734,
      "step": 6480
    },
    {
      "epoch": 0.5576013077518713,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009698319222117511,
      "loss": 0.9722,
      "step": 6481
    },
    {
      "epoch": 0.557687344059193,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009698226709108056,
      "loss": 0.989,
      "step": 6482
    },
    {
      "epoch": 0.5577733803665147,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00096981341823572,
      "loss": 1.4414,
      "step": 6483
    },
    {
      "epoch": 0.5578594166738363,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009698041641865216,
      "loss": 1.391,
      "step": 6484
    },
    {
      "epoch": 0.557945452981158,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009697949087632374,
      "loss": 1.1457,
      "step": 6485
    },
    {
      "epoch": 0.5580314892884798,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009697856519658945,
      "loss": 0.6899,
      "step": 6486
    },
    {
      "epoch": 0.5581175255958014,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009697763937945199,
      "loss": 0.9088,
      "step": 6487
    },
    {
      "epoch": 0.5582035619031231,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009697671342491406,
      "loss": 0.6906,
      "step": 6488
    },
    {
      "epoch": 0.5582895982104448,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009697578733297839,
      "loss": 1.3952,
      "step": 6489
    },
    {
      "epoch": 0.5583756345177665,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009697486110364768,
      "loss": 1.8136,
      "step": 6490
    },
    {
      "epoch": 0.5584616708250882,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009697393473692464,
      "loss": 1.0991,
      "step": 6491
    },
    {
      "epoch": 0.5585477071324099,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009697300823281196,
      "loss": 1.7685,
      "step": 6492
    },
    {
      "epoch": 0.5586337434397316,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009697208159131239,
      "loss": 1.4664,
      "step": 6493
    },
    {
      "epoch": 0.5587197797470532,
      "grad_norm": 0.9375,
      "learning_rate": 0.000969711548124286,
      "loss": 0.8709,
      "step": 6494
    },
    {
      "epoch": 0.558805816054375,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009697022789616333,
      "loss": 1.1446,
      "step": 6495
    },
    {
      "epoch": 0.5588918523616966,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009696930084251927,
      "loss": 1.2096,
      "step": 6496
    },
    {
      "epoch": 0.5589778886690183,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009696837365149915,
      "loss": 1.4139,
      "step": 6497
    },
    {
      "epoch": 0.55906392497634,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009696744632310566,
      "loss": 1.3477,
      "step": 6498
    },
    {
      "epoch": 0.5591499612836617,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009696651885734153,
      "loss": 1.6196,
      "step": 6499
    },
    {
      "epoch": 0.5592359975909834,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009696559125420948,
      "loss": 1.7057,
      "step": 6500
    },
    {
      "epoch": 0.559322033898305,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009696466351371218,
      "loss": 1.4226,
      "step": 6501
    },
    {
      "epoch": 0.5594080702056268,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009696373563585239,
      "loss": 1.4022,
      "step": 6502
    },
    {
      "epoch": 0.5594941065129485,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009696280762063281,
      "loss": 1.8594,
      "step": 6503
    },
    {
      "epoch": 0.5595801428202701,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009696187946805615,
      "loss": 1.0673,
      "step": 6504
    },
    {
      "epoch": 0.5596661791275919,
      "grad_norm": 1.5,
      "learning_rate": 0.0009696095117812511,
      "loss": 1.4383,
      "step": 6505
    },
    {
      "epoch": 0.5597522154349135,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009696002275084245,
      "loss": 1.0839,
      "step": 6506
    },
    {
      "epoch": 0.5598382517422352,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009695909418621084,
      "loss": 1.4602,
      "step": 6507
    },
    {
      "epoch": 0.559924288049557,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009695816548423301,
      "loss": 1.5213,
      "step": 6508
    },
    {
      "epoch": 0.5600103243568786,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009695723664491169,
      "loss": 1.1943,
      "step": 6509
    },
    {
      "epoch": 0.5600963606642003,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009695630766824956,
      "loss": 1.317,
      "step": 6510
    },
    {
      "epoch": 0.560182396971522,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009695537855424939,
      "loss": 1.7909,
      "step": 6511
    },
    {
      "epoch": 0.5602684332788437,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009695444930291386,
      "loss": 1.3492,
      "step": 6512
    },
    {
      "epoch": 0.5603544695861654,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009695351991424571,
      "loss": 1.4752,
      "step": 6513
    },
    {
      "epoch": 0.560440505893487,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009695259038824763,
      "loss": 1.1815,
      "step": 6514
    },
    {
      "epoch": 0.5605265422008088,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009695166072492237,
      "loss": 0.7608,
      "step": 6515
    },
    {
      "epoch": 0.5606125785081304,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009695073092427261,
      "loss": 0.969,
      "step": 6516
    },
    {
      "epoch": 0.5606986148154521,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009694980098630111,
      "loss": 1.4606,
      "step": 6517
    },
    {
      "epoch": 0.5607846511227738,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009694887091101057,
      "loss": 1.7978,
      "step": 6518
    },
    {
      "epoch": 0.5608706874300955,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009694794069840372,
      "loss": 1.4653,
      "step": 6519
    },
    {
      "epoch": 0.5609567237374172,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009694701034848325,
      "loss": 1.0566,
      "step": 6520
    },
    {
      "epoch": 0.5610427600447389,
      "grad_norm": 1.125,
      "learning_rate": 0.0009694607986125193,
      "loss": 1.1676,
      "step": 6521
    },
    {
      "epoch": 0.5611287963520606,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009694514923671246,
      "loss": 0.9827,
      "step": 6522
    },
    {
      "epoch": 0.5612148326593822,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009694421847486754,
      "loss": 1.0217,
      "step": 6523
    },
    {
      "epoch": 0.561300868966704,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009694328757571993,
      "loss": 1.0464,
      "step": 6524
    },
    {
      "epoch": 0.5613869052740257,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009694235653927233,
      "loss": 0.991,
      "step": 6525
    },
    {
      "epoch": 0.5614729415813473,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009694142536552746,
      "loss": 1.4043,
      "step": 6526
    },
    {
      "epoch": 0.561558977888669,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009694049405448805,
      "loss": 1.3346,
      "step": 6527
    },
    {
      "epoch": 0.5616450141959907,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009693956260615682,
      "loss": 0.9647,
      "step": 6528
    },
    {
      "epoch": 0.5617310505033124,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000969386310205365,
      "loss": 1.8044,
      "step": 6529
    },
    {
      "epoch": 0.5618170868106341,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009693769929762983,
      "loss": 0.7244,
      "step": 6530
    },
    {
      "epoch": 0.5619031231179558,
      "grad_norm": 3.4375,
      "learning_rate": 0.0009693676743743951,
      "loss": 1.5148,
      "step": 6531
    },
    {
      "epoch": 0.5619891594252775,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009693583543996827,
      "loss": 1.7867,
      "step": 6532
    },
    {
      "epoch": 0.5620751957325991,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009693490330521884,
      "loss": 2.17,
      "step": 6533
    },
    {
      "epoch": 0.5621612320399209,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009693397103319395,
      "loss": 1.288,
      "step": 6534
    },
    {
      "epoch": 0.5622472683472426,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000969330386238963,
      "loss": 1.1887,
      "step": 6535
    },
    {
      "epoch": 0.5623333046545642,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009693210607732868,
      "loss": 0.9789,
      "step": 6536
    },
    {
      "epoch": 0.562419340961886,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009693117339349375,
      "loss": 1.0001,
      "step": 6537
    },
    {
      "epoch": 0.5625053772692076,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009693024057239428,
      "loss": 1.1494,
      "step": 6538
    },
    {
      "epoch": 0.5625914135765293,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009692930761403297,
      "loss": 1.8393,
      "step": 6539
    },
    {
      "epoch": 0.5626774498838509,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009692837451841257,
      "loss": 0.7835,
      "step": 6540
    },
    {
      "epoch": 0.5627634861911727,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009692744128553579,
      "loss": 1.0417,
      "step": 6541
    },
    {
      "epoch": 0.5628495224984944,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009692650791540538,
      "loss": 1.0842,
      "step": 6542
    },
    {
      "epoch": 0.562935558805816,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009692557440802407,
      "loss": 1.3687,
      "step": 6543
    },
    {
      "epoch": 0.5630215951131378,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009692464076339458,
      "loss": 0.9186,
      "step": 6544
    },
    {
      "epoch": 0.5631076314204594,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009692370698151963,
      "loss": 1.263,
      "step": 6545
    },
    {
      "epoch": 0.5631936677277811,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009692277306240197,
      "loss": 2.3946,
      "step": 6546
    },
    {
      "epoch": 0.5632797040351029,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009692183900604432,
      "loss": 1.1374,
      "step": 6547
    },
    {
      "epoch": 0.5633657403424245,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009692090481244941,
      "loss": 1.4406,
      "step": 6548
    },
    {
      "epoch": 0.5634517766497462,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009691997048162,
      "loss": 1.7203,
      "step": 6549
    },
    {
      "epoch": 0.5635378129570678,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000969190360135588,
      "loss": 1.1794,
      "step": 6550
    },
    {
      "epoch": 0.5636238492643896,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009691810140826854,
      "loss": 1.0341,
      "step": 6551
    },
    {
      "epoch": 0.5637098855717113,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009691716666575195,
      "loss": 1.2972,
      "step": 6552
    },
    {
      "epoch": 0.5637959218790329,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009691623178601178,
      "loss": 1.795,
      "step": 6553
    },
    {
      "epoch": 0.5638819581863547,
      "grad_norm": 1.375,
      "learning_rate": 0.0009691529676905075,
      "loss": 1.64,
      "step": 6554
    },
    {
      "epoch": 0.5639679944936763,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000969143616148716,
      "loss": 1.3398,
      "step": 6555
    },
    {
      "epoch": 0.564054030800998,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009691342632347709,
      "loss": 1.2378,
      "step": 6556
    },
    {
      "epoch": 0.5641400671083198,
      "grad_norm": 1.375,
      "learning_rate": 0.000969124908948699,
      "loss": 1.4729,
      "step": 6557
    },
    {
      "epoch": 0.5642261034156414,
      "grad_norm": 1.375,
      "learning_rate": 0.0009691155532905282,
      "loss": 0.9569,
      "step": 6558
    },
    {
      "epoch": 0.5643121397229631,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009691061962602855,
      "loss": 0.8978,
      "step": 6559
    },
    {
      "epoch": 0.5643981760302847,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009690968378579985,
      "loss": 1.2709,
      "step": 6560
    },
    {
      "epoch": 0.5644842123376065,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009690874780836945,
      "loss": 1.6385,
      "step": 6561
    },
    {
      "epoch": 0.5645702486449281,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009690781169374008,
      "loss": 1.557,
      "step": 6562
    },
    {
      "epoch": 0.5646562849522498,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009690687544191448,
      "loss": 0.9019,
      "step": 6563
    },
    {
      "epoch": 0.5647423212595716,
      "grad_norm": 1.3125,
      "learning_rate": 0.000969059390528954,
      "loss": 1.0684,
      "step": 6564
    },
    {
      "epoch": 0.5648283575668932,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009690500252668557,
      "loss": 1.7792,
      "step": 6565
    },
    {
      "epoch": 0.5649143938742149,
      "grad_norm": 1.5,
      "learning_rate": 0.0009690406586328773,
      "loss": 1.531,
      "step": 6566
    },
    {
      "epoch": 0.5650004301815366,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009690312906270462,
      "loss": 1.1848,
      "step": 6567
    },
    {
      "epoch": 0.5650864664888583,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009690219212493898,
      "loss": 0.9037,
      "step": 6568
    },
    {
      "epoch": 0.56517250279618,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009690125504999355,
      "loss": 2.5674,
      "step": 6569
    },
    {
      "epoch": 0.5652585391035017,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009690031783787106,
      "loss": 1.4006,
      "step": 6570
    },
    {
      "epoch": 0.5653445754108234,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009689938048857427,
      "loss": 1.5102,
      "step": 6571
    },
    {
      "epoch": 0.565430611718145,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009689844300210593,
      "loss": 1.0921,
      "step": 6572
    },
    {
      "epoch": 0.5655166480254668,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009689750537846874,
      "loss": 1.7578,
      "step": 6573
    },
    {
      "epoch": 0.5656026843327885,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009689656761766547,
      "loss": 1.1286,
      "step": 6574
    },
    {
      "epoch": 0.5656887206401101,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009689562971969888,
      "loss": 1.5168,
      "step": 6575
    },
    {
      "epoch": 0.5657747569474318,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009689469168457168,
      "loss": 1.9466,
      "step": 6576
    },
    {
      "epoch": 0.5658607932547535,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009689375351228663,
      "loss": 0.974,
      "step": 6577
    },
    {
      "epoch": 0.5659468295620752,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009689281520284648,
      "loss": 1.1728,
      "step": 6578
    },
    {
      "epoch": 0.5660328658693969,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009689187675625395,
      "loss": 1.425,
      "step": 6579
    },
    {
      "epoch": 0.5661189021767186,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000968909381725118,
      "loss": 0.924,
      "step": 6580
    },
    {
      "epoch": 0.5662049384840403,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009688999945162279,
      "loss": 0.8263,
      "step": 6581
    },
    {
      "epoch": 0.5662909747913619,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009688906059358964,
      "loss": 1.5451,
      "step": 6582
    },
    {
      "epoch": 0.5663770110986837,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009688812159841511,
      "loss": 1.2095,
      "step": 6583
    },
    {
      "epoch": 0.5664630474060053,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009688718246610195,
      "loss": 1.6077,
      "step": 6584
    },
    {
      "epoch": 0.566549083713327,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009688624319665289,
      "loss": 1.0951,
      "step": 6585
    },
    {
      "epoch": 0.5666351200206488,
      "grad_norm": 1.25,
      "learning_rate": 0.000968853037900707,
      "loss": 1.4938,
      "step": 6586
    },
    {
      "epoch": 0.5667211563279704,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009688436424635812,
      "loss": 1.6488,
      "step": 6587
    },
    {
      "epoch": 0.5668071926352921,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009688342456551786,
      "loss": 1.1555,
      "step": 6588
    },
    {
      "epoch": 0.5668932289426137,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009688248474755272,
      "loss": 0.7314,
      "step": 6589
    },
    {
      "epoch": 0.5669792652499355,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009688154479246544,
      "loss": 1.0575,
      "step": 6590
    },
    {
      "epoch": 0.5670653015572572,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009688060470025876,
      "loss": 0.9389,
      "step": 6591
    },
    {
      "epoch": 0.5671513378645788,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009687966447093541,
      "loss": 1.7847,
      "step": 6592
    },
    {
      "epoch": 0.5672373741719006,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009687872410449819,
      "loss": 1.2405,
      "step": 6593
    },
    {
      "epoch": 0.5673234104792222,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009687778360094978,
      "loss": 1.2232,
      "step": 6594
    },
    {
      "epoch": 0.5674094467865439,
      "grad_norm": 1.5625,
      "learning_rate": 0.00096876842960293,
      "loss": 1.6266,
      "step": 6595
    },
    {
      "epoch": 0.5674954830938657,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009687590218253055,
      "loss": 1.2226,
      "step": 6596
    },
    {
      "epoch": 0.5675815194011873,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009687496126766522,
      "loss": 1.5185,
      "step": 6597
    },
    {
      "epoch": 0.567667555708509,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009687402021569975,
      "loss": 1.4875,
      "step": 6598
    },
    {
      "epoch": 0.5677535920158306,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009687307902663686,
      "loss": 1.7717,
      "step": 6599
    },
    {
      "epoch": 0.5678396283231524,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009687213770047936,
      "loss": 1.4187,
      "step": 6600
    },
    {
      "epoch": 0.5679256646304741,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009687119623722995,
      "loss": 0.9917,
      "step": 6601
    },
    {
      "epoch": 0.5680117009377957,
      "grad_norm": 1.375,
      "learning_rate": 0.0009687025463689142,
      "loss": 1.2892,
      "step": 6602
    },
    {
      "epoch": 0.5680977372451175,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009686931289946651,
      "loss": 1.7607,
      "step": 6603
    },
    {
      "epoch": 0.5681837735524391,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009686837102495798,
      "loss": 1.3822,
      "step": 6604
    },
    {
      "epoch": 0.5682698098597608,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009686742901336856,
      "loss": 1.481,
      "step": 6605
    },
    {
      "epoch": 0.5683558461670825,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009686648686470104,
      "loss": 0.9165,
      "step": 6606
    },
    {
      "epoch": 0.5684418824744042,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009686554457895817,
      "loss": 1.3415,
      "step": 6607
    },
    {
      "epoch": 0.5685279187817259,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009686460215614268,
      "loss": 1.6315,
      "step": 6608
    },
    {
      "epoch": 0.5686139550890476,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009686365959625736,
      "loss": 1.4965,
      "step": 6609
    },
    {
      "epoch": 0.5686999913963693,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009686271689930493,
      "loss": 1.8333,
      "step": 6610
    },
    {
      "epoch": 0.5687860277036909,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009686177406528818,
      "loss": 1.4009,
      "step": 6611
    },
    {
      "epoch": 0.5688720640110126,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009686083109420985,
      "loss": 0.9042,
      "step": 6612
    },
    {
      "epoch": 0.5689581003183344,
      "grad_norm": 3.015625,
      "learning_rate": 0.000968598879860727,
      "loss": 1.142,
      "step": 6613
    },
    {
      "epoch": 0.569044136625656,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009685894474087951,
      "loss": 1.1396,
      "step": 6614
    },
    {
      "epoch": 0.5691301729329777,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009685800135863301,
      "loss": 0.9015,
      "step": 6615
    },
    {
      "epoch": 0.5692162092402994,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009685705783933596,
      "loss": 1.0088,
      "step": 6616
    },
    {
      "epoch": 0.5693022455476211,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009685611418299114,
      "loss": 0.7666,
      "step": 6617
    },
    {
      "epoch": 0.5693882818549428,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009685517038960129,
      "loss": 1.8099,
      "step": 6618
    },
    {
      "epoch": 0.5694743181622645,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009685422645916918,
      "loss": 2.0704,
      "step": 6619
    },
    {
      "epoch": 0.5695603544695862,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009685328239169757,
      "loss": 1.3789,
      "step": 6620
    },
    {
      "epoch": 0.5696463907769078,
      "grad_norm": 1.75,
      "learning_rate": 0.0009685233818718924,
      "loss": 1.7994,
      "step": 6621
    },
    {
      "epoch": 0.5697324270842296,
      "grad_norm": 1.25,
      "learning_rate": 0.0009685139384564691,
      "loss": 1.3821,
      "step": 6622
    },
    {
      "epoch": 0.5698184633915513,
      "grad_norm": 1.25,
      "learning_rate": 0.0009685044936707337,
      "loss": 1.3621,
      "step": 6623
    },
    {
      "epoch": 0.5699044996988729,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009684950475147138,
      "loss": 0.8954,
      "step": 6624
    },
    {
      "epoch": 0.5699905360061946,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000968485599988437,
      "loss": 1.52,
      "step": 6625
    },
    {
      "epoch": 0.5700765723135163,
      "grad_norm": 1.234375,
      "learning_rate": 0.000968476151091931,
      "loss": 1.22,
      "step": 6626
    },
    {
      "epoch": 0.570162608620838,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009684667008252233,
      "loss": 1.6956,
      "step": 6627
    },
    {
      "epoch": 0.5702486449281596,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009684572491883417,
      "loss": 1.7606,
      "step": 6628
    },
    {
      "epoch": 0.5703346812354814,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009684477961813136,
      "loss": 1.5609,
      "step": 6629
    },
    {
      "epoch": 0.5704207175428031,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009684383418041669,
      "loss": 1.1193,
      "step": 6630
    },
    {
      "epoch": 0.5705067538501247,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009684288860569292,
      "loss": 2.4,
      "step": 6631
    },
    {
      "epoch": 0.5705927901574465,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000968419428939628,
      "loss": 1.186,
      "step": 6632
    },
    {
      "epoch": 0.5706788264647681,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009684099704522912,
      "loss": 1.2168,
      "step": 6633
    },
    {
      "epoch": 0.5707648627720898,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009684005105949462,
      "loss": 1.7715,
      "step": 6634
    },
    {
      "epoch": 0.5708508990794116,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009683910493676209,
      "loss": 1.7808,
      "step": 6635
    },
    {
      "epoch": 0.5709369353867332,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000968381586770343,
      "loss": 1.4079,
      "step": 6636
    },
    {
      "epoch": 0.5710229716940549,
      "grad_norm": 1.28125,
      "learning_rate": 0.00096837212280314,
      "loss": 1.6842,
      "step": 6637
    },
    {
      "epoch": 0.5711090080013765,
      "grad_norm": 1.125,
      "learning_rate": 0.0009683626574660394,
      "loss": 1.1386,
      "step": 6638
    },
    {
      "epoch": 0.5711950443086983,
      "grad_norm": 1.125,
      "learning_rate": 0.0009683531907590693,
      "loss": 1.4981,
      "step": 6639
    },
    {
      "epoch": 0.57128108061602,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009683437226822571,
      "loss": 1.1755,
      "step": 6640
    },
    {
      "epoch": 0.5713671169233416,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009683342532356308,
      "loss": 0.9539,
      "step": 6641
    },
    {
      "epoch": 0.5714531532306634,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009683247824192176,
      "loss": 2.2059,
      "step": 6642
    },
    {
      "epoch": 0.571539189537985,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009683153102330457,
      "loss": 1.4799,
      "step": 6643
    },
    {
      "epoch": 0.5716252258453067,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009683058366771425,
      "loss": 0.7987,
      "step": 6644
    },
    {
      "epoch": 0.5717112621526285,
      "grad_norm": 1.09375,
      "learning_rate": 0.000968296361751536,
      "loss": 1.2676,
      "step": 6645
    },
    {
      "epoch": 0.5717972984599501,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009682868854562535,
      "loss": 1.4038,
      "step": 6646
    },
    {
      "epoch": 0.5718833347672718,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000968277407791323,
      "loss": 0.9616,
      "step": 6647
    },
    {
      "epoch": 0.5719693710745934,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009682679287567722,
      "loss": 1.4932,
      "step": 6648
    },
    {
      "epoch": 0.5720554073819152,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009682584483526287,
      "loss": 0.9319,
      "step": 6649
    },
    {
      "epoch": 0.5721414436892368,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009682489665789204,
      "loss": 1.7734,
      "step": 6650
    },
    {
      "epoch": 0.5722274799965585,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009682394834356749,
      "loss": 2.1338,
      "step": 6651
    },
    {
      "epoch": 0.5723135163038803,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009682299989229201,
      "loss": 1.6486,
      "step": 6652
    },
    {
      "epoch": 0.5723995526112019,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009682205130406835,
      "loss": 1.3026,
      "step": 6653
    },
    {
      "epoch": 0.5724855889185236,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009682110257889929,
      "loss": 1.4427,
      "step": 6654
    },
    {
      "epoch": 0.5725716252258453,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009682015371678763,
      "loss": 1.3676,
      "step": 6655
    },
    {
      "epoch": 0.572657661533167,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009681920471773611,
      "loss": 1.0013,
      "step": 6656
    },
    {
      "epoch": 0.5727436978404887,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009681825558174753,
      "loss": 1.3165,
      "step": 6657
    },
    {
      "epoch": 0.5728297341478104,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009681730630882466,
      "loss": 1.4351,
      "step": 6658
    },
    {
      "epoch": 0.5729157704551321,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009681635689897025,
      "loss": 0.7262,
      "step": 6659
    },
    {
      "epoch": 0.5730018067624537,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009681540735218714,
      "loss": 1.4565,
      "step": 6660
    },
    {
      "epoch": 0.5730878430697754,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009681445766847805,
      "loss": 2.1065,
      "step": 6661
    },
    {
      "epoch": 0.5731738793770972,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009681350784784577,
      "loss": 1.11,
      "step": 6662
    },
    {
      "epoch": 0.5732599156844188,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009681255789029308,
      "loss": 1.1633,
      "step": 6663
    },
    {
      "epoch": 0.5733459519917405,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009681160779582279,
      "loss": 1.9728,
      "step": 6664
    },
    {
      "epoch": 0.5734319882990622,
      "grad_norm": 1.375,
      "learning_rate": 0.0009681065756443763,
      "loss": 1.397,
      "step": 6665
    },
    {
      "epoch": 0.5735180246063839,
      "grad_norm": 1.59375,
      "learning_rate": 0.000968097071961404,
      "loss": 1.8126,
      "step": 6666
    },
    {
      "epoch": 0.5736040609137056,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000968087566909339,
      "loss": 1.0379,
      "step": 6667
    },
    {
      "epoch": 0.5736900972210273,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009680780604882087,
      "loss": 1.858,
      "step": 6668
    },
    {
      "epoch": 0.573776133528349,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009680685526980411,
      "loss": 1.543,
      "step": 6669
    },
    {
      "epoch": 0.5738621698356706,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009680590435388641,
      "loss": 1.4573,
      "step": 6670
    },
    {
      "epoch": 0.5739482061429924,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009680495330107055,
      "loss": 2.2396,
      "step": 6671
    },
    {
      "epoch": 0.574034242450314,
      "grad_norm": 2.3125,
      "learning_rate": 0.000968040021113593,
      "loss": 1.4247,
      "step": 6672
    },
    {
      "epoch": 0.5741202787576357,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009680305078475543,
      "loss": 1.3833,
      "step": 6673
    },
    {
      "epoch": 0.5742063150649575,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009680209932126176,
      "loss": 1.0324,
      "step": 6674
    },
    {
      "epoch": 0.5742923513722791,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009680114772088104,
      "loss": 1.7406,
      "step": 6675
    },
    {
      "epoch": 0.5743783876796008,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009680019598361607,
      "loss": 1.1428,
      "step": 6676
    },
    {
      "epoch": 0.5744644239869224,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009679924410946963,
      "loss": 1.7352,
      "step": 6677
    },
    {
      "epoch": 0.5745504602942442,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009679829209844451,
      "loss": 1.6793,
      "step": 6678
    },
    {
      "epoch": 0.5746364966015659,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009679733995054347,
      "loss": 1.7995,
      "step": 6679
    },
    {
      "epoch": 0.5747225329088875,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009679638766576932,
      "loss": 1.4468,
      "step": 6680
    },
    {
      "epoch": 0.5748085692162093,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009679543524412484,
      "loss": 1.6003,
      "step": 6681
    },
    {
      "epoch": 0.5748946055235309,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009679448268561282,
      "loss": 2.0135,
      "step": 6682
    },
    {
      "epoch": 0.5749806418308526,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009679352999023603,
      "loss": 1.0272,
      "step": 6683
    },
    {
      "epoch": 0.5750666781381744,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009679257715799726,
      "loss": 2.4294,
      "step": 6684
    },
    {
      "epoch": 0.575152714445496,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009679162418889931,
      "loss": 0.9783,
      "step": 6685
    },
    {
      "epoch": 0.5752387507528177,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009679067108294495,
      "loss": 1.9568,
      "step": 6686
    },
    {
      "epoch": 0.5753247870601393,
      "grad_norm": 1.28125,
      "learning_rate": 0.00096789717840137,
      "loss": 1.1539,
      "step": 6687
    },
    {
      "epoch": 0.5754108233674611,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009678876446047819,
      "loss": 2.0016,
      "step": 6688
    },
    {
      "epoch": 0.5754968596747828,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009678781094397137,
      "loss": 2.2646,
      "step": 6689
    },
    {
      "epoch": 0.5755828959821044,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009678685729061932,
      "loss": 1.1025,
      "step": 6690
    },
    {
      "epoch": 0.5756689322894262,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009678590350042477,
      "loss": 1.151,
      "step": 6691
    },
    {
      "epoch": 0.5757549685967478,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009678494957339058,
      "loss": 1.7239,
      "step": 6692
    },
    {
      "epoch": 0.5758410049040695,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009678399550951951,
      "loss": 1.3805,
      "step": 6693
    },
    {
      "epoch": 0.5759270412113912,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009678304130881434,
      "loss": 1.3565,
      "step": 6694
    },
    {
      "epoch": 0.5760130775187129,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009678208697127787,
      "loss": 0.8831,
      "step": 6695
    },
    {
      "epoch": 0.5760991138260346,
      "grad_norm": 1.484375,
      "learning_rate": 0.000967811324969129,
      "loss": 1.7847,
      "step": 6696
    },
    {
      "epoch": 0.5761851501333562,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009678017788572222,
      "loss": 1.2952,
      "step": 6697
    },
    {
      "epoch": 0.576271186440678,
      "grad_norm": 1.125,
      "learning_rate": 0.0009677922313770862,
      "loss": 0.9473,
      "step": 6698
    },
    {
      "epoch": 0.5763572227479996,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009677826825287488,
      "loss": 1.6284,
      "step": 6699
    },
    {
      "epoch": 0.5764432590553213,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009677731323122382,
      "loss": 1.2379,
      "step": 6700
    },
    {
      "epoch": 0.5765292953626431,
      "grad_norm": 1.421875,
      "learning_rate": 0.000967763580727582,
      "loss": 1.8351,
      "step": 6701
    },
    {
      "epoch": 0.5766153316699647,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009677540277748085,
      "loss": 1.5066,
      "step": 6702
    },
    {
      "epoch": 0.5767013679772864,
      "grad_norm": 1.25,
      "learning_rate": 0.0009677444734539452,
      "loss": 0.8086,
      "step": 6703
    },
    {
      "epoch": 0.5767874042846081,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009677349177650203,
      "loss": 1.4412,
      "step": 6704
    },
    {
      "epoch": 0.5768734405919298,
      "grad_norm": 1.234375,
      "learning_rate": 0.000967725360708062,
      "loss": 1.5324,
      "step": 6705
    },
    {
      "epoch": 0.5769594768992515,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009677158022830978,
      "loss": 1.0142,
      "step": 6706
    },
    {
      "epoch": 0.5770455132065732,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009677062424901558,
      "loss": 1.2486,
      "step": 6707
    },
    {
      "epoch": 0.5771315495138949,
      "grad_norm": 1.109375,
      "learning_rate": 0.000967696681329264,
      "loss": 1.1532,
      "step": 6708
    },
    {
      "epoch": 0.5772175858212165,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009676871188004505,
      "loss": 2.2251,
      "step": 6709
    },
    {
      "epoch": 0.5773036221285383,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009676775549037431,
      "loss": 1.0851,
      "step": 6710
    },
    {
      "epoch": 0.57738965843586,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009676679896391698,
      "loss": 1.1988,
      "step": 6711
    },
    {
      "epoch": 0.5774756947431816,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009676584230067586,
      "loss": 0.8002,
      "step": 6712
    },
    {
      "epoch": 0.5775617310505033,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009676488550065375,
      "loss": 1.7039,
      "step": 6713
    },
    {
      "epoch": 0.577647767357825,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009676392856385343,
      "loss": 1.7302,
      "step": 6714
    },
    {
      "epoch": 0.5777338036651467,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009676297149027772,
      "loss": 0.7993,
      "step": 6715
    },
    {
      "epoch": 0.5778198399724683,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009676201427992941,
      "loss": 1.8632,
      "step": 6716
    },
    {
      "epoch": 0.5779058762797901,
      "grad_norm": 1.25,
      "learning_rate": 0.0009676105693281132,
      "loss": 1.6982,
      "step": 6717
    },
    {
      "epoch": 0.5779919125871118,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009676009944892621,
      "loss": 1.6313,
      "step": 6718
    },
    {
      "epoch": 0.5780779488944334,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009675914182827692,
      "loss": 1.1712,
      "step": 6719
    },
    {
      "epoch": 0.5781639852017552,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009675818407086623,
      "loss": 1.7462,
      "step": 6720
    },
    {
      "epoch": 0.5782500215090768,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009675722617669694,
      "loss": 1.3478,
      "step": 6721
    },
    {
      "epoch": 0.5783360578163985,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009675626814577187,
      "loss": 1.2344,
      "step": 6722
    },
    {
      "epoch": 0.5784220941237203,
      "grad_norm": 1.234375,
      "learning_rate": 0.000967553099780938,
      "loss": 0.8848,
      "step": 6723
    },
    {
      "epoch": 0.5785081304310419,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009675435167366554,
      "loss": 1.1049,
      "step": 6724
    },
    {
      "epoch": 0.5785941667383636,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000967533932324899,
      "loss": 1.8953,
      "step": 6725
    },
    {
      "epoch": 0.5786802030456852,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009675243465456967,
      "loss": 1.4229,
      "step": 6726
    },
    {
      "epoch": 0.578766239353007,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009675147593990768,
      "loss": 1.6201,
      "step": 6727
    },
    {
      "epoch": 0.5788522756603287,
      "grad_norm": 1.7734375,
      "learning_rate": 0.000967505170885067,
      "loss": 1.8389,
      "step": 6728
    },
    {
      "epoch": 0.5789383119676503,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009674955810036955,
      "loss": 0.8434,
      "step": 6729
    },
    {
      "epoch": 0.5790243482749721,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009674859897549903,
      "loss": 0.8686,
      "step": 6730
    },
    {
      "epoch": 0.5791103845822937,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009674763971389797,
      "loss": 0.7421,
      "step": 6731
    },
    {
      "epoch": 0.5791964208896154,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009674668031556914,
      "loss": 1.595,
      "step": 6732
    },
    {
      "epoch": 0.5792824571969372,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009674572078051535,
      "loss": 1.1394,
      "step": 6733
    },
    {
      "epoch": 0.5793684935042588,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009674476110873944,
      "loss": 1.1514,
      "step": 6734
    },
    {
      "epoch": 0.5794545298115805,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009674380130024419,
      "loss": 0.9015,
      "step": 6735
    },
    {
      "epoch": 0.5795405661189021,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009674284135503242,
      "loss": 1.3206,
      "step": 6736
    },
    {
      "epoch": 0.5796266024262239,
      "grad_norm": 1.5,
      "learning_rate": 0.0009674188127310692,
      "loss": 2.1027,
      "step": 6737
    },
    {
      "epoch": 0.5797126387335455,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009674092105447051,
      "loss": 1.4677,
      "step": 6738
    },
    {
      "epoch": 0.5797986750408672,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00096739960699126,
      "loss": 1.0604,
      "step": 6739
    },
    {
      "epoch": 0.579884711348189,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009673900020707618,
      "loss": 1.8937,
      "step": 6740
    },
    {
      "epoch": 0.5799707476555106,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009673803957832389,
      "loss": 1.5104,
      "step": 6741
    },
    {
      "epoch": 0.5800567839628323,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009673707881287192,
      "loss": 0.865,
      "step": 6742
    },
    {
      "epoch": 0.580142820270154,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009673611791072309,
      "loss": 0.8915,
      "step": 6743
    },
    {
      "epoch": 0.5802288565774757,
      "grad_norm": 1.296875,
      "learning_rate": 0.000967351568718802,
      "loss": 1.6129,
      "step": 6744
    },
    {
      "epoch": 0.5803148928847974,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009673419569634605,
      "loss": 0.7505,
      "step": 6745
    },
    {
      "epoch": 0.580400929192119,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009673323438412349,
      "loss": 2.2601,
      "step": 6746
    },
    {
      "epoch": 0.5804869654994408,
      "grad_norm": 2.0,
      "learning_rate": 0.000967322729352153,
      "loss": 1.7396,
      "step": 6747
    },
    {
      "epoch": 0.5805730018067624,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009673131134962429,
      "loss": 1.1382,
      "step": 6748
    },
    {
      "epoch": 0.5806590381140841,
      "grad_norm": 1.15625,
      "learning_rate": 0.000967303496273533,
      "loss": 0.9413,
      "step": 6749
    },
    {
      "epoch": 0.5807450744214059,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009672938776840512,
      "loss": 1.5526,
      "step": 6750
    },
    {
      "epoch": 0.5808311107287275,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009672842577278258,
      "loss": 0.992,
      "step": 6751
    },
    {
      "epoch": 0.5809171470360492,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009672746364048847,
      "loss": 1.5085,
      "step": 6752
    },
    {
      "epoch": 0.5810031833433709,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009672650137152561,
      "loss": 1.3494,
      "step": 6753
    },
    {
      "epoch": 0.5810892196506926,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009672553896589683,
      "loss": 1.6767,
      "step": 6754
    },
    {
      "epoch": 0.5811752559580143,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009672457642360494,
      "loss": 1.994,
      "step": 6755
    },
    {
      "epoch": 0.581261292265336,
      "grad_norm": 1.5,
      "learning_rate": 0.0009672361374465275,
      "loss": 1.457,
      "step": 6756
    },
    {
      "epoch": 0.5813473285726577,
      "grad_norm": 1.75,
      "learning_rate": 0.0009672265092904306,
      "loss": 1.1639,
      "step": 6757
    },
    {
      "epoch": 0.5814333648799793,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009672168797677873,
      "loss": 1.5666,
      "step": 6758
    },
    {
      "epoch": 0.581519401187301,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009672072488786254,
      "loss": 0.9406,
      "step": 6759
    },
    {
      "epoch": 0.5816054374946227,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000967197616622973,
      "loss": 1.8692,
      "step": 6760
    },
    {
      "epoch": 0.5816914738019444,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009671879830008587,
      "loss": 1.438,
      "step": 6761
    },
    {
      "epoch": 0.5817775101092661,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009671783480123102,
      "loss": 1.7044,
      "step": 6762
    },
    {
      "epoch": 0.5818635464165878,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000967168711657356,
      "loss": 1.8058,
      "step": 6763
    },
    {
      "epoch": 0.5819495827239095,
      "grad_norm": 3.6875,
      "learning_rate": 0.0009671590739360241,
      "loss": 1.4502,
      "step": 6764
    },
    {
      "epoch": 0.5820356190312311,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009671494348483429,
      "loss": 1.5507,
      "step": 6765
    },
    {
      "epoch": 0.5821216553385529,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009671397943943403,
      "loss": 0.9253,
      "step": 6766
    },
    {
      "epoch": 0.5822076916458746,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009671301525740448,
      "loss": 1.9181,
      "step": 6767
    },
    {
      "epoch": 0.5822937279531962,
      "grad_norm": 1.5,
      "learning_rate": 0.0009671205093874844,
      "loss": 1.5548,
      "step": 6768
    },
    {
      "epoch": 0.582379764260518,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009671108648346873,
      "loss": 1.3378,
      "step": 6769
    },
    {
      "epoch": 0.5824658005678396,
      "grad_norm": 3.25,
      "learning_rate": 0.0009671012189156819,
      "loss": 1.8368,
      "step": 6770
    },
    {
      "epoch": 0.5825518368751613,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009670915716304962,
      "loss": 1.1973,
      "step": 6771
    },
    {
      "epoch": 0.5826378731824831,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009670819229791584,
      "loss": 1.5545,
      "step": 6772
    },
    {
      "epoch": 0.5827239094898047,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000967072272961697,
      "loss": 1.0265,
      "step": 6773
    },
    {
      "epoch": 0.5828099457971264,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00096706262157814,
      "loss": 1.4238,
      "step": 6774
    },
    {
      "epoch": 0.582895982104448,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009670529688285155,
      "loss": 1.3767,
      "step": 6775
    },
    {
      "epoch": 0.5829820184117698,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009670433147128521,
      "loss": 1.1653,
      "step": 6776
    },
    {
      "epoch": 0.5830680547190915,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009670336592311777,
      "loss": 1.6864,
      "step": 6777
    },
    {
      "epoch": 0.5831540910264131,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009670240023835208,
      "loss": 2.3624,
      "step": 6778
    },
    {
      "epoch": 0.5832401273337349,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009670143441699095,
      "loss": 1.3833,
      "step": 6779
    },
    {
      "epoch": 0.5833261636410565,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000967004684590372,
      "loss": 0.9907,
      "step": 6780
    },
    {
      "epoch": 0.5834121999483782,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009669950236449366,
      "loss": 1.4253,
      "step": 6781
    },
    {
      "epoch": 0.5834982362556999,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009669853613336317,
      "loss": 2.0442,
      "step": 6782
    },
    {
      "epoch": 0.5835842725630216,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009669756976564854,
      "loss": 1.8714,
      "step": 6783
    },
    {
      "epoch": 0.5836703088703433,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000966966032613526,
      "loss": 0.7699,
      "step": 6784
    },
    {
      "epoch": 0.583756345177665,
      "grad_norm": 2.390625,
      "learning_rate": 0.0009669563662047818,
      "loss": 0.9617,
      "step": 6785
    },
    {
      "epoch": 0.5838423814849867,
      "grad_norm": 1.21875,
      "learning_rate": 0.000966946698430281,
      "loss": 0.6826,
      "step": 6786
    },
    {
      "epoch": 0.5839284177923083,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009669370292900519,
      "loss": 0.7757,
      "step": 6787
    },
    {
      "epoch": 0.58401445409963,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009669273587841229,
      "loss": 1.5357,
      "step": 6788
    },
    {
      "epoch": 0.5841004904069518,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009669176869125221,
      "loss": 1.3004,
      "step": 6789
    },
    {
      "epoch": 0.5841865267142734,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000966908013675278,
      "loss": 1.7501,
      "step": 6790
    },
    {
      "epoch": 0.5842725630215951,
      "grad_norm": 1.25,
      "learning_rate": 0.0009668983390724187,
      "loss": 0.6825,
      "step": 6791
    },
    {
      "epoch": 0.5843585993289168,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009668886631039724,
      "loss": 0.8027,
      "step": 6792
    },
    {
      "epoch": 0.5844446356362385,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009668789857699678,
      "loss": 1.2159,
      "step": 6793
    },
    {
      "epoch": 0.5845306719435602,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009668693070704329,
      "loss": 1.1508,
      "step": 6794
    },
    {
      "epoch": 0.5846167082508819,
      "grad_norm": 1.5,
      "learning_rate": 0.000966859627005396,
      "loss": 1.0556,
      "step": 6795
    },
    {
      "epoch": 0.5847027445582036,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009668499455748856,
      "loss": 2.3709,
      "step": 6796
    },
    {
      "epoch": 0.5847887808655252,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009668402627789298,
      "loss": 0.9218,
      "step": 6797
    },
    {
      "epoch": 0.584874817172847,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009668305786175571,
      "loss": 1.0041,
      "step": 6798
    },
    {
      "epoch": 0.5849608534801687,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009668208930907957,
      "loss": 1.7154,
      "step": 6799
    },
    {
      "epoch": 0.5850468897874903,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009668112061986741,
      "loss": 0.9809,
      "step": 6800
    },
    {
      "epoch": 0.585132926094812,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009668015179412203,
      "loss": 1.7634,
      "step": 6801
    },
    {
      "epoch": 0.5852189624021337,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009667918283184629,
      "loss": 0.7741,
      "step": 6802
    },
    {
      "epoch": 0.5853049987094554,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009667821373304305,
      "loss": 1.4514,
      "step": 6803
    },
    {
      "epoch": 0.585391035016777,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009667724449771507,
      "loss": 1.1217,
      "step": 6804
    },
    {
      "epoch": 0.5854770713240988,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009667627512586525,
      "loss": 1.2726,
      "step": 6805
    },
    {
      "epoch": 0.5855631076314205,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000966753056174964,
      "loss": 1.3697,
      "step": 6806
    },
    {
      "epoch": 0.5856491439387421,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009667433597261136,
      "loss": 1.5573,
      "step": 6807
    },
    {
      "epoch": 0.5857351802460639,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009667336619121296,
      "loss": 0.8815,
      "step": 6808
    },
    {
      "epoch": 0.5858212165533855,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009667239627330404,
      "loss": 1.3069,
      "step": 6809
    },
    {
      "epoch": 0.5859072528607072,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009667142621888745,
      "loss": 1.0308,
      "step": 6810
    },
    {
      "epoch": 0.585993289168029,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009667045602796601,
      "loss": 1.6815,
      "step": 6811
    },
    {
      "epoch": 0.5860793254753506,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009666948570054256,
      "loss": 1.5697,
      "step": 6812
    },
    {
      "epoch": 0.5861653617826723,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009666851523661995,
      "loss": 1.778,
      "step": 6813
    },
    {
      "epoch": 0.5862513980899939,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00096667544636201,
      "loss": 1.8848,
      "step": 6814
    },
    {
      "epoch": 0.5863374343973157,
      "grad_norm": 1.5,
      "learning_rate": 0.0009666657389928857,
      "loss": 2.2256,
      "step": 6815
    },
    {
      "epoch": 0.5864234707046374,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009666560302588548,
      "loss": 2.0034,
      "step": 6816
    },
    {
      "epoch": 0.586509507011959,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009666463201599456,
      "loss": 1.2114,
      "step": 6817
    },
    {
      "epoch": 0.5865955433192808,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009666366086961869,
      "loss": 1.1512,
      "step": 6818
    },
    {
      "epoch": 0.5866815796266024,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009666268958676068,
      "loss": 0.8146,
      "step": 6819
    },
    {
      "epoch": 0.5867676159339241,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009666171816742337,
      "loss": 1.4508,
      "step": 6820
    },
    {
      "epoch": 0.5868536522412459,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009666074661160962,
      "loss": 1.4616,
      "step": 6821
    },
    {
      "epoch": 0.5869396885485675,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009665977491932227,
      "loss": 1.6062,
      "step": 6822
    },
    {
      "epoch": 0.5870257248558892,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009665880309056414,
      "loss": 2.0139,
      "step": 6823
    },
    {
      "epoch": 0.5871117611632108,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009665783112533808,
      "loss": 1.1878,
      "step": 6824
    },
    {
      "epoch": 0.5871977974705326,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009665685902364696,
      "loss": 1.4632,
      "step": 6825
    },
    {
      "epoch": 0.5872838337778542,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009665588678549358,
      "loss": 1.4332,
      "step": 6826
    },
    {
      "epoch": 0.5873698700851759,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009665491441088081,
      "loss": 2.1393,
      "step": 6827
    },
    {
      "epoch": 0.5874559063924977,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009665394189981149,
      "loss": 0.8655,
      "step": 6828
    },
    {
      "epoch": 0.5875419426998193,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009665296925228847,
      "loss": 0.9182,
      "step": 6829
    },
    {
      "epoch": 0.587627979007141,
      "grad_norm": 1.125,
      "learning_rate": 0.0009665199646831459,
      "loss": 1.1005,
      "step": 6830
    },
    {
      "epoch": 0.5877140153144627,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009665102354789268,
      "loss": 1.0092,
      "step": 6831
    },
    {
      "epoch": 0.5878000516217844,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009665005049102561,
      "loss": 1.2221,
      "step": 6832
    },
    {
      "epoch": 0.5878860879291061,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009664907729771622,
      "loss": 1.745,
      "step": 6833
    },
    {
      "epoch": 0.5879721242364278,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009664810396796733,
      "loss": 1.2509,
      "step": 6834
    },
    {
      "epoch": 0.5880581605437495,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009664713050178183,
      "loss": 1.6729,
      "step": 6835
    },
    {
      "epoch": 0.5881441968510711,
      "grad_norm": 1.125,
      "learning_rate": 0.0009664615689916254,
      "loss": 0.9954,
      "step": 6836
    },
    {
      "epoch": 0.5882302331583928,
      "grad_norm": 1.21875,
      "learning_rate": 0.000966451831601123,
      "loss": 0.9731,
      "step": 6837
    },
    {
      "epoch": 0.5883162694657146,
      "grad_norm": 2.21875,
      "learning_rate": 0.00096644209284634,
      "loss": 1.1864,
      "step": 6838
    },
    {
      "epoch": 0.5884023057730362,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009664323527273044,
      "loss": 1.5635,
      "step": 6839
    },
    {
      "epoch": 0.5884883420803579,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009664226112440448,
      "loss": 1.7101,
      "step": 6840
    },
    {
      "epoch": 0.5885743783876796,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00096641286839659,
      "loss": 1.5592,
      "step": 6841
    },
    {
      "epoch": 0.5886604146950013,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009664031241849681,
      "loss": 1.6306,
      "step": 6842
    },
    {
      "epoch": 0.588746451002323,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009663933786092079,
      "loss": 1.236,
      "step": 6843
    },
    {
      "epoch": 0.5888324873096447,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009663836316693377,
      "loss": 1.5574,
      "step": 6844
    },
    {
      "epoch": 0.5889185236169664,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009663738833653861,
      "loss": 1.7584,
      "step": 6845
    },
    {
      "epoch": 0.589004559924288,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009663641336973816,
      "loss": 1.3468,
      "step": 6846
    },
    {
      "epoch": 0.5890905962316098,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009663543826653527,
      "loss": 1.5162,
      "step": 6847
    },
    {
      "epoch": 0.5891766325389314,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000966344630269328,
      "loss": 1.109,
      "step": 6848
    },
    {
      "epoch": 0.5892626688462531,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009663348765093358,
      "loss": 1.4231,
      "step": 6849
    },
    {
      "epoch": 0.5893487051535748,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009663251213854047,
      "loss": 2.0514,
      "step": 6850
    },
    {
      "epoch": 0.5894347414608965,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009663153648975636,
      "loss": 1.5167,
      "step": 6851
    },
    {
      "epoch": 0.5895207777682182,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009663056070458407,
      "loss": 1.841,
      "step": 6852
    },
    {
      "epoch": 0.5896068140755398,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009662958478302645,
      "loss": 1.0662,
      "step": 6853
    },
    {
      "epoch": 0.5896928503828616,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009662860872508637,
      "loss": 1.6685,
      "step": 6854
    },
    {
      "epoch": 0.5897788866901833,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009662763253076667,
      "loss": 1.3199,
      "step": 6855
    },
    {
      "epoch": 0.5898649229975049,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009662665620007022,
      "loss": 0.9865,
      "step": 6856
    },
    {
      "epoch": 0.5899509593048267,
      "grad_norm": 3.34375,
      "learning_rate": 0.0009662567973299987,
      "loss": 2.4028,
      "step": 6857
    },
    {
      "epoch": 0.5900369956121483,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009662470312955847,
      "loss": 1.2001,
      "step": 6858
    },
    {
      "epoch": 0.59012303191947,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009662372638974888,
      "loss": 1.0678,
      "step": 6859
    },
    {
      "epoch": 0.5902090682267918,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009662274951357394,
      "loss": 1.7999,
      "step": 6860
    },
    {
      "epoch": 0.5902951045341134,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009662177250103657,
      "loss": 1.272,
      "step": 6861
    },
    {
      "epoch": 0.5903811408414351,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009662079535213955,
      "loss": 2.0743,
      "step": 6862
    },
    {
      "epoch": 0.5904671771487567,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009661981806688577,
      "loss": 1.3338,
      "step": 6863
    },
    {
      "epoch": 0.5905532134560785,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000966188406452781,
      "loss": 1.6063,
      "step": 6864
    },
    {
      "epoch": 0.5906392497634002,
      "grad_norm": 1.03125,
      "learning_rate": 0.000966178630873194,
      "loss": 1.2223,
      "step": 6865
    },
    {
      "epoch": 0.5907252860707218,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000966168853930125,
      "loss": 1.6416,
      "step": 6866
    },
    {
      "epoch": 0.5908113223780436,
      "grad_norm": 1.25,
      "learning_rate": 0.0009661590756236028,
      "loss": 1.8497,
      "step": 6867
    },
    {
      "epoch": 0.5908973586853652,
      "grad_norm": 1.25,
      "learning_rate": 0.0009661492959536559,
      "loss": 1.8507,
      "step": 6868
    },
    {
      "epoch": 0.5909833949926869,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009661395149203132,
      "loss": 1.6413,
      "step": 6869
    },
    {
      "epoch": 0.5910694313000086,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009661297325236029,
      "loss": 1.9401,
      "step": 6870
    },
    {
      "epoch": 0.5911554676073303,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009661199487635538,
      "loss": 1.6254,
      "step": 6871
    },
    {
      "epoch": 0.591241503914652,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009661101636401946,
      "loss": 1.3271,
      "step": 6872
    },
    {
      "epoch": 0.5913275402219736,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009661003771535537,
      "loss": 1.8339,
      "step": 6873
    },
    {
      "epoch": 0.5914135765292954,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009660905893036599,
      "loss": 0.9803,
      "step": 6874
    },
    {
      "epoch": 0.591499612836617,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009660808000905418,
      "loss": 1.3368,
      "step": 6875
    },
    {
      "epoch": 0.5915856491439387,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000966071009514228,
      "loss": 1.0114,
      "step": 6876
    },
    {
      "epoch": 0.5916716854512605,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009660612175747471,
      "loss": 1.3146,
      "step": 6877
    },
    {
      "epoch": 0.5917577217585821,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009660514242721277,
      "loss": 0.7999,
      "step": 6878
    },
    {
      "epoch": 0.5918437580659038,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009660416296063988,
      "loss": 1.113,
      "step": 6879
    },
    {
      "epoch": 0.5919297943732255,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009660318335775885,
      "loss": 1.2927,
      "step": 6880
    },
    {
      "epoch": 0.5920158306805472,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009660220361857258,
      "loss": 0.9586,
      "step": 6881
    },
    {
      "epoch": 0.5921018669878689,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009660122374308393,
      "loss": 1.671,
      "step": 6882
    },
    {
      "epoch": 0.5921879032951906,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009660024373129576,
      "loss": 1.813,
      "step": 6883
    },
    {
      "epoch": 0.5922739396025123,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009659926358321097,
      "loss": 1.0877,
      "step": 6884
    },
    {
      "epoch": 0.5923599759098339,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009659828329883236,
      "loss": 1.2585,
      "step": 6885
    },
    {
      "epoch": 0.5924460122171556,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009659730287816284,
      "loss": 1.5566,
      "step": 6886
    },
    {
      "epoch": 0.5925320485244774,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009659632232120528,
      "loss": 1.6343,
      "step": 6887
    },
    {
      "epoch": 0.592618084831799,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009659534162796253,
      "loss": 1.503,
      "step": 6888
    },
    {
      "epoch": 0.5927041211391207,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009659436079843749,
      "loss": 1.6619,
      "step": 6889
    },
    {
      "epoch": 0.5927901574464424,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009659337983263298,
      "loss": 1.7371,
      "step": 6890
    },
    {
      "epoch": 0.5928761937537641,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009659239873055192,
      "loss": 1.599,
      "step": 6891
    },
    {
      "epoch": 0.5929622300610857,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009659141749219713,
      "loss": 1.4563,
      "step": 6892
    },
    {
      "epoch": 0.5930482663684075,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009659043611757153,
      "loss": 1.5085,
      "step": 6893
    },
    {
      "epoch": 0.5931343026757292,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009658945460667795,
      "loss": 1.6309,
      "step": 6894
    },
    {
      "epoch": 0.5932203389830508,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009658847295951929,
      "loss": 1.9274,
      "step": 6895
    },
    {
      "epoch": 0.5933063752903726,
      "grad_norm": 1.3125,
      "learning_rate": 0.000965874911760984,
      "loss": 1.5184,
      "step": 6896
    },
    {
      "epoch": 0.5933924115976942,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009658650925641816,
      "loss": 1.3428,
      "step": 6897
    },
    {
      "epoch": 0.5934784479050159,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009658552720048143,
      "loss": 1.5403,
      "step": 6898
    },
    {
      "epoch": 0.5935644842123377,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000965845450082911,
      "loss": 1.4984,
      "step": 6899
    },
    {
      "epoch": 0.5936505205196593,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009658356267985004,
      "loss": 1.1316,
      "step": 6900
    },
    {
      "epoch": 0.593736556826981,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009658258021516112,
      "loss": 1.563,
      "step": 6901
    },
    {
      "epoch": 0.5938225931343026,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009658159761422721,
      "loss": 1.3189,
      "step": 6902
    },
    {
      "epoch": 0.5939086294416244,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009658061487705119,
      "loss": 1.4108,
      "step": 6903
    },
    {
      "epoch": 0.5939946657489461,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009657963200363593,
      "loss": 1.0251,
      "step": 6904
    },
    {
      "epoch": 0.5940807020562677,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000965786489939843,
      "loss": 1.2661,
      "step": 6905
    },
    {
      "epoch": 0.5941667383635895,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009657766584809919,
      "loss": 1.08,
      "step": 6906
    },
    {
      "epoch": 0.5942527746709111,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009657668256598345,
      "loss": 1.2157,
      "step": 6907
    },
    {
      "epoch": 0.5943388109782328,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009657569914763998,
      "loss": 1.6297,
      "step": 6908
    },
    {
      "epoch": 0.5944248472855546,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009657471559307165,
      "loss": 1.6244,
      "step": 6909
    },
    {
      "epoch": 0.5945108835928762,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009657373190228134,
      "loss": 0.9312,
      "step": 6910
    },
    {
      "epoch": 0.5945969199001979,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009657274807527192,
      "loss": 2.1043,
      "step": 6911
    },
    {
      "epoch": 0.5946829562075195,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009657176411204627,
      "loss": 0.971,
      "step": 6912
    },
    {
      "epoch": 0.5947689925148413,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009657078001260726,
      "loss": 1.0303,
      "step": 6913
    },
    {
      "epoch": 0.5948550288221629,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009656979577695779,
      "loss": 1.9082,
      "step": 6914
    },
    {
      "epoch": 0.5949410651294846,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009656881140510072,
      "loss": 1.427,
      "step": 6915
    },
    {
      "epoch": 0.5950271014368064,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009656782689703892,
      "loss": 1.8348,
      "step": 6916
    },
    {
      "epoch": 0.595113137744128,
      "grad_norm": 1.171875,
      "learning_rate": 0.000965668422527753,
      "loss": 1.3419,
      "step": 6917
    },
    {
      "epoch": 0.5951991740514497,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0009656585747231273,
      "loss": 0.6819,
      "step": 6918
    },
    {
      "epoch": 0.5952852103587714,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009656487255565405,
      "loss": 1.8518,
      "step": 6919
    },
    {
      "epoch": 0.5953712466660931,
      "grad_norm": 1.359375,
      "learning_rate": 0.000965638875028022,
      "loss": 1.0246,
      "step": 6920
    },
    {
      "epoch": 0.5954572829734148,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009656290231376004,
      "loss": 2.3341,
      "step": 6921
    },
    {
      "epoch": 0.5955433192807364,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009656191698853043,
      "loss": 1.5561,
      "step": 6922
    },
    {
      "epoch": 0.5956293555880582,
      "grad_norm": 2.40625,
      "learning_rate": 0.0009656093152711627,
      "loss": 2.1099,
      "step": 6923
    },
    {
      "epoch": 0.5957153918953798,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009655994592952044,
      "loss": 1.1773,
      "step": 6924
    },
    {
      "epoch": 0.5958014282027015,
      "grad_norm": 1.75,
      "learning_rate": 0.0009655896019574584,
      "loss": 1.9122,
      "step": 6925
    },
    {
      "epoch": 0.5958874645100233,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009655797432579532,
      "loss": 1.3931,
      "step": 6926
    },
    {
      "epoch": 0.5959735008173449,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009655698831967179,
      "loss": 0.9835,
      "step": 6927
    },
    {
      "epoch": 0.5960595371246666,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009655600217737812,
      "loss": 1.8696,
      "step": 6928
    },
    {
      "epoch": 0.5961455734319883,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000965550158989172,
      "loss": 0.6895,
      "step": 6929
    },
    {
      "epoch": 0.59623160973931,
      "grad_norm": 1.296875,
      "learning_rate": 0.000965540294842919,
      "loss": 1.3114,
      "step": 6930
    },
    {
      "epoch": 0.5963176460466317,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009655304293350513,
      "loss": 1.5304,
      "step": 6931
    },
    {
      "epoch": 0.5964036823539534,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009655205624655977,
      "loss": 1.4192,
      "step": 6932
    },
    {
      "epoch": 0.5964897186612751,
      "grad_norm": 1.0,
      "learning_rate": 0.000965510694234587,
      "loss": 0.719,
      "step": 6933
    },
    {
      "epoch": 0.5965757549685967,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009655008246420479,
      "loss": 1.3409,
      "step": 6934
    },
    {
      "epoch": 0.5966617912759185,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009654909536880095,
      "loss": 1.481,
      "step": 6935
    },
    {
      "epoch": 0.5967478275832401,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009654810813725006,
      "loss": 1.6021,
      "step": 6936
    },
    {
      "epoch": 0.5968338638905618,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009654712076955501,
      "loss": 1.7209,
      "step": 6937
    },
    {
      "epoch": 0.5969199001978835,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009654613326571869,
      "loss": 1.3901,
      "step": 6938
    },
    {
      "epoch": 0.5970059365052052,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009654514562574397,
      "loss": 1.2436,
      "step": 6939
    },
    {
      "epoch": 0.5970919728125269,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009654415784963375,
      "loss": 1.1023,
      "step": 6940
    },
    {
      "epoch": 0.5971780091198485,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009654316993739091,
      "loss": 1.3251,
      "step": 6941
    },
    {
      "epoch": 0.5972640454271703,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009654218188901838,
      "loss": 0.7436,
      "step": 6942
    },
    {
      "epoch": 0.597350081734492,
      "grad_norm": 1.078125,
      "learning_rate": 0.00096541193704519,
      "loss": 1.3175,
      "step": 6943
    },
    {
      "epoch": 0.5974361180418136,
      "grad_norm": 1.5,
      "learning_rate": 0.0009654020538389569,
      "loss": 1.3191,
      "step": 6944
    },
    {
      "epoch": 0.5975221543491354,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009653921692715133,
      "loss": 0.8346,
      "step": 6945
    },
    {
      "epoch": 0.597608190656457,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000965382283342888,
      "loss": 1.3326,
      "step": 6946
    },
    {
      "epoch": 0.5976942269637787,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009653723960531103,
      "loss": 1.1096,
      "step": 6947
    },
    {
      "epoch": 0.5977802632711005,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009653625074022086,
      "loss": 1.0664,
      "step": 6948
    },
    {
      "epoch": 0.5978662995784221,
      "grad_norm": 1.5625,
      "learning_rate": 0.000965352617390212,
      "loss": 1.312,
      "step": 6949
    },
    {
      "epoch": 0.5979523358857438,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009653427260171498,
      "loss": 1.3537,
      "step": 6950
    },
    {
      "epoch": 0.5980383721930654,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009653328332830505,
      "loss": 1.3044,
      "step": 6951
    },
    {
      "epoch": 0.5981244085003872,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009653229391879431,
      "loss": 1.3169,
      "step": 6952
    },
    {
      "epoch": 0.5982104448077089,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009653130437318567,
      "loss": 1.5386,
      "step": 6953
    },
    {
      "epoch": 0.5982964811150305,
      "grad_norm": 1.265625,
      "learning_rate": 0.00096530314691482,
      "loss": 1.4142,
      "step": 6954
    },
    {
      "epoch": 0.5983825174223523,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009652932487368622,
      "loss": 1.3314,
      "step": 6955
    },
    {
      "epoch": 0.5984685537296739,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009652833491980122,
      "loss": 1.6347,
      "step": 6956
    },
    {
      "epoch": 0.5985545900369956,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009652734482982988,
      "loss": 1.3567,
      "step": 6957
    },
    {
      "epoch": 0.5986406263443173,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000965263546037751,
      "loss": 1.3354,
      "step": 6958
    },
    {
      "epoch": 0.598726662651639,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000965253642416398,
      "loss": 1.3668,
      "step": 6959
    },
    {
      "epoch": 0.5988126989589607,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009652437374342684,
      "loss": 1.0406,
      "step": 6960
    },
    {
      "epoch": 0.5988987352662823,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009652338310913915,
      "loss": 1.1409,
      "step": 6961
    },
    {
      "epoch": 0.5989847715736041,
      "grad_norm": 1.359375,
      "learning_rate": 0.000965223923387796,
      "loss": 1.3938,
      "step": 6962
    },
    {
      "epoch": 0.5990708078809257,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009652140143235111,
      "loss": 1.3126,
      "step": 6963
    },
    {
      "epoch": 0.5991568441882474,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009652041038985656,
      "loss": 1.8401,
      "step": 6964
    },
    {
      "epoch": 0.5992428804955692,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009651941921129887,
      "loss": 1.1749,
      "step": 6965
    },
    {
      "epoch": 0.5993289168028908,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009651842789668091,
      "loss": 1.2014,
      "step": 6966
    },
    {
      "epoch": 0.5994149531102125,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009651743644600562,
      "loss": 1.1611,
      "step": 6967
    },
    {
      "epoch": 0.5995009894175342,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009651644485927584,
      "loss": 1.4508,
      "step": 6968
    },
    {
      "epoch": 0.5995870257248559,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009651545313649453,
      "loss": 1.2756,
      "step": 6969
    },
    {
      "epoch": 0.5996730620321776,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009651446127766456,
      "loss": 1.181,
      "step": 6970
    },
    {
      "epoch": 0.5997590983394993,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009651346928278884,
      "loss": 1.4252,
      "step": 6971
    },
    {
      "epoch": 0.599845134646821,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009651247715187027,
      "loss": 0.6935,
      "step": 6972
    },
    {
      "epoch": 0.5999311709541426,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009651148488491175,
      "loss": 0.6706,
      "step": 6973
    },
    {
      "epoch": 0.6000172072614643,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009651049248191618,
      "loss": 1.8682,
      "step": 6974
    },
    {
      "epoch": 0.6001032435687861,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009650949994288646,
      "loss": 1.0311,
      "step": 6975
    },
    {
      "epoch": 0.6001892798761077,
      "grad_norm": 2.421875,
      "learning_rate": 0.0009650850726782551,
      "loss": 1.6995,
      "step": 6976
    },
    {
      "epoch": 0.6002753161834294,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009650751445673622,
      "loss": 1.1889,
      "step": 6977
    },
    {
      "epoch": 0.6003613524907511,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009650652150962149,
      "loss": 1.3453,
      "step": 6978
    },
    {
      "epoch": 0.6004473887980728,
      "grad_norm": 1.5,
      "learning_rate": 0.0009650552842648424,
      "loss": 1.2125,
      "step": 6979
    },
    {
      "epoch": 0.6005334251053944,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009650453520732734,
      "loss": 1.6205,
      "step": 6980
    },
    {
      "epoch": 0.6006194614127162,
      "grad_norm": 1.75,
      "learning_rate": 0.0009650354185215374,
      "loss": 1.4671,
      "step": 6981
    },
    {
      "epoch": 0.6007054977200379,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009650254836096633,
      "loss": 1.1354,
      "step": 6982
    },
    {
      "epoch": 0.6007915340273595,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009650155473376798,
      "loss": 1.3145,
      "step": 6983
    },
    {
      "epoch": 0.6008775703346813,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009650056097056166,
      "loss": 1.2781,
      "step": 6984
    },
    {
      "epoch": 0.6009636066420029,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009649956707135022,
      "loss": 1.5934,
      "step": 6985
    },
    {
      "epoch": 0.6010496429493246,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000964985730361366,
      "loss": 0.9467,
      "step": 6986
    },
    {
      "epoch": 0.6011356792566463,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009649757886492369,
      "loss": 1.1604,
      "step": 6987
    },
    {
      "epoch": 0.601221715563968,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009649658455771441,
      "loss": 1.1258,
      "step": 6988
    },
    {
      "epoch": 0.6013077518712897,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009649559011451167,
      "loss": 1.6089,
      "step": 6989
    },
    {
      "epoch": 0.6013937881786113,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009649459553531836,
      "loss": 1.3419,
      "step": 6990
    },
    {
      "epoch": 0.6014798244859331,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009649360082013742,
      "loss": 1.2129,
      "step": 6991
    },
    {
      "epoch": 0.6015658607932548,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009649260596897172,
      "loss": 1.8956,
      "step": 6992
    },
    {
      "epoch": 0.6016518971005764,
      "grad_norm": 1.8125,
      "learning_rate": 0.000964916109818242,
      "loss": 1.4687,
      "step": 6993
    },
    {
      "epoch": 0.6017379334078982,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009649061585869776,
      "loss": 1.2746,
      "step": 6994
    },
    {
      "epoch": 0.6018239697152198,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009648962059959531,
      "loss": 0.9433,
      "step": 6995
    },
    {
      "epoch": 0.6019100060225415,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009648862520451975,
      "loss": 1.2049,
      "step": 6996
    },
    {
      "epoch": 0.6019960423298633,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009648762967347402,
      "loss": 0.7622,
      "step": 6997
    },
    {
      "epoch": 0.6020820786371849,
      "grad_norm": 1.8671875,
      "learning_rate": 0.00096486634006461,
      "loss": 1.2949,
      "step": 6998
    },
    {
      "epoch": 0.6021681149445066,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009648563820348363,
      "loss": 1.6234,
      "step": 6999
    },
    {
      "epoch": 0.6022541512518282,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009648464226454479,
      "loss": 1.2622,
      "step": 7000
    },
    {
      "epoch": 0.60234018755915,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009648364618964744,
      "loss": 1.2325,
      "step": 7001
    },
    {
      "epoch": 0.6024262238664716,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009648264997879443,
      "loss": 0.6147,
      "step": 7002
    },
    {
      "epoch": 0.6025122601737933,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009648165363198874,
      "loss": 1.2472,
      "step": 7003
    },
    {
      "epoch": 0.6025982964811151,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009648065714923324,
      "loss": 1.8543,
      "step": 7004
    },
    {
      "epoch": 0.6026843327884367,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009647966053053087,
      "loss": 1.5223,
      "step": 7005
    },
    {
      "epoch": 0.6027703690957584,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009647866377588452,
      "loss": 0.7483,
      "step": 7006
    },
    {
      "epoch": 0.60285640540308,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009647766688529711,
      "loss": 0.9733,
      "step": 7007
    },
    {
      "epoch": 0.6029424417104018,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009647666985877158,
      "loss": 0.9816,
      "step": 7008
    },
    {
      "epoch": 0.6030284780177235,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009647567269631083,
      "loss": 1.2671,
      "step": 7009
    },
    {
      "epoch": 0.6031145143250451,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009647467539791776,
      "loss": 2.1521,
      "step": 7010
    },
    {
      "epoch": 0.6032005506323669,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009647367796359532,
      "loss": 1.8232,
      "step": 7011
    },
    {
      "epoch": 0.6032865869396885,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000964726803933464,
      "loss": 1.2784,
      "step": 7012
    },
    {
      "epoch": 0.6033726232470102,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009647168268717395,
      "loss": 1.0237,
      "step": 7013
    },
    {
      "epoch": 0.603458659554332,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009647068484508084,
      "loss": 2.2655,
      "step": 7014
    },
    {
      "epoch": 0.6035446958616536,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009646968686707003,
      "loss": 1.7864,
      "step": 7015
    },
    {
      "epoch": 0.6036307321689753,
      "grad_norm": 1.203125,
      "learning_rate": 0.000964686887531444,
      "loss": 1.9396,
      "step": 7016
    },
    {
      "epoch": 0.603716768476297,
      "grad_norm": 1.125,
      "learning_rate": 0.0009646769050330692,
      "loss": 1.7442,
      "step": 7017
    },
    {
      "epoch": 0.6038028047836187,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009646669211756047,
      "loss": 1.1969,
      "step": 7018
    },
    {
      "epoch": 0.6038888410909404,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00096465693595908,
      "loss": 1.731,
      "step": 7019
    },
    {
      "epoch": 0.6039748773982621,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009646469493835239,
      "loss": 1.138,
      "step": 7020
    },
    {
      "epoch": 0.6040609137055838,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009646369614489661,
      "loss": 0.7258,
      "step": 7021
    },
    {
      "epoch": 0.6041469500129054,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009646269721554353,
      "loss": 1.7827,
      "step": 7022
    },
    {
      "epoch": 0.6042329863202272,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009646169815029611,
      "loss": 1.8818,
      "step": 7023
    },
    {
      "epoch": 0.6043190226275488,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009646069894915726,
      "loss": 1.1458,
      "step": 7024
    },
    {
      "epoch": 0.6044050589348705,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009645969961212989,
      "loss": 1.4554,
      "step": 7025
    },
    {
      "epoch": 0.6044910952421922,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009645870013921694,
      "loss": 1.1222,
      "step": 7026
    },
    {
      "epoch": 0.6045771315495139,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009645770053042133,
      "loss": 1.1683,
      "step": 7027
    },
    {
      "epoch": 0.6046631678568356,
      "grad_norm": 1.625,
      "learning_rate": 0.0009645670078574599,
      "loss": 1.3883,
      "step": 7028
    },
    {
      "epoch": 0.6047492041641572,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009645570090519382,
      "loss": 1.0684,
      "step": 7029
    },
    {
      "epoch": 0.604835240471479,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009645470088876778,
      "loss": 1.2322,
      "step": 7030
    },
    {
      "epoch": 0.6049212767788007,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009645370073647075,
      "loss": 2.0497,
      "step": 7031
    },
    {
      "epoch": 0.6050073130861223,
      "grad_norm": 1.5,
      "learning_rate": 0.000964527004483057,
      "loss": 1.3192,
      "step": 7032
    },
    {
      "epoch": 0.6050933493934441,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009645170002427553,
      "loss": 1.386,
      "step": 7033
    },
    {
      "epoch": 0.6051793857007657,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009645069946438317,
      "loss": 1.2323,
      "step": 7034
    },
    {
      "epoch": 0.6052654220080874,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009644969876863155,
      "loss": 1.2161,
      "step": 7035
    },
    {
      "epoch": 0.6053514583154092,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009644869793702358,
      "loss": 1.2531,
      "step": 7036
    },
    {
      "epoch": 0.6054374946227308,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009644769696956221,
      "loss": 2.0921,
      "step": 7037
    },
    {
      "epoch": 0.6055235309300525,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009644669586625036,
      "loss": 1.6606,
      "step": 7038
    },
    {
      "epoch": 0.6056095672373741,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009644569462709099,
      "loss": 1.3508,
      "step": 7039
    },
    {
      "epoch": 0.6056956035446959,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009644469325208695,
      "loss": 1.5521,
      "step": 7040
    },
    {
      "epoch": 0.6057816398520176,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009644369174124124,
      "loss": 2.1392,
      "step": 7041
    },
    {
      "epoch": 0.6058676761593392,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009644269009455676,
      "loss": 1.6689,
      "step": 7042
    },
    {
      "epoch": 0.605953712466661,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009644168831203643,
      "loss": 1.8109,
      "step": 7043
    },
    {
      "epoch": 0.6060397487739826,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009644068639368322,
      "loss": 1.5455,
      "step": 7044
    },
    {
      "epoch": 0.6061257850813043,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009643968433950002,
      "loss": 1.7758,
      "step": 7045
    },
    {
      "epoch": 0.606211821388626,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009643868214948978,
      "loss": 2.0733,
      "step": 7046
    },
    {
      "epoch": 0.6062978576959477,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009643767982365542,
      "loss": 1.257,
      "step": 7047
    },
    {
      "epoch": 0.6063838940032694,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009643667736199988,
      "loss": 1.0933,
      "step": 7048
    },
    {
      "epoch": 0.606469930310591,
      "grad_norm": 1.625,
      "learning_rate": 0.000964356747645261,
      "loss": 1.1038,
      "step": 7049
    },
    {
      "epoch": 0.6065559666179128,
      "grad_norm": 1.25,
      "learning_rate": 0.0009643467203123699,
      "loss": 1.623,
      "step": 7050
    },
    {
      "epoch": 0.6066420029252344,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009643366916213549,
      "loss": 1.8672,
      "step": 7051
    },
    {
      "epoch": 0.6067280392325561,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009643266615722457,
      "loss": 1.0501,
      "step": 7052
    },
    {
      "epoch": 0.6068140755398779,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000964316630165071,
      "loss": 0.5546,
      "step": 7053
    },
    {
      "epoch": 0.6069001118471995,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009643065973998605,
      "loss": 1.8077,
      "step": 7054
    },
    {
      "epoch": 0.6069861481545212,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009642965632766437,
      "loss": 1.1242,
      "step": 7055
    },
    {
      "epoch": 0.6070721844618429,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009642865277954496,
      "loss": 1.2498,
      "step": 7056
    },
    {
      "epoch": 0.6071582207691646,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009642764909563077,
      "loss": 0.8925,
      "step": 7057
    },
    {
      "epoch": 0.6072442570764863,
      "grad_norm": 2.78125,
      "learning_rate": 0.0009642664527592474,
      "loss": 1.894,
      "step": 7058
    },
    {
      "epoch": 0.607330293383808,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009642564132042981,
      "loss": 1.3117,
      "step": 7059
    },
    {
      "epoch": 0.6074163296911297,
      "grad_norm": 1.375,
      "learning_rate": 0.000964246372291489,
      "loss": 1.3237,
      "step": 7060
    },
    {
      "epoch": 0.6075023659984513,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009642363300208494,
      "loss": 1.4454,
      "step": 7061
    },
    {
      "epoch": 0.607588402305773,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000964226286392409,
      "loss": 1.2849,
      "step": 7062
    },
    {
      "epoch": 0.6076744386130948,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009642162414061971,
      "loss": 1.9003,
      "step": 7063
    },
    {
      "epoch": 0.6077604749204164,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009642061950622427,
      "loss": 1.412,
      "step": 7064
    },
    {
      "epoch": 0.6078465112277381,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009641961473605757,
      "loss": 0.5819,
      "step": 7065
    },
    {
      "epoch": 0.6079325475350598,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009641860983012251,
      "loss": 1.3606,
      "step": 7066
    },
    {
      "epoch": 0.6080185838423815,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009641760478842205,
      "loss": 1.2362,
      "step": 7067
    },
    {
      "epoch": 0.6081046201497031,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009641659961095912,
      "loss": 1.7876,
      "step": 7068
    },
    {
      "epoch": 0.6081906564570249,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009641559429773667,
      "loss": 1.555,
      "step": 7069
    },
    {
      "epoch": 0.6082766927643466,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009641458884875765,
      "loss": 0.7136,
      "step": 7070
    },
    {
      "epoch": 0.6083627290716682,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009641358326402496,
      "loss": 1.6475,
      "step": 7071
    },
    {
      "epoch": 0.60844876537899,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009641257754354157,
      "loss": 1.6817,
      "step": 7072
    },
    {
      "epoch": 0.6085348016863116,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009641157168731042,
      "loss": 1.8386,
      "step": 7073
    },
    {
      "epoch": 0.6086208379936333,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009641056569533444,
      "loss": 1.254,
      "step": 7074
    },
    {
      "epoch": 0.608706874300955,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000964095595676166,
      "loss": 1.3603,
      "step": 7075
    },
    {
      "epoch": 0.6087929106082767,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009640855330415981,
      "loss": 1.765,
      "step": 7076
    },
    {
      "epoch": 0.6088789469155984,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009640754690496703,
      "loss": 1.8147,
      "step": 7077
    },
    {
      "epoch": 0.60896498322292,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009640654037004121,
      "loss": 1.1618,
      "step": 7078
    },
    {
      "epoch": 0.6090510195302418,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009640553369938529,
      "loss": 1.4703,
      "step": 7079
    },
    {
      "epoch": 0.6091370558375635,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009640452689300219,
      "loss": 1.3383,
      "step": 7080
    },
    {
      "epoch": 0.6092230921448851,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009640351995089488,
      "loss": 2.061,
      "step": 7081
    },
    {
      "epoch": 0.6093091284522069,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000964025128730663,
      "loss": 1.142,
      "step": 7082
    },
    {
      "epoch": 0.6093951647595285,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000964015056595194,
      "loss": 1.0835,
      "step": 7083
    },
    {
      "epoch": 0.6094812010668502,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009640049831025712,
      "loss": 1.1886,
      "step": 7084
    },
    {
      "epoch": 0.609567237374172,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000963994908252824,
      "loss": 1.0263,
      "step": 7085
    },
    {
      "epoch": 0.6096532736814936,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000963984832045982,
      "loss": 2.0957,
      "step": 7086
    },
    {
      "epoch": 0.6097393099888153,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009639747544820745,
      "loss": 1.5012,
      "step": 7087
    },
    {
      "epoch": 0.6098253462961369,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009639646755611312,
      "loss": 1.1061,
      "step": 7088
    },
    {
      "epoch": 0.6099113826034587,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009639545952831814,
      "loss": 1.2647,
      "step": 7089
    },
    {
      "epoch": 0.6099974189107803,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009639445136482548,
      "loss": 0.9793,
      "step": 7090
    },
    {
      "epoch": 0.610083455218102,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009639344306563804,
      "loss": 1.0745,
      "step": 7091
    },
    {
      "epoch": 0.6101694915254238,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009639243463075883,
      "loss": 1.1736,
      "step": 7092
    },
    {
      "epoch": 0.6102555278327454,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009639142606019075,
      "loss": 1.0586,
      "step": 7093
    },
    {
      "epoch": 0.6103415641400671,
      "grad_norm": 1.125,
      "learning_rate": 0.000963904173539368,
      "loss": 1.5523,
      "step": 7094
    },
    {
      "epoch": 0.6104276004473888,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009638940851199989,
      "loss": 1.2088,
      "step": 7095
    },
    {
      "epoch": 0.6105136367547105,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009638839953438296,
      "loss": 1.3225,
      "step": 7096
    },
    {
      "epoch": 0.6105996730620322,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009638739042108902,
      "loss": 1.2304,
      "step": 7097
    },
    {
      "epoch": 0.6106857093693538,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009638638117212095,
      "loss": 1.0813,
      "step": 7098
    },
    {
      "epoch": 0.6107717456766756,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009638537178748177,
      "loss": 1.8524,
      "step": 7099
    },
    {
      "epoch": 0.6108577819839972,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009638436226717437,
      "loss": 1.2881,
      "step": 7100
    },
    {
      "epoch": 0.6109438182913189,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009638335261120174,
      "loss": 1.0226,
      "step": 7101
    },
    {
      "epoch": 0.6110298545986407,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009638234281956684,
      "loss": 1.8899,
      "step": 7102
    },
    {
      "epoch": 0.6111158909059623,
      "grad_norm": 1.6875,
      "learning_rate": 0.000963813328922726,
      "loss": 1.8772,
      "step": 7103
    },
    {
      "epoch": 0.611201927213284,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009638032282932197,
      "loss": 1.6317,
      "step": 7104
    },
    {
      "epoch": 0.6112879635206057,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009637931263071793,
      "loss": 2.0446,
      "step": 7105
    },
    {
      "epoch": 0.6113739998279274,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009637830229646341,
      "loss": 2.0354,
      "step": 7106
    },
    {
      "epoch": 0.6114600361352491,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009637729182656138,
      "loss": 2.0923,
      "step": 7107
    },
    {
      "epoch": 0.6115460724425708,
      "grad_norm": 1.625,
      "learning_rate": 0.0009637628122101479,
      "loss": 1.1728,
      "step": 7108
    },
    {
      "epoch": 0.6116321087498925,
      "grad_norm": 1.25,
      "learning_rate": 0.0009637527047982661,
      "loss": 1.3414,
      "step": 7109
    },
    {
      "epoch": 0.6117181450572141,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009637425960299977,
      "loss": 1.3883,
      "step": 7110
    },
    {
      "epoch": 0.6118041813645358,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009637324859053724,
      "loss": 1.4643,
      "step": 7111
    },
    {
      "epoch": 0.6118902176718575,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0009637223744244198,
      "loss": 1.5456,
      "step": 7112
    },
    {
      "epoch": 0.6119762539791792,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009637122615871694,
      "loss": 1.8427,
      "step": 7113
    },
    {
      "epoch": 0.6120622902865009,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009637021473936509,
      "loss": 1.3484,
      "step": 7114
    },
    {
      "epoch": 0.6121483265938226,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009636920318438938,
      "loss": 1.1086,
      "step": 7115
    },
    {
      "epoch": 0.6122343629011443,
      "grad_norm": 1.25,
      "learning_rate": 0.0009636819149379277,
      "loss": 1.1874,
      "step": 7116
    },
    {
      "epoch": 0.6123203992084659,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009636717966757821,
      "loss": 1.7289,
      "step": 7117
    },
    {
      "epoch": 0.6124064355157877,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009636616770574867,
      "loss": 1.7555,
      "step": 7118
    },
    {
      "epoch": 0.6124924718231094,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009636515560830711,
      "loss": 1.3685,
      "step": 7119
    },
    {
      "epoch": 0.612578508130431,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009636414337525649,
      "loss": 1.2686,
      "step": 7120
    },
    {
      "epoch": 0.6126645444377528,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009636313100659976,
      "loss": 1.1727,
      "step": 7121
    },
    {
      "epoch": 0.6127505807450744,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000963621185023399,
      "loss": 1.0568,
      "step": 7122
    },
    {
      "epoch": 0.6128366170523961,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009636110586247984,
      "loss": 1.935,
      "step": 7123
    },
    {
      "epoch": 0.6129226533597179,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009636009308702258,
      "loss": 2.005,
      "step": 7124
    },
    {
      "epoch": 0.6130086896670395,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009635908017597106,
      "loss": 1.49,
      "step": 7125
    },
    {
      "epoch": 0.6130947259743612,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009635806712932824,
      "loss": 1.9066,
      "step": 7126
    },
    {
      "epoch": 0.6131807622816828,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009635705394709711,
      "loss": 2.0477,
      "step": 7127
    },
    {
      "epoch": 0.6132667985890046,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009635604062928058,
      "loss": 1.1021,
      "step": 7128
    },
    {
      "epoch": 0.6133528348963263,
      "grad_norm": 1.0,
      "learning_rate": 0.0009635502717588167,
      "loss": 0.9337,
      "step": 7129
    },
    {
      "epoch": 0.6134388712036479,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009635401358690332,
      "loss": 0.7946,
      "step": 7130
    },
    {
      "epoch": 0.6135249075109697,
      "grad_norm": 1.25,
      "learning_rate": 0.000963529998623485,
      "loss": 1.2328,
      "step": 7131
    },
    {
      "epoch": 0.6136109438182913,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009635198600222016,
      "loss": 1.1774,
      "step": 7132
    },
    {
      "epoch": 0.613696980125613,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009635097200652127,
      "loss": 0.7962,
      "step": 7133
    },
    {
      "epoch": 0.6137830164329346,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009634995787525482,
      "loss": 1.8,
      "step": 7134
    },
    {
      "epoch": 0.6138690527402564,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009634894360842374,
      "loss": 1.2294,
      "step": 7135
    },
    {
      "epoch": 0.6139550890475781,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009634792920603103,
      "loss": 2.3284,
      "step": 7136
    },
    {
      "epoch": 0.6140411253548997,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009634691466807963,
      "loss": 1.5226,
      "step": 7137
    },
    {
      "epoch": 0.6141271616622215,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009634589999457252,
      "loss": 0.7907,
      "step": 7138
    },
    {
      "epoch": 0.6142131979695431,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009634488518551268,
      "loss": 0.8271,
      "step": 7139
    },
    {
      "epoch": 0.6142992342768648,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009634387024090305,
      "loss": 1.2935,
      "step": 7140
    },
    {
      "epoch": 0.6143852705841866,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009634285516074662,
      "loss": 1.4177,
      "step": 7141
    },
    {
      "epoch": 0.6144713068915082,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009634183994504635,
      "loss": 1.4649,
      "step": 7142
    },
    {
      "epoch": 0.6145573431988299,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009634082459380521,
      "loss": 1.6682,
      "step": 7143
    },
    {
      "epoch": 0.6146433795061516,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009633980910702618,
      "loss": 1.8419,
      "step": 7144
    },
    {
      "epoch": 0.6147294158134733,
      "grad_norm": 1.125,
      "learning_rate": 0.0009633879348471219,
      "loss": 0.9874,
      "step": 7145
    },
    {
      "epoch": 0.614815452120795,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009633777772686627,
      "loss": 1.4853,
      "step": 7146
    },
    {
      "epoch": 0.6149014884281166,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009633676183349136,
      "loss": 1.0431,
      "step": 7147
    },
    {
      "epoch": 0.6149875247354384,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009633574580459043,
      "loss": 1.4656,
      "step": 7148
    },
    {
      "epoch": 0.61507356104276,
      "grad_norm": 1.375,
      "learning_rate": 0.0009633472964016645,
      "loss": 1.2396,
      "step": 7149
    },
    {
      "epoch": 0.6151595973500817,
      "grad_norm": 1.25,
      "learning_rate": 0.000963337133402224,
      "loss": 0.9155,
      "step": 7150
    },
    {
      "epoch": 0.6152456336574035,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009633269690476125,
      "loss": 1.9221,
      "step": 7151
    },
    {
      "epoch": 0.6153316699647251,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009633168033378599,
      "loss": 1.4294,
      "step": 7152
    },
    {
      "epoch": 0.6154177062720468,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009633066362729954,
      "loss": 2.3464,
      "step": 7153
    },
    {
      "epoch": 0.6155037425793685,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009632964678530494,
      "loss": 1.0022,
      "step": 7154
    },
    {
      "epoch": 0.6155897788866902,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009632862980780512,
      "loss": 1.6834,
      "step": 7155
    },
    {
      "epoch": 0.6156758151940118,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009632761269480308,
      "loss": 1.2837,
      "step": 7156
    },
    {
      "epoch": 0.6157618515013336,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009632659544630177,
      "loss": 1.5022,
      "step": 7157
    },
    {
      "epoch": 0.6158478878086553,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009632557806230418,
      "loss": 1.0731,
      "step": 7158
    },
    {
      "epoch": 0.6159339241159769,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009632456054281329,
      "loss": 0.992,
      "step": 7159
    },
    {
      "epoch": 0.6160199604232987,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009632354288783208,
      "loss": 1.2079,
      "step": 7160
    },
    {
      "epoch": 0.6161059967306203,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009632252509736351,
      "loss": 1.9514,
      "step": 7161
    },
    {
      "epoch": 0.616192033037942,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009632150717141054,
      "loss": 1.7987,
      "step": 7162
    },
    {
      "epoch": 0.6162780693452637,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000963204891099762,
      "loss": 0.9004,
      "step": 7163
    },
    {
      "epoch": 0.6163641056525854,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009631947091306341,
      "loss": 1.2596,
      "step": 7164
    },
    {
      "epoch": 0.6164501419599071,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009631845258067519,
      "loss": 1.6776,
      "step": 7165
    },
    {
      "epoch": 0.6165361782672287,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009631743411281451,
      "loss": 0.8087,
      "step": 7166
    },
    {
      "epoch": 0.6166222145745505,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009631641550948434,
      "loss": 0.9469,
      "step": 7167
    },
    {
      "epoch": 0.6167082508818722,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009631539677068766,
      "loss": 2.158,
      "step": 7168
    },
    {
      "epoch": 0.6167942871891938,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009631437789642747,
      "loss": 1.7296,
      "step": 7169
    },
    {
      "epoch": 0.6168803234965156,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009631335888670671,
      "loss": 1.9744,
      "step": 7170
    },
    {
      "epoch": 0.6169663598038372,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009631233974152839,
      "loss": 0.7601,
      "step": 7171
    },
    {
      "epoch": 0.6170523961111589,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009631132046089548,
      "loss": 1.7643,
      "step": 7172
    },
    {
      "epoch": 0.6171384324184807,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009631030104481098,
      "loss": 1.2366,
      "step": 7173
    },
    {
      "epoch": 0.6172244687258023,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009630928149327783,
      "loss": 2.3384,
      "step": 7174
    },
    {
      "epoch": 0.617310505033124,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009630826180629906,
      "loss": 1.3368,
      "step": 7175
    },
    {
      "epoch": 0.6173965413404456,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009630724198387762,
      "loss": 1.8464,
      "step": 7176
    },
    {
      "epoch": 0.6174825776477674,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009630622202601651,
      "loss": 1.1061,
      "step": 7177
    },
    {
      "epoch": 0.617568613955089,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000963052019327187,
      "loss": 1.429,
      "step": 7178
    },
    {
      "epoch": 0.6176546502624107,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009630418170398718,
      "loss": 1.6493,
      "step": 7179
    },
    {
      "epoch": 0.6177406865697325,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009630316133982496,
      "loss": 1.6791,
      "step": 7180
    },
    {
      "epoch": 0.6178267228770541,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009630214084023497,
      "loss": 1.8248,
      "step": 7181
    },
    {
      "epoch": 0.6179127591843758,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009630112020522022,
      "loss": 1.2743,
      "step": 7182
    },
    {
      "epoch": 0.6179987954916975,
      "grad_norm": 4.15625,
      "learning_rate": 0.000963000994347837,
      "loss": 1.3137,
      "step": 7183
    },
    {
      "epoch": 0.6180848317990192,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009629907852892841,
      "loss": 1.8096,
      "step": 7184
    },
    {
      "epoch": 0.6181708681063409,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009629805748765732,
      "loss": 0.505,
      "step": 7185
    },
    {
      "epoch": 0.6182569044136625,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009629703631097341,
      "loss": 1.4648,
      "step": 7186
    },
    {
      "epoch": 0.6183429407209843,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009629601499887967,
      "loss": 1.9743,
      "step": 7187
    },
    {
      "epoch": 0.6184289770283059,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009629499355137909,
      "loss": 1.2812,
      "step": 7188
    },
    {
      "epoch": 0.6185150133356276,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009629397196847466,
      "loss": 1.065,
      "step": 7189
    },
    {
      "epoch": 0.6186010496429494,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009629295025016936,
      "loss": 1.0879,
      "step": 7190
    },
    {
      "epoch": 0.618687085950271,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009629192839646619,
      "loss": 1.4704,
      "step": 7191
    },
    {
      "epoch": 0.6187731222575927,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009629090640736813,
      "loss": 1.1122,
      "step": 7192
    },
    {
      "epoch": 0.6188591585649144,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009628988428287818,
      "loss": 1.3302,
      "step": 7193
    },
    {
      "epoch": 0.6189451948722361,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009628886202299931,
      "loss": 1.6248,
      "step": 7194
    },
    {
      "epoch": 0.6190312311795578,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009628783962773453,
      "loss": 1.0634,
      "step": 7195
    },
    {
      "epoch": 0.6191172674868795,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009628681709708682,
      "loss": 1.5116,
      "step": 7196
    },
    {
      "epoch": 0.6192033037942012,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009628579443105918,
      "loss": 1.2718,
      "step": 7197
    },
    {
      "epoch": 0.6192893401015228,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009628477162965458,
      "loss": 1.8281,
      "step": 7198
    },
    {
      "epoch": 0.6193753764088445,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009628374869287601,
      "loss": 1.2416,
      "step": 7199
    },
    {
      "epoch": 0.6194614127161662,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009628272562072651,
      "loss": 1.1449,
      "step": 7200
    },
    {
      "epoch": 0.6195474490234879,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009628170241320901,
      "loss": 1.0778,
      "step": 7201
    },
    {
      "epoch": 0.6196334853308096,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009628067907032656,
      "loss": 2.1418,
      "step": 7202
    },
    {
      "epoch": 0.6197195216381313,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009627965559208211,
      "loss": 1.145,
      "step": 7203
    },
    {
      "epoch": 0.619805557945453,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009627863197847867,
      "loss": 1.655,
      "step": 7204
    },
    {
      "epoch": 0.6198915942527746,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009627760822951923,
      "loss": 1.8001,
      "step": 7205
    },
    {
      "epoch": 0.6199776305600964,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009627658434520678,
      "loss": 0.9813,
      "step": 7206
    },
    {
      "epoch": 0.6200636668674181,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009627556032554435,
      "loss": 1.298,
      "step": 7207
    },
    {
      "epoch": 0.6201497031747397,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009627453617053488,
      "loss": 1.836,
      "step": 7208
    },
    {
      "epoch": 0.6202357394820615,
      "grad_norm": 2.390625,
      "learning_rate": 0.0009627351188018138,
      "loss": 1.7202,
      "step": 7209
    },
    {
      "epoch": 0.6203217757893831,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009627248745448689,
      "loss": 1.5216,
      "step": 7210
    },
    {
      "epoch": 0.6204078120967048,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009627146289345435,
      "loss": 0.8978,
      "step": 7211
    },
    {
      "epoch": 0.6204938484040265,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009627043819708678,
      "loss": 0.8441,
      "step": 7212
    },
    {
      "epoch": 0.6205798847113482,
      "grad_norm": 2.203125,
      "learning_rate": 0.000962694133653872,
      "loss": 2.6294,
      "step": 7213
    },
    {
      "epoch": 0.6206659210186699,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009626838839835856,
      "loss": 1.8133,
      "step": 7214
    },
    {
      "epoch": 0.6207519573259915,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009626736329600389,
      "loss": 1.5707,
      "step": 7215
    },
    {
      "epoch": 0.6208379936333133,
      "grad_norm": 1.203125,
      "learning_rate": 0.000962663380583262,
      "loss": 1.5272,
      "step": 7216
    },
    {
      "epoch": 0.620924029940635,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009626531268532846,
      "loss": 0.9988,
      "step": 7217
    },
    {
      "epoch": 0.6210100662479566,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009626428717701366,
      "loss": 1.1861,
      "step": 7218
    },
    {
      "epoch": 0.6210961025552784,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009626326153338483,
      "loss": 1.1091,
      "step": 7219
    },
    {
      "epoch": 0.6211821388626,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009626223575444496,
      "loss": 0.9697,
      "step": 7220
    },
    {
      "epoch": 0.6212681751699217,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009626120984019704,
      "loss": 1.1772,
      "step": 7221
    },
    {
      "epoch": 0.6213542114772433,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000962601837906441,
      "loss": 1.8146,
      "step": 7222
    },
    {
      "epoch": 0.6214402477845651,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009625915760578909,
      "loss": 1.5085,
      "step": 7223
    },
    {
      "epoch": 0.6215262840918868,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009625813128563505,
      "loss": 1.8292,
      "step": 7224
    },
    {
      "epoch": 0.6216123203992084,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009625710483018499,
      "loss": 1.4075,
      "step": 7225
    },
    {
      "epoch": 0.6216983567065302,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009625607823944188,
      "loss": 1.0567,
      "step": 7226
    },
    {
      "epoch": 0.6217843930138518,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009625505151340875,
      "loss": 1.8916,
      "step": 7227
    },
    {
      "epoch": 0.6218704293211735,
      "grad_norm": 1.25,
      "learning_rate": 0.0009625402465208856,
      "loss": 1.0333,
      "step": 7228
    },
    {
      "epoch": 0.6219564656284953,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009625299765548437,
      "loss": 1.2891,
      "step": 7229
    },
    {
      "epoch": 0.6220425019358169,
      "grad_norm": 0.875,
      "learning_rate": 0.0009625197052359916,
      "loss": 0.653,
      "step": 7230
    },
    {
      "epoch": 0.6221285382431386,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009625094325643591,
      "loss": 1.4833,
      "step": 7231
    },
    {
      "epoch": 0.6222145745504603,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009624991585399767,
      "loss": 1.63,
      "step": 7232
    },
    {
      "epoch": 0.622300610857782,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009624888831628741,
      "loss": 2.1004,
      "step": 7233
    },
    {
      "epoch": 0.6223866471651037,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009624786064330815,
      "loss": 1.8166,
      "step": 7234
    },
    {
      "epoch": 0.6224726834724253,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009624683283506288,
      "loss": 1.0856,
      "step": 7235
    },
    {
      "epoch": 0.6225587197797471,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009624580489155463,
      "loss": 1.0361,
      "step": 7236
    },
    {
      "epoch": 0.6226447560870687,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009624477681278639,
      "loss": 0.943,
      "step": 7237
    },
    {
      "epoch": 0.6227307923943904,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009624374859876117,
      "loss": 1.0063,
      "step": 7238
    },
    {
      "epoch": 0.6228168287017122,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009624272024948199,
      "loss": 1.1008,
      "step": 7239
    },
    {
      "epoch": 0.6229028650090338,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009624169176495184,
      "loss": 1.9362,
      "step": 7240
    },
    {
      "epoch": 0.6229889013163555,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009624066314517374,
      "loss": 1.633,
      "step": 7241
    },
    {
      "epoch": 0.6230749376236772,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009623963439015069,
      "loss": 1.3263,
      "step": 7242
    },
    {
      "epoch": 0.6231609739309989,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000962386054998857,
      "loss": 1.3017,
      "step": 7243
    },
    {
      "epoch": 0.6232470102383205,
      "grad_norm": 1.25,
      "learning_rate": 0.0009623757647438179,
      "loss": 1.0216,
      "step": 7244
    },
    {
      "epoch": 0.6233330465456423,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009623654731364195,
      "loss": 1.7179,
      "step": 7245
    },
    {
      "epoch": 0.623419082852964,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009623551801766921,
      "loss": 0.8728,
      "step": 7246
    },
    {
      "epoch": 0.6235051191602856,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009623448858646658,
      "loss": 1.3559,
      "step": 7247
    },
    {
      "epoch": 0.6235911554676073,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009623345902003704,
      "loss": 1.9448,
      "step": 7248
    },
    {
      "epoch": 0.623677191774929,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009623242931838364,
      "loss": 0.7106,
      "step": 7249
    },
    {
      "epoch": 0.6237632280822507,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009623139948150937,
      "loss": 1.4236,
      "step": 7250
    },
    {
      "epoch": 0.6238492643895724,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009623036950941726,
      "loss": 1.2726,
      "step": 7251
    },
    {
      "epoch": 0.6239353006968941,
      "grad_norm": 1.265625,
      "learning_rate": 0.000962293394021103,
      "loss": 0.9816,
      "step": 7252
    },
    {
      "epoch": 0.6240213370042158,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009622830915959151,
      "loss": 1.2349,
      "step": 7253
    },
    {
      "epoch": 0.6241073733115374,
      "grad_norm": 1.3125,
      "learning_rate": 0.000962272787818639,
      "loss": 1.2204,
      "step": 7254
    },
    {
      "epoch": 0.6241934096188592,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009622624826893052,
      "loss": 0.9341,
      "step": 7255
    },
    {
      "epoch": 0.6242794459261809,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009622521762079432,
      "loss": 1.3617,
      "step": 7256
    },
    {
      "epoch": 0.6243654822335025,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009622418683745837,
      "loss": 1.2314,
      "step": 7257
    },
    {
      "epoch": 0.6244515185408243,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009622315591892566,
      "loss": 1.392,
      "step": 7258
    },
    {
      "epoch": 0.6245375548481459,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009622212486519921,
      "loss": 1.4174,
      "step": 7259
    },
    {
      "epoch": 0.6246235911554676,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009622109367628202,
      "loss": 1.0169,
      "step": 7260
    },
    {
      "epoch": 0.6247096274627894,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009622006235217714,
      "loss": 1.8498,
      "step": 7261
    },
    {
      "epoch": 0.624795663770111,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009621903089288756,
      "loss": 1.1374,
      "step": 7262
    },
    {
      "epoch": 0.6248817000774327,
      "grad_norm": 25.375,
      "learning_rate": 0.0009621799929841631,
      "loss": 1.7012,
      "step": 7263
    },
    {
      "epoch": 0.6249677363847543,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009621696756876639,
      "loss": 1.319,
      "step": 7264
    },
    {
      "epoch": 0.6250537726920761,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009621593570394083,
      "loss": 1.2406,
      "step": 7265
    },
    {
      "epoch": 0.6251398089993977,
      "grad_norm": 1.375,
      "learning_rate": 0.0009621490370394265,
      "loss": 1.8747,
      "step": 7266
    },
    {
      "epoch": 0.6252258453067194,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009621387156877486,
      "loss": 1.2751,
      "step": 7267
    },
    {
      "epoch": 0.6253118816140412,
      "grad_norm": 1.5,
      "learning_rate": 0.0009621283929844049,
      "loss": 1.7894,
      "step": 7268
    },
    {
      "epoch": 0.6253979179213628,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009621180689294255,
      "loss": 1.0485,
      "step": 7269
    },
    {
      "epoch": 0.6254839542286845,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009621077435228406,
      "loss": 1.3638,
      "step": 7270
    },
    {
      "epoch": 0.6255699905360061,
      "grad_norm": 1.5,
      "learning_rate": 0.0009620974167646806,
      "loss": 1.9662,
      "step": 7271
    },
    {
      "epoch": 0.6256560268433279,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009620870886549753,
      "loss": 0.76,
      "step": 7272
    },
    {
      "epoch": 0.6257420631506496,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009620767591937551,
      "loss": 1.3677,
      "step": 7273
    },
    {
      "epoch": 0.6258280994579712,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009620664283810505,
      "loss": 0.9934,
      "step": 7274
    },
    {
      "epoch": 0.625914135765293,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009620560962168913,
      "loss": 1.4124,
      "step": 7275
    },
    {
      "epoch": 0.6260001720726146,
      "grad_norm": 2.140625,
      "learning_rate": 0.000962045762701308,
      "loss": 1.7754,
      "step": 7276
    },
    {
      "epoch": 0.6260862083799363,
      "grad_norm": 1.125,
      "learning_rate": 0.0009620354278343305,
      "loss": 1.3602,
      "step": 7277
    },
    {
      "epoch": 0.6261722446872581,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009620250916159894,
      "loss": 1.1509,
      "step": 7278
    },
    {
      "epoch": 0.6262582809945797,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009620147540463146,
      "loss": 1.3595,
      "step": 7279
    },
    {
      "epoch": 0.6263443173019014,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009620044151253367,
      "loss": 1.0912,
      "step": 7280
    },
    {
      "epoch": 0.6264303536092231,
      "grad_norm": 1.125,
      "learning_rate": 0.0009619940748530857,
      "loss": 1.3751,
      "step": 7281
    },
    {
      "epoch": 0.6265163899165448,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009619837332295919,
      "loss": 1.8024,
      "step": 7282
    },
    {
      "epoch": 0.6266024262238665,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009619733902548855,
      "loss": 1.4197,
      "step": 7283
    },
    {
      "epoch": 0.6266884625311882,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009619630459289967,
      "loss": 1.5669,
      "step": 7284
    },
    {
      "epoch": 0.6267744988385099,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009619527002519559,
      "loss": 0.9984,
      "step": 7285
    },
    {
      "epoch": 0.6268605351458315,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009619423532237933,
      "loss": 1.6503,
      "step": 7286
    },
    {
      "epoch": 0.6269465714531532,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009619320048445391,
      "loss": 1.0898,
      "step": 7287
    },
    {
      "epoch": 0.6270326077604749,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009619216551142238,
      "loss": 0.9937,
      "step": 7288
    },
    {
      "epoch": 0.6271186440677966,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009619113040328773,
      "loss": 2.1097,
      "step": 7289
    },
    {
      "epoch": 0.6272046803751183,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009619009516005302,
      "loss": 2.0519,
      "step": 7290
    },
    {
      "epoch": 0.62729071668244,
      "grad_norm": 1.25,
      "learning_rate": 0.0009618905978172126,
      "loss": 2.0139,
      "step": 7291
    },
    {
      "epoch": 0.6273767529897617,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009618802426829548,
      "loss": 1.5216,
      "step": 7292
    },
    {
      "epoch": 0.6274627892970833,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009618698861977871,
      "loss": 1.4114,
      "step": 7293
    },
    {
      "epoch": 0.6275488256044051,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009618595283617399,
      "loss": 1.6513,
      "step": 7294
    },
    {
      "epoch": 0.6276348619117268,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009618491691748435,
      "loss": 1.7306,
      "step": 7295
    },
    {
      "epoch": 0.6277208982190484,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009618388086371279,
      "loss": 1.487,
      "step": 7296
    },
    {
      "epoch": 0.6278069345263702,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009618284467486237,
      "loss": 1.3355,
      "step": 7297
    },
    {
      "epoch": 0.6278929708336918,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000961818083509361,
      "loss": 1.1705,
      "step": 7298
    },
    {
      "epoch": 0.6279790071410135,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009618077189193704,
      "loss": 1.4332,
      "step": 7299
    },
    {
      "epoch": 0.6280650434483352,
      "grad_norm": 1.296875,
      "learning_rate": 0.000961797352978682,
      "loss": 1.987,
      "step": 7300
    },
    {
      "epoch": 0.6281510797556569,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000961786985687326,
      "loss": 1.192,
      "step": 7301
    },
    {
      "epoch": 0.6282371160629786,
      "grad_norm": 1.296875,
      "learning_rate": 0.000961776617045333,
      "loss": 1.5546,
      "step": 7302
    },
    {
      "epoch": 0.6283231523703002,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009617662470527333,
      "loss": 1.3542,
      "step": 7303
    },
    {
      "epoch": 0.628409188677622,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009617558757095571,
      "loss": 1.4365,
      "step": 7304
    },
    {
      "epoch": 0.6284952249849437,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009617455030158347,
      "loss": 2.106,
      "step": 7305
    },
    {
      "epoch": 0.6285812612922653,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009617351289715966,
      "loss": 1.0613,
      "step": 7306
    },
    {
      "epoch": 0.6286672975995871,
      "grad_norm": 1.9453125,
      "learning_rate": 0.000961724753576873,
      "loss": 1.3953,
      "step": 7307
    },
    {
      "epoch": 0.6287533339069087,
      "grad_norm": 2.265625,
      "learning_rate": 0.0009617143768316941,
      "loss": 1.1705,
      "step": 7308
    },
    {
      "epoch": 0.6288393702142304,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009617039987360907,
      "loss": 1.3495,
      "step": 7309
    },
    {
      "epoch": 0.628925406521552,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009616936192900928,
      "loss": 1.3705,
      "step": 7310
    },
    {
      "epoch": 0.6290114428288738,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009616832384937308,
      "loss": 1.4162,
      "step": 7311
    },
    {
      "epoch": 0.6290974791361955,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009616728563470353,
      "loss": 1.4439,
      "step": 7312
    },
    {
      "epoch": 0.6291835154435171,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009616624728500363,
      "loss": 1.2983,
      "step": 7313
    },
    {
      "epoch": 0.6292695517508389,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009616520880027645,
      "loss": 0.9615,
      "step": 7314
    },
    {
      "epoch": 0.6293555880581605,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009616417018052501,
      "loss": 0.9425,
      "step": 7315
    },
    {
      "epoch": 0.6294416243654822,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009616313142575234,
      "loss": 1.9343,
      "step": 7316
    },
    {
      "epoch": 0.629527660672804,
      "grad_norm": 2.59375,
      "learning_rate": 0.0009616209253596149,
      "loss": 2.1342,
      "step": 7317
    },
    {
      "epoch": 0.6296136969801256,
      "grad_norm": 1.125,
      "learning_rate": 0.0009616105351115551,
      "loss": 0.7353,
      "step": 7318
    },
    {
      "epoch": 0.6296997332874473,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009616001435133742,
      "loss": 1.6124,
      "step": 7319
    },
    {
      "epoch": 0.629785769594769,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009615897505651027,
      "loss": 1.9867,
      "step": 7320
    },
    {
      "epoch": 0.6298718059020907,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009615793562667708,
      "loss": 1.4903,
      "step": 7321
    },
    {
      "epoch": 0.6299578422094124,
      "grad_norm": 2.0,
      "learning_rate": 0.0009615689606184092,
      "loss": 1.7538,
      "step": 7322
    },
    {
      "epoch": 0.630043878516734,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009615585636200481,
      "loss": 0.9781,
      "step": 7323
    },
    {
      "epoch": 0.6301299148240558,
      "grad_norm": 1.09375,
      "learning_rate": 0.000961548165271718,
      "loss": 1.3617,
      "step": 7324
    },
    {
      "epoch": 0.6302159511313774,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009615377655734493,
      "loss": 1.1771,
      "step": 7325
    },
    {
      "epoch": 0.6303019874386991,
      "grad_norm": 1.375,
      "learning_rate": 0.0009615273645252725,
      "loss": 1.6976,
      "step": 7326
    },
    {
      "epoch": 0.6303880237460209,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009615169621272178,
      "loss": 1.7119,
      "step": 7327
    },
    {
      "epoch": 0.6304740600533425,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009615065583793157,
      "loss": 1.3265,
      "step": 7328
    },
    {
      "epoch": 0.6305600963606642,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009614961532815967,
      "loss": 1.1577,
      "step": 7329
    },
    {
      "epoch": 0.6306461326679859,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009614857468340912,
      "loss": 1.7346,
      "step": 7330
    },
    {
      "epoch": 0.6307321689753076,
      "grad_norm": 1.375,
      "learning_rate": 0.0009614753390368296,
      "loss": 2.006,
      "step": 7331
    },
    {
      "epoch": 0.6308182052826292,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009614649298898425,
      "loss": 1.4395,
      "step": 7332
    },
    {
      "epoch": 0.630904241589951,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009614545193931601,
      "loss": 1.5349,
      "step": 7333
    },
    {
      "epoch": 0.6309902778972727,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009614441075468131,
      "loss": 1.1738,
      "step": 7334
    },
    {
      "epoch": 0.6310763142045943,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009614336943508319,
      "loss": 1.1227,
      "step": 7335
    },
    {
      "epoch": 0.631162350511916,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009614232798052467,
      "loss": 1.2485,
      "step": 7336
    },
    {
      "epoch": 0.6312483868192377,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009614128639100882,
      "loss": 1.5422,
      "step": 7337
    },
    {
      "epoch": 0.6313344231265594,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009614024466653867,
      "loss": 1.0682,
      "step": 7338
    },
    {
      "epoch": 0.6314204594338811,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009613920280711731,
      "loss": 1.5457,
      "step": 7339
    },
    {
      "epoch": 0.6315064957412028,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009613816081274773,
      "loss": 0.8927,
      "step": 7340
    },
    {
      "epoch": 0.6315925320485245,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009613711868343303,
      "loss": 2.2002,
      "step": 7341
    },
    {
      "epoch": 0.6316785683558461,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000961360764191762,
      "loss": 0.6251,
      "step": 7342
    },
    {
      "epoch": 0.6317646046631679,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009613503401998033,
      "loss": 1.6267,
      "step": 7343
    },
    {
      "epoch": 0.6318506409704896,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009613399148584846,
      "loss": 1.3559,
      "step": 7344
    },
    {
      "epoch": 0.6319366772778112,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009613294881678364,
      "loss": 1.1367,
      "step": 7345
    },
    {
      "epoch": 0.632022713585133,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009613190601278892,
      "loss": 1.3527,
      "step": 7346
    },
    {
      "epoch": 0.6321087498924546,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009613086307386733,
      "loss": 1.1246,
      "step": 7347
    },
    {
      "epoch": 0.6321947861997763,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009612982000002195,
      "loss": 1.2129,
      "step": 7348
    },
    {
      "epoch": 0.632280822507098,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009612877679125582,
      "loss": 1.5654,
      "step": 7349
    },
    {
      "epoch": 0.6323668588144197,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009612773344757199,
      "loss": 1.6318,
      "step": 7350
    },
    {
      "epoch": 0.6324528951217414,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000961266899689735,
      "loss": 0.8661,
      "step": 7351
    },
    {
      "epoch": 0.632538931429063,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009612564635546341,
      "loss": 1.3577,
      "step": 7352
    },
    {
      "epoch": 0.6326249677363848,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000961246026070448,
      "loss": 0.8953,
      "step": 7353
    },
    {
      "epoch": 0.6327110040437064,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009612355872372067,
      "loss": 1.2896,
      "step": 7354
    },
    {
      "epoch": 0.6327970403510281,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009612251470549413,
      "loss": 1.1099,
      "step": 7355
    },
    {
      "epoch": 0.6328830766583499,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009612147055236819,
      "loss": 1.6184,
      "step": 7356
    },
    {
      "epoch": 0.6329691129656715,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009612042626434592,
      "loss": 1.2541,
      "step": 7357
    },
    {
      "epoch": 0.6330551492729932,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009611938184143036,
      "loss": 0.6975,
      "step": 7358
    },
    {
      "epoch": 0.6331411855803148,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000961183372836246,
      "loss": 1.3597,
      "step": 7359
    },
    {
      "epoch": 0.6332272218876366,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009611729259093165,
      "loss": 0.8434,
      "step": 7360
    },
    {
      "epoch": 0.6333132581949583,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009611624776335461,
      "loss": 0.7453,
      "step": 7361
    },
    {
      "epoch": 0.6333992945022799,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000961152028008965,
      "loss": 0.9533,
      "step": 7362
    },
    {
      "epoch": 0.6334853308096017,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000961141577035604,
      "loss": 0.6594,
      "step": 7363
    },
    {
      "epoch": 0.6335713671169233,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009611311247134935,
      "loss": 1.526,
      "step": 7364
    },
    {
      "epoch": 0.633657403424245,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009611206710426642,
      "loss": 1.7707,
      "step": 7365
    },
    {
      "epoch": 0.6337434397315668,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009611102160231467,
      "loss": 0.9472,
      "step": 7366
    },
    {
      "epoch": 0.6338294760388884,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009610997596549713,
      "loss": 1.4814,
      "step": 7367
    },
    {
      "epoch": 0.6339155123462101,
      "grad_norm": 1.453125,
      "learning_rate": 0.000961089301938169,
      "loss": 1.3044,
      "step": 7368
    },
    {
      "epoch": 0.6340015486535318,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009610788428727699,
      "loss": 1.238,
      "step": 7369
    },
    {
      "epoch": 0.6340875849608535,
      "grad_norm": 1.671875,
      "learning_rate": 0.000961068382458805,
      "loss": 1.1621,
      "step": 7370
    },
    {
      "epoch": 0.6341736212681752,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009610579206963049,
      "loss": 1.3374,
      "step": 7371
    },
    {
      "epoch": 0.6342596575754968,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009610474575852999,
      "loss": 1.0382,
      "step": 7372
    },
    {
      "epoch": 0.6343456938828186,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009610369931258207,
      "loss": 1.433,
      "step": 7373
    },
    {
      "epoch": 0.6344317301901402,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009610265273178979,
      "loss": 1.0997,
      "step": 7374
    },
    {
      "epoch": 0.6345177664974619,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009610160601615624,
      "loss": 1.6414,
      "step": 7375
    },
    {
      "epoch": 0.6346038028047836,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009610055916568445,
      "loss": 0.9663,
      "step": 7376
    },
    {
      "epoch": 0.6346898391121053,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009609951218037749,
      "loss": 2.26,
      "step": 7377
    },
    {
      "epoch": 0.634775875419427,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009609846506023843,
      "loss": 1.2988,
      "step": 7378
    },
    {
      "epoch": 0.6348619117267487,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009609741780527032,
      "loss": 2.1764,
      "step": 7379
    },
    {
      "epoch": 0.6349479480340704,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009609637041547622,
      "loss": 0.8001,
      "step": 7380
    },
    {
      "epoch": 0.635033984341392,
      "grad_norm": 1.1875,
      "learning_rate": 0.000960953228908592,
      "loss": 1.6121,
      "step": 7381
    },
    {
      "epoch": 0.6351200206487138,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009609427523142234,
      "loss": 1.1513,
      "step": 7382
    },
    {
      "epoch": 0.6352060569560355,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009609322743716869,
      "loss": 1.3514,
      "step": 7383
    },
    {
      "epoch": 0.6352920932633571,
      "grad_norm": 1.03125,
      "learning_rate": 0.000960921795081013,
      "loss": 1.5743,
      "step": 7384
    },
    {
      "epoch": 0.6353781295706789,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009609113144422326,
      "loss": 1.1873,
      "step": 7385
    },
    {
      "epoch": 0.6354641658780005,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009609008324553762,
      "loss": 0.6739,
      "step": 7386
    },
    {
      "epoch": 0.6355502021853222,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009608903491204746,
      "loss": 2.1827,
      "step": 7387
    },
    {
      "epoch": 0.635636238492644,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009608798644375583,
      "loss": 1.5183,
      "step": 7388
    },
    {
      "epoch": 0.6357222747999656,
      "grad_norm": 1.265625,
      "learning_rate": 0.000960869378406658,
      "loss": 1.611,
      "step": 7389
    },
    {
      "epoch": 0.6358083111072873,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009608588910278045,
      "loss": 1.4631,
      "step": 7390
    },
    {
      "epoch": 0.6358943474146089,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009608484023010283,
      "loss": 1.6646,
      "step": 7391
    },
    {
      "epoch": 0.6359803837219307,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00096083791222636,
      "loss": 1.7649,
      "step": 7392
    },
    {
      "epoch": 0.6360664200292524,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009608274208038306,
      "loss": 2.0304,
      "step": 7393
    },
    {
      "epoch": 0.636152456336574,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009608169280334707,
      "loss": 2.1281,
      "step": 7394
    },
    {
      "epoch": 0.6362384926438958,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009608064339153107,
      "loss": 1.9266,
      "step": 7395
    },
    {
      "epoch": 0.6363245289512174,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009607959384493816,
      "loss": 1.3078,
      "step": 7396
    },
    {
      "epoch": 0.6364105652585391,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000960785441635714,
      "loss": 1.5661,
      "step": 7397
    },
    {
      "epoch": 0.6364966015658607,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009607749434743387,
      "loss": 1.4093,
      "step": 7398
    },
    {
      "epoch": 0.6365826378731825,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009607644439652863,
      "loss": 1.487,
      "step": 7399
    },
    {
      "epoch": 0.6366686741805042,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009607539431085874,
      "loss": 2.137,
      "step": 7400
    },
    {
      "epoch": 0.6367547104878258,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009607434409042727,
      "loss": 1.2335,
      "step": 7401
    },
    {
      "epoch": 0.6368407467951476,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009607329373523734,
      "loss": 1.2356,
      "step": 7402
    },
    {
      "epoch": 0.6369267831024692,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009607224324529196,
      "loss": 1.2538,
      "step": 7403
    },
    {
      "epoch": 0.6370128194097909,
      "grad_norm": 1.0,
      "learning_rate": 0.0009607119262059425,
      "loss": 0.9795,
      "step": 7404
    },
    {
      "epoch": 0.6370988557171127,
      "grad_norm": 1.25,
      "learning_rate": 0.0009607014186114724,
      "loss": 1.0249,
      "step": 7405
    },
    {
      "epoch": 0.6371848920244343,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009606909096695405,
      "loss": 0.8187,
      "step": 7406
    },
    {
      "epoch": 0.637270928331756,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009606803993801769,
      "loss": 0.8197,
      "step": 7407
    },
    {
      "epoch": 0.6373569646390776,
      "grad_norm": 1.109375,
      "learning_rate": 0.000960669887743413,
      "loss": 0.864,
      "step": 7408
    },
    {
      "epoch": 0.6374430009463994,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009606593747592794,
      "loss": 1.2099,
      "step": 7409
    },
    {
      "epoch": 0.6375290372537211,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009606488604278065,
      "loss": 1.7247,
      "step": 7410
    },
    {
      "epoch": 0.6376150735610427,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009606383447490252,
      "loss": 1.0131,
      "step": 7411
    },
    {
      "epoch": 0.6377011098683645,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009606278277229665,
      "loss": 1.4168,
      "step": 7412
    },
    {
      "epoch": 0.6377871461756861,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009606173093496611,
      "loss": 2.2072,
      "step": 7413
    },
    {
      "epoch": 0.6378731824830078,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009606067896291393,
      "loss": 1.8834,
      "step": 7414
    },
    {
      "epoch": 0.6379592187903296,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009605962685614325,
      "loss": 1.3511,
      "step": 7415
    },
    {
      "epoch": 0.6380452550976512,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009605857461465712,
      "loss": 0.8112,
      "step": 7416
    },
    {
      "epoch": 0.6381312914049729,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009605752223845861,
      "loss": 1.09,
      "step": 7417
    },
    {
      "epoch": 0.6382173277122946,
      "grad_norm": 1.125,
      "learning_rate": 0.000960564697275508,
      "loss": 1.5505,
      "step": 7418
    },
    {
      "epoch": 0.6383033640196163,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000960554170819368,
      "loss": 1.9846,
      "step": 7419
    },
    {
      "epoch": 0.6383894003269379,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009605436430161963,
      "loss": 1.1548,
      "step": 7420
    },
    {
      "epoch": 0.6384754366342597,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000960533113866024,
      "loss": 1.2613,
      "step": 7421
    },
    {
      "epoch": 0.6385614729415814,
      "grad_norm": 2.90625,
      "learning_rate": 0.0009605225833688821,
      "loss": 1.6392,
      "step": 7422
    },
    {
      "epoch": 0.638647509248903,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009605120515248012,
      "loss": 1.2894,
      "step": 7423
    },
    {
      "epoch": 0.6387335455562247,
      "grad_norm": 1.21875,
      "learning_rate": 0.000960501518333812,
      "loss": 0.9992,
      "step": 7424
    },
    {
      "epoch": 0.6388195818635464,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009604909837959455,
      "loss": 0.9631,
      "step": 7425
    },
    {
      "epoch": 0.6389056181708681,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009604804479112323,
      "loss": 1.1777,
      "step": 7426
    },
    {
      "epoch": 0.6389916544781898,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009604699106797036,
      "loss": 1.2845,
      "step": 7427
    },
    {
      "epoch": 0.6390776907855115,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009604593721013898,
      "loss": 1.2979,
      "step": 7428
    },
    {
      "epoch": 0.6391637270928332,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009604488321763219,
      "loss": 1.5892,
      "step": 7429
    },
    {
      "epoch": 0.6392497634001548,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009604382909045308,
      "loss": 1.6607,
      "step": 7430
    },
    {
      "epoch": 0.6393357997074766,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009604277482860471,
      "loss": 0.8803,
      "step": 7431
    },
    {
      "epoch": 0.6394218360147983,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009604172043209019,
      "loss": 1.4845,
      "step": 7432
    },
    {
      "epoch": 0.6395078723221199,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009604066590091259,
      "loss": 1.6288,
      "step": 7433
    },
    {
      "epoch": 0.6395939086294417,
      "grad_norm": 1.109375,
      "learning_rate": 0.00096039611235075,
      "loss": 1.9025,
      "step": 7434
    },
    {
      "epoch": 0.6396799449367633,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000960385564345805,
      "loss": 0.8274,
      "step": 7435
    },
    {
      "epoch": 0.639765981244085,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009603750149943218,
      "loss": 1.7634,
      "step": 7436
    },
    {
      "epoch": 0.6398520175514067,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009603644642963311,
      "loss": 2.9661,
      "step": 7437
    },
    {
      "epoch": 0.6399380538587284,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009603539122518638,
      "loss": 1.3817,
      "step": 7438
    },
    {
      "epoch": 0.6400240901660501,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009603433588609512,
      "loss": 1.4057,
      "step": 7439
    },
    {
      "epoch": 0.6401101264733717,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009603328041236234,
      "loss": 0.8793,
      "step": 7440
    },
    {
      "epoch": 0.6401961627806935,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009603222480399119,
      "loss": 1.6866,
      "step": 7441
    },
    {
      "epoch": 0.6402821990880151,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009603116906098473,
      "loss": 1.5195,
      "step": 7442
    },
    {
      "epoch": 0.6403682353953368,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009603011318334606,
      "loss": 0.9641,
      "step": 7443
    },
    {
      "epoch": 0.6404542717026586,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009602905717107826,
      "loss": 1.9688,
      "step": 7444
    },
    {
      "epoch": 0.6405403080099802,
      "grad_norm": 1.5,
      "learning_rate": 0.000960280010241844,
      "loss": 1.9681,
      "step": 7445
    },
    {
      "epoch": 0.6406263443173019,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009602694474266763,
      "loss": 2.0149,
      "step": 7446
    },
    {
      "epoch": 0.6407123806246235,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009602588832653096,
      "loss": 1.7346,
      "step": 7447
    },
    {
      "epoch": 0.6407984169319453,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009602483177577753,
      "loss": 1.7435,
      "step": 7448
    },
    {
      "epoch": 0.640884453239267,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009602377509041043,
      "loss": 0.8611,
      "step": 7449
    },
    {
      "epoch": 0.6409704895465886,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009602271827043272,
      "loss": 1.3484,
      "step": 7450
    },
    {
      "epoch": 0.6410565258539104,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009602166131584752,
      "loss": 1.5812,
      "step": 7451
    },
    {
      "epoch": 0.641142562161232,
      "grad_norm": 1.328125,
      "learning_rate": 0.000960206042266579,
      "loss": 0.7986,
      "step": 7452
    },
    {
      "epoch": 0.6412285984685537,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009601954700286698,
      "loss": 1.6515,
      "step": 7453
    },
    {
      "epoch": 0.6413146347758755,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009601848964447783,
      "loss": 0.8639,
      "step": 7454
    },
    {
      "epoch": 0.6414006710831971,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009601743215149355,
      "loss": 1.6963,
      "step": 7455
    },
    {
      "epoch": 0.6414867073905188,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009601637452391722,
      "loss": 1.5469,
      "step": 7456
    },
    {
      "epoch": 0.6415727436978405,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009601531676175196,
      "loss": 1.2621,
      "step": 7457
    },
    {
      "epoch": 0.6416587800051622,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009601425886500084,
      "loss": 1.7404,
      "step": 7458
    },
    {
      "epoch": 0.6417448163124839,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009601320083366696,
      "loss": 0.9644,
      "step": 7459
    },
    {
      "epoch": 0.6418308526198055,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009601214266775342,
      "loss": 1.1252,
      "step": 7460
    },
    {
      "epoch": 0.6419168889271273,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009601108436726331,
      "loss": 0.9623,
      "step": 7461
    },
    {
      "epoch": 0.6420029252344489,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009601002593219972,
      "loss": 2.6052,
      "step": 7462
    },
    {
      "epoch": 0.6420889615417706,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009600896736256577,
      "loss": 1.31,
      "step": 7463
    },
    {
      "epoch": 0.6421749978490923,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009600790865836453,
      "loss": 1.7554,
      "step": 7464
    },
    {
      "epoch": 0.642261034156414,
      "grad_norm": 1.203125,
      "learning_rate": 0.000960068498195991,
      "loss": 1.3154,
      "step": 7465
    },
    {
      "epoch": 0.6423470704637357,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009600579084627258,
      "loss": 1.6648,
      "step": 7466
    },
    {
      "epoch": 0.6424331067710574,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009600473173838807,
      "loss": 0.9471,
      "step": 7467
    },
    {
      "epoch": 0.6425191430783791,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009600367249594868,
      "loss": 1.3854,
      "step": 7468
    },
    {
      "epoch": 0.6426051793857007,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009600261311895748,
      "loss": 0.9737,
      "step": 7469
    },
    {
      "epoch": 0.6426912156930225,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009600155360741759,
      "loss": 1.1541,
      "step": 7470
    },
    {
      "epoch": 0.6427772520003442,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000960004939613321,
      "loss": 1.0636,
      "step": 7471
    },
    {
      "epoch": 0.6428632883076658,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009599943418070412,
      "loss": 1.717,
      "step": 7472
    },
    {
      "epoch": 0.6429493246149875,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009599837426553674,
      "loss": 1.1794,
      "step": 7473
    },
    {
      "epoch": 0.6430353609223092,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009599731421583306,
      "loss": 1.4785,
      "step": 7474
    },
    {
      "epoch": 0.6431213972296309,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009599625403159618,
      "loss": 1.8039,
      "step": 7475
    },
    {
      "epoch": 0.6432074335369526,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000959951937128292,
      "loss": 1.026,
      "step": 7476
    },
    {
      "epoch": 0.6432934698442743,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009599413325953524,
      "loss": 1.0587,
      "step": 7477
    },
    {
      "epoch": 0.643379506151596,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009599307267171735,
      "loss": 1.3346,
      "step": 7478
    },
    {
      "epoch": 0.6434655424589176,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000959920119493787,
      "loss": 1.3001,
      "step": 7479
    },
    {
      "epoch": 0.6435515787662394,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009599095109252234,
      "loss": 0.5716,
      "step": 7480
    },
    {
      "epoch": 0.6436376150735611,
      "grad_norm": 1.203125,
      "learning_rate": 0.000959898901011514,
      "loss": 1.1808,
      "step": 7481
    },
    {
      "epoch": 0.6437236513808827,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009598882897526899,
      "loss": 1.4233,
      "step": 7482
    },
    {
      "epoch": 0.6438096876882045,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009598776771487819,
      "loss": 0.8992,
      "step": 7483
    },
    {
      "epoch": 0.6438957239955261,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009598670631998211,
      "loss": 1.1981,
      "step": 7484
    },
    {
      "epoch": 0.6439817603028478,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009598564479058384,
      "loss": 2.1927,
      "step": 7485
    },
    {
      "epoch": 0.6440677966101694,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009598458312668654,
      "loss": 1.5326,
      "step": 7486
    },
    {
      "epoch": 0.6441538329174912,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009598352132829326,
      "loss": 1.1981,
      "step": 7487
    },
    {
      "epoch": 0.6442398692248129,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009598245939540713,
      "loss": 1.2313,
      "step": 7488
    },
    {
      "epoch": 0.6443259055321345,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009598139732803124,
      "loss": 0.8544,
      "step": 7489
    },
    {
      "epoch": 0.6444119418394563,
      "grad_norm": 1.78125,
      "learning_rate": 0.000959803351261687,
      "loss": 1.5288,
      "step": 7490
    },
    {
      "epoch": 0.6444979781467779,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009597927278982264,
      "loss": 0.6952,
      "step": 7491
    },
    {
      "epoch": 0.6445840144540996,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009597821031899614,
      "loss": 1.142,
      "step": 7492
    },
    {
      "epoch": 0.6446700507614214,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009597714771369232,
      "loss": 1.004,
      "step": 7493
    },
    {
      "epoch": 0.644756087068743,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009597608497391428,
      "loss": 1.3171,
      "step": 7494
    },
    {
      "epoch": 0.6448421233760647,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009597502209966513,
      "loss": 1.2169,
      "step": 7495
    },
    {
      "epoch": 0.6449281596833863,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009597395909094799,
      "loss": 1.0051,
      "step": 7496
    },
    {
      "epoch": 0.6450141959907081,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009597289594776596,
      "loss": 2.3575,
      "step": 7497
    },
    {
      "epoch": 0.6451002322980298,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009597183267012216,
      "loss": 0.8971,
      "step": 7498
    },
    {
      "epoch": 0.6451862686053514,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009597076925801966,
      "loss": 1.9719,
      "step": 7499
    },
    {
      "epoch": 0.6452723049126732,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009596970571146163,
      "loss": 1.3431,
      "step": 7500
    },
    {
      "epoch": 0.6453583412199948,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009596864203045114,
      "loss": 1.426,
      "step": 7501
    },
    {
      "epoch": 0.6454443775273165,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009596757821499131,
      "loss": 1.5814,
      "step": 7502
    },
    {
      "epoch": 0.6455304138346383,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009596651426508527,
      "loss": 1.4601,
      "step": 7503
    },
    {
      "epoch": 0.6456164501419599,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009596545018073611,
      "loss": 1.076,
      "step": 7504
    },
    {
      "epoch": 0.6457024864492816,
      "grad_norm": 5.21875,
      "learning_rate": 0.0009596438596194693,
      "loss": 1.0748,
      "step": 7505
    },
    {
      "epoch": 0.6457885227566033,
      "grad_norm": 3.234375,
      "learning_rate": 0.0009596332160872088,
      "loss": 1.6021,
      "step": 7506
    },
    {
      "epoch": 0.645874559063925,
      "grad_norm": 1.25,
      "learning_rate": 0.0009596225712106103,
      "loss": 1.6132,
      "step": 7507
    },
    {
      "epoch": 0.6459605953712466,
      "grad_norm": 2.78125,
      "learning_rate": 0.0009596119249897055,
      "loss": 1.0805,
      "step": 7508
    },
    {
      "epoch": 0.6460466316785684,
      "grad_norm": 1.421875,
      "learning_rate": 0.000959601277424525,
      "loss": 1.7205,
      "step": 7509
    },
    {
      "epoch": 0.6461326679858901,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009595906285151002,
      "loss": 1.4723,
      "step": 7510
    },
    {
      "epoch": 0.6462187042932117,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009595799782614623,
      "loss": 1.044,
      "step": 7511
    },
    {
      "epoch": 0.6463047406005334,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009595693266636421,
      "loss": 1.0012,
      "step": 7512
    },
    {
      "epoch": 0.6463907769078551,
      "grad_norm": 2.265625,
      "learning_rate": 0.0009595586737216711,
      "loss": 1.0396,
      "step": 7513
    },
    {
      "epoch": 0.6464768132151768,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009595480194355805,
      "loss": 1.5778,
      "step": 7514
    },
    {
      "epoch": 0.6465628495224985,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009595373638054013,
      "loss": 1.1848,
      "step": 7515
    },
    {
      "epoch": 0.6466488858298202,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009595267068311645,
      "loss": 1.0373,
      "step": 7516
    },
    {
      "epoch": 0.6467349221371419,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009595160485129015,
      "loss": 0.8694,
      "step": 7517
    },
    {
      "epoch": 0.6468209584444635,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009595053888506436,
      "loss": 1.0097,
      "step": 7518
    },
    {
      "epoch": 0.6469069947517853,
      "grad_norm": 1.375,
      "learning_rate": 0.0009594947278444217,
      "loss": 1.4151,
      "step": 7519
    },
    {
      "epoch": 0.646993031059107,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000959484065494267,
      "loss": 1.4236,
      "step": 7520
    },
    {
      "epoch": 0.6470790673664286,
      "grad_norm": 1.875,
      "learning_rate": 0.0009594734018002109,
      "loss": 1.742,
      "step": 7521
    },
    {
      "epoch": 0.6471651036737504,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009594627367622843,
      "loss": 0.8226,
      "step": 7522
    },
    {
      "epoch": 0.647251139981072,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009594520703805188,
      "loss": 1.4772,
      "step": 7523
    },
    {
      "epoch": 0.6473371762883937,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009594414026549452,
      "loss": 0.6868,
      "step": 7524
    },
    {
      "epoch": 0.6474232125957154,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000959430733585595,
      "loss": 1.1976,
      "step": 7525
    },
    {
      "epoch": 0.6475092489030371,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009594200631724989,
      "loss": 2.0581,
      "step": 7526
    },
    {
      "epoch": 0.6475952852103588,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009594093914156889,
      "loss": 1.6005,
      "step": 7527
    },
    {
      "epoch": 0.6476813215176804,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009593987183151955,
      "loss": 1.5757,
      "step": 7528
    },
    {
      "epoch": 0.6477673578250022,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009593880438710503,
      "loss": 1.2315,
      "step": 7529
    },
    {
      "epoch": 0.6478533941323238,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009593773680832844,
      "loss": 1.8231,
      "step": 7530
    },
    {
      "epoch": 0.6479394304396455,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000959366690951929,
      "loss": 0.9492,
      "step": 7531
    },
    {
      "epoch": 0.6480254667469673,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009593560124770156,
      "loss": 1.1107,
      "step": 7532
    },
    {
      "epoch": 0.6481115030542889,
      "grad_norm": 1.265625,
      "learning_rate": 0.000959345332658575,
      "loss": 0.9932,
      "step": 7533
    },
    {
      "epoch": 0.6481975393616106,
      "grad_norm": 2.25,
      "learning_rate": 0.0009593346514966386,
      "loss": 2.0101,
      "step": 7534
    },
    {
      "epoch": 0.6482835756689322,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009593239689912376,
      "loss": 1.0581,
      "step": 7535
    },
    {
      "epoch": 0.648369611976254,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009593132851424036,
      "loss": 1.358,
      "step": 7536
    },
    {
      "epoch": 0.6484556482835757,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009593025999501674,
      "loss": 1.2944,
      "step": 7537
    },
    {
      "epoch": 0.6485416845908973,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009592919134145605,
      "loss": 1.1307,
      "step": 7538
    },
    {
      "epoch": 0.6486277208982191,
      "grad_norm": 1.234375,
      "learning_rate": 0.000959281225535614,
      "loss": 1.1392,
      "step": 7539
    },
    {
      "epoch": 0.6487137572055407,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009592705363133594,
      "loss": 1.5239,
      "step": 7540
    },
    {
      "epoch": 0.6487997935128624,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009592598457478276,
      "loss": 1.1187,
      "step": 7541
    },
    {
      "epoch": 0.6488858298201842,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009592491538390502,
      "loss": 1.3761,
      "step": 7542
    },
    {
      "epoch": 0.6489718661275058,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009592384605870584,
      "loss": 1.3663,
      "step": 7543
    },
    {
      "epoch": 0.6490579024348275,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009592277659918831,
      "loss": 1.74,
      "step": 7544
    },
    {
      "epoch": 0.6491439387421492,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009592170700535562,
      "loss": 1.4281,
      "step": 7545
    },
    {
      "epoch": 0.6492299750494709,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009592063727721085,
      "loss": 1.6098,
      "step": 7546
    },
    {
      "epoch": 0.6493160113567926,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009591956741475717,
      "loss": 1.451,
      "step": 7547
    },
    {
      "epoch": 0.6494020476641142,
      "grad_norm": 1.375,
      "learning_rate": 0.0009591849741799768,
      "loss": 1.8863,
      "step": 7548
    },
    {
      "epoch": 0.649488083971436,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009591742728693551,
      "loss": 1.2888,
      "step": 7549
    },
    {
      "epoch": 0.6495741202787576,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009591635702157379,
      "loss": 1.078,
      "step": 7550
    },
    {
      "epoch": 0.6496601565860793,
      "grad_norm": 1.375,
      "learning_rate": 0.0009591528662191566,
      "loss": 0.9337,
      "step": 7551
    },
    {
      "epoch": 0.649746192893401,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009591421608796425,
      "loss": 1.1505,
      "step": 7552
    },
    {
      "epoch": 0.6498322292007227,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009591314541972268,
      "loss": 1.7375,
      "step": 7553
    },
    {
      "epoch": 0.6499182655080444,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009591207461719409,
      "loss": 1.8291,
      "step": 7554
    },
    {
      "epoch": 0.6500043018153661,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009591100368038162,
      "loss": 1.4813,
      "step": 7555
    },
    {
      "epoch": 0.6500903381226878,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000959099326092884,
      "loss": 0.8024,
      "step": 7556
    },
    {
      "epoch": 0.6501763744300094,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009590886140391754,
      "loss": 1.301,
      "step": 7557
    },
    {
      "epoch": 0.6502624107373312,
      "grad_norm": 1.265625,
      "learning_rate": 0.000959077900642722,
      "loss": 1.8308,
      "step": 7558
    },
    {
      "epoch": 0.6503484470446529,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009590671859035549,
      "loss": 1.422,
      "step": 7559
    },
    {
      "epoch": 0.6504344833519745,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009590564698217056,
      "loss": 1.0624,
      "step": 7560
    },
    {
      "epoch": 0.6505205196592962,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009590457523972055,
      "loss": 1.856,
      "step": 7561
    },
    {
      "epoch": 0.6506065559666179,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0009590350336300857,
      "loss": 1.0574,
      "step": 7562
    },
    {
      "epoch": 0.6506925922739396,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009590243135203779,
      "loss": 1.3329,
      "step": 7563
    },
    {
      "epoch": 0.6507786285812613,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009590135920681132,
      "loss": 1.6844,
      "step": 7564
    },
    {
      "epoch": 0.650864664888583,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000959002869273323,
      "loss": 1.3811,
      "step": 7565
    },
    {
      "epoch": 0.6509507011959047,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009589921451360386,
      "loss": 1.3178,
      "step": 7566
    },
    {
      "epoch": 0.6510367375032263,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009589814196562914,
      "loss": 1.6061,
      "step": 7567
    },
    {
      "epoch": 0.6511227738105481,
      "grad_norm": 1.453125,
      "learning_rate": 0.000958970692834113,
      "loss": 1.2401,
      "step": 7568
    },
    {
      "epoch": 0.6512088101178698,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009589599646695344,
      "loss": 1.9373,
      "step": 7569
    },
    {
      "epoch": 0.6512948464251914,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009589492351625873,
      "loss": 1.2829,
      "step": 7570
    },
    {
      "epoch": 0.6513808827325132,
      "grad_norm": 1.484375,
      "learning_rate": 0.000958938504313303,
      "loss": 1.3069,
      "step": 7571
    },
    {
      "epoch": 0.6514669190398348,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009589277721217126,
      "loss": 1.3115,
      "step": 7572
    },
    {
      "epoch": 0.6515529553471565,
      "grad_norm": 1.609375,
      "learning_rate": 0.000958917038587848,
      "loss": 2.0722,
      "step": 7573
    },
    {
      "epoch": 0.6516389916544781,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009589063037117401,
      "loss": 1.5783,
      "step": 7574
    },
    {
      "epoch": 0.6517250279617999,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009588955674934206,
      "loss": 1.3857,
      "step": 7575
    },
    {
      "epoch": 0.6518110642691216,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009588848299329209,
      "loss": 1.2345,
      "step": 7576
    },
    {
      "epoch": 0.6518971005764432,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009588740910302723,
      "loss": 1.3459,
      "step": 7577
    },
    {
      "epoch": 0.651983136883765,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000958863350785506,
      "loss": 1.7321,
      "step": 7578
    },
    {
      "epoch": 0.6520691731910866,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009588526091986539,
      "loss": 1.9897,
      "step": 7579
    },
    {
      "epoch": 0.6521552094984083,
      "grad_norm": 1.25,
      "learning_rate": 0.0009588418662697471,
      "loss": 1.4711,
      "step": 7580
    },
    {
      "epoch": 0.6522412458057301,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000958831121998817,
      "loss": 1.4773,
      "step": 7581
    },
    {
      "epoch": 0.6523272821130517,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009588203763858951,
      "loss": 0.8008,
      "step": 7582
    },
    {
      "epoch": 0.6524133184203734,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009588096294310129,
      "loss": 0.5767,
      "step": 7583
    },
    {
      "epoch": 0.652499354727695,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009587988811342019,
      "loss": 0.8927,
      "step": 7584
    },
    {
      "epoch": 0.6525853910350168,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009587881314954931,
      "loss": 2.0612,
      "step": 7585
    },
    {
      "epoch": 0.6526714273423385,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009587773805149185,
      "loss": 1.6691,
      "step": 7586
    },
    {
      "epoch": 0.6527574636496601,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009587666281925092,
      "loss": 0.7226,
      "step": 7587
    },
    {
      "epoch": 0.6528434999569819,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009587558745282967,
      "loss": 1.9579,
      "step": 7588
    },
    {
      "epoch": 0.6529295362643035,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009587451195223126,
      "loss": 1.5653,
      "step": 7589
    },
    {
      "epoch": 0.6530155725716252,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009587343631745882,
      "loss": 1.034,
      "step": 7590
    },
    {
      "epoch": 0.653101608878947,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009587236054851549,
      "loss": 0.8315,
      "step": 7591
    },
    {
      "epoch": 0.6531876451862686,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009587128464540445,
      "loss": 1.8051,
      "step": 7592
    },
    {
      "epoch": 0.6532736814935903,
      "grad_norm": 1.421875,
      "learning_rate": 0.000958702086081288,
      "loss": 1.7403,
      "step": 7593
    },
    {
      "epoch": 0.653359717800912,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009586913243669172,
      "loss": 1.8738,
      "step": 7594
    },
    {
      "epoch": 0.6534457541082337,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009586805613109635,
      "loss": 1.8248,
      "step": 7595
    },
    {
      "epoch": 0.6535317904155553,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009586697969134584,
      "loss": 1.6625,
      "step": 7596
    },
    {
      "epoch": 0.653617826722877,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009586590311744332,
      "loss": 1.5746,
      "step": 7597
    },
    {
      "epoch": 0.6537038630301988,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009586482640939196,
      "loss": 1.3714,
      "step": 7598
    },
    {
      "epoch": 0.6537898993375204,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009586374956719492,
      "loss": 1.1464,
      "step": 7599
    },
    {
      "epoch": 0.6538759356448421,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009586267259085532,
      "loss": 1.2991,
      "step": 7600
    },
    {
      "epoch": 0.6539619719521638,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009586159548037632,
      "loss": 1.5244,
      "step": 7601
    },
    {
      "epoch": 0.6540480082594855,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009586051823576107,
      "loss": 1.3046,
      "step": 7602
    },
    {
      "epoch": 0.6541340445668072,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009585944085701273,
      "loss": 1.4273,
      "step": 7603
    },
    {
      "epoch": 0.6542200808741289,
      "grad_norm": 6.75,
      "learning_rate": 0.0009585836334413445,
      "loss": 1.355,
      "step": 7604
    },
    {
      "epoch": 0.6543061171814506,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009585728569712936,
      "loss": 1.7185,
      "step": 7605
    },
    {
      "epoch": 0.6543921534887722,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009585620791600064,
      "loss": 1.1896,
      "step": 7606
    },
    {
      "epoch": 0.654478189796094,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009585513000075144,
      "loss": 1.5677,
      "step": 7607
    },
    {
      "epoch": 0.6545642261034157,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009585405195138488,
      "loss": 0.8762,
      "step": 7608
    },
    {
      "epoch": 0.6546502624107373,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009585297376790414,
      "loss": 1.4747,
      "step": 7609
    },
    {
      "epoch": 0.654736298718059,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009585189545031239,
      "loss": 1.7423,
      "step": 7610
    },
    {
      "epoch": 0.6548223350253807,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009585081699861274,
      "loss": 1.4335,
      "step": 7611
    },
    {
      "epoch": 0.6549083713327024,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009584973841280837,
      "loss": 1.6759,
      "step": 7612
    },
    {
      "epoch": 0.6549944076400241,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009584865969290245,
      "loss": 1.7718,
      "step": 7613
    },
    {
      "epoch": 0.6550804439473458,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009584758083889811,
      "loss": 1.1175,
      "step": 7614
    },
    {
      "epoch": 0.6551664802546675,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009584650185079851,
      "loss": 1.4777,
      "step": 7615
    },
    {
      "epoch": 0.6552525165619891,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009584542272860681,
      "loss": 1.4845,
      "step": 7616
    },
    {
      "epoch": 0.6553385528693109,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009584434347232615,
      "loss": 1.8704,
      "step": 7617
    },
    {
      "epoch": 0.6554245891766325,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009584326408195973,
      "loss": 1.1069,
      "step": 7618
    },
    {
      "epoch": 0.6555106254839542,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009584218455751066,
      "loss": 1.9521,
      "step": 7619
    },
    {
      "epoch": 0.655596661791276,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009584110489898213,
      "loss": 1.1564,
      "step": 7620
    },
    {
      "epoch": 0.6556826980985976,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009584002510637727,
      "loss": 1.4152,
      "step": 7621
    },
    {
      "epoch": 0.6557687344059193,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009583894517969925,
      "loss": 1.0636,
      "step": 7622
    },
    {
      "epoch": 0.6558547707132409,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009583786511895125,
      "loss": 1.1155,
      "step": 7623
    },
    {
      "epoch": 0.6559408070205627,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009583678492413639,
      "loss": 1.8061,
      "step": 7624
    },
    {
      "epoch": 0.6560268433278844,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009583570459525785,
      "loss": 1.4539,
      "step": 7625
    },
    {
      "epoch": 0.656112879635206,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000958346241323188,
      "loss": 1.0851,
      "step": 7626
    },
    {
      "epoch": 0.6561989159425278,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009583354353532239,
      "loss": 2.0382,
      "step": 7627
    },
    {
      "epoch": 0.6562849522498494,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009583246280427175,
      "loss": 1.4297,
      "step": 7628
    },
    {
      "epoch": 0.6563709885571711,
      "grad_norm": 1.140625,
      "learning_rate": 0.000958313819391701,
      "loss": 1.2566,
      "step": 7629
    },
    {
      "epoch": 0.6564570248644929,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009583030094002056,
      "loss": 1.7164,
      "step": 7630
    },
    {
      "epoch": 0.6565430611718145,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009582921980682632,
      "loss": 1.0388,
      "step": 7631
    },
    {
      "epoch": 0.6566290974791362,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009582813853959051,
      "loss": 1.4659,
      "step": 7632
    },
    {
      "epoch": 0.6567151337864578,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009582705713831631,
      "loss": 1.4682,
      "step": 7633
    },
    {
      "epoch": 0.6568011700937796,
      "grad_norm": 0.875,
      "learning_rate": 0.0009582597560300688,
      "loss": 0.7038,
      "step": 7634
    },
    {
      "epoch": 0.6568872064011013,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009582489393366538,
      "loss": 1.6194,
      "step": 7635
    },
    {
      "epoch": 0.6569732427084229,
      "grad_norm": 1.375,
      "learning_rate": 0.0009582381213029498,
      "loss": 1.8147,
      "step": 7636
    },
    {
      "epoch": 0.6570592790157447,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009582273019289884,
      "loss": 1.268,
      "step": 7637
    },
    {
      "epoch": 0.6571453153230663,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009582164812148014,
      "loss": 1.707,
      "step": 7638
    },
    {
      "epoch": 0.657231351630388,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009582056591604202,
      "loss": 0.9815,
      "step": 7639
    },
    {
      "epoch": 0.6573173879377097,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009581948357658766,
      "loss": 1.5663,
      "step": 7640
    },
    {
      "epoch": 0.6574034242450314,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009581840110312023,
      "loss": 2.1219,
      "step": 7641
    },
    {
      "epoch": 0.6574894605523531,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009581731849564287,
      "loss": 1.5827,
      "step": 7642
    },
    {
      "epoch": 0.6575754968596748,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009581623575415876,
      "loss": 2.0318,
      "step": 7643
    },
    {
      "epoch": 0.6576615331669965,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009581515287867109,
      "loss": 1.5457,
      "step": 7644
    },
    {
      "epoch": 0.6577475694743181,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009581406986918301,
      "loss": 1.1509,
      "step": 7645
    },
    {
      "epoch": 0.6578336057816399,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009581298672569767,
      "loss": 1.2711,
      "step": 7646
    },
    {
      "epoch": 0.6579196420889616,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009581190344821826,
      "loss": 1.2504,
      "step": 7647
    },
    {
      "epoch": 0.6580056783962832,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009581082003674795,
      "loss": 1.482,
      "step": 7648
    },
    {
      "epoch": 0.658091714703605,
      "grad_norm": 1.515625,
      "learning_rate": 0.000958097364912899,
      "loss": 1.4256,
      "step": 7649
    },
    {
      "epoch": 0.6581777510109266,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009580865281184727,
      "loss": 1.4452,
      "step": 7650
    },
    {
      "epoch": 0.6582637873182483,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009580756899842325,
      "loss": 1.58,
      "step": 7651
    },
    {
      "epoch": 0.65834982362557,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009580648505102099,
      "loss": 0.805,
      "step": 7652
    },
    {
      "epoch": 0.6584358599328917,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009580540096964368,
      "loss": 1.0316,
      "step": 7653
    },
    {
      "epoch": 0.6585218962402134,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009580431675429446,
      "loss": 1.4711,
      "step": 7654
    },
    {
      "epoch": 0.658607932547535,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009580323240497655,
      "loss": 1.6358,
      "step": 7655
    },
    {
      "epoch": 0.6586939688548568,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009580214792169309,
      "loss": 1.4559,
      "step": 7656
    },
    {
      "epoch": 0.6587800051621785,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009580106330444723,
      "loss": 0.8596,
      "step": 7657
    },
    {
      "epoch": 0.6588660414695001,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009579997855324218,
      "loss": 1.2962,
      "step": 7658
    },
    {
      "epoch": 0.6589520777768219,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009579889366808113,
      "loss": 0.8391,
      "step": 7659
    },
    {
      "epoch": 0.6590381140841435,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009579780864896719,
      "loss": 1.3899,
      "step": 7660
    },
    {
      "epoch": 0.6591241503914652,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009579672349590356,
      "loss": 1.1146,
      "step": 7661
    },
    {
      "epoch": 0.6592101866987868,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009579563820889344,
      "loss": 1.2991,
      "step": 7662
    },
    {
      "epoch": 0.6592962230061086,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009579455278793998,
      "loss": 1.2885,
      "step": 7663
    },
    {
      "epoch": 0.6593822593134303,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009579346723304636,
      "loss": 1.2447,
      "step": 7664
    },
    {
      "epoch": 0.6594682956207519,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009579238154421575,
      "loss": 2.2082,
      "step": 7665
    },
    {
      "epoch": 0.6595543319280737,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009579129572145135,
      "loss": 2.2112,
      "step": 7666
    },
    {
      "epoch": 0.6596403682353953,
      "grad_norm": 1.140625,
      "learning_rate": 0.000957902097647563,
      "loss": 1.2068,
      "step": 7667
    },
    {
      "epoch": 0.659726404542717,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009578912367413379,
      "loss": 1.4734,
      "step": 7668
    },
    {
      "epoch": 0.6598124408500388,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00095788037449587,
      "loss": 1.0047,
      "step": 7669
    },
    {
      "epoch": 0.6598984771573604,
      "grad_norm": 1.75,
      "learning_rate": 0.000957869510911191,
      "loss": 1.1611,
      "step": 7670
    },
    {
      "epoch": 0.6599845134646821,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009578586459873329,
      "loss": 0.999,
      "step": 7671
    },
    {
      "epoch": 0.6600705497720037,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009578477797243273,
      "loss": 0.9106,
      "step": 7672
    },
    {
      "epoch": 0.6601565860793255,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009578369121222058,
      "loss": 1.8471,
      "step": 7673
    },
    {
      "epoch": 0.6602426223866472,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009578260431810004,
      "loss": 2.194,
      "step": 7674
    },
    {
      "epoch": 0.6603286586939688,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009578151729007431,
      "loss": 1.1352,
      "step": 7675
    },
    {
      "epoch": 0.6604146950012906,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009578043012814652,
      "loss": 0.9376,
      "step": 7676
    },
    {
      "epoch": 0.6605007313086122,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009577934283231989,
      "loss": 1.3979,
      "step": 7677
    },
    {
      "epoch": 0.6605867676159339,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009577825540259758,
      "loss": 1.0732,
      "step": 7678
    },
    {
      "epoch": 0.6606728039232557,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009577716783898278,
      "loss": 1.5276,
      "step": 7679
    },
    {
      "epoch": 0.6607588402305773,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009577608014147867,
      "loss": 1.1665,
      "step": 7680
    },
    {
      "epoch": 0.660844876537899,
      "grad_norm": 1.375,
      "learning_rate": 0.0009577499231008842,
      "loss": 2.0728,
      "step": 7681
    },
    {
      "epoch": 0.6609309128452207,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009577390434481523,
      "loss": 1.6313,
      "step": 7682
    },
    {
      "epoch": 0.6610169491525424,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009577281624566226,
      "loss": 1.3117,
      "step": 7683
    },
    {
      "epoch": 0.661102985459864,
      "grad_norm": 1.625,
      "learning_rate": 0.0009577172801263272,
      "loss": 1.8641,
      "step": 7684
    },
    {
      "epoch": 0.6611890217671857,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009577063964572977,
      "loss": 1.6703,
      "step": 7685
    },
    {
      "epoch": 0.6612750580745075,
      "grad_norm": 1.7421875,
      "learning_rate": 0.000957695511449566,
      "loss": 1.3239,
      "step": 7686
    },
    {
      "epoch": 0.6613610943818291,
      "grad_norm": 2.765625,
      "learning_rate": 0.0009576846251031639,
      "loss": 1.2616,
      "step": 7687
    },
    {
      "epoch": 0.6614471306891508,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009576737374181233,
      "loss": 1.2976,
      "step": 7688
    },
    {
      "epoch": 0.6615331669964725,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009576628483944761,
      "loss": 1.5109,
      "step": 7689
    },
    {
      "epoch": 0.6616192033037942,
      "grad_norm": 1.78125,
      "learning_rate": 0.000957651958032254,
      "loss": 1.4948,
      "step": 7690
    },
    {
      "epoch": 0.6617052396111159,
      "grad_norm": 8.0625,
      "learning_rate": 0.000957641066331489,
      "loss": 1.67,
      "step": 7691
    },
    {
      "epoch": 0.6617912759184376,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009576301732922129,
      "loss": 1.2291,
      "step": 7692
    },
    {
      "epoch": 0.6618773122257593,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009576192789144576,
      "loss": 1.4771,
      "step": 7693
    },
    {
      "epoch": 0.6619633485330809,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009576083831982547,
      "loss": 1.5439,
      "step": 7694
    },
    {
      "epoch": 0.6620493848404027,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009575974861436366,
      "loss": 1.963,
      "step": 7695
    },
    {
      "epoch": 0.6621354211477244,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009575865877506345,
      "loss": 0.9187,
      "step": 7696
    },
    {
      "epoch": 0.662221457455046,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009575756880192808,
      "loss": 1.7961,
      "step": 7697
    },
    {
      "epoch": 0.6623074937623677,
      "grad_norm": 1.375,
      "learning_rate": 0.0009575647869496071,
      "loss": 1.173,
      "step": 7698
    },
    {
      "epoch": 0.6623935300696894,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009575538845416455,
      "loss": 0.9471,
      "step": 7699
    },
    {
      "epoch": 0.6624795663770111,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009575429807954278,
      "loss": 2.0257,
      "step": 7700
    },
    {
      "epoch": 0.6625656026843328,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009575320757109857,
      "loss": 1.8094,
      "step": 7701
    },
    {
      "epoch": 0.6626516389916545,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009575211692883514,
      "loss": 1.2467,
      "step": 7702
    },
    {
      "epoch": 0.6627376752989762,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009575102615275567,
      "loss": 1.2196,
      "step": 7703
    },
    {
      "epoch": 0.6628237116062978,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009574993524286335,
      "loss": 1.081,
      "step": 7704
    },
    {
      "epoch": 0.6629097479136196,
      "grad_norm": 1.125,
      "learning_rate": 0.0009574884419916136,
      "loss": 1.4648,
      "step": 7705
    },
    {
      "epoch": 0.6629957842209412,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009574775302165289,
      "loss": 1.1692,
      "step": 7706
    },
    {
      "epoch": 0.6630818205282629,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009574666171034115,
      "loss": 1.3871,
      "step": 7707
    },
    {
      "epoch": 0.6631678568355847,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009574557026522932,
      "loss": 2.0289,
      "step": 7708
    },
    {
      "epoch": 0.6632538931429063,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009574447868632059,
      "loss": 1.1731,
      "step": 7709
    },
    {
      "epoch": 0.663339929450228,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009574338697361816,
      "loss": 1.7423,
      "step": 7710
    },
    {
      "epoch": 0.6634259657575496,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009574229512712522,
      "loss": 2.1209,
      "step": 7711
    },
    {
      "epoch": 0.6635120020648714,
      "grad_norm": 3.15625,
      "learning_rate": 0.0009574120314684495,
      "loss": 1.1512,
      "step": 7712
    },
    {
      "epoch": 0.6635980383721931,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009574011103278058,
      "loss": 1.5993,
      "step": 7713
    },
    {
      "epoch": 0.6636840746795147,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009573901878493527,
      "loss": 0.8279,
      "step": 7714
    },
    {
      "epoch": 0.6637701109868365,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009573792640331223,
      "loss": 0.8694,
      "step": 7715
    },
    {
      "epoch": 0.6638561472941581,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009573683388791463,
      "loss": 1.5178,
      "step": 7716
    },
    {
      "epoch": 0.6639421836014798,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009573574123874571,
      "loss": 1.9944,
      "step": 7717
    },
    {
      "epoch": 0.6640282199088016,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009573464845580863,
      "loss": 0.9501,
      "step": 7718
    },
    {
      "epoch": 0.6641142562161232,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009573355553910661,
      "loss": 1.4219,
      "step": 7719
    },
    {
      "epoch": 0.6642002925234449,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009573246248864283,
      "loss": 1.0816,
      "step": 7720
    },
    {
      "epoch": 0.6642863288307665,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009573136930442049,
      "loss": 1.0801,
      "step": 7721
    },
    {
      "epoch": 0.6643723651380883,
      "grad_norm": 1.375,
      "learning_rate": 0.0009573027598644278,
      "loss": 1.8329,
      "step": 7722
    },
    {
      "epoch": 0.66445840144541,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009572918253471293,
      "loss": 1.8363,
      "step": 7723
    },
    {
      "epoch": 0.6645444377527316,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009572808894923409,
      "loss": 0.7998,
      "step": 7724
    },
    {
      "epoch": 0.6646304740600534,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000957269952300095,
      "loss": 1.0142,
      "step": 7725
    },
    {
      "epoch": 0.664716510367375,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009572590137704233,
      "loss": 2.1846,
      "step": 7726
    },
    {
      "epoch": 0.6648025466746967,
      "grad_norm": 1.703125,
      "learning_rate": 0.000957248073903358,
      "loss": 2.2475,
      "step": 7727
    },
    {
      "epoch": 0.6648885829820184,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009572371326989309,
      "loss": 1.0148,
      "step": 7728
    },
    {
      "epoch": 0.6649746192893401,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009572261901571742,
      "loss": 1.2631,
      "step": 7729
    },
    {
      "epoch": 0.6650606555966618,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009572152462781198,
      "loss": 1.1664,
      "step": 7730
    },
    {
      "epoch": 0.6651466919039835,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009572043010617999,
      "loss": 1.372,
      "step": 7731
    },
    {
      "epoch": 0.6652327282113052,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009571933545082462,
      "loss": 1.2239,
      "step": 7732
    },
    {
      "epoch": 0.6653187645186268,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009571824066174908,
      "loss": 1.8475,
      "step": 7733
    },
    {
      "epoch": 0.6654048008259486,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009571714573895658,
      "loss": 2.0219,
      "step": 7734
    },
    {
      "epoch": 0.6654908371332703,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009571605068245032,
      "loss": 1.0181,
      "step": 7735
    },
    {
      "epoch": 0.6655768734405919,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009571495549223351,
      "loss": 1.5658,
      "step": 7736
    },
    {
      "epoch": 0.6656629097479136,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009571386016830934,
      "loss": 0.8122,
      "step": 7737
    },
    {
      "epoch": 0.6657489460552353,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0009571276471068102,
      "loss": 0.896,
      "step": 7738
    },
    {
      "epoch": 0.665834982362557,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009571166911935176,
      "loss": 1.7807,
      "step": 7739
    },
    {
      "epoch": 0.6659210186698787,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009571057339432476,
      "loss": 1.5512,
      "step": 7740
    },
    {
      "epoch": 0.6660070549772004,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009570947753560322,
      "loss": 1.552,
      "step": 7741
    },
    {
      "epoch": 0.6660930912845221,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009570838154319035,
      "loss": 1.1514,
      "step": 7742
    },
    {
      "epoch": 0.6661791275918437,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009570728541708934,
      "loss": 1.2044,
      "step": 7743
    },
    {
      "epoch": 0.6662651638991655,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009570618915730342,
      "loss": 1.5661,
      "step": 7744
    },
    {
      "epoch": 0.6663512002064872,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009570509276383579,
      "loss": 1.6435,
      "step": 7745
    },
    {
      "epoch": 0.6664372365138088,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009570399623668964,
      "loss": 1.7993,
      "step": 7746
    },
    {
      "epoch": 0.6665232728211306,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000957028995758682,
      "loss": 0.9866,
      "step": 7747
    },
    {
      "epoch": 0.6666093091284522,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009570180278137465,
      "loss": 1.9033,
      "step": 7748
    },
    {
      "epoch": 0.6666953454357739,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009570070585321225,
      "loss": 1.4014,
      "step": 7749
    },
    {
      "epoch": 0.6667813817430955,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009569960879138416,
      "loss": 1.5873,
      "step": 7750
    },
    {
      "epoch": 0.6668674180504173,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009569851159589359,
      "loss": 1.2845,
      "step": 7751
    },
    {
      "epoch": 0.666953454357739,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009569741426674376,
      "loss": 1.3513,
      "step": 7752
    },
    {
      "epoch": 0.6670394906650606,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009569631680393789,
      "loss": 1.4269,
      "step": 7753
    },
    {
      "epoch": 0.6671255269723824,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009569521920747919,
      "loss": 1.6129,
      "step": 7754
    },
    {
      "epoch": 0.667211563279704,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009569412147737085,
      "loss": 1.6795,
      "step": 7755
    },
    {
      "epoch": 0.6672975995870257,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009569302361361609,
      "loss": 1.1423,
      "step": 7756
    },
    {
      "epoch": 0.6673836358943475,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009569192561621812,
      "loss": 1.3891,
      "step": 7757
    },
    {
      "epoch": 0.6674696722016691,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009569082748518016,
      "loss": 1.6906,
      "step": 7758
    },
    {
      "epoch": 0.6675557085089908,
      "grad_norm": 0.796875,
      "learning_rate": 0.0009568972922050542,
      "loss": 0.8821,
      "step": 7759
    },
    {
      "epoch": 0.6676417448163124,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000956886308221971,
      "loss": 1.0218,
      "step": 7760
    },
    {
      "epoch": 0.6677277811236342,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009568753229025843,
      "loss": 1.2736,
      "step": 7761
    },
    {
      "epoch": 0.6678138174309559,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000956864336246926,
      "loss": 1.57,
      "step": 7762
    },
    {
      "epoch": 0.6678998537382775,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009568533482550286,
      "loss": 1.8472,
      "step": 7763
    },
    {
      "epoch": 0.6679858900455993,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009568423589269237,
      "loss": 1.8026,
      "step": 7764
    },
    {
      "epoch": 0.6680719263529209,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009568313682626439,
      "loss": 1.0483,
      "step": 7765
    },
    {
      "epoch": 0.6681579626602426,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009568203762622213,
      "loss": 1.33,
      "step": 7766
    },
    {
      "epoch": 0.6682439989675643,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009568093829256879,
      "loss": 1.7968,
      "step": 7767
    },
    {
      "epoch": 0.668330035274886,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009567983882530758,
      "loss": 1.8257,
      "step": 7768
    },
    {
      "epoch": 0.6684160715822077,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009567873922444174,
      "loss": 1.1447,
      "step": 7769
    },
    {
      "epoch": 0.6685021078895294,
      "grad_norm": 1.125,
      "learning_rate": 0.0009567763948997445,
      "loss": 1.0287,
      "step": 7770
    },
    {
      "epoch": 0.6685881441968511,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009567653962190896,
      "loss": 1.5621,
      "step": 7771
    },
    {
      "epoch": 0.6686741805041727,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009567543962024849,
      "loss": 1.4236,
      "step": 7772
    },
    {
      "epoch": 0.6687602168114944,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009567433948499622,
      "loss": 1.9988,
      "step": 7773
    },
    {
      "epoch": 0.6688462531188162,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009567323921615541,
      "loss": 1.1046,
      "step": 7774
    },
    {
      "epoch": 0.6689322894261378,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009567213881372925,
      "loss": 1.0487,
      "step": 7775
    },
    {
      "epoch": 0.6690183257334595,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009567103827772096,
      "loss": 1.3067,
      "step": 7776
    },
    {
      "epoch": 0.6691043620407812,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009566993760813376,
      "loss": 1.4068,
      "step": 7777
    },
    {
      "epoch": 0.6691903983481029,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009566883680497089,
      "loss": 0.8194,
      "step": 7778
    },
    {
      "epoch": 0.6692764346554246,
      "grad_norm": 1.375,
      "learning_rate": 0.0009566773586823555,
      "loss": 1.1998,
      "step": 7779
    },
    {
      "epoch": 0.6693624709627463,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009566663479793097,
      "loss": 1.5899,
      "step": 7780
    },
    {
      "epoch": 0.669448507270068,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009566553359406035,
      "loss": 0.9933,
      "step": 7781
    },
    {
      "epoch": 0.6695345435773896,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009566443225662694,
      "loss": 2.1056,
      "step": 7782
    },
    {
      "epoch": 0.6696205798847114,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009566333078563394,
      "loss": 2.0544,
      "step": 7783
    },
    {
      "epoch": 0.6697066161920331,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009566222918108458,
      "loss": 1.5006,
      "step": 7784
    },
    {
      "epoch": 0.6697926524993547,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009566112744298208,
      "loss": 1.7196,
      "step": 7785
    },
    {
      "epoch": 0.6698786888066764,
      "grad_norm": 1.625,
      "learning_rate": 0.0009566002557132966,
      "loss": 1.1442,
      "step": 7786
    },
    {
      "epoch": 0.6699647251139981,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009565892356613056,
      "loss": 0.9088,
      "step": 7787
    },
    {
      "epoch": 0.6700507614213198,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009565782142738797,
      "loss": 1.4318,
      "step": 7788
    },
    {
      "epoch": 0.6701367977286414,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009565671915510515,
      "loss": 0.9052,
      "step": 7789
    },
    {
      "epoch": 0.6702228340359632,
      "grad_norm": 2.4375,
      "learning_rate": 0.0009565561674928528,
      "loss": 0.8674,
      "step": 7790
    },
    {
      "epoch": 0.6703088703432849,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009565451420993163,
      "loss": 1.4737,
      "step": 7791
    },
    {
      "epoch": 0.6703949066506065,
      "grad_norm": 1.0625,
      "learning_rate": 0.000956534115370474,
      "loss": 1.2941,
      "step": 7792
    },
    {
      "epoch": 0.6704809429579283,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009565230873063581,
      "loss": 1.5255,
      "step": 7793
    },
    {
      "epoch": 0.6705669792652499,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009565120579070012,
      "loss": 1.9151,
      "step": 7794
    },
    {
      "epoch": 0.6706530155725716,
      "grad_norm": 1.125,
      "learning_rate": 0.0009565010271724351,
      "loss": 1.3019,
      "step": 7795
    },
    {
      "epoch": 0.6707390518798934,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009564899951026924,
      "loss": 2.3596,
      "step": 7796
    },
    {
      "epoch": 0.670825088187215,
      "grad_norm": 2.171875,
      "learning_rate": 0.0009564789616978052,
      "loss": 1.1587,
      "step": 7797
    },
    {
      "epoch": 0.6709111244945367,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009564679269578058,
      "loss": 1.7897,
      "step": 7798
    },
    {
      "epoch": 0.6709971608018583,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009564568908827266,
      "loss": 0.9268,
      "step": 7799
    },
    {
      "epoch": 0.6710831971091801,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009564458534725997,
      "loss": 1.8645,
      "step": 7800
    },
    {
      "epoch": 0.6711692334165018,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009564348147274575,
      "loss": 1.7421,
      "step": 7801
    },
    {
      "epoch": 0.6712552697238234,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009564237746473321,
      "loss": 1.4292,
      "step": 7802
    },
    {
      "epoch": 0.6713413060311452,
      "grad_norm": 1.375,
      "learning_rate": 0.0009564127332322561,
      "loss": 2.2768,
      "step": 7803
    },
    {
      "epoch": 0.6714273423384668,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009564016904822616,
      "loss": 1.9973,
      "step": 7804
    },
    {
      "epoch": 0.6715133786457885,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009563906463973808,
      "loss": 1.4927,
      "step": 7805
    },
    {
      "epoch": 0.6715994149531103,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009563796009776462,
      "loss": 1.0441,
      "step": 7806
    },
    {
      "epoch": 0.6716854512604319,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00095636855422309,
      "loss": 1.4425,
      "step": 7807
    },
    {
      "epoch": 0.6717714875677536,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009563575061337446,
      "loss": 1.2925,
      "step": 7808
    },
    {
      "epoch": 0.6718575238750752,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009563464567096422,
      "loss": 2.1088,
      "step": 7809
    },
    {
      "epoch": 0.671943560182397,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0009563354059508152,
      "loss": 1.1545,
      "step": 7810
    },
    {
      "epoch": 0.6720295964897186,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009563243538572957,
      "loss": 1.9001,
      "step": 7811
    },
    {
      "epoch": 0.6721156327970403,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009563133004291165,
      "loss": 1.431,
      "step": 7812
    },
    {
      "epoch": 0.6722016691043621,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009563022456663095,
      "loss": 1.5229,
      "step": 7813
    },
    {
      "epoch": 0.6722877054116837,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009562911895689072,
      "loss": 1.5044,
      "step": 7814
    },
    {
      "epoch": 0.6723737417190054,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009562801321369419,
      "loss": 2.0033,
      "step": 7815
    },
    {
      "epoch": 0.6724597780263271,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000956269073370446,
      "loss": 1.2507,
      "step": 7816
    },
    {
      "epoch": 0.6725458143336488,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009562580132694517,
      "loss": 1.5576,
      "step": 7817
    },
    {
      "epoch": 0.6726318506409705,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009562469518339916,
      "loss": 1.763,
      "step": 7818
    },
    {
      "epoch": 0.6727178869482922,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009562358890640978,
      "loss": 0.9623,
      "step": 7819
    },
    {
      "epoch": 0.6728039232556139,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009562248249598027,
      "loss": 1.8571,
      "step": 7820
    },
    {
      "epoch": 0.6728899595629355,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009562137595211388,
      "loss": 0.9984,
      "step": 7821
    },
    {
      "epoch": 0.6729759958702572,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009562026927481384,
      "loss": 1.5445,
      "step": 7822
    },
    {
      "epoch": 0.673062032177579,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009561916246408337,
      "loss": 1.7937,
      "step": 7823
    },
    {
      "epoch": 0.6731480684849006,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009561805551992575,
      "loss": 1.1397,
      "step": 7824
    },
    {
      "epoch": 0.6732341047922223,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009561694844234417,
      "loss": 1.9081,
      "step": 7825
    },
    {
      "epoch": 0.673320141099544,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009561584123134189,
      "loss": 1.1999,
      "step": 7826
    },
    {
      "epoch": 0.6734061774068657,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009561473388692215,
      "loss": 1.5786,
      "step": 7827
    },
    {
      "epoch": 0.6734922137141874,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009561362640908819,
      "loss": 1.2102,
      "step": 7828
    },
    {
      "epoch": 0.6735782500215091,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009561251879784322,
      "loss": 1.3652,
      "step": 7829
    },
    {
      "epoch": 0.6736642863288308,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009561141105319053,
      "loss": 1.507,
      "step": 7830
    },
    {
      "epoch": 0.6737503226361524,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009561030317513333,
      "loss": 1.151,
      "step": 7831
    },
    {
      "epoch": 0.6738363589434742,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0009560919516367485,
      "loss": 0.7926,
      "step": 7832
    },
    {
      "epoch": 0.6739223952507958,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009560808701881835,
      "loss": 1.0151,
      "step": 7833
    },
    {
      "epoch": 0.6740084315581175,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009560697874056708,
      "loss": 1.6317,
      "step": 7834
    },
    {
      "epoch": 0.6740944678654393,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009560587032892425,
      "loss": 1.198,
      "step": 7835
    },
    {
      "epoch": 0.6741805041727609,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009560476178389313,
      "loss": 1.4295,
      "step": 7836
    },
    {
      "epoch": 0.6742665404800826,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009560365310547694,
      "loss": 0.8005,
      "step": 7837
    },
    {
      "epoch": 0.6743525767874042,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009560254429367893,
      "loss": 1.9232,
      "step": 7838
    },
    {
      "epoch": 0.674438613094726,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009560143534850237,
      "loss": 0.8581,
      "step": 7839
    },
    {
      "epoch": 0.6745246494020477,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009560032626995048,
      "loss": 1.4814,
      "step": 7840
    },
    {
      "epoch": 0.6746106857093693,
      "grad_norm": 2.65625,
      "learning_rate": 0.0009559921705802648,
      "loss": 1.2923,
      "step": 7841
    },
    {
      "epoch": 0.6746967220166911,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009559810771273365,
      "loss": 1.015,
      "step": 7842
    },
    {
      "epoch": 0.6747827583240127,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009559699823407523,
      "loss": 0.7442,
      "step": 7843
    },
    {
      "epoch": 0.6748687946313344,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009559588862205446,
      "loss": 2.3233,
      "step": 7844
    },
    {
      "epoch": 0.6749548309386562,
      "grad_norm": 3.4375,
      "learning_rate": 0.0009559477887667458,
      "loss": 1.6737,
      "step": 7845
    },
    {
      "epoch": 0.6750408672459778,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009559366899793882,
      "loss": 1.1713,
      "step": 7846
    },
    {
      "epoch": 0.6751269035532995,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009559255898585046,
      "loss": 1.5532,
      "step": 7847
    },
    {
      "epoch": 0.6752129398606211,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009559144884041276,
      "loss": 1.641,
      "step": 7848
    },
    {
      "epoch": 0.6752989761679429,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009559033856162889,
      "loss": 1.9641,
      "step": 7849
    },
    {
      "epoch": 0.6753850124752646,
      "grad_norm": 1.625,
      "learning_rate": 0.0009558922814950217,
      "loss": 1.9775,
      "step": 7850
    },
    {
      "epoch": 0.6754710487825862,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009558811760403583,
      "loss": 1.4125,
      "step": 7851
    },
    {
      "epoch": 0.675557085089908,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009558700692523311,
      "loss": 1.1274,
      "step": 7852
    },
    {
      "epoch": 0.6756431213972296,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009558589611309726,
      "loss": 1.2112,
      "step": 7853
    },
    {
      "epoch": 0.6757291577045513,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009558478516763153,
      "loss": 0.9941,
      "step": 7854
    },
    {
      "epoch": 0.675815194011873,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009558367408883917,
      "loss": 1.6079,
      "step": 7855
    },
    {
      "epoch": 0.6759012303191947,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009558256287672342,
      "loss": 1.8872,
      "step": 7856
    },
    {
      "epoch": 0.6759872666265164,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009558145153128754,
      "loss": 1.688,
      "step": 7857
    },
    {
      "epoch": 0.676073302933838,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009558034005253479,
      "loss": 1.5278,
      "step": 7858
    },
    {
      "epoch": 0.6761593392411598,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009557922844046841,
      "loss": 1.0032,
      "step": 7859
    },
    {
      "epoch": 0.6762453755484814,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009557811669509165,
      "loss": 1.811,
      "step": 7860
    },
    {
      "epoch": 0.6763314118558031,
      "grad_norm": 1.375,
      "learning_rate": 0.0009557700481640774,
      "loss": 1.5351,
      "step": 7861
    },
    {
      "epoch": 0.6764174481631249,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009557589280441999,
      "loss": 0.8287,
      "step": 7862
    },
    {
      "epoch": 0.6765034844704465,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009557478065913158,
      "loss": 1.4235,
      "step": 7863
    },
    {
      "epoch": 0.6765895207777682,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009557366838054583,
      "loss": 1.1479,
      "step": 7864
    },
    {
      "epoch": 0.6766755570850899,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009557255596866596,
      "loss": 1.3555,
      "step": 7865
    },
    {
      "epoch": 0.6767615933924116,
      "grad_norm": 1.375,
      "learning_rate": 0.0009557144342349522,
      "loss": 1.3393,
      "step": 7866
    },
    {
      "epoch": 0.6768476296997333,
      "grad_norm": 1.375,
      "learning_rate": 0.0009557033074503687,
      "loss": 1.4501,
      "step": 7867
    },
    {
      "epoch": 0.676933666007055,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009556921793329417,
      "loss": 1.7397,
      "step": 7868
    },
    {
      "epoch": 0.6770197023143767,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009556810498827038,
      "loss": 1.0995,
      "step": 7869
    },
    {
      "epoch": 0.6771057386216983,
      "grad_norm": 1.25,
      "learning_rate": 0.0009556699190996872,
      "loss": 1.3379,
      "step": 7870
    },
    {
      "epoch": 0.67719177492902,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009556587869839248,
      "loss": 1.457,
      "step": 7871
    },
    {
      "epoch": 0.6772778112363418,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009556476535354493,
      "loss": 1.4482,
      "step": 7872
    },
    {
      "epoch": 0.6773638475436634,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009556365187542928,
      "loss": 1.3631,
      "step": 7873
    },
    {
      "epoch": 0.6774498838509851,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009556253826404882,
      "loss": 1.2452,
      "step": 7874
    },
    {
      "epoch": 0.6775359201583068,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009556142451940679,
      "loss": 1.4679,
      "step": 7875
    },
    {
      "epoch": 0.6776219564656285,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009556031064150646,
      "loss": 1.6005,
      "step": 7876
    },
    {
      "epoch": 0.6777079927729501,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009555919663035107,
      "loss": 1.3153,
      "step": 7877
    },
    {
      "epoch": 0.6777940290802719,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009555808248594391,
      "loss": 1.5204,
      "step": 7878
    },
    {
      "epoch": 0.6778800653875936,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009555696820828821,
      "loss": 0.9495,
      "step": 7879
    },
    {
      "epoch": 0.6779661016949152,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009555585379738723,
      "loss": 1.1225,
      "step": 7880
    },
    {
      "epoch": 0.678052138002237,
      "grad_norm": 1.0,
      "learning_rate": 0.0009555473925324425,
      "loss": 1.1092,
      "step": 7881
    },
    {
      "epoch": 0.6781381743095586,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009555362457586251,
      "loss": 1.9142,
      "step": 7882
    },
    {
      "epoch": 0.6782242106168803,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009555250976524528,
      "loss": 1.2156,
      "step": 7883
    },
    {
      "epoch": 0.678310246924202,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009555139482139582,
      "loss": 1.5768,
      "step": 7884
    },
    {
      "epoch": 0.6783962832315237,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009555027974431738,
      "loss": 1.5451,
      "step": 7885
    },
    {
      "epoch": 0.6784823195388454,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009554916453401325,
      "loss": 1.2506,
      "step": 7886
    },
    {
      "epoch": 0.678568355846167,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009554804919048666,
      "loss": 0.8196,
      "step": 7887
    },
    {
      "epoch": 0.6786543921534888,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009554693371374088,
      "loss": 1.425,
      "step": 7888
    },
    {
      "epoch": 0.6787404284608105,
      "grad_norm": 1.875,
      "learning_rate": 0.0009554581810377919,
      "loss": 1.4334,
      "step": 7889
    },
    {
      "epoch": 0.6788264647681321,
      "grad_norm": 2.453125,
      "learning_rate": 0.0009554470236060484,
      "loss": 1.7238,
      "step": 7890
    },
    {
      "epoch": 0.6789125010754539,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009554358648422108,
      "loss": 1.4817,
      "step": 7891
    },
    {
      "epoch": 0.6789985373827755,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009554247047463121,
      "loss": 1.0295,
      "step": 7892
    },
    {
      "epoch": 0.6790845736900972,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009554135433183844,
      "loss": 0.9391,
      "step": 7893
    },
    {
      "epoch": 0.679170609997419,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009554023805584609,
      "loss": 1.0832,
      "step": 7894
    },
    {
      "epoch": 0.6792566463047406,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009553912164665739,
      "loss": 1.1362,
      "step": 7895
    },
    {
      "epoch": 0.6793426826120623,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009553800510427562,
      "loss": 2.337,
      "step": 7896
    },
    {
      "epoch": 0.679428718919384,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009553688842870405,
      "loss": 1.1478,
      "step": 7897
    },
    {
      "epoch": 0.6795147552267057,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009553577161994592,
      "loss": 1.4124,
      "step": 7898
    },
    {
      "epoch": 0.6796007915340273,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009553465467800453,
      "loss": 1.7976,
      "step": 7899
    },
    {
      "epoch": 0.679686827841349,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009553353760288312,
      "loss": 1.944,
      "step": 7900
    },
    {
      "epoch": 0.6797728641486708,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009553242039458496,
      "loss": 1.7222,
      "step": 7901
    },
    {
      "epoch": 0.6798589004559924,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009553130305311335,
      "loss": 1.1958,
      "step": 7902
    },
    {
      "epoch": 0.6799449367633141,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009553018557847152,
      "loss": 1.5293,
      "step": 7903
    },
    {
      "epoch": 0.6800309730706358,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009552906797066275,
      "loss": 1.4355,
      "step": 7904
    },
    {
      "epoch": 0.6801170093779575,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009552795022969031,
      "loss": 1.6316,
      "step": 7905
    },
    {
      "epoch": 0.6802030456852792,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009552683235555748,
      "loss": 1.7336,
      "step": 7906
    },
    {
      "epoch": 0.6802890819926009,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009552571434826752,
      "loss": 1.1171,
      "step": 7907
    },
    {
      "epoch": 0.6803751182999226,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000955245962078237,
      "loss": 0.535,
      "step": 7908
    },
    {
      "epoch": 0.6804611546072442,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009552347793422929,
      "loss": 1.5205,
      "step": 7909
    },
    {
      "epoch": 0.680547190914566,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009552235952748755,
      "loss": 1.8838,
      "step": 7910
    },
    {
      "epoch": 0.6806332272218877,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009552124098760177,
      "loss": 1.5493,
      "step": 7911
    },
    {
      "epoch": 0.6807192635292093,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009552012231457521,
      "loss": 1.1077,
      "step": 7912
    },
    {
      "epoch": 0.680805299836531,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009551900350841114,
      "loss": 1.2083,
      "step": 7913
    },
    {
      "epoch": 0.6808913361438527,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009551788456911286,
      "loss": 0.996,
      "step": 7914
    },
    {
      "epoch": 0.6809773724511744,
      "grad_norm": 1.359375,
      "learning_rate": 0.000955167654966836,
      "loss": 1.6312,
      "step": 7915
    },
    {
      "epoch": 0.6810634087584961,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009551564629112666,
      "loss": 1.5845,
      "step": 7916
    },
    {
      "epoch": 0.6811494450658178,
      "grad_norm": 1.84375,
      "learning_rate": 0.000955145269524453,
      "loss": 1.8282,
      "step": 7917
    },
    {
      "epoch": 0.6812354813731395,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009551340748064281,
      "loss": 1.0128,
      "step": 7918
    },
    {
      "epoch": 0.6813215176804611,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009551228787572246,
      "loss": 1.734,
      "step": 7919
    },
    {
      "epoch": 0.6814075539877829,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000955111681376875,
      "loss": 1.1654,
      "step": 7920
    },
    {
      "epoch": 0.6814935902951045,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009551004826654124,
      "loss": 1.0487,
      "step": 7921
    },
    {
      "epoch": 0.6815796266024262,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009550892826228693,
      "loss": 1.9081,
      "step": 7922
    },
    {
      "epoch": 0.681665662909748,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009550780812492786,
      "loss": 1.5227,
      "step": 7923
    },
    {
      "epoch": 0.6817516992170696,
      "grad_norm": 1.125,
      "learning_rate": 0.000955066878544673,
      "loss": 1.1274,
      "step": 7924
    },
    {
      "epoch": 0.6818377355243913,
      "grad_norm": 1.625,
      "learning_rate": 0.0009550556745090855,
      "loss": 1.0062,
      "step": 7925
    },
    {
      "epoch": 0.6819237718317129,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009550444691425484,
      "loss": 1.4384,
      "step": 7926
    },
    {
      "epoch": 0.6820098081390347,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009550332624450948,
      "loss": 0.9068,
      "step": 7927
    },
    {
      "epoch": 0.6820958444463564,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009550220544167574,
      "loss": 1.1661,
      "step": 7928
    },
    {
      "epoch": 0.682181880753678,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000955010845057569,
      "loss": 1.9652,
      "step": 7929
    },
    {
      "epoch": 0.6822679170609998,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009549996343675624,
      "loss": 1.3808,
      "step": 7930
    },
    {
      "epoch": 0.6823539533683214,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009549884223467703,
      "loss": 1.5745,
      "step": 7931
    },
    {
      "epoch": 0.6824399896756431,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009549772089952255,
      "loss": 2.334,
      "step": 7932
    },
    {
      "epoch": 0.6825260259829649,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009549659943129611,
      "loss": 1.0452,
      "step": 7933
    },
    {
      "epoch": 0.6826120622902865,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009549547783000094,
      "loss": 1.6844,
      "step": 7934
    },
    {
      "epoch": 0.6826980985976082,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009549435609564037,
      "loss": 2.1249,
      "step": 7935
    },
    {
      "epoch": 0.6827841349049298,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009549323422821763,
      "loss": 1.9034,
      "step": 7936
    },
    {
      "epoch": 0.6828701712122516,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009549211222773604,
      "loss": 0.8149,
      "step": 7937
    },
    {
      "epoch": 0.6829562075195733,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009549099009419888,
      "loss": 1.6439,
      "step": 7938
    },
    {
      "epoch": 0.6830422438268949,
      "grad_norm": 1.21875,
      "learning_rate": 0.000954898678276094,
      "loss": 0.7503,
      "step": 7939
    },
    {
      "epoch": 0.6831282801342167,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009548874542797091,
      "loss": 1.5024,
      "step": 7940
    },
    {
      "epoch": 0.6832143164415383,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000954876228952867,
      "loss": 2.0112,
      "step": 7941
    },
    {
      "epoch": 0.68330035274886,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009548650022956003,
      "loss": 1.502,
      "step": 7942
    },
    {
      "epoch": 0.6833863890561817,
      "grad_norm": 1.484375,
      "learning_rate": 0.000954853774307942,
      "loss": 1.5783,
      "step": 7943
    },
    {
      "epoch": 0.6834724253635034,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009548425449899247,
      "loss": 2.0333,
      "step": 7944
    },
    {
      "epoch": 0.6835584616708251,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009548313143415817,
      "loss": 1.092,
      "step": 7945
    },
    {
      "epoch": 0.6836444979781467,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009548200823629453,
      "loss": 1.1215,
      "step": 7946
    },
    {
      "epoch": 0.6837305342854685,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009548088490540488,
      "loss": 2.1013,
      "step": 7947
    },
    {
      "epoch": 0.6838165705927901,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009547976144149246,
      "loss": 0.7144,
      "step": 7948
    },
    {
      "epoch": 0.6839026069001118,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009547863784456061,
      "loss": 1.8812,
      "step": 7949
    },
    {
      "epoch": 0.6839886432074336,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009547751411461259,
      "loss": 1.1996,
      "step": 7950
    },
    {
      "epoch": 0.6840746795147552,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009547639025165167,
      "loss": 1.0624,
      "step": 7951
    },
    {
      "epoch": 0.6841607158220769,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009547526625568116,
      "loss": 1.1794,
      "step": 7952
    },
    {
      "epoch": 0.6842467521293986,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009547414212670434,
      "loss": 2.2087,
      "step": 7953
    },
    {
      "epoch": 0.6843327884367203,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009547301786472451,
      "loss": 0.8842,
      "step": 7954
    },
    {
      "epoch": 0.684418824744042,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009547189346974493,
      "loss": 0.9277,
      "step": 7955
    },
    {
      "epoch": 0.6845048610513637,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009547076894176891,
      "loss": 1.1839,
      "step": 7956
    },
    {
      "epoch": 0.6845908973586854,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009546964428079974,
      "loss": 1.8542,
      "step": 7957
    },
    {
      "epoch": 0.684676933666007,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000954685194868407,
      "loss": 1.8791,
      "step": 7958
    },
    {
      "epoch": 0.6847629699733288,
      "grad_norm": 1.375,
      "learning_rate": 0.0009546739455989508,
      "loss": 2.1749,
      "step": 7959
    },
    {
      "epoch": 0.6848490062806505,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0009546626949996618,
      "loss": 0.7985,
      "step": 7960
    },
    {
      "epoch": 0.6849350425879721,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009546514430705728,
      "loss": 1.9532,
      "step": 7961
    },
    {
      "epoch": 0.6850210788952938,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009546401898117169,
      "loss": 1.8397,
      "step": 7962
    },
    {
      "epoch": 0.6851071152026155,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009546289352231267,
      "loss": 0.6444,
      "step": 7963
    },
    {
      "epoch": 0.6851931515099372,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009546176793048353,
      "loss": 1.3428,
      "step": 7964
    },
    {
      "epoch": 0.6852791878172588,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009546064220568757,
      "loss": 0.9329,
      "step": 7965
    },
    {
      "epoch": 0.6853652241245806,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009545951634792807,
      "loss": 1.8884,
      "step": 7966
    },
    {
      "epoch": 0.6854512604319023,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009545839035720831,
      "loss": 1.4144,
      "step": 7967
    },
    {
      "epoch": 0.6855372967392239,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009545726423353161,
      "loss": 1.9031,
      "step": 7968
    },
    {
      "epoch": 0.6856233330465457,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009545613797690126,
      "loss": 1.0684,
      "step": 7969
    },
    {
      "epoch": 0.6857093693538673,
      "grad_norm": 2.1875,
      "learning_rate": 0.0009545501158732055,
      "loss": 2.2044,
      "step": 7970
    },
    {
      "epoch": 0.685795405661189,
      "grad_norm": 1.375,
      "learning_rate": 0.0009545388506479275,
      "loss": 1.1463,
      "step": 7971
    },
    {
      "epoch": 0.6858814419685108,
      "grad_norm": 1.046875,
      "learning_rate": 0.000954527584093212,
      "loss": 0.8372,
      "step": 7972
    },
    {
      "epoch": 0.6859674782758324,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009545163162090917,
      "loss": 0.9575,
      "step": 7973
    },
    {
      "epoch": 0.6860535145831541,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009545050469955994,
      "loss": 1.3625,
      "step": 7974
    },
    {
      "epoch": 0.6861395508904757,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009544937764527683,
      "loss": 1.6232,
      "step": 7975
    },
    {
      "epoch": 0.6862255871977975,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009544825045806314,
      "loss": 0.7512,
      "step": 7976
    },
    {
      "epoch": 0.6863116235051192,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009544712313792215,
      "loss": 0.9164,
      "step": 7977
    },
    {
      "epoch": 0.6863976598124408,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009544599568485716,
      "loss": 0.9457,
      "step": 7978
    },
    {
      "epoch": 0.6864836961197626,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009544486809887147,
      "loss": 1.3079,
      "step": 7979
    },
    {
      "epoch": 0.6865697324270842,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009544374037996839,
      "loss": 1.4202,
      "step": 7980
    },
    {
      "epoch": 0.6866557687344059,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000954426125281512,
      "loss": 1.6114,
      "step": 7981
    },
    {
      "epoch": 0.6867418050417277,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009544148454342321,
      "loss": 1.8858,
      "step": 7982
    },
    {
      "epoch": 0.6868278413490493,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009544035642578771,
      "loss": 1.0727,
      "step": 7983
    },
    {
      "epoch": 0.686913877656371,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009543922817524802,
      "loss": 1.9214,
      "step": 7984
    },
    {
      "epoch": 0.6869999139636926,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000954380997918074,
      "loss": 2.3178,
      "step": 7985
    },
    {
      "epoch": 0.6870859502710144,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000954369712754692,
      "loss": 1.0578,
      "step": 7986
    },
    {
      "epoch": 0.687171986578336,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009543584262623668,
      "loss": 2.0429,
      "step": 7987
    },
    {
      "epoch": 0.6872580228856577,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009543471384411315,
      "loss": 1.0996,
      "step": 7988
    },
    {
      "epoch": 0.6873440591929795,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009543358492910195,
      "loss": 1.5407,
      "step": 7989
    },
    {
      "epoch": 0.6874300955003011,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009543245588120632,
      "loss": 1.2615,
      "step": 7990
    },
    {
      "epoch": 0.6875161318076228,
      "grad_norm": 1.0,
      "learning_rate": 0.000954313267004296,
      "loss": 1.1396,
      "step": 7991
    },
    {
      "epoch": 0.6876021681149445,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009543019738677509,
      "loss": 1.1044,
      "step": 7992
    },
    {
      "epoch": 0.6876882044222662,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009542906794024609,
      "loss": 1.3053,
      "step": 7993
    },
    {
      "epoch": 0.6877742407295879,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009542793836084591,
      "loss": 2.2511,
      "step": 7994
    },
    {
      "epoch": 0.6878602770369096,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009542680864857783,
      "loss": 2.175,
      "step": 7995
    },
    {
      "epoch": 0.6879463133442313,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009542567880344517,
      "loss": 1.6983,
      "step": 7996
    },
    {
      "epoch": 0.6880323496515529,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009542454882545124,
      "loss": 1.1642,
      "step": 7997
    },
    {
      "epoch": 0.6881183859588746,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009542341871459935,
      "loss": 1.0239,
      "step": 7998
    },
    {
      "epoch": 0.6882044222661964,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009542228847089277,
      "loss": 1.3746,
      "step": 7999
    },
    {
      "epoch": 0.688290458573518,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009542115809433485,
      "loss": 1.5243,
      "step": 8000
    },
    {
      "epoch": 0.6883764948808397,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009542002758492887,
      "loss": 1.0072,
      "step": 8001
    },
    {
      "epoch": 0.6884625311881614,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009541889694267814,
      "loss": 2.0973,
      "step": 8002
    },
    {
      "epoch": 0.6885485674954831,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009541776616758598,
      "loss": 1.8319,
      "step": 8003
    },
    {
      "epoch": 0.6886346038028048,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009541663525965567,
      "loss": 1.076,
      "step": 8004
    },
    {
      "epoch": 0.6887206401101265,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009541550421889054,
      "loss": 1.8703,
      "step": 8005
    },
    {
      "epoch": 0.6888066764174482,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000954143730452939,
      "loss": 1.4398,
      "step": 8006
    },
    {
      "epoch": 0.6888927127247698,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009541324173886905,
      "loss": 1.1178,
      "step": 8007
    },
    {
      "epoch": 0.6889787490320916,
      "grad_norm": 1.3125,
      "learning_rate": 0.000954121102996193,
      "loss": 0.8415,
      "step": 8008
    },
    {
      "epoch": 0.6890647853394132,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009541097872754796,
      "loss": 0.9645,
      "step": 8009
    },
    {
      "epoch": 0.6891508216467349,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009540984702265834,
      "loss": 2.3405,
      "step": 8010
    },
    {
      "epoch": 0.6892368579540566,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009540871518495376,
      "loss": 1.1883,
      "step": 8011
    },
    {
      "epoch": 0.6893228942613783,
      "grad_norm": 1.171875,
      "learning_rate": 0.000954075832144375,
      "loss": 1.0029,
      "step": 8012
    },
    {
      "epoch": 0.6894089305687,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009540645111111289,
      "loss": 1.5882,
      "step": 8013
    },
    {
      "epoch": 0.6894949668760216,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009540531887498324,
      "loss": 1.0314,
      "step": 8014
    },
    {
      "epoch": 0.6895810031833434,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009540418650605189,
      "loss": 1.397,
      "step": 8015
    },
    {
      "epoch": 0.6896670394906651,
      "grad_norm": 1.171875,
      "learning_rate": 0.000954030540043221,
      "loss": 0.9191,
      "step": 8016
    },
    {
      "epoch": 0.6897530757979867,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009540192136979721,
      "loss": 1.8065,
      "step": 8017
    },
    {
      "epoch": 0.6898391121053085,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009540078860248054,
      "loss": 0.8942,
      "step": 8018
    },
    {
      "epoch": 0.6899251484126301,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009539965570237539,
      "loss": 1.6901,
      "step": 8019
    },
    {
      "epoch": 0.6900111847199518,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009539852266948508,
      "loss": 1.7432,
      "step": 8020
    },
    {
      "epoch": 0.6900972210272736,
      "grad_norm": 1.015625,
      "learning_rate": 0.000953973895038129,
      "loss": 1.0304,
      "step": 8021
    },
    {
      "epoch": 0.6901832573345952,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009539625620536221,
      "loss": 1.1825,
      "step": 8022
    },
    {
      "epoch": 0.6902692936419169,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009539512277413629,
      "loss": 1.2399,
      "step": 8023
    },
    {
      "epoch": 0.6903553299492385,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009539398921013846,
      "loss": 1.4462,
      "step": 8024
    },
    {
      "epoch": 0.6904413662565603,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009539285551337205,
      "loss": 1.6999,
      "step": 8025
    },
    {
      "epoch": 0.690527402563882,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009539172168384036,
      "loss": 1.1545,
      "step": 8026
    },
    {
      "epoch": 0.6906134388712036,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009539058772154672,
      "loss": 1.7266,
      "step": 8027
    },
    {
      "epoch": 0.6906994751785254,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009538945362649443,
      "loss": 1.1838,
      "step": 8028
    },
    {
      "epoch": 0.690785511485847,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009538831939868682,
      "loss": 0.8658,
      "step": 8029
    },
    {
      "epoch": 0.6908715477931687,
      "grad_norm": 1.78125,
      "learning_rate": 0.000953871850381272,
      "loss": 1.122,
      "step": 8030
    },
    {
      "epoch": 0.6909575841004904,
      "grad_norm": 2.171875,
      "learning_rate": 0.0009538605054481888,
      "loss": 1.1717,
      "step": 8031
    },
    {
      "epoch": 0.6910436204078121,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009538491591876521,
      "loss": 1.3578,
      "step": 8032
    },
    {
      "epoch": 0.6911296567151338,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009538378115996948,
      "loss": 1.3277,
      "step": 8033
    },
    {
      "epoch": 0.6912156930224554,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009538264626843501,
      "loss": 1.3737,
      "step": 8034
    },
    {
      "epoch": 0.6913017293297772,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009538151124416514,
      "loss": 1.3475,
      "step": 8035
    },
    {
      "epoch": 0.6913877656370988,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009538037608716317,
      "loss": 1.4104,
      "step": 8036
    },
    {
      "epoch": 0.6914738019444205,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009537924079743242,
      "loss": 1.5971,
      "step": 8037
    },
    {
      "epoch": 0.6915598382517423,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009537810537497622,
      "loss": 0.851,
      "step": 8038
    },
    {
      "epoch": 0.6916458745590639,
      "grad_norm": 1.0,
      "learning_rate": 0.0009537696981979789,
      "loss": 1.0377,
      "step": 8039
    },
    {
      "epoch": 0.6917319108663856,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009537583413190074,
      "loss": 1.9395,
      "step": 8040
    },
    {
      "epoch": 0.6918179471737073,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009537469831128811,
      "loss": 1.897,
      "step": 8041
    },
    {
      "epoch": 0.691903983481029,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009537356235796331,
      "loss": 1.5963,
      "step": 8042
    },
    {
      "epoch": 0.6919900197883507,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009537242627192967,
      "loss": 1.7749,
      "step": 8043
    },
    {
      "epoch": 0.6920760560956724,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009537129005319049,
      "loss": 1.6863,
      "step": 8044
    },
    {
      "epoch": 0.6921620924029941,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009537015370174913,
      "loss": 1.1913,
      "step": 8045
    },
    {
      "epoch": 0.6922481287103157,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009536901721760889,
      "loss": 1.7189,
      "step": 8046
    },
    {
      "epoch": 0.6923341650176374,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000953678806007731,
      "loss": 1.7975,
      "step": 8047
    },
    {
      "epoch": 0.6924202013249592,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009536674385124508,
      "loss": 1.2193,
      "step": 8048
    },
    {
      "epoch": 0.6925062376322808,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009536560696902815,
      "loss": 1.1574,
      "step": 8049
    },
    {
      "epoch": 0.6925922739396025,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009536446995412564,
      "loss": 1.0448,
      "step": 8050
    },
    {
      "epoch": 0.6926783102469242,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009536333280654089,
      "loss": 1.5903,
      "step": 8051
    },
    {
      "epoch": 0.6927643465542459,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009536219552627722,
      "loss": 1.4525,
      "step": 8052
    },
    {
      "epoch": 0.6928503828615675,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009536105811333794,
      "loss": 1.6747,
      "step": 8053
    },
    {
      "epoch": 0.6929364191688893,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009535992056772639,
      "loss": 1.4529,
      "step": 8054
    },
    {
      "epoch": 0.693022455476211,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009535878288944588,
      "loss": 1.1499,
      "step": 8055
    },
    {
      "epoch": 0.6931084917835326,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009535764507849978,
      "loss": 1.7585,
      "step": 8056
    },
    {
      "epoch": 0.6931945280908544,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009535650713489137,
      "loss": 1.0525,
      "step": 8057
    },
    {
      "epoch": 0.693280564398176,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009535536905862401,
      "loss": 1.2927,
      "step": 8058
    },
    {
      "epoch": 0.6933666007054977,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009535423084970101,
      "loss": 1.66,
      "step": 8059
    },
    {
      "epoch": 0.6934526370128195,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009535309250812569,
      "loss": 1.2789,
      "step": 8060
    },
    {
      "epoch": 0.6935386733201411,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009535195403390142,
      "loss": 1.1527,
      "step": 8061
    },
    {
      "epoch": 0.6936247096274628,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000953508154270315,
      "loss": 1.5037,
      "step": 8062
    },
    {
      "epoch": 0.6937107459347844,
      "grad_norm": 1.25,
      "learning_rate": 0.0009534967668751927,
      "loss": 1.1381,
      "step": 8063
    },
    {
      "epoch": 0.6937967822421062,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009534853781536803,
      "loss": 1.4666,
      "step": 8064
    },
    {
      "epoch": 0.6938828185494279,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009534739881058115,
      "loss": 1.1639,
      "step": 8065
    },
    {
      "epoch": 0.6939688548567495,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009534625967316195,
      "loss": 1.6578,
      "step": 8066
    },
    {
      "epoch": 0.6940548911640713,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009534512040311376,
      "loss": 2.0593,
      "step": 8067
    },
    {
      "epoch": 0.6941409274713929,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009534398100043992,
      "loss": 1.7546,
      "step": 8068
    },
    {
      "epoch": 0.6942269637787146,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009534284146514374,
      "loss": 1.2718,
      "step": 8069
    },
    {
      "epoch": 0.6943130000860364,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009534170179722858,
      "loss": 1.54,
      "step": 8070
    },
    {
      "epoch": 0.694399036393358,
      "grad_norm": 1.125,
      "learning_rate": 0.0009534056199669775,
      "loss": 1.6677,
      "step": 8071
    },
    {
      "epoch": 0.6944850727006797,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009533942206355461,
      "loss": 1.9868,
      "step": 8072
    },
    {
      "epoch": 0.6945711090080013,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009533828199780245,
      "loss": 1.0682,
      "step": 8073
    },
    {
      "epoch": 0.6946571453153231,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009533714179944466,
      "loss": 0.5372,
      "step": 8074
    },
    {
      "epoch": 0.6947431816226447,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009533600146848453,
      "loss": 1.0709,
      "step": 8075
    },
    {
      "epoch": 0.6948292179299664,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009533486100492543,
      "loss": 1.6915,
      "step": 8076
    },
    {
      "epoch": 0.6949152542372882,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009533372040877068,
      "loss": 1.5011,
      "step": 8077
    },
    {
      "epoch": 0.6950012905446098,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000953325796800236,
      "loss": 0.8933,
      "step": 8078
    },
    {
      "epoch": 0.6950873268519315,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009533143881868755,
      "loss": 2.0248,
      "step": 8079
    },
    {
      "epoch": 0.6951733631592532,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009533029782476585,
      "loss": 0.9345,
      "step": 8080
    },
    {
      "epoch": 0.6952593994665749,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009532915669826185,
      "loss": 0.8354,
      "step": 8081
    },
    {
      "epoch": 0.6953454357738966,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009532801543917889,
      "loss": 1.2425,
      "step": 8082
    },
    {
      "epoch": 0.6954314720812182,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009532687404752028,
      "loss": 1.0526,
      "step": 8083
    },
    {
      "epoch": 0.69551750838854,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000953257325232894,
      "loss": 2.0174,
      "step": 8084
    },
    {
      "epoch": 0.6956035446958616,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009532459086648956,
      "loss": 1.1721,
      "step": 8085
    },
    {
      "epoch": 0.6956895810031833,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000953234490771241,
      "loss": 1.1283,
      "step": 8086
    },
    {
      "epoch": 0.6957756173105051,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009532230715519638,
      "loss": 0.9873,
      "step": 8087
    },
    {
      "epoch": 0.6958616536178267,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009532116510070972,
      "loss": 1.2371,
      "step": 8088
    },
    {
      "epoch": 0.6959476899251484,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009532002291366746,
      "loss": 1.6635,
      "step": 8089
    },
    {
      "epoch": 0.6960337262324701,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009531888059407295,
      "loss": 1.2603,
      "step": 8090
    },
    {
      "epoch": 0.6961197625397918,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009531773814192954,
      "loss": 1.2833,
      "step": 8091
    },
    {
      "epoch": 0.6962057988471135,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009531659555724054,
      "loss": 1.7084,
      "step": 8092
    },
    {
      "epoch": 0.6962918351544352,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009531545284000933,
      "loss": 1.8879,
      "step": 8093
    },
    {
      "epoch": 0.6963778714617569,
      "grad_norm": 48.25,
      "learning_rate": 0.0009531430999023922,
      "loss": 0.7836,
      "step": 8094
    },
    {
      "epoch": 0.6964639077690785,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009531316700793357,
      "loss": 1.4438,
      "step": 8095
    },
    {
      "epoch": 0.6965499440764003,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009531202389309573,
      "loss": 1.6394,
      "step": 8096
    },
    {
      "epoch": 0.6966359803837219,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009531088064572902,
      "loss": 1.5414,
      "step": 8097
    },
    {
      "epoch": 0.6967220166910436,
      "grad_norm": 1.421875,
      "learning_rate": 0.000953097372658368,
      "loss": 2.0179,
      "step": 8098
    },
    {
      "epoch": 0.6968080529983653,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009530859375342242,
      "loss": 1.8497,
      "step": 8099
    },
    {
      "epoch": 0.696894089305687,
      "grad_norm": 1.375,
      "learning_rate": 0.000953074501084892,
      "loss": 1.6121,
      "step": 8100
    },
    {
      "epoch": 0.6969801256130087,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009530630633104052,
      "loss": 1.868,
      "step": 8101
    },
    {
      "epoch": 0.6970661619203303,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009530516242107969,
      "loss": 0.8205,
      "step": 8102
    },
    {
      "epoch": 0.6971521982276521,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009530401837861006,
      "loss": 1.3584,
      "step": 8103
    },
    {
      "epoch": 0.6972382345349738,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009530287420363502,
      "loss": 1.51,
      "step": 8104
    },
    {
      "epoch": 0.6973242708422954,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009530172989615786,
      "loss": 2.25,
      "step": 8105
    },
    {
      "epoch": 0.6974103071496172,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009530058545618196,
      "loss": 1.1666,
      "step": 8106
    },
    {
      "epoch": 0.6974963434569388,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009529944088371065,
      "loss": 0.9616,
      "step": 8107
    },
    {
      "epoch": 0.6975823797642605,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009529829617874729,
      "loss": 2.019,
      "step": 8108
    },
    {
      "epoch": 0.6976684160715823,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009529715134129522,
      "loss": 1.791,
      "step": 8109
    },
    {
      "epoch": 0.6977544523789039,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000952960063713578,
      "loss": 1.0726,
      "step": 8110
    },
    {
      "epoch": 0.6978404886862256,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009529486126893836,
      "loss": 1.1026,
      "step": 8111
    },
    {
      "epoch": 0.6979265249935472,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0009529371603404027,
      "loss": 0.705,
      "step": 8112
    },
    {
      "epoch": 0.698012561300869,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009529257066666687,
      "loss": 2.1316,
      "step": 8113
    },
    {
      "epoch": 0.6980985976081907,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000952914251668215,
      "loss": 1.2988,
      "step": 8114
    },
    {
      "epoch": 0.6981846339155123,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009529027953450753,
      "loss": 1.7528,
      "step": 8115
    },
    {
      "epoch": 0.6982706702228341,
      "grad_norm": 1.03125,
      "learning_rate": 0.000952891337697283,
      "loss": 0.7791,
      "step": 8116
    },
    {
      "epoch": 0.6983567065301557,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009528798787248716,
      "loss": 0.9997,
      "step": 8117
    },
    {
      "epoch": 0.6984427428374774,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009528684184278746,
      "loss": 0.7488,
      "step": 8118
    },
    {
      "epoch": 0.698528779144799,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009528569568063257,
      "loss": 0.8777,
      "step": 8119
    },
    {
      "epoch": 0.6986148154521208,
      "grad_norm": 1.3125,
      "learning_rate": 0.000952845493860258,
      "loss": 0.8192,
      "step": 8120
    },
    {
      "epoch": 0.6987008517594425,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009528340295897055,
      "loss": 1.0711,
      "step": 8121
    },
    {
      "epoch": 0.6987868880667641,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009528225639947016,
      "loss": 1.3514,
      "step": 8122
    },
    {
      "epoch": 0.6988729243740859,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009528110970752797,
      "loss": 1.5872,
      "step": 8123
    },
    {
      "epoch": 0.6989589606814075,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009527996288314733,
      "loss": 1.7407,
      "step": 8124
    },
    {
      "epoch": 0.6990449969887292,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009527881592633161,
      "loss": 1.9671,
      "step": 8125
    },
    {
      "epoch": 0.699131033296051,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009527766883708418,
      "loss": 1.4333,
      "step": 8126
    },
    {
      "epoch": 0.6992170696033726,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009527652161540834,
      "loss": 1.6598,
      "step": 8127
    },
    {
      "epoch": 0.6993031059106943,
      "grad_norm": 3.0,
      "learning_rate": 0.0009527537426130751,
      "loss": 1.1438,
      "step": 8128
    },
    {
      "epoch": 0.699389142218016,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009527422677478498,
      "loss": 1.1925,
      "step": 8129
    },
    {
      "epoch": 0.6994751785253377,
      "grad_norm": 1.625,
      "learning_rate": 0.0009527307915584418,
      "loss": 1.2774,
      "step": 8130
    },
    {
      "epoch": 0.6995612148326594,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009527193140448843,
      "loss": 0.8662,
      "step": 8131
    },
    {
      "epoch": 0.699647251139981,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009527078352072105,
      "loss": 2.1028,
      "step": 8132
    },
    {
      "epoch": 0.6997332874473028,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009526963550454546,
      "loss": 1.3108,
      "step": 8133
    },
    {
      "epoch": 0.6998193237546244,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009526848735596498,
      "loss": 1.3937,
      "step": 8134
    },
    {
      "epoch": 0.6999053600619461,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009526733907498298,
      "loss": 1.3345,
      "step": 8135
    },
    {
      "epoch": 0.6999913963692679,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009526619066160283,
      "loss": 1.9128,
      "step": 8136
    },
    {
      "epoch": 0.7000774326765895,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009526504211582785,
      "loss": 1.467,
      "step": 8137
    },
    {
      "epoch": 0.7001634689839112,
      "grad_norm": 2.5,
      "learning_rate": 0.0009526389343766144,
      "loss": 1.7733,
      "step": 8138
    },
    {
      "epoch": 0.7002495052912329,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009526274462710695,
      "loss": 1.3394,
      "step": 8139
    },
    {
      "epoch": 0.7003355415985546,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009526159568416774,
      "loss": 0.8807,
      "step": 8140
    },
    {
      "epoch": 0.7004215779058762,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009526044660884714,
      "loss": 1.617,
      "step": 8141
    },
    {
      "epoch": 0.700507614213198,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009525929740114856,
      "loss": 1.0937,
      "step": 8142
    },
    {
      "epoch": 0.7005936505205197,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009525814806107533,
      "loss": 1.8619,
      "step": 8143
    },
    {
      "epoch": 0.7006796868278413,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009525699858863081,
      "loss": 1.1136,
      "step": 8144
    },
    {
      "epoch": 0.700765723135163,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009525584898381838,
      "loss": 1.0949,
      "step": 8145
    },
    {
      "epoch": 0.7008517594424847,
      "grad_norm": 1.8046875,
      "learning_rate": 0.000952546992466414,
      "loss": 1.7997,
      "step": 8146
    },
    {
      "epoch": 0.7009377957498064,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009525354937710322,
      "loss": 1.4539,
      "step": 8147
    },
    {
      "epoch": 0.7010238320571281,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009525239937520721,
      "loss": 1.464,
      "step": 8148
    },
    {
      "epoch": 0.7011098683644498,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009525124924095674,
      "loss": 1.1262,
      "step": 8149
    },
    {
      "epoch": 0.7011959046717715,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009525009897435514,
      "loss": 1.6277,
      "step": 8150
    },
    {
      "epoch": 0.7012819409790931,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009524894857540582,
      "loss": 1.3943,
      "step": 8151
    },
    {
      "epoch": 0.7013679772864149,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009524779804411215,
      "loss": 1.6467,
      "step": 8152
    },
    {
      "epoch": 0.7014540135937366,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009524664738047744,
      "loss": 1.2314,
      "step": 8153
    },
    {
      "epoch": 0.7015400499010582,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009524549658450509,
      "loss": 1.5656,
      "step": 8154
    },
    {
      "epoch": 0.70162608620838,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009524434565619846,
      "loss": 1.227,
      "step": 8155
    },
    {
      "epoch": 0.7017121225157016,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009524319459556092,
      "loss": 1.4727,
      "step": 8156
    },
    {
      "epoch": 0.7017981588230233,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009524204340259586,
      "loss": 1.15,
      "step": 8157
    },
    {
      "epoch": 0.7018841951303451,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009524089207730659,
      "loss": 1.8889,
      "step": 8158
    },
    {
      "epoch": 0.7019702314376667,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009523974061969653,
      "loss": 1.0999,
      "step": 8159
    },
    {
      "epoch": 0.7020562677449884,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009523858902976901,
      "loss": 0.8971,
      "step": 8160
    },
    {
      "epoch": 0.70214230405231,
      "grad_norm": 1.0,
      "learning_rate": 0.0009523743730752743,
      "loss": 0.863,
      "step": 8161
    },
    {
      "epoch": 0.7022283403596318,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009523628545297513,
      "loss": 1.6197,
      "step": 8162
    },
    {
      "epoch": 0.7023143766669534,
      "grad_norm": 1.609375,
      "learning_rate": 0.000952351334661155,
      "loss": 1.4877,
      "step": 8163
    },
    {
      "epoch": 0.7024004129742751,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000952339813469519,
      "loss": 1.1993,
      "step": 8164
    },
    {
      "epoch": 0.7024864492815969,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009523282909548772,
      "loss": 1.6539,
      "step": 8165
    },
    {
      "epoch": 0.7025724855889185,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009523167671172629,
      "loss": 0.8732,
      "step": 8166
    },
    {
      "epoch": 0.7026585218962402,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009523052419567101,
      "loss": 1.1046,
      "step": 8167
    },
    {
      "epoch": 0.7027445582035619,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009522937154732525,
      "loss": 1.8666,
      "step": 8168
    },
    {
      "epoch": 0.7028305945108836,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009522821876669236,
      "loss": 1.573,
      "step": 8169
    },
    {
      "epoch": 0.7029166308182053,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009522706585377574,
      "loss": 2.1036,
      "step": 8170
    },
    {
      "epoch": 0.703002667125527,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009522591280857873,
      "loss": 1.6734,
      "step": 8171
    },
    {
      "epoch": 0.7030887034328487,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009522475963110474,
      "loss": 1.024,
      "step": 8172
    },
    {
      "epoch": 0.7031747397401703,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009522360632135712,
      "loss": 1.9719,
      "step": 8173
    },
    {
      "epoch": 0.703260776047492,
      "grad_norm": 1.25,
      "learning_rate": 0.0009522245287933925,
      "loss": 0.9882,
      "step": 8174
    },
    {
      "epoch": 0.7033468123548138,
      "grad_norm": 1.1875,
      "learning_rate": 0.000952212993050545,
      "loss": 1.6315,
      "step": 8175
    },
    {
      "epoch": 0.7034328486621354,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009522014559850624,
      "loss": 0.9259,
      "step": 8176
    },
    {
      "epoch": 0.7035188849694571,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009521899175969784,
      "loss": 1.3922,
      "step": 8177
    },
    {
      "epoch": 0.7036049212767788,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009521783778863269,
      "loss": 2.2743,
      "step": 8178
    },
    {
      "epoch": 0.7036909575841005,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009521668368531416,
      "loss": 1.1131,
      "step": 8179
    },
    {
      "epoch": 0.7037769938914222,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009521552944974563,
      "loss": 0.9253,
      "step": 8180
    },
    {
      "epoch": 0.7038630301987439,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009521437508193048,
      "loss": 0.7217,
      "step": 8181
    },
    {
      "epoch": 0.7039490665060656,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009521322058187205,
      "loss": 0.9553,
      "step": 8182
    },
    {
      "epoch": 0.7040351028133872,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009521206594957376,
      "loss": 2.1729,
      "step": 8183
    },
    {
      "epoch": 0.704121139120709,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009521091118503898,
      "loss": 2.055,
      "step": 8184
    },
    {
      "epoch": 0.7042071754280306,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009520975628827106,
      "loss": 1.5674,
      "step": 8185
    },
    {
      "epoch": 0.7042932117353523,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009520860125927339,
      "loss": 0.9821,
      "step": 8186
    },
    {
      "epoch": 0.704379248042674,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009520744609804937,
      "loss": 0.7966,
      "step": 8187
    },
    {
      "epoch": 0.7044652843499957,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009520629080460236,
      "loss": 1.4204,
      "step": 8188
    },
    {
      "epoch": 0.7045513206573174,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009520513537893574,
      "loss": 1.4743,
      "step": 8189
    },
    {
      "epoch": 0.704637356964639,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009520397982105289,
      "loss": 1.8217,
      "step": 8190
    },
    {
      "epoch": 0.7047233932719608,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009520282413095718,
      "loss": 1.5912,
      "step": 8191
    },
    {
      "epoch": 0.7048094295792825,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009520166830865203,
      "loss": 2.0569,
      "step": 8192
    },
    {
      "epoch": 0.7048954658866041,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009520051235414076,
      "loss": 1.2715,
      "step": 8193
    },
    {
      "epoch": 0.7049815021939259,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009519935626742679,
      "loss": 1.5466,
      "step": 8194
    },
    {
      "epoch": 0.7050675385012475,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009519820004851349,
      "loss": 1.1315,
      "step": 8195
    },
    {
      "epoch": 0.7051535748085692,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009519704369740427,
      "loss": 1.7972,
      "step": 8196
    },
    {
      "epoch": 0.705239611115891,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009519588721410246,
      "loss": 1.422,
      "step": 8197
    },
    {
      "epoch": 0.7053256474232126,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009519473059861149,
      "loss": 1.307,
      "step": 8198
    },
    {
      "epoch": 0.7054116837305343,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009519357385093471,
      "loss": 0.7365,
      "step": 8199
    },
    {
      "epoch": 0.7054977200378559,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009519241697107553,
      "loss": 1.2113,
      "step": 8200
    },
    {
      "epoch": 0.7055837563451777,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000951912599590373,
      "loss": 1.1182,
      "step": 8201
    },
    {
      "epoch": 0.7056697926524994,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009519010281482343,
      "loss": 1.5419,
      "step": 8202
    },
    {
      "epoch": 0.705755828959821,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009518894553843731,
      "loss": 1.7912,
      "step": 8203
    },
    {
      "epoch": 0.7058418652671428,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009518778812988232,
      "loss": 1.6161,
      "step": 8204
    },
    {
      "epoch": 0.7059279015744644,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009518663058916181,
      "loss": 1.144,
      "step": 8205
    },
    {
      "epoch": 0.7060139378817861,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009518547291627921,
      "loss": 1.6432,
      "step": 8206
    },
    {
      "epoch": 0.7060999741891077,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009518431511123789,
      "loss": 0.8996,
      "step": 8207
    },
    {
      "epoch": 0.7061860104964295,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009518315717404123,
      "loss": 1.7789,
      "step": 8208
    },
    {
      "epoch": 0.7062720468037512,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009518199910469263,
      "loss": 1.3806,
      "step": 8209
    },
    {
      "epoch": 0.7063580831110728,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009518084090319545,
      "loss": 1.3713,
      "step": 8210
    },
    {
      "epoch": 0.7064441194183946,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009517968256955312,
      "loss": 1.5983,
      "step": 8211
    },
    {
      "epoch": 0.7065301557257162,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00095178524103769,
      "loss": 1.3269,
      "step": 8212
    },
    {
      "epoch": 0.7066161920330379,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009517736550584649,
      "loss": 0.8889,
      "step": 8213
    },
    {
      "epoch": 0.7067022283403597,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009517620677578897,
      "loss": 1.4503,
      "step": 8214
    },
    {
      "epoch": 0.7067882646476813,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009517504791359983,
      "loss": 1.4635,
      "step": 8215
    },
    {
      "epoch": 0.706874300955003,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009517388891928244,
      "loss": 0.9786,
      "step": 8216
    },
    {
      "epoch": 0.7069603372623247,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009517272979284024,
      "loss": 1.5327,
      "step": 8217
    },
    {
      "epoch": 0.7070463735696464,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009517157053427659,
      "loss": 1.2327,
      "step": 8218
    },
    {
      "epoch": 0.7071324098769681,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009517041114359486,
      "loss": 0.8966,
      "step": 8219
    },
    {
      "epoch": 0.7072184461842898,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009516925162079849,
      "loss": 1.6047,
      "step": 8220
    },
    {
      "epoch": 0.7073044824916115,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009516809196589082,
      "loss": 1.7226,
      "step": 8221
    },
    {
      "epoch": 0.7073905187989331,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009516693217887527,
      "loss": 1.244,
      "step": 8222
    },
    {
      "epoch": 0.7074765551062548,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009516577225975523,
      "loss": 0.8856,
      "step": 8223
    },
    {
      "epoch": 0.7075625914135766,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009516461220853409,
      "loss": 1.1121,
      "step": 8224
    },
    {
      "epoch": 0.7076486277208982,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009516345202521524,
      "loss": 1.1041,
      "step": 8225
    },
    {
      "epoch": 0.7077346640282199,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009516229170980208,
      "loss": 1.6967,
      "step": 8226
    },
    {
      "epoch": 0.7078207003355416,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00095161131262298,
      "loss": 1.0832,
      "step": 8227
    },
    {
      "epoch": 0.7079067366428633,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009515997068270641,
      "loss": 1.4161,
      "step": 8228
    },
    {
      "epoch": 0.7079927729501849,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009515880997103067,
      "loss": 0.7669,
      "step": 8229
    },
    {
      "epoch": 0.7080788092575067,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009515764912727421,
      "loss": 1.7355,
      "step": 8230
    },
    {
      "epoch": 0.7081648455648284,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009515648815144038,
      "loss": 1.019,
      "step": 8231
    },
    {
      "epoch": 0.70825088187215,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009515532704353262,
      "loss": 1.1193,
      "step": 8232
    },
    {
      "epoch": 0.7083369181794718,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009515416580355431,
      "loss": 1.9865,
      "step": 8233
    },
    {
      "epoch": 0.7084229544867934,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009515300443150886,
      "loss": 0.7118,
      "step": 8234
    },
    {
      "epoch": 0.7085089907941151,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009515184292739965,
      "loss": 1.3278,
      "step": 8235
    },
    {
      "epoch": 0.7085950271014368,
      "grad_norm": 1.25,
      "learning_rate": 0.0009515068129123007,
      "loss": 0.932,
      "step": 8236
    },
    {
      "epoch": 0.7086810634087585,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009514951952300354,
      "loss": 1.0255,
      "step": 8237
    },
    {
      "epoch": 0.7087670997160802,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009514835762272344,
      "loss": 1.0731,
      "step": 8238
    },
    {
      "epoch": 0.7088531360234018,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009514719559039318,
      "loss": 1.8021,
      "step": 8239
    },
    {
      "epoch": 0.7089391723307236,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009514603342601613,
      "loss": 1.8482,
      "step": 8240
    },
    {
      "epoch": 0.7090252086380453,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009514487112959574,
      "loss": 1.5538,
      "step": 8241
    },
    {
      "epoch": 0.7091112449453669,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009514370870113538,
      "loss": 1.254,
      "step": 8242
    },
    {
      "epoch": 0.7091972812526887,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009514254614063844,
      "loss": 1.9609,
      "step": 8243
    },
    {
      "epoch": 0.7092833175600103,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009514138344810834,
      "loss": 1.0413,
      "step": 8244
    },
    {
      "epoch": 0.709369353867332,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009514022062354847,
      "loss": 0.8304,
      "step": 8245
    },
    {
      "epoch": 0.7094553901746538,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009513905766696223,
      "loss": 1.9235,
      "step": 8246
    },
    {
      "epoch": 0.7095414264819754,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009513789457835303,
      "loss": 1.2626,
      "step": 8247
    },
    {
      "epoch": 0.7096274627892971,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009513673135772427,
      "loss": 1.3684,
      "step": 8248
    },
    {
      "epoch": 0.7097134990966187,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009513556800507934,
      "loss": 1.537,
      "step": 8249
    },
    {
      "epoch": 0.7097995354039405,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009513440452042166,
      "loss": 1.5265,
      "step": 8250
    },
    {
      "epoch": 0.7098855717112621,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009513324090375463,
      "loss": 0.8875,
      "step": 8251
    },
    {
      "epoch": 0.7099716080185838,
      "grad_norm": 4.59375,
      "learning_rate": 0.0009513207715508163,
      "loss": 1.4809,
      "step": 8252
    },
    {
      "epoch": 0.7100576443259056,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009513091327440611,
      "loss": 1.0808,
      "step": 8253
    },
    {
      "epoch": 0.7101436806332272,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009512974926173142,
      "loss": 0.8443,
      "step": 8254
    },
    {
      "epoch": 0.7102297169405489,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00095128585117061,
      "loss": 1.4907,
      "step": 8255
    },
    {
      "epoch": 0.7103157532478706,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009512742084039824,
      "loss": 1.7052,
      "step": 8256
    },
    {
      "epoch": 0.7104017895551923,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009512625643174657,
      "loss": 1.3531,
      "step": 8257
    },
    {
      "epoch": 0.710487825862514,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009512509189110936,
      "loss": 1.3808,
      "step": 8258
    },
    {
      "epoch": 0.7105738621698356,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009512392721849004,
      "loss": 1.0282,
      "step": 8259
    },
    {
      "epoch": 0.7106598984771574,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009512276241389199,
      "loss": 1.548,
      "step": 8260
    },
    {
      "epoch": 0.710745934784479,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009512159747731866,
      "loss": 0.975,
      "step": 8261
    },
    {
      "epoch": 0.7108319710918007,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009512043240877342,
      "loss": 0.9495,
      "step": 8262
    },
    {
      "epoch": 0.7109180073991225,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009511926720825969,
      "loss": 1.672,
      "step": 8263
    },
    {
      "epoch": 0.7110040437064441,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009511810187578089,
      "loss": 1.3118,
      "step": 8264
    },
    {
      "epoch": 0.7110900800137658,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000951169364113404,
      "loss": 0.8582,
      "step": 8265
    },
    {
      "epoch": 0.7111761163210875,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009511577081494167,
      "loss": 1.7267,
      "step": 8266
    },
    {
      "epoch": 0.7112621526284092,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009511460508658807,
      "loss": 1.4488,
      "step": 8267
    },
    {
      "epoch": 0.7113481889357309,
      "grad_norm": 1.375,
      "learning_rate": 0.0009511343922628301,
      "loss": 0.988,
      "step": 8268
    },
    {
      "epoch": 0.7114342252430526,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009511227323402994,
      "loss": 1.4835,
      "step": 8269
    },
    {
      "epoch": 0.7115202615503743,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009511110710983224,
      "loss": 0.8037,
      "step": 8270
    },
    {
      "epoch": 0.7116062978576959,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009510994085369331,
      "loss": 1.7838,
      "step": 8271
    },
    {
      "epoch": 0.7116923341650176,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009510877446561659,
      "loss": 0.9953,
      "step": 8272
    },
    {
      "epoch": 0.7117783704723393,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009510760794560547,
      "loss": 0.9246,
      "step": 8273
    },
    {
      "epoch": 0.711864406779661,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009510644129366337,
      "loss": 1.6764,
      "step": 8274
    },
    {
      "epoch": 0.7119504430869827,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009510527450979372,
      "loss": 0.9084,
      "step": 8275
    },
    {
      "epoch": 0.7120364793943044,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000951041075939999,
      "loss": 1.5256,
      "step": 8276
    },
    {
      "epoch": 0.7121225157016261,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009510294054628533,
      "loss": 1.2282,
      "step": 8277
    },
    {
      "epoch": 0.7122085520089477,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009510177336665344,
      "loss": 1.022,
      "step": 8278
    },
    {
      "epoch": 0.7122945883162695,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009510060605510764,
      "loss": 0.5859,
      "step": 8279
    },
    {
      "epoch": 0.7123806246235912,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009509943861165133,
      "loss": 1.2527,
      "step": 8280
    },
    {
      "epoch": 0.7124666609309128,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009509827103628793,
      "loss": 1.4907,
      "step": 8281
    },
    {
      "epoch": 0.7125526972382346,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009509710332902086,
      "loss": 0.9493,
      "step": 8282
    },
    {
      "epoch": 0.7126387335455562,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009509593548985354,
      "loss": 1.2829,
      "step": 8283
    },
    {
      "epoch": 0.7127247698528779,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009509476751878938,
      "loss": 1.4111,
      "step": 8284
    },
    {
      "epoch": 0.7128108061601997,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009509359941583179,
      "loss": 1.3001,
      "step": 8285
    },
    {
      "epoch": 0.7128968424675213,
      "grad_norm": 1.203125,
      "learning_rate": 0.000950924311809842,
      "loss": 1.4506,
      "step": 8286
    },
    {
      "epoch": 0.712982878774843,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009509126281425,
      "loss": 0.7143,
      "step": 8287
    },
    {
      "epoch": 0.7130689150821646,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009509009431563264,
      "loss": 1.2198,
      "step": 8288
    },
    {
      "epoch": 0.7131549513894864,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009508892568513553,
      "loss": 1.2699,
      "step": 8289
    },
    {
      "epoch": 0.7132409876968081,
      "grad_norm": 1.625,
      "learning_rate": 0.0009508775692276208,
      "loss": 1.6903,
      "step": 8290
    },
    {
      "epoch": 0.7133270240041297,
      "grad_norm": 1.203125,
      "learning_rate": 0.000950865880285157,
      "loss": 1.4092,
      "step": 8291
    },
    {
      "epoch": 0.7134130603114515,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009508541900239982,
      "loss": 1.4473,
      "step": 8292
    },
    {
      "epoch": 0.7134990966187731,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009508424984441785,
      "loss": 2.1927,
      "step": 8293
    },
    {
      "epoch": 0.7135851329260948,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009508308055457324,
      "loss": 1.3012,
      "step": 8294
    },
    {
      "epoch": 0.7136711692334164,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009508191113286937,
      "loss": 0.952,
      "step": 8295
    },
    {
      "epoch": 0.7137572055407382,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009508074157930969,
      "loss": 1.1783,
      "step": 8296
    },
    {
      "epoch": 0.7138432418480599,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000950795718938976,
      "loss": 2.1327,
      "step": 8297
    },
    {
      "epoch": 0.7139292781553815,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009507840207663654,
      "loss": 1.4267,
      "step": 8298
    },
    {
      "epoch": 0.7140153144627033,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009507723212752991,
      "loss": 1.7822,
      "step": 8299
    },
    {
      "epoch": 0.7141013507700249,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009507606204658115,
      "loss": 1.707,
      "step": 8300
    },
    {
      "epoch": 0.7141873870773466,
      "grad_norm": 1.5,
      "learning_rate": 0.0009507489183379368,
      "loss": 1.0206,
      "step": 8301
    },
    {
      "epoch": 0.7142734233846684,
      "grad_norm": 1.640625,
      "learning_rate": 0.000950737214891709,
      "loss": 1.0816,
      "step": 8302
    },
    {
      "epoch": 0.71435945969199,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009507255101271625,
      "loss": 1.3658,
      "step": 8303
    },
    {
      "epoch": 0.7144454959993117,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009507138040443317,
      "loss": 1.9152,
      "step": 8304
    },
    {
      "epoch": 0.7145315323066334,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009507020966432507,
      "loss": 1.2097,
      "step": 8305
    },
    {
      "epoch": 0.7146175686139551,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009506903879239537,
      "loss": 1.2018,
      "step": 8306
    },
    {
      "epoch": 0.7147036049212768,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009506786778864749,
      "loss": 1.6259,
      "step": 8307
    },
    {
      "epoch": 0.7147896412285984,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009506669665308487,
      "loss": 1.5975,
      "step": 8308
    },
    {
      "epoch": 0.7148756775359202,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009506552538571092,
      "loss": 1.3077,
      "step": 8309
    },
    {
      "epoch": 0.7149617138432418,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009506435398652907,
      "loss": 1.3813,
      "step": 8310
    },
    {
      "epoch": 0.7150477501505635,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009506318245554275,
      "loss": 1.4179,
      "step": 8311
    },
    {
      "epoch": 0.7151337864578853,
      "grad_norm": 1.4375,
      "learning_rate": 0.000950620107927554,
      "loss": 0.98,
      "step": 8312
    },
    {
      "epoch": 0.7152198227652069,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009506083899817042,
      "loss": 1.88,
      "step": 8313
    },
    {
      "epoch": 0.7153058590725286,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009505966707179126,
      "loss": 0.8313,
      "step": 8314
    },
    {
      "epoch": 0.7153918953798503,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009505849501362133,
      "loss": 1.8883,
      "step": 8315
    },
    {
      "epoch": 0.715477931687172,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009505732282366407,
      "loss": 0.9968,
      "step": 8316
    },
    {
      "epoch": 0.7155639679944936,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009505615050192289,
      "loss": 0.6986,
      "step": 8317
    },
    {
      "epoch": 0.7156500043018154,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009505497804840125,
      "loss": 1.7312,
      "step": 8318
    },
    {
      "epoch": 0.7157360406091371,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009505380546310254,
      "loss": 1.4293,
      "step": 8319
    },
    {
      "epoch": 0.7158220769164587,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009505263274603022,
      "loss": 1.4,
      "step": 8320
    },
    {
      "epoch": 0.7159081132237805,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009505145989718773,
      "loss": 1.1747,
      "step": 8321
    },
    {
      "epoch": 0.7159941495311021,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009505028691657847,
      "loss": 2.4036,
      "step": 8322
    },
    {
      "epoch": 0.7160801858384238,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009504911380420588,
      "loss": 0.8164,
      "step": 8323
    },
    {
      "epoch": 0.7161662221457455,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009504794056007339,
      "loss": 1.2007,
      "step": 8324
    },
    {
      "epoch": 0.7162522584530672,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009504676718418444,
      "loss": 2.0429,
      "step": 8325
    },
    {
      "epoch": 0.7163382947603889,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0009504559367654247,
      "loss": 1.1146,
      "step": 8326
    },
    {
      "epoch": 0.7164243310677105,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009504442003715088,
      "loss": 1.5844,
      "step": 8327
    },
    {
      "epoch": 0.7165103673750323,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009504324626601313,
      "loss": 1.4407,
      "step": 8328
    },
    {
      "epoch": 0.716596403682354,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009504207236313264,
      "loss": 1.3004,
      "step": 8329
    },
    {
      "epoch": 0.7166824399896756,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009504089832851286,
      "loss": 1.6633,
      "step": 8330
    },
    {
      "epoch": 0.7167684762969974,
      "grad_norm": 1.046875,
      "learning_rate": 0.000950397241621572,
      "loss": 1.1593,
      "step": 8331
    },
    {
      "epoch": 0.716854512604319,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000950385498640691,
      "loss": 1.9539,
      "step": 8332
    },
    {
      "epoch": 0.7169405489116407,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009503737543425201,
      "loss": 1.5606,
      "step": 8333
    },
    {
      "epoch": 0.7170265852189625,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009503620087270937,
      "loss": 0.8218,
      "step": 8334
    },
    {
      "epoch": 0.7171126215262841,
      "grad_norm": 2.4375,
      "learning_rate": 0.0009503502617944459,
      "loss": 1.5407,
      "step": 8335
    },
    {
      "epoch": 0.7171986578336058,
      "grad_norm": 1.25,
      "learning_rate": 0.0009503385135446112,
      "loss": 1.4283,
      "step": 8336
    },
    {
      "epoch": 0.7172846941409274,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009503267639776238,
      "loss": 1.4877,
      "step": 8337
    },
    {
      "epoch": 0.7173707304482492,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009503150130935183,
      "loss": 1.1667,
      "step": 8338
    },
    {
      "epoch": 0.7174567667555708,
      "grad_norm": 14.6875,
      "learning_rate": 0.000950303260892329,
      "loss": 2.3435,
      "step": 8339
    },
    {
      "epoch": 0.7175428030628925,
      "grad_norm": 2.53125,
      "learning_rate": 0.0009502915073740904,
      "loss": 1.5714,
      "step": 8340
    },
    {
      "epoch": 0.7176288393702143,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009502797525388363,
      "loss": 1.8233,
      "step": 8341
    },
    {
      "epoch": 0.7177148756775359,
      "grad_norm": 31.875,
      "learning_rate": 0.0009502679963866019,
      "loss": 1.4579,
      "step": 8342
    },
    {
      "epoch": 0.7178009119848576,
      "grad_norm": 2.734375,
      "learning_rate": 0.000950256238917421,
      "loss": 1.1186,
      "step": 8343
    },
    {
      "epoch": 0.7178869482921792,
      "grad_norm": 6.375,
      "learning_rate": 0.0009502444801313282,
      "loss": 1.1366,
      "step": 8344
    },
    {
      "epoch": 0.717972984599501,
      "grad_norm": 1.25,
      "learning_rate": 0.0009502327200283579,
      "loss": 1.3083,
      "step": 8345
    },
    {
      "epoch": 0.7180590209068227,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009502209586085443,
      "loss": 1.4805,
      "step": 8346
    },
    {
      "epoch": 0.7181450572141443,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009502091958719221,
      "loss": 1.5736,
      "step": 8347
    },
    {
      "epoch": 0.7182310935214661,
      "grad_norm": 51.5,
      "learning_rate": 0.0009501974318185254,
      "loss": 0.8756,
      "step": 8348
    },
    {
      "epoch": 0.7183171298287877,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000950185666448389,
      "loss": 1.5599,
      "step": 8349
    },
    {
      "epoch": 0.7184031661361094,
      "grad_norm": 1.3125,
      "learning_rate": 0.000950173899761547,
      "loss": 1.6831,
      "step": 8350
    },
    {
      "epoch": 0.7184892024434312,
      "grad_norm": 3.125,
      "learning_rate": 0.0009501621317580338,
      "loss": 1.5442,
      "step": 8351
    },
    {
      "epoch": 0.7185752387507528,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000950150362437884,
      "loss": 0.8375,
      "step": 8352
    },
    {
      "epoch": 0.7186612750580745,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009501385918011321,
      "loss": 1.7414,
      "step": 8353
    },
    {
      "epoch": 0.7187473113653962,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0009501268198478122,
      "loss": 0.635,
      "step": 8354
    },
    {
      "epoch": 0.7188333476727179,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009501150465779588,
      "loss": 0.7183,
      "step": 8355
    },
    {
      "epoch": 0.7189193839800396,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009501032719916065,
      "loss": 2.0221,
      "step": 8356
    },
    {
      "epoch": 0.7190054202873613,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009500914960887896,
      "loss": 0.9969,
      "step": 8357
    },
    {
      "epoch": 0.719091456594683,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009500797188695428,
      "loss": 1.1254,
      "step": 8358
    },
    {
      "epoch": 0.7191774929020046,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009500679403339003,
      "loss": 1.3524,
      "step": 8359
    },
    {
      "epoch": 0.7192635292093263,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009500561604818964,
      "loss": 3.3137,
      "step": 8360
    },
    {
      "epoch": 0.719349565516648,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009500443793135661,
      "loss": 0.9405,
      "step": 8361
    },
    {
      "epoch": 0.7194356018239697,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009500325968289434,
      "loss": 1.616,
      "step": 8362
    },
    {
      "epoch": 0.7195216381312914,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009500208130280627,
      "loss": 0.7496,
      "step": 8363
    },
    {
      "epoch": 0.7196076744386131,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009500090279109587,
      "loss": 1.8081,
      "step": 8364
    },
    {
      "epoch": 0.7196937107459348,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000949997241477666,
      "loss": 1.7822,
      "step": 8365
    },
    {
      "epoch": 0.7197797470532564,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009499854537282188,
      "loss": 1.3552,
      "step": 8366
    },
    {
      "epoch": 0.7198657833605782,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009499736646626517,
      "loss": 1.4157,
      "step": 8367
    },
    {
      "epoch": 0.7199518196678999,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009499618742809991,
      "loss": 0.8952,
      "step": 8368
    },
    {
      "epoch": 0.7200378559752215,
      "grad_norm": 1.25,
      "learning_rate": 0.0009499500825832955,
      "loss": 1.5625,
      "step": 8369
    },
    {
      "epoch": 0.7201238922825433,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009499382895695755,
      "loss": 0.8454,
      "step": 8370
    },
    {
      "epoch": 0.7202099285898649,
      "grad_norm": 1.125,
      "learning_rate": 0.0009499264952398734,
      "loss": 0.9993,
      "step": 8371
    },
    {
      "epoch": 0.7202959648971866,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009499146995942239,
      "loss": 1.1865,
      "step": 8372
    },
    {
      "epoch": 0.7203820012045083,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009499029026326613,
      "loss": 1.1357,
      "step": 8373
    },
    {
      "epoch": 0.72046803751183,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009498911043552203,
      "loss": 1.1975,
      "step": 8374
    },
    {
      "epoch": 0.7205540738191517,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009498793047619354,
      "loss": 0.8661,
      "step": 8375
    },
    {
      "epoch": 0.7206401101264733,
      "grad_norm": 2.53125,
      "learning_rate": 0.0009498675038528408,
      "loss": 1.5464,
      "step": 8376
    },
    {
      "epoch": 0.7207261464337951,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009498557016279714,
      "loss": 1.4077,
      "step": 8377
    },
    {
      "epoch": 0.7208121827411168,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009498438980873616,
      "loss": 1.8038,
      "step": 8378
    },
    {
      "epoch": 0.7208982190484384,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009498320932310458,
      "loss": 0.9907,
      "step": 8379
    },
    {
      "epoch": 0.7209842553557602,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009498202870590587,
      "loss": 1.2971,
      "step": 8380
    },
    {
      "epoch": 0.7210702916630818,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009498084795714347,
      "loss": 1.6962,
      "step": 8381
    },
    {
      "epoch": 0.7211563279704035,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009497966707682084,
      "loss": 0.8285,
      "step": 8382
    },
    {
      "epoch": 0.7212423642777251,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009497848606494143,
      "loss": 1.7074,
      "step": 8383
    },
    {
      "epoch": 0.7213284005850469,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009497730492150869,
      "loss": 1.2238,
      "step": 8384
    },
    {
      "epoch": 0.7214144368923686,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009497612364652608,
      "loss": 1.4355,
      "step": 8385
    },
    {
      "epoch": 0.7215004731996902,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009497494223999707,
      "loss": 1.1395,
      "step": 8386
    },
    {
      "epoch": 0.721586509507012,
      "grad_norm": 15.25,
      "learning_rate": 0.0009497376070192509,
      "loss": 1.6075,
      "step": 8387
    },
    {
      "epoch": 0.7216725458143336,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009497257903231362,
      "loss": 1.831,
      "step": 8388
    },
    {
      "epoch": 0.7217585821216553,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009497139723116608,
      "loss": 2.298,
      "step": 8389
    },
    {
      "epoch": 0.7218446184289771,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009497021529848598,
      "loss": 0.8937,
      "step": 8390
    },
    {
      "epoch": 0.7219306547362987,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009496903323427671,
      "loss": 1.7342,
      "step": 8391
    },
    {
      "epoch": 0.7220166910436204,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009496785103854179,
      "loss": 1.7395,
      "step": 8392
    },
    {
      "epoch": 0.722102727350942,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009496666871128464,
      "loss": 1.5917,
      "step": 8393
    },
    {
      "epoch": 0.7221887636582638,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009496548625250873,
      "loss": 0.8881,
      "step": 8394
    },
    {
      "epoch": 0.7222747999655855,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009496430366221754,
      "loss": 1.5129,
      "step": 8395
    },
    {
      "epoch": 0.7223608362729071,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009496312094041448,
      "loss": 0.9119,
      "step": 8396
    },
    {
      "epoch": 0.7224468725802289,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009496193808710304,
      "loss": 0.986,
      "step": 8397
    },
    {
      "epoch": 0.7225329088875505,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009496075510228668,
      "loss": 1.2033,
      "step": 8398
    },
    {
      "epoch": 0.7226189451948722,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009495957198596885,
      "loss": 1.3793,
      "step": 8399
    },
    {
      "epoch": 0.722704981502194,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009495838873815304,
      "loss": 1.3894,
      "step": 8400
    },
    {
      "epoch": 0.7227910178095156,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009495720535884265,
      "loss": 1.4675,
      "step": 8401
    },
    {
      "epoch": 0.7228770541168373,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009495602184804119,
      "loss": 1.4825,
      "step": 8402
    },
    {
      "epoch": 0.722963090424159,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009495483820575213,
      "loss": 1.0791,
      "step": 8403
    },
    {
      "epoch": 0.7230491267314807,
      "grad_norm": 1.1875,
      "learning_rate": 0.000949536544319789,
      "loss": 1.7007,
      "step": 8404
    },
    {
      "epoch": 0.7231351630388023,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009495247052672498,
      "loss": 1.7088,
      "step": 8405
    },
    {
      "epoch": 0.723221199346124,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000949512864899938,
      "loss": 1.1821,
      "step": 8406
    },
    {
      "epoch": 0.7233072356534458,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009495010232178888,
      "loss": 1.6924,
      "step": 8407
    },
    {
      "epoch": 0.7233932719607674,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009494891802211364,
      "loss": 1.0417,
      "step": 8408
    },
    {
      "epoch": 0.7234793082680891,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009494773359097155,
      "loss": 1.6767,
      "step": 8409
    },
    {
      "epoch": 0.7235653445754108,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009494654902836609,
      "loss": 0.9381,
      "step": 8410
    },
    {
      "epoch": 0.7236513808827325,
      "grad_norm": 1.125,
      "learning_rate": 0.0009494536433430072,
      "loss": 1.134,
      "step": 8411
    },
    {
      "epoch": 0.7237374171900542,
      "grad_norm": 1.03125,
      "learning_rate": 0.000949441795087789,
      "loss": 1.3862,
      "step": 8412
    },
    {
      "epoch": 0.7238234534973759,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009494299455180408,
      "loss": 1.1993,
      "step": 8413
    },
    {
      "epoch": 0.7239094898046976,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009494180946337976,
      "loss": 1.1916,
      "step": 8414
    },
    {
      "epoch": 0.7239955261120192,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009494062424350938,
      "loss": 2.0597,
      "step": 8415
    },
    {
      "epoch": 0.724081562419341,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009493943889219641,
      "loss": 1.2404,
      "step": 8416
    },
    {
      "epoch": 0.7241675987266627,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009493825340944433,
      "loss": 1.4354,
      "step": 8417
    },
    {
      "epoch": 0.7242536350339843,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009493706779525659,
      "loss": 1.2512,
      "step": 8418
    },
    {
      "epoch": 0.7243396713413061,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009493588204963668,
      "loss": 1.6449,
      "step": 8419
    },
    {
      "epoch": 0.7244257076486277,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009493469617258805,
      "loss": 0.9068,
      "step": 8420
    },
    {
      "epoch": 0.7245117439559494,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009493351016411415,
      "loss": 1.578,
      "step": 8421
    },
    {
      "epoch": 0.7245977802632712,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000949323240242185,
      "loss": 1.4147,
      "step": 8422
    },
    {
      "epoch": 0.7246838165705928,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009493113775290453,
      "loss": 1.253,
      "step": 8423
    },
    {
      "epoch": 0.7247698528779145,
      "grad_norm": 1.375,
      "learning_rate": 0.0009492995135017574,
      "loss": 1.554,
      "step": 8424
    },
    {
      "epoch": 0.7248558891852361,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009492876481603556,
      "loss": 0.9328,
      "step": 8425
    },
    {
      "epoch": 0.7249419254925579,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0009492757815048748,
      "loss": 0.6587,
      "step": 8426
    },
    {
      "epoch": 0.7250279617998795,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009492639135353498,
      "loss": 1.2211,
      "step": 8427
    },
    {
      "epoch": 0.7251139981072012,
      "grad_norm": 1.25,
      "learning_rate": 0.0009492520442518152,
      "loss": 1.278,
      "step": 8428
    },
    {
      "epoch": 0.725200034414523,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009492401736543057,
      "loss": 0.9581,
      "step": 8429
    },
    {
      "epoch": 0.7252860707218446,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009492283017428563,
      "loss": 1.8437,
      "step": 8430
    },
    {
      "epoch": 0.7253721070291663,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009492164285175011,
      "loss": 0.8246,
      "step": 8431
    },
    {
      "epoch": 0.725458143336488,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009492045539782754,
      "loss": 1.186,
      "step": 8432
    },
    {
      "epoch": 0.7255441796438097,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009491926781252138,
      "loss": 1.6071,
      "step": 8433
    },
    {
      "epoch": 0.7256302159511314,
      "grad_norm": 1.15625,
      "learning_rate": 0.000949180800958351,
      "loss": 1.7084,
      "step": 8434
    },
    {
      "epoch": 0.725716252258453,
      "grad_norm": 1.25,
      "learning_rate": 0.0009491689224777216,
      "loss": 1.1477,
      "step": 8435
    },
    {
      "epoch": 0.7258022885657748,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009491570426833606,
      "loss": 0.7671,
      "step": 8436
    },
    {
      "epoch": 0.7258883248730964,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009491451615753024,
      "loss": 1.072,
      "step": 8437
    },
    {
      "epoch": 0.7259743611804181,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009491332791535821,
      "loss": 0.9018,
      "step": 8438
    },
    {
      "epoch": 0.7260603974877399,
      "grad_norm": 1.375,
      "learning_rate": 0.0009491213954182343,
      "loss": 0.9962,
      "step": 8439
    },
    {
      "epoch": 0.7261464337950615,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009491095103692938,
      "loss": 1.717,
      "step": 8440
    },
    {
      "epoch": 0.7262324701023832,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009490976240067952,
      "loss": 2.011,
      "step": 8441
    },
    {
      "epoch": 0.7263185064097049,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009490857363307736,
      "loss": 1.2485,
      "step": 8442
    },
    {
      "epoch": 0.7264045427170266,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009490738473412635,
      "loss": 2.3573,
      "step": 8443
    },
    {
      "epoch": 0.7264905790243483,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009490619570382996,
      "loss": 1.6344,
      "step": 8444
    },
    {
      "epoch": 0.72657661533167,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009490500654219168,
      "loss": 1.6541,
      "step": 8445
    },
    {
      "epoch": 0.7266626516389917,
      "grad_norm": 1.4375,
      "learning_rate": 0.00094903817249215,
      "loss": 1.0277,
      "step": 8446
    },
    {
      "epoch": 0.7267486879463133,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000949026278249034,
      "loss": 1.1012,
      "step": 8447
    },
    {
      "epoch": 0.726834724253635,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009490143826926032,
      "loss": 1.0768,
      "step": 8448
    },
    {
      "epoch": 0.7269207605609567,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0009490024858228929,
      "loss": 0.6595,
      "step": 8449
    },
    {
      "epoch": 0.7270067968682784,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009489905876399375,
      "loss": 0.7354,
      "step": 8450
    },
    {
      "epoch": 0.7270928331756001,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000948978688143772,
      "loss": 1.6623,
      "step": 8451
    },
    {
      "epoch": 0.7271788694829218,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009489667873344312,
      "loss": 1.8654,
      "step": 8452
    },
    {
      "epoch": 0.7272649057902435,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009489548852119499,
      "loss": 1.3536,
      "step": 8453
    },
    {
      "epoch": 0.7273509420975651,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009489429817763626,
      "loss": 1.5849,
      "step": 8454
    },
    {
      "epoch": 0.7274369784048869,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009489310770277046,
      "loss": 1.5999,
      "step": 8455
    },
    {
      "epoch": 0.7275230147122086,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009489191709660106,
      "loss": 0.9145,
      "step": 8456
    },
    {
      "epoch": 0.7276090510195302,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009489072635913151,
      "loss": 1.453,
      "step": 8457
    },
    {
      "epoch": 0.727695087326852,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009488953549036534,
      "loss": 0.9909,
      "step": 8458
    },
    {
      "epoch": 0.7277811236341736,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009488834449030601,
      "loss": 0.8019,
      "step": 8459
    },
    {
      "epoch": 0.7278671599414953,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009488715335895699,
      "loss": 1.2895,
      "step": 8460
    },
    {
      "epoch": 0.727953196248817,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009488596209632177,
      "loss": 0.8384,
      "step": 8461
    },
    {
      "epoch": 0.7280392325561387,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009488477070240385,
      "loss": 2.0124,
      "step": 8462
    },
    {
      "epoch": 0.7281252688634604,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009488357917720672,
      "loss": 2.1571,
      "step": 8463
    },
    {
      "epoch": 0.728211305170782,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009488238752073382,
      "loss": 1.5608,
      "step": 8464
    },
    {
      "epoch": 0.7282973414781038,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009488119573298868,
      "loss": 2.0335,
      "step": 8465
    },
    {
      "epoch": 0.7283833777854255,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009488000381397478,
      "loss": 1.1565,
      "step": 8466
    },
    {
      "epoch": 0.7284694140927471,
      "grad_norm": 1.375,
      "learning_rate": 0.0009487881176369559,
      "loss": 1.0006,
      "step": 8467
    },
    {
      "epoch": 0.7285554504000689,
      "grad_norm": 1.359375,
      "learning_rate": 0.000948776195821546,
      "loss": 1.8333,
      "step": 8468
    },
    {
      "epoch": 0.7286414867073905,
      "grad_norm": 1.8125,
      "learning_rate": 0.000948764272693553,
      "loss": 1.596,
      "step": 8469
    },
    {
      "epoch": 0.7287275230147122,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009487523482530119,
      "loss": 1.0417,
      "step": 8470
    },
    {
      "epoch": 0.7288135593220338,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009487404224999574,
      "loss": 1.3527,
      "step": 8471
    },
    {
      "epoch": 0.7288995956293556,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009487284954344244,
      "loss": 1.1223,
      "step": 8472
    },
    {
      "epoch": 0.7289856319366773,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000948716567056448,
      "loss": 1.2337,
      "step": 8473
    },
    {
      "epoch": 0.7290716682439989,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009487046373660626,
      "loss": 0.7147,
      "step": 8474
    },
    {
      "epoch": 0.7291577045513207,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009486927063633036,
      "loss": 1.2178,
      "step": 8475
    },
    {
      "epoch": 0.7292437408586423,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009486807740482058,
      "loss": 2.0588,
      "step": 8476
    },
    {
      "epoch": 0.729329777165964,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009486688404208038,
      "loss": 1.0947,
      "step": 8477
    },
    {
      "epoch": 0.7294158134732858,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009486569054811327,
      "loss": 1.9876,
      "step": 8478
    },
    {
      "epoch": 0.7295018497806074,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009486449692292275,
      "loss": 1.3072,
      "step": 8479
    },
    {
      "epoch": 0.7295878860879291,
      "grad_norm": 1.34375,
      "learning_rate": 0.000948633031665123,
      "loss": 1.4693,
      "step": 8480
    },
    {
      "epoch": 0.7296739223952508,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009486210927888542,
      "loss": 1.0703,
      "step": 8481
    },
    {
      "epoch": 0.7297599587025725,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009486091526004559,
      "loss": 0.7528,
      "step": 8482
    },
    {
      "epoch": 0.7298459950098942,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000948597211099963,
      "loss": 1.663,
      "step": 8483
    },
    {
      "epoch": 0.7299320313172158,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009485852682874106,
      "loss": 1.7321,
      "step": 8484
    },
    {
      "epoch": 0.7300180676245376,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009485733241628335,
      "loss": 1.2192,
      "step": 8485
    },
    {
      "epoch": 0.7301041039318592,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009485613787262667,
      "loss": 1.5989,
      "step": 8486
    },
    {
      "epoch": 0.7301901402391809,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000948549431977745,
      "loss": 1.1179,
      "step": 8487
    },
    {
      "epoch": 0.7302761765465027,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009485374839173035,
      "loss": 0.907,
      "step": 8488
    },
    {
      "epoch": 0.7303622128538243,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009485255345449773,
      "loss": 1.4693,
      "step": 8489
    },
    {
      "epoch": 0.730448249161146,
      "grad_norm": 1.328125,
      "learning_rate": 0.000948513583860801,
      "loss": 1.2411,
      "step": 8490
    },
    {
      "epoch": 0.7305342854684677,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009485016318648096,
      "loss": 0.6988,
      "step": 8491
    },
    {
      "epoch": 0.7306203217757894,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009484896785570383,
      "loss": 1.3854,
      "step": 8492
    },
    {
      "epoch": 0.730706358083111,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009484777239375218,
      "loss": 1.3396,
      "step": 8493
    },
    {
      "epoch": 0.7307923943904328,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009484657680062952,
      "loss": 0.8207,
      "step": 8494
    },
    {
      "epoch": 0.7308784306977545,
      "grad_norm": 1.5,
      "learning_rate": 0.0009484538107633935,
      "loss": 1.9592,
      "step": 8495
    },
    {
      "epoch": 0.7309644670050761,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009484418522088516,
      "loss": 1.6314,
      "step": 8496
    },
    {
      "epoch": 0.7310505033123978,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009484298923427045,
      "loss": 1.4412,
      "step": 8497
    },
    {
      "epoch": 0.7311365396197195,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009484179311649872,
      "loss": 0.9078,
      "step": 8498
    },
    {
      "epoch": 0.7312225759270412,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009484059686757347,
      "loss": 1.9711,
      "step": 8499
    },
    {
      "epoch": 0.7313086122343629,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009483940048749818,
      "loss": 0.9172,
      "step": 8500
    },
    {
      "epoch": 0.7313946485416846,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009483820397627638,
      "loss": 1.425,
      "step": 8501
    },
    {
      "epoch": 0.7314806848490063,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009483700733391154,
      "loss": 1.3252,
      "step": 8502
    },
    {
      "epoch": 0.7315667211563279,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009483581056040719,
      "loss": 1.3112,
      "step": 8503
    },
    {
      "epoch": 0.7316527574636497,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000948346136557668,
      "loss": 1.3293,
      "step": 8504
    },
    {
      "epoch": 0.7317387937709714,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009483341661999389,
      "loss": 1.5685,
      "step": 8505
    },
    {
      "epoch": 0.731824830078293,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009483221945309196,
      "loss": 1.0782,
      "step": 8506
    },
    {
      "epoch": 0.7319108663856148,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009483102215506451,
      "loss": 1.0464,
      "step": 8507
    },
    {
      "epoch": 0.7319969026929364,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009482982472591504,
      "loss": 1.8472,
      "step": 8508
    },
    {
      "epoch": 0.7320829390002581,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009482862716564704,
      "loss": 1.7876,
      "step": 8509
    },
    {
      "epoch": 0.7321689753075799,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009482742947426403,
      "loss": 1.1269,
      "step": 8510
    },
    {
      "epoch": 0.7322550116149015,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000948262316517695,
      "loss": 1.2285,
      "step": 8511
    },
    {
      "epoch": 0.7323410479222232,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009482503369816698,
      "loss": 1.7403,
      "step": 8512
    },
    {
      "epoch": 0.7324270842295448,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009482383561345993,
      "loss": 1.046,
      "step": 8513
    },
    {
      "epoch": 0.7325131205368666,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009482263739765188,
      "loss": 1.4357,
      "step": 8514
    },
    {
      "epoch": 0.7325991568441882,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009482143905074636,
      "loss": 1.5579,
      "step": 8515
    },
    {
      "epoch": 0.7326851931515099,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009482024057274683,
      "loss": 1.104,
      "step": 8516
    },
    {
      "epoch": 0.7327712294588317,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009481904196365681,
      "loss": 1.5986,
      "step": 8517
    },
    {
      "epoch": 0.7328572657661533,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009481784322347981,
      "loss": 1.9516,
      "step": 8518
    },
    {
      "epoch": 0.732943302073475,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009481664435221935,
      "loss": 1.219,
      "step": 8519
    },
    {
      "epoch": 0.7330293383807966,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009481544534987892,
      "loss": 0.8202,
      "step": 8520
    },
    {
      "epoch": 0.7331153746881184,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009481424621646201,
      "loss": 0.7952,
      "step": 8521
    },
    {
      "epoch": 0.7332014109954401,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009481304695197214,
      "loss": 2.0211,
      "step": 8522
    },
    {
      "epoch": 0.7332874473027617,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009481184755641285,
      "loss": 1.7194,
      "step": 8523
    },
    {
      "epoch": 0.7333734836100835,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009481064802978761,
      "loss": 1.5946,
      "step": 8524
    },
    {
      "epoch": 0.7334595199174051,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009480944837209995,
      "loss": 1.6435,
      "step": 8525
    },
    {
      "epoch": 0.7335455562247268,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009480824858335335,
      "loss": 0.9383,
      "step": 8526
    },
    {
      "epoch": 0.7336315925320486,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009480704866355134,
      "loss": 1.002,
      "step": 8527
    },
    {
      "epoch": 0.7337176288393702,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009480584861269744,
      "loss": 2.3003,
      "step": 8528
    },
    {
      "epoch": 0.7338036651466919,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009480464843079515,
      "loss": 1.0556,
      "step": 8529
    },
    {
      "epoch": 0.7338897014540136,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009480344811784796,
      "loss": 0.9362,
      "step": 8530
    },
    {
      "epoch": 0.7339757377613353,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009480224767385941,
      "loss": 1.3424,
      "step": 8531
    },
    {
      "epoch": 0.734061774068657,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009480104709883299,
      "loss": 1.3404,
      "step": 8532
    },
    {
      "epoch": 0.7341478103759786,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009479984639277221,
      "loss": 1.0161,
      "step": 8533
    },
    {
      "epoch": 0.7342338466833004,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009479864555568062,
      "loss": 1.3483,
      "step": 8534
    },
    {
      "epoch": 0.734319882990622,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000947974445875617,
      "loss": 1.4815,
      "step": 8535
    },
    {
      "epoch": 0.7344059192979437,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009479624348841895,
      "loss": 1.9835,
      "step": 8536
    },
    {
      "epoch": 0.7344919556052654,
      "grad_norm": 1.0,
      "learning_rate": 0.0009479504225825591,
      "loss": 1.4619,
      "step": 8537
    },
    {
      "epoch": 0.7345779919125871,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009479384089707607,
      "loss": 0.8775,
      "step": 8538
    },
    {
      "epoch": 0.7346640282199088,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009479263940488299,
      "loss": 1.0717,
      "step": 8539
    },
    {
      "epoch": 0.7347500645272305,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009479143778168012,
      "loss": 1.4055,
      "step": 8540
    },
    {
      "epoch": 0.7348361008345522,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009479023602747103,
      "loss": 1.2069,
      "step": 8541
    },
    {
      "epoch": 0.7349221371418738,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009478903414225919,
      "loss": 1.8256,
      "step": 8542
    },
    {
      "epoch": 0.7350081734491956,
      "grad_norm": 1.125,
      "learning_rate": 0.0009478783212604814,
      "loss": 1.3009,
      "step": 8543
    },
    {
      "epoch": 0.7350942097565173,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009478662997884139,
      "loss": 0.8742,
      "step": 8544
    },
    {
      "epoch": 0.7351802460638389,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009478542770064248,
      "loss": 0.8259,
      "step": 8545
    },
    {
      "epoch": 0.7352662823711607,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009478422529145489,
      "loss": 1.34,
      "step": 8546
    },
    {
      "epoch": 0.7353523186784823,
      "grad_norm": 1.375,
      "learning_rate": 0.0009478302275128214,
      "loss": 1.7862,
      "step": 8547
    },
    {
      "epoch": 0.735438354985804,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009478182008012776,
      "loss": 1.5533,
      "step": 8548
    },
    {
      "epoch": 0.7355243912931257,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009478061727799528,
      "loss": 1.3568,
      "step": 8549
    },
    {
      "epoch": 0.7356104276004474,
      "grad_norm": 2.09375,
      "learning_rate": 0.0009477941434488821,
      "loss": 1.3679,
      "step": 8550
    },
    {
      "epoch": 0.7356964639077691,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009477821128081004,
      "loss": 2.2141,
      "step": 8551
    },
    {
      "epoch": 0.7357825002150907,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009477700808576433,
      "loss": 0.7632,
      "step": 8552
    },
    {
      "epoch": 0.7358685365224125,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009477580475975456,
      "loss": 1.7913,
      "step": 8553
    },
    {
      "epoch": 0.7359545728297342,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009477460130278428,
      "loss": 0.8585,
      "step": 8554
    },
    {
      "epoch": 0.7360406091370558,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00094773397714857,
      "loss": 1.6567,
      "step": 8555
    },
    {
      "epoch": 0.7361266454443776,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009477219399597625,
      "loss": 1.0618,
      "step": 8556
    },
    {
      "epoch": 0.7362126817516992,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009477099014614552,
      "loss": 1.9347,
      "step": 8557
    },
    {
      "epoch": 0.7362987180590209,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009476978616536837,
      "loss": 1.2126,
      "step": 8558
    },
    {
      "epoch": 0.7363847543663425,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009476858205364829,
      "loss": 1.1373,
      "step": 8559
    },
    {
      "epoch": 0.7364707906736643,
      "grad_norm": 1.671875,
      "learning_rate": 0.000947673778109888,
      "loss": 1.8997,
      "step": 8560
    },
    {
      "epoch": 0.736556826980986,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009476617343739346,
      "loss": 1.1445,
      "step": 8561
    },
    {
      "epoch": 0.7366428632883076,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009476496893286577,
      "loss": 1.2079,
      "step": 8562
    },
    {
      "epoch": 0.7367288995956294,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009476376429740925,
      "loss": 1.3913,
      "step": 8563
    },
    {
      "epoch": 0.736814935902951,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009476255953102742,
      "loss": 2.0684,
      "step": 8564
    },
    {
      "epoch": 0.7369009722102727,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009476135463372382,
      "loss": 1.376,
      "step": 8565
    },
    {
      "epoch": 0.7369870085175945,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009476014960550194,
      "loss": 1.038,
      "step": 8566
    },
    {
      "epoch": 0.7370730448249161,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009475894444636535,
      "loss": 1.0573,
      "step": 8567
    },
    {
      "epoch": 0.7371590811322378,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009475773915631753,
      "loss": 1.8175,
      "step": 8568
    },
    {
      "epoch": 0.7372451174395594,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009475653373536204,
      "loss": 1.3565,
      "step": 8569
    },
    {
      "epoch": 0.7373311537468812,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009475532818350241,
      "loss": 1.2832,
      "step": 8570
    },
    {
      "epoch": 0.7374171900542029,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009475412250074213,
      "loss": 1.3517,
      "step": 8571
    },
    {
      "epoch": 0.7375032263615245,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009475291668708476,
      "loss": 1.1994,
      "step": 8572
    },
    {
      "epoch": 0.7375892626688463,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009475171074253379,
      "loss": 2.4667,
      "step": 8573
    },
    {
      "epoch": 0.7376752989761679,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009475050466709278,
      "loss": 1.5866,
      "step": 8574
    },
    {
      "epoch": 0.7377613352834896,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009474929846076525,
      "loss": 1.3126,
      "step": 8575
    },
    {
      "epoch": 0.7378473715908114,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009474809212355473,
      "loss": 1.2345,
      "step": 8576
    },
    {
      "epoch": 0.737933407898133,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009474688565546474,
      "loss": 1.2529,
      "step": 8577
    },
    {
      "epoch": 0.7380194442054547,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009474567905649882,
      "loss": 1.877,
      "step": 8578
    },
    {
      "epoch": 0.7381054805127764,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009474447232666048,
      "loss": 1.4898,
      "step": 8579
    },
    {
      "epoch": 0.7381915168200981,
      "grad_norm": 1.125,
      "learning_rate": 0.0009474326546595326,
      "loss": 1.624,
      "step": 8580
    },
    {
      "epoch": 0.7382775531274197,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000947420584743807,
      "loss": 1.8477,
      "step": 8581
    },
    {
      "epoch": 0.7383635894347415,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009474085135194632,
      "loss": 1.2292,
      "step": 8582
    },
    {
      "epoch": 0.7384496257420632,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009473964409865364,
      "loss": 1.8208,
      "step": 8583
    },
    {
      "epoch": 0.7385356620493848,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009473843671450622,
      "loss": 1.0873,
      "step": 8584
    },
    {
      "epoch": 0.7386216983567065,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009473722919950756,
      "loss": 1.0144,
      "step": 8585
    },
    {
      "epoch": 0.7387077346640282,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000947360215536612,
      "loss": 1.2734,
      "step": 8586
    },
    {
      "epoch": 0.7387937709713499,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009473481377697069,
      "loss": 1.267,
      "step": 8587
    },
    {
      "epoch": 0.7388798072786716,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009473360586943954,
      "loss": 1.1969,
      "step": 8588
    },
    {
      "epoch": 0.7389658435859933,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009473239783107131,
      "loss": 1.8314,
      "step": 8589
    },
    {
      "epoch": 0.739051879893315,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000947311896618695,
      "loss": 0.7909,
      "step": 8590
    },
    {
      "epoch": 0.7391379162006366,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009472998136183766,
      "loss": 1.397,
      "step": 8591
    },
    {
      "epoch": 0.7392239525079584,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009472877293097933,
      "loss": 1.4759,
      "step": 8592
    },
    {
      "epoch": 0.7393099888152801,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009472756436929803,
      "loss": 1.1634,
      "step": 8593
    },
    {
      "epoch": 0.7393960251226017,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009472635567679733,
      "loss": 1.3452,
      "step": 8594
    },
    {
      "epoch": 0.7394820614299235,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009472514685348072,
      "loss": 1.2515,
      "step": 8595
    },
    {
      "epoch": 0.7395680977372451,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009472393789935175,
      "loss": 1.6296,
      "step": 8596
    },
    {
      "epoch": 0.7396541340445668,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009472272881441397,
      "loss": 1.1321,
      "step": 8597
    },
    {
      "epoch": 0.7397401703518885,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009472151959867089,
      "loss": 1.2686,
      "step": 8598
    },
    {
      "epoch": 0.7398262066592102,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009472031025212608,
      "loss": 2.0051,
      "step": 8599
    },
    {
      "epoch": 0.7399122429665319,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009471910077478305,
      "loss": 1.3162,
      "step": 8600
    },
    {
      "epoch": 0.7399982792738535,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009471789116664535,
      "loss": 1.4435,
      "step": 8601
    },
    {
      "epoch": 0.7400843155811753,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009471668142771652,
      "loss": 1.1117,
      "step": 8602
    },
    {
      "epoch": 0.7401703518884969,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009471547155800009,
      "loss": 0.8221,
      "step": 8603
    },
    {
      "epoch": 0.7402563881958186,
      "grad_norm": 1.28125,
      "learning_rate": 0.000947142615574996,
      "loss": 1.2435,
      "step": 8604
    },
    {
      "epoch": 0.7403424245031404,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009471305142621859,
      "loss": 1.4014,
      "step": 8605
    },
    {
      "epoch": 0.740428460810462,
      "grad_norm": 1.4375,
      "learning_rate": 0.000947118411641606,
      "loss": 1.3808,
      "step": 8606
    },
    {
      "epoch": 0.7405144971177837,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009471063077132918,
      "loss": 1.5558,
      "step": 8607
    },
    {
      "epoch": 0.7406005334251053,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009470942024772786,
      "loss": 1.4048,
      "step": 8608
    },
    {
      "epoch": 0.7406865697324271,
      "grad_norm": 1.25,
      "learning_rate": 0.0009470820959336017,
      "loss": 0.9317,
      "step": 8609
    },
    {
      "epoch": 0.7407726060397488,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009470699880822967,
      "loss": 1.2792,
      "step": 8610
    },
    {
      "epoch": 0.7408586423470704,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000947057878923399,
      "loss": 1.2992,
      "step": 8611
    },
    {
      "epoch": 0.7409446786543922,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009470457684569438,
      "loss": 1.4416,
      "step": 8612
    },
    {
      "epoch": 0.7410307149617138,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009470336566829667,
      "loss": 1.8805,
      "step": 8613
    },
    {
      "epoch": 0.7411167512690355,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009470215436015033,
      "loss": 2.0431,
      "step": 8614
    },
    {
      "epoch": 0.7412027875763573,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009470094292125886,
      "loss": 0.984,
      "step": 8615
    },
    {
      "epoch": 0.7412888238836789,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009469973135162584,
      "loss": 1.1235,
      "step": 8616
    },
    {
      "epoch": 0.7413748601910006,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009469851965125478,
      "loss": 1.0873,
      "step": 8617
    },
    {
      "epoch": 0.7414608964983223,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009469730782014926,
      "loss": 1.385,
      "step": 8618
    },
    {
      "epoch": 0.741546932805644,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009469609585831281,
      "loss": 1.441,
      "step": 8619
    },
    {
      "epoch": 0.7416329691129657,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009469488376574897,
      "loss": 1.2867,
      "step": 8620
    },
    {
      "epoch": 0.7417190054202873,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009469367154246128,
      "loss": 0.7958,
      "step": 8621
    },
    {
      "epoch": 0.7418050417276091,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000946924591884533,
      "loss": 1.0402,
      "step": 8622
    },
    {
      "epoch": 0.7418910780349307,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009469124670372855,
      "loss": 0.9981,
      "step": 8623
    },
    {
      "epoch": 0.7419771143422524,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009469003408829062,
      "loss": 1.1368,
      "step": 8624
    },
    {
      "epoch": 0.7420631506495741,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009468882134214303,
      "loss": 1.0038,
      "step": 8625
    },
    {
      "epoch": 0.7421491869568958,
      "grad_norm": 1.125,
      "learning_rate": 0.0009468760846528932,
      "loss": 1.0335,
      "step": 8626
    },
    {
      "epoch": 0.7422352232642175,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009468639545773305,
      "loss": 1.4522,
      "step": 8627
    },
    {
      "epoch": 0.7423212595715392,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009468518231947776,
      "loss": 1.5847,
      "step": 8628
    },
    {
      "epoch": 0.7424072958788609,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00094683969050527,
      "loss": 1.3298,
      "step": 8629
    },
    {
      "epoch": 0.7424933321861825,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009468275565088434,
      "loss": 1.5602,
      "step": 8630
    },
    {
      "epoch": 0.7425793684935043,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009468154212055328,
      "loss": 1.0704,
      "step": 8631
    },
    {
      "epoch": 0.742665404800826,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009468032845953743,
      "loss": 2.078,
      "step": 8632
    },
    {
      "epoch": 0.7427514411081476,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009467911466784029,
      "loss": 1.096,
      "step": 8633
    },
    {
      "epoch": 0.7428374774154693,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009467790074546544,
      "loss": 1.1485,
      "step": 8634
    },
    {
      "epoch": 0.742923513722791,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009467668669241641,
      "loss": 1.5816,
      "step": 8635
    },
    {
      "epoch": 0.7430095500301127,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009467547250869677,
      "loss": 1.281,
      "step": 8636
    },
    {
      "epoch": 0.7430955863374344,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009467425819431006,
      "loss": 1.5334,
      "step": 8637
    },
    {
      "epoch": 0.7431816226447561,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009467304374925983,
      "loss": 1.6815,
      "step": 8638
    },
    {
      "epoch": 0.7432676589520778,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009467182917354964,
      "loss": 1.5692,
      "step": 8639
    },
    {
      "epoch": 0.7433536952593994,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009467061446718302,
      "loss": 1.0513,
      "step": 8640
    },
    {
      "epoch": 0.7434397315667212,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009466939963016357,
      "loss": 2.3528,
      "step": 8641
    },
    {
      "epoch": 0.7435257678740429,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009466818466249479,
      "loss": 1.0921,
      "step": 8642
    },
    {
      "epoch": 0.7436118041813645,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009466696956418027,
      "loss": 1.4374,
      "step": 8643
    },
    {
      "epoch": 0.7436978404886863,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009466575433522354,
      "loss": 1.1839,
      "step": 8644
    },
    {
      "epoch": 0.7437838767960079,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009466453897562817,
      "loss": 1.5049,
      "step": 8645
    },
    {
      "epoch": 0.7438699131033296,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009466332348539772,
      "loss": 1.1991,
      "step": 8646
    },
    {
      "epoch": 0.7439559494106512,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009466210786453573,
      "loss": 2.0433,
      "step": 8647
    },
    {
      "epoch": 0.744041985717973,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009466089211304576,
      "loss": 1.9838,
      "step": 8648
    },
    {
      "epoch": 0.7441280220252947,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009465967623093136,
      "loss": 2.1608,
      "step": 8649
    },
    {
      "epoch": 0.7442140583326163,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000946584602181961,
      "loss": 1.2221,
      "step": 8650
    },
    {
      "epoch": 0.7443000946399381,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009465724407484353,
      "loss": 1.6851,
      "step": 8651
    },
    {
      "epoch": 0.7443861309472597,
      "grad_norm": 1.28125,
      "learning_rate": 0.000946560278008772,
      "loss": 1.0125,
      "step": 8652
    },
    {
      "epoch": 0.7444721672545814,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009465481139630067,
      "loss": 1.3112,
      "step": 8653
    },
    {
      "epoch": 0.7445582035619032,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000946535948611175,
      "loss": 1.8579,
      "step": 8654
    },
    {
      "epoch": 0.7446442398692248,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009465237819533128,
      "loss": 1.5368,
      "step": 8655
    },
    {
      "epoch": 0.7447302761765465,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009465116139894552,
      "loss": 0.9437,
      "step": 8656
    },
    {
      "epoch": 0.7448163124838681,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009464994447196379,
      "loss": 0.6517,
      "step": 8657
    },
    {
      "epoch": 0.7449023487911899,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009464872741438966,
      "loss": 1.5696,
      "step": 8658
    },
    {
      "epoch": 0.7449883850985116,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009464751022622668,
      "loss": 1.2459,
      "step": 8659
    },
    {
      "epoch": 0.7450744214058332,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009464629290747842,
      "loss": 1.1539,
      "step": 8660
    },
    {
      "epoch": 0.745160457713155,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009464507545814845,
      "loss": 1.2776,
      "step": 8661
    },
    {
      "epoch": 0.7452464940204766,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000946438578782403,
      "loss": 0.9314,
      "step": 8662
    },
    {
      "epoch": 0.7453325303277983,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009464264016775756,
      "loss": 2.0734,
      "step": 8663
    },
    {
      "epoch": 0.7454185666351201,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009464142232670377,
      "loss": 0.9184,
      "step": 8664
    },
    {
      "epoch": 0.7455046029424417,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009464020435508251,
      "loss": 1.8294,
      "step": 8665
    },
    {
      "epoch": 0.7455906392497634,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009463898625289733,
      "loss": 1.541,
      "step": 8666
    },
    {
      "epoch": 0.7456766755570851,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000946377680201518,
      "loss": 1.4212,
      "step": 8667
    },
    {
      "epoch": 0.7457627118644068,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009463654965684947,
      "loss": 1.8978,
      "step": 8668
    },
    {
      "epoch": 0.7458487481717284,
      "grad_norm": 1.625,
      "learning_rate": 0.0009463533116299393,
      "loss": 1.3194,
      "step": 8669
    },
    {
      "epoch": 0.7459347844790502,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009463411253858872,
      "loss": 1.4639,
      "step": 8670
    },
    {
      "epoch": 0.7460208207863719,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009463289378363741,
      "loss": 2.2948,
      "step": 8671
    },
    {
      "epoch": 0.7461068570936935,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009463167489814357,
      "loss": 1.5827,
      "step": 8672
    },
    {
      "epoch": 0.7461928934010152,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009463045588211076,
      "loss": 0.6607,
      "step": 8673
    },
    {
      "epoch": 0.7462789297083369,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009462923673554253,
      "loss": 1.5878,
      "step": 8674
    },
    {
      "epoch": 0.7463649660156586,
      "grad_norm": 1.875,
      "learning_rate": 0.0009462801745844249,
      "loss": 1.6327,
      "step": 8675
    },
    {
      "epoch": 0.7464510023229803,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009462679805081417,
      "loss": 1.3159,
      "step": 8676
    },
    {
      "epoch": 0.746537038630302,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009462557851266114,
      "loss": 1.6877,
      "step": 8677
    },
    {
      "epoch": 0.7466230749376237,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009462435884398697,
      "loss": 1.8293,
      "step": 8678
    },
    {
      "epoch": 0.7467091112449453,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009462313904479524,
      "loss": 1.5927,
      "step": 8679
    },
    {
      "epoch": 0.7467951475522671,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009462191911508949,
      "loss": 0.7604,
      "step": 8680
    },
    {
      "epoch": 0.7468811838595888,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009462069905487332,
      "loss": 0.7142,
      "step": 8681
    },
    {
      "epoch": 0.7469672201669104,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009461947886415027,
      "loss": 1.5604,
      "step": 8682
    },
    {
      "epoch": 0.7470532564742322,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009461825854292394,
      "loss": 0.8095,
      "step": 8683
    },
    {
      "epoch": 0.7471392927815538,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009461703809119787,
      "loss": 1.0754,
      "step": 8684
    },
    {
      "epoch": 0.7472253290888755,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009461581750897564,
      "loss": 1.9675,
      "step": 8685
    },
    {
      "epoch": 0.7473113653961972,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009461459679626083,
      "loss": 1.6224,
      "step": 8686
    },
    {
      "epoch": 0.7473974017035189,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009461337595305699,
      "loss": 1.6125,
      "step": 8687
    },
    {
      "epoch": 0.7474834380108406,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009461215497936771,
      "loss": 1.2727,
      "step": 8688
    },
    {
      "epoch": 0.7475694743181622,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009461093387519655,
      "loss": 1.036,
      "step": 8689
    },
    {
      "epoch": 0.747655510625484,
      "grad_norm": 1.1875,
      "learning_rate": 0.000946097126405471,
      "loss": 1.2655,
      "step": 8690
    },
    {
      "epoch": 0.7477415469328056,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000946084912754229,
      "loss": 1.6098,
      "step": 8691
    },
    {
      "epoch": 0.7478275832401273,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009460726977982754,
      "loss": 1.539,
      "step": 8692
    },
    {
      "epoch": 0.7479136195474491,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009460604815376459,
      "loss": 1.0789,
      "step": 8693
    },
    {
      "epoch": 0.7479996558547707,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009460482639723763,
      "loss": 1.8844,
      "step": 8694
    },
    {
      "epoch": 0.7480856921620924,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009460360451025023,
      "loss": 1.1316,
      "step": 8695
    },
    {
      "epoch": 0.748171728469414,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009460238249280595,
      "loss": 0.7669,
      "step": 8696
    },
    {
      "epoch": 0.7482577647767358,
      "grad_norm": 0.8125,
      "learning_rate": 0.0009460116034490837,
      "loss": 0.8274,
      "step": 8697
    },
    {
      "epoch": 0.7483438010840575,
      "grad_norm": 1.1875,
      "learning_rate": 0.000945999380665611,
      "loss": 1.5012,
      "step": 8698
    },
    {
      "epoch": 0.7484298373913791,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009459871565776767,
      "loss": 1.6481,
      "step": 8699
    },
    {
      "epoch": 0.7485158736987009,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009459749311853167,
      "loss": 1.2791,
      "step": 8700
    },
    {
      "epoch": 0.7486019100060225,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009459627044885668,
      "loss": 1.5136,
      "step": 8701
    },
    {
      "epoch": 0.7486879463133442,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009459504764874626,
      "loss": 1.2993,
      "step": 8702
    },
    {
      "epoch": 0.748773982620666,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009459382471820401,
      "loss": 0.9111,
      "step": 8703
    },
    {
      "epoch": 0.7488600189279876,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009459260165723349,
      "loss": 0.8933,
      "step": 8704
    },
    {
      "epoch": 0.7489460552353093,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009459137846583829,
      "loss": 1.4254,
      "step": 8705
    },
    {
      "epoch": 0.749032091542631,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009459015514402198,
      "loss": 1.3024,
      "step": 8706
    },
    {
      "epoch": 0.7491181278499527,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009458893169178813,
      "loss": 1.0928,
      "step": 8707
    },
    {
      "epoch": 0.7492041641572744,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009458770810914034,
      "loss": 1.048,
      "step": 8708
    },
    {
      "epoch": 0.749290200464596,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009458648439608216,
      "loss": 1.8851,
      "step": 8709
    },
    {
      "epoch": 0.7493762367719178,
      "grad_norm": 1.265625,
      "learning_rate": 0.000945852605526172,
      "loss": 1.1521,
      "step": 8710
    },
    {
      "epoch": 0.7494622730792394,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009458403657874901,
      "loss": 1.2783,
      "step": 8711
    },
    {
      "epoch": 0.7495483093865611,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009458281247448119,
      "loss": 0.8879,
      "step": 8712
    },
    {
      "epoch": 0.7496343456938828,
      "grad_norm": 1.125,
      "learning_rate": 0.0009458158823981732,
      "loss": 1.3668,
      "step": 8713
    },
    {
      "epoch": 0.7497203820012045,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009458036387476097,
      "loss": 1.0067,
      "step": 8714
    },
    {
      "epoch": 0.7498064183085262,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009457913937931572,
      "loss": 1.5188,
      "step": 8715
    },
    {
      "epoch": 0.7498924546158479,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009457791475348516,
      "loss": 1.9353,
      "step": 8716
    },
    {
      "epoch": 0.7499784909231696,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009457668999727288,
      "loss": 1.4946,
      "step": 8717
    },
    {
      "epoch": 0.7500645272304912,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009457546511068244,
      "loss": 1.3498,
      "step": 8718
    },
    {
      "epoch": 0.7500645272304912,
      "eval_loss": 1.3129265308380127,
      "eval_runtime": 354.8712,
      "eval_samples_per_second": 13.101,
      "eval_steps_per_second": 1.64,
      "step": 8718
    },
    {
      "epoch": 0.750150563537813,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009457424009371745,
      "loss": 1.4318,
      "step": 8719
    },
    {
      "epoch": 0.7502365998451347,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009457301494638147,
      "loss": 1.4759,
      "step": 8720
    },
    {
      "epoch": 0.7503226361524563,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009457178966867808,
      "loss": 1.4145,
      "step": 8721
    },
    {
      "epoch": 0.750408672459778,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009457056426061088,
      "loss": 1.8224,
      "step": 8722
    },
    {
      "epoch": 0.7504947087670997,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009456933872218344,
      "loss": 1.3676,
      "step": 8723
    },
    {
      "epoch": 0.7505807450744214,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009456811305339936,
      "loss": 1.1626,
      "step": 8724
    },
    {
      "epoch": 0.7506667813817431,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009456688725426221,
      "loss": 1.6579,
      "step": 8725
    },
    {
      "epoch": 0.7507528176890648,
      "grad_norm": 1.5,
      "learning_rate": 0.000945656613247756,
      "loss": 1.1143,
      "step": 8726
    },
    {
      "epoch": 0.7508388539963865,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000945644352649431,
      "loss": 1.6225,
      "step": 8727
    },
    {
      "epoch": 0.7509248903037081,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009456320907476828,
      "loss": 1.7301,
      "step": 8728
    },
    {
      "epoch": 0.7510109266110299,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009456198275425474,
      "loss": 0.584,
      "step": 8729
    },
    {
      "epoch": 0.7510969629183516,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009456075630340606,
      "loss": 1.5705,
      "step": 8730
    },
    {
      "epoch": 0.7511829992256732,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009455952972222586,
      "loss": 1.7114,
      "step": 8731
    },
    {
      "epoch": 0.751269035532995,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009455830301071768,
      "loss": 1.3793,
      "step": 8732
    },
    {
      "epoch": 0.7513550718403166,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009455707616888514,
      "loss": 1.1311,
      "step": 8733
    },
    {
      "epoch": 0.7514411081476383,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009455584919673181,
      "loss": 1.1057,
      "step": 8734
    },
    {
      "epoch": 0.7515271444549599,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009455462209426129,
      "loss": 1.4489,
      "step": 8735
    },
    {
      "epoch": 0.7516131807622817,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009455339486147716,
      "loss": 1.3227,
      "step": 8736
    },
    {
      "epoch": 0.7516992170696034,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009455216749838303,
      "loss": 0.913,
      "step": 8737
    },
    {
      "epoch": 0.751785253376925,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009455094000498246,
      "loss": 1.4755,
      "step": 8738
    },
    {
      "epoch": 0.7518712896842468,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009454971238127908,
      "loss": 1.3835,
      "step": 8739
    },
    {
      "epoch": 0.7519573259915684,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009454848462727643,
      "loss": 1.2898,
      "step": 8740
    },
    {
      "epoch": 0.7520433622988901,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009454725674297813,
      "loss": 1.024,
      "step": 8741
    },
    {
      "epoch": 0.7521293986062119,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009454602872838777,
      "loss": 1.296,
      "step": 8742
    },
    {
      "epoch": 0.7522154349135335,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009454480058350893,
      "loss": 1.8097,
      "step": 8743
    },
    {
      "epoch": 0.7523014712208552,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009454357230834524,
      "loss": 1.2534,
      "step": 8744
    },
    {
      "epoch": 0.7523875075281768,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009454234390290024,
      "loss": 1.9563,
      "step": 8745
    },
    {
      "epoch": 0.7524735438354986,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009454111536717756,
      "loss": 1.6596,
      "step": 8746
    },
    {
      "epoch": 0.7525595801428203,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009453988670118078,
      "loss": 1.3506,
      "step": 8747
    },
    {
      "epoch": 0.7526456164501419,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009453865790491349,
      "loss": 1.6405,
      "step": 8748
    },
    {
      "epoch": 0.7527316527574637,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009453742897837929,
      "loss": 1.8292,
      "step": 8749
    },
    {
      "epoch": 0.7528176890647853,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009453619992158176,
      "loss": 1.408,
      "step": 8750
    },
    {
      "epoch": 0.752903725372107,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009453497073452453,
      "loss": 2.0206,
      "step": 8751
    },
    {
      "epoch": 0.7529897616794288,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009453374141721115,
      "loss": 1.5589,
      "step": 8752
    },
    {
      "epoch": 0.7530757979867504,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009453251196964525,
      "loss": 1.452,
      "step": 8753
    },
    {
      "epoch": 0.7531618342940721,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000945312823918304,
      "loss": 1.0211,
      "step": 8754
    },
    {
      "epoch": 0.7532478706013938,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009453005268377023,
      "loss": 1.1257,
      "step": 8755
    },
    {
      "epoch": 0.7533339069087155,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009452882284546831,
      "loss": 1.4104,
      "step": 8756
    },
    {
      "epoch": 0.7534199432160371,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009452759287692824,
      "loss": 1.8691,
      "step": 8757
    },
    {
      "epoch": 0.7535059795233588,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009452636277815362,
      "loss": 1.3981,
      "step": 8758
    },
    {
      "epoch": 0.7535920158306806,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009452513254914805,
      "loss": 1.5505,
      "step": 8759
    },
    {
      "epoch": 0.7536780521380022,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009452390218991513,
      "loss": 1.2597,
      "step": 8760
    },
    {
      "epoch": 0.7537640884453239,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009452267170045845,
      "loss": 1.2189,
      "step": 8761
    },
    {
      "epoch": 0.7538501247526456,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009452144108078161,
      "loss": 1.5436,
      "step": 8762
    },
    {
      "epoch": 0.7539361610599673,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009452021033088822,
      "loss": 1.4173,
      "step": 8763
    },
    {
      "epoch": 0.754022197367289,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009451897945078187,
      "loss": 1.9707,
      "step": 8764
    },
    {
      "epoch": 0.7541082336746107,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009451774844046617,
      "loss": 1.9709,
      "step": 8765
    },
    {
      "epoch": 0.7541942699819324,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009451651729994472,
      "loss": 1.0936,
      "step": 8766
    },
    {
      "epoch": 0.754280306289254,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000945152860292211,
      "loss": 1.3647,
      "step": 8767
    },
    {
      "epoch": 0.7543663425965758,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009451405462829894,
      "loss": 1.5717,
      "step": 8768
    },
    {
      "epoch": 0.7544523789038975,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009451282309718181,
      "loss": 0.8934,
      "step": 8769
    },
    {
      "epoch": 0.7545384152112191,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009451159143587334,
      "loss": 1.2441,
      "step": 8770
    },
    {
      "epoch": 0.7546244515185409,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009451035964437713,
      "loss": 1.601,
      "step": 8771
    },
    {
      "epoch": 0.7547104878258625,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009450912772269677,
      "loss": 1.1766,
      "step": 8772
    },
    {
      "epoch": 0.7547965241331842,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009450789567083586,
      "loss": 1.5174,
      "step": 8773
    },
    {
      "epoch": 0.754882560440506,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009450666348879801,
      "loss": 1.7107,
      "step": 8774
    },
    {
      "epoch": 0.7549685967478276,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009450543117658684,
      "loss": 1.0707,
      "step": 8775
    },
    {
      "epoch": 0.7550546330551493,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009450419873420592,
      "loss": 1.5252,
      "step": 8776
    },
    {
      "epoch": 0.7551406693624709,
      "grad_norm": 1.296875,
      "learning_rate": 0.000945029661616589,
      "loss": 1.3781,
      "step": 8777
    },
    {
      "epoch": 0.7552267056697927,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009450173345894934,
      "loss": 1.8187,
      "step": 8778
    },
    {
      "epoch": 0.7553127419771143,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009450050062608087,
      "loss": 1.2156,
      "step": 8779
    },
    {
      "epoch": 0.755398778284436,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000944992676630571,
      "loss": 1.1866,
      "step": 8780
    },
    {
      "epoch": 0.7554848145917578,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009449803456988161,
      "loss": 2.2598,
      "step": 8781
    },
    {
      "epoch": 0.7555708508990794,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009449680134655802,
      "loss": 1.3641,
      "step": 8782
    },
    {
      "epoch": 0.7556568872064011,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009449556799308996,
      "loss": 1.8704,
      "step": 8783
    },
    {
      "epoch": 0.7557429235137227,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009449433450948102,
      "loss": 1.1497,
      "step": 8784
    },
    {
      "epoch": 0.7558289598210445,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009449310089573479,
      "loss": 1.3576,
      "step": 8785
    },
    {
      "epoch": 0.7559149961283662,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000944918671518549,
      "loss": 1.1699,
      "step": 8786
    },
    {
      "epoch": 0.7560010324356878,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009449063327784494,
      "loss": 0.836,
      "step": 8787
    },
    {
      "epoch": 0.7560870687430096,
      "grad_norm": 1.5,
      "learning_rate": 0.0009448939927370854,
      "loss": 1.1758,
      "step": 8788
    },
    {
      "epoch": 0.7561731050503312,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009448816513944929,
      "loss": 0.9845,
      "step": 8789
    },
    {
      "epoch": 0.7562591413576529,
      "grad_norm": 3.1875,
      "learning_rate": 0.0009448693087507084,
      "loss": 1.5168,
      "step": 8790
    },
    {
      "epoch": 0.7563451776649747,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009448569648057673,
      "loss": 1.3113,
      "step": 8791
    },
    {
      "epoch": 0.7564312139722963,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009448446195597063,
      "loss": 1.7638,
      "step": 8792
    },
    {
      "epoch": 0.756517250279618,
      "grad_norm": 27.0,
      "learning_rate": 0.0009448322730125614,
      "loss": 1.8362,
      "step": 8793
    },
    {
      "epoch": 0.7566032865869396,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009448199251643684,
      "loss": 2.1302,
      "step": 8794
    },
    {
      "epoch": 0.7566893228942614,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009448075760151637,
      "loss": 0.8278,
      "step": 8795
    },
    {
      "epoch": 0.7567753592015831,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009447952255649835,
      "loss": 1.2635,
      "step": 8796
    },
    {
      "epoch": 0.7568613955089047,
      "grad_norm": 7.3125,
      "learning_rate": 0.0009447828738138635,
      "loss": 1.3979,
      "step": 8797
    },
    {
      "epoch": 0.7569474318162265,
      "grad_norm": 4.4375,
      "learning_rate": 0.0009447705207618403,
      "loss": 1.4721,
      "step": 8798
    },
    {
      "epoch": 0.7570334681235481,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009447581664089497,
      "loss": 1.7344,
      "step": 8799
    },
    {
      "epoch": 0.7571195044308698,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009447458107552281,
      "loss": 1.1268,
      "step": 8800
    },
    {
      "epoch": 0.7572055407381915,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009447334538007115,
      "loss": 0.8736,
      "step": 8801
    },
    {
      "epoch": 0.7572915770455132,
      "grad_norm": 1.40625,
      "learning_rate": 0.000944721095545436,
      "loss": 1.7544,
      "step": 8802
    },
    {
      "epoch": 0.7573776133528349,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009447087359894377,
      "loss": 1.3457,
      "step": 8803
    },
    {
      "epoch": 0.7574636496601566,
      "grad_norm": 2.171875,
      "learning_rate": 0.000944696375132753,
      "loss": 1.5365,
      "step": 8804
    },
    {
      "epoch": 0.7575496859674783,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009446840129754177,
      "loss": 0.8452,
      "step": 8805
    },
    {
      "epoch": 0.7576357222747999,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009446716495174684,
      "loss": 0.6552,
      "step": 8806
    },
    {
      "epoch": 0.7577217585821217,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009446592847589409,
      "loss": 1.08,
      "step": 8807
    },
    {
      "epoch": 0.7578077948894434,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009446469186998714,
      "loss": 1.8695,
      "step": 8808
    },
    {
      "epoch": 0.757893831196765,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009446345513402961,
      "loss": 1.0502,
      "step": 8809
    },
    {
      "epoch": 0.7579798675040867,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009446221826802512,
      "loss": 1.0693,
      "step": 8810
    },
    {
      "epoch": 0.7580659038114084,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009446098127197731,
      "loss": 2.6776,
      "step": 8811
    },
    {
      "epoch": 0.7581519401187301,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009445974414588977,
      "loss": 1.6004,
      "step": 8812
    },
    {
      "epoch": 0.7582379764260518,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009445850688976612,
      "loss": 1.569,
      "step": 8813
    },
    {
      "epoch": 0.7583240127333735,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009445726950360998,
      "loss": 1.9643,
      "step": 8814
    },
    {
      "epoch": 0.7584100490406952,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009445603198742498,
      "loss": 1.4928,
      "step": 8815
    },
    {
      "epoch": 0.7584960853480168,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009445479434121473,
      "loss": 1.4003,
      "step": 8816
    },
    {
      "epoch": 0.7585821216553386,
      "grad_norm": 1.75,
      "learning_rate": 0.0009445355656498284,
      "loss": 2.1699,
      "step": 8817
    },
    {
      "epoch": 0.7586681579626603,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009445231865873297,
      "loss": 1.7668,
      "step": 8818
    },
    {
      "epoch": 0.7587541942699819,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000944510806224687,
      "loss": 0.941,
      "step": 8819
    },
    {
      "epoch": 0.7588402305773037,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009444984245619366,
      "loss": 1.2263,
      "step": 8820
    },
    {
      "epoch": 0.7589262668846253,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009444860415991148,
      "loss": 0.7116,
      "step": 8821
    },
    {
      "epoch": 0.759012303191947,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009444736573362578,
      "loss": 1.6274,
      "step": 8822
    },
    {
      "epoch": 0.7590983394992686,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009444612717734018,
      "loss": 1.1804,
      "step": 8823
    },
    {
      "epoch": 0.7591843758065904,
      "grad_norm": 1.0,
      "learning_rate": 0.000944448884910583,
      "loss": 0.9086,
      "step": 8824
    },
    {
      "epoch": 0.7592704121139121,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009444364967478376,
      "loss": 1.7603,
      "step": 8825
    },
    {
      "epoch": 0.7593564484212337,
      "grad_norm": 1.125,
      "learning_rate": 0.000944424107285202,
      "loss": 1.3374,
      "step": 8826
    },
    {
      "epoch": 0.7594424847285555,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009444117165227122,
      "loss": 1.5531,
      "step": 8827
    },
    {
      "epoch": 0.7595285210358771,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009443993244604046,
      "loss": 0.8258,
      "step": 8828
    },
    {
      "epoch": 0.7596145573431988,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009443869310983155,
      "loss": 1.3169,
      "step": 8829
    },
    {
      "epoch": 0.7597005936505206,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009443745364364808,
      "loss": 2.0771,
      "step": 8830
    },
    {
      "epoch": 0.7597866299578422,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009443621404749373,
      "loss": 1.058,
      "step": 8831
    },
    {
      "epoch": 0.7598726662651639,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000944349743213721,
      "loss": 1.1656,
      "step": 8832
    },
    {
      "epoch": 0.7599587025724855,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009443373446528678,
      "loss": 1.0487,
      "step": 8833
    },
    {
      "epoch": 0.7600447388798073,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009443249447924145,
      "loss": 1.0077,
      "step": 8834
    },
    {
      "epoch": 0.760130775187129,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009443125436323971,
      "loss": 1.793,
      "step": 8835
    },
    {
      "epoch": 0.7602168114944506,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009443001411728519,
      "loss": 1.0292,
      "step": 8836
    },
    {
      "epoch": 0.7603028478017724,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009442877374138153,
      "loss": 1.4306,
      "step": 8837
    },
    {
      "epoch": 0.760388884109094,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009442753323553233,
      "loss": 1.3372,
      "step": 8838
    },
    {
      "epoch": 0.7604749204164157,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009442629259974125,
      "loss": 0.9679,
      "step": 8839
    },
    {
      "epoch": 0.7605609567237375,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009442505183401191,
      "loss": 0.8094,
      "step": 8840
    },
    {
      "epoch": 0.7606469930310591,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009442381093834792,
      "loss": 0.9688,
      "step": 8841
    },
    {
      "epoch": 0.7607330293383808,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009442256991275293,
      "loss": 1.2127,
      "step": 8842
    },
    {
      "epoch": 0.7608190656457025,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009442132875723055,
      "loss": 2.2635,
      "step": 8843
    },
    {
      "epoch": 0.7609051019530242,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009442008747178445,
      "loss": 1.3526,
      "step": 8844
    },
    {
      "epoch": 0.7609911382603458,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009441884605641822,
      "loss": 1.2057,
      "step": 8845
    },
    {
      "epoch": 0.7610771745676675,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009441760451113549,
      "loss": 2.0963,
      "step": 8846
    },
    {
      "epoch": 0.7611632108749893,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009441636283593992,
      "loss": 1.3696,
      "step": 8847
    },
    {
      "epoch": 0.7612492471823109,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009441512103083512,
      "loss": 1.3009,
      "step": 8848
    },
    {
      "epoch": 0.7613352834896326,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009441387909582473,
      "loss": 1.2092,
      "step": 8849
    },
    {
      "epoch": 0.7614213197969543,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009441263703091239,
      "loss": 1.1723,
      "step": 8850
    },
    {
      "epoch": 0.761507356104276,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009441139483610173,
      "loss": 1.4685,
      "step": 8851
    },
    {
      "epoch": 0.7615933924115977,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009441015251139635,
      "loss": 1.3828,
      "step": 8852
    },
    {
      "epoch": 0.7616794287189194,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009440891005679993,
      "loss": 1.6158,
      "step": 8853
    },
    {
      "epoch": 0.7617654650262411,
      "grad_norm": 1.75,
      "learning_rate": 0.0009440766747231609,
      "loss": 1.6026,
      "step": 8854
    },
    {
      "epoch": 0.7618515013335627,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009440642475794845,
      "loss": 0.8544,
      "step": 8855
    },
    {
      "epoch": 0.7619375376408845,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009440518191370066,
      "loss": 1.0249,
      "step": 8856
    },
    {
      "epoch": 0.7620235739482062,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009440393893957635,
      "loss": 1.7343,
      "step": 8857
    },
    {
      "epoch": 0.7621096102555278,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009440269583557915,
      "loss": 1.5236,
      "step": 8858
    },
    {
      "epoch": 0.7621956465628495,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009440145260171268,
      "loss": 1.7479,
      "step": 8859
    },
    {
      "epoch": 0.7622816828701712,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009440020923798063,
      "loss": 1.5881,
      "step": 8860
    },
    {
      "epoch": 0.7623677191774929,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000943989657443866,
      "loss": 1.0559,
      "step": 8861
    },
    {
      "epoch": 0.7624537554848146,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009439772212093421,
      "loss": 1.3386,
      "step": 8862
    },
    {
      "epoch": 0.7625397917921363,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009439647836762714,
      "loss": 1.1017,
      "step": 8863
    },
    {
      "epoch": 0.762625828099458,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009439523448446899,
      "loss": 1.3127,
      "step": 8864
    },
    {
      "epoch": 0.7627118644067796,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009439399047146342,
      "loss": 1.2385,
      "step": 8865
    },
    {
      "epoch": 0.7627979007141014,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009439274632861406,
      "loss": 1.0048,
      "step": 8866
    },
    {
      "epoch": 0.762883937021423,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009439150205592455,
      "loss": 2.0215,
      "step": 8867
    },
    {
      "epoch": 0.7629699733287447,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009439025765339853,
      "loss": 1.1949,
      "step": 8868
    },
    {
      "epoch": 0.7630560096360665,
      "grad_norm": 1.25,
      "learning_rate": 0.0009438901312103965,
      "loss": 1.052,
      "step": 8869
    },
    {
      "epoch": 0.7631420459433881,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009438776845885153,
      "loss": 1.1079,
      "step": 8870
    },
    {
      "epoch": 0.7632280822507098,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009438652366683781,
      "loss": 1.1197,
      "step": 8871
    },
    {
      "epoch": 0.7633141185580314,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009438527874500215,
      "loss": 1.8479,
      "step": 8872
    },
    {
      "epoch": 0.7634001548653532,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009438403369334819,
      "loss": 1.4998,
      "step": 8873
    },
    {
      "epoch": 0.7634861911726749,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009438278851187956,
      "loss": 1.6009,
      "step": 8874
    },
    {
      "epoch": 0.7635722274799965,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009438154320059991,
      "loss": 1.4728,
      "step": 8875
    },
    {
      "epoch": 0.7636582637873183,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009438029775951286,
      "loss": 0.9674,
      "step": 8876
    },
    {
      "epoch": 0.7637443000946399,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009437905218862209,
      "loss": 1.1278,
      "step": 8877
    },
    {
      "epoch": 0.7638303364019616,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009437780648793121,
      "loss": 1.0431,
      "step": 8878
    },
    {
      "epoch": 0.7639163727092834,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009437656065744388,
      "loss": 1.2991,
      "step": 8879
    },
    {
      "epoch": 0.764002409016605,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0009437531469716375,
      "loss": 1.1149,
      "step": 8880
    },
    {
      "epoch": 0.7640884453239267,
      "grad_norm": 22.75,
      "learning_rate": 0.0009437406860709445,
      "loss": 1.2316,
      "step": 8881
    },
    {
      "epoch": 0.7641744816312483,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009437282238723963,
      "loss": 1.1243,
      "step": 8882
    },
    {
      "epoch": 0.7642605179385701,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009437157603760293,
      "loss": 0.9516,
      "step": 8883
    },
    {
      "epoch": 0.7643465542458918,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009437032955818801,
      "loss": 1.6764,
      "step": 8884
    },
    {
      "epoch": 0.7644325905532134,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009436908294899849,
      "loss": 1.438,
      "step": 8885
    },
    {
      "epoch": 0.7645186268605352,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009436783621003804,
      "loss": 1.0868,
      "step": 8886
    },
    {
      "epoch": 0.7646046631678568,
      "grad_norm": 1.25,
      "learning_rate": 0.000943665893413103,
      "loss": 1.2962,
      "step": 8887
    },
    {
      "epoch": 0.7646906994751785,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000943653423428189,
      "loss": 1.0887,
      "step": 8888
    },
    {
      "epoch": 0.7647767357825002,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009436409521456753,
      "loss": 1.0147,
      "step": 8889
    },
    {
      "epoch": 0.7648627720898219,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009436284795655978,
      "loss": 1.3377,
      "step": 8890
    },
    {
      "epoch": 0.7649488083971436,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009436160056879935,
      "loss": 2.1091,
      "step": 8891
    },
    {
      "epoch": 0.7650348447044653,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009436035305128986,
      "loss": 1.6093,
      "step": 8892
    },
    {
      "epoch": 0.765120881011787,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009435910540403496,
      "loss": 1.6328,
      "step": 8893
    },
    {
      "epoch": 0.7652069173191086,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009435785762703831,
      "loss": 0.8684,
      "step": 8894
    },
    {
      "epoch": 0.7652929536264304,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009435660972030355,
      "loss": 1.704,
      "step": 8895
    },
    {
      "epoch": 0.7653789899337521,
      "grad_norm": 1.0,
      "learning_rate": 0.0009435536168383434,
      "loss": 0.8676,
      "step": 8896
    },
    {
      "epoch": 0.7654650262410737,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009435411351763432,
      "loss": 0.916,
      "step": 8897
    },
    {
      "epoch": 0.7655510625483954,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009435286522170714,
      "loss": 1.2776,
      "step": 8898
    },
    {
      "epoch": 0.7656370988557171,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009435161679605647,
      "loss": 1.4895,
      "step": 8899
    },
    {
      "epoch": 0.7657231351630388,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009435036824068593,
      "loss": 1.7396,
      "step": 8900
    },
    {
      "epoch": 0.7658091714703605,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009434911955559919,
      "loss": 0.8918,
      "step": 8901
    },
    {
      "epoch": 0.7658952077776822,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000943478707407999,
      "loss": 1.4374,
      "step": 8902
    },
    {
      "epoch": 0.7659812440850039,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009434662179629173,
      "loss": 1.0365,
      "step": 8903
    },
    {
      "epoch": 0.7660672803923255,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009434537272207831,
      "loss": 1.394,
      "step": 8904
    },
    {
      "epoch": 0.7661533166996473,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009434412351816328,
      "loss": 0.9957,
      "step": 8905
    },
    {
      "epoch": 0.766239353006969,
      "grad_norm": 4.96875,
      "learning_rate": 0.0009434287418455032,
      "loss": 1.3204,
      "step": 8906
    },
    {
      "epoch": 0.7663253893142906,
      "grad_norm": 1.5625,
      "learning_rate": 0.000943416247212431,
      "loss": 1.4381,
      "step": 8907
    },
    {
      "epoch": 0.7664114256216124,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009434037512824523,
      "loss": 1.0019,
      "step": 8908
    },
    {
      "epoch": 0.766497461928934,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009433912540556039,
      "loss": 1.1851,
      "step": 8909
    },
    {
      "epoch": 0.7665834982362557,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009433787555319223,
      "loss": 1.0668,
      "step": 8910
    },
    {
      "epoch": 0.7666695345435773,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009433662557114442,
      "loss": 1.7011,
      "step": 8911
    },
    {
      "epoch": 0.7667555708508991,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009433537545942059,
      "loss": 0.9625,
      "step": 8912
    },
    {
      "epoch": 0.7668416071582208,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009433412521802441,
      "loss": 1.2064,
      "step": 8913
    },
    {
      "epoch": 0.7669276434655424,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009433287484695953,
      "loss": 1.148,
      "step": 8914
    },
    {
      "epoch": 0.7670136797728642,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009433162434622964,
      "loss": 1.5929,
      "step": 8915
    },
    {
      "epoch": 0.7670997160801858,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009433037371583834,
      "loss": 1.0501,
      "step": 8916
    },
    {
      "epoch": 0.7671857523875075,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009432912295578935,
      "loss": 1.9985,
      "step": 8917
    },
    {
      "epoch": 0.7672717886948293,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009432787206608627,
      "loss": 1.9222,
      "step": 8918
    },
    {
      "epoch": 0.7673578250021509,
      "grad_norm": 1.375,
      "learning_rate": 0.0009432662104673279,
      "loss": 1.9555,
      "step": 8919
    },
    {
      "epoch": 0.7674438613094726,
      "grad_norm": 3.34375,
      "learning_rate": 0.0009432536989773257,
      "loss": 1.6308,
      "step": 8920
    },
    {
      "epoch": 0.7675298976167942,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009432411861908927,
      "loss": 0.9231,
      "step": 8921
    },
    {
      "epoch": 0.767615933924116,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009432286721080654,
      "loss": 1.6065,
      "step": 8922
    },
    {
      "epoch": 0.7677019702314377,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009432161567288804,
      "loss": 1.5695,
      "step": 8923
    },
    {
      "epoch": 0.7677880065387593,
      "grad_norm": 1.25,
      "learning_rate": 0.0009432036400533743,
      "loss": 1.0152,
      "step": 8924
    },
    {
      "epoch": 0.7678740428460811,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009431911220815839,
      "loss": 1.6022,
      "step": 8925
    },
    {
      "epoch": 0.7679600791534027,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009431786028135455,
      "loss": 1.7157,
      "step": 8926
    },
    {
      "epoch": 0.7680461154607244,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009431660822492959,
      "loss": 1.0702,
      "step": 8927
    },
    {
      "epoch": 0.7681321517680462,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009431535603888718,
      "loss": 1.0944,
      "step": 8928
    },
    {
      "epoch": 0.7682181880753678,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009431410372323097,
      "loss": 1.6048,
      "step": 8929
    },
    {
      "epoch": 0.7683042243826895,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009431285127796462,
      "loss": 1.4639,
      "step": 8930
    },
    {
      "epoch": 0.7683902606900112,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009431159870309181,
      "loss": 1.4991,
      "step": 8931
    },
    {
      "epoch": 0.7684762969973329,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009431034599861616,
      "loss": 1.6335,
      "step": 8932
    },
    {
      "epoch": 0.7685623333046545,
      "grad_norm": 1.5,
      "learning_rate": 0.000943090931645414,
      "loss": 1.3194,
      "step": 8933
    },
    {
      "epoch": 0.7686483696119762,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009430784020087116,
      "loss": 1.3027,
      "step": 8934
    },
    {
      "epoch": 0.768734405919298,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009430658710760909,
      "loss": 1.7359,
      "step": 8935
    },
    {
      "epoch": 0.7688204422266196,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009430533388475886,
      "loss": 1.5317,
      "step": 8936
    },
    {
      "epoch": 0.7689064785339413,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009430408053232415,
      "loss": 1.4992,
      "step": 8937
    },
    {
      "epoch": 0.768992514841263,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009430282705030863,
      "loss": 1.2521,
      "step": 8938
    },
    {
      "epoch": 0.7690785511485847,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009430157343871597,
      "loss": 1.3293,
      "step": 8939
    },
    {
      "epoch": 0.7691645874559064,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009430031969754981,
      "loss": 0.8222,
      "step": 8940
    },
    {
      "epoch": 0.7692506237632281,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009429906582681383,
      "loss": 1.1532,
      "step": 8941
    },
    {
      "epoch": 0.7693366600705498,
      "grad_norm": 1.8203125,
      "learning_rate": 0.000942978118265117,
      "loss": 1.8005,
      "step": 8942
    },
    {
      "epoch": 0.7694226963778714,
      "grad_norm": 1.09375,
      "learning_rate": 0.000942965576966471,
      "loss": 1.0246,
      "step": 8943
    },
    {
      "epoch": 0.7695087326851932,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009429530343722366,
      "loss": 1.1852,
      "step": 8944
    },
    {
      "epoch": 0.7695947689925149,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009429404904824507,
      "loss": 1.6592,
      "step": 8945
    },
    {
      "epoch": 0.7696808052998365,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009429279452971502,
      "loss": 2.0413,
      "step": 8946
    },
    {
      "epoch": 0.7697668416071582,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009429153988163715,
      "loss": 1.4027,
      "step": 8947
    },
    {
      "epoch": 0.7698528779144799,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009429028510401515,
      "loss": 1.725,
      "step": 8948
    },
    {
      "epoch": 0.7699389142218016,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009428903019685266,
      "loss": 1.097,
      "step": 8949
    },
    {
      "epoch": 0.7700249505291233,
      "grad_norm": 1.125,
      "learning_rate": 0.000942877751601534,
      "loss": 1.8717,
      "step": 8950
    },
    {
      "epoch": 0.770110986836445,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009428651999392099,
      "loss": 1.1302,
      "step": 8951
    },
    {
      "epoch": 0.7701970231437667,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009428526469815912,
      "loss": 2.3215,
      "step": 8952
    },
    {
      "epoch": 0.7702830594510883,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009428400927287148,
      "loss": 2.0458,
      "step": 8953
    },
    {
      "epoch": 0.7703690957584101,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009428275371806172,
      "loss": 1.5231,
      "step": 8954
    },
    {
      "epoch": 0.7704551320657317,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009428149803373351,
      "loss": 1.4795,
      "step": 8955
    },
    {
      "epoch": 0.7705411683730534,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009428024221989054,
      "loss": 1.6478,
      "step": 8956
    },
    {
      "epoch": 0.7706272046803752,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009427898627653648,
      "loss": 1.0657,
      "step": 8957
    },
    {
      "epoch": 0.7707132409876968,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009427773020367497,
      "loss": 1.9974,
      "step": 8958
    },
    {
      "epoch": 0.7707992772950185,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009427647400130974,
      "loss": 1.5432,
      "step": 8959
    },
    {
      "epoch": 0.7708853136023401,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009427521766944442,
      "loss": 0.9964,
      "step": 8960
    },
    {
      "epoch": 0.7709713499096619,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009427396120808269,
      "loss": 1.0941,
      "step": 8961
    },
    {
      "epoch": 0.7710573862169836,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009427270461722826,
      "loss": 0.7946,
      "step": 8962
    },
    {
      "epoch": 0.7711434225243052,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009427144789688475,
      "loss": 1.3263,
      "step": 8963
    },
    {
      "epoch": 0.771229458831627,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009427019104705588,
      "loss": 1.0141,
      "step": 8964
    },
    {
      "epoch": 0.7713154951389486,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009426893406774532,
      "loss": 0.908,
      "step": 8965
    },
    {
      "epoch": 0.7714015314462703,
      "grad_norm": 6.71875,
      "learning_rate": 0.0009426767695895672,
      "loss": 1.3219,
      "step": 8966
    },
    {
      "epoch": 0.7714875677535921,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009426641972069378,
      "loss": 1.0183,
      "step": 8967
    },
    {
      "epoch": 0.7715736040609137,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009426516235296018,
      "loss": 1.2487,
      "step": 8968
    },
    {
      "epoch": 0.7716596403682354,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009426390485575959,
      "loss": 1.3609,
      "step": 8969
    },
    {
      "epoch": 0.771745676675557,
      "grad_norm": 7.59375,
      "learning_rate": 0.0009426264722909568,
      "loss": 0.7535,
      "step": 8970
    },
    {
      "epoch": 0.7718317129828788,
      "grad_norm": 6.65625,
      "learning_rate": 0.0009426138947297213,
      "loss": 1.6846,
      "step": 8971
    },
    {
      "epoch": 0.7719177492902005,
      "grad_norm": 16.125,
      "learning_rate": 0.0009426013158739264,
      "loss": 1.8215,
      "step": 8972
    },
    {
      "epoch": 0.7720037855975221,
      "grad_norm": 23.125,
      "learning_rate": 0.0009425887357236086,
      "loss": 0.9697,
      "step": 8973
    },
    {
      "epoch": 0.7720898219048439,
      "grad_norm": 41.75,
      "learning_rate": 0.0009425761542788049,
      "loss": 1.661,
      "step": 8974
    },
    {
      "epoch": 0.7721758582121655,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000942563571539552,
      "loss": 1.4015,
      "step": 8975
    },
    {
      "epoch": 0.7722618945194872,
      "grad_norm": 6.1875,
      "learning_rate": 0.0009425509875058866,
      "loss": 1.3884,
      "step": 8976
    },
    {
      "epoch": 0.7723479308268089,
      "grad_norm": 8.4375,
      "learning_rate": 0.0009425384021778459,
      "loss": 1.0501,
      "step": 8977
    },
    {
      "epoch": 0.7724339671341306,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009425258155554662,
      "loss": 1.7256,
      "step": 8978
    },
    {
      "epoch": 0.7725200034414523,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009425132276387846,
      "loss": 0.8009,
      "step": 8979
    },
    {
      "epoch": 0.772606039748774,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000942500638427838,
      "loss": 0.9766,
      "step": 8980
    },
    {
      "epoch": 0.7726920760560957,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009424880479226629,
      "loss": 1.6418,
      "step": 8981
    },
    {
      "epoch": 0.7727781123634173,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009424754561232966,
      "loss": 1.7081,
      "step": 8982
    },
    {
      "epoch": 0.772864148670739,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009424628630297755,
      "loss": 1.221,
      "step": 8983
    },
    {
      "epoch": 0.7729501849780608,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009424502686421366,
      "loss": 1.2321,
      "step": 8984
    },
    {
      "epoch": 0.7730362212853824,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009424376729604166,
      "loss": 1.9037,
      "step": 8985
    },
    {
      "epoch": 0.7731222575927041,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009424250759846526,
      "loss": 1.1633,
      "step": 8986
    },
    {
      "epoch": 0.7732082939000258,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009424124777148812,
      "loss": 1.5823,
      "step": 8987
    },
    {
      "epoch": 0.7732943302073475,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009423998781511394,
      "loss": 1.37,
      "step": 8988
    },
    {
      "epoch": 0.7733803665146692,
      "grad_norm": 1.796875,
      "learning_rate": 0.000942387277293464,
      "loss": 2.3796,
      "step": 8989
    },
    {
      "epoch": 0.7734664028219909,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009423746751418918,
      "loss": 1.3254,
      "step": 8990
    },
    {
      "epoch": 0.7735524391293126,
      "grad_norm": 2.515625,
      "learning_rate": 0.0009423620716964597,
      "loss": 2.1165,
      "step": 8991
    },
    {
      "epoch": 0.7736384754366342,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009423494669572046,
      "loss": 1.0414,
      "step": 8992
    },
    {
      "epoch": 0.773724511743956,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009423368609241633,
      "loss": 1.0099,
      "step": 8993
    },
    {
      "epoch": 0.7738105480512777,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009423242535973728,
      "loss": 0.8192,
      "step": 8994
    },
    {
      "epoch": 0.7738965843585993,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009423116449768699,
      "loss": 1.3799,
      "step": 8995
    },
    {
      "epoch": 0.773982620665921,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009422990350626914,
      "loss": 0.9108,
      "step": 8996
    },
    {
      "epoch": 0.7740686569732427,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009422864238548743,
      "loss": 1.1129,
      "step": 8997
    },
    {
      "epoch": 0.7741546932805644,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009422738113534554,
      "loss": 1.1197,
      "step": 8998
    },
    {
      "epoch": 0.774240729587886,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009422611975584717,
      "loss": 1.0817,
      "step": 8999
    },
    {
      "epoch": 0.7743267658952078,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009422485824699599,
      "loss": 1.7422,
      "step": 9000
    },
    {
      "epoch": 0.7744128022025295,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009422359660879572,
      "loss": 1.4343,
      "step": 9001
    },
    {
      "epoch": 0.7744988385098511,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009422233484125,
      "loss": 0.6745,
      "step": 9002
    },
    {
      "epoch": 0.7745848748171729,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009422107294436257,
      "loss": 0.9113,
      "step": 9003
    },
    {
      "epoch": 0.7746709111244945,
      "grad_norm": 1.234375,
      "learning_rate": 0.000942198109181371,
      "loss": 1.2941,
      "step": 9004
    },
    {
      "epoch": 0.7747569474318162,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009421854876257727,
      "loss": 1.43,
      "step": 9005
    },
    {
      "epoch": 0.774842983739138,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009421728647768681,
      "loss": 0.8394,
      "step": 9006
    },
    {
      "epoch": 0.7749290200464596,
      "grad_norm": 1.9375,
      "learning_rate": 0.0009421602406346936,
      "loss": 1.4021,
      "step": 9007
    },
    {
      "epoch": 0.7750150563537813,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009421476151992866,
      "loss": 2.1334,
      "step": 9008
    },
    {
      "epoch": 0.7751010926611029,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009421349884706838,
      "loss": 1.5472,
      "step": 9009
    },
    {
      "epoch": 0.7751871289684247,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009421223604489221,
      "loss": 2.1644,
      "step": 9010
    },
    {
      "epoch": 0.7752731652757464,
      "grad_norm": 1.75,
      "learning_rate": 0.0009421097311340384,
      "loss": 1.2885,
      "step": 9011
    },
    {
      "epoch": 0.775359201583068,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009420971005260699,
      "loss": 1.6388,
      "step": 9012
    },
    {
      "epoch": 0.7754452378903898,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009420844686250532,
      "loss": 1.5207,
      "step": 9013
    },
    {
      "epoch": 0.7755312741977114,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009420718354310255,
      "loss": 1.1312,
      "step": 9014
    },
    {
      "epoch": 0.7756173105050331,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009420592009440237,
      "loss": 1.6887,
      "step": 9015
    },
    {
      "epoch": 0.7757033468123549,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009420465651640846,
      "loss": 0.8407,
      "step": 9016
    },
    {
      "epoch": 0.7757893831196765,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009420339280912453,
      "loss": 2.0792,
      "step": 9017
    },
    {
      "epoch": 0.7758754194269982,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009420212897255427,
      "loss": 1.4755,
      "step": 9018
    },
    {
      "epoch": 0.7759614557343198,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009420086500670138,
      "loss": 1.8255,
      "step": 9019
    },
    {
      "epoch": 0.7760474920416416,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009419960091156956,
      "loss": 0.9561,
      "step": 9020
    },
    {
      "epoch": 0.7761335283489632,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009419833668716251,
      "loss": 1.9712,
      "step": 9021
    },
    {
      "epoch": 0.7762195646562849,
      "grad_norm": 1.375,
      "learning_rate": 0.0009419707233348392,
      "loss": 1.3686,
      "step": 9022
    },
    {
      "epoch": 0.7763056009636067,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009419580785053748,
      "loss": 0.8363,
      "step": 9023
    },
    {
      "epoch": 0.7763916372709283,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000941945432383269,
      "loss": 0.977,
      "step": 9024
    },
    {
      "epoch": 0.77647767357825,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009419327849685588,
      "loss": 1.1911,
      "step": 9025
    },
    {
      "epoch": 0.7765637098855717,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009419201362612813,
      "loss": 1.0768,
      "step": 9026
    },
    {
      "epoch": 0.7766497461928934,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009419074862614731,
      "loss": 1.2025,
      "step": 9027
    },
    {
      "epoch": 0.7767357825002151,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009418948349691716,
      "loss": 1.4904,
      "step": 9028
    },
    {
      "epoch": 0.7768218188075368,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009418821823844136,
      "loss": 1.7823,
      "step": 9029
    },
    {
      "epoch": 0.7769078551148585,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000941869528507236,
      "loss": 0.8426,
      "step": 9030
    },
    {
      "epoch": 0.7769938914221801,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009418568733376762,
      "loss": 1.809,
      "step": 9031
    },
    {
      "epoch": 0.7770799277295019,
      "grad_norm": 1.40625,
      "learning_rate": 0.000941844216875771,
      "loss": 1.6777,
      "step": 9032
    },
    {
      "epoch": 0.7771659640368236,
      "grad_norm": 0.875,
      "learning_rate": 0.0009418315591215573,
      "loss": 0.8606,
      "step": 9033
    },
    {
      "epoch": 0.7772520003441452,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009418189000750722,
      "loss": 2.1037,
      "step": 9034
    },
    {
      "epoch": 0.777338036651467,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009418062397363527,
      "loss": 1.9914,
      "step": 9035
    },
    {
      "epoch": 0.7774240729587886,
      "grad_norm": 1.578125,
      "learning_rate": 0.000941793578105436,
      "loss": 1.2941,
      "step": 9036
    },
    {
      "epoch": 0.7775101092661103,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000941780915182359,
      "loss": 1.4607,
      "step": 9037
    },
    {
      "epoch": 0.777596145573432,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009417682509671587,
      "loss": 1.278,
      "step": 9038
    },
    {
      "epoch": 0.7776821818807537,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009417555854598722,
      "loss": 1.7861,
      "step": 9039
    },
    {
      "epoch": 0.7777682181880754,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009417429186605364,
      "loss": 2.1397,
      "step": 9040
    },
    {
      "epoch": 0.777854254495397,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009417302505691887,
      "loss": 1.2345,
      "step": 9041
    },
    {
      "epoch": 0.7779402908027188,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009417175811858658,
      "loss": 1.1074,
      "step": 9042
    },
    {
      "epoch": 0.7780263271100404,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009417049105106049,
      "loss": 1.835,
      "step": 9043
    },
    {
      "epoch": 0.7781123634173621,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000941692238543443,
      "loss": 0.7965,
      "step": 9044
    },
    {
      "epoch": 0.7781983997246839,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009416795652844172,
      "loss": 1.992,
      "step": 9045
    },
    {
      "epoch": 0.7782844360320055,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009416668907335648,
      "loss": 1.838,
      "step": 9046
    },
    {
      "epoch": 0.7783704723393272,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009416542148909224,
      "loss": 1.9597,
      "step": 9047
    },
    {
      "epoch": 0.7784565086466488,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009416415377565274,
      "loss": 1.2264,
      "step": 9048
    },
    {
      "epoch": 0.7785425449539706,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009416288593304167,
      "loss": 1.4653,
      "step": 9049
    },
    {
      "epoch": 0.7786285812612923,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009416161796126277,
      "loss": 1.5431,
      "step": 9050
    },
    {
      "epoch": 0.7787146175686139,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009416034986031972,
      "loss": 0.8839,
      "step": 9051
    },
    {
      "epoch": 0.7788006538759357,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009415908163021622,
      "loss": 1.1484,
      "step": 9052
    },
    {
      "epoch": 0.7788866901832573,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00094157813270956,
      "loss": 1.9734,
      "step": 9053
    },
    {
      "epoch": 0.778972726490579,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009415654478254277,
      "loss": 1.1663,
      "step": 9054
    },
    {
      "epoch": 0.7790587627979008,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009415527616498023,
      "loss": 1.7351,
      "step": 9055
    },
    {
      "epoch": 0.7791447991052224,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009415400741827209,
      "loss": 0.9774,
      "step": 9056
    },
    {
      "epoch": 0.7792308354125441,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009415273854242208,
      "loss": 1.3726,
      "step": 9057
    },
    {
      "epoch": 0.7793168717198657,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009415146953743389,
      "loss": 1.5997,
      "step": 9058
    },
    {
      "epoch": 0.7794029080271875,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009415020040331124,
      "loss": 1.3465,
      "step": 9059
    },
    {
      "epoch": 0.7794889443345092,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009414893114005784,
      "loss": 2.0296,
      "step": 9060
    },
    {
      "epoch": 0.7795749806418308,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009414766174767739,
      "loss": 1.6803,
      "step": 9061
    },
    {
      "epoch": 0.7796610169491526,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009414639222617364,
      "loss": 1.4965,
      "step": 9062
    },
    {
      "epoch": 0.7797470532564742,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009414512257555026,
      "loss": 1.7725,
      "step": 9063
    },
    {
      "epoch": 0.7798330895637959,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009414385279581099,
      "loss": 1.4168,
      "step": 9064
    },
    {
      "epoch": 0.7799191258711176,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009414258288695954,
      "loss": 1.9493,
      "step": 9065
    },
    {
      "epoch": 0.7800051621784393,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009414131284899961,
      "loss": 1.1613,
      "step": 9066
    },
    {
      "epoch": 0.780091198485761,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009414004268193492,
      "loss": 1.1351,
      "step": 9067
    },
    {
      "epoch": 0.7801772347930827,
      "grad_norm": 8.3125,
      "learning_rate": 0.0009413877238576921,
      "loss": 1.9297,
      "step": 9068
    },
    {
      "epoch": 0.7802632711004044,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009413750196050617,
      "loss": 0.8098,
      "step": 9069
    },
    {
      "epoch": 0.780349307407726,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009413623140614951,
      "loss": 1.1236,
      "step": 9070
    },
    {
      "epoch": 0.7804353437150477,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009413496072270297,
      "loss": 1.6885,
      "step": 9071
    },
    {
      "epoch": 0.7805213800223695,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009413368991017024,
      "loss": 1.0242,
      "step": 9072
    },
    {
      "epoch": 0.7806074163296911,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009413241896855507,
      "loss": 1.1001,
      "step": 9073
    },
    {
      "epoch": 0.7806934526370128,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009413114789786113,
      "loss": 1.7437,
      "step": 9074
    },
    {
      "epoch": 0.7807794889443345,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009412987669809218,
      "loss": 0.7915,
      "step": 9075
    },
    {
      "epoch": 0.7808655252516562,
      "grad_norm": 0.8125,
      "learning_rate": 0.0009412860536925193,
      "loss": 0.89,
      "step": 9076
    },
    {
      "epoch": 0.7809515615589779,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000941273339113441,
      "loss": 0.7804,
      "step": 9077
    },
    {
      "epoch": 0.7810375978662996,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009412606232437237,
      "loss": 0.9993,
      "step": 9078
    },
    {
      "epoch": 0.7811236341736213,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009412479060834051,
      "loss": 1.1649,
      "step": 9079
    },
    {
      "epoch": 0.7812096704809429,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009412351876325222,
      "loss": 1.1236,
      "step": 9080
    },
    {
      "epoch": 0.7812957067882647,
      "grad_norm": 1.515625,
      "learning_rate": 0.000941222467891112,
      "loss": 1.2586,
      "step": 9081
    },
    {
      "epoch": 0.7813817430955864,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009412097468592121,
      "loss": 2.079,
      "step": 9082
    },
    {
      "epoch": 0.781467779402908,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009411970245368595,
      "loss": 1.2987,
      "step": 9083
    },
    {
      "epoch": 0.7815538157102297,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0009411843009240913,
      "loss": 0.6811,
      "step": 9084
    },
    {
      "epoch": 0.7816398520175514,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009411715760209449,
      "loss": 1.9095,
      "step": 9085
    },
    {
      "epoch": 0.7817258883248731,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009411588498274575,
      "loss": 0.9982,
      "step": 9086
    },
    {
      "epoch": 0.7818119246321947,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009411461223436661,
      "loss": 1.3852,
      "step": 9087
    },
    {
      "epoch": 0.7818979609395165,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009411333935696082,
      "loss": 1.704,
      "step": 9088
    },
    {
      "epoch": 0.7819839972468382,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009411206635053209,
      "loss": 1.1176,
      "step": 9089
    },
    {
      "epoch": 0.7820700335541598,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009411079321508415,
      "loss": 1.3228,
      "step": 9090
    },
    {
      "epoch": 0.7821560698614816,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009410951995062071,
      "loss": 1.4389,
      "step": 9091
    },
    {
      "epoch": 0.7822421061688032,
      "grad_norm": 1.5,
      "learning_rate": 0.0009410824655714551,
      "loss": 1.7831,
      "step": 9092
    },
    {
      "epoch": 0.7823281424761249,
      "grad_norm": 2.0,
      "learning_rate": 0.0009410697303466226,
      "loss": 1.027,
      "step": 9093
    },
    {
      "epoch": 0.7824141787834467,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009410569938317469,
      "loss": 2.0748,
      "step": 9094
    },
    {
      "epoch": 0.7825002150907683,
      "grad_norm": 2.453125,
      "learning_rate": 0.0009410442560268654,
      "loss": 1.1937,
      "step": 9095
    },
    {
      "epoch": 0.78258625139809,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009410315169320151,
      "loss": 1.3658,
      "step": 9096
    },
    {
      "epoch": 0.7826722877054116,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009410187765472336,
      "loss": 1.0218,
      "step": 9097
    },
    {
      "epoch": 0.7827583240127334,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009410060348725577,
      "loss": 2.289,
      "step": 9098
    },
    {
      "epoch": 0.7828443603200551,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009409932919080251,
      "loss": 1.3139,
      "step": 9099
    },
    {
      "epoch": 0.7829303966273767,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009409805476536729,
      "loss": 1.2968,
      "step": 9100
    },
    {
      "epoch": 0.7830164329346985,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009409678021095381,
      "loss": 1.3526,
      "step": 9101
    },
    {
      "epoch": 0.7831024692420201,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009409550552756586,
      "loss": 0.8825,
      "step": 9102
    },
    {
      "epoch": 0.7831885055493418,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009409423071520711,
      "loss": 0.9724,
      "step": 9103
    },
    {
      "epoch": 0.7832745418566636,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009409295577388131,
      "loss": 1.3425,
      "step": 9104
    },
    {
      "epoch": 0.7833605781639852,
      "grad_norm": 1.375,
      "learning_rate": 0.000940916807035922,
      "loss": 1.7327,
      "step": 9105
    },
    {
      "epoch": 0.7834466144713069,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009409040550434349,
      "loss": 1.1681,
      "step": 9106
    },
    {
      "epoch": 0.7835326507786285,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009408913017613891,
      "loss": 1.8879,
      "step": 9107
    },
    {
      "epoch": 0.7836186870859503,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009408785471898222,
      "loss": 1.3266,
      "step": 9108
    },
    {
      "epoch": 0.7837047233932719,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009408657913287712,
      "loss": 1.6966,
      "step": 9109
    },
    {
      "epoch": 0.7837907597005936,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009408530341782737,
      "loss": 1.4633,
      "step": 9110
    },
    {
      "epoch": 0.7838767960079154,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009408402757383665,
      "loss": 1.8067,
      "step": 9111
    },
    {
      "epoch": 0.783962832315237,
      "grad_norm": 6.03125,
      "learning_rate": 0.0009408275160090874,
      "loss": 1.912,
      "step": 9112
    },
    {
      "epoch": 0.7840488686225587,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009408147549904736,
      "loss": 1.4276,
      "step": 9113
    },
    {
      "epoch": 0.7841349049298804,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009408019926825622,
      "loss": 1.8298,
      "step": 9114
    },
    {
      "epoch": 0.7842209412372021,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009407892290853909,
      "loss": 1.6489,
      "step": 9115
    },
    {
      "epoch": 0.7843069775445238,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009407764641989968,
      "loss": 1.4928,
      "step": 9116
    },
    {
      "epoch": 0.7843930138518455,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009407636980234171,
      "loss": 1.1517,
      "step": 9117
    },
    {
      "epoch": 0.7844790501591672,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009407509305586894,
      "loss": 1.8522,
      "step": 9118
    },
    {
      "epoch": 0.7845650864664888,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009407381618048511,
      "loss": 1.6348,
      "step": 9119
    },
    {
      "epoch": 0.7846511227738105,
      "grad_norm": 1.5,
      "learning_rate": 0.0009407253917619392,
      "loss": 1.3646,
      "step": 9120
    },
    {
      "epoch": 0.7847371590811323,
      "grad_norm": 1.25,
      "learning_rate": 0.0009407126204299913,
      "loss": 1.5433,
      "step": 9121
    },
    {
      "epoch": 0.7848231953884539,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009406998478090447,
      "loss": 1.2309,
      "step": 9122
    },
    {
      "epoch": 0.7849092316957756,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009406870738991368,
      "loss": 1.58,
      "step": 9123
    },
    {
      "epoch": 0.7849952680030973,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009406742987003049,
      "loss": 1.0096,
      "step": 9124
    },
    {
      "epoch": 0.785081304310419,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009406615222125863,
      "loss": 1.3691,
      "step": 9125
    },
    {
      "epoch": 0.7851673406177407,
      "grad_norm": 1.25,
      "learning_rate": 0.0009406487444360186,
      "loss": 1.1161,
      "step": 9126
    },
    {
      "epoch": 0.7852533769250624,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009406359653706389,
      "loss": 1.2769,
      "step": 9127
    },
    {
      "epoch": 0.7853394132323841,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009406231850164848,
      "loss": 1.2533,
      "step": 9128
    },
    {
      "epoch": 0.7854254495397057,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009406104033735934,
      "loss": 2.3133,
      "step": 9129
    },
    {
      "epoch": 0.7855114858470275,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009405976204420024,
      "loss": 1.5692,
      "step": 9130
    },
    {
      "epoch": 0.7855975221543491,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000940584836221749,
      "loss": 1.2894,
      "step": 9131
    },
    {
      "epoch": 0.7856835584616708,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009405720507128707,
      "loss": 1.3212,
      "step": 9132
    },
    {
      "epoch": 0.7857695947689926,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009405592639154048,
      "loss": 1.9802,
      "step": 9133
    },
    {
      "epoch": 0.7858556310763142,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009405464758293889,
      "loss": 1.5109,
      "step": 9134
    },
    {
      "epoch": 0.7859416673836359,
      "grad_norm": 1.65625,
      "learning_rate": 0.00094053368645486,
      "loss": 1.706,
      "step": 9135
    },
    {
      "epoch": 0.7860277036909575,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009405208957918559,
      "loss": 1.8249,
      "step": 9136
    },
    {
      "epoch": 0.7861137399982793,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009405081038404138,
      "loss": 1.4673,
      "step": 9137
    },
    {
      "epoch": 0.786199776305601,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009404953106005712,
      "loss": 1.1882,
      "step": 9138
    },
    {
      "epoch": 0.7862858126129226,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009404825160723654,
      "loss": 1.8795,
      "step": 9139
    },
    {
      "epoch": 0.7863718489202444,
      "grad_norm": 1.25,
      "learning_rate": 0.0009404697202558339,
      "loss": 1.4062,
      "step": 9140
    },
    {
      "epoch": 0.786457885227566,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009404569231510143,
      "loss": 1.3532,
      "step": 9141
    },
    {
      "epoch": 0.7865439215348877,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009404441247579437,
      "loss": 1.5886,
      "step": 9142
    },
    {
      "epoch": 0.7866299578422095,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009404313250766599,
      "loss": 1.0555,
      "step": 9143
    },
    {
      "epoch": 0.7867159941495311,
      "grad_norm": 2.25,
      "learning_rate": 0.0009404185241071999,
      "loss": 1.1619,
      "step": 9144
    },
    {
      "epoch": 0.7868020304568528,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009404057218496014,
      "loss": 1.008,
      "step": 9145
    },
    {
      "epoch": 0.7868880667641744,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000940392918303902,
      "loss": 0.7819,
      "step": 9146
    },
    {
      "epoch": 0.7869741030714962,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009403801134701389,
      "loss": 1.3628,
      "step": 9147
    },
    {
      "epoch": 0.7870601393788179,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009403673073483494,
      "loss": 1.0959,
      "step": 9148
    },
    {
      "epoch": 0.7871461756861395,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009403544999385714,
      "loss": 1.0052,
      "step": 9149
    },
    {
      "epoch": 0.7872322119934613,
      "grad_norm": 1.375,
      "learning_rate": 0.000940341691240842,
      "loss": 1.4526,
      "step": 9150
    },
    {
      "epoch": 0.7873182483007829,
      "grad_norm": 1.125,
      "learning_rate": 0.0009403288812551988,
      "loss": 1.0688,
      "step": 9151
    },
    {
      "epoch": 0.7874042846081046,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009403160699816795,
      "loss": 2.0361,
      "step": 9152
    },
    {
      "epoch": 0.7874903209154263,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009403032574203211,
      "loss": 1.2783,
      "step": 9153
    },
    {
      "epoch": 0.787576357222748,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009402904435711614,
      "loss": 0.8513,
      "step": 9154
    },
    {
      "epoch": 0.7876623935300697,
      "grad_norm": 1.625,
      "learning_rate": 0.0009402776284342378,
      "loss": 0.8417,
      "step": 9155
    },
    {
      "epoch": 0.7877484298373914,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009402648120095877,
      "loss": 1.8268,
      "step": 9156
    },
    {
      "epoch": 0.7878344661447131,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009402519942972486,
      "loss": 1.6361,
      "step": 9157
    },
    {
      "epoch": 0.7879205024520347,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009402391752972581,
      "loss": 1.2629,
      "step": 9158
    },
    {
      "epoch": 0.7880065387593564,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009402263550096537,
      "loss": 1.7578,
      "step": 9159
    },
    {
      "epoch": 0.7880925750666782,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009402135334344728,
      "loss": 1.3911,
      "step": 9160
    },
    {
      "epoch": 0.7881786113739998,
      "grad_norm": 1.5,
      "learning_rate": 0.0009402007105717529,
      "loss": 1.9096,
      "step": 9161
    },
    {
      "epoch": 0.7882646476813215,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009401878864215315,
      "loss": 1.0506,
      "step": 9162
    },
    {
      "epoch": 0.7883506839886432,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009401750609838462,
      "loss": 1.4832,
      "step": 9163
    },
    {
      "epoch": 0.7884367202959649,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009401622342587345,
      "loss": 0.8835,
      "step": 9164
    },
    {
      "epoch": 0.7885227566032866,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009401494062462337,
      "loss": 0.8698,
      "step": 9165
    },
    {
      "epoch": 0.7886087929106083,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009401365769463817,
      "loss": 2.3742,
      "step": 9166
    },
    {
      "epoch": 0.78869482921793,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009401237463592156,
      "loss": 1.831,
      "step": 9167
    },
    {
      "epoch": 0.7887808655252516,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009401109144847734,
      "loss": 1.2706,
      "step": 9168
    },
    {
      "epoch": 0.7888669018325734,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009400980813230921,
      "loss": 1.7687,
      "step": 9169
    },
    {
      "epoch": 0.7889529381398951,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009400852468742097,
      "loss": 2.0346,
      "step": 9170
    },
    {
      "epoch": 0.7890389744472167,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009400724111381636,
      "loss": 1.2785,
      "step": 9171
    },
    {
      "epoch": 0.7891250107545384,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009400595741149911,
      "loss": 1.5053,
      "step": 9172
    },
    {
      "epoch": 0.7892110470618601,
      "grad_norm": 2.0,
      "learning_rate": 0.00094004673580473,
      "loss": 1.8687,
      "step": 9173
    },
    {
      "epoch": 0.7892970833691818,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009400338962074178,
      "loss": 1.087,
      "step": 9174
    },
    {
      "epoch": 0.7893831196765034,
      "grad_norm": 1.015625,
      "learning_rate": 0.000940021055323092,
      "loss": 1.6606,
      "step": 9175
    },
    {
      "epoch": 0.7894691559838252,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009400082131517902,
      "loss": 0.8637,
      "step": 9176
    },
    {
      "epoch": 0.7895551922911469,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00093999536969355,
      "loss": 1.037,
      "step": 9177
    },
    {
      "epoch": 0.7896412285984685,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009399825249484089,
      "loss": 1.0421,
      "step": 9178
    },
    {
      "epoch": 0.7897272649057903,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009399696789164045,
      "loss": 1.7527,
      "step": 9179
    },
    {
      "epoch": 0.7898133012131119,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009399568315975743,
      "loss": 1.2215,
      "step": 9180
    },
    {
      "epoch": 0.7898993375204336,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009399439829919559,
      "loss": 1.1982,
      "step": 9181
    },
    {
      "epoch": 0.7899853738277554,
      "grad_norm": 1.03125,
      "learning_rate": 0.000939931133099587,
      "loss": 0.7474,
      "step": 9182
    },
    {
      "epoch": 0.790071410135077,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009399182819205052,
      "loss": 1.1532,
      "step": 9183
    },
    {
      "epoch": 0.7901574464423987,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0009399054294547478,
      "loss": 1.1165,
      "step": 9184
    },
    {
      "epoch": 0.7902434827497203,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009398925757023527,
      "loss": 1.4283,
      "step": 9185
    },
    {
      "epoch": 0.7903295190570421,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009398797206633571,
      "loss": 1.6562,
      "step": 9186
    },
    {
      "epoch": 0.7904155553643638,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009398668643377992,
      "loss": 1.3132,
      "step": 9187
    },
    {
      "epoch": 0.7905015916716854,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009398540067257162,
      "loss": 0.7487,
      "step": 9188
    },
    {
      "epoch": 0.7905876279790072,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009398411478271456,
      "loss": 1.9608,
      "step": 9189
    },
    {
      "epoch": 0.7906736642863288,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009398282876421254,
      "loss": 1.3951,
      "step": 9190
    },
    {
      "epoch": 0.7907597005936505,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009398154261706928,
      "loss": 1.1463,
      "step": 9191
    },
    {
      "epoch": 0.7908457369009723,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009398025634128857,
      "loss": 1.5078,
      "step": 9192
    },
    {
      "epoch": 0.7909317732082939,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009397896993687419,
      "loss": 1.2244,
      "step": 9193
    },
    {
      "epoch": 0.7910178095156156,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009397768340382984,
      "loss": 2.4537,
      "step": 9194
    },
    {
      "epoch": 0.7911038458229372,
      "grad_norm": 0.6875,
      "learning_rate": 0.0009397639674215933,
      "loss": 1.0481,
      "step": 9195
    },
    {
      "epoch": 0.791189882130259,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009397510995186642,
      "loss": 1.3222,
      "step": 9196
    },
    {
      "epoch": 0.7912759184375806,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009397382303295485,
      "loss": 1.1271,
      "step": 9197
    },
    {
      "epoch": 0.7913619547449023,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009397253598542842,
      "loss": 1.3688,
      "step": 9198
    },
    {
      "epoch": 0.7914479910522241,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009397124880929086,
      "loss": 2.0631,
      "step": 9199
    },
    {
      "epoch": 0.7915340273595457,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009396996150454596,
      "loss": 1.0898,
      "step": 9200
    },
    {
      "epoch": 0.7916200636668674,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009396867407119747,
      "loss": 1.8055,
      "step": 9201
    },
    {
      "epoch": 0.7917060999741891,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009396738650924916,
      "loss": 1.4808,
      "step": 9202
    },
    {
      "epoch": 0.7917921362815108,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000939660988187048,
      "loss": 1.7535,
      "step": 9203
    },
    {
      "epoch": 0.7918781725888325,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009396481099956813,
      "loss": 0.9436,
      "step": 9204
    },
    {
      "epoch": 0.7919642088961542,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009396352305184297,
      "loss": 1.1363,
      "step": 9205
    },
    {
      "epoch": 0.7920502452034759,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009396223497553304,
      "loss": 1.7388,
      "step": 9206
    },
    {
      "epoch": 0.7921362815107975,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009396094677064211,
      "loss": 2.102,
      "step": 9207
    },
    {
      "epoch": 0.7922223178181192,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009395965843717398,
      "loss": 1.2203,
      "step": 9208
    },
    {
      "epoch": 0.792308354125441,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000939583699751324,
      "loss": 1.6831,
      "step": 9209
    },
    {
      "epoch": 0.7923943904327626,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009395708138452113,
      "loss": 0.8356,
      "step": 9210
    },
    {
      "epoch": 0.7924804267400843,
      "grad_norm": 1.5,
      "learning_rate": 0.0009395579266534394,
      "loss": 1.145,
      "step": 9211
    },
    {
      "epoch": 0.792566463047406,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009395450381760461,
      "loss": 1.4051,
      "step": 9212
    },
    {
      "epoch": 0.7926524993547277,
      "grad_norm": 1.375,
      "learning_rate": 0.000939532148413069,
      "loss": 1.256,
      "step": 9213
    },
    {
      "epoch": 0.7927385356620494,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009395192573645459,
      "loss": 0.78,
      "step": 9214
    },
    {
      "epoch": 0.7928245719693711,
      "grad_norm": 1.0,
      "learning_rate": 0.0009395063650305145,
      "loss": 1.593,
      "step": 9215
    },
    {
      "epoch": 0.7929106082766928,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009394934714110125,
      "loss": 1.3818,
      "step": 9216
    },
    {
      "epoch": 0.7929966445840144,
      "grad_norm": 2.125,
      "learning_rate": 0.0009394805765060774,
      "loss": 1.5472,
      "step": 9217
    },
    {
      "epoch": 0.7930826808913362,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009394676803157472,
      "loss": 1.7531,
      "step": 9218
    },
    {
      "epoch": 0.7931687171986578,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009394547828400595,
      "loss": 1.3811,
      "step": 9219
    },
    {
      "epoch": 0.7932547535059795,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009394418840790522,
      "loss": 0.877,
      "step": 9220
    },
    {
      "epoch": 0.7933407898133013,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009394289840327626,
      "loss": 1.5179,
      "step": 9221
    },
    {
      "epoch": 0.7934268261206229,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009394160827012288,
      "loss": 1.4923,
      "step": 9222
    },
    {
      "epoch": 0.7935128624279446,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009394031800844883,
      "loss": 1.0172,
      "step": 9223
    },
    {
      "epoch": 0.7935988987352662,
      "grad_norm": 1.625,
      "learning_rate": 0.000939390276182579,
      "loss": 1.6216,
      "step": 9224
    },
    {
      "epoch": 0.793684935042588,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009393773709955387,
      "loss": 1.1408,
      "step": 9225
    },
    {
      "epoch": 0.7937709713499097,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009393644645234049,
      "loss": 1.8414,
      "step": 9226
    },
    {
      "epoch": 0.7938570076572313,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009393515567662156,
      "loss": 1.5203,
      "step": 9227
    },
    {
      "epoch": 0.7939430439645531,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009393386477240085,
      "loss": 1.158,
      "step": 9228
    },
    {
      "epoch": 0.7940290802718747,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009393257373968213,
      "loss": 1.0065,
      "step": 9229
    },
    {
      "epoch": 0.7941151165791964,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009393128257846917,
      "loss": 1.6225,
      "step": 9230
    },
    {
      "epoch": 0.7942011528865182,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009392999128876574,
      "loss": 2.3031,
      "step": 9231
    },
    {
      "epoch": 0.7942871891938398,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009392869987057564,
      "loss": 0.9845,
      "step": 9232
    },
    {
      "epoch": 0.7943732255011615,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009392740832390264,
      "loss": 2.148,
      "step": 9233
    },
    {
      "epoch": 0.7944592618084831,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009392611664875051,
      "loss": 1.3168,
      "step": 9234
    },
    {
      "epoch": 0.7945452981158049,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009392482484512303,
      "loss": 1.4769,
      "step": 9235
    },
    {
      "epoch": 0.7946313344231266,
      "grad_norm": 2.296875,
      "learning_rate": 0.0009392353291302401,
      "loss": 1.508,
      "step": 9236
    },
    {
      "epoch": 0.7947173707304482,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009392224085245716,
      "loss": 1.7408,
      "step": 9237
    },
    {
      "epoch": 0.79480340703777,
      "grad_norm": 1.8203125,
      "learning_rate": 0.000939209486634263,
      "loss": 1.1256,
      "step": 9238
    },
    {
      "epoch": 0.7948894433450916,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009391965634593522,
      "loss": 1.8279,
      "step": 9239
    },
    {
      "epoch": 0.7949754796524133,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009391836389998769,
      "loss": 1.4811,
      "step": 9240
    },
    {
      "epoch": 0.795061515959735,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009391707132558748,
      "loss": 1.8473,
      "step": 9241
    },
    {
      "epoch": 0.7951475522670567,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009391577862273838,
      "loss": 0.8209,
      "step": 9242
    },
    {
      "epoch": 0.7952335885743784,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009391448579144417,
      "loss": 1.3523,
      "step": 9243
    },
    {
      "epoch": 0.7953196248817,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009391319283170862,
      "loss": 1.2708,
      "step": 9244
    },
    {
      "epoch": 0.7954056611890218,
      "grad_norm": 1.0,
      "learning_rate": 0.0009391189974353553,
      "loss": 1.0338,
      "step": 9245
    },
    {
      "epoch": 0.7954916974963434,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009391060652692869,
      "loss": 1.155,
      "step": 9246
    },
    {
      "epoch": 0.7955777338036651,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009390931318189184,
      "loss": 2.0922,
      "step": 9247
    },
    {
      "epoch": 0.7956637701109869,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000939080197084288,
      "loss": 1.3302,
      "step": 9248
    },
    {
      "epoch": 0.7957498064183085,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009390672610654334,
      "loss": 1.9495,
      "step": 9249
    },
    {
      "epoch": 0.7958358427256302,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009390543237623925,
      "loss": 1.3664,
      "step": 9250
    },
    {
      "epoch": 0.7959218790329519,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009390413851752031,
      "loss": 1.2231,
      "step": 9251
    },
    {
      "epoch": 0.7960079153402736,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000939028445303903,
      "loss": 1.2179,
      "step": 9252
    },
    {
      "epoch": 0.7960939516475953,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00093901550414853,
      "loss": 1.1498,
      "step": 9253
    },
    {
      "epoch": 0.796179987954917,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009390025617091221,
      "loss": 1.2419,
      "step": 9254
    },
    {
      "epoch": 0.7962660242622387,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009389896179857172,
      "loss": 1.8026,
      "step": 9255
    },
    {
      "epoch": 0.7963520605695603,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009389766729783528,
      "loss": 1.741,
      "step": 9256
    },
    {
      "epoch": 0.796438096876882,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009389637266870672,
      "loss": 1.5875,
      "step": 9257
    },
    {
      "epoch": 0.7965241331842038,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009389507791118981,
      "loss": 0.9814,
      "step": 9258
    },
    {
      "epoch": 0.7966101694915254,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009389378302528833,
      "loss": 1.4593,
      "step": 9259
    },
    {
      "epoch": 0.7966962057988471,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009389248801100606,
      "loss": 1.3727,
      "step": 9260
    },
    {
      "epoch": 0.7967822421061688,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009389119286834681,
      "loss": 1.6777,
      "step": 9261
    },
    {
      "epoch": 0.7968682784134905,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009388989759731435,
      "loss": 1.1747,
      "step": 9262
    },
    {
      "epoch": 0.7969543147208121,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009388860219791249,
      "loss": 1.4233,
      "step": 9263
    },
    {
      "epoch": 0.7970403510281339,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009388730667014499,
      "loss": 1.8838,
      "step": 9264
    },
    {
      "epoch": 0.7971263873354556,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009388601101401566,
      "loss": 1.1875,
      "step": 9265
    },
    {
      "epoch": 0.7972124236427772,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009388471522952829,
      "loss": 1.7395,
      "step": 9266
    },
    {
      "epoch": 0.797298459950099,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009388341931668664,
      "loss": 1.67,
      "step": 9267
    },
    {
      "epoch": 0.7973844962574206,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009388212327549455,
      "loss": 2.3107,
      "step": 9268
    },
    {
      "epoch": 0.7974705325647423,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009388082710595577,
      "loss": 0.9225,
      "step": 9269
    },
    {
      "epoch": 0.797556568872064,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000938795308080741,
      "loss": 1.0214,
      "step": 9270
    },
    {
      "epoch": 0.7976426051793857,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009387823438185334,
      "loss": 1.2963,
      "step": 9271
    },
    {
      "epoch": 0.7977286414867074,
      "grad_norm": 1.015625,
      "learning_rate": 0.000938769378272973,
      "loss": 1.1377,
      "step": 9272
    },
    {
      "epoch": 0.797814677794029,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009387564114440972,
      "loss": 2.0116,
      "step": 9273
    },
    {
      "epoch": 0.7979007141013508,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009387434433319444,
      "loss": 1.2796,
      "step": 9274
    },
    {
      "epoch": 0.7979867504086725,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009387304739365524,
      "loss": 1.213,
      "step": 9275
    },
    {
      "epoch": 0.7980727867159941,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009387175032579589,
      "loss": 1.1649,
      "step": 9276
    },
    {
      "epoch": 0.7981588230233159,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009387045312962022,
      "loss": 1.2704,
      "step": 9277
    },
    {
      "epoch": 0.7982448593306375,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00093869155805132,
      "loss": 1.4705,
      "step": 9278
    },
    {
      "epoch": 0.7983308956379592,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009386785835233503,
      "loss": 1.7416,
      "step": 9279
    },
    {
      "epoch": 0.798416931945281,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000938665607712331,
      "loss": 0.9462,
      "step": 9280
    },
    {
      "epoch": 0.7985029682526026,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009386526306183003,
      "loss": 1.2584,
      "step": 9281
    },
    {
      "epoch": 0.7985890045599243,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009386396522412958,
      "loss": 1.6271,
      "step": 9282
    },
    {
      "epoch": 0.7986750408672459,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009386266725813556,
      "loss": 1.7747,
      "step": 9283
    },
    {
      "epoch": 0.7987610771745677,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009386136916385179,
      "loss": 1.2704,
      "step": 9284
    },
    {
      "epoch": 0.7988471134818893,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009386007094128204,
      "loss": 1.628,
      "step": 9285
    },
    {
      "epoch": 0.798933149789211,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009385877259043009,
      "loss": 1.3725,
      "step": 9286
    },
    {
      "epoch": 0.7990191860965328,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009385747411129979,
      "loss": 1.1745,
      "step": 9287
    },
    {
      "epoch": 0.7991052224038544,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009385617550389489,
      "loss": 1.0297,
      "step": 9288
    },
    {
      "epoch": 0.7991912587111761,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009385487676821921,
      "loss": 1.6035,
      "step": 9289
    },
    {
      "epoch": 0.7992772950184978,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009385357790427654,
      "loss": 0.9041,
      "step": 9290
    },
    {
      "epoch": 0.7993633313258195,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009385227891207068,
      "loss": 1.5536,
      "step": 9291
    },
    {
      "epoch": 0.7994493676331412,
      "grad_norm": 1.25,
      "learning_rate": 0.0009385097979160546,
      "loss": 2.1965,
      "step": 9292
    },
    {
      "epoch": 0.7995354039404629,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0009384968054288463,
      "loss": 1.2622,
      "step": 9293
    },
    {
      "epoch": 0.7996214402477846,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009384838116591202,
      "loss": 1.691,
      "step": 9294
    },
    {
      "epoch": 0.7997074765551062,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009384708166069142,
      "loss": 1.8128,
      "step": 9295
    },
    {
      "epoch": 0.799793512862428,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009384578202722663,
      "loss": 1.4253,
      "step": 9296
    },
    {
      "epoch": 0.7998795491697497,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009384448226552146,
      "loss": 1.6031,
      "step": 9297
    },
    {
      "epoch": 0.7999655854770713,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009384318237557972,
      "loss": 1.2957,
      "step": 9298
    },
    {
      "epoch": 0.800051621784393,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009384188235740518,
      "loss": 1.624,
      "step": 9299
    },
    {
      "epoch": 0.8001376580917147,
      "grad_norm": 1.125,
      "learning_rate": 0.0009384058221100167,
      "loss": 0.6832,
      "step": 9300
    },
    {
      "epoch": 0.8002236943990364,
      "grad_norm": 4.40625,
      "learning_rate": 0.0009383928193637297,
      "loss": 1.4253,
      "step": 9301
    },
    {
      "epoch": 0.8003097307063581,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009383798153352291,
      "loss": 1.1627,
      "step": 9302
    },
    {
      "epoch": 0.8003957670136798,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009383668100245528,
      "loss": 1.304,
      "step": 9303
    },
    {
      "epoch": 0.8004818033210015,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009383538034317387,
      "loss": 1.2581,
      "step": 9304
    },
    {
      "epoch": 0.8005678396283231,
      "grad_norm": 1.234375,
      "learning_rate": 0.000938340795556825,
      "loss": 1.3626,
      "step": 9305
    },
    {
      "epoch": 0.8006538759356449,
      "grad_norm": 1.140625,
      "learning_rate": 0.00093832778639985,
      "loss": 1.0846,
      "step": 9306
    },
    {
      "epoch": 0.8007399122429665,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009383147759608512,
      "loss": 1.3039,
      "step": 9307
    },
    {
      "epoch": 0.8008259485502882,
      "grad_norm": 1.1875,
      "learning_rate": 0.000938301764239867,
      "loss": 1.5365,
      "step": 9308
    },
    {
      "epoch": 0.80091198485761,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009382887512369354,
      "loss": 0.7322,
      "step": 9309
    },
    {
      "epoch": 0.8009980211649316,
      "grad_norm": 2.109375,
      "learning_rate": 0.0009382757369520944,
      "loss": 1.0568,
      "step": 9310
    },
    {
      "epoch": 0.8010840574722533,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009382627213853822,
      "loss": 1.2149,
      "step": 9311
    },
    {
      "epoch": 0.8011700937795749,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009382497045368368,
      "loss": 1.6443,
      "step": 9312
    },
    {
      "epoch": 0.8012561300868967,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009382366864064962,
      "loss": 0.8997,
      "step": 9313
    },
    {
      "epoch": 0.8013421663942184,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009382236669943985,
      "loss": 1.6594,
      "step": 9314
    },
    {
      "epoch": 0.80142820270154,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009382106463005817,
      "loss": 0.8843,
      "step": 9315
    },
    {
      "epoch": 0.8015142390088618,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000938197624325084,
      "loss": 0.7082,
      "step": 9316
    },
    {
      "epoch": 0.8016002753161834,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009381846010679437,
      "loss": 1.995,
      "step": 9317
    },
    {
      "epoch": 0.8016863116235051,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009381715765291985,
      "loss": 1.7053,
      "step": 9318
    },
    {
      "epoch": 0.8017723479308269,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009381585507088867,
      "loss": 1.0349,
      "step": 9319
    },
    {
      "epoch": 0.8018583842381485,
      "grad_norm": 1.25,
      "learning_rate": 0.0009381455236070463,
      "loss": 2.2466,
      "step": 9320
    },
    {
      "epoch": 0.8019444205454702,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009381324952237156,
      "loss": 1.5774,
      "step": 9321
    },
    {
      "epoch": 0.8020304568527918,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0009381194655589325,
      "loss": 0.6863,
      "step": 9322
    },
    {
      "epoch": 0.8021164931601136,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009381064346127351,
      "loss": 1.0412,
      "step": 9323
    },
    {
      "epoch": 0.8022025294674353,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009380934023851615,
      "loss": 0.7706,
      "step": 9324
    },
    {
      "epoch": 0.8022885657747569,
      "grad_norm": 1.375,
      "learning_rate": 0.0009380803688762501,
      "loss": 1.8666,
      "step": 9325
    },
    {
      "epoch": 0.8023746020820787,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009380673340860387,
      "loss": 1.8214,
      "step": 9326
    },
    {
      "epoch": 0.8024606383894003,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009380542980145656,
      "loss": 1.4704,
      "step": 9327
    },
    {
      "epoch": 0.802546674696722,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009380412606618689,
      "loss": 1.0264,
      "step": 9328
    },
    {
      "epoch": 0.8026327110040437,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009380282220279866,
      "loss": 1.1443,
      "step": 9329
    },
    {
      "epoch": 0.8027187473113654,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000938015182112957,
      "loss": 1.6039,
      "step": 9330
    },
    {
      "epoch": 0.8028047836186871,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009380021409168182,
      "loss": 1.3669,
      "step": 9331
    },
    {
      "epoch": 0.8028908199260087,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009379890984396084,
      "loss": 1.682,
      "step": 9332
    },
    {
      "epoch": 0.8029768562333305,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009379760546813656,
      "loss": 1.1102,
      "step": 9333
    },
    {
      "epoch": 0.8030628925406521,
      "grad_norm": 1.109375,
      "learning_rate": 0.000937963009642128,
      "loss": 1.9449,
      "step": 9334
    },
    {
      "epoch": 0.8031489288479738,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009379499633219337,
      "loss": 1.8237,
      "step": 9335
    },
    {
      "epoch": 0.8032349651552956,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009379369157208211,
      "loss": 0.6263,
      "step": 9336
    },
    {
      "epoch": 0.8033210014626172,
      "grad_norm": 1.375,
      "learning_rate": 0.0009379238668388282,
      "loss": 1.6925,
      "step": 9337
    },
    {
      "epoch": 0.8034070377699389,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009379108166759931,
      "loss": 1.0043,
      "step": 9338
    },
    {
      "epoch": 0.8034930740772606,
      "grad_norm": 1.375,
      "learning_rate": 0.000937897765232354,
      "loss": 1.2026,
      "step": 9339
    },
    {
      "epoch": 0.8035791103845823,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009378847125079491,
      "loss": 1.2957,
      "step": 9340
    },
    {
      "epoch": 0.803665146691904,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009378716585028167,
      "loss": 0.9445,
      "step": 9341
    },
    {
      "epoch": 0.8037511829992257,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009378586032169947,
      "loss": 1.2213,
      "step": 9342
    },
    {
      "epoch": 0.8038372193065474,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009378455466505214,
      "loss": 2.1252,
      "step": 9343
    },
    {
      "epoch": 0.803923255613869,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009378324888034351,
      "loss": 2.0639,
      "step": 9344
    },
    {
      "epoch": 0.8040092919211907,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000937819429675774,
      "loss": 0.6979,
      "step": 9345
    },
    {
      "epoch": 0.8040953282285125,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009378063692675763,
      "loss": 0.9899,
      "step": 9346
    },
    {
      "epoch": 0.8041813645358341,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009377933075788798,
      "loss": 1.1258,
      "step": 9347
    },
    {
      "epoch": 0.8042674008431558,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009377802446097232,
      "loss": 1.8016,
      "step": 9348
    },
    {
      "epoch": 0.8043534371504775,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009377671803601446,
      "loss": 1.2216,
      "step": 9349
    },
    {
      "epoch": 0.8044394734577992,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009377541148301821,
      "loss": 1.5316,
      "step": 9350
    },
    {
      "epoch": 0.8045255097651208,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000937741048019874,
      "loss": 1.3989,
      "step": 9351
    },
    {
      "epoch": 0.8046115460724426,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009377279799292584,
      "loss": 1.4194,
      "step": 9352
    },
    {
      "epoch": 0.8046975823797643,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009377149105583736,
      "loss": 1.4284,
      "step": 9353
    },
    {
      "epoch": 0.8047836186870859,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009377018399072578,
      "loss": 1.5992,
      "step": 9354
    },
    {
      "epoch": 0.8048696549944077,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009376887679759492,
      "loss": 1.6242,
      "step": 9355
    },
    {
      "epoch": 0.8049556913017293,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009376756947644862,
      "loss": 1.1204,
      "step": 9356
    },
    {
      "epoch": 0.805041727609051,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009376626202729068,
      "loss": 0.9795,
      "step": 9357
    },
    {
      "epoch": 0.8051277639163728,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009376495445012494,
      "loss": 1.1429,
      "step": 9358
    },
    {
      "epoch": 0.8052138002236944,
      "grad_norm": 1.8125,
      "learning_rate": 0.0009376364674495523,
      "loss": 2.0407,
      "step": 9359
    },
    {
      "epoch": 0.8052998365310161,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009376233891178536,
      "loss": 1.2574,
      "step": 9360
    },
    {
      "epoch": 0.8053858728383377,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009376103095061914,
      "loss": 0.9676,
      "step": 9361
    },
    {
      "epoch": 0.8054719091456595,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009375972286146044,
      "loss": 0.7891,
      "step": 9362
    },
    {
      "epoch": 0.8055579454529812,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009375841464431305,
      "loss": 1.872,
      "step": 9363
    },
    {
      "epoch": 0.8056439817603028,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009375710629918082,
      "loss": 0.6775,
      "step": 9364
    },
    {
      "epoch": 0.8057300180676246,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009375579782606756,
      "loss": 1.4247,
      "step": 9365
    },
    {
      "epoch": 0.8058160543749462,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009375448922497709,
      "loss": 1.8558,
      "step": 9366
    },
    {
      "epoch": 0.8059020906822679,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009375318049591326,
      "loss": 1.1766,
      "step": 9367
    },
    {
      "epoch": 0.8059881269895897,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009375187163887988,
      "loss": 2.3056,
      "step": 9368
    },
    {
      "epoch": 0.8060741632969113,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009375056265388078,
      "loss": 2.0973,
      "step": 9369
    },
    {
      "epoch": 0.806160199604233,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009374925354091979,
      "loss": 1.0628,
      "step": 9370
    },
    {
      "epoch": 0.8062462359115546,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009374794430000076,
      "loss": 1.7341,
      "step": 9371
    },
    {
      "epoch": 0.8063322722188764,
      "grad_norm": 1.25,
      "learning_rate": 0.000937466349311275,
      "loss": 1.272,
      "step": 9372
    },
    {
      "epoch": 0.806418308526198,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009374532543430383,
      "loss": 0.9092,
      "step": 9373
    },
    {
      "epoch": 0.8065043448335197,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000937440158095336,
      "loss": 1.5123,
      "step": 9374
    },
    {
      "epoch": 0.8065903811408415,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009374270605682063,
      "loss": 1.3153,
      "step": 9375
    },
    {
      "epoch": 0.8066764174481631,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009374139617616874,
      "loss": 1.3759,
      "step": 9376
    },
    {
      "epoch": 0.8067624537554848,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009374008616758178,
      "loss": 1.7395,
      "step": 9377
    },
    {
      "epoch": 0.8068484900628065,
      "grad_norm": 1.375,
      "learning_rate": 0.0009373877603106359,
      "loss": 1.9509,
      "step": 9378
    },
    {
      "epoch": 0.8069345263701282,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009373746576661797,
      "loss": 1.893,
      "step": 9379
    },
    {
      "epoch": 0.8070205626774499,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009373615537424878,
      "loss": 0.8901,
      "step": 9380
    },
    {
      "epoch": 0.8071065989847716,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009373484485395982,
      "loss": 0.8201,
      "step": 9381
    },
    {
      "epoch": 0.8071926352920933,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009373353420575497,
      "loss": 1.1989,
      "step": 9382
    },
    {
      "epoch": 0.8072786715994149,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009373222342963803,
      "loss": 1.0604,
      "step": 9383
    },
    {
      "epoch": 0.8073647079067366,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009373091252561285,
      "loss": 1.9923,
      "step": 9384
    },
    {
      "epoch": 0.8074507442140584,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009372960149368324,
      "loss": 1.5398,
      "step": 9385
    },
    {
      "epoch": 0.80753678052138,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009372829033385306,
      "loss": 1.6252,
      "step": 9386
    },
    {
      "epoch": 0.8076228168287017,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009372697904612612,
      "loss": 0.9957,
      "step": 9387
    },
    {
      "epoch": 0.8077088531360234,
      "grad_norm": 1.53125,
      "learning_rate": 0.000937256676305063,
      "loss": 1.7091,
      "step": 9388
    },
    {
      "epoch": 0.8077948894433451,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009372435608699739,
      "loss": 2.3413,
      "step": 9389
    },
    {
      "epoch": 0.8078809257506668,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009372304441560323,
      "loss": 1.2856,
      "step": 9390
    },
    {
      "epoch": 0.8079669620579885,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009372173261632768,
      "loss": 1.3238,
      "step": 9391
    },
    {
      "epoch": 0.8080529983653102,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009372042068917457,
      "loss": 1.5843,
      "step": 9392
    },
    {
      "epoch": 0.8081390346726318,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009371910863414772,
      "loss": 1.5488,
      "step": 9393
    },
    {
      "epoch": 0.8082250709799536,
      "grad_norm": 2.015625,
      "learning_rate": 0.00093717796451251,
      "loss": 1.1498,
      "step": 9394
    },
    {
      "epoch": 0.8083111072872752,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009371648414048822,
      "loss": 1.2167,
      "step": 9395
    },
    {
      "epoch": 0.8083971435945969,
      "grad_norm": 2.15625,
      "learning_rate": 0.0009371517170186321,
      "loss": 1.4898,
      "step": 9396
    },
    {
      "epoch": 0.8084831799019186,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0009371385913537984,
      "loss": 0.4356,
      "step": 9397
    },
    {
      "epoch": 0.8085692162092403,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009371254644104193,
      "loss": 1.6381,
      "step": 9398
    },
    {
      "epoch": 0.808655252516562,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009371123361885333,
      "loss": 1.2232,
      "step": 9399
    },
    {
      "epoch": 0.8087412888238836,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009370992066881787,
      "loss": 1.9671,
      "step": 9400
    },
    {
      "epoch": 0.8088273251312054,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000937086075909394,
      "loss": 0.977,
      "step": 9401
    },
    {
      "epoch": 0.8089133614385271,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009370729438522175,
      "loss": 0.8536,
      "step": 9402
    },
    {
      "epoch": 0.8089993977458487,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009370598105166875,
      "loss": 1.0102,
      "step": 9403
    },
    {
      "epoch": 0.8090854340531705,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009370466759028427,
      "loss": 1.1012,
      "step": 9404
    },
    {
      "epoch": 0.8091714703604921,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009370335400107212,
      "loss": 1.4193,
      "step": 9405
    },
    {
      "epoch": 0.8092575066678138,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009370204028403618,
      "loss": 2.0851,
      "step": 9406
    },
    {
      "epoch": 0.8093435429751356,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009370072643918026,
      "loss": 0.7636,
      "step": 9407
    },
    {
      "epoch": 0.8094295792824572,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009369941246650822,
      "loss": 1.4708,
      "step": 9408
    },
    {
      "epoch": 0.8095156155897789,
      "grad_norm": 2.265625,
      "learning_rate": 0.000936980983660239,
      "loss": 1.7502,
      "step": 9409
    },
    {
      "epoch": 0.8096016518971005,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009369678413773113,
      "loss": 0.7707,
      "step": 9410
    },
    {
      "epoch": 0.8096876882044223,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009369546978163376,
      "loss": 1.5415,
      "step": 9411
    },
    {
      "epoch": 0.809773724511744,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009369415529773567,
      "loss": 1.7568,
      "step": 9412
    },
    {
      "epoch": 0.8098597608190656,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009369284068604064,
      "loss": 1.3403,
      "step": 9413
    },
    {
      "epoch": 0.8099457971263874,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009369152594655256,
      "loss": 0.7881,
      "step": 9414
    },
    {
      "epoch": 0.810031833433709,
      "grad_norm": 1.5,
      "learning_rate": 0.0009369021107927529,
      "loss": 1.5036,
      "step": 9415
    },
    {
      "epoch": 0.8101178697410307,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009368889608421262,
      "loss": 1.6702,
      "step": 9416
    },
    {
      "epoch": 0.8102039060483524,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009368758096136844,
      "loss": 1.6001,
      "step": 9417
    },
    {
      "epoch": 0.8102899423556741,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009368626571074657,
      "loss": 0.866,
      "step": 9418
    },
    {
      "epoch": 0.8103759786629958,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009368495033235088,
      "loss": 1.2409,
      "step": 9419
    },
    {
      "epoch": 0.8104620149703174,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000936836348261852,
      "loss": 2.0108,
      "step": 9420
    },
    {
      "epoch": 0.8105480512776392,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009368231919225339,
      "loss": 2.037,
      "step": 9421
    },
    {
      "epoch": 0.8106340875849608,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0009368100343055927,
      "loss": 0.8582,
      "step": 9422
    },
    {
      "epoch": 0.8107201238922825,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009367968754110674,
      "loss": 1.0722,
      "step": 9423
    },
    {
      "epoch": 0.8108061601996043,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0009367837152389961,
      "loss": 0.9182,
      "step": 9424
    },
    {
      "epoch": 0.8108921965069259,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009367705537894173,
      "loss": 1.1072,
      "step": 9425
    },
    {
      "epoch": 0.8109782328142476,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009367573910623696,
      "loss": 1.1253,
      "step": 9426
    },
    {
      "epoch": 0.8110642691215693,
      "grad_norm": 2.265625,
      "learning_rate": 0.0009367442270578917,
      "loss": 1.4121,
      "step": 9427
    },
    {
      "epoch": 0.811150305428891,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009367310617760216,
      "loss": 1.628,
      "step": 9428
    },
    {
      "epoch": 0.8112363417362127,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009367178952167981,
      "loss": 1.1552,
      "step": 9429
    },
    {
      "epoch": 0.8113223780435344,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009367047273802598,
      "loss": 1.7255,
      "step": 9430
    },
    {
      "epoch": 0.8114084143508561,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009366915582664451,
      "loss": 1.054,
      "step": 9431
    },
    {
      "epoch": 0.8114944506581777,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009366783878753926,
      "loss": 1.6608,
      "step": 9432
    },
    {
      "epoch": 0.8115804869654994,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009366652162071406,
      "loss": 0.7546,
      "step": 9433
    },
    {
      "epoch": 0.8116665232728212,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009366520432617278,
      "loss": 1.1253,
      "step": 9434
    },
    {
      "epoch": 0.8117525595801428,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009366388690391927,
      "loss": 1.3092,
      "step": 9435
    },
    {
      "epoch": 0.8118385958874645,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009366256935395738,
      "loss": 1.7484,
      "step": 9436
    },
    {
      "epoch": 0.8119246321947862,
      "grad_norm": 3.59375,
      "learning_rate": 0.0009366125167629098,
      "loss": 1.5333,
      "step": 9437
    },
    {
      "epoch": 0.8120106685021079,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009365993387092389,
      "loss": 1.8994,
      "step": 9438
    },
    {
      "epoch": 0.8120967048094295,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009365861593786,
      "loss": 0.7457,
      "step": 9439
    },
    {
      "epoch": 0.8121827411167513,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009365729787710316,
      "loss": 1.2284,
      "step": 9440
    },
    {
      "epoch": 0.812268777424073,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009365597968865719,
      "loss": 0.9047,
      "step": 9441
    },
    {
      "epoch": 0.8123548137313946,
      "grad_norm": 1.625,
      "learning_rate": 0.0009365466137252599,
      "loss": 1.1031,
      "step": 9442
    },
    {
      "epoch": 0.8124408500387164,
      "grad_norm": 1.375,
      "learning_rate": 0.0009365334292871336,
      "loss": 1.5097,
      "step": 9443
    },
    {
      "epoch": 0.812526886346038,
      "grad_norm": 264.0,
      "learning_rate": 0.0009365202435722323,
      "loss": 1.624,
      "step": 9444
    },
    {
      "epoch": 0.8126129226533597,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009365070565805941,
      "loss": 1.4984,
      "step": 9445
    },
    {
      "epoch": 0.8126989589606815,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009364938683122576,
      "loss": 1.2623,
      "step": 9446
    },
    {
      "epoch": 0.8127849952680031,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009364806787672615,
      "loss": 0.8647,
      "step": 9447
    },
    {
      "epoch": 0.8128710315753248,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009364674879456442,
      "loss": 1.3223,
      "step": 9448
    },
    {
      "epoch": 0.8129570678826464,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009364542958474445,
      "loss": 2.4134,
      "step": 9449
    },
    {
      "epoch": 0.8130431041899682,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009364411024727008,
      "loss": 1.2558,
      "step": 9450
    },
    {
      "epoch": 0.8131291404972899,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009364279078214517,
      "loss": 1.0115,
      "step": 9451
    },
    {
      "epoch": 0.8132151768046115,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000936414711893736,
      "loss": 1.634,
      "step": 9452
    },
    {
      "epoch": 0.8133012131119333,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009364015146895921,
      "loss": 1.4392,
      "step": 9453
    },
    {
      "epoch": 0.8133872494192549,
      "grad_norm": 1.125,
      "learning_rate": 0.0009363883162090585,
      "loss": 1.461,
      "step": 9454
    },
    {
      "epoch": 0.8134732857265766,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009363751164521741,
      "loss": 1.7501,
      "step": 9455
    },
    {
      "epoch": 0.8135593220338984,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009363619154189773,
      "loss": 1.6022,
      "step": 9456
    },
    {
      "epoch": 0.81364535834122,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009363487131095067,
      "loss": 1.0183,
      "step": 9457
    },
    {
      "epoch": 0.8137313946485417,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009363355095238012,
      "loss": 1.336,
      "step": 9458
    },
    {
      "epoch": 0.8138174309558633,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009363223046618991,
      "loss": 1.2345,
      "step": 9459
    },
    {
      "epoch": 0.8139034672631851,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009363090985238389,
      "loss": 1.2831,
      "step": 9460
    },
    {
      "epoch": 0.8139895035705067,
      "grad_norm": 1.625,
      "learning_rate": 0.0009362958911096598,
      "loss": 1.4625,
      "step": 9461
    },
    {
      "epoch": 0.8140755398778284,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009362826824193997,
      "loss": 1.5583,
      "step": 9462
    },
    {
      "epoch": 0.8141615761851502,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009362694724530979,
      "loss": 0.8226,
      "step": 9463
    },
    {
      "epoch": 0.8142476124924718,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009362562612107927,
      "loss": 1.7744,
      "step": 9464
    },
    {
      "epoch": 0.8143336487997935,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009362430486925227,
      "loss": 1.2134,
      "step": 9465
    },
    {
      "epoch": 0.8144196851071152,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009362298348983267,
      "loss": 1.2671,
      "step": 9466
    },
    {
      "epoch": 0.8145057214144369,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009362166198282431,
      "loss": 1.3164,
      "step": 9467
    },
    {
      "epoch": 0.8145917577217586,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000936203403482311,
      "loss": 2.1636,
      "step": 9468
    },
    {
      "epoch": 0.8146777940290802,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009361901858605686,
      "loss": 1.0682,
      "step": 9469
    },
    {
      "epoch": 0.814763830336402,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009361769669630548,
      "loss": 1.1475,
      "step": 9470
    },
    {
      "epoch": 0.8148498666437236,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009361637467898081,
      "loss": 1.2074,
      "step": 9471
    },
    {
      "epoch": 0.8149359029510453,
      "grad_norm": 1.25,
      "learning_rate": 0.0009361505253408673,
      "loss": 0.9008,
      "step": 9472
    },
    {
      "epoch": 0.8150219392583671,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009361373026162712,
      "loss": 0.6889,
      "step": 9473
    },
    {
      "epoch": 0.8151079755656887,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009361240786160581,
      "loss": 1.7441,
      "step": 9474
    },
    {
      "epoch": 0.8151940118730104,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000936110853340267,
      "loss": 1.4146,
      "step": 9475
    },
    {
      "epoch": 0.8152800481803321,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009360976267889365,
      "loss": 0.9876,
      "step": 9476
    },
    {
      "epoch": 0.8153660844876538,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009360843989621053,
      "loss": 1.3145,
      "step": 9477
    },
    {
      "epoch": 0.8154521207949755,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009360711698598119,
      "loss": 0.8792,
      "step": 9478
    },
    {
      "epoch": 0.8155381571022972,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009360579394820952,
      "loss": 1.9934,
      "step": 9479
    },
    {
      "epoch": 0.8156241934096189,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009360447078289939,
      "loss": 1.3475,
      "step": 9480
    },
    {
      "epoch": 0.8157102297169405,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009360314749005466,
      "loss": 1.3125,
      "step": 9481
    },
    {
      "epoch": 0.8157962660242623,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009360182406967919,
      "loss": 1.159,
      "step": 9482
    },
    {
      "epoch": 0.8158823023315839,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009360050052177689,
      "loss": 0.9913,
      "step": 9483
    },
    {
      "epoch": 0.8159683386389056,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009359917684635159,
      "loss": 2.2973,
      "step": 9484
    },
    {
      "epoch": 0.8160543749462273,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009359785304340717,
      "loss": 0.8927,
      "step": 9485
    },
    {
      "epoch": 0.816140411253549,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009359652911294752,
      "loss": 0.9402,
      "step": 9486
    },
    {
      "epoch": 0.8162264475608707,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009359520505497651,
      "loss": 1.2348,
      "step": 9487
    },
    {
      "epoch": 0.8163124838681923,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009359388086949798,
      "loss": 1.8663,
      "step": 9488
    },
    {
      "epoch": 0.8163985201755141,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009359255655651585,
      "loss": 2.0241,
      "step": 9489
    },
    {
      "epoch": 0.8164845564828358,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009359123211603395,
      "loss": 1.1396,
      "step": 9490
    },
    {
      "epoch": 0.8165705927901574,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009358990754805619,
      "loss": 0.9432,
      "step": 9491
    },
    {
      "epoch": 0.8166566290974792,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009358858285258643,
      "loss": 1.2237,
      "step": 9492
    },
    {
      "epoch": 0.8167426654048008,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009358725802962853,
      "loss": 1.0811,
      "step": 9493
    },
    {
      "epoch": 0.8168287017121225,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009358593307918639,
      "loss": 1.4043,
      "step": 9494
    },
    {
      "epoch": 0.8169147380194443,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009358460800126387,
      "loss": 0.912,
      "step": 9495
    },
    {
      "epoch": 0.8170007743267659,
      "grad_norm": 0.84375,
      "learning_rate": 0.0009358328279586485,
      "loss": 0.7812,
      "step": 9496
    },
    {
      "epoch": 0.8170868106340876,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009358195746299319,
      "loss": 1.0167,
      "step": 9497
    },
    {
      "epoch": 0.8171728469414092,
      "grad_norm": 1.171875,
      "learning_rate": 0.000935806320026528,
      "loss": 1.3168,
      "step": 9498
    },
    {
      "epoch": 0.817258883248731,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009357930641484752,
      "loss": 1.5222,
      "step": 9499
    },
    {
      "epoch": 0.8173449195560527,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009357798069958126,
      "loss": 1.8686,
      "step": 9500
    },
    {
      "epoch": 0.8174309558633743,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009357665485685787,
      "loss": 0.9585,
      "step": 9501
    },
    {
      "epoch": 0.8175169921706961,
      "grad_norm": 1.125,
      "learning_rate": 0.0009357532888668124,
      "loss": 0.8594,
      "step": 9502
    },
    {
      "epoch": 0.8176030284780177,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009357400278905526,
      "loss": 2.0494,
      "step": 9503
    },
    {
      "epoch": 0.8176890647853394,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009357267656398378,
      "loss": 2.0423,
      "step": 9504
    },
    {
      "epoch": 0.817775101092661,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000935713502114707,
      "loss": 0.9251,
      "step": 9505
    },
    {
      "epoch": 0.8178611373999828,
      "grad_norm": 1.265625,
      "learning_rate": 0.000935700237315199,
      "loss": 1.069,
      "step": 9506
    },
    {
      "epoch": 0.8179471737073045,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009356869712413526,
      "loss": 1.3135,
      "step": 9507
    },
    {
      "epoch": 0.8180332100146261,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009356737038932063,
      "loss": 1.7729,
      "step": 9508
    },
    {
      "epoch": 0.8181192463219479,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009356604352707995,
      "loss": 0.976,
      "step": 9509
    },
    {
      "epoch": 0.8182052826292695,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009356471653741705,
      "loss": 1.9295,
      "step": 9510
    },
    {
      "epoch": 0.8182913189365912,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009356338942033582,
      "loss": 1.7642,
      "step": 9511
    },
    {
      "epoch": 0.818377355243913,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009356206217584016,
      "loss": 1.0,
      "step": 9512
    },
    {
      "epoch": 0.8184633915512346,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009356073480393392,
      "loss": 1.3751,
      "step": 9513
    },
    {
      "epoch": 0.8185494278585563,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009355940730462102,
      "loss": 1.2744,
      "step": 9514
    },
    {
      "epoch": 0.818635464165878,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009355807967790533,
      "loss": 1.0376,
      "step": 9515
    },
    {
      "epoch": 0.8187215004731997,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009355675192379072,
      "loss": 0.9265,
      "step": 9516
    },
    {
      "epoch": 0.8188075367805214,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009355542404228108,
      "loss": 1.0771,
      "step": 9517
    },
    {
      "epoch": 0.818893573087843,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009355409603338029,
      "loss": 1.0516,
      "step": 9518
    },
    {
      "epoch": 0.8189796093951648,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009355276789709226,
      "loss": 1.6543,
      "step": 9519
    },
    {
      "epoch": 0.8190656457024864,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009355143963342084,
      "loss": 1.1768,
      "step": 9520
    },
    {
      "epoch": 0.8191516820098081,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009355011124236994,
      "loss": 1.4558,
      "step": 9521
    },
    {
      "epoch": 0.8192377183171299,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009354878272394342,
      "loss": 0.7984,
      "step": 9522
    },
    {
      "epoch": 0.8193237546244515,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009354745407814519,
      "loss": 1.1701,
      "step": 9523
    },
    {
      "epoch": 0.8194097909317732,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009354612530497913,
      "loss": 1.2272,
      "step": 9524
    },
    {
      "epoch": 0.8194958272390949,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009354479640444911,
      "loss": 0.9303,
      "step": 9525
    },
    {
      "epoch": 0.8195818635464166,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009354346737655904,
      "loss": 1.3686,
      "step": 9526
    },
    {
      "epoch": 0.8196678998537382,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009354213822131279,
      "loss": 1.4358,
      "step": 9527
    },
    {
      "epoch": 0.81975393616106,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009354080893871426,
      "loss": 1.0748,
      "step": 9528
    },
    {
      "epoch": 0.8198399724683817,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009353947952876733,
      "loss": 1.1061,
      "step": 9529
    },
    {
      "epoch": 0.8199260087757033,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009353814999147589,
      "loss": 0.778,
      "step": 9530
    },
    {
      "epoch": 0.820012045083025,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009353682032684383,
      "loss": 1.6997,
      "step": 9531
    },
    {
      "epoch": 0.8200980813903467,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009353549053487502,
      "loss": 1.9037,
      "step": 9532
    },
    {
      "epoch": 0.8201841176976684,
      "grad_norm": 1.625,
      "learning_rate": 0.0009353416061557339,
      "loss": 1.0193,
      "step": 9533
    },
    {
      "epoch": 0.8202701540049901,
      "grad_norm": 1.0,
      "learning_rate": 0.0009353283056894279,
      "loss": 1.1472,
      "step": 9534
    },
    {
      "epoch": 0.8203561903123118,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009353150039498715,
      "loss": 0.9469,
      "step": 9535
    },
    {
      "epoch": 0.8204422266196335,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009353017009371032,
      "loss": 1.0266,
      "step": 9536
    },
    {
      "epoch": 0.8205282629269551,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009352883966511623,
      "loss": 1.0575,
      "step": 9537
    },
    {
      "epoch": 0.8206142992342769,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009352750910920873,
      "loss": 1.9362,
      "step": 9538
    },
    {
      "epoch": 0.8207003355415986,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009352617842599174,
      "loss": 1.3753,
      "step": 9539
    },
    {
      "epoch": 0.8207863718489202,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009352484761546914,
      "loss": 1.0745,
      "step": 9540
    },
    {
      "epoch": 0.820872408156242,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009352351667764482,
      "loss": 1.0445,
      "step": 9541
    },
    {
      "epoch": 0.8209584444635636,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009352218561252268,
      "loss": 1.5685,
      "step": 9542
    },
    {
      "epoch": 0.8210444807708853,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009352085442010661,
      "loss": 0.8371,
      "step": 9543
    },
    {
      "epoch": 0.8211305170782071,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009351952310040052,
      "loss": 1.6576,
      "step": 9544
    },
    {
      "epoch": 0.8212165533855287,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009351819165340827,
      "loss": 1.6131,
      "step": 9545
    },
    {
      "epoch": 0.8213025896928504,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009351686007913378,
      "loss": 1.5509,
      "step": 9546
    },
    {
      "epoch": 0.821388626000172,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009351552837758095,
      "loss": 1.0172,
      "step": 9547
    },
    {
      "epoch": 0.8214746623074938,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009351419654875365,
      "loss": 1.3918,
      "step": 9548
    },
    {
      "epoch": 0.8215606986148154,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009351286459265579,
      "loss": 2.4884,
      "step": 9549
    },
    {
      "epoch": 0.8216467349221371,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009351153250929128,
      "loss": 1.8358,
      "step": 9550
    },
    {
      "epoch": 0.8217327712294589,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009351020029866397,
      "loss": 1.7489,
      "step": 9551
    },
    {
      "epoch": 0.8218188075367805,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000935088679607778,
      "loss": 1.3231,
      "step": 9552
    },
    {
      "epoch": 0.8219048438441022,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009350753549563667,
      "loss": 1.2799,
      "step": 9553
    },
    {
      "epoch": 0.8219908801514239,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0009350620290324443,
      "loss": 0.7734,
      "step": 9554
    },
    {
      "epoch": 0.8220769164587456,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009350487018360503,
      "loss": 1.2767,
      "step": 9555
    },
    {
      "epoch": 0.8221629527660673,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009350353733672234,
      "loss": 1.3012,
      "step": 9556
    },
    {
      "epoch": 0.822248989073389,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009350220436260026,
      "loss": 2.193,
      "step": 9557
    },
    {
      "epoch": 0.8223350253807107,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009350087126124269,
      "loss": 1.9163,
      "step": 9558
    },
    {
      "epoch": 0.8224210616880323,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009349953803265354,
      "loss": 0.8742,
      "step": 9559
    },
    {
      "epoch": 0.822507097995354,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009349820467683671,
      "loss": 1.3421,
      "step": 9560
    },
    {
      "epoch": 0.8225931343026758,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009349687119379608,
      "loss": 0.7608,
      "step": 9561
    },
    {
      "epoch": 0.8226791706099974,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009349553758353555,
      "loss": 2.045,
      "step": 9562
    },
    {
      "epoch": 0.8227652069173191,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009349420384605905,
      "loss": 2.2878,
      "step": 9563
    },
    {
      "epoch": 0.8228512432246408,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009349286998137045,
      "loss": 1.6653,
      "step": 9564
    },
    {
      "epoch": 0.8229372795319625,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009349153598947369,
      "loss": 1.8234,
      "step": 9565
    },
    {
      "epoch": 0.8230233158392842,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009349020187037262,
      "loss": 0.7515,
      "step": 9566
    },
    {
      "epoch": 0.8231093521466059,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009348886762407118,
      "loss": 1.756,
      "step": 9567
    },
    {
      "epoch": 0.8231953884539276,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009348753325057324,
      "loss": 1.638,
      "step": 9568
    },
    {
      "epoch": 0.8232814247612492,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009348619874988275,
      "loss": 0.905,
      "step": 9569
    },
    {
      "epoch": 0.823367461068571,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009348486412200358,
      "loss": 0.9881,
      "step": 9570
    },
    {
      "epoch": 0.8234534973758926,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009348352936693963,
      "loss": 1.6011,
      "step": 9571
    },
    {
      "epoch": 0.8235395336832143,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009348219448469481,
      "loss": 1.4592,
      "step": 9572
    },
    {
      "epoch": 0.823625569990536,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009348085947527305,
      "loss": 1.767,
      "step": 9573
    },
    {
      "epoch": 0.8237116062978577,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009347952433867822,
      "loss": 1.3465,
      "step": 9574
    },
    {
      "epoch": 0.8237976426051794,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009347818907491423,
      "loss": 1.1979,
      "step": 9575
    },
    {
      "epoch": 0.823883678912501,
      "grad_norm": 1.671875,
      "learning_rate": 0.00093476853683985,
      "loss": 1.1953,
      "step": 9576
    },
    {
      "epoch": 0.8239697152198228,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009347551816589444,
      "loss": 1.1733,
      "step": 9577
    },
    {
      "epoch": 0.8240557515271445,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009347418252064643,
      "loss": 1.1393,
      "step": 9578
    },
    {
      "epoch": 0.8241417878344661,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009347284674824489,
      "loss": 1.3652,
      "step": 9579
    },
    {
      "epoch": 0.8242278241417879,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009347151084869373,
      "loss": 1.3063,
      "step": 9580
    },
    {
      "epoch": 0.8243138604491095,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009347017482199685,
      "loss": 1.3283,
      "step": 9581
    },
    {
      "epoch": 0.8243998967564312,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009346883866815819,
      "loss": 1.6577,
      "step": 9582
    },
    {
      "epoch": 0.824485933063753,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000934675023871816,
      "loss": 1.5918,
      "step": 9583
    },
    {
      "epoch": 0.8245719693710746,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009346616597907103,
      "loss": 2.091,
      "step": 9584
    },
    {
      "epoch": 0.8246580056783963,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009346482944383039,
      "loss": 1.9712,
      "step": 9585
    },
    {
      "epoch": 0.8247440419857179,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009346349278146356,
      "loss": 1.1645,
      "step": 9586
    },
    {
      "epoch": 0.8248300782930397,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009346215599197447,
      "loss": 2.0869,
      "step": 9587
    },
    {
      "epoch": 0.8249161146003614,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009346081907536703,
      "loss": 2.0913,
      "step": 9588
    },
    {
      "epoch": 0.825002150907683,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009345948203164516,
      "loss": 0.9316,
      "step": 9589
    },
    {
      "epoch": 0.8250881872150048,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009345814486081274,
      "loss": 1.7779,
      "step": 9590
    },
    {
      "epoch": 0.8251742235223264,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009345680756287371,
      "loss": 1.509,
      "step": 9591
    },
    {
      "epoch": 0.8252602598296481,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009345547013783196,
      "loss": 1.124,
      "step": 9592
    },
    {
      "epoch": 0.8253462961369697,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009345413258569143,
      "loss": 1.3774,
      "step": 9593
    },
    {
      "epoch": 0.8254323324442915,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009345279490645599,
      "loss": 1.3058,
      "step": 9594
    },
    {
      "epoch": 0.8255183687516132,
      "grad_norm": 1.375,
      "learning_rate": 0.0009345145710012957,
      "loss": 1.561,
      "step": 9595
    },
    {
      "epoch": 0.8256044050589348,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009345011916671611,
      "loss": 1.2196,
      "step": 9596
    },
    {
      "epoch": 0.8256904413662566,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009344878110621951,
      "loss": 1.1,
      "step": 9597
    },
    {
      "epoch": 0.8257764776735782,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009344744291864364,
      "loss": 0.8514,
      "step": 9598
    },
    {
      "epoch": 0.8258625139808999,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009344610460399247,
      "loss": 1.8198,
      "step": 9599
    },
    {
      "epoch": 0.8259485502882217,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009344476616226992,
      "loss": 1.668,
      "step": 9600
    },
    {
      "epoch": 0.8260345865955433,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009344342759347984,
      "loss": 1.3205,
      "step": 9601
    },
    {
      "epoch": 0.826120622902865,
      "grad_norm": 1.046875,
      "learning_rate": 0.000934420888976262,
      "loss": 1.6372,
      "step": 9602
    },
    {
      "epoch": 0.8262066592101867,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009344075007471289,
      "loss": 1.5533,
      "step": 9603
    },
    {
      "epoch": 0.8262926955175084,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009343941112474383,
      "loss": 1.3037,
      "step": 9604
    },
    {
      "epoch": 0.8263787318248301,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009343807204772297,
      "loss": 0.6623,
      "step": 9605
    },
    {
      "epoch": 0.8264647681321518,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009343673284365417,
      "loss": 1.5187,
      "step": 9606
    },
    {
      "epoch": 0.8265508044394735,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009343539351254138,
      "loss": 1.764,
      "step": 9607
    },
    {
      "epoch": 0.8266368407467951,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009343405405438851,
      "loss": 0.9556,
      "step": 9608
    },
    {
      "epoch": 0.8267228770541168,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000934327144691995,
      "loss": 1.2294,
      "step": 9609
    },
    {
      "epoch": 0.8268089133614386,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009343137475697823,
      "loss": 1.0854,
      "step": 9610
    },
    {
      "epoch": 0.8268949496687602,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009343003491772864,
      "loss": 1.5866,
      "step": 9611
    },
    {
      "epoch": 0.8269809859760819,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009342869495145463,
      "loss": 1.0681,
      "step": 9612
    },
    {
      "epoch": 0.8270670222834036,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009342735485816014,
      "loss": 1.6877,
      "step": 9613
    },
    {
      "epoch": 0.8271530585907253,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009342601463784909,
      "loss": 1.6412,
      "step": 9614
    },
    {
      "epoch": 0.8272390948980469,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009342467429052539,
      "loss": 1.3319,
      "step": 9615
    },
    {
      "epoch": 0.8273251312053687,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009342333381619297,
      "loss": 1.3337,
      "step": 9616
    },
    {
      "epoch": 0.8274111675126904,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009342199321485574,
      "loss": 0.8778,
      "step": 9617
    },
    {
      "epoch": 0.827497203820012,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009342065248651763,
      "loss": 1.8092,
      "step": 9618
    },
    {
      "epoch": 0.8275832401273338,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009341931163118255,
      "loss": 1.2255,
      "step": 9619
    },
    {
      "epoch": 0.8276692764346554,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009341797064885442,
      "loss": 0.8998,
      "step": 9620
    },
    {
      "epoch": 0.8277553127419771,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009341662953953717,
      "loss": 1.3833,
      "step": 9621
    },
    {
      "epoch": 0.8278413490492988,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009341528830323475,
      "loss": 1.208,
      "step": 9622
    },
    {
      "epoch": 0.8279273853566205,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009341394693995103,
      "loss": 1.3929,
      "step": 9623
    },
    {
      "epoch": 0.8280134216639422,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009341260544968997,
      "loss": 1.7866,
      "step": 9624
    },
    {
      "epoch": 0.8280994579712638,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009341126383245549,
      "loss": 1.3805,
      "step": 9625
    },
    {
      "epoch": 0.8281854942785856,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009340992208825147,
      "loss": 1.8644,
      "step": 9626
    },
    {
      "epoch": 0.8282715305859073,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009340858021708191,
      "loss": 1.3164,
      "step": 9627
    },
    {
      "epoch": 0.8283575668932289,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009340723821895067,
      "loss": 1.3627,
      "step": 9628
    },
    {
      "epoch": 0.8284436032005507,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009340589609386172,
      "loss": 0.9308,
      "step": 9629
    },
    {
      "epoch": 0.8285296395078723,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009340455384181895,
      "loss": 1.885,
      "step": 9630
    },
    {
      "epoch": 0.828615675815194,
      "grad_norm": 1.09375,
      "learning_rate": 0.000934032114628263,
      "loss": 1.6398,
      "step": 9631
    },
    {
      "epoch": 0.8287017121225158,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009340186895688772,
      "loss": 1.4844,
      "step": 9632
    },
    {
      "epoch": 0.8287877484298374,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009340052632400709,
      "loss": 0.8824,
      "step": 9633
    },
    {
      "epoch": 0.8288737847371591,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009339918356418838,
      "loss": 1.0879,
      "step": 9634
    },
    {
      "epoch": 0.8289598210444807,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009339784067743549,
      "loss": 1.1318,
      "step": 9635
    },
    {
      "epoch": 0.8290458573518025,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009339649766375236,
      "loss": 2.2587,
      "step": 9636
    },
    {
      "epoch": 0.8291318936591241,
      "grad_norm": 1.0,
      "learning_rate": 0.0009339515452314292,
      "loss": 0.816,
      "step": 9637
    },
    {
      "epoch": 0.8292179299664458,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009339381125561109,
      "loss": 1.1099,
      "step": 9638
    },
    {
      "epoch": 0.8293039662737676,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009339246786116078,
      "loss": 1.7716,
      "step": 9639
    },
    {
      "epoch": 0.8293900025810892,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009339112433979596,
      "loss": 1.7286,
      "step": 9640
    },
    {
      "epoch": 0.8294760388884109,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009338978069152053,
      "loss": 1.1747,
      "step": 9641
    },
    {
      "epoch": 0.8295620751957326,
      "grad_norm": 75.0,
      "learning_rate": 0.0009338843691633845,
      "loss": 1.3267,
      "step": 9642
    },
    {
      "epoch": 0.8296481115030543,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009338709301425361,
      "loss": 1.3306,
      "step": 9643
    },
    {
      "epoch": 0.829734147810376,
      "grad_norm": 1.375,
      "learning_rate": 0.0009338574898526997,
      "loss": 1.4653,
      "step": 9644
    },
    {
      "epoch": 0.8298201841176976,
      "grad_norm": 2.4375,
      "learning_rate": 0.0009338440482939145,
      "loss": 1.1544,
      "step": 9645
    },
    {
      "epoch": 0.8299062204250194,
      "grad_norm": 1.125,
      "learning_rate": 0.00093383060546622,
      "loss": 1.6032,
      "step": 9646
    },
    {
      "epoch": 0.829992256732341,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009338171613696551,
      "loss": 0.7371,
      "step": 9647
    },
    {
      "epoch": 0.8300782930396627,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009338037160042595,
      "loss": 1.323,
      "step": 9648
    },
    {
      "epoch": 0.8301643293469845,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009337902693700724,
      "loss": 1.5323,
      "step": 9649
    },
    {
      "epoch": 0.8302503656543061,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009337768214671331,
      "loss": 1.5881,
      "step": 9650
    },
    {
      "epoch": 0.8303364019616278,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000933763372295481,
      "loss": 1.2305,
      "step": 9651
    },
    {
      "epoch": 0.8304224382689495,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009337499218551554,
      "loss": 1.7677,
      "step": 9652
    },
    {
      "epoch": 0.8305084745762712,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009337364701461956,
      "loss": 1.4283,
      "step": 9653
    },
    {
      "epoch": 0.8305945108835929,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009337230171686411,
      "loss": 1.4121,
      "step": 9654
    },
    {
      "epoch": 0.8306805471909146,
      "grad_norm": 1.3125,
      "learning_rate": 0.000933709562922531,
      "loss": 1.6848,
      "step": 9655
    },
    {
      "epoch": 0.8307665834982363,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009336961074079048,
      "loss": 1.3378,
      "step": 9656
    },
    {
      "epoch": 0.8308526198055579,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009336826506248019,
      "loss": 0.9516,
      "step": 9657
    },
    {
      "epoch": 0.8309386561128796,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009336691925732614,
      "loss": 1.5833,
      "step": 9658
    },
    {
      "epoch": 0.8310246924202013,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009336557332533231,
      "loss": 0.975,
      "step": 9659
    },
    {
      "epoch": 0.831110728727523,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000933642272665026,
      "loss": 1.2861,
      "step": 9660
    },
    {
      "epoch": 0.8311967650348447,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009336288108084096,
      "loss": 1.1813,
      "step": 9661
    },
    {
      "epoch": 0.8312828013421664,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009336153476835132,
      "loss": 1.7769,
      "step": 9662
    },
    {
      "epoch": 0.8313688376494881,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009336018832903763,
      "loss": 1.6564,
      "step": 9663
    },
    {
      "epoch": 0.8314548739568097,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009335884176290383,
      "loss": 0.9495,
      "step": 9664
    },
    {
      "epoch": 0.8315409102641315,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009335749506995385,
      "loss": 1.1512,
      "step": 9665
    },
    {
      "epoch": 0.8316269465714532,
      "grad_norm": 1.125,
      "learning_rate": 0.0009335614825019162,
      "loss": 1.304,
      "step": 9666
    },
    {
      "epoch": 0.8317129828787748,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009335480130362109,
      "loss": 2.072,
      "step": 9667
    },
    {
      "epoch": 0.8317990191860966,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009335345423024619,
      "loss": 1.0767,
      "step": 9668
    },
    {
      "epoch": 0.8318850554934182,
      "grad_norm": 1.625,
      "learning_rate": 0.0009335210703007089,
      "loss": 1.0539,
      "step": 9669
    },
    {
      "epoch": 0.8319710918007399,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009335075970309909,
      "loss": 0.9339,
      "step": 9670
    },
    {
      "epoch": 0.8320571281080616,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009334941224933476,
      "loss": 1.0506,
      "step": 9671
    },
    {
      "epoch": 0.8321431644153833,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009334806466878183,
      "loss": 0.8218,
      "step": 9672
    },
    {
      "epoch": 0.832229200722705,
      "grad_norm": 1.5,
      "learning_rate": 0.0009334671696144423,
      "loss": 1.1338,
      "step": 9673
    },
    {
      "epoch": 0.8323152370300266,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009334536912732593,
      "loss": 1.2138,
      "step": 9674
    },
    {
      "epoch": 0.8324012733373484,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009334402116643084,
      "loss": 1.4516,
      "step": 9675
    },
    {
      "epoch": 0.8324873096446701,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009334267307876293,
      "loss": 0.8415,
      "step": 9676
    },
    {
      "epoch": 0.8325733459519917,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009334132486432613,
      "loss": 1.113,
      "step": 9677
    },
    {
      "epoch": 0.8326593822593135,
      "grad_norm": 2.875,
      "learning_rate": 0.0009333997652312438,
      "loss": 1.8352,
      "step": 9678
    },
    {
      "epoch": 0.8327454185666351,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009333862805516163,
      "loss": 1.0866,
      "step": 9679
    },
    {
      "epoch": 0.8328314548739568,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009333727946044182,
      "loss": 1.8706,
      "step": 9680
    },
    {
      "epoch": 0.8329174911812784,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009333593073896889,
      "loss": 1.0999,
      "step": 9681
    },
    {
      "epoch": 0.8330035274886002,
      "grad_norm": 1.375,
      "learning_rate": 0.0009333458189074679,
      "loss": 1.237,
      "step": 9682
    },
    {
      "epoch": 0.8330895637959219,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009333323291577949,
      "loss": 1.3567,
      "step": 9683
    },
    {
      "epoch": 0.8331756001032435,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009333188381407088,
      "loss": 0.8872,
      "step": 9684
    },
    {
      "epoch": 0.8332616364105653,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009333053458562496,
      "loss": 1.0437,
      "step": 9685
    },
    {
      "epoch": 0.8333476727178869,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009332918523044564,
      "loss": 1.8571,
      "step": 9686
    },
    {
      "epoch": 0.8334337090252086,
      "grad_norm": 1.234375,
      "learning_rate": 0.000933278357485369,
      "loss": 1.3732,
      "step": 9687
    },
    {
      "epoch": 0.8335197453325304,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009332648613990265,
      "loss": 0.9675,
      "step": 9688
    },
    {
      "epoch": 0.833605781639852,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009332513640454685,
      "loss": 1.4053,
      "step": 9689
    },
    {
      "epoch": 0.8336918179471737,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009332378654247347,
      "loss": 1.4249,
      "step": 9690
    },
    {
      "epoch": 0.8337778542544954,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009332243655368642,
      "loss": 1.8334,
      "step": 9691
    },
    {
      "epoch": 0.8338638905618171,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009332108643818969,
      "loss": 1.6694,
      "step": 9692
    },
    {
      "epoch": 0.8339499268691388,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009331973619598719,
      "loss": 1.4022,
      "step": 9693
    },
    {
      "epoch": 0.8340359631764604,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000933183858270829,
      "loss": 1.2126,
      "step": 9694
    },
    {
      "epoch": 0.8341219994837822,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009331703533148074,
      "loss": 1.6818,
      "step": 9695
    },
    {
      "epoch": 0.8342080357911038,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009331568470918468,
      "loss": 1.1195,
      "step": 9696
    },
    {
      "epoch": 0.8342940720984255,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009331433396019868,
      "loss": 1.1734,
      "step": 9697
    },
    {
      "epoch": 0.8343801084057473,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009331298308452667,
      "loss": 0.9284,
      "step": 9698
    },
    {
      "epoch": 0.8344661447130689,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009331163208217262,
      "loss": 1.4155,
      "step": 9699
    },
    {
      "epoch": 0.8345521810203906,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009331028095314045,
      "loss": 2.0284,
      "step": 9700
    },
    {
      "epoch": 0.8346382173277123,
      "grad_norm": 1.5,
      "learning_rate": 0.0009330892969743414,
      "loss": 2.3628,
      "step": 9701
    },
    {
      "epoch": 0.834724253635034,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009330757831505763,
      "loss": 0.7471,
      "step": 9702
    },
    {
      "epoch": 0.8348102899423556,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009330622680601489,
      "loss": 1.8629,
      "step": 9703
    },
    {
      "epoch": 0.8348963262496774,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009330487517030983,
      "loss": 1.1358,
      "step": 9704
    },
    {
      "epoch": 0.8349823625569991,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009330352340794647,
      "loss": 1.1891,
      "step": 9705
    },
    {
      "epoch": 0.8350683988643207,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000933021715189287,
      "loss": 1.6428,
      "step": 9706
    },
    {
      "epoch": 0.8351544351716425,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009330081950326051,
      "loss": 1.2302,
      "step": 9707
    },
    {
      "epoch": 0.8352404714789641,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009329946736094584,
      "loss": 1.7913,
      "step": 9708
    },
    {
      "epoch": 0.8353265077862858,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009329811509198866,
      "loss": 1.0715,
      "step": 9709
    },
    {
      "epoch": 0.8354125440936075,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009329676269639291,
      "loss": 0.9374,
      "step": 9710
    },
    {
      "epoch": 0.8354985804009292,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009329541017416254,
      "loss": 1.4492,
      "step": 9711
    },
    {
      "epoch": 0.8355846167082509,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009329405752530151,
      "loss": 1.5248,
      "step": 9712
    },
    {
      "epoch": 0.8356706530155725,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009329270474981381,
      "loss": 1.664,
      "step": 9713
    },
    {
      "epoch": 0.8357566893228943,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009329135184770334,
      "loss": 1.6212,
      "step": 9714
    },
    {
      "epoch": 0.835842725630216,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009328999881897411,
      "loss": 1.6634,
      "step": 9715
    },
    {
      "epoch": 0.8359287619375376,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009328864566363003,
      "loss": 2.0536,
      "step": 9716
    },
    {
      "epoch": 0.8360147982448594,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000932872923816751,
      "loss": 2.1488,
      "step": 9717
    },
    {
      "epoch": 0.836100834552181,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0009328593897311325,
      "loss": 0.9957,
      "step": 9718
    },
    {
      "epoch": 0.8361868708595027,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009328458543794844,
      "loss": 1.0889,
      "step": 9719
    },
    {
      "epoch": 0.8362729071668245,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009328323177618465,
      "loss": 1.2988,
      "step": 9720
    },
    {
      "epoch": 0.8363589434741461,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009328187798782582,
      "loss": 0.8473,
      "step": 9721
    },
    {
      "epoch": 0.8364449797814678,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009328052407287591,
      "loss": 1.259,
      "step": 9722
    },
    {
      "epoch": 0.8365310160887894,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009327917003133889,
      "loss": 0.8505,
      "step": 9723
    },
    {
      "epoch": 0.8366170523961112,
      "grad_norm": 0.9375,
      "learning_rate": 0.000932778158632187,
      "loss": 0.7094,
      "step": 9724
    },
    {
      "epoch": 0.8367030887034328,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009327646156851935,
      "loss": 1.1284,
      "step": 9725
    },
    {
      "epoch": 0.8367891250107545,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009327510714724474,
      "loss": 0.868,
      "step": 9726
    },
    {
      "epoch": 0.8368751613180763,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009327375259939885,
      "loss": 0.9083,
      "step": 9727
    },
    {
      "epoch": 0.8369611976253979,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009327239792498568,
      "loss": 1.4571,
      "step": 9728
    },
    {
      "epoch": 0.8370472339327196,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009327104312400914,
      "loss": 1.4643,
      "step": 9729
    },
    {
      "epoch": 0.8371332702400412,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009326968819647322,
      "loss": 2.2456,
      "step": 9730
    },
    {
      "epoch": 0.837219306547363,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009326833314238187,
      "loss": 0.9415,
      "step": 9731
    },
    {
      "epoch": 0.8373053428546847,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009326697796173907,
      "loss": 1.1079,
      "step": 9732
    },
    {
      "epoch": 0.8373913791620063,
      "grad_norm": 1.75,
      "learning_rate": 0.0009326562265454878,
      "loss": 1.9452,
      "step": 9733
    },
    {
      "epoch": 0.8374774154693281,
      "grad_norm": 1.25,
      "learning_rate": 0.0009326426722081495,
      "loss": 1.5727,
      "step": 9734
    },
    {
      "epoch": 0.8375634517766497,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009326291166054155,
      "loss": 0.803,
      "step": 9735
    },
    {
      "epoch": 0.8376494880839714,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009326155597373256,
      "loss": 1.3608,
      "step": 9736
    },
    {
      "epoch": 0.8377355243912932,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009326020016039192,
      "loss": 1.8045,
      "step": 9737
    },
    {
      "epoch": 0.8378215606986148,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009325884422052361,
      "loss": 1.0352,
      "step": 9738
    },
    {
      "epoch": 0.8379075970059365,
      "grad_norm": 1.390625,
      "learning_rate": 0.000932574881541316,
      "loss": 1.7836,
      "step": 9739
    },
    {
      "epoch": 0.8379936333132582,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009325613196121985,
      "loss": 1.2162,
      "step": 9740
    },
    {
      "epoch": 0.8380796696205799,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009325477564179232,
      "loss": 1.5548,
      "step": 9741
    },
    {
      "epoch": 0.8381657059279016,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009325341919585299,
      "loss": 1.6485,
      "step": 9742
    },
    {
      "epoch": 0.8382517422352233,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009325206262340581,
      "loss": 1.5712,
      "step": 9743
    },
    {
      "epoch": 0.838337778542545,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009325070592445477,
      "loss": 1.3394,
      "step": 9744
    },
    {
      "epoch": 0.8384238148498666,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009324934909900382,
      "loss": 1.7998,
      "step": 9745
    },
    {
      "epoch": 0.8385098511571883,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009324799214705695,
      "loss": 1.4699,
      "step": 9746
    },
    {
      "epoch": 0.83859588746451,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000932466350686181,
      "loss": 1.5729,
      "step": 9747
    },
    {
      "epoch": 0.8386819237718317,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009324527786369126,
      "loss": 1.7291,
      "step": 9748
    },
    {
      "epoch": 0.8387679600791534,
      "grad_norm": 1.375,
      "learning_rate": 0.0009324392053228039,
      "loss": 1.2018,
      "step": 9749
    },
    {
      "epoch": 0.8388539963864751,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009324256307438947,
      "loss": 1.6038,
      "step": 9750
    },
    {
      "epoch": 0.8389400326937968,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009324120549002244,
      "loss": 1.611,
      "step": 9751
    },
    {
      "epoch": 0.8390260690011184,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009323984777918332,
      "loss": 1.5743,
      "step": 9752
    },
    {
      "epoch": 0.8391121053084402,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009323848994187605,
      "loss": 1.6802,
      "step": 9753
    },
    {
      "epoch": 0.8391981416157619,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009323713197810459,
      "loss": 1.3533,
      "step": 9754
    },
    {
      "epoch": 0.8392841779230835,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009323577388787294,
      "loss": 1.8475,
      "step": 9755
    },
    {
      "epoch": 0.8393702142304053,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009323441567118507,
      "loss": 1.3818,
      "step": 9756
    },
    {
      "epoch": 0.8394562505377269,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009323305732804492,
      "loss": 0.7975,
      "step": 9757
    },
    {
      "epoch": 0.8395422868450486,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000932316988584565,
      "loss": 1.5194,
      "step": 9758
    },
    {
      "epoch": 0.8396283231523703,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009323034026242378,
      "loss": 1.6182,
      "step": 9759
    },
    {
      "epoch": 0.839714359459692,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009322898153995071,
      "loss": 1.8563,
      "step": 9760
    },
    {
      "epoch": 0.8398003957670137,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009322762269104128,
      "loss": 1.3586,
      "step": 9761
    },
    {
      "epoch": 0.8398864320743353,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009322626371569945,
      "loss": 1.5272,
      "step": 9762
    },
    {
      "epoch": 0.8399724683816571,
      "grad_norm": 1.1875,
      "learning_rate": 0.000932249046139292,
      "loss": 1.2269,
      "step": 9763
    },
    {
      "epoch": 0.8400585046889788,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009322354538573454,
      "loss": 0.8562,
      "step": 9764
    },
    {
      "epoch": 0.8401445409963004,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000932221860311194,
      "loss": 0.841,
      "step": 9765
    },
    {
      "epoch": 0.8402305773036222,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009322082655008777,
      "loss": 0.7875,
      "step": 9766
    },
    {
      "epoch": 0.8403166136109438,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009321946694264363,
      "loss": 1.3997,
      "step": 9767
    },
    {
      "epoch": 0.8404026499182655,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009321810720879096,
      "loss": 2.2229,
      "step": 9768
    },
    {
      "epoch": 0.8404886862255871,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009321674734853372,
      "loss": 1.2341,
      "step": 9769
    },
    {
      "epoch": 0.8405747225329089,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009321538736187592,
      "loss": 1.6895,
      "step": 9770
    },
    {
      "epoch": 0.8406607588402306,
      "grad_norm": 1.515625,
      "learning_rate": 0.000932140272488215,
      "loss": 1.3035,
      "step": 9771
    },
    {
      "epoch": 0.8407467951475522,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009321266700937445,
      "loss": 1.6589,
      "step": 9772
    },
    {
      "epoch": 0.840832831454874,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009321130664353876,
      "loss": 1.207,
      "step": 9773
    },
    {
      "epoch": 0.8409188677621956,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009320994615131841,
      "loss": 1.4221,
      "step": 9774
    },
    {
      "epoch": 0.8410049040695173,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009320858553271736,
      "loss": 1.1529,
      "step": 9775
    },
    {
      "epoch": 0.8410909403768391,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009320722478773962,
      "loss": 1.0932,
      "step": 9776
    },
    {
      "epoch": 0.8411769766841607,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009320586391638912,
      "loss": 0.8642,
      "step": 9777
    },
    {
      "epoch": 0.8412630129914824,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000932045029186699,
      "loss": 1.4939,
      "step": 9778
    },
    {
      "epoch": 0.841349049298804,
      "grad_norm": 1.609375,
      "learning_rate": 0.000932031417945859,
      "loss": 1.4419,
      "step": 9779
    },
    {
      "epoch": 0.8414350856061258,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009320178054414111,
      "loss": 1.983,
      "step": 9780
    },
    {
      "epoch": 0.8415211219134475,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009320041916733952,
      "loss": 1.5714,
      "step": 9781
    },
    {
      "epoch": 0.8416071582207691,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009319905766418511,
      "loss": 1.1932,
      "step": 9782
    },
    {
      "epoch": 0.8416931945280909,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009319769603468184,
      "loss": 1.6422,
      "step": 9783
    },
    {
      "epoch": 0.8417792308354125,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009319633427883373,
      "loss": 0.9371,
      "step": 9784
    },
    {
      "epoch": 0.8418652671427342,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009319497239664474,
      "loss": 1.4704,
      "step": 9785
    },
    {
      "epoch": 0.841951303450056,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009319361038811884,
      "loss": 1.3533,
      "step": 9786
    },
    {
      "epoch": 0.8420373397573776,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009319224825326005,
      "loss": 1.279,
      "step": 9787
    },
    {
      "epoch": 0.8421233760646993,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009319088599207232,
      "loss": 1.1107,
      "step": 9788
    },
    {
      "epoch": 0.842209412372021,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009318952360455966,
      "loss": 0.9322,
      "step": 9789
    },
    {
      "epoch": 0.8422954486793427,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009318816109072603,
      "loss": 1.6057,
      "step": 9790
    },
    {
      "epoch": 0.8423814849866643,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009318679845057545,
      "loss": 1.175,
      "step": 9791
    },
    {
      "epoch": 0.842467521293986,
      "grad_norm": 0.796875,
      "learning_rate": 0.0009318543568411186,
      "loss": 1.649,
      "step": 9792
    },
    {
      "epoch": 0.8425535576013078,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009318407279133929,
      "loss": 1.2001,
      "step": 9793
    },
    {
      "epoch": 0.8426395939086294,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009318270977226169,
      "loss": 1.0795,
      "step": 9794
    },
    {
      "epoch": 0.8427256302159511,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009318134662688307,
      "loss": 1.4259,
      "step": 9795
    },
    {
      "epoch": 0.8428116665232728,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009317998335520741,
      "loss": 1.9512,
      "step": 9796
    },
    {
      "epoch": 0.8428977028305945,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009317861995723869,
      "loss": 1.9819,
      "step": 9797
    },
    {
      "epoch": 0.8429837391379162,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009317725643298092,
      "loss": 0.9481,
      "step": 9798
    },
    {
      "epoch": 0.8430697754452379,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009317589278243806,
      "loss": 0.874,
      "step": 9799
    },
    {
      "epoch": 0.8431558117525596,
      "grad_norm": 1.25,
      "learning_rate": 0.0009317452900561412,
      "loss": 1.169,
      "step": 9800
    },
    {
      "epoch": 0.8432418480598812,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009317316510251306,
      "loss": 1.3141,
      "step": 9801
    },
    {
      "epoch": 0.843327884367203,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009317180107313891,
      "loss": 1.4776,
      "step": 9802
    },
    {
      "epoch": 0.8434139206745247,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009317043691749562,
      "loss": 1.1664,
      "step": 9803
    },
    {
      "epoch": 0.8434999569818463,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009316907263558721,
      "loss": 1.401,
      "step": 9804
    },
    {
      "epoch": 0.8435859932891681,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009316770822741766,
      "loss": 0.6482,
      "step": 9805
    },
    {
      "epoch": 0.8436720295964897,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009316634369299095,
      "loss": 0.9656,
      "step": 9806
    },
    {
      "epoch": 0.8437580659038114,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009316497903231109,
      "loss": 1.3322,
      "step": 9807
    },
    {
      "epoch": 0.8438441022111332,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009316361424538207,
      "loss": 1.9042,
      "step": 9808
    },
    {
      "epoch": 0.8439301385184548,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009316224933220785,
      "loss": 0.9707,
      "step": 9809
    },
    {
      "epoch": 0.8440161748257765,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009316088429279246,
      "loss": 1.4757,
      "step": 9810
    },
    {
      "epoch": 0.8441022111330981,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009315951912713987,
      "loss": 1.3995,
      "step": 9811
    },
    {
      "epoch": 0.8441882474404199,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009315815383525408,
      "loss": 1.4133,
      "step": 9812
    },
    {
      "epoch": 0.8442742837477415,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009315678841713908,
      "loss": 1.3721,
      "step": 9813
    },
    {
      "epoch": 0.8443603200550632,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009315542287279888,
      "loss": 1.5845,
      "step": 9814
    },
    {
      "epoch": 0.844446356362385,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009315405720223746,
      "loss": 1.1871,
      "step": 9815
    },
    {
      "epoch": 0.8445323926697066,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009315269140545881,
      "loss": 1.2892,
      "step": 9816
    },
    {
      "epoch": 0.8446184289770283,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009315132548246694,
      "loss": 1.7976,
      "step": 9817
    },
    {
      "epoch": 0.84470446528435,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009314995943326583,
      "loss": 1.0041,
      "step": 9818
    },
    {
      "epoch": 0.8447905015916717,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009314859325785947,
      "loss": 1.2989,
      "step": 9819
    },
    {
      "epoch": 0.8448765378989934,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009314722695625188,
      "loss": 1.2319,
      "step": 9820
    },
    {
      "epoch": 0.844962574206315,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009314586052844705,
      "loss": 1.4292,
      "step": 9821
    },
    {
      "epoch": 0.8450486105136368,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009314449397444895,
      "loss": 1.1064,
      "step": 9822
    },
    {
      "epoch": 0.8451346468209584,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000931431272942616,
      "loss": 1.4726,
      "step": 9823
    },
    {
      "epoch": 0.8452206831282801,
      "grad_norm": 1.65625,
      "learning_rate": 0.00093141760487889,
      "loss": 1.573,
      "step": 9824
    },
    {
      "epoch": 0.8453067194356019,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009314039355533514,
      "loss": 1.9465,
      "step": 9825
    },
    {
      "epoch": 0.8453927557429235,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009313902649660401,
      "loss": 1.2959,
      "step": 9826
    },
    {
      "epoch": 0.8454787920502452,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009313765931169963,
      "loss": 0.9847,
      "step": 9827
    },
    {
      "epoch": 0.8455648283575669,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009313629200062599,
      "loss": 1.7368,
      "step": 9828
    },
    {
      "epoch": 0.8456508646648886,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009313492456338707,
      "loss": 1.6541,
      "step": 9829
    },
    {
      "epoch": 0.8457369009722103,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009313355699998689,
      "loss": 1.6065,
      "step": 9830
    },
    {
      "epoch": 0.845822937279532,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009313218931042944,
      "loss": 0.9308,
      "step": 9831
    },
    {
      "epoch": 0.8459089735868537,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009313082149471874,
      "loss": 1.6035,
      "step": 9832
    },
    {
      "epoch": 0.8459950098941753,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009312945355285877,
      "loss": 0.9514,
      "step": 9833
    },
    {
      "epoch": 0.846081046201497,
      "grad_norm": 1.5,
      "learning_rate": 0.0009312808548485353,
      "loss": 1.9644,
      "step": 9834
    },
    {
      "epoch": 0.8461670825088187,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009312671729070703,
      "loss": 1.1449,
      "step": 9835
    },
    {
      "epoch": 0.8462531188161404,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009312534897042327,
      "loss": 1.5477,
      "step": 9836
    },
    {
      "epoch": 0.8463391551234621,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009312398052400626,
      "loss": 2.592,
      "step": 9837
    },
    {
      "epoch": 0.8464251914307838,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009312261195145999,
      "loss": 1.3273,
      "step": 9838
    },
    {
      "epoch": 0.8465112277381055,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009312124325278845,
      "loss": 1.3426,
      "step": 9839
    },
    {
      "epoch": 0.8465972640454271,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009311987442799568,
      "loss": 1.1727,
      "step": 9840
    },
    {
      "epoch": 0.8466833003527489,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009311850547708566,
      "loss": 1.9245,
      "step": 9841
    },
    {
      "epoch": 0.8467693366600706,
      "grad_norm": 1.5,
      "learning_rate": 0.000931171364000624,
      "loss": 1.7848,
      "step": 9842
    },
    {
      "epoch": 0.8468553729673922,
      "grad_norm": 1.203125,
      "learning_rate": 0.000931157671969299,
      "loss": 1.4855,
      "step": 9843
    },
    {
      "epoch": 0.846941409274714,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009311439786769216,
      "loss": 1.0729,
      "step": 9844
    },
    {
      "epoch": 0.8470274455820356,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000931130284123532,
      "loss": 1.4159,
      "step": 9845
    },
    {
      "epoch": 0.8471134818893573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009311165883091701,
      "loss": 1.3791,
      "step": 9846
    },
    {
      "epoch": 0.847199518196679,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000931102891233876,
      "loss": 1.1139,
      "step": 9847
    },
    {
      "epoch": 0.8472855545040007,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009310891928976899,
      "loss": 1.303,
      "step": 9848
    },
    {
      "epoch": 0.8473715908113224,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009310754933006517,
      "loss": 0.8174,
      "step": 9849
    },
    {
      "epoch": 0.847457627118644,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009310617924428015,
      "loss": 1.728,
      "step": 9850
    },
    {
      "epoch": 0.8475436634259658,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009310480903241794,
      "loss": 1.2641,
      "step": 9851
    },
    {
      "epoch": 0.8476296997332875,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009310343869448254,
      "loss": 1.1129,
      "step": 9852
    },
    {
      "epoch": 0.8477157360406091,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009310206823047798,
      "loss": 1.7143,
      "step": 9853
    },
    {
      "epoch": 0.8478017723479309,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009310069764040825,
      "loss": 0.8462,
      "step": 9854
    },
    {
      "epoch": 0.8478878086552525,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009309932692427737,
      "loss": 0.885,
      "step": 9855
    },
    {
      "epoch": 0.8479738449625742,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000930979560820893,
      "loss": 0.8198,
      "step": 9856
    },
    {
      "epoch": 0.8480598812698958,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009309658511384813,
      "loss": 1.1507,
      "step": 9857
    },
    {
      "epoch": 0.8481459175772176,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009309521401955783,
      "loss": 1.5466,
      "step": 9858
    },
    {
      "epoch": 0.8482319538845393,
      "grad_norm": 1.140625,
      "learning_rate": 0.000930938427992224,
      "loss": 1.1456,
      "step": 9859
    },
    {
      "epoch": 0.8483179901918609,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009309247145284585,
      "loss": 0.9505,
      "step": 9860
    },
    {
      "epoch": 0.8484040264991827,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009309109998043222,
      "loss": 1.4336,
      "step": 9861
    },
    {
      "epoch": 0.8484900628065043,
      "grad_norm": 1.328125,
      "learning_rate": 0.000930897283819855,
      "loss": 1.3644,
      "step": 9862
    },
    {
      "epoch": 0.848576099113826,
      "grad_norm": 1.015625,
      "learning_rate": 0.000930883566575097,
      "loss": 1.1206,
      "step": 9863
    },
    {
      "epoch": 0.8486621354211478,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009308698480700883,
      "loss": 0.9039,
      "step": 9864
    },
    {
      "epoch": 0.8487481717284694,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009308561283048693,
      "loss": 1.0576,
      "step": 9865
    },
    {
      "epoch": 0.8488342080357911,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009308424072794798,
      "loss": 2.437,
      "step": 9866
    },
    {
      "epoch": 0.8489202443431128,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00093082868499396,
      "loss": 1.6396,
      "step": 9867
    },
    {
      "epoch": 0.8490062806504345,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009308149614483502,
      "loss": 1.6452,
      "step": 9868
    },
    {
      "epoch": 0.8490923169577562,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009308012366426903,
      "loss": 1.98,
      "step": 9869
    },
    {
      "epoch": 0.8491783532650778,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009307875105770207,
      "loss": 1.814,
      "step": 9870
    },
    {
      "epoch": 0.8492643895723996,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009307737832513814,
      "loss": 0.5751,
      "step": 9871
    },
    {
      "epoch": 0.8493504258797212,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009307600546658124,
      "loss": 1.6229,
      "step": 9872
    },
    {
      "epoch": 0.8494364621870429,
      "grad_norm": 1.5,
      "learning_rate": 0.0009307463248203543,
      "loss": 1.1539,
      "step": 9873
    },
    {
      "epoch": 0.8495224984943647,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009307325937150467,
      "loss": 1.6107,
      "step": 9874
    },
    {
      "epoch": 0.8496085348016863,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009307188613499303,
      "loss": 1.8588,
      "step": 9875
    },
    {
      "epoch": 0.849694571109008,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009307051277250447,
      "loss": 1.0529,
      "step": 9876
    },
    {
      "epoch": 0.8497806074163297,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009306913928404306,
      "loss": 1.0182,
      "step": 9877
    },
    {
      "epoch": 0.8498666437236514,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009306776566961278,
      "loss": 2.641,
      "step": 9878
    },
    {
      "epoch": 0.849952680030973,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009306639192921766,
      "loss": 0.9617,
      "step": 9879
    },
    {
      "epoch": 0.8500387163382948,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009306501806286174,
      "loss": 0.6473,
      "step": 9880
    },
    {
      "epoch": 0.8501247526456165,
      "grad_norm": 1.640625,
      "learning_rate": 0.00093063644070549,
      "loss": 1.3357,
      "step": 9881
    },
    {
      "epoch": 0.8502107889529381,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009306226995228348,
      "loss": 1.4111,
      "step": 9882
    },
    {
      "epoch": 0.8502968252602598,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000930608957080692,
      "loss": 1.8711,
      "step": 9883
    },
    {
      "epoch": 0.8503828615675815,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009305952133791018,
      "loss": 1.7825,
      "step": 9884
    },
    {
      "epoch": 0.8504688978749032,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009305814684181042,
      "loss": 1.5926,
      "step": 9885
    },
    {
      "epoch": 0.8505549341822249,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009305677221977396,
      "loss": 1.0689,
      "step": 9886
    },
    {
      "epoch": 0.8506409704895466,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009305539747180482,
      "loss": 1.0395,
      "step": 9887
    },
    {
      "epoch": 0.8507270067968683,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009305402259790701,
      "loss": 0.7475,
      "step": 9888
    },
    {
      "epoch": 0.8508130431041899,
      "grad_norm": 1.25,
      "learning_rate": 0.0009305264759808457,
      "loss": 1.4894,
      "step": 9889
    },
    {
      "epoch": 0.8508990794115117,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009305127247234149,
      "loss": 1.1169,
      "step": 9890
    },
    {
      "epoch": 0.8509851157188334,
      "grad_norm": 1.25,
      "learning_rate": 0.0009304989722068182,
      "loss": 1.4419,
      "step": 9891
    },
    {
      "epoch": 0.851071152026155,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009304852184310959,
      "loss": 0.885,
      "step": 9892
    },
    {
      "epoch": 0.8511571883334768,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009304714633962878,
      "loss": 1.7914,
      "step": 9893
    },
    {
      "epoch": 0.8512432246407984,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0009304577071024345,
      "loss": 0.7195,
      "step": 9894
    },
    {
      "epoch": 0.8513292609481201,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009304439495495761,
      "loss": 1.551,
      "step": 9895
    },
    {
      "epoch": 0.8514152972554418,
      "grad_norm": 1.484375,
      "learning_rate": 0.000930430190737753,
      "loss": 1.5084,
      "step": 9896
    },
    {
      "epoch": 0.8515013335627635,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009304164306670051,
      "loss": 1.0886,
      "step": 9897
    },
    {
      "epoch": 0.8515873698700852,
      "grad_norm": 1.34375,
      "learning_rate": 0.000930402669337373,
      "loss": 1.8093,
      "step": 9898
    },
    {
      "epoch": 0.8516734061774068,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009303889067488967,
      "loss": 1.5681,
      "step": 9899
    },
    {
      "epoch": 0.8517594424847286,
      "grad_norm": 2.0,
      "learning_rate": 0.0009303751429016166,
      "loss": 1.6806,
      "step": 9900
    },
    {
      "epoch": 0.8518454787920502,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009303613777955728,
      "loss": 0.9898,
      "step": 9901
    },
    {
      "epoch": 0.8519315150993719,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009303476114308058,
      "loss": 1.8111,
      "step": 9902
    },
    {
      "epoch": 0.8520175514066937,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009303338438073559,
      "loss": 1.6878,
      "step": 9903
    },
    {
      "epoch": 0.8521035877140153,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009303200749252629,
      "loss": 1.7028,
      "step": 9904
    },
    {
      "epoch": 0.852189624021337,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009303063047845675,
      "loss": 1.4837,
      "step": 9905
    },
    {
      "epoch": 0.8522756603286586,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009302925333853099,
      "loss": 1.5773,
      "step": 9906
    },
    {
      "epoch": 0.8523616966359804,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009302787607275302,
      "loss": 1.9535,
      "step": 9907
    },
    {
      "epoch": 0.8524477329433021,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000930264986811269,
      "loss": 1.6244,
      "step": 9908
    },
    {
      "epoch": 0.8525337692506237,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009302512116365662,
      "loss": 1.3992,
      "step": 9909
    },
    {
      "epoch": 0.8526198055579455,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009302374352034623,
      "loss": 1.23,
      "step": 9910
    },
    {
      "epoch": 0.8527058418652671,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009302236575119976,
      "loss": 1.2371,
      "step": 9911
    },
    {
      "epoch": 0.8527918781725888,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009302098785622124,
      "loss": 1.5012,
      "step": 9912
    },
    {
      "epoch": 0.8528779144799106,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009301960983541471,
      "loss": 0.6852,
      "step": 9913
    },
    {
      "epoch": 0.8529639507872322,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009301823168878417,
      "loss": 1.1167,
      "step": 9914
    },
    {
      "epoch": 0.8530499870945539,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009301685341633368,
      "loss": 1.3206,
      "step": 9915
    },
    {
      "epoch": 0.8531360234018756,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009301547501806726,
      "loss": 1.4543,
      "step": 9916
    },
    {
      "epoch": 0.8532220597091973,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009301409649398893,
      "loss": 1.1111,
      "step": 9917
    },
    {
      "epoch": 0.853308096016519,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009301271784410275,
      "loss": 1.2489,
      "step": 9918
    },
    {
      "epoch": 0.8533941323238406,
      "grad_norm": 1.125,
      "learning_rate": 0.0009301133906841272,
      "loss": 1.2277,
      "step": 9919
    },
    {
      "epoch": 0.8534801686311624,
      "grad_norm": 1.25,
      "learning_rate": 0.000930099601669229,
      "loss": 0.8665,
      "step": 9920
    },
    {
      "epoch": 0.853566204938484,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000930085811396373,
      "loss": 1.5448,
      "step": 9921
    },
    {
      "epoch": 0.8536522412458057,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009300720198655997,
      "loss": 1.2177,
      "step": 9922
    },
    {
      "epoch": 0.8537382775531274,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009300582270769495,
      "loss": 0.9346,
      "step": 9923
    },
    {
      "epoch": 0.8538243138604491,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009300444330304625,
      "loss": 1.5175,
      "step": 9924
    },
    {
      "epoch": 0.8539103501677708,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009300306377261791,
      "loss": 0.813,
      "step": 9925
    },
    {
      "epoch": 0.8539963864750925,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009300168411641398,
      "loss": 1.3686,
      "step": 9926
    },
    {
      "epoch": 0.8540824227824142,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009300030433443849,
      "loss": 1.1102,
      "step": 9927
    },
    {
      "epoch": 0.8541684590897358,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009299892442669549,
      "loss": 1.3895,
      "step": 9928
    },
    {
      "epoch": 0.8542544953970576,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009299754439318897,
      "loss": 0.8546,
      "step": 9929
    },
    {
      "epoch": 0.8543405317043793,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009299616423392301,
      "loss": 1.1419,
      "step": 9930
    },
    {
      "epoch": 0.8544265680117009,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009299478394890161,
      "loss": 1.6484,
      "step": 9931
    },
    {
      "epoch": 0.8545126043190227,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009299340353812886,
      "loss": 2.4642,
      "step": 9932
    },
    {
      "epoch": 0.8545986406263443,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009299202300160877,
      "loss": 0.7044,
      "step": 9933
    },
    {
      "epoch": 0.854684676933666,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009299064233934534,
      "loss": 1.5621,
      "step": 9934
    },
    {
      "epoch": 0.8547707132409877,
      "grad_norm": 1.125,
      "learning_rate": 0.0009298926155134267,
      "loss": 1.1685,
      "step": 9935
    },
    {
      "epoch": 0.8548567495483094,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009298788063760475,
      "loss": 0.9234,
      "step": 9936
    },
    {
      "epoch": 0.8549427858556311,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009298649959813565,
      "loss": 1.816,
      "step": 9937
    },
    {
      "epoch": 0.8550288221629527,
      "grad_norm": 1.5,
      "learning_rate": 0.000929851184329394,
      "loss": 1.2131,
      "step": 9938
    },
    {
      "epoch": 0.8551148584702745,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0009298373714202005,
      "loss": 0.7511,
      "step": 9939
    },
    {
      "epoch": 0.8552008947775962,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009298235572538161,
      "loss": 0.9595,
      "step": 9940
    },
    {
      "epoch": 0.8552869310849178,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009298097418302815,
      "loss": 0.6854,
      "step": 9941
    },
    {
      "epoch": 0.8553729673922396,
      "grad_norm": 1.15625,
      "learning_rate": 0.000929795925149637,
      "loss": 1.0006,
      "step": 9942
    },
    {
      "epoch": 0.8554590036995612,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009297821072119229,
      "loss": 1.1821,
      "step": 9943
    },
    {
      "epoch": 0.8555450400068829,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009297682880171799,
      "loss": 0.7663,
      "step": 9944
    },
    {
      "epoch": 0.8556310763142045,
      "grad_norm": 1.0,
      "learning_rate": 0.0009297544675654481,
      "loss": 1.4233,
      "step": 9945
    },
    {
      "epoch": 0.8557171126215263,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009297406458567681,
      "loss": 1.0183,
      "step": 9946
    },
    {
      "epoch": 0.855803148928848,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009297268228911803,
      "loss": 1.2156,
      "step": 9947
    },
    {
      "epoch": 0.8558891852361696,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009297129986687252,
      "loss": 1.1056,
      "step": 9948
    },
    {
      "epoch": 0.8559752215434914,
      "grad_norm": 1.0625,
      "learning_rate": 0.000929699173189443,
      "loss": 1.0404,
      "step": 9949
    },
    {
      "epoch": 0.856061257850813,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009296853464533743,
      "loss": 1.0136,
      "step": 9950
    },
    {
      "epoch": 0.8561472941581347,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009296715184605596,
      "loss": 1.3929,
      "step": 9951
    },
    {
      "epoch": 0.8562333304654565,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009296576892110393,
      "loss": 1.2795,
      "step": 9952
    },
    {
      "epoch": 0.8563193667727781,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009296438587048538,
      "loss": 1.3,
      "step": 9953
    },
    {
      "epoch": 0.8564054030800998,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009296300269420436,
      "loss": 1.1231,
      "step": 9954
    },
    {
      "epoch": 0.8564914393874214,
      "grad_norm": 1.125,
      "learning_rate": 0.0009296161939226491,
      "loss": 0.9383,
      "step": 9955
    },
    {
      "epoch": 0.8565774756947432,
      "grad_norm": 1.375,
      "learning_rate": 0.0009296023596467107,
      "loss": 0.9827,
      "step": 9956
    },
    {
      "epoch": 0.8566635120020649,
      "grad_norm": 0.875,
      "learning_rate": 0.0009295885241142691,
      "loss": 0.5971,
      "step": 9957
    },
    {
      "epoch": 0.8567495483093865,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009295746873253646,
      "loss": 1.4917,
      "step": 9958
    },
    {
      "epoch": 0.8568355846167083,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009295608492800376,
      "loss": 1.3183,
      "step": 9959
    },
    {
      "epoch": 0.8569216209240299,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009295470099783287,
      "loss": 0.8504,
      "step": 9960
    },
    {
      "epoch": 0.8570076572313516,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009295331694202782,
      "loss": 1.3545,
      "step": 9961
    },
    {
      "epoch": 0.8570936935386734,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009295193276059269,
      "loss": 0.9612,
      "step": 9962
    },
    {
      "epoch": 0.857179729845995,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009295054845353151,
      "loss": 1.2941,
      "step": 9963
    },
    {
      "epoch": 0.8572657661533167,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009294916402084832,
      "loss": 2.0058,
      "step": 9964
    },
    {
      "epoch": 0.8573518024606384,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000929477794625472,
      "loss": 1.4099,
      "step": 9965
    },
    {
      "epoch": 0.8574378387679601,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009294639477863216,
      "loss": 1.317,
      "step": 9966
    },
    {
      "epoch": 0.8575238750752817,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009294500996910726,
      "loss": 1.5616,
      "step": 9967
    },
    {
      "epoch": 0.8576099113826035,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009294362503397658,
      "loss": 1.4529,
      "step": 9968
    },
    {
      "epoch": 0.8576959476899252,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009294223997324413,
      "loss": 1.0801,
      "step": 9969
    },
    {
      "epoch": 0.8577819839972468,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009294085478691399,
      "loss": 1.2623,
      "step": 9970
    },
    {
      "epoch": 0.8578680203045685,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009293946947499019,
      "loss": 1.828,
      "step": 9971
    },
    {
      "epoch": 0.8579540566118902,
      "grad_norm": 1.359375,
      "learning_rate": 0.000929380840374768,
      "loss": 1.3503,
      "step": 9972
    },
    {
      "epoch": 0.8580400929192119,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009293669847437786,
      "loss": 1.1991,
      "step": 9973
    },
    {
      "epoch": 0.8581261292265336,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009293531278569744,
      "loss": 1.1566,
      "step": 9974
    },
    {
      "epoch": 0.8582121655338553,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009293392697143957,
      "loss": 1.46,
      "step": 9975
    },
    {
      "epoch": 0.858298201841177,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009293254103160834,
      "loss": 1.3567,
      "step": 9976
    },
    {
      "epoch": 0.8583842381484986,
      "grad_norm": 1.25,
      "learning_rate": 0.0009293115496620774,
      "loss": 1.5234,
      "step": 9977
    },
    {
      "epoch": 0.8584702744558204,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009292976877524188,
      "loss": 1.9548,
      "step": 9978
    },
    {
      "epoch": 0.8585563107631421,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009292838245871478,
      "loss": 0.5672,
      "step": 9979
    },
    {
      "epoch": 0.8586423470704637,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009292699601663052,
      "loss": 1.771,
      "step": 9980
    },
    {
      "epoch": 0.8587283833777855,
      "grad_norm": 3.25,
      "learning_rate": 0.0009292560944899316,
      "loss": 2.2343,
      "step": 9981
    },
    {
      "epoch": 0.8588144196851071,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009292422275580673,
      "loss": 1.5815,
      "step": 9982
    },
    {
      "epoch": 0.8589004559924288,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009292283593707529,
      "loss": 1.2931,
      "step": 9983
    },
    {
      "epoch": 0.8589864922997505,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009292144899280292,
      "loss": 1.0604,
      "step": 9984
    },
    {
      "epoch": 0.8590725286070722,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009292006192299365,
      "loss": 1.5432,
      "step": 9985
    },
    {
      "epoch": 0.8591585649143939,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009291867472765154,
      "loss": 1.6245,
      "step": 9986
    },
    {
      "epoch": 0.8592446012217155,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009291728740678064,
      "loss": 1.302,
      "step": 9987
    },
    {
      "epoch": 0.8593306375290373,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009291589996038506,
      "loss": 0.8235,
      "step": 9988
    },
    {
      "epoch": 0.8594166738363589,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009291451238846879,
      "loss": 1.3652,
      "step": 9989
    },
    {
      "epoch": 0.8595027101436806,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009291312469103593,
      "loss": 1.2497,
      "step": 9990
    },
    {
      "epoch": 0.8595887464510024,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009291173686809053,
      "loss": 1.1557,
      "step": 9991
    },
    {
      "epoch": 0.859674782758324,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009291034891963664,
      "loss": 1.0885,
      "step": 9992
    },
    {
      "epoch": 0.8597608190656457,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009290896084567831,
      "loss": 1.2016,
      "step": 9993
    },
    {
      "epoch": 0.8598468553729673,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009290757264621963,
      "loss": 1.2082,
      "step": 9994
    },
    {
      "epoch": 0.8599328916802891,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009290618432126465,
      "loss": 0.9617,
      "step": 9995
    },
    {
      "epoch": 0.8600189279876108,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009290479587081742,
      "loss": 1.0035,
      "step": 9996
    },
    {
      "epoch": 0.8601049642949324,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009290340729488201,
      "loss": 0.9556,
      "step": 9997
    },
    {
      "epoch": 0.8601910006022542,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009290201859346246,
      "loss": 1.4149,
      "step": 9998
    },
    {
      "epoch": 0.8602770369095758,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009290062976656287,
      "loss": 1.4883,
      "step": 9999
    },
    {
      "epoch": 0.8603630732168975,
      "grad_norm": 4.90625,
      "learning_rate": 0.0009289924081418728,
      "loss": 2.3887,
      "step": 10000
    },
    {
      "epoch": 0.8604491095242193,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009289785173633974,
      "loss": 0.9788,
      "step": 10001
    },
    {
      "epoch": 0.8605351458315409,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009289646253302434,
      "loss": 1.871,
      "step": 10002
    },
    {
      "epoch": 0.8606211821388626,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009289507320424513,
      "loss": 1.8241,
      "step": 10003
    },
    {
      "epoch": 0.8607072184461843,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009289368375000616,
      "loss": 1.7432,
      "step": 10004
    },
    {
      "epoch": 0.860793254753506,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009289229417031152,
      "loss": 1.497,
      "step": 10005
    },
    {
      "epoch": 0.8608792910608277,
      "grad_norm": 2.5,
      "learning_rate": 0.0009289090446516526,
      "loss": 2.0123,
      "step": 10006
    },
    {
      "epoch": 0.8609653273681493,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009288951463457145,
      "loss": 0.9572,
      "step": 10007
    },
    {
      "epoch": 0.8610513636754711,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009288812467853414,
      "loss": 1.3449,
      "step": 10008
    },
    {
      "epoch": 0.8611373999827927,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009288673459705741,
      "loss": 1.1389,
      "step": 10009
    },
    {
      "epoch": 0.8612234362901144,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009288534439014533,
      "loss": 0.8631,
      "step": 10010
    },
    {
      "epoch": 0.8613094725974361,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009288395405780194,
      "loss": 1.1509,
      "step": 10011
    },
    {
      "epoch": 0.8613955089047578,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009288256360003136,
      "loss": 1.191,
      "step": 10012
    },
    {
      "epoch": 0.8614815452120795,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009288117301683758,
      "loss": 1.3983,
      "step": 10013
    },
    {
      "epoch": 0.8615675815194012,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009287978230822474,
      "loss": 1.7204,
      "step": 10014
    },
    {
      "epoch": 0.8616536178267229,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009287839147419686,
      "loss": 0.784,
      "step": 10015
    },
    {
      "epoch": 0.8617396541340445,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009287700051475803,
      "loss": 1.8959,
      "step": 10016
    },
    {
      "epoch": 0.8618256904413663,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009287560942991231,
      "loss": 1.8048,
      "step": 10017
    },
    {
      "epoch": 0.861911726748688,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009287421821966377,
      "loss": 1.5203,
      "step": 10018
    },
    {
      "epoch": 0.8619977630560096,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009287282688401648,
      "loss": 1.874,
      "step": 10019
    },
    {
      "epoch": 0.8620837993633313,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009287143542297451,
      "loss": 1.9093,
      "step": 10020
    },
    {
      "epoch": 0.862169835670653,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009287004383654193,
      "loss": 0.8266,
      "step": 10021
    },
    {
      "epoch": 0.8622558719779747,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009286865212472282,
      "loss": 1.4548,
      "step": 10022
    },
    {
      "epoch": 0.8623419082852964,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009286726028752124,
      "loss": 2.1378,
      "step": 10023
    },
    {
      "epoch": 0.8624279445926181,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009286586832494126,
      "loss": 1.5673,
      "step": 10024
    },
    {
      "epoch": 0.8625139808999398,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009286447623698693,
      "loss": 1.4226,
      "step": 10025
    },
    {
      "epoch": 0.8626000172072614,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009286308402366237,
      "loss": 1.6209,
      "step": 10026
    },
    {
      "epoch": 0.8626860535145832,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009286169168497161,
      "loss": 2.0238,
      "step": 10027
    },
    {
      "epoch": 0.8627720898219049,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009286029922091875,
      "loss": 0.8284,
      "step": 10028
    },
    {
      "epoch": 0.8628581261292265,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009285890663150784,
      "loss": 0.81,
      "step": 10029
    },
    {
      "epoch": 0.8629441624365483,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009285751391674296,
      "loss": 1.2055,
      "step": 10030
    },
    {
      "epoch": 0.8630301987438699,
      "grad_norm": 1.234375,
      "learning_rate": 0.000928561210766282,
      "loss": 1.8183,
      "step": 10031
    },
    {
      "epoch": 0.8631162350511916,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009285472811116762,
      "loss": 1.0485,
      "step": 10032
    },
    {
      "epoch": 0.8632022713585132,
      "grad_norm": 2.171875,
      "learning_rate": 0.0009285333502036528,
      "loss": 1.0326,
      "step": 10033
    },
    {
      "epoch": 0.863288307665835,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009285194180422528,
      "loss": 1.3879,
      "step": 10034
    },
    {
      "epoch": 0.8633743439731567,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009285054846275169,
      "loss": 1.4643,
      "step": 10035
    },
    {
      "epoch": 0.8634603802804783,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0009284915499594857,
      "loss": 0.5864,
      "step": 10036
    },
    {
      "epoch": 0.8635464165878001,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009284776140382001,
      "loss": 1.3136,
      "step": 10037
    },
    {
      "epoch": 0.8636324528951217,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009284636768637008,
      "loss": 0.907,
      "step": 10038
    },
    {
      "epoch": 0.8637184892024434,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009284497384360285,
      "loss": 2.4007,
      "step": 10039
    },
    {
      "epoch": 0.8638045255097652,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009284357987552241,
      "loss": 0.8904,
      "step": 10040
    },
    {
      "epoch": 0.8638905618170868,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009284218578213284,
      "loss": 1.4298,
      "step": 10041
    },
    {
      "epoch": 0.8639765981244085,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009284079156343821,
      "loss": 1.6994,
      "step": 10042
    },
    {
      "epoch": 0.8640626344317301,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009283939721944259,
      "loss": 1.3479,
      "step": 10043
    },
    {
      "epoch": 0.8641486707390519,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009283800275015007,
      "loss": 1.016,
      "step": 10044
    },
    {
      "epoch": 0.8642347070463736,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000928366081555647,
      "loss": 1.4694,
      "step": 10045
    },
    {
      "epoch": 0.8643207433536952,
      "grad_norm": 1.609375,
      "learning_rate": 0.000928352134356906,
      "loss": 1.2683,
      "step": 10046
    },
    {
      "epoch": 0.864406779661017,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009283381859053184,
      "loss": 2.1738,
      "step": 10047
    },
    {
      "epoch": 0.8644928159683386,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009283242362009247,
      "loss": 1.4101,
      "step": 10048
    },
    {
      "epoch": 0.8645788522756603,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009283102852437661,
      "loss": 1.6724,
      "step": 10049
    },
    {
      "epoch": 0.8646648885829821,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009282963330338831,
      "loss": 0.9663,
      "step": 10050
    },
    {
      "epoch": 0.8647509248903037,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009282823795713169,
      "loss": 1.1766,
      "step": 10051
    },
    {
      "epoch": 0.8648369611976254,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009282684248561078,
      "loss": 0.954,
      "step": 10052
    },
    {
      "epoch": 0.864922997504947,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009282544688882968,
      "loss": 1.6181,
      "step": 10053
    },
    {
      "epoch": 0.8650090338122688,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009282405116679249,
      "loss": 0.8402,
      "step": 10054
    },
    {
      "epoch": 0.8650950701195904,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009282265531950327,
      "loss": 1.9009,
      "step": 10055
    },
    {
      "epoch": 0.8651811064269121,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009282125934696611,
      "loss": 2.2655,
      "step": 10056
    },
    {
      "epoch": 0.8652671427342339,
      "grad_norm": 1.015625,
      "learning_rate": 0.000928198632491851,
      "loss": 1.1572,
      "step": 10057
    },
    {
      "epoch": 0.8653531790415555,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009281846702616433,
      "loss": 1.6764,
      "step": 10058
    },
    {
      "epoch": 0.8654392153488772,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009281707067790786,
      "loss": 1.4747,
      "step": 10059
    },
    {
      "epoch": 0.8655252516561989,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009281567420441978,
      "loss": 1.1425,
      "step": 10060
    },
    {
      "epoch": 0.8656112879635206,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009281427760570418,
      "loss": 0.908,
      "step": 10061
    },
    {
      "epoch": 0.8656973242708423,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009281288088176516,
      "loss": 1.8615,
      "step": 10062
    },
    {
      "epoch": 0.865783360578164,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009281148403260677,
      "loss": 1.5775,
      "step": 10063
    },
    {
      "epoch": 0.8658693968854857,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009281008705823313,
      "loss": 1.2792,
      "step": 10064
    },
    {
      "epoch": 0.8659554331928073,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009280868995864831,
      "loss": 1.6308,
      "step": 10065
    },
    {
      "epoch": 0.8660414695001291,
      "grad_norm": 1.375,
      "learning_rate": 0.000928072927338564,
      "loss": 0.9731,
      "step": 10066
    },
    {
      "epoch": 0.8661275058074508,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009280589538386147,
      "loss": 1.4289,
      "step": 10067
    },
    {
      "epoch": 0.8662135421147724,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009280449790866763,
      "loss": 1.1176,
      "step": 10068
    },
    {
      "epoch": 0.8662995784220942,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009280310030827896,
      "loss": 1.734,
      "step": 10069
    },
    {
      "epoch": 0.8663856147294158,
      "grad_norm": 1.0,
      "learning_rate": 0.0009280170258269955,
      "loss": 0.721,
      "step": 10070
    },
    {
      "epoch": 0.8664716510367375,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009280030473193348,
      "loss": 0.913,
      "step": 10071
    },
    {
      "epoch": 0.8665576873440592,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009279890675598484,
      "loss": 1.247,
      "step": 10072
    },
    {
      "epoch": 0.8666437236513809,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009279750865485772,
      "loss": 1.1174,
      "step": 10073
    },
    {
      "epoch": 0.8667297599587026,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009279611042855622,
      "loss": 1.1583,
      "step": 10074
    },
    {
      "epoch": 0.8668157962660242,
      "grad_norm": 1.125,
      "learning_rate": 0.0009279471207708442,
      "loss": 1.0011,
      "step": 10075
    },
    {
      "epoch": 0.866901832573346,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000927933136004464,
      "loss": 1.1968,
      "step": 10076
    },
    {
      "epoch": 0.8669878688806676,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009279191499864627,
      "loss": 1.1587,
      "step": 10077
    },
    {
      "epoch": 0.8670739051879893,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009279051627168811,
      "loss": 1.6675,
      "step": 10078
    },
    {
      "epoch": 0.8671599414953111,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00092789117419576,
      "loss": 1.5245,
      "step": 10079
    },
    {
      "epoch": 0.8672459778026327,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009278771844231406,
      "loss": 1.931,
      "step": 10080
    },
    {
      "epoch": 0.8673320141099544,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009278631933990636,
      "loss": 1.0374,
      "step": 10081
    },
    {
      "epoch": 0.867418050417276,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00092784920112357,
      "loss": 1.3877,
      "step": 10082
    },
    {
      "epoch": 0.8675040867245978,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009278352075967007,
      "loss": 1.2062,
      "step": 10083
    },
    {
      "epoch": 0.8675901230319195,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009278212128184967,
      "loss": 1.0818,
      "step": 10084
    },
    {
      "epoch": 0.8676761593392411,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009278072167889986,
      "loss": 2.1635,
      "step": 10085
    },
    {
      "epoch": 0.8677621956465629,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009277932195082477,
      "loss": 1.0913,
      "step": 10086
    },
    {
      "epoch": 0.8678482319538845,
      "grad_norm": 1.203125,
      "learning_rate": 0.000927779220976285,
      "loss": 1.7733,
      "step": 10087
    },
    {
      "epoch": 0.8679342682612062,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009277652211931511,
      "loss": 1.2718,
      "step": 10088
    },
    {
      "epoch": 0.868020304568528,
      "grad_norm": 1.125,
      "learning_rate": 0.0009277512201588871,
      "loss": 1.4206,
      "step": 10089
    },
    {
      "epoch": 0.8681063408758496,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009277372178735342,
      "loss": 1.1131,
      "step": 10090
    },
    {
      "epoch": 0.8681923771831713,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009277232143371329,
      "loss": 1.6788,
      "step": 10091
    },
    {
      "epoch": 0.868278413490493,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009277092095497244,
      "loss": 1.566,
      "step": 10092
    },
    {
      "epoch": 0.8683644497978147,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009276952035113498,
      "loss": 1.9691,
      "step": 10093
    },
    {
      "epoch": 0.8684504861051364,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009276811962220499,
      "loss": 1.3955,
      "step": 10094
    },
    {
      "epoch": 0.868536522412458,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009276671876818656,
      "loss": 1.6364,
      "step": 10095
    },
    {
      "epoch": 0.8686225587197798,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009276531778908379,
      "loss": 1.6178,
      "step": 10096
    },
    {
      "epoch": 0.8687085950271014,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009276391668490079,
      "loss": 1.5824,
      "step": 10097
    },
    {
      "epoch": 0.8687946313344231,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009276251545564164,
      "loss": 0.8248,
      "step": 10098
    },
    {
      "epoch": 0.8688806676417448,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009276111410131047,
      "loss": 1.3866,
      "step": 10099
    },
    {
      "epoch": 0.8689667039490665,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009275971262191135,
      "loss": 2.1577,
      "step": 10100
    },
    {
      "epoch": 0.8690527402563882,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009275831101744838,
      "loss": 1.8618,
      "step": 10101
    },
    {
      "epoch": 0.8691387765637099,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009275690928792566,
      "loss": 1.2388,
      "step": 10102
    },
    {
      "epoch": 0.8692248128710316,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009275550743334731,
      "loss": 1.0435,
      "step": 10103
    },
    {
      "epoch": 0.8693108491783532,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009275410545371741,
      "loss": 1.6955,
      "step": 10104
    },
    {
      "epoch": 0.869396885485675,
      "grad_norm": 2.328125,
      "learning_rate": 0.0009275270334904007,
      "loss": 1.6813,
      "step": 10105
    },
    {
      "epoch": 0.8694829217929967,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000927513011193194,
      "loss": 1.0215,
      "step": 10106
    },
    {
      "epoch": 0.8695689581003183,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009274989876455946,
      "loss": 1.539,
      "step": 10107
    },
    {
      "epoch": 0.86965499440764,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009274849628476441,
      "loss": 1.142,
      "step": 10108
    },
    {
      "epoch": 0.8697410307149617,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000927470936799383,
      "loss": 2.0459,
      "step": 10109
    },
    {
      "epoch": 0.8698270670222834,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009274569095008526,
      "loss": 1.4096,
      "step": 10110
    },
    {
      "epoch": 0.8699131033296051,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009274428809520939,
      "loss": 0.9311,
      "step": 10111
    },
    {
      "epoch": 0.8699991396369268,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009274288511531479,
      "loss": 1.257,
      "step": 10112
    },
    {
      "epoch": 0.8700851759442485,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009274148201040557,
      "loss": 1.97,
      "step": 10113
    },
    {
      "epoch": 0.8701712122515701,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009274007878048582,
      "loss": 0.8069,
      "step": 10114
    },
    {
      "epoch": 0.8702572485588919,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009273867542555966,
      "loss": 1.6476,
      "step": 10115
    },
    {
      "epoch": 0.8703432848662136,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009273727194563119,
      "loss": 0.7881,
      "step": 10116
    },
    {
      "epoch": 0.8704293211735352,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009273586834070449,
      "loss": 1.6587,
      "step": 10117
    },
    {
      "epoch": 0.870515357480857,
      "grad_norm": 1.140625,
      "learning_rate": 0.000927344646107837,
      "loss": 0.9676,
      "step": 10118
    },
    {
      "epoch": 0.8706013937881786,
      "grad_norm": 4.78125,
      "learning_rate": 0.0009273306075587291,
      "loss": 1.518,
      "step": 10119
    },
    {
      "epoch": 0.8706874300955003,
      "grad_norm": 1.125,
      "learning_rate": 0.0009273165677597624,
      "loss": 1.2284,
      "step": 10120
    },
    {
      "epoch": 0.8707734664028219,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009273025267109779,
      "loss": 1.6034,
      "step": 10121
    },
    {
      "epoch": 0.8708595027101437,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009272884844124163,
      "loss": 1.2793,
      "step": 10122
    },
    {
      "epoch": 0.8709455390174654,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009272744408641191,
      "loss": 1.8751,
      "step": 10123
    },
    {
      "epoch": 0.871031575324787,
      "grad_norm": 1.375,
      "learning_rate": 0.0009272603960661274,
      "loss": 1.1083,
      "step": 10124
    },
    {
      "epoch": 0.8711176116321088,
      "grad_norm": 1.65625,
      "learning_rate": 0.000927246350018482,
      "loss": 1.865,
      "step": 10125
    },
    {
      "epoch": 0.8712036479394304,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009272323027212242,
      "loss": 0.953,
      "step": 10126
    },
    {
      "epoch": 0.8712896842467521,
      "grad_norm": 0.96875,
      "learning_rate": 0.000927218254174395,
      "loss": 0.7789,
      "step": 10127
    },
    {
      "epoch": 0.8713757205540739,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009272042043780356,
      "loss": 1.3844,
      "step": 10128
    },
    {
      "epoch": 0.8714617568613955,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009271901533321868,
      "loss": 0.9143,
      "step": 10129
    },
    {
      "epoch": 0.8715477931687172,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009271761010368899,
      "loss": 0.8458,
      "step": 10130
    },
    {
      "epoch": 0.8716338294760388,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000927162047492186,
      "loss": 1.539,
      "step": 10131
    },
    {
      "epoch": 0.8717198657833606,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009271479926981163,
      "loss": 0.6181,
      "step": 10132
    },
    {
      "epoch": 0.8718059020906823,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009271339366547218,
      "loss": 2.1364,
      "step": 10133
    },
    {
      "epoch": 0.8718919383980039,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009271198793620435,
      "loss": 1.361,
      "step": 10134
    },
    {
      "epoch": 0.8719779747053257,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009271058208201226,
      "loss": 0.9207,
      "step": 10135
    },
    {
      "epoch": 0.8720640110126473,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009270917610290003,
      "loss": 1.7072,
      "step": 10136
    },
    {
      "epoch": 0.872150047319969,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009270776999887177,
      "loss": 1.4272,
      "step": 10137
    },
    {
      "epoch": 0.8722360836272908,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009270636376993159,
      "loss": 1.0515,
      "step": 10138
    },
    {
      "epoch": 0.8723221199346124,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009270495741608361,
      "loss": 1.4198,
      "step": 10139
    },
    {
      "epoch": 0.8724081562419341,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009270355093733191,
      "loss": 1.2082,
      "step": 10140
    },
    {
      "epoch": 0.8724941925492558,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009270214433368065,
      "loss": 0.7889,
      "step": 10141
    },
    {
      "epoch": 0.8725802288565775,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009270073760513391,
      "loss": 1.2506,
      "step": 10142
    },
    {
      "epoch": 0.8726662651638991,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009269933075169581,
      "loss": 1.2288,
      "step": 10143
    },
    {
      "epoch": 0.8727523014712208,
      "grad_norm": 1.4375,
      "learning_rate": 0.000926979237733705,
      "loss": 1.1659,
      "step": 10144
    },
    {
      "epoch": 0.8728383377785426,
      "grad_norm": 2.453125,
      "learning_rate": 0.0009269651667016205,
      "loss": 1.0993,
      "step": 10145
    },
    {
      "epoch": 0.8729243740858642,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009269510944207459,
      "loss": 1.8523,
      "step": 10146
    },
    {
      "epoch": 0.8730104103931859,
      "grad_norm": 1.0,
      "learning_rate": 0.0009269370208911223,
      "loss": 0.7333,
      "step": 10147
    },
    {
      "epoch": 0.8730964467005076,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009269229461127911,
      "loss": 1.7118,
      "step": 10148
    },
    {
      "epoch": 0.8731824830078293,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009269088700857933,
      "loss": 1.4313,
      "step": 10149
    },
    {
      "epoch": 0.873268519315151,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009268947928101701,
      "loss": 1.1249,
      "step": 10150
    },
    {
      "epoch": 0.8733545556224727,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009268807142859627,
      "loss": 2.0972,
      "step": 10151
    },
    {
      "epoch": 0.8734405919297944,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009268666345132122,
      "loss": 1.9592,
      "step": 10152
    },
    {
      "epoch": 0.873526628237116,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009268525534919598,
      "loss": 1.6627,
      "step": 10153
    },
    {
      "epoch": 0.8736126645444378,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009268384712222466,
      "loss": 1.8987,
      "step": 10154
    },
    {
      "epoch": 0.8736987008517595,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000926824387704114,
      "loss": 1.4348,
      "step": 10155
    },
    {
      "epoch": 0.8737847371590811,
      "grad_norm": 1.265625,
      "learning_rate": 0.000926810302937603,
      "loss": 1.6993,
      "step": 10156
    },
    {
      "epoch": 0.8738707734664029,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009267962169227548,
      "loss": 1.3013,
      "step": 10157
    },
    {
      "epoch": 0.8739568097737245,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009267821296596109,
      "loss": 1.7156,
      "step": 10158
    },
    {
      "epoch": 0.8740428460810462,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009267680411482122,
      "loss": 0.8164,
      "step": 10159
    },
    {
      "epoch": 0.874128882388368,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009267539513885998,
      "loss": 2.0032,
      "step": 10160
    },
    {
      "epoch": 0.8742149186956896,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009267398603808154,
      "loss": 1.2107,
      "step": 10161
    },
    {
      "epoch": 0.8743009550030113,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009267257681248997,
      "loss": 1.1625,
      "step": 10162
    },
    {
      "epoch": 0.8743869913103329,
      "grad_norm": 2.8125,
      "learning_rate": 0.0009267116746208942,
      "loss": 0.9639,
      "step": 10163
    },
    {
      "epoch": 0.8744730276176547,
      "grad_norm": 1.375,
      "learning_rate": 0.00092669757986884,
      "loss": 0.8424,
      "step": 10164
    },
    {
      "epoch": 0.8745590639249763,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009266834838687784,
      "loss": 1.0893,
      "step": 10165
    },
    {
      "epoch": 0.874645100232298,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009266693866207505,
      "loss": 1.4416,
      "step": 10166
    },
    {
      "epoch": 0.8747311365396198,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009266552881247979,
      "loss": 1.6124,
      "step": 10167
    },
    {
      "epoch": 0.8748171728469414,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009266411883809611,
      "loss": 1.7367,
      "step": 10168
    },
    {
      "epoch": 0.8749032091542631,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009266270873892822,
      "loss": 1.3286,
      "step": 10169
    },
    {
      "epoch": 0.8749892454615847,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009266129851498018,
      "loss": 1.4798,
      "step": 10170
    },
    {
      "epoch": 0.8750752817689065,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009265988816625616,
      "loss": 1.8396,
      "step": 10171
    },
    {
      "epoch": 0.8751613180762282,
      "grad_norm": 0.875,
      "learning_rate": 0.0009265847769276026,
      "loss": 1.752,
      "step": 10172
    },
    {
      "epoch": 0.8752473543835498,
      "grad_norm": 1.296875,
      "learning_rate": 0.000926570670944966,
      "loss": 1.3893,
      "step": 10173
    },
    {
      "epoch": 0.8753333906908716,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009265565637146931,
      "loss": 1.6894,
      "step": 10174
    },
    {
      "epoch": 0.8754194269981932,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009265424552368254,
      "loss": 1.0726,
      "step": 10175
    },
    {
      "epoch": 0.8755054633055149,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009265283455114037,
      "loss": 0.8565,
      "step": 10176
    },
    {
      "epoch": 0.8755914996128367,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009265142345384698,
      "loss": 1.1792,
      "step": 10177
    },
    {
      "epoch": 0.8756775359201583,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009265001223180647,
      "loss": 2.1633,
      "step": 10178
    },
    {
      "epoch": 0.87576357222748,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009264860088502296,
      "loss": 1.0943,
      "step": 10179
    },
    {
      "epoch": 0.8758496085348016,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009264718941350058,
      "loss": 1.4158,
      "step": 10180
    },
    {
      "epoch": 0.8759356448421234,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009264577781724348,
      "loss": 1.2653,
      "step": 10181
    },
    {
      "epoch": 0.8760216811494451,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009264436609625577,
      "loss": 1.5433,
      "step": 10182
    },
    {
      "epoch": 0.8761077174567667,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009264295425054158,
      "loss": 1.0903,
      "step": 10183
    },
    {
      "epoch": 0.8761937537640885,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009264154228010504,
      "loss": 1.2261,
      "step": 10184
    },
    {
      "epoch": 0.8762797900714101,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009264013018495028,
      "loss": 1.2298,
      "step": 10185
    },
    {
      "epoch": 0.8763658263787318,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009263871796508144,
      "loss": 1.6139,
      "step": 10186
    },
    {
      "epoch": 0.8764518626860535,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009263730562050265,
      "loss": 0.9936,
      "step": 10187
    },
    {
      "epoch": 0.8765378989933752,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009263589315121802,
      "loss": 1.6064,
      "step": 10188
    },
    {
      "epoch": 0.8766239353006969,
      "grad_norm": 1.21875,
      "learning_rate": 0.000926344805572317,
      "loss": 1.1079,
      "step": 10189
    },
    {
      "epoch": 0.8767099716080186,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009263306783854782,
      "loss": 0.9907,
      "step": 10190
    },
    {
      "epoch": 0.8767960079153403,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000926316549951705,
      "loss": 1.8146,
      "step": 10191
    },
    {
      "epoch": 0.8768820442226619,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009263024202710388,
      "loss": 1.3866,
      "step": 10192
    },
    {
      "epoch": 0.8769680805299837,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009262882893435211,
      "loss": 1.7574,
      "step": 10193
    },
    {
      "epoch": 0.8770541168373054,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009262741571691929,
      "loss": 1.5095,
      "step": 10194
    },
    {
      "epoch": 0.877140153144627,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009262600237480957,
      "loss": 1.8853,
      "step": 10195
    },
    {
      "epoch": 0.8772261894519487,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009262458890802709,
      "loss": 1.6619,
      "step": 10196
    },
    {
      "epoch": 0.8773122257592704,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009262317531657598,
      "loss": 1.469,
      "step": 10197
    },
    {
      "epoch": 0.8773982620665921,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009262176160046037,
      "loss": 1.3767,
      "step": 10198
    },
    {
      "epoch": 0.8774842983739138,
      "grad_norm": 1.765625,
      "learning_rate": 0.0009262034775968438,
      "loss": 0.9843,
      "step": 10199
    },
    {
      "epoch": 0.8775703346812355,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009261893379425218,
      "loss": 1.1892,
      "step": 10200
    },
    {
      "epoch": 0.8776563709885572,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009261751970416788,
      "loss": 1.6509,
      "step": 10201
    },
    {
      "epoch": 0.8777424072958788,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009261610548943562,
      "loss": 1.2171,
      "step": 10202
    },
    {
      "epoch": 0.8778284436032006,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009261469115005955,
      "loss": 0.9477,
      "step": 10203
    },
    {
      "epoch": 0.8779144799105223,
      "grad_norm": 1.609375,
      "learning_rate": 0.000926132766860438,
      "loss": 1.0081,
      "step": 10204
    },
    {
      "epoch": 0.8780005162178439,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009261186209739249,
      "loss": 1.1965,
      "step": 10205
    },
    {
      "epoch": 0.8780865525251657,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009261044738410977,
      "loss": 1.463,
      "step": 10206
    },
    {
      "epoch": 0.8781725888324873,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009260903254619979,
      "loss": 0.7637,
      "step": 10207
    },
    {
      "epoch": 0.878258625139809,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009260761758366667,
      "loss": 1.0313,
      "step": 10208
    },
    {
      "epoch": 0.8783446614471306,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009260620249651457,
      "loss": 0.8426,
      "step": 10209
    },
    {
      "epoch": 0.8784306977544524,
      "grad_norm": 1.46875,
      "learning_rate": 0.000926047872847476,
      "loss": 1.2071,
      "step": 10210
    },
    {
      "epoch": 0.8785167340617741,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009260337194836991,
      "loss": 1.0362,
      "step": 10211
    },
    {
      "epoch": 0.8786027703690957,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009260195648738565,
      "loss": 0.9689,
      "step": 10212
    },
    {
      "epoch": 0.8786888066764175,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009260054090179894,
      "loss": 1.0298,
      "step": 10213
    },
    {
      "epoch": 0.8787748429837391,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009259912519161396,
      "loss": 1.9615,
      "step": 10214
    },
    {
      "epoch": 0.8788608792910608,
      "grad_norm": 1.234375,
      "learning_rate": 0.000925977093568348,
      "loss": 1.4733,
      "step": 10215
    },
    {
      "epoch": 0.8789469155983826,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009259629339746564,
      "loss": 0.8739,
      "step": 10216
    },
    {
      "epoch": 0.8790329519057042,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000925948773135106,
      "loss": 1.9837,
      "step": 10217
    },
    {
      "epoch": 0.8791189882130259,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009259346110497382,
      "loss": 0.8025,
      "step": 10218
    },
    {
      "epoch": 0.8792050245203475,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009259204477185946,
      "loss": 1.3527,
      "step": 10219
    },
    {
      "epoch": 0.8792910608276693,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009259062831417164,
      "loss": 0.8462,
      "step": 10220
    },
    {
      "epoch": 0.879377097134991,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009258921173191454,
      "loss": 0.8396,
      "step": 10221
    },
    {
      "epoch": 0.8794631334423126,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009258779502509227,
      "loss": 0.9115,
      "step": 10222
    },
    {
      "epoch": 0.8795491697496344,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009258637819370896,
      "loss": 1.8493,
      "step": 10223
    },
    {
      "epoch": 0.879635206056956,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000925849612377688,
      "loss": 1.8507,
      "step": 10224
    },
    {
      "epoch": 0.8797212423642777,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009258354415727591,
      "loss": 0.7778,
      "step": 10225
    },
    {
      "epoch": 0.8798072786715995,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009258212695223443,
      "loss": 1.4321,
      "step": 10226
    },
    {
      "epoch": 0.8798933149789211,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009258070962264851,
      "loss": 1.2182,
      "step": 10227
    },
    {
      "epoch": 0.8799793512862428,
      "grad_norm": 1.46875,
      "learning_rate": 0.000925792921685223,
      "loss": 2.3985,
      "step": 10228
    },
    {
      "epoch": 0.8800653875935645,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009257787458985993,
      "loss": 1.1215,
      "step": 10229
    },
    {
      "epoch": 0.8801514239008862,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009257645688666556,
      "loss": 1.2553,
      "step": 10230
    },
    {
      "epoch": 0.8802374602082078,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009257503905894334,
      "loss": 1.1925,
      "step": 10231
    },
    {
      "epoch": 0.8803234965155295,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009257362110669743,
      "loss": 2.1383,
      "step": 10232
    },
    {
      "epoch": 0.8804095328228513,
      "grad_norm": 0.8125,
      "learning_rate": 0.0009257220302993192,
      "loss": 0.9748,
      "step": 10233
    },
    {
      "epoch": 0.8804955691301729,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009257078482865103,
      "loss": 1.7761,
      "step": 10234
    },
    {
      "epoch": 0.8805816054374946,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009256936650285885,
      "loss": 1.3827,
      "step": 10235
    },
    {
      "epoch": 0.8806676417448163,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009256794805255957,
      "loss": 1.135,
      "step": 10236
    },
    {
      "epoch": 0.880753678052138,
      "grad_norm": 2.625,
      "learning_rate": 0.000925665294777573,
      "loss": 1.751,
      "step": 10237
    },
    {
      "epoch": 0.8808397143594597,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009256511077845623,
      "loss": 1.2453,
      "step": 10238
    },
    {
      "epoch": 0.8809257506667814,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009256369195466047,
      "loss": 1.5382,
      "step": 10239
    },
    {
      "epoch": 0.8810117869741031,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009256227300637421,
      "loss": 1.427,
      "step": 10240
    },
    {
      "epoch": 0.8810978232814247,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009256085393360156,
      "loss": 1.0804,
      "step": 10241
    },
    {
      "epoch": 0.8811838595887465,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009255943473634672,
      "loss": 1.2449,
      "step": 10242
    },
    {
      "epoch": 0.8812698958960682,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009255801541461379,
      "loss": 1.6595,
      "step": 10243
    },
    {
      "epoch": 0.8813559322033898,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009255659596840693,
      "loss": 1.3338,
      "step": 10244
    },
    {
      "epoch": 0.8814419685107115,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009255517639773032,
      "loss": 0.9791,
      "step": 10245
    },
    {
      "epoch": 0.8815280048180332,
      "grad_norm": 1.125,
      "learning_rate": 0.000925537567025881,
      "loss": 0.667,
      "step": 10246
    },
    {
      "epoch": 0.8816140411253549,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009255233688298441,
      "loss": 1.162,
      "step": 10247
    },
    {
      "epoch": 0.8817000774326766,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000925509169389234,
      "loss": 0.7873,
      "step": 10248
    },
    {
      "epoch": 0.8817861137399983,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009254949687040925,
      "loss": 1.0102,
      "step": 10249
    },
    {
      "epoch": 0.88187215004732,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009254807667744609,
      "loss": 0.8251,
      "step": 10250
    },
    {
      "epoch": 0.8819581863546416,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009254665636003809,
      "loss": 1.4351,
      "step": 10251
    },
    {
      "epoch": 0.8820442226619634,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009254523591818938,
      "loss": 1.227,
      "step": 10252
    },
    {
      "epoch": 0.882130258969285,
      "grad_norm": 1.25,
      "learning_rate": 0.0009254381535190413,
      "loss": 1.7572,
      "step": 10253
    },
    {
      "epoch": 0.8822162952766067,
      "grad_norm": 31.25,
      "learning_rate": 0.000925423946611865,
      "loss": 1.7051,
      "step": 10254
    },
    {
      "epoch": 0.8823023315839285,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0009254097384604065,
      "loss": 1.0754,
      "step": 10255
    },
    {
      "epoch": 0.8823883678912501,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009253955290647072,
      "loss": 2.2239,
      "step": 10256
    },
    {
      "epoch": 0.8824744041985718,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009253813184248085,
      "loss": 1.1739,
      "step": 10257
    },
    {
      "epoch": 0.8825604405058934,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009253671065407525,
      "loss": 1.0948,
      "step": 10258
    },
    {
      "epoch": 0.8826464768132152,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009253528934125802,
      "loss": 1.1399,
      "step": 10259
    },
    {
      "epoch": 0.8827325131205369,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009253386790403336,
      "loss": 1.7076,
      "step": 10260
    },
    {
      "epoch": 0.8828185494278585,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000925324463424054,
      "loss": 1.045,
      "step": 10261
    },
    {
      "epoch": 0.8829045857351803,
      "grad_norm": 1.265625,
      "learning_rate": 0.000925310246563783,
      "loss": 1.1649,
      "step": 10262
    },
    {
      "epoch": 0.8829906220425019,
      "grad_norm": 1.25,
      "learning_rate": 0.0009252960284595624,
      "loss": 1.2055,
      "step": 10263
    },
    {
      "epoch": 0.8830766583498236,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009252818091114335,
      "loss": 1.6394,
      "step": 10264
    },
    {
      "epoch": 0.8831626946571454,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009252675885194381,
      "loss": 0.7604,
      "step": 10265
    },
    {
      "epoch": 0.883248730964467,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009252533666836176,
      "loss": 2.0255,
      "step": 10266
    },
    {
      "epoch": 0.8833347672717887,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009252391436040138,
      "loss": 1.3224,
      "step": 10267
    },
    {
      "epoch": 0.8834208035791103,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009252249192806681,
      "loss": 1.716,
      "step": 10268
    },
    {
      "epoch": 0.8835068398864321,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009252106937136224,
      "loss": 1.7627,
      "step": 10269
    },
    {
      "epoch": 0.8835928761937538,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000925196466902918,
      "loss": 0.8414,
      "step": 10270
    },
    {
      "epoch": 0.8836789125010754,
      "grad_norm": 1.25,
      "learning_rate": 0.0009251822388485967,
      "loss": 1.3623,
      "step": 10271
    },
    {
      "epoch": 0.8837649488083972,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009251680095507,
      "loss": 1.3282,
      "step": 10272
    },
    {
      "epoch": 0.8838509851157188,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009251537790092695,
      "loss": 0.6384,
      "step": 10273
    },
    {
      "epoch": 0.8839370214230405,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000925139547224347,
      "loss": 1.1249,
      "step": 10274
    },
    {
      "epoch": 0.8840230577303622,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009251253141959739,
      "loss": 1.2065,
      "step": 10275
    },
    {
      "epoch": 0.8841090940376839,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000925111079924192,
      "loss": 1.2535,
      "step": 10276
    },
    {
      "epoch": 0.8841951303450056,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009250968444090428,
      "loss": 1.1378,
      "step": 10277
    },
    {
      "epoch": 0.8842811666523273,
      "grad_norm": 1.375,
      "learning_rate": 0.000925082607650568,
      "loss": 2.1796,
      "step": 10278
    },
    {
      "epoch": 0.884367202959649,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009250683696488094,
      "loss": 0.736,
      "step": 10279
    },
    {
      "epoch": 0.8844532392669706,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009250541304038084,
      "loss": 0.8509,
      "step": 10280
    },
    {
      "epoch": 0.8845392755742923,
      "grad_norm": 3.140625,
      "learning_rate": 0.0009250398899156068,
      "loss": 1.1813,
      "step": 10281
    },
    {
      "epoch": 0.8846253118816141,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009250256481842462,
      "loss": 1.2584,
      "step": 10282
    },
    {
      "epoch": 0.8847113481889357,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009250114052097681,
      "loss": 1.529,
      "step": 10283
    },
    {
      "epoch": 0.8847973844962574,
      "grad_norm": 6.5,
      "learning_rate": 0.0009249971609922145,
      "loss": 1.7189,
      "step": 10284
    },
    {
      "epoch": 0.8848834208035791,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009249829155316266,
      "loss": 2.239,
      "step": 10285
    },
    {
      "epoch": 0.8849694571109008,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009249686688280465,
      "loss": 1.0103,
      "step": 10286
    },
    {
      "epoch": 0.8850554934182225,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009249544208815157,
      "loss": 1.627,
      "step": 10287
    },
    {
      "epoch": 0.8851415297255442,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009249401716920758,
      "loss": 2.1204,
      "step": 10288
    },
    {
      "epoch": 0.8852275660328659,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009249259212597685,
      "loss": 0.9908,
      "step": 10289
    },
    {
      "epoch": 0.8853136023401875,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009249116695846356,
      "loss": 1.1521,
      "step": 10290
    },
    {
      "epoch": 0.8853996386475093,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009248974166667188,
      "loss": 1.7396,
      "step": 10291
    },
    {
      "epoch": 0.885485674954831,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009248831625060595,
      "loss": 1.4519,
      "step": 10292
    },
    {
      "epoch": 0.8855717112621526,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009248689071026996,
      "loss": 0.7236,
      "step": 10293
    },
    {
      "epoch": 0.8856577475694744,
      "grad_norm": 1.3125,
      "learning_rate": 0.000924854650456681,
      "loss": 1.642,
      "step": 10294
    },
    {
      "epoch": 0.885743783876796,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009248403925680449,
      "loss": 1.2963,
      "step": 10295
    },
    {
      "epoch": 0.8858298201841177,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009248261334368335,
      "loss": 0.6327,
      "step": 10296
    },
    {
      "epoch": 0.8859158564914393,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009248118730630882,
      "loss": 1.2257,
      "step": 10297
    },
    {
      "epoch": 0.8860018927987611,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009247976114468508,
      "loss": 1.0483,
      "step": 10298
    },
    {
      "epoch": 0.8860879291060828,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009247833485881629,
      "loss": 1.7463,
      "step": 10299
    },
    {
      "epoch": 0.8861739654134044,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009247690844870665,
      "loss": 1.6245,
      "step": 10300
    },
    {
      "epoch": 0.8862600017207262,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009247548191436031,
      "loss": 0.8782,
      "step": 10301
    },
    {
      "epoch": 0.8863460380280478,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009247405525578145,
      "loss": 1.2283,
      "step": 10302
    },
    {
      "epoch": 0.8864320743353695,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009247262847297422,
      "loss": 0.8767,
      "step": 10303
    },
    {
      "epoch": 0.8865181106426913,
      "grad_norm": 1.625,
      "learning_rate": 0.0009247120156594282,
      "loss": 1.8567,
      "step": 10304
    },
    {
      "epoch": 0.8866041469500129,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009246977453469143,
      "loss": 1.3152,
      "step": 10305
    },
    {
      "epoch": 0.8866901832573346,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009246834737922419,
      "loss": 1.3829,
      "step": 10306
    },
    {
      "epoch": 0.8867762195646562,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009246692009954531,
      "loss": 1.3244,
      "step": 10307
    },
    {
      "epoch": 0.886862255871978,
      "grad_norm": 1.0,
      "learning_rate": 0.0009246549269565894,
      "loss": 1.0344,
      "step": 10308
    },
    {
      "epoch": 0.8869482921792997,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009246406516756927,
      "loss": 1.2886,
      "step": 10309
    },
    {
      "epoch": 0.8870343284866213,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009246263751528046,
      "loss": 0.6746,
      "step": 10310
    },
    {
      "epoch": 0.8871203647939431,
      "grad_norm": 1.25,
      "learning_rate": 0.0009246120973879668,
      "loss": 1.4129,
      "step": 10311
    },
    {
      "epoch": 0.8872064011012647,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0009245978183812214,
      "loss": 1.1437,
      "step": 10312
    },
    {
      "epoch": 0.8872924374085864,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00092458353813261,
      "loss": 1.8432,
      "step": 10313
    },
    {
      "epoch": 0.8873784737159082,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009245692566421742,
      "loss": 1.4501,
      "step": 10314
    },
    {
      "epoch": 0.8874645100232298,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009245549739099559,
      "loss": 1.0483,
      "step": 10315
    },
    {
      "epoch": 0.8875505463305515,
      "grad_norm": 1.296875,
      "learning_rate": 0.000924540689935997,
      "loss": 1.6345,
      "step": 10316
    },
    {
      "epoch": 0.8876365826378732,
      "grad_norm": 1.640625,
      "learning_rate": 0.000924526404720339,
      "loss": 1.5014,
      "step": 10317
    },
    {
      "epoch": 0.8877226189451949,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009245121182630239,
      "loss": 1.0135,
      "step": 10318
    },
    {
      "epoch": 0.8878086552525165,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009244978305640934,
      "loss": 1.9855,
      "step": 10319
    },
    {
      "epoch": 0.8878946915598382,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009244835416235893,
      "loss": 2.3601,
      "step": 10320
    },
    {
      "epoch": 0.88798072786716,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009244692514415534,
      "loss": 1.5864,
      "step": 10321
    },
    {
      "epoch": 0.8880667641744816,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009244549600180274,
      "loss": 1.8479,
      "step": 10322
    },
    {
      "epoch": 0.8881528004818033,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009244406673530535,
      "loss": 1.2989,
      "step": 10323
    },
    {
      "epoch": 0.888238836789125,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009244263734466729,
      "loss": 0.5631,
      "step": 10324
    },
    {
      "epoch": 0.8883248730964467,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009244120782989278,
      "loss": 2.2136,
      "step": 10325
    },
    {
      "epoch": 0.8884109094037684,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009243977819098599,
      "loss": 1.0444,
      "step": 10326
    },
    {
      "epoch": 0.8884969457110901,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009243834842795111,
      "loss": 0.7619,
      "step": 10327
    },
    {
      "epoch": 0.8885829820184118,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000924369185407923,
      "loss": 0.5756,
      "step": 10328
    },
    {
      "epoch": 0.8886690183257334,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009243548852951378,
      "loss": 1.8154,
      "step": 10329
    },
    {
      "epoch": 0.8887550546330552,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009243405839411969,
      "loss": 1.598,
      "step": 10330
    },
    {
      "epoch": 0.8888410909403769,
      "grad_norm": 2.203125,
      "learning_rate": 0.0009243262813461424,
      "loss": 1.2218,
      "step": 10331
    },
    {
      "epoch": 0.8889271272476985,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009243119775100159,
      "loss": 0.8026,
      "step": 10332
    },
    {
      "epoch": 0.8890131635550202,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009242976724328595,
      "loss": 1.295,
      "step": 10333
    },
    {
      "epoch": 0.8890991998623419,
      "grad_norm": 1.0625,
      "learning_rate": 0.000924283366114715,
      "loss": 0.999,
      "step": 10334
    },
    {
      "epoch": 0.8891852361696636,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009242690585556241,
      "loss": 1.0464,
      "step": 10335
    },
    {
      "epoch": 0.8892712724769853,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009242547497556289,
      "loss": 0.9386,
      "step": 10336
    },
    {
      "epoch": 0.889357308784307,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009242404397147708,
      "loss": 1.586,
      "step": 10337
    },
    {
      "epoch": 0.8894433450916287,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009242261284330921,
      "loss": 1.9969,
      "step": 10338
    },
    {
      "epoch": 0.8895293813989503,
      "grad_norm": 2.859375,
      "learning_rate": 0.0009242118159106343,
      "loss": 1.7441,
      "step": 10339
    },
    {
      "epoch": 0.8896154177062721,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009241975021474396,
      "loss": 1.3557,
      "step": 10340
    },
    {
      "epoch": 0.8897014540135937,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009241831871435497,
      "loss": 1.7263,
      "step": 10341
    },
    {
      "epoch": 0.8897874903209154,
      "grad_norm": 1.375,
      "learning_rate": 0.0009241688708990064,
      "loss": 1.2741,
      "step": 10342
    },
    {
      "epoch": 0.8898735266282372,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009241545534138517,
      "loss": 1.0546,
      "step": 10343
    },
    {
      "epoch": 0.8899595629355588,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009241402346881274,
      "loss": 0.9968,
      "step": 10344
    },
    {
      "epoch": 0.8900455992428805,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009241259147218754,
      "loss": 1.0692,
      "step": 10345
    },
    {
      "epoch": 0.8901316355502021,
      "grad_norm": 4.125,
      "learning_rate": 0.0009241115935151377,
      "loss": 1.689,
      "step": 10346
    },
    {
      "epoch": 0.8902176718575239,
      "grad_norm": 4.21875,
      "learning_rate": 0.000924097271067956,
      "loss": 1.1649,
      "step": 10347
    },
    {
      "epoch": 0.8903037081648456,
      "grad_norm": 1.125,
      "learning_rate": 0.0009240829473803723,
      "loss": 1.2601,
      "step": 10348
    },
    {
      "epoch": 0.8903897444721672,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009240686224524283,
      "loss": 1.3537,
      "step": 10349
    },
    {
      "epoch": 0.890475780779489,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009240542962841661,
      "loss": 1.0109,
      "step": 10350
    },
    {
      "epoch": 0.8905618170868106,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009240399688756276,
      "loss": 1.875,
      "step": 10351
    },
    {
      "epoch": 0.8906478533941323,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009240256402268548,
      "loss": 1.1295,
      "step": 10352
    },
    {
      "epoch": 0.8907338897014541,
      "grad_norm": 4.0625,
      "learning_rate": 0.0009240113103378894,
      "loss": 1.8045,
      "step": 10353
    },
    {
      "epoch": 0.8908199260087757,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009239969792087733,
      "loss": 1.2522,
      "step": 10354
    },
    {
      "epoch": 0.8909059623160974,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009239826468395485,
      "loss": 1.1994,
      "step": 10355
    },
    {
      "epoch": 0.890991998623419,
      "grad_norm": 1.6953125,
      "learning_rate": 0.000923968313230257,
      "loss": 1.1922,
      "step": 10356
    },
    {
      "epoch": 0.8910780349307408,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009239539783809405,
      "loss": 1.4832,
      "step": 10357
    },
    {
      "epoch": 0.8911640712380625,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000923939642291641,
      "loss": 1.1942,
      "step": 10358
    },
    {
      "epoch": 0.8912501075453841,
      "grad_norm": 1.25,
      "learning_rate": 0.0009239253049624007,
      "loss": 1.3171,
      "step": 10359
    },
    {
      "epoch": 0.8913361438527059,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009239109663932614,
      "loss": 1.5112,
      "step": 10360
    },
    {
      "epoch": 0.8914221801600275,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009238966265842648,
      "loss": 0.8985,
      "step": 10361
    },
    {
      "epoch": 0.8915082164673492,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009238822855354531,
      "loss": 0.9253,
      "step": 10362
    },
    {
      "epoch": 0.8915942527746709,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009238679432468682,
      "loss": 1.5696,
      "step": 10363
    },
    {
      "epoch": 0.8916802890819926,
      "grad_norm": 1.125,
      "learning_rate": 0.000923853599718552,
      "loss": 0.911,
      "step": 10364
    },
    {
      "epoch": 0.8917663253893143,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009238392549505463,
      "loss": 1.7629,
      "step": 10365
    },
    {
      "epoch": 0.891852361696636,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009238249089428932,
      "loss": 1.3004,
      "step": 10366
    },
    {
      "epoch": 0.8919383980039577,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009238105616956349,
      "loss": 1.2961,
      "step": 10367
    },
    {
      "epoch": 0.8920244343112793,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009237962132088129,
      "loss": 2.0113,
      "step": 10368
    },
    {
      "epoch": 0.892110470618601,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009237818634824695,
      "loss": 1.8199,
      "step": 10369
    },
    {
      "epoch": 0.8921965069259228,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009237675125166467,
      "loss": 1.7791,
      "step": 10370
    },
    {
      "epoch": 0.8922825432332444,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009237531603113862,
      "loss": 0.9897,
      "step": 10371
    },
    {
      "epoch": 0.8923685795405661,
      "grad_norm": 1.421875,
      "learning_rate": 0.00092373880686673,
      "loss": 1.6447,
      "step": 10372
    },
    {
      "epoch": 0.8924546158478878,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0009237244521827204,
      "loss": 0.8906,
      "step": 10373
    },
    {
      "epoch": 0.8925406521552095,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009237100962593992,
      "loss": 1.5728,
      "step": 10374
    },
    {
      "epoch": 0.8926266884625312,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009236957390968083,
      "loss": 2.2593,
      "step": 10375
    },
    {
      "epoch": 0.8927127247698529,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009236813806949897,
      "loss": 0.6503,
      "step": 10376
    },
    {
      "epoch": 0.8927987610771746,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009236670210539855,
      "loss": 1.2984,
      "step": 10377
    },
    {
      "epoch": 0.8928847973844962,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009236526601738378,
      "loss": 1.442,
      "step": 10378
    },
    {
      "epoch": 0.892970833691818,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009236382980545883,
      "loss": 1.3462,
      "step": 10379
    },
    {
      "epoch": 0.8930568699991397,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009236239346962793,
      "loss": 1.2539,
      "step": 10380
    },
    {
      "epoch": 0.8931429063064613,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009236095700989527,
      "loss": 1.3118,
      "step": 10381
    },
    {
      "epoch": 0.893228942613783,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009235952042626503,
      "loss": 1.8898,
      "step": 10382
    },
    {
      "epoch": 0.8933149789211047,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009235808371874144,
      "loss": 1.2656,
      "step": 10383
    },
    {
      "epoch": 0.8934010152284264,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000923566468873287,
      "loss": 2.0336,
      "step": 10384
    },
    {
      "epoch": 0.893487051535748,
      "grad_norm": 36.25,
      "learning_rate": 0.0009235520993203099,
      "loss": 1.6786,
      "step": 10385
    },
    {
      "epoch": 0.8935730878430698,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009235377285285254,
      "loss": 1.6578,
      "step": 10386
    },
    {
      "epoch": 0.8936591241503915,
      "grad_norm": 1.625,
      "learning_rate": 0.0009235233564979754,
      "loss": 1.4011,
      "step": 10387
    },
    {
      "epoch": 0.8937451604577131,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000923508983228702,
      "loss": 1.1595,
      "step": 10388
    },
    {
      "epoch": 0.8938311967650349,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0009234946087207472,
      "loss": 0.5597,
      "step": 10389
    },
    {
      "epoch": 0.8939172330723565,
      "grad_norm": 1.8125,
      "learning_rate": 0.000923480232974153,
      "loss": 1.6108,
      "step": 10390
    },
    {
      "epoch": 0.8940032693796782,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009234658559889615,
      "loss": 0.7098,
      "step": 10391
    },
    {
      "epoch": 0.894089305687,
      "grad_norm": 1.5,
      "learning_rate": 0.0009234514777652147,
      "loss": 2.0756,
      "step": 10392
    },
    {
      "epoch": 0.8941753419943216,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009234370983029547,
      "loss": 0.7767,
      "step": 10393
    },
    {
      "epoch": 0.8942613783016433,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0009234227176022234,
      "loss": 1.0402,
      "step": 10394
    },
    {
      "epoch": 0.8943474146089649,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009234083356630631,
      "loss": 1.8945,
      "step": 10395
    },
    {
      "epoch": 0.8944334509162867,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009233939524855159,
      "loss": 1.2616,
      "step": 10396
    },
    {
      "epoch": 0.8945194872236084,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009233795680696237,
      "loss": 1.1754,
      "step": 10397
    },
    {
      "epoch": 0.89460552353093,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009233651824154286,
      "loss": 1.028,
      "step": 10398
    },
    {
      "epoch": 0.8946915598382518,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009233507955229726,
      "loss": 1.1916,
      "step": 10399
    },
    {
      "epoch": 0.8947775961455734,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009233364073922978,
      "loss": 0.9332,
      "step": 10400
    },
    {
      "epoch": 0.8948636324528951,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009233220180234464,
      "loss": 1.2232,
      "step": 10401
    },
    {
      "epoch": 0.8949496687602169,
      "grad_norm": 1.25,
      "learning_rate": 0.0009233076274164604,
      "loss": 1.5085,
      "step": 10402
    },
    {
      "epoch": 0.8950357050675385,
      "grad_norm": 2.453125,
      "learning_rate": 0.0009232932355713821,
      "loss": 1.6513,
      "step": 10403
    },
    {
      "epoch": 0.8951217413748602,
      "grad_norm": 2.75,
      "learning_rate": 0.0009232788424882532,
      "loss": 1.68,
      "step": 10404
    },
    {
      "epoch": 0.8952077776821818,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009232644481671161,
      "loss": 1.4305,
      "step": 10405
    },
    {
      "epoch": 0.8952938139895036,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009232500526080128,
      "loss": 2.0912,
      "step": 10406
    },
    {
      "epoch": 0.8953798502968252,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009232356558109854,
      "loss": 1.4107,
      "step": 10407
    },
    {
      "epoch": 0.8954658866041469,
      "grad_norm": 1.90625,
      "learning_rate": 0.0009232212577760761,
      "loss": 1.3902,
      "step": 10408
    },
    {
      "epoch": 0.8955519229114687,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009232068585033269,
      "loss": 1.1273,
      "step": 10409
    },
    {
      "epoch": 0.8956379592187903,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009231924579927799,
      "loss": 0.8547,
      "step": 10410
    },
    {
      "epoch": 0.895723995526112,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009231780562444773,
      "loss": 1.2102,
      "step": 10411
    },
    {
      "epoch": 0.8958100318334337,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009231636532584611,
      "loss": 1.5307,
      "step": 10412
    },
    {
      "epoch": 0.8958960681407554,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009231492490347736,
      "loss": 1.5513,
      "step": 10413
    },
    {
      "epoch": 0.8959821044480771,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009231348435734568,
      "loss": 1.5289,
      "step": 10414
    },
    {
      "epoch": 0.8960681407553988,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000923120436874553,
      "loss": 1.5121,
      "step": 10415
    },
    {
      "epoch": 0.8961541770627205,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009231060289381041,
      "loss": 1.0601,
      "step": 10416
    },
    {
      "epoch": 0.8962402133700421,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009230916197641523,
      "loss": 1.1051,
      "step": 10417
    },
    {
      "epoch": 0.8963262496773639,
      "grad_norm": 1.125,
      "learning_rate": 0.00092307720935274,
      "loss": 1.6908,
      "step": 10418
    },
    {
      "epoch": 0.8964122859846856,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000923062797703909,
      "loss": 1.447,
      "step": 10419
    },
    {
      "epoch": 0.8964983222920072,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009230483848177016,
      "loss": 0.9318,
      "step": 10420
    },
    {
      "epoch": 0.896584358599329,
      "grad_norm": 2.078125,
      "learning_rate": 0.00092303397069416,
      "loss": 1.9609,
      "step": 10421
    },
    {
      "epoch": 0.8966703949066506,
      "grad_norm": 1.125,
      "learning_rate": 0.0009230195553333263,
      "loss": 1.298,
      "step": 10422
    },
    {
      "epoch": 0.8967564312139723,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009230051387352427,
      "loss": 1.2231,
      "step": 10423
    },
    {
      "epoch": 0.896842467521294,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009229907208999513,
      "loss": 1.0946,
      "step": 10424
    },
    {
      "epoch": 0.8969285038286157,
      "grad_norm": 12.875,
      "learning_rate": 0.0009229763018274943,
      "loss": 1.6186,
      "step": 10425
    },
    {
      "epoch": 0.8970145401359374,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009229618815179138,
      "loss": 1.29,
      "step": 10426
    },
    {
      "epoch": 0.897100576443259,
      "grad_norm": 1.875,
      "learning_rate": 0.0009229474599712522,
      "loss": 1.5129,
      "step": 10427
    },
    {
      "epoch": 0.8971866127505808,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009229330371875515,
      "loss": 1.3287,
      "step": 10428
    },
    {
      "epoch": 0.8972726490579024,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009229186131668539,
      "loss": 1.2983,
      "step": 10429
    },
    {
      "epoch": 0.8973586853652241,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009229041879092017,
      "loss": 1.3922,
      "step": 10430
    },
    {
      "epoch": 0.8974447216725459,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009228897614146367,
      "loss": 0.9157,
      "step": 10431
    },
    {
      "epoch": 0.8975307579798675,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009228753336832016,
      "loss": 1.1319,
      "step": 10432
    },
    {
      "epoch": 0.8976167942871892,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009228609047149386,
      "loss": 1.3358,
      "step": 10433
    },
    {
      "epoch": 0.8977028305945108,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009228464745098895,
      "loss": 1.8129,
      "step": 10434
    },
    {
      "epoch": 0.8977888669018326,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009228320430680967,
      "loss": 0.9948,
      "step": 10435
    },
    {
      "epoch": 0.8978749032091543,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009228176103896024,
      "loss": 1.1625,
      "step": 10436
    },
    {
      "epoch": 0.8979609395164759,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009228031764744488,
      "loss": 1.172,
      "step": 10437
    },
    {
      "epoch": 0.8980469758237977,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009227887413226782,
      "loss": 1.4723,
      "step": 10438
    },
    {
      "epoch": 0.8981330121311193,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009227743049343327,
      "loss": 0.6827,
      "step": 10439
    },
    {
      "epoch": 0.898219048438441,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009227598673094546,
      "loss": 1.0475,
      "step": 10440
    },
    {
      "epoch": 0.8983050847457628,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009227454284480861,
      "loss": 1.3394,
      "step": 10441
    },
    {
      "epoch": 0.8983911210530844,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009227309883502694,
      "loss": 1.2269,
      "step": 10442
    },
    {
      "epoch": 0.8984771573604061,
      "grad_norm": 3.109375,
      "learning_rate": 0.000922716547016047,
      "loss": 1.1709,
      "step": 10443
    },
    {
      "epoch": 0.8985631936677277,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009227021044454607,
      "loss": 1.3614,
      "step": 10444
    },
    {
      "epoch": 0.8986492299750495,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000922687660638553,
      "loss": 1.0379,
      "step": 10445
    },
    {
      "epoch": 0.8987352662823712,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009226732155953662,
      "loss": 1.7727,
      "step": 10446
    },
    {
      "epoch": 0.8988213025896928,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009226587693159423,
      "loss": 1.2228,
      "step": 10447
    },
    {
      "epoch": 0.8989073388970146,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009226443218003237,
      "loss": 1.9143,
      "step": 10448
    },
    {
      "epoch": 0.8989933752043362,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009226298730485529,
      "loss": 1.3531,
      "step": 10449
    },
    {
      "epoch": 0.8990794115116579,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009226154230606717,
      "loss": 1.0186,
      "step": 10450
    },
    {
      "epoch": 0.8991654478189796,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009226009718367226,
      "loss": 1.1498,
      "step": 10451
    },
    {
      "epoch": 0.8992514841263013,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009225865193767479,
      "loss": 1.403,
      "step": 10452
    },
    {
      "epoch": 0.899337520433623,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009225720656807899,
      "loss": 1.5372,
      "step": 10453
    },
    {
      "epoch": 0.8994235567409447,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009225576107488907,
      "loss": 1.1602,
      "step": 10454
    },
    {
      "epoch": 0.8995095930482664,
      "grad_norm": 1.5,
      "learning_rate": 0.0009225431545810925,
      "loss": 1.5428,
      "step": 10455
    },
    {
      "epoch": 0.899595629355588,
      "grad_norm": 1.390625,
      "learning_rate": 0.000922528697177438,
      "loss": 1.4165,
      "step": 10456
    },
    {
      "epoch": 0.8996816656629097,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009225142385379691,
      "loss": 1.0172,
      "step": 10457
    },
    {
      "epoch": 0.8997677019702315,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009224997786627283,
      "loss": 1.6967,
      "step": 10458
    },
    {
      "epoch": 0.8998537382775531,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009224853175517578,
      "loss": 1.143,
      "step": 10459
    },
    {
      "epoch": 0.8999397745848748,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009224708552050999,
      "loss": 0.8618,
      "step": 10460
    },
    {
      "epoch": 0.9000258108921965,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009224563916227969,
      "loss": 1.1536,
      "step": 10461
    },
    {
      "epoch": 0.9001118471995182,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009224419268048912,
      "loss": 2.1927,
      "step": 10462
    },
    {
      "epoch": 0.9001978835068399,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009224274607514249,
      "loss": 1.1847,
      "step": 10463
    },
    {
      "epoch": 0.9002839198141616,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009224129934624405,
      "loss": 1.6831,
      "step": 10464
    },
    {
      "epoch": 0.9003699561214833,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009223985249379803,
      "loss": 1.9663,
      "step": 10465
    },
    {
      "epoch": 0.9004559924288049,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009223840551780865,
      "loss": 1.1553,
      "step": 10466
    },
    {
      "epoch": 0.9005420287361267,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009223695841828014,
      "loss": 1.5522,
      "step": 10467
    },
    {
      "epoch": 0.9006280650434484,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009223551119521676,
      "loss": 0.8112,
      "step": 10468
    },
    {
      "epoch": 0.90071410135077,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009223406384862272,
      "loss": 2.259,
      "step": 10469
    },
    {
      "epoch": 0.9008001376580917,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009223261637850227,
      "loss": 1.384,
      "step": 10470
    },
    {
      "epoch": 0.9008861739654134,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000922311687848596,
      "loss": 1.4007,
      "step": 10471
    },
    {
      "epoch": 0.9009722102727351,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009222972106769899,
      "loss": 1.001,
      "step": 10472
    },
    {
      "epoch": 0.9010582465800567,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009222827322702466,
      "loss": 0.8529,
      "step": 10473
    },
    {
      "epoch": 0.9011442828873785,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009222682526284084,
      "loss": 1.0403,
      "step": 10474
    },
    {
      "epoch": 0.9012303191947002,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009222537717515178,
      "loss": 1.744,
      "step": 10475
    },
    {
      "epoch": 0.9013163555020218,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009222392896396169,
      "loss": 2.0491,
      "step": 10476
    },
    {
      "epoch": 0.9014023918093436,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009222248062927483,
      "loss": 0.9404,
      "step": 10477
    },
    {
      "epoch": 0.9014884281166652,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009222103217109541,
      "loss": 1.9406,
      "step": 10478
    },
    {
      "epoch": 0.9015744644239869,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000922195835894277,
      "loss": 0.944,
      "step": 10479
    },
    {
      "epoch": 0.9016605007313087,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000922181348842759,
      "loss": 1.2945,
      "step": 10480
    },
    {
      "epoch": 0.9017465370386303,
      "grad_norm": 1.5,
      "learning_rate": 0.0009221668605564428,
      "loss": 1.3261,
      "step": 10481
    },
    {
      "epoch": 0.901832573345952,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009221523710353707,
      "loss": 2.0018,
      "step": 10482
    },
    {
      "epoch": 0.9019186096532736,
      "grad_norm": 1.65625,
      "learning_rate": 0.000922137880279585,
      "loss": 1.1767,
      "step": 10483
    },
    {
      "epoch": 0.9020046459605954,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000922123388289128,
      "loss": 1.1347,
      "step": 10484
    },
    {
      "epoch": 0.9020906822679171,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009221088950640421,
      "loss": 0.7969,
      "step": 10485
    },
    {
      "epoch": 0.9021767185752387,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009220944006043701,
      "loss": 1.9951,
      "step": 10486
    },
    {
      "epoch": 0.9022627548825605,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009220799049101538,
      "loss": 0.7441,
      "step": 10487
    },
    {
      "epoch": 0.9023487911898821,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000922065407981436,
      "loss": 1.8139,
      "step": 10488
    },
    {
      "epoch": 0.9024348274972038,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009220509098182589,
      "loss": 1.5181,
      "step": 10489
    },
    {
      "epoch": 0.9025208638045256,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009220364104206651,
      "loss": 1.2545,
      "step": 10490
    },
    {
      "epoch": 0.9026069001118472,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009220219097886967,
      "loss": 0.8978,
      "step": 10491
    },
    {
      "epoch": 0.9026929364191689,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009220074079223964,
      "loss": 1.1299,
      "step": 10492
    },
    {
      "epoch": 0.9027789727264905,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009219929048218064,
      "loss": 1.4532,
      "step": 10493
    },
    {
      "epoch": 0.9028650090338123,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009219784004869694,
      "loss": 1.8553,
      "step": 10494
    },
    {
      "epoch": 0.9029510453411339,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009219638949179276,
      "loss": 1.1071,
      "step": 10495
    },
    {
      "epoch": 0.9030370816484556,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009219493881147234,
      "loss": 1.8654,
      "step": 10496
    },
    {
      "epoch": 0.9031231179557774,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009219348800773992,
      "loss": 0.9922,
      "step": 10497
    },
    {
      "epoch": 0.903209154263099,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009219203708059978,
      "loss": 0.8306,
      "step": 10498
    },
    {
      "epoch": 0.9032951905704207,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009219058603005612,
      "loss": 0.7768,
      "step": 10499
    },
    {
      "epoch": 0.9033812268777424,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000921891348561132,
      "loss": 1.3143,
      "step": 10500
    },
    {
      "epoch": 0.9034672631850641,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009218768355877527,
      "loss": 1.4001,
      "step": 10501
    },
    {
      "epoch": 0.9035532994923858,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009218623213804658,
      "loss": 1.8323,
      "step": 10502
    },
    {
      "epoch": 0.9036393357997075,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009218478059393136,
      "loss": 1.5437,
      "step": 10503
    },
    {
      "epoch": 0.9037253721070292,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009218332892643387,
      "loss": 1.1366,
      "step": 10504
    },
    {
      "epoch": 0.9038114084143508,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009218187713555833,
      "loss": 0.9973,
      "step": 10505
    },
    {
      "epoch": 0.9038974447216725,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009218042522130901,
      "loss": 1.5723,
      "step": 10506
    },
    {
      "epoch": 0.9039834810289943,
      "grad_norm": 1.5,
      "learning_rate": 0.0009217897318369014,
      "loss": 1.5565,
      "step": 10507
    },
    {
      "epoch": 0.9040695173363159,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009217752102270599,
      "loss": 0.858,
      "step": 10508
    },
    {
      "epoch": 0.9041555536436376,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009217606873836079,
      "loss": 2.1263,
      "step": 10509
    },
    {
      "epoch": 0.9042415899509593,
      "grad_norm": 1.875,
      "learning_rate": 0.0009217461633065879,
      "loss": 2.0896,
      "step": 10510
    },
    {
      "epoch": 0.904327626258281,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009217316379960425,
      "loss": 1.3301,
      "step": 10511
    },
    {
      "epoch": 0.9044136625656027,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000921717111452014,
      "loss": 1.6255,
      "step": 10512
    },
    {
      "epoch": 0.9044996988729244,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009217025836745449,
      "loss": 1.3663,
      "step": 10513
    },
    {
      "epoch": 0.9045857351802461,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009216880546636779,
      "loss": 0.9588,
      "step": 10514
    },
    {
      "epoch": 0.9046717714875677,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009216735244194552,
      "loss": 1.8455,
      "step": 10515
    },
    {
      "epoch": 0.9047578077948895,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009216589929419196,
      "loss": 0.8257,
      "step": 10516
    },
    {
      "epoch": 0.9048438441022111,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009216444602311134,
      "loss": 1.0096,
      "step": 10517
    },
    {
      "epoch": 0.9049298804095328,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009216299262870791,
      "loss": 1.4591,
      "step": 10518
    },
    {
      "epoch": 0.9050159167168546,
      "grad_norm": 1.25,
      "learning_rate": 0.0009216153911098592,
      "loss": 1.3686,
      "step": 10519
    },
    {
      "epoch": 0.9051019530241762,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009216008546994965,
      "loss": 1.2832,
      "step": 10520
    },
    {
      "epoch": 0.9051879893314979,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009215863170560333,
      "loss": 1.1003,
      "step": 10521
    },
    {
      "epoch": 0.9052740256388195,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009215717781795119,
      "loss": 1.3477,
      "step": 10522
    },
    {
      "epoch": 0.9053600619461413,
      "grad_norm": 1.375,
      "learning_rate": 0.0009215572380699751,
      "loss": 1.6269,
      "step": 10523
    },
    {
      "epoch": 0.905446098253463,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009215426967274655,
      "loss": 1.2075,
      "step": 10524
    },
    {
      "epoch": 0.9055321345607846,
      "grad_norm": 1.625,
      "learning_rate": 0.0009215281541520254,
      "loss": 1.7449,
      "step": 10525
    },
    {
      "epoch": 0.9056181708681064,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009215136103436975,
      "loss": 0.6571,
      "step": 10526
    },
    {
      "epoch": 0.905704207175428,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009214990653025241,
      "loss": 0.8883,
      "step": 10527
    },
    {
      "epoch": 0.9057902434827497,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009214845190285481,
      "loss": 2.0309,
      "step": 10528
    },
    {
      "epoch": 0.9058762797900715,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009214699715218117,
      "loss": 1.2113,
      "step": 10529
    },
    {
      "epoch": 0.9059623160973931,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009214554227823577,
      "loss": 1.8055,
      "step": 10530
    },
    {
      "epoch": 0.9060483524047148,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009214408728102285,
      "loss": 1.085,
      "step": 10531
    },
    {
      "epoch": 0.9061343887120364,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009214263216054668,
      "loss": 1.5687,
      "step": 10532
    },
    {
      "epoch": 0.9062204250193582,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009214117691681151,
      "loss": 0.8866,
      "step": 10533
    },
    {
      "epoch": 0.9063064613266799,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009213972154982157,
      "loss": 1.9764,
      "step": 10534
    },
    {
      "epoch": 0.9063924976340015,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009213826605958118,
      "loss": 1.1407,
      "step": 10535
    },
    {
      "epoch": 0.9064785339413233,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009213681044609452,
      "loss": 1.5619,
      "step": 10536
    },
    {
      "epoch": 0.9065645702486449,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009213535470936591,
      "loss": 1.411,
      "step": 10537
    },
    {
      "epoch": 0.9066506065559666,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009213389884939957,
      "loss": 0.8838,
      "step": 10538
    },
    {
      "epoch": 0.9067366428632883,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009213244286619978,
      "loss": 1.2432,
      "step": 10539
    },
    {
      "epoch": 0.90682267917061,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009213098675977076,
      "loss": 0.7125,
      "step": 10540
    },
    {
      "epoch": 0.9069087154779317,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009212953053011683,
      "loss": 1.5908,
      "step": 10541
    },
    {
      "epoch": 0.9069947517852534,
      "grad_norm": 1.578125,
      "learning_rate": 0.000921280741772422,
      "loss": 1.0625,
      "step": 10542
    },
    {
      "epoch": 0.9070807880925751,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009212661770115115,
      "loss": 0.9798,
      "step": 10543
    },
    {
      "epoch": 0.9071668243998967,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009212516110184794,
      "loss": 1.5169,
      "step": 10544
    },
    {
      "epoch": 0.9072528607072184,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009212370437933681,
      "loss": 1.5749,
      "step": 10545
    },
    {
      "epoch": 0.9073388970145402,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009212224753362207,
      "loss": 1.4038,
      "step": 10546
    },
    {
      "epoch": 0.9074249333218618,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009212079056470792,
      "loss": 1.7903,
      "step": 10547
    },
    {
      "epoch": 0.9075109696291835,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009211933347259867,
      "loss": 1.1807,
      "step": 10548
    },
    {
      "epoch": 0.9075970059365052,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009211787625729855,
      "loss": 1.3365,
      "step": 10549
    },
    {
      "epoch": 0.9076830422438269,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009211641891881184,
      "loss": 1.1348,
      "step": 10550
    },
    {
      "epoch": 0.9077690785511486,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009211496145714277,
      "loss": 1.4344,
      "step": 10551
    },
    {
      "epoch": 0.9078551148584703,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009211350387229567,
      "loss": 1.4401,
      "step": 10552
    },
    {
      "epoch": 0.907941151165792,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009211204616427473,
      "loss": 1.7989,
      "step": 10553
    },
    {
      "epoch": 0.9080271874731136,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009211058833308426,
      "loss": 1.2152,
      "step": 10554
    },
    {
      "epoch": 0.9081132237804354,
      "grad_norm": 1.109375,
      "learning_rate": 0.000921091303787285,
      "loss": 1.1466,
      "step": 10555
    },
    {
      "epoch": 0.9081992600877571,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009210767230121174,
      "loss": 1.4541,
      "step": 10556
    },
    {
      "epoch": 0.9082852963950787,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009210621410053821,
      "loss": 1.085,
      "step": 10557
    },
    {
      "epoch": 0.9083713327024004,
      "grad_norm": 1.3125,
      "learning_rate": 0.000921047557767122,
      "loss": 1.2595,
      "step": 10558
    },
    {
      "epoch": 0.9084573690097221,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009210329732973797,
      "loss": 1.1968,
      "step": 10559
    },
    {
      "epoch": 0.9085434053170438,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009210183875961978,
      "loss": 1.4011,
      "step": 10560
    },
    {
      "epoch": 0.9086294416243654,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000921003800663619,
      "loss": 1.0311,
      "step": 10561
    },
    {
      "epoch": 0.9087154779316872,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009209892124996859,
      "loss": 0.8259,
      "step": 10562
    },
    {
      "epoch": 0.9088015142390089,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009209746231044414,
      "loss": 1.328,
      "step": 10563
    },
    {
      "epoch": 0.9088875505463305,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000920960032477928,
      "loss": 1.551,
      "step": 10564
    },
    {
      "epoch": 0.9089735868536523,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009209454406201883,
      "loss": 1.361,
      "step": 10565
    },
    {
      "epoch": 0.9090596231609739,
      "grad_norm": 1.296875,
      "learning_rate": 0.000920930847531265,
      "loss": 2.0072,
      "step": 10566
    },
    {
      "epoch": 0.9091456594682956,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009209162532112007,
      "loss": 1.891,
      "step": 10567
    },
    {
      "epoch": 0.9092316957756174,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009209016576600385,
      "loss": 1.296,
      "step": 10568
    },
    {
      "epoch": 0.909317732082939,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009208870608778208,
      "loss": 1.9042,
      "step": 10569
    },
    {
      "epoch": 0.9094037683902607,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009208724628645902,
      "loss": 1.8252,
      "step": 10570
    },
    {
      "epoch": 0.9094898046975823,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009208578636203895,
      "loss": 1.0138,
      "step": 10571
    },
    {
      "epoch": 0.9095758410049041,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009208432631452614,
      "loss": 1.1302,
      "step": 10572
    },
    {
      "epoch": 0.9096618773122258,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009208286614392487,
      "loss": 1.1199,
      "step": 10573
    },
    {
      "epoch": 0.9097479136195474,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009208140585023938,
      "loss": 1.5681,
      "step": 10574
    },
    {
      "epoch": 0.9098339499268692,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009207994543347399,
      "loss": 0.7962,
      "step": 10575
    },
    {
      "epoch": 0.9099199862341908,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009207848489363293,
      "loss": 0.7774,
      "step": 10576
    },
    {
      "epoch": 0.9100060225415125,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009207702423072048,
      "loss": 0.6096,
      "step": 10577
    },
    {
      "epoch": 0.9100920588488343,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009207556344474093,
      "loss": 1.1777,
      "step": 10578
    },
    {
      "epoch": 0.9101780951561559,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009207410253569854,
      "loss": 1.4482,
      "step": 10579
    },
    {
      "epoch": 0.9102641314634776,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009207264150359757,
      "loss": 1.2246,
      "step": 10580
    },
    {
      "epoch": 0.9103501677707992,
      "grad_norm": 1.109375,
      "learning_rate": 0.000920711803484423,
      "loss": 1.048,
      "step": 10581
    },
    {
      "epoch": 0.910436204078121,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009206971907023705,
      "loss": 1.337,
      "step": 10582
    },
    {
      "epoch": 0.9105222403854426,
      "grad_norm": 1.25,
      "learning_rate": 0.0009206825766898601,
      "loss": 1.152,
      "step": 10583
    },
    {
      "epoch": 0.9106082766927643,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009206679614469351,
      "loss": 1.5588,
      "step": 10584
    },
    {
      "epoch": 0.9106943130000861,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009206533449736383,
      "loss": 2.2864,
      "step": 10585
    },
    {
      "epoch": 0.9107803493074077,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009206387272700121,
      "loss": 1.5008,
      "step": 10586
    },
    {
      "epoch": 0.9108663856147294,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009206241083360995,
      "loss": 1.0995,
      "step": 10587
    },
    {
      "epoch": 0.9109524219220511,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009206094881719432,
      "loss": 1.0307,
      "step": 10588
    },
    {
      "epoch": 0.9110384582293728,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009205948667775859,
      "loss": 1.3867,
      "step": 10589
    },
    {
      "epoch": 0.9111244945366945,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009205802441530704,
      "loss": 1.4259,
      "step": 10590
    },
    {
      "epoch": 0.9112105308440162,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009205656202984396,
      "loss": 1.0521,
      "step": 10591
    },
    {
      "epoch": 0.9112965671513379,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009205509952137361,
      "loss": 1.4362,
      "step": 10592
    },
    {
      "epoch": 0.9113826034586595,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009205363688990027,
      "loss": 1.6253,
      "step": 10593
    },
    {
      "epoch": 0.9114686397659812,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009205217413542822,
      "loss": 1.1147,
      "step": 10594
    },
    {
      "epoch": 0.911554676073303,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009205071125796174,
      "loss": 0.8184,
      "step": 10595
    },
    {
      "epoch": 0.9116407123806246,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009204924825750512,
      "loss": 1.7649,
      "step": 10596
    },
    {
      "epoch": 0.9117267486879463,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009204778513406261,
      "loss": 1.3849,
      "step": 10597
    },
    {
      "epoch": 0.911812784995268,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009204632188763852,
      "loss": 1.0534,
      "step": 10598
    },
    {
      "epoch": 0.9118988213025897,
      "grad_norm": 1.546875,
      "learning_rate": 0.000920448585182371,
      "loss": 1.3803,
      "step": 10599
    },
    {
      "epoch": 0.9119848576099114,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009204339502586267,
      "loss": 1.0168,
      "step": 10600
    },
    {
      "epoch": 0.9120708939172331,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009204193141051947,
      "loss": 1.3934,
      "step": 10601
    },
    {
      "epoch": 0.9121569302245548,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009204046767221179,
      "loss": 1.3341,
      "step": 10602
    },
    {
      "epoch": 0.9122429665318764,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009203900381094393,
      "loss": 1.4674,
      "step": 10603
    },
    {
      "epoch": 0.9123290028391982,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009203753982672016,
      "loss": 1.1404,
      "step": 10604
    },
    {
      "epoch": 0.9124150391465198,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009203607571954476,
      "loss": 1.4747,
      "step": 10605
    },
    {
      "epoch": 0.9125010754538415,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009203461148942202,
      "loss": 1.8261,
      "step": 10606
    },
    {
      "epoch": 0.9125871117611632,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000920331471363562,
      "loss": 0.9805,
      "step": 10607
    },
    {
      "epoch": 0.9126731480684849,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009203168266035161,
      "loss": 1.3178,
      "step": 10608
    },
    {
      "epoch": 0.9127591843758066,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009203021806141252,
      "loss": 1.5089,
      "step": 10609
    },
    {
      "epoch": 0.9128452206831282,
      "grad_norm": 2.0,
      "learning_rate": 0.0009202875333954324,
      "loss": 1.7334,
      "step": 10610
    },
    {
      "epoch": 0.91293125699045,
      "grad_norm": 1.546875,
      "learning_rate": 0.00092027288494748,
      "loss": 1.6121,
      "step": 10611
    },
    {
      "epoch": 0.9130172932977717,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009202582352703112,
      "loss": 1.2739,
      "step": 10612
    },
    {
      "epoch": 0.9131033296050933,
      "grad_norm": 1.28125,
      "learning_rate": 0.000920243584363969,
      "loss": 1.6226,
      "step": 10613
    },
    {
      "epoch": 0.9131893659124151,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009202289322284958,
      "loss": 1.2017,
      "step": 10614
    },
    {
      "epoch": 0.9132754022197367,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009202142788639348,
      "loss": 1.4292,
      "step": 10615
    },
    {
      "epoch": 0.9133614385270584,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009201996242703288,
      "loss": 1.2248,
      "step": 10616
    },
    {
      "epoch": 0.9134474748343802,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009201849684477206,
      "loss": 0.9539,
      "step": 10617
    },
    {
      "epoch": 0.9135335111417018,
      "grad_norm": 1.828125,
      "learning_rate": 0.000920170311396153,
      "loss": 1.3052,
      "step": 10618
    },
    {
      "epoch": 0.9136195474490235,
      "grad_norm": 1.171875,
      "learning_rate": 0.000920155653115669,
      "loss": 1.368,
      "step": 10619
    },
    {
      "epoch": 0.9137055837563451,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009201409936063114,
      "loss": 1.0228,
      "step": 10620
    },
    {
      "epoch": 0.9137916200636669,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009201263328681233,
      "loss": 0.7002,
      "step": 10621
    },
    {
      "epoch": 0.9138776563709886,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009201116709011472,
      "loss": 1.8146,
      "step": 10622
    },
    {
      "epoch": 0.9139636926783102,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009200970077054263,
      "loss": 1.1671,
      "step": 10623
    },
    {
      "epoch": 0.914049728985632,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009200823432810033,
      "loss": 1.6631,
      "step": 10624
    },
    {
      "epoch": 0.9141357652929536,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009200676776279211,
      "loss": 1.5911,
      "step": 10625
    },
    {
      "epoch": 0.9142218016002753,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009200530107462226,
      "loss": 2.3613,
      "step": 10626
    },
    {
      "epoch": 0.914307837907597,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009200383426359508,
      "loss": 1.4198,
      "step": 10627
    },
    {
      "epoch": 0.9143938742149187,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009200236732971487,
      "loss": 1.6848,
      "step": 10628
    },
    {
      "epoch": 0.9144799105222404,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009200090027298588,
      "loss": 1.7498,
      "step": 10629
    },
    {
      "epoch": 0.914565946829562,
      "grad_norm": 1.5,
      "learning_rate": 0.0009199943309341243,
      "loss": 1.3306,
      "step": 10630
    },
    {
      "epoch": 0.9146519831368838,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009199796579099881,
      "loss": 1.7411,
      "step": 10631
    },
    {
      "epoch": 0.9147380194442054,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009199649836574932,
      "loss": 1.1468,
      "step": 10632
    },
    {
      "epoch": 0.9148240557515271,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009199503081766822,
      "loss": 1.6781,
      "step": 10633
    },
    {
      "epoch": 0.9149100920588489,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009199356314675984,
      "loss": 1.4455,
      "step": 10634
    },
    {
      "epoch": 0.9149961283661705,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009199209535302846,
      "loss": 1.6124,
      "step": 10635
    },
    {
      "epoch": 0.9150821646734922,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009199062743647835,
      "loss": 1.0484,
      "step": 10636
    },
    {
      "epoch": 0.9151682009808139,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009198915939711383,
      "loss": 1.0579,
      "step": 10637
    },
    {
      "epoch": 0.9152542372881356,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009198769123493917,
      "loss": 1.0584,
      "step": 10638
    },
    {
      "epoch": 0.9153402735954573,
      "grad_norm": 1.40625,
      "learning_rate": 0.000919862229499587,
      "loss": 1.8585,
      "step": 10639
    },
    {
      "epoch": 0.915426309902779,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009198475454217669,
      "loss": 0.877,
      "step": 10640
    },
    {
      "epoch": 0.9155123462101007,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009198328601159742,
      "loss": 1.2192,
      "step": 10641
    },
    {
      "epoch": 0.9155983825174223,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009198181735822523,
      "loss": 1.3007,
      "step": 10642
    },
    {
      "epoch": 0.915684418824744,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009198034858206438,
      "loss": 1.0978,
      "step": 10643
    },
    {
      "epoch": 0.9157704551320658,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009197887968311917,
      "loss": 1.2202,
      "step": 10644
    },
    {
      "epoch": 0.9158564914393874,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000919774106613939,
      "loss": 1.548,
      "step": 10645
    },
    {
      "epoch": 0.9159425277467091,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009197594151689287,
      "loss": 1.8671,
      "step": 10646
    },
    {
      "epoch": 0.9160285640540308,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009197447224962038,
      "loss": 1.5052,
      "step": 10647
    },
    {
      "epoch": 0.9161146003613525,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009197300285958073,
      "loss": 2.3586,
      "step": 10648
    },
    {
      "epoch": 0.9162006366686741,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009197153334677819,
      "loss": 1.9825,
      "step": 10649
    },
    {
      "epoch": 0.9162866729759959,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009197006371121708,
      "loss": 2.0927,
      "step": 10650
    },
    {
      "epoch": 0.9163727092833176,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009196859395290171,
      "loss": 1.8274,
      "step": 10651
    },
    {
      "epoch": 0.9164587455906392,
      "grad_norm": 1.375,
      "learning_rate": 0.0009196712407183636,
      "loss": 2.173,
      "step": 10652
    },
    {
      "epoch": 0.916544781897961,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009196565406802533,
      "loss": 1.3843,
      "step": 10653
    },
    {
      "epoch": 0.9166308182052826,
      "grad_norm": 1.25,
      "learning_rate": 0.0009196418394147293,
      "loss": 1.8368,
      "step": 10654
    },
    {
      "epoch": 0.9167168545126043,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009196271369218344,
      "loss": 0.6326,
      "step": 10655
    },
    {
      "epoch": 0.916802890819926,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009196124332016118,
      "loss": 1.5132,
      "step": 10656
    },
    {
      "epoch": 0.9168889271272477,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009195977282541046,
      "loss": 0.5608,
      "step": 10657
    },
    {
      "epoch": 0.9169749634345694,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009195830220793556,
      "loss": 1.0157,
      "step": 10658
    },
    {
      "epoch": 0.917060999741891,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009195683146774077,
      "loss": 1.7943,
      "step": 10659
    },
    {
      "epoch": 0.9171470360492128,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009195536060483041,
      "loss": 1.3845,
      "step": 10660
    },
    {
      "epoch": 0.9172330723565345,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009195388961920879,
      "loss": 1.4684,
      "step": 10661
    },
    {
      "epoch": 0.9173191086638561,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009195241851088021,
      "loss": 2.0533,
      "step": 10662
    },
    {
      "epoch": 0.9174051449711779,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009195094727984894,
      "loss": 1.5237,
      "step": 10663
    },
    {
      "epoch": 0.9174911812784995,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009194947592611933,
      "loss": 1.6358,
      "step": 10664
    },
    {
      "epoch": 0.9175772175858212,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009194800444969565,
      "loss": 1.0882,
      "step": 10665
    },
    {
      "epoch": 0.917663253893143,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009194653285058223,
      "loss": 1.7867,
      "step": 10666
    },
    {
      "epoch": 0.9177492902004646,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009194506112878334,
      "loss": 1.4989,
      "step": 10667
    },
    {
      "epoch": 0.9178353265077863,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009194358928430331,
      "loss": 0.9745,
      "step": 10668
    },
    {
      "epoch": 0.9179213628151079,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009194211731714646,
      "loss": 1.6065,
      "step": 10669
    },
    {
      "epoch": 0.9180073991224297,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009194064522731704,
      "loss": 2.2389,
      "step": 10670
    },
    {
      "epoch": 0.9180934354297513,
      "grad_norm": 1.0,
      "learning_rate": 0.0009193917301481942,
      "loss": 1.0478,
      "step": 10671
    },
    {
      "epoch": 0.918179471737073,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009193770067965786,
      "loss": 1.1557,
      "step": 10672
    },
    {
      "epoch": 0.9182655080443948,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009193622822183669,
      "loss": 1.386,
      "step": 10673
    },
    {
      "epoch": 0.9183515443517164,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000919347556413602,
      "loss": 0.8004,
      "step": 10674
    },
    {
      "epoch": 0.9184375806590381,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009193328293823271,
      "loss": 1.0864,
      "step": 10675
    },
    {
      "epoch": 0.9185236169663598,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009193181011245853,
      "loss": 1.1718,
      "step": 10676
    },
    {
      "epoch": 0.9186096532736815,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009193033716404194,
      "loss": 2.1442,
      "step": 10677
    },
    {
      "epoch": 0.9186956895810032,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009192886409298729,
      "loss": 1.0304,
      "step": 10678
    },
    {
      "epoch": 0.9187817258883249,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009192739089929885,
      "loss": 1.2493,
      "step": 10679
    },
    {
      "epoch": 0.9188677621956466,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009192591758298095,
      "loss": 0.9135,
      "step": 10680
    },
    {
      "epoch": 0.9189537985029682,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009192444414403791,
      "loss": 1.9168,
      "step": 10681
    },
    {
      "epoch": 0.91903983481029,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00091922970582474,
      "loss": 0.6847,
      "step": 10682
    },
    {
      "epoch": 0.9191258711176117,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009192149689829358,
      "loss": 0.8253,
      "step": 10683
    },
    {
      "epoch": 0.9192119074249333,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0009192002309150091,
      "loss": 0.6046,
      "step": 10684
    },
    {
      "epoch": 0.919297943732255,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009191854916210036,
      "loss": 1.2685,
      "step": 10685
    },
    {
      "epoch": 0.9193839800395767,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009191707511009617,
      "loss": 1.1673,
      "step": 10686
    },
    {
      "epoch": 0.9194700163468984,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009191560093549272,
      "loss": 1.3104,
      "step": 10687
    },
    {
      "epoch": 0.9195560526542201,
      "grad_norm": 2.046875,
      "learning_rate": 0.0009191412663829426,
      "loss": 1.3717,
      "step": 10688
    },
    {
      "epoch": 0.9196420889615418,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009191265221850515,
      "loss": 2.1353,
      "step": 10689
    },
    {
      "epoch": 0.9197281252688635,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009191117767612969,
      "loss": 1.5267,
      "step": 10690
    },
    {
      "epoch": 0.9198141615761851,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009190970301117216,
      "loss": 1.4793,
      "step": 10691
    },
    {
      "epoch": 0.9199001978835069,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009190822822363691,
      "loss": 0.967,
      "step": 10692
    },
    {
      "epoch": 0.9199862341908285,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009190675331352824,
      "loss": 1.1738,
      "step": 10693
    },
    {
      "epoch": 0.9200722704981502,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009190527828085049,
      "loss": 1.1085,
      "step": 10694
    },
    {
      "epoch": 0.920158306805472,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009190380312560793,
      "loss": 1.4029,
      "step": 10695
    },
    {
      "epoch": 0.9202443431127936,
      "grad_norm": 1.15625,
      "learning_rate": 0.000919023278478049,
      "loss": 1.7071,
      "step": 10696
    },
    {
      "epoch": 0.9203303794201153,
      "grad_norm": 2.859375,
      "learning_rate": 0.0009190085244744572,
      "loss": 2.208,
      "step": 10697
    },
    {
      "epoch": 0.9204164157274369,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009189937692453468,
      "loss": 1.8089,
      "step": 10698
    },
    {
      "epoch": 0.9205024520347587,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009189790127907613,
      "loss": 1.8951,
      "step": 10699
    },
    {
      "epoch": 0.9205884883420804,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009189642551107436,
      "loss": 0.7882,
      "step": 10700
    },
    {
      "epoch": 0.920674524649402,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009189494962053369,
      "loss": 0.9473,
      "step": 10701
    },
    {
      "epoch": 0.9207605609567238,
      "grad_norm": 2.515625,
      "learning_rate": 0.0009189347360745844,
      "loss": 1.5003,
      "step": 10702
    },
    {
      "epoch": 0.9208465972640454,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009189199747185293,
      "loss": 1.0837,
      "step": 10703
    },
    {
      "epoch": 0.9209326335713671,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009189052121372146,
      "loss": 1.9387,
      "step": 10704
    },
    {
      "epoch": 0.9210186698786889,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009188904483306838,
      "loss": 1.5514,
      "step": 10705
    },
    {
      "epoch": 0.9211047061860105,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009188756832989799,
      "loss": 1.8471,
      "step": 10706
    },
    {
      "epoch": 0.9211907424933322,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009188609170421461,
      "loss": 1.1645,
      "step": 10707
    },
    {
      "epoch": 0.9212767788006538,
      "grad_norm": 1.125,
      "learning_rate": 0.0009188461495602255,
      "loss": 1.1454,
      "step": 10708
    },
    {
      "epoch": 0.9213628151079756,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009188313808532613,
      "loss": 0.8947,
      "step": 10709
    },
    {
      "epoch": 0.9214488514152973,
      "grad_norm": 0.828125,
      "learning_rate": 0.000918816610921297,
      "loss": 0.6657,
      "step": 10710
    },
    {
      "epoch": 0.9215348877226189,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009188018397643755,
      "loss": 1.559,
      "step": 10711
    },
    {
      "epoch": 0.9216209240299407,
      "grad_norm": 1.640625,
      "learning_rate": 0.00091878706738254,
      "loss": 1.3349,
      "step": 10712
    },
    {
      "epoch": 0.9217069603372623,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009187722937758338,
      "loss": 1.4826,
      "step": 10713
    },
    {
      "epoch": 0.921792996644584,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009187575189443,
      "loss": 1.8164,
      "step": 10714
    },
    {
      "epoch": 0.9218790329519057,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000918742742887982,
      "loss": 0.8964,
      "step": 10715
    },
    {
      "epoch": 0.9219650692592274,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009187279656069229,
      "loss": 1.9216,
      "step": 10716
    },
    {
      "epoch": 0.9220511055665491,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000918713187101166,
      "loss": 1.9348,
      "step": 10717
    },
    {
      "epoch": 0.9221371418738707,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009186984073707544,
      "loss": 0.9057,
      "step": 10718
    },
    {
      "epoch": 0.9222231781811925,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009186836264157315,
      "loss": 1.2431,
      "step": 10719
    },
    {
      "epoch": 0.9223092144885141,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009186688442361403,
      "loss": 1.9163,
      "step": 10720
    },
    {
      "epoch": 0.9223952507958358,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009186540608320242,
      "loss": 0.6008,
      "step": 10721
    },
    {
      "epoch": 0.9224812871031576,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009186392762034264,
      "loss": 1.238,
      "step": 10722
    },
    {
      "epoch": 0.9225673234104792,
      "grad_norm": 1.59375,
      "learning_rate": 0.00091862449035039,
      "loss": 1.8399,
      "step": 10723
    },
    {
      "epoch": 0.9226533597178009,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009186097032729586,
      "loss": 1.9723,
      "step": 10724
    },
    {
      "epoch": 0.9227393960251226,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009185949149711751,
      "loss": 1.605,
      "step": 10725
    },
    {
      "epoch": 0.9228254323324443,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009185801254450829,
      "loss": 1.2969,
      "step": 10726
    },
    {
      "epoch": 0.922911468639766,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009185653346947253,
      "loss": 1.3046,
      "step": 10727
    },
    {
      "epoch": 0.9229975049470877,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009185505427201454,
      "loss": 0.893,
      "step": 10728
    },
    {
      "epoch": 0.9230835412544094,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009185357495213866,
      "loss": 0.8963,
      "step": 10729
    },
    {
      "epoch": 0.923169577561731,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009185209550984922,
      "loss": 1.2344,
      "step": 10730
    },
    {
      "epoch": 0.9232556138690527,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009185061594515054,
      "loss": 0.9788,
      "step": 10731
    },
    {
      "epoch": 0.9233416501763745,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009184913625804694,
      "loss": 2.1625,
      "step": 10732
    },
    {
      "epoch": 0.9234276864836961,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009184765644854276,
      "loss": 1.4369,
      "step": 10733
    },
    {
      "epoch": 0.9235137227910178,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009184617651664234,
      "loss": 1.255,
      "step": 10734
    },
    {
      "epoch": 0.9235997590983395,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009184469646234996,
      "loss": 2.3002,
      "step": 10735
    },
    {
      "epoch": 0.9236857954056612,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009184321628567001,
      "loss": 1.4434,
      "step": 10736
    },
    {
      "epoch": 0.9237718317129828,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009184173598660676,
      "loss": 1.4869,
      "step": 10737
    },
    {
      "epoch": 0.9238578680203046,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000918402555651646,
      "loss": 0.8566,
      "step": 10738
    },
    {
      "epoch": 0.9239439043276263,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000918387750213478,
      "loss": 1.6108,
      "step": 10739
    },
    {
      "epoch": 0.9240299406349479,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009183729435516075,
      "loss": 1.3865,
      "step": 10740
    },
    {
      "epoch": 0.9241159769422697,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009183581356660774,
      "loss": 0.9222,
      "step": 10741
    },
    {
      "epoch": 0.9242020132495913,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009183433265569312,
      "loss": 1.2737,
      "step": 10742
    },
    {
      "epoch": 0.924288049556913,
      "grad_norm": 1.4375,
      "learning_rate": 0.000918328516224212,
      "loss": 2.1048,
      "step": 10743
    },
    {
      "epoch": 0.9243740858642348,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009183137046679632,
      "loss": 1.6578,
      "step": 10744
    },
    {
      "epoch": 0.9244601221715564,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009182988918882283,
      "loss": 1.5405,
      "step": 10745
    },
    {
      "epoch": 0.9245461584788781,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009182840778850505,
      "loss": 1.1682,
      "step": 10746
    },
    {
      "epoch": 0.9246321947861997,
      "grad_norm": 1.625,
      "learning_rate": 0.000918269262658473,
      "loss": 1.0109,
      "step": 10747
    },
    {
      "epoch": 0.9247182310935215,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009182544462085394,
      "loss": 1.9474,
      "step": 10748
    },
    {
      "epoch": 0.9248042674008432,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009182396285352929,
      "loss": 2.23,
      "step": 10749
    },
    {
      "epoch": 0.9248903037081648,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009182248096387767,
      "loss": 1.7156,
      "step": 10750
    },
    {
      "epoch": 0.9249763400154866,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009182099895190343,
      "loss": 0.7939,
      "step": 10751
    },
    {
      "epoch": 0.9250623763228082,
      "grad_norm": 1.25,
      "learning_rate": 0.0009181951681761091,
      "loss": 2.057,
      "step": 10752
    },
    {
      "epoch": 0.9251484126301299,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009181803456100444,
      "loss": 1.9178,
      "step": 10753
    },
    {
      "epoch": 0.9252344489374517,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009181655218208834,
      "loss": 1.5946,
      "step": 10754
    },
    {
      "epoch": 0.9253204852447733,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009181506968086695,
      "loss": 1.912,
      "step": 10755
    },
    {
      "epoch": 0.925406521552095,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009181358705734463,
      "loss": 1.7894,
      "step": 10756
    },
    {
      "epoch": 0.9254925578594166,
      "grad_norm": 1.71875,
      "learning_rate": 0.000918121043115257,
      "loss": 2.2015,
      "step": 10757
    },
    {
      "epoch": 0.9255785941667384,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009181062144341448,
      "loss": 1.1209,
      "step": 10758
    },
    {
      "epoch": 0.92566463047406,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009180913845301535,
      "loss": 1.5754,
      "step": 10759
    },
    {
      "epoch": 0.9257506667813817,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000918076553403326,
      "loss": 1.7494,
      "step": 10760
    },
    {
      "epoch": 0.9258367030887035,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009180617210537061,
      "loss": 1.7759,
      "step": 10761
    },
    {
      "epoch": 0.9259227393960251,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009180468874813368,
      "loss": 1.2788,
      "step": 10762
    },
    {
      "epoch": 0.9260087757033468,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009180320526862618,
      "loss": 1.378,
      "step": 10763
    },
    {
      "epoch": 0.9260948120106685,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009180172166685241,
      "loss": 1.2463,
      "step": 10764
    },
    {
      "epoch": 0.9261808483179902,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009180023794281676,
      "loss": 0.8571,
      "step": 10765
    },
    {
      "epoch": 0.9262668846253119,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009179875409652352,
      "loss": 1.6392,
      "step": 10766
    },
    {
      "epoch": 0.9263529209326336,
      "grad_norm": 1.5,
      "learning_rate": 0.0009179727012797708,
      "loss": 1.4381,
      "step": 10767
    },
    {
      "epoch": 0.9264389572399553,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009179578603718173,
      "loss": 1.0975,
      "step": 10768
    },
    {
      "epoch": 0.9265249935472769,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009179430182414184,
      "loss": 1.3633,
      "step": 10769
    },
    {
      "epoch": 0.9266110298545986,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009179281748886177,
      "loss": 1.3924,
      "step": 10770
    },
    {
      "epoch": 0.9266970661619204,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009179133303134581,
      "loss": 1.2775,
      "step": 10771
    },
    {
      "epoch": 0.926783102469242,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009178984845159834,
      "loss": 0.8819,
      "step": 10772
    },
    {
      "epoch": 0.9268691387765637,
      "grad_norm": 1.78125,
      "learning_rate": 0.0009178836374962368,
      "loss": 1.6825,
      "step": 10773
    },
    {
      "epoch": 0.9269551750838854,
      "grad_norm": 1.390625,
      "learning_rate": 0.000917868789254262,
      "loss": 2.2159,
      "step": 10774
    },
    {
      "epoch": 0.9270412113912071,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009178539397901021,
      "loss": 1.0231,
      "step": 10775
    },
    {
      "epoch": 0.9271272476985288,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009178390891038009,
      "loss": 0.7769,
      "step": 10776
    },
    {
      "epoch": 0.9272132840058505,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0009178242371954015,
      "loss": 1.0488,
      "step": 10777
    },
    {
      "epoch": 0.9272993203131722,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009178093840649474,
      "loss": 1.4777,
      "step": 10778
    },
    {
      "epoch": 0.9273853566204938,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009177945297124821,
      "loss": 1.5574,
      "step": 10779
    },
    {
      "epoch": 0.9274713929278156,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009177796741380492,
      "loss": 1.687,
      "step": 10780
    },
    {
      "epoch": 0.9275574292351372,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009177648173416919,
      "loss": 1.8851,
      "step": 10781
    },
    {
      "epoch": 0.9276434655424589,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009177499593234538,
      "loss": 1.331,
      "step": 10782
    },
    {
      "epoch": 0.9277295018497806,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009177351000833784,
      "loss": 1.5876,
      "step": 10783
    },
    {
      "epoch": 0.9278155381571023,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000917720239621509,
      "loss": 1.3722,
      "step": 10784
    },
    {
      "epoch": 0.927901574464424,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009177053779378891,
      "loss": 1.4425,
      "step": 10785
    },
    {
      "epoch": 0.9279876107717456,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009176905150325623,
      "loss": 2.0007,
      "step": 10786
    },
    {
      "epoch": 0.9280736470790674,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009176756509055719,
      "loss": 0.7391,
      "step": 10787
    },
    {
      "epoch": 0.9281596833863891,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009176607855569614,
      "loss": 2.0341,
      "step": 10788
    },
    {
      "epoch": 0.9282457196937107,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009176459189867745,
      "loss": 0.9192,
      "step": 10789
    },
    {
      "epoch": 0.9283317560010325,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009176310511950545,
      "loss": 1.0043,
      "step": 10790
    },
    {
      "epoch": 0.9284177923083541,
      "grad_norm": 2.03125,
      "learning_rate": 0.0009176161821818447,
      "loss": 1.6153,
      "step": 10791
    },
    {
      "epoch": 0.9285038286156758,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000917601311947189,
      "loss": 2.3498,
      "step": 10792
    },
    {
      "epoch": 0.9285898649229976,
      "grad_norm": 1.25,
      "learning_rate": 0.0009175864404911306,
      "loss": 1.5076,
      "step": 10793
    },
    {
      "epoch": 0.9286759012303192,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009175715678137131,
      "loss": 1.5371,
      "step": 10794
    },
    {
      "epoch": 0.9287619375376409,
      "grad_norm": 1.28125,
      "learning_rate": 0.00091755669391498,
      "loss": 1.5816,
      "step": 10795
    },
    {
      "epoch": 0.9288479738449625,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009175418187949747,
      "loss": 1.2933,
      "step": 10796
    },
    {
      "epoch": 0.9289340101522843,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009175269424537408,
      "loss": 1.0577,
      "step": 10797
    },
    {
      "epoch": 0.929020046459606,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009175120648913218,
      "loss": 1.28,
      "step": 10798
    },
    {
      "epoch": 0.9291060827669276,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009174971861077612,
      "loss": 1.4706,
      "step": 10799
    },
    {
      "epoch": 0.9291921190742494,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009174823061031026,
      "loss": 1.527,
      "step": 10800
    },
    {
      "epoch": 0.929278155381571,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009174674248773894,
      "loss": 1.6325,
      "step": 10801
    },
    {
      "epoch": 0.9293641916888927,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009174525424306652,
      "loss": 0.8345,
      "step": 10802
    },
    {
      "epoch": 0.9294502279962144,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009174376587629735,
      "loss": 1.1323,
      "step": 10803
    },
    {
      "epoch": 0.9295362643035361,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009174227738743579,
      "loss": 1.3024,
      "step": 10804
    },
    {
      "epoch": 0.9296223006108578,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009174078877648617,
      "loss": 1.1328,
      "step": 10805
    },
    {
      "epoch": 0.9297083369181794,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009173930004345288,
      "loss": 1.4795,
      "step": 10806
    },
    {
      "epoch": 0.9297943732255012,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009173781118834023,
      "loss": 1.0232,
      "step": 10807
    },
    {
      "epoch": 0.9298804095328228,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009173632221115263,
      "loss": 0.996,
      "step": 10808
    },
    {
      "epoch": 0.9299664458401445,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009173483311189439,
      "loss": 1.3262,
      "step": 10809
    },
    {
      "epoch": 0.9300524821474663,
      "grad_norm": 1.875,
      "learning_rate": 0.0009173334389056989,
      "loss": 1.5717,
      "step": 10810
    },
    {
      "epoch": 0.9301385184547879,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009173185454718345,
      "loss": 1.2941,
      "step": 10811
    },
    {
      "epoch": 0.9302245547621096,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009173036508173946,
      "loss": 0.7983,
      "step": 10812
    },
    {
      "epoch": 0.9303105910694313,
      "grad_norm": 1.234375,
      "learning_rate": 0.000917288754942423,
      "loss": 1.1084,
      "step": 10813
    },
    {
      "epoch": 0.930396627376753,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0009172738578469627,
      "loss": 0.5672,
      "step": 10814
    },
    {
      "epoch": 0.9304826636840747,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009172589595310574,
      "loss": 1.2672,
      "step": 10815
    },
    {
      "epoch": 0.9305686999913964,
      "grad_norm": 1.09375,
      "learning_rate": 0.000917244059994751,
      "loss": 1.0341,
      "step": 10816
    },
    {
      "epoch": 0.9306547362987181,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009172291592380868,
      "loss": 0.8069,
      "step": 10817
    },
    {
      "epoch": 0.9307407726060397,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009172142572611084,
      "loss": 1.0377,
      "step": 10818
    },
    {
      "epoch": 0.9308268089133614,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009171993540638597,
      "loss": 1.3561,
      "step": 10819
    },
    {
      "epoch": 0.9309128452206832,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009171844496463837,
      "loss": 1.5303,
      "step": 10820
    },
    {
      "epoch": 0.9309988815280048,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009171695440087245,
      "loss": 1.6175,
      "step": 10821
    },
    {
      "epoch": 0.9310849178353265,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009171546371509254,
      "loss": 1.1838,
      "step": 10822
    },
    {
      "epoch": 0.9311709541426482,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009171397290730303,
      "loss": 1.7586,
      "step": 10823
    },
    {
      "epoch": 0.9312569904499699,
      "grad_norm": 6.8125,
      "learning_rate": 0.0009171248197750824,
      "loss": 1.3251,
      "step": 10824
    },
    {
      "epoch": 0.9313430267572915,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009171099092571257,
      "loss": 2.1074,
      "step": 10825
    },
    {
      "epoch": 0.9314290630646133,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009170949975192035,
      "loss": 0.9543,
      "step": 10826
    },
    {
      "epoch": 0.931515099371935,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009170800845613597,
      "loss": 1.0765,
      "step": 10827
    },
    {
      "epoch": 0.9316011356792566,
      "grad_norm": 3.09375,
      "learning_rate": 0.0009170651703836376,
      "loss": 1.5885,
      "step": 10828
    },
    {
      "epoch": 0.9316871719865784,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009170502549860813,
      "loss": 0.9808,
      "step": 10829
    },
    {
      "epoch": 0.9317732082939,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009170353383687337,
      "loss": 1.5517,
      "step": 10830
    },
    {
      "epoch": 0.9318592446012217,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009170204205316393,
      "loss": 1.1752,
      "step": 10831
    },
    {
      "epoch": 0.9319452809085434,
      "grad_norm": 1.40625,
      "learning_rate": 0.000917005501474841,
      "loss": 1.846,
      "step": 10832
    },
    {
      "epoch": 0.9320313172158651,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009169905811983827,
      "loss": 1.4134,
      "step": 10833
    },
    {
      "epoch": 0.9321173535231868,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009169756597023081,
      "loss": 1.024,
      "step": 10834
    },
    {
      "epoch": 0.9322033898305084,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009169607369866608,
      "loss": 0.9319,
      "step": 10835
    },
    {
      "epoch": 0.9322894261378302,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009169458130514844,
      "loss": 0.9771,
      "step": 10836
    },
    {
      "epoch": 0.9323754624451519,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009169308878968225,
      "loss": 1.7416,
      "step": 10837
    },
    {
      "epoch": 0.9324614987524735,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000916915961522719,
      "loss": 1.0018,
      "step": 10838
    },
    {
      "epoch": 0.9325475350597953,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009169010339292174,
      "loss": 1.0928,
      "step": 10839
    },
    {
      "epoch": 0.9326335713671169,
      "grad_norm": 1.625,
      "learning_rate": 0.0009168861051163613,
      "loss": 1.95,
      "step": 10840
    },
    {
      "epoch": 0.9327196076744386,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009168711750841945,
      "loss": 0.6358,
      "step": 10841
    },
    {
      "epoch": 0.9328056439817604,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009168562438327606,
      "loss": 1.1497,
      "step": 10842
    },
    {
      "epoch": 0.932891680289082,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000916841311362103,
      "loss": 1.5915,
      "step": 10843
    },
    {
      "epoch": 0.9329777165964037,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009168263776722659,
      "loss": 0.7901,
      "step": 10844
    },
    {
      "epoch": 0.9330637529037253,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009168114427632926,
      "loss": 1.5728,
      "step": 10845
    },
    {
      "epoch": 0.9331497892110471,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009167965066352269,
      "loss": 1.3114,
      "step": 10846
    },
    {
      "epoch": 0.9332358255183687,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009167815692881126,
      "loss": 0.953,
      "step": 10847
    },
    {
      "epoch": 0.9333218618256904,
      "grad_norm": 2.28125,
      "learning_rate": 0.0009167666307219931,
      "loss": 0.9761,
      "step": 10848
    },
    {
      "epoch": 0.9334078981330122,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009167516909369124,
      "loss": 1.6525,
      "step": 10849
    },
    {
      "epoch": 0.9334939344403338,
      "grad_norm": 1.046875,
      "learning_rate": 0.000916736749932914,
      "loss": 0.9762,
      "step": 10850
    },
    {
      "epoch": 0.9335799707476555,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009167218077100417,
      "loss": 0.4639,
      "step": 10851
    },
    {
      "epoch": 0.9336660070549772,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009167068642683392,
      "loss": 1.5782,
      "step": 10852
    },
    {
      "epoch": 0.9337520433622989,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00091669191960785,
      "loss": 0.9983,
      "step": 10853
    },
    {
      "epoch": 0.9338380796696206,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000916676973728618,
      "loss": 1.3752,
      "step": 10854
    },
    {
      "epoch": 0.9339241159769422,
      "grad_norm": 1.28125,
      "learning_rate": 0.000916662026630687,
      "loss": 1.8866,
      "step": 10855
    },
    {
      "epoch": 0.934010152284264,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009166470783141006,
      "loss": 1.8138,
      "step": 10856
    },
    {
      "epoch": 0.9340961885915856,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009166321287789024,
      "loss": 1.1966,
      "step": 10857
    },
    {
      "epoch": 0.9341822248989073,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009166171780251364,
      "loss": 0.5576,
      "step": 10858
    },
    {
      "epoch": 0.9342682612062291,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009166022260528461,
      "loss": 1.3963,
      "step": 10859
    },
    {
      "epoch": 0.9343542975135507,
      "grad_norm": 7.1875,
      "learning_rate": 0.0009165872728620754,
      "loss": 1.4782,
      "step": 10860
    },
    {
      "epoch": 0.9344403338208724,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009165723184528679,
      "loss": 1.2764,
      "step": 10861
    },
    {
      "epoch": 0.9345263701281941,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009165573628252674,
      "loss": 1.331,
      "step": 10862
    },
    {
      "epoch": 0.9346124064355158,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009165424059793174,
      "loss": 1.8703,
      "step": 10863
    },
    {
      "epoch": 0.9346984427428375,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009165274479150622,
      "loss": 0.899,
      "step": 10864
    },
    {
      "epoch": 0.9347844790501592,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009165124886325451,
      "loss": 1.2833,
      "step": 10865
    },
    {
      "epoch": 0.9348705153574809,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009164975281318099,
      "loss": 0.7515,
      "step": 10866
    },
    {
      "epoch": 0.9349565516648025,
      "grad_norm": 52.25,
      "learning_rate": 0.0009164825664129007,
      "loss": 1.2597,
      "step": 10867
    },
    {
      "epoch": 0.9350425879721243,
      "grad_norm": 10.5,
      "learning_rate": 0.0009164676034758608,
      "loss": 1.1798,
      "step": 10868
    },
    {
      "epoch": 0.9351286242794459,
      "grad_norm": 1.203125,
      "learning_rate": 0.000916452639320734,
      "loss": 1.3805,
      "step": 10869
    },
    {
      "epoch": 0.9352146605867676,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009164376739475645,
      "loss": 1.6304,
      "step": 10870
    },
    {
      "epoch": 0.9353006968940893,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009164227073563957,
      "loss": 1.3215,
      "step": 10871
    },
    {
      "epoch": 0.935386733201411,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009164077395472715,
      "loss": 1.3736,
      "step": 10872
    },
    {
      "epoch": 0.9354727695087327,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009163927705202355,
      "loss": 1.5334,
      "step": 10873
    },
    {
      "epoch": 0.9355588058160543,
      "grad_norm": 9.3125,
      "learning_rate": 0.0009163778002753318,
      "loss": 1.81,
      "step": 10874
    },
    {
      "epoch": 0.9356448421233761,
      "grad_norm": 1.203125,
      "learning_rate": 0.000916362828812604,
      "loss": 1.4788,
      "step": 10875
    },
    {
      "epoch": 0.9357308784306978,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009163478561320959,
      "loss": 1.3377,
      "step": 10876
    },
    {
      "epoch": 0.9358169147380194,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009163328822338512,
      "loss": 1.1346,
      "step": 10877
    },
    {
      "epoch": 0.9359029510453412,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009163179071179138,
      "loss": 1.6394,
      "step": 10878
    },
    {
      "epoch": 0.9359889873526628,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009163029307843276,
      "loss": 1.2848,
      "step": 10879
    },
    {
      "epoch": 0.9360750236599845,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009162879532331363,
      "loss": 1.6294,
      "step": 10880
    },
    {
      "epoch": 0.9361610599673063,
      "grad_norm": 2.421875,
      "learning_rate": 0.0009162729744643835,
      "loss": 1.1256,
      "step": 10881
    },
    {
      "epoch": 0.9362470962746279,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009162579944781134,
      "loss": 1.0299,
      "step": 10882
    },
    {
      "epoch": 0.9363331325819496,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009162430132743695,
      "loss": 1.4303,
      "step": 10883
    },
    {
      "epoch": 0.9364191688892712,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009162280308531958,
      "loss": 1.1287,
      "step": 10884
    },
    {
      "epoch": 0.936505205196593,
      "grad_norm": 2.421875,
      "learning_rate": 0.0009162130472146362,
      "loss": 1.2873,
      "step": 10885
    },
    {
      "epoch": 0.9365912415039147,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009161980623587343,
      "loss": 1.0575,
      "step": 10886
    },
    {
      "epoch": 0.9366772778112363,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009161830762855339,
      "loss": 1.6682,
      "step": 10887
    },
    {
      "epoch": 0.9367633141185581,
      "grad_norm": 1.28125,
      "learning_rate": 0.000916168088995079,
      "loss": 1.7253,
      "step": 10888
    },
    {
      "epoch": 0.9368493504258797,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009161531004874134,
      "loss": 1.1659,
      "step": 10889
    },
    {
      "epoch": 0.9369353867332014,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009161381107625809,
      "loss": 1.1806,
      "step": 10890
    },
    {
      "epoch": 0.937021423040523,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009161231198206253,
      "loss": 1.1652,
      "step": 10891
    },
    {
      "epoch": 0.9371074593478448,
      "grad_norm": 6.09375,
      "learning_rate": 0.0009161081276615906,
      "loss": 1.2768,
      "step": 10892
    },
    {
      "epoch": 0.9371934956551665,
      "grad_norm": 1.375,
      "learning_rate": 0.0009160931342855205,
      "loss": 1.1783,
      "step": 10893
    },
    {
      "epoch": 0.9372795319624881,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009160781396924591,
      "loss": 0.7646,
      "step": 10894
    },
    {
      "epoch": 0.9373655682698099,
      "grad_norm": 1.25,
      "learning_rate": 0.0009160631438824498,
      "loss": 1.1472,
      "step": 10895
    },
    {
      "epoch": 0.9374516045771315,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009160481468555369,
      "loss": 0.8351,
      "step": 10896
    },
    {
      "epoch": 0.9375376408844532,
      "grad_norm": 1.1875,
      "learning_rate": 0.000916033148611764,
      "loss": 1.9841,
      "step": 10897
    },
    {
      "epoch": 0.937623677191775,
      "grad_norm": 1.9140625,
      "learning_rate": 0.000916018149151175,
      "loss": 1.0706,
      "step": 10898
    },
    {
      "epoch": 0.9377097134990966,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009160031484738141,
      "loss": 1.239,
      "step": 10899
    },
    {
      "epoch": 0.9377957498064183,
      "grad_norm": 1.625,
      "learning_rate": 0.0009159881465797247,
      "loss": 1.0403,
      "step": 10900
    },
    {
      "epoch": 0.93788178611374,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009159731434689507,
      "loss": 1.437,
      "step": 10901
    },
    {
      "epoch": 0.9379678224210617,
      "grad_norm": 2.546875,
      "learning_rate": 0.0009159581391415363,
      "loss": 1.1407,
      "step": 10902
    },
    {
      "epoch": 0.9380538587283834,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0009159431335975253,
      "loss": 1.502,
      "step": 10903
    },
    {
      "epoch": 0.938139895035705,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009159281268369615,
      "loss": 1.2485,
      "step": 10904
    },
    {
      "epoch": 0.9382259313430268,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009159131188598888,
      "loss": 0.9027,
      "step": 10905
    },
    {
      "epoch": 0.9383119676503484,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009158981096663513,
      "loss": 1.0767,
      "step": 10906
    },
    {
      "epoch": 0.9383980039576701,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009158830992563924,
      "loss": 1.1328,
      "step": 10907
    },
    {
      "epoch": 0.9384840402649919,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009158680876300566,
      "loss": 0.9952,
      "step": 10908
    },
    {
      "epoch": 0.9385700765723135,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009158530747873873,
      "loss": 1.5292,
      "step": 10909
    },
    {
      "epoch": 0.9386561128796352,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009158380607284288,
      "loss": 0.6445,
      "step": 10910
    },
    {
      "epoch": 0.9387421491869569,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009158230454532248,
      "loss": 1.3386,
      "step": 10911
    },
    {
      "epoch": 0.9388281854942786,
      "grad_norm": 2.71875,
      "learning_rate": 0.0009158080289618192,
      "loss": 1.8475,
      "step": 10912
    },
    {
      "epoch": 0.9389142218016002,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000915793011254256,
      "loss": 1.0877,
      "step": 10913
    },
    {
      "epoch": 0.939000258108922,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009157779923305793,
      "loss": 1.3214,
      "step": 10914
    },
    {
      "epoch": 0.9390862944162437,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009157629721908326,
      "loss": 1.4041,
      "step": 10915
    },
    {
      "epoch": 0.9391723307235653,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009157479508350602,
      "loss": 1.1161,
      "step": 10916
    },
    {
      "epoch": 0.939258367030887,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009157329282633058,
      "loss": 1.1419,
      "step": 10917
    },
    {
      "epoch": 0.9393444033382087,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009157179044756136,
      "loss": 1.3475,
      "step": 10918
    },
    {
      "epoch": 0.9394304396455304,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009157028794720272,
      "loss": 1.5337,
      "step": 10919
    },
    {
      "epoch": 0.9395164759528521,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009156878532525909,
      "loss": 1.5247,
      "step": 10920
    },
    {
      "epoch": 0.9396025122601738,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009156728258173484,
      "loss": 0.8134,
      "step": 10921
    },
    {
      "epoch": 0.9396885485674955,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009156577971663435,
      "loss": 1.3168,
      "step": 10922
    },
    {
      "epoch": 0.9397745848748171,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009156427672996207,
      "loss": 1.5899,
      "step": 10923
    },
    {
      "epoch": 0.9398606211821389,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009156277362172236,
      "loss": 1.4561,
      "step": 10924
    },
    {
      "epoch": 0.9399466574894606,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009156127039191962,
      "loss": 1.5318,
      "step": 10925
    },
    {
      "epoch": 0.9400326937967822,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009155976704055823,
      "loss": 1.7705,
      "step": 10926
    },
    {
      "epoch": 0.940118730104104,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009155826356764262,
      "loss": 1.4895,
      "step": 10927
    },
    {
      "epoch": 0.9402047664114256,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009155675997317716,
      "loss": 0.9751,
      "step": 10928
    },
    {
      "epoch": 0.9402908027187473,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009155525625716628,
      "loss": 1.8449,
      "step": 10929
    },
    {
      "epoch": 0.9403768390260691,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009155375241961434,
      "loss": 1.2601,
      "step": 10930
    },
    {
      "epoch": 0.9404628753333907,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009155224846052577,
      "loss": 1.558,
      "step": 10931
    },
    {
      "epoch": 0.9405489116407124,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009155074437990494,
      "loss": 1.6258,
      "step": 10932
    },
    {
      "epoch": 0.940634947948034,
      "grad_norm": 2.6875,
      "learning_rate": 0.0009154924017775627,
      "loss": 1.6001,
      "step": 10933
    },
    {
      "epoch": 0.9407209842553558,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0009154773585408415,
      "loss": 1.2028,
      "step": 10934
    },
    {
      "epoch": 0.9408070205626774,
      "grad_norm": 1.375,
      "learning_rate": 0.0009154623140889298,
      "loss": 1.2062,
      "step": 10935
    },
    {
      "epoch": 0.9408930568699991,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009154472684218718,
      "loss": 1.3328,
      "step": 10936
    },
    {
      "epoch": 0.9409790931773209,
      "grad_norm": 1.15625,
      "learning_rate": 0.000915432221539711,
      "loss": 1.869,
      "step": 10937
    },
    {
      "epoch": 0.9410651294846425,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009154171734424921,
      "loss": 1.49,
      "step": 10938
    },
    {
      "epoch": 0.9411511657919642,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009154021241302586,
      "loss": 0.9919,
      "step": 10939
    },
    {
      "epoch": 0.9412372020992859,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009153870736030548,
      "loss": 1.5346,
      "step": 10940
    },
    {
      "epoch": 0.9413232384066076,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009153720218609245,
      "loss": 0.7082,
      "step": 10941
    },
    {
      "epoch": 0.9414092747139293,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009153569689039118,
      "loss": 1.258,
      "step": 10942
    },
    {
      "epoch": 0.941495311021251,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009153419147320609,
      "loss": 1.103,
      "step": 10943
    },
    {
      "epoch": 0.9415813473285727,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009153268593454156,
      "loss": 1.6677,
      "step": 10944
    },
    {
      "epoch": 0.9416673836358943,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009153118027440199,
      "loss": 0.775,
      "step": 10945
    },
    {
      "epoch": 0.941753419943216,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009152967449279181,
      "loss": 1.77,
      "step": 10946
    },
    {
      "epoch": 0.9418394562505378,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000915281685897154,
      "loss": 1.1738,
      "step": 10947
    },
    {
      "epoch": 0.9419254925578594,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009152666256517718,
      "loss": 1.4462,
      "step": 10948
    },
    {
      "epoch": 0.9420115288651811,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009152515641918153,
      "loss": 1.8521,
      "step": 10949
    },
    {
      "epoch": 0.9420975651725028,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009152365015173289,
      "loss": 2.0345,
      "step": 10950
    },
    {
      "epoch": 0.9421836014798245,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009152214376283565,
      "loss": 1.7814,
      "step": 10951
    },
    {
      "epoch": 0.9422696377871462,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009152063725249422,
      "loss": 1.1519,
      "step": 10952
    },
    {
      "epoch": 0.9423556740944679,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00091519130620713,
      "loss": 1.054,
      "step": 10953
    },
    {
      "epoch": 0.9424417104017896,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009151762386749638,
      "loss": 1.1135,
      "step": 10954
    },
    {
      "epoch": 0.9425277467091112,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009151611699284881,
      "loss": 1.5449,
      "step": 10955
    },
    {
      "epoch": 0.942613783016433,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009151460999677466,
      "loss": 0.9806,
      "step": 10956
    },
    {
      "epoch": 0.9426998193237546,
      "grad_norm": 1.25,
      "learning_rate": 0.0009151310287927834,
      "loss": 1.0986,
      "step": 10957
    },
    {
      "epoch": 0.9427858556310763,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009151159564036428,
      "loss": 1.7272,
      "step": 10958
    },
    {
      "epoch": 0.942871891938398,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009151008828003686,
      "loss": 1.7738,
      "step": 10959
    },
    {
      "epoch": 0.9429579282457197,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009150858079830052,
      "loss": 1.5979,
      "step": 10960
    },
    {
      "epoch": 0.9430439645530414,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009150707319515965,
      "loss": 1.4924,
      "step": 10961
    },
    {
      "epoch": 0.943130000860363,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009150556547061867,
      "loss": 1.4991,
      "step": 10962
    },
    {
      "epoch": 0.9432160371676848,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009150405762468199,
      "loss": 1.2169,
      "step": 10963
    },
    {
      "epoch": 0.9433020734750065,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00091502549657354,
      "loss": 1.6728,
      "step": 10964
    },
    {
      "epoch": 0.9433881097823281,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009150104156863912,
      "loss": 0.6294,
      "step": 10965
    },
    {
      "epoch": 0.9434741460896499,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009149953335854178,
      "loss": 0.9926,
      "step": 10966
    },
    {
      "epoch": 0.9435601823969715,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009149802502706637,
      "loss": 1.3312,
      "step": 10967
    },
    {
      "epoch": 0.9436462187042932,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009149651657421729,
      "loss": 1.6896,
      "step": 10968
    },
    {
      "epoch": 0.943732255011615,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009149500799999899,
      "loss": 1.6149,
      "step": 10969
    },
    {
      "epoch": 0.9438182913189366,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009149349930441586,
      "loss": 0.7749,
      "step": 10970
    },
    {
      "epoch": 0.9439043276262583,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000914919904874723,
      "loss": 1.2543,
      "step": 10971
    },
    {
      "epoch": 0.9439903639335799,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009149048154917277,
      "loss": 1.2891,
      "step": 10972
    },
    {
      "epoch": 0.9440764002409017,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009148897248952161,
      "loss": 1.4224,
      "step": 10973
    },
    {
      "epoch": 0.9441624365482234,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009148746330852331,
      "loss": 1.267,
      "step": 10974
    },
    {
      "epoch": 0.944248472855545,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009148595400618223,
      "loss": 1.7029,
      "step": 10975
    },
    {
      "epoch": 0.9443345091628668,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009148444458250281,
      "loss": 1.0715,
      "step": 10976
    },
    {
      "epoch": 0.9444205454701884,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0009148293503748946,
      "loss": 1.2916,
      "step": 10977
    },
    {
      "epoch": 0.9445065817775101,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009148142537114659,
      "loss": 1.1982,
      "step": 10978
    },
    {
      "epoch": 0.9445926180848317,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009147991558347862,
      "loss": 1.6013,
      "step": 10979
    },
    {
      "epoch": 0.9446786543921535,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009147840567448996,
      "loss": 1.7385,
      "step": 10980
    },
    {
      "epoch": 0.9447646906994752,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009147689564418502,
      "loss": 1.6232,
      "step": 10981
    },
    {
      "epoch": 0.9448507270067968,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009147538549256825,
      "loss": 1.0024,
      "step": 10982
    },
    {
      "epoch": 0.9449367633141186,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009147387521964402,
      "loss": 1.4475,
      "step": 10983
    },
    {
      "epoch": 0.9450227996214402,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009147236482541679,
      "loss": 1.162,
      "step": 10984
    },
    {
      "epoch": 0.9451088359287619,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009147085430989095,
      "loss": 0.9963,
      "step": 10985
    },
    {
      "epoch": 0.9451948722360837,
      "grad_norm": 1.125,
      "learning_rate": 0.0009146934367307092,
      "loss": 1.1516,
      "step": 10986
    },
    {
      "epoch": 0.9452809085434053,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009146783291496114,
      "loss": 1.1618,
      "step": 10987
    },
    {
      "epoch": 0.945366944850727,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00091466322035566,
      "loss": 1.7296,
      "step": 10988
    },
    {
      "epoch": 0.9454529811580487,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009146481103488995,
      "loss": 0.7809,
      "step": 10989
    },
    {
      "epoch": 0.9455390174653704,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009146329991293738,
      "loss": 1.4942,
      "step": 10990
    },
    {
      "epoch": 0.9456250537726921,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009146178866971273,
      "loss": 1.7857,
      "step": 10991
    },
    {
      "epoch": 0.9457110900800137,
      "grad_norm": 1.8515625,
      "learning_rate": 0.000914602773052204,
      "loss": 3.186,
      "step": 10992
    },
    {
      "epoch": 0.9457971263873355,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009145876581946483,
      "loss": 0.7822,
      "step": 10993
    },
    {
      "epoch": 0.9458831626946571,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009145725421245044,
      "loss": 1.4809,
      "step": 10994
    },
    {
      "epoch": 0.9459691990019788,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0009145574248418162,
      "loss": 0.8734,
      "step": 10995
    },
    {
      "epoch": 0.9460552353093006,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009145423063466284,
      "loss": 2.0516,
      "step": 10996
    },
    {
      "epoch": 0.9461412716166222,
      "grad_norm": 1.125,
      "learning_rate": 0.0009145271866389848,
      "loss": 1.6174,
      "step": 10997
    },
    {
      "epoch": 0.9462273079239439,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00091451206571893,
      "loss": 1.0815,
      "step": 10998
    },
    {
      "epoch": 0.9463133442312656,
      "grad_norm": 1.265625,
      "learning_rate": 0.000914496943586508,
      "loss": 1.7268,
      "step": 10999
    },
    {
      "epoch": 0.9463993805385873,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009144818202417628,
      "loss": 1.8083,
      "step": 11000
    },
    {
      "epoch": 0.9464854168459089,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009144666956847392,
      "loss": 1.1499,
      "step": 11001
    },
    {
      "epoch": 0.9465714531532307,
      "grad_norm": 1.828125,
      "learning_rate": 0.000914451569915481,
      "loss": 1.6903,
      "step": 11002
    },
    {
      "epoch": 0.9466574894605524,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0009144364429340326,
      "loss": 0.8796,
      "step": 11003
    },
    {
      "epoch": 0.946743525767874,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009144213147404381,
      "loss": 1.0963,
      "step": 11004
    },
    {
      "epoch": 0.9468295620751958,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009144061853347419,
      "loss": 0.7978,
      "step": 11005
    },
    {
      "epoch": 0.9469155983825174,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009143910547169883,
      "loss": 1.7282,
      "step": 11006
    },
    {
      "epoch": 0.9470016346898391,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009143759228872213,
      "loss": 1.0664,
      "step": 11007
    },
    {
      "epoch": 0.9470876709971608,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009143607898454856,
      "loss": 1.8189,
      "step": 11008
    },
    {
      "epoch": 0.9471737073044825,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009143456555918249,
      "loss": 1.4228,
      "step": 11009
    },
    {
      "epoch": 0.9472597436118042,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009143305201262837,
      "loss": 0.8192,
      "step": 11010
    },
    {
      "epoch": 0.9473457799191258,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009143153834489065,
      "loss": 1.1069,
      "step": 11011
    },
    {
      "epoch": 0.9474318162264476,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009143002455597373,
      "loss": 1.8498,
      "step": 11012
    },
    {
      "epoch": 0.9475178525337693,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009142851064588205,
      "loss": 1.8876,
      "step": 11013
    },
    {
      "epoch": 0.9476038888410909,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009142699661462004,
      "loss": 1.3527,
      "step": 11014
    },
    {
      "epoch": 0.9476899251484127,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009142548246219211,
      "loss": 1.3974,
      "step": 11015
    },
    {
      "epoch": 0.9477759614557343,
      "grad_norm": 1.609375,
      "learning_rate": 0.000914239681886027,
      "loss": 1.9271,
      "step": 11016
    },
    {
      "epoch": 0.947861997763056,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009142245379385626,
      "loss": 1.2081,
      "step": 11017
    },
    {
      "epoch": 0.9479480340703778,
      "grad_norm": 1.375,
      "learning_rate": 0.0009142093927795717,
      "loss": 1.8062,
      "step": 11018
    },
    {
      "epoch": 0.9480340703776994,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009141942464090991,
      "loss": 1.7695,
      "step": 11019
    },
    {
      "epoch": 0.9481201066850211,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009141790988271887,
      "loss": 0.7514,
      "step": 11020
    },
    {
      "epoch": 0.9482061429923427,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009141639500338851,
      "loss": 0.6982,
      "step": 11021
    },
    {
      "epoch": 0.9482921792996645,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009141488000292325,
      "loss": 1.3631,
      "step": 11022
    },
    {
      "epoch": 0.9483782156069861,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009141336488132752,
      "loss": 1.5265,
      "step": 11023
    },
    {
      "epoch": 0.9484642519143078,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009141184963860575,
      "loss": 1.4806,
      "step": 11024
    },
    {
      "epoch": 0.9485502882216296,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009141033427476238,
      "loss": 1.5718,
      "step": 11025
    },
    {
      "epoch": 0.9486363245289512,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009140881878980183,
      "loss": 1.7733,
      "step": 11026
    },
    {
      "epoch": 0.9487223608362729,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009140730318372854,
      "loss": 0.887,
      "step": 11027
    },
    {
      "epoch": 0.9488083971435946,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009140578745654694,
      "loss": 1.3359,
      "step": 11028
    },
    {
      "epoch": 0.9488944334509163,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009140427160826146,
      "loss": 1.4821,
      "step": 11029
    },
    {
      "epoch": 0.948980469758238,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0009140275563887656,
      "loss": 0.8028,
      "step": 11030
    },
    {
      "epoch": 0.9490665060655596,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009140123954839663,
      "loss": 1.0885,
      "step": 11031
    },
    {
      "epoch": 0.9491525423728814,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009139972333682614,
      "loss": 2.0403,
      "step": 11032
    },
    {
      "epoch": 0.949238578680203,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000913982070041695,
      "loss": 2.0775,
      "step": 11033
    },
    {
      "epoch": 0.9493246149875247,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009139669055043116,
      "loss": 1.2342,
      "step": 11034
    },
    {
      "epoch": 0.9494106512948465,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009139517397561554,
      "loss": 1.3824,
      "step": 11035
    },
    {
      "epoch": 0.9494966876021681,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000913936572797271,
      "loss": 1.4851,
      "step": 11036
    },
    {
      "epoch": 0.9495827239094898,
      "grad_norm": 1.0,
      "learning_rate": 0.0009139214046277027,
      "loss": 1.7135,
      "step": 11037
    },
    {
      "epoch": 0.9496687602168115,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009139062352474947,
      "loss": 1.2163,
      "step": 11038
    },
    {
      "epoch": 0.9497547965241332,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009138910646566915,
      "loss": 1.4122,
      "step": 11039
    },
    {
      "epoch": 0.9498408328314549,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009138758928553375,
      "loss": 1.2762,
      "step": 11040
    },
    {
      "epoch": 0.9499268691387766,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009138607198434768,
      "loss": 1.6335,
      "step": 11041
    },
    {
      "epoch": 0.9500129054460983,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009138455456211541,
      "loss": 1.0848,
      "step": 11042
    },
    {
      "epoch": 0.9500989417534199,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009138303701884137,
      "loss": 1.9096,
      "step": 11043
    },
    {
      "epoch": 0.9501849780607416,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0009138151935452998,
      "loss": 1.2875,
      "step": 11044
    },
    {
      "epoch": 0.9502710143680633,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009138000156918571,
      "loss": 2.1879,
      "step": 11045
    },
    {
      "epoch": 0.950357050675385,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009137848366281298,
      "loss": 1.3166,
      "step": 11046
    },
    {
      "epoch": 0.9504430869827067,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009137696563541625,
      "loss": 1.0014,
      "step": 11047
    },
    {
      "epoch": 0.9505291232900284,
      "grad_norm": 2.109375,
      "learning_rate": 0.000913754474869999,
      "loss": 2.408,
      "step": 11048
    },
    {
      "epoch": 0.9506151595973501,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0009137392921756845,
      "loss": 1.1192,
      "step": 11049
    },
    {
      "epoch": 0.9507011959046717,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009137241082712629,
      "loss": 1.4755,
      "step": 11050
    },
    {
      "epoch": 0.9507872322119935,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009137089231567788,
      "loss": 1.3793,
      "step": 11051
    },
    {
      "epoch": 0.9508732685193152,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009136937368322765,
      "loss": 1.2496,
      "step": 11052
    },
    {
      "epoch": 0.9509593048266368,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009136785492978006,
      "loss": 1.4501,
      "step": 11053
    },
    {
      "epoch": 0.9510453411339586,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009136633605533953,
      "loss": 0.8991,
      "step": 11054
    },
    {
      "epoch": 0.9511313774412802,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009136481705991051,
      "loss": 0.7627,
      "step": 11055
    },
    {
      "epoch": 0.9512174137486019,
      "grad_norm": 37.75,
      "learning_rate": 0.0009136329794349745,
      "loss": 1.4236,
      "step": 11056
    },
    {
      "epoch": 0.9513034500559236,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009136177870610478,
      "loss": 1.3068,
      "step": 11057
    },
    {
      "epoch": 0.9513894863632453,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009136025934773697,
      "loss": 1.1338,
      "step": 11058
    },
    {
      "epoch": 0.951475522670567,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009135873986839844,
      "loss": 0.8647,
      "step": 11059
    },
    {
      "epoch": 0.9515615589778886,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009135722026809364,
      "loss": 1.6353,
      "step": 11060
    },
    {
      "epoch": 0.9516475952852104,
      "grad_norm": 9.6875,
      "learning_rate": 0.0009135570054682699,
      "loss": 1.102,
      "step": 11061
    },
    {
      "epoch": 0.9517336315925321,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009135418070460298,
      "loss": 1.1627,
      "step": 11062
    },
    {
      "epoch": 0.9518196678998537,
      "grad_norm": 2.78125,
      "learning_rate": 0.0009135266074142603,
      "loss": 2.1521,
      "step": 11063
    },
    {
      "epoch": 0.9519057042071755,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009135114065730059,
      "loss": 1.1458,
      "step": 11064
    },
    {
      "epoch": 0.9519917405144971,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0009134962045223112,
      "loss": 1.0325,
      "step": 11065
    },
    {
      "epoch": 0.9520777768218188,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009134810012622203,
      "loss": 2.131,
      "step": 11066
    },
    {
      "epoch": 0.9521638131291404,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009134657967927779,
      "loss": 1.6748,
      "step": 11067
    },
    {
      "epoch": 0.9522498494364622,
      "grad_norm": 2.078125,
      "learning_rate": 0.0009134505911140286,
      "loss": 1.2143,
      "step": 11068
    },
    {
      "epoch": 0.9523358857437839,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009134353842260168,
      "loss": 1.1338,
      "step": 11069
    },
    {
      "epoch": 0.9524219220511055,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009134201761287867,
      "loss": 0.9289,
      "step": 11070
    },
    {
      "epoch": 0.9525079583584273,
      "grad_norm": 2.9375,
      "learning_rate": 0.0009134049668223832,
      "loss": 1.5705,
      "step": 11071
    },
    {
      "epoch": 0.9525939946657489,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009133897563068503,
      "loss": 1.5808,
      "step": 11072
    },
    {
      "epoch": 0.9526800309730706,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000913374544582233,
      "loss": 1.4012,
      "step": 11073
    },
    {
      "epoch": 0.9527660672803924,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009133593316485754,
      "loss": 1.6418,
      "step": 11074
    },
    {
      "epoch": 0.952852103587714,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009133441175059223,
      "loss": 1.0019,
      "step": 11075
    },
    {
      "epoch": 0.9529381398950357,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009133289021543179,
      "loss": 0.8808,
      "step": 11076
    },
    {
      "epoch": 0.9530241762023574,
      "grad_norm": 2.6875,
      "learning_rate": 0.0009133136855938069,
      "loss": 1.5574,
      "step": 11077
    },
    {
      "epoch": 0.9531102125096791,
      "grad_norm": 3.609375,
      "learning_rate": 0.0009132984678244338,
      "loss": 1.5629,
      "step": 11078
    },
    {
      "epoch": 0.9531962488170008,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0009132832488462431,
      "loss": 1.7536,
      "step": 11079
    },
    {
      "epoch": 0.9532822851243224,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0009132680286592791,
      "loss": 1.4397,
      "step": 11080
    },
    {
      "epoch": 0.9533683214316442,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009132528072635866,
      "loss": 1.0197,
      "step": 11081
    },
    {
      "epoch": 0.9534543577389658,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009132375846592101,
      "loss": 1.5007,
      "step": 11082
    },
    {
      "epoch": 0.9535403940462875,
      "grad_norm": 1.0625,
      "learning_rate": 0.000913222360846194,
      "loss": 1.292,
      "step": 11083
    },
    {
      "epoch": 0.9536264303536093,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009132071358245828,
      "loss": 0.817,
      "step": 11084
    },
    {
      "epoch": 0.9537124666609309,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000913191909594421,
      "loss": 1.2546,
      "step": 11085
    },
    {
      "epoch": 0.9537985029682526,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009131766821557535,
      "loss": 0.9545,
      "step": 11086
    },
    {
      "epoch": 0.9538845392755743,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009131614535086243,
      "loss": 1.0411,
      "step": 11087
    },
    {
      "epoch": 0.953970575582896,
      "grad_norm": 1.921875,
      "learning_rate": 0.0009131462236530784,
      "loss": 1.9387,
      "step": 11088
    },
    {
      "epoch": 0.9540566118902176,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009131309925891599,
      "loss": 0.7402,
      "step": 11089
    },
    {
      "epoch": 0.9541426481975394,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009131157603169137,
      "loss": 2.1784,
      "step": 11090
    },
    {
      "epoch": 0.9542286845048611,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009131005268363843,
      "loss": 1.3547,
      "step": 11091
    },
    {
      "epoch": 0.9543147208121827,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009130852921476162,
      "loss": 1.4044,
      "step": 11092
    },
    {
      "epoch": 0.9544007571195045,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000913070056250654,
      "loss": 1.1138,
      "step": 11093
    },
    {
      "epoch": 0.9544867934268261,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009130548191455421,
      "loss": 1.1928,
      "step": 11094
    },
    {
      "epoch": 0.9545728297341478,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009130395808323253,
      "loss": 0.991,
      "step": 11095
    },
    {
      "epoch": 0.9546588660414695,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000913024341311048,
      "loss": 1.0733,
      "step": 11096
    },
    {
      "epoch": 0.9547449023487912,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000913009100581755,
      "loss": 2.3124,
      "step": 11097
    },
    {
      "epoch": 0.9548309386561129,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009129938586444904,
      "loss": 1.5651,
      "step": 11098
    },
    {
      "epoch": 0.9549169749634345,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009129786154992994,
      "loss": 1.2051,
      "step": 11099
    },
    {
      "epoch": 0.9550030112707563,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000912963371146226,
      "loss": 1.3364,
      "step": 11100
    },
    {
      "epoch": 0.955089047578078,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009129481255853152,
      "loss": 1.3342,
      "step": 11101
    },
    {
      "epoch": 0.9551750838853996,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009129328788166114,
      "loss": 1.0198,
      "step": 11102
    },
    {
      "epoch": 0.9552611201927214,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009129176308401594,
      "loss": 2.0929,
      "step": 11103
    },
    {
      "epoch": 0.955347156500043,
      "grad_norm": 2.34375,
      "learning_rate": 0.0009129023816560034,
      "loss": 1.8647,
      "step": 11104
    },
    {
      "epoch": 0.9554331928073647,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009128871312641884,
      "loss": 1.2997,
      "step": 11105
    },
    {
      "epoch": 0.9555192291146865,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009128718796647587,
      "loss": 1.3328,
      "step": 11106
    },
    {
      "epoch": 0.9556052654220081,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009128566268577591,
      "loss": 1.4619,
      "step": 11107
    },
    {
      "epoch": 0.9556913017293298,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009128413728432342,
      "loss": 0.8823,
      "step": 11108
    },
    {
      "epoch": 0.9557773380366514,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009128261176212286,
      "loss": 1.147,
      "step": 11109
    },
    {
      "epoch": 0.9558633743439732,
      "grad_norm": 1.15625,
      "learning_rate": 0.000912810861191787,
      "loss": 1.2509,
      "step": 11110
    },
    {
      "epoch": 0.9559494106512948,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009127956035549536,
      "loss": 1.208,
      "step": 11111
    },
    {
      "epoch": 0.9560354469586165,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009127803447107737,
      "loss": 1.5332,
      "step": 11112
    },
    {
      "epoch": 0.9561214832659383,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009127650846592912,
      "loss": 1.5404,
      "step": 11113
    },
    {
      "epoch": 0.9562075195732599,
      "grad_norm": 2.640625,
      "learning_rate": 0.0009127498234005513,
      "loss": 1.5152,
      "step": 11114
    },
    {
      "epoch": 0.9562935558805816,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009127345609345985,
      "loss": 0.8942,
      "step": 11115
    },
    {
      "epoch": 0.9563795921879032,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009127192972614772,
      "loss": 1.2116,
      "step": 11116
    },
    {
      "epoch": 0.956465628495225,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009127040323812325,
      "loss": 1.1164,
      "step": 11117
    },
    {
      "epoch": 0.9565516648025467,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009126887662939087,
      "loss": 1.4039,
      "step": 11118
    },
    {
      "epoch": 0.9566377011098683,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009126734989995505,
      "loss": 1.4168,
      "step": 11119
    },
    {
      "epoch": 0.9567237374171901,
      "grad_norm": 8.6875,
      "learning_rate": 0.0009126582304982026,
      "loss": 1.7096,
      "step": 11120
    },
    {
      "epoch": 0.9568097737245117,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009126429607899094,
      "loss": 1.4339,
      "step": 11121
    },
    {
      "epoch": 0.9568958100318334,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009126276898747159,
      "loss": 1.3652,
      "step": 11122
    },
    {
      "epoch": 0.9569818463391552,
      "grad_norm": 3.90625,
      "learning_rate": 0.0009126124177526668,
      "loss": 2.0783,
      "step": 11123
    },
    {
      "epoch": 0.9570678826464768,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009125971444238065,
      "loss": 1.1443,
      "step": 11124
    },
    {
      "epoch": 0.9571539189537985,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009125818698881798,
      "loss": 1.456,
      "step": 11125
    },
    {
      "epoch": 0.9572399552611202,
      "grad_norm": 1.984375,
      "learning_rate": 0.0009125665941458315,
      "loss": 1.6217,
      "step": 11126
    },
    {
      "epoch": 0.9573259915684419,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009125513171968061,
      "loss": 1.4049,
      "step": 11127
    },
    {
      "epoch": 0.9574120278757636,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009125360390411482,
      "loss": 2.0213,
      "step": 11128
    },
    {
      "epoch": 0.9574980641830853,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009125207596789027,
      "loss": 1.8529,
      "step": 11129
    },
    {
      "epoch": 0.957584100490407,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009125054791101142,
      "loss": 1.8462,
      "step": 11130
    },
    {
      "epoch": 0.9576701367977286,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009124901973348275,
      "loss": 0.6705,
      "step": 11131
    },
    {
      "epoch": 0.9577561731050503,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009124749143530871,
      "loss": 1.3224,
      "step": 11132
    },
    {
      "epoch": 0.957842209412372,
      "grad_norm": 1.5,
      "learning_rate": 0.0009124596301649378,
      "loss": 1.1576,
      "step": 11133
    },
    {
      "epoch": 0.9579282457196937,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009124443447704242,
      "loss": 1.5279,
      "step": 11134
    },
    {
      "epoch": 0.9580142820270154,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009124290581695913,
      "loss": 1.1148,
      "step": 11135
    },
    {
      "epoch": 0.9581003183343371,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009124137703624836,
      "loss": 0.9052,
      "step": 11136
    },
    {
      "epoch": 0.9581863546416588,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009123984813491457,
      "loss": 1.4063,
      "step": 11137
    },
    {
      "epoch": 0.9582723909489804,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009123831911296227,
      "loss": 2.1387,
      "step": 11138
    },
    {
      "epoch": 0.9583584272563022,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009123678997039589,
      "loss": 1.5113,
      "step": 11139
    },
    {
      "epoch": 0.9584444635636239,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009123526070721991,
      "loss": 1.0216,
      "step": 11140
    },
    {
      "epoch": 0.9585304998709455,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009123373132343881,
      "loss": 1.8613,
      "step": 11141
    },
    {
      "epoch": 0.9586165361782673,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009123220181905709,
      "loss": 1.689,
      "step": 11142
    },
    {
      "epoch": 0.9587025724855889,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009123067219407919,
      "loss": 0.9395,
      "step": 11143
    },
    {
      "epoch": 0.9587886087929106,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009122914244850958,
      "loss": 1.5657,
      "step": 11144
    },
    {
      "epoch": 0.9588746451002323,
      "grad_norm": 33.0,
      "learning_rate": 0.0009122761258235275,
      "loss": 1.8111,
      "step": 11145
    },
    {
      "epoch": 0.958960681407554,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000912260825956132,
      "loss": 1.4507,
      "step": 11146
    },
    {
      "epoch": 0.9590467177148757,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009122455248829534,
      "loss": 1.1554,
      "step": 11147
    },
    {
      "epoch": 0.9591327540221973,
      "grad_norm": 1.328125,
      "learning_rate": 0.000912230222604037,
      "loss": 1.2085,
      "step": 11148
    },
    {
      "epoch": 0.9592187903295191,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009122149191194273,
      "loss": 1.4917,
      "step": 11149
    },
    {
      "epoch": 0.9593048266368408,
      "grad_norm": 28.5,
      "learning_rate": 0.0009121996144291691,
      "loss": 1.9488,
      "step": 11150
    },
    {
      "epoch": 0.9593908629441624,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009121843085333072,
      "loss": 1.5176,
      "step": 11151
    },
    {
      "epoch": 0.9594768992514842,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009121690014318865,
      "loss": 2.1875,
      "step": 11152
    },
    {
      "epoch": 0.9595629355588058,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009121536931249516,
      "loss": 1.1989,
      "step": 11153
    },
    {
      "epoch": 0.9596489718661275,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009121383836125472,
      "loss": 1.4406,
      "step": 11154
    },
    {
      "epoch": 0.9597350081734491,
      "grad_norm": 1.96875,
      "learning_rate": 0.0009121230728947183,
      "loss": 2.0847,
      "step": 11155
    },
    {
      "epoch": 0.9598210444807709,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009121077609715094,
      "loss": 1.9573,
      "step": 11156
    },
    {
      "epoch": 0.9599070807880926,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009120924478429656,
      "loss": 0.6308,
      "step": 11157
    },
    {
      "epoch": 0.9599931170954142,
      "grad_norm": 2.8125,
      "learning_rate": 0.0009120771335091316,
      "loss": 1.6967,
      "step": 11158
    },
    {
      "epoch": 0.960079153402736,
      "grad_norm": 1.5625,
      "learning_rate": 0.000912061817970052,
      "loss": 1.3893,
      "step": 11159
    },
    {
      "epoch": 0.9601651897100576,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009120465012257717,
      "loss": 0.9518,
      "step": 11160
    },
    {
      "epoch": 0.9602512260173793,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009120311832763356,
      "loss": 1.3824,
      "step": 11161
    },
    {
      "epoch": 0.9603372623247011,
      "grad_norm": 1.125,
      "learning_rate": 0.0009120158641217884,
      "loss": 1.5956,
      "step": 11162
    },
    {
      "epoch": 0.9604232986320227,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009120005437621749,
      "loss": 1.7921,
      "step": 11163
    },
    {
      "epoch": 0.9605093349393444,
      "grad_norm": 1.109375,
      "learning_rate": 0.00091198522219754,
      "loss": 1.0607,
      "step": 11164
    },
    {
      "epoch": 0.960595371246666,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009119698994279284,
      "loss": 1.5696,
      "step": 11165
    },
    {
      "epoch": 0.9606814075539878,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000911954575453385,
      "loss": 1.9199,
      "step": 11166
    },
    {
      "epoch": 0.9607674438613095,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009119392502739545,
      "loss": 1.7078,
      "step": 11167
    },
    {
      "epoch": 0.9608534801686311,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009119239238896819,
      "loss": 1.0343,
      "step": 11168
    },
    {
      "epoch": 0.9609395164759529,
      "grad_norm": 1.53125,
      "learning_rate": 0.000911908596300612,
      "loss": 1.5977,
      "step": 11169
    },
    {
      "epoch": 0.9610255527832745,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009118932675067895,
      "loss": 1.2634,
      "step": 11170
    },
    {
      "epoch": 0.9611115890905962,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009118779375082592,
      "loss": 1.4119,
      "step": 11171
    },
    {
      "epoch": 0.961197625397918,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009118626063050661,
      "loss": 1.0017,
      "step": 11172
    },
    {
      "epoch": 0.9612836617052396,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000911847273897255,
      "loss": 1.2621,
      "step": 11173
    },
    {
      "epoch": 0.9613696980125613,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009118319402848708,
      "loss": 1.2933,
      "step": 11174
    },
    {
      "epoch": 0.961455734319883,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009118166054679582,
      "loss": 0.7631,
      "step": 11175
    },
    {
      "epoch": 0.9615417706272047,
      "grad_norm": 1.8046875,
      "learning_rate": 0.000911801269446562,
      "loss": 2.0598,
      "step": 11176
    },
    {
      "epoch": 0.9616278069345263,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009117859322207274,
      "loss": 1.9044,
      "step": 11177
    },
    {
      "epoch": 0.961713843241848,
      "grad_norm": 1.375,
      "learning_rate": 0.000911770593790499,
      "loss": 1.2363,
      "step": 11178
    },
    {
      "epoch": 0.9617998795491698,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009117552541559216,
      "loss": 1.3786,
      "step": 11179
    },
    {
      "epoch": 0.9618859158564914,
      "grad_norm": 2.34375,
      "learning_rate": 0.0009117399133170402,
      "loss": 1.1941,
      "step": 11180
    },
    {
      "epoch": 0.9619719521638131,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009117245712738996,
      "loss": 2.1451,
      "step": 11181
    },
    {
      "epoch": 0.9620579884711348,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009117092280265447,
      "loss": 1.0545,
      "step": 11182
    },
    {
      "epoch": 0.9621440247784565,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009116938835750204,
      "loss": 0.706,
      "step": 11183
    },
    {
      "epoch": 0.9622300610857782,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009116785379193718,
      "loss": 1.8921,
      "step": 11184
    },
    {
      "epoch": 0.9623160973930999,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009116631910596431,
      "loss": 0.8954,
      "step": 11185
    },
    {
      "epoch": 0.9624021337004216,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009116478429958799,
      "loss": 1.5938,
      "step": 11186
    },
    {
      "epoch": 0.9624881700077432,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009116324937281269,
      "loss": 1.0518,
      "step": 11187
    },
    {
      "epoch": 0.962574206315065,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009116171432564286,
      "loss": 1.5502,
      "step": 11188
    },
    {
      "epoch": 0.9626602426223867,
      "grad_norm": 1.703125,
      "learning_rate": 0.0009116017915808306,
      "loss": 1.4498,
      "step": 11189
    },
    {
      "epoch": 0.9627462789297083,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000911586438701377,
      "loss": 1.9372,
      "step": 11190
    },
    {
      "epoch": 0.9628323152370301,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009115710846181134,
      "loss": 0.9426,
      "step": 11191
    },
    {
      "epoch": 0.9629183515443517,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009115557293310843,
      "loss": 1.2569,
      "step": 11192
    },
    {
      "epoch": 0.9630043878516734,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009115403728403348,
      "loss": 1.2644,
      "step": 11193
    },
    {
      "epoch": 0.9630904241589952,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0009115250151459098,
      "loss": 0.9778,
      "step": 11194
    },
    {
      "epoch": 0.9631764604663168,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009115096562478541,
      "loss": 1.0202,
      "step": 11195
    },
    {
      "epoch": 0.9632624967736385,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009114942961462127,
      "loss": 1.4306,
      "step": 11196
    },
    {
      "epoch": 0.9633485330809601,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009114789348410305,
      "loss": 1.8722,
      "step": 11197
    },
    {
      "epoch": 0.9634345693882819,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009114635723323526,
      "loss": 1.7388,
      "step": 11198
    },
    {
      "epoch": 0.9635206056956035,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009114482086202235,
      "loss": 1.8178,
      "step": 11199
    },
    {
      "epoch": 0.9636066420029252,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009114328437046886,
      "loss": 1.2667,
      "step": 11200
    },
    {
      "epoch": 0.963692678310247,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009114174775857925,
      "loss": 0.7434,
      "step": 11201
    },
    {
      "epoch": 0.9637787146175686,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009114021102635805,
      "loss": 1.2994,
      "step": 11202
    },
    {
      "epoch": 0.9638647509248903,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009113867417380972,
      "loss": 2.0627,
      "step": 11203
    },
    {
      "epoch": 0.963950787232212,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009113713720093877,
      "loss": 1.9725,
      "step": 11204
    },
    {
      "epoch": 0.9640368235395337,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009113560010774969,
      "loss": 0.8761,
      "step": 11205
    },
    {
      "epoch": 0.9641228598468554,
      "grad_norm": 1.734375,
      "learning_rate": 0.0009113406289424699,
      "loss": 1.9993,
      "step": 11206
    },
    {
      "epoch": 0.964208896154177,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009113252556043515,
      "loss": 1.1047,
      "step": 11207
    },
    {
      "epoch": 0.9642949324614988,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009113098810631868,
      "loss": 1.2291,
      "step": 11208
    },
    {
      "epoch": 0.9643809687688204,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009112945053190206,
      "loss": 1.3718,
      "step": 11209
    },
    {
      "epoch": 0.9644670050761421,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009112791283718979,
      "loss": 1.0145,
      "step": 11210
    },
    {
      "epoch": 0.9645530413834639,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000911263750221864,
      "loss": 1.1416,
      "step": 11211
    },
    {
      "epoch": 0.9646390776907855,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009112483708689634,
      "loss": 1.4749,
      "step": 11212
    },
    {
      "epoch": 0.9647251139981072,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009112329903132413,
      "loss": 1.197,
      "step": 11213
    },
    {
      "epoch": 0.9648111503054289,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009112176085547428,
      "loss": 1.1228,
      "step": 11214
    },
    {
      "epoch": 0.9648971866127506,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009112022255935127,
      "loss": 1.9187,
      "step": 11215
    },
    {
      "epoch": 0.9649832229200723,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009111868414295961,
      "loss": 0.712,
      "step": 11216
    },
    {
      "epoch": 0.965069259227394,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009111714560630379,
      "loss": 1.6898,
      "step": 11217
    },
    {
      "epoch": 0.9651552955347157,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009111560694938832,
      "loss": 1.2577,
      "step": 11218
    },
    {
      "epoch": 0.9652413318420373,
      "grad_norm": 1.28125,
      "learning_rate": 0.000911140681722177,
      "loss": 1.198,
      "step": 11219
    },
    {
      "epoch": 0.965327368149359,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009111252927479643,
      "loss": 0.752,
      "step": 11220
    },
    {
      "epoch": 0.9654134044566807,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009111099025712898,
      "loss": 0.5646,
      "step": 11221
    },
    {
      "epoch": 0.9654994407640024,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009110945111921991,
      "loss": 1.759,
      "step": 11222
    },
    {
      "epoch": 0.9655854770713241,
      "grad_norm": 2.5,
      "learning_rate": 0.0009110791186107369,
      "loss": 1.6317,
      "step": 11223
    },
    {
      "epoch": 0.9656715133786458,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009110637248269481,
      "loss": 1.5262,
      "step": 11224
    },
    {
      "epoch": 0.9657575496859675,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009110483298408778,
      "loss": 1.8151,
      "step": 11225
    },
    {
      "epoch": 0.9658435859932891,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009110329336525714,
      "loss": 1.1122,
      "step": 11226
    },
    {
      "epoch": 0.9659296223006109,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009110175362620731,
      "loss": 1.5234,
      "step": 11227
    },
    {
      "epoch": 0.9660156586079326,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009110021376694289,
      "loss": 1.7851,
      "step": 11228
    },
    {
      "epoch": 0.9661016949152542,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009109867378746833,
      "loss": 1.3047,
      "step": 11229
    },
    {
      "epoch": 0.966187731222576,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009109713368778813,
      "loss": 1.2105,
      "step": 11230
    },
    {
      "epoch": 0.9662737675298976,
      "grad_norm": 0.96875,
      "learning_rate": 0.000910955934679068,
      "loss": 1.0708,
      "step": 11231
    },
    {
      "epoch": 0.9663598038372193,
      "grad_norm": 1.0,
      "learning_rate": 0.0009109405312782887,
      "loss": 1.0764,
      "step": 11232
    },
    {
      "epoch": 0.966445840144541,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009109251266755881,
      "loss": 1.4689,
      "step": 11233
    },
    {
      "epoch": 0.9665318764518627,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009109097208710115,
      "loss": 1.7641,
      "step": 11234
    },
    {
      "epoch": 0.9666179127591844,
      "grad_norm": 1.203125,
      "learning_rate": 0.000910894313864604,
      "loss": 1.0185,
      "step": 11235
    },
    {
      "epoch": 0.966703949066506,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009108789056564104,
      "loss": 1.8674,
      "step": 11236
    },
    {
      "epoch": 0.9667899853738278,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009108634962464759,
      "loss": 0.8052,
      "step": 11237
    },
    {
      "epoch": 0.9668760216811495,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009108480856348457,
      "loss": 1.7355,
      "step": 11238
    },
    {
      "epoch": 0.9669620579884711,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009108326738215646,
      "loss": 1.2337,
      "step": 11239
    },
    {
      "epoch": 0.9670480942957929,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009108172608066781,
      "loss": 1.6361,
      "step": 11240
    },
    {
      "epoch": 0.9671341306031145,
      "grad_norm": 1.125,
      "learning_rate": 0.0009108018465902308,
      "loss": 2.058,
      "step": 11241
    },
    {
      "epoch": 0.9672201669104362,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000910786431172268,
      "loss": 1.6168,
      "step": 11242
    },
    {
      "epoch": 0.9673062032177578,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009107710145528347,
      "loss": 0.9077,
      "step": 11243
    },
    {
      "epoch": 0.9673922395250796,
      "grad_norm": 1.25,
      "learning_rate": 0.0009107555967319763,
      "loss": 1.6263,
      "step": 11244
    },
    {
      "epoch": 0.9674782758324013,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009107401777097375,
      "loss": 1.5949,
      "step": 11245
    },
    {
      "epoch": 0.9675643121397229,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009107247574861637,
      "loss": 1.2919,
      "step": 11246
    },
    {
      "epoch": 0.9676503484470447,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009107093360612997,
      "loss": 1.4376,
      "step": 11247
    },
    {
      "epoch": 0.9677363847543663,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009106939134351908,
      "loss": 1.4555,
      "step": 11248
    },
    {
      "epoch": 0.967822421061688,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009106784896078822,
      "loss": 1.6512,
      "step": 11249
    },
    {
      "epoch": 0.9679084573690098,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009106630645794189,
      "loss": 1.3365,
      "step": 11250
    },
    {
      "epoch": 0.9679944936763314,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009106476383498459,
      "loss": 0.9074,
      "step": 11251
    },
    {
      "epoch": 0.9680805299836531,
      "grad_norm": 1.375,
      "learning_rate": 0.0009106322109192084,
      "loss": 1.0104,
      "step": 11252
    },
    {
      "epoch": 0.9681665662909748,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009106167822875516,
      "loss": 1.5,
      "step": 11253
    },
    {
      "epoch": 0.9682526025982965,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009106013524549206,
      "loss": 1.1614,
      "step": 11254
    },
    {
      "epoch": 0.9683386389056182,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009105859214213605,
      "loss": 1.2594,
      "step": 11255
    },
    {
      "epoch": 0.9684246752129398,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009105704891869164,
      "loss": 1.3455,
      "step": 11256
    },
    {
      "epoch": 0.9685107115202616,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009105550557516335,
      "loss": 1.1775,
      "step": 11257
    },
    {
      "epoch": 0.9685967478275832,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009105396211155569,
      "loss": 1.4822,
      "step": 11258
    },
    {
      "epoch": 0.9686827841349049,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009105241852787317,
      "loss": 0.8706,
      "step": 11259
    },
    {
      "epoch": 0.9687688204422267,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009105087482412031,
      "loss": 1.6607,
      "step": 11260
    },
    {
      "epoch": 0.9688548567495483,
      "grad_norm": 1.25,
      "learning_rate": 0.0009104933100030164,
      "loss": 1.8146,
      "step": 11261
    },
    {
      "epoch": 0.96894089305687,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009104778705642164,
      "loss": 1.5017,
      "step": 11262
    },
    {
      "epoch": 0.9690269293641917,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009104624299248487,
      "loss": 0.9923,
      "step": 11263
    },
    {
      "epoch": 0.9691129656715134,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000910446988084958,
      "loss": 1.0827,
      "step": 11264
    },
    {
      "epoch": 0.969199001978835,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009104315450445898,
      "loss": 0.8296,
      "step": 11265
    },
    {
      "epoch": 0.9692850382861568,
      "grad_norm": 1.71875,
      "learning_rate": 0.0009104161008037891,
      "loss": 1.4529,
      "step": 11266
    },
    {
      "epoch": 0.9693710745934785,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009104006553626012,
      "loss": 1.1386,
      "step": 11267
    },
    {
      "epoch": 0.9694571109008001,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009103852087210711,
      "loss": 0.9901,
      "step": 11268
    },
    {
      "epoch": 0.9695431472081218,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009103697608792441,
      "loss": 1.562,
      "step": 11269
    },
    {
      "epoch": 0.9696291835154435,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0009103543118371655,
      "loss": 0.6886,
      "step": 11270
    },
    {
      "epoch": 0.9697152198227652,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009103388615948802,
      "loss": 1.8873,
      "step": 11271
    },
    {
      "epoch": 0.9698012561300869,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009103234101524336,
      "loss": 1.5805,
      "step": 11272
    },
    {
      "epoch": 0.9698872924374086,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009103079575098708,
      "loss": 1.2323,
      "step": 11273
    },
    {
      "epoch": 0.9699733287447303,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009102925036672371,
      "loss": 1.2268,
      "step": 11274
    },
    {
      "epoch": 0.9700593650520519,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009102770486245775,
      "loss": 1.1291,
      "step": 11275
    },
    {
      "epoch": 0.9701454013593737,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009102615923819374,
      "loss": 1.503,
      "step": 11276
    },
    {
      "epoch": 0.9702314376666954,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000910246134939362,
      "loss": 1.964,
      "step": 11277
    },
    {
      "epoch": 0.970317473974017,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009102306762968962,
      "loss": 1.3304,
      "step": 11278
    },
    {
      "epoch": 0.9704035102813388,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009102152164545857,
      "loss": 1.1014,
      "step": 11279
    },
    {
      "epoch": 0.9704895465886604,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009101997554124755,
      "loss": 1.5221,
      "step": 11280
    },
    {
      "epoch": 0.9705755828959821,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009101842931706106,
      "loss": 0.8117,
      "step": 11281
    },
    {
      "epoch": 0.9706616192033038,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009101688297290365,
      "loss": 2.1145,
      "step": 11282
    },
    {
      "epoch": 0.9707476555106255,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009101533650877983,
      "loss": 1.9452,
      "step": 11283
    },
    {
      "epoch": 0.9708336918179472,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009101378992469413,
      "loss": 1.35,
      "step": 11284
    },
    {
      "epoch": 0.9709197281252688,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009101224322065109,
      "loss": 1.4655,
      "step": 11285
    },
    {
      "epoch": 0.9710057644325906,
      "grad_norm": 1.453125,
      "learning_rate": 0.000910106963966552,
      "loss": 1.4966,
      "step": 11286
    },
    {
      "epoch": 0.9710918007399122,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009100914945271099,
      "loss": 1.2496,
      "step": 11287
    },
    {
      "epoch": 0.9711778370472339,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00091007602388823,
      "loss": 0.9208,
      "step": 11288
    },
    {
      "epoch": 0.9712638733545557,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009100605520499576,
      "loss": 1.7242,
      "step": 11289
    },
    {
      "epoch": 0.9713499096618773,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009100450790123377,
      "loss": 1.1899,
      "step": 11290
    },
    {
      "epoch": 0.971435945969199,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009100296047754158,
      "loss": 0.9933,
      "step": 11291
    },
    {
      "epoch": 0.9715219822765206,
      "grad_norm": 1.46875,
      "learning_rate": 0.000910014129339237,
      "loss": 1.0788,
      "step": 11292
    },
    {
      "epoch": 0.9716080185838424,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009099986527038467,
      "loss": 1.0501,
      "step": 11293
    },
    {
      "epoch": 0.9716940548911641,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009099831748692898,
      "loss": 1.4326,
      "step": 11294
    },
    {
      "epoch": 0.9717800911984857,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009099676958356121,
      "loss": 0.973,
      "step": 11295
    },
    {
      "epoch": 0.9718661275058075,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009099522156028586,
      "loss": 1.4695,
      "step": 11296
    },
    {
      "epoch": 0.9719521638131291,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009099367341710745,
      "loss": 0.7304,
      "step": 11297
    },
    {
      "epoch": 0.9720382001204508,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009099212515403052,
      "loss": 0.8854,
      "step": 11298
    },
    {
      "epoch": 0.9721242364277726,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009099057677105962,
      "loss": 0.8453,
      "step": 11299
    },
    {
      "epoch": 0.9722102727350942,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009098902826819923,
      "loss": 1.3976,
      "step": 11300
    },
    {
      "epoch": 0.9722963090424159,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009098747964545391,
      "loss": 0.9094,
      "step": 11301
    },
    {
      "epoch": 0.9723823453497376,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009098593090282819,
      "loss": 1.6809,
      "step": 11302
    },
    {
      "epoch": 0.9724683816570593,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009098438204032657,
      "loss": 1.7475,
      "step": 11303
    },
    {
      "epoch": 0.972554417964381,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009098283305795363,
      "loss": 1.0664,
      "step": 11304
    },
    {
      "epoch": 0.9726404542717026,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009098128395571386,
      "loss": 1.1986,
      "step": 11305
    },
    {
      "epoch": 0.9727264905790244,
      "grad_norm": 1.375,
      "learning_rate": 0.000909797347336118,
      "loss": 1.6341,
      "step": 11306
    },
    {
      "epoch": 0.972812526886346,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009097818539165201,
      "loss": 1.4353,
      "step": 11307
    },
    {
      "epoch": 0.9728985631936677,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009097663592983898,
      "loss": 1.1457,
      "step": 11308
    },
    {
      "epoch": 0.9729845995009894,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009097508634817726,
      "loss": 0.8384,
      "step": 11309
    },
    {
      "epoch": 0.9730706358083111,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009097353664667138,
      "loss": 0.8417,
      "step": 11310
    },
    {
      "epoch": 0.9731566721156328,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009097198682532589,
      "loss": 1.0047,
      "step": 11311
    },
    {
      "epoch": 0.9732427084229545,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000909704368841453,
      "loss": 1.0417,
      "step": 11312
    },
    {
      "epoch": 0.9733287447302762,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009096888682313415,
      "loss": 1.0501,
      "step": 11313
    },
    {
      "epoch": 0.9734147810375978,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009096733664229697,
      "loss": 0.9334,
      "step": 11314
    },
    {
      "epoch": 0.9735008173449196,
      "grad_norm": 1.203125,
      "learning_rate": 0.000909657863416383,
      "loss": 0.7545,
      "step": 11315
    },
    {
      "epoch": 0.9735868536522413,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009096423592116266,
      "loss": 1.3543,
      "step": 11316
    },
    {
      "epoch": 0.9736728899595629,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0009096268538087461,
      "loss": 1.0032,
      "step": 11317
    },
    {
      "epoch": 0.9737589262668847,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009096113472077867,
      "loss": 1.3798,
      "step": 11318
    },
    {
      "epoch": 0.9738449625742063,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009095958394087938,
      "loss": 0.729,
      "step": 11319
    },
    {
      "epoch": 0.973930998881528,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009095803304118126,
      "loss": 1.3646,
      "step": 11320
    },
    {
      "epoch": 0.9740170351888497,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009095648202168888,
      "loss": 1.0449,
      "step": 11321
    },
    {
      "epoch": 0.9741030714961714,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009095493088240674,
      "loss": 2.2377,
      "step": 11322
    },
    {
      "epoch": 0.9741891078034931,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000909533796233394,
      "loss": 0.7832,
      "step": 11323
    },
    {
      "epoch": 0.9742751441108147,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009095182824449139,
      "loss": 0.843,
      "step": 11324
    },
    {
      "epoch": 0.9743611804181365,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009095027674586724,
      "loss": 1.0524,
      "step": 11325
    },
    {
      "epoch": 0.9744472167254582,
      "grad_norm": 1.265625,
      "learning_rate": 0.000909487251274715,
      "loss": 1.5766,
      "step": 11326
    },
    {
      "epoch": 0.9745332530327798,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009094717338930872,
      "loss": 1.5016,
      "step": 11327
    },
    {
      "epoch": 0.9746192893401016,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009094562153138339,
      "loss": 1.7845,
      "step": 11328
    },
    {
      "epoch": 0.9747053256474232,
      "grad_norm": 1.5,
      "learning_rate": 0.0009094406955370009,
      "loss": 1.2956,
      "step": 11329
    },
    {
      "epoch": 0.9747913619547449,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009094251745626336,
      "loss": 1.4284,
      "step": 11330
    },
    {
      "epoch": 0.9748773982620665,
      "grad_norm": 0.921875,
      "learning_rate": 0.0009094096523907774,
      "loss": 0.8197,
      "step": 11331
    },
    {
      "epoch": 0.9749634345693883,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009093941290214773,
      "loss": 1.8824,
      "step": 11332
    },
    {
      "epoch": 0.97504947087671,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009093786044547792,
      "loss": 0.8087,
      "step": 11333
    },
    {
      "epoch": 0.9751355071840316,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009093630786907281,
      "loss": 0.8771,
      "step": 11334
    },
    {
      "epoch": 0.9752215434913534,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009093475517293698,
      "loss": 0.8448,
      "step": 11335
    },
    {
      "epoch": 0.975307579798675,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009093320235707494,
      "loss": 1.692,
      "step": 11336
    },
    {
      "epoch": 0.9753936161059967,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009093164942149126,
      "loss": 1.2365,
      "step": 11337
    },
    {
      "epoch": 0.9754796524133185,
      "grad_norm": 2.515625,
      "learning_rate": 0.0009093009636619045,
      "loss": 1.1886,
      "step": 11338
    },
    {
      "epoch": 0.9755656887206401,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009092854319117707,
      "loss": 1.2657,
      "step": 11339
    },
    {
      "epoch": 0.9756517250279618,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009092698989645568,
      "loss": 2.1667,
      "step": 11340
    },
    {
      "epoch": 0.9757377613352834,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009092543648203078,
      "loss": 1.2517,
      "step": 11341
    },
    {
      "epoch": 0.9758237976426052,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009092388294790696,
      "loss": 1.2562,
      "step": 11342
    },
    {
      "epoch": 0.9759098339499269,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009092232929408873,
      "loss": 0.9613,
      "step": 11343
    },
    {
      "epoch": 0.9759958702572485,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009092077552058064,
      "loss": 0.9503,
      "step": 11344
    },
    {
      "epoch": 0.9760819065645703,
      "grad_norm": 2.0,
      "learning_rate": 0.0009091922162738725,
      "loss": 1.0965,
      "step": 11345
    },
    {
      "epoch": 0.9761679428718919,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009091766761451309,
      "loss": 1.6833,
      "step": 11346
    },
    {
      "epoch": 0.9762539791792136,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009091611348196272,
      "loss": 1.8465,
      "step": 11347
    },
    {
      "epoch": 0.9763400154865354,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009091455922974065,
      "loss": 1.1297,
      "step": 11348
    },
    {
      "epoch": 0.976426051793857,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009091300485785148,
      "loss": 1.361,
      "step": 11349
    },
    {
      "epoch": 0.9765120881011787,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009091145036629971,
      "loss": 1.3931,
      "step": 11350
    },
    {
      "epoch": 0.9765981244085004,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009090989575508993,
      "loss": 2.2992,
      "step": 11351
    },
    {
      "epoch": 0.9766841607158221,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009090834102422662,
      "loss": 0.8574,
      "step": 11352
    },
    {
      "epoch": 0.9767701970231437,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000909067861737144,
      "loss": 1.7423,
      "step": 11353
    },
    {
      "epoch": 0.9768562333304655,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0009090523120355779,
      "loss": 1.3679,
      "step": 11354
    },
    {
      "epoch": 0.9769422696377872,
      "grad_norm": 1.125,
      "learning_rate": 0.0009090367611376131,
      "loss": 1.3961,
      "step": 11355
    },
    {
      "epoch": 0.9770283059451088,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009090212090432955,
      "loss": 1.1304,
      "step": 11356
    },
    {
      "epoch": 0.9771143422524305,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009090056557526704,
      "loss": 0.9836,
      "step": 11357
    },
    {
      "epoch": 0.9772003785597522,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009089901012657833,
      "loss": 1.7195,
      "step": 11358
    },
    {
      "epoch": 0.9772864148670739,
      "grad_norm": 1.859375,
      "learning_rate": 0.0009089745455826796,
      "loss": 1.6735,
      "step": 11359
    },
    {
      "epoch": 0.9773724511743956,
      "grad_norm": 1.359375,
      "learning_rate": 0.000908958988703405,
      "loss": 1.151,
      "step": 11360
    },
    {
      "epoch": 0.9774584874817173,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009089434306280049,
      "loss": 2.1851,
      "step": 11361
    },
    {
      "epoch": 0.977544523789039,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0009089278713565248,
      "loss": 1.0245,
      "step": 11362
    },
    {
      "epoch": 0.9776305600963606,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009089123108890101,
      "loss": 1.1564,
      "step": 11363
    },
    {
      "epoch": 0.9777165964036824,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009088967492255065,
      "loss": 1.228,
      "step": 11364
    },
    {
      "epoch": 0.9778026327110041,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009088811863660595,
      "loss": 1.6304,
      "step": 11365
    },
    {
      "epoch": 0.9778886690183257,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009088656223107144,
      "loss": 0.9339,
      "step": 11366
    },
    {
      "epoch": 0.9779747053256475,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009088500570595169,
      "loss": 1.6163,
      "step": 11367
    },
    {
      "epoch": 0.9780607416329691,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0009088344906125126,
      "loss": 1.0918,
      "step": 11368
    },
    {
      "epoch": 0.9781467779402908,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009088189229697469,
      "loss": 0.5149,
      "step": 11369
    },
    {
      "epoch": 0.9782328142476125,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009088033541312653,
      "loss": 1.9154,
      "step": 11370
    },
    {
      "epoch": 0.9783188505549342,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009087877840971135,
      "loss": 1.6018,
      "step": 11371
    },
    {
      "epoch": 0.9784048868622559,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009087722128673369,
      "loss": 1.2669,
      "step": 11372
    },
    {
      "epoch": 0.9784909231695775,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009087566404419811,
      "loss": 2.0145,
      "step": 11373
    },
    {
      "epoch": 0.9785769594768993,
      "grad_norm": 1.0,
      "learning_rate": 0.0009087410668210914,
      "loss": 1.3128,
      "step": 11374
    },
    {
      "epoch": 0.9786629957842209,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009087254920047139,
      "loss": 0.9643,
      "step": 11375
    },
    {
      "epoch": 0.9787490320915426,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009087099159928937,
      "loss": 1.3554,
      "step": 11376
    },
    {
      "epoch": 0.9788350683988644,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009086943387856764,
      "loss": 1.3207,
      "step": 11377
    },
    {
      "epoch": 0.978921104706186,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009086787603831077,
      "loss": 1.0194,
      "step": 11378
    },
    {
      "epoch": 0.9790071410135077,
      "grad_norm": 6.5625,
      "learning_rate": 0.0009086631807852332,
      "loss": 1.4943,
      "step": 11379
    },
    {
      "epoch": 0.9790931773208293,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009086475999920983,
      "loss": 0.9203,
      "step": 11380
    },
    {
      "epoch": 0.9791792136281511,
      "grad_norm": 1.75,
      "learning_rate": 0.0009086320180037486,
      "loss": 1.593,
      "step": 11381
    },
    {
      "epoch": 0.9792652499354728,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009086164348202298,
      "loss": 2.0132,
      "step": 11382
    },
    {
      "epoch": 0.9793512862427944,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009086008504415875,
      "loss": 1.0219,
      "step": 11383
    },
    {
      "epoch": 0.9794373225501162,
      "grad_norm": 1.109375,
      "learning_rate": 0.000908585264867867,
      "loss": 1.6196,
      "step": 11384
    },
    {
      "epoch": 0.9795233588574378,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009085696780991141,
      "loss": 1.7166,
      "step": 11385
    },
    {
      "epoch": 0.9796093951647595,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009085540901353744,
      "loss": 1.2796,
      "step": 11386
    },
    {
      "epoch": 0.9796954314720813,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009085385009766934,
      "loss": 0.9819,
      "step": 11387
    },
    {
      "epoch": 0.9797814677794029,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009085229106231168,
      "loss": 1.4253,
      "step": 11388
    },
    {
      "epoch": 0.9798675040867246,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00090850731907469,
      "loss": 1.4355,
      "step": 11389
    },
    {
      "epoch": 0.9799535403940463,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009084917263314588,
      "loss": 1.8219,
      "step": 11390
    },
    {
      "epoch": 0.980039576701368,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009084761323934689,
      "loss": 1.3462,
      "step": 11391
    },
    {
      "epoch": 0.9801256130086897,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009084605372607656,
      "loss": 1.3699,
      "step": 11392
    },
    {
      "epoch": 0.9802116493160113,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009084449409333947,
      "loss": 0.9785,
      "step": 11393
    },
    {
      "epoch": 0.9802976856233331,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009084293434114017,
      "loss": 1.0446,
      "step": 11394
    },
    {
      "epoch": 0.9803837219306547,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009084137446948324,
      "loss": 1.6568,
      "step": 11395
    },
    {
      "epoch": 0.9804697582379764,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009083981447837324,
      "loss": 1.1981,
      "step": 11396
    },
    {
      "epoch": 0.9805557945452981,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009083825436781472,
      "loss": 1.6348,
      "step": 11397
    },
    {
      "epoch": 0.9806418308526198,
      "grad_norm": 1.375,
      "learning_rate": 0.0009083669413781226,
      "loss": 1.9006,
      "step": 11398
    },
    {
      "epoch": 0.9807278671599415,
      "grad_norm": 0.765625,
      "learning_rate": 0.0009083513378837038,
      "loss": 1.2451,
      "step": 11399
    },
    {
      "epoch": 0.9808139034672632,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009083357331949371,
      "loss": 1.3307,
      "step": 11400
    },
    {
      "epoch": 0.9808999397745849,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009083201273118675,
      "loss": 1.2726,
      "step": 11401
    },
    {
      "epoch": 0.9809859760819065,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0009083045202345413,
      "loss": 0.8602,
      "step": 11402
    },
    {
      "epoch": 0.9810720123892283,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009082889119630034,
      "loss": 0.9807,
      "step": 11403
    },
    {
      "epoch": 0.98115804869655,
      "grad_norm": 1.0,
      "learning_rate": 0.0009082733024973001,
      "loss": 0.7796,
      "step": 11404
    },
    {
      "epoch": 0.9812440850038716,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009082576918374767,
      "loss": 1.1878,
      "step": 11405
    },
    {
      "epoch": 0.9813301213111933,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000908242079983579,
      "loss": 1.0649,
      "step": 11406
    },
    {
      "epoch": 0.981416157618515,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009082264669356526,
      "loss": 0.8836,
      "step": 11407
    },
    {
      "epoch": 0.9815021939258367,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009082108526937432,
      "loss": 1.8793,
      "step": 11408
    },
    {
      "epoch": 0.9815882302331584,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009081952372578964,
      "loss": 1.3499,
      "step": 11409
    },
    {
      "epoch": 0.9816742665404801,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009081796206281579,
      "loss": 2.018,
      "step": 11410
    },
    {
      "epoch": 0.9817603028478018,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009081640028045735,
      "loss": 1.3731,
      "step": 11411
    },
    {
      "epoch": 0.9818463391551234,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009081483837871887,
      "loss": 1.548,
      "step": 11412
    },
    {
      "epoch": 0.9819323754624452,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009081327635760493,
      "loss": 1.1126,
      "step": 11413
    },
    {
      "epoch": 0.9820184117697669,
      "grad_norm": 1.25,
      "learning_rate": 0.0009081171421712008,
      "loss": 1.3765,
      "step": 11414
    },
    {
      "epoch": 0.9821044480770885,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009081015195726892,
      "loss": 1.5636,
      "step": 11415
    },
    {
      "epoch": 0.9821904843844103,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009080858957805599,
      "loss": 1.6735,
      "step": 11416
    },
    {
      "epoch": 0.9822765206917319,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009080702707948588,
      "loss": 1.8095,
      "step": 11417
    },
    {
      "epoch": 0.9823625569990536,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009080546446156314,
      "loss": 1.6064,
      "step": 11418
    },
    {
      "epoch": 0.9824485933063752,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009080390172429238,
      "loss": 0.997,
      "step": 11419
    },
    {
      "epoch": 0.982534629613697,
      "grad_norm": 1.828125,
      "learning_rate": 0.0009080233886767813,
      "loss": 1.6122,
      "step": 11420
    },
    {
      "epoch": 0.9826206659210187,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009080077589172496,
      "loss": 1.312,
      "step": 11421
    },
    {
      "epoch": 0.9827067022283403,
      "grad_norm": 1.578125,
      "learning_rate": 0.0009079921279643747,
      "loss": 1.0869,
      "step": 11422
    },
    {
      "epoch": 0.9827927385356621,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009079764958182022,
      "loss": 1.9588,
      "step": 11423
    },
    {
      "epoch": 0.9828787748429837,
      "grad_norm": 1.6875,
      "learning_rate": 0.0009079608624787777,
      "loss": 1.3605,
      "step": 11424
    },
    {
      "epoch": 0.9829648111503054,
      "grad_norm": 8.6875,
      "learning_rate": 0.000907945227946147,
      "loss": 2.1185,
      "step": 11425
    },
    {
      "epoch": 0.9830508474576272,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009079295922203561,
      "loss": 1.6018,
      "step": 11426
    },
    {
      "epoch": 0.9831368837649488,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009079139553014501,
      "loss": 1.2499,
      "step": 11427
    },
    {
      "epoch": 0.9832229200722705,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0009078983171894755,
      "loss": 1.4192,
      "step": 11428
    },
    {
      "epoch": 0.9833089563795921,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009078826778844773,
      "loss": 1.4134,
      "step": 11429
    },
    {
      "epoch": 0.9833949926869139,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009078670373865019,
      "loss": 1.2567,
      "step": 11430
    },
    {
      "epoch": 0.9834810289942356,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009078513956955944,
      "loss": 1.4701,
      "step": 11431
    },
    {
      "epoch": 0.9835670653015572,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009078357528118014,
      "loss": 1.1544,
      "step": 11432
    },
    {
      "epoch": 0.983653101608879,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009078201087351676,
      "loss": 0.9338,
      "step": 11433
    },
    {
      "epoch": 0.9837391379162006,
      "grad_norm": 1.125,
      "learning_rate": 0.0009078044634657396,
      "loss": 0.5676,
      "step": 11434
    },
    {
      "epoch": 0.9838251742235223,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009077888170035629,
      "loss": 1.6819,
      "step": 11435
    },
    {
      "epoch": 0.9839112105308441,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000907773169348683,
      "loss": 1.1673,
      "step": 11436
    },
    {
      "epoch": 0.9839972468381657,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009077575205011461,
      "loss": 2.3948,
      "step": 11437
    },
    {
      "epoch": 0.9840832831454874,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009077418704609977,
      "loss": 1.0588,
      "step": 11438
    },
    {
      "epoch": 0.984169319452809,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009077262192282836,
      "loss": 0.9159,
      "step": 11439
    },
    {
      "epoch": 0.9842553557601308,
      "grad_norm": 0.890625,
      "learning_rate": 0.0009077105668030494,
      "loss": 1.2063,
      "step": 11440
    },
    {
      "epoch": 0.9843413920674524,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009076949131853414,
      "loss": 0.5378,
      "step": 11441
    },
    {
      "epoch": 0.9844274283747741,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009076792583752049,
      "loss": 1.797,
      "step": 11442
    },
    {
      "epoch": 0.9845134646820959,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000907663602372686,
      "loss": 1.5822,
      "step": 11443
    },
    {
      "epoch": 0.9845995009894175,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009076479451778302,
      "loss": 1.4343,
      "step": 11444
    },
    {
      "epoch": 0.9846855372967392,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009076322867906836,
      "loss": 1.3703,
      "step": 11445
    },
    {
      "epoch": 0.9847715736040609,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0009076166272112917,
      "loss": 1.3232,
      "step": 11446
    },
    {
      "epoch": 0.9848576099113826,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009076009664397005,
      "loss": 1.6575,
      "step": 11447
    },
    {
      "epoch": 0.9849436462187043,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009075853044759557,
      "loss": 1.0116,
      "step": 11448
    },
    {
      "epoch": 0.985029682526026,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009075696413201032,
      "loss": 1.2231,
      "step": 11449
    },
    {
      "epoch": 0.9851157188333477,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009075539769721888,
      "loss": 0.9389,
      "step": 11450
    },
    {
      "epoch": 0.9852017551406693,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0009075383114322582,
      "loss": 1.7483,
      "step": 11451
    },
    {
      "epoch": 0.9852877914479911,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009075226447003574,
      "loss": 0.9657,
      "step": 11452
    },
    {
      "epoch": 0.9853738277553128,
      "grad_norm": 1.46875,
      "learning_rate": 0.000907506976776532,
      "loss": 1.9488,
      "step": 11453
    },
    {
      "epoch": 0.9854598640626344,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000907491307660828,
      "loss": 1.5768,
      "step": 11454
    },
    {
      "epoch": 0.9855459003699562,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009074756373532913,
      "loss": 1.7063,
      "step": 11455
    },
    {
      "epoch": 0.9856319366772778,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009074599658539676,
      "loss": 1.2031,
      "step": 11456
    },
    {
      "epoch": 0.9857179729845995,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009074442931629026,
      "loss": 1.3355,
      "step": 11457
    },
    {
      "epoch": 0.9858040092919212,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009074286192801422,
      "loss": 1.1314,
      "step": 11458
    },
    {
      "epoch": 0.9858900455992429,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0009074129442057325,
      "loss": 2.1174,
      "step": 11459
    },
    {
      "epoch": 0.9859760819065646,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009073972679397191,
      "loss": 2.1756,
      "step": 11460
    },
    {
      "epoch": 0.9860621182138862,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009073815904821481,
      "loss": 1.1144,
      "step": 11461
    },
    {
      "epoch": 0.986148154521208,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009073659118330649,
      "loss": 1.3172,
      "step": 11462
    },
    {
      "epoch": 0.9862341908285296,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009073502319925158,
      "loss": 1.0933,
      "step": 11463
    },
    {
      "epoch": 0.9863202271358513,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009073345509605465,
      "loss": 0.9527,
      "step": 11464
    },
    {
      "epoch": 0.9864062634431731,
      "grad_norm": 1.0,
      "learning_rate": 0.0009073188687372029,
      "loss": 0.9981,
      "step": 11465
    },
    {
      "epoch": 0.9864922997504947,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009073031853225308,
      "loss": 1.1443,
      "step": 11466
    },
    {
      "epoch": 0.9865783360578164,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000907287500716576,
      "loss": 1.1981,
      "step": 11467
    },
    {
      "epoch": 0.986664372365138,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009072718149193845,
      "loss": 1.261,
      "step": 11468
    },
    {
      "epoch": 0.9867504086724598,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009072561279310023,
      "loss": 1.1902,
      "step": 11469
    },
    {
      "epoch": 0.9868364449797815,
      "grad_norm": 1.9140625,
      "learning_rate": 0.000907240439751475,
      "loss": 1.5605,
      "step": 11470
    },
    {
      "epoch": 0.9869224812871031,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009072247503808486,
      "loss": 0.6683,
      "step": 11471
    },
    {
      "epoch": 0.9870085175944249,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009072090598191691,
      "loss": 1.2131,
      "step": 11472
    },
    {
      "epoch": 0.9870945539017465,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009071933680664823,
      "loss": 1.1472,
      "step": 11473
    },
    {
      "epoch": 0.9871805902090682,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000907177675122834,
      "loss": 1.3414,
      "step": 11474
    },
    {
      "epoch": 0.98726662651639,
      "grad_norm": 2.0625,
      "learning_rate": 0.0009071619809882702,
      "loss": 2.1227,
      "step": 11475
    },
    {
      "epoch": 0.9873526628237116,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009071462856628368,
      "loss": 1.1025,
      "step": 11476
    },
    {
      "epoch": 0.9874386991310333,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009071305891465797,
      "loss": 0.9983,
      "step": 11477
    },
    {
      "epoch": 0.987524735438355,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009071148914395449,
      "loss": 2.1337,
      "step": 11478
    },
    {
      "epoch": 0.9876107717456767,
      "grad_norm": 1.4375,
      "learning_rate": 0.0009070991925417781,
      "loss": 1.4491,
      "step": 11479
    },
    {
      "epoch": 0.9876968080529984,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009070834924533254,
      "loss": 1.231,
      "step": 11480
    },
    {
      "epoch": 0.98778284436032,
      "grad_norm": 1.375,
      "learning_rate": 0.0009070677911742327,
      "loss": 1.2809,
      "step": 11481
    },
    {
      "epoch": 0.9878688806676418,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009070520887045458,
      "loss": 1.3441,
      "step": 11482
    },
    {
      "epoch": 0.9879549169749634,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009070363850443107,
      "loss": 1.3115,
      "step": 11483
    },
    {
      "epoch": 0.9880409532822851,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009070206801935735,
      "loss": 1.2083,
      "step": 11484
    },
    {
      "epoch": 0.9881269895896068,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009070049741523798,
      "loss": 1.4369,
      "step": 11485
    },
    {
      "epoch": 0.9882130258969285,
      "grad_norm": 1.625,
      "learning_rate": 0.0009069892669207758,
      "loss": 1.1907,
      "step": 11486
    },
    {
      "epoch": 0.9882990622042502,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009069735584988073,
      "loss": 1.5736,
      "step": 11487
    },
    {
      "epoch": 0.9883850985115719,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009069578488865202,
      "loss": 1.7315,
      "step": 11488
    },
    {
      "epoch": 0.9884711348188936,
      "grad_norm": 1.953125,
      "learning_rate": 0.0009069421380839606,
      "loss": 2.0677,
      "step": 11489
    },
    {
      "epoch": 0.9885571711262152,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009069264260911745,
      "loss": 0.9593,
      "step": 11490
    },
    {
      "epoch": 0.988643207433537,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0009069107129082076,
      "loss": 1.5468,
      "step": 11491
    },
    {
      "epoch": 0.9887292437408587,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009068949985351063,
      "loss": 0.6308,
      "step": 11492
    },
    {
      "epoch": 0.9888152800481803,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000906879282971916,
      "loss": 0.8817,
      "step": 11493
    },
    {
      "epoch": 0.988901316355502,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009068635662186829,
      "loss": 1.6255,
      "step": 11494
    },
    {
      "epoch": 0.9889873526628237,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009068478482754531,
      "loss": 1.9875,
      "step": 11495
    },
    {
      "epoch": 0.9890733889701454,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009068321291422725,
      "loss": 0.8276,
      "step": 11496
    },
    {
      "epoch": 0.9891594252774671,
      "grad_norm": 1.203125,
      "learning_rate": 0.000906816408819187,
      "loss": 1.3915,
      "step": 11497
    },
    {
      "epoch": 0.9892454615847888,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009068006873062427,
      "loss": 2.0294,
      "step": 11498
    },
    {
      "epoch": 0.9893314978921105,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009067849646034854,
      "loss": 1.7109,
      "step": 11499
    },
    {
      "epoch": 0.9894175341994321,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009067692407109611,
      "loss": 1.7284,
      "step": 11500
    },
    {
      "epoch": 0.9895035705067539,
      "grad_norm": 2.25,
      "learning_rate": 0.0009067535156287161,
      "loss": 1.0769,
      "step": 11501
    },
    {
      "epoch": 0.9895896068140756,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009067377893567961,
      "loss": 0.9375,
      "step": 11502
    },
    {
      "epoch": 0.9896756431213972,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009067220618952472,
      "loss": 1.0376,
      "step": 11503
    },
    {
      "epoch": 0.989761679428719,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009067063332441152,
      "loss": 0.7985,
      "step": 11504
    },
    {
      "epoch": 0.9898477157360406,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009066906034034466,
      "loss": 1.1603,
      "step": 11505
    },
    {
      "epoch": 0.9899337520433623,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009066748723732867,
      "loss": 1.6041,
      "step": 11506
    },
    {
      "epoch": 0.9900197883506839,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009066591401536822,
      "loss": 1.261,
      "step": 11507
    },
    {
      "epoch": 0.9901058246580057,
      "grad_norm": 1.0,
      "learning_rate": 0.0009066434067446787,
      "loss": 0.9687,
      "step": 11508
    },
    {
      "epoch": 0.9901918609653274,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009066276721463222,
      "loss": 0.6675,
      "step": 11509
    },
    {
      "epoch": 0.990277897272649,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000906611936358659,
      "loss": 0.8276,
      "step": 11510
    },
    {
      "epoch": 0.9903639335799708,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009065961993817348,
      "loss": 1.3505,
      "step": 11511
    },
    {
      "epoch": 0.9904499698872924,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009065804612155961,
      "loss": 1.0194,
      "step": 11512
    },
    {
      "epoch": 0.9905360061946141,
      "grad_norm": 0.65625,
      "learning_rate": 0.0009065647218602883,
      "loss": 0.364,
      "step": 11513
    },
    {
      "epoch": 0.9906220425019359,
      "grad_norm": 1.125,
      "learning_rate": 0.0009065489813158581,
      "loss": 1.1975,
      "step": 11514
    },
    {
      "epoch": 0.9907080788092575,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009065332395823509,
      "loss": 1.0836,
      "step": 11515
    },
    {
      "epoch": 0.9907941151165792,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009065174966598131,
      "loss": 1.2883,
      "step": 11516
    },
    {
      "epoch": 0.9908801514239008,
      "grad_norm": 1.125,
      "learning_rate": 0.0009065017525482908,
      "loss": 1.6077,
      "step": 11517
    },
    {
      "epoch": 0.9909661877312226,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009064860072478298,
      "loss": 1.396,
      "step": 11518
    },
    {
      "epoch": 0.9910522240385443,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0009064702607584764,
      "loss": 1.3591,
      "step": 11519
    },
    {
      "epoch": 0.9911382603458659,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009064545130802764,
      "loss": 1.6609,
      "step": 11520
    },
    {
      "epoch": 0.9912242966531877,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009064387642132761,
      "loss": 1.561,
      "step": 11521
    },
    {
      "epoch": 0.9913103329605093,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009064230141575215,
      "loss": 1.3065,
      "step": 11522
    },
    {
      "epoch": 0.991396369267831,
      "grad_norm": 1.375,
      "learning_rate": 0.0009064072629130586,
      "loss": 1.5942,
      "step": 11523
    },
    {
      "epoch": 0.9914824055751528,
      "grad_norm": 1.75,
      "learning_rate": 0.0009063915104799334,
      "loss": 0.9654,
      "step": 11524
    },
    {
      "epoch": 0.9915684418824744,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009063757568581923,
      "loss": 1.414,
      "step": 11525
    },
    {
      "epoch": 0.9916544781897961,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009063600020478809,
      "loss": 1.0289,
      "step": 11526
    },
    {
      "epoch": 0.9917405144971178,
      "grad_norm": 1.65625,
      "learning_rate": 0.0009063442460490456,
      "loss": 1.8934,
      "step": 11527
    },
    {
      "epoch": 0.9918265508044395,
      "grad_norm": 1.375,
      "learning_rate": 0.0009063284888617324,
      "loss": 1.4718,
      "step": 11528
    },
    {
      "epoch": 0.9919125871117611,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0009063127304859875,
      "loss": 1.5691,
      "step": 11529
    },
    {
      "epoch": 0.9919986234190828,
      "grad_norm": 1.0,
      "learning_rate": 0.0009062969709218568,
      "loss": 0.8894,
      "step": 11530
    },
    {
      "epoch": 0.9920846597264046,
      "grad_norm": 1.75,
      "learning_rate": 0.0009062812101693864,
      "loss": 1.9339,
      "step": 11531
    },
    {
      "epoch": 0.9921706960337262,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009062654482286227,
      "loss": 0.9854,
      "step": 11532
    },
    {
      "epoch": 0.9922567323410479,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009062496850996115,
      "loss": 1.1732,
      "step": 11533
    },
    {
      "epoch": 0.9923427686483696,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009062339207823989,
      "loss": 1.0078,
      "step": 11534
    },
    {
      "epoch": 0.9924288049556913,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009062181552770311,
      "loss": 0.9673,
      "step": 11535
    },
    {
      "epoch": 0.992514841263013,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0009062023885835542,
      "loss": 1.7655,
      "step": 11536
    },
    {
      "epoch": 0.9926008775703347,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009061866207020144,
      "loss": 1.4907,
      "step": 11537
    },
    {
      "epoch": 0.9926869138776564,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009061708516324576,
      "loss": 1.5853,
      "step": 11538
    },
    {
      "epoch": 0.992772950184978,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0009061550813749302,
      "loss": 0.5926,
      "step": 11539
    },
    {
      "epoch": 0.9928589864922998,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009061393099294782,
      "loss": 0.8377,
      "step": 11540
    },
    {
      "epoch": 0.9929450227996215,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009061235372961475,
      "loss": 1.2975,
      "step": 11541
    },
    {
      "epoch": 0.9930310591069431,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009061077634749846,
      "loss": 1.307,
      "step": 11542
    },
    {
      "epoch": 0.9931170954142648,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009060919884660355,
      "loss": 1.0957,
      "step": 11543
    },
    {
      "epoch": 0.9932031317215865,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009060762122693462,
      "loss": 1.3988,
      "step": 11544
    },
    {
      "epoch": 0.9932891680289082,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000906060434884963,
      "loss": 0.9353,
      "step": 11545
    },
    {
      "epoch": 0.9933752043362299,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000906044656312932,
      "loss": 1.9361,
      "step": 11546
    },
    {
      "epoch": 0.9934612406435516,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009060288765532994,
      "loss": 0.9885,
      "step": 11547
    },
    {
      "epoch": 0.9935472769508733,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009060130956061112,
      "loss": 1.2191,
      "step": 11548
    },
    {
      "epoch": 0.9936333132581949,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009059973134714137,
      "loss": 1.4536,
      "step": 11549
    },
    {
      "epoch": 0.9937193495655167,
      "grad_norm": 1.84375,
      "learning_rate": 0.0009059815301492531,
      "loss": 0.8103,
      "step": 11550
    },
    {
      "epoch": 0.9938053858728383,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009059657456396753,
      "loss": 1.3159,
      "step": 11551
    },
    {
      "epoch": 0.99389142218016,
      "grad_norm": 1.125,
      "learning_rate": 0.0009059499599427267,
      "loss": 1.7465,
      "step": 11552
    },
    {
      "epoch": 0.9939774584874818,
      "grad_norm": 1.15625,
      "learning_rate": 0.0009059341730584536,
      "loss": 1.7374,
      "step": 11553
    },
    {
      "epoch": 0.9940634947948034,
      "grad_norm": 1.390625,
      "learning_rate": 0.000905918384986902,
      "loss": 0.8946,
      "step": 11554
    },
    {
      "epoch": 0.9941495311021251,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000905902595728118,
      "loss": 1.7715,
      "step": 11555
    },
    {
      "epoch": 0.9942355674094467,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009058868052821478,
      "loss": 1.4908,
      "step": 11556
    },
    {
      "epoch": 0.9943216037167685,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009058710136490376,
      "loss": 1.6204,
      "step": 11557
    },
    {
      "epoch": 0.9944076400240902,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009058552208288336,
      "loss": 1.5966,
      "step": 11558
    },
    {
      "epoch": 0.9944936763314118,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009058394268215822,
      "loss": 0.7146,
      "step": 11559
    },
    {
      "epoch": 0.9945797126387336,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009058236316273292,
      "loss": 1.1252,
      "step": 11560
    },
    {
      "epoch": 0.9946657489460552,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009058078352461212,
      "loss": 0.9043,
      "step": 11561
    },
    {
      "epoch": 0.9947517852533769,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000905792037678004,
      "loss": 1.0704,
      "step": 11562
    },
    {
      "epoch": 0.9948378215606987,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009057762389230242,
      "loss": 1.5461,
      "step": 11563
    },
    {
      "epoch": 0.9949238578680203,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009057604389812277,
      "loss": 1.2295,
      "step": 11564
    },
    {
      "epoch": 0.995009894175342,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009057446378526608,
      "loss": 2.1779,
      "step": 11565
    },
    {
      "epoch": 0.9950959304826636,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009057288355373698,
      "loss": 1.5416,
      "step": 11566
    },
    {
      "epoch": 0.9951819667899854,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000905713032035401,
      "loss": 1.186,
      "step": 11567
    },
    {
      "epoch": 0.9952680030973071,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009056972273468003,
      "loss": 0.6126,
      "step": 11568
    },
    {
      "epoch": 0.9953540394046287,
      "grad_norm": 1.890625,
      "learning_rate": 0.0009056814214716142,
      "loss": 1.4657,
      "step": 11569
    },
    {
      "epoch": 0.9954400757119505,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009056656144098888,
      "loss": 1.5517,
      "step": 11570
    },
    {
      "epoch": 0.9955261120192721,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009056498061616705,
      "loss": 0.9445,
      "step": 11571
    },
    {
      "epoch": 0.9956121483265938,
      "grad_norm": 1.25,
      "learning_rate": 0.0009056339967270053,
      "loss": 2.0985,
      "step": 11572
    },
    {
      "epoch": 0.9956981846339155,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009056181861059396,
      "loss": 1.1792,
      "step": 11573
    },
    {
      "epoch": 0.9957842209412372,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009056023742985195,
      "loss": 1.6794,
      "step": 11574
    },
    {
      "epoch": 0.9958702572485589,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009055865613047914,
      "loss": 1.864,
      "step": 11575
    },
    {
      "epoch": 0.9959562935558806,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009055707471248014,
      "loss": 0.9399,
      "step": 11576
    },
    {
      "epoch": 0.9960423298632023,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009055549317585961,
      "loss": 0.7717,
      "step": 11577
    },
    {
      "epoch": 0.9961283661705239,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009055391152062212,
      "loss": 1.6073,
      "step": 11578
    },
    {
      "epoch": 0.9962144024778457,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009055232974677234,
      "loss": 1.4154,
      "step": 11579
    },
    {
      "epoch": 0.9963004387851674,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009055074785431488,
      "loss": 1.1913,
      "step": 11580
    },
    {
      "epoch": 0.996386475092489,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009054916584325438,
      "loss": 1.1256,
      "step": 11581
    },
    {
      "epoch": 0.9964725113998107,
      "grad_norm": 1.25,
      "learning_rate": 0.0009054758371359544,
      "loss": 1.379,
      "step": 11582
    },
    {
      "epoch": 0.9965585477071324,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009054600146534271,
      "loss": 1.2256,
      "step": 11583
    },
    {
      "epoch": 0.9966445840144541,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009054441909850082,
      "loss": 1.1607,
      "step": 11584
    },
    {
      "epoch": 0.9967306203217758,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0009054283661307439,
      "loss": 1.862,
      "step": 11585
    },
    {
      "epoch": 0.9968166566290975,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009054125400906803,
      "loss": 1.0353,
      "step": 11586
    },
    {
      "epoch": 0.9969026929364192,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009053967128648639,
      "loss": 1.567,
      "step": 11587
    },
    {
      "epoch": 0.9969887292437408,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009053808844533411,
      "loss": 1.0519,
      "step": 11588
    },
    {
      "epoch": 0.9970747655510626,
      "grad_norm": 1.265625,
      "learning_rate": 0.000905365054856158,
      "loss": 2.1248,
      "step": 11589
    },
    {
      "epoch": 0.9971608018583843,
      "grad_norm": 1.25,
      "learning_rate": 0.000905349224073361,
      "loss": 1.5947,
      "step": 11590
    },
    {
      "epoch": 0.9972468381657059,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009053333921049962,
      "loss": 1.6705,
      "step": 11591
    },
    {
      "epoch": 0.9973328744730277,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009053175589511102,
      "loss": 1.1814,
      "step": 11592
    },
    {
      "epoch": 0.9974189107803493,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0009053017246117491,
      "loss": 1.3977,
      "step": 11593
    },
    {
      "epoch": 0.997504947087671,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0009052858890869593,
      "loss": 1.1753,
      "step": 11594
    },
    {
      "epoch": 0.9975909833949926,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009052700523767872,
      "loss": 1.8052,
      "step": 11595
    },
    {
      "epoch": 0.9976770197023144,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009052542144812789,
      "loss": 1.368,
      "step": 11596
    },
    {
      "epoch": 0.9977630560096361,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009052383754004808,
      "loss": 1.9663,
      "step": 11597
    },
    {
      "epoch": 0.9978490923169577,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009052225351344394,
      "loss": 0.7936,
      "step": 11598
    },
    {
      "epoch": 0.9979351286242795,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009052066936832008,
      "loss": 1.6478,
      "step": 11599
    },
    {
      "epoch": 0.9980211649316011,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009051908510468114,
      "loss": 0.9641,
      "step": 11600
    },
    {
      "epoch": 0.9981072012389228,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009051750072253176,
      "loss": 1.6371,
      "step": 11601
    },
    {
      "epoch": 0.9981932375462446,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009051591622187659,
      "loss": 1.43,
      "step": 11602
    },
    {
      "epoch": 0.9982792738535662,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009051433160272022,
      "loss": 1.3548,
      "step": 11603
    },
    {
      "epoch": 0.9983653101608879,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009051274686506732,
      "loss": 1.1729,
      "step": 11604
    },
    {
      "epoch": 0.9984513464682095,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009051116200892251,
      "loss": 1.5391,
      "step": 11605
    },
    {
      "epoch": 0.9985373827755313,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009050957703429043,
      "loss": 0.7691,
      "step": 11606
    },
    {
      "epoch": 0.998623419082853,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009050799194117572,
      "loss": 0.8663,
      "step": 11607
    },
    {
      "epoch": 0.9987094553901746,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009050640672958302,
      "loss": 1.0132,
      "step": 11608
    },
    {
      "epoch": 0.9987954916974964,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009050482139951694,
      "loss": 0.8823,
      "step": 11609
    },
    {
      "epoch": 0.998881528004818,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009050323595098214,
      "loss": 1.4048,
      "step": 11610
    },
    {
      "epoch": 0.9989675643121397,
      "grad_norm": 1.0,
      "learning_rate": 0.0009050165038398326,
      "loss": 1.3682,
      "step": 11611
    },
    {
      "epoch": 0.9990536006194615,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009050006469852492,
      "loss": 1.2193,
      "step": 11612
    },
    {
      "epoch": 0.9991396369267831,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009049847889461178,
      "loss": 1.5503,
      "step": 11613
    },
    {
      "epoch": 0.9992256732341048,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009049689297224846,
      "loss": 1.2619,
      "step": 11614
    },
    {
      "epoch": 0.9993117095414265,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009049530693143959,
      "loss": 1.3607,
      "step": 11615
    },
    {
      "epoch": 0.9993977458487482,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009049372077218985,
      "loss": 1.1879,
      "step": 11616
    },
    {
      "epoch": 0.9994837821560698,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009049213449450384,
      "loss": 1.9225,
      "step": 11617
    },
    {
      "epoch": 0.9995698184633915,
      "grad_norm": 1.390625,
      "learning_rate": 0.000904905480983862,
      "loss": 1.8268,
      "step": 11618
    },
    {
      "epoch": 0.9996558547707133,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009048896158384159,
      "loss": 1.6249,
      "step": 11619
    },
    {
      "epoch": 0.9997418910780349,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009048737495087465,
      "loss": 1.4071,
      "step": 11620
    },
    {
      "epoch": 0.9998279273853566,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009048578819949001,
      "loss": 1.4172,
      "step": 11621
    },
    {
      "epoch": 0.9999139636926783,
      "grad_norm": 1.46875,
      "learning_rate": 0.000904842013296923,
      "loss": 2.0587,
      "step": 11622
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0009048261434148619,
      "loss": 1.0895,
      "step": 11623
    },
    {
      "epoch": 1.0000860363073216,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000904810272348763,
      "loss": 1.5439,
      "step": 11624
    },
    {
      "epoch": 1.0000860363073216,
      "eval_loss": 1.291654348373413,
      "eval_runtime": 354.4201,
      "eval_samples_per_second": 13.117,
      "eval_steps_per_second": 1.642,
      "step": 11624
    },
    {
      "epoch": 1.0001720726146435,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009047944000986728,
      "loss": 0.723,
      "step": 11625
    },
    {
      "epoch": 1.000258108921965,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009047785266646378,
      "loss": 1.2145,
      "step": 11626
    },
    {
      "epoch": 1.0003441452292867,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009047626520467043,
      "loss": 1.3457,
      "step": 11627
    },
    {
      "epoch": 1.0004301815366083,
      "grad_norm": 1.125,
      "learning_rate": 0.0009047467762449186,
      "loss": 1.2778,
      "step": 11628
    },
    {
      "epoch": 1.0005162178439302,
      "grad_norm": 1.375,
      "learning_rate": 0.0009047308992593275,
      "loss": 1.2767,
      "step": 11629
    },
    {
      "epoch": 1.0006022541512518,
      "grad_norm": 1.640625,
      "learning_rate": 0.0009047150210899772,
      "loss": 1.2747,
      "step": 11630
    },
    {
      "epoch": 1.0006882904585734,
      "grad_norm": 1.125,
      "learning_rate": 0.000904699141736914,
      "loss": 0.9856,
      "step": 11631
    },
    {
      "epoch": 1.0007743267658953,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009046832612001847,
      "loss": 1.3338,
      "step": 11632
    },
    {
      "epoch": 1.000860363073217,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009046673794798354,
      "loss": 1.679,
      "step": 11633
    },
    {
      "epoch": 1.0009463993805385,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009046514965759129,
      "loss": 1.4247,
      "step": 11634
    },
    {
      "epoch": 1.0010324356878604,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009046356124884634,
      "loss": 1.1128,
      "step": 11635
    },
    {
      "epoch": 1.001118471995182,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009046197272175334,
      "loss": 1.5448,
      "step": 11636
    },
    {
      "epoch": 1.0012045083025036,
      "grad_norm": 1.375,
      "learning_rate": 0.0009046038407631695,
      "loss": 0.8469,
      "step": 11637
    },
    {
      "epoch": 1.0012905446098253,
      "grad_norm": 1.390625,
      "learning_rate": 0.000904587953125418,
      "loss": 1.2167,
      "step": 11638
    },
    {
      "epoch": 1.001376580917147,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009045720643043254,
      "loss": 1.6185,
      "step": 11639
    },
    {
      "epoch": 1.0014626172244687,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009045561742999382,
      "loss": 1.7038,
      "step": 11640
    },
    {
      "epoch": 1.0015486535317903,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009045402831123028,
      "loss": 1.1094,
      "step": 11641
    },
    {
      "epoch": 1.0016346898391122,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009045243907414659,
      "loss": 0.8705,
      "step": 11642
    },
    {
      "epoch": 1.0017207261464338,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009045084971874737,
      "loss": 1.2403,
      "step": 11643
    },
    {
      "epoch": 1.0018067624537554,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009044926024503729,
      "loss": 1.4483,
      "step": 11644
    },
    {
      "epoch": 1.001892798761077,
      "grad_norm": 1.40625,
      "learning_rate": 0.00090447670653021,
      "loss": 1.8968,
      "step": 11645
    },
    {
      "epoch": 1.001978835068399,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009044608094270312,
      "loss": 1.1972,
      "step": 11646
    },
    {
      "epoch": 1.0020648713757205,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009044449111408832,
      "loss": 1.0452,
      "step": 11647
    },
    {
      "epoch": 1.0021509076830422,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009044290116718126,
      "loss": 1.2957,
      "step": 11648
    },
    {
      "epoch": 1.002236943990364,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009044131110198658,
      "loss": 1.5923,
      "step": 11649
    },
    {
      "epoch": 1.0023229802976856,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0009043972091850893,
      "loss": 1.0115,
      "step": 11650
    },
    {
      "epoch": 1.0024090166050073,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009043813061675296,
      "loss": 1.3472,
      "step": 11651
    },
    {
      "epoch": 1.002495052912329,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009043654019672331,
      "loss": 1.0846,
      "step": 11652
    },
    {
      "epoch": 1.0025810892196507,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009043494965842467,
      "loss": 0.8462,
      "step": 11653
    },
    {
      "epoch": 1.0026671255269723,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009043335900186164,
      "loss": 1.4562,
      "step": 11654
    },
    {
      "epoch": 1.002753161834294,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009043176822703891,
      "loss": 1.3069,
      "step": 11655
    },
    {
      "epoch": 1.0028391981416158,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009043017733396112,
      "loss": 1.8037,
      "step": 11656
    },
    {
      "epoch": 1.0029252344489374,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009042858632263293,
      "loss": 0.931,
      "step": 11657
    },
    {
      "epoch": 1.003011270756259,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009042699519305899,
      "loss": 1.3341,
      "step": 11658
    },
    {
      "epoch": 1.003097307063581,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0009042540394524395,
      "loss": 0.67,
      "step": 11659
    },
    {
      "epoch": 1.0031833433709025,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009042381257919246,
      "loss": 0.677,
      "step": 11660
    },
    {
      "epoch": 1.0032693796782242,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009042222109490918,
      "loss": 1.3658,
      "step": 11661
    },
    {
      "epoch": 1.003355415985546,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009042062949239877,
      "loss": 1.2545,
      "step": 11662
    },
    {
      "epoch": 1.0034414522928676,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009041903777166589,
      "loss": 0.7259,
      "step": 11663
    },
    {
      "epoch": 1.0035274886001893,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009041744593271517,
      "loss": 1.443,
      "step": 11664
    },
    {
      "epoch": 1.0036135249075109,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009041585397555128,
      "loss": 1.0657,
      "step": 11665
    },
    {
      "epoch": 1.0036995612148327,
      "grad_norm": 2.0,
      "learning_rate": 0.0009041426190017889,
      "loss": 1.7003,
      "step": 11666
    },
    {
      "epoch": 1.0037855975221543,
      "grad_norm": 1.125,
      "learning_rate": 0.0009041266970660263,
      "loss": 1.4729,
      "step": 11667
    },
    {
      "epoch": 1.003871633829476,
      "grad_norm": 0.765625,
      "learning_rate": 0.0009041107739482718,
      "loss": 0.769,
      "step": 11668
    },
    {
      "epoch": 1.0039576701367978,
      "grad_norm": 1.5,
      "learning_rate": 0.0009040948496485719,
      "loss": 0.8152,
      "step": 11669
    },
    {
      "epoch": 1.0040437064441194,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009040789241669731,
      "loss": 0.9839,
      "step": 11670
    },
    {
      "epoch": 1.004129742751441,
      "grad_norm": 1.375,
      "learning_rate": 0.0009040629975035219,
      "loss": 1.6616,
      "step": 11671
    },
    {
      "epoch": 1.0042157790587627,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009040470696582652,
      "loss": 1.2171,
      "step": 11672
    },
    {
      "epoch": 1.0043018153660845,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009040311406312495,
      "loss": 1.2677,
      "step": 11673
    },
    {
      "epoch": 1.0043878516734062,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009040152104225211,
      "loss": 1.1157,
      "step": 11674
    },
    {
      "epoch": 1.0044738879807278,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009039992790321269,
      "loss": 2.2018,
      "step": 11675
    },
    {
      "epoch": 1.0045599242880496,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009039833464601132,
      "loss": 1.0445,
      "step": 11676
    },
    {
      "epoch": 1.0046459605953713,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000903967412706527,
      "loss": 1.8945,
      "step": 11677
    },
    {
      "epoch": 1.0047319969026929,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009039514777714146,
      "loss": 0.5756,
      "step": 11678
    },
    {
      "epoch": 1.0048180332100147,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009039355416548227,
      "loss": 1.5448,
      "step": 11679
    },
    {
      "epoch": 1.0049040695173364,
      "grad_norm": 1.25,
      "learning_rate": 0.0009039196043567979,
      "loss": 1.0636,
      "step": 11680
    },
    {
      "epoch": 1.004990105824658,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009039036658773868,
      "loss": 1.1009,
      "step": 11681
    },
    {
      "epoch": 1.0050761421319796,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0009038877262166361,
      "loss": 1.7896,
      "step": 11682
    },
    {
      "epoch": 1.0051621784393014,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009038717853745923,
      "loss": 1.3742,
      "step": 11683
    },
    {
      "epoch": 1.005248214746623,
      "grad_norm": 1.71875,
      "learning_rate": 0.000903855843351302,
      "loss": 1.2914,
      "step": 11684
    },
    {
      "epoch": 1.0053342510539447,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000903839900146812,
      "loss": 0.8462,
      "step": 11685
    },
    {
      "epoch": 1.0054202873612665,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000903823955761169,
      "loss": 0.8517,
      "step": 11686
    },
    {
      "epoch": 1.0055063236685882,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009038080101944191,
      "loss": 0.7018,
      "step": 11687
    },
    {
      "epoch": 1.0055923599759098,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009037920634466095,
      "loss": 1.5471,
      "step": 11688
    },
    {
      "epoch": 1.0056783962832316,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009037761155177866,
      "loss": 1.5258,
      "step": 11689
    },
    {
      "epoch": 1.0057644325905533,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009037601664079972,
      "loss": 1.1158,
      "step": 11690
    },
    {
      "epoch": 1.0058504688978749,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0009037442161172877,
      "loss": 1.3336,
      "step": 11691
    },
    {
      "epoch": 1.0059365052051965,
      "grad_norm": 1.140625,
      "learning_rate": 0.000903728264645705,
      "loss": 1.0139,
      "step": 11692
    },
    {
      "epoch": 1.0060225415125184,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009037123119932957,
      "loss": 0.8185,
      "step": 11693
    },
    {
      "epoch": 1.00610857781984,
      "grad_norm": 4.625,
      "learning_rate": 0.0009036963581601062,
      "loss": 1.0031,
      "step": 11694
    },
    {
      "epoch": 1.0061946141271616,
      "grad_norm": 1.109375,
      "learning_rate": 0.0009036804031461835,
      "loss": 1.7688,
      "step": 11695
    },
    {
      "epoch": 1.0062806504344834,
      "grad_norm": 1.625,
      "learning_rate": 0.000903664446951574,
      "loss": 1.2512,
      "step": 11696
    },
    {
      "epoch": 1.006366686741805,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009036484895763247,
      "loss": 1.0357,
      "step": 11697
    },
    {
      "epoch": 1.0064527230491267,
      "grad_norm": 1.125,
      "learning_rate": 0.000903632531020482,
      "loss": 1.3125,
      "step": 11698
    },
    {
      "epoch": 1.0065387593564483,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009036165712840927,
      "loss": 1.1508,
      "step": 11699
    },
    {
      "epoch": 1.0066247956637702,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009036006103672032,
      "loss": 0.9442,
      "step": 11700
    },
    {
      "epoch": 1.0067108319710918,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009035846482698607,
      "loss": 1.036,
      "step": 11701
    },
    {
      "epoch": 1.0067968682784134,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009035686849921114,
      "loss": 1.3609,
      "step": 11702
    },
    {
      "epoch": 1.0068829045857353,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009035527205340022,
      "loss": 1.8296,
      "step": 11703
    },
    {
      "epoch": 1.0069689408930569,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00090353675489558,
      "loss": 1.2846,
      "step": 11704
    },
    {
      "epoch": 1.0070549772003785,
      "grad_norm": 1.421875,
      "learning_rate": 0.000903520788076891,
      "loss": 1.5229,
      "step": 11705
    },
    {
      "epoch": 1.0071410135077004,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009035048200779823,
      "loss": 1.5027,
      "step": 11706
    },
    {
      "epoch": 1.007227049815022,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009034888508989004,
      "loss": 1.0674,
      "step": 11707
    },
    {
      "epoch": 1.0073130861223436,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009034728805396922,
      "loss": 1.0475,
      "step": 11708
    },
    {
      "epoch": 1.0073991224296652,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009034569090004042,
      "loss": 1.8329,
      "step": 11709
    },
    {
      "epoch": 1.007485158736987,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009034409362810834,
      "loss": 1.4314,
      "step": 11710
    },
    {
      "epoch": 1.0075711950443087,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000903424962381776,
      "loss": 0.9807,
      "step": 11711
    },
    {
      "epoch": 1.0076572313516303,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009034089873025293,
      "loss": 2.1335,
      "step": 11712
    },
    {
      "epoch": 1.0077432676589522,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009033930110433897,
      "loss": 1.0728,
      "step": 11713
    },
    {
      "epoch": 1.0078293039662738,
      "grad_norm": 1.1875,
      "learning_rate": 0.000903377033604404,
      "loss": 1.4681,
      "step": 11714
    },
    {
      "epoch": 1.0079153402735954,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009033610549856189,
      "loss": 1.0811,
      "step": 11715
    },
    {
      "epoch": 1.008001376580917,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009033450751870812,
      "loss": 0.68,
      "step": 11716
    },
    {
      "epoch": 1.0080874128882389,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009033290942088376,
      "loss": 0.9337,
      "step": 11717
    },
    {
      "epoch": 1.0081734491955605,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009033131120509348,
      "loss": 1.0369,
      "step": 11718
    },
    {
      "epoch": 1.0082594855028821,
      "grad_norm": 1.484375,
      "learning_rate": 0.0009032971287134197,
      "loss": 2.0133,
      "step": 11719
    },
    {
      "epoch": 1.008345521810204,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009032811441963388,
      "loss": 0.549,
      "step": 11720
    },
    {
      "epoch": 1.0084315581175256,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009032651584997392,
      "loss": 1.4546,
      "step": 11721
    },
    {
      "epoch": 1.0085175944248472,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009032491716236671,
      "loss": 1.8492,
      "step": 11722
    },
    {
      "epoch": 1.008603630732169,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009032331835681697,
      "loss": 2.9025,
      "step": 11723
    },
    {
      "epoch": 1.0086896670394907,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0009032171943332938,
      "loss": 1.1899,
      "step": 11724
    },
    {
      "epoch": 1.0087757033468123,
      "grad_norm": 1.453125,
      "learning_rate": 0.000903201203919086,
      "loss": 1.5778,
      "step": 11725
    },
    {
      "epoch": 1.008861739654134,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000903185212325593,
      "loss": 1.8876,
      "step": 11726
    },
    {
      "epoch": 1.0089477759614558,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009031692195528618,
      "loss": 1.3155,
      "step": 11727
    },
    {
      "epoch": 1.0090338122687774,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009031532256009389,
      "loss": 1.124,
      "step": 11728
    },
    {
      "epoch": 1.009119848576099,
      "grad_norm": 1.078125,
      "learning_rate": 0.0009031372304698715,
      "loss": 0.9822,
      "step": 11729
    },
    {
      "epoch": 1.009205884883421,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009031212341597058,
      "loss": 1.93,
      "step": 11730
    },
    {
      "epoch": 1.0092919211907425,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009031052366704891,
      "loss": 1.2203,
      "step": 11731
    },
    {
      "epoch": 1.0093779574980641,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009030892380022678,
      "loss": 1.3984,
      "step": 11732
    },
    {
      "epoch": 1.0094639938053858,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000903073238155089,
      "loss": 0.8575,
      "step": 11733
    },
    {
      "epoch": 1.0095500301127076,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009030572371289993,
      "loss": 1.9086,
      "step": 11734
    },
    {
      "epoch": 1.0096360664200292,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009030412349240455,
      "loss": 0.8967,
      "step": 11735
    },
    {
      "epoch": 1.0097221027273509,
      "grad_norm": 0.953125,
      "learning_rate": 0.0009030252315402747,
      "loss": 1.0934,
      "step": 11736
    },
    {
      "epoch": 1.0098081390346727,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009030092269777335,
      "loss": 0.9819,
      "step": 11737
    },
    {
      "epoch": 1.0098941753419943,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009029932212364684,
      "loss": 1.0688,
      "step": 11738
    },
    {
      "epoch": 1.009980211649316,
      "grad_norm": 1.125,
      "learning_rate": 0.0009029772143165267,
      "loss": 1.0362,
      "step": 11739
    },
    {
      "epoch": 1.0100662479566378,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009029612062179551,
      "loss": 1.2646,
      "step": 11740
    },
    {
      "epoch": 1.0101522842639594,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009029451969408001,
      "loss": 1.7693,
      "step": 11741
    },
    {
      "epoch": 1.010238320571281,
      "grad_norm": 1.4375,
      "learning_rate": 0.000902929186485109,
      "loss": 2.093,
      "step": 11742
    },
    {
      "epoch": 1.0103243568786027,
      "grad_norm": 1.0,
      "learning_rate": 0.0009029131748509283,
      "loss": 1.0557,
      "step": 11743
    },
    {
      "epoch": 1.0104103931859245,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009028971620383048,
      "loss": 1.0139,
      "step": 11744
    },
    {
      "epoch": 1.0104964294932461,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009028811480472856,
      "loss": 0.6669,
      "step": 11745
    },
    {
      "epoch": 1.0105824658005678,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009028651328779174,
      "loss": 0.9487,
      "step": 11746
    },
    {
      "epoch": 1.0106685021078896,
      "grad_norm": 1.265625,
      "learning_rate": 0.000902849116530247,
      "loss": 2.1229,
      "step": 11747
    },
    {
      "epoch": 1.0107545384152112,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009028330990043214,
      "loss": 1.3668,
      "step": 11748
    },
    {
      "epoch": 1.0108405747225329,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009028170803001872,
      "loss": 1.9231,
      "step": 11749
    },
    {
      "epoch": 1.0109266110298547,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009028010604178913,
      "loss": 1.6249,
      "step": 11750
    },
    {
      "epoch": 1.0110126473371763,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0009027850393574808,
      "loss": 2.4539,
      "step": 11751
    },
    {
      "epoch": 1.011098683644498,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009027690171190024,
      "loss": 1.4234,
      "step": 11752
    },
    {
      "epoch": 1.0111847199518196,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009027529937025029,
      "loss": 1.5964,
      "step": 11753
    },
    {
      "epoch": 1.0112707562591414,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009027369691080292,
      "loss": 1.2947,
      "step": 11754
    },
    {
      "epoch": 1.011356792566463,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009027209433356282,
      "loss": 0.7305,
      "step": 11755
    },
    {
      "epoch": 1.0114428288737847,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009027049163853469,
      "loss": 0.7972,
      "step": 11756
    },
    {
      "epoch": 1.0115288651811065,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000902688888257232,
      "loss": 1.6722,
      "step": 11757
    },
    {
      "epoch": 1.0116149014884281,
      "grad_norm": 0.984375,
      "learning_rate": 0.0009026728589513302,
      "loss": 1.3706,
      "step": 11758
    },
    {
      "epoch": 1.0117009377957498,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0009026568284676888,
      "loss": 1.8475,
      "step": 11759
    },
    {
      "epoch": 1.0117869741030714,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0009026407968063544,
      "loss": 0.7354,
      "step": 11760
    },
    {
      "epoch": 1.0118730104103932,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009026247639673742,
      "loss": 1.8797,
      "step": 11761
    },
    {
      "epoch": 1.0119590467177149,
      "grad_norm": 1.75,
      "learning_rate": 0.0009026087299507945,
      "loss": 1.2659,
      "step": 11762
    },
    {
      "epoch": 1.0120450830250365,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009025926947566628,
      "loss": 1.9759,
      "step": 11763
    },
    {
      "epoch": 1.0121311193323583,
      "grad_norm": 1.125,
      "learning_rate": 0.0009025766583850257,
      "loss": 1.3557,
      "step": 11764
    },
    {
      "epoch": 1.01221715563968,
      "grad_norm": 1.59375,
      "learning_rate": 0.0009025606208359303,
      "loss": 1.109,
      "step": 11765
    },
    {
      "epoch": 1.0123031919470016,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009025445821094234,
      "loss": 1.957,
      "step": 11766
    },
    {
      "epoch": 1.0123892282543234,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009025285422055516,
      "loss": 1.415,
      "step": 11767
    },
    {
      "epoch": 1.012475264561645,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009025125011243623,
      "loss": 1.3988,
      "step": 11768
    },
    {
      "epoch": 1.0125613008689667,
      "grad_norm": 1.5,
      "learning_rate": 0.0009024964588659021,
      "loss": 1.2291,
      "step": 11769
    },
    {
      "epoch": 1.0126473371762883,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009024804154302182,
      "loss": 1.8905,
      "step": 11770
    },
    {
      "epoch": 1.0127333734836101,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0009024643708173574,
      "loss": 1.1921,
      "step": 11771
    },
    {
      "epoch": 1.0128194097909318,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0009024483250273664,
      "loss": 1.0918,
      "step": 11772
    },
    {
      "epoch": 1.0129054460982534,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009024322780602925,
      "loss": 1.2008,
      "step": 11773
    },
    {
      "epoch": 1.0129914824055752,
      "grad_norm": 1.125,
      "learning_rate": 0.0009024162299161822,
      "loss": 1.6965,
      "step": 11774
    },
    {
      "epoch": 1.0130775187128969,
      "grad_norm": 1.234375,
      "learning_rate": 0.000902400180595083,
      "loss": 1.6412,
      "step": 11775
    },
    {
      "epoch": 1.0131635550202185,
      "grad_norm": 1.125,
      "learning_rate": 0.0009023841300970413,
      "loss": 1.608,
      "step": 11776
    },
    {
      "epoch": 1.0132495913275403,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009023680784221045,
      "loss": 1.4295,
      "step": 11777
    },
    {
      "epoch": 1.013335627634862,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000902352025570319,
      "loss": 1.2963,
      "step": 11778
    },
    {
      "epoch": 1.0134216639421836,
      "grad_norm": 1.0,
      "learning_rate": 0.0009023359715417322,
      "loss": 0.8858,
      "step": 11779
    },
    {
      "epoch": 1.0135077002495052,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000902319916336391,
      "loss": 1.046,
      "step": 11780
    },
    {
      "epoch": 1.013593736556827,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009023038599543425,
      "loss": 1.2414,
      "step": 11781
    },
    {
      "epoch": 1.0136797728641487,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0009022878023956331,
      "loss": 1.0352,
      "step": 11782
    },
    {
      "epoch": 1.0137658091714703,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009022717436603104,
      "loss": 1.7017,
      "step": 11783
    },
    {
      "epoch": 1.0138518454787921,
      "grad_norm": 1.515625,
      "learning_rate": 0.000902255683748421,
      "loss": 2.0299,
      "step": 11784
    },
    {
      "epoch": 1.0139378817861138,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009022396226600118,
      "loss": 1.1392,
      "step": 11785
    },
    {
      "epoch": 1.0140239180934354,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009022235603951301,
      "loss": 1.0486,
      "step": 11786
    },
    {
      "epoch": 1.014109954400757,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009022074969538226,
      "loss": 1.4013,
      "step": 11787
    },
    {
      "epoch": 1.0141959907080789,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009021914323361365,
      "loss": 1.1296,
      "step": 11788
    },
    {
      "epoch": 1.0142820270154005,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009021753665421187,
      "loss": 1.1658,
      "step": 11789
    },
    {
      "epoch": 1.0143680633227221,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009021592995718161,
      "loss": 1.2151,
      "step": 11790
    },
    {
      "epoch": 1.014454099630044,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0009021432314252758,
      "loss": 1.3957,
      "step": 11791
    },
    {
      "epoch": 1.0145401359373656,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009021271621025447,
      "loss": 1.8189,
      "step": 11792
    },
    {
      "epoch": 1.0146261722446872,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0009021110916036699,
      "loss": 0.7654,
      "step": 11793
    },
    {
      "epoch": 1.014712208552009,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009020950199286983,
      "loss": 1.3225,
      "step": 11794
    },
    {
      "epoch": 1.0147982448593307,
      "grad_norm": 2.015625,
      "learning_rate": 0.0009020789470776771,
      "loss": 1.1242,
      "step": 11795
    },
    {
      "epoch": 1.0148842811666523,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009020628730506532,
      "loss": 1.5079,
      "step": 11796
    },
    {
      "epoch": 1.014970317473974,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009020467978476734,
      "loss": 1.6273,
      "step": 11797
    },
    {
      "epoch": 1.0150563537812958,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0009020307214687851,
      "loss": 1.6932,
      "step": 11798
    },
    {
      "epoch": 1.0151423900886174,
      "grad_norm": 1.328125,
      "learning_rate": 0.000902014643914035,
      "loss": 1.273,
      "step": 11799
    },
    {
      "epoch": 1.015228426395939,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009019985651834703,
      "loss": 1.2372,
      "step": 11800
    },
    {
      "epoch": 1.0153144627032609,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009019824852771379,
      "loss": 1.2144,
      "step": 11801
    },
    {
      "epoch": 1.0154004990105825,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000901966404195085,
      "loss": 1.6148,
      "step": 11802
    },
    {
      "epoch": 1.0154865353179041,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009019503219373586,
      "loss": 1.1905,
      "step": 11803
    },
    {
      "epoch": 1.0155725716252257,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0009019342385040057,
      "loss": 1.2727,
      "step": 11804
    },
    {
      "epoch": 1.0156586079325476,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009019181538950732,
      "loss": 1.3231,
      "step": 11805
    },
    {
      "epoch": 1.0157446442398692,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009019020681106083,
      "loss": 2.1046,
      "step": 11806
    },
    {
      "epoch": 1.0158306805471908,
      "grad_norm": 1.625,
      "learning_rate": 0.0009018859811506582,
      "loss": 1.5143,
      "step": 11807
    },
    {
      "epoch": 1.0159167168545127,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0009018698930152695,
      "loss": 1.7097,
      "step": 11808
    },
    {
      "epoch": 1.0160027531618343,
      "grad_norm": 1.75,
      "learning_rate": 0.0009018538037044896,
      "loss": 1.4202,
      "step": 11809
    },
    {
      "epoch": 1.016088789469156,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009018377132183655,
      "loss": 1.3511,
      "step": 11810
    },
    {
      "epoch": 1.0161748257764778,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009018216215569442,
      "loss": 1.9781,
      "step": 11811
    },
    {
      "epoch": 1.0162608620837994,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009018055287202728,
      "loss": 2.1266,
      "step": 11812
    },
    {
      "epoch": 1.016346898391121,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009017894347083986,
      "loss": 1.3095,
      "step": 11813
    },
    {
      "epoch": 1.0164329346984426,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009017733395213681,
      "loss": 1.6869,
      "step": 11814
    },
    {
      "epoch": 1.0165189710057645,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000901757243159229,
      "loss": 1.1731,
      "step": 11815
    },
    {
      "epoch": 1.0166050073130861,
      "grad_norm": 1.109375,
      "learning_rate": 0.000901741145622028,
      "loss": 1.2051,
      "step": 11816
    },
    {
      "epoch": 1.0166910436204077,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0009017250469098121,
      "loss": 1.7745,
      "step": 11817
    },
    {
      "epoch": 1.0167770799277296,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009017089470226289,
      "loss": 1.4242,
      "step": 11818
    },
    {
      "epoch": 1.0168631162350512,
      "grad_norm": 1.5,
      "learning_rate": 0.0009016928459605249,
      "loss": 0.9544,
      "step": 11819
    },
    {
      "epoch": 1.0169491525423728,
      "grad_norm": 1.546875,
      "learning_rate": 0.0009016767437235476,
      "loss": 1.659,
      "step": 11820
    },
    {
      "epoch": 1.0170351888496945,
      "grad_norm": 1.1875,
      "learning_rate": 0.000901660640311744,
      "loss": 0.7056,
      "step": 11821
    },
    {
      "epoch": 1.0171212251570163,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000901644535725161,
      "loss": 1.3436,
      "step": 11822
    },
    {
      "epoch": 1.017207261464338,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000901628429963846,
      "loss": 0.9035,
      "step": 11823
    },
    {
      "epoch": 1.0172932977716596,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009016123230278458,
      "loss": 1.6128,
      "step": 11824
    },
    {
      "epoch": 1.0173793340789814,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009015962149172077,
      "loss": 1.0099,
      "step": 11825
    },
    {
      "epoch": 1.017465370386303,
      "grad_norm": 1.125,
      "learning_rate": 0.0009015801056319788,
      "loss": 1.3418,
      "step": 11826
    },
    {
      "epoch": 1.0175514066936246,
      "grad_norm": 1.015625,
      "learning_rate": 0.0009015639951722061,
      "loss": 1.3931,
      "step": 11827
    },
    {
      "epoch": 1.0176374430009465,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000901547883537937,
      "loss": 1.7167,
      "step": 11828
    },
    {
      "epoch": 1.0177234793082681,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009015317707292183,
      "loss": 0.9704,
      "step": 11829
    },
    {
      "epoch": 1.0178095156155897,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009015156567460974,
      "loss": 1.2388,
      "step": 11830
    },
    {
      "epoch": 1.0178955519229114,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009014995415886211,
      "loss": 1.1369,
      "step": 11831
    },
    {
      "epoch": 1.0179815882302332,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009014834252568369,
      "loss": 0.7861,
      "step": 11832
    },
    {
      "epoch": 1.0180676245375548,
      "grad_norm": 1.0,
      "learning_rate": 0.0009014673077507916,
      "loss": 0.8764,
      "step": 11833
    },
    {
      "epoch": 1.0181536608448765,
      "grad_norm": 1.25,
      "learning_rate": 0.0009014511890705326,
      "loss": 1.3253,
      "step": 11834
    },
    {
      "epoch": 1.0182396971521983,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000901435069216107,
      "loss": 1.6962,
      "step": 11835
    },
    {
      "epoch": 1.01832573345952,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0009014189481875619,
      "loss": 1.4477,
      "step": 11836
    },
    {
      "epoch": 1.0184117697668416,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009014028259849443,
      "loss": 1.6029,
      "step": 11837
    },
    {
      "epoch": 1.0184978060741634,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009013867026083016,
      "loss": 1.1721,
      "step": 11838
    },
    {
      "epoch": 1.018583842381485,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009013705780576809,
      "loss": 1.062,
      "step": 11839
    },
    {
      "epoch": 1.0186698786888067,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009013544523331293,
      "loss": 1.4982,
      "step": 11840
    },
    {
      "epoch": 1.0187559149961283,
      "grad_norm": 1.359375,
      "learning_rate": 0.000901338325434694,
      "loss": 1.1464,
      "step": 11841
    },
    {
      "epoch": 1.0188419513034501,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009013221973624221,
      "loss": 1.7932,
      "step": 11842
    },
    {
      "epoch": 1.0189279876107717,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009013060681163607,
      "loss": 1.7316,
      "step": 11843
    },
    {
      "epoch": 1.0190140239180934,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0009012899376965573,
      "loss": 1.8568,
      "step": 11844
    },
    {
      "epoch": 1.0191000602254152,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009012738061030588,
      "loss": 1.1892,
      "step": 11845
    },
    {
      "epoch": 1.0191860965327368,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009012576733359126,
      "loss": 1.429,
      "step": 11846
    },
    {
      "epoch": 1.0192721328400585,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009012415393951656,
      "loss": 1.0097,
      "step": 11847
    },
    {
      "epoch": 1.01935816914738,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009012254042808652,
      "loss": 0.615,
      "step": 11848
    },
    {
      "epoch": 1.019444205454702,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0009012092679930583,
      "loss": 1.6762,
      "step": 11849
    },
    {
      "epoch": 1.0195302417620236,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009011931305317926,
      "loss": 1.0919,
      "step": 11850
    },
    {
      "epoch": 1.0196162780693452,
      "grad_norm": 1.09375,
      "learning_rate": 0.000901176991897115,
      "loss": 0.8606,
      "step": 11851
    },
    {
      "epoch": 1.019702314376667,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009011608520890726,
      "loss": 0.6646,
      "step": 11852
    },
    {
      "epoch": 1.0197883506839887,
      "grad_norm": 1.5625,
      "learning_rate": 0.0009011447111077129,
      "loss": 0.8288,
      "step": 11853
    },
    {
      "epoch": 1.0198743869913103,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009011285689530828,
      "loss": 0.8098,
      "step": 11854
    },
    {
      "epoch": 1.0199604232986321,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009011124256252296,
      "loss": 1.0504,
      "step": 11855
    },
    {
      "epoch": 1.0200464596059537,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009010962811242007,
      "loss": 1.5026,
      "step": 11856
    },
    {
      "epoch": 1.0201324959132754,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009010801354500431,
      "loss": 1.1325,
      "step": 11857
    },
    {
      "epoch": 1.020218532220597,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009010639886028042,
      "loss": 0.7464,
      "step": 11858
    },
    {
      "epoch": 1.0203045685279188,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009010478405825311,
      "loss": 0.8915,
      "step": 11859
    },
    {
      "epoch": 1.0203906048352405,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000901031691389271,
      "loss": 0.9289,
      "step": 11860
    },
    {
      "epoch": 1.020476641142562,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0009010155410230712,
      "loss": 1.6479,
      "step": 11861
    },
    {
      "epoch": 1.020562677449884,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000900999389483979,
      "loss": 1.6574,
      "step": 11862
    },
    {
      "epoch": 1.0206487137572056,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009009832367720416,
      "loss": 1.103,
      "step": 11863
    },
    {
      "epoch": 1.0207347500645272,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009009670828873061,
      "loss": 1.1472,
      "step": 11864
    },
    {
      "epoch": 1.020820786371849,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00090095092782982,
      "loss": 1.0864,
      "step": 11865
    },
    {
      "epoch": 1.0209068226791707,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009009347715996304,
      "loss": 1.4179,
      "step": 11866
    },
    {
      "epoch": 1.0209928589864923,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0009009186141967845,
      "loss": 0.6785,
      "step": 11867
    },
    {
      "epoch": 1.021078895293814,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009009024556213297,
      "loss": 0.8685,
      "step": 11868
    },
    {
      "epoch": 1.0211649316011358,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009008862958733131,
      "loss": 0.8649,
      "step": 11869
    },
    {
      "epoch": 1.0212509679084574,
      "grad_norm": 1.203125,
      "learning_rate": 0.000900870134952782,
      "loss": 0.8794,
      "step": 11870
    },
    {
      "epoch": 1.021337004215779,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009008539728597839,
      "loss": 1.2425,
      "step": 11871
    },
    {
      "epoch": 1.0214230405231008,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0009008378095943658,
      "loss": 1.5964,
      "step": 11872
    },
    {
      "epoch": 1.0215090768304225,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000900821645156575,
      "loss": 1.6743,
      "step": 11873
    },
    {
      "epoch": 1.021595113137744,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009008054795464588,
      "loss": 1.1235,
      "step": 11874
    },
    {
      "epoch": 1.0216811494450657,
      "grad_norm": 1.25,
      "learning_rate": 0.0009007893127640647,
      "loss": 1.0901,
      "step": 11875
    },
    {
      "epoch": 1.0217671857523876,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009007731448094398,
      "loss": 1.5253,
      "step": 11876
    },
    {
      "epoch": 1.0218532220597092,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009007569756826314,
      "loss": 1.3308,
      "step": 11877
    },
    {
      "epoch": 1.0219392583670308,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009007408053836867,
      "loss": 1.0887,
      "step": 11878
    },
    {
      "epoch": 1.0220252946743527,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0009007246339126532,
      "loss": 0.9751,
      "step": 11879
    },
    {
      "epoch": 1.0221113309816743,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009007084612695779,
      "loss": 1.156,
      "step": 11880
    },
    {
      "epoch": 1.022197367288996,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009006922874545085,
      "loss": 1.6608,
      "step": 11881
    },
    {
      "epoch": 1.0222834035963178,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009006761124674919,
      "loss": 1.2775,
      "step": 11882
    },
    {
      "epoch": 1.0223694399036394,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0009006599363085756,
      "loss": 1.0131,
      "step": 11883
    },
    {
      "epoch": 1.022455476210961,
      "grad_norm": 0.859375,
      "learning_rate": 0.000900643758977807,
      "loss": 0.7891,
      "step": 11884
    },
    {
      "epoch": 1.0225415125182826,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0009006275804752333,
      "loss": 1.1615,
      "step": 11885
    },
    {
      "epoch": 1.0226275488256045,
      "grad_norm": 1.125,
      "learning_rate": 0.0009006114008009019,
      "loss": 1.1166,
      "step": 11886
    },
    {
      "epoch": 1.022713585132926,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00090059521995486,
      "loss": 1.0365,
      "step": 11887
    },
    {
      "epoch": 1.0227996214402477,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000900579037937155,
      "loss": 1.2412,
      "step": 11888
    },
    {
      "epoch": 1.0228856577475696,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009005628547478343,
      "loss": 1.3411,
      "step": 11889
    },
    {
      "epoch": 1.0229716940548912,
      "grad_norm": 1.125,
      "learning_rate": 0.0009005466703869452,
      "loss": 0.9796,
      "step": 11890
    },
    {
      "epoch": 1.0230577303622128,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0009005304848545349,
      "loss": 0.4808,
      "step": 11891
    },
    {
      "epoch": 1.0231437666695344,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009005142981506507,
      "loss": 1.3216,
      "step": 11892
    },
    {
      "epoch": 1.0232298029768563,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009004981102753402,
      "loss": 1.4372,
      "step": 11893
    },
    {
      "epoch": 1.023315839284178,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009004819212286506,
      "loss": 0.8725,
      "step": 11894
    },
    {
      "epoch": 1.0234018755914995,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009004657310106292,
      "loss": 1.157,
      "step": 11895
    },
    {
      "epoch": 1.0234879118988214,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0009004495396213234,
      "loss": 1.3416,
      "step": 11896
    },
    {
      "epoch": 1.023573948206143,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009004333470607808,
      "loss": 1.1746,
      "step": 11897
    },
    {
      "epoch": 1.0236599845134646,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009004171533290484,
      "loss": 1.1956,
      "step": 11898
    },
    {
      "epoch": 1.0237460208207865,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009004009584261739,
      "loss": 2.0706,
      "step": 11899
    },
    {
      "epoch": 1.023832057128108,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009003847623522041,
      "loss": 1.4049,
      "step": 11900
    },
    {
      "epoch": 1.0239180934354297,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000900368565107187,
      "loss": 1.3203,
      "step": 11901
    },
    {
      "epoch": 1.0240041297427513,
      "grad_norm": 2.171875,
      "learning_rate": 0.0009003523666911697,
      "loss": 1.164,
      "step": 11902
    },
    {
      "epoch": 1.0240901660500732,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009003361671041995,
      "loss": 2.0322,
      "step": 11903
    },
    {
      "epoch": 1.0241762023573948,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009003199663463239,
      "loss": 1.3938,
      "step": 11904
    },
    {
      "epoch": 1.0242622386647164,
      "grad_norm": 1.09375,
      "learning_rate": 0.0009003037644175902,
      "loss": 1.3462,
      "step": 11905
    },
    {
      "epoch": 1.0243482749720383,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000900287561318046,
      "loss": 0.751,
      "step": 11906
    },
    {
      "epoch": 1.02443431127936,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009002713570477384,
      "loss": 1.2749,
      "step": 11907
    },
    {
      "epoch": 1.0245203475866815,
      "grad_norm": 1.265625,
      "learning_rate": 0.000900255151606715,
      "loss": 0.9907,
      "step": 11908
    },
    {
      "epoch": 1.0246063838940032,
      "grad_norm": 1.140625,
      "learning_rate": 0.0009002389449950231,
      "loss": 1.3095,
      "step": 11909
    },
    {
      "epoch": 1.024692420201325,
      "grad_norm": 1.140625,
      "learning_rate": 0.00090022273721271,
      "loss": 0.8196,
      "step": 11910
    },
    {
      "epoch": 1.0247784565086466,
      "grad_norm": 1.796875,
      "learning_rate": 0.0009002065282598235,
      "loss": 1.0037,
      "step": 11911
    },
    {
      "epoch": 1.0248644928159683,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009001903181364106,
      "loss": 1.3823,
      "step": 11912
    },
    {
      "epoch": 1.02495052912329,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0009001741068425189,
      "loss": 0.8997,
      "step": 11913
    },
    {
      "epoch": 1.0250365654306117,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009001578943781957,
      "loss": 1.7048,
      "step": 11914
    },
    {
      "epoch": 1.0251226017379333,
      "grad_norm": 1.046875,
      "learning_rate": 0.0009001416807434885,
      "loss": 0.8598,
      "step": 11915
    },
    {
      "epoch": 1.0252086380452552,
      "grad_norm": 1.0625,
      "learning_rate": 0.0009001254659384448,
      "loss": 1.1588,
      "step": 11916
    },
    {
      "epoch": 1.0252946743525768,
      "grad_norm": 225.0,
      "learning_rate": 0.0009001092499631118,
      "loss": 1.5663,
      "step": 11917
    },
    {
      "epoch": 1.0253807106598984,
      "grad_norm": 5.34375,
      "learning_rate": 0.000900093032817537,
      "loss": 1.2327,
      "step": 11918
    },
    {
      "epoch": 1.02546674696722,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009000768145017682,
      "loss": 1.2912,
      "step": 11919
    },
    {
      "epoch": 1.025552783274542,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009000605950158523,
      "loss": 1.1373,
      "step": 11920
    },
    {
      "epoch": 1.0256388195818635,
      "grad_norm": 11.25,
      "learning_rate": 0.000900044374359837,
      "loss": 1.4313,
      "step": 11921
    },
    {
      "epoch": 1.0257248558891852,
      "grad_norm": 2.921875,
      "learning_rate": 0.0009000281525337697,
      "loss": 0.6573,
      "step": 11922
    },
    {
      "epoch": 1.025810892196507,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009000119295376978,
      "loss": 1.1611,
      "step": 11923
    },
    {
      "epoch": 1.0258969285038286,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008999957053716688,
      "loss": 2.005,
      "step": 11924
    },
    {
      "epoch": 1.0259829648111503,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008999794800357303,
      "loss": 0.9888,
      "step": 11925
    },
    {
      "epoch": 1.026069001118472,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008999632535299296,
      "loss": 1.246,
      "step": 11926
    },
    {
      "epoch": 1.0261550374257937,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008999470258543141,
      "loss": 1.7873,
      "step": 11927
    },
    {
      "epoch": 1.0262410737331153,
      "grad_norm": 11.375,
      "learning_rate": 0.0008999307970089312,
      "loss": 1.8309,
      "step": 11928
    },
    {
      "epoch": 1.026327110040437,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008999145669938287,
      "loss": 1.3797,
      "step": 11929
    },
    {
      "epoch": 1.0264131463477588,
      "grad_norm": 9.9375,
      "learning_rate": 0.0008998983358090537,
      "loss": 1.5387,
      "step": 11930
    },
    {
      "epoch": 1.0264991826550804,
      "grad_norm": 31.25,
      "learning_rate": 0.0008998821034546541,
      "loss": 0.9721,
      "step": 11931
    },
    {
      "epoch": 1.026585218962402,
      "grad_norm": 137.0,
      "learning_rate": 0.000899865869930677,
      "loss": 1.0705,
      "step": 11932
    },
    {
      "epoch": 1.026671255269724,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008998496352371699,
      "loss": 0.7511,
      "step": 11933
    },
    {
      "epoch": 1.0267572915770455,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008998333993741805,
      "loss": 1.2519,
      "step": 11934
    },
    {
      "epoch": 1.0268433278843672,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008998171623417562,
      "loss": 0.9676,
      "step": 11935
    },
    {
      "epoch": 1.0269293641916888,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008998009241399443,
      "loss": 1.0356,
      "step": 11936
    },
    {
      "epoch": 1.0270154004990106,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008997846847687925,
      "loss": 1.2544,
      "step": 11937
    },
    {
      "epoch": 1.0271014368063323,
      "grad_norm": 7.3125,
      "learning_rate": 0.0008997684442283483,
      "loss": 1.3509,
      "step": 11938
    },
    {
      "epoch": 1.0271874731136539,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008997522025186592,
      "loss": 1.9056,
      "step": 11939
    },
    {
      "epoch": 1.0272735094209757,
      "grad_norm": 6.875,
      "learning_rate": 0.0008997359596397726,
      "loss": 1.9036,
      "step": 11940
    },
    {
      "epoch": 1.0273595457282974,
      "grad_norm": 3.4375,
      "learning_rate": 0.000899719715591736,
      "loss": 1.7702,
      "step": 11941
    },
    {
      "epoch": 1.027445582035619,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008997034703745972,
      "loss": 0.8732,
      "step": 11942
    },
    {
      "epoch": 1.0275316183429408,
      "grad_norm": 2.46875,
      "learning_rate": 0.0008996872239884033,
      "loss": 0.9322,
      "step": 11943
    },
    {
      "epoch": 1.0276176546502624,
      "grad_norm": 2.09375,
      "learning_rate": 0.000899670976433202,
      "loss": 1.0154,
      "step": 11944
    },
    {
      "epoch": 1.027703690957584,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008996547277090408,
      "loss": 1.6611,
      "step": 11945
    },
    {
      "epoch": 1.0277897272649057,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008996384778159673,
      "loss": 0.8929,
      "step": 11946
    },
    {
      "epoch": 1.0278757635722275,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008996222267540291,
      "loss": 1.2772,
      "step": 11947
    },
    {
      "epoch": 1.0279617998795492,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008996059745232734,
      "loss": 1.1072,
      "step": 11948
    },
    {
      "epoch": 1.0280478361868708,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008995897211237481,
      "loss": 1.373,
      "step": 11949
    },
    {
      "epoch": 1.0281338724941926,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008995734665555005,
      "loss": 0.8929,
      "step": 11950
    },
    {
      "epoch": 1.0282199088015143,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008995572108185782,
      "loss": 1.6409,
      "step": 11951
    },
    {
      "epoch": 1.0283059451088359,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008995409539130289,
      "loss": 1.4184,
      "step": 11952
    },
    {
      "epoch": 1.0283919814161577,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008995246958388998,
      "loss": 1.2384,
      "step": 11953
    },
    {
      "epoch": 1.0284780177234794,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0008995084365962388,
      "loss": 1.6735,
      "step": 11954
    },
    {
      "epoch": 1.028564054030801,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008994921761850934,
      "loss": 1.2221,
      "step": 11955
    },
    {
      "epoch": 1.0286500903381226,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000899475914605511,
      "loss": 1.6745,
      "step": 11956
    },
    {
      "epoch": 1.0287361266454444,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008994596518575392,
      "loss": 1.0048,
      "step": 11957
    },
    {
      "epoch": 1.028822162952766,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008994433879412256,
      "loss": 1.1326,
      "step": 11958
    },
    {
      "epoch": 1.0289081992600877,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008994271228566178,
      "loss": 0.8221,
      "step": 11959
    },
    {
      "epoch": 1.0289942355674095,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008994108566037634,
      "loss": 1.6318,
      "step": 11960
    },
    {
      "epoch": 1.0290802718747312,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00089939458918271,
      "loss": 1.4023,
      "step": 11961
    },
    {
      "epoch": 1.0291663081820528,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008993783205935049,
      "loss": 0.8181,
      "step": 11962
    },
    {
      "epoch": 1.0292523444893744,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000899362050836196,
      "loss": 0.8122,
      "step": 11963
    },
    {
      "epoch": 1.0293383807966963,
      "grad_norm": 1.78125,
      "learning_rate": 0.0008993457799108309,
      "loss": 1.5528,
      "step": 11964
    },
    {
      "epoch": 1.0294244171040179,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008993295078174568,
      "loss": 1.0559,
      "step": 11965
    },
    {
      "epoch": 1.0295104534113395,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008993132345561217,
      "loss": 1.5773,
      "step": 11966
    },
    {
      "epoch": 1.0295964897186614,
      "grad_norm": 1.21875,
      "learning_rate": 0.000899296960126873,
      "loss": 1.6395,
      "step": 11967
    },
    {
      "epoch": 1.029682526025983,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008992806845297585,
      "loss": 0.8902,
      "step": 11968
    },
    {
      "epoch": 1.0297685623333046,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008992644077648256,
      "loss": 0.7968,
      "step": 11969
    },
    {
      "epoch": 1.0298545986406265,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008992481298321218,
      "loss": 1.3213,
      "step": 11970
    },
    {
      "epoch": 1.029940634947948,
      "grad_norm": 0.93359375,
      "learning_rate": 0.000899231850731695,
      "loss": 0.7504,
      "step": 11971
    },
    {
      "epoch": 1.0300266712552697,
      "grad_norm": 1.8125,
      "learning_rate": 0.0008992155704635926,
      "loss": 1.4261,
      "step": 11972
    },
    {
      "epoch": 1.0301127075625913,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008991992890278625,
      "loss": 1.2092,
      "step": 11973
    },
    {
      "epoch": 1.0301987438699132,
      "grad_norm": 2.53125,
      "learning_rate": 0.0008991830064245519,
      "loss": 1.4548,
      "step": 11974
    },
    {
      "epoch": 1.0302847801772348,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008991667226537085,
      "loss": 1.0766,
      "step": 11975
    },
    {
      "epoch": 1.0303708164845564,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008991504377153804,
      "loss": 1.8995,
      "step": 11976
    },
    {
      "epoch": 1.0304568527918783,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008991341516096147,
      "loss": 1.4279,
      "step": 11977
    },
    {
      "epoch": 1.0305428890991999,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008991178643364595,
      "loss": 0.7487,
      "step": 11978
    },
    {
      "epoch": 1.0306289254065215,
      "grad_norm": 3.59375,
      "learning_rate": 0.000899101575895962,
      "loss": 0.828,
      "step": 11979
    },
    {
      "epoch": 1.0307149617138431,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00089908528628817,
      "loss": 1.142,
      "step": 11980
    },
    {
      "epoch": 1.030800998021165,
      "grad_norm": 2.640625,
      "learning_rate": 0.0008990689955131311,
      "loss": 2.1229,
      "step": 11981
    },
    {
      "epoch": 1.0308870343284866,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008990527035708931,
      "loss": 1.489,
      "step": 11982
    },
    {
      "epoch": 1.0309730706358082,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008990364104615034,
      "loss": 1.2153,
      "step": 11983
    },
    {
      "epoch": 1.03105910694313,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00089902011618501,
      "loss": 0.853,
      "step": 11984
    },
    {
      "epoch": 1.0311451432504517,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008990038207414603,
      "loss": 1.0724,
      "step": 11985
    },
    {
      "epoch": 1.0312311795577733,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008989875241309021,
      "loss": 0.6648,
      "step": 11986
    },
    {
      "epoch": 1.0313172158650952,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008989712263533828,
      "loss": 1.1645,
      "step": 11987
    },
    {
      "epoch": 1.0314032521724168,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008989549274089504,
      "loss": 0.9134,
      "step": 11988
    },
    {
      "epoch": 1.0314892884797384,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008989386272976522,
      "loss": 1.1379,
      "step": 11989
    },
    {
      "epoch": 1.03157532478706,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008989223260195365,
      "loss": 1.6123,
      "step": 11990
    },
    {
      "epoch": 1.031661361094382,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008989060235746503,
      "loss": 1.5095,
      "step": 11991
    },
    {
      "epoch": 1.0317473974017035,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008988897199630417,
      "loss": 1.1818,
      "step": 11992
    },
    {
      "epoch": 1.0318334337090251,
      "grad_norm": 1.25,
      "learning_rate": 0.0008988734151847582,
      "loss": 1.1859,
      "step": 11993
    },
    {
      "epoch": 1.031919470016347,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008988571092398474,
      "loss": 1.2503,
      "step": 11994
    },
    {
      "epoch": 1.0320055063236686,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008988408021283573,
      "loss": 1.2497,
      "step": 11995
    },
    {
      "epoch": 1.0320915426309902,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008988244938503354,
      "loss": 0.8246,
      "step": 11996
    },
    {
      "epoch": 1.0321775789383119,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008988081844058293,
      "loss": 1.323,
      "step": 11997
    },
    {
      "epoch": 1.0322636152456337,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000898791873794887,
      "loss": 1.1469,
      "step": 11998
    },
    {
      "epoch": 1.0323496515529553,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000898775562017556,
      "loss": 1.9879,
      "step": 11999
    },
    {
      "epoch": 1.032435687860277,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000898759249073884,
      "loss": 0.9897,
      "step": 12000
    },
    {
      "epoch": 1.0325217241675988,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008987429349639187,
      "loss": 1.5638,
      "step": 12001
    },
    {
      "epoch": 1.0326077604749204,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000898726619687708,
      "loss": 0.7889,
      "step": 12002
    },
    {
      "epoch": 1.032693796782242,
      "grad_norm": 1.25,
      "learning_rate": 0.0008987103032452994,
      "loss": 0.7389,
      "step": 12003
    },
    {
      "epoch": 1.032779833089564,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008986939856367406,
      "loss": 0.9854,
      "step": 12004
    },
    {
      "epoch": 1.0328658693968855,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008986776668620795,
      "loss": 1.0541,
      "step": 12005
    },
    {
      "epoch": 1.0329519057042071,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008986613469213638,
      "loss": 0.7362,
      "step": 12006
    },
    {
      "epoch": 1.0330379420115288,
      "grad_norm": 2.4375,
      "learning_rate": 0.0008986450258146412,
      "loss": 1.1407,
      "step": 12007
    },
    {
      "epoch": 1.0331239783188506,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008986287035419596,
      "loss": 1.1109,
      "step": 12008
    },
    {
      "epoch": 1.0332100146261722,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008986123801033663,
      "loss": 0.7693,
      "step": 12009
    },
    {
      "epoch": 1.0332960509334939,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008985960554989095,
      "loss": 1.5245,
      "step": 12010
    },
    {
      "epoch": 1.0333820872408157,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008985797297286366,
      "loss": 0.8583,
      "step": 12011
    },
    {
      "epoch": 1.0334681235481373,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008985634027925955,
      "loss": 1.5738,
      "step": 12012
    },
    {
      "epoch": 1.033554159855459,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008985470746908342,
      "loss": 0.7621,
      "step": 12013
    },
    {
      "epoch": 1.0336401961627808,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008985307454234,
      "loss": 0.7542,
      "step": 12014
    },
    {
      "epoch": 1.0337262324701024,
      "grad_norm": 1.453125,
      "learning_rate": 0.000898514414990341,
      "loss": 1.8403,
      "step": 12015
    },
    {
      "epoch": 1.033812268777424,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008984980833917047,
      "loss": 1.1686,
      "step": 12016
    },
    {
      "epoch": 1.0338983050847457,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008984817506275391,
      "loss": 1.7548,
      "step": 12017
    },
    {
      "epoch": 1.0339843413920675,
      "grad_norm": 1.15625,
      "learning_rate": 0.000898465416697892,
      "loss": 1.0453,
      "step": 12018
    },
    {
      "epoch": 1.0340703776993891,
      "grad_norm": 1.375,
      "learning_rate": 0.0008984490816028108,
      "loss": 0.8987,
      "step": 12019
    },
    {
      "epoch": 1.0341564140067108,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008984327453423437,
      "loss": 0.7958,
      "step": 12020
    },
    {
      "epoch": 1.0342424503140326,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008984164079165382,
      "loss": 0.5867,
      "step": 12021
    },
    {
      "epoch": 1.0343284866213542,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008984000693254424,
      "loss": 1.1736,
      "step": 12022
    },
    {
      "epoch": 1.0344145229286759,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008983837295691038,
      "loss": 0.9394,
      "step": 12023
    },
    {
      "epoch": 1.0345005592359975,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008983673886475701,
      "loss": 1.5988,
      "step": 12024
    },
    {
      "epoch": 1.0345865955433193,
      "grad_norm": 1.921875,
      "learning_rate": 0.0008983510465608895,
      "loss": 1.5921,
      "step": 12025
    },
    {
      "epoch": 1.034672631850641,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008983347033091094,
      "loss": 1.9324,
      "step": 12026
    },
    {
      "epoch": 1.0347586681579626,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008983183588922778,
      "loss": 0.9773,
      "step": 12027
    },
    {
      "epoch": 1.0348447044652844,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008983020133104425,
      "loss": 0.9609,
      "step": 12028
    },
    {
      "epoch": 1.034930740772606,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008982856665636512,
      "loss": 1.2186,
      "step": 12029
    },
    {
      "epoch": 1.0350167770799277,
      "grad_norm": 1.3125,
      "learning_rate": 0.000898269318651952,
      "loss": 0.8478,
      "step": 12030
    },
    {
      "epoch": 1.0351028133872495,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008982529695753923,
      "loss": 1.3749,
      "step": 12031
    },
    {
      "epoch": 1.0351888496945711,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008982366193340201,
      "loss": 1.5639,
      "step": 12032
    },
    {
      "epoch": 1.0352748860018928,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008982202679278834,
      "loss": 1.6864,
      "step": 12033
    },
    {
      "epoch": 1.0353609223092144,
      "grad_norm": 15.9375,
      "learning_rate": 0.0008982039153570297,
      "loss": 1.0562,
      "step": 12034
    },
    {
      "epoch": 1.0354469586165362,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008981875616215072,
      "loss": 0.7891,
      "step": 12035
    },
    {
      "epoch": 1.0355329949238579,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008981712067213634,
      "loss": 1.3766,
      "step": 12036
    },
    {
      "epoch": 1.0356190312311795,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008981548506566461,
      "loss": 0.9137,
      "step": 12037
    },
    {
      "epoch": 1.0357050675385013,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008981384934274035,
      "loss": 1.6832,
      "step": 12038
    },
    {
      "epoch": 1.035791103845823,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008981221350336832,
      "loss": 1.3501,
      "step": 12039
    },
    {
      "epoch": 1.0358771401531446,
      "grad_norm": 1.4375,
      "learning_rate": 0.000898105775475533,
      "loss": 0.9105,
      "step": 12040
    },
    {
      "epoch": 1.0359631764604664,
      "grad_norm": 6.75,
      "learning_rate": 0.000898089414753001,
      "loss": 1.0942,
      "step": 12041
    },
    {
      "epoch": 1.036049212767788,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008980730528661349,
      "loss": 0.8384,
      "step": 12042
    },
    {
      "epoch": 1.0361352490751097,
      "grad_norm": 4.71875,
      "learning_rate": 0.0008980566898149824,
      "loss": 1.5319,
      "step": 12043
    },
    {
      "epoch": 1.0362212853824313,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008980403255995915,
      "loss": 1.2653,
      "step": 12044
    },
    {
      "epoch": 1.0363073216897531,
      "grad_norm": 2.25,
      "learning_rate": 0.0008980239602200102,
      "loss": 1.6038,
      "step": 12045
    },
    {
      "epoch": 1.0363933579970748,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008980075936762861,
      "loss": 1.0623,
      "step": 12046
    },
    {
      "epoch": 1.0364793943043964,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008979912259684672,
      "loss": 1.191,
      "step": 12047
    },
    {
      "epoch": 1.0365654306117182,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008979748570966014,
      "loss": 1.9286,
      "step": 12048
    },
    {
      "epoch": 1.0366514669190399,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008979584870607366,
      "loss": 1.3991,
      "step": 12049
    },
    {
      "epoch": 1.0367375032263615,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008979421158609206,
      "loss": 0.9895,
      "step": 12050
    },
    {
      "epoch": 1.0368235395336831,
      "grad_norm": 2.359375,
      "learning_rate": 0.0008979257434972012,
      "loss": 0.6246,
      "step": 12051
    },
    {
      "epoch": 1.036909575841005,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008979093699696266,
      "loss": 0.9809,
      "step": 12052
    },
    {
      "epoch": 1.0369956121483266,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008978929952782442,
      "loss": 1.7081,
      "step": 12053
    },
    {
      "epoch": 1.0370816484556482,
      "grad_norm": 1.375,
      "learning_rate": 0.0008978766194231024,
      "loss": 1.1023,
      "step": 12054
    },
    {
      "epoch": 1.03716768476297,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008978602424042488,
      "loss": 1.2917,
      "step": 12055
    },
    {
      "epoch": 1.0372537210702917,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008978438642217312,
      "loss": 1.6859,
      "step": 12056
    },
    {
      "epoch": 1.0373397573776133,
      "grad_norm": 1.203125,
      "learning_rate": 0.000897827484875598,
      "loss": 0.9678,
      "step": 12057
    },
    {
      "epoch": 1.0374257936849351,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008978111043658966,
      "loss": 1.3194,
      "step": 12058
    },
    {
      "epoch": 1.0375118299922568,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008977947226926751,
      "loss": 1.04,
      "step": 12059
    },
    {
      "epoch": 1.0375978662995784,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008977783398559814,
      "loss": 1.0786,
      "step": 12060
    },
    {
      "epoch": 1.0376839026069,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008977619558558634,
      "loss": 1.4048,
      "step": 12061
    },
    {
      "epoch": 1.0377699389142219,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008977455706923692,
      "loss": 0.755,
      "step": 12062
    },
    {
      "epoch": 1.0378559752215435,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008977291843655464,
      "loss": 1.3824,
      "step": 12063
    },
    {
      "epoch": 1.0379420115288651,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008977127968754431,
      "loss": 1.0646,
      "step": 12064
    },
    {
      "epoch": 1.038028047836187,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008976964082221073,
      "loss": 1.0549,
      "step": 12065
    },
    {
      "epoch": 1.0381140841435086,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008976800184055868,
      "loss": 1.2709,
      "step": 12066
    },
    {
      "epoch": 1.0382001204508302,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008976636274259295,
      "loss": 1.1412,
      "step": 12067
    },
    {
      "epoch": 1.0382861567581518,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008976472352831835,
      "loss": 1.0163,
      "step": 12068
    },
    {
      "epoch": 1.0383721930654737,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008976308419773967,
      "loss": 1.4808,
      "step": 12069
    },
    {
      "epoch": 1.0384582293727953,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000897614447508617,
      "loss": 1.4583,
      "step": 12070
    },
    {
      "epoch": 1.038544265680117,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008975980518768924,
      "loss": 1.1219,
      "step": 12071
    },
    {
      "epoch": 1.0386303019874388,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008975816550822708,
      "loss": 1.0283,
      "step": 12072
    },
    {
      "epoch": 1.0387163382947604,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008975652571248003,
      "loss": 1.4605,
      "step": 12073
    },
    {
      "epoch": 1.038802374602082,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008975488580045286,
      "loss": 0.6868,
      "step": 12074
    },
    {
      "epoch": 1.0388884109094039,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008975324577215039,
      "loss": 0.8105,
      "step": 12075
    },
    {
      "epoch": 1.0389744472167255,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008975160562757741,
      "loss": 1.0035,
      "step": 12076
    },
    {
      "epoch": 1.0390604835240471,
      "grad_norm": 1.9921875,
      "learning_rate": 0.000897499653667387,
      "loss": 1.6795,
      "step": 12077
    },
    {
      "epoch": 1.0391465198313687,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008974832498963907,
      "loss": 1.6144,
      "step": 12078
    },
    {
      "epoch": 1.0392325561386906,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008974668449628334,
      "loss": 1.5126,
      "step": 12079
    },
    {
      "epoch": 1.0393185924460122,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008974504388667629,
      "loss": 1.3686,
      "step": 12080
    },
    {
      "epoch": 1.0394046287533338,
      "grad_norm": 1.0625,
      "learning_rate": 0.000897434031608227,
      "loss": 0.8991,
      "step": 12081
    },
    {
      "epoch": 1.0394906650606557,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008974176231872739,
      "loss": 0.9106,
      "step": 12082
    },
    {
      "epoch": 1.0395767013679773,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008974012136039515,
      "loss": 1.0681,
      "step": 12083
    },
    {
      "epoch": 1.039662737675299,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008973848028583078,
      "loss": 1.1409,
      "step": 12084
    },
    {
      "epoch": 1.0397487739826206,
      "grad_norm": 1.46875,
      "learning_rate": 0.000897368390950391,
      "loss": 1.7499,
      "step": 12085
    },
    {
      "epoch": 1.0398348102899424,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008973519778802489,
      "loss": 2.1184,
      "step": 12086
    },
    {
      "epoch": 1.039920846597264,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008973355636479294,
      "loss": 1.5451,
      "step": 12087
    },
    {
      "epoch": 1.0400068829045856,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008973191482534808,
      "loss": 1.5858,
      "step": 12088
    },
    {
      "epoch": 1.0400929192119075,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008973027316969511,
      "loss": 1.2993,
      "step": 12089
    },
    {
      "epoch": 1.0401789555192291,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008972863139783878,
      "loss": 0.6919,
      "step": 12090
    },
    {
      "epoch": 1.0402649918265507,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008972698950978396,
      "loss": 1.9573,
      "step": 12091
    },
    {
      "epoch": 1.0403510281338726,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008972534750553542,
      "loss": 0.9836,
      "step": 12092
    },
    {
      "epoch": 1.0404370644411942,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008972370538509795,
      "loss": 1.2616,
      "step": 12093
    },
    {
      "epoch": 1.0405231007485158,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008972206314847639,
      "loss": 0.7303,
      "step": 12094
    },
    {
      "epoch": 1.0406091370558375,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008972042079567551,
      "loss": 0.9924,
      "step": 12095
    },
    {
      "epoch": 1.0406951733631593,
      "grad_norm": 1.515625,
      "learning_rate": 0.000897187783267001,
      "loss": 1.1372,
      "step": 12096
    },
    {
      "epoch": 1.040781209670481,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008971713574155503,
      "loss": 1.1191,
      "step": 12097
    },
    {
      "epoch": 1.0408672459778026,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008971549304024503,
      "loss": 1.0955,
      "step": 12098
    },
    {
      "epoch": 1.0409532822851244,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008971385022277495,
      "loss": 1.2435,
      "step": 12099
    },
    {
      "epoch": 1.041039318592446,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008971220728914959,
      "loss": 2.4153,
      "step": 12100
    },
    {
      "epoch": 1.0411253548997677,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008971056423937374,
      "loss": 1.3882,
      "step": 12101
    },
    {
      "epoch": 1.0412113912070895,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008970892107345221,
      "loss": 1.5188,
      "step": 12102
    },
    {
      "epoch": 1.0412974275144111,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008970727779138981,
      "loss": 1.1167,
      "step": 12103
    },
    {
      "epoch": 1.0413834638217327,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008970563439319134,
      "loss": 0.9746,
      "step": 12104
    },
    {
      "epoch": 1.0414695001290544,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008970399087886161,
      "loss": 1.3928,
      "step": 12105
    },
    {
      "epoch": 1.0415555364363762,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008970234724840544,
      "loss": 1.6207,
      "step": 12106
    },
    {
      "epoch": 1.0416415727436978,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008970070350182762,
      "loss": 0.8711,
      "step": 12107
    },
    {
      "epoch": 1.0417276090510195,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008969905963913296,
      "loss": 1.0098,
      "step": 12108
    },
    {
      "epoch": 1.0418136453583413,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008969741566032627,
      "loss": 1.7914,
      "step": 12109
    },
    {
      "epoch": 1.041899681665663,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008969577156541237,
      "loss": 1.6121,
      "step": 12110
    },
    {
      "epoch": 1.0419857179729846,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008969412735439607,
      "loss": 1.5336,
      "step": 12111
    },
    {
      "epoch": 1.0420717542803062,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008969248302728214,
      "loss": 1.8722,
      "step": 12112
    },
    {
      "epoch": 1.042157790587628,
      "grad_norm": 1.5,
      "learning_rate": 0.0008969083858407544,
      "loss": 1.6726,
      "step": 12113
    },
    {
      "epoch": 1.0422438268949497,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0008968919402478075,
      "loss": 1.3577,
      "step": 12114
    },
    {
      "epoch": 1.0423298632022713,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008968754934940287,
      "loss": 1.0355,
      "step": 12115
    },
    {
      "epoch": 1.0424158995095931,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008968590455794665,
      "loss": 1.0818,
      "step": 12116
    },
    {
      "epoch": 1.0425019358169147,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008968425965041687,
      "loss": 1.617,
      "step": 12117
    },
    {
      "epoch": 1.0425879721242364,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008968261462681834,
      "loss": 1.3379,
      "step": 12118
    },
    {
      "epoch": 1.0426740084315582,
      "grad_norm": 1.09375,
      "learning_rate": 0.000896809694871559,
      "loss": 1.5352,
      "step": 12119
    },
    {
      "epoch": 1.0427600447388798,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008967932423143432,
      "loss": 0.9467,
      "step": 12120
    },
    {
      "epoch": 1.0428460810462015,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008967767885965844,
      "loss": 2.0124,
      "step": 12121
    },
    {
      "epoch": 1.042932117353523,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008967603337183307,
      "loss": 2.1904,
      "step": 12122
    },
    {
      "epoch": 1.043018153660845,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008967438776796301,
      "loss": 0.9301,
      "step": 12123
    },
    {
      "epoch": 1.0431041899681666,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008967274204805309,
      "loss": 1.4343,
      "step": 12124
    },
    {
      "epoch": 1.0431902262754882,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008967109621210812,
      "loss": 1.7767,
      "step": 12125
    },
    {
      "epoch": 1.04327626258281,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008966945026013289,
      "loss": 0.6177,
      "step": 12126
    },
    {
      "epoch": 1.0433622988901317,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008966780419213226,
      "loss": 1.7146,
      "step": 12127
    },
    {
      "epoch": 1.0434483351974533,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00089666158008111,
      "loss": 1.577,
      "step": 12128
    },
    {
      "epoch": 1.0435343715047751,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008966451170807395,
      "loss": 1.3356,
      "step": 12129
    },
    {
      "epoch": 1.0436204078120968,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008966286529202592,
      "loss": 1.0677,
      "step": 12130
    },
    {
      "epoch": 1.0437064441194184,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008966121875997171,
      "loss": 1.0878,
      "step": 12131
    },
    {
      "epoch": 1.04379248042674,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008965957211191616,
      "loss": 1.5787,
      "step": 12132
    },
    {
      "epoch": 1.0438785167340618,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008965792534786406,
      "loss": 0.9905,
      "step": 12133
    },
    {
      "epoch": 1.0439645530413835,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008965627846782027,
      "loss": 1.645,
      "step": 12134
    },
    {
      "epoch": 1.044050589348705,
      "grad_norm": 1.375,
      "learning_rate": 0.0008965463147178955,
      "loss": 0.7519,
      "step": 12135
    },
    {
      "epoch": 1.044136625656027,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008965298435977676,
      "loss": 1.4398,
      "step": 12136
    },
    {
      "epoch": 1.0442226619633486,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000896513371317867,
      "loss": 0.9008,
      "step": 12137
    },
    {
      "epoch": 1.0443086982706702,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008964968978782419,
      "loss": 1.2826,
      "step": 12138
    },
    {
      "epoch": 1.0443947345779918,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008964804232789405,
      "loss": 1.1045,
      "step": 12139
    },
    {
      "epoch": 1.0444807708853137,
      "grad_norm": 2.6875,
      "learning_rate": 0.0008964639475200109,
      "loss": 1.0929,
      "step": 12140
    },
    {
      "epoch": 1.0445668071926353,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008964474706015013,
      "loss": 1.5694,
      "step": 12141
    },
    {
      "epoch": 1.044652843499957,
      "grad_norm": 2.234375,
      "learning_rate": 0.0008964309925234602,
      "loss": 1.4006,
      "step": 12142
    },
    {
      "epoch": 1.0447388798072788,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008964145132859353,
      "loss": 1.3444,
      "step": 12143
    },
    {
      "epoch": 1.0448249161146004,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008963980328889753,
      "loss": 1.6182,
      "step": 12144
    },
    {
      "epoch": 1.044910952421922,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000896381551332628,
      "loss": 0.9599,
      "step": 12145
    },
    {
      "epoch": 1.0449969887292438,
      "grad_norm": 1.0,
      "learning_rate": 0.0008963650686169418,
      "loss": 1.0389,
      "step": 12146
    },
    {
      "epoch": 1.0450830250365655,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008963485847419647,
      "loss": 2.2108,
      "step": 12147
    },
    {
      "epoch": 1.045169061343887,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008963320997077453,
      "loss": 1.2076,
      "step": 12148
    },
    {
      "epoch": 1.0452550976512087,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008963156135143315,
      "loss": 1.3197,
      "step": 12149
    },
    {
      "epoch": 1.0453411339585306,
      "grad_norm": 2.15625,
      "learning_rate": 0.0008962991261617717,
      "loss": 1.4671,
      "step": 12150
    },
    {
      "epoch": 1.0454271702658522,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008962826376501139,
      "loss": 0.9759,
      "step": 12151
    },
    {
      "epoch": 1.0455132065731738,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008962661479794066,
      "loss": 1.6989,
      "step": 12152
    },
    {
      "epoch": 1.0455992428804957,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008962496571496976,
      "loss": 1.0404,
      "step": 12153
    },
    {
      "epoch": 1.0456852791878173,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008962331651610357,
      "loss": 1.4975,
      "step": 12154
    },
    {
      "epoch": 1.045771315495139,
      "grad_norm": 2.109375,
      "learning_rate": 0.0008962166720134688,
      "loss": 1.3943,
      "step": 12155
    },
    {
      "epoch": 1.0458573518024605,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008962001777070451,
      "loss": 2.0152,
      "step": 12156
    },
    {
      "epoch": 1.0459433881097824,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000896183682241813,
      "loss": 1.7318,
      "step": 12157
    },
    {
      "epoch": 1.046029424417104,
      "grad_norm": 1.84375,
      "learning_rate": 0.0008961671856178206,
      "loss": 1.532,
      "step": 12158
    },
    {
      "epoch": 1.0461154607244256,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008961506878351164,
      "loss": 1.8002,
      "step": 12159
    },
    {
      "epoch": 1.0462014970317475,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008961341888937484,
      "loss": 0.9184,
      "step": 12160
    },
    {
      "epoch": 1.046287533339069,
      "grad_norm": 1.125,
      "learning_rate": 0.0008961176887937648,
      "loss": 1.4436,
      "step": 12161
    },
    {
      "epoch": 1.0463735696463907,
      "grad_norm": 1.5,
      "learning_rate": 0.0008961011875352143,
      "loss": 1.3679,
      "step": 12162
    },
    {
      "epoch": 1.0464596059537126,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008960846851181445,
      "loss": 1.2527,
      "step": 12163
    },
    {
      "epoch": 1.0465456422610342,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008960681815426044,
      "loss": 1.5335,
      "step": 12164
    },
    {
      "epoch": 1.0466316785683558,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008960516768086417,
      "loss": 0.7732,
      "step": 12165
    },
    {
      "epoch": 1.0467177148756774,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000896035170916305,
      "loss": 2.2791,
      "step": 12166
    },
    {
      "epoch": 1.0468037511829993,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008960186638656422,
      "loss": 1.6503,
      "step": 12167
    },
    {
      "epoch": 1.046889787490321,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008960021556567021,
      "loss": 1.1139,
      "step": 12168
    },
    {
      "epoch": 1.0469758237976425,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008959856462895325,
      "loss": 1.3047,
      "step": 12169
    },
    {
      "epoch": 1.0470618601049644,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008959691357641821,
      "loss": 1.1669,
      "step": 12170
    },
    {
      "epoch": 1.047147896412286,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008959526240806988,
      "loss": 2.144,
      "step": 12171
    },
    {
      "epoch": 1.0472339327196076,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008959361112391311,
      "loss": 0.8133,
      "step": 12172
    },
    {
      "epoch": 1.0473199690269293,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008959195972395274,
      "loss": 0.5656,
      "step": 12173
    },
    {
      "epoch": 1.047406005334251,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008959030820819358,
      "loss": 1.1767,
      "step": 12174
    },
    {
      "epoch": 1.0474920416415727,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008958865657664048,
      "loss": 1.0124,
      "step": 12175
    },
    {
      "epoch": 1.0475780779488943,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008958700482929825,
      "loss": 1.7779,
      "step": 12176
    },
    {
      "epoch": 1.0476641142562162,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008958535296617173,
      "loss": 0.8739,
      "step": 12177
    },
    {
      "epoch": 1.0477501505635378,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008958370098726575,
      "loss": 1.0303,
      "step": 12178
    },
    {
      "epoch": 1.0478361868708594,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008958204889258515,
      "loss": 1.0803,
      "step": 12179
    },
    {
      "epoch": 1.0479222231781813,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008958039668213474,
      "loss": 1.8984,
      "step": 12180
    },
    {
      "epoch": 1.048008259485503,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008957874435591938,
      "loss": 1.04,
      "step": 12181
    },
    {
      "epoch": 1.0480942957928245,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008957709191394389,
      "loss": 0.9628,
      "step": 12182
    },
    {
      "epoch": 1.0481803321001462,
      "grad_norm": 1.9609375,
      "learning_rate": 0.000895754393562131,
      "loss": 1.5059,
      "step": 12183
    },
    {
      "epoch": 1.048266368407468,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008957378668273185,
      "loss": 0.9989,
      "step": 12184
    },
    {
      "epoch": 1.0483524047147896,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008957213389350497,
      "loss": 1.1005,
      "step": 12185
    },
    {
      "epoch": 1.0484384410221113,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000895704809885373,
      "loss": 1.2089,
      "step": 12186
    },
    {
      "epoch": 1.048524477329433,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008956882796783365,
      "loss": 1.109,
      "step": 12187
    },
    {
      "epoch": 1.0486105136367547,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008956717483139889,
      "loss": 1.2419,
      "step": 12188
    },
    {
      "epoch": 1.0486965499440764,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008956552157923783,
      "loss": 1.4414,
      "step": 12189
    },
    {
      "epoch": 1.0487825862513982,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008956386821135534,
      "loss": 1.093,
      "step": 12190
    },
    {
      "epoch": 1.0488686225587198,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008956221472775621,
      "loss": 1.0697,
      "step": 12191
    },
    {
      "epoch": 1.0489546588660414,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008956056112844528,
      "loss": 1.3164,
      "step": 12192
    },
    {
      "epoch": 1.049040695173363,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008955890741342743,
      "loss": 1.0834,
      "step": 12193
    },
    {
      "epoch": 1.049126731480685,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008955725358270745,
      "loss": 0.5262,
      "step": 12194
    },
    {
      "epoch": 1.0492127677880065,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000895555996362902,
      "loss": 0.9295,
      "step": 12195
    },
    {
      "epoch": 1.0492988040953282,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008955394557418054,
      "loss": 1.485,
      "step": 12196
    },
    {
      "epoch": 1.04938484040265,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008955229139638326,
      "loss": 1.3256,
      "step": 12197
    },
    {
      "epoch": 1.0494708767099716,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008955063710290322,
      "loss": 0.8203,
      "step": 12198
    },
    {
      "epoch": 1.0495569130172933,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008954898269374526,
      "loss": 0.9328,
      "step": 12199
    },
    {
      "epoch": 1.0496429493246149,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008954732816891422,
      "loss": 1.4023,
      "step": 12200
    },
    {
      "epoch": 1.0497289856319367,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008954567352841494,
      "loss": 1.4868,
      "step": 12201
    },
    {
      "epoch": 1.0498150219392584,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008954401877225225,
      "loss": 0.582,
      "step": 12202
    },
    {
      "epoch": 1.04990105824658,
      "grad_norm": 1.125,
      "learning_rate": 0.00089542363900431,
      "loss": 1.3829,
      "step": 12203
    },
    {
      "epoch": 1.0499870945539018,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008954070891295602,
      "loss": 1.6001,
      "step": 12204
    },
    {
      "epoch": 1.0500731308612234,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008953905380983216,
      "loss": 1.3481,
      "step": 12205
    },
    {
      "epoch": 1.050159167168545,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008953739859106425,
      "loss": 1.6731,
      "step": 12206
    },
    {
      "epoch": 1.050245203475867,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008953574325665716,
      "loss": 1.0703,
      "step": 12207
    },
    {
      "epoch": 1.0503312397831885,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008953408780661569,
      "loss": 0.8617,
      "step": 12208
    },
    {
      "epoch": 1.0504172760905102,
      "grad_norm": 1.84375,
      "learning_rate": 0.0008953243224094471,
      "loss": 1.2148,
      "step": 12209
    },
    {
      "epoch": 1.0505033123978318,
      "grad_norm": 1.0,
      "learning_rate": 0.0008953077655964905,
      "loss": 0.7361,
      "step": 12210
    },
    {
      "epoch": 1.0505893487051536,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008952912076273357,
      "loss": 1.2546,
      "step": 12211
    },
    {
      "epoch": 1.0506753850124753,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008952746485020309,
      "loss": 0.8747,
      "step": 12212
    },
    {
      "epoch": 1.0507614213197969,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008952580882206245,
      "loss": 1.4542,
      "step": 12213
    },
    {
      "epoch": 1.0508474576271187,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008952415267831653,
      "loss": 1.0184,
      "step": 12214
    },
    {
      "epoch": 1.0509334939344404,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008952249641897013,
      "loss": 0.7929,
      "step": 12215
    },
    {
      "epoch": 1.051019530241762,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008952084004402813,
      "loss": 1.3448,
      "step": 12216
    },
    {
      "epoch": 1.0511055665490838,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008951918355349536,
      "loss": 1.7588,
      "step": 12217
    },
    {
      "epoch": 1.0511916028564054,
      "grad_norm": 1.125,
      "learning_rate": 0.0008951752694737664,
      "loss": 0.7848,
      "step": 12218
    },
    {
      "epoch": 1.051277639163727,
      "grad_norm": 1.890625,
      "learning_rate": 0.0008951587022567686,
      "loss": 1.575,
      "step": 12219
    },
    {
      "epoch": 1.0513636754710487,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008951421338840084,
      "loss": 1.2371,
      "step": 12220
    },
    {
      "epoch": 1.0514497117783705,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008951255643555345,
      "loss": 1.4289,
      "step": 12221
    },
    {
      "epoch": 1.0515357480856922,
      "grad_norm": 1.0,
      "learning_rate": 0.000895108993671395,
      "loss": 1.0207,
      "step": 12222
    },
    {
      "epoch": 1.0516217843930138,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008950924218316385,
      "loss": 2.1258,
      "step": 12223
    },
    {
      "epoch": 1.0517078207003356,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008950758488363136,
      "loss": 1.858,
      "step": 12224
    },
    {
      "epoch": 1.0517938570076573,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008950592746854685,
      "loss": 1.1537,
      "step": 12225
    },
    {
      "epoch": 1.0518798933149789,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008950426993791521,
      "loss": 0.9663,
      "step": 12226
    },
    {
      "epoch": 1.0519659296223005,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008950261229174124,
      "loss": 1.8457,
      "step": 12227
    },
    {
      "epoch": 1.0520519659296224,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008950095453002983,
      "loss": 1.5295,
      "step": 12228
    },
    {
      "epoch": 1.052138002236944,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000894992966527858,
      "loss": 0.8498,
      "step": 12229
    },
    {
      "epoch": 1.0522240385442656,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008949763866001402,
      "loss": 1.1597,
      "step": 12230
    },
    {
      "epoch": 1.0523100748515875,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008949598055171933,
      "loss": 0.8289,
      "step": 12231
    },
    {
      "epoch": 1.052396111158909,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008949432232790656,
      "loss": 1.6278,
      "step": 12232
    },
    {
      "epoch": 1.0524821474662307,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008949266398858059,
      "loss": 1.4174,
      "step": 12233
    },
    {
      "epoch": 1.0525681837735525,
      "grad_norm": 1.125,
      "learning_rate": 0.0008949100553374625,
      "loss": 1.1019,
      "step": 12234
    },
    {
      "epoch": 1.0526542200808742,
      "grad_norm": 1.828125,
      "learning_rate": 0.0008948934696340842,
      "loss": 1.5017,
      "step": 12235
    },
    {
      "epoch": 1.0527402563881958,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008948768827757192,
      "loss": 1.594,
      "step": 12236
    },
    {
      "epoch": 1.0528262926955174,
      "grad_norm": 1.328125,
      "learning_rate": 0.000894860294762416,
      "loss": 1.4388,
      "step": 12237
    },
    {
      "epoch": 1.0529123290028393,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008948437055942233,
      "loss": 0.9047,
      "step": 12238
    },
    {
      "epoch": 1.0529983653101609,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008948271152711898,
      "loss": 1.6957,
      "step": 12239
    },
    {
      "epoch": 1.0530844016174825,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008948105237933635,
      "loss": 1.6745,
      "step": 12240
    },
    {
      "epoch": 1.0531704379248044,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008947939311607932,
      "loss": 1.1297,
      "step": 12241
    },
    {
      "epoch": 1.053256474232126,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008947773373735275,
      "loss": 1.0153,
      "step": 12242
    },
    {
      "epoch": 1.0533425105394476,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008947607424316149,
      "loss": 1.4023,
      "step": 12243
    },
    {
      "epoch": 1.0534285468467692,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008947441463351038,
      "loss": 1.2252,
      "step": 12244
    },
    {
      "epoch": 1.053514583154091,
      "grad_norm": 1.140625,
      "learning_rate": 0.000894727549084043,
      "loss": 1.8414,
      "step": 12245
    },
    {
      "epoch": 1.0536006194614127,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008947109506784808,
      "loss": 1.0467,
      "step": 12246
    },
    {
      "epoch": 1.0536866557687343,
      "grad_norm": 1.25,
      "learning_rate": 0.0008946943511184657,
      "loss": 1.8119,
      "step": 12247
    },
    {
      "epoch": 1.0537726920760562,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008946777504040466,
      "loss": 0.837,
      "step": 12248
    },
    {
      "epoch": 1.0538587283833778,
      "grad_norm": 1.25,
      "learning_rate": 0.0008946611485352718,
      "loss": 1.2423,
      "step": 12249
    },
    {
      "epoch": 1.0539447646906994,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008946445455121899,
      "loss": 1.1798,
      "step": 12250
    },
    {
      "epoch": 1.0540308009980213,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008946279413348495,
      "loss": 1.5619,
      "step": 12251
    },
    {
      "epoch": 1.054116837305343,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000894611336003299,
      "loss": 1.1905,
      "step": 12252
    },
    {
      "epoch": 1.0542028736126645,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008945947295175871,
      "loss": 1.0108,
      "step": 12253
    },
    {
      "epoch": 1.0542889099199861,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008945781218777625,
      "loss": 1.3085,
      "step": 12254
    },
    {
      "epoch": 1.054374946227308,
      "grad_norm": 2.09375,
      "learning_rate": 0.0008945615130838736,
      "loss": 1.3578,
      "step": 12255
    },
    {
      "epoch": 1.0544609825346296,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008945449031359689,
      "loss": 1.3262,
      "step": 12256
    },
    {
      "epoch": 1.0545470188419512,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008945282920340971,
      "loss": 0.8659,
      "step": 12257
    },
    {
      "epoch": 1.054633055149273,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008945116797783068,
      "loss": 1.4273,
      "step": 12258
    },
    {
      "epoch": 1.0547190914565947,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008944950663686466,
      "loss": 1.8267,
      "step": 12259
    },
    {
      "epoch": 1.0548051277639163,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008944784518051651,
      "loss": 1.5432,
      "step": 12260
    },
    {
      "epoch": 1.054891164071238,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008944618360879108,
      "loss": 0.6879,
      "step": 12261
    },
    {
      "epoch": 1.0549772003785598,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008944452192169324,
      "loss": 1.5813,
      "step": 12262
    },
    {
      "epoch": 1.0550632366858814,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008944286011922782,
      "loss": 1.5376,
      "step": 12263
    },
    {
      "epoch": 1.055149272993203,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008944119820139974,
      "loss": 1.121,
      "step": 12264
    },
    {
      "epoch": 1.055235309300525,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008943953616821381,
      "loss": 1.0874,
      "step": 12265
    },
    {
      "epoch": 1.0553213456078465,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008943787401967489,
      "loss": 1.0293,
      "step": 12266
    },
    {
      "epoch": 1.0554073819151681,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008943621175578789,
      "loss": 1.561,
      "step": 12267
    },
    {
      "epoch": 1.05549341822249,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008943454937655761,
      "loss": 1.4541,
      "step": 12268
    },
    {
      "epoch": 1.0555794545298116,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0008943288688198897,
      "loss": 1.6983,
      "step": 12269
    },
    {
      "epoch": 1.0556654908371332,
      "grad_norm": 1.265625,
      "learning_rate": 0.000894312242720868,
      "loss": 0.7688,
      "step": 12270
    },
    {
      "epoch": 1.0557515271444549,
      "grad_norm": 1.375,
      "learning_rate": 0.0008942956154685595,
      "loss": 1.4144,
      "step": 12271
    },
    {
      "epoch": 1.0558375634517767,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008942789870630132,
      "loss": 0.9341,
      "step": 12272
    },
    {
      "epoch": 1.0559235997590983,
      "grad_norm": 1.625,
      "learning_rate": 0.0008942623575042774,
      "loss": 2.2143,
      "step": 12273
    },
    {
      "epoch": 1.05600963606642,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008942457267924011,
      "loss": 1.031,
      "step": 12274
    },
    {
      "epoch": 1.0560956723737418,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008942290949274325,
      "loss": 1.1329,
      "step": 12275
    },
    {
      "epoch": 1.0561817086810634,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008942124619094208,
      "loss": 1.2004,
      "step": 12276
    },
    {
      "epoch": 1.056267744988385,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0008941958277384139,
      "loss": 1.4284,
      "step": 12277
    },
    {
      "epoch": 1.056353781295707,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008941791924144611,
      "loss": 2.073,
      "step": 12278
    },
    {
      "epoch": 1.0564398176030285,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008941625559376107,
      "loss": 0.9975,
      "step": 12279
    },
    {
      "epoch": 1.0565258539103501,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008941459183079116,
      "loss": 1.2177,
      "step": 12280
    },
    {
      "epoch": 1.0566118902176718,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008941292795254124,
      "loss": 1.8515,
      "step": 12281
    },
    {
      "epoch": 1.0566979265249936,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008941126395901615,
      "loss": 1.3728,
      "step": 12282
    },
    {
      "epoch": 1.0567839628323152,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008940959985022079,
      "loss": 1.8175,
      "step": 12283
    },
    {
      "epoch": 1.0568699991396369,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008940793562616003,
      "loss": 1.17,
      "step": 12284
    },
    {
      "epoch": 1.0569560354469587,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000894062712868387,
      "loss": 1.6386,
      "step": 12285
    },
    {
      "epoch": 1.0570420717542803,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000894046068322617,
      "loss": 2.117,
      "step": 12286
    },
    {
      "epoch": 1.057128108061602,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008940294226243388,
      "loss": 1.0576,
      "step": 12287
    },
    {
      "epoch": 1.0572141443689236,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008940127757736013,
      "loss": 1.3716,
      "step": 12288
    },
    {
      "epoch": 1.0573001806762454,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008939961277704531,
      "loss": 0.837,
      "step": 12289
    },
    {
      "epoch": 1.057386216983567,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008939794786149426,
      "loss": 1.1011,
      "step": 12290
    },
    {
      "epoch": 1.0574722532908887,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008939628283071191,
      "loss": 0.7877,
      "step": 12291
    },
    {
      "epoch": 1.0575582895982105,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008939461768470307,
      "loss": 1.8902,
      "step": 12292
    },
    {
      "epoch": 1.0576443259055321,
      "grad_norm": 1.0,
      "learning_rate": 0.0008939295242347264,
      "loss": 2.0809,
      "step": 12293
    },
    {
      "epoch": 1.0577303622128538,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008939128704702549,
      "loss": 1.3761,
      "step": 12294
    },
    {
      "epoch": 1.0578163985201756,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008938962155536647,
      "loss": 1.8033,
      "step": 12295
    },
    {
      "epoch": 1.0579024348274972,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008938795594850049,
      "loss": 1.3607,
      "step": 12296
    },
    {
      "epoch": 1.0579884711348189,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008938629022643238,
      "loss": 1.195,
      "step": 12297
    },
    {
      "epoch": 1.0580745074421405,
      "grad_norm": 0.8125,
      "learning_rate": 0.0008938462438916705,
      "loss": 0.906,
      "step": 12298
    },
    {
      "epoch": 1.0581605437494623,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008938295843670935,
      "loss": 1.0711,
      "step": 12299
    },
    {
      "epoch": 1.058246580056784,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008938129236906415,
      "loss": 1.2836,
      "step": 12300
    },
    {
      "epoch": 1.0583326163641056,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008937962618623633,
      "loss": 1.6403,
      "step": 12301
    },
    {
      "epoch": 1.0584186526714274,
      "grad_norm": 1.375,
      "learning_rate": 0.0008937795988823076,
      "loss": 1.7455,
      "step": 12302
    },
    {
      "epoch": 1.058504688978749,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008937629347505232,
      "loss": 1.0925,
      "step": 12303
    },
    {
      "epoch": 1.0585907252860707,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008937462694670589,
      "loss": 1.7506,
      "step": 12304
    },
    {
      "epoch": 1.0586767615933925,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008937296030319631,
      "loss": 0.8411,
      "step": 12305
    },
    {
      "epoch": 1.0587627979007141,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000893712935445285,
      "loss": 1.1037,
      "step": 12306
    },
    {
      "epoch": 1.0588488342080358,
      "grad_norm": 1.328125,
      "learning_rate": 0.000893696266707073,
      "loss": 1.7909,
      "step": 12307
    },
    {
      "epoch": 1.0589348705153574,
      "grad_norm": 1.25,
      "learning_rate": 0.0008936795968173761,
      "loss": 0.7887,
      "step": 12308
    },
    {
      "epoch": 1.0590209068226792,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008936629257762428,
      "loss": 1.9384,
      "step": 12309
    },
    {
      "epoch": 1.0591069431300009,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008936462535837222,
      "loss": 1.5457,
      "step": 12310
    },
    {
      "epoch": 1.0591929794373225,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008936295802398627,
      "loss": 2.2013,
      "step": 12311
    },
    {
      "epoch": 1.0592790157446443,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008936129057447133,
      "loss": 1.5698,
      "step": 12312
    },
    {
      "epoch": 1.059365052051966,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008935962300983226,
      "loss": 1.6505,
      "step": 12313
    },
    {
      "epoch": 1.0594510883592876,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008935795533007397,
      "loss": 1.2823,
      "step": 12314
    },
    {
      "epoch": 1.0595371246666092,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008935628753520131,
      "loss": 1.1183,
      "step": 12315
    },
    {
      "epoch": 1.059623160973931,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008935461962521915,
      "loss": 0.8675,
      "step": 12316
    },
    {
      "epoch": 1.0597091972812527,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008935295160013239,
      "loss": 0.9121,
      "step": 12317
    },
    {
      "epoch": 1.0597952335885743,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008935128345994589,
      "loss": 1.5589,
      "step": 12318
    },
    {
      "epoch": 1.0598812698958961,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008934961520466456,
      "loss": 1.4636,
      "step": 12319
    },
    {
      "epoch": 1.0599673062032178,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008934794683429323,
      "loss": 1.7822,
      "step": 12320
    },
    {
      "epoch": 1.0600533425105394,
      "grad_norm": 0.875,
      "learning_rate": 0.0008934627834883684,
      "loss": 0.8696,
      "step": 12321
    },
    {
      "epoch": 1.0601393788178612,
      "grad_norm": 1.796875,
      "learning_rate": 0.0008934460974830022,
      "loss": 1.3612,
      "step": 12322
    },
    {
      "epoch": 1.0602254151251829,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008934294103268828,
      "loss": 1.7018,
      "step": 12323
    },
    {
      "epoch": 1.0603114514325045,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008934127220200588,
      "loss": 1.1552,
      "step": 12324
    },
    {
      "epoch": 1.0603974877398261,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008933960325625789,
      "loss": 1.0418,
      "step": 12325
    },
    {
      "epoch": 1.060483524047148,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008933793419544924,
      "loss": 1.3522,
      "step": 12326
    },
    {
      "epoch": 1.0605695603544696,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008933626501958479,
      "loss": 1.0507,
      "step": 12327
    },
    {
      "epoch": 1.0606555966617912,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000893345957286694,
      "loss": 1.5468,
      "step": 12328
    },
    {
      "epoch": 1.060741632969113,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008933292632270796,
      "loss": 1.2071,
      "step": 12329
    },
    {
      "epoch": 1.0608276692764347,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008933125680170537,
      "loss": 1.8298,
      "step": 12330
    },
    {
      "epoch": 1.0609137055837563,
      "grad_norm": 1.265625,
      "learning_rate": 0.000893295871656665,
      "loss": 1.5425,
      "step": 12331
    },
    {
      "epoch": 1.060999741891078,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008932791741459625,
      "loss": 0.9748,
      "step": 12332
    },
    {
      "epoch": 1.0610857781983998,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008932624754849947,
      "loss": 0.8958,
      "step": 12333
    },
    {
      "epoch": 1.0611718145057214,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008932457756738109,
      "loss": 1.1694,
      "step": 12334
    },
    {
      "epoch": 1.061257850813043,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008932290747124594,
      "loss": 1.3975,
      "step": 12335
    },
    {
      "epoch": 1.0613438871203649,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008932123726009896,
      "loss": 0.645,
      "step": 12336
    },
    {
      "epoch": 1.0614299234276865,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008931956693394498,
      "loss": 1.0679,
      "step": 12337
    },
    {
      "epoch": 1.0615159597350081,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008931789649278893,
      "loss": 0.8644,
      "step": 12338
    },
    {
      "epoch": 1.06160199604233,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008931622593663568,
      "loss": 1.1691,
      "step": 12339
    },
    {
      "epoch": 1.0616880323496516,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008931455526549011,
      "loss": 1.5212,
      "step": 12340
    },
    {
      "epoch": 1.0617740686569732,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008931288447935712,
      "loss": 1.5851,
      "step": 12341
    },
    {
      "epoch": 1.0618601049642948,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008931121357824159,
      "loss": 1.2066,
      "step": 12342
    },
    {
      "epoch": 1.0619461412716167,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000893095425621484,
      "loss": 1.3926,
      "step": 12343
    },
    {
      "epoch": 1.0620321775789383,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008930787143108244,
      "loss": 1.3789,
      "step": 12344
    },
    {
      "epoch": 1.06211821388626,
      "grad_norm": 1.140625,
      "learning_rate": 0.000893062001850486,
      "loss": 1.0193,
      "step": 12345
    },
    {
      "epoch": 1.0622042501935818,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008930452882405178,
      "loss": 1.0633,
      "step": 12346
    },
    {
      "epoch": 1.0622902865009034,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008930285734809686,
      "loss": 1.6252,
      "step": 12347
    },
    {
      "epoch": 1.062376322808225,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008930118575718871,
      "loss": 0.9637,
      "step": 12348
    },
    {
      "epoch": 1.0624623591155467,
      "grad_norm": 1.5,
      "learning_rate": 0.0008929951405133225,
      "loss": 1.214,
      "step": 12349
    },
    {
      "epoch": 1.0625483954228685,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008929784223053234,
      "loss": 1.2436,
      "step": 12350
    },
    {
      "epoch": 1.0626344317301901,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000892961702947939,
      "loss": 1.023,
      "step": 12351
    },
    {
      "epoch": 1.0627204680375117,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008929449824412179,
      "loss": 1.3795,
      "step": 12352
    },
    {
      "epoch": 1.0628065043448336,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008929282607852093,
      "loss": 1.1188,
      "step": 12353
    },
    {
      "epoch": 1.0628925406521552,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008929115379799618,
      "loss": 1.2142,
      "step": 12354
    },
    {
      "epoch": 1.0629785769594768,
      "grad_norm": 1.125,
      "learning_rate": 0.0008928948140255246,
      "loss": 1.4978,
      "step": 12355
    },
    {
      "epoch": 1.0630646132667987,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008928780889219465,
      "loss": 0.9216,
      "step": 12356
    },
    {
      "epoch": 1.0631506495741203,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008928613626692763,
      "loss": 0.7035,
      "step": 12357
    },
    {
      "epoch": 1.063236685881442,
      "grad_norm": 1.625,
      "learning_rate": 0.000892844635267563,
      "loss": 2.0957,
      "step": 12358
    },
    {
      "epoch": 1.0633227221887636,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008928279067168557,
      "loss": 0.6482,
      "step": 12359
    },
    {
      "epoch": 1.0634087584960854,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008928111770172031,
      "loss": 1.3131,
      "step": 12360
    },
    {
      "epoch": 1.063494794803407,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008927944461686541,
      "loss": 1.6096,
      "step": 12361
    },
    {
      "epoch": 1.0635808311107287,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008927777141712578,
      "loss": 2.183,
      "step": 12362
    },
    {
      "epoch": 1.0636668674180505,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0008927609810250631,
      "loss": 1.3537,
      "step": 12363
    },
    {
      "epoch": 1.0637529037253721,
      "grad_norm": 1.359375,
      "learning_rate": 0.000892744246730119,
      "loss": 1.2846,
      "step": 12364
    },
    {
      "epoch": 1.0638389400326937,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008927275112864743,
      "loss": 1.0458,
      "step": 12365
    },
    {
      "epoch": 1.0639249763400156,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008927107746941778,
      "loss": 1.4578,
      "step": 12366
    },
    {
      "epoch": 1.0640110126473372,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000892694036953279,
      "loss": 1.9412,
      "step": 12367
    },
    {
      "epoch": 1.0640970489546588,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008926772980638264,
      "loss": 1.5278,
      "step": 12368
    },
    {
      "epoch": 1.0641830852619805,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000892660558025869,
      "loss": 1.2951,
      "step": 12369
    },
    {
      "epoch": 1.0642691215693023,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008926438168394558,
      "loss": 1.4666,
      "step": 12370
    },
    {
      "epoch": 1.064355157876624,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008926270745046359,
      "loss": 0.6644,
      "step": 12371
    },
    {
      "epoch": 1.0644411941839456,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008926103310214581,
      "loss": 1.7457,
      "step": 12372
    },
    {
      "epoch": 1.0645272304912674,
      "grad_norm": 1.71875,
      "learning_rate": 0.0008925935863899715,
      "loss": 1.6743,
      "step": 12373
    },
    {
      "epoch": 1.064613266798589,
      "grad_norm": 1.703125,
      "learning_rate": 0.000892576840610225,
      "loss": 1.2609,
      "step": 12374
    },
    {
      "epoch": 1.0646993031059107,
      "grad_norm": 1.953125,
      "learning_rate": 0.0008925600936822676,
      "loss": 2.0427,
      "step": 12375
    },
    {
      "epoch": 1.0647853394132323,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008925433456061483,
      "loss": 1.829,
      "step": 12376
    },
    {
      "epoch": 1.0648713757205541,
      "grad_norm": 1.234375,
      "learning_rate": 0.000892526596381916,
      "loss": 1.3318,
      "step": 12377
    },
    {
      "epoch": 1.0649574120278757,
      "grad_norm": 1.890625,
      "learning_rate": 0.0008925098460096198,
      "loss": 1.5589,
      "step": 12378
    },
    {
      "epoch": 1.0650434483351974,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008924930944893086,
      "loss": 1.4998,
      "step": 12379
    },
    {
      "epoch": 1.0651294846425192,
      "grad_norm": 1.125,
      "learning_rate": 0.0008924763418210316,
      "loss": 0.9994,
      "step": 12380
    },
    {
      "epoch": 1.0652155209498408,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008924595880048374,
      "loss": 1.2336,
      "step": 12381
    },
    {
      "epoch": 1.0653015572571625,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008924428330407753,
      "loss": 1.1182,
      "step": 12382
    },
    {
      "epoch": 1.0653875935644843,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008924260769288943,
      "loss": 1.1569,
      "step": 12383
    },
    {
      "epoch": 1.065473629871806,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008924093196692434,
      "loss": 1.2709,
      "step": 12384
    },
    {
      "epoch": 1.0655596661791276,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008923925612618713,
      "loss": 1.1309,
      "step": 12385
    },
    {
      "epoch": 1.0656457024864492,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008923758017068275,
      "loss": 0.8022,
      "step": 12386
    },
    {
      "epoch": 1.065731738793771,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008923590410041607,
      "loss": 1.3547,
      "step": 12387
    },
    {
      "epoch": 1.0658177751010927,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008923422791539201,
      "loss": 1.1966,
      "step": 12388
    },
    {
      "epoch": 1.0659038114084143,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008923255161561545,
      "loss": 1.4023,
      "step": 12389
    },
    {
      "epoch": 1.0659898477157361,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008923087520109133,
      "loss": 1.6407,
      "step": 12390
    },
    {
      "epoch": 1.0660758840230578,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008922919867182451,
      "loss": 1.194,
      "step": 12391
    },
    {
      "epoch": 1.0661619203303794,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008922752202781992,
      "loss": 1.3668,
      "step": 12392
    },
    {
      "epoch": 1.0662479566377012,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008922584526908245,
      "loss": 0.9165,
      "step": 12393
    },
    {
      "epoch": 1.0663339929450228,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008922416839561702,
      "loss": 1.2627,
      "step": 12394
    },
    {
      "epoch": 1.0664200292523445,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008922249140742853,
      "loss": 1.0945,
      "step": 12395
    },
    {
      "epoch": 1.066506065559666,
      "grad_norm": 5.0625,
      "learning_rate": 0.0008922081430452187,
      "loss": 0.9689,
      "step": 12396
    },
    {
      "epoch": 1.066592101866988,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008921913708690196,
      "loss": 0.6701,
      "step": 12397
    },
    {
      "epoch": 1.0666781381743096,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000892174597545737,
      "loss": 1.0889,
      "step": 12398
    },
    {
      "epoch": 1.0667641744816312,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008921578230754201,
      "loss": 1.2082,
      "step": 12399
    },
    {
      "epoch": 1.066850210788953,
      "grad_norm": 1.25,
      "learning_rate": 0.0008921410474581178,
      "loss": 1.3223,
      "step": 12400
    },
    {
      "epoch": 1.0669362470962747,
      "grad_norm": 1.046875,
      "learning_rate": 0.000892124270693879,
      "loss": 1.4906,
      "step": 12401
    },
    {
      "epoch": 1.0670222834035963,
      "grad_norm": 1.640625,
      "learning_rate": 0.000892107492782753,
      "loss": 1.6954,
      "step": 12402
    },
    {
      "epoch": 1.067108319710918,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008920907137247891,
      "loss": 1.2101,
      "step": 12403
    },
    {
      "epoch": 1.0671943560182398,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000892073933520036,
      "loss": 1.1675,
      "step": 12404
    },
    {
      "epoch": 1.0672803923255614,
      "grad_norm": 2.828125,
      "learning_rate": 0.0008920571521685429,
      "loss": 1.3269,
      "step": 12405
    },
    {
      "epoch": 1.067366428632883,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008920403696703589,
      "loss": 0.4023,
      "step": 12406
    },
    {
      "epoch": 1.0674524649402048,
      "grad_norm": 1.125,
      "learning_rate": 0.000892023586025533,
      "loss": 1.3009,
      "step": 12407
    },
    {
      "epoch": 1.0675385012475265,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008920068012341144,
      "loss": 1.4269,
      "step": 12408
    },
    {
      "epoch": 1.067624537554848,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008919900152961521,
      "loss": 0.737,
      "step": 12409
    },
    {
      "epoch": 1.0677105738621697,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008919732282116952,
      "loss": 1.8841,
      "step": 12410
    },
    {
      "epoch": 1.0677966101694916,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008919564399807929,
      "loss": 1.5654,
      "step": 12411
    },
    {
      "epoch": 1.0678826464768132,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008919396506034943,
      "loss": 0.8045,
      "step": 12412
    },
    {
      "epoch": 1.0679686827841348,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008919228600798483,
      "loss": 1.1832,
      "step": 12413
    },
    {
      "epoch": 1.0680547190914567,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008919060684099044,
      "loss": 1.1936,
      "step": 12414
    },
    {
      "epoch": 1.0681407553987783,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008918892755937114,
      "loss": 1.8729,
      "step": 12415
    },
    {
      "epoch": 1.0682267917061,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008918724816313184,
      "loss": 1.0857,
      "step": 12416
    },
    {
      "epoch": 1.0683128280134218,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008918556865227746,
      "loss": 1.0545,
      "step": 12417
    },
    {
      "epoch": 1.0683988643207434,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008918388902681291,
      "loss": 1.4952,
      "step": 12418
    },
    {
      "epoch": 1.068484900628065,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008918220928674312,
      "loss": 1.0531,
      "step": 12419
    },
    {
      "epoch": 1.0685709369353869,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008918052943207298,
      "loss": 1.5473,
      "step": 12420
    },
    {
      "epoch": 1.0686569732427085,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008917884946280741,
      "loss": 0.9835,
      "step": 12421
    },
    {
      "epoch": 1.06874300955003,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008917716937895133,
      "loss": 1.764,
      "step": 12422
    },
    {
      "epoch": 1.0688290458573517,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008917548918050964,
      "loss": 1.3618,
      "step": 12423
    },
    {
      "epoch": 1.0689150821646736,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008917380886748727,
      "loss": 1.0222,
      "step": 12424
    },
    {
      "epoch": 1.0690011184719952,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008917212843988913,
      "loss": 0.9991,
      "step": 12425
    },
    {
      "epoch": 1.0690871547793168,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008917044789772013,
      "loss": 0.6225,
      "step": 12426
    },
    {
      "epoch": 1.0691731910866387,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008916876724098519,
      "loss": 1.3087,
      "step": 12427
    },
    {
      "epoch": 1.0692592273939603,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008916708646968923,
      "loss": 1.0065,
      "step": 12428
    },
    {
      "epoch": 1.069345263701282,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008916540558383714,
      "loss": 1.4432,
      "step": 12429
    },
    {
      "epoch": 1.0694313000086035,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008916372458343387,
      "loss": 1.0138,
      "step": 12430
    },
    {
      "epoch": 1.0695173363159254,
      "grad_norm": 1.0,
      "learning_rate": 0.0008916204346848432,
      "loss": 1.5481,
      "step": 12431
    },
    {
      "epoch": 1.069603372623247,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008916036223899342,
      "loss": 1.7833,
      "step": 12432
    },
    {
      "epoch": 1.0696894089305686,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008915868089496606,
      "loss": 0.8785,
      "step": 12433
    },
    {
      "epoch": 1.0697754452378905,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008915699943640718,
      "loss": 1.0252,
      "step": 12434
    },
    {
      "epoch": 1.069861481545212,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000891553178633217,
      "loss": 1.8197,
      "step": 12435
    },
    {
      "epoch": 1.0699475178525337,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008915363617571452,
      "loss": 1.5379,
      "step": 12436
    },
    {
      "epoch": 1.0700335541598553,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0008915195437359057,
      "loss": 2.1696,
      "step": 12437
    },
    {
      "epoch": 1.0701195904671772,
      "grad_norm": 1.375,
      "learning_rate": 0.0008915027245695478,
      "loss": 1.7659,
      "step": 12438
    },
    {
      "epoch": 1.0702056267744988,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008914859042581203,
      "loss": 1.0955,
      "step": 12439
    },
    {
      "epoch": 1.0702916630818204,
      "grad_norm": 1.359375,
      "learning_rate": 0.000891469082801673,
      "loss": 1.2122,
      "step": 12440
    },
    {
      "epoch": 1.0703776993891423,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008914522602002546,
      "loss": 0.6345,
      "step": 12441
    },
    {
      "epoch": 1.070463735696464,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008914354364539145,
      "loss": 1.2528,
      "step": 12442
    },
    {
      "epoch": 1.0705497720037855,
      "grad_norm": 2.125,
      "learning_rate": 0.0008914186115627019,
      "loss": 1.4626,
      "step": 12443
    },
    {
      "epoch": 1.0706358083111074,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008914017855266658,
      "loss": 2.0414,
      "step": 12444
    },
    {
      "epoch": 1.070721844618429,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008913849583458559,
      "loss": 1.0625,
      "step": 12445
    },
    {
      "epoch": 1.0708078809257506,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008913681300203209,
      "loss": 0.9464,
      "step": 12446
    },
    {
      "epoch": 1.0708939172330723,
      "grad_norm": 8.75,
      "learning_rate": 0.0008913513005501104,
      "loss": 1.1884,
      "step": 12447
    },
    {
      "epoch": 1.070979953540394,
      "grad_norm": 1.25,
      "learning_rate": 0.0008913344699352732,
      "loss": 2.0051,
      "step": 12448
    },
    {
      "epoch": 1.0710659898477157,
      "grad_norm": 1.625,
      "learning_rate": 0.000891317638175859,
      "loss": 1.4061,
      "step": 12449
    },
    {
      "epoch": 1.0711520261550374,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008913008052719169,
      "loss": 1.8871,
      "step": 12450
    },
    {
      "epoch": 1.0712380624623592,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008912839712234959,
      "loss": 1.4121,
      "step": 12451
    },
    {
      "epoch": 1.0713240987696808,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008912671360306455,
      "loss": 1.5509,
      "step": 12452
    },
    {
      "epoch": 1.0714101350770024,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008912502996934146,
      "loss": 1.9231,
      "step": 12453
    },
    {
      "epoch": 1.0714961713843243,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008912334622118527,
      "loss": 1.0686,
      "step": 12454
    },
    {
      "epoch": 1.071582207691646,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008912166235860093,
      "loss": 1.2378,
      "step": 12455
    },
    {
      "epoch": 1.0716682439989675,
      "grad_norm": 2.609375,
      "learning_rate": 0.0008911997838159331,
      "loss": 1.329,
      "step": 12456
    },
    {
      "epoch": 1.0717542803062892,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008911829429016736,
      "loss": 1.6594,
      "step": 12457
    },
    {
      "epoch": 1.071840316613611,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008911661008432802,
      "loss": 1.6179,
      "step": 12458
    },
    {
      "epoch": 1.0719263529209326,
      "grad_norm": 1.125,
      "learning_rate": 0.000891149257640802,
      "loss": 1.2114,
      "step": 12459
    },
    {
      "epoch": 1.0720123892282543,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008911324132942884,
      "loss": 1.187,
      "step": 12460
    },
    {
      "epoch": 1.072098425535576,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008911155678037883,
      "loss": 1.1782,
      "step": 12461
    },
    {
      "epoch": 1.0721844618428977,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008910987211693516,
      "loss": 1.6614,
      "step": 12462
    },
    {
      "epoch": 1.0722704981502194,
      "grad_norm": 1.25,
      "learning_rate": 0.0008910818733910269,
      "loss": 1.4779,
      "step": 12463
    },
    {
      "epoch": 1.072356534457541,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008910650244688638,
      "loss": 0.6154,
      "step": 12464
    },
    {
      "epoch": 1.0724425707648628,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008910481744029118,
      "loss": 0.741,
      "step": 12465
    },
    {
      "epoch": 1.0725286070721844,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008910313231932197,
      "loss": 2.2791,
      "step": 12466
    },
    {
      "epoch": 1.072614643379506,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0008910144708398371,
      "loss": 0.9426,
      "step": 12467
    },
    {
      "epoch": 1.072700679686828,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008909976173428132,
      "loss": 1.1552,
      "step": 12468
    },
    {
      "epoch": 1.0727867159941495,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008909807627021973,
      "loss": 1.3658,
      "step": 12469
    },
    {
      "epoch": 1.0728727523014712,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008909639069180389,
      "loss": 1.1935,
      "step": 12470
    },
    {
      "epoch": 1.072958788608793,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008909470499903869,
      "loss": 1.2538,
      "step": 12471
    },
    {
      "epoch": 1.0730448249161146,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000890930191919291,
      "loss": 0.9052,
      "step": 12472
    },
    {
      "epoch": 1.0731308612234363,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008909133327048,
      "loss": 1.5843,
      "step": 12473
    },
    {
      "epoch": 1.0732168975307579,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008908964723469638,
      "loss": 1.3724,
      "step": 12474
    },
    {
      "epoch": 1.0733029338380797,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008908796108458314,
      "loss": 1.4243,
      "step": 12475
    },
    {
      "epoch": 1.0733889701454014,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008908627482014522,
      "loss": 1.3298,
      "step": 12476
    },
    {
      "epoch": 1.073475006452723,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008908458844138754,
      "loss": 1.0756,
      "step": 12477
    },
    {
      "epoch": 1.0735610427600448,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0008908290194831503,
      "loss": 0.915,
      "step": 12478
    },
    {
      "epoch": 1.0736470790673664,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008908121534093265,
      "loss": 0.8598,
      "step": 12479
    },
    {
      "epoch": 1.073733115374688,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008907952861924531,
      "loss": 1.0556,
      "step": 12480
    },
    {
      "epoch": 1.07381915168201,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008907784178325795,
      "loss": 1.5926,
      "step": 12481
    },
    {
      "epoch": 1.0739051879893315,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008907615483297551,
      "loss": 1.2868,
      "step": 12482
    },
    {
      "epoch": 1.0739912242966532,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000890744677684029,
      "loss": 2.0238,
      "step": 12483
    },
    {
      "epoch": 1.0740772606039748,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008907278058954509,
      "loss": 1.4005,
      "step": 12484
    },
    {
      "epoch": 1.0741632969112966,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008907109329640699,
      "loss": 1.4759,
      "step": 12485
    },
    {
      "epoch": 1.0742493332186183,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008906940588899353,
      "loss": 1.3808,
      "step": 12486
    },
    {
      "epoch": 1.0743353695259399,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008906771836730967,
      "loss": 0.9566,
      "step": 12487
    },
    {
      "epoch": 1.0744214058332617,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008906603073136032,
      "loss": 1.8667,
      "step": 12488
    },
    {
      "epoch": 1.0745074421405834,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008906434298115045,
      "loss": 1.9337,
      "step": 12489
    },
    {
      "epoch": 1.074593478447905,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008906265511668494,
      "loss": 1.2934,
      "step": 12490
    },
    {
      "epoch": 1.0746795147552266,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008906096713796878,
      "loss": 1.2585,
      "step": 12491
    },
    {
      "epoch": 1.0747655510625485,
      "grad_norm": 0.890625,
      "learning_rate": 0.0008905927904500691,
      "loss": 0.8273,
      "step": 12492
    },
    {
      "epoch": 1.07485158736987,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008905759083780421,
      "loss": 1.1562,
      "step": 12493
    },
    {
      "epoch": 1.0749376236771917,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008905590251636566,
      "loss": 2.0769,
      "step": 12494
    },
    {
      "epoch": 1.0750236599845135,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000890542140806962,
      "loss": 1.7155,
      "step": 12495
    },
    {
      "epoch": 1.0751096962918352,
      "grad_norm": 1.125,
      "learning_rate": 0.0008905252553080077,
      "loss": 1.0741,
      "step": 12496
    },
    {
      "epoch": 1.0751957325991568,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008905083686668427,
      "loss": 1.2699,
      "step": 12497
    },
    {
      "epoch": 1.0752817689064784,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008904914808835169,
      "loss": 0.9602,
      "step": 12498
    },
    {
      "epoch": 1.0753678052138003,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008904745919580795,
      "loss": 0.8554,
      "step": 12499
    },
    {
      "epoch": 1.075453841521122,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008904577018905797,
      "loss": 1.8327,
      "step": 12500
    },
    {
      "epoch": 1.0755398778284435,
      "grad_norm": 1.25,
      "learning_rate": 0.000890440810681067,
      "loss": 1.3415,
      "step": 12501
    },
    {
      "epoch": 1.0756259141357654,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000890423918329591,
      "loss": 1.3613,
      "step": 12502
    },
    {
      "epoch": 1.075711950443087,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008904070248362011,
      "loss": 0.8427,
      "step": 12503
    },
    {
      "epoch": 1.0757979867504086,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008903901302009463,
      "loss": 1.3699,
      "step": 12504
    },
    {
      "epoch": 1.0758840230577305,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008903732344238764,
      "loss": 1.9268,
      "step": 12505
    },
    {
      "epoch": 1.075970059365052,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008903563375050409,
      "loss": 0.8489,
      "step": 12506
    },
    {
      "epoch": 1.0760560956723737,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008903394394444888,
      "loss": 1.4527,
      "step": 12507
    },
    {
      "epoch": 1.0761421319796955,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008903225402422699,
      "loss": 1.2026,
      "step": 12508
    },
    {
      "epoch": 1.0762281682870172,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008903056398984334,
      "loss": 1.748,
      "step": 12509
    },
    {
      "epoch": 1.0763142045943388,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008902887384130288,
      "loss": 1.5523,
      "step": 12510
    },
    {
      "epoch": 1.0764002409016604,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008902718357861055,
      "loss": 1.4695,
      "step": 12511
    },
    {
      "epoch": 1.0764862772089823,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008902549320177132,
      "loss": 1.2262,
      "step": 12512
    },
    {
      "epoch": 1.076572313516304,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008902380271079009,
      "loss": 1.4521,
      "step": 12513
    },
    {
      "epoch": 1.0766583498236255,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008902211210567184,
      "loss": 0.9996,
      "step": 12514
    },
    {
      "epoch": 1.0767443861309474,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000890204213864215,
      "loss": 1.7327,
      "step": 12515
    },
    {
      "epoch": 1.076830422438269,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008901873055304401,
      "loss": 1.3564,
      "step": 12516
    },
    {
      "epoch": 1.0769164587455906,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008901703960554432,
      "loss": 1.6725,
      "step": 12517
    },
    {
      "epoch": 1.0770024950529122,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008901534854392739,
      "loss": 1.1508,
      "step": 12518
    },
    {
      "epoch": 1.077088531360234,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008901365736819814,
      "loss": 1.5563,
      "step": 12519
    },
    {
      "epoch": 1.0771745676675557,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008901196607836153,
      "loss": 0.7812,
      "step": 12520
    },
    {
      "epoch": 1.0772606039748773,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008901027467442252,
      "loss": 0.8242,
      "step": 12521
    },
    {
      "epoch": 1.0773466402821992,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0008900858315638604,
      "loss": 1.3719,
      "step": 12522
    },
    {
      "epoch": 1.0774326765895208,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008900689152425704,
      "loss": 1.2599,
      "step": 12523
    },
    {
      "epoch": 1.0775187128968424,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008900519977804047,
      "loss": 0.9392,
      "step": 12524
    },
    {
      "epoch": 1.077604749204164,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008900350791774126,
      "loss": 1.3673,
      "step": 12525
    },
    {
      "epoch": 1.077690785511486,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008900181594336439,
      "loss": 1.2659,
      "step": 12526
    },
    {
      "epoch": 1.0777768218188075,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000890001238549148,
      "loss": 1.3976,
      "step": 12527
    },
    {
      "epoch": 1.0778628581261291,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008899843165239743,
      "loss": 1.2002,
      "step": 12528
    },
    {
      "epoch": 1.077948894433451,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008899673933581721,
      "loss": 1.0377,
      "step": 12529
    },
    {
      "epoch": 1.0780349307407726,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008899504690517914,
      "loss": 1.095,
      "step": 12530
    },
    {
      "epoch": 1.0781209670480942,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008899335436048812,
      "loss": 1.8473,
      "step": 12531
    },
    {
      "epoch": 1.078207003355416,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008899166170174914,
      "loss": 1.1765,
      "step": 12532
    },
    {
      "epoch": 1.0782930396627377,
      "grad_norm": 1.375,
      "learning_rate": 0.0008898996892896712,
      "loss": 1.428,
      "step": 12533
    },
    {
      "epoch": 1.0783790759700593,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008898827604214703,
      "loss": 1.3932,
      "step": 12534
    },
    {
      "epoch": 1.078465112277381,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008898658304129383,
      "loss": 1.8228,
      "step": 12535
    },
    {
      "epoch": 1.0785511485847028,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008898488992641244,
      "loss": 0.8208,
      "step": 12536
    },
    {
      "epoch": 1.0786371848920244,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008898319669750782,
      "loss": 1.5745,
      "step": 12537
    },
    {
      "epoch": 1.078723221199346,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008898150335458494,
      "loss": 0.6996,
      "step": 12538
    },
    {
      "epoch": 1.078809257506668,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008897980989764875,
      "loss": 0.8402,
      "step": 12539
    },
    {
      "epoch": 1.0788952938139895,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008897811632670421,
      "loss": 1.8407,
      "step": 12540
    },
    {
      "epoch": 1.0789813301213111,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008897642264175622,
      "loss": 1.5365,
      "step": 12541
    },
    {
      "epoch": 1.079067366428633,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008897472884280981,
      "loss": 0.9635,
      "step": 12542
    },
    {
      "epoch": 1.0791534027359546,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008897303492986989,
      "loss": 0.7632,
      "step": 12543
    },
    {
      "epoch": 1.0792394390432762,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008897134090294143,
      "loss": 0.918,
      "step": 12544
    },
    {
      "epoch": 1.0793254753505979,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008896964676202937,
      "loss": 0.9368,
      "step": 12545
    },
    {
      "epoch": 1.0794115116579197,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008896795250713866,
      "loss": 1.8902,
      "step": 12546
    },
    {
      "epoch": 1.0794975479652413,
      "grad_norm": 1.125,
      "learning_rate": 0.0008896625813827428,
      "loss": 0.6237,
      "step": 12547
    },
    {
      "epoch": 1.079583584272563,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008896456365544116,
      "loss": 1.0305,
      "step": 12548
    },
    {
      "epoch": 1.0796696205798848,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008896286905864428,
      "loss": 0.6856,
      "step": 12549
    },
    {
      "epoch": 1.0797556568872064,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000889611743478886,
      "loss": 1.0041,
      "step": 12550
    },
    {
      "epoch": 1.079841693194528,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008895947952317904,
      "loss": 2.0095,
      "step": 12551
    },
    {
      "epoch": 1.0799277295018497,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000889577845845206,
      "loss": 1.1543,
      "step": 12552
    },
    {
      "epoch": 1.0800137658091715,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000889560895319182,
      "loss": 1.3007,
      "step": 12553
    },
    {
      "epoch": 1.0800998021164931,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008895439436537682,
      "loss": 1.4707,
      "step": 12554
    },
    {
      "epoch": 1.0801858384238148,
      "grad_norm": 1.5,
      "learning_rate": 0.0008895269908490141,
      "loss": 1.7825,
      "step": 12555
    },
    {
      "epoch": 1.0802718747311366,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008895100369049694,
      "loss": 0.9388,
      "step": 12556
    },
    {
      "epoch": 1.0803579110384582,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008894930818216835,
      "loss": 1.3756,
      "step": 12557
    },
    {
      "epoch": 1.0804439473457799,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008894761255992061,
      "loss": 1.3805,
      "step": 12558
    },
    {
      "epoch": 1.0805299836531017,
      "grad_norm": 1.125,
      "learning_rate": 0.0008894591682375867,
      "loss": 0.9011,
      "step": 12559
    },
    {
      "epoch": 1.0806160199604233,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008894422097368753,
      "loss": 1.2321,
      "step": 12560
    },
    {
      "epoch": 1.080702056267745,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000889425250097121,
      "loss": 1.9128,
      "step": 12561
    },
    {
      "epoch": 1.0807880925750666,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008894082893183735,
      "loss": 1.2829,
      "step": 12562
    },
    {
      "epoch": 1.0808741288823884,
      "grad_norm": 1.90625,
      "learning_rate": 0.0008893913274006824,
      "loss": 2.2789,
      "step": 12563
    },
    {
      "epoch": 1.08096016518971,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008893743643440977,
      "loss": 1.5175,
      "step": 12564
    },
    {
      "epoch": 1.0810462014970317,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008893574001486683,
      "loss": 0.9622,
      "step": 12565
    },
    {
      "epoch": 1.0811322378043535,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008893404348144447,
      "loss": 1.4023,
      "step": 12566
    },
    {
      "epoch": 1.0812182741116751,
      "grad_norm": 1.25,
      "learning_rate": 0.0008893234683414758,
      "loss": 0.6877,
      "step": 12567
    },
    {
      "epoch": 1.0813043104189968,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008893065007298115,
      "loss": 0.8898,
      "step": 12568
    },
    {
      "epoch": 1.0813903467263186,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008892895319795014,
      "loss": 1.2861,
      "step": 12569
    },
    {
      "epoch": 1.0814763830336402,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008892725620905951,
      "loss": 1.7028,
      "step": 12570
    },
    {
      "epoch": 1.0815624193409619,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008892555910631424,
      "loss": 0.7099,
      "step": 12571
    },
    {
      "epoch": 1.0816484556482835,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008892386188971926,
      "loss": 1.0717,
      "step": 12572
    },
    {
      "epoch": 1.0817344919556053,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008892216455927957,
      "loss": 1.9946,
      "step": 12573
    },
    {
      "epoch": 1.081820528262927,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008892046711500011,
      "loss": 0.8769,
      "step": 12574
    },
    {
      "epoch": 1.0819065645702486,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008891876955688586,
      "loss": 1.2135,
      "step": 12575
    },
    {
      "epoch": 1.0819926008775704,
      "grad_norm": 1.5,
      "learning_rate": 0.0008891707188494177,
      "loss": 1.3788,
      "step": 12576
    },
    {
      "epoch": 1.082078637184892,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008891537409917283,
      "loss": 1.5075,
      "step": 12577
    },
    {
      "epoch": 1.0821646734922137,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008891367619958397,
      "loss": 1.7403,
      "step": 12578
    },
    {
      "epoch": 1.0822507097995353,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008891197818618018,
      "loss": 1.4532,
      "step": 12579
    },
    {
      "epoch": 1.0823367461068572,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008891028005896643,
      "loss": 1.7043,
      "step": 12580
    },
    {
      "epoch": 1.0824227824141788,
      "grad_norm": 1.375,
      "learning_rate": 0.0008890858181794767,
      "loss": 1.2131,
      "step": 12581
    },
    {
      "epoch": 1.0825088187215004,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008890688346312888,
      "loss": 1.3056,
      "step": 12582
    },
    {
      "epoch": 1.0825948550288222,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008890518499451504,
      "loss": 1.6599,
      "step": 12583
    },
    {
      "epoch": 1.0826808913361439,
      "grad_norm": 1.859375,
      "learning_rate": 0.0008890348641211106,
      "loss": 1.7977,
      "step": 12584
    },
    {
      "epoch": 1.0827669276434655,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008890178771592198,
      "loss": 1.6246,
      "step": 12585
    },
    {
      "epoch": 1.0828529639507871,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008890008890595272,
      "loss": 1.1856,
      "step": 12586
    },
    {
      "epoch": 1.082939000258109,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008889838998220827,
      "loss": 0.7865,
      "step": 12587
    },
    {
      "epoch": 1.0830250365654306,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008889669094469361,
      "loss": 1.5886,
      "step": 12588
    },
    {
      "epoch": 1.0831110728727522,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008889499179341368,
      "loss": 1.1897,
      "step": 12589
    },
    {
      "epoch": 1.083197109180074,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008889329252837346,
      "loss": 2.09,
      "step": 12590
    },
    {
      "epoch": 1.0832831454873957,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008889159314957793,
      "loss": 2.0077,
      "step": 12591
    },
    {
      "epoch": 1.0833691817947173,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008888989365703206,
      "loss": 1.0954,
      "step": 12592
    },
    {
      "epoch": 1.0834552181020392,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008888819405074081,
      "loss": 2.0865,
      "step": 12593
    },
    {
      "epoch": 1.0835412544093608,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008888649433070915,
      "loss": 0.8699,
      "step": 12594
    },
    {
      "epoch": 1.0836272907166824,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008888479449694206,
      "loss": 1.3305,
      "step": 12595
    },
    {
      "epoch": 1.0837133270240042,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008888309454944452,
      "loss": 1.6809,
      "step": 12596
    },
    {
      "epoch": 1.0837993633313259,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008888139448822148,
      "loss": 0.9815,
      "step": 12597
    },
    {
      "epoch": 1.0838853996386475,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008887969431327792,
      "loss": 1.0838,
      "step": 12598
    },
    {
      "epoch": 1.0839714359459691,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008887799402461882,
      "loss": 1.625,
      "step": 12599
    },
    {
      "epoch": 1.084057472253291,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008887629362224914,
      "loss": 1.6113,
      "step": 12600
    },
    {
      "epoch": 1.0841435085606126,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008887459310617387,
      "loss": 1.0637,
      "step": 12601
    },
    {
      "epoch": 1.0842295448679342,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008887289247639798,
      "loss": 0.8473,
      "step": 12602
    },
    {
      "epoch": 1.084315581175256,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008887119173292643,
      "loss": 1.0438,
      "step": 12603
    },
    {
      "epoch": 1.0844016174825777,
      "grad_norm": 1.125,
      "learning_rate": 0.0008886949087576421,
      "loss": 0.9354,
      "step": 12604
    },
    {
      "epoch": 1.0844876537898993,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008886778990491631,
      "loss": 1.0219,
      "step": 12605
    },
    {
      "epoch": 1.084573690097221,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008886608882038764,
      "loss": 1.134,
      "step": 12606
    },
    {
      "epoch": 1.0846597264045428,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008886438762218326,
      "loss": 2.8315,
      "step": 12607
    },
    {
      "epoch": 1.0847457627118644,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008886268631030809,
      "loss": 0.6306,
      "step": 12608
    },
    {
      "epoch": 1.084831799019186,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008886098488476711,
      "loss": 1.2661,
      "step": 12609
    },
    {
      "epoch": 1.0849178353265079,
      "grad_norm": 1.5,
      "learning_rate": 0.0008885928334556532,
      "loss": 1.3716,
      "step": 12610
    },
    {
      "epoch": 1.0850038716338295,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008885758169270768,
      "loss": 1.162,
      "step": 12611
    },
    {
      "epoch": 1.0850899079411511,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008885587992619917,
      "loss": 1.5928,
      "step": 12612
    },
    {
      "epoch": 1.0851759442484727,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008885417804604476,
      "loss": 0.7548,
      "step": 12613
    },
    {
      "epoch": 1.0852619805557946,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008885247605224945,
      "loss": 0.9677,
      "step": 12614
    },
    {
      "epoch": 1.0853480168631162,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000888507739448182,
      "loss": 1.0033,
      "step": 12615
    },
    {
      "epoch": 1.0854340531704378,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00088849071723756,
      "loss": 0.9855,
      "step": 12616
    },
    {
      "epoch": 1.0855200894777597,
      "grad_norm": 1.125,
      "learning_rate": 0.000888473693890678,
      "loss": 1.6798,
      "step": 12617
    },
    {
      "epoch": 1.0856061257850813,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008884566694075861,
      "loss": 1.5607,
      "step": 12618
    },
    {
      "epoch": 1.085692162092403,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000888439643788334,
      "loss": 1.8491,
      "step": 12619
    },
    {
      "epoch": 1.0857781983997248,
      "grad_norm": 1.375,
      "learning_rate": 0.0008884226170329716,
      "loss": 1.1039,
      "step": 12620
    },
    {
      "epoch": 1.0858642347070464,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008884055891415485,
      "loss": 0.773,
      "step": 12621
    },
    {
      "epoch": 1.085950271014368,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008883885601141146,
      "loss": 1.2636,
      "step": 12622
    },
    {
      "epoch": 1.0860363073216897,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008883715299507196,
      "loss": 0.9941,
      "step": 12623
    },
    {
      "epoch": 1.0861223436290115,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0008883544986514136,
      "loss": 1.9371,
      "step": 12624
    },
    {
      "epoch": 1.0862083799363331,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000888337466216246,
      "loss": 1.4904,
      "step": 12625
    },
    {
      "epoch": 1.0862944162436547,
      "grad_norm": 1.125,
      "learning_rate": 0.000888320432645267,
      "loss": 1.4211,
      "step": 12626
    },
    {
      "epoch": 1.0863804525509766,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008883033979385262,
      "loss": 1.0703,
      "step": 12627
    },
    {
      "epoch": 1.0864664888582982,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008882863620960735,
      "loss": 1.4443,
      "step": 12628
    },
    {
      "epoch": 1.0865525251656198,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008882693251179587,
      "loss": 1.4059,
      "step": 12629
    },
    {
      "epoch": 1.0866385614729417,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008882522870042315,
      "loss": 2.1386,
      "step": 12630
    },
    {
      "epoch": 1.0867245977802633,
      "grad_norm": 1.375,
      "learning_rate": 0.0008882352477549421,
      "loss": 0.8931,
      "step": 12631
    },
    {
      "epoch": 1.086810634087585,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00088821820737014,
      "loss": 0.8129,
      "step": 12632
    },
    {
      "epoch": 1.0868966703949066,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008882011658498752,
      "loss": 1.0062,
      "step": 12633
    },
    {
      "epoch": 1.0869827067022284,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008881841231941974,
      "loss": 2.124,
      "step": 12634
    },
    {
      "epoch": 1.08706874300955,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008881670794031566,
      "loss": 1.6417,
      "step": 12635
    },
    {
      "epoch": 1.0871547793168717,
      "grad_norm": 1.375,
      "learning_rate": 0.0008881500344768027,
      "loss": 0.9267,
      "step": 12636
    },
    {
      "epoch": 1.0872408156241935,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008881329884151853,
      "loss": 0.8514,
      "step": 12637
    },
    {
      "epoch": 1.0873268519315151,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008881159412183544,
      "loss": 1.5446,
      "step": 12638
    },
    {
      "epoch": 1.0874128882388368,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008880988928863598,
      "loss": 1.7615,
      "step": 12639
    },
    {
      "epoch": 1.0874989245461584,
      "grad_norm": 1.859375,
      "learning_rate": 0.0008880818434192514,
      "loss": 1.0827,
      "step": 12640
    },
    {
      "epoch": 1.0875849608534802,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008880647928170793,
      "loss": 1.4839,
      "step": 12641
    },
    {
      "epoch": 1.0876709971608018,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008880477410798931,
      "loss": 0.9396,
      "step": 12642
    },
    {
      "epoch": 1.0877570334681235,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008880306882077426,
      "loss": 0.8226,
      "step": 12643
    },
    {
      "epoch": 1.0878430697754453,
      "grad_norm": 1.3125,
      "learning_rate": 0.000888013634200678,
      "loss": 1.1251,
      "step": 12644
    },
    {
      "epoch": 1.087929106082767,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008879965790587487,
      "loss": 1.8479,
      "step": 12645
    },
    {
      "epoch": 1.0880151423900886,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008879795227820051,
      "loss": 1.8346,
      "step": 12646
    },
    {
      "epoch": 1.0881011786974104,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008879624653704968,
      "loss": 1.8027,
      "step": 12647
    },
    {
      "epoch": 1.088187215004732,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008879454068242736,
      "loss": 0.814,
      "step": 12648
    },
    {
      "epoch": 1.0882732513120537,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008879283471433856,
      "loss": 0.7382,
      "step": 12649
    },
    {
      "epoch": 1.0883592876193753,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008879112863278828,
      "loss": 1.3637,
      "step": 12650
    },
    {
      "epoch": 1.0884453239266971,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008878942243778148,
      "loss": 0.8944,
      "step": 12651
    },
    {
      "epoch": 1.0885313602340188,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008878771612932317,
      "loss": 0.6476,
      "step": 12652
    },
    {
      "epoch": 1.0886173965413404,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008878600970741833,
      "loss": 1.3756,
      "step": 12653
    },
    {
      "epoch": 1.0887034328486622,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008878430317207194,
      "loss": 1.5023,
      "step": 12654
    },
    {
      "epoch": 1.0887894691559838,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008878259652328904,
      "loss": 0.7861,
      "step": 12655
    },
    {
      "epoch": 1.0888755054633055,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008878088976107455,
      "loss": 1.0739,
      "step": 12656
    },
    {
      "epoch": 1.0889615417706273,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008877918288543353,
      "loss": 1.4495,
      "step": 12657
    },
    {
      "epoch": 1.089047578077949,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008877747589637093,
      "loss": 0.958,
      "step": 12658
    },
    {
      "epoch": 1.0891336143852706,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008877576879389175,
      "loss": 1.4804,
      "step": 12659
    },
    {
      "epoch": 1.0892196506925922,
      "grad_norm": 1.078125,
      "learning_rate": 0.00088774061578001,
      "loss": 1.3526,
      "step": 12660
    },
    {
      "epoch": 1.089305686999914,
      "grad_norm": 1.9375,
      "learning_rate": 0.0008877235424870365,
      "loss": 1.5796,
      "step": 12661
    },
    {
      "epoch": 1.0893917233072357,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000887706468060047,
      "loss": 0.863,
      "step": 12662
    },
    {
      "epoch": 1.0894777596145573,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008876893924990916,
      "loss": 0.7555,
      "step": 12663
    },
    {
      "epoch": 1.0895637959218791,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008876723158042202,
      "loss": 1.1448,
      "step": 12664
    },
    {
      "epoch": 1.0896498322292008,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008876552379754825,
      "loss": 0.8466,
      "step": 12665
    },
    {
      "epoch": 1.0897358685365224,
      "grad_norm": 1.125,
      "learning_rate": 0.0008876381590129287,
      "loss": 1.5998,
      "step": 12666
    },
    {
      "epoch": 1.089821904843844,
      "grad_norm": 3.0,
      "learning_rate": 0.0008876210789166088,
      "loss": 2.7429,
      "step": 12667
    },
    {
      "epoch": 1.0899079411511658,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008876039976865724,
      "loss": 1.128,
      "step": 12668
    },
    {
      "epoch": 1.0899939774584875,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008875869153228699,
      "loss": 1.4726,
      "step": 12669
    },
    {
      "epoch": 1.090080013765809,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000887569831825551,
      "loss": 1.2156,
      "step": 12670
    },
    {
      "epoch": 1.090166050073131,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008875527471946656,
      "loss": 1.3877,
      "step": 12671
    },
    {
      "epoch": 1.0902520863804526,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008875356614302639,
      "loss": 1.9391,
      "step": 12672
    },
    {
      "epoch": 1.0903381226877742,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008875185745323957,
      "loss": 1.1326,
      "step": 12673
    },
    {
      "epoch": 1.0904241589950958,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008875014865011111,
      "loss": 1.5511,
      "step": 12674
    },
    {
      "epoch": 1.0905101953024177,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008874843973364599,
      "loss": 2.0622,
      "step": 12675
    },
    {
      "epoch": 1.0905962316097393,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008874673070384924,
      "loss": 1.026,
      "step": 12676
    },
    {
      "epoch": 1.090682267917061,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008874502156072581,
      "loss": 1.3177,
      "step": 12677
    },
    {
      "epoch": 1.0907683042243828,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008874331230428076,
      "loss": 1.3215,
      "step": 12678
    },
    {
      "epoch": 1.0908543405317044,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008874160293451903,
      "loss": 1.7402,
      "step": 12679
    },
    {
      "epoch": 1.090940376839026,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008873989345144567,
      "loss": 1.4959,
      "step": 12680
    },
    {
      "epoch": 1.0910264131463479,
      "grad_norm": 1.25,
      "learning_rate": 0.0008873818385506563,
      "loss": 1.0356,
      "step": 12681
    },
    {
      "epoch": 1.0911124494536695,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008873647414538395,
      "loss": 0.6545,
      "step": 12682
    },
    {
      "epoch": 1.091198485760991,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008873476432240562,
      "loss": 1.0023,
      "step": 12683
    },
    {
      "epoch": 1.091284522068313,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008873305438613563,
      "loss": 1.8906,
      "step": 12684
    },
    {
      "epoch": 1.0913705583756346,
      "grad_norm": 1.546875,
      "learning_rate": 0.00088731344336579,
      "loss": 1.1802,
      "step": 12685
    },
    {
      "epoch": 1.0914565946829562,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008872963417374071,
      "loss": 2.2996,
      "step": 12686
    },
    {
      "epoch": 1.0915426309902778,
      "grad_norm": 0.875,
      "learning_rate": 0.0008872792389762579,
      "loss": 0.6329,
      "step": 12687
    },
    {
      "epoch": 1.0916286672975997,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000887262135082392,
      "loss": 1.6989,
      "step": 12688
    },
    {
      "epoch": 1.0917147036049213,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008872450300558598,
      "loss": 1.7154,
      "step": 12689
    },
    {
      "epoch": 1.091800739912243,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008872279238967112,
      "loss": 1.7019,
      "step": 12690
    },
    {
      "epoch": 1.0918867762195648,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008872108166049961,
      "loss": 1.076,
      "step": 12691
    },
    {
      "epoch": 1.0919728125268864,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008871937081807649,
      "loss": 1.681,
      "step": 12692
    },
    {
      "epoch": 1.092058848834208,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008871765986240672,
      "loss": 0.9673,
      "step": 12693
    },
    {
      "epoch": 1.0921448851415296,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008871594879349534,
      "loss": 1.5559,
      "step": 12694
    },
    {
      "epoch": 1.0922309214488515,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008871423761134733,
      "loss": 1.2739,
      "step": 12695
    },
    {
      "epoch": 1.092316957756173,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008871252631596771,
      "loss": 1.5119,
      "step": 12696
    },
    {
      "epoch": 1.0924029940634947,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008871081490736148,
      "loss": 1.6789,
      "step": 12697
    },
    {
      "epoch": 1.0924890303708166,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008870910338553365,
      "loss": 1.2844,
      "step": 12698
    },
    {
      "epoch": 1.0925750666781382,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008870739175048921,
      "loss": 1.4285,
      "step": 12699
    },
    {
      "epoch": 1.0926611029854598,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008870568000223319,
      "loss": 1.676,
      "step": 12700
    },
    {
      "epoch": 1.0927471392927814,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008870396814077057,
      "loss": 1.4929,
      "step": 12701
    },
    {
      "epoch": 1.0928331756001033,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008870225616610639,
      "loss": 1.3744,
      "step": 12702
    },
    {
      "epoch": 1.092919211907425,
      "grad_norm": 2.390625,
      "learning_rate": 0.0008870054407824562,
      "loss": 1.6308,
      "step": 12703
    },
    {
      "epoch": 1.0930052482147465,
      "grad_norm": 1.109375,
      "learning_rate": 0.000886988318771933,
      "loss": 1.6443,
      "step": 12704
    },
    {
      "epoch": 1.0930912845220684,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000886971195629544,
      "loss": 1.0347,
      "step": 12705
    },
    {
      "epoch": 1.09317732082939,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008869540713553396,
      "loss": 1.5698,
      "step": 12706
    },
    {
      "epoch": 1.0932633571367116,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008869369459493699,
      "loss": 1.1501,
      "step": 12707
    },
    {
      "epoch": 1.0933493934440335,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008869198194116849,
      "loss": 1.0327,
      "step": 12708
    },
    {
      "epoch": 1.093435429751355,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008869026917423346,
      "loss": 1.4584,
      "step": 12709
    },
    {
      "epoch": 1.0935214660586767,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008868855629413692,
      "loss": 1.7805,
      "step": 12710
    },
    {
      "epoch": 1.0936075023659984,
      "grad_norm": 1.75,
      "learning_rate": 0.0008868684330088388,
      "loss": 1.4805,
      "step": 12711
    },
    {
      "epoch": 1.0936935386733202,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008868513019447934,
      "loss": 1.134,
      "step": 12712
    },
    {
      "epoch": 1.0937795749806418,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008868341697492832,
      "loss": 1.0786,
      "step": 12713
    },
    {
      "epoch": 1.0938656112879634,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008868170364223582,
      "loss": 1.7462,
      "step": 12714
    },
    {
      "epoch": 1.0939516475952853,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008867999019640689,
      "loss": 1.2845,
      "step": 12715
    },
    {
      "epoch": 1.094037683902607,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008867827663744649,
      "loss": 1.0098,
      "step": 12716
    },
    {
      "epoch": 1.0941237202099285,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008867656296535965,
      "loss": 1.1948,
      "step": 12717
    },
    {
      "epoch": 1.0942097565172504,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008867484918015139,
      "loss": 0.8697,
      "step": 12718
    },
    {
      "epoch": 1.094295792824572,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008867313528182672,
      "loss": 1.3134,
      "step": 12719
    },
    {
      "epoch": 1.0943818291318936,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0008867142127039065,
      "loss": 1.3786,
      "step": 12720
    },
    {
      "epoch": 1.0944678654392153,
      "grad_norm": 1.078125,
      "learning_rate": 0.000886697071458482,
      "loss": 1.6215,
      "step": 12721
    },
    {
      "epoch": 1.094553901746537,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008866799290820437,
      "loss": 1.3568,
      "step": 12722
    },
    {
      "epoch": 1.0946399380538587,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008866627855746419,
      "loss": 0.8623,
      "step": 12723
    },
    {
      "epoch": 1.0947259743611804,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008866456409363265,
      "loss": 1.5778,
      "step": 12724
    },
    {
      "epoch": 1.0948120106685022,
      "grad_norm": 2.796875,
      "learning_rate": 0.0008866284951671479,
      "loss": 0.9273,
      "step": 12725
    },
    {
      "epoch": 1.0948980469758238,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000886611348267156,
      "loss": 1.8883,
      "step": 12726
    },
    {
      "epoch": 1.0949840832831454,
      "grad_norm": 3.484375,
      "learning_rate": 0.0008865942002364013,
      "loss": 0.8175,
      "step": 12727
    },
    {
      "epoch": 1.095070119590467,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008865770510749338,
      "loss": 0.727,
      "step": 12728
    },
    {
      "epoch": 1.095156155897789,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008865599007828034,
      "loss": 1.1984,
      "step": 12729
    },
    {
      "epoch": 1.0952421922051105,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0008865427493600606,
      "loss": 1.0621,
      "step": 12730
    },
    {
      "epoch": 1.0953282285124322,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008865255968067554,
      "loss": 0.9681,
      "step": 12731
    },
    {
      "epoch": 1.095414264819754,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000886508443122938,
      "loss": 1.1664,
      "step": 12732
    },
    {
      "epoch": 1.0955003011270756,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008864912883086585,
      "loss": 1.0665,
      "step": 12733
    },
    {
      "epoch": 1.0955863374343973,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008864741323639674,
      "loss": 1.2804,
      "step": 12734
    },
    {
      "epoch": 1.095672373741719,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008864569752889143,
      "loss": 1.5556,
      "step": 12735
    },
    {
      "epoch": 1.0957584100490407,
      "grad_norm": 720.0,
      "learning_rate": 0.00088643981708355,
      "loss": 1.1628,
      "step": 12736
    },
    {
      "epoch": 1.0958444463563624,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008864226577479242,
      "loss": 0.6759,
      "step": 12737
    },
    {
      "epoch": 1.095930482663684,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008864054972820873,
      "loss": 1.8571,
      "step": 12738
    },
    {
      "epoch": 1.0960165189710058,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008863883356860896,
      "loss": 0.8618,
      "step": 12739
    },
    {
      "epoch": 1.0961025552783275,
      "grad_norm": 13.5,
      "learning_rate": 0.000886371172959981,
      "loss": 1.2893,
      "step": 12740
    },
    {
      "epoch": 1.096188591585649,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000886354009103812,
      "loss": 1.2951,
      "step": 12741
    },
    {
      "epoch": 1.096274627892971,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008863368441176325,
      "loss": 1.0554,
      "step": 12742
    },
    {
      "epoch": 1.0963606642002925,
      "grad_norm": 1.25,
      "learning_rate": 0.000886319678001493,
      "loss": 1.8852,
      "step": 12743
    },
    {
      "epoch": 1.0964467005076142,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008863025107554436,
      "loss": 0.8939,
      "step": 12744
    },
    {
      "epoch": 1.096532736814936,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008862853423795346,
      "loss": 1.7054,
      "step": 12745
    },
    {
      "epoch": 1.0966187731222576,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008862681728738158,
      "loss": 2.0304,
      "step": 12746
    },
    {
      "epoch": 1.0967048094295793,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008862510022383379,
      "loss": 1.3829,
      "step": 12747
    },
    {
      "epoch": 1.0967908457369009,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008862338304731509,
      "loss": 0.8752,
      "step": 12748
    },
    {
      "epoch": 1.0968768820442227,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008862166575783051,
      "loss": 0.6427,
      "step": 12749
    },
    {
      "epoch": 1.0969629183515444,
      "grad_norm": 1.0,
      "learning_rate": 0.0008861994835538506,
      "loss": 1.1428,
      "step": 12750
    },
    {
      "epoch": 1.097048954658866,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008861823083998377,
      "loss": 2.0356,
      "step": 12751
    },
    {
      "epoch": 1.0971349909661878,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008861651321163169,
      "loss": 1.5477,
      "step": 12752
    },
    {
      "epoch": 1.0972210272735095,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000886147954703338,
      "loss": 0.7224,
      "step": 12753
    },
    {
      "epoch": 1.097307063580831,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008861307761609514,
      "loss": 1.6044,
      "step": 12754
    },
    {
      "epoch": 1.0973930998881527,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008861135964892075,
      "loss": 1.7167,
      "step": 12755
    },
    {
      "epoch": 1.0974791361954745,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008860964156881564,
      "loss": 1.293,
      "step": 12756
    },
    {
      "epoch": 1.0975651725027962,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008860792337578483,
      "loss": 1.7976,
      "step": 12757
    },
    {
      "epoch": 1.0976512088101178,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008860620506983335,
      "loss": 0.9591,
      "step": 12758
    },
    {
      "epoch": 1.0977372451174396,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008860448665096624,
      "loss": 0.9003,
      "step": 12759
    },
    {
      "epoch": 1.0978232814247613,
      "grad_norm": 1.84375,
      "learning_rate": 0.0008860276811918851,
      "loss": 1.3631,
      "step": 12760
    },
    {
      "epoch": 1.097909317732083,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008860104947450519,
      "loss": 1.143,
      "step": 12761
    },
    {
      "epoch": 1.0979953540394045,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008859933071692132,
      "loss": 1.8407,
      "step": 12762
    },
    {
      "epoch": 1.0980813903467264,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008859761184644191,
      "loss": 1.2684,
      "step": 12763
    },
    {
      "epoch": 1.098167426654048,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008859589286307197,
      "loss": 1.744,
      "step": 12764
    },
    {
      "epoch": 1.0982534629613696,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008859417376681658,
      "loss": 1.0194,
      "step": 12765
    },
    {
      "epoch": 1.0983394992686915,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008859245455768071,
      "loss": 1.4774,
      "step": 12766
    },
    {
      "epoch": 1.098425535576013,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008859073523566944,
      "loss": 1.3659,
      "step": 12767
    },
    {
      "epoch": 1.0985115718833347,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008858901580078777,
      "loss": 0.9665,
      "step": 12768
    },
    {
      "epoch": 1.0985976081906565,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008858729625304072,
      "loss": 1.085,
      "step": 12769
    },
    {
      "epoch": 1.0986836444979782,
      "grad_norm": 1.8125,
      "learning_rate": 0.0008858557659243334,
      "loss": 0.7353,
      "step": 12770
    },
    {
      "epoch": 1.0987696808052998,
      "grad_norm": 2.625,
      "learning_rate": 0.0008858385681897066,
      "loss": 1.2493,
      "step": 12771
    },
    {
      "epoch": 1.0988557171126216,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000885821369326577,
      "loss": 0.845,
      "step": 12772
    },
    {
      "epoch": 1.0989417534199433,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008858041693349949,
      "loss": 1.388,
      "step": 12773
    },
    {
      "epoch": 1.099027789727265,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008857869682150106,
      "loss": 0.9118,
      "step": 12774
    },
    {
      "epoch": 1.0991138260345865,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008857697659666745,
      "loss": 1.781,
      "step": 12775
    },
    {
      "epoch": 1.0991998623419084,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0008857525625900369,
      "loss": 1.972,
      "step": 12776
    },
    {
      "epoch": 1.09928589864923,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008857353580851481,
      "loss": 0.7494,
      "step": 12777
    },
    {
      "epoch": 1.0993719349565516,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008857181524520583,
      "loss": 0.822,
      "step": 12778
    },
    {
      "epoch": 1.0994579712638735,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008857009456908181,
      "loss": 1.1169,
      "step": 12779
    },
    {
      "epoch": 1.099544007571195,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008856837378014775,
      "loss": 1.5295,
      "step": 12780
    },
    {
      "epoch": 1.0996300438785167,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000885666528784087,
      "loss": 1.8081,
      "step": 12781
    },
    {
      "epoch": 1.0997160801858383,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008856493186386969,
      "loss": 1.8957,
      "step": 12782
    },
    {
      "epoch": 1.0998021164931602,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008856321073653575,
      "loss": 1.1876,
      "step": 12783
    },
    {
      "epoch": 1.0998881528004818,
      "grad_norm": 1.125,
      "learning_rate": 0.0008856148949641192,
      "loss": 1.2101,
      "step": 12784
    },
    {
      "epoch": 1.0999741891078034,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008855976814350324,
      "loss": 0.9162,
      "step": 12785
    },
    {
      "epoch": 1.1000602254151253,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008855804667781475,
      "loss": 1.6116,
      "step": 12786
    },
    {
      "epoch": 1.100146261722447,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008855632509935145,
      "loss": 1.199,
      "step": 12787
    },
    {
      "epoch": 1.1002322980297685,
      "grad_norm": 1.578125,
      "learning_rate": 0.000885546034081184,
      "loss": 1.7057,
      "step": 12788
    },
    {
      "epoch": 1.1003183343370901,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008855288160412064,
      "loss": 2.0869,
      "step": 12789
    },
    {
      "epoch": 1.100404370644412,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008855115968736321,
      "loss": 1.2233,
      "step": 12790
    },
    {
      "epoch": 1.1004904069517336,
      "grad_norm": 1.8515625,
      "learning_rate": 0.000885494376578511,
      "loss": 1.2751,
      "step": 12791
    },
    {
      "epoch": 1.1005764432590552,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0008854771551558942,
      "loss": 0.9911,
      "step": 12792
    },
    {
      "epoch": 1.100662479566377,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008854599326058314,
      "loss": 1.9802,
      "step": 12793
    },
    {
      "epoch": 1.1007485158736987,
      "grad_norm": 2.0,
      "learning_rate": 0.0008854427089283735,
      "loss": 1.1917,
      "step": 12794
    },
    {
      "epoch": 1.1008345521810203,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008854254841235707,
      "loss": 1.8258,
      "step": 12795
    },
    {
      "epoch": 1.1009205884883422,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000885408258191473,
      "loss": 1.0882,
      "step": 12796
    },
    {
      "epoch": 1.1010066247956638,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008853910311321314,
      "loss": 1.1841,
      "step": 12797
    },
    {
      "epoch": 1.1010926611029854,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008853738029455958,
      "loss": 0.7398,
      "step": 12798
    },
    {
      "epoch": 1.101178697410307,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008853565736319169,
      "loss": 1.4485,
      "step": 12799
    },
    {
      "epoch": 1.101264733717629,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000885339343191145,
      "loss": 0.9368,
      "step": 12800
    },
    {
      "epoch": 1.1013507700249505,
      "grad_norm": 1.25,
      "learning_rate": 0.0008853221116233303,
      "loss": 1.4293,
      "step": 12801
    },
    {
      "epoch": 1.1014368063322721,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008853048789285234,
      "loss": 1.5971,
      "step": 12802
    },
    {
      "epoch": 1.101522842639594,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008852876451067749,
      "loss": 1.2869,
      "step": 12803
    },
    {
      "epoch": 1.1016088789469156,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008852704101581347,
      "loss": 0.5446,
      "step": 12804
    },
    {
      "epoch": 1.1016949152542372,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008852531740826536,
      "loss": 0.7268,
      "step": 12805
    },
    {
      "epoch": 1.101780951561559,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008852359368803818,
      "loss": 0.9914,
      "step": 12806
    },
    {
      "epoch": 1.1018669878688807,
      "grad_norm": 1.203125,
      "learning_rate": 0.00088521869855137,
      "loss": 1.322,
      "step": 12807
    },
    {
      "epoch": 1.1019530241762023,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008852014590956682,
      "loss": 1.4208,
      "step": 12808
    },
    {
      "epoch": 1.102039060483524,
      "grad_norm": 2.046875,
      "learning_rate": 0.0008851842185133272,
      "loss": 1.1598,
      "step": 12809
    },
    {
      "epoch": 1.1021250967908458,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008851669768043972,
      "loss": 1.8726,
      "step": 12810
    },
    {
      "epoch": 1.1022111330981674,
      "grad_norm": 1.71875,
      "learning_rate": 0.0008851497339689287,
      "loss": 1.9468,
      "step": 12811
    },
    {
      "epoch": 1.102297169405489,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008851324900069721,
      "loss": 1.6135,
      "step": 12812
    },
    {
      "epoch": 1.102383205712811,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008851152449185779,
      "loss": 1.1957,
      "step": 12813
    },
    {
      "epoch": 1.1024692420201325,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008850979987037966,
      "loss": 1.2746,
      "step": 12814
    },
    {
      "epoch": 1.1025552783274541,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008850807513626787,
      "loss": 1.3908,
      "step": 12815
    },
    {
      "epoch": 1.1026413146347758,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008850635028952742,
      "loss": 1.1741,
      "step": 12816
    },
    {
      "epoch": 1.1027273509420976,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008850462533016339,
      "loss": 1.679,
      "step": 12817
    },
    {
      "epoch": 1.1028133872494192,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008850290025818082,
      "loss": 1.7665,
      "step": 12818
    },
    {
      "epoch": 1.1028994235567409,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008850117507358476,
      "loss": 1.7465,
      "step": 12819
    },
    {
      "epoch": 1.1029854598640627,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008849944977638025,
      "loss": 1.291,
      "step": 12820
    },
    {
      "epoch": 1.1030714961713843,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008849772436657233,
      "loss": 1.8262,
      "step": 12821
    },
    {
      "epoch": 1.103157532478706,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008849599884416607,
      "loss": 1.3442,
      "step": 12822
    },
    {
      "epoch": 1.1032435687860278,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008849427320916648,
      "loss": 1.2173,
      "step": 12823
    },
    {
      "epoch": 1.1033296050933494,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008849254746157864,
      "loss": 1.508,
      "step": 12824
    },
    {
      "epoch": 1.103415641400671,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008849082160140758,
      "loss": 1.9829,
      "step": 12825
    },
    {
      "epoch": 1.1035016777079927,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008848909562865837,
      "loss": 0.9453,
      "step": 12826
    },
    {
      "epoch": 1.1035877140153145,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008848736954333603,
      "loss": 1.1877,
      "step": 12827
    },
    {
      "epoch": 1.1036737503226361,
      "grad_norm": 1.109375,
      "learning_rate": 0.000884856433454456,
      "loss": 1.3717,
      "step": 12828
    },
    {
      "epoch": 1.1037597866299578,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008848391703499217,
      "loss": 1.2763,
      "step": 12829
    },
    {
      "epoch": 1.1038458229372796,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008848219061198077,
      "loss": 0.8227,
      "step": 12830
    },
    {
      "epoch": 1.1039318592446012,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008848046407641644,
      "loss": 0.9161,
      "step": 12831
    },
    {
      "epoch": 1.1040178955519229,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008847873742830423,
      "loss": 0.8882,
      "step": 12832
    },
    {
      "epoch": 1.1041039318592447,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008847701066764921,
      "loss": 1.1835,
      "step": 12833
    },
    {
      "epoch": 1.1041899681665663,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008847528379445639,
      "loss": 1.2904,
      "step": 12834
    },
    {
      "epoch": 1.104276004473888,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008847355680873088,
      "loss": 1.4117,
      "step": 12835
    },
    {
      "epoch": 1.1043620407812096,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008847182971047769,
      "loss": 1.3135,
      "step": 12836
    },
    {
      "epoch": 1.1044480770885314,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008847010249970188,
      "loss": 1.0576,
      "step": 12837
    },
    {
      "epoch": 1.104534113395853,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008846837517640848,
      "loss": 1.5906,
      "step": 12838
    },
    {
      "epoch": 1.1046201497031747,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000884666477406026,
      "loss": 1.9842,
      "step": 12839
    },
    {
      "epoch": 1.1047061860104965,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008846492019228924,
      "loss": 1.6887,
      "step": 12840
    },
    {
      "epoch": 1.1047922223178182,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008846319253147346,
      "loss": 1.0767,
      "step": 12841
    },
    {
      "epoch": 1.1048782586251398,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008846146475816035,
      "loss": 1.0304,
      "step": 12842
    },
    {
      "epoch": 1.1049642949324614,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008845973687235491,
      "loss": 1.6482,
      "step": 12843
    },
    {
      "epoch": 1.1050503312397832,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008845800887406222,
      "loss": 1.3238,
      "step": 12844
    },
    {
      "epoch": 1.1051363675471049,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008845628076328734,
      "loss": 0.5242,
      "step": 12845
    },
    {
      "epoch": 1.1052224038544265,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008845455254003533,
      "loss": 1.3088,
      "step": 12846
    },
    {
      "epoch": 1.1053084401617483,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008845282420431122,
      "loss": 1.2903,
      "step": 12847
    },
    {
      "epoch": 1.10539447646907,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008845109575612008,
      "loss": 1.0965,
      "step": 12848
    },
    {
      "epoch": 1.1054805127763916,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008844936719546697,
      "loss": 1.1414,
      "step": 12849
    },
    {
      "epoch": 1.1055665490837132,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008844763852235693,
      "loss": 1.482,
      "step": 12850
    },
    {
      "epoch": 1.105652585391035,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008844590973679503,
      "loss": 1.3361,
      "step": 12851
    },
    {
      "epoch": 1.1057386216983567,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008844418083878632,
      "loss": 1.1942,
      "step": 12852
    },
    {
      "epoch": 1.1058246580056783,
      "grad_norm": 0.875,
      "learning_rate": 0.0008844245182833585,
      "loss": 1.1509,
      "step": 12853
    },
    {
      "epoch": 1.1059106943130002,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008844072270544869,
      "loss": 0.8539,
      "step": 12854
    },
    {
      "epoch": 1.1059967306203218,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000884389934701299,
      "loss": 0.7807,
      "step": 12855
    },
    {
      "epoch": 1.1060827669276434,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008843726412238451,
      "loss": 1.8132,
      "step": 12856
    },
    {
      "epoch": 1.1061688032349652,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008843553466221762,
      "loss": 0.5395,
      "step": 12857
    },
    {
      "epoch": 1.1062548395422869,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008843380508963425,
      "loss": 1.3979,
      "step": 12858
    },
    {
      "epoch": 1.1063408758496085,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008843207540463948,
      "loss": 0.8234,
      "step": 12859
    },
    {
      "epoch": 1.1064269121569303,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008843034560723839,
      "loss": 0.9655,
      "step": 12860
    },
    {
      "epoch": 1.106512948464252,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008842861569743599,
      "loss": 1.4934,
      "step": 12861
    },
    {
      "epoch": 1.1065989847715736,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008842688567523736,
      "loss": 0.9686,
      "step": 12862
    },
    {
      "epoch": 1.1066850210788952,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008842515554064757,
      "loss": 1.0207,
      "step": 12863
    },
    {
      "epoch": 1.106771057386217,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008842342529367166,
      "loss": 1.453,
      "step": 12864
    },
    {
      "epoch": 1.1068570936935387,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008842169493431472,
      "loss": 1.6315,
      "step": 12865
    },
    {
      "epoch": 1.1069431300008603,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008841996446258179,
      "loss": 1.0819,
      "step": 12866
    },
    {
      "epoch": 1.1070291663081822,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008841823387847794,
      "loss": 0.9961,
      "step": 12867
    },
    {
      "epoch": 1.1071152026155038,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008841650318200823,
      "loss": 0.9455,
      "step": 12868
    },
    {
      "epoch": 1.1072012389228254,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000884147723731777,
      "loss": 0.8904,
      "step": 12869
    },
    {
      "epoch": 1.107287275230147,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008841304145199145,
      "loss": 1.3342,
      "step": 12870
    },
    {
      "epoch": 1.1073733115374689,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008841131041845452,
      "loss": 1.3229,
      "step": 12871
    },
    {
      "epoch": 1.1074593478447905,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008840957927257198,
      "loss": 1.4359,
      "step": 12872
    },
    {
      "epoch": 1.1075453841521121,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008840784801434889,
      "loss": 1.1853,
      "step": 12873
    },
    {
      "epoch": 1.107631420459434,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008840611664379031,
      "loss": 1.2841,
      "step": 12874
    },
    {
      "epoch": 1.1077174567667556,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008840438516090131,
      "loss": 1.0521,
      "step": 12875
    },
    {
      "epoch": 1.1078034930740772,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008840265356568696,
      "loss": 0.7449,
      "step": 12876
    },
    {
      "epoch": 1.1078895293813988,
      "grad_norm": 1.28125,
      "learning_rate": 0.000884009218581523,
      "loss": 1.9558,
      "step": 12877
    },
    {
      "epoch": 1.1079755656887207,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008839919003830243,
      "loss": 1.0604,
      "step": 12878
    },
    {
      "epoch": 1.1080616019960423,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000883974581061424,
      "loss": 0.9379,
      "step": 12879
    },
    {
      "epoch": 1.108147638303364,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008839572606167725,
      "loss": 1.4902,
      "step": 12880
    },
    {
      "epoch": 1.1082336746106858,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008839399390491208,
      "loss": 1.2786,
      "step": 12881
    },
    {
      "epoch": 1.1083197109180074,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008839226163585195,
      "loss": 0.9615,
      "step": 12882
    },
    {
      "epoch": 1.108405747225329,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008839052925450193,
      "loss": 1.1516,
      "step": 12883
    },
    {
      "epoch": 1.1084917835326509,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008838879676086707,
      "loss": 1.4896,
      "step": 12884
    },
    {
      "epoch": 1.1085778198399725,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008838706415495244,
      "loss": 0.6104,
      "step": 12885
    },
    {
      "epoch": 1.1086638561472941,
      "grad_norm": 1.5,
      "learning_rate": 0.0008838533143676311,
      "loss": 1.0674,
      "step": 12886
    },
    {
      "epoch": 1.1087498924546157,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008838359860630416,
      "loss": 1.6022,
      "step": 12887
    },
    {
      "epoch": 1.1088359287619376,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008838186566358064,
      "loss": 1.2384,
      "step": 12888
    },
    {
      "epoch": 1.1089219650692592,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008838013260859764,
      "loss": 1.1505,
      "step": 12889
    },
    {
      "epoch": 1.1090080013765808,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008837839944136021,
      "loss": 1.1044,
      "step": 12890
    },
    {
      "epoch": 1.1090940376839027,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008837666616187343,
      "loss": 1.518,
      "step": 12891
    },
    {
      "epoch": 1.1091800739912243,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008837493277014237,
      "loss": 0.9238,
      "step": 12892
    },
    {
      "epoch": 1.109266110298546,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008837319926617208,
      "loss": 1.4021,
      "step": 12893
    },
    {
      "epoch": 1.1093521466058678,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008837146564996765,
      "loss": 1.5379,
      "step": 12894
    },
    {
      "epoch": 1.1094381829131894,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008836973192153415,
      "loss": 1.1751,
      "step": 12895
    },
    {
      "epoch": 1.109524219220511,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0008836799808087665,
      "loss": 0.6706,
      "step": 12896
    },
    {
      "epoch": 1.1096102555278327,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008836626412800022,
      "loss": 1.3218,
      "step": 12897
    },
    {
      "epoch": 1.1096962918351545,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008836453006290992,
      "loss": 2.2964,
      "step": 12898
    },
    {
      "epoch": 1.1097823281424761,
      "grad_norm": 1.125,
      "learning_rate": 0.0008836279588561082,
      "loss": 1.3498,
      "step": 12899
    },
    {
      "epoch": 1.1098683644497978,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008836106159610802,
      "loss": 1.0108,
      "step": 12900
    },
    {
      "epoch": 1.1099544007571196,
      "grad_norm": 1.125,
      "learning_rate": 0.0008835932719440658,
      "loss": 0.8122,
      "step": 12901
    },
    {
      "epoch": 1.1100404370644412,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008835759268051156,
      "loss": 0.8904,
      "step": 12902
    },
    {
      "epoch": 1.1101264733717628,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008835585805442804,
      "loss": 1.0264,
      "step": 12903
    },
    {
      "epoch": 1.1102125096790845,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008835412331616111,
      "loss": 1.1277,
      "step": 12904
    },
    {
      "epoch": 1.1102985459864063,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008835238846571581,
      "loss": 0.5955,
      "step": 12905
    },
    {
      "epoch": 1.110384582293728,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008835065350309724,
      "loss": 1.5199,
      "step": 12906
    },
    {
      "epoch": 1.1104706186010496,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008834891842831045,
      "loss": 1.7758,
      "step": 12907
    },
    {
      "epoch": 1.1105566549083714,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008834718324136055,
      "loss": 1.0708,
      "step": 12908
    },
    {
      "epoch": 1.110642691215693,
      "grad_norm": 1.328125,
      "learning_rate": 0.000883454479422526,
      "loss": 0.8502,
      "step": 12909
    },
    {
      "epoch": 1.1107287275230147,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008834371253099167,
      "loss": 1.5023,
      "step": 12910
    },
    {
      "epoch": 1.1108147638303365,
      "grad_norm": 1.25,
      "learning_rate": 0.0008834197700758284,
      "loss": 1.2406,
      "step": 12911
    },
    {
      "epoch": 1.1109008001376581,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008834024137203118,
      "loss": 1.9916,
      "step": 12912
    },
    {
      "epoch": 1.1109868364449798,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008833850562434177,
      "loss": 0.6896,
      "step": 12913
    },
    {
      "epoch": 1.1110728727523014,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008833676976451968,
      "loss": 1.3849,
      "step": 12914
    },
    {
      "epoch": 1.1111589090596232,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008833503379257002,
      "loss": 1.1199,
      "step": 12915
    },
    {
      "epoch": 1.1112449453669448,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008833329770849782,
      "loss": 1.6325,
      "step": 12916
    },
    {
      "epoch": 1.1113309816742665,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008833156151230818,
      "loss": 0.9626,
      "step": 12917
    },
    {
      "epoch": 1.1114170179815883,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008832982520400619,
      "loss": 1.1141,
      "step": 12918
    },
    {
      "epoch": 1.11150305428891,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008832808878359691,
      "loss": 1.0821,
      "step": 12919
    },
    {
      "epoch": 1.1115890905962316,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008832635225108542,
      "loss": 1.3827,
      "step": 12920
    },
    {
      "epoch": 1.1116751269035534,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000883246156064768,
      "loss": 1.2801,
      "step": 12921
    },
    {
      "epoch": 1.111761163210875,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008832287884977614,
      "loss": 1.5304,
      "step": 12922
    },
    {
      "epoch": 1.1118471995181967,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008832114198098852,
      "loss": 1.3315,
      "step": 12923
    },
    {
      "epoch": 1.1119332358255183,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00088319405000119,
      "loss": 1.2926,
      "step": 12924
    },
    {
      "epoch": 1.1120192721328401,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008831766790717269,
      "loss": 0.7976,
      "step": 12925
    },
    {
      "epoch": 1.1121053084401618,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008831593070215463,
      "loss": 1.7901,
      "step": 12926
    },
    {
      "epoch": 1.1121913447474834,
      "grad_norm": 2.515625,
      "learning_rate": 0.0008831419338506994,
      "loss": 1.7268,
      "step": 12927
    },
    {
      "epoch": 1.1122773810548052,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008831245595592367,
      "loss": 0.6024,
      "step": 12928
    },
    {
      "epoch": 1.1123634173621268,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008831071841472094,
      "loss": 1.1845,
      "step": 12929
    },
    {
      "epoch": 1.1124494536694485,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008830898076146681,
      "loss": 0.8517,
      "step": 12930
    },
    {
      "epoch": 1.11253548997677,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008830724299616634,
      "loss": 1.356,
      "step": 12931
    },
    {
      "epoch": 1.112621526284092,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008830550511882466,
      "loss": 1.0528,
      "step": 12932
    },
    {
      "epoch": 1.1127075625914136,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008830376712944681,
      "loss": 1.1376,
      "step": 12933
    },
    {
      "epoch": 1.1127935988987352,
      "grad_norm": 1.296875,
      "learning_rate": 0.000883020290280379,
      "loss": 1.9374,
      "step": 12934
    },
    {
      "epoch": 1.112879635206057,
      "grad_norm": 1.15625,
      "learning_rate": 0.00088300290814603,
      "loss": 1.2263,
      "step": 12935
    },
    {
      "epoch": 1.1129656715133787,
      "grad_norm": 1.40625,
      "learning_rate": 0.000882985524891472,
      "loss": 1.4583,
      "step": 12936
    },
    {
      "epoch": 1.1130517078207003,
      "grad_norm": 1.125,
      "learning_rate": 0.0008829681405167558,
      "loss": 1.4244,
      "step": 12937
    },
    {
      "epoch": 1.113137744128022,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008829507550219322,
      "loss": 1.6549,
      "step": 12938
    },
    {
      "epoch": 1.1132237804353438,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008829333684070522,
      "loss": 1.5854,
      "step": 12939
    },
    {
      "epoch": 1.1133098167426654,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008829159806721666,
      "loss": 1.5999,
      "step": 12940
    },
    {
      "epoch": 1.113395853049987,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008828985918173263,
      "loss": 1.5234,
      "step": 12941
    },
    {
      "epoch": 1.1134818893573089,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008828812018425819,
      "loss": 1.3022,
      "step": 12942
    },
    {
      "epoch": 1.1135679256646305,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008828638107479845,
      "loss": 1.4313,
      "step": 12943
    },
    {
      "epoch": 1.113653961971952,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000882846418533585,
      "loss": 1.2959,
      "step": 12944
    },
    {
      "epoch": 1.113739998279274,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008828290251994342,
      "loss": 1.232,
      "step": 12945
    },
    {
      "epoch": 1.1138260345865956,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008828116307455829,
      "loss": 1.5212,
      "step": 12946
    },
    {
      "epoch": 1.1139120708939172,
      "grad_norm": 1.421875,
      "learning_rate": 0.000882794235172082,
      "loss": 1.8987,
      "step": 12947
    },
    {
      "epoch": 1.113998107201239,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008827768384789825,
      "loss": 1.77,
      "step": 12948
    },
    {
      "epoch": 1.1140841435085607,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000882759440666335,
      "loss": 1.5173,
      "step": 12949
    },
    {
      "epoch": 1.1141701798158823,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008827420417341908,
      "loss": 1.472,
      "step": 12950
    },
    {
      "epoch": 1.114256216123204,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008827246416826005,
      "loss": 0.9284,
      "step": 12951
    },
    {
      "epoch": 1.1143422524305258,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000882707240511615,
      "loss": 2.1335,
      "step": 12952
    },
    {
      "epoch": 1.1144282887378474,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008826898382212853,
      "loss": 1.2943,
      "step": 12953
    },
    {
      "epoch": 1.114514325045169,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008826724348116624,
      "loss": 1.2018,
      "step": 12954
    },
    {
      "epoch": 1.1146003613524909,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008826550302827968,
      "loss": 1.18,
      "step": 12955
    },
    {
      "epoch": 1.1146863976598125,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008826376246347398,
      "loss": 1.865,
      "step": 12956
    },
    {
      "epoch": 1.114772433967134,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008826202178675421,
      "loss": 0.6773,
      "step": 12957
    },
    {
      "epoch": 1.1148584702744557,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008826028099812547,
      "loss": 1.6876,
      "step": 12958
    },
    {
      "epoch": 1.1149445065817776,
      "grad_norm": 1.5,
      "learning_rate": 0.0008825854009759285,
      "loss": 2.1463,
      "step": 12959
    },
    {
      "epoch": 1.1150305428890992,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008825679908516144,
      "loss": 2.038,
      "step": 12960
    },
    {
      "epoch": 1.1151165791964208,
      "grad_norm": 1.125,
      "learning_rate": 0.0008825505796083634,
      "loss": 1.3705,
      "step": 12961
    },
    {
      "epoch": 1.1152026155037427,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008825331672462263,
      "loss": 1.2885,
      "step": 12962
    },
    {
      "epoch": 1.1152886518110643,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008825157537652542,
      "loss": 2.2373,
      "step": 12963
    },
    {
      "epoch": 1.115374688118386,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008824983391654977,
      "loss": 0.8924,
      "step": 12964
    },
    {
      "epoch": 1.1154607244257075,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000882480923447008,
      "loss": 1.105,
      "step": 12965
    },
    {
      "epoch": 1.1155467607330294,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008824635066098361,
      "loss": 0.9854,
      "step": 12966
    },
    {
      "epoch": 1.115632797040351,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008824460886540327,
      "loss": 0.6189,
      "step": 12967
    },
    {
      "epoch": 1.1157188333476726,
      "grad_norm": 1.328125,
      "learning_rate": 0.000882428669579649,
      "loss": 0.9182,
      "step": 12968
    },
    {
      "epoch": 1.1158048696549945,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008824112493867358,
      "loss": 0.9784,
      "step": 12969
    },
    {
      "epoch": 1.115890905962316,
      "grad_norm": 1.015625,
      "learning_rate": 0.000882393828075344,
      "loss": 0.8428,
      "step": 12970
    },
    {
      "epoch": 1.1159769422696377,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008823764056455247,
      "loss": 1.826,
      "step": 12971
    },
    {
      "epoch": 1.1160629785769596,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008823589820973287,
      "loss": 1.2085,
      "step": 12972
    },
    {
      "epoch": 1.1161490148842812,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008823415574308072,
      "loss": 1.3701,
      "step": 12973
    },
    {
      "epoch": 1.1162350511916028,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008823241316460107,
      "loss": 1.1927,
      "step": 12974
    },
    {
      "epoch": 1.1163210874989244,
      "grad_norm": 0.890625,
      "learning_rate": 0.0008823067047429907,
      "loss": 0.8968,
      "step": 12975
    },
    {
      "epoch": 1.1164071238062463,
      "grad_norm": 2.0625,
      "learning_rate": 0.0008822892767217978,
      "loss": 1.3982,
      "step": 12976
    },
    {
      "epoch": 1.116493160113568,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008822718475824834,
      "loss": 1.8211,
      "step": 12977
    },
    {
      "epoch": 1.1165791964208895,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008822544173250979,
      "loss": 1.9381,
      "step": 12978
    },
    {
      "epoch": 1.1166652327282114,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008822369859496927,
      "loss": 0.788,
      "step": 12979
    },
    {
      "epoch": 1.116751269035533,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008822195534563187,
      "loss": 1.888,
      "step": 12980
    },
    {
      "epoch": 1.1168373053428546,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008822021198450267,
      "loss": 0.5307,
      "step": 12981
    },
    {
      "epoch": 1.1169233416501765,
      "grad_norm": 1.390625,
      "learning_rate": 0.000882184685115868,
      "loss": 1.4052,
      "step": 12982
    },
    {
      "epoch": 1.117009377957498,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008821672492688933,
      "loss": 1.728,
      "step": 12983
    },
    {
      "epoch": 1.1170954142648197,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008821498123041538,
      "loss": 0.8092,
      "step": 12984
    },
    {
      "epoch": 1.1171814505721414,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008821323742217004,
      "loss": 1.0629,
      "step": 12985
    },
    {
      "epoch": 1.1172674868794632,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008821149350215842,
      "loss": 0.6117,
      "step": 12986
    },
    {
      "epoch": 1.1173535231867848,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008820974947038559,
      "loss": 1.498,
      "step": 12987
    },
    {
      "epoch": 1.1174395594941064,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008820800532685669,
      "loss": 1.457,
      "step": 12988
    },
    {
      "epoch": 1.1175255958014283,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008820626107157681,
      "loss": 1.1926,
      "step": 12989
    },
    {
      "epoch": 1.11761163210875,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008820451670455104,
      "loss": 1.5117,
      "step": 12990
    },
    {
      "epoch": 1.1176976684160715,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008820277222578449,
      "loss": 1.1217,
      "step": 12991
    },
    {
      "epoch": 1.1177837047233932,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008820102763528226,
      "loss": 0.8621,
      "step": 12992
    },
    {
      "epoch": 1.117869741030715,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008819928293304946,
      "loss": 1.4308,
      "step": 12993
    },
    {
      "epoch": 1.1179557773380366,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008819753811909119,
      "loss": 1.8592,
      "step": 12994
    },
    {
      "epoch": 1.1180418136453583,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008819579319341253,
      "loss": 1.0554,
      "step": 12995
    },
    {
      "epoch": 1.11812784995268,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008819404815601862,
      "loss": 1.2429,
      "step": 12996
    },
    {
      "epoch": 1.1182138862600017,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008819230300691455,
      "loss": 0.9715,
      "step": 12997
    },
    {
      "epoch": 1.1182999225673234,
      "grad_norm": 1.53125,
      "learning_rate": 0.000881905577461054,
      "loss": 1.4836,
      "step": 12998
    },
    {
      "epoch": 1.1183859588746452,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008818881237359633,
      "loss": 0.8441,
      "step": 12999
    },
    {
      "epoch": 1.1184719951819668,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008818706688939238,
      "loss": 0.9807,
      "step": 13000
    },
    {
      "epoch": 1.1185580314892885,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000881853212934987,
      "loss": 1.3962,
      "step": 13001
    },
    {
      "epoch": 1.11864406779661,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008818357558592039,
      "loss": 1.6678,
      "step": 13002
    },
    {
      "epoch": 1.118730104103932,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008818182976666253,
      "loss": 1.309,
      "step": 13003
    },
    {
      "epoch": 1.1188161404112535,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008818008383573026,
      "loss": 1.051,
      "step": 13004
    },
    {
      "epoch": 1.1189021767185752,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008817833779312866,
      "loss": 1.9186,
      "step": 13005
    },
    {
      "epoch": 1.118988213025897,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008817659163886285,
      "loss": 0.8753,
      "step": 13006
    },
    {
      "epoch": 1.1190742493332186,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008817484537293794,
      "loss": 1.3787,
      "step": 13007
    },
    {
      "epoch": 1.1191602856405403,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008817309899535903,
      "loss": 1.0107,
      "step": 13008
    },
    {
      "epoch": 1.119246321947862,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008817135250613121,
      "loss": 1.3348,
      "step": 13009
    },
    {
      "epoch": 1.1193323582551837,
      "grad_norm": 1.25,
      "learning_rate": 0.0008816960590525963,
      "loss": 1.3226,
      "step": 13010
    },
    {
      "epoch": 1.1194183945625054,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008816785919274937,
      "loss": 0.7931,
      "step": 13011
    },
    {
      "epoch": 1.119504430869827,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008816611236860555,
      "loss": 1.1947,
      "step": 13012
    },
    {
      "epoch": 1.1195904671771488,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008816436543283326,
      "loss": 1.8418,
      "step": 13013
    },
    {
      "epoch": 1.1196765034844705,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008816261838543764,
      "loss": 1.7512,
      "step": 13014
    },
    {
      "epoch": 1.119762539791792,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008816087122642378,
      "loss": 1.3159,
      "step": 13015
    },
    {
      "epoch": 1.119848576099114,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008815912395579679,
      "loss": 0.9079,
      "step": 13016
    },
    {
      "epoch": 1.1199346124064355,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008815737657356178,
      "loss": 0.8131,
      "step": 13017
    },
    {
      "epoch": 1.1200206487137572,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008815562907972387,
      "loss": 0.7916,
      "step": 13018
    },
    {
      "epoch": 1.1201066850210788,
      "grad_norm": 3.0625,
      "learning_rate": 0.0008815388147428815,
      "loss": 0.7988,
      "step": 13019
    },
    {
      "epoch": 1.1201927213284006,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008815213375725978,
      "loss": 1.5818,
      "step": 13020
    },
    {
      "epoch": 1.1202787576357223,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008815038592864382,
      "loss": 1.1319,
      "step": 13021
    },
    {
      "epoch": 1.120364793943044,
      "grad_norm": 1.328125,
      "learning_rate": 0.000881486379884454,
      "loss": 1.4752,
      "step": 13022
    },
    {
      "epoch": 1.1204508302503657,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008814688993666964,
      "loss": 1.5227,
      "step": 13023
    },
    {
      "epoch": 1.1205368665576874,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008814514177332164,
      "loss": 1.4573,
      "step": 13024
    },
    {
      "epoch": 1.120622902865009,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008814339349840653,
      "loss": 0.8951,
      "step": 13025
    },
    {
      "epoch": 1.1207089391723306,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008814164511192939,
      "loss": 1.2538,
      "step": 13026
    },
    {
      "epoch": 1.1207949754796525,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008813989661389537,
      "loss": 0.7781,
      "step": 13027
    },
    {
      "epoch": 1.120881011786974,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008813814800430958,
      "loss": 1.2382,
      "step": 13028
    },
    {
      "epoch": 1.1209670480942957,
      "grad_norm": 1.0625,
      "learning_rate": 0.000881363992831771,
      "loss": 1.6867,
      "step": 13029
    },
    {
      "epoch": 1.1210530844016176,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008813465045050308,
      "loss": 1.1521,
      "step": 13030
    },
    {
      "epoch": 1.1211391207089392,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008813290150629263,
      "loss": 1.0393,
      "step": 13031
    },
    {
      "epoch": 1.1212251570162608,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008813115245055085,
      "loss": 0.8378,
      "step": 13032
    },
    {
      "epoch": 1.1213111933235826,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008812940328328288,
      "loss": 1.2648,
      "step": 13033
    },
    {
      "epoch": 1.1213972296309043,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008812765400449382,
      "loss": 1.202,
      "step": 13034
    },
    {
      "epoch": 1.121483265938226,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008812590461418877,
      "loss": 1.5347,
      "step": 13035
    },
    {
      "epoch": 1.1215693022455477,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008812415511237288,
      "loss": 0.9759,
      "step": 13036
    },
    {
      "epoch": 1.1216553385528694,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008812240549905125,
      "loss": 1.2575,
      "step": 13037
    },
    {
      "epoch": 1.121741374860191,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008812065577422899,
      "loss": 1.9436,
      "step": 13038
    },
    {
      "epoch": 1.1218274111675126,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008811890593791123,
      "loss": 0.9657,
      "step": 13039
    },
    {
      "epoch": 1.1219134474748345,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008811715599010308,
      "loss": 0.8151,
      "step": 13040
    },
    {
      "epoch": 1.121999483782156,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008811540593080966,
      "loss": 0.965,
      "step": 13041
    },
    {
      "epoch": 1.1220855200894777,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0008811365576003612,
      "loss": 2.0107,
      "step": 13042
    },
    {
      "epoch": 1.1221715563967996,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008811190547778751,
      "loss": 1.4781,
      "step": 13043
    },
    {
      "epoch": 1.1222575927041212,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008811015508406901,
      "loss": 1.3628,
      "step": 13044
    },
    {
      "epoch": 1.1223436290114428,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008810840457888572,
      "loss": 1.9815,
      "step": 13045
    },
    {
      "epoch": 1.1224296653187644,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008810665396224276,
      "loss": 0.9022,
      "step": 13046
    },
    {
      "epoch": 1.1225157016260863,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008810490323414522,
      "loss": 0.8449,
      "step": 13047
    },
    {
      "epoch": 1.122601737933408,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008810315239459827,
      "loss": 1.4681,
      "step": 13048
    },
    {
      "epoch": 1.1226877742407295,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008810140144360701,
      "loss": 1.4708,
      "step": 13049
    },
    {
      "epoch": 1.1227738105480514,
      "grad_norm": 1.25,
      "learning_rate": 0.0008809965038117656,
      "loss": 0.7653,
      "step": 13050
    },
    {
      "epoch": 1.122859846855373,
      "grad_norm": 1.125,
      "learning_rate": 0.0008809789920731205,
      "loss": 1.1855,
      "step": 13051
    },
    {
      "epoch": 1.1229458831626946,
      "grad_norm": 6.25,
      "learning_rate": 0.0008809614792201859,
      "loss": 1.4733,
      "step": 13052
    },
    {
      "epoch": 1.1230319194700162,
      "grad_norm": 1.234375,
      "learning_rate": 0.000880943965253013,
      "loss": 1.6705,
      "step": 13053
    },
    {
      "epoch": 1.123117955777338,
      "grad_norm": 3.328125,
      "learning_rate": 0.0008809264501716532,
      "loss": 1.8075,
      "step": 13054
    },
    {
      "epoch": 1.1232039920846597,
      "grad_norm": 1.0,
      "learning_rate": 0.0008809089339761576,
      "loss": 1.0329,
      "step": 13055
    },
    {
      "epoch": 1.1232900283919813,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008808914166665773,
      "loss": 1.2499,
      "step": 13056
    },
    {
      "epoch": 1.1233760646993032,
      "grad_norm": 2.1875,
      "learning_rate": 0.0008808738982429638,
      "loss": 1.2772,
      "step": 13057
    },
    {
      "epoch": 1.1234621010066248,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008808563787053681,
      "loss": 1.3157,
      "step": 13058
    },
    {
      "epoch": 1.1235481373139464,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008808388580538417,
      "loss": 1.0052,
      "step": 13059
    },
    {
      "epoch": 1.1236341736212683,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008808213362884357,
      "loss": 0.9555,
      "step": 13060
    },
    {
      "epoch": 1.12372020992859,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008808038134092013,
      "loss": 1.1687,
      "step": 13061
    },
    {
      "epoch": 1.1238062462359115,
      "grad_norm": 1.484375,
      "learning_rate": 0.00088078628941619,
      "loss": 1.3588,
      "step": 13062
    },
    {
      "epoch": 1.1238922825432331,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008807687643094526,
      "loss": 1.1392,
      "step": 13063
    },
    {
      "epoch": 1.123978318850555,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008807512380890408,
      "loss": 1.5749,
      "step": 13064
    },
    {
      "epoch": 1.1240643551578766,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008807337107550055,
      "loss": 1.7169,
      "step": 13065
    },
    {
      "epoch": 1.1241503914651982,
      "grad_norm": 1.5,
      "learning_rate": 0.0008807161823073984,
      "loss": 0.9624,
      "step": 13066
    },
    {
      "epoch": 1.12423642777252,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008806986527462704,
      "loss": 0.7245,
      "step": 13067
    },
    {
      "epoch": 1.1243224640798417,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008806811220716729,
      "loss": 1.3024,
      "step": 13068
    },
    {
      "epoch": 1.1244085003871633,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000880663590283657,
      "loss": 1.7569,
      "step": 13069
    },
    {
      "epoch": 1.1244945366944852,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008806460573822742,
      "loss": 0.8987,
      "step": 13070
    },
    {
      "epoch": 1.1245805730018068,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008806285233675757,
      "loss": 1.4615,
      "step": 13071
    },
    {
      "epoch": 1.1246666093091284,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000880610988239613,
      "loss": 2.1067,
      "step": 13072
    },
    {
      "epoch": 1.12475264561645,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000880593451998437,
      "loss": 1.4022,
      "step": 13073
    },
    {
      "epoch": 1.124838681923772,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008805759146440993,
      "loss": 0.959,
      "step": 13074
    },
    {
      "epoch": 1.1249247182310935,
      "grad_norm": 1.359375,
      "learning_rate": 0.000880558376176651,
      "loss": 0.875,
      "step": 13075
    },
    {
      "epoch": 1.1250107545384151,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008805408365961433,
      "loss": 0.8729,
      "step": 13076
    },
    {
      "epoch": 1.125096790845737,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008805232959026278,
      "loss": 1.3521,
      "step": 13077
    },
    {
      "epoch": 1.1251828271530586,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008805057540961556,
      "loss": 1.5278,
      "step": 13078
    },
    {
      "epoch": 1.1252688634603802,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008804882111767782,
      "loss": 0.8624,
      "step": 13079
    },
    {
      "epoch": 1.1253548997677019,
      "grad_norm": 1.125,
      "learning_rate": 0.0008804706671445467,
      "loss": 1.1163,
      "step": 13080
    },
    {
      "epoch": 1.1254409360750237,
      "grad_norm": 2.296875,
      "learning_rate": 0.0008804531219995127,
      "loss": 1.0957,
      "step": 13081
    },
    {
      "epoch": 1.1255269723823453,
      "grad_norm": 1.546875,
      "learning_rate": 0.000880435575741727,
      "loss": 1.6649,
      "step": 13082
    },
    {
      "epoch": 1.125613008689667,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008804180283712414,
      "loss": 1.1074,
      "step": 13083
    },
    {
      "epoch": 1.1256990449969888,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008804004798881069,
      "loss": 1.2566,
      "step": 13084
    },
    {
      "epoch": 1.1257850813043104,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008803829302923751,
      "loss": 1.0336,
      "step": 13085
    },
    {
      "epoch": 1.125871117611632,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008803653795840974,
      "loss": 1.328,
      "step": 13086
    },
    {
      "epoch": 1.1259571539189537,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008803478277633246,
      "loss": 1.3467,
      "step": 13087
    },
    {
      "epoch": 1.1260431902262755,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008803302748301087,
      "loss": 0.9673,
      "step": 13088
    },
    {
      "epoch": 1.1261292265335971,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008803127207845005,
      "loss": 0.9682,
      "step": 13089
    },
    {
      "epoch": 1.1262152628409188,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008802951656265517,
      "loss": 1.0803,
      "step": 13090
    },
    {
      "epoch": 1.1263012991482406,
      "grad_norm": 1.78125,
      "learning_rate": 0.0008802776093563133,
      "loss": 1.3401,
      "step": 13091
    },
    {
      "epoch": 1.1263873354555622,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008802600519738369,
      "loss": 1.0597,
      "step": 13092
    },
    {
      "epoch": 1.1264733717628839,
      "grad_norm": 1.125,
      "learning_rate": 0.0008802424934791739,
      "loss": 1.1189,
      "step": 13093
    },
    {
      "epoch": 1.1265594080702057,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008802249338723756,
      "loss": 1.0936,
      "step": 13094
    },
    {
      "epoch": 1.1266454443775273,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008802073731534932,
      "loss": 0.6008,
      "step": 13095
    },
    {
      "epoch": 1.126731480684849,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008801898113225781,
      "loss": 1.5408,
      "step": 13096
    },
    {
      "epoch": 1.1268175169921708,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008801722483796819,
      "loss": 0.8229,
      "step": 13097
    },
    {
      "epoch": 1.1269035532994924,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008801546843248559,
      "loss": 1.5423,
      "step": 13098
    },
    {
      "epoch": 1.126989589606814,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008801371191581514,
      "loss": 1.2947,
      "step": 13099
    },
    {
      "epoch": 1.1270756259141357,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008801195528796195,
      "loss": 1.0832,
      "step": 13100
    },
    {
      "epoch": 1.1271616622214575,
      "grad_norm": 1.0625,
      "learning_rate": 0.000880101985489312,
      "loss": 0.8991,
      "step": 13101
    },
    {
      "epoch": 1.1272476985287792,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008800844169872799,
      "loss": 2.1519,
      "step": 13102
    },
    {
      "epoch": 1.1273337348361008,
      "grad_norm": 1.96875,
      "learning_rate": 0.0008800668473735752,
      "loss": 1.1495,
      "step": 13103
    },
    {
      "epoch": 1.1274197711434226,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008800492766482486,
      "loss": 1.5557,
      "step": 13104
    },
    {
      "epoch": 1.1275058074507442,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008800317048113517,
      "loss": 1.0481,
      "step": 13105
    },
    {
      "epoch": 1.1275918437580659,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0008800141318629362,
      "loss": 1.312,
      "step": 13106
    },
    {
      "epoch": 1.1276778800653875,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008799965578030533,
      "loss": 1.6983,
      "step": 13107
    },
    {
      "epoch": 1.1277639163727093,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008799789826317543,
      "loss": 1.6041,
      "step": 13108
    },
    {
      "epoch": 1.127849952680031,
      "grad_norm": 1.125,
      "learning_rate": 0.0008799614063490906,
      "loss": 1.1502,
      "step": 13109
    },
    {
      "epoch": 1.1279359889873526,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008799438289551138,
      "loss": 1.4925,
      "step": 13110
    },
    {
      "epoch": 1.1280220252946744,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008799262504498751,
      "loss": 0.7465,
      "step": 13111
    },
    {
      "epoch": 1.128108061601996,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000879908670833426,
      "loss": 1.1226,
      "step": 13112
    },
    {
      "epoch": 1.1281940979093177,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008798910901058179,
      "loss": 1.0819,
      "step": 13113
    },
    {
      "epoch": 1.1282801342166393,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008798735082671023,
      "loss": 1.2441,
      "step": 13114
    },
    {
      "epoch": 1.1283661705239612,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008798559253173307,
      "loss": 0.9581,
      "step": 13115
    },
    {
      "epoch": 1.1284522068312828,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008798383412565542,
      "loss": 1.5034,
      "step": 13116
    },
    {
      "epoch": 1.1285382431386044,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008798207560848245,
      "loss": 1.2498,
      "step": 13117
    },
    {
      "epoch": 1.1286242794459262,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008798031698021931,
      "loss": 1.6352,
      "step": 13118
    },
    {
      "epoch": 1.1287103157532479,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008797855824087111,
      "loss": 1.6516,
      "step": 13119
    },
    {
      "epoch": 1.1287963520605695,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008797679939044303,
      "loss": 1.9111,
      "step": 13120
    },
    {
      "epoch": 1.1288823883678913,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008797504042894018,
      "loss": 1.0017,
      "step": 13121
    },
    {
      "epoch": 1.128968424675213,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008797328135636773,
      "loss": 0.8789,
      "step": 13122
    },
    {
      "epoch": 1.1290544609825346,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008797152217273082,
      "loss": 0.7403,
      "step": 13123
    },
    {
      "epoch": 1.1291404972898564,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008796976287803459,
      "loss": 0.7039,
      "step": 13124
    },
    {
      "epoch": 1.129226533597178,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008796800347228419,
      "loss": 0.4779,
      "step": 13125
    },
    {
      "epoch": 1.1293125699044997,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008796624395548476,
      "loss": 1.6852,
      "step": 13126
    },
    {
      "epoch": 1.1293986062118213,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008796448432764146,
      "loss": 1.1705,
      "step": 13127
    },
    {
      "epoch": 1.1294846425191432,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008796272458875943,
      "loss": 0.5872,
      "step": 13128
    },
    {
      "epoch": 1.1295706788264648,
      "grad_norm": 1.125,
      "learning_rate": 0.0008796096473884378,
      "loss": 1.1697,
      "step": 13129
    },
    {
      "epoch": 1.1296567151337864,
      "grad_norm": 1.828125,
      "learning_rate": 0.0008795920477789972,
      "loss": 1.7465,
      "step": 13130
    },
    {
      "epoch": 1.1297427514411083,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008795744470593236,
      "loss": 0.9323,
      "step": 13131
    },
    {
      "epoch": 1.1298287877484299,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008795568452294686,
      "loss": 1.5874,
      "step": 13132
    },
    {
      "epoch": 1.1299148240557515,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008795392422894837,
      "loss": 1.2737,
      "step": 13133
    },
    {
      "epoch": 1.1300008603630731,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008795216382394204,
      "loss": 1.4371,
      "step": 13134
    },
    {
      "epoch": 1.130086896670395,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00087950403307933,
      "loss": 0.7642,
      "step": 13135
    },
    {
      "epoch": 1.1301729329777166,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000879486426809264,
      "loss": 0.9016,
      "step": 13136
    },
    {
      "epoch": 1.1302589692850382,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008794688194292741,
      "loss": 2.1753,
      "step": 13137
    },
    {
      "epoch": 1.13034500559236,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008794512109394118,
      "loss": 1.074,
      "step": 13138
    },
    {
      "epoch": 1.1304310418996817,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008794336013397283,
      "loss": 0.8907,
      "step": 13139
    },
    {
      "epoch": 1.1305170782070033,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008794159906302757,
      "loss": 0.7088,
      "step": 13140
    },
    {
      "epoch": 1.130603114514325,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008793983788111049,
      "loss": 1.8182,
      "step": 13141
    },
    {
      "epoch": 1.1306891508216468,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008793807658822675,
      "loss": 1.424,
      "step": 13142
    },
    {
      "epoch": 1.1307751871289684,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008793631518438152,
      "loss": 1.8696,
      "step": 13143
    },
    {
      "epoch": 1.13086122343629,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008793455366957997,
      "loss": 1.7967,
      "step": 13144
    },
    {
      "epoch": 1.1309472597436119,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000879327920438272,
      "loss": 1.2226,
      "step": 13145
    },
    {
      "epoch": 1.1310332960509335,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008793103030712839,
      "loss": 2.0859,
      "step": 13146
    },
    {
      "epoch": 1.1311193323582551,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008792926845948872,
      "loss": 2.1334,
      "step": 13147
    },
    {
      "epoch": 1.1312053686655767,
      "grad_norm": 1.140625,
      "learning_rate": 0.000879275065009133,
      "loss": 1.3235,
      "step": 13148
    },
    {
      "epoch": 1.1312914049728986,
      "grad_norm": 0.890625,
      "learning_rate": 0.000879257444314073,
      "loss": 0.6451,
      "step": 13149
    },
    {
      "epoch": 1.1313774412802202,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008792398225097586,
      "loss": 1.6596,
      "step": 13150
    },
    {
      "epoch": 1.131463477587542,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008792221995962417,
      "loss": 1.3161,
      "step": 13151
    },
    {
      "epoch": 1.1315495138948637,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008792045755735736,
      "loss": 1.2177,
      "step": 13152
    },
    {
      "epoch": 1.1316355502021853,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008791869504418058,
      "loss": 0.8389,
      "step": 13153
    },
    {
      "epoch": 1.131721586509507,
      "grad_norm": 2.8125,
      "learning_rate": 0.0008791693242009899,
      "loss": 0.8777,
      "step": 13154
    },
    {
      "epoch": 1.1318076228168288,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008791516968511775,
      "loss": 1.938,
      "step": 13155
    },
    {
      "epoch": 1.1318936591241504,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00087913406839242,
      "loss": 1.7417,
      "step": 13156
    },
    {
      "epoch": 1.131979695431472,
      "grad_norm": 1.90625,
      "learning_rate": 0.0008791164388247694,
      "loss": 1.4964,
      "step": 13157
    },
    {
      "epoch": 1.1320657317387939,
      "grad_norm": 1.375,
      "learning_rate": 0.0008790988081482767,
      "loss": 1.9158,
      "step": 13158
    },
    {
      "epoch": 1.1321517680461155,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0008790811763629937,
      "loss": 1.2742,
      "step": 13159
    },
    {
      "epoch": 1.1322378043534371,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008790635434689721,
      "loss": 1.3283,
      "step": 13160
    },
    {
      "epoch": 1.1323238406607588,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008790459094662633,
      "loss": 1.7041,
      "step": 13161
    },
    {
      "epoch": 1.1324098769680806,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008790282743549189,
      "loss": 1.3711,
      "step": 13162
    },
    {
      "epoch": 1.1324959132754022,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008790106381349908,
      "loss": 1.2142,
      "step": 13163
    },
    {
      "epoch": 1.1325819495827238,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008789930008065299,
      "loss": 1.724,
      "step": 13164
    },
    {
      "epoch": 1.1326679858900457,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008789753623695884,
      "loss": 1.9531,
      "step": 13165
    },
    {
      "epoch": 1.1327540221973673,
      "grad_norm": 1.375,
      "learning_rate": 0.0008789577228242176,
      "loss": 1.0003,
      "step": 13166
    },
    {
      "epoch": 1.132840058504689,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008789400821704691,
      "loss": 1.0443,
      "step": 13167
    },
    {
      "epoch": 1.1329260948120106,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008789224404083947,
      "loss": 2.2269,
      "step": 13168
    },
    {
      "epoch": 1.1330121311193324,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008789047975380458,
      "loss": 1.3138,
      "step": 13169
    },
    {
      "epoch": 1.133098167426654,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008788871535594741,
      "loss": 1.3086,
      "step": 13170
    },
    {
      "epoch": 1.1331842037339757,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008788695084727312,
      "loss": 1.4031,
      "step": 13171
    },
    {
      "epoch": 1.1332702400412975,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008788518622778685,
      "loss": 1.041,
      "step": 13172
    },
    {
      "epoch": 1.1333562763486191,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000878834214974938,
      "loss": 2.1403,
      "step": 13173
    },
    {
      "epoch": 1.1334423126559408,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008788165665639909,
      "loss": 1.0756,
      "step": 13174
    },
    {
      "epoch": 1.1335283489632624,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008787989170450792,
      "loss": 0.9395,
      "step": 13175
    },
    {
      "epoch": 1.1336143852705842,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008787812664182541,
      "loss": 1.7273,
      "step": 13176
    },
    {
      "epoch": 1.1337004215779058,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008787636146835678,
      "loss": 1.4563,
      "step": 13177
    },
    {
      "epoch": 1.1337864578852275,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008787459618410713,
      "loss": 1.0011,
      "step": 13178
    },
    {
      "epoch": 1.1338724941925493,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008787283078908167,
      "loss": 1.3104,
      "step": 13179
    },
    {
      "epoch": 1.133958530499871,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008787106528328554,
      "loss": 1.0129,
      "step": 13180
    },
    {
      "epoch": 1.1340445668071926,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000878692996667239,
      "loss": 1.3212,
      "step": 13181
    },
    {
      "epoch": 1.1341306031145144,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008786753393940195,
      "loss": 0.9335,
      "step": 13182
    },
    {
      "epoch": 1.134216639421836,
      "grad_norm": 1.796875,
      "learning_rate": 0.000878657681013248,
      "loss": 2.2766,
      "step": 13183
    },
    {
      "epoch": 1.1343026757291577,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008786400215249765,
      "loss": 1.2109,
      "step": 13184
    },
    {
      "epoch": 1.1343887120364795,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008786223609292566,
      "loss": 1.3559,
      "step": 13185
    },
    {
      "epoch": 1.1344747483438011,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008786046992261399,
      "loss": 1.349,
      "step": 13186
    },
    {
      "epoch": 1.1345607846511228,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008785870364156782,
      "loss": 1.4809,
      "step": 13187
    },
    {
      "epoch": 1.1346468209584444,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008785693724979229,
      "loss": 1.1427,
      "step": 13188
    },
    {
      "epoch": 1.1347328572657662,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008785517074729259,
      "loss": 1.1534,
      "step": 13189
    },
    {
      "epoch": 1.1348188935730879,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008785340413407388,
      "loss": 1.3744,
      "step": 13190
    },
    {
      "epoch": 1.1349049298804095,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008785163741014132,
      "loss": 0.7656,
      "step": 13191
    },
    {
      "epoch": 1.1349909661877313,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008784987057550009,
      "loss": 1.0841,
      "step": 13192
    },
    {
      "epoch": 1.135077002495053,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008784810363015534,
      "loss": 1.2883,
      "step": 13193
    },
    {
      "epoch": 1.1351630388023746,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008784633657411225,
      "loss": 0.9502,
      "step": 13194
    },
    {
      "epoch": 1.1352490751096962,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00087844569407376,
      "loss": 0.9154,
      "step": 13195
    },
    {
      "epoch": 1.135335111417018,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008784280212995173,
      "loss": 1.4798,
      "step": 13196
    },
    {
      "epoch": 1.1354211477243397,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008784103474184462,
      "loss": 1.8644,
      "step": 13197
    },
    {
      "epoch": 1.1355071840316613,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008783926724305986,
      "loss": 0.9361,
      "step": 13198
    },
    {
      "epoch": 1.1355932203389831,
      "grad_norm": 1.109375,
      "learning_rate": 0.000878374996336026,
      "loss": 1.1968,
      "step": 13199
    },
    {
      "epoch": 1.1356792566463048,
      "grad_norm": 1.46875,
      "learning_rate": 0.00087835731913478,
      "loss": 1.2261,
      "step": 13200
    },
    {
      "epoch": 1.1357652929536264,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008783396408269124,
      "loss": 0.9063,
      "step": 13201
    },
    {
      "epoch": 1.135851329260948,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008783219614124751,
      "loss": 1.4746,
      "step": 13202
    },
    {
      "epoch": 1.1359373655682699,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008783042808915196,
      "loss": 1.1026,
      "step": 13203
    },
    {
      "epoch": 1.1360234018755915,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008782865992640975,
      "loss": 1.2558,
      "step": 13204
    },
    {
      "epoch": 1.136109438182913,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008782689165302608,
      "loss": 1.2034,
      "step": 13205
    },
    {
      "epoch": 1.136195474490235,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008782512326900612,
      "loss": 1.4684,
      "step": 13206
    },
    {
      "epoch": 1.1362815107975566,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008782335477435501,
      "loss": 0.9139,
      "step": 13207
    },
    {
      "epoch": 1.1363675471048782,
      "grad_norm": 1.375,
      "learning_rate": 0.0008782158616907795,
      "loss": 1.9888,
      "step": 13208
    },
    {
      "epoch": 1.1364535834122,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008781981745318011,
      "loss": 1.5183,
      "step": 13209
    },
    {
      "epoch": 1.1365396197195217,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008781804862666665,
      "loss": 1.3311,
      "step": 13210
    },
    {
      "epoch": 1.1366256560268433,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008781627968954275,
      "loss": 1.7592,
      "step": 13211
    },
    {
      "epoch": 1.1367116923341651,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000878145106418136,
      "loss": 1.4192,
      "step": 13212
    },
    {
      "epoch": 1.1367977286414868,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008781274148348437,
      "loss": 2.006,
      "step": 13213
    },
    {
      "epoch": 1.1368837649488084,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008781097221456022,
      "loss": 0.8724,
      "step": 13214
    },
    {
      "epoch": 1.13696980125613,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008780920283504632,
      "loss": 0.8142,
      "step": 13215
    },
    {
      "epoch": 1.1370558375634519,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008780743334494785,
      "loss": 1.6667,
      "step": 13216
    },
    {
      "epoch": 1.1371418738707735,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008780566374427001,
      "loss": 0.7918,
      "step": 13217
    },
    {
      "epoch": 1.137227910178095,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008780389403301794,
      "loss": 1.3628,
      "step": 13218
    },
    {
      "epoch": 1.137313946485417,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008780212421119685,
      "loss": 1.3879,
      "step": 13219
    },
    {
      "epoch": 1.1373999827927386,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008780035427881187,
      "loss": 1.1427,
      "step": 13220
    },
    {
      "epoch": 1.1374860191000602,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008779858423586823,
      "loss": 1.6423,
      "step": 13221
    },
    {
      "epoch": 1.1375720554073818,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008779681408237108,
      "loss": 1.0522,
      "step": 13222
    },
    {
      "epoch": 1.1376580917147037,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008779504381832559,
      "loss": 1.2038,
      "step": 13223
    },
    {
      "epoch": 1.1377441280220253,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008779327344373695,
      "loss": 1.0901,
      "step": 13224
    },
    {
      "epoch": 1.137830164329347,
      "grad_norm": 1.0,
      "learning_rate": 0.0008779150295861034,
      "loss": 0.9302,
      "step": 13225
    },
    {
      "epoch": 1.1379162006366688,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008778973236295094,
      "loss": 1.0491,
      "step": 13226
    },
    {
      "epoch": 1.1380022369439904,
      "grad_norm": 1.0625,
      "learning_rate": 0.000877879616567639,
      "loss": 1.6571,
      "step": 13227
    },
    {
      "epoch": 1.138088273251312,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008778619084005443,
      "loss": 1.7264,
      "step": 13228
    },
    {
      "epoch": 1.1381743095586336,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008778441991282771,
      "loss": 1.9126,
      "step": 13229
    },
    {
      "epoch": 1.1382603458659555,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008778264887508891,
      "loss": 1.1441,
      "step": 13230
    },
    {
      "epoch": 1.138346382173277,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008778087772684322,
      "loss": 1.2155,
      "step": 13231
    },
    {
      "epoch": 1.1384324184805987,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008777910646809579,
      "loss": 2.047,
      "step": 13232
    },
    {
      "epoch": 1.1385184547879206,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008777733509885183,
      "loss": 0.8368,
      "step": 13233
    },
    {
      "epoch": 1.1386044910952422,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008777556361911652,
      "loss": 1.3701,
      "step": 13234
    },
    {
      "epoch": 1.1386905274025638,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008777379202889502,
      "loss": 1.1307,
      "step": 13235
    },
    {
      "epoch": 1.1387765637098854,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008777202032819253,
      "loss": 0.9909,
      "step": 13236
    },
    {
      "epoch": 1.1388626000172073,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008777024851701423,
      "loss": 1.1896,
      "step": 13237
    },
    {
      "epoch": 1.138948636324529,
      "grad_norm": 1.75,
      "learning_rate": 0.0008776847659536529,
      "loss": 1.2343,
      "step": 13238
    },
    {
      "epoch": 1.1390346726318508,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008776670456325091,
      "loss": 1.0291,
      "step": 13239
    },
    {
      "epoch": 1.1391207089391724,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008776493242067626,
      "loss": 1.5512,
      "step": 13240
    },
    {
      "epoch": 1.139206745246494,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008776316016764652,
      "loss": 0.9094,
      "step": 13241
    },
    {
      "epoch": 1.1392927815538156,
      "grad_norm": 1.265625,
      "learning_rate": 0.000877613878041669,
      "loss": 1.2604,
      "step": 13242
    },
    {
      "epoch": 1.1393788178611375,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008775961533024255,
      "loss": 1.3192,
      "step": 13243
    },
    {
      "epoch": 1.139464854168459,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008775784274587869,
      "loss": 1.3154,
      "step": 13244
    },
    {
      "epoch": 1.1395508904757807,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008775607005108045,
      "loss": 1.2216,
      "step": 13245
    },
    {
      "epoch": 1.1396369267831026,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008775429724585306,
      "loss": 1.338,
      "step": 13246
    },
    {
      "epoch": 1.1397229630904242,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008775252433020171,
      "loss": 1.143,
      "step": 13247
    },
    {
      "epoch": 1.1398089993977458,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008775075130413155,
      "loss": 1.301,
      "step": 13248
    },
    {
      "epoch": 1.1398950357050674,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000877489781676478,
      "loss": 1.1814,
      "step": 13249
    },
    {
      "epoch": 1.1399810720123893,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008774720492075562,
      "loss": 1.6519,
      "step": 13250
    },
    {
      "epoch": 1.140067108319711,
      "grad_norm": 1.3125,
      "learning_rate": 0.000877454315634602,
      "loss": 1.0607,
      "step": 13251
    },
    {
      "epoch": 1.1401531446270325,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008774365809576674,
      "loss": 0.9445,
      "step": 13252
    },
    {
      "epoch": 1.1402391809343544,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008774188451768043,
      "loss": 0.9105,
      "step": 13253
    },
    {
      "epoch": 1.140325217241676,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008774011082920645,
      "loss": 0.8495,
      "step": 13254
    },
    {
      "epoch": 1.1404112535489976,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008773833703034998,
      "loss": 1.3046,
      "step": 13255
    },
    {
      "epoch": 1.1404972898563193,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008773656312111621,
      "loss": 0.9644,
      "step": 13256
    },
    {
      "epoch": 1.140583326163641,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008773478910151034,
      "loss": 1.2373,
      "step": 13257
    },
    {
      "epoch": 1.1406693624709627,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008773301497153754,
      "loss": 1.8872,
      "step": 13258
    },
    {
      "epoch": 1.1407553987782844,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008773124073120302,
      "loss": 1.2864,
      "step": 13259
    },
    {
      "epoch": 1.1408414350856062,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008772946638051194,
      "loss": 1.6598,
      "step": 13260
    },
    {
      "epoch": 1.1409274713929278,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008772769191946954,
      "loss": 1.4105,
      "step": 13261
    },
    {
      "epoch": 1.1410135077002495,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008772591734808095,
      "loss": 1.7598,
      "step": 13262
    },
    {
      "epoch": 1.141099544007571,
      "grad_norm": 1.046875,
      "learning_rate": 0.000877241426663514,
      "loss": 1.5066,
      "step": 13263
    },
    {
      "epoch": 1.141185580314893,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008772236787428606,
      "loss": 1.2844,
      "step": 13264
    },
    {
      "epoch": 1.1412716166222145,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008772059297189015,
      "loss": 1.5503,
      "step": 13265
    },
    {
      "epoch": 1.1413576529295362,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008771881795916885,
      "loss": 1.3339,
      "step": 13266
    },
    {
      "epoch": 1.141443689236858,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008771704283612731,
      "loss": 1.8059,
      "step": 13267
    },
    {
      "epoch": 1.1415297255441796,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008771526760277077,
      "loss": 2.039,
      "step": 13268
    },
    {
      "epoch": 1.1416157618515013,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008771349225910442,
      "loss": 1.1762,
      "step": 13269
    },
    {
      "epoch": 1.1417017981588231,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008771171680513343,
      "loss": 0.9647,
      "step": 13270
    },
    {
      "epoch": 1.1417878344661447,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00087709941240863,
      "loss": 2.2693,
      "step": 13271
    },
    {
      "epoch": 1.1418738707734664,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008770816556629833,
      "loss": 1.3377,
      "step": 13272
    },
    {
      "epoch": 1.1419599070807882,
      "grad_norm": 1.296875,
      "learning_rate": 0.000877063897814446,
      "loss": 2.0668,
      "step": 13273
    },
    {
      "epoch": 1.1420459433881098,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008770461388630703,
      "loss": 1.9368,
      "step": 13274
    },
    {
      "epoch": 1.1421319796954315,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008770283788089078,
      "loss": 1.9408,
      "step": 13275
    },
    {
      "epoch": 1.142218016002753,
      "grad_norm": 1.125,
      "learning_rate": 0.0008770106176520107,
      "loss": 2.1853,
      "step": 13276
    },
    {
      "epoch": 1.142304052310075,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008769928553924309,
      "loss": 1.3197,
      "step": 13277
    },
    {
      "epoch": 1.1423900886173965,
      "grad_norm": 1.5,
      "learning_rate": 0.0008769750920302205,
      "loss": 1.4917,
      "step": 13278
    },
    {
      "epoch": 1.1424761249247182,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008769573275654309,
      "loss": 1.3641,
      "step": 13279
    },
    {
      "epoch": 1.14256216123204,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008769395619981147,
      "loss": 0.8991,
      "step": 13280
    },
    {
      "epoch": 1.1426481975393616,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008769217953283234,
      "loss": 1.6493,
      "step": 13281
    },
    {
      "epoch": 1.1427342338466833,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008769040275561092,
      "loss": 1.3156,
      "step": 13282
    },
    {
      "epoch": 1.142820270154005,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000876886258681524,
      "loss": 2.6751,
      "step": 13283
    },
    {
      "epoch": 1.1429063064613267,
      "grad_norm": 1.5,
      "learning_rate": 0.00087686848870462,
      "loss": 1.0236,
      "step": 13284
    },
    {
      "epoch": 1.1429923427686484,
      "grad_norm": 1.0,
      "learning_rate": 0.0008768507176254487,
      "loss": 1.0559,
      "step": 13285
    },
    {
      "epoch": 1.14307837907597,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008768329454440627,
      "loss": 0.9991,
      "step": 13286
    },
    {
      "epoch": 1.1431644153832918,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008768151721605134,
      "loss": 1.6657,
      "step": 13287
    },
    {
      "epoch": 1.1432504516906135,
      "grad_norm": 79.0,
      "learning_rate": 0.0008767973977748529,
      "loss": 1.5389,
      "step": 13288
    },
    {
      "epoch": 1.143336487997935,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008767796222871333,
      "loss": 1.1823,
      "step": 13289
    },
    {
      "epoch": 1.1434225243052567,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008767618456974067,
      "loss": 1.6209,
      "step": 13290
    },
    {
      "epoch": 1.1435085606125786,
      "grad_norm": 1.375,
      "learning_rate": 0.000876744068005725,
      "loss": 2.0476,
      "step": 13291
    },
    {
      "epoch": 1.1435945969199002,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008767262892121401,
      "loss": 1.8397,
      "step": 13292
    },
    {
      "epoch": 1.1436806332272218,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008767085093167042,
      "loss": 1.37,
      "step": 13293
    },
    {
      "epoch": 1.1437666695345436,
      "grad_norm": 1.296875,
      "learning_rate": 0.000876690728319469,
      "loss": 1.5636,
      "step": 13294
    },
    {
      "epoch": 1.1438527058418653,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008766729462204867,
      "loss": 1.3878,
      "step": 13295
    },
    {
      "epoch": 1.143938742149187,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008766551630198094,
      "loss": 1.3056,
      "step": 13296
    },
    {
      "epoch": 1.1440247784565087,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008766373787174891,
      "loss": 1.2919,
      "step": 13297
    },
    {
      "epoch": 1.1441108147638304,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008766195933135776,
      "loss": 1.0428,
      "step": 13298
    },
    {
      "epoch": 1.144196851071152,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000876601806808127,
      "loss": 1.2549,
      "step": 13299
    },
    {
      "epoch": 1.1442828873784738,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008765840192011894,
      "loss": 0.8929,
      "step": 13300
    },
    {
      "epoch": 1.1443689236857955,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008765662304928167,
      "loss": 1.3758,
      "step": 13301
    },
    {
      "epoch": 1.144454959993117,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000876548440683061,
      "loss": 0.8943,
      "step": 13302
    },
    {
      "epoch": 1.1445409963004387,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008765306497719745,
      "loss": 0.9862,
      "step": 13303
    },
    {
      "epoch": 1.1446270326077606,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000876512857759609,
      "loss": 1.5485,
      "step": 13304
    },
    {
      "epoch": 1.1447130689150822,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008764950646460165,
      "loss": 1.2857,
      "step": 13305
    },
    {
      "epoch": 1.1447991052224038,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008764772704312494,
      "loss": 1.4897,
      "step": 13306
    },
    {
      "epoch": 1.1448851415297256,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008764594751153594,
      "loss": 0.6874,
      "step": 13307
    },
    {
      "epoch": 1.1449711778370473,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008764416786983986,
      "loss": 1.2611,
      "step": 13308
    },
    {
      "epoch": 1.145057214144369,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008764238811804193,
      "loss": 1.1881,
      "step": 13309
    },
    {
      "epoch": 1.1451432504516905,
      "grad_norm": 1.28125,
      "learning_rate": 0.000876406082561473,
      "loss": 1.106,
      "step": 13310
    },
    {
      "epoch": 1.1452292867590124,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008763882828416124,
      "loss": 1.2455,
      "step": 13311
    },
    {
      "epoch": 1.145315323066334,
      "grad_norm": 1.796875,
      "learning_rate": 0.0008763704820208893,
      "loss": 1.9387,
      "step": 13312
    },
    {
      "epoch": 1.1454013593736556,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008763526800993556,
      "loss": 1.525,
      "step": 13313
    },
    {
      "epoch": 1.1454873956809775,
      "grad_norm": 1.984375,
      "learning_rate": 0.0008763348770770635,
      "loss": 1.3319,
      "step": 13314
    },
    {
      "epoch": 1.145573431988299,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000876317072954065,
      "loss": 1.1606,
      "step": 13315
    },
    {
      "epoch": 1.1456594682956207,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008762992677304125,
      "loss": 1.2121,
      "step": 13316
    },
    {
      "epoch": 1.1457455046029423,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008762814614061578,
      "loss": 1.8777,
      "step": 13317
    },
    {
      "epoch": 1.1458315409102642,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008762636539813528,
      "loss": 0.9124,
      "step": 13318
    },
    {
      "epoch": 1.1459175772175858,
      "grad_norm": 1.375,
      "learning_rate": 0.0008762458454560499,
      "loss": 0.9737,
      "step": 13319
    },
    {
      "epoch": 1.1460036135249074,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008762280358303011,
      "loss": 1.3503,
      "step": 13320
    },
    {
      "epoch": 1.1460896498322293,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008762102251041584,
      "loss": 0.9022,
      "step": 13321
    },
    {
      "epoch": 1.146175686139551,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008761924132776739,
      "loss": 1.3935,
      "step": 13322
    },
    {
      "epoch": 1.1462617224468725,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008761746003508999,
      "loss": 0.6777,
      "step": 13323
    },
    {
      "epoch": 1.1463477587541941,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008761567863238884,
      "loss": 1.8225,
      "step": 13324
    },
    {
      "epoch": 1.146433795061516,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008761389711966914,
      "loss": 0.8616,
      "step": 13325
    },
    {
      "epoch": 1.1465198313688376,
      "grad_norm": 1.046875,
      "learning_rate": 0.000876121154969361,
      "loss": 0.8525,
      "step": 13326
    },
    {
      "epoch": 1.1466058676761595,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008761033376419494,
      "loss": 1.5129,
      "step": 13327
    },
    {
      "epoch": 1.146691903983481,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008760855192145087,
      "loss": 1.955,
      "step": 13328
    },
    {
      "epoch": 1.1467779402908027,
      "grad_norm": 1.21875,
      "learning_rate": 0.000876067699687091,
      "loss": 1.4232,
      "step": 13329
    },
    {
      "epoch": 1.1468639765981243,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008760498790597484,
      "loss": 1.0978,
      "step": 13330
    },
    {
      "epoch": 1.1469500129054462,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008760320573325331,
      "loss": 0.711,
      "step": 13331
    },
    {
      "epoch": 1.1470360492127678,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008760142345054971,
      "loss": 1.8288,
      "step": 13332
    },
    {
      "epoch": 1.1471220855200894,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008759964105786927,
      "loss": 0.8884,
      "step": 13333
    },
    {
      "epoch": 1.1472081218274113,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008759785855521718,
      "loss": 1.1139,
      "step": 13334
    },
    {
      "epoch": 1.147294158134733,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008759607594259867,
      "loss": 1.1957,
      "step": 13335
    },
    {
      "epoch": 1.1473801944420545,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008759429322001895,
      "loss": 1.9695,
      "step": 13336
    },
    {
      "epoch": 1.1474662307493761,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008759251038748324,
      "loss": 1.3566,
      "step": 13337
    },
    {
      "epoch": 1.147552267056698,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008759072744499674,
      "loss": 1.3007,
      "step": 13338
    },
    {
      "epoch": 1.1476383033640196,
      "grad_norm": 2.015625,
      "learning_rate": 0.0008758894439256468,
      "loss": 1.3507,
      "step": 13339
    },
    {
      "epoch": 1.1477243396713412,
      "grad_norm": 13.0,
      "learning_rate": 0.0008758716123019226,
      "loss": 1.1629,
      "step": 13340
    },
    {
      "epoch": 1.147810375978663,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000875853779578847,
      "loss": 0.8891,
      "step": 13341
    },
    {
      "epoch": 1.1478964122859847,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008758359457564724,
      "loss": 1.4689,
      "step": 13342
    },
    {
      "epoch": 1.1479824485933063,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008758181108348506,
      "loss": 1.9034,
      "step": 13343
    },
    {
      "epoch": 1.148068484900628,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008758002748140339,
      "loss": 1.0144,
      "step": 13344
    },
    {
      "epoch": 1.1481545212079498,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008757824376940746,
      "loss": 1.4488,
      "step": 13345
    },
    {
      "epoch": 1.1482405575152714,
      "grad_norm": 2.921875,
      "learning_rate": 0.0008757645994750248,
      "loss": 0.6234,
      "step": 13346
    },
    {
      "epoch": 1.148326593822593,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008757467601569365,
      "loss": 1.4917,
      "step": 13347
    },
    {
      "epoch": 1.148412630129915,
      "grad_norm": 1.484375,
      "learning_rate": 0.000875728919739862,
      "loss": 1.2913,
      "step": 13348
    },
    {
      "epoch": 1.1484986664372365,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008757110782238534,
      "loss": 2.1503,
      "step": 13349
    },
    {
      "epoch": 1.1485847027445582,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008756932356089631,
      "loss": 1.0318,
      "step": 13350
    },
    {
      "epoch": 1.1486707390518798,
      "grad_norm": 1.375,
      "learning_rate": 0.0008756753918952432,
      "loss": 1.5432,
      "step": 13351
    },
    {
      "epoch": 1.1487567753592016,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008756575470827457,
      "loss": 1.2906,
      "step": 13352
    },
    {
      "epoch": 1.1488428116665232,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008756397011715231,
      "loss": 1.6759,
      "step": 13353
    },
    {
      "epoch": 1.1489288479738449,
      "grad_norm": 1.859375,
      "learning_rate": 0.0008756218541616273,
      "loss": 2.0407,
      "step": 13354
    },
    {
      "epoch": 1.1490148842811667,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008756040060531106,
      "loss": 1.933,
      "step": 13355
    },
    {
      "epoch": 1.1491009205884883,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008755861568460253,
      "loss": 1.3966,
      "step": 13356
    },
    {
      "epoch": 1.14918695689581,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008755683065404236,
      "loss": 1.6233,
      "step": 13357
    },
    {
      "epoch": 1.1492729932031318,
      "grad_norm": 1.5,
      "learning_rate": 0.0008755504551363576,
      "loss": 0.9576,
      "step": 13358
    },
    {
      "epoch": 1.1493590295104534,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008755326026338796,
      "loss": 0.8906,
      "step": 13359
    },
    {
      "epoch": 1.149445065817775,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008755147490330417,
      "loss": 1.6883,
      "step": 13360
    },
    {
      "epoch": 1.149531102125097,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008754968943338962,
      "loss": 1.2726,
      "step": 13361
    },
    {
      "epoch": 1.1496171384324185,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008754790385364954,
      "loss": 1.381,
      "step": 13362
    },
    {
      "epoch": 1.1497031747397402,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008754611816408913,
      "loss": 2.0912,
      "step": 13363
    },
    {
      "epoch": 1.1497892110470618,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008754433236471364,
      "loss": 1.5954,
      "step": 13364
    },
    {
      "epoch": 1.1498752473543836,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008754254645552827,
      "loss": 0.8366,
      "step": 13365
    },
    {
      "epoch": 1.1499612836617052,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008754076043653825,
      "loss": 1.4776,
      "step": 13366
    },
    {
      "epoch": 1.1500473199690269,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008753897430774882,
      "loss": 1.1856,
      "step": 13367
    },
    {
      "epoch": 1.1501333562763487,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000875371880691652,
      "loss": 1.3913,
      "step": 13368
    },
    {
      "epoch": 1.1502193925836703,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008753540172079258,
      "loss": 1.3784,
      "step": 13369
    },
    {
      "epoch": 1.150305428890992,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008753361526263622,
      "loss": 1.9083,
      "step": 13370
    },
    {
      "epoch": 1.1503914651983136,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008753182869470133,
      "loss": 1.7313,
      "step": 13371
    },
    {
      "epoch": 1.1504775015056354,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008753004201699315,
      "loss": 0.5795,
      "step": 13372
    },
    {
      "epoch": 1.150563537812957,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008752825522951689,
      "loss": 1.1899,
      "step": 13373
    },
    {
      "epoch": 1.1506495741202787,
      "grad_norm": 1.71875,
      "learning_rate": 0.000875264683322778,
      "loss": 1.5119,
      "step": 13374
    },
    {
      "epoch": 1.1507356104276005,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008752468132528107,
      "loss": 0.9555,
      "step": 13375
    },
    {
      "epoch": 1.1508216467349222,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008752289420853194,
      "loss": 1.6055,
      "step": 13376
    },
    {
      "epoch": 1.1509076830422438,
      "grad_norm": 0.890625,
      "learning_rate": 0.0008752110698203566,
      "loss": 0.9435,
      "step": 13377
    },
    {
      "epoch": 1.1509937193495654,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008751931964579743,
      "loss": 1.2197,
      "step": 13378
    },
    {
      "epoch": 1.1510797556568872,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008751753219982248,
      "loss": 1.5453,
      "step": 13379
    },
    {
      "epoch": 1.1511657919642089,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008751574464411606,
      "loss": 1.0481,
      "step": 13380
    },
    {
      "epoch": 1.1512518282715305,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008751395697868337,
      "loss": 1.3271,
      "step": 13381
    },
    {
      "epoch": 1.1513378645788523,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008751216920352967,
      "loss": 1.4068,
      "step": 13382
    },
    {
      "epoch": 1.151423900886174,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008751038131866016,
      "loss": 0.9026,
      "step": 13383
    },
    {
      "epoch": 1.1515099371934956,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008750859332408008,
      "loss": 1.5561,
      "step": 13384
    },
    {
      "epoch": 1.1515959735008174,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008750680521979465,
      "loss": 1.5824,
      "step": 13385
    },
    {
      "epoch": 1.151682009808139,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008750501700580912,
      "loss": 1.2882,
      "step": 13386
    },
    {
      "epoch": 1.1517680461154607,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008750322868212871,
      "loss": 1.6838,
      "step": 13387
    },
    {
      "epoch": 1.1518540824227825,
      "grad_norm": 2.34375,
      "learning_rate": 0.0008750144024875866,
      "loss": 1.0932,
      "step": 13388
    },
    {
      "epoch": 1.1519401187301042,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008749965170570417,
      "loss": 1.033,
      "step": 13389
    },
    {
      "epoch": 1.1520261550374258,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008749786305297049,
      "loss": 0.8542,
      "step": 13390
    },
    {
      "epoch": 1.1521121913447474,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008749607429056288,
      "loss": 1.1788,
      "step": 13391
    },
    {
      "epoch": 1.1521982276520693,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008749428541848652,
      "loss": 1.5948,
      "step": 13392
    },
    {
      "epoch": 1.1522842639593909,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008749249643674669,
      "loss": 1.2094,
      "step": 13393
    },
    {
      "epoch": 1.1523703002667125,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008749070734534858,
      "loss": 1.5057,
      "step": 13394
    },
    {
      "epoch": 1.1524563365740343,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008748891814429745,
      "loss": 1.3655,
      "step": 13395
    },
    {
      "epoch": 1.152542372881356,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008748712883359851,
      "loss": 1.3536,
      "step": 13396
    },
    {
      "epoch": 1.1526284091886776,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008748533941325703,
      "loss": 1.696,
      "step": 13397
    },
    {
      "epoch": 1.1527144454959992,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008748354988327822,
      "loss": 1.1021,
      "step": 13398
    },
    {
      "epoch": 1.152800481803321,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008748176024366732,
      "loss": 1.2652,
      "step": 13399
    },
    {
      "epoch": 1.1528865181106427,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008747997049442955,
      "loss": 1.7682,
      "step": 13400
    },
    {
      "epoch": 1.1529725544179643,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008747818063557015,
      "loss": 2.1466,
      "step": 13401
    },
    {
      "epoch": 1.1530585907252862,
      "grad_norm": 2.203125,
      "learning_rate": 0.0008747639066709437,
      "loss": 0.6386,
      "step": 13402
    },
    {
      "epoch": 1.1531446270326078,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008747460058900745,
      "loss": 1.4827,
      "step": 13403
    },
    {
      "epoch": 1.1532306633399294,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008747281040131459,
      "loss": 1.3227,
      "step": 13404
    },
    {
      "epoch": 1.153316699647251,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008747102010402106,
      "loss": 0.9509,
      "step": 13405
    },
    {
      "epoch": 1.1534027359545729,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008746922969713208,
      "loss": 0.9374,
      "step": 13406
    },
    {
      "epoch": 1.1534887722618945,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008746743918065289,
      "loss": 1.192,
      "step": 13407
    },
    {
      "epoch": 1.1535748085692161,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008746564855458873,
      "loss": 1.2328,
      "step": 13408
    },
    {
      "epoch": 1.153660844876538,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008746385781894483,
      "loss": 1.2882,
      "step": 13409
    },
    {
      "epoch": 1.1537468811838596,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008746206697372642,
      "loss": 1.3664,
      "step": 13410
    },
    {
      "epoch": 1.1538329174911812,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008746027601893878,
      "loss": 0.9799,
      "step": 13411
    },
    {
      "epoch": 1.1539189537985028,
      "grad_norm": 1.546875,
      "learning_rate": 0.000874584849545871,
      "loss": 1.5737,
      "step": 13412
    },
    {
      "epoch": 1.1540049901058247,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008745669378067664,
      "loss": 2.0117,
      "step": 13413
    },
    {
      "epoch": 1.1540910264131463,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008745490249721264,
      "loss": 0.7937,
      "step": 13414
    },
    {
      "epoch": 1.1541770627204682,
      "grad_norm": 1.375,
      "learning_rate": 0.0008745311110420033,
      "loss": 1.5267,
      "step": 13415
    },
    {
      "epoch": 1.1542630990277898,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008745131960164496,
      "loss": 1.5196,
      "step": 13416
    },
    {
      "epoch": 1.1543491353351114,
      "grad_norm": 1.125,
      "learning_rate": 0.0008744952798955174,
      "loss": 1.6733,
      "step": 13417
    },
    {
      "epoch": 1.154435171642433,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008744773626792597,
      "loss": 1.3896,
      "step": 13418
    },
    {
      "epoch": 1.1545212079497549,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008744594443677283,
      "loss": 1.6616,
      "step": 13419
    },
    {
      "epoch": 1.1546072442570765,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000874441524960976,
      "loss": 1.5506,
      "step": 13420
    },
    {
      "epoch": 1.1546932805643981,
      "grad_norm": 1.234375,
      "learning_rate": 0.000874423604459055,
      "loss": 1.42,
      "step": 13421
    },
    {
      "epoch": 1.15477931687172,
      "grad_norm": 1.375,
      "learning_rate": 0.0008744056828620177,
      "loss": 1.1328,
      "step": 13422
    },
    {
      "epoch": 1.1548653531790416,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008743877601699167,
      "loss": 1.7974,
      "step": 13423
    },
    {
      "epoch": 1.1549513894863632,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008743698363828043,
      "loss": 1.2051,
      "step": 13424
    },
    {
      "epoch": 1.1550374257936848,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008743519115007331,
      "loss": 1.6354,
      "step": 13425
    },
    {
      "epoch": 1.1551234621010067,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000874333985523755,
      "loss": 1.1249,
      "step": 13426
    },
    {
      "epoch": 1.1552094984083283,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000874316058451923,
      "loss": 1.7308,
      "step": 13427
    },
    {
      "epoch": 1.15529553471565,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008742981302852893,
      "loss": 1.3997,
      "step": 13428
    },
    {
      "epoch": 1.1553815710229718,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008742802010239064,
      "loss": 0.7409,
      "step": 13429
    },
    {
      "epoch": 1.1554676073302934,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008742622706678267,
      "loss": 0.9228,
      "step": 13430
    },
    {
      "epoch": 1.155553643637615,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0008742443392171027,
      "loss": 1.4569,
      "step": 13431
    },
    {
      "epoch": 1.1556396799449367,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008742264066717866,
      "loss": 1.2854,
      "step": 13432
    },
    {
      "epoch": 1.1557257162522585,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008742084730319311,
      "loss": 1.4811,
      "step": 13433
    },
    {
      "epoch": 1.1558117525595801,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008741905382975887,
      "loss": 0.4073,
      "step": 13434
    },
    {
      "epoch": 1.1558977888669018,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008741726024688118,
      "loss": 1.5056,
      "step": 13435
    },
    {
      "epoch": 1.1559838251742236,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008741546655456527,
      "loss": 1.855,
      "step": 13436
    },
    {
      "epoch": 1.1560698614815452,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000874136727528164,
      "loss": 1.2689,
      "step": 13437
    },
    {
      "epoch": 1.1561558977888668,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008741187884163981,
      "loss": 1.3964,
      "step": 13438
    },
    {
      "epoch": 1.1562419340961885,
      "grad_norm": 3.3125,
      "learning_rate": 0.0008741008482104076,
      "loss": 1.7719,
      "step": 13439
    },
    {
      "epoch": 1.1563279704035103,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008740829069102448,
      "loss": 0.9514,
      "step": 13440
    },
    {
      "epoch": 1.156414006710832,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008740649645159623,
      "loss": 0.9413,
      "step": 13441
    },
    {
      "epoch": 1.1565000430181536,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008740470210276126,
      "loss": 1.0117,
      "step": 13442
    },
    {
      "epoch": 1.1565860793254754,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008740290764452478,
      "loss": 0.9846,
      "step": 13443
    },
    {
      "epoch": 1.156672115632797,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008740111307689211,
      "loss": 1.1513,
      "step": 13444
    },
    {
      "epoch": 1.1567581519401187,
      "grad_norm": 1.125,
      "learning_rate": 0.0008739931839986843,
      "loss": 1.3408,
      "step": 13445
    },
    {
      "epoch": 1.1568441882474405,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008739752361345903,
      "loss": 1.4322,
      "step": 13446
    },
    {
      "epoch": 1.1569302245547621,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008739572871766915,
      "loss": 2.2144,
      "step": 13447
    },
    {
      "epoch": 1.1570162608620838,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008739393371250402,
      "loss": 1.7104,
      "step": 13448
    },
    {
      "epoch": 1.1571022971694056,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008739213859796892,
      "loss": 1.8709,
      "step": 13449
    },
    {
      "epoch": 1.1571883334767272,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008739034337406908,
      "loss": 1.2976,
      "step": 13450
    },
    {
      "epoch": 1.1572743697840489,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008738854804080978,
      "loss": 1.4931,
      "step": 13451
    },
    {
      "epoch": 1.1573604060913705,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008738675259819622,
      "loss": 1.0832,
      "step": 13452
    },
    {
      "epoch": 1.1574464423986923,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008738495704623369,
      "loss": 0.8621,
      "step": 13453
    },
    {
      "epoch": 1.157532478706014,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008738316138492743,
      "loss": 1.3607,
      "step": 13454
    },
    {
      "epoch": 1.1576185150133356,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008738136561428269,
      "loss": 1.0195,
      "step": 13455
    },
    {
      "epoch": 1.1577045513206574,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008737956973430475,
      "loss": 1.6888,
      "step": 13456
    },
    {
      "epoch": 1.157790587627979,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008737777374499881,
      "loss": 1.3762,
      "step": 13457
    },
    {
      "epoch": 1.1578766239353007,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008737597764637016,
      "loss": 1.495,
      "step": 13458
    },
    {
      "epoch": 1.1579626602426223,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008737418143842405,
      "loss": 2.2706,
      "step": 13459
    },
    {
      "epoch": 1.1580486965499441,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008737238512116572,
      "loss": 1.2937,
      "step": 13460
    },
    {
      "epoch": 1.1581347328572658,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0008737058869460044,
      "loss": 1.0618,
      "step": 13461
    },
    {
      "epoch": 1.1582207691645874,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008736879215873346,
      "loss": 1.7671,
      "step": 13462
    },
    {
      "epoch": 1.1583068054719092,
      "grad_norm": 1.25,
      "learning_rate": 0.0008736699551357001,
      "loss": 1.6672,
      "step": 13463
    },
    {
      "epoch": 1.1583928417792309,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008736519875911538,
      "loss": 1.3531,
      "step": 13464
    },
    {
      "epoch": 1.1584788780865525,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008736340189537482,
      "loss": 1.7486,
      "step": 13465
    },
    {
      "epoch": 1.158564914393874,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008736160492235356,
      "loss": 1.4774,
      "step": 13466
    },
    {
      "epoch": 1.158650950701196,
      "grad_norm": 1.71875,
      "learning_rate": 0.0008735980784005689,
      "loss": 2.0662,
      "step": 13467
    },
    {
      "epoch": 1.1587369870085176,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008735801064849003,
      "loss": 1.8499,
      "step": 13468
    },
    {
      "epoch": 1.1588230233158392,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008735621334765826,
      "loss": 1.483,
      "step": 13469
    },
    {
      "epoch": 1.158909059623161,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008735441593756683,
      "loss": 1.5567,
      "step": 13470
    },
    {
      "epoch": 1.1589950959304827,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00087352618418221,
      "loss": 0.9394,
      "step": 13471
    },
    {
      "epoch": 1.1590811322378043,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008735082078962602,
      "loss": 1.407,
      "step": 13472
    },
    {
      "epoch": 1.1591671685451261,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008734902305178716,
      "loss": 1.4276,
      "step": 13473
    },
    {
      "epoch": 1.1592532048524478,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008734722520470968,
      "loss": 2.424,
      "step": 13474
    },
    {
      "epoch": 1.1593392411597694,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008734542724839881,
      "loss": 0.6661,
      "step": 13475
    },
    {
      "epoch": 1.1594252774670912,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008734362918285984,
      "loss": 1.9508,
      "step": 13476
    },
    {
      "epoch": 1.1595113137744129,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008734183100809803,
      "loss": 1.4341,
      "step": 13477
    },
    {
      "epoch": 1.1595973500817345,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008734003272411861,
      "loss": 1.4197,
      "step": 13478
    },
    {
      "epoch": 1.159683386389056,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008733823433092685,
      "loss": 1.6021,
      "step": 13479
    },
    {
      "epoch": 1.159769422696378,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008733643582852804,
      "loss": 1.6182,
      "step": 13480
    },
    {
      "epoch": 1.1598554590036996,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008733463721692739,
      "loss": 2.0252,
      "step": 13481
    },
    {
      "epoch": 1.1599414953110212,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000873328384961302,
      "loss": 0.9273,
      "step": 13482
    },
    {
      "epoch": 1.160027531618343,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008733103966614173,
      "loss": 1.2281,
      "step": 13483
    },
    {
      "epoch": 1.1601135679256647,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008732924072696722,
      "loss": 1.6889,
      "step": 13484
    },
    {
      "epoch": 1.1601996042329863,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008732744167861193,
      "loss": 1.9328,
      "step": 13485
    },
    {
      "epoch": 1.160285640540308,
      "grad_norm": 0.875,
      "learning_rate": 0.0008732564252108115,
      "loss": 1.4221,
      "step": 13486
    },
    {
      "epoch": 1.1603716768476298,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008732384325438012,
      "loss": 1.044,
      "step": 13487
    },
    {
      "epoch": 1.1604577131549514,
      "grad_norm": 1.453125,
      "learning_rate": 0.000873220438785141,
      "loss": 1.1578,
      "step": 13488
    },
    {
      "epoch": 1.160543749462273,
      "grad_norm": 1.375,
      "learning_rate": 0.0008732024439348835,
      "loss": 0.935,
      "step": 13489
    },
    {
      "epoch": 1.1606297857695949,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008731844479930817,
      "loss": 1.4612,
      "step": 13490
    },
    {
      "epoch": 1.1607158220769165,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008731664509597878,
      "loss": 1.1247,
      "step": 13491
    },
    {
      "epoch": 1.160801858384238,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008731484528350549,
      "loss": 0.989,
      "step": 13492
    },
    {
      "epoch": 1.1608878946915597,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008731304536189349,
      "loss": 0.8474,
      "step": 13493
    },
    {
      "epoch": 1.1609739309988816,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008731124533114811,
      "loss": 1.3204,
      "step": 13494
    },
    {
      "epoch": 1.1610599673062032,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008730944519127461,
      "loss": 1.1424,
      "step": 13495
    },
    {
      "epoch": 1.1611460036135248,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008730764494227822,
      "loss": 0.932,
      "step": 13496
    },
    {
      "epoch": 1.1612320399208467,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008730584458416423,
      "loss": 1.2228,
      "step": 13497
    },
    {
      "epoch": 1.1613180762281683,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008730404411693791,
      "loss": 1.1068,
      "step": 13498
    },
    {
      "epoch": 1.16140411253549,
      "grad_norm": 1.265625,
      "learning_rate": 0.000873022435406045,
      "loss": 0.8066,
      "step": 13499
    },
    {
      "epoch": 1.1614901488428115,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008730044285516929,
      "loss": 0.6994,
      "step": 13500
    },
    {
      "epoch": 1.1615761851501334,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008729864206063753,
      "loss": 2.2995,
      "step": 13501
    },
    {
      "epoch": 1.161662221457455,
      "grad_norm": 1.3125,
      "learning_rate": 0.000872968411570145,
      "loss": 1.2996,
      "step": 13502
    },
    {
      "epoch": 1.1617482577647769,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008729504014430547,
      "loss": 1.7795,
      "step": 13503
    },
    {
      "epoch": 1.1618342940720985,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000872932390225157,
      "loss": 1.0937,
      "step": 13504
    },
    {
      "epoch": 1.16192033037942,
      "grad_norm": 1.625,
      "learning_rate": 0.0008729143779165045,
      "loss": 1.6168,
      "step": 13505
    },
    {
      "epoch": 1.1620063666867417,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00087289636451715,
      "loss": 1.5547,
      "step": 13506
    },
    {
      "epoch": 1.1620924029940636,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008728783500271463,
      "loss": 0.8981,
      "step": 13507
    },
    {
      "epoch": 1.1621784393013852,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008728603344465458,
      "loss": 2.1709,
      "step": 13508
    },
    {
      "epoch": 1.1622644756087068,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008728423177754013,
      "loss": 1.6064,
      "step": 13509
    },
    {
      "epoch": 1.1623505119160287,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008728243000137657,
      "loss": 0.8435,
      "step": 13510
    },
    {
      "epoch": 1.1624365482233503,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008728062811616914,
      "loss": 1.72,
      "step": 13511
    },
    {
      "epoch": 1.162522584530672,
      "grad_norm": 2.28125,
      "learning_rate": 0.0008727882612192312,
      "loss": 1.7698,
      "step": 13512
    },
    {
      "epoch": 1.1626086208379935,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008727702401864382,
      "loss": 1.2643,
      "step": 13513
    },
    {
      "epoch": 1.1626946571453154,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008727522180633642,
      "loss": 1.7638,
      "step": 13514
    },
    {
      "epoch": 1.162780693452637,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008727341948500629,
      "loss": 1.15,
      "step": 13515
    },
    {
      "epoch": 1.1628667297599586,
      "grad_norm": 1.375,
      "learning_rate": 0.0008727161705465863,
      "loss": 1.586,
      "step": 13516
    },
    {
      "epoch": 1.1629527660672805,
      "grad_norm": 1.625,
      "learning_rate": 0.0008726981451529876,
      "loss": 1.0461,
      "step": 13517
    },
    {
      "epoch": 1.163038802374602,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008726801186693192,
      "loss": 1.7474,
      "step": 13518
    },
    {
      "epoch": 1.1631248386819237,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008726620910956339,
      "loss": 1.6383,
      "step": 13519
    },
    {
      "epoch": 1.1632108749892454,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008726440624319845,
      "loss": 1.2999,
      "step": 13520
    },
    {
      "epoch": 1.1632969112965672,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008726260326784237,
      "loss": 1.7853,
      "step": 13521
    },
    {
      "epoch": 1.1633829476038888,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008726080018350042,
      "loss": 0.8128,
      "step": 13522
    },
    {
      "epoch": 1.1634689839112105,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008725899699017788,
      "loss": 0.8926,
      "step": 13523
    },
    {
      "epoch": 1.1635550202185323,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008725719368788002,
      "loss": 1.0081,
      "step": 13524
    },
    {
      "epoch": 1.163641056525854,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008725539027661212,
      "loss": 1.5107,
      "step": 13525
    },
    {
      "epoch": 1.1637270928331755,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008725358675637944,
      "loss": 0.8632,
      "step": 13526
    },
    {
      "epoch": 1.1638131291404972,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008725178312718726,
      "loss": 1.9167,
      "step": 13527
    },
    {
      "epoch": 1.163899165447819,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008724997938904087,
      "loss": 1.5676,
      "step": 13528
    },
    {
      "epoch": 1.1639852017551406,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008724817554194553,
      "loss": 1.4316,
      "step": 13529
    },
    {
      "epoch": 1.1640712380624623,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008724637158590651,
      "loss": 1.0533,
      "step": 13530
    },
    {
      "epoch": 1.1641572743697841,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008724456752092912,
      "loss": 1.5165,
      "step": 13531
    },
    {
      "epoch": 1.1642433106771057,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000872427633470186,
      "loss": 1.9142,
      "step": 13532
    },
    {
      "epoch": 1.1643293469844274,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008724095906418024,
      "loss": 1.1722,
      "step": 13533
    },
    {
      "epoch": 1.1644153832917492,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008723915467241932,
      "loss": 0.6981,
      "step": 13534
    },
    {
      "epoch": 1.1645014195990708,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008723735017174113,
      "loss": 1.6534,
      "step": 13535
    },
    {
      "epoch": 1.1645874559063925,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000872355455621509,
      "loss": 1.5675,
      "step": 13536
    },
    {
      "epoch": 1.1646734922137143,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008723374084365396,
      "loss": 1.0075,
      "step": 13537
    },
    {
      "epoch": 1.164759528521036,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008723193601625558,
      "loss": 1.1414,
      "step": 13538
    },
    {
      "epoch": 1.1648455648283575,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008723013107996101,
      "loss": 1.7044,
      "step": 13539
    },
    {
      "epoch": 1.1649316011356792,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008722832603477556,
      "loss": 0.8838,
      "step": 13540
    },
    {
      "epoch": 1.165017637443001,
      "grad_norm": 1.34375,
      "learning_rate": 0.000872265208807045,
      "loss": 0.7892,
      "step": 13541
    },
    {
      "epoch": 1.1651036737503226,
      "grad_norm": 1.125,
      "learning_rate": 0.0008722471561775309,
      "loss": 1.907,
      "step": 13542
    },
    {
      "epoch": 1.1651897100576443,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008722291024592663,
      "loss": 1.1667,
      "step": 13543
    },
    {
      "epoch": 1.1652757463649661,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008722110476523042,
      "loss": 0.7541,
      "step": 13544
    },
    {
      "epoch": 1.1653617826722877,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008721929917566969,
      "loss": 1.8973,
      "step": 13545
    },
    {
      "epoch": 1.1654478189796094,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008721749347724977,
      "loss": 0.5902,
      "step": 13546
    },
    {
      "epoch": 1.165533855286931,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008721568766997589,
      "loss": 1.0697,
      "step": 13547
    },
    {
      "epoch": 1.1656198915942528,
      "grad_norm": 1.125,
      "learning_rate": 0.0008721388175385339,
      "loss": 1.3686,
      "step": 13548
    },
    {
      "epoch": 1.1657059279015745,
      "grad_norm": 1.0,
      "learning_rate": 0.0008721207572888752,
      "loss": 0.946,
      "step": 13549
    },
    {
      "epoch": 1.165791964208896,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008721026959508356,
      "loss": 0.624,
      "step": 13550
    },
    {
      "epoch": 1.165878000516218,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008720846335244679,
      "loss": 2.1407,
      "step": 13551
    },
    {
      "epoch": 1.1659640368235396,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008720665700098252,
      "loss": 1.9078,
      "step": 13552
    },
    {
      "epoch": 1.1660500731308612,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00087204850540696,
      "loss": 1.8056,
      "step": 13553
    },
    {
      "epoch": 1.1661361094381828,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008720304397159253,
      "loss": 1.6676,
      "step": 13554
    },
    {
      "epoch": 1.1662221457455046,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008720123729367739,
      "loss": 1.7723,
      "step": 13555
    },
    {
      "epoch": 1.1663081820528263,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008719943050695587,
      "loss": 1.5034,
      "step": 13556
    },
    {
      "epoch": 1.166394218360148,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008719762361143326,
      "loss": 1.3692,
      "step": 13557
    },
    {
      "epoch": 1.1664802546674697,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008719581660711482,
      "loss": 1.0446,
      "step": 13558
    },
    {
      "epoch": 1.1665662909747914,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008719400949400586,
      "loss": 1.4583,
      "step": 13559
    },
    {
      "epoch": 1.166652327282113,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008719220227211167,
      "loss": 1.408,
      "step": 13560
    },
    {
      "epoch": 1.1667383635894348,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008719039494143749,
      "loss": 1.6488,
      "step": 13561
    },
    {
      "epoch": 1.1668243998967565,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008718858750198866,
      "loss": 1.5103,
      "step": 13562
    },
    {
      "epoch": 1.166910436204078,
      "grad_norm": 1.625,
      "learning_rate": 0.0008718677995377043,
      "loss": 2.0272,
      "step": 13563
    },
    {
      "epoch": 1.1669964725114,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008718497229678812,
      "loss": 1.2487,
      "step": 13564
    },
    {
      "epoch": 1.1670825088187216,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008718316453104698,
      "loss": 1.6368,
      "step": 13565
    },
    {
      "epoch": 1.1671685451260432,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008718135665655232,
      "loss": 1.401,
      "step": 13566
    },
    {
      "epoch": 1.1672545814333648,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008717954867330941,
      "loss": 1.4365,
      "step": 13567
    },
    {
      "epoch": 1.1673406177406866,
      "grad_norm": 2.125,
      "learning_rate": 0.0008717774058132357,
      "loss": 1.5801,
      "step": 13568
    },
    {
      "epoch": 1.1674266540480083,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008717593238060007,
      "loss": 1.0634,
      "step": 13569
    },
    {
      "epoch": 1.16751269035533,
      "grad_norm": 1.109375,
      "learning_rate": 0.000871741240711442,
      "loss": 2.1,
      "step": 13570
    },
    {
      "epoch": 1.1675987266626517,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008717231565296124,
      "loss": 1.877,
      "step": 13571
    },
    {
      "epoch": 1.1676847629699734,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008717050712605649,
      "loss": 0.7865,
      "step": 13572
    },
    {
      "epoch": 1.167770799277295,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008716869849043523,
      "loss": 1.1983,
      "step": 13573
    },
    {
      "epoch": 1.1678568355846166,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008716688974610276,
      "loss": 1.2578,
      "step": 13574
    },
    {
      "epoch": 1.1679428718919385,
      "grad_norm": 1.125,
      "learning_rate": 0.0008716508089306437,
      "loss": 0.9939,
      "step": 13575
    },
    {
      "epoch": 1.16802890819926,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008716327193132534,
      "loss": 1.6892,
      "step": 13576
    },
    {
      "epoch": 1.1681149445065817,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008716146286089098,
      "loss": 1.4609,
      "step": 13577
    },
    {
      "epoch": 1.1682009808139036,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008715965368176655,
      "loss": 1.3167,
      "step": 13578
    },
    {
      "epoch": 1.1682870171212252,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008715784439395736,
      "loss": 2.2005,
      "step": 13579
    },
    {
      "epoch": 1.1683730534285468,
      "grad_norm": 1.125,
      "learning_rate": 0.0008715603499746871,
      "loss": 0.7462,
      "step": 13580
    },
    {
      "epoch": 1.1684590897358684,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008715422549230589,
      "loss": 2.0286,
      "step": 13581
    },
    {
      "epoch": 1.1685451260431903,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008715241587847418,
      "loss": 1.1782,
      "step": 13582
    },
    {
      "epoch": 1.168631162350512,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008715060615597887,
      "loss": 1.2742,
      "step": 13583
    },
    {
      "epoch": 1.1687171986578335,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008714879632482529,
      "loss": 1.2145,
      "step": 13584
    },
    {
      "epoch": 1.1688032349651554,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008714698638501867,
      "loss": 1.1045,
      "step": 13585
    },
    {
      "epoch": 1.168889271272477,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008714517633656437,
      "loss": 1.4262,
      "step": 13586
    },
    {
      "epoch": 1.1689753075797986,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008714336617946763,
      "loss": 1.0719,
      "step": 13587
    },
    {
      "epoch": 1.1690613438871202,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008714155591373378,
      "loss": 1.6045,
      "step": 13588
    },
    {
      "epoch": 1.169147380194442,
      "grad_norm": 1.125,
      "learning_rate": 0.0008713974553936811,
      "loss": 1.2003,
      "step": 13589
    },
    {
      "epoch": 1.1692334165017637,
      "grad_norm": 1.78125,
      "learning_rate": 0.0008713793505637589,
      "loss": 0.8005,
      "step": 13590
    },
    {
      "epoch": 1.1693194528090856,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008713612446476245,
      "loss": 1.4843,
      "step": 13591
    },
    {
      "epoch": 1.1694054891164072,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008713431376453306,
      "loss": 1.1085,
      "step": 13592
    },
    {
      "epoch": 1.1694915254237288,
      "grad_norm": 5.78125,
      "learning_rate": 0.0008713250295569302,
      "loss": 1.4398,
      "step": 13593
    },
    {
      "epoch": 1.1695775617310504,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008713069203824764,
      "loss": 1.3696,
      "step": 13594
    },
    {
      "epoch": 1.1696635980383723,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000871288810122022,
      "loss": 0.9838,
      "step": 13595
    },
    {
      "epoch": 1.169749634345694,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00087127069877562,
      "loss": 1.7048,
      "step": 13596
    },
    {
      "epoch": 1.1698356706530155,
      "grad_norm": 2.625,
      "learning_rate": 0.0008712525863433236,
      "loss": 1.1461,
      "step": 13597
    },
    {
      "epoch": 1.1699217069603374,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008712344728251855,
      "loss": 1.4036,
      "step": 13598
    },
    {
      "epoch": 1.170007743267659,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008712163582212587,
      "loss": 1.6685,
      "step": 13599
    },
    {
      "epoch": 1.1700937795749806,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008711982425315963,
      "loss": 1.2346,
      "step": 13600
    },
    {
      "epoch": 1.1701798158823022,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008711801257562513,
      "loss": 1.6435,
      "step": 13601
    },
    {
      "epoch": 1.170265852189624,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008711620078952764,
      "loss": 2.1806,
      "step": 13602
    },
    {
      "epoch": 1.1703518884969457,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008711438889487251,
      "loss": 1.281,
      "step": 13603
    },
    {
      "epoch": 1.1704379248042673,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008711257689166498,
      "loss": 0.6685,
      "step": 13604
    },
    {
      "epoch": 1.1705239611115892,
      "grad_norm": 1.40625,
      "learning_rate": 0.000871107647799104,
      "loss": 2.072,
      "step": 13605
    },
    {
      "epoch": 1.1706099974189108,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008710895255961406,
      "loss": 1.0657,
      "step": 13606
    },
    {
      "epoch": 1.1706960337262324,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008710714023078123,
      "loss": 1.6201,
      "step": 13607
    },
    {
      "epoch": 1.170782070033554,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008710532779341725,
      "loss": 0.9211,
      "step": 13608
    },
    {
      "epoch": 1.170868106340876,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000871035152475274,
      "loss": 1.219,
      "step": 13609
    },
    {
      "epoch": 1.1709541426481975,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008710170259311697,
      "loss": 1.3165,
      "step": 13610
    },
    {
      "epoch": 1.1710401789555192,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008709988983019128,
      "loss": 1.0592,
      "step": 13611
    },
    {
      "epoch": 1.171126215262841,
      "grad_norm": 1.75,
      "learning_rate": 0.0008709807695875565,
      "loss": 1.4111,
      "step": 13612
    },
    {
      "epoch": 1.1712122515701626,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008709626397881534,
      "loss": 1.6215,
      "step": 13613
    },
    {
      "epoch": 1.1712982878774842,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008709445089037566,
      "loss": 1.0424,
      "step": 13614
    },
    {
      "epoch": 1.1713843241848059,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008709263769344195,
      "loss": 0.7456,
      "step": 13615
    },
    {
      "epoch": 1.1714703604921277,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008709082438801948,
      "loss": 1.1536,
      "step": 13616
    },
    {
      "epoch": 1.1715563967994493,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008708901097411356,
      "loss": 0.9364,
      "step": 13617
    },
    {
      "epoch": 1.171642433106771,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008708719745172949,
      "loss": 1.5661,
      "step": 13618
    },
    {
      "epoch": 1.1717284694140928,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008708538382087259,
      "loss": 0.7862,
      "step": 13619
    },
    {
      "epoch": 1.1718145057214144,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008708357008154815,
      "loss": 1.4146,
      "step": 13620
    },
    {
      "epoch": 1.171900542028736,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008708175623376149,
      "loss": 1.2787,
      "step": 13621
    },
    {
      "epoch": 1.171986578336058,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000870799422775179,
      "loss": 1.0923,
      "step": 13622
    },
    {
      "epoch": 1.1720726146433795,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008707812821282269,
      "loss": 2.1992,
      "step": 13623
    },
    {
      "epoch": 1.1721586509507012,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008707631403968117,
      "loss": 1.5432,
      "step": 13624
    },
    {
      "epoch": 1.172244687258023,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008707449975809864,
      "loss": 1.7971,
      "step": 13625
    },
    {
      "epoch": 1.1723307235653446,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008707268536808039,
      "loss": 0.9852,
      "step": 13626
    },
    {
      "epoch": 1.1724167598726662,
      "grad_norm": 1.375,
      "learning_rate": 0.0008707087086963177,
      "loss": 1.9984,
      "step": 13627
    },
    {
      "epoch": 1.1725027961799879,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008706905626275805,
      "loss": 2.0006,
      "step": 13628
    },
    {
      "epoch": 1.1725888324873097,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008706724154746456,
      "loss": 1.1442,
      "step": 13629
    },
    {
      "epoch": 1.1726748687946313,
      "grad_norm": 1.09375,
      "learning_rate": 0.000870654267237566,
      "loss": 1.1642,
      "step": 13630
    },
    {
      "epoch": 1.172760905101953,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008706361179163947,
      "loss": 1.5001,
      "step": 13631
    },
    {
      "epoch": 1.1728469414092748,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008706179675111847,
      "loss": 1.4344,
      "step": 13632
    },
    {
      "epoch": 1.1729329777165964,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008705998160219894,
      "loss": 1.1327,
      "step": 13633
    },
    {
      "epoch": 1.173019014023918,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008705816634488616,
      "loss": 1.1965,
      "step": 13634
    },
    {
      "epoch": 1.1731050503312397,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008705635097918545,
      "loss": 1.0021,
      "step": 13635
    },
    {
      "epoch": 1.1731910866385615,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008705453550510214,
      "loss": 1.5919,
      "step": 13636
    },
    {
      "epoch": 1.1732771229458832,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008705271992264149,
      "loss": 1.4025,
      "step": 13637
    },
    {
      "epoch": 1.1733631592532048,
      "grad_norm": 2.03125,
      "learning_rate": 0.0008705090423180887,
      "loss": 1.4914,
      "step": 13638
    },
    {
      "epoch": 1.1734491955605266,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008704908843260954,
      "loss": 0.9562,
      "step": 13639
    },
    {
      "epoch": 1.1735352318678482,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008704727252504885,
      "loss": 1.951,
      "step": 13640
    },
    {
      "epoch": 1.1736212681751699,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000870454565091321,
      "loss": 1.059,
      "step": 13641
    },
    {
      "epoch": 1.1737073044824915,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008704364038486456,
      "loss": 1.3565,
      "step": 13642
    },
    {
      "epoch": 1.1737933407898133,
      "grad_norm": 1.171875,
      "learning_rate": 0.000870418241522516,
      "loss": 1.4059,
      "step": 13643
    },
    {
      "epoch": 1.173879377097135,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008704000781129851,
      "loss": 2.3177,
      "step": 13644
    },
    {
      "epoch": 1.1739654134044566,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008703819136201059,
      "loss": 1.0473,
      "step": 13645
    },
    {
      "epoch": 1.1740514497117784,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008703637480439317,
      "loss": 0.9283,
      "step": 13646
    },
    {
      "epoch": 1.1741374860191,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008703455813845156,
      "loss": 0.9112,
      "step": 13647
    },
    {
      "epoch": 1.1742235223264217,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008703274136419106,
      "loss": 0.9611,
      "step": 13648
    },
    {
      "epoch": 1.1743095586337435,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00087030924481617,
      "loss": 1.8096,
      "step": 13649
    },
    {
      "epoch": 1.1743955949410652,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000870291074907347,
      "loss": 0.8237,
      "step": 13650
    },
    {
      "epoch": 1.1744816312483868,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008702729039154944,
      "loss": 1.4754,
      "step": 13651
    },
    {
      "epoch": 1.1745676675557086,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008702547318406658,
      "loss": 1.3117,
      "step": 13652
    },
    {
      "epoch": 1.1746537038630303,
      "grad_norm": 1.3125,
      "learning_rate": 0.000870236558682914,
      "loss": 1.4683,
      "step": 13653
    },
    {
      "epoch": 1.1747397401703519,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008702183844422922,
      "loss": 0.9487,
      "step": 13654
    },
    {
      "epoch": 1.1748257764776735,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008702002091188536,
      "loss": 1.3166,
      "step": 13655
    },
    {
      "epoch": 1.1749118127849953,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008701820327126515,
      "loss": 1.1333,
      "step": 13656
    },
    {
      "epoch": 1.174997849092317,
      "grad_norm": 1.265625,
      "learning_rate": 0.000870163855223739,
      "loss": 1.7874,
      "step": 13657
    },
    {
      "epoch": 1.1750838853996386,
      "grad_norm": 1.25,
      "learning_rate": 0.0008701456766521692,
      "loss": 2.1282,
      "step": 13658
    },
    {
      "epoch": 1.1751699217069604,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008701274969979951,
      "loss": 1.6103,
      "step": 13659
    },
    {
      "epoch": 1.175255958014282,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008701093162612702,
      "loss": 1.3301,
      "step": 13660
    },
    {
      "epoch": 1.1753419943216037,
      "grad_norm": 1.0,
      "learning_rate": 0.0008700911344420476,
      "loss": 0.647,
      "step": 13661
    },
    {
      "epoch": 1.1754280306289253,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008700729515403803,
      "loss": 1.6571,
      "step": 13662
    },
    {
      "epoch": 1.1755140669362472,
      "grad_norm": 1.375,
      "learning_rate": 0.0008700547675563217,
      "loss": 1.9771,
      "step": 13663
    },
    {
      "epoch": 1.1756001032435688,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008700365824899248,
      "loss": 1.0743,
      "step": 13664
    },
    {
      "epoch": 1.1756861395508904,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008700183963412428,
      "loss": 1.3632,
      "step": 13665
    },
    {
      "epoch": 1.1757721758582123,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000870000209110329,
      "loss": 1.0532,
      "step": 13666
    },
    {
      "epoch": 1.1758582121655339,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008699820207972367,
      "loss": 1.3506,
      "step": 13667
    },
    {
      "epoch": 1.1759442484728555,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008699638314020187,
      "loss": 1.0189,
      "step": 13668
    },
    {
      "epoch": 1.1760302847801771,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008699456409247287,
      "loss": 1.2495,
      "step": 13669
    },
    {
      "epoch": 1.176116321087499,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008699274493654195,
      "loss": 1.1299,
      "step": 13670
    },
    {
      "epoch": 1.1762023573948206,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008699092567241444,
      "loss": 1.5485,
      "step": 13671
    },
    {
      "epoch": 1.1762883937021422,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008698910630009568,
      "loss": 1.6782,
      "step": 13672
    },
    {
      "epoch": 1.176374430009464,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008698728681959096,
      "loss": 0.5931,
      "step": 13673
    },
    {
      "epoch": 1.1764604663167857,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008698546723090564,
      "loss": 1.4194,
      "step": 13674
    },
    {
      "epoch": 1.1765465026241073,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008698364753404502,
      "loss": 1.3453,
      "step": 13675
    },
    {
      "epoch": 1.176632538931429,
      "grad_norm": 1.125,
      "learning_rate": 0.0008698182772901441,
      "loss": 1.3686,
      "step": 13676
    },
    {
      "epoch": 1.1767185752387508,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008698000781581915,
      "loss": 1.5365,
      "step": 13677
    },
    {
      "epoch": 1.1768046115460724,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008697818779446456,
      "loss": 1.0102,
      "step": 13678
    },
    {
      "epoch": 1.1768906478533943,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008697636766495598,
      "loss": 0.761,
      "step": 13679
    },
    {
      "epoch": 1.1769766841607159,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008697454742729869,
      "loss": 1.2315,
      "step": 13680
    },
    {
      "epoch": 1.1770627204680375,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008697272708149806,
      "loss": 1.2211,
      "step": 13681
    },
    {
      "epoch": 1.1771487567753591,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008697090662755937,
      "loss": 1.9849,
      "step": 13682
    },
    {
      "epoch": 1.177234793082681,
      "grad_norm": 1.1875,
      "learning_rate": 0.00086969086065488,
      "loss": 1.714,
      "step": 13683
    },
    {
      "epoch": 1.1773208293900026,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008696726539528923,
      "loss": 1.323,
      "step": 13684
    },
    {
      "epoch": 1.1774068656973242,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000869654446169684,
      "loss": 1.4729,
      "step": 13685
    },
    {
      "epoch": 1.177492902004646,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008696362373053081,
      "loss": 0.9117,
      "step": 13686
    },
    {
      "epoch": 1.1775789383119677,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008696180273598185,
      "loss": 1.5017,
      "step": 13687
    },
    {
      "epoch": 1.1776649746192893,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008695998163332678,
      "loss": 1.5726,
      "step": 13688
    },
    {
      "epoch": 1.177751010926611,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008695816042257095,
      "loss": 1.1226,
      "step": 13689
    },
    {
      "epoch": 1.1778370472339328,
      "grad_norm": 2.21875,
      "learning_rate": 0.000869563391037197,
      "loss": 1.4162,
      "step": 13690
    },
    {
      "epoch": 1.1779230835412544,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008695451767677835,
      "loss": 1.219,
      "step": 13691
    },
    {
      "epoch": 1.178009119848576,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0008695269614175221,
      "loss": 1.5921,
      "step": 13692
    },
    {
      "epoch": 1.1780951561558979,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008695087449864662,
      "loss": 1.0545,
      "step": 13693
    },
    {
      "epoch": 1.1781811924632195,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008694905274746692,
      "loss": 1.444,
      "step": 13694
    },
    {
      "epoch": 1.1782672287705411,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008694723088821842,
      "loss": 0.9842,
      "step": 13695
    },
    {
      "epoch": 1.1783532650778628,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008694540892090646,
      "loss": 0.722,
      "step": 13696
    },
    {
      "epoch": 1.1784393013851846,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008694358684553636,
      "loss": 1.449,
      "step": 13697
    },
    {
      "epoch": 1.1785253376925062,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008694176466211345,
      "loss": 0.9414,
      "step": 13698
    },
    {
      "epoch": 1.1786113739998278,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008693994237064307,
      "loss": 1.4555,
      "step": 13699
    },
    {
      "epoch": 1.1786974103071497,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008693811997113055,
      "loss": 2.0549,
      "step": 13700
    },
    {
      "epoch": 1.1787834466144713,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008693629746358121,
      "loss": 1.3018,
      "step": 13701
    },
    {
      "epoch": 1.178869482921793,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008693447484800036,
      "loss": 1.0066,
      "step": 13702
    },
    {
      "epoch": 1.1789555192291146,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008693265212439338,
      "loss": 1.4366,
      "step": 13703
    },
    {
      "epoch": 1.1790415555364364,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008693082929276557,
      "loss": 1.9384,
      "step": 13704
    },
    {
      "epoch": 1.179127591843758,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008692900635312226,
      "loss": 1.0819,
      "step": 13705
    },
    {
      "epoch": 1.1792136281510797,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000869271833054688,
      "loss": 1.0647,
      "step": 13706
    },
    {
      "epoch": 1.1792996644584015,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000869253601498105,
      "loss": 1.3733,
      "step": 13707
    },
    {
      "epoch": 1.1793857007657231,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008692353688615271,
      "loss": 1.3002,
      "step": 13708
    },
    {
      "epoch": 1.1794717370730448,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008692171351450075,
      "loss": 0.7528,
      "step": 13709
    },
    {
      "epoch": 1.1795577733803666,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008691989003485996,
      "loss": 1.3677,
      "step": 13710
    },
    {
      "epoch": 1.1796438096876882,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008691806644723568,
      "loss": 1.1995,
      "step": 13711
    },
    {
      "epoch": 1.1797298459950099,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008691624275163323,
      "loss": 1.5319,
      "step": 13712
    },
    {
      "epoch": 1.1798158823023317,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008691441894805794,
      "loss": 1.2876,
      "step": 13713
    },
    {
      "epoch": 1.1799019186096533,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008691259503651516,
      "loss": 1.2506,
      "step": 13714
    },
    {
      "epoch": 1.179987954916975,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008691077101701023,
      "loss": 1.9673,
      "step": 13715
    },
    {
      "epoch": 1.1800739912242966,
      "grad_norm": 1.375,
      "learning_rate": 0.0008690894688954847,
      "loss": 1.899,
      "step": 13716
    },
    {
      "epoch": 1.1801600275316184,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008690712265413521,
      "loss": 1.5462,
      "step": 13717
    },
    {
      "epoch": 1.18024606383894,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000869052983107758,
      "loss": 1.3306,
      "step": 13718
    },
    {
      "epoch": 1.1803321001462617,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008690347385947556,
      "loss": 0.9517,
      "step": 13719
    },
    {
      "epoch": 1.1804181364535835,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008690164930023986,
      "loss": 1.5187,
      "step": 13720
    },
    {
      "epoch": 1.1805041727609051,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008689982463307399,
      "loss": 1.551,
      "step": 13721
    },
    {
      "epoch": 1.1805902090682268,
      "grad_norm": 1.921875,
      "learning_rate": 0.0008689799985798333,
      "loss": 1.5889,
      "step": 13722
    },
    {
      "epoch": 1.1806762453755484,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008689617497497317,
      "loss": 1.058,
      "step": 13723
    },
    {
      "epoch": 1.1807622816828702,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008689434998404889,
      "loss": 0.8537,
      "step": 13724
    },
    {
      "epoch": 1.1808483179901919,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008689252488521582,
      "loss": 0.757,
      "step": 13725
    },
    {
      "epoch": 1.1809343542975135,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008689069967847927,
      "loss": 1.0234,
      "step": 13726
    },
    {
      "epoch": 1.1810203906048353,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008688887436384461,
      "loss": 1.4547,
      "step": 13727
    },
    {
      "epoch": 1.181106426912157,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008688704894131717,
      "loss": 1.5293,
      "step": 13728
    },
    {
      "epoch": 1.1811924632194786,
      "grad_norm": 1.0,
      "learning_rate": 0.0008688522341090228,
      "loss": 0.8781,
      "step": 13729
    },
    {
      "epoch": 1.1812784995268002,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008688339777260529,
      "loss": 1.7673,
      "step": 13730
    },
    {
      "epoch": 1.181364535834122,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008688157202643152,
      "loss": 0.8108,
      "step": 13731
    },
    {
      "epoch": 1.1814505721414437,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008687974617238634,
      "loss": 1.005,
      "step": 13732
    },
    {
      "epoch": 1.1815366084487653,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008687792021047507,
      "loss": 0.8489,
      "step": 13733
    },
    {
      "epoch": 1.1816226447560871,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008687609414070307,
      "loss": 1.7935,
      "step": 13734
    },
    {
      "epoch": 1.1817086810634088,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008687426796307564,
      "loss": 0.631,
      "step": 13735
    },
    {
      "epoch": 1.1817947173707304,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008687244167759818,
      "loss": 1.1324,
      "step": 13736
    },
    {
      "epoch": 1.1818807536780522,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008687061528427597,
      "loss": 1.0173,
      "step": 13737
    },
    {
      "epoch": 1.1819667899853739,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008686878878311439,
      "loss": 1.4313,
      "step": 13738
    },
    {
      "epoch": 1.1820528262926955,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008686696217411878,
      "loss": 1.9212,
      "step": 13739
    },
    {
      "epoch": 1.1821388626000173,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008686513545729446,
      "loss": 1.1565,
      "step": 13740
    },
    {
      "epoch": 1.182224898907339,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008686330863264682,
      "loss": 1.3787,
      "step": 13741
    },
    {
      "epoch": 1.1823109352146606,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008686148170018114,
      "loss": 1.4428,
      "step": 13742
    },
    {
      "epoch": 1.1823969715219822,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008685965465990282,
      "loss": 0.8327,
      "step": 13743
    },
    {
      "epoch": 1.182483007829304,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008685782751181717,
      "loss": 1.7515,
      "step": 13744
    },
    {
      "epoch": 1.1825690441366257,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008685600025592953,
      "loss": 0.9746,
      "step": 13745
    },
    {
      "epoch": 1.1826550804439473,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008685417289224527,
      "loss": 1.2213,
      "step": 13746
    },
    {
      "epoch": 1.1827411167512691,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008685234542076972,
      "loss": 1.0847,
      "step": 13747
    },
    {
      "epoch": 1.1828271530585908,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008685051784150823,
      "loss": 0.5146,
      "step": 13748
    },
    {
      "epoch": 1.1829131893659124,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008684869015446612,
      "loss": 1.1088,
      "step": 13749
    },
    {
      "epoch": 1.182999225673234,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008684686235964878,
      "loss": 1.1815,
      "step": 13750
    },
    {
      "epoch": 1.1830852619805559,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008684503445706153,
      "loss": 1.315,
      "step": 13751
    },
    {
      "epoch": 1.1831712982878775,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008684320644670973,
      "loss": 1.2227,
      "step": 13752
    },
    {
      "epoch": 1.183257334595199,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008684137832859871,
      "loss": 1.7995,
      "step": 13753
    },
    {
      "epoch": 1.183343370902521,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008683955010273381,
      "loss": 1.2546,
      "step": 13754
    },
    {
      "epoch": 1.1834294072098426,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008683772176912042,
      "loss": 1.2369,
      "step": 13755
    },
    {
      "epoch": 1.1835154435171642,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008683589332776383,
      "loss": 0.663,
      "step": 13756
    },
    {
      "epoch": 1.1836014798244858,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008683406477866942,
      "loss": 1.6833,
      "step": 13757
    },
    {
      "epoch": 1.1836875161318077,
      "grad_norm": 2.484375,
      "learning_rate": 0.0008683223612184254,
      "loss": 1.373,
      "step": 13758
    },
    {
      "epoch": 1.1837735524391293,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008683040735728853,
      "loss": 1.5577,
      "step": 13759
    },
    {
      "epoch": 1.183859588746451,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008682857848501276,
      "loss": 0.8984,
      "step": 13760
    },
    {
      "epoch": 1.1839456250537728,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008682674950502054,
      "loss": 1.6029,
      "step": 13761
    },
    {
      "epoch": 1.1840316613610944,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008682492041731725,
      "loss": 1.3859,
      "step": 13762
    },
    {
      "epoch": 1.184117697668416,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008682309122190822,
      "loss": 1.7961,
      "step": 13763
    },
    {
      "epoch": 1.1842037339757376,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008682126191879881,
      "loss": 1.272,
      "step": 13764
    },
    {
      "epoch": 1.1842897702830595,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008681943250799437,
      "loss": 1.8806,
      "step": 13765
    },
    {
      "epoch": 1.184375806590381,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008681760298950026,
      "loss": 1.0958,
      "step": 13766
    },
    {
      "epoch": 1.184461842897703,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008681577336332182,
      "loss": 0.8966,
      "step": 13767
    },
    {
      "epoch": 1.1845478792050246,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008681394362946439,
      "loss": 1.171,
      "step": 13768
    },
    {
      "epoch": 1.1846339155123462,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008681211378793336,
      "loss": 1.8038,
      "step": 13769
    },
    {
      "epoch": 1.1847199518196678,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008681028383873404,
      "loss": 1.038,
      "step": 13770
    },
    {
      "epoch": 1.1848059881269897,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008680845378187181,
      "loss": 1.2446,
      "step": 13771
    },
    {
      "epoch": 1.1848920244343113,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00086806623617352,
      "loss": 2.2557,
      "step": 13772
    },
    {
      "epoch": 1.184978060741633,
      "grad_norm": 1.796875,
      "learning_rate": 0.0008680479334517998,
      "loss": 2.3546,
      "step": 13773
    },
    {
      "epoch": 1.1850640970489548,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000868029629653611,
      "loss": 1.883,
      "step": 13774
    },
    {
      "epoch": 1.1851501333562764,
      "grad_norm": 1.421875,
      "learning_rate": 0.000868011324779007,
      "loss": 1.0008,
      "step": 13775
    },
    {
      "epoch": 1.185236169663598,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008679930188280416,
      "loss": 1.2108,
      "step": 13776
    },
    {
      "epoch": 1.1853222059709196,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008679747118007679,
      "loss": 1.0335,
      "step": 13777
    },
    {
      "epoch": 1.1854082422782415,
      "grad_norm": 1.140625,
      "learning_rate": 0.00086795640369724,
      "loss": 2.1311,
      "step": 13778
    },
    {
      "epoch": 1.185494278585563,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008679380945175112,
      "loss": 1.4578,
      "step": 13779
    },
    {
      "epoch": 1.1855803148928847,
      "grad_norm": 1.203125,
      "learning_rate": 0.000867919784261635,
      "loss": 0.7476,
      "step": 13780
    },
    {
      "epoch": 1.1856663512002066,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008679014729296649,
      "loss": 1.4361,
      "step": 13781
    },
    {
      "epoch": 1.1857523875075282,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008678831605216546,
      "loss": 1.3473,
      "step": 13782
    },
    {
      "epoch": 1.1858384238148498,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008678648470376576,
      "loss": 0.9961,
      "step": 13783
    },
    {
      "epoch": 1.1859244601221715,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008678465324777275,
      "loss": 1.2836,
      "step": 13784
    },
    {
      "epoch": 1.1860104964294933,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008678282168419179,
      "loss": 1.1639,
      "step": 13785
    },
    {
      "epoch": 1.186096532736815,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000867809900130282,
      "loss": 1.4501,
      "step": 13786
    },
    {
      "epoch": 1.1861825690441365,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008677915823428739,
      "loss": 1.1093,
      "step": 13787
    },
    {
      "epoch": 1.1862686053514584,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000867773263479747,
      "loss": 1.7293,
      "step": 13788
    },
    {
      "epoch": 1.18635464165878,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008677549435409548,
      "loss": 1.3454,
      "step": 13789
    },
    {
      "epoch": 1.1864406779661016,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008677366225265508,
      "loss": 2.138,
      "step": 13790
    },
    {
      "epoch": 1.1865267142734233,
      "grad_norm": 1.25,
      "learning_rate": 0.0008677183004365888,
      "loss": 0.9776,
      "step": 13791
    },
    {
      "epoch": 1.1866127505807451,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008676999772711223,
      "loss": 1.2244,
      "step": 13792
    },
    {
      "epoch": 1.1866987868880667,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008676816530302047,
      "loss": 1.103,
      "step": 13793
    },
    {
      "epoch": 1.1867848231953884,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008676633277138901,
      "loss": 0.7999,
      "step": 13794
    },
    {
      "epoch": 1.1868708595027102,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008676450013222315,
      "loss": 0.6739,
      "step": 13795
    },
    {
      "epoch": 1.1869568958100318,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008676266738552829,
      "loss": 0.956,
      "step": 13796
    },
    {
      "epoch": 1.1870429321173535,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008676083453130977,
      "loss": 2.4336,
      "step": 13797
    },
    {
      "epoch": 1.1871289684246753,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008675900156957297,
      "loss": 1.11,
      "step": 13798
    },
    {
      "epoch": 1.187215004731997,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008675716850032322,
      "loss": 1.659,
      "step": 13799
    },
    {
      "epoch": 1.1873010410393185,
      "grad_norm": 1.96875,
      "learning_rate": 0.0008675533532356592,
      "loss": 1.1729,
      "step": 13800
    },
    {
      "epoch": 1.1873870773466404,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008675350203930641,
      "loss": 1.0998,
      "step": 13801
    },
    {
      "epoch": 1.187473113653962,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008675166864755005,
      "loss": 1.4939,
      "step": 13802
    },
    {
      "epoch": 1.1875591499612836,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008674983514830221,
      "loss": 1.1979,
      "step": 13803
    },
    {
      "epoch": 1.1876451862686053,
      "grad_norm": 1.0,
      "learning_rate": 0.0008674800154156824,
      "loss": 1.762,
      "step": 13804
    },
    {
      "epoch": 1.1877312225759271,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008674616782735352,
      "loss": 1.4181,
      "step": 13805
    },
    {
      "epoch": 1.1878172588832487,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0008674433400566342,
      "loss": 1.2562,
      "step": 13806
    },
    {
      "epoch": 1.1879032951905704,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008674250007650327,
      "loss": 1.121,
      "step": 13807
    },
    {
      "epoch": 1.1879893314978922,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008674066603987847,
      "loss": 1.0213,
      "step": 13808
    },
    {
      "epoch": 1.1880753678052138,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008673883189579435,
      "loss": 1.0595,
      "step": 13809
    },
    {
      "epoch": 1.1881614041125355,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000867369976442563,
      "loss": 0.8322,
      "step": 13810
    },
    {
      "epoch": 1.188247440419857,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008673516328526967,
      "loss": 1.5346,
      "step": 13811
    },
    {
      "epoch": 1.188333476727179,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008673332881883984,
      "loss": 1.0372,
      "step": 13812
    },
    {
      "epoch": 1.1884195130345006,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008673149424497217,
      "loss": 1.4113,
      "step": 13813
    },
    {
      "epoch": 1.1885055493418222,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008672965956367201,
      "loss": 1.4713,
      "step": 13814
    },
    {
      "epoch": 1.188591585649144,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008672782477494474,
      "loss": 1.8858,
      "step": 13815
    },
    {
      "epoch": 1.1886776219564656,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008672598987879574,
      "loss": 1.547,
      "step": 13816
    },
    {
      "epoch": 1.1887636582637873,
      "grad_norm": 1.125,
      "learning_rate": 0.0008672415487523035,
      "loss": 1.1377,
      "step": 13817
    },
    {
      "epoch": 1.188849694571109,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008672231976425397,
      "loss": 1.3092,
      "step": 13818
    },
    {
      "epoch": 1.1889357308784307,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008672048454587192,
      "loss": 1.0501,
      "step": 13819
    },
    {
      "epoch": 1.1890217671857524,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008671864922008961,
      "loss": 1.1577,
      "step": 13820
    },
    {
      "epoch": 1.189107803493074,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008671681378691239,
      "loss": 2.0493,
      "step": 13821
    },
    {
      "epoch": 1.1891938398003958,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008671497824634563,
      "loss": 1.1523,
      "step": 13822
    },
    {
      "epoch": 1.1892798761077175,
      "grad_norm": 1.3125,
      "learning_rate": 0.000867131425983947,
      "loss": 0.7576,
      "step": 13823
    },
    {
      "epoch": 1.189365912415039,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008671130684306496,
      "loss": 0.9566,
      "step": 13824
    },
    {
      "epoch": 1.189451948722361,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008670947098036179,
      "loss": 1.4592,
      "step": 13825
    },
    {
      "epoch": 1.1895379850296826,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008670763501029058,
      "loss": 0.6572,
      "step": 13826
    },
    {
      "epoch": 1.1896240213370042,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008670579893285665,
      "loss": 1.1158,
      "step": 13827
    },
    {
      "epoch": 1.189710057644326,
      "grad_norm": 1.015625,
      "learning_rate": 0.000867039627480654,
      "loss": 0.8364,
      "step": 13828
    },
    {
      "epoch": 1.1897960939516476,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008670212645592221,
      "loss": 1.4148,
      "step": 13829
    },
    {
      "epoch": 1.1898821302589693,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008670029005643242,
      "loss": 0.8284,
      "step": 13830
    },
    {
      "epoch": 1.189968166566291,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008669845354960144,
      "loss": 1.4558,
      "step": 13831
    },
    {
      "epoch": 1.1900542028736127,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000866966169354346,
      "loss": 1.488,
      "step": 13832
    },
    {
      "epoch": 1.1901402391809344,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008669478021393731,
      "loss": 1.2338,
      "step": 13833
    },
    {
      "epoch": 1.190226275488256,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008669294338511493,
      "loss": 1.4953,
      "step": 13834
    },
    {
      "epoch": 1.1903123117955778,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008669110644897281,
      "loss": 1.2605,
      "step": 13835
    },
    {
      "epoch": 1.1903983481028995,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008668926940551635,
      "loss": 0.7566,
      "step": 13836
    },
    {
      "epoch": 1.190484384410221,
      "grad_norm": 1.09375,
      "learning_rate": 0.000866874322547509,
      "loss": 1.086,
      "step": 13837
    },
    {
      "epoch": 1.1905704207175427,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008668559499668185,
      "loss": 1.7172,
      "step": 13838
    },
    {
      "epoch": 1.1906564570248646,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008668375763131457,
      "loss": 1.4193,
      "step": 13839
    },
    {
      "epoch": 1.1907424933321862,
      "grad_norm": 1.78125,
      "learning_rate": 0.0008668192015865445,
      "loss": 2.02,
      "step": 13840
    },
    {
      "epoch": 1.1908285296395078,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008668008257870683,
      "loss": 1.6468,
      "step": 13841
    },
    {
      "epoch": 1.1909145659468297,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008667824489147711,
      "loss": 1.3702,
      "step": 13842
    },
    {
      "epoch": 1.1910006022541513,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008667640709697065,
      "loss": 1.0892,
      "step": 13843
    },
    {
      "epoch": 1.191086638561473,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008667456919519283,
      "loss": 0.6125,
      "step": 13844
    },
    {
      "epoch": 1.1911726748687945,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008667273118614905,
      "loss": 1.1784,
      "step": 13845
    },
    {
      "epoch": 1.1912587111761164,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008667089306984464,
      "loss": 1.1799,
      "step": 13846
    },
    {
      "epoch": 1.191344747483438,
      "grad_norm": 1.046875,
      "learning_rate": 0.00086669054846285,
      "loss": 0.8437,
      "step": 13847
    },
    {
      "epoch": 1.1914307837907596,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008666721651547552,
      "loss": 0.8605,
      "step": 13848
    },
    {
      "epoch": 1.1915168200980815,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008666537807742157,
      "loss": 1.5959,
      "step": 13849
    },
    {
      "epoch": 1.191602856405403,
      "grad_norm": 1.3125,
      "learning_rate": 0.000866635395321285,
      "loss": 1.3581,
      "step": 13850
    },
    {
      "epoch": 1.1916888927127247,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008666170087960172,
      "loss": 0.766,
      "step": 13851
    },
    {
      "epoch": 1.1917749290200463,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008665986211984658,
      "loss": 1.4246,
      "step": 13852
    },
    {
      "epoch": 1.1918609653273682,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008665802325286849,
      "loss": 1.4204,
      "step": 13853
    },
    {
      "epoch": 1.1919470016346898,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000866561842786728,
      "loss": 0.6273,
      "step": 13854
    },
    {
      "epoch": 1.1920330379420117,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008665434519726491,
      "loss": 1.712,
      "step": 13855
    },
    {
      "epoch": 1.1921190742493333,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008665250600865018,
      "loss": 1.5569,
      "step": 13856
    },
    {
      "epoch": 1.192205110556655,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00086650666712834,
      "loss": 1.6202,
      "step": 13857
    },
    {
      "epoch": 1.1922911468639765,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008664882730982175,
      "loss": 1.4187,
      "step": 13858
    },
    {
      "epoch": 1.1923771831712984,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008664698779961882,
      "loss": 1.0579,
      "step": 13859
    },
    {
      "epoch": 1.19246321947862,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008664514818223056,
      "loss": 1.1966,
      "step": 13860
    },
    {
      "epoch": 1.1925492557859416,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008664330845766237,
      "loss": 1.2853,
      "step": 13861
    },
    {
      "epoch": 1.1926352920932635,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008664146862591964,
      "loss": 1.4579,
      "step": 13862
    },
    {
      "epoch": 1.192721328400585,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008663962868700774,
      "loss": 1.2463,
      "step": 13863
    },
    {
      "epoch": 1.1928073647079067,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008663778864093205,
      "loss": 1.9927,
      "step": 13864
    },
    {
      "epoch": 1.1928934010152283,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008663594848769794,
      "loss": 0.7949,
      "step": 13865
    },
    {
      "epoch": 1.1929794373225502,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008663410822731082,
      "loss": 0.8877,
      "step": 13866
    },
    {
      "epoch": 1.1930654736298718,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008663226785977605,
      "loss": 1.4733,
      "step": 13867
    },
    {
      "epoch": 1.1931515099371934,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008663042738509904,
      "loss": 1.0509,
      "step": 13868
    },
    {
      "epoch": 1.1932375462445153,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008662858680328514,
      "loss": 0.8885,
      "step": 13869
    },
    {
      "epoch": 1.193323582551837,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008662674611433975,
      "loss": 0.9978,
      "step": 13870
    },
    {
      "epoch": 1.1934096188591585,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008662490531826825,
      "loss": 1.5357,
      "step": 13871
    },
    {
      "epoch": 1.1934956551664802,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008662306441507602,
      "loss": 1.1094,
      "step": 13872
    },
    {
      "epoch": 1.193581691473802,
      "grad_norm": 1.125,
      "learning_rate": 0.0008662122340476845,
      "loss": 0.9594,
      "step": 13873
    },
    {
      "epoch": 1.1936677277811236,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008661938228735094,
      "loss": 1.6328,
      "step": 13874
    },
    {
      "epoch": 1.1937537640884452,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008661754106282884,
      "loss": 0.8107,
      "step": 13875
    },
    {
      "epoch": 1.193839800395767,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008661569973120757,
      "loss": 1.5618,
      "step": 13876
    },
    {
      "epoch": 1.1939258367030887,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008661385829249248,
      "loss": 1.6051,
      "step": 13877
    },
    {
      "epoch": 1.1940118730104103,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008661201674668899,
      "loss": 1.2628,
      "step": 13878
    },
    {
      "epoch": 1.194097909317732,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008661017509380246,
      "loss": 0.8347,
      "step": 13879
    },
    {
      "epoch": 1.1941839456250538,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008660833333383831,
      "loss": 1.6786,
      "step": 13880
    },
    {
      "epoch": 1.1942699819323754,
      "grad_norm": 1.25,
      "learning_rate": 0.0008660649146680189,
      "loss": 0.9496,
      "step": 13881
    },
    {
      "epoch": 1.194356018239697,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000866046494926986,
      "loss": 0.9322,
      "step": 13882
    },
    {
      "epoch": 1.194442054547019,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008660280741153383,
      "loss": 1.2953,
      "step": 13883
    },
    {
      "epoch": 1.1945280908543405,
      "grad_norm": 1.375,
      "learning_rate": 0.0008660096522331298,
      "loss": 1.167,
      "step": 13884
    },
    {
      "epoch": 1.1946141271616622,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008659912292804141,
      "loss": 1.2497,
      "step": 13885
    },
    {
      "epoch": 1.194700163468984,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008659728052572453,
      "loss": 1.1219,
      "step": 13886
    },
    {
      "epoch": 1.1947861997763056,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008659543801636771,
      "loss": 1.0102,
      "step": 13887
    },
    {
      "epoch": 1.1948722360836272,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008659359539997637,
      "loss": 1.151,
      "step": 13888
    },
    {
      "epoch": 1.194958272390949,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008659175267655588,
      "loss": 1.2214,
      "step": 13889
    },
    {
      "epoch": 1.1950443086982707,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0008658990984611162,
      "loss": 1.0817,
      "step": 13890
    },
    {
      "epoch": 1.1951303450055923,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008658806690864899,
      "loss": 1.5642,
      "step": 13891
    },
    {
      "epoch": 1.195216381312914,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008658622386417339,
      "loss": 1.2449,
      "step": 13892
    },
    {
      "epoch": 1.1953024176202358,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008658438071269019,
      "loss": 1.3032,
      "step": 13893
    },
    {
      "epoch": 1.1953884539275574,
      "grad_norm": 1.015625,
      "learning_rate": 0.000865825374542048,
      "loss": 1.2623,
      "step": 13894
    },
    {
      "epoch": 1.195474490234879,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008658069408872259,
      "loss": 0.9744,
      "step": 13895
    },
    {
      "epoch": 1.195560526542201,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008657885061624897,
      "loss": 1.2762,
      "step": 13896
    },
    {
      "epoch": 1.1956465628495225,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008657700703678932,
      "loss": 1.2395,
      "step": 13897
    },
    {
      "epoch": 1.1957325991568442,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008657516335034906,
      "loss": 1.2148,
      "step": 13898
    },
    {
      "epoch": 1.1958186354641658,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008657331955693355,
      "loss": 1.1485,
      "step": 13899
    },
    {
      "epoch": 1.1959046717714876,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008657147565654819,
      "loss": 1.3562,
      "step": 13900
    },
    {
      "epoch": 1.1959907080788093,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008656963164919838,
      "loss": 1.3076,
      "step": 13901
    },
    {
      "epoch": 1.1960767443861309,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000865677875348895,
      "loss": 1.3683,
      "step": 13902
    },
    {
      "epoch": 1.1961627806934527,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008656594331362698,
      "loss": 1.9638,
      "step": 13903
    },
    {
      "epoch": 1.1962488170007743,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008656409898541615,
      "loss": 1.4896,
      "step": 13904
    },
    {
      "epoch": 1.196334853308096,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008656225455026247,
      "loss": 1.2276,
      "step": 13905
    },
    {
      "epoch": 1.1964208896154176,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008656041000817128,
      "loss": 1.5206,
      "step": 13906
    },
    {
      "epoch": 1.1965069259227394,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008655856535914802,
      "loss": 1.2495,
      "step": 13907
    },
    {
      "epoch": 1.196592962230061,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008655672060319807,
      "loss": 0.6474,
      "step": 13908
    },
    {
      "epoch": 1.1966789985373827,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008655487574032681,
      "loss": 1.3108,
      "step": 13909
    },
    {
      "epoch": 1.1967650348447045,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008655303077053966,
      "loss": 2.4369,
      "step": 13910
    },
    {
      "epoch": 1.1968510711520262,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008655118569384201,
      "loss": 1.4722,
      "step": 13911
    },
    {
      "epoch": 1.1969371074593478,
      "grad_norm": 2.0625,
      "learning_rate": 0.0008654934051023923,
      "loss": 1.7174,
      "step": 13912
    },
    {
      "epoch": 1.1970231437666696,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008654749521973675,
      "loss": 0.9258,
      "step": 13913
    },
    {
      "epoch": 1.1971091800739913,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008654564982233994,
      "loss": 1.5496,
      "step": 13914
    },
    {
      "epoch": 1.1971952163813129,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008654380431805423,
      "loss": 1.2236,
      "step": 13915
    },
    {
      "epoch": 1.1972812526886347,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008654195870688499,
      "loss": 1.5152,
      "step": 13916
    },
    {
      "epoch": 1.1973672889959563,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008654011298883763,
      "loss": 1.3323,
      "step": 13917
    },
    {
      "epoch": 1.197453325303278,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008653826716391755,
      "loss": 1.0367,
      "step": 13918
    },
    {
      "epoch": 1.1975393616105996,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008653642123213015,
      "loss": 1.5987,
      "step": 13919
    },
    {
      "epoch": 1.1976253979179214,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008653457519348081,
      "loss": 1.5606,
      "step": 13920
    },
    {
      "epoch": 1.197711434225243,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008653272904797496,
      "loss": 0.7791,
      "step": 13921
    },
    {
      "epoch": 1.1977974705325647,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008653088279561796,
      "loss": 2.3223,
      "step": 13922
    },
    {
      "epoch": 1.1978835068398865,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008652903643641526,
      "loss": 1.8957,
      "step": 13923
    },
    {
      "epoch": 1.1979695431472082,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008652718997037221,
      "loss": 1.0898,
      "step": 13924
    },
    {
      "epoch": 1.1980555794545298,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008652534339749423,
      "loss": 0.8378,
      "step": 13925
    },
    {
      "epoch": 1.1981416157618514,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008652349671778674,
      "loss": 0.9827,
      "step": 13926
    },
    {
      "epoch": 1.1982276520691733,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008652164993125512,
      "loss": 0.8664,
      "step": 13927
    },
    {
      "epoch": 1.1983136883764949,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008651980303790479,
      "loss": 1.0323,
      "step": 13928
    },
    {
      "epoch": 1.1983997246838165,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008651795603774112,
      "loss": 1.4759,
      "step": 13929
    },
    {
      "epoch": 1.1984857609911383,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008651610893076953,
      "loss": 1.0802,
      "step": 13930
    },
    {
      "epoch": 1.19857179729846,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008651426171699543,
      "loss": 1.7417,
      "step": 13931
    },
    {
      "epoch": 1.1986578336057816,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008651241439642422,
      "loss": 1.6719,
      "step": 13932
    },
    {
      "epoch": 1.1987438699131032,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000865105669690613,
      "loss": 1.247,
      "step": 13933
    },
    {
      "epoch": 1.198829906220425,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008650871943491206,
      "loss": 1.963,
      "step": 13934
    },
    {
      "epoch": 1.1989159425277467,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008650687179398193,
      "loss": 0.9867,
      "step": 13935
    },
    {
      "epoch": 1.1990019788350683,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008650502404627629,
      "loss": 0.9304,
      "step": 13936
    },
    {
      "epoch": 1.1990880151423902,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008650317619180057,
      "loss": 1.1703,
      "step": 13937
    },
    {
      "epoch": 1.1991740514497118,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008650132823056014,
      "loss": 1.2271,
      "step": 13938
    },
    {
      "epoch": 1.1992600877570334,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008649948016256044,
      "loss": 1.1491,
      "step": 13939
    },
    {
      "epoch": 1.199346124064355,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008649763198780684,
      "loss": 2.3052,
      "step": 13940
    },
    {
      "epoch": 1.1994321603716769,
      "grad_norm": 1.03125,
      "learning_rate": 0.000864957837063048,
      "loss": 1.3245,
      "step": 13941
    },
    {
      "epoch": 1.1995181966789985,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008649393531805965,
      "loss": 1.4168,
      "step": 13942
    },
    {
      "epoch": 1.1996042329863204,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008649208682307686,
      "loss": 1.7331,
      "step": 13943
    },
    {
      "epoch": 1.199690269293642,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008649023822136181,
      "loss": 0.8625,
      "step": 13944
    },
    {
      "epoch": 1.1997763056009636,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008648838951291991,
      "loss": 1.2537,
      "step": 13945
    },
    {
      "epoch": 1.1998623419082852,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008648654069775656,
      "loss": 1.931,
      "step": 13946
    },
    {
      "epoch": 1.199948378215607,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008648469177587719,
      "loss": 1.6097,
      "step": 13947
    },
    {
      "epoch": 1.2000344145229287,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008648284274728718,
      "loss": 0.8385,
      "step": 13948
    },
    {
      "epoch": 1.2001204508302503,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008648099361199196,
      "loss": 0.8393,
      "step": 13949
    },
    {
      "epoch": 1.2002064871375722,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008647914436999691,
      "loss": 1.0371,
      "step": 13950
    },
    {
      "epoch": 1.2002925234448938,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008647729502130748,
      "loss": 1.8067,
      "step": 13951
    },
    {
      "epoch": 1.2003785597522154,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008647544556592904,
      "loss": 1.983,
      "step": 13952
    },
    {
      "epoch": 1.200464596059537,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008647359600386701,
      "loss": 0.8933,
      "step": 13953
    },
    {
      "epoch": 1.2005506323668589,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008647174633512683,
      "loss": 1.002,
      "step": 13954
    },
    {
      "epoch": 1.2006366686741805,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008646989655971387,
      "loss": 1.5557,
      "step": 13955
    },
    {
      "epoch": 1.2007227049815021,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008646804667763355,
      "loss": 1.7225,
      "step": 13956
    },
    {
      "epoch": 1.200808741288824,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000864661966888913,
      "loss": 2.0627,
      "step": 13957
    },
    {
      "epoch": 1.2008947775961456,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008646434659349252,
      "loss": 0.7769,
      "step": 13958
    },
    {
      "epoch": 1.2009808139034672,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008646249639144261,
      "loss": 1.3422,
      "step": 13959
    },
    {
      "epoch": 1.2010668502107889,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008646064608274698,
      "loss": 1.2123,
      "step": 13960
    },
    {
      "epoch": 1.2011528865181107,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008645879566741106,
      "loss": 1.4235,
      "step": 13961
    },
    {
      "epoch": 1.2012389228254323,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008645694514544024,
      "loss": 2.1307,
      "step": 13962
    },
    {
      "epoch": 1.201324959132754,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008645509451683997,
      "loss": 1.876,
      "step": 13963
    },
    {
      "epoch": 1.2014109954400758,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008645324378161563,
      "loss": 1.42,
      "step": 13964
    },
    {
      "epoch": 1.2014970317473974,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008645139293977264,
      "loss": 0.8757,
      "step": 13965
    },
    {
      "epoch": 1.201583068054719,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008644954199131642,
      "loss": 2.0464,
      "step": 13966
    },
    {
      "epoch": 1.2016691043620407,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008644769093625237,
      "loss": 1.0108,
      "step": 13967
    },
    {
      "epoch": 1.2017551406693625,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008644583977458591,
      "loss": 0.8409,
      "step": 13968
    },
    {
      "epoch": 1.2018411769766841,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008644398850632247,
      "loss": 0.8265,
      "step": 13969
    },
    {
      "epoch": 1.2019272132840058,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008644213713146744,
      "loss": 1.263,
      "step": 13970
    },
    {
      "epoch": 1.2020132495913276,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008644028565002626,
      "loss": 0.9219,
      "step": 13971
    },
    {
      "epoch": 1.2020992858986492,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008643843406200431,
      "loss": 1.7668,
      "step": 13972
    },
    {
      "epoch": 1.2021853222059709,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008643658236740705,
      "loss": 1.9526,
      "step": 13973
    },
    {
      "epoch": 1.2022713585132927,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008643473056623986,
      "loss": 1.1995,
      "step": 13974
    },
    {
      "epoch": 1.2023573948206143,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008643287865850817,
      "loss": 1.1972,
      "step": 13975
    },
    {
      "epoch": 1.202443431127936,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008643102664421741,
      "loss": 1.9073,
      "step": 13976
    },
    {
      "epoch": 1.2025294674352578,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008642917452337296,
      "loss": 1.5753,
      "step": 13977
    },
    {
      "epoch": 1.2026155037425794,
      "grad_norm": 1.875,
      "learning_rate": 0.0008642732229598027,
      "loss": 1.734,
      "step": 13978
    },
    {
      "epoch": 1.202701540049901,
      "grad_norm": 1.375,
      "learning_rate": 0.0008642546996204476,
      "loss": 1.2968,
      "step": 13979
    },
    {
      "epoch": 1.2027875763572227,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008642361752157181,
      "loss": 1.1383,
      "step": 13980
    },
    {
      "epoch": 1.2028736126645445,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0008642176497456689,
      "loss": 1.4994,
      "step": 13981
    },
    {
      "epoch": 1.2029596489718661,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008641991232103537,
      "loss": 0.718,
      "step": 13982
    },
    {
      "epoch": 1.2030456852791878,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000864180595609827,
      "loss": 1.1923,
      "step": 13983
    },
    {
      "epoch": 1.2031317215865096,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008641620669441427,
      "loss": 1.9826,
      "step": 13984
    },
    {
      "epoch": 1.2032177578938312,
      "grad_norm": 1.8125,
      "learning_rate": 0.0008641435372133554,
      "loss": 1.7376,
      "step": 13985
    },
    {
      "epoch": 1.2033037942011529,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008641250064175191,
      "loss": 0.9973,
      "step": 13986
    },
    {
      "epoch": 1.2033898305084745,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008641064745566878,
      "loss": 1.4866,
      "step": 13987
    },
    {
      "epoch": 1.2034758668157963,
      "grad_norm": 1.0,
      "learning_rate": 0.000864087941630916,
      "loss": 0.8313,
      "step": 13988
    },
    {
      "epoch": 1.203561903123118,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008640694076402577,
      "loss": 1.129,
      "step": 13989
    },
    {
      "epoch": 1.2036479394304396,
      "grad_norm": 1.0,
      "learning_rate": 0.0008640508725847671,
      "loss": 0.9668,
      "step": 13990
    },
    {
      "epoch": 1.2037339757377614,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008640323364644988,
      "loss": 0.9743,
      "step": 13991
    },
    {
      "epoch": 1.203820012045083,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008640137992795065,
      "loss": 1.2096,
      "step": 13992
    },
    {
      "epoch": 1.2039060483524047,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008639952610298448,
      "loss": 1.5516,
      "step": 13993
    },
    {
      "epoch": 1.2039920846597263,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008639767217155674,
      "loss": 1.189,
      "step": 13994
    },
    {
      "epoch": 1.2040781209670481,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008639581813367292,
      "loss": 1.672,
      "step": 13995
    },
    {
      "epoch": 1.2041641572743698,
      "grad_norm": 1.375,
      "learning_rate": 0.0008639396398933841,
      "loss": 1.4857,
      "step": 13996
    },
    {
      "epoch": 1.2042501935816914,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008639210973855862,
      "loss": 1.8895,
      "step": 13997
    },
    {
      "epoch": 1.2043362298890132,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008639025538133897,
      "loss": 1.0067,
      "step": 13998
    },
    {
      "epoch": 1.2044222661963349,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008638840091768494,
      "loss": 1.03,
      "step": 13999
    },
    {
      "epoch": 1.2045083025036565,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000863865463476019,
      "loss": 1.3115,
      "step": 14000
    },
    {
      "epoch": 1.2045943388109783,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008638469167109528,
      "loss": 1.2392,
      "step": 14001
    },
    {
      "epoch": 1.2046803751183,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008638283688817052,
      "loss": 0.8142,
      "step": 14002
    },
    {
      "epoch": 1.2047664114256216,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008638098199883304,
      "loss": 1.3553,
      "step": 14003
    },
    {
      "epoch": 1.2048524477329434,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008637912700308826,
      "loss": 1.1212,
      "step": 14004
    },
    {
      "epoch": 1.204938484040265,
      "grad_norm": 2.484375,
      "learning_rate": 0.0008637727190094161,
      "loss": 1.3794,
      "step": 14005
    },
    {
      "epoch": 1.2050245203475867,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008637541669239853,
      "loss": 1.132,
      "step": 14006
    },
    {
      "epoch": 1.2051105566549083,
      "grad_norm": 1.203125,
      "learning_rate": 0.000863735613774644,
      "loss": 1.6208,
      "step": 14007
    },
    {
      "epoch": 1.2051965929622301,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000863717059561447,
      "loss": 1.8429,
      "step": 14008
    },
    {
      "epoch": 1.2052826292695518,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008636985042844482,
      "loss": 1.173,
      "step": 14009
    },
    {
      "epoch": 1.2053686655768734,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008636799479437021,
      "loss": 0.7761,
      "step": 14010
    },
    {
      "epoch": 1.2054547018841952,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008636613905392628,
      "loss": 1.7493,
      "step": 14011
    },
    {
      "epoch": 1.2055407381915169,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008636428320711847,
      "loss": 0.8371,
      "step": 14012
    },
    {
      "epoch": 1.2056267744988385,
      "grad_norm": 1.28125,
      "learning_rate": 0.000863624272539522,
      "loss": 1.6922,
      "step": 14013
    },
    {
      "epoch": 1.20571281080616,
      "grad_norm": 1.0,
      "learning_rate": 0.0008636057119443292,
      "loss": 1.5634,
      "step": 14014
    },
    {
      "epoch": 1.205798847113482,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008635871502856603,
      "loss": 0.9705,
      "step": 14015
    },
    {
      "epoch": 1.2058848834208036,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008635685875635697,
      "loss": 1.5416,
      "step": 14016
    },
    {
      "epoch": 1.2059709197281252,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008635500237781116,
      "loss": 1.0879,
      "step": 14017
    },
    {
      "epoch": 1.206056956035447,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008635314589293404,
      "loss": 0.9232,
      "step": 14018
    },
    {
      "epoch": 1.2061429923427687,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008635128930173106,
      "loss": 1.1333,
      "step": 14019
    },
    {
      "epoch": 1.2062290286500903,
      "grad_norm": 1.234375,
      "learning_rate": 0.000863494326042076,
      "loss": 1.0545,
      "step": 14020
    },
    {
      "epoch": 1.206315064957412,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008634757580036913,
      "loss": 1.358,
      "step": 14021
    },
    {
      "epoch": 1.2064011012647338,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008634571889022109,
      "loss": 0.8226,
      "step": 14022
    },
    {
      "epoch": 1.2064871375720554,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008634386187376888,
      "loss": 1.3322,
      "step": 14023
    },
    {
      "epoch": 1.206573173879377,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008634200475101792,
      "loss": 0.8659,
      "step": 14024
    },
    {
      "epoch": 1.2066592101866989,
      "grad_norm": 1.046875,
      "learning_rate": 0.000863401475219737,
      "loss": 1.2168,
      "step": 14025
    },
    {
      "epoch": 1.2067452464940205,
      "grad_norm": 1.25,
      "learning_rate": 0.0008633829018664159,
      "loss": 1.5515,
      "step": 14026
    },
    {
      "epoch": 1.206831282801342,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008633643274502705,
      "loss": 1.1762,
      "step": 14027
    },
    {
      "epoch": 1.2069173191086637,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008633457519713553,
      "loss": 2.1182,
      "step": 14028
    },
    {
      "epoch": 1.2070033554159856,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008633271754297242,
      "loss": 1.2639,
      "step": 14029
    },
    {
      "epoch": 1.2070893917233072,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008633085978254319,
      "loss": 1.8166,
      "step": 14030
    },
    {
      "epoch": 1.207175428030629,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008632900191585327,
      "loss": 1.6929,
      "step": 14031
    },
    {
      "epoch": 1.2072614643379507,
      "grad_norm": 1.5,
      "learning_rate": 0.0008632714394290807,
      "loss": 1.1176,
      "step": 14032
    },
    {
      "epoch": 1.2073475006452723,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008632528586371305,
      "loss": 1.5843,
      "step": 14033
    },
    {
      "epoch": 1.207433536952594,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008632342767827361,
      "loss": 1.4949,
      "step": 14034
    },
    {
      "epoch": 1.2075195732599158,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008632156938659523,
      "loss": 1.5771,
      "step": 14035
    },
    {
      "epoch": 1.2076056095672374,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008631971098868331,
      "loss": 1.9244,
      "step": 14036
    },
    {
      "epoch": 1.207691645874559,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008631785248454332,
      "loss": 2.5367,
      "step": 14037
    },
    {
      "epoch": 1.2077776821818809,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008631599387418068,
      "loss": 1.3814,
      "step": 14038
    },
    {
      "epoch": 1.2078637184892025,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000863141351576008,
      "loss": 1.3415,
      "step": 14039
    },
    {
      "epoch": 1.2079497547965241,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008631227633480914,
      "loss": 1.1423,
      "step": 14040
    },
    {
      "epoch": 1.2080357911038457,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008631041740581114,
      "loss": 2.0237,
      "step": 14041
    },
    {
      "epoch": 1.2081218274111676,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008630855837061223,
      "loss": 2.1345,
      "step": 14042
    },
    {
      "epoch": 1.2082078637184892,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008630669922921784,
      "loss": 1.2054,
      "step": 14043
    },
    {
      "epoch": 1.2082939000258108,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008630483998163343,
      "loss": 1.218,
      "step": 14044
    },
    {
      "epoch": 1.2083799363331327,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008630298062786442,
      "loss": 0.9149,
      "step": 14045
    },
    {
      "epoch": 1.2084659726404543,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008630112116791624,
      "loss": 1.068,
      "step": 14046
    },
    {
      "epoch": 1.208552008947776,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008629926160179435,
      "loss": 1.2478,
      "step": 14047
    },
    {
      "epoch": 1.2086380452550975,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008629740192950417,
      "loss": 1.3228,
      "step": 14048
    },
    {
      "epoch": 1.2087240815624194,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008629554215105116,
      "loss": 1.3809,
      "step": 14049
    },
    {
      "epoch": 1.208810117869741,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008629368226644074,
      "loss": 0.844,
      "step": 14050
    },
    {
      "epoch": 1.2088961541770626,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008629182227567838,
      "loss": 1.3822,
      "step": 14051
    },
    {
      "epoch": 1.2089821904843845,
      "grad_norm": 1.5,
      "learning_rate": 0.0008628996217876948,
      "loss": 1.2635,
      "step": 14052
    },
    {
      "epoch": 1.2090682267917061,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000862881019757195,
      "loss": 1.5853,
      "step": 14053
    },
    {
      "epoch": 1.2091542630990277,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008628624166653387,
      "loss": 1.2536,
      "step": 14054
    },
    {
      "epoch": 1.2092402994063494,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008628438125121805,
      "loss": 0.7338,
      "step": 14055
    },
    {
      "epoch": 1.2093263357136712,
      "grad_norm": 1.625,
      "learning_rate": 0.0008628252072977748,
      "loss": 1.0834,
      "step": 14056
    },
    {
      "epoch": 1.2094123720209928,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008628066010221759,
      "loss": 0.5983,
      "step": 14057
    },
    {
      "epoch": 1.2094984083283145,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008627879936854381,
      "loss": 1.6959,
      "step": 14058
    },
    {
      "epoch": 1.2095844446356363,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008627693852876161,
      "loss": 2.0887,
      "step": 14059
    },
    {
      "epoch": 1.209670480942958,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008627507758287643,
      "loss": 1.1053,
      "step": 14060
    },
    {
      "epoch": 1.2097565172502796,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008627321653089368,
      "loss": 1.0674,
      "step": 14061
    },
    {
      "epoch": 1.2098425535576014,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008627135537281885,
      "loss": 1.9047,
      "step": 14062
    },
    {
      "epoch": 1.209928589864923,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008626949410865735,
      "loss": 1.9652,
      "step": 14063
    },
    {
      "epoch": 1.2100146261722446,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008626763273841462,
      "loss": 1.4442,
      "step": 14064
    },
    {
      "epoch": 1.2101006624795665,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008626577126209613,
      "loss": 1.4457,
      "step": 14065
    },
    {
      "epoch": 1.2101866987868881,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008626390967970732,
      "loss": 1.6357,
      "step": 14066
    },
    {
      "epoch": 1.2102727350942097,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008626204799125362,
      "loss": 1.2767,
      "step": 14067
    },
    {
      "epoch": 1.2103587714015314,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008626018619674048,
      "loss": 2.1165,
      "step": 14068
    },
    {
      "epoch": 1.2104448077088532,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008625832429617335,
      "loss": 1.5439,
      "step": 14069
    },
    {
      "epoch": 1.2105308440161748,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008625646228955768,
      "loss": 1.118,
      "step": 14070
    },
    {
      "epoch": 1.2106168803234965,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000862546001768989,
      "loss": 0.9036,
      "step": 14071
    },
    {
      "epoch": 1.2107029166308183,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008625273795820248,
      "loss": 0.9265,
      "step": 14072
    },
    {
      "epoch": 1.21078895293814,
      "grad_norm": 1.25,
      "learning_rate": 0.0008625087563347382,
      "loss": 1.0055,
      "step": 14073
    },
    {
      "epoch": 1.2108749892454616,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008624901320271843,
      "loss": 1.373,
      "step": 14074
    },
    {
      "epoch": 1.2109610255527832,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008624715066594171,
      "loss": 1.5801,
      "step": 14075
    },
    {
      "epoch": 1.211047061860105,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008624528802314913,
      "loss": 1.3626,
      "step": 14076
    },
    {
      "epoch": 1.2111330981674266,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008624342527434614,
      "loss": 1.8517,
      "step": 14077
    },
    {
      "epoch": 1.2112191344747483,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008624156241953816,
      "loss": 1.0907,
      "step": 14078
    },
    {
      "epoch": 1.2113051707820701,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008623969945873068,
      "loss": 1.0941,
      "step": 14079
    },
    {
      "epoch": 1.2113912070893917,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008623783639192911,
      "loss": 0.839,
      "step": 14080
    },
    {
      "epoch": 1.2114772433967134,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008623597321913893,
      "loss": 1.1779,
      "step": 14081
    },
    {
      "epoch": 1.211563279704035,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008623410994036556,
      "loss": 1.6592,
      "step": 14082
    },
    {
      "epoch": 1.2116493160113568,
      "grad_norm": 3.640625,
      "learning_rate": 0.0008623224655561447,
      "loss": 1.5093,
      "step": 14083
    },
    {
      "epoch": 1.2117353523186785,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008623038306489112,
      "loss": 1.4426,
      "step": 14084
    },
    {
      "epoch": 1.211821388626,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008622851946820094,
      "loss": 0.4302,
      "step": 14085
    },
    {
      "epoch": 1.211907424933322,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008622665576554939,
      "loss": 1.5339,
      "step": 14086
    },
    {
      "epoch": 1.2119934612406436,
      "grad_norm": 0.75,
      "learning_rate": 0.0008622479195694191,
      "loss": 0.7858,
      "step": 14087
    },
    {
      "epoch": 1.2120794975479652,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008622292804238396,
      "loss": 1.5823,
      "step": 14088
    },
    {
      "epoch": 1.212165533855287,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008622106402188099,
      "loss": 1.4474,
      "step": 14089
    },
    {
      "epoch": 1.2122515701626086,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008621919989543845,
      "loss": 1.249,
      "step": 14090
    },
    {
      "epoch": 1.2123376064699303,
      "grad_norm": 2.546875,
      "learning_rate": 0.000862173356630618,
      "loss": 2.0811,
      "step": 14091
    },
    {
      "epoch": 1.2124236427772521,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008621547132475649,
      "loss": 1.4023,
      "step": 14092
    },
    {
      "epoch": 1.2125096790845737,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008621360688052795,
      "loss": 1.0907,
      "step": 14093
    },
    {
      "epoch": 1.2125957153918954,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008621174233038167,
      "loss": 1.1734,
      "step": 14094
    },
    {
      "epoch": 1.212681751699217,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008620987767432309,
      "loss": 1.6564,
      "step": 14095
    },
    {
      "epoch": 1.2127677880065388,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008620801291235765,
      "loss": 1.079,
      "step": 14096
    },
    {
      "epoch": 1.2128538243138605,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008620614804449083,
      "loss": 0.6848,
      "step": 14097
    },
    {
      "epoch": 1.212939860621182,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008620428307072806,
      "loss": 1.428,
      "step": 14098
    },
    {
      "epoch": 1.213025896928504,
      "grad_norm": 1.25,
      "learning_rate": 0.0008620241799107478,
      "loss": 1.6525,
      "step": 14099
    },
    {
      "epoch": 1.2131119332358256,
      "grad_norm": 0.9375,
      "learning_rate": 0.000862005528055365,
      "loss": 0.9467,
      "step": 14100
    },
    {
      "epoch": 1.2131979695431472,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008619868751411863,
      "loss": 1.7393,
      "step": 14101
    },
    {
      "epoch": 1.2132840058504688,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008619682211682665,
      "loss": 0.8005,
      "step": 14102
    },
    {
      "epoch": 1.2133700421577907,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008619495661366599,
      "loss": 1.7244,
      "step": 14103
    },
    {
      "epoch": 1.2134560784651123,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008619309100464213,
      "loss": 1.0978,
      "step": 14104
    },
    {
      "epoch": 1.213542114772434,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008619122528976051,
      "loss": 2.0537,
      "step": 14105
    },
    {
      "epoch": 1.2136281510797557,
      "grad_norm": 1.203125,
      "learning_rate": 0.000861893594690266,
      "loss": 1.6789,
      "step": 14106
    },
    {
      "epoch": 1.2137141873870774,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008618749354244586,
      "loss": 1.2861,
      "step": 14107
    },
    {
      "epoch": 1.213800223694399,
      "grad_norm": 1.375,
      "learning_rate": 0.0008618562751002372,
      "loss": 1.6446,
      "step": 14108
    },
    {
      "epoch": 1.2138862600017206,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008618376137176568,
      "loss": 1.5055,
      "step": 14109
    },
    {
      "epoch": 1.2139722963090425,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008618189512767715,
      "loss": 1.0889,
      "step": 14110
    },
    {
      "epoch": 1.214058332616364,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008618002877776364,
      "loss": 2.123,
      "step": 14111
    },
    {
      "epoch": 1.2141443689236857,
      "grad_norm": 2.21875,
      "learning_rate": 0.0008617816232203057,
      "loss": 1.0806,
      "step": 14112
    },
    {
      "epoch": 1.2142304052310076,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008617629576048343,
      "loss": 1.7393,
      "step": 14113
    },
    {
      "epoch": 1.2143164415383292,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008617442909312763,
      "loss": 1.4645,
      "step": 14114
    },
    {
      "epoch": 1.2144024778456508,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008617256231996868,
      "loss": 0.7325,
      "step": 14115
    },
    {
      "epoch": 1.2144885141529724,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008617069544101201,
      "loss": 1.7385,
      "step": 14116
    },
    {
      "epoch": 1.2145745504602943,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008616882845626309,
      "loss": 1.8202,
      "step": 14117
    },
    {
      "epoch": 1.214660586767616,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0008616696136572739,
      "loss": 1.1072,
      "step": 14118
    },
    {
      "epoch": 1.2147466230749377,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008616509416941036,
      "loss": 0.8845,
      "step": 14119
    },
    {
      "epoch": 1.2148326593822594,
      "grad_norm": 31.5,
      "learning_rate": 0.0008616322686731746,
      "loss": 1.5648,
      "step": 14120
    },
    {
      "epoch": 1.214918695689581,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008616135945945416,
      "loss": 1.8538,
      "step": 14121
    },
    {
      "epoch": 1.2150047319969026,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008615949194582591,
      "loss": 1.0825,
      "step": 14122
    },
    {
      "epoch": 1.2150907683042245,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008615762432643819,
      "loss": 1.0815,
      "step": 14123
    },
    {
      "epoch": 1.215176804611546,
      "grad_norm": 1.90625,
      "learning_rate": 0.0008615575660129645,
      "loss": 2.2689,
      "step": 14124
    },
    {
      "epoch": 1.2152628409188677,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008615388877040613,
      "loss": 0.5021,
      "step": 14125
    },
    {
      "epoch": 1.2153488772261896,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008615202083377273,
      "loss": 0.7426,
      "step": 14126
    },
    {
      "epoch": 1.2154349135335112,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008615015279140171,
      "loss": 1.4605,
      "step": 14127
    },
    {
      "epoch": 1.2155209498408328,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008614828464329851,
      "loss": 1.6873,
      "step": 14128
    },
    {
      "epoch": 1.2156069861481544,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008614641638946862,
      "loss": 1.3384,
      "step": 14129
    },
    {
      "epoch": 1.2156930224554763,
      "grad_norm": 1.765625,
      "learning_rate": 0.000861445480299175,
      "loss": 1.7534,
      "step": 14130
    },
    {
      "epoch": 1.215779058762798,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008614267956465058,
      "loss": 1.3301,
      "step": 14131
    },
    {
      "epoch": 1.2158650950701195,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008614081099367337,
      "loss": 2.4198,
      "step": 14132
    },
    {
      "epoch": 1.2159511313774414,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008613894231699131,
      "loss": 1.8074,
      "step": 14133
    },
    {
      "epoch": 1.216037167684763,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008613707353460988,
      "loss": 1.614,
      "step": 14134
    },
    {
      "epoch": 1.2161232039920846,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008613520464653453,
      "loss": 1.5465,
      "step": 14135
    },
    {
      "epoch": 1.2162092402994062,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008613333565277074,
      "loss": 2.1218,
      "step": 14136
    },
    {
      "epoch": 1.216295276606728,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008613146655332397,
      "loss": 1.404,
      "step": 14137
    },
    {
      "epoch": 1.2163813129140497,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008612959734819968,
      "loss": 0.6629,
      "step": 14138
    },
    {
      "epoch": 1.2164673492213713,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008612772803740335,
      "loss": 0.8078,
      "step": 14139
    },
    {
      "epoch": 1.2165533855286932,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008612585862094044,
      "loss": 1.8844,
      "step": 14140
    },
    {
      "epoch": 1.2166394218360148,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008612398909881643,
      "loss": 1.0936,
      "step": 14141
    },
    {
      "epoch": 1.2167254581433364,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008612211947103676,
      "loss": 1.0521,
      "step": 14142
    },
    {
      "epoch": 1.216811494450658,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008612024973760693,
      "loss": 0.6652,
      "step": 14143
    },
    {
      "epoch": 1.21689753075798,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008611837989853239,
      "loss": 1.3576,
      "step": 14144
    },
    {
      "epoch": 1.2169835670653015,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008611650995381861,
      "loss": 1.5528,
      "step": 14145
    },
    {
      "epoch": 1.2170696033726232,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008611463990347107,
      "loss": 1.3116,
      "step": 14146
    },
    {
      "epoch": 1.217155639679945,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008611276974749522,
      "loss": 1.5074,
      "step": 14147
    },
    {
      "epoch": 1.2172416759872666,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008611089948589657,
      "loss": 1.6248,
      "step": 14148
    },
    {
      "epoch": 1.2173277122945882,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008610902911868053,
      "loss": 1.5313,
      "step": 14149
    },
    {
      "epoch": 1.21741374860191,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008610715864585262,
      "loss": 1.3827,
      "step": 14150
    },
    {
      "epoch": 1.2174997849092317,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008610528806741828,
      "loss": 1.1069,
      "step": 14151
    },
    {
      "epoch": 1.2175858212165533,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008610341738338301,
      "loss": 0.7245,
      "step": 14152
    },
    {
      "epoch": 1.2176718575238752,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008610154659375226,
      "loss": 1.6828,
      "step": 14153
    },
    {
      "epoch": 1.2177578938311968,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008609967569853151,
      "loss": 1.1392,
      "step": 14154
    },
    {
      "epoch": 1.2178439301385184,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008609780469772622,
      "loss": 1.41,
      "step": 14155
    },
    {
      "epoch": 1.21792996644584,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008609593359134189,
      "loss": 1.1187,
      "step": 14156
    },
    {
      "epoch": 1.218016002753162,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008609406237938395,
      "loss": 1.8181,
      "step": 14157
    },
    {
      "epoch": 1.2181020390604835,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008609219106185792,
      "loss": 1.4243,
      "step": 14158
    },
    {
      "epoch": 1.2181880753678052,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008609031963876923,
      "loss": 1.5761,
      "step": 14159
    },
    {
      "epoch": 1.218274111675127,
      "grad_norm": 1.1875,
      "learning_rate": 0.000860884481101234,
      "loss": 1.6102,
      "step": 14160
    },
    {
      "epoch": 1.2183601479824486,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008608657647592586,
      "loss": 1.856,
      "step": 14161
    },
    {
      "epoch": 1.2184461842897703,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000860847047361821,
      "loss": 1.2379,
      "step": 14162
    },
    {
      "epoch": 1.2185322205970919,
      "grad_norm": 1.3125,
      "learning_rate": 0.000860828328908976,
      "loss": 1.2544,
      "step": 14163
    },
    {
      "epoch": 1.2186182569044137,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008608096094007785,
      "loss": 1.1572,
      "step": 14164
    },
    {
      "epoch": 1.2187042932117353,
      "grad_norm": 8.4375,
      "learning_rate": 0.0008607908888372828,
      "loss": 1.0722,
      "step": 14165
    },
    {
      "epoch": 1.218790329519057,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000860772167218544,
      "loss": 0.7759,
      "step": 14166
    },
    {
      "epoch": 1.2188763658263788,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008607534445446168,
      "loss": 1.3129,
      "step": 14167
    },
    {
      "epoch": 1.2189624021337004,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008607347208155559,
      "loss": 1.6256,
      "step": 14168
    },
    {
      "epoch": 1.219048438441022,
      "grad_norm": 1.75,
      "learning_rate": 0.0008607159960314162,
      "loss": 2.0561,
      "step": 14169
    },
    {
      "epoch": 1.2191344747483437,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008606972701922523,
      "loss": 1.5465,
      "step": 14170
    },
    {
      "epoch": 1.2192205110556655,
      "grad_norm": 1.609375,
      "learning_rate": 0.000860678543298119,
      "loss": 1.5341,
      "step": 14171
    },
    {
      "epoch": 1.2193065473629872,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008606598153490712,
      "loss": 1.4397,
      "step": 14172
    },
    {
      "epoch": 1.2193925836703088,
      "grad_norm": 1.0,
      "learning_rate": 0.0008606410863451634,
      "loss": 0.9829,
      "step": 14173
    },
    {
      "epoch": 1.2194786199776306,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008606223562864508,
      "loss": 1.2628,
      "step": 14174
    },
    {
      "epoch": 1.2195646562849523,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008606036251729879,
      "loss": 1.2123,
      "step": 14175
    },
    {
      "epoch": 1.2196506925922739,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008605848930048294,
      "loss": 0.8041,
      "step": 14176
    },
    {
      "epoch": 1.2197367288995957,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008605661597820303,
      "loss": 1.4602,
      "step": 14177
    },
    {
      "epoch": 1.2198227652069173,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008605474255046452,
      "loss": 1.5493,
      "step": 14178
    },
    {
      "epoch": 1.219908801514239,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008605286901727291,
      "loss": 1.6064,
      "step": 14179
    },
    {
      "epoch": 1.2199948378215608,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008605099537863368,
      "loss": 1.7193,
      "step": 14180
    },
    {
      "epoch": 1.2200808741288824,
      "grad_norm": 1.46875,
      "learning_rate": 0.000860491216345523,
      "loss": 1.8604,
      "step": 14181
    },
    {
      "epoch": 1.220166910436204,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008604724778503423,
      "loss": 0.9521,
      "step": 14182
    },
    {
      "epoch": 1.2202529467435257,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008604537383008499,
      "loss": 1.686,
      "step": 14183
    },
    {
      "epoch": 1.2203389830508475,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008604349976971004,
      "loss": 0.8744,
      "step": 14184
    },
    {
      "epoch": 1.2204250193581692,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008604162560391485,
      "loss": 1.5065,
      "step": 14185
    },
    {
      "epoch": 1.2205110556654908,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008603975133270492,
      "loss": 1.5433,
      "step": 14186
    },
    {
      "epoch": 1.2205970919728126,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008603787695608573,
      "loss": 1.2137,
      "step": 14187
    },
    {
      "epoch": 1.2206831282801343,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008603600247406275,
      "loss": 2.1162,
      "step": 14188
    },
    {
      "epoch": 1.2207691645874559,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008603412788664149,
      "loss": 1.0619,
      "step": 14189
    },
    {
      "epoch": 1.2208552008947775,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008603225319382739,
      "loss": 1.4853,
      "step": 14190
    },
    {
      "epoch": 1.2209412372020993,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008603037839562597,
      "loss": 1.2196,
      "step": 14191
    },
    {
      "epoch": 1.221027273509421,
      "grad_norm": 1.609375,
      "learning_rate": 0.000860285034920427,
      "loss": 1.2844,
      "step": 14192
    },
    {
      "epoch": 1.2211133098167426,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008602662848308307,
      "loss": 1.5383,
      "step": 14193
    },
    {
      "epoch": 1.2211993461240644,
      "grad_norm": 1.25,
      "learning_rate": 0.0008602475336875256,
      "loss": 0.6178,
      "step": 14194
    },
    {
      "epoch": 1.221285382431386,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008602287814905665,
      "loss": 1.2207,
      "step": 14195
    },
    {
      "epoch": 1.2213714187387077,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008602100282400081,
      "loss": 1.5391,
      "step": 14196
    },
    {
      "epoch": 1.2214574550460293,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008601912739359057,
      "loss": 0.9377,
      "step": 14197
    },
    {
      "epoch": 1.2215434913533512,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008601725185783137,
      "loss": 1.3994,
      "step": 14198
    },
    {
      "epoch": 1.2216295276606728,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008601537621672872,
      "loss": 1.3076,
      "step": 14199
    },
    {
      "epoch": 1.2217155639679944,
      "grad_norm": 2.125,
      "learning_rate": 0.0008601350047028808,
      "loss": 1.7705,
      "step": 14200
    },
    {
      "epoch": 1.2218016002753163,
      "grad_norm": 2.640625,
      "learning_rate": 0.0008601162461851498,
      "loss": 1.1675,
      "step": 14201
    },
    {
      "epoch": 1.2218876365826379,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008600974866141488,
      "loss": 1.251,
      "step": 14202
    },
    {
      "epoch": 1.2219736728899595,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008600787259899326,
      "loss": 1.5604,
      "step": 14203
    },
    {
      "epoch": 1.2220597091972811,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008600599643125562,
      "loss": 1.5785,
      "step": 14204
    },
    {
      "epoch": 1.222145745504603,
      "grad_norm": 1.625,
      "learning_rate": 0.0008600412015820745,
      "loss": 1.8217,
      "step": 14205
    },
    {
      "epoch": 1.2222317818119246,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008600224377985422,
      "loss": 2.081,
      "step": 14206
    },
    {
      "epoch": 1.2223178181192464,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008600036729620142,
      "loss": 1.3884,
      "step": 14207
    },
    {
      "epoch": 1.222403854426568,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008599849070725456,
      "loss": 1.612,
      "step": 14208
    },
    {
      "epoch": 1.2224898907338897,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008599661401301913,
      "loss": 1.0159,
      "step": 14209
    },
    {
      "epoch": 1.2225759270412113,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008599473721350059,
      "loss": 1.3185,
      "step": 14210
    },
    {
      "epoch": 1.2226619633485332,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008599286030870443,
      "loss": 0.6674,
      "step": 14211
    },
    {
      "epoch": 1.2227479996558548,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008599098329863617,
      "loss": 0.9478,
      "step": 14212
    },
    {
      "epoch": 1.2228340359631764,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008598910618330129,
      "loss": 1.1161,
      "step": 14213
    },
    {
      "epoch": 1.2229200722704983,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008598722896270525,
      "loss": 1.2268,
      "step": 14214
    },
    {
      "epoch": 1.2230061085778199,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000859853516368536,
      "loss": 1.6965,
      "step": 14215
    },
    {
      "epoch": 1.2230921448851415,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008598347420575176,
      "loss": 1.6201,
      "step": 14216
    },
    {
      "epoch": 1.2231781811924631,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008598159666940528,
      "loss": 0.941,
      "step": 14217
    },
    {
      "epoch": 1.223264217499785,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008597971902781961,
      "loss": 1.0022,
      "step": 14218
    },
    {
      "epoch": 1.2233502538071066,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008597784128100028,
      "loss": 1.622,
      "step": 14219
    },
    {
      "epoch": 1.2234362901144282,
      "grad_norm": 1.25,
      "learning_rate": 0.0008597596342895275,
      "loss": 1.1004,
      "step": 14220
    },
    {
      "epoch": 1.22352232642175,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008597408547168252,
      "loss": 1.8892,
      "step": 14221
    },
    {
      "epoch": 1.2236083627290717,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008597220740919509,
      "loss": 0.8954,
      "step": 14222
    },
    {
      "epoch": 1.2236943990363933,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008597032924149595,
      "loss": 0.9078,
      "step": 14223
    },
    {
      "epoch": 1.223780435343715,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000859684509685906,
      "loss": 0.9357,
      "step": 14224
    },
    {
      "epoch": 1.2238664716510368,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008596657259048451,
      "loss": 1.0775,
      "step": 14225
    },
    {
      "epoch": 1.2239525079583584,
      "grad_norm": 1.515625,
      "learning_rate": 0.000859646941071832,
      "loss": 1.6663,
      "step": 14226
    },
    {
      "epoch": 1.22403854426568,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008596281551869217,
      "loss": 1.0339,
      "step": 14227
    },
    {
      "epoch": 1.2241245805730019,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008596093682501686,
      "loss": 1.0633,
      "step": 14228
    },
    {
      "epoch": 1.2242106168803235,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008595905802616284,
      "loss": 2.0913,
      "step": 14229
    },
    {
      "epoch": 1.2242966531876451,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008595717912213555,
      "loss": 1.4494,
      "step": 14230
    },
    {
      "epoch": 1.2243826894949668,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008595530011294052,
      "loss": 1.0706,
      "step": 14231
    },
    {
      "epoch": 1.2244687258022886,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008595342099858321,
      "loss": 1.2257,
      "step": 14232
    },
    {
      "epoch": 1.2245547621096102,
      "grad_norm": 1.25,
      "learning_rate": 0.0008595154177906915,
      "loss": 1.5995,
      "step": 14233
    },
    {
      "epoch": 1.2246407984169319,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008594966245440382,
      "loss": 1.5576,
      "step": 14234
    },
    {
      "epoch": 1.2247268347242537,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008594778302459272,
      "loss": 1.3635,
      "step": 14235
    },
    {
      "epoch": 1.2248128710315753,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008594590348964135,
      "loss": 2.1891,
      "step": 14236
    },
    {
      "epoch": 1.224898907338897,
      "grad_norm": 1.4375,
      "learning_rate": 0.000859440238495552,
      "loss": 1.439,
      "step": 14237
    },
    {
      "epoch": 1.2249849436462188,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008594214410433976,
      "loss": 0.9573,
      "step": 14238
    },
    {
      "epoch": 1.2250709799535404,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008594026425400055,
      "loss": 0.9909,
      "step": 14239
    },
    {
      "epoch": 1.225157016260862,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008593838429854306,
      "loss": 1.7261,
      "step": 14240
    },
    {
      "epoch": 1.2252430525681839,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008593650423797277,
      "loss": 1.689,
      "step": 14241
    },
    {
      "epoch": 1.2253290888755055,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008593462407229521,
      "loss": 1.3057,
      "step": 14242
    },
    {
      "epoch": 1.2254151251828271,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008593274380151585,
      "loss": 1.0411,
      "step": 14243
    },
    {
      "epoch": 1.2255011614901488,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008593086342564022,
      "loss": 1.4166,
      "step": 14244
    },
    {
      "epoch": 1.2255871977974706,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008592898294467379,
      "loss": 1.4303,
      "step": 14245
    },
    {
      "epoch": 1.2256732341047922,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008592710235862208,
      "loss": 1.488,
      "step": 14246
    },
    {
      "epoch": 1.2257592704121139,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008592522166749057,
      "loss": 1.2518,
      "step": 14247
    },
    {
      "epoch": 1.2258453067194357,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008592334087128479,
      "loss": 1.1954,
      "step": 14248
    },
    {
      "epoch": 1.2259313430267573,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008592145997001021,
      "loss": 1.0936,
      "step": 14249
    },
    {
      "epoch": 1.226017379334079,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008591957896367236,
      "loss": 0.9432,
      "step": 14250
    },
    {
      "epoch": 1.2261034156414006,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008591769785227672,
      "loss": 1.4469,
      "step": 14251
    },
    {
      "epoch": 1.2261894519487224,
      "grad_norm": 1.765625,
      "learning_rate": 0.000859158166358288,
      "loss": 1.1639,
      "step": 14252
    },
    {
      "epoch": 1.226275488256044,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008591393531433412,
      "loss": 1.4049,
      "step": 14253
    },
    {
      "epoch": 1.2263615245633657,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008591205388779813,
      "loss": 1.0967,
      "step": 14254
    },
    {
      "epoch": 1.2264475608706875,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008591017235622638,
      "loss": 0.8969,
      "step": 14255
    },
    {
      "epoch": 1.2265335971780091,
      "grad_norm": 2.03125,
      "learning_rate": 0.0008590829071962437,
      "loss": 1.5219,
      "step": 14256
    },
    {
      "epoch": 1.2266196334853308,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008590640897799759,
      "loss": 1.6679,
      "step": 14257
    },
    {
      "epoch": 1.2267056697926524,
      "grad_norm": 1.890625,
      "learning_rate": 0.0008590452713135154,
      "loss": 1.6579,
      "step": 14258
    },
    {
      "epoch": 1.2267917060999742,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0008590264517969173,
      "loss": 1.6535,
      "step": 14259
    },
    {
      "epoch": 1.2268777424072959,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008590076312302368,
      "loss": 2.0727,
      "step": 14260
    },
    {
      "epoch": 1.2269637787146175,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008589888096135287,
      "loss": 1.287,
      "step": 14261
    },
    {
      "epoch": 1.2270498150219393,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008589699869468481,
      "loss": 1.3379,
      "step": 14262
    },
    {
      "epoch": 1.227135851329261,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008589511632302501,
      "loss": 1.4363,
      "step": 14263
    },
    {
      "epoch": 1.2272218876365826,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00085893233846379,
      "loss": 1.5535,
      "step": 14264
    },
    {
      "epoch": 1.2273079239439044,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008589135126475223,
      "loss": 1.5003,
      "step": 14265
    },
    {
      "epoch": 1.227393960251226,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008588946857815026,
      "loss": 0.9168,
      "step": 14266
    },
    {
      "epoch": 1.2274799965585477,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008588758578657856,
      "loss": 1.2533,
      "step": 14267
    },
    {
      "epoch": 1.2275660328658695,
      "grad_norm": 1.0,
      "learning_rate": 0.0008588570289004266,
      "loss": 0.8767,
      "step": 14268
    },
    {
      "epoch": 1.2276520691731911,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008588381988854806,
      "loss": 1.0473,
      "step": 14269
    },
    {
      "epoch": 1.2277381054805128,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008588193678210025,
      "loss": 1.433,
      "step": 14270
    },
    {
      "epoch": 1.2278241417878344,
      "grad_norm": 1.25,
      "learning_rate": 0.0008588005357070478,
      "loss": 1.5942,
      "step": 14271
    },
    {
      "epoch": 1.2279101780951562,
      "grad_norm": 1.015625,
      "learning_rate": 0.000858781702543671,
      "loss": 1.0323,
      "step": 14272
    },
    {
      "epoch": 1.2279962144024779,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008587628683309278,
      "loss": 1.612,
      "step": 14273
    },
    {
      "epoch": 1.2280822507097995,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008587440330688729,
      "loss": 1.0597,
      "step": 14274
    },
    {
      "epoch": 1.2281682870171213,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008587251967575614,
      "loss": 1.7471,
      "step": 14275
    },
    {
      "epoch": 1.228254323324443,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008587063593970485,
      "loss": 1.2962,
      "step": 14276
    },
    {
      "epoch": 1.2283403596317646,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008586875209873892,
      "loss": 1.4738,
      "step": 14277
    },
    {
      "epoch": 1.2284263959390862,
      "grad_norm": 1.375,
      "learning_rate": 0.0008586686815286388,
      "loss": 2.031,
      "step": 14278
    },
    {
      "epoch": 1.228512432246408,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008586498410208521,
      "loss": 1.7566,
      "step": 14279
    },
    {
      "epoch": 1.2285984685537297,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0008586309994640844,
      "loss": 1.2082,
      "step": 14280
    },
    {
      "epoch": 1.2286845048610513,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008586121568583909,
      "loss": 1.1184,
      "step": 14281
    },
    {
      "epoch": 1.2287705411683731,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008585933132038266,
      "loss": 1.9376,
      "step": 14282
    },
    {
      "epoch": 1.2288565774756948,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008585744685004464,
      "loss": 1.2224,
      "step": 14283
    },
    {
      "epoch": 1.2289426137830164,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008585556227483058,
      "loss": 1.5583,
      "step": 14284
    },
    {
      "epoch": 1.229028650090338,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008585367759474597,
      "loss": 1.0115,
      "step": 14285
    },
    {
      "epoch": 1.2291146863976599,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008585179280979631,
      "loss": 1.7185,
      "step": 14286
    },
    {
      "epoch": 1.2292007227049815,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008584990791998715,
      "loss": 1.5369,
      "step": 14287
    },
    {
      "epoch": 1.229286759012303,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008584802292532399,
      "loss": 1.5169,
      "step": 14288
    },
    {
      "epoch": 1.229372795319625,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008584613782581231,
      "loss": 1.7941,
      "step": 14289
    },
    {
      "epoch": 1.2294588316269466,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008584425262145765,
      "loss": 1.8159,
      "step": 14290
    },
    {
      "epoch": 1.2295448679342682,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008584236731226554,
      "loss": 1.4952,
      "step": 14291
    },
    {
      "epoch": 1.2296309042415898,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008584048189824148,
      "loss": 1.8944,
      "step": 14292
    },
    {
      "epoch": 1.2297169405489117,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008583859637939095,
      "loss": 1.2732,
      "step": 14293
    },
    {
      "epoch": 1.2298029768562333,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008583671075571953,
      "loss": 1.5585,
      "step": 14294
    },
    {
      "epoch": 1.2298890131635551,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008583482502723267,
      "loss": 1.44,
      "step": 14295
    },
    {
      "epoch": 1.2299750494708768,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008583293919393593,
      "loss": 1.7594,
      "step": 14296
    },
    {
      "epoch": 1.2300610857781984,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008583105325583483,
      "loss": 1.4065,
      "step": 14297
    },
    {
      "epoch": 1.23014712208552,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008582916721293484,
      "loss": 0.6519,
      "step": 14298
    },
    {
      "epoch": 1.2302331583928419,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008582728106524152,
      "loss": 1.9001,
      "step": 14299
    },
    {
      "epoch": 1.2303191947001635,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008582539481276036,
      "loss": 0.8056,
      "step": 14300
    },
    {
      "epoch": 1.2304052310074851,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008582350845549689,
      "loss": 1.0136,
      "step": 14301
    },
    {
      "epoch": 1.230491267314807,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008582162199345663,
      "loss": 1.2342,
      "step": 14302
    },
    {
      "epoch": 1.2305773036221286,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008581973542664509,
      "loss": 0.8128,
      "step": 14303
    },
    {
      "epoch": 1.2306633399294502,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008581784875506779,
      "loss": 1.4602,
      "step": 14304
    },
    {
      "epoch": 1.2307493762367718,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008581596197873026,
      "loss": 1.3861,
      "step": 14305
    },
    {
      "epoch": 1.2308354125440937,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008581407509763799,
      "loss": 1.4888,
      "step": 14306
    },
    {
      "epoch": 1.2309214488514153,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008581218811179654,
      "loss": 1.1629,
      "step": 14307
    },
    {
      "epoch": 1.231007485158737,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008581030102121137,
      "loss": 1.3461,
      "step": 14308
    },
    {
      "epoch": 1.2310935214660588,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008580841382588806,
      "loss": 1.8937,
      "step": 14309
    },
    {
      "epoch": 1.2311795577733804,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008580652652583209,
      "loss": 1.6109,
      "step": 14310
    },
    {
      "epoch": 1.231265594080702,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008580463912104901,
      "loss": 1.669,
      "step": 14311
    },
    {
      "epoch": 1.2313516303880236,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008580275161154432,
      "loss": 2.0909,
      "step": 14312
    },
    {
      "epoch": 1.2314376666953455,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008580086399732354,
      "loss": 0.8105,
      "step": 14313
    },
    {
      "epoch": 1.2315237030026671,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008579897627839219,
      "loss": 1.4296,
      "step": 14314
    },
    {
      "epoch": 1.2316097393099887,
      "grad_norm": 1.375,
      "learning_rate": 0.000857970884547558,
      "loss": 1.4735,
      "step": 14315
    },
    {
      "epoch": 1.2316957756173106,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000857952005264199,
      "loss": 1.5061,
      "step": 14316
    },
    {
      "epoch": 1.2317818119246322,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008579331249338999,
      "loss": 2.0149,
      "step": 14317
    },
    {
      "epoch": 1.2318678482319538,
      "grad_norm": 1.25,
      "learning_rate": 0.000857914243556716,
      "loss": 1.0129,
      "step": 14318
    },
    {
      "epoch": 1.2319538845392755,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008578953611327025,
      "loss": 1.1503,
      "step": 14319
    },
    {
      "epoch": 1.2320399208465973,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008578764776619148,
      "loss": 0.7961,
      "step": 14320
    },
    {
      "epoch": 1.232125957153919,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008578575931444079,
      "loss": 1.3964,
      "step": 14321
    },
    {
      "epoch": 1.2322119934612406,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008578387075802373,
      "loss": 1.9084,
      "step": 14322
    },
    {
      "epoch": 1.2322980297685624,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008578198209694579,
      "loss": 1.2283,
      "step": 14323
    },
    {
      "epoch": 1.232384066075884,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008578009333121253,
      "loss": 2.0746,
      "step": 14324
    },
    {
      "epoch": 1.2324701023832056,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008577820446082945,
      "loss": 1.3235,
      "step": 14325
    },
    {
      "epoch": 1.2325561386905275,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008577631548580207,
      "loss": 1.1236,
      "step": 14326
    },
    {
      "epoch": 1.2326421749978491,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008577442640613593,
      "loss": 1.431,
      "step": 14327
    },
    {
      "epoch": 1.2327282113051707,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008577253722183656,
      "loss": 0.9862,
      "step": 14328
    },
    {
      "epoch": 1.2328142476124926,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008577064793290945,
      "loss": 0.7448,
      "step": 14329
    },
    {
      "epoch": 1.2329002839198142,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008576875853936017,
      "loss": 1.2461,
      "step": 14330
    },
    {
      "epoch": 1.2329863202271358,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008576686904119423,
      "loss": 2.2278,
      "step": 14331
    },
    {
      "epoch": 1.2330723565344575,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008576497943841714,
      "loss": 1.4628,
      "step": 14332
    },
    {
      "epoch": 1.2331583928417793,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008576308973103446,
      "loss": 1.2607,
      "step": 14333
    },
    {
      "epoch": 1.233244429149101,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008576119991905169,
      "loss": 1.1097,
      "step": 14334
    },
    {
      "epoch": 1.2333304654564226,
      "grad_norm": 3.09375,
      "learning_rate": 0.0008575931000247436,
      "loss": 1.5825,
      "step": 14335
    },
    {
      "epoch": 1.2334165017637444,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008575741998130802,
      "loss": 0.4789,
      "step": 14336
    },
    {
      "epoch": 1.233502538071066,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008575552985555817,
      "loss": 1.5883,
      "step": 14337
    },
    {
      "epoch": 1.2335885743783876,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008575363962523035,
      "loss": 1.2369,
      "step": 14338
    },
    {
      "epoch": 1.2336746106857093,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000857517492903301,
      "loss": 1.1652,
      "step": 14339
    },
    {
      "epoch": 1.2337606469930311,
      "grad_norm": 1.8046875,
      "learning_rate": 0.000857498588508629,
      "loss": 1.7282,
      "step": 14340
    },
    {
      "epoch": 1.2338466833003527,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008574796830683436,
      "loss": 0.6704,
      "step": 14341
    },
    {
      "epoch": 1.2339327196076744,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008574607765824994,
      "loss": 1.0736,
      "step": 14342
    },
    {
      "epoch": 1.2340187559149962,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008574418690511519,
      "loss": 1.1771,
      "step": 14343
    },
    {
      "epoch": 1.2341047922223178,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008574229604743567,
      "loss": 1.8053,
      "step": 14344
    },
    {
      "epoch": 1.2341908285296395,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008574040508521686,
      "loss": 0.9718,
      "step": 14345
    },
    {
      "epoch": 1.234276864836961,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008573851401846433,
      "loss": 1.0796,
      "step": 14346
    },
    {
      "epoch": 1.234362901144283,
      "grad_norm": 1.203125,
      "learning_rate": 0.000857366228471836,
      "loss": 1.5126,
      "step": 14347
    },
    {
      "epoch": 1.2344489374516046,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008573473157138018,
      "loss": 1.3227,
      "step": 14348
    },
    {
      "epoch": 1.2345349737589262,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008573284019105964,
      "loss": 1.0801,
      "step": 14349
    },
    {
      "epoch": 1.234621010066248,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008573094870622748,
      "loss": 1.0854,
      "step": 14350
    },
    {
      "epoch": 1.2347070463735696,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008572905711688925,
      "loss": 1.4859,
      "step": 14351
    },
    {
      "epoch": 1.2347930826808913,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008572716542305048,
      "loss": 1.3348,
      "step": 14352
    },
    {
      "epoch": 1.2348791189882131,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008572527362471669,
      "loss": 0.8172,
      "step": 14353
    },
    {
      "epoch": 1.2349651552955347,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008572338172189343,
      "loss": 1.6948,
      "step": 14354
    },
    {
      "epoch": 1.2350511916028564,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008572148971458622,
      "loss": 1.2143,
      "step": 14355
    },
    {
      "epoch": 1.2351372279101782,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008571959760280062,
      "loss": 1.4244,
      "step": 14356
    },
    {
      "epoch": 1.2352232642174998,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008571770538654212,
      "loss": 1.4432,
      "step": 14357
    },
    {
      "epoch": 1.2353093005248215,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008571581306581628,
      "loss": 1.0384,
      "step": 14358
    },
    {
      "epoch": 1.235395336832143,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008571392064062864,
      "loss": 0.8167,
      "step": 14359
    },
    {
      "epoch": 1.235481373139465,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008571202811098473,
      "loss": 0.6177,
      "step": 14360
    },
    {
      "epoch": 1.2355674094467866,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008571013547689009,
      "loss": 1.5538,
      "step": 14361
    },
    {
      "epoch": 1.2356534457541082,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008570824273835024,
      "loss": 1.7838,
      "step": 14362
    },
    {
      "epoch": 1.23573948206143,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008570634989537073,
      "loss": 1.6576,
      "step": 14363
    },
    {
      "epoch": 1.2358255183687517,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008570445694795708,
      "loss": 1.0662,
      "step": 14364
    },
    {
      "epoch": 1.2359115546760733,
      "grad_norm": 1.25,
      "learning_rate": 0.0008570256389611485,
      "loss": 1.4572,
      "step": 14365
    },
    {
      "epoch": 1.235997590983395,
      "grad_norm": 1.921875,
      "learning_rate": 0.0008570067073984955,
      "loss": 1.5333,
      "step": 14366
    },
    {
      "epoch": 1.2360836272907167,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008569877747916675,
      "loss": 1.9043,
      "step": 14367
    },
    {
      "epoch": 1.2361696635980384,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008569688411407195,
      "loss": 0.9443,
      "step": 14368
    },
    {
      "epoch": 1.23625569990536,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008569499064457072,
      "loss": 1.3694,
      "step": 14369
    },
    {
      "epoch": 1.2363417362126818,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008569309707066858,
      "loss": 1.7447,
      "step": 14370
    },
    {
      "epoch": 1.2364277725200035,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008569120339237108,
      "loss": 1.1517,
      "step": 14371
    },
    {
      "epoch": 1.236513808827325,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008568930960968374,
      "loss": 1.3197,
      "step": 14372
    },
    {
      "epoch": 1.2365998451346467,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008568741572261212,
      "loss": 1.1339,
      "step": 14373
    },
    {
      "epoch": 1.2366858814419686,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008568552173116174,
      "loss": 1.07,
      "step": 14374
    },
    {
      "epoch": 1.2367719177492902,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008568362763533816,
      "loss": 1.3736,
      "step": 14375
    },
    {
      "epoch": 1.2368579540566118,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000856817334351469,
      "loss": 2.039,
      "step": 14376
    },
    {
      "epoch": 1.2369439903639337,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008567983913059352,
      "loss": 1.417,
      "step": 14377
    },
    {
      "epoch": 1.2370300266712553,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008567794472168353,
      "loss": 0.5668,
      "step": 14378
    },
    {
      "epoch": 1.237116062978577,
      "grad_norm": 0.96875,
      "learning_rate": 0.000856760502084225,
      "loss": 1.1243,
      "step": 14379
    },
    {
      "epoch": 1.2372020992858985,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008567415559081598,
      "loss": 1.2355,
      "step": 14380
    },
    {
      "epoch": 1.2372881355932204,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008567226086886948,
      "loss": 1.1075,
      "step": 14381
    },
    {
      "epoch": 1.237374171900542,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008567036604258855,
      "loss": 1.6135,
      "step": 14382
    },
    {
      "epoch": 1.2374602082078638,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008566847111197875,
      "loss": 1.3798,
      "step": 14383
    },
    {
      "epoch": 1.2375462445151855,
      "grad_norm": 1.7421875,
      "learning_rate": 0.000856665760770456,
      "loss": 1.1227,
      "step": 14384
    },
    {
      "epoch": 1.237632280822507,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008566468093779465,
      "loss": 1.1586,
      "step": 14385
    },
    {
      "epoch": 1.2377183171298287,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008566278569423145,
      "loss": 1.2196,
      "step": 14386
    },
    {
      "epoch": 1.2378043534371506,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008566089034636154,
      "loss": 1.0549,
      "step": 14387
    },
    {
      "epoch": 1.2378903897444722,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008565899489419045,
      "loss": 1.1253,
      "step": 14388
    },
    {
      "epoch": 1.2379764260517938,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008565709933772373,
      "loss": 1.6201,
      "step": 14389
    },
    {
      "epoch": 1.2380624623591157,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008565520367696695,
      "loss": 1.205,
      "step": 14390
    },
    {
      "epoch": 1.2381484986664373,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008565330791192561,
      "loss": 2.0585,
      "step": 14391
    },
    {
      "epoch": 1.238234534973759,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008565141204260529,
      "loss": 1.2515,
      "step": 14392
    },
    {
      "epoch": 1.2383205712810805,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008564951606901154,
      "loss": 1.3739,
      "step": 14393
    },
    {
      "epoch": 1.2384066075884024,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008564761999114987,
      "loss": 1.0164,
      "step": 14394
    },
    {
      "epoch": 1.238492643895724,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008564572380902585,
      "loss": 1.2564,
      "step": 14395
    },
    {
      "epoch": 1.2385786802030456,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008564382752264501,
      "loss": 2.0344,
      "step": 14396
    },
    {
      "epoch": 1.2386647165103675,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008564193113201293,
      "loss": 1.0461,
      "step": 14397
    },
    {
      "epoch": 1.238750752817689,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008564003463713512,
      "loss": 1.3348,
      "step": 14398
    },
    {
      "epoch": 1.2388367891250107,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008563813803801713,
      "loss": 1.1175,
      "step": 14399
    },
    {
      "epoch": 1.2389228254323323,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008563624133466454,
      "loss": 1.0685,
      "step": 14400
    },
    {
      "epoch": 1.2390088617396542,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008563434452708285,
      "loss": 1.6326,
      "step": 14401
    },
    {
      "epoch": 1.2390948980469758,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008563244761527765,
      "loss": 1.1073,
      "step": 14402
    },
    {
      "epoch": 1.2391809343542974,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008563055059925447,
      "loss": 1.6588,
      "step": 14403
    },
    {
      "epoch": 1.2392669706616193,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008562865347901886,
      "loss": 1.7023,
      "step": 14404
    },
    {
      "epoch": 1.239353006968941,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008562675625457636,
      "loss": 1.1671,
      "step": 14405
    },
    {
      "epoch": 1.2394390432762625,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008562485892593252,
      "loss": 0.8166,
      "step": 14406
    },
    {
      "epoch": 1.2395250795835842,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008562296149309292,
      "loss": 1.8469,
      "step": 14407
    },
    {
      "epoch": 1.239611115890906,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008562106395606307,
      "loss": 1.2521,
      "step": 14408
    },
    {
      "epoch": 1.2396971521982276,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008561916631484854,
      "loss": 1.221,
      "step": 14409
    },
    {
      "epoch": 1.2397831885055492,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008561726856945488,
      "loss": 1.5091,
      "step": 14410
    },
    {
      "epoch": 1.239869224812871,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008561537071988763,
      "loss": 1.3235,
      "step": 14411
    },
    {
      "epoch": 1.2399552611201927,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008561347276615234,
      "loss": 1.4925,
      "step": 14412
    },
    {
      "epoch": 1.2400412974275143,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008561157470825458,
      "loss": 1.0208,
      "step": 14413
    },
    {
      "epoch": 1.2401273337348362,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008560967654619989,
      "loss": 0.9976,
      "step": 14414
    },
    {
      "epoch": 1.2402133700421578,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008560777827999382,
      "loss": 1.113,
      "step": 14415
    },
    {
      "epoch": 1.2402994063494794,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008560587990964193,
      "loss": 1.5985,
      "step": 14416
    },
    {
      "epoch": 1.2403854426568013,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008560398143514975,
      "loss": 0.9966,
      "step": 14417
    },
    {
      "epoch": 1.240471478964123,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008560208285652287,
      "loss": 0.813,
      "step": 14418
    },
    {
      "epoch": 1.2405575152714445,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008560018417376681,
      "loss": 1.3231,
      "step": 14419
    },
    {
      "epoch": 1.2406435515787662,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008559828538688713,
      "loss": 1.2814,
      "step": 14420
    },
    {
      "epoch": 1.240729587886088,
      "grad_norm": 4.96875,
      "learning_rate": 0.0008559638649588938,
      "loss": 1.69,
      "step": 14421
    },
    {
      "epoch": 1.2408156241934096,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008559448750077915,
      "loss": 1.6814,
      "step": 14422
    },
    {
      "epoch": 1.2409016605007313,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008559258840156195,
      "loss": 1.2537,
      "step": 14423
    },
    {
      "epoch": 1.240987696808053,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008559068919824336,
      "loss": 1.1471,
      "step": 14424
    },
    {
      "epoch": 1.2410737331153747,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008558878989082892,
      "loss": 0.7791,
      "step": 14425
    },
    {
      "epoch": 1.2411597694226963,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008558689047932419,
      "loss": 0.9969,
      "step": 14426
    },
    {
      "epoch": 1.241245805730018,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008558499096373473,
      "loss": 0.8855,
      "step": 14427
    },
    {
      "epoch": 1.2413318420373398,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000855830913440661,
      "loss": 1.4532,
      "step": 14428
    },
    {
      "epoch": 1.2414178783446614,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008558119162032383,
      "loss": 2.1699,
      "step": 14429
    },
    {
      "epoch": 1.241503914651983,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008557929179251352,
      "loss": 1.1184,
      "step": 14430
    },
    {
      "epoch": 1.241589950959305,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008557739186064066,
      "loss": 2.0173,
      "step": 14431
    },
    {
      "epoch": 1.2416759872666265,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008557549182471089,
      "loss": 1.2183,
      "step": 14432
    },
    {
      "epoch": 1.2417620235739482,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000855735916847297,
      "loss": 1.1562,
      "step": 14433
    },
    {
      "epoch": 1.2418480598812698,
      "grad_norm": 1.203125,
      "learning_rate": 0.000855716914407027,
      "loss": 0.7956,
      "step": 14434
    },
    {
      "epoch": 1.2419340961885916,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008556979109263539,
      "loss": 0.9518,
      "step": 14435
    },
    {
      "epoch": 1.2420201324959133,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008556789064053337,
      "loss": 0.5845,
      "step": 14436
    },
    {
      "epoch": 1.2421061688032349,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008556599008440219,
      "loss": 1.2838,
      "step": 14437
    },
    {
      "epoch": 1.2421922051105567,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008556408942424741,
      "loss": 1.6601,
      "step": 14438
    },
    {
      "epoch": 1.2422782414178783,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008556218866007458,
      "loss": 0.9825,
      "step": 14439
    },
    {
      "epoch": 1.2423642777252,
      "grad_norm": 4.5625,
      "learning_rate": 0.0008556028779188927,
      "loss": 1.7258,
      "step": 14440
    },
    {
      "epoch": 1.2424503140325218,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008555838681969703,
      "loss": 1.767,
      "step": 14441
    },
    {
      "epoch": 1.2425363503398434,
      "grad_norm": 2.21875,
      "learning_rate": 0.0008555648574350342,
      "loss": 1.6916,
      "step": 14442
    },
    {
      "epoch": 1.242622386647165,
      "grad_norm": 1.8671875,
      "learning_rate": 0.00085554584563314,
      "loss": 1.3717,
      "step": 14443
    },
    {
      "epoch": 1.242708422954487,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008555268327913435,
      "loss": 1.2987,
      "step": 14444
    },
    {
      "epoch": 1.2427944592618085,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008555078189097001,
      "loss": 1.3476,
      "step": 14445
    },
    {
      "epoch": 1.2428804955691302,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008554888039882655,
      "loss": 1.6706,
      "step": 14446
    },
    {
      "epoch": 1.2429665318764518,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008554697880270952,
      "loss": 1.137,
      "step": 14447
    },
    {
      "epoch": 1.2430525681837736,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008554507710262449,
      "loss": 1.7471,
      "step": 14448
    },
    {
      "epoch": 1.2431386044910953,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008554317529857704,
      "loss": 1.5972,
      "step": 14449
    },
    {
      "epoch": 1.2432246407984169,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000855412733905727,
      "loss": 1.3161,
      "step": 14450
    },
    {
      "epoch": 1.2433106771057387,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008553937137861704,
      "loss": 2.1371,
      "step": 14451
    },
    {
      "epoch": 1.2433967134130604,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008553746926271566,
      "loss": 1.1518,
      "step": 14452
    },
    {
      "epoch": 1.243482749720382,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008553556704287408,
      "loss": 1.01,
      "step": 14453
    },
    {
      "epoch": 1.2435687860277036,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008553366471909787,
      "loss": 1.316,
      "step": 14454
    },
    {
      "epoch": 1.2436548223350254,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008553176229139261,
      "loss": 1.1238,
      "step": 14455
    },
    {
      "epoch": 1.243740858642347,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008552985975976384,
      "loss": 1.6353,
      "step": 14456
    },
    {
      "epoch": 1.2438268949496687,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008552795712421715,
      "loss": 1.5007,
      "step": 14457
    },
    {
      "epoch": 1.2439129312569905,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008552605438475812,
      "loss": 1.3901,
      "step": 14458
    },
    {
      "epoch": 1.2439989675643122,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008552415154139226,
      "loss": 1.568,
      "step": 14459
    },
    {
      "epoch": 1.2440850038716338,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008552224859412516,
      "loss": 1.1358,
      "step": 14460
    },
    {
      "epoch": 1.2441710401789554,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000855203455429624,
      "loss": 1.3075,
      "step": 14461
    },
    {
      "epoch": 1.2442570764862773,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008551844238790956,
      "loss": 1.8168,
      "step": 14462
    },
    {
      "epoch": 1.2443431127935989,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008551653912897216,
      "loss": 0.8584,
      "step": 14463
    },
    {
      "epoch": 1.2444291491009205,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008551463576615579,
      "loss": 1.3014,
      "step": 14464
    },
    {
      "epoch": 1.2445151854082424,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008551273229946603,
      "loss": 0.8927,
      "step": 14465
    },
    {
      "epoch": 1.244601221715564,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008551082872890843,
      "loss": 1.0834,
      "step": 14466
    },
    {
      "epoch": 1.2446872580228856,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008550892505448856,
      "loss": 1.3794,
      "step": 14467
    },
    {
      "epoch": 1.2447732943302072,
      "grad_norm": 1.375,
      "learning_rate": 0.00085507021276212,
      "loss": 1.6275,
      "step": 14468
    },
    {
      "epoch": 1.244859330637529,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008550511739408429,
      "loss": 1.6567,
      "step": 14469
    },
    {
      "epoch": 1.2449453669448507,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008550321340811104,
      "loss": 0.6554,
      "step": 14470
    },
    {
      "epoch": 1.2450314032521725,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008550130931829778,
      "loss": 1.5595,
      "step": 14471
    },
    {
      "epoch": 1.2451174395594942,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008549940512465008,
      "loss": 0.8387,
      "step": 14472
    },
    {
      "epoch": 1.2452034758668158,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008549750082717356,
      "loss": 2.2897,
      "step": 14473
    },
    {
      "epoch": 1.2452895121741374,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008549559642587373,
      "loss": 1.0913,
      "step": 14474
    },
    {
      "epoch": 1.2453755484814593,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000854936919207562,
      "loss": 1.4698,
      "step": 14475
    },
    {
      "epoch": 1.2454615847887809,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008549178731182651,
      "loss": 1.3767,
      "step": 14476
    },
    {
      "epoch": 1.2455476210961025,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008548988259909025,
      "loss": 0.7482,
      "step": 14477
    },
    {
      "epoch": 1.2456336574034244,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008548797778255299,
      "loss": 1.5151,
      "step": 14478
    },
    {
      "epoch": 1.245719693710746,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008548607286222029,
      "loss": 1.1273,
      "step": 14479
    },
    {
      "epoch": 1.2458057300180676,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008548416783809772,
      "loss": 1.5413,
      "step": 14480
    },
    {
      "epoch": 1.2458917663253892,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008548226271019088,
      "loss": 1.3133,
      "step": 14481
    },
    {
      "epoch": 1.245977802632711,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008548035747850531,
      "loss": 1.4558,
      "step": 14482
    },
    {
      "epoch": 1.2460638389400327,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008547845214304661,
      "loss": 0.9258,
      "step": 14483
    },
    {
      "epoch": 1.2461498752473543,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008547654670382032,
      "loss": 0.8841,
      "step": 14484
    },
    {
      "epoch": 1.2462359115546762,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008547464116083203,
      "loss": 0.9632,
      "step": 14485
    },
    {
      "epoch": 1.2463219478619978,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008547273551408733,
      "loss": 1.0763,
      "step": 14486
    },
    {
      "epoch": 1.2464079841693194,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008547082976359177,
      "loss": 1.4605,
      "step": 14487
    },
    {
      "epoch": 1.246494020476641,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008546892390935093,
      "loss": 0.9326,
      "step": 14488
    },
    {
      "epoch": 1.2465800567839629,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008546701795137039,
      "loss": 1.5809,
      "step": 14489
    },
    {
      "epoch": 1.2466660930912845,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008546511188965571,
      "loss": 1.5454,
      "step": 14490
    },
    {
      "epoch": 1.2467521293986061,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008546320572421249,
      "loss": 1.1173,
      "step": 14491
    },
    {
      "epoch": 1.246838165705928,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008546129945504629,
      "loss": 0.8502,
      "step": 14492
    },
    {
      "epoch": 1.2469242020132496,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008545939308216267,
      "loss": 1.6648,
      "step": 14493
    },
    {
      "epoch": 1.2470102383205712,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008545748660556723,
      "loss": 1.2546,
      "step": 14494
    },
    {
      "epoch": 1.2470962746278929,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008545558002526555,
      "loss": 1.4215,
      "step": 14495
    },
    {
      "epoch": 1.2471823109352147,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008545367334126318,
      "loss": 1.4086,
      "step": 14496
    },
    {
      "epoch": 1.2472683472425363,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000854517665535657,
      "loss": 0.8019,
      "step": 14497
    },
    {
      "epoch": 1.247354383549858,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008544985966217871,
      "loss": 1.1237,
      "step": 14498
    },
    {
      "epoch": 1.2474404198571798,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008544795266710778,
      "loss": 0.5431,
      "step": 14499
    },
    {
      "epoch": 1.2475264561645014,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008544604556835848,
      "loss": 1.1736,
      "step": 14500
    },
    {
      "epoch": 1.247612492471823,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0008544413836593639,
      "loss": 1.0052,
      "step": 14501
    },
    {
      "epoch": 1.2476985287791449,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008544223105984709,
      "loss": 1.4103,
      "step": 14502
    },
    {
      "epoch": 1.2477845650864665,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008544032365009615,
      "loss": 1.6046,
      "step": 14503
    },
    {
      "epoch": 1.2478706013937881,
      "grad_norm": 1.5,
      "learning_rate": 0.0008543841613668917,
      "loss": 1.1745,
      "step": 14504
    },
    {
      "epoch": 1.24795663770111,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000854365085196317,
      "loss": 0.7264,
      "step": 14505
    },
    {
      "epoch": 1.2480426740084316,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008543460079892934,
      "loss": 1.7504,
      "step": 14506
    },
    {
      "epoch": 1.2481287103157532,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008543269297458766,
      "loss": 1.5723,
      "step": 14507
    },
    {
      "epoch": 1.2482147466230749,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008543078504661226,
      "loss": 1.3648,
      "step": 14508
    },
    {
      "epoch": 1.2483007829303967,
      "grad_norm": 1.5,
      "learning_rate": 0.000854288770150087,
      "loss": 1.0797,
      "step": 14509
    },
    {
      "epoch": 1.2483868192377183,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008542696887978255,
      "loss": 0.5983,
      "step": 14510
    },
    {
      "epoch": 1.24847285554504,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008542506064093942,
      "loss": 1.0612,
      "step": 14511
    },
    {
      "epoch": 1.2485588918523618,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0008542315229848488,
      "loss": 0.8254,
      "step": 14512
    },
    {
      "epoch": 1.2486449281596834,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008542124385242449,
      "loss": 1.4657,
      "step": 14513
    },
    {
      "epoch": 1.248730964467005,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008541933530276386,
      "loss": 1.8692,
      "step": 14514
    },
    {
      "epoch": 1.2488170007743267,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008541742664950857,
      "loss": 1.2851,
      "step": 14515
    },
    {
      "epoch": 1.2489030370816485,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008541551789266419,
      "loss": 1.6238,
      "step": 14516
    },
    {
      "epoch": 1.2489890733889701,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008541360903223632,
      "loss": 0.9955,
      "step": 14517
    },
    {
      "epoch": 1.2490751096962918,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008541170006823052,
      "loss": 1.254,
      "step": 14518
    },
    {
      "epoch": 1.2491611460036136,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008540979100065239,
      "loss": 1.2899,
      "step": 14519
    },
    {
      "epoch": 1.2492471823109352,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0008540788182950749,
      "loss": 1.0933,
      "step": 14520
    },
    {
      "epoch": 1.2493332186182569,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008540597255480144,
      "loss": 1.2497,
      "step": 14521
    },
    {
      "epoch": 1.2494192549255785,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008540406317653981,
      "loss": 1.603,
      "step": 14522
    },
    {
      "epoch": 1.2495052912329003,
      "grad_norm": 1.9375,
      "learning_rate": 0.0008540215369472817,
      "loss": 1.7274,
      "step": 14523
    },
    {
      "epoch": 1.249591327540222,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008540024410937211,
      "loss": 1.245,
      "step": 14524
    },
    {
      "epoch": 1.2496773638475436,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008539833442047723,
      "loss": 0.998,
      "step": 14525
    },
    {
      "epoch": 1.2497634001548654,
      "grad_norm": 1.296875,
      "learning_rate": 0.000853964246280491,
      "loss": 1.372,
      "step": 14526
    },
    {
      "epoch": 1.249849436462187,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008539451473209332,
      "loss": 1.0295,
      "step": 14527
    },
    {
      "epoch": 1.2499354727695087,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008539260473261548,
      "loss": 1.2083,
      "step": 14528
    },
    {
      "epoch": 1.2500215090768303,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008539069462962114,
      "loss": 0.9832,
      "step": 14529
    },
    {
      "epoch": 1.2501075453841521,
      "grad_norm": 1.25,
      "learning_rate": 0.0008538878442311589,
      "loss": 1.4936,
      "step": 14530
    },
    {
      "epoch": 1.2501075453841521,
      "eval_loss": 1.2660362720489502,
      "eval_runtime": 354.4852,
      "eval_samples_per_second": 13.115,
      "eval_steps_per_second": 1.642,
      "step": 14530
    },
    {
      "epoch": 1.2501935816914738,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008538687411310535,
      "loss": 0.4562,
      "step": 14531
    },
    {
      "epoch": 1.2502796179987956,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008538496369959507,
      "loss": 1.3103,
      "step": 14532
    },
    {
      "epoch": 1.2503656543061172,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008538305318259066,
      "loss": 0.9855,
      "step": 14533
    },
    {
      "epoch": 1.2504516906134389,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000853811425620977,
      "loss": 0.7914,
      "step": 14534
    },
    {
      "epoch": 1.2505377269207605,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008537923183812177,
      "loss": 0.7033,
      "step": 14535
    },
    {
      "epoch": 1.2506237632280823,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008537732101066848,
      "loss": 1.5088,
      "step": 14536
    },
    {
      "epoch": 1.250709799535404,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008537541007974341,
      "loss": 1.1106,
      "step": 14537
    },
    {
      "epoch": 1.2507958358427256,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008537349904535215,
      "loss": 1.8085,
      "step": 14538
    },
    {
      "epoch": 1.2508818721500474,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008537158790750027,
      "loss": 1.9453,
      "step": 14539
    },
    {
      "epoch": 1.250967908457369,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008536967666619337,
      "loss": 1.0689,
      "step": 14540
    },
    {
      "epoch": 1.2510539447646907,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008536776532143707,
      "loss": 1.0678,
      "step": 14541
    },
    {
      "epoch": 1.2511399810720123,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008536585387323692,
      "loss": 1.1262,
      "step": 14542
    },
    {
      "epoch": 1.2512260173793341,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008536394232159852,
      "loss": 1.1391,
      "step": 14543
    },
    {
      "epoch": 1.2513120536866558,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008536203066652747,
      "loss": 1.4124,
      "step": 14544
    },
    {
      "epoch": 1.2513980899939774,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008536011890802935,
      "loss": 1.3236,
      "step": 14545
    },
    {
      "epoch": 1.2514841263012992,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008535820704610977,
      "loss": 1.0905,
      "step": 14546
    },
    {
      "epoch": 1.2515701626086209,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008535629508077432,
      "loss": 0.9834,
      "step": 14547
    },
    {
      "epoch": 1.2516561989159425,
      "grad_norm": 2.03125,
      "learning_rate": 0.0008535438301202857,
      "loss": 0.8297,
      "step": 14548
    },
    {
      "epoch": 1.251742235223264,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008535247083987813,
      "loss": 1.2162,
      "step": 14549
    },
    {
      "epoch": 1.251828271530586,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008535055856432859,
      "loss": 1.0345,
      "step": 14550
    },
    {
      "epoch": 1.2519143078379076,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008534864618538554,
      "loss": 0.6533,
      "step": 14551
    },
    {
      "epoch": 1.2520003441452292,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008534673370305457,
      "loss": 1.5909,
      "step": 14552
    },
    {
      "epoch": 1.252086380452551,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008534482111734128,
      "loss": 1.2149,
      "step": 14553
    },
    {
      "epoch": 1.2521724167598727,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008534290842825127,
      "loss": 1.1431,
      "step": 14554
    },
    {
      "epoch": 1.2522584530671943,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008534099563579011,
      "loss": 1.0756,
      "step": 14555
    },
    {
      "epoch": 1.252344489374516,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008533908273996343,
      "loss": 1.6529,
      "step": 14556
    },
    {
      "epoch": 1.2524305256818378,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008533716974077678,
      "loss": 2.0121,
      "step": 14557
    },
    {
      "epoch": 1.2525165619891594,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008533525663823583,
      "loss": 1.4118,
      "step": 14558
    },
    {
      "epoch": 1.2526025982964812,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008533334343234609,
      "loss": 0.6262,
      "step": 14559
    },
    {
      "epoch": 1.2526886346038029,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008533143012311319,
      "loss": 1.466,
      "step": 14560
    },
    {
      "epoch": 1.2527746709111245,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008532951671054274,
      "loss": 1.3347,
      "step": 14561
    },
    {
      "epoch": 1.2528607072184461,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008532760319464032,
      "loss": 1.3162,
      "step": 14562
    },
    {
      "epoch": 1.2529467435257677,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008532568957541153,
      "loss": 0.8204,
      "step": 14563
    },
    {
      "epoch": 1.2530327798330896,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008532377585286198,
      "loss": 1.0472,
      "step": 14564
    },
    {
      "epoch": 1.2531188161404112,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008532186202699725,
      "loss": 1.453,
      "step": 14565
    },
    {
      "epoch": 1.253204852447733,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008531994809782294,
      "loss": 0.8314,
      "step": 14566
    },
    {
      "epoch": 1.2532908887550547,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008531803406534465,
      "loss": 0.6252,
      "step": 14567
    },
    {
      "epoch": 1.2533769250623763,
      "grad_norm": 1.859375,
      "learning_rate": 0.0008531611992956799,
      "loss": 0.9207,
      "step": 14568
    },
    {
      "epoch": 1.253462961369698,
      "grad_norm": 1.125,
      "learning_rate": 0.0008531420569049853,
      "loss": 1.4358,
      "step": 14569
    },
    {
      "epoch": 1.2535489976770198,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000853122913481419,
      "loss": 1.8858,
      "step": 14570
    },
    {
      "epoch": 1.2536350339843414,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008531037690250368,
      "loss": 1.1183,
      "step": 14571
    },
    {
      "epoch": 1.253721070291663,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008530846235358948,
      "loss": 0.5012,
      "step": 14572
    },
    {
      "epoch": 1.2538071065989849,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008530654770140489,
      "loss": 1.3924,
      "step": 14573
    },
    {
      "epoch": 1.2538931429063065,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008530463294595552,
      "loss": 0.9263,
      "step": 14574
    },
    {
      "epoch": 1.2539791792136281,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008530271808724697,
      "loss": 1.5352,
      "step": 14575
    },
    {
      "epoch": 1.2540652155209497,
      "grad_norm": 1.625,
      "learning_rate": 0.0008530080312528481,
      "loss": 1.204,
      "step": 14576
    },
    {
      "epoch": 1.2541512518282716,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008529888806007469,
      "loss": 0.7476,
      "step": 14577
    },
    {
      "epoch": 1.2542372881355932,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008529697289162217,
      "loss": 1.8646,
      "step": 14578
    },
    {
      "epoch": 1.2543233244429148,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008529505761993288,
      "loss": 0.5064,
      "step": 14579
    },
    {
      "epoch": 1.2544093607502367,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008529314224501243,
      "loss": 1.4463,
      "step": 14580
    },
    {
      "epoch": 1.2544953970575583,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008529122676686638,
      "loss": 2.3695,
      "step": 14581
    },
    {
      "epoch": 1.25458143336488,
      "grad_norm": 5.34375,
      "learning_rate": 0.0008528931118550036,
      "loss": 1.5424,
      "step": 14582
    },
    {
      "epoch": 1.2546674696722016,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008528739550091996,
      "loss": 1.6261,
      "step": 14583
    },
    {
      "epoch": 1.2547535059795234,
      "grad_norm": 1.4375,
      "learning_rate": 0.000852854797131308,
      "loss": 1.7988,
      "step": 14584
    },
    {
      "epoch": 1.254839542286845,
      "grad_norm": 1.25,
      "learning_rate": 0.0008528356382213847,
      "loss": 1.3843,
      "step": 14585
    },
    {
      "epoch": 1.2549255785941669,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008528164782794858,
      "loss": 2.1871,
      "step": 14586
    },
    {
      "epoch": 1.2550116149014885,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008527973173056674,
      "loss": 0.7535,
      "step": 14587
    },
    {
      "epoch": 1.2550976512088101,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008527781552999853,
      "loss": 1.024,
      "step": 14588
    },
    {
      "epoch": 1.2551836875161317,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008527589922624959,
      "loss": 1.8057,
      "step": 14589
    },
    {
      "epoch": 1.2552697238234534,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000852739828193255,
      "loss": 1.0091,
      "step": 14590
    },
    {
      "epoch": 1.2553557601307752,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008527206630923186,
      "loss": 1.768,
      "step": 14591
    },
    {
      "epoch": 1.2554417964380968,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000852701496959743,
      "loss": 1.4983,
      "step": 14592
    },
    {
      "epoch": 1.2555278327454187,
      "grad_norm": 1.3125,
      "learning_rate": 0.000852682329795584,
      "loss": 0.9887,
      "step": 14593
    },
    {
      "epoch": 1.2556138690527403,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008526631615998979,
      "loss": 0.9245,
      "step": 14594
    },
    {
      "epoch": 1.255699905360062,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008526439923727404,
      "loss": 1.0832,
      "step": 14595
    },
    {
      "epoch": 1.2557859416673836,
      "grad_norm": 1.859375,
      "learning_rate": 0.000852624822114168,
      "loss": 1.065,
      "step": 14596
    },
    {
      "epoch": 1.2558719779747054,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008526056508242366,
      "loss": 1.7717,
      "step": 14597
    },
    {
      "epoch": 1.255958014282027,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008525864785030022,
      "loss": 1.066,
      "step": 14598
    },
    {
      "epoch": 1.2560440505893486,
      "grad_norm": 1.46875,
      "learning_rate": 0.000852567305150521,
      "loss": 1.1721,
      "step": 14599
    },
    {
      "epoch": 1.2561300868966705,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008525481307668489,
      "loss": 1.2187,
      "step": 14600
    },
    {
      "epoch": 1.2562161232039921,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008525289553520421,
      "loss": 0.8581,
      "step": 14601
    },
    {
      "epoch": 1.2563021595113137,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008525097789061567,
      "loss": 1.3912,
      "step": 14602
    },
    {
      "epoch": 1.2563881958186354,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008524906014292488,
      "loss": 1.0648,
      "step": 14603
    },
    {
      "epoch": 1.2564742321259572,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008524714229213744,
      "loss": 1.8204,
      "step": 14604
    },
    {
      "epoch": 1.2565602684332788,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008524522433825895,
      "loss": 1.9708,
      "step": 14605
    },
    {
      "epoch": 1.2566463047406005,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008524330628129505,
      "loss": 1.8185,
      "step": 14606
    },
    {
      "epoch": 1.2567323410479223,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008524138812125133,
      "loss": 0.8635,
      "step": 14607
    },
    {
      "epoch": 1.256818377355244,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000852394698581334,
      "loss": 0.979,
      "step": 14608
    },
    {
      "epoch": 1.2569044136625656,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008523755149194689,
      "loss": 1.4474,
      "step": 14609
    },
    {
      "epoch": 1.2569904499698872,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008523563302269737,
      "loss": 1.2249,
      "step": 14610
    },
    {
      "epoch": 1.257076486277209,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008523371445039049,
      "loss": 0.8119,
      "step": 14611
    },
    {
      "epoch": 1.2571625225845307,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008523179577503185,
      "loss": 1.3596,
      "step": 14612
    },
    {
      "epoch": 1.2572485588918525,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008522987699662706,
      "loss": 1.4206,
      "step": 14613
    },
    {
      "epoch": 1.2573345951991741,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008522795811518173,
      "loss": 1.6414,
      "step": 14614
    },
    {
      "epoch": 1.2574206315064957,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008522603913070147,
      "loss": 1.5222,
      "step": 14615
    },
    {
      "epoch": 1.2575066678138174,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008522412004319189,
      "loss": 1.2921,
      "step": 14616
    },
    {
      "epoch": 1.257592704121139,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008522220085265863,
      "loss": 1.5072,
      "step": 14617
    },
    {
      "epoch": 1.2576787404284608,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008522028155910726,
      "loss": 0.8398,
      "step": 14618
    },
    {
      "epoch": 1.2577647767357825,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008521836216254344,
      "loss": 1.6314,
      "step": 14619
    },
    {
      "epoch": 1.2578508130431043,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008521644266297275,
      "loss": 1.2205,
      "step": 14620
    },
    {
      "epoch": 1.257936849350426,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000852145230604008,
      "loss": 1.1365,
      "step": 14621
    },
    {
      "epoch": 1.2580228856577476,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008521260335483323,
      "loss": 1.4476,
      "step": 14622
    },
    {
      "epoch": 1.2581089219650692,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008521068354627565,
      "loss": 1.3305,
      "step": 14623
    },
    {
      "epoch": 1.258194958272391,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008520876363473367,
      "loss": 1.6504,
      "step": 14624
    },
    {
      "epoch": 1.2582809945797127,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008520684362021289,
      "loss": 1.2839,
      "step": 14625
    },
    {
      "epoch": 1.2583670308870343,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008520492350271896,
      "loss": 1.3465,
      "step": 14626
    },
    {
      "epoch": 1.2584530671943561,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008520300328225746,
      "loss": 1.9215,
      "step": 14627
    },
    {
      "epoch": 1.2585391035016777,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008520108295883402,
      "loss": 1.1174,
      "step": 14628
    },
    {
      "epoch": 1.2586251398089994,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008519916253245427,
      "loss": 2.2852,
      "step": 14629
    },
    {
      "epoch": 1.258711176116321,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008519724200312381,
      "loss": 1.4194,
      "step": 14630
    },
    {
      "epoch": 1.2587972124236428,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008519532137084826,
      "loss": 0.6109,
      "step": 14631
    },
    {
      "epoch": 1.2588832487309645,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008519340063563323,
      "loss": 1.5649,
      "step": 14632
    },
    {
      "epoch": 1.258969285038286,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008519147979748436,
      "loss": 0.8109,
      "step": 14633
    },
    {
      "epoch": 1.259055321345608,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008518955885640727,
      "loss": 2.0522,
      "step": 14634
    },
    {
      "epoch": 1.2591413576529296,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008518763781240755,
      "loss": 1.253,
      "step": 14635
    },
    {
      "epoch": 1.2592273939602512,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008518571666549082,
      "loss": 2.0258,
      "step": 14636
    },
    {
      "epoch": 1.2593134302675728,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008518379541566272,
      "loss": 1.6448,
      "step": 14637
    },
    {
      "epoch": 1.2593994665748947,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008518187406292888,
      "loss": 1.9169,
      "step": 14638
    },
    {
      "epoch": 1.2594855028822163,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008517995260729489,
      "loss": 1.2682,
      "step": 14639
    },
    {
      "epoch": 1.259571539189538,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008517803104876638,
      "loss": 0.8646,
      "step": 14640
    },
    {
      "epoch": 1.2596575754968597,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008517610938734896,
      "loss": 0.9635,
      "step": 14641
    },
    {
      "epoch": 1.2597436118041814,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008517418762304827,
      "loss": 1.146,
      "step": 14642
    },
    {
      "epoch": 1.259829648111503,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008517226575586993,
      "loss": 0.8627,
      "step": 14643
    },
    {
      "epoch": 1.2599156844188246,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008517034378581953,
      "loss": 0.9694,
      "step": 14644
    },
    {
      "epoch": 1.2600017207261465,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008516842171290276,
      "loss": 1.9186,
      "step": 14645
    },
    {
      "epoch": 1.260087757033468,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008516649953712515,
      "loss": 1.6437,
      "step": 14646
    },
    {
      "epoch": 1.26017379334079,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008516457725849239,
      "loss": 0.5647,
      "step": 14647
    },
    {
      "epoch": 1.2602598296481116,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008516265487701007,
      "loss": 1.1722,
      "step": 14648
    },
    {
      "epoch": 1.2603458659554332,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008516073239268383,
      "loss": 1.3005,
      "step": 14649
    },
    {
      "epoch": 1.2604319022627548,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008515880980551929,
      "loss": 1.0096,
      "step": 14650
    },
    {
      "epoch": 1.2605179385700764,
      "grad_norm": 1.375,
      "learning_rate": 0.0008515688711552205,
      "loss": 1.4317,
      "step": 14651
    },
    {
      "epoch": 1.2606039748773983,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008515496432269778,
      "loss": 1.1164,
      "step": 14652
    },
    {
      "epoch": 1.26069001118472,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008515304142705206,
      "loss": 1.5705,
      "step": 14653
    },
    {
      "epoch": 1.2607760474920418,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008515111842859053,
      "loss": 0.7527,
      "step": 14654
    },
    {
      "epoch": 1.2608620837993634,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008514919532731883,
      "loss": 0.9891,
      "step": 14655
    },
    {
      "epoch": 1.260948120106685,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008514727212324256,
      "loss": 1.6623,
      "step": 14656
    },
    {
      "epoch": 1.2610341564140066,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008514534881636735,
      "loss": 1.1288,
      "step": 14657
    },
    {
      "epoch": 1.2611201927213285,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008514342540669883,
      "loss": 2.0308,
      "step": 14658
    },
    {
      "epoch": 1.26120622902865,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008514150189424263,
      "loss": 1.0144,
      "step": 14659
    },
    {
      "epoch": 1.2612922653359717,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008513957827900438,
      "loss": 1.6819,
      "step": 14660
    },
    {
      "epoch": 1.2613783016432936,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008513765456098969,
      "loss": 1.7777,
      "step": 14661
    },
    {
      "epoch": 1.2614643379506152,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0008513573074020419,
      "loss": 1.1513,
      "step": 14662
    },
    {
      "epoch": 1.2615503742579368,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008513380681665352,
      "loss": 1.6295,
      "step": 14663
    },
    {
      "epoch": 1.2616364105652584,
      "grad_norm": 1.328125,
      "learning_rate": 0.000851318827903433,
      "loss": 1.2014,
      "step": 14664
    },
    {
      "epoch": 1.2617224468725803,
      "grad_norm": 2.046875,
      "learning_rate": 0.0008512995866127915,
      "loss": 2.1227,
      "step": 14665
    },
    {
      "epoch": 1.261808483179902,
      "grad_norm": 1.5,
      "learning_rate": 0.0008512803442946669,
      "loss": 1.4541,
      "step": 14666
    },
    {
      "epoch": 1.2618945194872235,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0008512611009491158,
      "loss": 1.7389,
      "step": 14667
    },
    {
      "epoch": 1.2619805557945454,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008512418565761942,
      "loss": 1.3539,
      "step": 14668
    },
    {
      "epoch": 1.262066592101867,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008512226111759585,
      "loss": 1.1236,
      "step": 14669
    },
    {
      "epoch": 1.2621526284091886,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008512033647484651,
      "loss": 1.3508,
      "step": 14670
    },
    {
      "epoch": 1.2622386647165103,
      "grad_norm": 1.390625,
      "learning_rate": 0.00085118411729377,
      "loss": 1.4286,
      "step": 14671
    },
    {
      "epoch": 1.262324701023832,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008511648688119297,
      "loss": 1.812,
      "step": 14672
    },
    {
      "epoch": 1.2624107373311537,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008511456193030005,
      "loss": 0.7198,
      "step": 14673
    },
    {
      "epoch": 1.2624967736384756,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008511263687670385,
      "loss": 0.9944,
      "step": 14674
    },
    {
      "epoch": 1.2625828099457972,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008511071172041004,
      "loss": 1.3513,
      "step": 14675
    },
    {
      "epoch": 1.2626688462531188,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008510878646142421,
      "loss": 1.1546,
      "step": 14676
    },
    {
      "epoch": 1.2627548825604404,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008510686109975201,
      "loss": 1.6599,
      "step": 14677
    },
    {
      "epoch": 1.262840918867762,
      "grad_norm": 110.0,
      "learning_rate": 0.0008510493563539907,
      "loss": 1.6543,
      "step": 14678
    },
    {
      "epoch": 1.262926955175084,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008510301006837103,
      "loss": 0.9969,
      "step": 14679
    },
    {
      "epoch": 1.2630129914824055,
      "grad_norm": 1.203125,
      "learning_rate": 0.000851010843986735,
      "loss": 1.2363,
      "step": 14680
    },
    {
      "epoch": 1.2630990277897274,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008509915862631214,
      "loss": 2.0763,
      "step": 14681
    },
    {
      "epoch": 1.263185064097049,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008509723275129255,
      "loss": 0.7838,
      "step": 14682
    },
    {
      "epoch": 1.2632711004043706,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008509530677362039,
      "loss": 1.2929,
      "step": 14683
    },
    {
      "epoch": 1.2633571367116923,
      "grad_norm": 1.25,
      "learning_rate": 0.0008509338069330129,
      "loss": 1.3579,
      "step": 14684
    },
    {
      "epoch": 1.263443173019014,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008509145451034086,
      "loss": 1.422,
      "step": 14685
    },
    {
      "epoch": 1.2635292093263357,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008508952822474476,
      "loss": 1.3212,
      "step": 14686
    },
    {
      "epoch": 1.2636152456336573,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008508760183651862,
      "loss": 1.7855,
      "step": 14687
    },
    {
      "epoch": 1.2637012819409792,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008508567534566806,
      "loss": 1.2515,
      "step": 14688
    },
    {
      "epoch": 1.2637873182483008,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008508374875219872,
      "loss": 1.4831,
      "step": 14689
    },
    {
      "epoch": 1.2638733545556224,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008508182205611626,
      "loss": 1.2431,
      "step": 14690
    },
    {
      "epoch": 1.263959390862944,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008507989525742627,
      "loss": 1.0709,
      "step": 14691
    },
    {
      "epoch": 1.264045427170266,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008507796835613442,
      "loss": 1.6748,
      "step": 14692
    },
    {
      "epoch": 1.2641314634775875,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008507604135224632,
      "loss": 0.7054,
      "step": 14693
    },
    {
      "epoch": 1.2642174997849092,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008507411424576764,
      "loss": 1.5708,
      "step": 14694
    },
    {
      "epoch": 1.264303536092231,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008507218703670399,
      "loss": 2.0599,
      "step": 14695
    },
    {
      "epoch": 1.2643895723995526,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008507025972506101,
      "loss": 1.7588,
      "step": 14696
    },
    {
      "epoch": 1.2644756087068743,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008506833231084434,
      "loss": 1.8393,
      "step": 14697
    },
    {
      "epoch": 1.2645616450141959,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008506640479405963,
      "loss": 1.3248,
      "step": 14698
    },
    {
      "epoch": 1.2646476813215177,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008506447717471251,
      "loss": 1.5447,
      "step": 14699
    },
    {
      "epoch": 1.2647337176288393,
      "grad_norm": 1.21875,
      "learning_rate": 0.000850625494528086,
      "loss": 1.0409,
      "step": 14700
    },
    {
      "epoch": 1.2648197539361612,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008506062162835355,
      "loss": 1.0621,
      "step": 14701
    },
    {
      "epoch": 1.2649057902434828,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008505869370135301,
      "loss": 0.8156,
      "step": 14702
    },
    {
      "epoch": 1.2649918265508044,
      "grad_norm": 1.0,
      "learning_rate": 0.0008505676567181259,
      "loss": 0.8265,
      "step": 14703
    },
    {
      "epoch": 1.265077862858126,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008505483753973798,
      "loss": 0.9697,
      "step": 14704
    },
    {
      "epoch": 1.2651638991654477,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008505290930513476,
      "loss": 1.3322,
      "step": 14705
    },
    {
      "epoch": 1.2652499354727695,
      "grad_norm": 1.640625,
      "learning_rate": 0.000850509809680086,
      "loss": 1.5415,
      "step": 14706
    },
    {
      "epoch": 1.2653359717800912,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008504905252836517,
      "loss": 0.7964,
      "step": 14707
    },
    {
      "epoch": 1.265422008087413,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008504712398621004,
      "loss": 0.5784,
      "step": 14708
    },
    {
      "epoch": 1.2655080443947346,
      "grad_norm": 1.09375,
      "learning_rate": 0.000850451953415489,
      "loss": 0.9127,
      "step": 14709
    },
    {
      "epoch": 1.2655940807020563,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000850432665943874,
      "loss": 1.4876,
      "step": 14710
    },
    {
      "epoch": 1.2656801170093779,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008504133774473115,
      "loss": 2.0199,
      "step": 14711
    },
    {
      "epoch": 1.2657661533166997,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008503940879258579,
      "loss": 0.9707,
      "step": 14712
    },
    {
      "epoch": 1.2658521896240214,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008503747973795698,
      "loss": 1.4679,
      "step": 14713
    },
    {
      "epoch": 1.265938225931343,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008503555058085035,
      "loss": 1.2111,
      "step": 14714
    },
    {
      "epoch": 1.2660242622386648,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008503362132127156,
      "loss": 1.5267,
      "step": 14715
    },
    {
      "epoch": 1.2661102985459864,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008503169195922624,
      "loss": 1.3414,
      "step": 14716
    },
    {
      "epoch": 1.266196334853308,
      "grad_norm": 2.328125,
      "learning_rate": 0.0008502976249472003,
      "loss": 1.234,
      "step": 14717
    },
    {
      "epoch": 1.2662823711606297,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008502783292775858,
      "loss": 1.0072,
      "step": 14718
    },
    {
      "epoch": 1.2663684074679515,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008502590325834753,
      "loss": 1.2502,
      "step": 14719
    },
    {
      "epoch": 1.2664544437752732,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008502397348649253,
      "loss": 1.7462,
      "step": 14720
    },
    {
      "epoch": 1.2665404800825948,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008502204361219921,
      "loss": 1.0311,
      "step": 14721
    },
    {
      "epoch": 1.2666265163899166,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008502011363547323,
      "loss": 1.1377,
      "step": 14722
    },
    {
      "epoch": 1.2667125526972383,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008501818355632023,
      "loss": 1.5148,
      "step": 14723
    },
    {
      "epoch": 1.2667985890045599,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008501625337474585,
      "loss": 1.5209,
      "step": 14724
    },
    {
      "epoch": 1.2668846253118815,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008501432309075573,
      "loss": 1.121,
      "step": 14725
    },
    {
      "epoch": 1.2669706616192034,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008501239270435555,
      "loss": 1.4341,
      "step": 14726
    },
    {
      "epoch": 1.267056697926525,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008501046221555091,
      "loss": 1.2107,
      "step": 14727
    },
    {
      "epoch": 1.2671427342338466,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008500853162434748,
      "loss": 1.1831,
      "step": 14728
    },
    {
      "epoch": 1.2672287705411684,
      "grad_norm": 1.078125,
      "learning_rate": 0.000850066009307509,
      "loss": 0.7794,
      "step": 14729
    },
    {
      "epoch": 1.26731480684849,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008500467013476682,
      "loss": 0.9271,
      "step": 14730
    },
    {
      "epoch": 1.2674008431558117,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008500273923640091,
      "loss": 1.4502,
      "step": 14731
    },
    {
      "epoch": 1.2674868794631333,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008500080823565878,
      "loss": 1.2671,
      "step": 14732
    },
    {
      "epoch": 1.2675729157704552,
      "grad_norm": 1.3125,
      "learning_rate": 0.000849988771325461,
      "loss": 1.1352,
      "step": 14733
    },
    {
      "epoch": 1.2676589520777768,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008499694592706851,
      "loss": 1.392,
      "step": 14734
    },
    {
      "epoch": 1.2677449883850986,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008499501461923164,
      "loss": 0.7771,
      "step": 14735
    },
    {
      "epoch": 1.2678310246924203,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008499308320904118,
      "loss": 1.131,
      "step": 14736
    },
    {
      "epoch": 1.2679170609997419,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008499115169650275,
      "loss": 1.5859,
      "step": 14737
    },
    {
      "epoch": 1.2680030973070635,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00084989220081622,
      "loss": 1.6916,
      "step": 14738
    },
    {
      "epoch": 1.2680891336143851,
      "grad_norm": 1.28125,
      "learning_rate": 0.000849872883644046,
      "loss": 1.0522,
      "step": 14739
    },
    {
      "epoch": 1.268175169921707,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008498535654485617,
      "loss": 1.0951,
      "step": 14740
    },
    {
      "epoch": 1.2682612062290286,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008498342462298239,
      "loss": 0.6596,
      "step": 14741
    },
    {
      "epoch": 1.2683472425363504,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008498149259878889,
      "loss": 1.2281,
      "step": 14742
    },
    {
      "epoch": 1.268433278843672,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008497956047228133,
      "loss": 1.4425,
      "step": 14743
    },
    {
      "epoch": 1.2685193151509937,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008497762824346536,
      "loss": 1.6202,
      "step": 14744
    },
    {
      "epoch": 1.2686053514583153,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008497569591234663,
      "loss": 1.617,
      "step": 14745
    },
    {
      "epoch": 1.2686913877656372,
      "grad_norm": 14.6875,
      "learning_rate": 0.000849737634789308,
      "loss": 1.1988,
      "step": 14746
    },
    {
      "epoch": 1.2687774240729588,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008497183094322349,
      "loss": 1.504,
      "step": 14747
    },
    {
      "epoch": 1.2688634603802804,
      "grad_norm": 1.84375,
      "learning_rate": 0.0008496989830523041,
      "loss": 0.8688,
      "step": 14748
    },
    {
      "epoch": 1.2689494966876023,
      "grad_norm": 3.15625,
      "learning_rate": 0.0008496796556495715,
      "loss": 1.2031,
      "step": 14749
    },
    {
      "epoch": 1.2690355329949239,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000849660327224094,
      "loss": 1.0951,
      "step": 14750
    },
    {
      "epoch": 1.2691215693022455,
      "grad_norm": 1.625,
      "learning_rate": 0.0008496409977759281,
      "loss": 1.2701,
      "step": 14751
    },
    {
      "epoch": 1.2692076056095671,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008496216673051303,
      "loss": 2.275,
      "step": 14752
    },
    {
      "epoch": 1.269293641916889,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008496023358117571,
      "loss": 1.4709,
      "step": 14753
    },
    {
      "epoch": 1.2693796782242106,
      "grad_norm": 1.8125,
      "learning_rate": 0.0008495830032958649,
      "loss": 1.5673,
      "step": 14754
    },
    {
      "epoch": 1.2694657145315322,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008495636697575106,
      "loss": 0.9027,
      "step": 14755
    },
    {
      "epoch": 1.269551750838854,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008495443351967505,
      "loss": 0.9957,
      "step": 14756
    },
    {
      "epoch": 1.2696377871461757,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008495249996136411,
      "loss": 0.8744,
      "step": 14757
    },
    {
      "epoch": 1.2697238234534973,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008495056630082392,
      "loss": 0.9908,
      "step": 14758
    },
    {
      "epoch": 1.269809859760819,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008494863253806013,
      "loss": 0.7859,
      "step": 14759
    },
    {
      "epoch": 1.2698958960681408,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0008494669867307836,
      "loss": 1.0152,
      "step": 14760
    },
    {
      "epoch": 1.2699819323754624,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008494476470588431,
      "loss": 0.9284,
      "step": 14761
    },
    {
      "epoch": 1.2700679686827843,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000849428306364836,
      "loss": 1.3633,
      "step": 14762
    },
    {
      "epoch": 1.270154004990106,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008494089646488192,
      "loss": 1.6173,
      "step": 14763
    },
    {
      "epoch": 1.2702400412974275,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008493896219108492,
      "loss": 1.7135,
      "step": 14764
    },
    {
      "epoch": 1.2703260776047491,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008493702781509824,
      "loss": 1.1296,
      "step": 14765
    },
    {
      "epoch": 1.2704121139120708,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008493509333692758,
      "loss": 1.0777,
      "step": 14766
    },
    {
      "epoch": 1.2704981502193926,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008493315875657852,
      "loss": 1.1283,
      "step": 14767
    },
    {
      "epoch": 1.2705841865267142,
      "grad_norm": 1.171875,
      "learning_rate": 0.000849312240740568,
      "loss": 1.7617,
      "step": 14768
    },
    {
      "epoch": 1.270670222834036,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008492928928936803,
      "loss": 1.7983,
      "step": 14769
    },
    {
      "epoch": 1.2707562591413577,
      "grad_norm": 1.375,
      "learning_rate": 0.0008492735440251787,
      "loss": 1.055,
      "step": 14770
    },
    {
      "epoch": 1.2708422954486793,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008492541941351201,
      "loss": 1.2525,
      "step": 14771
    },
    {
      "epoch": 1.270928331756001,
      "grad_norm": 2.328125,
      "learning_rate": 0.0008492348432235608,
      "loss": 1.6468,
      "step": 14772
    },
    {
      "epoch": 1.2710143680633228,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008492154912905576,
      "loss": 1.8769,
      "step": 14773
    },
    {
      "epoch": 1.2711004043706444,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008491961383361669,
      "loss": 1.2741,
      "step": 14774
    },
    {
      "epoch": 1.271186440677966,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008491767843604454,
      "loss": 1.565,
      "step": 14775
    },
    {
      "epoch": 1.271272476985288,
      "grad_norm": 1.375,
      "learning_rate": 0.0008491574293634499,
      "loss": 0.8176,
      "step": 14776
    },
    {
      "epoch": 1.2713585132926095,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008491380733452367,
      "loss": 1.5237,
      "step": 14777
    },
    {
      "epoch": 1.2714445495999311,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008491187163058624,
      "loss": 1.1746,
      "step": 14778
    },
    {
      "epoch": 1.2715305859072528,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008490993582453838,
      "loss": 1.7275,
      "step": 14779
    },
    {
      "epoch": 1.2716166222145746,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008490799991638578,
      "loss": 2.0328,
      "step": 14780
    },
    {
      "epoch": 1.2717026585218962,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008490606390613403,
      "loss": 1.369,
      "step": 14781
    },
    {
      "epoch": 1.2717886948292179,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008490412779378887,
      "loss": 1.1608,
      "step": 14782
    },
    {
      "epoch": 1.2718747311365397,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008490219157935589,
      "loss": 1.0209,
      "step": 14783
    },
    {
      "epoch": 1.2719607674438613,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000849002552628408,
      "loss": 1.1509,
      "step": 14784
    },
    {
      "epoch": 1.272046803751183,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008489831884424926,
      "loss": 2.1674,
      "step": 14785
    },
    {
      "epoch": 1.2721328400585046,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008489638232358691,
      "loss": 0.9729,
      "step": 14786
    },
    {
      "epoch": 1.2722188763658264,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008489444570085944,
      "loss": 1.7943,
      "step": 14787
    },
    {
      "epoch": 1.272304912673148,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008489250897607249,
      "loss": 0.8338,
      "step": 14788
    },
    {
      "epoch": 1.27239094898047,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008489057214923174,
      "loss": 1.4557,
      "step": 14789
    },
    {
      "epoch": 1.2724769852877915,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008488863522034287,
      "loss": 1.47,
      "step": 14790
    },
    {
      "epoch": 1.2725630215951131,
      "grad_norm": 1.625,
      "learning_rate": 0.0008488669818941152,
      "loss": 1.5467,
      "step": 14791
    },
    {
      "epoch": 1.2726490579024348,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008488476105644334,
      "loss": 1.1023,
      "step": 14792
    },
    {
      "epoch": 1.2727350942097564,
      "grad_norm": 1.0,
      "learning_rate": 0.0008488282382144404,
      "loss": 1.3273,
      "step": 14793
    },
    {
      "epoch": 1.2728211305170782,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008488088648441927,
      "loss": 1.8588,
      "step": 14794
    },
    {
      "epoch": 1.2729071668243999,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008487894904537468,
      "loss": 0.9806,
      "step": 14795
    },
    {
      "epoch": 1.2729932031317217,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008487701150431595,
      "loss": 1.8249,
      "step": 14796
    },
    {
      "epoch": 1.2730792394390433,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008487507386124874,
      "loss": 1.444,
      "step": 14797
    },
    {
      "epoch": 1.273165275746365,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008487313611617872,
      "loss": 0.8912,
      "step": 14798
    },
    {
      "epoch": 1.2732513120536866,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008487119826911158,
      "loss": 0.8224,
      "step": 14799
    },
    {
      "epoch": 1.2733373483610084,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0008486926032005295,
      "loss": 1.3772,
      "step": 14800
    },
    {
      "epoch": 1.27342338466833,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008486732226900851,
      "loss": 1.2019,
      "step": 14801
    },
    {
      "epoch": 1.2735094209756517,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008486538411598395,
      "loss": 1.5945,
      "step": 14802
    },
    {
      "epoch": 1.2735954572829735,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008486344586098491,
      "loss": 1.9728,
      "step": 14803
    },
    {
      "epoch": 1.2736814935902951,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008486150750401708,
      "loss": 0.855,
      "step": 14804
    },
    {
      "epoch": 1.2737675298976168,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008485956904508611,
      "loss": 1.5606,
      "step": 14805
    },
    {
      "epoch": 1.2738535662049384,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000848576304841977,
      "loss": 1.7867,
      "step": 14806
    },
    {
      "epoch": 1.2739396025122602,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008485569182135748,
      "loss": 1.1945,
      "step": 14807
    },
    {
      "epoch": 1.2740256388195819,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008485375305657115,
      "loss": 1.5443,
      "step": 14808
    },
    {
      "epoch": 1.2741116751269035,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008485181418984437,
      "loss": 0.7844,
      "step": 14809
    },
    {
      "epoch": 1.2741977114342253,
      "grad_norm": 1.1875,
      "learning_rate": 0.000848498752211828,
      "loss": 1.8546,
      "step": 14810
    },
    {
      "epoch": 1.274283747741547,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008484793615059214,
      "loss": 1.9053,
      "step": 14811
    },
    {
      "epoch": 1.2743697840488686,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008484599697807804,
      "loss": 1.2907,
      "step": 14812
    },
    {
      "epoch": 1.2744558203561902,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008484405770364617,
      "loss": 1.9964,
      "step": 14813
    },
    {
      "epoch": 1.274541856663512,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008484211832730222,
      "loss": 0.8209,
      "step": 14814
    },
    {
      "epoch": 1.2746278929708337,
      "grad_norm": 1.25,
      "learning_rate": 0.0008484017884905184,
      "loss": 1.3539,
      "step": 14815
    },
    {
      "epoch": 1.2747139292781553,
      "grad_norm": 1.0625,
      "learning_rate": 0.000848382392689007,
      "loss": 0.9638,
      "step": 14816
    },
    {
      "epoch": 1.2747999655854771,
      "grad_norm": 1.140625,
      "learning_rate": 0.000848362995868545,
      "loss": 1.5746,
      "step": 14817
    },
    {
      "epoch": 1.2748860018927988,
      "grad_norm": 1.203125,
      "learning_rate": 0.000848343598029189,
      "loss": 1.0464,
      "step": 14818
    },
    {
      "epoch": 1.2749720382001204,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008483241991709958,
      "loss": 0.6877,
      "step": 14819
    },
    {
      "epoch": 1.275058074507442,
      "grad_norm": 1.25,
      "learning_rate": 0.000848304799294022,
      "loss": 0.9283,
      "step": 14820
    },
    {
      "epoch": 1.2751441108147639,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008482853983983243,
      "loss": 1.0699,
      "step": 14821
    },
    {
      "epoch": 1.2752301471220855,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008482659964839596,
      "loss": 0.9832,
      "step": 14822
    },
    {
      "epoch": 1.2753161834294073,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008482465935509846,
      "loss": 1.7105,
      "step": 14823
    },
    {
      "epoch": 1.275402219736729,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008482271895994563,
      "loss": 2.0032,
      "step": 14824
    },
    {
      "epoch": 1.2754882560440506,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008482077846294308,
      "loss": 0.7548,
      "step": 14825
    },
    {
      "epoch": 1.2755742923513722,
      "grad_norm": 1.375,
      "learning_rate": 0.0008481883786409654,
      "loss": 1.2753,
      "step": 14826
    },
    {
      "epoch": 1.2756603286586938,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008481689716341168,
      "loss": 0.736,
      "step": 14827
    },
    {
      "epoch": 1.2757463649660157,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008481495636089416,
      "loss": 0.6954,
      "step": 14828
    },
    {
      "epoch": 1.2758324012733373,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008481301545654967,
      "loss": 1.6314,
      "step": 14829
    },
    {
      "epoch": 1.2759184375806591,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008481107445038387,
      "loss": 1.5166,
      "step": 14830
    },
    {
      "epoch": 1.2760044738879808,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008480913334240245,
      "loss": 0.9357,
      "step": 14831
    },
    {
      "epoch": 1.2760905101953024,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000848071921326111,
      "loss": 1.2376,
      "step": 14832
    },
    {
      "epoch": 1.276176546502624,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008480525082101547,
      "loss": 1.1273,
      "step": 14833
    },
    {
      "epoch": 1.2762625828099459,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008480330940762124,
      "loss": 1.2425,
      "step": 14834
    },
    {
      "epoch": 1.2763486191172675,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008480136789243411,
      "loss": 1.2693,
      "step": 14835
    },
    {
      "epoch": 1.2764346554245891,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008479942627545974,
      "loss": 1.4143,
      "step": 14836
    },
    {
      "epoch": 1.276520691731911,
      "grad_norm": 1.375,
      "learning_rate": 0.0008479748455670383,
      "loss": 2.3351,
      "step": 14837
    },
    {
      "epoch": 1.2766067280392326,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008479554273617204,
      "loss": 1.5492,
      "step": 14838
    },
    {
      "epoch": 1.2766927643465542,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008479360081387005,
      "loss": 1.0576,
      "step": 14839
    },
    {
      "epoch": 1.2767788006538758,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008479165878980355,
      "loss": 1.5476,
      "step": 14840
    },
    {
      "epoch": 1.2768648369611977,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008478971666397821,
      "loss": 1.249,
      "step": 14841
    },
    {
      "epoch": 1.2769508732685193,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008478777443639972,
      "loss": 0.6545,
      "step": 14842
    },
    {
      "epoch": 1.277036909575841,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008478583210707375,
      "loss": 1.238,
      "step": 14843
    },
    {
      "epoch": 1.2771229458831628,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008478388967600598,
      "loss": 1.1334,
      "step": 14844
    },
    {
      "epoch": 1.2772089821904844,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008478194714320213,
      "loss": 0.6903,
      "step": 14845
    },
    {
      "epoch": 1.277295018497806,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008478000450866782,
      "loss": 1.8517,
      "step": 14846
    },
    {
      "epoch": 1.2773810548051276,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008477806177240876,
      "loss": 1.3005,
      "step": 14847
    },
    {
      "epoch": 1.2774670911124495,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008477611893443064,
      "loss": 0.851,
      "step": 14848
    },
    {
      "epoch": 1.2775531274197711,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008477417599473914,
      "loss": 1.3009,
      "step": 14849
    },
    {
      "epoch": 1.277639163727093,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008477223295333995,
      "loss": 1.4826,
      "step": 14850
    },
    {
      "epoch": 1.2777252000344146,
      "grad_norm": 1.25,
      "learning_rate": 0.0008477028981023873,
      "loss": 1.7982,
      "step": 14851
    },
    {
      "epoch": 1.2778112363417362,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008476834656544116,
      "loss": 0.6504,
      "step": 14852
    },
    {
      "epoch": 1.2778972726490578,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008476640321895295,
      "loss": 1.4311,
      "step": 14853
    },
    {
      "epoch": 1.2779833089563795,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008476445977077978,
      "loss": 1.1186,
      "step": 14854
    },
    {
      "epoch": 1.2780693452637013,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008476251622092731,
      "loss": 1.0175,
      "step": 14855
    },
    {
      "epoch": 1.278155381571023,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008476057256940126,
      "loss": 1.3455,
      "step": 14856
    },
    {
      "epoch": 1.2782414178783448,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008475862881620728,
      "loss": 0.8896,
      "step": 14857
    },
    {
      "epoch": 1.2783274541856664,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008475668496135107,
      "loss": 0.8602,
      "step": 14858
    },
    {
      "epoch": 1.278413490492988,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008475474100483833,
      "loss": 1.1646,
      "step": 14859
    },
    {
      "epoch": 1.2784995268003096,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008475279694667473,
      "loss": 1.379,
      "step": 14860
    },
    {
      "epoch": 1.2785855631076315,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008475085278686594,
      "loss": 1.0118,
      "step": 14861
    },
    {
      "epoch": 1.2786715994149531,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008474890852541768,
      "loss": 0.7947,
      "step": 14862
    },
    {
      "epoch": 1.2787576357222747,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008474696416233562,
      "loss": 1.0467,
      "step": 14863
    },
    {
      "epoch": 1.2788436720295966,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008474501969762545,
      "loss": 1.7889,
      "step": 14864
    },
    {
      "epoch": 1.2789297083369182,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008474307513129285,
      "loss": 1.0339,
      "step": 14865
    },
    {
      "epoch": 1.2790157446442398,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008474113046334351,
      "loss": 1.1307,
      "step": 14866
    },
    {
      "epoch": 1.2791017809515615,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008473918569378312,
      "loss": 0.9769,
      "step": 14867
    },
    {
      "epoch": 1.2791878172588833,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008473724082261738,
      "loss": 0.9682,
      "step": 14868
    },
    {
      "epoch": 1.279273853566205,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008473529584985196,
      "loss": 1.7504,
      "step": 14869
    },
    {
      "epoch": 1.2793598898735266,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008473335077549255,
      "loss": 1.0669,
      "step": 14870
    },
    {
      "epoch": 1.2794459261808484,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008473140559954485,
      "loss": 1.0093,
      "step": 14871
    },
    {
      "epoch": 1.27953196248817,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008472946032201453,
      "loss": 1.5024,
      "step": 14872
    },
    {
      "epoch": 1.2796179987954917,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000847275149429073,
      "loss": 1.6207,
      "step": 14873
    },
    {
      "epoch": 1.2797040351028133,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008472556946222885,
      "loss": 1.2461,
      "step": 14874
    },
    {
      "epoch": 1.2797900714101351,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008472362387998487,
      "loss": 0.9221,
      "step": 14875
    },
    {
      "epoch": 1.2798761077174567,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008472167819618102,
      "loss": 0.7298,
      "step": 14876
    },
    {
      "epoch": 1.2799621440247786,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008471973241082303,
      "loss": 1.4753,
      "step": 14877
    },
    {
      "epoch": 1.2800481803321002,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008471778652391658,
      "loss": 1.7731,
      "step": 14878
    },
    {
      "epoch": 1.2801342166394218,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008471584053546735,
      "loss": 1.7506,
      "step": 14879
    },
    {
      "epoch": 1.2802202529467435,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008471389444548104,
      "loss": 1.5164,
      "step": 14880
    },
    {
      "epoch": 1.280306289254065,
      "grad_norm": 1.5,
      "learning_rate": 0.0008471194825396334,
      "loss": 2.0151,
      "step": 14881
    },
    {
      "epoch": 1.280392325561387,
      "grad_norm": 1.375,
      "learning_rate": 0.0008471000196091994,
      "loss": 1.6737,
      "step": 14882
    },
    {
      "epoch": 1.2804783618687086,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008470805556635652,
      "loss": 1.2465,
      "step": 14883
    },
    {
      "epoch": 1.2805643981760304,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008470610907027881,
      "loss": 1.4839,
      "step": 14884
    },
    {
      "epoch": 1.280650434483352,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008470416247269248,
      "loss": 2.0801,
      "step": 14885
    },
    {
      "epoch": 1.2807364707906737,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008470221577360321,
      "loss": 1.4925,
      "step": 14886
    },
    {
      "epoch": 1.2808225070979953,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008470026897301672,
      "loss": 1.2973,
      "step": 14887
    },
    {
      "epoch": 1.2809085434053171,
      "grad_norm": 1.8828125,
      "learning_rate": 0.000846983220709387,
      "loss": 1.0584,
      "step": 14888
    },
    {
      "epoch": 1.2809945797126387,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008469637506737482,
      "loss": 0.7982,
      "step": 14889
    },
    {
      "epoch": 1.2810806160199604,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000846944279623308,
      "loss": 0.8067,
      "step": 14890
    },
    {
      "epoch": 1.2811666523272822,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008469248075581231,
      "loss": 1.098,
      "step": 14891
    },
    {
      "epoch": 1.2812526886346038,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008469053344782508,
      "loss": 1.5043,
      "step": 14892
    },
    {
      "epoch": 1.2813387249419255,
      "grad_norm": 1.625,
      "learning_rate": 0.0008468858603837477,
      "loss": 1.5131,
      "step": 14893
    },
    {
      "epoch": 1.281424761249247,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008468663852746712,
      "loss": 1.9973,
      "step": 14894
    },
    {
      "epoch": 1.281510797556569,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008468469091510777,
      "loss": 1.8194,
      "step": 14895
    },
    {
      "epoch": 1.2815968338638906,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008468274320130246,
      "loss": 1.0051,
      "step": 14896
    },
    {
      "epoch": 1.2816828701712122,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008468079538605686,
      "loss": 1.1082,
      "step": 14897
    },
    {
      "epoch": 1.281768906478534,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008467884746937668,
      "loss": 1.6583,
      "step": 14898
    },
    {
      "epoch": 1.2818549427858557,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008467689945126764,
      "loss": 1.1482,
      "step": 14899
    },
    {
      "epoch": 1.2819409790931773,
      "grad_norm": 1.03125,
      "learning_rate": 0.000846749513317354,
      "loss": 0.8797,
      "step": 14900
    },
    {
      "epoch": 1.282027015400499,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008467300311078565,
      "loss": 1.5162,
      "step": 14901
    },
    {
      "epoch": 1.2821130517078208,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008467105478842414,
      "loss": 1.0405,
      "step": 14902
    },
    {
      "epoch": 1.2821990880151424,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008466910636465652,
      "loss": 1.0656,
      "step": 14903
    },
    {
      "epoch": 1.282285124322464,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008466715783948851,
      "loss": 1.7826,
      "step": 14904
    },
    {
      "epoch": 1.2823711606297858,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008466520921292579,
      "loss": 1.3614,
      "step": 14905
    },
    {
      "epoch": 1.2824571969371075,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000846632604849741,
      "loss": 1.6159,
      "step": 14906
    },
    {
      "epoch": 1.282543233244429,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000846613116556391,
      "loss": 0.8698,
      "step": 14907
    },
    {
      "epoch": 1.2826292695517507,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008465936272492651,
      "loss": 1.4271,
      "step": 14908
    },
    {
      "epoch": 1.2827153058590726,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008465741369284203,
      "loss": 1.093,
      "step": 14909
    },
    {
      "epoch": 1.2828013421663942,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008465546455939135,
      "loss": 1.1281,
      "step": 14910
    },
    {
      "epoch": 1.282887378473716,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008465351532458018,
      "loss": 1.3905,
      "step": 14911
    },
    {
      "epoch": 1.2829734147810377,
      "grad_norm": 1.75,
      "learning_rate": 0.0008465156598841422,
      "loss": 1.0403,
      "step": 14912
    },
    {
      "epoch": 1.2830594510883593,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008464961655089916,
      "loss": 0.9724,
      "step": 14913
    },
    {
      "epoch": 1.283145487395681,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008464766701204072,
      "loss": 1.2539,
      "step": 14914
    },
    {
      "epoch": 1.2832315237030025,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008464571737184458,
      "loss": 1.3152,
      "step": 14915
    },
    {
      "epoch": 1.2833175600103244,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008464376763031646,
      "loss": 0.8439,
      "step": 14916
    },
    {
      "epoch": 1.283403596317646,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008464181778746206,
      "loss": 1.4838,
      "step": 14917
    },
    {
      "epoch": 1.2834896326249678,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008463986784328708,
      "loss": 1.26,
      "step": 14918
    },
    {
      "epoch": 1.2835756689322895,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008463791779779722,
      "loss": 1.0254,
      "step": 14919
    },
    {
      "epoch": 1.283661705239611,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008463596765099819,
      "loss": 0.9167,
      "step": 14920
    },
    {
      "epoch": 1.2837477415469327,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008463401740289569,
      "loss": 1.2345,
      "step": 14921
    },
    {
      "epoch": 1.2838337778542546,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008463206705349543,
      "loss": 1.2777,
      "step": 14922
    },
    {
      "epoch": 1.2839198141615762,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008463011660280311,
      "loss": 1.9491,
      "step": 14923
    },
    {
      "epoch": 1.2840058504688978,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008462816605082442,
      "loss": 1.3731,
      "step": 14924
    },
    {
      "epoch": 1.2840918867762197,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000846262153975651,
      "loss": 1.9016,
      "step": 14925
    },
    {
      "epoch": 1.2841779230835413,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008462426464303081,
      "loss": 1.8098,
      "step": 14926
    },
    {
      "epoch": 1.284263959390863,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008462231378722728,
      "loss": 1.4397,
      "step": 14927
    },
    {
      "epoch": 1.2843499956981845,
      "grad_norm": 1.25,
      "learning_rate": 0.0008462036283016025,
      "loss": 1.1702,
      "step": 14928
    },
    {
      "epoch": 1.2844360320055064,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008461841177183537,
      "loss": 1.0835,
      "step": 14929
    },
    {
      "epoch": 1.284522068312828,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008461646061225836,
      "loss": 1.4349,
      "step": 14930
    },
    {
      "epoch": 1.2846081046201496,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008461450935143496,
      "loss": 1.4727,
      "step": 14931
    },
    {
      "epoch": 1.2846941409274715,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008461255798937082,
      "loss": 1.6681,
      "step": 14932
    },
    {
      "epoch": 1.284780177234793,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008461060652607171,
      "loss": 1.0268,
      "step": 14933
    },
    {
      "epoch": 1.2848662135421147,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008460865496154328,
      "loss": 1.5712,
      "step": 14934
    },
    {
      "epoch": 1.2849522498494363,
      "grad_norm": 1.3125,
      "learning_rate": 0.000846067032957913,
      "loss": 1.1861,
      "step": 14935
    },
    {
      "epoch": 1.2850382861567582,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008460475152882141,
      "loss": 1.3931,
      "step": 14936
    },
    {
      "epoch": 1.2851243224640798,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008460279966063937,
      "loss": 1.7105,
      "step": 14937
    },
    {
      "epoch": 1.2852103587714017,
      "grad_norm": 2.078125,
      "learning_rate": 0.0008460084769125087,
      "loss": 1.8905,
      "step": 14938
    },
    {
      "epoch": 1.2852963950787233,
      "grad_norm": 0.828125,
      "learning_rate": 0.000845988956206616,
      "loss": 0.6887,
      "step": 14939
    },
    {
      "epoch": 1.285382431386045,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008459694344887732,
      "loss": 1.1553,
      "step": 14940
    },
    {
      "epoch": 1.2854684676933665,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008459499117590369,
      "loss": 1.0875,
      "step": 14941
    },
    {
      "epoch": 1.2855545040006882,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008459303880174645,
      "loss": 2.6302,
      "step": 14942
    },
    {
      "epoch": 1.28564054030801,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008459108632641128,
      "loss": 1.5011,
      "step": 14943
    },
    {
      "epoch": 1.2857265766153316,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008458913374990393,
      "loss": 0.8296,
      "step": 14944
    },
    {
      "epoch": 1.2858126129226535,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008458718107223007,
      "loss": 1.7328,
      "step": 14945
    },
    {
      "epoch": 1.285898649229975,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008458522829339546,
      "loss": 1.1226,
      "step": 14946
    },
    {
      "epoch": 1.2859846855372967,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008458327541340575,
      "loss": 1.0649,
      "step": 14947
    },
    {
      "epoch": 1.2860707218446183,
      "grad_norm": 1.375,
      "learning_rate": 0.0008458132243226673,
      "loss": 1.6117,
      "step": 14948
    },
    {
      "epoch": 1.2861567581519402,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008457936934998403,
      "loss": 0.9965,
      "step": 14949
    },
    {
      "epoch": 1.2862427944592618,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008457741616656341,
      "loss": 0.8941,
      "step": 14950
    },
    {
      "epoch": 1.2863288307665834,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008457546288201058,
      "loss": 1.5701,
      "step": 14951
    },
    {
      "epoch": 1.2864148670739053,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008457350949633125,
      "loss": 1.2808,
      "step": 14952
    },
    {
      "epoch": 1.286500903381227,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008457155600953111,
      "loss": 1.0455,
      "step": 14953
    },
    {
      "epoch": 1.2865869396885485,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008456960242161591,
      "loss": 1.3383,
      "step": 14954
    },
    {
      "epoch": 1.2866729759958702,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008456764873259133,
      "loss": 0.9813,
      "step": 14955
    },
    {
      "epoch": 1.286759012303192,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008456569494246311,
      "loss": 0.9685,
      "step": 14956
    },
    {
      "epoch": 1.2868450486105136,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008456374105123696,
      "loss": 2.0182,
      "step": 14957
    },
    {
      "epoch": 1.2869310849178353,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008456178705891858,
      "loss": 2.073,
      "step": 14958
    },
    {
      "epoch": 1.287017121225157,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008455983296551371,
      "loss": 1.2819,
      "step": 14959
    },
    {
      "epoch": 1.2871031575324787,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008455787877102803,
      "loss": 2.1574,
      "step": 14960
    },
    {
      "epoch": 1.2871891938398003,
      "grad_norm": 1.5,
      "learning_rate": 0.000845559244754673,
      "loss": 1.0301,
      "step": 14961
    },
    {
      "epoch": 1.287275230147122,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008455397007883719,
      "loss": 0.7164,
      "step": 14962
    },
    {
      "epoch": 1.2873612664544438,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008455201558114343,
      "loss": 1.0927,
      "step": 14963
    },
    {
      "epoch": 1.2874473027617654,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008455006098239175,
      "loss": 0.9002,
      "step": 14964
    },
    {
      "epoch": 1.2875333390690873,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008454810628258788,
      "loss": 1.1514,
      "step": 14965
    },
    {
      "epoch": 1.287619375376409,
      "grad_norm": 1.234375,
      "learning_rate": 0.000845461514817375,
      "loss": 1.2414,
      "step": 14966
    },
    {
      "epoch": 1.2877054116837305,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008454419657984635,
      "loss": 0.6872,
      "step": 14967
    },
    {
      "epoch": 1.2877914479910522,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008454224157692017,
      "loss": 1.7338,
      "step": 14968
    },
    {
      "epoch": 1.2878774842983738,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008454028647296462,
      "loss": 0.8713,
      "step": 14969
    },
    {
      "epoch": 1.2879635206056956,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008453833126798546,
      "loss": 1.8541,
      "step": 14970
    },
    {
      "epoch": 1.2880495569130173,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008453637596198839,
      "loss": 1.0849,
      "step": 14971
    },
    {
      "epoch": 1.288135593220339,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008453442055497915,
      "loss": 1.742,
      "step": 14972
    },
    {
      "epoch": 1.2882216295276607,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008453246504696344,
      "loss": 1.7313,
      "step": 14973
    },
    {
      "epoch": 1.2883076658349824,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008453050943794698,
      "loss": 1.719,
      "step": 14974
    },
    {
      "epoch": 1.288393702142304,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008452855372793551,
      "loss": 0.6651,
      "step": 14975
    },
    {
      "epoch": 1.2884797384496258,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008452659791693474,
      "loss": 1.0194,
      "step": 14976
    },
    {
      "epoch": 1.2885657747569474,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008452464200495036,
      "loss": 1.2686,
      "step": 14977
    },
    {
      "epoch": 1.288651811064269,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008452268599198813,
      "loss": 0.7939,
      "step": 14978
    },
    {
      "epoch": 1.288737847371591,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008452072987805376,
      "loss": 1.4624,
      "step": 14979
    },
    {
      "epoch": 1.2888238836789125,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008451877366315297,
      "loss": 1.4428,
      "step": 14980
    },
    {
      "epoch": 1.2889099199862342,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008451681734729148,
      "loss": 1.4842,
      "step": 14981
    },
    {
      "epoch": 1.2889959562935558,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00084514860930475,
      "loss": 0.7687,
      "step": 14982
    },
    {
      "epoch": 1.2890819926008776,
      "grad_norm": 1.375,
      "learning_rate": 0.0008451290441270928,
      "loss": 1.1134,
      "step": 14983
    },
    {
      "epoch": 1.2891680289081993,
      "grad_norm": 2.046875,
      "learning_rate": 0.0008451094779400001,
      "loss": 2.0092,
      "step": 14984
    },
    {
      "epoch": 1.2892540652155209,
      "grad_norm": 1.125,
      "learning_rate": 0.0008450899107435295,
      "loss": 0.8146,
      "step": 14985
    },
    {
      "epoch": 1.2893401015228427,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008450703425377379,
      "loss": 0.9245,
      "step": 14986
    },
    {
      "epoch": 1.2894261378301644,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008450507733226829,
      "loss": 1.1299,
      "step": 14987
    },
    {
      "epoch": 1.289512174137486,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008450312030984212,
      "loss": 1.4379,
      "step": 14988
    },
    {
      "epoch": 1.2895982104448076,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008450116318650105,
      "loss": 1.1153,
      "step": 14989
    },
    {
      "epoch": 1.2896842467521294,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008449920596225078,
      "loss": 1.1248,
      "step": 14990
    },
    {
      "epoch": 1.289770283059451,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008449724863709704,
      "loss": 0.7921,
      "step": 14991
    },
    {
      "epoch": 1.2898563193667727,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008449529121104556,
      "loss": 1.7099,
      "step": 14992
    },
    {
      "epoch": 1.2899423556740945,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008449333368410206,
      "loss": 2.2776,
      "step": 14993
    },
    {
      "epoch": 1.2900283919814162,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008449137605627228,
      "loss": 1.2151,
      "step": 14994
    },
    {
      "epoch": 1.2901144282887378,
      "grad_norm": 1.78125,
      "learning_rate": 0.0008448941832756192,
      "loss": 1.9669,
      "step": 14995
    },
    {
      "epoch": 1.2902004645960594,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008448746049797674,
      "loss": 1.5047,
      "step": 14996
    },
    {
      "epoch": 1.2902865009033813,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008448550256752241,
      "loss": 1.0277,
      "step": 14997
    },
    {
      "epoch": 1.2903725372107029,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008448354453620471,
      "loss": 1.0582,
      "step": 14998
    },
    {
      "epoch": 1.2904585735180247,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008448158640402936,
      "loss": 1.9008,
      "step": 14999
    },
    {
      "epoch": 1.2905446098253464,
      "grad_norm": 1.125,
      "learning_rate": 0.0008447962817100206,
      "loss": 0.8522,
      "step": 15000
    },
    {
      "epoch": 1.290630646132668,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008447766983712855,
      "loss": 1.773,
      "step": 15001
    },
    {
      "epoch": 1.2907166824399896,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008447571140241458,
      "loss": 0.6571,
      "step": 15002
    },
    {
      "epoch": 1.2908027187473112,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008447375286686585,
      "loss": 1.4004,
      "step": 15003
    },
    {
      "epoch": 1.290888755054633,
      "grad_norm": 1.375,
      "learning_rate": 0.000844717942304881,
      "loss": 1.339,
      "step": 15004
    },
    {
      "epoch": 1.2909747913619547,
      "grad_norm": 2.03125,
      "learning_rate": 0.0008446983549328705,
      "loss": 1.1667,
      "step": 15005
    },
    {
      "epoch": 1.2910608276692765,
      "grad_norm": 1.125,
      "learning_rate": 0.0008446787665526844,
      "loss": 0.8445,
      "step": 15006
    },
    {
      "epoch": 1.2911468639765982,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00084465917716438,
      "loss": 1.7626,
      "step": 15007
    },
    {
      "epoch": 1.2912329002839198,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008446395867680144,
      "loss": 1.6541,
      "step": 15008
    },
    {
      "epoch": 1.2913189365912414,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000844619995363645,
      "loss": 0.8495,
      "step": 15009
    },
    {
      "epoch": 1.2914049728985633,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0008446004029513294,
      "loss": 1.6263,
      "step": 15010
    },
    {
      "epoch": 1.2914910092058849,
      "grad_norm": 2.25,
      "learning_rate": 0.0008445808095311245,
      "loss": 1.8772,
      "step": 15011
    },
    {
      "epoch": 1.2915770455132065,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008445612151030877,
      "loss": 1.0015,
      "step": 15012
    },
    {
      "epoch": 1.2916630818205284,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008445416196672763,
      "loss": 1.3059,
      "step": 15013
    },
    {
      "epoch": 1.29174911812785,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008445220232237478,
      "loss": 1.3532,
      "step": 15014
    },
    {
      "epoch": 1.2918351544351716,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008445024257725596,
      "loss": 0.8669,
      "step": 15015
    },
    {
      "epoch": 1.2919211907424932,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008444828273137684,
      "loss": 1.009,
      "step": 15016
    },
    {
      "epoch": 1.292007227049815,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008444632278474322,
      "loss": 1.3869,
      "step": 15017
    },
    {
      "epoch": 1.2920932633571367,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008444436273736078,
      "loss": 1.0355,
      "step": 15018
    },
    {
      "epoch": 1.2921792996644583,
      "grad_norm": 2.578125,
      "learning_rate": 0.000844424025892353,
      "loss": 1.6097,
      "step": 15019
    },
    {
      "epoch": 1.2922653359717802,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008444044234037247,
      "loss": 1.8134,
      "step": 15020
    },
    {
      "epoch": 1.2923513722791018,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008443848199077806,
      "loss": 1.7177,
      "step": 15021
    },
    {
      "epoch": 1.2924374085864234,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000844365215404578,
      "loss": 0.8011,
      "step": 15022
    },
    {
      "epoch": 1.292523444893745,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008443456098941739,
      "loss": 0.9785,
      "step": 15023
    },
    {
      "epoch": 1.292609481201067,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000844326003376626,
      "loss": 1.1018,
      "step": 15024
    },
    {
      "epoch": 1.2926955175083885,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008443063958519914,
      "loss": 1.4221,
      "step": 15025
    },
    {
      "epoch": 1.2927815538157104,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008442867873203277,
      "loss": 1.5782,
      "step": 15026
    },
    {
      "epoch": 1.292867590123032,
      "grad_norm": 2.15625,
      "learning_rate": 0.0008442671777816919,
      "loss": 2.3798,
      "step": 15027
    },
    {
      "epoch": 1.2929536264303536,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008442475672361419,
      "loss": 0.8882,
      "step": 15028
    },
    {
      "epoch": 1.2930396627376752,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008442279556837343,
      "loss": 1.2286,
      "step": 15029
    },
    {
      "epoch": 1.2931256990449969,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008442083431245271,
      "loss": 1.5983,
      "step": 15030
    },
    {
      "epoch": 1.2932117353523187,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008441887295585775,
      "loss": 1.2881,
      "step": 15031
    },
    {
      "epoch": 1.2932977716596403,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008441691149859427,
      "loss": 1.8295,
      "step": 15032
    },
    {
      "epoch": 1.2933838079669622,
      "grad_norm": 2.015625,
      "learning_rate": 0.0008441494994066802,
      "loss": 1.4524,
      "step": 15033
    },
    {
      "epoch": 1.2934698442742838,
      "grad_norm": 1.25,
      "learning_rate": 0.0008441298828208473,
      "loss": 1.1586,
      "step": 15034
    },
    {
      "epoch": 1.2935558805816054,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008441102652285015,
      "loss": 1.905,
      "step": 15035
    },
    {
      "epoch": 1.293641916888927,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008440906466297,
      "loss": 1.0676,
      "step": 15036
    },
    {
      "epoch": 1.293727953196249,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008440710270245004,
      "loss": 0.7144,
      "step": 15037
    },
    {
      "epoch": 1.2938139895035705,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008440514064129599,
      "loss": 1.1457,
      "step": 15038
    },
    {
      "epoch": 1.2939000258108921,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000844031784795136,
      "loss": 1.0197,
      "step": 15039
    },
    {
      "epoch": 1.293986062118214,
      "grad_norm": 1.8828125,
      "learning_rate": 0.000844012162171086,
      "loss": 0.5566,
      "step": 15040
    },
    {
      "epoch": 1.2940720984255356,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008439925385408673,
      "loss": 2.0158,
      "step": 15041
    },
    {
      "epoch": 1.2941581347328572,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008439729139045374,
      "loss": 0.9222,
      "step": 15042
    },
    {
      "epoch": 1.2942441710401789,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008439532882621537,
      "loss": 1.0773,
      "step": 15043
    },
    {
      "epoch": 1.2943302073475007,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008439336616137733,
      "loss": 1.1874,
      "step": 15044
    },
    {
      "epoch": 1.2944162436548223,
      "grad_norm": 1.125,
      "learning_rate": 0.000843914033959454,
      "loss": 1.6447,
      "step": 15045
    },
    {
      "epoch": 1.294502279962144,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000843894405299253,
      "loss": 1.6506,
      "step": 15046
    },
    {
      "epoch": 1.2945883162694658,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008438747756332277,
      "loss": 1.2456,
      "step": 15047
    },
    {
      "epoch": 1.2946743525767874,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008438551449614356,
      "loss": 0.76,
      "step": 15048
    },
    {
      "epoch": 1.294760388884109,
      "grad_norm": 1.5,
      "learning_rate": 0.0008438355132839342,
      "loss": 2.1452,
      "step": 15049
    },
    {
      "epoch": 1.2948464251914307,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008438158806007805,
      "loss": 1.2143,
      "step": 15050
    },
    {
      "epoch": 1.2949324614987525,
      "grad_norm": 1.78125,
      "learning_rate": 0.0008437962469120323,
      "loss": 1.5783,
      "step": 15051
    },
    {
      "epoch": 1.2950184978060741,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008437766122177473,
      "loss": 1.5349,
      "step": 15052
    },
    {
      "epoch": 1.295104534113396,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008437569765179822,
      "loss": 1.5312,
      "step": 15053
    },
    {
      "epoch": 1.2951905704207176,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008437373398127949,
      "loss": 1.3704,
      "step": 15054
    },
    {
      "epoch": 1.2952766067280392,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008437177021022427,
      "loss": 1.3401,
      "step": 15055
    },
    {
      "epoch": 1.2953626430353609,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008436980633863831,
      "loss": 1.1177,
      "step": 15056
    },
    {
      "epoch": 1.2954486793426825,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008436784236652735,
      "loss": 1.3893,
      "step": 15057
    },
    {
      "epoch": 1.2955347156500043,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008436587829389713,
      "loss": 1.5581,
      "step": 15058
    },
    {
      "epoch": 1.295620751957326,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008436391412075341,
      "loss": 1.2567,
      "step": 15059
    },
    {
      "epoch": 1.2957067882646478,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008436194984710192,
      "loss": 1.8801,
      "step": 15060
    },
    {
      "epoch": 1.2957928245719694,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008435998547294841,
      "loss": 1.3767,
      "step": 15061
    },
    {
      "epoch": 1.295878860879291,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008435802099829863,
      "loss": 0.8947,
      "step": 15062
    },
    {
      "epoch": 1.2959648971866127,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008435605642315833,
      "loss": 1.4555,
      "step": 15063
    },
    {
      "epoch": 1.2960509334939345,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008435409174753323,
      "loss": 1.1043,
      "step": 15064
    },
    {
      "epoch": 1.2961369698012561,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008435212697142909,
      "loss": 1.2059,
      "step": 15065
    },
    {
      "epoch": 1.2962230061085778,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008435016209485166,
      "loss": 1.2521,
      "step": 15066
    },
    {
      "epoch": 1.2963090424158996,
      "grad_norm": 1.265625,
      "learning_rate": 0.000843481971178067,
      "loss": 1.8021,
      "step": 15067
    },
    {
      "epoch": 1.2963950787232212,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008434623204029994,
      "loss": 1.5121,
      "step": 15068
    },
    {
      "epoch": 1.2964811150305429,
      "grad_norm": 1.25,
      "learning_rate": 0.0008434426686233714,
      "loss": 1.9067,
      "step": 15069
    },
    {
      "epoch": 1.2965671513378645,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008434230158392402,
      "loss": 0.8383,
      "step": 15070
    },
    {
      "epoch": 1.2966531876451863,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008434033620506636,
      "loss": 1.2176,
      "step": 15071
    },
    {
      "epoch": 1.296739223952508,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008433837072576989,
      "loss": 1.2369,
      "step": 15072
    },
    {
      "epoch": 1.2968252602598296,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008433640514604035,
      "loss": 0.8338,
      "step": 15073
    },
    {
      "epoch": 1.2969112965671514,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008433443946588352,
      "loss": 0.9677,
      "step": 15074
    },
    {
      "epoch": 1.296997332874473,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008433247368530512,
      "loss": 1.6526,
      "step": 15075
    },
    {
      "epoch": 1.2970833691817947,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008433050780431093,
      "loss": 1.4935,
      "step": 15076
    },
    {
      "epoch": 1.2971694054891163,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008432854182290668,
      "loss": 2.0031,
      "step": 15077
    },
    {
      "epoch": 1.2972554417964381,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000843265757410981,
      "loss": 2.3468,
      "step": 15078
    },
    {
      "epoch": 1.2973414781037598,
      "grad_norm": 1.5,
      "learning_rate": 0.0008432460955889098,
      "loss": 1.1594,
      "step": 15079
    },
    {
      "epoch": 1.2974275144110814,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008432264327629104,
      "loss": 1.5085,
      "step": 15080
    },
    {
      "epoch": 1.2975135507184032,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008432067689330407,
      "loss": 1.1922,
      "step": 15081
    },
    {
      "epoch": 1.2975995870257249,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008431871040993577,
      "loss": 1.2418,
      "step": 15082
    },
    {
      "epoch": 1.2976856233330465,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008431674382619192,
      "loss": 1.1301,
      "step": 15083
    },
    {
      "epoch": 1.2977716596403681,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008431477714207829,
      "loss": 0.9572,
      "step": 15084
    },
    {
      "epoch": 1.29785769594769,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008431281035760059,
      "loss": 0.8689,
      "step": 15085
    },
    {
      "epoch": 1.2979437322550116,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000843108434727646,
      "loss": 0.8515,
      "step": 15086
    },
    {
      "epoch": 1.2980297685623334,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008430887648757606,
      "loss": 1.5544,
      "step": 15087
    },
    {
      "epoch": 1.298115804869655,
      "grad_norm": 2.09375,
      "learning_rate": 0.0008430690940204073,
      "loss": 0.8898,
      "step": 15088
    },
    {
      "epoch": 1.2982018411769767,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008430494221616435,
      "loss": 0.9012,
      "step": 15089
    },
    {
      "epoch": 1.2982878774842983,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008430297492995272,
      "loss": 1.5332,
      "step": 15090
    },
    {
      "epoch": 1.29837391379162,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008430100754341154,
      "loss": 1.4844,
      "step": 15091
    },
    {
      "epoch": 1.2984599500989418,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008429904005654658,
      "loss": 1.1953,
      "step": 15092
    },
    {
      "epoch": 1.2985459864062634,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000842970724693636,
      "loss": 1.153,
      "step": 15093
    },
    {
      "epoch": 1.2986320227135852,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008429510478186836,
      "loss": 1.9029,
      "step": 15094
    },
    {
      "epoch": 1.2987180590209069,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000842931369940666,
      "loss": 1.5009,
      "step": 15095
    },
    {
      "epoch": 1.2988040953282285,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008429116910596409,
      "loss": 0.6967,
      "step": 15096
    },
    {
      "epoch": 1.2988901316355501,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008428920111756657,
      "loss": 1.2742,
      "step": 15097
    },
    {
      "epoch": 1.298976167942872,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008428723302887982,
      "loss": 1.8333,
      "step": 15098
    },
    {
      "epoch": 1.2990622042501936,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008428526483990959,
      "loss": 2.1786,
      "step": 15099
    },
    {
      "epoch": 1.2991482405575152,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000842832965506616,
      "loss": 1.5028,
      "step": 15100
    },
    {
      "epoch": 1.299234276864837,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008428132816114166,
      "loss": 0.7106,
      "step": 15101
    },
    {
      "epoch": 1.2993203131721587,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000842793596713555,
      "loss": 0.9676,
      "step": 15102
    },
    {
      "epoch": 1.2994063494794803,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008427739108130886,
      "loss": 1.1132,
      "step": 15103
    },
    {
      "epoch": 1.299492385786802,
      "grad_norm": 17.375,
      "learning_rate": 0.0008427542239100753,
      "loss": 1.5985,
      "step": 15104
    },
    {
      "epoch": 1.2995784220941238,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008427345360045726,
      "loss": 1.4064,
      "step": 15105
    },
    {
      "epoch": 1.2996644584014454,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008427148470966379,
      "loss": 1.9149,
      "step": 15106
    },
    {
      "epoch": 1.299750494708767,
      "grad_norm": 1.265625,
      "learning_rate": 0.000842695157186329,
      "loss": 1.987,
      "step": 15107
    },
    {
      "epoch": 1.2998365310160889,
      "grad_norm": 1.125,
      "learning_rate": 0.0008426754662737035,
      "loss": 0.9704,
      "step": 15108
    },
    {
      "epoch": 1.2999225673234105,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008426557743588188,
      "loss": 1.4988,
      "step": 15109
    },
    {
      "epoch": 1.3000086036307321,
      "grad_norm": 1.25,
      "learning_rate": 0.0008426360814417327,
      "loss": 1.1868,
      "step": 15110
    },
    {
      "epoch": 1.3000946399380537,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008426163875225027,
      "loss": 1.3584,
      "step": 15111
    },
    {
      "epoch": 1.3001806762453756,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008425966926011863,
      "loss": 1.7368,
      "step": 15112
    },
    {
      "epoch": 1.3002667125526972,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008425769966778411,
      "loss": 1.6499,
      "step": 15113
    },
    {
      "epoch": 1.300352748860019,
      "grad_norm": 1.125,
      "learning_rate": 0.0008425572997525251,
      "loss": 1.4497,
      "step": 15114
    },
    {
      "epoch": 1.3004387851673407,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008425376018252955,
      "loss": 1.6003,
      "step": 15115
    },
    {
      "epoch": 1.3005248214746623,
      "grad_norm": 1.359375,
      "learning_rate": 0.00084251790289621,
      "loss": 1.5317,
      "step": 15116
    },
    {
      "epoch": 1.300610857781984,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008424982029653263,
      "loss": 1.4683,
      "step": 15117
    },
    {
      "epoch": 1.3006968940893056,
      "grad_norm": 4.15625,
      "learning_rate": 0.0008424785020327019,
      "loss": 0.8624,
      "step": 15118
    },
    {
      "epoch": 1.3007829303966274,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008424588000983946,
      "loss": 1.5911,
      "step": 15119
    },
    {
      "epoch": 1.300868966703949,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008424390971624618,
      "loss": 1.5322,
      "step": 15120
    },
    {
      "epoch": 1.3009550030112709,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008424193932249613,
      "loss": 1.6253,
      "step": 15121
    },
    {
      "epoch": 1.3010410393185925,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008423996882859507,
      "loss": 0.9908,
      "step": 15122
    },
    {
      "epoch": 1.3011270756259141,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008423799823454876,
      "loss": 0.9067,
      "step": 15123
    },
    {
      "epoch": 1.3012131119332357,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008423602754036296,
      "loss": 1.253,
      "step": 15124
    },
    {
      "epoch": 1.3012991482405576,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008423405674604345,
      "loss": 1.5074,
      "step": 15125
    },
    {
      "epoch": 1.3013851845478792,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008423208585159596,
      "loss": 1.5616,
      "step": 15126
    },
    {
      "epoch": 1.3014712208552008,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008423011485702629,
      "loss": 1.1234,
      "step": 15127
    },
    {
      "epoch": 1.3015572571625227,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008422814376234018,
      "loss": 1.2291,
      "step": 15128
    },
    {
      "epoch": 1.3016432934698443,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008422617256754342,
      "loss": 0.8366,
      "step": 15129
    },
    {
      "epoch": 1.301729329777166,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008422420127264176,
      "loss": 1.182,
      "step": 15130
    },
    {
      "epoch": 1.3018153660844876,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008422222987764096,
      "loss": 1.4627,
      "step": 15131
    },
    {
      "epoch": 1.3019014023918094,
      "grad_norm": 1.078125,
      "learning_rate": 0.000842202583825468,
      "loss": 1.8821,
      "step": 15132
    },
    {
      "epoch": 1.301987438699131,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008421828678736504,
      "loss": 1.0543,
      "step": 15133
    },
    {
      "epoch": 1.3020734750064527,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008421631509210143,
      "loss": 2.1754,
      "step": 15134
    },
    {
      "epoch": 1.3021595113137745,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008421434329676179,
      "loss": 1.7861,
      "step": 15135
    },
    {
      "epoch": 1.3022455476210961,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008421237140135182,
      "loss": 0.6135,
      "step": 15136
    },
    {
      "epoch": 1.3023315839284177,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008421039940587732,
      "loss": 1.3696,
      "step": 15137
    },
    {
      "epoch": 1.3024176202357394,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008420842731034406,
      "loss": 0.768,
      "step": 15138
    },
    {
      "epoch": 1.3025036565430612,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008420645511475781,
      "loss": 1.0903,
      "step": 15139
    },
    {
      "epoch": 1.3025896928503828,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008420448281912431,
      "loss": 1.2938,
      "step": 15140
    },
    {
      "epoch": 1.3026757291577047,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0008420251042344937,
      "loss": 2.4316,
      "step": 15141
    },
    {
      "epoch": 1.3027617654650263,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008420053792773874,
      "loss": 1.6314,
      "step": 15142
    },
    {
      "epoch": 1.302847801772348,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008419856533199819,
      "loss": 0.8669,
      "step": 15143
    },
    {
      "epoch": 1.3029338380796696,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008419659263623348,
      "loss": 1.143,
      "step": 15144
    },
    {
      "epoch": 1.3030198743869912,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008419461984045039,
      "loss": 0.9554,
      "step": 15145
    },
    {
      "epoch": 1.303105910694313,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008419264694465469,
      "loss": 1.9512,
      "step": 15146
    },
    {
      "epoch": 1.3031919470016347,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008419067394885213,
      "loss": 1.3574,
      "step": 15147
    },
    {
      "epoch": 1.3032779833089565,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008418870085304851,
      "loss": 1.654,
      "step": 15148
    },
    {
      "epoch": 1.3033640196162781,
      "grad_norm": 1.21875,
      "learning_rate": 0.000841867276572496,
      "loss": 0.5764,
      "step": 15149
    },
    {
      "epoch": 1.3034500559235997,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008418475436146115,
      "loss": 2.1678,
      "step": 15150
    },
    {
      "epoch": 1.3035360922309214,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008418278096568895,
      "loss": 1.8201,
      "step": 15151
    },
    {
      "epoch": 1.3036221285382432,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008418080746993875,
      "loss": 1.0409,
      "step": 15152
    },
    {
      "epoch": 1.3037081648455648,
      "grad_norm": 1.5,
      "learning_rate": 0.0008417883387421634,
      "loss": 1.6029,
      "step": 15153
    },
    {
      "epoch": 1.3037942011528865,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008417686017852748,
      "loss": 1.886,
      "step": 15154
    },
    {
      "epoch": 1.3038802374602083,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008417488638287798,
      "loss": 1.1801,
      "step": 15155
    },
    {
      "epoch": 1.30396627376753,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008417291248727357,
      "loss": 1.6182,
      "step": 15156
    },
    {
      "epoch": 1.3040523100748516,
      "grad_norm": 2.015625,
      "learning_rate": 0.0008417093849172003,
      "loss": 1.4142,
      "step": 15157
    },
    {
      "epoch": 1.3041383463821732,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008416896439622315,
      "loss": 1.4418,
      "step": 15158
    },
    {
      "epoch": 1.304224382689495,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008416699020078869,
      "loss": 1.1219,
      "step": 15159
    },
    {
      "epoch": 1.3043104189968167,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008416501590542241,
      "loss": 1.1083,
      "step": 15160
    },
    {
      "epoch": 1.3043964553041383,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008416304151013012,
      "loss": 1.5908,
      "step": 15161
    },
    {
      "epoch": 1.3044824916114601,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000841610670149176,
      "loss": 1.0379,
      "step": 15162
    },
    {
      "epoch": 1.3045685279187818,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008415909241979056,
      "loss": 1.3677,
      "step": 15163
    },
    {
      "epoch": 1.3046545642261034,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008415711772475485,
      "loss": 1.4915,
      "step": 15164
    },
    {
      "epoch": 1.304740600533425,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008415514292981621,
      "loss": 1.0577,
      "step": 15165
    },
    {
      "epoch": 1.3048266368407468,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008415316803498041,
      "loss": 1.4699,
      "step": 15166
    },
    {
      "epoch": 1.3049126731480685,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008415119304025323,
      "loss": 1.9353,
      "step": 15167
    },
    {
      "epoch": 1.30499870945539,
      "grad_norm": 1.75,
      "learning_rate": 0.0008414921794564047,
      "loss": 1.6268,
      "step": 15168
    },
    {
      "epoch": 1.305084745762712,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008414724275114788,
      "loss": 2.1329,
      "step": 15169
    },
    {
      "epoch": 1.3051707820700336,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008414526745678123,
      "loss": 0.7973,
      "step": 15170
    },
    {
      "epoch": 1.3052568183773552,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008414329206254634,
      "loss": 1.0448,
      "step": 15171
    },
    {
      "epoch": 1.3053428546846768,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008414131656844896,
      "loss": 2.1328,
      "step": 15172
    },
    {
      "epoch": 1.3054288909919987,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008413934097449486,
      "loss": 1.6382,
      "step": 15173
    },
    {
      "epoch": 1.3055149272993203,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008413736528068982,
      "loss": 0.7885,
      "step": 15174
    },
    {
      "epoch": 1.3056009636066421,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008413538948703965,
      "loss": 0.996,
      "step": 15175
    },
    {
      "epoch": 1.3056869999139638,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008413341359355009,
      "loss": 1.9036,
      "step": 15176
    },
    {
      "epoch": 1.3057730362212854,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008413143760022693,
      "loss": 1.9285,
      "step": 15177
    },
    {
      "epoch": 1.305859072528607,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008412946150707595,
      "loss": 1.5529,
      "step": 15178
    },
    {
      "epoch": 1.3059451088359286,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008412748531410294,
      "loss": 1.188,
      "step": 15179
    },
    {
      "epoch": 1.3060311451432505,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008412550902131368,
      "loss": 1.6727,
      "step": 15180
    },
    {
      "epoch": 1.306117181450572,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008412353262871393,
      "loss": 0.8316,
      "step": 15181
    },
    {
      "epoch": 1.306203217757894,
      "grad_norm": 1.109375,
      "learning_rate": 0.000841215561363095,
      "loss": 1.4701,
      "step": 15182
    },
    {
      "epoch": 1.3062892540652156,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008411957954410614,
      "loss": 0.9459,
      "step": 15183
    },
    {
      "epoch": 1.3063752903725372,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008411760285210965,
      "loss": 0.8296,
      "step": 15184
    },
    {
      "epoch": 1.3064613266798588,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000841156260603258,
      "loss": 1.3341,
      "step": 15185
    },
    {
      "epoch": 1.3065473629871807,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008411364916876039,
      "loss": 1.4439,
      "step": 15186
    },
    {
      "epoch": 1.3066333992945023,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0008411167217741919,
      "loss": 1.1783,
      "step": 15187
    },
    {
      "epoch": 1.306719435601824,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008410969508630797,
      "loss": 1.1252,
      "step": 15188
    },
    {
      "epoch": 1.3068054719091458,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008410771789543255,
      "loss": 1.0432,
      "step": 15189
    },
    {
      "epoch": 1.3068915082164674,
      "grad_norm": 2.296875,
      "learning_rate": 0.0008410574060479867,
      "loss": 1.3664,
      "step": 15190
    },
    {
      "epoch": 1.306977544523789,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008410376321441213,
      "loss": 1.8149,
      "step": 15191
    },
    {
      "epoch": 1.3070635808311106,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008410178572427873,
      "loss": 0.8522,
      "step": 15192
    },
    {
      "epoch": 1.3071496171384325,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008409980813440423,
      "loss": 0.8345,
      "step": 15193
    },
    {
      "epoch": 1.307235653445754,
      "grad_norm": 1.828125,
      "learning_rate": 0.0008409783044479442,
      "loss": 1.0672,
      "step": 15194
    },
    {
      "epoch": 1.3073216897530757,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008409585265545509,
      "loss": 1.5706,
      "step": 15195
    },
    {
      "epoch": 1.3074077260603976,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0008409387476639203,
      "loss": 1.6409,
      "step": 15196
    },
    {
      "epoch": 1.3074937623677192,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00084091896777611,
      "loss": 1.1973,
      "step": 15197
    },
    {
      "epoch": 1.3075797986750408,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008408991868911782,
      "loss": 1.0989,
      "step": 15198
    },
    {
      "epoch": 1.3076658349823624,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008408794050091824,
      "loss": 1.4654,
      "step": 15199
    },
    {
      "epoch": 1.3077518712896843,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008408596221301808,
      "loss": 1.2579,
      "step": 15200
    },
    {
      "epoch": 1.307837907597006,
      "grad_norm": 1.40625,
      "learning_rate": 0.000840839838254231,
      "loss": 0.9634,
      "step": 15201
    },
    {
      "epoch": 1.3079239439043278,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008408200533813913,
      "loss": 0.7851,
      "step": 15202
    },
    {
      "epoch": 1.3080099802116494,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008408002675117188,
      "loss": 1.352,
      "step": 15203
    },
    {
      "epoch": 1.308096016518971,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008407804806452719,
      "loss": 1.3274,
      "step": 15204
    },
    {
      "epoch": 1.3081820528262926,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008407606927821083,
      "loss": 0.8215,
      "step": 15205
    },
    {
      "epoch": 1.3082680891336143,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008407409039222862,
      "loss": 0.8879,
      "step": 15206
    },
    {
      "epoch": 1.308354125440936,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008407211140658631,
      "loss": 0.7872,
      "step": 15207
    },
    {
      "epoch": 1.3084401617482577,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000840701323212897,
      "loss": 0.7712,
      "step": 15208
    },
    {
      "epoch": 1.3085261980555796,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008406815313634457,
      "loss": 2.126,
      "step": 15209
    },
    {
      "epoch": 1.3086122343629012,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008406617385175672,
      "loss": 0.9295,
      "step": 15210
    },
    {
      "epoch": 1.3086982706702228,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008406419446753196,
      "loss": 1.06,
      "step": 15211
    },
    {
      "epoch": 1.3087843069775444,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008406221498367604,
      "loss": 0.7566,
      "step": 15212
    },
    {
      "epoch": 1.3088703432848663,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008406023540019476,
      "loss": 1.3941,
      "step": 15213
    },
    {
      "epoch": 1.308956379592188,
      "grad_norm": 1.828125,
      "learning_rate": 0.0008405825571709393,
      "loss": 1.4265,
      "step": 15214
    },
    {
      "epoch": 1.3090424158995095,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008405627593437932,
      "loss": 1.1864,
      "step": 15215
    },
    {
      "epoch": 1.3091284522068314,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008405429605205673,
      "loss": 1.2285,
      "step": 15216
    },
    {
      "epoch": 1.309214488514153,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008405231607013193,
      "loss": 0.7562,
      "step": 15217
    },
    {
      "epoch": 1.3093005248214746,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008405033598861074,
      "loss": 0.8487,
      "step": 15218
    },
    {
      "epoch": 1.3093865611287963,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008404835580749895,
      "loss": 1.4396,
      "step": 15219
    },
    {
      "epoch": 1.309472597436118,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008404637552680232,
      "loss": 0.7872,
      "step": 15220
    },
    {
      "epoch": 1.3095586337434397,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008404439514652668,
      "loss": 1.2644,
      "step": 15221
    },
    {
      "epoch": 1.3096446700507614,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008404241466667779,
      "loss": 1.0711,
      "step": 15222
    },
    {
      "epoch": 1.3097307063580832,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008404043408726147,
      "loss": 2.1838,
      "step": 15223
    },
    {
      "epoch": 1.3098167426654048,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000840384534082835,
      "loss": 0.967,
      "step": 15224
    },
    {
      "epoch": 1.3099027789727264,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008403647262974967,
      "loss": 0.8775,
      "step": 15225
    },
    {
      "epoch": 1.309988815280048,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008403449175166577,
      "loss": 0.9954,
      "step": 15226
    },
    {
      "epoch": 1.31007485158737,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0008403251077403761,
      "loss": 0.8171,
      "step": 15227
    },
    {
      "epoch": 1.3101608878946915,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008403052969687099,
      "loss": 1.2499,
      "step": 15228
    },
    {
      "epoch": 1.3102469242020134,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008402854852017168,
      "loss": 0.9031,
      "step": 15229
    },
    {
      "epoch": 1.310332960509335,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008402656724394547,
      "loss": 1.6466,
      "step": 15230
    },
    {
      "epoch": 1.3104189968166566,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008402458586819817,
      "loss": 1.4675,
      "step": 15231
    },
    {
      "epoch": 1.3105050331239783,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008402260439293558,
      "loss": 1.2729,
      "step": 15232
    },
    {
      "epoch": 1.3105910694312999,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000840206228181635,
      "loss": 1.6811,
      "step": 15233
    },
    {
      "epoch": 1.3106771057386217,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000840186411438877,
      "loss": 1.865,
      "step": 15234
    },
    {
      "epoch": 1.3107631420459434,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008401665937011401,
      "loss": 1.1532,
      "step": 15235
    },
    {
      "epoch": 1.3108491783532652,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008401467749684819,
      "loss": 0.6076,
      "step": 15236
    },
    {
      "epoch": 1.3109352146605868,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0008401269552409605,
      "loss": 0.8442,
      "step": 15237
    },
    {
      "epoch": 1.3110212509679084,
      "grad_norm": 0.921875,
      "learning_rate": 0.000840107134518634,
      "loss": 1.822,
      "step": 15238
    },
    {
      "epoch": 1.31110728727523,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008400873128015603,
      "loss": 1.2478,
      "step": 15239
    },
    {
      "epoch": 1.311193323582552,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008400674900897973,
      "loss": 0.8249,
      "step": 15240
    },
    {
      "epoch": 1.3112793598898735,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008400476663834029,
      "loss": 1.4512,
      "step": 15241
    },
    {
      "epoch": 1.3113653961971952,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008400278416824354,
      "loss": 1.1126,
      "step": 15242
    },
    {
      "epoch": 1.311451432504517,
      "grad_norm": 1.8125,
      "learning_rate": 0.0008400080159869525,
      "loss": 1.3414,
      "step": 15243
    },
    {
      "epoch": 1.3115374688118386,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008399881892970124,
      "loss": 1.3329,
      "step": 15244
    },
    {
      "epoch": 1.3116235051191603,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008399683616126729,
      "loss": 1.3137,
      "step": 15245
    },
    {
      "epoch": 1.3117095414264819,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008399485329339919,
      "loss": 0.9333,
      "step": 15246
    },
    {
      "epoch": 1.3117955777338037,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008399287032610278,
      "loss": 0.9129,
      "step": 15247
    },
    {
      "epoch": 1.3118816140411254,
      "grad_norm": 1.25,
      "learning_rate": 0.0008399088725938384,
      "loss": 1.4312,
      "step": 15248
    },
    {
      "epoch": 1.311967650348447,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008398890409324815,
      "loss": 1.8843,
      "step": 15249
    },
    {
      "epoch": 1.3120536866557688,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008398692082770152,
      "loss": 2.0734,
      "step": 15250
    },
    {
      "epoch": 1.3121397229630904,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008398493746274978,
      "loss": 0.7984,
      "step": 15251
    },
    {
      "epoch": 1.312225759270412,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008398295399839868,
      "loss": 1.2957,
      "step": 15252
    },
    {
      "epoch": 1.3123117955777337,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008398097043465407,
      "loss": 1.1724,
      "step": 15253
    },
    {
      "epoch": 1.3123978318850555,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008397898677152172,
      "loss": 0.9281,
      "step": 15254
    },
    {
      "epoch": 1.3124838681923772,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008397700300900744,
      "loss": 1.3123,
      "step": 15255
    },
    {
      "epoch": 1.3125699044996988,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008397501914711705,
      "loss": 1.5991,
      "step": 15256
    },
    {
      "epoch": 1.3126559408070206,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008397303518585632,
      "loss": 1.3927,
      "step": 15257
    },
    {
      "epoch": 1.3127419771143423,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0008397105112523108,
      "loss": 1.4,
      "step": 15258
    },
    {
      "epoch": 1.3128280134216639,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008396906696524711,
      "loss": 1.2747,
      "step": 15259
    },
    {
      "epoch": 1.3129140497289855,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008396708270591023,
      "loss": 1.508,
      "step": 15260
    },
    {
      "epoch": 1.3130000860363074,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008396509834722626,
      "loss": 1.4731,
      "step": 15261
    },
    {
      "epoch": 1.313086122343629,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008396311388920095,
      "loss": 1.0311,
      "step": 15262
    },
    {
      "epoch": 1.3131721586509508,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008396112933184016,
      "loss": 1.1222,
      "step": 15263
    },
    {
      "epoch": 1.3132581949582725,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008395914467514967,
      "loss": 1.8785,
      "step": 15264
    },
    {
      "epoch": 1.313344231265594,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008395715991913529,
      "loss": 1.5161,
      "step": 15265
    },
    {
      "epoch": 1.3134302675729157,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008395517506380279,
      "loss": 1.3089,
      "step": 15266
    },
    {
      "epoch": 1.3135163038802373,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008395319010915804,
      "loss": 1.5451,
      "step": 15267
    },
    {
      "epoch": 1.3136023401875592,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000839512050552068,
      "loss": 1.6556,
      "step": 15268
    },
    {
      "epoch": 1.3136883764948808,
      "grad_norm": 1.703125,
      "learning_rate": 0.000839492199019549,
      "loss": 2.3084,
      "step": 15269
    },
    {
      "epoch": 1.3137744128022026,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008394723464940812,
      "loss": 1.12,
      "step": 15270
    },
    {
      "epoch": 1.3138604491095243,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008394524929757229,
      "loss": 1.4222,
      "step": 15271
    },
    {
      "epoch": 1.3139464854168459,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000839432638464532,
      "loss": 0.8476,
      "step": 15272
    },
    {
      "epoch": 1.3140325217241675,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008394127829605667,
      "loss": 1.2092,
      "step": 15273
    },
    {
      "epoch": 1.3141185580314894,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008393929264638851,
      "loss": 1.1093,
      "step": 15274
    },
    {
      "epoch": 1.314204594338811,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008393730689745451,
      "loss": 0.7389,
      "step": 15275
    },
    {
      "epoch": 1.3142906306461326,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008393532104926048,
      "loss": 1.0273,
      "step": 15276
    },
    {
      "epoch": 1.3143766669534545,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008393333510181225,
      "loss": 1.6206,
      "step": 15277
    },
    {
      "epoch": 1.314462703260776,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008393134905511561,
      "loss": 1.2428,
      "step": 15278
    },
    {
      "epoch": 1.3145487395680977,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008392936290917637,
      "loss": 1.2686,
      "step": 15279
    },
    {
      "epoch": 1.3146347758754193,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008392737666400034,
      "loss": 0.8516,
      "step": 15280
    },
    {
      "epoch": 1.3147208121827412,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008392539031959334,
      "loss": 1.1827,
      "step": 15281
    },
    {
      "epoch": 1.3148068484900628,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008392340387596116,
      "loss": 1.544,
      "step": 15282
    },
    {
      "epoch": 1.3148928847973844,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008392141733310962,
      "loss": 1.1287,
      "step": 15283
    },
    {
      "epoch": 1.3149789211047063,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008391943069104455,
      "loss": 1.1985,
      "step": 15284
    },
    {
      "epoch": 1.315064957412028,
      "grad_norm": 1.25,
      "learning_rate": 0.0008391744394977174,
      "loss": 1.0661,
      "step": 15285
    },
    {
      "epoch": 1.3151509937193495,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00083915457109297,
      "loss": 1.3387,
      "step": 15286
    },
    {
      "epoch": 1.3152370300266711,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008391347016962614,
      "loss": 1.5821,
      "step": 15287
    },
    {
      "epoch": 1.315323066333993,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008391148313076497,
      "loss": 1.4549,
      "step": 15288
    },
    {
      "epoch": 1.3154091026413146,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008390949599271932,
      "loss": 1.5286,
      "step": 15289
    },
    {
      "epoch": 1.3154951389486365,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008390750875549497,
      "loss": 1.1607,
      "step": 15290
    },
    {
      "epoch": 1.315581175255958,
      "grad_norm": 1.5,
      "learning_rate": 0.0008390552141909778,
      "loss": 1.4987,
      "step": 15291
    },
    {
      "epoch": 1.3156672115632797,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008390353398353352,
      "loss": 2.0173,
      "step": 15292
    },
    {
      "epoch": 1.3157532478706013,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008390154644880801,
      "loss": 1.2615,
      "step": 15293
    },
    {
      "epoch": 1.315839284177923,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008389955881492708,
      "loss": 0.896,
      "step": 15294
    },
    {
      "epoch": 1.3159253204852448,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008389757108189653,
      "loss": 0.9186,
      "step": 15295
    },
    {
      "epoch": 1.3160113567925664,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008389558324972217,
      "loss": 1.4286,
      "step": 15296
    },
    {
      "epoch": 1.3160973930998883,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008389359531840983,
      "loss": 1.5862,
      "step": 15297
    },
    {
      "epoch": 1.31618342940721,
      "grad_norm": 1.8203125,
      "learning_rate": 0.000838916072879653,
      "loss": 1.3927,
      "step": 15298
    },
    {
      "epoch": 1.3162694657145315,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008388961915839443,
      "loss": 2.203,
      "step": 15299
    },
    {
      "epoch": 1.3163555020218531,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008388763092970301,
      "loss": 1.4558,
      "step": 15300
    },
    {
      "epoch": 1.316441538329175,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008388564260189686,
      "loss": 0.76,
      "step": 15301
    },
    {
      "epoch": 1.3165275746364966,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008388365417498179,
      "loss": 1.1711,
      "step": 15302
    },
    {
      "epoch": 1.3166136109438182,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008388166564896363,
      "loss": 1.0182,
      "step": 15303
    },
    {
      "epoch": 1.31669964725114,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008387967702384818,
      "loss": 1.5903,
      "step": 15304
    },
    {
      "epoch": 1.3167856835584617,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008387768829964127,
      "loss": 1.5931,
      "step": 15305
    },
    {
      "epoch": 1.3168717198657833,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008387569947634873,
      "loss": 1.088,
      "step": 15306
    },
    {
      "epoch": 1.316957756173105,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008387371055397632,
      "loss": 1.1703,
      "step": 15307
    },
    {
      "epoch": 1.3170437924804268,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008387172153252991,
      "loss": 1.4946,
      "step": 15308
    },
    {
      "epoch": 1.3171298287877484,
      "grad_norm": 1.390625,
      "learning_rate": 0.000838697324120153,
      "loss": 1.5402,
      "step": 15309
    },
    {
      "epoch": 1.31721586509507,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008386774319243832,
      "loss": 1.4325,
      "step": 15310
    },
    {
      "epoch": 1.317301901402392,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008386575387380477,
      "loss": 1.7255,
      "step": 15311
    },
    {
      "epoch": 1.3173879377097135,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008386376445612047,
      "loss": 1.337,
      "step": 15312
    },
    {
      "epoch": 1.3174739740170351,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008386177493939125,
      "loss": 1.0527,
      "step": 15313
    },
    {
      "epoch": 1.3175600103243568,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008385978532362293,
      "loss": 1.6298,
      "step": 15314
    },
    {
      "epoch": 1.3176460466316786,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008385779560882132,
      "loss": 1.6119,
      "step": 15315
    },
    {
      "epoch": 1.3177320829390002,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008385580579499223,
      "loss": 1.5249,
      "step": 15316
    },
    {
      "epoch": 1.317818119246322,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008385381588214151,
      "loss": 1.108,
      "step": 15317
    },
    {
      "epoch": 1.3179041555536437,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008385182587027497,
      "loss": 1.9181,
      "step": 15318
    },
    {
      "epoch": 1.3179901918609653,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008384983575939841,
      "loss": 1.3762,
      "step": 15319
    },
    {
      "epoch": 1.318076228168287,
      "grad_norm": 1.25,
      "learning_rate": 0.0008384784554951767,
      "loss": 1.4391,
      "step": 15320
    },
    {
      "epoch": 1.3181622644756086,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0008384585524063854,
      "loss": 0.817,
      "step": 15321
    },
    {
      "epoch": 1.3182483007829304,
      "grad_norm": 1.109375,
      "learning_rate": 0.000838438648327669,
      "loss": 1.0872,
      "step": 15322
    },
    {
      "epoch": 1.318334337090252,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008384187432590853,
      "loss": 1.3659,
      "step": 15323
    },
    {
      "epoch": 1.318420373397574,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008383988372006924,
      "loss": 1.1722,
      "step": 15324
    },
    {
      "epoch": 1.3185064097048955,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008383789301525488,
      "loss": 1.4264,
      "step": 15325
    },
    {
      "epoch": 1.3185924460122171,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008383590221147128,
      "loss": 1.7858,
      "step": 15326
    },
    {
      "epoch": 1.3186784823195388,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008383391130872424,
      "loss": 1.9944,
      "step": 15327
    },
    {
      "epoch": 1.3187645186268606,
      "grad_norm": 1.875,
      "learning_rate": 0.0008383192030701958,
      "loss": 0.9465,
      "step": 15328
    },
    {
      "epoch": 1.3188505549341822,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008382992920636314,
      "loss": 1.2797,
      "step": 15329
    },
    {
      "epoch": 1.3189365912415039,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008382793800676074,
      "loss": 1.039,
      "step": 15330
    },
    {
      "epoch": 1.3190226275488257,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008382594670821819,
      "loss": 0.8032,
      "step": 15331
    },
    {
      "epoch": 1.3191086638561473,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008382395531074133,
      "loss": 1.32,
      "step": 15332
    },
    {
      "epoch": 1.319194700163469,
      "grad_norm": 1.828125,
      "learning_rate": 0.0008382196381433597,
      "loss": 2.5814,
      "step": 15333
    },
    {
      "epoch": 1.3192807364707906,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008381997221900795,
      "loss": 1.4443,
      "step": 15334
    },
    {
      "epoch": 1.3193667727781124,
      "grad_norm": 1.5625,
      "learning_rate": 0.000838179805247631,
      "loss": 1.8699,
      "step": 15335
    },
    {
      "epoch": 1.319452809085434,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008381598873160723,
      "loss": 0.9468,
      "step": 15336
    },
    {
      "epoch": 1.3195388453927557,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008381399683954616,
      "loss": 0.9552,
      "step": 15337
    },
    {
      "epoch": 1.3196248817000775,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008381200484858573,
      "loss": 1.0188,
      "step": 15338
    },
    {
      "epoch": 1.3197109180073991,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008381001275873177,
      "loss": 0.9377,
      "step": 15339
    },
    {
      "epoch": 1.3197969543147208,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008380802056999009,
      "loss": 1.6202,
      "step": 15340
    },
    {
      "epoch": 1.3198829906220424,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008380602828236652,
      "loss": 1.2163,
      "step": 15341
    },
    {
      "epoch": 1.3199690269293642,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008380403589586691,
      "loss": 0.7619,
      "step": 15342
    },
    {
      "epoch": 1.3200550632366859,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008380204341049705,
      "loss": 0.851,
      "step": 15343
    },
    {
      "epoch": 1.3201410995440075,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000838000508262628,
      "loss": 1.6299,
      "step": 15344
    },
    {
      "epoch": 1.3202271358513293,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008379805814316997,
      "loss": 1.1907,
      "step": 15345
    },
    {
      "epoch": 1.320313172158651,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000837960653612244,
      "loss": 1.4991,
      "step": 15346
    },
    {
      "epoch": 1.3203992084659726,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008379407248043189,
      "loss": 1.6898,
      "step": 15347
    },
    {
      "epoch": 1.3204852447732942,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008379207950079831,
      "loss": 1.4984,
      "step": 15348
    },
    {
      "epoch": 1.320571281080616,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008379008642232949,
      "loss": 0.6504,
      "step": 15349
    },
    {
      "epoch": 1.3206573173879377,
      "grad_norm": 1.46875,
      "learning_rate": 0.000837880932450312,
      "loss": 1.3604,
      "step": 15350
    },
    {
      "epoch": 1.3207433536952595,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008378609996890933,
      "loss": 1.119,
      "step": 15351
    },
    {
      "epoch": 1.3208293900025811,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008378410659396967,
      "loss": 1.878,
      "step": 15352
    },
    {
      "epoch": 1.3209154263099028,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008378211312021808,
      "loss": 1.641,
      "step": 15353
    },
    {
      "epoch": 1.3210014626172244,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008378011954766039,
      "loss": 1.8228,
      "step": 15354
    },
    {
      "epoch": 1.321087498924546,
      "grad_norm": 1.40625,
      "learning_rate": 0.000837781258763024,
      "loss": 1.1269,
      "step": 15355
    },
    {
      "epoch": 1.3211735352318679,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008377613210614997,
      "loss": 2.3471,
      "step": 15356
    },
    {
      "epoch": 1.3212595715391895,
      "grad_norm": 2.546875,
      "learning_rate": 0.0008377413823720893,
      "loss": 1.5466,
      "step": 15357
    },
    {
      "epoch": 1.3213456078465113,
      "grad_norm": 0.859375,
      "learning_rate": 0.000837721442694851,
      "loss": 0.8456,
      "step": 15358
    },
    {
      "epoch": 1.321431644153833,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008377015020298432,
      "loss": 1.748,
      "step": 15359
    },
    {
      "epoch": 1.3215176804611546,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008376815603771242,
      "loss": 0.9769,
      "step": 15360
    },
    {
      "epoch": 1.3216037167684762,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008376616177367522,
      "loss": 0.8752,
      "step": 15361
    },
    {
      "epoch": 1.321689753075798,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008376416741087857,
      "loss": 1.6918,
      "step": 15362
    },
    {
      "epoch": 1.3217757893831197,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008376217294932828,
      "loss": 1.2596,
      "step": 15363
    },
    {
      "epoch": 1.3218618256904413,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008376017838903023,
      "loss": 1.6036,
      "step": 15364
    },
    {
      "epoch": 1.3219478619977632,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008375818372999022,
      "loss": 1.3276,
      "step": 15365
    },
    {
      "epoch": 1.3220338983050848,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008375618897221406,
      "loss": 1.104,
      "step": 15366
    },
    {
      "epoch": 1.3221199346124064,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008375419411570763,
      "loss": 1.0333,
      "step": 15367
    },
    {
      "epoch": 1.322205970919728,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008375219916047676,
      "loss": 1.4135,
      "step": 15368
    },
    {
      "epoch": 1.3222920072270499,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008375020410652727,
      "loss": 1.6621,
      "step": 15369
    },
    {
      "epoch": 1.3223780435343715,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008374820895386497,
      "loss": 1.9213,
      "step": 15370
    },
    {
      "epoch": 1.3224640798416931,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008374621370249574,
      "loss": 1.0596,
      "step": 15371
    },
    {
      "epoch": 1.322550116149015,
      "grad_norm": 1.046875,
      "learning_rate": 0.000837442183524254,
      "loss": 0.7145,
      "step": 15372
    },
    {
      "epoch": 1.3226361524563366,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008374222290365978,
      "loss": 1.873,
      "step": 15373
    },
    {
      "epoch": 1.3227221887636582,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008374022735620473,
      "loss": 1.885,
      "step": 15374
    },
    {
      "epoch": 1.3228082250709798,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008373823171006607,
      "loss": 1.3892,
      "step": 15375
    },
    {
      "epoch": 1.3228942613783017,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008373623596524964,
      "loss": 1.3413,
      "step": 15376
    },
    {
      "epoch": 1.3229802976856233,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000837342401217613,
      "loss": 1.1248,
      "step": 15377
    },
    {
      "epoch": 1.3230663339929452,
      "grad_norm": 1.125,
      "learning_rate": 0.0008373224417960685,
      "loss": 0.9436,
      "step": 15378
    },
    {
      "epoch": 1.3231523703002668,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008373024813879214,
      "loss": 0.858,
      "step": 15379
    },
    {
      "epoch": 1.3232384066075884,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008372825199932304,
      "loss": 1.7342,
      "step": 15380
    },
    {
      "epoch": 1.32332444291491,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008372625576120534,
      "loss": 1.1279,
      "step": 15381
    },
    {
      "epoch": 1.3234104792222317,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008372425942444491,
      "loss": 1.2917,
      "step": 15382
    },
    {
      "epoch": 1.3234965155295535,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008372226298904757,
      "loss": 0.9485,
      "step": 15383
    },
    {
      "epoch": 1.3235825518368751,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008372026645501918,
      "loss": 1.8192,
      "step": 15384
    },
    {
      "epoch": 1.323668588144197,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008371826982236557,
      "loss": 0.8886,
      "step": 15385
    },
    {
      "epoch": 1.3237546244515186,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008371627309109258,
      "loss": 1.9017,
      "step": 15386
    },
    {
      "epoch": 1.3238406607588402,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008371427626120604,
      "loss": 1.6136,
      "step": 15387
    },
    {
      "epoch": 1.3239266970661618,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000837122793327118,
      "loss": 1.2879,
      "step": 15388
    },
    {
      "epoch": 1.3240127333734837,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000837102823056157,
      "loss": 1.919,
      "step": 15389
    },
    {
      "epoch": 1.3240987696808053,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008370828517992358,
      "loss": 1.5131,
      "step": 15390
    },
    {
      "epoch": 1.324184805988127,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008370628795564129,
      "loss": 1.236,
      "step": 15391
    },
    {
      "epoch": 1.3242708422954488,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008370429063277465,
      "loss": 0.748,
      "step": 15392
    },
    {
      "epoch": 1.3243568786027704,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008370229321132954,
      "loss": 1.0883,
      "step": 15393
    },
    {
      "epoch": 1.324442914910092,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008370029569131174,
      "loss": 0.6688,
      "step": 15394
    },
    {
      "epoch": 1.3245289512174137,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008369829807272715,
      "loss": 1.3332,
      "step": 15395
    },
    {
      "epoch": 1.3246149875247355,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000836963003555816,
      "loss": 1.4045,
      "step": 15396
    },
    {
      "epoch": 1.3247010238320571,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008369430253988093,
      "loss": 0.9842,
      "step": 15397
    },
    {
      "epoch": 1.3247870601393787,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008369230462563096,
      "loss": 0.9359,
      "step": 15398
    },
    {
      "epoch": 1.3248730964467006,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008369030661283755,
      "loss": 2.1011,
      "step": 15399
    },
    {
      "epoch": 1.3249591327540222,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008368830850150656,
      "loss": 0.8903,
      "step": 15400
    },
    {
      "epoch": 1.3250451690613438,
      "grad_norm": 1.09375,
      "learning_rate": 0.000836863102916438,
      "loss": 1.2085,
      "step": 15401
    },
    {
      "epoch": 1.3251312053686655,
      "grad_norm": 2.0,
      "learning_rate": 0.0008368431198325515,
      "loss": 1.6895,
      "step": 15402
    },
    {
      "epoch": 1.3252172416759873,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008368231357634644,
      "loss": 1.1853,
      "step": 15403
    },
    {
      "epoch": 1.325303277983309,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000836803150709235,
      "loss": 1.554,
      "step": 15404
    },
    {
      "epoch": 1.3253893142906308,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008367831646699221,
      "loss": 0.7883,
      "step": 15405
    },
    {
      "epoch": 1.3254753505979524,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008367631776455837,
      "loss": 1.0231,
      "step": 15406
    },
    {
      "epoch": 1.325561386905274,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008367431896362787,
      "loss": 1.2869,
      "step": 15407
    },
    {
      "epoch": 1.3256474232125957,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008367232006420653,
      "loss": 0.6747,
      "step": 15408
    },
    {
      "epoch": 1.3257334595199173,
      "grad_norm": 1.390625,
      "learning_rate": 0.000836703210663002,
      "loss": 1.2127,
      "step": 15409
    },
    {
      "epoch": 1.3258194958272391,
      "grad_norm": 1.90625,
      "learning_rate": 0.0008366832196991472,
      "loss": 1.9172,
      "step": 15410
    },
    {
      "epoch": 1.3259055321345607,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008366632277505597,
      "loss": 1.3331,
      "step": 15411
    },
    {
      "epoch": 1.3259915684418826,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008366432348172975,
      "loss": 1.8523,
      "step": 15412
    },
    {
      "epoch": 1.3260776047492042,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008366232408994194,
      "loss": 1.3329,
      "step": 15413
    },
    {
      "epoch": 1.3261636410565258,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008366032459969839,
      "loss": 1.3188,
      "step": 15414
    },
    {
      "epoch": 1.3262496773638475,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008365832501100492,
      "loss": 1.6676,
      "step": 15415
    },
    {
      "epoch": 1.3263357136711693,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008365632532386741,
      "loss": 1.6411,
      "step": 15416
    },
    {
      "epoch": 1.326421749978491,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008365432553829169,
      "loss": 1.0089,
      "step": 15417
    },
    {
      "epoch": 1.3265077862858126,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008365232565428362,
      "loss": 1.7693,
      "step": 15418
    },
    {
      "epoch": 1.3265938225931344,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008365032567184903,
      "loss": 1.2485,
      "step": 15419
    },
    {
      "epoch": 1.326679858900456,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008364832559099378,
      "loss": 1.452,
      "step": 15420
    },
    {
      "epoch": 1.3267658952077777,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008364632541172372,
      "loss": 0.9966,
      "step": 15421
    },
    {
      "epoch": 1.3268519315150993,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008364432513404472,
      "loss": 0.8416,
      "step": 15422
    },
    {
      "epoch": 1.3269379678224211,
      "grad_norm": 1.078125,
      "learning_rate": 0.000836423247579626,
      "loss": 1.7185,
      "step": 15423
    },
    {
      "epoch": 1.3270240041297428,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008364032428348323,
      "loss": 0.8059,
      "step": 15424
    },
    {
      "epoch": 1.3271100404370644,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008363832371061245,
      "loss": 1.1508,
      "step": 15425
    },
    {
      "epoch": 1.3271960767443862,
      "grad_norm": 1.0,
      "learning_rate": 0.000836363230393561,
      "loss": 0.8592,
      "step": 15426
    },
    {
      "epoch": 1.3272821130517078,
      "grad_norm": 1.0,
      "learning_rate": 0.0008363432226972005,
      "loss": 1.4429,
      "step": 15427
    },
    {
      "epoch": 1.3273681493590295,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008363232140171017,
      "loss": 1.3062,
      "step": 15428
    },
    {
      "epoch": 1.327454185666351,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008363032043533227,
      "loss": 0.9263,
      "step": 15429
    },
    {
      "epoch": 1.327540221973673,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008362831937059226,
      "loss": 0.9248,
      "step": 15430
    },
    {
      "epoch": 1.3276262582809946,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008362631820749591,
      "loss": 1.656,
      "step": 15431
    },
    {
      "epoch": 1.3277122945883162,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008362431694604915,
      "loss": 1.1865,
      "step": 15432
    },
    {
      "epoch": 1.327798330895638,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008362231558625779,
      "loss": 1.6003,
      "step": 15433
    },
    {
      "epoch": 1.3278843672029597,
      "grad_norm": 1.0,
      "learning_rate": 0.000836203141281277,
      "loss": 1.5647,
      "step": 15434
    },
    {
      "epoch": 1.3279704035102813,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008361831257166473,
      "loss": 1.0576,
      "step": 15435
    },
    {
      "epoch": 1.328056439817603,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008361631091687474,
      "loss": 1.2892,
      "step": 15436
    },
    {
      "epoch": 1.3281424761249248,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008361430916376357,
      "loss": 1.0794,
      "step": 15437
    },
    {
      "epoch": 1.3282285124322464,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008361230731233708,
      "loss": 1.5755,
      "step": 15438
    },
    {
      "epoch": 1.3283145487395682,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008361030536260113,
      "loss": 0.9725,
      "step": 15439
    },
    {
      "epoch": 1.3284005850468898,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008360830331456158,
      "loss": 1.2987,
      "step": 15440
    },
    {
      "epoch": 1.3284866213542115,
      "grad_norm": 1.125,
      "learning_rate": 0.0008360630116822428,
      "loss": 1.9749,
      "step": 15441
    },
    {
      "epoch": 1.328572657661533,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008360429892359508,
      "loss": 1.488,
      "step": 15442
    },
    {
      "epoch": 1.3286586939688547,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008360229658067984,
      "loss": 1.2889,
      "step": 15443
    },
    {
      "epoch": 1.3287447302761766,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008360029413948441,
      "loss": 0.9854,
      "step": 15444
    },
    {
      "epoch": 1.3288307665834982,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008359829160001466,
      "loss": 1.5683,
      "step": 15445
    },
    {
      "epoch": 1.32891680289082,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008359628896227646,
      "loss": 1.6696,
      "step": 15446
    },
    {
      "epoch": 1.3290028391981417,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008359428622627563,
      "loss": 1.2721,
      "step": 15447
    },
    {
      "epoch": 1.3290888755054633,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008359228339201805,
      "loss": 0.9891,
      "step": 15448
    },
    {
      "epoch": 1.329174911812785,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008359028045950959,
      "loss": 1.361,
      "step": 15449
    },
    {
      "epoch": 1.3292609481201068,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008358827742875607,
      "loss": 2.1469,
      "step": 15450
    },
    {
      "epoch": 1.3293469844274284,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008358627429976337,
      "loss": 0.9726,
      "step": 15451
    },
    {
      "epoch": 1.32943302073475,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008358427107253737,
      "loss": 1.271,
      "step": 15452
    },
    {
      "epoch": 1.3295190570420719,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008358226774708389,
      "loss": 0.8984,
      "step": 15453
    },
    {
      "epoch": 1.3296050933493935,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008358026432340882,
      "loss": 1.5598,
      "step": 15454
    },
    {
      "epoch": 1.329691129656715,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008357826080151801,
      "loss": 1.7291,
      "step": 15455
    },
    {
      "epoch": 1.3297771659640367,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008357625718141731,
      "loss": 1.8634,
      "step": 15456
    },
    {
      "epoch": 1.3298632022713586,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008357425346311259,
      "loss": 1.994,
      "step": 15457
    },
    {
      "epoch": 1.3299492385786802,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008357224964660972,
      "loss": 1.5082,
      "step": 15458
    },
    {
      "epoch": 1.3300352748860018,
      "grad_norm": 1.625,
      "learning_rate": 0.0008357024573191452,
      "loss": 1.4668,
      "step": 15459
    },
    {
      "epoch": 1.3301213111933237,
      "grad_norm": 1.46875,
      "learning_rate": 0.000835682417190329,
      "loss": 2.2955,
      "step": 15460
    },
    {
      "epoch": 1.3302073475006453,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008356623760797071,
      "loss": 1.7958,
      "step": 15461
    },
    {
      "epoch": 1.330293383807967,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008356423339873379,
      "loss": 1.064,
      "step": 15462
    },
    {
      "epoch": 1.3303794201152885,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008356222909132803,
      "loss": 1.4645,
      "step": 15463
    },
    {
      "epoch": 1.3304654564226104,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008356022468575926,
      "loss": 1.0867,
      "step": 15464
    },
    {
      "epoch": 1.330551492729932,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008355822018203336,
      "loss": 1.1583,
      "step": 15465
    },
    {
      "epoch": 1.3306375290372539,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008355621558015621,
      "loss": 1.1995,
      "step": 15466
    },
    {
      "epoch": 1.3307235653445755,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008355421088013363,
      "loss": 0.7769,
      "step": 15467
    },
    {
      "epoch": 1.330809601651897,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008355220608197152,
      "loss": 1.9158,
      "step": 15468
    },
    {
      "epoch": 1.3308956379592187,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008355020118567573,
      "loss": 1.2808,
      "step": 15469
    },
    {
      "epoch": 1.3309816742665403,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008354819619125213,
      "loss": 2.051,
      "step": 15470
    },
    {
      "epoch": 1.3310677105738622,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008354619109870661,
      "loss": 2.1808,
      "step": 15471
    },
    {
      "epoch": 1.3311537468811838,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008354418590804497,
      "loss": 1.553,
      "step": 15472
    },
    {
      "epoch": 1.3312397831885057,
      "grad_norm": 1.375,
      "learning_rate": 0.0008354218061927312,
      "loss": 1.5074,
      "step": 15473
    },
    {
      "epoch": 1.3313258194958273,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008354017523239691,
      "loss": 1.7625,
      "step": 15474
    },
    {
      "epoch": 1.331411855803149,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008353816974742221,
      "loss": 1.5258,
      "step": 15475
    },
    {
      "epoch": 1.3314978921104705,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000835361641643549,
      "loss": 1.9111,
      "step": 15476
    },
    {
      "epoch": 1.3315839284177924,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008353415848320083,
      "loss": 1.3414,
      "step": 15477
    },
    {
      "epoch": 1.331669964725114,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008353215270396585,
      "loss": 0.9666,
      "step": 15478
    },
    {
      "epoch": 1.3317560010324356,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008353014682665587,
      "loss": 1.0144,
      "step": 15479
    },
    {
      "epoch": 1.3318420373397575,
      "grad_norm": 2.78125,
      "learning_rate": 0.0008352814085127671,
      "loss": 1.7214,
      "step": 15480
    },
    {
      "epoch": 1.331928073647079,
      "grad_norm": 1.5,
      "learning_rate": 0.0008352613477783426,
      "loss": 1.5613,
      "step": 15481
    },
    {
      "epoch": 1.3320141099544007,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000835241286063344,
      "loss": 0.9524,
      "step": 15482
    },
    {
      "epoch": 1.3321001462617224,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008352212233678298,
      "loss": 1.6066,
      "step": 15483
    },
    {
      "epoch": 1.3321861825690442,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008352011596918587,
      "loss": 1.3946,
      "step": 15484
    },
    {
      "epoch": 1.3322722188763658,
      "grad_norm": 1.5,
      "learning_rate": 0.0008351810950354893,
      "loss": 1.5775,
      "step": 15485
    },
    {
      "epoch": 1.3323582551836874,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008351610293987806,
      "loss": 1.1133,
      "step": 15486
    },
    {
      "epoch": 1.3324442914910093,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000835140962781791,
      "loss": 1.2326,
      "step": 15487
    },
    {
      "epoch": 1.332530327798331,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008351208951845792,
      "loss": 1.6967,
      "step": 15488
    },
    {
      "epoch": 1.3326163641056525,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000835100826607204,
      "loss": 2.0513,
      "step": 15489
    },
    {
      "epoch": 1.3327024004129742,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000835080757049724,
      "loss": 1.2797,
      "step": 15490
    },
    {
      "epoch": 1.332788436720296,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008350606865121979,
      "loss": 1.3763,
      "step": 15491
    },
    {
      "epoch": 1.3328744730276176,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008350406149946845,
      "loss": 1.5507,
      "step": 15492
    },
    {
      "epoch": 1.3329605093349395,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008350205424972426,
      "loss": 1.6302,
      "step": 15493
    },
    {
      "epoch": 1.333046545642261,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008350004690199307,
      "loss": 1.1181,
      "step": 15494
    },
    {
      "epoch": 1.3331325819495827,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008349803945628076,
      "loss": 1.28,
      "step": 15495
    },
    {
      "epoch": 1.3332186182569044,
      "grad_norm": 1.28125,
      "learning_rate": 0.000834960319125932,
      "loss": 1.2065,
      "step": 15496
    },
    {
      "epoch": 1.333304654564226,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008349402427093626,
      "loss": 1.3007,
      "step": 15497
    },
    {
      "epoch": 1.3333906908715478,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008349201653131581,
      "loss": 0.5856,
      "step": 15498
    },
    {
      "epoch": 1.3334767271788694,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008349000869373773,
      "loss": 1.1593,
      "step": 15499
    },
    {
      "epoch": 1.3335627634861913,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008348800075820787,
      "loss": 1.8892,
      "step": 15500
    },
    {
      "epoch": 1.333648799793513,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008348599272473214,
      "loss": 0.9757,
      "step": 15501
    },
    {
      "epoch": 1.3337348361008345,
      "grad_norm": 1.25,
      "learning_rate": 0.0008348398459331639,
      "loss": 1.6503,
      "step": 15502
    },
    {
      "epoch": 1.3338208724081562,
      "grad_norm": 1.046875,
      "learning_rate": 0.000834819763639665,
      "loss": 1.0586,
      "step": 15503
    },
    {
      "epoch": 1.333906908715478,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008347996803668834,
      "loss": 0.9678,
      "step": 15504
    },
    {
      "epoch": 1.3339929450227996,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008347795961148777,
      "loss": 1.4974,
      "step": 15505
    },
    {
      "epoch": 1.3340789813301213,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008347595108837069,
      "loss": 1.4843,
      "step": 15506
    },
    {
      "epoch": 1.334165017637443,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008347394246734297,
      "loss": 1.4074,
      "step": 15507
    },
    {
      "epoch": 1.3342510539447647,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008347193374841048,
      "loss": 1.8491,
      "step": 15508
    },
    {
      "epoch": 1.3343370902520864,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008346992493157908,
      "loss": 1.8352,
      "step": 15509
    },
    {
      "epoch": 1.334423126559408,
      "grad_norm": 1.90625,
      "learning_rate": 0.0008346791601685467,
      "loss": 1.1385,
      "step": 15510
    },
    {
      "epoch": 1.3345091628667298,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008346590700424311,
      "loss": 1.6973,
      "step": 15511
    },
    {
      "epoch": 1.3345951991740514,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008346389789375029,
      "loss": 1.1748,
      "step": 15512
    },
    {
      "epoch": 1.334681235481373,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008346188868538207,
      "loss": 1.08,
      "step": 15513
    },
    {
      "epoch": 1.334767271788695,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008345987937914434,
      "loss": 1.7337,
      "step": 15514
    },
    {
      "epoch": 1.3348533080960165,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008345786997504296,
      "loss": 1.0836,
      "step": 15515
    },
    {
      "epoch": 1.3349393444033382,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008345586047308381,
      "loss": 0.897,
      "step": 15516
    },
    {
      "epoch": 1.3350253807106598,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008345385087327279,
      "loss": 1.4219,
      "step": 15517
    },
    {
      "epoch": 1.3351114170179816,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008345184117561575,
      "loss": 0.6693,
      "step": 15518
    },
    {
      "epoch": 1.3351974533253033,
      "grad_norm": 1.375,
      "learning_rate": 0.0008344983138011859,
      "loss": 1.6434,
      "step": 15519
    },
    {
      "epoch": 1.3352834896326249,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008344782148678718,
      "loss": 1.3825,
      "step": 15520
    },
    {
      "epoch": 1.3353695259399467,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000834458114956274,
      "loss": 1.1689,
      "step": 15521
    },
    {
      "epoch": 1.3354555622472684,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008344380140664512,
      "loss": 1.3883,
      "step": 15522
    },
    {
      "epoch": 1.33554159855459,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008344179121984623,
      "loss": 1.3791,
      "step": 15523
    },
    {
      "epoch": 1.3356276348619116,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000834397809352366,
      "loss": 1.45,
      "step": 15524
    },
    {
      "epoch": 1.3357136711692335,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008343777055282211,
      "loss": 1.7567,
      "step": 15525
    },
    {
      "epoch": 1.335799707476555,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008343576007260866,
      "loss": 1.2584,
      "step": 15526
    },
    {
      "epoch": 1.335885743783877,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000834337494946021,
      "loss": 1.6165,
      "step": 15527
    },
    {
      "epoch": 1.3359717800911985,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008343173881880834,
      "loss": 1.208,
      "step": 15528
    },
    {
      "epoch": 1.3360578163985202,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008342972804523324,
      "loss": 1.8893,
      "step": 15529
    },
    {
      "epoch": 1.3361438527058418,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008342771717388266,
      "loss": 1.2916,
      "step": 15530
    },
    {
      "epoch": 1.3362298890131634,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008342570620476254,
      "loss": 0.7735,
      "step": 15531
    },
    {
      "epoch": 1.3363159253204853,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008342369513787873,
      "loss": 1.7431,
      "step": 15532
    },
    {
      "epoch": 1.336401961627807,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000834216839732371,
      "loss": 1.2604,
      "step": 15533
    },
    {
      "epoch": 1.3364879979351287,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008341967271084356,
      "loss": 1.0479,
      "step": 15534
    },
    {
      "epoch": 1.3365740342424504,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008341766135070396,
      "loss": 2.0232,
      "step": 15535
    },
    {
      "epoch": 1.336660070549772,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000834156498928242,
      "loss": 1.7936,
      "step": 15536
    },
    {
      "epoch": 1.3367461068570936,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008341363833721017,
      "loss": 1.6954,
      "step": 15537
    },
    {
      "epoch": 1.3368321431644155,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008341162668386774,
      "loss": 1.0314,
      "step": 15538
    },
    {
      "epoch": 1.336918179471737,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008340961493280281,
      "loss": 1.3609,
      "step": 15539
    },
    {
      "epoch": 1.3370042157790587,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008340760308402125,
      "loss": 1.188,
      "step": 15540
    },
    {
      "epoch": 1.3370902520863805,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008340559113752895,
      "loss": 1.4066,
      "step": 15541
    },
    {
      "epoch": 1.3371762883937022,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000834035790933318,
      "loss": 1.7465,
      "step": 15542
    },
    {
      "epoch": 1.3372623247010238,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008340156695143566,
      "loss": 1.8564,
      "step": 15543
    },
    {
      "epoch": 1.3373483610083454,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008339955471184645,
      "loss": 1.5643,
      "step": 15544
    },
    {
      "epoch": 1.3374343973156673,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008339754237457002,
      "loss": 1.2489,
      "step": 15545
    },
    {
      "epoch": 1.337520433622989,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008339552993961231,
      "loss": 1.4384,
      "step": 15546
    },
    {
      "epoch": 1.3376064699303105,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008339351740697914,
      "loss": 1.5358,
      "step": 15547
    },
    {
      "epoch": 1.3376925062376324,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008339150477667641,
      "loss": 2.2354,
      "step": 15548
    },
    {
      "epoch": 1.337778542544954,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008338949204871005,
      "loss": 1.2671,
      "step": 15549
    },
    {
      "epoch": 1.3378645788522756,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008338747922308591,
      "loss": 0.5351,
      "step": 15550
    },
    {
      "epoch": 1.3379506151595972,
      "grad_norm": 1.015625,
      "learning_rate": 0.000833854662998099,
      "loss": 1.5112,
      "step": 15551
    },
    {
      "epoch": 1.338036651466919,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008338345327888788,
      "loss": 0.7655,
      "step": 15552
    },
    {
      "epoch": 1.3381226877742407,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008338144016032576,
      "loss": 1.9976,
      "step": 15553
    },
    {
      "epoch": 1.3382087240815626,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008337942694412942,
      "loss": 1.4629,
      "step": 15554
    },
    {
      "epoch": 1.3382947603888842,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008337741363030473,
      "loss": 1.307,
      "step": 15555
    },
    {
      "epoch": 1.3383807966962058,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008337540021885761,
      "loss": 1.4729,
      "step": 15556
    },
    {
      "epoch": 1.3384668330035274,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008337338670979394,
      "loss": 1.1625,
      "step": 15557
    },
    {
      "epoch": 1.338552869310849,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000833713731031196,
      "loss": 0.7896,
      "step": 15558
    },
    {
      "epoch": 1.338638905618171,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008336935939884049,
      "loss": 1.4685,
      "step": 15559
    },
    {
      "epoch": 1.3387249419254925,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008336734559696248,
      "loss": 0.749,
      "step": 15560
    },
    {
      "epoch": 1.3388109782328144,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000833653316974915,
      "loss": 0.9333,
      "step": 15561
    },
    {
      "epoch": 1.338897014540136,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008336331770043339,
      "loss": 1.0977,
      "step": 15562
    },
    {
      "epoch": 1.3389830508474576,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008336130360579406,
      "loss": 1.6724,
      "step": 15563
    },
    {
      "epoch": 1.3390690871547792,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008335928941357942,
      "loss": 1.4112,
      "step": 15564
    },
    {
      "epoch": 1.339155123462101,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0008335727512379534,
      "loss": 1.7173,
      "step": 15565
    },
    {
      "epoch": 1.3392411597694227,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008335526073644773,
      "loss": 1.3226,
      "step": 15566
    },
    {
      "epoch": 1.3393271960767443,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008335324625154246,
      "loss": 1.4028,
      "step": 15567
    },
    {
      "epoch": 1.3394132323840662,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008335123166908543,
      "loss": 2.2225,
      "step": 15568
    },
    {
      "epoch": 1.3394992686913878,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008334921698908253,
      "loss": 1.4418,
      "step": 15569
    },
    {
      "epoch": 1.3395853049987094,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008334720221153968,
      "loss": 1.6265,
      "step": 15570
    },
    {
      "epoch": 1.339671341306031,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008334518733646273,
      "loss": 2.1132,
      "step": 15571
    },
    {
      "epoch": 1.339757377613353,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008334317236385757,
      "loss": 1.3568,
      "step": 15572
    },
    {
      "epoch": 1.3398434139206745,
      "grad_norm": 1.375,
      "learning_rate": 0.0008334115729373015,
      "loss": 0.9384,
      "step": 15573
    },
    {
      "epoch": 1.3399294502279961,
      "grad_norm": 2.171875,
      "learning_rate": 0.0008333914212608632,
      "loss": 1.1339,
      "step": 15574
    },
    {
      "epoch": 1.340015486535318,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008333712686093198,
      "loss": 1.5351,
      "step": 15575
    },
    {
      "epoch": 1.3401015228426396,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008333511149827304,
      "loss": 1.2,
      "step": 15576
    },
    {
      "epoch": 1.3401875591499612,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008333309603811536,
      "loss": 1.3847,
      "step": 15577
    },
    {
      "epoch": 1.3402735954572829,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008333108048046487,
      "loss": 1.0299,
      "step": 15578
    },
    {
      "epoch": 1.3403596317646047,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008332906482532745,
      "loss": 1.0798,
      "step": 15579
    },
    {
      "epoch": 1.3404456680719263,
      "grad_norm": 1.375,
      "learning_rate": 0.00083327049072709,
      "loss": 1.7055,
      "step": 15580
    },
    {
      "epoch": 1.3405317043792482,
      "grad_norm": 0.84375,
      "learning_rate": 0.0008332503322261543,
      "loss": 0.6143,
      "step": 15581
    },
    {
      "epoch": 1.3406177406865698,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008332301727505258,
      "loss": 0.9802,
      "step": 15582
    },
    {
      "epoch": 1.3407037769938914,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008332100123002641,
      "loss": 1.3489,
      "step": 15583
    },
    {
      "epoch": 1.340789813301213,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008331898508754279,
      "loss": 1.3239,
      "step": 15584
    },
    {
      "epoch": 1.3408758496085347,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008331696884760762,
      "loss": 0.9446,
      "step": 15585
    },
    {
      "epoch": 1.3409618859158565,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008331495251022679,
      "loss": 0.8145,
      "step": 15586
    },
    {
      "epoch": 1.3410479222231781,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008331293607540622,
      "loss": 0.9219,
      "step": 15587
    },
    {
      "epoch": 1.3411339585305,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008331091954315177,
      "loss": 1.277,
      "step": 15588
    },
    {
      "epoch": 1.3412199948378216,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008330890291346936,
      "loss": 0.8032,
      "step": 15589
    },
    {
      "epoch": 1.3413060311451432,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008330688618636491,
      "loss": 1.8184,
      "step": 15590
    },
    {
      "epoch": 1.3413920674524649,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008330486936184427,
      "loss": 0.9952,
      "step": 15591
    },
    {
      "epoch": 1.3414781037597867,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008330285243991337,
      "loss": 1.0592,
      "step": 15592
    },
    {
      "epoch": 1.3415641400671083,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008330083542057811,
      "loss": 1.4846,
      "step": 15593
    },
    {
      "epoch": 1.34165017637443,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008329881830384438,
      "loss": 1.9386,
      "step": 15594
    },
    {
      "epoch": 1.3417362126817518,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008329680108971807,
      "loss": 1.1853,
      "step": 15595
    },
    {
      "epoch": 1.3418222489890734,
      "grad_norm": 116.5,
      "learning_rate": 0.0008329478377820512,
      "loss": 1.335,
      "step": 15596
    },
    {
      "epoch": 1.341908285296395,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008329276636931138,
      "loss": 1.4027,
      "step": 15597
    },
    {
      "epoch": 1.3419943216037167,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000832907488630428,
      "loss": 1.1904,
      "step": 15598
    },
    {
      "epoch": 1.3420803579110385,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008328873125940522,
      "loss": 0.9477,
      "step": 15599
    },
    {
      "epoch": 1.3421663942183601,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000832867135584046,
      "loss": 0.9408,
      "step": 15600
    },
    {
      "epoch": 1.3422524305256818,
      "grad_norm": 1.265625,
      "learning_rate": 0.000832846957600468,
      "loss": 1.0644,
      "step": 15601
    },
    {
      "epoch": 1.3423384668330036,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008328267786433776,
      "loss": 1.22,
      "step": 15602
    },
    {
      "epoch": 1.3424245031403252,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008328065987128335,
      "loss": 1.2685,
      "step": 15603
    },
    {
      "epoch": 1.3425105394476469,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008327864178088948,
      "loss": 1.6128,
      "step": 15604
    },
    {
      "epoch": 1.3425965757549685,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008327662359316205,
      "loss": 1.5358,
      "step": 15605
    },
    {
      "epoch": 1.3426826120622903,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008327460530810698,
      "loss": 1.3638,
      "step": 15606
    },
    {
      "epoch": 1.342768648369612,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008327258692573014,
      "loss": 0.7805,
      "step": 15607
    },
    {
      "epoch": 1.3428546846769336,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008327056844603748,
      "loss": 1.6498,
      "step": 15608
    },
    {
      "epoch": 1.3429407209842554,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008326854986903487,
      "loss": 2.0768,
      "step": 15609
    },
    {
      "epoch": 1.343026757291577,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000832665311947282,
      "loss": 1.3127,
      "step": 15610
    },
    {
      "epoch": 1.3431127935988987,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008326451242312343,
      "loss": 0.7076,
      "step": 15611
    },
    {
      "epoch": 1.3431988299062203,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008326249355422641,
      "loss": 1.7045,
      "step": 15612
    },
    {
      "epoch": 1.3432848662135422,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008326047458804307,
      "loss": 1.251,
      "step": 15613
    },
    {
      "epoch": 1.3433709025208638,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008325845552457932,
      "loss": 1.4593,
      "step": 15614
    },
    {
      "epoch": 1.3434569388281856,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008325643636384103,
      "loss": 1.5196,
      "step": 15615
    },
    {
      "epoch": 1.3435429751355072,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008325441710583417,
      "loss": 0.8335,
      "step": 15616
    },
    {
      "epoch": 1.3436290114428289,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008325239775056458,
      "loss": 1.9448,
      "step": 15617
    },
    {
      "epoch": 1.3437150477501505,
      "grad_norm": 1.234375,
      "learning_rate": 0.000832503782980382,
      "loss": 1.0673,
      "step": 15618
    },
    {
      "epoch": 1.3438010840574721,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008324835874826093,
      "loss": 1.2052,
      "step": 15619
    },
    {
      "epoch": 1.343887120364794,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008324633910123868,
      "loss": 1.004,
      "step": 15620
    },
    {
      "epoch": 1.3439731566721156,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008324431935697735,
      "loss": 1.4837,
      "step": 15621
    },
    {
      "epoch": 1.3440591929794374,
      "grad_norm": 3.140625,
      "learning_rate": 0.0008324229951548285,
      "loss": 1.1369,
      "step": 15622
    },
    {
      "epoch": 1.344145229286759,
      "grad_norm": 1.046875,
      "learning_rate": 0.000832402795767611,
      "loss": 0.7871,
      "step": 15623
    },
    {
      "epoch": 1.3442312655940807,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00083238259540818,
      "loss": 2.0181,
      "step": 15624
    },
    {
      "epoch": 1.3443173019014023,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008323623940765943,
      "loss": 1.8414,
      "step": 15625
    },
    {
      "epoch": 1.3444033382087242,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008323421917729134,
      "loss": 1.0692,
      "step": 15626
    },
    {
      "epoch": 1.3444893745160458,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008323219884971963,
      "loss": 1.7285,
      "step": 15627
    },
    {
      "epoch": 1.3445754108233674,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000832301784249502,
      "loss": 1.7246,
      "step": 15628
    },
    {
      "epoch": 1.3446614471306892,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008322815790298894,
      "loss": 1.9795,
      "step": 15629
    },
    {
      "epoch": 1.3447474834380109,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008322613728384181,
      "loss": 1.3319,
      "step": 15630
    },
    {
      "epoch": 1.3448335197453325,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008322411656751466,
      "loss": 1.6559,
      "step": 15631
    },
    {
      "epoch": 1.3449195560526541,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008322209575401345,
      "loss": 1.8834,
      "step": 15632
    },
    {
      "epoch": 1.345005592359976,
      "grad_norm": 2.109375,
      "learning_rate": 0.0008322007484334407,
      "loss": 1.6854,
      "step": 15633
    },
    {
      "epoch": 1.3450916286672976,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008321805383551244,
      "loss": 1.291,
      "step": 15634
    },
    {
      "epoch": 1.3451776649746192,
      "grad_norm": 10.3125,
      "learning_rate": 0.0008321603273052445,
      "loss": 0.8486,
      "step": 15635
    },
    {
      "epoch": 1.345263701281941,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008321401152838602,
      "loss": 1.1109,
      "step": 15636
    },
    {
      "epoch": 1.3453497375892627,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008321199022910308,
      "loss": 1.0036,
      "step": 15637
    },
    {
      "epoch": 1.3454357738965843,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008320996883268152,
      "loss": 1.1939,
      "step": 15638
    },
    {
      "epoch": 1.345521810203906,
      "grad_norm": 2.875,
      "learning_rate": 0.0008320794733912727,
      "loss": 1.6754,
      "step": 15639
    },
    {
      "epoch": 1.3456078465112278,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008320592574844621,
      "loss": 2.2684,
      "step": 15640
    },
    {
      "epoch": 1.3456938828185494,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008320390406064429,
      "loss": 1.163,
      "step": 15641
    },
    {
      "epoch": 1.3457799191258712,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008320188227572741,
      "loss": 1.3951,
      "step": 15642
    },
    {
      "epoch": 1.3458659554331929,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008319986039370149,
      "loss": 1.2054,
      "step": 15643
    },
    {
      "epoch": 1.3459519917405145,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008319783841457243,
      "loss": 1.0627,
      "step": 15644
    },
    {
      "epoch": 1.3460380280478361,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008319581633834614,
      "loss": 0.8232,
      "step": 15645
    },
    {
      "epoch": 1.3461240643551577,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008319379416502855,
      "loss": 1.2521,
      "step": 15646
    },
    {
      "epoch": 1.3462101006624796,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008319177189462556,
      "loss": 1.7644,
      "step": 15647
    },
    {
      "epoch": 1.3462961369698012,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008318974952714309,
      "loss": 1.1349,
      "step": 15648
    },
    {
      "epoch": 1.346382173277123,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008318772706258707,
      "loss": 1.5321,
      "step": 15649
    },
    {
      "epoch": 1.3464682095844447,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000831857045009634,
      "loss": 1.0359,
      "step": 15650
    },
    {
      "epoch": 1.3465542458917663,
      "grad_norm": 1.375,
      "learning_rate": 0.00083183681842278,
      "loss": 1.5005,
      "step": 15651
    },
    {
      "epoch": 1.346640282199088,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008318165908653679,
      "loss": 1.8799,
      "step": 15652
    },
    {
      "epoch": 1.3467263185064098,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008317963623374567,
      "loss": 0.937,
      "step": 15653
    },
    {
      "epoch": 1.3468123548137314,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008317761328391057,
      "loss": 1.3135,
      "step": 15654
    },
    {
      "epoch": 1.346898391121053,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008317559023703741,
      "loss": 1.0469,
      "step": 15655
    },
    {
      "epoch": 1.3469844274283749,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008317356709313208,
      "loss": 1.4407,
      "step": 15656
    },
    {
      "epoch": 1.3470704637356965,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008317154385220054,
      "loss": 1.6142,
      "step": 15657
    },
    {
      "epoch": 1.3471565000430181,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008316952051424869,
      "loss": 1.4585,
      "step": 15658
    },
    {
      "epoch": 1.3472425363503397,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008316749707928242,
      "loss": 1.1639,
      "step": 15659
    },
    {
      "epoch": 1.3473285726576616,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008316547354730768,
      "loss": 0.8516,
      "step": 15660
    },
    {
      "epoch": 1.3474146089649832,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008316344991833039,
      "loss": 1.2214,
      "step": 15661
    },
    {
      "epoch": 1.3475006452723048,
      "grad_norm": 2.046875,
      "learning_rate": 0.0008316142619235645,
      "loss": 1.8853,
      "step": 15662
    },
    {
      "epoch": 1.3475866815796267,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008315940236939179,
      "loss": 1.1072,
      "step": 15663
    },
    {
      "epoch": 1.3476727178869483,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008315737844944234,
      "loss": 0.9227,
      "step": 15664
    },
    {
      "epoch": 1.34775875419427,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008315535443251398,
      "loss": 1.7178,
      "step": 15665
    },
    {
      "epoch": 1.3478447905015916,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008315333031861268,
      "loss": 1.1804,
      "step": 15666
    },
    {
      "epoch": 1.3479308268089134,
      "grad_norm": 1.0,
      "learning_rate": 0.0008315130610774433,
      "loss": 1.1969,
      "step": 15667
    },
    {
      "epoch": 1.348016863116235,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008314928179991485,
      "loss": 1.0645,
      "step": 15668
    },
    {
      "epoch": 1.3481028994235569,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008314725739513018,
      "loss": 1.856,
      "step": 15669
    },
    {
      "epoch": 1.3481889357308785,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008314523289339622,
      "loss": 1.9728,
      "step": 15670
    },
    {
      "epoch": 1.3482749720382001,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000831432082947189,
      "loss": 1.3818,
      "step": 15671
    },
    {
      "epoch": 1.3483610083455217,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008314118359910415,
      "loss": 1.8937,
      "step": 15672
    },
    {
      "epoch": 1.3484470446528434,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008313915880655787,
      "loss": 1.4637,
      "step": 15673
    },
    {
      "epoch": 1.3485330809601652,
      "grad_norm": 1.125,
      "learning_rate": 0.00083137133917086,
      "loss": 1.7163,
      "step": 15674
    },
    {
      "epoch": 1.3486191172674868,
      "grad_norm": 1.125,
      "learning_rate": 0.0008313510893069447,
      "loss": 1.5401,
      "step": 15675
    },
    {
      "epoch": 1.3487051535748087,
      "grad_norm": 1.625,
      "learning_rate": 0.0008313308384738918,
      "loss": 1.9252,
      "step": 15676
    },
    {
      "epoch": 1.3487911898821303,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008313105866717607,
      "loss": 1.0367,
      "step": 15677
    },
    {
      "epoch": 1.348877226189452,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008312903339006105,
      "loss": 1.4773,
      "step": 15678
    },
    {
      "epoch": 1.3489632624967736,
      "grad_norm": 1.125,
      "learning_rate": 0.0008312700801605006,
      "loss": 1.3507,
      "step": 15679
    },
    {
      "epoch": 1.3490492988040954,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008312498254514901,
      "loss": 1.4558,
      "step": 15680
    },
    {
      "epoch": 1.349135335111417,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008312295697736383,
      "loss": 1.8707,
      "step": 15681
    },
    {
      "epoch": 1.3492213714187387,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0008312093131270044,
      "loss": 1.4861,
      "step": 15682
    },
    {
      "epoch": 1.3493074077260605,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008311890555116477,
      "loss": 1.6465,
      "step": 15683
    },
    {
      "epoch": 1.3493934440333821,
      "grad_norm": 1.0,
      "learning_rate": 0.0008311687969276273,
      "loss": 1.3191,
      "step": 15684
    },
    {
      "epoch": 1.3494794803407038,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008311485373750027,
      "loss": 1.3513,
      "step": 15685
    },
    {
      "epoch": 1.3495655166480254,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008311282768538329,
      "loss": 1.3109,
      "step": 15686
    },
    {
      "epoch": 1.3496515529553472,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008311080153641775,
      "loss": 1.3261,
      "step": 15687
    },
    {
      "epoch": 1.3497375892626688,
      "grad_norm": 1.0,
      "learning_rate": 0.0008310877529060954,
      "loss": 1.4369,
      "step": 15688
    },
    {
      "epoch": 1.3498236255699905,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008310674894796462,
      "loss": 1.8602,
      "step": 15689
    },
    {
      "epoch": 1.3499096618773123,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008310472250848888,
      "loss": 1.3268,
      "step": 15690
    },
    {
      "epoch": 1.349995698184634,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008310269597218827,
      "loss": 1.3985,
      "step": 15691
    },
    {
      "epoch": 1.3500817344919556,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000831006693390687,
      "loss": 0.9093,
      "step": 15692
    },
    {
      "epoch": 1.3501677707992772,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008309864260913613,
      "loss": 1.3716,
      "step": 15693
    },
    {
      "epoch": 1.350253807106599,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008309661578239647,
      "loss": 1.1539,
      "step": 15694
    },
    {
      "epoch": 1.3503398434139207,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008309458885885564,
      "loss": 1.535,
      "step": 15695
    },
    {
      "epoch": 1.3504258797212423,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008309256183851956,
      "loss": 1.2682,
      "step": 15696
    },
    {
      "epoch": 1.3505119160285641,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008309053472139418,
      "loss": 1.7589,
      "step": 15697
    },
    {
      "epoch": 1.3505979523358858,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008308850750748543,
      "loss": 1.0471,
      "step": 15698
    },
    {
      "epoch": 1.3506839886432074,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008308648019679923,
      "loss": 1.4568,
      "step": 15699
    },
    {
      "epoch": 1.350770024950529,
      "grad_norm": 2.0625,
      "learning_rate": 0.000830844527893415,
      "loss": 1.6193,
      "step": 15700
    },
    {
      "epoch": 1.3508560612578508,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008308242528511818,
      "loss": 0.8636,
      "step": 15701
    },
    {
      "epoch": 1.3509420975651725,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008308039768413521,
      "loss": 0.7991,
      "step": 15702
    },
    {
      "epoch": 1.3510281338724943,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008307836998639851,
      "loss": 1.2165,
      "step": 15703
    },
    {
      "epoch": 1.351114170179816,
      "grad_norm": 3.109375,
      "learning_rate": 0.00083076342191914,
      "loss": 1.4075,
      "step": 15704
    },
    {
      "epoch": 1.3512002064871376,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008307431430068764,
      "loss": 1.1783,
      "step": 15705
    },
    {
      "epoch": 1.3512862427944592,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008307228631272533,
      "loss": 1.1686,
      "step": 15706
    },
    {
      "epoch": 1.3513722791017808,
      "grad_norm": 1.25,
      "learning_rate": 0.0008307025822803302,
      "loss": 2.2653,
      "step": 15707
    },
    {
      "epoch": 1.3514583154091027,
      "grad_norm": 2.25,
      "learning_rate": 0.0008306823004661663,
      "loss": 1.0471,
      "step": 15708
    },
    {
      "epoch": 1.3515443517164243,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000830662017684821,
      "loss": 0.6776,
      "step": 15709
    },
    {
      "epoch": 1.3516303880237461,
      "grad_norm": 1.0,
      "learning_rate": 0.0008306417339363537,
      "loss": 1.2889,
      "step": 15710
    },
    {
      "epoch": 1.3517164243310678,
      "grad_norm": 1.25,
      "learning_rate": 0.0008306214492208235,
      "loss": 1.1865,
      "step": 15711
    },
    {
      "epoch": 1.3518024606383894,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008306011635382899,
      "loss": 1.4051,
      "step": 15712
    },
    {
      "epoch": 1.351888496945711,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008305808768888124,
      "loss": 2.0233,
      "step": 15713
    },
    {
      "epoch": 1.3519745332530329,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008305605892724499,
      "loss": 1.5598,
      "step": 15714
    },
    {
      "epoch": 1.3520605695603545,
      "grad_norm": 1.3125,
      "learning_rate": 0.000830540300689262,
      "loss": 1.5859,
      "step": 15715
    },
    {
      "epoch": 1.352146605867676,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008305200111393081,
      "loss": 0.9693,
      "step": 15716
    },
    {
      "epoch": 1.352232642174998,
      "grad_norm": 1.25,
      "learning_rate": 0.0008304997206226474,
      "loss": 1.3358,
      "step": 15717
    },
    {
      "epoch": 1.3523186784823196,
      "grad_norm": 1.25,
      "learning_rate": 0.0008304794291393393,
      "loss": 1.2945,
      "step": 15718
    },
    {
      "epoch": 1.3524047147896412,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008304591366894432,
      "loss": 0.9994,
      "step": 15719
    },
    {
      "epoch": 1.3524907510969628,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008304388432730182,
      "loss": 1.5034,
      "step": 15720
    },
    {
      "epoch": 1.3525767874042847,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008304185488901239,
      "loss": 1.188,
      "step": 15721
    },
    {
      "epoch": 1.3526628237116063,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008303982535408196,
      "loss": 0.9851,
      "step": 15722
    },
    {
      "epoch": 1.352748860018928,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008303779572251649,
      "loss": 1.7131,
      "step": 15723
    },
    {
      "epoch": 1.3528348963262498,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008303576599432188,
      "loss": 1.5939,
      "step": 15724
    },
    {
      "epoch": 1.3529209326335714,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008303373616950407,
      "loss": 1.1438,
      "step": 15725
    },
    {
      "epoch": 1.353006968940893,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008303170624806901,
      "loss": 1.6488,
      "step": 15726
    },
    {
      "epoch": 1.3530930052482146,
      "grad_norm": 1.0,
      "learning_rate": 0.0008302967623002264,
      "loss": 1.333,
      "step": 15727
    },
    {
      "epoch": 1.3531790415555365,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008302764611537089,
      "loss": 1.3714,
      "step": 15728
    },
    {
      "epoch": 1.353265077862858,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000830256159041197,
      "loss": 1.678,
      "step": 15729
    },
    {
      "epoch": 1.35335111417018,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008302358559627501,
      "loss": 1.5196,
      "step": 15730
    },
    {
      "epoch": 1.3534371504775016,
      "grad_norm": 3.921875,
      "learning_rate": 0.0008302155519184274,
      "loss": 1.0663,
      "step": 15731
    },
    {
      "epoch": 1.3535231867848232,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008301952469082886,
      "loss": 1.453,
      "step": 15732
    },
    {
      "epoch": 1.3536092230921448,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008301749409323928,
      "loss": 1.419,
      "step": 15733
    },
    {
      "epoch": 1.3536952593994664,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008301546339907995,
      "loss": 1.0866,
      "step": 15734
    },
    {
      "epoch": 1.3537812957067883,
      "grad_norm": 1.5,
      "learning_rate": 0.0008301343260835682,
      "loss": 1.9103,
      "step": 15735
    },
    {
      "epoch": 1.35386733201411,
      "grad_norm": 1.25,
      "learning_rate": 0.0008301140172107582,
      "loss": 0.9018,
      "step": 15736
    },
    {
      "epoch": 1.3539533683214318,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008300937073724288,
      "loss": 1.3536,
      "step": 15737
    },
    {
      "epoch": 1.3540394046287534,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008300733965686397,
      "loss": 1.4678,
      "step": 15738
    },
    {
      "epoch": 1.354125440936075,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008300530847994499,
      "loss": 1.6233,
      "step": 15739
    },
    {
      "epoch": 1.3542114772433966,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000830032772064919,
      "loss": 0.9776,
      "step": 15740
    },
    {
      "epoch": 1.3542975135507185,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008300124583651065,
      "loss": 1.3663,
      "step": 15741
    },
    {
      "epoch": 1.35438354985804,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008299921437000719,
      "loss": 1.4414,
      "step": 15742
    },
    {
      "epoch": 1.3544695861653617,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008299718280698743,
      "loss": 0.7214,
      "step": 15743
    },
    {
      "epoch": 1.3545556224726836,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008299515114745734,
      "loss": 1.4304,
      "step": 15744
    },
    {
      "epoch": 1.3546416587800052,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008299311939142283,
      "loss": 1.2066,
      "step": 15745
    },
    {
      "epoch": 1.3547276950873268,
      "grad_norm": 1.25,
      "learning_rate": 0.0008299108753888988,
      "loss": 1.5592,
      "step": 15746
    },
    {
      "epoch": 1.3548137313946484,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008298905558986441,
      "loss": 1.0084,
      "step": 15747
    },
    {
      "epoch": 1.3548997677019703,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008298702354435236,
      "loss": 1.853,
      "step": 15748
    },
    {
      "epoch": 1.354985804009292,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000829849914023597,
      "loss": 1.4058,
      "step": 15749
    },
    {
      "epoch": 1.3550718403166135,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0008298295916389233,
      "loss": 1.5858,
      "step": 15750
    },
    {
      "epoch": 1.3551578766239354,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008298092682895625,
      "loss": 1.0923,
      "step": 15751
    },
    {
      "epoch": 1.355243912931257,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008297889439755735,
      "loss": 1.3314,
      "step": 15752
    },
    {
      "epoch": 1.3553299492385786,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008297686186970162,
      "loss": 1.7251,
      "step": 15753
    },
    {
      "epoch": 1.3554159855459003,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008297482924539497,
      "loss": 0.6703,
      "step": 15754
    },
    {
      "epoch": 1.355502021853222,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008297279652464336,
      "loss": 1.8574,
      "step": 15755
    },
    {
      "epoch": 1.3555880581605437,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008297076370745271,
      "loss": 1.3524,
      "step": 15756
    },
    {
      "epoch": 1.3556740944678656,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008296873079382901,
      "loss": 1.8263,
      "step": 15757
    },
    {
      "epoch": 1.3557601307751872,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008296669778377817,
      "loss": 0.6732,
      "step": 15758
    },
    {
      "epoch": 1.3558461670825088,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0008296466467730617,
      "loss": 1.5986,
      "step": 15759
    },
    {
      "epoch": 1.3559322033898304,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008296263147441893,
      "loss": 1.2068,
      "step": 15760
    },
    {
      "epoch": 1.356018239697152,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000829605981751224,
      "loss": 1.2599,
      "step": 15761
    },
    {
      "epoch": 1.356104276004474,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008295856477942254,
      "loss": 1.6568,
      "step": 15762
    },
    {
      "epoch": 1.3561903123117955,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008295653128732529,
      "loss": 1.659,
      "step": 15763
    },
    {
      "epoch": 1.3562763486191174,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008295449769883658,
      "loss": 1.7369,
      "step": 15764
    },
    {
      "epoch": 1.356362384926439,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008295246401396239,
      "loss": 2.3106,
      "step": 15765
    },
    {
      "epoch": 1.3564484212337606,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008295043023270864,
      "loss": 1.1998,
      "step": 15766
    },
    {
      "epoch": 1.3565344575410823,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008294839635508129,
      "loss": 1.2728,
      "step": 15767
    },
    {
      "epoch": 1.356620493848404,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000829463623810863,
      "loss": 1.0897,
      "step": 15768
    },
    {
      "epoch": 1.3567065301557257,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000829443283107296,
      "loss": 1.1623,
      "step": 15769
    },
    {
      "epoch": 1.3567925664630474,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008294229414401715,
      "loss": 1.5101,
      "step": 15770
    },
    {
      "epoch": 1.3568786027703692,
      "grad_norm": 1.921875,
      "learning_rate": 0.0008294025988095488,
      "loss": 1.9944,
      "step": 15771
    },
    {
      "epoch": 1.3569646390776908,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008293822552154877,
      "loss": 0.7775,
      "step": 15772
    },
    {
      "epoch": 1.3570506753850125,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008293619106580477,
      "loss": 2.0593,
      "step": 15773
    },
    {
      "epoch": 1.357136711692334,
      "grad_norm": 1.984375,
      "learning_rate": 0.000829341565137288,
      "loss": 2.0719,
      "step": 15774
    },
    {
      "epoch": 1.357222747999656,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008293212186532684,
      "loss": 1.7759,
      "step": 15775
    },
    {
      "epoch": 1.3573087843069775,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008293008712060481,
      "loss": 1.4717,
      "step": 15776
    },
    {
      "epoch": 1.3573948206142992,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008292805227956868,
      "loss": 1.1822,
      "step": 15777
    },
    {
      "epoch": 1.357480856921621,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008292601734222442,
      "loss": 0.816,
      "step": 15778
    },
    {
      "epoch": 1.3575668932289426,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008292398230857795,
      "loss": 2.0645,
      "step": 15779
    },
    {
      "epoch": 1.3576529295362643,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008292194717863524,
      "loss": 1.1736,
      "step": 15780
    },
    {
      "epoch": 1.3577389658435859,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008291991195240223,
      "loss": 1.0373,
      "step": 15781
    },
    {
      "epoch": 1.3578250021509077,
      "grad_norm": 1.875,
      "learning_rate": 0.0008291787662988488,
      "loss": 1.9084,
      "step": 15782
    },
    {
      "epoch": 1.3579110384582294,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008291584121108916,
      "loss": 1.0311,
      "step": 15783
    },
    {
      "epoch": 1.357997074765551,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00082913805696021,
      "loss": 1.0942,
      "step": 15784
    },
    {
      "epoch": 1.3580831110728728,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008291177008468636,
      "loss": 0.9008,
      "step": 15785
    },
    {
      "epoch": 1.3581691473801945,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008290973437709117,
      "loss": 1.3386,
      "step": 15786
    },
    {
      "epoch": 1.358255183687516,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008290769857324143,
      "loss": 1.3091,
      "step": 15787
    },
    {
      "epoch": 1.3583412199948377,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008290566267314307,
      "loss": 1.1527,
      "step": 15788
    },
    {
      "epoch": 1.3584272563021595,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008290362667680204,
      "loss": 1.2175,
      "step": 15789
    },
    {
      "epoch": 1.3585132926094812,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000829015905842243,
      "loss": 1.4199,
      "step": 15790
    },
    {
      "epoch": 1.358599328916803,
      "grad_norm": 1.375,
      "learning_rate": 0.0008289955439541583,
      "loss": 1.5958,
      "step": 15791
    },
    {
      "epoch": 1.3586853652241246,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008289751811038254,
      "loss": 1.2095,
      "step": 15792
    },
    {
      "epoch": 1.3587714015314463,
      "grad_norm": 1.125,
      "learning_rate": 0.0008289548172913042,
      "loss": 1.1085,
      "step": 15793
    },
    {
      "epoch": 1.358857437838768,
      "grad_norm": 1.234375,
      "learning_rate": 0.000828934452516654,
      "loss": 1.6833,
      "step": 15794
    },
    {
      "epoch": 1.3589434741460895,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008289140867799345,
      "loss": 1.3084,
      "step": 15795
    },
    {
      "epoch": 1.3590295104534114,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008288937200812053,
      "loss": 1.2413,
      "step": 15796
    },
    {
      "epoch": 1.359115546760733,
      "grad_norm": 1.171875,
      "learning_rate": 0.000828873352420526,
      "loss": 0.9166,
      "step": 15797
    },
    {
      "epoch": 1.3592015830680548,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000828852983797956,
      "loss": 1.9536,
      "step": 15798
    },
    {
      "epoch": 1.3592876193753765,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008288326142135552,
      "loss": 1.3087,
      "step": 15799
    },
    {
      "epoch": 1.359373655682698,
      "grad_norm": 0.890625,
      "learning_rate": 0.0008288122436673828,
      "loss": 0.9032,
      "step": 15800
    },
    {
      "epoch": 1.3594596919900197,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008287918721594984,
      "loss": 1.1405,
      "step": 15801
    },
    {
      "epoch": 1.3595457282973415,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000828771499689962,
      "loss": 1.0753,
      "step": 15802
    },
    {
      "epoch": 1.3596317646046632,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008287511262588328,
      "loss": 0.7995,
      "step": 15803
    },
    {
      "epoch": 1.3597178009119848,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008287307518661704,
      "loss": 0.754,
      "step": 15804
    },
    {
      "epoch": 1.3598038372193066,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008287103765120345,
      "loss": 1.9524,
      "step": 15805
    },
    {
      "epoch": 1.3598898735266283,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008286900001964847,
      "loss": 1.6541,
      "step": 15806
    },
    {
      "epoch": 1.35997590983395,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008286696229195807,
      "loss": 1.4706,
      "step": 15807
    },
    {
      "epoch": 1.3600619461412715,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008286492446813818,
      "loss": 0.6879,
      "step": 15808
    },
    {
      "epoch": 1.3601479824485934,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008286288654819479,
      "loss": 1.5448,
      "step": 15809
    },
    {
      "epoch": 1.360234018755915,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008286084853213384,
      "loss": 0.985,
      "step": 15810
    },
    {
      "epoch": 1.3603200550632366,
      "grad_norm": 1.0,
      "learning_rate": 0.000828588104199613,
      "loss": 1.0454,
      "step": 15811
    },
    {
      "epoch": 1.3604060913705585,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008285677221168314,
      "loss": 2.078,
      "step": 15812
    },
    {
      "epoch": 1.36049212767788,
      "grad_norm": 1.25,
      "learning_rate": 0.0008285473390730532,
      "loss": 1.3238,
      "step": 15813
    },
    {
      "epoch": 1.3605781639852017,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008285269550683377,
      "loss": 2.0224,
      "step": 15814
    },
    {
      "epoch": 1.3606642002925233,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008285065701027448,
      "loss": 0.9918,
      "step": 15815
    },
    {
      "epoch": 1.3607502365998452,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000828486184176334,
      "loss": 1.6897,
      "step": 15816
    },
    {
      "epoch": 1.3608362729071668,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008284657972891652,
      "loss": 1.117,
      "step": 15817
    },
    {
      "epoch": 1.3609223092144886,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008284454094412976,
      "loss": 0.9908,
      "step": 15818
    },
    {
      "epoch": 1.3610083455218103,
      "grad_norm": 2.140625,
      "learning_rate": 0.0008284250206327912,
      "loss": 1.411,
      "step": 15819
    },
    {
      "epoch": 1.361094381829132,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008284046308637055,
      "loss": 1.4092,
      "step": 15820
    },
    {
      "epoch": 1.3611804181364535,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008283842401341001,
      "loss": 1.043,
      "step": 15821
    },
    {
      "epoch": 1.3612664544437751,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008283638484440348,
      "loss": 2.2888,
      "step": 15822
    },
    {
      "epoch": 1.361352490751097,
      "grad_norm": 1.109375,
      "learning_rate": 0.000828343455793569,
      "loss": 1.4349,
      "step": 15823
    },
    {
      "epoch": 1.3614385270584186,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008283230621827625,
      "loss": 1.1717,
      "step": 15824
    },
    {
      "epoch": 1.3615245633657405,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008283026676116748,
      "loss": 1.671,
      "step": 15825
    },
    {
      "epoch": 1.361610599673062,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008282822720803656,
      "loss": 1.8018,
      "step": 15826
    },
    {
      "epoch": 1.3616966359803837,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008282618755888946,
      "loss": 1.6441,
      "step": 15827
    },
    {
      "epoch": 1.3617826722877053,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008282414781373216,
      "loss": 0.7986,
      "step": 15828
    },
    {
      "epoch": 1.3618687085950272,
      "grad_norm": 0.84375,
      "learning_rate": 0.000828221079725706,
      "loss": 0.9975,
      "step": 15829
    },
    {
      "epoch": 1.3619547449023488,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008282006803541075,
      "loss": 0.7484,
      "step": 15830
    },
    {
      "epoch": 1.3620407812096704,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008281802800225861,
      "loss": 0.9624,
      "step": 15831
    },
    {
      "epoch": 1.3621268175169923,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008281598787312011,
      "loss": 2.1865,
      "step": 15832
    },
    {
      "epoch": 1.362212853824314,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008281394764800122,
      "loss": 0.8908,
      "step": 15833
    },
    {
      "epoch": 1.3622988901316355,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008281190732690792,
      "loss": 0.8684,
      "step": 15834
    },
    {
      "epoch": 1.3623849264389571,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008280986690984617,
      "loss": 1.7145,
      "step": 15835
    },
    {
      "epoch": 1.362470962746279,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008280782639682193,
      "loss": 0.7123,
      "step": 15836
    },
    {
      "epoch": 1.3625569990536006,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008280578578784122,
      "loss": 0.8094,
      "step": 15837
    },
    {
      "epoch": 1.3626430353609222,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008280374508290993,
      "loss": 0.999,
      "step": 15838
    },
    {
      "epoch": 1.362729071668244,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008280170428203407,
      "loss": 1.5128,
      "step": 15839
    },
    {
      "epoch": 1.3628151079755657,
      "grad_norm": 1.109375,
      "learning_rate": 0.000827996633852196,
      "loss": 1.4212,
      "step": 15840
    },
    {
      "epoch": 1.3629011442828873,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008279762239247251,
      "loss": 1.0489,
      "step": 15841
    },
    {
      "epoch": 1.362987180590209,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008279558130379875,
      "loss": 1.6087,
      "step": 15842
    },
    {
      "epoch": 1.3630732168975308,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008279354011920429,
      "loss": 1.8416,
      "step": 15843
    },
    {
      "epoch": 1.3631592532048524,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000827914988386951,
      "loss": 1.2283,
      "step": 15844
    },
    {
      "epoch": 1.3632452895121743,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008278945746227715,
      "loss": 1.0832,
      "step": 15845
    },
    {
      "epoch": 1.363331325819496,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008278741598995642,
      "loss": 1.6692,
      "step": 15846
    },
    {
      "epoch": 1.3634173621268175,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008278537442173888,
      "loss": 1.3886,
      "step": 15847
    },
    {
      "epoch": 1.3635033984341391,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000827833327576305,
      "loss": 1.8981,
      "step": 15848
    },
    {
      "epoch": 1.3635894347414608,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008278129099763724,
      "loss": 1.0279,
      "step": 15849
    },
    {
      "epoch": 1.3636754710487826,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008277924914176508,
      "loss": 2.2752,
      "step": 15850
    },
    {
      "epoch": 1.3637615073561042,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008277720719001997,
      "loss": 1.7177,
      "step": 15851
    },
    {
      "epoch": 1.363847543663426,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008277516514240792,
      "loss": 1.0649,
      "step": 15852
    },
    {
      "epoch": 1.3639335799707477,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000827731229989349,
      "loss": 1.7526,
      "step": 15853
    },
    {
      "epoch": 1.3640196162780693,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008277108075960686,
      "loss": 1.433,
      "step": 15854
    },
    {
      "epoch": 1.364105652585391,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008276903842442979,
      "loss": 1.0109,
      "step": 15855
    },
    {
      "epoch": 1.3641916888927128,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008276699599340965,
      "loss": 1.1224,
      "step": 15856
    },
    {
      "epoch": 1.3642777252000344,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008276495346655242,
      "loss": 1.6373,
      "step": 15857
    },
    {
      "epoch": 1.364363761507356,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008276291084386407,
      "loss": 0.5782,
      "step": 15858
    },
    {
      "epoch": 1.364449797814678,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008276086812535057,
      "loss": 1.6155,
      "step": 15859
    },
    {
      "epoch": 1.3645358341219995,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008275882531101791,
      "loss": 0.9997,
      "step": 15860
    },
    {
      "epoch": 1.3646218704293211,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008275678240087205,
      "loss": 1.1059,
      "step": 15861
    },
    {
      "epoch": 1.3647079067366428,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008275473939491899,
      "loss": 1.1506,
      "step": 15862
    },
    {
      "epoch": 1.3647939430439646,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008275269629316468,
      "loss": 1.703,
      "step": 15863
    },
    {
      "epoch": 1.3648799793512862,
      "grad_norm": 2.125,
      "learning_rate": 0.000827506530956151,
      "loss": 1.1807,
      "step": 15864
    },
    {
      "epoch": 1.3649660156586079,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008274860980227623,
      "loss": 0.8669,
      "step": 15865
    },
    {
      "epoch": 1.3650520519659297,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008274656641315406,
      "loss": 1.1405,
      "step": 15866
    },
    {
      "epoch": 1.3651380882732513,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008274452292825453,
      "loss": 1.8538,
      "step": 15867
    },
    {
      "epoch": 1.365224124580573,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008274247934758366,
      "loss": 1.2363,
      "step": 15868
    },
    {
      "epoch": 1.3653101608878946,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008274043567114737,
      "loss": 1.7934,
      "step": 15869
    },
    {
      "epoch": 1.3653961971952164,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000827383918989517,
      "loss": 1.1727,
      "step": 15870
    },
    {
      "epoch": 1.365482233502538,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000827363480310026,
      "loss": 1.7187,
      "step": 15871
    },
    {
      "epoch": 1.3655682698098597,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008273430406730604,
      "loss": 1.7325,
      "step": 15872
    },
    {
      "epoch": 1.3656543061171815,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008273226000786802,
      "loss": 1.3985,
      "step": 15873
    },
    {
      "epoch": 1.3657403424245032,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008273021585269449,
      "loss": 1.0376,
      "step": 15874
    },
    {
      "epoch": 1.3658263787318248,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008272817160179147,
      "loss": 0.572,
      "step": 15875
    },
    {
      "epoch": 1.3659124150391464,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008272612725516488,
      "loss": 1.251,
      "step": 15876
    },
    {
      "epoch": 1.3659984513464682,
      "grad_norm": 1.0,
      "learning_rate": 0.0008272408281282075,
      "loss": 1.7756,
      "step": 15877
    },
    {
      "epoch": 1.3660844876537899,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008272203827476506,
      "loss": 1.4454,
      "step": 15878
    },
    {
      "epoch": 1.3661705239611117,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008271999364100374,
      "loss": 0.7951,
      "step": 15879
    },
    {
      "epoch": 1.3662565602684333,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008271794891154283,
      "loss": 0.9766,
      "step": 15880
    },
    {
      "epoch": 1.366342596575755,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008271590408638826,
      "loss": 1.4373,
      "step": 15881
    },
    {
      "epoch": 1.3664286328830766,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008271385916554605,
      "loss": 0.8878,
      "step": 15882
    },
    {
      "epoch": 1.3665146691903982,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008271181414902216,
      "loss": 0.8275,
      "step": 15883
    },
    {
      "epoch": 1.36660070549772,
      "grad_norm": 1.0,
      "learning_rate": 0.0008270976903682256,
      "loss": 0.7166,
      "step": 15884
    },
    {
      "epoch": 1.3666867418050417,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008270772382895326,
      "loss": 1.446,
      "step": 15885
    },
    {
      "epoch": 1.3667727781123635,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008270567852542023,
      "loss": 0.8288,
      "step": 15886
    },
    {
      "epoch": 1.3668588144196852,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008270363312622946,
      "loss": 0.6103,
      "step": 15887
    },
    {
      "epoch": 1.3669448507270068,
      "grad_norm": 1.265625,
      "learning_rate": 0.000827015876313869,
      "loss": 1.7244,
      "step": 15888
    },
    {
      "epoch": 1.3670308870343284,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008269954204089857,
      "loss": 1.0962,
      "step": 15889
    },
    {
      "epoch": 1.3671169233416502,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008269749635477044,
      "loss": 1.2136,
      "step": 15890
    },
    {
      "epoch": 1.3672029596489719,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000826954505730085,
      "loss": 1.304,
      "step": 15891
    },
    {
      "epoch": 1.3672889959562935,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000826934046956187,
      "loss": 0.628,
      "step": 15892
    },
    {
      "epoch": 1.3673750322636153,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008269135872260708,
      "loss": 1.2474,
      "step": 15893
    },
    {
      "epoch": 1.367461068570937,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008268931265397958,
      "loss": 0.8491,
      "step": 15894
    },
    {
      "epoch": 1.3675471048782586,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008268726648974219,
      "loss": 1.5881,
      "step": 15895
    },
    {
      "epoch": 1.3676331411855802,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008268522022990092,
      "loss": 1.2031,
      "step": 15896
    },
    {
      "epoch": 1.367719177492902,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008268317387446173,
      "loss": 0.9885,
      "step": 15897
    },
    {
      "epoch": 1.3678052138002237,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008268112742343062,
      "loss": 1.3418,
      "step": 15898
    },
    {
      "epoch": 1.3678912501075453,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008267908087681355,
      "loss": 2.0752,
      "step": 15899
    },
    {
      "epoch": 1.3679772864148672,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008267703423461654,
      "loss": 1.1778,
      "step": 15900
    },
    {
      "epoch": 1.3680633227221888,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008267498749684555,
      "loss": 0.8565,
      "step": 15901
    },
    {
      "epoch": 1.3681493590295104,
      "grad_norm": 3.0625,
      "learning_rate": 0.0008267294066350659,
      "loss": 1.2348,
      "step": 15902
    },
    {
      "epoch": 1.368235395336832,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008267089373460562,
      "loss": 1.5638,
      "step": 15903
    },
    {
      "epoch": 1.3683214316441539,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008266884671014865,
      "loss": 1.9752,
      "step": 15904
    },
    {
      "epoch": 1.3684074679514755,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008266679959014165,
      "loss": 1.6211,
      "step": 15905
    },
    {
      "epoch": 1.3684935042587973,
      "grad_norm": 2.625,
      "learning_rate": 0.0008266475237459061,
      "loss": 2.0141,
      "step": 15906
    },
    {
      "epoch": 1.368579540566119,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008266270506350153,
      "loss": 1.1655,
      "step": 15907
    },
    {
      "epoch": 1.3686655768734406,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000826606576568804,
      "loss": 1.8993,
      "step": 15908
    },
    {
      "epoch": 1.3687516131807622,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008265861015473318,
      "loss": 1.8891,
      "step": 15909
    },
    {
      "epoch": 1.3688376494880838,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008265656255706589,
      "loss": 1.027,
      "step": 15910
    },
    {
      "epoch": 1.3689236857954057,
      "grad_norm": 1.03125,
      "learning_rate": 0.000826545148638845,
      "loss": 0.9329,
      "step": 15911
    },
    {
      "epoch": 1.3690097221027273,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00082652467075195,
      "loss": 0.8022,
      "step": 15912
    },
    {
      "epoch": 1.3690957584100492,
      "grad_norm": 1.109375,
      "learning_rate": 0.000826504191910034,
      "loss": 0.864,
      "step": 15913
    },
    {
      "epoch": 1.3691817947173708,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008264837121131566,
      "loss": 0.8563,
      "step": 15914
    },
    {
      "epoch": 1.3692678310246924,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0008264632313613779,
      "loss": 0.9956,
      "step": 15915
    },
    {
      "epoch": 1.369353867332014,
      "grad_norm": 2.078125,
      "learning_rate": 0.0008264427496547577,
      "loss": 1.6569,
      "step": 15916
    },
    {
      "epoch": 1.3694399036393359,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000826422266993356,
      "loss": 0.7864,
      "step": 15917
    },
    {
      "epoch": 1.3695259399466575,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008264017833772326,
      "loss": 1.1489,
      "step": 15918
    },
    {
      "epoch": 1.3696119762539791,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008263812988064475,
      "loss": 2.2106,
      "step": 15919
    },
    {
      "epoch": 1.369698012561301,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008263608132810605,
      "loss": 1.1661,
      "step": 15920
    },
    {
      "epoch": 1.3697840488686226,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008263403268011318,
      "loss": 1.4602,
      "step": 15921
    },
    {
      "epoch": 1.3698700851759442,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008263198393667209,
      "loss": 1.7896,
      "step": 15922
    },
    {
      "epoch": 1.3699561214832658,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000826299350977888,
      "loss": 1.7145,
      "step": 15923
    },
    {
      "epoch": 1.3700421577905877,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000826278861634693,
      "loss": 1.3121,
      "step": 15924
    },
    {
      "epoch": 1.3701281940979093,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008262583713371958,
      "loss": 1.3125,
      "step": 15925
    },
    {
      "epoch": 1.370214230405231,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008262378800854562,
      "loss": 0.7983,
      "step": 15926
    },
    {
      "epoch": 1.3703002667125528,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008262173878795342,
      "loss": 0.9665,
      "step": 15927
    },
    {
      "epoch": 1.3703863030198744,
      "grad_norm": 1.125,
      "learning_rate": 0.00082619689471949,
      "loss": 1.0971,
      "step": 15928
    },
    {
      "epoch": 1.370472339327196,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008261764006053834,
      "loss": 1.5881,
      "step": 15929
    },
    {
      "epoch": 1.3705583756345177,
      "grad_norm": 1.03125,
      "learning_rate": 0.000826155905537274,
      "loss": 0.9806,
      "step": 15930
    },
    {
      "epoch": 1.3706444119418395,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008261354095152222,
      "loss": 1.1987,
      "step": 15931
    },
    {
      "epoch": 1.3707304482491611,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008261149125392876,
      "loss": 1.0263,
      "step": 15932
    },
    {
      "epoch": 1.370816484556483,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008260944146095304,
      "loss": 1.5765,
      "step": 15933
    },
    {
      "epoch": 1.3709025208638046,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008260739157260104,
      "loss": 0.7378,
      "step": 15934
    },
    {
      "epoch": 1.3709885571711262,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008260534158887877,
      "loss": 1.2868,
      "step": 15935
    },
    {
      "epoch": 1.3710745934784478,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008260329150979221,
      "loss": 1.0526,
      "step": 15936
    },
    {
      "epoch": 1.3711606297857695,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008260124133534736,
      "loss": 1.3981,
      "step": 15937
    },
    {
      "epoch": 1.3712466660930913,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008259919106555023,
      "loss": 1.9412,
      "step": 15938
    },
    {
      "epoch": 1.371332702400413,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000825971407004068,
      "loss": 0.7625,
      "step": 15939
    },
    {
      "epoch": 1.3714187387077348,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008259509023992309,
      "loss": 1.2251,
      "step": 15940
    },
    {
      "epoch": 1.3715047750150564,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008259303968410506,
      "loss": 1.4536,
      "step": 15941
    },
    {
      "epoch": 1.371590811322378,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008259098903295873,
      "loss": 0.6269,
      "step": 15942
    },
    {
      "epoch": 1.3716768476296997,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008258893828649009,
      "loss": 0.6325,
      "step": 15943
    },
    {
      "epoch": 1.3717628839370215,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008258688744470514,
      "loss": 1.0444,
      "step": 15944
    },
    {
      "epoch": 1.3718489202443431,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000825848365076099,
      "loss": 0.6473,
      "step": 15945
    },
    {
      "epoch": 1.3719349565516648,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008258278547521036,
      "loss": 0.7987,
      "step": 15946
    },
    {
      "epoch": 1.3720209928589866,
      "grad_norm": 0.921875,
      "learning_rate": 0.000825807343475125,
      "loss": 1.8034,
      "step": 15947
    },
    {
      "epoch": 1.3721070291663082,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008257868312452232,
      "loss": 1.3871,
      "step": 15948
    },
    {
      "epoch": 1.3721930654736298,
      "grad_norm": 2.375,
      "learning_rate": 0.0008257663180624582,
      "loss": 1.4878,
      "step": 15949
    },
    {
      "epoch": 1.3722791017809515,
      "grad_norm": 1.0,
      "learning_rate": 0.0008257458039268902,
      "loss": 0.7404,
      "step": 15950
    },
    {
      "epoch": 1.3723651380882733,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008257252888385792,
      "loss": 0.9579,
      "step": 15951
    },
    {
      "epoch": 1.372451174395595,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008257047727975848,
      "loss": 1.6261,
      "step": 15952
    },
    {
      "epoch": 1.3725372107029166,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008256842558039676,
      "loss": 0.6506,
      "step": 15953
    },
    {
      "epoch": 1.3726232470102384,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008256637378577872,
      "loss": 1.1915,
      "step": 15954
    },
    {
      "epoch": 1.37270928331756,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008256432189591036,
      "loss": 0.8384,
      "step": 15955
    },
    {
      "epoch": 1.3727953196248817,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008256226991079769,
      "loss": 1.026,
      "step": 15956
    },
    {
      "epoch": 1.3728813559322033,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008256021783044672,
      "loss": 1.077,
      "step": 15957
    },
    {
      "epoch": 1.3729673922395251,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008255816565486345,
      "loss": 0.6517,
      "step": 15958
    },
    {
      "epoch": 1.3730534285468468,
      "grad_norm": 1.25,
      "learning_rate": 0.0008255611338405386,
      "loss": 1.8625,
      "step": 15959
    },
    {
      "epoch": 1.3731394648541684,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008255406101802398,
      "loss": 1.3803,
      "step": 15960
    },
    {
      "epoch": 1.3732255011614902,
      "grad_norm": 1.328125,
      "learning_rate": 0.000825520085567798,
      "loss": 1.7969,
      "step": 15961
    },
    {
      "epoch": 1.3733115374688118,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008254995600032733,
      "loss": 1.6466,
      "step": 15962
    },
    {
      "epoch": 1.3733975737761335,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008254790334867258,
      "loss": 1.3036,
      "step": 15963
    },
    {
      "epoch": 1.373483610083455,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008254585060182153,
      "loss": 1.2371,
      "step": 15964
    },
    {
      "epoch": 1.373569646390777,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000825437977597802,
      "loss": 1.111,
      "step": 15965
    },
    {
      "epoch": 1.3736556826980986,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000825417448225546,
      "loss": 0.6381,
      "step": 15966
    },
    {
      "epoch": 1.3737417190054204,
      "grad_norm": 0.8125,
      "learning_rate": 0.000825396917901507,
      "loss": 1.078,
      "step": 15967
    },
    {
      "epoch": 1.373827755312742,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008253763866257456,
      "loss": 1.4335,
      "step": 15968
    },
    {
      "epoch": 1.3739137916200637,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008253558543983214,
      "loss": 1.5027,
      "step": 15969
    },
    {
      "epoch": 1.3739998279273853,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008253353212192947,
      "loss": 0.7949,
      "step": 15970
    },
    {
      "epoch": 1.374085864234707,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008253147870887253,
      "loss": 1.2553,
      "step": 15971
    },
    {
      "epoch": 1.3741719005420288,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008252942520066735,
      "loss": 1.123,
      "step": 15972
    },
    {
      "epoch": 1.3742579368493504,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008252737159731993,
      "loss": 1.1007,
      "step": 15973
    },
    {
      "epoch": 1.3743439731566722,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008252531789883627,
      "loss": 0.7025,
      "step": 15974
    },
    {
      "epoch": 1.3744300094639939,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000825232641052224,
      "loss": 1.2883,
      "step": 15975
    },
    {
      "epoch": 1.3745160457713155,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008252121021648427,
      "loss": 1.5708,
      "step": 15976
    },
    {
      "epoch": 1.374602082078637,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008251915623262796,
      "loss": 2.1326,
      "step": 15977
    },
    {
      "epoch": 1.374688118385959,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008251710215365942,
      "loss": 1.2978,
      "step": 15978
    },
    {
      "epoch": 1.3747741546932806,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008251504797958469,
      "loss": 2.0594,
      "step": 15979
    },
    {
      "epoch": 1.3748601910006022,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008251299371040976,
      "loss": 1.9429,
      "step": 15980
    },
    {
      "epoch": 1.374946227307924,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008251093934614067,
      "loss": 1.1742,
      "step": 15981
    },
    {
      "epoch": 1.3750322636152457,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008250888488678339,
      "loss": 1.1818,
      "step": 15982
    },
    {
      "epoch": 1.3751182999225673,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008250683033234394,
      "loss": 0.9618,
      "step": 15983
    },
    {
      "epoch": 1.375204336229889,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008250477568282834,
      "loss": 2.2074,
      "step": 15984
    },
    {
      "epoch": 1.3752903725372108,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0008250272093824258,
      "loss": 1.2044,
      "step": 15985
    },
    {
      "epoch": 1.3753764088445324,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000825006660985927,
      "loss": 1.1886,
      "step": 15986
    },
    {
      "epoch": 1.375462445151854,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008249861116388468,
      "loss": 1.4998,
      "step": 15987
    },
    {
      "epoch": 1.3755484814591759,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008249655613412456,
      "loss": 0.4733,
      "step": 15988
    },
    {
      "epoch": 1.3756345177664975,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0008249450100931832,
      "loss": 1.8089,
      "step": 15989
    },
    {
      "epoch": 1.375720554073819,
      "grad_norm": 1.640625,
      "learning_rate": 0.00082492445789472,
      "loss": 1.4106,
      "step": 15990
    },
    {
      "epoch": 1.3758065903811407,
      "grad_norm": 1.125,
      "learning_rate": 0.0008249039047459157,
      "loss": 1.7861,
      "step": 15991
    },
    {
      "epoch": 1.3758926266884626,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008248833506468308,
      "loss": 1.5212,
      "step": 15992
    },
    {
      "epoch": 1.3759786629957842,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008248627955975254,
      "loss": 1.7176,
      "step": 15993
    },
    {
      "epoch": 1.376064699303106,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008248422395980593,
      "loss": 1.785,
      "step": 15994
    },
    {
      "epoch": 1.3761507356104277,
      "grad_norm": 1.125,
      "learning_rate": 0.0008248216826484929,
      "loss": 1.423,
      "step": 15995
    },
    {
      "epoch": 1.3762367719177493,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008248011247488862,
      "loss": 1.0727,
      "step": 15996
    },
    {
      "epoch": 1.376322808225071,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008247805658992995,
      "loss": 1.1858,
      "step": 15997
    },
    {
      "epoch": 1.3764088445323925,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008247600060997928,
      "loss": 1.6366,
      "step": 15998
    },
    {
      "epoch": 1.3764948808397144,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008247394453504261,
      "loss": 1.1878,
      "step": 15999
    },
    {
      "epoch": 1.376580917147036,
      "grad_norm": 1.8125,
      "learning_rate": 0.0008247188836512598,
      "loss": 2.055,
      "step": 16000
    },
    {
      "epoch": 1.3766669534543579,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008246983210023539,
      "loss": 1.3237,
      "step": 16001
    },
    {
      "epoch": 1.3767529897616795,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008246777574037686,
      "loss": 1.4865,
      "step": 16002
    },
    {
      "epoch": 1.376839026069001,
      "grad_norm": 1.078125,
      "learning_rate": 0.000824657192855564,
      "loss": 1.2554,
      "step": 16003
    },
    {
      "epoch": 1.3769250623763227,
      "grad_norm": 0.859375,
      "learning_rate": 0.0008246366273578,
      "loss": 1.091,
      "step": 16004
    },
    {
      "epoch": 1.3770110986836446,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008246160609105374,
      "loss": 0.7767,
      "step": 16005
    },
    {
      "epoch": 1.3770971349909662,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008245954935138358,
      "loss": 1.1303,
      "step": 16006
    },
    {
      "epoch": 1.3771831712982878,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008245749251677555,
      "loss": 1.6622,
      "step": 16007
    },
    {
      "epoch": 1.3772692076056097,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008245543558723567,
      "loss": 1.572,
      "step": 16008
    },
    {
      "epoch": 1.3773552439129313,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008245337856276995,
      "loss": 1.0363,
      "step": 16009
    },
    {
      "epoch": 1.377441280220253,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008245132144338441,
      "loss": 1.2449,
      "step": 16010
    },
    {
      "epoch": 1.3775273165275745,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008244926422908506,
      "loss": 0.9518,
      "step": 16011
    },
    {
      "epoch": 1.3776133528348964,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008244720691987792,
      "loss": 1.6626,
      "step": 16012
    },
    {
      "epoch": 1.377699389142218,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008244514951576903,
      "loss": 0.6659,
      "step": 16013
    },
    {
      "epoch": 1.3777854254495396,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008244309201676439,
      "loss": 1.1088,
      "step": 16014
    },
    {
      "epoch": 1.3778714617568615,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008244103442286998,
      "loss": 1.4199,
      "step": 16015
    },
    {
      "epoch": 1.377957498064183,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008243897673409189,
      "loss": 1.3444,
      "step": 16016
    },
    {
      "epoch": 1.3780435343715047,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008243691895043608,
      "loss": 0.9893,
      "step": 16017
    },
    {
      "epoch": 1.3781295706788264,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008243486107190861,
      "loss": 0.7593,
      "step": 16018
    },
    {
      "epoch": 1.3782156069861482,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008243280309851546,
      "loss": 1.3165,
      "step": 16019
    },
    {
      "epoch": 1.3783016432934698,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008243074503026268,
      "loss": 1.6412,
      "step": 16020
    },
    {
      "epoch": 1.3783876796007917,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008242868686715628,
      "loss": 2.0677,
      "step": 16021
    },
    {
      "epoch": 1.3784737159081133,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008242662860920227,
      "loss": 1.5563,
      "step": 16022
    },
    {
      "epoch": 1.378559752215435,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008242457025640669,
      "loss": 1.2002,
      "step": 16023
    },
    {
      "epoch": 1.3786457885227565,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008242251180877553,
      "loss": 1.25,
      "step": 16024
    },
    {
      "epoch": 1.3787318248300782,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008242045326631484,
      "loss": 0.7395,
      "step": 16025
    },
    {
      "epoch": 1.3788178611374,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008241839462903062,
      "loss": 1.2525,
      "step": 16026
    },
    {
      "epoch": 1.3789038974447216,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008241633589692891,
      "loss": 0.813,
      "step": 16027
    },
    {
      "epoch": 1.3789899337520435,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008241427707001573,
      "loss": 1.893,
      "step": 16028
    },
    {
      "epoch": 1.379075970059365,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008241221814829708,
      "loss": 1.3467,
      "step": 16029
    },
    {
      "epoch": 1.3791620063666867,
      "grad_norm": 1.15625,
      "learning_rate": 0.00082410159131779,
      "loss": 1.4145,
      "step": 16030
    },
    {
      "epoch": 1.3792480426740084,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008240810002046751,
      "loss": 0.7762,
      "step": 16031
    },
    {
      "epoch": 1.3793340789813302,
      "grad_norm": 1.25,
      "learning_rate": 0.0008240604081436863,
      "loss": 1.283,
      "step": 16032
    },
    {
      "epoch": 1.3794201152886518,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008240398151348838,
      "loss": 1.9588,
      "step": 16033
    },
    {
      "epoch": 1.3795061515959735,
      "grad_norm": 1.015625,
      "learning_rate": 0.000824019221178328,
      "loss": 0.8322,
      "step": 16034
    },
    {
      "epoch": 1.3795921879032953,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008239986262740788,
      "loss": 1.3011,
      "step": 16035
    },
    {
      "epoch": 1.379678224210617,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008239780304221966,
      "loss": 2.1322,
      "step": 16036
    },
    {
      "epoch": 1.3797642605179385,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008239574336227418,
      "loss": 1.2781,
      "step": 16037
    },
    {
      "epoch": 1.3798502968252602,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008239368358757746,
      "loss": 0.8656,
      "step": 16038
    },
    {
      "epoch": 1.379936333132582,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008239162371813551,
      "loss": 1.2703,
      "step": 16039
    },
    {
      "epoch": 1.3800223694399036,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008238956375395435,
      "loss": 1.5917,
      "step": 16040
    },
    {
      "epoch": 1.3801084057472253,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008238750369504003,
      "loss": 1.3294,
      "step": 16041
    },
    {
      "epoch": 1.380194442054547,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008238544354139854,
      "loss": 0.9113,
      "step": 16042
    },
    {
      "epoch": 1.3802804783618687,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008238338329303594,
      "loss": 1.4052,
      "step": 16043
    },
    {
      "epoch": 1.3803665146691904,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008238132294995824,
      "loss": 0.9213,
      "step": 16044
    },
    {
      "epoch": 1.380452550976512,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008237926251217148,
      "loss": 0.9037,
      "step": 16045
    },
    {
      "epoch": 1.3805385872838338,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008237720197968166,
      "loss": 1.6673,
      "step": 16046
    },
    {
      "epoch": 1.3806246235911555,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008237514135249482,
      "loss": 1.6327,
      "step": 16047
    },
    {
      "epoch": 1.380710659898477,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008237308063061699,
      "loss": 1.2595,
      "step": 16048
    },
    {
      "epoch": 1.380796696205799,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008237101981405419,
      "loss": 1.9728,
      "step": 16049
    },
    {
      "epoch": 1.3808827325131205,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008236895890281247,
      "loss": 0.8839,
      "step": 16050
    },
    {
      "epoch": 1.3809687688204422,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008236689789689784,
      "loss": 1.8691,
      "step": 16051
    },
    {
      "epoch": 1.3810548051277638,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008236483679631633,
      "loss": 1.094,
      "step": 16052
    },
    {
      "epoch": 1.3811408414350856,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008236277560107394,
      "loss": 1.0582,
      "step": 16053
    },
    {
      "epoch": 1.3812268777424073,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008236071431117673,
      "loss": 0.8668,
      "step": 16054
    },
    {
      "epoch": 1.3813129140497291,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008235865292663074,
      "loss": 1.7085,
      "step": 16055
    },
    {
      "epoch": 1.3813989503570507,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008235659144744199,
      "loss": 1.6066,
      "step": 16056
    },
    {
      "epoch": 1.3814849866643724,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0008235452987361649,
      "loss": 1.0643,
      "step": 16057
    },
    {
      "epoch": 1.381571022971694,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008235246820516029,
      "loss": 1.2883,
      "step": 16058
    },
    {
      "epoch": 1.3816570592790156,
      "grad_norm": 1.125,
      "learning_rate": 0.0008235040644207941,
      "loss": 1.6423,
      "step": 16059
    },
    {
      "epoch": 1.3817430955863375,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008234834458437989,
      "loss": 1.0641,
      "step": 16060
    },
    {
      "epoch": 1.381829131893659,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008234628263206774,
      "loss": 0.8062,
      "step": 16061
    },
    {
      "epoch": 1.381915168200981,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008234422058514901,
      "loss": 1.5345,
      "step": 16062
    },
    {
      "epoch": 1.3820012045083025,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008234215844362972,
      "loss": 1.7408,
      "step": 16063
    },
    {
      "epoch": 1.3820872408156242,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008234009620751591,
      "loss": 0.8053,
      "step": 16064
    },
    {
      "epoch": 1.3821732771229458,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008233803387681362,
      "loss": 1.1297,
      "step": 16065
    },
    {
      "epoch": 1.3822593134302676,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008233597145152886,
      "loss": 1.2484,
      "step": 16066
    },
    {
      "epoch": 1.3823453497375893,
      "grad_norm": 2.421875,
      "learning_rate": 0.0008233390893166767,
      "loss": 1.349,
      "step": 16067
    },
    {
      "epoch": 1.382431386044911,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008233184631723609,
      "loss": 0.7535,
      "step": 16068
    },
    {
      "epoch": 1.3825174223522327,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008232978360824015,
      "loss": 0.8449,
      "step": 16069
    },
    {
      "epoch": 1.3826034586595544,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008232772080468587,
      "loss": 1.1737,
      "step": 16070
    },
    {
      "epoch": 1.382689494966876,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008232565790657931,
      "loss": 1.082,
      "step": 16071
    },
    {
      "epoch": 1.3827755312741976,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008232359491392646,
      "loss": 2.1771,
      "step": 16072
    },
    {
      "epoch": 1.3828615675815195,
      "grad_norm": 1.625,
      "learning_rate": 0.0008232153182673341,
      "loss": 0.6552,
      "step": 16073
    },
    {
      "epoch": 1.382947603888841,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008231946864500615,
      "loss": 1.2264,
      "step": 16074
    },
    {
      "epoch": 1.3830336401961627,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008231740536875074,
      "loss": 0.8751,
      "step": 16075
    },
    {
      "epoch": 1.3831196765034846,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008231534199797321,
      "loss": 0.9724,
      "step": 16076
    },
    {
      "epoch": 1.3832057128108062,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008231327853267957,
      "loss": 2.0352,
      "step": 16077
    },
    {
      "epoch": 1.3832917491181278,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008231121497287588,
      "loss": 2.0348,
      "step": 16078
    },
    {
      "epoch": 1.3833777854254494,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008230915131856817,
      "loss": 1.7234,
      "step": 16079
    },
    {
      "epoch": 1.3834638217327713,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000823070875697625,
      "loss": 1.141,
      "step": 16080
    },
    {
      "epoch": 1.383549858040093,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008230502372646486,
      "loss": 1.5429,
      "step": 16081
    },
    {
      "epoch": 1.3836358943474147,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008230295978868132,
      "loss": 1.037,
      "step": 16082
    },
    {
      "epoch": 1.3837219306547364,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008230089575641788,
      "loss": 0.9691,
      "step": 16083
    },
    {
      "epoch": 1.383807966962058,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008229883162968063,
      "loss": 1.6271,
      "step": 16084
    },
    {
      "epoch": 1.3838940032693796,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008229676740847555,
      "loss": 1.7923,
      "step": 16085
    },
    {
      "epoch": 1.3839800395767012,
      "grad_norm": 1.875,
      "learning_rate": 0.0008229470309280874,
      "loss": 0.973,
      "step": 16086
    },
    {
      "epoch": 1.384066075884023,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000822926386826862,
      "loss": 0.6603,
      "step": 16087
    },
    {
      "epoch": 1.3841521121913447,
      "grad_norm": 1.75,
      "learning_rate": 0.0008229057417811395,
      "loss": 1.6242,
      "step": 16088
    },
    {
      "epoch": 1.3842381484986666,
      "grad_norm": 1.9375,
      "learning_rate": 0.0008228850957909807,
      "loss": 1.8168,
      "step": 16089
    },
    {
      "epoch": 1.3843241848059882,
      "grad_norm": 1.25,
      "learning_rate": 0.0008228644488564456,
      "loss": 1.4559,
      "step": 16090
    },
    {
      "epoch": 1.3844102211133098,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008228438009775949,
      "loss": 1.2456,
      "step": 16091
    },
    {
      "epoch": 1.3844962574206314,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008228231521544889,
      "loss": 1.4954,
      "step": 16092
    },
    {
      "epoch": 1.3845822937279533,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008228025023871879,
      "loss": 1.7623,
      "step": 16093
    },
    {
      "epoch": 1.384668330035275,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008227818516757525,
      "loss": 0.8079,
      "step": 16094
    },
    {
      "epoch": 1.3847543663425965,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008227612000202429,
      "loss": 0.9099,
      "step": 16095
    },
    {
      "epoch": 1.3848404026499184,
      "grad_norm": 1.25,
      "learning_rate": 0.0008227405474207195,
      "loss": 1.1634,
      "step": 16096
    },
    {
      "epoch": 1.38492643895724,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008227198938772428,
      "loss": 1.3075,
      "step": 16097
    },
    {
      "epoch": 1.3850124752645616,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008226992393898731,
      "loss": 2.094,
      "step": 16098
    },
    {
      "epoch": 1.3850985115718832,
      "grad_norm": 1.0,
      "learning_rate": 0.000822678583958671,
      "loss": 1.295,
      "step": 16099
    },
    {
      "epoch": 1.385184547879205,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008226579275836968,
      "loss": 0.8817,
      "step": 16100
    },
    {
      "epoch": 1.3852705841865267,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008226372702650109,
      "loss": 1.6812,
      "step": 16101
    },
    {
      "epoch": 1.3853566204938483,
      "grad_norm": 1.0,
      "learning_rate": 0.0008226166120026737,
      "loss": 1.2341,
      "step": 16102
    },
    {
      "epoch": 1.3854426568011702,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008225959527967455,
      "loss": 0.7609,
      "step": 16103
    },
    {
      "epoch": 1.3855286931084918,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0008225752926472871,
      "loss": 2.0027,
      "step": 16104
    },
    {
      "epoch": 1.3856147294158134,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008225546315543587,
      "loss": 1.1509,
      "step": 16105
    },
    {
      "epoch": 1.385700765723135,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008225339695180207,
      "loss": 1.9944,
      "step": 16106
    },
    {
      "epoch": 1.385786802030457,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008225133065383335,
      "loss": 1.4472,
      "step": 16107
    },
    {
      "epoch": 1.3858728383377785,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008224926426153577,
      "loss": 1.0054,
      "step": 16108
    },
    {
      "epoch": 1.3859588746451004,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008224719777491536,
      "loss": 0.841,
      "step": 16109
    },
    {
      "epoch": 1.386044910952422,
      "grad_norm": 4.5625,
      "learning_rate": 0.0008224513119397816,
      "loss": 1.6169,
      "step": 16110
    },
    {
      "epoch": 1.3861309472597436,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008224306451873022,
      "loss": 1.426,
      "step": 16111
    },
    {
      "epoch": 1.3862169835670652,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008224099774917762,
      "loss": 0.9612,
      "step": 16112
    },
    {
      "epoch": 1.3863030198743869,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008223893088532635,
      "loss": 1.0475,
      "step": 16113
    },
    {
      "epoch": 1.3863890561817087,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008223686392718247,
      "loss": 1.2178,
      "step": 16114
    },
    {
      "epoch": 1.3864750924890303,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008223479687475204,
      "loss": 1.5166,
      "step": 16115
    },
    {
      "epoch": 1.3865611287963522,
      "grad_norm": 1.3125,
      "learning_rate": 0.000822327297280411,
      "loss": 1.3126,
      "step": 16116
    },
    {
      "epoch": 1.3866471651036738,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008223066248705569,
      "loss": 1.0806,
      "step": 16117
    },
    {
      "epoch": 1.3867332014109954,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008222859515180186,
      "loss": 1.2129,
      "step": 16118
    },
    {
      "epoch": 1.386819237718317,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0008222652772228567,
      "loss": 1.5356,
      "step": 16119
    },
    {
      "epoch": 1.386905274025639,
      "grad_norm": 1.125,
      "learning_rate": 0.0008222446019851314,
      "loss": 0.9333,
      "step": 16120
    },
    {
      "epoch": 1.3869913103329605,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008222239258049033,
      "loss": 2.0798,
      "step": 16121
    },
    {
      "epoch": 1.3870773466402821,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008222032486822331,
      "loss": 0.9816,
      "step": 16122
    },
    {
      "epoch": 1.387163382947604,
      "grad_norm": 1.828125,
      "learning_rate": 0.0008221825706171809,
      "loss": 1.8559,
      "step": 16123
    },
    {
      "epoch": 1.3872494192549256,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008221618916098072,
      "loss": 0.9461,
      "step": 16124
    },
    {
      "epoch": 1.3873354555622472,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008221412116601729,
      "loss": 0.9676,
      "step": 16125
    },
    {
      "epoch": 1.3874214918695689,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008221205307683381,
      "loss": 0.6703,
      "step": 16126
    },
    {
      "epoch": 1.3875075281768907,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008220998489343634,
      "loss": 1.5402,
      "step": 16127
    },
    {
      "epoch": 1.3875935644842123,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008220791661583091,
      "loss": 0.8738,
      "step": 16128
    },
    {
      "epoch": 1.387679600791534,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008220584824402362,
      "loss": 1.4176,
      "step": 16129
    },
    {
      "epoch": 1.3877656370988558,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008220377977802047,
      "loss": 1.7634,
      "step": 16130
    },
    {
      "epoch": 1.3878516734061774,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0008220171121782753,
      "loss": 1.2675,
      "step": 16131
    },
    {
      "epoch": 1.387937709713499,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008219964256345085,
      "loss": 1.5186,
      "step": 16132
    },
    {
      "epoch": 1.3880237460208207,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008219757381489648,
      "loss": 1.172,
      "step": 16133
    },
    {
      "epoch": 1.3881097823281425,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008219550497217046,
      "loss": 1.0389,
      "step": 16134
    },
    {
      "epoch": 1.3881958186354642,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008219343603527885,
      "loss": 1.5422,
      "step": 16135
    },
    {
      "epoch": 1.3882818549427858,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008219136700422772,
      "loss": 0.7501,
      "step": 16136
    },
    {
      "epoch": 1.3883678912501076,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008218929787902307,
      "loss": 0.7345,
      "step": 16137
    },
    {
      "epoch": 1.3884539275574292,
      "grad_norm": 1.125,
      "learning_rate": 0.0008218722865967101,
      "loss": 1.5653,
      "step": 16138
    },
    {
      "epoch": 1.3885399638647509,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008218515934617756,
      "loss": 1.0959,
      "step": 16139
    },
    {
      "epoch": 1.3886260001720725,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008218308993854877,
      "loss": 1.753,
      "step": 16140
    },
    {
      "epoch": 1.3887120364793943,
      "grad_norm": 1.8828125,
      "learning_rate": 0.000821810204367907,
      "loss": 1.5442,
      "step": 16141
    },
    {
      "epoch": 1.388798072786716,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008217895084090942,
      "loss": 0.7547,
      "step": 16142
    },
    {
      "epoch": 1.3888841090940378,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008217688115091095,
      "loss": 1.8436,
      "step": 16143
    },
    {
      "epoch": 1.3889701454013594,
      "grad_norm": 1.125,
      "learning_rate": 0.0008217481136680139,
      "loss": 1.7721,
      "step": 16144
    },
    {
      "epoch": 1.389056181708681,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008217274148858674,
      "loss": 1.4049,
      "step": 16145
    },
    {
      "epoch": 1.3891422180160027,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008217067151627306,
      "loss": 1.7068,
      "step": 16146
    },
    {
      "epoch": 1.3892282543233243,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008216860144986645,
      "loss": 2.2947,
      "step": 16147
    },
    {
      "epoch": 1.3893142906306462,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008216653128937293,
      "loss": 0.9729,
      "step": 16148
    },
    {
      "epoch": 1.3894003269379678,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008216446103479856,
      "loss": 1.7848,
      "step": 16149
    },
    {
      "epoch": 1.3894863632452896,
      "grad_norm": 1.125,
      "learning_rate": 0.0008216239068614942,
      "loss": 1.4059,
      "step": 16150
    },
    {
      "epoch": 1.3895723995526112,
      "grad_norm": 0.890625,
      "learning_rate": 0.0008216032024343151,
      "loss": 1.2749,
      "step": 16151
    },
    {
      "epoch": 1.3896584358599329,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008215824970665093,
      "loss": 1.7111,
      "step": 16152
    },
    {
      "epoch": 1.3897444721672545,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008215617907581373,
      "loss": 1.7869,
      "step": 16153
    },
    {
      "epoch": 1.3898305084745763,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008215410835092596,
      "loss": 1.311,
      "step": 16154
    },
    {
      "epoch": 1.389916544781898,
      "grad_norm": 1.375,
      "learning_rate": 0.0008215203753199368,
      "loss": 1.5036,
      "step": 16155
    },
    {
      "epoch": 1.3900025810892196,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008214996661902292,
      "loss": 0.9568,
      "step": 16156
    },
    {
      "epoch": 1.3900886173965414,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008214789561201978,
      "loss": 0.881,
      "step": 16157
    },
    {
      "epoch": 1.390174653703863,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008214582451099029,
      "loss": 1.2917,
      "step": 16158
    },
    {
      "epoch": 1.3902606900111847,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008214375331594052,
      "loss": 0.971,
      "step": 16159
    },
    {
      "epoch": 1.3903467263185063,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008214168202687653,
      "loss": 1.1069,
      "step": 16160
    },
    {
      "epoch": 1.3904327626258282,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008213961064380437,
      "loss": 1.4645,
      "step": 16161
    },
    {
      "epoch": 1.3905187989331498,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008213753916673009,
      "loss": 1.4896,
      "step": 16162
    },
    {
      "epoch": 1.3906048352404714,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008213546759565976,
      "loss": 1.0293,
      "step": 16163
    },
    {
      "epoch": 1.3906908715477933,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008213339593059945,
      "loss": 1.0669,
      "step": 16164
    },
    {
      "epoch": 1.3907769078551149,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008213132417155518,
      "loss": 1.3558,
      "step": 16165
    },
    {
      "epoch": 1.3908629441624365,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008212925231853305,
      "loss": 0.9751,
      "step": 16166
    },
    {
      "epoch": 1.3909489804697581,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008212718037153912,
      "loss": 1.6825,
      "step": 16167
    },
    {
      "epoch": 1.39103501677708,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008212510833057943,
      "loss": 1.1256,
      "step": 16168
    },
    {
      "epoch": 1.3911210530844016,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008212303619566003,
      "loss": 1.392,
      "step": 16169
    },
    {
      "epoch": 1.3912070893917234,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008212096396678701,
      "loss": 0.6247,
      "step": 16170
    },
    {
      "epoch": 1.391293125699045,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008211889164396641,
      "loss": 0.9878,
      "step": 16171
    },
    {
      "epoch": 1.3913791620063667,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008211681922720431,
      "loss": 1.2643,
      "step": 16172
    },
    {
      "epoch": 1.3914651983136883,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008211474671650674,
      "loss": 1.0118,
      "step": 16173
    },
    {
      "epoch": 1.39155123462101,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008211267411187979,
      "loss": 0.6776,
      "step": 16174
    },
    {
      "epoch": 1.3916372709283318,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008211060141332953,
      "loss": 1.6172,
      "step": 16175
    },
    {
      "epoch": 1.3917233072356534,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008210852862086199,
      "loss": 0.9476,
      "step": 16176
    },
    {
      "epoch": 1.3918093435429753,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008210645573448325,
      "loss": 1.9247,
      "step": 16177
    },
    {
      "epoch": 1.3918953798502969,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008210438275419938,
      "loss": 1.7474,
      "step": 16178
    },
    {
      "epoch": 1.3919814161576185,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008210230968001643,
      "loss": 0.8054,
      "step": 16179
    },
    {
      "epoch": 1.3920674524649401,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008210023651194047,
      "loss": 0.7378,
      "step": 16180
    },
    {
      "epoch": 1.392153488772262,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008209816324997757,
      "loss": 1.3584,
      "step": 16181
    },
    {
      "epoch": 1.3922395250795836,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008209608989413375,
      "loss": 1.8513,
      "step": 16182
    },
    {
      "epoch": 1.3923255613869052,
      "grad_norm": 1.125,
      "learning_rate": 0.0008209401644441515,
      "loss": 1.2668,
      "step": 16183
    },
    {
      "epoch": 1.392411597694227,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008209194290082775,
      "loss": 1.7076,
      "step": 16184
    },
    {
      "epoch": 1.3924976340015487,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000820898692633777,
      "loss": 1.1485,
      "step": 16185
    },
    {
      "epoch": 1.3925836703088703,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008208779553207099,
      "loss": 1.29,
      "step": 16186
    },
    {
      "epoch": 1.392669706616192,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008208572170691375,
      "loss": 0.972,
      "step": 16187
    },
    {
      "epoch": 1.3927557429235138,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00082083647787912,
      "loss": 1.6743,
      "step": 16188
    },
    {
      "epoch": 1.3928417792308354,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008208157377507182,
      "loss": 0.779,
      "step": 16189
    },
    {
      "epoch": 1.392927815538157,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008207949966839928,
      "loss": 1.3855,
      "step": 16190
    },
    {
      "epoch": 1.3930138518454789,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008207742546790045,
      "loss": 1.0224,
      "step": 16191
    },
    {
      "epoch": 1.3930998881528005,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008207535117358138,
      "loss": 1.6518,
      "step": 16192
    },
    {
      "epoch": 1.3931859244601221,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008207327678544814,
      "loss": 1.3854,
      "step": 16193
    },
    {
      "epoch": 1.3932719607674438,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008207120230350682,
      "loss": 1.4713,
      "step": 16194
    },
    {
      "epoch": 1.3933579970747656,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008206912772776345,
      "loss": 1.354,
      "step": 16195
    },
    {
      "epoch": 1.3934440333820872,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008206705305822412,
      "loss": 1.2485,
      "step": 16196
    },
    {
      "epoch": 1.393530069689409,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008206497829489492,
      "loss": 1.4085,
      "step": 16197
    },
    {
      "epoch": 1.3936161059967307,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008206290343778186,
      "loss": 0.9133,
      "step": 16198
    },
    {
      "epoch": 1.3937021423040523,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008206082848689107,
      "loss": 0.7975,
      "step": 16199
    },
    {
      "epoch": 1.393788178611374,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000820587534422286,
      "loss": 1.2958,
      "step": 16200
    },
    {
      "epoch": 1.3938742149186956,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008205667830380049,
      "loss": 1.0928,
      "step": 16201
    },
    {
      "epoch": 1.3939602512260174,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008205460307161283,
      "loss": 1.109,
      "step": 16202
    },
    {
      "epoch": 1.394046287533339,
      "grad_norm": 1.1875,
      "learning_rate": 0.000820525277456717,
      "loss": 1.1231,
      "step": 16203
    },
    {
      "epoch": 1.3941323238406609,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008205045232598316,
      "loss": 1.0437,
      "step": 16204
    },
    {
      "epoch": 1.3942183601479825,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008204837681255328,
      "loss": 1.3626,
      "step": 16205
    },
    {
      "epoch": 1.3943043964553041,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008204630120538812,
      "loss": 1.1036,
      "step": 16206
    },
    {
      "epoch": 1.3943904327626258,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008204422550449378,
      "loss": 1.8357,
      "step": 16207
    },
    {
      "epoch": 1.3944764690699476,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008204214970987629,
      "loss": 1.7394,
      "step": 16208
    },
    {
      "epoch": 1.3945625053772692,
      "grad_norm": 2.015625,
      "learning_rate": 0.0008204007382154177,
      "loss": 1.8167,
      "step": 16209
    },
    {
      "epoch": 1.3946485416845908,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008203799783949625,
      "loss": 0.6567,
      "step": 16210
    },
    {
      "epoch": 1.3947345779919127,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008203592176374582,
      "loss": 1.6867,
      "step": 16211
    },
    {
      "epoch": 1.3948206142992343,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008203384559429654,
      "loss": 1.3959,
      "step": 16212
    },
    {
      "epoch": 1.394906650606556,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008203176933115451,
      "loss": 1.2594,
      "step": 16213
    },
    {
      "epoch": 1.3949926869138776,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008202969297432577,
      "loss": 1.3467,
      "step": 16214
    },
    {
      "epoch": 1.3950787232211994,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0008202761652381642,
      "loss": 1.1549,
      "step": 16215
    },
    {
      "epoch": 1.395164759528521,
      "grad_norm": 1.90625,
      "learning_rate": 0.0008202553997963252,
      "loss": 1.5417,
      "step": 16216
    },
    {
      "epoch": 1.3952507958358427,
      "grad_norm": 2.328125,
      "learning_rate": 0.0008202346334178014,
      "loss": 1.6997,
      "step": 16217
    },
    {
      "epoch": 1.3953368321431645,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008202138661026534,
      "loss": 1.6186,
      "step": 16218
    },
    {
      "epoch": 1.3954228684504861,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008201930978509424,
      "loss": 1.5261,
      "step": 16219
    },
    {
      "epoch": 1.3955089047578078,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008201723286627287,
      "loss": 1.0416,
      "step": 16220
    },
    {
      "epoch": 1.3955949410651294,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008201515585380732,
      "loss": 1.1717,
      "step": 16221
    },
    {
      "epoch": 1.3956809773724512,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008201307874770368,
      "loss": 1.293,
      "step": 16222
    },
    {
      "epoch": 1.3957670136797728,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008201100154796799,
      "loss": 0.809,
      "step": 16223
    },
    {
      "epoch": 1.3958530499870945,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008200892425460636,
      "loss": 2.037,
      "step": 16224
    },
    {
      "epoch": 1.3959390862944163,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008200684686762485,
      "loss": 0.9527,
      "step": 16225
    },
    {
      "epoch": 1.396025122601738,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008200476938702954,
      "loss": 0.8299,
      "step": 16226
    },
    {
      "epoch": 1.3961111589090596,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008200269181282651,
      "loss": 1.0046,
      "step": 16227
    },
    {
      "epoch": 1.3961971952163812,
      "grad_norm": 1.5,
      "learning_rate": 0.0008200061414502183,
      "loss": 1.4416,
      "step": 16228
    },
    {
      "epoch": 1.396283231523703,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008199853638362156,
      "loss": 1.5405,
      "step": 16229
    },
    {
      "epoch": 1.3963692678310247,
      "grad_norm": 2.203125,
      "learning_rate": 0.000819964585286318,
      "loss": 2.1645,
      "step": 16230
    },
    {
      "epoch": 1.3964553041383465,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008199438058005864,
      "loss": 1.6846,
      "step": 16231
    },
    {
      "epoch": 1.3965413404456681,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008199230253790812,
      "loss": 1.156,
      "step": 16232
    },
    {
      "epoch": 1.3966273767529898,
      "grad_norm": 1.125,
      "learning_rate": 0.0008199022440218634,
      "loss": 1.1822,
      "step": 16233
    },
    {
      "epoch": 1.3967134130603114,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008198814617289939,
      "loss": 0.7117,
      "step": 16234
    },
    {
      "epoch": 1.396799449367633,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008198606785005332,
      "loss": 1.2182,
      "step": 16235
    },
    {
      "epoch": 1.3968854856749549,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008198398943365422,
      "loss": 0.943,
      "step": 16236
    },
    {
      "epoch": 1.3969715219822765,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008198191092370817,
      "loss": 1.3085,
      "step": 16237
    },
    {
      "epoch": 1.3970575582895983,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008197983232022126,
      "loss": 1.5036,
      "step": 16238
    },
    {
      "epoch": 1.39714359459692,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008197775362319955,
      "loss": 0.6715,
      "step": 16239
    },
    {
      "epoch": 1.3972296309042416,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008197567483264914,
      "loss": 1.2794,
      "step": 16240
    },
    {
      "epoch": 1.3973156672115632,
      "grad_norm": 0.984375,
      "learning_rate": 0.000819735959485761,
      "loss": 0.9149,
      "step": 16241
    },
    {
      "epoch": 1.397401703518885,
      "grad_norm": 1.28125,
      "learning_rate": 0.000819715169709865,
      "loss": 1.2989,
      "step": 16242
    },
    {
      "epoch": 1.3974877398262067,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008196943789988645,
      "loss": 2.2349,
      "step": 16243
    },
    {
      "epoch": 1.3975737761335283,
      "grad_norm": 1.140625,
      "learning_rate": 0.00081967358735282,
      "loss": 1.342,
      "step": 16244
    },
    {
      "epoch": 1.3976598124408501,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008196527947717924,
      "loss": 1.795,
      "step": 16245
    },
    {
      "epoch": 1.3977458487481718,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008196320012558426,
      "loss": 1.0182,
      "step": 16246
    },
    {
      "epoch": 1.3978318850554934,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008196112068050313,
      "loss": 1.3932,
      "step": 16247
    },
    {
      "epoch": 1.397917921362815,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008195904114194195,
      "loss": 0.9789,
      "step": 16248
    },
    {
      "epoch": 1.3980039576701369,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008195696150990677,
      "loss": 1.1148,
      "step": 16249
    },
    {
      "epoch": 1.3980899939774585,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008195488178440371,
      "loss": 1.2704,
      "step": 16250
    },
    {
      "epoch": 1.39817603028478,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008195280196543884,
      "loss": 1.238,
      "step": 16251
    },
    {
      "epoch": 1.398262066592102,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008195072205301822,
      "loss": 1.7372,
      "step": 16252
    },
    {
      "epoch": 1.3983481028994236,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008194864204714796,
      "loss": 1.3217,
      "step": 16253
    },
    {
      "epoch": 1.3984341392067452,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0008194656194783415,
      "loss": 1.647,
      "step": 16254
    },
    {
      "epoch": 1.3985201755140668,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008194448175508285,
      "loss": 1.4942,
      "step": 16255
    },
    {
      "epoch": 1.3986062118213887,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008194240146890013,
      "loss": 0.9383,
      "step": 16256
    },
    {
      "epoch": 1.3986922481287103,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008194032108929212,
      "loss": 0.8201,
      "step": 16257
    },
    {
      "epoch": 1.3987782844360321,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008193824061626489,
      "loss": 1.6449,
      "step": 16258
    },
    {
      "epoch": 1.3988643207433538,
      "grad_norm": 1.265625,
      "learning_rate": 0.000819361600498245,
      "loss": 1.4756,
      "step": 16259
    },
    {
      "epoch": 1.3989503570506754,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008193407938997707,
      "loss": 0.9975,
      "step": 16260
    },
    {
      "epoch": 1.399036393357997,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008193199863672866,
      "loss": 1.1308,
      "step": 16261
    },
    {
      "epoch": 1.3991224296653186,
      "grad_norm": 1.0,
      "learning_rate": 0.0008192991779008537,
      "loss": 1.2859,
      "step": 16262
    },
    {
      "epoch": 1.3992084659726405,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008192783685005327,
      "loss": 1.4363,
      "step": 16263
    },
    {
      "epoch": 1.399294502279962,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008192575581663846,
      "loss": 1.6255,
      "step": 16264
    },
    {
      "epoch": 1.399380538587284,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008192367468984703,
      "loss": 1.4381,
      "step": 16265
    },
    {
      "epoch": 1.3994665748946056,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008192159346968506,
      "loss": 1.1332,
      "step": 16266
    },
    {
      "epoch": 1.3995526112019272,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008191951215615863,
      "loss": 2.2433,
      "step": 16267
    },
    {
      "epoch": 1.3996386475092488,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008191743074927385,
      "loss": 1.2263,
      "step": 16268
    },
    {
      "epoch": 1.3997246838165707,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008191534924903677,
      "loss": 0.8123,
      "step": 16269
    },
    {
      "epoch": 1.3998107201238923,
      "grad_norm": 1.25,
      "learning_rate": 0.0008191326765545351,
      "loss": 0.9348,
      "step": 16270
    },
    {
      "epoch": 1.399896756431214,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008191118596853015,
      "loss": 1.0009,
      "step": 16271
    },
    {
      "epoch": 1.3999827927385358,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008190910418827278,
      "loss": 0.9,
      "step": 16272
    },
    {
      "epoch": 1.4000688290458574,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000819070223146875,
      "loss": 1.6915,
      "step": 16273
    },
    {
      "epoch": 1.400154865353179,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008190494034778035,
      "loss": 1.838,
      "step": 16274
    },
    {
      "epoch": 1.4002409016605006,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008190285828755748,
      "loss": 1.4616,
      "step": 16275
    },
    {
      "epoch": 1.4003269379678225,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008190077613402496,
      "loss": 1.4702,
      "step": 16276
    },
    {
      "epoch": 1.400412974275144,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008189869388718885,
      "loss": 1.6137,
      "step": 16277
    },
    {
      "epoch": 1.4004990105824657,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008189661154705528,
      "loss": 1.3057,
      "step": 16278
    },
    {
      "epoch": 1.4005850468897876,
      "grad_norm": 1.765625,
      "learning_rate": 0.0008189452911363032,
      "loss": 1.249,
      "step": 16279
    },
    {
      "epoch": 1.4006710831971092,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008189244658692008,
      "loss": 1.2615,
      "step": 16280
    },
    {
      "epoch": 1.4007571195044308,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008189036396693062,
      "loss": 1.9133,
      "step": 16281
    },
    {
      "epoch": 1.4008431558117524,
      "grad_norm": 1.125,
      "learning_rate": 0.0008188828125366805,
      "loss": 1.7918,
      "step": 16282
    },
    {
      "epoch": 1.4009291921190743,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008188619844713846,
      "loss": 1.6188,
      "step": 16283
    },
    {
      "epoch": 1.401015228426396,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008188411554734795,
      "loss": 1.3427,
      "step": 16284
    },
    {
      "epoch": 1.4011012647337178,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008188203255430259,
      "loss": 1.2096,
      "step": 16285
    },
    {
      "epoch": 1.4011873010410394,
      "grad_norm": 1.625,
      "learning_rate": 0.0008187994946800849,
      "loss": 0.9284,
      "step": 16286
    },
    {
      "epoch": 1.401273337348361,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008187786628847173,
      "loss": 1.6295,
      "step": 16287
    },
    {
      "epoch": 1.4013593736556826,
      "grad_norm": 1.25,
      "learning_rate": 0.0008187578301569842,
      "loss": 2.0781,
      "step": 16288
    },
    {
      "epoch": 1.4014454099630043,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008187369964969465,
      "loss": 1.3126,
      "step": 16289
    },
    {
      "epoch": 1.401531446270326,
      "grad_norm": 25.75,
      "learning_rate": 0.0008187161619046651,
      "loss": 1.669,
      "step": 16290
    },
    {
      "epoch": 1.4016174825776477,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008186953263802008,
      "loss": 2.014,
      "step": 16291
    },
    {
      "epoch": 1.4017035188849696,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008186744899236147,
      "loss": 1.1065,
      "step": 16292
    },
    {
      "epoch": 1.4017895551922912,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008186536525349677,
      "loss": 1.3415,
      "step": 16293
    },
    {
      "epoch": 1.4018755914996128,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008186328142143207,
      "loss": 1.3109,
      "step": 16294
    },
    {
      "epoch": 1.4019616278069345,
      "grad_norm": 1.625,
      "learning_rate": 0.0008186119749617346,
      "loss": 1.1384,
      "step": 16295
    },
    {
      "epoch": 1.4020476641142563,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008185911347772705,
      "loss": 1.7391,
      "step": 16296
    },
    {
      "epoch": 1.402133700421578,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008185702936609895,
      "loss": 1.3549,
      "step": 16297
    },
    {
      "epoch": 1.4022197367288995,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008185494516129521,
      "loss": 1.0375,
      "step": 16298
    },
    {
      "epoch": 1.4023057730362214,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008185286086332198,
      "loss": 1.3631,
      "step": 16299
    },
    {
      "epoch": 1.402391809343543,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008185077647218529,
      "loss": 2.1236,
      "step": 16300
    },
    {
      "epoch": 1.4024778456508646,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000818486919878913,
      "loss": 1.3328,
      "step": 16301
    },
    {
      "epoch": 1.4025638819581863,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008184660741044607,
      "loss": 1.1715,
      "step": 16302
    },
    {
      "epoch": 1.4026499182655081,
      "grad_norm": 1.171875,
      "learning_rate": 0.000818445227398557,
      "loss": 2.0555,
      "step": 16303
    },
    {
      "epoch": 1.4027359545728297,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008184243797612631,
      "loss": 1.5684,
      "step": 16304
    },
    {
      "epoch": 1.4028219908801514,
      "grad_norm": 1.703125,
      "learning_rate": 0.0008184035311926397,
      "loss": 1.807,
      "step": 16305
    },
    {
      "epoch": 1.4029080271874732,
      "grad_norm": 1.109375,
      "learning_rate": 0.000818382681692748,
      "loss": 1.2689,
      "step": 16306
    },
    {
      "epoch": 1.4029940634947948,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008183618312616488,
      "loss": 1.1137,
      "step": 16307
    },
    {
      "epoch": 1.4030800998021165,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008183409798994032,
      "loss": 1.086,
      "step": 16308
    },
    {
      "epoch": 1.403166136109438,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008183201276060721,
      "loss": 1.8418,
      "step": 16309
    },
    {
      "epoch": 1.40325217241676,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008182992743817165,
      "loss": 1.3515,
      "step": 16310
    },
    {
      "epoch": 1.4033382087240815,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008182784202263975,
      "loss": 1.995,
      "step": 16311
    },
    {
      "epoch": 1.4034242450314032,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008182575651401762,
      "loss": 1.604,
      "step": 16312
    },
    {
      "epoch": 1.403510281338725,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008182367091231131,
      "loss": 1.1847,
      "step": 16313
    },
    {
      "epoch": 1.4035963176460466,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008182158521752697,
      "loss": 1.0238,
      "step": 16314
    },
    {
      "epoch": 1.4036823539533683,
      "grad_norm": 1.84375,
      "learning_rate": 0.0008181949942967068,
      "loss": 1.1534,
      "step": 16315
    },
    {
      "epoch": 1.40376839026069,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008181741354874855,
      "loss": 1.0302,
      "step": 16316
    },
    {
      "epoch": 1.4038544265680117,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008181532757476664,
      "loss": 1.5513,
      "step": 16317
    },
    {
      "epoch": 1.4039404628753334,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008181324150773111,
      "loss": 0.765,
      "step": 16318
    },
    {
      "epoch": 1.4040264991826552,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008181115534764805,
      "loss": 0.6783,
      "step": 16319
    },
    {
      "epoch": 1.4041125354899768,
      "grad_norm": 1.25,
      "learning_rate": 0.0008180906909452352,
      "loss": 1.3965,
      "step": 16320
    },
    {
      "epoch": 1.4041985717972985,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008180698274836367,
      "loss": 1.7146,
      "step": 16321
    },
    {
      "epoch": 1.40428460810462,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008180489630917457,
      "loss": 1.6472,
      "step": 16322
    },
    {
      "epoch": 1.4043706444119417,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008180280977696234,
      "loss": 1.0406,
      "step": 16323
    },
    {
      "epoch": 1.4044566807192636,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008180072315173307,
      "loss": 1.2049,
      "step": 16324
    },
    {
      "epoch": 1.4045427170265852,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008179863643349288,
      "loss": 1.022,
      "step": 16325
    },
    {
      "epoch": 1.404628753333907,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008179654962224785,
      "loss": 1.3557,
      "step": 16326
    },
    {
      "epoch": 1.4047147896412286,
      "grad_norm": 1.46875,
      "learning_rate": 0.000817944627180041,
      "loss": 1.1903,
      "step": 16327
    },
    {
      "epoch": 1.4048008259485503,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008179237572076774,
      "loss": 1.0055,
      "step": 16328
    },
    {
      "epoch": 1.404886862255872,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008179028863054485,
      "loss": 1.9719,
      "step": 16329
    },
    {
      "epoch": 1.4049728985631937,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008178820144734155,
      "loss": 1.3654,
      "step": 16330
    },
    {
      "epoch": 1.4050589348705154,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008178611417116395,
      "loss": 1.6218,
      "step": 16331
    },
    {
      "epoch": 1.405144971177837,
      "grad_norm": 1.625,
      "learning_rate": 0.0008178402680201815,
      "loss": 1.6391,
      "step": 16332
    },
    {
      "epoch": 1.4052310074851588,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008178193933991025,
      "loss": 1.4318,
      "step": 16333
    },
    {
      "epoch": 1.4053170437924805,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008177985178484637,
      "loss": 1.8503,
      "step": 16334
    },
    {
      "epoch": 1.405403080099802,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008177776413683257,
      "loss": 0.6081,
      "step": 16335
    },
    {
      "epoch": 1.4054891164071237,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008177567639587503,
      "loss": 1.1579,
      "step": 16336
    },
    {
      "epoch": 1.4055751527144456,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008177358856197981,
      "loss": 2.157,
      "step": 16337
    },
    {
      "epoch": 1.4056611890217672,
      "grad_norm": 3.234375,
      "learning_rate": 0.0008177150063515301,
      "loss": 1.2519,
      "step": 16338
    },
    {
      "epoch": 1.4057472253290888,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008176941261540074,
      "loss": 1.7959,
      "step": 16339
    },
    {
      "epoch": 1.4058332616364106,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008176732450272914,
      "loss": 1.2697,
      "step": 16340
    },
    {
      "epoch": 1.4059192979437323,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008176523629714428,
      "loss": 0.7173,
      "step": 16341
    },
    {
      "epoch": 1.406005334251054,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008176314799865228,
      "loss": 1.5039,
      "step": 16342
    },
    {
      "epoch": 1.4060913705583755,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008176105960725927,
      "loss": 2.1783,
      "step": 16343
    },
    {
      "epoch": 1.4061774068656974,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008175897112297132,
      "loss": 1.4762,
      "step": 16344
    },
    {
      "epoch": 1.406263443173019,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008175688254579457,
      "loss": 0.7371,
      "step": 16345
    },
    {
      "epoch": 1.4063494794803408,
      "grad_norm": 0.953125,
      "learning_rate": 0.000817547938757351,
      "loss": 1.8009,
      "step": 16346
    },
    {
      "epoch": 1.4064355157876625,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008175270511279904,
      "loss": 1.5522,
      "step": 16347
    },
    {
      "epoch": 1.406521552094984,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008175061625699249,
      "loss": 2.2262,
      "step": 16348
    },
    {
      "epoch": 1.4066075884023057,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008174852730832156,
      "loss": 1.4274,
      "step": 16349
    },
    {
      "epoch": 1.4066936247096273,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008174643826679236,
      "loss": 1.5702,
      "step": 16350
    },
    {
      "epoch": 1.4067796610169492,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008174434913241102,
      "loss": 1.6005,
      "step": 16351
    },
    {
      "epoch": 1.4068656973242708,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008174225990518362,
      "loss": 0.8877,
      "step": 16352
    },
    {
      "epoch": 1.4069517336315926,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0008174017058511629,
      "loss": 1.102,
      "step": 16353
    },
    {
      "epoch": 1.4070377699389143,
      "grad_norm": 1.25,
      "learning_rate": 0.0008173808117221513,
      "loss": 1.3763,
      "step": 16354
    },
    {
      "epoch": 1.407123806246236,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008173599166648624,
      "loss": 1.4929,
      "step": 16355
    },
    {
      "epoch": 1.4072098425535575,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008173390206793576,
      "loss": 0.8701,
      "step": 16356
    },
    {
      "epoch": 1.4072958788608794,
      "grad_norm": 1.421875,
      "learning_rate": 0.000817318123765698,
      "loss": 1.077,
      "step": 16357
    },
    {
      "epoch": 1.407381915168201,
      "grad_norm": 1.921875,
      "learning_rate": 0.0008172972259239445,
      "loss": 1.3375,
      "step": 16358
    },
    {
      "epoch": 1.4074679514755226,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008172763271541584,
      "loss": 0.9966,
      "step": 16359
    },
    {
      "epoch": 1.4075539877828445,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008172554274564006,
      "loss": 2.2318,
      "step": 16360
    },
    {
      "epoch": 1.407640024090166,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008172345268307324,
      "loss": 1.6051,
      "step": 16361
    },
    {
      "epoch": 1.4077260603974877,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008172136252772151,
      "loss": 1.0009,
      "step": 16362
    },
    {
      "epoch": 1.4078120967048093,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0008171927227959095,
      "loss": 1.5603,
      "step": 16363
    },
    {
      "epoch": 1.4078981330121312,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008171718193868769,
      "loss": 1.1131,
      "step": 16364
    },
    {
      "epoch": 1.4079841693194528,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008171509150501783,
      "loss": 0.7891,
      "step": 16365
    },
    {
      "epoch": 1.4080702056267744,
      "grad_norm": 1.140625,
      "learning_rate": 0.000817130009785875,
      "loss": 0.71,
      "step": 16366
    },
    {
      "epoch": 1.4081562419340963,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008171091035940282,
      "loss": 1.242,
      "step": 16367
    },
    {
      "epoch": 1.408242278241418,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008170881964746989,
      "loss": 1.4592,
      "step": 16368
    },
    {
      "epoch": 1.4083283145487395,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008170672884279482,
      "loss": 1.0588,
      "step": 16369
    },
    {
      "epoch": 1.4084143508560611,
      "grad_norm": 1.125,
      "learning_rate": 0.0008170463794538376,
      "loss": 0.8823,
      "step": 16370
    },
    {
      "epoch": 1.408500387163383,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008170254695524277,
      "loss": 1.7138,
      "step": 16371
    },
    {
      "epoch": 1.4085864234707046,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008170045587237802,
      "loss": 1.3885,
      "step": 16372
    },
    {
      "epoch": 1.4086724597780265,
      "grad_norm": 1.703125,
      "learning_rate": 0.000816983646967956,
      "loss": 1.57,
      "step": 16373
    },
    {
      "epoch": 1.408758496085348,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008169627342850161,
      "loss": 0.9105,
      "step": 16374
    },
    {
      "epoch": 1.4088445323926697,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008169418206750219,
      "loss": 1.2446,
      "step": 16375
    },
    {
      "epoch": 1.4089305686999913,
      "grad_norm": 1.125,
      "learning_rate": 0.0008169209061380346,
      "loss": 1.3545,
      "step": 16376
    },
    {
      "epoch": 1.409016605007313,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008168999906741153,
      "loss": 1.8741,
      "step": 16377
    },
    {
      "epoch": 1.4091026413146348,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008168790742833251,
      "loss": 0.9957,
      "step": 16378
    },
    {
      "epoch": 1.4091886776219564,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008168581569657252,
      "loss": 1.8197,
      "step": 16379
    },
    {
      "epoch": 1.4092747139292783,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008168372387213769,
      "loss": 1.0494,
      "step": 16380
    },
    {
      "epoch": 1.4093607502366,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0008168163195503413,
      "loss": 0.6007,
      "step": 16381
    },
    {
      "epoch": 1.4094467865439215,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008167953994526795,
      "loss": 1.5724,
      "step": 16382
    },
    {
      "epoch": 1.4095328228512432,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008167744784284529,
      "loss": 1.0974,
      "step": 16383
    },
    {
      "epoch": 1.409618859158565,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008167535564777224,
      "loss": 1.9732,
      "step": 16384
    },
    {
      "epoch": 1.4097048954658866,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008167326336005494,
      "loss": 1.3101,
      "step": 16385
    },
    {
      "epoch": 1.4097909317732082,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008167117097969952,
      "loss": 1.615,
      "step": 16386
    },
    {
      "epoch": 1.40987696808053,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008166907850671206,
      "loss": 0.9715,
      "step": 16387
    },
    {
      "epoch": 1.4099630043878517,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008166698594109872,
      "loss": 1.4729,
      "step": 16388
    },
    {
      "epoch": 1.4100490406951733,
      "grad_norm": 1.53125,
      "learning_rate": 0.000816648932828656,
      "loss": 1.3763,
      "step": 16389
    },
    {
      "epoch": 1.410135077002495,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008166280053201883,
      "loss": 1.0377,
      "step": 16390
    },
    {
      "epoch": 1.4102211133098168,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008166070768856453,
      "loss": 1.052,
      "step": 16391
    },
    {
      "epoch": 1.4103071496171384,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0008165861475250881,
      "loss": 1.8184,
      "step": 16392
    },
    {
      "epoch": 1.41039318592446,
      "grad_norm": 1.0625,
      "learning_rate": 0.000816565217238578,
      "loss": 1.898,
      "step": 16393
    },
    {
      "epoch": 1.410479222231782,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008165442860261763,
      "loss": 1.3468,
      "step": 16394
    },
    {
      "epoch": 1.4105652585391035,
      "grad_norm": 1.25,
      "learning_rate": 0.000816523353887944,
      "loss": 1.0011,
      "step": 16395
    },
    {
      "epoch": 1.4106512948464252,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008165024208239424,
      "loss": 1.0214,
      "step": 16396
    },
    {
      "epoch": 1.4107373311537468,
      "grad_norm": 1.265625,
      "learning_rate": 0.000816481486834233,
      "loss": 1.3208,
      "step": 16397
    },
    {
      "epoch": 1.4108233674610686,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008164605519188766,
      "loss": 1.0837,
      "step": 16398
    },
    {
      "epoch": 1.4109094037683902,
      "grad_norm": 1.859375,
      "learning_rate": 0.0008164396160779348,
      "loss": 1.3751,
      "step": 16399
    },
    {
      "epoch": 1.4109954400757119,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008164186793114685,
      "loss": 0.9613,
      "step": 16400
    },
    {
      "epoch": 1.4110814763830337,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008163977416195392,
      "loss": 0.6536,
      "step": 16401
    },
    {
      "epoch": 1.4111675126903553,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008163768030022081,
      "loss": 1.6592,
      "step": 16402
    },
    {
      "epoch": 1.411253548997677,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008163558634595362,
      "loss": 1.5194,
      "step": 16403
    },
    {
      "epoch": 1.4113395853049986,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008163349229915851,
      "loss": 1.9635,
      "step": 16404
    },
    {
      "epoch": 1.4114256216123204,
      "grad_norm": 2.0,
      "learning_rate": 0.0008163139815984158,
      "loss": 1.3786,
      "step": 16405
    },
    {
      "epoch": 1.411511657919642,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008162930392800895,
      "loss": 0.6701,
      "step": 16406
    },
    {
      "epoch": 1.411597694226964,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008162720960366678,
      "loss": 1.292,
      "step": 16407
    },
    {
      "epoch": 1.4116837305342855,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008162511518682116,
      "loss": 1.1405,
      "step": 16408
    },
    {
      "epoch": 1.4117697668416072,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008162302067747824,
      "loss": 1.2033,
      "step": 16409
    },
    {
      "epoch": 1.4118558031489288,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008162092607564412,
      "loss": 1.0172,
      "step": 16410
    },
    {
      "epoch": 1.4119418394562504,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008161883138132495,
      "loss": 1.4038,
      "step": 16411
    },
    {
      "epoch": 1.4120278757635722,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008161673659452685,
      "loss": 1.1452,
      "step": 16412
    },
    {
      "epoch": 1.4121139120708939,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008161464171525595,
      "loss": 1.3476,
      "step": 16413
    },
    {
      "epoch": 1.4121999483782157,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008161254674351835,
      "loss": 0.7857,
      "step": 16414
    },
    {
      "epoch": 1.4122859846855373,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008161045167932022,
      "loss": 0.9981,
      "step": 16415
    },
    {
      "epoch": 1.412372020992859,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008160835652266764,
      "loss": 1.5432,
      "step": 16416
    },
    {
      "epoch": 1.4124580573001806,
      "grad_norm": 1.0625,
      "learning_rate": 0.000816062612735668,
      "loss": 1.3014,
      "step": 16417
    },
    {
      "epoch": 1.4125440936075024,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008160416593202378,
      "loss": 1.6329,
      "step": 16418
    },
    {
      "epoch": 1.412630129914824,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008160207049804472,
      "loss": 1.4555,
      "step": 16419
    },
    {
      "epoch": 1.4127161662221457,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008159997497163573,
      "loss": 1.6506,
      "step": 16420
    },
    {
      "epoch": 1.4128022025294675,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008159787935280298,
      "loss": 0.8017,
      "step": 16421
    },
    {
      "epoch": 1.4128882388367892,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008159578364155258,
      "loss": 2.118,
      "step": 16422
    },
    {
      "epoch": 1.4129742751441108,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008159368783789065,
      "loss": 1.1977,
      "step": 16423
    },
    {
      "epoch": 1.4130603114514324,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008159159194182332,
      "loss": 1.5036,
      "step": 16424
    },
    {
      "epoch": 1.4131463477587543,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008158949595335674,
      "loss": 1.4337,
      "step": 16425
    },
    {
      "epoch": 1.4132323840660759,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008158739987249704,
      "loss": 1.2371,
      "step": 16426
    },
    {
      "epoch": 1.4133184203733975,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008158530369925031,
      "loss": 1.2601,
      "step": 16427
    },
    {
      "epoch": 1.4134044566807193,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008158320743362273,
      "loss": 1.4341,
      "step": 16428
    },
    {
      "epoch": 1.413490492988041,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008158111107562041,
      "loss": 1.6415,
      "step": 16429
    },
    {
      "epoch": 1.4135765292953626,
      "grad_norm": 1.890625,
      "learning_rate": 0.0008157901462524948,
      "loss": 1.8285,
      "step": 16430
    },
    {
      "epoch": 1.4136625656026842,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008157691808251607,
      "loss": 1.6,
      "step": 16431
    },
    {
      "epoch": 1.413748601910006,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008157482144742631,
      "loss": 1.0413,
      "step": 16432
    },
    {
      "epoch": 1.4138346382173277,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008157272471998636,
      "loss": 1.5908,
      "step": 16433
    },
    {
      "epoch": 1.4139206745246495,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008157062790020231,
      "loss": 1.7039,
      "step": 16434
    },
    {
      "epoch": 1.4140067108319712,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008156853098808032,
      "loss": 1.2375,
      "step": 16435
    },
    {
      "epoch": 1.4140927471392928,
      "grad_norm": 4.0625,
      "learning_rate": 0.000815664339836265,
      "loss": 1.421,
      "step": 16436
    },
    {
      "epoch": 1.4141787834466144,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008156433688684704,
      "loss": 0.7428,
      "step": 16437
    },
    {
      "epoch": 1.414264819753936,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008156223969774801,
      "loss": 1.0065,
      "step": 16438
    },
    {
      "epoch": 1.4143508560612579,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008156014241633556,
      "loss": 1.5738,
      "step": 16439
    },
    {
      "epoch": 1.4144368923685795,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008155804504261584,
      "loss": 1.5799,
      "step": 16440
    },
    {
      "epoch": 1.4145229286759013,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008155594757659497,
      "loss": 1.4072,
      "step": 16441
    },
    {
      "epoch": 1.414608964983223,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000815538500182791,
      "loss": 1.039,
      "step": 16442
    },
    {
      "epoch": 1.4146950012905446,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008155175236767434,
      "loss": 1.2147,
      "step": 16443
    },
    {
      "epoch": 1.4147810375978662,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008154965462478686,
      "loss": 1.1742,
      "step": 16444
    },
    {
      "epoch": 1.414867073905188,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008154755678962278,
      "loss": 1.5847,
      "step": 16445
    },
    {
      "epoch": 1.4149531102125097,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000815454588621882,
      "loss": 1.4084,
      "step": 16446
    },
    {
      "epoch": 1.4150391465198313,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008154336084248932,
      "loss": 1.9099,
      "step": 16447
    },
    {
      "epoch": 1.4151251828271532,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008154126273053224,
      "loss": 0.8349,
      "step": 16448
    },
    {
      "epoch": 1.4152112191344748,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008153916452632308,
      "loss": 0.842,
      "step": 16449
    },
    {
      "epoch": 1.4152972554417964,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008153706622986801,
      "loss": 1.4078,
      "step": 16450
    },
    {
      "epoch": 1.415383291749118,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008153496784117316,
      "loss": 1.7462,
      "step": 16451
    },
    {
      "epoch": 1.4154693280564399,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008153286936024467,
      "loss": 1.5104,
      "step": 16452
    },
    {
      "epoch": 1.4155553643637615,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008153077078708866,
      "loss": 1.2923,
      "step": 16453
    },
    {
      "epoch": 1.4156414006710831,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008152867212171126,
      "loss": 1.3446,
      "step": 16454
    },
    {
      "epoch": 1.415727436978405,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008152657336411866,
      "loss": 1.3018,
      "step": 16455
    },
    {
      "epoch": 1.4158134732857266,
      "grad_norm": 1.53125,
      "learning_rate": 0.0008152447451431694,
      "loss": 1.8962,
      "step": 16456
    },
    {
      "epoch": 1.4158995095930482,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008152237557231228,
      "loss": 1.1971,
      "step": 16457
    },
    {
      "epoch": 1.4159855459003698,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008152027653811078,
      "loss": 1.9582,
      "step": 16458
    },
    {
      "epoch": 1.4160715822076917,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0008151817741171862,
      "loss": 1.6692,
      "step": 16459
    },
    {
      "epoch": 1.4161576185150133,
      "grad_norm": 1.171875,
      "learning_rate": 0.000815160781931419,
      "loss": 1.2115,
      "step": 16460
    },
    {
      "epoch": 1.4162436548223352,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008151397888238681,
      "loss": 0.8968,
      "step": 16461
    },
    {
      "epoch": 1.4163296911296568,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008151187947945944,
      "loss": 1.2889,
      "step": 16462
    },
    {
      "epoch": 1.4164157274369784,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008150977998436596,
      "loss": 1.128,
      "step": 16463
    },
    {
      "epoch": 1.4165017637443,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008150768039711249,
      "loss": 1.7233,
      "step": 16464
    },
    {
      "epoch": 1.4165878000516217,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008150558071770519,
      "loss": 1.0369,
      "step": 16465
    },
    {
      "epoch": 1.4166738363589435,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008150348094615021,
      "loss": 1.1428,
      "step": 16466
    },
    {
      "epoch": 1.4167598726662651,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008150138108245366,
      "loss": 1.3385,
      "step": 16467
    },
    {
      "epoch": 1.416845908973587,
      "grad_norm": 1.015625,
      "learning_rate": 0.000814992811266217,
      "loss": 0.8943,
      "step": 16468
    },
    {
      "epoch": 1.4169319452809086,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008149718107866047,
      "loss": 1.1358,
      "step": 16469
    },
    {
      "epoch": 1.4170179815882302,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008149508093857612,
      "loss": 1.2822,
      "step": 16470
    },
    {
      "epoch": 1.4171040178955518,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008149298070637477,
      "loss": 0.9304,
      "step": 16471
    },
    {
      "epoch": 1.4171900542028737,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008149088038206257,
      "loss": 1.1386,
      "step": 16472
    },
    {
      "epoch": 1.4172760905101953,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008148877996564568,
      "loss": 1.0267,
      "step": 16473
    },
    {
      "epoch": 1.417362126817517,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008148667945713024,
      "loss": 1.1987,
      "step": 16474
    },
    {
      "epoch": 1.4174481631248388,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008148457885652238,
      "loss": 1.1831,
      "step": 16475
    },
    {
      "epoch": 1.4175341994321604,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008148247816382824,
      "loss": 1.8754,
      "step": 16476
    },
    {
      "epoch": 1.417620235739482,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008148037737905398,
      "loss": 1.6948,
      "step": 16477
    },
    {
      "epoch": 1.4177062720468037,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008147827650220574,
      "loss": 1.4573,
      "step": 16478
    },
    {
      "epoch": 1.4177923083541255,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008147617553328966,
      "loss": 1.1102,
      "step": 16479
    },
    {
      "epoch": 1.4178783446614471,
      "grad_norm": 1.125,
      "learning_rate": 0.000814740744723119,
      "loss": 1.8834,
      "step": 16480
    },
    {
      "epoch": 1.4179643809687688,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0008147197331927858,
      "loss": 2.149,
      "step": 16481
    },
    {
      "epoch": 1.4180504172760906,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008146987207419587,
      "loss": 1.7276,
      "step": 16482
    },
    {
      "epoch": 1.4181364535834122,
      "grad_norm": 1.390625,
      "learning_rate": 0.000814677707370699,
      "loss": 1.5131,
      "step": 16483
    },
    {
      "epoch": 1.4182224898907339,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008146566930790682,
      "loss": 1.9739,
      "step": 16484
    },
    {
      "epoch": 1.4183085261980555,
      "grad_norm": 2.1875,
      "learning_rate": 0.0008146356778671277,
      "loss": 1.6729,
      "step": 16485
    },
    {
      "epoch": 1.4183945625053773,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008146146617349393,
      "loss": 0.7257,
      "step": 16486
    },
    {
      "epoch": 1.418480598812699,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008145936446825639,
      "loss": 0.7611,
      "step": 16487
    },
    {
      "epoch": 1.4185666351200206,
      "grad_norm": 1.0,
      "learning_rate": 0.0008145726267100635,
      "loss": 1.8589,
      "step": 16488
    },
    {
      "epoch": 1.4186526714273424,
      "grad_norm": 1.0,
      "learning_rate": 0.0008145516078174993,
      "loss": 1.0634,
      "step": 16489
    },
    {
      "epoch": 1.418738707734664,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008145305880049328,
      "loss": 1.2104,
      "step": 16490
    },
    {
      "epoch": 1.4188247440419857,
      "grad_norm": 1.375,
      "learning_rate": 0.0008145095672724255,
      "loss": 1.1386,
      "step": 16491
    },
    {
      "epoch": 1.4189107803493073,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008144885456200387,
      "loss": 0.9973,
      "step": 16492
    },
    {
      "epoch": 1.4189968166566291,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008144675230478342,
      "loss": 1.6587,
      "step": 16493
    },
    {
      "epoch": 1.4190828529639508,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008144464995558734,
      "loss": 1.4516,
      "step": 16494
    },
    {
      "epoch": 1.4191688892712726,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008144254751442177,
      "loss": 1.4936,
      "step": 16495
    },
    {
      "epoch": 1.4192549255785942,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008144044498129287,
      "loss": 0.8404,
      "step": 16496
    },
    {
      "epoch": 1.4193409618859159,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008143834235620677,
      "loss": 0.8682,
      "step": 16497
    },
    {
      "epoch": 1.4194269981932375,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008143623963916964,
      "loss": 0.6025,
      "step": 16498
    },
    {
      "epoch": 1.419513034500559,
      "grad_norm": 1.375,
      "learning_rate": 0.0008143413683018763,
      "loss": 1.0265,
      "step": 16499
    },
    {
      "epoch": 1.419599070807881,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008143203392926688,
      "loss": 1.5495,
      "step": 16500
    },
    {
      "epoch": 1.4196851071152026,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008142993093641355,
      "loss": 0.6251,
      "step": 16501
    },
    {
      "epoch": 1.4197711434225244,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008142782785163377,
      "loss": 1.1492,
      "step": 16502
    },
    {
      "epoch": 1.419857179729846,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008142572467493372,
      "loss": 1.4702,
      "step": 16503
    },
    {
      "epoch": 1.4199432160371677,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008142362140631954,
      "loss": 1.2857,
      "step": 16504
    },
    {
      "epoch": 1.4200292523444893,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008142151804579736,
      "loss": 1.4095,
      "step": 16505
    },
    {
      "epoch": 1.4201152886518111,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008141941459337336,
      "loss": 1.0899,
      "step": 16506
    },
    {
      "epoch": 1.4202013249591328,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008141731104905368,
      "loss": 1.0634,
      "step": 16507
    },
    {
      "epoch": 1.4202873612664544,
      "grad_norm": 1.25,
      "learning_rate": 0.0008141520741284449,
      "loss": 1.1854,
      "step": 16508
    },
    {
      "epoch": 1.4203733975737762,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008141310368475191,
      "loss": 1.1383,
      "step": 16509
    },
    {
      "epoch": 1.4204594338810979,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008141099986478212,
      "loss": 2.2943,
      "step": 16510
    },
    {
      "epoch": 1.4205454701884195,
      "grad_norm": 1.796875,
      "learning_rate": 0.0008140889595294127,
      "loss": 1.1961,
      "step": 16511
    },
    {
      "epoch": 1.420631506495741,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008140679194923549,
      "loss": 1.4412,
      "step": 16512
    },
    {
      "epoch": 1.420717542803063,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008140468785367097,
      "loss": 1.4759,
      "step": 16513
    },
    {
      "epoch": 1.4208035791103846,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008140258366625384,
      "loss": 1.736,
      "step": 16514
    },
    {
      "epoch": 1.4208896154177062,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0008140047938699027,
      "loss": 1.497,
      "step": 16515
    },
    {
      "epoch": 1.420975651725028,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008139837501588638,
      "loss": 1.3868,
      "step": 16516
    },
    {
      "epoch": 1.4210616880323497,
      "grad_norm": 1.25,
      "learning_rate": 0.0008139627055294836,
      "loss": 1.2759,
      "step": 16517
    },
    {
      "epoch": 1.4211477243396713,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008139416599818235,
      "loss": 1.0309,
      "step": 16518
    },
    {
      "epoch": 1.421233760646993,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008139206135159453,
      "loss": 1.5476,
      "step": 16519
    },
    {
      "epoch": 1.4213197969543148,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008138995661319102,
      "loss": 1.5375,
      "step": 16520
    },
    {
      "epoch": 1.4214058332616364,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008138785178297801,
      "loss": 0.6651,
      "step": 16521
    },
    {
      "epoch": 1.4214918695689582,
      "grad_norm": 1.1875,
      "learning_rate": 0.000813857468609616,
      "loss": 1.4777,
      "step": 16522
    },
    {
      "epoch": 1.4215779058762799,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008138364184714803,
      "loss": 0.8856,
      "step": 16523
    },
    {
      "epoch": 1.4216639421836015,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008138153674154339,
      "loss": 1.4832,
      "step": 16524
    },
    {
      "epoch": 1.421749978490923,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0008137943154415386,
      "loss": 1.0453,
      "step": 16525
    },
    {
      "epoch": 1.4218360147982447,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008137732625498559,
      "loss": 0.7859,
      "step": 16526
    },
    {
      "epoch": 1.4219220511055666,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008137522087404474,
      "loss": 0.8908,
      "step": 16527
    },
    {
      "epoch": 1.4220080874128882,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008137311540133749,
      "loss": 1.1231,
      "step": 16528
    },
    {
      "epoch": 1.42209412372021,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008137100983686997,
      "loss": 1.5936,
      "step": 16529
    },
    {
      "epoch": 1.4221801600275317,
      "grad_norm": 1.125,
      "learning_rate": 0.0008136890418064835,
      "loss": 1.5603,
      "step": 16530
    },
    {
      "epoch": 1.4222661963348533,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008136679843267878,
      "loss": 1.7252,
      "step": 16531
    },
    {
      "epoch": 1.422352232642175,
      "grad_norm": 1.5,
      "learning_rate": 0.0008136469259296743,
      "loss": 1.6292,
      "step": 16532
    },
    {
      "epoch": 1.4224382689494968,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008136258666152045,
      "loss": 1.7691,
      "step": 16533
    },
    {
      "epoch": 1.4225243052568184,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008136048063834401,
      "loss": 0.6414,
      "step": 16534
    },
    {
      "epoch": 1.42261034156414,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008135837452344424,
      "loss": 1.5915,
      "step": 16535
    },
    {
      "epoch": 1.4226963778714619,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008135626831682735,
      "loss": 2.044,
      "step": 16536
    },
    {
      "epoch": 1.4227824141787835,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008135416201849946,
      "loss": 1.4626,
      "step": 16537
    },
    {
      "epoch": 1.422868450486105,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008135205562846675,
      "loss": 1.099,
      "step": 16538
    },
    {
      "epoch": 1.4229544867934267,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008134994914673538,
      "loss": 1.3118,
      "step": 16539
    },
    {
      "epoch": 1.4230405231007486,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008134784257331151,
      "loss": 1.2942,
      "step": 16540
    },
    {
      "epoch": 1.4231265594080702,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008134573590820128,
      "loss": 1.4031,
      "step": 16541
    },
    {
      "epoch": 1.4232125957153918,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008134362915141087,
      "loss": 0.7789,
      "step": 16542
    },
    {
      "epoch": 1.4232986320227137,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008134152230294645,
      "loss": 1.2132,
      "step": 16543
    },
    {
      "epoch": 1.4233846683300353,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008133941536281417,
      "loss": 1.0454,
      "step": 16544
    },
    {
      "epoch": 1.423470704637357,
      "grad_norm": 1.453125,
      "learning_rate": 0.000813373083310202,
      "loss": 1.8198,
      "step": 16545
    },
    {
      "epoch": 1.4235567409446785,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000813352012075707,
      "loss": 1.4928,
      "step": 16546
    },
    {
      "epoch": 1.4236427772520004,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008133309399247182,
      "loss": 2.4987,
      "step": 16547
    },
    {
      "epoch": 1.423728813559322,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008133098668572973,
      "loss": 1.4927,
      "step": 16548
    },
    {
      "epoch": 1.4238148498666439,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008132887928735062,
      "loss": 0.9845,
      "step": 16549
    },
    {
      "epoch": 1.4239008861739655,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008132677179734061,
      "loss": 0.7574,
      "step": 16550
    },
    {
      "epoch": 1.423986922481287,
      "grad_norm": 2.359375,
      "learning_rate": 0.000813246642157059,
      "loss": 2.1099,
      "step": 16551
    },
    {
      "epoch": 1.4240729587886087,
      "grad_norm": 1.125,
      "learning_rate": 0.0008132255654245263,
      "loss": 1.4053,
      "step": 16552
    },
    {
      "epoch": 1.4241589950959304,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008132044877758697,
      "loss": 1.0115,
      "step": 16553
    },
    {
      "epoch": 1.4242450314032522,
      "grad_norm": 1.28125,
      "learning_rate": 0.000813183409211151,
      "loss": 1.0896,
      "step": 16554
    },
    {
      "epoch": 1.4243310677105738,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008131623297304316,
      "loss": 1.1651,
      "step": 16555
    },
    {
      "epoch": 1.4244171040178957,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008131412493337734,
      "loss": 1.9139,
      "step": 16556
    },
    {
      "epoch": 1.4245031403252173,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008131201680212378,
      "loss": 1.2955,
      "step": 16557
    },
    {
      "epoch": 1.424589176632539,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008130990857928869,
      "loss": 1.5354,
      "step": 16558
    },
    {
      "epoch": 1.4246752129398605,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008130780026487819,
      "loss": 1.222,
      "step": 16559
    },
    {
      "epoch": 1.4247612492471824,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008130569185889847,
      "loss": 0.8648,
      "step": 16560
    },
    {
      "epoch": 1.424847285554504,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008130358336135568,
      "loss": 1.6162,
      "step": 16561
    },
    {
      "epoch": 1.4249333218618256,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00081301474772256,
      "loss": 1.1062,
      "step": 16562
    },
    {
      "epoch": 1.4250193581691475,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000812993660916056,
      "loss": 1.445,
      "step": 16563
    },
    {
      "epoch": 1.4251053944764691,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008129725731941062,
      "loss": 1.1236,
      "step": 16564
    },
    {
      "epoch": 1.4251914307837907,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008129514845567727,
      "loss": 1.2314,
      "step": 16565
    },
    {
      "epoch": 1.4252774670911124,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008129303950041169,
      "loss": 2.172,
      "step": 16566
    },
    {
      "epoch": 1.4253635033984342,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008129093045362006,
      "loss": 0.9516,
      "step": 16567
    },
    {
      "epoch": 1.4254495397057558,
      "grad_norm": 1.75,
      "learning_rate": 0.0008128882131530855,
      "loss": 1.3076,
      "step": 16568
    },
    {
      "epoch": 1.4255355760130775,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008128671208548332,
      "loss": 1.375,
      "step": 16569
    },
    {
      "epoch": 1.4256216123203993,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008128460276415053,
      "loss": 1.5378,
      "step": 16570
    },
    {
      "epoch": 1.425707648627721,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008128249335131637,
      "loss": 1.2568,
      "step": 16571
    },
    {
      "epoch": 1.4257936849350425,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008128038384698699,
      "loss": 1.6226,
      "step": 16572
    },
    {
      "epoch": 1.4258797212423642,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008127827425116858,
      "loss": 1.1082,
      "step": 16573
    },
    {
      "epoch": 1.425965757549686,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000812761645638673,
      "loss": 1.5704,
      "step": 16574
    },
    {
      "epoch": 1.4260517938570076,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008127405478508934,
      "loss": 1.8265,
      "step": 16575
    },
    {
      "epoch": 1.4261378301643293,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008127194491484082,
      "loss": 1.3431,
      "step": 16576
    },
    {
      "epoch": 1.4262238664716511,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008126983495312796,
      "loss": 1.3122,
      "step": 16577
    },
    {
      "epoch": 1.4263099027789727,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008126772489995692,
      "loss": 1.607,
      "step": 16578
    },
    {
      "epoch": 1.4263959390862944,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008126561475533386,
      "loss": 1.5996,
      "step": 16579
    },
    {
      "epoch": 1.426481975393616,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008126350451926497,
      "loss": 1.7345,
      "step": 16580
    },
    {
      "epoch": 1.4265680117009378,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008126139419175639,
      "loss": 2.1613,
      "step": 16581
    },
    {
      "epoch": 1.4266540480082595,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000812592837728143,
      "loss": 1.3393,
      "step": 16582
    },
    {
      "epoch": 1.4267400843155813,
      "grad_norm": 1.0,
      "learning_rate": 0.0008125717326244492,
      "loss": 1.4887,
      "step": 16583
    },
    {
      "epoch": 1.426826120622903,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008125506266065437,
      "loss": 1.0545,
      "step": 16584
    },
    {
      "epoch": 1.4269121569302246,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008125295196744884,
      "loss": 2.3828,
      "step": 16585
    },
    {
      "epoch": 1.4269981932375462,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008125084118283451,
      "loss": 0.8502,
      "step": 16586
    },
    {
      "epoch": 1.4270842295448678,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008124873030681755,
      "loss": 1.9521,
      "step": 16587
    },
    {
      "epoch": 1.4271702658521896,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0008124661933940412,
      "loss": 1.4507,
      "step": 16588
    },
    {
      "epoch": 1.4272563021595113,
      "grad_norm": 2.28125,
      "learning_rate": 0.0008124450828060041,
      "loss": 1.4813,
      "step": 16589
    },
    {
      "epoch": 1.4273423384668331,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008124239713041257,
      "loss": 1.5591,
      "step": 16590
    },
    {
      "epoch": 1.4274283747741547,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008124028588884682,
      "loss": 1.4205,
      "step": 16591
    },
    {
      "epoch": 1.4275144110814764,
      "grad_norm": 1.375,
      "learning_rate": 0.000812381745559093,
      "loss": 1.4861,
      "step": 16592
    },
    {
      "epoch": 1.427600447388798,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008123606313160621,
      "loss": 1.0074,
      "step": 16593
    },
    {
      "epoch": 1.4276864836961198,
      "grad_norm": 1.0,
      "learning_rate": 0.0008123395161594369,
      "loss": 0.7607,
      "step": 16594
    },
    {
      "epoch": 1.4277725200034415,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008123184000892794,
      "loss": 1.1653,
      "step": 16595
    },
    {
      "epoch": 1.427858556310763,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008122972831056515,
      "loss": 2.0216,
      "step": 16596
    },
    {
      "epoch": 1.427944592618085,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008122761652086146,
      "loss": 1.684,
      "step": 16597
    },
    {
      "epoch": 1.4280306289254066,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008122550463982308,
      "loss": 0.6978,
      "step": 16598
    },
    {
      "epoch": 1.4281166652327282,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008122339266745615,
      "loss": 1.197,
      "step": 16599
    },
    {
      "epoch": 1.4282027015400498,
      "grad_norm": 1.125,
      "learning_rate": 0.0008122128060376689,
      "loss": 1.1231,
      "step": 16600
    },
    {
      "epoch": 1.4282887378473716,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008121916844876145,
      "loss": 1.0212,
      "step": 16601
    },
    {
      "epoch": 1.4283747741546933,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008121705620244601,
      "loss": 1.9299,
      "step": 16602
    },
    {
      "epoch": 1.428460810462015,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008121494386482675,
      "loss": 1.2997,
      "step": 16603
    },
    {
      "epoch": 1.4285468467693367,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008121283143590986,
      "loss": 1.8322,
      "step": 16604
    },
    {
      "epoch": 1.4286328830766584,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008121071891570151,
      "loss": 0.9207,
      "step": 16605
    },
    {
      "epoch": 1.42871891938398,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008120860630420787,
      "loss": 0.9552,
      "step": 16606
    },
    {
      "epoch": 1.4288049556913016,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0008120649360143512,
      "loss": 0.8291,
      "step": 16607
    },
    {
      "epoch": 1.4288909919986235,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008120438080738945,
      "loss": 1.6313,
      "step": 16608
    },
    {
      "epoch": 1.428977028305945,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008120226792207704,
      "loss": 0.8688,
      "step": 16609
    },
    {
      "epoch": 1.429063064613267,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008120015494550405,
      "loss": 0.7505,
      "step": 16610
    },
    {
      "epoch": 1.4291491009205886,
      "grad_norm": 1.28125,
      "learning_rate": 0.000811980418776767,
      "loss": 0.9686,
      "step": 16611
    },
    {
      "epoch": 1.4292351372279102,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008119592871860113,
      "loss": 1.1008,
      "step": 16612
    },
    {
      "epoch": 1.4293211735352318,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008119381546828354,
      "loss": 1.7831,
      "step": 16613
    },
    {
      "epoch": 1.4294072098425534,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008119170212673011,
      "loss": 1.7307,
      "step": 16614
    },
    {
      "epoch": 1.4294932461498753,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008118958869394701,
      "loss": 1.4137,
      "step": 16615
    },
    {
      "epoch": 1.429579282457197,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008118747516994043,
      "loss": 1.3983,
      "step": 16616
    },
    {
      "epoch": 1.4296653187645187,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008118536155471655,
      "loss": 1.7418,
      "step": 16617
    },
    {
      "epoch": 1.4297513550718404,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008118324784828156,
      "loss": 1.0343,
      "step": 16618
    },
    {
      "epoch": 1.429837391379162,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0008118113405064162,
      "loss": 1.3014,
      "step": 16619
    },
    {
      "epoch": 1.4299234276864836,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008117902016180294,
      "loss": 1.6071,
      "step": 16620
    },
    {
      "epoch": 1.4300094639938055,
      "grad_norm": 0.921875,
      "learning_rate": 0.0008117690618177169,
      "loss": 1.4428,
      "step": 16621
    },
    {
      "epoch": 1.430095500301127,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008117479211055405,
      "loss": 0.9916,
      "step": 16622
    },
    {
      "epoch": 1.4301815366084487,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008117267794815621,
      "loss": 0.9659,
      "step": 16623
    },
    {
      "epoch": 1.4302675729157706,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008117056369458435,
      "loss": 1.3125,
      "step": 16624
    },
    {
      "epoch": 1.4303536092230922,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008116844934984466,
      "loss": 0.8106,
      "step": 16625
    },
    {
      "epoch": 1.4304396455304138,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000811663349139433,
      "loss": 1.5101,
      "step": 16626
    },
    {
      "epoch": 1.4305256818377354,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008116422038688647,
      "loss": 0.9708,
      "step": 16627
    },
    {
      "epoch": 1.4306117181450573,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008116210576868037,
      "loss": 1.3189,
      "step": 16628
    },
    {
      "epoch": 1.430697754452379,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008115999105933117,
      "loss": 1.13,
      "step": 16629
    },
    {
      "epoch": 1.4307837907597005,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008115787625884506,
      "loss": 1.9182,
      "step": 16630
    },
    {
      "epoch": 1.4308698270670224,
      "grad_norm": 1.421875,
      "learning_rate": 0.000811557613672282,
      "loss": 1.6475,
      "step": 16631
    },
    {
      "epoch": 1.430955863374344,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008115364638448681,
      "loss": 1.5724,
      "step": 16632
    },
    {
      "epoch": 1.4310418996816656,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008115153131062707,
      "loss": 1.657,
      "step": 16633
    },
    {
      "epoch": 1.4311279359889872,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0008114941614565516,
      "loss": 0.8638,
      "step": 16634
    },
    {
      "epoch": 1.431213972296309,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008114730088957727,
      "loss": 1.5012,
      "step": 16635
    },
    {
      "epoch": 1.4313000086036307,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008114518554239957,
      "loss": 2.0948,
      "step": 16636
    },
    {
      "epoch": 1.4313860449109526,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008114307010412827,
      "loss": 0.7699,
      "step": 16637
    },
    {
      "epoch": 1.4314720812182742,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008114095457476953,
      "loss": 1.2194,
      "step": 16638
    },
    {
      "epoch": 1.4315581175255958,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008113883895432956,
      "loss": 1.493,
      "step": 16639
    },
    {
      "epoch": 1.4316441538329174,
      "grad_norm": 1.453125,
      "learning_rate": 0.0008113672324281455,
      "loss": 1.5044,
      "step": 16640
    },
    {
      "epoch": 1.431730190140239,
      "grad_norm": 1.5,
      "learning_rate": 0.0008113460744023069,
      "loss": 1.6305,
      "step": 16641
    },
    {
      "epoch": 1.431816226447561,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008113249154658415,
      "loss": 0.7678,
      "step": 16642
    },
    {
      "epoch": 1.4319022627548825,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008113037556188112,
      "loss": 2.1968,
      "step": 16643
    },
    {
      "epoch": 1.4319882990622044,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008112825948612779,
      "loss": 1.6295,
      "step": 16644
    },
    {
      "epoch": 1.432074335369526,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008112614331933037,
      "loss": 1.2104,
      "step": 16645
    },
    {
      "epoch": 1.4321603716768476,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008112402706149502,
      "loss": 1.3023,
      "step": 16646
    },
    {
      "epoch": 1.4322464079841692,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008112191071262796,
      "loss": 1.2189,
      "step": 16647
    },
    {
      "epoch": 1.432332444291491,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008111979427273536,
      "loss": 1.8158,
      "step": 16648
    },
    {
      "epoch": 1.4324184805988127,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008111767774182341,
      "loss": 1.197,
      "step": 16649
    },
    {
      "epoch": 1.4325045169061343,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008111556111989831,
      "loss": 1.4203,
      "step": 16650
    },
    {
      "epoch": 1.4325905532134562,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008111344440696624,
      "loss": 0.9014,
      "step": 16651
    },
    {
      "epoch": 1.4326765895207778,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008111132760303339,
      "loss": 1.8188,
      "step": 16652
    },
    {
      "epoch": 1.4327626258280994,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008110921070810598,
      "loss": 1.6366,
      "step": 16653
    },
    {
      "epoch": 1.432848662135421,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008110709372219016,
      "loss": 1.0084,
      "step": 16654
    },
    {
      "epoch": 1.432934698442743,
      "grad_norm": 1.5,
      "learning_rate": 0.0008110497664529215,
      "loss": 1.2926,
      "step": 16655
    },
    {
      "epoch": 1.4330207347500645,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008110285947741812,
      "loss": 1.2626,
      "step": 16656
    },
    {
      "epoch": 1.4331067710573862,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008110074221857428,
      "loss": 1.5707,
      "step": 16657
    },
    {
      "epoch": 1.433192807364708,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008109862486876682,
      "loss": 1.1984,
      "step": 16658
    },
    {
      "epoch": 1.4332788436720296,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008109650742800194,
      "loss": 1.1012,
      "step": 16659
    },
    {
      "epoch": 1.4333648799793512,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008109438989628581,
      "loss": 1.4114,
      "step": 16660
    },
    {
      "epoch": 1.4334509162866729,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008109227227362466,
      "loss": 1.4501,
      "step": 16661
    },
    {
      "epoch": 1.4335369525939947,
      "grad_norm": 0.9375,
      "learning_rate": 0.0008109015456002465,
      "loss": 1.1271,
      "step": 16662
    },
    {
      "epoch": 1.4336229889013163,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008108803675549197,
      "loss": 1.3728,
      "step": 16663
    },
    {
      "epoch": 1.433709025208638,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008108591886003282,
      "loss": 1.3111,
      "step": 16664
    },
    {
      "epoch": 1.4337950615159598,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008108380087365343,
      "loss": 0.583,
      "step": 16665
    },
    {
      "epoch": 1.4338810978232814,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008108168279635995,
      "loss": 0.9051,
      "step": 16666
    },
    {
      "epoch": 1.433967134130603,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008107956462815861,
      "loss": 1.9783,
      "step": 16667
    },
    {
      "epoch": 1.4340531704379247,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008107744636905558,
      "loss": 1.9757,
      "step": 16668
    },
    {
      "epoch": 1.4341392067452465,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008107532801905706,
      "loss": 1.1085,
      "step": 16669
    },
    {
      "epoch": 1.4342252430525682,
      "grad_norm": 1.0,
      "learning_rate": 0.0008107320957816925,
      "loss": 1.0095,
      "step": 16670
    },
    {
      "epoch": 1.43431127935989,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008107109104639835,
      "loss": 1.2622,
      "step": 16671
    },
    {
      "epoch": 1.4343973156672116,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008106897242375055,
      "loss": 1.3575,
      "step": 16672
    },
    {
      "epoch": 1.4344833519745332,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008106685371023205,
      "loss": 1.1795,
      "step": 16673
    },
    {
      "epoch": 1.4345693882818549,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008106473490584905,
      "loss": 1.0511,
      "step": 16674
    },
    {
      "epoch": 1.4346554245891765,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008106261601060772,
      "loss": 1.7736,
      "step": 16675
    },
    {
      "epoch": 1.4347414608964983,
      "grad_norm": 1.015625,
      "learning_rate": 0.000810604970245143,
      "loss": 0.7292,
      "step": 16676
    },
    {
      "epoch": 1.43482749720382,
      "grad_norm": 1.125,
      "learning_rate": 0.0008105837794757496,
      "loss": 1.8452,
      "step": 16677
    },
    {
      "epoch": 1.4349135335111418,
      "grad_norm": 1.59375,
      "learning_rate": 0.0008105625877979591,
      "loss": 1.6652,
      "step": 16678
    },
    {
      "epoch": 1.4349995698184634,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008105413952118334,
      "loss": 1.1965,
      "step": 16679
    },
    {
      "epoch": 1.435085606125785,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008105202017174344,
      "loss": 0.8643,
      "step": 16680
    },
    {
      "epoch": 1.4351716424331067,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008104990073148244,
      "loss": 1.1645,
      "step": 16681
    },
    {
      "epoch": 1.4352576787404285,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000810477812004065,
      "loss": 1.6331,
      "step": 16682
    },
    {
      "epoch": 1.4353437150477502,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008104566157852185,
      "loss": 0.9218,
      "step": 16683
    },
    {
      "epoch": 1.4354297513550718,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008104354186583469,
      "loss": 1.337,
      "step": 16684
    },
    {
      "epoch": 1.4355157876623936,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008104142206235119,
      "loss": 1.3413,
      "step": 16685
    },
    {
      "epoch": 1.4356018239697153,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008103930216807757,
      "loss": 0.7342,
      "step": 16686
    },
    {
      "epoch": 1.4356878602770369,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008103718218302003,
      "loss": 1.3956,
      "step": 16687
    },
    {
      "epoch": 1.4357738965843585,
      "grad_norm": 1.25,
      "learning_rate": 0.0008103506210718476,
      "loss": 0.9281,
      "step": 16688
    },
    {
      "epoch": 1.4358599328916803,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008103294194057799,
      "loss": 1.0718,
      "step": 16689
    },
    {
      "epoch": 1.435945969199002,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008103082168320587,
      "loss": 1.4347,
      "step": 16690
    },
    {
      "epoch": 1.4360320055063236,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0008102870133507468,
      "loss": 0.9883,
      "step": 16691
    },
    {
      "epoch": 1.4361180418136454,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008102658089619053,
      "loss": 0.9816,
      "step": 16692
    },
    {
      "epoch": 1.436204078120967,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008102446036655968,
      "loss": 1.4154,
      "step": 16693
    },
    {
      "epoch": 1.4362901144282887,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008102233974618831,
      "loss": 1.6591,
      "step": 16694
    },
    {
      "epoch": 1.4363761507356103,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008102021903508265,
      "loss": 1.1985,
      "step": 16695
    },
    {
      "epoch": 1.4364621870429322,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008101809823324887,
      "loss": 0.7954,
      "step": 16696
    },
    {
      "epoch": 1.4365482233502538,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008101597734069317,
      "loss": 1.682,
      "step": 16697
    },
    {
      "epoch": 1.4366342596575756,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008101385635742179,
      "loss": 0.7226,
      "step": 16698
    },
    {
      "epoch": 1.4367202959648973,
      "grad_norm": 1.9296875,
      "learning_rate": 0.000810117352834409,
      "loss": 2.0887,
      "step": 16699
    },
    {
      "epoch": 1.4368063322722189,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0008100961411875672,
      "loss": 2.3425,
      "step": 16700
    },
    {
      "epoch": 1.4368923685795405,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008100749286337546,
      "loss": 1.2228,
      "step": 16701
    },
    {
      "epoch": 1.4369784048868621,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008100537151730329,
      "loss": 0.89,
      "step": 16702
    },
    {
      "epoch": 1.437064441194184,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008100325008054645,
      "loss": 1.9442,
      "step": 16703
    },
    {
      "epoch": 1.4371504775015056,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008100112855311115,
      "loss": 1.6718,
      "step": 16704
    },
    {
      "epoch": 1.4372365138088274,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008099900693500355,
      "loss": 1.3686,
      "step": 16705
    },
    {
      "epoch": 1.437322550116149,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000809968852262299,
      "loss": 1.3647,
      "step": 16706
    },
    {
      "epoch": 1.4374085864234707,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008099476342679639,
      "loss": 1.0041,
      "step": 16707
    },
    {
      "epoch": 1.4374946227307923,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008099264153670923,
      "loss": 0.9471,
      "step": 16708
    },
    {
      "epoch": 1.4375806590381142,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008099051955597461,
      "loss": 1.3967,
      "step": 16709
    },
    {
      "epoch": 1.4376666953454358,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008098839748459874,
      "loss": 1.4633,
      "step": 16710
    },
    {
      "epoch": 1.4377527316527574,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008098627532258785,
      "loss": 0.8982,
      "step": 16711
    },
    {
      "epoch": 1.4378387679600793,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008098415306994813,
      "loss": 1.2627,
      "step": 16712
    },
    {
      "epoch": 1.4379248042674009,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008098203072668578,
      "loss": 1.7092,
      "step": 16713
    },
    {
      "epoch": 1.4380108405747225,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008097990829280703,
      "loss": 0.995,
      "step": 16714
    },
    {
      "epoch": 1.4380968768820441,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008097778576831807,
      "loss": 1.025,
      "step": 16715
    },
    {
      "epoch": 1.438182913189366,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008097566315322509,
      "loss": 1.114,
      "step": 16716
    },
    {
      "epoch": 1.4382689494966876,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008097354044753434,
      "loss": 1.0372,
      "step": 16717
    },
    {
      "epoch": 1.4383549858040092,
      "grad_norm": 1.09375,
      "learning_rate": 0.00080971417651252,
      "loss": 1.7112,
      "step": 16718
    },
    {
      "epoch": 1.438441022111331,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008096929476438428,
      "loss": 1.2924,
      "step": 16719
    },
    {
      "epoch": 1.4385270584186527,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008096717178693742,
      "loss": 1.2345,
      "step": 16720
    },
    {
      "epoch": 1.4386130947259743,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008096504871891759,
      "loss": 1.4869,
      "step": 16721
    },
    {
      "epoch": 1.438699131033296,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008096292556033101,
      "loss": 1.6052,
      "step": 16722
    },
    {
      "epoch": 1.4387851673406178,
      "grad_norm": 1.21875,
      "learning_rate": 0.000809608023111839,
      "loss": 1.1327,
      "step": 16723
    },
    {
      "epoch": 1.4388712036479394,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008095867897148248,
      "loss": 1.0199,
      "step": 16724
    },
    {
      "epoch": 1.4389572399552613,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008095655554123293,
      "loss": 1.429,
      "step": 16725
    },
    {
      "epoch": 1.4390432762625829,
      "grad_norm": 1.25,
      "learning_rate": 0.0008095443202044147,
      "loss": 1.2087,
      "step": 16726
    },
    {
      "epoch": 1.4391293125699045,
      "grad_norm": 1.25,
      "learning_rate": 0.0008095230840911432,
      "loss": 1.5964,
      "step": 16727
    },
    {
      "epoch": 1.4392153488772261,
      "grad_norm": 1.140625,
      "learning_rate": 0.000809501847072577,
      "loss": 1.0431,
      "step": 16728
    },
    {
      "epoch": 1.4393013851845478,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000809480609148778,
      "loss": 1.8942,
      "step": 16729
    },
    {
      "epoch": 1.4393874214918696,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008094593703198085,
      "loss": 1.3806,
      "step": 16730
    },
    {
      "epoch": 1.4394734577991912,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008094381305857304,
      "loss": 2.285,
      "step": 16731
    },
    {
      "epoch": 1.439559494106513,
      "grad_norm": 1.421875,
      "learning_rate": 0.0008094168899466059,
      "loss": 1.3779,
      "step": 16732
    },
    {
      "epoch": 1.4396455304138347,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008093956484024974,
      "loss": 1.7712,
      "step": 16733
    },
    {
      "epoch": 1.4397315667211563,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008093744059534667,
      "loss": 1.1541,
      "step": 16734
    },
    {
      "epoch": 1.439817603028478,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008093531625995759,
      "loss": 1.6936,
      "step": 16735
    },
    {
      "epoch": 1.4399036393357998,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008093319183408873,
      "loss": 1.2022,
      "step": 16736
    },
    {
      "epoch": 1.4399896756431214,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008093106731774631,
      "loss": 1.1085,
      "step": 16737
    },
    {
      "epoch": 1.440075711950443,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008092894271093654,
      "loss": 1.6757,
      "step": 16738
    },
    {
      "epoch": 1.4401617482577649,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008092681801366562,
      "loss": 0.9827,
      "step": 16739
    },
    {
      "epoch": 1.4402477845650865,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008092469322593976,
      "loss": 1.0609,
      "step": 16740
    },
    {
      "epoch": 1.4403338208724081,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008092256834776519,
      "loss": 1.5296,
      "step": 16741
    },
    {
      "epoch": 1.4404198571797298,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008092044337914815,
      "loss": 1.3569,
      "step": 16742
    },
    {
      "epoch": 1.4405058934870516,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000809183183200948,
      "loss": 1.0525,
      "step": 16743
    },
    {
      "epoch": 1.4405919297943732,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008091619317061138,
      "loss": 1.2511,
      "step": 16744
    },
    {
      "epoch": 1.4406779661016949,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008091406793070412,
      "loss": 1.5103,
      "step": 16745
    },
    {
      "epoch": 1.4407640024090167,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008091194260037922,
      "loss": 1.9703,
      "step": 16746
    },
    {
      "epoch": 1.4408500387163383,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008090981717964289,
      "loss": 1.0958,
      "step": 16747
    },
    {
      "epoch": 1.44093607502366,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008090769166850139,
      "loss": 1.0129,
      "step": 16748
    },
    {
      "epoch": 1.4410221113309816,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008090556606696087,
      "loss": 2.0943,
      "step": 16749
    },
    {
      "epoch": 1.4411081476383034,
      "grad_norm": 1.03125,
      "learning_rate": 0.000809034403750276,
      "loss": 1.251,
      "step": 16750
    },
    {
      "epoch": 1.441194183945625,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008090131459270775,
      "loss": 0.9116,
      "step": 16751
    },
    {
      "epoch": 1.4412802202529467,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008089918872000759,
      "loss": 0.9486,
      "step": 16752
    },
    {
      "epoch": 1.4413662565602685,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000808970627569333,
      "loss": 0.9247,
      "step": 16753
    },
    {
      "epoch": 1.4414522928675901,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0008089493670349111,
      "loss": 1.6349,
      "step": 16754
    },
    {
      "epoch": 1.4415383291749118,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008089281055968725,
      "loss": 1.6944,
      "step": 16755
    },
    {
      "epoch": 1.4416243654822334,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000808906843255279,
      "loss": 1.3645,
      "step": 16756
    },
    {
      "epoch": 1.4417104017895552,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008088855800101935,
      "loss": 1.8547,
      "step": 16757
    },
    {
      "epoch": 1.4417964380968769,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008088643158616775,
      "loss": 1.6487,
      "step": 16758
    },
    {
      "epoch": 1.4418824744041987,
      "grad_norm": 19.5,
      "learning_rate": 0.0008088430508097935,
      "loss": 1.2365,
      "step": 16759
    },
    {
      "epoch": 1.4419685107115203,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008088217848546035,
      "loss": 0.7243,
      "step": 16760
    },
    {
      "epoch": 1.442054547018842,
      "grad_norm": 2.296875,
      "learning_rate": 0.00080880051799617,
      "loss": 1.5301,
      "step": 16761
    },
    {
      "epoch": 1.4421405833261636,
      "grad_norm": 1.484375,
      "learning_rate": 0.000808779250234555,
      "loss": 1.3759,
      "step": 16762
    },
    {
      "epoch": 1.4422266196334852,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008087579815698208,
      "loss": 0.739,
      "step": 16763
    },
    {
      "epoch": 1.442312655940807,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008087367120020295,
      "loss": 1.8758,
      "step": 16764
    },
    {
      "epoch": 1.4423986922481287,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008087154415312434,
      "loss": 1.4766,
      "step": 16765
    },
    {
      "epoch": 1.4424847285554505,
      "grad_norm": 1.5,
      "learning_rate": 0.0008086941701575246,
      "loss": 1.6177,
      "step": 16766
    },
    {
      "epoch": 1.4425707648627721,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008086728978809355,
      "loss": 1.5413,
      "step": 16767
    },
    {
      "epoch": 1.4426568011700938,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008086516247015381,
      "loss": 0.9797,
      "step": 16768
    },
    {
      "epoch": 1.4427428374774154,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008086303506193948,
      "loss": 0.6109,
      "step": 16769
    },
    {
      "epoch": 1.4428288737847372,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008086090756345677,
      "loss": 0.9861,
      "step": 16770
    },
    {
      "epoch": 1.4429149100920589,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008085877997471192,
      "loss": 1.9828,
      "step": 16771
    },
    {
      "epoch": 1.4430009463993805,
      "grad_norm": 1.265625,
      "learning_rate": 0.000808566522957111,
      "loss": 1.5704,
      "step": 16772
    },
    {
      "epoch": 1.4430869827067023,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008085452452646063,
      "loss": 1.128,
      "step": 16773
    },
    {
      "epoch": 1.443173019014024,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008085239666696663,
      "loss": 1.2776,
      "step": 16774
    },
    {
      "epoch": 1.4432590553213456,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000808502687172354,
      "loss": 1.3583,
      "step": 16775
    },
    {
      "epoch": 1.4433450916286672,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008084814067727311,
      "loss": 1.6095,
      "step": 16776
    },
    {
      "epoch": 1.443431127935989,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008084601254708604,
      "loss": 1.0135,
      "step": 16777
    },
    {
      "epoch": 1.4435171642433107,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008084388432668035,
      "loss": 1.1706,
      "step": 16778
    },
    {
      "epoch": 1.4436032005506323,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008084175601606231,
      "loss": 0.8603,
      "step": 16779
    },
    {
      "epoch": 1.4436892368579541,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008083962761523813,
      "loss": 1.517,
      "step": 16780
    },
    {
      "epoch": 1.4437752731652758,
      "grad_norm": 1.734375,
      "learning_rate": 0.0008083749912421405,
      "loss": 1.4601,
      "step": 16781
    },
    {
      "epoch": 1.4438613094725974,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008083537054299628,
      "loss": 0.8567,
      "step": 16782
    },
    {
      "epoch": 1.443947345779919,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008083324187159103,
      "loss": 0.7691,
      "step": 16783
    },
    {
      "epoch": 1.4440333820872409,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008083111311000456,
      "loss": 1.6932,
      "step": 16784
    },
    {
      "epoch": 1.4441194183945625,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008082898425824308,
      "loss": 1.5045,
      "step": 16785
    },
    {
      "epoch": 1.4442054547018843,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008082685531631282,
      "loss": 1.1395,
      "step": 16786
    },
    {
      "epoch": 1.444291491009206,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008082472628421998,
      "loss": 1.1033,
      "step": 16787
    },
    {
      "epoch": 1.4443775273165276,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008082259716197085,
      "loss": 1.087,
      "step": 16788
    },
    {
      "epoch": 1.4444635636238492,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008082046794957159,
      "loss": 1.0976,
      "step": 16789
    },
    {
      "epoch": 1.4445495999311708,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008081833864702846,
      "loss": 1.6326,
      "step": 16790
    },
    {
      "epoch": 1.4446356362384927,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008081620925434768,
      "loss": 1.7586,
      "step": 16791
    },
    {
      "epoch": 1.4447216725458143,
      "grad_norm": 11.625,
      "learning_rate": 0.0008081407977153548,
      "loss": 1.3265,
      "step": 16792
    },
    {
      "epoch": 1.4448077088531361,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008081195019859809,
      "loss": 1.072,
      "step": 16793
    },
    {
      "epoch": 1.4448937451604578,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008080982053554176,
      "loss": 1.6134,
      "step": 16794
    },
    {
      "epoch": 1.4449797814677794,
      "grad_norm": 1.75,
      "learning_rate": 0.0008080769078237267,
      "loss": 0.8785,
      "step": 16795
    },
    {
      "epoch": 1.445065817775101,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008080556093909709,
      "loss": 1.0207,
      "step": 16796
    },
    {
      "epoch": 1.4451518540824229,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008080343100572123,
      "loss": 1.6863,
      "step": 16797
    },
    {
      "epoch": 1.4452378903897445,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008080130098225134,
      "loss": 0.8623,
      "step": 16798
    },
    {
      "epoch": 1.445323926697066,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008079917086869362,
      "loss": 0.7011,
      "step": 16799
    },
    {
      "epoch": 1.445409963004388,
      "grad_norm": 2.25,
      "learning_rate": 0.0008079704066505431,
      "loss": 0.84,
      "step": 16800
    },
    {
      "epoch": 1.4454959993117096,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008079491037133966,
      "loss": 1.3555,
      "step": 16801
    },
    {
      "epoch": 1.4455820356190312,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008079277998755588,
      "loss": 0.8997,
      "step": 16802
    },
    {
      "epoch": 1.4456680719263528,
      "grad_norm": 1.421875,
      "learning_rate": 0.000807906495137092,
      "loss": 2.2216,
      "step": 16803
    },
    {
      "epoch": 1.4457541082336747,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008078851894980587,
      "loss": 0.7668,
      "step": 16804
    },
    {
      "epoch": 1.4458401445409963,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008078638829585211,
      "loss": 1.6371,
      "step": 16805
    },
    {
      "epoch": 1.445926180848318,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008078425755185415,
      "loss": 1.8506,
      "step": 16806
    },
    {
      "epoch": 1.4460122171556398,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008078212671781822,
      "loss": 1.4661,
      "step": 16807
    },
    {
      "epoch": 1.4460982534629614,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008077999579375056,
      "loss": 1.9097,
      "step": 16808
    },
    {
      "epoch": 1.446184289770283,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000807778647796574,
      "loss": 1.5685,
      "step": 16809
    },
    {
      "epoch": 1.4462703260776046,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008077573367554496,
      "loss": 1.2487,
      "step": 16810
    },
    {
      "epoch": 1.4463563623849265,
      "grad_norm": 1.609375,
      "learning_rate": 0.000807736024814195,
      "loss": 1.225,
      "step": 16811
    },
    {
      "epoch": 1.446442398692248,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008077147119728723,
      "loss": 2.0173,
      "step": 16812
    },
    {
      "epoch": 1.44652843499957,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008076933982315438,
      "loss": 0.8713,
      "step": 16813
    },
    {
      "epoch": 1.4466144713068916,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008076720835902721,
      "loss": 1.1487,
      "step": 16814
    },
    {
      "epoch": 1.4467005076142132,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008076507680491195,
      "loss": 1.4068,
      "step": 16815
    },
    {
      "epoch": 1.4467865439215348,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000807629451608148,
      "loss": 1.7069,
      "step": 16816
    },
    {
      "epoch": 1.4468725802288565,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0008076081342674203,
      "loss": 1.2706,
      "step": 16817
    },
    {
      "epoch": 1.4469586165361783,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008075868160269987,
      "loss": 0.738,
      "step": 16818
    },
    {
      "epoch": 1.4470446528435,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0008075654968869453,
      "loss": 1.1795,
      "step": 16819
    },
    {
      "epoch": 1.4471306891508218,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008075441768473228,
      "loss": 1.3518,
      "step": 16820
    },
    {
      "epoch": 1.4472167254581434,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008075228559081934,
      "loss": 1.0548,
      "step": 16821
    },
    {
      "epoch": 1.447302761765465,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0008075015340696194,
      "loss": 1.2904,
      "step": 16822
    },
    {
      "epoch": 1.4473887980727866,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008074802113316633,
      "loss": 0.9636,
      "step": 16823
    },
    {
      "epoch": 1.4474748343801085,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008074588876943872,
      "loss": 1.7453,
      "step": 16824
    },
    {
      "epoch": 1.4475608706874301,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0008074375631578538,
      "loss": 1.1796,
      "step": 16825
    },
    {
      "epoch": 1.4476469069947517,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008074162377221252,
      "loss": 1.0479,
      "step": 16826
    },
    {
      "epoch": 1.4477329433020736,
      "grad_norm": 1.0,
      "learning_rate": 0.0008073949113872641,
      "loss": 1.128,
      "step": 16827
    },
    {
      "epoch": 1.4478189796093952,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008073735841533325,
      "loss": 0.8797,
      "step": 16828
    },
    {
      "epoch": 1.4479050159167168,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008073522560203929,
      "loss": 1.8534,
      "step": 16829
    },
    {
      "epoch": 1.4479910522240385,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008073309269885079,
      "loss": 1.8475,
      "step": 16830
    },
    {
      "epoch": 1.4480770885313603,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008073095970577395,
      "loss": 1.412,
      "step": 16831
    },
    {
      "epoch": 1.448163124838682,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008072882662281505,
      "loss": 1.5269,
      "step": 16832
    },
    {
      "epoch": 1.4482491611460035,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008072669344998031,
      "loss": 1.2584,
      "step": 16833
    },
    {
      "epoch": 1.4483351974533254,
      "grad_norm": 1.84375,
      "learning_rate": 0.0008072456018727595,
      "loss": 1.2035,
      "step": 16834
    },
    {
      "epoch": 1.448421233760647,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008072242683470826,
      "loss": 1.0653,
      "step": 16835
    },
    {
      "epoch": 1.4485072700679686,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008072029339228341,
      "loss": 1.1961,
      "step": 16836
    },
    {
      "epoch": 1.4485933063752903,
      "grad_norm": 3.15625,
      "learning_rate": 0.0008071815986000769,
      "loss": 2.2048,
      "step": 16837
    },
    {
      "epoch": 1.4486793426826121,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008071602623788732,
      "loss": 1.3664,
      "step": 16838
    },
    {
      "epoch": 1.4487653789899337,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008071389252592855,
      "loss": 1.9381,
      "step": 16839
    },
    {
      "epoch": 1.4488514152972554,
      "grad_norm": 0.875,
      "learning_rate": 0.0008071175872413763,
      "loss": 1.8709,
      "step": 16840
    },
    {
      "epoch": 1.4489374516045772,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008070962483252078,
      "loss": 1.127,
      "step": 16841
    },
    {
      "epoch": 1.4490234879118988,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008070749085108424,
      "loss": 2.0023,
      "step": 16842
    },
    {
      "epoch": 1.4491095242192205,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008070535677983428,
      "loss": 0.9869,
      "step": 16843
    },
    {
      "epoch": 1.449195560526542,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008070322261877712,
      "loss": 1.3403,
      "step": 16844
    },
    {
      "epoch": 1.449281596833864,
      "grad_norm": 1.125,
      "learning_rate": 0.00080701088367919,
      "loss": 1.1824,
      "step": 16845
    },
    {
      "epoch": 1.4493676331411856,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008069895402726617,
      "loss": 1.2571,
      "step": 16846
    },
    {
      "epoch": 1.4494536694485074,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008069681959682487,
      "loss": 0.8795,
      "step": 16847
    },
    {
      "epoch": 1.449539705755829,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0008069468507660133,
      "loss": 1.112,
      "step": 16848
    },
    {
      "epoch": 1.4496257420631506,
      "grad_norm": 1.640625,
      "learning_rate": 0.0008069255046660181,
      "loss": 1.3001,
      "step": 16849
    },
    {
      "epoch": 1.4497117783704723,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008069041576683256,
      "loss": 0.9194,
      "step": 16850
    },
    {
      "epoch": 1.449797814677794,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0008068828097729981,
      "loss": 1.2441,
      "step": 16851
    },
    {
      "epoch": 1.4498838509851157,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000806861460980098,
      "loss": 1.3868,
      "step": 16852
    },
    {
      "epoch": 1.4499698872924374,
      "grad_norm": 1.125,
      "learning_rate": 0.0008068401112896879,
      "loss": 1.0826,
      "step": 16853
    },
    {
      "epoch": 1.4500559235997592,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008068187607018301,
      "loss": 1.2116,
      "step": 16854
    },
    {
      "epoch": 1.4501419599070808,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008067974092165872,
      "loss": 1.284,
      "step": 16855
    },
    {
      "epoch": 1.4502279962144025,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008067760568340215,
      "loss": 1.2081,
      "step": 16856
    },
    {
      "epoch": 1.450314032521724,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0008067547035541953,
      "loss": 1.1733,
      "step": 16857
    },
    {
      "epoch": 1.450400068829046,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008067333493771714,
      "loss": 1.1488,
      "step": 16858
    },
    {
      "epoch": 1.4504861051363676,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008067119943030122,
      "loss": 1.2273,
      "step": 16859
    },
    {
      "epoch": 1.4505721414436892,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00080669063833178,
      "loss": 1.3465,
      "step": 16860
    },
    {
      "epoch": 1.450658177751011,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008066692814635374,
      "loss": 1.1537,
      "step": 16861
    },
    {
      "epoch": 1.4507442140583326,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008066479236983466,
      "loss": 1.5275,
      "step": 16862
    },
    {
      "epoch": 1.4508302503656543,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008066265650362706,
      "loss": 1.1036,
      "step": 16863
    },
    {
      "epoch": 1.450916286672976,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008066052054773712,
      "loss": 0.9143,
      "step": 16864
    },
    {
      "epoch": 1.4510023229802977,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0008065838450217112,
      "loss": 1.2772,
      "step": 16865
    },
    {
      "epoch": 1.4510883592876194,
      "grad_norm": 1.4375,
      "learning_rate": 0.0008065624836693533,
      "loss": 1.6683,
      "step": 16866
    },
    {
      "epoch": 1.451174395594941,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008065411214203595,
      "loss": 0.9465,
      "step": 16867
    },
    {
      "epoch": 1.4512604319022628,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008065197582747928,
      "loss": 1.6238,
      "step": 16868
    },
    {
      "epoch": 1.4513464682095845,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008064983942327153,
      "loss": 1.7999,
      "step": 16869
    },
    {
      "epoch": 1.451432504516906,
      "grad_norm": 1.953125,
      "learning_rate": 0.0008064770292941897,
      "loss": 1.8957,
      "step": 16870
    },
    {
      "epoch": 1.4515185408242277,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008064556634592782,
      "loss": 1.0637,
      "step": 16871
    },
    {
      "epoch": 1.4516045771315496,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008064342967280436,
      "loss": 0.968,
      "step": 16872
    },
    {
      "epoch": 1.4516906134388712,
      "grad_norm": 0.90625,
      "learning_rate": 0.0008064129291005483,
      "loss": 0.6305,
      "step": 16873
    },
    {
      "epoch": 1.451776649746193,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0008063915605768548,
      "loss": 0.7782,
      "step": 16874
    },
    {
      "epoch": 1.4518626860535147,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008063701911570255,
      "loss": 1.3429,
      "step": 16875
    },
    {
      "epoch": 1.4519487223608363,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008063488208411231,
      "loss": 1.1203,
      "step": 16876
    },
    {
      "epoch": 1.452034758668158,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008063274496292099,
      "loss": 1.9817,
      "step": 16877
    },
    {
      "epoch": 1.4521207949754795,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008063060775213484,
      "loss": 1.2923,
      "step": 16878
    },
    {
      "epoch": 1.4522068312828014,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008062847045176013,
      "loss": 1.3334,
      "step": 16879
    },
    {
      "epoch": 1.452292867590123,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008062633306180309,
      "loss": 1.1731,
      "step": 16880
    },
    {
      "epoch": 1.4523789038974448,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008062419558227,
      "loss": 1.0533,
      "step": 16881
    },
    {
      "epoch": 1.4524649402047665,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008062205801316709,
      "loss": 1.1083,
      "step": 16882
    },
    {
      "epoch": 1.452550976512088,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008061992035450061,
      "loss": 1.866,
      "step": 16883
    },
    {
      "epoch": 1.4526370128194097,
      "grad_norm": 1.5,
      "learning_rate": 0.0008061778260627681,
      "loss": 1.1426,
      "step": 16884
    },
    {
      "epoch": 1.4527230491267316,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008061564476850196,
      "loss": 1.1262,
      "step": 16885
    },
    {
      "epoch": 1.4528090854340532,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000806135068411823,
      "loss": 0.9666,
      "step": 16886
    },
    {
      "epoch": 1.4528951217413748,
      "grad_norm": 1.140625,
      "learning_rate": 0.000806113688243241,
      "loss": 1.3266,
      "step": 16887
    },
    {
      "epoch": 1.4529811580486967,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008060923071793358,
      "loss": 1.5322,
      "step": 16888
    },
    {
      "epoch": 1.4530671943560183,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008060709252201703,
      "loss": 1.3362,
      "step": 16889
    },
    {
      "epoch": 1.45315323066334,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008060495423658068,
      "loss": 1.207,
      "step": 16890
    },
    {
      "epoch": 1.4532392669706615,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008060281586163079,
      "loss": 1.6316,
      "step": 16891
    },
    {
      "epoch": 1.4533253032779834,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0008060067739717362,
      "loss": 1.159,
      "step": 16892
    },
    {
      "epoch": 1.453411339585305,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008059853884321542,
      "loss": 1.3942,
      "step": 16893
    },
    {
      "epoch": 1.4534973758926266,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008059640019976243,
      "loss": 1.99,
      "step": 16894
    },
    {
      "epoch": 1.4535834121999485,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008059426146682094,
      "loss": 1.1271,
      "step": 16895
    },
    {
      "epoch": 1.45366944850727,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008059212264439717,
      "loss": 1.3538,
      "step": 16896
    },
    {
      "epoch": 1.4537554848145917,
      "grad_norm": 1.3125,
      "learning_rate": 0.000805899837324974,
      "loss": 1.3055,
      "step": 16897
    },
    {
      "epoch": 1.4538415211219133,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008058784473112788,
      "loss": 1.4814,
      "step": 16898
    },
    {
      "epoch": 1.4539275574292352,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008058570564029486,
      "loss": 1.5388,
      "step": 16899
    },
    {
      "epoch": 1.4540135937365568,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000805835664600046,
      "loss": 1.6029,
      "step": 16900
    },
    {
      "epoch": 1.4540996300438787,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0008058142719026335,
      "loss": 0.928,
      "step": 16901
    },
    {
      "epoch": 1.4541856663512003,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008057928783107738,
      "loss": 0.9201,
      "step": 16902
    },
    {
      "epoch": 1.454271702658522,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008057714838245293,
      "loss": 1.2117,
      "step": 16903
    },
    {
      "epoch": 1.4543577389658435,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008057500884439628,
      "loss": 1.7701,
      "step": 16904
    },
    {
      "epoch": 1.4544437752731652,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008057286921691367,
      "loss": 1.891,
      "step": 16905
    },
    {
      "epoch": 1.454529811580487,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008057072950001137,
      "loss": 1.7086,
      "step": 16906
    },
    {
      "epoch": 1.4546158478878086,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008056858969369562,
      "loss": 1.2438,
      "step": 16907
    },
    {
      "epoch": 1.4547018841951305,
      "grad_norm": 1.140625,
      "learning_rate": 0.000805664497979727,
      "loss": 1.2624,
      "step": 16908
    },
    {
      "epoch": 1.454787920502452,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008056430981284886,
      "loss": 1.2291,
      "step": 16909
    },
    {
      "epoch": 1.4548739568097737,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008056216973833035,
      "loss": 1.1159,
      "step": 16910
    },
    {
      "epoch": 1.4549599931170953,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008056002957442344,
      "loss": 1.6251,
      "step": 16911
    },
    {
      "epoch": 1.4550460294244172,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008055788932113439,
      "loss": 1.4135,
      "step": 16912
    },
    {
      "epoch": 1.4551320657317388,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008055574897846947,
      "loss": 2.0021,
      "step": 16913
    },
    {
      "epoch": 1.4552181020390604,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000805536085464349,
      "loss": 1.5615,
      "step": 16914
    },
    {
      "epoch": 1.4553041383463823,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008055146802503699,
      "loss": 0.8071,
      "step": 16915
    },
    {
      "epoch": 1.455390174653704,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008054932741428196,
      "loss": 1.2827,
      "step": 16916
    },
    {
      "epoch": 1.4554762109610255,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008054718671417612,
      "loss": 0.9074,
      "step": 16917
    },
    {
      "epoch": 1.4555622472683472,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008054504592472566,
      "loss": 1.108,
      "step": 16918
    },
    {
      "epoch": 1.455648283575669,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008054290504593691,
      "loss": 0.8005,
      "step": 16919
    },
    {
      "epoch": 1.4557343198829906,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008054076407781608,
      "loss": 0.6089,
      "step": 16920
    },
    {
      "epoch": 1.4558203561903122,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008053862302036947,
      "loss": 0.9125,
      "step": 16921
    },
    {
      "epoch": 1.455906392497634,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008053648187360332,
      "loss": 1.3902,
      "step": 16922
    },
    {
      "epoch": 1.4559924288049557,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000805343406375239,
      "loss": 0.8403,
      "step": 16923
    },
    {
      "epoch": 1.4560784651122773,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008053219931213748,
      "loss": 1.5435,
      "step": 16924
    },
    {
      "epoch": 1.456164501419599,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0008053005789745031,
      "loss": 1.5202,
      "step": 16925
    },
    {
      "epoch": 1.4562505377269208,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008052791639346866,
      "loss": 1.2015,
      "step": 16926
    },
    {
      "epoch": 1.4563365740342424,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008052577480019879,
      "loss": 1.2206,
      "step": 16927
    },
    {
      "epoch": 1.456422610341564,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008052363311764697,
      "loss": 0.8727,
      "step": 16928
    },
    {
      "epoch": 1.456508646648886,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008052149134581946,
      "loss": 1.1644,
      "step": 16929
    },
    {
      "epoch": 1.4565946829562075,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008051934948472252,
      "loss": 1.2791,
      "step": 16930
    },
    {
      "epoch": 1.4566807192635292,
      "grad_norm": 1.671875,
      "learning_rate": 0.0008051720753436242,
      "loss": 2.316,
      "step": 16931
    },
    {
      "epoch": 1.4567667555708508,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008051506549474542,
      "loss": 0.958,
      "step": 16932
    },
    {
      "epoch": 1.4568527918781726,
      "grad_norm": 2.140625,
      "learning_rate": 0.0008051292336587779,
      "loss": 2.2868,
      "step": 16933
    },
    {
      "epoch": 1.4569388281854943,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000805107811477658,
      "loss": 1.6721,
      "step": 16934
    },
    {
      "epoch": 1.457024864492816,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000805086388404157,
      "loss": 1.5743,
      "step": 16935
    },
    {
      "epoch": 1.4571109008001377,
      "grad_norm": 2.25,
      "learning_rate": 0.0008050649644383377,
      "loss": 1.2691,
      "step": 16936
    },
    {
      "epoch": 1.4571969371074593,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008050435395802627,
      "loss": 1.0396,
      "step": 16937
    },
    {
      "epoch": 1.457282973414781,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008050221138299947,
      "loss": 0.7185,
      "step": 16938
    },
    {
      "epoch": 1.4573690097221026,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008050006871875963,
      "loss": 1.7961,
      "step": 16939
    },
    {
      "epoch": 1.4574550460294244,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008049792596531302,
      "loss": 1.3522,
      "step": 16940
    },
    {
      "epoch": 1.457541082336746,
      "grad_norm": 1.125,
      "learning_rate": 0.000804957831226659,
      "loss": 1.7707,
      "step": 16941
    },
    {
      "epoch": 1.457627118644068,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008049364019082455,
      "loss": 0.8602,
      "step": 16942
    },
    {
      "epoch": 1.4577131549513895,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0008049149716979525,
      "loss": 1.3557,
      "step": 16943
    },
    {
      "epoch": 1.4577991912587112,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008048935405958423,
      "loss": 1.6386,
      "step": 16944
    },
    {
      "epoch": 1.4578852275660328,
      "grad_norm": 1.125,
      "learning_rate": 0.0008048721086019779,
      "loss": 1.8037,
      "step": 16945
    },
    {
      "epoch": 1.4579712638733546,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008048506757164219,
      "loss": 1.1982,
      "step": 16946
    },
    {
      "epoch": 1.4580573001806763,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0008048292419392369,
      "loss": 1.2018,
      "step": 16947
    },
    {
      "epoch": 1.4581433364879979,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008048078072704856,
      "loss": 1.498,
      "step": 16948
    },
    {
      "epoch": 1.4582293727953197,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000804786371710231,
      "loss": 1.0069,
      "step": 16949
    },
    {
      "epoch": 1.4583154091026413,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008047649352585354,
      "loss": 0.7172,
      "step": 16950
    },
    {
      "epoch": 1.458401445409963,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0008047434979154615,
      "loss": 0.9564,
      "step": 16951
    },
    {
      "epoch": 1.4584874817172846,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008047220596810723,
      "loss": 0.7414,
      "step": 16952
    },
    {
      "epoch": 1.4585735180246064,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008047006205554303,
      "loss": 1.1289,
      "step": 16953
    },
    {
      "epoch": 1.458659554331928,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008046791805385983,
      "loss": 2.1818,
      "step": 16954
    },
    {
      "epoch": 1.4587455906392497,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0008046577396306387,
      "loss": 1.6779,
      "step": 16955
    },
    {
      "epoch": 1.4588316269465715,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008046362978316147,
      "loss": 0.7492,
      "step": 16956
    },
    {
      "epoch": 1.4589176632538932,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008046148551415888,
      "loss": 0.9524,
      "step": 16957
    },
    {
      "epoch": 1.4590036995612148,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008045934115606236,
      "loss": 0.7272,
      "step": 16958
    },
    {
      "epoch": 1.4590897358685364,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008045719670887819,
      "loss": 0.971,
      "step": 16959
    },
    {
      "epoch": 1.4591757721758583,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008045505217261264,
      "loss": 1.0177,
      "step": 16960
    },
    {
      "epoch": 1.4592618084831799,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008045290754727199,
      "loss": 0.6931,
      "step": 16961
    },
    {
      "epoch": 1.4593478447905017,
      "grad_norm": 1.25,
      "learning_rate": 0.0008045076283286251,
      "loss": 1.5018,
      "step": 16962
    },
    {
      "epoch": 1.4594338810978233,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008044861802939048,
      "loss": 1.6065,
      "step": 16963
    },
    {
      "epoch": 1.459519917405145,
      "grad_norm": 0.984375,
      "learning_rate": 0.0008044647313686215,
      "loss": 1.0364,
      "step": 16964
    },
    {
      "epoch": 1.4596059537124666,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008044432815528381,
      "loss": 1.4863,
      "step": 16965
    },
    {
      "epoch": 1.4596919900197882,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008044218308466174,
      "loss": 1.3801,
      "step": 16966
    },
    {
      "epoch": 1.45977802632711,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0008044003792500219,
      "loss": 1.2765,
      "step": 16967
    },
    {
      "epoch": 1.4598640626344317,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008043789267631146,
      "loss": 0.7618,
      "step": 16968
    },
    {
      "epoch": 1.4599500989417535,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008043574733859583,
      "loss": 1.3667,
      "step": 16969
    },
    {
      "epoch": 1.4600361352490752,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008043360191186153,
      "loss": 1.2433,
      "step": 16970
    },
    {
      "epoch": 1.4601221715563968,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008043145639611487,
      "loss": 1.6075,
      "step": 16971
    },
    {
      "epoch": 1.4602082078637184,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0008042931079136213,
      "loss": 0.4306,
      "step": 16972
    },
    {
      "epoch": 1.4602942441710403,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008042716509760957,
      "loss": 0.9912,
      "step": 16973
    },
    {
      "epoch": 1.4603802804783619,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008042501931486345,
      "loss": 1.2945,
      "step": 16974
    },
    {
      "epoch": 1.4604663167856835,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000804228734431301,
      "loss": 1.0738,
      "step": 16975
    },
    {
      "epoch": 1.4605523530930054,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008042072748241574,
      "loss": 0.9524,
      "step": 16976
    },
    {
      "epoch": 1.460638389400327,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008041858143272667,
      "loss": 1.4492,
      "step": 16977
    },
    {
      "epoch": 1.4607244257076486,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008041643529406917,
      "loss": 1.2103,
      "step": 16978
    },
    {
      "epoch": 1.4608104620149702,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008041428906644949,
      "loss": 1.5435,
      "step": 16979
    },
    {
      "epoch": 1.460896498322292,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008041214274987396,
      "loss": 0.785,
      "step": 16980
    },
    {
      "epoch": 1.4609825346296137,
      "grad_norm": 1.125,
      "learning_rate": 0.0008040999634434883,
      "loss": 1.571,
      "step": 16981
    },
    {
      "epoch": 1.4610685709369353,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008040784984988035,
      "loss": 1.713,
      "step": 16982
    },
    {
      "epoch": 1.4611546072442572,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008040570326647485,
      "loss": 1.3294,
      "step": 16983
    },
    {
      "epoch": 1.4612406435515788,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0008040355659413856,
      "loss": 1.5615,
      "step": 16984
    },
    {
      "epoch": 1.4613266798589004,
      "grad_norm": 1.5625,
      "learning_rate": 0.0008040140983287778,
      "loss": 1.3241,
      "step": 16985
    },
    {
      "epoch": 1.461412716166222,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008039926298269881,
      "loss": 1.1613,
      "step": 16986
    },
    {
      "epoch": 1.4614987524735439,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008039711604360789,
      "loss": 2.122,
      "step": 16987
    },
    {
      "epoch": 1.4615847887808655,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008039496901561132,
      "loss": 1.25,
      "step": 16988
    },
    {
      "epoch": 1.4616708250881874,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008039282189871538,
      "loss": 1.7561,
      "step": 16989
    },
    {
      "epoch": 1.461756861395509,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008039067469292636,
      "loss": 1.8585,
      "step": 16990
    },
    {
      "epoch": 1.4618428977028306,
      "grad_norm": 1.125,
      "learning_rate": 0.0008038852739825051,
      "loss": 1.8143,
      "step": 16991
    },
    {
      "epoch": 1.4619289340101522,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008038638001469413,
      "loss": 1.0033,
      "step": 16992
    },
    {
      "epoch": 1.4620149703174738,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000803842325422635,
      "loss": 0.9897,
      "step": 16993
    },
    {
      "epoch": 1.4621010066247957,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000803820849809649,
      "loss": 1.987,
      "step": 16994
    },
    {
      "epoch": 1.4621870429321173,
      "grad_norm": 1.484375,
      "learning_rate": 0.000803799373308046,
      "loss": 1.4138,
      "step": 16995
    },
    {
      "epoch": 1.4622730792394392,
      "grad_norm": 1.1875,
      "learning_rate": 0.000803777895917889,
      "loss": 1.6777,
      "step": 16996
    },
    {
      "epoch": 1.4623591155467608,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008037564176392408,
      "loss": 1.6057,
      "step": 16997
    },
    {
      "epoch": 1.4624451518540824,
      "grad_norm": 1.984375,
      "learning_rate": 0.0008037349384721642,
      "loss": 1.396,
      "step": 16998
    },
    {
      "epoch": 1.462531188161404,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008037134584167219,
      "loss": 1.1708,
      "step": 16999
    },
    {
      "epoch": 1.4626172244687259,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008036919774729766,
      "loss": 0.7682,
      "step": 17000
    },
    {
      "epoch": 1.4627032607760475,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008036704956409916,
      "loss": 1.0673,
      "step": 17001
    },
    {
      "epoch": 1.4627892970833691,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008036490129208294,
      "loss": 1.3068,
      "step": 17002
    },
    {
      "epoch": 1.462875333390691,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008036275293125528,
      "loss": 0.7887,
      "step": 17003
    },
    {
      "epoch": 1.4629613696980126,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008036060448162247,
      "loss": 2.226,
      "step": 17004
    },
    {
      "epoch": 1.4630474060053342,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008035845594319082,
      "loss": 0.892,
      "step": 17005
    },
    {
      "epoch": 1.4631334423126559,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008035630731596657,
      "loss": 1.1463,
      "step": 17006
    },
    {
      "epoch": 1.4632194786199777,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008035415859995605,
      "loss": 1.415,
      "step": 17007
    },
    {
      "epoch": 1.4633055149272993,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008035200979516549,
      "loss": 1.2419,
      "step": 17008
    },
    {
      "epoch": 1.463391551234621,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0008034986090160121,
      "loss": 1.4534,
      "step": 17009
    },
    {
      "epoch": 1.4634775875419428,
      "grad_norm": 1.09375,
      "learning_rate": 0.000803477119192695,
      "loss": 1.7207,
      "step": 17010
    },
    {
      "epoch": 1.4635636238492644,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008034556284817664,
      "loss": 1.0193,
      "step": 17011
    },
    {
      "epoch": 1.463649660156586,
      "grad_norm": 1.109375,
      "learning_rate": 0.000803434136883289,
      "loss": 1.7125,
      "step": 17012
    },
    {
      "epoch": 1.4637356964639077,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008034126443973257,
      "loss": 0.9031,
      "step": 17013
    },
    {
      "epoch": 1.4638217327712295,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008033911510239395,
      "loss": 1.7892,
      "step": 17014
    },
    {
      "epoch": 1.4639077690785511,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008033696567631932,
      "loss": 1.9905,
      "step": 17015
    },
    {
      "epoch": 1.4639938053858728,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008033481616151497,
      "loss": 1.5386,
      "step": 17016
    },
    {
      "epoch": 1.4640798416931946,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008033266655798717,
      "loss": 1.4573,
      "step": 17017
    },
    {
      "epoch": 1.4641658780005162,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008033051686574224,
      "loss": 1.242,
      "step": 17018
    },
    {
      "epoch": 1.4642519143078379,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0008032836708478644,
      "loss": 0.8352,
      "step": 17019
    },
    {
      "epoch": 1.4643379506151595,
      "grad_norm": 1.125,
      "learning_rate": 0.0008032621721512605,
      "loss": 1.2981,
      "step": 17020
    },
    {
      "epoch": 1.4644239869224813,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008032406725676737,
      "loss": 1.7817,
      "step": 17021
    },
    {
      "epoch": 1.464510023229803,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008032191720971671,
      "loss": 1.6264,
      "step": 17022
    },
    {
      "epoch": 1.4645960595371248,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008031976707398034,
      "loss": 0.7351,
      "step": 17023
    },
    {
      "epoch": 1.4646820958444464,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008031761684956453,
      "loss": 1.2197,
      "step": 17024
    },
    {
      "epoch": 1.464768132151768,
      "grad_norm": 1.546875,
      "learning_rate": 0.0008031546653647562,
      "loss": 0.7383,
      "step": 17025
    },
    {
      "epoch": 1.4648541684590897,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008031331613471983,
      "loss": 1.4256,
      "step": 17026
    },
    {
      "epoch": 1.4649402047664113,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000803111656443035,
      "loss": 1.9249,
      "step": 17027
    },
    {
      "epoch": 1.4650262410737331,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008030901506523292,
      "loss": 0.9346,
      "step": 17028
    },
    {
      "epoch": 1.4651122773810548,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0008030686439751435,
      "loss": 1.8512,
      "step": 17029
    },
    {
      "epoch": 1.4651983136883766,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008030471364115409,
      "loss": 1.9037,
      "step": 17030
    },
    {
      "epoch": 1.4652843499956982,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0008030256279615844,
      "loss": 1.5151,
      "step": 17031
    },
    {
      "epoch": 1.4653703863030199,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0008030041186253369,
      "loss": 0.7517,
      "step": 17032
    },
    {
      "epoch": 1.4654564226103415,
      "grad_norm": 1.0,
      "learning_rate": 0.0008029826084028613,
      "loss": 0.6976,
      "step": 17033
    },
    {
      "epoch": 1.4655424589176633,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008029610972942205,
      "loss": 1.9335,
      "step": 17034
    },
    {
      "epoch": 1.465628495224985,
      "grad_norm": 1.5,
      "learning_rate": 0.0008029395852994774,
      "loss": 0.7618,
      "step": 17035
    },
    {
      "epoch": 1.4657145315323066,
      "grad_norm": 0.90625,
      "learning_rate": 0.000802918072418695,
      "loss": 0.7364,
      "step": 17036
    },
    {
      "epoch": 1.4658005678396284,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008028965586519361,
      "loss": 1.8757,
      "step": 17037
    },
    {
      "epoch": 1.46588660414695,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008028750439992636,
      "loss": 1.9181,
      "step": 17038
    },
    {
      "epoch": 1.4659726404542717,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008028535284607406,
      "loss": 1.3991,
      "step": 17039
    },
    {
      "epoch": 1.4660586767615933,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0008028320120364299,
      "loss": 1.1073,
      "step": 17040
    },
    {
      "epoch": 1.4661447130689151,
      "grad_norm": 1.515625,
      "learning_rate": 0.0008028104947263946,
      "loss": 1.8089,
      "step": 17041
    },
    {
      "epoch": 1.4662307493762368,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008027889765306973,
      "loss": 1.1148,
      "step": 17042
    },
    {
      "epoch": 1.4663167856835584,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0008027674574494012,
      "loss": 1.2352,
      "step": 17043
    },
    {
      "epoch": 1.4664028219908802,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008027459374825692,
      "loss": 0.9973,
      "step": 17044
    },
    {
      "epoch": 1.4664888582982019,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000802724416630264,
      "loss": 1.4089,
      "step": 17045
    },
    {
      "epoch": 1.4665748946055235,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008027028948925491,
      "loss": 1.8417,
      "step": 17046
    },
    {
      "epoch": 1.466660930912845,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008026813722694869,
      "loss": 0.9892,
      "step": 17047
    },
    {
      "epoch": 1.466746967220167,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008026598487611406,
      "loss": 1.4936,
      "step": 17048
    },
    {
      "epoch": 1.4668330035274886,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008026383243675731,
      "loss": 0.8538,
      "step": 17049
    },
    {
      "epoch": 1.4669190398348104,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008026167990888473,
      "loss": 1.4046,
      "step": 17050
    },
    {
      "epoch": 1.467005076142132,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008025952729250264,
      "loss": 0.8,
      "step": 17051
    },
    {
      "epoch": 1.4670911124494537,
      "grad_norm": 1.375,
      "learning_rate": 0.000802573745876173,
      "loss": 1.6917,
      "step": 17052
    },
    {
      "epoch": 1.4671771487567753,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008025522179423504,
      "loss": 1.2334,
      "step": 17053
    },
    {
      "epoch": 1.467263185064097,
      "grad_norm": 1.125,
      "learning_rate": 0.0008025306891236212,
      "loss": 0.9848,
      "step": 17054
    },
    {
      "epoch": 1.4673492213714188,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008025091594200486,
      "loss": 0.7883,
      "step": 17055
    },
    {
      "epoch": 1.4674352576787404,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0008024876288316956,
      "loss": 1.7534,
      "step": 17056
    },
    {
      "epoch": 1.4675212939860622,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008024660973586253,
      "loss": 0.7482,
      "step": 17057
    },
    {
      "epoch": 1.4676073302933839,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0008024445650009003,
      "loss": 1.2777,
      "step": 17058
    },
    {
      "epoch": 1.4676933666007055,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0008024230317585839,
      "loss": 1.353,
      "step": 17059
    },
    {
      "epoch": 1.467779402908027,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0008024014976317388,
      "loss": 1.4763,
      "step": 17060
    },
    {
      "epoch": 1.467865439215349,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008023799626204282,
      "loss": 1.6776,
      "step": 17061
    },
    {
      "epoch": 1.4679514755226706,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008023584267247149,
      "loss": 1.8232,
      "step": 17062
    },
    {
      "epoch": 1.4680375118299922,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008023368899446622,
      "loss": 0.6761,
      "step": 17063
    },
    {
      "epoch": 1.468123548137314,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008023153522803328,
      "loss": 1.4101,
      "step": 17064
    },
    {
      "epoch": 1.4682095844446357,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008022938137317898,
      "loss": 1.2742,
      "step": 17065
    },
    {
      "epoch": 1.4682956207519573,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008022722742990962,
      "loss": 1.0066,
      "step": 17066
    },
    {
      "epoch": 1.468381657059279,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008022507339823149,
      "loss": 0.8434,
      "step": 17067
    },
    {
      "epoch": 1.4684676933666008,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008022291927815091,
      "loss": 1.3167,
      "step": 17068
    },
    {
      "epoch": 1.4685537296739224,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008022076506967417,
      "loss": 1.7992,
      "step": 17069
    },
    {
      "epoch": 1.468639765981244,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0008021861077280757,
      "loss": 0.6217,
      "step": 17070
    },
    {
      "epoch": 1.4687258022885659,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008021645638755739,
      "loss": 1.3665,
      "step": 17071
    },
    {
      "epoch": 1.4688118385958875,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0008021430191392998,
      "loss": 1.174,
      "step": 17072
    },
    {
      "epoch": 1.468897874903209,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008021214735193158,
      "loss": 1.2579,
      "step": 17073
    },
    {
      "epoch": 1.4689839112105307,
      "grad_norm": 2.0625,
      "learning_rate": 0.0008020999270156857,
      "loss": 1.4263,
      "step": 17074
    },
    {
      "epoch": 1.4690699475178526,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008020783796284718,
      "loss": 0.9277,
      "step": 17075
    },
    {
      "epoch": 1.4691559838251742,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008020568313577374,
      "loss": 0.6934,
      "step": 17076
    },
    {
      "epoch": 1.469242020132496,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008020352822035455,
      "loss": 1.4636,
      "step": 17077
    },
    {
      "epoch": 1.4693280564398177,
      "grad_norm": 1.0625,
      "learning_rate": 0.0008020137321659592,
      "loss": 1.2253,
      "step": 17078
    },
    {
      "epoch": 1.4694140927471393,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008019921812450415,
      "loss": 1.0554,
      "step": 17079
    },
    {
      "epoch": 1.469500129054461,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0008019706294408553,
      "loss": 0.6168,
      "step": 17080
    },
    {
      "epoch": 1.4695861653617825,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008019490767534639,
      "loss": 1.0559,
      "step": 17081
    },
    {
      "epoch": 1.4696722016691044,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008019275231829299,
      "loss": 1.1864,
      "step": 17082
    },
    {
      "epoch": 1.469758237976426,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0008019059687293168,
      "loss": 1.6239,
      "step": 17083
    },
    {
      "epoch": 1.4698442742837479,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008018844133926875,
      "loss": 0.6054,
      "step": 17084
    },
    {
      "epoch": 1.4699303105910695,
      "grad_norm": 1.1875,
      "learning_rate": 0.000801862857173105,
      "loss": 1.5095,
      "step": 17085
    },
    {
      "epoch": 1.4700163468983911,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008018413000706323,
      "loss": 1.938,
      "step": 17086
    },
    {
      "epoch": 1.4701023832057127,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0008018197420853325,
      "loss": 0.9065,
      "step": 17087
    },
    {
      "epoch": 1.4701884195130346,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008017981832172686,
      "loss": 1.3353,
      "step": 17088
    },
    {
      "epoch": 1.4702744558203562,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008017766234665037,
      "loss": 0.8856,
      "step": 17089
    },
    {
      "epoch": 1.4703604921276778,
      "grad_norm": 0.96875,
      "learning_rate": 0.0008017550628331009,
      "loss": 0.7806,
      "step": 17090
    },
    {
      "epoch": 1.4704465284349997,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0008017335013171235,
      "loss": 1.7195,
      "step": 17091
    },
    {
      "epoch": 1.4705325647423213,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008017119389186338,
      "loss": 0.9829,
      "step": 17092
    },
    {
      "epoch": 1.470618601049643,
      "grad_norm": 1.46875,
      "learning_rate": 0.0008016903756376956,
      "loss": 1.5286,
      "step": 17093
    },
    {
      "epoch": 1.4707046373569646,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0008016688114743719,
      "loss": 1.5358,
      "step": 17094
    },
    {
      "epoch": 1.4707906736642864,
      "grad_norm": 2.078125,
      "learning_rate": 0.0008016472464287255,
      "loss": 1.1306,
      "step": 17095
    },
    {
      "epoch": 1.470876709971608,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008016256805008194,
      "loss": 0.9437,
      "step": 17096
    },
    {
      "epoch": 1.4709627462789296,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008016041136907169,
      "loss": 0.8757,
      "step": 17097
    },
    {
      "epoch": 1.4710487825862515,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0008015825459984811,
      "loss": 1.5566,
      "step": 17098
    },
    {
      "epoch": 1.4711348188935731,
      "grad_norm": 1.25,
      "learning_rate": 0.000801560977424175,
      "loss": 1.853,
      "step": 17099
    },
    {
      "epoch": 1.4712208552008947,
      "grad_norm": 2.21875,
      "learning_rate": 0.0008015394079678617,
      "loss": 1.6489,
      "step": 17100
    },
    {
      "epoch": 1.4713068915082164,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008015178376296044,
      "loss": 1.0328,
      "step": 17101
    },
    {
      "epoch": 1.4713929278155382,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008014962664094658,
      "loss": 1.3071,
      "step": 17102
    },
    {
      "epoch": 1.4714789641228598,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008014746943075095,
      "loss": 1.4437,
      "step": 17103
    },
    {
      "epoch": 1.4715650004301815,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008014531213237983,
      "loss": 2.0561,
      "step": 17104
    },
    {
      "epoch": 1.4716510367375033,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0008014315474583952,
      "loss": 0.9289,
      "step": 17105
    },
    {
      "epoch": 1.471737073044825,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008014099727113635,
      "loss": 1.1109,
      "step": 17106
    },
    {
      "epoch": 1.4718231093521466,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008013883970827662,
      "loss": 1.1311,
      "step": 17107
    },
    {
      "epoch": 1.4719091456594682,
      "grad_norm": 47.0,
      "learning_rate": 0.0008013668205726666,
      "loss": 1.431,
      "step": 17108
    },
    {
      "epoch": 1.47199518196679,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008013452431811277,
      "loss": 2.0512,
      "step": 17109
    },
    {
      "epoch": 1.4720812182741116,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008013236649082124,
      "loss": 0.6675,
      "step": 17110
    },
    {
      "epoch": 1.4721672545814335,
      "grad_norm": 34.5,
      "learning_rate": 0.0008013020857539841,
      "loss": 0.732,
      "step": 17111
    },
    {
      "epoch": 1.4722532908887551,
      "grad_norm": 1.5,
      "learning_rate": 0.0008012805057185057,
      "loss": 1.3443,
      "step": 17112
    },
    {
      "epoch": 1.4723393271960767,
      "grad_norm": 1.03125,
      "learning_rate": 0.0008012589248018405,
      "loss": 0.8196,
      "step": 17113
    },
    {
      "epoch": 1.4724253635033984,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008012373430040516,
      "loss": 1.4846,
      "step": 17114
    },
    {
      "epoch": 1.47251139981072,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000801215760325202,
      "loss": 0.5247,
      "step": 17115
    },
    {
      "epoch": 1.4725974361180418,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008011941767653548,
      "loss": 1.9681,
      "step": 17116
    },
    {
      "epoch": 1.4726834724253635,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008011725923245732,
      "loss": 1.689,
      "step": 17117
    },
    {
      "epoch": 1.4727695087326853,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008011510070029204,
      "loss": 1.3014,
      "step": 17118
    },
    {
      "epoch": 1.472855545040007,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008011294208004596,
      "loss": 0.654,
      "step": 17119
    },
    {
      "epoch": 1.4729415813473286,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008011078337172537,
      "loss": 0.8819,
      "step": 17120
    },
    {
      "epoch": 1.4730276176546502,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0008010862457533659,
      "loss": 1.1728,
      "step": 17121
    },
    {
      "epoch": 1.473113653961972,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008010646569088595,
      "loss": 0.8778,
      "step": 17122
    },
    {
      "epoch": 1.4731996902692936,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008010430671837974,
      "loss": 1.8775,
      "step": 17123
    },
    {
      "epoch": 1.4732857265766153,
      "grad_norm": 1.6875,
      "learning_rate": 0.000801021476578243,
      "loss": 1.5937,
      "step": 17124
    },
    {
      "epoch": 1.4733717628839371,
      "grad_norm": 1.0,
      "learning_rate": 0.0008009998850922593,
      "loss": 0.7079,
      "step": 17125
    },
    {
      "epoch": 1.4734577991912587,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0008009782927259093,
      "loss": 0.9931,
      "step": 17126
    },
    {
      "epoch": 1.4735438354985804,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008009566994792565,
      "loss": 1.2504,
      "step": 17127
    },
    {
      "epoch": 1.473629871805902,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008009351053523639,
      "loss": 0.777,
      "step": 17128
    },
    {
      "epoch": 1.4737159081132238,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008009135103452947,
      "loss": 1.8026,
      "step": 17129
    },
    {
      "epoch": 1.4738019444205455,
      "grad_norm": 1.015625,
      "learning_rate": 0.000800891914458112,
      "loss": 1.3278,
      "step": 17130
    },
    {
      "epoch": 1.473887980727867,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008008703176908787,
      "loss": 0.7755,
      "step": 17131
    },
    {
      "epoch": 1.473974017035189,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008008487200436585,
      "loss": 0.7413,
      "step": 17132
    },
    {
      "epoch": 1.4740600533425106,
      "grad_norm": 1.375,
      "learning_rate": 0.0008008271215165141,
      "loss": 1.1904,
      "step": 17133
    },
    {
      "epoch": 1.4741460896498322,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008008055221095092,
      "loss": 1.3655,
      "step": 17134
    },
    {
      "epoch": 1.4742321259571538,
      "grad_norm": 1.484375,
      "learning_rate": 0.0008007839218227064,
      "loss": 1.5296,
      "step": 17135
    },
    {
      "epoch": 1.4743181622644757,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008007623206561692,
      "loss": 0.7163,
      "step": 17136
    },
    {
      "epoch": 1.4744041985717973,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008007407186099607,
      "loss": 1.6302,
      "step": 17137
    },
    {
      "epoch": 1.4744902348791191,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000800719115684144,
      "loss": 1.0589,
      "step": 17138
    },
    {
      "epoch": 1.4745762711864407,
      "grad_norm": 1.078125,
      "learning_rate": 0.0008006975118787826,
      "loss": 1.1984,
      "step": 17139
    },
    {
      "epoch": 1.4746623074937624,
      "grad_norm": 1.046875,
      "learning_rate": 0.0008006759071939394,
      "loss": 1.9414,
      "step": 17140
    },
    {
      "epoch": 1.474748343801084,
      "grad_norm": 1.140625,
      "learning_rate": 0.0008006543016296774,
      "loss": 1.2452,
      "step": 17141
    },
    {
      "epoch": 1.4748343801084056,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008006326951860603,
      "loss": 1.9201,
      "step": 17142
    },
    {
      "epoch": 1.4749204164157275,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000800611087863151,
      "loss": 1.5402,
      "step": 17143
    },
    {
      "epoch": 1.475006452723049,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0008005894796610125,
      "loss": 1.2084,
      "step": 17144
    },
    {
      "epoch": 1.475092489030371,
      "grad_norm": 1.015625,
      "learning_rate": 0.0008005678705797084,
      "loss": 0.6997,
      "step": 17145
    },
    {
      "epoch": 1.4751785253376926,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008005462606193018,
      "loss": 0.9596,
      "step": 17146
    },
    {
      "epoch": 1.4752645616450142,
      "grad_norm": 1.328125,
      "learning_rate": 0.0008005246497798558,
      "loss": 1.6028,
      "step": 17147
    },
    {
      "epoch": 1.4753505979523358,
      "grad_norm": 1.15625,
      "learning_rate": 0.0008005030380614336,
      "loss": 1.1646,
      "step": 17148
    },
    {
      "epoch": 1.4754366342596577,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008004814254640987,
      "loss": 1.1463,
      "step": 17149
    },
    {
      "epoch": 1.4755226705669793,
      "grad_norm": 1.390625,
      "learning_rate": 0.0008004598119879138,
      "loss": 1.5717,
      "step": 17150
    },
    {
      "epoch": 1.475608706874301,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008004381976329425,
      "loss": 1.7932,
      "step": 17151
    },
    {
      "epoch": 1.4756947431816227,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008004165823992477,
      "loss": 1.2238,
      "step": 17152
    },
    {
      "epoch": 1.4757807794889444,
      "grad_norm": 1.5,
      "learning_rate": 0.000800394966286893,
      "loss": 2.0694,
      "step": 17153
    },
    {
      "epoch": 1.475866815796266,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0008003733492959415,
      "loss": 0.9348,
      "step": 17154
    },
    {
      "epoch": 1.4759528521035876,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008003517314264563,
      "loss": 1.5427,
      "step": 17155
    },
    {
      "epoch": 1.4760388884109095,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0008003301126785008,
      "loss": 1.5453,
      "step": 17156
    },
    {
      "epoch": 1.476124924718231,
      "grad_norm": 1.359375,
      "learning_rate": 0.000800308493052138,
      "loss": 1.4755,
      "step": 17157
    },
    {
      "epoch": 1.4762109610255527,
      "grad_norm": 1.40625,
      "learning_rate": 0.0008002868725474312,
      "loss": 0.9481,
      "step": 17158
    },
    {
      "epoch": 1.4762969973328746,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008002652511644438,
      "loss": 1.0642,
      "step": 17159
    },
    {
      "epoch": 1.4763830336401962,
      "grad_norm": 1.34375,
      "learning_rate": 0.000800243628903239,
      "loss": 2.0786,
      "step": 17160
    },
    {
      "epoch": 1.4764690699475178,
      "grad_norm": 1.109375,
      "learning_rate": 0.0008002220057638798,
      "loss": 1.1605,
      "step": 17161
    },
    {
      "epoch": 1.4765551062548394,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008002003817464299,
      "loss": 1.483,
      "step": 17162
    },
    {
      "epoch": 1.4766411425621613,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0008001787568509519,
      "loss": 1.6121,
      "step": 17163
    },
    {
      "epoch": 1.476727178869483,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008001571310775098,
      "loss": 1.0845,
      "step": 17164
    },
    {
      "epoch": 1.4768132151768047,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0008001355044261662,
      "loss": 0.8413,
      "step": 17165
    },
    {
      "epoch": 1.4768992514841264,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008001138768969848,
      "loss": 1.7728,
      "step": 17166
    },
    {
      "epoch": 1.476985287791448,
      "grad_norm": 1.609375,
      "learning_rate": 0.0008000922484900285,
      "loss": 1.1587,
      "step": 17167
    },
    {
      "epoch": 1.4770713240987696,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008000706192053609,
      "loss": 0.9703,
      "step": 17168
    },
    {
      "epoch": 1.4771573604060912,
      "grad_norm": 1.8125,
      "learning_rate": 0.000800048989043045,
      "loss": 1.3716,
      "step": 17169
    },
    {
      "epoch": 1.477243396713413,
      "grad_norm": 1.0,
      "learning_rate": 0.0008000273580031443,
      "loss": 1.2588,
      "step": 17170
    },
    {
      "epoch": 1.4773294330207347,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008000057260857218,
      "loss": 1.1686,
      "step": 17171
    },
    {
      "epoch": 1.4774154693280566,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007999840932908409,
      "loss": 0.8196,
      "step": 17172
    },
    {
      "epoch": 1.4775015056353782,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000799962459618565,
      "loss": 0.9779,
      "step": 17173
    },
    {
      "epoch": 1.4775875419426998,
      "grad_norm": 1.125,
      "learning_rate": 0.000799940825068957,
      "loss": 0.967,
      "step": 17174
    },
    {
      "epoch": 1.4776735782500214,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007999191896420806,
      "loss": 1.1241,
      "step": 17175
    },
    {
      "epoch": 1.4777596145573433,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007998975533379988,
      "loss": 1.584,
      "step": 17176
    },
    {
      "epoch": 1.477845650864665,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007998759161567751,
      "loss": 1.7257,
      "step": 17177
    },
    {
      "epoch": 1.4779316871719865,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007998542780984727,
      "loss": 1.9126,
      "step": 17178
    },
    {
      "epoch": 1.4780177234793084,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007998326391631547,
      "loss": 0.8416,
      "step": 17179
    },
    {
      "epoch": 1.47810375978663,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007998109993508847,
      "loss": 0.9818,
      "step": 17180
    },
    {
      "epoch": 1.4781897960939516,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007997893586617257,
      "loss": 0.7127,
      "step": 17181
    },
    {
      "epoch": 1.4782758324012732,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0007997677170957411,
      "loss": 0.6278,
      "step": 17182
    },
    {
      "epoch": 1.478361868708595,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0007997460746529944,
      "loss": 2.0772,
      "step": 17183
    },
    {
      "epoch": 1.4784479050159167,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007997244313335487,
      "loss": 0.838,
      "step": 17184
    },
    {
      "epoch": 1.4785339413232383,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007997027871374671,
      "loss": 1.8459,
      "step": 17185
    },
    {
      "epoch": 1.4786199776305602,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007996811420648135,
      "loss": 1.0854,
      "step": 17186
    },
    {
      "epoch": 1.4787060139378818,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007996594961156506,
      "loss": 0.7002,
      "step": 17187
    },
    {
      "epoch": 1.4787920502452034,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000799637849290042,
      "loss": 1.1941,
      "step": 17188
    },
    {
      "epoch": 1.478878086552525,
      "grad_norm": 1.484375,
      "learning_rate": 0.000799616201588051,
      "loss": 1.6078,
      "step": 17189
    },
    {
      "epoch": 1.478964122859847,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007995945530097408,
      "loss": 1.4366,
      "step": 17190
    },
    {
      "epoch": 1.4790501591671685,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007995729035551748,
      "loss": 1.1078,
      "step": 17191
    },
    {
      "epoch": 1.4791361954744902,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007995512532244165,
      "loss": 1.8409,
      "step": 17192
    },
    {
      "epoch": 1.479222231781812,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007995296020175289,
      "loss": 2.0326,
      "step": 17193
    },
    {
      "epoch": 1.4793082680891336,
      "grad_norm": 2.15625,
      "learning_rate": 0.0007995079499345755,
      "loss": 0.8888,
      "step": 17194
    },
    {
      "epoch": 1.4793943043964553,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007994862969756196,
      "loss": 1.0,
      "step": 17195
    },
    {
      "epoch": 1.4794803407037769,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007994646431407245,
      "loss": 1.8997,
      "step": 17196
    },
    {
      "epoch": 1.4795663770110987,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007994429884299537,
      "loss": 0.9513,
      "step": 17197
    },
    {
      "epoch": 1.4796524133184203,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007994213328433703,
      "loss": 1.3893,
      "step": 17198
    },
    {
      "epoch": 1.4797384496257422,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007993996763810376,
      "loss": 1.4292,
      "step": 17199
    },
    {
      "epoch": 1.4798244859330638,
      "grad_norm": 1.0,
      "learning_rate": 0.0007993780190430191,
      "loss": 1.0072,
      "step": 17200
    },
    {
      "epoch": 1.4799105222403854,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007993563608293784,
      "loss": 1.7457,
      "step": 17201
    },
    {
      "epoch": 1.479996558547707,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007993347017401782,
      "loss": 1.0392,
      "step": 17202
    },
    {
      "epoch": 1.4800825948550287,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007993130417754824,
      "loss": 0.914,
      "step": 17203
    },
    {
      "epoch": 1.4801686311623505,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007992913809353543,
      "loss": 1.77,
      "step": 17204
    },
    {
      "epoch": 1.4802546674696722,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000799269719219857,
      "loss": 1.444,
      "step": 17205
    },
    {
      "epoch": 1.480340703776994,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007992480566290539,
      "loss": 1.0379,
      "step": 17206
    },
    {
      "epoch": 1.4804267400843156,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007992263931630084,
      "loss": 1.8493,
      "step": 17207
    },
    {
      "epoch": 1.4805127763916373,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000799204728821784,
      "loss": 0.8943,
      "step": 17208
    },
    {
      "epoch": 1.4805988126989589,
      "grad_norm": 2.5,
      "learning_rate": 0.000799183063605444,
      "loss": 1.0279,
      "step": 17209
    },
    {
      "epoch": 1.4806848490062807,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007991613975140517,
      "loss": 0.9545,
      "step": 17210
    },
    {
      "epoch": 1.4807708853136023,
      "grad_norm": 1.25,
      "learning_rate": 0.0007991397305476705,
      "loss": 1.7455,
      "step": 17211
    },
    {
      "epoch": 1.480856921620924,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007991180627063637,
      "loss": 1.5157,
      "step": 17212
    },
    {
      "epoch": 1.4809429579282458,
      "grad_norm": 1.125,
      "learning_rate": 0.0007990963939901948,
      "loss": 2.1196,
      "step": 17213
    },
    {
      "epoch": 1.4810289942355674,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007990747243992271,
      "loss": 1.8271,
      "step": 17214
    },
    {
      "epoch": 1.481115030542889,
      "grad_norm": 1.1875,
      "learning_rate": 0.000799053053933524,
      "loss": 0.8534,
      "step": 17215
    },
    {
      "epoch": 1.4812010668502107,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007990313825931489,
      "loss": 1.17,
      "step": 17216
    },
    {
      "epoch": 1.4812871031575325,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007990097103781651,
      "loss": 1.2652,
      "step": 17217
    },
    {
      "epoch": 1.4813731394648542,
      "grad_norm": 1.46875,
      "learning_rate": 0.000798988037288636,
      "loss": 1.8081,
      "step": 17218
    },
    {
      "epoch": 1.4814591757721758,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007989663633246252,
      "loss": 1.5648,
      "step": 17219
    },
    {
      "epoch": 1.4815452120794976,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007989446884861959,
      "loss": 0.9274,
      "step": 17220
    },
    {
      "epoch": 1.4816312483868193,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007989230127734116,
      "loss": 1.2709,
      "step": 17221
    },
    {
      "epoch": 1.4817172846941409,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007989013361863355,
      "loss": 1.4433,
      "step": 17222
    },
    {
      "epoch": 1.4818033210014625,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007988796587250311,
      "loss": 1.3383,
      "step": 17223
    },
    {
      "epoch": 1.4818893573087843,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007988579803895618,
      "loss": 1.5685,
      "step": 17224
    },
    {
      "epoch": 1.481975393616106,
      "grad_norm": 1.3125,
      "learning_rate": 0.000798836301179991,
      "loss": 1.872,
      "step": 17225
    },
    {
      "epoch": 1.4820614299234278,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007988146210963823,
      "loss": 0.7881,
      "step": 17226
    },
    {
      "epoch": 1.4821474662307494,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007987929401387989,
      "loss": 1.744,
      "step": 17227
    },
    {
      "epoch": 1.482233502538071,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007987712583073044,
      "loss": 0.7188,
      "step": 17228
    },
    {
      "epoch": 1.4823195388453927,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007987495756019618,
      "loss": 0.7069,
      "step": 17229
    },
    {
      "epoch": 1.4824055751527143,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007987278920228349,
      "loss": 1.3708,
      "step": 17230
    },
    {
      "epoch": 1.4824916114600362,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007987062075699872,
      "loss": 1.3442,
      "step": 17231
    },
    {
      "epoch": 1.4825776477673578,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007986845222434818,
      "loss": 1.2702,
      "step": 17232
    },
    {
      "epoch": 1.4826636840746796,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007986628360433821,
      "loss": 0.9157,
      "step": 17233
    },
    {
      "epoch": 1.4827497203820013,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007986411489697518,
      "loss": 1.1176,
      "step": 17234
    },
    {
      "epoch": 1.4828357566893229,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007986194610226543,
      "loss": 1.4338,
      "step": 17235
    },
    {
      "epoch": 1.4829217929966445,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007985977722021529,
      "loss": 1.3246,
      "step": 17236
    },
    {
      "epoch": 1.4830078293039664,
      "grad_norm": 1.0625,
      "learning_rate": 0.000798576082508311,
      "loss": 1.0007,
      "step": 17237
    },
    {
      "epoch": 1.483093865611288,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007985543919411924,
      "loss": 0.8549,
      "step": 17238
    },
    {
      "epoch": 1.4831799019186096,
      "grad_norm": 1.28125,
      "learning_rate": 0.00079853270050086,
      "loss": 1.3254,
      "step": 17239
    },
    {
      "epoch": 1.4832659382259314,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007985110081873777,
      "loss": 1.0488,
      "step": 17240
    },
    {
      "epoch": 1.483351974533253,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007984893150008085,
      "loss": 0.8739,
      "step": 17241
    },
    {
      "epoch": 1.4834380108405747,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007984676209412164,
      "loss": 1.0814,
      "step": 17242
    },
    {
      "epoch": 1.4835240471478963,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007984459260086644,
      "loss": 1.4167,
      "step": 17243
    },
    {
      "epoch": 1.4836100834552182,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007984242302032162,
      "loss": 1.5536,
      "step": 17244
    },
    {
      "epoch": 1.4836961197625398,
      "grad_norm": 0.75,
      "learning_rate": 0.0007984025335249351,
      "loss": 0.832,
      "step": 17245
    },
    {
      "epoch": 1.4837821560698614,
      "grad_norm": 1.125,
      "learning_rate": 0.0007983808359738846,
      "loss": 1.9045,
      "step": 17246
    },
    {
      "epoch": 1.4838681923771833,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007983591375501282,
      "loss": 1.1563,
      "step": 17247
    },
    {
      "epoch": 1.4839542286845049,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007983374382537294,
      "loss": 1.1612,
      "step": 17248
    },
    {
      "epoch": 1.4840402649918265,
      "grad_norm": 2.0,
      "learning_rate": 0.0007983157380847517,
      "loss": 0.8019,
      "step": 17249
    },
    {
      "epoch": 1.4841263012991481,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007982940370432583,
      "loss": 0.8216,
      "step": 17250
    },
    {
      "epoch": 1.48421233760647,
      "grad_norm": 1.390625,
      "learning_rate": 0.000798272335129313,
      "loss": 1.932,
      "step": 17251
    },
    {
      "epoch": 1.4842983739137916,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000798250632342979,
      "loss": 1.1591,
      "step": 17252
    },
    {
      "epoch": 1.4843844102211134,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007982289286843201,
      "loss": 1.044,
      "step": 17253
    },
    {
      "epoch": 1.484470446528435,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007982072241533993,
      "loss": 1.7051,
      "step": 17254
    },
    {
      "epoch": 1.4845564828357567,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007981855187502804,
      "loss": 1.595,
      "step": 17255
    },
    {
      "epoch": 1.4846425191430783,
      "grad_norm": 1.328125,
      "learning_rate": 0.000798163812475027,
      "loss": 1.2254,
      "step": 17256
    },
    {
      "epoch": 1.4847285554504,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007981421053277025,
      "loss": 1.108,
      "step": 17257
    },
    {
      "epoch": 1.4848145917577218,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007981203973083702,
      "loss": 1.3249,
      "step": 17258
    },
    {
      "epoch": 1.4849006280650434,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007980986884170937,
      "loss": 1.1388,
      "step": 17259
    },
    {
      "epoch": 1.4849866643723653,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007980769786539365,
      "loss": 1.3566,
      "step": 17260
    },
    {
      "epoch": 1.4850727006796869,
      "grad_norm": 1.0,
      "learning_rate": 0.0007980552680189624,
      "loss": 1.3496,
      "step": 17261
    },
    {
      "epoch": 1.4851587369870085,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007980335565122343,
      "loss": 0.9913,
      "step": 17262
    },
    {
      "epoch": 1.4852447732943301,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007980118441338162,
      "loss": 1.9086,
      "step": 17263
    },
    {
      "epoch": 1.485330809601652,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007979901308837712,
      "loss": 1.5117,
      "step": 17264
    },
    {
      "epoch": 1.4854168459089736,
      "grad_norm": 1.5,
      "learning_rate": 0.0007979684167621632,
      "loss": 1.5021,
      "step": 17265
    },
    {
      "epoch": 1.4855028822162952,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007979467017690555,
      "loss": 2.1573,
      "step": 17266
    },
    {
      "epoch": 1.485588918523617,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007979249859045117,
      "loss": 0.9157,
      "step": 17267
    },
    {
      "epoch": 1.4856749548309387,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007979032691685952,
      "loss": 1.443,
      "step": 17268
    },
    {
      "epoch": 1.4857609911382603,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007978815515613695,
      "loss": 1.2119,
      "step": 17269
    },
    {
      "epoch": 1.485847027445582,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007978598330828982,
      "loss": 1.1158,
      "step": 17270
    },
    {
      "epoch": 1.4859330637529038,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007978381137332449,
      "loss": 0.7255,
      "step": 17271
    },
    {
      "epoch": 1.4860191000602254,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007978163935124731,
      "loss": 1.5642,
      "step": 17272
    },
    {
      "epoch": 1.486105136367547,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000797794672420646,
      "loss": 0.943,
      "step": 17273
    },
    {
      "epoch": 1.4861911726748689,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007977729504578277,
      "loss": 1.4356,
      "step": 17274
    },
    {
      "epoch": 1.4862772089821905,
      "grad_norm": 1.25,
      "learning_rate": 0.0007977512276240812,
      "loss": 1.3221,
      "step": 17275
    },
    {
      "epoch": 1.4863632452895121,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007977295039194706,
      "loss": 1.6475,
      "step": 17276
    },
    {
      "epoch": 1.4864492815968338,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007977077793440588,
      "loss": 0.8829,
      "step": 17277
    },
    {
      "epoch": 1.4865353179041556,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007976860538979097,
      "loss": 1.2326,
      "step": 17278
    },
    {
      "epoch": 1.4866213542114772,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007976643275810868,
      "loss": 0.9753,
      "step": 17279
    },
    {
      "epoch": 1.4867073905187989,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007976426003936536,
      "loss": 1.2026,
      "step": 17280
    },
    {
      "epoch": 1.4867934268261207,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007976208723356738,
      "loss": 1.3759,
      "step": 17281
    },
    {
      "epoch": 1.4868794631334423,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007975991434072108,
      "loss": 2.0698,
      "step": 17282
    },
    {
      "epoch": 1.486965499440764,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007975774136083283,
      "loss": 1.3351,
      "step": 17283
    },
    {
      "epoch": 1.4870515357480856,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007975556829390895,
      "loss": 1.2275,
      "step": 17284
    },
    {
      "epoch": 1.4871375720554074,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007975339513995583,
      "loss": 1.0714,
      "step": 17285
    },
    {
      "epoch": 1.487223608362729,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007975122189897981,
      "loss": 1.5642,
      "step": 17286
    },
    {
      "epoch": 1.487309644670051,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007974904857098727,
      "loss": 1.2183,
      "step": 17287
    },
    {
      "epoch": 1.4873956809773725,
      "grad_norm": 1.0,
      "learning_rate": 0.0007974687515598454,
      "loss": 1.4315,
      "step": 17288
    },
    {
      "epoch": 1.4874817172846941,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007974470165397798,
      "loss": 1.4244,
      "step": 17289
    },
    {
      "epoch": 1.4875677535920158,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007974252806497395,
      "loss": 1.0202,
      "step": 17290
    },
    {
      "epoch": 1.4876537898993374,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007974035438897882,
      "loss": 1.0206,
      "step": 17291
    },
    {
      "epoch": 1.4877398262066592,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0007973818062599892,
      "loss": 1.425,
      "step": 17292
    },
    {
      "epoch": 1.4878258625139809,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0007973600677604064,
      "loss": 1.2968,
      "step": 17293
    },
    {
      "epoch": 1.4879118988213027,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007973383283911033,
      "loss": 1.1725,
      "step": 17294
    },
    {
      "epoch": 1.4879979351286243,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007973165881521434,
      "loss": 1.3107,
      "step": 17295
    },
    {
      "epoch": 1.488083971435946,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007972948470435901,
      "loss": 1.7245,
      "step": 17296
    },
    {
      "epoch": 1.4881700077432676,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007972731050655073,
      "loss": 1.235,
      "step": 17297
    },
    {
      "epoch": 1.4882560440505894,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007972513622179587,
      "loss": 1.4511,
      "step": 17298
    },
    {
      "epoch": 1.488342080357911,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007972296185010075,
      "loss": 2.2989,
      "step": 17299
    },
    {
      "epoch": 1.4884281166652327,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007972078739147175,
      "loss": 1.9144,
      "step": 17300
    },
    {
      "epoch": 1.4885141529725545,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007971861284591521,
      "loss": 1.1767,
      "step": 17301
    },
    {
      "epoch": 1.4886001892798761,
      "grad_norm": 1.8125,
      "learning_rate": 0.0007971643821343753,
      "loss": 1.6271,
      "step": 17302
    },
    {
      "epoch": 1.4886862255871978,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007971426349404504,
      "loss": 1.1755,
      "step": 17303
    },
    {
      "epoch": 1.4887722618945194,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007971208868774411,
      "loss": 1.7188,
      "step": 17304
    },
    {
      "epoch": 1.4888582982018412,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007970991379454109,
      "loss": 0.9621,
      "step": 17305
    },
    {
      "epoch": 1.4889443345091629,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007970773881444236,
      "loss": 1.5526,
      "step": 17306
    },
    {
      "epoch": 1.4890303708164845,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007970556374745428,
      "loss": 1.7579,
      "step": 17307
    },
    {
      "epoch": 1.4891164071238063,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007970338859358319,
      "loss": 1.0441,
      "step": 17308
    },
    {
      "epoch": 1.489202443431128,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007970121335283547,
      "loss": 1.1162,
      "step": 17309
    },
    {
      "epoch": 1.4892884797384496,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007969903802521747,
      "loss": 0.8716,
      "step": 17310
    },
    {
      "epoch": 1.4893745160457712,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007969686261073556,
      "loss": 0.9675,
      "step": 17311
    },
    {
      "epoch": 1.489460552353093,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007969468710939612,
      "loss": 1.0154,
      "step": 17312
    },
    {
      "epoch": 1.4895465886604147,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007969251152120547,
      "loss": 1.6819,
      "step": 17313
    },
    {
      "epoch": 1.4896326249677365,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007969033584617002,
      "loss": 1.2849,
      "step": 17314
    },
    {
      "epoch": 1.4897186612750581,
      "grad_norm": 1.09375,
      "learning_rate": 0.000796881600842961,
      "loss": 0.737,
      "step": 17315
    },
    {
      "epoch": 1.4898046975823798,
      "grad_norm": 19.875,
      "learning_rate": 0.0007968598423559009,
      "loss": 1.6287,
      "step": 17316
    },
    {
      "epoch": 1.4898907338897014,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007968380830005835,
      "loss": 0.9704,
      "step": 17317
    },
    {
      "epoch": 1.489976770197023,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007968163227770725,
      "loss": 1.5936,
      "step": 17318
    },
    {
      "epoch": 1.4900628065043449,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007967945616854315,
      "loss": 1.2666,
      "step": 17319
    },
    {
      "epoch": 1.4901488428116665,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000796772799725724,
      "loss": 1.6761,
      "step": 17320
    },
    {
      "epoch": 1.4902348791189883,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007967510368980138,
      "loss": 1.4517,
      "step": 17321
    },
    {
      "epoch": 1.49032091542631,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007967292732023645,
      "loss": 1.1058,
      "step": 17322
    },
    {
      "epoch": 1.4904069517336316,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00079670750863884,
      "loss": 0.9305,
      "step": 17323
    },
    {
      "epoch": 1.4904929880409532,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007966857432075036,
      "loss": 1.1112,
      "step": 17324
    },
    {
      "epoch": 1.490579024348275,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007966639769084189,
      "loss": 1.3273,
      "step": 17325
    },
    {
      "epoch": 1.4906650606555967,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007966422097416501,
      "loss": 1.0096,
      "step": 17326
    },
    {
      "epoch": 1.4907510969629183,
      "grad_norm": 1.125,
      "learning_rate": 0.0007966204417072604,
      "loss": 1.5243,
      "step": 17327
    },
    {
      "epoch": 1.4908371332702401,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007965986728053137,
      "loss": 0.9,
      "step": 17328
    },
    {
      "epoch": 1.4909231695775618,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007965769030358733,
      "loss": 0.8228,
      "step": 17329
    },
    {
      "epoch": 1.4910092058848834,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007965551323990033,
      "loss": 1.5927,
      "step": 17330
    },
    {
      "epoch": 1.491095242192205,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007965333608947673,
      "loss": 1.7371,
      "step": 17331
    },
    {
      "epoch": 1.4911812784995269,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007965115885232288,
      "loss": 1.3008,
      "step": 17332
    },
    {
      "epoch": 1.4912673148068485,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007964898152844515,
      "loss": 1.3542,
      "step": 17333
    },
    {
      "epoch": 1.4913533511141701,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007964680411784992,
      "loss": 1.5415,
      "step": 17334
    },
    {
      "epoch": 1.491439387421492,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0007964462662054356,
      "loss": 1.0181,
      "step": 17335
    },
    {
      "epoch": 1.4915254237288136,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007964244903653244,
      "loss": 1.7407,
      "step": 17336
    },
    {
      "epoch": 1.4916114600361352,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0007964027136582291,
      "loss": 1.5208,
      "step": 17337
    },
    {
      "epoch": 1.4916974963434568,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007963809360842135,
      "loss": 0.5954,
      "step": 17338
    },
    {
      "epoch": 1.4917835326507787,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007963591576433414,
      "loss": 1.453,
      "step": 17339
    },
    {
      "epoch": 1.4918695689581003,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007963373783356762,
      "loss": 1.5576,
      "step": 17340
    },
    {
      "epoch": 1.4919556052654221,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007963155981612819,
      "loss": 1.2061,
      "step": 17341
    },
    {
      "epoch": 1.4920416415727438,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007962938171202223,
      "loss": 2.0265,
      "step": 17342
    },
    {
      "epoch": 1.4921276778800654,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007962720352125607,
      "loss": 1.0761,
      "step": 17343
    },
    {
      "epoch": 1.492213714187387,
      "grad_norm": 1.046875,
      "learning_rate": 0.000796250252438361,
      "loss": 1.1376,
      "step": 17344
    },
    {
      "epoch": 1.4922997504947086,
      "grad_norm": 0.859375,
      "learning_rate": 0.000796228468797687,
      "loss": 0.8438,
      "step": 17345
    },
    {
      "epoch": 1.4923857868020305,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007962066842906024,
      "loss": 1.2606,
      "step": 17346
    },
    {
      "epoch": 1.4924718231093521,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007961848989171708,
      "loss": 1.2084,
      "step": 17347
    },
    {
      "epoch": 1.492557859416674,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000796163112677456,
      "loss": 1.273,
      "step": 17348
    },
    {
      "epoch": 1.4926438957239956,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007961413255715217,
      "loss": 1.1174,
      "step": 17349
    },
    {
      "epoch": 1.4927299320313172,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007961195375994314,
      "loss": 0.7077,
      "step": 17350
    },
    {
      "epoch": 1.4928159683386388,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007960977487612495,
      "loss": 1.8739,
      "step": 17351
    },
    {
      "epoch": 1.4929020046459607,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007960759590570389,
      "loss": 2.0357,
      "step": 17352
    },
    {
      "epoch": 1.4929880409532823,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007960541684868638,
      "loss": 1.0032,
      "step": 17353
    },
    {
      "epoch": 1.493074077260604,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007960323770507877,
      "loss": 0.7934,
      "step": 17354
    },
    {
      "epoch": 1.4931601135679258,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007960105847488747,
      "loss": 1.2294,
      "step": 17355
    },
    {
      "epoch": 1.4932461498752474,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007959887915811883,
      "loss": 1.3246,
      "step": 17356
    },
    {
      "epoch": 1.493332186182569,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007959669975477921,
      "loss": 1.7841,
      "step": 17357
    },
    {
      "epoch": 1.4934182224898906,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007959452026487502,
      "loss": 0.686,
      "step": 17358
    },
    {
      "epoch": 1.4935042587972125,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000795923406884126,
      "loss": 1.5943,
      "step": 17359
    },
    {
      "epoch": 1.4935902951045341,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007959016102539834,
      "loss": 0.9248,
      "step": 17360
    },
    {
      "epoch": 1.4936763314118557,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007958798127583862,
      "loss": 1.064,
      "step": 17361
    },
    {
      "epoch": 1.4937623677191776,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007958580143973979,
      "loss": 1.6351,
      "step": 17362
    },
    {
      "epoch": 1.4938484040264992,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007958362151710827,
      "loss": 0.6975,
      "step": 17363
    },
    {
      "epoch": 1.4939344403338208,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007958144150795039,
      "loss": 0.665,
      "step": 17364
    },
    {
      "epoch": 1.4940204766411425,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007957926141227257,
      "loss": 1.3361,
      "step": 17365
    },
    {
      "epoch": 1.4941065129484643,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007957708123008114,
      "loss": 1.2697,
      "step": 17366
    },
    {
      "epoch": 1.494192549255786,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007957490096138251,
      "loss": 1.3484,
      "step": 17367
    },
    {
      "epoch": 1.4942785855631076,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007957272060618304,
      "loss": 0.626,
      "step": 17368
    },
    {
      "epoch": 1.4943646218704294,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007957054016448912,
      "loss": 1.1855,
      "step": 17369
    },
    {
      "epoch": 1.494450658177751,
      "grad_norm": 1.640625,
      "learning_rate": 0.0007956835963630713,
      "loss": 1.6126,
      "step": 17370
    },
    {
      "epoch": 1.4945366944850726,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007956617902164343,
      "loss": 1.8353,
      "step": 17371
    },
    {
      "epoch": 1.4946227307923943,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007956399832050441,
      "loss": 0.7417,
      "step": 17372
    },
    {
      "epoch": 1.4947087670997161,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007956181753289645,
      "loss": 0.9049,
      "step": 17373
    },
    {
      "epoch": 1.4947948034070377,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007955963665882591,
      "loss": 1.3127,
      "step": 17374
    },
    {
      "epoch": 1.4948808397143596,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007955745569829918,
      "loss": 1.3928,
      "step": 17375
    },
    {
      "epoch": 1.4949668760216812,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007955527465132264,
      "loss": 1.3226,
      "step": 17376
    },
    {
      "epoch": 1.4950529123290028,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007955309351790268,
      "loss": 1.0633,
      "step": 17377
    },
    {
      "epoch": 1.4951389486363245,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007955091229804568,
      "loss": 1.4456,
      "step": 17378
    },
    {
      "epoch": 1.495224984943646,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007954873099175798,
      "loss": 0.8555,
      "step": 17379
    },
    {
      "epoch": 1.495311021250968,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007954654959904602,
      "loss": 0.8527,
      "step": 17380
    },
    {
      "epoch": 1.4953970575582896,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007954436811991613,
      "loss": 1.2675,
      "step": 17381
    },
    {
      "epoch": 1.4954830938656114,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007954218655437471,
      "loss": 1.5665,
      "step": 17382
    },
    {
      "epoch": 1.495569130172933,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007954000490242813,
      "loss": 0.8201,
      "step": 17383
    },
    {
      "epoch": 1.4956551664802546,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000795378231640828,
      "loss": 1.269,
      "step": 17384
    },
    {
      "epoch": 1.4957412027875763,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007953564133934505,
      "loss": 0.9358,
      "step": 17385
    },
    {
      "epoch": 1.4958272390948981,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007953345942822134,
      "loss": 1.3122,
      "step": 17386
    },
    {
      "epoch": 1.4959132754022197,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007953127743071797,
      "loss": 1.9197,
      "step": 17387
    },
    {
      "epoch": 1.4959993117095414,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007952909534684139,
      "loss": 1.9457,
      "step": 17388
    },
    {
      "epoch": 1.4960853480168632,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007952691317659791,
      "loss": 1.5759,
      "step": 17389
    },
    {
      "epoch": 1.4961713843241848,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007952473091999399,
      "loss": 0.8029,
      "step": 17390
    },
    {
      "epoch": 1.4962574206315065,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007952254857703595,
      "loss": 0.8791,
      "step": 17391
    },
    {
      "epoch": 1.496343456938828,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007952036614773019,
      "loss": 1.7799,
      "step": 17392
    },
    {
      "epoch": 1.49642949324615,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007951818363208312,
      "loss": 0.8211,
      "step": 17393
    },
    {
      "epoch": 1.4965155295534716,
      "grad_norm": 1.375,
      "learning_rate": 0.0007951600103010111,
      "loss": 1.251,
      "step": 17394
    },
    {
      "epoch": 1.4966015658607932,
      "grad_norm": 2.109375,
      "learning_rate": 0.0007951381834179052,
      "loss": 1.8813,
      "step": 17395
    },
    {
      "epoch": 1.496687602168115,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007951163556715775,
      "loss": 0.6741,
      "step": 17396
    },
    {
      "epoch": 1.4967736384754367,
      "grad_norm": 1.078125,
      "learning_rate": 0.000795094527062092,
      "loss": 0.8297,
      "step": 17397
    },
    {
      "epoch": 1.4968596747827583,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007950726975895122,
      "loss": 1.2129,
      "step": 17398
    },
    {
      "epoch": 1.49694571109008,
      "grad_norm": 1.828125,
      "learning_rate": 0.0007950508672539023,
      "loss": 1.5368,
      "step": 17399
    },
    {
      "epoch": 1.4970317473974017,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007950290360553261,
      "loss": 1.1174,
      "step": 17400
    },
    {
      "epoch": 1.4971177837047234,
      "grad_norm": 1.25,
      "learning_rate": 0.0007950072039938472,
      "loss": 1.4452,
      "step": 17401
    },
    {
      "epoch": 1.4972038200120452,
      "grad_norm": 1.25,
      "learning_rate": 0.0007949853710695295,
      "loss": 0.8576,
      "step": 17402
    },
    {
      "epoch": 1.4972898563193668,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007949635372824371,
      "loss": 1.3168,
      "step": 17403
    },
    {
      "epoch": 1.4973758926266885,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007949417026326336,
      "loss": 1.0612,
      "step": 17404
    },
    {
      "epoch": 1.49746192893401,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007949198671201831,
      "loss": 1.6907,
      "step": 17405
    },
    {
      "epoch": 1.4975479652413317,
      "grad_norm": 1.0,
      "learning_rate": 0.0007948980307451493,
      "loss": 1.4926,
      "step": 17406
    },
    {
      "epoch": 1.4976340015486536,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007948761935075963,
      "loss": 0.8851,
      "step": 17407
    },
    {
      "epoch": 1.4977200378559752,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007948543554075876,
      "loss": 1.1257,
      "step": 17408
    },
    {
      "epoch": 1.497806074163297,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007948325164451872,
      "loss": 1.0178,
      "step": 17409
    },
    {
      "epoch": 1.4978921104706187,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007948106766204592,
      "loss": 1.6546,
      "step": 17410
    },
    {
      "epoch": 1.4979781467779403,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007947888359334671,
      "loss": 0.7655,
      "step": 17411
    },
    {
      "epoch": 1.498064183085262,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007947669943842753,
      "loss": 1.2772,
      "step": 17412
    },
    {
      "epoch": 1.4981502193925837,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000794745151972947,
      "loss": 1.9033,
      "step": 17413
    },
    {
      "epoch": 1.4982362556999054,
      "grad_norm": 3.390625,
      "learning_rate": 0.0007947233086995467,
      "loss": 1.6738,
      "step": 17414
    },
    {
      "epoch": 1.498322292007227,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007947014645641379,
      "loss": 2.1099,
      "step": 17415
    },
    {
      "epoch": 1.4984083283145488,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007946796195667848,
      "loss": 1.1256,
      "step": 17416
    },
    {
      "epoch": 1.4984943646218705,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007946577737075511,
      "loss": 0.842,
      "step": 17417
    },
    {
      "epoch": 1.498580400929192,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007946359269865006,
      "loss": 0.8986,
      "step": 17418
    },
    {
      "epoch": 1.4986664372365137,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007946140794036974,
      "loss": 1.3276,
      "step": 17419
    },
    {
      "epoch": 1.4987524735438356,
      "grad_norm": 1.125,
      "learning_rate": 0.0007945922309592053,
      "loss": 1.6579,
      "step": 17420
    },
    {
      "epoch": 1.4988385098511572,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000794570381653088,
      "loss": 1.911,
      "step": 17421
    },
    {
      "epoch": 1.4989245461584788,
      "grad_norm": 1.28125,
      "learning_rate": 0.00079454853148541,
      "loss": 1.2558,
      "step": 17422
    },
    {
      "epoch": 1.4990105824658007,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007945266804562346,
      "loss": 1.396,
      "step": 17423
    },
    {
      "epoch": 1.4990966187731223,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000794504828565626,
      "loss": 1.3537,
      "step": 17424
    },
    {
      "epoch": 1.499182655080444,
      "grad_norm": 1.0625,
      "learning_rate": 0.000794482975813648,
      "loss": 0.929,
      "step": 17425
    },
    {
      "epoch": 1.4992686913877655,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007944611222003647,
      "loss": 1.0025,
      "step": 17426
    },
    {
      "epoch": 1.4993547276950874,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007944392677258398,
      "loss": 1.204,
      "step": 17427
    },
    {
      "epoch": 1.499440764002409,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007944174123901373,
      "loss": 1.5472,
      "step": 17428
    },
    {
      "epoch": 1.4995268003097308,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007943955561933213,
      "loss": 1.6989,
      "step": 17429
    },
    {
      "epoch": 1.4996128366170525,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007943736991354553,
      "loss": 1.582,
      "step": 17430
    },
    {
      "epoch": 1.499698872924374,
      "grad_norm": 1.25,
      "learning_rate": 0.0007943518412166038,
      "loss": 1.0036,
      "step": 17431
    },
    {
      "epoch": 1.4997849092316957,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007943299824368301,
      "loss": 1.5603,
      "step": 17432
    },
    {
      "epoch": 1.4998709455390173,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007943081227961985,
      "loss": 0.958,
      "step": 17433
    },
    {
      "epoch": 1.4999569818463392,
      "grad_norm": 0.859375,
      "learning_rate": 0.000794286262294773,
      "loss": 0.6166,
      "step": 17434
    },
    {
      "epoch": 1.5000430181536608,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007942644009326173,
      "loss": 0.7161,
      "step": 17435
    },
    {
      "epoch": 1.5001290544609827,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007942425387097955,
      "loss": 0.6686,
      "step": 17436
    },
    {
      "epoch": 1.5001290544609827,
      "eval_loss": 1.2417023181915283,
      "eval_runtime": 356.5124,
      "eval_samples_per_second": 13.04,
      "eval_steps_per_second": 1.632,
      "step": 17436
    },
    {
      "epoch": 1.5002150907683043,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007942206756263715,
      "loss": 1.2041,
      "step": 17437
    },
    {
      "epoch": 1.500301127075626,
      "grad_norm": 1.765625,
      "learning_rate": 0.0007941988116824093,
      "loss": 1.3743,
      "step": 17438
    },
    {
      "epoch": 1.5003871633829475,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007941769468779727,
      "loss": 1.2934,
      "step": 17439
    },
    {
      "epoch": 1.5004731996902692,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007941550812131257,
      "loss": 1.4311,
      "step": 17440
    },
    {
      "epoch": 1.500559235997591,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007941332146879325,
      "loss": 1.4189,
      "step": 17441
    },
    {
      "epoch": 1.5006452723049126,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007941113473024568,
      "loss": 1.367,
      "step": 17442
    },
    {
      "epoch": 1.5007313086122345,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007940894790567626,
      "loss": 1.2861,
      "step": 17443
    },
    {
      "epoch": 1.500817344919556,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007940676099509138,
      "loss": 0.7506,
      "step": 17444
    },
    {
      "epoch": 1.5009033812268777,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007940457399849743,
      "loss": 1.1163,
      "step": 17445
    },
    {
      "epoch": 1.5009894175341993,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007940238691590085,
      "loss": 1.4207,
      "step": 17446
    },
    {
      "epoch": 1.501075453841521,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007940019974730798,
      "loss": 1.6413,
      "step": 17447
    },
    {
      "epoch": 1.5011614901488428,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007939801249272527,
      "loss": 0.8462,
      "step": 17448
    },
    {
      "epoch": 1.5012475264561647,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007939582515215908,
      "loss": 1.5384,
      "step": 17449
    },
    {
      "epoch": 1.5013335627634863,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000793936377256158,
      "loss": 1.6664,
      "step": 17450
    },
    {
      "epoch": 1.501419599070808,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007939145021310188,
      "loss": 1.8335,
      "step": 17451
    },
    {
      "epoch": 1.5015056353781295,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007938926261462366,
      "loss": 0.9547,
      "step": 17452
    },
    {
      "epoch": 1.5015916716854512,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007938707493018758,
      "loss": 0.5591,
      "step": 17453
    },
    {
      "epoch": 1.501677707992773,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000793848871598,
      "loss": 1.3085,
      "step": 17454
    },
    {
      "epoch": 1.5017637443000946,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007938269930346736,
      "loss": 1.168,
      "step": 17455
    },
    {
      "epoch": 1.5018497806074165,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007938051136119602,
      "loss": 1.051,
      "step": 17456
    },
    {
      "epoch": 1.501935816914738,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007937832333299242,
      "loss": 2.0763,
      "step": 17457
    },
    {
      "epoch": 1.5020218532220597,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007937613521886291,
      "loss": 1.2667,
      "step": 17458
    },
    {
      "epoch": 1.5021078895293813,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007937394701881394,
      "loss": 1.7013,
      "step": 17459
    },
    {
      "epoch": 1.502193925836703,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007937175873285187,
      "loss": 0.7395,
      "step": 17460
    },
    {
      "epoch": 1.5022799621440248,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007936957036098315,
      "loss": 1.0095,
      "step": 17461
    },
    {
      "epoch": 1.5023659984513464,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007936738190321413,
      "loss": 1.084,
      "step": 17462
    },
    {
      "epoch": 1.5024520347586683,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007936519335955122,
      "loss": 1.9839,
      "step": 17463
    },
    {
      "epoch": 1.50253807106599,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007936300473000084,
      "loss": 0.9046,
      "step": 17464
    },
    {
      "epoch": 1.5026241073733115,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0007936081601456938,
      "loss": 1.1164,
      "step": 17465
    },
    {
      "epoch": 1.5027101436806332,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007935862721326326,
      "loss": 1.6736,
      "step": 17466
    },
    {
      "epoch": 1.5027961799879548,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007935643832608884,
      "loss": 1.314,
      "step": 17467
    },
    {
      "epoch": 1.5028822162952766,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007935424935305255,
      "loss": 1.5891,
      "step": 17468
    },
    {
      "epoch": 1.5029682526025983,
      "grad_norm": 1.234375,
      "learning_rate": 0.000793520602941608,
      "loss": 0.7723,
      "step": 17469
    },
    {
      "epoch": 1.50305428890992,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007934987114941998,
      "loss": 1.7049,
      "step": 17470
    },
    {
      "epoch": 1.5031403252172417,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000793476819188365,
      "loss": 1.0996,
      "step": 17471
    },
    {
      "epoch": 1.5032263615245633,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007934549260241676,
      "loss": 1.4234,
      "step": 17472
    },
    {
      "epoch": 1.503312397831885,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007934330320016716,
      "loss": 1.7721,
      "step": 17473
    },
    {
      "epoch": 1.5033984341392066,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000793411137120941,
      "loss": 1.7069,
      "step": 17474
    },
    {
      "epoch": 1.5034844704465284,
      "grad_norm": 1.0,
      "learning_rate": 0.0007933892413820398,
      "loss": 1.743,
      "step": 17475
    },
    {
      "epoch": 1.5035705067538503,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007933673447850322,
      "loss": 1.7427,
      "step": 17476
    },
    {
      "epoch": 1.503656543061172,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007933454473299822,
      "loss": 0.9637,
      "step": 17477
    },
    {
      "epoch": 1.5037425793684935,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007933235490169538,
      "loss": 1.1233,
      "step": 17478
    },
    {
      "epoch": 1.5038286156758152,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000793301649846011,
      "loss": 1.6077,
      "step": 17479
    },
    {
      "epoch": 1.5039146519831368,
      "grad_norm": 1.0,
      "learning_rate": 0.0007932797498172181,
      "loss": 1.7836,
      "step": 17480
    },
    {
      "epoch": 1.5040006882904584,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007932578489306388,
      "loss": 0.9768,
      "step": 17481
    },
    {
      "epoch": 1.5040867245977803,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007932359471863374,
      "loss": 1.1287,
      "step": 17482
    },
    {
      "epoch": 1.504172760905102,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007932140445843778,
      "loss": 0.9926,
      "step": 17483
    },
    {
      "epoch": 1.5042587972124237,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007931921411248244,
      "loss": 1.5833,
      "step": 17484
    },
    {
      "epoch": 1.5043448335197454,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007931702368077408,
      "loss": 1.073,
      "step": 17485
    },
    {
      "epoch": 1.504430869827067,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007931483316331913,
      "loss": 1.6819,
      "step": 17486
    },
    {
      "epoch": 1.5045169061343886,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00079312642560124,
      "loss": 1.517,
      "step": 17487
    },
    {
      "epoch": 1.5046029424417104,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007931045187119508,
      "loss": 1.4955,
      "step": 17488
    },
    {
      "epoch": 1.504688978749032,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007930826109653881,
      "loss": 1.4184,
      "step": 17489
    },
    {
      "epoch": 1.504775015056354,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007930607023616155,
      "loss": 1.0862,
      "step": 17490
    },
    {
      "epoch": 1.5048610513636755,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007930387929006975,
      "loss": 1.7931,
      "step": 17491
    },
    {
      "epoch": 1.5049470876709972,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000793016882582698,
      "loss": 1.5959,
      "step": 17492
    },
    {
      "epoch": 1.5050331239783188,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007929949714076812,
      "loss": 0.9074,
      "step": 17493
    },
    {
      "epoch": 1.5051191602856404,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007929730593757109,
      "loss": 0.9995,
      "step": 17494
    },
    {
      "epoch": 1.5052051965929623,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007929511464868515,
      "loss": 0.909,
      "step": 17495
    },
    {
      "epoch": 1.5052912329002839,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007929292327411668,
      "loss": 1.0451,
      "step": 17496
    },
    {
      "epoch": 1.5053772692076057,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007929073181387212,
      "loss": 1.0802,
      "step": 17497
    },
    {
      "epoch": 1.5054633055149274,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007928854026795787,
      "loss": 1.5516,
      "step": 17498
    },
    {
      "epoch": 1.505549341822249,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007928634863638034,
      "loss": 1.4977,
      "step": 17499
    },
    {
      "epoch": 1.5056353781295706,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007928415691914592,
      "loss": 1.1102,
      "step": 17500
    },
    {
      "epoch": 1.5057214144368922,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007928196511626105,
      "loss": 1.1561,
      "step": 17501
    },
    {
      "epoch": 1.505807450744214,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007927977322773213,
      "loss": 1.4004,
      "step": 17502
    },
    {
      "epoch": 1.5058934870515357,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007927758125356556,
      "loss": 0.6978,
      "step": 17503
    },
    {
      "epoch": 1.5059795233588575,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007927538919376776,
      "loss": 1.0873,
      "step": 17504
    },
    {
      "epoch": 1.5060655596661792,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007927319704834515,
      "loss": 1.053,
      "step": 17505
    },
    {
      "epoch": 1.5061515959735008,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007927100481730411,
      "loss": 2.0841,
      "step": 17506
    },
    {
      "epoch": 1.5062376322808224,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007926881250065108,
      "loss": 1.0239,
      "step": 17507
    },
    {
      "epoch": 1.506323668588144,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007926662009839248,
      "loss": 1.459,
      "step": 17508
    },
    {
      "epoch": 1.5064097048954659,
      "grad_norm": 0.9375,
      "learning_rate": 0.000792644276105347,
      "loss": 1.2417,
      "step": 17509
    },
    {
      "epoch": 1.5064957412027877,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007926223503708416,
      "loss": 1.3647,
      "step": 17510
    },
    {
      "epoch": 1.5065817775101094,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007926004237804728,
      "loss": 0.7395,
      "step": 17511
    },
    {
      "epoch": 1.506667813817431,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007925784963343045,
      "loss": 0.7068,
      "step": 17512
    },
    {
      "epoch": 1.5067538501247526,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007925565680324012,
      "loss": 1.1243,
      "step": 17513
    },
    {
      "epoch": 1.5068398864320742,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007925346388748266,
      "loss": 2.1852,
      "step": 17514
    },
    {
      "epoch": 1.506925922739396,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007925127088616453,
      "loss": 1.1966,
      "step": 17515
    },
    {
      "epoch": 1.5070119590467177,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007924907779929211,
      "loss": 1.6255,
      "step": 17516
    },
    {
      "epoch": 1.5070979953540395,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007924688462687184,
      "loss": 1.0345,
      "step": 17517
    },
    {
      "epoch": 1.5071840316613612,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007924469136891012,
      "loss": 1.3166,
      "step": 17518
    },
    {
      "epoch": 1.5072700679686828,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007924249802541334,
      "loss": 1.0993,
      "step": 17519
    },
    {
      "epoch": 1.5073561042760044,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007924030459638797,
      "loss": 0.8985,
      "step": 17520
    },
    {
      "epoch": 1.507442140583326,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007923811108184038,
      "loss": 0.4941,
      "step": 17521
    },
    {
      "epoch": 1.5075281768906479,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00079235917481777,
      "loss": 1.3339,
      "step": 17522
    },
    {
      "epoch": 1.5076142131979695,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007923372379620426,
      "loss": 0.9677,
      "step": 17523
    },
    {
      "epoch": 1.5077002495052914,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007923153002512857,
      "loss": 0.9838,
      "step": 17524
    },
    {
      "epoch": 1.507786285812613,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007922933616855632,
      "loss": 1.7953,
      "step": 17525
    },
    {
      "epoch": 1.5078723221199346,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007922714222649395,
      "loss": 1.0311,
      "step": 17526
    },
    {
      "epoch": 1.5079583584272562,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007922494819894788,
      "loss": 1.0175,
      "step": 17527
    },
    {
      "epoch": 1.5080443947345779,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007922275408592453,
      "loss": 1.6238,
      "step": 17528
    },
    {
      "epoch": 1.5081304310418997,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007922055988743027,
      "loss": 0.8168,
      "step": 17529
    },
    {
      "epoch": 1.5082164673492213,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007921836560347157,
      "loss": 0.9202,
      "step": 17530
    },
    {
      "epoch": 1.5083025036565432,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007921617123405485,
      "loss": 0.6605,
      "step": 17531
    },
    {
      "epoch": 1.5083885399638648,
      "grad_norm": 1.859375,
      "learning_rate": 0.0007921397677918651,
      "loss": 1.7327,
      "step": 17532
    },
    {
      "epoch": 1.5084745762711864,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007921178223887295,
      "loss": 1.5069,
      "step": 17533
    },
    {
      "epoch": 1.508560612578508,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007920958761312064,
      "loss": 0.8727,
      "step": 17534
    },
    {
      "epoch": 1.5086466488858297,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007920739290193595,
      "loss": 1.9063,
      "step": 17535
    },
    {
      "epoch": 1.5087326851931515,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0007920519810532532,
      "loss": 1.5733,
      "step": 17536
    },
    {
      "epoch": 1.5088187215004734,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0007920300322329514,
      "loss": 1.7368,
      "step": 17537
    },
    {
      "epoch": 1.508904757807795,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007920080825585189,
      "loss": 0.8716,
      "step": 17538
    },
    {
      "epoch": 1.5089907941151166,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007919861320300195,
      "loss": 1.3408,
      "step": 17539
    },
    {
      "epoch": 1.5090768304224382,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007919641806475173,
      "loss": 0.984,
      "step": 17540
    },
    {
      "epoch": 1.5091628667297599,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007919422284110766,
      "loss": 1.6079,
      "step": 17541
    },
    {
      "epoch": 1.5092489030370817,
      "grad_norm": 1.1875,
      "learning_rate": 0.000791920275320762,
      "loss": 1.033,
      "step": 17542
    },
    {
      "epoch": 1.5093349393444033,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007918983213766371,
      "loss": 1.1106,
      "step": 17543
    },
    {
      "epoch": 1.5094209756517252,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007918763665787664,
      "loss": 0.9583,
      "step": 17544
    },
    {
      "epoch": 1.5095070119590468,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007918544109272141,
      "loss": 1.3324,
      "step": 17545
    },
    {
      "epoch": 1.5095930482663684,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007918324544220445,
      "loss": 1.8674,
      "step": 17546
    },
    {
      "epoch": 1.50967908457369,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007918104970633216,
      "loss": 1.2737,
      "step": 17547
    },
    {
      "epoch": 1.5097651208810117,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007917885388511099,
      "loss": 0.7091,
      "step": 17548
    },
    {
      "epoch": 1.5098511571883335,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007917665797854734,
      "loss": 1.1362,
      "step": 17549
    },
    {
      "epoch": 1.5099371934956551,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007917446198664764,
      "loss": 1.3272,
      "step": 17550
    },
    {
      "epoch": 1.510023229802977,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007917226590941831,
      "loss": 1.5063,
      "step": 17551
    },
    {
      "epoch": 1.5101092661102986,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007917006974686578,
      "loss": 2.0317,
      "step": 17552
    },
    {
      "epoch": 1.5101953024176202,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007916787349899646,
      "loss": 0.6531,
      "step": 17553
    },
    {
      "epoch": 1.5102813387249419,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000791656771658168,
      "loss": 1.3976,
      "step": 17554
    },
    {
      "epoch": 1.5103673750322635,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000791634807473332,
      "loss": 1.6447,
      "step": 17555
    },
    {
      "epoch": 1.5104534113395853,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007916128424355208,
      "loss": 1.7368,
      "step": 17556
    },
    {
      "epoch": 1.510539447646907,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007915908765447988,
      "loss": 1.0585,
      "step": 17557
    },
    {
      "epoch": 1.5106254839542288,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007915689098012302,
      "loss": 1.4651,
      "step": 17558
    },
    {
      "epoch": 1.5107115202615504,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007915469422048793,
      "loss": 1.7063,
      "step": 17559
    },
    {
      "epoch": 1.510797556568872,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007915249737558103,
      "loss": 1.824,
      "step": 17560
    },
    {
      "epoch": 1.5108835928761937,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0007915030044540874,
      "loss": 1.0176,
      "step": 17561
    },
    {
      "epoch": 1.5109696291835153,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000791481034299775,
      "loss": 1.1765,
      "step": 17562
    },
    {
      "epoch": 1.5110556654908371,
      "grad_norm": 1.375,
      "learning_rate": 0.0007914590632929372,
      "loss": 1.0789,
      "step": 17563
    },
    {
      "epoch": 1.511141701798159,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007914370914336382,
      "loss": 1.5804,
      "step": 17564
    },
    {
      "epoch": 1.5112277381054806,
      "grad_norm": 1.125,
      "learning_rate": 0.0007914151187219425,
      "loss": 1.0908,
      "step": 17565
    },
    {
      "epoch": 1.5113137744128022,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007913931451579144,
      "loss": 1.1886,
      "step": 17566
    },
    {
      "epoch": 1.5113998107201239,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007913711707416178,
      "loss": 1.7321,
      "step": 17567
    },
    {
      "epoch": 1.5114858470274455,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007913491954731173,
      "loss": 1.4079,
      "step": 17568
    },
    {
      "epoch": 1.511571883334767,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000791327219352477,
      "loss": 1.2677,
      "step": 17569
    },
    {
      "epoch": 1.511657919642089,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007913052423797613,
      "loss": 1.5025,
      "step": 17570
    },
    {
      "epoch": 1.5117439559494108,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007912832645550344,
      "loss": 1.6416,
      "step": 17571
    },
    {
      "epoch": 1.5118299922567324,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007912612858783605,
      "loss": 1.5034,
      "step": 17572
    },
    {
      "epoch": 1.511916028564054,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007912393063498043,
      "loss": 1.0024,
      "step": 17573
    },
    {
      "epoch": 1.5120020648713757,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007912173259694295,
      "loss": 1.401,
      "step": 17574
    },
    {
      "epoch": 1.5120881011786973,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007911953447373007,
      "loss": 1.6586,
      "step": 17575
    },
    {
      "epoch": 1.5121741374860191,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007911733626534822,
      "loss": 1.8326,
      "step": 17576
    },
    {
      "epoch": 1.5122601737933408,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007911513797180383,
      "loss": 1.5078,
      "step": 17577
    },
    {
      "epoch": 1.5123462101006626,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007911293959310331,
      "loss": 0.9092,
      "step": 17578
    },
    {
      "epoch": 1.5124322464079842,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007911074112925312,
      "loss": 0.8595,
      "step": 17579
    },
    {
      "epoch": 1.5125182827153059,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007910854258025966,
      "loss": 0.9195,
      "step": 17580
    },
    {
      "epoch": 1.5126043190226275,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007910634394612937,
      "loss": 1.6812,
      "step": 17581
    },
    {
      "epoch": 1.512690355329949,
      "grad_norm": 1.484375,
      "learning_rate": 0.000791041452268687,
      "loss": 2.0293,
      "step": 17582
    },
    {
      "epoch": 1.512776391637271,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007910194642248405,
      "loss": 1.3033,
      "step": 17583
    },
    {
      "epoch": 1.5128624279445926,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007909974753298188,
      "loss": 1.4688,
      "step": 17584
    },
    {
      "epoch": 1.5129484642519144,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007909754855836862,
      "loss": 1.1505,
      "step": 17585
    },
    {
      "epoch": 1.513034500559236,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007909534949865066,
      "loss": 1.2873,
      "step": 17586
    },
    {
      "epoch": 1.5131205368665577,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007909315035383446,
      "loss": 1.6993,
      "step": 17587
    },
    {
      "epoch": 1.5132065731738793,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007909095112392648,
      "loss": 1.8491,
      "step": 17588
    },
    {
      "epoch": 1.513292609481201,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007908875180893311,
      "loss": 1.6212,
      "step": 17589
    },
    {
      "epoch": 1.5133786457885228,
      "grad_norm": 1.125,
      "learning_rate": 0.000790865524088608,
      "loss": 1.1494,
      "step": 17590
    },
    {
      "epoch": 1.5134646820958444,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0007908435292371599,
      "loss": 0.7744,
      "step": 17591
    },
    {
      "epoch": 1.5135507184031662,
      "grad_norm": 1.078125,
      "learning_rate": 0.000790821533535051,
      "loss": 1.0448,
      "step": 17592
    },
    {
      "epoch": 1.5136367547104879,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007907995369823458,
      "loss": 1.1199,
      "step": 17593
    },
    {
      "epoch": 1.5137227910178095,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007907775395791082,
      "loss": 0.7132,
      "step": 17594
    },
    {
      "epoch": 1.5138088273251311,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007907555413254031,
      "loss": 1.1043,
      "step": 17595
    },
    {
      "epoch": 1.5138948636324527,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007907335422212945,
      "loss": 1.7032,
      "step": 17596
    },
    {
      "epoch": 1.5139808999397746,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007907115422668468,
      "loss": 1.5152,
      "step": 17597
    },
    {
      "epoch": 1.5140669362470964,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007906895414621245,
      "loss": 1.2896,
      "step": 17598
    },
    {
      "epoch": 1.514152972554418,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007906675398071918,
      "loss": 1.5679,
      "step": 17599
    },
    {
      "epoch": 1.5142390088617397,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007906455373021128,
      "loss": 1.5669,
      "step": 17600
    },
    {
      "epoch": 1.5143250451690613,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007906235339469525,
      "loss": 1.1021,
      "step": 17601
    },
    {
      "epoch": 1.514411081476383,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007906015297417748,
      "loss": 1.6809,
      "step": 17602
    },
    {
      "epoch": 1.5144971177837048,
      "grad_norm": 1.375,
      "learning_rate": 0.0007905795246866441,
      "loss": 2.1767,
      "step": 17603
    },
    {
      "epoch": 1.5145831540910264,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007905575187816248,
      "loss": 1.0044,
      "step": 17604
    },
    {
      "epoch": 1.5146691903983482,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007905355120267812,
      "loss": 0.9253,
      "step": 17605
    },
    {
      "epoch": 1.5147552267056699,
      "grad_norm": 2.140625,
      "learning_rate": 0.0007905135044221777,
      "loss": 1.4782,
      "step": 17606
    },
    {
      "epoch": 1.5148412630129915,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007904914959678788,
      "loss": 0.7859,
      "step": 17607
    },
    {
      "epoch": 1.5149272993203131,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007904694866639487,
      "loss": 0.8266,
      "step": 17608
    },
    {
      "epoch": 1.5150133356276347,
      "grad_norm": 1.265625,
      "learning_rate": 0.000790447476510452,
      "loss": 2.0554,
      "step": 17609
    },
    {
      "epoch": 1.5150993719349566,
      "grad_norm": 1.25,
      "learning_rate": 0.0007904254655074529,
      "loss": 1.6276,
      "step": 17610
    },
    {
      "epoch": 1.5151854082422782,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007904034536550157,
      "loss": 0.9136,
      "step": 17611
    },
    {
      "epoch": 1.5152714445496,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007903814409532049,
      "loss": 1.2002,
      "step": 17612
    },
    {
      "epoch": 1.5153574808569217,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007903594274020849,
      "loss": 1.0156,
      "step": 17613
    },
    {
      "epoch": 1.5154435171642433,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007903374130017201,
      "loss": 1.3322,
      "step": 17614
    },
    {
      "epoch": 1.515529553471565,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007903153977521746,
      "loss": 0.6593,
      "step": 17615
    },
    {
      "epoch": 1.5156155897788866,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007902933816535132,
      "loss": 1.8938,
      "step": 17616
    },
    {
      "epoch": 1.5157016260862084,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007902713647058002,
      "loss": 1.7069,
      "step": 17617
    },
    {
      "epoch": 1.51578766239353,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007902493469090997,
      "loss": 1.2873,
      "step": 17618
    },
    {
      "epoch": 1.5158736987008519,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007902273282634764,
      "loss": 1.5272,
      "step": 17619
    },
    {
      "epoch": 1.5159597350081735,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007902053087689946,
      "loss": 0.7642,
      "step": 17620
    },
    {
      "epoch": 1.5160457713154951,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007901832884257189,
      "loss": 1.3555,
      "step": 17621
    },
    {
      "epoch": 1.5161318076228167,
      "grad_norm": 1.25,
      "learning_rate": 0.0007901612672337133,
      "loss": 1.7643,
      "step": 17622
    },
    {
      "epoch": 1.5162178439301384,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007901392451930424,
      "loss": 1.1211,
      "step": 17623
    },
    {
      "epoch": 1.5163038802374602,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007901172223037708,
      "loss": 1.2832,
      "step": 17624
    },
    {
      "epoch": 1.516389916544782,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007900951985659627,
      "loss": 1.3293,
      "step": 17625
    },
    {
      "epoch": 1.5164759528521037,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007900731739796825,
      "loss": 1.5242,
      "step": 17626
    },
    {
      "epoch": 1.5165619891594253,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007900511485449947,
      "loss": 1.051,
      "step": 17627
    },
    {
      "epoch": 1.516648025466747,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007900291222619638,
      "loss": 0.9767,
      "step": 17628
    },
    {
      "epoch": 1.5167340617740686,
      "grad_norm": 1.4375,
      "learning_rate": 0.000790007095130654,
      "loss": 1.2842,
      "step": 17629
    },
    {
      "epoch": 1.5168200980813904,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007899850671511299,
      "loss": 1.3791,
      "step": 17630
    },
    {
      "epoch": 1.516906134388712,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007899630383234559,
      "loss": 1.4835,
      "step": 17631
    },
    {
      "epoch": 1.5169921706960339,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007899410086476963,
      "loss": 1.4225,
      "step": 17632
    },
    {
      "epoch": 1.5170782070033555,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007899189781239158,
      "loss": 1.0095,
      "step": 17633
    },
    {
      "epoch": 1.5171642433106771,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007898969467521786,
      "loss": 1.154,
      "step": 17634
    },
    {
      "epoch": 1.5172502796179987,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007898749145325493,
      "loss": 0.9278,
      "step": 17635
    },
    {
      "epoch": 1.5173363159253204,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007898528814650921,
      "loss": 1.6111,
      "step": 17636
    },
    {
      "epoch": 1.5174223522326422,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007898308475498717,
      "loss": 0.855,
      "step": 17637
    },
    {
      "epoch": 1.5175083885399638,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007898088127869524,
      "loss": 1.6472,
      "step": 17638
    },
    {
      "epoch": 1.5175944248472857,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007897867771763987,
      "loss": 1.9688,
      "step": 17639
    },
    {
      "epoch": 1.5176804611546073,
      "grad_norm": 1.15625,
      "learning_rate": 0.000789764740718275,
      "loss": 0.6943,
      "step": 17640
    },
    {
      "epoch": 1.517766497461929,
      "grad_norm": 0.875,
      "learning_rate": 0.0007897427034126459,
      "loss": 1.0543,
      "step": 17641
    },
    {
      "epoch": 1.5178525337692506,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007897206652595756,
      "loss": 1.8853,
      "step": 17642
    },
    {
      "epoch": 1.5179385700765722,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0007896986262591286,
      "loss": 1.1627,
      "step": 17643
    },
    {
      "epoch": 1.518024606383894,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007896765864113698,
      "loss": 1.473,
      "step": 17644
    },
    {
      "epoch": 1.5181106426912157,
      "grad_norm": 1.328125,
      "learning_rate": 0.000789654545716363,
      "loss": 1.891,
      "step": 17645
    },
    {
      "epoch": 1.5181966789985375,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007896325041741732,
      "loss": 1.3798,
      "step": 17646
    },
    {
      "epoch": 1.5182827153058591,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007896104617848645,
      "loss": 1.4056,
      "step": 17647
    },
    {
      "epoch": 1.5183687516131807,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007895884185485018,
      "loss": 1.4931,
      "step": 17648
    },
    {
      "epoch": 1.5184547879205024,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000789566374465149,
      "loss": 1.3536,
      "step": 17649
    },
    {
      "epoch": 1.518540824227824,
      "grad_norm": 1.234375,
      "learning_rate": 0.000789544329534871,
      "loss": 1.6308,
      "step": 17650
    },
    {
      "epoch": 1.5186268605351458,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000789522283757732,
      "loss": 1.8991,
      "step": 17651
    },
    {
      "epoch": 1.5187128968424677,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007895002371337969,
      "loss": 1.824,
      "step": 17652
    },
    {
      "epoch": 1.5187989331497893,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007894781896631297,
      "loss": 0.8089,
      "step": 17653
    },
    {
      "epoch": 1.518884969457111,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007894561413457952,
      "loss": 1.9974,
      "step": 17654
    },
    {
      "epoch": 1.5189710057644326,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007894340921818578,
      "loss": 1.1458,
      "step": 17655
    },
    {
      "epoch": 1.5190570420717542,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007894120421713819,
      "loss": 1.3966,
      "step": 17656
    },
    {
      "epoch": 1.5191430783790758,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007893899913144322,
      "loss": 1.1957,
      "step": 17657
    },
    {
      "epoch": 1.5192291146863977,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007893679396110729,
      "loss": 1.098,
      "step": 17658
    },
    {
      "epoch": 1.5193151509937195,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007893458870613687,
      "loss": 1.7332,
      "step": 17659
    },
    {
      "epoch": 1.5194011873010411,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007893238336653841,
      "loss": 0.8955,
      "step": 17660
    },
    {
      "epoch": 1.5194872236083627,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007893017794231834,
      "loss": 0.9911,
      "step": 17661
    },
    {
      "epoch": 1.5195732599156844,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007892797243348313,
      "loss": 1.3717,
      "step": 17662
    },
    {
      "epoch": 1.519659296223006,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007892576684003925,
      "loss": 1.555,
      "step": 17663
    },
    {
      "epoch": 1.5197453325303278,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000789235611619931,
      "loss": 1.8141,
      "step": 17664
    },
    {
      "epoch": 1.5198313688376495,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007892135539935117,
      "loss": 1.1034,
      "step": 17665
    },
    {
      "epoch": 1.5199174051449713,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007891914955211991,
      "loss": 2.1926,
      "step": 17666
    },
    {
      "epoch": 1.520003441452293,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007891694362030574,
      "loss": 1.3298,
      "step": 17667
    },
    {
      "epoch": 1.5200894777596146,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007891473760391517,
      "loss": 1.6044,
      "step": 17668
    },
    {
      "epoch": 1.5201755140669362,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007891253150295459,
      "loss": 0.7444,
      "step": 17669
    },
    {
      "epoch": 1.5202615503742578,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007891032531743047,
      "loss": 1.1717,
      "step": 17670
    },
    {
      "epoch": 1.5203475866815797,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000789081190473493,
      "loss": 1.1722,
      "step": 17671
    },
    {
      "epoch": 1.5204336229889013,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007890591269271748,
      "loss": 0.6826,
      "step": 17672
    },
    {
      "epoch": 1.5205196592962231,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007890370625354148,
      "loss": 1.582,
      "step": 17673
    },
    {
      "epoch": 1.5206056956035447,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007890149972982778,
      "loss": 1.9148,
      "step": 17674
    },
    {
      "epoch": 1.5206917319108664,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007889929312158282,
      "loss": 0.9965,
      "step": 17675
    },
    {
      "epoch": 1.520777768218188,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007889708642881304,
      "loss": 1.2707,
      "step": 17676
    },
    {
      "epoch": 1.5208638045255096,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007889487965152489,
      "loss": 1.1324,
      "step": 17677
    },
    {
      "epoch": 1.5209498408328315,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007889267278972485,
      "loss": 0.9585,
      "step": 17678
    },
    {
      "epoch": 1.521035877140153,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007889046584341937,
      "loss": 1.7572,
      "step": 17679
    },
    {
      "epoch": 1.521121913447475,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007888825881261488,
      "loss": 2.1529,
      "step": 17680
    },
    {
      "epoch": 1.5212079497547966,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007888605169731784,
      "loss": 1.2942,
      "step": 17681
    },
    {
      "epoch": 1.5212939860621182,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007888384449753474,
      "loss": 1.781,
      "step": 17682
    },
    {
      "epoch": 1.5213800223694398,
      "grad_norm": 1.5625,
      "learning_rate": 0.00078881637213272,
      "loss": 1.5943,
      "step": 17683
    },
    {
      "epoch": 1.5214660586767614,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007887942984453609,
      "loss": 1.6583,
      "step": 17684
    },
    {
      "epoch": 1.5215520949840833,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007887722239133348,
      "loss": 1.3782,
      "step": 17685
    },
    {
      "epoch": 1.5216381312914051,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007887501485367059,
      "loss": 1.612,
      "step": 17686
    },
    {
      "epoch": 1.5217241675987268,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007887280723155391,
      "loss": 1.969,
      "step": 17687
    },
    {
      "epoch": 1.5218102039060484,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007887059952498989,
      "loss": 0.9384,
      "step": 17688
    },
    {
      "epoch": 1.52189624021337,
      "grad_norm": 1.25,
      "learning_rate": 0.0007886839173398497,
      "loss": 1.0118,
      "step": 17689
    },
    {
      "epoch": 1.5219822765206916,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007886618385854561,
      "loss": 1.0741,
      "step": 17690
    },
    {
      "epoch": 1.5220683128280135,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007886397589867829,
      "loss": 0.8186,
      "step": 17691
    },
    {
      "epoch": 1.522154349135335,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007886176785438945,
      "loss": 1.2155,
      "step": 17692
    },
    {
      "epoch": 1.522240385442657,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007885955972568556,
      "loss": 1.2895,
      "step": 17693
    },
    {
      "epoch": 1.5223264217499786,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007885735151257307,
      "loss": 0.9944,
      "step": 17694
    },
    {
      "epoch": 1.5224124580573002,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007885514321505843,
      "loss": 0.784,
      "step": 17695
    },
    {
      "epoch": 1.5224984943646218,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000788529348331481,
      "loss": 1.2021,
      "step": 17696
    },
    {
      "epoch": 1.5225845306719434,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007885072636684857,
      "loss": 1.3593,
      "step": 17697
    },
    {
      "epoch": 1.5226705669792653,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007884851781616626,
      "loss": 0.9157,
      "step": 17698
    },
    {
      "epoch": 1.522756603286587,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007884630918110764,
      "loss": 1.4087,
      "step": 17699
    },
    {
      "epoch": 1.5228426395939088,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000788441004616792,
      "loss": 0.9217,
      "step": 17700
    },
    {
      "epoch": 1.5229286759012304,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007884189165788735,
      "loss": 1.9003,
      "step": 17701
    },
    {
      "epoch": 1.523014712208552,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007883968276973861,
      "loss": 1.166,
      "step": 17702
    },
    {
      "epoch": 1.5231007485158736,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007883747379723938,
      "loss": 1.0034,
      "step": 17703
    },
    {
      "epoch": 1.5231867848231953,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007883526474039615,
      "loss": 1.8702,
      "step": 17704
    },
    {
      "epoch": 1.523272821130517,
      "grad_norm": 1.640625,
      "learning_rate": 0.0007883305559921539,
      "loss": 1.3926,
      "step": 17705
    },
    {
      "epoch": 1.5233588574378387,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007883084637370355,
      "loss": 1.8696,
      "step": 17706
    },
    {
      "epoch": 1.5234448937451606,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007882863706386707,
      "loss": 1.1489,
      "step": 17707
    },
    {
      "epoch": 1.5235309300524822,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007882642766971245,
      "loss": 0.7006,
      "step": 17708
    },
    {
      "epoch": 1.5236169663598038,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007882421819124614,
      "loss": 0.7626,
      "step": 17709
    },
    {
      "epoch": 1.5237030026671254,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007882200862847459,
      "loss": 1.8291,
      "step": 17710
    },
    {
      "epoch": 1.523789038974447,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007881979898140428,
      "loss": 1.3132,
      "step": 17711
    },
    {
      "epoch": 1.523875075281769,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007881758925004166,
      "loss": 1.818,
      "step": 17712
    },
    {
      "epoch": 1.5239611115890908,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007881537943439319,
      "loss": 1.8289,
      "step": 17713
    },
    {
      "epoch": 1.5240471478964124,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007881316953446534,
      "loss": 0.7588,
      "step": 17714
    },
    {
      "epoch": 1.524133184203734,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007881095955026456,
      "loss": 1.8053,
      "step": 17715
    },
    {
      "epoch": 1.5242192205110556,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007880874948179735,
      "loss": 1.1738,
      "step": 17716
    },
    {
      "epoch": 1.5243052568183773,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007880653932907013,
      "loss": 0.9685,
      "step": 17717
    },
    {
      "epoch": 1.524391293125699,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000788043290920894,
      "loss": 1.0739,
      "step": 17718
    },
    {
      "epoch": 1.5244773294330207,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007880211877086161,
      "loss": 1.6757,
      "step": 17719
    },
    {
      "epoch": 1.5245633657403426,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007879990836539321,
      "loss": 1.2396,
      "step": 17720
    },
    {
      "epoch": 1.5246494020476642,
      "grad_norm": 1.09375,
      "learning_rate": 0.000787976978756907,
      "loss": 1.194,
      "step": 17721
    },
    {
      "epoch": 1.5247354383549858,
      "grad_norm": 1.125,
      "learning_rate": 0.0007879548730176051,
      "loss": 2.173,
      "step": 17722
    },
    {
      "epoch": 1.5248214746623074,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007879327664360912,
      "loss": 1.043,
      "step": 17723
    },
    {
      "epoch": 1.524907510969629,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007879106590124301,
      "loss": 1.6729,
      "step": 17724
    },
    {
      "epoch": 1.524993547276951,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007878885507466861,
      "loss": 1.6229,
      "step": 17725
    },
    {
      "epoch": 1.5250795835842725,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007878664416389242,
      "loss": 1.3927,
      "step": 17726
    },
    {
      "epoch": 1.5251656198915944,
      "grad_norm": 1.28125,
      "learning_rate": 0.000787844331689209,
      "loss": 2.2279,
      "step": 17727
    },
    {
      "epoch": 1.525251656198916,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007878222208976051,
      "loss": 0.9833,
      "step": 17728
    },
    {
      "epoch": 1.5253376925062376,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007878001092641771,
      "loss": 1.0661,
      "step": 17729
    },
    {
      "epoch": 1.5254237288135593,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007877779967889898,
      "loss": 1.5466,
      "step": 17730
    },
    {
      "epoch": 1.5255097651208809,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007877558834721078,
      "loss": 1.4426,
      "step": 17731
    },
    {
      "epoch": 1.5255958014282027,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007877337693135959,
      "loss": 1.3298,
      "step": 17732
    },
    {
      "epoch": 1.5256818377355243,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007877116543135186,
      "loss": 1.3828,
      "step": 17733
    },
    {
      "epoch": 1.5257678740428462,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007876895384719407,
      "loss": 1.5533,
      "step": 17734
    },
    {
      "epoch": 1.5258539103501678,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007876674217889267,
      "loss": 1.4939,
      "step": 17735
    },
    {
      "epoch": 1.5259399466574894,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007876453042645416,
      "loss": 0.8465,
      "step": 17736
    },
    {
      "epoch": 1.526025982964811,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007876231858988498,
      "loss": 0.6684,
      "step": 17737
    },
    {
      "epoch": 1.5261120192721327,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007876010666919162,
      "loss": 1.1367,
      "step": 17738
    },
    {
      "epoch": 1.5261980555794545,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007875789466438056,
      "loss": 1.1862,
      "step": 17739
    },
    {
      "epoch": 1.5262840918867764,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007875568257545824,
      "loss": 1.0613,
      "step": 17740
    },
    {
      "epoch": 1.526370128194098,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007875347040243113,
      "loss": 1.0245,
      "step": 17741
    },
    {
      "epoch": 1.5264561645014196,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007875125814530571,
      "loss": 1.5287,
      "step": 17742
    },
    {
      "epoch": 1.5265422008087413,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007874904580408846,
      "loss": 0.8885,
      "step": 17743
    },
    {
      "epoch": 1.5266282371160629,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007874683337878585,
      "loss": 1.158,
      "step": 17744
    },
    {
      "epoch": 1.5267142734233845,
      "grad_norm": 1.125,
      "learning_rate": 0.0007874462086940433,
      "loss": 1.4959,
      "step": 17745
    },
    {
      "epoch": 1.5268003097307064,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007874240827595039,
      "loss": 0.7628,
      "step": 17746
    },
    {
      "epoch": 1.5268863460380282,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007874019559843049,
      "loss": 1.5511,
      "step": 17747
    },
    {
      "epoch": 1.5269723823453498,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007873798283685112,
      "loss": 0.8806,
      "step": 17748
    },
    {
      "epoch": 1.5270584186526714,
      "grad_norm": 1.125,
      "learning_rate": 0.0007873576999121873,
      "loss": 0.833,
      "step": 17749
    },
    {
      "epoch": 1.527144454959993,
      "grad_norm": 1.25,
      "learning_rate": 0.000787335570615398,
      "loss": 1.648,
      "step": 17750
    },
    {
      "epoch": 1.5272304912673147,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000787313440478208,
      "loss": 1.117,
      "step": 17751
    },
    {
      "epoch": 1.5273165275746365,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007872913095006824,
      "loss": 1.2551,
      "step": 17752
    },
    {
      "epoch": 1.5274025638819582,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007872691776828853,
      "loss": 1.6204,
      "step": 17753
    },
    {
      "epoch": 1.52748860018928,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007872470450248816,
      "loss": 1.1611,
      "step": 17754
    },
    {
      "epoch": 1.5275746364966016,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007872249115267364,
      "loss": 0.5971,
      "step": 17755
    },
    {
      "epoch": 1.5276606728039233,
      "grad_norm": 1.15625,
      "learning_rate": 0.000787202777188514,
      "loss": 1.296,
      "step": 17756
    },
    {
      "epoch": 1.5277467091112449,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007871806420102795,
      "loss": 1.2939,
      "step": 17757
    },
    {
      "epoch": 1.5278327454185665,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007871585059920974,
      "loss": 0.7946,
      "step": 17758
    },
    {
      "epoch": 1.5279187817258884,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007871363691340326,
      "loss": 2.0519,
      "step": 17759
    },
    {
      "epoch": 1.52800481803321,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007871142314361496,
      "loss": 1.2468,
      "step": 17760
    },
    {
      "epoch": 1.5280908543405318,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007870920928985136,
      "loss": 1.6353,
      "step": 17761
    },
    {
      "epoch": 1.5281768906478534,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007870699535211889,
      "loss": 1.1165,
      "step": 17762
    },
    {
      "epoch": 1.528262926955175,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007870478133042405,
      "loss": 1.5355,
      "step": 17763
    },
    {
      "epoch": 1.5283489632624967,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000787025672247733,
      "loss": 1.8705,
      "step": 17764
    },
    {
      "epoch": 1.5284349995698183,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007870035303517312,
      "loss": 1.5448,
      "step": 17765
    },
    {
      "epoch": 1.5285210358771402,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007869813876162998,
      "loss": 0.9773,
      "step": 17766
    },
    {
      "epoch": 1.5286070721844618,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007869592440415037,
      "loss": 1.483,
      "step": 17767
    },
    {
      "epoch": 1.5286931084917836,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007869370996274078,
      "loss": 0.7104,
      "step": 17768
    },
    {
      "epoch": 1.5287791447991053,
      "grad_norm": 1.25,
      "learning_rate": 0.0007869149543740766,
      "loss": 1.006,
      "step": 17769
    },
    {
      "epoch": 1.5288651811064269,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000786892808281575,
      "loss": 1.0932,
      "step": 17770
    },
    {
      "epoch": 1.5289512174137485,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007868706613499676,
      "loss": 0.9366,
      "step": 17771
    },
    {
      "epoch": 1.5290372537210701,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007868485135793193,
      "loss": 0.6166,
      "step": 17772
    },
    {
      "epoch": 1.529123290028392,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000786826364969695,
      "loss": 1.4692,
      "step": 17773
    },
    {
      "epoch": 1.5292093263357138,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007868042155211592,
      "loss": 0.7141,
      "step": 17774
    },
    {
      "epoch": 1.5292953626430354,
      "grad_norm": 1.375,
      "learning_rate": 0.0007867820652337771,
      "loss": 2.1068,
      "step": 17775
    },
    {
      "epoch": 1.529381398950357,
      "grad_norm": 1.234375,
      "learning_rate": 0.000786759914107613,
      "loss": 1.1662,
      "step": 17776
    },
    {
      "epoch": 1.5294674352576787,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007867377621427321,
      "loss": 0.9364,
      "step": 17777
    },
    {
      "epoch": 1.5295534715650003,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007867156093391989,
      "loss": 1.9933,
      "step": 17778
    },
    {
      "epoch": 1.5296395078723222,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007866934556970784,
      "loss": 1.2229,
      "step": 17779
    },
    {
      "epoch": 1.5297255441796438,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007866713012164352,
      "loss": 1.8266,
      "step": 17780
    },
    {
      "epoch": 1.5298115804869656,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007866491458973344,
      "loss": 2.3269,
      "step": 17781
    },
    {
      "epoch": 1.5298976167942873,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007866269897398403,
      "loss": 1.569,
      "step": 17782
    },
    {
      "epoch": 1.5299836531016089,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007866048327440183,
      "loss": 1.3161,
      "step": 17783
    },
    {
      "epoch": 1.5300696894089305,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007865826749099327,
      "loss": 0.8378,
      "step": 17784
    },
    {
      "epoch": 1.5301557257162521,
      "grad_norm": 1.25,
      "learning_rate": 0.0007865605162376486,
      "loss": 1.0232,
      "step": 17785
    },
    {
      "epoch": 1.530241762023574,
      "grad_norm": 1.25,
      "learning_rate": 0.0007865383567272308,
      "loss": 1.4717,
      "step": 17786
    },
    {
      "epoch": 1.5303277983308956,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007865161963787439,
      "loss": 1.003,
      "step": 17787
    },
    {
      "epoch": 1.5304138346382175,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000786494035192253,
      "loss": 2.0115,
      "step": 17788
    },
    {
      "epoch": 1.530499870945539,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007864718731678225,
      "loss": 1.7359,
      "step": 17789
    },
    {
      "epoch": 1.5305859072528607,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007864497103055177,
      "loss": 1.1996,
      "step": 17790
    },
    {
      "epoch": 1.5306719435601823,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007864275466054031,
      "loss": 1.3941,
      "step": 17791
    },
    {
      "epoch": 1.530757979867504,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007864053820675436,
      "loss": 1.1607,
      "step": 17792
    },
    {
      "epoch": 1.5308440161748258,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007863832166920042,
      "loss": 1.2392,
      "step": 17793
    },
    {
      "epoch": 1.5309300524821474,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007863610504788495,
      "loss": 1.1853,
      "step": 17794
    },
    {
      "epoch": 1.5310160887894693,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007863388834281444,
      "loss": 1.4315,
      "step": 17795
    },
    {
      "epoch": 1.531102125096791,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007863167155399539,
      "loss": 1.6944,
      "step": 17796
    },
    {
      "epoch": 1.5311881614041125,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007862945468143425,
      "loss": 1.9362,
      "step": 17797
    },
    {
      "epoch": 1.5312741977114341,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007862723772513754,
      "loss": 0.8882,
      "step": 17798
    },
    {
      "epoch": 1.5313602340187558,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007862502068511172,
      "loss": 1.6309,
      "step": 17799
    },
    {
      "epoch": 1.5314462703260776,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0007862280356136328,
      "loss": 1.2636,
      "step": 17800
    },
    {
      "epoch": 1.5315323066333995,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007862058635389872,
      "loss": 1.3331,
      "step": 17801
    },
    {
      "epoch": 1.531618342940721,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007861836906272449,
      "loss": 1.0828,
      "step": 17802
    },
    {
      "epoch": 1.5317043792480427,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007861615168784711,
      "loss": 1.6725,
      "step": 17803
    },
    {
      "epoch": 1.5317904155553643,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007861393422927306,
      "loss": 0.8887,
      "step": 17804
    },
    {
      "epoch": 1.531876451862686,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000786117166870088,
      "loss": 1.1844,
      "step": 17805
    },
    {
      "epoch": 1.5319624881700078,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007860949906106083,
      "loss": 0.9997,
      "step": 17806
    },
    {
      "epoch": 1.5320485244773294,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007860728135143565,
      "loss": 1.1367,
      "step": 17807
    },
    {
      "epoch": 1.5321345607846513,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007860506355813975,
      "loss": 1.5916,
      "step": 17808
    },
    {
      "epoch": 1.532220597091973,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0007860284568117957,
      "loss": 1.0845,
      "step": 17809
    },
    {
      "epoch": 1.5323066333992945,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007860062772056164,
      "loss": 1.5468,
      "step": 17810
    },
    {
      "epoch": 1.5323926697066161,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007859840967629245,
      "loss": 0.8928,
      "step": 17811
    },
    {
      "epoch": 1.5324787060139378,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0007859619154837848,
      "loss": 1.5521,
      "step": 17812
    },
    {
      "epoch": 1.5325647423212596,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007859397333682619,
      "loss": 1.436,
      "step": 17813
    },
    {
      "epoch": 1.5326507786285812,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007859175504164209,
      "loss": 1.0579,
      "step": 17814
    },
    {
      "epoch": 1.532736814935903,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007858953666283269,
      "loss": 1.2545,
      "step": 17815
    },
    {
      "epoch": 1.5328228512432247,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007858731820040443,
      "loss": 0.9519,
      "step": 17816
    },
    {
      "epoch": 1.5329088875505463,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007858509965436383,
      "loss": 1.0036,
      "step": 17817
    },
    {
      "epoch": 1.532994923857868,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007858288102471737,
      "loss": 1.353,
      "step": 17818
    },
    {
      "epoch": 1.5330809601651896,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007858066231147156,
      "loss": 1.8195,
      "step": 17819
    },
    {
      "epoch": 1.5331669964725114,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007857844351463286,
      "loss": 1.0176,
      "step": 17820
    },
    {
      "epoch": 1.533253032779833,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007857622463420777,
      "loss": 1.8913,
      "step": 17821
    },
    {
      "epoch": 1.533339069087155,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007857400567020279,
      "loss": 1.6302,
      "step": 17822
    },
    {
      "epoch": 1.5334251053944765,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007857178662262439,
      "loss": 1.195,
      "step": 17823
    },
    {
      "epoch": 1.5335111417017981,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007856956749147908,
      "loss": 0.701,
      "step": 17824
    },
    {
      "epoch": 1.5335971780091198,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007856734827677333,
      "loss": 0.8839,
      "step": 17825
    },
    {
      "epoch": 1.5336832143164414,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007856512897851365,
      "loss": 1.1488,
      "step": 17826
    },
    {
      "epoch": 1.5337692506237632,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007856290959670651,
      "loss": 0.9337,
      "step": 17827
    },
    {
      "epoch": 1.533855286931085,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007856069013135842,
      "loss": 1.7806,
      "step": 17828
    },
    {
      "epoch": 1.5339413232384067,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007855847058247589,
      "loss": 1.5179,
      "step": 17829
    },
    {
      "epoch": 1.5340273595457283,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007855625095006535,
      "loss": 1.3326,
      "step": 17830
    },
    {
      "epoch": 1.53411339585305,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007855403123413336,
      "loss": 1.5624,
      "step": 17831
    },
    {
      "epoch": 1.5341994321603716,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007855181143468637,
      "loss": 1.1398,
      "step": 17832
    },
    {
      "epoch": 1.5342854684676932,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000785495915517309,
      "loss": 1.2402,
      "step": 17833
    },
    {
      "epoch": 1.534371504775015,
      "grad_norm": 1.171875,
      "learning_rate": 0.000785473715852734,
      "loss": 1.394,
      "step": 17834
    },
    {
      "epoch": 1.534457541082337,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007854515153532041,
      "loss": 1.4422,
      "step": 17835
    },
    {
      "epoch": 1.5345435773896585,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000785429314018784,
      "loss": 1.5729,
      "step": 17836
    },
    {
      "epoch": 1.5346296136969801,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007854071118495386,
      "loss": 1.451,
      "step": 17837
    },
    {
      "epoch": 1.5347156500043018,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007853849088455329,
      "loss": 1.1578,
      "step": 17838
    },
    {
      "epoch": 1.5348016863116234,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000785362705006832,
      "loss": 1.71,
      "step": 17839
    },
    {
      "epoch": 1.5348877226189452,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007853405003335006,
      "loss": 1.8298,
      "step": 17840
    },
    {
      "epoch": 1.5349737589262669,
      "grad_norm": 2.0625,
      "learning_rate": 0.0007853182948256037,
      "loss": 1.3679,
      "step": 17841
    },
    {
      "epoch": 1.5350597952335887,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007852960884832063,
      "loss": 1.3316,
      "step": 17842
    },
    {
      "epoch": 1.5351458315409103,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007852738813063733,
      "loss": 1.4038,
      "step": 17843
    },
    {
      "epoch": 1.535231867848232,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007852516732951698,
      "loss": 1.5401,
      "step": 17844
    },
    {
      "epoch": 1.5353179041555536,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007852294644496605,
      "loss": 0.959,
      "step": 17845
    },
    {
      "epoch": 1.5354039404628752,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007852072547699105,
      "loss": 1.7281,
      "step": 17846
    },
    {
      "epoch": 1.535489976770197,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007851850442559847,
      "loss": 0.8583,
      "step": 17847
    },
    {
      "epoch": 1.5355760130775187,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007851628329079482,
      "loss": 1.1254,
      "step": 17848
    },
    {
      "epoch": 1.5356620493848405,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007851406207258658,
      "loss": 1.2534,
      "step": 17849
    },
    {
      "epoch": 1.5357480856921621,
      "grad_norm": 1.25,
      "learning_rate": 0.0007851184077098026,
      "loss": 1.2139,
      "step": 17850
    },
    {
      "epoch": 1.5358341219994838,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007850961938598236,
      "loss": 1.8682,
      "step": 17851
    },
    {
      "epoch": 1.5359201583068054,
      "grad_norm": 1.0,
      "learning_rate": 0.0007850739791759935,
      "loss": 0.7128,
      "step": 17852
    },
    {
      "epoch": 1.536006194614127,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0007850517636583776,
      "loss": 0.5957,
      "step": 17853
    },
    {
      "epoch": 1.5360922309214489,
      "grad_norm": 1.25,
      "learning_rate": 0.0007850295473070407,
      "loss": 1.2655,
      "step": 17854
    },
    {
      "epoch": 1.5361782672287705,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007850073301220479,
      "loss": 0.637,
      "step": 17855
    },
    {
      "epoch": 1.5362643035360923,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007849851121034639,
      "loss": 1.0874,
      "step": 17856
    },
    {
      "epoch": 1.536350339843414,
      "grad_norm": 2.046875,
      "learning_rate": 0.0007849628932513539,
      "loss": 0.7752,
      "step": 17857
    },
    {
      "epoch": 1.5364363761507356,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007849406735657829,
      "loss": 1.0284,
      "step": 17858
    },
    {
      "epoch": 1.5365224124580572,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007849184530468159,
      "loss": 1.1152,
      "step": 17859
    },
    {
      "epoch": 1.5366084487653788,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007848962316945178,
      "loss": 1.2034,
      "step": 17860
    },
    {
      "epoch": 1.5366944850727007,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007848740095089537,
      "loss": 0.6529,
      "step": 17861
    },
    {
      "epoch": 1.5367805213800225,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007848517864901886,
      "loss": 1.2619,
      "step": 17862
    },
    {
      "epoch": 1.5368665576873441,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007848295626382871,
      "loss": 1.588,
      "step": 17863
    },
    {
      "epoch": 1.5369525939946658,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007848073379533147,
      "loss": 1.2867,
      "step": 17864
    },
    {
      "epoch": 1.5370386303019874,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007847851124353362,
      "loss": 0.9037,
      "step": 17865
    },
    {
      "epoch": 1.537124666609309,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007847628860844167,
      "loss": 1.0776,
      "step": 17866
    },
    {
      "epoch": 1.5372107029166309,
      "grad_norm": 0.96875,
      "learning_rate": 0.000784740658900621,
      "loss": 1.1498,
      "step": 17867
    },
    {
      "epoch": 1.5372967392239525,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007847184308840146,
      "loss": 0.7935,
      "step": 17868
    },
    {
      "epoch": 1.5373827755312743,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007846962020346618,
      "loss": 1.5825,
      "step": 17869
    },
    {
      "epoch": 1.537468811838596,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007846739723526282,
      "loss": 1.2022,
      "step": 17870
    },
    {
      "epoch": 1.5375548481459176,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007846517418379783,
      "loss": 2.0167,
      "step": 17871
    },
    {
      "epoch": 1.5376408844532392,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007846295104907777,
      "loss": 1.3834,
      "step": 17872
    },
    {
      "epoch": 1.5377269207605608,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007846072783110912,
      "loss": 1.1806,
      "step": 17873
    },
    {
      "epoch": 1.5378129570678827,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007845850452989836,
      "loss": 1.5819,
      "step": 17874
    },
    {
      "epoch": 1.5378989933752043,
      "grad_norm": 1.9609375,
      "learning_rate": 0.00078456281145452,
      "loss": 1.8042,
      "step": 17875
    },
    {
      "epoch": 1.5379850296825262,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007845405767777657,
      "loss": 1.3053,
      "step": 17876
    },
    {
      "epoch": 1.5380710659898478,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007845183412687855,
      "loss": 1.0058,
      "step": 17877
    },
    {
      "epoch": 1.5381571022971694,
      "grad_norm": 2.0625,
      "learning_rate": 0.0007844961049276445,
      "loss": 1.5638,
      "step": 17878
    },
    {
      "epoch": 1.538243138604491,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007844738677544077,
      "loss": 0.8801,
      "step": 17879
    },
    {
      "epoch": 1.5383291749118126,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007844516297491402,
      "loss": 1.2454,
      "step": 17880
    },
    {
      "epoch": 1.5384152112191345,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007844293909119069,
      "loss": 1.5695,
      "step": 17881
    },
    {
      "epoch": 1.5385012475264561,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007844071512427731,
      "loss": 0.8944,
      "step": 17882
    },
    {
      "epoch": 1.538587283833778,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007843849107418035,
      "loss": 1.6038,
      "step": 17883
    },
    {
      "epoch": 1.5386733201410996,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007843626694090635,
      "loss": 1.0548,
      "step": 17884
    },
    {
      "epoch": 1.5387593564484212,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000784340427244618,
      "loss": 0.9628,
      "step": 17885
    },
    {
      "epoch": 1.5388453927557428,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000784318184248532,
      "loss": 0.7652,
      "step": 17886
    },
    {
      "epoch": 1.5389314290630645,
      "grad_norm": 1.375,
      "learning_rate": 0.0007842959404208706,
      "loss": 1.2466,
      "step": 17887
    },
    {
      "epoch": 1.5390174653703863,
      "grad_norm": 0.953125,
      "learning_rate": 0.000784273695761699,
      "loss": 0.7059,
      "step": 17888
    },
    {
      "epoch": 1.5391035016777082,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007842514502710819,
      "loss": 1.0415,
      "step": 17889
    },
    {
      "epoch": 1.5391895379850298,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007842292039490847,
      "loss": 0.7062,
      "step": 17890
    },
    {
      "epoch": 1.5392755742923514,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007842069567957724,
      "loss": 1.1553,
      "step": 17891
    },
    {
      "epoch": 1.539361610599673,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00078418470881121,
      "loss": 0.9043,
      "step": 17892
    },
    {
      "epoch": 1.5394476469069946,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007841624599954625,
      "loss": 1.533,
      "step": 17893
    },
    {
      "epoch": 1.5395336832143165,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007841402103485953,
      "loss": 1.2744,
      "step": 17894
    },
    {
      "epoch": 1.5396197195216381,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000784117959870673,
      "loss": 1.7188,
      "step": 17895
    },
    {
      "epoch": 1.53970575582896,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007840957085617611,
      "loss": 1.3217,
      "step": 17896
    },
    {
      "epoch": 1.5397917921362816,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007840734564219244,
      "loss": 0.9852,
      "step": 17897
    },
    {
      "epoch": 1.5398778284436032,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007840512034512281,
      "loss": 1.0098,
      "step": 17898
    },
    {
      "epoch": 1.5399638647509248,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007840289496497374,
      "loss": 0.5623,
      "step": 17899
    },
    {
      "epoch": 1.5400499010582465,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007840066950175171,
      "loss": 1.1635,
      "step": 17900
    },
    {
      "epoch": 1.5401359373655683,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007839844395546325,
      "loss": 0.9166,
      "step": 17901
    },
    {
      "epoch": 1.54022197367289,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007839621832611487,
      "loss": 1.0129,
      "step": 17902
    },
    {
      "epoch": 1.5403080099802118,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007839399261371307,
      "loss": 1.3838,
      "step": 17903
    },
    {
      "epoch": 1.5403940462875334,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007839176681826436,
      "loss": 1.5516,
      "step": 17904
    },
    {
      "epoch": 1.540480082594855,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007838954093977526,
      "loss": 1.4349,
      "step": 17905
    },
    {
      "epoch": 1.5405661189021767,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007838731497825227,
      "loss": 1.4096,
      "step": 17906
    },
    {
      "epoch": 1.5406521552094983,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007838508893370191,
      "loss": 0.8846,
      "step": 17907
    },
    {
      "epoch": 1.5407381915168201,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007838286280613068,
      "loss": 1.1506,
      "step": 17908
    },
    {
      "epoch": 1.5408242278241417,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007838063659554511,
      "loss": 1.929,
      "step": 17909
    },
    {
      "epoch": 1.5409102641314636,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007837841030195168,
      "loss": 1.0179,
      "step": 17910
    },
    {
      "epoch": 1.5409963004387852,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007837618392535692,
      "loss": 0.983,
      "step": 17911
    },
    {
      "epoch": 1.5410823367461068,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007837395746576736,
      "loss": 1.6812,
      "step": 17912
    },
    {
      "epoch": 1.5411683730534285,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007837173092318947,
      "loss": 1.9751,
      "step": 17913
    },
    {
      "epoch": 1.54125440936075,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000783695042976298,
      "loss": 2.2785,
      "step": 17914
    },
    {
      "epoch": 1.541340445668072,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007836727758909484,
      "loss": 0.8393,
      "step": 17915
    },
    {
      "epoch": 1.5414264819753938,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007836505079759112,
      "loss": 1.3827,
      "step": 17916
    },
    {
      "epoch": 1.5415125182827154,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007836282392312512,
      "loss": 1.0813,
      "step": 17917
    },
    {
      "epoch": 1.541598554590037,
      "grad_norm": 1.71875,
      "learning_rate": 0.0007836059696570341,
      "loss": 1.6758,
      "step": 17918
    },
    {
      "epoch": 1.5416845908973587,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007835836992533245,
      "loss": 2.0346,
      "step": 17919
    },
    {
      "epoch": 1.5417706272046803,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007835614280201877,
      "loss": 0.6248,
      "step": 17920
    },
    {
      "epoch": 1.541856663512002,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000783539155957689,
      "loss": 1.3061,
      "step": 17921
    },
    {
      "epoch": 1.5419426998193237,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007835168830658932,
      "loss": 0.7877,
      "step": 17922
    },
    {
      "epoch": 1.5420287361266456,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007834946093448658,
      "loss": 0.9006,
      "step": 17923
    },
    {
      "epoch": 1.5421147724339672,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007834723347946719,
      "loss": 1.6027,
      "step": 17924
    },
    {
      "epoch": 1.5422008087412888,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007834500594153764,
      "loss": 1.2961,
      "step": 17925
    },
    {
      "epoch": 1.5422868450486105,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007834277832070446,
      "loss": 1.0059,
      "step": 17926
    },
    {
      "epoch": 1.542372881355932,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0007834055061697418,
      "loss": 1.6504,
      "step": 17927
    },
    {
      "epoch": 1.542458917663254,
      "grad_norm": 2.0625,
      "learning_rate": 0.0007833832283035329,
      "loss": 1.5034,
      "step": 17928
    },
    {
      "epoch": 1.5425449539705756,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007833609496084833,
      "loss": 0.9795,
      "step": 17929
    },
    {
      "epoch": 1.5426309902778974,
      "grad_norm": 1.375,
      "learning_rate": 0.0007833386700846579,
      "loss": 1.896,
      "step": 17930
    },
    {
      "epoch": 1.542717026585219,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007833163897321221,
      "loss": 0.8549,
      "step": 17931
    },
    {
      "epoch": 1.5428030628925407,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007832941085509408,
      "loss": 1.3545,
      "step": 17932
    },
    {
      "epoch": 1.5428890991998623,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007832718265411795,
      "loss": 1.8175,
      "step": 17933
    },
    {
      "epoch": 1.542975135507184,
      "grad_norm": 1.28125,
      "learning_rate": 0.000783249543702903,
      "loss": 1.4211,
      "step": 17934
    },
    {
      "epoch": 1.5430611718145057,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000783227260036177,
      "loss": 1.0014,
      "step": 17935
    },
    {
      "epoch": 1.5431472081218274,
      "grad_norm": 1.0,
      "learning_rate": 0.0007832049755410659,
      "loss": 1.3349,
      "step": 17936
    },
    {
      "epoch": 1.5432332444291492,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007831826902176358,
      "loss": 1.4683,
      "step": 17937
    },
    {
      "epoch": 1.5433192807364708,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000783160404065951,
      "loss": 2.0911,
      "step": 17938
    },
    {
      "epoch": 1.5434053170437925,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007831381170860773,
      "loss": 0.9747,
      "step": 17939
    },
    {
      "epoch": 1.543491353351114,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007831158292780796,
      "loss": 1.6895,
      "step": 17940
    },
    {
      "epoch": 1.5435773896584357,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007830935406420232,
      "loss": 1.2516,
      "step": 17941
    },
    {
      "epoch": 1.5436634259657576,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007830712511779733,
      "loss": 1.1623,
      "step": 17942
    },
    {
      "epoch": 1.5437494622730792,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000783048960885995,
      "loss": 1.5203,
      "step": 17943
    },
    {
      "epoch": 1.543835498580401,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007830266697661536,
      "loss": 1.0487,
      "step": 17944
    },
    {
      "epoch": 1.5439215348877227,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007830043778185142,
      "loss": 0.6928,
      "step": 17945
    },
    {
      "epoch": 1.5440075711950443,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007829820850431419,
      "loss": 1.3849,
      "step": 17946
    },
    {
      "epoch": 1.544093607502366,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007829597914401022,
      "loss": 1.8218,
      "step": 17947
    },
    {
      "epoch": 1.5441796438096875,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007829374970094601,
      "loss": 1.8483,
      "step": 17948
    },
    {
      "epoch": 1.5442656801170094,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007829152017512809,
      "loss": 0.5277,
      "step": 17949
    },
    {
      "epoch": 1.5443517164243312,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007828929056656297,
      "loss": 1.1936,
      "step": 17950
    },
    {
      "epoch": 1.5444377527316528,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0007828706087525718,
      "loss": 1.3994,
      "step": 17951
    },
    {
      "epoch": 1.5445237890389745,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007828483110121723,
      "loss": 1.7191,
      "step": 17952
    },
    {
      "epoch": 1.544609825346296,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007828260124444966,
      "loss": 1.0652,
      "step": 17953
    },
    {
      "epoch": 1.5446958616536177,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007828037130496098,
      "loss": 1.811,
      "step": 17954
    },
    {
      "epoch": 1.5447818979609396,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007827814128275772,
      "loss": 0.7603,
      "step": 17955
    },
    {
      "epoch": 1.5448679342682612,
      "grad_norm": 1.125,
      "learning_rate": 0.0007827591117784638,
      "loss": 1.4083,
      "step": 17956
    },
    {
      "epoch": 1.544953970575583,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007827368099023351,
      "loss": 1.8884,
      "step": 17957
    },
    {
      "epoch": 1.5450400068829047,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007827145071992562,
      "loss": 1.3699,
      "step": 17958
    },
    {
      "epoch": 1.5451260431902263,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007826922036692922,
      "loss": 1.515,
      "step": 17959
    },
    {
      "epoch": 1.545212079497548,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007826698993125087,
      "loss": 1.1579,
      "step": 17960
    },
    {
      "epoch": 1.5452981158048695,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007826475941289708,
      "loss": 1.1202,
      "step": 17961
    },
    {
      "epoch": 1.5453841521121914,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0007826252881187435,
      "loss": 1.6395,
      "step": 17962
    },
    {
      "epoch": 1.545470188419513,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007826029812818922,
      "loss": 1.3049,
      "step": 17963
    },
    {
      "epoch": 1.5455562247268348,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007825806736184821,
      "loss": 1.489,
      "step": 17964
    },
    {
      "epoch": 1.5456422610341565,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007825583651285786,
      "loss": 2.0079,
      "step": 17965
    },
    {
      "epoch": 1.545728297341478,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007825360558122468,
      "loss": 0.8964,
      "step": 17966
    },
    {
      "epoch": 1.5458143336487997,
      "grad_norm": 1.125,
      "learning_rate": 0.0007825137456695519,
      "loss": 1.2073,
      "step": 17967
    },
    {
      "epoch": 1.5459003699561213,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007824914347005594,
      "loss": 1.1784,
      "step": 17968
    },
    {
      "epoch": 1.5459864062634432,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007824691229053342,
      "loss": 1.2005,
      "step": 17969
    },
    {
      "epoch": 1.5460724425707648,
      "grad_norm": 1.53125,
      "learning_rate": 0.000782446810283942,
      "loss": 1.1929,
      "step": 17970
    },
    {
      "epoch": 1.5461584788780867,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007824244968364475,
      "loss": 0.6319,
      "step": 17971
    },
    {
      "epoch": 1.5462445151854083,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007824021825629166,
      "loss": 1.3877,
      "step": 17972
    },
    {
      "epoch": 1.54633055149273,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007823798674634141,
      "loss": 1.0808,
      "step": 17973
    },
    {
      "epoch": 1.5464165878000515,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007823575515380054,
      "loss": 0.8758,
      "step": 17974
    },
    {
      "epoch": 1.5465026241073732,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007823352347867556,
      "loss": 0.6776,
      "step": 17975
    },
    {
      "epoch": 1.546588660414695,
      "grad_norm": 5.78125,
      "learning_rate": 0.0007823129172097304,
      "loss": 1.0268,
      "step": 17976
    },
    {
      "epoch": 1.5466746967220169,
      "grad_norm": 1.9375,
      "learning_rate": 0.0007822905988069948,
      "loss": 1.1838,
      "step": 17977
    },
    {
      "epoch": 1.5467607330293385,
      "grad_norm": 1.109375,
      "learning_rate": 0.000782268279578614,
      "loss": 1.2733,
      "step": 17978
    },
    {
      "epoch": 1.54684676933666,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007822459595246534,
      "loss": 1.3768,
      "step": 17979
    },
    {
      "epoch": 1.5469328056439817,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007822236386451783,
      "loss": 0.5827,
      "step": 17980
    },
    {
      "epoch": 1.5470188419513033,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007822013169402539,
      "loss": 1.7959,
      "step": 17981
    },
    {
      "epoch": 1.5471048782586252,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007821789944099457,
      "loss": 1.2868,
      "step": 17982
    },
    {
      "epoch": 1.5471909145659468,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007821566710543187,
      "loss": 0.7865,
      "step": 17983
    },
    {
      "epoch": 1.5472769508732687,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007821343468734384,
      "loss": 0.9128,
      "step": 17984
    },
    {
      "epoch": 1.5473629871805903,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007821120218673697,
      "loss": 1.6168,
      "step": 17985
    },
    {
      "epoch": 1.547449023487912,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007820896960361785,
      "loss": 1.0904,
      "step": 17986
    },
    {
      "epoch": 1.5475350597952335,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007820673693799298,
      "loss": 1.902,
      "step": 17987
    },
    {
      "epoch": 1.5476210961025552,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0007820450418986886,
      "loss": 1.9309,
      "step": 17988
    },
    {
      "epoch": 1.547707132409877,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007820227135925208,
      "loss": 1.305,
      "step": 17989
    },
    {
      "epoch": 1.5477931687171986,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007820003844614913,
      "loss": 1.046,
      "step": 17990
    },
    {
      "epoch": 1.5478792050245205,
      "grad_norm": 2.125,
      "learning_rate": 0.0007819780545056657,
      "loss": 1.2624,
      "step": 17991
    },
    {
      "epoch": 1.547965241331842,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007819557237251089,
      "loss": 1.1688,
      "step": 17992
    },
    {
      "epoch": 1.5480512776391637,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007819333921198866,
      "loss": 0.9076,
      "step": 17993
    },
    {
      "epoch": 1.5481373139464853,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007819110596900638,
      "loss": 1.458,
      "step": 17994
    },
    {
      "epoch": 1.548223350253807,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007818887264357062,
      "loss": 1.0176,
      "step": 17995
    },
    {
      "epoch": 1.5483093865611288,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007818663923568787,
      "loss": 0.5177,
      "step": 17996
    },
    {
      "epoch": 1.5483954228684504,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000781844057453647,
      "loss": 2.1153,
      "step": 17997
    },
    {
      "epoch": 1.5484814591757723,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007818217217260762,
      "loss": 1.2305,
      "step": 17998
    },
    {
      "epoch": 1.548567495483094,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007817993851742315,
      "loss": 0.8779,
      "step": 17999
    },
    {
      "epoch": 1.5486535317904155,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007817770477981786,
      "loss": 1.3227,
      "step": 18000
    },
    {
      "epoch": 1.5487395680977372,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007817547095979825,
      "loss": 1.6441,
      "step": 18001
    },
    {
      "epoch": 1.5488256044050588,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007817323705737087,
      "loss": 1.6184,
      "step": 18002
    },
    {
      "epoch": 1.5489116407123806,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007817100307254226,
      "loss": 1.3196,
      "step": 18003
    },
    {
      "epoch": 1.5489976770197025,
      "grad_norm": 2.140625,
      "learning_rate": 0.0007816876900531895,
      "loss": 2.2164,
      "step": 18004
    },
    {
      "epoch": 1.549083713327024,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007816653485570747,
      "loss": 1.1368,
      "step": 18005
    },
    {
      "epoch": 1.5491697496343457,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007816430062371433,
      "loss": 1.282,
      "step": 18006
    },
    {
      "epoch": 1.5492557859416674,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000781620663093461,
      "loss": 1.0402,
      "step": 18007
    },
    {
      "epoch": 1.549341822248989,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007815983191260932,
      "loss": 1.2612,
      "step": 18008
    },
    {
      "epoch": 1.5494278585563106,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007815759743351047,
      "loss": 1.9072,
      "step": 18009
    },
    {
      "epoch": 1.5495138948636324,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0007815536287205616,
      "loss": 1.1806,
      "step": 18010
    },
    {
      "epoch": 1.5495999311709543,
      "grad_norm": 6.53125,
      "learning_rate": 0.0007815312822825286,
      "loss": 1.2437,
      "step": 18011
    },
    {
      "epoch": 1.549685967478276,
      "grad_norm": 1.125,
      "learning_rate": 0.0007815089350210715,
      "loss": 1.1503,
      "step": 18012
    },
    {
      "epoch": 1.5497720037855975,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007814865869362556,
      "loss": 0.967,
      "step": 18013
    },
    {
      "epoch": 1.5498580400929192,
      "grad_norm": 9.0625,
      "learning_rate": 0.0007814642380281462,
      "loss": 1.8278,
      "step": 18014
    },
    {
      "epoch": 1.5499440764002408,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007814418882968085,
      "loss": 0.9304,
      "step": 18015
    },
    {
      "epoch": 1.5500301127075626,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007814195377423081,
      "loss": 1.176,
      "step": 18016
    },
    {
      "epoch": 1.5501161490148843,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007813971863647102,
      "loss": 1.3128,
      "step": 18017
    },
    {
      "epoch": 1.550202185322206,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007813748341640804,
      "loss": 2.1247,
      "step": 18018
    },
    {
      "epoch": 1.5502882216295277,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007813524811404838,
      "loss": 1.354,
      "step": 18019
    },
    {
      "epoch": 1.5503742579368494,
      "grad_norm": 1.90625,
      "learning_rate": 0.000781330127293986,
      "loss": 1.2752,
      "step": 18020
    },
    {
      "epoch": 1.550460294244171,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0007813077726246523,
      "loss": 0.9241,
      "step": 18021
    },
    {
      "epoch": 1.5505463305514926,
      "grad_norm": 2.046875,
      "learning_rate": 0.000781285417132548,
      "loss": 1.0841,
      "step": 18022
    },
    {
      "epoch": 1.5506323668588144,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007812630608177387,
      "loss": 1.0037,
      "step": 18023
    },
    {
      "epoch": 1.550718403166136,
      "grad_norm": 6.84375,
      "learning_rate": 0.0007812407036802894,
      "loss": 1.3999,
      "step": 18024
    },
    {
      "epoch": 1.550804439473458,
      "grad_norm": 6.09375,
      "learning_rate": 0.000781218345720266,
      "loss": 1.0451,
      "step": 18025
    },
    {
      "epoch": 1.5508904757807795,
      "grad_norm": 9.9375,
      "learning_rate": 0.0007811959869377334,
      "loss": 0.8711,
      "step": 18026
    },
    {
      "epoch": 1.5509765120881012,
      "grad_norm": 91.5,
      "learning_rate": 0.0007811736273327574,
      "loss": 1.6554,
      "step": 18027
    },
    {
      "epoch": 1.5510625483954228,
      "grad_norm": 268.0,
      "learning_rate": 0.0007811512669054033,
      "loss": 1.6875,
      "step": 18028
    },
    {
      "epoch": 1.5511485847027444,
      "grad_norm": 7200.0,
      "learning_rate": 0.0007811289056557362,
      "loss": 1.6501,
      "step": 18029
    },
    {
      "epoch": 1.5512346210100663,
      "grad_norm": 6.8125,
      "learning_rate": 0.0007811065435838218,
      "loss": 1.3177,
      "step": 18030
    },
    {
      "epoch": 1.5513206573173879,
      "grad_norm": 5.5625,
      "learning_rate": 0.0007810841806897255,
      "loss": 1.8344,
      "step": 18031
    },
    {
      "epoch": 1.5514066936247097,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007810618169735125,
      "loss": 1.0975,
      "step": 18032
    },
    {
      "epoch": 1.5514927299320314,
      "grad_norm": 77.0,
      "learning_rate": 0.0007810394524352484,
      "loss": 0.661,
      "step": 18033
    },
    {
      "epoch": 1.551578766239353,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007810170870749986,
      "loss": 1.3533,
      "step": 18034
    },
    {
      "epoch": 1.5516648025466746,
      "grad_norm": 2.140625,
      "learning_rate": 0.0007809947208928285,
      "loss": 1.9034,
      "step": 18035
    },
    {
      "epoch": 1.5517508388539962,
      "grad_norm": 11.0625,
      "learning_rate": 0.0007809723538888036,
      "loss": 1.2446,
      "step": 18036
    },
    {
      "epoch": 1.551836875161318,
      "grad_norm": 3.484375,
      "learning_rate": 0.0007809499860629889,
      "loss": 1.2921,
      "step": 18037
    },
    {
      "epoch": 1.55192291146864,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007809276174154505,
      "loss": 1.068,
      "step": 18038
    },
    {
      "epoch": 1.5520089477759615,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007809052479462533,
      "loss": 1.9723,
      "step": 18039
    },
    {
      "epoch": 1.5520949840832832,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007808828776554628,
      "loss": 1.0482,
      "step": 18040
    },
    {
      "epoch": 1.5521810203906048,
      "grad_norm": 2.71875,
      "learning_rate": 0.0007808605065431445,
      "loss": 1.3075,
      "step": 18041
    },
    {
      "epoch": 1.5522670566979264,
      "grad_norm": 8.6875,
      "learning_rate": 0.0007808381346093641,
      "loss": 1.5441,
      "step": 18042
    },
    {
      "epoch": 1.5523530930052483,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007808157618541864,
      "loss": 1.0923,
      "step": 18043
    },
    {
      "epoch": 1.5524391293125699,
      "grad_norm": 2.859375,
      "learning_rate": 0.0007807933882776775,
      "loss": 2.1001,
      "step": 18044
    },
    {
      "epoch": 1.5525251656198917,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007807710138799024,
      "loss": 1.1246,
      "step": 18045
    },
    {
      "epoch": 1.5526112019272134,
      "grad_norm": 64.0,
      "learning_rate": 0.0007807486386609267,
      "loss": 1.6248,
      "step": 18046
    },
    {
      "epoch": 1.552697238234535,
      "grad_norm": 29.875,
      "learning_rate": 0.000780726262620816,
      "loss": 1.4346,
      "step": 18047
    },
    {
      "epoch": 1.5527832745418566,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0007807038857596353,
      "loss": 1.3368,
      "step": 18048
    },
    {
      "epoch": 1.5528693108491782,
      "grad_norm": 1.25,
      "learning_rate": 0.0007806815080774505,
      "loss": 1.6084,
      "step": 18049
    },
    {
      "epoch": 1.5529553471565,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007806591295743268,
      "loss": 0.8673,
      "step": 18050
    },
    {
      "epoch": 1.5530413834638217,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007806367502503299,
      "loss": 1.5897,
      "step": 18051
    },
    {
      "epoch": 1.5531274197711435,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007806143701055249,
      "loss": 0.83,
      "step": 18052
    },
    {
      "epoch": 1.5532134560784652,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007805919891399776,
      "loss": 0.9149,
      "step": 18053
    },
    {
      "epoch": 1.5532994923857868,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007805696073537532,
      "loss": 1.4353,
      "step": 18054
    },
    {
      "epoch": 1.5533855286931084,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0007805472247469173,
      "loss": 1.9468,
      "step": 18055
    },
    {
      "epoch": 1.55347156500043,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007805248413195354,
      "loss": 0.8414,
      "step": 18056
    },
    {
      "epoch": 1.553557601307752,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007805024570716728,
      "loss": 1.1179,
      "step": 18057
    },
    {
      "epoch": 1.5536436376150735,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007804800720033951,
      "loss": 1.1068,
      "step": 18058
    },
    {
      "epoch": 1.5537296739223954,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007804576861147677,
      "loss": 1.7909,
      "step": 18059
    },
    {
      "epoch": 1.553815710229717,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007804352994058561,
      "loss": 1.3732,
      "step": 18060
    },
    {
      "epoch": 1.5539017465370386,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007804129118767258,
      "loss": 1.407,
      "step": 18061
    },
    {
      "epoch": 1.5539877828443602,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007803905235274425,
      "loss": 1.8579,
      "step": 18062
    },
    {
      "epoch": 1.5540738191516819,
      "grad_norm": 0.9375,
      "learning_rate": 0.000780368134358071,
      "loss": 1.2707,
      "step": 18063
    },
    {
      "epoch": 1.5541598554590037,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007803457443686777,
      "loss": 0.9875,
      "step": 18064
    },
    {
      "epoch": 1.5542458917663255,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007803233535593273,
      "loss": 1.2486,
      "step": 18065
    },
    {
      "epoch": 1.5543319280736472,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007803009619300859,
      "loss": 0.8036,
      "step": 18066
    },
    {
      "epoch": 1.5544179643809688,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007802785694810184,
      "loss": 1.3246,
      "step": 18067
    },
    {
      "epoch": 1.5545040006882904,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007802561762121907,
      "loss": 1.2743,
      "step": 18068
    },
    {
      "epoch": 1.554590036995612,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007802337821236682,
      "loss": 1.2893,
      "step": 18069
    },
    {
      "epoch": 1.554676073302934,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007802113872155163,
      "loss": 1.4667,
      "step": 18070
    },
    {
      "epoch": 1.5547621096102555,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007801889914878007,
      "loss": 1.3857,
      "step": 18071
    },
    {
      "epoch": 1.5548481459175774,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007801665949405869,
      "loss": 1.4338,
      "step": 18072
    },
    {
      "epoch": 1.554934182224899,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007801441975739401,
      "loss": 1.2541,
      "step": 18073
    },
    {
      "epoch": 1.5550202185322206,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007801217993879261,
      "loss": 1.6818,
      "step": 18074
    },
    {
      "epoch": 1.5551062548395422,
      "grad_norm": 1.125,
      "learning_rate": 0.0007800994003826104,
      "loss": 0.6519,
      "step": 18075
    },
    {
      "epoch": 1.5551922911468639,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007800770005580583,
      "loss": 1.3579,
      "step": 18076
    },
    {
      "epoch": 1.5552783274541857,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007800545999143356,
      "loss": 1.7791,
      "step": 18077
    },
    {
      "epoch": 1.5553643637615073,
      "grad_norm": 2.296875,
      "learning_rate": 0.0007800321984515076,
      "loss": 1.3981,
      "step": 18078
    },
    {
      "epoch": 1.5554504000688292,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007800097961696398,
      "loss": 0.857,
      "step": 18079
    },
    {
      "epoch": 1.5555364363761508,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007799873930687978,
      "loss": 1.3688,
      "step": 18080
    },
    {
      "epoch": 1.5556224726834724,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007799649891490471,
      "loss": 2.8011,
      "step": 18081
    },
    {
      "epoch": 1.555708508990794,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007799425844104533,
      "loss": 1.7282,
      "step": 18082
    },
    {
      "epoch": 1.5557945452981157,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000779920178853082,
      "loss": 1.7903,
      "step": 18083
    },
    {
      "epoch": 1.5558805816054375,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007798977724769985,
      "loss": 1.3828,
      "step": 18084
    },
    {
      "epoch": 1.5559666179127591,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0007798753652822685,
      "loss": 2.0682,
      "step": 18085
    },
    {
      "epoch": 1.556052654220081,
      "grad_norm": 2.625,
      "learning_rate": 0.0007798529572689574,
      "loss": 1.1701,
      "step": 18086
    },
    {
      "epoch": 1.5561386905274026,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007798305484371309,
      "loss": 1.7511,
      "step": 18087
    },
    {
      "epoch": 1.5562247268347242,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007798081387868544,
      "loss": 1.5149,
      "step": 18088
    },
    {
      "epoch": 1.5563107631420459,
      "grad_norm": 1.8125,
      "learning_rate": 0.0007797857283181937,
      "loss": 1.5613,
      "step": 18089
    },
    {
      "epoch": 1.5563967994493675,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007797633170312139,
      "loss": 0.8008,
      "step": 18090
    },
    {
      "epoch": 1.5564828357566893,
      "grad_norm": 1.046875,
      "learning_rate": 0.000779740904925981,
      "loss": 0.8044,
      "step": 18091
    },
    {
      "epoch": 1.5565688720640112,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007797184920025604,
      "loss": 1.1758,
      "step": 18092
    },
    {
      "epoch": 1.5566549083713328,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0007796960782610174,
      "loss": 0.9461,
      "step": 18093
    },
    {
      "epoch": 1.5567409446786544,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007796736637014179,
      "loss": 1.892,
      "step": 18094
    },
    {
      "epoch": 1.556826980985976,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007796512483238273,
      "loss": 0.9915,
      "step": 18095
    },
    {
      "epoch": 1.5569130172932977,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007796288321283113,
      "loss": 1.1301,
      "step": 18096
    },
    {
      "epoch": 1.5569990536006193,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007796064151149352,
      "loss": 0.8165,
      "step": 18097
    },
    {
      "epoch": 1.5570850899079411,
      "grad_norm": 1.78125,
      "learning_rate": 0.0007795839972837648,
      "loss": 1.3868,
      "step": 18098
    },
    {
      "epoch": 1.557171126215263,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007795615786348655,
      "loss": 0.8249,
      "step": 18099
    },
    {
      "epoch": 1.5572571625225846,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007795391591683031,
      "loss": 1.3002,
      "step": 18100
    },
    {
      "epoch": 1.5573431988299062,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000779516738884143,
      "loss": 0.9862,
      "step": 18101
    },
    {
      "epoch": 1.5574292351372279,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007794943177824507,
      "loss": 1.0587,
      "step": 18102
    },
    {
      "epoch": 1.5575152714445495,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007794718958632921,
      "loss": 1.4954,
      "step": 18103
    },
    {
      "epoch": 1.5576013077518713,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007794494731267322,
      "loss": 1.3656,
      "step": 18104
    },
    {
      "epoch": 1.557687344059193,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000779427049572837,
      "loss": 0.6629,
      "step": 18105
    },
    {
      "epoch": 1.5577733803665148,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007794046252016723,
      "loss": 1.0159,
      "step": 18106
    },
    {
      "epoch": 1.5578594166738364,
      "grad_norm": 1.640625,
      "learning_rate": 0.0007793822000133031,
      "loss": 2.0078,
      "step": 18107
    },
    {
      "epoch": 1.557945452981158,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007793597740077955,
      "loss": 1.1649,
      "step": 18108
    },
    {
      "epoch": 1.5580314892884797,
      "grad_norm": 1.0625,
      "learning_rate": 0.000779337347185215,
      "loss": 1.6026,
      "step": 18109
    },
    {
      "epoch": 1.5581175255958013,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007793149195456268,
      "loss": 1.9715,
      "step": 18110
    },
    {
      "epoch": 1.5582035619031231,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000779292491089097,
      "loss": 1.252,
      "step": 18111
    },
    {
      "epoch": 1.5582895982104448,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007792700618156909,
      "loss": 1.3644,
      "step": 18112
    },
    {
      "epoch": 1.5583756345177666,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007792476317254742,
      "loss": 1.097,
      "step": 18113
    },
    {
      "epoch": 1.5584616708250882,
      "grad_norm": 8.6875,
      "learning_rate": 0.0007792252008185126,
      "loss": 1.3222,
      "step": 18114
    },
    {
      "epoch": 1.5585477071324099,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007792027690948713,
      "loss": 1.345,
      "step": 18115
    },
    {
      "epoch": 1.5586337434397315,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007791803365546164,
      "loss": 1.179,
      "step": 18116
    },
    {
      "epoch": 1.5587197797470531,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007791579031978134,
      "loss": 1.4082,
      "step": 18117
    },
    {
      "epoch": 1.558805816054375,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007791354690245276,
      "loss": 1.7202,
      "step": 18118
    },
    {
      "epoch": 1.5588918523616966,
      "grad_norm": 1.03125,
      "learning_rate": 0.000779113034034825,
      "loss": 1.2387,
      "step": 18119
    },
    {
      "epoch": 1.5589778886690184,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007790905982287711,
      "loss": 1.6194,
      "step": 18120
    },
    {
      "epoch": 1.55906392497634,
      "grad_norm": 1.625,
      "learning_rate": 0.0007790681616064313,
      "loss": 1.2491,
      "step": 18121
    },
    {
      "epoch": 1.5591499612836617,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007790457241678715,
      "loss": 1.5541,
      "step": 18122
    },
    {
      "epoch": 1.5592359975909833,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007790232859131572,
      "loss": 1.3064,
      "step": 18123
    },
    {
      "epoch": 1.559322033898305,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0007790008468423541,
      "loss": 0.7014,
      "step": 18124
    },
    {
      "epoch": 1.5594080702056268,
      "grad_norm": 2.171875,
      "learning_rate": 0.0007789784069555277,
      "loss": 1.6455,
      "step": 18125
    },
    {
      "epoch": 1.5594941065129486,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007789559662527438,
      "loss": 1.1101,
      "step": 18126
    },
    {
      "epoch": 1.5595801428202702,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007789335247340679,
      "loss": 0.8856,
      "step": 18127
    },
    {
      "epoch": 1.5596661791275919,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007789110823995659,
      "loss": 1.4605,
      "step": 18128
    },
    {
      "epoch": 1.5597522154349135,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007788886392493031,
      "loss": 1.3005,
      "step": 18129
    },
    {
      "epoch": 1.5598382517422351,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007788661952833452,
      "loss": 1.2742,
      "step": 18130
    },
    {
      "epoch": 1.559924288049557,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007788437505017578,
      "loss": 2.0101,
      "step": 18131
    },
    {
      "epoch": 1.5600103243568786,
      "grad_norm": 1.03125,
      "learning_rate": 0.000778821304904607,
      "loss": 1.4346,
      "step": 18132
    },
    {
      "epoch": 1.5600963606642004,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007787988584919579,
      "loss": 1.2816,
      "step": 18133
    },
    {
      "epoch": 1.560182396971522,
      "grad_norm": 13.5625,
      "learning_rate": 0.0007787764112638764,
      "loss": 1.6525,
      "step": 18134
    },
    {
      "epoch": 1.5602684332788437,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000778753963220428,
      "loss": 1.7391,
      "step": 18135
    },
    {
      "epoch": 1.5603544695861653,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007787315143616786,
      "loss": 1.0926,
      "step": 18136
    },
    {
      "epoch": 1.560440505893487,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0007787090646876937,
      "loss": 0.9365,
      "step": 18137
    },
    {
      "epoch": 1.5605265422008088,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007786866141985389,
      "loss": 1.0165,
      "step": 18138
    },
    {
      "epoch": 1.5606125785081304,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007786641628942801,
      "loss": 1.3819,
      "step": 18139
    },
    {
      "epoch": 1.5606986148154522,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007786417107749828,
      "loss": 0.9351,
      "step": 18140
    },
    {
      "epoch": 1.5607846511227739,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007786192578407127,
      "loss": 1.2187,
      "step": 18141
    },
    {
      "epoch": 1.5608706874300955,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007785968040915353,
      "loss": 0.8751,
      "step": 18142
    },
    {
      "epoch": 1.5609567237374171,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007785743495275166,
      "loss": 0.64,
      "step": 18143
    },
    {
      "epoch": 1.5610427600447387,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007785518941487219,
      "loss": 0.8639,
      "step": 18144
    },
    {
      "epoch": 1.5611287963520606,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007785294379552175,
      "loss": 0.8581,
      "step": 18145
    },
    {
      "epoch": 1.5612148326593822,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007785069809470682,
      "loss": 1.1677,
      "step": 18146
    },
    {
      "epoch": 1.561300868966704,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007784845231243405,
      "loss": 1.4201,
      "step": 18147
    },
    {
      "epoch": 1.5613869052740257,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007784620644870995,
      "loss": 1.3575,
      "step": 18148
    },
    {
      "epoch": 1.5614729415813473,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007784396050354112,
      "loss": 0.7402,
      "step": 18149
    },
    {
      "epoch": 1.561558977888669,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007784171447693414,
      "loss": 1.4908,
      "step": 18150
    },
    {
      "epoch": 1.5616450141959906,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007783946836889552,
      "loss": 2.0546,
      "step": 18151
    },
    {
      "epoch": 1.5617310505033124,
      "grad_norm": 1.1875,
      "learning_rate": 0.000778372221794319,
      "loss": 0.6092,
      "step": 18152
    },
    {
      "epoch": 1.5618170868106342,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007783497590854981,
      "loss": 1.9751,
      "step": 18153
    },
    {
      "epoch": 1.5619031231179559,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007783272955625584,
      "loss": 1.2929,
      "step": 18154
    },
    {
      "epoch": 1.5619891594252775,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007783048312255653,
      "loss": 1.4494,
      "step": 18155
    },
    {
      "epoch": 1.5620751957325991,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007782823660745847,
      "loss": 2.102,
      "step": 18156
    },
    {
      "epoch": 1.5621612320399207,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007782599001096824,
      "loss": 0.9289,
      "step": 18157
    },
    {
      "epoch": 1.5622472683472426,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007782374333309241,
      "loss": 0.8394,
      "step": 18158
    },
    {
      "epoch": 1.5623333046545642,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007782149657383753,
      "loss": 0.7296,
      "step": 18159
    },
    {
      "epoch": 1.562419340961886,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007781924973321017,
      "loss": 1.3951,
      "step": 18160
    },
    {
      "epoch": 1.5625053772692077,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007781700281121693,
      "loss": 1.7826,
      "step": 18161
    },
    {
      "epoch": 1.5625914135765293,
      "grad_norm": 1.0,
      "learning_rate": 0.0007781475580786437,
      "loss": 1.1803,
      "step": 18162
    },
    {
      "epoch": 1.562677449883851,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007781250872315906,
      "loss": 1.0809,
      "step": 18163
    },
    {
      "epoch": 1.5627634861911726,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007781026155710757,
      "loss": 1.4096,
      "step": 18164
    },
    {
      "epoch": 1.5628495224984944,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007780801430971647,
      "loss": 1.6487,
      "step": 18165
    },
    {
      "epoch": 1.562935558805816,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007780576698099233,
      "loss": 1.7699,
      "step": 18166
    },
    {
      "epoch": 1.5630215951131379,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007780351957094174,
      "loss": 1.6954,
      "step": 18167
    },
    {
      "epoch": 1.5631076314204595,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007780127207957126,
      "loss": 1.4083,
      "step": 18168
    },
    {
      "epoch": 1.5631936677277811,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007779902450688747,
      "loss": 1.3849,
      "step": 18169
    },
    {
      "epoch": 1.5632797040351027,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007779677685289694,
      "loss": 0.909,
      "step": 18170
    },
    {
      "epoch": 1.5633657403424244,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007779452911760624,
      "loss": 0.857,
      "step": 18171
    },
    {
      "epoch": 1.5634517766497462,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007779228130102194,
      "loss": 0.7539,
      "step": 18172
    },
    {
      "epoch": 1.5635378129570678,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007779003340315063,
      "loss": 0.9231,
      "step": 18173
    },
    {
      "epoch": 1.5636238492643897,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007778778542399888,
      "loss": 1.3445,
      "step": 18174
    },
    {
      "epoch": 1.5637098855717113,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007778553736357325,
      "loss": 1.8523,
      "step": 18175
    },
    {
      "epoch": 1.563795921879033,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007778328922188035,
      "loss": 1.4017,
      "step": 18176
    },
    {
      "epoch": 1.5638819581863546,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007778104099892671,
      "loss": 0.8652,
      "step": 18177
    },
    {
      "epoch": 1.5639679944936762,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007777879269471895,
      "loss": 0.887,
      "step": 18178
    },
    {
      "epoch": 1.564054030800998,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000777765443092636,
      "loss": 1.5524,
      "step": 18179
    },
    {
      "epoch": 1.5641400671083199,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007777429584256728,
      "loss": 1.7597,
      "step": 18180
    },
    {
      "epoch": 1.5642261034156415,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007777204729463653,
      "loss": 1.1048,
      "step": 18181
    },
    {
      "epoch": 1.5643121397229631,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007776979866547795,
      "loss": 0.993,
      "step": 18182
    },
    {
      "epoch": 1.5643981760302847,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007776754995509811,
      "loss": 1.3366,
      "step": 18183
    },
    {
      "epoch": 1.5644842123376064,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007776530116350359,
      "loss": 1.1404,
      "step": 18184
    },
    {
      "epoch": 1.564570248644928,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007776305229070095,
      "loss": 1.1428,
      "step": 18185
    },
    {
      "epoch": 1.5646562849522498,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007776080333669681,
      "loss": 1.1893,
      "step": 18186
    },
    {
      "epoch": 1.5647423212595717,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000777585543014977,
      "loss": 1.4617,
      "step": 18187
    },
    {
      "epoch": 1.5648283575668933,
      "grad_norm": 1.640625,
      "learning_rate": 0.0007775630518511021,
      "loss": 1.3699,
      "step": 18188
    },
    {
      "epoch": 1.564914393874215,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007775405598754096,
      "loss": 1.1839,
      "step": 18189
    },
    {
      "epoch": 1.5650004301815366,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007775180670879645,
      "loss": 1.3308,
      "step": 18190
    },
    {
      "epoch": 1.5650864664888582,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007774955734888334,
      "loss": 1.5664,
      "step": 18191
    },
    {
      "epoch": 1.56517250279618,
      "grad_norm": 1.0,
      "learning_rate": 0.0007774730790780814,
      "loss": 1.0895,
      "step": 18192
    },
    {
      "epoch": 1.5652585391035017,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007774505838557749,
      "loss": 1.22,
      "step": 18193
    },
    {
      "epoch": 1.5653445754108235,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007774280878219791,
      "loss": 0.936,
      "step": 18194
    },
    {
      "epoch": 1.5654306117181451,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007774055909767604,
      "loss": 1.6493,
      "step": 18195
    },
    {
      "epoch": 1.5655166480254668,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007773830933201841,
      "loss": 1.5432,
      "step": 18196
    },
    {
      "epoch": 1.5656026843327884,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007773605948523161,
      "loss": 1.2608,
      "step": 18197
    },
    {
      "epoch": 1.56568872064011,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007773380955732225,
      "loss": 1.2571,
      "step": 18198
    },
    {
      "epoch": 1.5657747569474318,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007773155954829689,
      "loss": 0.9103,
      "step": 18199
    },
    {
      "epoch": 1.5658607932547535,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007772930945816212,
      "loss": 1.1977,
      "step": 18200
    },
    {
      "epoch": 1.5659468295620753,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000777270592869245,
      "loss": 1.2381,
      "step": 18201
    },
    {
      "epoch": 1.566032865869397,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007772480903459062,
      "loss": 1.44,
      "step": 18202
    },
    {
      "epoch": 1.5661189021767186,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007772255870116707,
      "loss": 1.3813,
      "step": 18203
    },
    {
      "epoch": 1.5662049384840402,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007772030828666044,
      "loss": 0.9612,
      "step": 18204
    },
    {
      "epoch": 1.5662909747913618,
      "grad_norm": 0.875,
      "learning_rate": 0.0007771805779107728,
      "loss": 1.0061,
      "step": 18205
    },
    {
      "epoch": 1.5663770110986837,
      "grad_norm": 1.34375,
      "learning_rate": 0.000777158072144242,
      "loss": 2.0774,
      "step": 18206
    },
    {
      "epoch": 1.5664630474060053,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007771355655670777,
      "loss": 1.1008,
      "step": 18207
    },
    {
      "epoch": 1.5665490837133271,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000777113058179346,
      "loss": 1.1684,
      "step": 18208
    },
    {
      "epoch": 1.5666351200206488,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007770905499811123,
      "loss": 1.0139,
      "step": 18209
    },
    {
      "epoch": 1.5667211563279704,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007770680409724427,
      "loss": 1.4375,
      "step": 18210
    },
    {
      "epoch": 1.566807192635292,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007770455311534028,
      "loss": 1.5839,
      "step": 18211
    },
    {
      "epoch": 1.5668932289426136,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007770230205240587,
      "loss": 0.923,
      "step": 18212
    },
    {
      "epoch": 1.5669792652499355,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007770005090844762,
      "loss": 1.1303,
      "step": 18213
    },
    {
      "epoch": 1.5670653015572573,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007769779968347211,
      "loss": 1.0629,
      "step": 18214
    },
    {
      "epoch": 1.567151337864579,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007769554837748591,
      "loss": 1.6251,
      "step": 18215
    },
    {
      "epoch": 1.5672373741719006,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007769329699049562,
      "loss": 0.8287,
      "step": 18216
    },
    {
      "epoch": 1.5673234104792222,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007769104552250783,
      "loss": 1.0072,
      "step": 18217
    },
    {
      "epoch": 1.5674094467865438,
      "grad_norm": 1.125,
      "learning_rate": 0.0007768879397352912,
      "loss": 1.011,
      "step": 18218
    },
    {
      "epoch": 1.5674954830938657,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007768654234356607,
      "loss": 1.8843,
      "step": 18219
    },
    {
      "epoch": 1.5675815194011873,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007768429063262526,
      "loss": 0.9309,
      "step": 18220
    },
    {
      "epoch": 1.5676675557085091,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000776820388407133,
      "loss": 1.4304,
      "step": 18221
    },
    {
      "epoch": 1.5677535920158308,
      "grad_norm": 1.125,
      "learning_rate": 0.0007767978696783674,
      "loss": 1.917,
      "step": 18222
    },
    {
      "epoch": 1.5678396283231524,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000776775350140022,
      "loss": 1.4448,
      "step": 18223
    },
    {
      "epoch": 1.567925664630474,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007767528297921624,
      "loss": 0.7276,
      "step": 18224
    },
    {
      "epoch": 1.5680117009377956,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007767303086348549,
      "loss": 1.1131,
      "step": 18225
    },
    {
      "epoch": 1.5680977372451175,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007767077866681648,
      "loss": 1.7785,
      "step": 18226
    },
    {
      "epoch": 1.568183773552439,
      "grad_norm": 1.25,
      "learning_rate": 0.0007766852638921583,
      "loss": 1.0747,
      "step": 18227
    },
    {
      "epoch": 1.568269809859761,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007766627403069013,
      "loss": 1.0627,
      "step": 18228
    },
    {
      "epoch": 1.5683558461670826,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007766402159124595,
      "loss": 2.0571,
      "step": 18229
    },
    {
      "epoch": 1.5684418824744042,
      "grad_norm": 1.125,
      "learning_rate": 0.0007766176907088989,
      "loss": 1.7574,
      "step": 18230
    },
    {
      "epoch": 1.5685279187817258,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007765951646962852,
      "loss": 1.882,
      "step": 18231
    },
    {
      "epoch": 1.5686139550890474,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007765726378746847,
      "loss": 1.8081,
      "step": 18232
    },
    {
      "epoch": 1.5686999913963693,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007765501102441628,
      "loss": 0.9677,
      "step": 18233
    },
    {
      "epoch": 1.568786027703691,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007765275818047858,
      "loss": 1.5939,
      "step": 18234
    },
    {
      "epoch": 1.5688720640110128,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007765050525566192,
      "loss": 2.1266,
      "step": 18235
    },
    {
      "epoch": 1.5689581003183344,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007764825224997294,
      "loss": 0.9924,
      "step": 18236
    },
    {
      "epoch": 1.569044136625656,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007764599916341817,
      "loss": 0.9812,
      "step": 18237
    },
    {
      "epoch": 1.5691301729329776,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007764374599600424,
      "loss": 1.3594,
      "step": 18238
    },
    {
      "epoch": 1.5692162092402993,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007764149274773772,
      "loss": 1.2773,
      "step": 18239
    },
    {
      "epoch": 1.569302245547621,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007763923941862524,
      "loss": 1.1254,
      "step": 18240
    },
    {
      "epoch": 1.569388281854943,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007763698600867333,
      "loss": 1.4652,
      "step": 18241
    },
    {
      "epoch": 1.5694743181622646,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007763473251788864,
      "loss": 1.5407,
      "step": 18242
    },
    {
      "epoch": 1.5695603544695862,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007763247894627771,
      "loss": 1.9774,
      "step": 18243
    },
    {
      "epoch": 1.5696463907769078,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007763022529384716,
      "loss": 1.6363,
      "step": 18244
    },
    {
      "epoch": 1.5697324270842294,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007762797156060357,
      "loss": 0.9141,
      "step": 18245
    },
    {
      "epoch": 1.5698184633915513,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007762571774655354,
      "loss": 1.4977,
      "step": 18246
    },
    {
      "epoch": 1.569904499698873,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007762346385170364,
      "loss": 1.4851,
      "step": 18247
    },
    {
      "epoch": 1.5699905360061948,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000776212098760605,
      "loss": 0.7555,
      "step": 18248
    },
    {
      "epoch": 1.5700765723135164,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000776189558196307,
      "loss": 1.4413,
      "step": 18249
    },
    {
      "epoch": 1.570162608620838,
      "grad_norm": 1.375,
      "learning_rate": 0.0007761670168242081,
      "loss": 1.7411,
      "step": 18250
    },
    {
      "epoch": 1.5702486449281596,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007761444746443746,
      "loss": 1.4539,
      "step": 18251
    },
    {
      "epoch": 1.5703346812354813,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007761219316568719,
      "loss": 1.1852,
      "step": 18252
    },
    {
      "epoch": 1.570420717542803,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007760993878617664,
      "loss": 1.6256,
      "step": 18253
    },
    {
      "epoch": 1.5705067538501247,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007760768432591239,
      "loss": 0.4991,
      "step": 18254
    },
    {
      "epoch": 1.5705927901574466,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007760542978490104,
      "loss": 0.9795,
      "step": 18255
    },
    {
      "epoch": 1.5706788264647682,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007760317516314917,
      "loss": 1.4452,
      "step": 18256
    },
    {
      "epoch": 1.5707648627720898,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007760092046066339,
      "loss": 1.5112,
      "step": 18257
    },
    {
      "epoch": 1.5708508990794114,
      "grad_norm": 1.625,
      "learning_rate": 0.0007759866567745026,
      "loss": 0.9682,
      "step": 18258
    },
    {
      "epoch": 1.570936935386733,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007759641081351643,
      "loss": 1.0861,
      "step": 18259
    },
    {
      "epoch": 1.571022971694055,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007759415586886846,
      "loss": 0.8717,
      "step": 18260
    },
    {
      "epoch": 1.5711090080013765,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007759190084351294,
      "loss": 1.6934,
      "step": 18261
    },
    {
      "epoch": 1.5711950443086984,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007758964573745649,
      "loss": 1.0726,
      "step": 18262
    },
    {
      "epoch": 1.57128108061602,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007758739055070568,
      "loss": 1.1974,
      "step": 18263
    },
    {
      "epoch": 1.5713671169233416,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007758513528326712,
      "loss": 1.4027,
      "step": 18264
    },
    {
      "epoch": 1.5714531532306633,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000775828799351474,
      "loss": 1.0819,
      "step": 18265
    },
    {
      "epoch": 1.5715391895379849,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007758062450635313,
      "loss": 1.3001,
      "step": 18266
    },
    {
      "epoch": 1.5716252258453067,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007757836899689089,
      "loss": 0.8253,
      "step": 18267
    },
    {
      "epoch": 1.5717112621526286,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007757611340676729,
      "loss": 1.2895,
      "step": 18268
    },
    {
      "epoch": 1.5717972984599502,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007757385773598893,
      "loss": 0.9899,
      "step": 18269
    },
    {
      "epoch": 1.5718833347672718,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007757160198456238,
      "loss": 2.2401,
      "step": 18270
    },
    {
      "epoch": 1.5719693710745934,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007756934615249425,
      "loss": 1.6526,
      "step": 18271
    },
    {
      "epoch": 1.572055407381915,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007756709023979117,
      "loss": 1.1008,
      "step": 18272
    },
    {
      "epoch": 1.5721414436892367,
      "grad_norm": 1.390625,
      "learning_rate": 0.000775648342464597,
      "loss": 1.2348,
      "step": 18273
    },
    {
      "epoch": 1.5722274799965585,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0007756257817250645,
      "loss": 1.4582,
      "step": 18274
    },
    {
      "epoch": 1.5723135163038804,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007756032201793803,
      "loss": 1.0985,
      "step": 18275
    },
    {
      "epoch": 1.572399552611202,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007755806578276101,
      "loss": 1.0444,
      "step": 18276
    },
    {
      "epoch": 1.5724855889185236,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007755580946698202,
      "loss": 2.2027,
      "step": 18277
    },
    {
      "epoch": 1.5725716252258453,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007755355307060766,
      "loss": 1.1011,
      "step": 18278
    },
    {
      "epoch": 1.5726576615331669,
      "grad_norm": 1.78125,
      "learning_rate": 0.000775512965936445,
      "loss": 1.442,
      "step": 18279
    },
    {
      "epoch": 1.5727436978404887,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007754904003609915,
      "loss": 1.3747,
      "step": 18280
    },
    {
      "epoch": 1.5728297341478104,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007754678339797822,
      "loss": 1.167,
      "step": 18281
    },
    {
      "epoch": 1.5729157704551322,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007754452667928831,
      "loss": 0.7999,
      "step": 18282
    },
    {
      "epoch": 1.5730018067624538,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007754226988003602,
      "loss": 0.6866,
      "step": 18283
    },
    {
      "epoch": 1.5730878430697754,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007754001300022795,
      "loss": 0.8571,
      "step": 18284
    },
    {
      "epoch": 1.573173879377097,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007753775603987068,
      "loss": 1.4305,
      "step": 18285
    },
    {
      "epoch": 1.5732599156844187,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007753549899897084,
      "loss": 1.8053,
      "step": 18286
    },
    {
      "epoch": 1.5733459519917405,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007753324187753502,
      "loss": 1.5705,
      "step": 18287
    },
    {
      "epoch": 1.5734319882990622,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007753098467556983,
      "loss": 1.1556,
      "step": 18288
    },
    {
      "epoch": 1.573518024606384,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007752872739308185,
      "loss": 1.421,
      "step": 18289
    },
    {
      "epoch": 1.5736040609137056,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007752647003007771,
      "loss": 1.3064,
      "step": 18290
    },
    {
      "epoch": 1.5736900972210273,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00077524212586564,
      "loss": 0.6508,
      "step": 18291
    },
    {
      "epoch": 1.5737761335283489,
      "grad_norm": 1.109375,
      "learning_rate": 0.000775219550625473,
      "loss": 1.931,
      "step": 18292
    },
    {
      "epoch": 1.5738621698356705,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007751969745803424,
      "loss": 1.3559,
      "step": 18293
    },
    {
      "epoch": 1.5739482061429924,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007751743977303144,
      "loss": 2.0748,
      "step": 18294
    },
    {
      "epoch": 1.574034242450314,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0007751518200754546,
      "loss": 1.1983,
      "step": 18295
    },
    {
      "epoch": 1.5741202787576358,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007751292416158293,
      "loss": 1.2076,
      "step": 18296
    },
    {
      "epoch": 1.5742063150649575,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007751066623515045,
      "loss": 1.4737,
      "step": 18297
    },
    {
      "epoch": 1.574292351372279,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007750840822825462,
      "loss": 1.1602,
      "step": 18298
    },
    {
      "epoch": 1.5743783876796007,
      "grad_norm": 1.375,
      "learning_rate": 0.0007750615014090205,
      "loss": 0.8951,
      "step": 18299
    },
    {
      "epoch": 1.5744644239869223,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007750389197309933,
      "loss": 1.1109,
      "step": 18300
    },
    {
      "epoch": 1.5745504602942442,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007750163372485309,
      "loss": 0.7902,
      "step": 18301
    },
    {
      "epoch": 1.574636496601566,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007749937539616991,
      "loss": 1.7056,
      "step": 18302
    },
    {
      "epoch": 1.5747225329088876,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007749711698705642,
      "loss": 0.9111,
      "step": 18303
    },
    {
      "epoch": 1.5748085692162093,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007749485849751918,
      "loss": 1.5893,
      "step": 18304
    },
    {
      "epoch": 1.5748946055235309,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007749259992756485,
      "loss": 1.1156,
      "step": 18305
    },
    {
      "epoch": 1.5749806418308525,
      "grad_norm": 1.015625,
      "learning_rate": 0.000774903412772,
      "loss": 1.002,
      "step": 18306
    },
    {
      "epoch": 1.5750666781381744,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007748808254643126,
      "loss": 1.8134,
      "step": 18307
    },
    {
      "epoch": 1.575152714445496,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007748582373526521,
      "loss": 0.8794,
      "step": 18308
    },
    {
      "epoch": 1.5752387507528178,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007748356484370848,
      "loss": 1.6026,
      "step": 18309
    },
    {
      "epoch": 1.5753247870601395,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007748130587176767,
      "loss": 1.2999,
      "step": 18310
    },
    {
      "epoch": 1.575410823367461,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007747904681944939,
      "loss": 1.1561,
      "step": 18311
    },
    {
      "epoch": 1.5754968596747827,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007747678768676023,
      "loss": 1.0446,
      "step": 18312
    },
    {
      "epoch": 1.5755828959821043,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007747452847370682,
      "loss": 0.8899,
      "step": 18313
    },
    {
      "epoch": 1.5756689322894262,
      "grad_norm": 1.640625,
      "learning_rate": 0.0007747226918029575,
      "loss": 1.133,
      "step": 18314
    },
    {
      "epoch": 1.5757549685967478,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007747000980653365,
      "loss": 1.2376,
      "step": 18315
    },
    {
      "epoch": 1.5758410049040696,
      "grad_norm": 1.640625,
      "learning_rate": 0.000774677503524271,
      "loss": 1.15,
      "step": 18316
    },
    {
      "epoch": 1.5759270412113913,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007746549081798272,
      "loss": 1.8259,
      "step": 18317
    },
    {
      "epoch": 1.576013077518713,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007746323120320713,
      "loss": 1.4748,
      "step": 18318
    },
    {
      "epoch": 1.5760991138260345,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007746097150810692,
      "loss": 1.6509,
      "step": 18319
    },
    {
      "epoch": 1.5761851501333561,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007745871173268872,
      "loss": 0.6402,
      "step": 18320
    },
    {
      "epoch": 1.576271186440678,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007745645187695914,
      "loss": 1.0308,
      "step": 18321
    },
    {
      "epoch": 1.5763572227479996,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007745419194092475,
      "loss": 1.3454,
      "step": 18322
    },
    {
      "epoch": 1.5764432590553215,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007745193192459221,
      "loss": 0.7268,
      "step": 18323
    },
    {
      "epoch": 1.576529295362643,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007744967182796809,
      "loss": 1.7371,
      "step": 18324
    },
    {
      "epoch": 1.5766153316699647,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007744741165105904,
      "loss": 2.4074,
      "step": 18325
    },
    {
      "epoch": 1.5767013679772863,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007744515139387162,
      "loss": 1.3562,
      "step": 18326
    },
    {
      "epoch": 1.576787404284608,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007744289105641249,
      "loss": 1.3022,
      "step": 18327
    },
    {
      "epoch": 1.5768734405919298,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007744063063868825,
      "loss": 1.2891,
      "step": 18328
    },
    {
      "epoch": 1.5769594768992516,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007743837014070547,
      "loss": 0.9773,
      "step": 18329
    },
    {
      "epoch": 1.5770455132065733,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007743610956247082,
      "loss": 0.8344,
      "step": 18330
    },
    {
      "epoch": 1.577131549513895,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007743384890399088,
      "loss": 0.9127,
      "step": 18331
    },
    {
      "epoch": 1.5772175858212165,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007743158816527228,
      "loss": 1.7016,
      "step": 18332
    },
    {
      "epoch": 1.5773036221285381,
      "grad_norm": 0.984375,
      "learning_rate": 0.000774293273463216,
      "loss": 0.8286,
      "step": 18333
    },
    {
      "epoch": 1.57738965843586,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007742706644714548,
      "loss": 1.2289,
      "step": 18334
    },
    {
      "epoch": 1.5774756947431816,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007742480546775052,
      "loss": 1.4677,
      "step": 18335
    },
    {
      "epoch": 1.5775617310505035,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007742254440814334,
      "loss": 1.1045,
      "step": 18336
    },
    {
      "epoch": 1.577647767357825,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007742028326833056,
      "loss": 1.3781,
      "step": 18337
    },
    {
      "epoch": 1.5777338036651467,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007741802204831878,
      "loss": 0.7765,
      "step": 18338
    },
    {
      "epoch": 1.5778198399724683,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007741576074811462,
      "loss": 1.0768,
      "step": 18339
    },
    {
      "epoch": 1.57790587627979,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007741349936772468,
      "loss": 1.4336,
      "step": 18340
    },
    {
      "epoch": 1.5779919125871118,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0007741123790715558,
      "loss": 1.3456,
      "step": 18341
    },
    {
      "epoch": 1.5780779488944334,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007740897636641398,
      "loss": 1.3676,
      "step": 18342
    },
    {
      "epoch": 1.5781639852017553,
      "grad_norm": 0.875,
      "learning_rate": 0.0007740671474550642,
      "loss": 1.1584,
      "step": 18343
    },
    {
      "epoch": 1.578250021509077,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007740445304443955,
      "loss": 1.8093,
      "step": 18344
    },
    {
      "epoch": 1.5783360578163985,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007740219126321999,
      "loss": 1.3945,
      "step": 18345
    },
    {
      "epoch": 1.5784220941237201,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007739992940185435,
      "loss": 1.368,
      "step": 18346
    },
    {
      "epoch": 1.5785081304310418,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007739766746034926,
      "loss": 1.4881,
      "step": 18347
    },
    {
      "epoch": 1.5785941667383636,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007739540543871131,
      "loss": 1.7906,
      "step": 18348
    },
    {
      "epoch": 1.5786802030456852,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007739314333694711,
      "loss": 1.3143,
      "step": 18349
    },
    {
      "epoch": 1.578766239353007,
      "grad_norm": 1.0,
      "learning_rate": 0.0007739088115506332,
      "loss": 1.6502,
      "step": 18350
    },
    {
      "epoch": 1.5788522756603287,
      "grad_norm": 5.53125,
      "learning_rate": 0.0007738861889306652,
      "loss": 1.0318,
      "step": 18351
    },
    {
      "epoch": 1.5789383119676503,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007738635655096333,
      "loss": 1.4835,
      "step": 18352
    },
    {
      "epoch": 1.579024348274972,
      "grad_norm": 1.03125,
      "learning_rate": 0.000773840941287604,
      "loss": 1.6673,
      "step": 18353
    },
    {
      "epoch": 1.5791103845822936,
      "grad_norm": 1.34375,
      "learning_rate": 0.000773818316264643,
      "loss": 2.1921,
      "step": 18354
    },
    {
      "epoch": 1.5791964208896154,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007737956904408166,
      "loss": 0.8919,
      "step": 18355
    },
    {
      "epoch": 1.5792824571969373,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007737730638161911,
      "loss": 1.6125,
      "step": 18356
    },
    {
      "epoch": 1.579368493504259,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007737504363908328,
      "loss": 1.2768,
      "step": 18357
    },
    {
      "epoch": 1.5794545298115805,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007737278081648075,
      "loss": 0.7596,
      "step": 18358
    },
    {
      "epoch": 1.5795405661189021,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007737051791381817,
      "loss": 0.9037,
      "step": 18359
    },
    {
      "epoch": 1.5796266024262238,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007736825493110214,
      "loss": 2.1628,
      "step": 18360
    },
    {
      "epoch": 1.5797126387335454,
      "grad_norm": 0.8125,
      "learning_rate": 0.000773659918683393,
      "loss": 0.5335,
      "step": 18361
    },
    {
      "epoch": 1.5797986750408672,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007736372872553625,
      "loss": 0.8613,
      "step": 18362
    },
    {
      "epoch": 1.579884711348189,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007736146550269961,
      "loss": 1.1332,
      "step": 18363
    },
    {
      "epoch": 1.5799707476555107,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007735920219983601,
      "loss": 1.583,
      "step": 18364
    },
    {
      "epoch": 1.5800567839628323,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007735693881695207,
      "loss": 1.6455,
      "step": 18365
    },
    {
      "epoch": 1.580142820270154,
      "grad_norm": 1.15625,
      "learning_rate": 0.000773546753540544,
      "loss": 1.7131,
      "step": 18366
    },
    {
      "epoch": 1.5802288565774756,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007735241181114962,
      "loss": 1.2892,
      "step": 18367
    },
    {
      "epoch": 1.5803148928847974,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007735014818824437,
      "loss": 1.4668,
      "step": 18368
    },
    {
      "epoch": 1.580400929192119,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007734788448534525,
      "loss": 0.9501,
      "step": 18369
    },
    {
      "epoch": 1.580486965499441,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007734562070245887,
      "loss": 2.0022,
      "step": 18370
    },
    {
      "epoch": 1.5805730018067625,
      "grad_norm": 1.125,
      "learning_rate": 0.0007734335683959188,
      "loss": 0.9325,
      "step": 18371
    },
    {
      "epoch": 1.5806590381140841,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000773410928967509,
      "loss": 1.0368,
      "step": 18372
    },
    {
      "epoch": 1.5807450744214058,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007733882887394252,
      "loss": 0.6868,
      "step": 18373
    },
    {
      "epoch": 1.5808311107287274,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000773365647711734,
      "loss": 1.4237,
      "step": 18374
    },
    {
      "epoch": 1.5809171470360492,
      "grad_norm": 1.125,
      "learning_rate": 0.0007733430058845014,
      "loss": 1.051,
      "step": 18375
    },
    {
      "epoch": 1.5810031833433709,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007733203632577936,
      "loss": 0.8011,
      "step": 18376
    },
    {
      "epoch": 1.5810892196506927,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007732977198316771,
      "loss": 1.4232,
      "step": 18377
    },
    {
      "epoch": 1.5811752559580143,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007732750756062177,
      "loss": 0.9613,
      "step": 18378
    },
    {
      "epoch": 1.581261292265336,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007732524305814819,
      "loss": 2.0189,
      "step": 18379
    },
    {
      "epoch": 1.5813473285726576,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007732297847575359,
      "loss": 1.1422,
      "step": 18380
    },
    {
      "epoch": 1.5814333648799792,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007732071381344459,
      "loss": 1.0876,
      "step": 18381
    },
    {
      "epoch": 1.581519401187301,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007731844907122781,
      "loss": 1.3193,
      "step": 18382
    },
    {
      "epoch": 1.5816054374946227,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007731618424910989,
      "loss": 1.1977,
      "step": 18383
    },
    {
      "epoch": 1.5816914738019445,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007731391934709745,
      "loss": 0.7001,
      "step": 18384
    },
    {
      "epoch": 1.5817775101092661,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007731165436519709,
      "loss": 1.9554,
      "step": 18385
    },
    {
      "epoch": 1.5818635464165878,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007730938930341546,
      "loss": 1.5697,
      "step": 18386
    },
    {
      "epoch": 1.5819495827239094,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007730712416175919,
      "loss": 1.175,
      "step": 18387
    },
    {
      "epoch": 1.582035619031231,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007730485894023489,
      "loss": 1.2589,
      "step": 18388
    },
    {
      "epoch": 1.5821216553385529,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007730259363884917,
      "loss": 1.0322,
      "step": 18389
    },
    {
      "epoch": 1.5822076916458747,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007730032825760868,
      "loss": 1.0214,
      "step": 18390
    },
    {
      "epoch": 1.5822937279531963,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007729806279652005,
      "loss": 1.7009,
      "step": 18391
    },
    {
      "epoch": 1.582379764260518,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000772957972555899,
      "loss": 0.9404,
      "step": 18392
    },
    {
      "epoch": 1.5824658005678396,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007729353163482485,
      "loss": 0.9392,
      "step": 18393
    },
    {
      "epoch": 1.5825518368751612,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007729126593423151,
      "loss": 1.0972,
      "step": 18394
    },
    {
      "epoch": 1.582637873182483,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007728900015381654,
      "loss": 0.9196,
      "step": 18395
    },
    {
      "epoch": 1.5827239094898047,
      "grad_norm": 1.71875,
      "learning_rate": 0.0007728673429358655,
      "loss": 2.2695,
      "step": 18396
    },
    {
      "epoch": 1.5828099457971265,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007728446835354815,
      "loss": 1.8236,
      "step": 18397
    },
    {
      "epoch": 1.5828959821044482,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007728220233370801,
      "loss": 0.9743,
      "step": 18398
    },
    {
      "epoch": 1.5829820184117698,
      "grad_norm": 1.75,
      "learning_rate": 0.0007727993623407272,
      "loss": 0.9977,
      "step": 18399
    },
    {
      "epoch": 1.5830680547190914,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007727767005464893,
      "loss": 1.5356,
      "step": 18400
    },
    {
      "epoch": 1.583154091026413,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0007727540379544326,
      "loss": 1.4737,
      "step": 18401
    },
    {
      "epoch": 1.5832401273337349,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0007727313745646234,
      "loss": 1.7851,
      "step": 18402
    },
    {
      "epoch": 1.5833261636410565,
      "grad_norm": 1.1875,
      "learning_rate": 0.000772708710377128,
      "loss": 1.4669,
      "step": 18403
    },
    {
      "epoch": 1.5834121999483783,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007726860453920126,
      "loss": 1.3812,
      "step": 18404
    },
    {
      "epoch": 1.5834982362557,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007726633796093436,
      "loss": 1.0534,
      "step": 18405
    },
    {
      "epoch": 1.5835842725630216,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007726407130291871,
      "loss": 0.7164,
      "step": 18406
    },
    {
      "epoch": 1.5836703088703432,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007726180456516096,
      "loss": 1.2896,
      "step": 18407
    },
    {
      "epoch": 1.5837563451776648,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007725953774766776,
      "loss": 1.1803,
      "step": 18408
    },
    {
      "epoch": 1.5838423814849867,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007725727085044568,
      "loss": 1.6362,
      "step": 18409
    },
    {
      "epoch": 1.5839284177923083,
      "grad_norm": 1.265625,
      "learning_rate": 0.000772550038735014,
      "loss": 1.6557,
      "step": 18410
    },
    {
      "epoch": 1.5840144540996302,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007725273681684153,
      "loss": 1.7527,
      "step": 18411
    },
    {
      "epoch": 1.5841004904069518,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007725046968047271,
      "loss": 1.3734,
      "step": 18412
    },
    {
      "epoch": 1.5841865267142734,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007724820246440157,
      "loss": 0.7811,
      "step": 18413
    },
    {
      "epoch": 1.584272563021595,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007724593516863472,
      "loss": 0.9396,
      "step": 18414
    },
    {
      "epoch": 1.5843585993289167,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007724366779317882,
      "loss": 1.4141,
      "step": 18415
    },
    {
      "epoch": 1.5844446356362385,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007724140033804048,
      "loss": 1.4194,
      "step": 18416
    },
    {
      "epoch": 1.5845306719435603,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007723913280322637,
      "loss": 1.1962,
      "step": 18417
    },
    {
      "epoch": 1.584616708250882,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007723686518874307,
      "loss": 1.77,
      "step": 18418
    },
    {
      "epoch": 1.5847027445582036,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007723459749459723,
      "loss": 1.7763,
      "step": 18419
    },
    {
      "epoch": 1.5847887808655252,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007723232972079551,
      "loss": 0.996,
      "step": 18420
    },
    {
      "epoch": 1.5848748171728468,
      "grad_norm": 1.390625,
      "learning_rate": 0.000772300618673445,
      "loss": 1.2102,
      "step": 18421
    },
    {
      "epoch": 1.5849608534801687,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007722779393425088,
      "loss": 1.7507,
      "step": 18422
    },
    {
      "epoch": 1.5850468897874903,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007722552592152123,
      "loss": 1.7378,
      "step": 18423
    },
    {
      "epoch": 1.5851329260948122,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007722325782916225,
      "loss": 1.0239,
      "step": 18424
    },
    {
      "epoch": 1.5852189624021338,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007722098965718049,
      "loss": 1.2002,
      "step": 18425
    },
    {
      "epoch": 1.5853049987094554,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007721872140558266,
      "loss": 0.8351,
      "step": 18426
    },
    {
      "epoch": 1.585391035016777,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007721645307437537,
      "loss": 1.0955,
      "step": 18427
    },
    {
      "epoch": 1.5854770713240987,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007721418466356523,
      "loss": 1.125,
      "step": 18428
    },
    {
      "epoch": 1.5855631076314205,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007721191617315889,
      "loss": 0.6678,
      "step": 18429
    },
    {
      "epoch": 1.5856491439387421,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00077209647603163,
      "loss": 1.5617,
      "step": 18430
    },
    {
      "epoch": 1.585735180246064,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007720737895358417,
      "loss": 0.7571,
      "step": 18431
    },
    {
      "epoch": 1.5858212165533856,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007720511022442907,
      "loss": 1.4955,
      "step": 18432
    },
    {
      "epoch": 1.5859072528607072,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007720284141570429,
      "loss": 1.6369,
      "step": 18433
    },
    {
      "epoch": 1.5859932891680288,
      "grad_norm": 1.234375,
      "learning_rate": 0.000772005725274165,
      "loss": 1.3337,
      "step": 18434
    },
    {
      "epoch": 1.5860793254753505,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007719830355957233,
      "loss": 0.7959,
      "step": 18435
    },
    {
      "epoch": 1.5861653617826723,
      "grad_norm": 1.9453125,
      "learning_rate": 0.000771960345121784,
      "loss": 1.2534,
      "step": 18436
    },
    {
      "epoch": 1.586251398089994,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007719376538524136,
      "loss": 1.0321,
      "step": 18437
    },
    {
      "epoch": 1.5863374343973158,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007719149617876785,
      "loss": 1.2469,
      "step": 18438
    },
    {
      "epoch": 1.5864234707046374,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007718922689276451,
      "loss": 1.1575,
      "step": 18439
    },
    {
      "epoch": 1.586509507011959,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007718695752723795,
      "loss": 1.1255,
      "step": 18440
    },
    {
      "epoch": 1.5865955433192807,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007718468808219484,
      "loss": 1.4982,
      "step": 18441
    },
    {
      "epoch": 1.5866815796266023,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007718241855764181,
      "loss": 1.6405,
      "step": 18442
    },
    {
      "epoch": 1.5867676159339241,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007718014895358548,
      "loss": 1.5568,
      "step": 18443
    },
    {
      "epoch": 1.586853652241246,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0007717787927003251,
      "loss": 0.7491,
      "step": 18444
    },
    {
      "epoch": 1.5869396885485676,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007717560950698952,
      "loss": 0.6435,
      "step": 18445
    },
    {
      "epoch": 1.5870257248558892,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007717333966446316,
      "loss": 0.8862,
      "step": 18446
    },
    {
      "epoch": 1.5871117611632108,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007717106974246008,
      "loss": 1.3045,
      "step": 18447
    },
    {
      "epoch": 1.5871977974705325,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007716879974098688,
      "loss": 0.7283,
      "step": 18448
    },
    {
      "epoch": 1.587283833777854,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007716652966005025,
      "loss": 1.0077,
      "step": 18449
    },
    {
      "epoch": 1.587369870085176,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000771642594996568,
      "loss": 1.4509,
      "step": 18450
    },
    {
      "epoch": 1.5874559063924978,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007716198925981316,
      "loss": 1.6353,
      "step": 18451
    },
    {
      "epoch": 1.5875419426998194,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007715971894052599,
      "loss": 1.0508,
      "step": 18452
    },
    {
      "epoch": 1.587627979007141,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007715744854180192,
      "loss": 1.7469,
      "step": 18453
    },
    {
      "epoch": 1.5877140153144627,
      "grad_norm": 1.234375,
      "learning_rate": 0.000771551780636476,
      "loss": 1.2753,
      "step": 18454
    },
    {
      "epoch": 1.5878000516217843,
      "grad_norm": 1.0,
      "learning_rate": 0.0007715290750606967,
      "loss": 1.8354,
      "step": 18455
    },
    {
      "epoch": 1.5878860879291061,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007715063686907476,
      "loss": 1.2942,
      "step": 18456
    },
    {
      "epoch": 1.5879721242364278,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007714836615266952,
      "loss": 0.9613,
      "step": 18457
    },
    {
      "epoch": 1.5880581605437496,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007714609535686059,
      "loss": 1.2131,
      "step": 18458
    },
    {
      "epoch": 1.5881441968510712,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007714382448165461,
      "loss": 1.4659,
      "step": 18459
    },
    {
      "epoch": 1.5882302331583928,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007714155352705823,
      "loss": 1.3773,
      "step": 18460
    },
    {
      "epoch": 1.5883162694657145,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007713928249307807,
      "loss": 0.7113,
      "step": 18461
    },
    {
      "epoch": 1.588402305773036,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000771370113797208,
      "loss": 1.1128,
      "step": 18462
    },
    {
      "epoch": 1.588488342080358,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007713474018699304,
      "loss": 1.0709,
      "step": 18463
    },
    {
      "epoch": 1.5885743783876796,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007713246891490144,
      "loss": 1.0532,
      "step": 18464
    },
    {
      "epoch": 1.5886604146950014,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007713019756345266,
      "loss": 1.7857,
      "step": 18465
    },
    {
      "epoch": 1.588746451002323,
      "grad_norm": 2.046875,
      "learning_rate": 0.0007712792613265332,
      "loss": 1.5894,
      "step": 18466
    },
    {
      "epoch": 1.5888324873096447,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007712565462251006,
      "loss": 1.2373,
      "step": 18467
    },
    {
      "epoch": 1.5889185236169663,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007712338303302956,
      "loss": 1.6564,
      "step": 18468
    },
    {
      "epoch": 1.589004559924288,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007712111136421843,
      "loss": 1.1802,
      "step": 18469
    },
    {
      "epoch": 1.5890905962316098,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007711883961608332,
      "loss": 1.377,
      "step": 18470
    },
    {
      "epoch": 1.5891766325389314,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007711656778863088,
      "loss": 0.6192,
      "step": 18471
    },
    {
      "epoch": 1.5892626688462532,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007711429588186776,
      "loss": 1.851,
      "step": 18472
    },
    {
      "epoch": 1.5893487051535748,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007711202389580059,
      "loss": 0.935,
      "step": 18473
    },
    {
      "epoch": 1.5894347414608965,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007710975183043603,
      "loss": 1.33,
      "step": 18474
    },
    {
      "epoch": 1.589520777768218,
      "grad_norm": 1.421875,
      "learning_rate": 0.000771074796857807,
      "loss": 1.1825,
      "step": 18475
    },
    {
      "epoch": 1.5896068140755397,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007710520746184127,
      "loss": 1.996,
      "step": 18476
    },
    {
      "epoch": 1.5896928503828616,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007710293515862438,
      "loss": 1.1968,
      "step": 18477
    },
    {
      "epoch": 1.5897788866901834,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000771006627761367,
      "loss": 0.9162,
      "step": 18478
    },
    {
      "epoch": 1.589864922997505,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007709839031438484,
      "loss": 1.306,
      "step": 18479
    },
    {
      "epoch": 1.5899509593048267,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007709611777337544,
      "loss": 1.4753,
      "step": 18480
    },
    {
      "epoch": 1.5900369956121483,
      "grad_norm": 1.234375,
      "learning_rate": 0.000770938451531152,
      "loss": 2.5312,
      "step": 18481
    },
    {
      "epoch": 1.59012303191947,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000770915724536107,
      "loss": 1.3137,
      "step": 18482
    },
    {
      "epoch": 1.5902090682267918,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007708929967486864,
      "loss": 0.932,
      "step": 18483
    },
    {
      "epoch": 1.5902951045341134,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007708702681689563,
      "loss": 1.074,
      "step": 18484
    },
    {
      "epoch": 1.5903811408414352,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007708475387969833,
      "loss": 0.6451,
      "step": 18485
    },
    {
      "epoch": 1.5904671771487568,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007708248086328342,
      "loss": 1.9116,
      "step": 18486
    },
    {
      "epoch": 1.5905532134560785,
      "grad_norm": 3.34375,
      "learning_rate": 0.000770802077676575,
      "loss": 2.5155,
      "step": 18487
    },
    {
      "epoch": 1.5906392497634,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007707793459282725,
      "loss": 0.7933,
      "step": 18488
    },
    {
      "epoch": 1.5907252860707217,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007707566133879931,
      "loss": 2.0472,
      "step": 18489
    },
    {
      "epoch": 1.5908113223780436,
      "grad_norm": 1.765625,
      "learning_rate": 0.0007707338800558031,
      "loss": 1.4359,
      "step": 18490
    },
    {
      "epoch": 1.5908973586853652,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007707111459317693,
      "loss": 1.4776,
      "step": 18491
    },
    {
      "epoch": 1.590983394992687,
      "grad_norm": 1.078125,
      "learning_rate": 0.000770688411015958,
      "loss": 1.1011,
      "step": 18492
    },
    {
      "epoch": 1.5910694313000087,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007706656753084357,
      "loss": 1.4328,
      "step": 18493
    },
    {
      "epoch": 1.5911554676073303,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000770642938809269,
      "loss": 1.6419,
      "step": 18494
    },
    {
      "epoch": 1.591241503914652,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007706202015185243,
      "loss": 1.5054,
      "step": 18495
    },
    {
      "epoch": 1.5913275402219735,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007705974634362682,
      "loss": 2.1079,
      "step": 18496
    },
    {
      "epoch": 1.5914135765292954,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007705747245625672,
      "loss": 0.8635,
      "step": 18497
    },
    {
      "epoch": 1.591499612836617,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007705519848974875,
      "loss": 1.5454,
      "step": 18498
    },
    {
      "epoch": 1.5915856491439389,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007705292444410961,
      "loss": 1.0039,
      "step": 18499
    },
    {
      "epoch": 1.5916716854512605,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007705065031934592,
      "loss": 1.4191,
      "step": 18500
    },
    {
      "epoch": 1.591757721758582,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007704837611546434,
      "loss": 1.0457,
      "step": 18501
    },
    {
      "epoch": 1.5918437580659037,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007704610183247152,
      "loss": 1.3942,
      "step": 18502
    },
    {
      "epoch": 1.5919297943732253,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007704382747037411,
      "loss": 1.277,
      "step": 18503
    },
    {
      "epoch": 1.5920158306805472,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007704155302917877,
      "loss": 1.0165,
      "step": 18504
    },
    {
      "epoch": 1.592101866987869,
      "grad_norm": 1.375,
      "learning_rate": 0.0007703927850889215,
      "loss": 1.4745,
      "step": 18505
    },
    {
      "epoch": 1.5921879032951907,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007703700390952089,
      "loss": 1.2249,
      "step": 18506
    },
    {
      "epoch": 1.5922739396025123,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007703472923107166,
      "loss": 1.4437,
      "step": 18507
    },
    {
      "epoch": 1.592359975909834,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007703245447355109,
      "loss": 1.4731,
      "step": 18508
    },
    {
      "epoch": 1.5924460122171555,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007703017963696586,
      "loss": 0.9654,
      "step": 18509
    },
    {
      "epoch": 1.5925320485244774,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007702790472132262,
      "loss": 1.5666,
      "step": 18510
    },
    {
      "epoch": 1.592618084831799,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007702562972662799,
      "loss": 0.9621,
      "step": 18511
    },
    {
      "epoch": 1.5927041211391209,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007702335465288867,
      "loss": 0.7033,
      "step": 18512
    },
    {
      "epoch": 1.5927901574464425,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007702107950011128,
      "loss": 2.049,
      "step": 18513
    },
    {
      "epoch": 1.592876193753764,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000770188042683025,
      "loss": 1.595,
      "step": 18514
    },
    {
      "epoch": 1.5929622300610857,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007701652895746898,
      "loss": 1.1536,
      "step": 18515
    },
    {
      "epoch": 1.5930482663684074,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007701425356761735,
      "loss": 1.0368,
      "step": 18516
    },
    {
      "epoch": 1.5931343026757292,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007701197809875429,
      "loss": 1.6252,
      "step": 18517
    },
    {
      "epoch": 1.5932203389830508,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007700970255088644,
      "loss": 1.8798,
      "step": 18518
    },
    {
      "epoch": 1.5933063752903727,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007700742692402047,
      "loss": 1.5167,
      "step": 18519
    },
    {
      "epoch": 1.5933924115976943,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007700515121816302,
      "loss": 1.0409,
      "step": 18520
    },
    {
      "epoch": 1.593478447905016,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007700287543332076,
      "loss": 1.698,
      "step": 18521
    },
    {
      "epoch": 1.5935644842123375,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007700059956950035,
      "loss": 1.2445,
      "step": 18522
    },
    {
      "epoch": 1.5936505205196592,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007699832362670845,
      "loss": 1.3294,
      "step": 18523
    },
    {
      "epoch": 1.593736556826981,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007699604760495167,
      "loss": 1.5157,
      "step": 18524
    },
    {
      "epoch": 1.5938225931343026,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007699377150423672,
      "loss": 0.7201,
      "step": 18525
    },
    {
      "epoch": 1.5939086294416245,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007699149532457023,
      "loss": 0.8297,
      "step": 18526
    },
    {
      "epoch": 1.593994665748946,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007698921906595886,
      "loss": 1.7506,
      "step": 18527
    },
    {
      "epoch": 1.5940807020562677,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007698694272840929,
      "loss": 1.8463,
      "step": 18528
    },
    {
      "epoch": 1.5941667383635894,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0007698466631192815,
      "loss": 0.9648,
      "step": 18529
    },
    {
      "epoch": 1.594252774670911,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007698238981652212,
      "loss": 2.2218,
      "step": 18530
    },
    {
      "epoch": 1.5943388109782328,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007698011324219785,
      "loss": 1.9534,
      "step": 18531
    },
    {
      "epoch": 1.5944248472855547,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007697783658896198,
      "loss": 1.3809,
      "step": 18532
    },
    {
      "epoch": 1.5945108835928763,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007697555985682118,
      "loss": 1.2543,
      "step": 18533
    },
    {
      "epoch": 1.594596919900198,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007697328304578214,
      "loss": 0.9996,
      "step": 18534
    },
    {
      "epoch": 1.5946829562075195,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007697100615585146,
      "loss": 1.1368,
      "step": 18535
    },
    {
      "epoch": 1.5947689925148412,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007696872918703585,
      "loss": 0.9547,
      "step": 18536
    },
    {
      "epoch": 1.5948550288221628,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007696645213934197,
      "loss": 1.621,
      "step": 18537
    },
    {
      "epoch": 1.5949410651294846,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007696417501277645,
      "loss": 1.1611,
      "step": 18538
    },
    {
      "epoch": 1.5950271014368065,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007696189780734595,
      "loss": 1.3544,
      "step": 18539
    },
    {
      "epoch": 1.595113137744128,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007695962052305716,
      "loss": 1.27,
      "step": 18540
    },
    {
      "epoch": 1.5951991740514497,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007695734315991672,
      "loss": 0.8272,
      "step": 18541
    },
    {
      "epoch": 1.5952852103587714,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007695506571793127,
      "loss": 1.3972,
      "step": 18542
    },
    {
      "epoch": 1.595371246666093,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007695278819710753,
      "loss": 1.033,
      "step": 18543
    },
    {
      "epoch": 1.5954572829734148,
      "grad_norm": 1.0625,
      "learning_rate": 0.000769505105974521,
      "loss": 1.1028,
      "step": 18544
    },
    {
      "epoch": 1.5955433192807364,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000769482329189717,
      "loss": 1.3763,
      "step": 18545
    },
    {
      "epoch": 1.5956293555880583,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007694595516167293,
      "loss": 1.9445,
      "step": 18546
    },
    {
      "epoch": 1.59571539189538,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007694367732556249,
      "loss": 1.1649,
      "step": 18547
    },
    {
      "epoch": 1.5958014282027015,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007694139941064703,
      "loss": 1.279,
      "step": 18548
    },
    {
      "epoch": 1.5958874645100232,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007693912141693322,
      "loss": 1.795,
      "step": 18549
    },
    {
      "epoch": 1.5959735008173448,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007693684334442772,
      "loss": 1.3085,
      "step": 18550
    },
    {
      "epoch": 1.5960595371246666,
      "grad_norm": 1.203125,
      "learning_rate": 0.000769345651931372,
      "loss": 1.1686,
      "step": 18551
    },
    {
      "epoch": 1.5961455734319883,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007693228696306831,
      "loss": 1.2741,
      "step": 18552
    },
    {
      "epoch": 1.59623160973931,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007693000865422771,
      "loss": 1.4776,
      "step": 18553
    },
    {
      "epoch": 1.5963176460466317,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007692773026662207,
      "loss": 1.102,
      "step": 18554
    },
    {
      "epoch": 1.5964036823539534,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007692545180025807,
      "loss": 0.9243,
      "step": 18555
    },
    {
      "epoch": 1.596489718661275,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007692317325514235,
      "loss": 1.2976,
      "step": 18556
    },
    {
      "epoch": 1.5965757549685966,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000769208946312816,
      "loss": 1.1501,
      "step": 18557
    },
    {
      "epoch": 1.5966617912759185,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007691861592868246,
      "loss": 1.564,
      "step": 18558
    },
    {
      "epoch": 1.59674782758324,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000769163371473516,
      "loss": 1.4881,
      "step": 18559
    },
    {
      "epoch": 1.596833863890562,
      "grad_norm": 0.8125,
      "learning_rate": 0.000769140582872957,
      "loss": 0.9641,
      "step": 18560
    },
    {
      "epoch": 1.5969199001978835,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007691177934852139,
      "loss": 0.9637,
      "step": 18561
    },
    {
      "epoch": 1.5970059365052052,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007690950033103539,
      "loss": 1.2585,
      "step": 18562
    },
    {
      "epoch": 1.5970919728125268,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007690722123484432,
      "loss": 1.1182,
      "step": 18563
    },
    {
      "epoch": 1.5971780091198484,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007690494205995486,
      "loss": 1.3603,
      "step": 18564
    },
    {
      "epoch": 1.5972640454271703,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007690266280637367,
      "loss": 1.0384,
      "step": 18565
    },
    {
      "epoch": 1.597350081734492,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007690038347410744,
      "loss": 2.0837,
      "step": 18566
    },
    {
      "epoch": 1.5974361180418137,
      "grad_norm": 1.484375,
      "learning_rate": 0.000768981040631628,
      "loss": 1.0849,
      "step": 18567
    },
    {
      "epoch": 1.5975221543491354,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007689582457354645,
      "loss": 1.7229,
      "step": 18568
    },
    {
      "epoch": 1.597608190656457,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007689354500526504,
      "loss": 1.625,
      "step": 18569
    },
    {
      "epoch": 1.5976942269637786,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007689126535832525,
      "loss": 1.2385,
      "step": 18570
    },
    {
      "epoch": 1.5977802632711005,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007688898563273373,
      "loss": 1.0248,
      "step": 18571
    },
    {
      "epoch": 1.597866299578422,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007688670582849717,
      "loss": 1.4298,
      "step": 18572
    },
    {
      "epoch": 1.597952335885744,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000768844259456222,
      "loss": 1.6051,
      "step": 18573
    },
    {
      "epoch": 1.5980383721930655,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007688214598411553,
      "loss": 0.6994,
      "step": 18574
    },
    {
      "epoch": 1.5981244085003872,
      "grad_norm": 1.375,
      "learning_rate": 0.0007687986594398381,
      "loss": 1.3665,
      "step": 18575
    },
    {
      "epoch": 1.5982104448077088,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007687758582523371,
      "loss": 1.0186,
      "step": 18576
    },
    {
      "epoch": 1.5982964811150304,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000768753056278719,
      "loss": 1.3889,
      "step": 18577
    },
    {
      "epoch": 1.5983825174223523,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007687302535190505,
      "loss": 1.8884,
      "step": 18578
    },
    {
      "epoch": 1.598468553729674,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007687074499733982,
      "loss": 1.1048,
      "step": 18579
    },
    {
      "epoch": 1.5985545900369957,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007686846456418288,
      "loss": 1.3673,
      "step": 18580
    },
    {
      "epoch": 1.5986406263443174,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007686618405244092,
      "loss": 1.4082,
      "step": 18581
    },
    {
      "epoch": 1.598726662651639,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000768639034621206,
      "loss": 1.3057,
      "step": 18582
    },
    {
      "epoch": 1.5988126989589606,
      "grad_norm": 1.125,
      "learning_rate": 0.0007686162279322857,
      "loss": 2.3049,
      "step": 18583
    },
    {
      "epoch": 1.5988987352662822,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007685934204577154,
      "loss": 1.271,
      "step": 18584
    },
    {
      "epoch": 1.598984771573604,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007685706121975613,
      "loss": 0.9488,
      "step": 18585
    },
    {
      "epoch": 1.5990708078809257,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007685478031518907,
      "loss": 0.5376,
      "step": 18586
    },
    {
      "epoch": 1.5991568441882476,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007685249933207699,
      "loss": 1.1965,
      "step": 18587
    },
    {
      "epoch": 1.5992428804955692,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007685021827042656,
      "loss": 1.4351,
      "step": 18588
    },
    {
      "epoch": 1.5993289168028908,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007684793713024447,
      "loss": 1.4183,
      "step": 18589
    },
    {
      "epoch": 1.5994149531102124,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007684565591153739,
      "loss": 1.4408,
      "step": 18590
    },
    {
      "epoch": 1.599500989417534,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007684337461431197,
      "loss": 1.7611,
      "step": 18591
    },
    {
      "epoch": 1.599587025724856,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007684109323857493,
      "loss": 1.1628,
      "step": 18592
    },
    {
      "epoch": 1.5996730620321777,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000768388117843329,
      "loss": 1.3979,
      "step": 18593
    },
    {
      "epoch": 1.5997590983394994,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007683653025159256,
      "loss": 1.4239,
      "step": 18594
    },
    {
      "epoch": 1.599845134646821,
      "grad_norm": 1.0,
      "learning_rate": 0.000768342486403606,
      "loss": 0.929,
      "step": 18595
    },
    {
      "epoch": 1.5999311709541426,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007683196695064367,
      "loss": 1.3331,
      "step": 18596
    },
    {
      "epoch": 1.6000172072614642,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007682968518244846,
      "loss": 0.9528,
      "step": 18597
    },
    {
      "epoch": 1.600103243568786,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0007682740333578162,
      "loss": 0.8275,
      "step": 18598
    },
    {
      "epoch": 1.6001892798761077,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007682512141064988,
      "loss": 1.5225,
      "step": 18599
    },
    {
      "epoch": 1.6002753161834296,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007682283940705986,
      "loss": 1.4621,
      "step": 18600
    },
    {
      "epoch": 1.6003613524907512,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007682055732501825,
      "loss": 1.5891,
      "step": 18601
    },
    {
      "epoch": 1.6004473887980728,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007681827516453171,
      "loss": 0.9692,
      "step": 18602
    },
    {
      "epoch": 1.6005334251053944,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007681599292560695,
      "loss": 0.9896,
      "step": 18603
    },
    {
      "epoch": 1.600619461412716,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007681371060825062,
      "loss": 1.1788,
      "step": 18604
    },
    {
      "epoch": 1.600705497720038,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000768114282124694,
      "loss": 0.8865,
      "step": 18605
    },
    {
      "epoch": 1.6007915340273595,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007680914573826997,
      "loss": 1.5688,
      "step": 18606
    },
    {
      "epoch": 1.6008775703346814,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007680686318565901,
      "loss": 1.4476,
      "step": 18607
    },
    {
      "epoch": 1.600963606642003,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007680458055464318,
      "loss": 0.8279,
      "step": 18608
    },
    {
      "epoch": 1.6010496429493246,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007680229784522917,
      "loss": 2.0688,
      "step": 18609
    },
    {
      "epoch": 1.6011356792566462,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007680001505742365,
      "loss": 0.9829,
      "step": 18610
    },
    {
      "epoch": 1.6012217155639679,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007679773219123329,
      "loss": 1.2956,
      "step": 18611
    },
    {
      "epoch": 1.6013077518712897,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007679544924666479,
      "loss": 0.5443,
      "step": 18612
    },
    {
      "epoch": 1.6013937881786113,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000767931662237248,
      "loss": 1.5406,
      "step": 18613
    },
    {
      "epoch": 1.6014798244859332,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007679088312242002,
      "loss": 1.0404,
      "step": 18614
    },
    {
      "epoch": 1.6015658607932548,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007678859994275711,
      "loss": 1.7644,
      "step": 18615
    },
    {
      "epoch": 1.6016518971005764,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007678631668474276,
      "loss": 1.4931,
      "step": 18616
    },
    {
      "epoch": 1.601737933407898,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007678403334838364,
      "loss": 0.9535,
      "step": 18617
    },
    {
      "epoch": 1.6018239697152197,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007678174993368642,
      "loss": 0.7918,
      "step": 18618
    },
    {
      "epoch": 1.6019100060225415,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007677946644065781,
      "loss": 1.2882,
      "step": 18619
    },
    {
      "epoch": 1.6019960423298634,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007677718286930446,
      "loss": 1.2155,
      "step": 18620
    },
    {
      "epoch": 1.602082078637185,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007677489921963306,
      "loss": 1.5187,
      "step": 18621
    },
    {
      "epoch": 1.6021681149445066,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007677261549165027,
      "loss": 1.3354,
      "step": 18622
    },
    {
      "epoch": 1.6022541512518282,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007677033168536282,
      "loss": 0.9905,
      "step": 18623
    },
    {
      "epoch": 1.6023401875591499,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007676804780077733,
      "loss": 0.8454,
      "step": 18624
    },
    {
      "epoch": 1.6024262238664715,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007676576383790052,
      "loss": 1.2518,
      "step": 18625
    },
    {
      "epoch": 1.6025122601737933,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007676347979673905,
      "loss": 1.1387,
      "step": 18626
    },
    {
      "epoch": 1.6025982964811152,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007676119567729962,
      "loss": 1.1218,
      "step": 18627
    },
    {
      "epoch": 1.6026843327884368,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007675891147958889,
      "loss": 2.0502,
      "step": 18628
    },
    {
      "epoch": 1.6027703690957584,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007675662720361354,
      "loss": 1.2024,
      "step": 18629
    },
    {
      "epoch": 1.60285640540308,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007675434284938026,
      "loss": 0.979,
      "step": 18630
    },
    {
      "epoch": 1.6029424417104017,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007675205841689574,
      "loss": 1.2603,
      "step": 18631
    },
    {
      "epoch": 1.6030284780177235,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007674977390616664,
      "loss": 0.9163,
      "step": 18632
    },
    {
      "epoch": 1.6031145143250451,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007674748931719966,
      "loss": 0.9767,
      "step": 18633
    },
    {
      "epoch": 1.603200550632367,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007674520465000147,
      "loss": 1.6464,
      "step": 18634
    },
    {
      "epoch": 1.6032865869396886,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007674291990457878,
      "loss": 1.0307,
      "step": 18635
    },
    {
      "epoch": 1.6033726232470102,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007674063508093823,
      "loss": 1.5216,
      "step": 18636
    },
    {
      "epoch": 1.6034586595543319,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007673835017908652,
      "loss": 1.0648,
      "step": 18637
    },
    {
      "epoch": 1.6035446958616535,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007673606519903036,
      "loss": 1.6747,
      "step": 18638
    },
    {
      "epoch": 1.6036307321689753,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007673378014077639,
      "loss": 1.3459,
      "step": 18639
    },
    {
      "epoch": 1.603716768476297,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007673149500433131,
      "loss": 1.3403,
      "step": 18640
    },
    {
      "epoch": 1.6038028047836188,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000767292097897018,
      "loss": 2.4102,
      "step": 18641
    },
    {
      "epoch": 1.6038888410909404,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007672692449689456,
      "loss": 1.8213,
      "step": 18642
    },
    {
      "epoch": 1.603974877398262,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007672463912591627,
      "loss": 0.9408,
      "step": 18643
    },
    {
      "epoch": 1.6040609137055837,
      "grad_norm": 1.0,
      "learning_rate": 0.0007672235367677361,
      "loss": 0.8053,
      "step": 18644
    },
    {
      "epoch": 1.6041469500129053,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007672006814947325,
      "loss": 1.2103,
      "step": 18645
    },
    {
      "epoch": 1.6042329863202272,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007671778254402191,
      "loss": 1.1549,
      "step": 18646
    },
    {
      "epoch": 1.6043190226275488,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007671549686042623,
      "loss": 0.9279,
      "step": 18647
    },
    {
      "epoch": 1.6044050589348706,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007671321109869292,
      "loss": 1.5076,
      "step": 18648
    },
    {
      "epoch": 1.6044910952421922,
      "grad_norm": 1.25,
      "learning_rate": 0.0007671092525882866,
      "loss": 0.9426,
      "step": 18649
    },
    {
      "epoch": 1.6045771315495139,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007670863934084016,
      "loss": 1.1236,
      "step": 18650
    },
    {
      "epoch": 1.6046631678568355,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007670635334473406,
      "loss": 1.1609,
      "step": 18651
    },
    {
      "epoch": 1.6047492041641571,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007670406727051708,
      "loss": 1.5826,
      "step": 18652
    },
    {
      "epoch": 1.604835240471479,
      "grad_norm": 0.96875,
      "learning_rate": 0.000767017811181959,
      "loss": 1.216,
      "step": 18653
    },
    {
      "epoch": 1.6049212767788008,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007669949488777719,
      "loss": 1.5064,
      "step": 18654
    },
    {
      "epoch": 1.6050073130861224,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007669720857926768,
      "loss": 0.6013,
      "step": 18655
    },
    {
      "epoch": 1.605093349393444,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00076694922192674,
      "loss": 1.0459,
      "step": 18656
    },
    {
      "epoch": 1.6051793857007657,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007669263572800289,
      "loss": 0.9952,
      "step": 18657
    },
    {
      "epoch": 1.6052654220080873,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007669034918526098,
      "loss": 0.9947,
      "step": 18658
    },
    {
      "epoch": 1.6053514583154092,
      "grad_norm": 1.15625,
      "learning_rate": 0.00076688062564455,
      "loss": 1.3549,
      "step": 18659
    },
    {
      "epoch": 1.6054374946227308,
      "grad_norm": 1.78125,
      "learning_rate": 0.0007668577586559163,
      "loss": 1.7933,
      "step": 18660
    },
    {
      "epoch": 1.6055235309300526,
      "grad_norm": 1.953125,
      "learning_rate": 0.0007668348908867757,
      "loss": 1.5033,
      "step": 18661
    },
    {
      "epoch": 1.6056095672373742,
      "grad_norm": 1.125,
      "learning_rate": 0.0007668120223371947,
      "loss": 2.1632,
      "step": 18662
    },
    {
      "epoch": 1.6056956035446959,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007667891530072406,
      "loss": 1.0846,
      "step": 18663
    },
    {
      "epoch": 1.6057816398520175,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007667662828969801,
      "loss": 1.2944,
      "step": 18664
    },
    {
      "epoch": 1.6058676761593391,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00076674341200648,
      "loss": 1.5442,
      "step": 18665
    },
    {
      "epoch": 1.605953712466661,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007667205403358073,
      "loss": 1.1155,
      "step": 18666
    },
    {
      "epoch": 1.6060397487739826,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007666976678850291,
      "loss": 1.2623,
      "step": 18667
    },
    {
      "epoch": 1.6061257850813044,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007666747946542119,
      "loss": 0.8958,
      "step": 18668
    },
    {
      "epoch": 1.606211821388626,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007666519206434229,
      "loss": 1.0134,
      "step": 18669
    },
    {
      "epoch": 1.6062978576959477,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007666290458527287,
      "loss": 0.9594,
      "step": 18670
    },
    {
      "epoch": 1.6063838940032693,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007666061702821967,
      "loss": 1.1759,
      "step": 18671
    },
    {
      "epoch": 1.606469930310591,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007665832939318933,
      "loss": 1.1411,
      "step": 18672
    },
    {
      "epoch": 1.6065559666179128,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007665604168018855,
      "loss": 0.6648,
      "step": 18673
    },
    {
      "epoch": 1.6066420029252344,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007665375388922407,
      "loss": 1.2679,
      "step": 18674
    },
    {
      "epoch": 1.6067280392325562,
      "grad_norm": 1.734375,
      "learning_rate": 0.0007665146602030251,
      "loss": 1.283,
      "step": 18675
    },
    {
      "epoch": 1.6068140755398779,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007664917807343061,
      "loss": 1.4675,
      "step": 18676
    },
    {
      "epoch": 1.6069001118471995,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007664689004861502,
      "loss": 0.8079,
      "step": 18677
    },
    {
      "epoch": 1.6069861481545211,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007664460194586249,
      "loss": 1.2938,
      "step": 18678
    },
    {
      "epoch": 1.6070721844618427,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007664231376517968,
      "loss": 1.0338,
      "step": 18679
    },
    {
      "epoch": 1.6071582207691646,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007664002550657327,
      "loss": 1.8081,
      "step": 18680
    },
    {
      "epoch": 1.6072442570764864,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007663773717004999,
      "loss": 1.3047,
      "step": 18681
    },
    {
      "epoch": 1.607330293383808,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007663544875561647,
      "loss": 2.2141,
      "step": 18682
    },
    {
      "epoch": 1.6074163296911297,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007663316026327949,
      "loss": 2.0151,
      "step": 18683
    },
    {
      "epoch": 1.6075023659984513,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007663087169304568,
      "loss": 0.7378,
      "step": 18684
    },
    {
      "epoch": 1.607588402305773,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007662858304492173,
      "loss": 0.8685,
      "step": 18685
    },
    {
      "epoch": 1.6076744386130948,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007662629431891437,
      "loss": 1.6561,
      "step": 18686
    },
    {
      "epoch": 1.6077604749204164,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007662400551503027,
      "loss": 1.2172,
      "step": 18687
    },
    {
      "epoch": 1.6078465112277383,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007662171663327613,
      "loss": 0.9593,
      "step": 18688
    },
    {
      "epoch": 1.6079325475350599,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007661942767365865,
      "loss": 1.0564,
      "step": 18689
    },
    {
      "epoch": 1.6080185838423815,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007661713863618454,
      "loss": 0.7393,
      "step": 18690
    },
    {
      "epoch": 1.6081046201497031,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007661484952086045,
      "loss": 0.6803,
      "step": 18691
    },
    {
      "epoch": 1.6081906564570247,
      "grad_norm": 1.109375,
      "learning_rate": 0.000766125603276931,
      "loss": 0.7622,
      "step": 18692
    },
    {
      "epoch": 1.6082766927643466,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007661027105668921,
      "loss": 2.0509,
      "step": 18693
    },
    {
      "epoch": 1.6083627290716682,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007660798170785545,
      "loss": 0.9478,
      "step": 18694
    },
    {
      "epoch": 1.60844876537899,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007660569228119849,
      "loss": 0.585,
      "step": 18695
    },
    {
      "epoch": 1.6085348016863117,
      "grad_norm": 2.140625,
      "learning_rate": 0.0007660340277672507,
      "loss": 1.3155,
      "step": 18696
    },
    {
      "epoch": 1.6086208379936333,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007660111319444188,
      "loss": 1.1092,
      "step": 18697
    },
    {
      "epoch": 1.608706874300955,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007659882353435561,
      "loss": 1.4576,
      "step": 18698
    },
    {
      "epoch": 1.6087929106082766,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007659653379647294,
      "loss": 1.5681,
      "step": 18699
    },
    {
      "epoch": 1.6088789469155984,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007659424398080059,
      "loss": 1.0764,
      "step": 18700
    },
    {
      "epoch": 1.60896498322292,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007659195408734525,
      "loss": 1.4028,
      "step": 18701
    },
    {
      "epoch": 1.6090510195302419,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007658966411611361,
      "loss": 1.7574,
      "step": 18702
    },
    {
      "epoch": 1.6091370558375635,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007658737406711237,
      "loss": 1.6553,
      "step": 18703
    },
    {
      "epoch": 1.6092230921448851,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007658508394034825,
      "loss": 1.5803,
      "step": 18704
    },
    {
      "epoch": 1.6093091284522067,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007658279373582791,
      "loss": 0.8537,
      "step": 18705
    },
    {
      "epoch": 1.6093951647595284,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007658050345355808,
      "loss": 2.0689,
      "step": 18706
    },
    {
      "epoch": 1.6094812010668502,
      "grad_norm": 1.25,
      "learning_rate": 0.0007657821309354544,
      "loss": 1.7617,
      "step": 18707
    },
    {
      "epoch": 1.609567237374172,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007657592265579669,
      "loss": 1.4195,
      "step": 18708
    },
    {
      "epoch": 1.6096532736814937,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007657363214031856,
      "loss": 1.0547,
      "step": 18709
    },
    {
      "epoch": 1.6097393099888153,
      "grad_norm": 1.875,
      "learning_rate": 0.000765713415471177,
      "loss": 1.5579,
      "step": 18710
    },
    {
      "epoch": 1.609825346296137,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007656905087620085,
      "loss": 1.291,
      "step": 18711
    },
    {
      "epoch": 1.6099113826034586,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007656676012757469,
      "loss": 1.1131,
      "step": 18712
    },
    {
      "epoch": 1.6099974189107802,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007656446930124592,
      "loss": 1.5246,
      "step": 18713
    },
    {
      "epoch": 1.610083455218102,
      "grad_norm": 2.203125,
      "learning_rate": 0.0007656217839722124,
      "loss": 1.2343,
      "step": 18714
    },
    {
      "epoch": 1.6101694915254239,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007655988741550735,
      "loss": 1.3726,
      "step": 18715
    },
    {
      "epoch": 1.6102555278327455,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007655759635611096,
      "loss": 1.6012,
      "step": 18716
    },
    {
      "epoch": 1.6103415641400671,
      "grad_norm": 1.0,
      "learning_rate": 0.0007655530521903877,
      "loss": 1.0626,
      "step": 18717
    },
    {
      "epoch": 1.6104276004473888,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007655301400429748,
      "loss": 1.2616,
      "step": 18718
    },
    {
      "epoch": 1.6105136367547104,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007655072271189378,
      "loss": 1.4487,
      "step": 18719
    },
    {
      "epoch": 1.6105996730620322,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007654843134183438,
      "loss": 0.7333,
      "step": 18720
    },
    {
      "epoch": 1.6106857093693538,
      "grad_norm": 1.125,
      "learning_rate": 0.00076546139894126,
      "loss": 1.27,
      "step": 18721
    },
    {
      "epoch": 1.6107717456766757,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000765438483687753,
      "loss": 1.2422,
      "step": 18722
    },
    {
      "epoch": 1.6108577819839973,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007654155676578901,
      "loss": 1.7997,
      "step": 18723
    },
    {
      "epoch": 1.610943818291319,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007653926508517383,
      "loss": 1.6819,
      "step": 18724
    },
    {
      "epoch": 1.6110298545986406,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007653697332693646,
      "loss": 1.0068,
      "step": 18725
    },
    {
      "epoch": 1.6111158909059622,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007653468149108363,
      "loss": 1.2961,
      "step": 18726
    },
    {
      "epoch": 1.611201927213284,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007653238957762198,
      "loss": 1.1164,
      "step": 18727
    },
    {
      "epoch": 1.6112879635206057,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007653009758655828,
      "loss": 1.6517,
      "step": 18728
    },
    {
      "epoch": 1.6113739998279275,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007652780551789919,
      "loss": 0.7686,
      "step": 18729
    },
    {
      "epoch": 1.6114600361352491,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007652551337165144,
      "loss": 1.6519,
      "step": 18730
    },
    {
      "epoch": 1.6115460724425708,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007652322114782171,
      "loss": 0.9691,
      "step": 18731
    },
    {
      "epoch": 1.6116321087498924,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007652092884641671,
      "loss": 2.0549,
      "step": 18732
    },
    {
      "epoch": 1.611718145057214,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007651863646744314,
      "loss": 1.145,
      "step": 18733
    },
    {
      "epoch": 1.6118041813645358,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007651634401090775,
      "loss": 1.1579,
      "step": 18734
    },
    {
      "epoch": 1.6118902176718575,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007651405147681721,
      "loss": 0.9893,
      "step": 18735
    },
    {
      "epoch": 1.6119762539791793,
      "grad_norm": 0.953125,
      "learning_rate": 0.000765117588651782,
      "loss": 1.6611,
      "step": 18736
    },
    {
      "epoch": 1.612062290286501,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007650946617599745,
      "loss": 1.5216,
      "step": 18737
    },
    {
      "epoch": 1.6121483265938226,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007650717340928169,
      "loss": 1.0412,
      "step": 18738
    },
    {
      "epoch": 1.6122343629011442,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007650488056503759,
      "loss": 1.4248,
      "step": 18739
    },
    {
      "epoch": 1.6123203992084658,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000765025876432719,
      "loss": 1.5498,
      "step": 18740
    },
    {
      "epoch": 1.6124064355157877,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007650029464399125,
      "loss": 0.728,
      "step": 18741
    },
    {
      "epoch": 1.6124924718231095,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007649800156720242,
      "loss": 1.1071,
      "step": 18742
    },
    {
      "epoch": 1.6125785081304311,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007649570841291208,
      "loss": 1.6783,
      "step": 18743
    },
    {
      "epoch": 1.6126645444377528,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007649341518112694,
      "loss": 1.4897,
      "step": 18744
    },
    {
      "epoch": 1.6127505807450744,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007649112187185372,
      "loss": 1.3711,
      "step": 18745
    },
    {
      "epoch": 1.612836617052396,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0007648882848509912,
      "loss": 0.7123,
      "step": 18746
    },
    {
      "epoch": 1.6129226533597179,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007648653502086986,
      "loss": 2.4554,
      "step": 18747
    },
    {
      "epoch": 1.6130086896670395,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007648424147917261,
      "loss": 1.8812,
      "step": 18748
    },
    {
      "epoch": 1.6130947259743613,
      "grad_norm": 1.25,
      "learning_rate": 0.0007648194786001413,
      "loss": 1.2033,
      "step": 18749
    },
    {
      "epoch": 1.613180762281683,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007647965416340111,
      "loss": 0.7678,
      "step": 18750
    },
    {
      "epoch": 1.6132667985890046,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007647736038934021,
      "loss": 1.0427,
      "step": 18751
    },
    {
      "epoch": 1.6133528348963262,
      "grad_norm": 1.453125,
      "learning_rate": 0.000764750665378382,
      "loss": 1.2301,
      "step": 18752
    },
    {
      "epoch": 1.6134388712036478,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007647277260890178,
      "loss": 1.1911,
      "step": 18753
    },
    {
      "epoch": 1.6135249075109697,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007647047860253765,
      "loss": 0.8176,
      "step": 18754
    },
    {
      "epoch": 1.6136109438182913,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007646818451875251,
      "loss": 1.6087,
      "step": 18755
    },
    {
      "epoch": 1.6136969801256131,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007646589035755307,
      "loss": 1.5302,
      "step": 18756
    },
    {
      "epoch": 1.6137830164329348,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007646359611894607,
      "loss": 0.7562,
      "step": 18757
    },
    {
      "epoch": 1.6138690527402564,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007646130180293818,
      "loss": 1.8281,
      "step": 18758
    },
    {
      "epoch": 1.613955089047578,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007645900740953612,
      "loss": 1.0488,
      "step": 18759
    },
    {
      "epoch": 1.6140411253548996,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007645671293874664,
      "loss": 0.7857,
      "step": 18760
    },
    {
      "epoch": 1.6141271616622215,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007645441839057641,
      "loss": 1.0202,
      "step": 18761
    },
    {
      "epoch": 1.614213197969543,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007645212376503214,
      "loss": 1.1668,
      "step": 18762
    },
    {
      "epoch": 1.614299234276865,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007644982906212056,
      "loss": 1.6047,
      "step": 18763
    },
    {
      "epoch": 1.6143852705841866,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007644753428184837,
      "loss": 0.6887,
      "step": 18764
    },
    {
      "epoch": 1.6144713068915082,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007644523942422231,
      "loss": 2.0113,
      "step": 18765
    },
    {
      "epoch": 1.6145573431988298,
      "grad_norm": 2.296875,
      "learning_rate": 0.0007644294448924902,
      "loss": 1.6711,
      "step": 18766
    },
    {
      "epoch": 1.6146433795061514,
      "grad_norm": 1.328125,
      "learning_rate": 0.000764406494769353,
      "loss": 1.5771,
      "step": 18767
    },
    {
      "epoch": 1.6147294158134733,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007643835438728781,
      "loss": 0.5897,
      "step": 18768
    },
    {
      "epoch": 1.6148154521207951,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007643605922031327,
      "loss": 1.0313,
      "step": 18769
    },
    {
      "epoch": 1.6149014884281168,
      "grad_norm": 1.34375,
      "learning_rate": 0.000764337639760184,
      "loss": 1.1655,
      "step": 18770
    },
    {
      "epoch": 1.6149875247354384,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007643146865440991,
      "loss": 1.7163,
      "step": 18771
    },
    {
      "epoch": 1.61507356104276,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007642917325549453,
      "loss": 1.4057,
      "step": 18772
    },
    {
      "epoch": 1.6151595973500816,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007642687777927895,
      "loss": 0.8571,
      "step": 18773
    },
    {
      "epoch": 1.6152456336574035,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007642458222576988,
      "loss": 1.242,
      "step": 18774
    },
    {
      "epoch": 1.615331669964725,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007642228659497406,
      "loss": 1.0674,
      "step": 18775
    },
    {
      "epoch": 1.615417706272047,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007641999088689819,
      "loss": 0.94,
      "step": 18776
    },
    {
      "epoch": 1.6155037425793686,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007641769510154899,
      "loss": 1.1732,
      "step": 18777
    },
    {
      "epoch": 1.6155897788866902,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007641539923893315,
      "loss": 1.9231,
      "step": 18778
    },
    {
      "epoch": 1.6156758151940118,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007641310329905742,
      "loss": 0.8961,
      "step": 18779
    },
    {
      "epoch": 1.6157618515013334,
      "grad_norm": 0.84375,
      "learning_rate": 0.000764108072819285,
      "loss": 0.7772,
      "step": 18780
    },
    {
      "epoch": 1.6158478878086553,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007640851118755309,
      "loss": 1.04,
      "step": 18781
    },
    {
      "epoch": 1.615933924115977,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0007640621501593794,
      "loss": 0.9027,
      "step": 18782
    },
    {
      "epoch": 1.6160199604232988,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007640391876708974,
      "loss": 1.5273,
      "step": 18783
    },
    {
      "epoch": 1.6161059967306204,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007640162244101521,
      "loss": 2.0596,
      "step": 18784
    },
    {
      "epoch": 1.616192033037942,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007639932603772106,
      "loss": 1.0008,
      "step": 18785
    },
    {
      "epoch": 1.6162780693452636,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007639702955721403,
      "loss": 1.2048,
      "step": 18786
    },
    {
      "epoch": 1.6163641056525853,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007639473299950084,
      "loss": 1.1021,
      "step": 18787
    },
    {
      "epoch": 1.616450141959907,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007639243636458814,
      "loss": 1.1127,
      "step": 18788
    },
    {
      "epoch": 1.6165361782672287,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007639013965248274,
      "loss": 1.3542,
      "step": 18789
    },
    {
      "epoch": 1.6166222145745506,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007638784286319129,
      "loss": 1.1067,
      "step": 18790
    },
    {
      "epoch": 1.6167082508818722,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007638554599672055,
      "loss": 1.0857,
      "step": 18791
    },
    {
      "epoch": 1.6167942871891938,
      "grad_norm": 1.25,
      "learning_rate": 0.0007638324905307721,
      "loss": 2.2859,
      "step": 18792
    },
    {
      "epoch": 1.6168803234965154,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007638095203226801,
      "loss": 1.2282,
      "step": 18793
    },
    {
      "epoch": 1.616966359803837,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007637865493429962,
      "loss": 0.7771,
      "step": 18794
    },
    {
      "epoch": 1.617052396111159,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007637635775917884,
      "loss": 0.9145,
      "step": 18795
    },
    {
      "epoch": 1.6171384324184808,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007637406050691233,
      "loss": 1.3561,
      "step": 18796
    },
    {
      "epoch": 1.6172244687258024,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000763717631775068,
      "loss": 1.3106,
      "step": 18797
    },
    {
      "epoch": 1.617310505033124,
      "grad_norm": 1.375,
      "learning_rate": 0.0007636946577096902,
      "loss": 1.5839,
      "step": 18798
    },
    {
      "epoch": 1.6173965413404456,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007636716828730567,
      "loss": 1.3816,
      "step": 18799
    },
    {
      "epoch": 1.6174825776477673,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0007636487072652347,
      "loss": 1.0928,
      "step": 18800
    },
    {
      "epoch": 1.6175686139550889,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007636257308862917,
      "loss": 0.9914,
      "step": 18801
    },
    {
      "epoch": 1.6176546502624107,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007636027537362946,
      "loss": 1.1416,
      "step": 18802
    },
    {
      "epoch": 1.6177406865697326,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007635797758153107,
      "loss": 1.2936,
      "step": 18803
    },
    {
      "epoch": 1.6178267228770542,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0007635567971234072,
      "loss": 0.526,
      "step": 18804
    },
    {
      "epoch": 1.6179127591843758,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007635338176606515,
      "loss": 1.2613,
      "step": 18805
    },
    {
      "epoch": 1.6179987954916975,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007635108374271105,
      "loss": 1.6528,
      "step": 18806
    },
    {
      "epoch": 1.618084831799019,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007634878564228516,
      "loss": 1.423,
      "step": 18807
    },
    {
      "epoch": 1.618170868106341,
      "grad_norm": 1.828125,
      "learning_rate": 0.0007634648746479418,
      "loss": 1.7148,
      "step": 18808
    },
    {
      "epoch": 1.6182569044136625,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007634418921024487,
      "loss": 1.0007,
      "step": 18809
    },
    {
      "epoch": 1.6183429407209844,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007634189087864392,
      "loss": 1.1536,
      "step": 18810
    },
    {
      "epoch": 1.618428977028306,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007633959246999807,
      "loss": 1.7256,
      "step": 18811
    },
    {
      "epoch": 1.6185150133356276,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007633729398431402,
      "loss": 1.2542,
      "step": 18812
    },
    {
      "epoch": 1.6186010496429493,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007633499542159852,
      "loss": 1.999,
      "step": 18813
    },
    {
      "epoch": 1.6186870859502709,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007633269678185827,
      "loss": 1.104,
      "step": 18814
    },
    {
      "epoch": 1.6187731222575927,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007633039806510001,
      "loss": 1.6787,
      "step": 18815
    },
    {
      "epoch": 1.6188591585649144,
      "grad_norm": 1.125,
      "learning_rate": 0.0007632809927133045,
      "loss": 0.9066,
      "step": 18816
    },
    {
      "epoch": 1.6189451948722362,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007632580040055633,
      "loss": 2.244,
      "step": 18817
    },
    {
      "epoch": 1.6190312311795578,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007632350145278437,
      "loss": 1.384,
      "step": 18818
    },
    {
      "epoch": 1.6191172674868795,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007632120242802127,
      "loss": 1.1017,
      "step": 18819
    },
    {
      "epoch": 1.619203303794201,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007631890332627378,
      "loss": 1.9471,
      "step": 18820
    },
    {
      "epoch": 1.6192893401015227,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007631660414754861,
      "loss": 1.8457,
      "step": 18821
    },
    {
      "epoch": 1.6193753764088445,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000763143048918525,
      "loss": 1.7936,
      "step": 18822
    },
    {
      "epoch": 1.6194614127161662,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007631200555919217,
      "loss": 2.231,
      "step": 18823
    },
    {
      "epoch": 1.619547449023488,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007630970614957433,
      "loss": 1.0984,
      "step": 18824
    },
    {
      "epoch": 1.6196334853308096,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007630740666300572,
      "loss": 1.4668,
      "step": 18825
    },
    {
      "epoch": 1.6197195216381313,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007630510709949308,
      "loss": 1.0966,
      "step": 18826
    },
    {
      "epoch": 1.619805557945453,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007630280745904307,
      "loss": 1.4333,
      "step": 18827
    },
    {
      "epoch": 1.6198915942527745,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007630050774166251,
      "loss": 1.3909,
      "step": 18828
    },
    {
      "epoch": 1.6199776305600964,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007629820794735807,
      "loss": 1.4033,
      "step": 18829
    },
    {
      "epoch": 1.6200636668674182,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007629590807613647,
      "loss": 1.3702,
      "step": 18830
    },
    {
      "epoch": 1.6201497031747398,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007629360812800447,
      "loss": 1.4877,
      "step": 18831
    },
    {
      "epoch": 1.6202357394820615,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007629130810296878,
      "loss": 0.8634,
      "step": 18832
    },
    {
      "epoch": 1.620321775789383,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007628900800103613,
      "loss": 0.8055,
      "step": 18833
    },
    {
      "epoch": 1.6204078120967047,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007628670782221323,
      "loss": 0.7817,
      "step": 18834
    },
    {
      "epoch": 1.6204938484040265,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007628440756650683,
      "loss": 0.7638,
      "step": 18835
    },
    {
      "epoch": 1.6205798847113482,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007628210723392365,
      "loss": 1.4832,
      "step": 18836
    },
    {
      "epoch": 1.62066592101867,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007627980682447042,
      "loss": 1.1908,
      "step": 18837
    },
    {
      "epoch": 1.6207519573259916,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007627750633815386,
      "loss": 0.8674,
      "step": 18838
    },
    {
      "epoch": 1.6208379936333133,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000762752057749807,
      "loss": 1.2084,
      "step": 18839
    },
    {
      "epoch": 1.620924029940635,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007627290513495768,
      "loss": 1.6071,
      "step": 18840
    },
    {
      "epoch": 1.6210100662479565,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007627060441809153,
      "loss": 1.373,
      "step": 18841
    },
    {
      "epoch": 1.6210961025552784,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007626830362438897,
      "loss": 1.7167,
      "step": 18842
    },
    {
      "epoch": 1.6211821388626,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0007626600275385672,
      "loss": 2.2541,
      "step": 18843
    },
    {
      "epoch": 1.6212681751699218,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007626370180650152,
      "loss": 0.9246,
      "step": 18844
    },
    {
      "epoch": 1.6213542114772435,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000762614007823301,
      "loss": 1.0958,
      "step": 18845
    },
    {
      "epoch": 1.621440247784565,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007625909968134919,
      "loss": 0.6242,
      "step": 18846
    },
    {
      "epoch": 1.6215262840918867,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007625679850356554,
      "loss": 1.5468,
      "step": 18847
    },
    {
      "epoch": 1.6216123203992083,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007625449724898586,
      "loss": 1.573,
      "step": 18848
    },
    {
      "epoch": 1.6216983567065302,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007625219591761686,
      "loss": 1.468,
      "step": 18849
    },
    {
      "epoch": 1.6217843930138518,
      "grad_norm": 1.125,
      "learning_rate": 0.0007624989450946529,
      "loss": 1.143,
      "step": 18850
    },
    {
      "epoch": 1.6218704293211736,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007624759302453791,
      "loss": 1.2006,
      "step": 18851
    },
    {
      "epoch": 1.6219564656284953,
      "grad_norm": 3.609375,
      "learning_rate": 0.0007624529146284141,
      "loss": 1.9699,
      "step": 18852
    },
    {
      "epoch": 1.622042501935817,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007624298982438254,
      "loss": 1.2766,
      "step": 18853
    },
    {
      "epoch": 1.6221285382431385,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007624068810916802,
      "loss": 0.9074,
      "step": 18854
    },
    {
      "epoch": 1.6222145745504601,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007623838631720462,
      "loss": 1.4462,
      "step": 18855
    },
    {
      "epoch": 1.622300610857782,
      "grad_norm": 1.359375,
      "learning_rate": 0.00076236084448499,
      "loss": 1.9875,
      "step": 18856
    },
    {
      "epoch": 1.6223866471651038,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007623378250305795,
      "loss": 1.1047,
      "step": 18857
    },
    {
      "epoch": 1.6224726834724255,
      "grad_norm": 1.40625,
      "learning_rate": 0.000762314804808882,
      "loss": 2.1108,
      "step": 18858
    },
    {
      "epoch": 1.622558719779747,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007622917838199648,
      "loss": 0.8409,
      "step": 18859
    },
    {
      "epoch": 1.6226447560870687,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007622687620638948,
      "loss": 1.8393,
      "step": 18860
    },
    {
      "epoch": 1.6227307923943903,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007622457395407401,
      "loss": 0.9488,
      "step": 18861
    },
    {
      "epoch": 1.6228168287017122,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007622227162505672,
      "loss": 1.3986,
      "step": 18862
    },
    {
      "epoch": 1.6229028650090338,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007621996921934442,
      "loss": 1.0749,
      "step": 18863
    },
    {
      "epoch": 1.6229889013163556,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007621766673694377,
      "loss": 2.0643,
      "step": 18864
    },
    {
      "epoch": 1.6230749376236773,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007621536417786159,
      "loss": 1.0447,
      "step": 18865
    },
    {
      "epoch": 1.623160973930999,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007621306154210455,
      "loss": 0.8251,
      "step": 18866
    },
    {
      "epoch": 1.6232470102383205,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000762107588296794,
      "loss": 0.5279,
      "step": 18867
    },
    {
      "epoch": 1.6233330465456421,
      "grad_norm": 1.375,
      "learning_rate": 0.0007620845604059288,
      "loss": 1.2031,
      "step": 18868
    },
    {
      "epoch": 1.623419082852964,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007620615317485171,
      "loss": 1.1196,
      "step": 18869
    },
    {
      "epoch": 1.6235051191602856,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007620385023246266,
      "loss": 1.2187,
      "step": 18870
    },
    {
      "epoch": 1.6235911554676075,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007620154721343243,
      "loss": 1.186,
      "step": 18871
    },
    {
      "epoch": 1.623677191774929,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007619924411776777,
      "loss": 1.2887,
      "step": 18872
    },
    {
      "epoch": 1.6237632280822507,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007619694094547542,
      "loss": 1.32,
      "step": 18873
    },
    {
      "epoch": 1.6238492643895723,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007619463769656213,
      "loss": 1.5651,
      "step": 18874
    },
    {
      "epoch": 1.623935300696894,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000761923343710346,
      "loss": 1.149,
      "step": 18875
    },
    {
      "epoch": 1.6240213370042158,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007619003096889959,
      "loss": 1.9316,
      "step": 18876
    },
    {
      "epoch": 1.6241073733115374,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007618772749016383,
      "loss": 1.0579,
      "step": 18877
    },
    {
      "epoch": 1.6241934096188593,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007618542393483406,
      "loss": 1.224,
      "step": 18878
    },
    {
      "epoch": 1.624279445926181,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007618312030291703,
      "loss": 1.6316,
      "step": 18879
    },
    {
      "epoch": 1.6243654822335025,
      "grad_norm": 1.5,
      "learning_rate": 0.0007618081659441947,
      "loss": 2.1298,
      "step": 18880
    },
    {
      "epoch": 1.6244515185408241,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007617851280934809,
      "loss": 1.8117,
      "step": 18881
    },
    {
      "epoch": 1.6245375548481458,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007617620894770966,
      "loss": 0.782,
      "step": 18882
    },
    {
      "epoch": 1.6246235911554676,
      "grad_norm": 1.03125,
      "learning_rate": 0.000761739050095109,
      "loss": 0.9822,
      "step": 18883
    },
    {
      "epoch": 1.6247096274627895,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007617160099475857,
      "loss": 1.0074,
      "step": 18884
    },
    {
      "epoch": 1.624795663770111,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007616929690345941,
      "loss": 0.8444,
      "step": 18885
    },
    {
      "epoch": 1.6248817000774327,
      "grad_norm": 1.375,
      "learning_rate": 0.0007616699273562012,
      "loss": 1.145,
      "step": 18886
    },
    {
      "epoch": 1.6249677363847543,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007616468849124749,
      "loss": 1.0229,
      "step": 18887
    },
    {
      "epoch": 1.625053772692076,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007616238417034822,
      "loss": 1.6858,
      "step": 18888
    },
    {
      "epoch": 1.6251398089993976,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007616007977292907,
      "loss": 1.5647,
      "step": 18889
    },
    {
      "epoch": 1.6252258453067194,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007615777529899676,
      "loss": 1.2818,
      "step": 18890
    },
    {
      "epoch": 1.6253118816140413,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007615547074855807,
      "loss": 2.026,
      "step": 18891
    },
    {
      "epoch": 1.625397917921363,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007615316612161971,
      "loss": 1.6449,
      "step": 18892
    },
    {
      "epoch": 1.6254839542286845,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007615086141818841,
      "loss": 0.968,
      "step": 18893
    },
    {
      "epoch": 1.6255699905360061,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007614855663827094,
      "loss": 1.2424,
      "step": 18894
    },
    {
      "epoch": 1.6256560268433278,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0007614625178187402,
      "loss": 1.7982,
      "step": 18895
    },
    {
      "epoch": 1.6257420631506496,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007614394684900441,
      "loss": 0.8019,
      "step": 18896
    },
    {
      "epoch": 1.6258280994579712,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007614164183966881,
      "loss": 1.6197,
      "step": 18897
    },
    {
      "epoch": 1.625914135765293,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007613933675387402,
      "loss": 1.5814,
      "step": 18898
    },
    {
      "epoch": 1.6260001720726147,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007613703159162675,
      "loss": 1.4552,
      "step": 18899
    },
    {
      "epoch": 1.6260862083799363,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007613472635293373,
      "loss": 1.3432,
      "step": 18900
    },
    {
      "epoch": 1.626172244687258,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007613242103780174,
      "loss": 1.6143,
      "step": 18901
    },
    {
      "epoch": 1.6262582809945796,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007613011564623749,
      "loss": 1.139,
      "step": 18902
    },
    {
      "epoch": 1.6263443173019014,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007612781017824774,
      "loss": 1.1446,
      "step": 18903
    },
    {
      "epoch": 1.626430353609223,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007612550463383923,
      "loss": 1.0193,
      "step": 18904
    },
    {
      "epoch": 1.626516389916545,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007612319901301867,
      "loss": 1.0992,
      "step": 18905
    },
    {
      "epoch": 1.6266024262238665,
      "grad_norm": 3.59375,
      "learning_rate": 0.0007612089331579286,
      "loss": 1.3058,
      "step": 18906
    },
    {
      "epoch": 1.6266884625311882,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007611858754216852,
      "loss": 1.018,
      "step": 18907
    },
    {
      "epoch": 1.6267744988385098,
      "grad_norm": 0.875,
      "learning_rate": 0.0007611628169215239,
      "loss": 1.0396,
      "step": 18908
    },
    {
      "epoch": 1.6268605351458314,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007611397576575121,
      "loss": 1.9198,
      "step": 18909
    },
    {
      "epoch": 1.6269465714531532,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007611166976297173,
      "loss": 0.9664,
      "step": 18910
    },
    {
      "epoch": 1.6270326077604749,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007610936368382069,
      "loss": 0.8163,
      "step": 18911
    },
    {
      "epoch": 1.6271186440677967,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007610705752830484,
      "loss": 1.0818,
      "step": 18912
    },
    {
      "epoch": 1.6272046803751183,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007610475129643093,
      "loss": 0.815,
      "step": 18913
    },
    {
      "epoch": 1.62729071668244,
      "grad_norm": 1.640625,
      "learning_rate": 0.000761024449882057,
      "loss": 1.5342,
      "step": 18914
    },
    {
      "epoch": 1.6273767529897616,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007610013860363589,
      "loss": 1.2524,
      "step": 18915
    },
    {
      "epoch": 1.6274627892970832,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007609783214272825,
      "loss": 1.3827,
      "step": 18916
    },
    {
      "epoch": 1.627548825604405,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007609552560548954,
      "loss": 1.34,
      "step": 18917
    },
    {
      "epoch": 1.627634861911727,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007609321899192647,
      "loss": 0.7903,
      "step": 18918
    },
    {
      "epoch": 1.6277208982190485,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007609091230204582,
      "loss": 1.1799,
      "step": 18919
    },
    {
      "epoch": 1.6278069345263702,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007608860553585433,
      "loss": 1.5577,
      "step": 18920
    },
    {
      "epoch": 1.6278929708336918,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007608629869335873,
      "loss": 1.6279,
      "step": 18921
    },
    {
      "epoch": 1.6279790071410134,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007608399177456579,
      "loss": 1.2192,
      "step": 18922
    },
    {
      "epoch": 1.6280650434483352,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007608168477948226,
      "loss": 1.543,
      "step": 18923
    },
    {
      "epoch": 1.6281510797556569,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007607937770811486,
      "loss": 1.1928,
      "step": 18924
    },
    {
      "epoch": 1.6282371160629787,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007607707056047035,
      "loss": 0.555,
      "step": 18925
    },
    {
      "epoch": 1.6283231523703003,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000760747633365555,
      "loss": 1.0122,
      "step": 18926
    },
    {
      "epoch": 1.628409188677622,
      "grad_norm": 1.4375,
      "learning_rate": 0.00076072456036377,
      "loss": 2.0956,
      "step": 18927
    },
    {
      "epoch": 1.6284952249849436,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007607014865994167,
      "loss": 1.101,
      "step": 18928
    },
    {
      "epoch": 1.6285812612922652,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007606784120725623,
      "loss": 1.8776,
      "step": 18929
    },
    {
      "epoch": 1.628667297599587,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007606553367832741,
      "loss": 2.2974,
      "step": 18930
    },
    {
      "epoch": 1.6287533339069087,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007606322607316196,
      "loss": 0.9014,
      "step": 18931
    },
    {
      "epoch": 1.6288393702142305,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007606091839176666,
      "loss": 1.4897,
      "step": 18932
    },
    {
      "epoch": 1.6289254065215522,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007605861063414824,
      "loss": 1.3545,
      "step": 18933
    },
    {
      "epoch": 1.6290114428288738,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007605630280031345,
      "loss": 0.7908,
      "step": 18934
    },
    {
      "epoch": 1.6290974791361954,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007605399489026904,
      "loss": 1.5345,
      "step": 18935
    },
    {
      "epoch": 1.629183515443517,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007605168690402175,
      "loss": 1.1403,
      "step": 18936
    },
    {
      "epoch": 1.6292695517508389,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007604937884157836,
      "loss": 1.2269,
      "step": 18937
    },
    {
      "epoch": 1.6293555880581605,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007604707070294561,
      "loss": 0.9901,
      "step": 18938
    },
    {
      "epoch": 1.6294416243654823,
      "grad_norm": 1.734375,
      "learning_rate": 0.0007604476248813023,
      "loss": 1.6032,
      "step": 18939
    },
    {
      "epoch": 1.629527660672804,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00076042454197139,
      "loss": 1.4139,
      "step": 18940
    },
    {
      "epoch": 1.6296136969801256,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007604014582997864,
      "loss": 2.0995,
      "step": 18941
    },
    {
      "epoch": 1.6296997332874472,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007603783738665593,
      "loss": 1.2095,
      "step": 18942
    },
    {
      "epoch": 1.6297857695947688,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007603552886717759,
      "loss": 1.0851,
      "step": 18943
    },
    {
      "epoch": 1.6298718059020907,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007603322027155042,
      "loss": 1.6377,
      "step": 18944
    },
    {
      "epoch": 1.6299578422094125,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007603091159978113,
      "loss": 1.2273,
      "step": 18945
    },
    {
      "epoch": 1.6300438785167342,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007602860285187647,
      "loss": 1.1977,
      "step": 18946
    },
    {
      "epoch": 1.6301299148240558,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007602629402784323,
      "loss": 1.8672,
      "step": 18947
    },
    {
      "epoch": 1.6302159511313774,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007602398512768815,
      "loss": 0.6632,
      "step": 18948
    },
    {
      "epoch": 1.630301987438699,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007602167615141795,
      "loss": 1.4565,
      "step": 18949
    },
    {
      "epoch": 1.6303880237460209,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007601936709903943,
      "loss": 1.8859,
      "step": 18950
    },
    {
      "epoch": 1.6304740600533425,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007601705797055931,
      "loss": 0.6486,
      "step": 18951
    },
    {
      "epoch": 1.6305600963606643,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007601474876598436,
      "loss": 1.0437,
      "step": 18952
    },
    {
      "epoch": 1.630646132667986,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007601243948532132,
      "loss": 1.0652,
      "step": 18953
    },
    {
      "epoch": 1.6307321689753076,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007601013012857696,
      "loss": 1.7732,
      "step": 18954
    },
    {
      "epoch": 1.6308182052826292,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007600782069575803,
      "loss": 1.073,
      "step": 18955
    },
    {
      "epoch": 1.6309042415899508,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007600551118687127,
      "loss": 1.3544,
      "step": 18956
    },
    {
      "epoch": 1.6309902778972727,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007600320160192347,
      "loss": 1.528,
      "step": 18957
    },
    {
      "epoch": 1.6310763142045943,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007600089194092134,
      "loss": 0.8589,
      "step": 18958
    },
    {
      "epoch": 1.6311623505119162,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007599858220387167,
      "loss": 2.0536,
      "step": 18959
    },
    {
      "epoch": 1.6312483868192378,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000759962723907812,
      "loss": 1.8461,
      "step": 18960
    },
    {
      "epoch": 1.6313344231265594,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007599396250165669,
      "loss": 0.9642,
      "step": 18961
    },
    {
      "epoch": 1.631420459433881,
      "grad_norm": 1.125,
      "learning_rate": 0.0007599165253650491,
      "loss": 0.9309,
      "step": 18962
    },
    {
      "epoch": 1.6315064957412027,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007598934249533258,
      "loss": 1.0848,
      "step": 18963
    },
    {
      "epoch": 1.6315925320485245,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007598703237814649,
      "loss": 1.5375,
      "step": 18964
    },
    {
      "epoch": 1.6316785683558461,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007598472218495338,
      "loss": 1.1773,
      "step": 18965
    },
    {
      "epoch": 1.631764604663168,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007598241191576001,
      "loss": 0.9461,
      "step": 18966
    },
    {
      "epoch": 1.6318506409704896,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007598010157057315,
      "loss": 0.995,
      "step": 18967
    },
    {
      "epoch": 1.6319366772778112,
      "grad_norm": 5.0,
      "learning_rate": 0.0007597779114939952,
      "loss": 1.3142,
      "step": 18968
    },
    {
      "epoch": 1.6320227135851328,
      "grad_norm": 1.125,
      "learning_rate": 0.0007597548065224593,
      "loss": 1.0408,
      "step": 18969
    },
    {
      "epoch": 1.6321087498924545,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007597317007911911,
      "loss": 1.8124,
      "step": 18970
    },
    {
      "epoch": 1.6321947861997763,
      "grad_norm": 1.609375,
      "learning_rate": 0.000759708594300258,
      "loss": 1.8489,
      "step": 18971
    },
    {
      "epoch": 1.6322808225070982,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007596854870497279,
      "loss": 0.9127,
      "step": 18972
    },
    {
      "epoch": 1.6323668588144198,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007596623790396681,
      "loss": 1.1271,
      "step": 18973
    },
    {
      "epoch": 1.6324528951217414,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0007596392702701464,
      "loss": 1.8768,
      "step": 18974
    },
    {
      "epoch": 1.632538931429063,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007596161607412304,
      "loss": 1.2048,
      "step": 18975
    },
    {
      "epoch": 1.6326249677363847,
      "grad_norm": 1.5,
      "learning_rate": 0.0007595930504529876,
      "loss": 0.9446,
      "step": 18976
    },
    {
      "epoch": 1.6327110040437063,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0007595699394054857,
      "loss": 0.923,
      "step": 18977
    },
    {
      "epoch": 1.6327970403510281,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007595468275987923,
      "loss": 1.2692,
      "step": 18978
    },
    {
      "epoch": 1.63288307665835,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007595237150329747,
      "loss": 1.3138,
      "step": 18979
    },
    {
      "epoch": 1.6329691129656716,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007595006017081006,
      "loss": 1.1602,
      "step": 18980
    },
    {
      "epoch": 1.6330551492729932,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007594774876242379,
      "loss": 1.5985,
      "step": 18981
    },
    {
      "epoch": 1.6331411855803148,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000759454372781454,
      "loss": 1.2858,
      "step": 18982
    },
    {
      "epoch": 1.6332272218876365,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007594312571798166,
      "loss": 1.5079,
      "step": 18983
    },
    {
      "epoch": 1.6333132581949583,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007594081408193931,
      "loss": 1.7409,
      "step": 18984
    },
    {
      "epoch": 1.63339929450228,
      "grad_norm": 2.0625,
      "learning_rate": 0.0007593850237002513,
      "loss": 1.1158,
      "step": 18985
    },
    {
      "epoch": 1.6334853308096018,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007593619058224588,
      "loss": 0.9305,
      "step": 18986
    },
    {
      "epoch": 1.6335713671169234,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007593387871860831,
      "loss": 1.6837,
      "step": 18987
    },
    {
      "epoch": 1.633657403424245,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007593156677911919,
      "loss": 0.8125,
      "step": 18988
    },
    {
      "epoch": 1.6337434397315667,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007592925476378527,
      "loss": 1.1429,
      "step": 18989
    },
    {
      "epoch": 1.6338294760388883,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007592694267261333,
      "loss": 1.5911,
      "step": 18990
    },
    {
      "epoch": 1.6339155123462101,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007592463050561014,
      "loss": 1.649,
      "step": 18991
    },
    {
      "epoch": 1.6340015486535318,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007592231826278242,
      "loss": 1.6285,
      "step": 18992
    },
    {
      "epoch": 1.6340875849608536,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007592000594413697,
      "loss": 1.3435,
      "step": 18993
    },
    {
      "epoch": 1.6341736212681752,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007591769354968056,
      "loss": 1.1168,
      "step": 18994
    },
    {
      "epoch": 1.6342596575754968,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007591538107941992,
      "loss": 1.0991,
      "step": 18995
    },
    {
      "epoch": 1.6343456938828185,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007591306853336184,
      "loss": 1.4233,
      "step": 18996
    },
    {
      "epoch": 1.63443173019014,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007591075591151308,
      "loss": 0.7896,
      "step": 18997
    },
    {
      "epoch": 1.634517766497462,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007590844321388037,
      "loss": 1.5139,
      "step": 18998
    },
    {
      "epoch": 1.6346038028047836,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007590613044047053,
      "loss": 1.5906,
      "step": 18999
    },
    {
      "epoch": 1.6346898391121054,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007590381759129029,
      "loss": 0.8573,
      "step": 19000
    },
    {
      "epoch": 1.634775875419427,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007590150466634643,
      "loss": 1.5245,
      "step": 19001
    },
    {
      "epoch": 1.6348619117267487,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000758991916656457,
      "loss": 1.6342,
      "step": 19002
    },
    {
      "epoch": 1.6349479480340703,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007589687858919486,
      "loss": 1.001,
      "step": 19003
    },
    {
      "epoch": 1.635033984341392,
      "grad_norm": 1.0,
      "learning_rate": 0.000758945654370007,
      "loss": 0.7174,
      "step": 19004
    },
    {
      "epoch": 1.6351200206487138,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007589225220906996,
      "loss": 1.8034,
      "step": 19005
    },
    {
      "epoch": 1.6352060569560356,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007588993890540943,
      "loss": 1.7907,
      "step": 19006
    },
    {
      "epoch": 1.6352920932633572,
      "grad_norm": 1.5,
      "learning_rate": 0.0007588762552602586,
      "loss": 1.9399,
      "step": 19007
    },
    {
      "epoch": 1.6353781295706789,
      "grad_norm": 1.375,
      "learning_rate": 0.0007588531207092602,
      "loss": 1.537,
      "step": 19008
    },
    {
      "epoch": 1.6354641658780005,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007588299854011667,
      "loss": 1.206,
      "step": 19009
    },
    {
      "epoch": 1.635550202185322,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007588068493360458,
      "loss": 0.7667,
      "step": 19010
    },
    {
      "epoch": 1.635636238492644,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007587837125139654,
      "loss": 2.0085,
      "step": 19011
    },
    {
      "epoch": 1.6357222747999656,
      "grad_norm": 1.25,
      "learning_rate": 0.0007587605749349929,
      "loss": 1.6005,
      "step": 19012
    },
    {
      "epoch": 1.6358083111072874,
      "grad_norm": 1.109375,
      "learning_rate": 0.000758737436599196,
      "loss": 1.2501,
      "step": 19013
    },
    {
      "epoch": 1.635894347414609,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007587142975066423,
      "loss": 1.4653,
      "step": 19014
    },
    {
      "epoch": 1.6359803837219307,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007586911576573999,
      "loss": 1.8275,
      "step": 19015
    },
    {
      "epoch": 1.6360664200292523,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007586680170515361,
      "loss": 1.1555,
      "step": 19016
    },
    {
      "epoch": 1.636152456336574,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0007586448756891185,
      "loss": 1.5674,
      "step": 19017
    },
    {
      "epoch": 1.6362384926438958,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000758621733570215,
      "loss": 1.2403,
      "step": 19018
    },
    {
      "epoch": 1.6363245289512174,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007585985906948933,
      "loss": 0.5451,
      "step": 19019
    },
    {
      "epoch": 1.6364105652585392,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007585754470632209,
      "loss": 1.0807,
      "step": 19020
    },
    {
      "epoch": 1.6364966015658609,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007585523026752656,
      "loss": 1.1682,
      "step": 19021
    },
    {
      "epoch": 1.6365826378731825,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007585291575310952,
      "loss": 1.1514,
      "step": 19022
    },
    {
      "epoch": 1.636668674180504,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007585060116307772,
      "loss": 1.4971,
      "step": 19023
    },
    {
      "epoch": 1.6367547104878257,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007584828649743794,
      "loss": 0.8908,
      "step": 19024
    },
    {
      "epoch": 1.6368407467951476,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007584597175619697,
      "loss": 0.9162,
      "step": 19025
    },
    {
      "epoch": 1.6369267831024692,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007584365693936154,
      "loss": 1.9083,
      "step": 19026
    },
    {
      "epoch": 1.637012819409791,
      "grad_norm": 1.0,
      "learning_rate": 0.0007584134204693845,
      "loss": 1.3441,
      "step": 19027
    },
    {
      "epoch": 1.6370988557171127,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007583902707893445,
      "loss": 1.7303,
      "step": 19028
    },
    {
      "epoch": 1.6371848920244343,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007583671203535632,
      "loss": 1.921,
      "step": 19029
    },
    {
      "epoch": 1.637270928331756,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007583439691621084,
      "loss": 1.8823,
      "step": 19030
    },
    {
      "epoch": 1.6373569646390775,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007583208172150476,
      "loss": 1.3791,
      "step": 19031
    },
    {
      "epoch": 1.6374430009463994,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007582976645124488,
      "loss": 0.7497,
      "step": 19032
    },
    {
      "epoch": 1.6375290372537212,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007582745110543794,
      "loss": 1.5579,
      "step": 19033
    },
    {
      "epoch": 1.6376150735610429,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007582513568409076,
      "loss": 1.3392,
      "step": 19034
    },
    {
      "epoch": 1.6377011098683645,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007582282018721006,
      "loss": 1.012,
      "step": 19035
    },
    {
      "epoch": 1.637787146175686,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007582050461480263,
      "loss": 1.969,
      "step": 19036
    },
    {
      "epoch": 1.6378731824830077,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0007581818896687524,
      "loss": 0.7461,
      "step": 19037
    },
    {
      "epoch": 1.6379592187903296,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000758158732434347,
      "loss": 1.3623,
      "step": 19038
    },
    {
      "epoch": 1.6380452550976512,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007581355744448773,
      "loss": 0.8287,
      "step": 19039
    },
    {
      "epoch": 1.638131291404973,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007581124157004112,
      "loss": 1.468,
      "step": 19040
    },
    {
      "epoch": 1.6382173277122947,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007580892562010167,
      "loss": 0.77,
      "step": 19041
    },
    {
      "epoch": 1.6383033640196163,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007580660959467611,
      "loss": 1.0261,
      "step": 19042
    },
    {
      "epoch": 1.638389400326938,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007580429349377122,
      "loss": 1.3562,
      "step": 19043
    },
    {
      "epoch": 1.6384754366342595,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007580197731739383,
      "loss": 0.8471,
      "step": 19044
    },
    {
      "epoch": 1.6385614729415814,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007579966106555067,
      "loss": 1.1622,
      "step": 19045
    },
    {
      "epoch": 1.638647509248903,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007579734473824851,
      "loss": 1.1151,
      "step": 19046
    },
    {
      "epoch": 1.6387335455562249,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007579502833549409,
      "loss": 1.5852,
      "step": 19047
    },
    {
      "epoch": 1.6388195818635465,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007579271185729428,
      "loss": 1.9913,
      "step": 19048
    },
    {
      "epoch": 1.638905618170868,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007579039530365581,
      "loss": 1.2787,
      "step": 19049
    },
    {
      "epoch": 1.6389916544781897,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007578807867458544,
      "loss": 0.7478,
      "step": 19050
    },
    {
      "epoch": 1.6390776907855114,
      "grad_norm": 1.125,
      "learning_rate": 0.0007578576197008992,
      "loss": 1.0445,
      "step": 19051
    },
    {
      "epoch": 1.6391637270928332,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000757834451901761,
      "loss": 1.3715,
      "step": 19052
    },
    {
      "epoch": 1.6392497634001548,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007578112833485071,
      "loss": 1.0219,
      "step": 19053
    },
    {
      "epoch": 1.6393357997074767,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007577881140412053,
      "loss": 1.2182,
      "step": 19054
    },
    {
      "epoch": 1.6394218360147983,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007577649439799235,
      "loss": 1.5023,
      "step": 19055
    },
    {
      "epoch": 1.63950787232212,
      "grad_norm": 1.5,
      "learning_rate": 0.0007577417731647293,
      "loss": 1.2446,
      "step": 19056
    },
    {
      "epoch": 1.6395939086294415,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007577186015956905,
      "loss": 1.1328,
      "step": 19057
    },
    {
      "epoch": 1.6396799449367632,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007576954292728749,
      "loss": 1.0977,
      "step": 19058
    },
    {
      "epoch": 1.639765981244085,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007576722561963504,
      "loss": 1.4389,
      "step": 19059
    },
    {
      "epoch": 1.6398520175514069,
      "grad_norm": 1.125,
      "learning_rate": 0.0007576490823661846,
      "loss": 1.6985,
      "step": 19060
    },
    {
      "epoch": 1.6399380538587285,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007576259077824454,
      "loss": 1.4691,
      "step": 19061
    },
    {
      "epoch": 1.64002409016605,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007576027324452004,
      "loss": 1.3787,
      "step": 19062
    },
    {
      "epoch": 1.6401101264733717,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007575795563545176,
      "loss": 1.7001,
      "step": 19063
    },
    {
      "epoch": 1.6401961627806934,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007575563795104648,
      "loss": 0.6553,
      "step": 19064
    },
    {
      "epoch": 1.640282199088015,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007575332019131094,
      "loss": 1.4989,
      "step": 19065
    },
    {
      "epoch": 1.6403682353953368,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007575100235625195,
      "loss": 0.964,
      "step": 19066
    },
    {
      "epoch": 1.6404542717026587,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000757486844458763,
      "loss": 1.1852,
      "step": 19067
    },
    {
      "epoch": 1.6405403080099803,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007574636646019077,
      "loss": 1.6536,
      "step": 19068
    },
    {
      "epoch": 1.640626344317302,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000757440483992021,
      "loss": 1.0065,
      "step": 19069
    },
    {
      "epoch": 1.6407123806246235,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007574173026291711,
      "loss": 1.4222,
      "step": 19070
    },
    {
      "epoch": 1.6407984169319452,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007573941205134256,
      "loss": 2.0256,
      "step": 19071
    },
    {
      "epoch": 1.640884453239267,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007573709376448523,
      "loss": 1.0821,
      "step": 19072
    },
    {
      "epoch": 1.6409704895465886,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007573477540235192,
      "loss": 1.4385,
      "step": 19073
    },
    {
      "epoch": 1.6410565258539105,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007573245696494939,
      "loss": 1.36,
      "step": 19074
    },
    {
      "epoch": 1.641142562161232,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007573013845228443,
      "loss": 1.133,
      "step": 19075
    },
    {
      "epoch": 1.6412285984685537,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007572781986436382,
      "loss": 1.3075,
      "step": 19076
    },
    {
      "epoch": 1.6413146347758754,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007572550120119433,
      "loss": 1.1468,
      "step": 19077
    },
    {
      "epoch": 1.641400671083197,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007572318246278275,
      "loss": 1.3082,
      "step": 19078
    },
    {
      "epoch": 1.6414867073905188,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007572086364913589,
      "loss": 0.6565,
      "step": 19079
    },
    {
      "epoch": 1.6415727436978405,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007571854476026049,
      "loss": 1.112,
      "step": 19080
    },
    {
      "epoch": 1.6416587800051623,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007571622579616335,
      "loss": 0.96,
      "step": 19081
    },
    {
      "epoch": 1.641744816312484,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007571390675685126,
      "loss": 0.9992,
      "step": 19082
    },
    {
      "epoch": 1.6418308526198055,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007571158764233098,
      "loss": 0.6624,
      "step": 19083
    },
    {
      "epoch": 1.6419168889271272,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000757092684526093,
      "loss": 1.6657,
      "step": 19084
    },
    {
      "epoch": 1.6420029252344488,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007570694918769302,
      "loss": 1.1047,
      "step": 19085
    },
    {
      "epoch": 1.6420889615417706,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007570462984758892,
      "loss": 1.6392,
      "step": 19086
    },
    {
      "epoch": 1.6421749978490923,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007570231043230377,
      "loss": 0.8555,
      "step": 19087
    },
    {
      "epoch": 1.6422610341564141,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007569999094184435,
      "loss": 1.1456,
      "step": 19088
    },
    {
      "epoch": 1.6423470704637357,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007569767137621748,
      "loss": 1.1656,
      "step": 19089
    },
    {
      "epoch": 1.6424331067710574,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000756953517354299,
      "loss": 1.8907,
      "step": 19090
    },
    {
      "epoch": 1.642519143078379,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007569303201948842,
      "loss": 0.8562,
      "step": 19091
    },
    {
      "epoch": 1.6426051793857006,
      "grad_norm": 1.765625,
      "learning_rate": 0.0007569071222839983,
      "loss": 1.134,
      "step": 19092
    },
    {
      "epoch": 1.6426912156930225,
      "grad_norm": 0.828125,
      "learning_rate": 0.0007568839236217089,
      "loss": 0.5326,
      "step": 19093
    },
    {
      "epoch": 1.6427772520003443,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007568607242080841,
      "loss": 1.0539,
      "step": 19094
    },
    {
      "epoch": 1.642863288307666,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007568375240431913,
      "loss": 1.3155,
      "step": 19095
    },
    {
      "epoch": 1.6429493246149875,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007568143231270989,
      "loss": 1.3815,
      "step": 19096
    },
    {
      "epoch": 1.6430353609223092,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007567911214598747,
      "loss": 0.9803,
      "step": 19097
    },
    {
      "epoch": 1.6431213972296308,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007567679190415864,
      "loss": 1.2258,
      "step": 19098
    },
    {
      "epoch": 1.6432074335369526,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007567447158723016,
      "loss": 1.7205,
      "step": 19099
    },
    {
      "epoch": 1.6432934698442743,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007567215119520887,
      "loss": 1.333,
      "step": 19100
    },
    {
      "epoch": 1.6433795061515961,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007566983072810152,
      "loss": 1.389,
      "step": 19101
    },
    {
      "epoch": 1.6434655424589177,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007566751018591491,
      "loss": 0.8807,
      "step": 19102
    },
    {
      "epoch": 1.6435515787662394,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007566518956865582,
      "loss": 2.0337,
      "step": 19103
    },
    {
      "epoch": 1.643637615073561,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007566286887633105,
      "loss": 1.1108,
      "step": 19104
    },
    {
      "epoch": 1.6437236513808826,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007566054810894739,
      "loss": 1.6626,
      "step": 19105
    },
    {
      "epoch": 1.6438096876882045,
      "grad_norm": 0.796875,
      "learning_rate": 0.000756582272665116,
      "loss": 1.3107,
      "step": 19106
    },
    {
      "epoch": 1.643895723995526,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007565590634903048,
      "loss": 1.6317,
      "step": 19107
    },
    {
      "epoch": 1.643981760302848,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007565358535651084,
      "loss": 1.4891,
      "step": 19108
    },
    {
      "epoch": 1.6440677966101696,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007565126428895943,
      "loss": 0.7828,
      "step": 19109
    },
    {
      "epoch": 1.6441538329174912,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007564894314638309,
      "loss": 1.806,
      "step": 19110
    },
    {
      "epoch": 1.6442398692248128,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007564662192878855,
      "loss": 1.5346,
      "step": 19111
    },
    {
      "epoch": 1.6443259055321344,
      "grad_norm": 0.875,
      "learning_rate": 0.0007564430063618266,
      "loss": 0.8528,
      "step": 19112
    },
    {
      "epoch": 1.6444119418394563,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007564197926857216,
      "loss": 1.2414,
      "step": 19113
    },
    {
      "epoch": 1.644497978146778,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007563965782596385,
      "loss": 2.0665,
      "step": 19114
    },
    {
      "epoch": 1.6445840144540997,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007563733630836455,
      "loss": 0.892,
      "step": 19115
    },
    {
      "epoch": 1.6446700507614214,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007563501471578102,
      "loss": 1.6634,
      "step": 19116
    },
    {
      "epoch": 1.644756087068743,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007563269304822005,
      "loss": 1.8411,
      "step": 19117
    },
    {
      "epoch": 1.6448421233760646,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007563037130568844,
      "loss": 2.0619,
      "step": 19118
    },
    {
      "epoch": 1.6449281596833862,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007562804948819297,
      "loss": 1.3206,
      "step": 19119
    },
    {
      "epoch": 1.645014195990708,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007562572759574047,
      "loss": 1.5485,
      "step": 19120
    },
    {
      "epoch": 1.64510023229803,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007562340562833767,
      "loss": 1.4471,
      "step": 19121
    },
    {
      "epoch": 1.6451862686053516,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007562108358599139,
      "loss": 1.188,
      "step": 19122
    },
    {
      "epoch": 1.6452723049126732,
      "grad_norm": 11.875,
      "learning_rate": 0.0007561876146870843,
      "loss": 1.5678,
      "step": 19123
    },
    {
      "epoch": 1.6453583412199948,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000756164392764956,
      "loss": 1.0968,
      "step": 19124
    },
    {
      "epoch": 1.6454443775273164,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007561411700935965,
      "loss": 1.6142,
      "step": 19125
    },
    {
      "epoch": 1.6455304138346383,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007561179466730738,
      "loss": 1.9421,
      "step": 19126
    },
    {
      "epoch": 1.64561645014196,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007560947225034561,
      "loss": 1.969,
      "step": 19127
    },
    {
      "epoch": 1.6457024864492817,
      "grad_norm": 1.390625,
      "learning_rate": 0.000756071497584811,
      "loss": 1.365,
      "step": 19128
    },
    {
      "epoch": 1.6457885227566034,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007560482719172065,
      "loss": 0.9725,
      "step": 19129
    },
    {
      "epoch": 1.645874559063925,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007560250455007107,
      "loss": 1.6517,
      "step": 19130
    },
    {
      "epoch": 1.6459605953712466,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007560018183353915,
      "loss": 1.3735,
      "step": 19131
    },
    {
      "epoch": 1.6460466316785682,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007559785904213168,
      "loss": 1.2094,
      "step": 19132
    },
    {
      "epoch": 1.64613266798589,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007559553617585544,
      "loss": 1.124,
      "step": 19133
    },
    {
      "epoch": 1.6462187042932117,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007559321323471724,
      "loss": 0.8548,
      "step": 19134
    },
    {
      "epoch": 1.6463047406005336,
      "grad_norm": 2.625,
      "learning_rate": 0.0007559089021872387,
      "loss": 1.4013,
      "step": 19135
    },
    {
      "epoch": 1.6463907769078552,
      "grad_norm": 0.71875,
      "learning_rate": 0.0007558856712788211,
      "loss": 0.517,
      "step": 19136
    },
    {
      "epoch": 1.6464768132151768,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007558624396219877,
      "loss": 0.7599,
      "step": 19137
    },
    {
      "epoch": 1.6465628495224984,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007558392072168067,
      "loss": 1.0202,
      "step": 19138
    },
    {
      "epoch": 1.64664888582982,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007558159740633455,
      "loss": 1.5598,
      "step": 19139
    },
    {
      "epoch": 1.646734922137142,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007557927401616724,
      "loss": 1.2623,
      "step": 19140
    },
    {
      "epoch": 1.6468209584444635,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007557695055118552,
      "loss": 1.8992,
      "step": 19141
    },
    {
      "epoch": 1.6469069947517854,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007557462701139621,
      "loss": 1.23,
      "step": 19142
    },
    {
      "epoch": 1.646993031059107,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007557230339680608,
      "loss": 1.2185,
      "step": 19143
    },
    {
      "epoch": 1.6470790673664286,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007556997970742194,
      "loss": 1.4168,
      "step": 19144
    },
    {
      "epoch": 1.6471651036737502,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007556765594325059,
      "loss": 1.1544,
      "step": 19145
    },
    {
      "epoch": 1.6472511399810719,
      "grad_norm": 1.125,
      "learning_rate": 0.0007556533210429881,
      "loss": 1.2967,
      "step": 19146
    },
    {
      "epoch": 1.6473371762883937,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000755630081905734,
      "loss": 0.6538,
      "step": 19147
    },
    {
      "epoch": 1.6474232125957156,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007556068420208117,
      "loss": 1.1338,
      "step": 19148
    },
    {
      "epoch": 1.6475092489030372,
      "grad_norm": 2.03125,
      "learning_rate": 0.0007555836013882891,
      "loss": 1.7731,
      "step": 19149
    },
    {
      "epoch": 1.6475952852103588,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007555603600082342,
      "loss": 1.2441,
      "step": 19150
    },
    {
      "epoch": 1.6476813215176804,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007555371178807149,
      "loss": 0.9074,
      "step": 19151
    },
    {
      "epoch": 1.647767357825002,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007555138750057992,
      "loss": 1.1651,
      "step": 19152
    },
    {
      "epoch": 1.6478533941323237,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007554906313835552,
      "loss": 1.3564,
      "step": 19153
    },
    {
      "epoch": 1.6479394304396455,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007554673870140509,
      "loss": 1.5845,
      "step": 19154
    },
    {
      "epoch": 1.6480254667469674,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007554441418973539,
      "loss": 1.1847,
      "step": 19155
    },
    {
      "epoch": 1.648111503054289,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007554208960335327,
      "loss": 2.0119,
      "step": 19156
    },
    {
      "epoch": 1.6481975393616106,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007553976494226549,
      "loss": 1.075,
      "step": 19157
    },
    {
      "epoch": 1.6482835756689322,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007553744020647886,
      "loss": 1.5094,
      "step": 19158
    },
    {
      "epoch": 1.6483696119762539,
      "grad_norm": 1.0625,
      "learning_rate": 0.000755351153960002,
      "loss": 1.0879,
      "step": 19159
    },
    {
      "epoch": 1.6484556482835757,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007553279051083628,
      "loss": 1.639,
      "step": 19160
    },
    {
      "epoch": 1.6485416845908973,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007553046555099393,
      "loss": 1.1862,
      "step": 19161
    },
    {
      "epoch": 1.6486277208982192,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007552814051647994,
      "loss": 0.9501,
      "step": 19162
    },
    {
      "epoch": 1.6487137572055408,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007552581540730108,
      "loss": 0.8596,
      "step": 19163
    },
    {
      "epoch": 1.6487997935128624,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007552349022346418,
      "loss": 1.2112,
      "step": 19164
    },
    {
      "epoch": 1.648885829820184,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007552116496497604,
      "loss": 1.1072,
      "step": 19165
    },
    {
      "epoch": 1.6489718661275057,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007551883963184346,
      "loss": 1.7214,
      "step": 19166
    },
    {
      "epoch": 1.6490579024348275,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007551651422407321,
      "loss": 1.3778,
      "step": 19167
    },
    {
      "epoch": 1.6491439387421492,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007551418874167216,
      "loss": 0.8013,
      "step": 19168
    },
    {
      "epoch": 1.649229975049471,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007551186318464706,
      "loss": 2.0279,
      "step": 19169
    },
    {
      "epoch": 1.6493160113567926,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000755095375530047,
      "loss": 1.6237,
      "step": 19170
    },
    {
      "epoch": 1.6494020476641142,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007550721184675192,
      "loss": 1.5307,
      "step": 19171
    },
    {
      "epoch": 1.6494880839714359,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000755048860658955,
      "loss": 1.07,
      "step": 19172
    },
    {
      "epoch": 1.6495741202787575,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007550256021044226,
      "loss": 1.0415,
      "step": 19173
    },
    {
      "epoch": 1.6496601565860793,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007550023428039898,
      "loss": 1.8773,
      "step": 19174
    },
    {
      "epoch": 1.649746192893401,
      "grad_norm": 1.375,
      "learning_rate": 0.0007549790827577247,
      "loss": 0.9986,
      "step": 19175
    },
    {
      "epoch": 1.6498322292007228,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007549558219656955,
      "loss": 2.111,
      "step": 19176
    },
    {
      "epoch": 1.6499182655080444,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00075493256042797,
      "loss": 1.2962,
      "step": 19177
    },
    {
      "epoch": 1.650004301815366,
      "grad_norm": 1.25,
      "learning_rate": 0.0007549092981446164,
      "loss": 1.4522,
      "step": 19178
    },
    {
      "epoch": 1.6500903381226877,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007548860351157027,
      "loss": 1.259,
      "step": 19179
    },
    {
      "epoch": 1.6501763744300093,
      "grad_norm": 1.015625,
      "learning_rate": 0.000754862771341297,
      "loss": 0.7033,
      "step": 19180
    },
    {
      "epoch": 1.6502624107373312,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000754839506821467,
      "loss": 0.7577,
      "step": 19181
    },
    {
      "epoch": 1.650348447044653,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007548162415562811,
      "loss": 1.278,
      "step": 19182
    },
    {
      "epoch": 1.6504344833519746,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007547929755458074,
      "loss": 1.0258,
      "step": 19183
    },
    {
      "epoch": 1.6505205196592962,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007547697087901137,
      "loss": 1.4232,
      "step": 19184
    },
    {
      "epoch": 1.6506065559666179,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007547464412892682,
      "loss": 1.9022,
      "step": 19185
    },
    {
      "epoch": 1.6506925922739395,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007547231730433389,
      "loss": 1.666,
      "step": 19186
    },
    {
      "epoch": 1.6507786285812613,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007546999040523939,
      "loss": 0.9178,
      "step": 19187
    },
    {
      "epoch": 1.650864664888583,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007546766343165013,
      "loss": 1.0793,
      "step": 19188
    },
    {
      "epoch": 1.6509507011959048,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007546533638357287,
      "loss": 1.6253,
      "step": 19189
    },
    {
      "epoch": 1.6510367375032264,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007546300926101448,
      "loss": 1.417,
      "step": 19190
    },
    {
      "epoch": 1.651122773810548,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007546068206398175,
      "loss": 1.0824,
      "step": 19191
    },
    {
      "epoch": 1.6512088101178697,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007545835479248144,
      "loss": 1.0358,
      "step": 19192
    },
    {
      "epoch": 1.6512948464251913,
      "grad_norm": 3.359375,
      "learning_rate": 0.0007545602744652043,
      "loss": 1.2396,
      "step": 19193
    },
    {
      "epoch": 1.6513808827325132,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007545370002610547,
      "loss": 1.1052,
      "step": 19194
    },
    {
      "epoch": 1.6514669190398348,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000754513725312434,
      "loss": 1.0128,
      "step": 19195
    },
    {
      "epoch": 1.6515529553471566,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007544904496194102,
      "loss": 0.6674,
      "step": 19196
    },
    {
      "epoch": 1.6516389916544783,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007544671731820511,
      "loss": 2.1232,
      "step": 19197
    },
    {
      "epoch": 1.6517250279617999,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007544438960004252,
      "loss": 1.4097,
      "step": 19198
    },
    {
      "epoch": 1.6518110642691215,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007544206180746004,
      "loss": 1.5999,
      "step": 19199
    },
    {
      "epoch": 1.6518971005764431,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007543973394046447,
      "loss": 1.3465,
      "step": 19200
    },
    {
      "epoch": 1.651983136883765,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007543740599906262,
      "loss": 1.3225,
      "step": 19201
    },
    {
      "epoch": 1.6520691731910866,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007543507798326133,
      "loss": 1.3595,
      "step": 19202
    },
    {
      "epoch": 1.6521552094984084,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007543274989306736,
      "loss": 1.0328,
      "step": 19203
    },
    {
      "epoch": 1.65224124580573,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007543042172848754,
      "loss": 1.403,
      "step": 19204
    },
    {
      "epoch": 1.6523272821130517,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000754280934895287,
      "loss": 1.1852,
      "step": 19205
    },
    {
      "epoch": 1.6524133184203733,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007542576517619762,
      "loss": 1.2463,
      "step": 19206
    },
    {
      "epoch": 1.652499354727695,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007542343678850112,
      "loss": 1.3306,
      "step": 19207
    },
    {
      "epoch": 1.6525853910350168,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007542110832644602,
      "loss": 1.5014,
      "step": 19208
    },
    {
      "epoch": 1.6526714273423386,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007541877979003913,
      "loss": 1.1526,
      "step": 19209
    },
    {
      "epoch": 1.6527574636496603,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007541645117928724,
      "loss": 0.7958,
      "step": 19210
    },
    {
      "epoch": 1.6528434999569819,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007541412249419717,
      "loss": 0.9238,
      "step": 19211
    },
    {
      "epoch": 1.6529295362643035,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007541179373477575,
      "loss": 0.7307,
      "step": 19212
    },
    {
      "epoch": 1.6530155725716251,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007540946490102975,
      "loss": 1.9818,
      "step": 19213
    },
    {
      "epoch": 1.653101608878947,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007540713599296602,
      "loss": 2.1314,
      "step": 19214
    },
    {
      "epoch": 1.6531876451862686,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007540480701059136,
      "loss": 0.9205,
      "step": 19215
    },
    {
      "epoch": 1.6532736814935904,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007540247795391259,
      "loss": 1.1137,
      "step": 19216
    },
    {
      "epoch": 1.653359717800912,
      "grad_norm": 1.140625,
      "learning_rate": 0.000754001488229365,
      "loss": 0.8901,
      "step": 19217
    },
    {
      "epoch": 1.6534457541082337,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007539781961766994,
      "loss": 1.1556,
      "step": 19218
    },
    {
      "epoch": 1.6535317904155553,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007539549033811965,
      "loss": 1.1497,
      "step": 19219
    },
    {
      "epoch": 1.653617826722877,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007539316098429253,
      "loss": 0.8916,
      "step": 19220
    },
    {
      "epoch": 1.6537038630301988,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007539083155619535,
      "loss": 1.1985,
      "step": 19221
    },
    {
      "epoch": 1.6537898993375204,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007538850205383492,
      "loss": 1.2856,
      "step": 19222
    },
    {
      "epoch": 1.6538759356448423,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007538617247721804,
      "loss": 1.8674,
      "step": 19223
    },
    {
      "epoch": 1.6539619719521639,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007538384282635155,
      "loss": 1.3624,
      "step": 19224
    },
    {
      "epoch": 1.6540480082594855,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007538151310124228,
      "loss": 1.8068,
      "step": 19225
    },
    {
      "epoch": 1.6541340445668071,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00075379183301897,
      "loss": 1.6722,
      "step": 19226
    },
    {
      "epoch": 1.6542200808741288,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007537685342832254,
      "loss": 0.8044,
      "step": 19227
    },
    {
      "epoch": 1.6543061171814506,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007537452348052574,
      "loss": 1.5135,
      "step": 19228
    },
    {
      "epoch": 1.6543921534887722,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007537219345851338,
      "loss": 1.0963,
      "step": 19229
    },
    {
      "epoch": 1.654478189796094,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007536986336229229,
      "loss": 2.2451,
      "step": 19230
    },
    {
      "epoch": 1.6545642261034157,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007536753319186928,
      "loss": 1.0401,
      "step": 19231
    },
    {
      "epoch": 1.6546502624107373,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007536520294725117,
      "loss": 1.3587,
      "step": 19232
    },
    {
      "epoch": 1.654736298718059,
      "grad_norm": 1.078125,
      "learning_rate": 0.000753628726284448,
      "loss": 1.0647,
      "step": 19233
    },
    {
      "epoch": 1.6548223350253806,
      "grad_norm": 1.0,
      "learning_rate": 0.0007536054223545691,
      "loss": 1.193,
      "step": 19234
    },
    {
      "epoch": 1.6549083713327024,
      "grad_norm": 1.359375,
      "learning_rate": 0.000753582117682944,
      "loss": 1.2533,
      "step": 19235
    },
    {
      "epoch": 1.6549944076400243,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007535588122696406,
      "loss": 1.4239,
      "step": 19236
    },
    {
      "epoch": 1.6550804439473459,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007535355061147269,
      "loss": 1.2137,
      "step": 19237
    },
    {
      "epoch": 1.6551664802546675,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007535121992182711,
      "loss": 0.9555,
      "step": 19238
    },
    {
      "epoch": 1.6552525165619891,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007534888915803415,
      "loss": 1.947,
      "step": 19239
    },
    {
      "epoch": 1.6553385528693108,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007534655832010062,
      "loss": 1.3341,
      "step": 19240
    },
    {
      "epoch": 1.6554245891766324,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007534422740803332,
      "loss": 1.0268,
      "step": 19241
    },
    {
      "epoch": 1.6555106254839542,
      "grad_norm": 0.984375,
      "learning_rate": 0.000753418964218391,
      "loss": 1.0643,
      "step": 19242
    },
    {
      "epoch": 1.655596661791276,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007533956536152476,
      "loss": 1.8514,
      "step": 19243
    },
    {
      "epoch": 1.6556826980985977,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007533723422709711,
      "loss": 0.8623,
      "step": 19244
    },
    {
      "epoch": 1.6557687344059193,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007533490301856299,
      "loss": 1.7536,
      "step": 19245
    },
    {
      "epoch": 1.655854770713241,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000753325717359292,
      "loss": 1.2693,
      "step": 19246
    },
    {
      "epoch": 1.6559408070205626,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007533024037920256,
      "loss": 1.1413,
      "step": 19247
    },
    {
      "epoch": 1.6560268433278844,
      "grad_norm": 0.96875,
      "learning_rate": 0.000753279089483899,
      "loss": 0.609,
      "step": 19248
    },
    {
      "epoch": 1.656112879635206,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007532557744349803,
      "loss": 1.079,
      "step": 19249
    },
    {
      "epoch": 1.6561989159425279,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007532324586453376,
      "loss": 3.5411,
      "step": 19250
    },
    {
      "epoch": 1.6562849522498495,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007532091421150394,
      "loss": 0.5994,
      "step": 19251
    },
    {
      "epoch": 1.6563709885571711,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007531858248441537,
      "loss": 1.2245,
      "step": 19252
    },
    {
      "epoch": 1.6564570248644928,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007531625068327486,
      "loss": 1.3038,
      "step": 19253
    },
    {
      "epoch": 1.6565430611718144,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007531391880808924,
      "loss": 1.7252,
      "step": 19254
    },
    {
      "epoch": 1.6566290974791362,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0007531158685886533,
      "loss": 0.7531,
      "step": 19255
    },
    {
      "epoch": 1.6567151337864578,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007530925483560995,
      "loss": 0.8312,
      "step": 19256
    },
    {
      "epoch": 1.6568011700937797,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007530692273832993,
      "loss": 1.6464,
      "step": 19257
    },
    {
      "epoch": 1.6568872064011013,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007530459056703209,
      "loss": 1.4478,
      "step": 19258
    },
    {
      "epoch": 1.656973242708423,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007530225832172324,
      "loss": 0.9648,
      "step": 19259
    },
    {
      "epoch": 1.6570592790157446,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007529992600241019,
      "loss": 0.8875,
      "step": 19260
    },
    {
      "epoch": 1.6571453153230662,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007529759360909979,
      "loss": 1.087,
      "step": 19261
    },
    {
      "epoch": 1.657231351630388,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007529526114179886,
      "loss": 0.8218,
      "step": 19262
    },
    {
      "epoch": 1.6573173879377097,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007529292860051417,
      "loss": 1.1996,
      "step": 19263
    },
    {
      "epoch": 1.6574034242450315,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007529059598525263,
      "loss": 1.5184,
      "step": 19264
    },
    {
      "epoch": 1.6574894605523531,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007528826329602098,
      "loss": 1.252,
      "step": 19265
    },
    {
      "epoch": 1.6575754968596748,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007528593053282611,
      "loss": 1.1617,
      "step": 19266
    },
    {
      "epoch": 1.6576615331669964,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007528359769567479,
      "loss": 1.0336,
      "step": 19267
    },
    {
      "epoch": 1.657747569474318,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007528126478457387,
      "loss": 1.613,
      "step": 19268
    },
    {
      "epoch": 1.6578336057816399,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007527893179953016,
      "loss": 1.8791,
      "step": 19269
    },
    {
      "epoch": 1.6579196420889617,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0007527659874055051,
      "loss": 1.2635,
      "step": 19270
    },
    {
      "epoch": 1.6580056783962833,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000752742656076417,
      "loss": 0.766,
      "step": 19271
    },
    {
      "epoch": 1.658091714703605,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007527193240081058,
      "loss": 1.4331,
      "step": 19272
    },
    {
      "epoch": 1.6581777510109266,
      "grad_norm": 1.140625,
      "learning_rate": 0.00075269599120064,
      "loss": 1.2741,
      "step": 19273
    },
    {
      "epoch": 1.6582637873182482,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007526726576540875,
      "loss": 1.4271,
      "step": 19274
    },
    {
      "epoch": 1.65834982362557,
      "grad_norm": 1.5,
      "learning_rate": 0.0007526493233685163,
      "loss": 1.017,
      "step": 19275
    },
    {
      "epoch": 1.6584358599328917,
      "grad_norm": 6.0,
      "learning_rate": 0.0007526259883439952,
      "loss": 1.6907,
      "step": 19276
    },
    {
      "epoch": 1.6585218962402135,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007526026525805922,
      "loss": 1.4263,
      "step": 19277
    },
    {
      "epoch": 1.6586079325475351,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007525793160783757,
      "loss": 1.3891,
      "step": 19278
    },
    {
      "epoch": 1.6586939688548568,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007525559788374137,
      "loss": 1.9278,
      "step": 19279
    },
    {
      "epoch": 1.6587800051621784,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007525326408577746,
      "loss": 1.2825,
      "step": 19280
    },
    {
      "epoch": 1.6588660414695,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007525093021395267,
      "loss": 2.359,
      "step": 19281
    },
    {
      "epoch": 1.6589520777768219,
      "grad_norm": 1.125,
      "learning_rate": 0.0007524859626827381,
      "loss": 0.9021,
      "step": 19282
    },
    {
      "epoch": 1.6590381140841435,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007524626224874771,
      "loss": 1.8483,
      "step": 19283
    },
    {
      "epoch": 1.6591241503914653,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007524392815538121,
      "loss": 1.8013,
      "step": 19284
    },
    {
      "epoch": 1.659210186698787,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007524159398818114,
      "loss": 1.2397,
      "step": 19285
    },
    {
      "epoch": 1.6592962230061086,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007523925974715431,
      "loss": 0.7056,
      "step": 19286
    },
    {
      "epoch": 1.6593822593134302,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007523692543230753,
      "loss": 1.9981,
      "step": 19287
    },
    {
      "epoch": 1.6594682956207518,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007523459104364769,
      "loss": 0.7369,
      "step": 19288
    },
    {
      "epoch": 1.6595543319280737,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007523225658118156,
      "loss": 1.4108,
      "step": 19289
    },
    {
      "epoch": 1.6596403682353953,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007522992204491599,
      "loss": 1.2117,
      "step": 19290
    },
    {
      "epoch": 1.6597264045427171,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007522758743485781,
      "loss": 0.9201,
      "step": 19291
    },
    {
      "epoch": 1.6598124408500388,
      "grad_norm": 15.5,
      "learning_rate": 0.0007522525275101383,
      "loss": 0.8121,
      "step": 19292
    },
    {
      "epoch": 1.6598984771573604,
      "grad_norm": 3.28125,
      "learning_rate": 0.0007522291799339091,
      "loss": 1.2288,
      "step": 19293
    },
    {
      "epoch": 1.659984513464682,
      "grad_norm": 1.125,
      "learning_rate": 0.0007522058316199585,
      "loss": 1.357,
      "step": 19294
    },
    {
      "epoch": 1.6600705497720036,
      "grad_norm": 2.1875,
      "learning_rate": 0.0007521824825683549,
      "loss": 1.5226,
      "step": 19295
    },
    {
      "epoch": 1.6601565860793255,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007521591327791666,
      "loss": 0.9287,
      "step": 19296
    },
    {
      "epoch": 1.6602426223866473,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007521357822524617,
      "loss": 1.3027,
      "step": 19297
    },
    {
      "epoch": 1.660328658693969,
      "grad_norm": 6.96875,
      "learning_rate": 0.0007521124309883089,
      "loss": 1.5913,
      "step": 19298
    },
    {
      "epoch": 1.6604146950012906,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007520890789867763,
      "loss": 1.7777,
      "step": 19299
    },
    {
      "epoch": 1.6605007313086122,
      "grad_norm": 3.09375,
      "learning_rate": 0.0007520657262479321,
      "loss": 1.051,
      "step": 19300
    },
    {
      "epoch": 1.6605867676159338,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007520423727718446,
      "loss": 1.4294,
      "step": 19301
    },
    {
      "epoch": 1.6606728039232557,
      "grad_norm": 3.640625,
      "learning_rate": 0.0007520190185585822,
      "loss": 0.9727,
      "step": 19302
    },
    {
      "epoch": 1.6607588402305773,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007519956636082133,
      "loss": 1.3689,
      "step": 19303
    },
    {
      "epoch": 1.6608448765378991,
      "grad_norm": 1.140625,
      "learning_rate": 0.000751972307920806,
      "loss": 1.9132,
      "step": 19304
    },
    {
      "epoch": 1.6609309128452208,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007519489514964287,
      "loss": 1.2826,
      "step": 19305
    },
    {
      "epoch": 1.6610169491525424,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007519255943351497,
      "loss": 1.372,
      "step": 19306
    },
    {
      "epoch": 1.661102985459864,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007519022364370374,
      "loss": 1.6886,
      "step": 19307
    },
    {
      "epoch": 1.6611890217671856,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007518788778021602,
      "loss": 1.5491,
      "step": 19308
    },
    {
      "epoch": 1.6612750580745075,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007518555184305858,
      "loss": 1.5821,
      "step": 19309
    },
    {
      "epoch": 1.661361094381829,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007518321583223834,
      "loss": 1.298,
      "step": 19310
    },
    {
      "epoch": 1.661447130689151,
      "grad_norm": 11.25,
      "learning_rate": 0.0007518087974776209,
      "loss": 1.1452,
      "step": 19311
    },
    {
      "epoch": 1.6615331669964726,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007517854358963665,
      "loss": 0.8412,
      "step": 19312
    },
    {
      "epoch": 1.6616192033037942,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007517620735786886,
      "loss": 0.6081,
      "step": 19313
    },
    {
      "epoch": 1.6617052396111158,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007517387105246558,
      "loss": 2.2855,
      "step": 19314
    },
    {
      "epoch": 1.6617912759184374,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007517153467343361,
      "loss": 1.3548,
      "step": 19315
    },
    {
      "epoch": 1.6618773122257593,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007516919822077981,
      "loss": 0.8731,
      "step": 19316
    },
    {
      "epoch": 1.661963348533081,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00075166861694511,
      "loss": 2.0739,
      "step": 19317
    },
    {
      "epoch": 1.6620493848404028,
      "grad_norm": 1.21875,
      "learning_rate": 0.00075164525094634,
      "loss": 1.3411,
      "step": 19318
    },
    {
      "epoch": 1.6621354211477244,
      "grad_norm": 104.5,
      "learning_rate": 0.0007516218842115567,
      "loss": 1.0912,
      "step": 19319
    },
    {
      "epoch": 1.662221457455046,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007515985167408281,
      "loss": 1.2127,
      "step": 19320
    },
    {
      "epoch": 1.6623074937623676,
      "grad_norm": 1.546875,
      "learning_rate": 0.000751575148534223,
      "loss": 1.8654,
      "step": 19321
    },
    {
      "epoch": 1.6623935300696893,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007515517795918094,
      "loss": 1.2787,
      "step": 19322
    },
    {
      "epoch": 1.662479566377011,
      "grad_norm": 1.125,
      "learning_rate": 0.0007515284099136559,
      "loss": 2.0617,
      "step": 19323
    },
    {
      "epoch": 1.662565602684333,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007515050394998307,
      "loss": 1.1529,
      "step": 19324
    },
    {
      "epoch": 1.6626516389916546,
      "grad_norm": 1.28125,
      "learning_rate": 0.000751481668350402,
      "loss": 1.3566,
      "step": 19325
    },
    {
      "epoch": 1.6627376752989762,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007514582964654386,
      "loss": 1.4451,
      "step": 19326
    },
    {
      "epoch": 1.6628237116062978,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007514349238450082,
      "loss": 0.8465,
      "step": 19327
    },
    {
      "epoch": 1.6629097479136195,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007514115504891799,
      "loss": 1.4469,
      "step": 19328
    },
    {
      "epoch": 1.662995784220941,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007513881763980216,
      "loss": 1.9419,
      "step": 19329
    },
    {
      "epoch": 1.663081820528263,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007513648015716018,
      "loss": 0.9307,
      "step": 19330
    },
    {
      "epoch": 1.6631678568355848,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007513414260099888,
      "loss": 1.5645,
      "step": 19331
    },
    {
      "epoch": 1.6632538931429064,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000751318049713251,
      "loss": 0.9048,
      "step": 19332
    },
    {
      "epoch": 1.663339929450228,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007512946726814569,
      "loss": 1.0229,
      "step": 19333
    },
    {
      "epoch": 1.6634259657575496,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007512712949146747,
      "loss": 1.4172,
      "step": 19334
    },
    {
      "epoch": 1.6635120020648713,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007512479164129728,
      "loss": 0.6546,
      "step": 19335
    },
    {
      "epoch": 1.663598038372193,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007512245371764195,
      "loss": 1.1596,
      "step": 19336
    },
    {
      "epoch": 1.6636840746795147,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007512011572050835,
      "loss": 1.3939,
      "step": 19337
    },
    {
      "epoch": 1.6637701109868366,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007511777764990328,
      "loss": 1.4659,
      "step": 19338
    },
    {
      "epoch": 1.6638561472941582,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007511543950583362,
      "loss": 0.6989,
      "step": 19339
    },
    {
      "epoch": 1.6639421836014798,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007511310128830616,
      "loss": 1.5001,
      "step": 19340
    },
    {
      "epoch": 1.6640282199088015,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007511076299732777,
      "loss": 1.1425,
      "step": 19341
    },
    {
      "epoch": 1.664114256216123,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007510842463290528,
      "loss": 0.9935,
      "step": 19342
    },
    {
      "epoch": 1.664200292523445,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007510608619504554,
      "loss": 2.1441,
      "step": 19343
    },
    {
      "epoch": 1.6642863288307665,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007510374768375537,
      "loss": 1.1258,
      "step": 19344
    },
    {
      "epoch": 1.6643723651380884,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007510140909904162,
      "loss": 0.9014,
      "step": 19345
    },
    {
      "epoch": 1.66445840144541,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007509907044091113,
      "loss": 0.8142,
      "step": 19346
    },
    {
      "epoch": 1.6645444377527316,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007509673170937075,
      "loss": 1.0517,
      "step": 19347
    },
    {
      "epoch": 1.6646304740600533,
      "grad_norm": 0.984375,
      "learning_rate": 0.000750943929044273,
      "loss": 0.8476,
      "step": 19348
    },
    {
      "epoch": 1.664716510367375,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0007509205402608765,
      "loss": 1.6202,
      "step": 19349
    },
    {
      "epoch": 1.6648025466746967,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000750897150743586,
      "loss": 1.075,
      "step": 19350
    },
    {
      "epoch": 1.6648885829820184,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007508737604924702,
      "loss": 1.2786,
      "step": 19351
    },
    {
      "epoch": 1.6649746192893402,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007508503695075975,
      "loss": 1.0106,
      "step": 19352
    },
    {
      "epoch": 1.6650606555966618,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007508269777890362,
      "loss": 1.0003,
      "step": 19353
    },
    {
      "epoch": 1.6651466919039835,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007508035853368549,
      "loss": 1.8586,
      "step": 19354
    },
    {
      "epoch": 1.665232728211305,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007507801921511218,
      "loss": 0.8045,
      "step": 19355
    },
    {
      "epoch": 1.6653187645186267,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007507567982319055,
      "loss": 1.4676,
      "step": 19356
    },
    {
      "epoch": 1.6654048008259486,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007507334035792741,
      "loss": 1.4119,
      "step": 19357
    },
    {
      "epoch": 1.6654908371332704,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007507100081932964,
      "loss": 0.9526,
      "step": 19358
    },
    {
      "epoch": 1.665576873440592,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007506866120740407,
      "loss": 0.7802,
      "step": 19359
    },
    {
      "epoch": 1.6656629097479136,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007506632152215753,
      "loss": 1.4708,
      "step": 19360
    },
    {
      "epoch": 1.6657489460552353,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007506398176359688,
      "loss": 1.3144,
      "step": 19361
    },
    {
      "epoch": 1.665834982362557,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007506164193172897,
      "loss": 1.6929,
      "step": 19362
    },
    {
      "epoch": 1.6659210186698787,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007505930202656063,
      "loss": 1.3021,
      "step": 19363
    },
    {
      "epoch": 1.6660070549772004,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007505696204809871,
      "loss": 0.8348,
      "step": 19364
    },
    {
      "epoch": 1.6660930912845222,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007505462199635003,
      "loss": 1.9232,
      "step": 19365
    },
    {
      "epoch": 1.6661791275918438,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007505228187132146,
      "loss": 1.2671,
      "step": 19366
    },
    {
      "epoch": 1.6662651638991655,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007504994167301984,
      "loss": 1.3825,
      "step": 19367
    },
    {
      "epoch": 1.666351200206487,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007504760140145202,
      "loss": 1.5344,
      "step": 19368
    },
    {
      "epoch": 1.6664372365138087,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007504526105662481,
      "loss": 1.4304,
      "step": 19369
    },
    {
      "epoch": 1.6665232728211306,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007504292063854512,
      "loss": 1.8258,
      "step": 19370
    },
    {
      "epoch": 1.6666093091284522,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007504058014721974,
      "loss": 0.7591,
      "step": 19371
    },
    {
      "epoch": 1.666695345435774,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007503823958265552,
      "loss": 0.9858,
      "step": 19372
    },
    {
      "epoch": 1.6667813817430956,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007503589894485934,
      "loss": 1.3348,
      "step": 19373
    },
    {
      "epoch": 1.6668674180504173,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007503355823383802,
      "loss": 0.7898,
      "step": 19374
    },
    {
      "epoch": 1.666953454357739,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007503121744959842,
      "loss": 1.204,
      "step": 19375
    },
    {
      "epoch": 1.6670394906650605,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007502887659214736,
      "loss": 1.5315,
      "step": 19376
    },
    {
      "epoch": 1.6671255269723824,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007502653566149171,
      "loss": 0.9152,
      "step": 19377
    },
    {
      "epoch": 1.667211563279704,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007502419465763832,
      "loss": 1.141,
      "step": 19378
    },
    {
      "epoch": 1.6672975995870258,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007502185358059401,
      "loss": 0.7708,
      "step": 19379
    },
    {
      "epoch": 1.6673836358943475,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007501951243036564,
      "loss": 0.6032,
      "step": 19380
    },
    {
      "epoch": 1.667469672201669,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007501717120696008,
      "loss": 1.2741,
      "step": 19381
    },
    {
      "epoch": 1.6675557085089907,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007501482991038415,
      "loss": 1.3412,
      "step": 19382
    },
    {
      "epoch": 1.6676417448163123,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007501248854064471,
      "loss": 1.5071,
      "step": 19383
    },
    {
      "epoch": 1.6677277811236342,
      "grad_norm": 1.078125,
      "learning_rate": 0.000750101470977486,
      "loss": 1.2001,
      "step": 19384
    },
    {
      "epoch": 1.667813817430956,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007500780558170269,
      "loss": 1.8674,
      "step": 19385
    },
    {
      "epoch": 1.6678998537382776,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007500546399251382,
      "loss": 0.9524,
      "step": 19386
    },
    {
      "epoch": 1.6679858900455993,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000750031223301888,
      "loss": 1.0992,
      "step": 19387
    },
    {
      "epoch": 1.668071926352921,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007500078059473452,
      "loss": 1.3395,
      "step": 19388
    },
    {
      "epoch": 1.6681579626602425,
      "grad_norm": 1.625,
      "learning_rate": 0.0007499843878615783,
      "loss": 1.3819,
      "step": 19389
    },
    {
      "epoch": 1.6682439989675641,
      "grad_norm": 1.0,
      "learning_rate": 0.0007499609690446558,
      "loss": 0.9093,
      "step": 19390
    },
    {
      "epoch": 1.668330035274886,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007499375494966457,
      "loss": 0.9219,
      "step": 19391
    },
    {
      "epoch": 1.6684160715822078,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007499141292176171,
      "loss": 1.5779,
      "step": 19392
    },
    {
      "epoch": 1.6685021078895295,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007498907082076383,
      "loss": 1.5032,
      "step": 19393
    },
    {
      "epoch": 1.668588144196851,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007498672864667777,
      "loss": 2.2263,
      "step": 19394
    },
    {
      "epoch": 1.6686741805041727,
      "grad_norm": 1.125,
      "learning_rate": 0.0007498438639951039,
      "loss": 1.5902,
      "step": 19395
    },
    {
      "epoch": 1.6687602168114943,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007498204407926856,
      "loss": 1.349,
      "step": 19396
    },
    {
      "epoch": 1.6688462531188162,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007497970168595909,
      "loss": 1.3978,
      "step": 19397
    },
    {
      "epoch": 1.6689322894261378,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007497735921958884,
      "loss": 1.9787,
      "step": 19398
    },
    {
      "epoch": 1.6690183257334597,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007497501668016468,
      "loss": 0.6839,
      "step": 19399
    },
    {
      "epoch": 1.6691043620407813,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007497267406769345,
      "loss": 1.5546,
      "step": 19400
    },
    {
      "epoch": 1.669190398348103,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007497033138218202,
      "loss": 0.6696,
      "step": 19401
    },
    {
      "epoch": 1.6692764346554245,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007496798862363722,
      "loss": 1.2919,
      "step": 19402
    },
    {
      "epoch": 1.6693624709627461,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007496564579206591,
      "loss": 1.6312,
      "step": 19403
    },
    {
      "epoch": 1.669448507270068,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007496330288747495,
      "loss": 0.9562,
      "step": 19404
    },
    {
      "epoch": 1.6695345435773896,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0007496095990987118,
      "loss": 1.6586,
      "step": 19405
    },
    {
      "epoch": 1.6696205798847115,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007495861685926144,
      "loss": 1.7434,
      "step": 19406
    },
    {
      "epoch": 1.669706616192033,
      "grad_norm": 2.015625,
      "learning_rate": 0.0007495627373565262,
      "loss": 1.9652,
      "step": 19407
    },
    {
      "epoch": 1.6697926524993547,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007495393053905154,
      "loss": 1.328,
      "step": 19408
    },
    {
      "epoch": 1.6698786888066763,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007495158726946507,
      "loss": 1.4077,
      "step": 19409
    },
    {
      "epoch": 1.669964725113998,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007494924392690007,
      "loss": 1.5491,
      "step": 19410
    },
    {
      "epoch": 1.6700507614213198,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007494690051136337,
      "loss": 1.381,
      "step": 19411
    },
    {
      "epoch": 1.6701367977286414,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007494455702286185,
      "loss": 1.9729,
      "step": 19412
    },
    {
      "epoch": 1.6702228340359633,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007494221346140234,
      "loss": 1.787,
      "step": 19413
    },
    {
      "epoch": 1.670308870343285,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007493986982699171,
      "loss": 1.3998,
      "step": 19414
    },
    {
      "epoch": 1.6703949066506065,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007493752611963682,
      "loss": 1.5441,
      "step": 19415
    },
    {
      "epoch": 1.6704809429579281,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007493518233934451,
      "loss": 1.1284,
      "step": 19416
    },
    {
      "epoch": 1.6705669792652498,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007493283848612162,
      "loss": 1.5741,
      "step": 19417
    },
    {
      "epoch": 1.6706530155725716,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007493049455997506,
      "loss": 0.9085,
      "step": 19418
    },
    {
      "epoch": 1.6707390518798935,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007492815056091165,
      "loss": 1.3789,
      "step": 19419
    },
    {
      "epoch": 1.670825088187215,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007492580648893823,
      "loss": 0.5882,
      "step": 19420
    },
    {
      "epoch": 1.6709111244945367,
      "grad_norm": 11.8125,
      "learning_rate": 0.0007492346234406167,
      "loss": 1.4352,
      "step": 19421
    },
    {
      "epoch": 1.6709971608018583,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007492111812628885,
      "loss": 1.2192,
      "step": 19422
    },
    {
      "epoch": 1.67108319710918,
      "grad_norm": 1.125,
      "learning_rate": 0.000749187738356266,
      "loss": 1.1314,
      "step": 19423
    },
    {
      "epoch": 1.6711692334165018,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007491642947208177,
      "loss": 1.4118,
      "step": 19424
    },
    {
      "epoch": 1.6712552697238234,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007491408503566125,
      "loss": 1.7099,
      "step": 19425
    },
    {
      "epoch": 1.6713413060311453,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007491174052637186,
      "loss": 0.5927,
      "step": 19426
    },
    {
      "epoch": 1.671427342338467,
      "grad_norm": 1.6875,
      "learning_rate": 0.000749093959442205,
      "loss": 0.8627,
      "step": 19427
    },
    {
      "epoch": 1.6715133786457885,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007490705128921397,
      "loss": 0.9605,
      "step": 19428
    },
    {
      "epoch": 1.6715994149531102,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007490470656135917,
      "loss": 0.8622,
      "step": 19429
    },
    {
      "epoch": 1.6716854512604318,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007490236176066296,
      "loss": 1.7935,
      "step": 19430
    },
    {
      "epoch": 1.6717714875677536,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007490001688713218,
      "loss": 1.2056,
      "step": 19431
    },
    {
      "epoch": 1.6718575238750752,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007489767194077368,
      "loss": 1.7141,
      "step": 19432
    },
    {
      "epoch": 1.671943560182397,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007489532692159433,
      "loss": 0.9891,
      "step": 19433
    },
    {
      "epoch": 1.6720295964897187,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00074892981829601,
      "loss": 0.8341,
      "step": 19434
    },
    {
      "epoch": 1.6721156327970403,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007489063666480054,
      "loss": 0.8521,
      "step": 19435
    },
    {
      "epoch": 1.672201669104362,
      "grad_norm": 1.234375,
      "learning_rate": 0.000748882914271998,
      "loss": 2.04,
      "step": 19436
    },
    {
      "epoch": 1.6722877054116836,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007488594611680566,
      "loss": 0.5452,
      "step": 19437
    },
    {
      "epoch": 1.6723737417190054,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007488360073362497,
      "loss": 1.0422,
      "step": 19438
    },
    {
      "epoch": 1.672459778026327,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0007488125527766457,
      "loss": 1.4406,
      "step": 19439
    },
    {
      "epoch": 1.672545814333649,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0007487890974893135,
      "loss": 1.3946,
      "step": 19440
    },
    {
      "epoch": 1.6726318506409705,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007487656414743217,
      "loss": 1.3182,
      "step": 19441
    },
    {
      "epoch": 1.6727178869482922,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007487421847317386,
      "loss": 1.4687,
      "step": 19442
    },
    {
      "epoch": 1.6728039232556138,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007487187272616331,
      "loss": 1.385,
      "step": 19443
    },
    {
      "epoch": 1.6728899595629354,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007486952690640736,
      "loss": 0.91,
      "step": 19444
    },
    {
      "epoch": 1.6729759958702572,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007486718101391288,
      "loss": 1.4758,
      "step": 19445
    },
    {
      "epoch": 1.673062032177579,
      "grad_norm": 1.25,
      "learning_rate": 0.0007486483504868674,
      "loss": 1.5699,
      "step": 19446
    },
    {
      "epoch": 1.6731480684849007,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007486248901073578,
      "loss": 1.1595,
      "step": 19447
    },
    {
      "epoch": 1.6732341047922223,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007486014290006689,
      "loss": 1.8113,
      "step": 19448
    },
    {
      "epoch": 1.673320141099544,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007485779671668691,
      "loss": 0.8608,
      "step": 19449
    },
    {
      "epoch": 1.6734061774068656,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007485545046060271,
      "loss": 0.632,
      "step": 19450
    },
    {
      "epoch": 1.6734922137141874,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007485310413182116,
      "loss": 1.7708,
      "step": 19451
    },
    {
      "epoch": 1.673578250021509,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007485075773034911,
      "loss": 1.4519,
      "step": 19452
    },
    {
      "epoch": 1.673664286328831,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007484841125619342,
      "loss": 2.0715,
      "step": 19453
    },
    {
      "epoch": 1.6737503226361525,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007484606470936095,
      "loss": 0.5543,
      "step": 19454
    },
    {
      "epoch": 1.6738363589434742,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007484371808985858,
      "loss": 1.5885,
      "step": 19455
    },
    {
      "epoch": 1.6739223952507958,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007484137139769318,
      "loss": 1.6403,
      "step": 19456
    },
    {
      "epoch": 1.6740084315581174,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000748390246328716,
      "loss": 1.5872,
      "step": 19457
    },
    {
      "epoch": 1.6740944678654393,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007483667779540069,
      "loss": 1.3278,
      "step": 19458
    },
    {
      "epoch": 1.6741805041727609,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007483433088528732,
      "loss": 1.767,
      "step": 19459
    },
    {
      "epoch": 1.6742665404800827,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007483198390253839,
      "loss": 1.0533,
      "step": 19460
    },
    {
      "epoch": 1.6743525767874043,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007482963684716074,
      "loss": 1.67,
      "step": 19461
    },
    {
      "epoch": 1.674438613094726,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007482728971916119,
      "loss": 1.5612,
      "step": 19462
    },
    {
      "epoch": 1.6745246494020476,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007482494251854668,
      "loss": 1.1499,
      "step": 19463
    },
    {
      "epoch": 1.6746106857093692,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007482259524532403,
      "loss": 1.3057,
      "step": 19464
    },
    {
      "epoch": 1.674696722016691,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007482024789950011,
      "loss": 1.014,
      "step": 19465
    },
    {
      "epoch": 1.6747827583240127,
      "grad_norm": 1.328125,
      "learning_rate": 0.000748179004810818,
      "loss": 1.4595,
      "step": 19466
    },
    {
      "epoch": 1.6748687946313345,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007481555299007597,
      "loss": 1.6051,
      "step": 19467
    },
    {
      "epoch": 1.6749548309386562,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007481320542648946,
      "loss": 1.9478,
      "step": 19468
    },
    {
      "epoch": 1.6750408672459778,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007481085779032915,
      "loss": 1.6457,
      "step": 19469
    },
    {
      "epoch": 1.6751269035532994,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007480851008160191,
      "loss": 1.4631,
      "step": 19470
    },
    {
      "epoch": 1.675212939860621,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007480616230031461,
      "loss": 0.7364,
      "step": 19471
    },
    {
      "epoch": 1.6752989761679429,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007480381444647409,
      "loss": 1.1864,
      "step": 19472
    },
    {
      "epoch": 1.6753850124752647,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007480146652008723,
      "loss": 1.5278,
      "step": 19473
    },
    {
      "epoch": 1.6754710487825863,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007479911852116093,
      "loss": 1.1338,
      "step": 19474
    },
    {
      "epoch": 1.675557085089908,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007479677044970204,
      "loss": 1.3056,
      "step": 19475
    },
    {
      "epoch": 1.6756431213972296,
      "grad_norm": 1.484375,
      "learning_rate": 0.000747944223057174,
      "loss": 1.618,
      "step": 19476
    },
    {
      "epoch": 1.6757291577045512,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007479207408921389,
      "loss": 1.1228,
      "step": 19477
    },
    {
      "epoch": 1.6758151940118728,
      "grad_norm": 1.109375,
      "learning_rate": 0.000747897258001984,
      "loss": 0.9059,
      "step": 19478
    },
    {
      "epoch": 1.6759012303191947,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007478737743867778,
      "loss": 1.1576,
      "step": 19479
    },
    {
      "epoch": 1.6759872666265165,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007478502900465888,
      "loss": 1.0533,
      "step": 19480
    },
    {
      "epoch": 1.6760733029338382,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007478268049814862,
      "loss": 1.6612,
      "step": 19481
    },
    {
      "epoch": 1.6761593392411598,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007478033191915384,
      "loss": 1.2347,
      "step": 19482
    },
    {
      "epoch": 1.6762453755484814,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007477798326768141,
      "loss": 1.6126,
      "step": 19483
    },
    {
      "epoch": 1.676331411855803,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007477563454373817,
      "loss": 0.7234,
      "step": 19484
    },
    {
      "epoch": 1.6764174481631249,
      "grad_norm": 1.5,
      "learning_rate": 0.0007477328574733104,
      "loss": 0.9337,
      "step": 19485
    },
    {
      "epoch": 1.6765034844704465,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007477093687846686,
      "loss": 1.3411,
      "step": 19486
    },
    {
      "epoch": 1.6765895207777683,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007476858793715251,
      "loss": 0.7661,
      "step": 19487
    },
    {
      "epoch": 1.67667555708509,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007476623892339486,
      "loss": 1.6092,
      "step": 19488
    },
    {
      "epoch": 1.6767615933924116,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007476388983720078,
      "loss": 1.1565,
      "step": 19489
    },
    {
      "epoch": 1.6768476296997332,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007476154067857713,
      "loss": 1.2336,
      "step": 19490
    },
    {
      "epoch": 1.6769336660070548,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0007475919144753078,
      "loss": 1.6762,
      "step": 19491
    },
    {
      "epoch": 1.6770197023143767,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007475684214406862,
      "loss": 1.0688,
      "step": 19492
    },
    {
      "epoch": 1.6771057386216983,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007475449276819752,
      "loss": 1.8521,
      "step": 19493
    },
    {
      "epoch": 1.6771917749290202,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007475214331992434,
      "loss": 1.4432,
      "step": 19494
    },
    {
      "epoch": 1.6772778112363418,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007474979379925595,
      "loss": 1.8505,
      "step": 19495
    },
    {
      "epoch": 1.6773638475436634,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007474744420619923,
      "loss": 1.6353,
      "step": 19496
    },
    {
      "epoch": 1.677449883850985,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007474509454076105,
      "loss": 2.1403,
      "step": 19497
    },
    {
      "epoch": 1.6775359201583067,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007474274480294828,
      "loss": 1.0221,
      "step": 19498
    },
    {
      "epoch": 1.6776219564656285,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007474039499276779,
      "loss": 1.216,
      "step": 19499
    },
    {
      "epoch": 1.6777079927729501,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007473804511022646,
      "loss": 1.4014,
      "step": 19500
    },
    {
      "epoch": 1.677794029080272,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007473569515533116,
      "loss": 1.301,
      "step": 19501
    },
    {
      "epoch": 1.6778800653875936,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007473334512808876,
      "loss": 0.878,
      "step": 19502
    },
    {
      "epoch": 1.6779661016949152,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007473099502850613,
      "loss": 1.1128,
      "step": 19503
    },
    {
      "epoch": 1.6780521380022368,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007472864485659017,
      "loss": 1.7597,
      "step": 19504
    },
    {
      "epoch": 1.6781381743095585,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007472629461234773,
      "loss": 1.2849,
      "step": 19505
    },
    {
      "epoch": 1.6782242106168803,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007472394429578566,
      "loss": 1.3684,
      "step": 19506
    },
    {
      "epoch": 1.6783102469242022,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007472159390691088,
      "loss": 1.6453,
      "step": 19507
    },
    {
      "epoch": 1.6783962832315238,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007471924344573024,
      "loss": 0.7173,
      "step": 19508
    },
    {
      "epoch": 1.6784823195388454,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007471689291225062,
      "loss": 0.9649,
      "step": 19509
    },
    {
      "epoch": 1.678568355846167,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007471454230647891,
      "loss": 1.1726,
      "step": 19510
    },
    {
      "epoch": 1.6786543921534887,
      "grad_norm": 1.625,
      "learning_rate": 0.0007471219162842195,
      "loss": 0.9024,
      "step": 19511
    },
    {
      "epoch": 1.6787404284608105,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007470984087808665,
      "loss": 1.5603,
      "step": 19512
    },
    {
      "epoch": 1.6788264647681321,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007470749005547986,
      "loss": 0.7504,
      "step": 19513
    },
    {
      "epoch": 1.678912501075454,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007470513916060848,
      "loss": 1.4992,
      "step": 19514
    },
    {
      "epoch": 1.6789985373827756,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007470278819347937,
      "loss": 1.0754,
      "step": 19515
    },
    {
      "epoch": 1.6790845736900972,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000747004371540994,
      "loss": 0.9853,
      "step": 19516
    },
    {
      "epoch": 1.6791706099974189,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007469808604247547,
      "loss": 1.8449,
      "step": 19517
    },
    {
      "epoch": 1.6792566463047405,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007469573485861442,
      "loss": 1.021,
      "step": 19518
    },
    {
      "epoch": 1.6793426826120623,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007469338360252318,
      "loss": 1.1151,
      "step": 19519
    },
    {
      "epoch": 1.679428718919384,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007469103227420857,
      "loss": 1.2858,
      "step": 19520
    },
    {
      "epoch": 1.6795147552267058,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007468868087367749,
      "loss": 0.9549,
      "step": 19521
    },
    {
      "epoch": 1.6796007915340274,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007468632940093683,
      "loss": 1.2097,
      "step": 19522
    },
    {
      "epoch": 1.679686827841349,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007468397785599346,
      "loss": 1.0189,
      "step": 19523
    },
    {
      "epoch": 1.6797728641486707,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0007468162623885427,
      "loss": 2.0453,
      "step": 19524
    },
    {
      "epoch": 1.6798589004559923,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000746792745495261,
      "loss": 1.6865,
      "step": 19525
    },
    {
      "epoch": 1.6799449367633141,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007467692278801585,
      "loss": 1.1452,
      "step": 19526
    },
    {
      "epoch": 1.6800309730706358,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007467457095433042,
      "loss": 1.2134,
      "step": 19527
    },
    {
      "epoch": 1.6801170093779576,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007467221904847665,
      "loss": 1.0029,
      "step": 19528
    },
    {
      "epoch": 1.6802030456852792,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007466986707046144,
      "loss": 0.664,
      "step": 19529
    },
    {
      "epoch": 1.6802890819926009,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007466751502029168,
      "loss": 1.9114,
      "step": 19530
    },
    {
      "epoch": 1.6803751182999225,
      "grad_norm": 1.71875,
      "learning_rate": 0.0007466516289797423,
      "loss": 1.2125,
      "step": 19531
    },
    {
      "epoch": 1.680461154607244,
      "grad_norm": 1.0,
      "learning_rate": 0.0007466281070351599,
      "loss": 0.8892,
      "step": 19532
    },
    {
      "epoch": 1.680547190914566,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000746604584369238,
      "loss": 1.1943,
      "step": 19533
    },
    {
      "epoch": 1.6806332272218878,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007465810609820457,
      "loss": 1.3283,
      "step": 19534
    },
    {
      "epoch": 1.6807192635292094,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007465575368736518,
      "loss": 1.2695,
      "step": 19535
    },
    {
      "epoch": 1.680805299836531,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007465340120441251,
      "loss": 1.6442,
      "step": 19536
    },
    {
      "epoch": 1.6808913361438527,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007465104864935343,
      "loss": 1.2092,
      "step": 19537
    },
    {
      "epoch": 1.6809773724511743,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007464869602219483,
      "loss": 1.4536,
      "step": 19538
    },
    {
      "epoch": 1.6810634087584961,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007464634332294359,
      "loss": 0.966,
      "step": 19539
    },
    {
      "epoch": 1.6811494450658178,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007464399055160658,
      "loss": 1.2819,
      "step": 19540
    },
    {
      "epoch": 1.6812354813731396,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000746416377081907,
      "loss": 1.2266,
      "step": 19541
    },
    {
      "epoch": 1.6813215176804612,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007463928479270283,
      "loss": 1.0393,
      "step": 19542
    },
    {
      "epoch": 1.6814075539877829,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007463693180514983,
      "loss": 1.0912,
      "step": 19543
    },
    {
      "epoch": 1.6814935902951045,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000746345787455386,
      "loss": 1.0909,
      "step": 19544
    },
    {
      "epoch": 1.681579626602426,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007463222561387602,
      "loss": 1.2617,
      "step": 19545
    },
    {
      "epoch": 1.681665662909748,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007462987241016897,
      "loss": 0.8643,
      "step": 19546
    },
    {
      "epoch": 1.6817516992170696,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007462751913442433,
      "loss": 1.0147,
      "step": 19547
    },
    {
      "epoch": 1.6818377355243914,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007462516578664898,
      "loss": 1.6468,
      "step": 19548
    },
    {
      "epoch": 1.681923771831713,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007462281236684982,
      "loss": 0.5881,
      "step": 19549
    },
    {
      "epoch": 1.6820098081390347,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007462045887503372,
      "loss": 1.6099,
      "step": 19550
    },
    {
      "epoch": 1.6820958444463563,
      "grad_norm": 1.125,
      "learning_rate": 0.0007461810531120758,
      "loss": 1.1108,
      "step": 19551
    },
    {
      "epoch": 1.682181880753678,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0007461575167537825,
      "loss": 1.6607,
      "step": 19552
    },
    {
      "epoch": 1.6822679170609998,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007461339796755264,
      "loss": 2.0472,
      "step": 19553
    },
    {
      "epoch": 1.6823539533683214,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007461104418773763,
      "loss": 1.4432,
      "step": 19554
    },
    {
      "epoch": 1.6824399896756432,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007460869033594009,
      "loss": 0.9832,
      "step": 19555
    },
    {
      "epoch": 1.6825260259829649,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007460633641216693,
      "loss": 0.7649,
      "step": 19556
    },
    {
      "epoch": 1.6826120622902865,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007460398241642503,
      "loss": 1.6445,
      "step": 19557
    },
    {
      "epoch": 1.682698098597608,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007460162834872126,
      "loss": 1.187,
      "step": 19558
    },
    {
      "epoch": 1.6827841349049297,
      "grad_norm": 1.15625,
      "learning_rate": 0.000745992742090625,
      "loss": 0.8812,
      "step": 19559
    },
    {
      "epoch": 1.6828701712122516,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007459691999745566,
      "loss": 0.9586,
      "step": 19560
    },
    {
      "epoch": 1.6829562075195734,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007459456571390761,
      "loss": 0.8784,
      "step": 19561
    },
    {
      "epoch": 1.683042243826895,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007459221135842522,
      "loss": 1.3814,
      "step": 19562
    },
    {
      "epoch": 1.6831282801342167,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007458985693101542,
      "loss": 1.376,
      "step": 19563
    },
    {
      "epoch": 1.6832143164415383,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007458750243168506,
      "loss": 1.063,
      "step": 19564
    },
    {
      "epoch": 1.68330035274886,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007458514786044104,
      "loss": 1.542,
      "step": 19565
    },
    {
      "epoch": 1.6833863890561815,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007458279321729025,
      "loss": 1.574,
      "step": 19566
    },
    {
      "epoch": 1.6834724253635034,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007458043850223956,
      "loss": 0.8249,
      "step": 19567
    },
    {
      "epoch": 1.6835584616708252,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007457808371529587,
      "loss": 1.8492,
      "step": 19568
    },
    {
      "epoch": 1.6836444979781469,
      "grad_norm": 1.125,
      "learning_rate": 0.0007457572885646608,
      "loss": 1.4847,
      "step": 19569
    },
    {
      "epoch": 1.6837305342854685,
      "grad_norm": 1.921875,
      "learning_rate": 0.0007457337392575704,
      "loss": 1.3754,
      "step": 19570
    },
    {
      "epoch": 1.68381657059279,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007457101892317567,
      "loss": 0.8871,
      "step": 19571
    },
    {
      "epoch": 1.6839026069001117,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007456866384872886,
      "loss": 1.1534,
      "step": 19572
    },
    {
      "epoch": 1.6839886432074336,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007456630870242347,
      "loss": 1.3978,
      "step": 19573
    },
    {
      "epoch": 1.6840746795147552,
      "grad_norm": 1.8125,
      "learning_rate": 0.0007456395348426641,
      "loss": 1.8659,
      "step": 19574
    },
    {
      "epoch": 1.684160715822077,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007456159819426457,
      "loss": 0.5988,
      "step": 19575
    },
    {
      "epoch": 1.6842467521293987,
      "grad_norm": 0.875,
      "learning_rate": 0.0007455924283242484,
      "loss": 0.8422,
      "step": 19576
    },
    {
      "epoch": 1.6843327884367203,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007455688739875407,
      "loss": 1.6414,
      "step": 19577
    },
    {
      "epoch": 1.684418824744042,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000745545318932592,
      "loss": 0.7891,
      "step": 19578
    },
    {
      "epoch": 1.6845048610513635,
      "grad_norm": 1.203125,
      "learning_rate": 0.000745521763159471,
      "loss": 1.159,
      "step": 19579
    },
    {
      "epoch": 1.6845908973586854,
      "grad_norm": 1.71875,
      "learning_rate": 0.0007454982066682467,
      "loss": 1.7609,
      "step": 19580
    },
    {
      "epoch": 1.684676933666007,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007454746494589877,
      "loss": 1.3228,
      "step": 19581
    },
    {
      "epoch": 1.6847629699733289,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007454510915317631,
      "loss": 1.1691,
      "step": 19582
    },
    {
      "epoch": 1.6848490062806505,
      "grad_norm": 2.671875,
      "learning_rate": 0.0007454275328866419,
      "loss": 1.7143,
      "step": 19583
    },
    {
      "epoch": 1.684935042587972,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007454039735236928,
      "loss": 1.3701,
      "step": 19584
    },
    {
      "epoch": 1.6850210788952937,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000745380413442985,
      "loss": 1.6732,
      "step": 19585
    },
    {
      "epoch": 1.6851071152026154,
      "grad_norm": 1.3125,
      "learning_rate": 0.000745356852644587,
      "loss": 1.206,
      "step": 19586
    },
    {
      "epoch": 1.6851931515099372,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007453332911285679,
      "loss": 1.1325,
      "step": 19587
    },
    {
      "epoch": 1.6852791878172588,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007453097288949969,
      "loss": 1.7421,
      "step": 19588
    },
    {
      "epoch": 1.6853652241245807,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007452861659439424,
      "loss": 1.5122,
      "step": 19589
    },
    {
      "epoch": 1.6854512604319023,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007452626022754738,
      "loss": 1.1662,
      "step": 19590
    },
    {
      "epoch": 1.685537296739224,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007452390378896596,
      "loss": 1.0663,
      "step": 19591
    },
    {
      "epoch": 1.6856233330465455,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007452154727865688,
      "loss": 1.3655,
      "step": 19592
    },
    {
      "epoch": 1.6857093693538672,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007451919069662707,
      "loss": 1.3471,
      "step": 19593
    },
    {
      "epoch": 1.685795405661189,
      "grad_norm": 2.25,
      "learning_rate": 0.0007451683404288338,
      "loss": 1.4106,
      "step": 19594
    },
    {
      "epoch": 1.6858814419685109,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007451447731743274,
      "loss": 1.4889,
      "step": 19595
    },
    {
      "epoch": 1.6859674782758325,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00074512120520282,
      "loss": 0.553,
      "step": 19596
    },
    {
      "epoch": 1.6860535145831541,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007450976365143809,
      "loss": 0.9033,
      "step": 19597
    },
    {
      "epoch": 1.6861395508904757,
      "grad_norm": 1.125,
      "learning_rate": 0.0007450740671090788,
      "loss": 1.7839,
      "step": 19598
    },
    {
      "epoch": 1.6862255871977974,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007450504969869826,
      "loss": 1.3504,
      "step": 19599
    },
    {
      "epoch": 1.6863116235051192,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007450269261481615,
      "loss": 1.3636,
      "step": 19600
    },
    {
      "epoch": 1.6863976598124408,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007450033545926841,
      "loss": 1.4418,
      "step": 19601
    },
    {
      "epoch": 1.6864836961197627,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007449797823206198,
      "loss": 0.8764,
      "step": 19602
    },
    {
      "epoch": 1.6865697324270843,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007449562093320372,
      "loss": 0.7691,
      "step": 19603
    },
    {
      "epoch": 1.686655768734406,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007449326356270053,
      "loss": 1.5586,
      "step": 19604
    },
    {
      "epoch": 1.6867418050417275,
      "grad_norm": 1.0,
      "learning_rate": 0.0007449090612055931,
      "loss": 1.195,
      "step": 19605
    },
    {
      "epoch": 1.6868278413490492,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007448854860678696,
      "loss": 1.249,
      "step": 19606
    },
    {
      "epoch": 1.686913877656371,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007448619102139036,
      "loss": 1.4033,
      "step": 19607
    },
    {
      "epoch": 1.6869999139636926,
      "grad_norm": 1.453125,
      "learning_rate": 0.000744838333643764,
      "loss": 1.6821,
      "step": 19608
    },
    {
      "epoch": 1.6870859502710145,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00074481475635752,
      "loss": 1.0971,
      "step": 19609
    },
    {
      "epoch": 1.6871719865783361,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007447911783552406,
      "loss": 1.5564,
      "step": 19610
    },
    {
      "epoch": 1.6872580228856577,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007447675996369945,
      "loss": 1.2766,
      "step": 19611
    },
    {
      "epoch": 1.6873440591929794,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007447440202028508,
      "loss": 1.1697,
      "step": 19612
    },
    {
      "epoch": 1.687430095500301,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007447204400528784,
      "loss": 1.3268,
      "step": 19613
    },
    {
      "epoch": 1.6875161318076228,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007446968591871463,
      "loss": 1.0018,
      "step": 19614
    },
    {
      "epoch": 1.6876021681149445,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007446732776057234,
      "loss": 0.8843,
      "step": 19615
    },
    {
      "epoch": 1.6876882044222663,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000744649695308679,
      "loss": 1.2973,
      "step": 19616
    },
    {
      "epoch": 1.687774240729588,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007446261122960817,
      "loss": 0.8066,
      "step": 19617
    },
    {
      "epoch": 1.6878602770369096,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007446025285680005,
      "loss": 1.6423,
      "step": 19618
    },
    {
      "epoch": 1.6879463133442312,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007445789441245046,
      "loss": 0.9899,
      "step": 19619
    },
    {
      "epoch": 1.6880323496515528,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007445553589656626,
      "loss": 1.1513,
      "step": 19620
    },
    {
      "epoch": 1.6881183859588746,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000744531773091544,
      "loss": 1.4125,
      "step": 19621
    },
    {
      "epoch": 1.6882044222661965,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007445081865022175,
      "loss": 1.4043,
      "step": 19622
    },
    {
      "epoch": 1.6882904585735181,
      "grad_norm": 1.09375,
      "learning_rate": 0.000744484599197752,
      "loss": 0.9466,
      "step": 19623
    },
    {
      "epoch": 1.6883764948808397,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007444610111782167,
      "loss": 1.0638,
      "step": 19624
    },
    {
      "epoch": 1.6884625311881614,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007444374224436805,
      "loss": 0.8999,
      "step": 19625
    },
    {
      "epoch": 1.688548567495483,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007444138329942123,
      "loss": 1.3683,
      "step": 19626
    },
    {
      "epoch": 1.6886346038028048,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0007443902428298813,
      "loss": 0.9154,
      "step": 19627
    },
    {
      "epoch": 1.6887206401101265,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0007443666519507562,
      "loss": 1.5637,
      "step": 19628
    },
    {
      "epoch": 1.6888066764174483,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007443430603569064,
      "loss": 1.1222,
      "step": 19629
    },
    {
      "epoch": 1.68889271272477,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007443194680484003,
      "loss": 0.5394,
      "step": 19630
    },
    {
      "epoch": 1.6889787490320916,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007442958750253076,
      "loss": 1.362,
      "step": 19631
    },
    {
      "epoch": 1.6890647853394132,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007442722812876968,
      "loss": 1.6863,
      "step": 19632
    },
    {
      "epoch": 1.6891508216467348,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007442486868356372,
      "loss": 1.4427,
      "step": 19633
    },
    {
      "epoch": 1.6892368579540566,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007442250916691976,
      "loss": 1.2078,
      "step": 19634
    },
    {
      "epoch": 1.6893228942613783,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007442014957884472,
      "loss": 1.5226,
      "step": 19635
    },
    {
      "epoch": 1.6894089305687001,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007441778991934549,
      "loss": 0.8032,
      "step": 19636
    },
    {
      "epoch": 1.6894949668760217,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007441543018842897,
      "loss": 1.12,
      "step": 19637
    },
    {
      "epoch": 1.6895810031833434,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007441307038610205,
      "loss": 1.0859,
      "step": 19638
    },
    {
      "epoch": 1.689667039490665,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007441071051237167,
      "loss": 1.4485,
      "step": 19639
    },
    {
      "epoch": 1.6897530757979866,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000744083505672447,
      "loss": 0.905,
      "step": 19640
    },
    {
      "epoch": 1.6898391121053085,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007440599055072805,
      "loss": 1.4577,
      "step": 19641
    },
    {
      "epoch": 1.68992514841263,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007440363046282865,
      "loss": 1.1961,
      "step": 19642
    },
    {
      "epoch": 1.690011184719952,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007440127030355335,
      "loss": 0.8004,
      "step": 19643
    },
    {
      "epoch": 1.6900972210272736,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007439891007290909,
      "loss": 1.2665,
      "step": 19644
    },
    {
      "epoch": 1.6901832573345952,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007439654977090275,
      "loss": 1.327,
      "step": 19645
    },
    {
      "epoch": 1.6902692936419168,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007439418939754126,
      "loss": 1.1351,
      "step": 19646
    },
    {
      "epoch": 1.6903553299492384,
      "grad_norm": 1.078125,
      "learning_rate": 0.000743918289528315,
      "loss": 1.2201,
      "step": 19647
    },
    {
      "epoch": 1.6904413662565603,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007438946843678039,
      "loss": 1.1606,
      "step": 19648
    },
    {
      "epoch": 1.6905274025638821,
      "grad_norm": 1.125,
      "learning_rate": 0.0007438710784939484,
      "loss": 1.046,
      "step": 19649
    },
    {
      "epoch": 1.6906134388712037,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007438474719068173,
      "loss": 1.2964,
      "step": 19650
    },
    {
      "epoch": 1.6906994751785254,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007438238646064798,
      "loss": 0.8738,
      "step": 19651
    },
    {
      "epoch": 1.690785511485847,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007438002565930048,
      "loss": 1.2599,
      "step": 19652
    },
    {
      "epoch": 1.6908715477931686,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007437766478664617,
      "loss": 0.7469,
      "step": 19653
    },
    {
      "epoch": 1.6909575841004902,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007437530384269192,
      "loss": 1.7538,
      "step": 19654
    },
    {
      "epoch": 1.691043620407812,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007437294282744464,
      "loss": 1.1076,
      "step": 19655
    },
    {
      "epoch": 1.691129656715134,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007437058174091125,
      "loss": 2.0958,
      "step": 19656
    },
    {
      "epoch": 1.6912156930224556,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007436822058309864,
      "loss": 1.4014,
      "step": 19657
    },
    {
      "epoch": 1.6913017293297772,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007436585935401374,
      "loss": 1.1938,
      "step": 19658
    },
    {
      "epoch": 1.6913877656370988,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007436349805366343,
      "loss": 1.9822,
      "step": 19659
    },
    {
      "epoch": 1.6914738019444204,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007436113668205462,
      "loss": 1.316,
      "step": 19660
    },
    {
      "epoch": 1.6915598382517423,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0007435877523919423,
      "loss": 1.4489,
      "step": 19661
    },
    {
      "epoch": 1.691645874559064,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007435641372508917,
      "loss": 1.329,
      "step": 19662
    },
    {
      "epoch": 1.6917319108663857,
      "grad_norm": 1.0,
      "learning_rate": 0.0007435405213974633,
      "loss": 1.2831,
      "step": 19663
    },
    {
      "epoch": 1.6918179471737074,
      "grad_norm": 1.78125,
      "learning_rate": 0.0007435169048317262,
      "loss": 1.6575,
      "step": 19664
    },
    {
      "epoch": 1.691903983481029,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007434932875537496,
      "loss": 0.9779,
      "step": 19665
    },
    {
      "epoch": 1.6919900197883506,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007434696695636024,
      "loss": 1.2722,
      "step": 19666
    },
    {
      "epoch": 1.6920760560956722,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007434460508613538,
      "loss": 1.0892,
      "step": 19667
    },
    {
      "epoch": 1.692162092402994,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007434224314470728,
      "loss": 1.6154,
      "step": 19668
    },
    {
      "epoch": 1.6922481287103157,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007433988113208285,
      "loss": 1.2703,
      "step": 19669
    },
    {
      "epoch": 1.6923341650176376,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007433751904826902,
      "loss": 1.6807,
      "step": 19670
    },
    {
      "epoch": 1.6924202013249592,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007433515689327266,
      "loss": 1.1718,
      "step": 19671
    },
    {
      "epoch": 1.6925062376322808,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007433279466710071,
      "loss": 1.4485,
      "step": 19672
    },
    {
      "epoch": 1.6925922739396024,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007433043236976007,
      "loss": 0.9564,
      "step": 19673
    },
    {
      "epoch": 1.692678310246924,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007432807000125765,
      "loss": 1.3268,
      "step": 19674
    },
    {
      "epoch": 1.692764346554246,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007432570756160034,
      "loss": 1.5438,
      "step": 19675
    },
    {
      "epoch": 1.6928503828615675,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007432334505079508,
      "loss": 1.1134,
      "step": 19676
    },
    {
      "epoch": 1.6929364191688894,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007432098246884876,
      "loss": 0.8299,
      "step": 19677
    },
    {
      "epoch": 1.693022455476211,
      "grad_norm": 1.4375,
      "learning_rate": 0.000743186198157683,
      "loss": 1.759,
      "step": 19678
    },
    {
      "epoch": 1.6931084917835326,
      "grad_norm": 1.046875,
      "learning_rate": 0.000743162570915606,
      "loss": 1.439,
      "step": 19679
    },
    {
      "epoch": 1.6931945280908542,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007431389429623259,
      "loss": 1.0033,
      "step": 19680
    },
    {
      "epoch": 1.6932805643981759,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007431153142979117,
      "loss": 1.2884,
      "step": 19681
    },
    {
      "epoch": 1.6933666007054977,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007430916849224323,
      "loss": 1.9487,
      "step": 19682
    },
    {
      "epoch": 1.6934526370128196,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007430680548359571,
      "loss": 0.9901,
      "step": 19683
    },
    {
      "epoch": 1.6935386733201412,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007430444240385552,
      "loss": 1.3851,
      "step": 19684
    },
    {
      "epoch": 1.6936247096274628,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007430207925302956,
      "loss": 1.0038,
      "step": 19685
    },
    {
      "epoch": 1.6937107459347844,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007429971603112474,
      "loss": 1.2934,
      "step": 19686
    },
    {
      "epoch": 1.693796782242106,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007429735273814797,
      "loss": 1.0743,
      "step": 19687
    },
    {
      "epoch": 1.693882818549428,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007429498937410618,
      "loss": 1.6943,
      "step": 19688
    },
    {
      "epoch": 1.6939688548567495,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007429262593900626,
      "loss": 1.2789,
      "step": 19689
    },
    {
      "epoch": 1.6940548911640714,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007429026243285514,
      "loss": 1.9161,
      "step": 19690
    },
    {
      "epoch": 1.694140927471393,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007428789885565972,
      "loss": 2.4746,
      "step": 19691
    },
    {
      "epoch": 1.6942269637787146,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007428553520742693,
      "loss": 0.9417,
      "step": 19692
    },
    {
      "epoch": 1.6943130000860362,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007428317148816367,
      "loss": 0.696,
      "step": 19693
    },
    {
      "epoch": 1.6943990363933579,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007428080769787684,
      "loss": 1.5648,
      "step": 19694
    },
    {
      "epoch": 1.6944850727006797,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007427844383657339,
      "loss": 1.4742,
      "step": 19695
    },
    {
      "epoch": 1.6945711090080013,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007427607990426022,
      "loss": 1.7423,
      "step": 19696
    },
    {
      "epoch": 1.6946571453153232,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007427371590094422,
      "loss": 1.0898,
      "step": 19697
    },
    {
      "epoch": 1.6947431816226448,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007427135182663233,
      "loss": 1.4175,
      "step": 19698
    },
    {
      "epoch": 1.6948292179299664,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007426898768133146,
      "loss": 2.1745,
      "step": 19699
    },
    {
      "epoch": 1.694915254237288,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007426662346504851,
      "loss": 1.3001,
      "step": 19700
    },
    {
      "epoch": 1.6950012905446097,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007426425917779041,
      "loss": 1.3611,
      "step": 19701
    },
    {
      "epoch": 1.6950873268519315,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007426189481956407,
      "loss": 1.2154,
      "step": 19702
    },
    {
      "epoch": 1.6951733631592532,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007425953039037642,
      "loss": 1.1046,
      "step": 19703
    },
    {
      "epoch": 1.695259399466575,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007425716589023434,
      "loss": 0.9908,
      "step": 19704
    },
    {
      "epoch": 1.6953454357738966,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007425480131914478,
      "loss": 1.1096,
      "step": 19705
    },
    {
      "epoch": 1.6954314720812182,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007425243667711464,
      "loss": 1.7614,
      "step": 19706
    },
    {
      "epoch": 1.6955175083885399,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007425007196415086,
      "loss": 1.011,
      "step": 19707
    },
    {
      "epoch": 1.6956035446958615,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000742477071802603,
      "loss": 0.9314,
      "step": 19708
    },
    {
      "epoch": 1.6956895810031833,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0007424534232544993,
      "loss": 2.2228,
      "step": 19709
    },
    {
      "epoch": 1.6957756173105052,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007424297739972665,
      "loss": 2.1641,
      "step": 19710
    },
    {
      "epoch": 1.6958616536178268,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007424061240309739,
      "loss": 1.4488,
      "step": 19711
    },
    {
      "epoch": 1.6959476899251484,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007423824733556901,
      "loss": 1.1998,
      "step": 19712
    },
    {
      "epoch": 1.69603372623247,
      "grad_norm": 1.59375,
      "learning_rate": 0.000742358821971485,
      "loss": 1.2556,
      "step": 19713
    },
    {
      "epoch": 1.6961197625397917,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007423351698784275,
      "loss": 1.6474,
      "step": 19714
    },
    {
      "epoch": 1.6962057988471135,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007423115170765868,
      "loss": 1.4673,
      "step": 19715
    },
    {
      "epoch": 1.6962918351544352,
      "grad_norm": 1.375,
      "learning_rate": 0.000742287863566032,
      "loss": 0.6879,
      "step": 19716
    },
    {
      "epoch": 1.696377871461757,
      "grad_norm": 1.0,
      "learning_rate": 0.0007422642093468323,
      "loss": 1.7439,
      "step": 19717
    },
    {
      "epoch": 1.6964639077690786,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007422405544190569,
      "loss": 1.4211,
      "step": 19718
    },
    {
      "epoch": 1.6965499440764003,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000742216898782775,
      "loss": 0.8211,
      "step": 19719
    },
    {
      "epoch": 1.6966359803837219,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007421932424380558,
      "loss": 1.8202,
      "step": 19720
    },
    {
      "epoch": 1.6967220166910435,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007421695853849684,
      "loss": 1.2674,
      "step": 19721
    },
    {
      "epoch": 1.6968080529983653,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007421459276235822,
      "loss": 1.2465,
      "step": 19722
    },
    {
      "epoch": 1.696894089305687,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007421222691539659,
      "loss": 1.8016,
      "step": 19723
    },
    {
      "epoch": 1.6969801256130088,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007420986099761895,
      "loss": 0.8253,
      "step": 19724
    },
    {
      "epoch": 1.6970661619203304,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007420749500903215,
      "loss": 1.129,
      "step": 19725
    },
    {
      "epoch": 1.697152198227652,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007420512894964316,
      "loss": 0.9622,
      "step": 19726
    },
    {
      "epoch": 1.6972382345349737,
      "grad_norm": 1.25,
      "learning_rate": 0.0007420276281945884,
      "loss": 1.2347,
      "step": 19727
    },
    {
      "epoch": 1.6973242708422953,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007420039661848617,
      "loss": 1.3741,
      "step": 19728
    },
    {
      "epoch": 1.6974103071496172,
      "grad_norm": 26.0,
      "learning_rate": 0.0007419803034673203,
      "loss": 1.4742,
      "step": 19729
    },
    {
      "epoch": 1.6974963434569388,
      "grad_norm": 21.125,
      "learning_rate": 0.0007419566400420337,
      "loss": 1.3715,
      "step": 19730
    },
    {
      "epoch": 1.6975823797642606,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007419329759090708,
      "loss": 0.8601,
      "step": 19731
    },
    {
      "epoch": 1.6976684160715823,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007419093110685013,
      "loss": 0.8302,
      "step": 19732
    },
    {
      "epoch": 1.6977544523789039,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007418856455203939,
      "loss": 1.3529,
      "step": 19733
    },
    {
      "epoch": 1.6978404886862255,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007418619792648181,
      "loss": 1.0223,
      "step": 19734
    },
    {
      "epoch": 1.6979265249935471,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007418383123018431,
      "loss": 0.7494,
      "step": 19735
    },
    {
      "epoch": 1.698012561300869,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000741814644631538,
      "loss": 0.9267,
      "step": 19736
    },
    {
      "epoch": 1.6980985976081908,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007417909762539721,
      "loss": 1.1652,
      "step": 19737
    },
    {
      "epoch": 1.6981846339155124,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007417673071692146,
      "loss": 2.0762,
      "step": 19738
    },
    {
      "epoch": 1.698270670222834,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007417436373773347,
      "loss": 1.7231,
      "step": 19739
    },
    {
      "epoch": 1.6983567065301557,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007417199668784019,
      "loss": 1.9031,
      "step": 19740
    },
    {
      "epoch": 1.6984427428374773,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007416962956724851,
      "loss": 1.3864,
      "step": 19741
    },
    {
      "epoch": 1.698528779144799,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007416726237596535,
      "loss": 1.329,
      "step": 19742
    },
    {
      "epoch": 1.6986148154521208,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007416489511399767,
      "loss": 1.3894,
      "step": 19743
    },
    {
      "epoch": 1.6987008517594426,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007416252778135236,
      "loss": 0.4851,
      "step": 19744
    },
    {
      "epoch": 1.6987868880667643,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007416016037803634,
      "loss": 1.2627,
      "step": 19745
    },
    {
      "epoch": 1.6988729243740859,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007415779290405657,
      "loss": 1.4628,
      "step": 19746
    },
    {
      "epoch": 1.6989589606814075,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007415542535941996,
      "loss": 1.2593,
      "step": 19747
    },
    {
      "epoch": 1.6990449969887291,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007415305774413342,
      "loss": 1.2831,
      "step": 19748
    },
    {
      "epoch": 1.699131033296051,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007415069005820387,
      "loss": 1.3199,
      "step": 19749
    },
    {
      "epoch": 1.6992170696033726,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007414832230163826,
      "loss": 0.9306,
      "step": 19750
    },
    {
      "epoch": 1.6993031059106944,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007414595447444351,
      "loss": 1.3651,
      "step": 19751
    },
    {
      "epoch": 1.699389142218016,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007414358657662655,
      "loss": 1.4745,
      "step": 19752
    },
    {
      "epoch": 1.6994751785253377,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007414121860819426,
      "loss": 1.5155,
      "step": 19753
    },
    {
      "epoch": 1.6995612148326593,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007413885056915362,
      "loss": 1.0775,
      "step": 19754
    },
    {
      "epoch": 1.699647251139981,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007413648245951155,
      "loss": 1.1241,
      "step": 19755
    },
    {
      "epoch": 1.6997332874473028,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007413411427927495,
      "loss": 1.2686,
      "step": 19756
    },
    {
      "epoch": 1.6998193237546244,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007413174602845075,
      "loss": 1.0435,
      "step": 19757
    },
    {
      "epoch": 1.6999053600619463,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007412937770704589,
      "loss": 1.3167,
      "step": 19758
    },
    {
      "epoch": 1.6999913963692679,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007412700931506731,
      "loss": 1.5861,
      "step": 19759
    },
    {
      "epoch": 1.7000774326765895,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007412464085252191,
      "loss": 1.6607,
      "step": 19760
    },
    {
      "epoch": 1.7001634689839111,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007412227231941662,
      "loss": 1.4054,
      "step": 19761
    },
    {
      "epoch": 1.7002495052912328,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007411990371575839,
      "loss": 1.9414,
      "step": 19762
    },
    {
      "epoch": 1.7003355415985546,
      "grad_norm": 1.0,
      "learning_rate": 0.0007411753504155412,
      "loss": 1.3783,
      "step": 19763
    },
    {
      "epoch": 1.7004215779058762,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007411516629681074,
      "loss": 1.3043,
      "step": 19764
    },
    {
      "epoch": 1.700507614213198,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007411279748153521,
      "loss": 1.4361,
      "step": 19765
    },
    {
      "epoch": 1.7005936505205197,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007411042859573443,
      "loss": 1.6211,
      "step": 19766
    },
    {
      "epoch": 1.7006796868278413,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007410805963941534,
      "loss": 0.7635,
      "step": 19767
    },
    {
      "epoch": 1.700765723135163,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007410569061258485,
      "loss": 1.4239,
      "step": 19768
    },
    {
      "epoch": 1.7008517594424846,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007410332151524992,
      "loss": 0.9007,
      "step": 19769
    },
    {
      "epoch": 1.7009377957498064,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007410095234741746,
      "loss": 0.9426,
      "step": 19770
    },
    {
      "epoch": 1.7010238320571283,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000740985831090944,
      "loss": 1.2678,
      "step": 19771
    },
    {
      "epoch": 1.7011098683644499,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007409621380028766,
      "loss": 0.6793,
      "step": 19772
    },
    {
      "epoch": 1.7011959046717715,
      "grad_norm": 1.4375,
      "learning_rate": 0.000740938444210042,
      "loss": 1.8709,
      "step": 19773
    },
    {
      "epoch": 1.7012819409790931,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007409147497125092,
      "loss": 0.79,
      "step": 19774
    },
    {
      "epoch": 1.7013679772864148,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007408910545103477,
      "loss": 1.3373,
      "step": 19775
    },
    {
      "epoch": 1.7014540135937366,
      "grad_norm": 1.375,
      "learning_rate": 0.0007408673586036265,
      "loss": 1.1784,
      "step": 19776
    },
    {
      "epoch": 1.7015400499010582,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007408436619924152,
      "loss": 1.7285,
      "step": 19777
    },
    {
      "epoch": 1.70162608620838,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007408199646767832,
      "loss": 0.9205,
      "step": 19778
    },
    {
      "epoch": 1.7017121225157017,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007407962666567996,
      "loss": 0.9143,
      "step": 19779
    },
    {
      "epoch": 1.7017981588230233,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007407725679325337,
      "loss": 1.6537,
      "step": 19780
    },
    {
      "epoch": 1.701884195130345,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007407488685040548,
      "loss": 1.0933,
      "step": 19781
    },
    {
      "epoch": 1.7019702314376666,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007407251683714324,
      "loss": 0.6975,
      "step": 19782
    },
    {
      "epoch": 1.7020562677449884,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007407014675347355,
      "loss": 1.61,
      "step": 19783
    },
    {
      "epoch": 1.70214230405231,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007406777659940339,
      "loss": 1.1272,
      "step": 19784
    },
    {
      "epoch": 1.7022283403596319,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007406540637493965,
      "loss": 0.8615,
      "step": 19785
    },
    {
      "epoch": 1.7023143766669535,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007406303608008927,
      "loss": 0.7941,
      "step": 19786
    },
    {
      "epoch": 1.7024004129742751,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007406066571485918,
      "loss": 1.1378,
      "step": 19787
    },
    {
      "epoch": 1.7024864492815968,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007405829527925635,
      "loss": 2.0762,
      "step": 19788
    },
    {
      "epoch": 1.7025724855889184,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007405592477328768,
      "loss": 0.8563,
      "step": 19789
    },
    {
      "epoch": 1.7026585218962402,
      "grad_norm": 1.734375,
      "learning_rate": 0.0007405355419696009,
      "loss": 2.2063,
      "step": 19790
    },
    {
      "epoch": 1.7027445582035619,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007405118355028055,
      "loss": 0.6605,
      "step": 19791
    },
    {
      "epoch": 1.7028305945108837,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007404881283325597,
      "loss": 0.6823,
      "step": 19792
    },
    {
      "epoch": 1.7029166308182053,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0007404644204589329,
      "loss": 0.6125,
      "step": 19793
    },
    {
      "epoch": 1.703002667125527,
      "grad_norm": 1.0,
      "learning_rate": 0.0007404407118819945,
      "loss": 1.3772,
      "step": 19794
    },
    {
      "epoch": 1.7030887034328486,
      "grad_norm": 1.71875,
      "learning_rate": 0.0007404170026018137,
      "loss": 1.734,
      "step": 19795
    },
    {
      "epoch": 1.7031747397401702,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00074039329261846,
      "loss": 1.7198,
      "step": 19796
    },
    {
      "epoch": 1.703260776047492,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007403695819320026,
      "loss": 1.6155,
      "step": 19797
    },
    {
      "epoch": 1.7033468123548139,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0007403458705425109,
      "loss": 1.2079,
      "step": 19798
    },
    {
      "epoch": 1.7034328486621355,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007403221584500544,
      "loss": 0.7381,
      "step": 19799
    },
    {
      "epoch": 1.7035188849694571,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007402984456547022,
      "loss": 1.2426,
      "step": 19800
    },
    {
      "epoch": 1.7036049212767788,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007402747321565237,
      "loss": 1.4959,
      "step": 19801
    },
    {
      "epoch": 1.7036909575841004,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007402510179555886,
      "loss": 1.4488,
      "step": 19802
    },
    {
      "epoch": 1.7037769938914222,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007402273030519659,
      "loss": 1.3049,
      "step": 19803
    },
    {
      "epoch": 1.7038630301987439,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000740203587445725,
      "loss": 1.0399,
      "step": 19804
    },
    {
      "epoch": 1.7039490665060657,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007401798711369353,
      "loss": 1.0054,
      "step": 19805
    },
    {
      "epoch": 1.7040351028133873,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007401561541256662,
      "loss": 0.7919,
      "step": 19806
    },
    {
      "epoch": 1.704121139120709,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007401324364119872,
      "loss": 1.4607,
      "step": 19807
    },
    {
      "epoch": 1.7042071754280306,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007401087179959673,
      "loss": 0.8289,
      "step": 19808
    },
    {
      "epoch": 1.7042932117353522,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007400849988776762,
      "loss": 2.1782,
      "step": 19809
    },
    {
      "epoch": 1.704379248042674,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007400612790571833,
      "loss": 1.3438,
      "step": 19810
    },
    {
      "epoch": 1.7044652843499957,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007400375585345577,
      "loss": 1.6976,
      "step": 19811
    },
    {
      "epoch": 1.7045513206573175,
      "grad_norm": 1.5,
      "learning_rate": 0.0007400138373098691,
      "loss": 1.3828,
      "step": 19812
    },
    {
      "epoch": 1.7046373569646391,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007399901153831863,
      "loss": 0.5714,
      "step": 19813
    },
    {
      "epoch": 1.7047233932719608,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007399663927545794,
      "loss": 1.0679,
      "step": 19814
    },
    {
      "epoch": 1.7048094295792824,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007399426694241175,
      "loss": 1.366,
      "step": 19815
    },
    {
      "epoch": 1.704895465886604,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00073991894539187,
      "loss": 1.4795,
      "step": 19816
    },
    {
      "epoch": 1.7049815021939259,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007398952206579061,
      "loss": 1.2533,
      "step": 19817
    },
    {
      "epoch": 1.7050675385012475,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007398714952222954,
      "loss": 0.9807,
      "step": 19818
    },
    {
      "epoch": 1.7051535748085693,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007398477690851071,
      "loss": 0.8018,
      "step": 19819
    },
    {
      "epoch": 1.705239611115891,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007398240422464108,
      "loss": 1.919,
      "step": 19820
    },
    {
      "epoch": 1.7053256474232126,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000739800314706276,
      "loss": 1.5712,
      "step": 19821
    },
    {
      "epoch": 1.7054116837305342,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007397765864647716,
      "loss": 1.4073,
      "step": 19822
    },
    {
      "epoch": 1.7054977200378558,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007397528575219675,
      "loss": 1.3519,
      "step": 19823
    },
    {
      "epoch": 1.7055837563451777,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007397291278779328,
      "loss": 1.1844,
      "step": 19824
    },
    {
      "epoch": 1.7056697926524995,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007397053975327372,
      "loss": 1.6862,
      "step": 19825
    },
    {
      "epoch": 1.7057558289598211,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007396816664864499,
      "loss": 0.742,
      "step": 19826
    },
    {
      "epoch": 1.7058418652671428,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0007396579347391404,
      "loss": 1.9574,
      "step": 19827
    },
    {
      "epoch": 1.7059279015744644,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007396342022908777,
      "loss": 0.8889,
      "step": 19828
    },
    {
      "epoch": 1.706013937881786,
      "grad_norm": 1.0,
      "learning_rate": 0.0007396104691417319,
      "loss": 1.181,
      "step": 19829
    },
    {
      "epoch": 1.7060999741891076,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000739586735291772,
      "loss": 0.9704,
      "step": 19830
    },
    {
      "epoch": 1.7061860104964295,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007395630007410676,
      "loss": 1.2352,
      "step": 19831
    },
    {
      "epoch": 1.7062720468037513,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007395392654896878,
      "loss": 0.6459,
      "step": 19832
    },
    {
      "epoch": 1.706358083111073,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007395155295377024,
      "loss": 1.5593,
      "step": 19833
    },
    {
      "epoch": 1.7064441194183946,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007394917928851807,
      "loss": 1.2089,
      "step": 19834
    },
    {
      "epoch": 1.7065301557257162,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007394680555321919,
      "loss": 1.0571,
      "step": 19835
    },
    {
      "epoch": 1.7066161920330378,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007394443174788057,
      "loss": 1.3828,
      "step": 19836
    },
    {
      "epoch": 1.7067022283403597,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007394205787250915,
      "loss": 1.9311,
      "step": 19837
    },
    {
      "epoch": 1.7067882646476813,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007393968392711186,
      "loss": 1.4507,
      "step": 19838
    },
    {
      "epoch": 1.7068743009550031,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007393730991169565,
      "loss": 1.0184,
      "step": 19839
    },
    {
      "epoch": 1.7069603372623248,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007393493582626747,
      "loss": 0.9765,
      "step": 19840
    },
    {
      "epoch": 1.7070463735696464,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007393256167083426,
      "loss": 1.2757,
      "step": 19841
    },
    {
      "epoch": 1.707132409876968,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007393018744540294,
      "loss": 1.5599,
      "step": 19842
    },
    {
      "epoch": 1.7072184461842896,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007392781314998049,
      "loss": 1.0244,
      "step": 19843
    },
    {
      "epoch": 1.7073044824916115,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007392543878457384,
      "loss": 1.2817,
      "step": 19844
    },
    {
      "epoch": 1.707390518798933,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007392306434918994,
      "loss": 1.3099,
      "step": 19845
    },
    {
      "epoch": 1.707476555106255,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007392068984383573,
      "loss": 1.0598,
      "step": 19846
    },
    {
      "epoch": 1.7075625914135766,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007391831526851815,
      "loss": 2.2104,
      "step": 19847
    },
    {
      "epoch": 1.7076486277208982,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007391594062324415,
      "loss": 1.5978,
      "step": 19848
    },
    {
      "epoch": 1.7077346640282198,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007391356590802068,
      "loss": 1.0719,
      "step": 19849
    },
    {
      "epoch": 1.7078207003355415,
      "grad_norm": 1.125,
      "learning_rate": 0.0007391119112285466,
      "loss": 1.5171,
      "step": 19850
    },
    {
      "epoch": 1.7079067366428633,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007390881626775308,
      "loss": 0.7186,
      "step": 19851
    },
    {
      "epoch": 1.707992772950185,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007390644134272285,
      "loss": 1.2477,
      "step": 19852
    },
    {
      "epoch": 1.7080788092575068,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007390406634777093,
      "loss": 1.2252,
      "step": 19853
    },
    {
      "epoch": 1.7081648455648284,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007390169128290427,
      "loss": 2.2908,
      "step": 19854
    },
    {
      "epoch": 1.70825088187215,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007389931614812981,
      "loss": 1.2318,
      "step": 19855
    },
    {
      "epoch": 1.7083369181794716,
      "grad_norm": 1.953125,
      "learning_rate": 0.0007389694094345451,
      "loss": 1.4014,
      "step": 19856
    },
    {
      "epoch": 1.7084229544867933,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007389456566888528,
      "loss": 1.418,
      "step": 19857
    },
    {
      "epoch": 1.7085089907941151,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007389219032442911,
      "loss": 0.9949,
      "step": 19858
    },
    {
      "epoch": 1.708595027101437,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0007388981491009293,
      "loss": 0.5865,
      "step": 19859
    },
    {
      "epoch": 1.7086810634087586,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007388743942588369,
      "loss": 1.2489,
      "step": 19860
    },
    {
      "epoch": 1.7087670997160802,
      "grad_norm": 2.328125,
      "learning_rate": 0.0007388506387180832,
      "loss": 1.1936,
      "step": 19861
    },
    {
      "epoch": 1.7088531360234018,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007388268824787378,
      "loss": 1.6007,
      "step": 19862
    },
    {
      "epoch": 1.7089391723307235,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007388031255408705,
      "loss": 1.6124,
      "step": 19863
    },
    {
      "epoch": 1.7090252086380453,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007387793679045503,
      "loss": 1.6099,
      "step": 19864
    },
    {
      "epoch": 1.709111244945367,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007387556095698469,
      "loss": 1.2765,
      "step": 19865
    },
    {
      "epoch": 1.7091972812526888,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007387318505368299,
      "loss": 1.6784,
      "step": 19866
    },
    {
      "epoch": 1.7092833175600104,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007387080908055686,
      "loss": 1.8258,
      "step": 19867
    },
    {
      "epoch": 1.709369353867332,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007386843303761326,
      "loss": 1.8491,
      "step": 19868
    },
    {
      "epoch": 1.7094553901746536,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007386605692485912,
      "loss": 0.7042,
      "step": 19869
    },
    {
      "epoch": 1.7095414264819753,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007386368074230142,
      "loss": 0.6402,
      "step": 19870
    },
    {
      "epoch": 1.7096274627892971,
      "grad_norm": 1.28125,
      "learning_rate": 0.000738613044899471,
      "loss": 1.3281,
      "step": 19871
    },
    {
      "epoch": 1.7097134990966187,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007385892816780309,
      "loss": 1.3215,
      "step": 19872
    },
    {
      "epoch": 1.7097995354039406,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007385655177587635,
      "loss": 0.9687,
      "step": 19873
    },
    {
      "epoch": 1.7098855717112622,
      "grad_norm": 1.75,
      "learning_rate": 0.0007385417531417385,
      "loss": 1.8373,
      "step": 19874
    },
    {
      "epoch": 1.7099716080185838,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007385179878270254,
      "loss": 1.4769,
      "step": 19875
    },
    {
      "epoch": 1.7100576443259055,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007384942218146933,
      "loss": 1.3647,
      "step": 19876
    },
    {
      "epoch": 1.710143680633227,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007384704551048121,
      "loss": 1.2719,
      "step": 19877
    },
    {
      "epoch": 1.710229716940549,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007384466876974512,
      "loss": 1.0735,
      "step": 19878
    },
    {
      "epoch": 1.7103157532478706,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007384229195926801,
      "loss": 1.1819,
      "step": 19879
    },
    {
      "epoch": 1.7104017895551924,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007383991507905683,
      "loss": 0.5848,
      "step": 19880
    },
    {
      "epoch": 1.710487825862514,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007383753812911854,
      "loss": 1.6001,
      "step": 19881
    },
    {
      "epoch": 1.7105738621698356,
      "grad_norm": 1.0625,
      "learning_rate": 0.000738351611094601,
      "loss": 0.9906,
      "step": 19882
    },
    {
      "epoch": 1.7106598984771573,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007383278402008842,
      "loss": 1.6708,
      "step": 19883
    },
    {
      "epoch": 1.710745934784479,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000738304068610105,
      "loss": 0.8481,
      "step": 19884
    },
    {
      "epoch": 1.7108319710918007,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007382802963223328,
      "loss": 1.6248,
      "step": 19885
    },
    {
      "epoch": 1.7109180073991226,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000738256523337637,
      "loss": 0.7852,
      "step": 19886
    },
    {
      "epoch": 1.7110040437064442,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007382327496560872,
      "loss": 0.7406,
      "step": 19887
    },
    {
      "epoch": 1.7110900800137658,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007382089752777528,
      "loss": 0.9705,
      "step": 19888
    },
    {
      "epoch": 1.7111761163210875,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007381852002027037,
      "loss": 1.314,
      "step": 19889
    },
    {
      "epoch": 1.711262152628409,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007381614244310092,
      "loss": 1.0974,
      "step": 19890
    },
    {
      "epoch": 1.711348188935731,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007381376479627386,
      "loss": 1.2874,
      "step": 19891
    },
    {
      "epoch": 1.7114342252430526,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000738113870797962,
      "loss": 1.5219,
      "step": 19892
    },
    {
      "epoch": 1.7115202615503744,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007380900929367485,
      "loss": 0.6554,
      "step": 19893
    },
    {
      "epoch": 1.711606297857696,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007380663143791679,
      "loss": 1.8237,
      "step": 19894
    },
    {
      "epoch": 1.7116923341650176,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007380425351252895,
      "loss": 0.5934,
      "step": 19895
    },
    {
      "epoch": 1.7117783704723393,
      "grad_norm": 0.875,
      "learning_rate": 0.000738018755175183,
      "loss": 1.3354,
      "step": 19896
    },
    {
      "epoch": 1.711864406779661,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007379949745289179,
      "loss": 1.5105,
      "step": 19897
    },
    {
      "epoch": 1.7119504430869827,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007379711931865639,
      "loss": 1.6586,
      "step": 19898
    },
    {
      "epoch": 1.7120364793943044,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007379474111481903,
      "loss": 1.3231,
      "step": 19899
    },
    {
      "epoch": 1.7121225157016262,
      "grad_norm": 1.046875,
      "learning_rate": 0.000737923628413867,
      "loss": 0.9585,
      "step": 19900
    },
    {
      "epoch": 1.7122085520089478,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007378998449836631,
      "loss": 1.1544,
      "step": 19901
    },
    {
      "epoch": 1.7122945883162695,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007378760608576487,
      "loss": 0.9518,
      "step": 19902
    },
    {
      "epoch": 1.712380624623591,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007378522760358929,
      "loss": 1.3926,
      "step": 19903
    },
    {
      "epoch": 1.7124666609309127,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007378284905184655,
      "loss": 1.1357,
      "step": 19904
    },
    {
      "epoch": 1.7125526972382346,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007378047043054361,
      "loss": 1.2892,
      "step": 19905
    },
    {
      "epoch": 1.7126387335455562,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000737780917396874,
      "loss": 1.3443,
      "step": 19906
    },
    {
      "epoch": 1.712724769852878,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007377571297928491,
      "loss": 0.8071,
      "step": 19907
    },
    {
      "epoch": 1.7128108061601997,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007377333414934308,
      "loss": 0.6658,
      "step": 19908
    },
    {
      "epoch": 1.7128968424675213,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007377095524986889,
      "loss": 1.6666,
      "step": 19909
    },
    {
      "epoch": 1.712982878774843,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007376857628086925,
      "loss": 1.6666,
      "step": 19910
    },
    {
      "epoch": 1.7130689150821645,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007376619724235116,
      "loss": 1.2385,
      "step": 19911
    },
    {
      "epoch": 1.7131549513894864,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007376381813432157,
      "loss": 1.0298,
      "step": 19912
    },
    {
      "epoch": 1.7132409876968082,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007376143895678743,
      "loss": 1.3481,
      "step": 19913
    },
    {
      "epoch": 1.7133270240041298,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007375905970975569,
      "loss": 1.2281,
      "step": 19914
    },
    {
      "epoch": 1.7134130603114515,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007375668039323333,
      "loss": 1.3705,
      "step": 19915
    },
    {
      "epoch": 1.713499096618773,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007375430100722731,
      "loss": 1.2632,
      "step": 19916
    },
    {
      "epoch": 1.7135851329260947,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007375192155174458,
      "loss": 1.3792,
      "step": 19917
    },
    {
      "epoch": 1.7136711692334163,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007374954202679207,
      "loss": 1.0319,
      "step": 19918
    },
    {
      "epoch": 1.7137572055407382,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007374716243237679,
      "loss": 0.9733,
      "step": 19919
    },
    {
      "epoch": 1.71384324184806,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007374478276850569,
      "loss": 0.9017,
      "step": 19920
    },
    {
      "epoch": 1.7139292781553817,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007374240303518569,
      "loss": 1.5012,
      "step": 19921
    },
    {
      "epoch": 1.7140153144627033,
      "grad_norm": 1.265625,
      "learning_rate": 0.000737400232324238,
      "loss": 1.3368,
      "step": 19922
    },
    {
      "epoch": 1.714101350770025,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007373764336022696,
      "loss": 1.1284,
      "step": 19923
    },
    {
      "epoch": 1.7141873870773465,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007373526341860212,
      "loss": 0.8118,
      "step": 19924
    },
    {
      "epoch": 1.7142734233846684,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007373288340755626,
      "loss": 0.9365,
      "step": 19925
    },
    {
      "epoch": 1.71435945969199,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007373050332709632,
      "loss": 1.4966,
      "step": 19926
    },
    {
      "epoch": 1.7144454959993118,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007372812317722929,
      "loss": 1.3634,
      "step": 19927
    },
    {
      "epoch": 1.7145315323066335,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007372574295796211,
      "loss": 1.7554,
      "step": 19928
    },
    {
      "epoch": 1.714617568613955,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007372336266930173,
      "loss": 1.1304,
      "step": 19929
    },
    {
      "epoch": 1.7147036049212767,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007372098231125514,
      "loss": 1.051,
      "step": 19930
    },
    {
      "epoch": 1.7147896412285983,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007371860188382929,
      "loss": 1.0445,
      "step": 19931
    },
    {
      "epoch": 1.7148756775359202,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007371622138703113,
      "loss": 0.7155,
      "step": 19932
    },
    {
      "epoch": 1.7149617138432418,
      "grad_norm": 2.34375,
      "learning_rate": 0.0007371384082086765,
      "loss": 1.6056,
      "step": 19933
    },
    {
      "epoch": 1.7150477501505637,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007371146018534579,
      "loss": 0.7506,
      "step": 19934
    },
    {
      "epoch": 1.7151337864578853,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007370907948047253,
      "loss": 1.7105,
      "step": 19935
    },
    {
      "epoch": 1.715219822765207,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007370669870625481,
      "loss": 1.437,
      "step": 19936
    },
    {
      "epoch": 1.7153058590725285,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007370431786269962,
      "loss": 0.6268,
      "step": 19937
    },
    {
      "epoch": 1.7153918953798502,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007370193694981391,
      "loss": 1.6417,
      "step": 19938
    },
    {
      "epoch": 1.715477931687172,
      "grad_norm": 1.125,
      "learning_rate": 0.0007369955596760464,
      "loss": 0.9691,
      "step": 19939
    },
    {
      "epoch": 1.7155639679944936,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0007369717491607877,
      "loss": 1.006,
      "step": 19940
    },
    {
      "epoch": 1.7156500043018155,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007369479379524329,
      "loss": 1.4977,
      "step": 19941
    },
    {
      "epoch": 1.715736040609137,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007369241260510513,
      "loss": 1.7726,
      "step": 19942
    },
    {
      "epoch": 1.7158220769164587,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007369003134567129,
      "loss": 1.2347,
      "step": 19943
    },
    {
      "epoch": 1.7159081132237803,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007368765001694869,
      "loss": 1.0458,
      "step": 19944
    },
    {
      "epoch": 1.715994149531102,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007368526861894433,
      "loss": 0.8592,
      "step": 19945
    },
    {
      "epoch": 1.7160801858384238,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007368288715166518,
      "loss": 2.0665,
      "step": 19946
    },
    {
      "epoch": 1.7161662221457457,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007368050561511818,
      "loss": 1.0788,
      "step": 19947
    },
    {
      "epoch": 1.7162522584530673,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007367812400931031,
      "loss": 0.7668,
      "step": 19948
    },
    {
      "epoch": 1.716338294760389,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007367574233424853,
      "loss": 1.0639,
      "step": 19949
    },
    {
      "epoch": 1.7164243310677105,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007367336058993981,
      "loss": 1.0333,
      "step": 19950
    },
    {
      "epoch": 1.7165103673750322,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007367097877639111,
      "loss": 0.9576,
      "step": 19951
    },
    {
      "epoch": 1.716596403682354,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000736685968936094,
      "loss": 0.8253,
      "step": 19952
    },
    {
      "epoch": 1.7166824399896756,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007366621494160165,
      "loss": 1.6682,
      "step": 19953
    },
    {
      "epoch": 1.7167684762969975,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007366383292037483,
      "loss": 1.1019,
      "step": 19954
    },
    {
      "epoch": 1.716854512604319,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007366145082993589,
      "loss": 1.6241,
      "step": 19955
    },
    {
      "epoch": 1.7169405489116407,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007365906867029181,
      "loss": 1.1945,
      "step": 19956
    },
    {
      "epoch": 1.7170265852189623,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007365668644144957,
      "loss": 1.3649,
      "step": 19957
    },
    {
      "epoch": 1.717112621526284,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007365430414341612,
      "loss": 1.2753,
      "step": 19958
    },
    {
      "epoch": 1.7171986578336058,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007365192177619842,
      "loss": 1.3863,
      "step": 19959
    },
    {
      "epoch": 1.7172846941409274,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007364953933980345,
      "loss": 0.5536,
      "step": 19960
    },
    {
      "epoch": 1.7173707304482493,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007364715683423818,
      "loss": 1.5895,
      "step": 19961
    },
    {
      "epoch": 1.717456766755571,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007364477425950958,
      "loss": 0.9457,
      "step": 19962
    },
    {
      "epoch": 1.7175428030628925,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007364239161562461,
      "loss": 0.8267,
      "step": 19963
    },
    {
      "epoch": 1.7176288393702142,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007364000890259024,
      "loss": 1.5112,
      "step": 19964
    },
    {
      "epoch": 1.7177148756775358,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007363762612041345,
      "loss": 1.2804,
      "step": 19965
    },
    {
      "epoch": 1.7178009119848576,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007363524326910118,
      "loss": 0.7018,
      "step": 19966
    },
    {
      "epoch": 1.7178869482921792,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007363286034866045,
      "loss": 1.6284,
      "step": 19967
    },
    {
      "epoch": 1.717972984599501,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0007363047735909818,
      "loss": 1.6479,
      "step": 19968
    },
    {
      "epoch": 1.7180590209068227,
      "grad_norm": 0.875,
      "learning_rate": 0.0007362809430042136,
      "loss": 1.56,
      "step": 19969
    },
    {
      "epoch": 1.7181450572141443,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007362571117263696,
      "loss": 1.3827,
      "step": 19970
    },
    {
      "epoch": 1.718231093521466,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007362332797575195,
      "loss": 1.1444,
      "step": 19971
    },
    {
      "epoch": 1.7183171298287876,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007362094470977332,
      "loss": 1.0664,
      "step": 19972
    },
    {
      "epoch": 1.7184031661361094,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00073618561374708,
      "loss": 1.493,
      "step": 19973
    },
    {
      "epoch": 1.7184892024434313,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007361617797056299,
      "loss": 1.1405,
      "step": 19974
    },
    {
      "epoch": 1.718575238750753,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007361379449734525,
      "loss": 1.0856,
      "step": 19975
    },
    {
      "epoch": 1.7186612750580745,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007361141095506174,
      "loss": 0.8245,
      "step": 19976
    },
    {
      "epoch": 1.7187473113653962,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007360902734371946,
      "loss": 1.3369,
      "step": 19977
    },
    {
      "epoch": 1.7188333476727178,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007360664366332537,
      "loss": 1.174,
      "step": 19978
    },
    {
      "epoch": 1.7189193839800396,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007360425991388643,
      "loss": 1.0045,
      "step": 19979
    },
    {
      "epoch": 1.7190054202873613,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007360187609540962,
      "loss": 1.0821,
      "step": 19980
    },
    {
      "epoch": 1.719091456594683,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007359949220790192,
      "loss": 1.659,
      "step": 19981
    },
    {
      "epoch": 1.7191774929020047,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007359710825137029,
      "loss": 1.3117,
      "step": 19982
    },
    {
      "epoch": 1.7192635292093263,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007359472422582172,
      "loss": 1.3993,
      "step": 19983
    },
    {
      "epoch": 1.719349565516648,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007359234013126315,
      "loss": 0.7956,
      "step": 19984
    },
    {
      "epoch": 1.7194356018239696,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007358995596770158,
      "loss": 1.4754,
      "step": 19985
    },
    {
      "epoch": 1.7195216381312914,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007358757173514398,
      "loss": 1.6295,
      "step": 19986
    },
    {
      "epoch": 1.719607674438613,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0007358518743359732,
      "loss": 0.5485,
      "step": 19987
    },
    {
      "epoch": 1.719693710745935,
      "grad_norm": 1.125,
      "learning_rate": 0.0007358280306306857,
      "loss": 1.9268,
      "step": 19988
    },
    {
      "epoch": 1.7197797470532565,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007358041862356471,
      "loss": 1.3796,
      "step": 19989
    },
    {
      "epoch": 1.7198657833605782,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007357803411509272,
      "loss": 1.844,
      "step": 19990
    },
    {
      "epoch": 1.7199518196678998,
      "grad_norm": 1.0,
      "learning_rate": 0.0007357564953765956,
      "loss": 1.1308,
      "step": 19991
    },
    {
      "epoch": 1.7200378559752214,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007357326489127221,
      "loss": 1.0193,
      "step": 19992
    },
    {
      "epoch": 1.7201238922825433,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007357088017593764,
      "loss": 0.7204,
      "step": 19993
    },
    {
      "epoch": 1.7202099285898649,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007356849539166284,
      "loss": 1.5067,
      "step": 19994
    },
    {
      "epoch": 1.7202959648971867,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007356611053845476,
      "loss": 1.5725,
      "step": 19995
    },
    {
      "epoch": 1.7203820012045083,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000735637256163204,
      "loss": 1.9618,
      "step": 19996
    },
    {
      "epoch": 1.72046803751183,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007356134062526672,
      "loss": 1.9151,
      "step": 19997
    },
    {
      "epoch": 1.7205540738191516,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007355895556530073,
      "loss": 1.3618,
      "step": 19998
    },
    {
      "epoch": 1.7206401101264732,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007355657043642933,
      "loss": 1.1527,
      "step": 19999
    },
    {
      "epoch": 1.720726146433795,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007355418523865958,
      "loss": 1.7277,
      "step": 20000
    },
    {
      "epoch": 1.720812182741117,
      "grad_norm": 1.453125,
      "learning_rate": 0.000735517999719984,
      "loss": 1.5321,
      "step": 20001
    },
    {
      "epoch": 1.7208982190484385,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000735494146364528,
      "loss": 0.694,
      "step": 20002
    },
    {
      "epoch": 1.7209842553557602,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007354702923202972,
      "loss": 1.5847,
      "step": 20003
    },
    {
      "epoch": 1.7210702916630818,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007354464375873617,
      "loss": 2.1024,
      "step": 20004
    },
    {
      "epoch": 1.7211563279704034,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007354225821657913,
      "loss": 1.6037,
      "step": 20005
    },
    {
      "epoch": 1.721242364277725,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007353987260556555,
      "loss": 1.7288,
      "step": 20006
    },
    {
      "epoch": 1.7213284005850469,
      "grad_norm": 1.625,
      "learning_rate": 0.0007353748692570241,
      "loss": 1.6746,
      "step": 20007
    },
    {
      "epoch": 1.7214144368923687,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007353510117699671,
      "loss": 1.2877,
      "step": 20008
    },
    {
      "epoch": 1.7215004731996904,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007353271535945542,
      "loss": 0.6937,
      "step": 20009
    },
    {
      "epoch": 1.721586509507012,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007353032947308551,
      "loss": 0.6872,
      "step": 20010
    },
    {
      "epoch": 1.7216725458143336,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007352794351789396,
      "loss": 1.0023,
      "step": 20011
    },
    {
      "epoch": 1.7217585821216552,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007352555749388776,
      "loss": 0.7334,
      "step": 20012
    },
    {
      "epoch": 1.721844618428977,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007352317140107387,
      "loss": 1.4339,
      "step": 20013
    },
    {
      "epoch": 1.7219306547362987,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007352078523945928,
      "loss": 1.6628,
      "step": 20014
    },
    {
      "epoch": 1.7220166910436205,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007351839900905097,
      "loss": 1.5182,
      "step": 20015
    },
    {
      "epoch": 1.7221027273509422,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007351601270985593,
      "loss": 0.7269,
      "step": 20016
    },
    {
      "epoch": 1.7221887636582638,
      "grad_norm": 1.125,
      "learning_rate": 0.0007351362634188112,
      "loss": 1.4574,
      "step": 20017
    },
    {
      "epoch": 1.7222747999655854,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007351123990513351,
      "loss": 1.5137,
      "step": 20018
    },
    {
      "epoch": 1.722360836272907,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007350885339962012,
      "loss": 0.9178,
      "step": 20019
    },
    {
      "epoch": 1.7224468725802289,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007350646682534789,
      "loss": 1.3585,
      "step": 20020
    },
    {
      "epoch": 1.7225329088875505,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007350408018232382,
      "loss": 1.9123,
      "step": 20021
    },
    {
      "epoch": 1.7226189451948724,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007350169347055489,
      "loss": 0.7964,
      "step": 20022
    },
    {
      "epoch": 1.722704981502194,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007349930669004807,
      "loss": 1.1964,
      "step": 20023
    },
    {
      "epoch": 1.7227910178095156,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007349691984081035,
      "loss": 0.9508,
      "step": 20024
    },
    {
      "epoch": 1.7228770541168372,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007349453292284871,
      "loss": 0.9195,
      "step": 20025
    },
    {
      "epoch": 1.7229630904241588,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007349214593617012,
      "loss": 1.0118,
      "step": 20026
    },
    {
      "epoch": 1.7230491267314807,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007348975888078159,
      "loss": 1.0073,
      "step": 20027
    },
    {
      "epoch": 1.7231351630388023,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007348737175669008,
      "loss": 1.326,
      "step": 20028
    },
    {
      "epoch": 1.7232211993461242,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007348498456390258,
      "loss": 1.3483,
      "step": 20029
    },
    {
      "epoch": 1.7233072356534458,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007348259730242605,
      "loss": 1.8874,
      "step": 20030
    },
    {
      "epoch": 1.7233932719607674,
      "grad_norm": 0.921875,
      "learning_rate": 0.000734802099722675,
      "loss": 1.0207,
      "step": 20031
    },
    {
      "epoch": 1.723479308268089,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007347782257343392,
      "loss": 1.5998,
      "step": 20032
    },
    {
      "epoch": 1.7235653445754107,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007347543510593225,
      "loss": 0.6997,
      "step": 20033
    },
    {
      "epoch": 1.7236513808827325,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007347304756976951,
      "loss": 1.8574,
      "step": 20034
    },
    {
      "epoch": 1.7237374171900544,
      "grad_norm": 1.25,
      "learning_rate": 0.0007347065996495265,
      "loss": 1.6493,
      "step": 20035
    },
    {
      "epoch": 1.723823453497376,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007346827229148871,
      "loss": 1.6253,
      "step": 20036
    },
    {
      "epoch": 1.7239094898046976,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007346588454938461,
      "loss": 1.5381,
      "step": 20037
    },
    {
      "epoch": 1.7239955261120192,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007346349673864738,
      "loss": 1.0266,
      "step": 20038
    },
    {
      "epoch": 1.7240815624193409,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007346110885928398,
      "loss": 1.6794,
      "step": 20039
    },
    {
      "epoch": 1.7241675987266627,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007345872091130139,
      "loss": 0.6318,
      "step": 20040
    },
    {
      "epoch": 1.7242536350339843,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007345633289470661,
      "loss": 1.6262,
      "step": 20041
    },
    {
      "epoch": 1.7243396713413062,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007345394480950662,
      "loss": 0.8997,
      "step": 20042
    },
    {
      "epoch": 1.7244257076486278,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000734515566557084,
      "loss": 1.36,
      "step": 20043
    },
    {
      "epoch": 1.7245117439559494,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007344916843331895,
      "loss": 0.7734,
      "step": 20044
    },
    {
      "epoch": 1.724597780263271,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007344678014234522,
      "loss": 0.9948,
      "step": 20045
    },
    {
      "epoch": 1.7246838165705927,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007344439178279423,
      "loss": 1.4352,
      "step": 20046
    },
    {
      "epoch": 1.7247698528779145,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007344200335467296,
      "loss": 1.2332,
      "step": 20047
    },
    {
      "epoch": 1.7248558891852361,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007343961485798838,
      "loss": 0.9041,
      "step": 20048
    },
    {
      "epoch": 1.724941925492558,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007343722629274749,
      "loss": 1.307,
      "step": 20049
    },
    {
      "epoch": 1.7250279617998796,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007343483765895727,
      "loss": 1.1617,
      "step": 20050
    },
    {
      "epoch": 1.7251139981072012,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007343244895662471,
      "loss": 1.2645,
      "step": 20051
    },
    {
      "epoch": 1.7252000344145229,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007343006018575678,
      "loss": 1.2746,
      "step": 20052
    },
    {
      "epoch": 1.7252860707218445,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000734276713463605,
      "loss": 1.4415,
      "step": 20053
    },
    {
      "epoch": 1.7253721070291663,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007342528243844285,
      "loss": 1.1579,
      "step": 20054
    },
    {
      "epoch": 1.725458143336488,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007342289346201077,
      "loss": 1.8635,
      "step": 20055
    },
    {
      "epoch": 1.7255441796438098,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000734205044170713,
      "loss": 0.9358,
      "step": 20056
    },
    {
      "epoch": 1.7256302159511314,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007341811530363141,
      "loss": 1.6377,
      "step": 20057
    },
    {
      "epoch": 1.725716252258453,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007341572612169809,
      "loss": 1.7145,
      "step": 20058
    },
    {
      "epoch": 1.7258022885657747,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007341333687127832,
      "loss": 1.0122,
      "step": 20059
    },
    {
      "epoch": 1.7258883248730963,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000734109475523791,
      "loss": 1.2495,
      "step": 20060
    },
    {
      "epoch": 1.7259743611804181,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000734085581650074,
      "loss": 0.8401,
      "step": 20061
    },
    {
      "epoch": 1.72606039748774,
      "grad_norm": 2.421875,
      "learning_rate": 0.0007340616870917023,
      "loss": 1.7628,
      "step": 20062
    },
    {
      "epoch": 1.7261464337950616,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0007340377918487455,
      "loss": 0.91,
      "step": 20063
    },
    {
      "epoch": 1.7262324701023832,
      "grad_norm": 1.625,
      "learning_rate": 0.0007340138959212737,
      "loss": 1.9501,
      "step": 20064
    },
    {
      "epoch": 1.7263185064097049,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007339899993093569,
      "loss": 1.0427,
      "step": 20065
    },
    {
      "epoch": 1.7264045427170265,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000733966102013065,
      "loss": 1.7715,
      "step": 20066
    },
    {
      "epoch": 1.7264905790243483,
      "grad_norm": 4.21875,
      "learning_rate": 0.0007339422040324675,
      "loss": 1.1647,
      "step": 20067
    },
    {
      "epoch": 1.72657661533167,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007339183053676346,
      "loss": 1.2943,
      "step": 20068
    },
    {
      "epoch": 1.7266626516389918,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007338944060186364,
      "loss": 1.3719,
      "step": 20069
    },
    {
      "epoch": 1.7267486879463134,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007338705059855423,
      "loss": 1.5085,
      "step": 20070
    },
    {
      "epoch": 1.726834724253635,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007338466052684224,
      "loss": 1.1736,
      "step": 20071
    },
    {
      "epoch": 1.7269207605609567,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007338227038673469,
      "loss": 1.0711,
      "step": 20072
    },
    {
      "epoch": 1.7270067968682783,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007337988017823852,
      "loss": 1.0544,
      "step": 20073
    },
    {
      "epoch": 1.7270928331756001,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007337748990136076,
      "loss": 1.2125,
      "step": 20074
    },
    {
      "epoch": 1.7271788694829218,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007337509955610839,
      "loss": 1.3254,
      "step": 20075
    },
    {
      "epoch": 1.7272649057902436,
      "grad_norm": 0.6875,
      "learning_rate": 0.0007337270914248841,
      "loss": 0.4866,
      "step": 20076
    },
    {
      "epoch": 1.7273509420975652,
      "grad_norm": 1.1875,
      "learning_rate": 0.000733703186605078,
      "loss": 0.7662,
      "step": 20077
    },
    {
      "epoch": 1.7274369784048869,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007336792811017353,
      "loss": 1.3892,
      "step": 20078
    },
    {
      "epoch": 1.7275230147122085,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007336553749149263,
      "loss": 1.6559,
      "step": 20079
    },
    {
      "epoch": 1.72760905101953,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007336314680447208,
      "loss": 1.1971,
      "step": 20080
    },
    {
      "epoch": 1.727695087326852,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007336075604911887,
      "loss": 1.6393,
      "step": 20081
    },
    {
      "epoch": 1.7277811236341736,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007335836522543997,
      "loss": 1.2937,
      "step": 20082
    },
    {
      "epoch": 1.7278671599414954,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007335597433344242,
      "loss": 1.1364,
      "step": 20083
    },
    {
      "epoch": 1.727953196248817,
      "grad_norm": 1.125,
      "learning_rate": 0.0007335358337313318,
      "loss": 1.3389,
      "step": 20084
    },
    {
      "epoch": 1.7280392325561387,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007335119234451926,
      "loss": 1.8341,
      "step": 20085
    },
    {
      "epoch": 1.7281252688634603,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007334880124760763,
      "loss": 1.031,
      "step": 20086
    },
    {
      "epoch": 1.728211305170782,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007334641008240531,
      "loss": 1.1973,
      "step": 20087
    },
    {
      "epoch": 1.7282973414781038,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007334401884891927,
      "loss": 1.2119,
      "step": 20088
    },
    {
      "epoch": 1.7283833777854256,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007334162754715652,
      "loss": 2.0424,
      "step": 20089
    },
    {
      "epoch": 1.7284694140927472,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007333923617712405,
      "loss": 1.5443,
      "step": 20090
    },
    {
      "epoch": 1.7285554504000689,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007333684473882885,
      "loss": 1.2618,
      "step": 20091
    },
    {
      "epoch": 1.7286414867073905,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007333445323227792,
      "loss": 2.8064,
      "step": 20092
    },
    {
      "epoch": 1.728727523014712,
      "grad_norm": 1.125,
      "learning_rate": 0.0007333206165747824,
      "loss": 1.6696,
      "step": 20093
    },
    {
      "epoch": 1.7288135593220337,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007332967001443684,
      "loss": 1.1166,
      "step": 20094
    },
    {
      "epoch": 1.7288995956293556,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007332727830316069,
      "loss": 0.6536,
      "step": 20095
    },
    {
      "epoch": 1.7289856319366774,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007332488652365678,
      "loss": 1.9329,
      "step": 20096
    },
    {
      "epoch": 1.729071668243999,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007332249467593211,
      "loss": 1.3089,
      "step": 20097
    },
    {
      "epoch": 1.7291577045513207,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007332010275999368,
      "loss": 0.6032,
      "step": 20098
    },
    {
      "epoch": 1.7292437408586423,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007331771077584849,
      "loss": 1.1332,
      "step": 20099
    },
    {
      "epoch": 1.729329777165964,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007331531872350353,
      "loss": 1.6627,
      "step": 20100
    },
    {
      "epoch": 1.7294158134732858,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007331292660296578,
      "loss": 1.5759,
      "step": 20101
    },
    {
      "epoch": 1.7295018497806074,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007331053441424227,
      "loss": 0.9103,
      "step": 20102
    },
    {
      "epoch": 1.7295878860879292,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007330814215733999,
      "loss": 2.0143,
      "step": 20103
    },
    {
      "epoch": 1.7296739223952509,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007330574983226591,
      "loss": 1.3901,
      "step": 20104
    },
    {
      "epoch": 1.7297599587025725,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007330335743902705,
      "loss": 1.107,
      "step": 20105
    },
    {
      "epoch": 1.729845995009894,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000733009649776304,
      "loss": 1.5972,
      "step": 20106
    },
    {
      "epoch": 1.7299320313172157,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007329857244808297,
      "loss": 1.9065,
      "step": 20107
    },
    {
      "epoch": 1.7300180676245376,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007329617985039174,
      "loss": 1.6214,
      "step": 20108
    },
    {
      "epoch": 1.7301041039318592,
      "grad_norm": 1.03125,
      "learning_rate": 0.000732937871845637,
      "loss": 1.1954,
      "step": 20109
    },
    {
      "epoch": 1.730190140239181,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007329139445060588,
      "loss": 0.9832,
      "step": 20110
    },
    {
      "epoch": 1.7302761765465027,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007328900164852528,
      "loss": 0.9197,
      "step": 20111
    },
    {
      "epoch": 1.7303622128538243,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007328660877832884,
      "loss": 1.1843,
      "step": 20112
    },
    {
      "epoch": 1.730448249161146,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007328421584002362,
      "loss": 1.2686,
      "step": 20113
    },
    {
      "epoch": 1.7305342854684675,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007328182283361659,
      "loss": 1.3536,
      "step": 20114
    },
    {
      "epoch": 1.7306203217757894,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007327942975911477,
      "loss": 0.8524,
      "step": 20115
    },
    {
      "epoch": 1.730706358083111,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007327703661652513,
      "loss": 1.2044,
      "step": 20116
    },
    {
      "epoch": 1.7307923943904329,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007327464340585469,
      "loss": 1.7108,
      "step": 20117
    },
    {
      "epoch": 1.7308784306977545,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007327225012711045,
      "loss": 1.7439,
      "step": 20118
    },
    {
      "epoch": 1.7309644670050761,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007326985678029939,
      "loss": 1.2938,
      "step": 20119
    },
    {
      "epoch": 1.7310505033123977,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007326746336542853,
      "loss": 1.6584,
      "step": 20120
    },
    {
      "epoch": 1.7311365396197194,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007326506988250488,
      "loss": 1.6468,
      "step": 20121
    },
    {
      "epoch": 1.7312225759270412,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007326267633153542,
      "loss": 1.5159,
      "step": 20122
    },
    {
      "epoch": 1.731308612234363,
      "grad_norm": 0.6875,
      "learning_rate": 0.0007326028271252713,
      "loss": 0.676,
      "step": 20123
    },
    {
      "epoch": 1.7313946485416847,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007325788902548707,
      "loss": 1.3283,
      "step": 20124
    },
    {
      "epoch": 1.7314806848490063,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007325549527042219,
      "loss": 1.2221,
      "step": 20125
    },
    {
      "epoch": 1.731566721156328,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007325310144733952,
      "loss": 0.9755,
      "step": 20126
    },
    {
      "epoch": 1.7316527574636496,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007325070755624604,
      "loss": 1.6772,
      "step": 20127
    },
    {
      "epoch": 1.7317387937709714,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007324831359714878,
      "loss": 0.8464,
      "step": 20128
    },
    {
      "epoch": 1.731824830078293,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007324591957005471,
      "loss": 0.953,
      "step": 20129
    },
    {
      "epoch": 1.7319108663856149,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007324352547497084,
      "loss": 0.7979,
      "step": 20130
    },
    {
      "epoch": 1.7319969026929365,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000732411313119042,
      "loss": 0.9768,
      "step": 20131
    },
    {
      "epoch": 1.7320829390002581,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007323873708086175,
      "loss": 1.7942,
      "step": 20132
    },
    {
      "epoch": 1.7321689753075797,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007323634278185053,
      "loss": 1.5628,
      "step": 20133
    },
    {
      "epoch": 1.7322550116149014,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007323394841487751,
      "loss": 1.1864,
      "step": 20134
    },
    {
      "epoch": 1.7323410479222232,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007323155397994972,
      "loss": 1.7227,
      "step": 20135
    },
    {
      "epoch": 1.7324270842295448,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007322915947707416,
      "loss": 1.394,
      "step": 20136
    },
    {
      "epoch": 1.7325131205368667,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007322676490625781,
      "loss": 1.5356,
      "step": 20137
    },
    {
      "epoch": 1.7325991568441883,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007322437026750769,
      "loss": 1.8591,
      "step": 20138
    },
    {
      "epoch": 1.73268519315151,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007322197556083081,
      "loss": 0.81,
      "step": 20139
    },
    {
      "epoch": 1.7327712294588316,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007321958078623417,
      "loss": 1.4631,
      "step": 20140
    },
    {
      "epoch": 1.7328572657661532,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0007321718594372477,
      "loss": 0.5808,
      "step": 20141
    },
    {
      "epoch": 1.732943302073475,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007321479103330962,
      "loss": 0.7843,
      "step": 20142
    },
    {
      "epoch": 1.7330293383807966,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0007321239605499572,
      "loss": 0.8764,
      "step": 20143
    },
    {
      "epoch": 1.7331153746881185,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007321000100879009,
      "loss": 0.8129,
      "step": 20144
    },
    {
      "epoch": 1.7332014109954401,
      "grad_norm": 1.71875,
      "learning_rate": 0.000732076058946997,
      "loss": 1.832,
      "step": 20145
    },
    {
      "epoch": 1.7332874473027617,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007320521071273158,
      "loss": 0.8071,
      "step": 20146
    },
    {
      "epoch": 1.7333734836100834,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007320281546289273,
      "loss": 1.1118,
      "step": 20147
    },
    {
      "epoch": 1.733459519917405,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007320042014519017,
      "loss": 1.9249,
      "step": 20148
    },
    {
      "epoch": 1.7335455562247268,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007319802475963088,
      "loss": 1.4064,
      "step": 20149
    },
    {
      "epoch": 1.7336315925320487,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007319562930622189,
      "loss": 0.8911,
      "step": 20150
    },
    {
      "epoch": 1.7337176288393703,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007319323378497019,
      "loss": 1.7552,
      "step": 20151
    },
    {
      "epoch": 1.733803665146692,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007319083819588279,
      "loss": 2.1461,
      "step": 20152
    },
    {
      "epoch": 1.7338897014540136,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000731884425389667,
      "loss": 1.1746,
      "step": 20153
    },
    {
      "epoch": 1.7339757377613352,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007318604681422893,
      "loss": 1.1985,
      "step": 20154
    },
    {
      "epoch": 1.734061774068657,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007318365102167648,
      "loss": 0.8921,
      "step": 20155
    },
    {
      "epoch": 1.7341478103759786,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007318125516131635,
      "loss": 1.5075,
      "step": 20156
    },
    {
      "epoch": 1.7342338466833005,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007317885923315557,
      "loss": 0.7412,
      "step": 20157
    },
    {
      "epoch": 1.7343198829906221,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007317646323720113,
      "loss": 1.6804,
      "step": 20158
    },
    {
      "epoch": 1.7344059192979437,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007317406717346005,
      "loss": 1.1912,
      "step": 20159
    },
    {
      "epoch": 1.7344919556052654,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007317167104193932,
      "loss": 1.4595,
      "step": 20160
    },
    {
      "epoch": 1.734577991912587,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007316927484264596,
      "loss": 1.5765,
      "step": 20161
    },
    {
      "epoch": 1.7346640282199088,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007316687857558699,
      "loss": 1.3229,
      "step": 20162
    },
    {
      "epoch": 1.7347500645272305,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000731644822407694,
      "loss": 1.8465,
      "step": 20163
    },
    {
      "epoch": 1.7348361008345523,
      "grad_norm": 1.25,
      "learning_rate": 0.0007316208583820019,
      "loss": 1.33,
      "step": 20164
    },
    {
      "epoch": 1.734922137141874,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007315968936788639,
      "loss": 1.5468,
      "step": 20165
    },
    {
      "epoch": 1.7350081734491956,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007315729282983501,
      "loss": 1.1271,
      "step": 20166
    },
    {
      "epoch": 1.7350942097565172,
      "grad_norm": 0.875,
      "learning_rate": 0.0007315489622405306,
      "loss": 0.8099,
      "step": 20167
    },
    {
      "epoch": 1.7351802460638388,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007315249955054753,
      "loss": 1.0287,
      "step": 20168
    },
    {
      "epoch": 1.7352662823711607,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007315010280932544,
      "loss": 1.4351,
      "step": 20169
    },
    {
      "epoch": 1.7353523186784823,
      "grad_norm": 1.4375,
      "learning_rate": 0.000731477060003938,
      "loss": 1.1161,
      "step": 20170
    },
    {
      "epoch": 1.7354383549858041,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007314530912375963,
      "loss": 0.7908,
      "step": 20171
    },
    {
      "epoch": 1.7355243912931257,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007314291217942992,
      "loss": 1.669,
      "step": 20172
    },
    {
      "epoch": 1.7356104276004474,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007314051516741171,
      "loss": 1.0658,
      "step": 20173
    },
    {
      "epoch": 1.735696463907769,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007313811808771198,
      "loss": 1.7048,
      "step": 20174
    },
    {
      "epoch": 1.7357825002150906,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007313572094033775,
      "loss": 1.8334,
      "step": 20175
    },
    {
      "epoch": 1.7358685365224125,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007313332372529606,
      "loss": 1.0389,
      "step": 20176
    },
    {
      "epoch": 1.7359545728297343,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007313092644259388,
      "loss": 1.7669,
      "step": 20177
    },
    {
      "epoch": 1.736040609137056,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007312852909223825,
      "loss": 1.7239,
      "step": 20178
    },
    {
      "epoch": 1.7361266454443776,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007312613167423615,
      "loss": 0.8542,
      "step": 20179
    },
    {
      "epoch": 1.7362126817516992,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007312373418859463,
      "loss": 1.2587,
      "step": 20180
    },
    {
      "epoch": 1.7362987180590208,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007312133663532068,
      "loss": 1.2476,
      "step": 20181
    },
    {
      "epoch": 1.7363847543663424,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0007311893901442133,
      "loss": 2.0304,
      "step": 20182
    },
    {
      "epoch": 1.7364707906736643,
      "grad_norm": 1.0,
      "learning_rate": 0.0007311654132590355,
      "loss": 0.7741,
      "step": 20183
    },
    {
      "epoch": 1.7365568269809861,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007311414356977442,
      "loss": 1.0986,
      "step": 20184
    },
    {
      "epoch": 1.7366428632883077,
      "grad_norm": 0.90625,
      "learning_rate": 0.000731117457460409,
      "loss": 0.6863,
      "step": 20185
    },
    {
      "epoch": 1.7367288995956294,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007310934785471,
      "loss": 1.5472,
      "step": 20186
    },
    {
      "epoch": 1.736814935902951,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007310694989578877,
      "loss": 0.9358,
      "step": 20187
    },
    {
      "epoch": 1.7369009722102726,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000731045518692842,
      "loss": 2.26,
      "step": 20188
    },
    {
      "epoch": 1.7369870085175945,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007310215377520332,
      "loss": 1.6171,
      "step": 20189
    },
    {
      "epoch": 1.737073044824916,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007309975561355311,
      "loss": 1.3791,
      "step": 20190
    },
    {
      "epoch": 1.737159081132238,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007309735738434063,
      "loss": 1.807,
      "step": 20191
    },
    {
      "epoch": 1.7372451174395596,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007309495908757286,
      "loss": 1.2648,
      "step": 20192
    },
    {
      "epoch": 1.7373311537468812,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007309256072325683,
      "loss": 0.9074,
      "step": 20193
    },
    {
      "epoch": 1.7374171900542028,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007309016229139954,
      "loss": 2.022,
      "step": 20194
    },
    {
      "epoch": 1.7375032263615244,
      "grad_norm": 1.125,
      "learning_rate": 0.0007308776379200804,
      "loss": 1.7924,
      "step": 20195
    },
    {
      "epoch": 1.7375892626688463,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007308536522508931,
      "loss": 1.7104,
      "step": 20196
    },
    {
      "epoch": 1.737675298976168,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007308296659065038,
      "loss": 1.7274,
      "step": 20197
    },
    {
      "epoch": 1.7377613352834897,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007308056788869825,
      "loss": 0.6816,
      "step": 20198
    },
    {
      "epoch": 1.7378473715908114,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007307816911923995,
      "loss": 1.5873,
      "step": 20199
    },
    {
      "epoch": 1.737933407898133,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007307577028228251,
      "loss": 1.3782,
      "step": 20200
    },
    {
      "epoch": 1.7380194442054546,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007307337137783292,
      "loss": 1.142,
      "step": 20201
    },
    {
      "epoch": 1.7381054805127762,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000730709724058982,
      "loss": 1.1117,
      "step": 20202
    },
    {
      "epoch": 1.738191516820098,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007306857336648538,
      "loss": 1.107,
      "step": 20203
    },
    {
      "epoch": 1.7382775531274197,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007306617425960147,
      "loss": 1.0016,
      "step": 20204
    },
    {
      "epoch": 1.7383635894347416,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007306377508525348,
      "loss": 1.2883,
      "step": 20205
    },
    {
      "epoch": 1.7384496257420632,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007306137584344845,
      "loss": 1.6517,
      "step": 20206
    },
    {
      "epoch": 1.7385356620493848,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007305897653419335,
      "loss": 1.5976,
      "step": 20207
    },
    {
      "epoch": 1.7386216983567064,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007305657715749527,
      "loss": 0.7704,
      "step": 20208
    },
    {
      "epoch": 1.738707734664028,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007305417771336115,
      "loss": 1.1367,
      "step": 20209
    },
    {
      "epoch": 1.73879377097135,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007305177820179806,
      "loss": 1.1518,
      "step": 20210
    },
    {
      "epoch": 1.7388798072786718,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007304937862281299,
      "loss": 1.2552,
      "step": 20211
    },
    {
      "epoch": 1.7389658435859934,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0007304697897641298,
      "loss": 0.7863,
      "step": 20212
    },
    {
      "epoch": 1.739051879893315,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007304457926260503,
      "loss": 0.8621,
      "step": 20213
    },
    {
      "epoch": 1.7391379162006366,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007304217948139619,
      "loss": 0.9286,
      "step": 20214
    },
    {
      "epoch": 1.7392239525079582,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007303977963279344,
      "loss": 1.67,
      "step": 20215
    },
    {
      "epoch": 1.73930998881528,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007303737971680382,
      "loss": 1.5251,
      "step": 20216
    },
    {
      "epoch": 1.7393960251226017,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007303497973343433,
      "loss": 1.5089,
      "step": 20217
    },
    {
      "epoch": 1.7394820614299236,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007303257968269203,
      "loss": 0.9707,
      "step": 20218
    },
    {
      "epoch": 1.7395680977372452,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000730301795645839,
      "loss": 1.5542,
      "step": 20219
    },
    {
      "epoch": 1.7396541340445668,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007302777937911696,
      "loss": 0.9061,
      "step": 20220
    },
    {
      "epoch": 1.7397401703518884,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007302537912629827,
      "loss": 1.9445,
      "step": 20221
    },
    {
      "epoch": 1.73982620665921,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007302297880613482,
      "loss": 1.9314,
      "step": 20222
    },
    {
      "epoch": 1.739912242966532,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007302057841863364,
      "loss": 0.9839,
      "step": 20223
    },
    {
      "epoch": 1.7399982792738535,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007301817796380171,
      "loss": 1.6411,
      "step": 20224
    },
    {
      "epoch": 1.7400843155811754,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007301577744164611,
      "loss": 1.44,
      "step": 20225
    },
    {
      "epoch": 1.740170351888497,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007301337685217384,
      "loss": 1.2681,
      "step": 20226
    },
    {
      "epoch": 1.7402563881958186,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0007301097619539192,
      "loss": 1.0183,
      "step": 20227
    },
    {
      "epoch": 1.7403424245031403,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007300857547130736,
      "loss": 1.3888,
      "step": 20228
    },
    {
      "epoch": 1.7404284608104619,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007300617467992721,
      "loss": 0.7826,
      "step": 20229
    },
    {
      "epoch": 1.7405144971177837,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007300377382125845,
      "loss": 1.4013,
      "step": 20230
    },
    {
      "epoch": 1.7406005334251053,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007300137289530813,
      "loss": 1.1112,
      "step": 20231
    },
    {
      "epoch": 1.7406865697324272,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007299897190208327,
      "loss": 0.9577,
      "step": 20232
    },
    {
      "epoch": 1.7407726060397488,
      "grad_norm": 0.96875,
      "learning_rate": 0.000729965708415909,
      "loss": 1.4069,
      "step": 20233
    },
    {
      "epoch": 1.7408586423470704,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007299416971383803,
      "loss": 1.7542,
      "step": 20234
    },
    {
      "epoch": 1.740944678654392,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007299176851883168,
      "loss": 1.435,
      "step": 20235
    },
    {
      "epoch": 1.7410307149617137,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007298936725657888,
      "loss": 0.9543,
      "step": 20236
    },
    {
      "epoch": 1.7411167512690355,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007298696592708666,
      "loss": 1.0744,
      "step": 20237
    },
    {
      "epoch": 1.7412027875763574,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007298456453036203,
      "loss": 1.8848,
      "step": 20238
    },
    {
      "epoch": 1.741288823883679,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007298216306641201,
      "loss": 1.3807,
      "step": 20239
    },
    {
      "epoch": 1.7413748601910006,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007297976153524363,
      "loss": 1.0514,
      "step": 20240
    },
    {
      "epoch": 1.7414608964983223,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007297735993686394,
      "loss": 1.3856,
      "step": 20241
    },
    {
      "epoch": 1.7415469328056439,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007297495827127993,
      "loss": 0.9154,
      "step": 20242
    },
    {
      "epoch": 1.7416329691129657,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007297255653849863,
      "loss": 1.1852,
      "step": 20243
    },
    {
      "epoch": 1.7417190054202873,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007297015473852708,
      "loss": 0.8617,
      "step": 20244
    },
    {
      "epoch": 1.7418050417276092,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000729677528713723,
      "loss": 1.1463,
      "step": 20245
    },
    {
      "epoch": 1.7418910780349308,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007296535093704128,
      "loss": 1.2077,
      "step": 20246
    },
    {
      "epoch": 1.7419771143422524,
      "grad_norm": 0.859375,
      "learning_rate": 0.000729629489355411,
      "loss": 0.9026,
      "step": 20247
    },
    {
      "epoch": 1.742063150649574,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007296054686687875,
      "loss": 1.3504,
      "step": 20248
    },
    {
      "epoch": 1.7421491869568957,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007295814473106127,
      "loss": 0.9298,
      "step": 20249
    },
    {
      "epoch": 1.7422352232642175,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007295574252809568,
      "loss": 1.1749,
      "step": 20250
    },
    {
      "epoch": 1.7423212595715392,
      "grad_norm": 1.125,
      "learning_rate": 0.0007295334025798901,
      "loss": 0.7876,
      "step": 20251
    },
    {
      "epoch": 1.742407295878861,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007295093792074828,
      "loss": 1.0889,
      "step": 20252
    },
    {
      "epoch": 1.7424933321861826,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007294853551638054,
      "loss": 0.8699,
      "step": 20253
    },
    {
      "epoch": 1.7425793684935043,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007294613304489277,
      "loss": 1.6235,
      "step": 20254
    },
    {
      "epoch": 1.7426654048008259,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007294373050629203,
      "loss": 1.5837,
      "step": 20255
    },
    {
      "epoch": 1.7427514411081475,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007294132790058535,
      "loss": 1.2002,
      "step": 20256
    },
    {
      "epoch": 1.7428374774154693,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007293892522777975,
      "loss": 1.0164,
      "step": 20257
    },
    {
      "epoch": 1.742923513722791,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007293652248788224,
      "loss": 0.9369,
      "step": 20258
    },
    {
      "epoch": 1.7430095500301128,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007293411968089987,
      "loss": 1.1234,
      "step": 20259
    },
    {
      "epoch": 1.7430955863374344,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007293171680683966,
      "loss": 1.5943,
      "step": 20260
    },
    {
      "epoch": 1.743181622644756,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007292931386570864,
      "loss": 1.0989,
      "step": 20261
    },
    {
      "epoch": 1.7432676589520777,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007292691085751384,
      "loss": 1.135,
      "step": 20262
    },
    {
      "epoch": 1.7433536952593993,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007292450778226228,
      "loss": 0.8586,
      "step": 20263
    },
    {
      "epoch": 1.7434397315667212,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007292210463996099,
      "loss": 1.918,
      "step": 20264
    },
    {
      "epoch": 1.743525767874043,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007291970143061701,
      "loss": 2.0129,
      "step": 20265
    },
    {
      "epoch": 1.7436118041813646,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007291729815423735,
      "loss": 1.3986,
      "step": 20266
    },
    {
      "epoch": 1.7436978404886863,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007291489481082905,
      "loss": 1.3237,
      "step": 20267
    },
    {
      "epoch": 1.7437838767960079,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007291249140039915,
      "loss": 1.9757,
      "step": 20268
    },
    {
      "epoch": 1.7438699131033295,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007291008792295466,
      "loss": 1.5923,
      "step": 20269
    },
    {
      "epoch": 1.7439559494106511,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007290768437850263,
      "loss": 1.5392,
      "step": 20270
    },
    {
      "epoch": 1.744041985717973,
      "grad_norm": 1.75,
      "learning_rate": 0.0007290528076705008,
      "loss": 1.4711,
      "step": 20271
    },
    {
      "epoch": 1.7441280220252948,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007290287708860403,
      "loss": 1.1857,
      "step": 20272
    },
    {
      "epoch": 1.7442140583326164,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007290047334317151,
      "loss": 1.1608,
      "step": 20273
    },
    {
      "epoch": 1.744300094639938,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0007289806953075957,
      "loss": 0.7596,
      "step": 20274
    },
    {
      "epoch": 1.7443861309472597,
      "grad_norm": 2.0625,
      "learning_rate": 0.0007289566565137523,
      "loss": 1.8944,
      "step": 20275
    },
    {
      "epoch": 1.7444721672545813,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007289326170502551,
      "loss": 1.3945,
      "step": 20276
    },
    {
      "epoch": 1.7445582035619032,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007289085769171747,
      "loss": 1.3219,
      "step": 20277
    },
    {
      "epoch": 1.7446442398692248,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007288845361145812,
      "loss": 1.4846,
      "step": 20278
    },
    {
      "epoch": 1.7447302761765466,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007288604946425449,
      "loss": 1.705,
      "step": 20279
    },
    {
      "epoch": 1.7448163124838683,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007288364525011361,
      "loss": 1.0267,
      "step": 20280
    },
    {
      "epoch": 1.7449023487911899,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007288124096904253,
      "loss": 0.9042,
      "step": 20281
    },
    {
      "epoch": 1.7449883850985115,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007287883662104826,
      "loss": 0.7988,
      "step": 20282
    },
    {
      "epoch": 1.7450744214058331,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007287643220613785,
      "loss": 0.812,
      "step": 20283
    },
    {
      "epoch": 1.745160457713155,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007287402772431832,
      "loss": 1.7263,
      "step": 20284
    },
    {
      "epoch": 1.7452464940204766,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000728716231755967,
      "loss": 0.8595,
      "step": 20285
    },
    {
      "epoch": 1.7453325303277984,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007286921855998006,
      "loss": 1.9086,
      "step": 20286
    },
    {
      "epoch": 1.74541856663512,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007286681387747538,
      "loss": 0.9468,
      "step": 20287
    },
    {
      "epoch": 1.7455046029424417,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007286440912808972,
      "loss": 1.2233,
      "step": 20288
    },
    {
      "epoch": 1.7455906392497633,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000728620043118301,
      "loss": 0.8865,
      "step": 20289
    },
    {
      "epoch": 1.745676675557085,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007285959942870358,
      "loss": 1.4987,
      "step": 20290
    },
    {
      "epoch": 1.7457627118644068,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007285719447871717,
      "loss": 1.4723,
      "step": 20291
    },
    {
      "epoch": 1.7458487481717284,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007285478946187791,
      "loss": 1.1946,
      "step": 20292
    },
    {
      "epoch": 1.7459347844790503,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007285238437819285,
      "loss": 1.1205,
      "step": 20293
    },
    {
      "epoch": 1.7460208207863719,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007284997922766901,
      "loss": 1.1445,
      "step": 20294
    },
    {
      "epoch": 1.7461068570936935,
      "grad_norm": 1.40625,
      "learning_rate": 0.000728475740103134,
      "loss": 1.2167,
      "step": 20295
    },
    {
      "epoch": 1.7461928934010151,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007284516872613309,
      "loss": 1.1571,
      "step": 20296
    },
    {
      "epoch": 1.7462789297083368,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007284276337513511,
      "loss": 0.9082,
      "step": 20297
    },
    {
      "epoch": 1.7463649660156586,
      "grad_norm": 1.28125,
      "learning_rate": 0.000728403579573265,
      "loss": 0.9261,
      "step": 20298
    },
    {
      "epoch": 1.7464510023229805,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007283795247271427,
      "loss": 0.675,
      "step": 20299
    },
    {
      "epoch": 1.746537038630302,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007283554692130548,
      "loss": 1.2662,
      "step": 20300
    },
    {
      "epoch": 1.7466230749376237,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007283314130310715,
      "loss": 1.9198,
      "step": 20301
    },
    {
      "epoch": 1.7467091112449453,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007283073561812632,
      "loss": 1.726,
      "step": 20302
    },
    {
      "epoch": 1.746795147552267,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007282832986637003,
      "loss": 1.5797,
      "step": 20303
    },
    {
      "epoch": 1.7468811838595888,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007282592404784531,
      "loss": 1.4944,
      "step": 20304
    },
    {
      "epoch": 1.7469672201669104,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007282351816255921,
      "loss": 1.0696,
      "step": 20305
    },
    {
      "epoch": 1.7470532564742323,
      "grad_norm": 1.125,
      "learning_rate": 0.0007282111221051876,
      "loss": 1.4495,
      "step": 20306
    },
    {
      "epoch": 1.7471392927815539,
      "grad_norm": 1.5,
      "learning_rate": 0.0007281870619173099,
      "loss": 1.3256,
      "step": 20307
    },
    {
      "epoch": 1.7472253290888755,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007281630010620295,
      "loss": 0.8689,
      "step": 20308
    },
    {
      "epoch": 1.7473113653961971,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007281389395394167,
      "loss": 1.398,
      "step": 20309
    },
    {
      "epoch": 1.7473974017035188,
      "grad_norm": 1.25,
      "learning_rate": 0.0007281148773495417,
      "loss": 0.7579,
      "step": 20310
    },
    {
      "epoch": 1.7474834380108406,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007280908144924752,
      "loss": 1.0076,
      "step": 20311
    },
    {
      "epoch": 1.7475694743181622,
      "grad_norm": 1.375,
      "learning_rate": 0.0007280667509682874,
      "loss": 1.2958,
      "step": 20312
    },
    {
      "epoch": 1.747655510625484,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007280426867770489,
      "loss": 1.3151,
      "step": 20313
    },
    {
      "epoch": 1.7477415469328057,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007280186219188295,
      "loss": 1.2953,
      "step": 20314
    },
    {
      "epoch": 1.7478275832401273,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0007279945563937002,
      "loss": 1.2992,
      "step": 20315
    },
    {
      "epoch": 1.747913619547449,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007279704902017312,
      "loss": 1.038,
      "step": 20316
    },
    {
      "epoch": 1.7479996558547706,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007279464233429928,
      "loss": 1.4004,
      "step": 20317
    },
    {
      "epoch": 1.7480856921620924,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007279223558175554,
      "loss": 1.0887,
      "step": 20318
    },
    {
      "epoch": 1.748171728469414,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007278982876254895,
      "loss": 1.2146,
      "step": 20319
    },
    {
      "epoch": 1.748257764776736,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007278742187668654,
      "loss": 1.4882,
      "step": 20320
    },
    {
      "epoch": 1.7483438010840575,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007278501492417536,
      "loss": 1.0794,
      "step": 20321
    },
    {
      "epoch": 1.7484298373913791,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007278260790502243,
      "loss": 1.0887,
      "step": 20322
    },
    {
      "epoch": 1.7485158736987008,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007278020081923481,
      "loss": 1.5689,
      "step": 20323
    },
    {
      "epoch": 1.7486019100060224,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007277779366681954,
      "loss": 0.8694,
      "step": 20324
    },
    {
      "epoch": 1.7486879463133442,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0007277538644778364,
      "loss": 1.347,
      "step": 20325
    },
    {
      "epoch": 1.748773982620666,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007277297916213417,
      "loss": 1.6174,
      "step": 20326
    },
    {
      "epoch": 1.7488600189279877,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007277057180987818,
      "loss": 0.5876,
      "step": 20327
    },
    {
      "epoch": 1.7489460552353093,
      "grad_norm": 1.0,
      "learning_rate": 0.0007276816439102268,
      "loss": 1.2093,
      "step": 20328
    },
    {
      "epoch": 1.749032091542631,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007276575690557472,
      "loss": 1.3807,
      "step": 20329
    },
    {
      "epoch": 1.7491181278499526,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007276334935354136,
      "loss": 1.4811,
      "step": 20330
    },
    {
      "epoch": 1.7492041641572744,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007276094173492963,
      "loss": 1.1118,
      "step": 20331
    },
    {
      "epoch": 1.749290200464596,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007275853404974656,
      "loss": 1.7726,
      "step": 20332
    },
    {
      "epoch": 1.749376236771918,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007275612629799921,
      "loss": 1.2594,
      "step": 20333
    },
    {
      "epoch": 1.7494622730792395,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007275371847969463,
      "loss": 1.7384,
      "step": 20334
    },
    {
      "epoch": 1.7495483093865611,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007275131059483984,
      "loss": 1.8092,
      "step": 20335
    },
    {
      "epoch": 1.7496343456938828,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007274890264344189,
      "loss": 0.924,
      "step": 20336
    },
    {
      "epoch": 1.7497203820012044,
      "grad_norm": 1.84375,
      "learning_rate": 0.0007274649462550783,
      "loss": 1.7861,
      "step": 20337
    },
    {
      "epoch": 1.7498064183085262,
      "grad_norm": 1.65625,
      "learning_rate": 0.000727440865410447,
      "loss": 1.3099,
      "step": 20338
    },
    {
      "epoch": 1.7498924546158479,
      "grad_norm": 2.203125,
      "learning_rate": 0.0007274167839005952,
      "loss": 1.7089,
      "step": 20339
    },
    {
      "epoch": 1.7499784909231697,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0007273927017255936,
      "loss": 1.1906,
      "step": 20340
    },
    {
      "epoch": 1.7500645272304913,
      "grad_norm": 1.828125,
      "learning_rate": 0.0007273686188855127,
      "loss": 1.1084,
      "step": 20341
    },
    {
      "epoch": 1.750150563537813,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007273445353804228,
      "loss": 0.7053,
      "step": 20342
    },
    {
      "epoch": 1.750150563537813,
      "eval_loss": 1.2168831825256348,
      "eval_runtime": 350.4189,
      "eval_samples_per_second": 13.267,
      "eval_steps_per_second": 1.661,
      "step": 20342
    },
    {
      "epoch": 1.7502365998451346,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007273204512103944,
      "loss": 1.0751,
      "step": 20343
    },
    {
      "epoch": 1.7503226361524562,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007272963663754976,
      "loss": 0.9011,
      "step": 20344
    },
    {
      "epoch": 1.750408672459778,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007272722808758035,
      "loss": 1.9252,
      "step": 20345
    },
    {
      "epoch": 1.7504947087670997,
      "grad_norm": 2.1875,
      "learning_rate": 0.0007272481947113819,
      "loss": 1.6464,
      "step": 20346
    },
    {
      "epoch": 1.7505807450744215,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007272241078823036,
      "loss": 1.3206,
      "step": 20347
    },
    {
      "epoch": 1.7506667813817431,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007272000203886391,
      "loss": 0.9436,
      "step": 20348
    },
    {
      "epoch": 1.7507528176890648,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007271759322304587,
      "loss": 0.9886,
      "step": 20349
    },
    {
      "epoch": 1.7508388539963864,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007271518434078328,
      "loss": 1.4118,
      "step": 20350
    },
    {
      "epoch": 1.750924890303708,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007271277539208318,
      "loss": 1.5186,
      "step": 20351
    },
    {
      "epoch": 1.7510109266110299,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007271036637695265,
      "loss": 1.2035,
      "step": 20352
    },
    {
      "epoch": 1.7510969629183517,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007270795729539871,
      "loss": 1.1357,
      "step": 20353
    },
    {
      "epoch": 1.7511829992256733,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007270554814742843,
      "loss": 0.7973,
      "step": 20354
    },
    {
      "epoch": 1.751269035532995,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000727031389330488,
      "loss": 0.8935,
      "step": 20355
    },
    {
      "epoch": 1.7513550718403166,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007270072965226693,
      "loss": 0.7168,
      "step": 20356
    },
    {
      "epoch": 1.7514411081476382,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007269832030508984,
      "loss": 1.535,
      "step": 20357
    },
    {
      "epoch": 1.7515271444549598,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007269591089152456,
      "loss": 1.4537,
      "step": 20358
    },
    {
      "epoch": 1.7516131807622817,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007269350141157816,
      "loss": 1.2208,
      "step": 20359
    },
    {
      "epoch": 1.7516992170696035,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000726910918652577,
      "loss": 0.7563,
      "step": 20360
    },
    {
      "epoch": 1.7517852533769251,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007268868225257019,
      "loss": 1.611,
      "step": 20361
    },
    {
      "epoch": 1.7518712896842468,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000726862725735227,
      "loss": 1.7297,
      "step": 20362
    },
    {
      "epoch": 1.7519573259915684,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007268386282812229,
      "loss": 1.4281,
      "step": 20363
    },
    {
      "epoch": 1.75204336229889,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0007268145301637599,
      "loss": 1.521,
      "step": 20364
    },
    {
      "epoch": 1.7521293986062119,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007267904313829085,
      "loss": 0.854,
      "step": 20365
    },
    {
      "epoch": 1.7522154349135335,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000726766331938739,
      "loss": 1.5664,
      "step": 20366
    },
    {
      "epoch": 1.7523014712208553,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007267422318313223,
      "loss": 0.9909,
      "step": 20367
    },
    {
      "epoch": 1.752387507528177,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007267181310607286,
      "loss": 1.1039,
      "step": 20368
    },
    {
      "epoch": 1.7524735438354986,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007266940296270285,
      "loss": 0.5063,
      "step": 20369
    },
    {
      "epoch": 1.7525595801428202,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007266699275302924,
      "loss": 2.0113,
      "step": 20370
    },
    {
      "epoch": 1.7526456164501418,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007266458247705908,
      "loss": 0.8835,
      "step": 20371
    },
    {
      "epoch": 1.7527316527574637,
      "grad_norm": 1.125,
      "learning_rate": 0.0007266217213479945,
      "loss": 1.4329,
      "step": 20372
    },
    {
      "epoch": 1.7528176890647853,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007265976172625734,
      "loss": 0.9938,
      "step": 20373
    },
    {
      "epoch": 1.7529037253721071,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007265735125143984,
      "loss": 0.8182,
      "step": 20374
    },
    {
      "epoch": 1.7529897616794288,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007265494071035402,
      "loss": 1.4518,
      "step": 20375
    },
    {
      "epoch": 1.7530757979867504,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007265253010300687,
      "loss": 1.3915,
      "step": 20376
    },
    {
      "epoch": 1.753161834294072,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000726501194294055,
      "loss": 1.7011,
      "step": 20377
    },
    {
      "epoch": 1.7532478706013936,
      "grad_norm": 1.125,
      "learning_rate": 0.0007264770868955692,
      "loss": 1.1901,
      "step": 20378
    },
    {
      "epoch": 1.7533339069087155,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007264529788346821,
      "loss": 1.4184,
      "step": 20379
    },
    {
      "epoch": 1.7534199432160371,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007264288701114641,
      "loss": 0.9091,
      "step": 20380
    },
    {
      "epoch": 1.753505979523359,
      "grad_norm": 0.828125,
      "learning_rate": 0.0007264047607259855,
      "loss": 0.5464,
      "step": 20381
    },
    {
      "epoch": 1.7535920158306806,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007263806506783171,
      "loss": 1.531,
      "step": 20382
    },
    {
      "epoch": 1.7536780521380022,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007263565399685295,
      "loss": 0.9868,
      "step": 20383
    },
    {
      "epoch": 1.7537640884453238,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007263324285966927,
      "loss": 1.1135,
      "step": 20384
    },
    {
      "epoch": 1.7538501247526455,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007263083165628779,
      "loss": 1.2158,
      "step": 20385
    },
    {
      "epoch": 1.7539361610599673,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007262842038671551,
      "loss": 1.4726,
      "step": 20386
    },
    {
      "epoch": 1.7540221973672891,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007262600905095951,
      "loss": 0.7954,
      "step": 20387
    },
    {
      "epoch": 1.7541082336746108,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007262359764902683,
      "loss": 0.9892,
      "step": 20388
    },
    {
      "epoch": 1.7541942699819324,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007262118618092452,
      "loss": 0.5732,
      "step": 20389
    },
    {
      "epoch": 1.754280306289254,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007261877464665965,
      "loss": 1.1255,
      "step": 20390
    },
    {
      "epoch": 1.7543663425965756,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007261636304623926,
      "loss": 1.2818,
      "step": 20391
    },
    {
      "epoch": 1.7544523789038975,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007261395137967042,
      "loss": 0.7352,
      "step": 20392
    },
    {
      "epoch": 1.7545384152112191,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007261153964696014,
      "loss": 1.4543,
      "step": 20393
    },
    {
      "epoch": 1.754624451518541,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007260912784811553,
      "loss": 1.8009,
      "step": 20394
    },
    {
      "epoch": 1.7547104878258626,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007260671598314361,
      "loss": 1.195,
      "step": 20395
    },
    {
      "epoch": 1.7547965241331842,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007260430405205144,
      "loss": 0.9816,
      "step": 20396
    },
    {
      "epoch": 1.7548825604405058,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007260189205484609,
      "loss": 1.5247,
      "step": 20397
    },
    {
      "epoch": 1.7549685967478275,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007259947999153458,
      "loss": 1.5853,
      "step": 20398
    },
    {
      "epoch": 1.7550546330551493,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007259706786212401,
      "loss": 1.7735,
      "step": 20399
    },
    {
      "epoch": 1.755140669362471,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007259465566662139,
      "loss": 1.9386,
      "step": 20400
    },
    {
      "epoch": 1.7552267056697928,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000725922434050338,
      "loss": 1.5353,
      "step": 20401
    },
    {
      "epoch": 1.7553127419771144,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000725898310773683,
      "loss": 0.8515,
      "step": 20402
    },
    {
      "epoch": 1.755398778284436,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007258741868363194,
      "loss": 1.1166,
      "step": 20403
    },
    {
      "epoch": 1.7554848145917576,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007258500622383177,
      "loss": 0.8935,
      "step": 20404
    },
    {
      "epoch": 1.7555708508990793,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007258259369797485,
      "loss": 1.6638,
      "step": 20405
    },
    {
      "epoch": 1.7556568872064011,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007258018110606824,
      "loss": 1.054,
      "step": 20406
    },
    {
      "epoch": 1.7557429235137227,
      "grad_norm": 1.125,
      "learning_rate": 0.0007257776844811898,
      "loss": 1.1413,
      "step": 20407
    },
    {
      "epoch": 1.7558289598210446,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007257535572413414,
      "loss": 1.7632,
      "step": 20408
    },
    {
      "epoch": 1.7559149961283662,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007257294293412078,
      "loss": 1.4015,
      "step": 20409
    },
    {
      "epoch": 1.7560010324356878,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007257053007808596,
      "loss": 1.0265,
      "step": 20410
    },
    {
      "epoch": 1.7560870687430095,
      "grad_norm": 23.25,
      "learning_rate": 0.0007256811715603671,
      "loss": 1.2759,
      "step": 20411
    },
    {
      "epoch": 1.756173105050331,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007256570416798012,
      "loss": 1.6253,
      "step": 20412
    },
    {
      "epoch": 1.756259141357653,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007256329111392323,
      "loss": 0.9433,
      "step": 20413
    },
    {
      "epoch": 1.7563451776649748,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000725608779938731,
      "loss": 1.8263,
      "step": 20414
    },
    {
      "epoch": 1.7564312139722964,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007255846480783679,
      "loss": 1.6489,
      "step": 20415
    },
    {
      "epoch": 1.756517250279618,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007255605155582135,
      "loss": 1.5106,
      "step": 20416
    },
    {
      "epoch": 1.7566032865869396,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007255363823783388,
      "loss": 1.0371,
      "step": 20417
    },
    {
      "epoch": 1.7566893228942613,
      "grad_norm": 1.125,
      "learning_rate": 0.0007255122485388137,
      "loss": 1.8343,
      "step": 20418
    },
    {
      "epoch": 1.7567753592015831,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007254881140397091,
      "loss": 1.1511,
      "step": 20419
    },
    {
      "epoch": 1.7568613955089047,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000725463978881096,
      "loss": 0.7823,
      "step": 20420
    },
    {
      "epoch": 1.7569474318162266,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007254398430630443,
      "loss": 1.4118,
      "step": 20421
    },
    {
      "epoch": 1.7570334681235482,
      "grad_norm": 1.328125,
      "learning_rate": 0.000725415706585625,
      "loss": 1.3293,
      "step": 20422
    },
    {
      "epoch": 1.7571195044308698,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007253915694489086,
      "loss": 0.7145,
      "step": 20423
    },
    {
      "epoch": 1.7572055407381915,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007253674316529657,
      "loss": 1.2304,
      "step": 20424
    },
    {
      "epoch": 1.757291577045513,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007253432931978668,
      "loss": 1.6887,
      "step": 20425
    },
    {
      "epoch": 1.757377613352835,
      "grad_norm": 1.25,
      "learning_rate": 0.0007253191540836826,
      "loss": 1.9053,
      "step": 20426
    },
    {
      "epoch": 1.7574636496601566,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007252950143104838,
      "loss": 0.8047,
      "step": 20427
    },
    {
      "epoch": 1.7575496859674784,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007252708738783409,
      "loss": 2.3192,
      "step": 20428
    },
    {
      "epoch": 1.7576357222748,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007252467327873244,
      "loss": 0.6148,
      "step": 20429
    },
    {
      "epoch": 1.7577217585821217,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007252225910375052,
      "loss": 0.8688,
      "step": 20430
    },
    {
      "epoch": 1.7578077948894433,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007251984486289536,
      "loss": 1.0152,
      "step": 20431
    },
    {
      "epoch": 1.757893831196765,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007251743055617404,
      "loss": 0.9098,
      "step": 20432
    },
    {
      "epoch": 1.7579798675040867,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000725150161835936,
      "loss": 1.2483,
      "step": 20433
    },
    {
      "epoch": 1.7580659038114084,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007251260174516112,
      "loss": 0.8567,
      "step": 20434
    },
    {
      "epoch": 1.7581519401187302,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007251018724088366,
      "loss": 2.0563,
      "step": 20435
    },
    {
      "epoch": 1.7582379764260518,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007250777267076829,
      "loss": 1.5161,
      "step": 20436
    },
    {
      "epoch": 1.7583240127333735,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007250535803482206,
      "loss": 1.5194,
      "step": 20437
    },
    {
      "epoch": 1.758410049040695,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007250294333305203,
      "loss": 1.3381,
      "step": 20438
    },
    {
      "epoch": 1.7584960853480167,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007250052856546527,
      "loss": 1.8511,
      "step": 20439
    },
    {
      "epoch": 1.7585821216553386,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007249811373206885,
      "loss": 1.6674,
      "step": 20440
    },
    {
      "epoch": 1.7586681579626604,
      "grad_norm": 1.171875,
      "learning_rate": 0.000724956988328698,
      "loss": 1.161,
      "step": 20441
    },
    {
      "epoch": 1.758754194269982,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007249328386787522,
      "loss": 1.9212,
      "step": 20442
    },
    {
      "epoch": 1.7588402305773037,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007249086883709216,
      "loss": 1.8022,
      "step": 20443
    },
    {
      "epoch": 1.7589262668846253,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007248845374052767,
      "loss": 1.0972,
      "step": 20444
    },
    {
      "epoch": 1.759012303191947,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007248603857818886,
      "loss": 1.832,
      "step": 20445
    },
    {
      "epoch": 1.7590983394992685,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007248362335008273,
      "loss": 0.9822,
      "step": 20446
    },
    {
      "epoch": 1.7591843758065904,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007248120805621638,
      "loss": 0.5309,
      "step": 20447
    },
    {
      "epoch": 1.7592704121139122,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007247879269659687,
      "loss": 0.9054,
      "step": 20448
    },
    {
      "epoch": 1.7593564484212338,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007247637727123127,
      "loss": 0.7396,
      "step": 20449
    },
    {
      "epoch": 1.7594424847285555,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007247396178012663,
      "loss": 0.8415,
      "step": 20450
    },
    {
      "epoch": 1.759528521035877,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007247154622329002,
      "loss": 0.8975,
      "step": 20451
    },
    {
      "epoch": 1.7596145573431987,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000724691306007285,
      "loss": 0.6221,
      "step": 20452
    },
    {
      "epoch": 1.7597005936505206,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007246671491244916,
      "loss": 1.8946,
      "step": 20453
    },
    {
      "epoch": 1.7597866299578422,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007246429915845904,
      "loss": 1.7679,
      "step": 20454
    },
    {
      "epoch": 1.759872666265164,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007246188333876523,
      "loss": 0.625,
      "step": 20455
    },
    {
      "epoch": 1.7599587025724857,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007245946745337476,
      "loss": 0.9783,
      "step": 20456
    },
    {
      "epoch": 1.7600447388798073,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007245705150229473,
      "loss": 1.0864,
      "step": 20457
    },
    {
      "epoch": 1.760130775187129,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007245463548553218,
      "loss": 1.8875,
      "step": 20458
    },
    {
      "epoch": 1.7602168114944505,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007245221940309419,
      "loss": 0.919,
      "step": 20459
    },
    {
      "epoch": 1.7603028478017724,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007244980325498783,
      "loss": 1.0618,
      "step": 20460
    },
    {
      "epoch": 1.760388884109094,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0007244738704122017,
      "loss": 0.7897,
      "step": 20461
    },
    {
      "epoch": 1.7604749204164158,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007244497076179826,
      "loss": 1.4434,
      "step": 20462
    },
    {
      "epoch": 1.7605609567237375,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007244255441672916,
      "loss": 1.3213,
      "step": 20463
    },
    {
      "epoch": 1.760646993031059,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0007244013800601998,
      "loss": 1.0996,
      "step": 20464
    },
    {
      "epoch": 1.7607330293383807,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007243772152967776,
      "loss": 1.9017,
      "step": 20465
    },
    {
      "epoch": 1.7608190656457023,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007243530498770956,
      "loss": 0.5621,
      "step": 20466
    },
    {
      "epoch": 1.7609051019530242,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007243288838012244,
      "loss": 1.0341,
      "step": 20467
    },
    {
      "epoch": 1.7609911382603458,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007243047170692351,
      "loss": 0.9697,
      "step": 20468
    },
    {
      "epoch": 1.7610771745676677,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007242805496811979,
      "loss": 1.9183,
      "step": 20469
    },
    {
      "epoch": 1.7611632108749893,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007242563816371839,
      "loss": 0.8843,
      "step": 20470
    },
    {
      "epoch": 1.761249247182311,
      "grad_norm": 3.078125,
      "learning_rate": 0.0007242322129372634,
      "loss": 1.3962,
      "step": 20471
    },
    {
      "epoch": 1.7613352834896325,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007242080435815075,
      "loss": 2.0534,
      "step": 20472
    },
    {
      "epoch": 1.7614213197969542,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007241838735699866,
      "loss": 1.447,
      "step": 20473
    },
    {
      "epoch": 1.761507356104276,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007241597029027714,
      "loss": 1.7625,
      "step": 20474
    },
    {
      "epoch": 1.7615933924115978,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007241355315799327,
      "loss": 1.4374,
      "step": 20475
    },
    {
      "epoch": 1.7616794287189195,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007241113596015412,
      "loss": 1.347,
      "step": 20476
    },
    {
      "epoch": 1.761765465026241,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007240871869676676,
      "loss": 1.0695,
      "step": 20477
    },
    {
      "epoch": 1.7618515013335627,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007240630136783825,
      "loss": 1.116,
      "step": 20478
    },
    {
      "epoch": 1.7619375376408843,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007240388397337566,
      "loss": 2.0361,
      "step": 20479
    },
    {
      "epoch": 1.7620235739482062,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0007240146651338608,
      "loss": 0.743,
      "step": 20480
    },
    {
      "epoch": 1.7621096102555278,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007239904898787657,
      "loss": 1.4157,
      "step": 20481
    },
    {
      "epoch": 1.7621956465628497,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007239663139685417,
      "loss": 1.4351,
      "step": 20482
    },
    {
      "epoch": 1.7622816828701713,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007239421374032601,
      "loss": 1.2686,
      "step": 20483
    },
    {
      "epoch": 1.762367719177493,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007239179601829913,
      "loss": 1.0449,
      "step": 20484
    },
    {
      "epoch": 1.7624537554848145,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007238937823078058,
      "loss": 1.5793,
      "step": 20485
    },
    {
      "epoch": 1.7625397917921362,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007238696037777746,
      "loss": 1.0373,
      "step": 20486
    },
    {
      "epoch": 1.762625828099458,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007238454245929684,
      "loss": 1.5545,
      "step": 20487
    },
    {
      "epoch": 1.7627118644067796,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007238212447534578,
      "loss": 1.6205,
      "step": 20488
    },
    {
      "epoch": 1.7627979007141015,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007237970642593137,
      "loss": 1.2465,
      "step": 20489
    },
    {
      "epoch": 1.762883937021423,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007237728831106065,
      "loss": 1.2983,
      "step": 20490
    },
    {
      "epoch": 1.7629699733287447,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007237487013074073,
      "loss": 0.8502,
      "step": 20491
    },
    {
      "epoch": 1.7630560096360663,
      "grad_norm": 1.25,
      "learning_rate": 0.0007237245188497867,
      "loss": 1.7504,
      "step": 20492
    },
    {
      "epoch": 1.763142045943388,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007237003357378153,
      "loss": 1.2393,
      "step": 20493
    },
    {
      "epoch": 1.7632280822507098,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007236761519715642,
      "loss": 1.0618,
      "step": 20494
    },
    {
      "epoch": 1.7633141185580314,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007236519675511035,
      "loss": 1.1,
      "step": 20495
    },
    {
      "epoch": 1.7634001548653533,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007236277824765045,
      "loss": 1.0158,
      "step": 20496
    },
    {
      "epoch": 1.763486191172675,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0007236035967478376,
      "loss": 1.2556,
      "step": 20497
    },
    {
      "epoch": 1.7635722274799965,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007235794103651737,
      "loss": 1.1962,
      "step": 20498
    },
    {
      "epoch": 1.7636582637873182,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007235552233285836,
      "loss": 1.0423,
      "step": 20499
    },
    {
      "epoch": 1.7637443000946398,
      "grad_norm": 1.375,
      "learning_rate": 0.0007235310356381379,
      "loss": 1.1823,
      "step": 20500
    },
    {
      "epoch": 1.7638303364019616,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007235068472939074,
      "loss": 1.6602,
      "step": 20501
    },
    {
      "epoch": 1.7639163727092835,
      "grad_norm": 0.75,
      "learning_rate": 0.0007234826582959628,
      "loss": 0.7932,
      "step": 20502
    },
    {
      "epoch": 1.764002409016605,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007234584686443749,
      "loss": 0.9363,
      "step": 20503
    },
    {
      "epoch": 1.7640884453239267,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007234342783392144,
      "loss": 1.0748,
      "step": 20504
    },
    {
      "epoch": 1.7641744816312483,
      "grad_norm": 1.125,
      "learning_rate": 0.0007234100873805523,
      "loss": 1.2514,
      "step": 20505
    },
    {
      "epoch": 1.76426051793857,
      "grad_norm": 1.21875,
      "learning_rate": 0.000723385895768459,
      "loss": 0.7829,
      "step": 20506
    },
    {
      "epoch": 1.7643465542458918,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007233617035030053,
      "loss": 1.6151,
      "step": 20507
    },
    {
      "epoch": 1.7644325905532134,
      "grad_norm": 1.0625,
      "learning_rate": 0.000723337510584262,
      "loss": 1.0428,
      "step": 20508
    },
    {
      "epoch": 1.7645186268605353,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007233133170123,
      "loss": 1.0718,
      "step": 20509
    },
    {
      "epoch": 1.764604663167857,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007232891227871901,
      "loss": 1.0079,
      "step": 20510
    },
    {
      "epoch": 1.7646906994751785,
      "grad_norm": 1.125,
      "learning_rate": 0.000723264927909003,
      "loss": 1.1978,
      "step": 20511
    },
    {
      "epoch": 1.7647767357825002,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007232407323778092,
      "loss": 1.2093,
      "step": 20512
    },
    {
      "epoch": 1.7648627720898218,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007232165361936798,
      "loss": 1.4086,
      "step": 20513
    },
    {
      "epoch": 1.7649488083971436,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007231923393566855,
      "loss": 1.0887,
      "step": 20514
    },
    {
      "epoch": 1.7650348447044653,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007231681418668969,
      "loss": 1.3159,
      "step": 20515
    },
    {
      "epoch": 1.765120881011787,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000723143943724385,
      "loss": 1.2867,
      "step": 20516
    },
    {
      "epoch": 1.7652069173191087,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007231197449292203,
      "loss": 1.1623,
      "step": 20517
    },
    {
      "epoch": 1.7652929536264304,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007230955454814738,
      "loss": 1.991,
      "step": 20518
    },
    {
      "epoch": 1.765378989933752,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007230713453812162,
      "loss": 1.0542,
      "step": 20519
    },
    {
      "epoch": 1.7654650262410736,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007230471446285184,
      "loss": 1.4142,
      "step": 20520
    },
    {
      "epoch": 1.7655510625483954,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0007230229432234511,
      "loss": 2.0105,
      "step": 20521
    },
    {
      "epoch": 1.765637098855717,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000722998741166085,
      "loss": 0.931,
      "step": 20522
    },
    {
      "epoch": 1.765723135163039,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007229745384564909,
      "loss": 0.7616,
      "step": 20523
    },
    {
      "epoch": 1.7658091714703605,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007229503350947397,
      "loss": 1.2273,
      "step": 20524
    },
    {
      "epoch": 1.7658952077776822,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007229261310809021,
      "loss": 1.385,
      "step": 20525
    },
    {
      "epoch": 1.7659812440850038,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007229019264150489,
      "loss": 1.1234,
      "step": 20526
    },
    {
      "epoch": 1.7660672803923254,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000722877721097251,
      "loss": 1.6121,
      "step": 20527
    },
    {
      "epoch": 1.7661533166996473,
      "grad_norm": 1.359375,
      "learning_rate": 0.000722853515127579,
      "loss": 1.2115,
      "step": 20528
    },
    {
      "epoch": 1.766239353006969,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007228293085061039,
      "loss": 1.7767,
      "step": 20529
    },
    {
      "epoch": 1.7663253893142907,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007228051012328965,
      "loss": 1.3796,
      "step": 20530
    },
    {
      "epoch": 1.7664114256216124,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007227808933080273,
      "loss": 2.1007,
      "step": 20531
    },
    {
      "epoch": 1.766497461928934,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007227566847315674,
      "loss": 0.7402,
      "step": 20532
    },
    {
      "epoch": 1.7665834982362556,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007227324755035877,
      "loss": 1.2743,
      "step": 20533
    },
    {
      "epoch": 1.7666695345435772,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007227082656241586,
      "loss": 1.4834,
      "step": 20534
    },
    {
      "epoch": 1.766755570850899,
      "grad_norm": 1.125,
      "learning_rate": 0.0007226840550933512,
      "loss": 1.7747,
      "step": 20535
    },
    {
      "epoch": 1.766841607158221,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007226598439112363,
      "loss": 1.1576,
      "step": 20536
    },
    {
      "epoch": 1.7669276434655425,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007226356320778846,
      "loss": 1.3645,
      "step": 20537
    },
    {
      "epoch": 1.7670136797728642,
      "grad_norm": 1.078125,
      "learning_rate": 0.000722611419593367,
      "loss": 1.2876,
      "step": 20538
    },
    {
      "epoch": 1.7670997160801858,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007225872064577543,
      "loss": 1.5582,
      "step": 20539
    },
    {
      "epoch": 1.7671857523875074,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007225629926711172,
      "loss": 1.5211,
      "step": 20540
    },
    {
      "epoch": 1.7672717886948293,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007225387782335268,
      "loss": 0.6351,
      "step": 20541
    },
    {
      "epoch": 1.7673578250021509,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007225145631450538,
      "loss": 1.0395,
      "step": 20542
    },
    {
      "epoch": 1.7674438613094727,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007224903474057688,
      "loss": 0.9209,
      "step": 20543
    },
    {
      "epoch": 1.7675298976167944,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007224661310157428,
      "loss": 1.3288,
      "step": 20544
    },
    {
      "epoch": 1.767615933924116,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007224419139750468,
      "loss": 1.0085,
      "step": 20545
    },
    {
      "epoch": 1.7677019702314376,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007224176962837512,
      "loss": 1.6037,
      "step": 20546
    },
    {
      "epoch": 1.7677880065387592,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007223934779419273,
      "loss": 0.7494,
      "step": 20547
    },
    {
      "epoch": 1.767874042846081,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007223692589496457,
      "loss": 1.4092,
      "step": 20548
    },
    {
      "epoch": 1.7679600791534027,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007223450393069773,
      "loss": 2.0556,
      "step": 20549
    },
    {
      "epoch": 1.7680461154607245,
      "grad_norm": 1.25,
      "learning_rate": 0.0007223208190139928,
      "loss": 1.1845,
      "step": 20550
    },
    {
      "epoch": 1.7681321517680462,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007222965980707634,
      "loss": 1.3417,
      "step": 20551
    },
    {
      "epoch": 1.7682181880753678,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007222723764773594,
      "loss": 0.7745,
      "step": 20552
    },
    {
      "epoch": 1.7683042243826894,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007222481542338519,
      "loss": 1.3336,
      "step": 20553
    },
    {
      "epoch": 1.768390260690011,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007222239313403121,
      "loss": 0.582,
      "step": 20554
    },
    {
      "epoch": 1.7684762969973329,
      "grad_norm": 0.8125,
      "learning_rate": 0.0007221997077968103,
      "loss": 1.3268,
      "step": 20555
    },
    {
      "epoch": 1.7685623333046545,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007221754836034174,
      "loss": 1.1976,
      "step": 20556
    },
    {
      "epoch": 1.7686483696119764,
      "grad_norm": 4.65625,
      "learning_rate": 0.0007221512587602048,
      "loss": 1.6611,
      "step": 20557
    },
    {
      "epoch": 1.768734405919298,
      "grad_norm": 1.375,
      "learning_rate": 0.0007221270332672427,
      "loss": 1.1848,
      "step": 20558
    },
    {
      "epoch": 1.7688204422266196,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007221028071246024,
      "loss": 1.2972,
      "step": 20559
    },
    {
      "epoch": 1.7689064785339412,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007220785803323545,
      "loss": 1.2471,
      "step": 20560
    },
    {
      "epoch": 1.7689925148412629,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007220543528905698,
      "loss": 0.8467,
      "step": 20561
    },
    {
      "epoch": 1.7690785511485847,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007220301247993196,
      "loss": 1.9582,
      "step": 20562
    },
    {
      "epoch": 1.7691645874559065,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007220058960586742,
      "loss": 1.6209,
      "step": 20563
    },
    {
      "epoch": 1.7692506237632282,
      "grad_norm": 1.921875,
      "learning_rate": 0.0007219816666687048,
      "loss": 1.349,
      "step": 20564
    },
    {
      "epoch": 1.7693366600705498,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007219574366294823,
      "loss": 1.4134,
      "step": 20565
    },
    {
      "epoch": 1.7694226963778714,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007219332059410775,
      "loss": 1.4839,
      "step": 20566
    },
    {
      "epoch": 1.769508732685193,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007219089746035611,
      "loss": 1.0844,
      "step": 20567
    },
    {
      "epoch": 1.7695947689925149,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007218847426170041,
      "loss": 2.3704,
      "step": 20568
    },
    {
      "epoch": 1.7696808052998365,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007218605099814775,
      "loss": 0.99,
      "step": 20569
    },
    {
      "epoch": 1.7697668416071584,
      "grad_norm": 1.15625,
      "learning_rate": 0.000721836276697052,
      "loss": 0.8795,
      "step": 20570
    },
    {
      "epoch": 1.76985287791448,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007218120427637984,
      "loss": 1.5196,
      "step": 20571
    },
    {
      "epoch": 1.7699389142218016,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007217878081817879,
      "loss": 1.8908,
      "step": 20572
    },
    {
      "epoch": 1.7700249505291232,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007217635729510911,
      "loss": 1.1751,
      "step": 20573
    },
    {
      "epoch": 1.7701109868364449,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007217393370717792,
      "loss": 1.5443,
      "step": 20574
    },
    {
      "epoch": 1.7701970231437667,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007217151005439226,
      "loss": 0.6668,
      "step": 20575
    },
    {
      "epoch": 1.7702830594510883,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007216908633675925,
      "loss": 1.0686,
      "step": 20576
    },
    {
      "epoch": 1.7703690957584102,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007216666255428598,
      "loss": 1.0939,
      "step": 20577
    },
    {
      "epoch": 1.7704551320657318,
      "grad_norm": 1.515625,
      "learning_rate": 0.0007216423870697955,
      "loss": 1.2634,
      "step": 20578
    },
    {
      "epoch": 1.7705411683730534,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00072161814794847,
      "loss": 1.5576,
      "step": 20579
    },
    {
      "epoch": 1.770627204680375,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007215939081789546,
      "loss": 1.2066,
      "step": 20580
    },
    {
      "epoch": 1.7707132409876967,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007215696677613201,
      "loss": 0.8347,
      "step": 20581
    },
    {
      "epoch": 1.7707992772950185,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007215454266956376,
      "loss": 1.8183,
      "step": 20582
    },
    {
      "epoch": 1.7708853136023401,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007215211849819776,
      "loss": 1.3908,
      "step": 20583
    },
    {
      "epoch": 1.770971349909662,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007214969426204115,
      "loss": 1.5687,
      "step": 20584
    },
    {
      "epoch": 1.7710573862169836,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007214726996110097,
      "loss": 1.1403,
      "step": 20585
    },
    {
      "epoch": 1.7711434225243052,
      "grad_norm": 1.5,
      "learning_rate": 0.0007214484559538433,
      "loss": 1.1566,
      "step": 20586
    },
    {
      "epoch": 1.7712294588316269,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007214242116489833,
      "loss": 1.7177,
      "step": 20587
    },
    {
      "epoch": 1.7713154951389485,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007213999666965006,
      "loss": 1.369,
      "step": 20588
    },
    {
      "epoch": 1.7714015314462703,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000721375721096466,
      "loss": 1.5875,
      "step": 20589
    },
    {
      "epoch": 1.7714875677535922,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007213514748489505,
      "loss": 1.0942,
      "step": 20590
    },
    {
      "epoch": 1.7715736040609138,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0007213272279540249,
      "loss": 0.7784,
      "step": 20591
    },
    {
      "epoch": 1.7716596403682354,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007213029804117603,
      "loss": 1.6218,
      "step": 20592
    },
    {
      "epoch": 1.771745676675557,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007212787322222275,
      "loss": 0.8405,
      "step": 20593
    },
    {
      "epoch": 1.7718317129828787,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007212544833854975,
      "loss": 0.6645,
      "step": 20594
    },
    {
      "epoch": 1.7719177492902005,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000721230233901641,
      "loss": 1.19,
      "step": 20595
    },
    {
      "epoch": 1.7720037855975221,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007212059837707293,
      "loss": 1.4929,
      "step": 20596
    },
    {
      "epoch": 1.772089821904844,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000721181732992833,
      "loss": 1.1768,
      "step": 20597
    },
    {
      "epoch": 1.7721758582121656,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000721157481568023,
      "loss": 0.909,
      "step": 20598
    },
    {
      "epoch": 1.7722618945194872,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007211332294963706,
      "loss": 1.3904,
      "step": 20599
    },
    {
      "epoch": 1.7723479308268089,
      "grad_norm": 1.5,
      "learning_rate": 0.0007211089767779465,
      "loss": 1.6658,
      "step": 20600
    },
    {
      "epoch": 1.7724339671341305,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007210847234128216,
      "loss": 1.5828,
      "step": 20601
    },
    {
      "epoch": 1.7725200034414523,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007210604694010669,
      "loss": 1.0122,
      "step": 20602
    },
    {
      "epoch": 1.772606039748774,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007210362147427533,
      "loss": 1.0173,
      "step": 20603
    },
    {
      "epoch": 1.7726920760560958,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007210119594379519,
      "loss": 2.0648,
      "step": 20604
    },
    {
      "epoch": 1.7727781123634174,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007209877034867331,
      "loss": 1.446,
      "step": 20605
    },
    {
      "epoch": 1.772864148670739,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007209634468891686,
      "loss": 0.7692,
      "step": 20606
    },
    {
      "epoch": 1.7729501849780607,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000720939189645329,
      "loss": 0.8361,
      "step": 20607
    },
    {
      "epoch": 1.7730362212853823,
      "grad_norm": 1.125,
      "learning_rate": 0.0007209149317552852,
      "loss": 0.8855,
      "step": 20608
    },
    {
      "epoch": 1.7731222575927041,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000720890673219108,
      "loss": 1.2547,
      "step": 20609
    },
    {
      "epoch": 1.7732082939000258,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007208664140368688,
      "loss": 0.8774,
      "step": 20610
    },
    {
      "epoch": 1.7732943302073476,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007208421542086382,
      "loss": 0.8406,
      "step": 20611
    },
    {
      "epoch": 1.7733803665146692,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007208178937344871,
      "loss": 1.1324,
      "step": 20612
    },
    {
      "epoch": 1.7734664028219909,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007207936326144868,
      "loss": 1.599,
      "step": 20613
    },
    {
      "epoch": 1.7735524391293125,
      "grad_norm": 0.890625,
      "learning_rate": 0.000720769370848708,
      "loss": 1.2754,
      "step": 20614
    },
    {
      "epoch": 1.773638475436634,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007207451084372218,
      "loss": 1.0926,
      "step": 20615
    },
    {
      "epoch": 1.773724511743956,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007207208453800989,
      "loss": 1.5926,
      "step": 20616
    },
    {
      "epoch": 1.7738105480512778,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007206965816774107,
      "loss": 1.2369,
      "step": 20617
    },
    {
      "epoch": 1.7738965843585994,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007206723173292277,
      "loss": 1.0164,
      "step": 20618
    },
    {
      "epoch": 1.773982620665921,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007206480523356212,
      "loss": 0.9882,
      "step": 20619
    },
    {
      "epoch": 1.7740686569732427,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007206237866966619,
      "loss": 0.9123,
      "step": 20620
    },
    {
      "epoch": 1.7741546932805643,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007205995204124212,
      "loss": 1.358,
      "step": 20621
    },
    {
      "epoch": 1.774240729587886,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007205752534829697,
      "loss": 0.8467,
      "step": 20622
    },
    {
      "epoch": 1.7743267658952078,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007205509859083785,
      "loss": 0.703,
      "step": 20623
    },
    {
      "epoch": 1.7744128022025296,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007205267176887184,
      "loss": 1.8354,
      "step": 20624
    },
    {
      "epoch": 1.7744988385098512,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007205024488240607,
      "loss": 1.5366,
      "step": 20625
    },
    {
      "epoch": 1.7745848748171729,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007204781793144763,
      "loss": 1.0453,
      "step": 20626
    },
    {
      "epoch": 1.7746709111244945,
      "grad_norm": 1.09375,
      "learning_rate": 0.000720453909160036,
      "loss": 1.6476,
      "step": 20627
    },
    {
      "epoch": 1.7747569474318161,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007204296383608109,
      "loss": 1.7943,
      "step": 20628
    },
    {
      "epoch": 1.774842983739138,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000720405366916872,
      "loss": 0.9985,
      "step": 20629
    },
    {
      "epoch": 1.7749290200464596,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007203810948282903,
      "loss": 0.8576,
      "step": 20630
    },
    {
      "epoch": 1.7750150563537814,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007203568220951367,
      "loss": 1.4809,
      "step": 20631
    },
    {
      "epoch": 1.775101092661103,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007203325487174822,
      "loss": 1.0457,
      "step": 20632
    },
    {
      "epoch": 1.7751871289684247,
      "grad_norm": 0.953125,
      "learning_rate": 0.000720308274695398,
      "loss": 0.835,
      "step": 20633
    },
    {
      "epoch": 1.7752731652757463,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007202840000289548,
      "loss": 0.4552,
      "step": 20634
    },
    {
      "epoch": 1.775359201583068,
      "grad_norm": 1.125,
      "learning_rate": 0.0007202597247182237,
      "loss": 1.4293,
      "step": 20635
    },
    {
      "epoch": 1.7754452378903898,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007202354487632757,
      "loss": 1.2623,
      "step": 20636
    },
    {
      "epoch": 1.7755312741977114,
      "grad_norm": 1.71875,
      "learning_rate": 0.0007202111721641821,
      "loss": 1.2953,
      "step": 20637
    },
    {
      "epoch": 1.7756173105050332,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007201868949210135,
      "loss": 0.854,
      "step": 20638
    },
    {
      "epoch": 1.7757033468123549,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007201626170338411,
      "loss": 1.1215,
      "step": 20639
    },
    {
      "epoch": 1.7757893831196765,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007201383385027358,
      "loss": 1.0181,
      "step": 20640
    },
    {
      "epoch": 1.7758754194269981,
      "grad_norm": 1.25,
      "learning_rate": 0.0007201140593277687,
      "loss": 0.9089,
      "step": 20641
    },
    {
      "epoch": 1.7759614557343197,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007200897795090109,
      "loss": 1.012,
      "step": 20642
    },
    {
      "epoch": 1.7760474920416416,
      "grad_norm": 1.125,
      "learning_rate": 0.0007200654990465331,
      "loss": 0.9799,
      "step": 20643
    },
    {
      "epoch": 1.7761335283489632,
      "grad_norm": 0.8125,
      "learning_rate": 0.0007200412179404068,
      "loss": 1.0137,
      "step": 20644
    },
    {
      "epoch": 1.776219564656285,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0007200169361907025,
      "loss": 0.938,
      "step": 20645
    },
    {
      "epoch": 1.7763056009636067,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007199926537974914,
      "loss": 1.4421,
      "step": 20646
    },
    {
      "epoch": 1.7763916372709283,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007199683707608447,
      "loss": 1.0985,
      "step": 20647
    },
    {
      "epoch": 1.77647767357825,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007199440870808334,
      "loss": 0.9464,
      "step": 20648
    },
    {
      "epoch": 1.7765637098855716,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007199198027575284,
      "loss": 1.573,
      "step": 20649
    },
    {
      "epoch": 1.7766497461928934,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007198955177910005,
      "loss": 1.9098,
      "step": 20650
    },
    {
      "epoch": 1.7767357825002152,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007198712321813211,
      "loss": 1.4854,
      "step": 20651
    },
    {
      "epoch": 1.7768218188075369,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007198469459285613,
      "loss": 1.3185,
      "step": 20652
    },
    {
      "epoch": 1.7769078551148585,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007198226590327916,
      "loss": 0.8956,
      "step": 20653
    },
    {
      "epoch": 1.7769938914221801,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007197983714940836,
      "loss": 1.1262,
      "step": 20654
    },
    {
      "epoch": 1.7770799277295017,
      "grad_norm": 1.21875,
      "learning_rate": 0.000719774083312508,
      "loss": 1.0813,
      "step": 20655
    },
    {
      "epoch": 1.7771659640368236,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007197497944881361,
      "loss": 1.2617,
      "step": 20656
    },
    {
      "epoch": 1.7772520003441452,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007197255050210387,
      "loss": 0.8749,
      "step": 20657
    },
    {
      "epoch": 1.777338036651467,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007197012149112869,
      "loss": 0.6855,
      "step": 20658
    },
    {
      "epoch": 1.7774240729587887,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007196769241589518,
      "loss": 0.9222,
      "step": 20659
    },
    {
      "epoch": 1.7775101092661103,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007196526327641044,
      "loss": 2.2434,
      "step": 20660
    },
    {
      "epoch": 1.777596145573432,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007196283407268158,
      "loss": 1.3857,
      "step": 20661
    },
    {
      "epoch": 1.7776821818807536,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007196040480471571,
      "loss": 0.9231,
      "step": 20662
    },
    {
      "epoch": 1.7777682181880754,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007195797547251993,
      "loss": 1.0742,
      "step": 20663
    },
    {
      "epoch": 1.777854254495397,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007195554607610133,
      "loss": 1.3798,
      "step": 20664
    },
    {
      "epoch": 1.7779402908027189,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007195311661546701,
      "loss": 1.0509,
      "step": 20665
    },
    {
      "epoch": 1.7780263271100405,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007195068709062412,
      "loss": 1.1046,
      "step": 20666
    },
    {
      "epoch": 1.7781123634173621,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007194825750157974,
      "loss": 1.1437,
      "step": 20667
    },
    {
      "epoch": 1.7781983997246837,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007194582784834096,
      "loss": 1.2473,
      "step": 20668
    },
    {
      "epoch": 1.7782844360320054,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007194339813091492,
      "loss": 0.7015,
      "step": 20669
    },
    {
      "epoch": 1.7783704723393272,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000719409683493087,
      "loss": 1.1108,
      "step": 20670
    },
    {
      "epoch": 1.7784565086466488,
      "grad_norm": 1.390625,
      "learning_rate": 0.000719385385035294,
      "loss": 0.7032,
      "step": 20671
    },
    {
      "epoch": 1.7785425449539707,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007193610859358416,
      "loss": 1.4237,
      "step": 20672
    },
    {
      "epoch": 1.7786285812612923,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007193367861948006,
      "loss": 1.0993,
      "step": 20673
    },
    {
      "epoch": 1.778714617568614,
      "grad_norm": 1.0,
      "learning_rate": 0.0007193124858122422,
      "loss": 1.5325,
      "step": 20674
    },
    {
      "epoch": 1.7788006538759356,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0007192881847882374,
      "loss": 0.9736,
      "step": 20675
    },
    {
      "epoch": 1.7788866901832572,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007192638831228573,
      "loss": 0.5451,
      "step": 20676
    },
    {
      "epoch": 1.778972726490579,
      "grad_norm": 1.328125,
      "learning_rate": 0.000719239580816173,
      "loss": 1.5795,
      "step": 20677
    },
    {
      "epoch": 1.7790587627979009,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007192152778682557,
      "loss": 1.1931,
      "step": 20678
    },
    {
      "epoch": 1.7791447991052225,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007191909742791762,
      "loss": 1.1144,
      "step": 20679
    },
    {
      "epoch": 1.7792308354125441,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007191666700490057,
      "loss": 1.8213,
      "step": 20680
    },
    {
      "epoch": 1.7793168717198657,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007191423651778154,
      "loss": 0.7242,
      "step": 20681
    },
    {
      "epoch": 1.7794029080271874,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007191180596656763,
      "loss": 0.8421,
      "step": 20682
    },
    {
      "epoch": 1.7794889443345092,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007190937535126596,
      "loss": 1.6195,
      "step": 20683
    },
    {
      "epoch": 1.7795749806418308,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000719069446718836,
      "loss": 1.1073,
      "step": 20684
    },
    {
      "epoch": 1.7796610169491527,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000719045139284277,
      "loss": 1.0711,
      "step": 20685
    },
    {
      "epoch": 1.7797470532564743,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007190208312090537,
      "loss": 1.2401,
      "step": 20686
    },
    {
      "epoch": 1.779833089563796,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007189965224932369,
      "loss": 1.3752,
      "step": 20687
    },
    {
      "epoch": 1.7799191258711176,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000718972213136898,
      "loss": 0.7869,
      "step": 20688
    },
    {
      "epoch": 1.7800051621784392,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0007189479031401079,
      "loss": 0.6201,
      "step": 20689
    },
    {
      "epoch": 1.780091198485761,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007189235925029378,
      "loss": 1.6603,
      "step": 20690
    },
    {
      "epoch": 1.7801772347930827,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007188992812254586,
      "loss": 1.3729,
      "step": 20691
    },
    {
      "epoch": 1.7802632711004045,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007188749693077417,
      "loss": 1.2344,
      "step": 20692
    },
    {
      "epoch": 1.7803493074077261,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007188506567498582,
      "loss": 1.4512,
      "step": 20693
    },
    {
      "epoch": 1.7804353437150477,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007188263435518789,
      "loss": 1.5361,
      "step": 20694
    },
    {
      "epoch": 1.7805213800223694,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007188020297138751,
      "loss": 1.2849,
      "step": 20695
    },
    {
      "epoch": 1.780607416329691,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007187777152359181,
      "loss": 1.3098,
      "step": 20696
    },
    {
      "epoch": 1.7806934526370128,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007187534001180788,
      "loss": 0.7681,
      "step": 20697
    },
    {
      "epoch": 1.7807794889443345,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0007187290843604282,
      "loss": 1.8419,
      "step": 20698
    },
    {
      "epoch": 1.7808655252516563,
      "grad_norm": 1.875,
      "learning_rate": 0.0007187047679630376,
      "loss": 0.7724,
      "step": 20699
    },
    {
      "epoch": 1.780951561558978,
      "grad_norm": 1.703125,
      "learning_rate": 0.0007186804509259783,
      "loss": 0.6336,
      "step": 20700
    },
    {
      "epoch": 1.7810375978662996,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007186561332493211,
      "loss": 0.7969,
      "step": 20701
    },
    {
      "epoch": 1.7811236341736212,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000718631814933137,
      "loss": 1.5042,
      "step": 20702
    },
    {
      "epoch": 1.7812096704809428,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007186074959774977,
      "loss": 1.4785,
      "step": 20703
    },
    {
      "epoch": 1.7812957067882647,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007185831763824738,
      "loss": 1.2196,
      "step": 20704
    },
    {
      "epoch": 1.7813817430955865,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007185588561481368,
      "loss": 1.2154,
      "step": 20705
    },
    {
      "epoch": 1.7814677794029081,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007185345352745574,
      "loss": 1.104,
      "step": 20706
    },
    {
      "epoch": 1.7815538157102297,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007185102137618071,
      "loss": 1.2677,
      "step": 20707
    },
    {
      "epoch": 1.7816398520175514,
      "grad_norm": 1.03125,
      "learning_rate": 0.000718485891609957,
      "loss": 0.8146,
      "step": 20708
    },
    {
      "epoch": 1.781725888324873,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007184615688190781,
      "loss": 0.8299,
      "step": 20709
    },
    {
      "epoch": 1.7818119246321946,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0007184372453892415,
      "loss": 1.8602,
      "step": 20710
    },
    {
      "epoch": 1.7818979609395165,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007184129213205184,
      "loss": 0.9764,
      "step": 20711
    },
    {
      "epoch": 1.7819839972468383,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007183885966129803,
      "loss": 1.1362,
      "step": 20712
    },
    {
      "epoch": 1.78207003355416,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007183642712666978,
      "loss": 1.1922,
      "step": 20713
    },
    {
      "epoch": 1.7821560698614816,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007183399452817423,
      "loss": 1.7146,
      "step": 20714
    },
    {
      "epoch": 1.7822421061688032,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000718315618658185,
      "loss": 1.5471,
      "step": 20715
    },
    {
      "epoch": 1.7823281424761248,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000718291291396097,
      "loss": 1.7822,
      "step": 20716
    },
    {
      "epoch": 1.7824141787834467,
      "grad_norm": 2.3125,
      "learning_rate": 0.0007182669634955493,
      "loss": 1.99,
      "step": 20717
    },
    {
      "epoch": 1.7825002150907683,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007182426349566131,
      "loss": 1.2057,
      "step": 20718
    },
    {
      "epoch": 1.7825862513980901,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007182183057793599,
      "loss": 1.5432,
      "step": 20719
    },
    {
      "epoch": 1.7826722877054118,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007181939759638605,
      "loss": 0.8774,
      "step": 20720
    },
    {
      "epoch": 1.7827583240127334,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000718169645510186,
      "loss": 1.6446,
      "step": 20721
    },
    {
      "epoch": 1.782844360320055,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007181453144184079,
      "loss": 1.0687,
      "step": 20722
    },
    {
      "epoch": 1.7829303966273766,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007181209826885972,
      "loss": 1.0799,
      "step": 20723
    },
    {
      "epoch": 1.7830164329346985,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007180966503208249,
      "loss": 0.7009,
      "step": 20724
    },
    {
      "epoch": 1.78310246924202,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007180723173151625,
      "loss": 1.004,
      "step": 20725
    },
    {
      "epoch": 1.783188505549342,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007180479836716808,
      "loss": 1.969,
      "step": 20726
    },
    {
      "epoch": 1.7832745418566636,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007180236493904513,
      "loss": 1.4696,
      "step": 20727
    },
    {
      "epoch": 1.7833605781639852,
      "grad_norm": 1.25,
      "learning_rate": 0.0007179993144715449,
      "loss": 1.5354,
      "step": 20728
    },
    {
      "epoch": 1.7834466144713068,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007179749789150328,
      "loss": 1.8281,
      "step": 20729
    },
    {
      "epoch": 1.7835326507786284,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007179506427209865,
      "loss": 1.2012,
      "step": 20730
    },
    {
      "epoch": 1.7836186870859503,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000717926305889477,
      "loss": 1.7676,
      "step": 20731
    },
    {
      "epoch": 1.783704723393272,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007179019684205752,
      "loss": 1.571,
      "step": 20732
    },
    {
      "epoch": 1.7837907597005938,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0007178776303143527,
      "loss": 1.2758,
      "step": 20733
    },
    {
      "epoch": 1.7838767960079154,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007178532915708806,
      "loss": 1.7921,
      "step": 20734
    },
    {
      "epoch": 1.783962832315237,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007178289521902298,
      "loss": 2.2577,
      "step": 20735
    },
    {
      "epoch": 1.7840488686225586,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007178046121724717,
      "loss": 1.7576,
      "step": 20736
    },
    {
      "epoch": 1.7841349049298802,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007177802715176776,
      "loss": 1.1589,
      "step": 20737
    },
    {
      "epoch": 1.784220941237202,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007177559302259185,
      "loss": 1.3194,
      "step": 20738
    },
    {
      "epoch": 1.784306977544524,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0007177315882972656,
      "loss": 1.0397,
      "step": 20739
    },
    {
      "epoch": 1.7843930138518456,
      "grad_norm": 1.5,
      "learning_rate": 0.0007177072457317901,
      "loss": 1.2101,
      "step": 20740
    },
    {
      "epoch": 1.7844790501591672,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007176829025295633,
      "loss": 0.8879,
      "step": 20741
    },
    {
      "epoch": 1.7845650864664888,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007176585586906566,
      "loss": 0.7972,
      "step": 20742
    },
    {
      "epoch": 1.7846511227738104,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007176342142151405,
      "loss": 1.3676,
      "step": 20743
    },
    {
      "epoch": 1.7847371590811323,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000717609869103087,
      "loss": 1.3126,
      "step": 20744
    },
    {
      "epoch": 1.784823195388454,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007175855233545668,
      "loss": 0.947,
      "step": 20745
    },
    {
      "epoch": 1.7849092316957758,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007175611769696513,
      "loss": 1.2579,
      "step": 20746
    },
    {
      "epoch": 1.7849952680030974,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007175368299484115,
      "loss": 1.7883,
      "step": 20747
    },
    {
      "epoch": 1.785081304310419,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000717512482290919,
      "loss": 1.2977,
      "step": 20748
    },
    {
      "epoch": 1.7851673406177406,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007174881339972448,
      "loss": 1.2477,
      "step": 20749
    },
    {
      "epoch": 1.7852533769250623,
      "grad_norm": 1.21875,
      "learning_rate": 0.00071746378506746,
      "loss": 1.1261,
      "step": 20750
    },
    {
      "epoch": 1.785339413232384,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007174394355016359,
      "loss": 1.7096,
      "step": 20751
    },
    {
      "epoch": 1.7854254495397057,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007174150852998439,
      "loss": 1.1617,
      "step": 20752
    },
    {
      "epoch": 1.7855114858470276,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000717390734462155,
      "loss": 2.0025,
      "step": 20753
    },
    {
      "epoch": 1.7855975221543492,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007173663829886405,
      "loss": 1.0393,
      "step": 20754
    },
    {
      "epoch": 1.7856835584616708,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007173420308793715,
      "loss": 0.9856,
      "step": 20755
    },
    {
      "epoch": 1.7857695947689924,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007173176781344194,
      "loss": 1.0306,
      "step": 20756
    },
    {
      "epoch": 1.785855631076314,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007172933247538554,
      "loss": 1.5548,
      "step": 20757
    },
    {
      "epoch": 1.785941667383636,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007172689707377505,
      "loss": 1.5995,
      "step": 20758
    },
    {
      "epoch": 1.7860277036909575,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007172446160861763,
      "loss": 0.9193,
      "step": 20759
    },
    {
      "epoch": 1.7861137399982794,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007172202607992038,
      "loss": 1.3094,
      "step": 20760
    },
    {
      "epoch": 1.786199776305601,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007171959048769043,
      "loss": 1.0914,
      "step": 20761
    },
    {
      "epoch": 1.7862858126129226,
      "grad_norm": 1.328125,
      "learning_rate": 0.000717171548319349,
      "loss": 1.6488,
      "step": 20762
    },
    {
      "epoch": 1.7863718489202443,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007171471911266092,
      "loss": 1.0106,
      "step": 20763
    },
    {
      "epoch": 1.7864578852275659,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000717122833298756,
      "loss": 0.7232,
      "step": 20764
    },
    {
      "epoch": 1.7865439215348877,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007170984748358607,
      "loss": 1.3914,
      "step": 20765
    },
    {
      "epoch": 1.7866299578422096,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007170741157379947,
      "loss": 1.2216,
      "step": 20766
    },
    {
      "epoch": 1.7867159941495312,
      "grad_norm": 1.125,
      "learning_rate": 0.0007170497560052291,
      "loss": 1.3486,
      "step": 20767
    },
    {
      "epoch": 1.7868020304568528,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0007170253956376352,
      "loss": 0.6397,
      "step": 20768
    },
    {
      "epoch": 1.7868880667641744,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007170010346352842,
      "loss": 2.012,
      "step": 20769
    },
    {
      "epoch": 1.786974103071496,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007169766729982474,
      "loss": 1.1806,
      "step": 20770
    },
    {
      "epoch": 1.787060139378818,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000716952310726596,
      "loss": 1.7827,
      "step": 20771
    },
    {
      "epoch": 1.7871461756861395,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007169279478204014,
      "loss": 1.0806,
      "step": 20772
    },
    {
      "epoch": 1.7872322119934614,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007169035842797346,
      "loss": 1.2846,
      "step": 20773
    },
    {
      "epoch": 1.787318248300783,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007168792201046672,
      "loss": 1.1354,
      "step": 20774
    },
    {
      "epoch": 1.7874042846081046,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007168548552952701,
      "loss": 1.648,
      "step": 20775
    },
    {
      "epoch": 1.7874903209154263,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007168304898516147,
      "loss": 1.2452,
      "step": 20776
    },
    {
      "epoch": 1.7875763572227479,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007168061237737723,
      "loss": 1.5141,
      "step": 20777
    },
    {
      "epoch": 1.7876623935300697,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007167817570618143,
      "loss": 0.9573,
      "step": 20778
    },
    {
      "epoch": 1.7877484298373914,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007167573897158116,
      "loss": 1.6018,
      "step": 20779
    },
    {
      "epoch": 1.7878344661447132,
      "grad_norm": 1.125,
      "learning_rate": 0.0007167330217358358,
      "loss": 1.1119,
      "step": 20780
    },
    {
      "epoch": 1.7879205024520348,
      "grad_norm": 1.265625,
      "learning_rate": 0.000716708653121958,
      "loss": 1.5862,
      "step": 20781
    },
    {
      "epoch": 1.7880065387593564,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0007166842838742496,
      "loss": 0.8764,
      "step": 20782
    },
    {
      "epoch": 1.788092575066678,
      "grad_norm": 1.09375,
      "learning_rate": 0.000716659913992782,
      "loss": 1.9035,
      "step": 20783
    },
    {
      "epoch": 1.7881786113739997,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007166355434776259,
      "loss": 1.5706,
      "step": 20784
    },
    {
      "epoch": 1.7882646476813215,
      "grad_norm": 1.4375,
      "learning_rate": 0.0007166111723288531,
      "loss": 1.7826,
      "step": 20785
    },
    {
      "epoch": 1.7883506839886432,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000716586800546535,
      "loss": 1.745,
      "step": 20786
    },
    {
      "epoch": 1.788436720295965,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007165624281307424,
      "loss": 1.1201,
      "step": 20787
    },
    {
      "epoch": 1.7885227566032866,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007165380550815468,
      "loss": 1.5439,
      "step": 20788
    },
    {
      "epoch": 1.7886087929106083,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007165136813990197,
      "loss": 1.1601,
      "step": 20789
    },
    {
      "epoch": 1.7886948292179299,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007164893070832321,
      "loss": 1.2742,
      "step": 20790
    },
    {
      "epoch": 1.7887808655252515,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007164649321342553,
      "loss": 0.6509,
      "step": 20791
    },
    {
      "epoch": 1.7888669018325734,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007164405565521606,
      "loss": 1.3424,
      "step": 20792
    },
    {
      "epoch": 1.7889529381398952,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007164161803370195,
      "loss": 0.698,
      "step": 20793
    },
    {
      "epoch": 1.7890389744472168,
      "grad_norm": 0.875,
      "learning_rate": 0.0007163918034889032,
      "loss": 0.6411,
      "step": 20794
    },
    {
      "epoch": 1.7891250107545384,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007163674260078829,
      "loss": 1.6691,
      "step": 20795
    },
    {
      "epoch": 1.78921104706186,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00071634304789403,
      "loss": 1.7197,
      "step": 20796
    },
    {
      "epoch": 1.7892970833691817,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007163186691474157,
      "loss": 0.746,
      "step": 20797
    },
    {
      "epoch": 1.7893831196765033,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007162942897681115,
      "loss": 0.9274,
      "step": 20798
    },
    {
      "epoch": 1.7894691559838252,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007162699097561883,
      "loss": 0.8129,
      "step": 20799
    },
    {
      "epoch": 1.789555192291147,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007162455291117179,
      "loss": 1.569,
      "step": 20800
    },
    {
      "epoch": 1.7896412285984686,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007162211478347714,
      "loss": 1.1426,
      "step": 20801
    },
    {
      "epoch": 1.7897272649057903,
      "grad_norm": 1.015625,
      "learning_rate": 0.00071619676592542,
      "loss": 0.9688,
      "step": 20802
    },
    {
      "epoch": 1.7898133012131119,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007161723833837352,
      "loss": 1.4676,
      "step": 20803
    },
    {
      "epoch": 1.7898993375204335,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000716148000209788,
      "loss": 2.0,
      "step": 20804
    },
    {
      "epoch": 1.7899853738277554,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007161236164036501,
      "loss": 1.1518,
      "step": 20805
    },
    {
      "epoch": 1.790071410135077,
      "grad_norm": 1.0,
      "learning_rate": 0.0007160992319653927,
      "loss": 1.0948,
      "step": 20806
    },
    {
      "epoch": 1.7901574464423988,
      "grad_norm": 1.1875,
      "learning_rate": 0.000716074846895087,
      "loss": 0.9296,
      "step": 20807
    },
    {
      "epoch": 1.7902434827497204,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007160504611928044,
      "loss": 1.0795,
      "step": 20808
    },
    {
      "epoch": 1.790329519057042,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007160260748586163,
      "loss": 1.3131,
      "step": 20809
    },
    {
      "epoch": 1.7904155553643637,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007160016878925939,
      "loss": 0.7431,
      "step": 20810
    },
    {
      "epoch": 1.7905015916716853,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007159773002948085,
      "loss": 1.3281,
      "step": 20811
    },
    {
      "epoch": 1.7905876279790072,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007159529120653315,
      "loss": 1.6739,
      "step": 20812
    },
    {
      "epoch": 1.7906736642863288,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007159285232042344,
      "loss": 1.4889,
      "step": 20813
    },
    {
      "epoch": 1.7907597005936506,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007159041337115881,
      "loss": 1.5611,
      "step": 20814
    },
    {
      "epoch": 1.7908457369009723,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007158797435874645,
      "loss": 0.8236,
      "step": 20815
    },
    {
      "epoch": 1.7909317732082939,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007158553528319345,
      "loss": 1.3345,
      "step": 20816
    },
    {
      "epoch": 1.7910178095156155,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007158309614450696,
      "loss": 1.2287,
      "step": 20817
    },
    {
      "epoch": 1.7911038458229371,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000715806569426941,
      "loss": 2.1305,
      "step": 20818
    },
    {
      "epoch": 1.791189882130259,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007157821767776203,
      "loss": 1.4361,
      "step": 20819
    },
    {
      "epoch": 1.7912759184375806,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007157577834971787,
      "loss": 1.4671,
      "step": 20820
    },
    {
      "epoch": 1.7913619547449025,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007157333895856874,
      "loss": 1.2298,
      "step": 20821
    },
    {
      "epoch": 1.791447991052224,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007157089950432179,
      "loss": 1.3751,
      "step": 20822
    },
    {
      "epoch": 1.7915340273595457,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007156845998698416,
      "loss": 1.1176,
      "step": 20823
    },
    {
      "epoch": 1.7916200636668673,
      "grad_norm": 0.96875,
      "learning_rate": 0.00071566020406563,
      "loss": 0.9868,
      "step": 20824
    },
    {
      "epoch": 1.791706099974189,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007156358076306539,
      "loss": 1.1346,
      "step": 20825
    },
    {
      "epoch": 1.7917921362815108,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007156114105649851,
      "loss": 1.4328,
      "step": 20826
    },
    {
      "epoch": 1.7918781725888326,
      "grad_norm": 1.15625,
      "learning_rate": 0.000715587012868695,
      "loss": 1.4445,
      "step": 20827
    },
    {
      "epoch": 1.7919642088961543,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007155626145418548,
      "loss": 0.78,
      "step": 20828
    },
    {
      "epoch": 1.7920502452034759,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007155382155845357,
      "loss": 0.8125,
      "step": 20829
    },
    {
      "epoch": 1.7921362815107975,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007155138159968094,
      "loss": 1.8046,
      "step": 20830
    },
    {
      "epoch": 1.7922223178181191,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007154894157787471,
      "loss": 1.3937,
      "step": 20831
    },
    {
      "epoch": 1.792308354125441,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007154650149304202,
      "loss": 1.004,
      "step": 20832
    },
    {
      "epoch": 1.7923943904327626,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007154406134518998,
      "loss": 1.3813,
      "step": 20833
    },
    {
      "epoch": 1.7924804267400845,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007154162113432578,
      "loss": 1.1369,
      "step": 20834
    },
    {
      "epoch": 1.792566463047406,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007153918086045652,
      "loss": 1.9007,
      "step": 20835
    },
    {
      "epoch": 1.7926524993547277,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007153674052358934,
      "loss": 0.9282,
      "step": 20836
    },
    {
      "epoch": 1.7927385356620493,
      "grad_norm": 2.59375,
      "learning_rate": 0.0007153430012373136,
      "loss": 1.0284,
      "step": 20837
    },
    {
      "epoch": 1.792824571969371,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007153185966088978,
      "loss": 1.1333,
      "step": 20838
    },
    {
      "epoch": 1.7929106082766928,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007152941913507167,
      "loss": 1.9048,
      "step": 20839
    },
    {
      "epoch": 1.7929966445840144,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000715269785462842,
      "loss": 1.038,
      "step": 20840
    },
    {
      "epoch": 1.7930826808913363,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007152453789453451,
      "loss": 1.3244,
      "step": 20841
    },
    {
      "epoch": 1.793168717198658,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0007152209717982975,
      "loss": 0.9324,
      "step": 20842
    },
    {
      "epoch": 1.7932547535059795,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007151965640217702,
      "loss": 1.5138,
      "step": 20843
    },
    {
      "epoch": 1.7933407898133011,
      "grad_norm": 1.25,
      "learning_rate": 0.0007151721556158348,
      "loss": 1.2379,
      "step": 20844
    },
    {
      "epoch": 1.7934268261206228,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007151477465805628,
      "loss": 2.1075,
      "step": 20845
    },
    {
      "epoch": 1.7935128624279446,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007151233369160255,
      "loss": 1.5372,
      "step": 20846
    },
    {
      "epoch": 1.7935988987352662,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007150989266222943,
      "loss": 0.8239,
      "step": 20847
    },
    {
      "epoch": 1.793684935042588,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007150745156994404,
      "loss": 1.1099,
      "step": 20848
    },
    {
      "epoch": 1.7937709713499097,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007150501041475355,
      "loss": 0.9582,
      "step": 20849
    },
    {
      "epoch": 1.7938570076572313,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007150256919666509,
      "loss": 1.1374,
      "step": 20850
    },
    {
      "epoch": 1.793943043964553,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007150012791568578,
      "loss": 1.7203,
      "step": 20851
    },
    {
      "epoch": 1.7940290802718746,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000714976865718228,
      "loss": 1.1641,
      "step": 20852
    },
    {
      "epoch": 1.7941151165791964,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007149524516508325,
      "loss": 1.3942,
      "step": 20853
    },
    {
      "epoch": 1.7942011528865183,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007149280369547431,
      "loss": 1.9696,
      "step": 20854
    },
    {
      "epoch": 1.79428718919384,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007149036216300307,
      "loss": 1.5768,
      "step": 20855
    },
    {
      "epoch": 1.7943732255011615,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007148792056767671,
      "loss": 1.6425,
      "step": 20856
    },
    {
      "epoch": 1.7944592618084831,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007148547890950238,
      "loss": 1.463,
      "step": 20857
    },
    {
      "epoch": 1.7945452981158048,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007148303718848718,
      "loss": 1.4707,
      "step": 20858
    },
    {
      "epoch": 1.7946313344231266,
      "grad_norm": 1.5,
      "learning_rate": 0.0007148059540463828,
      "loss": 1.7876,
      "step": 20859
    },
    {
      "epoch": 1.7947173707304482,
      "grad_norm": 0.859375,
      "learning_rate": 0.0007147815355796281,
      "loss": 0.8619,
      "step": 20860
    },
    {
      "epoch": 1.79480340703777,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007147571164846793,
      "loss": 1.3919,
      "step": 20861
    },
    {
      "epoch": 1.7948894433450917,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007147326967616078,
      "loss": 1.3035,
      "step": 20862
    },
    {
      "epoch": 1.7949754796524133,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007147082764104847,
      "loss": 0.8454,
      "step": 20863
    },
    {
      "epoch": 1.795061515959735,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007146838554313818,
      "loss": 2.0332,
      "step": 20864
    },
    {
      "epoch": 1.7951475522670566,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007146594338243704,
      "loss": 0.9499,
      "step": 20865
    },
    {
      "epoch": 1.7952335885743784,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007146350115895218,
      "loss": 1.0057,
      "step": 20866
    },
    {
      "epoch": 1.7953196248817,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007146105887269076,
      "loss": 1.1692,
      "step": 20867
    },
    {
      "epoch": 1.795405661189022,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000714586165236599,
      "loss": 1.3659,
      "step": 20868
    },
    {
      "epoch": 1.7954916974963435,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007145617411186679,
      "loss": 0.9708,
      "step": 20869
    },
    {
      "epoch": 1.7955777338036651,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007145373163731853,
      "loss": 1.0671,
      "step": 20870
    },
    {
      "epoch": 1.7956637701109868,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007145128910002227,
      "loss": 0.7268,
      "step": 20871
    },
    {
      "epoch": 1.7957498064183084,
      "grad_norm": 0.828125,
      "learning_rate": 0.0007144884649998517,
      "loss": 1.7423,
      "step": 20872
    },
    {
      "epoch": 1.7958358427256302,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007144640383721436,
      "loss": 0.9311,
      "step": 20873
    },
    {
      "epoch": 1.7959218790329519,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007144396111171699,
      "loss": 1.3976,
      "step": 20874
    },
    {
      "epoch": 1.7960079153402737,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007144151832350021,
      "loss": 1.6848,
      "step": 20875
    },
    {
      "epoch": 1.7960939516475953,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007143907547257115,
      "loss": 0.8974,
      "step": 20876
    },
    {
      "epoch": 1.796179987954917,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007143663255893696,
      "loss": 1.5473,
      "step": 20877
    },
    {
      "epoch": 1.7962660242622386,
      "grad_norm": 0.953125,
      "learning_rate": 0.000714341895826048,
      "loss": 0.7151,
      "step": 20878
    },
    {
      "epoch": 1.7963520605695602,
      "grad_norm": 1.125,
      "learning_rate": 0.000714317465435818,
      "loss": 1.2958,
      "step": 20879
    },
    {
      "epoch": 1.796438096876882,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007142930344187511,
      "loss": 1.4767,
      "step": 20880
    },
    {
      "epoch": 1.796524133184204,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007142686027749188,
      "loss": 1.4434,
      "step": 20881
    },
    {
      "epoch": 1.7966101694915255,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007142441705043925,
      "loss": 1.4856,
      "step": 20882
    },
    {
      "epoch": 1.7966962057988471,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007142197376072437,
      "loss": 0.5324,
      "step": 20883
    },
    {
      "epoch": 1.7967822421061688,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007141953040835439,
      "loss": 1.0807,
      "step": 20884
    },
    {
      "epoch": 1.7968682784134904,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007141708699333643,
      "loss": 1.5052,
      "step": 20885
    },
    {
      "epoch": 1.796954314720812,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007141464351567765,
      "loss": 1.5474,
      "step": 20886
    },
    {
      "epoch": 1.7970403510281339,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007141219997538523,
      "loss": 1.5577,
      "step": 20887
    },
    {
      "epoch": 1.7971263873354557,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007140975637246627,
      "loss": 0.7342,
      "step": 20888
    },
    {
      "epoch": 1.7972124236427773,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007140731270692794,
      "loss": 2.1972,
      "step": 20889
    },
    {
      "epoch": 1.797298459950099,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0007140486897877739,
      "loss": 1.5156,
      "step": 20890
    },
    {
      "epoch": 1.7973844962574206,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007140242518802176,
      "loss": 1.0755,
      "step": 20891
    },
    {
      "epoch": 1.7974705325647422,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007139998133466821,
      "loss": 0.9802,
      "step": 20892
    },
    {
      "epoch": 1.797556568872064,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007139753741872384,
      "loss": 1.7313,
      "step": 20893
    },
    {
      "epoch": 1.7976426051793857,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007139509344019587,
      "loss": 1.1175,
      "step": 20894
    },
    {
      "epoch": 1.7977286414867075,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000713926493990914,
      "loss": 1.6901,
      "step": 20895
    },
    {
      "epoch": 1.7978146777940291,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007139020529541759,
      "loss": 0.9819,
      "step": 20896
    },
    {
      "epoch": 1.7979007141013508,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007138776112918158,
      "loss": 1.2496,
      "step": 20897
    },
    {
      "epoch": 1.7979867504086724,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007138531690039054,
      "loss": 1.2893,
      "step": 20898
    },
    {
      "epoch": 1.798072786715994,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007138287260905161,
      "loss": 1.37,
      "step": 20899
    },
    {
      "epoch": 1.7981588230233159,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007138042825517193,
      "loss": 1.4652,
      "step": 20900
    },
    {
      "epoch": 1.7982448593306375,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007137798383875867,
      "loss": 1.7032,
      "step": 20901
    },
    {
      "epoch": 1.7983308956379593,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007137553935981896,
      "loss": 1.0294,
      "step": 20902
    },
    {
      "epoch": 1.798416931945281,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007137309481835994,
      "loss": 0.6209,
      "step": 20903
    },
    {
      "epoch": 1.7985029682526026,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007137065021438878,
      "loss": 2.189,
      "step": 20904
    },
    {
      "epoch": 1.7985890045599242,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007136820554791263,
      "loss": 0.9846,
      "step": 20905
    },
    {
      "epoch": 1.7986750408672458,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007136576081893863,
      "loss": 1.5202,
      "step": 20906
    },
    {
      "epoch": 1.7987610771745677,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007136331602747394,
      "loss": 1.4892,
      "step": 20907
    },
    {
      "epoch": 1.7988471134818893,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000713608711735257,
      "loss": 1.5308,
      "step": 20908
    },
    {
      "epoch": 1.7989331497892111,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007135842625710107,
      "loss": 1.1201,
      "step": 20909
    },
    {
      "epoch": 1.7990191860965328,
      "grad_norm": 1.265625,
      "learning_rate": 0.000713559812782072,
      "loss": 1.1334,
      "step": 20910
    },
    {
      "epoch": 1.7991052224038544,
      "grad_norm": 1.5625,
      "learning_rate": 0.0007135353623685123,
      "loss": 1.6683,
      "step": 20911
    },
    {
      "epoch": 1.799191258711176,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007135109113304033,
      "loss": 1.1099,
      "step": 20912
    },
    {
      "epoch": 1.7992772950184976,
      "grad_norm": 1.25,
      "learning_rate": 0.0007134864596678162,
      "loss": 1.3045,
      "step": 20913
    },
    {
      "epoch": 1.7993633313258195,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007134620073808229,
      "loss": 0.897,
      "step": 20914
    },
    {
      "epoch": 1.7994493676331413,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007134375544694948,
      "loss": 1.5087,
      "step": 20915
    },
    {
      "epoch": 1.799535403940463,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007134131009339033,
      "loss": 1.5136,
      "step": 20916
    },
    {
      "epoch": 1.7996214402477846,
      "grad_norm": 0.875,
      "learning_rate": 0.00071338864677412,
      "loss": 1.016,
      "step": 20917
    },
    {
      "epoch": 1.7997074765551062,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007133641919902163,
      "loss": 1.2408,
      "step": 20918
    },
    {
      "epoch": 1.7997935128624278,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007133397365822639,
      "loss": 0.996,
      "step": 20919
    },
    {
      "epoch": 1.7998795491697497,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007133152805503342,
      "loss": 1.6211,
      "step": 20920
    },
    {
      "epoch": 1.7999655854770713,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007132908238944991,
      "loss": 1.5903,
      "step": 20921
    },
    {
      "epoch": 1.8000516217843932,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0007132663666148296,
      "loss": 1.185,
      "step": 20922
    },
    {
      "epoch": 1.8001376580917148,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007132419087113972,
      "loss": 1.6791,
      "step": 20923
    },
    {
      "epoch": 1.8002236943990364,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007132174501842741,
      "loss": 1.6719,
      "step": 20924
    },
    {
      "epoch": 1.800309730706358,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007131929910335313,
      "loss": 1.6455,
      "step": 20925
    },
    {
      "epoch": 1.8003957670136796,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007131685312592404,
      "loss": 1.6907,
      "step": 20926
    },
    {
      "epoch": 1.8004818033210015,
      "grad_norm": 1.53125,
      "learning_rate": 0.000713144070861473,
      "loss": 2.0448,
      "step": 20927
    },
    {
      "epoch": 1.8005678396283231,
      "grad_norm": 1.796875,
      "learning_rate": 0.0007131196098403008,
      "loss": 2.1213,
      "step": 20928
    },
    {
      "epoch": 1.800653875935645,
      "grad_norm": 1.25,
      "learning_rate": 0.000713095148195795,
      "loss": 1.5274,
      "step": 20929
    },
    {
      "epoch": 1.8007399122429666,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007130706859280274,
      "loss": 0.8203,
      "step": 20930
    },
    {
      "epoch": 1.8008259485502882,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007130462230370696,
      "loss": 0.8717,
      "step": 20931
    },
    {
      "epoch": 1.8009119848576098,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007130217595229928,
      "loss": 0.7368,
      "step": 20932
    },
    {
      "epoch": 1.8009980211649315,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007129972953858689,
      "loss": 1.6784,
      "step": 20933
    },
    {
      "epoch": 1.8010840574722533,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007129728306257692,
      "loss": 1.662,
      "step": 20934
    },
    {
      "epoch": 1.801170093779575,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007129483652427658,
      "loss": 0.9906,
      "step": 20935
    },
    {
      "epoch": 1.8012561300868968,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007129238992369295,
      "loss": 1.3948,
      "step": 20936
    },
    {
      "epoch": 1.8013421663942184,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007128994326083323,
      "loss": 1.914,
      "step": 20937
    },
    {
      "epoch": 1.80142820270154,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007128749653570457,
      "loss": 1.2618,
      "step": 20938
    },
    {
      "epoch": 1.8015142390088617,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007128504974831412,
      "loss": 1.4155,
      "step": 20939
    },
    {
      "epoch": 1.8016002753161833,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007128260289866904,
      "loss": 1.3124,
      "step": 20940
    },
    {
      "epoch": 1.8016863116235051,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007128015598677648,
      "loss": 1.1225,
      "step": 20941
    },
    {
      "epoch": 1.801772347930827,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000712777090126436,
      "loss": 1.2692,
      "step": 20942
    },
    {
      "epoch": 1.8018583842381486,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007127526197627757,
      "loss": 0.711,
      "step": 20943
    },
    {
      "epoch": 1.8019444205454702,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007127281487768553,
      "loss": 1.6075,
      "step": 20944
    },
    {
      "epoch": 1.8020304568527918,
      "grad_norm": 7.125,
      "learning_rate": 0.0007127036771687465,
      "loss": 1.7574,
      "step": 20945
    },
    {
      "epoch": 1.8021164931601135,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007126792049385208,
      "loss": 1.4741,
      "step": 20946
    },
    {
      "epoch": 1.8022025294674353,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007126547320862498,
      "loss": 0.7052,
      "step": 20947
    },
    {
      "epoch": 1.802288565774757,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007126302586120051,
      "loss": 0.7941,
      "step": 20948
    },
    {
      "epoch": 1.8023746020820788,
      "grad_norm": 1.09375,
      "learning_rate": 0.000712605784515858,
      "loss": 1.5951,
      "step": 20949
    },
    {
      "epoch": 1.8024606383894004,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007125813097978805,
      "loss": 1.1844,
      "step": 20950
    },
    {
      "epoch": 1.802546674696722,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000712556834458144,
      "loss": 0.7694,
      "step": 20951
    },
    {
      "epoch": 1.8026327110040437,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007125323584967201,
      "loss": 1.0052,
      "step": 20952
    },
    {
      "epoch": 1.8027187473113653,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007125078819136805,
      "loss": 1.7071,
      "step": 20953
    },
    {
      "epoch": 1.8028047836186871,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007124834047090964,
      "loss": 1.2891,
      "step": 20954
    },
    {
      "epoch": 1.8028908199260087,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007124589268830401,
      "loss": 1.7093,
      "step": 20955
    },
    {
      "epoch": 1.8029768562333306,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007124344484355824,
      "loss": 1.0519,
      "step": 20956
    },
    {
      "epoch": 1.8030628925406522,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007124099693667954,
      "loss": 1.2769,
      "step": 20957
    },
    {
      "epoch": 1.8031489288479738,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007123854896767506,
      "loss": 0.9568,
      "step": 20958
    },
    {
      "epoch": 1.8032349651552955,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007123610093655194,
      "loss": 1.4244,
      "step": 20959
    },
    {
      "epoch": 1.803321001462617,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007123365284331736,
      "loss": 0.8318,
      "step": 20960
    },
    {
      "epoch": 1.803407037769939,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0007123120468797848,
      "loss": 0.7104,
      "step": 20961
    },
    {
      "epoch": 1.8034930740772606,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007122875647054246,
      "loss": 0.9648,
      "step": 20962
    },
    {
      "epoch": 1.8035791103845824,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007122630819101644,
      "loss": 0.8898,
      "step": 20963
    },
    {
      "epoch": 1.803665146691904,
      "grad_norm": 1.09375,
      "learning_rate": 0.000712238598494076,
      "loss": 1.7204,
      "step": 20964
    },
    {
      "epoch": 1.8037511829992257,
      "grad_norm": 1.6875,
      "learning_rate": 0.0007122141144572311,
      "loss": 1.7991,
      "step": 20965
    },
    {
      "epoch": 1.8038372193065473,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000712189629799701,
      "loss": 1.7613,
      "step": 20966
    },
    {
      "epoch": 1.803923255613869,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007121651445215577,
      "loss": 1.1402,
      "step": 20967
    },
    {
      "epoch": 1.8040092919211907,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007121406586228725,
      "loss": 0.9594,
      "step": 20968
    },
    {
      "epoch": 1.8040953282285126,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0007121161721037171,
      "loss": 0.8927,
      "step": 20969
    },
    {
      "epoch": 1.8041813645358342,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007120916849641632,
      "loss": 1.0394,
      "step": 20970
    },
    {
      "epoch": 1.8042674008431558,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007120671972042825,
      "loss": 1.2257,
      "step": 20971
    },
    {
      "epoch": 1.8043534371504775,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007120427088241464,
      "loss": 1.0572,
      "step": 20972
    },
    {
      "epoch": 1.804439473457799,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007120182198238265,
      "loss": 0.7722,
      "step": 20973
    },
    {
      "epoch": 1.8045255097651207,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007119937302033948,
      "loss": 1.0778,
      "step": 20974
    },
    {
      "epoch": 1.8046115460724426,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007119692399629223,
      "loss": 0.6509,
      "step": 20975
    },
    {
      "epoch": 1.8046975823797644,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007119447491024813,
      "loss": 1.4154,
      "step": 20976
    },
    {
      "epoch": 1.804783618687086,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007119202576221431,
      "loss": 1.2358,
      "step": 20977
    },
    {
      "epoch": 1.8048696549944077,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007118957655219793,
      "loss": 1.3595,
      "step": 20978
    },
    {
      "epoch": 1.8049556913017293,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007118712728020615,
      "loss": 1.0906,
      "step": 20979
    },
    {
      "epoch": 1.805041727609051,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007118467794624616,
      "loss": 1.5541,
      "step": 20980
    },
    {
      "epoch": 1.8051277639163728,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000711822285503251,
      "loss": 1.7755,
      "step": 20981
    },
    {
      "epoch": 1.8052138002236944,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007117977909245014,
      "loss": 1.1994,
      "step": 20982
    },
    {
      "epoch": 1.8052998365310162,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007117732957262845,
      "loss": 1.0697,
      "step": 20983
    },
    {
      "epoch": 1.8053858728383378,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007117487999086719,
      "loss": 1.7016,
      "step": 20984
    },
    {
      "epoch": 1.8054719091456595,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007117243034717352,
      "loss": 0.8792,
      "step": 20985
    },
    {
      "epoch": 1.805557945452981,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000711699806415546,
      "loss": 0.7159,
      "step": 20986
    },
    {
      "epoch": 1.8056439817603027,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007116753087401762,
      "loss": 1.1951,
      "step": 20987
    },
    {
      "epoch": 1.8057300180676246,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007116508104456971,
      "loss": 1.6487,
      "step": 20988
    },
    {
      "epoch": 1.8058160543749462,
      "grad_norm": 1.640625,
      "learning_rate": 0.0007116263115321808,
      "loss": 1.5009,
      "step": 20989
    },
    {
      "epoch": 1.805902090682268,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0007116018119996986,
      "loss": 1.0552,
      "step": 20990
    },
    {
      "epoch": 1.8059881269895897,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007115773118483221,
      "loss": 0.9183,
      "step": 20991
    },
    {
      "epoch": 1.8060741632969113,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007115528110781233,
      "loss": 1.6441,
      "step": 20992
    },
    {
      "epoch": 1.806160199604233,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007115283096891737,
      "loss": 1.4262,
      "step": 20993
    },
    {
      "epoch": 1.8062462359115545,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007115038076815447,
      "loss": 1.7834,
      "step": 20994
    },
    {
      "epoch": 1.8063322722188764,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007114793050553085,
      "loss": 1.3496,
      "step": 20995
    },
    {
      "epoch": 1.806418308526198,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007114548018105362,
      "loss": 0.9878,
      "step": 20996
    },
    {
      "epoch": 1.8065043448335198,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007114302979472998,
      "loss": 1.3968,
      "step": 20997
    },
    {
      "epoch": 1.8065903811408415,
      "grad_norm": 1.046875,
      "learning_rate": 0.000711405793465671,
      "loss": 1.9787,
      "step": 20998
    },
    {
      "epoch": 1.806676417448163,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007113812883657214,
      "loss": 1.0951,
      "step": 20999
    },
    {
      "epoch": 1.8067624537554847,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007113567826475224,
      "loss": 1.5054,
      "step": 21000
    },
    {
      "epoch": 1.8068484900628063,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007113322763111461,
      "loss": 1.3242,
      "step": 21001
    },
    {
      "epoch": 1.8069345263701282,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007113077693566641,
      "loss": 1.1678,
      "step": 21002
    },
    {
      "epoch": 1.80702056267745,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007112832617841478,
      "loss": 2.1014,
      "step": 21003
    },
    {
      "epoch": 1.8071065989847717,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007112587535936691,
      "loss": 0.9421,
      "step": 21004
    },
    {
      "epoch": 1.8071926352920933,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007112342447852995,
      "loss": 0.7715,
      "step": 21005
    },
    {
      "epoch": 1.807278671599415,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007112097353591109,
      "loss": 0.7546,
      "step": 21006
    },
    {
      "epoch": 1.8073647079067365,
      "grad_norm": 1.421875,
      "learning_rate": 0.000711185225315175,
      "loss": 1.5616,
      "step": 21007
    },
    {
      "epoch": 1.8074507442140584,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007111607146535634,
      "loss": 1.6212,
      "step": 21008
    },
    {
      "epoch": 1.80753678052138,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007111362033743475,
      "loss": 1.4571,
      "step": 21009
    },
    {
      "epoch": 1.8076228168287019,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007111116914775996,
      "loss": 1.1277,
      "step": 21010
    },
    {
      "epoch": 1.8077088531360235,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007110871789633909,
      "loss": 1.1518,
      "step": 21011
    },
    {
      "epoch": 1.807794889443345,
      "grad_norm": 1.125,
      "learning_rate": 0.0007110626658317934,
      "loss": 1.2263,
      "step": 21012
    },
    {
      "epoch": 1.8078809257506667,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007110381520828785,
      "loss": 1.3603,
      "step": 21013
    },
    {
      "epoch": 1.8079669620579883,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000711013637716718,
      "loss": 2.1144,
      "step": 21014
    },
    {
      "epoch": 1.8080529983653102,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007109891227333838,
      "loss": 0.9779,
      "step": 21015
    },
    {
      "epoch": 1.8081390346726318,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007109646071329475,
      "loss": 1.1882,
      "step": 21016
    },
    {
      "epoch": 1.8082250709799537,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007109400909154805,
      "loss": 1.9982,
      "step": 21017
    },
    {
      "epoch": 1.8083111072872753,
      "grad_norm": 1.140625,
      "learning_rate": 0.000710915574081055,
      "loss": 1.2486,
      "step": 21018
    },
    {
      "epoch": 1.808397143594597,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007108910566297424,
      "loss": 1.7807,
      "step": 21019
    },
    {
      "epoch": 1.8084831799019185,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007108665385616145,
      "loss": 0.9658,
      "step": 21020
    },
    {
      "epoch": 1.8085692162092402,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007108420198767429,
      "loss": 1.3857,
      "step": 21021
    },
    {
      "epoch": 1.808655252516562,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007108175005751993,
      "loss": 0.9503,
      "step": 21022
    },
    {
      "epoch": 1.8087412888238836,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007107929806570557,
      "loss": 1.8178,
      "step": 21023
    },
    {
      "epoch": 1.8088273251312055,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007107684601223835,
      "loss": 1.188,
      "step": 21024
    },
    {
      "epoch": 1.808913361438527,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007107439389712547,
      "loss": 1.175,
      "step": 21025
    },
    {
      "epoch": 1.8089993977458487,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007107194172037408,
      "loss": 1.2555,
      "step": 21026
    },
    {
      "epoch": 1.8090854340531703,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007106948948199136,
      "loss": 0.9522,
      "step": 21027
    },
    {
      "epoch": 1.809171470360492,
      "grad_norm": 2.125,
      "learning_rate": 0.0007106703718198446,
      "loss": 0.7473,
      "step": 21028
    },
    {
      "epoch": 1.8092575066678138,
      "grad_norm": 1.0,
      "learning_rate": 0.000710645848203606,
      "loss": 1.4584,
      "step": 21029
    },
    {
      "epoch": 1.8093435429751357,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007106213239712692,
      "loss": 0.8998,
      "step": 21030
    },
    {
      "epoch": 1.8094295792824573,
      "grad_norm": 1.234375,
      "learning_rate": 0.000710596799122906,
      "loss": 1.7299,
      "step": 21031
    },
    {
      "epoch": 1.809515615589779,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000710572273658588,
      "loss": 1.1699,
      "step": 21032
    },
    {
      "epoch": 1.8096016518971005,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007105477475783873,
      "loss": 1.1132,
      "step": 21033
    },
    {
      "epoch": 1.8096876882044222,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007105232208823752,
      "loss": 1.0943,
      "step": 21034
    },
    {
      "epoch": 1.809773724511744,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007104986935706236,
      "loss": 0.9608,
      "step": 21035
    },
    {
      "epoch": 1.8098597608190656,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007104741656432043,
      "loss": 1.0158,
      "step": 21036
    },
    {
      "epoch": 1.8099457971263875,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007104496371001892,
      "loss": 1.9444,
      "step": 21037
    },
    {
      "epoch": 1.810031833433709,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007104251079416495,
      "loss": 1.3502,
      "step": 21038
    },
    {
      "epoch": 1.8101178697410307,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007104005781676574,
      "loss": 1.5946,
      "step": 21039
    },
    {
      "epoch": 1.8102039060483524,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007103760477782847,
      "loss": 1.3609,
      "step": 21040
    },
    {
      "epoch": 1.810289942355674,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007103515167736029,
      "loss": 1.1374,
      "step": 21041
    },
    {
      "epoch": 1.8103759786629958,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007103269851536836,
      "loss": 0.8559,
      "step": 21042
    },
    {
      "epoch": 1.8104620149703174,
      "grad_norm": 1.4375,
      "learning_rate": 0.000710302452918599,
      "loss": 0.8356,
      "step": 21043
    },
    {
      "epoch": 1.8105480512776393,
      "grad_norm": 1.25,
      "learning_rate": 0.0007102779200684206,
      "loss": 1.5035,
      "step": 21044
    },
    {
      "epoch": 1.810634087584961,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007102533866032201,
      "loss": 0.8382,
      "step": 21045
    },
    {
      "epoch": 1.8107201238922825,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007102288525230693,
      "loss": 1.426,
      "step": 21046
    },
    {
      "epoch": 1.8108061601996042,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007102043178280402,
      "loss": 0.7149,
      "step": 21047
    },
    {
      "epoch": 1.8108921965069258,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007101797825182043,
      "loss": 1.3962,
      "step": 21048
    },
    {
      "epoch": 1.8109782328142476,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007101552465936334,
      "loss": 1.2905,
      "step": 21049
    },
    {
      "epoch": 1.8110642691215693,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007101307100543992,
      "loss": 1.1806,
      "step": 21050
    },
    {
      "epoch": 1.811150305428891,
      "grad_norm": 1.25,
      "learning_rate": 0.0007101061729005736,
      "loss": 1.2056,
      "step": 21051
    },
    {
      "epoch": 1.8112363417362127,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007100816351322283,
      "loss": 1.4275,
      "step": 21052
    },
    {
      "epoch": 1.8113223780435344,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000710057096749435,
      "loss": 2.0479,
      "step": 21053
    },
    {
      "epoch": 1.811408414350856,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007100325577522656,
      "loss": 0.704,
      "step": 21054
    },
    {
      "epoch": 1.8114944506581776,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007100080181407919,
      "loss": 1.0172,
      "step": 21055
    },
    {
      "epoch": 1.8115804869654994,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007099834779150855,
      "loss": 1.3855,
      "step": 21056
    },
    {
      "epoch": 1.8116665232728213,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007099589370752182,
      "loss": 1.2875,
      "step": 21057
    },
    {
      "epoch": 1.811752559580143,
      "grad_norm": 1.375,
      "learning_rate": 0.0007099343956212619,
      "loss": 1.1802,
      "step": 21058
    },
    {
      "epoch": 1.8118385958874645,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007099098535532884,
      "loss": 0.9935,
      "step": 21059
    },
    {
      "epoch": 1.8119246321947862,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007098853108713694,
      "loss": 0.7605,
      "step": 21060
    },
    {
      "epoch": 1.8120106685021078,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007098607675755765,
      "loss": 1.73,
      "step": 21061
    },
    {
      "epoch": 1.8120967048094294,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007098362236659819,
      "loss": 1.6861,
      "step": 21062
    },
    {
      "epoch": 1.8121827411167513,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000709811679142657,
      "loss": 1.1016,
      "step": 21063
    },
    {
      "epoch": 1.812268777424073,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007097871340056739,
      "loss": 1.0893,
      "step": 21064
    },
    {
      "epoch": 1.8123548137313947,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000709762588255104,
      "loss": 0.7695,
      "step": 21065
    },
    {
      "epoch": 1.8124408500387164,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007097380418910195,
      "loss": 0.8259,
      "step": 21066
    },
    {
      "epoch": 1.812526886346038,
      "grad_norm": 1.109375,
      "learning_rate": 0.000709713494913492,
      "loss": 1.2275,
      "step": 21067
    },
    {
      "epoch": 1.8126129226533596,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007096889473225934,
      "loss": 1.5266,
      "step": 21068
    },
    {
      "epoch": 1.8126989589606815,
      "grad_norm": 1.125,
      "learning_rate": 0.0007096643991183952,
      "loss": 1.2617,
      "step": 21069
    },
    {
      "epoch": 1.812784995268003,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007096398503009696,
      "loss": 1.2116,
      "step": 21070
    },
    {
      "epoch": 1.812871031575325,
      "grad_norm": 1.375,
      "learning_rate": 0.0007096153008703882,
      "loss": 1.9526,
      "step": 21071
    },
    {
      "epoch": 1.8129570678826465,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007095907508267227,
      "loss": 1.3419,
      "step": 21072
    },
    {
      "epoch": 1.8130431041899682,
      "grad_norm": 1.359375,
      "learning_rate": 0.000709566200170045,
      "loss": 1.7229,
      "step": 21073
    },
    {
      "epoch": 1.8131291404972898,
      "grad_norm": 1.09375,
      "learning_rate": 0.000709541648900427,
      "loss": 2.0462,
      "step": 21074
    },
    {
      "epoch": 1.8132151768046114,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007095170970179405,
      "loss": 0.9588,
      "step": 21075
    },
    {
      "epoch": 1.8133012131119333,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000709492544522657,
      "loss": 1.4684,
      "step": 21076
    },
    {
      "epoch": 1.8133872494192549,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007094679914146488,
      "loss": 1.6564,
      "step": 21077
    },
    {
      "epoch": 1.8134732857265767,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0007094434376939873,
      "loss": 1.952,
      "step": 21078
    },
    {
      "epoch": 1.8135593220338984,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007094188833607445,
      "loss": 1.03,
      "step": 21079
    },
    {
      "epoch": 1.81364535834122,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007093943284149923,
      "loss": 1.4478,
      "step": 21080
    },
    {
      "epoch": 1.8137313946485416,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007093697728568024,
      "loss": 0.9757,
      "step": 21081
    },
    {
      "epoch": 1.8138174309558632,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007093452166862466,
      "loss": 0.6947,
      "step": 21082
    },
    {
      "epoch": 1.813903467263185,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007093206599033967,
      "loss": 0.9016,
      "step": 21083
    },
    {
      "epoch": 1.8139895035705067,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007092961025083246,
      "loss": 1.2246,
      "step": 21084
    },
    {
      "epoch": 1.8140755398778285,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007092715445011022,
      "loss": 0.8825,
      "step": 21085
    },
    {
      "epoch": 1.8141615761851502,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007092469858818012,
      "loss": 0.9507,
      "step": 21086
    },
    {
      "epoch": 1.8142476124924718,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007092224266504934,
      "loss": 0.9339,
      "step": 21087
    },
    {
      "epoch": 1.8143336487997934,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0007091978668072507,
      "loss": 1.5363,
      "step": 21088
    },
    {
      "epoch": 1.814419685107115,
      "grad_norm": 1.109375,
      "learning_rate": 0.000709173306352145,
      "loss": 1.3468,
      "step": 21089
    },
    {
      "epoch": 1.814505721414437,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007091487452852481,
      "loss": 1.7985,
      "step": 21090
    },
    {
      "epoch": 1.8145917577217587,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007091241836066315,
      "loss": 0.9353,
      "step": 21091
    },
    {
      "epoch": 1.8146777940290804,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007090996213163676,
      "loss": 1.277,
      "step": 21092
    },
    {
      "epoch": 1.814763830336402,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000709075058414528,
      "loss": 1.4066,
      "step": 21093
    },
    {
      "epoch": 1.8148498666437236,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007090504949011844,
      "loss": 1.2988,
      "step": 21094
    },
    {
      "epoch": 1.8149359029510452,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007090259307764089,
      "loss": 1.4618,
      "step": 21095
    },
    {
      "epoch": 1.815021939258367,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007090013660402731,
      "loss": 1.3308,
      "step": 21096
    },
    {
      "epoch": 1.8151079755656887,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000708976800692849,
      "loss": 1.4236,
      "step": 21097
    },
    {
      "epoch": 1.8151940118730105,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007089522347342082,
      "loss": 0.6962,
      "step": 21098
    },
    {
      "epoch": 1.8152800481803322,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007089276681644229,
      "loss": 0.8662,
      "step": 21099
    },
    {
      "epoch": 1.8153660844876538,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007089031009835649,
      "loss": 0.907,
      "step": 21100
    },
    {
      "epoch": 1.8154521207949754,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007088785331917058,
      "loss": 0.9742,
      "step": 21101
    },
    {
      "epoch": 1.815538157102297,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007088539647889178,
      "loss": 1.457,
      "step": 21102
    },
    {
      "epoch": 1.815624193409619,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007088293957752724,
      "loss": 1.2902,
      "step": 21103
    },
    {
      "epoch": 1.8157102297169405,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007088048261508417,
      "loss": 1.2689,
      "step": 21104
    },
    {
      "epoch": 1.8157962660242624,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007087802559156975,
      "loss": 1.0039,
      "step": 21105
    },
    {
      "epoch": 1.815882302331584,
      "grad_norm": 0.96875,
      "learning_rate": 0.0007087556850699116,
      "loss": 1.6064,
      "step": 21106
    },
    {
      "epoch": 1.8159683386389056,
      "grad_norm": 1.09375,
      "learning_rate": 0.000708731113613556,
      "loss": 0.8781,
      "step": 21107
    },
    {
      "epoch": 1.8160543749462272,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007087065415467025,
      "loss": 1.6627,
      "step": 21108
    },
    {
      "epoch": 1.8161404112535489,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007086819688694228,
      "loss": 1.9938,
      "step": 21109
    },
    {
      "epoch": 1.8162264475608707,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007086573955817891,
      "loss": 0.8239,
      "step": 21110
    },
    {
      "epoch": 1.8163124838681923,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007086328216838731,
      "loss": 0.807,
      "step": 21111
    },
    {
      "epoch": 1.8163985201755142,
      "grad_norm": 1.125,
      "learning_rate": 0.0007086082471757467,
      "loss": 1.4875,
      "step": 21112
    },
    {
      "epoch": 1.8164845564828358,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007085836720574815,
      "loss": 1.3119,
      "step": 21113
    },
    {
      "epoch": 1.8165705927901574,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007085590963291499,
      "loss": 1.7253,
      "step": 21114
    },
    {
      "epoch": 1.816656629097479,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007085345199908234,
      "loss": 1.1454,
      "step": 21115
    },
    {
      "epoch": 1.8167426654048007,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007085099430425739,
      "loss": 2.039,
      "step": 21116
    },
    {
      "epoch": 1.8168287017121225,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007084853654844734,
      "loss": 0.4842,
      "step": 21117
    },
    {
      "epoch": 1.8169147380194444,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007084607873165938,
      "loss": 1.3539,
      "step": 21118
    },
    {
      "epoch": 1.817000774326766,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000708436208539007,
      "loss": 0.8068,
      "step": 21119
    },
    {
      "epoch": 1.8170868106340876,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007084116291517849,
      "loss": 0.9738,
      "step": 21120
    },
    {
      "epoch": 1.8171728469414092,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007083870491549991,
      "loss": 0.8641,
      "step": 21121
    },
    {
      "epoch": 1.8172588832487309,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007083624685487218,
      "loss": 1.1082,
      "step": 21122
    },
    {
      "epoch": 1.8173449195560527,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007083378873330248,
      "loss": 0.9757,
      "step": 21123
    },
    {
      "epoch": 1.8174309558633743,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007083133055079802,
      "loss": 1.32,
      "step": 21124
    },
    {
      "epoch": 1.8175169921706962,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007082887230736593,
      "loss": 2.1212,
      "step": 21125
    },
    {
      "epoch": 1.8176030284780178,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007082641400301346,
      "loss": 1.7061,
      "step": 21126
    },
    {
      "epoch": 1.8176890647853394,
      "grad_norm": 1.359375,
      "learning_rate": 0.000708239556377478,
      "loss": 1.5786,
      "step": 21127
    },
    {
      "epoch": 1.817775101092661,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007082149721157608,
      "loss": 0.8402,
      "step": 21128
    },
    {
      "epoch": 1.8178611373999827,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007081903872450555,
      "loss": 1.1683,
      "step": 21129
    },
    {
      "epoch": 1.8179471737073045,
      "grad_norm": 1.484375,
      "learning_rate": 0.0007081658017654339,
      "loss": 0.6504,
      "step": 21130
    },
    {
      "epoch": 1.8180332100146261,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007081412156769677,
      "loss": 1.3289,
      "step": 21131
    },
    {
      "epoch": 1.818119246321948,
      "grad_norm": 1.375,
      "learning_rate": 0.0007081166289797289,
      "loss": 1.6838,
      "step": 21132
    },
    {
      "epoch": 1.8182052826292696,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007080920416737895,
      "loss": 0.9999,
      "step": 21133
    },
    {
      "epoch": 1.8182913189365912,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007080674537592213,
      "loss": 1.0386,
      "step": 21134
    },
    {
      "epoch": 1.8183773552439129,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007080428652360963,
      "loss": 1.4436,
      "step": 21135
    },
    {
      "epoch": 1.8184633915512345,
      "grad_norm": 0.921875,
      "learning_rate": 0.0007080182761044863,
      "loss": 1.323,
      "step": 21136
    },
    {
      "epoch": 1.8185494278585563,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0007079936863644634,
      "loss": 0.8402,
      "step": 21137
    },
    {
      "epoch": 1.818635464165878,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007079690960160994,
      "loss": 0.6466,
      "step": 21138
    },
    {
      "epoch": 1.8187215004731998,
      "grad_norm": 1.25,
      "learning_rate": 0.0007079445050594663,
      "loss": 1.4245,
      "step": 21139
    },
    {
      "epoch": 1.8188075367805214,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007079199134946359,
      "loss": 1.6471,
      "step": 21140
    },
    {
      "epoch": 1.818893573087843,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007078953213216801,
      "loss": 0.7419,
      "step": 21141
    },
    {
      "epoch": 1.8189796093951647,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007078707285406711,
      "loss": 0.8415,
      "step": 21142
    },
    {
      "epoch": 1.8190656457024863,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007078461351516805,
      "loss": 1.2166,
      "step": 21143
    },
    {
      "epoch": 1.8191516820098081,
      "grad_norm": 1.25,
      "learning_rate": 0.0007078215411547805,
      "loss": 1.9495,
      "step": 21144
    },
    {
      "epoch": 1.81923771831713,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007077969465500429,
      "loss": 0.949,
      "step": 21145
    },
    {
      "epoch": 1.8193237546244516,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007077723513375396,
      "loss": 1.3205,
      "step": 21146
    },
    {
      "epoch": 1.8194097909317732,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007077477555173425,
      "loss": 1.3325,
      "step": 21147
    },
    {
      "epoch": 1.8194958272390949,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007077231590895236,
      "loss": 0.9738,
      "step": 21148
    },
    {
      "epoch": 1.8195818635464165,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007076985620541552,
      "loss": 1.2304,
      "step": 21149
    },
    {
      "epoch": 1.8196678998537381,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0007076739644113085,
      "loss": 0.7374,
      "step": 21150
    },
    {
      "epoch": 1.81975393616106,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000707649366161056,
      "loss": 1.1046,
      "step": 21151
    },
    {
      "epoch": 1.8198399724683818,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007076247673034696,
      "loss": 1.4649,
      "step": 21152
    },
    {
      "epoch": 1.8199260087757034,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007076001678386209,
      "loss": 1.068,
      "step": 21153
    },
    {
      "epoch": 1.820012045083025,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007075755677665822,
      "loss": 0.777,
      "step": 21154
    },
    {
      "epoch": 1.8200980813903467,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007075509670874254,
      "loss": 1.8957,
      "step": 21155
    },
    {
      "epoch": 1.8201841176976683,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007075263658012223,
      "loss": 1.0082,
      "step": 21156
    },
    {
      "epoch": 1.8202701540049901,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000707501763908045,
      "loss": 1.247,
      "step": 21157
    },
    {
      "epoch": 1.8203561903123118,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007074771614079653,
      "loss": 1.6172,
      "step": 21158
    },
    {
      "epoch": 1.8204422266196336,
      "grad_norm": 0.828125,
      "learning_rate": 0.0007074525583010554,
      "loss": 0.7487,
      "step": 21159
    },
    {
      "epoch": 1.8205282629269552,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007074279545873871,
      "loss": 1.6658,
      "step": 21160
    },
    {
      "epoch": 1.8206142992342769,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007074033502670323,
      "loss": 1.0832,
      "step": 21161
    },
    {
      "epoch": 1.8207003355415985,
      "grad_norm": 1.453125,
      "learning_rate": 0.000707378745340063,
      "loss": 1.0266,
      "step": 21162
    },
    {
      "epoch": 1.8207863718489201,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007073541398065513,
      "loss": 0.8506,
      "step": 21163
    },
    {
      "epoch": 1.820872408156242,
      "grad_norm": 1.859375,
      "learning_rate": 0.0007073295336665691,
      "loss": 1.8851,
      "step": 21164
    },
    {
      "epoch": 1.8209584444635636,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0007073049269201882,
      "loss": 0.5725,
      "step": 21165
    },
    {
      "epoch": 1.8210444807708854,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007072803195674809,
      "loss": 1.294,
      "step": 21166
    },
    {
      "epoch": 1.821130517078207,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007072557116085188,
      "loss": 2.3497,
      "step": 21167
    },
    {
      "epoch": 1.8212165533855287,
      "grad_norm": 1.46875,
      "learning_rate": 0.0007072311030433741,
      "loss": 0.8824,
      "step": 21168
    },
    {
      "epoch": 1.8213025896928503,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007072064938721187,
      "loss": 1.5517,
      "step": 21169
    },
    {
      "epoch": 1.821388626000172,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007071818840948247,
      "loss": 2.166,
      "step": 21170
    },
    {
      "epoch": 1.8214746623074938,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007071572737115639,
      "loss": 1.698,
      "step": 21171
    },
    {
      "epoch": 1.8215606986148154,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007071326627224084,
      "loss": 1.0913,
      "step": 21172
    },
    {
      "epoch": 1.8216467349221372,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00070710805112743,
      "loss": 1.1935,
      "step": 21173
    },
    {
      "epoch": 1.8217327712294589,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000707083438926701,
      "loss": 1.3616,
      "step": 21174
    },
    {
      "epoch": 1.8218188075367805,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007070588261202933,
      "loss": 1.3274,
      "step": 21175
    },
    {
      "epoch": 1.8219048438441021,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007070342127082787,
      "loss": 0.644,
      "step": 21176
    },
    {
      "epoch": 1.8219908801514237,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007070095986907292,
      "loss": 1.2085,
      "step": 21177
    },
    {
      "epoch": 1.8220769164587456,
      "grad_norm": 1.078125,
      "learning_rate": 0.000706984984067717,
      "loss": 1.1794,
      "step": 21178
    },
    {
      "epoch": 1.8221629527660674,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007069603688393139,
      "loss": 1.1143,
      "step": 21179
    },
    {
      "epoch": 1.822248989073389,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007069357530055921,
      "loss": 0.9802,
      "step": 21180
    },
    {
      "epoch": 1.8223350253807107,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007069111365666233,
      "loss": 1.0822,
      "step": 21181
    },
    {
      "epoch": 1.8224210616880323,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007068865195224798,
      "loss": 1.0358,
      "step": 21182
    },
    {
      "epoch": 1.822507097995354,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007068619018732334,
      "loss": 1.8454,
      "step": 21183
    },
    {
      "epoch": 1.8225931343026758,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007068372836189562,
      "loss": 1.4987,
      "step": 21184
    },
    {
      "epoch": 1.8226791706099974,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007068126647597202,
      "loss": 0.916,
      "step": 21185
    },
    {
      "epoch": 1.8227652069173192,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007067880452955972,
      "loss": 1.5814,
      "step": 21186
    },
    {
      "epoch": 1.8228512432246409,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0007067634252266597,
      "loss": 0.9438,
      "step": 21187
    },
    {
      "epoch": 1.8229372795319625,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007067388045529791,
      "loss": 0.8117,
      "step": 21188
    },
    {
      "epoch": 1.8230233158392841,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007067141832746279,
      "loss": 0.8732,
      "step": 21189
    },
    {
      "epoch": 1.8231093521466057,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007066895613916778,
      "loss": 1.002,
      "step": 21190
    },
    {
      "epoch": 1.8231953884539276,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000706664938904201,
      "loss": 1.2312,
      "step": 21191
    },
    {
      "epoch": 1.8232814247612492,
      "grad_norm": 1.3125,
      "learning_rate": 0.0007066403158122694,
      "loss": 1.9374,
      "step": 21192
    },
    {
      "epoch": 1.823367461068571,
      "grad_norm": 1.296875,
      "learning_rate": 0.000706615692115955,
      "loss": 1.4603,
      "step": 21193
    },
    {
      "epoch": 1.8234534973758927,
      "grad_norm": 1.03125,
      "learning_rate": 0.00070659106781533,
      "loss": 0.8084,
      "step": 21194
    },
    {
      "epoch": 1.8235395336832143,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007065664429104664,
      "loss": 1.5939,
      "step": 21195
    },
    {
      "epoch": 1.823625569990536,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000706541817401436,
      "loss": 0.6954,
      "step": 21196
    },
    {
      "epoch": 1.8237116062978576,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000706517191288311,
      "loss": 1.3109,
      "step": 21197
    },
    {
      "epoch": 1.8237976426051794,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007064925645711633,
      "loss": 1.7448,
      "step": 21198
    },
    {
      "epoch": 1.823883678912501,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000706467937250065,
      "loss": 1.6578,
      "step": 21199
    },
    {
      "epoch": 1.8239697152198229,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007064433093250882,
      "loss": 1.4247,
      "step": 21200
    },
    {
      "epoch": 1.8240557515271445,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007064186807963049,
      "loss": 1.5183,
      "step": 21201
    },
    {
      "epoch": 1.8241417878344661,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007063940516637871,
      "loss": 0.8027,
      "step": 21202
    },
    {
      "epoch": 1.8242278241417877,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007063694219276068,
      "loss": 0.9042,
      "step": 21203
    },
    {
      "epoch": 1.8243138604491094,
      "grad_norm": 1.453125,
      "learning_rate": 0.000706344791587836,
      "loss": 1.8686,
      "step": 21204
    },
    {
      "epoch": 1.8243998967564312,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000706320160644547,
      "loss": 1.0583,
      "step": 21205
    },
    {
      "epoch": 1.824485933063753,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007062955290978115,
      "loss": 1.4435,
      "step": 21206
    },
    {
      "epoch": 1.8245719693710747,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007062708969477017,
      "loss": 1.2435,
      "step": 21207
    },
    {
      "epoch": 1.8246580056783963,
      "grad_norm": 1.375,
      "learning_rate": 0.0007062462641942898,
      "loss": 1.2858,
      "step": 21208
    },
    {
      "epoch": 1.824744041985718,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007062216308376476,
      "loss": 1.5337,
      "step": 21209
    },
    {
      "epoch": 1.8248300782930396,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007061969968778472,
      "loss": 1.869,
      "step": 21210
    },
    {
      "epoch": 1.8249161146003614,
      "grad_norm": 1.921875,
      "learning_rate": 0.0007061723623149606,
      "loss": 1.7493,
      "step": 21211
    },
    {
      "epoch": 1.825002150907683,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007061477271490602,
      "loss": 1.4845,
      "step": 21212
    },
    {
      "epoch": 1.8250881872150049,
      "grad_norm": 1.453125,
      "learning_rate": 0.0007061230913802176,
      "loss": 1.1083,
      "step": 21213
    },
    {
      "epoch": 1.8251742235223265,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000706098455008505,
      "loss": 1.1145,
      "step": 21214
    },
    {
      "epoch": 1.8252602598296481,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007060738180339947,
      "loss": 1.285,
      "step": 21215
    },
    {
      "epoch": 1.8253462961369697,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007060491804567585,
      "loss": 1.3848,
      "step": 21216
    },
    {
      "epoch": 1.8254323324442914,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007060245422768686,
      "loss": 0.5138,
      "step": 21217
    },
    {
      "epoch": 1.8255183687516132,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007059999034943967,
      "loss": 0.8196,
      "step": 21218
    },
    {
      "epoch": 1.8256044050589348,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007059752641094154,
      "loss": 0.9728,
      "step": 21219
    },
    {
      "epoch": 1.8256904413662567,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007059506241219964,
      "loss": 0.4646,
      "step": 21220
    },
    {
      "epoch": 1.8257764776735783,
      "grad_norm": 1.03125,
      "learning_rate": 0.000705925983532212,
      "loss": 1.8529,
      "step": 21221
    },
    {
      "epoch": 1.8258625139809,
      "grad_norm": 1.125,
      "learning_rate": 0.0007059013423401339,
      "loss": 1.0252,
      "step": 21222
    },
    {
      "epoch": 1.8259485502882216,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007058767005458346,
      "loss": 1.2126,
      "step": 21223
    },
    {
      "epoch": 1.8260345865955432,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000705852058149386,
      "loss": 0.8538,
      "step": 21224
    },
    {
      "epoch": 1.826120622902865,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007058274151508601,
      "loss": 1.3269,
      "step": 21225
    },
    {
      "epoch": 1.8262066592101867,
      "grad_norm": 1.125,
      "learning_rate": 0.0007058027715503291,
      "loss": 1.2852,
      "step": 21226
    },
    {
      "epoch": 1.8262926955175085,
      "grad_norm": 0.9375,
      "learning_rate": 0.000705778127347865,
      "loss": 0.8337,
      "step": 21227
    },
    {
      "epoch": 1.8263787318248301,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007057534825435399,
      "loss": 1.2463,
      "step": 21228
    },
    {
      "epoch": 1.8264647681321518,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007057288371374258,
      "loss": 1.6827,
      "step": 21229
    },
    {
      "epoch": 1.8265508044394734,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007057041911295949,
      "loss": 1.4498,
      "step": 21230
    },
    {
      "epoch": 1.826636840746795,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0007056795445201193,
      "loss": 1.2635,
      "step": 21231
    },
    {
      "epoch": 1.8267228770541168,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000705654897309071,
      "loss": 1.2664,
      "step": 21232
    },
    {
      "epoch": 1.8268089133614387,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007056302494965221,
      "loss": 1.1289,
      "step": 21233
    },
    {
      "epoch": 1.8268949496687603,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000705605601082545,
      "loss": 1.4718,
      "step": 21234
    },
    {
      "epoch": 1.826980985976082,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0007055809520672112,
      "loss": 1.2616,
      "step": 21235
    },
    {
      "epoch": 1.8270670222834036,
      "grad_norm": 1.5,
      "learning_rate": 0.0007055563024505931,
      "loss": 1.3555,
      "step": 21236
    },
    {
      "epoch": 1.8271530585907252,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007055316522327628,
      "loss": 1.5845,
      "step": 21237
    },
    {
      "epoch": 1.8272390948980468,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007055070014137926,
      "loss": 1.3359,
      "step": 21238
    },
    {
      "epoch": 1.8273251312053687,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007054823499937541,
      "loss": 1.3109,
      "step": 21239
    },
    {
      "epoch": 1.8274111675126905,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007054576979727197,
      "loss": 1.1444,
      "step": 21240
    },
    {
      "epoch": 1.8274972038200121,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007054330453507617,
      "loss": 0.7065,
      "step": 21241
    },
    {
      "epoch": 1.8275832401273338,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007054083921279519,
      "loss": 0.8671,
      "step": 21242
    },
    {
      "epoch": 1.8276692764346554,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007053837383043626,
      "loss": 1.6463,
      "step": 21243
    },
    {
      "epoch": 1.827755312741977,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007053590838800657,
      "loss": 1.3071,
      "step": 21244
    },
    {
      "epoch": 1.8278413490492988,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0007053344288551334,
      "loss": 0.8783,
      "step": 21245
    },
    {
      "epoch": 1.8279273853566205,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007053097732296379,
      "loss": 0.9449,
      "step": 21246
    },
    {
      "epoch": 1.8280134216639423,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007052851170036512,
      "loss": 0.8265,
      "step": 21247
    },
    {
      "epoch": 1.828099457971264,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007052604601772455,
      "loss": 0.9616,
      "step": 21248
    },
    {
      "epoch": 1.8281854942785856,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007052358027504928,
      "loss": 1.6471,
      "step": 21249
    },
    {
      "epoch": 1.8282715305859072,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007052111447234654,
      "loss": 1.1393,
      "step": 21250
    },
    {
      "epoch": 1.8283575668932288,
      "grad_norm": 2.046875,
      "learning_rate": 0.0007051864860962352,
      "loss": 1.4666,
      "step": 21251
    },
    {
      "epoch": 1.8284436032005507,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007051618268688745,
      "loss": 1.58,
      "step": 21252
    },
    {
      "epoch": 1.8285296395078723,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007051371670414553,
      "loss": 1.206,
      "step": 21253
    },
    {
      "epoch": 1.8286156758151941,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00070511250661405,
      "loss": 1.6465,
      "step": 21254
    },
    {
      "epoch": 1.8287017121225158,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007050878455867303,
      "loss": 0.976,
      "step": 21255
    },
    {
      "epoch": 1.8287877484298374,
      "grad_norm": 1.53125,
      "learning_rate": 0.0007050631839595686,
      "loss": 1.401,
      "step": 21256
    },
    {
      "epoch": 1.828873784737159,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000705038521732637,
      "loss": 1.8985,
      "step": 21257
    },
    {
      "epoch": 1.8289598210444806,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007050138589060075,
      "loss": 1.3068,
      "step": 21258
    },
    {
      "epoch": 1.8290458573518025,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007049891954797523,
      "loss": 1.0407,
      "step": 21259
    },
    {
      "epoch": 1.829131893659124,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0007049645314539437,
      "loss": 1.2773,
      "step": 21260
    },
    {
      "epoch": 1.829217929966446,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007049398668286537,
      "loss": 1.2773,
      "step": 21261
    },
    {
      "epoch": 1.8293039662737676,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007049152016039544,
      "loss": 1.4479,
      "step": 21262
    },
    {
      "epoch": 1.8293900025810892,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000704890535779918,
      "loss": 1.073,
      "step": 21263
    },
    {
      "epoch": 1.8294760388884108,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0007048658693566166,
      "loss": 1.0549,
      "step": 21264
    },
    {
      "epoch": 1.8295620751957324,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007048412023341224,
      "loss": 0.9351,
      "step": 21265
    },
    {
      "epoch": 1.8296481115030543,
      "grad_norm": 1.125,
      "learning_rate": 0.0007048165347125076,
      "loss": 1.9237,
      "step": 21266
    },
    {
      "epoch": 1.8297341478103761,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007047918664918443,
      "loss": 1.1803,
      "step": 21267
    },
    {
      "epoch": 1.8298201841176978,
      "grad_norm": 1.875,
      "learning_rate": 0.0007047671976722045,
      "loss": 1.8974,
      "step": 21268
    },
    {
      "epoch": 1.8299062204250194,
      "grad_norm": 2.203125,
      "learning_rate": 0.0007047425282536605,
      "loss": 1.4622,
      "step": 21269
    },
    {
      "epoch": 1.829992256732341,
      "grad_norm": 1.671875,
      "learning_rate": 0.0007047178582362843,
      "loss": 1.5245,
      "step": 21270
    },
    {
      "epoch": 1.8300782930396626,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007046931876201483,
      "loss": 0.8692,
      "step": 21271
    },
    {
      "epoch": 1.8301643293469845,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007046685164053246,
      "loss": 1.6109,
      "step": 21272
    },
    {
      "epoch": 1.830250365654306,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007046438445918854,
      "loss": 0.8208,
      "step": 21273
    },
    {
      "epoch": 1.830336401961628,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007046191721799025,
      "loss": 1.2685,
      "step": 21274
    },
    {
      "epoch": 1.8304224382689496,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007045944991694484,
      "loss": 1.8487,
      "step": 21275
    },
    {
      "epoch": 1.8305084745762712,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0007045698255605952,
      "loss": 0.8617,
      "step": 21276
    },
    {
      "epoch": 1.8305945108835928,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007045451513534152,
      "loss": 1.2424,
      "step": 21277
    },
    {
      "epoch": 1.8306805471909144,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00070452047654798,
      "loss": 0.7644,
      "step": 21278
    },
    {
      "epoch": 1.8307665834982363,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007044958011443625,
      "loss": 1.8018,
      "step": 21279
    },
    {
      "epoch": 1.830852619805558,
      "grad_norm": 1.125,
      "learning_rate": 0.0007044711251426346,
      "loss": 1.4853,
      "step": 21280
    },
    {
      "epoch": 1.8309386561128798,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007044464485428682,
      "loss": 1.1845,
      "step": 21281
    },
    {
      "epoch": 1.8310246924202014,
      "grad_norm": 0.9375,
      "learning_rate": 0.0007044217713451359,
      "loss": 0.8412,
      "step": 21282
    },
    {
      "epoch": 1.831110728727523,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007043970935495096,
      "loss": 1.7868,
      "step": 21283
    },
    {
      "epoch": 1.8311967650348446,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007043724151560615,
      "loss": 1.8853,
      "step": 21284
    },
    {
      "epoch": 1.8312828013421663,
      "grad_norm": 1.140625,
      "learning_rate": 0.0007043477361648638,
      "loss": 0.9635,
      "step": 21285
    },
    {
      "epoch": 1.831368837649488,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007043230565759889,
      "loss": 1.1083,
      "step": 21286
    },
    {
      "epoch": 1.8314548739568097,
      "grad_norm": 1.375,
      "learning_rate": 0.0007042983763895087,
      "loss": 1.1586,
      "step": 21287
    },
    {
      "epoch": 1.8315409102641316,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007042736956054955,
      "loss": 1.2103,
      "step": 21288
    },
    {
      "epoch": 1.8316269465714532,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0007042490142240214,
      "loss": 0.9189,
      "step": 21289
    },
    {
      "epoch": 1.8317129828787748,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007042243322451587,
      "loss": 1.1683,
      "step": 21290
    },
    {
      "epoch": 1.8317990191860964,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007041996496689795,
      "loss": 0.6183,
      "step": 21291
    },
    {
      "epoch": 1.831885055493418,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007041749664955562,
      "loss": 0.8751,
      "step": 21292
    },
    {
      "epoch": 1.83197109180074,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007041502827249606,
      "loss": 0.8754,
      "step": 21293
    },
    {
      "epoch": 1.8320571281080618,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007041255983572654,
      "loss": 1.457,
      "step": 21294
    },
    {
      "epoch": 1.8321431644153834,
      "grad_norm": 1.59375,
      "learning_rate": 0.0007041009133925423,
      "loss": 1.5418,
      "step": 21295
    },
    {
      "epoch": 1.832229200722705,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007040762278308638,
      "loss": 1.3971,
      "step": 21296
    },
    {
      "epoch": 1.8323152370300266,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007040515416723021,
      "loss": 1.8577,
      "step": 21297
    },
    {
      "epoch": 1.8324012733373483,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0007040268549169293,
      "loss": 1.1586,
      "step": 21298
    },
    {
      "epoch": 1.83248730964467,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007040021675648176,
      "loss": 1.2114,
      "step": 21299
    },
    {
      "epoch": 1.8325733459519917,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007039774796160391,
      "loss": 0.957,
      "step": 21300
    },
    {
      "epoch": 1.8326593822593136,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007039527910706662,
      "loss": 1.2646,
      "step": 21301
    },
    {
      "epoch": 1.8327454185666352,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0007039281019287711,
      "loss": 0.7503,
      "step": 21302
    },
    {
      "epoch": 1.8328314548739568,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007039034121904261,
      "loss": 1.0181,
      "step": 21303
    },
    {
      "epoch": 1.8329174911812784,
      "grad_norm": 1.078125,
      "learning_rate": 0.000703878721855703,
      "loss": 1.0766,
      "step": 21304
    },
    {
      "epoch": 1.8330035274886,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0007038540309246746,
      "loss": 1.43,
      "step": 21305
    },
    {
      "epoch": 1.833089563795922,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007038293393974126,
      "loss": 1.3163,
      "step": 21306
    },
    {
      "epoch": 1.8331756001032435,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007038046472739895,
      "loss": 0.996,
      "step": 21307
    },
    {
      "epoch": 1.8332616364105654,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007037799545544774,
      "loss": 1.7379,
      "step": 21308
    },
    {
      "epoch": 1.833347672717887,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007037552612389486,
      "loss": 0.928,
      "step": 21309
    },
    {
      "epoch": 1.8334337090252086,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007037305673274753,
      "loss": 1.9283,
      "step": 21310
    },
    {
      "epoch": 1.8335197453325303,
      "grad_norm": 1.609375,
      "learning_rate": 0.0007037058728201296,
      "loss": 1.0649,
      "step": 21311
    },
    {
      "epoch": 1.8336057816398519,
      "grad_norm": 1.40625,
      "learning_rate": 0.0007036811777169841,
      "loss": 2.0598,
      "step": 21312
    },
    {
      "epoch": 1.8336918179471737,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007036564820181106,
      "loss": 1.1681,
      "step": 21313
    },
    {
      "epoch": 1.8337778542544954,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007036317857235815,
      "loss": 1.0226,
      "step": 21314
    },
    {
      "epoch": 1.8338638905618172,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007036070888334689,
      "loss": 1.5095,
      "step": 21315
    },
    {
      "epoch": 1.8339499268691388,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007035823913478454,
      "loss": 0.7094,
      "step": 21316
    },
    {
      "epoch": 1.8340359631764604,
      "grad_norm": 1.421875,
      "learning_rate": 0.0007035576932667829,
      "loss": 1.795,
      "step": 21317
    },
    {
      "epoch": 1.834121999483782,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007035329945903537,
      "loss": 0.8259,
      "step": 21318
    },
    {
      "epoch": 1.8342080357911037,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007035082953186302,
      "loss": 1.6002,
      "step": 21319
    },
    {
      "epoch": 1.8342940720984255,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007034835954516842,
      "loss": 1.0626,
      "step": 21320
    },
    {
      "epoch": 1.8343801084057474,
      "grad_norm": 0.84375,
      "learning_rate": 0.0007034588949895886,
      "loss": 1.0959,
      "step": 21321
    },
    {
      "epoch": 1.834466144713069,
      "grad_norm": 1.015625,
      "learning_rate": 0.0007034341939324152,
      "loss": 1.2204,
      "step": 21322
    },
    {
      "epoch": 1.8345521810203906,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007034094922802363,
      "loss": 1.8282,
      "step": 21323
    },
    {
      "epoch": 1.8346382173277123,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007033847900331242,
      "loss": 2.0816,
      "step": 21324
    },
    {
      "epoch": 1.8347242536350339,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007033600871911513,
      "loss": 0.7987,
      "step": 21325
    },
    {
      "epoch": 1.8348102899423555,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007033353837543894,
      "loss": 1.1344,
      "step": 21326
    },
    {
      "epoch": 1.8348963262496774,
      "grad_norm": 0.984375,
      "learning_rate": 0.0007033106797229113,
      "loss": 1.7159,
      "step": 21327
    },
    {
      "epoch": 1.8349823625569992,
      "grad_norm": 1.09375,
      "learning_rate": 0.000703285975096789,
      "loss": 1.3339,
      "step": 21328
    },
    {
      "epoch": 1.8350683988643208,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0007032612698760947,
      "loss": 1.6086,
      "step": 21329
    },
    {
      "epoch": 1.8351544351716425,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007032365640609006,
      "loss": 1.2651,
      "step": 21330
    },
    {
      "epoch": 1.835240471478964,
      "grad_norm": 0.8125,
      "learning_rate": 0.0007032118576512792,
      "loss": 0.6766,
      "step": 21331
    },
    {
      "epoch": 1.8353265077862857,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007031871506473027,
      "loss": 1.5175,
      "step": 21332
    },
    {
      "epoch": 1.8354125440936075,
      "grad_norm": 1.625,
      "learning_rate": 0.0007031624430490433,
      "loss": 2.0228,
      "step": 21333
    },
    {
      "epoch": 1.8354985804009292,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007031377348565731,
      "loss": 0.6241,
      "step": 21334
    },
    {
      "epoch": 1.835584616708251,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007031130260699647,
      "loss": 1.5278,
      "step": 21335
    },
    {
      "epoch": 1.8356706530155726,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0007030883166892901,
      "loss": 1.186,
      "step": 21336
    },
    {
      "epoch": 1.8357566893228943,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007030636067146217,
      "loss": 2.1898,
      "step": 21337
    },
    {
      "epoch": 1.8358427256302159,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007030388961460316,
      "loss": 1.0567,
      "step": 21338
    },
    {
      "epoch": 1.8359287619375375,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007030141849835924,
      "loss": 1.4123,
      "step": 21339
    },
    {
      "epoch": 1.8360147982448594,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0007029894732273761,
      "loss": 0.808,
      "step": 21340
    },
    {
      "epoch": 1.836100834552181,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007029647608774552,
      "loss": 0.8903,
      "step": 21341
    },
    {
      "epoch": 1.8361868708595028,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007029400479339017,
      "loss": 1.1862,
      "step": 21342
    },
    {
      "epoch": 1.8362729071668245,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007029153343967883,
      "loss": 1.5177,
      "step": 21343
    },
    {
      "epoch": 1.836358943474146,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007028906202661867,
      "loss": 1.4108,
      "step": 21344
    },
    {
      "epoch": 1.8364449797814677,
      "grad_norm": 1.078125,
      "learning_rate": 0.0007028659055421696,
      "loss": 1.0371,
      "step": 21345
    },
    {
      "epoch": 1.8365310160887893,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007028411902248092,
      "loss": 1.1659,
      "step": 21346
    },
    {
      "epoch": 1.8366170523961112,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007028164743141779,
      "loss": 1.0213,
      "step": 21347
    },
    {
      "epoch": 1.8367030887034328,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007027917578103477,
      "loss": 1.1162,
      "step": 21348
    },
    {
      "epoch": 1.8367891250107546,
      "grad_norm": 1.0,
      "learning_rate": 0.000702767040713391,
      "loss": 0.7297,
      "step": 21349
    },
    {
      "epoch": 1.8368751613180763,
      "grad_norm": 1.0,
      "learning_rate": 0.0007027423230233802,
      "loss": 0.8121,
      "step": 21350
    },
    {
      "epoch": 1.836961197625398,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007027176047403876,
      "loss": 1.0902,
      "step": 21351
    },
    {
      "epoch": 1.8370472339327195,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007026928858644852,
      "loss": 1.2017,
      "step": 21352
    },
    {
      "epoch": 1.8371332702400411,
      "grad_norm": 1.796875,
      "learning_rate": 0.0007026681663957459,
      "loss": 1.4918,
      "step": 21353
    },
    {
      "epoch": 1.837219306547363,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007026434463342414,
      "loss": 0.9636,
      "step": 21354
    },
    {
      "epoch": 1.8373053428546848,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007026187256800443,
      "loss": 1.4561,
      "step": 21355
    },
    {
      "epoch": 1.8373913791620065,
      "grad_norm": 1.125,
      "learning_rate": 0.0007025940044332266,
      "loss": 1.1165,
      "step": 21356
    },
    {
      "epoch": 1.837477415469328,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007025692825938611,
      "loss": 1.0827,
      "step": 21357
    },
    {
      "epoch": 1.8375634517766497,
      "grad_norm": 0.796875,
      "learning_rate": 0.0007025445601620198,
      "loss": 0.777,
      "step": 21358
    },
    {
      "epoch": 1.8376494880839713,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000702519837137775,
      "loss": 0.8453,
      "step": 21359
    },
    {
      "epoch": 1.8377355243912932,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007024951135211991,
      "loss": 0.9128,
      "step": 21360
    },
    {
      "epoch": 1.8378215606986148,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0007024703893123644,
      "loss": 1.2145,
      "step": 21361
    },
    {
      "epoch": 1.8379075970059366,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007024456645113433,
      "loss": 1.0213,
      "step": 21362
    },
    {
      "epoch": 1.8379936333132583,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007024209391182078,
      "loss": 0.9926,
      "step": 21363
    },
    {
      "epoch": 1.83807966962058,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007023962131330305,
      "loss": 1.1288,
      "step": 21364
    },
    {
      "epoch": 1.8381657059279015,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007023714865558837,
      "loss": 2.1272,
      "step": 21365
    },
    {
      "epoch": 1.8382517422352231,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007023467593868396,
      "loss": 1.5293,
      "step": 21366
    },
    {
      "epoch": 1.838337778542545,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007023220316259705,
      "loss": 1.874,
      "step": 21367
    },
    {
      "epoch": 1.8384238148498666,
      "grad_norm": 2.078125,
      "learning_rate": 0.000702297303273349,
      "loss": 1.3926,
      "step": 21368
    },
    {
      "epoch": 1.8385098511571885,
      "grad_norm": 0.921875,
      "learning_rate": 0.000702272574329047,
      "loss": 0.9188,
      "step": 21369
    },
    {
      "epoch": 1.83859588746451,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007022478447931373,
      "loss": 1.3852,
      "step": 21370
    },
    {
      "epoch": 1.8386819237718317,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007022231146656917,
      "loss": 1.1204,
      "step": 21371
    },
    {
      "epoch": 1.8387679600791533,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0007021983839467831,
      "loss": 1.878,
      "step": 21372
    },
    {
      "epoch": 1.838853996386475,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0007021736526364834,
      "loss": 1.5292,
      "step": 21373
    },
    {
      "epoch": 1.8389400326937968,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007021489207348652,
      "loss": 1.1563,
      "step": 21374
    },
    {
      "epoch": 1.8390260690011184,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007021241882420005,
      "loss": 1.5596,
      "step": 21375
    },
    {
      "epoch": 1.8391121053084403,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0007020994551579621,
      "loss": 1.2716,
      "step": 21376
    },
    {
      "epoch": 1.839198141615762,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0007020747214828221,
      "loss": 1.3386,
      "step": 21377
    },
    {
      "epoch": 1.8392841779230835,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007020499872166527,
      "loss": 1.2437,
      "step": 21378
    },
    {
      "epoch": 1.8393702142304051,
      "grad_norm": 1.125,
      "learning_rate": 0.0007020252523595264,
      "loss": 0.8005,
      "step": 21379
    },
    {
      "epoch": 1.8394562505377268,
      "grad_norm": 1.359375,
      "learning_rate": 0.0007020005169115156,
      "loss": 1.4817,
      "step": 21380
    },
    {
      "epoch": 1.8395422868450486,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0007019757808726925,
      "loss": 2.3746,
      "step": 21381
    },
    {
      "epoch": 1.8396283231523705,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007019510442431296,
      "loss": 0.8973,
      "step": 21382
    },
    {
      "epoch": 1.839714359459692,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007019263070228991,
      "loss": 0.6959,
      "step": 21383
    },
    {
      "epoch": 1.8398003957670137,
      "grad_norm": 1.5,
      "learning_rate": 0.0007019015692120736,
      "loss": 1.1056,
      "step": 21384
    },
    {
      "epoch": 1.8398864320743353,
      "grad_norm": 1.34375,
      "learning_rate": 0.000701876830810725,
      "loss": 1.1014,
      "step": 21385
    },
    {
      "epoch": 1.839972468381657,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007018520918189261,
      "loss": 1.1105,
      "step": 21386
    },
    {
      "epoch": 1.8400585046889788,
      "grad_norm": 1.125,
      "learning_rate": 0.0007018273522367491,
      "loss": 1.1011,
      "step": 21387
    },
    {
      "epoch": 1.8401445409963004,
      "grad_norm": 1.125,
      "learning_rate": 0.0007018026120642663,
      "loss": 1.1563,
      "step": 21388
    },
    {
      "epoch": 1.8402305773036223,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007017778713015502,
      "loss": 0.9254,
      "step": 21389
    },
    {
      "epoch": 1.840316613610944,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007017531299486729,
      "loss": 2.1063,
      "step": 21390
    },
    {
      "epoch": 1.8404026499182655,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007017283880057071,
      "loss": 1.8886,
      "step": 21391
    },
    {
      "epoch": 1.8404886862255871,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0007017036454727249,
      "loss": 1.1734,
      "step": 21392
    },
    {
      "epoch": 1.8405747225329088,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007016789023497987,
      "loss": 1.3938,
      "step": 21393
    },
    {
      "epoch": 1.8406607588402306,
      "grad_norm": 1.046875,
      "learning_rate": 0.000701654158637001,
      "loss": 1.1485,
      "step": 21394
    },
    {
      "epoch": 1.8407467951475522,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007016294143344042,
      "loss": 0.9993,
      "step": 21395
    },
    {
      "epoch": 1.840832831454874,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0007016046694420805,
      "loss": 0.8093,
      "step": 21396
    },
    {
      "epoch": 1.8409188677621957,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0007015799239601024,
      "loss": 0.8919,
      "step": 21397
    },
    {
      "epoch": 1.8410049040695173,
      "grad_norm": 1.09375,
      "learning_rate": 0.0007015551778885421,
      "loss": 0.9117,
      "step": 21398
    },
    {
      "epoch": 1.841090940376839,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007015304312274721,
      "loss": 1.4381,
      "step": 21399
    },
    {
      "epoch": 1.8411769766841606,
      "grad_norm": 1.046875,
      "learning_rate": 0.000701505683976965,
      "loss": 1.0003,
      "step": 21400
    },
    {
      "epoch": 1.8412630129914824,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0007014809361370929,
      "loss": 1.1755,
      "step": 21401
    },
    {
      "epoch": 1.841349049298804,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0007014561877079282,
      "loss": 1.3462,
      "step": 21402
    },
    {
      "epoch": 1.841435085606126,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007014314386895433,
      "loss": 1.3561,
      "step": 21403
    },
    {
      "epoch": 1.8415211219134475,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007014066890820106,
      "loss": 1.6308,
      "step": 21404
    },
    {
      "epoch": 1.8416071582207691,
      "grad_norm": 0.875,
      "learning_rate": 0.0007013819388854025,
      "loss": 1.13,
      "step": 21405
    },
    {
      "epoch": 1.8416931945280908,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0007013571880997915,
      "loss": 1.1918,
      "step": 21406
    },
    {
      "epoch": 1.8417792308354124,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0007013324367252499,
      "loss": 0.7542,
      "step": 21407
    },
    {
      "epoch": 1.8418652671427342,
      "grad_norm": 1.140625,
      "learning_rate": 0.00070130768476185,
      "loss": 1.5465,
      "step": 21408
    },
    {
      "epoch": 1.841951303450056,
      "grad_norm": 1.15625,
      "learning_rate": 0.0007012829322096642,
      "loss": 1.2847,
      "step": 21409
    },
    {
      "epoch": 1.8420373397573777,
      "grad_norm": 2.265625,
      "learning_rate": 0.0007012581790687651,
      "loss": 2.459,
      "step": 21410
    },
    {
      "epoch": 1.8421233760646993,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000701233425339225,
      "loss": 1.3017,
      "step": 21411
    },
    {
      "epoch": 1.842209412372021,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007012086710211161,
      "loss": 1.3293,
      "step": 21412
    },
    {
      "epoch": 1.8422954486793426,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0007011839161145109,
      "loss": 0.7888,
      "step": 21413
    },
    {
      "epoch": 1.8423814849866642,
      "grad_norm": 0.890625,
      "learning_rate": 0.0007011591606194822,
      "loss": 0.7597,
      "step": 21414
    },
    {
      "epoch": 1.842467521293986,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007011344045361018,
      "loss": 2.1567,
      "step": 21415
    },
    {
      "epoch": 1.842553557601308,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007011096478644425,
      "loss": 0.9068,
      "step": 21416
    },
    {
      "epoch": 1.8426395939086295,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007010848906045766,
      "loss": 1.7384,
      "step": 21417
    },
    {
      "epoch": 1.8427256302159511,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007010601327565766,
      "loss": 1.3418,
      "step": 21418
    },
    {
      "epoch": 1.8428116665232728,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007010353743205147,
      "loss": 1.5363,
      "step": 21419
    },
    {
      "epoch": 1.8428977028305944,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0007010106152964634,
      "loss": 1.6417,
      "step": 21420
    },
    {
      "epoch": 1.8429837391379162,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0007009858556844952,
      "loss": 1.8454,
      "step": 21421
    },
    {
      "epoch": 1.8430697754452379,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007009610954846825,
      "loss": 1.2501,
      "step": 21422
    },
    {
      "epoch": 1.8431558117525597,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007009363346970975,
      "loss": 1.2181,
      "step": 21423
    },
    {
      "epoch": 1.8432418480598813,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000700911573321813,
      "loss": 0.8781,
      "step": 21424
    },
    {
      "epoch": 1.843327884367203,
      "grad_norm": 1.046875,
      "learning_rate": 0.0007008868113589011,
      "loss": 0.8559,
      "step": 21425
    },
    {
      "epoch": 1.8434139206745246,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007008620488084345,
      "loss": 1.0523,
      "step": 21426
    },
    {
      "epoch": 1.8434999569818462,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0007008372856704853,
      "loss": 1.0435,
      "step": 21427
    },
    {
      "epoch": 1.843585993289168,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0007008125219451261,
      "loss": 1.13,
      "step": 21428
    },
    {
      "epoch": 1.8436720295964897,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007007877576324295,
      "loss": 1.6152,
      "step": 21429
    },
    {
      "epoch": 1.8437580659038115,
      "grad_norm": 1.0625,
      "learning_rate": 0.0007007629927324676,
      "loss": 1.9133,
      "step": 21430
    },
    {
      "epoch": 1.8438441022111332,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000700738227245313,
      "loss": 1.3536,
      "step": 21431
    },
    {
      "epoch": 1.8439301385184548,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0007007134611710381,
      "loss": 0.9325,
      "step": 21432
    },
    {
      "epoch": 1.8440161748257764,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0007006886945097155,
      "loss": 1.2532,
      "step": 21433
    },
    {
      "epoch": 1.844102211133098,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0007006639272614175,
      "loss": 1.5959,
      "step": 21434
    },
    {
      "epoch": 1.8441882474404199,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007006391594262164,
      "loss": 0.7061,
      "step": 21435
    },
    {
      "epoch": 1.8442742837477415,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007006143910041849,
      "loss": 0.6576,
      "step": 21436
    },
    {
      "epoch": 1.8443603200550633,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007005896219953953,
      "loss": 1.741,
      "step": 21437
    },
    {
      "epoch": 1.844446356362385,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0007005648523999199,
      "loss": 0.9473,
      "step": 21438
    },
    {
      "epoch": 1.8445323926697066,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007005400822178316,
      "loss": 1.4579,
      "step": 21439
    },
    {
      "epoch": 1.8446184289770282,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007005153114492024,
      "loss": 0.9839,
      "step": 21440
    },
    {
      "epoch": 1.8447044652843498,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0007004905400941051,
      "loss": 1.0632,
      "step": 21441
    },
    {
      "epoch": 1.8447905015916717,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007004657681526117,
      "loss": 1.3602,
      "step": 21442
    },
    {
      "epoch": 1.8448765378989935,
      "grad_norm": 1.0625,
      "learning_rate": 0.000700440995624795,
      "loss": 2.085,
      "step": 21443
    },
    {
      "epoch": 1.8449625742063152,
      "grad_norm": 1.171875,
      "learning_rate": 0.0007004162225107274,
      "loss": 1.2271,
      "step": 21444
    },
    {
      "epoch": 1.8450486105136368,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0007003914488104815,
      "loss": 1.6396,
      "step": 21445
    },
    {
      "epoch": 1.8451346468209584,
      "grad_norm": 3.3125,
      "learning_rate": 0.0007003666745241293,
      "loss": 0.8243,
      "step": 21446
    },
    {
      "epoch": 1.84522068312828,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007003418996517437,
      "loss": 1.5976,
      "step": 21447
    },
    {
      "epoch": 1.8453067194356019,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000700317124193397,
      "loss": 1.0897,
      "step": 21448
    },
    {
      "epoch": 1.8453927557429235,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0007002923481491616,
      "loss": 1.4112,
      "step": 21449
    },
    {
      "epoch": 1.8454787920502453,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0007002675715191099,
      "loss": 1.0298,
      "step": 21450
    },
    {
      "epoch": 1.845564828357567,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007002427943033148,
      "loss": 2.3031,
      "step": 21451
    },
    {
      "epoch": 1.8456508646648886,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007002180165018483,
      "loss": 1.4472,
      "step": 21452
    },
    {
      "epoch": 1.8457369009722102,
      "grad_norm": 2.078125,
      "learning_rate": 0.0007001932381147831,
      "loss": 1.3834,
      "step": 21453
    },
    {
      "epoch": 1.8458229372795318,
      "grad_norm": 1.125,
      "learning_rate": 0.0007001684591421915,
      "loss": 1.0461,
      "step": 21454
    },
    {
      "epoch": 1.8459089735868537,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007001436795841463,
      "loss": 0.8721,
      "step": 21455
    },
    {
      "epoch": 1.8459950098941753,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007001188994407196,
      "loss": 0.7991,
      "step": 21456
    },
    {
      "epoch": 1.8460810462014972,
      "grad_norm": 1.390625,
      "learning_rate": 0.000700094118711984,
      "loss": 1.5033,
      "step": 21457
    },
    {
      "epoch": 1.8461670825088188,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0007000693373980123,
      "loss": 0.9727,
      "step": 21458
    },
    {
      "epoch": 1.8462531188161404,
      "grad_norm": 1.578125,
      "learning_rate": 0.0007000445554988764,
      "loss": 1.0559,
      "step": 21459
    },
    {
      "epoch": 1.846339155123462,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007000197730146492,
      "loss": 1.4763,
      "step": 21460
    },
    {
      "epoch": 1.8464251914307837,
      "grad_norm": 1.28125,
      "learning_rate": 0.000699994989945403,
      "loss": 1.2364,
      "step": 21461
    },
    {
      "epoch": 1.8465112277381055,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006999702062912105,
      "loss": 1.8468,
      "step": 21462
    },
    {
      "epoch": 1.8465972640454271,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000699945422052144,
      "loss": 1.2958,
      "step": 21463
    },
    {
      "epoch": 1.846683300352749,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006999206372282758,
      "loss": 1.4105,
      "step": 21464
    },
    {
      "epoch": 1.8467693366600706,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000699895851819679,
      "loss": 0.6607,
      "step": 21465
    },
    {
      "epoch": 1.8468553729673922,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006998710658264255,
      "loss": 1.5913,
      "step": 21466
    },
    {
      "epoch": 1.8469414092747138,
      "grad_norm": 1.25,
      "learning_rate": 0.0006998462792485879,
      "loss": 1.341,
      "step": 21467
    },
    {
      "epoch": 1.8470274455820355,
      "grad_norm": 1.125,
      "learning_rate": 0.0006998214920862389,
      "loss": 1.0348,
      "step": 21468
    },
    {
      "epoch": 1.8471134818893573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006997967043394508,
      "loss": 1.6274,
      "step": 21469
    },
    {
      "epoch": 1.8471995181966792,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006997719160082964,
      "loss": 1.8087,
      "step": 21470
    },
    {
      "epoch": 1.8472855545040008,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000699747127092848,
      "loss": 1.5198,
      "step": 21471
    },
    {
      "epoch": 1.8473715908113224,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006997223375931779,
      "loss": 0.9903,
      "step": 21472
    },
    {
      "epoch": 1.847457627118644,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006996975475093589,
      "loss": 0.6334,
      "step": 21473
    },
    {
      "epoch": 1.8475436634259657,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0006996727568414634,
      "loss": 1.5435,
      "step": 21474
    },
    {
      "epoch": 1.8476296997332875,
      "grad_norm": 1.296875,
      "learning_rate": 0.000699647965589564,
      "loss": 1.4621,
      "step": 21475
    },
    {
      "epoch": 1.8477157360406091,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006996231737537331,
      "loss": 1.478,
      "step": 21476
    },
    {
      "epoch": 1.847801772347931,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006995983813340432,
      "loss": 1.382,
      "step": 21477
    },
    {
      "epoch": 1.8478878086552526,
      "grad_norm": 1.9296875,
      "learning_rate": 0.000699573588330567,
      "loss": 1.8473,
      "step": 21478
    },
    {
      "epoch": 1.8479738449625742,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006995487947433766,
      "loss": 1.6093,
      "step": 21479
    },
    {
      "epoch": 1.8480598812698958,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006995240005725449,
      "loss": 1.4305,
      "step": 21480
    },
    {
      "epoch": 1.8481459175772175,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006994992058181444,
      "loss": 1.7747,
      "step": 21481
    },
    {
      "epoch": 1.8482319538845393,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006994744104802475,
      "loss": 1.2595,
      "step": 21482
    },
    {
      "epoch": 1.848317990191861,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006994496145589266,
      "loss": 1.1334,
      "step": 21483
    },
    {
      "epoch": 1.8484040264991828,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006994248180542545,
      "loss": 1.4257,
      "step": 21484
    },
    {
      "epoch": 1.8484900628065044,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006994000209663036,
      "loss": 1.7664,
      "step": 21485
    },
    {
      "epoch": 1.848576099113826,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006993752232951464,
      "loss": 0.9334,
      "step": 21486
    },
    {
      "epoch": 1.8486621354211477,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006993504250408553,
      "loss": 2.0334,
      "step": 21487
    },
    {
      "epoch": 1.8487481717284693,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006993256262035031,
      "loss": 1.5526,
      "step": 21488
    },
    {
      "epoch": 1.8488342080357911,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006993008267831624,
      "loss": 1.4546,
      "step": 21489
    },
    {
      "epoch": 1.8489202443431128,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006992760267799052,
      "loss": 1.3667,
      "step": 21490
    },
    {
      "epoch": 1.8490062806504346,
      "grad_norm": 1.0,
      "learning_rate": 0.0006992512261938046,
      "loss": 0.914,
      "step": 21491
    },
    {
      "epoch": 1.8490923169577562,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006992264250249328,
      "loss": 1.0562,
      "step": 21492
    },
    {
      "epoch": 1.8491783532650778,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0006992016232733626,
      "loss": 0.5287,
      "step": 21493
    },
    {
      "epoch": 1.8492643895723995,
      "grad_norm": 1.703125,
      "learning_rate": 0.0006991768209391663,
      "loss": 2.2304,
      "step": 21494
    },
    {
      "epoch": 1.849350425879721,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006991520180224164,
      "loss": 1.0862,
      "step": 21495
    },
    {
      "epoch": 1.849436462187043,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006991272145231858,
      "loss": 0.7989,
      "step": 21496
    },
    {
      "epoch": 1.8495224984943648,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006991024104415468,
      "loss": 1.2888,
      "step": 21497
    },
    {
      "epoch": 1.8496085348016864,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006990776057775719,
      "loss": 1.141,
      "step": 21498
    },
    {
      "epoch": 1.849694571109008,
      "grad_norm": 1.625,
      "learning_rate": 0.0006990528005313337,
      "loss": 1.0311,
      "step": 21499
    },
    {
      "epoch": 1.8497806074163297,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006990279947029049,
      "loss": 0.7132,
      "step": 21500
    },
    {
      "epoch": 1.8498666437236513,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006990031882923578,
      "loss": 1.4181,
      "step": 21501
    },
    {
      "epoch": 1.849952680030973,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006989783812997651,
      "loss": 0.699,
      "step": 21502
    },
    {
      "epoch": 1.8500387163382948,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006989535737251993,
      "loss": 1.3569,
      "step": 21503
    },
    {
      "epoch": 1.8501247526456166,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006989287655687331,
      "loss": 2.2463,
      "step": 21504
    },
    {
      "epoch": 1.8502107889529382,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006989039568304388,
      "loss": 1.1355,
      "step": 21505
    },
    {
      "epoch": 1.8502968252602598,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006988791475103891,
      "loss": 1.3584,
      "step": 21506
    },
    {
      "epoch": 1.8503828615675815,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006988543376086568,
      "loss": 1.4491,
      "step": 21507
    },
    {
      "epoch": 1.850468897874903,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006988295271253141,
      "loss": 1.6298,
      "step": 21508
    },
    {
      "epoch": 1.850554934182225,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006988047160604338,
      "loss": 0.829,
      "step": 21509
    },
    {
      "epoch": 1.8506409704895466,
      "grad_norm": 1.625,
      "learning_rate": 0.0006987799044140883,
      "loss": 1.251,
      "step": 21510
    },
    {
      "epoch": 1.8507270067968684,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006987550921863501,
      "loss": 1.1541,
      "step": 21511
    },
    {
      "epoch": 1.85081304310419,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006987302793772922,
      "loss": 1.2816,
      "step": 21512
    },
    {
      "epoch": 1.8508990794115117,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006987054659869866,
      "loss": 1.0045,
      "step": 21513
    },
    {
      "epoch": 1.8509851157188333,
      "grad_norm": 1.6875,
      "learning_rate": 0.0006986806520155063,
      "loss": 1.9752,
      "step": 21514
    },
    {
      "epoch": 1.851071152026155,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006986558374629237,
      "loss": 1.4362,
      "step": 21515
    },
    {
      "epoch": 1.8511571883334768,
      "grad_norm": 1.5,
      "learning_rate": 0.0006986310223293115,
      "loss": 1.6459,
      "step": 21516
    },
    {
      "epoch": 1.8512432246407984,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006986062066147421,
      "loss": 1.149,
      "step": 21517
    },
    {
      "epoch": 1.8513292609481202,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006985813903192882,
      "loss": 1.2353,
      "step": 21518
    },
    {
      "epoch": 1.8514152972554418,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006985565734430224,
      "loss": 2.0584,
      "step": 21519
    },
    {
      "epoch": 1.8515013335627635,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000698531755986017,
      "loss": 1.617,
      "step": 21520
    },
    {
      "epoch": 1.851587369870085,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006985069379483451,
      "loss": 1.7559,
      "step": 21521
    },
    {
      "epoch": 1.8516734061774067,
      "grad_norm": 2.03125,
      "learning_rate": 0.000698482119330079,
      "loss": 1.4972,
      "step": 21522
    },
    {
      "epoch": 1.8517594424847286,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000698457300131291,
      "loss": 1.8552,
      "step": 21523
    },
    {
      "epoch": 1.8518454787920502,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006984324803520543,
      "loss": 1.0615,
      "step": 21524
    },
    {
      "epoch": 1.851931515099372,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006984076599924412,
      "loss": 0.7357,
      "step": 21525
    },
    {
      "epoch": 1.8520175514066937,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006983828390525242,
      "loss": 1.7844,
      "step": 21526
    },
    {
      "epoch": 1.8521035877140153,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006983580175323759,
      "loss": 0.9536,
      "step": 21527
    },
    {
      "epoch": 1.852189624021337,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006983331954320691,
      "loss": 0.9218,
      "step": 21528
    },
    {
      "epoch": 1.8522756603286585,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006983083727516761,
      "loss": 1.5658,
      "step": 21529
    },
    {
      "epoch": 1.8523616966359804,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006982835494912699,
      "loss": 1.2473,
      "step": 21530
    },
    {
      "epoch": 1.8524477329433022,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006982587256509228,
      "loss": 2.1178,
      "step": 21531
    },
    {
      "epoch": 1.8525337692506239,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006982339012307074,
      "loss": 0.9624,
      "step": 21532
    },
    {
      "epoch": 1.8526198055579455,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006982090762306965,
      "loss": 1.1474,
      "step": 21533
    },
    {
      "epoch": 1.852705841865267,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006981842506509626,
      "loss": 0.9683,
      "step": 21534
    },
    {
      "epoch": 1.8527918781725887,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006981594244915782,
      "loss": 1.1327,
      "step": 21535
    },
    {
      "epoch": 1.8528779144799106,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006981345977526162,
      "loss": 1.3618,
      "step": 21536
    },
    {
      "epoch": 1.8529639507872322,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006981097704341489,
      "loss": 0.6957,
      "step": 21537
    },
    {
      "epoch": 1.853049987094554,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006980849425362492,
      "loss": 1.1225,
      "step": 21538
    },
    {
      "epoch": 1.8531360234018757,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006980601140589894,
      "loss": 1.3164,
      "step": 21539
    },
    {
      "epoch": 1.8532220597091973,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006980352850024423,
      "loss": 0.8378,
      "step": 21540
    },
    {
      "epoch": 1.853308096016519,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0006980104553666806,
      "loss": 2.1491,
      "step": 21541
    },
    {
      "epoch": 1.8533941323238405,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006979856251517767,
      "loss": 1.3486,
      "step": 21542
    },
    {
      "epoch": 1.8534801686311624,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0006979607943578035,
      "loss": 1.2839,
      "step": 21543
    },
    {
      "epoch": 1.853566204938484,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006979359629848334,
      "loss": 1.3592,
      "step": 21544
    },
    {
      "epoch": 1.8536522412458059,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006979111310329391,
      "loss": 1.3526,
      "step": 21545
    },
    {
      "epoch": 1.8537382775531275,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006978862985021933,
      "loss": 0.8503,
      "step": 21546
    },
    {
      "epoch": 1.853824313860449,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006978614653926684,
      "loss": 1.5014,
      "step": 21547
    },
    {
      "epoch": 1.8539103501677707,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006978366317044373,
      "loss": 0.775,
      "step": 21548
    },
    {
      "epoch": 1.8539963864750924,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006978117974375726,
      "loss": 1.9954,
      "step": 21549
    },
    {
      "epoch": 1.8540824227824142,
      "grad_norm": 1.125,
      "learning_rate": 0.0006977869625921468,
      "loss": 1.2281,
      "step": 21550
    },
    {
      "epoch": 1.8541684590897358,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006977621271682326,
      "loss": 1.5023,
      "step": 21551
    },
    {
      "epoch": 1.8542544953970577,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006977372911659027,
      "loss": 1.393,
      "step": 21552
    },
    {
      "epoch": 1.8543405317043793,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006977124545852297,
      "loss": 1.1834,
      "step": 21553
    },
    {
      "epoch": 1.854426568011701,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006976876174262861,
      "loss": 0.9823,
      "step": 21554
    },
    {
      "epoch": 1.8545126043190225,
      "grad_norm": 1.5,
      "learning_rate": 0.0006976627796891447,
      "loss": 2.1394,
      "step": 21555
    },
    {
      "epoch": 1.8545986406263442,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006976379413738781,
      "loss": 1.0921,
      "step": 21556
    },
    {
      "epoch": 1.854684676933666,
      "grad_norm": 1.515625,
      "learning_rate": 0.0006976131024805591,
      "loss": 1.2614,
      "step": 21557
    },
    {
      "epoch": 1.8547707132409879,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006975882630092601,
      "loss": 1.3068,
      "step": 21558
    },
    {
      "epoch": 1.8548567495483095,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006975634229600538,
      "loss": 1.3323,
      "step": 21559
    },
    {
      "epoch": 1.854942785855631,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006975385823330129,
      "loss": 1.1077,
      "step": 21560
    },
    {
      "epoch": 1.8550288221629527,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006975137411282102,
      "loss": 2.0938,
      "step": 21561
    },
    {
      "epoch": 1.8551148584702744,
      "grad_norm": 1.0,
      "learning_rate": 0.000697488899345718,
      "loss": 0.9587,
      "step": 21562
    },
    {
      "epoch": 1.8552008947775962,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006974640569856094,
      "loss": 1.0313,
      "step": 21563
    },
    {
      "epoch": 1.8552869310849178,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006974392140479569,
      "loss": 1.2811,
      "step": 21564
    },
    {
      "epoch": 1.8553729673922397,
      "grad_norm": 1.125,
      "learning_rate": 0.000697414370532833,
      "loss": 0.9324,
      "step": 21565
    },
    {
      "epoch": 1.8554590036995613,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006973895264403103,
      "loss": 0.9501,
      "step": 21566
    },
    {
      "epoch": 1.855545040006883,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006973646817704617,
      "loss": 1.7252,
      "step": 21567
    },
    {
      "epoch": 1.8556310763142045,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00069733983652336,
      "loss": 1.3819,
      "step": 21568
    },
    {
      "epoch": 1.8557171126215262,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006973149906990774,
      "loss": 1.5801,
      "step": 21569
    },
    {
      "epoch": 1.855803148928848,
      "grad_norm": 1.1875,
      "learning_rate": 0.000697290144297687,
      "loss": 1.087,
      "step": 21570
    },
    {
      "epoch": 1.8558891852361696,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006972652973192612,
      "loss": 1.2457,
      "step": 21571
    },
    {
      "epoch": 1.8559752215434915,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006972404497638729,
      "loss": 1.3773,
      "step": 21572
    },
    {
      "epoch": 1.856061257850813,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006972156016315944,
      "loss": 1.4166,
      "step": 21573
    },
    {
      "epoch": 1.8561472941581347,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006971907529224989,
      "loss": 1.1173,
      "step": 21574
    },
    {
      "epoch": 1.8562333304654564,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006971659036366587,
      "loss": 0.7267,
      "step": 21575
    },
    {
      "epoch": 1.856319366772778,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006971410537741465,
      "loss": 1.5024,
      "step": 21576
    },
    {
      "epoch": 1.8564054030800998,
      "grad_norm": 1.25,
      "learning_rate": 0.0006971162033350351,
      "loss": 2.0159,
      "step": 21577
    },
    {
      "epoch": 1.8564914393874214,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006970913523193971,
      "loss": 0.9884,
      "step": 21578
    },
    {
      "epoch": 1.8565774756947433,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006970665007273054,
      "loss": 1.518,
      "step": 21579
    },
    {
      "epoch": 1.856663512002065,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006970416485588323,
      "loss": 1.0569,
      "step": 21580
    },
    {
      "epoch": 1.8567495483093865,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006970167958140508,
      "loss": 0.9361,
      "step": 21581
    },
    {
      "epoch": 1.8568355846167082,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0006969919424930336,
      "loss": 1.0327,
      "step": 21582
    },
    {
      "epoch": 1.8569216209240298,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006969670885958531,
      "loss": 1.3512,
      "step": 21583
    },
    {
      "epoch": 1.8570076572313516,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006969422341225822,
      "loss": 1.0092,
      "step": 21584
    },
    {
      "epoch": 1.8570936935386735,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006969173790732937,
      "loss": 0.9957,
      "step": 21585
    },
    {
      "epoch": 1.857179729845995,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00069689252344806,
      "loss": 1.7058,
      "step": 21586
    },
    {
      "epoch": 1.8572657661533167,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006968676672469541,
      "loss": 1.1602,
      "step": 21587
    },
    {
      "epoch": 1.8573518024606384,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006968428104700484,
      "loss": 0.8986,
      "step": 21588
    },
    {
      "epoch": 1.85743783876796,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006968179531174159,
      "loss": 1.8521,
      "step": 21589
    },
    {
      "epoch": 1.8575238750752816,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006967930951891292,
      "loss": 1.1403,
      "step": 21590
    },
    {
      "epoch": 1.8576099113826035,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006967682366852608,
      "loss": 0.9936,
      "step": 21591
    },
    {
      "epoch": 1.8576959476899253,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006967433776058837,
      "loss": 0.9975,
      "step": 21592
    },
    {
      "epoch": 1.857781983997247,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006967185179510704,
      "loss": 0.791,
      "step": 21593
    },
    {
      "epoch": 1.8578680203045685,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006966936577208939,
      "loss": 1.8291,
      "step": 21594
    },
    {
      "epoch": 1.8579540566118902,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006966687969154264,
      "loss": 1.0388,
      "step": 21595
    },
    {
      "epoch": 1.8580400929192118,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006966439355347412,
      "loss": 1.5637,
      "step": 21596
    },
    {
      "epoch": 1.8581261292265336,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006966190735789105,
      "loss": 1.6197,
      "step": 21597
    },
    {
      "epoch": 1.8582121655338553,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006965942110480074,
      "loss": 0.9453,
      "step": 21598
    },
    {
      "epoch": 1.858298201841177,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006965693479421043,
      "loss": 0.7091,
      "step": 21599
    },
    {
      "epoch": 1.8583842381484987,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006965444842612742,
      "loss": 1.0938,
      "step": 21600
    },
    {
      "epoch": 1.8584702744558204,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006965196200055899,
      "loss": 0.8272,
      "step": 21601
    },
    {
      "epoch": 1.858556310763142,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006964947551751237,
      "loss": 2.1004,
      "step": 21602
    },
    {
      "epoch": 1.8586423470704636,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006964698897699486,
      "loss": 1.5227,
      "step": 21603
    },
    {
      "epoch": 1.8587283833777855,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006964450237901372,
      "loss": 1.9435,
      "step": 21604
    },
    {
      "epoch": 1.858814419685107,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006964201572357626,
      "loss": 1.557,
      "step": 21605
    },
    {
      "epoch": 1.858900455992429,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006963952901068971,
      "loss": 1.2148,
      "step": 21606
    },
    {
      "epoch": 1.8589864922997505,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006963704224036134,
      "loss": 1.0689,
      "step": 21607
    },
    {
      "epoch": 1.8590725286070722,
      "grad_norm": 1.0,
      "learning_rate": 0.0006963455541259847,
      "loss": 1.9481,
      "step": 21608
    },
    {
      "epoch": 1.8591585649143938,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006963206852740832,
      "loss": 1.2991,
      "step": 21609
    },
    {
      "epoch": 1.8592446012217154,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006962958158479818,
      "loss": 1.2578,
      "step": 21610
    },
    {
      "epoch": 1.8593306375290373,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006962709458477537,
      "loss": 1.8892,
      "step": 21611
    },
    {
      "epoch": 1.859416673836359,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006962460752734709,
      "loss": 1.2199,
      "step": 21612
    },
    {
      "epoch": 1.8595027101436807,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006962212041252068,
      "loss": 1.2068,
      "step": 21613
    },
    {
      "epoch": 1.8595887464510024,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006961963324030334,
      "loss": 1.0561,
      "step": 21614
    },
    {
      "epoch": 1.859674782758324,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006961714601070242,
      "loss": 1.3754,
      "step": 21615
    },
    {
      "epoch": 1.8597608190656456,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006961465872372516,
      "loss": 0.8016,
      "step": 21616
    },
    {
      "epoch": 1.8598468553729672,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006961217137937885,
      "loss": 1.2262,
      "step": 21617
    },
    {
      "epoch": 1.859932891680289,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006960968397767073,
      "loss": 0.8649,
      "step": 21618
    },
    {
      "epoch": 1.860018927987611,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006960719651860809,
      "loss": 1.0366,
      "step": 21619
    },
    {
      "epoch": 1.8601049642949326,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006960470900219824,
      "loss": 1.2625,
      "step": 21620
    },
    {
      "epoch": 1.8601910006022542,
      "grad_norm": 1.8125,
      "learning_rate": 0.0006960222142844843,
      "loss": 1.4101,
      "step": 21621
    },
    {
      "epoch": 1.8602770369095758,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006959973379736592,
      "loss": 0.907,
      "step": 21622
    },
    {
      "epoch": 1.8603630732168974,
      "grad_norm": 1.4375,
      "learning_rate": 0.00069597246108958,
      "loss": 1.586,
      "step": 21623
    },
    {
      "epoch": 1.8604491095242193,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006959475836323196,
      "loss": 1.3067,
      "step": 21624
    },
    {
      "epoch": 1.860535145831541,
      "grad_norm": 1.0,
      "learning_rate": 0.0006959227056019504,
      "loss": 1.0797,
      "step": 21625
    },
    {
      "epoch": 1.8606211821388627,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006958978269985456,
      "loss": 0.8395,
      "step": 21626
    },
    {
      "epoch": 1.8607072184461844,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006958729478221778,
      "loss": 0.7973,
      "step": 21627
    },
    {
      "epoch": 1.860793254753506,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006958480680729196,
      "loss": 0.9075,
      "step": 21628
    },
    {
      "epoch": 1.8608792910608276,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006958231877508439,
      "loss": 0.9669,
      "step": 21629
    },
    {
      "epoch": 1.8609653273681492,
      "grad_norm": 1.0,
      "learning_rate": 0.0006957983068560234,
      "loss": 1.6148,
      "step": 21630
    },
    {
      "epoch": 1.861051363675471,
      "grad_norm": 1.0,
      "learning_rate": 0.000695773425388531,
      "loss": 1.5771,
      "step": 21631
    },
    {
      "epoch": 1.8611373999827927,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006957485433484394,
      "loss": 1.0291,
      "step": 21632
    },
    {
      "epoch": 1.8612234362901146,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006957236607358215,
      "loss": 1.5536,
      "step": 21633
    },
    {
      "epoch": 1.8613094725974362,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006956987775507497,
      "loss": 0.6644,
      "step": 21634
    },
    {
      "epoch": 1.8613955089047578,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006956738937932972,
      "loss": 1.4098,
      "step": 21635
    },
    {
      "epoch": 1.8614815452120794,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006956490094635367,
      "loss": 1.5734,
      "step": 21636
    },
    {
      "epoch": 1.861567581519401,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006956241245615408,
      "loss": 1.6662,
      "step": 21637
    },
    {
      "epoch": 1.861653617826723,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006955992390873825,
      "loss": 1.3262,
      "step": 21638
    },
    {
      "epoch": 1.8617396541340445,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006955743530411342,
      "loss": 1.6221,
      "step": 21639
    },
    {
      "epoch": 1.8618256904413664,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006955494664228691,
      "loss": 0.7721,
      "step": 21640
    },
    {
      "epoch": 1.861911726748688,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006955245792326599,
      "loss": 0.8523,
      "step": 21641
    },
    {
      "epoch": 1.8619977630560096,
      "grad_norm": 1.96875,
      "learning_rate": 0.0006954996914705794,
      "loss": 1.2057,
      "step": 21642
    },
    {
      "epoch": 1.8620837993633312,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006954748031367001,
      "loss": 1.7468,
      "step": 21643
    },
    {
      "epoch": 1.8621698356706529,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000695449914231095,
      "loss": 1.5063,
      "step": 21644
    },
    {
      "epoch": 1.8622558719779747,
      "grad_norm": 11.4375,
      "learning_rate": 0.0006954250247538372,
      "loss": 1.1655,
      "step": 21645
    },
    {
      "epoch": 1.8623419082852966,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006954001347049991,
      "loss": 0.7568,
      "step": 21646
    },
    {
      "epoch": 1.8624279445926182,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006953752440846535,
      "loss": 1.5301,
      "step": 21647
    },
    {
      "epoch": 1.8625139808999398,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006953503528928733,
      "loss": 1.7193,
      "step": 21648
    },
    {
      "epoch": 1.8626000172072614,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006953254611297315,
      "loss": 1.1065,
      "step": 21649
    },
    {
      "epoch": 1.862686053514583,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006953005687953006,
      "loss": 0.7902,
      "step": 21650
    },
    {
      "epoch": 1.862772089821905,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006952756758896535,
      "loss": 0.967,
      "step": 21651
    },
    {
      "epoch": 1.8628581261292265,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006952507824128629,
      "loss": 1.7205,
      "step": 21652
    },
    {
      "epoch": 1.8629441624365484,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006952258883650018,
      "loss": 1.2544,
      "step": 21653
    },
    {
      "epoch": 1.86303019874387,
      "grad_norm": 1.125,
      "learning_rate": 0.0006952009937461432,
      "loss": 1.2515,
      "step": 21654
    },
    {
      "epoch": 1.8631162350511916,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006951760985563593,
      "loss": 1.0886,
      "step": 21655
    },
    {
      "epoch": 1.8632022713585132,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006951512027957237,
      "loss": 1.5081,
      "step": 21656
    },
    {
      "epoch": 1.8632883076658349,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0006951263064643084,
      "loss": 1.3649,
      "step": 21657
    },
    {
      "epoch": 1.8633743439731567,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006951014095621868,
      "loss": 0.9945,
      "step": 21658
    },
    {
      "epoch": 1.8634603802804783,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006950765120894315,
      "loss": 0.579,
      "step": 21659
    },
    {
      "epoch": 1.8635464165878002,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006950516140461153,
      "loss": 1.021,
      "step": 21660
    },
    {
      "epoch": 1.8636324528951218,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006950267154323111,
      "loss": 1.5331,
      "step": 21661
    },
    {
      "epoch": 1.8637184892024434,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006950018162480917,
      "loss": 1.1113,
      "step": 21662
    },
    {
      "epoch": 1.863804525509765,
      "grad_norm": 1.5,
      "learning_rate": 0.0006949769164935298,
      "loss": 1.3138,
      "step": 21663
    },
    {
      "epoch": 1.8638905618170867,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006949520161686984,
      "loss": 1.1946,
      "step": 21664
    },
    {
      "epoch": 1.8639765981244085,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006949271152736704,
      "loss": 1.4732,
      "step": 21665
    },
    {
      "epoch": 1.8640626344317301,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006949022138085183,
      "loss": 1.0781,
      "step": 21666
    },
    {
      "epoch": 1.864148670739052,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006948773117733152,
      "loss": 1.4989,
      "step": 21667
    },
    {
      "epoch": 1.8642347070463736,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006948524091681339,
      "loss": 0.9956,
      "step": 21668
    },
    {
      "epoch": 1.8643207433536952,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006948275059930472,
      "loss": 1.1404,
      "step": 21669
    },
    {
      "epoch": 1.8644067796610169,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006948026022481279,
      "loss": 1.0532,
      "step": 21670
    },
    {
      "epoch": 1.8644928159683385,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006947776979334488,
      "loss": 1.8696,
      "step": 21671
    },
    {
      "epoch": 1.8645788522756603,
      "grad_norm": 1.578125,
      "learning_rate": 0.0006947527930490829,
      "loss": 1.373,
      "step": 21672
    },
    {
      "epoch": 1.8646648885829822,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006947278875951031,
      "loss": 0.8136,
      "step": 21673
    },
    {
      "epoch": 1.8647509248903038,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006947029815715819,
      "loss": 1.284,
      "step": 21674
    },
    {
      "epoch": 1.8648369611976254,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006946780749785924,
      "loss": 1.081,
      "step": 21675
    },
    {
      "epoch": 1.864922997504947,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006946531678162074,
      "loss": 1.098,
      "step": 21676
    },
    {
      "epoch": 1.8650090338122687,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006946282600844997,
      "loss": 0.7499,
      "step": 21677
    },
    {
      "epoch": 1.8650950701195903,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006946033517835423,
      "loss": 1.479,
      "step": 21678
    },
    {
      "epoch": 1.8651811064269121,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006945784429134078,
      "loss": 1.4113,
      "step": 21679
    },
    {
      "epoch": 1.865267142734234,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006945535334741693,
      "loss": 1.0409,
      "step": 21680
    },
    {
      "epoch": 1.8653531790415556,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006945286234658995,
      "loss": 1.2503,
      "step": 21681
    },
    {
      "epoch": 1.8654392153488772,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006945037128886714,
      "loss": 0.7847,
      "step": 21682
    },
    {
      "epoch": 1.8655252516561989,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006944788017425578,
      "loss": 1.3994,
      "step": 21683
    },
    {
      "epoch": 1.8656112879635205,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006944538900276314,
      "loss": 0.7456,
      "step": 21684
    },
    {
      "epoch": 1.8656973242708423,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006944289777439652,
      "loss": 1.504,
      "step": 21685
    },
    {
      "epoch": 1.865783360578164,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006944040648916321,
      "loss": 1.3001,
      "step": 21686
    },
    {
      "epoch": 1.8658693968854858,
      "grad_norm": 1.3125,
      "learning_rate": 0.000694379151470705,
      "loss": 1.671,
      "step": 21687
    },
    {
      "epoch": 1.8659554331928074,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006943542374812564,
      "loss": 0.5815,
      "step": 21688
    },
    {
      "epoch": 1.866041469500129,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006943293229233598,
      "loss": 0.7824,
      "step": 21689
    },
    {
      "epoch": 1.8661275058074507,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006943044077970875,
      "loss": 1.1816,
      "step": 21690
    },
    {
      "epoch": 1.8662135421147723,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006942794921025126,
      "loss": 0.9244,
      "step": 21691
    },
    {
      "epoch": 1.8662995784220942,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006942545758397082,
      "loss": 0.7796,
      "step": 21692
    },
    {
      "epoch": 1.8663856147294158,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006942296590087466,
      "loss": 0.9697,
      "step": 21693
    },
    {
      "epoch": 1.8664716510367376,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006942047416097013,
      "loss": 1.3379,
      "step": 21694
    },
    {
      "epoch": 1.8665576873440592,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006941798236426448,
      "loss": 1.0434,
      "step": 21695
    },
    {
      "epoch": 1.8666437236513809,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006941549051076501,
      "loss": 0.9819,
      "step": 21696
    },
    {
      "epoch": 1.8667297599587025,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00069412998600479,
      "loss": 1.0028,
      "step": 21697
    },
    {
      "epoch": 1.8668157962660241,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006941050663341376,
      "loss": 2.0599,
      "step": 21698
    },
    {
      "epoch": 1.866901832573346,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006940801460957655,
      "loss": 1.2037,
      "step": 21699
    },
    {
      "epoch": 1.8669878688806676,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0006940552252897468,
      "loss": 1.2254,
      "step": 21700
    },
    {
      "epoch": 1.8670739051879894,
      "grad_norm": 1.5,
      "learning_rate": 0.0006940303039161543,
      "loss": 1.8509,
      "step": 21701
    },
    {
      "epoch": 1.867159941495311,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006940053819750609,
      "loss": 1.1733,
      "step": 21702
    },
    {
      "epoch": 1.8672459778026327,
      "grad_norm": 1.125,
      "learning_rate": 0.0006939804594665393,
      "loss": 0.9545,
      "step": 21703
    },
    {
      "epoch": 1.8673320141099543,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006939555363906629,
      "loss": 1.8271,
      "step": 21704
    },
    {
      "epoch": 1.867418050417276,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000693930612747504,
      "loss": 0.991,
      "step": 21705
    },
    {
      "epoch": 1.8675040867245978,
      "grad_norm": 1.390625,
      "learning_rate": 0.000693905688537136,
      "loss": 1.329,
      "step": 21706
    },
    {
      "epoch": 1.8675901230319196,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006938807637596314,
      "loss": 1.2007,
      "step": 21707
    },
    {
      "epoch": 1.8676761593392412,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006938558384150632,
      "loss": 0.7737,
      "step": 21708
    },
    {
      "epoch": 1.8677621956465629,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006938309125035046,
      "loss": 1.7789,
      "step": 21709
    },
    {
      "epoch": 1.8678482319538845,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006938059860250281,
      "loss": 1.657,
      "step": 21710
    },
    {
      "epoch": 1.8679342682612061,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006937810589797069,
      "loss": 1.7522,
      "step": 21711
    },
    {
      "epoch": 1.868020304568528,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006937561313676139,
      "loss": 0.7903,
      "step": 21712
    },
    {
      "epoch": 1.8681063408758496,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006937312031888217,
      "loss": 1.2308,
      "step": 21713
    },
    {
      "epoch": 1.8681923771831714,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006937062744434035,
      "loss": 1.1726,
      "step": 21714
    },
    {
      "epoch": 1.868278413490493,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006936813451314321,
      "loss": 1.609,
      "step": 21715
    },
    {
      "epoch": 1.8683644497978147,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006936564152529804,
      "loss": 1.8726,
      "step": 21716
    },
    {
      "epoch": 1.8684504861051363,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006936314848081215,
      "loss": 1.082,
      "step": 21717
    },
    {
      "epoch": 1.868536522412458,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006936065537969281,
      "loss": 1.342,
      "step": 21718
    },
    {
      "epoch": 1.8686225587197798,
      "grad_norm": 1.1875,
      "learning_rate": 0.000693581622219473,
      "loss": 0.9388,
      "step": 21719
    },
    {
      "epoch": 1.8687085950271014,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006935566900758294,
      "loss": 1.3553,
      "step": 21720
    },
    {
      "epoch": 1.8687946313344233,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006935317573660702,
      "loss": 2.0173,
      "step": 21721
    },
    {
      "epoch": 1.8688806676417449,
      "grad_norm": 1.125,
      "learning_rate": 0.0006935068240902682,
      "loss": 1.9364,
      "step": 21722
    },
    {
      "epoch": 1.8689667039490665,
      "grad_norm": 1.375,
      "learning_rate": 0.0006934818902484964,
      "loss": 1.2847,
      "step": 21723
    },
    {
      "epoch": 1.8690527402563881,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006934569558408277,
      "loss": 1.3315,
      "step": 21724
    },
    {
      "epoch": 1.8691387765637097,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000693432020867335,
      "loss": 1.0161,
      "step": 21725
    },
    {
      "epoch": 1.8692248128710316,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006934070853280911,
      "loss": 0.933,
      "step": 21726
    },
    {
      "epoch": 1.8693108491783532,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006933821492231693,
      "loss": 2.1641,
      "step": 21727
    },
    {
      "epoch": 1.869396885485675,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006933572125526424,
      "loss": 1.2327,
      "step": 21728
    },
    {
      "epoch": 1.8694829217929967,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006933322753165832,
      "loss": 1.7448,
      "step": 21729
    },
    {
      "epoch": 1.8695689581003183,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006933073375150646,
      "loss": 0.8396,
      "step": 21730
    },
    {
      "epoch": 1.86965499440764,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006932823991481597,
      "loss": 0.6852,
      "step": 21731
    },
    {
      "epoch": 1.8697410307149616,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006932574602159414,
      "loss": 1.6035,
      "step": 21732
    },
    {
      "epoch": 1.8698270670222834,
      "grad_norm": 1.703125,
      "learning_rate": 0.0006932325207184826,
      "loss": 1.1653,
      "step": 21733
    },
    {
      "epoch": 1.8699131033296053,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006932075806558561,
      "loss": 1.6535,
      "step": 21734
    },
    {
      "epoch": 1.8699991396369269,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006931826400281352,
      "loss": 1.5044,
      "step": 21735
    },
    {
      "epoch": 1.8700851759442485,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006931576988353926,
      "loss": 0.8207,
      "step": 21736
    },
    {
      "epoch": 1.8701712122515701,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006931327570777012,
      "loss": 0.5467,
      "step": 21737
    },
    {
      "epoch": 1.8702572485588917,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006931078147551342,
      "loss": 1.461,
      "step": 21738
    },
    {
      "epoch": 1.8703432848662136,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006930828718677643,
      "loss": 1.4389,
      "step": 21739
    },
    {
      "epoch": 1.8704293211735352,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006930579284156646,
      "loss": 1.7775,
      "step": 21740
    },
    {
      "epoch": 1.870515357480857,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000693032984398908,
      "loss": 2.0525,
      "step": 21741
    },
    {
      "epoch": 1.8706013937881787,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006930080398175675,
      "loss": 0.6064,
      "step": 21742
    },
    {
      "epoch": 1.8706874300955003,
      "grad_norm": 1.125,
      "learning_rate": 0.000692983094671716,
      "loss": 1.5343,
      "step": 21743
    },
    {
      "epoch": 1.870773466402822,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006929581489614262,
      "loss": 1.5986,
      "step": 21744
    },
    {
      "epoch": 1.8708595027101436,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0006929332026867716,
      "loss": 1.3452,
      "step": 21745
    },
    {
      "epoch": 1.8709455390174654,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006929082558478249,
      "loss": 0.8308,
      "step": 21746
    },
    {
      "epoch": 1.871031575324787,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006928833084446592,
      "loss": 1.2749,
      "step": 21747
    },
    {
      "epoch": 1.8711176116321089,
      "grad_norm": 1.671875,
      "learning_rate": 0.0006928583604773472,
      "loss": 0.816,
      "step": 21748
    },
    {
      "epoch": 1.8712036479394305,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000692833411945962,
      "loss": 1.8391,
      "step": 21749
    },
    {
      "epoch": 1.8712896842467521,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006928084628505765,
      "loss": 1.7991,
      "step": 21750
    },
    {
      "epoch": 1.8713757205540738,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006927835131912639,
      "loss": 1.1871,
      "step": 21751
    },
    {
      "epoch": 1.8714617568613954,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000692758562968097,
      "loss": 1.7502,
      "step": 21752
    },
    {
      "epoch": 1.8715477931687172,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006927336121811487,
      "loss": 1.6998,
      "step": 21753
    },
    {
      "epoch": 1.8716338294760388,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006927086608304921,
      "loss": 1.5362,
      "step": 21754
    },
    {
      "epoch": 1.8717198657833607,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006926837089162002,
      "loss": 1.6093,
      "step": 21755
    },
    {
      "epoch": 1.8718059020906823,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006926587564383457,
      "loss": 1.6181,
      "step": 21756
    },
    {
      "epoch": 1.871891938398004,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006926338033970021,
      "loss": 0.89,
      "step": 21757
    },
    {
      "epoch": 1.8719779747053256,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000692608849792242,
      "loss": 0.8895,
      "step": 21758
    },
    {
      "epoch": 1.8720640110126472,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006925838956241384,
      "loss": 1.588,
      "step": 21759
    },
    {
      "epoch": 1.872150047319969,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006925589408927645,
      "loss": 0.7593,
      "step": 21760
    },
    {
      "epoch": 1.8722360836272909,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006925339855981932,
      "loss": 0.9566,
      "step": 21761
    },
    {
      "epoch": 1.8723221199346125,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006925090297404972,
      "loss": 0.9793,
      "step": 21762
    },
    {
      "epoch": 1.8724081562419341,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006924840733197497,
      "loss": 0.9047,
      "step": 21763
    },
    {
      "epoch": 1.8724941925492558,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006924591163360239,
      "loss": 1.5115,
      "step": 21764
    },
    {
      "epoch": 1.8725802288565774,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006924341587893926,
      "loss": 1.7416,
      "step": 21765
    },
    {
      "epoch": 1.872666265163899,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006924092006799288,
      "loss": 1.4429,
      "step": 21766
    },
    {
      "epoch": 1.8727523014712208,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006923842420077053,
      "loss": 1.6342,
      "step": 21767
    },
    {
      "epoch": 1.8728383377785427,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006923592827727956,
      "loss": 0.9538,
      "step": 21768
    },
    {
      "epoch": 1.8729243740858643,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006923343229752724,
      "loss": 0.8201,
      "step": 21769
    },
    {
      "epoch": 1.873010410393186,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006923093626152086,
      "loss": 0.9854,
      "step": 21770
    },
    {
      "epoch": 1.8730964467005076,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006922844016926772,
      "loss": 2.1907,
      "step": 21771
    },
    {
      "epoch": 1.8731824830078292,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006922594402077515,
      "loss": 1.6015,
      "step": 21772
    },
    {
      "epoch": 1.873268519315151,
      "grad_norm": 1.25,
      "learning_rate": 0.0006922344781605044,
      "loss": 1.1226,
      "step": 21773
    },
    {
      "epoch": 1.8733545556224727,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006922095155510087,
      "loss": 1.1721,
      "step": 21774
    },
    {
      "epoch": 1.8734405919297945,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006921845523793376,
      "loss": 1.206,
      "step": 21775
    },
    {
      "epoch": 1.8735266282371161,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006921595886455641,
      "loss": 0.7733,
      "step": 21776
    },
    {
      "epoch": 1.8736126645444378,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006921346243497611,
      "loss": 0.8366,
      "step": 21777
    },
    {
      "epoch": 1.8736987008517594,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006921096594920017,
      "loss": 1.7012,
      "step": 21778
    },
    {
      "epoch": 1.873784737159081,
      "grad_norm": 1.9375,
      "learning_rate": 0.0006920846940723589,
      "loss": 2.0264,
      "step": 21779
    },
    {
      "epoch": 1.8738707734664029,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006920597280909058,
      "loss": 1.1245,
      "step": 21780
    },
    {
      "epoch": 1.8739568097737245,
      "grad_norm": 2.015625,
      "learning_rate": 0.0006920347615477153,
      "loss": 1.1684,
      "step": 21781
    },
    {
      "epoch": 1.8740428460810463,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006920097944428604,
      "loss": 1.7409,
      "step": 21782
    },
    {
      "epoch": 1.874128882388368,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006919848267764141,
      "loss": 1.031,
      "step": 21783
    },
    {
      "epoch": 1.8742149186956896,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006919598585484498,
      "loss": 1.336,
      "step": 21784
    },
    {
      "epoch": 1.8743009550030112,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006919348897590402,
      "loss": 0.7406,
      "step": 21785
    },
    {
      "epoch": 1.8743869913103328,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006919099204082582,
      "loss": 1.553,
      "step": 21786
    },
    {
      "epoch": 1.8744730276176547,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0006918849504961771,
      "loss": 1.1706,
      "step": 21787
    },
    {
      "epoch": 1.8745590639249763,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00069185998002287,
      "loss": 0.5171,
      "step": 21788
    },
    {
      "epoch": 1.8746451002322981,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006918350089884095,
      "loss": 0.9594,
      "step": 21789
    },
    {
      "epoch": 1.8747311365396198,
      "grad_norm": 0.953125,
      "learning_rate": 0.000691810037392869,
      "loss": 1.3373,
      "step": 21790
    },
    {
      "epoch": 1.8748171728469414,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006917850652363215,
      "loss": 1.0794,
      "step": 21791
    },
    {
      "epoch": 1.874903209154263,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006917600925188401,
      "loss": 0.7129,
      "step": 21792
    },
    {
      "epoch": 1.8749892454615846,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006917351192404974,
      "loss": 1.1999,
      "step": 21793
    },
    {
      "epoch": 1.8750752817689065,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006917101454013671,
      "loss": 2.0873,
      "step": 21794
    },
    {
      "epoch": 1.8751613180762283,
      "grad_norm": 1.71875,
      "learning_rate": 0.0006916851710015217,
      "loss": 1.8778,
      "step": 21795
    },
    {
      "epoch": 1.87524735438355,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006916601960410346,
      "loss": 1.548,
      "step": 21796
    },
    {
      "epoch": 1.8753333906908716,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006916352205199785,
      "loss": 0.8671,
      "step": 21797
    },
    {
      "epoch": 1.8754194269981932,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006916102444384269,
      "loss": 1.8895,
      "step": 21798
    },
    {
      "epoch": 1.8755054633055148,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006915852677964524,
      "loss": 1.0754,
      "step": 21799
    },
    {
      "epoch": 1.8755914996128367,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006915602905941283,
      "loss": 0.8925,
      "step": 21800
    },
    {
      "epoch": 1.8756775359201583,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006915353128315276,
      "loss": 1.3577,
      "step": 21801
    },
    {
      "epoch": 1.8757635722274801,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006915103345087235,
      "loss": 1.0817,
      "step": 21802
    },
    {
      "epoch": 1.8758496085348018,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006914853556257887,
      "loss": 0.8433,
      "step": 21803
    },
    {
      "epoch": 1.8759356448421234,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006914603761827967,
      "loss": 1.117,
      "step": 21804
    },
    {
      "epoch": 1.876021681149445,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006914353961798202,
      "loss": 0.8108,
      "step": 21805
    },
    {
      "epoch": 1.8761077174567666,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006914104156169324,
      "loss": 1.1783,
      "step": 21806
    },
    {
      "epoch": 1.8761937537640885,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006913854344942064,
      "loss": 0.9613,
      "step": 21807
    },
    {
      "epoch": 1.87627979007141,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006913604528117152,
      "loss": 0.6022,
      "step": 21808
    },
    {
      "epoch": 1.876365826378732,
      "grad_norm": 1.15625,
      "learning_rate": 0.000691335470569532,
      "loss": 1.2375,
      "step": 21809
    },
    {
      "epoch": 1.8764518626860536,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006913104877677296,
      "loss": 1.6775,
      "step": 21810
    },
    {
      "epoch": 1.8765378989933752,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006912855044063814,
      "loss": 0.8572,
      "step": 21811
    },
    {
      "epoch": 1.8766239353006968,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006912605204855601,
      "loss": 1.0777,
      "step": 21812
    },
    {
      "epoch": 1.8767099716080184,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006912355360053391,
      "loss": 1.3129,
      "step": 21813
    },
    {
      "epoch": 1.8767960079153403,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006912105509657914,
      "loss": 1.0935,
      "step": 21814
    },
    {
      "epoch": 1.876882044222662,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00069118556536699,
      "loss": 1.4241,
      "step": 21815
    },
    {
      "epoch": 1.8769680805299838,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006911605792090079,
      "loss": 0.8597,
      "step": 21816
    },
    {
      "epoch": 1.8770541168373054,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006911355924919183,
      "loss": 1.2314,
      "step": 21817
    },
    {
      "epoch": 1.877140153144627,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006911106052157942,
      "loss": 1.8301,
      "step": 21818
    },
    {
      "epoch": 1.8772261894519486,
      "grad_norm": 1.21875,
      "learning_rate": 0.000691085617380709,
      "loss": 1.4556,
      "step": 21819
    },
    {
      "epoch": 1.8773122257592703,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006910606289867353,
      "loss": 1.1555,
      "step": 21820
    },
    {
      "epoch": 1.877398262066592,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006910356400339466,
      "loss": 1.5042,
      "step": 21821
    },
    {
      "epoch": 1.877484298373914,
      "grad_norm": 2.140625,
      "learning_rate": 0.0006910106505224156,
      "loss": 1.6241,
      "step": 21822
    },
    {
      "epoch": 1.8775703346812356,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006909856604522158,
      "loss": 1.4668,
      "step": 21823
    },
    {
      "epoch": 1.8776563709885572,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006909606698234199,
      "loss": 0.6623,
      "step": 21824
    },
    {
      "epoch": 1.8777424072958788,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006909356786361014,
      "loss": 0.8814,
      "step": 21825
    },
    {
      "epoch": 1.8778284436032004,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006909106868903331,
      "loss": 0.8379,
      "step": 21826
    },
    {
      "epoch": 1.8779144799105223,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006908856945861881,
      "loss": 1.3241,
      "step": 21827
    },
    {
      "epoch": 1.878000516217844,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006908607017237396,
      "loss": 1.8282,
      "step": 21828
    },
    {
      "epoch": 1.8780865525251658,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006908357083030607,
      "loss": 1.0905,
      "step": 21829
    },
    {
      "epoch": 1.8781725888324874,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006908107143242245,
      "loss": 0.9463,
      "step": 21830
    },
    {
      "epoch": 1.878258625139809,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006907857197873038,
      "loss": 0.9304,
      "step": 21831
    },
    {
      "epoch": 1.8783446614471306,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006907607246923723,
      "loss": 1.2023,
      "step": 21832
    },
    {
      "epoch": 1.8784306977544523,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006907357290395026,
      "loss": 0.761,
      "step": 21833
    },
    {
      "epoch": 1.878516734061774,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000690710732828768,
      "loss": 1.5337,
      "step": 21834
    },
    {
      "epoch": 1.8786027703690957,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006906857360602417,
      "loss": 0.7417,
      "step": 21835
    },
    {
      "epoch": 1.8786888066764176,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006906607387339967,
      "loss": 0.6791,
      "step": 21836
    },
    {
      "epoch": 1.8787748429837392,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006906357408501061,
      "loss": 1.5909,
      "step": 21837
    },
    {
      "epoch": 1.8788608792910608,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000690610742408643,
      "loss": 1.2826,
      "step": 21838
    },
    {
      "epoch": 1.8789469155983824,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006905857434096805,
      "loss": 0.8387,
      "step": 21839
    },
    {
      "epoch": 1.879032951905704,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006905607438532919,
      "loss": 0.986,
      "step": 21840
    },
    {
      "epoch": 1.879118988213026,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006905357437395501,
      "loss": 1.6143,
      "step": 21841
    },
    {
      "epoch": 1.8792050245203475,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006905107430685282,
      "loss": 1.2909,
      "step": 21842
    },
    {
      "epoch": 1.8792910608276694,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006904857418402997,
      "loss": 1.6444,
      "step": 21843
    },
    {
      "epoch": 1.879377097134991,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006904607400549373,
      "loss": 0.9129,
      "step": 21844
    },
    {
      "epoch": 1.8794631334423126,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006904357377125143,
      "loss": 1.9127,
      "step": 21845
    },
    {
      "epoch": 1.8795491697496343,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0006904107348131037,
      "loss": 1.2537,
      "step": 21846
    },
    {
      "epoch": 1.8796352060569559,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006903857313567789,
      "loss": 0.867,
      "step": 21847
    },
    {
      "epoch": 1.8797212423642777,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006903607273436127,
      "loss": 1.0737,
      "step": 21848
    },
    {
      "epoch": 1.8798072786715996,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006903357227736785,
      "loss": 0.7395,
      "step": 21849
    },
    {
      "epoch": 1.8798933149789212,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006903107176470493,
      "loss": 1.4442,
      "step": 21850
    },
    {
      "epoch": 1.8799793512862428,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006902857119637984,
      "loss": 1.151,
      "step": 21851
    },
    {
      "epoch": 1.8800653875935645,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006902607057239987,
      "loss": 1.7357,
      "step": 21852
    },
    {
      "epoch": 1.880151423900886,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006902356989277233,
      "loss": 1.7207,
      "step": 21853
    },
    {
      "epoch": 1.8802374602082077,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006902106915750457,
      "loss": 0.9244,
      "step": 21854
    },
    {
      "epoch": 1.8803234965155295,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006901856836660387,
      "loss": 1.2793,
      "step": 21855
    },
    {
      "epoch": 1.8804095328228514,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006901606752007755,
      "loss": 0.9377,
      "step": 21856
    },
    {
      "epoch": 1.880495569130173,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006901356661793293,
      "loss": 1.6791,
      "step": 21857
    },
    {
      "epoch": 1.8805816054374946,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006901106566017735,
      "loss": 1.403,
      "step": 21858
    },
    {
      "epoch": 1.8806676417448163,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006900856464681809,
      "loss": 2.0947,
      "step": 21859
    },
    {
      "epoch": 1.880753678052138,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006900606357786247,
      "loss": 0.8837,
      "step": 21860
    },
    {
      "epoch": 1.8808397143594597,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000690035624533178,
      "loss": 0.7327,
      "step": 21861
    },
    {
      "epoch": 1.8809257506667814,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006900106127319142,
      "loss": 0.757,
      "step": 21862
    },
    {
      "epoch": 1.8810117869741032,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006899856003749063,
      "loss": 0.9282,
      "step": 21863
    },
    {
      "epoch": 1.8810978232814248,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006899605874622273,
      "loss": 0.7773,
      "step": 21864
    },
    {
      "epoch": 1.8811838595887465,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006899355739939508,
      "loss": 2.0374,
      "step": 21865
    },
    {
      "epoch": 1.881269895896068,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006899105599701495,
      "loss": 1.6216,
      "step": 21866
    },
    {
      "epoch": 1.8813559322033897,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006898855453908968,
      "loss": 1.2629,
      "step": 21867
    },
    {
      "epoch": 1.8814419685107115,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006898605302562658,
      "loss": 1.7051,
      "step": 21868
    },
    {
      "epoch": 1.8815280048180332,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006898355145663297,
      "loss": 1.3677,
      "step": 21869
    },
    {
      "epoch": 1.881614041125355,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006898104983211617,
      "loss": 1.5465,
      "step": 21870
    },
    {
      "epoch": 1.8817000774326766,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006897854815208349,
      "loss": 1.478,
      "step": 21871
    },
    {
      "epoch": 1.8817861137399983,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0006897604641654223,
      "loss": 1.227,
      "step": 21872
    },
    {
      "epoch": 1.88187215004732,
      "grad_norm": 1.125,
      "learning_rate": 0.0006897354462549974,
      "loss": 0.863,
      "step": 21873
    },
    {
      "epoch": 1.8819581863546415,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006897104277896332,
      "loss": 0.8696,
      "step": 21874
    },
    {
      "epoch": 1.8820442226619634,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006896854087694029,
      "loss": 0.9884,
      "step": 21875
    },
    {
      "epoch": 1.882130258969285,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006896603891943796,
      "loss": 1.129,
      "step": 21876
    },
    {
      "epoch": 1.8822162952766068,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006896353690646367,
      "loss": 1.0303,
      "step": 21877
    },
    {
      "epoch": 1.8823023315839285,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006896103483802471,
      "loss": 1.6166,
      "step": 21878
    },
    {
      "epoch": 1.88238836789125,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000689585327141284,
      "loss": 0.935,
      "step": 21879
    },
    {
      "epoch": 1.8824744041985717,
      "grad_norm": 1.65625,
      "learning_rate": 0.000689560305347821,
      "loss": 1.4333,
      "step": 21880
    },
    {
      "epoch": 1.8825604405058933,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006895352829999307,
      "loss": 1.0938,
      "step": 21881
    },
    {
      "epoch": 1.8826464768132152,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006895102600976868,
      "loss": 1.7205,
      "step": 21882
    },
    {
      "epoch": 1.882732513120537,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006894852366411621,
      "loss": 1.4235,
      "step": 21883
    },
    {
      "epoch": 1.8828185494278586,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006894602126304299,
      "loss": 0.8425,
      "step": 21884
    },
    {
      "epoch": 1.8829045857351803,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006894351880655635,
      "loss": 0.8815,
      "step": 21885
    },
    {
      "epoch": 1.882990622042502,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000689410162946636,
      "loss": 0.8824,
      "step": 21886
    },
    {
      "epoch": 1.8830766583498235,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006893851372737205,
      "loss": 1.2189,
      "step": 21887
    },
    {
      "epoch": 1.8831626946571454,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006893601110468905,
      "loss": 0.9542,
      "step": 21888
    },
    {
      "epoch": 1.883248730964467,
      "grad_norm": 1.796875,
      "learning_rate": 0.0006893350842662189,
      "loss": 1.1018,
      "step": 21889
    },
    {
      "epoch": 1.8833347672717888,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006893100569317789,
      "loss": 1.8102,
      "step": 21890
    },
    {
      "epoch": 1.8834208035791105,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006892850290436441,
      "loss": 1.8249,
      "step": 21891
    },
    {
      "epoch": 1.883506839886432,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006892600006018871,
      "loss": 1.467,
      "step": 21892
    },
    {
      "epoch": 1.8835928761937537,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006892349716065816,
      "loss": 0.7958,
      "step": 21893
    },
    {
      "epoch": 1.8836789125010753,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006892099420578006,
      "loss": 1.7681,
      "step": 21894
    },
    {
      "epoch": 1.8837649488083972,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006891849119556172,
      "loss": 1.0152,
      "step": 21895
    },
    {
      "epoch": 1.8838509851157188,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006891598813001048,
      "loss": 1.1975,
      "step": 21896
    },
    {
      "epoch": 1.8839370214230406,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006891348500913365,
      "loss": 1.7589,
      "step": 21897
    },
    {
      "epoch": 1.8840230577303623,
      "grad_norm": 1.515625,
      "learning_rate": 0.0006891098183293856,
      "loss": 1.2903,
      "step": 21898
    },
    {
      "epoch": 1.884109094037684,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006890847860143252,
      "loss": 1.1795,
      "step": 21899
    },
    {
      "epoch": 1.8841951303450055,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006890597531462285,
      "loss": 1.0879,
      "step": 21900
    },
    {
      "epoch": 1.8842811666523271,
      "grad_norm": 1.140625,
      "learning_rate": 0.000689034719725169,
      "loss": 2.353,
      "step": 21901
    },
    {
      "epoch": 1.884367202959649,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006890096857512196,
      "loss": 1.0469,
      "step": 21902
    },
    {
      "epoch": 1.8844532392669706,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006889846512244536,
      "loss": 1.3431,
      "step": 21903
    },
    {
      "epoch": 1.8845392755742925,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006889596161449444,
      "loss": 1.7675,
      "step": 21904
    },
    {
      "epoch": 1.884625311881614,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006889345805127649,
      "loss": 0.6752,
      "step": 21905
    },
    {
      "epoch": 1.8847113481889357,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006889095443279885,
      "loss": 1.8236,
      "step": 21906
    },
    {
      "epoch": 1.8847973844962573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006888845075906886,
      "loss": 1.1309,
      "step": 21907
    },
    {
      "epoch": 1.884883420803579,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006888594703009382,
      "loss": 0.7161,
      "step": 21908
    },
    {
      "epoch": 1.8849694571109008,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006888344324588106,
      "loss": 1.7966,
      "step": 21909
    },
    {
      "epoch": 1.8850554934182226,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006888093940643789,
      "loss": 1.6417,
      "step": 21910
    },
    {
      "epoch": 1.8851415297255443,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006887843551177163,
      "loss": 1.1394,
      "step": 21911
    },
    {
      "epoch": 1.885227566032866,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006887593156188964,
      "loss": 1.2638,
      "step": 21912
    },
    {
      "epoch": 1.8853136023401875,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000688734275567992,
      "loss": 1.7256,
      "step": 21913
    },
    {
      "epoch": 1.8853996386475091,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006887092349650768,
      "loss": 1.6824,
      "step": 21914
    },
    {
      "epoch": 1.885485674954831,
      "grad_norm": 1.5,
      "learning_rate": 0.0006886841938102237,
      "loss": 1.4761,
      "step": 21915
    },
    {
      "epoch": 1.8855717112621526,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006886591521035061,
      "loss": 1.5029,
      "step": 21916
    },
    {
      "epoch": 1.8856577475694745,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006886341098449971,
      "loss": 1.8726,
      "step": 21917
    },
    {
      "epoch": 1.885743783876796,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00068860906703477,
      "loss": 1.1742,
      "step": 21918
    },
    {
      "epoch": 1.8858298201841177,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006885840236728982,
      "loss": 1.4807,
      "step": 21919
    },
    {
      "epoch": 1.8859158564914393,
      "grad_norm": 0.875,
      "learning_rate": 0.0006885589797594546,
      "loss": 1.42,
      "step": 21920
    },
    {
      "epoch": 1.886001892798761,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006885339352945128,
      "loss": 1.349,
      "step": 21921
    },
    {
      "epoch": 1.8860879291060828,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006885088902781459,
      "loss": 1.4222,
      "step": 21922
    },
    {
      "epoch": 1.8861739654134044,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006884838447104271,
      "loss": 1.2265,
      "step": 21923
    },
    {
      "epoch": 1.8862600017207263,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006884587985914297,
      "loss": 1.0732,
      "step": 21924
    },
    {
      "epoch": 1.886346038028048,
      "grad_norm": 0.921875,
      "learning_rate": 0.000688433751921227,
      "loss": 0.5456,
      "step": 21925
    },
    {
      "epoch": 1.8864320743353695,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006884087046998922,
      "loss": 0.9727,
      "step": 21926
    },
    {
      "epoch": 1.8865181106426911,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006883836569274987,
      "loss": 1.6062,
      "step": 21927
    },
    {
      "epoch": 1.8866041469500128,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006883586086041195,
      "loss": 0.7855,
      "step": 21928
    },
    {
      "epoch": 1.8866901832573346,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006883335597298279,
      "loss": 1.7327,
      "step": 21929
    },
    {
      "epoch": 1.8867762195646562,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006883085103046976,
      "loss": 1.1768,
      "step": 21930
    },
    {
      "epoch": 1.886862255871978,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006882834603288013,
      "loss": 0.7568,
      "step": 21931
    },
    {
      "epoch": 1.8869482921792997,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006882584098022124,
      "loss": 1.4736,
      "step": 21932
    },
    {
      "epoch": 1.8870343284866213,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006882333587250045,
      "loss": 1.189,
      "step": 21933
    },
    {
      "epoch": 1.887120364793943,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006882083070972506,
      "loss": 1.8029,
      "step": 21934
    },
    {
      "epoch": 1.8872064011012646,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006881832549190239,
      "loss": 1.4842,
      "step": 21935
    },
    {
      "epoch": 1.8872924374085864,
      "grad_norm": 0.765625,
      "learning_rate": 0.0006881582021903978,
      "loss": 0.8378,
      "step": 21936
    },
    {
      "epoch": 1.8873784737159083,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006881331489114456,
      "loss": 1.3329,
      "step": 21937
    },
    {
      "epoch": 1.88746451002323,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006881080950822406,
      "loss": 1.2081,
      "step": 21938
    },
    {
      "epoch": 1.8875505463305515,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006880830407028557,
      "loss": 1.9804,
      "step": 21939
    },
    {
      "epoch": 1.8876365826378732,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006880579857733649,
      "loss": 0.6335,
      "step": 21940
    },
    {
      "epoch": 1.8877226189451948,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006880329302938409,
      "loss": 1.5352,
      "step": 21941
    },
    {
      "epoch": 1.8878086552525164,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006880078742643572,
      "loss": 1.2892,
      "step": 21942
    },
    {
      "epoch": 1.8878946915598382,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006879828176849868,
      "loss": 1.5092,
      "step": 21943
    },
    {
      "epoch": 1.88798072786716,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006879577605558033,
      "loss": 1.0189,
      "step": 21944
    },
    {
      "epoch": 1.8880667641744817,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006879327028768801,
      "loss": 1.2724,
      "step": 21945
    },
    {
      "epoch": 1.8881528004818033,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006879076446482901,
      "loss": 1.1818,
      "step": 21946
    },
    {
      "epoch": 1.888238836789125,
      "grad_norm": 1.78125,
      "learning_rate": 0.0006878825858701069,
      "loss": 1.5275,
      "step": 21947
    },
    {
      "epoch": 1.8883248730964466,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006878575265424036,
      "loss": 1.3822,
      "step": 21948
    },
    {
      "epoch": 1.8884109094037684,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006878324666652535,
      "loss": 1.4499,
      "step": 21949
    },
    {
      "epoch": 1.88849694571109,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006878074062387303,
      "loss": 1.3641,
      "step": 21950
    },
    {
      "epoch": 1.888582982018412,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006877823452629067,
      "loss": 1.0874,
      "step": 21951
    },
    {
      "epoch": 1.8886690183257335,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006877572837378563,
      "loss": 1.149,
      "step": 21952
    },
    {
      "epoch": 1.8887550546330552,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006877322216636523,
      "loss": 0.8877,
      "step": 21953
    },
    {
      "epoch": 1.8888410909403768,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006877071590403681,
      "loss": 1.6788,
      "step": 21954
    },
    {
      "epoch": 1.8889271272476984,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006876820958680769,
      "loss": 1.3907,
      "step": 21955
    },
    {
      "epoch": 1.8890131635550202,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006876570321468523,
      "loss": 0.972,
      "step": 21956
    },
    {
      "epoch": 1.8890991998623419,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006876319678767673,
      "loss": 1.1441,
      "step": 21957
    },
    {
      "epoch": 1.8891852361696637,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006876069030578951,
      "loss": 1.1352,
      "step": 21958
    },
    {
      "epoch": 1.8892712724769853,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006875818376903093,
      "loss": 1.3092,
      "step": 21959
    },
    {
      "epoch": 1.889357308784307,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006875567717740833,
      "loss": 1.3901,
      "step": 21960
    },
    {
      "epoch": 1.8894433450916286,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006875317053092899,
      "loss": 0.9867,
      "step": 21961
    },
    {
      "epoch": 1.8895293813989502,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006875066382960029,
      "loss": 1.8889,
      "step": 21962
    },
    {
      "epoch": 1.889615417706272,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006874815707342955,
      "loss": 1.1585,
      "step": 21963
    },
    {
      "epoch": 1.8897014540135937,
      "grad_norm": 1.09375,
      "learning_rate": 0.000687456502624241,
      "loss": 1.4112,
      "step": 21964
    },
    {
      "epoch": 1.8897874903209155,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006874314339659124,
      "loss": 1.3736,
      "step": 21965
    },
    {
      "epoch": 1.8898735266282372,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006874063647593836,
      "loss": 1.8017,
      "step": 21966
    },
    {
      "epoch": 1.8899595629355588,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006873812950047276,
      "loss": 0.813,
      "step": 21967
    },
    {
      "epoch": 1.8900455992428804,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006873562247020176,
      "loss": 1.2217,
      "step": 21968
    },
    {
      "epoch": 1.890131635550202,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006873311538513272,
      "loss": 1.2345,
      "step": 21969
    },
    {
      "epoch": 1.8902176718575239,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0006873060824527295,
      "loss": 1.4931,
      "step": 21970
    },
    {
      "epoch": 1.8903037081648457,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006872810105062981,
      "loss": 1.3657,
      "step": 21971
    },
    {
      "epoch": 1.8903897444721673,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006872559380121061,
      "loss": 1.4513,
      "step": 21972
    },
    {
      "epoch": 1.890475780779489,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006872308649702268,
      "loss": 1.8333,
      "step": 21973
    },
    {
      "epoch": 1.8905618170868106,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006872057913807338,
      "loss": 1.5687,
      "step": 21974
    },
    {
      "epoch": 1.8906478533941322,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006871807172437002,
      "loss": 1.6464,
      "step": 21975
    },
    {
      "epoch": 1.890733889701454,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006871556425591994,
      "loss": 1.1624,
      "step": 21976
    },
    {
      "epoch": 1.8908199260087757,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006871305673273047,
      "loss": 1.4782,
      "step": 21977
    },
    {
      "epoch": 1.8909059623160975,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006871054915480897,
      "loss": 1.3883,
      "step": 21978
    },
    {
      "epoch": 1.8909919986234192,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006870804152216272,
      "loss": 1.2751,
      "step": 21979
    },
    {
      "epoch": 1.8910780349307408,
      "grad_norm": 3.90625,
      "learning_rate": 0.000687055338347991,
      "loss": 1.0631,
      "step": 21980
    },
    {
      "epoch": 1.8911640712380624,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006870302609272545,
      "loss": 1.5865,
      "step": 21981
    },
    {
      "epoch": 1.891250107545384,
      "grad_norm": 1.734375,
      "learning_rate": 0.0006870051829594906,
      "loss": 1.0199,
      "step": 21982
    },
    {
      "epoch": 1.8913361438527059,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006869801044447732,
      "loss": 0.7109,
      "step": 21983
    },
    {
      "epoch": 1.8914221801600275,
      "grad_norm": 2.015625,
      "learning_rate": 0.000686955025383175,
      "loss": 1.7719,
      "step": 21984
    },
    {
      "epoch": 1.8915082164673493,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006869299457747699,
      "loss": 0.7755,
      "step": 21985
    },
    {
      "epoch": 1.891594252774671,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006869048656196311,
      "loss": 1.5039,
      "step": 21986
    },
    {
      "epoch": 1.8916802890819926,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006868797849178318,
      "loss": 1.3068,
      "step": 21987
    },
    {
      "epoch": 1.8917663253893142,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006868547036694455,
      "loss": 0.8441,
      "step": 21988
    },
    {
      "epoch": 1.8918523616966358,
      "grad_norm": 2.109375,
      "learning_rate": 0.0006868296218745456,
      "loss": 1.8644,
      "step": 21989
    },
    {
      "epoch": 1.8919383980039577,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006868045395332054,
      "loss": 0.6478,
      "step": 21990
    },
    {
      "epoch": 1.8920244343112793,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006867794566454983,
      "loss": 0.5164,
      "step": 21991
    },
    {
      "epoch": 1.8921104706186012,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006867543732114974,
      "loss": 1.5634,
      "step": 21992
    },
    {
      "epoch": 1.8921965069259228,
      "grad_norm": 2.140625,
      "learning_rate": 0.0006867292892312765,
      "loss": 1.186,
      "step": 21993
    },
    {
      "epoch": 1.8922825432332444,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006867042047049087,
      "loss": 1.0561,
      "step": 21994
    },
    {
      "epoch": 1.892368579540566,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006866791196324673,
      "loss": 0.973,
      "step": 21995
    },
    {
      "epoch": 1.8924546158478877,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000686654034014026,
      "loss": 1.6994,
      "step": 21996
    },
    {
      "epoch": 1.8925406521552095,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006866289478496578,
      "loss": 0.9359,
      "step": 21997
    },
    {
      "epoch": 1.8926266884625313,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006866038611394363,
      "loss": 1.8305,
      "step": 21998
    },
    {
      "epoch": 1.892712724769853,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006865787738834346,
      "loss": 1.8809,
      "step": 21999
    },
    {
      "epoch": 1.8927987610771746,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006865536860817264,
      "loss": 1.1438,
      "step": 22000
    },
    {
      "epoch": 1.8928847973844962,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000686528597734385,
      "loss": 1.4988,
      "step": 22001
    },
    {
      "epoch": 1.8929708336918178,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006865035088414837,
      "loss": 1.3355,
      "step": 22002
    },
    {
      "epoch": 1.8930568699991397,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006864784194030957,
      "loss": 0.9424,
      "step": 22003
    },
    {
      "epoch": 1.8931429063064613,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000686453329419295,
      "loss": 1.4465,
      "step": 22004
    },
    {
      "epoch": 1.8932289426137832,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0006864282388901544,
      "loss": 1.3213,
      "step": 22005
    },
    {
      "epoch": 1.8933149789211048,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006864031478157474,
      "loss": 1.0906,
      "step": 22006
    },
    {
      "epoch": 1.8934010152284264,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006863780561961474,
      "loss": 0.8594,
      "step": 22007
    },
    {
      "epoch": 1.893487051535748,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006863529640314279,
      "loss": 1.6639,
      "step": 22008
    },
    {
      "epoch": 1.8935730878430697,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006863278713216623,
      "loss": 1.4875,
      "step": 22009
    },
    {
      "epoch": 1.8936591241503915,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006863027780669238,
      "loss": 1.3923,
      "step": 22010
    },
    {
      "epoch": 1.8937451604577131,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000686277684267286,
      "loss": 1.0713,
      "step": 22011
    },
    {
      "epoch": 1.893831196765035,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006862525899228221,
      "loss": 0.766,
      "step": 22012
    },
    {
      "epoch": 1.8939172330723566,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006862274950336057,
      "loss": 0.8316,
      "step": 22013
    },
    {
      "epoch": 1.8940032693796782,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00068620239959971,
      "loss": 1.4551,
      "step": 22014
    },
    {
      "epoch": 1.8940893056869998,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006861773036212086,
      "loss": 0.9406,
      "step": 22015
    },
    {
      "epoch": 1.8941753419943215,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006861522070981749,
      "loss": 0.738,
      "step": 22016
    },
    {
      "epoch": 1.8942613783016433,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0006861271100306819,
      "loss": 0.7727,
      "step": 22017
    },
    {
      "epoch": 1.894347414608965,
      "grad_norm": 1.65625,
      "learning_rate": 0.0006861020124188035,
      "loss": 1.4095,
      "step": 22018
    },
    {
      "epoch": 1.8944334509162868,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006860769142626129,
      "loss": 0.8462,
      "step": 22019
    },
    {
      "epoch": 1.8945194872236084,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006860518155621835,
      "loss": 1.0597,
      "step": 22020
    },
    {
      "epoch": 1.89460552353093,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006860267163175886,
      "loss": 1.9022,
      "step": 22021
    },
    {
      "epoch": 1.8946915598382517,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006860016165289017,
      "loss": 0.986,
      "step": 22022
    },
    {
      "epoch": 1.8947775961455733,
      "grad_norm": 1.6875,
      "learning_rate": 0.0006859765161961964,
      "loss": 1.2196,
      "step": 22023
    },
    {
      "epoch": 1.8948636324528951,
      "grad_norm": 0.828125,
      "learning_rate": 0.000685951415319546,
      "loss": 1.3283,
      "step": 22024
    },
    {
      "epoch": 1.894949668760217,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006859263138990237,
      "loss": 0.7494,
      "step": 22025
    },
    {
      "epoch": 1.8950357050675386,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006859012119347032,
      "loss": 0.7762,
      "step": 22026
    },
    {
      "epoch": 1.8951217413748602,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006858761094266579,
      "loss": 1.2757,
      "step": 22027
    },
    {
      "epoch": 1.8952077776821818,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006858510063749611,
      "loss": 0.7957,
      "step": 22028
    },
    {
      "epoch": 1.8952938139895035,
      "grad_norm": 1.515625,
      "learning_rate": 0.0006858259027796861,
      "loss": 1.3203,
      "step": 22029
    },
    {
      "epoch": 1.895379850296825,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006858007986409066,
      "loss": 1.8619,
      "step": 22030
    },
    {
      "epoch": 1.895465886604147,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000685775693958696,
      "loss": 1.8794,
      "step": 22031
    },
    {
      "epoch": 1.8955519229114688,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006857505887331274,
      "loss": 1.8521,
      "step": 22032
    },
    {
      "epoch": 1.8956379592187904,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006857254829642746,
      "loss": 1.446,
      "step": 22033
    },
    {
      "epoch": 1.895723995526112,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006857003766522109,
      "loss": 1.1779,
      "step": 22034
    },
    {
      "epoch": 1.8958100318334337,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006856752697970097,
      "loss": 1.3666,
      "step": 22035
    },
    {
      "epoch": 1.8958960681407553,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006856501623987443,
      "loss": 1.521,
      "step": 22036
    },
    {
      "epoch": 1.8959821044480771,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006856250544574885,
      "loss": 1.4799,
      "step": 22037
    },
    {
      "epoch": 1.8960681407553988,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006855999459733154,
      "loss": 1.4648,
      "step": 22038
    },
    {
      "epoch": 1.8961541770627206,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006855748369462985,
      "loss": 1.0301,
      "step": 22039
    },
    {
      "epoch": 1.8962402133700422,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006855497273765113,
      "loss": 2.127,
      "step": 22040
    },
    {
      "epoch": 1.8963262496773639,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006855246172640273,
      "loss": 0.9913,
      "step": 22041
    },
    {
      "epoch": 1.8964122859846855,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006854995066089197,
      "loss": 1.5412,
      "step": 22042
    },
    {
      "epoch": 1.896498322292007,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006854743954112625,
      "loss": 2.1294,
      "step": 22043
    },
    {
      "epoch": 1.896584358599329,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006854492836711283,
      "loss": 1.032,
      "step": 22044
    },
    {
      "epoch": 1.8966703949066506,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006854241713885914,
      "loss": 1.6936,
      "step": 22045
    },
    {
      "epoch": 1.8967564312139724,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006853990585637248,
      "loss": 1.4452,
      "step": 22046
    },
    {
      "epoch": 1.896842467521294,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000685373945196602,
      "loss": 0.9496,
      "step": 22047
    },
    {
      "epoch": 1.8969285038286157,
      "grad_norm": 1.0,
      "learning_rate": 0.0006853488312872963,
      "loss": 1.2913,
      "step": 22048
    },
    {
      "epoch": 1.8970145401359373,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006853237168358815,
      "loss": 1.6437,
      "step": 22049
    },
    {
      "epoch": 1.897100576443259,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006852986018424308,
      "loss": 1.8807,
      "step": 22050
    },
    {
      "epoch": 1.8971866127505808,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006852734863070176,
      "loss": 1.0319,
      "step": 22051
    },
    {
      "epoch": 1.8972726490579024,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006852483702297157,
      "loss": 1.367,
      "step": 22052
    },
    {
      "epoch": 1.8973586853652242,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006852232536105982,
      "loss": 1.3245,
      "step": 22053
    },
    {
      "epoch": 1.8974447216725459,
      "grad_norm": 512.0,
      "learning_rate": 0.0006851981364497388,
      "loss": 1.3389,
      "step": 22054
    },
    {
      "epoch": 1.8975307579798675,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006851730187472107,
      "loss": 2.2031,
      "step": 22055
    },
    {
      "epoch": 1.897616794287189,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006851479005030878,
      "loss": 1.4635,
      "step": 22056
    },
    {
      "epoch": 1.8977028305945107,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006851227817174431,
      "loss": 1.839,
      "step": 22057
    },
    {
      "epoch": 1.8977888669018326,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006850976623903503,
      "loss": 1.1151,
      "step": 22058
    },
    {
      "epoch": 1.8978749032091544,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006850725425218828,
      "loss": 0.9869,
      "step": 22059
    },
    {
      "epoch": 1.897960939516476,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006850474221121142,
      "loss": 0.683,
      "step": 22060
    },
    {
      "epoch": 1.8980469758237977,
      "grad_norm": 1.25,
      "learning_rate": 0.0006850223011611178,
      "loss": 1.9649,
      "step": 22061
    },
    {
      "epoch": 1.8981330121311193,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006849971796689673,
      "loss": 0.395,
      "step": 22062
    },
    {
      "epoch": 1.898219048438441,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006849720576357359,
      "loss": 0.6561,
      "step": 22063
    },
    {
      "epoch": 1.8983050847457628,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006849469350614973,
      "loss": 0.7614,
      "step": 22064
    },
    {
      "epoch": 1.8983911210530844,
      "grad_norm": 1.375,
      "learning_rate": 0.0006849218119463249,
      "loss": 1.9429,
      "step": 22065
    },
    {
      "epoch": 1.8984771573604062,
      "grad_norm": 1.25,
      "learning_rate": 0.000684896688290292,
      "loss": 2.0663,
      "step": 22066
    },
    {
      "epoch": 1.8985631936677279,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006848715640934724,
      "loss": 0.9494,
      "step": 22067
    },
    {
      "epoch": 1.8986492299750495,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006848464393559393,
      "loss": 0.656,
      "step": 22068
    },
    {
      "epoch": 1.898735266282371,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006848213140777665,
      "loss": 1.494,
      "step": 22069
    },
    {
      "epoch": 1.8988213025896927,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006847961882590271,
      "loss": 1.5052,
      "step": 22070
    },
    {
      "epoch": 1.8989073388970146,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006847710618997949,
      "loss": 0.9717,
      "step": 22071
    },
    {
      "epoch": 1.8989933752043362,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006847459350001432,
      "loss": 1.4698,
      "step": 22072
    },
    {
      "epoch": 1.899079411511658,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006847208075601457,
      "loss": 1.0879,
      "step": 22073
    },
    {
      "epoch": 1.8991654478189797,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006846956795798758,
      "loss": 0.7055,
      "step": 22074
    },
    {
      "epoch": 1.8992514841263013,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006846705510594068,
      "loss": 1.2527,
      "step": 22075
    },
    {
      "epoch": 1.899337520433623,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006846454219988124,
      "loss": 0.9697,
      "step": 22076
    },
    {
      "epoch": 1.8994235567409445,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000684620292398166,
      "loss": 1.1812,
      "step": 22077
    },
    {
      "epoch": 1.8995095930482664,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006845951622575411,
      "loss": 1.1813,
      "step": 22078
    },
    {
      "epoch": 1.899595629355588,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006845700315770115,
      "loss": 1.0565,
      "step": 22079
    },
    {
      "epoch": 1.8996816656629099,
      "grad_norm": 1.125,
      "learning_rate": 0.0006845449003566503,
      "loss": 1.2256,
      "step": 22080
    },
    {
      "epoch": 1.8997677019702315,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006845197685965311,
      "loss": 1.2767,
      "step": 22081
    },
    {
      "epoch": 1.899853738277553,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006844946362967276,
      "loss": 1.1581,
      "step": 22082
    },
    {
      "epoch": 1.8999397745848747,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006844695034573132,
      "loss": 1.198,
      "step": 22083
    },
    {
      "epoch": 1.9000258108921964,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006844443700783613,
      "loss": 1.3433,
      "step": 22084
    },
    {
      "epoch": 1.9001118471995182,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006844192361599454,
      "loss": 1.0274,
      "step": 22085
    },
    {
      "epoch": 1.90019788350684,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006843941017021392,
      "loss": 1.5358,
      "step": 22086
    },
    {
      "epoch": 1.9002839198141617,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006843689667050161,
      "loss": 2.2441,
      "step": 22087
    },
    {
      "epoch": 1.9003699561214833,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006843438311686497,
      "loss": 1.0598,
      "step": 22088
    },
    {
      "epoch": 1.900455992428805,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006843186950931133,
      "loss": 1.1272,
      "step": 22089
    },
    {
      "epoch": 1.9005420287361265,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006842935584784806,
      "loss": 1.6356,
      "step": 22090
    },
    {
      "epoch": 1.9006280650434484,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006842684213248252,
      "loss": 1.2164,
      "step": 22091
    },
    {
      "epoch": 1.90071410135077,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006842432836322204,
      "loss": 1.2036,
      "step": 22092
    },
    {
      "epoch": 1.9008001376580919,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00068421814540074,
      "loss": 1.8698,
      "step": 22093
    },
    {
      "epoch": 1.9008861739654135,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006841930066304572,
      "loss": 1.5885,
      "step": 22094
    },
    {
      "epoch": 1.900972210272735,
      "grad_norm": 1.125,
      "learning_rate": 0.0006841678673214456,
      "loss": 1.4049,
      "step": 22095
    },
    {
      "epoch": 1.9010582465800567,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006841427274737788,
      "loss": 0.8891,
      "step": 22096
    },
    {
      "epoch": 1.9011442828873784,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006841175870875304,
      "loss": 1.212,
      "step": 22097
    },
    {
      "epoch": 1.9012303191947002,
      "grad_norm": 1.0625,
      "learning_rate": 0.000684092446162774,
      "loss": 1.2322,
      "step": 22098
    },
    {
      "epoch": 1.9013163555020218,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006840673046995828,
      "loss": 1.0235,
      "step": 22099
    },
    {
      "epoch": 1.9014023918093437,
      "grad_norm": 2.09375,
      "learning_rate": 0.0006840421626980307,
      "loss": 1.3423,
      "step": 22100
    },
    {
      "epoch": 1.9014884281166653,
      "grad_norm": 0.84375,
      "learning_rate": 0.000684017020158191,
      "loss": 0.9255,
      "step": 22101
    },
    {
      "epoch": 1.901574464423987,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006839918770801375,
      "loss": 0.9993,
      "step": 22102
    },
    {
      "epoch": 1.9016605007313085,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006839667334639433,
      "loss": 1.1757,
      "step": 22103
    },
    {
      "epoch": 1.9017465370386302,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006839415893096821,
      "loss": 1.2732,
      "step": 22104
    },
    {
      "epoch": 1.901832573345952,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006839164446174277,
      "loss": 1.1859,
      "step": 22105
    },
    {
      "epoch": 1.9019186096532736,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006838912993872534,
      "loss": 1.0916,
      "step": 22106
    },
    {
      "epoch": 1.9020046459605955,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006838661536192329,
      "loss": 0.9889,
      "step": 22107
    },
    {
      "epoch": 1.902090682267917,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006838410073134396,
      "loss": 0.9794,
      "step": 22108
    },
    {
      "epoch": 1.9021767185752387,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006838158604699472,
      "loss": 1.0591,
      "step": 22109
    },
    {
      "epoch": 1.9022627548825604,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000683790713088829,
      "loss": 1.5075,
      "step": 22110
    },
    {
      "epoch": 1.902348791189882,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006837655651701587,
      "loss": 1.4213,
      "step": 22111
    },
    {
      "epoch": 1.9024348274972038,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00068374041671401,
      "loss": 1.3652,
      "step": 22112
    },
    {
      "epoch": 1.9025208638045257,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006837152677204564,
      "loss": 1.6203,
      "step": 22113
    },
    {
      "epoch": 1.9026069001118473,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000683690118189571,
      "loss": 1.2169,
      "step": 22114
    },
    {
      "epoch": 1.902692936419169,
      "grad_norm": 1.046875,
      "learning_rate": 0.000683664968121428,
      "loss": 1.6283,
      "step": 22115
    },
    {
      "epoch": 1.9027789727264905,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006836398175161007,
      "loss": 1.7238,
      "step": 22116
    },
    {
      "epoch": 1.9028650090338122,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006836146663736627,
      "loss": 0.5742,
      "step": 22117
    },
    {
      "epoch": 1.9029510453411338,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006835895146941873,
      "loss": 1.5622,
      "step": 22118
    },
    {
      "epoch": 1.9030370816484556,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006835643624777483,
      "loss": 1.5047,
      "step": 22119
    },
    {
      "epoch": 1.9031231179557775,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006835392097244194,
      "loss": 0.9799,
      "step": 22120
    },
    {
      "epoch": 1.9032091542630991,
      "grad_norm": 1.0625,
      "learning_rate": 0.000683514056434274,
      "loss": 1.4423,
      "step": 22121
    },
    {
      "epoch": 1.9032951905704207,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006834889026073856,
      "loss": 0.8571,
      "step": 22122
    },
    {
      "epoch": 1.9033812268777424,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006834637482438278,
      "loss": 1.4185,
      "step": 22123
    },
    {
      "epoch": 1.903467263185064,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006834385933436744,
      "loss": 0.9708,
      "step": 22124
    },
    {
      "epoch": 1.9035532994923858,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006834134379069988,
      "loss": 1.3118,
      "step": 22125
    },
    {
      "epoch": 1.9036393357997075,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006833882819338743,
      "loss": 1.2238,
      "step": 22126
    },
    {
      "epoch": 1.9037253721070293,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006833631254243749,
      "loss": 1.329,
      "step": 22127
    },
    {
      "epoch": 1.903811408414351,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000683337968378574,
      "loss": 1.4336,
      "step": 22128
    },
    {
      "epoch": 1.9038974447216725,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006833128107965452,
      "loss": 1.1764,
      "step": 22129
    },
    {
      "epoch": 1.9039834810289942,
      "grad_norm": 1.40625,
      "learning_rate": 0.000683287652678362,
      "loss": 2.1734,
      "step": 22130
    },
    {
      "epoch": 1.9040695173363158,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006832624940240982,
      "loss": 1.2319,
      "step": 22131
    },
    {
      "epoch": 1.9041555536436376,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006832373348338272,
      "loss": 0.8645,
      "step": 22132
    },
    {
      "epoch": 1.9042415899509593,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006832121751076226,
      "loss": 1.6016,
      "step": 22133
    },
    {
      "epoch": 1.9043276262582811,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006831870148455579,
      "loss": 0.728,
      "step": 22134
    },
    {
      "epoch": 1.9044136625656027,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006831618540477071,
      "loss": 1.1993,
      "step": 22135
    },
    {
      "epoch": 1.9044996988729244,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006831366927141434,
      "loss": 1.3551,
      "step": 22136
    },
    {
      "epoch": 1.904585735180246,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006831115308449403,
      "loss": 1.4633,
      "step": 22137
    },
    {
      "epoch": 1.9046717714875676,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006830863684401718,
      "loss": 1.4316,
      "step": 22138
    },
    {
      "epoch": 1.9047578077948895,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006830612054999112,
      "loss": 1.5528,
      "step": 22139
    },
    {
      "epoch": 1.904843844102211,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006830360420242323,
      "loss": 1.287,
      "step": 22140
    },
    {
      "epoch": 1.904929880409533,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006830108780132084,
      "loss": 0.8917,
      "step": 22141
    },
    {
      "epoch": 1.9050159167168546,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006829857134669133,
      "loss": 0.8851,
      "step": 22142
    },
    {
      "epoch": 1.9051019530241762,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006829605483854207,
      "loss": 1.6345,
      "step": 22143
    },
    {
      "epoch": 1.9051879893314978,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006829353827688038,
      "loss": 0.87,
      "step": 22144
    },
    {
      "epoch": 1.9052740256388194,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006829102166171368,
      "loss": 1.0563,
      "step": 22145
    },
    {
      "epoch": 1.9053600619461413,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006828850499304929,
      "loss": 1.3028,
      "step": 22146
    },
    {
      "epoch": 1.9054460982534631,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006828598827089458,
      "loss": 1.256,
      "step": 22147
    },
    {
      "epoch": 1.9055321345607847,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006828347149525692,
      "loss": 1.2238,
      "step": 22148
    },
    {
      "epoch": 1.9056181708681064,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006828095466614363,
      "loss": 1.5329,
      "step": 22149
    },
    {
      "epoch": 1.905704207175428,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006827843778356213,
      "loss": 1.5584,
      "step": 22150
    },
    {
      "epoch": 1.9057902434827496,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006827592084751974,
      "loss": 0.6173,
      "step": 22151
    },
    {
      "epoch": 1.9058762797900715,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006827340385802384,
      "loss": 1.0983,
      "step": 22152
    },
    {
      "epoch": 1.905962316097393,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006827088681508179,
      "loss": 0.7542,
      "step": 22153
    },
    {
      "epoch": 1.906048352404715,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006826836971870096,
      "loss": 1.3869,
      "step": 22154
    },
    {
      "epoch": 1.9061343887120366,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006826585256888869,
      "loss": 1.0832,
      "step": 22155
    },
    {
      "epoch": 1.9062204250193582,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006826333536565234,
      "loss": 0.8042,
      "step": 22156
    },
    {
      "epoch": 1.9063064613266798,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006826081810899931,
      "loss": 1.1597,
      "step": 22157
    },
    {
      "epoch": 1.9063924976340014,
      "grad_norm": 2.015625,
      "learning_rate": 0.0006825830079893693,
      "loss": 1.2351,
      "step": 22158
    },
    {
      "epoch": 1.9064785339413233,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006825578343547258,
      "loss": 1.5326,
      "step": 22159
    },
    {
      "epoch": 1.906564570248645,
      "grad_norm": 0.859375,
      "learning_rate": 0.000682532660186136,
      "loss": 0.89,
      "step": 22160
    },
    {
      "epoch": 1.9066506065559667,
      "grad_norm": 1.125,
      "learning_rate": 0.0006825074854836738,
      "loss": 1.8311,
      "step": 22161
    },
    {
      "epoch": 1.9067366428632884,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006824823102474127,
      "loss": 1.9459,
      "step": 22162
    },
    {
      "epoch": 1.90682267917061,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006824571344774262,
      "loss": 1.6217,
      "step": 22163
    },
    {
      "epoch": 1.9069087154779316,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006824319581737881,
      "loss": 1.5285,
      "step": 22164
    },
    {
      "epoch": 1.9069947517852532,
      "grad_norm": 1.328125,
      "learning_rate": 0.000682406781336572,
      "loss": 1.8294,
      "step": 22165
    },
    {
      "epoch": 1.907080788092575,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006823816039658516,
      "loss": 1.8582,
      "step": 22166
    },
    {
      "epoch": 1.9071668243998967,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006823564260617004,
      "loss": 0.8629,
      "step": 22167
    },
    {
      "epoch": 1.9072528607072186,
      "grad_norm": 1.703125,
      "learning_rate": 0.0006823312476241923,
      "loss": 1.87,
      "step": 22168
    },
    {
      "epoch": 1.9073388970145402,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006823060686534006,
      "loss": 1.8637,
      "step": 22169
    },
    {
      "epoch": 1.9074249333218618,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006822808891493991,
      "loss": 0.9126,
      "step": 22170
    },
    {
      "epoch": 1.9075109696291834,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006822557091122614,
      "loss": 2.1666,
      "step": 22171
    },
    {
      "epoch": 1.907597005936505,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006822305285420614,
      "loss": 1.5211,
      "step": 22172
    },
    {
      "epoch": 1.907683042243827,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006822053474388724,
      "loss": 1.4545,
      "step": 22173
    },
    {
      "epoch": 1.9077690785511487,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0006821801658027682,
      "loss": 0.8509,
      "step": 22174
    },
    {
      "epoch": 1.9078551148584704,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006821549836338224,
      "loss": 1.1547,
      "step": 22175
    },
    {
      "epoch": 1.907941151165792,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006821298009321089,
      "loss": 1.2165,
      "step": 22176
    },
    {
      "epoch": 1.9080271874731136,
      "grad_norm": 1.6875,
      "learning_rate": 0.000682104617697701,
      "loss": 1.5829,
      "step": 22177
    },
    {
      "epoch": 1.9081132237804352,
      "grad_norm": 1.578125,
      "learning_rate": 0.0006820794339306725,
      "loss": 1.3741,
      "step": 22178
    },
    {
      "epoch": 1.908199260087757,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006820542496310971,
      "loss": 1.6349,
      "step": 22179
    },
    {
      "epoch": 1.9082852963950787,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006820290647990486,
      "loss": 0.8087,
      "step": 22180
    },
    {
      "epoch": 1.9083713327024006,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006820038794346003,
      "loss": 1.1208,
      "step": 22181
    },
    {
      "epoch": 1.9084573690097222,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006819786935378261,
      "loss": 1.9898,
      "step": 22182
    },
    {
      "epoch": 1.9085434053170438,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006819535071087994,
      "loss": 1.4156,
      "step": 22183
    },
    {
      "epoch": 1.9086294416243654,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006819283201475945,
      "loss": 0.3529,
      "step": 22184
    },
    {
      "epoch": 1.908715477931687,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006819031326542842,
      "loss": 1.1097,
      "step": 22185
    },
    {
      "epoch": 1.908801514239009,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006818779446289429,
      "loss": 1.0521,
      "step": 22186
    },
    {
      "epoch": 1.9088875505463305,
      "grad_norm": 1.359375,
      "learning_rate": 0.000681852756071644,
      "loss": 2.0143,
      "step": 22187
    },
    {
      "epoch": 1.9089735868536524,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000681827566982461,
      "loss": 1.8627,
      "step": 22188
    },
    {
      "epoch": 1.909059623160974,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006818023773614678,
      "loss": 0.9938,
      "step": 22189
    },
    {
      "epoch": 1.9091456594682956,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006817771872087379,
      "loss": 1.117,
      "step": 22190
    },
    {
      "epoch": 1.9092316957756172,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006817519965243453,
      "loss": 1.7406,
      "step": 22191
    },
    {
      "epoch": 1.9093177320829389,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0006817268053083634,
      "loss": 0.6521,
      "step": 22192
    },
    {
      "epoch": 1.9094037683902607,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006817016135608658,
      "loss": 1.1137,
      "step": 22193
    },
    {
      "epoch": 1.9094898046975823,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006816764212819264,
      "loss": 1.3161,
      "step": 22194
    },
    {
      "epoch": 1.9095758410049042,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006816512284716188,
      "loss": 1.0471,
      "step": 22195
    },
    {
      "epoch": 1.9096618773122258,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006816260351300168,
      "loss": 1.0975,
      "step": 22196
    },
    {
      "epoch": 1.9097479136195474,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006816008412571938,
      "loss": 1.1428,
      "step": 22197
    },
    {
      "epoch": 1.909833949926869,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006815756468532237,
      "loss": 1.429,
      "step": 22198
    },
    {
      "epoch": 1.9099199862341907,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006815504519181802,
      "loss": 1.5681,
      "step": 22199
    },
    {
      "epoch": 1.9100060225415125,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006815252564521371,
      "loss": 1.2216,
      "step": 22200
    },
    {
      "epoch": 1.9100920588488344,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006815000604551676,
      "loss": 2.0144,
      "step": 22201
    },
    {
      "epoch": 1.910178095156156,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006814748639273458,
      "loss": 1.5204,
      "step": 22202
    },
    {
      "epoch": 1.9102641314634776,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006814496668687454,
      "loss": 0.9849,
      "step": 22203
    },
    {
      "epoch": 1.9103501677707992,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00068142446927944,
      "loss": 0.4828,
      "step": 22204
    },
    {
      "epoch": 1.9104362040781209,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006813992711595033,
      "loss": 1.4594,
      "step": 22205
    },
    {
      "epoch": 1.9105222403854425,
      "grad_norm": 1.5,
      "learning_rate": 0.000681374072509009,
      "loss": 1.027,
      "step": 22206
    },
    {
      "epoch": 1.9106082766927643,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006813488733280309,
      "loss": 0.887,
      "step": 22207
    },
    {
      "epoch": 1.9106943130000862,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006813236736166423,
      "loss": 0.7568,
      "step": 22208
    },
    {
      "epoch": 1.9107803493074078,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006812984733749176,
      "loss": 0.8207,
      "step": 22209
    },
    {
      "epoch": 1.9108663856147294,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006812732726029301,
      "loss": 0.7767,
      "step": 22210
    },
    {
      "epoch": 1.910952421922051,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006812480713007534,
      "loss": 1.3036,
      "step": 22211
    },
    {
      "epoch": 1.9110384582293727,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006812228694684612,
      "loss": 0.9454,
      "step": 22212
    },
    {
      "epoch": 1.9111244945366945,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006811976671061276,
      "loss": 1.017,
      "step": 22213
    },
    {
      "epoch": 1.9112105308440162,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006811724642138261,
      "loss": 1.749,
      "step": 22214
    },
    {
      "epoch": 1.911296567151338,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006811472607916303,
      "loss": 1.5,
      "step": 22215
    },
    {
      "epoch": 1.9113826034586596,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006811220568396139,
      "loss": 0.9679,
      "step": 22216
    },
    {
      "epoch": 1.9114686397659812,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006810968523578508,
      "loss": 1.0844,
      "step": 22217
    },
    {
      "epoch": 1.9115546760733029,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006810716473464145,
      "loss": 1.11,
      "step": 22218
    },
    {
      "epoch": 1.9116407123806245,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000681046441805379,
      "loss": 1.7795,
      "step": 22219
    },
    {
      "epoch": 1.9117267486879463,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006810212357348178,
      "loss": 1.0317,
      "step": 22220
    },
    {
      "epoch": 1.911812784995268,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006809960291348047,
      "loss": 1.2244,
      "step": 22221
    },
    {
      "epoch": 1.9118988213025898,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0006809708220054135,
      "loss": 1.6516,
      "step": 22222
    },
    {
      "epoch": 1.9119848576099114,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006809456143467177,
      "loss": 1.124,
      "step": 22223
    },
    {
      "epoch": 1.912070893917233,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0006809204061587913,
      "loss": 1.0857,
      "step": 22224
    },
    {
      "epoch": 1.9121569302245547,
      "grad_norm": 1.0,
      "learning_rate": 0.0006808951974417077,
      "loss": 1.6699,
      "step": 22225
    },
    {
      "epoch": 1.9122429665318763,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000680869988195541,
      "loss": 0.8458,
      "step": 22226
    },
    {
      "epoch": 1.9123290028391982,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006808447784203646,
      "loss": 1.0087,
      "step": 22227
    },
    {
      "epoch": 1.9124150391465198,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006808195681162527,
      "loss": 1.2444,
      "step": 22228
    },
    {
      "epoch": 1.9125010754538416,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006807943572832786,
      "loss": 1.0441,
      "step": 22229
    },
    {
      "epoch": 1.9125871117611632,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006807691459215161,
      "loss": 1.7465,
      "step": 22230
    },
    {
      "epoch": 1.9126731480684849,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000680743934031039,
      "loss": 0.9889,
      "step": 22231
    },
    {
      "epoch": 1.9127591843758065,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000680718721611921,
      "loss": 0.7862,
      "step": 22232
    },
    {
      "epoch": 1.9128452206831281,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006806935086642361,
      "loss": 1.4511,
      "step": 22233
    },
    {
      "epoch": 1.91293125699045,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006806682951880576,
      "loss": 1.1603,
      "step": 22234
    },
    {
      "epoch": 1.9130172932977718,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006806430811834596,
      "loss": 1.0053,
      "step": 22235
    },
    {
      "epoch": 1.9131033296050934,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006806178666505156,
      "loss": 1.7737,
      "step": 22236
    },
    {
      "epoch": 1.913189365912415,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006805926515892997,
      "loss": 1.1765,
      "step": 22237
    },
    {
      "epoch": 1.9132754022197367,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006805674359998852,
      "loss": 1.2238,
      "step": 22238
    },
    {
      "epoch": 1.9133614385270583,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006805422198823461,
      "loss": 1.1685,
      "step": 22239
    },
    {
      "epoch": 1.9134474748343802,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006805170032367563,
      "loss": 1.3024,
      "step": 22240
    },
    {
      "epoch": 1.9135335111417018,
      "grad_norm": 1.453125,
      "learning_rate": 0.000680491786063189,
      "loss": 1.6833,
      "step": 22241
    },
    {
      "epoch": 1.9136195474490236,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006804665683617187,
      "loss": 0.9321,
      "step": 22242
    },
    {
      "epoch": 1.9137055837563453,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006804413501324186,
      "loss": 1.2042,
      "step": 22243
    },
    {
      "epoch": 1.9137916200636669,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006804161313753627,
      "loss": 1.2213,
      "step": 22244
    },
    {
      "epoch": 1.9138776563709885,
      "grad_norm": 1.6875,
      "learning_rate": 0.0006803909120906246,
      "loss": 1.0809,
      "step": 22245
    },
    {
      "epoch": 1.9139636926783101,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006803656922782783,
      "loss": 1.171,
      "step": 22246
    },
    {
      "epoch": 1.914049728985632,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006803404719383974,
      "loss": 0.9067,
      "step": 22247
    },
    {
      "epoch": 1.9141357652929536,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006803152510710558,
      "loss": 1.134,
      "step": 22248
    },
    {
      "epoch": 1.9142218016002754,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006802900296763269,
      "loss": 1.2775,
      "step": 22249
    },
    {
      "epoch": 1.914307837907597,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000680264807754285,
      "loss": 1.5275,
      "step": 22250
    },
    {
      "epoch": 1.9143938742149187,
      "grad_norm": 1.5,
      "learning_rate": 0.0006802395853050034,
      "loss": 2.2997,
      "step": 22251
    },
    {
      "epoch": 1.9144799105222403,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0006802143623285561,
      "loss": 0.543,
      "step": 22252
    },
    {
      "epoch": 1.914565946829562,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000680189138825017,
      "loss": 1.3462,
      "step": 22253
    },
    {
      "epoch": 1.9146519831368838,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006801639147944596,
      "loss": 1.5515,
      "step": 22254
    },
    {
      "epoch": 1.9147380194442054,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006801386902369577,
      "loss": 1.7661,
      "step": 22255
    },
    {
      "epoch": 1.9148240557515273,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006801134651525853,
      "loss": 1.0169,
      "step": 22256
    },
    {
      "epoch": 1.9149100920588489,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006800882395414159,
      "loss": 1.4026,
      "step": 22257
    },
    {
      "epoch": 1.9149961283661705,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006800630134035236,
      "loss": 1.3261,
      "step": 22258
    },
    {
      "epoch": 1.9150821646734921,
      "grad_norm": 0.890625,
      "learning_rate": 0.000680037786738982,
      "loss": 1.4327,
      "step": 22259
    },
    {
      "epoch": 1.9151682009808138,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006800125595478648,
      "loss": 1.3939,
      "step": 22260
    },
    {
      "epoch": 1.9152542372881356,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006799873318302459,
      "loss": 0.9894,
      "step": 22261
    },
    {
      "epoch": 1.9153402735954574,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000679962103586199,
      "loss": 0.7541,
      "step": 22262
    },
    {
      "epoch": 1.915426309902779,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006799368748157979,
      "loss": 1.397,
      "step": 22263
    },
    {
      "epoch": 1.9155123462101007,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006799116455191165,
      "loss": 1.2961,
      "step": 22264
    },
    {
      "epoch": 1.9155983825174223,
      "grad_norm": 2.21875,
      "learning_rate": 0.0006798864156962285,
      "loss": 1.3461,
      "step": 22265
    },
    {
      "epoch": 1.915684418824744,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006798611853472079,
      "loss": 1.0303,
      "step": 22266
    },
    {
      "epoch": 1.9157704551320658,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0006798359544721282,
      "loss": 0.7252,
      "step": 22267
    },
    {
      "epoch": 1.9158564914393874,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006798107230710633,
      "loss": 1.5196,
      "step": 22268
    },
    {
      "epoch": 1.9159425277467093,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000679785491144087,
      "loss": 1.2528,
      "step": 22269
    },
    {
      "epoch": 1.9160285640540309,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006797602586912732,
      "loss": 1.8503,
      "step": 22270
    },
    {
      "epoch": 1.9161146003613525,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006797350257126955,
      "loss": 1.4401,
      "step": 22271
    },
    {
      "epoch": 1.9162006366686741,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006797097922084278,
      "loss": 0.7382,
      "step": 22272
    },
    {
      "epoch": 1.9162866729759958,
      "grad_norm": 1.125,
      "learning_rate": 0.0006796845581785439,
      "loss": 1.2706,
      "step": 22273
    },
    {
      "epoch": 1.9163727092833176,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006796593236231177,
      "loss": 2.0321,
      "step": 22274
    },
    {
      "epoch": 1.9164587455906392,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006796340885422227,
      "loss": 1.2203,
      "step": 22275
    },
    {
      "epoch": 1.916544781897961,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006796088529359332,
      "loss": 1.4029,
      "step": 22276
    },
    {
      "epoch": 1.9166308182052827,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006795836168043227,
      "loss": 1.3781,
      "step": 22277
    },
    {
      "epoch": 1.9167168545126043,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000679558380147465,
      "loss": 1.604,
      "step": 22278
    },
    {
      "epoch": 1.916802890819926,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006795331429654339,
      "loss": 0.7754,
      "step": 22279
    },
    {
      "epoch": 1.9168889271272476,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006795079052583032,
      "loss": 0.9084,
      "step": 22280
    },
    {
      "epoch": 1.9169749634345694,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006794826670261469,
      "loss": 0.9741,
      "step": 22281
    },
    {
      "epoch": 1.917060999741891,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006794574282690388,
      "loss": 1.2601,
      "step": 22282
    },
    {
      "epoch": 1.9171470360492129,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006794321889870525,
      "loss": 1.1444,
      "step": 22283
    },
    {
      "epoch": 1.9172330723565345,
      "grad_norm": 1.15625,
      "learning_rate": 0.000679406949180262,
      "loss": 1.7164,
      "step": 22284
    },
    {
      "epoch": 1.9173191086638561,
      "grad_norm": 0.96875,
      "learning_rate": 0.000679381708848741,
      "loss": 0.7947,
      "step": 22285
    },
    {
      "epoch": 1.9174051449711778,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006793564679925635,
      "loss": 1.2975,
      "step": 22286
    },
    {
      "epoch": 1.9174911812784994,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006793312266118031,
      "loss": 1.2675,
      "step": 22287
    },
    {
      "epoch": 1.9175772175858212,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006793059847065338,
      "loss": 1.1011,
      "step": 22288
    },
    {
      "epoch": 1.917663253893143,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006792807422768295,
      "loss": 1.5475,
      "step": 22289
    },
    {
      "epoch": 1.9177492902004647,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006792554993227636,
      "loss": 1.4001,
      "step": 22290
    },
    {
      "epoch": 1.9178353265077863,
      "grad_norm": 1.125,
      "learning_rate": 0.0006792302558444104,
      "loss": 1.3319,
      "step": 22291
    },
    {
      "epoch": 1.917921362815108,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006792050118418436,
      "loss": 0.5125,
      "step": 22292
    },
    {
      "epoch": 1.9180073991224296,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000679179767315137,
      "loss": 1.2543,
      "step": 22293
    },
    {
      "epoch": 1.9180934354297512,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006791545222643642,
      "loss": 2.0543,
      "step": 22294
    },
    {
      "epoch": 1.918179471737073,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006791292766895996,
      "loss": 1.8971,
      "step": 22295
    },
    {
      "epoch": 1.9182655080443949,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006791040305909166,
      "loss": 0.7848,
      "step": 22296
    },
    {
      "epoch": 1.9183515443517165,
      "grad_norm": 0.984375,
      "learning_rate": 0.000679078783968389,
      "loss": 1.4957,
      "step": 22297
    },
    {
      "epoch": 1.9184375806590381,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006790535368220908,
      "loss": 1.5433,
      "step": 22298
    },
    {
      "epoch": 1.9185236169663598,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006790282891520959,
      "loss": 1.2446,
      "step": 22299
    },
    {
      "epoch": 1.9186096532736814,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006790030409584781,
      "loss": 0.8036,
      "step": 22300
    },
    {
      "epoch": 1.9186956895810032,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006789777922413112,
      "loss": 1.5155,
      "step": 22301
    },
    {
      "epoch": 1.9187817258883249,
      "grad_norm": 1.09375,
      "learning_rate": 0.000678952543000669,
      "loss": 1.154,
      "step": 22302
    },
    {
      "epoch": 1.9188677621956467,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0006789272932366255,
      "loss": 1.3248,
      "step": 22303
    },
    {
      "epoch": 1.9189537985029683,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006789020429492545,
      "loss": 0.9148,
      "step": 22304
    },
    {
      "epoch": 1.91903983481029,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006788767921386297,
      "loss": 0.8889,
      "step": 22305
    },
    {
      "epoch": 1.9191258711176116,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000678851540804825,
      "loss": 1.4828,
      "step": 22306
    },
    {
      "epoch": 1.9192119074249332,
      "grad_norm": 0.875,
      "learning_rate": 0.0006788262889479145,
      "loss": 1.2042,
      "step": 22307
    },
    {
      "epoch": 1.919297943732255,
      "grad_norm": 1.34375,
      "learning_rate": 0.000678801036567972,
      "loss": 1.4201,
      "step": 22308
    },
    {
      "epoch": 1.9193839800395767,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006787757836650709,
      "loss": 1.1583,
      "step": 22309
    },
    {
      "epoch": 1.9194700163468985,
      "grad_norm": 1.0,
      "learning_rate": 0.0006787505302392856,
      "loss": 1.346,
      "step": 22310
    },
    {
      "epoch": 1.9195560526542201,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006787252762906899,
      "loss": 0.9894,
      "step": 22311
    },
    {
      "epoch": 1.9196420889615418,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006787000218193574,
      "loss": 1.3142,
      "step": 22312
    },
    {
      "epoch": 1.9197281252688634,
      "grad_norm": 0.84375,
      "learning_rate": 0.000678674766825362,
      "loss": 1.0664,
      "step": 22313
    },
    {
      "epoch": 1.919814161576185,
      "grad_norm": 1.921875,
      "learning_rate": 0.0006786495113087777,
      "loss": 1.6245,
      "step": 22314
    },
    {
      "epoch": 1.9199001978835069,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006786242552696784,
      "loss": 1.7103,
      "step": 22315
    },
    {
      "epoch": 1.9199862341908285,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006785989987081378,
      "loss": 1.2533,
      "step": 22316
    },
    {
      "epoch": 1.9200722704981503,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006785737416242299,
      "loss": 1.0594,
      "step": 22317
    },
    {
      "epoch": 1.920158306805472,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006785484840180285,
      "loss": 1.3585,
      "step": 22318
    },
    {
      "epoch": 1.9202443431127936,
      "grad_norm": 1.609375,
      "learning_rate": 0.0006785232258896077,
      "loss": 2.2951,
      "step": 22319
    },
    {
      "epoch": 1.9203303794201152,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000678497967239041,
      "loss": 1.707,
      "step": 22320
    },
    {
      "epoch": 1.9204164157274368,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006784727080664023,
      "loss": 1.2221,
      "step": 22321
    },
    {
      "epoch": 1.9205024520347587,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006784474483717659,
      "loss": 1.5434,
      "step": 22322
    },
    {
      "epoch": 1.9205884883420805,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006784221881552055,
      "loss": 1.6164,
      "step": 22323
    },
    {
      "epoch": 1.9206745246494021,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006783969274167947,
      "loss": 1.7604,
      "step": 22324
    },
    {
      "epoch": 1.9207605609567238,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006783716661566077,
      "loss": 0.8775,
      "step": 22325
    },
    {
      "epoch": 1.9208465972640454,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006783464043747183,
      "loss": 1.2147,
      "step": 22326
    },
    {
      "epoch": 1.920932633571367,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006783211420712002,
      "loss": 1.0316,
      "step": 22327
    },
    {
      "epoch": 1.9210186698786889,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006782958792461275,
      "loss": 1.7094,
      "step": 22328
    },
    {
      "epoch": 1.9211047061860105,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006782706158995741,
      "loss": 1.1734,
      "step": 22329
    },
    {
      "epoch": 1.9211907424933323,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006782453520316138,
      "loss": 1.2788,
      "step": 22330
    },
    {
      "epoch": 1.921276778800654,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006782200876423204,
      "loss": 1.2876,
      "step": 22331
    },
    {
      "epoch": 1.9213628151079756,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006781948227317679,
      "loss": 0.9645,
      "step": 22332
    },
    {
      "epoch": 1.9214488514152972,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006781695573000304,
      "loss": 1.046,
      "step": 22333
    },
    {
      "epoch": 1.9215348877226188,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006781442913471815,
      "loss": 1.0,
      "step": 22334
    },
    {
      "epoch": 1.9216209240299407,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006781190248732951,
      "loss": 1.5665,
      "step": 22335
    },
    {
      "epoch": 1.9217069603372623,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006780937578784452,
      "loss": 1.3255,
      "step": 22336
    },
    {
      "epoch": 1.9217929966445841,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006780684903627058,
      "loss": 0.9329,
      "step": 22337
    },
    {
      "epoch": 1.9218790329519058,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006780432223261507,
      "loss": 1.1908,
      "step": 22338
    },
    {
      "epoch": 1.9219650692592274,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0006780179537688537,
      "loss": 1.0455,
      "step": 22339
    },
    {
      "epoch": 1.922051105566549,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006779926846908888,
      "loss": 0.9308,
      "step": 22340
    },
    {
      "epoch": 1.9221371418738706,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006779674150923301,
      "loss": 1.6312,
      "step": 22341
    },
    {
      "epoch": 1.9222231781811925,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006779421449732512,
      "loss": 1.0723,
      "step": 22342
    },
    {
      "epoch": 1.922309214488514,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006779168743337261,
      "loss": 1.3394,
      "step": 22343
    },
    {
      "epoch": 1.922395250795836,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006778916031738288,
      "loss": 1.2234,
      "step": 22344
    },
    {
      "epoch": 1.9224812871031576,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006778663314936332,
      "loss": 0.8947,
      "step": 22345
    },
    {
      "epoch": 1.9225673234104792,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000677841059293213,
      "loss": 1.3759,
      "step": 22346
    },
    {
      "epoch": 1.9226533597178008,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006778157865726424,
      "loss": 1.1706,
      "step": 22347
    },
    {
      "epoch": 1.9227393960251224,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006777905133319952,
      "loss": 1.6186,
      "step": 22348
    },
    {
      "epoch": 1.9228254323324443,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006777652395713455,
      "loss": 0.8078,
      "step": 22349
    },
    {
      "epoch": 1.9229114686397661,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006777399652907667,
      "loss": 1.5647,
      "step": 22350
    },
    {
      "epoch": 1.9229975049470878,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006777146904903334,
      "loss": 1.6776,
      "step": 22351
    },
    {
      "epoch": 1.9230835412544094,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006776894151701191,
      "loss": 1.7369,
      "step": 22352
    },
    {
      "epoch": 1.923169577561731,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006776641393301978,
      "loss": 1.1828,
      "step": 22353
    },
    {
      "epoch": 1.9232556138690526,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006776388629706432,
      "loss": 0.8021,
      "step": 22354
    },
    {
      "epoch": 1.9233416501763745,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0006776135860915298,
      "loss": 0.8469,
      "step": 22355
    },
    {
      "epoch": 1.923427686483696,
      "grad_norm": 1.375,
      "learning_rate": 0.000677588308692931,
      "loss": 1.952,
      "step": 22356
    },
    {
      "epoch": 1.923513722791018,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006775630307749212,
      "loss": 2.0212,
      "step": 22357
    },
    {
      "epoch": 1.9235997590983396,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006775377523375737,
      "loss": 1.3588,
      "step": 22358
    },
    {
      "epoch": 1.9236857954056612,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006775124733809631,
      "loss": 1.1364,
      "step": 22359
    },
    {
      "epoch": 1.9237718317129828,
      "grad_norm": 1.046875,
      "learning_rate": 0.000677487193905163,
      "loss": 1.2729,
      "step": 22360
    },
    {
      "epoch": 1.9238578680203045,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006774619139102473,
      "loss": 1.4566,
      "step": 22361
    },
    {
      "epoch": 1.9239439043276263,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006774366333962901,
      "loss": 1.2186,
      "step": 22362
    },
    {
      "epoch": 1.924029940634948,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006774113523633651,
      "loss": 1.0588,
      "step": 22363
    },
    {
      "epoch": 1.9241159769422698,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006773860708115466,
      "loss": 1.0298,
      "step": 22364
    },
    {
      "epoch": 1.9242020132495914,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006773607887409081,
      "loss": 0.9407,
      "step": 22365
    },
    {
      "epoch": 1.924288049556913,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006773355061515241,
      "loss": 1.5814,
      "step": 22366
    },
    {
      "epoch": 1.9243740858642346,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000677310223043468,
      "loss": 0.8304,
      "step": 22367
    },
    {
      "epoch": 1.9244601221715563,
      "grad_norm": 1.0,
      "learning_rate": 0.0006772849394168141,
      "loss": 1.1611,
      "step": 22368
    },
    {
      "epoch": 1.924546158478878,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006772596552716361,
      "loss": 0.9852,
      "step": 22369
    },
    {
      "epoch": 1.9246321947861997,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006772343706080082,
      "loss": 0.649,
      "step": 22370
    },
    {
      "epoch": 1.9247182310935216,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006772090854260043,
      "loss": 1.308,
      "step": 22371
    },
    {
      "epoch": 1.9248042674008432,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000677183799725698,
      "loss": 1.8591,
      "step": 22372
    },
    {
      "epoch": 1.9248903037081648,
      "grad_norm": 1.125,
      "learning_rate": 0.0006771585135071637,
      "loss": 1.2724,
      "step": 22373
    },
    {
      "epoch": 1.9249763400154865,
      "grad_norm": 1.0,
      "learning_rate": 0.0006771332267704752,
      "loss": 0.9411,
      "step": 22374
    },
    {
      "epoch": 1.925062376322808,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006771079395157066,
      "loss": 1.1478,
      "step": 22375
    },
    {
      "epoch": 1.92514841263013,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006770826517429316,
      "loss": 0.8642,
      "step": 22376
    },
    {
      "epoch": 1.9252344489374518,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006770573634522244,
      "loss": 1.4681,
      "step": 22377
    },
    {
      "epoch": 1.9253204852447734,
      "grad_norm": 1.8125,
      "learning_rate": 0.0006770320746436587,
      "loss": 1.1658,
      "step": 22378
    },
    {
      "epoch": 1.925406521552095,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006770067853173086,
      "loss": 1.4762,
      "step": 22379
    },
    {
      "epoch": 1.9254925578594166,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006769814954732482,
      "loss": 1.2468,
      "step": 22380
    },
    {
      "epoch": 1.9255785941667383,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006769562051115512,
      "loss": 1.7491,
      "step": 22381
    },
    {
      "epoch": 1.92566463047406,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006769309142322919,
      "loss": 0.812,
      "step": 22382
    },
    {
      "epoch": 1.9257506667813817,
      "grad_norm": 1.015625,
      "learning_rate": 0.000676905622835544,
      "loss": 0.8903,
      "step": 22383
    },
    {
      "epoch": 1.9258367030887036,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006768803309213816,
      "loss": 0.8165,
      "step": 22384
    },
    {
      "epoch": 1.9259227393960252,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006768550384898785,
      "loss": 1.1391,
      "step": 22385
    },
    {
      "epoch": 1.9260087757033468,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000676829745541109,
      "loss": 0.8479,
      "step": 22386
    },
    {
      "epoch": 1.9260948120106685,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006768044520751467,
      "loss": 0.6973,
      "step": 22387
    },
    {
      "epoch": 1.92618084831799,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006767791580920659,
      "loss": 0.699,
      "step": 22388
    },
    {
      "epoch": 1.926266884625312,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006767538635919404,
      "loss": 1.5356,
      "step": 22389
    },
    {
      "epoch": 1.9263529209326336,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006767285685748442,
      "loss": 1.7722,
      "step": 22390
    },
    {
      "epoch": 1.9264389572399554,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006767032730408513,
      "loss": 1.2352,
      "step": 22391
    },
    {
      "epoch": 1.926524993547277,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006766779769900356,
      "loss": 0.5897,
      "step": 22392
    },
    {
      "epoch": 1.9266110298545986,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006766526804224713,
      "loss": 1.0593,
      "step": 22393
    },
    {
      "epoch": 1.9266970661619203,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006766273833382324,
      "loss": 2.0347,
      "step": 22394
    },
    {
      "epoch": 1.926783102469242,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006766020857373924,
      "loss": 1.08,
      "step": 22395
    },
    {
      "epoch": 1.9268691387765637,
      "grad_norm": 1.0625,
      "learning_rate": 0.000676576787620026,
      "loss": 1.019,
      "step": 22396
    },
    {
      "epoch": 1.9269551750838854,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006765514889862067,
      "loss": 1.8562,
      "step": 22397
    },
    {
      "epoch": 1.9270412113912072,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006765261898360087,
      "loss": 2.1607,
      "step": 22398
    },
    {
      "epoch": 1.9271272476985288,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006765008901695057,
      "loss": 1.2707,
      "step": 22399
    },
    {
      "epoch": 1.9272132840058505,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006764755899867723,
      "loss": 1.1073,
      "step": 22400
    },
    {
      "epoch": 1.927299320313172,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006764502892878818,
      "loss": 1.6738,
      "step": 22401
    },
    {
      "epoch": 1.9273853566204937,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006764249880729087,
      "loss": 2.0276,
      "step": 22402
    },
    {
      "epoch": 1.9274713929278156,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006763996863419267,
      "loss": 1.0114,
      "step": 22403
    },
    {
      "epoch": 1.9275574292351372,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00067637438409501,
      "loss": 0.9902,
      "step": 22404
    },
    {
      "epoch": 1.927643465542459,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006763490813322326,
      "loss": 1.6172,
      "step": 22405
    },
    {
      "epoch": 1.9277295018497806,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006763237780536682,
      "loss": 1.0431,
      "step": 22406
    },
    {
      "epoch": 1.9278155381571023,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006762984742593913,
      "loss": 0.7277,
      "step": 22407
    },
    {
      "epoch": 1.927901574464424,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006762731699494755,
      "loss": 0.7346,
      "step": 22408
    },
    {
      "epoch": 1.9279876107717455,
      "grad_norm": 1.25,
      "learning_rate": 0.0006762478651239951,
      "loss": 1.3768,
      "step": 22409
    },
    {
      "epoch": 1.9280736470790674,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006762225597830236,
      "loss": 1.3137,
      "step": 22410
    },
    {
      "epoch": 1.9281596833863892,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006761972539266357,
      "loss": 0.8135,
      "step": 22411
    },
    {
      "epoch": 1.9282457196937108,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000676171947554905,
      "loss": 1.1533,
      "step": 22412
    },
    {
      "epoch": 1.9283317560010325,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006761466406679057,
      "loss": 0.54,
      "step": 22413
    },
    {
      "epoch": 1.928417792308354,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006761213332657118,
      "loss": 0.9312,
      "step": 22414
    },
    {
      "epoch": 1.9285038286156757,
      "grad_norm": 3.09375,
      "learning_rate": 0.0006760960253483972,
      "loss": 1.6693,
      "step": 22415
    },
    {
      "epoch": 1.9285898649229976,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000676070716916036,
      "loss": 1.3203,
      "step": 22416
    },
    {
      "epoch": 1.9286759012303192,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006760454079687019,
      "loss": 0.8102,
      "step": 22417
    },
    {
      "epoch": 1.928761937537641,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006760200985064696,
      "loss": 0.6798,
      "step": 22418
    },
    {
      "epoch": 1.9288479738449626,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006759947885294126,
      "loss": 1.8131,
      "step": 22419
    },
    {
      "epoch": 1.9289340101522843,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006759694780376052,
      "loss": 1.5516,
      "step": 22420
    },
    {
      "epoch": 1.929020046459606,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006759441670311211,
      "loss": 1.7245,
      "step": 22421
    },
    {
      "epoch": 1.9291060827669275,
      "grad_norm": 1.5,
      "learning_rate": 0.0006759188555100346,
      "loss": 2.2424,
      "step": 22422
    },
    {
      "epoch": 1.9291921190742494,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006758935434744197,
      "loss": 1.017,
      "step": 22423
    },
    {
      "epoch": 1.929278155381571,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0006758682309243503,
      "loss": 1.6937,
      "step": 22424
    },
    {
      "epoch": 1.9293641916888928,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006758429178599006,
      "loss": 0.8248,
      "step": 22425
    },
    {
      "epoch": 1.9294502279962145,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006758176042811445,
      "loss": 1.0311,
      "step": 22426
    },
    {
      "epoch": 1.929536264303536,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006757922901881562,
      "loss": 1.1687,
      "step": 22427
    },
    {
      "epoch": 1.9296223006108577,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006757669755810096,
      "loss": 1.5489,
      "step": 22428
    },
    {
      "epoch": 1.9297083369181793,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006757416604597787,
      "loss": 1.5511,
      "step": 22429
    },
    {
      "epoch": 1.9297943732255012,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006757163448245377,
      "loss": 0.6041,
      "step": 22430
    },
    {
      "epoch": 1.9298804095328228,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006756910286753606,
      "loss": 1.6583,
      "step": 22431
    },
    {
      "epoch": 1.9299664458401447,
      "grad_norm": 1.125,
      "learning_rate": 0.0006756657120123214,
      "loss": 1.5905,
      "step": 22432
    },
    {
      "epoch": 1.9300524821474663,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006756403948354941,
      "loss": 0.9179,
      "step": 22433
    },
    {
      "epoch": 1.930138518454788,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006756150771449528,
      "loss": 0.9342,
      "step": 22434
    },
    {
      "epoch": 1.9302245547621095,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006755897589407717,
      "loss": 1.3785,
      "step": 22435
    },
    {
      "epoch": 1.9303105910694311,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006755644402230245,
      "loss": 1.2451,
      "step": 22436
    },
    {
      "epoch": 1.930396627376753,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0006755391209917856,
      "loss": 1.1449,
      "step": 22437
    },
    {
      "epoch": 1.9304826636840748,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000675513801247129,
      "loss": 0.9754,
      "step": 22438
    },
    {
      "epoch": 1.9305686999913965,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006754884809891285,
      "loss": 1.3925,
      "step": 22439
    },
    {
      "epoch": 1.930654736298718,
      "grad_norm": 1.25,
      "learning_rate": 0.0006754631602178582,
      "loss": 1.2582,
      "step": 22440
    },
    {
      "epoch": 1.9307407726060397,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006754378389333926,
      "loss": 2.2142,
      "step": 22441
    },
    {
      "epoch": 1.9308268089133613,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006754125171358053,
      "loss": 2.0371,
      "step": 22442
    },
    {
      "epoch": 1.9309128452206832,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006753871948251704,
      "loss": 1.4047,
      "step": 22443
    },
    {
      "epoch": 1.9309988815280048,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006753618720015622,
      "loss": 1.2127,
      "step": 22444
    },
    {
      "epoch": 1.9310849178353267,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006753365486650547,
      "loss": 1.2279,
      "step": 22445
    },
    {
      "epoch": 1.9311709541426483,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006753112248157218,
      "loss": 1.5639,
      "step": 22446
    },
    {
      "epoch": 1.93125699044997,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0006752859004536376,
      "loss": 0.9089,
      "step": 22447
    },
    {
      "epoch": 1.9313430267572915,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006752605755788761,
      "loss": 1.0533,
      "step": 22448
    },
    {
      "epoch": 1.9314290630646131,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006752352501915118,
      "loss": 0.8946,
      "step": 22449
    },
    {
      "epoch": 1.931515099371935,
      "grad_norm": 1.90625,
      "learning_rate": 0.0006752099242916184,
      "loss": 1.7352,
      "step": 22450
    },
    {
      "epoch": 1.9316011356792566,
      "grad_norm": 1.125,
      "learning_rate": 0.0006751845978792699,
      "loss": 0.7785,
      "step": 22451
    },
    {
      "epoch": 1.9316871719865785,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006751592709545406,
      "loss": 1.7969,
      "step": 22452
    },
    {
      "epoch": 1.9317732082939,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006751339435175045,
      "loss": 0.7824,
      "step": 22453
    },
    {
      "epoch": 1.9318592446012217,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006751086155682358,
      "loss": 0.6747,
      "step": 22454
    },
    {
      "epoch": 1.9319452809085433,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006750832871068083,
      "loss": 1.9288,
      "step": 22455
    },
    {
      "epoch": 1.932031317215865,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006750579581332963,
      "loss": 2.2939,
      "step": 22456
    },
    {
      "epoch": 1.9321173535231868,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006750326286477739,
      "loss": 1.4135,
      "step": 22457
    },
    {
      "epoch": 1.9322033898305084,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006750072986503149,
      "loss": 1.5602,
      "step": 22458
    },
    {
      "epoch": 1.9322894261378303,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006749819681409938,
      "loss": 1.7023,
      "step": 22459
    },
    {
      "epoch": 1.932375462445152,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006749566371198844,
      "loss": 0.9863,
      "step": 22460
    },
    {
      "epoch": 1.9324614987524735,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006749313055870608,
      "loss": 0.8301,
      "step": 22461
    },
    {
      "epoch": 1.9325475350597952,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006749059735425972,
      "loss": 1.9309,
      "step": 22462
    },
    {
      "epoch": 1.9326335713671168,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006748806409865676,
      "loss": 1.8864,
      "step": 22463
    },
    {
      "epoch": 1.9327196076744386,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006748553079190463,
      "loss": 1.3024,
      "step": 22464
    },
    {
      "epoch": 1.9328056439817605,
      "grad_norm": 1.25,
      "learning_rate": 0.0006748299743401069,
      "loss": 1.4762,
      "step": 22465
    },
    {
      "epoch": 1.932891680289082,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006748046402498241,
      "loss": 1.862,
      "step": 22466
    },
    {
      "epoch": 1.9329777165964037,
      "grad_norm": 0.75,
      "learning_rate": 0.0006747793056482717,
      "loss": 1.2702,
      "step": 22467
    },
    {
      "epoch": 1.9330637529037253,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006747539705355237,
      "loss": 0.6383,
      "step": 22468
    },
    {
      "epoch": 1.933149789211047,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006747286349116546,
      "loss": 1.367,
      "step": 22469
    },
    {
      "epoch": 1.9332358255183686,
      "grad_norm": 0.875,
      "learning_rate": 0.0006747032987767379,
      "loss": 0.7233,
      "step": 22470
    },
    {
      "epoch": 1.9333218618256904,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006746779621308482,
      "loss": 1.4083,
      "step": 22471
    },
    {
      "epoch": 1.9334078981330123,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006746526249740595,
      "loss": 1.291,
      "step": 22472
    },
    {
      "epoch": 1.933493934440334,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006746272873064456,
      "loss": 1.6533,
      "step": 22473
    },
    {
      "epoch": 1.9335799707476555,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006746019491280811,
      "loss": 1.4159,
      "step": 22474
    },
    {
      "epoch": 1.9336660070549772,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006745766104390397,
      "loss": 1.7711,
      "step": 22475
    },
    {
      "epoch": 1.9337520433622988,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006745512712393958,
      "loss": 1.2725,
      "step": 22476
    },
    {
      "epoch": 1.9338380796696206,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006745259315292232,
      "loss": 2.1676,
      "step": 22477
    },
    {
      "epoch": 1.9339241159769422,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006745005913085963,
      "loss": 1.4423,
      "step": 22478
    },
    {
      "epoch": 1.934010152284264,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006744752505775892,
      "loss": 0.8335,
      "step": 22479
    },
    {
      "epoch": 1.9340961885915857,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006744499093362758,
      "loss": 1.4722,
      "step": 22480
    },
    {
      "epoch": 1.9341822248989073,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006744245675847303,
      "loss": 1.0269,
      "step": 22481
    },
    {
      "epoch": 1.934268261206229,
      "grad_norm": 1.625,
      "learning_rate": 0.000674399225323027,
      "loss": 1.3413,
      "step": 22482
    },
    {
      "epoch": 1.9343542975135506,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006743738825512398,
      "loss": 1.1682,
      "step": 22483
    },
    {
      "epoch": 1.9344403338208724,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006743485392694429,
      "loss": 0.9654,
      "step": 22484
    },
    {
      "epoch": 1.934526370128194,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006743231954777104,
      "loss": 2.0634,
      "step": 22485
    },
    {
      "epoch": 1.934612406435516,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006742978511761164,
      "loss": 1.7201,
      "step": 22486
    },
    {
      "epoch": 1.9346984427428375,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006742725063647353,
      "loss": 1.5055,
      "step": 22487
    },
    {
      "epoch": 1.9347844790501592,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006742471610436409,
      "loss": 1.1353,
      "step": 22488
    },
    {
      "epoch": 1.9348705153574808,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006742218152129074,
      "loss": 1.5958,
      "step": 22489
    },
    {
      "epoch": 1.9349565516648024,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006741964688726091,
      "loss": 0.648,
      "step": 22490
    },
    {
      "epoch": 1.9350425879721243,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006741711220228199,
      "loss": 1.6074,
      "step": 22491
    },
    {
      "epoch": 1.9351286242794459,
      "grad_norm": 0.875,
      "learning_rate": 0.0006741457746636141,
      "loss": 0.9119,
      "step": 22492
    },
    {
      "epoch": 1.9352146605867677,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006741204267950658,
      "loss": 1.2625,
      "step": 22493
    },
    {
      "epoch": 1.9353006968940893,
      "grad_norm": 1.140625,
      "learning_rate": 0.000674095078417249,
      "loss": 1.2781,
      "step": 22494
    },
    {
      "epoch": 1.935386733201411,
      "grad_norm": 1.28125,
      "learning_rate": 0.000674069729530238,
      "loss": 0.9963,
      "step": 22495
    },
    {
      "epoch": 1.9354727695087326,
      "grad_norm": 1.0,
      "learning_rate": 0.000674044380134107,
      "loss": 1.309,
      "step": 22496
    },
    {
      "epoch": 1.9355588058160542,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006740190302289299,
      "loss": 1.5724,
      "step": 22497
    },
    {
      "epoch": 1.935644842123376,
      "grad_norm": 1.734375,
      "learning_rate": 0.0006739936798147811,
      "loss": 2.299,
      "step": 22498
    },
    {
      "epoch": 1.935730878430698,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006739683288917345,
      "loss": 0.7786,
      "step": 22499
    },
    {
      "epoch": 1.9358169147380195,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006739429774598644,
      "loss": 1.2781,
      "step": 22500
    },
    {
      "epoch": 1.9359029510453412,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000673917625519245,
      "loss": 1.0347,
      "step": 22501
    },
    {
      "epoch": 1.9359889873526628,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006738922730699504,
      "loss": 0.9487,
      "step": 22502
    },
    {
      "epoch": 1.9360750236599844,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006738669201120547,
      "loss": 1.6316,
      "step": 22503
    },
    {
      "epoch": 1.9361610599673063,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006738415666456319,
      "loss": 1.187,
      "step": 22504
    },
    {
      "epoch": 1.9362470962746279,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006738162126707566,
      "loss": 1.4542,
      "step": 22505
    },
    {
      "epoch": 1.9363331325819497,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006737908581875026,
      "loss": 1.0443,
      "step": 22506
    },
    {
      "epoch": 1.9364191688892713,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006737655031959439,
      "loss": 1.5455,
      "step": 22507
    },
    {
      "epoch": 1.936505205196593,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006737401476961551,
      "loss": 0.9365,
      "step": 22508
    },
    {
      "epoch": 1.9365912415039146,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006737147916882102,
      "loss": 1.445,
      "step": 22509
    },
    {
      "epoch": 1.9366772778112362,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006736894351721832,
      "loss": 1.1537,
      "step": 22510
    },
    {
      "epoch": 1.936763314118558,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006736640781481485,
      "loss": 1.6983,
      "step": 22511
    },
    {
      "epoch": 1.9368493504258797,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006736387206161801,
      "loss": 0.6357,
      "step": 22512
    },
    {
      "epoch": 1.9369353867332015,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006736133625763521,
      "loss": 1.1941,
      "step": 22513
    },
    {
      "epoch": 1.9370214230405232,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006735880040287389,
      "loss": 1.5542,
      "step": 22514
    },
    {
      "epoch": 1.9371074593478448,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006735626449734145,
      "loss": 1.1392,
      "step": 22515
    },
    {
      "epoch": 1.9371934956551664,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006735372854104532,
      "loss": 1.7061,
      "step": 22516
    },
    {
      "epoch": 1.937279531962488,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006735119253399291,
      "loss": 0.9761,
      "step": 22517
    },
    {
      "epoch": 1.9373655682698099,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006734865647619162,
      "loss": 0.7115,
      "step": 22518
    },
    {
      "epoch": 1.9374516045771315,
      "grad_norm": 1.15625,
      "learning_rate": 0.000673461203676489,
      "loss": 1.4997,
      "step": 22519
    },
    {
      "epoch": 1.9375376408844533,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006734358420837214,
      "loss": 1.419,
      "step": 22520
    },
    {
      "epoch": 1.937623677191775,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006734104799836877,
      "loss": 1.4383,
      "step": 22521
    },
    {
      "epoch": 1.9377097134990966,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006733851173764622,
      "loss": 1.074,
      "step": 22522
    },
    {
      "epoch": 1.9377957498064182,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006733597542621188,
      "loss": 1.9039,
      "step": 22523
    },
    {
      "epoch": 1.9378817861137398,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006733343906407319,
      "loss": 1.2065,
      "step": 22524
    },
    {
      "epoch": 1.9379678224210617,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006733090265123757,
      "loss": 1.6941,
      "step": 22525
    },
    {
      "epoch": 1.9380538587283835,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006732836618771242,
      "loss": 1.5642,
      "step": 22526
    },
    {
      "epoch": 1.9381398950357052,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006732582967350518,
      "loss": 0.8583,
      "step": 22527
    },
    {
      "epoch": 1.9382259313430268,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006732329310862325,
      "loss": 0.9836,
      "step": 22528
    },
    {
      "epoch": 1.9383119676503484,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006732075649307405,
      "loss": 0.8155,
      "step": 22529
    },
    {
      "epoch": 1.93839800395767,
      "grad_norm": 1.515625,
      "learning_rate": 0.0006731821982686501,
      "loss": 1.7113,
      "step": 22530
    },
    {
      "epoch": 1.9384840402649919,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006731568311000355,
      "loss": 1.4348,
      "step": 22531
    },
    {
      "epoch": 1.9385700765723135,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000673131463424971,
      "loss": 1.3326,
      "step": 22532
    },
    {
      "epoch": 1.9386561128796354,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006731060952435304,
      "loss": 1.862,
      "step": 22533
    },
    {
      "epoch": 1.938742149186957,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006730807265557882,
      "loss": 1.8637,
      "step": 22534
    },
    {
      "epoch": 1.9388281854942786,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006730553573618185,
      "loss": 1.4942,
      "step": 22535
    },
    {
      "epoch": 1.9389142218016002,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006730299876616956,
      "loss": 0.9841,
      "step": 22536
    },
    {
      "epoch": 1.9390002581089218,
      "grad_norm": 1.375,
      "learning_rate": 0.0006730046174554937,
      "loss": 1.1901,
      "step": 22537
    },
    {
      "epoch": 1.9390862944162437,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006729792467432869,
      "loss": 0.8732,
      "step": 22538
    },
    {
      "epoch": 1.9391723307235653,
      "grad_norm": 0.875,
      "learning_rate": 0.0006729538755251493,
      "loss": 0.7704,
      "step": 22539
    },
    {
      "epoch": 1.9392583670308872,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006729285038011555,
      "loss": 1.2163,
      "step": 22540
    },
    {
      "epoch": 1.9393444033382088,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006729031315713792,
      "loss": 1.5104,
      "step": 22541
    },
    {
      "epoch": 1.9394304396455304,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000672877758835895,
      "loss": 1.7363,
      "step": 22542
    },
    {
      "epoch": 1.939516475952852,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006728523855947771,
      "loss": 1.2718,
      "step": 22543
    },
    {
      "epoch": 1.9396025122601737,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006728270118480994,
      "loss": 1.5396,
      "step": 22544
    },
    {
      "epoch": 1.9396885485674955,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006728016375959362,
      "loss": 1.4492,
      "step": 22545
    },
    {
      "epoch": 1.9397745848748171,
      "grad_norm": 1.015625,
      "learning_rate": 0.000672776262838362,
      "loss": 1.0238,
      "step": 22546
    },
    {
      "epoch": 1.939860621182139,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006727508875754509,
      "loss": 1.4525,
      "step": 22547
    },
    {
      "epoch": 1.9399466574894606,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006727255118072769,
      "loss": 2.1294,
      "step": 22548
    },
    {
      "epoch": 1.9400326937967822,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006727001355339145,
      "loss": 1.1141,
      "step": 22549
    },
    {
      "epoch": 1.9401187301041039,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006726747587554378,
      "loss": 1.131,
      "step": 22550
    },
    {
      "epoch": 1.9402047664114255,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000672649381471921,
      "loss": 1.436,
      "step": 22551
    },
    {
      "epoch": 1.9402908027187473,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006726240036834381,
      "loss": 1.3627,
      "step": 22552
    },
    {
      "epoch": 1.9403768390260692,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006725986253900638,
      "loss": 1.4852,
      "step": 22553
    },
    {
      "epoch": 1.9404628753333908,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006725732465918721,
      "loss": 1.207,
      "step": 22554
    },
    {
      "epoch": 1.9405489116407124,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006725478672889372,
      "loss": 1.1692,
      "step": 22555
    },
    {
      "epoch": 1.940634947948034,
      "grad_norm": 2.078125,
      "learning_rate": 0.0006725224874813331,
      "loss": 1.5081,
      "step": 22556
    },
    {
      "epoch": 1.9407209842553557,
      "grad_norm": 1.125,
      "learning_rate": 0.0006724971071691346,
      "loss": 0.9338,
      "step": 22557
    },
    {
      "epoch": 1.9408070205626773,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006724717263524154,
      "loss": 1.318,
      "step": 22558
    },
    {
      "epoch": 1.9408930568699991,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006724463450312501,
      "loss": 1.6337,
      "step": 22559
    },
    {
      "epoch": 1.940979093177321,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006724209632057126,
      "loss": 1.3323,
      "step": 22560
    },
    {
      "epoch": 1.9410651294846426,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006723955808758775,
      "loss": 1.032,
      "step": 22561
    },
    {
      "epoch": 1.9411511657919642,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006723701980418189,
      "loss": 1.3061,
      "step": 22562
    },
    {
      "epoch": 1.9412372020992859,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006723448147036107,
      "loss": 1.7465,
      "step": 22563
    },
    {
      "epoch": 1.9413232384066075,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006723194308613276,
      "loss": 0.7211,
      "step": 22564
    },
    {
      "epoch": 1.9414092747139293,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006722940465150438,
      "loss": 1.4968,
      "step": 22565
    },
    {
      "epoch": 1.941495311021251,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006722686616648333,
      "loss": 0.8771,
      "step": 22566
    },
    {
      "epoch": 1.9415813473285728,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006722432763107704,
      "loss": 1.6285,
      "step": 22567
    },
    {
      "epoch": 1.9416673836358944,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006722178904529296,
      "loss": 0.8839,
      "step": 22568
    },
    {
      "epoch": 1.941753419943216,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006721925040913849,
      "loss": 1.6632,
      "step": 22569
    },
    {
      "epoch": 1.9418394562505377,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006721671172262105,
      "loss": 1.2155,
      "step": 22570
    },
    {
      "epoch": 1.9419254925578593,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006721417298574809,
      "loss": 1.8173,
      "step": 22571
    },
    {
      "epoch": 1.9420115288651811,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006721163419852701,
      "loss": 1.4342,
      "step": 22572
    },
    {
      "epoch": 1.9420975651725028,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006720909536096527,
      "loss": 0.9602,
      "step": 22573
    },
    {
      "epoch": 1.9421836014798246,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006720655647307024,
      "loss": 2.121,
      "step": 22574
    },
    {
      "epoch": 1.9422696377871462,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000672040175348494,
      "loss": 0.6876,
      "step": 22575
    },
    {
      "epoch": 1.9423556740944679,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006720147854631015,
      "loss": 1.4951,
      "step": 22576
    },
    {
      "epoch": 1.9424417104017895,
      "grad_norm": 1.25,
      "learning_rate": 0.0006719893950745992,
      "loss": 1.927,
      "step": 22577
    },
    {
      "epoch": 1.942527746709111,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0006719640041830614,
      "loss": 1.073,
      "step": 22578
    },
    {
      "epoch": 1.942613783016433,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006719386127885623,
      "loss": 0.991,
      "step": 22579
    },
    {
      "epoch": 1.9426998193237546,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006719132208911762,
      "loss": 0.7631,
      "step": 22580
    },
    {
      "epoch": 1.9427858556310764,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006718878284909776,
      "loss": 1.1542,
      "step": 22581
    },
    {
      "epoch": 1.942871891938398,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006718624355880402,
      "loss": 1.8676,
      "step": 22582
    },
    {
      "epoch": 1.9429579282457197,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006718370421824386,
      "loss": 0.9874,
      "step": 22583
    },
    {
      "epoch": 1.9430439645530413,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006718116482742473,
      "loss": 1.0813,
      "step": 22584
    },
    {
      "epoch": 1.943130000860363,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006717862538635402,
      "loss": 1.1549,
      "step": 22585
    },
    {
      "epoch": 1.9432160371676848,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006717608589503917,
      "loss": 1.3386,
      "step": 22586
    },
    {
      "epoch": 1.9433020734750066,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006717354635348762,
      "loss": 1.5475,
      "step": 22587
    },
    {
      "epoch": 1.9433881097823282,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006717100676170679,
      "loss": 1.1381,
      "step": 22588
    },
    {
      "epoch": 1.9434741460896499,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006716846711970408,
      "loss": 1.0947,
      "step": 22589
    },
    {
      "epoch": 1.9435601823969715,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006716592742748698,
      "loss": 1.6195,
      "step": 22590
    },
    {
      "epoch": 1.943646218704293,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006716338768506285,
      "loss": 1.7982,
      "step": 22591
    },
    {
      "epoch": 1.943732255011615,
      "grad_norm": 1.0,
      "learning_rate": 0.0006716084789243915,
      "loss": 0.825,
      "step": 22592
    },
    {
      "epoch": 1.9438182913189366,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006715830804962332,
      "loss": 1.1701,
      "step": 22593
    },
    {
      "epoch": 1.9439043276262584,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006715576815662276,
      "loss": 0.6553,
      "step": 22594
    },
    {
      "epoch": 1.94399036393358,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006715322821344493,
      "loss": 1.4806,
      "step": 22595
    },
    {
      "epoch": 1.9440764002409017,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006715068822009724,
      "loss": 1.1861,
      "step": 22596
    },
    {
      "epoch": 1.9441624365482233,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006714814817658712,
      "loss": 1.1824,
      "step": 22597
    },
    {
      "epoch": 1.944248472855545,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006714560808292201,
      "loss": 1.4413,
      "step": 22598
    },
    {
      "epoch": 1.9443345091628668,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006714306793910934,
      "loss": 1.2246,
      "step": 22599
    },
    {
      "epoch": 1.9444205454701884,
      "grad_norm": 9.0625,
      "learning_rate": 0.0006714052774515651,
      "loss": 0.9941,
      "step": 22600
    },
    {
      "epoch": 1.9445065817775102,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006713798750107098,
      "loss": 1.3524,
      "step": 22601
    },
    {
      "epoch": 1.9445926180848319,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006713544720686015,
      "loss": 1.3672,
      "step": 22602
    },
    {
      "epoch": 1.9446786543921535,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006713290686253151,
      "loss": 1.4574,
      "step": 22603
    },
    {
      "epoch": 1.944764690699475,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006713036646809241,
      "loss": 1.0922,
      "step": 22604
    },
    {
      "epoch": 1.9448507270067967,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006712782602355034,
      "loss": 1.1499,
      "step": 22605
    },
    {
      "epoch": 1.9449367633141186,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006712528552891271,
      "loss": 1.675,
      "step": 22606
    },
    {
      "epoch": 1.9450227996214402,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006712274498418694,
      "loss": 1.3605,
      "step": 22607
    },
    {
      "epoch": 1.945108835928762,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006712020438938047,
      "loss": 0.7394,
      "step": 22608
    },
    {
      "epoch": 1.9451948722360837,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006711766374450075,
      "loss": 1.3185,
      "step": 22609
    },
    {
      "epoch": 1.9452809085434053,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006711512304955518,
      "loss": 0.9196,
      "step": 22610
    },
    {
      "epoch": 1.945366944850727,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006711258230455119,
      "loss": 0.8208,
      "step": 22611
    },
    {
      "epoch": 1.9454529811580485,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006711004150949624,
      "loss": 1.772,
      "step": 22612
    },
    {
      "epoch": 1.9455390174653704,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006710750066439775,
      "loss": 1.3438,
      "step": 22613
    },
    {
      "epoch": 1.9456250537726922,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006710495976926314,
      "loss": 1.5202,
      "step": 22614
    },
    {
      "epoch": 1.9457110900800139,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006710241882409986,
      "loss": 1.5649,
      "step": 22615
    },
    {
      "epoch": 1.9457971263873355,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006709987782891532,
      "loss": 1.1801,
      "step": 22616
    },
    {
      "epoch": 1.945883162694657,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006709733678371696,
      "loss": 1.4822,
      "step": 22617
    },
    {
      "epoch": 1.9459691990019787,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006709479568851224,
      "loss": 1.8388,
      "step": 22618
    },
    {
      "epoch": 1.9460552353093006,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006709225454330854,
      "loss": 0.7332,
      "step": 22619
    },
    {
      "epoch": 1.9461412716166222,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006708971334811333,
      "loss": 0.8431,
      "step": 22620
    },
    {
      "epoch": 1.946227307923944,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006708717210293405,
      "loss": 1.7092,
      "step": 22621
    },
    {
      "epoch": 1.9463133442312657,
      "grad_norm": 1.046875,
      "learning_rate": 0.000670846308077781,
      "loss": 1.9026,
      "step": 22622
    },
    {
      "epoch": 1.9463993805385873,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000670820894626529,
      "loss": 0.8569,
      "step": 22623
    },
    {
      "epoch": 1.946485416845909,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006707954806756594,
      "loss": 1.4091,
      "step": 22624
    },
    {
      "epoch": 1.9465714531532305,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006707700662252463,
      "loss": 1.4464,
      "step": 22625
    },
    {
      "epoch": 1.9466574894605524,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006707446512753637,
      "loss": 1.1557,
      "step": 22626
    },
    {
      "epoch": 1.946743525767874,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006707192358260865,
      "loss": 0.8286,
      "step": 22627
    },
    {
      "epoch": 1.9468295620751959,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006706938198774886,
      "loss": 1.0404,
      "step": 22628
    },
    {
      "epoch": 1.9469155983825175,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006706684034296445,
      "loss": 1.8162,
      "step": 22629
    },
    {
      "epoch": 1.9470016346898391,
      "grad_norm": 1.25,
      "learning_rate": 0.0006706429864826285,
      "loss": 1.4612,
      "step": 22630
    },
    {
      "epoch": 1.9470876709971607,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006706175690365148,
      "loss": 1.9996,
      "step": 22631
    },
    {
      "epoch": 1.9471737073044824,
      "grad_norm": 1.0,
      "learning_rate": 0.000670592151091378,
      "loss": 1.4896,
      "step": 22632
    },
    {
      "epoch": 1.9472597436118042,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006705667326472925,
      "loss": 1.8368,
      "step": 22633
    },
    {
      "epoch": 1.9473457799191258,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006705413137043322,
      "loss": 0.7774,
      "step": 22634
    },
    {
      "epoch": 1.9474318162264477,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000670515894262572,
      "loss": 1.5162,
      "step": 22635
    },
    {
      "epoch": 1.9475178525337693,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006704904743220859,
      "loss": 1.7205,
      "step": 22636
    },
    {
      "epoch": 1.947603888841091,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006704650538829482,
      "loss": 0.6715,
      "step": 22637
    },
    {
      "epoch": 1.9476899251484125,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006704396329452334,
      "loss": 0.7918,
      "step": 22638
    },
    {
      "epoch": 1.9477759614557342,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000670414211509016,
      "loss": 1.4067,
      "step": 22639
    },
    {
      "epoch": 1.947861997763056,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00067038878957437,
      "loss": 1.2923,
      "step": 22640
    },
    {
      "epoch": 1.9479480340703779,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00067036336714137,
      "loss": 1.0117,
      "step": 22641
    },
    {
      "epoch": 1.9480340703776995,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006703379442100902,
      "loss": 1.1353,
      "step": 22642
    },
    {
      "epoch": 1.9481201066850211,
      "grad_norm": 1.125,
      "learning_rate": 0.0006703125207806051,
      "loss": 0.8877,
      "step": 22643
    },
    {
      "epoch": 1.9482061429923427,
      "grad_norm": 1.234375,
      "learning_rate": 0.000670287096852989,
      "loss": 1.4666,
      "step": 22644
    },
    {
      "epoch": 1.9482921792996644,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006702616724273163,
      "loss": 1.3361,
      "step": 22645
    },
    {
      "epoch": 1.948378215606986,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006702362475036612,
      "loss": 1.227,
      "step": 22646
    },
    {
      "epoch": 1.9484642519143078,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006702108220820984,
      "loss": 2.126,
      "step": 22647
    },
    {
      "epoch": 1.9485502882216297,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006701853961627019,
      "loss": 1.721,
      "step": 22648
    },
    {
      "epoch": 1.9486363245289513,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006701599697455462,
      "loss": 0.7552,
      "step": 22649
    },
    {
      "epoch": 1.948722360836273,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006701345428307056,
      "loss": 1.648,
      "step": 22650
    },
    {
      "epoch": 1.9488083971435946,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0006701091154182547,
      "loss": 1.298,
      "step": 22651
    },
    {
      "epoch": 1.9488944334509162,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006700836875082678,
      "loss": 1.6864,
      "step": 22652
    },
    {
      "epoch": 1.948980469758238,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006700582591008189,
      "loss": 1.1383,
      "step": 22653
    },
    {
      "epoch": 1.9490665060655596,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006700328301959829,
      "loss": 0.7706,
      "step": 22654
    },
    {
      "epoch": 1.9491525423728815,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006700074007938339,
      "loss": 0.8836,
      "step": 22655
    },
    {
      "epoch": 1.9492385786802031,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006699819708944464,
      "loss": 0.7484,
      "step": 22656
    },
    {
      "epoch": 1.9493246149875247,
      "grad_norm": 2.921875,
      "learning_rate": 0.0006699565404978945,
      "loss": 1.3481,
      "step": 22657
    },
    {
      "epoch": 1.9494106512948464,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006699311096042529,
      "loss": 1.3286,
      "step": 22658
    },
    {
      "epoch": 1.949496687602168,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006699056782135958,
      "loss": 0.9071,
      "step": 22659
    },
    {
      "epoch": 1.9495827239094898,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006698802463259977,
      "loss": 1.3259,
      "step": 22660
    },
    {
      "epoch": 1.9496687602168115,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0006698548139415328,
      "loss": 0.61,
      "step": 22661
    },
    {
      "epoch": 1.9497547965241333,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006698293810602758,
      "loss": 0.9832,
      "step": 22662
    },
    {
      "epoch": 1.949840832831455,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006698039476823008,
      "loss": 1.819,
      "step": 22663
    },
    {
      "epoch": 1.9499268691387766,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006697785138076822,
      "loss": 1.4613,
      "step": 22664
    },
    {
      "epoch": 1.9500129054460982,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006697530794364945,
      "loss": 0.8398,
      "step": 22665
    },
    {
      "epoch": 1.9500989417534198,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006697276445688122,
      "loss": 1.6733,
      "step": 22666
    },
    {
      "epoch": 1.9501849780607416,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006697022092047093,
      "loss": 1.1462,
      "step": 22667
    },
    {
      "epoch": 1.9502710143680633,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006696767733442606,
      "loss": 0.9734,
      "step": 22668
    },
    {
      "epoch": 1.9503570506753851,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006696513369875404,
      "loss": 1.5733,
      "step": 22669
    },
    {
      "epoch": 1.9504430869827067,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006696259001346228,
      "loss": 0.9356,
      "step": 22670
    },
    {
      "epoch": 1.9505291232900284,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006696004627855827,
      "loss": 1.5188,
      "step": 22671
    },
    {
      "epoch": 1.95061515959735,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000669575024940494,
      "loss": 0.7508,
      "step": 22672
    },
    {
      "epoch": 1.9507011959046716,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006695495865994314,
      "loss": 1.1807,
      "step": 22673
    },
    {
      "epoch": 1.9507872322119935,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006695241477624694,
      "loss": 0.9475,
      "step": 22674
    },
    {
      "epoch": 1.9508732685193153,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006694987084296819,
      "loss": 1.6611,
      "step": 22675
    },
    {
      "epoch": 1.950959304826637,
      "grad_norm": 1.0625,
      "learning_rate": 0.000669473268601144,
      "loss": 1.0652,
      "step": 22676
    },
    {
      "epoch": 1.9510453411339586,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006694478282769297,
      "loss": 1.1199,
      "step": 22677
    },
    {
      "epoch": 1.9511313774412802,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006694223874571134,
      "loss": 1.8363,
      "step": 22678
    },
    {
      "epoch": 1.9512174137486018,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006693969461417695,
      "loss": 1.1297,
      "step": 22679
    },
    {
      "epoch": 1.9513034500559236,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006693715043309725,
      "loss": 1.1176,
      "step": 22680
    },
    {
      "epoch": 1.9513894863632453,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006693460620247969,
      "loss": 0.9559,
      "step": 22681
    },
    {
      "epoch": 1.9514755226705671,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006693206192233168,
      "loss": 0.7794,
      "step": 22682
    },
    {
      "epoch": 1.9515615589778887,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006692951759266069,
      "loss": 1.0503,
      "step": 22683
    },
    {
      "epoch": 1.9516475952852104,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006692697321347416,
      "loss": 1.4417,
      "step": 22684
    },
    {
      "epoch": 1.951733631592532,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006692442878477953,
      "loss": 1.2402,
      "step": 22685
    },
    {
      "epoch": 1.9518196678998536,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006692188430658423,
      "loss": 0.9279,
      "step": 22686
    },
    {
      "epoch": 1.9519057042071755,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006691933977889569,
      "loss": 1.3958,
      "step": 22687
    },
    {
      "epoch": 1.951991740514497,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000669167952017214,
      "loss": 1.6608,
      "step": 22688
    },
    {
      "epoch": 1.952077776821819,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006691425057506877,
      "loss": 1.0213,
      "step": 22689
    },
    {
      "epoch": 1.9521638131291406,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006691170589894523,
      "loss": 0.9701,
      "step": 22690
    },
    {
      "epoch": 1.9522498494364622,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006690916117335825,
      "loss": 1.217,
      "step": 22691
    },
    {
      "epoch": 1.9523358857437838,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006690661639831526,
      "loss": 1.0524,
      "step": 22692
    },
    {
      "epoch": 1.9524219220511054,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006690407157382372,
      "loss": 1.4598,
      "step": 22693
    },
    {
      "epoch": 1.9525079583584273,
      "grad_norm": 1.125,
      "learning_rate": 0.0006690152669989104,
      "loss": 0.7394,
      "step": 22694
    },
    {
      "epoch": 1.952593994665749,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006689898177652469,
      "loss": 1.7442,
      "step": 22695
    },
    {
      "epoch": 1.9526800309730707,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006689643680373211,
      "loss": 1.2749,
      "step": 22696
    },
    {
      "epoch": 1.9527660672803924,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006689389178152073,
      "loss": 0.8932,
      "step": 22697
    },
    {
      "epoch": 1.952852103587714,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006689134670989801,
      "loss": 1.5751,
      "step": 22698
    },
    {
      "epoch": 1.9529381398950356,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006688880158887137,
      "loss": 0.5831,
      "step": 22699
    },
    {
      "epoch": 1.9530241762023572,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006688625641844829,
      "loss": 1.0971,
      "step": 22700
    },
    {
      "epoch": 1.953110212509679,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006688371119863617,
      "loss": 1.1241,
      "step": 22701
    },
    {
      "epoch": 1.953196248817001,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000668811659294425,
      "loss": 1.4012,
      "step": 22702
    },
    {
      "epoch": 1.9532822851243226,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000668786206108747,
      "loss": 0.712,
      "step": 22703
    },
    {
      "epoch": 1.9533683214316442,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006687607524294021,
      "loss": 0.5901,
      "step": 22704
    },
    {
      "epoch": 1.9534543577389658,
      "grad_norm": 1.6875,
      "learning_rate": 0.0006687352982564648,
      "loss": 1.8816,
      "step": 22705
    },
    {
      "epoch": 1.9535403940462874,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006687098435900097,
      "loss": 0.8303,
      "step": 22706
    },
    {
      "epoch": 1.9536264303536093,
      "grad_norm": 0.84375,
      "learning_rate": 0.000668684388430111,
      "loss": 1.1166,
      "step": 22707
    },
    {
      "epoch": 1.953712466660931,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006686589327768433,
      "loss": 1.3854,
      "step": 22708
    },
    {
      "epoch": 1.9537985029682527,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006686334766302811,
      "loss": 1.0489,
      "step": 22709
    },
    {
      "epoch": 1.9538845392755744,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006686080199904987,
      "loss": 1.4758,
      "step": 22710
    },
    {
      "epoch": 1.953970575582896,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006685825628575706,
      "loss": 1.2097,
      "step": 22711
    },
    {
      "epoch": 1.9540566118902176,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006685571052315715,
      "loss": 1.4887,
      "step": 22712
    },
    {
      "epoch": 1.9541426481975392,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006685316471125753,
      "loss": 1.8305,
      "step": 22713
    },
    {
      "epoch": 1.954228684504861,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006685061885006571,
      "loss": 0.9304,
      "step": 22714
    },
    {
      "epoch": 1.9543147208121827,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000668480729395891,
      "loss": 1.9591,
      "step": 22715
    },
    {
      "epoch": 1.9544007571195046,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006684552697983514,
      "loss": 1.4244,
      "step": 22716
    },
    {
      "epoch": 1.9544867934268262,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006684298097081131,
      "loss": 1.2949,
      "step": 22717
    },
    {
      "epoch": 1.9545728297341478,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006684043491252502,
      "loss": 1.473,
      "step": 22718
    },
    {
      "epoch": 1.9546588660414694,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006683788880498375,
      "loss": 1.6885,
      "step": 22719
    },
    {
      "epoch": 1.954744902348791,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000668353426481949,
      "loss": 0.8635,
      "step": 22720
    },
    {
      "epoch": 1.954830938656113,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006683279644216597,
      "loss": 0.7868,
      "step": 22721
    },
    {
      "epoch": 1.9549169749634345,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000668302501869044,
      "loss": 1.1396,
      "step": 22722
    },
    {
      "epoch": 1.9550030112707564,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006682770388241759,
      "loss": 1.3776,
      "step": 22723
    },
    {
      "epoch": 1.955089047578078,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006682515752871303,
      "loss": 0.9979,
      "step": 22724
    },
    {
      "epoch": 1.9551750838853996,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006682261112579816,
      "loss": 1.1741,
      "step": 22725
    },
    {
      "epoch": 1.9552611201927212,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006682006467368043,
      "loss": 1.8645,
      "step": 22726
    },
    {
      "epoch": 1.9553471565000429,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006681751817236728,
      "loss": 1.5044,
      "step": 22727
    },
    {
      "epoch": 1.9554331928073647,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006681497162186614,
      "loss": 1.1061,
      "step": 22728
    },
    {
      "epoch": 1.9555192291146866,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000668124250221845,
      "loss": 1.5786,
      "step": 22729
    },
    {
      "epoch": 1.9556052654220082,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006680987837332978,
      "loss": 1.3127,
      "step": 22730
    },
    {
      "epoch": 1.9556913017293298,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006680733167530943,
      "loss": 1.3254,
      "step": 22731
    },
    {
      "epoch": 1.9557773380366514,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006680478492813091,
      "loss": 1.2224,
      "step": 22732
    },
    {
      "epoch": 1.955863374343973,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006680223813180167,
      "loss": 1.3979,
      "step": 22733
    },
    {
      "epoch": 1.9559494106512947,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006679969128632915,
      "loss": 1.8362,
      "step": 22734
    },
    {
      "epoch": 1.9560354469586165,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006679714439172077,
      "loss": 0.7317,
      "step": 22735
    },
    {
      "epoch": 1.9561214832659384,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006679459744798404,
      "loss": 1.5626,
      "step": 22736
    },
    {
      "epoch": 1.95620751957326,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006679205045512637,
      "loss": 1.1121,
      "step": 22737
    },
    {
      "epoch": 1.9562935558805816,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006678950341315522,
      "loss": 1.7289,
      "step": 22738
    },
    {
      "epoch": 1.9563795921879032,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006678695632207804,
      "loss": 1.2961,
      "step": 22739
    },
    {
      "epoch": 1.9564656284952249,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006678440918190228,
      "loss": 0.981,
      "step": 22740
    },
    {
      "epoch": 1.9565516648025467,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006678186199263539,
      "loss": 2.0163,
      "step": 22741
    },
    {
      "epoch": 1.9566377011098683,
      "grad_norm": 1.5,
      "learning_rate": 0.0006677931475428479,
      "loss": 1.9453,
      "step": 22742
    },
    {
      "epoch": 1.9567237374171902,
      "grad_norm": 1.65625,
      "learning_rate": 0.0006677676746685797,
      "loss": 1.323,
      "step": 22743
    },
    {
      "epoch": 1.9568097737245118,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006677422013036239,
      "loss": 1.392,
      "step": 22744
    },
    {
      "epoch": 1.9568958100318334,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006677167274480546,
      "loss": 1.414,
      "step": 22745
    },
    {
      "epoch": 1.956981846339155,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006676912531019464,
      "loss": 1.4572,
      "step": 22746
    },
    {
      "epoch": 1.9570678826464767,
      "grad_norm": 1.0,
      "learning_rate": 0.0006676657782653741,
      "loss": 1.941,
      "step": 22747
    },
    {
      "epoch": 1.9571539189537985,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000667640302938412,
      "loss": 2.0074,
      "step": 22748
    },
    {
      "epoch": 1.9572399552611202,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006676148271211347,
      "loss": 0.8201,
      "step": 22749
    },
    {
      "epoch": 1.957325991568442,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006675893508136163,
      "loss": 2.1869,
      "step": 22750
    },
    {
      "epoch": 1.9574120278757636,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006675638740159319,
      "loss": 0.8719,
      "step": 22751
    },
    {
      "epoch": 1.9574980641830853,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006675383967281557,
      "loss": 1.6539,
      "step": 22752
    },
    {
      "epoch": 1.9575841004904069,
      "grad_norm": 1.0,
      "learning_rate": 0.0006675129189503624,
      "loss": 2.0889,
      "step": 22753
    },
    {
      "epoch": 1.9576701367977285,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006674874406826262,
      "loss": 1.0172,
      "step": 22754
    },
    {
      "epoch": 1.9577561731050503,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006674619619250218,
      "loss": 1.0072,
      "step": 22755
    },
    {
      "epoch": 1.957842209412372,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006674364826776238,
      "loss": 0.4948,
      "step": 22756
    },
    {
      "epoch": 1.9579282457196938,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006674110029405066,
      "loss": 1.5899,
      "step": 22757
    },
    {
      "epoch": 1.9580142820270154,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006673855227137448,
      "loss": 1.6481,
      "step": 22758
    },
    {
      "epoch": 1.958100318334337,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006673600419974131,
      "loss": 1.0043,
      "step": 22759
    },
    {
      "epoch": 1.9581863546416587,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006673345607915856,
      "loss": 1.5526,
      "step": 22760
    },
    {
      "epoch": 1.9582723909489803,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006673090790963371,
      "loss": 1.836,
      "step": 22761
    },
    {
      "epoch": 1.9583584272563022,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000667283596911742,
      "loss": 0.825,
      "step": 22762
    },
    {
      "epoch": 1.958444463563624,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006672581142378751,
      "loss": 1.497,
      "step": 22763
    },
    {
      "epoch": 1.9585304998709456,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006672326310748106,
      "loss": 1.6693,
      "step": 22764
    },
    {
      "epoch": 1.9586165361782673,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006672071474226231,
      "loss": 1.0598,
      "step": 22765
    },
    {
      "epoch": 1.9587025724855889,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006671816632813874,
      "loss": 1.5486,
      "step": 22766
    },
    {
      "epoch": 1.9587886087929105,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006671561786511779,
      "loss": 1.2503,
      "step": 22767
    },
    {
      "epoch": 1.9588746451002323,
      "grad_norm": 1.375,
      "learning_rate": 0.0006671306935320691,
      "loss": 1.3917,
      "step": 22768
    },
    {
      "epoch": 1.958960681407554,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006671052079241352,
      "loss": 0.8967,
      "step": 22769
    },
    {
      "epoch": 1.9590467177148758,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0006670797218274513,
      "loss": 1.6299,
      "step": 22770
    },
    {
      "epoch": 1.9591327540221974,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006670542352420918,
      "loss": 1.2512,
      "step": 22771
    },
    {
      "epoch": 1.959218790329519,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006670287481681309,
      "loss": 1.6576,
      "step": 22772
    },
    {
      "epoch": 1.9593048266368407,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006670032606056435,
      "loss": 0.7505,
      "step": 22773
    },
    {
      "epoch": 1.9593908629441623,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006669777725547041,
      "loss": 0.7409,
      "step": 22774
    },
    {
      "epoch": 1.9594768992514842,
      "grad_norm": 1.25,
      "learning_rate": 0.0006669522840153872,
      "loss": 2.0111,
      "step": 22775
    },
    {
      "epoch": 1.9595629355588058,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006669267949877673,
      "loss": 0.8102,
      "step": 22776
    },
    {
      "epoch": 1.9596489718661276,
      "grad_norm": 1.0,
      "learning_rate": 0.000666901305471919,
      "loss": 0.7078,
      "step": 22777
    },
    {
      "epoch": 1.9597350081734493,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006668758154679169,
      "loss": 0.5572,
      "step": 22778
    },
    {
      "epoch": 1.9598210444807709,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006668503249758353,
      "loss": 1.7273,
      "step": 22779
    },
    {
      "epoch": 1.9599070807880925,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006668248339957491,
      "loss": 0.936,
      "step": 22780
    },
    {
      "epoch": 1.9599931170954141,
      "grad_norm": 1.734375,
      "learning_rate": 0.0006667993425277328,
      "loss": 1.3093,
      "step": 22781
    },
    {
      "epoch": 1.960079153402736,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006667738505718607,
      "loss": 1.6619,
      "step": 22782
    },
    {
      "epoch": 1.9601651897100576,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006667483581282076,
      "loss": 1.2152,
      "step": 22783
    },
    {
      "epoch": 1.9602512260173794,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006667228651968479,
      "loss": 1.7939,
      "step": 22784
    },
    {
      "epoch": 1.960337262324701,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006666973717778564,
      "loss": 1.3131,
      "step": 22785
    },
    {
      "epoch": 1.9604232986320227,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006666718778713074,
      "loss": 0.5467,
      "step": 22786
    },
    {
      "epoch": 1.9605093349393443,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006666463834772754,
      "loss": 1.4761,
      "step": 22787
    },
    {
      "epoch": 1.960595371246666,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006666208885958355,
      "loss": 1.485,
      "step": 22788
    },
    {
      "epoch": 1.9606814075539878,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006665953932270618,
      "loss": 1.4659,
      "step": 22789
    },
    {
      "epoch": 1.9607674438613096,
      "grad_norm": 1.0,
      "learning_rate": 0.0006665698973710288,
      "loss": 0.7344,
      "step": 22790
    },
    {
      "epoch": 1.9608534801686313,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006665444010278114,
      "loss": 0.933,
      "step": 22791
    },
    {
      "epoch": 1.9609395164759529,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000666518904197484,
      "loss": 1.3228,
      "step": 22792
    },
    {
      "epoch": 1.9610255527832745,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006664934068801212,
      "loss": 0.7615,
      "step": 22793
    },
    {
      "epoch": 1.9611115890905961,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006664679090757975,
      "loss": 1.2291,
      "step": 22794
    },
    {
      "epoch": 1.961197625397918,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006664424107845876,
      "loss": 1.3213,
      "step": 22795
    },
    {
      "epoch": 1.9612836617052396,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006664169120065659,
      "loss": 0.9656,
      "step": 22796
    },
    {
      "epoch": 1.9613696980125614,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006663914127418072,
      "loss": 1.4798,
      "step": 22797
    },
    {
      "epoch": 1.961455734319883,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006663659129903859,
      "loss": 1.0846,
      "step": 22798
    },
    {
      "epoch": 1.9615417706272047,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006663404127523766,
      "loss": 1.2784,
      "step": 22799
    },
    {
      "epoch": 1.9616278069345263,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006663149120278542,
      "loss": 1.3376,
      "step": 22800
    },
    {
      "epoch": 1.961713843241848,
      "grad_norm": 0.84375,
      "learning_rate": 0.000666289410816893,
      "loss": 0.726,
      "step": 22801
    },
    {
      "epoch": 1.9617998795491698,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006662639091195674,
      "loss": 1.1867,
      "step": 22802
    },
    {
      "epoch": 1.9618859158564914,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006662384069359524,
      "loss": 1.4774,
      "step": 22803
    },
    {
      "epoch": 1.9619719521638133,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006662129042661224,
      "loss": 1.4292,
      "step": 22804
    },
    {
      "epoch": 1.9620579884711349,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006661874011101519,
      "loss": 1.2461,
      "step": 22805
    },
    {
      "epoch": 1.9621440247784565,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006661618974681154,
      "loss": 1.3256,
      "step": 22806
    },
    {
      "epoch": 1.9622300610857781,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006661363933400879,
      "loss": 1.1368,
      "step": 22807
    },
    {
      "epoch": 1.9623160973930998,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006661108887261438,
      "loss": 1.7937,
      "step": 22808
    },
    {
      "epoch": 1.9624021337004216,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006660853836263575,
      "loss": 1.6894,
      "step": 22809
    },
    {
      "epoch": 1.9624881700077432,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006660598780408038,
      "loss": 1.3392,
      "step": 22810
    },
    {
      "epoch": 1.962574206315065,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006660343719695572,
      "loss": 0.4543,
      "step": 22811
    },
    {
      "epoch": 1.9626602426223867,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006660088654126925,
      "loss": 1.4169,
      "step": 22812
    },
    {
      "epoch": 1.9627462789297083,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006659833583702841,
      "loss": 1.1888,
      "step": 22813
    },
    {
      "epoch": 1.96283231523703,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006659578508424065,
      "loss": 0.9034,
      "step": 22814
    },
    {
      "epoch": 1.9629183515443516,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006659323428291346,
      "loss": 1.182,
      "step": 22815
    },
    {
      "epoch": 1.9630043878516734,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006659068343305429,
      "loss": 1.6198,
      "step": 22816
    },
    {
      "epoch": 1.9630904241589953,
      "grad_norm": 1.109375,
      "learning_rate": 0.000665881325346706,
      "loss": 1.1846,
      "step": 22817
    },
    {
      "epoch": 1.9631764604663169,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006658558158776983,
      "loss": 1.5667,
      "step": 22818
    },
    {
      "epoch": 1.9632624967736385,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006658303059235947,
      "loss": 0.8572,
      "step": 22819
    },
    {
      "epoch": 1.9633485330809601,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006658047954844698,
      "loss": 2.2842,
      "step": 22820
    },
    {
      "epoch": 1.9634345693882818,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006657792845603981,
      "loss": 1.6506,
      "step": 22821
    },
    {
      "epoch": 1.9635206056956034,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006657537731514541,
      "loss": 1.2744,
      "step": 22822
    },
    {
      "epoch": 1.9636066420029252,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006657282612577127,
      "loss": 1.6159,
      "step": 22823
    },
    {
      "epoch": 1.963692678310247,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006657027488792483,
      "loss": 0.9202,
      "step": 22824
    },
    {
      "epoch": 1.9637787146175687,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006656772360161355,
      "loss": 1.8158,
      "step": 22825
    },
    {
      "epoch": 1.9638647509248903,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006656517226684491,
      "loss": 1.8388,
      "step": 22826
    },
    {
      "epoch": 1.963950787232212,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006656262088362637,
      "loss": 0.654,
      "step": 22827
    },
    {
      "epoch": 1.9640368235395336,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006656006945196537,
      "loss": 0.9474,
      "step": 22828
    },
    {
      "epoch": 1.9641228598468554,
      "grad_norm": 1.015625,
      "learning_rate": 0.000665575179718694,
      "loss": 1.8614,
      "step": 22829
    },
    {
      "epoch": 1.964208896154177,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006655496644334591,
      "loss": 1.2433,
      "step": 22830
    },
    {
      "epoch": 1.9642949324614989,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006655241486640236,
      "loss": 1.3788,
      "step": 22831
    },
    {
      "epoch": 1.9643809687688205,
      "grad_norm": 1.515625,
      "learning_rate": 0.000665498632410462,
      "loss": 1.2279,
      "step": 22832
    },
    {
      "epoch": 1.9644670050761421,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006654731156728493,
      "loss": 1.9022,
      "step": 22833
    },
    {
      "epoch": 1.9645530413834638,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006654475984512599,
      "loss": 0.9489,
      "step": 22834
    },
    {
      "epoch": 1.9646390776907854,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006654220807457684,
      "loss": 1.2973,
      "step": 22835
    },
    {
      "epoch": 1.9647251139981072,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006653965625564493,
      "loss": 1.6396,
      "step": 22836
    },
    {
      "epoch": 1.9648111503054289,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006653710438833776,
      "loss": 1.7073,
      "step": 22837
    },
    {
      "epoch": 1.9648971866127507,
      "grad_norm": 1.0,
      "learning_rate": 0.000665345524726628,
      "loss": 0.9015,
      "step": 22838
    },
    {
      "epoch": 1.9649832229200723,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006653200050862747,
      "loss": 1.1055,
      "step": 22839
    },
    {
      "epoch": 1.965069259227394,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006652944849623923,
      "loss": 0.5603,
      "step": 22840
    },
    {
      "epoch": 1.9651552955347156,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000665268964355056,
      "loss": 0.7017,
      "step": 22841
    },
    {
      "epoch": 1.9652413318420372,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00066524344326434,
      "loss": 1.2142,
      "step": 22842
    },
    {
      "epoch": 1.965327368149359,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006652179216903191,
      "loss": 1.2413,
      "step": 22843
    },
    {
      "epoch": 1.9654134044566807,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006651923996330677,
      "loss": 0.8017,
      "step": 22844
    },
    {
      "epoch": 1.9654994407640025,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000665166877092661,
      "loss": 0.816,
      "step": 22845
    },
    {
      "epoch": 1.9655854770713241,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006651413540691732,
      "loss": 0.7842,
      "step": 22846
    },
    {
      "epoch": 1.9656715133786458,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006651158305626788,
      "loss": 1.3851,
      "step": 22847
    },
    {
      "epoch": 1.9657575496859674,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006650903065732531,
      "loss": 0.8766,
      "step": 22848
    },
    {
      "epoch": 1.965843585993289,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006650647821009702,
      "loss": 1.962,
      "step": 22849
    },
    {
      "epoch": 1.9659296223006109,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006650392571459049,
      "loss": 1.432,
      "step": 22850
    },
    {
      "epoch": 1.9660156586079327,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006650137317081318,
      "loss": 1.0336,
      "step": 22851
    },
    {
      "epoch": 1.9661016949152543,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006649882057877257,
      "loss": 1.329,
      "step": 22852
    },
    {
      "epoch": 1.966187731222576,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006649626793847612,
      "loss": 0.946,
      "step": 22853
    },
    {
      "epoch": 1.9662737675298976,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006649371524993129,
      "loss": 1.3158,
      "step": 22854
    },
    {
      "epoch": 1.9663598038372192,
      "grad_norm": 1.0,
      "learning_rate": 0.0006649116251314555,
      "loss": 1.5418,
      "step": 22855
    },
    {
      "epoch": 1.966445840144541,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006648860972812637,
      "loss": 1.0076,
      "step": 22856
    },
    {
      "epoch": 1.9665318764518627,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006648605689488122,
      "loss": 0.8282,
      "step": 22857
    },
    {
      "epoch": 1.9666179127591845,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006648350401341754,
      "loss": 1.407,
      "step": 22858
    },
    {
      "epoch": 1.9667039490665061,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006648095108374284,
      "loss": 1.4146,
      "step": 22859
    },
    {
      "epoch": 1.9667899853738278,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006647839810586455,
      "loss": 1.5774,
      "step": 22860
    },
    {
      "epoch": 1.9668760216811494,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006647584507979016,
      "loss": 1.465,
      "step": 22861
    },
    {
      "epoch": 1.966962057988471,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006647329200552712,
      "loss": 0.9328,
      "step": 22862
    },
    {
      "epoch": 1.9670480942957929,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000664707388830829,
      "loss": 0.8445,
      "step": 22863
    },
    {
      "epoch": 1.9671341306031145,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0006646818571246499,
      "loss": 1.2117,
      "step": 22864
    },
    {
      "epoch": 1.9672201669104363,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006646563249368083,
      "loss": 1.1007,
      "step": 22865
    },
    {
      "epoch": 1.967306203217758,
      "grad_norm": 16.75,
      "learning_rate": 0.0006646307922673789,
      "loss": 0.9407,
      "step": 22866
    },
    {
      "epoch": 1.9673922395250796,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006646052591164365,
      "loss": 1.3799,
      "step": 22867
    },
    {
      "epoch": 1.9674782758324012,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006645797254840557,
      "loss": 1.0161,
      "step": 22868
    },
    {
      "epoch": 1.9675643121397228,
      "grad_norm": 166.0,
      "learning_rate": 0.0006645541913703112,
      "loss": 1.3438,
      "step": 22869
    },
    {
      "epoch": 1.9676503484470447,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006645286567752779,
      "loss": 1.014,
      "step": 22870
    },
    {
      "epoch": 1.9677363847543663,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006645031216990302,
      "loss": 1.7697,
      "step": 22871
    },
    {
      "epoch": 1.9678224210616881,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006644775861416426,
      "loss": 1.3239,
      "step": 22872
    },
    {
      "epoch": 1.9679084573690098,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006644520501031903,
      "loss": 1.007,
      "step": 22873
    },
    {
      "epoch": 1.9679944936763314,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006644265135837477,
      "loss": 2.0282,
      "step": 22874
    },
    {
      "epoch": 1.968080529983653,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006644009765833895,
      "loss": 1.7495,
      "step": 22875
    },
    {
      "epoch": 1.9681665662909746,
      "grad_norm": 1.125,
      "learning_rate": 0.0006643754391021904,
      "loss": 1.3838,
      "step": 22876
    },
    {
      "epoch": 1.9682526025982965,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006643499011402251,
      "loss": 0.8155,
      "step": 22877
    },
    {
      "epoch": 1.9683386389056183,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006643243626975684,
      "loss": 2.1365,
      "step": 22878
    },
    {
      "epoch": 1.96842467521294,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006642988237742949,
      "loss": 1.6308,
      "step": 22879
    },
    {
      "epoch": 1.9685107115202616,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006642732843704792,
      "loss": 0.8861,
      "step": 22880
    },
    {
      "epoch": 1.9685967478275832,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006642477444861961,
      "loss": 1.0773,
      "step": 22881
    },
    {
      "epoch": 1.9686827841349048,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006642222041215203,
      "loss": 0.7913,
      "step": 22882
    },
    {
      "epoch": 1.9687688204422267,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006641966632765265,
      "loss": 1.9118,
      "step": 22883
    },
    {
      "epoch": 1.9688548567495483,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006641711219512894,
      "loss": 0.7067,
      "step": 22884
    },
    {
      "epoch": 1.9689408930568701,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006641455801458837,
      "loss": 1.628,
      "step": 22885
    },
    {
      "epoch": 1.9690269293641918,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006641200378603841,
      "loss": 1.0103,
      "step": 22886
    },
    {
      "epoch": 1.9691129656715134,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006640944950948653,
      "loss": 0.8563,
      "step": 22887
    },
    {
      "epoch": 1.969199001978835,
      "grad_norm": 0.921875,
      "learning_rate": 0.000664068951849402,
      "loss": 1.2258,
      "step": 22888
    },
    {
      "epoch": 1.9692850382861566,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000664043408124069,
      "loss": 0.8935,
      "step": 22889
    },
    {
      "epoch": 1.9693710745934785,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006640178639189408,
      "loss": 1.2147,
      "step": 22890
    },
    {
      "epoch": 1.9694571109008001,
      "grad_norm": 1.515625,
      "learning_rate": 0.0006639923192340924,
      "loss": 1.1382,
      "step": 22891
    },
    {
      "epoch": 1.969543147208122,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006639667740695981,
      "loss": 1.3863,
      "step": 22892
    },
    {
      "epoch": 1.9696291835154436,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006639412284255332,
      "loss": 1.0655,
      "step": 22893
    },
    {
      "epoch": 1.9697152198227652,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006639156823019719,
      "loss": 0.7655,
      "step": 22894
    },
    {
      "epoch": 1.9698012561300868,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006638901356989893,
      "loss": 0.9679,
      "step": 22895
    },
    {
      "epoch": 1.9698872924374085,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006638645886166597,
      "loss": 1.2406,
      "step": 22896
    },
    {
      "epoch": 1.9699733287447303,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006638390410550581,
      "loss": 0.9249,
      "step": 22897
    },
    {
      "epoch": 1.970059365052052,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006638134930142593,
      "loss": 1.7131,
      "step": 22898
    },
    {
      "epoch": 1.9701454013593738,
      "grad_norm": 0.875,
      "learning_rate": 0.0006637879444943378,
      "loss": 1.3438,
      "step": 22899
    },
    {
      "epoch": 1.9702314376666954,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006637623954953685,
      "loss": 1.85,
      "step": 22900
    },
    {
      "epoch": 1.970317473974017,
      "grad_norm": 1.640625,
      "learning_rate": 0.0006637368460174259,
      "loss": 1.7059,
      "step": 22901
    },
    {
      "epoch": 1.9704035102813386,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000663711296060585,
      "loss": 1.2331,
      "step": 22902
    },
    {
      "epoch": 1.9704895465886603,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006636857456249202,
      "loss": 1.3203,
      "step": 22903
    },
    {
      "epoch": 1.9705755828959821,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006636601947105067,
      "loss": 2.0316,
      "step": 22904
    },
    {
      "epoch": 1.970661619203304,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006636346433174188,
      "loss": 1.3209,
      "step": 22905
    },
    {
      "epoch": 1.9707476555106256,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006636090914457314,
      "loss": 1.9764,
      "step": 22906
    },
    {
      "epoch": 1.9708336918179472,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006635835390955192,
      "loss": 0.9473,
      "step": 22907
    },
    {
      "epoch": 1.9709197281252688,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006635579862668571,
      "loss": 1.665,
      "step": 22908
    },
    {
      "epoch": 1.9710057644325905,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006635324329598196,
      "loss": 1.1388,
      "step": 22909
    },
    {
      "epoch": 1.971091800739912,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006635068791744815,
      "loss": 1.6372,
      "step": 22910
    },
    {
      "epoch": 1.971177837047234,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006634813249109177,
      "loss": 1.163,
      "step": 22911
    },
    {
      "epoch": 1.9712638733545558,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006634557701692027,
      "loss": 1.1034,
      "step": 22912
    },
    {
      "epoch": 1.9713499096618774,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006634302149494114,
      "loss": 1.2428,
      "step": 22913
    },
    {
      "epoch": 1.971435945969199,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006634046592516185,
      "loss": 1.2465,
      "step": 22914
    },
    {
      "epoch": 1.9715219822765206,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006633791030758988,
      "loss": 0.6441,
      "step": 22915
    },
    {
      "epoch": 1.9716080185838423,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006633535464223271,
      "loss": 0.8231,
      "step": 22916
    },
    {
      "epoch": 1.9716940548911641,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006633279892909778,
      "loss": 1.201,
      "step": 22917
    },
    {
      "epoch": 1.9717800911984857,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006633024316819259,
      "loss": 0.8376,
      "step": 22918
    },
    {
      "epoch": 1.9718661275058076,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006632768735952464,
      "loss": 1.0717,
      "step": 22919
    },
    {
      "epoch": 1.9719521638131292,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006632513150310137,
      "loss": 1.0005,
      "step": 22920
    },
    {
      "epoch": 1.9720382001204508,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006632257559893027,
      "loss": 1.5404,
      "step": 22921
    },
    {
      "epoch": 1.9721242364277725,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000663200196470188,
      "loss": 0.8086,
      "step": 22922
    },
    {
      "epoch": 1.972210272735094,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006631746364737445,
      "loss": 1.3868,
      "step": 22923
    },
    {
      "epoch": 1.972296309042416,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006631490760000469,
      "loss": 1.0285,
      "step": 22924
    },
    {
      "epoch": 1.9723823453497376,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006631235150491699,
      "loss": 1.461,
      "step": 22925
    },
    {
      "epoch": 1.9724683816570594,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006630979536211885,
      "loss": 0.4267,
      "step": 22926
    },
    {
      "epoch": 1.972554417964381,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006630723917161772,
      "loss": 1.2953,
      "step": 22927
    },
    {
      "epoch": 1.9726404542717026,
      "grad_norm": 1.125,
      "learning_rate": 0.0006630468293342109,
      "loss": 1.2499,
      "step": 22928
    },
    {
      "epoch": 1.9727264905790243,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006630212664753644,
      "loss": 1.1066,
      "step": 22929
    },
    {
      "epoch": 1.972812526886346,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006629957031397122,
      "loss": 1.2708,
      "step": 22930
    },
    {
      "epoch": 1.9728985631936677,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006629701393273295,
      "loss": 1.8449,
      "step": 22931
    },
    {
      "epoch": 1.9729845995009894,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006629445750382909,
      "loss": 1.2357,
      "step": 22932
    },
    {
      "epoch": 1.9730706358083112,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006629190102726708,
      "loss": 1.2764,
      "step": 22933
    },
    {
      "epoch": 1.9731566721156328,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006628934450305445,
      "loss": 1.5404,
      "step": 22934
    },
    {
      "epoch": 1.9732427084229545,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006628678793119865,
      "loss": 1.3138,
      "step": 22935
    },
    {
      "epoch": 1.973328744730276,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006628423131170717,
      "loss": 1.0234,
      "step": 22936
    },
    {
      "epoch": 1.9734147810375977,
      "grad_norm": 2.171875,
      "learning_rate": 0.0006628167464458747,
      "loss": 1.6428,
      "step": 22937
    },
    {
      "epoch": 1.9735008173449196,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006627911792984704,
      "loss": 0.815,
      "step": 22938
    },
    {
      "epoch": 1.9735868536522414,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006627656116749336,
      "loss": 1.4586,
      "step": 22939
    },
    {
      "epoch": 1.973672889959563,
      "grad_norm": 1.546875,
      "learning_rate": 0.000662740043575339,
      "loss": 1.8045,
      "step": 22940
    },
    {
      "epoch": 1.9737589262668847,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006627144749997615,
      "loss": 1.95,
      "step": 22941
    },
    {
      "epoch": 1.9738449625742063,
      "grad_norm": 1.125,
      "learning_rate": 0.000662688905948276,
      "loss": 1.2325,
      "step": 22942
    },
    {
      "epoch": 1.973930998881528,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006626633364209567,
      "loss": 0.8971,
      "step": 22943
    },
    {
      "epoch": 1.9740170351888497,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006626377664178789,
      "loss": 0.5309,
      "step": 22944
    },
    {
      "epoch": 1.9741030714961714,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006626121959391174,
      "loss": 1.4395,
      "step": 22945
    },
    {
      "epoch": 1.9741891078034932,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0006625866249847465,
      "loss": 1.0668,
      "step": 22946
    },
    {
      "epoch": 1.9742751441108148,
      "grad_norm": 1.125,
      "learning_rate": 0.0006625610535548418,
      "loss": 1.4994,
      "step": 22947
    },
    {
      "epoch": 1.9743611804181365,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006625354816494773,
      "loss": 1.9794,
      "step": 22948
    },
    {
      "epoch": 1.974447216725458,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006625099092687283,
      "loss": 1.494,
      "step": 22949
    },
    {
      "epoch": 1.9745332530327797,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006624843364126694,
      "loss": 1.0255,
      "step": 22950
    },
    {
      "epoch": 1.9746192893401016,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006624587630813754,
      "loss": 1.1167,
      "step": 22951
    },
    {
      "epoch": 1.9747053256474232,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0006624331892749211,
      "loss": 1.8106,
      "step": 22952
    },
    {
      "epoch": 1.974791361954745,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006624076149933814,
      "loss": 1.383,
      "step": 22953
    },
    {
      "epoch": 1.9748773982620667,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006623820402368308,
      "loss": 1.0206,
      "step": 22954
    },
    {
      "epoch": 1.9749634345693883,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006623564650053445,
      "loss": 0.7618,
      "step": 22955
    },
    {
      "epoch": 1.97504947087671,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000662330889298997,
      "loss": 0.8161,
      "step": 22956
    },
    {
      "epoch": 1.9751355071840315,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006623053131178633,
      "loss": 1.0897,
      "step": 22957
    },
    {
      "epoch": 1.9752215434913534,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006622797364620181,
      "loss": 1.4173,
      "step": 22958
    },
    {
      "epoch": 1.975307579798675,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006622541593315362,
      "loss": 0.9079,
      "step": 22959
    },
    {
      "epoch": 1.9753936161059968,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006622285817264926,
      "loss": 0.5908,
      "step": 22960
    },
    {
      "epoch": 1.9754796524133185,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006622030036469618,
      "loss": 1.4201,
      "step": 22961
    },
    {
      "epoch": 1.97556568872064,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006621774250930188,
      "loss": 1.5871,
      "step": 22962
    },
    {
      "epoch": 1.9756517250279617,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006621518460647384,
      "loss": 0.9363,
      "step": 22963
    },
    {
      "epoch": 1.9757377613352833,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006621262665621952,
      "loss": 1.4751,
      "step": 22964
    },
    {
      "epoch": 1.9758237976426052,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006621006865854644,
      "loss": 1.3655,
      "step": 22965
    },
    {
      "epoch": 1.975909833949927,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006620751061346207,
      "loss": 1.5001,
      "step": 22966
    },
    {
      "epoch": 1.9759958702572487,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006620495252097386,
      "loss": 1.7835,
      "step": 22967
    },
    {
      "epoch": 1.9760819065645703,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006620239438108932,
      "loss": 0.8385,
      "step": 22968
    },
    {
      "epoch": 1.976167942871892,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006619983619381595,
      "loss": 2.4917,
      "step": 22969
    },
    {
      "epoch": 1.9762539791792135,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006619727795916119,
      "loss": 1.4659,
      "step": 22970
    },
    {
      "epoch": 1.9763400154865354,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006619471967713254,
      "loss": 2.1925,
      "step": 22971
    },
    {
      "epoch": 1.976426051793857,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006619216134773749,
      "loss": 0.4983,
      "step": 22972
    },
    {
      "epoch": 1.9765120881011788,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006618960297098353,
      "loss": 0.7499,
      "step": 22973
    },
    {
      "epoch": 1.9765981244085005,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006618704454687811,
      "loss": 1.1434,
      "step": 22974
    },
    {
      "epoch": 1.976684160715822,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006618448607542874,
      "loss": 0.8448,
      "step": 22975
    },
    {
      "epoch": 1.9767701970231437,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006618192755664291,
      "loss": 1.3909,
      "step": 22976
    },
    {
      "epoch": 1.9768562333304653,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006617936899052807,
      "loss": 0.9201,
      "step": 22977
    },
    {
      "epoch": 1.9769422696377872,
      "grad_norm": 1.640625,
      "learning_rate": 0.0006617681037709172,
      "loss": 1.6262,
      "step": 22978
    },
    {
      "epoch": 1.9770283059451088,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006617425171634137,
      "loss": 0.6413,
      "step": 22979
    },
    {
      "epoch": 1.9771143422524307,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006617169300828446,
      "loss": 1.0332,
      "step": 22980
    },
    {
      "epoch": 1.9772003785597523,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006616913425292848,
      "loss": 1.0258,
      "step": 22981
    },
    {
      "epoch": 1.977286414867074,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006616657545028095,
      "loss": 1.3492,
      "step": 22982
    },
    {
      "epoch": 1.9773724511743955,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006616401660034932,
      "loss": 1.7021,
      "step": 22983
    },
    {
      "epoch": 1.9774584874817172,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006616145770314111,
      "loss": 1.9185,
      "step": 22984
    },
    {
      "epoch": 1.977544523789039,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0006615889875866375,
      "loss": 0.8911,
      "step": 22985
    },
    {
      "epoch": 1.9776305600963606,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006615633976692476,
      "loss": 1.0829,
      "step": 22986
    },
    {
      "epoch": 1.9777165964036825,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006615378072793162,
      "loss": 1.705,
      "step": 22987
    },
    {
      "epoch": 1.977802632711004,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006615122164169183,
      "loss": 1.0854,
      "step": 22988
    },
    {
      "epoch": 1.9778886690183257,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006614866250821283,
      "loss": 1.0958,
      "step": 22989
    },
    {
      "epoch": 1.9779747053256473,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006614610332750215,
      "loss": 0.7504,
      "step": 22990
    },
    {
      "epoch": 1.978060741632969,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006614354409956725,
      "loss": 1.0967,
      "step": 22991
    },
    {
      "epoch": 1.9781467779402908,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006614098482441564,
      "loss": 1.3184,
      "step": 22992
    },
    {
      "epoch": 1.9782328142476127,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006613842550205476,
      "loss": 1.3597,
      "step": 22993
    },
    {
      "epoch": 1.9783188505549343,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006613586613249214,
      "loss": 1.2683,
      "step": 22994
    },
    {
      "epoch": 1.978404886862256,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006613330671573525,
      "loss": 1.3625,
      "step": 22995
    },
    {
      "epoch": 1.9784909231695775,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006613074725179158,
      "loss": 1.3118,
      "step": 22996
    },
    {
      "epoch": 1.9785769594768992,
      "grad_norm": 1.796875,
      "learning_rate": 0.0006612818774066862,
      "loss": 1.1194,
      "step": 22997
    },
    {
      "epoch": 1.9786629957842208,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006612562818237382,
      "loss": 1.4248,
      "step": 22998
    },
    {
      "epoch": 1.9787490320915426,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006612306857691473,
      "loss": 1.7353,
      "step": 22999
    },
    {
      "epoch": 1.9788350683988645,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006612050892429876,
      "loss": 1.5592,
      "step": 23000
    },
    {
      "epoch": 1.978921104706186,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006611794922453345,
      "loss": 0.7755,
      "step": 23001
    },
    {
      "epoch": 1.9790071410135077,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006611538947762628,
      "loss": 1.8911,
      "step": 23002
    },
    {
      "epoch": 1.9790931773208293,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006611282968358473,
      "loss": 1.0259,
      "step": 23003
    },
    {
      "epoch": 1.979179213628151,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006611026984241628,
      "loss": 2.0144,
      "step": 23004
    },
    {
      "epoch": 1.9792652499354728,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006610770995412842,
      "loss": 1.177,
      "step": 23005
    },
    {
      "epoch": 1.9793512862427944,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006610515001872866,
      "loss": 1.2806,
      "step": 23006
    },
    {
      "epoch": 1.9794373225501163,
      "grad_norm": 1.125,
      "learning_rate": 0.0006610259003622447,
      "loss": 2.5015,
      "step": 23007
    },
    {
      "epoch": 1.979523358857438,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006610003000662332,
      "loss": 1.0343,
      "step": 23008
    },
    {
      "epoch": 1.9796093951647595,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006609746992993271,
      "loss": 0.675,
      "step": 23009
    },
    {
      "epoch": 1.9796954314720812,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006609490980616016,
      "loss": 1.2084,
      "step": 23010
    },
    {
      "epoch": 1.9797814677794028,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000660923496353131,
      "loss": 0.8386,
      "step": 23011
    },
    {
      "epoch": 1.9798675040867246,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006608978941739905,
      "loss": 1.6645,
      "step": 23012
    },
    {
      "epoch": 1.9799535403940463,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006608722915242551,
      "loss": 1.0882,
      "step": 23013
    },
    {
      "epoch": 1.980039576701368,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006608466884039994,
      "loss": 1.5087,
      "step": 23014
    },
    {
      "epoch": 1.9801256130086897,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006608210848132985,
      "loss": 1.1697,
      "step": 23015
    },
    {
      "epoch": 1.9802116493160113,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006607954807522271,
      "loss": 1.5442,
      "step": 23016
    },
    {
      "epoch": 1.980297685623333,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006607698762208604,
      "loss": 0.5926,
      "step": 23017
    },
    {
      "epoch": 1.9803837219306546,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000660744271219273,
      "loss": 1.8145,
      "step": 23018
    },
    {
      "epoch": 1.9804697582379764,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006607186657475396,
      "loss": 1.186,
      "step": 23019
    },
    {
      "epoch": 1.980555794545298,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006606930598057356,
      "loss": 2.3362,
      "step": 23020
    },
    {
      "epoch": 1.98064183085262,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006606674533939357,
      "loss": 1.6716,
      "step": 23021
    },
    {
      "epoch": 1.9807278671599415,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006606418465122146,
      "loss": 0.9933,
      "step": 23022
    },
    {
      "epoch": 1.9808139034672632,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006606162391606474,
      "loss": 1.8579,
      "step": 23023
    },
    {
      "epoch": 1.9808999397745848,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006605906313393089,
      "loss": 2.242,
      "step": 23024
    },
    {
      "epoch": 1.9809859760819064,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006605650230482742,
      "loss": 1.7502,
      "step": 23025
    },
    {
      "epoch": 1.9810720123892283,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000660539414287618,
      "loss": 1.6293,
      "step": 23026
    },
    {
      "epoch": 1.98115804869655,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006605138050574151,
      "loss": 1.2166,
      "step": 23027
    },
    {
      "epoch": 1.9812440850038717,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006604881953577406,
      "loss": 1.6805,
      "step": 23028
    },
    {
      "epoch": 1.9813301213111933,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006604625851886694,
      "loss": 1.407,
      "step": 23029
    },
    {
      "epoch": 1.981416157618515,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006604369745502762,
      "loss": 1.2561,
      "step": 23030
    },
    {
      "epoch": 1.9815021939258366,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006604113634426362,
      "loss": 0.9924,
      "step": 23031
    },
    {
      "epoch": 1.9815882302331584,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000660385751865824,
      "loss": 1.7221,
      "step": 23032
    },
    {
      "epoch": 1.98167426654048,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006603601398199149,
      "loss": 1.5571,
      "step": 23033
    },
    {
      "epoch": 1.981760302847802,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006603345273049835,
      "loss": 0.6647,
      "step": 23034
    },
    {
      "epoch": 1.9818463391551235,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006603089143211047,
      "loss": 1.594,
      "step": 23035
    },
    {
      "epoch": 1.9819323754624452,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006602833008683536,
      "loss": 1.2844,
      "step": 23036
    },
    {
      "epoch": 1.9820184117697668,
      "grad_norm": 3.0625,
      "learning_rate": 0.0006602576869468049,
      "loss": 1.3382,
      "step": 23037
    },
    {
      "epoch": 1.9821044480770884,
      "grad_norm": 1.125,
      "learning_rate": 0.0006602320725565337,
      "loss": 1.4208,
      "step": 23038
    },
    {
      "epoch": 1.9821904843844103,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006602064576976148,
      "loss": 1.7215,
      "step": 23039
    },
    {
      "epoch": 1.9822765206917319,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006601808423701232,
      "loss": 0.8665,
      "step": 23040
    },
    {
      "epoch": 1.9823625569990537,
      "grad_norm": 1.125,
      "learning_rate": 0.0006601552265741339,
      "loss": 1.6366,
      "step": 23041
    },
    {
      "epoch": 1.9824485933063754,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006601296103097216,
      "loss": 0.8868,
      "step": 23042
    },
    {
      "epoch": 1.982534629613697,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006601039935769613,
      "loss": 1.3124,
      "step": 23043
    },
    {
      "epoch": 1.9826206659210186,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006600783763759281,
      "loss": 1.1181,
      "step": 23044
    },
    {
      "epoch": 1.9827067022283402,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006600527587066966,
      "loss": 0.8775,
      "step": 23045
    },
    {
      "epoch": 1.982792738535662,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000660027140569342,
      "loss": 0.9234,
      "step": 23046
    },
    {
      "epoch": 1.9828787748429837,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006600015219639391,
      "loss": 1.3075,
      "step": 23047
    },
    {
      "epoch": 1.9829648111503055,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000659975902890563,
      "loss": 1.0734,
      "step": 23048
    },
    {
      "epoch": 1.9830508474576272,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006599502833492883,
      "loss": 1.9586,
      "step": 23049
    },
    {
      "epoch": 1.9831368837649488,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006599246633401903,
      "loss": 1.0473,
      "step": 23050
    },
    {
      "epoch": 1.9832229200722704,
      "grad_norm": 4.28125,
      "learning_rate": 0.0006598990428633438,
      "loss": 1.2158,
      "step": 23051
    },
    {
      "epoch": 1.983308956379592,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006598734219188235,
      "loss": 1.4574,
      "step": 23052
    },
    {
      "epoch": 1.9833949926869139,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006598478005067046,
      "loss": 0.9776,
      "step": 23053
    },
    {
      "epoch": 1.9834810289942357,
      "grad_norm": 1.140625,
      "learning_rate": 0.000659822178627062,
      "loss": 1.326,
      "step": 23054
    },
    {
      "epoch": 1.9835670653015574,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006597965562799707,
      "loss": 0.793,
      "step": 23055
    },
    {
      "epoch": 1.983653101608879,
      "grad_norm": 3.421875,
      "learning_rate": 0.0006597709334655055,
      "loss": 2.0547,
      "step": 23056
    },
    {
      "epoch": 1.9837391379162006,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006597453101837412,
      "loss": 1.1399,
      "step": 23057
    },
    {
      "epoch": 1.9838251742235222,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006597196864347531,
      "loss": 1.5363,
      "step": 23058
    },
    {
      "epoch": 1.983911210530844,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006596940622186161,
      "loss": 1.4066,
      "step": 23059
    },
    {
      "epoch": 1.9839972468381657,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006596684375354047,
      "loss": 1.213,
      "step": 23060
    },
    {
      "epoch": 1.9840832831454875,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0006596428123851945,
      "loss": 0.9957,
      "step": 23061
    },
    {
      "epoch": 1.9841693194528092,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00065961718676806,
      "loss": 1.304,
      "step": 23062
    },
    {
      "epoch": 1.9842553557601308,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006595915606840764,
      "loss": 0.9926,
      "step": 23063
    },
    {
      "epoch": 1.9843413920674524,
      "grad_norm": 1.609375,
      "learning_rate": 0.0006595659341333182,
      "loss": 2.083,
      "step": 23064
    },
    {
      "epoch": 1.984427428374774,
      "grad_norm": 1.125,
      "learning_rate": 0.000659540307115861,
      "loss": 1.3085,
      "step": 23065
    },
    {
      "epoch": 1.9845134646820959,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006595146796317793,
      "loss": 1.6591,
      "step": 23066
    },
    {
      "epoch": 1.9845995009894175,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006594890516811482,
      "loss": 1.246,
      "step": 23067
    },
    {
      "epoch": 1.9846855372967394,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006594634232640426,
      "loss": 1.4682,
      "step": 23068
    },
    {
      "epoch": 1.984771573604061,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006594377943805375,
      "loss": 1.2515,
      "step": 23069
    },
    {
      "epoch": 1.9848576099113826,
      "grad_norm": 1.140625,
      "learning_rate": 0.000659412165030708,
      "loss": 0.9631,
      "step": 23070
    },
    {
      "epoch": 1.9849436462187042,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0006593865352146287,
      "loss": 0.8946,
      "step": 23071
    },
    {
      "epoch": 1.9850296825260259,
      "grad_norm": 1.078125,
      "learning_rate": 0.000659360904932375,
      "loss": 1.1014,
      "step": 23072
    },
    {
      "epoch": 1.9851157188333477,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006593352741840216,
      "loss": 1.3842,
      "step": 23073
    },
    {
      "epoch": 1.9852017551406693,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006593096429696434,
      "loss": 1.3787,
      "step": 23074
    },
    {
      "epoch": 1.9852877914479912,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006592840112893154,
      "loss": 1.625,
      "step": 23075
    },
    {
      "epoch": 1.9853738277553128,
      "grad_norm": 1.0,
      "learning_rate": 0.0006592583791431129,
      "loss": 0.8985,
      "step": 23076
    },
    {
      "epoch": 1.9854598640626344,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006592327465311104,
      "loss": 1.9159,
      "step": 23077
    },
    {
      "epoch": 1.985545900369956,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006592071134533833,
      "loss": 1.5801,
      "step": 23078
    },
    {
      "epoch": 1.9856319366772777,
      "grad_norm": 1.1875,
      "learning_rate": 0.000659181479910006,
      "loss": 1.1686,
      "step": 23079
    },
    {
      "epoch": 1.9857179729845995,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006591558459010543,
      "loss": 1.5935,
      "step": 23080
    },
    {
      "epoch": 1.9858040092919214,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006591302114266024,
      "loss": 1.5503,
      "step": 23081
    },
    {
      "epoch": 1.985890045599243,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006591045764867257,
      "loss": 1.1463,
      "step": 23082
    },
    {
      "epoch": 1.9859760819065646,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006590789410814989,
      "loss": 1.1401,
      "step": 23083
    },
    {
      "epoch": 1.9860621182138862,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006590533052109973,
      "loss": 1.1929,
      "step": 23084
    },
    {
      "epoch": 1.9861481545212079,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006590276688752957,
      "loss": 1.2093,
      "step": 23085
    },
    {
      "epoch": 1.9862341908285295,
      "grad_norm": 1.71875,
      "learning_rate": 0.000659002032074469,
      "loss": 1.6008,
      "step": 23086
    },
    {
      "epoch": 1.9863202271358513,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006589763948085924,
      "loss": 1.5054,
      "step": 23087
    },
    {
      "epoch": 1.9864062634431732,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006589507570777407,
      "loss": 1.6117,
      "step": 23088
    },
    {
      "epoch": 1.9864922997504948,
      "grad_norm": 1.9609375,
      "learning_rate": 0.000658925118881989,
      "loss": 1.2354,
      "step": 23089
    },
    {
      "epoch": 1.9865783360578164,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006588994802214121,
      "loss": 1.7387,
      "step": 23090
    },
    {
      "epoch": 1.986664372365138,
      "grad_norm": 1.625,
      "learning_rate": 0.0006588738410960853,
      "loss": 1.3428,
      "step": 23091
    },
    {
      "epoch": 1.9867504086724597,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006588482015060834,
      "loss": 1.6981,
      "step": 23092
    },
    {
      "epoch": 1.9868364449797815,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006588225614514812,
      "loss": 1.2215,
      "step": 23093
    },
    {
      "epoch": 1.9869224812871031,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000658796920932354,
      "loss": 1.4482,
      "step": 23094
    },
    {
      "epoch": 1.987008517594425,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0006587712799487768,
      "loss": 1.3038,
      "step": 23095
    },
    {
      "epoch": 1.9870945539017466,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006587456385008243,
      "loss": 1.1897,
      "step": 23096
    },
    {
      "epoch": 1.9871805902090682,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000658719996588572,
      "loss": 1.9116,
      "step": 23097
    },
    {
      "epoch": 1.9872666265163899,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006586943542120943,
      "loss": 0.7342,
      "step": 23098
    },
    {
      "epoch": 1.9873526628237115,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006586687113714666,
      "loss": 1.4695,
      "step": 23099
    },
    {
      "epoch": 1.9874386991310333,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006586430680667637,
      "loss": 1.0635,
      "step": 23100
    },
    {
      "epoch": 1.987524735438355,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006586174242980606,
      "loss": 0.9756,
      "step": 23101
    },
    {
      "epoch": 1.9876107717456768,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006585917800654325,
      "loss": 1.0856,
      "step": 23102
    },
    {
      "epoch": 1.9876968080529984,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006585661353689544,
      "loss": 0.9545,
      "step": 23103
    },
    {
      "epoch": 1.98778284436032,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006585404902087011,
      "loss": 1.3035,
      "step": 23104
    },
    {
      "epoch": 1.9878688806676417,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006585148445847475,
      "loss": 1.7085,
      "step": 23105
    },
    {
      "epoch": 1.9879549169749633,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006584891984971691,
      "loss": 0.7783,
      "step": 23106
    },
    {
      "epoch": 1.9880409532822851,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006584635519460405,
      "loss": 1.3576,
      "step": 23107
    },
    {
      "epoch": 1.9881269895896068,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006584379049314369,
      "loss": 1.9925,
      "step": 23108
    },
    {
      "epoch": 1.9882130258969286,
      "grad_norm": 1.765625,
      "learning_rate": 0.0006584122574534331,
      "loss": 1.2471,
      "step": 23109
    },
    {
      "epoch": 1.9882990622042502,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006583866095121044,
      "loss": 1.4212,
      "step": 23110
    },
    {
      "epoch": 1.9883850985115719,
      "grad_norm": 1.640625,
      "learning_rate": 0.0006583609611075257,
      "loss": 1.4055,
      "step": 23111
    },
    {
      "epoch": 1.9884711348188935,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006583353122397717,
      "loss": 1.0311,
      "step": 23112
    },
    {
      "epoch": 1.988557171126215,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006583096629089178,
      "loss": 1.0039,
      "step": 23113
    },
    {
      "epoch": 1.988643207433537,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006582840131150392,
      "loss": 1.2362,
      "step": 23114
    },
    {
      "epoch": 1.9887292437408588,
      "grad_norm": 0.875,
      "learning_rate": 0.0006582583628582104,
      "loss": 1.4006,
      "step": 23115
    },
    {
      "epoch": 1.9888152800481804,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006582327121385067,
      "loss": 1.8627,
      "step": 23116
    },
    {
      "epoch": 1.988901316355502,
      "grad_norm": 1.125,
      "learning_rate": 0.0006582070609560031,
      "loss": 1.3526,
      "step": 23117
    },
    {
      "epoch": 1.9889873526628237,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006581814093107747,
      "loss": 2.0793,
      "step": 23118
    },
    {
      "epoch": 1.9890733889701453,
      "grad_norm": 1.671875,
      "learning_rate": 0.0006581557572028964,
      "loss": 1.8663,
      "step": 23119
    },
    {
      "epoch": 1.9891594252774671,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006581301046324433,
      "loss": 1.8423,
      "step": 23120
    },
    {
      "epoch": 1.9892454615847888,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006581044515994903,
      "loss": 1.6266,
      "step": 23121
    },
    {
      "epoch": 1.9893314978921106,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006580787981041126,
      "loss": 0.8308,
      "step": 23122
    },
    {
      "epoch": 1.9894175341994322,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006580531441463852,
      "loss": 1.1718,
      "step": 23123
    },
    {
      "epoch": 1.9895035705067539,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000658027489726383,
      "loss": 1.1688,
      "step": 23124
    },
    {
      "epoch": 1.9895896068140755,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006580018348441812,
      "loss": 1.0693,
      "step": 23125
    },
    {
      "epoch": 1.989675643121397,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006579761794998547,
      "loss": 1.2893,
      "step": 23126
    },
    {
      "epoch": 1.989761679428719,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006579505236934786,
      "loss": 0.7944,
      "step": 23127
    },
    {
      "epoch": 1.9898477157360406,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006579248674251279,
      "loss": 2.1795,
      "step": 23128
    },
    {
      "epoch": 1.9899337520433624,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006578992106948778,
      "loss": 0.9339,
      "step": 23129
    },
    {
      "epoch": 1.990019788350684,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0006578735535028031,
      "loss": 1.8921,
      "step": 23130
    },
    {
      "epoch": 1.9901058246580057,
      "grad_norm": 1.109375,
      "learning_rate": 0.000657847895848979,
      "loss": 0.8224,
      "step": 23131
    },
    {
      "epoch": 1.9901918609653273,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006578222377334807,
      "loss": 1.7487,
      "step": 23132
    },
    {
      "epoch": 1.990277897272649,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006577965791563827,
      "loss": 0.8997,
      "step": 23133
    },
    {
      "epoch": 1.9903639335799708,
      "grad_norm": 2.140625,
      "learning_rate": 0.0006577709201177606,
      "loss": 1.2915,
      "step": 23134
    },
    {
      "epoch": 1.9904499698872924,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006577452606176893,
      "loss": 1.0905,
      "step": 23135
    },
    {
      "epoch": 1.9905360061946142,
      "grad_norm": 1.125,
      "learning_rate": 0.0006577196006562436,
      "loss": 1.1274,
      "step": 23136
    },
    {
      "epoch": 1.9906220425019359,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006576939402334989,
      "loss": 0.6887,
      "step": 23137
    },
    {
      "epoch": 1.9907080788092575,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006576682793495301,
      "loss": 1.0961,
      "step": 23138
    },
    {
      "epoch": 1.990794115116579,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006576426180044121,
      "loss": 0.9967,
      "step": 23139
    },
    {
      "epoch": 1.9908801514239007,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006576169561982203,
      "loss": 1.1565,
      "step": 23140
    },
    {
      "epoch": 1.9909661877312226,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006575912939310295,
      "loss": 1.35,
      "step": 23141
    },
    {
      "epoch": 1.9910522240385444,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006575656312029145,
      "loss": 1.0454,
      "step": 23142
    },
    {
      "epoch": 1.991138260345866,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006575399680139511,
      "loss": 0.9187,
      "step": 23143
    },
    {
      "epoch": 1.9912242966531877,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006575143043642137,
      "loss": 1.6386,
      "step": 23144
    },
    {
      "epoch": 1.9913103329605093,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006574886402537779,
      "loss": 0.7764,
      "step": 23145
    },
    {
      "epoch": 1.991396369267831,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000657462975682718,
      "loss": 1.7887,
      "step": 23146
    },
    {
      "epoch": 1.9914824055751528,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006574373106511097,
      "loss": 1.6739,
      "step": 23147
    },
    {
      "epoch": 1.9915684418824744,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006574116451590282,
      "loss": 1.4563,
      "step": 23148
    },
    {
      "epoch": 1.9916544781897962,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006573859792065479,
      "loss": 1.0648,
      "step": 23149
    },
    {
      "epoch": 1.9917405144971179,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006573603127937442,
      "loss": 1.6842,
      "step": 23150
    },
    {
      "epoch": 1.9918265508044395,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006573346459206923,
      "loss": 1.8988,
      "step": 23151
    },
    {
      "epoch": 1.9919125871117611,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006573089785874671,
      "loss": 1.5059,
      "step": 23152
    },
    {
      "epoch": 1.9919986234190827,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006572833107941439,
      "loss": 2.2155,
      "step": 23153
    },
    {
      "epoch": 1.9920846597264046,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006572576425407974,
      "loss": 1.0164,
      "step": 23154
    },
    {
      "epoch": 1.9921706960337262,
      "grad_norm": 0.984375,
      "learning_rate": 0.000657231973827503,
      "loss": 1.037,
      "step": 23155
    },
    {
      "epoch": 1.992256732341048,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006572063046543357,
      "loss": 1.1528,
      "step": 23156
    },
    {
      "epoch": 1.9923427686483697,
      "grad_norm": 1.25,
      "learning_rate": 0.0006571806350213704,
      "loss": 1.1316,
      "step": 23157
    },
    {
      "epoch": 1.9924288049556913,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006571549649286824,
      "loss": 1.0264,
      "step": 23158
    },
    {
      "epoch": 1.992514841263013,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006571292943763468,
      "loss": 1.0591,
      "step": 23159
    },
    {
      "epoch": 1.9926008775703345,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006571036233644384,
      "loss": 1.0051,
      "step": 23160
    },
    {
      "epoch": 1.9926869138776564,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006570779518930324,
      "loss": 1.4212,
      "step": 23161
    },
    {
      "epoch": 1.992772950184978,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006570522799622041,
      "loss": 0.9559,
      "step": 23162
    },
    {
      "epoch": 1.9928589864922999,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006570266075720284,
      "loss": 0.9365,
      "step": 23163
    },
    {
      "epoch": 1.9929450227996215,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006570009347225803,
      "loss": 1.5193,
      "step": 23164
    },
    {
      "epoch": 1.9930310591069431,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006569752614139352,
      "loss": 1.4162,
      "step": 23165
    },
    {
      "epoch": 1.9931170954142647,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006569495876461678,
      "loss": 1.3635,
      "step": 23166
    },
    {
      "epoch": 1.9932031317215864,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006569239134193534,
      "loss": 0.8515,
      "step": 23167
    },
    {
      "epoch": 1.9932891680289082,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006568982387335671,
      "loss": 1.309,
      "step": 23168
    },
    {
      "epoch": 1.99337520433623,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006568725635888841,
      "loss": 1.0172,
      "step": 23169
    },
    {
      "epoch": 1.9934612406435517,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006568468879853791,
      "loss": 0.7296,
      "step": 23170
    },
    {
      "epoch": 1.9935472769508733,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006568212119231277,
      "loss": 1.5974,
      "step": 23171
    },
    {
      "epoch": 1.993633313258195,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006567955354022046,
      "loss": 1.6076,
      "step": 23172
    },
    {
      "epoch": 1.9937193495655166,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006567698584226852,
      "loss": 1.5826,
      "step": 23173
    },
    {
      "epoch": 1.9938053858728382,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006567441809846444,
      "loss": 2.0784,
      "step": 23174
    },
    {
      "epoch": 1.99389142218016,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0006567185030881574,
      "loss": 1.6679,
      "step": 23175
    },
    {
      "epoch": 1.9939774584874819,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006566928247332991,
      "loss": 1.536,
      "step": 23176
    },
    {
      "epoch": 1.9940634947948035,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006566671459201449,
      "loss": 0.6592,
      "step": 23177
    },
    {
      "epoch": 1.9941495311021251,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006566414666487697,
      "loss": 1.0805,
      "step": 23178
    },
    {
      "epoch": 1.9942355674094467,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006566157869192488,
      "loss": 1.2281,
      "step": 23179
    },
    {
      "epoch": 1.9943216037167684,
      "grad_norm": 0.90625,
      "learning_rate": 0.000656590106731657,
      "loss": 1.3665,
      "step": 23180
    },
    {
      "epoch": 1.9944076400240902,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0006565644260860697,
      "loss": 0.9701,
      "step": 23181
    },
    {
      "epoch": 1.9944936763314118,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000656538744982562,
      "loss": 0.7927,
      "step": 23182
    },
    {
      "epoch": 1.9945797126387337,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006565130634212087,
      "loss": 0.9401,
      "step": 23183
    },
    {
      "epoch": 1.9946657489460553,
      "grad_norm": 2.5,
      "learning_rate": 0.0006564873814020852,
      "loss": 1.9654,
      "step": 23184
    },
    {
      "epoch": 1.994751785253377,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006564616989252665,
      "loss": 0.7788,
      "step": 23185
    },
    {
      "epoch": 1.9948378215606986,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006564360159908278,
      "loss": 1.514,
      "step": 23186
    },
    {
      "epoch": 1.9949238578680202,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006564103325988441,
      "loss": 0.7936,
      "step": 23187
    },
    {
      "epoch": 1.995009894175342,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006563846487493908,
      "loss": 1.1116,
      "step": 23188
    },
    {
      "epoch": 1.9950959304826636,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006563589644425428,
      "loss": 1.1123,
      "step": 23189
    },
    {
      "epoch": 1.9951819667899855,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006563332796783752,
      "loss": 1.325,
      "step": 23190
    },
    {
      "epoch": 1.9952680030973071,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000656307594456963,
      "loss": 1.6134,
      "step": 23191
    },
    {
      "epoch": 1.9953540394046287,
      "grad_norm": 1.125,
      "learning_rate": 0.0006562819087783815,
      "loss": 0.8886,
      "step": 23192
    },
    {
      "epoch": 1.9954400757119504,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006562562226427061,
      "loss": 1.5886,
      "step": 23193
    },
    {
      "epoch": 1.995526112019272,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006562305360500115,
      "loss": 1.3437,
      "step": 23194
    },
    {
      "epoch": 1.9956121483265938,
      "grad_norm": 1.71875,
      "learning_rate": 0.0006562048490003728,
      "loss": 1.3497,
      "step": 23195
    },
    {
      "epoch": 1.9956981846339155,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006561791614938655,
      "loss": 1.812,
      "step": 23196
    },
    {
      "epoch": 1.9957842209412373,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006561534735305644,
      "loss": 1.7286,
      "step": 23197
    },
    {
      "epoch": 1.995870257248559,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006561277851105448,
      "loss": 0.7399,
      "step": 23198
    },
    {
      "epoch": 1.9959562935558806,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006561020962338819,
      "loss": 0.7113,
      "step": 23199
    },
    {
      "epoch": 1.9960423298632022,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006560764069006507,
      "loss": 1.6339,
      "step": 23200
    },
    {
      "epoch": 1.9961283661705238,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0006560507171109262,
      "loss": 1.3289,
      "step": 23201
    },
    {
      "epoch": 1.9962144024778457,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006560250268647838,
      "loss": 2.207,
      "step": 23202
    },
    {
      "epoch": 1.9963004387851675,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006559993361622986,
      "loss": 1.6314,
      "step": 23203
    },
    {
      "epoch": 1.9963864750924891,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006559736450035455,
      "loss": 1.6475,
      "step": 23204
    },
    {
      "epoch": 1.9964725113998107,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006559479533885999,
      "loss": 1.1483,
      "step": 23205
    },
    {
      "epoch": 1.9965585477071324,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006559222613175369,
      "loss": 0.8935,
      "step": 23206
    },
    {
      "epoch": 1.996644584014454,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006558965687904317,
      "loss": 1.2192,
      "step": 23207
    },
    {
      "epoch": 1.9967306203217758,
      "grad_norm": 1.125,
      "learning_rate": 0.0006558708758073593,
      "loss": 1.6555,
      "step": 23208
    },
    {
      "epoch": 1.9968166566290975,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006558451823683949,
      "loss": 1.2425,
      "step": 23209
    },
    {
      "epoch": 1.9969026929364193,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006558194884736136,
      "loss": 1.3205,
      "step": 23210
    },
    {
      "epoch": 1.996988729243741,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006557937941230907,
      "loss": 1.5605,
      "step": 23211
    },
    {
      "epoch": 1.9970747655510626,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006557680993169012,
      "loss": 1.1241,
      "step": 23212
    },
    {
      "epoch": 1.9971608018583842,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006557424040551202,
      "loss": 1.2215,
      "step": 23213
    },
    {
      "epoch": 1.9972468381657058,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006557167083378231,
      "loss": 1.3767,
      "step": 23214
    },
    {
      "epoch": 1.9973328744730277,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006556910121650848,
      "loss": 0.7709,
      "step": 23215
    },
    {
      "epoch": 1.9974189107803493,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006556653155369807,
      "loss": 1.4007,
      "step": 23216
    },
    {
      "epoch": 1.9975049470876711,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006556396184535857,
      "loss": 1.232,
      "step": 23217
    },
    {
      "epoch": 1.9975909833949927,
      "grad_norm": 1.0,
      "learning_rate": 0.0006556139209149752,
      "loss": 0.9544,
      "step": 23218
    },
    {
      "epoch": 1.9976770197023144,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006555882229212243,
      "loss": 1.4343,
      "step": 23219
    },
    {
      "epoch": 1.997763056009636,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006555625244724079,
      "loss": 1.1976,
      "step": 23220
    },
    {
      "epoch": 1.9978490923169576,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006555368255686015,
      "loss": 1.2768,
      "step": 23221
    },
    {
      "epoch": 1.9979351286242795,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006555111262098802,
      "loss": 2.1668,
      "step": 23222
    },
    {
      "epoch": 1.998021164931601,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006554854263963189,
      "loss": 1.3474,
      "step": 23223
    },
    {
      "epoch": 1.998107201238923,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006554597261279931,
      "loss": 2.025,
      "step": 23224
    },
    {
      "epoch": 1.9981932375462446,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006554340254049778,
      "loss": 1.4297,
      "step": 23225
    },
    {
      "epoch": 1.9982792738535662,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006554083242273482,
      "loss": 0.8343,
      "step": 23226
    },
    {
      "epoch": 1.9983653101608878,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006553826225951795,
      "loss": 0.7938,
      "step": 23227
    },
    {
      "epoch": 1.9984513464682094,
      "grad_norm": 1.125,
      "learning_rate": 0.0006553569205085468,
      "loss": 0.9555,
      "step": 23228
    },
    {
      "epoch": 1.9985373827755313,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006553312179675255,
      "loss": 0.53,
      "step": 23229
    },
    {
      "epoch": 1.9986234190828531,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006553055149721904,
      "loss": 0.7612,
      "step": 23230
    },
    {
      "epoch": 1.9987094553901747,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000655279811522617,
      "loss": 0.7041,
      "step": 23231
    },
    {
      "epoch": 1.9987954916974964,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006552541076188802,
      "loss": 1.263,
      "step": 23232
    },
    {
      "epoch": 1.998881528004818,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006552284032610555,
      "loss": 1.2597,
      "step": 23233
    },
    {
      "epoch": 1.9989675643121396,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006552026984492179,
      "loss": 1.8315,
      "step": 23234
    },
    {
      "epoch": 1.9990536006194615,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006551769931834425,
      "loss": 1.2845,
      "step": 23235
    },
    {
      "epoch": 1.999139636926783,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006551512874638046,
      "loss": 1.126,
      "step": 23236
    },
    {
      "epoch": 1.999225673234105,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006551255812903793,
      "loss": 0.7291,
      "step": 23237
    },
    {
      "epoch": 1.9993117095414266,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006550998746632419,
      "loss": 1.8383,
      "step": 23238
    },
    {
      "epoch": 1.9993977458487482,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006550741675824676,
      "loss": 1.4097,
      "step": 23239
    },
    {
      "epoch": 1.9994837821560698,
      "grad_norm": 1.0,
      "learning_rate": 0.0006550484600481314,
      "loss": 0.9109,
      "step": 23240
    },
    {
      "epoch": 1.9995698184633914,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006550227520603087,
      "loss": 1.4548,
      "step": 23241
    },
    {
      "epoch": 1.9996558547707133,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006549970436190745,
      "loss": 1.0238,
      "step": 23242
    },
    {
      "epoch": 1.999741891078035,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006549713347245042,
      "loss": 1.2155,
      "step": 23243
    },
    {
      "epoch": 1.9998279273853568,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006549456253766727,
      "loss": 0.9345,
      "step": 23244
    },
    {
      "epoch": 1.9999139636926784,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006549199155756557,
      "loss": 1.0816,
      "step": 23245
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.265625,
      "learning_rate": 0.0006548942053215278,
      "loss": 1.1366,
      "step": 23246
    },
    {
      "epoch": 2.0000860363073216,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006548684946143646,
      "loss": 0.8885,
      "step": 23247
    },
    {
      "epoch": 2.0001720726146432,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006548427834542412,
      "loss": 1.1847,
      "step": 23248
    },
    {
      "epoch": 2.0001720726146432,
      "eval_loss": 1.1906558275222778,
      "eval_runtime": 352.7214,
      "eval_samples_per_second": 13.18,
      "eval_steps_per_second": 1.65,
      "step": 23248
    },
    {
      "epoch": 2.000258108921965,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006548170718412329,
      "loss": 1.4021,
      "step": 23249
    },
    {
      "epoch": 2.000344145229287,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006547913597754145,
      "loss": 0.5963,
      "step": 23250
    },
    {
      "epoch": 2.0004301815366086,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006547656472568615,
      "loss": 1.0648,
      "step": 23251
    },
    {
      "epoch": 2.00051621784393,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006547399342856493,
      "loss": 1.8776,
      "step": 23252
    },
    {
      "epoch": 2.000602254151252,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006547142208618527,
      "loss": 1.443,
      "step": 23253
    },
    {
      "epoch": 2.0006882904585734,
      "grad_norm": 2.53125,
      "learning_rate": 0.000654688506985547,
      "loss": 1.5554,
      "step": 23254
    },
    {
      "epoch": 2.000774326765895,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006546627926568078,
      "loss": 1.301,
      "step": 23255
    },
    {
      "epoch": 2.0008603630732167,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006546370778757099,
      "loss": 1.7657,
      "step": 23256
    },
    {
      "epoch": 2.0009463993805388,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006546113626423286,
      "loss": 1.2881,
      "step": 23257
    },
    {
      "epoch": 2.0010324356878604,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000654585646956739,
      "loss": 1.0726,
      "step": 23258
    },
    {
      "epoch": 2.001118471995182,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006545599308190167,
      "loss": 0.898,
      "step": 23259
    },
    {
      "epoch": 2.0012045083025036,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006545342142292365,
      "loss": 1.5085,
      "step": 23260
    },
    {
      "epoch": 2.0012905446098253,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006545084971874737,
      "loss": 1.6289,
      "step": 23261
    },
    {
      "epoch": 2.001376580917147,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006544827796938036,
      "loss": 1.836,
      "step": 23262
    },
    {
      "epoch": 2.001462617224469,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006544570617483017,
      "loss": 0.8087,
      "step": 23263
    },
    {
      "epoch": 2.0015486535317906,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006544313433510427,
      "loss": 1.0902,
      "step": 23264
    },
    {
      "epoch": 2.001634689839112,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006544056245021022,
      "loss": 1.1552,
      "step": 23265
    },
    {
      "epoch": 2.001720726146434,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006543799052015551,
      "loss": 1.22,
      "step": 23266
    },
    {
      "epoch": 2.0018067624537554,
      "grad_norm": 1.125,
      "learning_rate": 0.0006543541854494769,
      "loss": 1.4344,
      "step": 23267
    },
    {
      "epoch": 2.001892798761077,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006543284652459427,
      "loss": 1.4622,
      "step": 23268
    },
    {
      "epoch": 2.0019788350683987,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0006543027445910277,
      "loss": 1.3144,
      "step": 23269
    },
    {
      "epoch": 2.0020648713757208,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006542770234848073,
      "loss": 1.1303,
      "step": 23270
    },
    {
      "epoch": 2.0021509076830424,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006542513019273566,
      "loss": 1.1117,
      "step": 23271
    },
    {
      "epoch": 2.002236943990364,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006542255799187509,
      "loss": 1.9826,
      "step": 23272
    },
    {
      "epoch": 2.0023229802976856,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006541998574590652,
      "loss": 0.6895,
      "step": 23273
    },
    {
      "epoch": 2.0024090166050073,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006541741345483749,
      "loss": 1.1062,
      "step": 23274
    },
    {
      "epoch": 2.002495052912329,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006541484111867554,
      "loss": 1.5039,
      "step": 23275
    },
    {
      "epoch": 2.0025810892196505,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006541226873742817,
      "loss": 1.3399,
      "step": 23276
    },
    {
      "epoch": 2.0026671255269726,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006540969631110291,
      "loss": 1.3275,
      "step": 23277
    },
    {
      "epoch": 2.002753161834294,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006540712383970728,
      "loss": 0.5419,
      "step": 23278
    },
    {
      "epoch": 2.002839198141616,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006540455132324883,
      "loss": 0.9395,
      "step": 23279
    },
    {
      "epoch": 2.0029252344489374,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006540197876173505,
      "loss": 1.3556,
      "step": 23280
    },
    {
      "epoch": 2.003011270756259,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006539940615517347,
      "loss": 1.0325,
      "step": 23281
    },
    {
      "epoch": 2.0030973070635807,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006539683350357163,
      "loss": 1.2679,
      "step": 23282
    },
    {
      "epoch": 2.0031833433709023,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0006539426080693706,
      "loss": 0.6206,
      "step": 23283
    },
    {
      "epoch": 2.0032693796782244,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006539168806527724,
      "loss": 0.8051,
      "step": 23284
    },
    {
      "epoch": 2.003355415985546,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006538911527859975,
      "loss": 1.5413,
      "step": 23285
    },
    {
      "epoch": 2.0034414522928676,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006538654244691209,
      "loss": 1.4725,
      "step": 23286
    },
    {
      "epoch": 2.0035274886001893,
      "grad_norm": 1.125,
      "learning_rate": 0.0006538396957022178,
      "loss": 0.5461,
      "step": 23287
    },
    {
      "epoch": 2.003613524907511,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006538139664853635,
      "loss": 1.2447,
      "step": 23288
    },
    {
      "epoch": 2.0036995612148325,
      "grad_norm": 1.0,
      "learning_rate": 0.0006537882368186332,
      "loss": 1.6289,
      "step": 23289
    },
    {
      "epoch": 2.003785597522154,
      "grad_norm": 1.125,
      "learning_rate": 0.0006537625067021024,
      "loss": 0.8685,
      "step": 23290
    },
    {
      "epoch": 2.003871633829476,
      "grad_norm": 1.0625,
      "learning_rate": 0.000653736776135846,
      "loss": 1.6568,
      "step": 23291
    },
    {
      "epoch": 2.003957670136798,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006537110451199395,
      "loss": 1.3184,
      "step": 23292
    },
    {
      "epoch": 2.0040437064441194,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006536853136544581,
      "loss": 0.9298,
      "step": 23293
    },
    {
      "epoch": 2.004129742751441,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000653659581739477,
      "loss": 1.2181,
      "step": 23294
    },
    {
      "epoch": 2.0042157790587627,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006536338493750715,
      "loss": 0.6446,
      "step": 23295
    },
    {
      "epoch": 2.0043018153660843,
      "grad_norm": 1.046875,
      "learning_rate": 0.000653608116561317,
      "loss": 1.1303,
      "step": 23296
    },
    {
      "epoch": 2.0043878516734064,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006535823832982885,
      "loss": 1.3815,
      "step": 23297
    },
    {
      "epoch": 2.004473887980728,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006535566495860614,
      "loss": 0.7914,
      "step": 23298
    },
    {
      "epoch": 2.0045599242880496,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006535309154247111,
      "loss": 0.9875,
      "step": 23299
    },
    {
      "epoch": 2.0046459605953713,
      "grad_norm": 1.125,
      "learning_rate": 0.0006535051808143125,
      "loss": 1.3817,
      "step": 23300
    },
    {
      "epoch": 2.004731996902693,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006534794457549413,
      "loss": 0.72,
      "step": 23301
    },
    {
      "epoch": 2.0048180332100145,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006534537102466726,
      "loss": 1.5861,
      "step": 23302
    },
    {
      "epoch": 2.004904069517336,
      "grad_norm": 1.5,
      "learning_rate": 0.0006534279742895815,
      "loss": 1.9775,
      "step": 23303
    },
    {
      "epoch": 2.004990105824658,
      "grad_norm": 0.875,
      "learning_rate": 0.0006534022378837434,
      "loss": 0.9903,
      "step": 23304
    },
    {
      "epoch": 2.00507614213198,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006533765010292337,
      "loss": 0.5908,
      "step": 23305
    },
    {
      "epoch": 2.0051621784393014,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006533507637261276,
      "loss": 1.2295,
      "step": 23306
    },
    {
      "epoch": 2.005248214746623,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006533250259745002,
      "loss": 0.8843,
      "step": 23307
    },
    {
      "epoch": 2.0053342510539447,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000653299287774427,
      "loss": 1.1296,
      "step": 23308
    },
    {
      "epoch": 2.0054202873612663,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006532735491259833,
      "loss": 0.9765,
      "step": 23309
    },
    {
      "epoch": 2.005506323668588,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000653247810029244,
      "loss": 0.907,
      "step": 23310
    },
    {
      "epoch": 2.00559235997591,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006532220704842848,
      "loss": 1.5527,
      "step": 23311
    },
    {
      "epoch": 2.0056783962832316,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000653196330491181,
      "loss": 1.3181,
      "step": 23312
    },
    {
      "epoch": 2.0057644325905533,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006531705900500076,
      "loss": 0.7322,
      "step": 23313
    },
    {
      "epoch": 2.005850468897875,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00065314484916084,
      "loss": 1.6979,
      "step": 23314
    },
    {
      "epoch": 2.0059365052051965,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006531191078237534,
      "loss": 1.0592,
      "step": 23315
    },
    {
      "epoch": 2.006022541512518,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006530933660388233,
      "loss": 0.689,
      "step": 23316
    },
    {
      "epoch": 2.0061085778198398,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006530676238061249,
      "loss": 1.2,
      "step": 23317
    },
    {
      "epoch": 2.006194614127162,
      "grad_norm": 8.75,
      "learning_rate": 0.0006530418811257335,
      "loss": 1.2526,
      "step": 23318
    },
    {
      "epoch": 2.0062806504344834,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006530161379977243,
      "loss": 1.0533,
      "step": 23319
    },
    {
      "epoch": 2.006366686741805,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006529903944221727,
      "loss": 0.7436,
      "step": 23320
    },
    {
      "epoch": 2.0064527230491267,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006529646503991539,
      "loss": 1.3026,
      "step": 23321
    },
    {
      "epoch": 2.0065387593564483,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006529389059287432,
      "loss": 1.7129,
      "step": 23322
    },
    {
      "epoch": 2.00662479566377,
      "grad_norm": 1.234375,
      "learning_rate": 0.000652913161011016,
      "loss": 0.8966,
      "step": 23323
    },
    {
      "epoch": 2.006710831971092,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006528874156460476,
      "loss": 1.4161,
      "step": 23324
    },
    {
      "epoch": 2.0067968682784136,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006528616698339132,
      "loss": 1.8488,
      "step": 23325
    },
    {
      "epoch": 2.0068829045857353,
      "grad_norm": 1.0,
      "learning_rate": 0.0006528359235746881,
      "loss": 1.5762,
      "step": 23326
    },
    {
      "epoch": 2.006968940893057,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006528101768684478,
      "loss": 1.6188,
      "step": 23327
    },
    {
      "epoch": 2.0070549772003785,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006527844297152674,
      "loss": 0.7567,
      "step": 23328
    },
    {
      "epoch": 2.0071410135077,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006527586821152221,
      "loss": 1.587,
      "step": 23329
    },
    {
      "epoch": 2.0072270498150218,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006527329340683875,
      "loss": 1.2222,
      "step": 23330
    },
    {
      "epoch": 2.007313086122344,
      "grad_norm": 1.0,
      "learning_rate": 0.0006527071855748388,
      "loss": 1.4493,
      "step": 23331
    },
    {
      "epoch": 2.0073991224296655,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006526814366346512,
      "loss": 0.9835,
      "step": 23332
    },
    {
      "epoch": 2.007485158736987,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006526556872479001,
      "loss": 1.3899,
      "step": 23333
    },
    {
      "epoch": 2.0075711950443087,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006526299374146607,
      "loss": 1.8212,
      "step": 23334
    },
    {
      "epoch": 2.0076572313516303,
      "grad_norm": 1.125,
      "learning_rate": 0.0006526041871350085,
      "loss": 0.7978,
      "step": 23335
    },
    {
      "epoch": 2.007743267658952,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000652578436409019,
      "loss": 0.8802,
      "step": 23336
    },
    {
      "epoch": 2.0078293039662736,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006525526852367669,
      "loss": 1.104,
      "step": 23337
    },
    {
      "epoch": 2.0079153402735956,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006525269336183279,
      "loss": 1.2498,
      "step": 23338
    },
    {
      "epoch": 2.0080013765809173,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006525011815537774,
      "loss": 1.331,
      "step": 23339
    },
    {
      "epoch": 2.008087412888239,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006524754290431905,
      "loss": 1.4854,
      "step": 23340
    },
    {
      "epoch": 2.0081734491955605,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006524496760866429,
      "loss": 1.6444,
      "step": 23341
    },
    {
      "epoch": 2.008259485502882,
      "grad_norm": 1.0,
      "learning_rate": 0.0006524239226842095,
      "loss": 1.8156,
      "step": 23342
    },
    {
      "epoch": 2.0083455218102038,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006523981688359657,
      "loss": 1.0412,
      "step": 23343
    },
    {
      "epoch": 2.0084315581175254,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006523724145419868,
      "loss": 1.811,
      "step": 23344
    },
    {
      "epoch": 2.0085175944248475,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006523466598023484,
      "loss": 1.2632,
      "step": 23345
    },
    {
      "epoch": 2.008603630732169,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006523209046171256,
      "loss": 1.194,
      "step": 23346
    },
    {
      "epoch": 2.0086896670394907,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006522951489863938,
      "loss": 1.2124,
      "step": 23347
    },
    {
      "epoch": 2.0087757033468123,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006522693929102283,
      "loss": 0.8128,
      "step": 23348
    },
    {
      "epoch": 2.008861739654134,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006522436363887044,
      "loss": 1.2564,
      "step": 23349
    },
    {
      "epoch": 2.0089477759614556,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006522178794218975,
      "loss": 1.9743,
      "step": 23350
    },
    {
      "epoch": 2.0090338122687776,
      "grad_norm": 5.0625,
      "learning_rate": 0.0006521921220098828,
      "loss": 0.5485,
      "step": 23351
    },
    {
      "epoch": 2.0091198485760993,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006521663641527358,
      "loss": 1.177,
      "step": 23352
    },
    {
      "epoch": 2.009205884883421,
      "grad_norm": 0.65625,
      "learning_rate": 0.0006521406058505318,
      "loss": 1.4001,
      "step": 23353
    },
    {
      "epoch": 2.0092919211907425,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006521148471033462,
      "loss": 0.9556,
      "step": 23354
    },
    {
      "epoch": 2.009377957498064,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006520890879112542,
      "loss": 0.8319,
      "step": 23355
    },
    {
      "epoch": 2.0094639938053858,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006520633282743311,
      "loss": 1.0893,
      "step": 23356
    },
    {
      "epoch": 2.0095500301127074,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006520375681926525,
      "loss": 0.4188,
      "step": 23357
    },
    {
      "epoch": 2.0096360664200295,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006520118076662935,
      "loss": 0.8243,
      "step": 23358
    },
    {
      "epoch": 2.009722102727351,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006519860466953296,
      "loss": 1.4916,
      "step": 23359
    },
    {
      "epoch": 2.0098081390346727,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006519602852798359,
      "loss": 0.6891,
      "step": 23360
    },
    {
      "epoch": 2.0098941753419943,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006519345234198881,
      "loss": 0.6677,
      "step": 23361
    },
    {
      "epoch": 2.009980211649316,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006519087611155614,
      "loss": 1.0975,
      "step": 23362
    },
    {
      "epoch": 2.0100662479566376,
      "grad_norm": 2.40625,
      "learning_rate": 0.0006518829983669309,
      "loss": 1.597,
      "step": 23363
    },
    {
      "epoch": 2.010152284263959,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006518572351740723,
      "loss": 1.0545,
      "step": 23364
    },
    {
      "epoch": 2.0102383205712813,
      "grad_norm": 0.875,
      "learning_rate": 0.0006518314715370609,
      "loss": 1.3383,
      "step": 23365
    },
    {
      "epoch": 2.010324356878603,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006518057074559718,
      "loss": 0.9816,
      "step": 23366
    },
    {
      "epoch": 2.0104103931859245,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006517799429308806,
      "loss": 0.7616,
      "step": 23367
    },
    {
      "epoch": 2.010496429493246,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006517541779618626,
      "loss": 0.9238,
      "step": 23368
    },
    {
      "epoch": 2.0105824658005678,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006517284125489933,
      "loss": 1.2758,
      "step": 23369
    },
    {
      "epoch": 2.0106685021078894,
      "grad_norm": 1.0,
      "learning_rate": 0.0006517026466923477,
      "loss": 1.4202,
      "step": 23370
    },
    {
      "epoch": 2.010754538415211,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006516768803920013,
      "loss": 1.4018,
      "step": 23371
    },
    {
      "epoch": 2.010840574722533,
      "grad_norm": 1.125,
      "learning_rate": 0.0006516511136480296,
      "loss": 1.2747,
      "step": 23372
    },
    {
      "epoch": 2.0109266110298547,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006516253464605079,
      "loss": 1.589,
      "step": 23373
    },
    {
      "epoch": 2.0110126473371763,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006515995788295116,
      "loss": 1.4531,
      "step": 23374
    },
    {
      "epoch": 2.011098683644498,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006515738107551161,
      "loss": 1.3232,
      "step": 23375
    },
    {
      "epoch": 2.0111847199518196,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006515480422373965,
      "loss": 0.9354,
      "step": 23376
    },
    {
      "epoch": 2.011270756259141,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006515222732764286,
      "loss": 1.3249,
      "step": 23377
    },
    {
      "epoch": 2.0113567925664633,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006514965038722872,
      "loss": 1.5476,
      "step": 23378
    },
    {
      "epoch": 2.011442828873785,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006514707340250481,
      "loss": 0.8056,
      "step": 23379
    },
    {
      "epoch": 2.0115288651811065,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006514449637347867,
      "loss": 1.1798,
      "step": 23380
    },
    {
      "epoch": 2.011614901488428,
      "grad_norm": 1.21875,
      "learning_rate": 0.000651419193001578,
      "loss": 1.4743,
      "step": 23381
    },
    {
      "epoch": 2.0117009377957498,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006513934218254978,
      "loss": 0.9294,
      "step": 23382
    },
    {
      "epoch": 2.0117869741030714,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006513676502066212,
      "loss": 1.0091,
      "step": 23383
    },
    {
      "epoch": 2.011873010410393,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006513418781450238,
      "loss": 1.4961,
      "step": 23384
    },
    {
      "epoch": 2.011959046717715,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006513161056407806,
      "loss": 1.556,
      "step": 23385
    },
    {
      "epoch": 2.0120450830250367,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006512903326939674,
      "loss": 0.7069,
      "step": 23386
    },
    {
      "epoch": 2.0121311193323583,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0006512645593046595,
      "loss": 1.7971,
      "step": 23387
    },
    {
      "epoch": 2.01221715563968,
      "grad_norm": 5.84375,
      "learning_rate": 0.000651238785472932,
      "loss": 1.3618,
      "step": 23388
    },
    {
      "epoch": 2.0123031919470016,
      "grad_norm": 1.671875,
      "learning_rate": 0.0006512130111988603,
      "loss": 1.5414,
      "step": 23389
    },
    {
      "epoch": 2.012389228254323,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006511872364825202,
      "loss": 1.0079,
      "step": 23390
    },
    {
      "epoch": 2.012475264561645,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006511614613239869,
      "loss": 0.9852,
      "step": 23391
    },
    {
      "epoch": 2.012561300868967,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006511356857233356,
      "loss": 1.1329,
      "step": 23392
    },
    {
      "epoch": 2.0126473371762885,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006511099096806418,
      "loss": 1.3498,
      "step": 23393
    },
    {
      "epoch": 2.01273337348361,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000651084133195981,
      "loss": 0.9577,
      "step": 23394
    },
    {
      "epoch": 2.0128194097909318,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006510583562694283,
      "loss": 2.0343,
      "step": 23395
    },
    {
      "epoch": 2.0129054460982534,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006510325789010594,
      "loss": 1.195,
      "step": 23396
    },
    {
      "epoch": 2.012991482405575,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006510068010909496,
      "loss": 1.4344,
      "step": 23397
    },
    {
      "epoch": 2.0130775187128966,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006509810228391744,
      "loss": 0.8975,
      "step": 23398
    },
    {
      "epoch": 2.0131635550202187,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000650955244145809,
      "loss": 1.1785,
      "step": 23399
    },
    {
      "epoch": 2.0132495913275403,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006509294650109287,
      "loss": 1.4615,
      "step": 23400
    },
    {
      "epoch": 2.013335627634862,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006509036854346092,
      "loss": 1.0095,
      "step": 23401
    },
    {
      "epoch": 2.0134216639421836,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006508779054169258,
      "loss": 0.7338,
      "step": 23402
    },
    {
      "epoch": 2.013507700249505,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006508521249579539,
      "loss": 1.145,
      "step": 23403
    },
    {
      "epoch": 2.013593736556827,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006508263440577687,
      "loss": 1.3668,
      "step": 23404
    },
    {
      "epoch": 2.0136797728641485,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006508005627164459,
      "loss": 0.6705,
      "step": 23405
    },
    {
      "epoch": 2.0137658091714705,
      "grad_norm": 1.0,
      "learning_rate": 0.0006507747809340608,
      "loss": 0.8765,
      "step": 23406
    },
    {
      "epoch": 2.013851845478792,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006507489987106887,
      "loss": 1.2318,
      "step": 23407
    },
    {
      "epoch": 2.0139378817861138,
      "grad_norm": 1.171875,
      "learning_rate": 0.000650723216046405,
      "loss": 1.9308,
      "step": 23408
    },
    {
      "epoch": 2.0140239180934354,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006506974329412855,
      "loss": 1.2698,
      "step": 23409
    },
    {
      "epoch": 2.014109954400757,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006506716493954051,
      "loss": 1.2413,
      "step": 23410
    },
    {
      "epoch": 2.0141959907080786,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006506458654088397,
      "loss": 2.1182,
      "step": 23411
    },
    {
      "epoch": 2.0142820270154007,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006506200809816641,
      "loss": 0.6698,
      "step": 23412
    },
    {
      "epoch": 2.0143680633227223,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006505942961139541,
      "loss": 1.5135,
      "step": 23413
    },
    {
      "epoch": 2.014454099630044,
      "grad_norm": 1.625,
      "learning_rate": 0.0006505685108057853,
      "loss": 1.5006,
      "step": 23414
    },
    {
      "epoch": 2.0145401359373656,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006505427250572326,
      "loss": 1.3914,
      "step": 23415
    },
    {
      "epoch": 2.014626172244687,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000650516938868372,
      "loss": 1.4599,
      "step": 23416
    },
    {
      "epoch": 2.014712208552009,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006504911522392786,
      "loss": 1.0901,
      "step": 23417
    },
    {
      "epoch": 2.0147982448593305,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006504653651700277,
      "loss": 1.073,
      "step": 23418
    },
    {
      "epoch": 2.0148842811666525,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006504395776606948,
      "loss": 1.0597,
      "step": 23419
    },
    {
      "epoch": 2.014970317473974,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006504137897113555,
      "loss": 0.7057,
      "step": 23420
    },
    {
      "epoch": 2.0150563537812958,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006503880013220854,
      "loss": 0.6857,
      "step": 23421
    },
    {
      "epoch": 2.0151423900886174,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006503622124929591,
      "loss": 1.0729,
      "step": 23422
    },
    {
      "epoch": 2.015228426395939,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000650336423224053,
      "loss": 1.5799,
      "step": 23423
    },
    {
      "epoch": 2.0153144627032606,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006503106335154419,
      "loss": 1.0476,
      "step": 23424
    },
    {
      "epoch": 2.0154004990105823,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006502848433672015,
      "loss": 0.7864,
      "step": 23425
    },
    {
      "epoch": 2.0154865353179043,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006502590527794071,
      "loss": 0.6609,
      "step": 23426
    },
    {
      "epoch": 2.015572571625226,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006502332617521342,
      "loss": 0.6558,
      "step": 23427
    },
    {
      "epoch": 2.0156586079325476,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006502074702854583,
      "loss": 1.99,
      "step": 23428
    },
    {
      "epoch": 2.015744644239869,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006501816783794547,
      "loss": 1.3674,
      "step": 23429
    },
    {
      "epoch": 2.015830680547191,
      "grad_norm": 0.875,
      "learning_rate": 0.0006501558860341989,
      "loss": 0.9134,
      "step": 23430
    },
    {
      "epoch": 2.0159167168545125,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006501300932497665,
      "loss": 0.9481,
      "step": 23431
    },
    {
      "epoch": 2.016002753161834,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006501043000262326,
      "loss": 1.4144,
      "step": 23432
    },
    {
      "epoch": 2.016088789469156,
      "grad_norm": 1.125,
      "learning_rate": 0.0006500785063636729,
      "loss": 0.851,
      "step": 23433
    },
    {
      "epoch": 2.0161748257764778,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006500527122621627,
      "loss": 1.4656,
      "step": 23434
    },
    {
      "epoch": 2.0162608620837994,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006500269177217774,
      "loss": 1.3574,
      "step": 23435
    },
    {
      "epoch": 2.016346898391121,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006500011227425928,
      "loss": 0.9704,
      "step": 23436
    },
    {
      "epoch": 2.0164329346984426,
      "grad_norm": 1.375,
      "learning_rate": 0.0006499753273246839,
      "loss": 1.2112,
      "step": 23437
    },
    {
      "epoch": 2.0165189710057643,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006499495314681264,
      "loss": 1.3469,
      "step": 23438
    },
    {
      "epoch": 2.0166050073130863,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006499237351729957,
      "loss": 1.4227,
      "step": 23439
    },
    {
      "epoch": 2.016691043620408,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006498979384393672,
      "loss": 1.6614,
      "step": 23440
    },
    {
      "epoch": 2.0167770799277296,
      "grad_norm": 1.375,
      "learning_rate": 0.0006498721412673163,
      "loss": 1.4873,
      "step": 23441
    },
    {
      "epoch": 2.016863116235051,
      "grad_norm": 1.125,
      "learning_rate": 0.0006498463436569186,
      "loss": 1.9531,
      "step": 23442
    },
    {
      "epoch": 2.016949152542373,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006498205456082496,
      "loss": 0.926,
      "step": 23443
    },
    {
      "epoch": 2.0170351888496945,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006497947471213845,
      "loss": 0.6803,
      "step": 23444
    },
    {
      "epoch": 2.017121225157016,
      "grad_norm": 2.890625,
      "learning_rate": 0.0006497689481963989,
      "loss": 1.4225,
      "step": 23445
    },
    {
      "epoch": 2.017207261464338,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006497431488333683,
      "loss": 1.4556,
      "step": 23446
    },
    {
      "epoch": 2.0172932977716598,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006497173490323682,
      "loss": 1.4004,
      "step": 23447
    },
    {
      "epoch": 2.0173793340789814,
      "grad_norm": 1.671875,
      "learning_rate": 0.000649691548793474,
      "loss": 1.4239,
      "step": 23448
    },
    {
      "epoch": 2.017465370386303,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006496657481167608,
      "loss": 1.0258,
      "step": 23449
    },
    {
      "epoch": 2.0175514066936246,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006496399470023046,
      "loss": 0.8956,
      "step": 23450
    },
    {
      "epoch": 2.0176374430009463,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006496141454501808,
      "loss": 1.5805,
      "step": 23451
    },
    {
      "epoch": 2.017723479308268,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006495883434604646,
      "loss": 0.8985,
      "step": 23452
    },
    {
      "epoch": 2.01780951561559,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006495625410332316,
      "loss": 1.4893,
      "step": 23453
    },
    {
      "epoch": 2.0178955519229116,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006495367381685573,
      "loss": 1.4527,
      "step": 23454
    },
    {
      "epoch": 2.017981588230233,
      "grad_norm": 1.0,
      "learning_rate": 0.000649510934866517,
      "loss": 1.5455,
      "step": 23455
    },
    {
      "epoch": 2.018067624537555,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006494851311271864,
      "loss": 1.2967,
      "step": 23456
    },
    {
      "epoch": 2.0181536608448765,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006494593269506408,
      "loss": 1.2308,
      "step": 23457
    },
    {
      "epoch": 2.018239697152198,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006494335223369558,
      "loss": 1.3068,
      "step": 23458
    },
    {
      "epoch": 2.0183257334595197,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0006494077172862068,
      "loss": 1.3715,
      "step": 23459
    },
    {
      "epoch": 2.018411769766842,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006493819117984692,
      "loss": 0.7678,
      "step": 23460
    },
    {
      "epoch": 2.0184978060741634,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006493561058738186,
      "loss": 1.5434,
      "step": 23461
    },
    {
      "epoch": 2.018583842381485,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006493302995123306,
      "loss": 0.9807,
      "step": 23462
    },
    {
      "epoch": 2.0186698786888067,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006493044927140803,
      "loss": 1.3654,
      "step": 23463
    },
    {
      "epoch": 2.0187559149961283,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006492786854791434,
      "loss": 1.373,
      "step": 23464
    },
    {
      "epoch": 2.01884195130345,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006492528778075955,
      "loss": 1.8855,
      "step": 23465
    },
    {
      "epoch": 2.0189279876107715,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000649227069699512,
      "loss": 1.0924,
      "step": 23466
    },
    {
      "epoch": 2.0190140239180936,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006492012611549683,
      "loss": 1.5406,
      "step": 23467
    },
    {
      "epoch": 2.019100060225415,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006491754521740397,
      "loss": 0.7971,
      "step": 23468
    },
    {
      "epoch": 2.019186096532737,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006491496427568021,
      "loss": 0.8247,
      "step": 23469
    },
    {
      "epoch": 2.0192721328400585,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006491238329033309,
      "loss": 0.9536,
      "step": 23470
    },
    {
      "epoch": 2.01935816914738,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006490980226137014,
      "loss": 1.3968,
      "step": 23471
    },
    {
      "epoch": 2.0194442054547017,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0006490722118879891,
      "loss": 0.4946,
      "step": 23472
    },
    {
      "epoch": 2.019530241762024,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006490464007262698,
      "loss": 0.6874,
      "step": 23473
    },
    {
      "epoch": 2.0196162780693454,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006490205891286185,
      "loss": 1.2298,
      "step": 23474
    },
    {
      "epoch": 2.019702314376667,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000648994777095111,
      "loss": 1.2041,
      "step": 23475
    },
    {
      "epoch": 2.0197883506839887,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006489689646258229,
      "loss": 1.0695,
      "step": 23476
    },
    {
      "epoch": 2.0198743869913103,
      "grad_norm": 1.125,
      "learning_rate": 0.0006489431517208295,
      "loss": 1.4123,
      "step": 23477
    },
    {
      "epoch": 2.019960423298632,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006489173383802063,
      "loss": 1.2751,
      "step": 23478
    },
    {
      "epoch": 2.0200464596059535,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006488915246040289,
      "loss": 1.0885,
      "step": 23479
    },
    {
      "epoch": 2.0201324959132756,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006488657103923727,
      "loss": 1.3481,
      "step": 23480
    },
    {
      "epoch": 2.020218532220597,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006488398957453132,
      "loss": 0.9213,
      "step": 23481
    },
    {
      "epoch": 2.020304568527919,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006488140806629262,
      "loss": 1.3869,
      "step": 23482
    },
    {
      "epoch": 2.0203906048352405,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006487882651452867,
      "loss": 1.6297,
      "step": 23483
    },
    {
      "epoch": 2.020476641142562,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006487624491924705,
      "loss": 1.041,
      "step": 23484
    },
    {
      "epoch": 2.0205626774498837,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006487366328045532,
      "loss": 1.825,
      "step": 23485
    },
    {
      "epoch": 2.0206487137572053,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00064871081598161,
      "loss": 1.0589,
      "step": 23486
    },
    {
      "epoch": 2.0207347500645274,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006486849987237166,
      "loss": 0.788,
      "step": 23487
    },
    {
      "epoch": 2.020820786371849,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006486591810309488,
      "loss": 1.2585,
      "step": 23488
    },
    {
      "epoch": 2.0209068226791707,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006486333629033816,
      "loss": 2.2527,
      "step": 23489
    },
    {
      "epoch": 2.0209928589864923,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006486075443410906,
      "loss": 1.6875,
      "step": 23490
    },
    {
      "epoch": 2.021078895293814,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006485817253441516,
      "loss": 1.3755,
      "step": 23491
    },
    {
      "epoch": 2.0211649316011355,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006485559059126399,
      "loss": 0.6448,
      "step": 23492
    },
    {
      "epoch": 2.021250967908457,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000648530086046631,
      "loss": 1.1405,
      "step": 23493
    },
    {
      "epoch": 2.021337004215779,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006485042657462006,
      "loss": 1.2822,
      "step": 23494
    },
    {
      "epoch": 2.021423040523101,
      "grad_norm": 0.875,
      "learning_rate": 0.000648478445011424,
      "loss": 1.4449,
      "step": 23495
    },
    {
      "epoch": 2.0215090768304225,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006484526238423769,
      "loss": 1.0207,
      "step": 23496
    },
    {
      "epoch": 2.021595113137744,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006484268022391347,
      "loss": 1.3121,
      "step": 23497
    },
    {
      "epoch": 2.0216811494450657,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006484009802017731,
      "loss": 0.7436,
      "step": 23498
    },
    {
      "epoch": 2.0217671857523873,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006483751577303674,
      "loss": 1.6117,
      "step": 23499
    },
    {
      "epoch": 2.0218532220597094,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006483493348249931,
      "loss": 1.9014,
      "step": 23500
    },
    {
      "epoch": 2.021939258367031,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006483235114857259,
      "loss": 1.2691,
      "step": 23501
    },
    {
      "epoch": 2.0220252946743527,
      "grad_norm": 1.625,
      "learning_rate": 0.0006482976877126414,
      "loss": 1.4545,
      "step": 23502
    },
    {
      "epoch": 2.0221113309816743,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006482718635058149,
      "loss": 0.7137,
      "step": 23503
    },
    {
      "epoch": 2.022197367288996,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006482460388653222,
      "loss": 1.3326,
      "step": 23504
    },
    {
      "epoch": 2.0222834035963175,
      "grad_norm": 1.578125,
      "learning_rate": 0.0006482202137912384,
      "loss": 1.4901,
      "step": 23505
    },
    {
      "epoch": 2.022369439903639,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006481943882836396,
      "loss": 1.1547,
      "step": 23506
    },
    {
      "epoch": 2.0224554762109612,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006481685623426008,
      "loss": 0.8929,
      "step": 23507
    },
    {
      "epoch": 2.022541512518283,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006481427359681979,
      "loss": 1.1544,
      "step": 23508
    },
    {
      "epoch": 2.0226275488256045,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006481169091605062,
      "loss": 1.6275,
      "step": 23509
    },
    {
      "epoch": 2.022713585132926,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006480910819196016,
      "loss": 1.669,
      "step": 23510
    },
    {
      "epoch": 2.0227996214402477,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006480652542455591,
      "loss": 1.8687,
      "step": 23511
    },
    {
      "epoch": 2.0228856577475693,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006480394261384546,
      "loss": 0.6039,
      "step": 23512
    },
    {
      "epoch": 2.022971694054891,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006480135975983636,
      "loss": 1.313,
      "step": 23513
    },
    {
      "epoch": 2.023057730362213,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006479877686253616,
      "loss": 1.1569,
      "step": 23514
    },
    {
      "epoch": 2.0231437666695347,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006479619392195243,
      "loss": 0.8904,
      "step": 23515
    },
    {
      "epoch": 2.0232298029768563,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006479361093809268,
      "loss": 0.8063,
      "step": 23516
    },
    {
      "epoch": 2.023315839284178,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006479102791096452,
      "loss": 1.3603,
      "step": 23517
    },
    {
      "epoch": 2.0234018755914995,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006478844484057546,
      "loss": 1.3794,
      "step": 23518
    },
    {
      "epoch": 2.023487911898821,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000647858617269331,
      "loss": 1.3777,
      "step": 23519
    },
    {
      "epoch": 2.023573948206143,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006478327857004496,
      "loss": 0.7362,
      "step": 23520
    },
    {
      "epoch": 2.023659984513465,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006478069536991859,
      "loss": 0.9999,
      "step": 23521
    },
    {
      "epoch": 2.0237460208207865,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000647781121265616,
      "loss": 1.144,
      "step": 23522
    },
    {
      "epoch": 2.023832057128108,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006477552883998147,
      "loss": 1.1842,
      "step": 23523
    },
    {
      "epoch": 2.0239180934354297,
      "grad_norm": 1.25,
      "learning_rate": 0.000647729455101858,
      "loss": 1.6397,
      "step": 23524
    },
    {
      "epoch": 2.0240041297427513,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006477036213718215,
      "loss": 1.4937,
      "step": 23525
    },
    {
      "epoch": 2.024090166050073,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006476777872097806,
      "loss": 1.4103,
      "step": 23526
    },
    {
      "epoch": 2.024176202357395,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006476519526158106,
      "loss": 0.8244,
      "step": 23527
    },
    {
      "epoch": 2.0242622386647167,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006476261175899877,
      "loss": 0.9532,
      "step": 23528
    },
    {
      "epoch": 2.0243482749720383,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000647600282132387,
      "loss": 0.8674,
      "step": 23529
    },
    {
      "epoch": 2.02443431127936,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006475744462430842,
      "loss": 1.035,
      "step": 23530
    },
    {
      "epoch": 2.0245203475866815,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006475486099221548,
      "loss": 0.9078,
      "step": 23531
    },
    {
      "epoch": 2.024606383894003,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006475227731696744,
      "loss": 0.8212,
      "step": 23532
    },
    {
      "epoch": 2.024692420201325,
      "grad_norm": 1.84375,
      "learning_rate": 0.0006474969359857187,
      "loss": 1.0805,
      "step": 23533
    },
    {
      "epoch": 2.024778456508647,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000647471098370363,
      "loss": 1.0112,
      "step": 23534
    },
    {
      "epoch": 2.0248644928159685,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006474452603236831,
      "loss": 1.7772,
      "step": 23535
    },
    {
      "epoch": 2.02495052912329,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006474194218457545,
      "loss": 1.2873,
      "step": 23536
    },
    {
      "epoch": 2.0250365654306117,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006473935829366527,
      "loss": 1.3071,
      "step": 23537
    },
    {
      "epoch": 2.0251226017379333,
      "grad_norm": 1.375,
      "learning_rate": 0.0006473677435964534,
      "loss": 1.366,
      "step": 23538
    },
    {
      "epoch": 2.025208638045255,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006473419038252321,
      "loss": 1.061,
      "step": 23539
    },
    {
      "epoch": 2.0252946743525766,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006473160636230644,
      "loss": 1.0983,
      "step": 23540
    },
    {
      "epoch": 2.0253807106598987,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006472902229900259,
      "loss": 1.1803,
      "step": 23541
    },
    {
      "epoch": 2.0254667469672203,
      "grad_norm": 1.296875,
      "learning_rate": 0.000647264381926192,
      "loss": 1.5396,
      "step": 23542
    },
    {
      "epoch": 2.025552783274542,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006472385404316385,
      "loss": 0.7714,
      "step": 23543
    },
    {
      "epoch": 2.0256388195818635,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006472126985064409,
      "loss": 1.0013,
      "step": 23544
    },
    {
      "epoch": 2.025724855889185,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006471868561506749,
      "loss": 1.148,
      "step": 23545
    },
    {
      "epoch": 2.025810892196507,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006471610133644158,
      "loss": 1.0362,
      "step": 23546
    },
    {
      "epoch": 2.0258969285038284,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006471351701477395,
      "loss": 1.5035,
      "step": 23547
    },
    {
      "epoch": 2.0259829648111505,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006471093265007214,
      "loss": 1.2269,
      "step": 23548
    },
    {
      "epoch": 2.026069001118472,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000647083482423437,
      "loss": 1.3315,
      "step": 23549
    },
    {
      "epoch": 2.0261550374257937,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006470576379159622,
      "loss": 1.3559,
      "step": 23550
    },
    {
      "epoch": 2.0262410737331153,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006470317929783724,
      "loss": 1.7226,
      "step": 23551
    },
    {
      "epoch": 2.026327110040437,
      "grad_norm": 1.65625,
      "learning_rate": 0.0006470059476107432,
      "loss": 1.4233,
      "step": 23552
    },
    {
      "epoch": 2.0264131463477586,
      "grad_norm": 1.28125,
      "learning_rate": 0.00064698010181315,
      "loss": 1.85,
      "step": 23553
    },
    {
      "epoch": 2.0264991826550807,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006469542555856688,
      "loss": 1.5872,
      "step": 23554
    },
    {
      "epoch": 2.0265852189624023,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006469284089283748,
      "loss": 1.1634,
      "step": 23555
    },
    {
      "epoch": 2.026671255269724,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006469025618413439,
      "loss": 0.699,
      "step": 23556
    },
    {
      "epoch": 2.0267572915770455,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006468767143246514,
      "loss": 1.5714,
      "step": 23557
    },
    {
      "epoch": 2.026843327884367,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006468508663783735,
      "loss": 2.1673,
      "step": 23558
    },
    {
      "epoch": 2.026929364191689,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000646825018002585,
      "loss": 0.6138,
      "step": 23559
    },
    {
      "epoch": 2.0270154004990104,
      "grad_norm": 0.90625,
      "learning_rate": 0.000646799169197362,
      "loss": 0.831,
      "step": 23560
    },
    {
      "epoch": 2.0271014368063325,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00064677331996278,
      "loss": 0.7145,
      "step": 23561
    },
    {
      "epoch": 2.027187473113654,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006467474702989145,
      "loss": 0.8819,
      "step": 23562
    },
    {
      "epoch": 2.0272735094209757,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006467216202058414,
      "loss": 0.5705,
      "step": 23563
    },
    {
      "epoch": 2.0273595457282974,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006466957696836359,
      "loss": 0.8308,
      "step": 23564
    },
    {
      "epoch": 2.027445582035619,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006466699187323738,
      "loss": 1.5608,
      "step": 23565
    },
    {
      "epoch": 2.0275316183429406,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006466440673521307,
      "loss": 1.5876,
      "step": 23566
    },
    {
      "epoch": 2.0276176546502622,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006466182155429825,
      "loss": 0.8098,
      "step": 23567
    },
    {
      "epoch": 2.0277036909575843,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0006465923633050041,
      "loss": 2.0293,
      "step": 23568
    },
    {
      "epoch": 2.027789727264906,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006465665106382719,
      "loss": 1.3799,
      "step": 23569
    },
    {
      "epoch": 2.0278757635722275,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006465406575428611,
      "loss": 0.7874,
      "step": 23570
    },
    {
      "epoch": 2.027961799879549,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006465148040188473,
      "loss": 1.1244,
      "step": 23571
    },
    {
      "epoch": 2.028047836186871,
      "grad_norm": 13.4375,
      "learning_rate": 0.0006464889500663062,
      "loss": 1.111,
      "step": 23572
    },
    {
      "epoch": 2.0281338724941924,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006464630956853135,
      "loss": 1.6391,
      "step": 23573
    },
    {
      "epoch": 2.028219908801514,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006464372408759444,
      "loss": 1.1602,
      "step": 23574
    },
    {
      "epoch": 2.028305945108836,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006464113856382752,
      "loss": 1.0337,
      "step": 23575
    },
    {
      "epoch": 2.0283919814161577,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006463855299723811,
      "loss": 0.6717,
      "step": 23576
    },
    {
      "epoch": 2.0284780177234794,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006463596738783377,
      "loss": 1.0536,
      "step": 23577
    },
    {
      "epoch": 2.028564054030801,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006463338173562209,
      "loss": 1.1981,
      "step": 23578
    },
    {
      "epoch": 2.0286500903381226,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006463079604061059,
      "loss": 1.4788,
      "step": 23579
    },
    {
      "epoch": 2.0287361266454442,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006462821030280688,
      "loss": 1.5216,
      "step": 23580
    },
    {
      "epoch": 2.028822162952766,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006462562452221849,
      "loss": 1.6308,
      "step": 23581
    },
    {
      "epoch": 2.028908199260088,
      "grad_norm": 1.03125,
      "learning_rate": 0.00064623038698853,
      "loss": 1.5931,
      "step": 23582
    },
    {
      "epoch": 2.0289942355674095,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006462045283271795,
      "loss": 2.1701,
      "step": 23583
    },
    {
      "epoch": 2.029080271874731,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006461786692382092,
      "loss": 0.9753,
      "step": 23584
    },
    {
      "epoch": 2.029166308182053,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006461528097216948,
      "loss": 1.4174,
      "step": 23585
    },
    {
      "epoch": 2.0292523444893744,
      "grad_norm": 0.828125,
      "learning_rate": 0.000646126949777712,
      "loss": 1.3115,
      "step": 23586
    },
    {
      "epoch": 2.029338380796696,
      "grad_norm": 1.21875,
      "learning_rate": 0.000646101089406336,
      "loss": 1.5458,
      "step": 23587
    },
    {
      "epoch": 2.029424417104018,
      "grad_norm": 1.375,
      "learning_rate": 0.000646075228607643,
      "loss": 1.3086,
      "step": 23588
    },
    {
      "epoch": 2.0295104534113397,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006460493673817083,
      "loss": 1.5869,
      "step": 23589
    },
    {
      "epoch": 2.0295964897186614,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006460235057286074,
      "loss": 1.7647,
      "step": 23590
    },
    {
      "epoch": 2.029682526025983,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006459976436484164,
      "loss": 1.6963,
      "step": 23591
    },
    {
      "epoch": 2.0297685623333046,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006459717811412105,
      "loss": 0.7574,
      "step": 23592
    },
    {
      "epoch": 2.0298545986406262,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006459459182070656,
      "loss": 1.1392,
      "step": 23593
    },
    {
      "epoch": 2.029940634947948,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006459200548460573,
      "loss": 1.207,
      "step": 23594
    },
    {
      "epoch": 2.03002667125527,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006458941910582612,
      "loss": 0.6985,
      "step": 23595
    },
    {
      "epoch": 2.0301127075625915,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000645868326843753,
      "loss": 1.3103,
      "step": 23596
    },
    {
      "epoch": 2.030198743869913,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006458424622026083,
      "loss": 1.3632,
      "step": 23597
    },
    {
      "epoch": 2.030284780177235,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006458165971349027,
      "loss": 1.7537,
      "step": 23598
    },
    {
      "epoch": 2.0303708164845564,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000645790731640712,
      "loss": 1.1439,
      "step": 23599
    },
    {
      "epoch": 2.030456852791878,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006457648657201118,
      "loss": 1.122,
      "step": 23600
    },
    {
      "epoch": 2.0305428890991997,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006457389993731777,
      "loss": 0.9022,
      "step": 23601
    },
    {
      "epoch": 2.0306289254065217,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006457131325999853,
      "loss": 1.7744,
      "step": 23602
    },
    {
      "epoch": 2.0307149617138434,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006456872654006103,
      "loss": 0.7233,
      "step": 23603
    },
    {
      "epoch": 2.030800998021165,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006456613977751284,
      "loss": 1.0014,
      "step": 23604
    },
    {
      "epoch": 2.0308870343284866,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006456355297236152,
      "loss": 1.9215,
      "step": 23605
    },
    {
      "epoch": 2.0309730706358082,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006456096612461465,
      "loss": 0.9262,
      "step": 23606
    },
    {
      "epoch": 2.03105910694313,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006455837923427978,
      "loss": 1.4896,
      "step": 23607
    },
    {
      "epoch": 2.0311451432504515,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006455579230136451,
      "loss": 1.6869,
      "step": 23608
    },
    {
      "epoch": 2.0312311795577735,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006455320532587633,
      "loss": 1.0619,
      "step": 23609
    },
    {
      "epoch": 2.031317215865095,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006455061830782288,
      "loss": 1.2308,
      "step": 23610
    },
    {
      "epoch": 2.031403252172417,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006454803124721169,
      "loss": 0.9761,
      "step": 23611
    },
    {
      "epoch": 2.0314892884797384,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006454544414405036,
      "loss": 1.3532,
      "step": 23612
    },
    {
      "epoch": 2.03157532478706,
      "grad_norm": 1.375,
      "learning_rate": 0.0006454285699834642,
      "loss": 1.1653,
      "step": 23613
    },
    {
      "epoch": 2.0316613610943817,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006454026981010746,
      "loss": 1.0083,
      "step": 23614
    },
    {
      "epoch": 2.0317473974017037,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006453768257934102,
      "loss": 1.1206,
      "step": 23615
    },
    {
      "epoch": 2.0318334337090254,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000645350953060547,
      "loss": 0.7464,
      "step": 23616
    },
    {
      "epoch": 2.031919470016347,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006453250799025605,
      "loss": 1.3972,
      "step": 23617
    },
    {
      "epoch": 2.0320055063236686,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006452992063195265,
      "loss": 0.7764,
      "step": 23618
    },
    {
      "epoch": 2.0320915426309902,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006452733323115206,
      "loss": 1.1656,
      "step": 23619
    },
    {
      "epoch": 2.032177578938312,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006452474578786183,
      "loss": 1.3788,
      "step": 23620
    },
    {
      "epoch": 2.0322636152456335,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006452215830208954,
      "loss": 1.1977,
      "step": 23621
    },
    {
      "epoch": 2.0323496515529555,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006451957077384279,
      "loss": 1.0511,
      "step": 23622
    },
    {
      "epoch": 2.032435687860277,
      "grad_norm": 1.109375,
      "learning_rate": 0.000645169832031291,
      "loss": 1.2203,
      "step": 23623
    },
    {
      "epoch": 2.032521724167599,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006451439558995604,
      "loss": 1.4825,
      "step": 23624
    },
    {
      "epoch": 2.0326077604749204,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006451180793433123,
      "loss": 1.2743,
      "step": 23625
    },
    {
      "epoch": 2.032693796782242,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006450922023626218,
      "loss": 0.8478,
      "step": 23626
    },
    {
      "epoch": 2.0327798330895637,
      "grad_norm": 1.171875,
      "learning_rate": 0.000645066324957565,
      "loss": 1.4629,
      "step": 23627
    },
    {
      "epoch": 2.0328658693968853,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006450404471282172,
      "loss": 1.199,
      "step": 23628
    },
    {
      "epoch": 2.0329519057042074,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006450145688746544,
      "loss": 1.0627,
      "step": 23629
    },
    {
      "epoch": 2.033037942011529,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006449886901969522,
      "loss": 1.1881,
      "step": 23630
    },
    {
      "epoch": 2.0331239783188506,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006449628110951864,
      "loss": 1.5923,
      "step": 23631
    },
    {
      "epoch": 2.0332100146261722,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006449369315694323,
      "loss": 1.2074,
      "step": 23632
    },
    {
      "epoch": 2.033296050933494,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000644911051619766,
      "loss": 1.0301,
      "step": 23633
    },
    {
      "epoch": 2.0333820872408155,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006448851712462633,
      "loss": 1.3049,
      "step": 23634
    },
    {
      "epoch": 2.033468123548137,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006448592904489994,
      "loss": 0.8001,
      "step": 23635
    },
    {
      "epoch": 2.033554159855459,
      "grad_norm": 1.125,
      "learning_rate": 0.0006448334092280502,
      "loss": 0.9429,
      "step": 23636
    },
    {
      "epoch": 2.033640196162781,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006448075275834916,
      "loss": 0.8308,
      "step": 23637
    },
    {
      "epoch": 2.0337262324701024,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000644781645515399,
      "loss": 1.3991,
      "step": 23638
    },
    {
      "epoch": 2.033812268777424,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006447557630238483,
      "loss": 1.4474,
      "step": 23639
    },
    {
      "epoch": 2.0338983050847457,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006447298801089152,
      "loss": 1.7794,
      "step": 23640
    },
    {
      "epoch": 2.0339843413920673,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006447039967706754,
      "loss": 0.8478,
      "step": 23641
    },
    {
      "epoch": 2.034070377699389,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006446781130092045,
      "loss": 1.9269,
      "step": 23642
    },
    {
      "epoch": 2.034156414006711,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006446522288245782,
      "loss": 0.8007,
      "step": 23643
    },
    {
      "epoch": 2.0342424503140326,
      "grad_norm": 0.875,
      "learning_rate": 0.0006446263442168723,
      "loss": 0.6222,
      "step": 23644
    },
    {
      "epoch": 2.0343284866213542,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006446004591861625,
      "loss": 1.7789,
      "step": 23645
    },
    {
      "epoch": 2.034414522928676,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006445745737325244,
      "loss": 1.0995,
      "step": 23646
    },
    {
      "epoch": 2.0345005592359975,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006445486878560338,
      "loss": 1.9255,
      "step": 23647
    },
    {
      "epoch": 2.034586595543319,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006445228015567663,
      "loss": 1.3707,
      "step": 23648
    },
    {
      "epoch": 2.034672631850641,
      "grad_norm": 0.796875,
      "learning_rate": 0.000644496914834798,
      "loss": 0.9151,
      "step": 23649
    },
    {
      "epoch": 2.034758668157963,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006444710276902042,
      "loss": 1.6312,
      "step": 23650
    },
    {
      "epoch": 2.0348447044652844,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006444451401230606,
      "loss": 1.6213,
      "step": 23651
    },
    {
      "epoch": 2.034930740772606,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006444192521334432,
      "loss": 1.7272,
      "step": 23652
    },
    {
      "epoch": 2.0350167770799277,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006443933637214277,
      "loss": 1.5937,
      "step": 23653
    },
    {
      "epoch": 2.0351028133872493,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006443674748870894,
      "loss": 1.4188,
      "step": 23654
    },
    {
      "epoch": 2.035188849694571,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006443415856305045,
      "loss": 1.0178,
      "step": 23655
    },
    {
      "epoch": 2.035274886001893,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006443156959517485,
      "loss": 1.0885,
      "step": 23656
    },
    {
      "epoch": 2.0353609223092146,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006442898058508972,
      "loss": 1.3834,
      "step": 23657
    },
    {
      "epoch": 2.0354469586165362,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006442639153280262,
      "loss": 1.5246,
      "step": 23658
    },
    {
      "epoch": 2.035532994923858,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006442380243832114,
      "loss": 1.4952,
      "step": 23659
    },
    {
      "epoch": 2.0356190312311795,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006442121330165284,
      "loss": 0.7921,
      "step": 23660
    },
    {
      "epoch": 2.035705067538501,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006441862412280528,
      "loss": 0.9107,
      "step": 23661
    },
    {
      "epoch": 2.0357911038458227,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006441603490178607,
      "loss": 1.0652,
      "step": 23662
    },
    {
      "epoch": 2.035877140153145,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006441344563860275,
      "loss": 0.9759,
      "step": 23663
    },
    {
      "epoch": 2.0359631764604664,
      "grad_norm": 30.25,
      "learning_rate": 0.0006441085633326291,
      "loss": 1.1762,
      "step": 23664
    },
    {
      "epoch": 2.036049212767788,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006440826698577411,
      "loss": 0.7825,
      "step": 23665
    },
    {
      "epoch": 2.0361352490751097,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006440567759614393,
      "loss": 0.9189,
      "step": 23666
    },
    {
      "epoch": 2.0362212853824313,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006440308816437996,
      "loss": 0.6416,
      "step": 23667
    },
    {
      "epoch": 2.036307321689753,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006440049869048975,
      "loss": 1.8523,
      "step": 23668
    },
    {
      "epoch": 2.0363933579970745,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006439790917448087,
      "loss": 0.8369,
      "step": 23669
    },
    {
      "epoch": 2.0364793943043966,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006439531961636093,
      "loss": 0.7643,
      "step": 23670
    },
    {
      "epoch": 2.0365654306117182,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006439273001613747,
      "loss": 0.6752,
      "step": 23671
    },
    {
      "epoch": 2.03665146691904,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006439014037381807,
      "loss": 0.9282,
      "step": 23672
    },
    {
      "epoch": 2.0367375032263615,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006438755068941031,
      "loss": 1.0361,
      "step": 23673
    },
    {
      "epoch": 2.036823539533683,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006438496096292177,
      "loss": 0.979,
      "step": 23674
    },
    {
      "epoch": 2.0369095758410047,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006438237119436001,
      "loss": 2.0642,
      "step": 23675
    },
    {
      "epoch": 2.036995612148327,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006437978138373262,
      "loss": 0.9112,
      "step": 23676
    },
    {
      "epoch": 2.0370816484556484,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006437719153104714,
      "loss": 1.3801,
      "step": 23677
    },
    {
      "epoch": 2.03716768476297,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006437460163631119,
      "loss": 1.3341,
      "step": 23678
    },
    {
      "epoch": 2.0372537210702917,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006437201169953233,
      "loss": 1.715,
      "step": 23679
    },
    {
      "epoch": 2.0373397573776133,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006436942172071813,
      "loss": 1.4003,
      "step": 23680
    },
    {
      "epoch": 2.037425793684935,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006436683169987616,
      "loss": 1.3929,
      "step": 23681
    },
    {
      "epoch": 2.0375118299922566,
      "grad_norm": 1.1875,
      "learning_rate": 0.00064364241637014,
      "loss": 1.5218,
      "step": 23682
    },
    {
      "epoch": 2.0375978662995786,
      "grad_norm": 1.25,
      "learning_rate": 0.0006436165153213924,
      "loss": 1.4687,
      "step": 23683
    },
    {
      "epoch": 2.0376839026069002,
      "grad_norm": 0.765625,
      "learning_rate": 0.0006435906138525942,
      "loss": 0.7507,
      "step": 23684
    },
    {
      "epoch": 2.037769938914222,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006435647119638215,
      "loss": 1.7381,
      "step": 23685
    },
    {
      "epoch": 2.0378559752215435,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006435388096551499,
      "loss": 1.7865,
      "step": 23686
    },
    {
      "epoch": 2.037942011528865,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006435129069266554,
      "loss": 1.2872,
      "step": 23687
    },
    {
      "epoch": 2.0380280478361867,
      "grad_norm": 1.0,
      "learning_rate": 0.0006434870037784133,
      "loss": 1.1599,
      "step": 23688
    },
    {
      "epoch": 2.0381140841435084,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006434611002104996,
      "loss": 1.4898,
      "step": 23689
    },
    {
      "epoch": 2.0382001204508304,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006434351962229903,
      "loss": 1.4021,
      "step": 23690
    },
    {
      "epoch": 2.038286156758152,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006434092918159609,
      "loss": 1.5895,
      "step": 23691
    },
    {
      "epoch": 2.0383721930654737,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006433833869894871,
      "loss": 0.8629,
      "step": 23692
    },
    {
      "epoch": 2.0384582293727953,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000643357481743645,
      "loss": 1.9438,
      "step": 23693
    },
    {
      "epoch": 2.038544265680117,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006433315760785099,
      "loss": 1.8273,
      "step": 23694
    },
    {
      "epoch": 2.0386303019874386,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006433056699941579,
      "loss": 1.4629,
      "step": 23695
    },
    {
      "epoch": 2.03871633829476,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006432797634906646,
      "loss": 0.8381,
      "step": 23696
    },
    {
      "epoch": 2.0388023746020822,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000643253856568106,
      "loss": 2.2141,
      "step": 23697
    },
    {
      "epoch": 2.038888410909404,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006432279492265578,
      "loss": 1.4335,
      "step": 23698
    },
    {
      "epoch": 2.0389744472167255,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006432020414660955,
      "loss": 1.688,
      "step": 23699
    },
    {
      "epoch": 2.039060483524047,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006431761332867951,
      "loss": 1.1451,
      "step": 23700
    },
    {
      "epoch": 2.0391465198313687,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006431502246887326,
      "loss": 1.8849,
      "step": 23701
    },
    {
      "epoch": 2.0392325561386904,
      "grad_norm": 1.78125,
      "learning_rate": 0.0006431243156719833,
      "loss": 1.5551,
      "step": 23702
    },
    {
      "epoch": 2.0393185924460124,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006430984062366232,
      "loss": 1.1013,
      "step": 23703
    },
    {
      "epoch": 2.039404628753334,
      "grad_norm": 1.65625,
      "learning_rate": 0.0006430724963827282,
      "loss": 1.3781,
      "step": 23704
    },
    {
      "epoch": 2.0394906650606557,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006430465861103739,
      "loss": 1.1082,
      "step": 23705
    },
    {
      "epoch": 2.0395767013679773,
      "grad_norm": 1.5,
      "learning_rate": 0.0006430206754196362,
      "loss": 1.0256,
      "step": 23706
    },
    {
      "epoch": 2.039662737675299,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006429947643105908,
      "loss": 1.0991,
      "step": 23707
    },
    {
      "epoch": 2.0397487739826206,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006429688527833136,
      "loss": 1.4381,
      "step": 23708
    },
    {
      "epoch": 2.039834810289942,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006429429408378803,
      "loss": 1.0867,
      "step": 23709
    },
    {
      "epoch": 2.0399208465972642,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006429170284743666,
      "loss": 1.2542,
      "step": 23710
    },
    {
      "epoch": 2.040006882904586,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006428911156928486,
      "loss": 1.0437,
      "step": 23711
    },
    {
      "epoch": 2.0400929192119075,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006428652024934017,
      "loss": 1.8092,
      "step": 23712
    },
    {
      "epoch": 2.040178955519229,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006428392888761018,
      "loss": 1.1222,
      "step": 23713
    },
    {
      "epoch": 2.0402649918265507,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000642813374841025,
      "loss": 1.5083,
      "step": 23714
    },
    {
      "epoch": 2.0403510281338724,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006427874603882466,
      "loss": 0.9695,
      "step": 23715
    },
    {
      "epoch": 2.040437064441194,
      "grad_norm": 1.703125,
      "learning_rate": 0.0006427615455178429,
      "loss": 1.2742,
      "step": 23716
    },
    {
      "epoch": 2.040523100748516,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006427356302298894,
      "loss": 1.8907,
      "step": 23717
    },
    {
      "epoch": 2.0406091370558377,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006427097145244617,
      "loss": 0.9939,
      "step": 23718
    },
    {
      "epoch": 2.0406951733631593,
      "grad_norm": 1.0625,
      "learning_rate": 0.000642683798401636,
      "loss": 1.0084,
      "step": 23719
    },
    {
      "epoch": 2.040781209670481,
      "grad_norm": 1.28125,
      "learning_rate": 0.000642657881861488,
      "loss": 1.4557,
      "step": 23720
    },
    {
      "epoch": 2.0408672459778026,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006426319649040932,
      "loss": 1.1595,
      "step": 23721
    },
    {
      "epoch": 2.040953282285124,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006426060475295279,
      "loss": 1.3304,
      "step": 23722
    },
    {
      "epoch": 2.041039318592446,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006425801297378675,
      "loss": 1.5698,
      "step": 23723
    },
    {
      "epoch": 2.041125354899768,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006425542115291881,
      "loss": 1.0718,
      "step": 23724
    },
    {
      "epoch": 2.0412113912070895,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006425282929035652,
      "loss": 1.4458,
      "step": 23725
    },
    {
      "epoch": 2.041297427514411,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006425023738610749,
      "loss": 0.609,
      "step": 23726
    },
    {
      "epoch": 2.0413834638217327,
      "grad_norm": 1.5,
      "learning_rate": 0.0006424764544017928,
      "loss": 2.0184,
      "step": 23727
    },
    {
      "epoch": 2.0414695001290544,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006424505345257949,
      "loss": 1.9306,
      "step": 23728
    },
    {
      "epoch": 2.041555536436376,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006424246142331567,
      "loss": 1.0175,
      "step": 23729
    },
    {
      "epoch": 2.041641572743698,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006423986935239543,
      "loss": 0.7637,
      "step": 23730
    },
    {
      "epoch": 2.0417276090510197,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006423727723982634,
      "loss": 1.3719,
      "step": 23731
    },
    {
      "epoch": 2.0418136453583413,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006423468508561599,
      "loss": 1.6299,
      "step": 23732
    },
    {
      "epoch": 2.041899681665663,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006423209288977194,
      "loss": 0.9492,
      "step": 23733
    },
    {
      "epoch": 2.0419857179729846,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000642295006523018,
      "loss": 1.0215,
      "step": 23734
    },
    {
      "epoch": 2.042071754280306,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006422690837321314,
      "loss": 1.0606,
      "step": 23735
    },
    {
      "epoch": 2.042157790587628,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006422431605251352,
      "loss": 1.2801,
      "step": 23736
    },
    {
      "epoch": 2.04224382689495,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006422172369021055,
      "loss": 0.8401,
      "step": 23737
    },
    {
      "epoch": 2.0423298632022715,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006421913128631182,
      "loss": 1.1312,
      "step": 23738
    },
    {
      "epoch": 2.042415899509593,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006421653884082489,
      "loss": 1.2787,
      "step": 23739
    },
    {
      "epoch": 2.0425019358169147,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006421394635375734,
      "loss": 1.5122,
      "step": 23740
    },
    {
      "epoch": 2.0425879721242364,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006421135382511676,
      "loss": 0.8056,
      "step": 23741
    },
    {
      "epoch": 2.042674008431558,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006420876125491073,
      "loss": 0.8086,
      "step": 23742
    },
    {
      "epoch": 2.0427600447388796,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006420616864314686,
      "loss": 1.0243,
      "step": 23743
    },
    {
      "epoch": 2.0428460810462017,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006420357598983269,
      "loss": 0.606,
      "step": 23744
    },
    {
      "epoch": 2.0429321173535233,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006420098329497582,
      "loss": 1.2739,
      "step": 23745
    },
    {
      "epoch": 2.043018153660845,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006419839055858386,
      "loss": 1.0426,
      "step": 23746
    },
    {
      "epoch": 2.0431041899681666,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006419579778066435,
      "loss": 1.3676,
      "step": 23747
    },
    {
      "epoch": 2.043190226275488,
      "grad_norm": 1.25,
      "learning_rate": 0.0006419320496122488,
      "loss": 1.4699,
      "step": 23748
    },
    {
      "epoch": 2.04327626258281,
      "grad_norm": 1.96875,
      "learning_rate": 0.0006419061210027308,
      "loss": 1.6477,
      "step": 23749
    },
    {
      "epoch": 2.0433622988901314,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006418801919781648,
      "loss": 1.152,
      "step": 23750
    },
    {
      "epoch": 2.0434483351974535,
      "grad_norm": 1.625,
      "learning_rate": 0.0006418542625386268,
      "loss": 1.4727,
      "step": 23751
    },
    {
      "epoch": 2.043534371504775,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006418283326841927,
      "loss": 0.8846,
      "step": 23752
    },
    {
      "epoch": 2.0436204078120968,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006418024024149384,
      "loss": 1.4077,
      "step": 23753
    },
    {
      "epoch": 2.0437064441194184,
      "grad_norm": 1.125,
      "learning_rate": 0.0006417764717309395,
      "loss": 0.8619,
      "step": 23754
    },
    {
      "epoch": 2.04379248042674,
      "grad_norm": 1.328125,
      "learning_rate": 0.000641750540632272,
      "loss": 0.9852,
      "step": 23755
    },
    {
      "epoch": 2.0438785167340616,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006417246091190118,
      "loss": 1.8016,
      "step": 23756
    },
    {
      "epoch": 2.0439645530413832,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006416986771912347,
      "loss": 1.5129,
      "step": 23757
    },
    {
      "epoch": 2.0440505893487053,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006416727448490165,
      "loss": 0.7322,
      "step": 23758
    },
    {
      "epoch": 2.044136625656027,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006416468120924331,
      "loss": 1.0249,
      "step": 23759
    },
    {
      "epoch": 2.0442226619633486,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006416208789215603,
      "loss": 1.8512,
      "step": 23760
    },
    {
      "epoch": 2.04430869827067,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000641594945336474,
      "loss": 1.1381,
      "step": 23761
    },
    {
      "epoch": 2.044394734577992,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006415690113372501,
      "loss": 1.4944,
      "step": 23762
    },
    {
      "epoch": 2.0444807708853134,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0006415430769239642,
      "loss": 0.9446,
      "step": 23763
    },
    {
      "epoch": 2.0445668071926355,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006415171420966926,
      "loss": 1.2007,
      "step": 23764
    },
    {
      "epoch": 2.044652843499957,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006414912068555106,
      "loss": 0.7079,
      "step": 23765
    },
    {
      "epoch": 2.0447388798072788,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006414652712004944,
      "loss": 1.6755,
      "step": 23766
    },
    {
      "epoch": 2.0448249161146004,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006414393351317198,
      "loss": 1.2181,
      "step": 23767
    },
    {
      "epoch": 2.044910952421922,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006414133986492628,
      "loss": 0.665,
      "step": 23768
    },
    {
      "epoch": 2.0449969887292436,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006413874617531989,
      "loss": 1.904,
      "step": 23769
    },
    {
      "epoch": 2.0450830250365652,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006413615244436043,
      "loss": 1.0659,
      "step": 23770
    },
    {
      "epoch": 2.0451690613438873,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0006413355867205548,
      "loss": 0.9414,
      "step": 23771
    },
    {
      "epoch": 2.045255097651209,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006413096485841261,
      "loss": 1.2133,
      "step": 23772
    },
    {
      "epoch": 2.0453411339585306,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006412837100343942,
      "loss": 1.4307,
      "step": 23773
    },
    {
      "epoch": 2.045427170265852,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006412577710714348,
      "loss": 1.2741,
      "step": 23774
    },
    {
      "epoch": 2.045513206573174,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006412318316953241,
      "loss": 1.2275,
      "step": 23775
    },
    {
      "epoch": 2.0455992428804954,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006412058919061376,
      "loss": 0.8123,
      "step": 23776
    },
    {
      "epoch": 2.045685279187817,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006411799517039513,
      "loss": 0.6677,
      "step": 23777
    },
    {
      "epoch": 2.045771315495139,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006411540110888411,
      "loss": 1.2813,
      "step": 23778
    },
    {
      "epoch": 2.0458573518024608,
      "grad_norm": 0.93359375,
      "learning_rate": 0.000641128070060883,
      "loss": 0.8821,
      "step": 23779
    },
    {
      "epoch": 2.0459433881097824,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006411021286201527,
      "loss": 0.7866,
      "step": 23780
    },
    {
      "epoch": 2.046029424417104,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006410761867667261,
      "loss": 1.4977,
      "step": 23781
    },
    {
      "epoch": 2.0461154607244256,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000641050244500679,
      "loss": 0.6438,
      "step": 23782
    },
    {
      "epoch": 2.0462014970317473,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006410243018220876,
      "loss": 0.9501,
      "step": 23783
    },
    {
      "epoch": 2.046287533339069,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006409983587310275,
      "loss": 1.5539,
      "step": 23784
    },
    {
      "epoch": 2.046373569646391,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006409724152275743,
      "loss": 0.9634,
      "step": 23785
    },
    {
      "epoch": 2.0464596059537126,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006409464713118045,
      "loss": 1.1262,
      "step": 23786
    },
    {
      "epoch": 2.046545642261034,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006409205269837936,
      "loss": 1.589,
      "step": 23787
    },
    {
      "epoch": 2.046631678568356,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006408945822436177,
      "loss": 1.3201,
      "step": 23788
    },
    {
      "epoch": 2.0467177148756774,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006408686370913524,
      "loss": 0.7853,
      "step": 23789
    },
    {
      "epoch": 2.046803751182999,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006408426915270738,
      "loss": 1.2052,
      "step": 23790
    },
    {
      "epoch": 2.046889787490321,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006408167455508577,
      "loss": 1.904,
      "step": 23791
    },
    {
      "epoch": 2.0469758237976428,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006407907991627799,
      "loss": 1.2915,
      "step": 23792
    },
    {
      "epoch": 2.0470618601049644,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006407648523629166,
      "loss": 1.4609,
      "step": 23793
    },
    {
      "epoch": 2.047147896412286,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006407389051513434,
      "loss": 1.3905,
      "step": 23794
    },
    {
      "epoch": 2.0472339327196076,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006407129575281361,
      "loss": 0.8819,
      "step": 23795
    },
    {
      "epoch": 2.0473199690269293,
      "grad_norm": 1.4375,
      "learning_rate": 0.000640687009493371,
      "loss": 0.961,
      "step": 23796
    },
    {
      "epoch": 2.047406005334251,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006406610610471235,
      "loss": 0.7478,
      "step": 23797
    },
    {
      "epoch": 2.047492041641573,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00064063511218947,
      "loss": 1.1774,
      "step": 23798
    },
    {
      "epoch": 2.0475780779488946,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006406091629204861,
      "loss": 1.13,
      "step": 23799
    },
    {
      "epoch": 2.047664114256216,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006405832132402477,
      "loss": 1.6695,
      "step": 23800
    },
    {
      "epoch": 2.047750150563538,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006405572631488308,
      "loss": 1.0632,
      "step": 23801
    },
    {
      "epoch": 2.0478361868708594,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006405313126463112,
      "loss": 0.5661,
      "step": 23802
    },
    {
      "epoch": 2.047922223178181,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006405053617327648,
      "loss": 1.5011,
      "step": 23803
    },
    {
      "epoch": 2.0480082594855027,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006404794104082676,
      "loss": 1.6137,
      "step": 23804
    },
    {
      "epoch": 2.0480942957928248,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006404534586728955,
      "loss": 1.2312,
      "step": 23805
    },
    {
      "epoch": 2.0481803321001464,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006404275065267244,
      "loss": 1.4661,
      "step": 23806
    },
    {
      "epoch": 2.048266368407468,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006404015539698299,
      "loss": 0.8091,
      "step": 23807
    },
    {
      "epoch": 2.0483524047147896,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006403756010022883,
      "loss": 1.3474,
      "step": 23808
    },
    {
      "epoch": 2.0484384410221113,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006403496476241754,
      "loss": 1.5076,
      "step": 23809
    },
    {
      "epoch": 2.048524477329433,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006403236938355671,
      "loss": 1.2591,
      "step": 23810
    },
    {
      "epoch": 2.0486105136367545,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006402977396365391,
      "loss": 1.7084,
      "step": 23811
    },
    {
      "epoch": 2.0486965499440766,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006402717850271677,
      "loss": 0.8934,
      "step": 23812
    },
    {
      "epoch": 2.048782586251398,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006402458300075286,
      "loss": 0.8506,
      "step": 23813
    },
    {
      "epoch": 2.04886862255872,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006402198745776976,
      "loss": 1.6057,
      "step": 23814
    },
    {
      "epoch": 2.0489546588660414,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006401939187377508,
      "loss": 1.3292,
      "step": 23815
    },
    {
      "epoch": 2.049040695173363,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000640167962487764,
      "loss": 0.8576,
      "step": 23816
    },
    {
      "epoch": 2.0491267314806847,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006401420058278134,
      "loss": 0.5525,
      "step": 23817
    },
    {
      "epoch": 2.0492127677880063,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006401160487579745,
      "loss": 0.684,
      "step": 23818
    },
    {
      "epoch": 2.0492988040953284,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006400900912783233,
      "loss": 1.1349,
      "step": 23819
    },
    {
      "epoch": 2.04938484040265,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006400641333889359,
      "loss": 1.1957,
      "step": 23820
    },
    {
      "epoch": 2.0494708767099716,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006400381750898883,
      "loss": 1.0732,
      "step": 23821
    },
    {
      "epoch": 2.0495569130172933,
      "grad_norm": 1.125,
      "learning_rate": 0.0006400122163812561,
      "loss": 0.7888,
      "step": 23822
    },
    {
      "epoch": 2.049642949324615,
      "grad_norm": 1.25,
      "learning_rate": 0.0006399862572631157,
      "loss": 1.1439,
      "step": 23823
    },
    {
      "epoch": 2.0497289856319365,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006399602977355425,
      "loss": 1.0486,
      "step": 23824
    },
    {
      "epoch": 2.0498150219392586,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006399343377986126,
      "loss": 1.656,
      "step": 23825
    },
    {
      "epoch": 2.04990105824658,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006399083774524019,
      "loss": 1.3587,
      "step": 23826
    },
    {
      "epoch": 2.049987094553902,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006398824166969866,
      "loss": 1.0618,
      "step": 23827
    },
    {
      "epoch": 2.0500731308612234,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0006398564555324424,
      "loss": 1.0523,
      "step": 23828
    },
    {
      "epoch": 2.050159167168545,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006398304939588453,
      "loss": 1.1112,
      "step": 23829
    },
    {
      "epoch": 2.0502452034758667,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006398045319762711,
      "loss": 0.8566,
      "step": 23830
    },
    {
      "epoch": 2.0503312397831883,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006397785695847959,
      "loss": 1.6477,
      "step": 23831
    },
    {
      "epoch": 2.0504172760905104,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006397526067844955,
      "loss": 1.1184,
      "step": 23832
    },
    {
      "epoch": 2.050503312397832,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006397266435754459,
      "loss": 0.7739,
      "step": 23833
    },
    {
      "epoch": 2.0505893487051536,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000639700679957723,
      "loss": 1.2586,
      "step": 23834
    },
    {
      "epoch": 2.0506753850124753,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000639674715931403,
      "loss": 0.8793,
      "step": 23835
    },
    {
      "epoch": 2.050761421319797,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006396487514965615,
      "loss": 1.5697,
      "step": 23836
    },
    {
      "epoch": 2.0508474576271185,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006396227866532746,
      "loss": 1.2581,
      "step": 23837
    },
    {
      "epoch": 2.05093349393444,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006395968214016181,
      "loss": 1.0679,
      "step": 23838
    },
    {
      "epoch": 2.051019530241762,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006395708557416682,
      "loss": 1.5015,
      "step": 23839
    },
    {
      "epoch": 2.051105566549084,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006395448896735007,
      "loss": 0.7773,
      "step": 23840
    },
    {
      "epoch": 2.0511916028564054,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006395189231971913,
      "loss": 1.1385,
      "step": 23841
    },
    {
      "epoch": 2.051277639163727,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006394929563128163,
      "loss": 1.2448,
      "step": 23842
    },
    {
      "epoch": 2.0513636754710487,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006394669890204517,
      "loss": 1.1275,
      "step": 23843
    },
    {
      "epoch": 2.0514497117783703,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006394410213201731,
      "loss": 1.319,
      "step": 23844
    },
    {
      "epoch": 2.051535748085692,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006394150532120567,
      "loss": 1.4061,
      "step": 23845
    },
    {
      "epoch": 2.051621784393014,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006393890846961783,
      "loss": 1.0838,
      "step": 23846
    },
    {
      "epoch": 2.0517078207003356,
      "grad_norm": 1.25,
      "learning_rate": 0.000639363115772614,
      "loss": 1.531,
      "step": 23847
    },
    {
      "epoch": 2.0517938570076573,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006393371464414397,
      "loss": 1.0954,
      "step": 23848
    },
    {
      "epoch": 2.051879893314979,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006393111767027313,
      "loss": 1.542,
      "step": 23849
    },
    {
      "epoch": 2.0519659296223005,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006392852065565649,
      "loss": 1.3663,
      "step": 23850
    },
    {
      "epoch": 2.052051965929622,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0006392592360030163,
      "loss": 1.6561,
      "step": 23851
    },
    {
      "epoch": 2.052138002236944,
      "grad_norm": 0.875,
      "learning_rate": 0.0006392332650421615,
      "loss": 0.7288,
      "step": 23852
    },
    {
      "epoch": 2.052224038544266,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006392072936740764,
      "loss": 1.3161,
      "step": 23853
    },
    {
      "epoch": 2.0523100748515875,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006391813218988372,
      "loss": 0.9903,
      "step": 23854
    },
    {
      "epoch": 2.052396111158909,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006391553497165197,
      "loss": 0.9542,
      "step": 23855
    },
    {
      "epoch": 2.0524821474662307,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006391293771271997,
      "loss": 1.0998,
      "step": 23856
    },
    {
      "epoch": 2.0525681837735523,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006391034041309534,
      "loss": 1.1508,
      "step": 23857
    },
    {
      "epoch": 2.052654220080874,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006390774307278569,
      "loss": 1.1485,
      "step": 23858
    },
    {
      "epoch": 2.052740256388196,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006390514569179857,
      "loss": 0.8794,
      "step": 23859
    },
    {
      "epoch": 2.0528262926955176,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006390254827014162,
      "loss": 0.6749,
      "step": 23860
    },
    {
      "epoch": 2.0529123290028393,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000638999508078224,
      "loss": 1.7413,
      "step": 23861
    },
    {
      "epoch": 2.052998365310161,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006389735330484855,
      "loss": 1.3246,
      "step": 23862
    },
    {
      "epoch": 2.0530844016174825,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006389475576122763,
      "loss": 2.1476,
      "step": 23863
    },
    {
      "epoch": 2.053170437924804,
      "grad_norm": 1.25,
      "learning_rate": 0.0006389215817696726,
      "loss": 1.4377,
      "step": 23864
    },
    {
      "epoch": 2.0532564742321258,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006388956055207502,
      "loss": 2.0856,
      "step": 23865
    },
    {
      "epoch": 2.053342510539448,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006388696288655853,
      "loss": 1.0788,
      "step": 23866
    },
    {
      "epoch": 2.0534285468467695,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006388436518042536,
      "loss": 0.9096,
      "step": 23867
    },
    {
      "epoch": 2.053514583154091,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006388176743368313,
      "loss": 1.0317,
      "step": 23868
    },
    {
      "epoch": 2.0536006194614127,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006387916964633942,
      "loss": 0.7857,
      "step": 23869
    },
    {
      "epoch": 2.0536866557687343,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006387657181840186,
      "loss": 1.0861,
      "step": 23870
    },
    {
      "epoch": 2.053772692076056,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006387397394987801,
      "loss": 0.6878,
      "step": 23871
    },
    {
      "epoch": 2.0538587283833776,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006387137604077548,
      "loss": 1.4399,
      "step": 23872
    },
    {
      "epoch": 2.0539447646906996,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006386877809110189,
      "loss": 0.7403,
      "step": 23873
    },
    {
      "epoch": 2.0540308009980213,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006386618010086481,
      "loss": 1.3273,
      "step": 23874
    },
    {
      "epoch": 2.054116837305343,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006386358207007185,
      "loss": 1.298,
      "step": 23875
    },
    {
      "epoch": 2.0542028736126645,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006386098399873061,
      "loss": 1.7096,
      "step": 23876
    },
    {
      "epoch": 2.054288909919986,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0006385838588684869,
      "loss": 2.9973,
      "step": 23877
    },
    {
      "epoch": 2.0543749462273078,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006385578773443368,
      "loss": 1.0448,
      "step": 23878
    },
    {
      "epoch": 2.05446098253463,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000638531895414932,
      "loss": 1.9023,
      "step": 23879
    },
    {
      "epoch": 2.0545470188419515,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006385059130803483,
      "loss": 1.4982,
      "step": 23880
    },
    {
      "epoch": 2.054633055149273,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006384799303406617,
      "loss": 1.2565,
      "step": 23881
    },
    {
      "epoch": 2.0547190914565947,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006384539471959482,
      "loss": 0.8846,
      "step": 23882
    },
    {
      "epoch": 2.0548051277639163,
      "grad_norm": 1.5,
      "learning_rate": 0.0006384279636462839,
      "loss": 1.9369,
      "step": 23883
    },
    {
      "epoch": 2.054891164071238,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006384019796917448,
      "loss": 0.846,
      "step": 23884
    },
    {
      "epoch": 2.0549772003785596,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006383759953324066,
      "loss": 0.977,
      "step": 23885
    },
    {
      "epoch": 2.0550632366858816,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006383500105683457,
      "loss": 1.2841,
      "step": 23886
    },
    {
      "epoch": 2.0551492729932033,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006383240253996378,
      "loss": 1.5811,
      "step": 23887
    },
    {
      "epoch": 2.055235309300525,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006382980398263591,
      "loss": 1.2322,
      "step": 23888
    },
    {
      "epoch": 2.0553213456078465,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006382720538485856,
      "loss": 1.7489,
      "step": 23889
    },
    {
      "epoch": 2.055407381915168,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006382460674663931,
      "loss": 0.8712,
      "step": 23890
    },
    {
      "epoch": 2.0554934182224898,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006382200806798579,
      "loss": 1.2338,
      "step": 23891
    },
    {
      "epoch": 2.0555794545298114,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006381940934890558,
      "loss": 1.7973,
      "step": 23892
    },
    {
      "epoch": 2.0556654908371335,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006381681058940628,
      "loss": 1.5011,
      "step": 23893
    },
    {
      "epoch": 2.055751527144455,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006381421178949549,
      "loss": 2.0545,
      "step": 23894
    },
    {
      "epoch": 2.0558375634517767,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006381161294918083,
      "loss": 0.9804,
      "step": 23895
    },
    {
      "epoch": 2.0559235997590983,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006380901406846988,
      "loss": 1.2892,
      "step": 23896
    },
    {
      "epoch": 2.05600963606642,
      "grad_norm": 1.125,
      "learning_rate": 0.0006380641514737024,
      "loss": 0.8792,
      "step": 23897
    },
    {
      "epoch": 2.0560956723737416,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006380381618588955,
      "loss": 1.5954,
      "step": 23898
    },
    {
      "epoch": 2.056181708681063,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006380121718403537,
      "loss": 0.9311,
      "step": 23899
    },
    {
      "epoch": 2.0562677449883853,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006379861814181531,
      "loss": 1.4129,
      "step": 23900
    },
    {
      "epoch": 2.056353781295707,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006379601905923696,
      "loss": 1.4561,
      "step": 23901
    },
    {
      "epoch": 2.0564398176030285,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006379341993630796,
      "loss": 1.4251,
      "step": 23902
    },
    {
      "epoch": 2.05652585391035,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006379082077303588,
      "loss": 1.0906,
      "step": 23903
    },
    {
      "epoch": 2.0566118902176718,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006378822156942834,
      "loss": 1.4611,
      "step": 23904
    },
    {
      "epoch": 2.0566979265249934,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006378562232549293,
      "loss": 1.2898,
      "step": 23905
    },
    {
      "epoch": 2.0567839628323155,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006378302304123724,
      "loss": 1.4549,
      "step": 23906
    },
    {
      "epoch": 2.056869999139637,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006378042371666891,
      "loss": 0.6089,
      "step": 23907
    },
    {
      "epoch": 2.0569560354469587,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006377782435179551,
      "loss": 1.3621,
      "step": 23908
    },
    {
      "epoch": 2.0570420717542803,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006377522494662466,
      "loss": 1.0881,
      "step": 23909
    },
    {
      "epoch": 2.057128108061602,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006377262550116395,
      "loss": 0.6919,
      "step": 23910
    },
    {
      "epoch": 2.0572141443689236,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006377002601542099,
      "loss": 1.6151,
      "step": 23911
    },
    {
      "epoch": 2.057300180676245,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006376742648940338,
      "loss": 1.1655,
      "step": 23912
    },
    {
      "epoch": 2.0573862169835673,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006376482692311874,
      "loss": 0.9451,
      "step": 23913
    },
    {
      "epoch": 2.057472253290889,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006376222731657464,
      "loss": 1.0758,
      "step": 23914
    },
    {
      "epoch": 2.0575582895982105,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006375962766977871,
      "loss": 0.8333,
      "step": 23915
    },
    {
      "epoch": 2.057644325905532,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006375702798273854,
      "loss": 1.076,
      "step": 23916
    },
    {
      "epoch": 2.0577303622128538,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006375442825546174,
      "loss": 1.0827,
      "step": 23917
    },
    {
      "epoch": 2.0578163985201754,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006375182848795592,
      "loss": 1.3773,
      "step": 23918
    },
    {
      "epoch": 2.057902434827497,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006374922868022867,
      "loss": 1.6863,
      "step": 23919
    },
    {
      "epoch": 2.057988471134819,
      "grad_norm": 1.125,
      "learning_rate": 0.000637466288322876,
      "loss": 0.8853,
      "step": 23920
    },
    {
      "epoch": 2.0580745074421407,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006374402894414032,
      "loss": 0.9655,
      "step": 23921
    },
    {
      "epoch": 2.0581605437494623,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006374142901579443,
      "loss": 0.6227,
      "step": 23922
    },
    {
      "epoch": 2.058246580056784,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006373882904725752,
      "loss": 0.7272,
      "step": 23923
    },
    {
      "epoch": 2.0583326163641056,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000637362290385372,
      "loss": 1.4349,
      "step": 23924
    },
    {
      "epoch": 2.058418652671427,
      "grad_norm": 1.234375,
      "learning_rate": 0.000637336289896411,
      "loss": 0.984,
      "step": 23925
    },
    {
      "epoch": 2.058504688978749,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006373102890057681,
      "loss": 1.9304,
      "step": 23926
    },
    {
      "epoch": 2.058590725286071,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006372842877135191,
      "loss": 1.0922,
      "step": 23927
    },
    {
      "epoch": 2.0586767615933925,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006372582860197404,
      "loss": 0.6641,
      "step": 23928
    },
    {
      "epoch": 2.058762797900714,
      "grad_norm": 1.546875,
      "learning_rate": 0.000637232283924508,
      "loss": 1.8703,
      "step": 23929
    },
    {
      "epoch": 2.0588488342080358,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006372062814278977,
      "loss": 1.4673,
      "step": 23930
    },
    {
      "epoch": 2.0589348705153574,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006371802785299856,
      "loss": 1.6786,
      "step": 23931
    },
    {
      "epoch": 2.059020906822679,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006371542752308481,
      "loss": 1.7308,
      "step": 23932
    },
    {
      "epoch": 2.0591069431300006,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006371282715305611,
      "loss": 1.0707,
      "step": 23933
    },
    {
      "epoch": 2.0591929794373227,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006371022674292003,
      "loss": 1.6521,
      "step": 23934
    },
    {
      "epoch": 2.0592790157446443,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006370762629268422,
      "loss": 0.9193,
      "step": 23935
    },
    {
      "epoch": 2.059365052051966,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006370502580235626,
      "loss": 0.7291,
      "step": 23936
    },
    {
      "epoch": 2.0594510883592876,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006370242527194377,
      "loss": 0.54,
      "step": 23937
    },
    {
      "epoch": 2.059537124666609,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006369982470145434,
      "loss": 0.9978,
      "step": 23938
    },
    {
      "epoch": 2.059623160973931,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006369722409089561,
      "loss": 1.2148,
      "step": 23939
    },
    {
      "epoch": 2.059709197281253,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006369462344027515,
      "loss": 1.882,
      "step": 23940
    },
    {
      "epoch": 2.0597952335885745,
      "grad_norm": 1.125,
      "learning_rate": 0.0006369202274960059,
      "loss": 1.4141,
      "step": 23941
    },
    {
      "epoch": 2.059881269895896,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006368942201887951,
      "loss": 1.5978,
      "step": 23942
    },
    {
      "epoch": 2.0599673062032178,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006368682124811954,
      "loss": 1.2058,
      "step": 23943
    },
    {
      "epoch": 2.0600533425105394,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006368422043732828,
      "loss": 0.5541,
      "step": 23944
    },
    {
      "epoch": 2.060139378817861,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006368161958651334,
      "loss": 1.3651,
      "step": 23945
    },
    {
      "epoch": 2.0602254151251826,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006367901869568231,
      "loss": 0.7266,
      "step": 23946
    },
    {
      "epoch": 2.0603114514325047,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006367641776484284,
      "loss": 2.0559,
      "step": 23947
    },
    {
      "epoch": 2.0603974877398263,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006367381679400248,
      "loss": 1.8844,
      "step": 23948
    },
    {
      "epoch": 2.060483524047148,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006367121578316888,
      "loss": 1.4946,
      "step": 23949
    },
    {
      "epoch": 2.0605695603544696,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006366861473234962,
      "loss": 1.2888,
      "step": 23950
    },
    {
      "epoch": 2.060655596661791,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006366601364155233,
      "loss": 1.3137,
      "step": 23951
    },
    {
      "epoch": 2.060741632969113,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000636634125107846,
      "loss": 0.8501,
      "step": 23952
    },
    {
      "epoch": 2.0608276692764345,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006366081134005406,
      "loss": 0.7898,
      "step": 23953
    },
    {
      "epoch": 2.0609137055837565,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006365821012936829,
      "loss": 0.8854,
      "step": 23954
    },
    {
      "epoch": 2.060999741891078,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006365560887873491,
      "loss": 1.8327,
      "step": 23955
    },
    {
      "epoch": 2.0610857781983998,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006365300758816155,
      "loss": 1.3417,
      "step": 23956
    },
    {
      "epoch": 2.0611718145057214,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006365040625765578,
      "loss": 1.9048,
      "step": 23957
    },
    {
      "epoch": 2.061257850813043,
      "grad_norm": 1.609375,
      "learning_rate": 0.0006364780488722524,
      "loss": 1.1909,
      "step": 23958
    },
    {
      "epoch": 2.0613438871203646,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006364520347687751,
      "loss": 1.5864,
      "step": 23959
    },
    {
      "epoch": 2.0614299234276863,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006364260202662022,
      "loss": 1.1217,
      "step": 23960
    },
    {
      "epoch": 2.0615159597350083,
      "grad_norm": 1.25,
      "learning_rate": 0.0006364000053646098,
      "loss": 1.1675,
      "step": 23961
    },
    {
      "epoch": 2.06160199604233,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006363739900640737,
      "loss": 1.0167,
      "step": 23962
    },
    {
      "epoch": 2.0616880323496516,
      "grad_norm": 4.3125,
      "learning_rate": 0.0006363479743646704,
      "loss": 1.791,
      "step": 23963
    },
    {
      "epoch": 2.061774068656973,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006363219582664757,
      "loss": 1.5314,
      "step": 23964
    },
    {
      "epoch": 2.061860104964295,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006362959417695658,
      "loss": 1.3942,
      "step": 23965
    },
    {
      "epoch": 2.0619461412716165,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006362699248740169,
      "loss": 1.3563,
      "step": 23966
    },
    {
      "epoch": 2.0620321775789385,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006362439075799049,
      "loss": 1.3515,
      "step": 23967
    },
    {
      "epoch": 2.06211821388626,
      "grad_norm": 1.375,
      "learning_rate": 0.0006362178898873058,
      "loss": 1.1632,
      "step": 23968
    },
    {
      "epoch": 2.062204250193582,
      "grad_norm": 0.9375,
      "learning_rate": 0.000636191871796296,
      "loss": 1.7445,
      "step": 23969
    },
    {
      "epoch": 2.0622902865009034,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006361658533069515,
      "loss": 0.7168,
      "step": 23970
    },
    {
      "epoch": 2.062376322808225,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006361398344193484,
      "loss": 2.1384,
      "step": 23971
    },
    {
      "epoch": 2.0624623591155467,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006361138151335625,
      "loss": 1.3951,
      "step": 23972
    },
    {
      "epoch": 2.0625483954228683,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006360877954496706,
      "loss": 1.7074,
      "step": 23973
    },
    {
      "epoch": 2.0626344317301903,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006360617753677481,
      "loss": 0.9216,
      "step": 23974
    },
    {
      "epoch": 2.062720468037512,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006360357548878713,
      "loss": 0.6388,
      "step": 23975
    },
    {
      "epoch": 2.0628065043448336,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006360097340101166,
      "loss": 0.8161,
      "step": 23976
    },
    {
      "epoch": 2.062892540652155,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006359837127345596,
      "loss": 0.9526,
      "step": 23977
    },
    {
      "epoch": 2.062978576959477,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000635957691061277,
      "loss": 0.9739,
      "step": 23978
    },
    {
      "epoch": 2.0630646132667985,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006359316689903443,
      "loss": 0.7799,
      "step": 23979
    },
    {
      "epoch": 2.06315064957412,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000635905646521838,
      "loss": 1.271,
      "step": 23980
    },
    {
      "epoch": 2.063236685881442,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006358796236558342,
      "loss": 0.8785,
      "step": 23981
    },
    {
      "epoch": 2.063322722188764,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006358536003924089,
      "loss": 1.4884,
      "step": 23982
    },
    {
      "epoch": 2.0634087584960854,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006358275767316382,
      "loss": 1.1716,
      "step": 23983
    },
    {
      "epoch": 2.063494794803407,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006358015526735983,
      "loss": 0.7676,
      "step": 23984
    },
    {
      "epoch": 2.0635808311107287,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006357755282183653,
      "loss": 1.4718,
      "step": 23985
    },
    {
      "epoch": 2.0636668674180503,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006357495033660152,
      "loss": 1.7395,
      "step": 23986
    },
    {
      "epoch": 2.063752903725372,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006357234781166241,
      "loss": 1.1645,
      "step": 23987
    },
    {
      "epoch": 2.063838940032694,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006356974524702685,
      "loss": 1.7882,
      "step": 23988
    },
    {
      "epoch": 2.0639249763400156,
      "grad_norm": 0.890625,
      "learning_rate": 0.000635671426427024,
      "loss": 0.9303,
      "step": 23989
    },
    {
      "epoch": 2.064011012647337,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006356453999869672,
      "loss": 1.6294,
      "step": 23990
    },
    {
      "epoch": 2.064097048954659,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006356193731501738,
      "loss": 1.2311,
      "step": 23991
    },
    {
      "epoch": 2.0641830852619805,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006355933459167201,
      "loss": 1.0536,
      "step": 23992
    },
    {
      "epoch": 2.064269121569302,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006355673182866824,
      "loss": 1.4336,
      "step": 23993
    },
    {
      "epoch": 2.0643551578766237,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006355412902601366,
      "loss": 1.2927,
      "step": 23994
    },
    {
      "epoch": 2.064441194183946,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006355152618371586,
      "loss": 1.3323,
      "step": 23995
    },
    {
      "epoch": 2.0645272304912674,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006354892330178252,
      "loss": 0.9806,
      "step": 23996
    },
    {
      "epoch": 2.064613266798589,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006354632038022122,
      "loss": 0.8947,
      "step": 23997
    },
    {
      "epoch": 2.0646993031059107,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006354371741903953,
      "loss": 0.8675,
      "step": 23998
    },
    {
      "epoch": 2.0647853394132323,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000635411144182451,
      "loss": 1.3749,
      "step": 23999
    },
    {
      "epoch": 2.064871375720554,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006353851137784558,
      "loss": 1.6541,
      "step": 24000
    },
    {
      "epoch": 2.064957412027876,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006353590829784854,
      "loss": 1.1589,
      "step": 24001
    },
    {
      "epoch": 2.0650434483351976,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006353330517826157,
      "loss": 1.0285,
      "step": 24002
    },
    {
      "epoch": 2.065129484642519,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006353070201909235,
      "loss": 1.0948,
      "step": 24003
    },
    {
      "epoch": 2.065215520949841,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006352809882034844,
      "loss": 0.711,
      "step": 24004
    },
    {
      "epoch": 2.0653015572571625,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006352549558203748,
      "loss": 1.4105,
      "step": 24005
    },
    {
      "epoch": 2.065387593564484,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006352289230416706,
      "loss": 1.1911,
      "step": 24006
    },
    {
      "epoch": 2.0654736298718057,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006352028898674483,
      "loss": 1.6617,
      "step": 24007
    },
    {
      "epoch": 2.065559666179128,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006351768562977838,
      "loss": 2.2869,
      "step": 24008
    },
    {
      "epoch": 2.0656457024864494,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006351508223327531,
      "loss": 1.4509,
      "step": 24009
    },
    {
      "epoch": 2.065731738793771,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006351247879724328,
      "loss": 1.0587,
      "step": 24010
    },
    {
      "epoch": 2.0658177751010927,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006350987532168985,
      "loss": 1.8393,
      "step": 24011
    },
    {
      "epoch": 2.0659038114084143,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006350727180662269,
      "loss": 1.1411,
      "step": 24012
    },
    {
      "epoch": 2.065989847715736,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006350466825204937,
      "loss": 1.8445,
      "step": 24013
    },
    {
      "epoch": 2.0660758840230575,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006350206465797753,
      "loss": 1.5413,
      "step": 24014
    },
    {
      "epoch": 2.0661619203303796,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006349946102441477,
      "loss": 0.8581,
      "step": 24015
    },
    {
      "epoch": 2.066247956637701,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006349685735136872,
      "loss": 1.0688,
      "step": 24016
    },
    {
      "epoch": 2.066333992945023,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006349425363884697,
      "loss": 1.7102,
      "step": 24017
    },
    {
      "epoch": 2.0664200292523445,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006349164988685716,
      "loss": 0.8398,
      "step": 24018
    },
    {
      "epoch": 2.066506065559666,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006348904609540691,
      "loss": 1.2031,
      "step": 24019
    },
    {
      "epoch": 2.0665921018669877,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006348644226450382,
      "loss": 1.3745,
      "step": 24020
    },
    {
      "epoch": 2.06667813817431,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006348383839415551,
      "loss": 1.3135,
      "step": 24021
    },
    {
      "epoch": 2.0667641744816314,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006348123448436959,
      "loss": 1.0734,
      "step": 24022
    },
    {
      "epoch": 2.066850210788953,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006347863053515368,
      "loss": 0.8672,
      "step": 24023
    },
    {
      "epoch": 2.0669362470962747,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006347602654651539,
      "loss": 0.7995,
      "step": 24024
    },
    {
      "epoch": 2.0670222834035963,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006347342251846235,
      "loss": 1.6559,
      "step": 24025
    },
    {
      "epoch": 2.067108319710918,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006347081845100218,
      "loss": 1.6059,
      "step": 24026
    },
    {
      "epoch": 2.0671943560182395,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006346821434414249,
      "loss": 1.0527,
      "step": 24027
    },
    {
      "epoch": 2.0672803923255616,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006346561019789087,
      "loss": 1.0156,
      "step": 24028
    },
    {
      "epoch": 2.0673664286328832,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006346300601225498,
      "loss": 1.2385,
      "step": 24029
    },
    {
      "epoch": 2.067452464940205,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000634604017872424,
      "loss": 0.6711,
      "step": 24030
    },
    {
      "epoch": 2.0675385012475265,
      "grad_norm": 0.875,
      "learning_rate": 0.0006345779752286076,
      "loss": 0.7566,
      "step": 24031
    },
    {
      "epoch": 2.067624537554848,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006345519321911769,
      "loss": 0.9487,
      "step": 24032
    },
    {
      "epoch": 2.0677105738621697,
      "grad_norm": 0.90625,
      "learning_rate": 0.000634525888760208,
      "loss": 0.8516,
      "step": 24033
    },
    {
      "epoch": 2.0677966101694913,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000634499844935777,
      "loss": 1.3363,
      "step": 24034
    },
    {
      "epoch": 2.0678826464768134,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00063447380071796,
      "loss": 1.9646,
      "step": 24035
    },
    {
      "epoch": 2.067968682784135,
      "grad_norm": 1.5,
      "learning_rate": 0.0006344477561068334,
      "loss": 1.1995,
      "step": 24036
    },
    {
      "epoch": 2.0680547190914567,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006344217111024733,
      "loss": 1.2766,
      "step": 24037
    },
    {
      "epoch": 2.0681407553987783,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006343956657049557,
      "loss": 1.4729,
      "step": 24038
    },
    {
      "epoch": 2.0682267917061,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006343696199143569,
      "loss": 0.7998,
      "step": 24039
    },
    {
      "epoch": 2.0683128280134215,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006343435737307533,
      "loss": 1.3177,
      "step": 24040
    },
    {
      "epoch": 2.068398864320743,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006343175271542209,
      "loss": 1.4554,
      "step": 24041
    },
    {
      "epoch": 2.0684849006280652,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006342914801848356,
      "loss": 1.1485,
      "step": 24042
    },
    {
      "epoch": 2.068570936935387,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000634265432822674,
      "loss": 1.1243,
      "step": 24043
    },
    {
      "epoch": 2.0686569732427085,
      "grad_norm": 0.890625,
      "learning_rate": 0.000634239385067812,
      "loss": 0.8599,
      "step": 24044
    },
    {
      "epoch": 2.06874300955003,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006342133369203262,
      "loss": 1.7182,
      "step": 24045
    },
    {
      "epoch": 2.0688290458573517,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006341872883802922,
      "loss": 1.1961,
      "step": 24046
    },
    {
      "epoch": 2.0689150821646733,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006341612394477865,
      "loss": 1.0612,
      "step": 24047
    },
    {
      "epoch": 2.069001118471995,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006341351901228853,
      "loss": 0.8614,
      "step": 24048
    },
    {
      "epoch": 2.069087154779317,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000634109140405665,
      "loss": 0.7415,
      "step": 24049
    },
    {
      "epoch": 2.0691731910866387,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006340830902962012,
      "loss": 1.9905,
      "step": 24050
    },
    {
      "epoch": 2.0692592273939603,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006340570397945707,
      "loss": 1.254,
      "step": 24051
    },
    {
      "epoch": 2.069345263701282,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006340309889008493,
      "loss": 1.2046,
      "step": 24052
    },
    {
      "epoch": 2.0694313000086035,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006340049376151133,
      "loss": 1.2364,
      "step": 24053
    },
    {
      "epoch": 2.069517336315925,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006339788859374387,
      "loss": 1.1996,
      "step": 24054
    },
    {
      "epoch": 2.069603372623247,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0006339528338679023,
      "loss": 1.7416,
      "step": 24055
    },
    {
      "epoch": 2.069689408930569,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006339267814065798,
      "loss": 1.6891,
      "step": 24056
    },
    {
      "epoch": 2.0697754452378905,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006339007285535475,
      "loss": 1.5663,
      "step": 24057
    },
    {
      "epoch": 2.069861481545212,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006338746753088815,
      "loss": 1.3726,
      "step": 24058
    },
    {
      "epoch": 2.0699475178525337,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006338486216726583,
      "loss": 1.0704,
      "step": 24059
    },
    {
      "epoch": 2.0700335541598553,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006338225676449539,
      "loss": 1.4226,
      "step": 24060
    },
    {
      "epoch": 2.070119590467177,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006337965132258444,
      "loss": 1.736,
      "step": 24061
    },
    {
      "epoch": 2.070205626774499,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006337704584154062,
      "loss": 1.3807,
      "step": 24062
    },
    {
      "epoch": 2.0702916630818207,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006337444032137155,
      "loss": 1.3117,
      "step": 24063
    },
    {
      "epoch": 2.0703776993891423,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006337183476208483,
      "loss": 1.1069,
      "step": 24064
    },
    {
      "epoch": 2.070463735696464,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006336922916368808,
      "loss": 0.9746,
      "step": 24065
    },
    {
      "epoch": 2.0705497720037855,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006336662352618896,
      "loss": 0.8314,
      "step": 24066
    },
    {
      "epoch": 2.070635808311107,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006336401784959505,
      "loss": 0.8154,
      "step": 24067
    },
    {
      "epoch": 2.070721844618429,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006336141213391401,
      "loss": 0.7571,
      "step": 24068
    },
    {
      "epoch": 2.070807880925751,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006335880637915342,
      "loss": 1.0211,
      "step": 24069
    },
    {
      "epoch": 2.0708939172330725,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006335620058532093,
      "loss": 1.4673,
      "step": 24070
    },
    {
      "epoch": 2.070979953540394,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006335359475242414,
      "loss": 1.3741,
      "step": 24071
    },
    {
      "epoch": 2.0710659898477157,
      "grad_norm": 1.0,
      "learning_rate": 0.0006335098888047067,
      "loss": 1.7592,
      "step": 24072
    },
    {
      "epoch": 2.0711520261550374,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006334838296946817,
      "loss": 1.5876,
      "step": 24073
    },
    {
      "epoch": 2.071238062462359,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006334577701942424,
      "loss": 1.0877,
      "step": 24074
    },
    {
      "epoch": 2.0713240987696806,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006334317103034653,
      "loss": 1.144,
      "step": 24075
    },
    {
      "epoch": 2.0714101350770027,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006334056500224262,
      "loss": 1.9235,
      "step": 24076
    },
    {
      "epoch": 2.0714961713843243,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006333795893512014,
      "loss": 0.9436,
      "step": 24077
    },
    {
      "epoch": 2.071582207691646,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006333535282898675,
      "loss": 1.2061,
      "step": 24078
    },
    {
      "epoch": 2.0716682439989675,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006333274668385003,
      "loss": 1.919,
      "step": 24079
    },
    {
      "epoch": 2.071754280306289,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006333014049971763,
      "loss": 1.9505,
      "step": 24080
    },
    {
      "epoch": 2.071840316613611,
      "grad_norm": 1.25,
      "learning_rate": 0.0006332753427659715,
      "loss": 0.6396,
      "step": 24081
    },
    {
      "epoch": 2.071926352920933,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006332492801449622,
      "loss": 1.0048,
      "step": 24082
    },
    {
      "epoch": 2.0720123892282545,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006332232171342249,
      "loss": 1.0479,
      "step": 24083
    },
    {
      "epoch": 2.072098425535576,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006331971537338355,
      "loss": 1.0657,
      "step": 24084
    },
    {
      "epoch": 2.0721844618428977,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006331710899438702,
      "loss": 1.4385,
      "step": 24085
    },
    {
      "epoch": 2.0722704981502194,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006331450257644056,
      "loss": 1.0027,
      "step": 24086
    },
    {
      "epoch": 2.072356534457541,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006331189611955175,
      "loss": 0.8127,
      "step": 24087
    },
    {
      "epoch": 2.0724425707648626,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006330928962372826,
      "loss": 1.0652,
      "step": 24088
    },
    {
      "epoch": 2.0725286070721847,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006330668308897765,
      "loss": 1.0663,
      "step": 24089
    },
    {
      "epoch": 2.0726146433795063,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006330407651530761,
      "loss": 1.8421,
      "step": 24090
    },
    {
      "epoch": 2.072700679686828,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006330146990272571,
      "loss": 1.1352,
      "step": 24091
    },
    {
      "epoch": 2.0727867159941495,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006329886325123959,
      "loss": 1.0973,
      "step": 24092
    },
    {
      "epoch": 2.072872752301471,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006329625656085691,
      "loss": 1.1901,
      "step": 24093
    },
    {
      "epoch": 2.072958788608793,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006329364983158525,
      "loss": 1.0251,
      "step": 24094
    },
    {
      "epoch": 2.0730448249161144,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006329104306343226,
      "loss": 1.9019,
      "step": 24095
    },
    {
      "epoch": 2.0731308612234365,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006328843625640555,
      "loss": 1.5029,
      "step": 24096
    },
    {
      "epoch": 2.073216897530758,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006328582941051274,
      "loss": 1.0485,
      "step": 24097
    },
    {
      "epoch": 2.0733029338380797,
      "grad_norm": 1.515625,
      "learning_rate": 0.0006328322252576147,
      "loss": 1.7124,
      "step": 24098
    },
    {
      "epoch": 2.0733889701454014,
      "grad_norm": 2.171875,
      "learning_rate": 0.0006328061560215935,
      "loss": 1.2588,
      "step": 24099
    },
    {
      "epoch": 2.073475006452723,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006327800863971402,
      "loss": 1.292,
      "step": 24100
    },
    {
      "epoch": 2.0735610427600446,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006327540163843311,
      "loss": 0.8933,
      "step": 24101
    },
    {
      "epoch": 2.0736470790673662,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006327279459832423,
      "loss": 0.9263,
      "step": 24102
    },
    {
      "epoch": 2.0737331153746883,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006327018751939497,
      "loss": 0.7904,
      "step": 24103
    },
    {
      "epoch": 2.07381915168201,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006326758040165304,
      "loss": 1.6067,
      "step": 24104
    },
    {
      "epoch": 2.0739051879893315,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006326497324510599,
      "loss": 1.2326,
      "step": 24105
    },
    {
      "epoch": 2.073991224296653,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006326236604976146,
      "loss": 1.3075,
      "step": 24106
    },
    {
      "epoch": 2.074077260603975,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006325975881562712,
      "loss": 1.6694,
      "step": 24107
    },
    {
      "epoch": 2.0741632969112964,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006325715154271056,
      "loss": 0.546,
      "step": 24108
    },
    {
      "epoch": 2.074249333218618,
      "grad_norm": 0.78125,
      "learning_rate": 0.000632545442310194,
      "loss": 0.5851,
      "step": 24109
    },
    {
      "epoch": 2.07433536952594,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006325193688056126,
      "loss": 1.2073,
      "step": 24110
    },
    {
      "epoch": 2.0744214058332617,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006324932949134381,
      "loss": 1.3957,
      "step": 24111
    },
    {
      "epoch": 2.0745074421405834,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006324672206337464,
      "loss": 1.5652,
      "step": 24112
    },
    {
      "epoch": 2.074593478447905,
      "grad_norm": 1.734375,
      "learning_rate": 0.0006324411459666139,
      "loss": 1.322,
      "step": 24113
    },
    {
      "epoch": 2.0746795147552266,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006324150709121166,
      "loss": 1.3112,
      "step": 24114
    },
    {
      "epoch": 2.0747655510625482,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006323889954703312,
      "loss": 1.7864,
      "step": 24115
    },
    {
      "epoch": 2.0748515873698703,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006323629196413336,
      "loss": 1.2617,
      "step": 24116
    },
    {
      "epoch": 2.074937623677192,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006323368434252003,
      "loss": 0.842,
      "step": 24117
    },
    {
      "epoch": 2.0750236599845135,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006323107668220072,
      "loss": 0.8081,
      "step": 24118
    },
    {
      "epoch": 2.075109696291835,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006322846898318312,
      "loss": 1.3581,
      "step": 24119
    },
    {
      "epoch": 2.075195732599157,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006322586124547481,
      "loss": 0.8967,
      "step": 24120
    },
    {
      "epoch": 2.0752817689064784,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006322325346908342,
      "loss": 0.8134,
      "step": 24121
    },
    {
      "epoch": 2.0753678052138,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006322064565401659,
      "loss": 1.2444,
      "step": 24122
    },
    {
      "epoch": 2.075453841521122,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006321803780028195,
      "loss": 1.0988,
      "step": 24123
    },
    {
      "epoch": 2.0755398778284437,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006321542990788712,
      "loss": 1.7457,
      "step": 24124
    },
    {
      "epoch": 2.0756259141357654,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006321282197683971,
      "loss": 1.1295,
      "step": 24125
    },
    {
      "epoch": 2.075711950443087,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006321021400714737,
      "loss": 1.1086,
      "step": 24126
    },
    {
      "epoch": 2.0757979867504086,
      "grad_norm": 0.875,
      "learning_rate": 0.0006320760599881774,
      "loss": 0.6255,
      "step": 24127
    },
    {
      "epoch": 2.0758840230577302,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006320499795185842,
      "loss": 0.8189,
      "step": 24128
    },
    {
      "epoch": 2.075970059365052,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006320238986627703,
      "loss": 0.8255,
      "step": 24129
    },
    {
      "epoch": 2.076056095672374,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006319978174208124,
      "loss": 1.2448,
      "step": 24130
    },
    {
      "epoch": 2.0761421319796955,
      "grad_norm": 2.09375,
      "learning_rate": 0.0006319717357927866,
      "loss": 0.7665,
      "step": 24131
    },
    {
      "epoch": 2.076228168287017,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006319456537787692,
      "loss": 0.7303,
      "step": 24132
    },
    {
      "epoch": 2.076314204594339,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006319195713788361,
      "loss": 1.1596,
      "step": 24133
    },
    {
      "epoch": 2.0764002409016604,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000631893488593064,
      "loss": 1.5759,
      "step": 24134
    },
    {
      "epoch": 2.076486277208982,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006318674054215293,
      "loss": 1.1171,
      "step": 24135
    },
    {
      "epoch": 2.0765723135163037,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000631841321864308,
      "loss": 1.0533,
      "step": 24136
    },
    {
      "epoch": 2.0766583498236257,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006318152379214764,
      "loss": 1.2485,
      "step": 24137
    },
    {
      "epoch": 2.0767443861309474,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006317891535931111,
      "loss": 1.5084,
      "step": 24138
    },
    {
      "epoch": 2.076830422438269,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006317630688792879,
      "loss": 1.6651,
      "step": 24139
    },
    {
      "epoch": 2.0769164587455906,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006317369837800834,
      "loss": 1.6396,
      "step": 24140
    },
    {
      "epoch": 2.0770024950529122,
      "grad_norm": 0.921875,
      "learning_rate": 0.000631710898295574,
      "loss": 0.5094,
      "step": 24141
    },
    {
      "epoch": 2.077088531360234,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006316848124258357,
      "loss": 0.9904,
      "step": 24142
    },
    {
      "epoch": 2.077174567667556,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006316587261709448,
      "loss": 1.175,
      "step": 24143
    },
    {
      "epoch": 2.0772606039748776,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006316326395309779,
      "loss": 1.3241,
      "step": 24144
    },
    {
      "epoch": 2.077346640282199,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006316065525060112,
      "loss": 1.3691,
      "step": 24145
    },
    {
      "epoch": 2.077432676589521,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006315804650961209,
      "loss": 0.778,
      "step": 24146
    },
    {
      "epoch": 2.0775187128968424,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006315543773013832,
      "loss": 0.9439,
      "step": 24147
    },
    {
      "epoch": 2.077604749204164,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006315282891218746,
      "loss": 0.9441,
      "step": 24148
    },
    {
      "epoch": 2.0776907855114857,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006315022005576713,
      "loss": 1.3101,
      "step": 24149
    },
    {
      "epoch": 2.0777768218188077,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006314761116088496,
      "loss": 1.3394,
      "step": 24150
    },
    {
      "epoch": 2.0778628581261294,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000631450022275486,
      "loss": 1.1246,
      "step": 24151
    },
    {
      "epoch": 2.077948894433451,
      "grad_norm": 1.484375,
      "learning_rate": 0.0006314239325576565,
      "loss": 1.3525,
      "step": 24152
    },
    {
      "epoch": 2.0780349307407726,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006313978424554378,
      "loss": 1.2656,
      "step": 24153
    },
    {
      "epoch": 2.0781209670480942,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006313717519689057,
      "loss": 0.8874,
      "step": 24154
    },
    {
      "epoch": 2.078207003355416,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006313456610981368,
      "loss": 0.8723,
      "step": 24155
    },
    {
      "epoch": 2.0782930396627375,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006313195698432075,
      "loss": 1.2368,
      "step": 24156
    },
    {
      "epoch": 2.0783790759700596,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000631293478204194,
      "loss": 1.065,
      "step": 24157
    },
    {
      "epoch": 2.078465112277381,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006312673861811727,
      "loss": 1.2894,
      "step": 24158
    },
    {
      "epoch": 2.078551148584703,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006312412937742195,
      "loss": 1.151,
      "step": 24159
    },
    {
      "epoch": 2.0786371848920244,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006312152009834113,
      "loss": 1.3449,
      "step": 24160
    },
    {
      "epoch": 2.078723221199346,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006311891078088241,
      "loss": 0.968,
      "step": 24161
    },
    {
      "epoch": 2.0788092575066677,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006311630142505343,
      "loss": 1.2252,
      "step": 24162
    },
    {
      "epoch": 2.0788952938139893,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006311369203086182,
      "loss": 1.4895,
      "step": 24163
    },
    {
      "epoch": 2.0789813301213114,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000631110825983152,
      "loss": 1.3067,
      "step": 24164
    },
    {
      "epoch": 2.079067366428633,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006310847312742122,
      "loss": 0.6891,
      "step": 24165
    },
    {
      "epoch": 2.0791534027359546,
      "grad_norm": 1.25,
      "learning_rate": 0.000631058636181875,
      "loss": 1.7082,
      "step": 24166
    },
    {
      "epoch": 2.0792394390432762,
      "grad_norm": 1.125,
      "learning_rate": 0.0006310325407062169,
      "loss": 1.037,
      "step": 24167
    },
    {
      "epoch": 2.079325475350598,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006310064448473141,
      "loss": 1.3165,
      "step": 24168
    },
    {
      "epoch": 2.0794115116579195,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006309803486052429,
      "loss": 0.6563,
      "step": 24169
    },
    {
      "epoch": 2.079497547965241,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006309542519800795,
      "loss": 1.2157,
      "step": 24170
    },
    {
      "epoch": 2.079583584272563,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006309281549719005,
      "loss": 1.0034,
      "step": 24171
    },
    {
      "epoch": 2.079669620579885,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006309020575807822,
      "loss": 0.909,
      "step": 24172
    },
    {
      "epoch": 2.0797556568872064,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006308759598068008,
      "loss": 0.8407,
      "step": 24173
    },
    {
      "epoch": 2.079841693194528,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006308498616500325,
      "loss": 1.0901,
      "step": 24174
    },
    {
      "epoch": 2.0799277295018497,
      "grad_norm": 1.03125,
      "learning_rate": 0.000630823763110554,
      "loss": 1.4925,
      "step": 24175
    },
    {
      "epoch": 2.0800137658091713,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006307976641884414,
      "loss": 1.1129,
      "step": 24176
    },
    {
      "epoch": 2.0800998021164934,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006307715648837711,
      "loss": 1.0652,
      "step": 24177
    },
    {
      "epoch": 2.080185838423815,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006307454651966193,
      "loss": 1.4403,
      "step": 24178
    },
    {
      "epoch": 2.0802718747311366,
      "grad_norm": 1.0,
      "learning_rate": 0.0006307193651270626,
      "loss": 0.7941,
      "step": 24179
    },
    {
      "epoch": 2.0803579110384582,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000630693264675177,
      "loss": 1.0096,
      "step": 24180
    },
    {
      "epoch": 2.08044394734578,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006306671638410391,
      "loss": 1.315,
      "step": 24181
    },
    {
      "epoch": 2.0805299836531015,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006306410626247252,
      "loss": 0.6941,
      "step": 24182
    },
    {
      "epoch": 2.080616019960423,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006306149610263114,
      "loss": 1.1503,
      "step": 24183
    },
    {
      "epoch": 2.080702056267745,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006305888590458744,
      "loss": 1.3705,
      "step": 24184
    },
    {
      "epoch": 2.080788092575067,
      "grad_norm": 1.0,
      "learning_rate": 0.0006305627566834904,
      "loss": 0.9321,
      "step": 24185
    },
    {
      "epoch": 2.0808741288823884,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006305366539392358,
      "loss": 2.2333,
      "step": 24186
    },
    {
      "epoch": 2.08096016518971,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006305105508131868,
      "loss": 1.4251,
      "step": 24187
    },
    {
      "epoch": 2.0810462014970317,
      "grad_norm": 1.5,
      "learning_rate": 0.0006304844473054198,
      "loss": 1.5084,
      "step": 24188
    },
    {
      "epoch": 2.0811322378043533,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006304583434160111,
      "loss": 0.761,
      "step": 24189
    },
    {
      "epoch": 2.081218274111675,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006304322391450372,
      "loss": 1.4159,
      "step": 24190
    },
    {
      "epoch": 2.081304310418997,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006304061344925744,
      "loss": 1.0461,
      "step": 24191
    },
    {
      "epoch": 2.0813903467263186,
      "grad_norm": 1.125,
      "learning_rate": 0.000630380029458699,
      "loss": 1.2093,
      "step": 24192
    },
    {
      "epoch": 2.0814763830336402,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006303539240434874,
      "loss": 0.8419,
      "step": 24193
    },
    {
      "epoch": 2.081562419340962,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006303278182470158,
      "loss": 0.8298,
      "step": 24194
    },
    {
      "epoch": 2.0816484556482835,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006303017120693608,
      "loss": 1.4159,
      "step": 24195
    },
    {
      "epoch": 2.081734491955605,
      "grad_norm": 1.25,
      "learning_rate": 0.0006302756055105985,
      "loss": 1.5422,
      "step": 24196
    },
    {
      "epoch": 2.081820528262927,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006302494985708056,
      "loss": 0.6959,
      "step": 24197
    },
    {
      "epoch": 2.081906564570249,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006302233912500581,
      "loss": 0.7913,
      "step": 24198
    },
    {
      "epoch": 2.0819926008775704,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006301972835484324,
      "loss": 0.639,
      "step": 24199
    },
    {
      "epoch": 2.082078637184892,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006301711754660052,
      "loss": 0.7133,
      "step": 24200
    },
    {
      "epoch": 2.0821646734922137,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006301450670028525,
      "loss": 1.4964,
      "step": 24201
    },
    {
      "epoch": 2.0822507097995353,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006301189581590508,
      "loss": 1.9213,
      "step": 24202
    },
    {
      "epoch": 2.082336746106857,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006300928489346765,
      "loss": 0.7524,
      "step": 24203
    },
    {
      "epoch": 2.082422782414179,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006300667393298058,
      "loss": 0.5015,
      "step": 24204
    },
    {
      "epoch": 2.0825088187215006,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006300406293445154,
      "loss": 1.6771,
      "step": 24205
    },
    {
      "epoch": 2.0825948550288222,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006300145189788814,
      "loss": 1.4088,
      "step": 24206
    },
    {
      "epoch": 2.082680891336144,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006299884082329801,
      "loss": 1.3352,
      "step": 24207
    },
    {
      "epoch": 2.0827669276434655,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000629962297106888,
      "loss": 1.2906,
      "step": 24208
    },
    {
      "epoch": 2.082852963950787,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006299361856006815,
      "loss": 0.6192,
      "step": 24209
    },
    {
      "epoch": 2.0829390002581087,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000629910073714437,
      "loss": 1.3969,
      "step": 24210
    },
    {
      "epoch": 2.083025036565431,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006298839614482306,
      "loss": 0.8654,
      "step": 24211
    },
    {
      "epoch": 2.0831110728727524,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006298578488021392,
      "loss": 1.1603,
      "step": 24212
    },
    {
      "epoch": 2.083197109180074,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006298317357762386,
      "loss": 1.2527,
      "step": 24213
    },
    {
      "epoch": 2.0832831454873957,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006298056223706055,
      "loss": 1.0557,
      "step": 24214
    },
    {
      "epoch": 2.0833691817947173,
      "grad_norm": 1.171875,
      "learning_rate": 0.000629779508585316,
      "loss": 1.5912,
      "step": 24215
    },
    {
      "epoch": 2.083455218102039,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000629753394420447,
      "loss": 2.1367,
      "step": 24216
    },
    {
      "epoch": 2.0835412544093606,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006297272798760745,
      "loss": 1.4502,
      "step": 24217
    },
    {
      "epoch": 2.0836272907166826,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006297011649522747,
      "loss": 0.8374,
      "step": 24218
    },
    {
      "epoch": 2.0837133270240042,
      "grad_norm": 1.0,
      "learning_rate": 0.0006296750496491243,
      "loss": 0.9811,
      "step": 24219
    },
    {
      "epoch": 2.083799363331326,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006296489339666997,
      "loss": 1.2292,
      "step": 24220
    },
    {
      "epoch": 2.0838853996386475,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006296228179050771,
      "loss": 1.7237,
      "step": 24221
    },
    {
      "epoch": 2.083971435945969,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006295967014643331,
      "loss": 0.8838,
      "step": 24222
    },
    {
      "epoch": 2.0840574722532907,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006295705846445439,
      "loss": 1.1374,
      "step": 24223
    },
    {
      "epoch": 2.0841435085606124,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006295444674457859,
      "loss": 1.1173,
      "step": 24224
    },
    {
      "epoch": 2.0842295448679344,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006295183498681355,
      "loss": 1.9497,
      "step": 24225
    },
    {
      "epoch": 2.084315581175256,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006294922319116692,
      "loss": 1.4631,
      "step": 24226
    },
    {
      "epoch": 2.0844016174825777,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006294661135764632,
      "loss": 0.7834,
      "step": 24227
    },
    {
      "epoch": 2.0844876537898993,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006294399948625941,
      "loss": 1.2571,
      "step": 24228
    },
    {
      "epoch": 2.084573690097221,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006294138757701382,
      "loss": 0.9114,
      "step": 24229
    },
    {
      "epoch": 2.0846597264045426,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006293877562991716,
      "loss": 1.0041,
      "step": 24230
    },
    {
      "epoch": 2.084745762711864,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006293616364497714,
      "loss": 0.6538,
      "step": 24231
    },
    {
      "epoch": 2.0848317990191862,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006293355162220133,
      "loss": 1.6573,
      "step": 24232
    },
    {
      "epoch": 2.084917835326508,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000629309395615974,
      "loss": 0.5221,
      "step": 24233
    },
    {
      "epoch": 2.0850038716338295,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006292832746317299,
      "loss": 1.1395,
      "step": 24234
    },
    {
      "epoch": 2.085089907941151,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006292571532693575,
      "loss": 0.7548,
      "step": 24235
    },
    {
      "epoch": 2.0851759442484727,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000629231031528933,
      "loss": 1.2333,
      "step": 24236
    },
    {
      "epoch": 2.0852619805557944,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006292049094105327,
      "loss": 1.3446,
      "step": 24237
    },
    {
      "epoch": 2.0853480168631164,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006291787869142333,
      "loss": 1.4051,
      "step": 24238
    },
    {
      "epoch": 2.085434053170438,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006291526640401111,
      "loss": 1.4348,
      "step": 24239
    },
    {
      "epoch": 2.0855200894777597,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006291265407882423,
      "loss": 1.3731,
      "step": 24240
    },
    {
      "epoch": 2.0856061257850813,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006291004171587036,
      "loss": 1.2497,
      "step": 24241
    },
    {
      "epoch": 2.085692162092403,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006290742931515714,
      "loss": 1.3031,
      "step": 24242
    },
    {
      "epoch": 2.0857781983997246,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006290481687669219,
      "loss": 0.7707,
      "step": 24243
    },
    {
      "epoch": 2.085864234707046,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006290220440048317,
      "loss": 0.9144,
      "step": 24244
    },
    {
      "epoch": 2.0859502710143683,
      "grad_norm": 1.0,
      "learning_rate": 0.0006289959188653768,
      "loss": 0.809,
      "step": 24245
    },
    {
      "epoch": 2.08603630732169,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006289697933486342,
      "loss": 0.6782,
      "step": 24246
    },
    {
      "epoch": 2.0861223436290115,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00062894366745468,
      "loss": 0.9229,
      "step": 24247
    },
    {
      "epoch": 2.086208379936333,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006289175411835906,
      "loss": 1.0721,
      "step": 24248
    },
    {
      "epoch": 2.0862944162436547,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006288914145354426,
      "loss": 1.1723,
      "step": 24249
    },
    {
      "epoch": 2.0863804525509764,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006288652875103122,
      "loss": 1.2944,
      "step": 24250
    },
    {
      "epoch": 2.086466488858298,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006288391601082759,
      "loss": 1.1778,
      "step": 24251
    },
    {
      "epoch": 2.08655252516562,
      "grad_norm": 1.203125,
      "learning_rate": 0.00062881303232941,
      "loss": 0.8499,
      "step": 24252
    },
    {
      "epoch": 2.0866385614729417,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006287869041737912,
      "loss": 1.1467,
      "step": 24253
    },
    {
      "epoch": 2.0867245977802633,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006287607756414957,
      "loss": 0.8914,
      "step": 24254
    },
    {
      "epoch": 2.086810634087585,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006287346467325999,
      "loss": 1.1715,
      "step": 24255
    },
    {
      "epoch": 2.0868966703949066,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006287085174471802,
      "loss": 1.1535,
      "step": 24256
    },
    {
      "epoch": 2.086982706702228,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006286823877853133,
      "loss": 1.1057,
      "step": 24257
    },
    {
      "epoch": 2.0870687430095503,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006286562577470753,
      "loss": 1.3049,
      "step": 24258
    },
    {
      "epoch": 2.087154779316872,
      "grad_norm": 1.078125,
      "learning_rate": 0.000628630127332543,
      "loss": 0.8568,
      "step": 24259
    },
    {
      "epoch": 2.0872408156241935,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006286039965417925,
      "loss": 1.0239,
      "step": 24260
    },
    {
      "epoch": 2.087326851931515,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006285778653749003,
      "loss": 0.7957,
      "step": 24261
    },
    {
      "epoch": 2.0874128882388368,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006285517338319429,
      "loss": 0.8746,
      "step": 24262
    },
    {
      "epoch": 2.0874989245461584,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006285256019129964,
      "loss": 1.2778,
      "step": 24263
    },
    {
      "epoch": 2.08758496085348,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0006284994696181379,
      "loss": 1.6469,
      "step": 24264
    },
    {
      "epoch": 2.087670997160802,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006284733369474432,
      "loss": 0.9987,
      "step": 24265
    },
    {
      "epoch": 2.0877570334681237,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006284472039009892,
      "loss": 1.2934,
      "step": 24266
    },
    {
      "epoch": 2.0878430697754453,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006284210704788518,
      "loss": 0.6269,
      "step": 24267
    },
    {
      "epoch": 2.087929106082767,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006283949366811079,
      "loss": 1.6361,
      "step": 24268
    },
    {
      "epoch": 2.0880151423900886,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006283688025078339,
      "loss": 1.5403,
      "step": 24269
    },
    {
      "epoch": 2.08810117869741,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000628342667959106,
      "loss": 1.2539,
      "step": 24270
    },
    {
      "epoch": 2.088187215004732,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006283165330350007,
      "loss": 0.982,
      "step": 24271
    },
    {
      "epoch": 2.088273251312054,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006282903977355946,
      "loss": 1.1545,
      "step": 24272
    },
    {
      "epoch": 2.0883592876193755,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000628264262060964,
      "loss": 1.0718,
      "step": 24273
    },
    {
      "epoch": 2.088445323926697,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006282381260111854,
      "loss": 2.1629,
      "step": 24274
    },
    {
      "epoch": 2.0885313602340188,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006282119895863352,
      "loss": 0.7008,
      "step": 24275
    },
    {
      "epoch": 2.0886173965413404,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006281858527864897,
      "loss": 1.291,
      "step": 24276
    },
    {
      "epoch": 2.088703432848662,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006281597156117258,
      "loss": 1.2923,
      "step": 24277
    },
    {
      "epoch": 2.0887894691559836,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006281335780621196,
      "loss": 1.1746,
      "step": 24278
    },
    {
      "epoch": 2.0888755054633057,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006281074401377475,
      "loss": 1.3008,
      "step": 24279
    },
    {
      "epoch": 2.0889615417706273,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000628081301838686,
      "loss": 1.2075,
      "step": 24280
    },
    {
      "epoch": 2.089047578077949,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006280551631650118,
      "loss": 1.2734,
      "step": 24281
    },
    {
      "epoch": 2.0891336143852706,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000628029024116801,
      "loss": 2.2911,
      "step": 24282
    },
    {
      "epoch": 2.089219650692592,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006280028846941302,
      "loss": 1.3282,
      "step": 24283
    },
    {
      "epoch": 2.089305686999914,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000627976744897076,
      "loss": 1.5413,
      "step": 24284
    },
    {
      "epoch": 2.0893917233072354,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006279506047257147,
      "loss": 1.5031,
      "step": 24285
    },
    {
      "epoch": 2.0894777596145575,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006279244641801226,
      "loss": 1.1611,
      "step": 24286
    },
    {
      "epoch": 2.089563795921879,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006278983232603765,
      "loss": 1.0956,
      "step": 24287
    },
    {
      "epoch": 2.0896498322292008,
      "grad_norm": 1.609375,
      "learning_rate": 0.0006278721819665526,
      "loss": 1.4591,
      "step": 24288
    },
    {
      "epoch": 2.0897358685365224,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006278460402987274,
      "loss": 1.739,
      "step": 24289
    },
    {
      "epoch": 2.089821904843844,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006278198982569774,
      "loss": 1.2655,
      "step": 24290
    },
    {
      "epoch": 2.0899079411511656,
      "grad_norm": 3.09375,
      "learning_rate": 0.0006277937558413791,
      "loss": 1.5986,
      "step": 24291
    },
    {
      "epoch": 2.0899939774584877,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006277676130520089,
      "loss": 1.2045,
      "step": 24292
    },
    {
      "epoch": 2.0900800137658093,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006277414698889432,
      "loss": 1.9441,
      "step": 24293
    },
    {
      "epoch": 2.090166050073131,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006277153263522586,
      "loss": 0.4786,
      "step": 24294
    },
    {
      "epoch": 2.0902520863804526,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006276891824420317,
      "loss": 1.0611,
      "step": 24295
    },
    {
      "epoch": 2.090338122687774,
      "grad_norm": 46.25,
      "learning_rate": 0.0006276630381583386,
      "loss": 0.9492,
      "step": 24296
    },
    {
      "epoch": 2.090424158995096,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006276368935012558,
      "loss": 1.1328,
      "step": 24297
    },
    {
      "epoch": 2.0905101953024174,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006276107484708602,
      "loss": 0.4694,
      "step": 24298
    },
    {
      "epoch": 2.0905962316097395,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006275846030672278,
      "loss": 0.9747,
      "step": 24299
    },
    {
      "epoch": 2.090682267917061,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006275584572904354,
      "loss": 0.7806,
      "step": 24300
    },
    {
      "epoch": 2.0907683042243828,
      "grad_norm": 1.0625,
      "learning_rate": 0.000627532311140559,
      "loss": 1.0531,
      "step": 24301
    },
    {
      "epoch": 2.0908543405317044,
      "grad_norm": 2.390625,
      "learning_rate": 0.0006275061646176757,
      "loss": 1.4172,
      "step": 24302
    },
    {
      "epoch": 2.090940376839026,
      "grad_norm": 2.796875,
      "learning_rate": 0.0006274800177218616,
      "loss": 1.2031,
      "step": 24303
    },
    {
      "epoch": 2.0910264131463476,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006274538704531932,
      "loss": 1.4026,
      "step": 24304
    },
    {
      "epoch": 2.0911124494536693,
      "grad_norm": 18.875,
      "learning_rate": 0.000627427722811747,
      "loss": 1.6082,
      "step": 24305
    },
    {
      "epoch": 2.0911984857609913,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006274015747975996,
      "loss": 1.4014,
      "step": 24306
    },
    {
      "epoch": 2.091284522068313,
      "grad_norm": 2.84375,
      "learning_rate": 0.0006273754264108273,
      "loss": 1.0273,
      "step": 24307
    },
    {
      "epoch": 2.0913705583756346,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006273492776515066,
      "loss": 1.2137,
      "step": 24308
    },
    {
      "epoch": 2.091456594682956,
      "grad_norm": 13.625,
      "learning_rate": 0.0006273231285197141,
      "loss": 1.6804,
      "step": 24309
    },
    {
      "epoch": 2.091542630990278,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006272969790155263,
      "loss": 1.1543,
      "step": 24310
    },
    {
      "epoch": 2.0916286672975994,
      "grad_norm": 1.125,
      "learning_rate": 0.0006272708291390196,
      "loss": 1.6281,
      "step": 24311
    },
    {
      "epoch": 2.091714703604921,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006272446788902703,
      "loss": 0.6965,
      "step": 24312
    },
    {
      "epoch": 2.091800739912243,
      "grad_norm": 0.8125,
      "learning_rate": 0.000627218528269355,
      "loss": 1.3442,
      "step": 24313
    },
    {
      "epoch": 2.0918867762195648,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006271923772763504,
      "loss": 0.921,
      "step": 24314
    },
    {
      "epoch": 2.0919728125268864,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006271662259113329,
      "loss": 1.8779,
      "step": 24315
    },
    {
      "epoch": 2.092058848834208,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006271400741743787,
      "loss": 0.8868,
      "step": 24316
    },
    {
      "epoch": 2.0921448851415296,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006271139220655647,
      "loss": 1.1162,
      "step": 24317
    },
    {
      "epoch": 2.0922309214488513,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006270877695849672,
      "loss": 1.5613,
      "step": 24318
    },
    {
      "epoch": 2.0923169577561733,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006270616167326625,
      "loss": 1.2276,
      "step": 24319
    },
    {
      "epoch": 2.092402994063495,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006270354635087275,
      "loss": 1.4542,
      "step": 24320
    },
    {
      "epoch": 2.0924890303708166,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006270093099132384,
      "loss": 0.5676,
      "step": 24321
    },
    {
      "epoch": 2.092575066678138,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006269831559462717,
      "loss": 1.2773,
      "step": 24322
    },
    {
      "epoch": 2.09266110298546,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006269570016079041,
      "loss": 1.0763,
      "step": 24323
    },
    {
      "epoch": 2.0927471392927814,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006269308468982119,
      "loss": 1.0671,
      "step": 24324
    },
    {
      "epoch": 2.092833175600103,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006269046918172717,
      "loss": 1.6602,
      "step": 24325
    },
    {
      "epoch": 2.092919211907425,
      "grad_norm": 0.90625,
      "learning_rate": 0.00062687853636516,
      "loss": 1.4547,
      "step": 24326
    },
    {
      "epoch": 2.0930052482147468,
      "grad_norm": 1.03125,
      "learning_rate": 0.000626852380541953,
      "loss": 0.9934,
      "step": 24327
    },
    {
      "epoch": 2.0930912845220684,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006268262243477276,
      "loss": 1.2229,
      "step": 24328
    },
    {
      "epoch": 2.09317732082939,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006268000677825602,
      "loss": 0.9161,
      "step": 24329
    },
    {
      "epoch": 2.0932633571367116,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006267739108465274,
      "loss": 1.329,
      "step": 24330
    },
    {
      "epoch": 2.0933493934440333,
      "grad_norm": 0.875,
      "learning_rate": 0.0006267477535397052,
      "loss": 1.0173,
      "step": 24331
    },
    {
      "epoch": 2.093435429751355,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006267215958621708,
      "loss": 1.3769,
      "step": 24332
    },
    {
      "epoch": 2.093521466058677,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006266954378140003,
      "loss": 1.3603,
      "step": 24333
    },
    {
      "epoch": 2.0936075023659986,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006266692793952702,
      "loss": 1.801,
      "step": 24334
    },
    {
      "epoch": 2.09369353867332,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006266431206060571,
      "loss": 1.3939,
      "step": 24335
    },
    {
      "epoch": 2.093779574980642,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006266169614464376,
      "loss": 1.6297,
      "step": 24336
    },
    {
      "epoch": 2.0938656112879634,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006265908019164881,
      "loss": 0.8388,
      "step": 24337
    },
    {
      "epoch": 2.093951647595285,
      "grad_norm": 1.109375,
      "learning_rate": 0.000626564642016285,
      "loss": 0.8929,
      "step": 24338
    },
    {
      "epoch": 2.0940376839026067,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006265384817459052,
      "loss": 1.4136,
      "step": 24339
    },
    {
      "epoch": 2.0941237202099288,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006265123211054247,
      "loss": 1.5948,
      "step": 24340
    },
    {
      "epoch": 2.0942097565172504,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006264861600949203,
      "loss": 1.6124,
      "step": 24341
    },
    {
      "epoch": 2.094295792824572,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006264599987144685,
      "loss": 0.8755,
      "step": 24342
    },
    {
      "epoch": 2.0943818291318936,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006264338369641459,
      "loss": 1.1215,
      "step": 24343
    },
    {
      "epoch": 2.0944678654392153,
      "grad_norm": 2.09375,
      "learning_rate": 0.0006264076748440288,
      "loss": 1.3188,
      "step": 24344
    },
    {
      "epoch": 2.094553901746537,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006263815123541939,
      "loss": 0.8126,
      "step": 24345
    },
    {
      "epoch": 2.0946399380538585,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006263553494947177,
      "loss": 1.0777,
      "step": 24346
    },
    {
      "epoch": 2.0947259743611806,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006263291862656766,
      "loss": 1.6618,
      "step": 24347
    },
    {
      "epoch": 2.094812010668502,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006263030226671473,
      "loss": 1.3421,
      "step": 24348
    },
    {
      "epoch": 2.094898046975824,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000626276858699206,
      "loss": 1.4737,
      "step": 24349
    },
    {
      "epoch": 2.0949840832831454,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006262506943619295,
      "loss": 1.0121,
      "step": 24350
    },
    {
      "epoch": 2.095070119590467,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0006262245296553943,
      "loss": 0.8992,
      "step": 24351
    },
    {
      "epoch": 2.0951561558977887,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000626198364579677,
      "loss": 1.2495,
      "step": 24352
    },
    {
      "epoch": 2.0952421922051108,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006261721991348539,
      "loss": 0.682,
      "step": 24353
    },
    {
      "epoch": 2.0953282285124324,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006261460333210018,
      "loss": 1.5198,
      "step": 24354
    },
    {
      "epoch": 2.095414264819754,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006261198671381971,
      "loss": 0.8702,
      "step": 24355
    },
    {
      "epoch": 2.0955003011270756,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006260937005865163,
      "loss": 0.9141,
      "step": 24356
    },
    {
      "epoch": 2.0955863374343973,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006260675336660357,
      "loss": 1.268,
      "step": 24357
    },
    {
      "epoch": 2.095672373741719,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006260413663768324,
      "loss": 1.1672,
      "step": 24358
    },
    {
      "epoch": 2.0957584100490405,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006260151987189825,
      "loss": 1.5584,
      "step": 24359
    },
    {
      "epoch": 2.0958444463563626,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006259890306925626,
      "loss": 1.8804,
      "step": 24360
    },
    {
      "epoch": 2.095930482663684,
      "grad_norm": 1.71875,
      "learning_rate": 0.0006259628622976494,
      "loss": 1.485,
      "step": 24361
    },
    {
      "epoch": 2.096016518971006,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006259366935343193,
      "loss": 1.0374,
      "step": 24362
    },
    {
      "epoch": 2.0961025552783275,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006259105244026488,
      "loss": 1.6275,
      "step": 24363
    },
    {
      "epoch": 2.096188591585649,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006258843549027145,
      "loss": 0.77,
      "step": 24364
    },
    {
      "epoch": 2.0962746278929707,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000625858185034593,
      "loss": 1.2484,
      "step": 24365
    },
    {
      "epoch": 2.0963606642002923,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006258320147983607,
      "loss": 1.9462,
      "step": 24366
    },
    {
      "epoch": 2.0964467005076144,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006258058441940944,
      "loss": 2.0053,
      "step": 24367
    },
    {
      "epoch": 2.096532736814936,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006257796732218701,
      "loss": 0.731,
      "step": 24368
    },
    {
      "epoch": 2.0966187731222576,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000625753501881765,
      "loss": 1.832,
      "step": 24369
    },
    {
      "epoch": 2.0967048094295793,
      "grad_norm": 1.0,
      "learning_rate": 0.0006257273301738556,
      "loss": 1.4968,
      "step": 24370
    },
    {
      "epoch": 2.096790845736901,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006257011580982179,
      "loss": 0.8992,
      "step": 24371
    },
    {
      "epoch": 2.0968768820442225,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006256749856549288,
      "loss": 0.9926,
      "step": 24372
    },
    {
      "epoch": 2.0969629183515446,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0006256488128440648,
      "loss": 2.0625,
      "step": 24373
    },
    {
      "epoch": 2.097048954658866,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006256226396657025,
      "loss": 0.5882,
      "step": 24374
    },
    {
      "epoch": 2.097134990966188,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006255964661199182,
      "loss": 1.301,
      "step": 24375
    },
    {
      "epoch": 2.0972210272735095,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006255702922067889,
      "loss": 2.0983,
      "step": 24376
    },
    {
      "epoch": 2.097307063580831,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000625544117926391,
      "loss": 0.7925,
      "step": 24377
    },
    {
      "epoch": 2.0973930998881527,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000625517943278801,
      "loss": 1.2486,
      "step": 24378
    },
    {
      "epoch": 2.0974791361954743,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006254917682640952,
      "loss": 1.4188,
      "step": 24379
    },
    {
      "epoch": 2.0975651725027964,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006254655928823505,
      "loss": 1.4411,
      "step": 24380
    },
    {
      "epoch": 2.097651208810118,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006254394171336433,
      "loss": 0.4841,
      "step": 24381
    },
    {
      "epoch": 2.0977372451174396,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006254132410180501,
      "loss": 0.8896,
      "step": 24382
    },
    {
      "epoch": 2.0978232814247613,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006253870645356478,
      "loss": 0.9822,
      "step": 24383
    },
    {
      "epoch": 2.097909317732083,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006253608876865126,
      "loss": 1.1171,
      "step": 24384
    },
    {
      "epoch": 2.0979953540394045,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006253347104707212,
      "loss": 1.498,
      "step": 24385
    },
    {
      "epoch": 2.098081390346726,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00062530853288835,
      "loss": 1.1597,
      "step": 24386
    },
    {
      "epoch": 2.098167426654048,
      "grad_norm": 1.5,
      "learning_rate": 0.0006252823549394757,
      "loss": 0.8484,
      "step": 24387
    },
    {
      "epoch": 2.09825346296137,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006252561766241751,
      "loss": 0.9432,
      "step": 24388
    },
    {
      "epoch": 2.0983394992686915,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006252299979425246,
      "loss": 0.7315,
      "step": 24389
    },
    {
      "epoch": 2.098425535576013,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006252038188946005,
      "loss": 1.5675,
      "step": 24390
    },
    {
      "epoch": 2.0985115718833347,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006251776394804796,
      "loss": 1.3492,
      "step": 24391
    },
    {
      "epoch": 2.0985976081906563,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006251514597002386,
      "loss": 1.0344,
      "step": 24392
    },
    {
      "epoch": 2.098683644497978,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006251252795539538,
      "loss": 2.2312,
      "step": 24393
    },
    {
      "epoch": 2.0987696808053,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006250990990417018,
      "loss": 1.1366,
      "step": 24394
    },
    {
      "epoch": 2.0988557171126216,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006250729181635595,
      "loss": 1.0803,
      "step": 24395
    },
    {
      "epoch": 2.0989417534199433,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006250467369196031,
      "loss": 0.8143,
      "step": 24396
    },
    {
      "epoch": 2.099027789727265,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006250205553099094,
      "loss": 0.9811,
      "step": 24397
    },
    {
      "epoch": 2.0991138260345865,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0006249943733345548,
      "loss": 1.9137,
      "step": 24398
    },
    {
      "epoch": 2.099199862341908,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000624968190993616,
      "loss": 0.9977,
      "step": 24399
    },
    {
      "epoch": 2.0992858986492298,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006249420082871695,
      "loss": 1.8324,
      "step": 24400
    },
    {
      "epoch": 2.099371934956552,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000624915825215292,
      "loss": 0.789,
      "step": 24401
    },
    {
      "epoch": 2.0994579712638735,
      "grad_norm": 1.21875,
      "learning_rate": 0.00062488964177806,
      "loss": 1.1205,
      "step": 24402
    },
    {
      "epoch": 2.099544007571195,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006248634579755501,
      "loss": 0.9518,
      "step": 24403
    },
    {
      "epoch": 2.0996300438785167,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006248372738078389,
      "loss": 1.1932,
      "step": 24404
    },
    {
      "epoch": 2.0997160801858383,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006248110892750027,
      "loss": 0.7324,
      "step": 24405
    },
    {
      "epoch": 2.09980211649316,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006247849043771185,
      "loss": 1.4214,
      "step": 24406
    },
    {
      "epoch": 2.0998881528004816,
      "grad_norm": 1.5,
      "learning_rate": 0.0006247587191142627,
      "loss": 1.6795,
      "step": 24407
    },
    {
      "epoch": 2.0999741891078036,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000624732533486512,
      "loss": 1.6368,
      "step": 24408
    },
    {
      "epoch": 2.1000602254151253,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006247063474939428,
      "loss": 1.365,
      "step": 24409
    },
    {
      "epoch": 2.100146261722447,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000624680161136632,
      "loss": 1.6471,
      "step": 24410
    },
    {
      "epoch": 2.1002322980297685,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006246539744146558,
      "loss": 1.073,
      "step": 24411
    },
    {
      "epoch": 2.10031833433709,
      "grad_norm": 1.125,
      "learning_rate": 0.000624627787328091,
      "loss": 0.9431,
      "step": 24412
    },
    {
      "epoch": 2.1004043706444118,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006246015998770142,
      "loss": 1.219,
      "step": 24413
    },
    {
      "epoch": 2.100490406951734,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006245754120615018,
      "loss": 1.055,
      "step": 24414
    },
    {
      "epoch": 2.1005764432590555,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006245492238816307,
      "loss": 0.9761,
      "step": 24415
    },
    {
      "epoch": 2.100662479566377,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006245230353374772,
      "loss": 0.9222,
      "step": 24416
    },
    {
      "epoch": 2.1007485158736987,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006244968464291182,
      "loss": 1.3701,
      "step": 24417
    },
    {
      "epoch": 2.1008345521810203,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006244706571566301,
      "loss": 2.1143,
      "step": 24418
    },
    {
      "epoch": 2.100920588488342,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006244444675200895,
      "loss": 1.5983,
      "step": 24419
    },
    {
      "epoch": 2.1010066247956636,
      "grad_norm": 1.9609375,
      "learning_rate": 0.000624418277519573,
      "loss": 1.3683,
      "step": 24420
    },
    {
      "epoch": 2.1010926611029856,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006243920871551574,
      "loss": 0.5754,
      "step": 24421
    },
    {
      "epoch": 2.1011786974103073,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000624365896426919,
      "loss": 1.7806,
      "step": 24422
    },
    {
      "epoch": 2.101264733717629,
      "grad_norm": 1.625,
      "learning_rate": 0.0006243397053349347,
      "loss": 1.2867,
      "step": 24423
    },
    {
      "epoch": 2.1013507700249505,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006243135138792806,
      "loss": 0.9489,
      "step": 24424
    },
    {
      "epoch": 2.101436806332272,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006242873220600338,
      "loss": 1.0418,
      "step": 24425
    },
    {
      "epoch": 2.1015228426395938,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006242611298772709,
      "loss": 1.2149,
      "step": 24426
    },
    {
      "epoch": 2.1016088789469154,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006242349373310683,
      "loss": 1.1428,
      "step": 24427
    },
    {
      "epoch": 2.1016949152542375,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006242087444215025,
      "loss": 1.0269,
      "step": 24428
    },
    {
      "epoch": 2.101780951561559,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006241825511486505,
      "loss": 1.2452,
      "step": 24429
    },
    {
      "epoch": 2.1018669878688807,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006241563575125887,
      "loss": 1.1881,
      "step": 24430
    },
    {
      "epoch": 2.1019530241762023,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006241301635133936,
      "loss": 1.0759,
      "step": 24431
    },
    {
      "epoch": 2.102039060483524,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006241039691511419,
      "loss": 1.2506,
      "step": 24432
    },
    {
      "epoch": 2.1021250967908456,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006240777744259103,
      "loss": 1.2639,
      "step": 24433
    },
    {
      "epoch": 2.1022111330981677,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006240515793377754,
      "loss": 0.7099,
      "step": 24434
    },
    {
      "epoch": 2.1022971694054893,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006240253838868136,
      "loss": 1.6997,
      "step": 24435
    },
    {
      "epoch": 2.102383205712811,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006239991880731017,
      "loss": 2.1124,
      "step": 24436
    },
    {
      "epoch": 2.1024692420201325,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006239729918967163,
      "loss": 0.6205,
      "step": 24437
    },
    {
      "epoch": 2.102555278327454,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006239467953577341,
      "loss": 0.9709,
      "step": 24438
    },
    {
      "epoch": 2.1026413146347758,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006239205984562315,
      "loss": 1.5399,
      "step": 24439
    },
    {
      "epoch": 2.1027273509420974,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006238944011922854,
      "loss": 1.7776,
      "step": 24440
    },
    {
      "epoch": 2.1028133872494195,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006238682035659722,
      "loss": 0.8474,
      "step": 24441
    },
    {
      "epoch": 2.102899423556741,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006238420055773684,
      "loss": 1.3652,
      "step": 24442
    },
    {
      "epoch": 2.1029854598640627,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006238158072265509,
      "loss": 2.0359,
      "step": 24443
    },
    {
      "epoch": 2.1030714961713843,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006237896085135964,
      "loss": 1.8002,
      "step": 24444
    },
    {
      "epoch": 2.103157532478706,
      "grad_norm": 1.0,
      "learning_rate": 0.0006237634094385813,
      "loss": 1.2795,
      "step": 24445
    },
    {
      "epoch": 2.1032435687860276,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006237372100015825,
      "loss": 1.1099,
      "step": 24446
    },
    {
      "epoch": 2.103329605093349,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006237110102026762,
      "loss": 1.1127,
      "step": 24447
    },
    {
      "epoch": 2.1034156414006713,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006236848100419392,
      "loss": 1.688,
      "step": 24448
    },
    {
      "epoch": 2.103501677707993,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006236586095194483,
      "loss": 1.1227,
      "step": 24449
    },
    {
      "epoch": 2.1035877140153145,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006236324086352801,
      "loss": 1.1913,
      "step": 24450
    },
    {
      "epoch": 2.103673750322636,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000623606207389511,
      "loss": 1.0257,
      "step": 24451
    },
    {
      "epoch": 2.1037597866299578,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006235800057822179,
      "loss": 1.0747,
      "step": 24452
    },
    {
      "epoch": 2.1038458229372794,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006235538038134773,
      "loss": 1.3059,
      "step": 24453
    },
    {
      "epoch": 2.103931859244601,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006235276014833656,
      "loss": 1.2134,
      "step": 24454
    },
    {
      "epoch": 2.104017895551923,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00062350139879196,
      "loss": 0.6527,
      "step": 24455
    },
    {
      "epoch": 2.1041039318592447,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006234751957393367,
      "loss": 1.4352,
      "step": 24456
    },
    {
      "epoch": 2.1041899681665663,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006234489923255725,
      "loss": 0.8245,
      "step": 24457
    },
    {
      "epoch": 2.104276004473888,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000623422788550744,
      "loss": 1.3125,
      "step": 24458
    },
    {
      "epoch": 2.1043620407812096,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006233965844149278,
      "loss": 1.8678,
      "step": 24459
    },
    {
      "epoch": 2.104448077088531,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006233703799182008,
      "loss": 0.7548,
      "step": 24460
    },
    {
      "epoch": 2.104534113395853,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006233441750606391,
      "loss": 2.0102,
      "step": 24461
    },
    {
      "epoch": 2.104620149703175,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006233179698423198,
      "loss": 1.3245,
      "step": 24462
    },
    {
      "epoch": 2.1047061860104965,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006232917642633195,
      "loss": 0.6427,
      "step": 24463
    },
    {
      "epoch": 2.104792222317818,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006232655583237148,
      "loss": 1.289,
      "step": 24464
    },
    {
      "epoch": 2.1048782586251398,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006232393520235822,
      "loss": 0.7063,
      "step": 24465
    },
    {
      "epoch": 2.1049642949324614,
      "grad_norm": 1.0,
      "learning_rate": 0.0006232131453629985,
      "loss": 1.2607,
      "step": 24466
    },
    {
      "epoch": 2.105050331239783,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006231869383420403,
      "loss": 0.5384,
      "step": 24467
    },
    {
      "epoch": 2.105136367547105,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006231607309607844,
      "loss": 0.4798,
      "step": 24468
    },
    {
      "epoch": 2.1052224038544267,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006231345232193071,
      "loss": 1.0631,
      "step": 24469
    },
    {
      "epoch": 2.1053084401617483,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006231083151176853,
      "loss": 1.1778,
      "step": 24470
    },
    {
      "epoch": 2.10539447646907,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006230821066559958,
      "loss": 1.6887,
      "step": 24471
    },
    {
      "epoch": 2.1054805127763916,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006230558978343149,
      "loss": 1.2284,
      "step": 24472
    },
    {
      "epoch": 2.105566549083713,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006230296886527196,
      "loss": 1.7034,
      "step": 24473
    },
    {
      "epoch": 2.105652585391035,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006230034791112862,
      "loss": 1.408,
      "step": 24474
    },
    {
      "epoch": 2.105738621698357,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006229772692100918,
      "loss": 1.048,
      "step": 24475
    },
    {
      "epoch": 2.1058246580056785,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006229510589492125,
      "loss": 1.2507,
      "step": 24476
    },
    {
      "epoch": 2.105910694313,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006229248483287256,
      "loss": 0.8679,
      "step": 24477
    },
    {
      "epoch": 2.1059967306203218,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006228986373487072,
      "loss": 1.3249,
      "step": 24478
    },
    {
      "epoch": 2.1060827669276434,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006228724260092341,
      "loss": 1.5957,
      "step": 24479
    },
    {
      "epoch": 2.106168803234965,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006228462143103832,
      "loss": 0.718,
      "step": 24480
    },
    {
      "epoch": 2.1062548395422866,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000622820002252231,
      "loss": 1.3124,
      "step": 24481
    },
    {
      "epoch": 2.1063408758496087,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006227937898348542,
      "loss": 0.7534,
      "step": 24482
    },
    {
      "epoch": 2.1064269121569303,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006227675770583294,
      "loss": 1.0191,
      "step": 24483
    },
    {
      "epoch": 2.106512948464252,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006227413639227333,
      "loss": 1.4032,
      "step": 24484
    },
    {
      "epoch": 2.1065989847715736,
      "grad_norm": 1.5,
      "learning_rate": 0.0006227151504281426,
      "loss": 1.8512,
      "step": 24485
    },
    {
      "epoch": 2.106685021078895,
      "grad_norm": 1.1875,
      "learning_rate": 0.000622688936574634,
      "loss": 1.3376,
      "step": 24486
    },
    {
      "epoch": 2.106771057386217,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006226627223622842,
      "loss": 1.4671,
      "step": 24487
    },
    {
      "epoch": 2.1068570936935385,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006226365077911696,
      "loss": 0.7048,
      "step": 24488
    },
    {
      "epoch": 2.1069431300008605,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006226102928613672,
      "loss": 1.316,
      "step": 24489
    },
    {
      "epoch": 2.107029166308182,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006225840775729534,
      "loss": 1.6925,
      "step": 24490
    },
    {
      "epoch": 2.107115202615504,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006225578619260052,
      "loss": 0.9783,
      "step": 24491
    },
    {
      "epoch": 2.1072012389228254,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000622531645920599,
      "loss": 0.7677,
      "step": 24492
    },
    {
      "epoch": 2.107287275230147,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006225054295568115,
      "loss": 2.0491,
      "step": 24493
    },
    {
      "epoch": 2.1073733115374687,
      "grad_norm": 2.84375,
      "learning_rate": 0.0006224792128347197,
      "loss": 1.9774,
      "step": 24494
    },
    {
      "epoch": 2.1074593478447907,
      "grad_norm": 1.125,
      "learning_rate": 0.0006224529957543996,
      "loss": 1.8128,
      "step": 24495
    },
    {
      "epoch": 2.1075453841521123,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006224267783159287,
      "loss": 2.0485,
      "step": 24496
    },
    {
      "epoch": 2.107631420459434,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006224005605193831,
      "loss": 1.0494,
      "step": 24497
    },
    {
      "epoch": 2.1077174567667556,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006223743423648396,
      "loss": 0.8681,
      "step": 24498
    },
    {
      "epoch": 2.107803493074077,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000622348123852375,
      "loss": 0.9393,
      "step": 24499
    },
    {
      "epoch": 2.107889529381399,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006223219049820659,
      "loss": 0.8381,
      "step": 24500
    },
    {
      "epoch": 2.1079755656887205,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006222956857539891,
      "loss": 1.1659,
      "step": 24501
    },
    {
      "epoch": 2.1080616019960425,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006222694661682213,
      "loss": 0.7404,
      "step": 24502
    },
    {
      "epoch": 2.108147638303364,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006222432462248389,
      "loss": 1.449,
      "step": 24503
    },
    {
      "epoch": 2.108233674610686,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006222170259239188,
      "loss": 1.537,
      "step": 24504
    },
    {
      "epoch": 2.1083197109180074,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006221908052655378,
      "loss": 1.3259,
      "step": 24505
    },
    {
      "epoch": 2.108405747225329,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006221645842497723,
      "loss": 2.2362,
      "step": 24506
    },
    {
      "epoch": 2.1084917835326507,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006221383628766993,
      "loss": 0.7702,
      "step": 24507
    },
    {
      "epoch": 2.1085778198399723,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006221121411463953,
      "loss": 1.6408,
      "step": 24508
    },
    {
      "epoch": 2.1086638561472943,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006220859190589371,
      "loss": 1.001,
      "step": 24509
    },
    {
      "epoch": 2.108749892454616,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006220596966144013,
      "loss": 1.711,
      "step": 24510
    },
    {
      "epoch": 2.1088359287619376,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006220334738128645,
      "loss": 0.6623,
      "step": 24511
    },
    {
      "epoch": 2.108921965069259,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006220072506544035,
      "loss": 1.3017,
      "step": 24512
    },
    {
      "epoch": 2.109008001376581,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006219810271390951,
      "loss": 0.8695,
      "step": 24513
    },
    {
      "epoch": 2.1090940376839025,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000621954803267016,
      "loss": 1.4485,
      "step": 24514
    },
    {
      "epoch": 2.109180073991224,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006219285790382428,
      "loss": 1.7752,
      "step": 24515
    },
    {
      "epoch": 2.109266110298546,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006219023544528522,
      "loss": 1.1162,
      "step": 24516
    },
    {
      "epoch": 2.109352146605868,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006218761295109209,
      "loss": 0.9856,
      "step": 24517
    },
    {
      "epoch": 2.1094381829131894,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006218499042125255,
      "loss": 1.0619,
      "step": 24518
    },
    {
      "epoch": 2.109524219220511,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006218236785577431,
      "loss": 0.7976,
      "step": 24519
    },
    {
      "epoch": 2.1096102555278327,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006217974525466499,
      "loss": 1.0576,
      "step": 24520
    },
    {
      "epoch": 2.1096962918351543,
      "grad_norm": 1.125,
      "learning_rate": 0.0006217712261793229,
      "loss": 0.9553,
      "step": 24521
    },
    {
      "epoch": 2.109782328142476,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006217449994558388,
      "loss": 1.025,
      "step": 24522
    },
    {
      "epoch": 2.109868364449798,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0006217187723762743,
      "loss": 1.2393,
      "step": 24523
    },
    {
      "epoch": 2.1099544007571196,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006216925449407059,
      "loss": 1.564,
      "step": 24524
    },
    {
      "epoch": 2.110040437064441,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006216663171492105,
      "loss": 0.9039,
      "step": 24525
    },
    {
      "epoch": 2.110126473371763,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006216400890018649,
      "loss": 1.4241,
      "step": 24526
    },
    {
      "epoch": 2.1102125096790845,
      "grad_norm": 1.25,
      "learning_rate": 0.0006216138604987456,
      "loss": 0.9553,
      "step": 24527
    },
    {
      "epoch": 2.110298545986406,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006215876316399295,
      "loss": 1.1091,
      "step": 24528
    },
    {
      "epoch": 2.110384582293728,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006215614024254932,
      "loss": 1.2109,
      "step": 24529
    },
    {
      "epoch": 2.11047061860105,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006215351728555134,
      "loss": 0.7212,
      "step": 24530
    },
    {
      "epoch": 2.1105566549083714,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000621508942930067,
      "loss": 1.3265,
      "step": 24531
    },
    {
      "epoch": 2.110642691215693,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006214827126492305,
      "loss": 0.9209,
      "step": 24532
    },
    {
      "epoch": 2.1107287275230147,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006214564820130805,
      "loss": 0.7446,
      "step": 24533
    },
    {
      "epoch": 2.1108147638303363,
      "grad_norm": 0.75,
      "learning_rate": 0.0006214302510216942,
      "loss": 0.8865,
      "step": 24534
    },
    {
      "epoch": 2.110900800137658,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006214040196751478,
      "loss": 1.0113,
      "step": 24535
    },
    {
      "epoch": 2.11098683644498,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0006213777879735182,
      "loss": 0.6984,
      "step": 24536
    },
    {
      "epoch": 2.1110728727523016,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006213515559168824,
      "loss": 1.0322,
      "step": 24537
    },
    {
      "epoch": 2.1111589090596232,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006213253235053169,
      "loss": 1.6211,
      "step": 24538
    },
    {
      "epoch": 2.111244945366945,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006212990907388983,
      "loss": 1.8587,
      "step": 24539
    },
    {
      "epoch": 2.1113309816742665,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0006212728576177035,
      "loss": 1.4965,
      "step": 24540
    },
    {
      "epoch": 2.111417017981588,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006212466241418092,
      "loss": 0.5281,
      "step": 24541
    },
    {
      "epoch": 2.1115030542889097,
      "grad_norm": 1.25,
      "learning_rate": 0.0006212203903112921,
      "loss": 1.7251,
      "step": 24542
    },
    {
      "epoch": 2.111589090596232,
      "grad_norm": 1.09375,
      "learning_rate": 0.000621194156126229,
      "loss": 0.9046,
      "step": 24543
    },
    {
      "epoch": 2.1116751269035534,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006211679215866963,
      "loss": 0.7314,
      "step": 24544
    },
    {
      "epoch": 2.111761163210875,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006211416866927713,
      "loss": 1.0582,
      "step": 24545
    },
    {
      "epoch": 2.1118471995181967,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006211154514445303,
      "loss": 1.5889,
      "step": 24546
    },
    {
      "epoch": 2.1119332358255183,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006210892158420501,
      "loss": 1.6686,
      "step": 24547
    },
    {
      "epoch": 2.11201927213284,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006210629798854076,
      "loss": 0.9185,
      "step": 24548
    },
    {
      "epoch": 2.112105308440162,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006210367435746794,
      "loss": 1.4779,
      "step": 24549
    },
    {
      "epoch": 2.1121913447474836,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006210105069099424,
      "loss": 1.2226,
      "step": 24550
    },
    {
      "epoch": 2.1122773810548052,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006209842698912728,
      "loss": 1.3716,
      "step": 24551
    },
    {
      "epoch": 2.112363417362127,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006209580325187481,
      "loss": 1.0943,
      "step": 24552
    },
    {
      "epoch": 2.1124494536694485,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006209317947924446,
      "loss": 0.4169,
      "step": 24553
    },
    {
      "epoch": 2.11253548997677,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006209055567124391,
      "loss": 1.0465,
      "step": 24554
    },
    {
      "epoch": 2.1126215262840917,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006208793182788083,
      "loss": 1.4209,
      "step": 24555
    },
    {
      "epoch": 2.112707562591414,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006208530794916291,
      "loss": 1.737,
      "step": 24556
    },
    {
      "epoch": 2.1127935988987354,
      "grad_norm": 1.390625,
      "learning_rate": 0.000620826840350978,
      "loss": 1.2042,
      "step": 24557
    },
    {
      "epoch": 2.112879635206057,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006208006008569322,
      "loss": 1.0775,
      "step": 24558
    },
    {
      "epoch": 2.1129656715133787,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006207743610095678,
      "loss": 1.6363,
      "step": 24559
    },
    {
      "epoch": 2.1130517078207003,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006207481208089621,
      "loss": 2.1841,
      "step": 24560
    },
    {
      "epoch": 2.113137744128022,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006207218802551915,
      "loss": 0.9937,
      "step": 24561
    },
    {
      "epoch": 2.1132237804353435,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006206956393483328,
      "loss": 0.6541,
      "step": 24562
    },
    {
      "epoch": 2.1133098167426656,
      "grad_norm": 1.296875,
      "learning_rate": 0.000620669398088463,
      "loss": 1.3532,
      "step": 24563
    },
    {
      "epoch": 2.1133958530499872,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006206431564756588,
      "loss": 1.5827,
      "step": 24564
    },
    {
      "epoch": 2.113481889357309,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006206169145099966,
      "loss": 1.3599,
      "step": 24565
    },
    {
      "epoch": 2.1135679256646305,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006205906721915534,
      "loss": 0.7554,
      "step": 24566
    },
    {
      "epoch": 2.113653961971952,
      "grad_norm": 7.40625,
      "learning_rate": 0.000620564429520406,
      "loss": 1.3761,
      "step": 24567
    },
    {
      "epoch": 2.1137399982792737,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006205381864966312,
      "loss": 1.0677,
      "step": 24568
    },
    {
      "epoch": 2.1138260345865953,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006205119431203055,
      "loss": 1.2331,
      "step": 24569
    },
    {
      "epoch": 2.1139120708939174,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006204856993915058,
      "loss": 0.8253,
      "step": 24570
    },
    {
      "epoch": 2.113998107201239,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006204594553103087,
      "loss": 1.4039,
      "step": 24571
    },
    {
      "epoch": 2.1140841435085607,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006204332108767915,
      "loss": 1.2374,
      "step": 24572
    },
    {
      "epoch": 2.1141701798158823,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006204069660910304,
      "loss": 0.8027,
      "step": 24573
    },
    {
      "epoch": 2.114256216123204,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006203807209531023,
      "loss": 1.1342,
      "step": 24574
    },
    {
      "epoch": 2.1143422524305255,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006203544754630841,
      "loss": 1.8737,
      "step": 24575
    },
    {
      "epoch": 2.114428288737847,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006203282296210524,
      "loss": 1.4094,
      "step": 24576
    },
    {
      "epoch": 2.1145143250451692,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000620301983427084,
      "loss": 1.1819,
      "step": 24577
    },
    {
      "epoch": 2.114600361352491,
      "grad_norm": 1.75,
      "learning_rate": 0.0006202757368812558,
      "loss": 1.3267,
      "step": 24578
    },
    {
      "epoch": 2.1146863976598125,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006202494899836444,
      "loss": 1.1814,
      "step": 24579
    },
    {
      "epoch": 2.114772433967134,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0006202232427343266,
      "loss": 1.1975,
      "step": 24580
    },
    {
      "epoch": 2.1148584702744557,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006201969951333793,
      "loss": 1.4344,
      "step": 24581
    },
    {
      "epoch": 2.1149445065817774,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006201707471808791,
      "loss": 1.7295,
      "step": 24582
    },
    {
      "epoch": 2.115030542889099,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0006201444988769028,
      "loss": 1.5845,
      "step": 24583
    },
    {
      "epoch": 2.115116579196421,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006201182502215272,
      "loss": 1.3761,
      "step": 24584
    },
    {
      "epoch": 2.1152026155037427,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006200920012148292,
      "loss": 0.945,
      "step": 24585
    },
    {
      "epoch": 2.1152886518110643,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006200657518568853,
      "loss": 1.2862,
      "step": 24586
    },
    {
      "epoch": 2.115374688118386,
      "grad_norm": 2.078125,
      "learning_rate": 0.0006200395021477727,
      "loss": 1.7178,
      "step": 24587
    },
    {
      "epoch": 2.1154607244257075,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006200132520875675,
      "loss": 1.4793,
      "step": 24588
    },
    {
      "epoch": 2.115546760733029,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006199870016763471,
      "loss": 0.7995,
      "step": 24589
    },
    {
      "epoch": 2.1156327970403512,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006199607509141881,
      "loss": 0.9039,
      "step": 24590
    },
    {
      "epoch": 2.115718833347673,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006199344998011671,
      "loss": 1.1573,
      "step": 24591
    },
    {
      "epoch": 2.1158048696549945,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000619908248337361,
      "loss": 1.2617,
      "step": 24592
    },
    {
      "epoch": 2.115890905962316,
      "grad_norm": 1.625,
      "learning_rate": 0.0006198819965228467,
      "loss": 1.9036,
      "step": 24593
    },
    {
      "epoch": 2.1159769422696377,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006198557443577009,
      "loss": 1.3287,
      "step": 24594
    },
    {
      "epoch": 2.1160629785769594,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006198294918420003,
      "loss": 1.568,
      "step": 24595
    },
    {
      "epoch": 2.116149014884281,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006198032389758217,
      "loss": 1.6462,
      "step": 24596
    },
    {
      "epoch": 2.116235051191603,
      "grad_norm": 1.296875,
      "learning_rate": 0.000619776985759242,
      "loss": 1.1183,
      "step": 24597
    },
    {
      "epoch": 2.1163210874989247,
      "grad_norm": 1.25,
      "learning_rate": 0.000619750732192338,
      "loss": 1.4953,
      "step": 24598
    },
    {
      "epoch": 2.1164071238062463,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006197244782751863,
      "loss": 0.6496,
      "step": 24599
    },
    {
      "epoch": 2.116493160113568,
      "grad_norm": 1.0,
      "learning_rate": 0.0006196982240078637,
      "loss": 0.8475,
      "step": 24600
    },
    {
      "epoch": 2.1165791964208895,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006196719693904473,
      "loss": 1.0984,
      "step": 24601
    },
    {
      "epoch": 2.116665232728211,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006196457144230134,
      "loss": 1.4632,
      "step": 24602
    },
    {
      "epoch": 2.116751269035533,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006196194591056392,
      "loss": 1.9648,
      "step": 24603
    },
    {
      "epoch": 2.116837305342855,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006195932034384013,
      "loss": 1.6791,
      "step": 24604
    },
    {
      "epoch": 2.1169233416501765,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006195669474213767,
      "loss": 0.6296,
      "step": 24605
    },
    {
      "epoch": 2.117009377957498,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006195406910546418,
      "loss": 0.7276,
      "step": 24606
    },
    {
      "epoch": 2.1170954142648197,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006195144343382737,
      "loss": 1.1697,
      "step": 24607
    },
    {
      "epoch": 2.1171814505721414,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006194881772723491,
      "loss": 1.5526,
      "step": 24608
    },
    {
      "epoch": 2.117267486879463,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006194619198569449,
      "loss": 0.9386,
      "step": 24609
    },
    {
      "epoch": 2.117353523186785,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006194356620921377,
      "loss": 0.8871,
      "step": 24610
    },
    {
      "epoch": 2.1174395594941067,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006194094039780045,
      "loss": 0.9626,
      "step": 24611
    },
    {
      "epoch": 2.1175255958014283,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006193831455146221,
      "loss": 1.9211,
      "step": 24612
    },
    {
      "epoch": 2.11761163210875,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006193568867020672,
      "loss": 1.0887,
      "step": 24613
    },
    {
      "epoch": 2.1176976684160715,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006193306275404166,
      "loss": 1.5033,
      "step": 24614
    },
    {
      "epoch": 2.117783704723393,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000619304368029747,
      "loss": 0.8021,
      "step": 24615
    },
    {
      "epoch": 2.117869741030715,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006192781081701354,
      "loss": 1.4389,
      "step": 24616
    },
    {
      "epoch": 2.117955777338037,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006192518479616586,
      "loss": 1.2698,
      "step": 24617
    },
    {
      "epoch": 2.1180418136453585,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006192255874043932,
      "loss": 1.4715,
      "step": 24618
    },
    {
      "epoch": 2.11812784995268,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006191993264984163,
      "loss": 1.4151,
      "step": 24619
    },
    {
      "epoch": 2.1182138862600017,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006191730652438045,
      "loss": 0.7066,
      "step": 24620
    },
    {
      "epoch": 2.1182999225673234,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006191468036406346,
      "loss": 1.0613,
      "step": 24621
    },
    {
      "epoch": 2.118385958874645,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006191205416889835,
      "loss": 0.7449,
      "step": 24622
    },
    {
      "epoch": 2.1184719951819666,
      "grad_norm": 1.046875,
      "learning_rate": 0.000619094279388928,
      "loss": 0.9037,
      "step": 24623
    },
    {
      "epoch": 2.1185580314892887,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006190680167405449,
      "loss": 1.0254,
      "step": 24624
    },
    {
      "epoch": 2.1186440677966103,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006190417537439109,
      "loss": 0.7716,
      "step": 24625
    },
    {
      "epoch": 2.118730104103932,
      "grad_norm": 1.3125,
      "learning_rate": 0.000619015490399103,
      "loss": 1.8273,
      "step": 24626
    },
    {
      "epoch": 2.1188161404112535,
      "grad_norm": 1.0625,
      "learning_rate": 0.000618989226706198,
      "loss": 1.1323,
      "step": 24627
    },
    {
      "epoch": 2.118902176718575,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006189629626652726,
      "loss": 1.2104,
      "step": 24628
    },
    {
      "epoch": 2.118988213025897,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006189366982764038,
      "loss": 1.7765,
      "step": 24629
    },
    {
      "epoch": 2.1190742493332184,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000618910433539668,
      "loss": 1.4083,
      "step": 24630
    },
    {
      "epoch": 2.1191602856405405,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006188841684551425,
      "loss": 1.4423,
      "step": 24631
    },
    {
      "epoch": 2.119246321947862,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000618857903022904,
      "loss": 1.0319,
      "step": 24632
    },
    {
      "epoch": 2.1193323582551837,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006188316372430291,
      "loss": 1.8465,
      "step": 24633
    },
    {
      "epoch": 2.1194183945625054,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006188053711155948,
      "loss": 1.0261,
      "step": 24634
    },
    {
      "epoch": 2.119504430869827,
      "grad_norm": 1.484375,
      "learning_rate": 0.000618779104640678,
      "loss": 1.2018,
      "step": 24635
    },
    {
      "epoch": 2.1195904671771486,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006187528378183554,
      "loss": 1.4189,
      "step": 24636
    },
    {
      "epoch": 2.1196765034844702,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006187265706487036,
      "loss": 1.226,
      "step": 24637
    },
    {
      "epoch": 2.1197625397917923,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006187003031318,
      "loss": 0.9954,
      "step": 24638
    },
    {
      "epoch": 2.119848576099114,
      "grad_norm": 1.0,
      "learning_rate": 0.000618674035267721,
      "loss": 0.6442,
      "step": 24639
    },
    {
      "epoch": 2.1199346124064355,
      "grad_norm": 1.765625,
      "learning_rate": 0.0006186477670565434,
      "loss": 1.4637,
      "step": 24640
    },
    {
      "epoch": 2.120020648713757,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006186214984983442,
      "loss": 1.9288,
      "step": 24641
    },
    {
      "epoch": 2.120106685021079,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006185952295932003,
      "loss": 1.6429,
      "step": 24642
    },
    {
      "epoch": 2.1201927213284004,
      "grad_norm": 1.125,
      "learning_rate": 0.0006185689603411885,
      "loss": 0.8633,
      "step": 24643
    },
    {
      "epoch": 2.1202787576357225,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006185426907423852,
      "loss": 0.6804,
      "step": 24644
    },
    {
      "epoch": 2.120364793943044,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006185164207968678,
      "loss": 1.1166,
      "step": 24645
    },
    {
      "epoch": 2.1204508302503657,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006184901505047129,
      "loss": 0.9919,
      "step": 24646
    },
    {
      "epoch": 2.1205368665576874,
      "grad_norm": 1.6875,
      "learning_rate": 0.0006184638798659974,
      "loss": 1.1001,
      "step": 24647
    },
    {
      "epoch": 2.120622902865009,
      "grad_norm": 1.109375,
      "learning_rate": 0.000618437608880798,
      "loss": 0.9434,
      "step": 24648
    },
    {
      "epoch": 2.1207089391723306,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006184113375491917,
      "loss": 1.3717,
      "step": 24649
    },
    {
      "epoch": 2.1207949754796522,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006183850658712553,
      "loss": 1.6366,
      "step": 24650
    },
    {
      "epoch": 2.1208810117869743,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006183587938470656,
      "loss": 1.2163,
      "step": 24651
    },
    {
      "epoch": 2.120967048094296,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006183325214766994,
      "loss": 1.3744,
      "step": 24652
    },
    {
      "epoch": 2.1210530844016176,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006183062487602335,
      "loss": 1.7135,
      "step": 24653
    },
    {
      "epoch": 2.121139120708939,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000618279975697745,
      "loss": 1.3429,
      "step": 24654
    },
    {
      "epoch": 2.121225157016261,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006182537022893106,
      "loss": 0.8951,
      "step": 24655
    },
    {
      "epoch": 2.1213111933235824,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006182274285350069,
      "loss": 0.9582,
      "step": 24656
    },
    {
      "epoch": 2.121397229630904,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006182011544349112,
      "loss": 1.0264,
      "step": 24657
    },
    {
      "epoch": 2.121483265938226,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006181748799891,
      "loss": 0.965,
      "step": 24658
    },
    {
      "epoch": 2.1215693022455477,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006181486051976501,
      "loss": 1.179,
      "step": 24659
    },
    {
      "epoch": 2.1216553385528694,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006181223300606388,
      "loss": 1.8163,
      "step": 24660
    },
    {
      "epoch": 2.121741374860191,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0006180960545781425,
      "loss": 1.3178,
      "step": 24661
    },
    {
      "epoch": 2.1218274111675126,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006180697787502383,
      "loss": 1.4546,
      "step": 24662
    },
    {
      "epoch": 2.1219134474748342,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006180435025770028,
      "loss": 1.8312,
      "step": 24663
    },
    {
      "epoch": 2.121999483782156,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006180172260585131,
      "loss": 0.6741,
      "step": 24664
    },
    {
      "epoch": 2.122085520089478,
      "grad_norm": 1.09375,
      "learning_rate": 0.000617990949194846,
      "loss": 0.8359,
      "step": 24665
    },
    {
      "epoch": 2.1221715563967996,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006179646719860784,
      "loss": 0.9459,
      "step": 24666
    },
    {
      "epoch": 2.122257592704121,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006179383944322868,
      "loss": 1.6555,
      "step": 24667
    },
    {
      "epoch": 2.122343629011443,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006179121165335486,
      "loss": 0.4765,
      "step": 24668
    },
    {
      "epoch": 2.1224296653187644,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006178858382899404,
      "loss": 0.7939,
      "step": 24669
    },
    {
      "epoch": 2.122515701626086,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006178595597015389,
      "loss": 0.6494,
      "step": 24670
    },
    {
      "epoch": 2.122601737933408,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000617833280768421,
      "loss": 1.3404,
      "step": 24671
    },
    {
      "epoch": 2.1226877742407297,
      "grad_norm": 1.125,
      "learning_rate": 0.000617807001490664,
      "loss": 1.4159,
      "step": 24672
    },
    {
      "epoch": 2.1227738105480514,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006177807218683442,
      "loss": 1.8303,
      "step": 24673
    },
    {
      "epoch": 2.122859846855373,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006177544419015388,
      "loss": 1.4799,
      "step": 24674
    },
    {
      "epoch": 2.1229458831626946,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006177281615903246,
      "loss": 1.5741,
      "step": 24675
    },
    {
      "epoch": 2.1230319194700162,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006177018809347783,
      "loss": 0.8462,
      "step": 24676
    },
    {
      "epoch": 2.123117955777338,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006176755999349771,
      "loss": 1.1241,
      "step": 24677
    },
    {
      "epoch": 2.12320399208466,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006176493185909973,
      "loss": 0.5429,
      "step": 24678
    },
    {
      "epoch": 2.1232900283919816,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006176230369029164,
      "loss": 0.95,
      "step": 24679
    },
    {
      "epoch": 2.123376064699303,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006175967548708109,
      "loss": 0.9782,
      "step": 24680
    },
    {
      "epoch": 2.123462101006625,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006175704724947577,
      "loss": 1.2515,
      "step": 24681
    },
    {
      "epoch": 2.1235481373139464,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006175441897748337,
      "loss": 0.9121,
      "step": 24682
    },
    {
      "epoch": 2.123634173621268,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000617517906711116,
      "loss": 1.6795,
      "step": 24683
    },
    {
      "epoch": 2.1237202099285897,
      "grad_norm": 1.65625,
      "learning_rate": 0.0006174916233036812,
      "loss": 1.0747,
      "step": 24684
    },
    {
      "epoch": 2.1238062462359117,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006174653395526063,
      "loss": 0.8919,
      "step": 24685
    },
    {
      "epoch": 2.1238922825432334,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000617439055457968,
      "loss": 1.5717,
      "step": 24686
    },
    {
      "epoch": 2.123978318850555,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006174127710198434,
      "loss": 1.032,
      "step": 24687
    },
    {
      "epoch": 2.1240643551578766,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006173864862383093,
      "loss": 1.7808,
      "step": 24688
    },
    {
      "epoch": 2.1241503914651982,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0006173602011134425,
      "loss": 1.1044,
      "step": 24689
    },
    {
      "epoch": 2.12423642777252,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00061733391564532,
      "loss": 0.9348,
      "step": 24690
    },
    {
      "epoch": 2.1243224640798415,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006173076298340186,
      "loss": 1.7736,
      "step": 24691
    },
    {
      "epoch": 2.1244085003871636,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006172813436796151,
      "loss": 1.2441,
      "step": 24692
    },
    {
      "epoch": 2.124494536694485,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006172550571821864,
      "loss": 1.6233,
      "step": 24693
    },
    {
      "epoch": 2.124580573001807,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006172287703418098,
      "loss": 1.0482,
      "step": 24694
    },
    {
      "epoch": 2.1246666093091284,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006172024831585616,
      "loss": 1.2885,
      "step": 24695
    },
    {
      "epoch": 2.12475264561645,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006171761956325191,
      "loss": 1.3846,
      "step": 24696
    },
    {
      "epoch": 2.1248386819237717,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006171499077637588,
      "loss": 0.8101,
      "step": 24697
    },
    {
      "epoch": 2.1249247182310933,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000617123619552358,
      "loss": 1.4314,
      "step": 24698
    },
    {
      "epoch": 2.1250107545384154,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0006170973309983932,
      "loss": 0.6537,
      "step": 24699
    },
    {
      "epoch": 2.125096790845737,
      "grad_norm": 1.0,
      "learning_rate": 0.0006170710421019416,
      "loss": 1.8932,
      "step": 24700
    },
    {
      "epoch": 2.1251828271530586,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006170447528630798,
      "loss": 1.5533,
      "step": 24701
    },
    {
      "epoch": 2.1252688634603802,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006170184632818852,
      "loss": 1.203,
      "step": 24702
    },
    {
      "epoch": 2.125354899767702,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006169921733584341,
      "loss": 1.0204,
      "step": 24703
    },
    {
      "epoch": 2.1254409360750235,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0006169658830928037,
      "loss": 0.6549,
      "step": 24704
    },
    {
      "epoch": 2.125526972382345,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006169395924850708,
      "loss": 1.0801,
      "step": 24705
    },
    {
      "epoch": 2.125613008689667,
      "grad_norm": 1.25,
      "learning_rate": 0.0006169133015353124,
      "loss": 1.1658,
      "step": 24706
    },
    {
      "epoch": 2.125699044996989,
      "grad_norm": 0.984375,
      "learning_rate": 0.0006168870102436054,
      "loss": 1.0582,
      "step": 24707
    },
    {
      "epoch": 2.1257850813043104,
      "grad_norm": 1.0,
      "learning_rate": 0.0006168607186100264,
      "loss": 1.3177,
      "step": 24708
    },
    {
      "epoch": 2.125871117611632,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006168344266346529,
      "loss": 0.8811,
      "step": 24709
    },
    {
      "epoch": 2.1259571539189537,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006168081343175611,
      "loss": 0.8915,
      "step": 24710
    },
    {
      "epoch": 2.1260431902262753,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006167818416588284,
      "loss": 1.0184,
      "step": 24711
    },
    {
      "epoch": 2.1261292265335974,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006167555486585313,
      "loss": 1.1058,
      "step": 24712
    },
    {
      "epoch": 2.126215262840919,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0006167292553167471,
      "loss": 1.9211,
      "step": 24713
    },
    {
      "epoch": 2.1263012991482406,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006167029616335526,
      "loss": 0.8025,
      "step": 24714
    },
    {
      "epoch": 2.1263873354555622,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006166766676090245,
      "loss": 0.7357,
      "step": 24715
    },
    {
      "epoch": 2.126473371762884,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006166503732432399,
      "loss": 0.9776,
      "step": 24716
    },
    {
      "epoch": 2.1265594080702055,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006166240785362757,
      "loss": 0.8454,
      "step": 24717
    },
    {
      "epoch": 2.126645444377527,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006165977834882088,
      "loss": 1.329,
      "step": 24718
    },
    {
      "epoch": 2.126731480684849,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006165714880991158,
      "loss": 0.9185,
      "step": 24719
    },
    {
      "epoch": 2.126817516992171,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006165451923690739,
      "loss": 1.3941,
      "step": 24720
    },
    {
      "epoch": 2.1269035532994924,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006165188962981602,
      "loss": 1.7264,
      "step": 24721
    },
    {
      "epoch": 2.126989589606814,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006164925998864512,
      "loss": 1.6853,
      "step": 24722
    },
    {
      "epoch": 2.1270756259141357,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000616466303134024,
      "loss": 0.9192,
      "step": 24723
    },
    {
      "epoch": 2.1271616622214573,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006164400060409557,
      "loss": 0.9148,
      "step": 24724
    },
    {
      "epoch": 2.1272476985287794,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006164137086073229,
      "loss": 1.6908,
      "step": 24725
    },
    {
      "epoch": 2.127333734836101,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006163874108332028,
      "loss": 1.1248,
      "step": 24726
    },
    {
      "epoch": 2.1274197711434226,
      "grad_norm": 1.25,
      "learning_rate": 0.0006163611127186719,
      "loss": 1.0052,
      "step": 24727
    },
    {
      "epoch": 2.1275058074507442,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006163348142638076,
      "loss": 1.2711,
      "step": 24728
    },
    {
      "epoch": 2.127591843758066,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006163085154686865,
      "loss": 0.5296,
      "step": 24729
    },
    {
      "epoch": 2.1276778800653875,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006162822163333856,
      "loss": 0.7893,
      "step": 24730
    },
    {
      "epoch": 2.127763916372709,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006162559168579819,
      "loss": 1.1464,
      "step": 24731
    },
    {
      "epoch": 2.127849952680031,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006162296170425522,
      "loss": 1.0785,
      "step": 24732
    },
    {
      "epoch": 2.127935988987353,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006162033168871735,
      "loss": 0.9393,
      "step": 24733
    },
    {
      "epoch": 2.1280220252946744,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006161770163919226,
      "loss": 0.9426,
      "step": 24734
    },
    {
      "epoch": 2.128108061601996,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006161507155568767,
      "loss": 1.0543,
      "step": 24735
    },
    {
      "epoch": 2.1281940979093177,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006161244143821124,
      "loss": 1.3996,
      "step": 24736
    },
    {
      "epoch": 2.1282801342166393,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006160981128677068,
      "loss": 1.797,
      "step": 24737
    },
    {
      "epoch": 2.128366170523961,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006160718110137367,
      "loss": 1.379,
      "step": 24738
    },
    {
      "epoch": 2.128452206831283,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006160455088202792,
      "loss": 1.6789,
      "step": 24739
    },
    {
      "epoch": 2.1285382431386046,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006160192062874113,
      "loss": 0.955,
      "step": 24740
    },
    {
      "epoch": 2.1286242794459262,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006159929034152097,
      "loss": 0.6945,
      "step": 24741
    },
    {
      "epoch": 2.128710315753248,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006159666002037514,
      "loss": 1.4646,
      "step": 24742
    },
    {
      "epoch": 2.1287963520605695,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006159402966531134,
      "loss": 0.7124,
      "step": 24743
    },
    {
      "epoch": 2.128882388367891,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006159139927633724,
      "loss": 1.678,
      "step": 24744
    },
    {
      "epoch": 2.1289684246752127,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006158876885346056,
      "loss": 0.7354,
      "step": 24745
    },
    {
      "epoch": 2.129054460982535,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006158613839668901,
      "loss": 1.2492,
      "step": 24746
    },
    {
      "epoch": 2.1291404972898564,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006158350790603024,
      "loss": 1.9894,
      "step": 24747
    },
    {
      "epoch": 2.129226533597178,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006158087738149196,
      "loss": 0.7302,
      "step": 24748
    },
    {
      "epoch": 2.1293125699044997,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006157824682308185,
      "loss": 1.597,
      "step": 24749
    },
    {
      "epoch": 2.1293986062118213,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006157561623080766,
      "loss": 2.0179,
      "step": 24750
    },
    {
      "epoch": 2.129484642519143,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006157298560467701,
      "loss": 0.8647,
      "step": 24751
    },
    {
      "epoch": 2.1295706788264646,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006157035494469764,
      "loss": 1.0361,
      "step": 24752
    },
    {
      "epoch": 2.1296567151337866,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006156772425087722,
      "loss": 1.257,
      "step": 24753
    },
    {
      "epoch": 2.1297427514411083,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006156509352322348,
      "loss": 0.8502,
      "step": 24754
    },
    {
      "epoch": 2.12982878774843,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0006156246276174407,
      "loss": 1.5704,
      "step": 24755
    },
    {
      "epoch": 2.1299148240557515,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006155983196644671,
      "loss": 2.1232,
      "step": 24756
    },
    {
      "epoch": 2.130000860363073,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006155720113733909,
      "loss": 0.662,
      "step": 24757
    },
    {
      "epoch": 2.1300868966703947,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006155457027442891,
      "loss": 0.5561,
      "step": 24758
    },
    {
      "epoch": 2.1301729329777164,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006155193937772385,
      "loss": 1.5114,
      "step": 24759
    },
    {
      "epoch": 2.1302589692850384,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006154930844723162,
      "loss": 1.6231,
      "step": 24760
    },
    {
      "epoch": 2.13034500559236,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006154667748295991,
      "loss": 1.6808,
      "step": 24761
    },
    {
      "epoch": 2.1304310418996817,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006154404648491641,
      "loss": 0.5522,
      "step": 24762
    },
    {
      "epoch": 2.1305170782070033,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006154141545310883,
      "loss": 1.4959,
      "step": 24763
    },
    {
      "epoch": 2.130603114514325,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006153878438754485,
      "loss": 1.1301,
      "step": 24764
    },
    {
      "epoch": 2.1306891508216466,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006153615328823216,
      "loss": 1.1486,
      "step": 24765
    },
    {
      "epoch": 2.1307751871289686,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006153352215517848,
      "loss": 1.0861,
      "step": 24766
    },
    {
      "epoch": 2.1308612234362903,
      "grad_norm": 0.765625,
      "learning_rate": 0.0006153089098839148,
      "loss": 0.9465,
      "step": 24767
    },
    {
      "epoch": 2.130947259743612,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006152825978787887,
      "loss": 1.0513,
      "step": 24768
    },
    {
      "epoch": 2.1310332960509335,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006152562855364834,
      "loss": 1.0854,
      "step": 24769
    },
    {
      "epoch": 2.131119332358255,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006152299728570759,
      "loss": 1.4926,
      "step": 24770
    },
    {
      "epoch": 2.1312053686655767,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006152036598406432,
      "loss": 1.5442,
      "step": 24771
    },
    {
      "epoch": 2.1312914049728984,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006151773464872621,
      "loss": 1.5555,
      "step": 24772
    },
    {
      "epoch": 2.1313774412802204,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006151510327970098,
      "loss": 0.9605,
      "step": 24773
    },
    {
      "epoch": 2.131463477587542,
      "grad_norm": 1.078125,
      "learning_rate": 0.000615124718769963,
      "loss": 0.7557,
      "step": 24774
    },
    {
      "epoch": 2.1315495138948637,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006150984044061986,
      "loss": 1.3048,
      "step": 24775
    },
    {
      "epoch": 2.1316355502021853,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006150720897057939,
      "loss": 1.7633,
      "step": 24776
    },
    {
      "epoch": 2.131721586509507,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000615045774668826,
      "loss": 1.1654,
      "step": 24777
    },
    {
      "epoch": 2.1318076228168286,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006150194592953714,
      "loss": 1.0232,
      "step": 24778
    },
    {
      "epoch": 2.13189365912415,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006149931435855071,
      "loss": 1.2235,
      "step": 24779
    },
    {
      "epoch": 2.1319796954314723,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006149668275393103,
      "loss": 1.7454,
      "step": 24780
    },
    {
      "epoch": 2.132065731738794,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000614940511156858,
      "loss": 1.0513,
      "step": 24781
    },
    {
      "epoch": 2.1321517680461155,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006149141944382271,
      "loss": 0.996,
      "step": 24782
    },
    {
      "epoch": 2.132237804353437,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0006148878773834943,
      "loss": 0.7982,
      "step": 24783
    },
    {
      "epoch": 2.1323238406607588,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006148615599927369,
      "loss": 0.9449,
      "step": 24784
    },
    {
      "epoch": 2.1324098769680804,
      "grad_norm": 1.109375,
      "learning_rate": 0.000614835242266032,
      "loss": 1.4502,
      "step": 24785
    },
    {
      "epoch": 2.1324959132754024,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006148089242034561,
      "loss": 0.7913,
      "step": 24786
    },
    {
      "epoch": 2.132581949582724,
      "grad_norm": 1.25,
      "learning_rate": 0.0006147826058050863,
      "loss": 1.2005,
      "step": 24787
    },
    {
      "epoch": 2.1326679858900457,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006147562870709999,
      "loss": 1.3756,
      "step": 24788
    },
    {
      "epoch": 2.1327540221973673,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006147299680012738,
      "loss": 1.5172,
      "step": 24789
    },
    {
      "epoch": 2.132840058504689,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006147036485959846,
      "loss": 0.7868,
      "step": 24790
    },
    {
      "epoch": 2.1329260948120106,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006146773288552097,
      "loss": 1.3889,
      "step": 24791
    },
    {
      "epoch": 2.133012131119332,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006146510087790259,
      "loss": 1.9204,
      "step": 24792
    },
    {
      "epoch": 2.1330981674266543,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00061462468836751,
      "loss": 0.9852,
      "step": 24793
    },
    {
      "epoch": 2.133184203733976,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006145983676207394,
      "loss": 1.0944,
      "step": 24794
    },
    {
      "epoch": 2.1332702400412975,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006145720465387908,
      "loss": 1.2457,
      "step": 24795
    },
    {
      "epoch": 2.133356276348619,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006145457251217414,
      "loss": 0.9537,
      "step": 24796
    },
    {
      "epoch": 2.1334423126559408,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006145194033696679,
      "loss": 0.9313,
      "step": 24797
    },
    {
      "epoch": 2.1335283489632624,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006144930812826472,
      "loss": 1.0347,
      "step": 24798
    },
    {
      "epoch": 2.133614385270584,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006144667588607567,
      "loss": 0.9464,
      "step": 24799
    },
    {
      "epoch": 2.133700421577906,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006144404361040733,
      "loss": 1.6985,
      "step": 24800
    },
    {
      "epoch": 2.1337864578852277,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006144141130126737,
      "loss": 1.6996,
      "step": 24801
    },
    {
      "epoch": 2.1338724941925493,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0006143877895866352,
      "loss": 1.7825,
      "step": 24802
    },
    {
      "epoch": 2.133958530499871,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006143614658260347,
      "loss": 0.6732,
      "step": 24803
    },
    {
      "epoch": 2.1340445668071926,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006143351417309491,
      "loss": 1.7007,
      "step": 24804
    },
    {
      "epoch": 2.134130603114514,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006143088173014554,
      "loss": 1.1735,
      "step": 24805
    },
    {
      "epoch": 2.134216639421836,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006142824925376306,
      "loss": 0.9448,
      "step": 24806
    },
    {
      "epoch": 2.134302675729158,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006142561674395519,
      "loss": 1.2237,
      "step": 24807
    },
    {
      "epoch": 2.1343887120364795,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000614229842007296,
      "loss": 1.3421,
      "step": 24808
    },
    {
      "epoch": 2.134474748343801,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006142035162409401,
      "loss": 0.6793,
      "step": 24809
    },
    {
      "epoch": 2.1345607846511228,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006141771901405611,
      "loss": 1.4978,
      "step": 24810
    },
    {
      "epoch": 2.1346468209584444,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006141508637062359,
      "loss": 0.4692,
      "step": 24811
    },
    {
      "epoch": 2.134732857265766,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006141245369380417,
      "loss": 1.2139,
      "step": 24812
    },
    {
      "epoch": 2.1348188935730876,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006140982098360554,
      "loss": 0.7389,
      "step": 24813
    },
    {
      "epoch": 2.1349049298804097,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006140718824003542,
      "loss": 1.3471,
      "step": 24814
    },
    {
      "epoch": 2.1349909661877313,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006140455546310149,
      "loss": 1.1419,
      "step": 24815
    },
    {
      "epoch": 2.135077002495053,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006140192265281143,
      "loss": 0.8027,
      "step": 24816
    },
    {
      "epoch": 2.1351630388023746,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006139928980917297,
      "loss": 1.5861,
      "step": 24817
    },
    {
      "epoch": 2.135249075109696,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006139665693219383,
      "loss": 1.3671,
      "step": 24818
    },
    {
      "epoch": 2.135335111417018,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006139402402188166,
      "loss": 0.8718,
      "step": 24819
    },
    {
      "epoch": 2.1354211477243394,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000613913910782442,
      "loss": 1.3144,
      "step": 24820
    },
    {
      "epoch": 2.1355071840316615,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006138875810128914,
      "loss": 1.1196,
      "step": 24821
    },
    {
      "epoch": 2.135593220338983,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006138612509102416,
      "loss": 0.6103,
      "step": 24822
    },
    {
      "epoch": 2.1356792566463048,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006138349204745699,
      "loss": 1.6914,
      "step": 24823
    },
    {
      "epoch": 2.1357652929536264,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006138085897059531,
      "loss": 0.9228,
      "step": 24824
    },
    {
      "epoch": 2.135851329260948,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006137822586044684,
      "loss": 1.6888,
      "step": 24825
    },
    {
      "epoch": 2.1359373655682696,
      "grad_norm": 1.125,
      "learning_rate": 0.0006137559271701927,
      "loss": 1.3376,
      "step": 24826
    },
    {
      "epoch": 2.1360234018755917,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006137295954032029,
      "loss": 0.9962,
      "step": 24827
    },
    {
      "epoch": 2.1361094381829133,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006137032633035764,
      "loss": 1.7606,
      "step": 24828
    },
    {
      "epoch": 2.136195474490235,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006136769308713899,
      "loss": 1.2617,
      "step": 24829
    },
    {
      "epoch": 2.1362815107975566,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006136505981067204,
      "loss": 1.1486,
      "step": 24830
    },
    {
      "epoch": 2.136367547104878,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006136242650096451,
      "loss": 1.3883,
      "step": 24831
    },
    {
      "epoch": 2.1364535834122,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006135979315802407,
      "loss": 1.245,
      "step": 24832
    },
    {
      "epoch": 2.1365396197195214,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006135715978185847,
      "loss": 1.7095,
      "step": 24833
    },
    {
      "epoch": 2.1366256560268435,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006135452637247538,
      "loss": 1.4411,
      "step": 24834
    },
    {
      "epoch": 2.136711692334165,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000613518929298825,
      "loss": 0.9554,
      "step": 24835
    },
    {
      "epoch": 2.1367977286414868,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006134925945408755,
      "loss": 1.0726,
      "step": 24836
    },
    {
      "epoch": 2.1368837649488084,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006134662594509822,
      "loss": 1.1376,
      "step": 24837
    },
    {
      "epoch": 2.13696980125613,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006134399240292222,
      "loss": 1.2788,
      "step": 24838
    },
    {
      "epoch": 2.1370558375634516,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006134135882756724,
      "loss": 0.9351,
      "step": 24839
    },
    {
      "epoch": 2.1371418738707737,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00061338725219041,
      "loss": 1.191,
      "step": 24840
    },
    {
      "epoch": 2.1372279101780953,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006133609157735119,
      "loss": 1.1093,
      "step": 24841
    },
    {
      "epoch": 2.137313946485417,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006133345790250552,
      "loss": 1.3405,
      "step": 24842
    },
    {
      "epoch": 2.1373999827927386,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006133082419451168,
      "loss": 0.8941,
      "step": 24843
    },
    {
      "epoch": 2.13748601910006,
      "grad_norm": 1.1875,
      "learning_rate": 0.000613281904533774,
      "loss": 0.6447,
      "step": 24844
    },
    {
      "epoch": 2.137572055407382,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006132555667911036,
      "loss": 1.2042,
      "step": 24845
    },
    {
      "epoch": 2.1376580917147034,
      "grad_norm": 1.125,
      "learning_rate": 0.0006132292287171825,
      "loss": 1.0143,
      "step": 24846
    },
    {
      "epoch": 2.1377441280220255,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006132028903120882,
      "loss": 0.4757,
      "step": 24847
    },
    {
      "epoch": 2.137830164329347,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006131765515758974,
      "loss": 0.7387,
      "step": 24848
    },
    {
      "epoch": 2.1379162006366688,
      "grad_norm": 1.546875,
      "learning_rate": 0.0006131502125086871,
      "loss": 1.5576,
      "step": 24849
    },
    {
      "epoch": 2.1380022369439904,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006131238731105344,
      "loss": 1.5458,
      "step": 24850
    },
    {
      "epoch": 2.138088273251312,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006130975333815164,
      "loss": 1.6644,
      "step": 24851
    },
    {
      "epoch": 2.1381743095586336,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006130711933217103,
      "loss": 0.9051,
      "step": 24852
    },
    {
      "epoch": 2.1382603458659553,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006130448529311929,
      "loss": 0.7914,
      "step": 24853
    },
    {
      "epoch": 2.1383463821732773,
      "grad_norm": 1.578125,
      "learning_rate": 0.000613018512210041,
      "loss": 1.54,
      "step": 24854
    },
    {
      "epoch": 2.138432418480599,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006129921711583322,
      "loss": 1.1064,
      "step": 24855
    },
    {
      "epoch": 2.1385184547879206,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006129658297761432,
      "loss": 1.1336,
      "step": 24856
    },
    {
      "epoch": 2.138604491095242,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006129394880635511,
      "loss": 0.9238,
      "step": 24857
    },
    {
      "epoch": 2.138690527402564,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006129131460206331,
      "loss": 0.5595,
      "step": 24858
    },
    {
      "epoch": 2.1387765637098854,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006128868036474659,
      "loss": 1.168,
      "step": 24859
    },
    {
      "epoch": 2.138862600017207,
      "grad_norm": 1.3125,
      "learning_rate": 0.000612860460944127,
      "loss": 0.7816,
      "step": 24860
    },
    {
      "epoch": 2.138948636324529,
      "grad_norm": 1.140625,
      "learning_rate": 0.000612834117910693,
      "loss": 0.6395,
      "step": 24861
    },
    {
      "epoch": 2.1390346726318508,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0006128077745472413,
      "loss": 0.7901,
      "step": 24862
    },
    {
      "epoch": 2.1391207089391724,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006127814308538487,
      "loss": 1.4292,
      "step": 24863
    },
    {
      "epoch": 2.139206745246494,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006127550868305923,
      "loss": 0.8574,
      "step": 24864
    },
    {
      "epoch": 2.1392927815538156,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006127287424775492,
      "loss": 1.4554,
      "step": 24865
    },
    {
      "epoch": 2.1393788178611373,
      "grad_norm": 1.25,
      "learning_rate": 0.0006127023977947966,
      "loss": 1.114,
      "step": 24866
    },
    {
      "epoch": 2.139464854168459,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006126760527824112,
      "loss": 0.7866,
      "step": 24867
    },
    {
      "epoch": 2.139550890475781,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0006126497074404704,
      "loss": 1.462,
      "step": 24868
    },
    {
      "epoch": 2.1396369267831026,
      "grad_norm": 1.25,
      "learning_rate": 0.0006126233617690511,
      "loss": 1.2548,
      "step": 24869
    },
    {
      "epoch": 2.139722963090424,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006125970157682304,
      "loss": 0.8787,
      "step": 24870
    },
    {
      "epoch": 2.139808999397746,
      "grad_norm": 1.375,
      "learning_rate": 0.0006125706694380853,
      "loss": 0.8621,
      "step": 24871
    },
    {
      "epoch": 2.1398950357050674,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006125443227786929,
      "loss": 1.2321,
      "step": 24872
    },
    {
      "epoch": 2.139981072012389,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006125179757901302,
      "loss": 1.8785,
      "step": 24873
    },
    {
      "epoch": 2.1400671083197107,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006124916284724743,
      "loss": 0.9108,
      "step": 24874
    },
    {
      "epoch": 2.1401531446270328,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006124652808258024,
      "loss": 1.337,
      "step": 24875
    },
    {
      "epoch": 2.1402391809343544,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006124389328501912,
      "loss": 0.7841,
      "step": 24876
    },
    {
      "epoch": 2.140325217241676,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006124125845457181,
      "loss": 0.8751,
      "step": 24877
    },
    {
      "epoch": 2.1404112535489976,
      "grad_norm": 1.5625,
      "learning_rate": 0.0006123862359124602,
      "loss": 1.4905,
      "step": 24878
    },
    {
      "epoch": 2.1404972898563193,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006123598869504943,
      "loss": 0.8583,
      "step": 24879
    },
    {
      "epoch": 2.140583326163641,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0006123335376598975,
      "loss": 1.547,
      "step": 24880
    },
    {
      "epoch": 2.1406693624709625,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006123071880407469,
      "loss": 1.1091,
      "step": 24881
    },
    {
      "epoch": 2.1407553987782846,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006122808380931198,
      "loss": 1.2603,
      "step": 24882
    },
    {
      "epoch": 2.140841435085606,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006122544878170929,
      "loss": 1.4581,
      "step": 24883
    },
    {
      "epoch": 2.140927471392928,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006122281372127436,
      "loss": 1.6757,
      "step": 24884
    },
    {
      "epoch": 2.1410135077002495,
      "grad_norm": 1.125,
      "learning_rate": 0.0006122017862801489,
      "loss": 0.9233,
      "step": 24885
    },
    {
      "epoch": 2.141099544007571,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006121754350193857,
      "loss": 1.2042,
      "step": 24886
    },
    {
      "epoch": 2.1411855803148927,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006121490834305311,
      "loss": 1.4501,
      "step": 24887
    },
    {
      "epoch": 2.1412716166222148,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006121227315136622,
      "loss": 0.6991,
      "step": 24888
    },
    {
      "epoch": 2.1413576529295364,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006120963792688564,
      "loss": 0.7416,
      "step": 24889
    },
    {
      "epoch": 2.141443689236858,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006120700266961902,
      "loss": 1.0963,
      "step": 24890
    },
    {
      "epoch": 2.1415297255441796,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006120436737957411,
      "loss": 0.9049,
      "step": 24891
    },
    {
      "epoch": 2.1416157618515013,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006120173205675861,
      "loss": 1.4357,
      "step": 24892
    },
    {
      "epoch": 2.141701798158823,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0006119909670118021,
      "loss": 1.1274,
      "step": 24893
    },
    {
      "epoch": 2.1417878344661445,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006119646131284664,
      "loss": 1.0391,
      "step": 24894
    },
    {
      "epoch": 2.1418738707734666,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006119382589176559,
      "loss": 0.6824,
      "step": 24895
    },
    {
      "epoch": 2.141959907080788,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006119119043794478,
      "loss": 1.1757,
      "step": 24896
    },
    {
      "epoch": 2.14204594338811,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006118855495139191,
      "loss": 0.4669,
      "step": 24897
    },
    {
      "epoch": 2.1421319796954315,
      "grad_norm": 1.125,
      "learning_rate": 0.000611859194321147,
      "loss": 1.0001,
      "step": 24898
    },
    {
      "epoch": 2.142218016002753,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006118328388012086,
      "loss": 1.0722,
      "step": 24899
    },
    {
      "epoch": 2.1423040523100747,
      "grad_norm": 1.5,
      "learning_rate": 0.0006118064829541808,
      "loss": 1.4787,
      "step": 24900
    },
    {
      "epoch": 2.1423900886173968,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006117801267801408,
      "loss": 0.8036,
      "step": 24901
    },
    {
      "epoch": 2.1424761249247184,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006117537702791655,
      "loss": 0.9894,
      "step": 24902
    },
    {
      "epoch": 2.14256216123204,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006117274134513324,
      "loss": 1.8311,
      "step": 24903
    },
    {
      "epoch": 2.1426481975393616,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006117010562967181,
      "loss": 1.761,
      "step": 24904
    },
    {
      "epoch": 2.1427342338466833,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006116746988154001,
      "loss": 1.5265,
      "step": 24905
    },
    {
      "epoch": 2.142820270154005,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006116483410074552,
      "loss": 0.5285,
      "step": 24906
    },
    {
      "epoch": 2.1429063064613265,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006116219828729608,
      "loss": 0.9571,
      "step": 24907
    },
    {
      "epoch": 2.1429923427686486,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006115956244119937,
      "loss": 1.7072,
      "step": 24908
    },
    {
      "epoch": 2.14307837907597,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006115692656246312,
      "loss": 1.0542,
      "step": 24909
    },
    {
      "epoch": 2.143164415383292,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006115429065109501,
      "loss": 1.2684,
      "step": 24910
    },
    {
      "epoch": 2.1432504516906135,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006115165470710279,
      "loss": 1.4387,
      "step": 24911
    },
    {
      "epoch": 2.143336487997935,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006114901873049414,
      "loss": 1.5432,
      "step": 24912
    },
    {
      "epoch": 2.1434225243052567,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006114638272127677,
      "loss": 0.9926,
      "step": 24913
    },
    {
      "epoch": 2.1435085606125783,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006114374667945842,
      "loss": 1.7242,
      "step": 24914
    },
    {
      "epoch": 2.1435945969199004,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0006114111060504676,
      "loss": 1.26,
      "step": 24915
    },
    {
      "epoch": 2.143680633227222,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006113847449804953,
      "loss": 0.6799,
      "step": 24916
    },
    {
      "epoch": 2.1437666695345436,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006113583835847442,
      "loss": 1.2931,
      "step": 24917
    },
    {
      "epoch": 2.1438527058418653,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006113320218632915,
      "loss": 2.054,
      "step": 24918
    },
    {
      "epoch": 2.143938742149187,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006113056598162143,
      "loss": 1.0385,
      "step": 24919
    },
    {
      "epoch": 2.1440247784565085,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006112792974435898,
      "loss": 0.9824,
      "step": 24920
    },
    {
      "epoch": 2.14411081476383,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006112529347454948,
      "loss": 1.0799,
      "step": 24921
    },
    {
      "epoch": 2.144196851071152,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006112265717220067,
      "loss": 1.3098,
      "step": 24922
    },
    {
      "epoch": 2.144282887378474,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006112002083732025,
      "loss": 1.1401,
      "step": 24923
    },
    {
      "epoch": 2.1443689236857955,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006111738446991593,
      "loss": 1.1931,
      "step": 24924
    },
    {
      "epoch": 2.144454959993117,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006111474806999542,
      "loss": 0.6745,
      "step": 24925
    },
    {
      "epoch": 2.1445409963004387,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006111211163756644,
      "loss": 0.6603,
      "step": 24926
    },
    {
      "epoch": 2.1446270326077603,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000611094751726367,
      "loss": 0.626,
      "step": 24927
    },
    {
      "epoch": 2.144713068915082,
      "grad_norm": 0.90625,
      "learning_rate": 0.000611068386752139,
      "loss": 1.4673,
      "step": 24928
    },
    {
      "epoch": 2.144799105222404,
      "grad_norm": 1.375,
      "learning_rate": 0.0006110420214530574,
      "loss": 1.7655,
      "step": 24929
    },
    {
      "epoch": 2.1448851415297256,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006110156558291997,
      "loss": 0.7561,
      "step": 24930
    },
    {
      "epoch": 2.1449711778370473,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006109892898806429,
      "loss": 1.0519,
      "step": 24931
    },
    {
      "epoch": 2.145057214144369,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006109629236074638,
      "loss": 1.023,
      "step": 24932
    },
    {
      "epoch": 2.1451432504516905,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006109365570097397,
      "loss": 0.9969,
      "step": 24933
    },
    {
      "epoch": 2.145229286759012,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006109101900875479,
      "loss": 0.9848,
      "step": 24934
    },
    {
      "epoch": 2.1453153230663338,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006108838228409653,
      "loss": 1.7437,
      "step": 24935
    },
    {
      "epoch": 2.145401359373656,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000610857455270069,
      "loss": 1.6736,
      "step": 24936
    },
    {
      "epoch": 2.1454873956809775,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006108310873749363,
      "loss": 1.1632,
      "step": 24937
    },
    {
      "epoch": 2.145573431988299,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006108047191556443,
      "loss": 1.126,
      "step": 24938
    },
    {
      "epoch": 2.1456594682956207,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006107783506122699,
      "loss": 0.8316,
      "step": 24939
    },
    {
      "epoch": 2.1457455046029423,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006107519817448904,
      "loss": 1.3788,
      "step": 24940
    },
    {
      "epoch": 2.145831540910264,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006107256125535828,
      "loss": 1.559,
      "step": 24941
    },
    {
      "epoch": 2.145917577217586,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006106992430384245,
      "loss": 1.4483,
      "step": 24942
    },
    {
      "epoch": 2.1460036135249076,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006106728731994925,
      "loss": 0.7276,
      "step": 24943
    },
    {
      "epoch": 2.1460896498322293,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006106465030368636,
      "loss": 0.9254,
      "step": 24944
    },
    {
      "epoch": 2.146175686139551,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006106201325506154,
      "loss": 1.2353,
      "step": 24945
    },
    {
      "epoch": 2.1462617224468725,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006105937617408248,
      "loss": 0.4795,
      "step": 24946
    },
    {
      "epoch": 2.146347758754194,
      "grad_norm": 1.59375,
      "learning_rate": 0.0006105673906075689,
      "loss": 1.0989,
      "step": 24947
    },
    {
      "epoch": 2.1464337950615158,
      "grad_norm": 2.09375,
      "learning_rate": 0.000610541019150925,
      "loss": 1.1423,
      "step": 24948
    },
    {
      "epoch": 2.146519831368838,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006105146473709699,
      "loss": 1.2173,
      "step": 24949
    },
    {
      "epoch": 2.1466058676761595,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006104882752677813,
      "loss": 1.1434,
      "step": 24950
    },
    {
      "epoch": 2.146691903983481,
      "grad_norm": 2.59375,
      "learning_rate": 0.0006104619028414357,
      "loss": 2.1926,
      "step": 24951
    },
    {
      "epoch": 2.1467779402908027,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006104355300920105,
      "loss": 1.3509,
      "step": 24952
    },
    {
      "epoch": 2.1468639765981243,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000610409157019583,
      "loss": 0.8484,
      "step": 24953
    },
    {
      "epoch": 2.146950012905446,
      "grad_norm": 55.75,
      "learning_rate": 0.0006103827836242301,
      "loss": 1.3387,
      "step": 24954
    },
    {
      "epoch": 2.1470360492127676,
      "grad_norm": 0.78125,
      "learning_rate": 0.000610356409906029,
      "loss": 0.6088,
      "step": 24955
    },
    {
      "epoch": 2.1471220855200897,
      "grad_norm": 1.0625,
      "learning_rate": 0.000610330035865057,
      "loss": 0.8514,
      "step": 24956
    },
    {
      "epoch": 2.1472081218274113,
      "grad_norm": 8.9375,
      "learning_rate": 0.0006103036615013911,
      "loss": 1.4967,
      "step": 24957
    },
    {
      "epoch": 2.147294158134733,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006102772868151083,
      "loss": 1.9629,
      "step": 24958
    },
    {
      "epoch": 2.1473801944420545,
      "grad_norm": 130.0,
      "learning_rate": 0.0006102509118062859,
      "loss": 1.3803,
      "step": 24959
    },
    {
      "epoch": 2.147466230749376,
      "grad_norm": 3.59375,
      "learning_rate": 0.0006102245364750011,
      "loss": 0.9419,
      "step": 24960
    },
    {
      "epoch": 2.1475522670566978,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0006101981608213309,
      "loss": 1.0763,
      "step": 24961
    },
    {
      "epoch": 2.14763830336402,
      "grad_norm": 12.25,
      "learning_rate": 0.0006101717848453525,
      "loss": 1.6621,
      "step": 24962
    },
    {
      "epoch": 2.1477243396713415,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006101454085471431,
      "loss": 0.8078,
      "step": 24963
    },
    {
      "epoch": 2.147810375978663,
      "grad_norm": 1.296875,
      "learning_rate": 0.00061011903192678,
      "loss": 1.0935,
      "step": 24964
    },
    {
      "epoch": 2.1478964122859847,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00061009265498434,
      "loss": 0.6305,
      "step": 24965
    },
    {
      "epoch": 2.1479824485933063,
      "grad_norm": 2.59375,
      "learning_rate": 0.0006100662777199002,
      "loss": 0.5869,
      "step": 24966
    },
    {
      "epoch": 2.148068484900628,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006100399001335382,
      "loss": 1.2573,
      "step": 24967
    },
    {
      "epoch": 2.1481545212079496,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006100135222253308,
      "loss": 1.5621,
      "step": 24968
    },
    {
      "epoch": 2.1482405575152717,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006099871439953551,
      "loss": 1.908,
      "step": 24969
    },
    {
      "epoch": 2.1483265938225933,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006099607654436886,
      "loss": 1.4846,
      "step": 24970
    },
    {
      "epoch": 2.148412630129915,
      "grad_norm": 1.25,
      "learning_rate": 0.0006099343865704082,
      "loss": 1.2013,
      "step": 24971
    },
    {
      "epoch": 2.1484986664372365,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006099080073755912,
      "loss": 1.1294,
      "step": 24972
    },
    {
      "epoch": 2.148584702744558,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006098816278593144,
      "loss": 0.5187,
      "step": 24973
    },
    {
      "epoch": 2.1486707390518798,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006098552480216554,
      "loss": 0.7706,
      "step": 24974
    },
    {
      "epoch": 2.1487567753592014,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0006098288678626911,
      "loss": 0.8654,
      "step": 24975
    },
    {
      "epoch": 2.1488428116665235,
      "grad_norm": 1.375,
      "learning_rate": 0.0006098024873824987,
      "loss": 1.401,
      "step": 24976
    },
    {
      "epoch": 2.148928847973845,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006097761065811555,
      "loss": 1.7187,
      "step": 24977
    },
    {
      "epoch": 2.1490148842811667,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006097497254587383,
      "loss": 0.808,
      "step": 24978
    },
    {
      "epoch": 2.1491009205884883,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006097233440153247,
      "loss": 0.8137,
      "step": 24979
    },
    {
      "epoch": 2.14918695689581,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006096969622509917,
      "loss": 0.8644,
      "step": 24980
    },
    {
      "epoch": 2.1492729932031316,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006096705801658162,
      "loss": 0.6171,
      "step": 24981
    },
    {
      "epoch": 2.149359029510453,
      "grad_norm": 0.875,
      "learning_rate": 0.0006096441977598758,
      "loss": 1.0433,
      "step": 24982
    },
    {
      "epoch": 2.1494450658177753,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006096178150332474,
      "loss": 1.4333,
      "step": 24983
    },
    {
      "epoch": 2.149531102125097,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006095914319860082,
      "loss": 1.3591,
      "step": 24984
    },
    {
      "epoch": 2.1496171384324185,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006095650486182353,
      "loss": 1.426,
      "step": 24985
    },
    {
      "epoch": 2.14970317473974,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000609538664930006,
      "loss": 1.3909,
      "step": 24986
    },
    {
      "epoch": 2.1497892110470618,
      "grad_norm": 1.53125,
      "learning_rate": 0.0006095122809213975,
      "loss": 1.368,
      "step": 24987
    },
    {
      "epoch": 2.1498752473543834,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006094858965924866,
      "loss": 1.2117,
      "step": 24988
    },
    {
      "epoch": 2.149961283661705,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006094595119433509,
      "loss": 0.9145,
      "step": 24989
    },
    {
      "epoch": 2.150047319969027,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006094331269740675,
      "loss": 1.4948,
      "step": 24990
    },
    {
      "epoch": 2.1501333562763487,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0006094067416847135,
      "loss": 0.4328,
      "step": 24991
    },
    {
      "epoch": 2.1502193925836703,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006093803560753659,
      "loss": 1.1572,
      "step": 24992
    },
    {
      "epoch": 2.150305428890992,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006093539701461022,
      "loss": 0.6015,
      "step": 24993
    },
    {
      "epoch": 2.1503914651983136,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006093275838969992,
      "loss": 1.1777,
      "step": 24994
    },
    {
      "epoch": 2.150477501505635,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006093011973281344,
      "loss": 1.2214,
      "step": 24995
    },
    {
      "epoch": 2.150563537812957,
      "grad_norm": 1.609375,
      "learning_rate": 0.0006092748104395847,
      "loss": 1.677,
      "step": 24996
    },
    {
      "epoch": 2.150649574120279,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006092484232314277,
      "loss": 1.4668,
      "step": 24997
    },
    {
      "epoch": 2.1507356104276005,
      "grad_norm": 1.125,
      "learning_rate": 0.0006092220357037403,
      "loss": 1.9976,
      "step": 24998
    },
    {
      "epoch": 2.150821646734922,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006091956478565996,
      "loss": 0.8903,
      "step": 24999
    },
    {
      "epoch": 2.150907683042244,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006091692596900827,
      "loss": 1.2047,
      "step": 25000
    },
    {
      "epoch": 2.1509937193495654,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006091428712042673,
      "loss": 1.2619,
      "step": 25001
    },
    {
      "epoch": 2.151079755656887,
      "grad_norm": 1.21875,
      "learning_rate": 0.00060911648239923,
      "loss": 1.4514,
      "step": 25002
    },
    {
      "epoch": 2.151165791964209,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006090900932750481,
      "loss": 1.3536,
      "step": 25003
    },
    {
      "epoch": 2.1512518282715307,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006090637038317991,
      "loss": 0.897,
      "step": 25004
    },
    {
      "epoch": 2.1513378645788523,
      "grad_norm": 1.109375,
      "learning_rate": 0.00060903731406956,
      "loss": 0.6513,
      "step": 25005
    },
    {
      "epoch": 2.151423900886174,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006090109239884078,
      "loss": 0.7892,
      "step": 25006
    },
    {
      "epoch": 2.1515099371934956,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0006089845335884199,
      "loss": 0.781,
      "step": 25007
    },
    {
      "epoch": 2.151595973500817,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006089581428696735,
      "loss": 1.7725,
      "step": 25008
    },
    {
      "epoch": 2.151682009808139,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006089317518322457,
      "loss": 0.8303,
      "step": 25009
    },
    {
      "epoch": 2.151768046115461,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006089053604762136,
      "loss": 1.2936,
      "step": 25010
    },
    {
      "epoch": 2.1518540824227825,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006088789688016548,
      "loss": 0.8599,
      "step": 25011
    },
    {
      "epoch": 2.151940118730104,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006088525768086459,
      "loss": 1.1321,
      "step": 25012
    },
    {
      "epoch": 2.152026155037426,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006088261844972645,
      "loss": 1.3588,
      "step": 25013
    },
    {
      "epoch": 2.1521121913447474,
      "grad_norm": 0.875,
      "learning_rate": 0.0006087997918675876,
      "loss": 1.2442,
      "step": 25014
    },
    {
      "epoch": 2.152198227652069,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006087733989196924,
      "loss": 0.8485,
      "step": 25015
    },
    {
      "epoch": 2.152284263959391,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006087470056536563,
      "loss": 0.8823,
      "step": 25016
    },
    {
      "epoch": 2.1523703002667127,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006087206120695564,
      "loss": 1.0761,
      "step": 25017
    },
    {
      "epoch": 2.1524563365740343,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006086942181674698,
      "loss": 1.0346,
      "step": 25018
    },
    {
      "epoch": 2.152542372881356,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006086678239474736,
      "loss": 1.2978,
      "step": 25019
    },
    {
      "epoch": 2.1526284091886776,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006086414294096454,
      "loss": 0.7338,
      "step": 25020
    },
    {
      "epoch": 2.152714445495999,
      "grad_norm": 1.53125,
      "learning_rate": 0.000608615034554062,
      "loss": 0.8964,
      "step": 25021
    },
    {
      "epoch": 2.152800481803321,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006085886393808009,
      "loss": 0.9082,
      "step": 25022
    },
    {
      "epoch": 2.152886518110643,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006085622438899389,
      "loss": 1.6226,
      "step": 25023
    },
    {
      "epoch": 2.1529725544179645,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0006085358480815537,
      "loss": 0.8982,
      "step": 25024
    },
    {
      "epoch": 2.153058590725286,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006085094519557222,
      "loss": 1.6302,
      "step": 25025
    },
    {
      "epoch": 2.153144627032608,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006084830555125216,
      "loss": 1.0483,
      "step": 25026
    },
    {
      "epoch": 2.1532306633399294,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0006084566587520291,
      "loss": 1.9877,
      "step": 25027
    },
    {
      "epoch": 2.153316699647251,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006084302616743222,
      "loss": 1.7796,
      "step": 25028
    },
    {
      "epoch": 2.1534027359545727,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0006084038642794776,
      "loss": 1.4045,
      "step": 25029
    },
    {
      "epoch": 2.1534887722618947,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006083774665675729,
      "loss": 1.5219,
      "step": 25030
    },
    {
      "epoch": 2.1535748085692163,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006083510685386852,
      "loss": 1.1462,
      "step": 25031
    },
    {
      "epoch": 2.153660844876538,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006083246701928917,
      "loss": 1.4764,
      "step": 25032
    },
    {
      "epoch": 2.1537468811838596,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006082982715302694,
      "loss": 1.8641,
      "step": 25033
    },
    {
      "epoch": 2.153832917491181,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0006082718725508959,
      "loss": 1.3226,
      "step": 25034
    },
    {
      "epoch": 2.153918953798503,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0006082454732548483,
      "loss": 1.451,
      "step": 25035
    },
    {
      "epoch": 2.1540049901058245,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006082190736422037,
      "loss": 1.3911,
      "step": 25036
    },
    {
      "epoch": 2.1540910264131465,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006081926737130393,
      "loss": 1.1468,
      "step": 25037
    },
    {
      "epoch": 2.154177062720468,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006081662734674323,
      "loss": 1.7352,
      "step": 25038
    },
    {
      "epoch": 2.15426309902779,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006081398729054601,
      "loss": 1.7541,
      "step": 25039
    },
    {
      "epoch": 2.1543491353351114,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006081134720272,
      "loss": 1.1734,
      "step": 25040
    },
    {
      "epoch": 2.154435171642433,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0006080870708327287,
      "loss": 1.843,
      "step": 25041
    },
    {
      "epoch": 2.1545212079497547,
      "grad_norm": 1.453125,
      "learning_rate": 0.0006080606693221238,
      "loss": 1.2086,
      "step": 25042
    },
    {
      "epoch": 2.1546072442570763,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006080342674954625,
      "loss": 0.6396,
      "step": 25043
    },
    {
      "epoch": 2.1546932805643983,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006080078653528221,
      "loss": 0.7473,
      "step": 25044
    },
    {
      "epoch": 2.15477931687172,
      "grad_norm": 2.71875,
      "learning_rate": 0.0006079814628942795,
      "loss": 1.0965,
      "step": 25045
    },
    {
      "epoch": 2.1548653531790416,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006079550601199123,
      "loss": 0.8411,
      "step": 25046
    },
    {
      "epoch": 2.1549513894863632,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006079286570297975,
      "loss": 2.1097,
      "step": 25047
    },
    {
      "epoch": 2.155037425793685,
      "grad_norm": 3.234375,
      "learning_rate": 0.0006079022536240122,
      "loss": 1.1279,
      "step": 25048
    },
    {
      "epoch": 2.1551234621010065,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000607875849902634,
      "loss": 0.9667,
      "step": 25049
    },
    {
      "epoch": 2.155209498408328,
      "grad_norm": 2.421875,
      "learning_rate": 0.0006078494458657399,
      "loss": 1.6046,
      "step": 25050
    },
    {
      "epoch": 2.15529553471565,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006078230415134069,
      "loss": 0.7675,
      "step": 25051
    },
    {
      "epoch": 2.155381571022972,
      "grad_norm": 1.25,
      "learning_rate": 0.0006077966368457126,
      "loss": 1.5234,
      "step": 25052
    },
    {
      "epoch": 2.1554676073302934,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006077702318627343,
      "loss": 1.5955,
      "step": 25053
    },
    {
      "epoch": 2.155553643637615,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006077438265645489,
      "loss": 1.3835,
      "step": 25054
    },
    {
      "epoch": 2.1556396799449367,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006077174209512337,
      "loss": 1.4997,
      "step": 25055
    },
    {
      "epoch": 2.1557257162522583,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000607691015022866,
      "loss": 1.2865,
      "step": 25056
    },
    {
      "epoch": 2.15581175255958,
      "grad_norm": 0.875,
      "learning_rate": 0.0006076646087795232,
      "loss": 0.7414,
      "step": 25057
    },
    {
      "epoch": 2.155897788866902,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006076382022212822,
      "loss": 1.7207,
      "step": 25058
    },
    {
      "epoch": 2.1559838251742236,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006076117953482203,
      "loss": 1.078,
      "step": 25059
    },
    {
      "epoch": 2.1560698614815452,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006075853881604151,
      "loss": 1.2063,
      "step": 25060
    },
    {
      "epoch": 2.156155897788867,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006075589806579434,
      "loss": 1.2308,
      "step": 25061
    },
    {
      "epoch": 2.1562419340961885,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006075325728408827,
      "loss": 1.5272,
      "step": 25062
    },
    {
      "epoch": 2.15632797040351,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00060750616470931,
      "loss": 1.2842,
      "step": 25063
    },
    {
      "epoch": 2.156414006710832,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006074797562633028,
      "loss": 0.9407,
      "step": 25064
    },
    {
      "epoch": 2.156500043018154,
      "grad_norm": 12.0625,
      "learning_rate": 0.0006074533475029381,
      "loss": 1.7032,
      "step": 25065
    },
    {
      "epoch": 2.1565860793254754,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006074269384282932,
      "loss": 0.9709,
      "step": 25066
    },
    {
      "epoch": 2.156672115632797,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006074005290394455,
      "loss": 1.654,
      "step": 25067
    },
    {
      "epoch": 2.1567581519401187,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006073741193364723,
      "loss": 1.04,
      "step": 25068
    },
    {
      "epoch": 2.1568441882474403,
      "grad_norm": 1.5,
      "learning_rate": 0.0006073477093194506,
      "loss": 1.6079,
      "step": 25069
    },
    {
      "epoch": 2.156930224554762,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0006073212989884575,
      "loss": 1.3184,
      "step": 25070
    },
    {
      "epoch": 2.157016260862084,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006072948883435706,
      "loss": 1.6651,
      "step": 25071
    },
    {
      "epoch": 2.1571022971694056,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0006072684773848671,
      "loss": 1.6629,
      "step": 25072
    },
    {
      "epoch": 2.1571883334767272,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006072420661124242,
      "loss": 1.1378,
      "step": 25073
    },
    {
      "epoch": 2.157274369784049,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006072156545263189,
      "loss": 1.0187,
      "step": 25074
    },
    {
      "epoch": 2.1573604060913705,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006071892426266288,
      "loss": 1.5587,
      "step": 25075
    },
    {
      "epoch": 2.157446442398692,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006071628304134311,
      "loss": 0.8973,
      "step": 25076
    },
    {
      "epoch": 2.157532478706014,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006071364178868028,
      "loss": 1.3545,
      "step": 25077
    },
    {
      "epoch": 2.157618515013336,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006071100050468215,
      "loss": 0.5268,
      "step": 25078
    },
    {
      "epoch": 2.1577045513206574,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0006070835918935641,
      "loss": 1.0671,
      "step": 25079
    },
    {
      "epoch": 2.157790587627979,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006070571784271081,
      "loss": 1.1657,
      "step": 25080
    },
    {
      "epoch": 2.1578766239353007,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006070307646475307,
      "loss": 0.9883,
      "step": 25081
    },
    {
      "epoch": 2.1579626602426223,
      "grad_norm": 1.4375,
      "learning_rate": 0.000607004350554909,
      "loss": 1.5567,
      "step": 25082
    },
    {
      "epoch": 2.158048696549944,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006069779361493205,
      "loss": 1.7368,
      "step": 25083
    },
    {
      "epoch": 2.158134732857266,
      "grad_norm": 0.890625,
      "learning_rate": 0.0006069515214308423,
      "loss": 0.8703,
      "step": 25084
    },
    {
      "epoch": 2.1582207691645876,
      "grad_norm": 1.640625,
      "learning_rate": 0.0006069251063995516,
      "loss": 0.8637,
      "step": 25085
    },
    {
      "epoch": 2.1583068054719092,
      "grad_norm": 0.953125,
      "learning_rate": 0.000606898691055526,
      "loss": 1.6273,
      "step": 25086
    },
    {
      "epoch": 2.158392841779231,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006068722753988423,
      "loss": 1.2204,
      "step": 25087
    },
    {
      "epoch": 2.1584788780865525,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006068458594295781,
      "loss": 1.0469,
      "step": 25088
    },
    {
      "epoch": 2.158564914393874,
      "grad_norm": 0.875,
      "learning_rate": 0.0006068194431478104,
      "loss": 1.0583,
      "step": 25089
    },
    {
      "epoch": 2.1586509507011957,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006067930265536168,
      "loss": 0.8799,
      "step": 25090
    },
    {
      "epoch": 2.158736987008518,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0006067666096470743,
      "loss": 1.2,
      "step": 25091
    },
    {
      "epoch": 2.1588230233158394,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006067401924282602,
      "loss": 1.0512,
      "step": 25092
    },
    {
      "epoch": 2.158909059623161,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006067137748972518,
      "loss": 1.3167,
      "step": 25093
    },
    {
      "epoch": 2.1589950959304827,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006066873570541264,
      "loss": 1.7692,
      "step": 25094
    },
    {
      "epoch": 2.1590811322378043,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006066609388989613,
      "loss": 1.6573,
      "step": 25095
    },
    {
      "epoch": 2.159167168545126,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006066345204318336,
      "loss": 1.109,
      "step": 25096
    },
    {
      "epoch": 2.1592532048524475,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006066081016528205,
      "loss": 1.8086,
      "step": 25097
    },
    {
      "epoch": 2.1593392411597696,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0006065816825619997,
      "loss": 1.4673,
      "step": 25098
    },
    {
      "epoch": 2.1594252774670912,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006065552631594481,
      "loss": 1.5145,
      "step": 25099
    },
    {
      "epoch": 2.159511313774413,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006065288434452431,
      "loss": 1.8385,
      "step": 25100
    },
    {
      "epoch": 2.1595973500817345,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006065024234194619,
      "loss": 1.9599,
      "step": 25101
    },
    {
      "epoch": 2.159683386389056,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006064760030821819,
      "loss": 1.7394,
      "step": 25102
    },
    {
      "epoch": 2.1597694226963777,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006064495824334804,
      "loss": 0.9207,
      "step": 25103
    },
    {
      "epoch": 2.1598554590036994,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006064231614734343,
      "loss": 1.0107,
      "step": 25104
    },
    {
      "epoch": 2.1599414953110214,
      "grad_norm": 1.0,
      "learning_rate": 0.0006063967402021213,
      "loss": 1.0257,
      "step": 25105
    },
    {
      "epoch": 2.160027531618343,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006063703186196186,
      "loss": 1.3134,
      "step": 25106
    },
    {
      "epoch": 2.1601135679256647,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006063438967260032,
      "loss": 1.8356,
      "step": 25107
    },
    {
      "epoch": 2.1601996042329863,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006063174745213526,
      "loss": 1.113,
      "step": 25108
    },
    {
      "epoch": 2.160285640540308,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006062910520057443,
      "loss": 1.7268,
      "step": 25109
    },
    {
      "epoch": 2.1603716768476295,
      "grad_norm": 0.875,
      "learning_rate": 0.0006062646291792552,
      "loss": 1.3215,
      "step": 25110
    },
    {
      "epoch": 2.160457713154951,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006062382060419628,
      "loss": 1.2526,
      "step": 25111
    },
    {
      "epoch": 2.1605437494622732,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0006062117825939441,
      "loss": 1.1935,
      "step": 25112
    },
    {
      "epoch": 2.160629785769595,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006061853588352768,
      "loss": 1.6489,
      "step": 25113
    },
    {
      "epoch": 2.1607158220769165,
      "grad_norm": 1.296875,
      "learning_rate": 0.000606158934766038,
      "loss": 0.8477,
      "step": 25114
    },
    {
      "epoch": 2.160801858384238,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006061325103863048,
      "loss": 1.21,
      "step": 25115
    },
    {
      "epoch": 2.1608878946915597,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0006061060856961547,
      "loss": 1.1757,
      "step": 25116
    },
    {
      "epoch": 2.1609739309988814,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0006060796606956651,
      "loss": 1.2991,
      "step": 25117
    },
    {
      "epoch": 2.1610599673062034,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006060532353849131,
      "loss": 0.6979,
      "step": 25118
    },
    {
      "epoch": 2.161146003613525,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006060268097639759,
      "loss": 1.4783,
      "step": 25119
    },
    {
      "epoch": 2.1612320399208467,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006060003838329309,
      "loss": 1.6581,
      "step": 25120
    },
    {
      "epoch": 2.1613180762281683,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006059739575918554,
      "loss": 1.8101,
      "step": 25121
    },
    {
      "epoch": 2.16140411253549,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006059475310408266,
      "loss": 0.6754,
      "step": 25122
    },
    {
      "epoch": 2.1614901488428115,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000605921104179922,
      "loss": 1.1926,
      "step": 25123
    },
    {
      "epoch": 2.161576185150133,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006058946770092187,
      "loss": 1.6642,
      "step": 25124
    },
    {
      "epoch": 2.1616622214574552,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006058682495287942,
      "loss": 0.6997,
      "step": 25125
    },
    {
      "epoch": 2.161748257764777,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0006058418217387254,
      "loss": 1.2054,
      "step": 25126
    },
    {
      "epoch": 2.1618342940720985,
      "grad_norm": 1.3125,
      "learning_rate": 0.00060581539363909,
      "loss": 2.1113,
      "step": 25127
    },
    {
      "epoch": 2.16192033037942,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006057889652299653,
      "loss": 1.3142,
      "step": 25128
    },
    {
      "epoch": 2.1620063666867417,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006057625365114283,
      "loss": 0.7697,
      "step": 25129
    },
    {
      "epoch": 2.1620924029940634,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006057361074835564,
      "loss": 1.2617,
      "step": 25130
    },
    {
      "epoch": 2.162178439301385,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006057096781464271,
      "loss": 1.3054,
      "step": 25131
    },
    {
      "epoch": 2.162264475608707,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006056832485001175,
      "loss": 0.8418,
      "step": 25132
    },
    {
      "epoch": 2.1623505119160287,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000605656818544705,
      "loss": 1.5447,
      "step": 25133
    },
    {
      "epoch": 2.1624365482233503,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006056303882802668,
      "loss": 1.1725,
      "step": 25134
    },
    {
      "epoch": 2.162522584530672,
      "grad_norm": 1.125,
      "learning_rate": 0.0006056039577068802,
      "loss": 1.0477,
      "step": 25135
    },
    {
      "epoch": 2.1626086208379935,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006055775268246226,
      "loss": 1.2692,
      "step": 25136
    },
    {
      "epoch": 2.162694657145315,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006055510956335713,
      "loss": 0.7127,
      "step": 25137
    },
    {
      "epoch": 2.1627806934526372,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006055246641338035,
      "loss": 1.0585,
      "step": 25138
    },
    {
      "epoch": 2.162866729759959,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006054982323253967,
      "loss": 0.7609,
      "step": 25139
    },
    {
      "epoch": 2.1629527660672805,
      "grad_norm": 1.25,
      "learning_rate": 0.000605471800208428,
      "loss": 0.716,
      "step": 25140
    },
    {
      "epoch": 2.163038802374602,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006054453677829748,
      "loss": 1.6414,
      "step": 25141
    },
    {
      "epoch": 2.1631248386819237,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006054189350491143,
      "loss": 1.0275,
      "step": 25142
    },
    {
      "epoch": 2.1632108749892454,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006053925020069242,
      "loss": 1.7163,
      "step": 25143
    },
    {
      "epoch": 2.163296911296567,
      "grad_norm": 1.734375,
      "learning_rate": 0.0006053660686564813,
      "loss": 1.3722,
      "step": 25144
    },
    {
      "epoch": 2.163382947603889,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006053396349978632,
      "loss": 0.9029,
      "step": 25145
    },
    {
      "epoch": 2.1634689839112107,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000605313201031147,
      "loss": 1.4353,
      "step": 25146
    },
    {
      "epoch": 2.1635550202185323,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006052867667564104,
      "loss": 2.0234,
      "step": 25147
    },
    {
      "epoch": 2.163641056525854,
      "grad_norm": 0.921875,
      "learning_rate": 0.0006052603321737305,
      "loss": 1.3083,
      "step": 25148
    },
    {
      "epoch": 2.1637270928331755,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006052338972831844,
      "loss": 0.7829,
      "step": 25149
    },
    {
      "epoch": 2.163813129140497,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006052074620848498,
      "loss": 1.3938,
      "step": 25150
    },
    {
      "epoch": 2.163899165447819,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006051810265788038,
      "loss": 1.1981,
      "step": 25151
    },
    {
      "epoch": 2.163985201755141,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006051545907651238,
      "loss": 0.8547,
      "step": 25152
    },
    {
      "epoch": 2.1640712380624625,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006051281546438869,
      "loss": 1.2892,
      "step": 25153
    },
    {
      "epoch": 2.164157274369784,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006051017182151707,
      "loss": 0.9348,
      "step": 25154
    },
    {
      "epoch": 2.1642433106771057,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006050752814790526,
      "loss": 1.1043,
      "step": 25155
    },
    {
      "epoch": 2.1643293469844274,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006050488444356094,
      "loss": 0.5886,
      "step": 25156
    },
    {
      "epoch": 2.164415383291749,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006050224070849189,
      "loss": 2.0773,
      "step": 25157
    },
    {
      "epoch": 2.1645014195990706,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006049959694270584,
      "loss": 1.1575,
      "step": 25158
    },
    {
      "epoch": 2.1645874559063927,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000604969531462105,
      "loss": 1.4584,
      "step": 25159
    },
    {
      "epoch": 2.1646734922137143,
      "grad_norm": 1.25,
      "learning_rate": 0.0006049430931901361,
      "loss": 1.3458,
      "step": 25160
    },
    {
      "epoch": 2.164759528521036,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006049166546112291,
      "loss": 0.8548,
      "step": 25161
    },
    {
      "epoch": 2.1648455648283575,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006048902157254613,
      "loss": 1.2219,
      "step": 25162
    },
    {
      "epoch": 2.164931601135679,
      "grad_norm": 5.96875,
      "learning_rate": 0.0006048637765329099,
      "loss": 1.5745,
      "step": 25163
    },
    {
      "epoch": 2.165017637443001,
      "grad_norm": 1.65625,
      "learning_rate": 0.0006048373370336524,
      "loss": 1.7583,
      "step": 25164
    },
    {
      "epoch": 2.1651036737503224,
      "grad_norm": 0.8125,
      "learning_rate": 0.0006048108972277662,
      "loss": 1.0161,
      "step": 25165
    },
    {
      "epoch": 2.1651897100576445,
      "grad_norm": 1.0,
      "learning_rate": 0.0006047844571153285,
      "loss": 1.2586,
      "step": 25166
    },
    {
      "epoch": 2.165275746364966,
      "grad_norm": 2.8125,
      "learning_rate": 0.0006047580166964167,
      "loss": 1.1671,
      "step": 25167
    },
    {
      "epoch": 2.1653617826722877,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006047315759711079,
      "loss": 1.7545,
      "step": 25168
    },
    {
      "epoch": 2.1654478189796094,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006047051349394796,
      "loss": 1.4326,
      "step": 25169
    },
    {
      "epoch": 2.165533855286931,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006046786936016093,
      "loss": 0.908,
      "step": 25170
    },
    {
      "epoch": 2.1656198915942526,
      "grad_norm": 1.25,
      "learning_rate": 0.000604652251957574,
      "loss": 1.2416,
      "step": 25171
    },
    {
      "epoch": 2.1657059279015742,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0006046258100074513,
      "loss": 1.5241,
      "step": 25172
    },
    {
      "epoch": 2.1657919642088963,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006045993677513186,
      "loss": 1.4948,
      "step": 25173
    },
    {
      "epoch": 2.165878000516218,
      "grad_norm": 1.109375,
      "learning_rate": 0.000604572925189253,
      "loss": 0.8269,
      "step": 25174
    },
    {
      "epoch": 2.1659640368235396,
      "grad_norm": 1.203125,
      "learning_rate": 0.0006045464823213318,
      "loss": 1.0805,
      "step": 25175
    },
    {
      "epoch": 2.166050073130861,
      "grad_norm": 1.34375,
      "learning_rate": 0.0006045200391476326,
      "loss": 1.1456,
      "step": 25176
    },
    {
      "epoch": 2.166136109438183,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006044935956682326,
      "loss": 0.8355,
      "step": 25177
    },
    {
      "epoch": 2.1662221457455044,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0006044671518832092,
      "loss": 1.9671,
      "step": 25178
    },
    {
      "epoch": 2.1663081820528265,
      "grad_norm": 2.359375,
      "learning_rate": 0.0006044407077926395,
      "loss": 1.4718,
      "step": 25179
    },
    {
      "epoch": 2.166394218360148,
      "grad_norm": 1.8125,
      "learning_rate": 0.0006044142633966012,
      "loss": 1.2672,
      "step": 25180
    },
    {
      "epoch": 2.1664802546674697,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006043878186951714,
      "loss": 0.8159,
      "step": 25181
    },
    {
      "epoch": 2.1665662909747914,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006043613736884277,
      "loss": 1.5674,
      "step": 25182
    },
    {
      "epoch": 2.166652327282113,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006043349283764471,
      "loss": 1.3522,
      "step": 25183
    },
    {
      "epoch": 2.1667383635894346,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006043084827593073,
      "loss": 1.4019,
      "step": 25184
    },
    {
      "epoch": 2.1668243998967562,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006042820368370854,
      "loss": 1.1764,
      "step": 25185
    },
    {
      "epoch": 2.1669104362040783,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006042555906098587,
      "loss": 1.3875,
      "step": 25186
    },
    {
      "epoch": 2.1669964725114,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006042291440777048,
      "loss": 1.2522,
      "step": 25187
    },
    {
      "epoch": 2.1670825088187216,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006042026972407009,
      "loss": 0.703,
      "step": 25188
    },
    {
      "epoch": 2.167168545126043,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006041762500989244,
      "loss": 1.6011,
      "step": 25189
    },
    {
      "epoch": 2.167254581433365,
      "grad_norm": 0.90625,
      "learning_rate": 0.0006041498026524525,
      "loss": 0.7801,
      "step": 25190
    },
    {
      "epoch": 2.1673406177406864,
      "grad_norm": 1.25,
      "learning_rate": 0.000604123354901363,
      "loss": 1.3124,
      "step": 25191
    },
    {
      "epoch": 2.1674266540480085,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006040969068457327,
      "loss": 1.699,
      "step": 25192
    },
    {
      "epoch": 2.16751269035533,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006040704584856393,
      "loss": 1.261,
      "step": 25193
    },
    {
      "epoch": 2.1675987266626517,
      "grad_norm": 47.5,
      "learning_rate": 0.0006040440098211598,
      "loss": 1.8133,
      "step": 25194
    },
    {
      "epoch": 2.1676847629699734,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000604017560852372,
      "loss": 1.0593,
      "step": 25195
    },
    {
      "epoch": 2.167770799277295,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006039911115793531,
      "loss": 1.5309,
      "step": 25196
    },
    {
      "epoch": 2.1678568355846166,
      "grad_norm": 0.875,
      "learning_rate": 0.0006039646620021803,
      "loss": 1.1427,
      "step": 25197
    },
    {
      "epoch": 2.1679428718919382,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006039382121209311,
      "loss": 0.9752,
      "step": 25198
    },
    {
      "epoch": 2.1680289081992603,
      "grad_norm": 7.125,
      "learning_rate": 0.000603911761935683,
      "loss": 0.6848,
      "step": 25199
    },
    {
      "epoch": 2.168114944506582,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0006038853114465131,
      "loss": 1.0501,
      "step": 25200
    },
    {
      "epoch": 2.1682009808139036,
      "grad_norm": 1.703125,
      "learning_rate": 0.0006038588606534989,
      "loss": 0.7289,
      "step": 25201
    },
    {
      "epoch": 2.168287017121225,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006038324095567175,
      "loss": 1.28,
      "step": 25202
    },
    {
      "epoch": 2.168373053428547,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0006038059581562468,
      "loss": 1.1963,
      "step": 25203
    },
    {
      "epoch": 2.1684590897358684,
      "grad_norm": 1.0625,
      "learning_rate": 0.0006037795064521639,
      "loss": 0.9136,
      "step": 25204
    },
    {
      "epoch": 2.16854512604319,
      "grad_norm": 1.359375,
      "learning_rate": 0.0006037530544445457,
      "loss": 1.3928,
      "step": 25205
    },
    {
      "epoch": 2.168631162350512,
      "grad_norm": 1.046875,
      "learning_rate": 0.0006037266021334704,
      "loss": 1.5519,
      "step": 25206
    },
    {
      "epoch": 2.1687171986578337,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0006037001495190149,
      "loss": 1.2814,
      "step": 25207
    },
    {
      "epoch": 2.1688032349651554,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006036736966012566,
      "loss": 1.1074,
      "step": 25208
    },
    {
      "epoch": 2.168889271272477,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006036472433802729,
      "loss": 1.7634,
      "step": 25209
    },
    {
      "epoch": 2.1689753075797986,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006036207898561411,
      "loss": 1.4565,
      "step": 25210
    },
    {
      "epoch": 2.1690613438871202,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006035943360289388,
      "loss": 0.4991,
      "step": 25211
    },
    {
      "epoch": 2.169147380194442,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000603567881898743,
      "loss": 0.621,
      "step": 25212
    },
    {
      "epoch": 2.169233416501764,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006035414274656313,
      "loss": 1.384,
      "step": 25213
    },
    {
      "epoch": 2.1693194528090856,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0006035149727296813,
      "loss": 1.6107,
      "step": 25214
    },
    {
      "epoch": 2.169405489116407,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006034885176909701,
      "loss": 0.7494,
      "step": 25215
    },
    {
      "epoch": 2.169491525423729,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006034620623495748,
      "loss": 1.2493,
      "step": 25216
    },
    {
      "epoch": 2.1695775617310504,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006034356067055734,
      "loss": 1.1667,
      "step": 25217
    },
    {
      "epoch": 2.169663598038372,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006034091507590428,
      "loss": 1.34,
      "step": 25218
    },
    {
      "epoch": 2.1697496343456937,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006033826945100606,
      "loss": 0.9399,
      "step": 25219
    },
    {
      "epoch": 2.1698356706530157,
      "grad_norm": 1.0625,
      "learning_rate": 0.000603356237958704,
      "loss": 1.2776,
      "step": 25220
    },
    {
      "epoch": 2.1699217069603374,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006033297811050508,
      "loss": 1.1326,
      "step": 25221
    },
    {
      "epoch": 2.170007743267659,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0006033033239491778,
      "loss": 1.2928,
      "step": 25222
    },
    {
      "epoch": 2.1700937795749806,
      "grad_norm": 0.96875,
      "learning_rate": 0.000603276866491163,
      "loss": 0.7837,
      "step": 25223
    },
    {
      "epoch": 2.1701798158823022,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006032504087310831,
      "loss": 1.6999,
      "step": 25224
    },
    {
      "epoch": 2.170265852189624,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000603223950669016,
      "loss": 1.4903,
      "step": 25225
    },
    {
      "epoch": 2.1703518884969455,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0006031974923050389,
      "loss": 1.3063,
      "step": 25226
    },
    {
      "epoch": 2.1704379248042676,
      "grad_norm": 1.9375,
      "learning_rate": 0.0006031710336392292,
      "loss": 1.4116,
      "step": 25227
    },
    {
      "epoch": 2.170523961111589,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006031445746716644,
      "loss": 1.1366,
      "step": 25228
    },
    {
      "epoch": 2.170609997418911,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0006031181154024217,
      "loss": 0.605,
      "step": 25229
    },
    {
      "epoch": 2.1706960337262324,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006030916558315785,
      "loss": 1.8565,
      "step": 25230
    },
    {
      "epoch": 2.170782070033554,
      "grad_norm": 1.78125,
      "learning_rate": 0.0006030651959592122,
      "loss": 1.0641,
      "step": 25231
    },
    {
      "epoch": 2.1708681063408757,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006030387357854005,
      "loss": 1.2013,
      "step": 25232
    },
    {
      "epoch": 2.1709541426481973,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006030122753102204,
      "loss": 1.2198,
      "step": 25233
    },
    {
      "epoch": 2.1710401789555194,
      "grad_norm": 1.671875,
      "learning_rate": 0.0006029858145337495,
      "loss": 1.7242,
      "step": 25234
    },
    {
      "epoch": 2.171126215262841,
      "grad_norm": 1.015625,
      "learning_rate": 0.0006029593534560649,
      "loss": 0.5974,
      "step": 25235
    },
    {
      "epoch": 2.1712122515701626,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006029328920772445,
      "loss": 0.7877,
      "step": 25236
    },
    {
      "epoch": 2.1712982878774842,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006029064303973652,
      "loss": 1.2379,
      "step": 25237
    },
    {
      "epoch": 2.171384324184806,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0006028799684165046,
      "loss": 0.9776,
      "step": 25238
    },
    {
      "epoch": 2.1714703604921275,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006028535061347401,
      "loss": 1.0191,
      "step": 25239
    },
    {
      "epoch": 2.1715563967994496,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006028270435521492,
      "loss": 1.1941,
      "step": 25240
    },
    {
      "epoch": 2.171642433106771,
      "grad_norm": 1.4375,
      "learning_rate": 0.0006028005806688093,
      "loss": 1.1012,
      "step": 25241
    },
    {
      "epoch": 2.171728469414093,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006027741174847974,
      "loss": 0.8675,
      "step": 25242
    },
    {
      "epoch": 2.1718145057214144,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006027476540001914,
      "loss": 1.338,
      "step": 25243
    },
    {
      "epoch": 2.171900542028736,
      "grad_norm": 1.15625,
      "learning_rate": 0.0006027211902150686,
      "loss": 0.5639,
      "step": 25244
    },
    {
      "epoch": 2.1719865783360577,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006026947261295061,
      "loss": 1.326,
      "step": 25245
    },
    {
      "epoch": 2.1720726146433793,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006026682617435815,
      "loss": 0.9207,
      "step": 25246
    },
    {
      "epoch": 2.1721586509507014,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006026417970573724,
      "loss": 1.4841,
      "step": 25247
    },
    {
      "epoch": 2.172244687258023,
      "grad_norm": 1.375,
      "learning_rate": 0.0006026153320709557,
      "loss": 1.6435,
      "step": 25248
    },
    {
      "epoch": 2.1723307235653446,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0006025888667844094,
      "loss": 1.1009,
      "step": 25249
    },
    {
      "epoch": 2.1724167598726662,
      "grad_norm": 0.953125,
      "learning_rate": 0.0006025624011978104,
      "loss": 1.2484,
      "step": 25250
    },
    {
      "epoch": 2.172502796179988,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006025359353112364,
      "loss": 1.4972,
      "step": 25251
    },
    {
      "epoch": 2.1725888324873095,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0006025094691247647,
      "loss": 1.6446,
      "step": 25252
    },
    {
      "epoch": 2.1726748687946316,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006024830026384728,
      "loss": 1.2999,
      "step": 25253
    },
    {
      "epoch": 2.172760905101953,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0006024565358524381,
      "loss": 1.6066,
      "step": 25254
    },
    {
      "epoch": 2.172846941409275,
      "grad_norm": 2.0625,
      "learning_rate": 0.0006024300687667378,
      "loss": 1.0356,
      "step": 25255
    },
    {
      "epoch": 2.1729329777165964,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0006024036013814496,
      "loss": 1.2574,
      "step": 25256
    },
    {
      "epoch": 2.173019014023918,
      "grad_norm": 1.140625,
      "learning_rate": 0.0006023771336966508,
      "loss": 1.0469,
      "step": 25257
    },
    {
      "epoch": 2.1731050503312397,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006023506657124187,
      "loss": 0.777,
      "step": 25258
    },
    {
      "epoch": 2.1731910866385613,
      "grad_norm": 0.859375,
      "learning_rate": 0.0006023241974288308,
      "loss": 1.2244,
      "step": 25259
    },
    {
      "epoch": 2.1732771229458834,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006022977288459647,
      "loss": 1.4796,
      "step": 25260
    },
    {
      "epoch": 2.173363159253205,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006022712599638973,
      "loss": 1.3015,
      "step": 25261
    },
    {
      "epoch": 2.1734491955605266,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006022447907827066,
      "loss": 1.4,
      "step": 25262
    },
    {
      "epoch": 2.1735352318678482,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0006022183213024699,
      "loss": 1.3233,
      "step": 25263
    },
    {
      "epoch": 2.17362126817517,
      "grad_norm": 0.9375,
      "learning_rate": 0.0006021918515232643,
      "loss": 1.4171,
      "step": 25264
    },
    {
      "epoch": 2.1737073044824915,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006021653814451674,
      "loss": 1.0852,
      "step": 25265
    },
    {
      "epoch": 2.173793340789813,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006021389110682566,
      "loss": 1.4489,
      "step": 25266
    },
    {
      "epoch": 2.173879377097135,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0006021124403926094,
      "loss": 0.9194,
      "step": 25267
    },
    {
      "epoch": 2.173965413404457,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0006020859694183033,
      "loss": 1.2035,
      "step": 25268
    },
    {
      "epoch": 2.1740514497117784,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006020594981454155,
      "loss": 1.3578,
      "step": 25269
    },
    {
      "epoch": 2.1741374860191,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006020330265740234,
      "loss": 1.2831,
      "step": 25270
    },
    {
      "epoch": 2.1742235223264217,
      "grad_norm": 1.25,
      "learning_rate": 0.0006020065547042048,
      "loss": 1.3475,
      "step": 25271
    },
    {
      "epoch": 2.1743095586337433,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006019800825360366,
      "loss": 1.5512,
      "step": 25272
    },
    {
      "epoch": 2.174395594941065,
      "grad_norm": 1.09375,
      "learning_rate": 0.0006019536100695967,
      "loss": 1.4009,
      "step": 25273
    },
    {
      "epoch": 2.174481631248387,
      "grad_norm": 1.046875,
      "learning_rate": 0.000601927137304962,
      "loss": 1.3565,
      "step": 25274
    },
    {
      "epoch": 2.1745676675557086,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006019006642422105,
      "loss": 0.9563,
      "step": 25275
    },
    {
      "epoch": 2.1746537038630303,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0006018741908814194,
      "loss": 1.1671,
      "step": 25276
    },
    {
      "epoch": 2.174739740170352,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006018477172226661,
      "loss": 1.4893,
      "step": 25277
    },
    {
      "epoch": 2.1748257764776735,
      "grad_norm": 1.234375,
      "learning_rate": 0.000601821243266028,
      "loss": 0.8833,
      "step": 25278
    },
    {
      "epoch": 2.174911812784995,
      "grad_norm": 1.890625,
      "learning_rate": 0.0006017947690115826,
      "loss": 1.7099,
      "step": 25279
    },
    {
      "epoch": 2.1749978490923167,
      "grad_norm": 1.109375,
      "learning_rate": 0.0006017682944594072,
      "loss": 1.0779,
      "step": 25280
    },
    {
      "epoch": 2.175083885399639,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006017418196095795,
      "loss": 1.1997,
      "step": 25281
    },
    {
      "epoch": 2.1751699217069604,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0006017153444621766,
      "loss": 1.8783,
      "step": 25282
    },
    {
      "epoch": 2.175255958014282,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006016888690172761,
      "loss": 1.1779,
      "step": 25283
    },
    {
      "epoch": 2.1753419943216037,
      "grad_norm": 1.21875,
      "learning_rate": 0.0006016623932749554,
      "loss": 1.0002,
      "step": 25284
    },
    {
      "epoch": 2.1754280306289253,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0006016359172352921,
      "loss": 0.9799,
      "step": 25285
    },
    {
      "epoch": 2.175514066936247,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006016094408983635,
      "loss": 1.3249,
      "step": 25286
    },
    {
      "epoch": 2.1756001032435686,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006015829642642469,
      "loss": 1.7207,
      "step": 25287
    },
    {
      "epoch": 2.1756861395508906,
      "grad_norm": 1.203125,
      "learning_rate": 0.00060155648733302,
      "loss": 1.3463,
      "step": 25288
    },
    {
      "epoch": 2.1757721758582123,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006015300101047603,
      "loss": 0.752,
      "step": 25289
    },
    {
      "epoch": 2.175858212165534,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006015035325795448,
      "loss": 0.7393,
      "step": 25290
    },
    {
      "epoch": 2.1759442484728555,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0006014770547574513,
      "loss": 1.3205,
      "step": 25291
    },
    {
      "epoch": 2.176030284780177,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006014505766385571,
      "loss": 1.4363,
      "step": 25292
    },
    {
      "epoch": 2.1761163210874988,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006014240982229398,
      "loss": 1.7375,
      "step": 25293
    },
    {
      "epoch": 2.176202357394821,
      "grad_norm": 1.25,
      "learning_rate": 0.0006013976195106767,
      "loss": 1.3996,
      "step": 25294
    },
    {
      "epoch": 2.1762883937021424,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006013711405018451,
      "loss": 1.1281,
      "step": 25295
    },
    {
      "epoch": 2.176374430009464,
      "grad_norm": 1.390625,
      "learning_rate": 0.0006013446611965228,
      "loss": 1.4492,
      "step": 25296
    },
    {
      "epoch": 2.1764604663167857,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006013181815947873,
      "loss": 1.7466,
      "step": 25297
    },
    {
      "epoch": 2.1765465026241073,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0006012917016967154,
      "loss": 0.8717,
      "step": 25298
    },
    {
      "epoch": 2.176632538931429,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006012652215023853,
      "loss": 1.2156,
      "step": 25299
    },
    {
      "epoch": 2.1767185752387506,
      "grad_norm": 0.890625,
      "learning_rate": 0.000601238741011874,
      "loss": 1.1939,
      "step": 25300
    },
    {
      "epoch": 2.1768046115460726,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0006012122602252591,
      "loss": 1.203,
      "step": 25301
    },
    {
      "epoch": 2.1768906478533943,
      "grad_norm": 0.96875,
      "learning_rate": 0.0006011857791426179,
      "loss": 0.8116,
      "step": 25302
    },
    {
      "epoch": 2.176976684160716,
      "grad_norm": 2.28125,
      "learning_rate": 0.0006011592977640282,
      "loss": 1.0165,
      "step": 25303
    },
    {
      "epoch": 2.1770627204680375,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000601132816089567,
      "loss": 1.8354,
      "step": 25304
    },
    {
      "epoch": 2.177148756775359,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006011063341193122,
      "loss": 1.996,
      "step": 25305
    },
    {
      "epoch": 2.1772347930826808,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006010798518533407,
      "loss": 1.144,
      "step": 25306
    },
    {
      "epoch": 2.1773208293900024,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006010533692917306,
      "loss": 0.9407,
      "step": 25307
    },
    {
      "epoch": 2.1774068656973244,
      "grad_norm": 1.015625,
      "learning_rate": 0.000601026886434559,
      "loss": 0.6779,
      "step": 25308
    },
    {
      "epoch": 2.177492902004646,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006010004032819032,
      "loss": 1.6607,
      "step": 25309
    },
    {
      "epoch": 2.1775789383119677,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006009739198338411,
      "loss": 0.8983,
      "step": 25310
    },
    {
      "epoch": 2.1776649746192893,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006009474360904498,
      "loss": 1.9219,
      "step": 25311
    },
    {
      "epoch": 2.177751010926611,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000600920952051807,
      "loss": 1.0864,
      "step": 25312
    },
    {
      "epoch": 2.1778370472339326,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0006008944677179899,
      "loss": 1.0541,
      "step": 25313
    },
    {
      "epoch": 2.1779230835412546,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0006008679830890761,
      "loss": 0.4562,
      "step": 25314
    },
    {
      "epoch": 2.1780091198485763,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0006008414981651432,
      "loss": 1.3813,
      "step": 25315
    },
    {
      "epoch": 2.178095156155898,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006008150129462685,
      "loss": 1.2809,
      "step": 25316
    },
    {
      "epoch": 2.1781811924632195,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0006007885274325294,
      "loss": 2.058,
      "step": 25317
    },
    {
      "epoch": 2.178267228770541,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0006007620416240036,
      "loss": 1.1723,
      "step": 25318
    },
    {
      "epoch": 2.1783532650778628,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0006007355555207683,
      "loss": 0.6532,
      "step": 25319
    },
    {
      "epoch": 2.1784393013851844,
      "grad_norm": 1.0,
      "learning_rate": 0.0006007090691229012,
      "loss": 1.1659,
      "step": 25320
    },
    {
      "epoch": 2.1785253376925064,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0006006825824304796,
      "loss": 1.5267,
      "step": 25321
    },
    {
      "epoch": 2.178611373999828,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000600656095443581,
      "loss": 1.3431,
      "step": 25322
    },
    {
      "epoch": 2.1786974103071497,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000600629608162283,
      "loss": 1.4834,
      "step": 25323
    },
    {
      "epoch": 2.1787834466144713,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006006031205866627,
      "loss": 1.1723,
      "step": 25324
    },
    {
      "epoch": 2.178869482921793,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0006005766327167981,
      "loss": 0.9361,
      "step": 25325
    },
    {
      "epoch": 2.1789555192291146,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0006005501445527665,
      "loss": 0.9208,
      "step": 25326
    },
    {
      "epoch": 2.179041555536436,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0006005236560946451,
      "loss": 0.9442,
      "step": 25327
    },
    {
      "epoch": 2.1791275918437583,
      "grad_norm": 1.25,
      "learning_rate": 0.0006004971673425116,
      "loss": 1.4023,
      "step": 25328
    },
    {
      "epoch": 2.17921362815108,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006004706782964433,
      "loss": 1.5443,
      "step": 25329
    },
    {
      "epoch": 2.1792996644584015,
      "grad_norm": 0.875,
      "learning_rate": 0.000600444188956518,
      "loss": 1.3115,
      "step": 25330
    },
    {
      "epoch": 2.179385700765723,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0006004176993228129,
      "loss": 1.4103,
      "step": 25331
    },
    {
      "epoch": 2.1794717370730448,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0006003912093954055,
      "loss": 1.7857,
      "step": 25332
    },
    {
      "epoch": 2.1795577733803664,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006003647191743734,
      "loss": 0.9639,
      "step": 25333
    },
    {
      "epoch": 2.179643809687688,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006003382286597941,
      "loss": 1.2878,
      "step": 25334
    },
    {
      "epoch": 2.17972984599501,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0006003117378517448,
      "loss": 1.06,
      "step": 25335
    },
    {
      "epoch": 2.1798158823023317,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006002852467503032,
      "loss": 0.9067,
      "step": 25336
    },
    {
      "epoch": 2.1799019186096533,
      "grad_norm": 1.53125,
      "learning_rate": 0.000600258755355547,
      "loss": 1.5691,
      "step": 25337
    },
    {
      "epoch": 2.179987954916975,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006002322636675533,
      "loss": 1.3908,
      "step": 25338
    },
    {
      "epoch": 2.1800739912242966,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0006002057716863997,
      "loss": 0.9557,
      "step": 25339
    },
    {
      "epoch": 2.180160027531618,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006001792794121638,
      "loss": 0.9055,
      "step": 25340
    },
    {
      "epoch": 2.18024606383894,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006001527868449229,
      "loss": 1.7349,
      "step": 25341
    },
    {
      "epoch": 2.180332100146262,
      "grad_norm": 1.8125,
      "learning_rate": 0.0006001262939847546,
      "loss": 1.4201,
      "step": 25342
    },
    {
      "epoch": 2.1804181364535835,
      "grad_norm": 1.234375,
      "learning_rate": 0.0006000998008317362,
      "loss": 0.883,
      "step": 25343
    },
    {
      "epoch": 2.180504172760905,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006000733073859456,
      "loss": 0.9834,
      "step": 25344
    },
    {
      "epoch": 2.1805902090682268,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00060004681364746,
      "loss": 2.2629,
      "step": 25345
    },
    {
      "epoch": 2.1806762453755484,
      "grad_norm": 1.078125,
      "learning_rate": 0.000600020319616357,
      "loss": 0.9606,
      "step": 25346
    },
    {
      "epoch": 2.18076228168287,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005999938252927137,
      "loss": 1.5804,
      "step": 25347
    },
    {
      "epoch": 2.1808483179901916,
      "grad_norm": 1.125,
      "learning_rate": 0.000599967330676608,
      "loss": 1.2798,
      "step": 25348
    },
    {
      "epoch": 2.1809343542975137,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005999408357681175,
      "loss": 1.0673,
      "step": 25349
    },
    {
      "epoch": 2.1810203906048353,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005999143405673195,
      "loss": 1.5968,
      "step": 25350
    },
    {
      "epoch": 2.181106426912157,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005998878450742912,
      "loss": 1.3575,
      "step": 25351
    },
    {
      "epoch": 2.1811924632194786,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005998613492891105,
      "loss": 1.0301,
      "step": 25352
    },
    {
      "epoch": 2.1812784995268,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005998348532118549,
      "loss": 1.0904,
      "step": 25353
    },
    {
      "epoch": 2.181364535834122,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005998083568426017,
      "loss": 1.1316,
      "step": 25354
    },
    {
      "epoch": 2.181450572141444,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005997818601814284,
      "loss": 1.2074,
      "step": 25355
    },
    {
      "epoch": 2.1815366084487655,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005997553632284127,
      "loss": 0.5821,
      "step": 25356
    },
    {
      "epoch": 2.181622644756087,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005997288659836319,
      "loss": 0.8963,
      "step": 25357
    },
    {
      "epoch": 2.1817086810634088,
      "grad_norm": 1.125,
      "learning_rate": 0.0005997023684471635,
      "loss": 0.5221,
      "step": 25358
    },
    {
      "epoch": 2.1817947173707304,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005996758706190852,
      "loss": 0.8769,
      "step": 25359
    },
    {
      "epoch": 2.181880753678052,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005996493724994744,
      "loss": 1.5935,
      "step": 25360
    },
    {
      "epoch": 2.1819667899853736,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005996228740884084,
      "loss": 1.2944,
      "step": 25361
    },
    {
      "epoch": 2.1820528262926957,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005995963753859649,
      "loss": 1.0529,
      "step": 25362
    },
    {
      "epoch": 2.1821388626000173,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005995698763922214,
      "loss": 1.2013,
      "step": 25363
    },
    {
      "epoch": 2.182224898907339,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005995433771072554,
      "loss": 0.7066,
      "step": 25364
    },
    {
      "epoch": 2.1823109352146606,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005995168775311442,
      "loss": 1.4704,
      "step": 25365
    },
    {
      "epoch": 2.182396971521982,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005994903776639657,
      "loss": 1.4133,
      "step": 25366
    },
    {
      "epoch": 2.182483007829304,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0005994638775057971,
      "loss": 2.0583,
      "step": 25367
    },
    {
      "epoch": 2.182569044136626,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005994373770567161,
      "loss": 0.7043,
      "step": 25368
    },
    {
      "epoch": 2.1826550804439475,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005994108763168001,
      "loss": 1.2994,
      "step": 25369
    },
    {
      "epoch": 2.182741116751269,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005993843752861265,
      "loss": 1.3181,
      "step": 25370
    },
    {
      "epoch": 2.1828271530585908,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005993578739647731,
      "loss": 1.5207,
      "step": 25371
    },
    {
      "epoch": 2.1829131893659124,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005993313723528171,
      "loss": 0.9441,
      "step": 25372
    },
    {
      "epoch": 2.182999225673234,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005993048704503361,
      "loss": 1.7086,
      "step": 25373
    },
    {
      "epoch": 2.1830852619805556,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005992783682574079,
      "loss": 0.9375,
      "step": 25374
    },
    {
      "epoch": 2.1831712982878777,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005992518657741096,
      "loss": 1.6869,
      "step": 25375
    },
    {
      "epoch": 2.1832573345951993,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000599225363000519,
      "loss": 1.47,
      "step": 25376
    },
    {
      "epoch": 2.183343370902521,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005991988599367134,
      "loss": 1.5074,
      "step": 25377
    },
    {
      "epoch": 2.1834294072098426,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005991723565827706,
      "loss": 1.4541,
      "step": 25378
    },
    {
      "epoch": 2.183515443517164,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005991458529387678,
      "loss": 1.5896,
      "step": 25379
    },
    {
      "epoch": 2.183601479824486,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005991193490047826,
      "loss": 1.4164,
      "step": 25380
    },
    {
      "epoch": 2.1836875161318074,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005990928447808928,
      "loss": 1.164,
      "step": 25381
    },
    {
      "epoch": 2.1837735524391295,
      "grad_norm": 1.25,
      "learning_rate": 0.0005990663402671755,
      "loss": 1.6472,
      "step": 25382
    },
    {
      "epoch": 2.183859588746451,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005990398354637085,
      "loss": 1.4828,
      "step": 25383
    },
    {
      "epoch": 2.1839456250537728,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005990133303705692,
      "loss": 1.2636,
      "step": 25384
    },
    {
      "epoch": 2.1840316613610944,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005989868249878352,
      "loss": 1.0874,
      "step": 25385
    },
    {
      "epoch": 2.184117697668416,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000598960319315584,
      "loss": 1.1611,
      "step": 25386
    },
    {
      "epoch": 2.1842037339757376,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005989338133538932,
      "loss": 0.8515,
      "step": 25387
    },
    {
      "epoch": 2.1842897702830593,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00059890730710284,
      "loss": 1.3758,
      "step": 25388
    },
    {
      "epoch": 2.1843758065903813,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005988808005625023,
      "loss": 0.9317,
      "step": 25389
    },
    {
      "epoch": 2.184461842897703,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005988542937329577,
      "loss": 1.1332,
      "step": 25390
    },
    {
      "epoch": 2.1845478792050246,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005988277866142833,
      "loss": 1.6825,
      "step": 25391
    },
    {
      "epoch": 2.184633915512346,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005988012792065568,
      "loss": 0.5514,
      "step": 25392
    },
    {
      "epoch": 2.184719951819668,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005987747715098558,
      "loss": 1.0039,
      "step": 25393
    },
    {
      "epoch": 2.1848059881269895,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005987482635242579,
      "loss": 0.835,
      "step": 25394
    },
    {
      "epoch": 2.184892024434311,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005987217552498404,
      "loss": 1.6112,
      "step": 25395
    },
    {
      "epoch": 2.184978060741633,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005986952466866811,
      "loss": 0.9725,
      "step": 25396
    },
    {
      "epoch": 2.1850640970489548,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005986687378348575,
      "loss": 0.4947,
      "step": 25397
    },
    {
      "epoch": 2.1851501333562764,
      "grad_norm": 0.71875,
      "learning_rate": 0.0005986422286944469,
      "loss": 0.4717,
      "step": 25398
    },
    {
      "epoch": 2.185236169663598,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005986157192655267,
      "loss": 1.3871,
      "step": 25399
    },
    {
      "epoch": 2.1853222059709196,
      "grad_norm": 1.5,
      "learning_rate": 0.0005985892095481751,
      "loss": 1.4573,
      "step": 25400
    },
    {
      "epoch": 2.1854082422782413,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005985626995424691,
      "loss": 1.2747,
      "step": 25401
    },
    {
      "epoch": 2.185494278585563,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005985361892484863,
      "loss": 1.3271,
      "step": 25402
    },
    {
      "epoch": 2.185580314892885,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005985096786663043,
      "loss": 1.2157,
      "step": 25403
    },
    {
      "epoch": 2.1856663512002066,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005984831677960006,
      "loss": 0.8888,
      "step": 25404
    },
    {
      "epoch": 2.185752387507528,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000598456656637653,
      "loss": 1.1692,
      "step": 25405
    },
    {
      "epoch": 2.18583842381485,
      "grad_norm": 1.484375,
      "learning_rate": 0.0005984301451913387,
      "loss": 0.9594,
      "step": 25406
    },
    {
      "epoch": 2.1859244601221715,
      "grad_norm": 1.375,
      "learning_rate": 0.0005984036334571353,
      "loss": 1.2784,
      "step": 25407
    },
    {
      "epoch": 2.186010496429493,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005983771214351206,
      "loss": 0.9203,
      "step": 25408
    },
    {
      "epoch": 2.1860965327368147,
      "grad_norm": 1.734375,
      "learning_rate": 0.0005983506091253719,
      "loss": 1.3068,
      "step": 25409
    },
    {
      "epoch": 2.1861825690441368,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0005983240965279668,
      "loss": 1.4145,
      "step": 25410
    },
    {
      "epoch": 2.1862686053514584,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005982975836429827,
      "loss": 1.5984,
      "step": 25411
    },
    {
      "epoch": 2.18635464165878,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005982710704704974,
      "loss": 1.7472,
      "step": 25412
    },
    {
      "epoch": 2.1864406779661016,
      "grad_norm": 1.125,
      "learning_rate": 0.0005982445570105883,
      "loss": 1.2926,
      "step": 25413
    },
    {
      "epoch": 2.1865267142734233,
      "grad_norm": 1.171875,
      "learning_rate": 0.000598218043263333,
      "loss": 0.8912,
      "step": 25414
    },
    {
      "epoch": 2.186612750580745,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005981915292288091,
      "loss": 1.0793,
      "step": 25415
    },
    {
      "epoch": 2.186698786888067,
      "grad_norm": 1.4375,
      "learning_rate": 0.000598165014907094,
      "loss": 1.2715,
      "step": 25416
    },
    {
      "epoch": 2.1867848231953886,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005981385002982653,
      "loss": 1.2041,
      "step": 25417
    },
    {
      "epoch": 2.18687085950271,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005981119854024005,
      "loss": 2.0504,
      "step": 25418
    },
    {
      "epoch": 2.186956895810032,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005980854702195774,
      "loss": 0.8226,
      "step": 25419
    },
    {
      "epoch": 2.1870429321173535,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005980589547498733,
      "loss": 1.1448,
      "step": 25420
    },
    {
      "epoch": 2.187128968424675,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005980324389933658,
      "loss": 1.1471,
      "step": 25421
    },
    {
      "epoch": 2.1872150047319967,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005980059229501323,
      "loss": 0.6402,
      "step": 25422
    },
    {
      "epoch": 2.1873010410393188,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005979794066202508,
      "loss": 1.9694,
      "step": 25423
    },
    {
      "epoch": 2.1873870773466404,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005979528900037986,
      "loss": 1.6599,
      "step": 25424
    },
    {
      "epoch": 2.187473113653962,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005979263731008532,
      "loss": 0.9078,
      "step": 25425
    },
    {
      "epoch": 2.1875591499612836,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005978998559114921,
      "loss": 0.923,
      "step": 25426
    },
    {
      "epoch": 2.1876451862686053,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000597873338435793,
      "loss": 1.3433,
      "step": 25427
    },
    {
      "epoch": 2.187731222575927,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005978468206738335,
      "loss": 1.0043,
      "step": 25428
    },
    {
      "epoch": 2.187817258883249,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005978203026256911,
      "loss": 1.0976,
      "step": 25429
    },
    {
      "epoch": 2.1879032951905706,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005977937842914433,
      "loss": 1.0884,
      "step": 25430
    },
    {
      "epoch": 2.187989331497892,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005977672656711677,
      "loss": 0.5931,
      "step": 25431
    },
    {
      "epoch": 2.188075367805214,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005977407467649418,
      "loss": 1.163,
      "step": 25432
    },
    {
      "epoch": 2.1881614041125355,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0005977142275728433,
      "loss": 2.0309,
      "step": 25433
    },
    {
      "epoch": 2.188247440419857,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005976877080949496,
      "loss": 0.8968,
      "step": 25434
    },
    {
      "epoch": 2.1883334767271787,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005976611883313386,
      "loss": 1.1489,
      "step": 25435
    },
    {
      "epoch": 2.1884195130345008,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005976346682820872,
      "loss": 1.0595,
      "step": 25436
    },
    {
      "epoch": 2.1885055493418224,
      "grad_norm": 2.140625,
      "learning_rate": 0.0005976081479472737,
      "loss": 1.3267,
      "step": 25437
    },
    {
      "epoch": 2.188591585649144,
      "grad_norm": 1.25,
      "learning_rate": 0.0005975816273269753,
      "loss": 0.7485,
      "step": 25438
    },
    {
      "epoch": 2.1886776219564656,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005975551064212696,
      "loss": 1.6424,
      "step": 25439
    },
    {
      "epoch": 2.1887636582637873,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005975285852302341,
      "loss": 1.2551,
      "step": 25440
    },
    {
      "epoch": 2.188849694571109,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005975020637539464,
      "loss": 1.2947,
      "step": 25441
    },
    {
      "epoch": 2.1889357308784305,
      "grad_norm": 1.25,
      "learning_rate": 0.0005974755419924843,
      "loss": 1.3695,
      "step": 25442
    },
    {
      "epoch": 2.1890217671857526,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005974490199459252,
      "loss": 0.6745,
      "step": 25443
    },
    {
      "epoch": 2.189107803493074,
      "grad_norm": 1.0,
      "learning_rate": 0.0005974224976143467,
      "loss": 1.008,
      "step": 25444
    },
    {
      "epoch": 2.189193839800396,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005973959749978262,
      "loss": 0.9601,
      "step": 25445
    },
    {
      "epoch": 2.1892798761077175,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005973694520964416,
      "loss": 0.7886,
      "step": 25446
    },
    {
      "epoch": 2.189365912415039,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00059734292891027,
      "loss": 0.8481,
      "step": 25447
    },
    {
      "epoch": 2.1894519487223607,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005973164054393895,
      "loss": 0.7457,
      "step": 25448
    },
    {
      "epoch": 2.1895379850296823,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005972898816838775,
      "loss": 0.7753,
      "step": 25449
    },
    {
      "epoch": 2.1896240213370044,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005972633576438114,
      "loss": 1.3977,
      "step": 25450
    },
    {
      "epoch": 2.189710057644326,
      "grad_norm": 1.125,
      "learning_rate": 0.0005972368333192687,
      "loss": 1.3321,
      "step": 25451
    },
    {
      "epoch": 2.1897960939516476,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005972103087103274,
      "loss": 1.2552,
      "step": 25452
    },
    {
      "epoch": 2.1898821302589693,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005971837838170648,
      "loss": 1.5567,
      "step": 25453
    },
    {
      "epoch": 2.189968166566291,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005971572586395586,
      "loss": 1.6319,
      "step": 25454
    },
    {
      "epoch": 2.1900542028736125,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005971307331778862,
      "loss": 0.9204,
      "step": 25455
    },
    {
      "epoch": 2.190140239180934,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005971042074321255,
      "loss": 1.9055,
      "step": 25456
    },
    {
      "epoch": 2.190226275488256,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005970776814023537,
      "loss": 1.0786,
      "step": 25457
    },
    {
      "epoch": 2.190312311795578,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005970511550886485,
      "loss": 1.8572,
      "step": 25458
    },
    {
      "epoch": 2.1903983481028995,
      "grad_norm": 1.0,
      "learning_rate": 0.0005970246284910877,
      "loss": 1.181,
      "step": 25459
    },
    {
      "epoch": 2.190484384410221,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005969981016097487,
      "loss": 1.4184,
      "step": 25460
    },
    {
      "epoch": 2.1905704207175427,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005969715744447091,
      "loss": 1.2053,
      "step": 25461
    },
    {
      "epoch": 2.1906564570248643,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005969450469960465,
      "loss": 1.9423,
      "step": 25462
    },
    {
      "epoch": 2.190742493332186,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005969185192638384,
      "loss": 1.1019,
      "step": 25463
    },
    {
      "epoch": 2.190828529639508,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005968919912481627,
      "loss": 1.7342,
      "step": 25464
    },
    {
      "epoch": 2.1909145659468297,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005968654629490966,
      "loss": 1.6653,
      "step": 25465
    },
    {
      "epoch": 2.1910006022541513,
      "grad_norm": 1.0625,
      "learning_rate": 0.000596838934366718,
      "loss": 1.2019,
      "step": 25466
    },
    {
      "epoch": 2.191086638561473,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005968124055011041,
      "loss": 1.1758,
      "step": 25467
    },
    {
      "epoch": 2.1911726748687945,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000596785876352333,
      "loss": 1.3933,
      "step": 25468
    },
    {
      "epoch": 2.191258711176116,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005967593469204819,
      "loss": 0.8304,
      "step": 25469
    },
    {
      "epoch": 2.191344747483438,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005967328172056286,
      "loss": 0.8537,
      "step": 25470
    },
    {
      "epoch": 2.19143078379076,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005967062872078506,
      "loss": 0.6365,
      "step": 25471
    },
    {
      "epoch": 2.1915168200980815,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005966797569272256,
      "loss": 1.0098,
      "step": 25472
    },
    {
      "epoch": 2.191602856405403,
      "grad_norm": 1.25,
      "learning_rate": 0.0005966532263638309,
      "loss": 1.4866,
      "step": 25473
    },
    {
      "epoch": 2.1916888927127247,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005966266955177444,
      "loss": 1.8478,
      "step": 25474
    },
    {
      "epoch": 2.1917749290200463,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005966001643890437,
      "loss": 0.8643,
      "step": 25475
    },
    {
      "epoch": 2.191860965327368,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005965736329778062,
      "loss": 1.7482,
      "step": 25476
    },
    {
      "epoch": 2.19194700163469,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005965471012841096,
      "loss": 1.1696,
      "step": 25477
    },
    {
      "epoch": 2.1920330379420117,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005965205693080315,
      "loss": 1.36,
      "step": 25478
    },
    {
      "epoch": 2.1921190742493333,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005964940370496496,
      "loss": 0.7996,
      "step": 25479
    },
    {
      "epoch": 2.192205110556655,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005964675045090414,
      "loss": 1.0478,
      "step": 25480
    },
    {
      "epoch": 2.1922911468639765,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005964409716862846,
      "loss": 0.7996,
      "step": 25481
    },
    {
      "epoch": 2.192377183171298,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005964144385814565,
      "loss": 0.9426,
      "step": 25482
    },
    {
      "epoch": 2.1924632194786198,
      "grad_norm": 0.953125,
      "learning_rate": 0.000596387905194635,
      "loss": 0.9558,
      "step": 25483
    },
    {
      "epoch": 2.192549255785942,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005963613715258977,
      "loss": 0.9189,
      "step": 25484
    },
    {
      "epoch": 2.1926352920932635,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000596334837575322,
      "loss": 1.1156,
      "step": 25485
    },
    {
      "epoch": 2.192721328400585,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005963083033429856,
      "loss": 0.8835,
      "step": 25486
    },
    {
      "epoch": 2.1928073647079067,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005962817688289665,
      "loss": 1.1734,
      "step": 25487
    },
    {
      "epoch": 2.1928934010152283,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005962552340333417,
      "loss": 1.3305,
      "step": 25488
    },
    {
      "epoch": 2.19297943732255,
      "grad_norm": 1.625,
      "learning_rate": 0.0005962286989561889,
      "loss": 0.8476,
      "step": 25489
    },
    {
      "epoch": 2.193065473629872,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005962021635975862,
      "loss": 1.4745,
      "step": 25490
    },
    {
      "epoch": 2.1931515099371937,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005961756279576107,
      "loss": 1.3363,
      "step": 25491
    },
    {
      "epoch": 2.1932375462445153,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005961490920363402,
      "loss": 1.1597,
      "step": 25492
    },
    {
      "epoch": 2.193323582551837,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005961225558338523,
      "loss": 1.4532,
      "step": 25493
    },
    {
      "epoch": 2.1934096188591585,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005960960193502246,
      "loss": 1.2848,
      "step": 25494
    },
    {
      "epoch": 2.19349565516648,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005960694825855348,
      "loss": 2.1328,
      "step": 25495
    },
    {
      "epoch": 2.1935816914738018,
      "grad_norm": 1.125,
      "learning_rate": 0.0005960429455398604,
      "loss": 0.8122,
      "step": 25496
    },
    {
      "epoch": 2.193667727781124,
      "grad_norm": 1.125,
      "learning_rate": 0.0005960164082132789,
      "loss": 1.2188,
      "step": 25497
    },
    {
      "epoch": 2.1937537640884455,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005959898706058684,
      "loss": 0.8791,
      "step": 25498
    },
    {
      "epoch": 2.193839800395767,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005959633327177061,
      "loss": 1.2753,
      "step": 25499
    },
    {
      "epoch": 2.1939258367030887,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005959367945488697,
      "loss": 0.4997,
      "step": 25500
    },
    {
      "epoch": 2.1940118730104103,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0005959102560994367,
      "loss": 0.7805,
      "step": 25501
    },
    {
      "epoch": 2.194097909317732,
      "grad_norm": 1.265625,
      "learning_rate": 0.000595883717369485,
      "loss": 1.9879,
      "step": 25502
    },
    {
      "epoch": 2.1941839456250536,
      "grad_norm": 1.234375,
      "learning_rate": 0.000595857178359092,
      "loss": 1.6415,
      "step": 25503
    },
    {
      "epoch": 2.1942699819323757,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005958306390683355,
      "loss": 1.0095,
      "step": 25504
    },
    {
      "epoch": 2.1943560182396973,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000595804099497293,
      "loss": 0.5291,
      "step": 25505
    },
    {
      "epoch": 2.194442054547019,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005957775596460421,
      "loss": 1.1824,
      "step": 25506
    },
    {
      "epoch": 2.1945280908543405,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005957510195146603,
      "loss": 1.5545,
      "step": 25507
    },
    {
      "epoch": 2.194614127161662,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005957244791032257,
      "loss": 1.6869,
      "step": 25508
    },
    {
      "epoch": 2.1947001634689838,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0005956979384118154,
      "loss": 1.8228,
      "step": 25509
    },
    {
      "epoch": 2.1947861997763054,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005956713974405074,
      "loss": 1.3352,
      "step": 25510
    },
    {
      "epoch": 2.1948722360836275,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000595644856189379,
      "loss": 1.4043,
      "step": 25511
    },
    {
      "epoch": 2.194958272390949,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005956183146585081,
      "loss": 0.9414,
      "step": 25512
    },
    {
      "epoch": 2.1950443086982707,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005955917728479723,
      "loss": 1.581,
      "step": 25513
    },
    {
      "epoch": 2.1951303450055923,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000595565230757849,
      "loss": 1.5663,
      "step": 25514
    },
    {
      "epoch": 2.195216381312914,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005955386883882158,
      "loss": 1.9598,
      "step": 25515
    },
    {
      "epoch": 2.1953024176202356,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005955121457391509,
      "loss": 0.9639,
      "step": 25516
    },
    {
      "epoch": 2.195388453927557,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0005954856028107314,
      "loss": 1.3389,
      "step": 25517
    },
    {
      "epoch": 2.1954744902348793,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005954590596030352,
      "loss": 1.4491,
      "step": 25518
    },
    {
      "epoch": 2.195560526542201,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005954325161161395,
      "loss": 0.8829,
      "step": 25519
    },
    {
      "epoch": 2.1956465628495225,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005954059723501226,
      "loss": 0.6937,
      "step": 25520
    },
    {
      "epoch": 2.195732599156844,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005953794283050616,
      "loss": 1.1372,
      "step": 25521
    },
    {
      "epoch": 2.195818635464166,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005953528839810345,
      "loss": 0.8049,
      "step": 25522
    },
    {
      "epoch": 2.1959046717714874,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005953263393781186,
      "loss": 0.85,
      "step": 25523
    },
    {
      "epoch": 2.195990708078809,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005952997944963917,
      "loss": 1.0249,
      "step": 25524
    },
    {
      "epoch": 2.196076744386131,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005952732493359316,
      "loss": 1.4725,
      "step": 25525
    },
    {
      "epoch": 2.1961627806934527,
      "grad_norm": 1.0,
      "learning_rate": 0.0005952467038968154,
      "loss": 0.9727,
      "step": 25526
    },
    {
      "epoch": 2.1962488170007743,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005952201581791214,
      "loss": 1.1155,
      "step": 25527
    },
    {
      "epoch": 2.196334853308096,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000595193612182927,
      "loss": 1.4414,
      "step": 25528
    },
    {
      "epoch": 2.1964208896154176,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005951670659083097,
      "loss": 1.0121,
      "step": 25529
    },
    {
      "epoch": 2.196506925922739,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005951405193553472,
      "loss": 1.0666,
      "step": 25530
    },
    {
      "epoch": 2.1965929622300613,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0005951139725241172,
      "loss": 1.4286,
      "step": 25531
    },
    {
      "epoch": 2.196678998537383,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005950874254146974,
      "loss": 0.714,
      "step": 25532
    },
    {
      "epoch": 2.1967650348447045,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005950608780271653,
      "loss": 0.7283,
      "step": 25533
    },
    {
      "epoch": 2.196851071152026,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005950343303615986,
      "loss": 0.7974,
      "step": 25534
    },
    {
      "epoch": 2.196937107459348,
      "grad_norm": 1.171875,
      "learning_rate": 0.000595007782418075,
      "loss": 1.4658,
      "step": 25535
    },
    {
      "epoch": 2.1970231437666694,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005949812341966723,
      "loss": 0.7342,
      "step": 25536
    },
    {
      "epoch": 2.197109180073991,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005949546856974679,
      "loss": 1.4554,
      "step": 25537
    },
    {
      "epoch": 2.197195216381313,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005949281369205392,
      "loss": 2.2888,
      "step": 25538
    },
    {
      "epoch": 2.1972812526886347,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005949015878659645,
      "loss": 0.9321,
      "step": 25539
    },
    {
      "epoch": 2.1973672889959563,
      "grad_norm": 1.0625,
      "learning_rate": 0.000594875038533821,
      "loss": 1.3725,
      "step": 25540
    },
    {
      "epoch": 2.197453325303278,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005948484889241864,
      "loss": 0.6528,
      "step": 25541
    },
    {
      "epoch": 2.1975393616105996,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005948219390371385,
      "loss": 1.0499,
      "step": 25542
    },
    {
      "epoch": 2.197625397917921,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005947953888727548,
      "loss": 1.0511,
      "step": 25543
    },
    {
      "epoch": 2.1977114342252433,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0005947688384311131,
      "loss": 1.3757,
      "step": 25544
    },
    {
      "epoch": 2.197797470532565,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005947422877122908,
      "loss": 1.2499,
      "step": 25545
    },
    {
      "epoch": 2.1978835068398865,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005947157367163661,
      "loss": 1.1266,
      "step": 25546
    },
    {
      "epoch": 2.197969543147208,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000594689185443416,
      "loss": 1.1109,
      "step": 25547
    },
    {
      "epoch": 2.19805557945453,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005946626338935184,
      "loss": 1.2652,
      "step": 25548
    },
    {
      "epoch": 2.1981416157618514,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005946360820667512,
      "loss": 1.1086,
      "step": 25549
    },
    {
      "epoch": 2.198227652069173,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005946095299631919,
      "loss": 1.0733,
      "step": 25550
    },
    {
      "epoch": 2.198313688376495,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005945829775829179,
      "loss": 0.928,
      "step": 25551
    },
    {
      "epoch": 2.1983997246838167,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005945564249260072,
      "loss": 1.1284,
      "step": 25552
    },
    {
      "epoch": 2.1984857609911383,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005945298719925374,
      "loss": 0.6126,
      "step": 25553
    },
    {
      "epoch": 2.19857179729846,
      "grad_norm": 2.5,
      "learning_rate": 0.000594503318782586,
      "loss": 1.9084,
      "step": 25554
    },
    {
      "epoch": 2.1986578336057816,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005944767652962308,
      "loss": 1.5548,
      "step": 25555
    },
    {
      "epoch": 2.198743869913103,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005944502115335495,
      "loss": 1.705,
      "step": 25556
    },
    {
      "epoch": 2.198829906220425,
      "grad_norm": 1.125,
      "learning_rate": 0.0005944236574946197,
      "loss": 1.2201,
      "step": 25557
    },
    {
      "epoch": 2.198915942527747,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005943971031795191,
      "loss": 1.6082,
      "step": 25558
    },
    {
      "epoch": 2.1990019788350685,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005943705485883253,
      "loss": 1.6757,
      "step": 25559
    },
    {
      "epoch": 2.19908801514239,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005943439937211159,
      "loss": 1.1808,
      "step": 25560
    },
    {
      "epoch": 2.199174051449712,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005943174385779687,
      "loss": 0.7261,
      "step": 25561
    },
    {
      "epoch": 2.1992600877570334,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005942908831589614,
      "loss": 1.615,
      "step": 25562
    },
    {
      "epoch": 2.199346124064355,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005942643274641716,
      "loss": 1.1975,
      "step": 25563
    },
    {
      "epoch": 2.1994321603716767,
      "grad_norm": 2.359375,
      "learning_rate": 0.000594237771493677,
      "loss": 1.3544,
      "step": 25564
    },
    {
      "epoch": 2.1995181966789987,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005942112152475552,
      "loss": 1.5032,
      "step": 25565
    },
    {
      "epoch": 2.1996042329863204,
      "grad_norm": 1.25,
      "learning_rate": 0.0005941846587258839,
      "loss": 1.1167,
      "step": 25566
    },
    {
      "epoch": 2.199690269293642,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005941581019287407,
      "loss": 0.5929,
      "step": 25567
    },
    {
      "epoch": 2.1997763056009636,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005941315448562036,
      "loss": 0.8004,
      "step": 25568
    },
    {
      "epoch": 2.1998623419082852,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005941049875083498,
      "loss": 1.4809,
      "step": 25569
    },
    {
      "epoch": 2.199948378215607,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005940784298852573,
      "loss": 1.1603,
      "step": 25570
    },
    {
      "epoch": 2.2000344145229285,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005940518719870037,
      "loss": 1.6141,
      "step": 25571
    },
    {
      "epoch": 2.2001204508302505,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005940253138136667,
      "loss": 1.3467,
      "step": 25572
    },
    {
      "epoch": 2.200206487137572,
      "grad_norm": 1.15625,
      "learning_rate": 0.000593998755365324,
      "loss": 1.5772,
      "step": 25573
    },
    {
      "epoch": 2.200292523444894,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005939721966420531,
      "loss": 1.3969,
      "step": 25574
    },
    {
      "epoch": 2.2003785597522154,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005939456376439318,
      "loss": 0.7615,
      "step": 25575
    },
    {
      "epoch": 2.200464596059537,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005939190783710379,
      "loss": 0.9765,
      "step": 25576
    },
    {
      "epoch": 2.2005506323668587,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000593892518823449,
      "loss": 1.0107,
      "step": 25577
    },
    {
      "epoch": 2.2006366686741803,
      "grad_norm": 1.25,
      "learning_rate": 0.0005938659590012426,
      "loss": 0.9812,
      "step": 25578
    },
    {
      "epoch": 2.2007227049815024,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005938393989044964,
      "loss": 1.0133,
      "step": 25579
    },
    {
      "epoch": 2.200808741288824,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005938128385332885,
      "loss": 1.1942,
      "step": 25580
    },
    {
      "epoch": 2.2008947775961456,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005937862778876962,
      "loss": 0.7632,
      "step": 25581
    },
    {
      "epoch": 2.2009808139034672,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005937597169677972,
      "loss": 1.2439,
      "step": 25582
    },
    {
      "epoch": 2.201066850210789,
      "grad_norm": 1.125,
      "learning_rate": 0.0005937331557736693,
      "loss": 1.4187,
      "step": 25583
    },
    {
      "epoch": 2.2011528865181105,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005937065943053902,
      "loss": 0.6613,
      "step": 25584
    },
    {
      "epoch": 2.201238922825432,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005936800325630375,
      "loss": 0.9172,
      "step": 25585
    },
    {
      "epoch": 2.201324959132754,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005936534705466888,
      "loss": 1.5141,
      "step": 25586
    },
    {
      "epoch": 2.201410995440076,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005936269082564221,
      "loss": 1.1498,
      "step": 25587
    },
    {
      "epoch": 2.2014970317473974,
      "grad_norm": 1.125,
      "learning_rate": 0.0005936003456923147,
      "loss": 2.0136,
      "step": 25588
    },
    {
      "epoch": 2.201583068054719,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005935737828544447,
      "loss": 1.4246,
      "step": 25589
    },
    {
      "epoch": 2.2016691043620407,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005935472197428895,
      "loss": 0.9156,
      "step": 25590
    },
    {
      "epoch": 2.2017551406693623,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005935206563577269,
      "loss": 1.5326,
      "step": 25591
    },
    {
      "epoch": 2.2018411769766844,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005934940926990345,
      "loss": 1.7189,
      "step": 25592
    },
    {
      "epoch": 2.201927213284006,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005934675287668903,
      "loss": 0.8724,
      "step": 25593
    },
    {
      "epoch": 2.2020132495913276,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005934409645613714,
      "loss": 0.8383,
      "step": 25594
    },
    {
      "epoch": 2.2020992858986492,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005934144000825561,
      "loss": 0.9584,
      "step": 25595
    },
    {
      "epoch": 2.202185322205971,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005933878353305218,
      "loss": 1.5137,
      "step": 25596
    },
    {
      "epoch": 2.2022713585132925,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005933612703053462,
      "loss": 1.6121,
      "step": 25597
    },
    {
      "epoch": 2.202357394820614,
      "grad_norm": 1.25,
      "learning_rate": 0.000593334705007107,
      "loss": 1.6443,
      "step": 25598
    },
    {
      "epoch": 2.202443431127936,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000593308139435882,
      "loss": 0.8851,
      "step": 25599
    },
    {
      "epoch": 2.202529467435258,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005932815735917489,
      "loss": 0.7271,
      "step": 25600
    },
    {
      "epoch": 2.2026155037425794,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005932550074747852,
      "loss": 1.1609,
      "step": 25601
    },
    {
      "epoch": 2.202701540049901,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005932284410850689,
      "loss": 0.9059,
      "step": 25602
    },
    {
      "epoch": 2.2027875763572227,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005932018744226775,
      "loss": 1.2289,
      "step": 25603
    },
    {
      "epoch": 2.2028736126645443,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005931753074876886,
      "loss": 0.693,
      "step": 25604
    },
    {
      "epoch": 2.2029596489718664,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005931487402801802,
      "loss": 1.2443,
      "step": 25605
    },
    {
      "epoch": 2.203045685279188,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005931221728002297,
      "loss": 1.0061,
      "step": 25606
    },
    {
      "epoch": 2.2031317215865096,
      "grad_norm": 1.0,
      "learning_rate": 0.0005930956050479151,
      "loss": 0.7739,
      "step": 25607
    },
    {
      "epoch": 2.2032177578938312,
      "grad_norm": 1.0,
      "learning_rate": 0.000593069037023314,
      "loss": 1.5438,
      "step": 25608
    },
    {
      "epoch": 2.203303794201153,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005930424687265039,
      "loss": 0.9417,
      "step": 25609
    },
    {
      "epoch": 2.2033898305084745,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005930159001575628,
      "loss": 1.1679,
      "step": 25610
    },
    {
      "epoch": 2.203475866815796,
      "grad_norm": 1.46875,
      "learning_rate": 0.0005929893313165683,
      "loss": 1.543,
      "step": 25611
    },
    {
      "epoch": 2.203561903123118,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005929627622035979,
      "loss": 0.3965,
      "step": 25612
    },
    {
      "epoch": 2.20364793943044,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005929361928187297,
      "loss": 0.8429,
      "step": 25613
    },
    {
      "epoch": 2.2037339757377614,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005929096231620413,
      "loss": 2.5089,
      "step": 25614
    },
    {
      "epoch": 2.203820012045083,
      "grad_norm": 1.046875,
      "learning_rate": 0.00059288305323361,
      "loss": 1.1165,
      "step": 25615
    },
    {
      "epoch": 2.2039060483524047,
      "grad_norm": 1.0625,
      "learning_rate": 0.000592856483033514,
      "loss": 1.6282,
      "step": 25616
    },
    {
      "epoch": 2.2039920846597263,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005928299125618308,
      "loss": 1.8375,
      "step": 25617
    },
    {
      "epoch": 2.204078120967048,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005928033418186384,
      "loss": 1.0871,
      "step": 25618
    },
    {
      "epoch": 2.20416415727437,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005927767708040141,
      "loss": 1.1622,
      "step": 25619
    },
    {
      "epoch": 2.2042501935816916,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005927501995180356,
      "loss": 1.0173,
      "step": 25620
    },
    {
      "epoch": 2.2043362298890132,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000592723627960781,
      "loss": 1.3412,
      "step": 25621
    },
    {
      "epoch": 2.204422266196335,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000592697056132328,
      "loss": 1.5652,
      "step": 25622
    },
    {
      "epoch": 2.2045083025036565,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005926704840327538,
      "loss": 1.4956,
      "step": 25623
    },
    {
      "epoch": 2.204594338810978,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005926439116621366,
      "loss": 1.7097,
      "step": 25624
    },
    {
      "epoch": 2.2046803751182997,
      "grad_norm": 1.5,
      "learning_rate": 0.000592617339020554,
      "loss": 1.5374,
      "step": 25625
    },
    {
      "epoch": 2.204766411425622,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005925907661080836,
      "loss": 0.73,
      "step": 25626
    },
    {
      "epoch": 2.2048524477329434,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005925641929248033,
      "loss": 1.5199,
      "step": 25627
    },
    {
      "epoch": 2.204938484040265,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005925376194707906,
      "loss": 1.1736,
      "step": 25628
    },
    {
      "epoch": 2.2050245203475867,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005925110457461236,
      "loss": 1.2652,
      "step": 25629
    },
    {
      "epoch": 2.2051105566549083,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005924844717508797,
      "loss": 1.0798,
      "step": 25630
    },
    {
      "epoch": 2.20519659296223,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005924578974851367,
      "loss": 1.5238,
      "step": 25631
    },
    {
      "epoch": 2.2052826292695515,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005924313229489723,
      "loss": 1.8168,
      "step": 25632
    },
    {
      "epoch": 2.2053686655768736,
      "grad_norm": 1.125,
      "learning_rate": 0.0005924047481424643,
      "loss": 0.7745,
      "step": 25633
    },
    {
      "epoch": 2.2054547018841952,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005923781730656904,
      "loss": 1.5564,
      "step": 25634
    },
    {
      "epoch": 2.205540738191517,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005923515977187281,
      "loss": 1.6164,
      "step": 25635
    },
    {
      "epoch": 2.2056267744988385,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005923250221016556,
      "loss": 0.82,
      "step": 25636
    },
    {
      "epoch": 2.20571281080616,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005922984462145503,
      "loss": 0.9326,
      "step": 25637
    },
    {
      "epoch": 2.2057988471134817,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005922718700574898,
      "loss": 1.4588,
      "step": 25638
    },
    {
      "epoch": 2.2058848834208034,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005922452936305523,
      "loss": 1.3309,
      "step": 25639
    },
    {
      "epoch": 2.2059709197281254,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005922187169338152,
      "loss": 0.8885,
      "step": 25640
    },
    {
      "epoch": 2.206056956035447,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005921921399673562,
      "loss": 1.8999,
      "step": 25641
    },
    {
      "epoch": 2.2061429923427687,
      "grad_norm": 1.234375,
      "learning_rate": 0.000592165562731253,
      "loss": 1.6215,
      "step": 25642
    },
    {
      "epoch": 2.2062290286500903,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005921389852255836,
      "loss": 1.0548,
      "step": 25643
    },
    {
      "epoch": 2.206315064957412,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005921124074504257,
      "loss": 1.01,
      "step": 25644
    },
    {
      "epoch": 2.2064011012647335,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005920858294058568,
      "loss": 2.3742,
      "step": 25645
    },
    {
      "epoch": 2.2064871375720556,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005920592510919548,
      "loss": 0.8851,
      "step": 25646
    },
    {
      "epoch": 2.2065731738793772,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005920326725087975,
      "loss": 0.8313,
      "step": 25647
    },
    {
      "epoch": 2.206659210186699,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005920060936564623,
      "loss": 1.1084,
      "step": 25648
    },
    {
      "epoch": 2.2067452464940205,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005919795145350273,
      "loss": 0.9291,
      "step": 25649
    },
    {
      "epoch": 2.206831282801342,
      "grad_norm": 1.03125,
      "learning_rate": 0.00059195293514457,
      "loss": 1.5817,
      "step": 25650
    },
    {
      "epoch": 2.2069173191086637,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005919263554851684,
      "loss": 1.1113,
      "step": 25651
    },
    {
      "epoch": 2.2070033554159854,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005918997755569001,
      "loss": 1.0153,
      "step": 25652
    },
    {
      "epoch": 2.2070893917233074,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005918731953598426,
      "loss": 0.9774,
      "step": 25653
    },
    {
      "epoch": 2.207175428030629,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005918466148940741,
      "loss": 0.8662,
      "step": 25654
    },
    {
      "epoch": 2.2072614643379507,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005918200341596721,
      "loss": 0.9946,
      "step": 25655
    },
    {
      "epoch": 2.2073475006452723,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005917934531567143,
      "loss": 1.0678,
      "step": 25656
    },
    {
      "epoch": 2.207433536952594,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005917668718852785,
      "loss": 1.723,
      "step": 25657
    },
    {
      "epoch": 2.2075195732599155,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005917402903454425,
      "loss": 1.5111,
      "step": 25658
    },
    {
      "epoch": 2.207605609567237,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005917137085372839,
      "loss": 1.5894,
      "step": 25659
    },
    {
      "epoch": 2.2076916458745592,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005916871264608805,
      "loss": 1.7941,
      "step": 25660
    },
    {
      "epoch": 2.207777682181881,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005916605441163102,
      "loss": 1.3843,
      "step": 25661
    },
    {
      "epoch": 2.2078637184892025,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005916339615036506,
      "loss": 1.0951,
      "step": 25662
    },
    {
      "epoch": 2.207949754796524,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005916073786229794,
      "loss": 1.5142,
      "step": 25663
    },
    {
      "epoch": 2.2080357911038457,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005915807954743746,
      "loss": 1.2475,
      "step": 25664
    },
    {
      "epoch": 2.2081218274111674,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0005915542120579138,
      "loss": 0.6822,
      "step": 25665
    },
    {
      "epoch": 2.2082078637184894,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005915276283736746,
      "loss": 1.4642,
      "step": 25666
    },
    {
      "epoch": 2.208293900025811,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005915010444217349,
      "loss": 1.1823,
      "step": 25667
    },
    {
      "epoch": 2.2083799363331327,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005914744602021724,
      "loss": 1.3251,
      "step": 25668
    },
    {
      "epoch": 2.2084659726404543,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000591447875715065,
      "loss": 1.1375,
      "step": 25669
    },
    {
      "epoch": 2.208552008947776,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005914212909604904,
      "loss": 1.5189,
      "step": 25670
    },
    {
      "epoch": 2.2086380452550975,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005913947059385261,
      "loss": 1.4052,
      "step": 25671
    },
    {
      "epoch": 2.208724081562419,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005913681206492502,
      "loss": 1.2865,
      "step": 25672
    },
    {
      "epoch": 2.2088101178697412,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005913415350927404,
      "loss": 0.7794,
      "step": 25673
    },
    {
      "epoch": 2.208896154177063,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005913149492690743,
      "loss": 1.6192,
      "step": 25674
    },
    {
      "epoch": 2.2089821904843845,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005912883631783297,
      "loss": 1.4745,
      "step": 25675
    },
    {
      "epoch": 2.209068226791706,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005912617768205844,
      "loss": 0.9198,
      "step": 25676
    },
    {
      "epoch": 2.2091542630990277,
      "grad_norm": 1.046875,
      "learning_rate": 0.000591235190195916,
      "loss": 1.616,
      "step": 25677
    },
    {
      "epoch": 2.2092402994063494,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005912086033044027,
      "loss": 1.8196,
      "step": 25678
    },
    {
      "epoch": 2.209326335713671,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005911820161461218,
      "loss": 1.4693,
      "step": 25679
    },
    {
      "epoch": 2.209412372020993,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005911554287211511,
      "loss": 1.2105,
      "step": 25680
    },
    {
      "epoch": 2.2094984083283147,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005911288410295688,
      "loss": 1.562,
      "step": 25681
    },
    {
      "epoch": 2.2095844446356363,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005911022530714523,
      "loss": 1.6529,
      "step": 25682
    },
    {
      "epoch": 2.209670480942958,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005910756648468792,
      "loss": 1.5446,
      "step": 25683
    },
    {
      "epoch": 2.2097565172502796,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005910490763559277,
      "loss": 1.4051,
      "step": 25684
    },
    {
      "epoch": 2.209842553557601,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0005910224875986754,
      "loss": 0.5135,
      "step": 25685
    },
    {
      "epoch": 2.209928589864923,
      "grad_norm": 1.0,
      "learning_rate": 0.0005909958985752,
      "loss": 1.8724,
      "step": 25686
    },
    {
      "epoch": 2.210014626172245,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005909693092855791,
      "loss": 1.028,
      "step": 25687
    },
    {
      "epoch": 2.2101006624795665,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005909427197298909,
      "loss": 1.1583,
      "step": 25688
    },
    {
      "epoch": 2.210186698786888,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005909161299082129,
      "loss": 1.33,
      "step": 25689
    },
    {
      "epoch": 2.2102727350942097,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005908895398206229,
      "loss": 0.7599,
      "step": 25690
    },
    {
      "epoch": 2.2103587714015314,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005908629494671984,
      "loss": 0.7302,
      "step": 25691
    },
    {
      "epoch": 2.210444807708853,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005908363588480178,
      "loss": 0.6605,
      "step": 25692
    },
    {
      "epoch": 2.2105308440161746,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005908097679631583,
      "loss": 1.4769,
      "step": 25693
    },
    {
      "epoch": 2.2106168803234967,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005907831768126979,
      "loss": 1.1526,
      "step": 25694
    },
    {
      "epoch": 2.2107029166308183,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005907565853967145,
      "loss": 1.3141,
      "step": 25695
    },
    {
      "epoch": 2.21078895293814,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005907299937152856,
      "loss": 1.4083,
      "step": 25696
    },
    {
      "epoch": 2.2108749892454616,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005907034017684894,
      "loss": 1.2263,
      "step": 25697
    },
    {
      "epoch": 2.210961025552783,
      "grad_norm": 1.6640625,
      "learning_rate": 0.000590676809556403,
      "loss": 1.5206,
      "step": 25698
    },
    {
      "epoch": 2.211047061860105,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005906502170791048,
      "loss": 0.9111,
      "step": 25699
    },
    {
      "epoch": 2.2111330981674264,
      "grad_norm": 0.75,
      "learning_rate": 0.0005906236243366723,
      "loss": 0.7201,
      "step": 25700
    },
    {
      "epoch": 2.2112191344747485,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005905970313291833,
      "loss": 0.9005,
      "step": 25701
    },
    {
      "epoch": 2.21130517078207,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005905704380567155,
      "loss": 1.0283,
      "step": 25702
    },
    {
      "epoch": 2.2113912070893917,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005905438445193469,
      "loss": 1.2456,
      "step": 25703
    },
    {
      "epoch": 2.2114772433967134,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005905172507171552,
      "loss": 0.9032,
      "step": 25704
    },
    {
      "epoch": 2.211563279704035,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005904906566502182,
      "loss": 0.9455,
      "step": 25705
    },
    {
      "epoch": 2.2116493160113566,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005904640623186133,
      "loss": 1.6006,
      "step": 25706
    },
    {
      "epoch": 2.2117353523186787,
      "grad_norm": 1.78125,
      "learning_rate": 0.000590437467722419,
      "loss": 1.8506,
      "step": 25707
    },
    {
      "epoch": 2.2118213886260003,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005904108728617126,
      "loss": 1.4075,
      "step": 25708
    },
    {
      "epoch": 2.211907424933322,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005903842777365718,
      "loss": 0.6787,
      "step": 25709
    },
    {
      "epoch": 2.2119934612406436,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005903576823470748,
      "loss": 2.1355,
      "step": 25710
    },
    {
      "epoch": 2.212079497547965,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005903310866932991,
      "loss": 1.4423,
      "step": 25711
    },
    {
      "epoch": 2.212165533855287,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005903044907753224,
      "loss": 1.5302,
      "step": 25712
    },
    {
      "epoch": 2.2122515701626084,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005902778945932227,
      "loss": 1.3755,
      "step": 25713
    },
    {
      "epoch": 2.2123376064699305,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005902512981470777,
      "loss": 1.6547,
      "step": 25714
    },
    {
      "epoch": 2.212423642777252,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005902247014369652,
      "loss": 0.5204,
      "step": 25715
    },
    {
      "epoch": 2.2125096790845737,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000590198104462963,
      "loss": 1.5476,
      "step": 25716
    },
    {
      "epoch": 2.2125957153918954,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005901715072251489,
      "loss": 1.2531,
      "step": 25717
    },
    {
      "epoch": 2.212681751699217,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005901449097236008,
      "loss": 1.1708,
      "step": 25718
    },
    {
      "epoch": 2.2127677880065386,
      "grad_norm": 1.0,
      "learning_rate": 0.0005901183119583962,
      "loss": 1.0723,
      "step": 25719
    },
    {
      "epoch": 2.2128538243138607,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005900917139296133,
      "loss": 0.6791,
      "step": 25720
    },
    {
      "epoch": 2.2129398606211823,
      "grad_norm": 1.0,
      "learning_rate": 0.0005900651156373293,
      "loss": 1.7517,
      "step": 25721
    },
    {
      "epoch": 2.213025896928504,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005900385170816226,
      "loss": 0.9494,
      "step": 25722
    },
    {
      "epoch": 2.2131119332358256,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005900119182625707,
      "loss": 0.6569,
      "step": 25723
    },
    {
      "epoch": 2.213197969543147,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005899853191802515,
      "loss": 0.8273,
      "step": 25724
    },
    {
      "epoch": 2.213284005850469,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005899587198347426,
      "loss": 1.4172,
      "step": 25725
    },
    {
      "epoch": 2.2133700421577904,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005899321202261221,
      "loss": 1.4358,
      "step": 25726
    },
    {
      "epoch": 2.2134560784651125,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005899055203544676,
      "loss": 1.431,
      "step": 25727
    },
    {
      "epoch": 2.213542114772434,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005898789202198568,
      "loss": 0.9977,
      "step": 25728
    },
    {
      "epoch": 2.2136281510797557,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005898523198223678,
      "loss": 1.2818,
      "step": 25729
    },
    {
      "epoch": 2.2137141873870774,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005898257191620782,
      "loss": 1.6863,
      "step": 25730
    },
    {
      "epoch": 2.213800223694399,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005897991182390659,
      "loss": 2.028,
      "step": 25731
    },
    {
      "epoch": 2.2138862600017206,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005897725170534084,
      "loss": 1.1895,
      "step": 25732
    },
    {
      "epoch": 2.2139722963090422,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005897459156051839,
      "loss": 1.577,
      "step": 25733
    },
    {
      "epoch": 2.2140583326163643,
      "grad_norm": 77.5,
      "learning_rate": 0.0005897193138944701,
      "loss": 1.3177,
      "step": 25734
    },
    {
      "epoch": 2.214144368923686,
      "grad_norm": 1.6875,
      "learning_rate": 0.0005896927119213446,
      "loss": 1.4301,
      "step": 25735
    },
    {
      "epoch": 2.2142304052310076,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005896661096858856,
      "loss": 1.8631,
      "step": 25736
    },
    {
      "epoch": 2.214316441538329,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005896395071881705,
      "loss": 2.0337,
      "step": 25737
    },
    {
      "epoch": 2.214402477845651,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005896129044282773,
      "loss": 2.0561,
      "step": 25738
    },
    {
      "epoch": 2.2144885141529724,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005895863014062838,
      "loss": 1.1346,
      "step": 25739
    },
    {
      "epoch": 2.214574550460294,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005895596981222678,
      "loss": 0.6378,
      "step": 25740
    },
    {
      "epoch": 2.214660586767616,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005895330945763071,
      "loss": 0.7427,
      "step": 25741
    },
    {
      "epoch": 2.2147466230749377,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005895064907684796,
      "loss": 1.4369,
      "step": 25742
    },
    {
      "epoch": 2.2148326593822594,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005894798866988628,
      "loss": 0.5547,
      "step": 25743
    },
    {
      "epoch": 2.214918695689581,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005894532823675349,
      "loss": 1.301,
      "step": 25744
    },
    {
      "epoch": 2.2150047319969026,
      "grad_norm": 1.484375,
      "learning_rate": 0.0005894266777745735,
      "loss": 1.353,
      "step": 25745
    },
    {
      "epoch": 2.2150907683042242,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005894000729200566,
      "loss": 0.9476,
      "step": 25746
    },
    {
      "epoch": 2.215176804611546,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005893734678040617,
      "loss": 0.9537,
      "step": 25747
    },
    {
      "epoch": 2.215262840918868,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005893468624266669,
      "loss": 1.3798,
      "step": 25748
    },
    {
      "epoch": 2.2153488772261896,
      "grad_norm": 1.65625,
      "learning_rate": 0.0005893202567879499,
      "loss": 0.8776,
      "step": 25749
    },
    {
      "epoch": 2.215434913533511,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005892936508879883,
      "loss": 1.1286,
      "step": 25750
    },
    {
      "epoch": 2.215520949840833,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005892670447268603,
      "loss": 1.1928,
      "step": 25751
    },
    {
      "epoch": 2.2156069861481544,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005892404383046437,
      "loss": 1.496,
      "step": 25752
    },
    {
      "epoch": 2.215693022455476,
      "grad_norm": 0.984375,
      "learning_rate": 0.000589213831621416,
      "loss": 1.7906,
      "step": 25753
    },
    {
      "epoch": 2.2157790587627977,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005891872246772551,
      "loss": 1.8977,
      "step": 25754
    },
    {
      "epoch": 2.2158650950701198,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005891606174722391,
      "loss": 1.6744,
      "step": 25755
    },
    {
      "epoch": 2.2159511313774414,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005891340100064455,
      "loss": 1.9648,
      "step": 25756
    },
    {
      "epoch": 2.216037167684763,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005891074022799524,
      "loss": 1.395,
      "step": 25757
    },
    {
      "epoch": 2.2161232039920846,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005890807942928373,
      "loss": 0.8573,
      "step": 25758
    },
    {
      "epoch": 2.2162092402994062,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005890541860451784,
      "loss": 0.8921,
      "step": 25759
    },
    {
      "epoch": 2.216295276606728,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005890275775370532,
      "loss": 1.6819,
      "step": 25760
    },
    {
      "epoch": 2.2163813129140495,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005890009687685398,
      "loss": 1.617,
      "step": 25761
    },
    {
      "epoch": 2.2164673492213716,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005889743597397157,
      "loss": 1.1827,
      "step": 25762
    },
    {
      "epoch": 2.216553385528693,
      "grad_norm": 1.15625,
      "learning_rate": 0.000588947750450659,
      "loss": 1.2436,
      "step": 25763
    },
    {
      "epoch": 2.216639421836015,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005889211409014475,
      "loss": 0.8289,
      "step": 25764
    },
    {
      "epoch": 2.2167254581433364,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005888945310921586,
      "loss": 1.3715,
      "step": 25765
    },
    {
      "epoch": 2.216811494450658,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005888679210228709,
      "loss": 0.8923,
      "step": 25766
    },
    {
      "epoch": 2.2168975307579797,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005888413106936617,
      "loss": 1.3146,
      "step": 25767
    },
    {
      "epoch": 2.2169835670653018,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000588814700104609,
      "loss": 0.8538,
      "step": 25768
    },
    {
      "epoch": 2.2170696033726234,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005887880892557903,
      "loss": 1.1832,
      "step": 25769
    },
    {
      "epoch": 2.217155639679945,
      "grad_norm": 1.125,
      "learning_rate": 0.0005887614781472838,
      "loss": 1.4593,
      "step": 25770
    },
    {
      "epoch": 2.2172416759872666,
      "grad_norm": 1.0,
      "learning_rate": 0.0005887348667791675,
      "loss": 0.7757,
      "step": 25771
    },
    {
      "epoch": 2.2173277122945882,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005887082551515186,
      "loss": 1.4064,
      "step": 25772
    },
    {
      "epoch": 2.21741374860191,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005886816432644154,
      "loss": 2.126,
      "step": 25773
    },
    {
      "epoch": 2.2174997849092315,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005886550311179358,
      "loss": 0.632,
      "step": 25774
    },
    {
      "epoch": 2.2175858212165536,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005886284187121575,
      "loss": 1.5653,
      "step": 25775
    },
    {
      "epoch": 2.217671857523875,
      "grad_norm": 1.5625,
      "learning_rate": 0.0005886018060471583,
      "loss": 1.0822,
      "step": 25776
    },
    {
      "epoch": 2.217757893831197,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005885751931230159,
      "loss": 1.3975,
      "step": 25777
    },
    {
      "epoch": 2.2178439301385184,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005885485799398084,
      "loss": 0.9085,
      "step": 25778
    },
    {
      "epoch": 2.21792996644584,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005885219664976134,
      "loss": 1.4335,
      "step": 25779
    },
    {
      "epoch": 2.2180160027531617,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000588495352796509,
      "loss": 1.137,
      "step": 25780
    },
    {
      "epoch": 2.2181020390604838,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005884687388365728,
      "loss": 0.784,
      "step": 25781
    },
    {
      "epoch": 2.2181880753678054,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000588442124617883,
      "loss": 1.1423,
      "step": 25782
    },
    {
      "epoch": 2.218274111675127,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005884155101405169,
      "loss": 0.6902,
      "step": 25783
    },
    {
      "epoch": 2.2183601479824486,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005883888954045526,
      "loss": 1.2951,
      "step": 25784
    },
    {
      "epoch": 2.2184461842897703,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005883622804100683,
      "loss": 0.7233,
      "step": 25785
    },
    {
      "epoch": 2.218532220597092,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005883356651571412,
      "loss": 1.0931,
      "step": 25786
    },
    {
      "epoch": 2.2186182569044135,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005883090496458497,
      "loss": 1.4547,
      "step": 25787
    },
    {
      "epoch": 2.2187042932117356,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005882824338762711,
      "loss": 0.7068,
      "step": 25788
    },
    {
      "epoch": 2.218790329519057,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005882558178484838,
      "loss": 0.8395,
      "step": 25789
    },
    {
      "epoch": 2.218876365826379,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005882292015625654,
      "loss": 1.4015,
      "step": 25790
    },
    {
      "epoch": 2.2189624021337004,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005882025850185936,
      "loss": 0.9857,
      "step": 25791
    },
    {
      "epoch": 2.219048438441022,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005881759682166463,
      "loss": 1.3518,
      "step": 25792
    },
    {
      "epoch": 2.2191344747483437,
      "grad_norm": 1.0,
      "learning_rate": 0.0005881493511568017,
      "loss": 1.177,
      "step": 25793
    },
    {
      "epoch": 2.2192205110556653,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005881227338391373,
      "loss": 1.2273,
      "step": 25794
    },
    {
      "epoch": 2.2193065473629874,
      "grad_norm": 0.984375,
      "learning_rate": 0.000588096116263731,
      "loss": 1.2857,
      "step": 25795
    },
    {
      "epoch": 2.219392583670309,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005880694984306607,
      "loss": 1.1565,
      "step": 25796
    },
    {
      "epoch": 2.2194786199776306,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005880428803400043,
      "loss": 1.2778,
      "step": 25797
    },
    {
      "epoch": 2.2195646562849523,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0005880162619918397,
      "loss": 1.616,
      "step": 25798
    },
    {
      "epoch": 2.219650692592274,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005879896433862444,
      "loss": 1.206,
      "step": 25799
    },
    {
      "epoch": 2.2197367288995955,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005879630245232966,
      "loss": 1.5628,
      "step": 25800
    },
    {
      "epoch": 2.219822765206917,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005879364054030741,
      "loss": 1.109,
      "step": 25801
    },
    {
      "epoch": 2.219908801514239,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005879097860256548,
      "loss": 1.1025,
      "step": 25802
    },
    {
      "epoch": 2.219994837821561,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005878831663911163,
      "loss": 0.7875,
      "step": 25803
    },
    {
      "epoch": 2.2200808741288824,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005878565464995368,
      "loss": 1.7746,
      "step": 25804
    },
    {
      "epoch": 2.220166910436204,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005878299263509939,
      "loss": 0.7478,
      "step": 25805
    },
    {
      "epoch": 2.2202529467435257,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005878033059455655,
      "loss": 1.1271,
      "step": 25806
    },
    {
      "epoch": 2.2203389830508473,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005877766852833295,
      "loss": 1.0983,
      "step": 25807
    },
    {
      "epoch": 2.220425019358169,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005877500643643639,
      "loss": 1.7395,
      "step": 25808
    },
    {
      "epoch": 2.220511055665491,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005877234431887463,
      "loss": 1.2937,
      "step": 25809
    },
    {
      "epoch": 2.2205970919728126,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005876968217565548,
      "loss": 0.7748,
      "step": 25810
    },
    {
      "epoch": 2.2206831282801343,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000587670200067867,
      "loss": 1.6731,
      "step": 25811
    },
    {
      "epoch": 2.220769164587456,
      "grad_norm": 0.78125,
      "learning_rate": 0.000587643578122761,
      "loss": 0.8017,
      "step": 25812
    },
    {
      "epoch": 2.2208552008947775,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005876169559213147,
      "loss": 1.1644,
      "step": 25813
    },
    {
      "epoch": 2.220941237202099,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005875903334636056,
      "loss": 1.0065,
      "step": 25814
    },
    {
      "epoch": 2.2210272735094208,
      "grad_norm": 1.046875,
      "learning_rate": 0.000587563710749712,
      "loss": 1.0967,
      "step": 25815
    },
    {
      "epoch": 2.221113309816743,
      "grad_norm": 1.625,
      "learning_rate": 0.0005875370877797115,
      "loss": 2.3093,
      "step": 25816
    },
    {
      "epoch": 2.2211993461240644,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005875104645536822,
      "loss": 1.4232,
      "step": 25817
    },
    {
      "epoch": 2.221285382431386,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005874838410717016,
      "loss": 1.5106,
      "step": 25818
    },
    {
      "epoch": 2.2213714187387077,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005874572173338479,
      "loss": 0.9288,
      "step": 25819
    },
    {
      "epoch": 2.2214574550460293,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005874305933401988,
      "loss": 1.0476,
      "step": 25820
    },
    {
      "epoch": 2.221543491353351,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005874039690908321,
      "loss": 1.016,
      "step": 25821
    },
    {
      "epoch": 2.221629527660673,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000587377344585826,
      "loss": 2.0098,
      "step": 25822
    },
    {
      "epoch": 2.2217155639679946,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0005873507198252581,
      "loss": 2.0355,
      "step": 25823
    },
    {
      "epoch": 2.2218016002753163,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005873240948092063,
      "loss": 0.7154,
      "step": 25824
    },
    {
      "epoch": 2.221887636582638,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005872974695377484,
      "loss": 1.3985,
      "step": 25825
    },
    {
      "epoch": 2.2219736728899595,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005872708440109626,
      "loss": 0.657,
      "step": 25826
    },
    {
      "epoch": 2.222059709197281,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005872442182289265,
      "loss": 1.3577,
      "step": 25827
    },
    {
      "epoch": 2.2221457455046028,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005872175921917179,
      "loss": 1.1335,
      "step": 25828
    },
    {
      "epoch": 2.222231781811925,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005871909658994148,
      "loss": 0.6667,
      "step": 25829
    },
    {
      "epoch": 2.2223178181192464,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005871643393520951,
      "loss": 1.8395,
      "step": 25830
    },
    {
      "epoch": 2.222403854426568,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005871377125498368,
      "loss": 0.9235,
      "step": 25831
    },
    {
      "epoch": 2.2224898907338897,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005871110854927176,
      "loss": 1.4066,
      "step": 25832
    },
    {
      "epoch": 2.2225759270412113,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005870844581808153,
      "loss": 1.1366,
      "step": 25833
    },
    {
      "epoch": 2.222661963348533,
      "grad_norm": 1.296875,
      "learning_rate": 0.000587057830614208,
      "loss": 1.3022,
      "step": 25834
    },
    {
      "epoch": 2.2227479996558546,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005870312027929735,
      "loss": 1.0577,
      "step": 25835
    },
    {
      "epoch": 2.2228340359631766,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005870045747171894,
      "loss": 1.499,
      "step": 25836
    },
    {
      "epoch": 2.2229200722704983,
      "grad_norm": 1.375,
      "learning_rate": 0.0005869779463869342,
      "loss": 1.0553,
      "step": 25837
    },
    {
      "epoch": 2.22300610857782,
      "grad_norm": 0.75,
      "learning_rate": 0.0005869513178022854,
      "loss": 0.6814,
      "step": 25838
    },
    {
      "epoch": 2.2230921448851415,
      "grad_norm": 1.984375,
      "learning_rate": 0.0005869246889633207,
      "loss": 0.7222,
      "step": 25839
    },
    {
      "epoch": 2.223178181192463,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0005868980598701183,
      "loss": 0.9918,
      "step": 25840
    },
    {
      "epoch": 2.2232642174997848,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005868714305227559,
      "loss": 0.9973,
      "step": 25841
    },
    {
      "epoch": 2.223350253807107,
      "grad_norm": 1.125,
      "learning_rate": 0.0005868448009213116,
      "loss": 1.3966,
      "step": 25842
    },
    {
      "epoch": 2.2234362901144284,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005868181710658631,
      "loss": 1.2435,
      "step": 25843
    },
    {
      "epoch": 2.22352232642175,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005867915409564882,
      "loss": 0.5024,
      "step": 25844
    },
    {
      "epoch": 2.2236083627290717,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005867649105932651,
      "loss": 1.8019,
      "step": 25845
    },
    {
      "epoch": 2.2236943990363933,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005867382799762713,
      "loss": 0.8623,
      "step": 25846
    },
    {
      "epoch": 2.223780435343715,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005867116491055851,
      "loss": 1.419,
      "step": 25847
    },
    {
      "epoch": 2.2238664716510366,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005866850179812841,
      "loss": 0.6412,
      "step": 25848
    },
    {
      "epoch": 2.2239525079583586,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005866583866034463,
      "loss": 0.5676,
      "step": 25849
    },
    {
      "epoch": 2.2240385442656803,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005866317549721495,
      "loss": 0.7718,
      "step": 25850
    },
    {
      "epoch": 2.224124580573002,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005866051230874718,
      "loss": 1.3501,
      "step": 25851
    },
    {
      "epoch": 2.2242106168803235,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005865784909494907,
      "loss": 0.8203,
      "step": 25852
    },
    {
      "epoch": 2.224296653187645,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005865518585582847,
      "loss": 1.6206,
      "step": 25853
    },
    {
      "epoch": 2.2243826894949668,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005865252259139312,
      "loss": 1.1022,
      "step": 25854
    },
    {
      "epoch": 2.2244687258022884,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005864985930165083,
      "loss": 1.1968,
      "step": 25855
    },
    {
      "epoch": 2.2245547621096105,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005864719598660937,
      "loss": 0.7584,
      "step": 25856
    },
    {
      "epoch": 2.224640798416932,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005864453264627656,
      "loss": 1.4332,
      "step": 25857
    },
    {
      "epoch": 2.2247268347242537,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005864186928066017,
      "loss": 1.3798,
      "step": 25858
    },
    {
      "epoch": 2.2248128710315753,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005863920588976797,
      "loss": 1.6066,
      "step": 25859
    },
    {
      "epoch": 2.224898907338897,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005863654247360781,
      "loss": 0.8419,
      "step": 25860
    },
    {
      "epoch": 2.2249849436462186,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005863387903218742,
      "loss": 1.0104,
      "step": 25861
    },
    {
      "epoch": 2.22507097995354,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005863121556551461,
      "loss": 1.7196,
      "step": 25862
    },
    {
      "epoch": 2.2251570162608623,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005862855207359718,
      "loss": 1.1626,
      "step": 25863
    },
    {
      "epoch": 2.225243052568184,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005862588855644292,
      "loss": 0.9507,
      "step": 25864
    },
    {
      "epoch": 2.2253290888755055,
      "grad_norm": 43.25,
      "learning_rate": 0.0005862322501405959,
      "loss": 1.4678,
      "step": 25865
    },
    {
      "epoch": 2.225415125182827,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005862056144645502,
      "loss": 0.6946,
      "step": 25866
    },
    {
      "epoch": 2.2255011614901488,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005861789785363698,
      "loss": 0.7801,
      "step": 25867
    },
    {
      "epoch": 2.2255871977974704,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005861523423561327,
      "loss": 1.595,
      "step": 25868
    },
    {
      "epoch": 2.225673234104792,
      "grad_norm": 1.90625,
      "learning_rate": 0.0005861257059239169,
      "loss": 1.7663,
      "step": 25869
    },
    {
      "epoch": 2.225759270412114,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005860990692398,
      "loss": 1.8971,
      "step": 25870
    },
    {
      "epoch": 2.2258453067194357,
      "grad_norm": 1.140625,
      "learning_rate": 0.00058607243230386,
      "loss": 1.2421,
      "step": 25871
    },
    {
      "epoch": 2.2259313430267573,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000586045795116175,
      "loss": 0.8563,
      "step": 25872
    },
    {
      "epoch": 2.226017379334079,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005860191576768228,
      "loss": 0.9043,
      "step": 25873
    },
    {
      "epoch": 2.2261034156414006,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005859925199858812,
      "loss": 0.8869,
      "step": 25874
    },
    {
      "epoch": 2.226189451948722,
      "grad_norm": 1.5,
      "learning_rate": 0.0005859658820434282,
      "loss": 1.7676,
      "step": 25875
    },
    {
      "epoch": 2.226275488256044,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005859392438495417,
      "loss": 1.1274,
      "step": 25876
    },
    {
      "epoch": 2.226361524563366,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005859126054042997,
      "loss": 1.3151,
      "step": 25877
    },
    {
      "epoch": 2.2264475608706875,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005858859667077801,
      "loss": 1.1896,
      "step": 25878
    },
    {
      "epoch": 2.226533597178009,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005858593277600607,
      "loss": 1.507,
      "step": 25879
    },
    {
      "epoch": 2.2266196334853308,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005858326885612195,
      "loss": 0.5677,
      "step": 25880
    },
    {
      "epoch": 2.2267056697926524,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005858060491113342,
      "loss": 1.2362,
      "step": 25881
    },
    {
      "epoch": 2.226791706099974,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000585779409410483,
      "loss": 1.2067,
      "step": 25882
    },
    {
      "epoch": 2.226877742407296,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005857527694587437,
      "loss": 0.9231,
      "step": 25883
    },
    {
      "epoch": 2.2269637787146177,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005857261292561942,
      "loss": 0.4435,
      "step": 25884
    },
    {
      "epoch": 2.2270498150219393,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005856994888029125,
      "loss": 0.6821,
      "step": 25885
    },
    {
      "epoch": 2.227135851329261,
      "grad_norm": 1.5,
      "learning_rate": 0.0005856728480989764,
      "loss": 1.6911,
      "step": 25886
    },
    {
      "epoch": 2.2272218876365826,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000585646207144464,
      "loss": 1.4477,
      "step": 25887
    },
    {
      "epoch": 2.227307923943904,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005856195659394531,
      "loss": 1.2674,
      "step": 25888
    },
    {
      "epoch": 2.227393960251226,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005855929244840215,
      "loss": 1.7604,
      "step": 25889
    },
    {
      "epoch": 2.227479996558548,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005855662827782473,
      "loss": 1.7945,
      "step": 25890
    },
    {
      "epoch": 2.2275660328658695,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005855396408222085,
      "loss": 1.3982,
      "step": 25891
    },
    {
      "epoch": 2.227652069173191,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005855129986159826,
      "loss": 0.9773,
      "step": 25892
    },
    {
      "epoch": 2.2277381054805128,
      "grad_norm": 1.125,
      "learning_rate": 0.000585486356159648,
      "loss": 1.2361,
      "step": 25893
    },
    {
      "epoch": 2.2278241417878344,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005854597134532824,
      "loss": 1.9742,
      "step": 25894
    },
    {
      "epoch": 2.227910178095156,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005854330704969638,
      "loss": 1.5205,
      "step": 25895
    },
    {
      "epoch": 2.227996214402478,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005854064272907699,
      "loss": 2.104,
      "step": 25896
    },
    {
      "epoch": 2.2280822507097997,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005853797838347791,
      "loss": 0.961,
      "step": 25897
    },
    {
      "epoch": 2.2281682870171213,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005853531401290689,
      "loss": 0.8464,
      "step": 25898
    },
    {
      "epoch": 2.228254323324443,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005853264961737175,
      "loss": 1.0006,
      "step": 25899
    },
    {
      "epoch": 2.2283403596317646,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005852998519688024,
      "loss": 1.0119,
      "step": 25900
    },
    {
      "epoch": 2.228426395939086,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000585273207514402,
      "loss": 1.0529,
      "step": 25901
    },
    {
      "epoch": 2.228512432246408,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000585246562810594,
      "loss": 1.2109,
      "step": 25902
    },
    {
      "epoch": 2.22859846855373,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005852199178574564,
      "loss": 0.9861,
      "step": 25903
    },
    {
      "epoch": 2.2286845048610515,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005851932726550671,
      "loss": 1.2801,
      "step": 25904
    },
    {
      "epoch": 2.228770541168373,
      "grad_norm": 1.140625,
      "learning_rate": 0.000585166627203504,
      "loss": 1.1767,
      "step": 25905
    },
    {
      "epoch": 2.2288565774756948,
      "grad_norm": 1.375,
      "learning_rate": 0.0005851399815028452,
      "loss": 1.2715,
      "step": 25906
    },
    {
      "epoch": 2.2289426137830164,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005851133355531684,
      "loss": 1.0619,
      "step": 25907
    },
    {
      "epoch": 2.229028650090338,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005850866893545518,
      "loss": 0.9085,
      "step": 25908
    },
    {
      "epoch": 2.2291146863976596,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005850600429070731,
      "loss": 0.9879,
      "step": 25909
    },
    {
      "epoch": 2.2292007227049817,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005850333962108104,
      "loss": 1.5772,
      "step": 25910
    },
    {
      "epoch": 2.2292867590123033,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005850067492658413,
      "loss": 0.7152,
      "step": 25911
    },
    {
      "epoch": 2.229372795319625,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005849801020722441,
      "loss": 2.0785,
      "step": 25912
    },
    {
      "epoch": 2.2294588316269466,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005849534546300968,
      "loss": 1.3613,
      "step": 25913
    },
    {
      "epoch": 2.229544867934268,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005849268069394771,
      "loss": 0.6497,
      "step": 25914
    },
    {
      "epoch": 2.22963090424159,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005849001590004629,
      "loss": 1.358,
      "step": 25915
    },
    {
      "epoch": 2.2297169405489115,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005848735108131323,
      "loss": 0.7775,
      "step": 25916
    },
    {
      "epoch": 2.2298029768562335,
      "grad_norm": 1.828125,
      "learning_rate": 0.0005848468623775632,
      "loss": 1.6678,
      "step": 25917
    },
    {
      "epoch": 2.229889013163555,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005848202136938336,
      "loss": 1.4803,
      "step": 25918
    },
    {
      "epoch": 2.2299750494708768,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005847935647620212,
      "loss": 1.613,
      "step": 25919
    },
    {
      "epoch": 2.2300610857781984,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005847669155822041,
      "loss": 0.5357,
      "step": 25920
    },
    {
      "epoch": 2.23014712208552,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005847402661544605,
      "loss": 1.1167,
      "step": 25921
    },
    {
      "epoch": 2.2302331583928416,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000584713616478868,
      "loss": 1.5204,
      "step": 25922
    },
    {
      "epoch": 2.2303191947001633,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005846869665555045,
      "loss": 0.9257,
      "step": 25923
    },
    {
      "epoch": 2.2304052310074853,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005846603163844482,
      "loss": 1.1677,
      "step": 25924
    },
    {
      "epoch": 2.230491267314807,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005846336659657771,
      "loss": 1.3925,
      "step": 25925
    },
    {
      "epoch": 2.2305773036221286,
      "grad_norm": 1.125,
      "learning_rate": 0.0005846070152995687,
      "loss": 0.9532,
      "step": 25926
    },
    {
      "epoch": 2.23066333992945,
      "grad_norm": 9.9375,
      "learning_rate": 0.0005845803643859015,
      "loss": 0.873,
      "step": 25927
    },
    {
      "epoch": 2.230749376236772,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005845537132248531,
      "loss": 1.545,
      "step": 25928
    },
    {
      "epoch": 2.2308354125440935,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005845270618165015,
      "loss": 0.6793,
      "step": 25929
    },
    {
      "epoch": 2.230921448851415,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005845004101609246,
      "loss": 0.8555,
      "step": 25930
    },
    {
      "epoch": 2.231007485158737,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005844737582582004,
      "loss": 1.7496,
      "step": 25931
    },
    {
      "epoch": 2.2310935214660588,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005844471061084071,
      "loss": 1.2278,
      "step": 25932
    },
    {
      "epoch": 2.2311795577733804,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005844204537116222,
      "loss": 0.9626,
      "step": 25933
    },
    {
      "epoch": 2.231265594080702,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005843938010679241,
      "loss": 1.0289,
      "step": 25934
    },
    {
      "epoch": 2.2313516303880236,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005843671481773905,
      "loss": 1.0719,
      "step": 25935
    },
    {
      "epoch": 2.2314376666953453,
      "grad_norm": 1.5625,
      "learning_rate": 0.0005843404950400994,
      "loss": 1.2095,
      "step": 25936
    },
    {
      "epoch": 2.231523703002667,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005843138416561286,
      "loss": 2.1797,
      "step": 25937
    },
    {
      "epoch": 2.231609739309989,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005842871880255565,
      "loss": 1.9104,
      "step": 25938
    },
    {
      "epoch": 2.2316957756173106,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005842605341484605,
      "loss": 0.6854,
      "step": 25939
    },
    {
      "epoch": 2.231781811924632,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005842338800249189,
      "loss": 1.4013,
      "step": 25940
    },
    {
      "epoch": 2.231867848231954,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005842072256550097,
      "loss": 1.1057,
      "step": 25941
    },
    {
      "epoch": 2.2319538845392755,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0005841805710388106,
      "loss": 1.4084,
      "step": 25942
    },
    {
      "epoch": 2.232039920846597,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005841539161763998,
      "loss": 1.2739,
      "step": 25943
    },
    {
      "epoch": 2.232125957153919,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005841272610678552,
      "loss": 1.0778,
      "step": 25944
    },
    {
      "epoch": 2.2322119934612408,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005841006057132546,
      "loss": 1.2372,
      "step": 25945
    },
    {
      "epoch": 2.2322980297685624,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005840739501126762,
      "loss": 0.8317,
      "step": 25946
    },
    {
      "epoch": 2.232384066075884,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005840472942661978,
      "loss": 0.8504,
      "step": 25947
    },
    {
      "epoch": 2.2324701023832056,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005840206381738974,
      "loss": 1.0553,
      "step": 25948
    },
    {
      "epoch": 2.2325561386905273,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005839939818358528,
      "loss": 0.9568,
      "step": 25949
    },
    {
      "epoch": 2.232642174997849,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005839673252521424,
      "loss": 1.5179,
      "step": 25950
    },
    {
      "epoch": 2.232728211305171,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005839406684228438,
      "loss": 1.7584,
      "step": 25951
    },
    {
      "epoch": 2.2328142476124926,
      "grad_norm": 5.875,
      "learning_rate": 0.000583914011348035,
      "loss": 1.5615,
      "step": 25952
    },
    {
      "epoch": 2.232900283919814,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005838873540277941,
      "loss": 1.6416,
      "step": 25953
    },
    {
      "epoch": 2.232986320227136,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0005838606964621991,
      "loss": 1.5646,
      "step": 25954
    },
    {
      "epoch": 2.2330723565344575,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005838340386513279,
      "loss": 1.8036,
      "step": 25955
    },
    {
      "epoch": 2.233158392841779,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005838073805952582,
      "loss": 2.0886,
      "step": 25956
    },
    {
      "epoch": 2.233244429149101,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005837807222940683,
      "loss": 1.145,
      "step": 25957
    },
    {
      "epoch": 2.2333304654564228,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005837540637478361,
      "loss": 2.1896,
      "step": 25958
    },
    {
      "epoch": 2.2334165017637444,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005837274049566396,
      "loss": 0.659,
      "step": 25959
    },
    {
      "epoch": 2.233502538071066,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005837007459205567,
      "loss": 1.4267,
      "step": 25960
    },
    {
      "epoch": 2.2335885743783876,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005836740866396653,
      "loss": 1.0265,
      "step": 25961
    },
    {
      "epoch": 2.2336746106857093,
      "grad_norm": 1.671875,
      "learning_rate": 0.0005836474271140437,
      "loss": 0.8787,
      "step": 25962
    },
    {
      "epoch": 2.233760646993031,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005836207673437696,
      "loss": 0.886,
      "step": 25963
    },
    {
      "epoch": 2.233846683300353,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005835941073289208,
      "loss": 1.7133,
      "step": 25964
    },
    {
      "epoch": 2.2339327196076746,
      "grad_norm": 1.25,
      "learning_rate": 0.0005835674470695757,
      "loss": 0.8682,
      "step": 25965
    },
    {
      "epoch": 2.234018755914996,
      "grad_norm": 1.34375,
      "learning_rate": 0.000583540786565812,
      "loss": 1.3226,
      "step": 25966
    },
    {
      "epoch": 2.234104792222318,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005835141258177076,
      "loss": 1.1497,
      "step": 25967
    },
    {
      "epoch": 2.2341908285296395,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005834874648253409,
      "loss": 1.0348,
      "step": 25968
    },
    {
      "epoch": 2.234276864836961,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005834608035887895,
      "loss": 1.028,
      "step": 25969
    },
    {
      "epoch": 2.2343629011442827,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0005834341421081315,
      "loss": 2.0551,
      "step": 25970
    },
    {
      "epoch": 2.2344489374516048,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005834074803834447,
      "loss": 0.7072,
      "step": 25971
    },
    {
      "epoch": 2.2345349737589264,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005833808184148074,
      "loss": 1.5192,
      "step": 25972
    },
    {
      "epoch": 2.234621010066248,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005833541562022975,
      "loss": 1.1158,
      "step": 25973
    },
    {
      "epoch": 2.2347070463735696,
      "grad_norm": 1.125,
      "learning_rate": 0.0005833274937459927,
      "loss": 1.0921,
      "step": 25974
    },
    {
      "epoch": 2.2347930826808913,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005833008310459713,
      "loss": 0.9377,
      "step": 25975
    },
    {
      "epoch": 2.234879118988213,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005832741681023111,
      "loss": 1.0426,
      "step": 25976
    },
    {
      "epoch": 2.2349651552955345,
      "grad_norm": 1.25,
      "learning_rate": 0.0005832475049150902,
      "loss": 0.7317,
      "step": 25977
    },
    {
      "epoch": 2.2350511916028566,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005832208414843865,
      "loss": 2.0983,
      "step": 25978
    },
    {
      "epoch": 2.235137227910178,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005831941778102781,
      "loss": 1.4226,
      "step": 25979
    },
    {
      "epoch": 2.2352232642175,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0005831675138928429,
      "loss": 0.7975,
      "step": 25980
    },
    {
      "epoch": 2.2353093005248215,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005831408497321589,
      "loss": 0.4181,
      "step": 25981
    },
    {
      "epoch": 2.235395336832143,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000583114185328304,
      "loss": 1.4453,
      "step": 25982
    },
    {
      "epoch": 2.2354813731394647,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005830875206813562,
      "loss": 0.7859,
      "step": 25983
    },
    {
      "epoch": 2.2355674094467863,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005830608557913937,
      "loss": 0.8954,
      "step": 25984
    },
    {
      "epoch": 2.2356534457541084,
      "grad_norm": 1.25,
      "learning_rate": 0.0005830341906584945,
      "loss": 1.4559,
      "step": 25985
    },
    {
      "epoch": 2.23573948206143,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005830075252827361,
      "loss": 0.9791,
      "step": 25986
    },
    {
      "epoch": 2.2358255183687517,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005829808596641971,
      "loss": 1.3537,
      "step": 25987
    },
    {
      "epoch": 2.2359115546760733,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005829541938029552,
      "loss": 1.1434,
      "step": 25988
    },
    {
      "epoch": 2.235997590983395,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005829275276990883,
      "loss": 0.9805,
      "step": 25989
    },
    {
      "epoch": 2.2360836272907165,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005829008613526745,
      "loss": 1.402,
      "step": 25990
    },
    {
      "epoch": 2.236169663598038,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000582874194763792,
      "loss": 1.5608,
      "step": 25991
    },
    {
      "epoch": 2.23625569990536,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005828475279325185,
      "loss": 0.7811,
      "step": 25992
    },
    {
      "epoch": 2.236341736212682,
      "grad_norm": 1.859375,
      "learning_rate": 0.0005828208608589319,
      "loss": 1.2801,
      "step": 25993
    },
    {
      "epoch": 2.2364277725200035,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005827941935431104,
      "loss": 0.6924,
      "step": 25994
    },
    {
      "epoch": 2.236513808827325,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005827675259851323,
      "loss": 1.426,
      "step": 25995
    },
    {
      "epoch": 2.2365998451346467,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005827408581850752,
      "loss": 0.8191,
      "step": 25996
    },
    {
      "epoch": 2.2366858814419683,
      "grad_norm": 1.40625,
      "learning_rate": 0.000582714190143017,
      "loss": 1.0452,
      "step": 25997
    },
    {
      "epoch": 2.2367719177492904,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005826875218590359,
      "loss": 1.43,
      "step": 25998
    },
    {
      "epoch": 2.236857954056612,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00058266085333321,
      "loss": 1.4363,
      "step": 25999
    },
    {
      "epoch": 2.2369439903639337,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005826341845656172,
      "loss": 0.9225,
      "step": 26000
    },
    {
      "epoch": 2.2370300266712553,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005826075155563354,
      "loss": 0.9994,
      "step": 26001
    },
    {
      "epoch": 2.237116062978577,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005825808463054427,
      "loss": 1.0763,
      "step": 26002
    },
    {
      "epoch": 2.2372020992858985,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0005825541768130171,
      "loss": 0.3096,
      "step": 26003
    },
    {
      "epoch": 2.23728813559322,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005825275070791367,
      "loss": 0.6662,
      "step": 26004
    },
    {
      "epoch": 2.237374171900542,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005825008371038793,
      "loss": 1.0019,
      "step": 26005
    },
    {
      "epoch": 2.237460208207864,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005824741668873229,
      "loss": 1.3331,
      "step": 26006
    },
    {
      "epoch": 2.2375462445151855,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005824474964295457,
      "loss": 0.5889,
      "step": 26007
    },
    {
      "epoch": 2.237632280822507,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005824208257306255,
      "loss": 0.8908,
      "step": 26008
    },
    {
      "epoch": 2.2377183171298287,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005823941547906407,
      "loss": 1.5064,
      "step": 26009
    },
    {
      "epoch": 2.2378043534371503,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005823674836096688,
      "loss": 0.8283,
      "step": 26010
    },
    {
      "epoch": 2.237890389744472,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000582340812187788,
      "loss": 1.4146,
      "step": 26011
    },
    {
      "epoch": 2.237976426051794,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005823141405250763,
      "loss": 1.9349,
      "step": 26012
    },
    {
      "epoch": 2.2380624623591157,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005822874686216118,
      "loss": 0.6877,
      "step": 26013
    },
    {
      "epoch": 2.2381484986664373,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005822607964774725,
      "loss": 1.2865,
      "step": 26014
    },
    {
      "epoch": 2.238234534973759,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005822341240927364,
      "loss": 0.7725,
      "step": 26015
    },
    {
      "epoch": 2.2383205712810805,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005822074514674813,
      "loss": 0.9704,
      "step": 26016
    },
    {
      "epoch": 2.238406607588402,
      "grad_norm": 1.59375,
      "learning_rate": 0.0005821807786017857,
      "loss": 1.6162,
      "step": 26017
    },
    {
      "epoch": 2.238492643895724,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005821541054957272,
      "loss": 1.1849,
      "step": 26018
    },
    {
      "epoch": 2.238578680203046,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005821274321493838,
      "loss": 0.9526,
      "step": 26019
    },
    {
      "epoch": 2.2386647165103675,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005821007585628336,
      "loss": 1.1783,
      "step": 26020
    },
    {
      "epoch": 2.238750752817689,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005820740847361547,
      "loss": 0.653,
      "step": 26021
    },
    {
      "epoch": 2.2388367891250107,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005820474106694253,
      "loss": 1.3438,
      "step": 26022
    },
    {
      "epoch": 2.2389228254323323,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0005820207363627229,
      "loss": 1.4161,
      "step": 26023
    },
    {
      "epoch": 2.239008861739654,
      "grad_norm": 1.0,
      "learning_rate": 0.000581994061816126,
      "loss": 1.3307,
      "step": 26024
    },
    {
      "epoch": 2.239094898046976,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005819673870297122,
      "loss": 1.2462,
      "step": 26025
    },
    {
      "epoch": 2.2391809343542977,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005819407120035599,
      "loss": 0.7277,
      "step": 26026
    },
    {
      "epoch": 2.2392669706616193,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005819140367377467,
      "loss": 1.438,
      "step": 26027
    },
    {
      "epoch": 2.239353006968941,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005818873612323512,
      "loss": 1.0225,
      "step": 26028
    },
    {
      "epoch": 2.2394390432762625,
      "grad_norm": 1.125,
      "learning_rate": 0.0005818606854874511,
      "loss": 1.2684,
      "step": 26029
    },
    {
      "epoch": 2.239525079583584,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005818340095031241,
      "loss": 2.3227,
      "step": 26030
    },
    {
      "epoch": 2.239611115890906,
      "grad_norm": 1.375,
      "learning_rate": 0.0005818073332794487,
      "loss": 1.0809,
      "step": 26031
    },
    {
      "epoch": 2.239697152198228,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005817806568165028,
      "loss": 1.789,
      "step": 26032
    },
    {
      "epoch": 2.2397831885055495,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005817539801143644,
      "loss": 1.2359,
      "step": 26033
    },
    {
      "epoch": 2.239869224812871,
      "grad_norm": 0.671875,
      "learning_rate": 0.0005817273031731115,
      "loss": 0.7327,
      "step": 26034
    },
    {
      "epoch": 2.2399552611201927,
      "grad_norm": 1.359375,
      "learning_rate": 0.000581700625992822,
      "loss": 1.584,
      "step": 26035
    },
    {
      "epoch": 2.2400412974275143,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005816739485735743,
      "loss": 1.1056,
      "step": 26036
    },
    {
      "epoch": 2.240127333734836,
      "grad_norm": 1.265625,
      "learning_rate": 0.000581647270915446,
      "loss": 1.2254,
      "step": 26037
    },
    {
      "epoch": 2.2402133700421576,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005816205930185152,
      "loss": 1.7752,
      "step": 26038
    },
    {
      "epoch": 2.2402994063494797,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005815939148828603,
      "loss": 1.0389,
      "step": 26039
    },
    {
      "epoch": 2.2403854426568013,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000581567236508559,
      "loss": 1.0466,
      "step": 26040
    },
    {
      "epoch": 2.240471478964123,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005815405578956894,
      "loss": 1.4399,
      "step": 26041
    },
    {
      "epoch": 2.2405575152714445,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005815138790443294,
      "loss": 1.002,
      "step": 26042
    },
    {
      "epoch": 2.240643551578766,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005814871999545574,
      "loss": 0.6505,
      "step": 26043
    },
    {
      "epoch": 2.240729587886088,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005814605206264509,
      "loss": 1.3559,
      "step": 26044
    },
    {
      "epoch": 2.2408156241934094,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005814338410600883,
      "loss": 0.8083,
      "step": 26045
    },
    {
      "epoch": 2.2409016605007315,
      "grad_norm": 1.0,
      "learning_rate": 0.0005814071612555477,
      "loss": 0.6836,
      "step": 26046
    },
    {
      "epoch": 2.240987696808053,
      "grad_norm": 1.203125,
      "learning_rate": 0.000581380481212907,
      "loss": 1.1235,
      "step": 26047
    },
    {
      "epoch": 2.2410737331153747,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005813538009322441,
      "loss": 1.3882,
      "step": 26048
    },
    {
      "epoch": 2.2411597694226963,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005813271204136371,
      "loss": 0.8878,
      "step": 26049
    },
    {
      "epoch": 2.241245805730018,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005813004396571641,
      "loss": 0.9199,
      "step": 26050
    },
    {
      "epoch": 2.2413318420373396,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005812737586629032,
      "loss": 1.605,
      "step": 26051
    },
    {
      "epoch": 2.241417878344661,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005812470774309324,
      "loss": 0.6971,
      "step": 26052
    },
    {
      "epoch": 2.2415039146519833,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005812203959613296,
      "loss": 1.1192,
      "step": 26053
    },
    {
      "epoch": 2.241589950959305,
      "grad_norm": 1.0,
      "learning_rate": 0.000581193714254173,
      "loss": 0.6396,
      "step": 26054
    },
    {
      "epoch": 2.2416759872666265,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005811670323095406,
      "loss": 0.7218,
      "step": 26055
    },
    {
      "epoch": 2.241762023573948,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005811403501275104,
      "loss": 0.6428,
      "step": 26056
    },
    {
      "epoch": 2.24184805988127,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005811136677081603,
      "loss": 1.1165,
      "step": 26057
    },
    {
      "epoch": 2.2419340961885914,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005810869850515687,
      "loss": 1.1326,
      "step": 26058
    },
    {
      "epoch": 2.2420201324959135,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005810603021578135,
      "loss": 1.5731,
      "step": 26059
    },
    {
      "epoch": 2.242106168803235,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005810336190269725,
      "loss": 1.4661,
      "step": 26060
    },
    {
      "epoch": 2.2421922051105567,
      "grad_norm": 2.8125,
      "learning_rate": 0.000581006935659124,
      "loss": 1.5852,
      "step": 26061
    },
    {
      "epoch": 2.2422782414178783,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005809802520543459,
      "loss": 0.8487,
      "step": 26062
    },
    {
      "epoch": 2.2423642777252,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005809535682127165,
      "loss": 0.796,
      "step": 26063
    },
    {
      "epoch": 2.2424503140325216,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005809268841343133,
      "loss": 0.545,
      "step": 26064
    },
    {
      "epoch": 2.242536350339843,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005809001998192151,
      "loss": 1.5215,
      "step": 26065
    },
    {
      "epoch": 2.2426223866471653,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005808735152674993,
      "loss": 1.3683,
      "step": 26066
    },
    {
      "epoch": 2.242708422954487,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005808468304792444,
      "loss": 1.3712,
      "step": 26067
    },
    {
      "epoch": 2.2427944592618085,
      "grad_norm": 1.0,
      "learning_rate": 0.0005808201454545279,
      "loss": 1.2352,
      "step": 26068
    },
    {
      "epoch": 2.24288049556913,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005807934601934284,
      "loss": 1.3966,
      "step": 26069
    },
    {
      "epoch": 2.242966531876452,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005807667746960238,
      "loss": 1.0793,
      "step": 26070
    },
    {
      "epoch": 2.2430525681837734,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005807400889623921,
      "loss": 1.103,
      "step": 26071
    },
    {
      "epoch": 2.2431386044910955,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005807134029926112,
      "loss": 1.2941,
      "step": 26072
    },
    {
      "epoch": 2.243224640798417,
      "grad_norm": 2.46875,
      "learning_rate": 0.0005806867167867595,
      "loss": 1.5115,
      "step": 26073
    },
    {
      "epoch": 2.2433106771057387,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005806600303449146,
      "loss": 1.4012,
      "step": 26074
    },
    {
      "epoch": 2.2433967134130604,
      "grad_norm": 1.5,
      "learning_rate": 0.0005806333436671551,
      "loss": 0.8946,
      "step": 26075
    },
    {
      "epoch": 2.243482749720382,
      "grad_norm": 1.0,
      "learning_rate": 0.0005806066567535584,
      "loss": 0.6842,
      "step": 26076
    },
    {
      "epoch": 2.2435687860277036,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005805799696042031,
      "loss": 0.7979,
      "step": 26077
    },
    {
      "epoch": 2.2436548223350252,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005805532822191671,
      "loss": 1.1356,
      "step": 26078
    },
    {
      "epoch": 2.2437408586423473,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005805265945985283,
      "loss": 0.8626,
      "step": 26079
    },
    {
      "epoch": 2.243826894949669,
      "grad_norm": 1.046875,
      "learning_rate": 0.000580499906742365,
      "loss": 1.1737,
      "step": 26080
    },
    {
      "epoch": 2.2439129312569905,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005804732186507551,
      "loss": 0.8981,
      "step": 26081
    },
    {
      "epoch": 2.243998967564312,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005804465303237767,
      "loss": 1.6944,
      "step": 26082
    },
    {
      "epoch": 2.244085003871634,
      "grad_norm": 0.75,
      "learning_rate": 0.0005804198417615076,
      "loss": 0.8703,
      "step": 26083
    },
    {
      "epoch": 2.2441710401789554,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005803931529640264,
      "loss": 0.6088,
      "step": 26084
    },
    {
      "epoch": 2.244257076486277,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005803664639314108,
      "loss": 1.4126,
      "step": 26085
    },
    {
      "epoch": 2.244343112793599,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005803397746637389,
      "loss": 1.5879,
      "step": 26086
    },
    {
      "epoch": 2.2444291491009207,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005803130851610886,
      "loss": 0.7025,
      "step": 26087
    },
    {
      "epoch": 2.2445151854082424,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005802863954235384,
      "loss": 1.4005,
      "step": 26088
    },
    {
      "epoch": 2.244601221715564,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000580259705451166,
      "loss": 0.7301,
      "step": 26089
    },
    {
      "epoch": 2.2446872580228856,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005802330152440497,
      "loss": 0.9551,
      "step": 26090
    },
    {
      "epoch": 2.2447732943302072,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005802063248022672,
      "loss": 1.0153,
      "step": 26091
    },
    {
      "epoch": 2.244859330637529,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005801796341258969,
      "loss": 1.4846,
      "step": 26092
    },
    {
      "epoch": 2.244945366944851,
      "grad_norm": 1.671875,
      "learning_rate": 0.0005801529432150168,
      "loss": 1.6588,
      "step": 26093
    },
    {
      "epoch": 2.2450314032521725,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005801262520697049,
      "loss": 1.0631,
      "step": 26094
    },
    {
      "epoch": 2.245117439559494,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005800995606900395,
      "loss": 0.9915,
      "step": 26095
    },
    {
      "epoch": 2.245203475866816,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005800728690760982,
      "loss": 0.9849,
      "step": 26096
    },
    {
      "epoch": 2.2452895121741374,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005800461772279595,
      "loss": 1.0629,
      "step": 26097
    },
    {
      "epoch": 2.245375548481459,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005800194851457012,
      "loss": 1.293,
      "step": 26098
    },
    {
      "epoch": 2.2454615847887807,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005799927928294014,
      "loss": 0.9352,
      "step": 26099
    },
    {
      "epoch": 2.2455476210961027,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005799661002791383,
      "loss": 1.5169,
      "step": 26100
    },
    {
      "epoch": 2.2456336574034244,
      "grad_norm": 1.125,
      "learning_rate": 0.0005799394074949899,
      "loss": 0.581,
      "step": 26101
    },
    {
      "epoch": 2.245719693710746,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005799127144770344,
      "loss": 1.433,
      "step": 26102
    },
    {
      "epoch": 2.2458057300180676,
      "grad_norm": 1.375,
      "learning_rate": 0.0005798860212253497,
      "loss": 1.2547,
      "step": 26103
    },
    {
      "epoch": 2.2458917663253892,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005798593277400139,
      "loss": 1.0725,
      "step": 26104
    },
    {
      "epoch": 2.245977802632711,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000579832634021105,
      "loss": 0.8162,
      "step": 26105
    },
    {
      "epoch": 2.2460638389400325,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005798059400687011,
      "loss": 1.5034,
      "step": 26106
    },
    {
      "epoch": 2.2461498752473545,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005797792458828806,
      "loss": 0.63,
      "step": 26107
    },
    {
      "epoch": 2.246235911554676,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005797525514637213,
      "loss": 2.0373,
      "step": 26108
    },
    {
      "epoch": 2.246321947861998,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005797258568113011,
      "loss": 1.7009,
      "step": 26109
    },
    {
      "epoch": 2.2464079841693194,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005796991619256985,
      "loss": 1.3476,
      "step": 26110
    },
    {
      "epoch": 2.246494020476641,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005796724668069912,
      "loss": 1.6123,
      "step": 26111
    },
    {
      "epoch": 2.2465800567839627,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005796457714552576,
      "loss": 1.1687,
      "step": 26112
    },
    {
      "epoch": 2.2466660930912843,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005796190758705754,
      "loss": 1.214,
      "step": 26113
    },
    {
      "epoch": 2.2467521293986064,
      "grad_norm": 1.21875,
      "learning_rate": 0.000579592380053023,
      "loss": 1.8844,
      "step": 26114
    },
    {
      "epoch": 2.246838165705928,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005795656840026784,
      "loss": 0.8435,
      "step": 26115
    },
    {
      "epoch": 2.2469242020132496,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005795389877196195,
      "loss": 1.0866,
      "step": 26116
    },
    {
      "epoch": 2.2470102383205712,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005795122912039247,
      "loss": 0.9832,
      "step": 26117
    },
    {
      "epoch": 2.247096274627893,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005794855944556719,
      "loss": 2.1015,
      "step": 26118
    },
    {
      "epoch": 2.2471823109352145,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0005794588974749391,
      "loss": 1.5592,
      "step": 26119
    },
    {
      "epoch": 2.2472683472425365,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005794322002618045,
      "loss": 1.9656,
      "step": 26120
    },
    {
      "epoch": 2.247354383549858,
      "grad_norm": 1.546875,
      "learning_rate": 0.000579405502816346,
      "loss": 1.3702,
      "step": 26121
    },
    {
      "epoch": 2.24744041985718,
      "grad_norm": 1.8125,
      "learning_rate": 0.0005793788051386421,
      "loss": 1.657,
      "step": 26122
    },
    {
      "epoch": 2.2475264561645014,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005793521072287707,
      "loss": 0.9963,
      "step": 26123
    },
    {
      "epoch": 2.247612492471823,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005793254090868096,
      "loss": 1.8358,
      "step": 26124
    },
    {
      "epoch": 2.2476985287791447,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005792987107128371,
      "loss": 1.1703,
      "step": 26125
    },
    {
      "epoch": 2.2477845650864663,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005792720121069314,
      "loss": 0.6912,
      "step": 26126
    },
    {
      "epoch": 2.2478706013937884,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005792453132691705,
      "loss": 0.9048,
      "step": 26127
    },
    {
      "epoch": 2.24795663770111,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005792186141996323,
      "loss": 0.9862,
      "step": 26128
    },
    {
      "epoch": 2.2480426740084316,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005791919148983953,
      "loss": 1.2846,
      "step": 26129
    },
    {
      "epoch": 2.2481287103157532,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005791652153655372,
      "loss": 1.9059,
      "step": 26130
    },
    {
      "epoch": 2.248214746623075,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005791385156011364,
      "loss": 1.5007,
      "step": 26131
    },
    {
      "epoch": 2.2483007829303965,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005791118156052706,
      "loss": 0.8615,
      "step": 26132
    },
    {
      "epoch": 2.2483868192377185,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005790851153780182,
      "loss": 1.0566,
      "step": 26133
    },
    {
      "epoch": 2.24847285554504,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005790584149194573,
      "loss": 1.5258,
      "step": 26134
    },
    {
      "epoch": 2.248558891852362,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005790317142296658,
      "loss": 1.1073,
      "step": 26135
    },
    {
      "epoch": 2.2486449281596834,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005790050133087221,
      "loss": 1.3552,
      "step": 26136
    },
    {
      "epoch": 2.248730964467005,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005789783121567041,
      "loss": 1.3244,
      "step": 26137
    },
    {
      "epoch": 2.2488170007743267,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005789516107736897,
      "loss": 2.2237,
      "step": 26138
    },
    {
      "epoch": 2.2489030370816483,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005789249091597571,
      "loss": 0.926,
      "step": 26139
    },
    {
      "epoch": 2.2489890733889704,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005788982073149847,
      "loss": 0.7915,
      "step": 26140
    },
    {
      "epoch": 2.249075109696292,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005788715052394504,
      "loss": 0.9526,
      "step": 26141
    },
    {
      "epoch": 2.2491611460036136,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005788448029332323,
      "loss": 0.8358,
      "step": 26142
    },
    {
      "epoch": 2.2492471823109352,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005788181003964082,
      "loss": 1.1846,
      "step": 26143
    },
    {
      "epoch": 2.249333218618257,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005787913976290569,
      "loss": 0.8643,
      "step": 26144
    },
    {
      "epoch": 2.2494192549255785,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005787646946312559,
      "loss": 0.7279,
      "step": 26145
    },
    {
      "epoch": 2.2495052912329,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005787379914030835,
      "loss": 0.7035,
      "step": 26146
    },
    {
      "epoch": 2.249591327540222,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005787112879446177,
      "loss": 1.0988,
      "step": 26147
    },
    {
      "epoch": 2.249677363847544,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005786845842559367,
      "loss": 1.0311,
      "step": 26148
    },
    {
      "epoch": 2.2497634001548654,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005786578803371187,
      "loss": 1.5757,
      "step": 26149
    },
    {
      "epoch": 2.249849436462187,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005786311761882416,
      "loss": 0.8421,
      "step": 26150
    },
    {
      "epoch": 2.2499354727695087,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005786044718093837,
      "loss": 0.9587,
      "step": 26151
    },
    {
      "epoch": 2.2500215090768303,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005785777672006231,
      "loss": 1.0316,
      "step": 26152
    },
    {
      "epoch": 2.250107545384152,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005785510623620377,
      "loss": 1.5819,
      "step": 26153
    },
    {
      "epoch": 2.250193581691474,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005785243572937055,
      "loss": 1.7956,
      "step": 26154
    },
    {
      "epoch": 2.250193581691474,
      "eval_loss": 1.1643381118774414,
      "eval_runtime": 355.5701,
      "eval_samples_per_second": 13.075,
      "eval_steps_per_second": 1.637,
      "step": 26154
    },
    {
      "epoch": 2.2502796179987956,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005784976519957051,
      "loss": 1.5499,
      "step": 26155
    },
    {
      "epoch": 2.2503656543061172,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005784709464681144,
      "loss": 1.8016,
      "step": 26156
    },
    {
      "epoch": 2.250451690613439,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005784442407110114,
      "loss": 1.9023,
      "step": 26157
    },
    {
      "epoch": 2.2505377269207605,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000578417534724474,
      "loss": 0.9046,
      "step": 26158
    },
    {
      "epoch": 2.250623763228082,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005783908285085807,
      "loss": 0.534,
      "step": 26159
    },
    {
      "epoch": 2.2507097995354037,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005783641220634093,
      "loss": 1.1519,
      "step": 26160
    },
    {
      "epoch": 2.250795835842726,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005783374153890384,
      "loss": 0.8945,
      "step": 26161
    },
    {
      "epoch": 2.2508818721500474,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005783107084855457,
      "loss": 1.7291,
      "step": 26162
    },
    {
      "epoch": 2.250967908457369,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005782840013530093,
      "loss": 1.7736,
      "step": 26163
    },
    {
      "epoch": 2.2510539447646907,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005782572939915077,
      "loss": 1.9229,
      "step": 26164
    },
    {
      "epoch": 2.2511399810720123,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005782305864011184,
      "loss": 0.9902,
      "step": 26165
    },
    {
      "epoch": 2.251226017379334,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005782038785819199,
      "loss": 1.4259,
      "step": 26166
    },
    {
      "epoch": 2.2513120536866555,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005781771705339904,
      "loss": 0.9749,
      "step": 26167
    },
    {
      "epoch": 2.2513980899939776,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005781504622574078,
      "loss": 0.6054,
      "step": 26168
    },
    {
      "epoch": 2.2514841263012992,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005781237537522503,
      "loss": 1.3318,
      "step": 26169
    },
    {
      "epoch": 2.251570162608621,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005780970450185961,
      "loss": 1.9959,
      "step": 26170
    },
    {
      "epoch": 2.2516561989159425,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005780703360565233,
      "loss": 1.4212,
      "step": 26171
    },
    {
      "epoch": 2.251742235223264,
      "grad_norm": 1.125,
      "learning_rate": 0.0005780436268661098,
      "loss": 0.7353,
      "step": 26172
    },
    {
      "epoch": 2.2518282715305857,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005780169174474338,
      "loss": 1.0307,
      "step": 26173
    },
    {
      "epoch": 2.2519143078379074,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005779902078005736,
      "loss": 0.7782,
      "step": 26174
    },
    {
      "epoch": 2.2520003441452294,
      "grad_norm": 1.5,
      "learning_rate": 0.0005779634979256073,
      "loss": 0.7039,
      "step": 26175
    },
    {
      "epoch": 2.252086380452551,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005779367878226127,
      "loss": 1.1947,
      "step": 26176
    },
    {
      "epoch": 2.2521724167598727,
      "grad_norm": 1.375,
      "learning_rate": 0.0005779100774916683,
      "loss": 1.4258,
      "step": 26177
    },
    {
      "epoch": 2.2522584530671943,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005778833669328521,
      "loss": 1.5781,
      "step": 26178
    },
    {
      "epoch": 2.252344489374516,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005778566561462421,
      "loss": 1.1155,
      "step": 26179
    },
    {
      "epoch": 2.2524305256818375,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005778299451319167,
      "loss": 1.0764,
      "step": 26180
    },
    {
      "epoch": 2.2525165619891596,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005778032338899536,
      "loss": 1.348,
      "step": 26181
    },
    {
      "epoch": 2.2526025982964812,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005777765224204314,
      "loss": 0.7954,
      "step": 26182
    },
    {
      "epoch": 2.252688634603803,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000577749810723428,
      "loss": 0.8545,
      "step": 26183
    },
    {
      "epoch": 2.2527746709111245,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005777230987990212,
      "loss": 1.371,
      "step": 26184
    },
    {
      "epoch": 2.252860707218446,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005776963866472898,
      "loss": 1.3296,
      "step": 26185
    },
    {
      "epoch": 2.2529467435257677,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005776696742683114,
      "loss": 0.8375,
      "step": 26186
    },
    {
      "epoch": 2.25303277983309,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005776429616621645,
      "loss": 1.5522,
      "step": 26187
    },
    {
      "epoch": 2.2531188161404114,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005776162488289267,
      "loss": 0.8984,
      "step": 26188
    },
    {
      "epoch": 2.253204852447733,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005775895357686768,
      "loss": 1.1138,
      "step": 26189
    },
    {
      "epoch": 2.2532908887550547,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005775628224814925,
      "loss": 1.6221,
      "step": 26190
    },
    {
      "epoch": 2.2533769250623763,
      "grad_norm": 1.0,
      "learning_rate": 0.0005775361089674519,
      "loss": 1.2058,
      "step": 26191
    },
    {
      "epoch": 2.253462961369698,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0005775093952266334,
      "loss": 1.6984,
      "step": 26192
    },
    {
      "epoch": 2.2535489976770195,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005774826812591149,
      "loss": 1.5443,
      "step": 26193
    },
    {
      "epoch": 2.2536350339843416,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005774559670649748,
      "loss": 1.7142,
      "step": 26194
    },
    {
      "epoch": 2.2537210702916632,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005774292526442908,
      "loss": 1.1899,
      "step": 26195
    },
    {
      "epoch": 2.253807106598985,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005774025379971415,
      "loss": 1.2876,
      "step": 26196
    },
    {
      "epoch": 2.2538931429063065,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005773758231236047,
      "loss": 1.0801,
      "step": 26197
    },
    {
      "epoch": 2.253979179213628,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005773491080237588,
      "loss": 0.7415,
      "step": 26198
    },
    {
      "epoch": 2.2540652155209497,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005773223926976816,
      "loss": 0.6968,
      "step": 26199
    },
    {
      "epoch": 2.2541512518282714,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005772956771454516,
      "loss": 1.2649,
      "step": 26200
    },
    {
      "epoch": 2.2542372881355934,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005772689613671468,
      "loss": 0.9747,
      "step": 26201
    },
    {
      "epoch": 2.254323324442915,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005772422453628453,
      "loss": 0.9816,
      "step": 26202
    },
    {
      "epoch": 2.2544093607502367,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005772155291326253,
      "loss": 2.1596,
      "step": 26203
    },
    {
      "epoch": 2.2544953970575583,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005771888126765648,
      "loss": 0.9614,
      "step": 26204
    },
    {
      "epoch": 2.25458143336488,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000577162095994742,
      "loss": 0.6109,
      "step": 26205
    },
    {
      "epoch": 2.2546674696722016,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005771353790872351,
      "loss": 0.8611,
      "step": 26206
    },
    {
      "epoch": 2.254753505979523,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005771086619541223,
      "loss": 1.606,
      "step": 26207
    },
    {
      "epoch": 2.2548395422868452,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005770819445954816,
      "loss": 1.837,
      "step": 26208
    },
    {
      "epoch": 2.254925578594167,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005770552270113913,
      "loss": 1.7695,
      "step": 26209
    },
    {
      "epoch": 2.2550116149014885,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005770285092019293,
      "loss": 1.279,
      "step": 26210
    },
    {
      "epoch": 2.25509765120881,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005770017911671739,
      "loss": 1.2164,
      "step": 26211
    },
    {
      "epoch": 2.2551836875161317,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005769750729072035,
      "loss": 1.4169,
      "step": 26212
    },
    {
      "epoch": 2.2552697238234534,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005769483544220957,
      "loss": 1.3252,
      "step": 26213
    },
    {
      "epoch": 2.255355760130775,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005769216357119289,
      "loss": 1.061,
      "step": 26214
    },
    {
      "epoch": 2.255441796438097,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005768949167767814,
      "loss": 0.8148,
      "step": 26215
    },
    {
      "epoch": 2.2555278327454187,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005768681976167312,
      "loss": 1.3594,
      "step": 26216
    },
    {
      "epoch": 2.2556138690527403,
      "grad_norm": 1.375,
      "learning_rate": 0.0005768414782318564,
      "loss": 1.3734,
      "step": 26217
    },
    {
      "epoch": 2.255699905360062,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005768147586222352,
      "loss": 1.1432,
      "step": 26218
    },
    {
      "epoch": 2.2557859416673836,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005767880387879459,
      "loss": 0.8009,
      "step": 26219
    },
    {
      "epoch": 2.255871977974705,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005767613187290665,
      "loss": 1.0518,
      "step": 26220
    },
    {
      "epoch": 2.255958014282027,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005767345984456751,
      "loss": 1.9134,
      "step": 26221
    },
    {
      "epoch": 2.256044050589349,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005767078779378501,
      "loss": 2.0982,
      "step": 26222
    },
    {
      "epoch": 2.2561300868966705,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005766811572056694,
      "loss": 0.7159,
      "step": 26223
    },
    {
      "epoch": 2.256216123203992,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005766544362492112,
      "loss": 0.8068,
      "step": 26224
    },
    {
      "epoch": 2.2563021595113137,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005766277150685537,
      "loss": 1.6549,
      "step": 26225
    },
    {
      "epoch": 2.2563881958186354,
      "grad_norm": 1.0625,
      "learning_rate": 0.000576600993663775,
      "loss": 1.4328,
      "step": 26226
    },
    {
      "epoch": 2.256474232125957,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005765742720349533,
      "loss": 1.5106,
      "step": 26227
    },
    {
      "epoch": 2.2565602684332786,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005765475501821667,
      "loss": 1.4473,
      "step": 26228
    },
    {
      "epoch": 2.2566463047406007,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005765208281054934,
      "loss": 1.1431,
      "step": 26229
    },
    {
      "epoch": 2.2567323410479223,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005764941058050117,
      "loss": 1.6891,
      "step": 26230
    },
    {
      "epoch": 2.256818377355244,
      "grad_norm": 1.125,
      "learning_rate": 0.0005764673832807995,
      "loss": 0.9932,
      "step": 26231
    },
    {
      "epoch": 2.2569044136625656,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000576440660532935,
      "loss": 1.1573,
      "step": 26232
    },
    {
      "epoch": 2.256990449969887,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005764139375614967,
      "loss": 1.2575,
      "step": 26233
    },
    {
      "epoch": 2.257076486277209,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005763872143665622,
      "loss": 1.1858,
      "step": 26234
    },
    {
      "epoch": 2.2571625225845304,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005763604909482101,
      "loss": 0.9348,
      "step": 26235
    },
    {
      "epoch": 2.2572485588918525,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005763337673065184,
      "loss": 1.0854,
      "step": 26236
    },
    {
      "epoch": 2.257334595199174,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005763070434415652,
      "loss": 1.1323,
      "step": 26237
    },
    {
      "epoch": 2.2574206315064957,
      "grad_norm": 1.125,
      "learning_rate": 0.0005762803193534289,
      "loss": 1.9855,
      "step": 26238
    },
    {
      "epoch": 2.2575066678138174,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005762535950421875,
      "loss": 1.754,
      "step": 26239
    },
    {
      "epoch": 2.257592704121139,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005762268705079189,
      "loss": 1.8411,
      "step": 26240
    },
    {
      "epoch": 2.2576787404284606,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005762001457507018,
      "loss": 1.5316,
      "step": 26241
    },
    {
      "epoch": 2.2577647767357827,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000576173420770614,
      "loss": 1.1607,
      "step": 26242
    },
    {
      "epoch": 2.2578508130431043,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005761466955677338,
      "loss": 1.6838,
      "step": 26243
    },
    {
      "epoch": 2.257936849350426,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005761199701421391,
      "loss": 0.861,
      "step": 26244
    },
    {
      "epoch": 2.2580228856577476,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005760932444939087,
      "loss": 2.2115,
      "step": 26245
    },
    {
      "epoch": 2.258108921965069,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005760665186231201,
      "loss": 1.7531,
      "step": 26246
    },
    {
      "epoch": 2.258194958272391,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005760397925298517,
      "loss": 0.6389,
      "step": 26247
    },
    {
      "epoch": 2.258280994579713,
      "grad_norm": 1.25,
      "learning_rate": 0.0005760130662141818,
      "loss": 2.002,
      "step": 26248
    },
    {
      "epoch": 2.2583670308870345,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005759863396761884,
      "loss": 1.2501,
      "step": 26249
    },
    {
      "epoch": 2.258453067194356,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005759596129159498,
      "loss": 1.0748,
      "step": 26250
    },
    {
      "epoch": 2.2585391035016777,
      "grad_norm": 0.9375,
      "learning_rate": 0.000575932885933544,
      "loss": 0.6615,
      "step": 26251
    },
    {
      "epoch": 2.2586251398089994,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005759061587290494,
      "loss": 1.6391,
      "step": 26252
    },
    {
      "epoch": 2.258711176116321,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000575879431302544,
      "loss": 0.9141,
      "step": 26253
    },
    {
      "epoch": 2.2587972124236426,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005758527036541059,
      "loss": 1.0208,
      "step": 26254
    },
    {
      "epoch": 2.2588832487309647,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005758259757838134,
      "loss": 0.794,
      "step": 26255
    },
    {
      "epoch": 2.2589692850382863,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005757992476917448,
      "loss": 0.6279,
      "step": 26256
    },
    {
      "epoch": 2.259055321345608,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005757725193779782,
      "loss": 1.0978,
      "step": 26257
    },
    {
      "epoch": 2.2591413576529296,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005757457908425917,
      "loss": 1.1567,
      "step": 26258
    },
    {
      "epoch": 2.259227393960251,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005757190620856634,
      "loss": 1.5535,
      "step": 26259
    },
    {
      "epoch": 2.259313430267573,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005756923331072716,
      "loss": 0.8632,
      "step": 26260
    },
    {
      "epoch": 2.2593994665748944,
      "grad_norm": 1.59375,
      "learning_rate": 0.0005756656039074946,
      "loss": 1.4048,
      "step": 26261
    },
    {
      "epoch": 2.2594855028822165,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005756388744864101,
      "loss": 1.103,
      "step": 26262
    },
    {
      "epoch": 2.259571539189538,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005756121448440969,
      "loss": 1.2515,
      "step": 26263
    },
    {
      "epoch": 2.2596575754968597,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005755854149806327,
      "loss": 1.5833,
      "step": 26264
    },
    {
      "epoch": 2.2597436118041814,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000575558684896096,
      "loss": 1.5264,
      "step": 26265
    },
    {
      "epoch": 2.259829648111503,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005755319545905646,
      "loss": 0.8596,
      "step": 26266
    },
    {
      "epoch": 2.2599156844188246,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005755052240641172,
      "loss": 1.5856,
      "step": 26267
    },
    {
      "epoch": 2.2600017207261462,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005754784933168315,
      "loss": 1.7386,
      "step": 26268
    },
    {
      "epoch": 2.2600877570334683,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005754517623487861,
      "loss": 1.0941,
      "step": 26269
    },
    {
      "epoch": 2.26017379334079,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005754250311600589,
      "loss": 0.9125,
      "step": 26270
    },
    {
      "epoch": 2.2602598296481116,
      "grad_norm": 2.140625,
      "learning_rate": 0.0005753982997507279,
      "loss": 1.4726,
      "step": 26271
    },
    {
      "epoch": 2.260345865955433,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005753715681208717,
      "loss": 1.5583,
      "step": 26272
    },
    {
      "epoch": 2.260431902262755,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0005753448362705684,
      "loss": 1.6059,
      "step": 26273
    },
    {
      "epoch": 2.2605179385700764,
      "grad_norm": 0.921875,
      "learning_rate": 0.000575318104199896,
      "loss": 1.1253,
      "step": 26274
    },
    {
      "epoch": 2.260603974877398,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000575291371908933,
      "loss": 1.1904,
      "step": 26275
    },
    {
      "epoch": 2.26069001118472,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005752646393977572,
      "loss": 1.0356,
      "step": 26276
    },
    {
      "epoch": 2.2607760474920418,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005752379066664469,
      "loss": 1.0566,
      "step": 26277
    },
    {
      "epoch": 2.2608620837993634,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005752111737150806,
      "loss": 1.4239,
      "step": 26278
    },
    {
      "epoch": 2.260948120106685,
      "grad_norm": 1.25,
      "learning_rate": 0.0005751844405437361,
      "loss": 1.3503,
      "step": 26279
    },
    {
      "epoch": 2.2610341564140066,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005751577071524918,
      "loss": 1.2905,
      "step": 26280
    },
    {
      "epoch": 2.2611201927213282,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005751309735414256,
      "loss": 1.6174,
      "step": 26281
    },
    {
      "epoch": 2.26120622902865,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005751042397106162,
      "loss": 1.2129,
      "step": 26282
    },
    {
      "epoch": 2.261292265335972,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005750775056601414,
      "loss": 1.8311,
      "step": 26283
    },
    {
      "epoch": 2.2613783016432936,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005750507713900795,
      "loss": 1.1607,
      "step": 26284
    },
    {
      "epoch": 2.261464337950615,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005750240369005086,
      "loss": 1.2003,
      "step": 26285
    },
    {
      "epoch": 2.261550374257937,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005749973021915071,
      "loss": 1.1503,
      "step": 26286
    },
    {
      "epoch": 2.2616364105652584,
      "grad_norm": 1.125,
      "learning_rate": 0.0005749705672631531,
      "loss": 1.1198,
      "step": 26287
    },
    {
      "epoch": 2.26172244687258,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005749438321155246,
      "loss": 1.599,
      "step": 26288
    },
    {
      "epoch": 2.2618084831799017,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005749170967487002,
      "loss": 1.4034,
      "step": 26289
    },
    {
      "epoch": 2.2618945194872238,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005748903611627576,
      "loss": 1.186,
      "step": 26290
    },
    {
      "epoch": 2.2619805557945454,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005748636253577754,
      "loss": 1.4754,
      "step": 26291
    },
    {
      "epoch": 2.262066592101867,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005748368893338317,
      "loss": 1.9884,
      "step": 26292
    },
    {
      "epoch": 2.2621526284091886,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005748101530910046,
      "loss": 0.7067,
      "step": 26293
    },
    {
      "epoch": 2.2622386647165103,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005747834166293724,
      "loss": 1.1493,
      "step": 26294
    },
    {
      "epoch": 2.262324701023832,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005747566799490132,
      "loss": 1.394,
      "step": 26295
    },
    {
      "epoch": 2.2624107373311535,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005747299430500051,
      "loss": 1.495,
      "step": 26296
    },
    {
      "epoch": 2.2624967736384756,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005747032059324268,
      "loss": 0.9852,
      "step": 26297
    },
    {
      "epoch": 2.262582809945797,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005746764685963559,
      "loss": 1.747,
      "step": 26298
    },
    {
      "epoch": 2.262668846253119,
      "grad_norm": 1.171875,
      "learning_rate": 0.000574649731041871,
      "loss": 1.2815,
      "step": 26299
    },
    {
      "epoch": 2.2627548825604404,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00057462299326905,
      "loss": 1.0949,
      "step": 26300
    },
    {
      "epoch": 2.262840918867762,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005745962552779713,
      "loss": 1.168,
      "step": 26301
    },
    {
      "epoch": 2.262926955175084,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005745695170687132,
      "loss": 0.7232,
      "step": 26302
    },
    {
      "epoch": 2.2630129914824058,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005745427786413535,
      "loss": 0.5861,
      "step": 26303
    },
    {
      "epoch": 2.2630990277897274,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005745160399959709,
      "loss": 1.6588,
      "step": 26304
    },
    {
      "epoch": 2.263185064097049,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005744893011326433,
      "loss": 0.9913,
      "step": 26305
    },
    {
      "epoch": 2.2632711004043706,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000574462562051449,
      "loss": 1.7564,
      "step": 26306
    },
    {
      "epoch": 2.2633571367116923,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005744358227524661,
      "loss": 1.1131,
      "step": 26307
    },
    {
      "epoch": 2.263443173019014,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005744090832357729,
      "loss": 1.0786,
      "step": 26308
    },
    {
      "epoch": 2.263529209326336,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005743823435014477,
      "loss": 1.1708,
      "step": 26309
    },
    {
      "epoch": 2.2636152456336576,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005743556035495686,
      "loss": 1.0464,
      "step": 26310
    },
    {
      "epoch": 2.263701281940979,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005743288633802137,
      "loss": 1.7459,
      "step": 26311
    },
    {
      "epoch": 2.263787318248301,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005743021229934615,
      "loss": 1.3336,
      "step": 26312
    },
    {
      "epoch": 2.2638733545556224,
      "grad_norm": 1.5,
      "learning_rate": 0.0005742753823893901,
      "loss": 1.6789,
      "step": 26313
    },
    {
      "epoch": 2.263959390862944,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005742486415680775,
      "loss": 1.5947,
      "step": 26314
    },
    {
      "epoch": 2.2640454271702657,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000574221900529602,
      "loss": 0.4259,
      "step": 26315
    },
    {
      "epoch": 2.2641314634775878,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000574195159274042,
      "loss": 0.7387,
      "step": 26316
    },
    {
      "epoch": 2.2642174997849094,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005741684178014756,
      "loss": 0.6834,
      "step": 26317
    },
    {
      "epoch": 2.264303536092231,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005741416761119809,
      "loss": 1.1664,
      "step": 26318
    },
    {
      "epoch": 2.2643895723995526,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005741149342056363,
      "loss": 1.1164,
      "step": 26319
    },
    {
      "epoch": 2.2644756087068743,
      "grad_norm": 0.921875,
      "learning_rate": 0.00057408819208252,
      "loss": 1.0104,
      "step": 26320
    },
    {
      "epoch": 2.264561645014196,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00057406144974271,
      "loss": 1.529,
      "step": 26321
    },
    {
      "epoch": 2.2646476813215175,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005740347071862848,
      "loss": 1.3907,
      "step": 26322
    },
    {
      "epoch": 2.2647337176288396,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005740079644133223,
      "loss": 0.9651,
      "step": 26323
    },
    {
      "epoch": 2.264819753936161,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005739812214239011,
      "loss": 1.647,
      "step": 26324
    },
    {
      "epoch": 2.264905790243483,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005739544782180991,
      "loss": 1.5361,
      "step": 26325
    },
    {
      "epoch": 2.2649918265508044,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005739277347959946,
      "loss": 1.4502,
      "step": 26326
    },
    {
      "epoch": 2.265077862858126,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005739009911576658,
      "loss": 0.7252,
      "step": 26327
    },
    {
      "epoch": 2.2651638991654477,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005738742473031911,
      "loss": 0.9288,
      "step": 26328
    },
    {
      "epoch": 2.2652499354727693,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005738475032326486,
      "loss": 1.0395,
      "step": 26329
    },
    {
      "epoch": 2.2653359717800914,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005738207589461164,
      "loss": 1.3182,
      "step": 26330
    },
    {
      "epoch": 2.265422008087413,
      "grad_norm": 1.390625,
      "learning_rate": 0.000573794014443673,
      "loss": 0.8208,
      "step": 26331
    },
    {
      "epoch": 2.2655080443947346,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005737672697253964,
      "loss": 0.9633,
      "step": 26332
    },
    {
      "epoch": 2.2655940807020563,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005737405247913649,
      "loss": 1.7247,
      "step": 26333
    },
    {
      "epoch": 2.265680117009378,
      "grad_norm": 2.015625,
      "learning_rate": 0.0005737137796416566,
      "loss": 1.613,
      "step": 26334
    },
    {
      "epoch": 2.2657661533166995,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005736870342763499,
      "loss": 0.9342,
      "step": 26335
    },
    {
      "epoch": 2.265852189624021,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000573660288695523,
      "loss": 0.7992,
      "step": 26336
    },
    {
      "epoch": 2.265938225931343,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005736335428992539,
      "loss": 1.8035,
      "step": 26337
    },
    {
      "epoch": 2.266024262238665,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005736067968876213,
      "loss": 1.3111,
      "step": 26338
    },
    {
      "epoch": 2.2661102985459864,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005735800506607029,
      "loss": 1.2588,
      "step": 26339
    },
    {
      "epoch": 2.266196334853308,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005735533042185774,
      "loss": 0.8753,
      "step": 26340
    },
    {
      "epoch": 2.2662823711606297,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005735265575613225,
      "loss": 0.9389,
      "step": 26341
    },
    {
      "epoch": 2.2663684074679513,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005734998106890169,
      "loss": 2.2489,
      "step": 26342
    },
    {
      "epoch": 2.266454443775273,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005734730636017387,
      "loss": 0.4782,
      "step": 26343
    },
    {
      "epoch": 2.266540480082595,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005734463162995659,
      "loss": 1.3568,
      "step": 26344
    },
    {
      "epoch": 2.2666265163899166,
      "grad_norm": 1.171875,
      "learning_rate": 0.000573419568782577,
      "loss": 1.67,
      "step": 26345
    },
    {
      "epoch": 2.2667125526972383,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005733928210508501,
      "loss": 1.1743,
      "step": 26346
    },
    {
      "epoch": 2.26679858900456,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005733660731044635,
      "loss": 1.0324,
      "step": 26347
    },
    {
      "epoch": 2.2668846253118815,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005733393249434955,
      "loss": 1.7471,
      "step": 26348
    },
    {
      "epoch": 2.266970661619203,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005733125765680241,
      "loss": 1.0746,
      "step": 26349
    },
    {
      "epoch": 2.2670566979265248,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005732858279781277,
      "loss": 1.9637,
      "step": 26350
    },
    {
      "epoch": 2.267142734233847,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005732590791738846,
      "loss": 1.1126,
      "step": 26351
    },
    {
      "epoch": 2.2672287705411684,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005732323301553728,
      "loss": 2.0716,
      "step": 26352
    },
    {
      "epoch": 2.26731480684849,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005732055809226707,
      "loss": 1.6822,
      "step": 26353
    },
    {
      "epoch": 2.2674008431558117,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005731788314758566,
      "loss": 0.6021,
      "step": 26354
    },
    {
      "epoch": 2.2674868794631333,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005731520818150087,
      "loss": 2.0456,
      "step": 26355
    },
    {
      "epoch": 2.267572915770455,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000573125331940205,
      "loss": 1.0521,
      "step": 26356
    },
    {
      "epoch": 2.267658952077777,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005730985818515241,
      "loss": 1.3746,
      "step": 26357
    },
    {
      "epoch": 2.2677449883850986,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000573071831549044,
      "loss": 1.2476,
      "step": 26358
    },
    {
      "epoch": 2.2678310246924203,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005730450810328429,
      "loss": 1.4924,
      "step": 26359
    },
    {
      "epoch": 2.267917060999742,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005730183303029993,
      "loss": 0.7775,
      "step": 26360
    },
    {
      "epoch": 2.2680030973070635,
      "grad_norm": 0.75,
      "learning_rate": 0.0005729915793595912,
      "loss": 0.6002,
      "step": 26361
    },
    {
      "epoch": 2.268089133614385,
      "grad_norm": 2.28125,
      "learning_rate": 0.000572964828202697,
      "loss": 1.4765,
      "step": 26362
    },
    {
      "epoch": 2.268175169921707,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005729380768323948,
      "loss": 1.1311,
      "step": 26363
    },
    {
      "epoch": 2.268261206229029,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005729113252487628,
      "loss": 1.3027,
      "step": 26364
    },
    {
      "epoch": 2.2683472425363504,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005728845734518795,
      "loss": 1.6954,
      "step": 26365
    },
    {
      "epoch": 2.268433278843672,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005728578214418231,
      "loss": 1.0088,
      "step": 26366
    },
    {
      "epoch": 2.2685193151509937,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005728310692186715,
      "loss": 1.1656,
      "step": 26367
    },
    {
      "epoch": 2.2686053514583153,
      "grad_norm": 1.375,
      "learning_rate": 0.0005728043167825033,
      "loss": 1.4931,
      "step": 26368
    },
    {
      "epoch": 2.268691387765637,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005727775641333968,
      "loss": 1.021,
      "step": 26369
    },
    {
      "epoch": 2.268777424072959,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005727508112714299,
      "loss": 1.4181,
      "step": 26370
    },
    {
      "epoch": 2.2688634603802806,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005727240581966809,
      "loss": 1.2408,
      "step": 26371
    },
    {
      "epoch": 2.2689494966876023,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005726973049092284,
      "loss": 0.9739,
      "step": 26372
    },
    {
      "epoch": 2.269035532994924,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005726705514091504,
      "loss": 0.5246,
      "step": 26373
    },
    {
      "epoch": 2.2691215693022455,
      "grad_norm": 1.0,
      "learning_rate": 0.0005726437976965251,
      "loss": 2.1953,
      "step": 26374
    },
    {
      "epoch": 2.269207605609567,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005726170437714308,
      "loss": 1.5844,
      "step": 26375
    },
    {
      "epoch": 2.2692936419168888,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005725902896339458,
      "loss": 1.0881,
      "step": 26376
    },
    {
      "epoch": 2.269379678224211,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005725635352841484,
      "loss": 1.11,
      "step": 26377
    },
    {
      "epoch": 2.2694657145315325,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005725367807221166,
      "loss": 0.899,
      "step": 26378
    },
    {
      "epoch": 2.269551750838854,
      "grad_norm": 1.984375,
      "learning_rate": 0.000572510025947929,
      "loss": 1.3435,
      "step": 26379
    },
    {
      "epoch": 2.2696377871461757,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005724832709616636,
      "loss": 0.8366,
      "step": 26380
    },
    {
      "epoch": 2.2697238234534973,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005724565157633987,
      "loss": 1.422,
      "step": 26381
    },
    {
      "epoch": 2.269809859760819,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005724297603532125,
      "loss": 1.1415,
      "step": 26382
    },
    {
      "epoch": 2.2698958960681406,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005724030047311834,
      "loss": 1.6186,
      "step": 26383
    },
    {
      "epoch": 2.2699819323754626,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005723762488973898,
      "loss": 1.0774,
      "step": 26384
    },
    {
      "epoch": 2.2700679686827843,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005723494928519097,
      "loss": 1.0389,
      "step": 26385
    },
    {
      "epoch": 2.270154004990106,
      "grad_norm": 1.5625,
      "learning_rate": 0.0005723227365948212,
      "loss": 1.8031,
      "step": 26386
    },
    {
      "epoch": 2.2702400412974275,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005722959801262029,
      "loss": 1.3412,
      "step": 26387
    },
    {
      "epoch": 2.270326077604749,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005722692234461328,
      "loss": 0.7516,
      "step": 26388
    },
    {
      "epoch": 2.2704121139120708,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005722424665546893,
      "loss": 1.2735,
      "step": 26389
    },
    {
      "epoch": 2.2704981502193924,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005722157094519507,
      "loss": 1.3887,
      "step": 26390
    },
    {
      "epoch": 2.2705841865267145,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005721889521379952,
      "loss": 1.649,
      "step": 26391
    },
    {
      "epoch": 2.270670222834036,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005721621946129011,
      "loss": 1.6442,
      "step": 26392
    },
    {
      "epoch": 2.2707562591413577,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005721354368767466,
      "loss": 1.878,
      "step": 26393
    },
    {
      "epoch": 2.2708422954486793,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005721086789296098,
      "loss": 1.2114,
      "step": 26394
    },
    {
      "epoch": 2.270928331756001,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005720819207715695,
      "loss": 0.9253,
      "step": 26395
    },
    {
      "epoch": 2.2710143680633226,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005720551624027033,
      "loss": 0.7907,
      "step": 26396
    },
    {
      "epoch": 2.271100404370644,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005720284038230897,
      "loss": 0.9424,
      "step": 26397
    },
    {
      "epoch": 2.2711864406779663,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005720016450328073,
      "loss": 1.7318,
      "step": 26398
    },
    {
      "epoch": 2.271272476985288,
      "grad_norm": 0.75,
      "learning_rate": 0.000571974886031934,
      "loss": 1.0048,
      "step": 26399
    },
    {
      "epoch": 2.2713585132926095,
      "grad_norm": 1.046875,
      "learning_rate": 0.000571948126820548,
      "loss": 1.2958,
      "step": 26400
    },
    {
      "epoch": 2.271444549599931,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005719213673987277,
      "loss": 1.1331,
      "step": 26401
    },
    {
      "epoch": 2.2715305859072528,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005718946077665516,
      "loss": 1.081,
      "step": 26402
    },
    {
      "epoch": 2.2716166222145744,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005718678479240978,
      "loss": 1.0964,
      "step": 26403
    },
    {
      "epoch": 2.271702658521896,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005718410878714443,
      "loss": 1.0479,
      "step": 26404
    },
    {
      "epoch": 2.271788694829218,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005718143276086696,
      "loss": 1.0499,
      "step": 26405
    },
    {
      "epoch": 2.2718747311365397,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005717875671358521,
      "loss": 0.7471,
      "step": 26406
    },
    {
      "epoch": 2.2719607674438613,
      "grad_norm": 0.96875,
      "learning_rate": 0.00057176080645307,
      "loss": 0.879,
      "step": 26407
    },
    {
      "epoch": 2.272046803751183,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005717340455604012,
      "loss": 1.1326,
      "step": 26408
    },
    {
      "epoch": 2.2721328400585046,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005717072844579245,
      "loss": 1.0747,
      "step": 26409
    },
    {
      "epoch": 2.272218876365826,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005716805231457179,
      "loss": 0.9617,
      "step": 26410
    },
    {
      "epoch": 2.272304912673148,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005716537616238598,
      "loss": 0.9698,
      "step": 26411
    },
    {
      "epoch": 2.27239094898047,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005716269998924282,
      "loss": 0.9784,
      "step": 26412
    },
    {
      "epoch": 2.2724769852877915,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005716002379515016,
      "loss": 0.9549,
      "step": 26413
    },
    {
      "epoch": 2.272563021595113,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005715734758011584,
      "loss": 1.0705,
      "step": 26414
    },
    {
      "epoch": 2.2726490579024348,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005715467134414765,
      "loss": 0.7966,
      "step": 26415
    },
    {
      "epoch": 2.2727350942097564,
      "grad_norm": 1.375,
      "learning_rate": 0.0005715199508725344,
      "loss": 0.9469,
      "step": 26416
    },
    {
      "epoch": 2.272821130517078,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005714931880944104,
      "loss": 0.8184,
      "step": 26417
    },
    {
      "epoch": 2.2729071668244,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005714664251071828,
      "loss": 1.2708,
      "step": 26418
    },
    {
      "epoch": 2.2729932031317217,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0005714396619109297,
      "loss": 1.8697,
      "step": 26419
    },
    {
      "epoch": 2.2730792394390433,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005714128985057294,
      "loss": 1.2693,
      "step": 26420
    },
    {
      "epoch": 2.273165275746365,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005713861348916603,
      "loss": 1.3656,
      "step": 26421
    },
    {
      "epoch": 2.2732513120536866,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005713593710688009,
      "loss": 0.8032,
      "step": 26422
    },
    {
      "epoch": 2.273337348361008,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005713326070372289,
      "loss": 0.4816,
      "step": 26423
    },
    {
      "epoch": 2.2734233846683303,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005713058427970231,
      "loss": 1.3012,
      "step": 26424
    },
    {
      "epoch": 2.273509420975652,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005712790783482615,
      "loss": 1.1303,
      "step": 26425
    },
    {
      "epoch": 2.2735954572829735,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005712523136910224,
      "loss": 0.7979,
      "step": 26426
    },
    {
      "epoch": 2.273681493590295,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005712255488253842,
      "loss": 1.0754,
      "step": 26427
    },
    {
      "epoch": 2.2737675298976168,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005711987837514252,
      "loss": 0.8829,
      "step": 26428
    },
    {
      "epoch": 2.2738535662049384,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005711720184692237,
      "loss": 0.7866,
      "step": 26429
    },
    {
      "epoch": 2.27393960251226,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005711452529788576,
      "loss": 0.9273,
      "step": 26430
    },
    {
      "epoch": 2.274025638819582,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005711184872804056,
      "loss": 1.1298,
      "step": 26431
    },
    {
      "epoch": 2.2741116751269037,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005710917213739458,
      "loss": 1.4425,
      "step": 26432
    },
    {
      "epoch": 2.2741977114342253,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005710649552595567,
      "loss": 1.5806,
      "step": 26433
    },
    {
      "epoch": 2.274283747741547,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005710381889373162,
      "loss": 0.4501,
      "step": 26434
    },
    {
      "epoch": 2.2743697840488686,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000571011422407303,
      "loss": 1.0016,
      "step": 26435
    },
    {
      "epoch": 2.27445582035619,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005709846556695952,
      "loss": 1.3799,
      "step": 26436
    },
    {
      "epoch": 2.274541856663512,
      "grad_norm": 1.140625,
      "learning_rate": 0.000570957888724271,
      "loss": 1.0323,
      "step": 26437
    },
    {
      "epoch": 2.274627892970834,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005709311215714085,
      "loss": 0.9904,
      "step": 26438
    },
    {
      "epoch": 2.2747139292781555,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005709043542110868,
      "loss": 0.4385,
      "step": 26439
    },
    {
      "epoch": 2.274799965585477,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005708775866433834,
      "loss": 1.3143,
      "step": 26440
    },
    {
      "epoch": 2.2748860018927988,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005708508188683769,
      "loss": 1.5867,
      "step": 26441
    },
    {
      "epoch": 2.2749720382001204,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005708240508861454,
      "loss": 1.1901,
      "step": 26442
    },
    {
      "epoch": 2.275058074507442,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005707972826967674,
      "loss": 1.7554,
      "step": 26443
    },
    {
      "epoch": 2.2751441108147636,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005707705143003213,
      "loss": 1.2499,
      "step": 26444
    },
    {
      "epoch": 2.2752301471220857,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005707437456968851,
      "loss": 1.565,
      "step": 26445
    },
    {
      "epoch": 2.2753161834294073,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000570716976886537,
      "loss": 1.0187,
      "step": 26446
    },
    {
      "epoch": 2.275402219736729,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005706902078693556,
      "loss": 1.6494,
      "step": 26447
    },
    {
      "epoch": 2.2754882560440506,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005706634386454193,
      "loss": 1.0791,
      "step": 26448
    },
    {
      "epoch": 2.275574292351372,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000570636669214806,
      "loss": 1.0158,
      "step": 26449
    },
    {
      "epoch": 2.275660328658694,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005706098995775941,
      "loss": 0.4691,
      "step": 26450
    },
    {
      "epoch": 2.2757463649660155,
      "grad_norm": 8.8125,
      "learning_rate": 0.0005705831297338622,
      "loss": 1.4812,
      "step": 26451
    },
    {
      "epoch": 2.2758324012733375,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005705563596836883,
      "loss": 0.8054,
      "step": 26452
    },
    {
      "epoch": 2.275918437580659,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005705295894271506,
      "loss": 0.9169,
      "step": 26453
    },
    {
      "epoch": 2.2760044738879808,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005705028189643277,
      "loss": 1.4767,
      "step": 26454
    },
    {
      "epoch": 2.2760905101953024,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005704760482952979,
      "loss": 0.927,
      "step": 26455
    },
    {
      "epoch": 2.276176546502624,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000570449277420139,
      "loss": 0.9255,
      "step": 26456
    },
    {
      "epoch": 2.2762625828099456,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005704225063389299,
      "loss": 1.0967,
      "step": 26457
    },
    {
      "epoch": 2.2763486191172673,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005703957350517487,
      "loss": 1.2254,
      "step": 26458
    },
    {
      "epoch": 2.2764346554245893,
      "grad_norm": 1.5,
      "learning_rate": 0.0005703689635586736,
      "loss": 2.0457,
      "step": 26459
    },
    {
      "epoch": 2.276520691731911,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005703421918597831,
      "loss": 1.0538,
      "step": 26460
    },
    {
      "epoch": 2.2766067280392326,
      "grad_norm": 1.375,
      "learning_rate": 0.0005703154199551551,
      "loss": 1.4047,
      "step": 26461
    },
    {
      "epoch": 2.276692764346554,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005702886478448684,
      "loss": 0.9135,
      "step": 26462
    },
    {
      "epoch": 2.276778800653876,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005702618755290012,
      "loss": 1.5958,
      "step": 26463
    },
    {
      "epoch": 2.2768648369611975,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005702351030076313,
      "loss": 1.0576,
      "step": 26464
    },
    {
      "epoch": 2.276950873268519,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0005702083302808376,
      "loss": 0.7725,
      "step": 26465
    },
    {
      "epoch": 2.277036909575841,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005701815573486981,
      "loss": 0.7645,
      "step": 26466
    },
    {
      "epoch": 2.2771229458831628,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005701547842112913,
      "loss": 1.9896,
      "step": 26467
    },
    {
      "epoch": 2.2772089821904844,
      "grad_norm": 1.375,
      "learning_rate": 0.0005701280108686954,
      "loss": 1.8732,
      "step": 26468
    },
    {
      "epoch": 2.277295018497806,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005701012373209886,
      "loss": 1.6383,
      "step": 26469
    },
    {
      "epoch": 2.2773810548051276,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005700744635682494,
      "loss": 1.0473,
      "step": 26470
    },
    {
      "epoch": 2.2774670911124493,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000570047689610556,
      "loss": 0.635,
      "step": 26471
    },
    {
      "epoch": 2.277553127419771,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005700209154479867,
      "loss": 1.3777,
      "step": 26472
    },
    {
      "epoch": 2.277639163727093,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005699941410806199,
      "loss": 1.1926,
      "step": 26473
    },
    {
      "epoch": 2.2777252000344146,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005699673665085339,
      "loss": 1.0058,
      "step": 26474
    },
    {
      "epoch": 2.277811236341736,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005699405917318068,
      "loss": 1.5137,
      "step": 26475
    },
    {
      "epoch": 2.277897272649058,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005699138167505172,
      "loss": 0.9898,
      "step": 26476
    },
    {
      "epoch": 2.2779833089563795,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005698870415647431,
      "loss": 1.1438,
      "step": 26477
    },
    {
      "epoch": 2.2780693452637015,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005698602661745632,
      "loss": 1.4836,
      "step": 26478
    },
    {
      "epoch": 2.278155381571023,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005698334905800555,
      "loss": 1.0855,
      "step": 26479
    },
    {
      "epoch": 2.2782414178783448,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005698067147812986,
      "loss": 1.069,
      "step": 26480
    },
    {
      "epoch": 2.2783274541856664,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005697799387783704,
      "loss": 0.8752,
      "step": 26481
    },
    {
      "epoch": 2.278413490492988,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005697531625713496,
      "loss": 1.8237,
      "step": 26482
    },
    {
      "epoch": 2.2784995268003096,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005697263861603142,
      "loss": 1.548,
      "step": 26483
    },
    {
      "epoch": 2.2785855631076313,
      "grad_norm": 1.125,
      "learning_rate": 0.0005696996095453428,
      "loss": 1.6101,
      "step": 26484
    },
    {
      "epoch": 2.2786715994149533,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005696728327265137,
      "loss": 0.5521,
      "step": 26485
    },
    {
      "epoch": 2.278757635722275,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000569646055703905,
      "loss": 1.279,
      "step": 26486
    },
    {
      "epoch": 2.2788436720295966,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005696192784775951,
      "loss": 0.8831,
      "step": 26487
    },
    {
      "epoch": 2.278929708336918,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005695925010476625,
      "loss": 1.2187,
      "step": 26488
    },
    {
      "epoch": 2.27901574464424,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005695657234141852,
      "loss": 1.3869,
      "step": 26489
    },
    {
      "epoch": 2.2791017809515615,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005695389455772417,
      "loss": 1.3306,
      "step": 26490
    },
    {
      "epoch": 2.279187817258883,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005695121675369103,
      "loss": 0.887,
      "step": 26491
    },
    {
      "epoch": 2.279273853566205,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005694853892932695,
      "loss": 1.8182,
      "step": 26492
    },
    {
      "epoch": 2.279359889873527,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005694586108463972,
      "loss": 1.1476,
      "step": 26493
    },
    {
      "epoch": 2.2794459261808484,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005694318321963722,
      "loss": 2.0641,
      "step": 26494
    },
    {
      "epoch": 2.27953196248817,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005694050533432724,
      "loss": 1.4281,
      "step": 26495
    },
    {
      "epoch": 2.2796179987954917,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005693782742871765,
      "loss": 1.0358,
      "step": 26496
    },
    {
      "epoch": 2.2797040351028133,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005693514950281626,
      "loss": 1.3882,
      "step": 26497
    },
    {
      "epoch": 2.279790071410135,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005693247155663091,
      "loss": 1.2687,
      "step": 26498
    },
    {
      "epoch": 2.279876107717457,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005692979359016941,
      "loss": 1.7954,
      "step": 26499
    },
    {
      "epoch": 2.2799621440247786,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005692711560343962,
      "loss": 0.6911,
      "step": 26500
    },
    {
      "epoch": 2.2800481803321,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005692443759644938,
      "loss": 1.3712,
      "step": 26501
    },
    {
      "epoch": 2.280134216639422,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005692175956920649,
      "loss": 0.8155,
      "step": 26502
    },
    {
      "epoch": 2.2802202529467435,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000569190815217188,
      "loss": 1.1128,
      "step": 26503
    },
    {
      "epoch": 2.280306289254065,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005691640345399416,
      "loss": 1.2465,
      "step": 26504
    },
    {
      "epoch": 2.2803923255613867,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005691372536604036,
      "loss": 0.9876,
      "step": 26505
    },
    {
      "epoch": 2.280478361868709,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005691104725786527,
      "loss": 1.1395,
      "step": 26506
    },
    {
      "epoch": 2.2805643981760304,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005690836912947672,
      "loss": 0.8828,
      "step": 26507
    },
    {
      "epoch": 2.280650434483352,
      "grad_norm": 1.125,
      "learning_rate": 0.0005690569098088251,
      "loss": 1.6629,
      "step": 26508
    },
    {
      "epoch": 2.2807364707906737,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005690301281209052,
      "loss": 0.9639,
      "step": 26509
    },
    {
      "epoch": 2.2808225070979953,
      "grad_norm": 1.0,
      "learning_rate": 0.0005690033462310855,
      "loss": 1.4276,
      "step": 26510
    },
    {
      "epoch": 2.280908543405317,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005689765641394445,
      "loss": 0.9326,
      "step": 26511
    },
    {
      "epoch": 2.2809945797126385,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005689497818460602,
      "loss": 0.6555,
      "step": 26512
    },
    {
      "epoch": 2.2810806160199606,
      "grad_norm": 3.59375,
      "learning_rate": 0.0005689229993510115,
      "loss": 1.1226,
      "step": 26513
    },
    {
      "epoch": 2.281166652327282,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005688962166543763,
      "loss": 1.9313,
      "step": 26514
    },
    {
      "epoch": 2.281252688634604,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005688694337562331,
      "loss": 1.3548,
      "step": 26515
    },
    {
      "epoch": 2.2813387249419255,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005688426506566603,
      "loss": 1.0234,
      "step": 26516
    },
    {
      "epoch": 2.281424761249247,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000568815867355736,
      "loss": 1.0702,
      "step": 26517
    },
    {
      "epoch": 2.2815107975565687,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005687890838535387,
      "loss": 0.8586,
      "step": 26518
    },
    {
      "epoch": 2.2815968338638903,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005687623001501467,
      "loss": 1.1259,
      "step": 26519
    },
    {
      "epoch": 2.2816828701712124,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005687355162456384,
      "loss": 1.5381,
      "step": 26520
    },
    {
      "epoch": 2.281768906478534,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000568708732140092,
      "loss": 1.6365,
      "step": 26521
    },
    {
      "epoch": 2.2818549427858557,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005686819478335861,
      "loss": 1.4922,
      "step": 26522
    },
    {
      "epoch": 2.2819409790931773,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005686551633261988,
      "loss": 1.1657,
      "step": 26523
    },
    {
      "epoch": 2.282027015400499,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0005686283786180083,
      "loss": 0.4773,
      "step": 26524
    },
    {
      "epoch": 2.2821130517078205,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005686015937090932,
      "loss": 1.2158,
      "step": 26525
    },
    {
      "epoch": 2.282199088015142,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005685748085995319,
      "loss": 1.4951,
      "step": 26526
    },
    {
      "epoch": 2.282285124322464,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005685480232894026,
      "loss": 0.7606,
      "step": 26527
    },
    {
      "epoch": 2.282371160629786,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005685212377787836,
      "loss": 0.7913,
      "step": 26528
    },
    {
      "epoch": 2.2824571969371075,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005684944520677534,
      "loss": 1.5989,
      "step": 26529
    },
    {
      "epoch": 2.282543233244429,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005684676661563902,
      "loss": 1.0107,
      "step": 26530
    },
    {
      "epoch": 2.2826292695517507,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005684408800447723,
      "loss": 1.5473,
      "step": 26531
    },
    {
      "epoch": 2.2827153058590723,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005684140937329782,
      "loss": 1.7604,
      "step": 26532
    },
    {
      "epoch": 2.2828013421663944,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005683873072210862,
      "loss": 0.7215,
      "step": 26533
    },
    {
      "epoch": 2.282887378473716,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005683605205091747,
      "loss": 2.1371,
      "step": 26534
    },
    {
      "epoch": 2.2829734147810377,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005683337335973217,
      "loss": 1.3726,
      "step": 26535
    },
    {
      "epoch": 2.2830594510883593,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005683069464856062,
      "loss": 1.317,
      "step": 26536
    },
    {
      "epoch": 2.283145487395681,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005682801591741061,
      "loss": 1.7345,
      "step": 26537
    },
    {
      "epoch": 2.2832315237030025,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005682533716628997,
      "loss": 0.7421,
      "step": 26538
    },
    {
      "epoch": 2.2833175600103246,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005682265839520653,
      "loss": 0.7583,
      "step": 26539
    },
    {
      "epoch": 2.2834035963176462,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005681997960416816,
      "loss": 0.9679,
      "step": 26540
    },
    {
      "epoch": 2.283489632624968,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005681730079318268,
      "loss": 1.1425,
      "step": 26541
    },
    {
      "epoch": 2.2835756689322895,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005681462196225792,
      "loss": 0.9432,
      "step": 26542
    },
    {
      "epoch": 2.283661705239611,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000568119431114017,
      "loss": 1.0743,
      "step": 26543
    },
    {
      "epoch": 2.2837477415469327,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005680926424062189,
      "loss": 1.5005,
      "step": 26544
    },
    {
      "epoch": 2.2838337778542543,
      "grad_norm": 1.109375,
      "learning_rate": 0.000568065853499263,
      "loss": 0.8459,
      "step": 26545
    },
    {
      "epoch": 2.2839198141615764,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005680390643932276,
      "loss": 1.7733,
      "step": 26546
    },
    {
      "epoch": 2.284005850468898,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005680122750881913,
      "loss": 1.6143,
      "step": 26547
    },
    {
      "epoch": 2.2840918867762197,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005679854855842322,
      "loss": 1.6845,
      "step": 26548
    },
    {
      "epoch": 2.2841779230835413,
      "grad_norm": 1.109375,
      "learning_rate": 0.000567958695881429,
      "loss": 1.2589,
      "step": 26549
    },
    {
      "epoch": 2.284263959390863,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005679319059798595,
      "loss": 1.4956,
      "step": 26550
    },
    {
      "epoch": 2.2843499956981845,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005679051158796027,
      "loss": 1.4828,
      "step": 26551
    },
    {
      "epoch": 2.284436032005506,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005678783255807366,
      "loss": 0.9797,
      "step": 26552
    },
    {
      "epoch": 2.2845220683128282,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005678515350833396,
      "loss": 1.6773,
      "step": 26553
    },
    {
      "epoch": 2.28460810462015,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005678247443874899,
      "loss": 1.0496,
      "step": 26554
    },
    {
      "epoch": 2.2846941409274715,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005677979534932661,
      "loss": 1.51,
      "step": 26555
    },
    {
      "epoch": 2.284780177234793,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005677711624007465,
      "loss": 1.3915,
      "step": 26556
    },
    {
      "epoch": 2.2848662135421147,
      "grad_norm": 1.25,
      "learning_rate": 0.0005677443711100094,
      "loss": 1.815,
      "step": 26557
    },
    {
      "epoch": 2.2849522498494363,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005677175796211332,
      "loss": 1.3036,
      "step": 26558
    },
    {
      "epoch": 2.285038286156758,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005676907879341962,
      "loss": 0.5329,
      "step": 26559
    },
    {
      "epoch": 2.28512432246408,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000567663996049277,
      "loss": 1.6403,
      "step": 26560
    },
    {
      "epoch": 2.2852103587714017,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005676372039664536,
      "loss": 0.8151,
      "step": 26561
    },
    {
      "epoch": 2.2852963950787233,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005676104116858046,
      "loss": 1.0557,
      "step": 26562
    },
    {
      "epoch": 2.285382431386045,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005675836192074082,
      "loss": 1.7088,
      "step": 26563
    },
    {
      "epoch": 2.2854684676933665,
      "grad_norm": 1.1875,
      "learning_rate": 0.000567556826531343,
      "loss": 1.8372,
      "step": 26564
    },
    {
      "epoch": 2.285554504000688,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005675300336576871,
      "loss": 1.1619,
      "step": 26565
    },
    {
      "epoch": 2.28564054030801,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000567503240586519,
      "loss": 0.698,
      "step": 26566
    },
    {
      "epoch": 2.285726576615332,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000567476447317917,
      "loss": 1.2423,
      "step": 26567
    },
    {
      "epoch": 2.2858126129226535,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005674496538519596,
      "loss": 1.2088,
      "step": 26568
    },
    {
      "epoch": 2.285898649229975,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000567422860188725,
      "loss": 1.1986,
      "step": 26569
    },
    {
      "epoch": 2.2859846855372967,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005673960663282918,
      "loss": 0.7623,
      "step": 26570
    },
    {
      "epoch": 2.2860707218446183,
      "grad_norm": 1.640625,
      "learning_rate": 0.000567369272270738,
      "loss": 1.3543,
      "step": 26571
    },
    {
      "epoch": 2.28615675815194,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005673424780161425,
      "loss": 0.6687,
      "step": 26572
    },
    {
      "epoch": 2.2862427944592616,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005673156835645831,
      "loss": 1.7279,
      "step": 26573
    },
    {
      "epoch": 2.2863288307665837,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005672888889161385,
      "loss": 1.3823,
      "step": 26574
    },
    {
      "epoch": 2.2864148670739053,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000567262094070887,
      "loss": 0.7609,
      "step": 26575
    },
    {
      "epoch": 2.286500903381227,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005672352990289068,
      "loss": 1.905,
      "step": 26576
    },
    {
      "epoch": 2.2865869396885485,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005672085037902766,
      "loss": 1.2367,
      "step": 26577
    },
    {
      "epoch": 2.28667297599587,
      "grad_norm": 1.0,
      "learning_rate": 0.0005671817083550745,
      "loss": 1.4205,
      "step": 26578
    },
    {
      "epoch": 2.286759012303192,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005671549127233791,
      "loss": 1.5028,
      "step": 26579
    },
    {
      "epoch": 2.2868450486105134,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005671281168952684,
      "loss": 1.1047,
      "step": 26580
    },
    {
      "epoch": 2.2869310849178355,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0005671013208708213,
      "loss": 0.7947,
      "step": 26581
    },
    {
      "epoch": 2.287017121225157,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005670745246501158,
      "loss": 0.7693,
      "step": 26582
    },
    {
      "epoch": 2.2871031575324787,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005670477282332302,
      "loss": 1.4829,
      "step": 26583
    },
    {
      "epoch": 2.2871891938398003,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005670209316202433,
      "loss": 1.9908,
      "step": 26584
    },
    {
      "epoch": 2.287275230147122,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005669941348112328,
      "loss": 0.8893,
      "step": 26585
    },
    {
      "epoch": 2.2873612664544436,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005669673378062779,
      "loss": 1.8081,
      "step": 26586
    },
    {
      "epoch": 2.2874473027617652,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005669405406054564,
      "loss": 1.9056,
      "step": 26587
    },
    {
      "epoch": 2.2875333390690873,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005669137432088468,
      "loss": 1.4619,
      "step": 26588
    },
    {
      "epoch": 2.287619375376409,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005668869456165275,
      "loss": 0.7635,
      "step": 26589
    },
    {
      "epoch": 2.2877054116837305,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000566860147828577,
      "loss": 0.6315,
      "step": 26590
    },
    {
      "epoch": 2.287791447991052,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005668333498450735,
      "loss": 0.6967,
      "step": 26591
    },
    {
      "epoch": 2.287877484298374,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005668065516660955,
      "loss": 1.2199,
      "step": 26592
    },
    {
      "epoch": 2.2879635206056954,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005667797532917214,
      "loss": 1.141,
      "step": 26593
    },
    {
      "epoch": 2.2880495569130175,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005667529547220294,
      "loss": 1.6321,
      "step": 26594
    },
    {
      "epoch": 2.288135593220339,
      "grad_norm": 1.03125,
      "learning_rate": 0.000566726155957098,
      "loss": 1.3814,
      "step": 26595
    },
    {
      "epoch": 2.2882216295276607,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005666993569970056,
      "loss": 0.8564,
      "step": 26596
    },
    {
      "epoch": 2.2883076658349824,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005666725578418306,
      "loss": 0.9812,
      "step": 26597
    },
    {
      "epoch": 2.288393702142304,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005666457584916513,
      "loss": 0.8028,
      "step": 26598
    },
    {
      "epoch": 2.2884797384496256,
      "grad_norm": 1.75,
      "learning_rate": 0.0005666189589465459,
      "loss": 1.4673,
      "step": 26599
    },
    {
      "epoch": 2.2885657747569477,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005665921592065932,
      "loss": 1.7079,
      "step": 26600
    },
    {
      "epoch": 2.2886518110642693,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005665653592718715,
      "loss": 1.5916,
      "step": 26601
    },
    {
      "epoch": 2.288737847371591,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005665385591424588,
      "loss": 1.5474,
      "step": 26602
    },
    {
      "epoch": 2.2888238836789125,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000566511758818434,
      "loss": 0.8668,
      "step": 26603
    },
    {
      "epoch": 2.288909919986234,
      "grad_norm": 1.46875,
      "learning_rate": 0.000566484958299875,
      "loss": 1.4217,
      "step": 26604
    },
    {
      "epoch": 2.288995956293556,
      "grad_norm": 1.375,
      "learning_rate": 0.0005664581575868606,
      "loss": 1.7378,
      "step": 26605
    },
    {
      "epoch": 2.2890819926008774,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005664313566794689,
      "loss": 1.0352,
      "step": 26606
    },
    {
      "epoch": 2.2891680289081995,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005664045555777786,
      "loss": 1.0602,
      "step": 26607
    },
    {
      "epoch": 2.289254065215521,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005663777542818677,
      "loss": 1.7588,
      "step": 26608
    },
    {
      "epoch": 2.2893401015228427,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005663509527918149,
      "loss": 0.7068,
      "step": 26609
    },
    {
      "epoch": 2.2894261378301644,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005663241511076984,
      "loss": 0.6392,
      "step": 26610
    },
    {
      "epoch": 2.289512174137486,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005662973492295966,
      "loss": 1.0159,
      "step": 26611
    },
    {
      "epoch": 2.2895982104448076,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005662705471575882,
      "loss": 1.4838,
      "step": 26612
    },
    {
      "epoch": 2.2896842467521292,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005662437448917512,
      "loss": 1.1721,
      "step": 26613
    },
    {
      "epoch": 2.2897702830594513,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005662169424321639,
      "loss": 0.8248,
      "step": 26614
    },
    {
      "epoch": 2.289856319366773,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005661901397789052,
      "loss": 1.227,
      "step": 26615
    },
    {
      "epoch": 2.2899423556740945,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0005661633369320532,
      "loss": 1.8553,
      "step": 26616
    },
    {
      "epoch": 2.290028391981416,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005661365338916861,
      "loss": 1.083,
      "step": 26617
    },
    {
      "epoch": 2.290114428288738,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005661097306578827,
      "loss": 1.0345,
      "step": 26618
    },
    {
      "epoch": 2.2902004645960594,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005660829272307212,
      "loss": 1.3689,
      "step": 26619
    },
    {
      "epoch": 2.290286500903381,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005660561236102799,
      "loss": 0.9763,
      "step": 26620
    },
    {
      "epoch": 2.290372537210703,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005660293197966373,
      "loss": 1.1597,
      "step": 26621
    },
    {
      "epoch": 2.2904585735180247,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005660025157898717,
      "loss": 0.8362,
      "step": 26622
    },
    {
      "epoch": 2.2905446098253464,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005659757115900616,
      "loss": 1.4214,
      "step": 26623
    },
    {
      "epoch": 2.290630646132668,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005659489071972855,
      "loss": 0.9148,
      "step": 26624
    },
    {
      "epoch": 2.2907166824399896,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005659221026116216,
      "loss": 0.9724,
      "step": 26625
    },
    {
      "epoch": 2.2908027187473112,
      "grad_norm": 2.328125,
      "learning_rate": 0.0005658952978331484,
      "loss": 0.9347,
      "step": 26626
    },
    {
      "epoch": 2.290888755054633,
      "grad_norm": 1.375,
      "learning_rate": 0.0005658684928619443,
      "loss": 1.0811,
      "step": 26627
    },
    {
      "epoch": 2.290974791361955,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005658416876980877,
      "loss": 0.9791,
      "step": 26628
    },
    {
      "epoch": 2.2910608276692765,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005658148823416568,
      "loss": 0.5993,
      "step": 26629
    },
    {
      "epoch": 2.291146863976598,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005657880767927303,
      "loss": 1.5399,
      "step": 26630
    },
    {
      "epoch": 2.29123290028392,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005657612710513864,
      "loss": 0.9312,
      "step": 26631
    },
    {
      "epoch": 2.2913189365912414,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005657344651177035,
      "loss": 1.2033,
      "step": 26632
    },
    {
      "epoch": 2.291404972898563,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005657076589917603,
      "loss": 1.3993,
      "step": 26633
    },
    {
      "epoch": 2.2914910092058847,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005656808526736349,
      "loss": 1.7761,
      "step": 26634
    },
    {
      "epoch": 2.2915770455132067,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005656540461634057,
      "loss": 0.7121,
      "step": 26635
    },
    {
      "epoch": 2.2916630818205284,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005656272394611513,
      "loss": 1.5787,
      "step": 26636
    },
    {
      "epoch": 2.29174911812785,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005656004325669498,
      "loss": 1.3676,
      "step": 26637
    },
    {
      "epoch": 2.2918351544351716,
      "grad_norm": 1.296875,
      "learning_rate": 0.00056557362548088,
      "loss": 1.3132,
      "step": 26638
    },
    {
      "epoch": 2.2919211907424932,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005655468182030199,
      "loss": 0.9093,
      "step": 26639
    },
    {
      "epoch": 2.292007227049815,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005655200107334483,
      "loss": 0.9161,
      "step": 26640
    },
    {
      "epoch": 2.2920932633571365,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005654932030722432,
      "loss": 1.6791,
      "step": 26641
    },
    {
      "epoch": 2.2921792996644585,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005654663952194834,
      "loss": 1.1448,
      "step": 26642
    },
    {
      "epoch": 2.29226533597178,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005654395871752471,
      "loss": 0.9739,
      "step": 26643
    },
    {
      "epoch": 2.292351372279102,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005654127789396126,
      "loss": 1.5433,
      "step": 26644
    },
    {
      "epoch": 2.2924374085864234,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005653859705126585,
      "loss": 1.1292,
      "step": 26645
    },
    {
      "epoch": 2.292523444893745,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005653591618944632,
      "loss": 1.9248,
      "step": 26646
    },
    {
      "epoch": 2.2926094812010667,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005653323530851051,
      "loss": 0.682,
      "step": 26647
    },
    {
      "epoch": 2.2926955175083883,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005653055440846625,
      "loss": 1.05,
      "step": 26648
    },
    {
      "epoch": 2.2927815538157104,
      "grad_norm": 1.5234375,
      "learning_rate": 0.000565278734893214,
      "loss": 1.6155,
      "step": 26649
    },
    {
      "epoch": 2.292867590123032,
      "grad_norm": 1.0,
      "learning_rate": 0.0005652519255108378,
      "loss": 1.2635,
      "step": 26650
    },
    {
      "epoch": 2.2929536264303536,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005652251159376124,
      "loss": 1.029,
      "step": 26651
    },
    {
      "epoch": 2.2930396627376752,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0005651983061736162,
      "loss": 1.4063,
      "step": 26652
    },
    {
      "epoch": 2.293125699044997,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005651714962189278,
      "loss": 1.6819,
      "step": 26653
    },
    {
      "epoch": 2.293211735352319,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005651446860736254,
      "loss": 0.9637,
      "step": 26654
    },
    {
      "epoch": 2.2932977716596405,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005651178757377873,
      "loss": 0.8168,
      "step": 26655
    },
    {
      "epoch": 2.293383807966962,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005650910652114921,
      "loss": 1.1082,
      "step": 26656
    },
    {
      "epoch": 2.293469844274284,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005650642544948183,
      "loss": 0.8337,
      "step": 26657
    },
    {
      "epoch": 2.2935558805816054,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005650374435878442,
      "loss": 0.8286,
      "step": 26658
    },
    {
      "epoch": 2.293641916888927,
      "grad_norm": 1.125,
      "learning_rate": 0.0005650106324906481,
      "loss": 0.7621,
      "step": 26659
    },
    {
      "epoch": 2.2937279531962487,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005649838212033086,
      "loss": 0.8713,
      "step": 26660
    },
    {
      "epoch": 2.2938139895035707,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005649570097259042,
      "loss": 0.8047,
      "step": 26661
    },
    {
      "epoch": 2.2939000258108924,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000564930198058513,
      "loss": 0.7038,
      "step": 26662
    },
    {
      "epoch": 2.293986062118214,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0005649033862012138,
      "loss": 0.7028,
      "step": 26663
    },
    {
      "epoch": 2.2940720984255356,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005648765741540847,
      "loss": 1.1777,
      "step": 26664
    },
    {
      "epoch": 2.2941581347328572,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005648497619172042,
      "loss": 1.7796,
      "step": 26665
    },
    {
      "epoch": 2.294244171040179,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005648229494906507,
      "loss": 1.0245,
      "step": 26666
    },
    {
      "epoch": 2.2943302073475005,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005647961368745028,
      "loss": 1.2867,
      "step": 26667
    },
    {
      "epoch": 2.2944162436548226,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005647693240688388,
      "loss": 1.1032,
      "step": 26668
    },
    {
      "epoch": 2.294502279962144,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005647425110737372,
      "loss": 0.8666,
      "step": 26669
    },
    {
      "epoch": 2.294588316269466,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005647156978892761,
      "loss": 1.0818,
      "step": 26670
    },
    {
      "epoch": 2.2946743525767874,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005646888845155344,
      "loss": 0.7725,
      "step": 26671
    },
    {
      "epoch": 2.294760388884109,
      "grad_norm": 1.375,
      "learning_rate": 0.0005646620709525902,
      "loss": 1.5306,
      "step": 26672
    },
    {
      "epoch": 2.2948464251914307,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000564635257200522,
      "loss": 1.7248,
      "step": 26673
    },
    {
      "epoch": 2.2949324614987523,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005646084432594083,
      "loss": 0.8459,
      "step": 26674
    },
    {
      "epoch": 2.2950184978060744,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005645816291293275,
      "loss": 1.1318,
      "step": 26675
    },
    {
      "epoch": 2.295104534113396,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000564554814810358,
      "loss": 0.8369,
      "step": 26676
    },
    {
      "epoch": 2.2951905704207176,
      "grad_norm": 1.375,
      "learning_rate": 0.0005645280003025781,
      "loss": 1.9062,
      "step": 26677
    },
    {
      "epoch": 2.2952766067280392,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005645011856060663,
      "loss": 0.8208,
      "step": 26678
    },
    {
      "epoch": 2.295362643035361,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005644743707209013,
      "loss": 0.5679,
      "step": 26679
    },
    {
      "epoch": 2.2954486793426825,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005644475556471612,
      "loss": 1.3374,
      "step": 26680
    },
    {
      "epoch": 2.295534715650004,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005644207403849244,
      "loss": 1.8223,
      "step": 26681
    },
    {
      "epoch": 2.295620751957326,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005643939249342696,
      "loss": 1.3242,
      "step": 26682
    },
    {
      "epoch": 2.295706788264648,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005643671092952752,
      "loss": 0.4742,
      "step": 26683
    },
    {
      "epoch": 2.2957928245719694,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005643402934680194,
      "loss": 0.7054,
      "step": 26684
    },
    {
      "epoch": 2.295878860879291,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005643134774525806,
      "loss": 1.3354,
      "step": 26685
    },
    {
      "epoch": 2.2959648971866127,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005642866612490377,
      "loss": 1.0132,
      "step": 26686
    },
    {
      "epoch": 2.2960509334939343,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005642598448574687,
      "loss": 1.0259,
      "step": 26687
    },
    {
      "epoch": 2.296136969801256,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005642330282779521,
      "loss": 0.8375,
      "step": 26688
    },
    {
      "epoch": 2.296223006108578,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005642062115105664,
      "loss": 1.505,
      "step": 26689
    },
    {
      "epoch": 2.2963090424158996,
      "grad_norm": 1.40625,
      "learning_rate": 0.00056417939455539,
      "loss": 1.2483,
      "step": 26690
    },
    {
      "epoch": 2.2963950787232212,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005641525774125014,
      "loss": 1.7253,
      "step": 26691
    },
    {
      "epoch": 2.296481115030543,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005641257600819789,
      "loss": 0.767,
      "step": 26692
    },
    {
      "epoch": 2.2965671513378645,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005640989425639011,
      "loss": 0.8396,
      "step": 26693
    },
    {
      "epoch": 2.296653187645186,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005640721248583464,
      "loss": 1.4676,
      "step": 26694
    },
    {
      "epoch": 2.2967392239525077,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005640453069653931,
      "loss": 0.7543,
      "step": 26695
    },
    {
      "epoch": 2.29682526025983,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005640184888851198,
      "loss": 1.4721,
      "step": 26696
    },
    {
      "epoch": 2.2969112965671514,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005639916706176049,
      "loss": 1.4835,
      "step": 26697
    },
    {
      "epoch": 2.296997332874473,
      "grad_norm": 1.375,
      "learning_rate": 0.0005639648521629267,
      "loss": 1.8361,
      "step": 26698
    },
    {
      "epoch": 2.2970833691817947,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005639380335211639,
      "loss": 1.5332,
      "step": 26699
    },
    {
      "epoch": 2.2971694054891163,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005639112146923946,
      "loss": 1.0961,
      "step": 26700
    },
    {
      "epoch": 2.297255441796438,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005638843956766976,
      "loss": 1.0364,
      "step": 26701
    },
    {
      "epoch": 2.2973414781037595,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000563857576474151,
      "loss": 0.7719,
      "step": 26702
    },
    {
      "epoch": 2.2974275144110816,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005638307570848335,
      "loss": 0.8254,
      "step": 26703
    },
    {
      "epoch": 2.2975135507184032,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005638039375088234,
      "loss": 1.6656,
      "step": 26704
    },
    {
      "epoch": 2.297599587025725,
      "grad_norm": 1.0,
      "learning_rate": 0.0005637771177461993,
      "loss": 0.6332,
      "step": 26705
    },
    {
      "epoch": 2.2976856233330465,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005637502977970395,
      "loss": 1.2118,
      "step": 26706
    },
    {
      "epoch": 2.297771659640368,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005637234776614224,
      "loss": 1.3729,
      "step": 26707
    },
    {
      "epoch": 2.2978576959476897,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005636966573394266,
      "loss": 1.7588,
      "step": 26708
    },
    {
      "epoch": 2.297943732255012,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005636698368311305,
      "loss": 1.3936,
      "step": 26709
    },
    {
      "epoch": 2.2980297685623334,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005636430161366125,
      "loss": 0.7836,
      "step": 26710
    },
    {
      "epoch": 2.298115804869655,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005636161952559509,
      "loss": 1.1805,
      "step": 26711
    },
    {
      "epoch": 2.2982018411769767,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005635893741892244,
      "loss": 1.03,
      "step": 26712
    },
    {
      "epoch": 2.2982878774842983,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005635625529365114,
      "loss": 2.1,
      "step": 26713
    },
    {
      "epoch": 2.29837391379162,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005635357314978901,
      "loss": 0.7788,
      "step": 26714
    },
    {
      "epoch": 2.298459950098942,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005635089098734394,
      "loss": 1.2495,
      "step": 26715
    },
    {
      "epoch": 2.2985459864062636,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005634820880632373,
      "loss": 1.7868,
      "step": 26716
    },
    {
      "epoch": 2.2986320227135852,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005634552660673626,
      "loss": 0.7201,
      "step": 26717
    },
    {
      "epoch": 2.298718059020907,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005634284438858936,
      "loss": 0.7432,
      "step": 26718
    },
    {
      "epoch": 2.2988040953282285,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005634016215189085,
      "loss": 0.9879,
      "step": 26719
    },
    {
      "epoch": 2.29889013163555,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005633747989664861,
      "loss": 1.654,
      "step": 26720
    },
    {
      "epoch": 2.2989761679428717,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005633479762287049,
      "loss": 0.8886,
      "step": 26721
    },
    {
      "epoch": 2.299062204250194,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000563321153305643,
      "loss": 0.6887,
      "step": 26722
    },
    {
      "epoch": 2.2991482405575154,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0005632943301973792,
      "loss": 1.3967,
      "step": 26723
    },
    {
      "epoch": 2.299234276864837,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005632675069039917,
      "loss": 1.3456,
      "step": 26724
    },
    {
      "epoch": 2.2993203131721587,
      "grad_norm": 1.25,
      "learning_rate": 0.000563240683425559,
      "loss": 1.2825,
      "step": 26725
    },
    {
      "epoch": 2.2994063494794803,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005632138597621597,
      "loss": 1.2771,
      "step": 26726
    },
    {
      "epoch": 2.299492385786802,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000563187035913872,
      "loss": 1.3511,
      "step": 26727
    },
    {
      "epoch": 2.2995784220941236,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005631602118807747,
      "loss": 0.7787,
      "step": 26728
    },
    {
      "epoch": 2.2996644584014456,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005631333876629459,
      "loss": 1.4199,
      "step": 26729
    },
    {
      "epoch": 2.2997504947087672,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005631065632604644,
      "loss": 1.2658,
      "step": 26730
    },
    {
      "epoch": 2.299836531016089,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005630797386734083,
      "loss": 1.5928,
      "step": 26731
    },
    {
      "epoch": 2.2999225673234105,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005630529139018563,
      "loss": 1.0959,
      "step": 26732
    },
    {
      "epoch": 2.300008603630732,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005630260889458868,
      "loss": 1.6225,
      "step": 26733
    },
    {
      "epoch": 2.3000946399380537,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005629992638055782,
      "loss": 1.3194,
      "step": 26734
    },
    {
      "epoch": 2.3001806762453754,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005629724384810091,
      "loss": 1.2105,
      "step": 26735
    },
    {
      "epoch": 2.3002667125526974,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005629456129722577,
      "loss": 1.2019,
      "step": 26736
    },
    {
      "epoch": 2.300352748860019,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005629187872794027,
      "loss": 0.7437,
      "step": 26737
    },
    {
      "epoch": 2.3004387851673407,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005628919614025226,
      "loss": 0.8029,
      "step": 26738
    },
    {
      "epoch": 2.3005248214746623,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005628651353416957,
      "loss": 0.7825,
      "step": 26739
    },
    {
      "epoch": 2.300610857781984,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005628383090970006,
      "loss": 0.9284,
      "step": 26740
    },
    {
      "epoch": 2.3006968940893056,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005628114826685154,
      "loss": 0.696,
      "step": 26741
    },
    {
      "epoch": 2.300782930396627,
      "grad_norm": 1.0,
      "learning_rate": 0.0005627846560563191,
      "loss": 1.0898,
      "step": 26742
    },
    {
      "epoch": 2.3008689667039492,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005627578292604897,
      "loss": 0.9484,
      "step": 26743
    },
    {
      "epoch": 2.300955003011271,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005627310022811059,
      "loss": 1.8382,
      "step": 26744
    },
    {
      "epoch": 2.3010410393185925,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005627041751182461,
      "loss": 1.7448,
      "step": 26745
    },
    {
      "epoch": 2.301127075625914,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005626773477719889,
      "loss": 1.412,
      "step": 26746
    },
    {
      "epoch": 2.3012131119332357,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005626505202424126,
      "loss": 1.4087,
      "step": 26747
    },
    {
      "epoch": 2.3012991482405574,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005626236925295956,
      "loss": 1.3419,
      "step": 26748
    },
    {
      "epoch": 2.301385184547879,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005625968646336166,
      "loss": 0.7949,
      "step": 26749
    },
    {
      "epoch": 2.301471220855201,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005625700365545538,
      "loss": 1.312,
      "step": 26750
    },
    {
      "epoch": 2.3015572571625227,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005625432082924861,
      "loss": 1.0412,
      "step": 26751
    },
    {
      "epoch": 2.3016432934698443,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005625163798474912,
      "loss": 2.1555,
      "step": 26752
    },
    {
      "epoch": 2.301729329777166,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005624895512196484,
      "loss": 1.6116,
      "step": 26753
    },
    {
      "epoch": 2.3018153660844876,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005624627224090358,
      "loss": 1.3761,
      "step": 26754
    },
    {
      "epoch": 2.301901402391809,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005624358934157318,
      "loss": 0.8483,
      "step": 26755
    },
    {
      "epoch": 2.301987438699131,
      "grad_norm": 1.03125,
      "learning_rate": 0.000562409064239815,
      "loss": 0.836,
      "step": 26756
    },
    {
      "epoch": 2.302073475006453,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005623822348813639,
      "loss": 1.0799,
      "step": 26757
    },
    {
      "epoch": 2.3021595113137745,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005623554053404568,
      "loss": 1.9507,
      "step": 26758
    },
    {
      "epoch": 2.302245547621096,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005623285756171722,
      "loss": 1.7146,
      "step": 26759
    },
    {
      "epoch": 2.3023315839284177,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005623017457115888,
      "loss": 1.5923,
      "step": 26760
    },
    {
      "epoch": 2.3024176202357394,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005622749156237849,
      "loss": 1.7769,
      "step": 26761
    },
    {
      "epoch": 2.302503656543061,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005622480853538389,
      "loss": 0.6047,
      "step": 26762
    },
    {
      "epoch": 2.3025896928503826,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005622212549018293,
      "loss": 1.0107,
      "step": 26763
    },
    {
      "epoch": 2.3026757291577047,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005621944242678347,
      "loss": 0.9403,
      "step": 26764
    },
    {
      "epoch": 2.3027617654650263,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005621675934519336,
      "loss": 1.4427,
      "step": 26765
    },
    {
      "epoch": 2.302847801772348,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005621407624542043,
      "loss": 1.8625,
      "step": 26766
    },
    {
      "epoch": 2.3029338380796696,
      "grad_norm": 1.0,
      "learning_rate": 0.0005621139312747253,
      "loss": 0.8451,
      "step": 26767
    },
    {
      "epoch": 2.303019874386991,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005620870999135752,
      "loss": 1.2821,
      "step": 26768
    },
    {
      "epoch": 2.303105910694313,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005620602683708325,
      "loss": 1.8223,
      "step": 26769
    },
    {
      "epoch": 2.303191947001635,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0005620334366465753,
      "loss": 0.5876,
      "step": 26770
    },
    {
      "epoch": 2.3032779833089565,
      "grad_norm": 1.0,
      "learning_rate": 0.0005620066047408825,
      "loss": 1.9878,
      "step": 26771
    },
    {
      "epoch": 2.303364019616278,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005619797726538325,
      "loss": 1.96,
      "step": 26772
    },
    {
      "epoch": 2.3034500559235997,
      "grad_norm": 1.0,
      "learning_rate": 0.0005619529403855037,
      "loss": 1.4958,
      "step": 26773
    },
    {
      "epoch": 2.3035360922309214,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005619261079359746,
      "loss": 1.6874,
      "step": 26774
    },
    {
      "epoch": 2.303622128538243,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005618992753053236,
      "loss": 1.792,
      "step": 26775
    },
    {
      "epoch": 2.303708164845565,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005618724424936295,
      "loss": 1.9321,
      "step": 26776
    },
    {
      "epoch": 2.3037942011528867,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005618456095009703,
      "loss": 1.145,
      "step": 26777
    },
    {
      "epoch": 2.3038802374602083,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005618187763274247,
      "loss": 1.6555,
      "step": 26778
    },
    {
      "epoch": 2.30396627376753,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005617919429730713,
      "loss": 1.3696,
      "step": 26779
    },
    {
      "epoch": 2.3040523100748516,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005617651094379885,
      "loss": 0.7909,
      "step": 26780
    },
    {
      "epoch": 2.304138346382173,
      "grad_norm": 1.0,
      "learning_rate": 0.0005617382757222549,
      "loss": 0.9422,
      "step": 26781
    },
    {
      "epoch": 2.304224382689495,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005617114418259486,
      "loss": 0.7349,
      "step": 26782
    },
    {
      "epoch": 2.304310418996817,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005616846077491484,
      "loss": 1.2843,
      "step": 26783
    },
    {
      "epoch": 2.3043964553041385,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005616577734919329,
      "loss": 0.7352,
      "step": 26784
    },
    {
      "epoch": 2.30448249161146,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005616309390543801,
      "loss": 0.9945,
      "step": 26785
    },
    {
      "epoch": 2.3045685279187818,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005616041044365689,
      "loss": 2.0186,
      "step": 26786
    },
    {
      "epoch": 2.3046545642261034,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005615772696385778,
      "loss": 1.1401,
      "step": 26787
    },
    {
      "epoch": 2.304740600533425,
      "grad_norm": 1.0,
      "learning_rate": 0.0005615504346604852,
      "loss": 0.9581,
      "step": 26788
    },
    {
      "epoch": 2.3048266368407466,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005615235995023693,
      "loss": 0.65,
      "step": 26789
    },
    {
      "epoch": 2.3049126731480687,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000561496764164309,
      "loss": 1.4589,
      "step": 26790
    },
    {
      "epoch": 2.3049987094553903,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005614699286463824,
      "loss": 1.4551,
      "step": 26791
    },
    {
      "epoch": 2.305084745762712,
      "grad_norm": 4.5625,
      "learning_rate": 0.0005614430929486686,
      "loss": 1.285,
      "step": 26792
    },
    {
      "epoch": 2.3051707820700336,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005614162570712454,
      "loss": 0.8485,
      "step": 26793
    },
    {
      "epoch": 2.305256818377355,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0005613894210141917,
      "loss": 1.6613,
      "step": 26794
    },
    {
      "epoch": 2.305342854684677,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000561362584777586,
      "loss": 1.9315,
      "step": 26795
    },
    {
      "epoch": 2.3054288909919984,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005613357483615066,
      "loss": 0.5669,
      "step": 26796
    },
    {
      "epoch": 2.3055149272993205,
      "grad_norm": 0.953125,
      "learning_rate": 0.000561308911766032,
      "loss": 1.0025,
      "step": 26797
    },
    {
      "epoch": 2.305600963606642,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005612820749912407,
      "loss": 2.066,
      "step": 26798
    },
    {
      "epoch": 2.3056869999139638,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005612552380372113,
      "loss": 0.9242,
      "step": 26799
    },
    {
      "epoch": 2.3057730362212854,
      "grad_norm": 1.125,
      "learning_rate": 0.0005612284009040223,
      "loss": 1.1294,
      "step": 26800
    },
    {
      "epoch": 2.305859072528607,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0005612015635917521,
      "loss": 0.9148,
      "step": 26801
    },
    {
      "epoch": 2.3059451088359286,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005611747261004792,
      "loss": 1.3692,
      "step": 26802
    },
    {
      "epoch": 2.3060311451432502,
      "grad_norm": 0.75,
      "learning_rate": 0.0005611478884302823,
      "loss": 0.9592,
      "step": 26803
    },
    {
      "epoch": 2.3061171814505723,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005611210505812395,
      "loss": 1.5235,
      "step": 26804
    },
    {
      "epoch": 2.306203217757894,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005610942125534297,
      "loss": 0.9333,
      "step": 26805
    },
    {
      "epoch": 2.3062892540652156,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005610673743469311,
      "loss": 1.4487,
      "step": 26806
    },
    {
      "epoch": 2.306375290372537,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005610405359618223,
      "loss": 0.7974,
      "step": 26807
    },
    {
      "epoch": 2.306461326679859,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005610136973981816,
      "loss": 0.6276,
      "step": 26808
    },
    {
      "epoch": 2.3065473629871804,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000560986858656088,
      "loss": 0.9095,
      "step": 26809
    },
    {
      "epoch": 2.306633399294502,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005609600197356197,
      "loss": 0.7583,
      "step": 26810
    },
    {
      "epoch": 2.306719435601824,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005609331806368552,
      "loss": 1.1126,
      "step": 26811
    },
    {
      "epoch": 2.3068054719091458,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005609063413598728,
      "loss": 0.9285,
      "step": 26812
    },
    {
      "epoch": 2.3068915082164674,
      "grad_norm": 2.40625,
      "learning_rate": 0.0005608795019047514,
      "loss": 1.8101,
      "step": 26813
    },
    {
      "epoch": 2.306977544523789,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005608526622715692,
      "loss": 1.5644,
      "step": 26814
    },
    {
      "epoch": 2.3070635808311106,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005608258224604047,
      "loss": 1.839,
      "step": 26815
    },
    {
      "epoch": 2.3071496171384323,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005607989824713366,
      "loss": 0.9933,
      "step": 26816
    },
    {
      "epoch": 2.307235653445754,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005607721423044434,
      "loss": 0.7371,
      "step": 26817
    },
    {
      "epoch": 2.307321689753076,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005607453019598035,
      "loss": 0.4461,
      "step": 26818
    },
    {
      "epoch": 2.3074077260603976,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005607184614374952,
      "loss": 0.8496,
      "step": 26819
    },
    {
      "epoch": 2.307493762367719,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005606916207375974,
      "loss": 1.1449,
      "step": 26820
    },
    {
      "epoch": 2.307579798675041,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005606647798601883,
      "loss": 1.1071,
      "step": 26821
    },
    {
      "epoch": 2.3076658349823624,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005606379388053466,
      "loss": 0.9702,
      "step": 26822
    },
    {
      "epoch": 2.307751871289684,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005606110975731507,
      "loss": 0.6943,
      "step": 26823
    },
    {
      "epoch": 2.3078379075970057,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005605842561636791,
      "loss": 1.1126,
      "step": 26824
    },
    {
      "epoch": 2.3079239439043278,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005605574145770104,
      "loss": 1.2903,
      "step": 26825
    },
    {
      "epoch": 2.3080099802116494,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005605305728132228,
      "loss": 1.6618,
      "step": 26826
    },
    {
      "epoch": 2.308096016518971,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005605037308723951,
      "loss": 0.5711,
      "step": 26827
    },
    {
      "epoch": 2.3081820528262926,
      "grad_norm": 0.859375,
      "learning_rate": 0.000560476888754606,
      "loss": 0.936,
      "step": 26828
    },
    {
      "epoch": 2.3082680891336143,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005604500464599337,
      "loss": 0.7006,
      "step": 26829
    },
    {
      "epoch": 2.3083541254409363,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005604232039884567,
      "loss": 1.2345,
      "step": 26830
    },
    {
      "epoch": 2.308440161748258,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005603963613402535,
      "loss": 1.7846,
      "step": 26831
    },
    {
      "epoch": 2.3085261980555796,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005603695185154028,
      "loss": 1.2174,
      "step": 26832
    },
    {
      "epoch": 2.308612234362901,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005603426755139831,
      "loss": 0.8563,
      "step": 26833
    },
    {
      "epoch": 2.308698270670223,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005603158323360726,
      "loss": 1.4293,
      "step": 26834
    },
    {
      "epoch": 2.3087843069775444,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005602889889817502,
      "loss": 0.6046,
      "step": 26835
    },
    {
      "epoch": 2.308870343284866,
      "grad_norm": 1.125,
      "learning_rate": 0.0005602621454510942,
      "loss": 0.9659,
      "step": 26836
    },
    {
      "epoch": 2.308956379592188,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005602353017441831,
      "loss": 1.8429,
      "step": 26837
    },
    {
      "epoch": 2.3090424158995098,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005602084578610955,
      "loss": 0.9772,
      "step": 26838
    },
    {
      "epoch": 2.3091284522068314,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005601816138019097,
      "loss": 0.7568,
      "step": 26839
    },
    {
      "epoch": 2.309214488514153,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005601547695667045,
      "loss": 1.7517,
      "step": 26840
    },
    {
      "epoch": 2.3093005248214746,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005601279251555583,
      "loss": 1.648,
      "step": 26841
    },
    {
      "epoch": 2.3093865611287963,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005601010805685497,
      "loss": 1.1354,
      "step": 26842
    },
    {
      "epoch": 2.309472597436118,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000560074235805757,
      "loss": 1.391,
      "step": 26843
    },
    {
      "epoch": 2.30955863374344,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005600473908672588,
      "loss": 1.3319,
      "step": 26844
    },
    {
      "epoch": 2.3096446700507616,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005600205457531339,
      "loss": 0.8705,
      "step": 26845
    },
    {
      "epoch": 2.309730706358083,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005599937004634603,
      "loss": 1.5442,
      "step": 26846
    },
    {
      "epoch": 2.309816742665405,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000559966854998317,
      "loss": 1.4749,
      "step": 26847
    },
    {
      "epoch": 2.3099027789727264,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005599400093577822,
      "loss": 1.4633,
      "step": 26848
    },
    {
      "epoch": 2.309988815280048,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005599131635419345,
      "loss": 0.4587,
      "step": 26849
    },
    {
      "epoch": 2.3100748515873697,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005598863175508526,
      "loss": 1.1231,
      "step": 26850
    },
    {
      "epoch": 2.3101608878946918,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005598594713846148,
      "loss": 0.9724,
      "step": 26851
    },
    {
      "epoch": 2.3102469242020134,
      "grad_norm": 1.0,
      "learning_rate": 0.0005598326250432997,
      "loss": 1.3411,
      "step": 26852
    },
    {
      "epoch": 2.310332960509335,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005598057785269856,
      "loss": 1.3803,
      "step": 26853
    },
    {
      "epoch": 2.3104189968166566,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005597789318357516,
      "loss": 1.4217,
      "step": 26854
    },
    {
      "epoch": 2.3105050331239783,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005597520849696757,
      "loss": 1.1113,
      "step": 26855
    },
    {
      "epoch": 2.3105910694313,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005597252379288364,
      "loss": 0.837,
      "step": 26856
    },
    {
      "epoch": 2.3106771057386215,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005596983907133125,
      "loss": 1.4051,
      "step": 26857
    },
    {
      "epoch": 2.3107631420459436,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005596715433231825,
      "loss": 1.2202,
      "step": 26858
    },
    {
      "epoch": 2.310849178353265,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005596446957585249,
      "loss": 1.1786,
      "step": 26859
    },
    {
      "epoch": 2.310935214660587,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005596178480194179,
      "loss": 0.7822,
      "step": 26860
    },
    {
      "epoch": 2.3110212509679084,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005595910001059406,
      "loss": 0.7041,
      "step": 26861
    },
    {
      "epoch": 2.31110728727523,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000559564152018171,
      "loss": 1.0824,
      "step": 26862
    },
    {
      "epoch": 2.3111933235825517,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005595373037561879,
      "loss": 0.9092,
      "step": 26863
    },
    {
      "epoch": 2.3112793598898733,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005595104553200696,
      "loss": 0.9922,
      "step": 26864
    },
    {
      "epoch": 2.3113653961971954,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005594836067098949,
      "loss": 0.8216,
      "step": 26865
    },
    {
      "epoch": 2.311451432504517,
      "grad_norm": 1.25,
      "learning_rate": 0.0005594567579257423,
      "loss": 1.2996,
      "step": 26866
    },
    {
      "epoch": 2.3115374688118386,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005594299089676905,
      "loss": 0.9614,
      "step": 26867
    },
    {
      "epoch": 2.3116235051191603,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005594030598358173,
      "loss": 0.8266,
      "step": 26868
    },
    {
      "epoch": 2.311709541426482,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000559376210530202,
      "loss": 1.2666,
      "step": 26869
    },
    {
      "epoch": 2.3117955777338035,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005593493610509227,
      "loss": 1.9609,
      "step": 26870
    },
    {
      "epoch": 2.311881614041125,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000559322511398058,
      "loss": 0.7942,
      "step": 26871
    },
    {
      "epoch": 2.311967650348447,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005592956615716867,
      "loss": 1.7868,
      "step": 26872
    },
    {
      "epoch": 2.312053686655769,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005592688115718869,
      "loss": 1.1778,
      "step": 26873
    },
    {
      "epoch": 2.3121397229630904,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005592419613987375,
      "loss": 1.7917,
      "step": 26874
    },
    {
      "epoch": 2.312225759270412,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005592151110523168,
      "loss": 0.7032,
      "step": 26875
    },
    {
      "epoch": 2.3123117955777337,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005591882605327036,
      "loss": 1.8836,
      "step": 26876
    },
    {
      "epoch": 2.3123978318850553,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000559161409839976,
      "loss": 1.5863,
      "step": 26877
    },
    {
      "epoch": 2.312483868192377,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005591345589742129,
      "loss": 0.6924,
      "step": 26878
    },
    {
      "epoch": 2.312569904499699,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005591077079354927,
      "loss": 1.1424,
      "step": 26879
    },
    {
      "epoch": 2.3126559408070206,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000559080856723894,
      "loss": 0.519,
      "step": 26880
    },
    {
      "epoch": 2.3127419771143423,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005590540053394952,
      "loss": 1.3071,
      "step": 26881
    },
    {
      "epoch": 2.312828013421664,
      "grad_norm": 0.765625,
      "learning_rate": 0.000559027153782375,
      "loss": 0.7486,
      "step": 26882
    },
    {
      "epoch": 2.3129140497289855,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005590003020526118,
      "loss": 2.051,
      "step": 26883
    },
    {
      "epoch": 2.313000086036307,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005589734501502841,
      "loss": 2.0141,
      "step": 26884
    },
    {
      "epoch": 2.313086122343629,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005589465980754708,
      "loss": 2.028,
      "step": 26885
    },
    {
      "epoch": 2.313172158650951,
      "grad_norm": 1.21875,
      "learning_rate": 0.00055891974582825,
      "loss": 1.7292,
      "step": 26886
    },
    {
      "epoch": 2.3132581949582725,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005588928934087003,
      "loss": 1.0621,
      "step": 26887
    },
    {
      "epoch": 2.313344231265594,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005588660408169005,
      "loss": 1.0894,
      "step": 26888
    },
    {
      "epoch": 2.3134302675729157,
      "grad_norm": 1.046875,
      "learning_rate": 0.000558839188052929,
      "loss": 0.9581,
      "step": 26889
    },
    {
      "epoch": 2.3135163038802373,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005588123351168642,
      "loss": 1.8235,
      "step": 26890
    },
    {
      "epoch": 2.3136023401875594,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005587854820087848,
      "loss": 0.874,
      "step": 26891
    },
    {
      "epoch": 2.313688376494881,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005587586287287693,
      "loss": 1.3673,
      "step": 26892
    },
    {
      "epoch": 2.3137744128022026,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005587317752768962,
      "loss": 1.4018,
      "step": 26893
    },
    {
      "epoch": 2.3138604491095243,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005587049216532442,
      "loss": 1.2306,
      "step": 26894
    },
    {
      "epoch": 2.313946485416846,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005586780678578916,
      "loss": 0.6147,
      "step": 26895
    },
    {
      "epoch": 2.3140325217241675,
      "grad_norm": 1.890625,
      "learning_rate": 0.0005586512138909171,
      "loss": 1.2946,
      "step": 26896
    },
    {
      "epoch": 2.314118558031489,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005586243597523992,
      "loss": 1.615,
      "step": 26897
    },
    {
      "epoch": 2.314204594338811,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005585975054424165,
      "loss": 1.5811,
      "step": 26898
    },
    {
      "epoch": 2.314290630646133,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005585706509610474,
      "loss": 1.5242,
      "step": 26899
    },
    {
      "epoch": 2.3143766669534545,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005585437963083706,
      "loss": 1.3285,
      "step": 26900
    },
    {
      "epoch": 2.314462703260776,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005585169414844646,
      "loss": 1.4614,
      "step": 26901
    },
    {
      "epoch": 2.3145487395680977,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005584900864894078,
      "loss": 1.0352,
      "step": 26902
    },
    {
      "epoch": 2.3146347758754193,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000558463231323279,
      "loss": 1.1612,
      "step": 26903
    },
    {
      "epoch": 2.314720812182741,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005584363759861566,
      "loss": 1.5942,
      "step": 26904
    },
    {
      "epoch": 2.314806848490063,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005584095204781191,
      "loss": 0.5549,
      "step": 26905
    },
    {
      "epoch": 2.3148928847973846,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005583826647992452,
      "loss": 1.0115,
      "step": 26906
    },
    {
      "epoch": 2.3149789211047063,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005583558089496133,
      "loss": 1.8258,
      "step": 26907
    },
    {
      "epoch": 2.315064957412028,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005583289529293022,
      "loss": 1.121,
      "step": 26908
    },
    {
      "epoch": 2.3151509937193495,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005583020967383901,
      "loss": 1.9074,
      "step": 26909
    },
    {
      "epoch": 2.315237030026671,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005582752403769556,
      "loss": 1.1247,
      "step": 26910
    },
    {
      "epoch": 2.3153230663339928,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005582483838450775,
      "loss": 1.1722,
      "step": 26911
    },
    {
      "epoch": 2.315409102641315,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005582215271428342,
      "loss": 0.9387,
      "step": 26912
    },
    {
      "epoch": 2.3154951389486365,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005581946702703041,
      "loss": 1.4761,
      "step": 26913
    },
    {
      "epoch": 2.315581175255958,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005581678132275662,
      "loss": 0.8625,
      "step": 26914
    },
    {
      "epoch": 2.3156672115632797,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005581409560146987,
      "loss": 0.9444,
      "step": 26915
    },
    {
      "epoch": 2.3157532478706013,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005581140986317799,
      "loss": 0.6737,
      "step": 26916
    },
    {
      "epoch": 2.315839284177923,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005580872410788888,
      "loss": 1.2102,
      "step": 26917
    },
    {
      "epoch": 2.3159253204852446,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005580603833561039,
      "loss": 1.2698,
      "step": 26918
    },
    {
      "epoch": 2.3160113567925666,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005580335254635038,
      "loss": 0.9288,
      "step": 26919
    },
    {
      "epoch": 2.3160973930998883,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0005580066674011666,
      "loss": 1.0256,
      "step": 26920
    },
    {
      "epoch": 2.31618342940721,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005579798091691713,
      "loss": 1.0326,
      "step": 26921
    },
    {
      "epoch": 2.3162694657145315,
      "grad_norm": 2.265625,
      "learning_rate": 0.0005579529507675964,
      "loss": 1.3747,
      "step": 26922
    },
    {
      "epoch": 2.316355502021853,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005579260921965204,
      "loss": 0.817,
      "step": 26923
    },
    {
      "epoch": 2.3164415383291748,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005578992334560219,
      "loss": 1.0466,
      "step": 26924
    },
    {
      "epoch": 2.3165275746364964,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005578723745461794,
      "loss": 1.5372,
      "step": 26925
    },
    {
      "epoch": 2.3166136109438185,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005578455154670713,
      "loss": 1.3892,
      "step": 26926
    },
    {
      "epoch": 2.31669964725114,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005578186562187764,
      "loss": 0.6488,
      "step": 26927
    },
    {
      "epoch": 2.3167856835584617,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005577917968013731,
      "loss": 1.2915,
      "step": 26928
    },
    {
      "epoch": 2.3168717198657833,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005577649372149403,
      "loss": 1.3458,
      "step": 26929
    },
    {
      "epoch": 2.316957756173105,
      "grad_norm": 1.15625,
      "learning_rate": 0.000557738077459556,
      "loss": 2.0697,
      "step": 26930
    },
    {
      "epoch": 2.3170437924804266,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005577112175352991,
      "loss": 0.8234,
      "step": 26931
    },
    {
      "epoch": 2.317129828787748,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005576843574422483,
      "loss": 1.4016,
      "step": 26932
    },
    {
      "epoch": 2.3172158650950703,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005576574971804818,
      "loss": 0.9369,
      "step": 26933
    },
    {
      "epoch": 2.317301901402392,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005576306367500785,
      "loss": 1.1018,
      "step": 26934
    },
    {
      "epoch": 2.3173879377097135,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005576037761511166,
      "loss": 1.301,
      "step": 26935
    },
    {
      "epoch": 2.317473974017035,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005575769153836748,
      "loss": 1.5556,
      "step": 26936
    },
    {
      "epoch": 2.3175600103243568,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005575500544478321,
      "loss": 1.2093,
      "step": 26937
    },
    {
      "epoch": 2.3176460466316784,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005575231933436663,
      "loss": 0.8814,
      "step": 26938
    },
    {
      "epoch": 2.317732082939,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005574963320712564,
      "loss": 0.8772,
      "step": 26939
    },
    {
      "epoch": 2.317818119246322,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005574694706306811,
      "loss": 0.6112,
      "step": 26940
    },
    {
      "epoch": 2.3179041555536437,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005574426090220187,
      "loss": 1.242,
      "step": 26941
    },
    {
      "epoch": 2.3179901918609653,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005574157472453479,
      "loss": 1.0602,
      "step": 26942
    },
    {
      "epoch": 2.318076228168287,
      "grad_norm": 1.234375,
      "learning_rate": 0.000557388885300747,
      "loss": 1.1911,
      "step": 26943
    },
    {
      "epoch": 2.3181622644756086,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000557362023188295,
      "loss": 1.0659,
      "step": 26944
    },
    {
      "epoch": 2.31824830078293,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005573351609080701,
      "loss": 2.1157,
      "step": 26945
    },
    {
      "epoch": 2.3183343370902523,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000557308298460151,
      "loss": 1.1636,
      "step": 26946
    },
    {
      "epoch": 2.318420373397574,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005572814358446163,
      "loss": 0.7341,
      "step": 26947
    },
    {
      "epoch": 2.3185064097048955,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005572545730615446,
      "loss": 1.567,
      "step": 26948
    },
    {
      "epoch": 2.318592446012217,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005572277101110144,
      "loss": 1.8109,
      "step": 26949
    },
    {
      "epoch": 2.3186784823195388,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005572008469931041,
      "loss": 1.0631,
      "step": 26950
    },
    {
      "epoch": 2.3187645186268604,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005571739837078926,
      "loss": 1.5218,
      "step": 26951
    },
    {
      "epoch": 2.3188505549341825,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005571471202554584,
      "loss": 1.301,
      "step": 26952
    },
    {
      "epoch": 2.318936591241504,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0005571202566358798,
      "loss": 0.4238,
      "step": 26953
    },
    {
      "epoch": 2.3190226275488257,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005570933928492358,
      "loss": 1.4395,
      "step": 26954
    },
    {
      "epoch": 2.3191086638561473,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005570665288956043,
      "loss": 0.9393,
      "step": 26955
    },
    {
      "epoch": 2.319194700163469,
      "grad_norm": 2.0625,
      "learning_rate": 0.0005570396647750647,
      "loss": 1.3629,
      "step": 26956
    },
    {
      "epoch": 2.3192807364707906,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0005570128004876951,
      "loss": 2.116,
      "step": 26957
    },
    {
      "epoch": 2.319366772778112,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005569859360335741,
      "loss": 1.2143,
      "step": 26958
    },
    {
      "epoch": 2.3194528090854343,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005569590714127803,
      "loss": 1.5264,
      "step": 26959
    },
    {
      "epoch": 2.319538845392756,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005569322066253924,
      "loss": 1.4829,
      "step": 26960
    },
    {
      "epoch": 2.3196248817000775,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005569053416714886,
      "loss": 1.1235,
      "step": 26961
    },
    {
      "epoch": 2.319710918007399,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005568784765511481,
      "loss": 1.4809,
      "step": 26962
    },
    {
      "epoch": 2.3197969543147208,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005568516112644489,
      "loss": 1.9329,
      "step": 26963
    },
    {
      "epoch": 2.3198829906220424,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005568247458114699,
      "loss": 0.736,
      "step": 26964
    },
    {
      "epoch": 2.319969026929364,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005567978801922895,
      "loss": 1.7379,
      "step": 26965
    },
    {
      "epoch": 2.320055063236686,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005567710144069864,
      "loss": 1.5774,
      "step": 26966
    },
    {
      "epoch": 2.3201410995440077,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000556744148455639,
      "loss": 1.2442,
      "step": 26967
    },
    {
      "epoch": 2.3202271358513293,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005567172823383261,
      "loss": 1.694,
      "step": 26968
    },
    {
      "epoch": 2.320313172158651,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005566904160551262,
      "loss": 1.668,
      "step": 26969
    },
    {
      "epoch": 2.3203992084659726,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005566635496061179,
      "loss": 1.9395,
      "step": 26970
    },
    {
      "epoch": 2.320485244773294,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005566366829913796,
      "loss": 0.8729,
      "step": 26971
    },
    {
      "epoch": 2.320571281080616,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0005566098162109901,
      "loss": 1.1813,
      "step": 26972
    },
    {
      "epoch": 2.320657317387938,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005565829492650277,
      "loss": 0.8411,
      "step": 26973
    },
    {
      "epoch": 2.3207433536952595,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005565560821535714,
      "loss": 0.6728,
      "step": 26974
    },
    {
      "epoch": 2.320829390002581,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005565292148766996,
      "loss": 0.6322,
      "step": 26975
    },
    {
      "epoch": 2.3209154263099028,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005565023474344906,
      "loss": 0.8282,
      "step": 26976
    },
    {
      "epoch": 2.3210014626172244,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005564754798270236,
      "loss": 2.3055,
      "step": 26977
    },
    {
      "epoch": 2.321087498924546,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005564486120543764,
      "loss": 1.2753,
      "step": 26978
    },
    {
      "epoch": 2.3211735352318676,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005564217441166283,
      "loss": 0.6692,
      "step": 26979
    },
    {
      "epoch": 2.3212595715391897,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005563948760138574,
      "loss": 0.7019,
      "step": 26980
    },
    {
      "epoch": 2.3213456078465113,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005563680077461425,
      "loss": 1.1131,
      "step": 26981
    },
    {
      "epoch": 2.321431644153833,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005563411393135622,
      "loss": 0.8629,
      "step": 26982
    },
    {
      "epoch": 2.3215176804611546,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000556314270716195,
      "loss": 1.7175,
      "step": 26983
    },
    {
      "epoch": 2.321603716768476,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005562874019541194,
      "loss": 1.1523,
      "step": 26984
    },
    {
      "epoch": 2.321689753075798,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005562605330274142,
      "loss": 0.8219,
      "step": 26985
    },
    {
      "epoch": 2.3217757893831195,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000556233663936158,
      "loss": 0.9446,
      "step": 26986
    },
    {
      "epoch": 2.3218618256904415,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005562067946804289,
      "loss": 1.535,
      "step": 26987
    },
    {
      "epoch": 2.321947861997763,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005561799252603062,
      "loss": 1.0573,
      "step": 26988
    },
    {
      "epoch": 2.3220338983050848,
      "grad_norm": 1.5625,
      "learning_rate": 0.0005561530556758681,
      "loss": 1.5611,
      "step": 26989
    },
    {
      "epoch": 2.3221199346124064,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005561261859271931,
      "loss": 0.8738,
      "step": 26990
    },
    {
      "epoch": 2.322205970919728,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00055609931601436,
      "loss": 1.9257,
      "step": 26991
    },
    {
      "epoch": 2.3222920072270496,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005560724459374471,
      "loss": 0.6401,
      "step": 26992
    },
    {
      "epoch": 2.3223780435343713,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005560455756965334,
      "loss": 1.5704,
      "step": 26993
    },
    {
      "epoch": 2.3224640798416933,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005560187052916974,
      "loss": 1.0363,
      "step": 26994
    },
    {
      "epoch": 2.322550116149015,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005559918347230174,
      "loss": 1.6127,
      "step": 26995
    },
    {
      "epoch": 2.3226361524563366,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005559649639905723,
      "loss": 0.8192,
      "step": 26996
    },
    {
      "epoch": 2.322722188763658,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005559380930944405,
      "loss": 1.0945,
      "step": 26997
    },
    {
      "epoch": 2.32280822507098,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005559112220347007,
      "loss": 1.4164,
      "step": 26998
    },
    {
      "epoch": 2.3228942613783015,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005558843508114313,
      "loss": 1.7998,
      "step": 26999
    },
    {
      "epoch": 2.322980297685623,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005558574794247112,
      "loss": 1.632,
      "step": 27000
    },
    {
      "epoch": 2.323066333992945,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005558306078746189,
      "loss": 1.3522,
      "step": 27001
    },
    {
      "epoch": 2.323152370300267,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005558037361612327,
      "loss": 1.1994,
      "step": 27002
    },
    {
      "epoch": 2.3232384066075884,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005557768642846316,
      "loss": 1.2049,
      "step": 27003
    },
    {
      "epoch": 2.32332444291491,
      "grad_norm": 1.15625,
      "learning_rate": 0.000555749992244894,
      "loss": 1.238,
      "step": 27004
    },
    {
      "epoch": 2.3234104792222317,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005557231200420985,
      "loss": 0.8716,
      "step": 27005
    },
    {
      "epoch": 2.3234965155295537,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005556962476763236,
      "loss": 1.205,
      "step": 27006
    },
    {
      "epoch": 2.3235825518368753,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005556693751476481,
      "loss": 1.1079,
      "step": 27007
    },
    {
      "epoch": 2.323668588144197,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005556425024561507,
      "loss": 0.7685,
      "step": 27008
    },
    {
      "epoch": 2.3237546244515186,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005556156296019094,
      "loss": 0.8951,
      "step": 27009
    },
    {
      "epoch": 2.32384066075884,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005555887565850034,
      "loss": 1.0329,
      "step": 27010
    },
    {
      "epoch": 2.323926697066162,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005555618834055112,
      "loss": 2.2001,
      "step": 27011
    },
    {
      "epoch": 2.3240127333734835,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005555350100635111,
      "loss": 0.8936,
      "step": 27012
    },
    {
      "epoch": 2.3240987696808055,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005555081365590822,
      "loss": 0.7286,
      "step": 27013
    },
    {
      "epoch": 2.324184805988127,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005554812628923024,
      "loss": 0.7533,
      "step": 27014
    },
    {
      "epoch": 2.324270842295449,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000555454389063251,
      "loss": 1.2208,
      "step": 27015
    },
    {
      "epoch": 2.3243568786027704,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005554275150720061,
      "loss": 0.6893,
      "step": 27016
    },
    {
      "epoch": 2.324442914910092,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005554006409186467,
      "loss": 0.7733,
      "step": 27017
    },
    {
      "epoch": 2.3245289512174137,
      "grad_norm": 0.890625,
      "learning_rate": 0.000555373766603251,
      "loss": 0.7689,
      "step": 27018
    },
    {
      "epoch": 2.3246149875247353,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000555346892125898,
      "loss": 1.0595,
      "step": 27019
    },
    {
      "epoch": 2.3247010238320573,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000555320017486666,
      "loss": 0.7597,
      "step": 27020
    },
    {
      "epoch": 2.324787060139379,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005552931426856336,
      "loss": 2.0938,
      "step": 27021
    },
    {
      "epoch": 2.3248730964467006,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005552662677228797,
      "loss": 1.169,
      "step": 27022
    },
    {
      "epoch": 2.324959132754022,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005552393925984827,
      "loss": 0.914,
      "step": 27023
    },
    {
      "epoch": 2.325045169061344,
      "grad_norm": 1.1875,
      "learning_rate": 0.000555212517312521,
      "loss": 0.8204,
      "step": 27024
    },
    {
      "epoch": 2.3251312053686655,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005551856418650736,
      "loss": 1.1578,
      "step": 27025
    },
    {
      "epoch": 2.325217241675987,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000555158766256219,
      "loss": 0.7931,
      "step": 27026
    },
    {
      "epoch": 2.325303277983309,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005551318904860357,
      "loss": 1.4015,
      "step": 27027
    },
    {
      "epoch": 2.325389314290631,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005551050145546022,
      "loss": 1.1237,
      "step": 27028
    },
    {
      "epoch": 2.3254753505979524,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005550781384619973,
      "loss": 1.1092,
      "step": 27029
    },
    {
      "epoch": 2.325561386905274,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005550512622082997,
      "loss": 1.1139,
      "step": 27030
    },
    {
      "epoch": 2.3256474232125957,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005550243857935877,
      "loss": 1.0356,
      "step": 27031
    },
    {
      "epoch": 2.3257334595199173,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005549975092179401,
      "loss": 1.1874,
      "step": 27032
    },
    {
      "epoch": 2.325819495827239,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005549706324814354,
      "loss": 1.3215,
      "step": 27033
    },
    {
      "epoch": 2.325905532134561,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005549437555841526,
      "loss": 1.2373,
      "step": 27034
    },
    {
      "epoch": 2.3259915684418826,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005549168785261698,
      "loss": 0.6733,
      "step": 27035
    },
    {
      "epoch": 2.326077604749204,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005548900013075658,
      "loss": 1.8306,
      "step": 27036
    },
    {
      "epoch": 2.326163641056526,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005548631239284192,
      "loss": 1.0706,
      "step": 27037
    },
    {
      "epoch": 2.3262496773638475,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005548362463888087,
      "loss": 1.1341,
      "step": 27038
    },
    {
      "epoch": 2.326335713671169,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005548093686888128,
      "loss": 1.2636,
      "step": 27039
    },
    {
      "epoch": 2.3264217499784907,
      "grad_norm": 1.125,
      "learning_rate": 0.0005547824908285102,
      "loss": 1.5105,
      "step": 27040
    },
    {
      "epoch": 2.326507786285813,
      "grad_norm": 4.28125,
      "learning_rate": 0.0005547556128079794,
      "loss": 1.4859,
      "step": 27041
    },
    {
      "epoch": 2.3265938225931344,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005547287346272992,
      "loss": 1.2355,
      "step": 27042
    },
    {
      "epoch": 2.326679858900456,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0005547018562865479,
      "loss": 1.1568,
      "step": 27043
    },
    {
      "epoch": 2.3267658952077777,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005546749777858045,
      "loss": 1.1429,
      "step": 27044
    },
    {
      "epoch": 2.3268519315150993,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005546480991251474,
      "loss": 0.8243,
      "step": 27045
    },
    {
      "epoch": 2.326937967822421,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005546212203046552,
      "loss": 0.8129,
      "step": 27046
    },
    {
      "epoch": 2.3270240041297425,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005545943413244063,
      "loss": 1.5257,
      "step": 27047
    },
    {
      "epoch": 2.3271100404370646,
      "grad_norm": 1.296875,
      "learning_rate": 0.00055456746218448,
      "loss": 1.2683,
      "step": 27048
    },
    {
      "epoch": 2.327196076744386,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005545405828849542,
      "loss": 0.8287,
      "step": 27049
    },
    {
      "epoch": 2.327282113051708,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000554513703425908,
      "loss": 1.3179,
      "step": 27050
    },
    {
      "epoch": 2.3273681493590295,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005544868238074196,
      "loss": 1.1331,
      "step": 27051
    },
    {
      "epoch": 2.327454185666351,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000554459944029568,
      "loss": 1.8768,
      "step": 27052
    },
    {
      "epoch": 2.3275402219736727,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005544330640924317,
      "loss": 1.3617,
      "step": 27053
    },
    {
      "epoch": 2.3276262582809943,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005544061839960892,
      "loss": 1.4465,
      "step": 27054
    },
    {
      "epoch": 2.3277122945883164,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005543793037406191,
      "loss": 1.1201,
      "step": 27055
    },
    {
      "epoch": 2.327798330895638,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005543524233261002,
      "loss": 1.1848,
      "step": 27056
    },
    {
      "epoch": 2.3278843672029597,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005543255427526111,
      "loss": 1.8508,
      "step": 27057
    },
    {
      "epoch": 2.3279704035102813,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005542986620202301,
      "loss": 1.2049,
      "step": 27058
    },
    {
      "epoch": 2.328056439817603,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005542717811290364,
      "loss": 1.6398,
      "step": 27059
    },
    {
      "epoch": 2.3281424761249245,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005542449000791082,
      "loss": 1.1862,
      "step": 27060
    },
    {
      "epoch": 2.3282285124322466,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005542180188705242,
      "loss": 1.1867,
      "step": 27061
    },
    {
      "epoch": 2.3283145487395682,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000554191137503363,
      "loss": 0.8256,
      "step": 27062
    },
    {
      "epoch": 2.32840058504689,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005541642559777032,
      "loss": 1.7938,
      "step": 27063
    },
    {
      "epoch": 2.3284866213542115,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005541373742936237,
      "loss": 1.8506,
      "step": 27064
    },
    {
      "epoch": 2.328572657661533,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0005541104924512026,
      "loss": 1.6661,
      "step": 27065
    },
    {
      "epoch": 2.3286586939688547,
      "grad_norm": 0.875,
      "learning_rate": 0.0005540836104505189,
      "loss": 1.3788,
      "step": 27066
    },
    {
      "epoch": 2.328744730276177,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005540567282916513,
      "loss": 0.9471,
      "step": 27067
    },
    {
      "epoch": 2.3288307665834984,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005540298459746783,
      "loss": 0.6313,
      "step": 27068
    },
    {
      "epoch": 2.32891680289082,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005540029634996785,
      "loss": 0.5468,
      "step": 27069
    },
    {
      "epoch": 2.3290028391981417,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005539760808667305,
      "loss": 1.4133,
      "step": 27070
    },
    {
      "epoch": 2.3290888755054633,
      "grad_norm": 1.125,
      "learning_rate": 0.000553949198075913,
      "loss": 1.1317,
      "step": 27071
    },
    {
      "epoch": 2.329174911812785,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005539223151273046,
      "loss": 1.1889,
      "step": 27072
    },
    {
      "epoch": 2.3292609481201065,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005538954320209837,
      "loss": 1.6703,
      "step": 27073
    },
    {
      "epoch": 2.3293469844274286,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005538685487570293,
      "loss": 1.0608,
      "step": 27074
    },
    {
      "epoch": 2.3294330207347502,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005538416653355199,
      "loss": 0.9339,
      "step": 27075
    },
    {
      "epoch": 2.329519057042072,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005538147817565341,
      "loss": 1.1286,
      "step": 27076
    },
    {
      "epoch": 2.3296050933493935,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005537878980201503,
      "loss": 1.3159,
      "step": 27077
    },
    {
      "epoch": 2.329691129656715,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005537610141264477,
      "loss": 1.0714,
      "step": 27078
    },
    {
      "epoch": 2.3297771659640367,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005537341300755045,
      "loss": 1.2401,
      "step": 27079
    },
    {
      "epoch": 2.3298632022713583,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005537072458673993,
      "loss": 0.8219,
      "step": 27080
    },
    {
      "epoch": 2.3299492385786804,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000553680361502211,
      "loss": 1.0254,
      "step": 27081
    },
    {
      "epoch": 2.330035274886002,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005536534769800179,
      "loss": 1.4558,
      "step": 27082
    },
    {
      "epoch": 2.3301213111933237,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000553626592300899,
      "loss": 1.6483,
      "step": 27083
    },
    {
      "epoch": 2.3302073475006453,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005535997074649326,
      "loss": 1.4491,
      "step": 27084
    },
    {
      "epoch": 2.330293383807967,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005535728224721974,
      "loss": 1.4735,
      "step": 27085
    },
    {
      "epoch": 2.3303794201152885,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005535459373227724,
      "loss": 1.6194,
      "step": 27086
    },
    {
      "epoch": 2.33046545642261,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005535190520167358,
      "loss": 1.077,
      "step": 27087
    },
    {
      "epoch": 2.3305514927299322,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005534921665541662,
      "loss": 0.977,
      "step": 27088
    },
    {
      "epoch": 2.330637529037254,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005534652809351427,
      "loss": 1.2156,
      "step": 27089
    },
    {
      "epoch": 2.3307235653445755,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005534383951597436,
      "loss": 1.1374,
      "step": 27090
    },
    {
      "epoch": 2.330809601651897,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005534115092280476,
      "loss": 0.7087,
      "step": 27091
    },
    {
      "epoch": 2.3308956379592187,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005533846231401331,
      "loss": 0.8608,
      "step": 27092
    },
    {
      "epoch": 2.3309816742665403,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005533577368960792,
      "loss": 1.3347,
      "step": 27093
    },
    {
      "epoch": 2.331067710573862,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005533308504959642,
      "loss": 0.8494,
      "step": 27094
    },
    {
      "epoch": 2.331153746881184,
      "grad_norm": 1.0,
      "learning_rate": 0.0005533039639398669,
      "loss": 0.9842,
      "step": 27095
    },
    {
      "epoch": 2.3312397831885057,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005532770772278658,
      "loss": 1.8088,
      "step": 27096
    },
    {
      "epoch": 2.3313258194958273,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005532501903600396,
      "loss": 1.4874,
      "step": 27097
    },
    {
      "epoch": 2.331411855803149,
      "grad_norm": 1.125,
      "learning_rate": 0.0005532233033364671,
      "loss": 0.919,
      "step": 27098
    },
    {
      "epoch": 2.3314978921104705,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005531964161572266,
      "loss": 1.2582,
      "step": 27099
    },
    {
      "epoch": 2.331583928417792,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005531695288223971,
      "loss": 1.4624,
      "step": 27100
    },
    {
      "epoch": 2.331669964725114,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000553142641332057,
      "loss": 1.9687,
      "step": 27101
    },
    {
      "epoch": 2.331756001032436,
      "grad_norm": 1.03125,
      "learning_rate": 0.000553115753686285,
      "loss": 1.4076,
      "step": 27102
    },
    {
      "epoch": 2.3318420373397575,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005530888658851597,
      "loss": 1.1065,
      "step": 27103
    },
    {
      "epoch": 2.331928073647079,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005530619779287597,
      "loss": 1.2769,
      "step": 27104
    },
    {
      "epoch": 2.3320141099544007,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005530350898171639,
      "loss": 1.1318,
      "step": 27105
    },
    {
      "epoch": 2.3321001462617224,
      "grad_norm": 1.25,
      "learning_rate": 0.0005530082015504507,
      "loss": 0.8697,
      "step": 27106
    },
    {
      "epoch": 2.332186182569044,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005529813131286988,
      "loss": 0.4585,
      "step": 27107
    },
    {
      "epoch": 2.3322722188763656,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005529544245519869,
      "loss": 1.5774,
      "step": 27108
    },
    {
      "epoch": 2.3323582551836877,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005529275358203937,
      "loss": 1.1521,
      "step": 27109
    },
    {
      "epoch": 2.3324442914910093,
      "grad_norm": 1.796875,
      "learning_rate": 0.0005529006469339977,
      "loss": 1.5836,
      "step": 27110
    },
    {
      "epoch": 2.332530327798331,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005528737578928775,
      "loss": 1.4997,
      "step": 27111
    },
    {
      "epoch": 2.3326163641056525,
      "grad_norm": 1.296875,
      "learning_rate": 0.000552846868697112,
      "loss": 1.1708,
      "step": 27112
    },
    {
      "epoch": 2.332702400412974,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005528199793467796,
      "loss": 0.7661,
      "step": 27113
    },
    {
      "epoch": 2.332788436720296,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0005527930898419591,
      "loss": 1.4181,
      "step": 27114
    },
    {
      "epoch": 2.3328744730276174,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005527662001827289,
      "loss": 2.0108,
      "step": 27115
    },
    {
      "epoch": 2.3329605093349395,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000552739310369168,
      "loss": 1.4238,
      "step": 27116
    },
    {
      "epoch": 2.333046545642261,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005527124204013549,
      "loss": 1.5245,
      "step": 27117
    },
    {
      "epoch": 2.3331325819495827,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005526855302793681,
      "loss": 1.0188,
      "step": 27118
    },
    {
      "epoch": 2.3332186182569044,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005526586400032865,
      "loss": 0.9811,
      "step": 27119
    },
    {
      "epoch": 2.333304654564226,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005526317495731885,
      "loss": 0.7737,
      "step": 27120
    },
    {
      "epoch": 2.3333906908715476,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005526048589891528,
      "loss": 1.1602,
      "step": 27121
    },
    {
      "epoch": 2.3334767271788697,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005525779682512582,
      "loss": 1.4444,
      "step": 27122
    },
    {
      "epoch": 2.3335627634861913,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005525510773595833,
      "loss": 1.3013,
      "step": 27123
    },
    {
      "epoch": 2.333648799793513,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005525241863142068,
      "loss": 1.5134,
      "step": 27124
    },
    {
      "epoch": 2.3337348361008345,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005524972951152072,
      "loss": 0.8042,
      "step": 27125
    },
    {
      "epoch": 2.333820872408156,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005524704037626631,
      "loss": 1.4319,
      "step": 27126
    },
    {
      "epoch": 2.333906908715478,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005524435122566534,
      "loss": 1.401,
      "step": 27127
    },
    {
      "epoch": 2.3339929450228,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005524166205972567,
      "loss": 1.1405,
      "step": 27128
    },
    {
      "epoch": 2.3340789813301215,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005523897287845514,
      "loss": 1.25,
      "step": 27129
    },
    {
      "epoch": 2.334165017637443,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005523628368186164,
      "loss": 0.8413,
      "step": 27130
    },
    {
      "epoch": 2.3342510539447647,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005523359446995303,
      "loss": 1.6562,
      "step": 27131
    },
    {
      "epoch": 2.3343370902520864,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005523090524273719,
      "loss": 1.245,
      "step": 27132
    },
    {
      "epoch": 2.334423126559408,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005522821600022194,
      "loss": 1.8523,
      "step": 27133
    },
    {
      "epoch": 2.3345091628667296,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005522552674241519,
      "loss": 1.3316,
      "step": 27134
    },
    {
      "epoch": 2.3345951991740517,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000552228374693248,
      "loss": 1.8317,
      "step": 27135
    },
    {
      "epoch": 2.3346812354813733,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000552201481809586,
      "loss": 1.5379,
      "step": 27136
    },
    {
      "epoch": 2.334767271788695,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000552174588773245,
      "loss": 1.1582,
      "step": 27137
    },
    {
      "epoch": 2.3348533080960165,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0005521476955843035,
      "loss": 1.2092,
      "step": 27138
    },
    {
      "epoch": 2.334939344403338,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005521208022428401,
      "loss": 0.9745,
      "step": 27139
    },
    {
      "epoch": 2.33502538071066,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005520939087489334,
      "loss": 1.202,
      "step": 27140
    },
    {
      "epoch": 2.3351114170179814,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005520670151026621,
      "loss": 0.6765,
      "step": 27141
    },
    {
      "epoch": 2.3351974533253035,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005520401213041051,
      "loss": 0.9396,
      "step": 27142
    },
    {
      "epoch": 2.335283489632625,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005520132273533408,
      "loss": 1.3038,
      "step": 27143
    },
    {
      "epoch": 2.3353695259399467,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005519863332504479,
      "loss": 1.0917,
      "step": 27144
    },
    {
      "epoch": 2.3354555622472684,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005519594389955052,
      "loss": 0.8693,
      "step": 27145
    },
    {
      "epoch": 2.33554159855459,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005519325445885911,
      "loss": 1.1854,
      "step": 27146
    },
    {
      "epoch": 2.3356276348619116,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005519056500297846,
      "loss": 1.1004,
      "step": 27147
    },
    {
      "epoch": 2.3357136711692332,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005518787553191639,
      "loss": 0.8095,
      "step": 27148
    },
    {
      "epoch": 2.3357997074765553,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000551851860456808,
      "loss": 1.6655,
      "step": 27149
    },
    {
      "epoch": 2.335885743783877,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005518249654427956,
      "loss": 1.0173,
      "step": 27150
    },
    {
      "epoch": 2.3359717800911985,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005517980702772053,
      "loss": 1.6675,
      "step": 27151
    },
    {
      "epoch": 2.33605781639852,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005517711749601156,
      "loss": 1.1937,
      "step": 27152
    },
    {
      "epoch": 2.336143852705842,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005517442794916053,
      "loss": 1.5169,
      "step": 27153
    },
    {
      "epoch": 2.3362298890131634,
      "grad_norm": 1.1875,
      "learning_rate": 0.000551717383871753,
      "loss": 1.358,
      "step": 27154
    },
    {
      "epoch": 2.336315925320485,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005516904881006375,
      "loss": 1.147,
      "step": 27155
    },
    {
      "epoch": 2.336401961627807,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005516635921783373,
      "loss": 1.8896,
      "step": 27156
    },
    {
      "epoch": 2.3364879979351287,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005516366961049313,
      "loss": 1.0552,
      "step": 27157
    },
    {
      "epoch": 2.3365740342424504,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005516097998804978,
      "loss": 1.3945,
      "step": 27158
    },
    {
      "epoch": 2.336660070549772,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005515829035051158,
      "loss": 0.7647,
      "step": 27159
    },
    {
      "epoch": 2.3367461068570936,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005515560069788637,
      "loss": 0.8737,
      "step": 27160
    },
    {
      "epoch": 2.3368321431644152,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005515291103018205,
      "loss": 1.7212,
      "step": 27161
    },
    {
      "epoch": 2.336918179471737,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005515022134740645,
      "loss": 1.2006,
      "step": 27162
    },
    {
      "epoch": 2.337004215779059,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005514753164956746,
      "loss": 1.5534,
      "step": 27163
    },
    {
      "epoch": 2.3370902520863805,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005514484193667294,
      "loss": 1.2955,
      "step": 27164
    },
    {
      "epoch": 2.337176288393702,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0005514215220873077,
      "loss": 1.8427,
      "step": 27165
    },
    {
      "epoch": 2.337262324701024,
      "grad_norm": 0.9375,
      "learning_rate": 0.000551394624657488,
      "loss": 1.1607,
      "step": 27166
    },
    {
      "epoch": 2.3373483610083454,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005513677270773488,
      "loss": 0.897,
      "step": 27167
    },
    {
      "epoch": 2.337434397315667,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005513408293469692,
      "loss": 1.9098,
      "step": 27168
    },
    {
      "epoch": 2.3375204336229887,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005513139314664278,
      "loss": 1.2085,
      "step": 27169
    },
    {
      "epoch": 2.3376064699303107,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005512870334358029,
      "loss": 0.9013,
      "step": 27170
    },
    {
      "epoch": 2.3376925062376324,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005512601352551734,
      "loss": 1.0751,
      "step": 27171
    },
    {
      "epoch": 2.337778542544954,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005512332369246181,
      "loss": 1.6119,
      "step": 27172
    },
    {
      "epoch": 2.3378645788522756,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005512063384442155,
      "loss": 1.4203,
      "step": 27173
    },
    {
      "epoch": 2.3379506151595972,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005511794398140443,
      "loss": 1.1738,
      "step": 27174
    },
    {
      "epoch": 2.338036651466919,
      "grad_norm": 1.25,
      "learning_rate": 0.0005511525410341831,
      "loss": 1.395,
      "step": 27175
    },
    {
      "epoch": 2.3381226877742405,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005511256421047108,
      "loss": 0.8341,
      "step": 27176
    },
    {
      "epoch": 2.3382087240815626,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005510987430257058,
      "loss": 1.2659,
      "step": 27177
    },
    {
      "epoch": 2.338294760388884,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000551071843797247,
      "loss": 1.4038,
      "step": 27178
    },
    {
      "epoch": 2.338380796696206,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005510449444194131,
      "loss": 0.8916,
      "step": 27179
    },
    {
      "epoch": 2.3384668330035274,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005510180448922826,
      "loss": 1.5699,
      "step": 27180
    },
    {
      "epoch": 2.338552869310849,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005509911452159341,
      "loss": 0.6243,
      "step": 27181
    },
    {
      "epoch": 2.338638905618171,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005509642453904465,
      "loss": 1.8985,
      "step": 27182
    },
    {
      "epoch": 2.3387249419254927,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005509373454158986,
      "loss": 0.9992,
      "step": 27183
    },
    {
      "epoch": 2.3388109782328144,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005509104452923687,
      "loss": 0.6744,
      "step": 27184
    },
    {
      "epoch": 2.338897014540136,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005508835450199355,
      "loss": 0.9156,
      "step": 27185
    },
    {
      "epoch": 2.3389830508474576,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000550856644598678,
      "loss": 0.7207,
      "step": 27186
    },
    {
      "epoch": 2.3390690871547792,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005508297440286747,
      "loss": 0.6678,
      "step": 27187
    },
    {
      "epoch": 2.339155123462101,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005508028433100043,
      "loss": 1.862,
      "step": 27188
    },
    {
      "epoch": 2.339241159769423,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005507759424427454,
      "loss": 1.5596,
      "step": 27189
    },
    {
      "epoch": 2.3393271960767446,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005507490414269768,
      "loss": 1.407,
      "step": 27190
    },
    {
      "epoch": 2.339413232384066,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005507221402627771,
      "loss": 0.9934,
      "step": 27191
    },
    {
      "epoch": 2.339499268691388,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000550695238950225,
      "loss": 1.0458,
      "step": 27192
    },
    {
      "epoch": 2.3395853049987094,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005506683374893992,
      "loss": 0.7421,
      "step": 27193
    },
    {
      "epoch": 2.339671341306031,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005506414358803783,
      "loss": 0.9357,
      "step": 27194
    },
    {
      "epoch": 2.3397573776133527,
      "grad_norm": 1.140625,
      "learning_rate": 0.000550614534123241,
      "loss": 1.2377,
      "step": 27195
    },
    {
      "epoch": 2.3398434139206747,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005505876322180662,
      "loss": 0.9725,
      "step": 27196
    },
    {
      "epoch": 2.3399294502279964,
      "grad_norm": 1.0,
      "learning_rate": 0.0005505607301649323,
      "loss": 0.7935,
      "step": 27197
    },
    {
      "epoch": 2.340015486535318,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005505338279639182,
      "loss": 2.2933,
      "step": 27198
    },
    {
      "epoch": 2.3401015228426396,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005505069256151024,
      "loss": 0.8646,
      "step": 27199
    },
    {
      "epoch": 2.3401875591499612,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005504800231185636,
      "loss": 1.2304,
      "step": 27200
    },
    {
      "epoch": 2.340273595457283,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005504531204743807,
      "loss": 1.0473,
      "step": 27201
    },
    {
      "epoch": 2.3403596317646045,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005504262176826322,
      "loss": 1.3689,
      "step": 27202
    },
    {
      "epoch": 2.3404456680719266,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005503993147433967,
      "loss": 0.8072,
      "step": 27203
    },
    {
      "epoch": 2.340531704379248,
      "grad_norm": 1.25,
      "learning_rate": 0.0005503724116567531,
      "loss": 1.2783,
      "step": 27204
    },
    {
      "epoch": 2.34061774068657,
      "grad_norm": 1.15625,
      "learning_rate": 0.00055034550842278,
      "loss": 1.9576,
      "step": 27205
    },
    {
      "epoch": 2.3407037769938914,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005503186050415562,
      "loss": 1.3036,
      "step": 27206
    },
    {
      "epoch": 2.340789813301213,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005502917015131602,
      "loss": 1.2973,
      "step": 27207
    },
    {
      "epoch": 2.3408758496085347,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005502647978376706,
      "loss": 1.6177,
      "step": 27208
    },
    {
      "epoch": 2.3409618859158563,
      "grad_norm": 1.0,
      "learning_rate": 0.0005502378940151663,
      "loss": 1.8847,
      "step": 27209
    },
    {
      "epoch": 2.3410479222231784,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000550210990045726,
      "loss": 1.1098,
      "step": 27210
    },
    {
      "epoch": 2.3411339585305,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005501840859294283,
      "loss": 0.9549,
      "step": 27211
    },
    {
      "epoch": 2.3412199948378216,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005501571816663519,
      "loss": 1.2941,
      "step": 27212
    },
    {
      "epoch": 2.3413060311451432,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005501302772565755,
      "loss": 1.6216,
      "step": 27213
    },
    {
      "epoch": 2.341392067452465,
      "grad_norm": 1.0,
      "learning_rate": 0.0005501033727001777,
      "loss": 0.8923,
      "step": 27214
    },
    {
      "epoch": 2.3414781037597865,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005500764679972375,
      "loss": 1.3279,
      "step": 27215
    },
    {
      "epoch": 2.341564140067108,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005500495631478331,
      "loss": 1.2995,
      "step": 27216
    },
    {
      "epoch": 2.34165017637443,
      "grad_norm": 1.46875,
      "learning_rate": 0.0005500226581520437,
      "loss": 1.9652,
      "step": 27217
    },
    {
      "epoch": 2.341736212681752,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005499957530099476,
      "loss": 1.3593,
      "step": 27218
    },
    {
      "epoch": 2.3418222489890734,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005499688477216235,
      "loss": 1.2171,
      "step": 27219
    },
    {
      "epoch": 2.341908285296395,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005499419422871505,
      "loss": 1.3979,
      "step": 27220
    },
    {
      "epoch": 2.3419943216037167,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005499150367066072,
      "loss": 1.5197,
      "step": 27221
    },
    {
      "epoch": 2.3420803579110383,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005498881309800718,
      "loss": 1.6412,
      "step": 27222
    },
    {
      "epoch": 2.34216639421836,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005498612251076234,
      "loss": 1.4401,
      "step": 27223
    },
    {
      "epoch": 2.342252430525682,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005498343190893407,
      "loss": 0.9945,
      "step": 27224
    },
    {
      "epoch": 2.3423384668330036,
      "grad_norm": 1.125,
      "learning_rate": 0.0005498074129253023,
      "loss": 1.0562,
      "step": 27225
    },
    {
      "epoch": 2.3424245031403252,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005497805066155867,
      "loss": 1.6558,
      "step": 27226
    },
    {
      "epoch": 2.342510539447647,
      "grad_norm": 0.984375,
      "learning_rate": 0.000549753600160273,
      "loss": 1.0613,
      "step": 27227
    },
    {
      "epoch": 2.3425965757549685,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005497266935594396,
      "loss": 0.8409,
      "step": 27228
    },
    {
      "epoch": 2.34268261206229,
      "grad_norm": 1.375,
      "learning_rate": 0.0005496997868131654,
      "loss": 1.1088,
      "step": 27229
    },
    {
      "epoch": 2.3427686483696117,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005496728799215288,
      "loss": 2.0765,
      "step": 27230
    },
    {
      "epoch": 2.342854684676934,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005496459728846088,
      "loss": 0.8781,
      "step": 27231
    },
    {
      "epoch": 2.3429407209842554,
      "grad_norm": 1.0,
      "learning_rate": 0.000549619065702484,
      "loss": 0.9839,
      "step": 27232
    },
    {
      "epoch": 2.343026757291577,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005495921583752331,
      "loss": 0.8996,
      "step": 27233
    },
    {
      "epoch": 2.3431127935988987,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005495652509029346,
      "loss": 1.2914,
      "step": 27234
    },
    {
      "epoch": 2.3431988299062203,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005495383432856676,
      "loss": 1.2612,
      "step": 27235
    },
    {
      "epoch": 2.343284866213542,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005495114355235106,
      "loss": 1.5071,
      "step": 27236
    },
    {
      "epoch": 2.343370902520864,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005494845276165422,
      "loss": 1.0788,
      "step": 27237
    },
    {
      "epoch": 2.3434569388281856,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005494576195648411,
      "loss": 0.5616,
      "step": 27238
    },
    {
      "epoch": 2.3435429751355072,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005494307113684862,
      "loss": 1.6459,
      "step": 27239
    },
    {
      "epoch": 2.343629011442829,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005494038030275561,
      "loss": 1.2835,
      "step": 27240
    },
    {
      "epoch": 2.3437150477501505,
      "grad_norm": 1.65625,
      "learning_rate": 0.0005493768945421294,
      "loss": 1.0561,
      "step": 27241
    },
    {
      "epoch": 2.343801084057472,
      "grad_norm": 1.078125,
      "learning_rate": 0.000549349985912285,
      "loss": 1.2782,
      "step": 27242
    },
    {
      "epoch": 2.343887120364794,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005493230771381012,
      "loss": 1.4774,
      "step": 27243
    },
    {
      "epoch": 2.343973156672116,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005492961682196573,
      "loss": 1.6019,
      "step": 27244
    },
    {
      "epoch": 2.3440591929794374,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005492692591570316,
      "loss": 1.331,
      "step": 27245
    },
    {
      "epoch": 2.344145229286759,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005492423499503029,
      "loss": 1.1022,
      "step": 27246
    },
    {
      "epoch": 2.3442312655940807,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005492154405995498,
      "loss": 0.9761,
      "step": 27247
    },
    {
      "epoch": 2.3443173019014023,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005491885311048513,
      "loss": 1.3331,
      "step": 27248
    },
    {
      "epoch": 2.344403338208724,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005491616214662858,
      "loss": 1.2936,
      "step": 27249
    },
    {
      "epoch": 2.344489374516046,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005491347116839321,
      "loss": 1.2648,
      "step": 27250
    },
    {
      "epoch": 2.3445754108233676,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005491078017578689,
      "loss": 1.777,
      "step": 27251
    },
    {
      "epoch": 2.3446614471306892,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005490808916881751,
      "loss": 0.9434,
      "step": 27252
    },
    {
      "epoch": 2.344747483438011,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000549053981474929,
      "loss": 2.1648,
      "step": 27253
    },
    {
      "epoch": 2.3448335197453325,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005490270711182097,
      "loss": 0.8466,
      "step": 27254
    },
    {
      "epoch": 2.344919556052654,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005490001606180958,
      "loss": 1.2524,
      "step": 27255
    },
    {
      "epoch": 2.3450055923599757,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005489732499746658,
      "loss": 1.1165,
      "step": 27256
    },
    {
      "epoch": 2.345091628667298,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005489463391879987,
      "loss": 1.138,
      "step": 27257
    },
    {
      "epoch": 2.3451776649746194,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000548919428258173,
      "loss": 1.5506,
      "step": 27258
    },
    {
      "epoch": 2.345263701281941,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005488925171852675,
      "loss": 1.0386,
      "step": 27259
    },
    {
      "epoch": 2.3453497375892627,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005488656059693609,
      "loss": 1.5316,
      "step": 27260
    },
    {
      "epoch": 2.3454357738965843,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005488386946105319,
      "loss": 1.3019,
      "step": 27261
    },
    {
      "epoch": 2.345521810203906,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005488117831088594,
      "loss": 1.1592,
      "step": 27262
    },
    {
      "epoch": 2.3456078465112276,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005487848714644216,
      "loss": 1.3529,
      "step": 27263
    },
    {
      "epoch": 2.3456938828185496,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005487579596772978,
      "loss": 1.2351,
      "step": 27264
    },
    {
      "epoch": 2.3457799191258712,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005487310477475663,
      "loss": 1.7283,
      "step": 27265
    },
    {
      "epoch": 2.345865955433193,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005487041356753061,
      "loss": 1.2303,
      "step": 27266
    },
    {
      "epoch": 2.3459519917405145,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005486772234605955,
      "loss": 0.6425,
      "step": 27267
    },
    {
      "epoch": 2.346038028047836,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0005486503111035139,
      "loss": 0.3774,
      "step": 27268
    },
    {
      "epoch": 2.3461240643551577,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005486233986041392,
      "loss": 1.7859,
      "step": 27269
    },
    {
      "epoch": 2.3462101006624794,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005485964859625507,
      "loss": 1.644,
      "step": 27270
    },
    {
      "epoch": 2.3462961369698014,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000548569573178827,
      "loss": 1.1935,
      "step": 27271
    },
    {
      "epoch": 2.346382173277123,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005485426602530466,
      "loss": 1.1128,
      "step": 27272
    },
    {
      "epoch": 2.3464682095844447,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005485157471852885,
      "loss": 1.0996,
      "step": 27273
    },
    {
      "epoch": 2.3465542458917663,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005484888339756313,
      "loss": 1.5864,
      "step": 27274
    },
    {
      "epoch": 2.346640282199088,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005484619206241535,
      "loss": 1.1228,
      "step": 27275
    },
    {
      "epoch": 2.3467263185064096,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005484350071309341,
      "loss": 1.4182,
      "step": 27276
    },
    {
      "epoch": 2.346812354813731,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005484080934960518,
      "loss": 1.1484,
      "step": 27277
    },
    {
      "epoch": 2.3468983911210533,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005483811797195852,
      "loss": 1.5549,
      "step": 27278
    },
    {
      "epoch": 2.346984427428375,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000548354265801613,
      "loss": 1.609,
      "step": 27279
    },
    {
      "epoch": 2.3470704637356965,
      "grad_norm": 1.34375,
      "learning_rate": 0.000548327351742214,
      "loss": 1.7817,
      "step": 27280
    },
    {
      "epoch": 2.347156500043018,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005483004375414669,
      "loss": 1.2612,
      "step": 27281
    },
    {
      "epoch": 2.3472425363503397,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005482735231994503,
      "loss": 1.3525,
      "step": 27282
    },
    {
      "epoch": 2.3473285726576614,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005482466087162432,
      "loss": 1.7768,
      "step": 27283
    },
    {
      "epoch": 2.347414608964983,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005482196940919242,
      "loss": 1.2791,
      "step": 27284
    },
    {
      "epoch": 2.347500645272305,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005481927793265719,
      "loss": 1.335,
      "step": 27285
    },
    {
      "epoch": 2.3475866815796267,
      "grad_norm": 1.125,
      "learning_rate": 0.000548165864420265,
      "loss": 0.5034,
      "step": 27286
    },
    {
      "epoch": 2.3476727178869483,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005481389493730824,
      "loss": 1.7706,
      "step": 27287
    },
    {
      "epoch": 2.34775875419427,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005481120341851026,
      "loss": 1.5629,
      "step": 27288
    },
    {
      "epoch": 2.3478447905015916,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005480851188564047,
      "loss": 1.4459,
      "step": 27289
    },
    {
      "epoch": 2.347930826808913,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0005480582033870669,
      "loss": 1.4447,
      "step": 27290
    },
    {
      "epoch": 2.348016863116235,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005480312877771683,
      "loss": 0.9478,
      "step": 27291
    },
    {
      "epoch": 2.348102899423557,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005480043720267878,
      "loss": 0.8982,
      "step": 27292
    },
    {
      "epoch": 2.3481889357308785,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005479774561360036,
      "loss": 0.9523,
      "step": 27293
    },
    {
      "epoch": 2.3482749720382,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005479505401048947,
      "loss": 1.0292,
      "step": 27294
    },
    {
      "epoch": 2.3483610083455217,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005479236239335398,
      "loss": 0.9165,
      "step": 27295
    },
    {
      "epoch": 2.3484470446528434,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005478967076220176,
      "loss": 1.2501,
      "step": 27296
    },
    {
      "epoch": 2.348533080960165,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005478697911704068,
      "loss": 0.9796,
      "step": 27297
    },
    {
      "epoch": 2.348619117267487,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005478428745787862,
      "loss": 0.8754,
      "step": 27298
    },
    {
      "epoch": 2.3487051535748087,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005478159578472346,
      "loss": 1.1558,
      "step": 27299
    },
    {
      "epoch": 2.3487911898821303,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005477890409758306,
      "loss": 0.8967,
      "step": 27300
    },
    {
      "epoch": 2.348877226189452,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0005477621239646528,
      "loss": 1.7815,
      "step": 27301
    },
    {
      "epoch": 2.3489632624967736,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005477352068137803,
      "loss": 1.7077,
      "step": 27302
    },
    {
      "epoch": 2.349049298804095,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005477082895232916,
      "loss": 1.3651,
      "step": 27303
    },
    {
      "epoch": 2.3491353351114173,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005476813720932653,
      "loss": 0.8124,
      "step": 27304
    },
    {
      "epoch": 2.349221371418739,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005476544545237801,
      "loss": 1.3814,
      "step": 27305
    },
    {
      "epoch": 2.3493074077260605,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005476275368149151,
      "loss": 1.966,
      "step": 27306
    },
    {
      "epoch": 2.349393444033382,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005476006189667487,
      "loss": 1.3611,
      "step": 27307
    },
    {
      "epoch": 2.3494794803407038,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005475737009793599,
      "loss": 1.2047,
      "step": 27308
    },
    {
      "epoch": 2.3495655166480254,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005475467828528273,
      "loss": 1.035,
      "step": 27309
    },
    {
      "epoch": 2.349651552955347,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005475198645872295,
      "loss": 1.0835,
      "step": 27310
    },
    {
      "epoch": 2.349737589262669,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005474929461826454,
      "loss": 1.4741,
      "step": 27311
    },
    {
      "epoch": 2.3498236255699907,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005474660276391536,
      "loss": 1.5717,
      "step": 27312
    },
    {
      "epoch": 2.3499096618773123,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005474391089568327,
      "loss": 1.1881,
      "step": 27313
    },
    {
      "epoch": 2.349995698184634,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000547412190135762,
      "loss": 1.4173,
      "step": 27314
    },
    {
      "epoch": 2.3500817344919556,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005473852711760197,
      "loss": 1.02,
      "step": 27315
    },
    {
      "epoch": 2.350167770799277,
      "grad_norm": 1.375,
      "learning_rate": 0.0005473583520776847,
      "loss": 1.2063,
      "step": 27316
    },
    {
      "epoch": 2.350253807106599,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005473314328408357,
      "loss": 1.1083,
      "step": 27317
    },
    {
      "epoch": 2.350339843413921,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005473045134655516,
      "loss": 1.5284,
      "step": 27318
    },
    {
      "epoch": 2.3504258797212425,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000547277593951911,
      "loss": 1.2372,
      "step": 27319
    },
    {
      "epoch": 2.350511916028564,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005472506742999925,
      "loss": 1.5571,
      "step": 27320
    },
    {
      "epoch": 2.3505979523358858,
      "grad_norm": 0.875,
      "learning_rate": 0.000547223754509875,
      "loss": 0.8801,
      "step": 27321
    },
    {
      "epoch": 2.3506839886432074,
      "grad_norm": 1.125,
      "learning_rate": 0.0005471968345816373,
      "loss": 0.922,
      "step": 27322
    },
    {
      "epoch": 2.350770024950529,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005471699145153578,
      "loss": 1.2473,
      "step": 27323
    },
    {
      "epoch": 2.3508560612578506,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005471429943111156,
      "loss": 0.8487,
      "step": 27324
    },
    {
      "epoch": 2.3509420975651727,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005471160739689893,
      "loss": 1.0046,
      "step": 27325
    },
    {
      "epoch": 2.3510281338724943,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005470891534890579,
      "loss": 1.3531,
      "step": 27326
    },
    {
      "epoch": 2.351114170179816,
      "grad_norm": 1.0,
      "learning_rate": 0.0005470622328713996,
      "loss": 0.7553,
      "step": 27327
    },
    {
      "epoch": 2.3512002064871376,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0005470353121160935,
      "loss": 0.7886,
      "step": 27328
    },
    {
      "epoch": 2.351286242794459,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0005470083912232182,
      "loss": 1.6846,
      "step": 27329
    },
    {
      "epoch": 2.351372279101781,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005469814701928526,
      "loss": 0.7879,
      "step": 27330
    },
    {
      "epoch": 2.3514583154091024,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005469545490250753,
      "loss": 1.6393,
      "step": 27331
    },
    {
      "epoch": 2.3515443517164245,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005469276277199651,
      "loss": 1.453,
      "step": 27332
    },
    {
      "epoch": 2.351630388023746,
      "grad_norm": 1.0,
      "learning_rate": 0.0005469007062776005,
      "loss": 1.2816,
      "step": 27333
    },
    {
      "epoch": 2.3517164243310678,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005468737846980608,
      "loss": 1.0971,
      "step": 27334
    },
    {
      "epoch": 2.3518024606383894,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005468468629814242,
      "loss": 1.7384,
      "step": 27335
    },
    {
      "epoch": 2.351888496945711,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005468199411277697,
      "loss": 1.4941,
      "step": 27336
    },
    {
      "epoch": 2.3519745332530326,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000546793019137176,
      "loss": 1.1661,
      "step": 27337
    },
    {
      "epoch": 2.3520605695603543,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005467660970097217,
      "loss": 0.8683,
      "step": 27338
    },
    {
      "epoch": 2.3521466058676763,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005467391747454857,
      "loss": 1.3973,
      "step": 27339
    },
    {
      "epoch": 2.352232642174998,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005467122523445468,
      "loss": 1.2411,
      "step": 27340
    },
    {
      "epoch": 2.3523186784823196,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005466853298069836,
      "loss": 0.8054,
      "step": 27341
    },
    {
      "epoch": 2.352404714789641,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005466584071328747,
      "loss": 1.6009,
      "step": 27342
    },
    {
      "epoch": 2.352490751096963,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005466314843222993,
      "loss": 1.1267,
      "step": 27343
    },
    {
      "epoch": 2.3525767874042844,
      "grad_norm": 4.1875,
      "learning_rate": 0.0005466045613753358,
      "loss": 1.9825,
      "step": 27344
    },
    {
      "epoch": 2.352662823711606,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000546577638292063,
      "loss": 0.7144,
      "step": 27345
    },
    {
      "epoch": 2.352748860018928,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005465507150725596,
      "loss": 0.7777,
      "step": 27346
    },
    {
      "epoch": 2.3528348963262498,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005465237917169044,
      "loss": 0.6016,
      "step": 27347
    },
    {
      "epoch": 2.3529209326335714,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005464968682251763,
      "loss": 0.8252,
      "step": 27348
    },
    {
      "epoch": 2.353006968940893,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0005464699445974539,
      "loss": 0.7354,
      "step": 27349
    },
    {
      "epoch": 2.3530930052482146,
      "grad_norm": 1.375,
      "learning_rate": 0.0005464430208338158,
      "loss": 1.2983,
      "step": 27350
    },
    {
      "epoch": 2.3531790415555363,
      "grad_norm": 1.5625,
      "learning_rate": 0.0005464160969343411,
      "loss": 1.5508,
      "step": 27351
    },
    {
      "epoch": 2.353265077862858,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005463891728991083,
      "loss": 1.5005,
      "step": 27352
    },
    {
      "epoch": 2.35335111417018,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005463622487281959,
      "loss": 1.167,
      "step": 27353
    },
    {
      "epoch": 2.3534371504775016,
      "grad_norm": 1.125,
      "learning_rate": 0.0005463353244216833,
      "loss": 0.6453,
      "step": 27354
    },
    {
      "epoch": 2.353523186784823,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005463083999796488,
      "loss": 0.6822,
      "step": 27355
    },
    {
      "epoch": 2.353609223092145,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005462814754021712,
      "loss": 0.8115,
      "step": 27356
    },
    {
      "epoch": 2.3536952593994664,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005462545506893291,
      "loss": 1.1911,
      "step": 27357
    },
    {
      "epoch": 2.3537812957067885,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005462276258412018,
      "loss": 0.6396,
      "step": 27358
    },
    {
      "epoch": 2.35386733201411,
      "grad_norm": 4.40625,
      "learning_rate": 0.0005462007008578675,
      "loss": 1.619,
      "step": 27359
    },
    {
      "epoch": 2.3539533683214318,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005461737757394052,
      "loss": 0.7285,
      "step": 27360
    },
    {
      "epoch": 2.3540394046287534,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005461468504858933,
      "loss": 1.2996,
      "step": 27361
    },
    {
      "epoch": 2.354125440936075,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005461199250974111,
      "loss": 1.3527,
      "step": 27362
    },
    {
      "epoch": 2.3542114772433966,
      "grad_norm": 1.3125,
      "learning_rate": 0.000546092999574037,
      "loss": 1.8399,
      "step": 27363
    },
    {
      "epoch": 2.3542975135507183,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005460660739158499,
      "loss": 1.315,
      "step": 27364
    },
    {
      "epoch": 2.3543835498580403,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005460391481229284,
      "loss": 2.0192,
      "step": 27365
    },
    {
      "epoch": 2.354469586165362,
      "grad_norm": 4.625,
      "learning_rate": 0.0005460122221953515,
      "loss": 1.2483,
      "step": 27366
    },
    {
      "epoch": 2.3545556224726836,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005459852961331978,
      "loss": 1.0308,
      "step": 27367
    },
    {
      "epoch": 2.354641658780005,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005459583699365461,
      "loss": 1.109,
      "step": 27368
    },
    {
      "epoch": 2.354727695087327,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005459314436054748,
      "loss": 1.5752,
      "step": 27369
    },
    {
      "epoch": 2.3548137313946484,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005459045171400631,
      "loss": 1.1228,
      "step": 27370
    },
    {
      "epoch": 2.35489976770197,
      "grad_norm": 1.734375,
      "learning_rate": 0.0005458775905403898,
      "loss": 1.4577,
      "step": 27371
    },
    {
      "epoch": 2.354985804009292,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005458506638065332,
      "loss": 0.8563,
      "step": 27372
    },
    {
      "epoch": 2.3550718403166138,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005458237369385725,
      "loss": 0.6656,
      "step": 27373
    },
    {
      "epoch": 2.3551578766239354,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005457968099365863,
      "loss": 1.3576,
      "step": 27374
    },
    {
      "epoch": 2.355243912931257,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005457698828006533,
      "loss": 0.6316,
      "step": 27375
    },
    {
      "epoch": 2.3553299492385786,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005457429555308521,
      "loss": 1.2043,
      "step": 27376
    },
    {
      "epoch": 2.3554159855459003,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005457160281272619,
      "loss": 1.5518,
      "step": 27377
    },
    {
      "epoch": 2.355502021853222,
      "grad_norm": 1.125,
      "learning_rate": 0.0005456891005899612,
      "loss": 0.9197,
      "step": 27378
    },
    {
      "epoch": 2.355588058160544,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005456621729190286,
      "loss": 1.38,
      "step": 27379
    },
    {
      "epoch": 2.3556740944678656,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005456352451145431,
      "loss": 1.3228,
      "step": 27380
    },
    {
      "epoch": 2.355760130775187,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005456083171765834,
      "loss": 1.1616,
      "step": 27381
    },
    {
      "epoch": 2.355846167082509,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005455813891052282,
      "loss": 1.1453,
      "step": 27382
    },
    {
      "epoch": 2.3559322033898304,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005455544609005565,
      "loss": 1.7298,
      "step": 27383
    },
    {
      "epoch": 2.356018239697152,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005455275325626466,
      "loss": 1.8788,
      "step": 27384
    },
    {
      "epoch": 2.3561042760044737,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005455006040915776,
      "loss": 1.0135,
      "step": 27385
    },
    {
      "epoch": 2.3561903123117958,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005454736754874283,
      "loss": 0.8731,
      "step": 27386
    },
    {
      "epoch": 2.3562763486191174,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005454467467502772,
      "loss": 1.1068,
      "step": 27387
    },
    {
      "epoch": 2.356362384926439,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005454198178802033,
      "loss": 1.2573,
      "step": 27388
    },
    {
      "epoch": 2.3564484212337606,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005453928888772853,
      "loss": 0.9456,
      "step": 27389
    },
    {
      "epoch": 2.3565344575410823,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005453659597416019,
      "loss": 0.6592,
      "step": 27390
    },
    {
      "epoch": 2.356620493848404,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005453390304732317,
      "loss": 1.4621,
      "step": 27391
    },
    {
      "epoch": 2.3567065301557255,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005453121010722538,
      "loss": 0.6369,
      "step": 27392
    },
    {
      "epoch": 2.3567925664630476,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005452851715387469,
      "loss": 1.6511,
      "step": 27393
    },
    {
      "epoch": 2.356878602770369,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005452582418727894,
      "loss": 0.725,
      "step": 27394
    },
    {
      "epoch": 2.356964639077691,
      "grad_norm": 1.125,
      "learning_rate": 0.0005452313120744608,
      "loss": 1.0407,
      "step": 27395
    },
    {
      "epoch": 2.3570506753850125,
      "grad_norm": 1.1875,
      "learning_rate": 0.000545204382143839,
      "loss": 1.3331,
      "step": 27396
    },
    {
      "epoch": 2.357136711692334,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005451774520810035,
      "loss": 0.8048,
      "step": 27397
    },
    {
      "epoch": 2.3572227479996557,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005451505218860324,
      "loss": 1.4814,
      "step": 27398
    },
    {
      "epoch": 2.3573087843069773,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005451235915590049,
      "loss": 1.7864,
      "step": 27399
    },
    {
      "epoch": 2.3573948206142994,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005450966610999998,
      "loss": 1.3944,
      "step": 27400
    },
    {
      "epoch": 2.357480856921621,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005450697305090957,
      "loss": 1.2127,
      "step": 27401
    },
    {
      "epoch": 2.3575668932289426,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005450427997863714,
      "loss": 1.2399,
      "step": 27402
    },
    {
      "epoch": 2.3576529295362643,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005450158689319057,
      "loss": 1.6489,
      "step": 27403
    },
    {
      "epoch": 2.357738965843586,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005449889379457773,
      "loss": 1.722,
      "step": 27404
    },
    {
      "epoch": 2.3578250021509075,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000544962006828065,
      "loss": 0.9261,
      "step": 27405
    },
    {
      "epoch": 2.357911038458229,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005449350755788475,
      "loss": 0.7499,
      "step": 27406
    },
    {
      "epoch": 2.357997074765551,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005449081441982038,
      "loss": 0.804,
      "step": 27407
    },
    {
      "epoch": 2.358083111072873,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005448812126862124,
      "loss": 1.1142,
      "step": 27408
    },
    {
      "epoch": 2.3581691473801945,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005448542810429523,
      "loss": 0.7545,
      "step": 27409
    },
    {
      "epoch": 2.358255183687516,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005448273492685019,
      "loss": 1.1419,
      "step": 27410
    },
    {
      "epoch": 2.3583412199948377,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005448004173629406,
      "loss": 0.957,
      "step": 27411
    },
    {
      "epoch": 2.3584272563021593,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005447734853263465,
      "loss": 0.8742,
      "step": 27412
    },
    {
      "epoch": 2.3585132926094814,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005447465531587986,
      "loss": 1.1773,
      "step": 27413
    },
    {
      "epoch": 2.358599328916803,
      "grad_norm": 1.0625,
      "learning_rate": 0.000544719620860376,
      "loss": 1.0496,
      "step": 27414
    },
    {
      "epoch": 2.3586853652241246,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005446926884311571,
      "loss": 1.3191,
      "step": 27415
    },
    {
      "epoch": 2.3587714015314463,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005446657558712208,
      "loss": 0.551,
      "step": 27416
    },
    {
      "epoch": 2.358857437838768,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005446388231806456,
      "loss": 1.2314,
      "step": 27417
    },
    {
      "epoch": 2.3589434741460895,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005446118903595108,
      "loss": 1.0754,
      "step": 27418
    },
    {
      "epoch": 2.3590295104534116,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005445849574078948,
      "loss": 0.7085,
      "step": 27419
    },
    {
      "epoch": 2.359115546760733,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005445580243258765,
      "loss": 0.8822,
      "step": 27420
    },
    {
      "epoch": 2.359201583068055,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005445310911135346,
      "loss": 1.5694,
      "step": 27421
    },
    {
      "epoch": 2.3592876193753765,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0005445041577709481,
      "loss": 0.6801,
      "step": 27422
    },
    {
      "epoch": 2.359373655682698,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005444772242981955,
      "loss": 0.6675,
      "step": 27423
    },
    {
      "epoch": 2.3594596919900197,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005444502906953555,
      "loss": 0.6403,
      "step": 27424
    },
    {
      "epoch": 2.3595457282973413,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005444233569625072,
      "loss": 0.9206,
      "step": 27425
    },
    {
      "epoch": 2.3596317646046634,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005443964230997292,
      "loss": 1.4161,
      "step": 27426
    },
    {
      "epoch": 2.359717800911985,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005443694891071004,
      "loss": 1.0687,
      "step": 27427
    },
    {
      "epoch": 2.3598038372193066,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005443425549846993,
      "loss": 0.9878,
      "step": 27428
    },
    {
      "epoch": 2.3598898735266283,
      "grad_norm": 0.875,
      "learning_rate": 0.0005443156207326048,
      "loss": 1.0867,
      "step": 27429
    },
    {
      "epoch": 2.35997590983395,
      "grad_norm": 1.265625,
      "learning_rate": 0.000544288686350896,
      "loss": 1.5679,
      "step": 27430
    },
    {
      "epoch": 2.3600619461412715,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005442617518396513,
      "loss": 1.4188,
      "step": 27431
    },
    {
      "epoch": 2.360147982448593,
      "grad_norm": 1.25,
      "learning_rate": 0.0005442348171989495,
      "loss": 1.7925,
      "step": 27432
    },
    {
      "epoch": 2.360234018755915,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005442078824288695,
      "loss": 1.695,
      "step": 27433
    },
    {
      "epoch": 2.360320055063237,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005441809475294901,
      "loss": 0.7729,
      "step": 27434
    },
    {
      "epoch": 2.3604060913705585,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005441540125008898,
      "loss": 1.709,
      "step": 27435
    },
    {
      "epoch": 2.36049212767788,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005441270773431477,
      "loss": 2.0289,
      "step": 27436
    },
    {
      "epoch": 2.3605781639852017,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005441001420563427,
      "loss": 1.4055,
      "step": 27437
    },
    {
      "epoch": 2.3606642002925233,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005440732066405531,
      "loss": 1.6967,
      "step": 27438
    },
    {
      "epoch": 2.360750236599845,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005440462710958582,
      "loss": 0.7008,
      "step": 27439
    },
    {
      "epoch": 2.360836272907167,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005440193354223363,
      "loss": 0.9375,
      "step": 27440
    },
    {
      "epoch": 2.3609223092144886,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005439923996200664,
      "loss": 0.5145,
      "step": 27441
    },
    {
      "epoch": 2.3610083455218103,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005439654636891274,
      "loss": 1.2324,
      "step": 27442
    },
    {
      "epoch": 2.361094381829132,
      "grad_norm": 1.46875,
      "learning_rate": 0.000543938527629598,
      "loss": 0.9251,
      "step": 27443
    },
    {
      "epoch": 2.3611804181364535,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005439115914415569,
      "loss": 1.2494,
      "step": 27444
    },
    {
      "epoch": 2.361266454443775,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005438846551250829,
      "loss": 0.8825,
      "step": 27445
    },
    {
      "epoch": 2.3613524907510968,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005438577186802549,
      "loss": 0.9986,
      "step": 27446
    },
    {
      "epoch": 2.361438527058419,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005438307821071514,
      "loss": 1.6644,
      "step": 27447
    },
    {
      "epoch": 2.3615245633657405,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005438038454058516,
      "loss": 1.5032,
      "step": 27448
    },
    {
      "epoch": 2.361610599673062,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005437769085764341,
      "loss": 1.2929,
      "step": 27449
    },
    {
      "epoch": 2.3616966359803837,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005437499716189774,
      "loss": 0.9585,
      "step": 27450
    },
    {
      "epoch": 2.3617826722877053,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005437230345335608,
      "loss": 1.2594,
      "step": 27451
    },
    {
      "epoch": 2.361868708595027,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005436960973202627,
      "loss": 1.4947,
      "step": 27452
    },
    {
      "epoch": 2.3619547449023486,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005436691599791621,
      "loss": 1.8257,
      "step": 27453
    },
    {
      "epoch": 2.3620407812096706,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005436422225103375,
      "loss": 1.1703,
      "step": 27454
    },
    {
      "epoch": 2.3621268175169923,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000543615284913868,
      "loss": 0.5611,
      "step": 27455
    },
    {
      "epoch": 2.362212853824314,
      "grad_norm": 1.25,
      "learning_rate": 0.0005435883471898323,
      "loss": 1.2941,
      "step": 27456
    },
    {
      "epoch": 2.3622988901316355,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005435614093383093,
      "loss": 0.9251,
      "step": 27457
    },
    {
      "epoch": 2.362384926438957,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005435344713593773,
      "loss": 1.496,
      "step": 27458
    },
    {
      "epoch": 2.3624709627462788,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005435075332531157,
      "loss": 1.6359,
      "step": 27459
    },
    {
      "epoch": 2.3625569990536004,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005434805950196031,
      "loss": 0.7313,
      "step": 27460
    },
    {
      "epoch": 2.3626430353609225,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000543453656658918,
      "loss": 1.1621,
      "step": 27461
    },
    {
      "epoch": 2.362729071668244,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005434267181711395,
      "loss": 1.1437,
      "step": 27462
    },
    {
      "epoch": 2.3628151079755657,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005433997795563463,
      "loss": 1.0886,
      "step": 27463
    },
    {
      "epoch": 2.3629011442828873,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005433728408146172,
      "loss": 0.9285,
      "step": 27464
    },
    {
      "epoch": 2.362987180590209,
      "grad_norm": 1.125,
      "learning_rate": 0.0005433459019460311,
      "loss": 1.7886,
      "step": 27465
    },
    {
      "epoch": 2.3630732168975306,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005433189629506664,
      "loss": 1.2237,
      "step": 27466
    },
    {
      "epoch": 2.363159253204852,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005432920238286023,
      "loss": 1.142,
      "step": 27467
    },
    {
      "epoch": 2.3632452895121743,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005432650845799175,
      "loss": 1.4724,
      "step": 27468
    },
    {
      "epoch": 2.363331325819496,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005432381452046906,
      "loss": 1.434,
      "step": 27469
    },
    {
      "epoch": 2.3634173621268175,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005432112057030007,
      "loss": 1.4234,
      "step": 27470
    },
    {
      "epoch": 2.363503398434139,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005431842660749264,
      "loss": 1.0917,
      "step": 27471
    },
    {
      "epoch": 2.3635894347414608,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005431573263205464,
      "loss": 0.9663,
      "step": 27472
    },
    {
      "epoch": 2.3636754710487824,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005431303864399395,
      "loss": 1.4092,
      "step": 27473
    },
    {
      "epoch": 2.3637615073561045,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0005431034464331849,
      "loss": 0.9591,
      "step": 27474
    },
    {
      "epoch": 2.363847543663426,
      "grad_norm": 0.8125,
      "learning_rate": 0.000543076506300361,
      "loss": 0.7994,
      "step": 27475
    },
    {
      "epoch": 2.3639335799707477,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005430495660415466,
      "loss": 0.8659,
      "step": 27476
    },
    {
      "epoch": 2.3640196162780693,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005430226256568207,
      "loss": 1.4919,
      "step": 27477
    },
    {
      "epoch": 2.364105652585391,
      "grad_norm": 1.546875,
      "learning_rate": 0.000542995685146262,
      "loss": 1.1549,
      "step": 27478
    },
    {
      "epoch": 2.3641916888927126,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005429687445099493,
      "loss": 1.3738,
      "step": 27479
    },
    {
      "epoch": 2.3642777252000347,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005429418037479613,
      "loss": 1.143,
      "step": 27480
    },
    {
      "epoch": 2.3643637615073563,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005429148628603768,
      "loss": 1.1267,
      "step": 27481
    },
    {
      "epoch": 2.364449797814678,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005428879218472749,
      "loss": 1.2689,
      "step": 27482
    },
    {
      "epoch": 2.3645358341219995,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000542860980708734,
      "loss": 0.8475,
      "step": 27483
    },
    {
      "epoch": 2.364621870429321,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000542834039444833,
      "loss": 0.9195,
      "step": 27484
    },
    {
      "epoch": 2.3647079067366428,
      "grad_norm": 0.90625,
      "learning_rate": 0.000542807098055651,
      "loss": 1.0161,
      "step": 27485
    },
    {
      "epoch": 2.3647939430439644,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005427801565412665,
      "loss": 1.2718,
      "step": 27486
    },
    {
      "epoch": 2.3648799793512865,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005427532149017583,
      "loss": 1.0408,
      "step": 27487
    },
    {
      "epoch": 2.364966015658608,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005427262731372051,
      "loss": 0.6934,
      "step": 27488
    },
    {
      "epoch": 2.3650520519659297,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005426993312476861,
      "loss": 0.853,
      "step": 27489
    },
    {
      "epoch": 2.3651380882732513,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005426723892332798,
      "loss": 1.5975,
      "step": 27490
    },
    {
      "epoch": 2.365224124580573,
      "grad_norm": 0.890625,
      "learning_rate": 0.000542645447094065,
      "loss": 1.3457,
      "step": 27491
    },
    {
      "epoch": 2.3653101608878946,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005426185048301206,
      "loss": 1.0422,
      "step": 27492
    },
    {
      "epoch": 2.365396197195216,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005425915624415252,
      "loss": 0.8712,
      "step": 27493
    },
    {
      "epoch": 2.3654822335025383,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005425646199283582,
      "loss": 1.3913,
      "step": 27494
    },
    {
      "epoch": 2.36556826980986,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005425376772906975,
      "loss": 1.6953,
      "step": 27495
    },
    {
      "epoch": 2.3656543061171815,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0005425107345286227,
      "loss": 1.4906,
      "step": 27496
    },
    {
      "epoch": 2.365740342424503,
      "grad_norm": 1.046875,
      "learning_rate": 0.000542483791642212,
      "loss": 1.0713,
      "step": 27497
    },
    {
      "epoch": 2.3658263787318248,
      "grad_norm": 1.46875,
      "learning_rate": 0.0005424568486315448,
      "loss": 1.3989,
      "step": 27498
    },
    {
      "epoch": 2.3659124150391464,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005424299054966992,
      "loss": 1.1267,
      "step": 27499
    },
    {
      "epoch": 2.365998451346468,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005424029622377546,
      "loss": 1.4176,
      "step": 27500
    },
    {
      "epoch": 2.36608448765379,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005423760188547896,
      "loss": 1.4901,
      "step": 27501
    },
    {
      "epoch": 2.3661705239611117,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005423490753478829,
      "loss": 1.3128,
      "step": 27502
    },
    {
      "epoch": 2.3662565602684333,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005423221317171134,
      "loss": 0.7803,
      "step": 27503
    },
    {
      "epoch": 2.366342596575755,
      "grad_norm": 1.125,
      "learning_rate": 0.0005422951879625599,
      "loss": 1.1414,
      "step": 27504
    },
    {
      "epoch": 2.3664286328830766,
      "grad_norm": 1.125,
      "learning_rate": 0.0005422682440843013,
      "loss": 1.2991,
      "step": 27505
    },
    {
      "epoch": 2.366514669190398,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0005422413000824161,
      "loss": 1.0793,
      "step": 27506
    },
    {
      "epoch": 2.36660070549772,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005422143559569836,
      "loss": 0.5538,
      "step": 27507
    },
    {
      "epoch": 2.366686741805042,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005421874117080821,
      "loss": 1.1679,
      "step": 27508
    },
    {
      "epoch": 2.3667727781123635,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005421604673357906,
      "loss": 1.8758,
      "step": 27509
    },
    {
      "epoch": 2.366858814419685,
      "grad_norm": 0.984375,
      "learning_rate": 0.000542133522840188,
      "loss": 1.1326,
      "step": 27510
    },
    {
      "epoch": 2.3669448507270068,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0005421065782213531,
      "loss": 0.9817,
      "step": 27511
    },
    {
      "epoch": 2.3670308870343284,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005420796334793647,
      "loss": 1.6318,
      "step": 27512
    },
    {
      "epoch": 2.36711692334165,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005420526886143013,
      "loss": 0.9925,
      "step": 27513
    },
    {
      "epoch": 2.3672029596489716,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005420257436262421,
      "loss": 0.5879,
      "step": 27514
    },
    {
      "epoch": 2.3672889959562937,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005419987985152658,
      "loss": 1.3433,
      "step": 27515
    },
    {
      "epoch": 2.3673750322636153,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005419718532814513,
      "loss": 1.0051,
      "step": 27516
    },
    {
      "epoch": 2.367461068570937,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005419449079248771,
      "loss": 0.9046,
      "step": 27517
    },
    {
      "epoch": 2.3675471048782586,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005419179624456223,
      "loss": 1.4856,
      "step": 27518
    },
    {
      "epoch": 2.36763314118558,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005418910168437655,
      "loss": 0.6485,
      "step": 27519
    },
    {
      "epoch": 2.367719177492902,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005418640711193859,
      "loss": 0.8494,
      "step": 27520
    },
    {
      "epoch": 2.3678052138002235,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005418371252725618,
      "loss": 1.365,
      "step": 27521
    },
    {
      "epoch": 2.3678912501075455,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005418101793033722,
      "loss": 0.9229,
      "step": 27522
    },
    {
      "epoch": 2.367977286414867,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005417832332118961,
      "loss": 0.9626,
      "step": 27523
    },
    {
      "epoch": 2.368063322722189,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005417562869982121,
      "loss": 1.542,
      "step": 27524
    },
    {
      "epoch": 2.3681493590295104,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005417293406623991,
      "loss": 1.6459,
      "step": 27525
    },
    {
      "epoch": 2.368235395336832,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005417023942045358,
      "loss": 1.0968,
      "step": 27526
    },
    {
      "epoch": 2.3683214316441537,
      "grad_norm": 1.125,
      "learning_rate": 0.0005416754476247013,
      "loss": 1.1004,
      "step": 27527
    },
    {
      "epoch": 2.3684074679514753,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000541648500922974,
      "loss": 0.7433,
      "step": 27528
    },
    {
      "epoch": 2.3684935042587973,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005416215540994331,
      "loss": 0.8801,
      "step": 27529
    },
    {
      "epoch": 2.368579540566119,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005415946071541572,
      "loss": 1.5679,
      "step": 27530
    },
    {
      "epoch": 2.3686655768734406,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005415676600872252,
      "loss": 1.3695,
      "step": 27531
    },
    {
      "epoch": 2.368751613180762,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005415407128987158,
      "loss": 2.0554,
      "step": 27532
    },
    {
      "epoch": 2.368837649488084,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005415137655887079,
      "loss": 0.8018,
      "step": 27533
    },
    {
      "epoch": 2.368923685795406,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005414868181572804,
      "loss": 0.763,
      "step": 27534
    },
    {
      "epoch": 2.3690097221027275,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000541459870604512,
      "loss": 1.2236,
      "step": 27535
    },
    {
      "epoch": 2.369095758410049,
      "grad_norm": 1.59375,
      "learning_rate": 0.0005414329229304815,
      "loss": 1.0104,
      "step": 27536
    },
    {
      "epoch": 2.369181794717371,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005414059751352678,
      "loss": 0.5557,
      "step": 27537
    },
    {
      "epoch": 2.3692678310246924,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005413790272189496,
      "loss": 1.2348,
      "step": 27538
    },
    {
      "epoch": 2.369353867332014,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005413520791816059,
      "loss": 1.4093,
      "step": 27539
    },
    {
      "epoch": 2.3694399036393357,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005413251310233153,
      "loss": 1.1354,
      "step": 27540
    },
    {
      "epoch": 2.3695259399466577,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005412981827441568,
      "loss": 0.7303,
      "step": 27541
    },
    {
      "epoch": 2.3696119762539793,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005412712343442093,
      "loss": 0.8155,
      "step": 27542
    },
    {
      "epoch": 2.369698012561301,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005412442858235512,
      "loss": 1.9504,
      "step": 27543
    },
    {
      "epoch": 2.3697840488686226,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005412173371822616,
      "loss": 1.9707,
      "step": 27544
    },
    {
      "epoch": 2.369870085175944,
      "grad_norm": 1.375,
      "learning_rate": 0.0005411903884204194,
      "loss": 1.3236,
      "step": 27545
    },
    {
      "epoch": 2.369956121483266,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005411634395381033,
      "loss": 1.0804,
      "step": 27546
    },
    {
      "epoch": 2.3700421577905875,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005411364905353921,
      "loss": 1.1818,
      "step": 27547
    },
    {
      "epoch": 2.3701281940979095,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005411095414123647,
      "loss": 1.4837,
      "step": 27548
    },
    {
      "epoch": 2.370214230405231,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005410825921690999,
      "loss": 0.933,
      "step": 27549
    },
    {
      "epoch": 2.370300266712553,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005410556428056767,
      "loss": 1.1438,
      "step": 27550
    },
    {
      "epoch": 2.3703863030198744,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005410286933221735,
      "loss": 1.0175,
      "step": 27551
    },
    {
      "epoch": 2.370472339327196,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005410017437186693,
      "loss": 1.2278,
      "step": 27552
    },
    {
      "epoch": 2.3705583756345177,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005409747939952432,
      "loss": 0.9287,
      "step": 27553
    },
    {
      "epoch": 2.3706444119418393,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005409478441519736,
      "loss": 1.3137,
      "step": 27554
    },
    {
      "epoch": 2.3707304482491613,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005409208941889395,
      "loss": 1.1191,
      "step": 27555
    },
    {
      "epoch": 2.370816484556483,
      "grad_norm": 1.671875,
      "learning_rate": 0.0005408939441062199,
      "loss": 1.1971,
      "step": 27556
    },
    {
      "epoch": 2.3709025208638046,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005408669939038935,
      "loss": 0.7473,
      "step": 27557
    },
    {
      "epoch": 2.370988557171126,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000540840043582039,
      "loss": 1.6046,
      "step": 27558
    },
    {
      "epoch": 2.371074593478448,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005408130931407353,
      "loss": 1.0342,
      "step": 27559
    },
    {
      "epoch": 2.3711606297857695,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005407861425800612,
      "loss": 0.94,
      "step": 27560
    },
    {
      "epoch": 2.371246666093091,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005407591919000956,
      "loss": 1.9425,
      "step": 27561
    },
    {
      "epoch": 2.371332702400413,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005407322411009173,
      "loss": 0.9309,
      "step": 27562
    },
    {
      "epoch": 2.371418738707735,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005407052901826052,
      "loss": 1.0707,
      "step": 27563
    },
    {
      "epoch": 2.3715047750150564,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005406783391452379,
      "loss": 1.2896,
      "step": 27564
    },
    {
      "epoch": 2.371590811322378,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005406513879888943,
      "loss": 1.3349,
      "step": 27565
    },
    {
      "epoch": 2.3716768476296997,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005406244367136536,
      "loss": 2.2517,
      "step": 27566
    },
    {
      "epoch": 2.3717628839370213,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005405974853195941,
      "loss": 1.3446,
      "step": 27567
    },
    {
      "epoch": 2.371848920244343,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005405705338067949,
      "loss": 1.3774,
      "step": 27568
    },
    {
      "epoch": 2.371934956551665,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005405435821753348,
      "loss": 1.3559,
      "step": 27569
    },
    {
      "epoch": 2.3720209928589866,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005405166304252925,
      "loss": 0.7515,
      "step": 27570
    },
    {
      "epoch": 2.3721070291663082,
      "grad_norm": 1.3125,
      "learning_rate": 0.000540489678556747,
      "loss": 1.0692,
      "step": 27571
    },
    {
      "epoch": 2.37219306547363,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005404627265697772,
      "loss": 1.5031,
      "step": 27572
    },
    {
      "epoch": 2.3722791017809515,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005404357744644617,
      "loss": 0.7961,
      "step": 27573
    },
    {
      "epoch": 2.372365138088273,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005404088222408794,
      "loss": 1.9572,
      "step": 27574
    },
    {
      "epoch": 2.3724511743955947,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005403818698991091,
      "loss": 0.9889,
      "step": 27575
    },
    {
      "epoch": 2.372537210702917,
      "grad_norm": 1.96875,
      "learning_rate": 0.0005403549174392298,
      "loss": 1.5599,
      "step": 27576
    },
    {
      "epoch": 2.3726232470102384,
      "grad_norm": 1.0,
      "learning_rate": 0.0005403279648613202,
      "loss": 1.2388,
      "step": 27577
    },
    {
      "epoch": 2.37270928331756,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000540301012165459,
      "loss": 0.7971,
      "step": 27578
    },
    {
      "epoch": 2.3727953196248817,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005402740593517253,
      "loss": 0.991,
      "step": 27579
    },
    {
      "epoch": 2.3728813559322033,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000540247106420198,
      "loss": 1.2429,
      "step": 27580
    },
    {
      "epoch": 2.372967392239525,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005402201533709555,
      "loss": 0.6826,
      "step": 27581
    },
    {
      "epoch": 2.3730534285468465,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005401932002040769,
      "loss": 1.6532,
      "step": 27582
    },
    {
      "epoch": 2.3731394648541686,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005401662469196409,
      "loss": 1.5343,
      "step": 27583
    },
    {
      "epoch": 2.3732255011614902,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005401392935177265,
      "loss": 0.8477,
      "step": 27584
    },
    {
      "epoch": 2.373311537468812,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005401123399984125,
      "loss": 0.8124,
      "step": 27585
    },
    {
      "epoch": 2.3733975737761335,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005400853863617778,
      "loss": 0.7287,
      "step": 27586
    },
    {
      "epoch": 2.373483610083455,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005400584326079012,
      "loss": 1.0445,
      "step": 27587
    },
    {
      "epoch": 2.3735696463907767,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005400314787368613,
      "loss": 0.6816,
      "step": 27588
    },
    {
      "epoch": 2.373655682698099,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000540004524748737,
      "loss": 1.4298,
      "step": 27589
    },
    {
      "epoch": 2.3737417190054204,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005399775706436075,
      "loss": 1.3432,
      "step": 27590
    },
    {
      "epoch": 2.373827755312742,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005399506164215513,
      "loss": 1.0797,
      "step": 27591
    },
    {
      "epoch": 2.3739137916200637,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005399236620826474,
      "loss": 1.7702,
      "step": 27592
    },
    {
      "epoch": 2.3739998279273853,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005398967076269744,
      "loss": 0.8436,
      "step": 27593
    },
    {
      "epoch": 2.374085864234707,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005398697530546114,
      "loss": 1.4014,
      "step": 27594
    },
    {
      "epoch": 2.374171900542029,
      "grad_norm": 1.375,
      "learning_rate": 0.0005398427983656371,
      "loss": 1.569,
      "step": 27595
    },
    {
      "epoch": 2.3742579368493506,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005398158435601303,
      "loss": 1.7147,
      "step": 27596
    },
    {
      "epoch": 2.3743439731566722,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005397888886381699,
      "loss": 1.3422,
      "step": 27597
    },
    {
      "epoch": 2.374430009463994,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005397619335998348,
      "loss": 2.0312,
      "step": 27598
    },
    {
      "epoch": 2.3745160457713155,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005397349784452038,
      "loss": 1.0786,
      "step": 27599
    },
    {
      "epoch": 2.374602082078637,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005397080231743556,
      "loss": 1.5111,
      "step": 27600
    },
    {
      "epoch": 2.3746881183859587,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005396810677873693,
      "loss": 0.8546,
      "step": 27601
    },
    {
      "epoch": 2.374774154693281,
      "grad_norm": 1.125,
      "learning_rate": 0.0005396541122843235,
      "loss": 1.805,
      "step": 27602
    },
    {
      "epoch": 2.3748601910006024,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005396271566652972,
      "loss": 1.5471,
      "step": 27603
    },
    {
      "epoch": 2.374946227307924,
      "grad_norm": 1.34375,
      "learning_rate": 0.000539600200930369,
      "loss": 1.3885,
      "step": 27604
    },
    {
      "epoch": 2.3750322636152457,
      "grad_norm": 1.0,
      "learning_rate": 0.0005395732450796182,
      "loss": 1.7304,
      "step": 27605
    },
    {
      "epoch": 2.3751182999225673,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005395462891131231,
      "loss": 1.2275,
      "step": 27606
    },
    {
      "epoch": 2.375204336229889,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005395193330309629,
      "loss": 1.5523,
      "step": 27607
    },
    {
      "epoch": 2.3752903725372105,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005394923768332162,
      "loss": 1.8118,
      "step": 27608
    },
    {
      "epoch": 2.3753764088445326,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005394654205199623,
      "loss": 1.1575,
      "step": 27609
    },
    {
      "epoch": 2.3754624451518542,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005394384640912795,
      "loss": 0.8269,
      "step": 27610
    },
    {
      "epoch": 2.375548481459176,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005394115075472469,
      "loss": 0.9261,
      "step": 27611
    },
    {
      "epoch": 2.3756345177664975,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005393845508879433,
      "loss": 1.0598,
      "step": 27612
    },
    {
      "epoch": 2.375720554073819,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005393575941134475,
      "loss": 1.0409,
      "step": 27613
    },
    {
      "epoch": 2.3758065903811407,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005393306372238386,
      "loss": 0.9021,
      "step": 27614
    },
    {
      "epoch": 2.3758926266884624,
      "grad_norm": 1.234375,
      "learning_rate": 0.000539303680219195,
      "loss": 1.2081,
      "step": 27615
    },
    {
      "epoch": 2.3759786629957844,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005392767230995958,
      "loss": 0.8182,
      "step": 27616
    },
    {
      "epoch": 2.376064699303106,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00053924976586512,
      "loss": 0.9607,
      "step": 27617
    },
    {
      "epoch": 2.3761507356104277,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005392228085158461,
      "loss": 1.6747,
      "step": 27618
    },
    {
      "epoch": 2.3762367719177493,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005391958510518531,
      "loss": 1.0289,
      "step": 27619
    },
    {
      "epoch": 2.376322808225071,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005391688934732198,
      "loss": 0.9938,
      "step": 27620
    },
    {
      "epoch": 2.3764088445323925,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005391419357800253,
      "loss": 0.7988,
      "step": 27621
    },
    {
      "epoch": 2.376494880839714,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005391149779723482,
      "loss": 1.3669,
      "step": 27622
    },
    {
      "epoch": 2.3765809171470362,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005390880200502674,
      "loss": 1.4871,
      "step": 27623
    },
    {
      "epoch": 2.376666953454358,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005390610620138618,
      "loss": 1.2307,
      "step": 27624
    },
    {
      "epoch": 2.3767529897616795,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005390341038632101,
      "loss": 0.7561,
      "step": 27625
    },
    {
      "epoch": 2.376839026069001,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005390071455983912,
      "loss": 1.1559,
      "step": 27626
    },
    {
      "epoch": 2.3769250623763227,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005389801872194839,
      "loss": 0.7337,
      "step": 27627
    },
    {
      "epoch": 2.3770110986836444,
      "grad_norm": 1.125,
      "learning_rate": 0.0005389532287265674,
      "loss": 1.26,
      "step": 27628
    },
    {
      "epoch": 2.377097134990966,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005389262701197202,
      "loss": 1.6881,
      "step": 27629
    },
    {
      "epoch": 2.377183171298288,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005388993113990211,
      "loss": 1.7018,
      "step": 27630
    },
    {
      "epoch": 2.3772692076056097,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005388723525645491,
      "loss": 1.4683,
      "step": 27631
    },
    {
      "epoch": 2.3773552439129313,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005388453936163832,
      "loss": 0.9622,
      "step": 27632
    },
    {
      "epoch": 2.377441280220253,
      "grad_norm": 1.125,
      "learning_rate": 0.000538818434554602,
      "loss": 1.4156,
      "step": 27633
    },
    {
      "epoch": 2.3775273165275745,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005387914753792843,
      "loss": 0.9258,
      "step": 27634
    },
    {
      "epoch": 2.377613352834896,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005387645160905093,
      "loss": 0.8147,
      "step": 27635
    },
    {
      "epoch": 2.377699389142218,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005387375566883554,
      "loss": 1.2255,
      "step": 27636
    },
    {
      "epoch": 2.37778542544954,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005387105971729017,
      "loss": 1.4949,
      "step": 27637
    },
    {
      "epoch": 2.3778714617568615,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005386836375442272,
      "loss": 0.846,
      "step": 27638
    },
    {
      "epoch": 2.377957498064183,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005386566778024104,
      "loss": 1.1425,
      "step": 27639
    },
    {
      "epoch": 2.3780435343715047,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005386297179475304,
      "loss": 1.1289,
      "step": 27640
    },
    {
      "epoch": 2.3781295706788264,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005386027579796662,
      "loss": 0.8054,
      "step": 27641
    },
    {
      "epoch": 2.378215606986148,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005385757978988961,
      "loss": 0.5162,
      "step": 27642
    },
    {
      "epoch": 2.3783016432934696,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005385488377052995,
      "loss": 1.5483,
      "step": 27643
    },
    {
      "epoch": 2.3783876796007917,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0005385218773989551,
      "loss": 1.7621,
      "step": 27644
    },
    {
      "epoch": 2.3784737159081133,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005384949169799415,
      "loss": 1.4746,
      "step": 27645
    },
    {
      "epoch": 2.378559752215435,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005384679564483379,
      "loss": 0.9619,
      "step": 27646
    },
    {
      "epoch": 2.3786457885227565,
      "grad_norm": 1.28125,
      "learning_rate": 0.000538440995804223,
      "loss": 1.3446,
      "step": 27647
    },
    {
      "epoch": 2.378731824830078,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005384140350476757,
      "loss": 0.7515,
      "step": 27648
    },
    {
      "epoch": 2.3788178611374,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005383870741787746,
      "loss": 0.7,
      "step": 27649
    },
    {
      "epoch": 2.378903897444722,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005383601131975991,
      "loss": 1.497,
      "step": 27650
    },
    {
      "epoch": 2.3789899337520435,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005383331521042275,
      "loss": 1.1356,
      "step": 27651
    },
    {
      "epoch": 2.379075970059365,
      "grad_norm": 2.0625,
      "learning_rate": 0.0005383061908987389,
      "loss": 0.9222,
      "step": 27652
    },
    {
      "epoch": 2.3791620063666867,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005382792295812122,
      "loss": 1.0093,
      "step": 27653
    },
    {
      "epoch": 2.3792480426740084,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005382522681517262,
      "loss": 0.779,
      "step": 27654
    },
    {
      "epoch": 2.37933407898133,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005382253066103597,
      "loss": 1.0813,
      "step": 27655
    },
    {
      "epoch": 2.379420115288652,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005381983449571916,
      "loss": 1.575,
      "step": 27656
    },
    {
      "epoch": 2.3795061515959737,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005381713831923007,
      "loss": 1.143,
      "step": 27657
    },
    {
      "epoch": 2.3795921879032953,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005381444213157661,
      "loss": 1.4309,
      "step": 27658
    },
    {
      "epoch": 2.379678224210617,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005381174593276666,
      "loss": 1.2372,
      "step": 27659
    },
    {
      "epoch": 2.3797642605179385,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005380904972280807,
      "loss": 1.183,
      "step": 27660
    },
    {
      "epoch": 2.37985029682526,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005380635350170875,
      "loss": 1.3302,
      "step": 27661
    },
    {
      "epoch": 2.379936333132582,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000538036572694766,
      "loss": 1.814,
      "step": 27662
    },
    {
      "epoch": 2.380022369439904,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005380096102611949,
      "loss": 0.7294,
      "step": 27663
    },
    {
      "epoch": 2.3801084057472255,
      "grad_norm": 1.390625,
      "learning_rate": 0.000537982647716453,
      "loss": 1.4648,
      "step": 27664
    },
    {
      "epoch": 2.380194442054547,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005379556850606194,
      "loss": 0.9918,
      "step": 27665
    },
    {
      "epoch": 2.3802804783618687,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005379287222937727,
      "loss": 0.7673,
      "step": 27666
    },
    {
      "epoch": 2.3803665146691904,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005379017594159917,
      "loss": 1.5575,
      "step": 27667
    },
    {
      "epoch": 2.380452550976512,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005378747964273557,
      "loss": 1.6724,
      "step": 27668
    },
    {
      "epoch": 2.3805385872838336,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005378478333279432,
      "loss": 1.104,
      "step": 27669
    },
    {
      "epoch": 2.3806246235911557,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005378208701178331,
      "loss": 1.7603,
      "step": 27670
    },
    {
      "epoch": 2.3807106598984773,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005377939067971043,
      "loss": 1.6867,
      "step": 27671
    },
    {
      "epoch": 2.380796696205799,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005377669433658357,
      "loss": 1.6434,
      "step": 27672
    },
    {
      "epoch": 2.3808827325131205,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005377399798241062,
      "loss": 1.4424,
      "step": 27673
    },
    {
      "epoch": 2.380968768820442,
      "grad_norm": 1.0,
      "learning_rate": 0.0005377130161719945,
      "loss": 1.4268,
      "step": 27674
    },
    {
      "epoch": 2.381054805127764,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005376860524095796,
      "loss": 1.8746,
      "step": 27675
    },
    {
      "epoch": 2.3811408414350854,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005376590885369403,
      "loss": 0.9859,
      "step": 27676
    },
    {
      "epoch": 2.3812268777424075,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0005376321245541554,
      "loss": 1.3041,
      "step": 27677
    },
    {
      "epoch": 2.381312914049729,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005376051604613041,
      "loss": 1.0418,
      "step": 27678
    },
    {
      "epoch": 2.3813989503570507,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005375781962584648,
      "loss": 1.2228,
      "step": 27679
    },
    {
      "epoch": 2.3814849866643724,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005375512319457168,
      "loss": 1.4516,
      "step": 27680
    },
    {
      "epoch": 2.381571022971694,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005375242675231386,
      "loss": 0.9789,
      "step": 27681
    },
    {
      "epoch": 2.3816570592790156,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005374973029908092,
      "loss": 0.7052,
      "step": 27682
    },
    {
      "epoch": 2.3817430955863372,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005374703383488076,
      "loss": 0.8843,
      "step": 27683
    },
    {
      "epoch": 2.3818291318936593,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005374433735972124,
      "loss": 1.0174,
      "step": 27684
    },
    {
      "epoch": 2.381915168200981,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005374164087361028,
      "loss": 0.8166,
      "step": 27685
    },
    {
      "epoch": 2.3820012045083025,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005373894437655572,
      "loss": 1.7047,
      "step": 27686
    },
    {
      "epoch": 2.382087240815624,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000537362478685655,
      "loss": 1.0028,
      "step": 27687
    },
    {
      "epoch": 2.382173277122946,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005373355134964747,
      "loss": 1.4218,
      "step": 27688
    },
    {
      "epoch": 2.3822593134302674,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005373085481980953,
      "loss": 1.1134,
      "step": 27689
    },
    {
      "epoch": 2.382345349737589,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005372815827905956,
      "loss": 2.1467,
      "step": 27690
    },
    {
      "epoch": 2.382431386044911,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005372546172740547,
      "loss": 1.3811,
      "step": 27691
    },
    {
      "epoch": 2.3825174223522327,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005372276516485512,
      "loss": 1.3522,
      "step": 27692
    },
    {
      "epoch": 2.3826034586595544,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005372006859141641,
      "loss": 1.0355,
      "step": 27693
    },
    {
      "epoch": 2.382689494966876,
      "grad_norm": 1.34375,
      "learning_rate": 0.000537173720070972,
      "loss": 0.9166,
      "step": 27694
    },
    {
      "epoch": 2.3827755312741976,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005371467541190543,
      "loss": 1.6764,
      "step": 27695
    },
    {
      "epoch": 2.3828615675815192,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005371197880584894,
      "loss": 1.287,
      "step": 27696
    },
    {
      "epoch": 2.382947603888841,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005370928218893564,
      "loss": 1.7321,
      "step": 27697
    },
    {
      "epoch": 2.383033640196163,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005370658556117339,
      "loss": 0.8387,
      "step": 27698
    },
    {
      "epoch": 2.3831196765034846,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005370388892257013,
      "loss": 1.5798,
      "step": 27699
    },
    {
      "epoch": 2.383205712810806,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005370119227313371,
      "loss": 1.0212,
      "step": 27700
    },
    {
      "epoch": 2.383291749118128,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00053698495612872,
      "loss": 1.2019,
      "step": 27701
    },
    {
      "epoch": 2.3833777854254494,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005369579894179293,
      "loss": 1.373,
      "step": 27702
    },
    {
      "epoch": 2.383463821732771,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005369310225990436,
      "loss": 1.1329,
      "step": 27703
    },
    {
      "epoch": 2.3835498580400927,
      "grad_norm": 1.0625,
      "learning_rate": 0.000536904055672142,
      "loss": 1.2407,
      "step": 27704
    },
    {
      "epoch": 2.3836358943474147,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005368770886373028,
      "loss": 1.8348,
      "step": 27705
    },
    {
      "epoch": 2.3837219306547364,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005368501214946057,
      "loss": 0.9389,
      "step": 27706
    },
    {
      "epoch": 2.383807966962058,
      "grad_norm": 2.0625,
      "learning_rate": 0.000536823154244129,
      "loss": 1.5484,
      "step": 27707
    },
    {
      "epoch": 2.3838940032693796,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005367961868859517,
      "loss": 1.8187,
      "step": 27708
    },
    {
      "epoch": 2.3839800395767012,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005367692194201528,
      "loss": 0.8196,
      "step": 27709
    },
    {
      "epoch": 2.3840660758840233,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005367422518468111,
      "loss": 0.5638,
      "step": 27710
    },
    {
      "epoch": 2.384152112191345,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005367152841660055,
      "loss": 0.9671,
      "step": 27711
    },
    {
      "epoch": 2.3842381484986666,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005366883163778146,
      "loss": 1.297,
      "step": 27712
    },
    {
      "epoch": 2.384324184805988,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005366613484823176,
      "loss": 0.7372,
      "step": 27713
    },
    {
      "epoch": 2.38441022111331,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005366343804795934,
      "loss": 1.734,
      "step": 27714
    },
    {
      "epoch": 2.3844962574206314,
      "grad_norm": 1.875,
      "learning_rate": 0.0005366074123697207,
      "loss": 1.4429,
      "step": 27715
    },
    {
      "epoch": 2.384582293727953,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005365804441527783,
      "loss": 1.8619,
      "step": 27716
    },
    {
      "epoch": 2.384668330035275,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005365534758288453,
      "loss": 0.7237,
      "step": 27717
    },
    {
      "epoch": 2.3847543663425967,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005365265073980007,
      "loss": 1.7979,
      "step": 27718
    },
    {
      "epoch": 2.3848404026499184,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000536499538860323,
      "loss": 1.4928,
      "step": 27719
    },
    {
      "epoch": 2.38492643895724,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005364725702158913,
      "loss": 1.6148,
      "step": 27720
    },
    {
      "epoch": 2.3850124752645616,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005364456014647844,
      "loss": 1.1476,
      "step": 27721
    },
    {
      "epoch": 2.3850985115718832,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005364186326070811,
      "loss": 1.811,
      "step": 27722
    },
    {
      "epoch": 2.385184547879205,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005363916636428605,
      "loss": 1.0171,
      "step": 27723
    },
    {
      "epoch": 2.385270584186527,
      "grad_norm": 2.015625,
      "learning_rate": 0.0005363646945722014,
      "loss": 1.0041,
      "step": 27724
    },
    {
      "epoch": 2.3853566204938486,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005363377253951825,
      "loss": 1.1148,
      "step": 27725
    },
    {
      "epoch": 2.38544265680117,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000536310756111883,
      "loss": 0.8454,
      "step": 27726
    },
    {
      "epoch": 2.385528693108492,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005362837867223814,
      "loss": 1.4054,
      "step": 27727
    },
    {
      "epoch": 2.3856147294158134,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005362568172267569,
      "loss": 0.6543,
      "step": 27728
    },
    {
      "epoch": 2.385700765723135,
      "grad_norm": 1.25,
      "learning_rate": 0.0005362298476250883,
      "loss": 1.3868,
      "step": 27729
    },
    {
      "epoch": 2.3857868020304567,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005362028779174544,
      "loss": 1.4695,
      "step": 27730
    },
    {
      "epoch": 2.3858728383377787,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005361759081039339,
      "loss": 1.2574,
      "step": 27731
    },
    {
      "epoch": 2.3859588746451004,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005361489381846062,
      "loss": 1.042,
      "step": 27732
    },
    {
      "epoch": 2.386044910952422,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005361219681595498,
      "loss": 1.1168,
      "step": 27733
    },
    {
      "epoch": 2.3861309472597436,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005360949980288437,
      "loss": 1.2141,
      "step": 27734
    },
    {
      "epoch": 2.3862169835670652,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005360680277925665,
      "loss": 1.3446,
      "step": 27735
    },
    {
      "epoch": 2.386303019874387,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005360410574507977,
      "loss": 0.7314,
      "step": 27736
    },
    {
      "epoch": 2.3863890561817085,
      "grad_norm": 1.90625,
      "learning_rate": 0.0005360140870036156,
      "loss": 1.5811,
      "step": 27737
    },
    {
      "epoch": 2.3864750924890306,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005359871164510994,
      "loss": 0.6031,
      "step": 27738
    },
    {
      "epoch": 2.386561128796352,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005359601457933279,
      "loss": 1.4258,
      "step": 27739
    },
    {
      "epoch": 2.386647165103674,
      "grad_norm": 1.453125,
      "learning_rate": 0.00053593317503038,
      "loss": 1.7012,
      "step": 27740
    },
    {
      "epoch": 2.3867332014109954,
      "grad_norm": 1.0,
      "learning_rate": 0.0005359062041623343,
      "loss": 1.3333,
      "step": 27741
    },
    {
      "epoch": 2.386819237718317,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005358792331892701,
      "loss": 1.3722,
      "step": 27742
    },
    {
      "epoch": 2.3869052740256387,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0005358522621112662,
      "loss": 0.6941,
      "step": 27743
    },
    {
      "epoch": 2.3869913103329603,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0005358252909284014,
      "loss": 1.5374,
      "step": 27744
    },
    {
      "epoch": 2.3870773466402824,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005357983196407545,
      "loss": 1.4011,
      "step": 27745
    },
    {
      "epoch": 2.387163382947604,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005357713482484045,
      "loss": 1.7502,
      "step": 27746
    },
    {
      "epoch": 2.3872494192549256,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005357443767514303,
      "loss": 1.3982,
      "step": 27747
    },
    {
      "epoch": 2.3873354555622472,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005357174051499107,
      "loss": 1.6658,
      "step": 27748
    },
    {
      "epoch": 2.387421491869569,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005356904334439245,
      "loss": 0.5992,
      "step": 27749
    },
    {
      "epoch": 2.3875075281768905,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005356634616335509,
      "loss": 1.1946,
      "step": 27750
    },
    {
      "epoch": 2.387593564484212,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005356364897188686,
      "loss": 0.7772,
      "step": 27751
    },
    {
      "epoch": 2.387679600791534,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0005356095176999565,
      "loss": 1.0376,
      "step": 27752
    },
    {
      "epoch": 2.387765637098856,
      "grad_norm": 1.125,
      "learning_rate": 0.0005355825455768935,
      "loss": 1.2544,
      "step": 27753
    },
    {
      "epoch": 2.3878516734061774,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005355555733497584,
      "loss": 1.5324,
      "step": 27754
    },
    {
      "epoch": 2.387937709713499,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005355286010186302,
      "loss": 0.4716,
      "step": 27755
    },
    {
      "epoch": 2.3880237460208207,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005355016285835877,
      "loss": 1.2245,
      "step": 27756
    },
    {
      "epoch": 2.3881097823281423,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00053547465604471,
      "loss": 0.7791,
      "step": 27757
    },
    {
      "epoch": 2.388195818635464,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005354476834020757,
      "loss": 1.0552,
      "step": 27758
    },
    {
      "epoch": 2.388281854942786,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005354207106557639,
      "loss": 1.0953,
      "step": 27759
    },
    {
      "epoch": 2.3883678912501076,
      "grad_norm": 1.125,
      "learning_rate": 0.0005353937378058533,
      "loss": 1.2591,
      "step": 27760
    },
    {
      "epoch": 2.3884539275574292,
      "grad_norm": 1.125,
      "learning_rate": 0.0005353667648524229,
      "loss": 0.7165,
      "step": 27761
    },
    {
      "epoch": 2.388539963864751,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005353397917955517,
      "loss": 0.8375,
      "step": 27762
    },
    {
      "epoch": 2.3886260001720725,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005353128186353185,
      "loss": 1.1818,
      "step": 27763
    },
    {
      "epoch": 2.388712036479394,
      "grad_norm": 1.453125,
      "learning_rate": 0.000535285845371802,
      "loss": 2.0705,
      "step": 27764
    },
    {
      "epoch": 2.388798072786716,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005352588720050815,
      "loss": 0.8269,
      "step": 27765
    },
    {
      "epoch": 2.388884109094038,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005352318985352355,
      "loss": 1.0536,
      "step": 27766
    },
    {
      "epoch": 2.3889701454013594,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005352049249623432,
      "loss": 0.8937,
      "step": 27767
    },
    {
      "epoch": 2.389056181708681,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000535177951286483,
      "loss": 0.5657,
      "step": 27768
    },
    {
      "epoch": 2.3891422180160027,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005351509775077343,
      "loss": 1.532,
      "step": 27769
    },
    {
      "epoch": 2.3892282543233243,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000535124003626176,
      "loss": 1.3172,
      "step": 27770
    },
    {
      "epoch": 2.3893142906306464,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005350970296418867,
      "loss": 0.9405,
      "step": 27771
    },
    {
      "epoch": 2.389400326937968,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005350700555549454,
      "loss": 0.8135,
      "step": 27772
    },
    {
      "epoch": 2.3894863632452896,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000535043081365431,
      "loss": 0.8557,
      "step": 27773
    },
    {
      "epoch": 2.3895723995526112,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005350161070734225,
      "loss": 0.8355,
      "step": 27774
    },
    {
      "epoch": 2.389658435859933,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005349891326789986,
      "loss": 0.9164,
      "step": 27775
    },
    {
      "epoch": 2.3897444721672545,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005349621581822383,
      "loss": 1.1431,
      "step": 27776
    },
    {
      "epoch": 2.389830508474576,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005349351835832205,
      "loss": 1.0279,
      "step": 27777
    },
    {
      "epoch": 2.389916544781898,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000534908208882024,
      "loss": 1.0333,
      "step": 27778
    },
    {
      "epoch": 2.39000258108922,
      "grad_norm": 1.59375,
      "learning_rate": 0.0005348812340787279,
      "loss": 1.1738,
      "step": 27779
    },
    {
      "epoch": 2.3900886173965414,
      "grad_norm": 1.234375,
      "learning_rate": 0.000534854259173411,
      "loss": 1.157,
      "step": 27780
    },
    {
      "epoch": 2.390174653703863,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005348272841661521,
      "loss": 1.0237,
      "step": 27781
    },
    {
      "epoch": 2.3902606900111847,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0005348003090570303,
      "loss": 0.9915,
      "step": 27782
    },
    {
      "epoch": 2.3903467263185063,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005347733338461241,
      "loss": 1.54,
      "step": 27783
    },
    {
      "epoch": 2.390432762625828,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005347463585335128,
      "loss": 1.2152,
      "step": 27784
    },
    {
      "epoch": 2.39051879893315,
      "grad_norm": 1.25,
      "learning_rate": 0.0005347193831192753,
      "loss": 1.9697,
      "step": 27785
    },
    {
      "epoch": 2.3906048352404716,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00053469240760349,
      "loss": 1.3772,
      "step": 27786
    },
    {
      "epoch": 2.3906908715477933,
      "grad_norm": 1.25,
      "learning_rate": 0.0005346654319862364,
      "loss": 0.7882,
      "step": 27787
    },
    {
      "epoch": 2.390776907855115,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005346384562675931,
      "loss": 0.5989,
      "step": 27788
    },
    {
      "epoch": 2.3908629441624365,
      "grad_norm": 3.71875,
      "learning_rate": 0.0005346114804476392,
      "loss": 0.8126,
      "step": 27789
    },
    {
      "epoch": 2.390948980469758,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005345845045264533,
      "loss": 1.3597,
      "step": 27790
    },
    {
      "epoch": 2.3910350167770797,
      "grad_norm": 1.125,
      "learning_rate": 0.0005345575285041145,
      "loss": 1.6436,
      "step": 27791
    },
    {
      "epoch": 2.391121053084402,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005345305523807016,
      "loss": 1.6127,
      "step": 27792
    },
    {
      "epoch": 2.3912070893917234,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005345035761562937,
      "loss": 0.8575,
      "step": 27793
    },
    {
      "epoch": 2.391293125699045,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005344765998309693,
      "loss": 1.3861,
      "step": 27794
    },
    {
      "epoch": 2.3913791620063667,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005344496234048076,
      "loss": 1.4551,
      "step": 27795
    },
    {
      "epoch": 2.3914651983136883,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005344226468778876,
      "loss": 1.1281,
      "step": 27796
    },
    {
      "epoch": 2.39155123462101,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005343956702502881,
      "loss": 0.9625,
      "step": 27797
    },
    {
      "epoch": 2.3916372709283316,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005343686935220878,
      "loss": 1.5534,
      "step": 27798
    },
    {
      "epoch": 2.3917233072356536,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005343417166933657,
      "loss": 0.9402,
      "step": 27799
    },
    {
      "epoch": 2.3918093435429753,
      "grad_norm": 1.171875,
      "learning_rate": 0.000534314739764201,
      "loss": 1.4794,
      "step": 27800
    },
    {
      "epoch": 2.391895379850297,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005342877627346721,
      "loss": 1.6459,
      "step": 27801
    },
    {
      "epoch": 2.3919814161576185,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005342607856048582,
      "loss": 1.0607,
      "step": 27802
    },
    {
      "epoch": 2.39206745246494,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005342338083748384,
      "loss": 1.293,
      "step": 27803
    },
    {
      "epoch": 2.3921534887722617,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005342068310446912,
      "loss": 0.8205,
      "step": 27804
    },
    {
      "epoch": 2.3922395250795834,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005341798536144956,
      "loss": 0.9376,
      "step": 27805
    },
    {
      "epoch": 2.3923255613869054,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005341528760843306,
      "loss": 0.9106,
      "step": 27806
    },
    {
      "epoch": 2.392411597694227,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005341258984542752,
      "loss": 1.4041,
      "step": 27807
    },
    {
      "epoch": 2.3924976340015487,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005340989207244082,
      "loss": 1.5876,
      "step": 27808
    },
    {
      "epoch": 2.3925836703088703,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005340719428948085,
      "loss": 1.6462,
      "step": 27809
    },
    {
      "epoch": 2.392669706616192,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005340449649655547,
      "loss": 1.354,
      "step": 27810
    },
    {
      "epoch": 2.3927557429235136,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005340179869367263,
      "loss": 1.445,
      "step": 27811
    },
    {
      "epoch": 2.392841779230835,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005339910088084018,
      "loss": 1.9083,
      "step": 27812
    },
    {
      "epoch": 2.3929278155381573,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005339640305806601,
      "loss": 1.1777,
      "step": 27813
    },
    {
      "epoch": 2.393013851845479,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005339370522535804,
      "loss": 1.7327,
      "step": 27814
    },
    {
      "epoch": 2.3930998881528005,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005339100738272415,
      "loss": 1.1724,
      "step": 27815
    },
    {
      "epoch": 2.393185924460122,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005338830953017221,
      "loss": 1.0771,
      "step": 27816
    },
    {
      "epoch": 2.3932719607674438,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005338561166771011,
      "loss": 1.2191,
      "step": 27817
    },
    {
      "epoch": 2.3933579970747654,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005338291379534577,
      "loss": 0.7976,
      "step": 27818
    },
    {
      "epoch": 2.393444033382087,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005338021591308706,
      "loss": 0.5854,
      "step": 27819
    },
    {
      "epoch": 2.393530069689409,
      "grad_norm": 1.25,
      "learning_rate": 0.0005337751802094187,
      "loss": 1.6166,
      "step": 27820
    },
    {
      "epoch": 2.3936161059967307,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005337482011891811,
      "loss": 0.558,
      "step": 27821
    },
    {
      "epoch": 2.3937021423040523,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005337212220702366,
      "loss": 0.5554,
      "step": 27822
    },
    {
      "epoch": 2.393788178611374,
      "grad_norm": 2.09375,
      "learning_rate": 0.0005336942428526639,
      "loss": 1.7762,
      "step": 27823
    },
    {
      "epoch": 2.3938742149186956,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005336672635365422,
      "loss": 1.5985,
      "step": 27824
    },
    {
      "epoch": 2.393960251226017,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005336402841219502,
      "loss": 1.2275,
      "step": 27825
    },
    {
      "epoch": 2.3940462875333393,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000533613304608967,
      "loss": 1.5295,
      "step": 27826
    },
    {
      "epoch": 2.394132323840661,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005335863249976715,
      "loss": 1.4887,
      "step": 27827
    },
    {
      "epoch": 2.3942183601479825,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005335593452881424,
      "loss": 1.1015,
      "step": 27828
    },
    {
      "epoch": 2.394304396455304,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005335323654804589,
      "loss": 0.7583,
      "step": 27829
    },
    {
      "epoch": 2.3943904327626258,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0005335053855746996,
      "loss": 1.168,
      "step": 27830
    },
    {
      "epoch": 2.3944764690699474,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005334784055709438,
      "loss": 1.033,
      "step": 27831
    },
    {
      "epoch": 2.3945625053772694,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005334514254692698,
      "loss": 1.594,
      "step": 27832
    },
    {
      "epoch": 2.394648541684591,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005334244452697572,
      "loss": 0.7317,
      "step": 27833
    },
    {
      "epoch": 2.3947345779919127,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005333974649724845,
      "loss": 1.05,
      "step": 27834
    },
    {
      "epoch": 2.3948206142992343,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005333704845775306,
      "loss": 1.3508,
      "step": 27835
    },
    {
      "epoch": 2.394906650606556,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005333435040849745,
      "loss": 1.6448,
      "step": 27836
    },
    {
      "epoch": 2.3949926869138776,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005333165234948954,
      "loss": 0.8552,
      "step": 27837
    },
    {
      "epoch": 2.395078723221199,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005332895428073718,
      "loss": 2.002,
      "step": 27838
    },
    {
      "epoch": 2.3951647595285213,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005332625620224826,
      "loss": 1.0768,
      "step": 27839
    },
    {
      "epoch": 2.395250795835843,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005332355811403072,
      "loss": 1.7033,
      "step": 27840
    },
    {
      "epoch": 2.3953368321431645,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000533208600160924,
      "loss": 1.1481,
      "step": 27841
    },
    {
      "epoch": 2.395422868450486,
      "grad_norm": 1.375,
      "learning_rate": 0.0005331816190844121,
      "loss": 1.3048,
      "step": 27842
    },
    {
      "epoch": 2.3955089047578078,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005331546379108504,
      "loss": 1.153,
      "step": 27843
    },
    {
      "epoch": 2.3955949410651294,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005331276566403179,
      "loss": 1.5456,
      "step": 27844
    },
    {
      "epoch": 2.395680977372451,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0005331006752728934,
      "loss": 1.5342,
      "step": 27845
    },
    {
      "epoch": 2.395767013679773,
      "grad_norm": 0.828125,
      "learning_rate": 0.000533073693808656,
      "loss": 0.9267,
      "step": 27846
    },
    {
      "epoch": 2.3958530499870947,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005330467122476842,
      "loss": 0.8138,
      "step": 27847
    },
    {
      "epoch": 2.3959390862944163,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005330197305900573,
      "loss": 2.0422,
      "step": 27848
    },
    {
      "epoch": 2.396025122601738,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005329927488358544,
      "loss": 0.6493,
      "step": 27849
    },
    {
      "epoch": 2.3961111589090596,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005329657669851538,
      "loss": 1.6828,
      "step": 27850
    },
    {
      "epoch": 2.396197195216381,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005329387850380348,
      "loss": 1.3087,
      "step": 27851
    },
    {
      "epoch": 2.396283231523703,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005329118029945763,
      "loss": 1.7975,
      "step": 27852
    },
    {
      "epoch": 2.396369267831025,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005328848208548572,
      "loss": 1.5191,
      "step": 27853
    },
    {
      "epoch": 2.3964553041383465,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005328578386189564,
      "loss": 1.359,
      "step": 27854
    },
    {
      "epoch": 2.396541340445668,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005328308562869527,
      "loss": 1.4466,
      "step": 27855
    },
    {
      "epoch": 2.3966273767529898,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005328038738589253,
      "loss": 1.1835,
      "step": 27856
    },
    {
      "epoch": 2.3967134130603114,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005327768913349529,
      "loss": 1.6135,
      "step": 27857
    },
    {
      "epoch": 2.396799449367633,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005327499087151143,
      "loss": 0.8044,
      "step": 27858
    },
    {
      "epoch": 2.3968854856749546,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005327229259994888,
      "loss": 1.8864,
      "step": 27859
    },
    {
      "epoch": 2.3969715219822767,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000532695943188155,
      "loss": 1.957,
      "step": 27860
    },
    {
      "epoch": 2.3970575582895983,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005326689602811919,
      "loss": 0.5258,
      "step": 27861
    },
    {
      "epoch": 2.39714359459692,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005326419772786785,
      "loss": 1.4122,
      "step": 27862
    },
    {
      "epoch": 2.3972296309042416,
      "grad_norm": 1.125,
      "learning_rate": 0.0005326149941806935,
      "loss": 1.0723,
      "step": 27863
    },
    {
      "epoch": 2.397315667211563,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005325880109873164,
      "loss": 1.6823,
      "step": 27864
    },
    {
      "epoch": 2.397401703518885,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005325610276986254,
      "loss": 1.237,
      "step": 27865
    },
    {
      "epoch": 2.3974877398262064,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005325340443146997,
      "loss": 0.7972,
      "step": 27866
    },
    {
      "epoch": 2.3975737761335285,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005325070608356184,
      "loss": 0.9297,
      "step": 27867
    },
    {
      "epoch": 2.39765981244085,
      "grad_norm": 0.75,
      "learning_rate": 0.0005324800772614602,
      "loss": 0.4075,
      "step": 27868
    },
    {
      "epoch": 2.3977458487481718,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005324530935923039,
      "loss": 1.4703,
      "step": 27869
    },
    {
      "epoch": 2.3978318850554934,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005324261098282289,
      "loss": 0.6855,
      "step": 27870
    },
    {
      "epoch": 2.397917921362815,
      "grad_norm": 10.4375,
      "learning_rate": 0.0005323991259693137,
      "loss": 1.4348,
      "step": 27871
    },
    {
      "epoch": 2.3980039576701366,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005323721420156376,
      "loss": 2.0591,
      "step": 27872
    },
    {
      "epoch": 2.3980899939774583,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000532345157967279,
      "loss": 1.3867,
      "step": 27873
    },
    {
      "epoch": 2.3981760302847803,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005323181738243172,
      "loss": 0.9154,
      "step": 27874
    },
    {
      "epoch": 2.398262066592102,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000532291189586831,
      "loss": 1.2954,
      "step": 27875
    },
    {
      "epoch": 2.3983481028994236,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005322642052548992,
      "loss": 1.4472,
      "step": 27876
    },
    {
      "epoch": 2.398434139206745,
      "grad_norm": 1.0,
      "learning_rate": 0.000532237220828601,
      "loss": 1.4316,
      "step": 27877
    },
    {
      "epoch": 2.398520175514067,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005322102363080153,
      "loss": 0.9242,
      "step": 27878
    },
    {
      "epoch": 2.3986062118213884,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005321832516932208,
      "loss": 0.6609,
      "step": 27879
    },
    {
      "epoch": 2.39869224812871,
      "grad_norm": 1.5,
      "learning_rate": 0.0005321562669842966,
      "loss": 1.6934,
      "step": 27880
    },
    {
      "epoch": 2.398778284436032,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005321292821813215,
      "loss": 1.882,
      "step": 27881
    },
    {
      "epoch": 2.3988643207433538,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005321022972843746,
      "loss": 0.903,
      "step": 27882
    },
    {
      "epoch": 2.3989503570506754,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005320753122935348,
      "loss": 0.8987,
      "step": 27883
    },
    {
      "epoch": 2.399036393357997,
      "grad_norm": 1.125,
      "learning_rate": 0.0005320483272088808,
      "loss": 0.7906,
      "step": 27884
    },
    {
      "epoch": 2.3991224296653186,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005320213420304916,
      "loss": 1.1114,
      "step": 27885
    },
    {
      "epoch": 2.3992084659726407,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005319943567584464,
      "loss": 1.0151,
      "step": 27886
    },
    {
      "epoch": 2.3992945022799623,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005319673713928239,
      "loss": 1.1339,
      "step": 27887
    },
    {
      "epoch": 2.399380538587284,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000531940385933703,
      "loss": 1.0864,
      "step": 27888
    },
    {
      "epoch": 2.3994665748946056,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005319134003811626,
      "loss": 0.8428,
      "step": 27889
    },
    {
      "epoch": 2.399552611201927,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005318864147352819,
      "loss": 1.3716,
      "step": 27890
    },
    {
      "epoch": 2.399638647509249,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005318594289961395,
      "loss": 0.8652,
      "step": 27891
    },
    {
      "epoch": 2.3997246838165704,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005318324431638146,
      "loss": 1.1004,
      "step": 27892
    },
    {
      "epoch": 2.3998107201238925,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005318054572383858,
      "loss": 1.3754,
      "step": 27893
    },
    {
      "epoch": 2.399896756431214,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005317784712199324,
      "loss": 0.8368,
      "step": 27894
    },
    {
      "epoch": 2.3999827927385358,
      "grad_norm": 0.921875,
      "learning_rate": 0.000531751485108533,
      "loss": 1.5809,
      "step": 27895
    },
    {
      "epoch": 2.4000688290458574,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005317244989042668,
      "loss": 1.2618,
      "step": 27896
    },
    {
      "epoch": 2.400154865353179,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005316975126072125,
      "loss": 0.9955,
      "step": 27897
    },
    {
      "epoch": 2.4002409016605006,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005316705262174492,
      "loss": 1.7098,
      "step": 27898
    },
    {
      "epoch": 2.4003269379678223,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005316435397350556,
      "loss": 1.0521,
      "step": 27899
    },
    {
      "epoch": 2.4004129742751443,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005316165531601112,
      "loss": 0.7829,
      "step": 27900
    },
    {
      "epoch": 2.400499010582466,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005315895664926942,
      "loss": 1.0529,
      "step": 27901
    },
    {
      "epoch": 2.4005850468897876,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005315625797328841,
      "loss": 1.7217,
      "step": 27902
    },
    {
      "epoch": 2.400671083197109,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005315355928807593,
      "loss": 1.0967,
      "step": 27903
    },
    {
      "epoch": 2.400757119504431,
      "grad_norm": 1.125,
      "learning_rate": 0.0005315086059363992,
      "loss": 1.6376,
      "step": 27904
    },
    {
      "epoch": 2.4008431558117524,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005314816188998827,
      "loss": 1.3621,
      "step": 27905
    },
    {
      "epoch": 2.400929192119074,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005314546317712884,
      "loss": 1.5219,
      "step": 27906
    },
    {
      "epoch": 2.401015228426396,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005314276445506954,
      "loss": 1.4969,
      "step": 27907
    },
    {
      "epoch": 2.4011012647337178,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005314006572381826,
      "loss": 0.7323,
      "step": 27908
    },
    {
      "epoch": 2.4011873010410394,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005313736698338292,
      "loss": 1.5879,
      "step": 27909
    },
    {
      "epoch": 2.401273337348361,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0005313466823377138,
      "loss": 1.269,
      "step": 27910
    },
    {
      "epoch": 2.4013593736556826,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005313196947499155,
      "loss": 1.1114,
      "step": 27911
    },
    {
      "epoch": 2.4014454099630043,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005312927070705132,
      "loss": 0.8296,
      "step": 27912
    },
    {
      "epoch": 2.401531446270326,
      "grad_norm": 1.625,
      "learning_rate": 0.0005312657192995857,
      "loss": 1.4554,
      "step": 27913
    },
    {
      "epoch": 2.401617482577648,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000531238731437212,
      "loss": 1.1333,
      "step": 27914
    },
    {
      "epoch": 2.4017035188849696,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005312117434834713,
      "loss": 1.3249,
      "step": 27915
    },
    {
      "epoch": 2.401789555192291,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005311847554384421,
      "loss": 1.7606,
      "step": 27916
    },
    {
      "epoch": 2.401875591499613,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005311577673022036,
      "loss": 1.3502,
      "step": 27917
    },
    {
      "epoch": 2.4019616278069345,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005311307790748348,
      "loss": 1.242,
      "step": 27918
    },
    {
      "epoch": 2.402047664114256,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005311037907564146,
      "loss": 1.1531,
      "step": 27919
    },
    {
      "epoch": 2.4021337004215777,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005310768023470217,
      "loss": 1.0306,
      "step": 27920
    },
    {
      "epoch": 2.4022197367288998,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005310498138467354,
      "loss": 0.7853,
      "step": 27921
    },
    {
      "epoch": 2.4023057730362214,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000531022825255634,
      "loss": 1.063,
      "step": 27922
    },
    {
      "epoch": 2.402391809343543,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0005309958365737972,
      "loss": 0.6502,
      "step": 27923
    },
    {
      "epoch": 2.4024778456508646,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005309688478013037,
      "loss": 2.0746,
      "step": 27924
    },
    {
      "epoch": 2.4025638819581863,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005309418589382321,
      "loss": 1.3779,
      "step": 27925
    },
    {
      "epoch": 2.402649918265508,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005309148699846616,
      "loss": 2.0195,
      "step": 27926
    },
    {
      "epoch": 2.4027359545728295,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005308878809406713,
      "loss": 1.5073,
      "step": 27927
    },
    {
      "epoch": 2.4028219908801516,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005308608918063399,
      "loss": 0.98,
      "step": 27928
    },
    {
      "epoch": 2.402908027187473,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005308339025817463,
      "loss": 0.8004,
      "step": 27929
    },
    {
      "epoch": 2.402994063494795,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005308069132669696,
      "loss": 1.9299,
      "step": 27930
    },
    {
      "epoch": 2.4030800998021165,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005307799238620887,
      "loss": 1.2956,
      "step": 27931
    },
    {
      "epoch": 2.403166136109438,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005307529343671824,
      "loss": 0.8154,
      "step": 27932
    },
    {
      "epoch": 2.4032521724167597,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005307259447823299,
      "loss": 1.017,
      "step": 27933
    },
    {
      "epoch": 2.4033382087240813,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005306989551076097,
      "loss": 1.296,
      "step": 27934
    },
    {
      "epoch": 2.4034242450314034,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005306719653431013,
      "loss": 1.309,
      "step": 27935
    },
    {
      "epoch": 2.403510281338725,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005306449754888834,
      "loss": 1.2629,
      "step": 27936
    },
    {
      "epoch": 2.4035963176460466,
      "grad_norm": 1.46875,
      "learning_rate": 0.0005306179855450346,
      "loss": 1.0374,
      "step": 27937
    },
    {
      "epoch": 2.4036823539533683,
      "grad_norm": 1.484375,
      "learning_rate": 0.0005305909955116343,
      "loss": 1.7887,
      "step": 27938
    },
    {
      "epoch": 2.40376839026069,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005305640053887615,
      "loss": 2.0594,
      "step": 27939
    },
    {
      "epoch": 2.4038544265680115,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005305370151764947,
      "loss": 1.3455,
      "step": 27940
    },
    {
      "epoch": 2.4039404628753336,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005305100248749131,
      "loss": 1.5586,
      "step": 27941
    },
    {
      "epoch": 2.404026499182655,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005304830344840956,
      "loss": 1.7321,
      "step": 27942
    },
    {
      "epoch": 2.404112535489977,
      "grad_norm": 1.25,
      "learning_rate": 0.0005304560440041213,
      "loss": 1.573,
      "step": 27943
    },
    {
      "epoch": 2.4041985717972985,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005304290534350689,
      "loss": 0.7312,
      "step": 27944
    },
    {
      "epoch": 2.40428460810462,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0005304020627770174,
      "loss": 1.5545,
      "step": 27945
    },
    {
      "epoch": 2.4043706444119417,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005303750720300458,
      "loss": 1.2572,
      "step": 27946
    },
    {
      "epoch": 2.4044566807192638,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005303480811942329,
      "loss": 1.6949,
      "step": 27947
    },
    {
      "epoch": 2.4045427170265854,
      "grad_norm": 1.03125,
      "learning_rate": 0.000530321090269658,
      "loss": 0.6642,
      "step": 27948
    },
    {
      "epoch": 2.404628753333907,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005302940992563997,
      "loss": 1.3169,
      "step": 27949
    },
    {
      "epoch": 2.4047147896412286,
      "grad_norm": 0.96875,
      "learning_rate": 0.000530267108154537,
      "loss": 0.8552,
      "step": 27950
    },
    {
      "epoch": 2.4048008259485503,
      "grad_norm": 1.5625,
      "learning_rate": 0.0005302401169641489,
      "loss": 2.4937,
      "step": 27951
    },
    {
      "epoch": 2.404886862255872,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005302131256853145,
      "loss": 1.3377,
      "step": 27952
    },
    {
      "epoch": 2.4049728985631935,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005301861343181124,
      "loss": 0.6033,
      "step": 27953
    },
    {
      "epoch": 2.4050589348705156,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005301591428626218,
      "loss": 1.425,
      "step": 27954
    },
    {
      "epoch": 2.405144971177837,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005301321513189215,
      "loss": 0.9214,
      "step": 27955
    },
    {
      "epoch": 2.405231007485159,
      "grad_norm": 1.25,
      "learning_rate": 0.0005301051596870906,
      "loss": 1.1031,
      "step": 27956
    },
    {
      "epoch": 2.4053170437924805,
      "grad_norm": 1.0625,
      "learning_rate": 0.000530078167967208,
      "loss": 0.973,
      "step": 27957
    },
    {
      "epoch": 2.405403080099802,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005300511761593526,
      "loss": 1.1673,
      "step": 27958
    },
    {
      "epoch": 2.4054891164071237,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005300241842636032,
      "loss": 1.6252,
      "step": 27959
    },
    {
      "epoch": 2.4055751527144453,
      "grad_norm": 1.171875,
      "learning_rate": 0.000529997192280039,
      "loss": 1.5918,
      "step": 27960
    },
    {
      "epoch": 2.4056611890217674,
      "grad_norm": 0.90625,
      "learning_rate": 0.000529970200208739,
      "loss": 0.5593,
      "step": 27961
    },
    {
      "epoch": 2.405747225329089,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005299432080497818,
      "loss": 0.9634,
      "step": 27962
    },
    {
      "epoch": 2.4058332616364106,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005299162158032465,
      "loss": 1.2881,
      "step": 27963
    },
    {
      "epoch": 2.4059192979437323,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005298892234692123,
      "loss": 1.4329,
      "step": 27964
    },
    {
      "epoch": 2.406005334251054,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005298622310477577,
      "loss": 0.8897,
      "step": 27965
    },
    {
      "epoch": 2.4060913705583755,
      "grad_norm": 1.6328125,
      "learning_rate": 0.000529835238538962,
      "loss": 1.3368,
      "step": 27966
    },
    {
      "epoch": 2.406177406865697,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005298082459429041,
      "loss": 1.2885,
      "step": 27967
    },
    {
      "epoch": 2.406263443173019,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005297812532596629,
      "loss": 0.7987,
      "step": 27968
    },
    {
      "epoch": 2.406349479480341,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005297542604893172,
      "loss": 1.1522,
      "step": 27969
    },
    {
      "epoch": 2.4064355157876625,
      "grad_norm": 0.859375,
      "learning_rate": 0.000529727267631946,
      "loss": 0.6555,
      "step": 27970
    },
    {
      "epoch": 2.406521552094984,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005297002746876284,
      "loss": 1.2232,
      "step": 27971
    },
    {
      "epoch": 2.4066075884023057,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005296732816564433,
      "loss": 0.8565,
      "step": 27972
    },
    {
      "epoch": 2.4066936247096273,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005296462885384696,
      "loss": 1.0136,
      "step": 27973
    },
    {
      "epoch": 2.406779661016949,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005296192953337862,
      "loss": 1.2495,
      "step": 27974
    },
    {
      "epoch": 2.406865697324271,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005295923020424724,
      "loss": 1.0419,
      "step": 27975
    },
    {
      "epoch": 2.4069517336315926,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005295653086646066,
      "loss": 1.2169,
      "step": 27976
    },
    {
      "epoch": 2.4070377699389143,
      "grad_norm": 1.96875,
      "learning_rate": 0.0005295383152002682,
      "loss": 1.328,
      "step": 27977
    },
    {
      "epoch": 2.407123806246236,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005295113216495358,
      "loss": 1.0533,
      "step": 27978
    },
    {
      "epoch": 2.4072098425535575,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005294843280124887,
      "loss": 1.5702,
      "step": 27979
    },
    {
      "epoch": 2.407295878860879,
      "grad_norm": 1.125,
      "learning_rate": 0.0005294573342892055,
      "loss": 1.2481,
      "step": 27980
    },
    {
      "epoch": 2.4073819151682008,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005294303404797654,
      "loss": 1.0899,
      "step": 27981
    },
    {
      "epoch": 2.407467951475523,
      "grad_norm": 1.0,
      "learning_rate": 0.0005294033465842473,
      "loss": 1.1437,
      "step": 27982
    },
    {
      "epoch": 2.4075539877828445,
      "grad_norm": 1.5,
      "learning_rate": 0.0005293763526027302,
      "loss": 1.2737,
      "step": 27983
    },
    {
      "epoch": 2.407640024090166,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005293493585352928,
      "loss": 2.0117,
      "step": 27984
    },
    {
      "epoch": 2.4077260603974877,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005293223643820143,
      "loss": 1.5124,
      "step": 27985
    },
    {
      "epoch": 2.4078120967048093,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005292953701429737,
      "loss": 1.2429,
      "step": 27986
    },
    {
      "epoch": 2.407898133012131,
      "grad_norm": 1.375,
      "learning_rate": 0.0005292683758182498,
      "loss": 1.1158,
      "step": 27987
    },
    {
      "epoch": 2.4079841693194526,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005292413814079215,
      "loss": 0.6054,
      "step": 27988
    },
    {
      "epoch": 2.4080702056267747,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005292143869120679,
      "loss": 1.2924,
      "step": 27989
    },
    {
      "epoch": 2.4081562419340963,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005291873923307679,
      "loss": 1.2467,
      "step": 27990
    },
    {
      "epoch": 2.408242278241418,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005291603976641004,
      "loss": 0.9184,
      "step": 27991
    },
    {
      "epoch": 2.4083283145487395,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005291334029121445,
      "loss": 1.1081,
      "step": 27992
    },
    {
      "epoch": 2.408414350856061,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000529106408074979,
      "loss": 1.7556,
      "step": 27993
    },
    {
      "epoch": 2.4085003871633828,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000529079413152683,
      "loss": 1.3045,
      "step": 27994
    },
    {
      "epoch": 2.4085864234707044,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005290524181453353,
      "loss": 1.1788,
      "step": 27995
    },
    {
      "epoch": 2.4086724597780265,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005290254230530148,
      "loss": 1.1221,
      "step": 27996
    },
    {
      "epoch": 2.408758496085348,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005289984278758009,
      "loss": 1.6305,
      "step": 27997
    },
    {
      "epoch": 2.4088445323926697,
      "grad_norm": 0.96875,
      "learning_rate": 0.000528971432613772,
      "loss": 0.9686,
      "step": 27998
    },
    {
      "epoch": 2.4089305686999913,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005289444372670074,
      "loss": 1.0568,
      "step": 27999
    },
    {
      "epoch": 2.409016605007313,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005289174418355859,
      "loss": 0.5935,
      "step": 28000
    },
    {
      "epoch": 2.4091026413146346,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005288904463195865,
      "loss": 0.6313,
      "step": 28001
    },
    {
      "epoch": 2.4091886776219567,
      "grad_norm": 1.25,
      "learning_rate": 0.0005288634507190882,
      "loss": 1.6391,
      "step": 28002
    },
    {
      "epoch": 2.4092747139292783,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005288364550341698,
      "loss": 0.7578,
      "step": 28003
    },
    {
      "epoch": 2.4093607502366,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005288094592649105,
      "loss": 1.7105,
      "step": 28004
    },
    {
      "epoch": 2.4094467865439215,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005287824634113891,
      "loss": 0.9809,
      "step": 28005
    },
    {
      "epoch": 2.409532822851243,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005287554674736848,
      "loss": 1.0968,
      "step": 28006
    },
    {
      "epoch": 2.4096188591585648,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005287284714518759,
      "loss": 1.7336,
      "step": 28007
    },
    {
      "epoch": 2.409704895465887,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005287014753460421,
      "loss": 1.2073,
      "step": 28008
    },
    {
      "epoch": 2.4097909317732085,
      "grad_norm": 1.125,
      "learning_rate": 0.000528674479156262,
      "loss": 1.7675,
      "step": 28009
    },
    {
      "epoch": 2.40987696808053,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005286474828826148,
      "loss": 0.7466,
      "step": 28010
    },
    {
      "epoch": 2.4099630043878517,
      "grad_norm": 1.6484375,
      "learning_rate": 0.000528620486525179,
      "loss": 1.963,
      "step": 28011
    },
    {
      "epoch": 2.4100490406951733,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005285934900840341,
      "loss": 1.05,
      "step": 28012
    },
    {
      "epoch": 2.410135077002495,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005285664935592586,
      "loss": 1.2588,
      "step": 28013
    },
    {
      "epoch": 2.4102211133098166,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005285394969509319,
      "loss": 1.0891,
      "step": 28014
    },
    {
      "epoch": 2.4103071496171387,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005285125002591325,
      "loss": 0.8997,
      "step": 28015
    },
    {
      "epoch": 2.4103931859244603,
      "grad_norm": 1.375,
      "learning_rate": 0.0005284855034839395,
      "loss": 1.7131,
      "step": 28016
    },
    {
      "epoch": 2.410479222231782,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005284585066254321,
      "loss": 0.6291,
      "step": 28017
    },
    {
      "epoch": 2.4105652585391035,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005284315096836892,
      "loss": 1.1314,
      "step": 28018
    },
    {
      "epoch": 2.410651294846425,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005284045126587896,
      "loss": 0.6859,
      "step": 28019
    },
    {
      "epoch": 2.4107373311537468,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005283775155508122,
      "loss": 0.8782,
      "step": 28020
    },
    {
      "epoch": 2.4108233674610684,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005283505183598363,
      "loss": 0.92,
      "step": 28021
    },
    {
      "epoch": 2.4109094037683905,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005283235210859404,
      "loss": 1.4678,
      "step": 28022
    },
    {
      "epoch": 2.410995440075712,
      "grad_norm": 1.0,
      "learning_rate": 0.0005282965237292039,
      "loss": 1.2827,
      "step": 28023
    },
    {
      "epoch": 2.4110814763830337,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005282695262897054,
      "loss": 1.2464,
      "step": 28024
    },
    {
      "epoch": 2.4111675126903553,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005282425287675242,
      "loss": 1.5738,
      "step": 28025
    },
    {
      "epoch": 2.411253548997677,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000528215531162739,
      "loss": 1.1332,
      "step": 28026
    },
    {
      "epoch": 2.4113395853049986,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005281885334754288,
      "loss": 1.3041,
      "step": 28027
    },
    {
      "epoch": 2.41142562161232,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005281615357056727,
      "loss": 0.9258,
      "step": 28028
    },
    {
      "epoch": 2.4115116579196423,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005281345378535497,
      "loss": 1.1058,
      "step": 28029
    },
    {
      "epoch": 2.411597694226964,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005281075399191384,
      "loss": 2.2513,
      "step": 28030
    },
    {
      "epoch": 2.4116837305342855,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005280805419025182,
      "loss": 1.6829,
      "step": 28031
    },
    {
      "epoch": 2.411769766841607,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005280535438037679,
      "loss": 1.2557,
      "step": 28032
    },
    {
      "epoch": 2.411855803148929,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005280265456229664,
      "loss": 1.0865,
      "step": 28033
    },
    {
      "epoch": 2.4119418394562504,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005279995473601926,
      "loss": 1.6919,
      "step": 28034
    },
    {
      "epoch": 2.412027875763572,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005279725490155257,
      "loss": 1.7471,
      "step": 28035
    },
    {
      "epoch": 2.412113912070894,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005279455505890445,
      "loss": 1.3351,
      "step": 28036
    },
    {
      "epoch": 2.4121999483782157,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005279185520808279,
      "loss": 1.3223,
      "step": 28037
    },
    {
      "epoch": 2.4122859846855373,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005278915534909551,
      "loss": 1.3387,
      "step": 28038
    },
    {
      "epoch": 2.412372020992859,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005278645548195049,
      "loss": 0.7872,
      "step": 28039
    },
    {
      "epoch": 2.4124580573001806,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0005278375560665563,
      "loss": 0.6441,
      "step": 28040
    },
    {
      "epoch": 2.412544093607502,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005278105572321882,
      "loss": 0.7109,
      "step": 28041
    },
    {
      "epoch": 2.412630129914824,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005277835583164796,
      "loss": 0.7692,
      "step": 28042
    },
    {
      "epoch": 2.412716166222146,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005277565593195096,
      "loss": 1.3743,
      "step": 28043
    },
    {
      "epoch": 2.4128022025294675,
      "grad_norm": 1.7578125,
      "learning_rate": 0.000527729560241357,
      "loss": 1.0896,
      "step": 28044
    },
    {
      "epoch": 2.412888238836789,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005277025610821008,
      "loss": 1.0363,
      "step": 28045
    },
    {
      "epoch": 2.412974275144111,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00052767556184182,
      "loss": 0.6799,
      "step": 28046
    },
    {
      "epoch": 2.4130603114514324,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005276485625205937,
      "loss": 1.3031,
      "step": 28047
    },
    {
      "epoch": 2.413146347758754,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005276215631185008,
      "loss": 1.5985,
      "step": 28048
    },
    {
      "epoch": 2.4132323840660757,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005275945636356199,
      "loss": 1.3304,
      "step": 28049
    },
    {
      "epoch": 2.4133184203733977,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005275675640720305,
      "loss": 0.6617,
      "step": 28050
    },
    {
      "epoch": 2.4134044566807193,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005275405644278112,
      "loss": 0.9732,
      "step": 28051
    },
    {
      "epoch": 2.413490492988041,
      "grad_norm": 2.296875,
      "learning_rate": 0.000527513564703041,
      "loss": 1.3832,
      "step": 28052
    },
    {
      "epoch": 2.4135765292953626,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005274865648977992,
      "loss": 1.1996,
      "step": 28053
    },
    {
      "epoch": 2.413662565602684,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005274595650121645,
      "loss": 1.3098,
      "step": 28054
    },
    {
      "epoch": 2.413748601910006,
      "grad_norm": 1.0625,
      "learning_rate": 0.000527432565046216,
      "loss": 1.314,
      "step": 28055
    },
    {
      "epoch": 2.4138346382173275,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005274055650000323,
      "loss": 1.7585,
      "step": 28056
    },
    {
      "epoch": 2.4139206745246495,
      "grad_norm": 1.375,
      "learning_rate": 0.000527378564873693,
      "loss": 1.0956,
      "step": 28057
    },
    {
      "epoch": 2.414006710831971,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005273515646672765,
      "loss": 1.0901,
      "step": 28058
    },
    {
      "epoch": 2.414092747139293,
      "grad_norm": 1.78125,
      "learning_rate": 0.000527324564380862,
      "loss": 1.2089,
      "step": 28059
    },
    {
      "epoch": 2.4141787834466144,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005272975640145285,
      "loss": 1.0971,
      "step": 28060
    },
    {
      "epoch": 2.414264819753936,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005272705635683549,
      "loss": 1.0886,
      "step": 28061
    },
    {
      "epoch": 2.414350856061258,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005272435630424202,
      "loss": 1.1391,
      "step": 28062
    },
    {
      "epoch": 2.4144368923685797,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005272165624368035,
      "loss": 0.8942,
      "step": 28063
    },
    {
      "epoch": 2.4145229286759013,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005271895617515835,
      "loss": 1.1792,
      "step": 28064
    },
    {
      "epoch": 2.414608964983223,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005271625609868393,
      "loss": 0.9837,
      "step": 28065
    },
    {
      "epoch": 2.4146950012905446,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005271355601426501,
      "loss": 1.0696,
      "step": 28066
    },
    {
      "epoch": 2.414781037597866,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005271085592190944,
      "loss": 0.9066,
      "step": 28067
    },
    {
      "epoch": 2.414867073905188,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005270815582162518,
      "loss": 0.6387,
      "step": 28068
    },
    {
      "epoch": 2.41495311021251,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005270545571342005,
      "loss": 1.313,
      "step": 28069
    },
    {
      "epoch": 2.4150391465198315,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005270275559730202,
      "loss": 1.1288,
      "step": 28070
    },
    {
      "epoch": 2.415125182827153,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005270005547327893,
      "loss": 1.7123,
      "step": 28071
    },
    {
      "epoch": 2.415211219134475,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005269735534135872,
      "loss": 1.5336,
      "step": 28072
    },
    {
      "epoch": 2.4152972554417964,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005269465520154927,
      "loss": 1.0941,
      "step": 28073
    },
    {
      "epoch": 2.415383291749118,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005269195505385847,
      "loss": 1.6409,
      "step": 28074
    },
    {
      "epoch": 2.4154693280564397,
      "grad_norm": 1.5,
      "learning_rate": 0.0005268925489829423,
      "loss": 1.7886,
      "step": 28075
    },
    {
      "epoch": 2.4155553643637617,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005268655473486444,
      "loss": 0.9037,
      "step": 28076
    },
    {
      "epoch": 2.4156414006710833,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00052683854563577,
      "loss": 0.985,
      "step": 28077
    },
    {
      "epoch": 2.415727436978405,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005268115438443981,
      "loss": 0.9484,
      "step": 28078
    },
    {
      "epoch": 2.4158134732857266,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005267845419746075,
      "loss": 0.523,
      "step": 28079
    },
    {
      "epoch": 2.4158995095930482,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005267575400264776,
      "loss": 0.9779,
      "step": 28080
    },
    {
      "epoch": 2.41598554590037,
      "grad_norm": 1.0625,
      "learning_rate": 0.000526730538000087,
      "loss": 0.9027,
      "step": 28081
    },
    {
      "epoch": 2.4160715822076915,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005267035358955147,
      "loss": 1.3183,
      "step": 28082
    },
    {
      "epoch": 2.4161576185150135,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005266765337128397,
      "loss": 1.1231,
      "step": 28083
    },
    {
      "epoch": 2.416243654822335,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005266495314521411,
      "loss": 1.1482,
      "step": 28084
    },
    {
      "epoch": 2.416329691129657,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005266225291134978,
      "loss": 1.2645,
      "step": 28085
    },
    {
      "epoch": 2.4164157274369784,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005265955266969887,
      "loss": 1.0331,
      "step": 28086
    },
    {
      "epoch": 2.4165017637443,
      "grad_norm": 1.28125,
      "learning_rate": 0.000526568524202693,
      "loss": 2.3063,
      "step": 28087
    },
    {
      "epoch": 2.4165878000516217,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005265415216306895,
      "loss": 1.6237,
      "step": 28088
    },
    {
      "epoch": 2.4166738363589433,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005265145189810573,
      "loss": 1.379,
      "step": 28089
    },
    {
      "epoch": 2.4167598726662654,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000526487516253875,
      "loss": 0.839,
      "step": 28090
    },
    {
      "epoch": 2.416845908973587,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005264605134492222,
      "loss": 1.0506,
      "step": 28091
    },
    {
      "epoch": 2.4169319452809086,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005264335105671774,
      "loss": 1.7326,
      "step": 28092
    },
    {
      "epoch": 2.4170179815882302,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005264065076078196,
      "loss": 1.3666,
      "step": 28093
    },
    {
      "epoch": 2.417104017895552,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005263795045712281,
      "loss": 0.7239,
      "step": 28094
    },
    {
      "epoch": 2.4171900542028735,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005263525014574816,
      "loss": 0.6047,
      "step": 28095
    },
    {
      "epoch": 2.417276090510195,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005263254982666592,
      "loss": 0.7831,
      "step": 28096
    },
    {
      "epoch": 2.417362126817517,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005262984949988397,
      "loss": 0.6961,
      "step": 28097
    },
    {
      "epoch": 2.417448163124839,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005262714916541022,
      "loss": 0.9343,
      "step": 28098
    },
    {
      "epoch": 2.4175341994321604,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000526244488232526,
      "loss": 1.0552,
      "step": 28099
    },
    {
      "epoch": 2.417620235739482,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005262174847341894,
      "loss": 1.1513,
      "step": 28100
    },
    {
      "epoch": 2.4177062720468037,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005261904811591719,
      "loss": 1.2095,
      "step": 28101
    },
    {
      "epoch": 2.4177923083541253,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005261634775075524,
      "loss": 0.8832,
      "step": 28102
    },
    {
      "epoch": 2.417878344661447,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005261364737794099,
      "loss": 1.1076,
      "step": 28103
    },
    {
      "epoch": 2.417964380968769,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005261094699748231,
      "loss": 0.8671,
      "step": 28104
    },
    {
      "epoch": 2.4180504172760906,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005260824660938713,
      "loss": 0.6969,
      "step": 28105
    },
    {
      "epoch": 2.4181364535834122,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005260554621366333,
      "loss": 0.8347,
      "step": 28106
    },
    {
      "epoch": 2.418222489890734,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005260284581031882,
      "loss": 0.9616,
      "step": 28107
    },
    {
      "epoch": 2.4183085261980555,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005260014539936147,
      "loss": 1.0761,
      "step": 28108
    },
    {
      "epoch": 2.418394562505377,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005259744498079924,
      "loss": 1.4494,
      "step": 28109
    },
    {
      "epoch": 2.4184805988126987,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005259474455463996,
      "loss": 0.9631,
      "step": 28110
    },
    {
      "epoch": 2.418566635120021,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005259204412089157,
      "loss": 1.3754,
      "step": 28111
    },
    {
      "epoch": 2.4186526714273424,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005258934367956194,
      "loss": 0.8684,
      "step": 28112
    },
    {
      "epoch": 2.418738707734664,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00052586643230659,
      "loss": 1.5213,
      "step": 28113
    },
    {
      "epoch": 2.4188247440419857,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005258394277419063,
      "loss": 0.7605,
      "step": 28114
    },
    {
      "epoch": 2.4189107803493073,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005258124231016472,
      "loss": 0.9864,
      "step": 28115
    },
    {
      "epoch": 2.418996816656629,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005257854183858918,
      "loss": 0.9457,
      "step": 28116
    },
    {
      "epoch": 2.419082852963951,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005257584135947191,
      "loss": 1.1898,
      "step": 28117
    },
    {
      "epoch": 2.4191688892712726,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005257314087282082,
      "loss": 1.3823,
      "step": 28118
    },
    {
      "epoch": 2.4192549255785942,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005257044037864375,
      "loss": 0.4825,
      "step": 28119
    },
    {
      "epoch": 2.419340961885916,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005256773987694867,
      "loss": 0.9862,
      "step": 28120
    },
    {
      "epoch": 2.4194269981932375,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005256503936774346,
      "loss": 1.3488,
      "step": 28121
    },
    {
      "epoch": 2.419513034500559,
      "grad_norm": 1.15625,
      "learning_rate": 0.00052562338851036,
      "loss": 1.7661,
      "step": 28122
    },
    {
      "epoch": 2.419599070807881,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005255963832683419,
      "loss": 1.7341,
      "step": 28123
    },
    {
      "epoch": 2.419685107115203,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005255693779514594,
      "loss": 1.2287,
      "step": 28124
    },
    {
      "epoch": 2.4197711434225244,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005255423725597915,
      "loss": 0.5318,
      "step": 28125
    },
    {
      "epoch": 2.419857179729846,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005255153670934171,
      "loss": 1.55,
      "step": 28126
    },
    {
      "epoch": 2.4199432160371677,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005254883615524152,
      "loss": 1.2044,
      "step": 28127
    },
    {
      "epoch": 2.4200292523444893,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0005254613559368648,
      "loss": 1.4455,
      "step": 28128
    },
    {
      "epoch": 2.420115288651811,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000525434350246845,
      "loss": 1.6287,
      "step": 28129
    },
    {
      "epoch": 2.420201324959133,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005254073444824345,
      "loss": 1.7752,
      "step": 28130
    },
    {
      "epoch": 2.4202873612664546,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005253803386437125,
      "loss": 0.7001,
      "step": 28131
    },
    {
      "epoch": 2.4203733975737762,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005253533327307579,
      "loss": 0.7526,
      "step": 28132
    },
    {
      "epoch": 2.420459433881098,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005253263267436499,
      "loss": 1.1716,
      "step": 28133
    },
    {
      "epoch": 2.4205454701884195,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005252993206824671,
      "loss": 1.2993,
      "step": 28134
    },
    {
      "epoch": 2.420631506495741,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005252723145472888,
      "loss": 1.2845,
      "step": 28135
    },
    {
      "epoch": 2.4207175428030627,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005252453083381941,
      "loss": 1.0832,
      "step": 28136
    },
    {
      "epoch": 2.420803579110385,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0005252183020552615,
      "loss": 0.9637,
      "step": 28137
    },
    {
      "epoch": 2.4208896154177064,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005251912956985704,
      "loss": 0.6428,
      "step": 28138
    },
    {
      "epoch": 2.420975651725028,
      "grad_norm": 1.75,
      "learning_rate": 0.0005251642892681995,
      "loss": 1.9844,
      "step": 28139
    },
    {
      "epoch": 2.4210616880323497,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005251372827642281,
      "loss": 0.609,
      "step": 28140
    },
    {
      "epoch": 2.4211477243396713,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000525110276186735,
      "loss": 0.961,
      "step": 28141
    },
    {
      "epoch": 2.421233760646993,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005250832695357993,
      "loss": 1.0551,
      "step": 28142
    },
    {
      "epoch": 2.4213197969543145,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005250562628114999,
      "loss": 1.7582,
      "step": 28143
    },
    {
      "epoch": 2.4214058332616366,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005250292560139157,
      "loss": 1.9388,
      "step": 28144
    },
    {
      "epoch": 2.4214918695689582,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005250022491431258,
      "loss": 1.2105,
      "step": 28145
    },
    {
      "epoch": 2.42157790587628,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005249752421992093,
      "loss": 1.5198,
      "step": 28146
    },
    {
      "epoch": 2.4216639421836015,
      "grad_norm": 1.0625,
      "learning_rate": 0.000524948235182245,
      "loss": 0.977,
      "step": 28147
    },
    {
      "epoch": 2.421749978490923,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000524921228092312,
      "loss": 1.4667,
      "step": 28148
    },
    {
      "epoch": 2.4218360147982447,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005248942209294891,
      "loss": 1.4884,
      "step": 28149
    },
    {
      "epoch": 2.4219220511055664,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0005248672136938557,
      "loss": 1.3294,
      "step": 28150
    },
    {
      "epoch": 2.4220080874128884,
      "grad_norm": 1.125,
      "learning_rate": 0.0005248402063854905,
      "loss": 1.165,
      "step": 28151
    },
    {
      "epoch": 2.42209412372021,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005248131990044725,
      "loss": 1.0937,
      "step": 28152
    },
    {
      "epoch": 2.4221801600275317,
      "grad_norm": 1.0,
      "learning_rate": 0.0005247861915508805,
      "loss": 0.6796,
      "step": 28153
    },
    {
      "epoch": 2.4222661963348533,
      "grad_norm": 0.890625,
      "learning_rate": 0.000524759184024794,
      "loss": 0.9396,
      "step": 28154
    },
    {
      "epoch": 2.422352232642175,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005247321764262915,
      "loss": 0.8117,
      "step": 28155
    },
    {
      "epoch": 2.4224382689494965,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005247051687554522,
      "loss": 1.882,
      "step": 28156
    },
    {
      "epoch": 2.422524305256818,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005246781610123551,
      "loss": 1.3978,
      "step": 28157
    },
    {
      "epoch": 2.4226103415641402,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005246511531970793,
      "loss": 0.8206,
      "step": 28158
    },
    {
      "epoch": 2.422696377871462,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005246241453097036,
      "loss": 1.1107,
      "step": 28159
    },
    {
      "epoch": 2.4227824141787835,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005245971373503071,
      "loss": 1.0901,
      "step": 28160
    },
    {
      "epoch": 2.422868450486105,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005245701293189686,
      "loss": 0.8086,
      "step": 28161
    },
    {
      "epoch": 2.4229544867934267,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005245431212157674,
      "loss": 1.2489,
      "step": 28162
    },
    {
      "epoch": 2.4230405231007484,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005245161130407824,
      "loss": 1.5347,
      "step": 28163
    },
    {
      "epoch": 2.42312655940807,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005244891047940924,
      "loss": 1.9946,
      "step": 28164
    },
    {
      "epoch": 2.423212595715392,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005244620964757765,
      "loss": 1.1156,
      "step": 28165
    },
    {
      "epoch": 2.4232986320227137,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005244350880859139,
      "loss": 1.5399,
      "step": 28166
    },
    {
      "epoch": 2.4233846683300353,
      "grad_norm": 1.25,
      "learning_rate": 0.0005244080796245834,
      "loss": 2.0861,
      "step": 28167
    },
    {
      "epoch": 2.423470704637357,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005243810710918638,
      "loss": 1.3162,
      "step": 28168
    },
    {
      "epoch": 2.4235567409446785,
      "grad_norm": 1.0,
      "learning_rate": 0.0005243540624878346,
      "loss": 0.9826,
      "step": 28169
    },
    {
      "epoch": 2.423642777252,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005243270538125744,
      "loss": 0.9882,
      "step": 28170
    },
    {
      "epoch": 2.423728813559322,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005243000450661623,
      "loss": 1.005,
      "step": 28171
    },
    {
      "epoch": 2.423814849866644,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005242730362486771,
      "loss": 1.171,
      "step": 28172
    },
    {
      "epoch": 2.4239008861739655,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005242460273601982,
      "loss": 1.4668,
      "step": 28173
    },
    {
      "epoch": 2.423986922481287,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005242190184008042,
      "loss": 1.7915,
      "step": 28174
    },
    {
      "epoch": 2.4240729587886087,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005241920093705746,
      "loss": 0.9659,
      "step": 28175
    },
    {
      "epoch": 2.4241589950959304,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005241650002695877,
      "loss": 0.98,
      "step": 28176
    },
    {
      "epoch": 2.424245031403252,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005241379910979231,
      "loss": 1.7258,
      "step": 28177
    },
    {
      "epoch": 2.424331067710574,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005241109818556595,
      "loss": 1.0087,
      "step": 28178
    },
    {
      "epoch": 2.4244171040178957,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005240839725428759,
      "loss": 1.1281,
      "step": 28179
    },
    {
      "epoch": 2.4245031403252173,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005240569631596516,
      "loss": 1.1458,
      "step": 28180
    },
    {
      "epoch": 2.424589176632539,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005240299537060653,
      "loss": 1.7025,
      "step": 28181
    },
    {
      "epoch": 2.4246752129398605,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005240029441821959,
      "loss": 2.2865,
      "step": 28182
    },
    {
      "epoch": 2.424761249247182,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005239759345881227,
      "loss": 1.1065,
      "step": 28183
    },
    {
      "epoch": 2.4248472855545042,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005239489249239244,
      "loss": 0.8545,
      "step": 28184
    },
    {
      "epoch": 2.424933321861826,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005239219151896803,
      "loss": 0.4771,
      "step": 28185
    },
    {
      "epoch": 2.4250193581691475,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005238949053854693,
      "loss": 0.9549,
      "step": 28186
    },
    {
      "epoch": 2.425105394476469,
      "grad_norm": 1.6875,
      "learning_rate": 0.0005238678955113702,
      "loss": 1.7856,
      "step": 28187
    },
    {
      "epoch": 2.4251914307837907,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005238408855674622,
      "loss": 1.4888,
      "step": 28188
    },
    {
      "epoch": 2.4252774670911124,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005238138755538242,
      "loss": 2.283,
      "step": 28189
    },
    {
      "epoch": 2.425363503398434,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005237868654705352,
      "loss": 1.2029,
      "step": 28190
    },
    {
      "epoch": 2.425449539705756,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005237598553176745,
      "loss": 1.1454,
      "step": 28191
    },
    {
      "epoch": 2.4255355760130777,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005237328450953206,
      "loss": 1.2413,
      "step": 28192
    },
    {
      "epoch": 2.4256216123203993,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005237058348035527,
      "loss": 1.1634,
      "step": 28193
    },
    {
      "epoch": 2.425707648627721,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00052367882444245,
      "loss": 0.9494,
      "step": 28194
    },
    {
      "epoch": 2.4257936849350425,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005236518140120913,
      "loss": 1.0503,
      "step": 28195
    },
    {
      "epoch": 2.425879721242364,
      "grad_norm": 17.5,
      "learning_rate": 0.0005236248035125557,
      "loss": 1.8632,
      "step": 28196
    },
    {
      "epoch": 2.425965757549686,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005235977929439221,
      "loss": 1.4736,
      "step": 28197
    },
    {
      "epoch": 2.426051793857008,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005235707823062695,
      "loss": 1.906,
      "step": 28198
    },
    {
      "epoch": 2.4261378301643295,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005235437715996771,
      "loss": 0.7177,
      "step": 28199
    },
    {
      "epoch": 2.426223866471651,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005235167608242235,
      "loss": 0.914,
      "step": 28200
    },
    {
      "epoch": 2.4263099027789727,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005234897499799882,
      "loss": 1.1761,
      "step": 28201
    },
    {
      "epoch": 2.4263959390862944,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005234627390670498,
      "loss": 1.3207,
      "step": 28202
    },
    {
      "epoch": 2.426481975393616,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005234357280854875,
      "loss": 1.0377,
      "step": 28203
    },
    {
      "epoch": 2.4265680117009376,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005234087170353801,
      "loss": 2.0802,
      "step": 28204
    },
    {
      "epoch": 2.4266540480082597,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005233817059168068,
      "loss": 1.1404,
      "step": 28205
    },
    {
      "epoch": 2.4267400843155813,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005233546947298469,
      "loss": 1.5893,
      "step": 28206
    },
    {
      "epoch": 2.426826120622903,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005233276834745787,
      "loss": 0.9188,
      "step": 28207
    },
    {
      "epoch": 2.4269121569302246,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005233006721510818,
      "loss": 0.9794,
      "step": 28208
    },
    {
      "epoch": 2.426998193237546,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005232736607594347,
      "loss": 1.106,
      "step": 28209
    },
    {
      "epoch": 2.427084229544868,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005232466492997169,
      "loss": 0.8016,
      "step": 28210
    },
    {
      "epoch": 2.4271702658521894,
      "grad_norm": 1.140625,
      "learning_rate": 0.000523219637772007,
      "loss": 1.058,
      "step": 28211
    },
    {
      "epoch": 2.4272563021595115,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005231926261763843,
      "loss": 0.77,
      "step": 28212
    },
    {
      "epoch": 2.427342338466833,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005231656145129275,
      "loss": 1.8734,
      "step": 28213
    },
    {
      "epoch": 2.4274283747741547,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000523138602781716,
      "loss": 1.1847,
      "step": 28214
    },
    {
      "epoch": 2.4275144110814764,
      "grad_norm": 1.25,
      "learning_rate": 0.0005231115909828285,
      "loss": 1.6208,
      "step": 28215
    },
    {
      "epoch": 2.427600447388798,
      "grad_norm": 1.0,
      "learning_rate": 0.0005230845791163441,
      "loss": 0.6819,
      "step": 28216
    },
    {
      "epoch": 2.4276864836961196,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005230575671823417,
      "loss": 1.2354,
      "step": 28217
    },
    {
      "epoch": 2.4277725200034412,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005230305551809006,
      "loss": 1.4425,
      "step": 28218
    },
    {
      "epoch": 2.4278585563107633,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005230035431120995,
      "loss": 1.4572,
      "step": 28219
    },
    {
      "epoch": 2.427944592618085,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005229765309760174,
      "loss": 1.1655,
      "step": 28220
    },
    {
      "epoch": 2.4280306289254066,
      "grad_norm": 1.5,
      "learning_rate": 0.0005229495187727335,
      "loss": 0.9269,
      "step": 28221
    },
    {
      "epoch": 2.428116665232728,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005229225065023268,
      "loss": 1.3516,
      "step": 28222
    },
    {
      "epoch": 2.42820270154005,
      "grad_norm": 1.0,
      "learning_rate": 0.0005228954941648762,
      "loss": 0.8095,
      "step": 28223
    },
    {
      "epoch": 2.4282887378473714,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005228684817604607,
      "loss": 0.9909,
      "step": 28224
    },
    {
      "epoch": 2.428374774154693,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005228414692891594,
      "loss": 1.3789,
      "step": 28225
    },
    {
      "epoch": 2.428460810462015,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005228144567510512,
      "loss": 0.743,
      "step": 28226
    },
    {
      "epoch": 2.4285468467693367,
      "grad_norm": 1.40625,
      "learning_rate": 0.0005227874441462152,
      "loss": 1.5678,
      "step": 28227
    },
    {
      "epoch": 2.4286328830766584,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005227604314747302,
      "loss": 1.5886,
      "step": 28228
    },
    {
      "epoch": 2.42871891938398,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005227334187366755,
      "loss": 0.8584,
      "step": 28229
    },
    {
      "epoch": 2.4288049556913016,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005227064059321298,
      "loss": 0.8455,
      "step": 28230
    },
    {
      "epoch": 2.4288909919986232,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005226793930611725,
      "loss": 1.0438,
      "step": 28231
    },
    {
      "epoch": 2.428977028305945,
      "grad_norm": 1.953125,
      "learning_rate": 0.0005226523801238822,
      "loss": 1.9994,
      "step": 28232
    },
    {
      "epoch": 2.429063064613267,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005226253671203382,
      "loss": 0.8768,
      "step": 28233
    },
    {
      "epoch": 2.4291491009205886,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005225983540506194,
      "loss": 0.7503,
      "step": 28234
    },
    {
      "epoch": 2.42923513722791,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005225713409148048,
      "loss": 0.7798,
      "step": 28235
    },
    {
      "epoch": 2.429321173535232,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005225443277129734,
      "loss": 0.7232,
      "step": 28236
    },
    {
      "epoch": 2.4294072098425534,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005225173144452044,
      "loss": 1.1678,
      "step": 28237
    },
    {
      "epoch": 2.4294932461498755,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005224903011115764,
      "loss": 1.3715,
      "step": 28238
    },
    {
      "epoch": 2.429579282457197,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005224632877121687,
      "loss": 1.5097,
      "step": 28239
    },
    {
      "epoch": 2.4296653187645187,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005224362742470603,
      "loss": 0.9213,
      "step": 28240
    },
    {
      "epoch": 2.4297513550718404,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005224092607163301,
      "loss": 1.1784,
      "step": 28241
    },
    {
      "epoch": 2.429837391379162,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005223822471200573,
      "loss": 0.945,
      "step": 28242
    },
    {
      "epoch": 2.4299234276864836,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005223552334583206,
      "loss": 1.1545,
      "step": 28243
    },
    {
      "epoch": 2.4300094639938052,
      "grad_norm": 1.25,
      "learning_rate": 0.0005223282197311992,
      "loss": 1.3925,
      "step": 28244
    },
    {
      "epoch": 2.4300955003011273,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005223012059387723,
      "loss": 1.2173,
      "step": 28245
    },
    {
      "epoch": 2.430181536608449,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005222741920811185,
      "loss": 1.3325,
      "step": 28246
    },
    {
      "epoch": 2.4302675729157706,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005222471781583171,
      "loss": 1.1517,
      "step": 28247
    },
    {
      "epoch": 2.430353609223092,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005222201641704471,
      "loss": 0.6952,
      "step": 28248
    },
    {
      "epoch": 2.430439645530414,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005221931501175874,
      "loss": 0.838,
      "step": 28249
    },
    {
      "epoch": 2.4305256818377354,
      "grad_norm": 1.03125,
      "learning_rate": 0.000522166135999817,
      "loss": 1.1592,
      "step": 28250
    },
    {
      "epoch": 2.430611718145057,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005221391218172149,
      "loss": 1.1743,
      "step": 28251
    },
    {
      "epoch": 2.430697754452379,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005221121075698604,
      "loss": 0.7845,
      "step": 28252
    },
    {
      "epoch": 2.4307837907597007,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005220850932578322,
      "loss": 1.1884,
      "step": 28253
    },
    {
      "epoch": 2.4308698270670224,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005220580788812092,
      "loss": 1.7032,
      "step": 28254
    },
    {
      "epoch": 2.430955863374344,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005220310644400707,
      "loss": 1.3711,
      "step": 28255
    },
    {
      "epoch": 2.4310418996816656,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005220040499344958,
      "loss": 1.4339,
      "step": 28256
    },
    {
      "epoch": 2.4311279359889872,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005219770353645633,
      "loss": 1.0516,
      "step": 28257
    },
    {
      "epoch": 2.431213972296309,
      "grad_norm": 1.109375,
      "learning_rate": 0.000521950020730352,
      "loss": 1.238,
      "step": 28258
    },
    {
      "epoch": 2.431300008603631,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005219230060319414,
      "loss": 1.3014,
      "step": 28259
    },
    {
      "epoch": 2.4313860449109526,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005218959912694103,
      "loss": 0.9402,
      "step": 28260
    },
    {
      "epoch": 2.431472081218274,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005218689764428375,
      "loss": 1.6188,
      "step": 28261
    },
    {
      "epoch": 2.431558117525596,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005218419615523022,
      "loss": 1.7066,
      "step": 28262
    },
    {
      "epoch": 2.4316441538329174,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005218149465978835,
      "loss": 1.1829,
      "step": 28263
    },
    {
      "epoch": 2.431730190140239,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005217879315796604,
      "loss": 2.1669,
      "step": 28264
    },
    {
      "epoch": 2.4318162264475607,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005217609164977117,
      "loss": 1.3774,
      "step": 28265
    },
    {
      "epoch": 2.4319022627548827,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005217339013521167,
      "loss": 1.3273,
      "step": 28266
    },
    {
      "epoch": 2.4319882990622044,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005217068861429541,
      "loss": 1.0408,
      "step": 28267
    },
    {
      "epoch": 2.432074335369526,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0005216798708703031,
      "loss": 1.6113,
      "step": 28268
    },
    {
      "epoch": 2.4321603716768476,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005216528555342427,
      "loss": 0.5646,
      "step": 28269
    },
    {
      "epoch": 2.4322464079841692,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005216258401348519,
      "loss": 1.0614,
      "step": 28270
    },
    {
      "epoch": 2.432332444291491,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0005215988246722098,
      "loss": 1.866,
      "step": 28271
    },
    {
      "epoch": 2.4324184805988125,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005215718091463954,
      "loss": 0.9527,
      "step": 28272
    },
    {
      "epoch": 2.4325045169061346,
      "grad_norm": 1.46875,
      "learning_rate": 0.0005215447935574874,
      "loss": 1.1855,
      "step": 28273
    },
    {
      "epoch": 2.432590553213456,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005215177779055654,
      "loss": 1.1391,
      "step": 28274
    },
    {
      "epoch": 2.432676589520778,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005214907621907079,
      "loss": 1.0882,
      "step": 28275
    },
    {
      "epoch": 2.4327626258280994,
      "grad_norm": 1.109375,
      "learning_rate": 0.000521463746412994,
      "loss": 1.3935,
      "step": 28276
    },
    {
      "epoch": 2.432848662135421,
      "grad_norm": 1.15625,
      "learning_rate": 0.000521436730572503,
      "loss": 1.7604,
      "step": 28277
    },
    {
      "epoch": 2.4329346984427427,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005214097146693137,
      "loss": 1.194,
      "step": 28278
    },
    {
      "epoch": 2.4330207347500643,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000521382698703505,
      "loss": 1.2416,
      "step": 28279
    },
    {
      "epoch": 2.4331067710573864,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005213556826751562,
      "loss": 1.333,
      "step": 28280
    },
    {
      "epoch": 2.433192807364708,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005213286665843461,
      "loss": 1.1197,
      "step": 28281
    },
    {
      "epoch": 2.4332788436720296,
      "grad_norm": 1.25,
      "learning_rate": 0.000521301650431154,
      "loss": 1.1657,
      "step": 28282
    },
    {
      "epoch": 2.4333648799793512,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005212746342156586,
      "loss": 1.1059,
      "step": 28283
    },
    {
      "epoch": 2.433450916286673,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000521247617937939,
      "loss": 0.9672,
      "step": 28284
    },
    {
      "epoch": 2.4335369525939945,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005212206015980742,
      "loss": 0.9872,
      "step": 28285
    },
    {
      "epoch": 2.433622988901316,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005211935851961433,
      "loss": 1.1526,
      "step": 28286
    },
    {
      "epoch": 2.433709025208638,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005211665687322254,
      "loss": 0.7853,
      "step": 28287
    },
    {
      "epoch": 2.43379506151596,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005211395522063992,
      "loss": 0.7844,
      "step": 28288
    },
    {
      "epoch": 2.4338810978232814,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000521112535618744,
      "loss": 0.6824,
      "step": 28289
    },
    {
      "epoch": 2.433967134130603,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005210855189693389,
      "loss": 1.18,
      "step": 28290
    },
    {
      "epoch": 2.4340531704379247,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005210585022582626,
      "loss": 0.98,
      "step": 28291
    },
    {
      "epoch": 2.4341392067452463,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005210314854855943,
      "loss": 1.0386,
      "step": 28292
    },
    {
      "epoch": 2.4342252430525684,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000521004468651413,
      "loss": 1.2394,
      "step": 28293
    },
    {
      "epoch": 2.43431127935989,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005209774517557976,
      "loss": 1.1877,
      "step": 28294
    },
    {
      "epoch": 2.4343973156672116,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005209504347988274,
      "loss": 1.5369,
      "step": 28295
    },
    {
      "epoch": 2.4344833519745332,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005209234177805812,
      "loss": 1.3452,
      "step": 28296
    },
    {
      "epoch": 2.434569388281855,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005208964007011381,
      "loss": 1.5534,
      "step": 28297
    },
    {
      "epoch": 2.4346554245891765,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005208693835605772,
      "loss": 0.9146,
      "step": 28298
    },
    {
      "epoch": 2.4347414608964986,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005208423663589772,
      "loss": 1.1565,
      "step": 28299
    },
    {
      "epoch": 2.43482749720382,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005208153490964173,
      "loss": 0.9502,
      "step": 28300
    },
    {
      "epoch": 2.434913533511142,
      "grad_norm": 1.25,
      "learning_rate": 0.0005207883317729768,
      "loss": 1.2718,
      "step": 28301
    },
    {
      "epoch": 2.4349995698184634,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005207613143887343,
      "loss": 1.2065,
      "step": 28302
    },
    {
      "epoch": 2.435085606125785,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005207342969437691,
      "loss": 1.0545,
      "step": 28303
    },
    {
      "epoch": 2.4351716424331067,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005207072794381599,
      "loss": 1.5755,
      "step": 28304
    },
    {
      "epoch": 2.4352576787404283,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005206802618719862,
      "loss": 1.6955,
      "step": 28305
    },
    {
      "epoch": 2.4353437150477504,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005206532442453266,
      "loss": 1.3818,
      "step": 28306
    },
    {
      "epoch": 2.435429751355072,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005206262265582602,
      "loss": 1.5317,
      "step": 28307
    },
    {
      "epoch": 2.4355157876623936,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005205992088108662,
      "loss": 1.9381,
      "step": 28308
    },
    {
      "epoch": 2.4356018239697153,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005205721910032236,
      "loss": 1.4311,
      "step": 28309
    },
    {
      "epoch": 2.435687860277037,
      "grad_norm": 1.640625,
      "learning_rate": 0.0005205451731354111,
      "loss": 1.6174,
      "step": 28310
    },
    {
      "epoch": 2.4357738965843585,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005205181552075083,
      "loss": 1.2722,
      "step": 28311
    },
    {
      "epoch": 2.43585993289168,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005204911372195937,
      "loss": 0.9739,
      "step": 28312
    },
    {
      "epoch": 2.435945969199002,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005204641191717466,
      "loss": 0.983,
      "step": 28313
    },
    {
      "epoch": 2.436032005506324,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005204371010640458,
      "loss": 1.7499,
      "step": 28314
    },
    {
      "epoch": 2.4361180418136454,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005204100828965706,
      "loss": 0.8114,
      "step": 28315
    },
    {
      "epoch": 2.436204078120967,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005203830646693997,
      "loss": 0.8309,
      "step": 28316
    },
    {
      "epoch": 2.4362901144282887,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005203560463826124,
      "loss": 1.4604,
      "step": 28317
    },
    {
      "epoch": 2.4363761507356103,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005203290280362876,
      "loss": 1.2528,
      "step": 28318
    },
    {
      "epoch": 2.436462187042932,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005203020096305044,
      "loss": 1.2764,
      "step": 28319
    },
    {
      "epoch": 2.436548223350254,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005202749911653418,
      "loss": 0.87,
      "step": 28320
    },
    {
      "epoch": 2.4366342596575756,
      "grad_norm": 2.71875,
      "learning_rate": 0.0005202479726408787,
      "loss": 1.2677,
      "step": 28321
    },
    {
      "epoch": 2.4367202959648973,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005202209540571942,
      "loss": 0.9436,
      "step": 28322
    },
    {
      "epoch": 2.436806332272219,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005201939354143674,
      "loss": 1.111,
      "step": 28323
    },
    {
      "epoch": 2.4368923685795405,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005201669167124773,
      "loss": 1.0141,
      "step": 28324
    },
    {
      "epoch": 2.436978404886862,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005201398979516028,
      "loss": 1.4639,
      "step": 28325
    },
    {
      "epoch": 2.4370644411941838,
      "grad_norm": 2.015625,
      "learning_rate": 0.000520112879131823,
      "loss": 1.452,
      "step": 28326
    },
    {
      "epoch": 2.437150477501506,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005200858602532172,
      "loss": 1.9464,
      "step": 28327
    },
    {
      "epoch": 2.4372365138088274,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005200588413158639,
      "loss": 1.7506,
      "step": 28328
    },
    {
      "epoch": 2.437322550116149,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005200318223198423,
      "loss": 1.135,
      "step": 28329
    },
    {
      "epoch": 2.4374085864234707,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005200048032652318,
      "loss": 1.1372,
      "step": 28330
    },
    {
      "epoch": 2.4374946227307923,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005199777841521111,
      "loss": 0.8957,
      "step": 28331
    },
    {
      "epoch": 2.437580659038114,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005199507649805591,
      "loss": 1.813,
      "step": 28332
    },
    {
      "epoch": 2.4376666953454356,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005199237457506551,
      "loss": 0.9807,
      "step": 28333
    },
    {
      "epoch": 2.4377527316527576,
      "grad_norm": 0.828125,
      "learning_rate": 0.000519896726462478,
      "loss": 0.6911,
      "step": 28334
    },
    {
      "epoch": 2.4378387679600793,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005198697071161068,
      "loss": 2.0945,
      "step": 28335
    },
    {
      "epoch": 2.437924804267401,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005198426877116206,
      "loss": 1.2952,
      "step": 28336
    },
    {
      "epoch": 2.4380108405747225,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005198156682490984,
      "loss": 1.7015,
      "step": 28337
    },
    {
      "epoch": 2.438096876882044,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005197886487286193,
      "loss": 1.0221,
      "step": 28338
    },
    {
      "epoch": 2.4381829131893658,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005197616291502623,
      "loss": 1.8469,
      "step": 28339
    },
    {
      "epoch": 2.4382689494966874,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005197346095141061,
      "loss": 0.9081,
      "step": 28340
    },
    {
      "epoch": 2.4383549858040094,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005197075898202301,
      "loss": 1.5175,
      "step": 28341
    },
    {
      "epoch": 2.438441022111331,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005196805700687134,
      "loss": 1.1882,
      "step": 28342
    },
    {
      "epoch": 2.4385270584186527,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005196535502596348,
      "loss": 0.8251,
      "step": 28343
    },
    {
      "epoch": 2.4386130947259743,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005196265303930731,
      "loss": 1.3612,
      "step": 28344
    },
    {
      "epoch": 2.438699131033296,
      "grad_norm": 1.328125,
      "learning_rate": 0.000519599510469108,
      "loss": 1.7873,
      "step": 28345
    },
    {
      "epoch": 2.4387851673406176,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000519572490487818,
      "loss": 0.651,
      "step": 28346
    },
    {
      "epoch": 2.438871203647939,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005195454704492823,
      "loss": 1.5769,
      "step": 28347
    },
    {
      "epoch": 2.4389572399552613,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005195184503535799,
      "loss": 0.7802,
      "step": 28348
    },
    {
      "epoch": 2.439043276262583,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005194914302007898,
      "loss": 1.0303,
      "step": 28349
    },
    {
      "epoch": 2.4391293125699045,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005194644099909911,
      "loss": 0.9191,
      "step": 28350
    },
    {
      "epoch": 2.439215348877226,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005194373897242627,
      "loss": 1.1084,
      "step": 28351
    },
    {
      "epoch": 2.4393013851845478,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005194103694006838,
      "loss": 1.7134,
      "step": 28352
    },
    {
      "epoch": 2.4393874214918694,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005193833490203334,
      "loss": 1.2828,
      "step": 28353
    },
    {
      "epoch": 2.4394734577991914,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005193563285832904,
      "loss": 1.0691,
      "step": 28354
    },
    {
      "epoch": 2.439559494106513,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005193293080896338,
      "loss": 1.0764,
      "step": 28355
    },
    {
      "epoch": 2.4396455304138347,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005193022875394429,
      "loss": 0.8772,
      "step": 28356
    },
    {
      "epoch": 2.4397315667211563,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005192752669327964,
      "loss": 1.4984,
      "step": 28357
    },
    {
      "epoch": 2.439817603028478,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005192482462697735,
      "loss": 1.0837,
      "step": 28358
    },
    {
      "epoch": 2.4399036393357996,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005192212255504533,
      "loss": 1.6425,
      "step": 28359
    },
    {
      "epoch": 2.4399896756431216,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005191942047749147,
      "loss": 1.1326,
      "step": 28360
    },
    {
      "epoch": 2.4400757119504433,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005191671839432369,
      "loss": 0.9014,
      "step": 28361
    },
    {
      "epoch": 2.440161748257765,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0005191401630554989,
      "loss": 1.9048,
      "step": 28362
    },
    {
      "epoch": 2.4402477845650865,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005191131421117794,
      "loss": 1.2412,
      "step": 28363
    },
    {
      "epoch": 2.440333820872408,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005190861211121578,
      "loss": 1.6476,
      "step": 28364
    },
    {
      "epoch": 2.4404198571797298,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000519059100056713,
      "loss": 1.8218,
      "step": 28365
    },
    {
      "epoch": 2.4405058934870514,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000519032078945524,
      "loss": 0.9049,
      "step": 28366
    },
    {
      "epoch": 2.4405919297943734,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005190050577786699,
      "loss": 0.4495,
      "step": 28367
    },
    {
      "epoch": 2.440677966101695,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005189780365562296,
      "loss": 1.6935,
      "step": 28368
    },
    {
      "epoch": 2.4407640024090167,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005189510152782825,
      "loss": 1.1617,
      "step": 28369
    },
    {
      "epoch": 2.4408500387163383,
      "grad_norm": 1.0,
      "learning_rate": 0.0005189239939449071,
      "loss": 1.7181,
      "step": 28370
    },
    {
      "epoch": 2.44093607502366,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005188969725561829,
      "loss": 1.2356,
      "step": 28371
    },
    {
      "epoch": 2.4410221113309816,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005188699511121886,
      "loss": 1.3694,
      "step": 28372
    },
    {
      "epoch": 2.441108147638303,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005188429296130033,
      "loss": 1.7531,
      "step": 28373
    },
    {
      "epoch": 2.4411941839456253,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005188159080587063,
      "loss": 1.2426,
      "step": 28374
    },
    {
      "epoch": 2.441280220252947,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005187888864493763,
      "loss": 0.7132,
      "step": 28375
    },
    {
      "epoch": 2.4413662565602685,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005187618647850925,
      "loss": 1.3466,
      "step": 28376
    },
    {
      "epoch": 2.44145229286759,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005187348430659338,
      "loss": 1.5352,
      "step": 28377
    },
    {
      "epoch": 2.4415383291749118,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005187078212919793,
      "loss": 0.8549,
      "step": 28378
    },
    {
      "epoch": 2.4416243654822334,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005186807994633082,
      "loss": 0.8163,
      "step": 28379
    },
    {
      "epoch": 2.441710401789555,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005186537775799993,
      "loss": 1.8457,
      "step": 28380
    },
    {
      "epoch": 2.441796438096877,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005186267556421318,
      "loss": 0.7088,
      "step": 28381
    },
    {
      "epoch": 2.4418824744041987,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005185997336497846,
      "loss": 1.2156,
      "step": 28382
    },
    {
      "epoch": 2.4419685107115203,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005185727116030369,
      "loss": 0.6607,
      "step": 28383
    },
    {
      "epoch": 2.442054547018842,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005185456895019677,
      "loss": 0.8003,
      "step": 28384
    },
    {
      "epoch": 2.4421405833261636,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005185186673466556,
      "loss": 0.9245,
      "step": 28385
    },
    {
      "epoch": 2.442226619633485,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005184916451371804,
      "loss": 1.0159,
      "step": 28386
    },
    {
      "epoch": 2.442312655940807,
      "grad_norm": 0.875,
      "learning_rate": 0.0005184646228736206,
      "loss": 0.5825,
      "step": 28387
    },
    {
      "epoch": 2.442398692248129,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005184376005560553,
      "loss": 1.5113,
      "step": 28388
    },
    {
      "epoch": 2.4424847285554505,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005184105781845636,
      "loss": 0.7025,
      "step": 28389
    },
    {
      "epoch": 2.442570764862772,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005183835557592245,
      "loss": 0.6166,
      "step": 28390
    },
    {
      "epoch": 2.4426568011700938,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005183565332801173,
      "loss": 1.4224,
      "step": 28391
    },
    {
      "epoch": 2.4427428374774154,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005183295107473206,
      "loss": 1.7991,
      "step": 28392
    },
    {
      "epoch": 2.442828873784737,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005183024881609137,
      "loss": 1.2579,
      "step": 28393
    },
    {
      "epoch": 2.4429149100920586,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005182754655209757,
      "loss": 0.8417,
      "step": 28394
    },
    {
      "epoch": 2.4430009463993807,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005182484428275854,
      "loss": 1.1528,
      "step": 28395
    },
    {
      "epoch": 2.4430869827067023,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005182214200808218,
      "loss": 1.2657,
      "step": 28396
    },
    {
      "epoch": 2.443173019014024,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005181943972807643,
      "loss": 0.8609,
      "step": 28397
    },
    {
      "epoch": 2.4432590553213456,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005181673744274918,
      "loss": 1.176,
      "step": 28398
    },
    {
      "epoch": 2.443345091628667,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0005181403515210831,
      "loss": 1.0648,
      "step": 28399
    },
    {
      "epoch": 2.443431127935989,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005181133285616174,
      "loss": 0.965,
      "step": 28400
    },
    {
      "epoch": 2.4435171642433104,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005180863055491738,
      "loss": 0.9787,
      "step": 28401
    },
    {
      "epoch": 2.4436032005506325,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005180592824838313,
      "loss": 1.7803,
      "step": 28402
    },
    {
      "epoch": 2.443689236857954,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005180322593656689,
      "loss": 1.1107,
      "step": 28403
    },
    {
      "epoch": 2.4437752731652758,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005180052361947656,
      "loss": 1.5556,
      "step": 28404
    },
    {
      "epoch": 2.4438613094725974,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005179782129712006,
      "loss": 1.2,
      "step": 28405
    },
    {
      "epoch": 2.443947345779919,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005179511896950526,
      "loss": 0.822,
      "step": 28406
    },
    {
      "epoch": 2.4440333820872406,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005179241663664009,
      "loss": 1.2015,
      "step": 28407
    },
    {
      "epoch": 2.4441194183945623,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005178971429853247,
      "loss": 1.272,
      "step": 28408
    },
    {
      "epoch": 2.4442054547018843,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005178701195519028,
      "loss": 0.9477,
      "step": 28409
    },
    {
      "epoch": 2.444291491009206,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005178430960662142,
      "loss": 1.0373,
      "step": 28410
    },
    {
      "epoch": 2.4443775273165276,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000517816072528338,
      "loss": 1.6538,
      "step": 28411
    },
    {
      "epoch": 2.444463563623849,
      "grad_norm": 1.25,
      "learning_rate": 0.0005177890489383532,
      "loss": 1.8043,
      "step": 28412
    },
    {
      "epoch": 2.444549599931171,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000517762025296339,
      "loss": 0.984,
      "step": 28413
    },
    {
      "epoch": 2.444635636238493,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0005177350016023742,
      "loss": 1.7625,
      "step": 28414
    },
    {
      "epoch": 2.4447216725458145,
      "grad_norm": 1.25,
      "learning_rate": 0.000517707977856538,
      "loss": 1.5532,
      "step": 28415
    },
    {
      "epoch": 2.444807708853136,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005176809540589095,
      "loss": 0.7798,
      "step": 28416
    },
    {
      "epoch": 2.4448937451604578,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005176539302095677,
      "loss": 0.902,
      "step": 28417
    },
    {
      "epoch": 2.4449797814677794,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005176269063085915,
      "loss": 0.6363,
      "step": 28418
    },
    {
      "epoch": 2.445065817775101,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00051759988235606,
      "loss": 0.8305,
      "step": 28419
    },
    {
      "epoch": 2.4451518540824226,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005175728583520523,
      "loss": 1.2862,
      "step": 28420
    },
    {
      "epoch": 2.4452378903897447,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005175458342966475,
      "loss": 1.2046,
      "step": 28421
    },
    {
      "epoch": 2.4453239266970663,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005175188101899243,
      "loss": 1.3445,
      "step": 28422
    },
    {
      "epoch": 2.445409963004388,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005174917860319622,
      "loss": 1.7542,
      "step": 28423
    },
    {
      "epoch": 2.4454959993117096,
      "grad_norm": 1.4375,
      "learning_rate": 0.00051746476182284,
      "loss": 1.6596,
      "step": 28424
    },
    {
      "epoch": 2.445582035619031,
      "grad_norm": 0.90625,
      "learning_rate": 0.0005174377375626368,
      "loss": 1.2316,
      "step": 28425
    },
    {
      "epoch": 2.445668071926353,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005174107132514314,
      "loss": 0.8751,
      "step": 28426
    },
    {
      "epoch": 2.4457541082336745,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005173836888893033,
      "loss": 0.962,
      "step": 28427
    },
    {
      "epoch": 2.4458401445409965,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005173566644763312,
      "loss": 1.2513,
      "step": 28428
    },
    {
      "epoch": 2.445926180848318,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005173296400125942,
      "loss": 1.1835,
      "step": 28429
    },
    {
      "epoch": 2.4460122171556398,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0005173026154981714,
      "loss": 0.7159,
      "step": 28430
    },
    {
      "epoch": 2.4460982534629614,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005172755909331418,
      "loss": 0.6706,
      "step": 28431
    },
    {
      "epoch": 2.446184289770283,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005172485663175845,
      "loss": 1.2972,
      "step": 28432
    },
    {
      "epoch": 2.4462703260776046,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005172215416515784,
      "loss": 1.4722,
      "step": 28433
    },
    {
      "epoch": 2.4463563623849263,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005171945169352026,
      "loss": 1.322,
      "step": 28434
    },
    {
      "epoch": 2.4464423986922483,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005171674921685363,
      "loss": 0.9831,
      "step": 28435
    },
    {
      "epoch": 2.44652843499957,
      "grad_norm": 1.125,
      "learning_rate": 0.0005171404673516586,
      "loss": 1.4554,
      "step": 28436
    },
    {
      "epoch": 2.4466144713068916,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000517113442484648,
      "loss": 0.6292,
      "step": 28437
    },
    {
      "epoch": 2.446700507614213,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005170864175675841,
      "loss": 0.9655,
      "step": 28438
    },
    {
      "epoch": 2.446786543921535,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005170593926005458,
      "loss": 1.3985,
      "step": 28439
    },
    {
      "epoch": 2.4468725802288565,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000517032367583612,
      "loss": 1.1177,
      "step": 28440
    },
    {
      "epoch": 2.446958616536178,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005170053425168618,
      "loss": 1.6494,
      "step": 28441
    },
    {
      "epoch": 2.4470446528435,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005169783174003744,
      "loss": 0.8835,
      "step": 28442
    },
    {
      "epoch": 2.4471306891508218,
      "grad_norm": 1.625,
      "learning_rate": 0.0005169512922342288,
      "loss": 1.566,
      "step": 28443
    },
    {
      "epoch": 2.4472167254581434,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005169242670185039,
      "loss": 1.0485,
      "step": 28444
    },
    {
      "epoch": 2.447302761765465,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005168972417532785,
      "loss": 1.5572,
      "step": 28445
    },
    {
      "epoch": 2.4473887980727866,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005168702164386323,
      "loss": 0.8567,
      "step": 28446
    },
    {
      "epoch": 2.4474748343801083,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000516843191074644,
      "loss": 1.2999,
      "step": 28447
    },
    {
      "epoch": 2.44756087068743,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005168161656613924,
      "loss": 0.7141,
      "step": 28448
    },
    {
      "epoch": 2.447646906994752,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000516789140198957,
      "loss": 0.7951,
      "step": 28449
    },
    {
      "epoch": 2.4477329433020736,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005167621146874166,
      "loss": 1.0836,
      "step": 28450
    },
    {
      "epoch": 2.447818979609395,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005167350891268503,
      "loss": 0.743,
      "step": 28451
    },
    {
      "epoch": 2.447905015916717,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005167080635173368,
      "loss": 0.9105,
      "step": 28452
    },
    {
      "epoch": 2.4479910522240385,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005166810378589558,
      "loss": 1.3257,
      "step": 28453
    },
    {
      "epoch": 2.44807708853136,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005166540121517859,
      "loss": 0.7778,
      "step": 28454
    },
    {
      "epoch": 2.4481631248386817,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005166269863959063,
      "loss": 1.4159,
      "step": 28455
    },
    {
      "epoch": 2.4482491611460038,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005165999605913959,
      "loss": 0.9923,
      "step": 28456
    },
    {
      "epoch": 2.4483351974533254,
      "grad_norm": 1.1875,
      "learning_rate": 0.000516572934738334,
      "loss": 1.4171,
      "step": 28457
    },
    {
      "epoch": 2.448421233760647,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005165459088367994,
      "loss": 1.1804,
      "step": 28458
    },
    {
      "epoch": 2.4485072700679686,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005165188828868713,
      "loss": 1.2665,
      "step": 28459
    },
    {
      "epoch": 2.4485933063752903,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005164918568886285,
      "loss": 1.1118,
      "step": 28460
    },
    {
      "epoch": 2.448679342682612,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005164648308421504,
      "loss": 0.7873,
      "step": 28461
    },
    {
      "epoch": 2.4487653789899335,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005164378047475159,
      "loss": 2.213,
      "step": 28462
    },
    {
      "epoch": 2.4488514152972556,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005164107786048039,
      "loss": 1.5195,
      "step": 28463
    },
    {
      "epoch": 2.448937451604577,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005163837524140936,
      "loss": 1.0738,
      "step": 28464
    },
    {
      "epoch": 2.449023487911899,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000516356726175464,
      "loss": 0.665,
      "step": 28465
    },
    {
      "epoch": 2.4491095242192205,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005163296998889942,
      "loss": 1.8572,
      "step": 28466
    },
    {
      "epoch": 2.449195560526542,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005163026735547632,
      "loss": 1.2549,
      "step": 28467
    },
    {
      "epoch": 2.4492815968338637,
      "grad_norm": 1.515625,
      "learning_rate": 0.00051627564717285,
      "loss": 1.36,
      "step": 28468
    },
    {
      "epoch": 2.4493676331411858,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005162486207433338,
      "loss": 0.9259,
      "step": 28469
    },
    {
      "epoch": 2.4494536694485074,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005162215942662935,
      "loss": 1.2337,
      "step": 28470
    },
    {
      "epoch": 2.449539705755829,
      "grad_norm": 1.140625,
      "learning_rate": 0.000516194567741808,
      "loss": 1.1436,
      "step": 28471
    },
    {
      "epoch": 2.4496257420631506,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005161675411699567,
      "loss": 1.1182,
      "step": 28472
    },
    {
      "epoch": 2.4497117783704723,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005161405145508185,
      "loss": 1.6307,
      "step": 28473
    },
    {
      "epoch": 2.449797814677794,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005161134878844725,
      "loss": 0.6628,
      "step": 28474
    },
    {
      "epoch": 2.449883850985116,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005160864611709976,
      "loss": 0.6412,
      "step": 28475
    },
    {
      "epoch": 2.4499698872924376,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005160594344104729,
      "loss": 1.2512,
      "step": 28476
    },
    {
      "epoch": 2.450055923599759,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005160324076029776,
      "loss": 0.9318,
      "step": 28477
    },
    {
      "epoch": 2.450141959907081,
      "grad_norm": 1.125,
      "learning_rate": 0.0005160053807485904,
      "loss": 1.044,
      "step": 28478
    },
    {
      "epoch": 2.4502279962144025,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005159783538473908,
      "loss": 0.8063,
      "step": 28479
    },
    {
      "epoch": 2.450314032521724,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005159513268994575,
      "loss": 1.2932,
      "step": 28480
    },
    {
      "epoch": 2.4504000688290457,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0005159242999048698,
      "loss": 0.7601,
      "step": 28481
    },
    {
      "epoch": 2.4504861051363678,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005158972728637064,
      "loss": 1.7919,
      "step": 28482
    },
    {
      "epoch": 2.4505721414436894,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005158702457760468,
      "loss": 0.949,
      "step": 28483
    },
    {
      "epoch": 2.450658177751011,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005158432186419698,
      "loss": 1.1919,
      "step": 28484
    },
    {
      "epoch": 2.4507442140583326,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005158161914615542,
      "loss": 1.227,
      "step": 28485
    },
    {
      "epoch": 2.4508302503656543,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005157891642348795,
      "loss": 0.6397,
      "step": 28486
    },
    {
      "epoch": 2.450916286672976,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005157621369620246,
      "loss": 0.7989,
      "step": 28487
    },
    {
      "epoch": 2.4510023229802975,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005157351096430685,
      "loss": 0.7051,
      "step": 28488
    },
    {
      "epoch": 2.4510883592876196,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005157080822780902,
      "loss": 0.585,
      "step": 28489
    },
    {
      "epoch": 2.451174395594941,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005156810548671688,
      "loss": 1.3999,
      "step": 28490
    },
    {
      "epoch": 2.451260431902263,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005156540274103835,
      "loss": 1.5668,
      "step": 28491
    },
    {
      "epoch": 2.4513464682095845,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005156269999078132,
      "loss": 0.4539,
      "step": 28492
    },
    {
      "epoch": 2.451432504516906,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0005155999723595367,
      "loss": 1.0332,
      "step": 28493
    },
    {
      "epoch": 2.4515185408242277,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005155729447656335,
      "loss": 1.4086,
      "step": 28494
    },
    {
      "epoch": 2.4516045771315493,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005155459171261826,
      "loss": 2.0279,
      "step": 28495
    },
    {
      "epoch": 2.4516906134388714,
      "grad_norm": 2.234375,
      "learning_rate": 0.0005155188894412628,
      "loss": 1.1072,
      "step": 28496
    },
    {
      "epoch": 2.451776649746193,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005154918617109532,
      "loss": 1.4361,
      "step": 28497
    },
    {
      "epoch": 2.4518626860535147,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000515464833935333,
      "loss": 1.5009,
      "step": 28498
    },
    {
      "epoch": 2.4519487223608363,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005154378061144813,
      "loss": 0.5942,
      "step": 28499
    },
    {
      "epoch": 2.452034758668158,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005154107782484768,
      "loss": 1.1103,
      "step": 28500
    },
    {
      "epoch": 2.4521207949754795,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005153837503373987,
      "loss": 1.0504,
      "step": 28501
    },
    {
      "epoch": 2.452206831282801,
      "grad_norm": 0.875,
      "learning_rate": 0.0005153567223813263,
      "loss": 0.9969,
      "step": 28502
    },
    {
      "epoch": 2.452292867590123,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005153296943803385,
      "loss": 1.0667,
      "step": 28503
    },
    {
      "epoch": 2.452378903897445,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005153026663345142,
      "loss": 1.1251,
      "step": 28504
    },
    {
      "epoch": 2.4524649402047665,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005152756382439327,
      "loss": 0.8702,
      "step": 28505
    },
    {
      "epoch": 2.452550976512088,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005152486101086728,
      "loss": 0.9709,
      "step": 28506
    },
    {
      "epoch": 2.4526370128194097,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005152215819288138,
      "loss": 1.6834,
      "step": 28507
    },
    {
      "epoch": 2.4527230491267313,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005151945537044345,
      "loss": 1.3763,
      "step": 28508
    },
    {
      "epoch": 2.452809085434053,
      "grad_norm": 0.953125,
      "learning_rate": 0.000515167525435614,
      "loss": 0.9243,
      "step": 28509
    },
    {
      "epoch": 2.452895121741375,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005151404971224317,
      "loss": 1.6473,
      "step": 28510
    },
    {
      "epoch": 2.4529811580486967,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005151134687649664,
      "loss": 1.5254,
      "step": 28511
    },
    {
      "epoch": 2.4530671943560183,
      "grad_norm": 1.15625,
      "learning_rate": 0.000515086440363297,
      "loss": 2.0723,
      "step": 28512
    },
    {
      "epoch": 2.45315323066334,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005150594119175027,
      "loss": 1.6247,
      "step": 28513
    },
    {
      "epoch": 2.4532392669706615,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005150323834276627,
      "loss": 1.0636,
      "step": 28514
    },
    {
      "epoch": 2.453325303277983,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005150053548938556,
      "loss": 0.7396,
      "step": 28515
    },
    {
      "epoch": 2.4534113395853048,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005149783263161611,
      "loss": 0.7089,
      "step": 28516
    },
    {
      "epoch": 2.453497375892627,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005149512976946577,
      "loss": 1.4475,
      "step": 28517
    },
    {
      "epoch": 2.4535834121999485,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005149242690294247,
      "loss": 1.4022,
      "step": 28518
    },
    {
      "epoch": 2.45366944850727,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005148972403205411,
      "loss": 1.1614,
      "step": 28519
    },
    {
      "epoch": 2.4537554848145917,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000514870211568086,
      "loss": 1.6315,
      "step": 28520
    },
    {
      "epoch": 2.4538415211219133,
      "grad_norm": 0.875,
      "learning_rate": 0.0005148431827721385,
      "loss": 0.89,
      "step": 28521
    },
    {
      "epoch": 2.453927557429235,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005148161539327775,
      "loss": 1.0517,
      "step": 28522
    },
    {
      "epoch": 2.4540135937365566,
      "grad_norm": 1.09375,
      "learning_rate": 0.000514789125050082,
      "loss": 0.7161,
      "step": 28523
    },
    {
      "epoch": 2.4540996300438787,
      "grad_norm": 1.71875,
      "learning_rate": 0.0005147620961241314,
      "loss": 0.998,
      "step": 28524
    },
    {
      "epoch": 2.4541856663512003,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005147350671550044,
      "loss": 0.7863,
      "step": 28525
    },
    {
      "epoch": 2.454271702658522,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005147080381427802,
      "loss": 1.74,
      "step": 28526
    },
    {
      "epoch": 2.4543577389658435,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005146810090875379,
      "loss": 1.1115,
      "step": 28527
    },
    {
      "epoch": 2.454443775273165,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005146539799893565,
      "loss": 0.6131,
      "step": 28528
    },
    {
      "epoch": 2.4545298115804868,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005146269508483151,
      "loss": 1.1734,
      "step": 28529
    },
    {
      "epoch": 2.454615847887809,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005145999216644927,
      "loss": 1.6398,
      "step": 28530
    },
    {
      "epoch": 2.4547018841951305,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005145728924379683,
      "loss": 1.7017,
      "step": 28531
    },
    {
      "epoch": 2.454787920502452,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000514545863168821,
      "loss": 1.3773,
      "step": 28532
    },
    {
      "epoch": 2.4548739568097737,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005145188338571301,
      "loss": 1.7244,
      "step": 28533
    },
    {
      "epoch": 2.4549599931170953,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005144918045029742,
      "loss": 1.0789,
      "step": 28534
    },
    {
      "epoch": 2.455046029424417,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005144647751064327,
      "loss": 1.8888,
      "step": 28535
    },
    {
      "epoch": 2.455132065731739,
      "grad_norm": 1.25,
      "learning_rate": 0.0005144377456675845,
      "loss": 1.0069,
      "step": 28536
    },
    {
      "epoch": 2.4552181020390607,
      "grad_norm": 1.125,
      "learning_rate": 0.0005144107161865087,
      "loss": 1.0707,
      "step": 28537
    },
    {
      "epoch": 2.4553041383463823,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005143836866632843,
      "loss": 0.9251,
      "step": 28538
    },
    {
      "epoch": 2.455390174653704,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005143566570979905,
      "loss": 1.1528,
      "step": 28539
    },
    {
      "epoch": 2.4554762109610255,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005143296274907063,
      "loss": 1.1746,
      "step": 28540
    },
    {
      "epoch": 2.455562247268347,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005143025978415107,
      "loss": 0.5167,
      "step": 28541
    },
    {
      "epoch": 2.4556482835756688,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005142755681504827,
      "loss": 1.3008,
      "step": 28542
    },
    {
      "epoch": 2.455734319882991,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005142485384177016,
      "loss": 1.2412,
      "step": 28543
    },
    {
      "epoch": 2.4558203561903125,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005142215086432462,
      "loss": 1.0722,
      "step": 28544
    },
    {
      "epoch": 2.455906392497634,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005141944788271956,
      "loss": 1.168,
      "step": 28545
    },
    {
      "epoch": 2.4559924288049557,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005141674489696289,
      "loss": 1.2236,
      "step": 28546
    },
    {
      "epoch": 2.4560784651122773,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005141404190706253,
      "loss": 1.0671,
      "step": 28547
    },
    {
      "epoch": 2.456164501419599,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005141133891302637,
      "loss": 1.9872,
      "step": 28548
    },
    {
      "epoch": 2.4562505377269206,
      "grad_norm": 1.265625,
      "learning_rate": 0.000514086359148623,
      "loss": 1.3556,
      "step": 28549
    },
    {
      "epoch": 2.4563365740342427,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005140593291257827,
      "loss": 0.5974,
      "step": 28550
    },
    {
      "epoch": 2.4564226103415643,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005140322990618215,
      "loss": 0.9401,
      "step": 28551
    },
    {
      "epoch": 2.456508646648886,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005140052689568185,
      "loss": 1.4128,
      "step": 28552
    },
    {
      "epoch": 2.4565946829562075,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005139782388108528,
      "loss": 1.3565,
      "step": 28553
    },
    {
      "epoch": 2.456680719263529,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005139512086240036,
      "loss": 1.4393,
      "step": 28554
    },
    {
      "epoch": 2.456766755570851,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005139241783963498,
      "loss": 1.4154,
      "step": 28555
    },
    {
      "epoch": 2.4568527918781724,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005138971481279704,
      "loss": 1.6615,
      "step": 28556
    },
    {
      "epoch": 2.4569388281854945,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005138701178189444,
      "loss": 0.3581,
      "step": 28557
    },
    {
      "epoch": 2.457024864492816,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005138430874693513,
      "loss": 1.4711,
      "step": 28558
    },
    {
      "epoch": 2.4571109008001377,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005138160570792698,
      "loss": 1.1635,
      "step": 28559
    },
    {
      "epoch": 2.4571969371074593,
      "grad_norm": 1.859375,
      "learning_rate": 0.0005137890266487788,
      "loss": 1.3416,
      "step": 28560
    },
    {
      "epoch": 2.457282973414781,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005137619961779578,
      "loss": 1.3323,
      "step": 28561
    },
    {
      "epoch": 2.4573690097221026,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005137349656668855,
      "loss": 1.3457,
      "step": 28562
    },
    {
      "epoch": 2.457455046029424,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005137079351156411,
      "loss": 1.3314,
      "step": 28563
    },
    {
      "epoch": 2.4575410823367463,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005136809045243037,
      "loss": 1.4924,
      "step": 28564
    },
    {
      "epoch": 2.457627118644068,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005136538738929523,
      "loss": 0.7171,
      "step": 28565
    },
    {
      "epoch": 2.4577131549513895,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005136268432216662,
      "loss": 1.1163,
      "step": 28566
    },
    {
      "epoch": 2.457799191258711,
      "grad_norm": 1.109375,
      "learning_rate": 0.000513599812510524,
      "loss": 1.4191,
      "step": 28567
    },
    {
      "epoch": 2.457885227566033,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000513572781759605,
      "loss": 1.1073,
      "step": 28568
    },
    {
      "epoch": 2.4579712638733544,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005135457509689884,
      "loss": 1.6737,
      "step": 28569
    },
    {
      "epoch": 2.458057300180676,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000513518720138753,
      "loss": 1.0363,
      "step": 28570
    },
    {
      "epoch": 2.458143336487998,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000513491689268978,
      "loss": 2.2148,
      "step": 28571
    },
    {
      "epoch": 2.4582293727953197,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005134646583597423,
      "loss": 0.5577,
      "step": 28572
    },
    {
      "epoch": 2.4583154091026413,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005134376274111253,
      "loss": 0.6428,
      "step": 28573
    },
    {
      "epoch": 2.458401445409963,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005134105964232058,
      "loss": 0.9857,
      "step": 28574
    },
    {
      "epoch": 2.4584874817172846,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005133835653960628,
      "loss": 1.225,
      "step": 28575
    },
    {
      "epoch": 2.458573518024606,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005133565343297756,
      "loss": 1.4031,
      "step": 28576
    },
    {
      "epoch": 2.458659554331928,
      "grad_norm": 1.015625,
      "learning_rate": 0.000513329503224423,
      "loss": 1.6489,
      "step": 28577
    },
    {
      "epoch": 2.45874559063925,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005133024720800844,
      "loss": 1.2291,
      "step": 28578
    },
    {
      "epoch": 2.4588316269465715,
      "grad_norm": 0.875,
      "learning_rate": 0.0005132754408968385,
      "loss": 0.9416,
      "step": 28579
    },
    {
      "epoch": 2.458917663253893,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005132484096747645,
      "loss": 0.8007,
      "step": 28580
    },
    {
      "epoch": 2.459003699561215,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005132213784139415,
      "loss": 1.235,
      "step": 28581
    },
    {
      "epoch": 2.4590897358685364,
      "grad_norm": 1.6875,
      "learning_rate": 0.0005131943471144485,
      "loss": 1.3442,
      "step": 28582
    },
    {
      "epoch": 2.459175772175858,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005131673157763646,
      "loss": 1.3797,
      "step": 28583
    },
    {
      "epoch": 2.4592618084831797,
      "grad_norm": 0.984375,
      "learning_rate": 0.000513140284399769,
      "loss": 0.9736,
      "step": 28584
    },
    {
      "epoch": 2.4593478447905017,
      "grad_norm": 2.796875,
      "learning_rate": 0.0005131132529847406,
      "loss": 1.0109,
      "step": 28585
    },
    {
      "epoch": 2.4594338810978233,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005130862215313585,
      "loss": 0.6769,
      "step": 28586
    },
    {
      "epoch": 2.459519917405145,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005130591900397016,
      "loss": 1.2295,
      "step": 28587
    },
    {
      "epoch": 2.4596059537124666,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005130321585098492,
      "loss": 0.9506,
      "step": 28588
    },
    {
      "epoch": 2.459691990019788,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005130051269418804,
      "loss": 1.4889,
      "step": 28589
    },
    {
      "epoch": 2.4597780263271103,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005129780953358739,
      "loss": 1.1167,
      "step": 28590
    },
    {
      "epoch": 2.459864062634432,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005129510636919092,
      "loss": 1.2204,
      "step": 28591
    },
    {
      "epoch": 2.4599500989417535,
      "grad_norm": 1.03125,
      "learning_rate": 0.000512924032010065,
      "loss": 1.4404,
      "step": 28592
    },
    {
      "epoch": 2.460036135249075,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005128970002904207,
      "loss": 1.0694,
      "step": 28593
    },
    {
      "epoch": 2.460122171556397,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005128699685330549,
      "loss": 1.4739,
      "step": 28594
    },
    {
      "epoch": 2.4602082078637184,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005128429367380472,
      "loss": 1.0691,
      "step": 28595
    },
    {
      "epoch": 2.46029424417104,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005128159049054762,
      "loss": 1.5885,
      "step": 28596
    },
    {
      "epoch": 2.460380280478362,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005127888730354213,
      "loss": 1.0319,
      "step": 28597
    },
    {
      "epoch": 2.4604663167856837,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005127618411279615,
      "loss": 1.1533,
      "step": 28598
    },
    {
      "epoch": 2.4605523530930054,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005127348091831755,
      "loss": 1.2085,
      "step": 28599
    },
    {
      "epoch": 2.460638389400327,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005127077772011429,
      "loss": 0.8314,
      "step": 28600
    },
    {
      "epoch": 2.4607244257076486,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005126807451819426,
      "loss": 1.6874,
      "step": 28601
    },
    {
      "epoch": 2.4608104620149702,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005126537131256533,
      "loss": 1.654,
      "step": 28602
    },
    {
      "epoch": 2.460896498322292,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005126266810323546,
      "loss": 1.5794,
      "step": 28603
    },
    {
      "epoch": 2.460982534629614,
      "grad_norm": 0.875,
      "learning_rate": 0.0005125996489021253,
      "loss": 1.4395,
      "step": 28604
    },
    {
      "epoch": 2.4610685709369355,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005125726167350444,
      "loss": 0.6671,
      "step": 28605
    },
    {
      "epoch": 2.461154607244257,
      "grad_norm": 0.96875,
      "learning_rate": 0.000512545584531191,
      "loss": 1.3056,
      "step": 28606
    },
    {
      "epoch": 2.461240643551579,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005125185522906443,
      "loss": 0.8154,
      "step": 28607
    },
    {
      "epoch": 2.4613266798589004,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005124915200134833,
      "loss": 0.8445,
      "step": 28608
    },
    {
      "epoch": 2.461412716166222,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005124644876997869,
      "loss": 1.5293,
      "step": 28609
    },
    {
      "epoch": 2.4614987524735437,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005124374553496344,
      "loss": 1.2736,
      "step": 28610
    },
    {
      "epoch": 2.4615847887808657,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0005124104229631048,
      "loss": 1.1205,
      "step": 28611
    },
    {
      "epoch": 2.4616708250881874,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005123833905402771,
      "loss": 1.1377,
      "step": 28612
    },
    {
      "epoch": 2.461756861395509,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005123563580812303,
      "loss": 1.347,
      "step": 28613
    },
    {
      "epoch": 2.4618428977028306,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005123293255860436,
      "loss": 0.9414,
      "step": 28614
    },
    {
      "epoch": 2.4619289340101522,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000512302293054796,
      "loss": 1.3453,
      "step": 28615
    },
    {
      "epoch": 2.462014970317474,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005122752604875667,
      "loss": 0.7696,
      "step": 28616
    },
    {
      "epoch": 2.4621010066247955,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0005122482278844345,
      "loss": 1.3517,
      "step": 28617
    },
    {
      "epoch": 2.4621870429321175,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005122211952454786,
      "loss": 0.73,
      "step": 28618
    },
    {
      "epoch": 2.462273079239439,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005121941625707782,
      "loss": 0.9484,
      "step": 28619
    },
    {
      "epoch": 2.462359115546761,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005121671298604122,
      "loss": 2.0166,
      "step": 28620
    },
    {
      "epoch": 2.4624451518540824,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005121400971144596,
      "loss": 0.8996,
      "step": 28621
    },
    {
      "epoch": 2.462531188161404,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005121130643329999,
      "loss": 1.4567,
      "step": 28622
    },
    {
      "epoch": 2.4626172244687257,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005120860315161116,
      "loss": 1.0517,
      "step": 28623
    },
    {
      "epoch": 2.4627032607760473,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000512058998663874,
      "loss": 0.9345,
      "step": 28624
    },
    {
      "epoch": 2.4627892970833694,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005120319657763662,
      "loss": 1.1447,
      "step": 28625
    },
    {
      "epoch": 2.462875333390691,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005120049328536673,
      "loss": 1.5207,
      "step": 28626
    },
    {
      "epoch": 2.4629613696980126,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005119778998958563,
      "loss": 0.9799,
      "step": 28627
    },
    {
      "epoch": 2.4630474060053342,
      "grad_norm": 1.65625,
      "learning_rate": 0.0005119508669030122,
      "loss": 0.8231,
      "step": 28628
    },
    {
      "epoch": 2.463133442312656,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005119238338752142,
      "loss": 1.1939,
      "step": 28629
    },
    {
      "epoch": 2.4632194786199775,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005118968008125414,
      "loss": 1.4225,
      "step": 28630
    },
    {
      "epoch": 2.463305514927299,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005118697677150725,
      "loss": 0.7241,
      "step": 28631
    },
    {
      "epoch": 2.463391551234621,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000511842734582887,
      "loss": 0.8059,
      "step": 28632
    },
    {
      "epoch": 2.463477587541943,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005118157014160638,
      "loss": 0.8698,
      "step": 28633
    },
    {
      "epoch": 2.4635636238492644,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005117886682146819,
      "loss": 0.9597,
      "step": 28634
    },
    {
      "epoch": 2.463649660156586,
      "grad_norm": 1.0,
      "learning_rate": 0.0005117616349788205,
      "loss": 1.614,
      "step": 28635
    },
    {
      "epoch": 2.4637356964639077,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005117346017085585,
      "loss": 1.1109,
      "step": 28636
    },
    {
      "epoch": 2.4638217327712293,
      "grad_norm": 0.84375,
      "learning_rate": 0.0005117075684039753,
      "loss": 1.1673,
      "step": 28637
    },
    {
      "epoch": 2.463907769078551,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005116805350651497,
      "loss": 1.0835,
      "step": 28638
    },
    {
      "epoch": 2.463993805385873,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005116535016921606,
      "loss": 0.5429,
      "step": 28639
    },
    {
      "epoch": 2.4640798416931946,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005116264682850874,
      "loss": 1.0477,
      "step": 28640
    },
    {
      "epoch": 2.4641658780005162,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005115994348440091,
      "loss": 1.0751,
      "step": 28641
    },
    {
      "epoch": 2.464251914307838,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005115724013690047,
      "loss": 1.0903,
      "step": 28642
    },
    {
      "epoch": 2.4643379506151595,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005115453678601531,
      "loss": 1.168,
      "step": 28643
    },
    {
      "epoch": 2.464423986922481,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005115183343175337,
      "loss": 1.1378,
      "step": 28644
    },
    {
      "epoch": 2.464510023229803,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005114913007412255,
      "loss": 1.0879,
      "step": 28645
    },
    {
      "epoch": 2.464596059537125,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005114642671313072,
      "loss": 0.5282,
      "step": 28646
    },
    {
      "epoch": 2.4646820958444464,
      "grad_norm": 1.671875,
      "learning_rate": 0.0005114372334878584,
      "loss": 1.4641,
      "step": 28647
    },
    {
      "epoch": 2.464768132151768,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0005114101998109578,
      "loss": 1.1516,
      "step": 28648
    },
    {
      "epoch": 2.4648541684590897,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005113831661006847,
      "loss": 1.2233,
      "step": 28649
    },
    {
      "epoch": 2.4649402047664113,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005113561323571179,
      "loss": 1.4289,
      "step": 28650
    },
    {
      "epoch": 2.4650262410737334,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005113290985803367,
      "loss": 1.2429,
      "step": 28651
    },
    {
      "epoch": 2.465112277381055,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005113020647704202,
      "loss": 1.9855,
      "step": 28652
    },
    {
      "epoch": 2.4651983136883766,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005112750309274472,
      "loss": 0.9639,
      "step": 28653
    },
    {
      "epoch": 2.4652843499956982,
      "grad_norm": 1.0625,
      "learning_rate": 0.000511247997051497,
      "loss": 1.7479,
      "step": 28654
    },
    {
      "epoch": 2.46537038630302,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005112209631426485,
      "loss": 1.8219,
      "step": 28655
    },
    {
      "epoch": 2.4654564226103415,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005111939292009809,
      "loss": 1.9267,
      "step": 28656
    },
    {
      "epoch": 2.465542458917663,
      "grad_norm": 1.125,
      "learning_rate": 0.0005111668952265732,
      "loss": 1.2703,
      "step": 28657
    },
    {
      "epoch": 2.465628495224985,
      "grad_norm": 1.75,
      "learning_rate": 0.0005111398612195045,
      "loss": 1.3337,
      "step": 28658
    },
    {
      "epoch": 2.465714531532307,
      "grad_norm": 1.4375,
      "learning_rate": 0.000511112827179854,
      "loss": 1.3777,
      "step": 28659
    },
    {
      "epoch": 2.4658005678396284,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0005110857931077006,
      "loss": 1.7568,
      "step": 28660
    },
    {
      "epoch": 2.46588660414695,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0005110587590031234,
      "loss": 1.3175,
      "step": 28661
    },
    {
      "epoch": 2.4659726404542717,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005110317248662015,
      "loss": 0.96,
      "step": 28662
    },
    {
      "epoch": 2.4660586767615933,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005110046906970139,
      "loss": 1.1863,
      "step": 28663
    },
    {
      "epoch": 2.466144713068915,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005109776564956398,
      "loss": 1.5311,
      "step": 28664
    },
    {
      "epoch": 2.466230749376237,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005109506222621581,
      "loss": 1.5219,
      "step": 28665
    },
    {
      "epoch": 2.4663167856835586,
      "grad_norm": 1.375,
      "learning_rate": 0.0005109235879966479,
      "loss": 1.2925,
      "step": 28666
    },
    {
      "epoch": 2.4664028219908802,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005108965536991885,
      "loss": 1.9142,
      "step": 28667
    },
    {
      "epoch": 2.466488858298202,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005108695193698588,
      "loss": 1.1641,
      "step": 28668
    },
    {
      "epoch": 2.4665748946055235,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005108424850087377,
      "loss": 1.6949,
      "step": 28669
    },
    {
      "epoch": 2.466660930912845,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005108154506159045,
      "loss": 1.6367,
      "step": 28670
    },
    {
      "epoch": 2.4667469672201667,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005107884161914384,
      "loss": 1.0833,
      "step": 28671
    },
    {
      "epoch": 2.466833003527489,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000510761381735418,
      "loss": 0.8303,
      "step": 28672
    },
    {
      "epoch": 2.4669190398348104,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005107343472479227,
      "loss": 1.1694,
      "step": 28673
    },
    {
      "epoch": 2.467005076142132,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005107073127290316,
      "loss": 0.6651,
      "step": 28674
    },
    {
      "epoch": 2.4670911124494537,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005106802781788236,
      "loss": 1.6253,
      "step": 28675
    },
    {
      "epoch": 2.4671771487567753,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005106532435973781,
      "loss": 1.3694,
      "step": 28676
    },
    {
      "epoch": 2.467263185064097,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005106262089847737,
      "loss": 1.6863,
      "step": 28677
    },
    {
      "epoch": 2.4673492213714185,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005105991743410898,
      "loss": 1.0353,
      "step": 28678
    },
    {
      "epoch": 2.4674352576787406,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005105721396664054,
      "loss": 1.0912,
      "step": 28679
    },
    {
      "epoch": 2.4675212939860622,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005105451049607994,
      "loss": 0.7865,
      "step": 28680
    },
    {
      "epoch": 2.467607330293384,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005105180702243512,
      "loss": 1.1305,
      "step": 28681
    },
    {
      "epoch": 2.4676933666007055,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005104910354571397,
      "loss": 1.6743,
      "step": 28682
    },
    {
      "epoch": 2.467779402908027,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005104640006592439,
      "loss": 1.4491,
      "step": 28683
    },
    {
      "epoch": 2.4678654392153487,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005104369658307428,
      "loss": 1.1042,
      "step": 28684
    },
    {
      "epoch": 2.4679514755226704,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005104099309717158,
      "loss": 1.1519,
      "step": 28685
    },
    {
      "epoch": 2.4680375118299924,
      "grad_norm": 1.125,
      "learning_rate": 0.0005103828960822418,
      "loss": 1.0507,
      "step": 28686
    },
    {
      "epoch": 2.468123548137314,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005103558611623996,
      "loss": 0.8902,
      "step": 28687
    },
    {
      "epoch": 2.4682095844446357,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005103288262122687,
      "loss": 1.27,
      "step": 28688
    },
    {
      "epoch": 2.4682956207519573,
      "grad_norm": 1.140625,
      "learning_rate": 0.000510301791231928,
      "loss": 0.8578,
      "step": 28689
    },
    {
      "epoch": 2.468381657059279,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005102747562214566,
      "loss": 1.3164,
      "step": 28690
    },
    {
      "epoch": 2.4684676933666005,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005102477211809334,
      "loss": 0.8267,
      "step": 28691
    },
    {
      "epoch": 2.468553729673922,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005102206861104376,
      "loss": 1.7492,
      "step": 28692
    },
    {
      "epoch": 2.4686397659812442,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005101936510100483,
      "loss": 1.6607,
      "step": 28693
    },
    {
      "epoch": 2.468725802288566,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005101666158798447,
      "loss": 1.2882,
      "step": 28694
    },
    {
      "epoch": 2.4688118385958875,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005101395807199055,
      "loss": 1.3459,
      "step": 28695
    },
    {
      "epoch": 2.468897874903209,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005101125455303101,
      "loss": 1.9074,
      "step": 28696
    },
    {
      "epoch": 2.4689839112105307,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005100855103111376,
      "loss": 0.6887,
      "step": 28697
    },
    {
      "epoch": 2.4690699475178524,
      "grad_norm": 1.421875,
      "learning_rate": 0.000510058475062467,
      "loss": 0.7969,
      "step": 28698
    },
    {
      "epoch": 2.469155983825174,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005100314397843771,
      "loss": 0.4875,
      "step": 28699
    },
    {
      "epoch": 2.469242020132496,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005100044044769472,
      "loss": 0.9226,
      "step": 28700
    },
    {
      "epoch": 2.4693280564398177,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005099773691402563,
      "loss": 0.6001,
      "step": 28701
    },
    {
      "epoch": 2.4694140927471393,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005099503337743837,
      "loss": 1.4471,
      "step": 28702
    },
    {
      "epoch": 2.469500129054461,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005099232983794083,
      "loss": 0.8709,
      "step": 28703
    },
    {
      "epoch": 2.4695861653617825,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005098962629554091,
      "loss": 1.1915,
      "step": 28704
    },
    {
      "epoch": 2.469672201669104,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005098692275024655,
      "loss": 1.2356,
      "step": 28705
    },
    {
      "epoch": 2.4697582379764262,
      "grad_norm": 0.921875,
      "learning_rate": 0.0005098421920206559,
      "loss": 1.183,
      "step": 28706
    },
    {
      "epoch": 2.469844274283748,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005098151565100602,
      "loss": 0.6261,
      "step": 28707
    },
    {
      "epoch": 2.4699303105910695,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005097881209707569,
      "loss": 1.7072,
      "step": 28708
    },
    {
      "epoch": 2.470016346898391,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005097610854028253,
      "loss": 1.3533,
      "step": 28709
    },
    {
      "epoch": 2.4701023832057127,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005097340498063441,
      "loss": 0.814,
      "step": 28710
    },
    {
      "epoch": 2.4701884195130344,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005097070141813931,
      "loss": 0.7386,
      "step": 28711
    },
    {
      "epoch": 2.4702744558203564,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005096799785280508,
      "loss": 1.2981,
      "step": 28712
    },
    {
      "epoch": 2.470360492127678,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005096529428463965,
      "loss": 1.186,
      "step": 28713
    },
    {
      "epoch": 2.4704465284349997,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005096259071365091,
      "loss": 1.384,
      "step": 28714
    },
    {
      "epoch": 2.4705325647423213,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000509598871398468,
      "loss": 0.9645,
      "step": 28715
    },
    {
      "epoch": 2.470618601049643,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000509571835632352,
      "loss": 2.0541,
      "step": 28716
    },
    {
      "epoch": 2.4707046373569646,
      "grad_norm": 0.984375,
      "learning_rate": 0.0005095447998382402,
      "loss": 0.8646,
      "step": 28717
    },
    {
      "epoch": 2.470790673664286,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005095177640162118,
      "loss": 1.3395,
      "step": 28718
    },
    {
      "epoch": 2.4708767099716082,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005094907281663457,
      "loss": 0.8625,
      "step": 28719
    },
    {
      "epoch": 2.47096274627893,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005094636922887212,
      "loss": 1.2424,
      "step": 28720
    },
    {
      "epoch": 2.4710487825862515,
      "grad_norm": 0.96875,
      "learning_rate": 0.000509436656383417,
      "loss": 1.1017,
      "step": 28721
    },
    {
      "epoch": 2.471134818893573,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005094096204505125,
      "loss": 0.9929,
      "step": 28722
    },
    {
      "epoch": 2.4712208552008947,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005093825844900868,
      "loss": 1.7707,
      "step": 28723
    },
    {
      "epoch": 2.4713068915082164,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005093555485022189,
      "loss": 1.2,
      "step": 28724
    },
    {
      "epoch": 2.471392927815538,
      "grad_norm": 1.609375,
      "learning_rate": 0.0005093285124869876,
      "loss": 1.0829,
      "step": 28725
    },
    {
      "epoch": 2.47147896412286,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005093014764444724,
      "loss": 0.7229,
      "step": 28726
    },
    {
      "epoch": 2.4715650004301817,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005092744403747522,
      "loss": 1.4365,
      "step": 28727
    },
    {
      "epoch": 2.4716510367375033,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000509247404277906,
      "loss": 1.2205,
      "step": 28728
    },
    {
      "epoch": 2.471737073044825,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0005092203681540127,
      "loss": 1.6569,
      "step": 28729
    },
    {
      "epoch": 2.4718231093521466,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000509193332003152,
      "loss": 0.5782,
      "step": 28730
    },
    {
      "epoch": 2.471909145659468,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005091662958254025,
      "loss": 1.3432,
      "step": 28731
    },
    {
      "epoch": 2.47199518196679,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005091392596208434,
      "loss": 1.1894,
      "step": 28732
    },
    {
      "epoch": 2.472081218274112,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005091122233895535,
      "loss": 0.9774,
      "step": 28733
    },
    {
      "epoch": 2.4721672545814335,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005090851871316123,
      "loss": 1.2404,
      "step": 28734
    },
    {
      "epoch": 2.472253290888755,
      "grad_norm": 0.875,
      "learning_rate": 0.0005090581508470989,
      "loss": 0.7925,
      "step": 28735
    },
    {
      "epoch": 2.4723393271960767,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005090311145360917,
      "loss": 0.984,
      "step": 28736
    },
    {
      "epoch": 2.4724253635033984,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005090040781986706,
      "loss": 2.1739,
      "step": 28737
    },
    {
      "epoch": 2.47251139981072,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005089770418349142,
      "loss": 1.2582,
      "step": 28738
    },
    {
      "epoch": 2.4725974361180416,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005089500054449018,
      "loss": 0.4732,
      "step": 28739
    },
    {
      "epoch": 2.4726834724253637,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005089229690287123,
      "loss": 0.7228,
      "step": 28740
    },
    {
      "epoch": 2.4727695087326853,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005088959325864248,
      "loss": 1.2515,
      "step": 28741
    },
    {
      "epoch": 2.472855545040007,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005088688961181185,
      "loss": 1.0768,
      "step": 28742
    },
    {
      "epoch": 2.4729415813473286,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0005088418596238723,
      "loss": 0.9987,
      "step": 28743
    },
    {
      "epoch": 2.47302761765465,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005088148231037656,
      "loss": 1.9471,
      "step": 28744
    },
    {
      "epoch": 2.473113653961972,
      "grad_norm": 0.9375,
      "learning_rate": 0.000508787786557877,
      "loss": 1.0405,
      "step": 28745
    },
    {
      "epoch": 2.4731996902692934,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005087607499862861,
      "loss": 0.9413,
      "step": 28746
    },
    {
      "epoch": 2.4732857265766155,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005087337133890715,
      "loss": 0.7911,
      "step": 28747
    },
    {
      "epoch": 2.473371762883937,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0005087066767663124,
      "loss": 1.2718,
      "step": 28748
    },
    {
      "epoch": 2.4734577991912587,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005086796401180883,
      "loss": 0.9591,
      "step": 28749
    },
    {
      "epoch": 2.4735438354985804,
      "grad_norm": 1.703125,
      "learning_rate": 0.0005086526034444776,
      "loss": 1.4367,
      "step": 28750
    },
    {
      "epoch": 2.473629871805902,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005086255667455598,
      "loss": 0.9521,
      "step": 28751
    },
    {
      "epoch": 2.4737159081132236,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005085985300214141,
      "loss": 1.3053,
      "step": 28752
    },
    {
      "epoch": 2.4738019444205452,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005085714932721193,
      "loss": 0.9174,
      "step": 28753
    },
    {
      "epoch": 2.4738879807278673,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005085444564977545,
      "loss": 1.4536,
      "step": 28754
    },
    {
      "epoch": 2.473974017035189,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005085174196983988,
      "loss": 1.2486,
      "step": 28755
    },
    {
      "epoch": 2.4740600533425106,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005084903828741312,
      "loss": 1.2978,
      "step": 28756
    },
    {
      "epoch": 2.474146089649832,
      "grad_norm": 1.9375,
      "learning_rate": 0.0005084633460250311,
      "loss": 1.7308,
      "step": 28757
    },
    {
      "epoch": 2.474232125957154,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005084363091511771,
      "loss": 1.2228,
      "step": 28758
    },
    {
      "epoch": 2.4743181622644754,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005084092722526487,
      "loss": 1.4572,
      "step": 28759
    },
    {
      "epoch": 2.474404198571797,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005083822353295249,
      "loss": 1.4259,
      "step": 28760
    },
    {
      "epoch": 2.474490234879119,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005083551983818846,
      "loss": 1.4332,
      "step": 28761
    },
    {
      "epoch": 2.4745762711864407,
      "grad_norm": 0.875,
      "learning_rate": 0.000508328161409807,
      "loss": 0.6743,
      "step": 28762
    },
    {
      "epoch": 2.4746623074937624,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0005083011244133711,
      "loss": 0.8289,
      "step": 28763
    },
    {
      "epoch": 2.474748343801084,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000508274087392656,
      "loss": 1.6013,
      "step": 28764
    },
    {
      "epoch": 2.4748343801084056,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000508247050347741,
      "loss": 0.9414,
      "step": 28765
    },
    {
      "epoch": 2.4749204164157277,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005082200132787046,
      "loss": 1.4098,
      "step": 28766
    },
    {
      "epoch": 2.4750064527230493,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005081929761856266,
      "loss": 2.0144,
      "step": 28767
    },
    {
      "epoch": 2.475092489030371,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005081659390685856,
      "loss": 0.6196,
      "step": 28768
    },
    {
      "epoch": 2.4751785253376926,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005081389019276609,
      "loss": 1.085,
      "step": 28769
    },
    {
      "epoch": 2.475264561645014,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005081118647629313,
      "loss": 1.4126,
      "step": 28770
    },
    {
      "epoch": 2.475350597952336,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005080848275744763,
      "loss": 1.6932,
      "step": 28771
    },
    {
      "epoch": 2.4754366342596574,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0005080577903623746,
      "loss": 0.8777,
      "step": 28772
    },
    {
      "epoch": 2.4755226705669795,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005080307531267054,
      "loss": 1.0612,
      "step": 28773
    },
    {
      "epoch": 2.475608706874301,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000508003715867548,
      "loss": 1.143,
      "step": 28774
    },
    {
      "epoch": 2.4756947431816227,
      "grad_norm": 1.25,
      "learning_rate": 0.0005079766785849812,
      "loss": 1.112,
      "step": 28775
    },
    {
      "epoch": 2.4757807794889444,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005079496412790841,
      "loss": 1.6488,
      "step": 28776
    },
    {
      "epoch": 2.475866815796266,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005079226039499359,
      "loss": 1.2776,
      "step": 28777
    },
    {
      "epoch": 2.4759528521035876,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005078955665976157,
      "loss": 1.6423,
      "step": 28778
    },
    {
      "epoch": 2.4760388884109092,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005078685292222025,
      "loss": 0.8393,
      "step": 28779
    },
    {
      "epoch": 2.4761249247182313,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005078414918237753,
      "loss": 1.2979,
      "step": 28780
    },
    {
      "epoch": 2.476210961025553,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005078144544024133,
      "loss": 0.813,
      "step": 28781
    },
    {
      "epoch": 2.4762969973328746,
      "grad_norm": 1.171875,
      "learning_rate": 0.0005077874169581954,
      "loss": 1.2875,
      "step": 28782
    },
    {
      "epoch": 2.476383033640196,
      "grad_norm": 1.3984375,
      "learning_rate": 0.000507760379491201,
      "loss": 1.0029,
      "step": 28783
    },
    {
      "epoch": 2.476469069947518,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005077333420015089,
      "loss": 0.7449,
      "step": 28784
    },
    {
      "epoch": 2.4765551062548394,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005077063044891982,
      "loss": 1.6149,
      "step": 28785
    },
    {
      "epoch": 2.476641142562161,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005076792669543482,
      "loss": 0.8883,
      "step": 28786
    },
    {
      "epoch": 2.476727178869483,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005076522293970379,
      "loss": 1.152,
      "step": 28787
    },
    {
      "epoch": 2.4768132151768047,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005076251918173462,
      "loss": 1.1314,
      "step": 28788
    },
    {
      "epoch": 2.4768992514841264,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005075981542153522,
      "loss": 1.0721,
      "step": 28789
    },
    {
      "epoch": 2.476985287791448,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005075711165911354,
      "loss": 1.0742,
      "step": 28790
    },
    {
      "epoch": 2.4770713240987696,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0005075440789447743,
      "loss": 0.9685,
      "step": 28791
    },
    {
      "epoch": 2.4771573604060912,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0005075170412763482,
      "loss": 1.3968,
      "step": 28792
    },
    {
      "epoch": 2.477243396713413,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005074900035859363,
      "loss": 1.9997,
      "step": 28793
    },
    {
      "epoch": 2.477329433020735,
      "grad_norm": 1.0,
      "learning_rate": 0.0005074629658736176,
      "loss": 1.1867,
      "step": 28794
    },
    {
      "epoch": 2.4774154693280566,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005074359281394713,
      "loss": 1.0701,
      "step": 28795
    },
    {
      "epoch": 2.477501505635378,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005074088903835762,
      "loss": 1.7907,
      "step": 28796
    },
    {
      "epoch": 2.4775875419427,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0005073818526060115,
      "loss": 0.9574,
      "step": 28797
    },
    {
      "epoch": 2.4776735782500214,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005073548148068564,
      "loss": 1.5399,
      "step": 28798
    },
    {
      "epoch": 2.477759614557343,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005073277769861899,
      "loss": 1.344,
      "step": 28799
    },
    {
      "epoch": 2.4778456508646647,
      "grad_norm": 1.3125,
      "learning_rate": 0.000507300739144091,
      "loss": 1.0824,
      "step": 28800
    },
    {
      "epoch": 2.4779316871719868,
      "grad_norm": 1.0625,
      "learning_rate": 0.000507273701280639,
      "loss": 1.229,
      "step": 28801
    },
    {
      "epoch": 2.4780177234793084,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005072466633959129,
      "loss": 1.2919,
      "step": 28802
    },
    {
      "epoch": 2.47810375978663,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005072196254899914,
      "loss": 1.5532,
      "step": 28803
    },
    {
      "epoch": 2.4781897960939516,
      "grad_norm": 1.125,
      "learning_rate": 0.0005071925875629541,
      "loss": 1.5404,
      "step": 28804
    },
    {
      "epoch": 2.4782758324012732,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005071655496148798,
      "loss": 1.1531,
      "step": 28805
    },
    {
      "epoch": 2.478361868708595,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005071385116458478,
      "loss": 1.3797,
      "step": 28806
    },
    {
      "epoch": 2.4784479050159165,
      "grad_norm": 1.375,
      "learning_rate": 0.0005071114736559369,
      "loss": 0.78,
      "step": 28807
    },
    {
      "epoch": 2.4785339413232386,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0005070844356452264,
      "loss": 1.3697,
      "step": 28808
    },
    {
      "epoch": 2.47861997763056,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005070573976137953,
      "loss": 1.4874,
      "step": 28809
    },
    {
      "epoch": 2.478706013937882,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005070303595617227,
      "loss": 0.7976,
      "step": 28810
    },
    {
      "epoch": 2.4787920502452034,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0005070033214890875,
      "loss": 1.8311,
      "step": 28811
    },
    {
      "epoch": 2.478878086552525,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005069762833959693,
      "loss": 1.6988,
      "step": 28812
    },
    {
      "epoch": 2.4789641228598467,
      "grad_norm": 1.125,
      "learning_rate": 0.0005069492452824466,
      "loss": 0.6383,
      "step": 28813
    },
    {
      "epoch": 2.4790501591671683,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005069222071485985,
      "loss": 1.2263,
      "step": 28814
    },
    {
      "epoch": 2.4791361954744904,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005068951689945046,
      "loss": 1.3755,
      "step": 28815
    },
    {
      "epoch": 2.479222231781812,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0005068681308202436,
      "loss": 1.2082,
      "step": 28816
    },
    {
      "epoch": 2.4793082680891336,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005068410926258948,
      "loss": 1.626,
      "step": 28817
    },
    {
      "epoch": 2.4793943043964553,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005068140544115369,
      "loss": 1.6748,
      "step": 28818
    },
    {
      "epoch": 2.479480340703777,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005067870161772492,
      "loss": 1.3024,
      "step": 28819
    },
    {
      "epoch": 2.4795663770110985,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005067599779231109,
      "loss": 0.5888,
      "step": 28820
    },
    {
      "epoch": 2.4796524133184206,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000506732939649201,
      "loss": 0.7725,
      "step": 28821
    },
    {
      "epoch": 2.479738449625742,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005067059013555983,
      "loss": 1.0651,
      "step": 28822
    },
    {
      "epoch": 2.479824485933064,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005066788630423826,
      "loss": 0.7193,
      "step": 28823
    },
    {
      "epoch": 2.4799105222403854,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005066518247096322,
      "loss": 1.236,
      "step": 28824
    },
    {
      "epoch": 2.479996558547707,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0005066247863574267,
      "loss": 0.6891,
      "step": 28825
    },
    {
      "epoch": 2.4800825948550287,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005065977479858448,
      "loss": 1.2151,
      "step": 28826
    },
    {
      "epoch": 2.4801686311623508,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005065707095949659,
      "loss": 1.3935,
      "step": 28827
    },
    {
      "epoch": 2.4802546674696724,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005065436711848689,
      "loss": 1.1704,
      "step": 28828
    },
    {
      "epoch": 2.480340703776994,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000506516632755633,
      "loss": 1.0363,
      "step": 28829
    },
    {
      "epoch": 2.4804267400843156,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005064895943073371,
      "loss": 0.912,
      "step": 28830
    },
    {
      "epoch": 2.4805127763916373,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0005064625558400606,
      "loss": 0.6913,
      "step": 28831
    },
    {
      "epoch": 2.480598812698959,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005064355173538821,
      "loss": 0.6802,
      "step": 28832
    },
    {
      "epoch": 2.4806848490062805,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005064084788488811,
      "loss": 1.5193,
      "step": 28833
    },
    {
      "epoch": 2.4807708853136026,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0005063814403251366,
      "loss": 1.2576,
      "step": 28834
    },
    {
      "epoch": 2.480856921620924,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005063544017827275,
      "loss": 0.6394,
      "step": 28835
    },
    {
      "epoch": 2.480942957928246,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005063273632217333,
      "loss": 0.945,
      "step": 28836
    },
    {
      "epoch": 2.4810289942355674,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005063003246422324,
      "loss": 2.5738,
      "step": 28837
    },
    {
      "epoch": 2.481115030542889,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005062732860443045,
      "loss": 1.0223,
      "step": 28838
    },
    {
      "epoch": 2.4812010668502107,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005062462474280285,
      "loss": 0.9678,
      "step": 28839
    },
    {
      "epoch": 2.4812871031575323,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005062192087934833,
      "loss": 1.798,
      "step": 28840
    },
    {
      "epoch": 2.4813731394648544,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000506192170140748,
      "loss": 1.3061,
      "step": 28841
    },
    {
      "epoch": 2.481459175772176,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005061651314699021,
      "loss": 0.7414,
      "step": 28842
    },
    {
      "epoch": 2.4815452120794976,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005061380927810243,
      "loss": 0.917,
      "step": 28843
    },
    {
      "epoch": 2.4816312483868193,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005061110540741938,
      "loss": 1.2063,
      "step": 28844
    },
    {
      "epoch": 2.481717284694141,
      "grad_norm": 1.0,
      "learning_rate": 0.0005060840153494894,
      "loss": 0.9645,
      "step": 28845
    },
    {
      "epoch": 2.4818033210014625,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005060569766069908,
      "loss": 1.9304,
      "step": 28846
    },
    {
      "epoch": 2.481889357308784,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005060299378467765,
      "loss": 1.1191,
      "step": 28847
    },
    {
      "epoch": 2.481975393616106,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005060028990689258,
      "loss": 1.4935,
      "step": 28848
    },
    {
      "epoch": 2.482061429923428,
      "grad_norm": 1.1875,
      "learning_rate": 0.000505975860273518,
      "loss": 1.6765,
      "step": 28849
    },
    {
      "epoch": 2.4821474662307494,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005059488214606318,
      "loss": 0.8149,
      "step": 28850
    },
    {
      "epoch": 2.482233502538071,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005059217826303465,
      "loss": 1.2596,
      "step": 28851
    },
    {
      "epoch": 2.4823195388453927,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000505894743782741,
      "loss": 0.7961,
      "step": 28852
    },
    {
      "epoch": 2.4824055751527143,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005058677049178947,
      "loss": 2.1972,
      "step": 28853
    },
    {
      "epoch": 2.482491611460036,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005058406660358865,
      "loss": 1.3361,
      "step": 28854
    },
    {
      "epoch": 2.482577647767358,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0005058136271367953,
      "loss": 0.7451,
      "step": 28855
    },
    {
      "epoch": 2.4826636840746796,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005057865882207006,
      "loss": 1.4142,
      "step": 28856
    },
    {
      "epoch": 2.4827497203820013,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0005057595492876813,
      "loss": 0.6466,
      "step": 28857
    },
    {
      "epoch": 2.482835756689323,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005057325103378161,
      "loss": 1.3767,
      "step": 28858
    },
    {
      "epoch": 2.4829217929966445,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005057054713711848,
      "loss": 1.8501,
      "step": 28859
    },
    {
      "epoch": 2.483007829303966,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005056784323878657,
      "loss": 1.0593,
      "step": 28860
    },
    {
      "epoch": 2.4830938656112878,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005056513933879385,
      "loss": 1.4825,
      "step": 28861
    },
    {
      "epoch": 2.48317990191861,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005056243543714822,
      "loss": 1.2593,
      "step": 28862
    },
    {
      "epoch": 2.4832659382259314,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005055973153385756,
      "loss": 0.7509,
      "step": 28863
    },
    {
      "epoch": 2.483351974533253,
      "grad_norm": 1.453125,
      "learning_rate": 0.000505570276289298,
      "loss": 1.6345,
      "step": 28864
    },
    {
      "epoch": 2.4834380108405747,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005055432372237284,
      "loss": 1.115,
      "step": 28865
    },
    {
      "epoch": 2.4835240471478963,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000505516198141946,
      "loss": 1.0895,
      "step": 28866
    },
    {
      "epoch": 2.483610083455218,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005054891590440296,
      "loss": 2.266,
      "step": 28867
    },
    {
      "epoch": 2.4836961197625396,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005054621199300586,
      "loss": 0.9336,
      "step": 28868
    },
    {
      "epoch": 2.4837821560698616,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000505435080800112,
      "loss": 1.3584,
      "step": 28869
    },
    {
      "epoch": 2.4838681923771833,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005054080416542688,
      "loss": 1.4533,
      "step": 28870
    },
    {
      "epoch": 2.483954228684505,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005053810024926081,
      "loss": 0.9779,
      "step": 28871
    },
    {
      "epoch": 2.4840402649918265,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005053539633152091,
      "loss": 1.0522,
      "step": 28872
    },
    {
      "epoch": 2.484126301299148,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005053269241221507,
      "loss": 1.145,
      "step": 28873
    },
    {
      "epoch": 2.4842123376064698,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005052998849135122,
      "loss": 1.506,
      "step": 28874
    },
    {
      "epoch": 2.4842983739137914,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005052728456893724,
      "loss": 1.0441,
      "step": 28875
    },
    {
      "epoch": 2.4843844102211134,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005052458064498107,
      "loss": 1.3718,
      "step": 28876
    },
    {
      "epoch": 2.484470446528435,
      "grad_norm": 1.09375,
      "learning_rate": 0.000505218767194906,
      "loss": 0.9493,
      "step": 28877
    },
    {
      "epoch": 2.4845564828357567,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005051917279247373,
      "loss": 0.7189,
      "step": 28878
    },
    {
      "epoch": 2.4846425191430783,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005051646886393839,
      "loss": 1.4747,
      "step": 28879
    },
    {
      "epoch": 2.4847285554504,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005051376493389249,
      "loss": 1.3097,
      "step": 28880
    },
    {
      "epoch": 2.4848145917577216,
      "grad_norm": 1.375,
      "learning_rate": 0.0005051106100234391,
      "loss": 1.2983,
      "step": 28881
    },
    {
      "epoch": 2.4849006280650436,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005050835706930058,
      "loss": 1.4718,
      "step": 28882
    },
    {
      "epoch": 2.4849866643723653,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0005050565313477042,
      "loss": 0.6257,
      "step": 28883
    },
    {
      "epoch": 2.485072700679687,
      "grad_norm": 1.625,
      "learning_rate": 0.000505029491987613,
      "loss": 1.5305,
      "step": 28884
    },
    {
      "epoch": 2.4851587369870085,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005050024526128118,
      "loss": 1.7499,
      "step": 28885
    },
    {
      "epoch": 2.48524477329433,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005049754132233791,
      "loss": 0.7564,
      "step": 28886
    },
    {
      "epoch": 2.4853308096016518,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005049483738193943,
      "loss": 0.986,
      "step": 28887
    },
    {
      "epoch": 2.485416845908974,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005049213344009367,
      "loss": 1.0314,
      "step": 28888
    },
    {
      "epoch": 2.4855028822162955,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000504894294968085,
      "loss": 1.1318,
      "step": 28889
    },
    {
      "epoch": 2.485588918523617,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005048672555209185,
      "loss": 0.6403,
      "step": 28890
    },
    {
      "epoch": 2.4856749548309387,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005048402160595162,
      "loss": 0.7669,
      "step": 28891
    },
    {
      "epoch": 2.4857609911382603,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005048131765839572,
      "loss": 1.2137,
      "step": 28892
    },
    {
      "epoch": 2.485847027445582,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005047861370943205,
      "loss": 1.4193,
      "step": 28893
    },
    {
      "epoch": 2.4859330637529036,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005047590975906854,
      "loss": 1.317,
      "step": 28894
    },
    {
      "epoch": 2.4860191000602256,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005047320580731309,
      "loss": 1.0007,
      "step": 28895
    },
    {
      "epoch": 2.4861051363675473,
      "grad_norm": 1.375,
      "learning_rate": 0.0005047050185417359,
      "loss": 1.2361,
      "step": 28896
    },
    {
      "epoch": 2.486191172674869,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005046779789965797,
      "loss": 0.8521,
      "step": 28897
    },
    {
      "epoch": 2.4862772089821905,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005046509394377414,
      "loss": 1.3001,
      "step": 28898
    },
    {
      "epoch": 2.486363245289512,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0005046238998653,
      "loss": 0.9707,
      "step": 28899
    },
    {
      "epoch": 2.4864492815968338,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005045968602793345,
      "loss": 1.3543,
      "step": 28900
    },
    {
      "epoch": 2.4865353179041554,
      "grad_norm": 1.234375,
      "learning_rate": 0.000504569820679924,
      "loss": 1.2085,
      "step": 28901
    },
    {
      "epoch": 2.4866213542114775,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005045427810671479,
      "loss": 0.8133,
      "step": 28902
    },
    {
      "epoch": 2.486707390518799,
      "grad_norm": 1.203125,
      "learning_rate": 0.000504515741441085,
      "loss": 0.5662,
      "step": 28903
    },
    {
      "epoch": 2.4867934268261207,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005044887018018142,
      "loss": 1.0493,
      "step": 28904
    },
    {
      "epoch": 2.4868794631334423,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005044616621494152,
      "loss": 1.5019,
      "step": 28905
    },
    {
      "epoch": 2.486965499440764,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0005044346224839664,
      "loss": 0.4739,
      "step": 28906
    },
    {
      "epoch": 2.4870515357480856,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005044075828055473,
      "loss": 0.6444,
      "step": 28907
    },
    {
      "epoch": 2.487137572055407,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005043805431142369,
      "loss": 0.9755,
      "step": 28908
    },
    {
      "epoch": 2.4872236083627293,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0005043535034101142,
      "loss": 0.9011,
      "step": 28909
    },
    {
      "epoch": 2.487309644670051,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005043264636932584,
      "loss": 0.7827,
      "step": 28910
    },
    {
      "epoch": 2.4873956809773725,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0005042994239637485,
      "loss": 1.0077,
      "step": 28911
    },
    {
      "epoch": 2.487481717284694,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005042723842216637,
      "loss": 0.769,
      "step": 28912
    },
    {
      "epoch": 2.4875677535920158,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005042453444670828,
      "loss": 1.0544,
      "step": 28913
    },
    {
      "epoch": 2.4876537898993374,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005042183047000852,
      "loss": 0.9603,
      "step": 28914
    },
    {
      "epoch": 2.487739826206659,
      "grad_norm": 0.953125,
      "learning_rate": 0.0005041912649207499,
      "loss": 0.5915,
      "step": 28915
    },
    {
      "epoch": 2.487825862513981,
      "grad_norm": 1.140625,
      "learning_rate": 0.0005041642251291561,
      "loss": 1.3065,
      "step": 28916
    },
    {
      "epoch": 2.4879118988213027,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005041371853253825,
      "loss": 1.6988,
      "step": 28917
    },
    {
      "epoch": 2.4879979351286243,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005041101455095086,
      "loss": 1.2864,
      "step": 28918
    },
    {
      "epoch": 2.488083971435946,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005040831056816133,
      "loss": 1.6019,
      "step": 28919
    },
    {
      "epoch": 2.4881700077432676,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005040560658417757,
      "loss": 0.8786,
      "step": 28920
    },
    {
      "epoch": 2.488256044050589,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005040290259900749,
      "loss": 0.9565,
      "step": 28921
    },
    {
      "epoch": 2.488342080357911,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005040019861265901,
      "loss": 1.815,
      "step": 28922
    },
    {
      "epoch": 2.488428116665233,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0005039749462514001,
      "loss": 1.1948,
      "step": 28923
    },
    {
      "epoch": 2.4885141529725545,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0005039479063645842,
      "loss": 1.2794,
      "step": 28924
    },
    {
      "epoch": 2.488600189279876,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005039208664662215,
      "loss": 1.0825,
      "step": 28925
    },
    {
      "epoch": 2.4886862255871978,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005038938265563909,
      "loss": 0.925,
      "step": 28926
    },
    {
      "epoch": 2.4887722618945194,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005038667866351717,
      "loss": 1.1259,
      "step": 28927
    },
    {
      "epoch": 2.488858298201841,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005038397467026429,
      "loss": 1.1905,
      "step": 28928
    },
    {
      "epoch": 2.4889443345091626,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005038127067588836,
      "loss": 1.0733,
      "step": 28929
    },
    {
      "epoch": 2.4890303708164847,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005037856668039728,
      "loss": 0.8923,
      "step": 28930
    },
    {
      "epoch": 2.4891164071238063,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0005037586268379898,
      "loss": 0.9316,
      "step": 28931
    },
    {
      "epoch": 2.489202443431128,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005037315868610132,
      "loss": 0.6031,
      "step": 28932
    },
    {
      "epoch": 2.4892884797384496,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005037045468731229,
      "loss": 1.399,
      "step": 28933
    },
    {
      "epoch": 2.489374516045771,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005036775068743972,
      "loss": 1.2943,
      "step": 28934
    },
    {
      "epoch": 2.489460552353093,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0005036504668649157,
      "loss": 1.0488,
      "step": 28935
    },
    {
      "epoch": 2.4895465886604145,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005036234268447573,
      "loss": 0.9939,
      "step": 28936
    },
    {
      "epoch": 2.4896326249677365,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000503596386814001,
      "loss": 0.6617,
      "step": 28937
    },
    {
      "epoch": 2.489718661275058,
      "grad_norm": 2.234375,
      "learning_rate": 0.0005035693467727258,
      "loss": 1.4533,
      "step": 28938
    },
    {
      "epoch": 2.4898046975823798,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005035423067210112,
      "loss": 1.3347,
      "step": 28939
    },
    {
      "epoch": 2.4898907338897014,
      "grad_norm": 1.234375,
      "learning_rate": 0.000503515266658936,
      "loss": 1.5172,
      "step": 28940
    },
    {
      "epoch": 2.489976770197023,
      "grad_norm": 1.484375,
      "learning_rate": 0.0005034882265865793,
      "loss": 1.1389,
      "step": 28941
    },
    {
      "epoch": 2.490062806504345,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0005034611865040202,
      "loss": 0.8444,
      "step": 28942
    },
    {
      "epoch": 2.4901488428116667,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005034341464113378,
      "loss": 1.4028,
      "step": 28943
    },
    {
      "epoch": 2.4902348791189883,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0005034071063086112,
      "loss": 1.3305,
      "step": 28944
    },
    {
      "epoch": 2.49032091542631,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005033800661959195,
      "loss": 1.8817,
      "step": 28945
    },
    {
      "epoch": 2.4904069517336316,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005033530260733416,
      "loss": 2.094,
      "step": 28946
    },
    {
      "epoch": 2.490492988040953,
      "grad_norm": 1.453125,
      "learning_rate": 0.000503325985940957,
      "loss": 1.2872,
      "step": 28947
    },
    {
      "epoch": 2.490579024348275,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005032989457988445,
      "loss": 1.0471,
      "step": 28948
    },
    {
      "epoch": 2.490665060655597,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005032719056470831,
      "loss": 1.7748,
      "step": 28949
    },
    {
      "epoch": 2.4907510969629185,
      "grad_norm": 0.921875,
      "learning_rate": 0.000503244865485752,
      "loss": 1.2287,
      "step": 28950
    },
    {
      "epoch": 2.49083713327024,
      "grad_norm": 1.390625,
      "learning_rate": 0.0005032178253149304,
      "loss": 1.3299,
      "step": 28951
    },
    {
      "epoch": 2.4909231695775618,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005031907851346973,
      "loss": 1.0723,
      "step": 28952
    },
    {
      "epoch": 2.4910092058848834,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005031637449451317,
      "loss": 2.2685,
      "step": 28953
    },
    {
      "epoch": 2.491095242192205,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005031367047463127,
      "loss": 1.7979,
      "step": 28954
    },
    {
      "epoch": 2.4911812784995266,
      "grad_norm": 1.015625,
      "learning_rate": 0.0005031096645383196,
      "loss": 1.9301,
      "step": 28955
    },
    {
      "epoch": 2.4912673148068487,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005030826243212313,
      "loss": 1.1116,
      "step": 28956
    },
    {
      "epoch": 2.4913533511141703,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005030555840951268,
      "loss": 0.5399,
      "step": 28957
    },
    {
      "epoch": 2.491439387421492,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005030285438600854,
      "loss": 1.3124,
      "step": 28958
    },
    {
      "epoch": 2.4915254237288136,
      "grad_norm": 1.421875,
      "learning_rate": 0.0005030015036161863,
      "loss": 0.8493,
      "step": 28959
    },
    {
      "epoch": 2.491611460036135,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000502974463363508,
      "loss": 0.9598,
      "step": 28960
    },
    {
      "epoch": 2.491697496343457,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005029474231021303,
      "loss": 1.439,
      "step": 28961
    },
    {
      "epoch": 2.4917835326507785,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005029203828321318,
      "loss": 1.2118,
      "step": 28962
    },
    {
      "epoch": 2.4918695689581005,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005028933425535917,
      "loss": 1.4685,
      "step": 28963
    },
    {
      "epoch": 2.491955605265422,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005028663022665893,
      "loss": 0.7205,
      "step": 28964
    },
    {
      "epoch": 2.4920416415727438,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005028392619712033,
      "loss": 1.6087,
      "step": 28965
    },
    {
      "epoch": 2.4921276778800654,
      "grad_norm": 1.1875,
      "learning_rate": 0.0005028122216675133,
      "loss": 1.185,
      "step": 28966
    },
    {
      "epoch": 2.492213714187387,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005027851813555978,
      "loss": 0.8854,
      "step": 28967
    },
    {
      "epoch": 2.4922997504947086,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0005027581410355364,
      "loss": 1.8174,
      "step": 28968
    },
    {
      "epoch": 2.4923857868020303,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005027311007074078,
      "loss": 1.3553,
      "step": 28969
    },
    {
      "epoch": 2.4924718231093523,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005027040603712915,
      "loss": 1.1808,
      "step": 28970
    },
    {
      "epoch": 2.492557859416674,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0005026770200272661,
      "loss": 0.805,
      "step": 28971
    },
    {
      "epoch": 2.4926438957239956,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005026499796754111,
      "loss": 2.0146,
      "step": 28972
    },
    {
      "epoch": 2.492729932031317,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005026229393158054,
      "loss": 1.1565,
      "step": 28973
    },
    {
      "epoch": 2.492815968338639,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005025958989485281,
      "loss": 1.4495,
      "step": 28974
    },
    {
      "epoch": 2.4929020046459605,
      "grad_norm": 1.0625,
      "learning_rate": 0.0005025688585736582,
      "loss": 1.6532,
      "step": 28975
    },
    {
      "epoch": 2.492988040953282,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005025418181912749,
      "loss": 1.9451,
      "step": 28976
    },
    {
      "epoch": 2.493074077260604,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0005025147778014575,
      "loss": 1.0236,
      "step": 28977
    },
    {
      "epoch": 2.4931601135679258,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005024877374042848,
      "loss": 1.1869,
      "step": 28978
    },
    {
      "epoch": 2.4932461498752474,
      "grad_norm": 0.96875,
      "learning_rate": 0.0005024606969998358,
      "loss": 0.8691,
      "step": 28979
    },
    {
      "epoch": 2.493332186182569,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0005024336565881898,
      "loss": 1.3466,
      "step": 28980
    },
    {
      "epoch": 2.4934182224898906,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005024066161694259,
      "loss": 0.8593,
      "step": 28981
    },
    {
      "epoch": 2.4935042587972123,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0005023795757436229,
      "loss": 0.7031,
      "step": 28982
    },
    {
      "epoch": 2.493590295104534,
      "grad_norm": 1.453125,
      "learning_rate": 0.0005023525353108604,
      "loss": 1.3679,
      "step": 28983
    },
    {
      "epoch": 2.493676331411856,
      "grad_norm": 1.375,
      "learning_rate": 0.000502325494871217,
      "loss": 1.4941,
      "step": 28984
    },
    {
      "epoch": 2.4937623677191776,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005022984544247721,
      "loss": 1.7038,
      "step": 28985
    },
    {
      "epoch": 2.493848404026499,
      "grad_norm": 1.546875,
      "learning_rate": 0.0005022714139716045,
      "loss": 1.8045,
      "step": 28986
    },
    {
      "epoch": 2.493934440333821,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005022443735117934,
      "loss": 1.3411,
      "step": 28987
    },
    {
      "epoch": 2.4940204766411425,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0005022173330454182,
      "loss": 0.6059,
      "step": 28988
    },
    {
      "epoch": 2.494106512948464,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005021902925725577,
      "loss": 1.3563,
      "step": 28989
    },
    {
      "epoch": 2.4941925492557857,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005021632520932908,
      "loss": 1.7343,
      "step": 28990
    },
    {
      "epoch": 2.4942785855631078,
      "grad_norm": 0.9375,
      "learning_rate": 0.000502136211607697,
      "loss": 0.9185,
      "step": 28991
    },
    {
      "epoch": 2.4943646218704294,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005021091711158551,
      "loss": 1.7352,
      "step": 28992
    },
    {
      "epoch": 2.494450658177751,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005020821306178445,
      "loss": 1.6673,
      "step": 28993
    },
    {
      "epoch": 2.4945366944850726,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005020550901137438,
      "loss": 0.8343,
      "step": 28994
    },
    {
      "epoch": 2.4946227307923943,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005020280496036325,
      "loss": 1.0187,
      "step": 28995
    },
    {
      "epoch": 2.494708767099716,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005020010090875895,
      "loss": 1.2887,
      "step": 28996
    },
    {
      "epoch": 2.494794803407038,
      "grad_norm": 1.1875,
      "learning_rate": 0.000501973968565694,
      "loss": 1.6572,
      "step": 28997
    },
    {
      "epoch": 2.4948808397143596,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0005019469280380249,
      "loss": 1.2981,
      "step": 28998
    },
    {
      "epoch": 2.494966876021681,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005019198875046615,
      "loss": 0.8896,
      "step": 28999
    },
    {
      "epoch": 2.495052912329003,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005018928469656829,
      "loss": 1.2927,
      "step": 29000
    },
    {
      "epoch": 2.4951389486363245,
      "grad_norm": 1.0,
      "learning_rate": 0.000501865806421168,
      "loss": 1.4721,
      "step": 29001
    },
    {
      "epoch": 2.495224984943646,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005018387658711959,
      "loss": 0.5204,
      "step": 29002
    },
    {
      "epoch": 2.495311021250968,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005018117253158458,
      "loss": 1.4774,
      "step": 29003
    },
    {
      "epoch": 2.4953970575582898,
      "grad_norm": 1.578125,
      "learning_rate": 0.0005017846847551969,
      "loss": 1.2291,
      "step": 29004
    },
    {
      "epoch": 2.4954830938656114,
      "grad_norm": 1.140625,
      "learning_rate": 0.000501757644189328,
      "loss": 1.4406,
      "step": 29005
    },
    {
      "epoch": 2.495569130172933,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0005017306036183182,
      "loss": 1.4628,
      "step": 29006
    },
    {
      "epoch": 2.4956551664802546,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000501703563042247,
      "loss": 0.8096,
      "step": 29007
    },
    {
      "epoch": 2.4957412027875763,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0005016765224611931,
      "loss": 1.5545,
      "step": 29008
    },
    {
      "epoch": 2.495827239094898,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005016494818752356,
      "loss": 0.9306,
      "step": 29009
    },
    {
      "epoch": 2.49591327540222,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0005016224412844539,
      "loss": 1.055,
      "step": 29010
    },
    {
      "epoch": 2.4959993117095416,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0005015954006889268,
      "loss": 0.5776,
      "step": 29011
    },
    {
      "epoch": 2.496085348016863,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0005015683600887334,
      "loss": 0.9469,
      "step": 29012
    },
    {
      "epoch": 2.496171384324185,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005015413194839528,
      "loss": 1.6867,
      "step": 29013
    },
    {
      "epoch": 2.4962574206315065,
      "grad_norm": 1.15625,
      "learning_rate": 0.0005015142788746642,
      "loss": 0.8357,
      "step": 29014
    },
    {
      "epoch": 2.496343456938828,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005014872382609468,
      "loss": 1.8224,
      "step": 29015
    },
    {
      "epoch": 2.4964294932461497,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005014601976428792,
      "loss": 1.2605,
      "step": 29016
    },
    {
      "epoch": 2.496515529553472,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000501433157020541,
      "loss": 1.4163,
      "step": 29017
    },
    {
      "epoch": 2.4966015658607934,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005014061163940111,
      "loss": 0.5139,
      "step": 29018
    },
    {
      "epoch": 2.496687602168115,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005013790757633686,
      "loss": 0.7036,
      "step": 29019
    },
    {
      "epoch": 2.4967736384754367,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005013520351286924,
      "loss": 0.8404,
      "step": 29020
    },
    {
      "epoch": 2.4968596747827583,
      "grad_norm": 1.109375,
      "learning_rate": 0.000501324994490062,
      "loss": 1.4512,
      "step": 29021
    },
    {
      "epoch": 2.49694571109008,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0005012979538475562,
      "loss": 1.7172,
      "step": 29022
    },
    {
      "epoch": 2.4970317473974015,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005012709132012539,
      "loss": 1.3495,
      "step": 29023
    },
    {
      "epoch": 2.4971177837047236,
      "grad_norm": 0.859375,
      "learning_rate": 0.0005012438725512345,
      "loss": 0.9019,
      "step": 29024
    },
    {
      "epoch": 2.497203820012045,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005012168318975772,
      "loss": 1.5541,
      "step": 29025
    },
    {
      "epoch": 2.497289856319367,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0005011897912403607,
      "loss": 0.9041,
      "step": 29026
    },
    {
      "epoch": 2.4973758926266885,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005011627505796645,
      "loss": 0.9939,
      "step": 29027
    },
    {
      "epoch": 2.49746192893401,
      "grad_norm": 0.828125,
      "learning_rate": 0.0005011357099155673,
      "loss": 1.1768,
      "step": 29028
    },
    {
      "epoch": 2.4975479652413317,
      "grad_norm": 1.890625,
      "learning_rate": 0.0005011086692481485,
      "loss": 1.0719,
      "step": 29029
    },
    {
      "epoch": 2.4976340015486533,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000501081628577487,
      "loss": 1.1612,
      "step": 29030
    },
    {
      "epoch": 2.4977200378559754,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000501054587903662,
      "loss": 1.0294,
      "step": 29031
    },
    {
      "epoch": 2.497806074163297,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0005010275472267524,
      "loss": 1.2103,
      "step": 29032
    },
    {
      "epoch": 2.4978921104706187,
      "grad_norm": 1.53125,
      "learning_rate": 0.0005010005065468376,
      "loss": 1.1072,
      "step": 29033
    },
    {
      "epoch": 2.4979781467779403,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0005009734658639965,
      "loss": 1.0079,
      "step": 29034
    },
    {
      "epoch": 2.498064183085262,
      "grad_norm": 1.265625,
      "learning_rate": 0.000500946425178308,
      "loss": 1.5579,
      "step": 29035
    },
    {
      "epoch": 2.4981502193925835,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005009193844898516,
      "loss": 1.9628,
      "step": 29036
    },
    {
      "epoch": 2.498236255699905,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000500892343798706,
      "loss": 0.6024,
      "step": 29037
    },
    {
      "epoch": 2.498322292007227,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005008653031049507,
      "loss": 1.8494,
      "step": 29038
    },
    {
      "epoch": 2.498408328314549,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005008382624086643,
      "loss": 0.8761,
      "step": 29039
    },
    {
      "epoch": 2.4984943646218705,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005008112217099263,
      "loss": 1.2368,
      "step": 29040
    },
    {
      "epoch": 2.498580400929192,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005007841810088157,
      "loss": 1.0891,
      "step": 29041
    },
    {
      "epoch": 2.4986664372365137,
      "grad_norm": 1.09375,
      "learning_rate": 0.0005007571403054113,
      "loss": 1.5092,
      "step": 29042
    },
    {
      "epoch": 2.4987524735438353,
      "grad_norm": 1.671875,
      "learning_rate": 0.0005007300995997925,
      "loss": 1.5302,
      "step": 29043
    },
    {
      "epoch": 2.498838509851157,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0005007030588920384,
      "loss": 1.0438,
      "step": 29044
    },
    {
      "epoch": 2.498924546158479,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000500676018182228,
      "loss": 1.2158,
      "step": 29045
    },
    {
      "epoch": 2.4990105824658007,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0005006489774704401,
      "loss": 1.189,
      "step": 29046
    },
    {
      "epoch": 2.4990966187731223,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005006219367567543,
      "loss": 1.7041,
      "step": 29047
    },
    {
      "epoch": 2.499182655080444,
      "grad_norm": 1.03125,
      "learning_rate": 0.0005005948960412495,
      "loss": 1.3136,
      "step": 29048
    },
    {
      "epoch": 2.4992686913877655,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005005678553240047,
      "loss": 1.2564,
      "step": 29049
    },
    {
      "epoch": 2.499354727695087,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005005408146050989,
      "loss": 0.9023,
      "step": 29050
    },
    {
      "epoch": 2.4994407640024088,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005005137738846115,
      "loss": 1.3504,
      "step": 29051
    },
    {
      "epoch": 2.499526800309731,
      "grad_norm": 0.9375,
      "learning_rate": 0.0005004867331626214,
      "loss": 0.6964,
      "step": 29052
    },
    {
      "epoch": 2.4996128366170525,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005004596924392076,
      "loss": 1.3031,
      "step": 29053
    },
    {
      "epoch": 2.499698872924374,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005004326517144493,
      "loss": 1.4987,
      "step": 29054
    },
    {
      "epoch": 2.4997849092316957,
      "grad_norm": 1.078125,
      "learning_rate": 0.0005004056109884256,
      "loss": 0.8299,
      "step": 29055
    },
    {
      "epoch": 2.4998709455390173,
      "grad_norm": 1.046875,
      "learning_rate": 0.0005003785702612157,
      "loss": 1.0506,
      "step": 29056
    },
    {
      "epoch": 2.499956981846339,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0005003515295328983,
      "loss": 1.3039,
      "step": 29057
    },
    {
      "epoch": 2.5000430181536606,
      "grad_norm": 1.0625,
      "learning_rate": 0.000500324488803553,
      "loss": 1.1114,
      "step": 29058
    },
    {
      "epoch": 2.5001290544609827,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005002974480732585,
      "loss": 1.5409,
      "step": 29059
    },
    {
      "epoch": 2.5002150907683043,
      "grad_norm": 1.703125,
      "learning_rate": 0.000500270407342094,
      "loss": 1.6187,
      "step": 29060
    },
    {
      "epoch": 2.5002150907683043,
      "eval_loss": 1.1449687480926514,
      "eval_runtime": 374.9922,
      "eval_samples_per_second": 12.398,
      "eval_steps_per_second": 1.552,
      "step": 29060
    },
    {
      "epoch": 2.500301127075626,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0005002433666101386,
      "loss": 0.6465,
      "step": 29061
    },
    {
      "epoch": 2.5003871633829475,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0005002163258774713,
      "loss": 0.7357,
      "step": 29062
    },
    {
      "epoch": 2.500473199690269,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0005001892851441715,
      "loss": 1.6505,
      "step": 29063
    },
    {
      "epoch": 2.5005592359975912,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0005001622444103181,
      "loss": 1.2196,
      "step": 29064
    },
    {
      "epoch": 2.5006452723049124,
      "grad_norm": 1.234375,
      "learning_rate": 0.00050013520367599,
      "loss": 1.22,
      "step": 29065
    },
    {
      "epoch": 2.5007313086122345,
      "grad_norm": 1.109375,
      "learning_rate": 0.0005001081629412666,
      "loss": 2.0631,
      "step": 29066
    },
    {
      "epoch": 2.500817344919556,
      "grad_norm": 2.328125,
      "learning_rate": 0.0005000811222062268,
      "loss": 1.0015,
      "step": 29067
    },
    {
      "epoch": 2.5009033812268777,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0005000540814709497,
      "loss": 1.0685,
      "step": 29068
    },
    {
      "epoch": 2.5009894175341993,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005000270407355145,
      "loss": 1.1284,
      "step": 29069
    },
    {
      "epoch": 2.501075453841521,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005,
      "loss": 0.7688,
      "step": 29070
    },
    {
      "epoch": 2.501161490148843,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004999729592644857,
      "loss": 1.9116,
      "step": 29071
    },
    {
      "epoch": 2.5012475264561647,
      "grad_norm": 1.0,
      "learning_rate": 0.0004999459185290504,
      "loss": 0.9204,
      "step": 29072
    },
    {
      "epoch": 2.5013335627634863,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004999188777937733,
      "loss": 1.1306,
      "step": 29073
    },
    {
      "epoch": 2.501419599070808,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004998918370587336,
      "loss": 1.1732,
      "step": 29074
    },
    {
      "epoch": 2.5015056353781295,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004998647963240101,
      "loss": 1.9443,
      "step": 29075
    },
    {
      "epoch": 2.501591671685451,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000499837755589682,
      "loss": 1.1047,
      "step": 29076
    },
    {
      "epoch": 2.501677707992773,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004998107148558286,
      "loss": 1.1079,
      "step": 29077
    },
    {
      "epoch": 2.501763744300095,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004997836741225287,
      "loss": 0.9678,
      "step": 29078
    },
    {
      "epoch": 2.5018497806074165,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004997566333898615,
      "loss": 1.4192,
      "step": 29079
    },
    {
      "epoch": 2.501935816914738,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004997295926579061,
      "loss": 0.8376,
      "step": 29080
    },
    {
      "epoch": 2.5020218532220597,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004997025519267417,
      "loss": 1.4346,
      "step": 29081
    },
    {
      "epoch": 2.5021078895293813,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004996755111964472,
      "loss": 1.1778,
      "step": 29082
    },
    {
      "epoch": 2.502193925836703,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004996484704671017,
      "loss": 1.6057,
      "step": 29083
    },
    {
      "epoch": 2.5022799621440246,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004996214297387845,
      "loss": 1.8342,
      "step": 29084
    },
    {
      "epoch": 2.5023659984513467,
      "grad_norm": 1.53125,
      "learning_rate": 0.0004995943890115745,
      "loss": 1.3089,
      "step": 29085
    },
    {
      "epoch": 2.5024520347586683,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0004995673482855508,
      "loss": 1.1385,
      "step": 29086
    },
    {
      "epoch": 2.50253807106599,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004995403075607925,
      "loss": 1.3117,
      "step": 29087
    },
    {
      "epoch": 2.5026241073733115,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004995132668373788,
      "loss": 1.0723,
      "step": 29088
    },
    {
      "epoch": 2.502710143680633,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0004994862261153884,
      "loss": 0.6017,
      "step": 29089
    },
    {
      "epoch": 2.502796179987955,
      "grad_norm": 1.875,
      "learning_rate": 0.000499459185394901,
      "loss": 1.1192,
      "step": 29090
    },
    {
      "epoch": 2.5028822162952764,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004994321446759953,
      "loss": 0.9946,
      "step": 29091
    },
    {
      "epoch": 2.5029682526025985,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004994051039587504,
      "loss": 1.8217,
      "step": 29092
    },
    {
      "epoch": 2.50305428890992,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004993780632432456,
      "loss": 0.7148,
      "step": 29093
    },
    {
      "epoch": 2.5031403252172417,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004993510225295598,
      "loss": 0.6745,
      "step": 29094
    },
    {
      "epoch": 2.5032263615245633,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000499323981817772,
      "loss": 0.8696,
      "step": 29095
    },
    {
      "epoch": 2.503312397831885,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004992969411079616,
      "loss": 0.4514,
      "step": 29096
    },
    {
      "epoch": 2.5033984341392066,
      "grad_norm": 1.5625,
      "learning_rate": 0.0004992699004002075,
      "loss": 1.5915,
      "step": 29097
    },
    {
      "epoch": 2.503484470446528,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004992428596945887,
      "loss": 1.2632,
      "step": 29098
    },
    {
      "epoch": 2.5035705067538503,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004992158189911845,
      "loss": 0.9836,
      "step": 29099
    },
    {
      "epoch": 2.503656543061172,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004991887782900737,
      "loss": 1.113,
      "step": 29100
    },
    {
      "epoch": 2.5037425793684935,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004991617375913357,
      "loss": 0.7076,
      "step": 29101
    },
    {
      "epoch": 2.503828615675815,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004991346968950494,
      "loss": 1.1983,
      "step": 29102
    },
    {
      "epoch": 2.503914651983137,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000499107656201294,
      "loss": 0.4086,
      "step": 29103
    },
    {
      "epoch": 2.5040006882904584,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004990806155101485,
      "loss": 1.4144,
      "step": 29104
    },
    {
      "epoch": 2.50408672459778,
      "grad_norm": 1.203125,
      "learning_rate": 0.000499053574821692,
      "loss": 1.4195,
      "step": 29105
    },
    {
      "epoch": 2.504172760905102,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004990265341360036,
      "loss": 1.1781,
      "step": 29106
    },
    {
      "epoch": 2.5042587972124237,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004989994934531625,
      "loss": 0.9387,
      "step": 29107
    },
    {
      "epoch": 2.5043448335197454,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004989724527732476,
      "loss": 1.0042,
      "step": 29108
    },
    {
      "epoch": 2.504430869827067,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004989454120963382,
      "loss": 0.9441,
      "step": 29109
    },
    {
      "epoch": 2.5045169061343886,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000498918371422513,
      "loss": 1.398,
      "step": 29110
    },
    {
      "epoch": 2.5046029424417107,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004988913307518516,
      "loss": 1.1841,
      "step": 29111
    },
    {
      "epoch": 2.504688978749032,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004988642900844327,
      "loss": 0.9127,
      "step": 29112
    },
    {
      "epoch": 2.504775015056354,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004988372494203355,
      "loss": 1.2302,
      "step": 29113
    },
    {
      "epoch": 2.5048610513636755,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004988102087596393,
      "loss": 1.2285,
      "step": 29114
    },
    {
      "epoch": 2.504947087670997,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004987831681024229,
      "loss": 1.2545,
      "step": 29115
    },
    {
      "epoch": 2.505033123978319,
      "grad_norm": 1.484375,
      "learning_rate": 0.0004987561274487656,
      "loss": 0.9536,
      "step": 29116
    },
    {
      "epoch": 2.5051191602856404,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004987290867987462,
      "loss": 1.4913,
      "step": 29117
    },
    {
      "epoch": 2.5052051965929625,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004987020461524441,
      "loss": 1.4007,
      "step": 29118
    },
    {
      "epoch": 2.5052912329002837,
      "grad_norm": 1.671875,
      "learning_rate": 0.0004986750055099382,
      "loss": 1.5255,
      "step": 29119
    },
    {
      "epoch": 2.5053772692076057,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004986479648713077,
      "loss": 1.0181,
      "step": 29120
    },
    {
      "epoch": 2.5054633055149274,
      "grad_norm": 1.25,
      "learning_rate": 0.0004986209242366315,
      "loss": 1.8932,
      "step": 29121
    },
    {
      "epoch": 2.505549341822249,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000498593883605989,
      "loss": 1.4119,
      "step": 29122
    },
    {
      "epoch": 2.5056353781295706,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004985668429794591,
      "loss": 1.7299,
      "step": 29123
    },
    {
      "epoch": 2.5057214144368922,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004985398023571208,
      "loss": 0.7841,
      "step": 29124
    },
    {
      "epoch": 2.5058074507442143,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004985127617390535,
      "loss": 1.8456,
      "step": 29125
    },
    {
      "epoch": 2.5058934870515355,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004984857211253357,
      "loss": 0.9089,
      "step": 29126
    },
    {
      "epoch": 2.5059795233588575,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004984586805160472,
      "loss": 1.7869,
      "step": 29127
    },
    {
      "epoch": 2.506065559666179,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004984316399112667,
      "loss": 1.659,
      "step": 29128
    },
    {
      "epoch": 2.506151595973501,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004984045993110731,
      "loss": 1.3956,
      "step": 29129
    },
    {
      "epoch": 2.5062376322808224,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004983775587155461,
      "loss": 1.4134,
      "step": 29130
    },
    {
      "epoch": 2.506323668588144,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004983505181247643,
      "loss": 1.193,
      "step": 29131
    },
    {
      "epoch": 2.506409704895466,
      "grad_norm": 1.359375,
      "learning_rate": 0.000498323477538807,
      "loss": 1.3387,
      "step": 29132
    },
    {
      "epoch": 2.5064957412027877,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004982964369577529,
      "loss": 0.9607,
      "step": 29133
    },
    {
      "epoch": 2.5065817775101094,
      "grad_norm": 1.125,
      "learning_rate": 0.0004982693963816817,
      "loss": 1.329,
      "step": 29134
    },
    {
      "epoch": 2.506667813817431,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004982423558106721,
      "loss": 1.1929,
      "step": 29135
    },
    {
      "epoch": 2.5067538501247526,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004982153152448032,
      "loss": 1.6175,
      "step": 29136
    },
    {
      "epoch": 2.5068398864320742,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0004981882746841542,
      "loss": 1.9159,
      "step": 29137
    },
    {
      "epoch": 2.506925922739396,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004981612341288041,
      "loss": 0.846,
      "step": 29138
    },
    {
      "epoch": 2.507011959046718,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004981341935788321,
      "loss": 0.9782,
      "step": 29139
    },
    {
      "epoch": 2.5070979953540395,
      "grad_norm": 1.25,
      "learning_rate": 0.0004981071530343172,
      "loss": 0.8276,
      "step": 29140
    },
    {
      "epoch": 2.507184031661361,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004980801124953385,
      "loss": 0.9226,
      "step": 29141
    },
    {
      "epoch": 2.507270067968683,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0004980530719619752,
      "loss": 0.9733,
      "step": 29142
    },
    {
      "epoch": 2.5073561042760044,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000498026031434306,
      "loss": 1.2998,
      "step": 29143
    },
    {
      "epoch": 2.507442140583326,
      "grad_norm": 1.25,
      "learning_rate": 0.0004979989909124105,
      "loss": 0.8387,
      "step": 29144
    },
    {
      "epoch": 2.5075281768906477,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0004979719503963677,
      "loss": 0.6816,
      "step": 29145
    },
    {
      "epoch": 2.5076142131979697,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004979449098862562,
      "loss": 0.6916,
      "step": 29146
    },
    {
      "epoch": 2.5077002495052914,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004979178693821556,
      "loss": 1.1075,
      "step": 29147
    },
    {
      "epoch": 2.507786285812613,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004978908288841449,
      "loss": 0.8905,
      "step": 29148
    },
    {
      "epoch": 2.5078723221199346,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004978637883923031,
      "loss": 1.4845,
      "step": 29149
    },
    {
      "epoch": 2.5079583584272562,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004978367479067093,
      "loss": 1.9529,
      "step": 29150
    },
    {
      "epoch": 2.508044394734578,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004978097074274425,
      "loss": 1.3798,
      "step": 29151
    },
    {
      "epoch": 2.5081304310418995,
      "grad_norm": 0.875,
      "learning_rate": 0.000497782666954582,
      "loss": 1.3671,
      "step": 29152
    },
    {
      "epoch": 2.5082164673492215,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004977556264882067,
      "loss": 0.8486,
      "step": 29153
    },
    {
      "epoch": 2.508302503656543,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004977285860283956,
      "loss": 1.435,
      "step": 29154
    },
    {
      "epoch": 2.508388539963865,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004977015455752281,
      "loss": 1.2043,
      "step": 29155
    },
    {
      "epoch": 2.5084745762711864,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004976745051287832,
      "loss": 1.5471,
      "step": 29156
    },
    {
      "epoch": 2.508560612578508,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004976474646891399,
      "loss": 0.5932,
      "step": 29157
    },
    {
      "epoch": 2.5086466488858297,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004976204242563771,
      "loss": 1.0125,
      "step": 29158
    },
    {
      "epoch": 2.5087326851931513,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004975933838305744,
      "loss": 0.8116,
      "step": 29159
    },
    {
      "epoch": 2.5088187215004734,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004975663434118104,
      "loss": 0.7495,
      "step": 29160
    },
    {
      "epoch": 2.508904757807795,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004975393030001643,
      "loss": 1.0181,
      "step": 29161
    },
    {
      "epoch": 2.5089907941151166,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004975122625957154,
      "loss": 1.2601,
      "step": 29162
    },
    {
      "epoch": 2.5090768304224382,
      "grad_norm": 1.375,
      "learning_rate": 0.0004974852221985424,
      "loss": 2.0703,
      "step": 29163
    },
    {
      "epoch": 2.50916286672976,
      "grad_norm": 1.421875,
      "learning_rate": 0.000497458181808725,
      "loss": 1.4058,
      "step": 29164
    },
    {
      "epoch": 2.509248903037082,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004974311414263418,
      "loss": 1.1767,
      "step": 29165
    },
    {
      "epoch": 2.509334939344403,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004974041010514719,
      "loss": 0.6256,
      "step": 29166
    },
    {
      "epoch": 2.509420975651725,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004973770606841947,
      "loss": 1.1633,
      "step": 29167
    },
    {
      "epoch": 2.509507011959047,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004973500203245889,
      "loss": 1.0573,
      "step": 29168
    },
    {
      "epoch": 2.5095930482663684,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000497322979972734,
      "loss": 1.0402,
      "step": 29169
    },
    {
      "epoch": 2.50967908457369,
      "grad_norm": 1.125,
      "learning_rate": 0.0004972959396287085,
      "loss": 1.0568,
      "step": 29170
    },
    {
      "epoch": 2.5097651208810117,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004972688992925921,
      "loss": 1.2215,
      "step": 29171
    },
    {
      "epoch": 2.5098511571883337,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004972418589644636,
      "loss": 1.4831,
      "step": 29172
    },
    {
      "epoch": 2.509937193495655,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004972148186444022,
      "loss": 1.3161,
      "step": 29173
    },
    {
      "epoch": 2.510023229802977,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004971877783324868,
      "loss": 1.2277,
      "step": 29174
    },
    {
      "epoch": 2.5101092661102986,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004971607380287967,
      "loss": 1.2592,
      "step": 29175
    },
    {
      "epoch": 2.5101953024176202,
      "grad_norm": 1.484375,
      "learning_rate": 0.0004971336977334108,
      "loss": 1.0882,
      "step": 29176
    },
    {
      "epoch": 2.510281338724942,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004971066574464083,
      "loss": 1.6378,
      "step": 29177
    },
    {
      "epoch": 2.5103673750322635,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004970796171678683,
      "loss": 0.7371,
      "step": 29178
    },
    {
      "epoch": 2.5104534113395855,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004970525768978699,
      "loss": 1.9921,
      "step": 29179
    },
    {
      "epoch": 2.5105394476469067,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004970255366364921,
      "loss": 0.4948,
      "step": 29180
    },
    {
      "epoch": 2.510625483954229,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004969984963838138,
      "loss": 1.5711,
      "step": 29181
    },
    {
      "epoch": 2.5107115202615504,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004969714561399146,
      "loss": 0.5646,
      "step": 29182
    },
    {
      "epoch": 2.510797556568872,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004969444159048733,
      "loss": 0.8028,
      "step": 29183
    },
    {
      "epoch": 2.5108835928761937,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004969173756787687,
      "loss": 1.2763,
      "step": 29184
    },
    {
      "epoch": 2.5109696291835153,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004968903354616805,
      "loss": 1.8493,
      "step": 29185
    },
    {
      "epoch": 2.5110556654908374,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0004968632952536873,
      "loss": 1.1946,
      "step": 29186
    },
    {
      "epoch": 2.511141701798159,
      "grad_norm": 1.578125,
      "learning_rate": 0.0004968362550548685,
      "loss": 1.7317,
      "step": 29187
    },
    {
      "epoch": 2.5112277381054806,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004968092148653028,
      "loss": 1.5076,
      "step": 29188
    },
    {
      "epoch": 2.5113137744128022,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004967821746850697,
      "loss": 0.7575,
      "step": 29189
    },
    {
      "epoch": 2.511399810720124,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004967551345142481,
      "loss": 1.1711,
      "step": 29190
    },
    {
      "epoch": 2.5114858470274455,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004967280943529171,
      "loss": 1.0724,
      "step": 29191
    },
    {
      "epoch": 2.511571883334767,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004967010542011557,
      "loss": 1.2016,
      "step": 29192
    },
    {
      "epoch": 2.511657919642089,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004966740140590432,
      "loss": 1.8919,
      "step": 29193
    },
    {
      "epoch": 2.511743955949411,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004966469739266585,
      "loss": 0.7349,
      "step": 29194
    },
    {
      "epoch": 2.5118299922567324,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004966199338040806,
      "loss": 1.0277,
      "step": 29195
    },
    {
      "epoch": 2.511916028564054,
      "grad_norm": 1.375,
      "learning_rate": 0.000496592893691389,
      "loss": 2.1972,
      "step": 29196
    },
    {
      "epoch": 2.5120020648713757,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004965658535886625,
      "loss": 1.9107,
      "step": 29197
    },
    {
      "epoch": 2.5120881011786973,
      "grad_norm": 1.3125,
      "learning_rate": 0.00049653881349598,
      "loss": 0.968,
      "step": 29198
    },
    {
      "epoch": 2.512174137486019,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004965117734134208,
      "loss": 0.9015,
      "step": 29199
    },
    {
      "epoch": 2.512260173793341,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004964847333410639,
      "loss": 1.4564,
      "step": 29200
    },
    {
      "epoch": 2.5123462101006626,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004964576932789887,
      "loss": 0.8093,
      "step": 29201
    },
    {
      "epoch": 2.5124322464079842,
      "grad_norm": 1.25,
      "learning_rate": 0.0004964306532272741,
      "loss": 1.1326,
      "step": 29202
    },
    {
      "epoch": 2.512518282715306,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004964036131859991,
      "loss": 1.5664,
      "step": 29203
    },
    {
      "epoch": 2.5126043190226275,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004963765731552428,
      "loss": 1.871,
      "step": 29204
    },
    {
      "epoch": 2.512690355329949,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004963495331350842,
      "loss": 1.3378,
      "step": 29205
    },
    {
      "epoch": 2.5127763916372707,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004963224931256027,
      "loss": 1.387,
      "step": 29206
    },
    {
      "epoch": 2.512862427944593,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004962954531268772,
      "loss": 1.4789,
      "step": 29207
    },
    {
      "epoch": 2.5129484642519144,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004962684131389866,
      "loss": 1.6775,
      "step": 29208
    },
    {
      "epoch": 2.513034500559236,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004962413731620103,
      "loss": 1.1692,
      "step": 29209
    },
    {
      "epoch": 2.5131205368665577,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0004962143331960272,
      "loss": 1.5198,
      "step": 29210
    },
    {
      "epoch": 2.5132065731738793,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0004961872932411164,
      "loss": 0.5961,
      "step": 29211
    },
    {
      "epoch": 2.513292609481201,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004961602532973571,
      "loss": 1.0765,
      "step": 29212
    },
    {
      "epoch": 2.5133786457885225,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004961332133648285,
      "loss": 1.0668,
      "step": 29213
    },
    {
      "epoch": 2.5134646820958446,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004961061734436091,
      "loss": 1.2115,
      "step": 29214
    },
    {
      "epoch": 2.5135507184031662,
      "grad_norm": 1.375,
      "learning_rate": 0.0004960791335337786,
      "loss": 1.6458,
      "step": 29215
    },
    {
      "epoch": 2.513636754710488,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004960520936354159,
      "loss": 0.9877,
      "step": 29216
    },
    {
      "epoch": 2.5137227910178095,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004960250537486001,
      "loss": 1.2308,
      "step": 29217
    },
    {
      "epoch": 2.513808827325131,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00049599801387341,
      "loss": 1.1205,
      "step": 29218
    },
    {
      "epoch": 2.5138948636324527,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004959709740099251,
      "loss": 1.2197,
      "step": 29219
    },
    {
      "epoch": 2.5139808999397744,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004959439341582244,
      "loss": 1.4309,
      "step": 29220
    },
    {
      "epoch": 2.5140669362470964,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004959168943183868,
      "loss": 0.6276,
      "step": 29221
    },
    {
      "epoch": 2.514152972554418,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004958898544904914,
      "loss": 0.9218,
      "step": 29222
    },
    {
      "epoch": 2.5142390088617397,
      "grad_norm": 1.75,
      "learning_rate": 0.0004958628146746175,
      "loss": 1.4758,
      "step": 29223
    },
    {
      "epoch": 2.5143250451690613,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004958357748708441,
      "loss": 1.0089,
      "step": 29224
    },
    {
      "epoch": 2.514411081476383,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004958087350792501,
      "loss": 1.1298,
      "step": 29225
    },
    {
      "epoch": 2.514497117783705,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004957816952999148,
      "loss": 0.7009,
      "step": 29226
    },
    {
      "epoch": 2.514583154091026,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004957546555329173,
      "loss": 1.2739,
      "step": 29227
    },
    {
      "epoch": 2.5146691903983482,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004957276157783366,
      "loss": 1.2005,
      "step": 29228
    },
    {
      "epoch": 2.51475522670567,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004957005760362516,
      "loss": 0.8816,
      "step": 29229
    },
    {
      "epoch": 2.5148412630129915,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004956735363067418,
      "loss": 1.5314,
      "step": 29230
    },
    {
      "epoch": 2.514927299320313,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004956464965898859,
      "loss": 1.355,
      "step": 29231
    },
    {
      "epoch": 2.5150133356276347,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004956194568857633,
      "loss": 1.823,
      "step": 29232
    },
    {
      "epoch": 2.515099371934957,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004955924171944528,
      "loss": 0.8909,
      "step": 29233
    },
    {
      "epoch": 2.515185408242278,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004955653775160337,
      "loss": 0.6842,
      "step": 29234
    },
    {
      "epoch": 2.5152714445496,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004955383378505851,
      "loss": 0.8612,
      "step": 29235
    },
    {
      "epoch": 2.5153574808569217,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004955112981981858,
      "loss": 1.2107,
      "step": 29236
    },
    {
      "epoch": 2.5154435171642433,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004954842585589151,
      "loss": 1.1979,
      "step": 29237
    },
    {
      "epoch": 2.515529553471565,
      "grad_norm": 1.09375,
      "learning_rate": 0.000495457218932852,
      "loss": 1.5941,
      "step": 29238
    },
    {
      "epoch": 2.5156155897788866,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004954301793200759,
      "loss": 1.1742,
      "step": 29239
    },
    {
      "epoch": 2.5157016260862086,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004954031397206655,
      "loss": 1.1917,
      "step": 29240
    },
    {
      "epoch": 2.51578766239353,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004953761001347001,
      "loss": 1.0214,
      "step": 29241
    },
    {
      "epoch": 2.515873698700852,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004953490605622586,
      "loss": 1.3042,
      "step": 29242
    },
    {
      "epoch": 2.5159597350081735,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004953220210034203,
      "loss": 1.4141,
      "step": 29243
    },
    {
      "epoch": 2.516045771315495,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004952949814582641,
      "loss": 1.7244,
      "step": 29244
    },
    {
      "epoch": 2.5161318076228167,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004952679419268692,
      "loss": 1.2023,
      "step": 29245
    },
    {
      "epoch": 2.5162178439301384,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004952409024093147,
      "loss": 0.7336,
      "step": 29246
    },
    {
      "epoch": 2.5163038802374604,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004952138629056795,
      "loss": 1.5237,
      "step": 29247
    },
    {
      "epoch": 2.516389916544782,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004951868234160428,
      "loss": 1.1278,
      "step": 29248
    },
    {
      "epoch": 2.5164759528521037,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0004951597839404839,
      "loss": 1.0141,
      "step": 29249
    },
    {
      "epoch": 2.5165619891594253,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004951327444790816,
      "loss": 1.2579,
      "step": 29250
    },
    {
      "epoch": 2.516648025466747,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004951057050319151,
      "loss": 1.6396,
      "step": 29251
    },
    {
      "epoch": 2.5167340617740686,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004950786655990633,
      "loss": 0.5009,
      "step": 29252
    },
    {
      "epoch": 2.51682009808139,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004950516261806057,
      "loss": 1.7573,
      "step": 29253
    },
    {
      "epoch": 2.5169061343887122,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004950245867766209,
      "loss": 1.7786,
      "step": 29254
    },
    {
      "epoch": 2.516992170696034,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004949975473871884,
      "loss": 1.5516,
      "step": 29255
    },
    {
      "epoch": 2.5170782070033555,
      "grad_norm": 1.328125,
      "learning_rate": 0.000494970508012387,
      "loss": 1.5165,
      "step": 29256
    },
    {
      "epoch": 2.517164243310677,
      "grad_norm": 1.359375,
      "learning_rate": 0.000494943468652296,
      "loss": 1.5786,
      "step": 29257
    },
    {
      "epoch": 2.5172502796179987,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004949164293069942,
      "loss": 1.501,
      "step": 29258
    },
    {
      "epoch": 2.5173363159253204,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004948893899765609,
      "loss": 0.8342,
      "step": 29259
    },
    {
      "epoch": 2.517422352232642,
      "grad_norm": 1.25,
      "learning_rate": 0.0004948623506610753,
      "loss": 1.1505,
      "step": 29260
    },
    {
      "epoch": 2.517508388539964,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004948353113606161,
      "loss": 0.9408,
      "step": 29261
    },
    {
      "epoch": 2.5175944248472857,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004948082720752628,
      "loss": 1.5317,
      "step": 29262
    },
    {
      "epoch": 2.5176804611546073,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004947812328050941,
      "loss": 0.927,
      "step": 29263
    },
    {
      "epoch": 2.517766497461929,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004947541935501895,
      "loss": 1.0517,
      "step": 29264
    },
    {
      "epoch": 2.5178525337692506,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004947271543106277,
      "loss": 1.0299,
      "step": 29265
    },
    {
      "epoch": 2.517938570076572,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004947001150864879,
      "loss": 0.6699,
      "step": 29266
    },
    {
      "epoch": 2.518024606383894,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004946730758778494,
      "loss": 0.8413,
      "step": 29267
    },
    {
      "epoch": 2.518110642691216,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004946460366847911,
      "loss": 1.2557,
      "step": 29268
    },
    {
      "epoch": 2.5181966789985375,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004946189975073921,
      "loss": 0.9275,
      "step": 29269
    },
    {
      "epoch": 2.518282715305859,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004945919583457313,
      "loss": 0.7251,
      "step": 29270
    },
    {
      "epoch": 2.5183687516131807,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004945649191998882,
      "loss": 1.1357,
      "step": 29271
    },
    {
      "epoch": 2.5184547879205024,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004945378800699416,
      "loss": 0.9737,
      "step": 29272
    },
    {
      "epoch": 2.518540824227824,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004945108409559704,
      "loss": 1.1302,
      "step": 29273
    },
    {
      "epoch": 2.5186268605351456,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004944838018580541,
      "loss": 1.1803,
      "step": 29274
    },
    {
      "epoch": 2.5187128968424677,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004944567627762715,
      "loss": 1.4712,
      "step": 29275
    },
    {
      "epoch": 2.5187989331497893,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000494429723710702,
      "loss": 1.3601,
      "step": 29276
    },
    {
      "epoch": 2.518884969457111,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004944026846614244,
      "loss": 1.6541,
      "step": 29277
    },
    {
      "epoch": 2.5189710057644326,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004943756456285178,
      "loss": 1.3303,
      "step": 29278
    },
    {
      "epoch": 2.519057042071754,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004943486066120614,
      "loss": 1.7801,
      "step": 29279
    },
    {
      "epoch": 2.519143078379076,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004943215676121343,
      "loss": 1.0508,
      "step": 29280
    },
    {
      "epoch": 2.5192291146863974,
      "grad_norm": 1.796875,
      "learning_rate": 0.0004942945286288154,
      "loss": 0.7855,
      "step": 29281
    },
    {
      "epoch": 2.5193151509937195,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004942674896621838,
      "loss": 1.114,
      "step": 29282
    },
    {
      "epoch": 2.519401187301041,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004942404507123189,
      "loss": 1.3423,
      "step": 29283
    },
    {
      "epoch": 2.5194872236083627,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004942134117792994,
      "loss": 0.5826,
      "step": 29284
    },
    {
      "epoch": 2.5195732599156844,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004941863728632047,
      "loss": 1.0247,
      "step": 29285
    },
    {
      "epoch": 2.519659296223006,
      "grad_norm": 1.546875,
      "learning_rate": 0.0004941593339641135,
      "loss": 1.7113,
      "step": 29286
    },
    {
      "epoch": 2.519745332530328,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0004941322950821053,
      "loss": 1.0161,
      "step": 29287
    },
    {
      "epoch": 2.5198313688376492,
      "grad_norm": 1.5,
      "learning_rate": 0.000494105256217259,
      "loss": 1.2762,
      "step": 29288
    },
    {
      "epoch": 2.5199174051449713,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004940782173696535,
      "loss": 0.9747,
      "step": 29289
    },
    {
      "epoch": 2.520003441452293,
      "grad_norm": 1.578125,
      "learning_rate": 0.0004940511785393683,
      "loss": 1.7154,
      "step": 29290
    },
    {
      "epoch": 2.5200894777596146,
      "grad_norm": 1.0,
      "learning_rate": 0.0004940241397264821,
      "loss": 1.3552,
      "step": 29291
    },
    {
      "epoch": 2.520175514066936,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004939971009310743,
      "loss": 1.487,
      "step": 29292
    },
    {
      "epoch": 2.520261550374258,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004939700621532236,
      "loss": 0.8833,
      "step": 29293
    },
    {
      "epoch": 2.52034758668158,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004939430233930094,
      "loss": 1.334,
      "step": 29294
    },
    {
      "epoch": 2.520433622988901,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004939159846505106,
      "loss": 1.5143,
      "step": 29295
    },
    {
      "epoch": 2.520519659296223,
      "grad_norm": 1.625,
      "learning_rate": 0.0004938889459258064,
      "loss": 1.5934,
      "step": 29296
    },
    {
      "epoch": 2.5206056956035447,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004938619072189758,
      "loss": 1.284,
      "step": 29297
    },
    {
      "epoch": 2.5206917319108664,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004938348685300981,
      "loss": 1.2848,
      "step": 29298
    },
    {
      "epoch": 2.520777768218188,
      "grad_norm": 1.125,
      "learning_rate": 0.000493807829859252,
      "loss": 0.58,
      "step": 29299
    },
    {
      "epoch": 2.5208638045255096,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004937807912065168,
      "loss": 1.8122,
      "step": 29300
    },
    {
      "epoch": 2.5209498408328317,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004937537525719717,
      "loss": 1.1123,
      "step": 29301
    },
    {
      "epoch": 2.521035877140153,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004937267139556957,
      "loss": 1.2456,
      "step": 29302
    },
    {
      "epoch": 2.521121913447475,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004936996753577678,
      "loss": 1.0738,
      "step": 29303
    },
    {
      "epoch": 2.5212079497547966,
      "grad_norm": 1.125,
      "learning_rate": 0.0004936726367782671,
      "loss": 1.6941,
      "step": 29304
    },
    {
      "epoch": 2.521293986062118,
      "grad_norm": 1.484375,
      "learning_rate": 0.0004936455982172726,
      "loss": 1.3839,
      "step": 29305
    },
    {
      "epoch": 2.52138002236944,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004936185596748637,
      "loss": 1.4707,
      "step": 29306
    },
    {
      "epoch": 2.5214660586767614,
      "grad_norm": 1.078125,
      "learning_rate": 0.000493591521151119,
      "loss": 1.1797,
      "step": 29307
    },
    {
      "epoch": 2.5215520949840835,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000493564482646118,
      "loss": 0.9152,
      "step": 29308
    },
    {
      "epoch": 2.521638131291405,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004935374441599397,
      "loss": 1.1039,
      "step": 29309
    },
    {
      "epoch": 2.5217241675987268,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004935104056926631,
      "loss": 1.4229,
      "step": 29310
    },
    {
      "epoch": 2.5218102039060484,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004934833672443672,
      "loss": 1.0569,
      "step": 29311
    },
    {
      "epoch": 2.52189624021337,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004934563288151311,
      "loss": 1.1481,
      "step": 29312
    },
    {
      "epoch": 2.5219822765206916,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004934292904050341,
      "loss": 1.2713,
      "step": 29313
    },
    {
      "epoch": 2.5220683128280132,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004934022520141552,
      "loss": 0.8083,
      "step": 29314
    },
    {
      "epoch": 2.5221543491353353,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004933752136425735,
      "loss": 1.0484,
      "step": 29315
    },
    {
      "epoch": 2.522240385442657,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004933481752903677,
      "loss": 1.6697,
      "step": 29316
    },
    {
      "epoch": 2.5223264217499786,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0004933211369576175,
      "loss": 1.1596,
      "step": 29317
    },
    {
      "epoch": 2.5224124580573,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004932940986444016,
      "loss": 1.0092,
      "step": 29318
    },
    {
      "epoch": 2.522498494364622,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000493267060350799,
      "loss": 1.7748,
      "step": 29319
    },
    {
      "epoch": 2.5225845306719434,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004932400220768891,
      "loss": 1.541,
      "step": 29320
    },
    {
      "epoch": 2.522670566979265,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004932129838227508,
      "loss": 1.5176,
      "step": 29321
    },
    {
      "epoch": 2.522756603286587,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0004931859455884633,
      "loss": 1.3181,
      "step": 29322
    },
    {
      "epoch": 2.5228426395939088,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004931589073741054,
      "loss": 0.9832,
      "step": 29323
    },
    {
      "epoch": 2.5229286759012304,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004931318691797564,
      "loss": 0.6214,
      "step": 29324
    },
    {
      "epoch": 2.523014712208552,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004931048310054955,
      "loss": 1.1182,
      "step": 29325
    },
    {
      "epoch": 2.5231007485158736,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004930777928514015,
      "loss": 1.405,
      "step": 29326
    },
    {
      "epoch": 2.5231867848231953,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004930507547175535,
      "loss": 0.8042,
      "step": 29327
    },
    {
      "epoch": 2.523272821130517,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004930237166040309,
      "loss": 0.8473,
      "step": 29328
    },
    {
      "epoch": 2.523358857437839,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004929966785109125,
      "loss": 1.0531,
      "step": 29329
    },
    {
      "epoch": 2.5234448937451606,
      "grad_norm": 1.0,
      "learning_rate": 0.0004929696404382774,
      "loss": 1.0126,
      "step": 29330
    },
    {
      "epoch": 2.523530930052482,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004929426023862047,
      "loss": 1.2159,
      "step": 29331
    },
    {
      "epoch": 2.523616966359804,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004929155643547737,
      "loss": 1.8367,
      "step": 29332
    },
    {
      "epoch": 2.5237030026671254,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004928885263440632,
      "loss": 1.2805,
      "step": 29333
    },
    {
      "epoch": 2.523789038974447,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0004928614883541523,
      "loss": 1.0787,
      "step": 29334
    },
    {
      "epoch": 2.5238750752817687,
      "grad_norm": 2.0625,
      "learning_rate": 0.0004928344503851203,
      "loss": 1.5777,
      "step": 29335
    },
    {
      "epoch": 2.5239611115890908,
      "grad_norm": 1.03125,
      "learning_rate": 0.000492807412437046,
      "loss": 1.013,
      "step": 29336
    },
    {
      "epoch": 2.5240471478964124,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004927803745100087,
      "loss": 0.7175,
      "step": 29337
    },
    {
      "epoch": 2.524133184203734,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004927533366040873,
      "loss": 0.9452,
      "step": 29338
    },
    {
      "epoch": 2.5242192205110556,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004927262987193612,
      "loss": 1.1847,
      "step": 29339
    },
    {
      "epoch": 2.5243052568183773,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004926992608559091,
      "loss": 1.5313,
      "step": 29340
    },
    {
      "epoch": 2.5243912931256993,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004926722230138102,
      "loss": 0.8522,
      "step": 29341
    },
    {
      "epoch": 2.5244773294330205,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004926451851931437,
      "loss": 1.7861,
      "step": 29342
    },
    {
      "epoch": 2.5245633657403426,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004926181473939886,
      "loss": 1.7138,
      "step": 29343
    },
    {
      "epoch": 2.524649402047664,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004925911096164239,
      "loss": 0.5231,
      "step": 29344
    },
    {
      "epoch": 2.524735438354986,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004925640718605289,
      "loss": 1.7028,
      "step": 29345
    },
    {
      "epoch": 2.5248214746623074,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004925370341263825,
      "loss": 1.7951,
      "step": 29346
    },
    {
      "epoch": 2.524907510969629,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004925099964140638,
      "loss": 1.5731,
      "step": 29347
    },
    {
      "epoch": 2.524993547276951,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004924829587236518,
      "loss": 0.7115,
      "step": 29348
    },
    {
      "epoch": 2.5250795835842723,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004924559210552258,
      "loss": 0.7897,
      "step": 29349
    },
    {
      "epoch": 2.5251656198915944,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0004924288834088647,
      "loss": 1.1065,
      "step": 29350
    },
    {
      "epoch": 2.525251656198916,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004924018457846477,
      "loss": 1.4279,
      "step": 29351
    },
    {
      "epoch": 2.5253376925062376,
      "grad_norm": 1.125,
      "learning_rate": 0.0004923748081826539,
      "loss": 1.062,
      "step": 29352
    },
    {
      "epoch": 2.5254237288135593,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000492347770602962,
      "loss": 0.8733,
      "step": 29353
    },
    {
      "epoch": 2.525509765120881,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004923207330456518,
      "loss": 1.5278,
      "step": 29354
    },
    {
      "epoch": 2.525595801428203,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004922936955108018,
      "loss": 1.1129,
      "step": 29355
    },
    {
      "epoch": 2.525681837735524,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004922666579984912,
      "loss": 1.4493,
      "step": 29356
    },
    {
      "epoch": 2.525767874042846,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004922396205087991,
      "loss": 1.1654,
      "step": 29357
    },
    {
      "epoch": 2.525853910350168,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004922125830418046,
      "loss": 1.3684,
      "step": 29358
    },
    {
      "epoch": 2.5259399466574894,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004921855455975869,
      "loss": 1.4701,
      "step": 29359
    },
    {
      "epoch": 2.526025982964811,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004921585081762248,
      "loss": 0.9888,
      "step": 29360
    },
    {
      "epoch": 2.5261120192721327,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004921314707777976,
      "loss": 0.8153,
      "step": 29361
    },
    {
      "epoch": 2.5261980555794548,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004921044334023843,
      "loss": 0.6615,
      "step": 29362
    },
    {
      "epoch": 2.5262840918867764,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004920773960500641,
      "loss": 1.6805,
      "step": 29363
    },
    {
      "epoch": 2.526370128194098,
      "grad_norm": 1.0,
      "learning_rate": 0.0004920503587209159,
      "loss": 1.1686,
      "step": 29364
    },
    {
      "epoch": 2.5264561645014196,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004920233214150189,
      "loss": 0.8026,
      "step": 29365
    },
    {
      "epoch": 2.5265422008087413,
      "grad_norm": 1.09375,
      "learning_rate": 0.000491996284132452,
      "loss": 1.4788,
      "step": 29366
    },
    {
      "epoch": 2.526628237116063,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004919692468732947,
      "loss": 0.8192,
      "step": 29367
    },
    {
      "epoch": 2.5267142734233845,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004919422096376254,
      "loss": 1.1022,
      "step": 29368
    },
    {
      "epoch": 2.5268003097307066,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004919151724255238,
      "loss": 1.9111,
      "step": 29369
    },
    {
      "epoch": 2.526886346038028,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004918881352370687,
      "loss": 1.6527,
      "step": 29370
    },
    {
      "epoch": 2.52697238234535,
      "grad_norm": 1.78125,
      "learning_rate": 0.0004918610980723392,
      "loss": 1.5968,
      "step": 29371
    },
    {
      "epoch": 2.5270584186526714,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004918340609314146,
      "loss": 0.7387,
      "step": 29372
    },
    {
      "epoch": 2.527144454959993,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004918070238143736,
      "loss": 1.0137,
      "step": 29373
    },
    {
      "epoch": 2.5272304912673147,
      "grad_norm": 1.421875,
      "learning_rate": 0.0004917799867212955,
      "loss": 1.3398,
      "step": 29374
    },
    {
      "epoch": 2.5273165275746363,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004917529496522592,
      "loss": 0.9363,
      "step": 29375
    },
    {
      "epoch": 2.5274025638819584,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0004917259126073441,
      "loss": 1.343,
      "step": 29376
    },
    {
      "epoch": 2.52748860018928,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004916988755866291,
      "loss": 1.4108,
      "step": 29377
    },
    {
      "epoch": 2.5275746364966016,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004916718385901931,
      "loss": 1.3102,
      "step": 29378
    },
    {
      "epoch": 2.5276606728039233,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004916448016181155,
      "loss": 1.1403,
      "step": 29379
    },
    {
      "epoch": 2.527746709111245,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004916177646704752,
      "loss": 0.8441,
      "step": 29380
    },
    {
      "epoch": 2.5278327454185665,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004915907277473514,
      "loss": 0.8905,
      "step": 29381
    },
    {
      "epoch": 2.527918781725888,
      "grad_norm": 1.203125,
      "learning_rate": 0.000491563690848823,
      "loss": 1.5654,
      "step": 29382
    },
    {
      "epoch": 2.52800481803321,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004915366539749691,
      "loss": 1.5366,
      "step": 29383
    },
    {
      "epoch": 2.528090854340532,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000491509617125869,
      "loss": 0.8656,
      "step": 29384
    },
    {
      "epoch": 2.5281768906478534,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004914825803016013,
      "loss": 1.463,
      "step": 29385
    },
    {
      "epoch": 2.528262926955175,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004914555435022456,
      "loss": 1.1016,
      "step": 29386
    },
    {
      "epoch": 2.5283489632624967,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004914285067278807,
      "loss": 0.8342,
      "step": 29387
    },
    {
      "epoch": 2.5284349995698183,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004914014699785859,
      "loss": 0.8404,
      "step": 29388
    },
    {
      "epoch": 2.52852103587714,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004913744332544401,
      "loss": 0.9708,
      "step": 29389
    },
    {
      "epoch": 2.528607072184462,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004913473965555224,
      "loss": 1.1374,
      "step": 29390
    },
    {
      "epoch": 2.5286931084917836,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004913203598819118,
      "loss": 1.6334,
      "step": 29391
    },
    {
      "epoch": 2.5287791447991053,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004912933232336875,
      "loss": 0.9257,
      "step": 29392
    },
    {
      "epoch": 2.528865181106427,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004912662866109286,
      "loss": 1.2534,
      "step": 29393
    },
    {
      "epoch": 2.5289512174137485,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000491239250013714,
      "loss": 1.2887,
      "step": 29394
    },
    {
      "epoch": 2.52903725372107,
      "grad_norm": 2.046875,
      "learning_rate": 0.000491212213442123,
      "loss": 1.649,
      "step": 29395
    },
    {
      "epoch": 2.5291232900283918,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004911851768962345,
      "loss": 1.1578,
      "step": 29396
    },
    {
      "epoch": 2.529209326335714,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004911581403761277,
      "loss": 2.3593,
      "step": 29397
    },
    {
      "epoch": 2.5292953626430354,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0004911311038818815,
      "loss": 0.509,
      "step": 29398
    },
    {
      "epoch": 2.529381398950357,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004911040674135753,
      "loss": 0.8669,
      "step": 29399
    },
    {
      "epoch": 2.5294674352576787,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004910770309712878,
      "loss": 0.9788,
      "step": 29400
    },
    {
      "epoch": 2.5295534715650003,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004910499945550982,
      "loss": 1.256,
      "step": 29401
    },
    {
      "epoch": 2.5296395078723224,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004910229581650858,
      "loss": 1.4901,
      "step": 29402
    },
    {
      "epoch": 2.5297255441796436,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004909959218013294,
      "loss": 1.3899,
      "step": 29403
    },
    {
      "epoch": 2.5298115804869656,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004909688854639083,
      "loss": 0.9454,
      "step": 29404
    },
    {
      "epoch": 2.5298976167942873,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0004909418491529013,
      "loss": 0.8251,
      "step": 29405
    },
    {
      "epoch": 2.529983653101609,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004909148128683877,
      "loss": 0.807,
      "step": 29406
    },
    {
      "epoch": 2.5300696894089305,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004908877766104465,
      "loss": 0.9841,
      "step": 29407
    },
    {
      "epoch": 2.530155725716252,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004908607403791567,
      "loss": 1.2631,
      "step": 29408
    },
    {
      "epoch": 2.530241762023574,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004908337041745976,
      "loss": 1.0219,
      "step": 29409
    },
    {
      "epoch": 2.5303277983308954,
      "grad_norm": 1.140625,
      "learning_rate": 0.000490806667996848,
      "loss": 1.5296,
      "step": 29410
    },
    {
      "epoch": 2.5304138346382175,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004907796318459873,
      "loss": 1.254,
      "step": 29411
    },
    {
      "epoch": 2.530499870945539,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004907525957220942,
      "loss": 1.3933,
      "step": 29412
    },
    {
      "epoch": 2.5305859072528607,
      "grad_norm": 0.84375,
      "learning_rate": 0.000490725559625248,
      "loss": 1.245,
      "step": 29413
    },
    {
      "epoch": 2.5306719435601823,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004906985235555278,
      "loss": 1.3656,
      "step": 29414
    },
    {
      "epoch": 2.530757979867504,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004906714875130126,
      "loss": 0.8787,
      "step": 29415
    },
    {
      "epoch": 2.530844016174826,
      "grad_norm": 0.875,
      "learning_rate": 0.0004906444514977814,
      "loss": 0.8129,
      "step": 29416
    },
    {
      "epoch": 2.530930052482147,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004906174155099134,
      "loss": 1.1396,
      "step": 29417
    },
    {
      "epoch": 2.5310160887894693,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004905903795494876,
      "loss": 1.7074,
      "step": 29418
    },
    {
      "epoch": 2.531102125096791,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004905633436165831,
      "loss": 1.7569,
      "step": 29419
    },
    {
      "epoch": 2.5311881614041125,
      "grad_norm": 0.875,
      "learning_rate": 0.0004905363077112791,
      "loss": 0.9977,
      "step": 29420
    },
    {
      "epoch": 2.531274197711434,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004905092718336545,
      "loss": 1.3436,
      "step": 29421
    },
    {
      "epoch": 2.5313602340187558,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004904822359837883,
      "loss": 1.5461,
      "step": 29422
    },
    {
      "epoch": 2.531446270326078,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004904552001617598,
      "loss": 1.45,
      "step": 29423
    },
    {
      "epoch": 2.5315323066333995,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000490428164367648,
      "loss": 0.9649,
      "step": 29424
    },
    {
      "epoch": 2.531618342940721,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000490401128601532,
      "loss": 1.6552,
      "step": 29425
    },
    {
      "epoch": 2.5317043792480427,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004903740928634908,
      "loss": 1.4476,
      "step": 29426
    },
    {
      "epoch": 2.5317904155553643,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004903470571536036,
      "loss": 1.7014,
      "step": 29427
    },
    {
      "epoch": 2.531876451862686,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004903200214719491,
      "loss": 1.3132,
      "step": 29428
    },
    {
      "epoch": 2.5319624881700076,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004902929858186069,
      "loss": 1.1763,
      "step": 29429
    },
    {
      "epoch": 2.5320485244773296,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004902659501936558,
      "loss": 1.3006,
      "step": 29430
    },
    {
      "epoch": 2.5321345607846513,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004902389145971748,
      "loss": 0.8151,
      "step": 29431
    },
    {
      "epoch": 2.532220597091973,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004902118790292432,
      "loss": 0.9973,
      "step": 29432
    },
    {
      "epoch": 2.5323066333992945,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00049018484348994,
      "loss": 0.9203,
      "step": 29433
    },
    {
      "epoch": 2.532392669706616,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004901578079793441,
      "loss": 1.1113,
      "step": 29434
    },
    {
      "epoch": 2.5324787060139378,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004901307724975347,
      "loss": 1.4221,
      "step": 29435
    },
    {
      "epoch": 2.5325647423212594,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004901037370445909,
      "loss": 1.2478,
      "step": 29436
    },
    {
      "epoch": 2.5326507786285815,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004900767016205918,
      "loss": 1.3281,
      "step": 29437
    },
    {
      "epoch": 2.532736814935903,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004900496662256163,
      "loss": 1.8872,
      "step": 29438
    },
    {
      "epoch": 2.5328228512432247,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004900226308597437,
      "loss": 1.2976,
      "step": 29439
    },
    {
      "epoch": 2.5329088875505463,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004899955955230529,
      "loss": 1.5424,
      "step": 29440
    },
    {
      "epoch": 2.532994923857868,
      "grad_norm": 0.84375,
      "learning_rate": 0.000489968560215623,
      "loss": 1.1707,
      "step": 29441
    },
    {
      "epoch": 2.5330809601651896,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004899415249375331,
      "loss": 1.0437,
      "step": 29442
    },
    {
      "epoch": 2.533166996472511,
      "grad_norm": 1.65625,
      "learning_rate": 0.0004899144896888624,
      "loss": 1.0721,
      "step": 29443
    },
    {
      "epoch": 2.5332530327798333,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004898874544696899,
      "loss": 1.2035,
      "step": 29444
    },
    {
      "epoch": 2.533339069087155,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004898604192800945,
      "loss": 1.3736,
      "step": 29445
    },
    {
      "epoch": 2.5334251053944765,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004898333841201553,
      "loss": 0.8436,
      "step": 29446
    },
    {
      "epoch": 2.533511141701798,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004898063489899517,
      "loss": 0.4396,
      "step": 29447
    },
    {
      "epoch": 2.5335971780091198,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004897793138895625,
      "loss": 1.0168,
      "step": 29448
    },
    {
      "epoch": 2.5336832143164414,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004897522788190667,
      "loss": 1.2332,
      "step": 29449
    },
    {
      "epoch": 2.533769250623763,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004897252437785436,
      "loss": 0.8876,
      "step": 29450
    },
    {
      "epoch": 2.533855286931085,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004896982087680722,
      "loss": 1.4975,
      "step": 29451
    },
    {
      "epoch": 2.5339413232384067,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004896711737877315,
      "loss": 0.8647,
      "step": 29452
    },
    {
      "epoch": 2.5340273595457283,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004896441388376005,
      "loss": 1.1258,
      "step": 29453
    },
    {
      "epoch": 2.53411339585305,
      "grad_norm": 0.734375,
      "learning_rate": 0.0004896171039177584,
      "loss": 0.9067,
      "step": 29454
    },
    {
      "epoch": 2.5341994321603716,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004895900690282843,
      "loss": 1.0591,
      "step": 29455
    },
    {
      "epoch": 2.534285468467693,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004895630341692573,
      "loss": 1.8774,
      "step": 29456
    },
    {
      "epoch": 2.534371504775015,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004895359993407562,
      "loss": 0.8767,
      "step": 29457
    },
    {
      "epoch": 2.534457541082337,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004895089645428604,
      "loss": 1.3352,
      "step": 29458
    },
    {
      "epoch": 2.5345435773896585,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004894819297756487,
      "loss": 0.9521,
      "step": 29459
    },
    {
      "epoch": 2.53462961369698,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0004894548950392005,
      "loss": 0.6209,
      "step": 29460
    },
    {
      "epoch": 2.5347156500043018,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0004894278603335947,
      "loss": 1.5209,
      "step": 29461
    },
    {
      "epoch": 2.5348016863116234,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004894008256589101,
      "loss": 1.3781,
      "step": 29462
    },
    {
      "epoch": 2.5348877226189455,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004893737910152263,
      "loss": 1.7213,
      "step": 29463
    },
    {
      "epoch": 2.5349737589262666,
      "grad_norm": 1.484375,
      "learning_rate": 0.000489346756402622,
      "loss": 2.5937,
      "step": 29464
    },
    {
      "epoch": 2.5350597952335887,
      "grad_norm": 1.125,
      "learning_rate": 0.0004893197218211763,
      "loss": 1.5343,
      "step": 29465
    },
    {
      "epoch": 2.5351458315409103,
      "grad_norm": 1.390625,
      "learning_rate": 0.0004892926872709683,
      "loss": 0.7071,
      "step": 29466
    },
    {
      "epoch": 2.535231867848232,
      "grad_norm": 1.9375,
      "learning_rate": 0.0004892656527520773,
      "loss": 1.2625,
      "step": 29467
    },
    {
      "epoch": 2.5353179041555536,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004892386182645821,
      "loss": 1.2482,
      "step": 29468
    },
    {
      "epoch": 2.535403940462875,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004892115838085617,
      "loss": 1.5913,
      "step": 29469
    },
    {
      "epoch": 2.5354899767701973,
      "grad_norm": 16.75,
      "learning_rate": 0.0004891845493840955,
      "loss": 1.5141,
      "step": 29470
    },
    {
      "epoch": 2.5355760130775185,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004891575149912624,
      "loss": 1.1503,
      "step": 29471
    },
    {
      "epoch": 2.5356620493848405,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004891304806301413,
      "loss": 0.7596,
      "step": 29472
    },
    {
      "epoch": 2.535748085692162,
      "grad_norm": 1.375,
      "learning_rate": 0.0004891034463008116,
      "loss": 1.8418,
      "step": 29473
    },
    {
      "epoch": 2.5358341219994838,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004890764120033521,
      "loss": 1.2388,
      "step": 29474
    },
    {
      "epoch": 2.5359201583068054,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004890493777378421,
      "loss": 0.8352,
      "step": 29475
    },
    {
      "epoch": 2.536006194614127,
      "grad_norm": 2.296875,
      "learning_rate": 0.0004890223435043604,
      "loss": 1.5003,
      "step": 29476
    },
    {
      "epoch": 2.536092230921449,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004889953093029862,
      "loss": 1.5879,
      "step": 29477
    },
    {
      "epoch": 2.5361782672287703,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004889682751337987,
      "loss": 1.7737,
      "step": 29478
    },
    {
      "epoch": 2.5362643035360923,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004889412409968768,
      "loss": 1.1232,
      "step": 29479
    },
    {
      "epoch": 2.536350339843414,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004889142068922994,
      "loss": 1.233,
      "step": 29480
    },
    {
      "epoch": 2.5364363761507356,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000488887172820146,
      "loss": 1.8922,
      "step": 29481
    },
    {
      "epoch": 2.536522412458057,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004888601387804955,
      "loss": 0.9153,
      "step": 29482
    },
    {
      "epoch": 2.536608448765379,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004888331047734268,
      "loss": 1.7824,
      "step": 29483
    },
    {
      "epoch": 2.536694485072701,
      "grad_norm": 1.484375,
      "learning_rate": 0.0004888060707990193,
      "loss": 1.2574,
      "step": 29484
    },
    {
      "epoch": 2.5367805213800225,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004887790368573516,
      "loss": 0.9876,
      "step": 29485
    },
    {
      "epoch": 2.536866557687344,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004887520029485033,
      "loss": 0.8308,
      "step": 29486
    },
    {
      "epoch": 2.5369525939946658,
      "grad_norm": 1.015625,
      "learning_rate": 0.000488724969072553,
      "loss": 1.2775,
      "step": 29487
    },
    {
      "epoch": 2.5370386303019874,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00048869793522958,
      "loss": 1.1138,
      "step": 29488
    },
    {
      "epoch": 2.537124666609309,
      "grad_norm": 1.25,
      "learning_rate": 0.0004886709014196635,
      "loss": 0.8752,
      "step": 29489
    },
    {
      "epoch": 2.5372107029166306,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004886438676428821,
      "loss": 1.301,
      "step": 29490
    },
    {
      "epoch": 2.5372967392239527,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004886168338993155,
      "loss": 0.6773,
      "step": 29491
    },
    {
      "epoch": 2.5373827755312743,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004885898001890423,
      "loss": 0.8531,
      "step": 29492
    },
    {
      "epoch": 2.537468811838596,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004885627665121417,
      "loss": 1.2902,
      "step": 29493
    },
    {
      "epoch": 2.5375548481459176,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004885357328686928,
      "loss": 0.8369,
      "step": 29494
    },
    {
      "epoch": 2.537640884453239,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004885086992587745,
      "loss": 0.8315,
      "step": 29495
    },
    {
      "epoch": 2.537726920760561,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004884816656824662,
      "loss": 1.3694,
      "step": 29496
    },
    {
      "epoch": 2.5378129570678825,
      "grad_norm": 0.765625,
      "learning_rate": 0.0004884546321398468,
      "loss": 1.2735,
      "step": 29497
    },
    {
      "epoch": 2.5378989933752045,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004884275986309954,
      "loss": 1.6536,
      "step": 29498
    },
    {
      "epoch": 2.537985029682526,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004884005651559909,
      "loss": 0.9922,
      "step": 29499
    },
    {
      "epoch": 2.5380710659898478,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004883735317149125,
      "loss": 1.3539,
      "step": 29500
    },
    {
      "epoch": 2.5381571022971694,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004883464983078394,
      "loss": 0.8069,
      "step": 29501
    },
    {
      "epoch": 2.538243138604491,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004883194649348505,
      "loss": 1.2104,
      "step": 29502
    },
    {
      "epoch": 2.5383291749118126,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004882924315960247,
      "loss": 1.3179,
      "step": 29503
    },
    {
      "epoch": 2.5384152112191343,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004882653982914414,
      "loss": 0.9581,
      "step": 29504
    },
    {
      "epoch": 2.5385012475264563,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00048823836502117957,
      "loss": 1.0718,
      "step": 29505
    },
    {
      "epoch": 2.538587283833778,
      "grad_norm": 0.765625,
      "learning_rate": 0.0004882113317853181,
      "loss": 0.6402,
      "step": 29506
    },
    {
      "epoch": 2.5386733201410996,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00048818429858393624,
      "loss": 1.4649,
      "step": 29507
    },
    {
      "epoch": 2.538759356448421,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00048815726541711306,
      "loss": 1.203,
      "step": 29508
    },
    {
      "epoch": 2.538845392755743,
      "grad_norm": 1.3125,
      "learning_rate": 0.00048813023228492765,
      "loss": 1.3324,
      "step": 29509
    },
    {
      "epoch": 2.5389314290630645,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00048810319918745883,
      "loss": 1.4674,
      "step": 29510
    },
    {
      "epoch": 2.539017465370386,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0004880761661247859,
      "loss": 0.8864,
      "step": 29511
    },
    {
      "epoch": 2.539103501677708,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00048804913309698797,
      "loss": 1.07,
      "step": 29512
    },
    {
      "epoch": 2.5391895379850298,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004880221001041438,
      "loss": 1.2411,
      "step": 29513
    },
    {
      "epoch": 2.5392755742923514,
      "grad_norm": 2.375,
      "learning_rate": 0.0004879950671463328,
      "loss": 0.9096,
      "step": 29514
    },
    {
      "epoch": 2.539361610599673,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004879680342236339,
      "loss": 1.3849,
      "step": 29515
    },
    {
      "epoch": 2.5394476469069946,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004879410013361261,
      "loss": 2.1553,
      "step": 29516
    },
    {
      "epoch": 2.5395336832143167,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004879139684838885,
      "loss": 1.8075,
      "step": 29517
    },
    {
      "epoch": 2.539619719521638,
      "grad_norm": 1.28125,
      "learning_rate": 0.00048788693566700024,
      "loss": 2.064,
      "step": 29518
    },
    {
      "epoch": 2.53970575582896,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00048785990288554036,
      "loss": 1.2787,
      "step": 29519
    },
    {
      "epoch": 2.5397917921362816,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004878328701395878,
      "loss": 1.2621,
      "step": 29520
    },
    {
      "epoch": 2.539877828443603,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00048780583742922184,
      "loss": 1.3992,
      "step": 29521
    },
    {
      "epoch": 2.539963864750925,
      "grad_norm": 1.140625,
      "learning_rate": 0.00048777880475452145,
      "loss": 0.9666,
      "step": 29522
    },
    {
      "epoch": 2.5400499010582465,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00048775177211556563,
      "loss": 0.7713,
      "step": 29523
    },
    {
      "epoch": 2.5401359373655685,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004877247395124335,
      "loss": 0.9506,
      "step": 29524
    },
    {
      "epoch": 2.5402219736728897,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004876977069452041,
      "loss": 1.3475,
      "step": 29525
    },
    {
      "epoch": 2.540308009980212,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004876706744139566,
      "loss": 0.8452,
      "step": 29526
    },
    {
      "epoch": 2.5403940462875334,
      "grad_norm": 1.46875,
      "learning_rate": 0.00048764364191876994,
      "loss": 2.2596,
      "step": 29527
    },
    {
      "epoch": 2.540480082594855,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0004876166094597231,
      "loss": 1.7079,
      "step": 29528
    },
    {
      "epoch": 2.5405661189021767,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004875895770368954,
      "loss": 1.202,
      "step": 29529
    },
    {
      "epoch": 2.5406521552094983,
      "grad_norm": 1.09375,
      "learning_rate": 0.00048756254465036575,
      "loss": 1.3697,
      "step": 29530
    },
    {
      "epoch": 2.5407381915168203,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00048753551230021314,
      "loss": 2.3238,
      "step": 29531
    },
    {
      "epoch": 2.5408242278241415,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00048750847998651673,
      "loss": 1.3486,
      "step": 29532
    },
    {
      "epoch": 2.5409102641314636,
      "grad_norm": 1.25,
      "learning_rate": 0.0004874814477093556,
      "loss": 1.3378,
      "step": 29533
    },
    {
      "epoch": 2.540996300438785,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004874544154688089,
      "loss": 1.6494,
      "step": 29534
    },
    {
      "epoch": 2.541082336746107,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004874273832649556,
      "loss": 2.0476,
      "step": 29535
    },
    {
      "epoch": 2.5411683730534285,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004874003510978746,
      "loss": 1.3645,
      "step": 29536
    },
    {
      "epoch": 2.54125440936075,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004873733189676453,
      "loss": 1.4343,
      "step": 29537
    },
    {
      "epoch": 2.541340445668072,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004873462868743466,
      "loss": 0.8591,
      "step": 29538
    },
    {
      "epoch": 2.541426481975394,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004873192548180575,
      "loss": 0.915,
      "step": 29539
    },
    {
      "epoch": 2.5415125182827154,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000487292222798857,
      "loss": 1.1288,
      "step": 29540
    },
    {
      "epoch": 2.541598554590037,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00048726519081682444,
      "loss": 1.668,
      "step": 29541
    },
    {
      "epoch": 2.5416845908973587,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00048723815887203865,
      "loss": 1.7807,
      "step": 29542
    },
    {
      "epoch": 2.5417706272046803,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004872111269645788,
      "loss": 1.6265,
      "step": 29543
    },
    {
      "epoch": 2.541856663512002,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0004871840950945238,
      "loss": 1.3929,
      "step": 29544
    },
    {
      "epoch": 2.541942699819324,
      "grad_norm": 1.03125,
      "learning_rate": 0.00048715706326195297,
      "loss": 0.8974,
      "step": 29545
    },
    {
      "epoch": 2.5420287361266456,
      "grad_norm": 1.34375,
      "learning_rate": 0.00048713003146694516,
      "loss": 1.6983,
      "step": 29546
    },
    {
      "epoch": 2.542114772433967,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004871029997095795,
      "loss": 1.1683,
      "step": 29547
    },
    {
      "epoch": 2.542200808741289,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00048707596798993507,
      "loss": 1.1828,
      "step": 29548
    },
    {
      "epoch": 2.5422868450486105,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00048704893630809097,
      "loss": 1.4381,
      "step": 29549
    },
    {
      "epoch": 2.542372881355932,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004870219046641262,
      "loss": 1.5667,
      "step": 29550
    },
    {
      "epoch": 2.5424589176632537,
      "grad_norm": 1.125,
      "learning_rate": 0.00048699487305811976,
      "loss": 1.451,
      "step": 29551
    },
    {
      "epoch": 2.542544953970576,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004869678414901508,
      "loss": 0.8998,
      "step": 29552
    },
    {
      "epoch": 2.5426309902778974,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004869408099602985,
      "loss": 0.8898,
      "step": 29553
    },
    {
      "epoch": 2.542717026585219,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004869137784686416,
      "loss": 1.3891,
      "step": 29554
    },
    {
      "epoch": 2.5428030628925407,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004868867470152595,
      "loss": 0.6627,
      "step": 29555
    },
    {
      "epoch": 2.5428890991998623,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00048685971560023106,
      "loss": 1.4059,
      "step": 29556
    },
    {
      "epoch": 2.542975135507184,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004868326842236354,
      "loss": 1.3624,
      "step": 29557
    },
    {
      "epoch": 2.5430611718145055,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004868056528855515,
      "loss": 1.1982,
      "step": 29558
    },
    {
      "epoch": 2.5431472081218276,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0004867786215860586,
      "loss": 0.8292,
      "step": 29559
    },
    {
      "epoch": 2.543233244429149,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004867515903252357,
      "loss": 1.4463,
      "step": 29560
    },
    {
      "epoch": 2.543319280736471,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004867245591031617,
      "loss": 1.2566,
      "step": 29561
    },
    {
      "epoch": 2.5434053170437925,
      "grad_norm": 1.078125,
      "learning_rate": 0.00048669752791991586,
      "loss": 0.7556,
      "step": 29562
    },
    {
      "epoch": 2.543491353351114,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00048667049677557714,
      "loss": 1.0048,
      "step": 29563
    },
    {
      "epoch": 2.5435773896584357,
      "grad_norm": 1.046875,
      "learning_rate": 0.00048664346567022464,
      "loss": 1.5989,
      "step": 29564
    },
    {
      "epoch": 2.5436634259657573,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00048661643460393736,
      "loss": 1.6902,
      "step": 29565
    },
    {
      "epoch": 2.5437494622730794,
      "grad_norm": 1.109375,
      "learning_rate": 0.00048658940357679443,
      "loss": 0.601,
      "step": 29566
    },
    {
      "epoch": 2.543835498580401,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004865623725888749,
      "loss": 1.0575,
      "step": 29567
    },
    {
      "epoch": 2.5439215348877227,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004865353416402579,
      "loss": 1.5507,
      "step": 29568
    },
    {
      "epoch": 2.5440075711950443,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004865083107310221,
      "loss": 1.5717,
      "step": 29569
    },
    {
      "epoch": 2.544093607502366,
      "grad_norm": 1.640625,
      "learning_rate": 0.000486481279861247,
      "loss": 1.2392,
      "step": 29570
    },
    {
      "epoch": 2.5441796438096875,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004864542490310117,
      "loss": 1.0108,
      "step": 29571
    },
    {
      "epoch": 2.544265680117009,
      "grad_norm": 1.125,
      "learning_rate": 0.000486427218240395,
      "loss": 1.6186,
      "step": 29572
    },
    {
      "epoch": 2.5443517164243312,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00048640018748947604,
      "loss": 1.7413,
      "step": 29573
    },
    {
      "epoch": 2.544437752731653,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004863731567783338,
      "loss": 0.6746,
      "step": 29574
    },
    {
      "epoch": 2.5445237890389745,
      "grad_norm": 1.015625,
      "learning_rate": 0.00048634612610704757,
      "loss": 0.8486,
      "step": 29575
    },
    {
      "epoch": 2.544609825346296,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00048631909547569623,
      "loss": 1.2437,
      "step": 29576
    },
    {
      "epoch": 2.5446958616536177,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004862920648843588,
      "loss": 1.0347,
      "step": 29577
    },
    {
      "epoch": 2.54478189796094,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00048626503433311447,
      "loss": 0.9464,
      "step": 29578
    },
    {
      "epoch": 2.544867934268261,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004862380038220423,
      "loss": 1.0584,
      "step": 29579
    },
    {
      "epoch": 2.544953970575583,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004862109733512213,
      "loss": 1.3608,
      "step": 29580
    },
    {
      "epoch": 2.5450400068829047,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004861839429207304,
      "loss": 1.666,
      "step": 29581
    },
    {
      "epoch": 2.5451260431902263,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004861569125306488,
      "loss": 1.8112,
      "step": 29582
    },
    {
      "epoch": 2.545212079497548,
      "grad_norm": 0.984375,
      "learning_rate": 0.00048612988218105566,
      "loss": 1.8324,
      "step": 29583
    },
    {
      "epoch": 2.5452981158048695,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00048610285187202976,
      "loss": 1.1316,
      "step": 29584
    },
    {
      "epoch": 2.5453841521121916,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004860758216036504,
      "loss": 1.4993,
      "step": 29585
    },
    {
      "epoch": 2.545470188419513,
      "grad_norm": 1.25,
      "learning_rate": 0.00048604879137599657,
      "loss": 1.1315,
      "step": 29586
    },
    {
      "epoch": 2.545556224726835,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004860217611891473,
      "loss": 1.6229,
      "step": 29587
    },
    {
      "epoch": 2.5456422610341565,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00048599473104318157,
      "loss": 1.9948,
      "step": 29588
    },
    {
      "epoch": 2.545728297341478,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00048596770093817867,
      "loss": 1.0394,
      "step": 29589
    },
    {
      "epoch": 2.5458143336487997,
      "grad_norm": 1.53125,
      "learning_rate": 0.0004859406708742174,
      "loss": 1.9674,
      "step": 29590
    },
    {
      "epoch": 2.5459003699561213,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00048591364085137705,
      "loss": 1.1426,
      "step": 29591
    },
    {
      "epoch": 2.5459864062634434,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004858866108697364,
      "loss": 1.9648,
      "step": 29592
    },
    {
      "epoch": 2.5460724425707646,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0004858595809293748,
      "loss": 0.9894,
      "step": 29593
    },
    {
      "epoch": 2.5461584788780867,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004858325510303712,
      "loss": 0.7133,
      "step": 29594
    },
    {
      "epoch": 2.5462445151854083,
      "grad_norm": 0.9375,
      "learning_rate": 0.00048580552117280447,
      "loss": 0.7551,
      "step": 29595
    },
    {
      "epoch": 2.54633055149273,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000485778491356754,
      "loss": 1.4945,
      "step": 29596
    },
    {
      "epoch": 2.5464165878000515,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004857514615822986,
      "loss": 1.1518,
      "step": 29597
    },
    {
      "epoch": 2.546502624107373,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00048572443184951745,
      "loss": 1.6456,
      "step": 29598
    },
    {
      "epoch": 2.5465886604146952,
      "grad_norm": 1.140625,
      "learning_rate": 0.00048569740215848945,
      "loss": 1.2258,
      "step": 29599
    },
    {
      "epoch": 2.546674696722017,
      "grad_norm": 0.953125,
      "learning_rate": 0.00048567037250929387,
      "loss": 0.8869,
      "step": 29600
    },
    {
      "epoch": 2.5467607330293385,
      "grad_norm": 1.609375,
      "learning_rate": 0.00048564334290200964,
      "loss": 1.3692,
      "step": 29601
    },
    {
      "epoch": 2.54684676933666,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00048561631333671575,
      "loss": 0.7746,
      "step": 29602
    },
    {
      "epoch": 2.5469328056439817,
      "grad_norm": 0.875,
      "learning_rate": 0.0004855892838134915,
      "loss": 1.3795,
      "step": 29603
    },
    {
      "epoch": 2.5470188419513033,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00048556225433241573,
      "loss": 1.6051,
      "step": 29604
    },
    {
      "epoch": 2.547104878258625,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0004855352248935676,
      "loss": 1.6557,
      "step": 29605
    },
    {
      "epoch": 2.547190914565947,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00048550819549702585,
      "loss": 0.7669,
      "step": 29606
    },
    {
      "epoch": 2.5472769508732687,
      "grad_norm": 1.25,
      "learning_rate": 0.00048548116614287,
      "loss": 1.3661,
      "step": 29607
    },
    {
      "epoch": 2.5473629871805903,
      "grad_norm": 1.71875,
      "learning_rate": 0.0004854541368311789,
      "loss": 1.1787,
      "step": 29608
    },
    {
      "epoch": 2.547449023487912,
      "grad_norm": 1.125,
      "learning_rate": 0.00048542710756203174,
      "loss": 1.3034,
      "step": 29609
    },
    {
      "epoch": 2.5475350597952335,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004854000783355074,
      "loss": 1.8497,
      "step": 29610
    },
    {
      "epoch": 2.547621096102555,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0004853730491516849,
      "loss": 0.8404,
      "step": 29611
    },
    {
      "epoch": 2.547707132409877,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004853460200106435,
      "loss": 1.5739,
      "step": 29612
    },
    {
      "epoch": 2.547793168717199,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004853189909124621,
      "loss": 1.0874,
      "step": 29613
    },
    {
      "epoch": 2.5478792050245205,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00048529196185721985,
      "loss": 1.3616,
      "step": 29614
    },
    {
      "epoch": 2.547965241331842,
      "grad_norm": 1.1875,
      "learning_rate": 0.00048526493284499565,
      "loss": 1.7566,
      "step": 29615
    },
    {
      "epoch": 2.5480512776391637,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004852379038758687,
      "loss": 1.4014,
      "step": 29616
    },
    {
      "epoch": 2.5481373139464853,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00048521087494991804,
      "loss": 1.0651,
      "step": 29617
    },
    {
      "epoch": 2.548223350253807,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004851838460672226,
      "loss": 1.7735,
      "step": 29618
    },
    {
      "epoch": 2.5483093865611286,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00048515681722786165,
      "loss": 0.859,
      "step": 29619
    },
    {
      "epoch": 2.5483954228684507,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004851297884319141,
      "loss": 1.2808,
      "step": 29620
    },
    {
      "epoch": 2.5484814591757723,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000485102759679459,
      "loss": 1.0416,
      "step": 29621
    },
    {
      "epoch": 2.548567495483094,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004850757309705753,
      "loss": 0.6081,
      "step": 29622
    },
    {
      "epoch": 2.5486535317904155,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004850487023053424,
      "loss": 1.5627,
      "step": 29623
    },
    {
      "epoch": 2.548739568097737,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004850216736838391,
      "loss": 0.8609,
      "step": 29624
    },
    {
      "epoch": 2.548825604405059,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004849946451061443,
      "loss": 0.9106,
      "step": 29625
    },
    {
      "epoch": 2.5489116407123804,
      "grad_norm": 1.203125,
      "learning_rate": 0.00048496761657233745,
      "loss": 1.264,
      "step": 29626
    },
    {
      "epoch": 2.5489976770197025,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00048494058808249736,
      "loss": 1.2816,
      "step": 29627
    },
    {
      "epoch": 2.549083713327024,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00048491355963670315,
      "loss": 1.2478,
      "step": 29628
    },
    {
      "epoch": 2.5491697496343457,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004848865312350337,
      "loss": 1.8271,
      "step": 29629
    },
    {
      "epoch": 2.5492557859416674,
      "grad_norm": 1.5,
      "learning_rate": 0.0004848595028775683,
      "loss": 1.4407,
      "step": 29630
    },
    {
      "epoch": 2.549341822248989,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004848324745643859,
      "loss": 0.8788,
      "step": 29631
    },
    {
      "epoch": 2.5494278585563106,
      "grad_norm": 1.21875,
      "learning_rate": 0.00048480544629556565,
      "loss": 1.2989,
      "step": 29632
    },
    {
      "epoch": 2.5495138948636322,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004847784180711864,
      "loss": 0.6512,
      "step": 29633
    },
    {
      "epoch": 2.5495999311709543,
      "grad_norm": 1.15625,
      "learning_rate": 0.00048475138989132735,
      "loss": 1.1364,
      "step": 29634
    },
    {
      "epoch": 2.549685967478276,
      "grad_norm": 1.515625,
      "learning_rate": 0.00048472436175606756,
      "loss": 1.8821,
      "step": 29635
    },
    {
      "epoch": 2.5497720037855975,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004846973336654859,
      "loss": 0.9452,
      "step": 29636
    },
    {
      "epoch": 2.549858040092919,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00048467030561966166,
      "loss": 0.9992,
      "step": 29637
    },
    {
      "epoch": 2.549944076400241,
      "grad_norm": 1.09375,
      "learning_rate": 0.00048464327761867386,
      "loss": 0.9635,
      "step": 29638
    },
    {
      "epoch": 2.550030112707563,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00048461624966260146,
      "loss": 1.0138,
      "step": 29639
    },
    {
      "epoch": 2.550116149014884,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004845892217515234,
      "loss": 1.0078,
      "step": 29640
    },
    {
      "epoch": 2.550202185322206,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00048456219388551904,
      "loss": 0.4022,
      "step": 29641
    },
    {
      "epoch": 2.5502882216295277,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004845351660646672,
      "loss": 1.1359,
      "step": 29642
    },
    {
      "epoch": 2.5503742579368494,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00048450813828904675,
      "loss": 1.5844,
      "step": 29643
    },
    {
      "epoch": 2.550460294244171,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004844811105587373,
      "loss": 1.3153,
      "step": 29644
    },
    {
      "epoch": 2.5505463305514926,
      "grad_norm": 1.171875,
      "learning_rate": 0.00048445408287381745,
      "loss": 1.0758,
      "step": 29645
    },
    {
      "epoch": 2.5506323668588147,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004844270552343664,
      "loss": 1.02,
      "step": 29646
    },
    {
      "epoch": 2.550718403166136,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004844000276404633,
      "loss": 0.7373,
      "step": 29647
    },
    {
      "epoch": 2.550804439473458,
      "grad_norm": 1.234375,
      "learning_rate": 0.00048437300009218686,
      "loss": 1.5917,
      "step": 29648
    },
    {
      "epoch": 2.5508904757807795,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00048434597258961655,
      "loss": 0.8773,
      "step": 29649
    },
    {
      "epoch": 2.550976512088101,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004843189451328312,
      "loss": 1.573,
      "step": 29650
    },
    {
      "epoch": 2.551062548395423,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004842919177219099,
      "loss": 1.2253,
      "step": 29651
    },
    {
      "epoch": 2.5511485847027444,
      "grad_norm": 0.9375,
      "learning_rate": 0.00048426489035693156,
      "loss": 1.1057,
      "step": 29652
    },
    {
      "epoch": 2.5512346210100665,
      "grad_norm": 1.25,
      "learning_rate": 0.00048423786303797544,
      "loss": 1.2208,
      "step": 29653
    },
    {
      "epoch": 2.5513206573173877,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004842108357651205,
      "loss": 1.5018,
      "step": 29654
    },
    {
      "epoch": 2.5514066936247097,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004841838085384459,
      "loss": 1.766,
      "step": 29655
    },
    {
      "epoch": 2.5514927299320314,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004841567813580304,
      "loss": 1.8591,
      "step": 29656
    },
    {
      "epoch": 2.551578766239353,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00048412975422395334,
      "loss": 2.0257,
      "step": 29657
    },
    {
      "epoch": 2.5516648025466746,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00048410272713629365,
      "loss": 1.1056,
      "step": 29658
    },
    {
      "epoch": 2.5517508388539962,
      "grad_norm": 0.765625,
      "learning_rate": 0.0004840757000951303,
      "loss": 1.1713,
      "step": 29659
    },
    {
      "epoch": 2.5518368751613183,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004840486731005426,
      "loss": 0.9575,
      "step": 29660
    },
    {
      "epoch": 2.55192291146864,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004840216461526093,
      "loss": 0.8473,
      "step": 29661
    },
    {
      "epoch": 2.5520089477759615,
      "grad_norm": 1.171875,
      "learning_rate": 0.00048399461925140966,
      "loss": 1.8922,
      "step": 29662
    },
    {
      "epoch": 2.552094984083283,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004839675923970225,
      "loss": 0.5866,
      "step": 29663
    },
    {
      "epoch": 2.552181020390605,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004839405655895271,
      "loss": 1.307,
      "step": 29664
    },
    {
      "epoch": 2.5522670566979264,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004839135388290025,
      "loss": 1.5051,
      "step": 29665
    },
    {
      "epoch": 2.552353093005248,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004838865121155276,
      "loss": 1.5663,
      "step": 29666
    },
    {
      "epoch": 2.55243912931257,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004838594854491815,
      "loss": 1.199,
      "step": 29667
    },
    {
      "epoch": 2.5525251656198917,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00048383245883004336,
      "loss": 0.8853,
      "step": 29668
    },
    {
      "epoch": 2.5526112019272134,
      "grad_norm": 1.25,
      "learning_rate": 0.00048380543225819204,
      "loss": 0.81,
      "step": 29669
    },
    {
      "epoch": 2.552697238234535,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004837784057337067,
      "loss": 1.4399,
      "step": 29670
    },
    {
      "epoch": 2.5527832745418566,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00048375137925666637,
      "loss": 0.5304,
      "step": 29671
    },
    {
      "epoch": 2.5528693108491782,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00048372435282715014,
      "loss": 1.302,
      "step": 29672
    },
    {
      "epoch": 2.5529553471565,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004836973264452369,
      "loss": 1.9076,
      "step": 29673
    },
    {
      "epoch": 2.553041383463822,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004836703001110059,
      "loss": 2.1879,
      "step": 29674
    },
    {
      "epoch": 2.5531274197711435,
      "grad_norm": 1.125,
      "learning_rate": 0.0004836432738245361,
      "loss": 1.1054,
      "step": 29675
    },
    {
      "epoch": 2.553213456078465,
      "grad_norm": 1.140625,
      "learning_rate": 0.00048361624758590656,
      "loss": 1.0463,
      "step": 29676
    },
    {
      "epoch": 2.553299492385787,
      "grad_norm": 1.375,
      "learning_rate": 0.00048358922139519616,
      "loss": 0.6431,
      "step": 29677
    },
    {
      "epoch": 2.5533855286931084,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004835621952524843,
      "loss": 1.2321,
      "step": 29678
    },
    {
      "epoch": 2.55347156500043,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004835351691578497,
      "loss": 1.0838,
      "step": 29679
    },
    {
      "epoch": 2.5535576013077517,
      "grad_norm": 1.0,
      "learning_rate": 0.0004835081431113714,
      "loss": 1.0974,
      "step": 29680
    },
    {
      "epoch": 2.5536436376150737,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004834811171131288,
      "loss": 0.7825,
      "step": 29681
    },
    {
      "epoch": 2.5537296739223954,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00048345409116320055,
      "loss": 1.3839,
      "step": 29682
    },
    {
      "epoch": 2.553815710229717,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00048342706526166595,
      "loss": 0.9585,
      "step": 29683
    },
    {
      "epoch": 2.5539017465370386,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004834000394086041,
      "loss": 0.9971,
      "step": 29684
    },
    {
      "epoch": 2.5539877828443602,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00048337301360409376,
      "loss": 1.0791,
      "step": 29685
    },
    {
      "epoch": 2.554073819151682,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000483345987848214,
      "loss": 0.6236,
      "step": 29686
    },
    {
      "epoch": 2.5541598554590035,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004833189621410442,
      "loss": 0.8499,
      "step": 29687
    },
    {
      "epoch": 2.5542458917663255,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004832919364826632,
      "loss": 1.4221,
      "step": 29688
    },
    {
      "epoch": 2.554331928073647,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0004832649108731498,
      "loss": 0.4035,
      "step": 29689
    },
    {
      "epoch": 2.554417964380969,
      "grad_norm": 1.71875,
      "learning_rate": 0.0004832378853125835,
      "loss": 1.3061,
      "step": 29690
    },
    {
      "epoch": 2.5545040006882904,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004832108598010431,
      "loss": 1.0354,
      "step": 29691
    },
    {
      "epoch": 2.554590036995612,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004831838343386077,
      "loss": 1.0685,
      "step": 29692
    },
    {
      "epoch": 2.554676073302934,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004831568089253561,
      "loss": 1.5646,
      "step": 29693
    },
    {
      "epoch": 2.5547621096102553,
      "grad_norm": 1.3125,
      "learning_rate": 0.00048312978356136776,
      "loss": 0.9125,
      "step": 29694
    },
    {
      "epoch": 2.5548481459175774,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004831027582467215,
      "loss": 0.9571,
      "step": 29695
    },
    {
      "epoch": 2.554934182224899,
      "grad_norm": 1.7578125,
      "learning_rate": 0.00048307573298149626,
      "loss": 1.8491,
      "step": 29696
    },
    {
      "epoch": 2.5550202185322206,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00048304870776577137,
      "loss": 1.4793,
      "step": 29697
    },
    {
      "epoch": 2.5551062548395422,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004830216825996256,
      "loss": 2.0629,
      "step": 29698
    },
    {
      "epoch": 2.555192291146864,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00048299465748313824,
      "loss": 1.0239,
      "step": 29699
    },
    {
      "epoch": 2.555278327454186,
      "grad_norm": 0.765625,
      "learning_rate": 0.00048296763241638803,
      "loss": 0.5016,
      "step": 29700
    },
    {
      "epoch": 2.555364363761507,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004829406073994543,
      "loss": 1.2888,
      "step": 29701
    },
    {
      "epoch": 2.555450400068829,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00048291358243241595,
      "loss": 1.194,
      "step": 29702
    },
    {
      "epoch": 2.555536436376151,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004828865575153521,
      "loss": 1.0412,
      "step": 29703
    },
    {
      "epoch": 2.5556224726834724,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004828595326483416,
      "loss": 1.201,
      "step": 29704
    },
    {
      "epoch": 2.555708508990794,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00048283250783146377,
      "loss": 0.9924,
      "step": 29705
    },
    {
      "epoch": 2.5557945452981157,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004828054830647974,
      "loss": 1.4867,
      "step": 29706
    },
    {
      "epoch": 2.5558805816054377,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004827784583484217,
      "loss": 0.9237,
      "step": 29707
    },
    {
      "epoch": 2.555966617912759,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004827514336824157,
      "loss": 1.7157,
      "step": 29708
    },
    {
      "epoch": 2.556052654220081,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00048272440906685837,
      "loss": 0.7058,
      "step": 29709
    },
    {
      "epoch": 2.5561386905274026,
      "grad_norm": 1.09375,
      "learning_rate": 0.00048269738450182884,
      "loss": 1.8017,
      "step": 29710
    },
    {
      "epoch": 2.5562247268347242,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004826703599874059,
      "loss": 1.7862,
      "step": 29711
    },
    {
      "epoch": 2.556310763142046,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000482643335523669,
      "loss": 1.357,
      "step": 29712
    },
    {
      "epoch": 2.5563967994493675,
      "grad_norm": 1.328125,
      "learning_rate": 0.00048261631111069694,
      "loss": 1.1359,
      "step": 29713
    },
    {
      "epoch": 2.5564828357566896,
      "grad_norm": 1.71875,
      "learning_rate": 0.0004825892867485686,
      "loss": 1.7456,
      "step": 29714
    },
    {
      "epoch": 2.556568872064011,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004825622624373634,
      "loss": 1.2908,
      "step": 29715
    },
    {
      "epoch": 2.556654908371333,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00048253523817716015,
      "loss": 0.9508,
      "step": 29716
    },
    {
      "epoch": 2.5567409446786544,
      "grad_norm": 0.96875,
      "learning_rate": 0.00048250821396803777,
      "loss": 1.0093,
      "step": 29717
    },
    {
      "epoch": 2.556826980985976,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00048248118981007563,
      "loss": 1.419,
      "step": 29718
    },
    {
      "epoch": 2.5569130172932977,
      "grad_norm": 0.6875,
      "learning_rate": 0.0004824541657033527,
      "loss": 0.2705,
      "step": 29719
    },
    {
      "epoch": 2.5569990536006193,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00048242714164794767,
      "loss": 1.513,
      "step": 29720
    },
    {
      "epoch": 2.5570850899079414,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00048240011764393996,
      "loss": 1.5028,
      "step": 29721
    },
    {
      "epoch": 2.557171126215263,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004823730936914086,
      "loss": 1.6718,
      "step": 29722
    },
    {
      "epoch": 2.5572571625225846,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00048234606979043226,
      "loss": 0.5861,
      "step": 29723
    },
    {
      "epoch": 2.5573431988299062,
      "grad_norm": 1.078125,
      "learning_rate": 0.00048231904594109044,
      "loss": 1.4567,
      "step": 29724
    },
    {
      "epoch": 2.557429235137228,
      "grad_norm": 1.34375,
      "learning_rate": 0.000482292022143462,
      "loss": 1.2081,
      "step": 29725
    },
    {
      "epoch": 2.5575152714445495,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00048226499839762585,
      "loss": 0.7039,
      "step": 29726
    },
    {
      "epoch": 2.557601307751871,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00048223797470366104,
      "loss": 0.7924,
      "step": 29727
    },
    {
      "epoch": 2.557687344059193,
      "grad_norm": 1.359375,
      "learning_rate": 0.00048221095106164687,
      "loss": 1.2037,
      "step": 29728
    },
    {
      "epoch": 2.557773380366515,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004821839274716622,
      "loss": 1.9843,
      "step": 29729
    },
    {
      "epoch": 2.5578594166738364,
      "grad_norm": 1.25,
      "learning_rate": 0.0004821569039337859,
      "loss": 1.81,
      "step": 29730
    },
    {
      "epoch": 2.557945452981158,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004821298804480974,
      "loss": 0.8194,
      "step": 29731
    },
    {
      "epoch": 2.5580314892884797,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00048210285701467536,
      "loss": 1.5503,
      "step": 29732
    },
    {
      "epoch": 2.5581175255958013,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00048207583363359916,
      "loss": 0.9881,
      "step": 29733
    },
    {
      "epoch": 2.558203561903123,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004820488103049474,
      "loss": 1.4231,
      "step": 29734
    },
    {
      "epoch": 2.558289598210445,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004820217870287996,
      "loss": 0.6877,
      "step": 29735
    },
    {
      "epoch": 2.5583756345177666,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004819947638052345,
      "loss": 0.8637,
      "step": 29736
    },
    {
      "epoch": 2.5584616708250882,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004819677406343312,
      "loss": 1.1245,
      "step": 29737
    },
    {
      "epoch": 2.55854770713241,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004819407175161688,
      "loss": 1.4191,
      "step": 29738
    },
    {
      "epoch": 2.5586337434397315,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00048191369445082626,
      "loss": 1.1538,
      "step": 29739
    },
    {
      "epoch": 2.558719779747053,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004818866714383827,
      "loss": 1.5186,
      "step": 29740
    },
    {
      "epoch": 2.5588058160543747,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00048185964847891695,
      "loss": 0.6828,
      "step": 29741
    },
    {
      "epoch": 2.558891852361697,
      "grad_norm": 1.015625,
      "learning_rate": 0.00048183262557250836,
      "loss": 1.0561,
      "step": 29742
    },
    {
      "epoch": 2.5589778886690184,
      "grad_norm": 1.046875,
      "learning_rate": 0.00048180560271923577,
      "loss": 1.0815,
      "step": 29743
    },
    {
      "epoch": 2.55906392497634,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004817785799191783,
      "loss": 1.4726,
      "step": 29744
    },
    {
      "epoch": 2.5591499612836617,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00048175155717241475,
      "loss": 1.1736,
      "step": 29745
    },
    {
      "epoch": 2.5592359975909833,
      "grad_norm": 1.28125,
      "learning_rate": 0.00048172453447902446,
      "loss": 2.1553,
      "step": 29746
    },
    {
      "epoch": 2.559322033898305,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0004816975118390864,
      "loss": 1.1857,
      "step": 29747
    },
    {
      "epoch": 2.5594080702056266,
      "grad_norm": 1.109375,
      "learning_rate": 0.00048167048925267945,
      "loss": 0.8961,
      "step": 29748
    },
    {
      "epoch": 2.5594941065129486,
      "grad_norm": 1.078125,
      "learning_rate": 0.00048164346671988287,
      "loss": 0.6609,
      "step": 29749
    },
    {
      "epoch": 2.5595801428202702,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00048161644424077555,
      "loss": 1.0956,
      "step": 29750
    },
    {
      "epoch": 2.559666179127592,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00048158942181543657,
      "loss": 1.8195,
      "step": 29751
    },
    {
      "epoch": 2.5597522154349135,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004815623994439449,
      "loss": 2.0213,
      "step": 29752
    },
    {
      "epoch": 2.559838251742235,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004815353771263796,
      "loss": 1.2519,
      "step": 29753
    },
    {
      "epoch": 2.559924288049557,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00048150835486281964,
      "loss": 0.9086,
      "step": 29754
    },
    {
      "epoch": 2.5600103243568784,
      "grad_norm": 1.078125,
      "learning_rate": 0.00048148133265334424,
      "loss": 1.15,
      "step": 29755
    },
    {
      "epoch": 2.5600963606642004,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004814543104980325,
      "loss": 1.5616,
      "step": 29756
    },
    {
      "epoch": 2.560182396971522,
      "grad_norm": 1.609375,
      "learning_rate": 0.00048142728839696303,
      "loss": 0.9824,
      "step": 29757
    },
    {
      "epoch": 2.5602684332788437,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004814002663502153,
      "loss": 1.3483,
      "step": 29758
    },
    {
      "epoch": 2.5603544695861653,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004813732443578682,
      "loss": 1.4214,
      "step": 29759
    },
    {
      "epoch": 2.560440505893487,
      "grad_norm": 1.609375,
      "learning_rate": 0.0004813462224200006,
      "loss": 1.0911,
      "step": 29760
    },
    {
      "epoch": 2.560526542200809,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00048131920053669176,
      "loss": 1.2005,
      "step": 29761
    },
    {
      "epoch": 2.56061257850813,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004812921787080207,
      "loss": 1.6099,
      "step": 29762
    },
    {
      "epoch": 2.5606986148154522,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004812651569340663,
      "loss": 1.0132,
      "step": 29763
    },
    {
      "epoch": 2.560784651122774,
      "grad_norm": 1.25,
      "learning_rate": 0.0004812381352149076,
      "loss": 1.5358,
      "step": 29764
    },
    {
      "epoch": 2.5608706874300955,
      "grad_norm": 1.25,
      "learning_rate": 0.0004812111135506238,
      "loss": 1.5561,
      "step": 29765
    },
    {
      "epoch": 2.560956723737417,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004811840919412939,
      "loss": 0.8008,
      "step": 29766
    },
    {
      "epoch": 2.5610427600447387,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004811570703869968,
      "loss": 0.9334,
      "step": 29767
    },
    {
      "epoch": 2.561128796352061,
      "grad_norm": 0.90625,
      "learning_rate": 0.00048113004888781147,
      "loss": 1.2842,
      "step": 29768
    },
    {
      "epoch": 2.561214832659382,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004811030274438172,
      "loss": 1.3586,
      "step": 29769
    },
    {
      "epoch": 2.561300868966704,
      "grad_norm": 0.828125,
      "learning_rate": 0.000481076006055093,
      "loss": 0.5345,
      "step": 29770
    },
    {
      "epoch": 2.5613869052740257,
      "grad_norm": 1.921875,
      "learning_rate": 0.0004810489847217176,
      "loss": 1.2502,
      "step": 29771
    },
    {
      "epoch": 2.5614729415813473,
      "grad_norm": 1.140625,
      "learning_rate": 0.00048102196344377033,
      "loss": 1.0187,
      "step": 29772
    },
    {
      "epoch": 2.561558977888669,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004809949422213302,
      "loss": 1.3966,
      "step": 29773
    },
    {
      "epoch": 2.5616450141959906,
      "grad_norm": 1.3125,
      "learning_rate": 0.00048096792105447613,
      "loss": 1.7036,
      "step": 29774
    },
    {
      "epoch": 2.5617310505033126,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0004809408999432871,
      "loss": 0.5955,
      "step": 29775
    },
    {
      "epoch": 2.5618170868106342,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004809138788878423,
      "loss": 0.8635,
      "step": 29776
    },
    {
      "epoch": 2.561903123117956,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004808868578882207,
      "loss": 1.2127,
      "step": 29777
    },
    {
      "epoch": 2.5619891594252775,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00048085983694450126,
      "loss": 1.1767,
      "step": 29778
    },
    {
      "epoch": 2.562075195732599,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004808328160567632,
      "loss": 1.6097,
      "step": 29779
    },
    {
      "epoch": 2.5621612320399207,
      "grad_norm": 1.140625,
      "learning_rate": 0.00048080579522508533,
      "loss": 0.8864,
      "step": 29780
    },
    {
      "epoch": 2.5622472683472424,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004807787744495468,
      "loss": 1.7691,
      "step": 29781
    },
    {
      "epoch": 2.5623333046545644,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004807517537302265,
      "loss": 1.2141,
      "step": 29782
    },
    {
      "epoch": 2.562419340961886,
      "grad_norm": 1.9375,
      "learning_rate": 0.0004807247330672037,
      "loss": 1.7761,
      "step": 29783
    },
    {
      "epoch": 2.5625053772692077,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004806977124605574,
      "loss": 1.3406,
      "step": 29784
    },
    {
      "epoch": 2.5625914135765293,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00048067069191036645,
      "loss": 0.6496,
      "step": 29785
    },
    {
      "epoch": 2.562677449883851,
      "grad_norm": 1.515625,
      "learning_rate": 0.0004806436714167098,
      "loss": 1.472,
      "step": 29786
    },
    {
      "epoch": 2.5627634861911726,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0004806166509796668,
      "loss": 0.808,
      "step": 29787
    },
    {
      "epoch": 2.562849522498494,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004805896305993164,
      "loss": 0.8746,
      "step": 29788
    },
    {
      "epoch": 2.5629355588058162,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004805626102757374,
      "loss": 1.7982,
      "step": 29789
    },
    {
      "epoch": 2.563021595113138,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004805355900090091,
      "loss": 1.1942,
      "step": 29790
    },
    {
      "epoch": 2.5631076314204595,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00048050856979921014,
      "loss": 1.2088,
      "step": 29791
    },
    {
      "epoch": 2.563193667727781,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004804815496464201,
      "loss": 1.3458,
      "step": 29792
    },
    {
      "epoch": 2.5632797040351027,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00048045452955071773,
      "loss": 1.1004,
      "step": 29793
    },
    {
      "epoch": 2.5633657403424244,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00048042750951218197,
      "loss": 1.3791,
      "step": 29794
    },
    {
      "epoch": 2.563451776649746,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000480400489530892,
      "loss": 1.2306,
      "step": 29795
    },
    {
      "epoch": 2.563537812957068,
      "grad_norm": 2.15625,
      "learning_rate": 0.0004803734696069267,
      "loss": 1.4191,
      "step": 29796
    },
    {
      "epoch": 2.5636238492643897,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004803464497403653,
      "loss": 1.4606,
      "step": 29797
    },
    {
      "epoch": 2.5637098855717113,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00048031942993128664,
      "loss": 1.456,
      "step": 29798
    },
    {
      "epoch": 2.563795921879033,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00048029241017976987,
      "loss": 0.581,
      "step": 29799
    },
    {
      "epoch": 2.5638819581863546,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00048026539048589396,
      "loss": 1.3514,
      "step": 29800
    },
    {
      "epoch": 2.563967994493676,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004802383708497379,
      "loss": 1.317,
      "step": 29801
    },
    {
      "epoch": 2.564054030800998,
      "grad_norm": 1.15625,
      "learning_rate": 0.00048021135127138077,
      "loss": 0.797,
      "step": 29802
    },
    {
      "epoch": 2.56414006710832,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00048018433175090163,
      "loss": 1.5739,
      "step": 29803
    },
    {
      "epoch": 2.5642261034156415,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004801573122883795,
      "loss": 0.8442,
      "step": 29804
    },
    {
      "epoch": 2.564312139722963,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004801302928838932,
      "loss": 0.9681,
      "step": 29805
    },
    {
      "epoch": 2.5643981760302847,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00048010327353752207,
      "loss": 1.5942,
      "step": 29806
    },
    {
      "epoch": 2.5644842123376064,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000480076254249345,
      "loss": 1.4111,
      "step": 29807
    },
    {
      "epoch": 2.564570248644928,
      "grad_norm": 1.046875,
      "learning_rate": 0.00048004923501944105,
      "loss": 0.8397,
      "step": 29808
    },
    {
      "epoch": 2.5646562849522496,
      "grad_norm": 1.046875,
      "learning_rate": 0.000480022215847889,
      "loss": 1.1617,
      "step": 29809
    },
    {
      "epoch": 2.5647423212595717,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0004799951967347683,
      "loss": 1.119,
      "step": 29810
    },
    {
      "epoch": 2.5648283575668933,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004799681776801577,
      "loss": 1.8297,
      "step": 29811
    },
    {
      "epoch": 2.564914393874215,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004799411586841362,
      "loss": 1.7139,
      "step": 29812
    },
    {
      "epoch": 2.5650004301815366,
      "grad_norm": 1.15625,
      "learning_rate": 0.000479914139746783,
      "loss": 0.8039,
      "step": 29813
    },
    {
      "epoch": 2.565086466488858,
      "grad_norm": 0.828125,
      "learning_rate": 0.00047988712086817703,
      "loss": 0.9536,
      "step": 29814
    },
    {
      "epoch": 2.5651725027961803,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004798601020483973,
      "loss": 1.0752,
      "step": 29815
    },
    {
      "epoch": 2.5652585391035014,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004798330832875228,
      "loss": 1.0881,
      "step": 29816
    },
    {
      "epoch": 2.5653445754108235,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004798060645856327,
      "loss": 1.5007,
      "step": 29817
    },
    {
      "epoch": 2.565430611718145,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00047977904594280593,
      "loss": 1.0346,
      "step": 29818
    },
    {
      "epoch": 2.5655166480254668,
      "grad_norm": 2.1875,
      "learning_rate": 0.0004797520273591214,
      "loss": 1.3198,
      "step": 29819
    },
    {
      "epoch": 2.5656026843327884,
      "grad_norm": 1.234375,
      "learning_rate": 0.00047972500883465834,
      "loss": 1.7287,
      "step": 29820
    },
    {
      "epoch": 2.56568872064011,
      "grad_norm": 3.03125,
      "learning_rate": 0.00047969799036949573,
      "loss": 1.1281,
      "step": 29821
    },
    {
      "epoch": 2.565774756947432,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004796709719637125,
      "loss": 1.047,
      "step": 29822
    },
    {
      "epoch": 2.5658607932547532,
      "grad_norm": 1.15625,
      "learning_rate": 0.00047964395361738767,
      "loss": 1.2434,
      "step": 29823
    },
    {
      "epoch": 2.5659468295620753,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0004796169353306004,
      "loss": 1.1684,
      "step": 29824
    },
    {
      "epoch": 2.566032865869397,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00047958991710342965,
      "loss": 0.8414,
      "step": 29825
    },
    {
      "epoch": 2.5661189021767186,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00047956289893595427,
      "loss": 1.4375,
      "step": 29826
    },
    {
      "epoch": 2.56620493848404,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047953588082825355,
      "loss": 1.5469,
      "step": 29827
    },
    {
      "epoch": 2.566290974791362,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004795088627804062,
      "loss": 1.0615,
      "step": 29828
    },
    {
      "epoch": 2.566377011098684,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004794818447924917,
      "loss": 1.7061,
      "step": 29829
    },
    {
      "epoch": 2.566463047406005,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047945482686458877,
      "loss": 0.7792,
      "step": 29830
    },
    {
      "epoch": 2.566549083713327,
      "grad_norm": 0.984375,
      "learning_rate": 0.00047942780899677646,
      "loss": 1.2798,
      "step": 29831
    },
    {
      "epoch": 2.5666351200206488,
      "grad_norm": 0.890625,
      "learning_rate": 0.00047940079118913374,
      "loss": 0.9763,
      "step": 29832
    },
    {
      "epoch": 2.5667211563279704,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00047937377344173975,
      "loss": 0.9859,
      "step": 29833
    },
    {
      "epoch": 2.566807192635292,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00047934675575467354,
      "loss": 0.603,
      "step": 29834
    },
    {
      "epoch": 2.5668932289426136,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00047931973812801387,
      "loss": 1.3777,
      "step": 29835
    },
    {
      "epoch": 2.5669792652499357,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004792927205618401,
      "loss": 1.7193,
      "step": 29836
    },
    {
      "epoch": 2.5670653015572573,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00047926570305623106,
      "loss": 1.6703,
      "step": 29837
    },
    {
      "epoch": 2.567151337864579,
      "grad_norm": 1.078125,
      "learning_rate": 0.00047923868561126583,
      "loss": 1.3861,
      "step": 29838
    },
    {
      "epoch": 2.5672373741719006,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047921166822702325,
      "loss": 1.2027,
      "step": 29839
    },
    {
      "epoch": 2.567323410479222,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004791846509035827,
      "loss": 1.3503,
      "step": 29840
    },
    {
      "epoch": 2.567409446786544,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004791576336410229,
      "loss": 2.1076,
      "step": 29841
    },
    {
      "epoch": 2.5674954830938654,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004791306164394229,
      "loss": 1.2421,
      "step": 29842
    },
    {
      "epoch": 2.5675815194011875,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00047910359929886193,
      "loss": 1.1042,
      "step": 29843
    },
    {
      "epoch": 2.567667555708509,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004790765822194189,
      "loss": 0.8876,
      "step": 29844
    },
    {
      "epoch": 2.5677535920158308,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004790495652011727,
      "loss": 0.9249,
      "step": 29845
    },
    {
      "epoch": 2.5678396283231524,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004790225482442024,
      "loss": 1.2001,
      "step": 29846
    },
    {
      "epoch": 2.567925664630474,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004789955313485871,
      "loss": 1.2094,
      "step": 29847
    },
    {
      "epoch": 2.5680117009377956,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004789685145144058,
      "loss": 1.3105,
      "step": 29848
    },
    {
      "epoch": 2.5680977372451173,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00047894149774173746,
      "loss": 1.4435,
      "step": 29849
    },
    {
      "epoch": 2.5681837735524393,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00047891448103066126,
      "loss": 1.3713,
      "step": 29850
    },
    {
      "epoch": 2.568269809859761,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00047888746438125605,
      "loss": 1.0523,
      "step": 29851
    },
    {
      "epoch": 2.5683558461670826,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004788604477936009,
      "loss": 1.6107,
      "step": 29852
    },
    {
      "epoch": 2.568441882474404,
      "grad_norm": 0.875,
      "learning_rate": 0.0004788334312677747,
      "loss": 1.2445,
      "step": 29853
    },
    {
      "epoch": 2.568527918781726,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00047880641480385676,
      "loss": 1.4052,
      "step": 29854
    },
    {
      "epoch": 2.5686139550890474,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00047877939840192596,
      "loss": 0.9154,
      "step": 29855
    },
    {
      "epoch": 2.568699991396369,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00047875238206206123,
      "loss": 1.5962,
      "step": 29856
    },
    {
      "epoch": 2.568786027703691,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00047872536578434156,
      "loss": 1.824,
      "step": 29857
    },
    {
      "epoch": 2.5688720640110128,
      "grad_norm": 0.65625,
      "learning_rate": 0.00047869834956884615,
      "loss": 0.5433,
      "step": 29858
    },
    {
      "epoch": 2.5689581003183344,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00047867133341565393,
      "loss": 1.084,
      "step": 29859
    },
    {
      "epoch": 2.569044136625656,
      "grad_norm": 0.875,
      "learning_rate": 0.0004786443173248439,
      "loss": 0.8838,
      "step": 29860
    },
    {
      "epoch": 2.5691301729329776,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00047861730129649505,
      "loss": 0.8525,
      "step": 29861
    },
    {
      "epoch": 2.5692162092402993,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00047859028533068647,
      "loss": 1.0403,
      "step": 29862
    },
    {
      "epoch": 2.569302245547621,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047856326942749723,
      "loss": 1.7611,
      "step": 29863
    },
    {
      "epoch": 2.569388281854943,
      "grad_norm": 1.59375,
      "learning_rate": 0.0004785362535870061,
      "loss": 1.6744,
      "step": 29864
    },
    {
      "epoch": 2.5694743181622646,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004785092378092921,
      "loss": 1.4522,
      "step": 29865
    },
    {
      "epoch": 2.569560354469586,
      "grad_norm": 1.4375,
      "learning_rate": 0.00047848222209443466,
      "loss": 1.5106,
      "step": 29866
    },
    {
      "epoch": 2.569646390776908,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004784552064425125,
      "loss": 1.6298,
      "step": 29867
    },
    {
      "epoch": 2.5697324270842294,
      "grad_norm": 1.5,
      "learning_rate": 0.0004784281908536047,
      "loss": 1.4972,
      "step": 29868
    },
    {
      "epoch": 2.5698184633915515,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00047840117532779014,
      "loss": 0.6136,
      "step": 29869
    },
    {
      "epoch": 2.5699044996988727,
      "grad_norm": 1.25,
      "learning_rate": 0.000478374159865148,
      "loss": 1.4273,
      "step": 29870
    },
    {
      "epoch": 2.5699905360061948,
      "grad_norm": 1.078125,
      "learning_rate": 0.00047834714446575734,
      "loss": 1.084,
      "step": 29871
    },
    {
      "epoch": 2.5700765723135164,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004783201291296969,
      "loss": 1.4324,
      "step": 29872
    },
    {
      "epoch": 2.570162608620838,
      "grad_norm": 0.875,
      "learning_rate": 0.000478293113857046,
      "loss": 0.8592,
      "step": 29873
    },
    {
      "epoch": 2.5702486449281596,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00047826609864788346,
      "loss": 1.0156,
      "step": 29874
    },
    {
      "epoch": 2.5703346812354813,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004782390835022884,
      "loss": 0.8112,
      "step": 29875
    },
    {
      "epoch": 2.5704207175428033,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00047821206842033965,
      "loss": 0.7259,
      "step": 29876
    },
    {
      "epoch": 2.5705067538501245,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004781850534021165,
      "loss": 2.189,
      "step": 29877
    },
    {
      "epoch": 2.5705927901574466,
      "grad_norm": 0.796875,
      "learning_rate": 0.00047815803844769784,
      "loss": 0.7473,
      "step": 29878
    },
    {
      "epoch": 2.570678826464768,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00047813102355716264,
      "loss": 1.4467,
      "step": 29879
    },
    {
      "epoch": 2.57076486277209,
      "grad_norm": 0.796875,
      "learning_rate": 0.00047810400873058985,
      "loss": 0.6606,
      "step": 29880
    },
    {
      "epoch": 2.5708508990794114,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00047807699396805865,
      "loss": 1.1353,
      "step": 29881
    },
    {
      "epoch": 2.570936935386733,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000478049979269648,
      "loss": 1.024,
      "step": 29882
    },
    {
      "epoch": 2.571022971694055,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00047802296463543683,
      "loss": 1.3092,
      "step": 29883
    },
    {
      "epoch": 2.5711090080013763,
      "grad_norm": 1.1875,
      "learning_rate": 0.00047799595006550423,
      "loss": 1.2313,
      "step": 29884
    },
    {
      "epoch": 2.5711950443086984,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004779689355599293,
      "loss": 0.6428,
      "step": 29885
    },
    {
      "epoch": 2.57128108061602,
      "grad_norm": 1.265625,
      "learning_rate": 0.00047794192111879084,
      "loss": 1.1566,
      "step": 29886
    },
    {
      "epoch": 2.5713671169233416,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00047791490674216797,
      "loss": 0.7503,
      "step": 29887
    },
    {
      "epoch": 2.5714531532306633,
      "grad_norm": 1.0625,
      "learning_rate": 0.00047788789243013974,
      "loss": 1.4047,
      "step": 29888
    },
    {
      "epoch": 2.571539189537985,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004778608781827851,
      "loss": 1.0542,
      "step": 29889
    },
    {
      "epoch": 2.571625225845307,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004778338640001831,
      "loss": 1.1087,
      "step": 29890
    },
    {
      "epoch": 2.5717112621526286,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00047780684988241276,
      "loss": 1.868,
      "step": 29891
    },
    {
      "epoch": 2.57179729845995,
      "grad_norm": 1.046875,
      "learning_rate": 0.00047777983582955304,
      "loss": 1.5128,
      "step": 29892
    },
    {
      "epoch": 2.571883334767272,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000477752821841683,
      "loss": 0.9701,
      "step": 29893
    },
    {
      "epoch": 2.5719693710745934,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00047772580791888156,
      "loss": 1.2992,
      "step": 29894
    },
    {
      "epoch": 2.572055407381915,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00047769879406122783,
      "loss": 1.1784,
      "step": 29895
    },
    {
      "epoch": 2.5721414436892367,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004776717802688009,
      "loss": 0.5815,
      "step": 29896
    },
    {
      "epoch": 2.5722274799965588,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004776447665416796,
      "loss": 1.0328,
      "step": 29897
    },
    {
      "epoch": 2.5723135163038804,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004776177528799429,
      "loss": 1.6923,
      "step": 29898
    },
    {
      "epoch": 2.572399552611202,
      "grad_norm": 1.28125,
      "learning_rate": 0.00047759073928367005,
      "loss": 1.5285,
      "step": 29899
    },
    {
      "epoch": 2.5724855889185236,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004775637257529399,
      "loss": 1.311,
      "step": 29900
    },
    {
      "epoch": 2.5725716252258453,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004775367122878314,
      "loss": 1.2616,
      "step": 29901
    },
    {
      "epoch": 2.572657661533167,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004775096988884237,
      "loss": 0.9629,
      "step": 29902
    },
    {
      "epoch": 2.5727436978404885,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004774826855547957,
      "loss": 1.387,
      "step": 29903
    },
    {
      "epoch": 2.5728297341478106,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004774556722870265,
      "loss": 1.5727,
      "step": 29904
    },
    {
      "epoch": 2.572915770455132,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004774286590851952,
      "loss": 1.0399,
      "step": 29905
    },
    {
      "epoch": 2.573001806762454,
      "grad_norm": 1.328125,
      "learning_rate": 0.00047740164594938054,
      "loss": 1.9798,
      "step": 29906
    },
    {
      "epoch": 2.5730878430697754,
      "grad_norm": 1.546875,
      "learning_rate": 0.00047737463287966174,
      "loss": 1.6482,
      "step": 29907
    },
    {
      "epoch": 2.573173879377097,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00047734761987611776,
      "loss": 0.9426,
      "step": 29908
    },
    {
      "epoch": 2.5732599156844187,
      "grad_norm": 1.15625,
      "learning_rate": 0.00047732060693882764,
      "loss": 1.154,
      "step": 29909
    },
    {
      "epoch": 2.5733459519917403,
      "grad_norm": 1.1875,
      "learning_rate": 0.00047729359406787015,
      "loss": 1.4054,
      "step": 29910
    },
    {
      "epoch": 2.5734319882990624,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00047726658126332464,
      "loss": 1.0945,
      "step": 29911
    },
    {
      "epoch": 2.573518024606384,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004772395685252699,
      "loss": 0.9753,
      "step": 29912
    },
    {
      "epoch": 2.5736040609137056,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004772125558537849,
      "loss": 1.4144,
      "step": 29913
    },
    {
      "epoch": 2.5736900972210273,
      "grad_norm": 0.90625,
      "learning_rate": 0.00047718554324894884,
      "loss": 1.0761,
      "step": 29914
    },
    {
      "epoch": 2.573776133528349,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004771585307108407,
      "loss": 1.5863,
      "step": 29915
    },
    {
      "epoch": 2.5738621698356705,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004771315182395394,
      "loss": 1.1524,
      "step": 29916
    },
    {
      "epoch": 2.573948206142992,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00047710450583512385,
      "loss": 1.067,
      "step": 29917
    },
    {
      "epoch": 2.574034242450314,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004770774934976732,
      "loss": 0.9706,
      "step": 29918
    },
    {
      "epoch": 2.574120278757636,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00047705048122726644,
      "loss": 1.0995,
      "step": 29919
    },
    {
      "epoch": 2.5742063150649575,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00047702346902398266,
      "loss": 1.6911,
      "step": 29920
    },
    {
      "epoch": 2.574292351372279,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00047699645688790054,
      "loss": 1.2777,
      "step": 29921
    },
    {
      "epoch": 2.5743783876796007,
      "grad_norm": 0.828125,
      "learning_rate": 0.00047696944481909946,
      "loss": 1.7551,
      "step": 29922
    },
    {
      "epoch": 2.5744644239869223,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004769424328176583,
      "loss": 1.0539,
      "step": 29923
    },
    {
      "epoch": 2.574550460294244,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00047691542088365595,
      "loss": 0.9353,
      "step": 29924
    },
    {
      "epoch": 2.574636496601566,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00047688840901717155,
      "loss": 1.327,
      "step": 29925
    },
    {
      "epoch": 2.5747225329088876,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00047686139721828406,
      "loss": 0.9562,
      "step": 29926
    },
    {
      "epoch": 2.5748085692162093,
      "grad_norm": 2.265625,
      "learning_rate": 0.0004768343854870726,
      "loss": 1.3363,
      "step": 29927
    },
    {
      "epoch": 2.574894605523531,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004768073738236158,
      "loss": 1.0826,
      "step": 29928
    },
    {
      "epoch": 2.5749806418308525,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004767803622279931,
      "loss": 0.707,
      "step": 29929
    },
    {
      "epoch": 2.5750666781381746,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00047675335070028333,
      "loss": 1.2775,
      "step": 29930
    },
    {
      "epoch": 2.5751527144454958,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00047672633924056534,
      "loss": 0.7965,
      "step": 29931
    },
    {
      "epoch": 2.575238750752818,
      "grad_norm": 1.109375,
      "learning_rate": 0.00047669932784891845,
      "loss": 0.7371,
      "step": 29932
    },
    {
      "epoch": 2.5753247870601395,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004766723165254214,
      "loss": 1.4955,
      "step": 29933
    },
    {
      "epoch": 2.575410823367461,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004766453052701534,
      "loss": 0.8986,
      "step": 29934
    },
    {
      "epoch": 2.5754968596747827,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004766182940831931,
      "loss": 1.4982,
      "step": 29935
    },
    {
      "epoch": 2.5755828959821043,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00047659128296461995,
      "loss": 1.7742,
      "step": 29936
    },
    {
      "epoch": 2.5756689322894264,
      "grad_norm": 1.046875,
      "learning_rate": 0.00047656427191451273,
      "loss": 1.2367,
      "step": 29937
    },
    {
      "epoch": 2.5757549685967476,
      "grad_norm": 0.78125,
      "learning_rate": 0.00047653726093295034,
      "loss": 0.6152,
      "step": 29938
    },
    {
      "epoch": 2.5758410049040696,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00047651025002001194,
      "loss": 0.7395,
      "step": 29939
    },
    {
      "epoch": 2.5759270412113913,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004764832391757764,
      "loss": 1.2169,
      "step": 29940
    },
    {
      "epoch": 2.576013077518713,
      "grad_norm": 1.015625,
      "learning_rate": 0.000476456228400323,
      "loss": 1.0564,
      "step": 29941
    },
    {
      "epoch": 2.5760991138260345,
      "grad_norm": 1.5625,
      "learning_rate": 0.0004764292176937305,
      "loss": 1.6979,
      "step": 29942
    },
    {
      "epoch": 2.576185150133356,
      "grad_norm": 0.875,
      "learning_rate": 0.00047640220705607797,
      "loss": 0.7566,
      "step": 29943
    },
    {
      "epoch": 2.576271186440678,
      "grad_norm": 2.359375,
      "learning_rate": 0.00047637519648744434,
      "loss": 0.8706,
      "step": 29944
    },
    {
      "epoch": 2.5763572227479994,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004763481859879087,
      "loss": 1.8436,
      "step": 29945
    },
    {
      "epoch": 2.5764432590553215,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00047632117555755,
      "loss": 0.5887,
      "step": 29946
    },
    {
      "epoch": 2.576529295362643,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00047629416519644717,
      "loss": 1.2624,
      "step": 29947
    },
    {
      "epoch": 2.5766153316699647,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00047626715490467947,
      "loss": 1.1891,
      "step": 29948
    },
    {
      "epoch": 2.5767013679772863,
      "grad_norm": 1.125,
      "learning_rate": 0.0004762401446823257,
      "loss": 1.8406,
      "step": 29949
    },
    {
      "epoch": 2.576787404284608,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00047621313452946484,
      "loss": 0.8746,
      "step": 29950
    },
    {
      "epoch": 2.57687344059193,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0004761861244461759,
      "loss": 0.4599,
      "step": 29951
    },
    {
      "epoch": 2.5769594768992516,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004761591144325379,
      "loss": 1.1999,
      "step": 29952
    },
    {
      "epoch": 2.5770455132065733,
      "grad_norm": 1.078125,
      "learning_rate": 0.00047613210448863,
      "loss": 1.0013,
      "step": 29953
    },
    {
      "epoch": 2.577131549513895,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00047610509461453083,
      "loss": 1.8953,
      "step": 29954
    },
    {
      "epoch": 2.5772175858212165,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004760780848103198,
      "loss": 1.2632,
      "step": 29955
    },
    {
      "epoch": 2.577303622128538,
      "grad_norm": 1.375,
      "learning_rate": 0.00047605107507607563,
      "loss": 1.3544,
      "step": 29956
    },
    {
      "epoch": 2.5773896584358598,
      "grad_norm": 1.0,
      "learning_rate": 0.0004760240654118775,
      "loss": 1.3316,
      "step": 29957
    },
    {
      "epoch": 2.577475694743182,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00047599705581780413,
      "loss": 1.1988,
      "step": 29958
    },
    {
      "epoch": 2.5775617310505035,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00047597004629393487,
      "loss": 0.8677,
      "step": 29959
    },
    {
      "epoch": 2.577647767357825,
      "grad_norm": 1.03125,
      "learning_rate": 0.00047594303684034853,
      "loss": 0.9497,
      "step": 29960
    },
    {
      "epoch": 2.5777338036651467,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004759160274571241,
      "loss": 1.7148,
      "step": 29961
    },
    {
      "epoch": 2.5778198399724683,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00047588901814434053,
      "loss": 0.5011,
      "step": 29962
    },
    {
      "epoch": 2.57790587627979,
      "grad_norm": 1.109375,
      "learning_rate": 0.00047586200890207696,
      "loss": 1.0248,
      "step": 29963
    },
    {
      "epoch": 2.5779919125871116,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004758349997304124,
      "loss": 0.6818,
      "step": 29964
    },
    {
      "epoch": 2.5780779488944336,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00047580799062942557,
      "loss": 1.1047,
      "step": 29965
    },
    {
      "epoch": 2.5781639852017553,
      "grad_norm": 1.265625,
      "learning_rate": 0.00047578098159919584,
      "loss": 1.6033,
      "step": 29966
    },
    {
      "epoch": 2.578250021509077,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000475753972639802,
      "loss": 1.3784,
      "step": 29967
    },
    {
      "epoch": 2.5783360578163985,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004757269637513231,
      "loss": 1.088,
      "step": 29968
    },
    {
      "epoch": 2.57842209412372,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00047569995493383795,
      "loss": 0.7228,
      "step": 29969
    },
    {
      "epoch": 2.5785081304310418,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00047567294618742584,
      "loss": 1.008,
      "step": 29970
    },
    {
      "epoch": 2.5785941667383634,
      "grad_norm": 1.078125,
      "learning_rate": 0.00047564593751216563,
      "loss": 0.8418,
      "step": 29971
    },
    {
      "epoch": 2.5786802030456855,
      "grad_norm": 1.0625,
      "learning_rate": 0.00047561892890813625,
      "loss": 1.4457,
      "step": 29972
    },
    {
      "epoch": 2.578766239353007,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004755919203754168,
      "loss": 0.9842,
      "step": 29973
    },
    {
      "epoch": 2.5788522756603287,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004755649119140862,
      "loss": 1.6458,
      "step": 29974
    },
    {
      "epoch": 2.5789383119676503,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004755379035242236,
      "loss": 0.9034,
      "step": 29975
    },
    {
      "epoch": 2.579024348274972,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004755108952059076,
      "loss": 1.4338,
      "step": 29976
    },
    {
      "epoch": 2.5791103845822936,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047548388695921757,
      "loss": 1.3943,
      "step": 29977
    },
    {
      "epoch": 2.579196420889615,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004754568787842325,
      "loss": 0.9078,
      "step": 29978
    },
    {
      "epoch": 2.5792824571969373,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0004754298706810313,
      "loss": 1.6601,
      "step": 29979
    },
    {
      "epoch": 2.579368493504259,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00047540286264969304,
      "loss": 0.8447,
      "step": 29980
    },
    {
      "epoch": 2.5794545298115805,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004753758546902964,
      "loss": 1.5521,
      "step": 29981
    },
    {
      "epoch": 2.579540566118902,
      "grad_norm": 1.328125,
      "learning_rate": 0.00047534884680292076,
      "loss": 0.8679,
      "step": 29982
    },
    {
      "epoch": 2.5796266024262238,
      "grad_norm": 0.984375,
      "learning_rate": 0.00047532183898764494,
      "loss": 0.8334,
      "step": 29983
    },
    {
      "epoch": 2.5797126387335454,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004752948312445479,
      "loss": 0.7615,
      "step": 29984
    },
    {
      "epoch": 2.579798675040867,
      "grad_norm": 1.015625,
      "learning_rate": 0.00047526782357370855,
      "loss": 1.1634,
      "step": 29985
    },
    {
      "epoch": 2.579884711348189,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00047524081597520617,
      "loss": 0.6098,
      "step": 29986
    },
    {
      "epoch": 2.5799707476555107,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004752138084491196,
      "loss": 1.0546,
      "step": 29987
    },
    {
      "epoch": 2.5800567839628323,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00047518680099552766,
      "loss": 0.7749,
      "step": 29988
    },
    {
      "epoch": 2.580142820270154,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004751597936145097,
      "loss": 1.1695,
      "step": 29989
    },
    {
      "epoch": 2.5802288565774756,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00047513278630614435,
      "loss": 1.0054,
      "step": 29990
    },
    {
      "epoch": 2.5803148928847977,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004751057790705109,
      "loss": 1.1605,
      "step": 29991
    },
    {
      "epoch": 2.580400929192119,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00047507877190768803,
      "loss": 1.1641,
      "step": 29992
    },
    {
      "epoch": 2.580486965499441,
      "grad_norm": 1.046875,
      "learning_rate": 0.00047505176481775507,
      "loss": 0.8294,
      "step": 29993
    },
    {
      "epoch": 2.5805730018067625,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004750247578007908,
      "loss": 1.0797,
      "step": 29994
    },
    {
      "epoch": 2.580659038114084,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004749977508568742,
      "loss": 0.9049,
      "step": 29995
    },
    {
      "epoch": 2.5807450744214058,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00047497074398608434,
      "loss": 1.4422,
      "step": 29996
    },
    {
      "epoch": 2.5808311107287274,
      "grad_norm": 1.4375,
      "learning_rate": 0.00047494373718850026,
      "loss": 2.0783,
      "step": 29997
    },
    {
      "epoch": 2.5809171470360495,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004749167304642009,
      "loss": 1.2336,
      "step": 29998
    },
    {
      "epoch": 2.5810031833433706,
      "grad_norm": 1.5546875,
      "learning_rate": 0.000474889723813265,
      "loss": 1.4774,
      "step": 29999
    },
    {
      "epoch": 2.5810892196506927,
      "grad_norm": 1.1875,
      "learning_rate": 0.00047486271723577196,
      "loss": 0.9009,
      "step": 30000
    },
    {
      "epoch": 2.5811752559580143,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004748357107318006,
      "loss": 0.8483,
      "step": 30001
    },
    {
      "epoch": 2.581261292265336,
      "grad_norm": 1.03125,
      "learning_rate": 0.00047480870430142974,
      "loss": 1.647,
      "step": 30002
    },
    {
      "epoch": 2.5813473285726576,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00047478169794473864,
      "loss": 1.0766,
      "step": 30003
    },
    {
      "epoch": 2.581433364879979,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004747546916618062,
      "loss": 1.7027,
      "step": 30004
    },
    {
      "epoch": 2.5815194011873013,
      "grad_norm": 1.265625,
      "learning_rate": 0.00047472768545271134,
      "loss": 1.3165,
      "step": 30005
    },
    {
      "epoch": 2.5816054374946225,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000474700679317533,
      "loss": 0.7647,
      "step": 30006
    },
    {
      "epoch": 2.5816914738019445,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004746736732563503,
      "loss": 1.3389,
      "step": 30007
    },
    {
      "epoch": 2.581777510109266,
      "grad_norm": 1.140625,
      "learning_rate": 0.00047464666726924226,
      "loss": 1.846,
      "step": 30008
    },
    {
      "epoch": 2.5818635464165878,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00047461966135628773,
      "loss": 1.4789,
      "step": 30009
    },
    {
      "epoch": 2.5819495827239094,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004745926555175657,
      "loss": 1.6325,
      "step": 30010
    },
    {
      "epoch": 2.582035619031231,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00047456564975315523,
      "loss": 2.1258,
      "step": 30011
    },
    {
      "epoch": 2.582121655338553,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004745386440631354,
      "loss": 1.1586,
      "step": 30012
    },
    {
      "epoch": 2.5822076916458747,
      "grad_norm": 1.375,
      "learning_rate": 0.00047451163844758484,
      "loss": 1.6299,
      "step": 30013
    },
    {
      "epoch": 2.5822937279531963,
      "grad_norm": 0.90625,
      "learning_rate": 0.00047448463290658294,
      "loss": 1.1177,
      "step": 30014
    },
    {
      "epoch": 2.582379764260518,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004744576274402084,
      "loss": 0.8283,
      "step": 30015
    },
    {
      "epoch": 2.5824658005678396,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004744306220485405,
      "loss": 1.3999,
      "step": 30016
    },
    {
      "epoch": 2.582551836875161,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004744036167316581,
      "loss": 1.1289,
      "step": 30017
    },
    {
      "epoch": 2.582637873182483,
      "grad_norm": 1.296875,
      "learning_rate": 0.00047437661148963994,
      "loss": 1.4181,
      "step": 30018
    },
    {
      "epoch": 2.582723909489805,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004743496063225654,
      "loss": 1.2377,
      "step": 30019
    },
    {
      "epoch": 2.5828099457971265,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00047432260123051324,
      "loss": 1.3398,
      "step": 30020
    },
    {
      "epoch": 2.582895982104448,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004742955962135624,
      "loss": 0.9727,
      "step": 30021
    },
    {
      "epoch": 2.5829820184117698,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00047426859127179196,
      "loss": 1.7263,
      "step": 30022
    },
    {
      "epoch": 2.5830680547190914,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004742415864052809,
      "loss": 0.7242,
      "step": 30023
    },
    {
      "epoch": 2.583154091026413,
      "grad_norm": 0.828125,
      "learning_rate": 0.00047421458161410826,
      "loss": 1.0107,
      "step": 30024
    },
    {
      "epoch": 2.5832401273337346,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0004741875768983528,
      "loss": 1.4437,
      "step": 30025
    },
    {
      "epoch": 2.5833261636410567,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004741605722580938,
      "loss": 0.864,
      "step": 30026
    },
    {
      "epoch": 2.5834121999483783,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004741335676934101,
      "loss": 1.2218,
      "step": 30027
    },
    {
      "epoch": 2.5834982362557,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004741065632043806,
      "loss": 1.6737,
      "step": 30028
    },
    {
      "epoch": 2.5835842725630216,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00047407955879108434,
      "loss": 0.8301,
      "step": 30029
    },
    {
      "epoch": 2.583670308870343,
      "grad_norm": 1.25,
      "learning_rate": 0.0004740525544536004,
      "loss": 1.1358,
      "step": 30030
    },
    {
      "epoch": 2.583756345177665,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004740255501920078,
      "loss": 1.0968,
      "step": 30031
    },
    {
      "epoch": 2.5838423814849865,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00047399854600638523,
      "loss": 1.3362,
      "step": 30032
    },
    {
      "epoch": 2.5839284177923085,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004739715418968119,
      "loss": 1.9452,
      "step": 30033
    },
    {
      "epoch": 2.58401445409963,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00047394453786336676,
      "loss": 1.48,
      "step": 30034
    },
    {
      "epoch": 2.5841004904069518,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00047391753390612886,
      "loss": 0.8413,
      "step": 30035
    },
    {
      "epoch": 2.5841865267142734,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004738905300251769,
      "loss": 1.2949,
      "step": 30036
    },
    {
      "epoch": 2.584272563021595,
      "grad_norm": 1.0,
      "learning_rate": 0.00047386352622059025,
      "loss": 1.821,
      "step": 30037
    },
    {
      "epoch": 2.5843585993289167,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004738365224924477,
      "loss": 1.0987,
      "step": 30038
    },
    {
      "epoch": 2.5844446356362383,
      "grad_norm": 1.140625,
      "learning_rate": 0.00047380951884082816,
      "loss": 1.1205,
      "step": 30039
    },
    {
      "epoch": 2.5845306719435603,
      "grad_norm": 0.875,
      "learning_rate": 0.0004737825152658106,
      "loss": 0.697,
      "step": 30040
    },
    {
      "epoch": 2.584616708250882,
      "grad_norm": 0.921875,
      "learning_rate": 0.00047375551176747423,
      "loss": 0.7069,
      "step": 30041
    },
    {
      "epoch": 2.5847027445582036,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00047372850834589785,
      "loss": 1.4204,
      "step": 30042
    },
    {
      "epoch": 2.584788780865525,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004737015050011604,
      "loss": 1.1719,
      "step": 30043
    },
    {
      "epoch": 2.584874817172847,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000473674501733341,
      "loss": 1.2076,
      "step": 30044
    },
    {
      "epoch": 2.584960853480169,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00047364749854251863,
      "loss": 0.8603,
      "step": 30045
    },
    {
      "epoch": 2.58504688978749,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004736204954287722,
      "loss": 1.0418,
      "step": 30046
    },
    {
      "epoch": 2.585132926094812,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004735934923921805,
      "loss": 1.1468,
      "step": 30047
    },
    {
      "epoch": 2.585218962402134,
      "grad_norm": 1.90625,
      "learning_rate": 0.0004735664894328228,
      "loss": 0.9039,
      "step": 30048
    },
    {
      "epoch": 2.5853049987094554,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00047353948655077805,
      "loss": 1.4018,
      "step": 30049
    },
    {
      "epoch": 2.585391035016777,
      "grad_norm": 1.140625,
      "learning_rate": 0.00047351248374612496,
      "loss": 1.6837,
      "step": 30050
    },
    {
      "epoch": 2.5854770713240987,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00047348548101894283,
      "loss": 0.9118,
      "step": 30051
    },
    {
      "epoch": 2.5855631076314207,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004734584783693104,
      "loss": 1.6497,
      "step": 30052
    },
    {
      "epoch": 2.585649143938742,
      "grad_norm": 1.03125,
      "learning_rate": 0.00047343147579730695,
      "loss": 1.3923,
      "step": 30053
    },
    {
      "epoch": 2.585735180246064,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00047340447330301123,
      "loss": 1.7609,
      "step": 30054
    },
    {
      "epoch": 2.5858212165533856,
      "grad_norm": 1.21875,
      "learning_rate": 0.00047337747088650224,
      "loss": 1.3287,
      "step": 30055
    },
    {
      "epoch": 2.585907252860707,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004733504685478589,
      "loss": 0.8326,
      "step": 30056
    },
    {
      "epoch": 2.585993289168029,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00047332346628716035,
      "loss": 1.0776,
      "step": 30057
    },
    {
      "epoch": 2.5860793254753505,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004732964641044855,
      "loss": 0.884,
      "step": 30058
    },
    {
      "epoch": 2.5861653617826725,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004732694619999131,
      "loss": 1.0652,
      "step": 30059
    },
    {
      "epoch": 2.5862513980899937,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004732424599735225,
      "loss": 1.9646,
      "step": 30060
    },
    {
      "epoch": 2.586337434397316,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00047321545802539244,
      "loss": 1.2573,
      "step": 30061
    },
    {
      "epoch": 2.5864234707046374,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00047318845615560204,
      "loss": 1.474,
      "step": 30062
    },
    {
      "epoch": 2.586509507011959,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00047316145436423004,
      "loss": 1.0045,
      "step": 30063
    },
    {
      "epoch": 2.5865955433192807,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004731344526513557,
      "loss": 1.1748,
      "step": 30064
    },
    {
      "epoch": 2.5866815796266023,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004731074510170578,
      "loss": 1.6611,
      "step": 30065
    },
    {
      "epoch": 2.5867676159339243,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004730804494614153,
      "loss": 2.0559,
      "step": 30066
    },
    {
      "epoch": 2.586853652241246,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00047305344798450737,
      "loss": 0.5756,
      "step": 30067
    },
    {
      "epoch": 2.5869396885485676,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047302644658641285,
      "loss": 1.8656,
      "step": 30068
    },
    {
      "epoch": 2.587025724855889,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004729994452672108,
      "loss": 1.4868,
      "step": 30069
    },
    {
      "epoch": 2.587111761163211,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004729724440269799,
      "loss": 1.1169,
      "step": 30070
    },
    {
      "epoch": 2.5871977974705325,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004729454428657995,
      "loss": 1.3907,
      "step": 30071
    },
    {
      "epoch": 2.587283833777854,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00047291844178374843,
      "loss": 1.0629,
      "step": 30072
    },
    {
      "epoch": 2.587369870085176,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004728914407809056,
      "loss": 1.0394,
      "step": 30073
    },
    {
      "epoch": 2.587455906392498,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00047286443985735,
      "loss": 0.9604,
      "step": 30074
    },
    {
      "epoch": 2.5875419426998194,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004728374390131607,
      "loss": 1.4061,
      "step": 30075
    },
    {
      "epoch": 2.587627979007141,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004728104382484166,
      "loss": 1.103,
      "step": 30076
    },
    {
      "epoch": 2.5877140153144627,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004727834375631966,
      "loss": 1.3382,
      "step": 30077
    },
    {
      "epoch": 2.5878000516217843,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004727564369575799,
      "loss": 0.7608,
      "step": 30078
    },
    {
      "epoch": 2.587886087929106,
      "grad_norm": 1.03125,
      "learning_rate": 0.00047272943643164524,
      "loss": 1.6841,
      "step": 30079
    },
    {
      "epoch": 2.587972124236428,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004727024359854717,
      "loss": 0.9936,
      "step": 30080
    },
    {
      "epoch": 2.5880581605437496,
      "grad_norm": 1.0625,
      "learning_rate": 0.00047267543561913813,
      "loss": 1.3805,
      "step": 30081
    },
    {
      "epoch": 2.588144196851071,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00047264843533272375,
      "loss": 1.21,
      "step": 30082
    },
    {
      "epoch": 2.588230233158393,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004726214351263073,
      "loss": 0.8883,
      "step": 30083
    },
    {
      "epoch": 2.5883162694657145,
      "grad_norm": 1.234375,
      "learning_rate": 0.00047259443499996775,
      "loss": 1.234,
      "step": 30084
    },
    {
      "epoch": 2.588402305773036,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004725674349537842,
      "loss": 0.6543,
      "step": 30085
    },
    {
      "epoch": 2.5884883420803577,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00047254043498783565,
      "loss": 0.906,
      "step": 30086
    },
    {
      "epoch": 2.58857437838768,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004725134351022007,
      "loss": 1.3602,
      "step": 30087
    },
    {
      "epoch": 2.5886604146950014,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004724864352969589,
      "loss": 1.4556,
      "step": 30088
    },
    {
      "epoch": 2.588746451002323,
      "grad_norm": 1.5625,
      "learning_rate": 0.0004724594355721888,
      "loss": 1.3074,
      "step": 30089
    },
    {
      "epoch": 2.5888324873096447,
      "grad_norm": 1.109375,
      "learning_rate": 0.00047243243592796956,
      "loss": 1.5055,
      "step": 30090
    },
    {
      "epoch": 2.5889185236169663,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004724054363643801,
      "loss": 0.8692,
      "step": 30091
    },
    {
      "epoch": 2.589004559924288,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00047237843688149937,
      "loss": 1.6613,
      "step": 30092
    },
    {
      "epoch": 2.5890905962316095,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00047235143747940625,
      "loss": 1.6598,
      "step": 30093
    },
    {
      "epoch": 2.5891766325389316,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004723244381581799,
      "loss": 0.9036,
      "step": 30094
    },
    {
      "epoch": 2.5892626688462532,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00047229743891789915,
      "loss": 0.9011,
      "step": 30095
    },
    {
      "epoch": 2.589348705153575,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004722704397586431,
      "loss": 1.5782,
      "step": 30096
    },
    {
      "epoch": 2.5894347414608965,
      "grad_norm": 1.3125,
      "learning_rate": 0.00047224344068049043,
      "loss": 1.6034,
      "step": 30097
    },
    {
      "epoch": 2.589520777768218,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004722164416835204,
      "loss": 0.6027,
      "step": 30098
    },
    {
      "epoch": 2.5896068140755397,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004721894427678119,
      "loss": 1.1859,
      "step": 30099
    },
    {
      "epoch": 2.5896928503828613,
      "grad_norm": 1.28125,
      "learning_rate": 0.00047216244393344375,
      "loss": 1.8211,
      "step": 30100
    },
    {
      "epoch": 2.5897788866901834,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00047213544518049516,
      "loss": 1.6385,
      "step": 30101
    },
    {
      "epoch": 2.589864922997505,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000472108446509045,
      "loss": 1.588,
      "step": 30102
    },
    {
      "epoch": 2.5899509593048267,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004720814479191722,
      "loss": 1.1302,
      "step": 30103
    },
    {
      "epoch": 2.5900369956121483,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0004720544494109556,
      "loss": 1.3052,
      "step": 30104
    },
    {
      "epoch": 2.59012303191947,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004720274509844744,
      "loss": 0.9246,
      "step": 30105
    },
    {
      "epoch": 2.590209068226792,
      "grad_norm": 1.0,
      "learning_rate": 0.00047200045263980745,
      "loss": 1.2461,
      "step": 30106
    },
    {
      "epoch": 2.590295104534113,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00047197345437703365,
      "loss": 1.3578,
      "step": 30107
    },
    {
      "epoch": 2.5903811408414352,
      "grad_norm": 1.28125,
      "learning_rate": 0.00047194645619623215,
      "loss": 1.1322,
      "step": 30108
    },
    {
      "epoch": 2.590467177148757,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00047191945809748183,
      "loss": 0.9379,
      "step": 30109
    },
    {
      "epoch": 2.5905532134560785,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004718924600808616,
      "loss": 0.9123,
      "step": 30110
    },
    {
      "epoch": 2.5906392497634,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004718654621464504,
      "loss": 1.4951,
      "step": 30111
    },
    {
      "epoch": 2.5907252860707217,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004718384642943273,
      "loss": 0.9362,
      "step": 30112
    },
    {
      "epoch": 2.590811322378044,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004718114665245713,
      "loss": 1.3563,
      "step": 30113
    },
    {
      "epoch": 2.590897358685365,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004717844688372611,
      "loss": 1.3421,
      "step": 30114
    },
    {
      "epoch": 2.590983394992687,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000471757471232476,
      "loss": 0.8469,
      "step": 30115
    },
    {
      "epoch": 2.5910694313000087,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004717304737102947,
      "loss": 1.1225,
      "step": 30116
    },
    {
      "epoch": 2.5911554676073303,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00047170347627079637,
      "loss": 0.5621,
      "step": 30117
    },
    {
      "epoch": 2.591241503914652,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004716764789140597,
      "loss": 1.3385,
      "step": 30118
    },
    {
      "epoch": 2.5913275402219735,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000471649481640164,
      "loss": 1.7904,
      "step": 30119
    },
    {
      "epoch": 2.5914135765292956,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000471622484449188,
      "loss": 0.5962,
      "step": 30120
    },
    {
      "epoch": 2.591499612836617,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004715954873412107,
      "loss": 1.4493,
      "step": 30121
    },
    {
      "epoch": 2.591585649143939,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00047156849031631094,
      "loss": 1.1798,
      "step": 30122
    },
    {
      "epoch": 2.5916716854512605,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000471541493374568,
      "loss": 0.989,
      "step": 30123
    },
    {
      "epoch": 2.591757721758582,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004715144965160604,
      "loss": 1.6271,
      "step": 30124
    },
    {
      "epoch": 2.5918437580659037,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004714874997408676,
      "loss": 0.725,
      "step": 30125
    },
    {
      "epoch": 2.5919297943732253,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00047146050304906826,
      "loss": 1.4236,
      "step": 30126
    },
    {
      "epoch": 2.5920158306805474,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00047143350644074134,
      "loss": 0.6561,
      "step": 30127
    },
    {
      "epoch": 2.592101866987869,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00047140650991596596,
      "loss": 1.6612,
      "step": 30128
    },
    {
      "epoch": 2.5921879032951907,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000471379513474821,
      "loss": 0.9709,
      "step": 30129
    },
    {
      "epoch": 2.5922739396025123,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00047135251711738523,
      "loss": 0.626,
      "step": 30130
    },
    {
      "epoch": 2.592359975909834,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004713255208437379,
      "loss": 0.9299,
      "step": 30131
    },
    {
      "epoch": 2.5924460122171555,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004712985246539579,
      "loss": 1.3232,
      "step": 30132
    },
    {
      "epoch": 2.592532048524477,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004712715285481241,
      "loss": 0.9162,
      "step": 30133
    },
    {
      "epoch": 2.5926180848317992,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00047124453252631537,
      "loss": 1.3168,
      "step": 30134
    },
    {
      "epoch": 2.592704121139121,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004712175365886109,
      "loss": 0.502,
      "step": 30135
    },
    {
      "epoch": 2.5927901574464425,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00047119054073508954,
      "loss": 1.3377,
      "step": 30136
    },
    {
      "epoch": 2.592876193753764,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00047116354496583025,
      "loss": 0.774,
      "step": 30137
    },
    {
      "epoch": 2.5929622300610857,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00047113654928091186,
      "loss": 1.0711,
      "step": 30138
    },
    {
      "epoch": 2.5930482663684074,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004711095536804136,
      "loss": 0.6583,
      "step": 30139
    },
    {
      "epoch": 2.593134302675729,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00047108255816441427,
      "loss": 1.1417,
      "step": 30140
    },
    {
      "epoch": 2.593220338983051,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00047105556273299265,
      "loss": 1.4432,
      "step": 30141
    },
    {
      "epoch": 2.5933063752903727,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004710285673862281,
      "loss": 1.149,
      "step": 30142
    },
    {
      "epoch": 2.5933924115976943,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004710015721241992,
      "loss": 1.3283,
      "step": 30143
    },
    {
      "epoch": 2.593478447905016,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004709745769469852,
      "loss": 1.4961,
      "step": 30144
    },
    {
      "epoch": 2.5935644842123375,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004709475818546648,
      "loss": 0.783,
      "step": 30145
    },
    {
      "epoch": 2.593650520519659,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00047092058684731716,
      "loss": 0.8622,
      "step": 30146
    },
    {
      "epoch": 2.593736556826981,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004708935919250211,
      "loss": 1.2177,
      "step": 30147
    },
    {
      "epoch": 2.593822593134303,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00047086659708785557,
      "loss": 1.2039,
      "step": 30148
    },
    {
      "epoch": 2.5939086294416245,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00047083960233589966,
      "loss": 1.1501,
      "step": 30149
    },
    {
      "epoch": 2.593994665748946,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00047081260766923224,
      "loss": 1.0409,
      "step": 30150
    },
    {
      "epoch": 2.5940807020562677,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004707856130879323,
      "loss": 0.6498,
      "step": 30151
    },
    {
      "epoch": 2.5941667383635894,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004707586185920786,
      "loss": 1.3919,
      "step": 30152
    },
    {
      "epoch": 2.594252774670911,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00047073162418175034,
      "loss": 0.9339,
      "step": 30153
    },
    {
      "epoch": 2.5943388109782326,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004707046298570265,
      "loss": 1.5746,
      "step": 30154
    },
    {
      "epoch": 2.5944248472855547,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00047067763561798573,
      "loss": 0.8522,
      "step": 30155
    },
    {
      "epoch": 2.5945108835928763,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00047065064146470734,
      "loss": 1.2719,
      "step": 30156
    },
    {
      "epoch": 2.594596919900198,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00047062364739727005,
      "loss": 1.3758,
      "step": 30157
    },
    {
      "epoch": 2.5946829562075195,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00047059665341575287,
      "loss": 1.4397,
      "step": 30158
    },
    {
      "epoch": 2.594768992514841,
      "grad_norm": 0.71875,
      "learning_rate": 0.00047056965952023465,
      "loss": 1.0272,
      "step": 30159
    },
    {
      "epoch": 2.594855028822163,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004705426657107946,
      "loss": 1.1408,
      "step": 30160
    },
    {
      "epoch": 2.5949410651294844,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047051567198751133,
      "loss": 1.3506,
      "step": 30161
    },
    {
      "epoch": 2.5950271014368065,
      "grad_norm": 0.9375,
      "learning_rate": 0.00047048867835046415,
      "loss": 1.2448,
      "step": 30162
    },
    {
      "epoch": 2.595113137744128,
      "grad_norm": 1.53125,
      "learning_rate": 0.0004704616847997319,
      "loss": 1.0191,
      "step": 30163
    },
    {
      "epoch": 2.5951991740514497,
      "grad_norm": 0.859375,
      "learning_rate": 0.00047043469133539327,
      "loss": 0.6404,
      "step": 30164
    },
    {
      "epoch": 2.5952852103587714,
      "grad_norm": 1.03125,
      "learning_rate": 0.00047040769795752765,
      "loss": 1.4233,
      "step": 30165
    },
    {
      "epoch": 2.595371246666093,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0004703807046662137,
      "loss": 1.2242,
      "step": 30166
    },
    {
      "epoch": 2.595457282973415,
      "grad_norm": 1.03125,
      "learning_rate": 0.00047035371146153044,
      "loss": 1.0944,
      "step": 30167
    },
    {
      "epoch": 2.5955433192807362,
      "grad_norm": 1.53125,
      "learning_rate": 0.00047032671834355664,
      "loss": 1.263,
      "step": 30168
    },
    {
      "epoch": 2.5956293555880583,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004702997253123716,
      "loss": 0.898,
      "step": 30169
    },
    {
      "epoch": 2.59571539189538,
      "grad_norm": 1.28125,
      "learning_rate": 0.00047027273236805405,
      "loss": 1.5957,
      "step": 30170
    },
    {
      "epoch": 2.5958014282027015,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004702457395106829,
      "loss": 0.987,
      "step": 30171
    },
    {
      "epoch": 2.595887464510023,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00047021874674033727,
      "loss": 0.9229,
      "step": 30172
    },
    {
      "epoch": 2.595973500817345,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000470191754057096,
      "loss": 1.2019,
      "step": 30173
    },
    {
      "epoch": 2.596059537124667,
      "grad_norm": 1.015625,
      "learning_rate": 0.000470164761461038,
      "loss": 0.962,
      "step": 30174
    },
    {
      "epoch": 2.596145573431988,
      "grad_norm": 1.09375,
      "learning_rate": 0.00047013776895224224,
      "loss": 1.6038,
      "step": 30175
    },
    {
      "epoch": 2.59623160973931,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00047011077653078784,
      "loss": 0.8656,
      "step": 30176
    },
    {
      "epoch": 2.5963176460466317,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00047008378419675353,
      "loss": 0.9525,
      "step": 30177
    },
    {
      "epoch": 2.5964036823539534,
      "grad_norm": 1.796875,
      "learning_rate": 0.00047005679195021824,
      "loss": 1.127,
      "step": 30178
    },
    {
      "epoch": 2.596489718661275,
      "grad_norm": 1.265625,
      "learning_rate": 0.00047002979979126117,
      "loss": 1.3952,
      "step": 30179
    },
    {
      "epoch": 2.5965757549685966,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00047000280771996103,
      "loss": 1.091,
      "step": 30180
    },
    {
      "epoch": 2.5966617912759187,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00046997581573639686,
      "loss": 1.2185,
      "step": 30181
    },
    {
      "epoch": 2.59674782758324,
      "grad_norm": 1.171875,
      "learning_rate": 0.00046994882384064754,
      "loss": 1.087,
      "step": 30182
    },
    {
      "epoch": 2.596833863890562,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004699218320327921,
      "loss": 0.7509,
      "step": 30183
    },
    {
      "epoch": 2.5969199001978835,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004698948403129095,
      "loss": 0.9755,
      "step": 30184
    },
    {
      "epoch": 2.597005936505205,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004698678486810786,
      "loss": 0.8594,
      "step": 30185
    },
    {
      "epoch": 2.597091972812527,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004698408571373783,
      "loss": 1.5879,
      "step": 30186
    },
    {
      "epoch": 2.5971780091198484,
      "grad_norm": 0.875,
      "learning_rate": 0.00046981386568188774,
      "loss": 0.8609,
      "step": 30187
    },
    {
      "epoch": 2.5972640454271705,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004697868743146857,
      "loss": 0.853,
      "step": 30188
    },
    {
      "epoch": 2.597350081734492,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00046975988303585113,
      "loss": 1.7277,
      "step": 30189
    },
    {
      "epoch": 2.5974361180418137,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004697328918454631,
      "loss": 1.7078,
      "step": 30190
    },
    {
      "epoch": 2.5975221543491354,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004697059007436005,
      "loss": 1.8389,
      "step": 30191
    },
    {
      "epoch": 2.597608190656457,
      "grad_norm": 1.203125,
      "learning_rate": 0.00046967890973034223,
      "loss": 1.396,
      "step": 30192
    },
    {
      "epoch": 2.5976942269637786,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004696519188057671,
      "loss": 1.6536,
      "step": 30193
    },
    {
      "epoch": 2.5977802632711002,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004696249279699544,
      "loss": 0.4744,
      "step": 30194
    },
    {
      "epoch": 2.5978662995784223,
      "grad_norm": 1.125,
      "learning_rate": 0.00046959793722298285,
      "loss": 1.9459,
      "step": 30195
    },
    {
      "epoch": 2.597952335885744,
      "grad_norm": 1.1875,
      "learning_rate": 0.00046957094656493125,
      "loss": 1.5758,
      "step": 30196
    },
    {
      "epoch": 2.5980383721930655,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0004695439559958789,
      "loss": 1.4392,
      "step": 30197
    },
    {
      "epoch": 2.598124408500387,
      "grad_norm": 1.03125,
      "learning_rate": 0.00046951696551590427,
      "loss": 1.8102,
      "step": 30198
    },
    {
      "epoch": 2.598210444807709,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00046948997512508687,
      "loss": 1.8805,
      "step": 30199
    },
    {
      "epoch": 2.5982964811150304,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004694629848235053,
      "loss": 0.9365,
      "step": 30200
    },
    {
      "epoch": 2.598382517422352,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004694359946112385,
      "loss": 1.2389,
      "step": 30201
    },
    {
      "epoch": 2.598468553729674,
      "grad_norm": 1.0,
      "learning_rate": 0.0004694090044883655,
      "loss": 0.7808,
      "step": 30202
    },
    {
      "epoch": 2.5985545900369957,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00046938201445496534,
      "loss": 1.1444,
      "step": 30203
    },
    {
      "epoch": 2.5986406263443174,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004693550245111168,
      "loss": 1.0482,
      "step": 30204
    },
    {
      "epoch": 2.598726662651639,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00046932803465689864,
      "loss": 1.5078,
      "step": 30205
    },
    {
      "epoch": 2.5988126989589606,
      "grad_norm": 1.046875,
      "learning_rate": 0.00046930104489239025,
      "loss": 1.7358,
      "step": 30206
    },
    {
      "epoch": 2.5988987352662822,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004692740552176703,
      "loss": 1.2982,
      "step": 30207
    },
    {
      "epoch": 2.598984771573604,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004692470656328177,
      "loss": 1.7817,
      "step": 30208
    },
    {
      "epoch": 2.599070807880926,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004692200761379114,
      "loss": 1.1919,
      "step": 30209
    },
    {
      "epoch": 2.5991568441882476,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00046919308673303044,
      "loss": 1.2995,
      "step": 30210
    },
    {
      "epoch": 2.599242880495569,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0004691660974182538,
      "loss": 0.7118,
      "step": 30211
    },
    {
      "epoch": 2.599328916802891,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004691391081936602,
      "loss": 0.7953,
      "step": 30212
    },
    {
      "epoch": 2.5994149531102124,
      "grad_norm": 1.125,
      "learning_rate": 0.0004691121190593288,
      "loss": 1.0722,
      "step": 30213
    },
    {
      "epoch": 2.599500989417534,
      "grad_norm": 1.890625,
      "learning_rate": 0.0004690851300153384,
      "loss": 1.5115,
      "step": 30214
    },
    {
      "epoch": 2.5995870257248557,
      "grad_norm": 1.234375,
      "learning_rate": 0.00046905814106176796,
      "loss": 1.131,
      "step": 30215
    },
    {
      "epoch": 2.5996730620321777,
      "grad_norm": 1.390625,
      "learning_rate": 0.0004690311521986964,
      "loss": 1.3031,
      "step": 30216
    },
    {
      "epoch": 2.5997590983394994,
      "grad_norm": 1.40625,
      "learning_rate": 0.00046900416342620277,
      "loss": 1.1807,
      "step": 30217
    },
    {
      "epoch": 2.599845134646821,
      "grad_norm": 1.140625,
      "learning_rate": 0.000468977174744366,
      "loss": 1.2678,
      "step": 30218
    },
    {
      "epoch": 2.5999311709541426,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004689501861532648,
      "loss": 0.6634,
      "step": 30219
    },
    {
      "epoch": 2.6000172072614642,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00046892319765297834,
      "loss": 1.3429,
      "step": 30220
    },
    {
      "epoch": 2.6001032435687863,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00046889620924358557,
      "loss": 1.5569,
      "step": 30221
    },
    {
      "epoch": 2.6001892798761075,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004688692209251653,
      "loss": 1.1491,
      "step": 30222
    },
    {
      "epoch": 2.6002753161834296,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00046884223269779633,
      "loss": 0.7809,
      "step": 30223
    },
    {
      "epoch": 2.600361352490751,
      "grad_norm": 1.03125,
      "learning_rate": 0.000468815244561558,
      "loss": 0.6181,
      "step": 30224
    },
    {
      "epoch": 2.600447388798073,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004687882565165289,
      "loss": 1.0678,
      "step": 30225
    },
    {
      "epoch": 2.6005334251053944,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00046876126856278816,
      "loss": 1.2671,
      "step": 30226
    },
    {
      "epoch": 2.600619461412716,
      "grad_norm": 1.109375,
      "learning_rate": 0.00046873428070041447,
      "loss": 0.9891,
      "step": 30227
    },
    {
      "epoch": 2.600705497720038,
      "grad_norm": 1.390625,
      "learning_rate": 0.00046870729292948705,
      "loss": 1.2806,
      "step": 30228
    },
    {
      "epoch": 2.6007915340273593,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004686803052500848,
      "loss": 1.5547,
      "step": 30229
    },
    {
      "epoch": 2.6008775703346814,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004686533176622863,
      "loss": 0.8326,
      "step": 30230
    },
    {
      "epoch": 2.600963606642003,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00046862633016617097,
      "loss": 1.2152,
      "step": 30231
    },
    {
      "epoch": 2.6010496429493246,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004685993427618175,
      "loss": 1.2178,
      "step": 30232
    },
    {
      "epoch": 2.6011356792566462,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004685723554493048,
      "loss": 0.8265,
      "step": 30233
    },
    {
      "epoch": 2.601221715563968,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004685453682287118,
      "loss": 1.7189,
      "step": 30234
    },
    {
      "epoch": 2.60130775187129,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00046851838110011735,
      "loss": 1.4376,
      "step": 30235
    },
    {
      "epoch": 2.601393788178611,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004684913940636007,
      "loss": 1.9508,
      "step": 30236
    },
    {
      "epoch": 2.601479824485933,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00046846440711924064,
      "loss": 1.8493,
      "step": 30237
    },
    {
      "epoch": 2.601565860793255,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00046843742026711607,
      "loss": 1.2127,
      "step": 30238
    },
    {
      "epoch": 2.6016518971005764,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00046841043350730573,
      "loss": 1.0264,
      "step": 30239
    },
    {
      "epoch": 2.601737933407898,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004683834468398889,
      "loss": 1.1037,
      "step": 30240
    },
    {
      "epoch": 2.6018239697152197,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0004683564602649443,
      "loss": 1.2979,
      "step": 30241
    },
    {
      "epoch": 2.6019100060225417,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004683294737825508,
      "loss": 0.711,
      "step": 30242
    },
    {
      "epoch": 2.6019960423298634,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004683024873927875,
      "loss": 1.9235,
      "step": 30243
    },
    {
      "epoch": 2.602082078637185,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004682755010957333,
      "loss": 0.9882,
      "step": 30244
    },
    {
      "epoch": 2.6021681149445066,
      "grad_norm": 1.21875,
      "learning_rate": 0.00046824851489146713,
      "loss": 1.6572,
      "step": 30245
    },
    {
      "epoch": 2.6022541512518282,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00046822152878006775,
      "loss": 1.1509,
      "step": 30246
    },
    {
      "epoch": 2.60234018755915,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00046819454276161425,
      "loss": 1.3198,
      "step": 30247
    },
    {
      "epoch": 2.6024262238664715,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00046816755683618563,
      "loss": 1.0416,
      "step": 30248
    },
    {
      "epoch": 2.6025122601737936,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00046814057100386065,
      "loss": 1.107,
      "step": 30249
    },
    {
      "epoch": 2.602598296481115,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004681135852647182,
      "loss": 1.1006,
      "step": 30250
    },
    {
      "epoch": 2.602684332788437,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00046808659961883743,
      "loss": 2.1443,
      "step": 30251
    },
    {
      "epoch": 2.6027703690957584,
      "grad_norm": 1.0,
      "learning_rate": 0.00046805961406629716,
      "loss": 1.9731,
      "step": 30252
    },
    {
      "epoch": 2.60285640540308,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00046803262860717614,
      "loss": 1.3812,
      "step": 30253
    },
    {
      "epoch": 2.6029424417104017,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004680056432415536,
      "loss": 1.7055,
      "step": 30254
    },
    {
      "epoch": 2.6030284780177233,
      "grad_norm": 1.21875,
      "learning_rate": 0.00046797865796950844,
      "loss": 1.2602,
      "step": 30255
    },
    {
      "epoch": 2.6031145143250454,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004679516727911194,
      "loss": 1.5756,
      "step": 30256
    },
    {
      "epoch": 2.603200550632367,
      "grad_norm": 0.9375,
      "learning_rate": 0.00046792468770646533,
      "loss": 1.101,
      "step": 30257
    },
    {
      "epoch": 2.6032865869396886,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004678977027156255,
      "loss": 0.6382,
      "step": 30258
    },
    {
      "epoch": 2.6033726232470102,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004678707178186786,
      "loss": 1.023,
      "step": 30259
    },
    {
      "epoch": 2.603458659554332,
      "grad_norm": 1.21875,
      "learning_rate": 0.00046784373301570343,
      "loss": 1.4224,
      "step": 30260
    },
    {
      "epoch": 2.6035446958616535,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0004678167483067793,
      "loss": 1.6829,
      "step": 30261
    },
    {
      "epoch": 2.603630732168975,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004677897636919849,
      "loss": 1.1957,
      "step": 30262
    },
    {
      "epoch": 2.603716768476297,
      "grad_norm": 1.046875,
      "learning_rate": 0.00046776277917139915,
      "loss": 1.0294,
      "step": 30263
    },
    {
      "epoch": 2.603802804783619,
      "grad_norm": 1.125,
      "learning_rate": 0.0004677357947451009,
      "loss": 1.9165,
      "step": 30264
    },
    {
      "epoch": 2.6038888410909404,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00046770881041316925,
      "loss": 0.6184,
      "step": 30265
    },
    {
      "epoch": 2.603974877398262,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004676818261756831,
      "loss": 1.0238,
      "step": 30266
    },
    {
      "epoch": 2.6040609137055837,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004676548420327212,
      "loss": 1.0733,
      "step": 30267
    },
    {
      "epoch": 2.6041469500129053,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004676278579843627,
      "loss": 1.0429,
      "step": 30268
    },
    {
      "epoch": 2.604232986320227,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004676008740306864,
      "loss": 1.4426,
      "step": 30269
    },
    {
      "epoch": 2.604319022627549,
      "grad_norm": 1.203125,
      "learning_rate": 0.00046757389017177125,
      "loss": 1.1658,
      "step": 30270
    },
    {
      "epoch": 2.6044050589348706,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00046754690640769597,
      "loss": 1.5351,
      "step": 30271
    },
    {
      "epoch": 2.6044910952421922,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004675199227385399,
      "loss": 1.2976,
      "step": 30272
    },
    {
      "epoch": 2.604577131549514,
      "grad_norm": 0.78125,
      "learning_rate": 0.00046749293916438164,
      "loss": 0.6201,
      "step": 30273
    },
    {
      "epoch": 2.6046631678568355,
      "grad_norm": 1.140625,
      "learning_rate": 0.00046746595568530024,
      "loss": 0.7226,
      "step": 30274
    },
    {
      "epoch": 2.604749204164157,
      "grad_norm": 1.046875,
      "learning_rate": 0.00046743897230137465,
      "loss": 1.1646,
      "step": 30275
    },
    {
      "epoch": 2.6048352404714787,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004674119890126837,
      "loss": 1.3369,
      "step": 30276
    },
    {
      "epoch": 2.604921276778801,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004673850058193063,
      "loss": 1.1656,
      "step": 30277
    },
    {
      "epoch": 2.6050073130861224,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00046735802272132154,
      "loss": 0.6204,
      "step": 30278
    },
    {
      "epoch": 2.605093349393444,
      "grad_norm": 1.234375,
      "learning_rate": 0.00046733103971880813,
      "loss": 1.8234,
      "step": 30279
    },
    {
      "epoch": 2.6051793857007657,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046730405681184503,
      "loss": 1.1958,
      "step": 30280
    },
    {
      "epoch": 2.6052654220080873,
      "grad_norm": 1.546875,
      "learning_rate": 0.0004672770740005113,
      "loss": 1.0853,
      "step": 30281
    },
    {
      "epoch": 2.6053514583154094,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00046725009128488574,
      "loss": 0.7844,
      "step": 30282
    },
    {
      "epoch": 2.6054374946227306,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004672231086650472,
      "loss": 1.7644,
      "step": 30283
    },
    {
      "epoch": 2.6055235309300526,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00046719612614107476,
      "loss": 0.7834,
      "step": 30284
    },
    {
      "epoch": 2.6056095672373742,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004671691437130473,
      "loss": 0.9291,
      "step": 30285
    },
    {
      "epoch": 2.605695603544696,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004671421613810437,
      "loss": 1.2588,
      "step": 30286
    },
    {
      "epoch": 2.6057816398520175,
      "grad_norm": 0.921875,
      "learning_rate": 0.00046711517914514283,
      "loss": 1.1574,
      "step": 30287
    },
    {
      "epoch": 2.605867676159339,
      "grad_norm": 1.125,
      "learning_rate": 0.0004670881970054237,
      "loss": 1.134,
      "step": 30288
    },
    {
      "epoch": 2.605953712466661,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004670612149619653,
      "loss": 0.8552,
      "step": 30289
    },
    {
      "epoch": 2.6060397487739824,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004670342330148462,
      "loss": 1.1326,
      "step": 30290
    },
    {
      "epoch": 2.6061257850813044,
      "grad_norm": 1.171875,
      "learning_rate": 0.00046700725116414577,
      "loss": 1.3407,
      "step": 30291
    },
    {
      "epoch": 2.606211821388626,
      "grad_norm": 1.015625,
      "learning_rate": 0.00046698026940994263,
      "loss": 1.218,
      "step": 30292
    },
    {
      "epoch": 2.6062978576959477,
      "grad_norm": 0.921875,
      "learning_rate": 0.00046695328775231583,
      "loss": 1.3307,
      "step": 30293
    },
    {
      "epoch": 2.6063838940032693,
      "grad_norm": 1.1875,
      "learning_rate": 0.00046692630619134414,
      "loss": 0.8511,
      "step": 30294
    },
    {
      "epoch": 2.606469930310591,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046689932472710664,
      "loss": 1.0293,
      "step": 30295
    },
    {
      "epoch": 2.606555966617913,
      "grad_norm": 0.640625,
      "learning_rate": 0.0004668723433596822,
      "loss": 0.8095,
      "step": 30296
    },
    {
      "epoch": 2.606642002925234,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00046684536208914975,
      "loss": 0.8129,
      "step": 30297
    },
    {
      "epoch": 2.6067280392325562,
      "grad_norm": 1.09375,
      "learning_rate": 0.000466818380915588,
      "loss": 0.6825,
      "step": 30298
    },
    {
      "epoch": 2.606814075539878,
      "grad_norm": 1.515625,
      "learning_rate": 0.00046679139983907617,
      "loss": 0.8167,
      "step": 30299
    },
    {
      "epoch": 2.6069001118471995,
      "grad_norm": 1.03125,
      "learning_rate": 0.000466764418859693,
      "loss": 1.6859,
      "step": 30300
    },
    {
      "epoch": 2.606986148154521,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00046673743797751737,
      "loss": 1.4751,
      "step": 30301
    },
    {
      "epoch": 2.6070721844618427,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004667104571926284,
      "loss": 1.6639,
      "step": 30302
    },
    {
      "epoch": 2.607158220769165,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00046668347650510476,
      "loss": 1.0485,
      "step": 30303
    },
    {
      "epoch": 2.6072442570764864,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004666564959150256,
      "loss": 1.6269,
      "step": 30304
    },
    {
      "epoch": 2.607330293383808,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004666295154224695,
      "loss": 0.9394,
      "step": 30305
    },
    {
      "epoch": 2.6074163296911297,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004666025350275157,
      "loss": 1.2461,
      "step": 30306
    },
    {
      "epoch": 2.6075023659984513,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000466575554730243,
      "loss": 1.2414,
      "step": 30307
    },
    {
      "epoch": 2.607588402305773,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004665485745307301,
      "loss": 1.6629,
      "step": 30308
    },
    {
      "epoch": 2.6076744386130946,
      "grad_norm": 1.65625,
      "learning_rate": 0.0004665215944290564,
      "loss": 1.3357,
      "step": 30309
    },
    {
      "epoch": 2.6077604749204166,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004664946144253003,
      "loss": 1.2802,
      "step": 30310
    },
    {
      "epoch": 2.6078465112277383,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004664676345195411,
      "loss": 1.0503,
      "step": 30311
    },
    {
      "epoch": 2.60793254753506,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004664406547118575,
      "loss": 1.0958,
      "step": 30312
    },
    {
      "epoch": 2.6080185838423815,
      "grad_norm": 1.0,
      "learning_rate": 0.00046641367500232855,
      "loss": 0.9506,
      "step": 30313
    },
    {
      "epoch": 2.608104620149703,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004663866953910329,
      "loss": 0.7527,
      "step": 30314
    },
    {
      "epoch": 2.6081906564570247,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004663597158780497,
      "loss": 1.8827,
      "step": 30315
    },
    {
      "epoch": 2.6082766927643464,
      "grad_norm": 0.890625,
      "learning_rate": 0.00046633273646345784,
      "loss": 1.1027,
      "step": 30316
    },
    {
      "epoch": 2.6083627290716684,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00046630575714733607,
      "loss": 1.5216,
      "step": 30317
    },
    {
      "epoch": 2.60844876537899,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004662787779297635,
      "loss": 1.1014,
      "step": 30318
    },
    {
      "epoch": 2.6085348016863117,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00046625179881081897,
      "loss": 0.7349,
      "step": 30319
    },
    {
      "epoch": 2.6086208379936333,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00046622481979058134,
      "loss": 1.1546,
      "step": 30320
    },
    {
      "epoch": 2.608706874300955,
      "grad_norm": 1.0,
      "learning_rate": 0.00046619784086912943,
      "loss": 0.9454,
      "step": 30321
    },
    {
      "epoch": 2.6087929106082766,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004661708620465424,
      "loss": 1.0909,
      "step": 30322
    },
    {
      "epoch": 2.608878946915598,
      "grad_norm": 1.03125,
      "learning_rate": 0.000466143883322899,
      "loss": 0.938,
      "step": 30323
    },
    {
      "epoch": 2.6089649832229203,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004661169046982781,
      "loss": 1.1209,
      "step": 30324
    },
    {
      "epoch": 2.609051019530242,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004660899261727587,
      "loss": 1.6477,
      "step": 30325
    },
    {
      "epoch": 2.6091370558375635,
      "grad_norm": 0.90625,
      "learning_rate": 0.00046606294774641964,
      "loss": 1.2761,
      "step": 30326
    },
    {
      "epoch": 2.609223092144885,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046603596941933995,
      "loss": 1.098,
      "step": 30327
    },
    {
      "epoch": 2.6093091284522067,
      "grad_norm": 0.921875,
      "learning_rate": 0.00046600899119159827,
      "loss": 1.3016,
      "step": 30328
    },
    {
      "epoch": 2.6093951647595284,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004659820130632738,
      "loss": 0.9162,
      "step": 30329
    },
    {
      "epoch": 2.60948120106685,
      "grad_norm": 1.078125,
      "learning_rate": 0.00046595503503444534,
      "loss": 1.6371,
      "step": 30330
    },
    {
      "epoch": 2.609567237374172,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004659280571051917,
      "loss": 1.3368,
      "step": 30331
    },
    {
      "epoch": 2.6096532736814937,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00046590107927559196,
      "loss": 1.0273,
      "step": 30332
    },
    {
      "epoch": 2.6097393099888153,
      "grad_norm": 1.25,
      "learning_rate": 0.0004658741015457249,
      "loss": 1.6694,
      "step": 30333
    },
    {
      "epoch": 2.609825346296137,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004658471239156695,
      "loss": 1.1215,
      "step": 30334
    },
    {
      "epoch": 2.6099113826034586,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00046582014638550445,
      "loss": 1.419,
      "step": 30335
    },
    {
      "epoch": 2.60999741891078,
      "grad_norm": 1.125,
      "learning_rate": 0.000465793168955309,
      "loss": 1.3141,
      "step": 30336
    },
    {
      "epoch": 2.610083455218102,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004657661916251618,
      "loss": 1.42,
      "step": 30337
    },
    {
      "epoch": 2.610169491525424,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0004657392143951419,
      "loss": 0.6592,
      "step": 30338
    },
    {
      "epoch": 2.6102555278327455,
      "grad_norm": 1.234375,
      "learning_rate": 0.00046571223726532795,
      "loss": 1.3324,
      "step": 30339
    },
    {
      "epoch": 2.610341564140067,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00046568526023579924,
      "loss": 1.5662,
      "step": 30340
    },
    {
      "epoch": 2.6104276004473888,
      "grad_norm": 1.59375,
      "learning_rate": 0.0004656582833066344,
      "loss": 1.9407,
      "step": 30341
    },
    {
      "epoch": 2.6105136367547104,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00046563130647791237,
      "loss": 0.7929,
      "step": 30342
    },
    {
      "epoch": 2.6105996730620324,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004656043297497121,
      "loss": 0.6812,
      "step": 30343
    },
    {
      "epoch": 2.6106857093693536,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004655773531221125,
      "loss": 1.0925,
      "step": 30344
    },
    {
      "epoch": 2.6107717456766757,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00046555037659519227,
      "loss": 1.2663,
      "step": 30345
    },
    {
      "epoch": 2.6108577819839973,
      "grad_norm": 1.140625,
      "learning_rate": 0.00046552340016903075,
      "loss": 1.7223,
      "step": 30346
    },
    {
      "epoch": 2.610943818291319,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00046549642384370636,
      "loss": 1.5754,
      "step": 30347
    },
    {
      "epoch": 2.6110298545986406,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004654694476192984,
      "loss": 0.929,
      "step": 30348
    },
    {
      "epoch": 2.611115890905962,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004654424714958855,
      "loss": 1.3396,
      "step": 30349
    },
    {
      "epoch": 2.6112019272132843,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004654154954735468,
      "loss": 1.1697,
      "step": 30350
    },
    {
      "epoch": 2.6112879635206054,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00046538851955236083,
      "loss": 1.4272,
      "step": 30351
    },
    {
      "epoch": 2.6113739998279275,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004653615437324068,
      "loss": 0.9994,
      "step": 30352
    },
    {
      "epoch": 2.611460036135249,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004653345680137636,
      "loss": 1.1618,
      "step": 30353
    },
    {
      "epoch": 2.6115460724425708,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00046530759239650987,
      "loss": 0.9446,
      "step": 30354
    },
    {
      "epoch": 2.6116321087498924,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004652806168807249,
      "loss": 2.1157,
      "step": 30355
    },
    {
      "epoch": 2.611718145057214,
      "grad_norm": 0.890625,
      "learning_rate": 0.00046525364146648724,
      "loss": 0.8344,
      "step": 30356
    },
    {
      "epoch": 2.611804181364536,
      "grad_norm": 1.078125,
      "learning_rate": 0.000465226666153876,
      "loss": 0.899,
      "step": 30357
    },
    {
      "epoch": 2.6118902176718573,
      "grad_norm": 1.0,
      "learning_rate": 0.00046519969094296985,
      "loss": 1.0936,
      "step": 30358
    },
    {
      "epoch": 2.6119762539791793,
      "grad_norm": 1.015625,
      "learning_rate": 0.00046517271583384795,
      "loss": 1.2813,
      "step": 30359
    },
    {
      "epoch": 2.612062290286501,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004651457408265891,
      "loss": 1.0289,
      "step": 30360
    },
    {
      "epoch": 2.6121483265938226,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004651187659212722,
      "loss": 0.9625,
      "step": 30361
    },
    {
      "epoch": 2.612234362901144,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000465091791117976,
      "loss": 1.3803,
      "step": 30362
    },
    {
      "epoch": 2.612320399208466,
      "grad_norm": 0.890625,
      "learning_rate": 0.00046506481641677956,
      "loss": 1.5971,
      "step": 30363
    },
    {
      "epoch": 2.612406435515788,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00046503784181776183,
      "loss": 1.1487,
      "step": 30364
    },
    {
      "epoch": 2.6124924718231095,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00046501086732100144,
      "loss": 1.4911,
      "step": 30365
    },
    {
      "epoch": 2.612578508130431,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004649838929265776,
      "loss": 1.823,
      "step": 30366
    },
    {
      "epoch": 2.6126645444377528,
      "grad_norm": 1.15625,
      "learning_rate": 0.00046495691863456906,
      "loss": 1.5943,
      "step": 30367
    },
    {
      "epoch": 2.6127505807450744,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00046492994444505476,
      "loss": 1.7381,
      "step": 30368
    },
    {
      "epoch": 2.612836617052396,
      "grad_norm": 0.8125,
      "learning_rate": 0.00046490297035811336,
      "loss": 0.641,
      "step": 30369
    },
    {
      "epoch": 2.6129226533597176,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004648759963738241,
      "loss": 0.9831,
      "step": 30370
    },
    {
      "epoch": 2.6130086896670397,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00046484902249226573,
      "loss": 0.8785,
      "step": 30371
    },
    {
      "epoch": 2.6130947259743613,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00046482204871351704,
      "loss": 0.8678,
      "step": 30372
    },
    {
      "epoch": 2.613180762281683,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004647950750376571,
      "loss": 0.9553,
      "step": 30373
    },
    {
      "epoch": 2.6132667985890046,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00046476810146476466,
      "loss": 1.4609,
      "step": 30374
    },
    {
      "epoch": 2.613352834896326,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00046474112799491877,
      "loss": 1.0501,
      "step": 30375
    },
    {
      "epoch": 2.613438871203648,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004647141546281981,
      "loss": 0.8509,
      "step": 30376
    },
    {
      "epoch": 2.6135249075109694,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0004646871813646817,
      "loss": 1.384,
      "step": 30377
    },
    {
      "epoch": 2.6136109438182915,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004646602082044484,
      "loss": 1.1319,
      "step": 30378
    },
    {
      "epoch": 2.613696980125613,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00046463323514757724,
      "loss": 1.7338,
      "step": 30379
    },
    {
      "epoch": 2.6137830164329348,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004646062621941468,
      "loss": 1.1627,
      "step": 30380
    },
    {
      "epoch": 2.6138690527402564,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004645792893442363,
      "loss": 1.6593,
      "step": 30381
    },
    {
      "epoch": 2.613955089047578,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004645523165979243,
      "loss": 1.5471,
      "step": 30382
    },
    {
      "epoch": 2.6140411253548996,
      "grad_norm": 1.1875,
      "learning_rate": 0.00046452534395529006,
      "loss": 1.2983,
      "step": 30383
    },
    {
      "epoch": 2.6141271616622213,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00046449837141641226,
      "loss": 0.8213,
      "step": 30384
    },
    {
      "epoch": 2.6142131979695433,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00046447139898136974,
      "loss": 2.103,
      "step": 30385
    },
    {
      "epoch": 2.614299234276865,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004644444266502416,
      "loss": 2.0276,
      "step": 30386
    },
    {
      "epoch": 2.6143852705841866,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004644174544231066,
      "loss": 1.2209,
      "step": 30387
    },
    {
      "epoch": 2.614471306891508,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004643904823000435,
      "loss": 1.5097,
      "step": 30388
    },
    {
      "epoch": 2.61455734319883,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0004643635102811314,
      "loss": 0.9639,
      "step": 30389
    },
    {
      "epoch": 2.6146433795061514,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004643365383664491,
      "loss": 0.6847,
      "step": 30390
    },
    {
      "epoch": 2.614729415813473,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00046430956655607553,
      "loss": 1.252,
      "step": 30391
    },
    {
      "epoch": 2.614815452120795,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004642825948500894,
      "loss": 1.9339,
      "step": 30392
    },
    {
      "epoch": 2.6149014884281168,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00046425562324856984,
      "loss": 0.8152,
      "step": 30393
    },
    {
      "epoch": 2.6149875247354384,
      "grad_norm": 1.03125,
      "learning_rate": 0.00046422865175159563,
      "loss": 1.5241,
      "step": 30394
    },
    {
      "epoch": 2.61507356104276,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00046420168035924553,
      "loss": 0.8132,
      "step": 30395
    },
    {
      "epoch": 2.6151595973500816,
      "grad_norm": 1.125,
      "learning_rate": 0.0004641747090715987,
      "loss": 1.3571,
      "step": 30396
    },
    {
      "epoch": 2.6152456336574037,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004641477378887339,
      "loss": 0.8182,
      "step": 30397
    },
    {
      "epoch": 2.615331669964725,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00046412076681073,
      "loss": 1.6155,
      "step": 30398
    },
    {
      "epoch": 2.615417706272047,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00046409379583766563,
      "loss": 0.5958,
      "step": 30399
    },
    {
      "epoch": 2.6155037425793686,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00046406682496962014,
      "loss": 1.2654,
      "step": 30400
    },
    {
      "epoch": 2.61558977888669,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004640398542066722,
      "loss": 1.1207,
      "step": 30401
    },
    {
      "epoch": 2.615675815194012,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004640128835489007,
      "loss": 1.1113,
      "step": 30402
    },
    {
      "epoch": 2.6157618515013334,
      "grad_norm": 1.03125,
      "learning_rate": 0.00046398591299638437,
      "loss": 1.9582,
      "step": 30403
    },
    {
      "epoch": 2.6158478878086555,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004639589425492024,
      "loss": 1.085,
      "step": 30404
    },
    {
      "epoch": 2.6159339241159767,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00046393197220743343,
      "loss": 0.8419,
      "step": 30405
    },
    {
      "epoch": 2.6160199604232988,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004639050019711564,
      "loss": 1.7471,
      "step": 30406
    },
    {
      "epoch": 2.6161059967306204,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004638780318404503,
      "loss": 1.3129,
      "step": 30407
    },
    {
      "epoch": 2.616192033037942,
      "grad_norm": 1.09375,
      "learning_rate": 0.00046385106181539394,
      "loss": 1.4239,
      "step": 30408
    },
    {
      "epoch": 2.6162780693452636,
      "grad_norm": 1.21875,
      "learning_rate": 0.00046382409189606617,
      "loss": 1.5465,
      "step": 30409
    },
    {
      "epoch": 2.6163641056525853,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00046379712208254585,
      "loss": 0.808,
      "step": 30410
    },
    {
      "epoch": 2.6164501419599073,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004637701523749119,
      "loss": 1.6195,
      "step": 30411
    },
    {
      "epoch": 2.6165361782672285,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004637431827732433,
      "loss": 1.5564,
      "step": 30412
    },
    {
      "epoch": 2.6166222145745506,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004637162132776187,
      "loss": 1.3121,
      "step": 30413
    },
    {
      "epoch": 2.616708250881872,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004636892438881172,
      "loss": 1.0422,
      "step": 30414
    },
    {
      "epoch": 2.616794287189194,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004636622746048176,
      "loss": 1.173,
      "step": 30415
    },
    {
      "epoch": 2.6168803234965154,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004636353054277989,
      "loss": 1.5431,
      "step": 30416
    },
    {
      "epoch": 2.616966359803837,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00046360833635713963,
      "loss": 0.561,
      "step": 30417
    },
    {
      "epoch": 2.617052396111159,
      "grad_norm": 1.5703125,
      "learning_rate": 0.000463581367392919,
      "loss": 1.0827,
      "step": 30418
    },
    {
      "epoch": 2.6171384324184808,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00046355439853521565,
      "loss": 1.9305,
      "step": 30419
    },
    {
      "epoch": 2.6172244687258024,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004635274297841088,
      "loss": 0.9434,
      "step": 30420
    },
    {
      "epoch": 2.617310505033124,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00046350046113967705,
      "loss": 1.8228,
      "step": 30421
    },
    {
      "epoch": 2.6173965413404456,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004634734926019993,
      "loss": 1.3469,
      "step": 30422
    },
    {
      "epoch": 2.6174825776477673,
      "grad_norm": 1.265625,
      "learning_rate": 0.00046344652417115457,
      "loss": 0.9547,
      "step": 30423
    },
    {
      "epoch": 2.617568613955089,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004634195558472217,
      "loss": 0.985,
      "step": 30424
    },
    {
      "epoch": 2.617654650262411,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0004633925876302794,
      "loss": 2.1156,
      "step": 30425
    },
    {
      "epoch": 2.6177406865697326,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00046336561952040663,
      "loss": 1.0284,
      "step": 30426
    },
    {
      "epoch": 2.617826722877054,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004633386515176824,
      "loss": 1.1186,
      "step": 30427
    },
    {
      "epoch": 2.617912759184376,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004633116836221855,
      "loss": 1.5762,
      "step": 30428
    },
    {
      "epoch": 2.6179987954916975,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004632847158339947,
      "loss": 1.884,
      "step": 30429
    },
    {
      "epoch": 2.618084831799019,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000463257748153189,
      "loss": 1.5706,
      "step": 30430
    },
    {
      "epoch": 2.6181708681063407,
      "grad_norm": 1.453125,
      "learning_rate": 0.0004632307805798473,
      "loss": 1.9401,
      "step": 30431
    },
    {
      "epoch": 2.6182569044136628,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00046320381311404835,
      "loss": 1.7517,
      "step": 30432
    },
    {
      "epoch": 2.6183429407209844,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00046317684575587103,
      "loss": 2.1371,
      "step": 30433
    },
    {
      "epoch": 2.618428977028306,
      "grad_norm": 1.0,
      "learning_rate": 0.0004631498785053944,
      "loss": 0.6877,
      "step": 30434
    },
    {
      "epoch": 2.6185150133356276,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00046312291136269717,
      "loss": 0.6338,
      "step": 30435
    },
    {
      "epoch": 2.6186010496429493,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00046309594432785816,
      "loss": 1.4297,
      "step": 30436
    },
    {
      "epoch": 2.618687085950271,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004630689774009564,
      "loss": 1.4369,
      "step": 30437
    },
    {
      "epoch": 2.6187731222575925,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004630420105820708,
      "loss": 0.9259,
      "step": 30438
    },
    {
      "epoch": 2.6188591585649146,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004630150438712801,
      "loss": 1.8488,
      "step": 30439
    },
    {
      "epoch": 2.618945194872236,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046298807726866305,
      "loss": 0.6818,
      "step": 30440
    },
    {
      "epoch": 2.619031231179558,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004629611107742988,
      "loss": 1.0147,
      "step": 30441
    },
    {
      "epoch": 2.6191172674868795,
      "grad_norm": 1.359375,
      "learning_rate": 0.00046293414438826607,
      "loss": 1.6713,
      "step": 30442
    },
    {
      "epoch": 2.619203303794201,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004629071781106437,
      "loss": 1.3889,
      "step": 30443
    },
    {
      "epoch": 2.6192893401015227,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00046288021194151077,
      "loss": 1.4081,
      "step": 30444
    },
    {
      "epoch": 2.6193753764088443,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004628532458809459,
      "loss": 0.8994,
      "step": 30445
    },
    {
      "epoch": 2.6194614127161664,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004628262799290281,
      "loss": 1.6162,
      "step": 30446
    },
    {
      "epoch": 2.619547449023488,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004627993140858361,
      "loss": 1.5885,
      "step": 30447
    },
    {
      "epoch": 2.6196334853308096,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000462772348351449,
      "loss": 0.9273,
      "step": 30448
    },
    {
      "epoch": 2.6197195216381313,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004627453827259455,
      "loss": 1.4337,
      "step": 30449
    },
    {
      "epoch": 2.619805557945453,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004627184172094045,
      "loss": 0.9127,
      "step": 30450
    },
    {
      "epoch": 2.6198915942527745,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004626914518019048,
      "loss": 1.2776,
      "step": 30451
    },
    {
      "epoch": 2.619977630560096,
      "grad_norm": 1.15625,
      "learning_rate": 0.00046266448650352545,
      "loss": 0.9185,
      "step": 30452
    },
    {
      "epoch": 2.620063666867418,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004626375213143452,
      "loss": 1.0762,
      "step": 30453
    },
    {
      "epoch": 2.62014970317474,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00046261055623444285,
      "loss": 1.3945,
      "step": 30454
    },
    {
      "epoch": 2.6202357394820615,
      "grad_norm": 1.046875,
      "learning_rate": 0.00046258359126389746,
      "loss": 1.4488,
      "step": 30455
    },
    {
      "epoch": 2.620321775789383,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004625566264027876,
      "loss": 1.0095,
      "step": 30456
    },
    {
      "epoch": 2.6204078120967047,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004625296616511925,
      "loss": 0.9857,
      "step": 30457
    },
    {
      "epoch": 2.6204938484040268,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00046250269700919087,
      "loss": 1.9081,
      "step": 30458
    },
    {
      "epoch": 2.620579884711348,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00046247573247686134,
      "loss": 1.7009,
      "step": 30459
    },
    {
      "epoch": 2.62066592101867,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004624487680542833,
      "loss": 1.6953,
      "step": 30460
    },
    {
      "epoch": 2.6207519573259916,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046242180374153516,
      "loss": 1.0796,
      "step": 30461
    },
    {
      "epoch": 2.6208379936333133,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000462394839538696,
      "loss": 1.3052,
      "step": 30462
    },
    {
      "epoch": 2.620924029940635,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004623678754458445,
      "loss": 1.5241,
      "step": 30463
    },
    {
      "epoch": 2.6210100662479565,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00046234091146305975,
      "loss": 1.0667,
      "step": 30464
    },
    {
      "epoch": 2.6210961025552786,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004623139475904205,
      "loss": 1.3128,
      "step": 30465
    },
    {
      "epoch": 2.6211821388625998,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00046228698382800557,
      "loss": 1.2994,
      "step": 30466
    },
    {
      "epoch": 2.621268175169922,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0004622600201758939,
      "loss": 0.4804,
      "step": 30467
    },
    {
      "epoch": 2.6213542114772435,
      "grad_norm": 0.8125,
      "learning_rate": 0.00046223305663416436,
      "loss": 0.6783,
      "step": 30468
    },
    {
      "epoch": 2.621440247784565,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004622060932028958,
      "loss": 1.4976,
      "step": 30469
    },
    {
      "epoch": 2.6215262840918867,
      "grad_norm": 1.046875,
      "learning_rate": 0.000462179129882167,
      "loss": 1.0598,
      "step": 30470
    },
    {
      "epoch": 2.6216123203992083,
      "grad_norm": 0.890625,
      "learning_rate": 0.00046215216667205687,
      "loss": 1.4873,
      "step": 30471
    },
    {
      "epoch": 2.6216983567065304,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00046212520357264444,
      "loss": 1.0437,
      "step": 30472
    },
    {
      "epoch": 2.6217843930138516,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00046209824058400837,
      "loss": 1.1541,
      "step": 30473
    },
    {
      "epoch": 2.6218704293211736,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004620712777062274,
      "loss": 0.9967,
      "step": 30474
    },
    {
      "epoch": 2.6219564656284953,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004620443149393808,
      "loss": 1.3749,
      "step": 30475
    },
    {
      "epoch": 2.622042501935817,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00046201735228354707,
      "loss": 1.6209,
      "step": 30476
    },
    {
      "epoch": 2.6221285382431385,
      "grad_norm": 0.96875,
      "learning_rate": 0.00046199038973880515,
      "loss": 0.7866,
      "step": 30477
    },
    {
      "epoch": 2.62221457455046,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004619634273052341,
      "loss": 0.9438,
      "step": 30478
    },
    {
      "epoch": 2.622300610857782,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00046193646498291256,
      "loss": 0.4974,
      "step": 30479
    },
    {
      "epoch": 2.622386647165104,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00046190950277191946,
      "loss": 0.6248,
      "step": 30480
    },
    {
      "epoch": 2.6224726834724255,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00046188254067233356,
      "loss": 1.0604,
      "step": 30481
    },
    {
      "epoch": 2.622558719779747,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046185557868423394,
      "loss": 0.8037,
      "step": 30482
    },
    {
      "epoch": 2.6226447560870687,
      "grad_norm": 1.25,
      "learning_rate": 0.0004618286168076993,
      "loss": 1.6332,
      "step": 30483
    },
    {
      "epoch": 2.6227307923943903,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0004618016550428085,
      "loss": 1.0663,
      "step": 30484
    },
    {
      "epoch": 2.622816828701712,
      "grad_norm": 1.265625,
      "learning_rate": 0.00046177469338964044,
      "loss": 1.918,
      "step": 30485
    },
    {
      "epoch": 2.622902865009034,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000461747731848274,
      "loss": 1.3099,
      "step": 30486
    },
    {
      "epoch": 2.6229889013163556,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000461720770418788,
      "loss": 0.7793,
      "step": 30487
    },
    {
      "epoch": 2.6230749376236773,
      "grad_norm": 1.21875,
      "learning_rate": 0.00046169380910126125,
      "loss": 0.8638,
      "step": 30488
    },
    {
      "epoch": 2.623160973930999,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004616668478957727,
      "loss": 1.046,
      "step": 30489
    },
    {
      "epoch": 2.6232470102383205,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004616398868024012,
      "loss": 1.3986,
      "step": 30490
    },
    {
      "epoch": 2.623333046545642,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00046161292582122554,
      "loss": 1.1498,
      "step": 30491
    },
    {
      "epoch": 2.6234190828529638,
      "grad_norm": 1.8125,
      "learning_rate": 0.0004615859649523245,
      "loss": 0.9025,
      "step": 30492
    },
    {
      "epoch": 2.623505119160286,
      "grad_norm": 1.046875,
      "learning_rate": 0.000461559004195777,
      "loss": 1.7181,
      "step": 30493
    },
    {
      "epoch": 2.6235911554676075,
      "grad_norm": 1.046875,
      "learning_rate": 0.00046153204355166207,
      "loss": 0.5036,
      "step": 30494
    },
    {
      "epoch": 2.623677191774929,
      "grad_norm": 1.265625,
      "learning_rate": 0.00046150508302005846,
      "loss": 1.6536,
      "step": 30495
    },
    {
      "epoch": 2.6237632280822507,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000461478122601045,
      "loss": 1.3298,
      "step": 30496
    },
    {
      "epoch": 2.6238492643895723,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004614511622947004,
      "loss": 0.7349,
      "step": 30497
    },
    {
      "epoch": 2.623935300696894,
      "grad_norm": 1.515625,
      "learning_rate": 0.0004614242021011038,
      "loss": 1.8902,
      "step": 30498
    },
    {
      "epoch": 2.6240213370042156,
      "grad_norm": 1.34375,
      "learning_rate": 0.00046139724202033395,
      "loss": 1.5685,
      "step": 30499
    },
    {
      "epoch": 2.6241073733115376,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00046137028205246944,
      "loss": 1.3895,
      "step": 30500
    },
    {
      "epoch": 2.6241934096188593,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004613433221975896,
      "loss": 0.8211,
      "step": 30501
    },
    {
      "epoch": 2.624279445926181,
      "grad_norm": 1.0625,
      "learning_rate": 0.00046131636245577283,
      "loss": 0.8093,
      "step": 30502
    },
    {
      "epoch": 2.6243654822335025,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004612894028270983,
      "loss": 1.2038,
      "step": 30503
    },
    {
      "epoch": 2.624451518540824,
      "grad_norm": 0.9375,
      "learning_rate": 0.00046126244331164465,
      "loss": 1.728,
      "step": 30504
    },
    {
      "epoch": 2.6245375548481458,
      "grad_norm": 1.09375,
      "learning_rate": 0.00046123548390949087,
      "loss": 1.4534,
      "step": 30505
    },
    {
      "epoch": 2.6246235911554674,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0004612085246207158,
      "loss": 0.8809,
      "step": 30506
    },
    {
      "epoch": 2.6247096274627895,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004611815654453981,
      "loss": 1.5988,
      "step": 30507
    },
    {
      "epoch": 2.624795663770111,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004611546063836169,
      "loss": 0.6597,
      "step": 30508
    },
    {
      "epoch": 2.6248817000774327,
      "grad_norm": 1.453125,
      "learning_rate": 0.00046112764743545093,
      "loss": 2.2351,
      "step": 30509
    },
    {
      "epoch": 2.6249677363847543,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000461100688600979,
      "loss": 1.604,
      "step": 30510
    },
    {
      "epoch": 2.625053772692076,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004610737298802799,
      "loss": 0.7735,
      "step": 30511
    },
    {
      "epoch": 2.6251398089993976,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004610467712734327,
      "loss": 0.6942,
      "step": 30512
    },
    {
      "epoch": 2.625225845306719,
      "grad_norm": 1.734375,
      "learning_rate": 0.0004610198127805161,
      "loss": 0.9215,
      "step": 30513
    },
    {
      "epoch": 2.6253118816140413,
      "grad_norm": 1.09375,
      "learning_rate": 0.000460992854401609,
      "loss": 1.0274,
      "step": 30514
    },
    {
      "epoch": 2.625397917921363,
      "grad_norm": 1.015625,
      "learning_rate": 0.00046096589613679,
      "loss": 1.2716,
      "step": 30515
    },
    {
      "epoch": 2.6254839542286845,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004609389379861384,
      "loss": 1.5678,
      "step": 30516
    },
    {
      "epoch": 2.625569990536006,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00046091197994973273,
      "loss": 0.518,
      "step": 30517
    },
    {
      "epoch": 2.6256560268433278,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004608850220276518,
      "loss": 1.2214,
      "step": 30518
    },
    {
      "epoch": 2.62574206315065,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00046085806421997474,
      "loss": 1.2778,
      "step": 30519
    },
    {
      "epoch": 2.625828099457971,
      "grad_norm": 0.75,
      "learning_rate": 0.00046083110652678016,
      "loss": 1.4362,
      "step": 30520
    },
    {
      "epoch": 2.625914135765293,
      "grad_norm": 1.171875,
      "learning_rate": 0.00046080414894814704,
      "loss": 1.561,
      "step": 30521
    },
    {
      "epoch": 2.6260001720726147,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00046077719148415405,
      "loss": 1.2669,
      "step": 30522
    },
    {
      "epoch": 2.6260862083799363,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004607502341348802,
      "loss": 1.3512,
      "step": 30523
    },
    {
      "epoch": 2.626172244687258,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004607232769004043,
      "loss": 1.5838,
      "step": 30524
    },
    {
      "epoch": 2.6262582809945796,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004606963197808051,
      "loss": 1.5902,
      "step": 30525
    },
    {
      "epoch": 2.6263443173019017,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004606693627761617,
      "loss": 1.0361,
      "step": 30526
    },
    {
      "epoch": 2.626430353609223,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004606424058865526,
      "loss": 0.7208,
      "step": 30527
    },
    {
      "epoch": 2.626516389916545,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004606154491120569,
      "loss": 1.2734,
      "step": 30528
    },
    {
      "epoch": 2.6266024262238665,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004605884924527533,
      "loss": 1.1852,
      "step": 30529
    },
    {
      "epoch": 2.626688462531188,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004605615359087204,
      "loss": 1.1988,
      "step": 30530
    },
    {
      "epoch": 2.6267744988385098,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004605345794800377,
      "loss": 1.8766,
      "step": 30531
    },
    {
      "epoch": 2.6268605351458314,
      "grad_norm": 1.453125,
      "learning_rate": 0.0004605076231667836,
      "loss": 1.2824,
      "step": 30532
    },
    {
      "epoch": 2.6269465714531535,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004604806669690371,
      "loss": 0.8617,
      "step": 30533
    },
    {
      "epoch": 2.6270326077604746,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004604537108868768,
      "loss": 0.7818,
      "step": 30534
    },
    {
      "epoch": 2.6271186440677967,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004604267549203819,
      "loss": 1.7424,
      "step": 30535
    },
    {
      "epoch": 2.6272046803751183,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000460399799069631,
      "loss": 1.0472,
      "step": 30536
    },
    {
      "epoch": 2.62729071668244,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004603728433347029,
      "loss": 1.2361,
      "step": 30537
    },
    {
      "epoch": 2.6273767529897616,
      "grad_norm": 1.125,
      "learning_rate": 0.00046034588771567655,
      "loss": 1.859,
      "step": 30538
    },
    {
      "epoch": 2.627462789297083,
      "grad_norm": 1.328125,
      "learning_rate": 0.00046031893221263074,
      "loss": 1.6676,
      "step": 30539
    },
    {
      "epoch": 2.6275488256044053,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00046029197682564445,
      "loss": 1.0064,
      "step": 30540
    },
    {
      "epoch": 2.627634861911727,
      "grad_norm": 1.703125,
      "learning_rate": 0.0004602650215547962,
      "loss": 1.9994,
      "step": 30541
    },
    {
      "epoch": 2.6277208982190485,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004602380664001652,
      "loss": 1.4681,
      "step": 30542
    },
    {
      "epoch": 2.62780693452637,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00046021111136183016,
      "loss": 1.5762,
      "step": 30543
    },
    {
      "epoch": 2.6278929708336918,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00046018415643986983,
      "loss": 2.0196,
      "step": 30544
    },
    {
      "epoch": 2.6279790071410134,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046015720163436295,
      "loss": 1.154,
      "step": 30545
    },
    {
      "epoch": 2.628065043448335,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004601302469453887,
      "loss": 1.0817,
      "step": 30546
    },
    {
      "epoch": 2.628151079755657,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004601032923730257,
      "loss": 1.6827,
      "step": 30547
    },
    {
      "epoch": 2.6282371160629787,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004600763379173527,
      "loss": 1.5615,
      "step": 30548
    },
    {
      "epoch": 2.6283231523703003,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004600493835784487,
      "loss": 0.8931,
      "step": 30549
    },
    {
      "epoch": 2.628409188677622,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00046002242935639254,
      "loss": 0.9521,
      "step": 30550
    },
    {
      "epoch": 2.6284952249849436,
      "grad_norm": 0.953125,
      "learning_rate": 0.00045999547525126303,
      "loss": 1.029,
      "step": 30551
    },
    {
      "epoch": 2.628581261292265,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004599685212631388,
      "loss": 0.8708,
      "step": 30552
    },
    {
      "epoch": 2.628667297599587,
      "grad_norm": 1.25,
      "learning_rate": 0.00045994156739209897,
      "loss": 1.5373,
      "step": 30553
    },
    {
      "epoch": 2.628753333906909,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004599146136382223,
      "loss": 1.7899,
      "step": 30554
    },
    {
      "epoch": 2.6288393702142305,
      "grad_norm": 1.015625,
      "learning_rate": 0.00045988766000158753,
      "loss": 1.9166,
      "step": 30555
    },
    {
      "epoch": 2.628925406521552,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00045986070648227354,
      "loss": 1.0934,
      "step": 30556
    },
    {
      "epoch": 2.629011442828874,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00045983375308035924,
      "loss": 1.4385,
      "step": 30557
    },
    {
      "epoch": 2.6290974791361954,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004598067997959234,
      "loss": 1.79,
      "step": 30558
    },
    {
      "epoch": 2.629183515443517,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0004597798466290447,
      "loss": 0.6854,
      "step": 30559
    },
    {
      "epoch": 2.6292695517508387,
      "grad_norm": 1.0,
      "learning_rate": 0.0004597528935798023,
      "loss": 0.9358,
      "step": 30560
    },
    {
      "epoch": 2.6293555880581607,
      "grad_norm": 1.625,
      "learning_rate": 0.00045972594064827484,
      "loss": 1.6122,
      "step": 30561
    },
    {
      "epoch": 2.6294416243654823,
      "grad_norm": 0.890625,
      "learning_rate": 0.00045969898783454114,
      "loss": 1.2518,
      "step": 30562
    },
    {
      "epoch": 2.629527660672804,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00045967203513867994,
      "loss": 1.4628,
      "step": 30563
    },
    {
      "epoch": 2.6296136969801256,
      "grad_norm": 1.28125,
      "learning_rate": 0.00045964508256077035,
      "loss": 1.1848,
      "step": 30564
    },
    {
      "epoch": 2.629699733287447,
      "grad_norm": 1.265625,
      "learning_rate": 0.000459618130100891,
      "loss": 1.345,
      "step": 30565
    },
    {
      "epoch": 2.629785769594769,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00045959117775912076,
      "loss": 1.6601,
      "step": 30566
    },
    {
      "epoch": 2.6298718059020905,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004595642255355384,
      "loss": 1.6498,
      "step": 30567
    },
    {
      "epoch": 2.6299578422094125,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004595372734302228,
      "loss": 0.8719,
      "step": 30568
    },
    {
      "epoch": 2.630043878516734,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00045951032144325287,
      "loss": 0.7559,
      "step": 30569
    },
    {
      "epoch": 2.630129914824056,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004594833695747075,
      "loss": 1.2491,
      "step": 30570
    },
    {
      "epoch": 2.6302159511313774,
      "grad_norm": 1.8125,
      "learning_rate": 0.0004594564178246652,
      "loss": 1.4858,
      "step": 30571
    },
    {
      "epoch": 2.630301987438699,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00045942946619320516,
      "loss": 1.0033,
      "step": 30572
    },
    {
      "epoch": 2.630388023746021,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00045940251468040596,
      "loss": 0.6849,
      "step": 30573
    },
    {
      "epoch": 2.6304740600533423,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00045937556328634653,
      "loss": 1.7298,
      "step": 30574
    },
    {
      "epoch": 2.6305600963606643,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004593486120111056,
      "loss": 0.9284,
      "step": 30575
    },
    {
      "epoch": 2.630646132667986,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00045932166085476214,
      "loss": 1.0348,
      "step": 30576
    },
    {
      "epoch": 2.6307321689753076,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000459294709817395,
      "loss": 1.2739,
      "step": 30577
    },
    {
      "epoch": 2.630818205282629,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004592677588990828,
      "loss": 0.9938,
      "step": 30578
    },
    {
      "epoch": 2.630904241589951,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004592408080999044,
      "loss": 1.785,
      "step": 30579
    },
    {
      "epoch": 2.630990277897273,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00045921385741993885,
      "loss": 0.8159,
      "step": 30580
    },
    {
      "epoch": 2.631076314204594,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004591869068592649,
      "loss": 1.039,
      "step": 30581
    },
    {
      "epoch": 2.631162350511916,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004591599564179611,
      "loss": 0.6663,
      "step": 30582
    },
    {
      "epoch": 2.631248386819238,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00045913300609610666,
      "loss": 0.8351,
      "step": 30583
    },
    {
      "epoch": 2.6313344231265594,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004591060558937802,
      "loss": 0.9921,
      "step": 30584
    },
    {
      "epoch": 2.631420459433881,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00045907910581106053,
      "loss": 1.4505,
      "step": 30585
    },
    {
      "epoch": 2.6315064957412027,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00045905215584802643,
      "loss": 0.9654,
      "step": 30586
    },
    {
      "epoch": 2.6315925320485247,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00045902520600475693,
      "loss": 0.834,
      "step": 30587
    },
    {
      "epoch": 2.631678568355846,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00045899825628133075,
      "loss": 1.4439,
      "step": 30588
    },
    {
      "epoch": 2.631764604663168,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0004589713066778266,
      "loss": 1.4883,
      "step": 30589
    },
    {
      "epoch": 2.6318506409704896,
      "grad_norm": 1.28125,
      "learning_rate": 0.00045894435719432347,
      "loss": 1.4574,
      "step": 30590
    },
    {
      "epoch": 2.631936677277811,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004589174078309001,
      "loss": 1.7257,
      "step": 30591
    },
    {
      "epoch": 2.632022713585133,
      "grad_norm": 1.0625,
      "learning_rate": 0.00045889045858763536,
      "loss": 1.0331,
      "step": 30592
    },
    {
      "epoch": 2.6321087498924545,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004588635094646079,
      "loss": 1.0354,
      "step": 30593
    },
    {
      "epoch": 2.6321947861997765,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004588365604618968,
      "loss": 0.5782,
      "step": 30594
    },
    {
      "epoch": 2.632280822507098,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004588096115795808,
      "loss": 1.3269,
      "step": 30595
    },
    {
      "epoch": 2.63236685881442,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00045878266281773844,
      "loss": 1.4889,
      "step": 30596
    },
    {
      "epoch": 2.6324528951217414,
      "grad_norm": 1.140625,
      "learning_rate": 0.000458755714176449,
      "loss": 1.2615,
      "step": 30597
    },
    {
      "epoch": 2.632538931429063,
      "grad_norm": 1.125,
      "learning_rate": 0.00045872876565579104,
      "loss": 0.8739,
      "step": 30598
    },
    {
      "epoch": 2.6326249677363847,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00045870181725584335,
      "loss": 1.2544,
      "step": 30599
    },
    {
      "epoch": 2.6327110040437063,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00045867486897668483,
      "loss": 1.2932,
      "step": 30600
    },
    {
      "epoch": 2.6327970403510284,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004586479208183943,
      "loss": 0.9161,
      "step": 30601
    },
    {
      "epoch": 2.63288307665835,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00045862097278105056,
      "loss": 0.8319,
      "step": 30602
    },
    {
      "epoch": 2.6329691129656716,
      "grad_norm": 1.03125,
      "learning_rate": 0.00045859402486473245,
      "loss": 1.5789,
      "step": 30603
    },
    {
      "epoch": 2.6330551492729932,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004585670770695186,
      "loss": 0.9615,
      "step": 30604
    },
    {
      "epoch": 2.633141185580315,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00045854012939548793,
      "loss": 0.7798,
      "step": 30605
    },
    {
      "epoch": 2.6332272218876365,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00045851318184271956,
      "loss": 0.9618,
      "step": 30606
    },
    {
      "epoch": 2.633313258194958,
      "grad_norm": 1.15625,
      "learning_rate": 0.00045848623441129205,
      "loss": 1.3619,
      "step": 30607
    },
    {
      "epoch": 2.63339929450228,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00045845928710128424,
      "loss": 0.9068,
      "step": 30608
    },
    {
      "epoch": 2.633485330809602,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00045843233991277484,
      "loss": 0.7224,
      "step": 30609
    },
    {
      "epoch": 2.6335713671169234,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004584053928458428,
      "loss": 1.3597,
      "step": 30610
    },
    {
      "epoch": 2.633657403424245,
      "grad_norm": 2.015625,
      "learning_rate": 0.0004583784459005669,
      "loss": 0.6295,
      "step": 30611
    },
    {
      "epoch": 2.6337434397315667,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004583514990770259,
      "loss": 1.8253,
      "step": 30612
    },
    {
      "epoch": 2.6338294760388883,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004583245523752988,
      "loss": 1.2803,
      "step": 30613
    },
    {
      "epoch": 2.63391551234621,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00045829760579546423,
      "loss": 0.8981,
      "step": 30614
    },
    {
      "epoch": 2.634001548653532,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00045827065933760105,
      "loss": 1.6642,
      "step": 30615
    },
    {
      "epoch": 2.6340875849608536,
      "grad_norm": 2.21875,
      "learning_rate": 0.000458243713001788,
      "loss": 1.305,
      "step": 30616
    },
    {
      "epoch": 2.6341736212681752,
      "grad_norm": 1.34375,
      "learning_rate": 0.000458216766788104,
      "loss": 1.0896,
      "step": 30617
    },
    {
      "epoch": 2.634259657575497,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004581898206966279,
      "loss": 1.4134,
      "step": 30618
    },
    {
      "epoch": 2.6343456938828185,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004581628747274383,
      "loss": 2.0741,
      "step": 30619
    },
    {
      "epoch": 2.63443173019014,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004581359288806143,
      "loss": 1.397,
      "step": 30620
    },
    {
      "epoch": 2.6345177664974617,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004581089831562345,
      "loss": 1.0578,
      "step": 30621
    },
    {
      "epoch": 2.634603802804784,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004580820375543778,
      "loss": 1.4271,
      "step": 30622
    },
    {
      "epoch": 2.6346898391121054,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00045805509207512293,
      "loss": 0.9454,
      "step": 30623
    },
    {
      "epoch": 2.634775875419427,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004580281467185488,
      "loss": 1.0378,
      "step": 30624
    },
    {
      "epoch": 2.6348619117267487,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004580012014847342,
      "loss": 1.3686,
      "step": 30625
    },
    {
      "epoch": 2.6349479480340703,
      "grad_norm": 1.28125,
      "learning_rate": 0.000457974256373758,
      "loss": 2.1152,
      "step": 30626
    },
    {
      "epoch": 2.635033984341392,
      "grad_norm": 1.140625,
      "learning_rate": 0.00045794731138569875,
      "loss": 0.9526,
      "step": 30627
    },
    {
      "epoch": 2.6351200206487135,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004579203665206355,
      "loss": 1.6237,
      "step": 30628
    },
    {
      "epoch": 2.6352060569560356,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00045789342177864707,
      "loss": 1.5098,
      "step": 30629
    },
    {
      "epoch": 2.6352920932633572,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000457866477159812,
      "loss": 0.7664,
      "step": 30630
    },
    {
      "epoch": 2.635378129570679,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004578395326642095,
      "loss": 1.1862,
      "step": 30631
    },
    {
      "epoch": 2.6354641658780005,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004578125882919181,
      "loss": 1.5391,
      "step": 30632
    },
    {
      "epoch": 2.635550202185322,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004577856440430167,
      "loss": 1.0636,
      "step": 30633
    },
    {
      "epoch": 2.635636238492644,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00045775869991758396,
      "loss": 1.3882,
      "step": 30634
    },
    {
      "epoch": 2.6357222747999653,
      "grad_norm": 1.046875,
      "learning_rate": 0.00045773175591569886,
      "loss": 1.423,
      "step": 30635
    },
    {
      "epoch": 2.6358083111072874,
      "grad_norm": 1.421875,
      "learning_rate": 0.00045770481203744024,
      "loss": 1.5009,
      "step": 30636
    },
    {
      "epoch": 2.635894347414609,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004576778682828867,
      "loss": 0.9704,
      "step": 30637
    },
    {
      "epoch": 2.6359803837219307,
      "grad_norm": 0.90625,
      "learning_rate": 0.00045765092465211723,
      "loss": 1.5219,
      "step": 30638
    },
    {
      "epoch": 2.6360664200292523,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004576239811452106,
      "loss": 1.4628,
      "step": 30639
    },
    {
      "epoch": 2.636152456336574,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004575970377622456,
      "loss": 0.8185,
      "step": 30640
    },
    {
      "epoch": 2.636238492643896,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004575700945033007,
      "loss": 0.9428,
      "step": 30641
    },
    {
      "epoch": 2.636324528951217,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00045754315136845524,
      "loss": 1.9289,
      "step": 30642
    },
    {
      "epoch": 2.6364105652585392,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004575162083577879,
      "loss": 1.309,
      "step": 30643
    },
    {
      "epoch": 2.636496601565861,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00045748926547137735,
      "loss": 0.8103,
      "step": 30644
    },
    {
      "epoch": 2.6365826378731825,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00045746232270930243,
      "loss": 0.923,
      "step": 30645
    },
    {
      "epoch": 2.636668674180504,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00045743538007164187,
      "loss": 1.8313,
      "step": 30646
    },
    {
      "epoch": 2.6367547104878257,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004574084375584746,
      "loss": 1.0119,
      "step": 30647
    },
    {
      "epoch": 2.636840746795148,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004573814951698794,
      "loss": 0.5985,
      "step": 30648
    },
    {
      "epoch": 2.636926783102469,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004573545529059351,
      "loss": 1.5266,
      "step": 30649
    },
    {
      "epoch": 2.637012819409791,
      "grad_norm": 1.265625,
      "learning_rate": 0.00045732761076672023,
      "loss": 1.1379,
      "step": 30650
    },
    {
      "epoch": 2.6370988557171127,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00045730066875231395,
      "loss": 1.1184,
      "step": 30651
    },
    {
      "epoch": 2.6371848920244343,
      "grad_norm": 1.0,
      "learning_rate": 0.00045727372686279495,
      "loss": 1.2294,
      "step": 30652
    },
    {
      "epoch": 2.637270928331756,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004572467850982418,
      "loss": 1.059,
      "step": 30653
    },
    {
      "epoch": 2.6373569646390775,
      "grad_norm": 1.25,
      "learning_rate": 0.00045721984345873366,
      "loss": 1.6163,
      "step": 30654
    },
    {
      "epoch": 2.6374430009463996,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00045719290194434916,
      "loss": 1.2558,
      "step": 30655
    },
    {
      "epoch": 2.6375290372537212,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00045716596055516703,
      "loss": 1.0567,
      "step": 30656
    },
    {
      "epoch": 2.637615073561043,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004571390192912661,
      "loss": 1.8649,
      "step": 30657
    },
    {
      "epoch": 2.6377011098683645,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00045711207815272525,
      "loss": 1.0017,
      "step": 30658
    },
    {
      "epoch": 2.637787146175686,
      "grad_norm": 1.3125,
      "learning_rate": 0.00045708513713962326,
      "loss": 1.3427,
      "step": 30659
    },
    {
      "epoch": 2.6378731824830077,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00045705819625203875,
      "loss": 0.7454,
      "step": 30660
    },
    {
      "epoch": 2.6379592187903294,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004570312554900508,
      "loss": 1.0525,
      "step": 30661
    },
    {
      "epoch": 2.6380452550976514,
      "grad_norm": 1.21875,
      "learning_rate": 0.00045700431485373814,
      "loss": 1.6177,
      "step": 30662
    },
    {
      "epoch": 2.638131291404973,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00045697737434317935,
      "loss": 0.8761,
      "step": 30663
    },
    {
      "epoch": 2.6382173277122947,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004569504339584534,
      "loss": 1.4072,
      "step": 30664
    },
    {
      "epoch": 2.6383033640196163,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0004569234936996391,
      "loss": 1.0347,
      "step": 30665
    },
    {
      "epoch": 2.638389400326938,
      "grad_norm": 1.28125,
      "learning_rate": 0.00045689655356681525,
      "loss": 1.465,
      "step": 30666
    },
    {
      "epoch": 2.6384754366342595,
      "grad_norm": 1.1875,
      "learning_rate": 0.00045686961356006055,
      "loss": 1.1123,
      "step": 30667
    },
    {
      "epoch": 2.638561472941581,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00045684267367945375,
      "loss": 0.7243,
      "step": 30668
    },
    {
      "epoch": 2.6386475092489032,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004568157339250739,
      "loss": 0.6403,
      "step": 30669
    },
    {
      "epoch": 2.638733545556225,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004567887942969995,
      "loss": 1.2181,
      "step": 30670
    },
    {
      "epoch": 2.6388195818635465,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004567618547953095,
      "loss": 0.6593,
      "step": 30671
    },
    {
      "epoch": 2.638905618170868,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00045673491542008265,
      "loss": 0.9032,
      "step": 30672
    },
    {
      "epoch": 2.6389916544781897,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00045670797617139783,
      "loss": 1.3153,
      "step": 30673
    },
    {
      "epoch": 2.6390776907855114,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004566810370493337,
      "loss": 1.0401,
      "step": 30674
    },
    {
      "epoch": 2.639163727092833,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004566540980539691,
      "loss": 1.175,
      "step": 30675
    },
    {
      "epoch": 2.639249763400155,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0004566271591853829,
      "loss": 0.5134,
      "step": 30676
    },
    {
      "epoch": 2.6393357997074767,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00045660022044365377,
      "loss": 1.2674,
      "step": 30677
    },
    {
      "epoch": 2.6394218360147983,
      "grad_norm": 1.21875,
      "learning_rate": 0.00045657328182886043,
      "loss": 1.3682,
      "step": 30678
    },
    {
      "epoch": 2.63950787232212,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00045654634334108197,
      "loss": 0.6731,
      "step": 30679
    },
    {
      "epoch": 2.6395939086294415,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004565194049803969,
      "loss": 0.7459,
      "step": 30680
    },
    {
      "epoch": 2.639679944936763,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004564924667468842,
      "loss": 1.5906,
      "step": 30681
    },
    {
      "epoch": 2.639765981244085,
      "grad_norm": 1.0,
      "learning_rate": 0.00045646552864062257,
      "loss": 1.173,
      "step": 30682
    },
    {
      "epoch": 2.639852017551407,
      "grad_norm": 1.203125,
      "learning_rate": 0.00045643859066169076,
      "loss": 1.1975,
      "step": 30683
    },
    {
      "epoch": 2.6399380538587285,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004564116528101676,
      "loss": 1.3398,
      "step": 30684
    },
    {
      "epoch": 2.64002409016605,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00045638471508613196,
      "loss": 0.874,
      "step": 30685
    },
    {
      "epoch": 2.6401101264733717,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004563577774896625,
      "loss": 1.1444,
      "step": 30686
    },
    {
      "epoch": 2.6401961627806934,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000456330840020838,
      "loss": 1.3305,
      "step": 30687
    },
    {
      "epoch": 2.640282199088015,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00045630390267973735,
      "loss": 0.8882,
      "step": 30688
    },
    {
      "epoch": 2.6403682353953366,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004562769654664393,
      "loss": 0.9546,
      "step": 30689
    },
    {
      "epoch": 2.6404542717026587,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004562500283810227,
      "loss": 0.8436,
      "step": 30690
    },
    {
      "epoch": 2.6405403080099803,
      "grad_norm": 1.125,
      "learning_rate": 0.00045622309142356605,
      "loss": 1.0759,
      "step": 30691
    },
    {
      "epoch": 2.640626344317302,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004561961545941485,
      "loss": 1.7516,
      "step": 30692
    },
    {
      "epoch": 2.6407123806246235,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00045616921789284865,
      "loss": 1.3183,
      "step": 30693
    },
    {
      "epoch": 2.640798416931945,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004561422813197452,
      "loss": 0.5741,
      "step": 30694
    },
    {
      "epoch": 2.6408844532392672,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004561153448749172,
      "loss": 1.0762,
      "step": 30695
    },
    {
      "epoch": 2.6409704895465884,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00045608840855844326,
      "loss": 1.477,
      "step": 30696
    },
    {
      "epoch": 2.6410565258539105,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004560614723704022,
      "loss": 1.3858,
      "step": 30697
    },
    {
      "epoch": 2.641142562161232,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004560345363108727,
      "loss": 1.5771,
      "step": 30698
    },
    {
      "epoch": 2.6412285984685537,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004560076003799336,
      "loss": 1.0293,
      "step": 30699
    },
    {
      "epoch": 2.6413146347758754,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004559806645776639,
      "loss": 1.5953,
      "step": 30700
    },
    {
      "epoch": 2.641400671083197,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00045595372890414195,
      "loss": 1.3681,
      "step": 30701
    },
    {
      "epoch": 2.641486707390519,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00045592679335944693,
      "loss": 1.2912,
      "step": 30702
    },
    {
      "epoch": 2.6415727436978402,
      "grad_norm": 1.0,
      "learning_rate": 0.0004558998579436575,
      "loss": 0.8996,
      "step": 30703
    },
    {
      "epoch": 2.6416587800051623,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004558729226568523,
      "loss": 1.2023,
      "step": 30704
    },
    {
      "epoch": 2.641744816312484,
      "grad_norm": 1.203125,
      "learning_rate": 0.00045584598749911023,
      "loss": 1.2701,
      "step": 30705
    },
    {
      "epoch": 2.6418308526198055,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00045581905247051014,
      "loss": 0.9315,
      "step": 30706
    },
    {
      "epoch": 2.641916888927127,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00045579211757113073,
      "loss": 1.3258,
      "step": 30707
    },
    {
      "epoch": 2.642002925234449,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004557651828010506,
      "loss": 0.7773,
      "step": 30708
    },
    {
      "epoch": 2.642088961541771,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004557382481603489,
      "loss": 1.5023,
      "step": 30709
    },
    {
      "epoch": 2.642174997849092,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00045571131364910417,
      "loss": 1.6238,
      "step": 30710
    },
    {
      "epoch": 2.642261034156414,
      "grad_norm": 0.8125,
      "learning_rate": 0.00045568437926739525,
      "loss": 1.0061,
      "step": 30711
    },
    {
      "epoch": 2.6423470704637357,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004556574450153008,
      "loss": 1.0699,
      "step": 30712
    },
    {
      "epoch": 2.6424331067710574,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004556305108928998,
      "loss": 1.8983,
      "step": 30713
    },
    {
      "epoch": 2.642519143078379,
      "grad_norm": 1.125,
      "learning_rate": 0.00045560357690027094,
      "loss": 1.5839,
      "step": 30714
    },
    {
      "epoch": 2.6426051793857006,
      "grad_norm": 1.28125,
      "learning_rate": 0.00045557664303749274,
      "loss": 1.0991,
      "step": 30715
    },
    {
      "epoch": 2.6426912156930227,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00045554970930464445,
      "loss": 0.8193,
      "step": 30716
    },
    {
      "epoch": 2.6427772520003443,
      "grad_norm": 1.078125,
      "learning_rate": 0.00045552277570180453,
      "loss": 1.5153,
      "step": 30717
    },
    {
      "epoch": 2.642863288307666,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00045549584222905195,
      "loss": 0.9394,
      "step": 30718
    },
    {
      "epoch": 2.6429493246149875,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004554689088864653,
      "loss": 1.2001,
      "step": 30719
    },
    {
      "epoch": 2.643035360922309,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004554419756741235,
      "loss": 1.6572,
      "step": 30720
    },
    {
      "epoch": 2.643121397229631,
      "grad_norm": 1.25,
      "learning_rate": 0.00045541504259210517,
      "loss": 2.2633,
      "step": 30721
    },
    {
      "epoch": 2.6432074335369524,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004553881096404892,
      "loss": 1.7204,
      "step": 30722
    },
    {
      "epoch": 2.6432934698442745,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00045536117681935434,
      "loss": 0.9112,
      "step": 30723
    },
    {
      "epoch": 2.643379506151596,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004553342441287793,
      "loss": 1.1772,
      "step": 30724
    },
    {
      "epoch": 2.6434655424589177,
      "grad_norm": 1.25,
      "learning_rate": 0.000455307311568843,
      "loss": 1.3803,
      "step": 30725
    },
    {
      "epoch": 2.6435515787662394,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004552803791396241,
      "loss": 1.1547,
      "step": 30726
    },
    {
      "epoch": 2.643637615073561,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004552534468412014,
      "loss": 1.8368,
      "step": 30727
    },
    {
      "epoch": 2.6437236513808826,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004552265146736536,
      "loss": 0.9375,
      "step": 30728
    },
    {
      "epoch": 2.6438096876882042,
      "grad_norm": 0.859375,
      "learning_rate": 0.00045519958263705963,
      "loss": 1.1465,
      "step": 30729
    },
    {
      "epoch": 2.6438957239955263,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004551726507314981,
      "loss": 1.6197,
      "step": 30730
    },
    {
      "epoch": 2.643981760302848,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00045514571895704793,
      "loss": 2.135,
      "step": 30731
    },
    {
      "epoch": 2.6440677966101696,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00045511878731378764,
      "loss": 1.6443,
      "step": 30732
    },
    {
      "epoch": 2.644153832917491,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00045509185580179637,
      "loss": 1.2427,
      "step": 30733
    },
    {
      "epoch": 2.644239869224813,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00045506492442115256,
      "loss": 0.9311,
      "step": 30734
    },
    {
      "epoch": 2.6443259055321344,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00045503799317193503,
      "loss": 0.7948,
      "step": 30735
    },
    {
      "epoch": 2.644411941839456,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00045501106205422283,
      "loss": 0.6256,
      "step": 30736
    },
    {
      "epoch": 2.644497978146778,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00045498413106809444,
      "loss": 1.8502,
      "step": 30737
    },
    {
      "epoch": 2.6445840144540997,
      "grad_norm": 1.15625,
      "learning_rate": 0.00045495720021362874,
      "loss": 0.9938,
      "step": 30738
    },
    {
      "epoch": 2.6446700507614214,
      "grad_norm": 1.09375,
      "learning_rate": 0.00045493026949090435,
      "loss": 1.0726,
      "step": 30739
    },
    {
      "epoch": 2.644756087068743,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00045490333890000023,
      "loss": 1.8258,
      "step": 30740
    },
    {
      "epoch": 2.6448421233760646,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00045487640844099516,
      "loss": 1.5428,
      "step": 30741
    },
    {
      "epoch": 2.6449281596833862,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004548494781139676,
      "loss": 1.4091,
      "step": 30742
    },
    {
      "epoch": 2.645014195990708,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00045482254791899676,
      "loss": 1.1552,
      "step": 30743
    },
    {
      "epoch": 2.64510023229803,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00045479561785616106,
      "loss": 1.4187,
      "step": 30744
    },
    {
      "epoch": 2.6451862686053516,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0004547686879255395,
      "loss": 2.0083,
      "step": 30745
    },
    {
      "epoch": 2.645272304912673,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004547417581272105,
      "loss": 1.167,
      "step": 30746
    },
    {
      "epoch": 2.645358341219995,
      "grad_norm": 0.984375,
      "learning_rate": 0.00045471482846125326,
      "loss": 0.7137,
      "step": 30747
    },
    {
      "epoch": 2.6454443775273164,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004546878989277463,
      "loss": 1.0988,
      "step": 30748
    },
    {
      "epoch": 2.6455304138346385,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004546609695267684,
      "loss": 1.4849,
      "step": 30749
    },
    {
      "epoch": 2.6456164501419597,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004546340402583983,
      "loss": 1.1662,
      "step": 30750
    },
    {
      "epoch": 2.6457024864492817,
      "grad_norm": 1.25,
      "learning_rate": 0.0004546071111227149,
      "loss": 1.4786,
      "step": 30751
    },
    {
      "epoch": 2.6457885227566034,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004545801821197967,
      "loss": 1.0609,
      "step": 30752
    },
    {
      "epoch": 2.645874559063925,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00045455325324972274,
      "loss": 1.2233,
      "step": 30753
    },
    {
      "epoch": 2.6459605953712466,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004545263245125718,
      "loss": 0.6414,
      "step": 30754
    },
    {
      "epoch": 2.6460466316785682,
      "grad_norm": 0.96875,
      "learning_rate": 0.00045449939590842234,
      "loss": 0.8801,
      "step": 30755
    },
    {
      "epoch": 2.6461326679858903,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0004544724674373534,
      "loss": 0.7735,
      "step": 30756
    },
    {
      "epoch": 2.6462187042932115,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004544455390994436,
      "loss": 1.7532,
      "step": 30757
    },
    {
      "epoch": 2.6463047406005336,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004544186108947717,
      "loss": 1.7701,
      "step": 30758
    },
    {
      "epoch": 2.646390776907855,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004543916828234166,
      "loss": 1.0041,
      "step": 30759
    },
    {
      "epoch": 2.646476813215177,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00045436475488545695,
      "loss": 1.2771,
      "step": 30760
    },
    {
      "epoch": 2.6465628495224984,
      "grad_norm": 1.390625,
      "learning_rate": 0.0004543378270809715,
      "loss": 1.2438,
      "step": 30761
    },
    {
      "epoch": 2.64664888582982,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004543108994100389,
      "loss": 1.1811,
      "step": 30762
    },
    {
      "epoch": 2.646734922137142,
      "grad_norm": 0.953125,
      "learning_rate": 0.00045428397187273814,
      "loss": 0.9165,
      "step": 30763
    },
    {
      "epoch": 2.6468209584444633,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004542570444691479,
      "loss": 0.9862,
      "step": 30764
    },
    {
      "epoch": 2.6469069947517854,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004542301171993468,
      "loss": 1.7006,
      "step": 30765
    },
    {
      "epoch": 2.646993031059107,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0004542031900634138,
      "loss": 0.5466,
      "step": 30766
    },
    {
      "epoch": 2.6470790673664286,
      "grad_norm": 1.125,
      "learning_rate": 0.00045417626306142755,
      "loss": 0.9153,
      "step": 30767
    },
    {
      "epoch": 2.6471651036737502,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004541493361934669,
      "loss": 1.7346,
      "step": 30768
    },
    {
      "epoch": 2.647251139981072,
      "grad_norm": 1.25,
      "learning_rate": 0.0004541224094596103,
      "loss": 1.0453,
      "step": 30769
    },
    {
      "epoch": 2.647337176288394,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004540954828599369,
      "loss": 1.4907,
      "step": 30770
    },
    {
      "epoch": 2.6474232125957156,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00045406855639452525,
      "loss": 1.495,
      "step": 30771
    },
    {
      "epoch": 2.647509248903037,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004540416300634541,
      "loss": 2.1056,
      "step": 30772
    },
    {
      "epoch": 2.647595285210359,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00045401470386680233,
      "loss": 0.7417,
      "step": 30773
    },
    {
      "epoch": 2.6476813215176804,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004539877778046486,
      "loss": 1.4834,
      "step": 30774
    },
    {
      "epoch": 2.647767357825002,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004539608518770716,
      "loss": 1.3865,
      "step": 30775
    },
    {
      "epoch": 2.6478533941323237,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004539339260841501,
      "loss": 1.0042,
      "step": 30776
    },
    {
      "epoch": 2.6479394304396457,
      "grad_norm": 0.953125,
      "learning_rate": 0.00045390700042596306,
      "loss": 1.2382,
      "step": 30777
    },
    {
      "epoch": 2.6480254667469674,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00045388007490258905,
      "loss": 0.9088,
      "step": 30778
    },
    {
      "epoch": 2.648111503054289,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0004538531495141068,
      "loss": 1.8177,
      "step": 30779
    },
    {
      "epoch": 2.6481975393616106,
      "grad_norm": 1.296875,
      "learning_rate": 0.00045382622426059507,
      "loss": 0.9198,
      "step": 30780
    },
    {
      "epoch": 2.6482835756689322,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0004537992991421327,
      "loss": 1.3617,
      "step": 30781
    },
    {
      "epoch": 2.648369611976254,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00045377237415879847,
      "loss": 0.7602,
      "step": 30782
    },
    {
      "epoch": 2.6484556482835755,
      "grad_norm": 1.578125,
      "learning_rate": 0.00045374544931067084,
      "loss": 0.9674,
      "step": 30783
    },
    {
      "epoch": 2.6485416845908976,
      "grad_norm": 1.140625,
      "learning_rate": 0.00045371852459782897,
      "loss": 1.5359,
      "step": 30784
    },
    {
      "epoch": 2.648627720898219,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00045369160002035143,
      "loss": 0.8711,
      "step": 30785
    },
    {
      "epoch": 2.648713757205541,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004536646755783169,
      "loss": 0.935,
      "step": 30786
    },
    {
      "epoch": 2.6487997935128624,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00045363775127180405,
      "loss": 1.8736,
      "step": 30787
    },
    {
      "epoch": 2.648885829820184,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00045361082710089197,
      "loss": 1.7433,
      "step": 30788
    },
    {
      "epoch": 2.6489718661275057,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004535839030656589,
      "loss": 1.7752,
      "step": 30789
    },
    {
      "epoch": 2.6490579024348273,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004535569791661841,
      "loss": 0.9489,
      "step": 30790
    },
    {
      "epoch": 2.6491439387421494,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00045353005540254616,
      "loss": 0.6564,
      "step": 30791
    },
    {
      "epoch": 2.649229975049471,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004535031317748236,
      "loss": 1.2474,
      "step": 30792
    },
    {
      "epoch": 2.6493160113567926,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004534762082830955,
      "loss": 1.2858,
      "step": 30793
    },
    {
      "epoch": 2.6494020476641142,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004534492849274404,
      "loss": 0.5385,
      "step": 30794
    },
    {
      "epoch": 2.649488083971436,
      "grad_norm": 1.015625,
      "learning_rate": 0.00045342236170793697,
      "loss": 1.4689,
      "step": 30795
    },
    {
      "epoch": 2.6495741202787575,
      "grad_norm": 1.1875,
      "learning_rate": 0.00045339543862466425,
      "loss": 1.0576,
      "step": 30796
    },
    {
      "epoch": 2.649660156586079,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004533685156777008,
      "loss": 0.7863,
      "step": 30797
    },
    {
      "epoch": 2.649746192893401,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004533415928671253,
      "loss": 0.9811,
      "step": 30798
    },
    {
      "epoch": 2.649832229200723,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004533146701930165,
      "loss": 1.9731,
      "step": 30799
    },
    {
      "epoch": 2.6499182655080444,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004532877476554533,
      "loss": 0.9364,
      "step": 30800
    },
    {
      "epoch": 2.650004301815366,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0004532608252545144,
      "loss": 1.7233,
      "step": 30801
    },
    {
      "epoch": 2.6500903381226877,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00045323390299027846,
      "loss": 0.5978,
      "step": 30802
    },
    {
      "epoch": 2.6501763744300093,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004532069808628241,
      "loss": 1.1006,
      "step": 30803
    },
    {
      "epoch": 2.650262410737331,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004531800588722304,
      "loss": 1.0412,
      "step": 30804
    },
    {
      "epoch": 2.650348447044653,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00045315313701857593,
      "loss": 1.1854,
      "step": 30805
    },
    {
      "epoch": 2.6504344833519746,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00045312621530193924,
      "loss": 0.9918,
      "step": 30806
    },
    {
      "epoch": 2.6505205196592962,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004530992937223994,
      "loss": 0.948,
      "step": 30807
    },
    {
      "epoch": 2.650606555966618,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00045307237228003503,
      "loss": 1.5871,
      "step": 30808
    },
    {
      "epoch": 2.6506925922739395,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004530454509749249,
      "loss": 1.9521,
      "step": 30809
    },
    {
      "epoch": 2.6507786285812616,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004530185298071475,
      "loss": 0.6047,
      "step": 30810
    },
    {
      "epoch": 2.6508646648885827,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00045299160877678183,
      "loss": 0.6725,
      "step": 30811
    },
    {
      "epoch": 2.650950701195905,
      "grad_norm": 1.734375,
      "learning_rate": 0.00045296468788390666,
      "loss": 1.2566,
      "step": 30812
    },
    {
      "epoch": 2.6510367375032264,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004529377671286005,
      "loss": 0.861,
      "step": 30813
    },
    {
      "epoch": 2.651122773810548,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004529108465109423,
      "loss": 0.9403,
      "step": 30814
    },
    {
      "epoch": 2.6512088101178697,
      "grad_norm": 1.359375,
      "learning_rate": 0.00045288392603101067,
      "loss": 0.816,
      "step": 30815
    },
    {
      "epoch": 2.6512948464251913,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004528570056888845,
      "loss": 1.2556,
      "step": 30816
    },
    {
      "epoch": 2.6513808827325134,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004528300854846422,
      "loss": 0.9624,
      "step": 30817
    },
    {
      "epoch": 2.6514669190398346,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004528031654183629,
      "loss": 1.5492,
      "step": 30818
    },
    {
      "epoch": 2.6515529553471566,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004527762454901252,
      "loss": 1.0717,
      "step": 30819
    },
    {
      "epoch": 2.6516389916544783,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004527493257000077,
      "loss": 1.066,
      "step": 30820
    },
    {
      "epoch": 2.6517250279618,
      "grad_norm": 0.703125,
      "learning_rate": 0.00045272240604808916,
      "loss": 1.073,
      "step": 30821
    },
    {
      "epoch": 2.6518110642691215,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00045269548653444855,
      "loss": 0.8036,
      "step": 30822
    },
    {
      "epoch": 2.651897100576443,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004526685671591644,
      "loss": 0.8988,
      "step": 30823
    },
    {
      "epoch": 2.651983136883765,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004526416479223154,
      "loss": 1.3523,
      "step": 30824
    },
    {
      "epoch": 2.6520691731910864,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004526147288239804,
      "loss": 1.3754,
      "step": 30825
    },
    {
      "epoch": 2.6521552094984084,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00045258780986423796,
      "loss": 0.6344,
      "step": 30826
    },
    {
      "epoch": 2.65224124580573,
      "grad_norm": 1.03125,
      "learning_rate": 0.00045256089104316717,
      "loss": 1.0931,
      "step": 30827
    },
    {
      "epoch": 2.6523272821130517,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00045253397236084654,
      "loss": 1.3282,
      "step": 30828
    },
    {
      "epoch": 2.6524133184203733,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004525070538173547,
      "loss": 1.2268,
      "step": 30829
    },
    {
      "epoch": 2.652499354727695,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004524801354127706,
      "loss": 0.8881,
      "step": 30830
    },
    {
      "epoch": 2.652585391035017,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004524532171471728,
      "loss": 1.235,
      "step": 30831
    },
    {
      "epoch": 2.6526714273423386,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00045242629902064014,
      "loss": 1.3847,
      "step": 30832
    },
    {
      "epoch": 2.6527574636496603,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004523993810332512,
      "loss": 1.6234,
      "step": 30833
    },
    {
      "epoch": 2.652843499956982,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004523724631850849,
      "loss": 0.8523,
      "step": 30834
    },
    {
      "epoch": 2.6529295362643035,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004523455454762199,
      "loss": 0.6104,
      "step": 30835
    },
    {
      "epoch": 2.653015572571625,
      "grad_norm": 1.015625,
      "learning_rate": 0.00045231862790673486,
      "loss": 1.4884,
      "step": 30836
    },
    {
      "epoch": 2.6531016088789467,
      "grad_norm": 1.0625,
      "learning_rate": 0.00045229171047670856,
      "loss": 1.0642,
      "step": 30837
    },
    {
      "epoch": 2.653187645186269,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00045226479318621985,
      "loss": 1.9549,
      "step": 30838
    },
    {
      "epoch": 2.6532736814935904,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004522378760353472,
      "loss": 1.6115,
      "step": 30839
    },
    {
      "epoch": 2.653359717800912,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004522109590241694,
      "loss": 1.3762,
      "step": 30840
    },
    {
      "epoch": 2.6534457541082337,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00045218404215276544,
      "loss": 1.0561,
      "step": 30841
    },
    {
      "epoch": 2.6535317904155553,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004521571254212138,
      "loss": 1.0984,
      "step": 30842
    },
    {
      "epoch": 2.653617826722877,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004521302088295933,
      "loss": 1.3028,
      "step": 30843
    },
    {
      "epoch": 2.6537038630301986,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004521032923779825,
      "loss": 0.6426,
      "step": 30844
    },
    {
      "epoch": 2.6537898993375206,
      "grad_norm": 1.171875,
      "learning_rate": 0.00045207637606646035,
      "loss": 1.397,
      "step": 30845
    },
    {
      "epoch": 2.6538759356448423,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004520494598951055,
      "loss": 1.0951,
      "step": 30846
    },
    {
      "epoch": 2.653961971952164,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004520225438639965,
      "loss": 0.8939,
      "step": 30847
    },
    {
      "epoch": 2.6540480082594855,
      "grad_norm": 1.125,
      "learning_rate": 0.00045199562797321234,
      "loss": 1.2863,
      "step": 30848
    },
    {
      "epoch": 2.654134044566807,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0004519687122228316,
      "loss": 0.5968,
      "step": 30849
    },
    {
      "epoch": 2.6542200808741288,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00045194179661293315,
      "loss": 0.9187,
      "step": 30850
    },
    {
      "epoch": 2.6543061171814504,
      "grad_norm": 0.8125,
      "learning_rate": 0.00045191488114359546,
      "loss": 0.6618,
      "step": 30851
    },
    {
      "epoch": 2.6543921534887724,
      "grad_norm": 1.6875,
      "learning_rate": 0.00045188796581489745,
      "loss": 1.3399,
      "step": 30852
    },
    {
      "epoch": 2.654478189796094,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00045186105062691777,
      "loss": 1.1523,
      "step": 30853
    },
    {
      "epoch": 2.6545642261034157,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004518341355797351,
      "loss": 0.8585,
      "step": 30854
    },
    {
      "epoch": 2.6546502624107373,
      "grad_norm": 1.75,
      "learning_rate": 0.00045180722067342835,
      "loss": 1.793,
      "step": 30855
    },
    {
      "epoch": 2.654736298718059,
      "grad_norm": 1.15625,
      "learning_rate": 0.00045178030590807604,
      "loss": 1.3336,
      "step": 30856
    },
    {
      "epoch": 2.6548223350253806,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00045175339128375696,
      "loss": 1.5397,
      "step": 30857
    },
    {
      "epoch": 2.654908371332702,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00045172647680054973,
      "loss": 1.2525,
      "step": 30858
    },
    {
      "epoch": 2.6549944076400243,
      "grad_norm": 1.078125,
      "learning_rate": 0.00045169956245853327,
      "loss": 1.1683,
      "step": 30859
    },
    {
      "epoch": 2.655080443947346,
      "grad_norm": 1.296875,
      "learning_rate": 0.00045167264825778623,
      "loss": 1.6368,
      "step": 30860
    },
    {
      "epoch": 2.6551664802546675,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004516457341983872,
      "loss": 1.5347,
      "step": 30861
    },
    {
      "epoch": 2.655252516561989,
      "grad_norm": 1.078125,
      "learning_rate": 0.00045161882028041503,
      "loss": 1.29,
      "step": 30862
    },
    {
      "epoch": 2.6553385528693108,
      "grad_norm": 0.890625,
      "learning_rate": 0.00045159190650394823,
      "loss": 0.8816,
      "step": 30863
    },
    {
      "epoch": 2.6554245891766324,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004515649928690659,
      "loss": 0.9639,
      "step": 30864
    },
    {
      "epoch": 2.655510625483954,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004515380793758465,
      "loss": 1.4033,
      "step": 30865
    },
    {
      "epoch": 2.655596661791276,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004515111660243689,
      "loss": 1.4504,
      "step": 30866
    },
    {
      "epoch": 2.6556826980985977,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004514842528147115,
      "loss": 1.2558,
      "step": 30867
    },
    {
      "epoch": 2.6557687344059193,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00045145733974695333,
      "loss": 0.9174,
      "step": 30868
    },
    {
      "epoch": 2.655854770713241,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004514304268211731,
      "loss": 0.8879,
      "step": 30869
    },
    {
      "epoch": 2.6559408070205626,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00045140351403744923,
      "loss": 1.3901,
      "step": 30870
    },
    {
      "epoch": 2.6560268433278846,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004513766013958607,
      "loss": 0.9914,
      "step": 30871
    },
    {
      "epoch": 2.656112879635206,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004513496888964863,
      "loss": 1.7595,
      "step": 30872
    },
    {
      "epoch": 2.656198915942528,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00045132277653940443,
      "loss": 1.1725,
      "step": 30873
    },
    {
      "epoch": 2.6562849522498495,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000451295864324694,
      "loss": 1.5427,
      "step": 30874
    },
    {
      "epoch": 2.656370988557171,
      "grad_norm": 1.4375,
      "learning_rate": 0.00045126895225243376,
      "loss": 1.3429,
      "step": 30875
    },
    {
      "epoch": 2.6564570248644928,
      "grad_norm": 0.984375,
      "learning_rate": 0.00045124204032270234,
      "loss": 1.0086,
      "step": 30876
    },
    {
      "epoch": 2.6565430611718144,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004512151285355783,
      "loss": 0.7324,
      "step": 30877
    },
    {
      "epoch": 2.6566290974791364,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0004511882168911408,
      "loss": 1.4344,
      "step": 30878
    },
    {
      "epoch": 2.6567151337864576,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00045116130538946816,
      "loss": 1.264,
      "step": 30879
    },
    {
      "epoch": 2.6568011700937797,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004511343940306392,
      "loss": 1.1849,
      "step": 30880
    },
    {
      "epoch": 2.6568872064011013,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004511074828147325,
      "loss": 1.283,
      "step": 30881
    },
    {
      "epoch": 2.656973242708423,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00045108057174182703,
      "loss": 1.1639,
      "step": 30882
    },
    {
      "epoch": 2.6570592790157446,
      "grad_norm": 0.9375,
      "learning_rate": 0.00045105366081200145,
      "loss": 0.7542,
      "step": 30883
    },
    {
      "epoch": 2.657145315323066,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004510267500253343,
      "loss": 1.215,
      "step": 30884
    },
    {
      "epoch": 2.6572313516303883,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004509998393819043,
      "loss": 1.1765,
      "step": 30885
    },
    {
      "epoch": 2.6573173879377094,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004509729288817904,
      "loss": 1.0277,
      "step": 30886
    },
    {
      "epoch": 2.6574034242450315,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004509460185250711,
      "loss": 1.8112,
      "step": 30887
    },
    {
      "epoch": 2.657489460552353,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00045091910831182504,
      "loss": 0.6852,
      "step": 30888
    },
    {
      "epoch": 2.6575754968596748,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004508921982421312,
      "loss": 0.4581,
      "step": 30889
    },
    {
      "epoch": 2.6576615331669964,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00045086528831606807,
      "loss": 0.885,
      "step": 30890
    },
    {
      "epoch": 2.657747569474318,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004508383785337144,
      "loss": 0.8666,
      "step": 30891
    },
    {
      "epoch": 2.65783360578164,
      "grad_norm": 1.03125,
      "learning_rate": 0.00045081146889514885,
      "loss": 1.2612,
      "step": 30892
    },
    {
      "epoch": 2.6579196420889617,
      "grad_norm": 1.15625,
      "learning_rate": 0.00045078455940045025,
      "loss": 1.2274,
      "step": 30893
    },
    {
      "epoch": 2.6580056783962833,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004507576500496973,
      "loss": 0.515,
      "step": 30894
    },
    {
      "epoch": 2.658091714703605,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00045073074084296856,
      "loss": 1.3908,
      "step": 30895
    },
    {
      "epoch": 2.6581777510109266,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00045070383178034284,
      "loss": 1.238,
      "step": 30896
    },
    {
      "epoch": 2.658263787318248,
      "grad_norm": 1.0,
      "learning_rate": 0.0004506769228618989,
      "loss": 1.7626,
      "step": 30897
    },
    {
      "epoch": 2.65834982362557,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00045065001408771535,
      "loss": 1.6931,
      "step": 30898
    },
    {
      "epoch": 2.658435859932892,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004506231054578708,
      "loss": 1.1243,
      "step": 30899
    },
    {
      "epoch": 2.6585218962402135,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00045059619697244394,
      "loss": 1.5556,
      "step": 30900
    },
    {
      "epoch": 2.658607932547535,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004505692886315138,
      "loss": 1.0416,
      "step": 30901
    },
    {
      "epoch": 2.6586939688548568,
      "grad_norm": 1.109375,
      "learning_rate": 0.00045054238043515886,
      "loss": 1.1137,
      "step": 30902
    },
    {
      "epoch": 2.6587800051621784,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004505154723834579,
      "loss": 1.5693,
      "step": 30903
    },
    {
      "epoch": 2.6588660414695,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004504885644764894,
      "loss": 0.9577,
      "step": 30904
    },
    {
      "epoch": 2.6589520777768216,
      "grad_norm": 1.171875,
      "learning_rate": 0.00045046165671433236,
      "loss": 1.1571,
      "step": 30905
    },
    {
      "epoch": 2.6590381140841437,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00045043474909706535,
      "loss": 1.1905,
      "step": 30906
    },
    {
      "epoch": 2.6591241503914653,
      "grad_norm": 0.984375,
      "learning_rate": 0.000450407841624767,
      "loss": 0.812,
      "step": 30907
    },
    {
      "epoch": 2.659210186698787,
      "grad_norm": 1.484375,
      "learning_rate": 0.00045038093429751604,
      "loss": 1.1431,
      "step": 30908
    },
    {
      "epoch": 2.6592962230061086,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00045035402711539124,
      "loss": 0.9541,
      "step": 30909
    },
    {
      "epoch": 2.65938225931343,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00045032712007847124,
      "loss": 0.9905,
      "step": 30910
    },
    {
      "epoch": 2.659468295620752,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0004503002131868347,
      "loss": 0.4776,
      "step": 30911
    },
    {
      "epoch": 2.6595543319280734,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004502733064405605,
      "loss": 0.9904,
      "step": 30912
    },
    {
      "epoch": 2.6596403682353955,
      "grad_norm": 1.1875,
      "learning_rate": 0.00045024639983972715,
      "loss": 1.4806,
      "step": 30913
    },
    {
      "epoch": 2.659726404542717,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004502194933844134,
      "loss": 1.4103,
      "step": 30914
    },
    {
      "epoch": 2.6598124408500388,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004501925870746979,
      "loss": 1.0289,
      "step": 30915
    },
    {
      "epoch": 2.6598984771573604,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00045016568091065944,
      "loss": 1.6166,
      "step": 30916
    },
    {
      "epoch": 2.659984513464682,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0004501387748923767,
      "loss": 0.6085,
      "step": 30917
    },
    {
      "epoch": 2.6600705497720036,
      "grad_norm": 1.03125,
      "learning_rate": 0.00045011186901992823,
      "loss": 0.9976,
      "step": 30918
    },
    {
      "epoch": 2.6601565860793253,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000450084963293393,
      "loss": 1.1515,
      "step": 30919
    },
    {
      "epoch": 2.6602426223866473,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00045005805771284945,
      "loss": 0.7491,
      "step": 30920
    },
    {
      "epoch": 2.660328658693969,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004500311522783764,
      "loss": 2.0101,
      "step": 30921
    },
    {
      "epoch": 2.6604146950012906,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004500042469900525,
      "loss": 0.9548,
      "step": 30922
    },
    {
      "epoch": 2.660500731308612,
      "grad_norm": 1.234375,
      "learning_rate": 0.00044997734184795647,
      "loss": 1.254,
      "step": 30923
    },
    {
      "epoch": 2.660586767615934,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000449950436852167,
      "loss": 1.0099,
      "step": 30924
    },
    {
      "epoch": 2.660672803923256,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00044992353200276267,
      "loss": 0.9666,
      "step": 30925
    },
    {
      "epoch": 2.660758840230577,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00044989662729982236,
      "loss": 1.0597,
      "step": 30926
    },
    {
      "epoch": 2.660844876537899,
      "grad_norm": 0.96875,
      "learning_rate": 0.00044986972274342466,
      "loss": 0.7765,
      "step": 30927
    },
    {
      "epoch": 2.6609309128452208,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004498428183336483,
      "loss": 1.2132,
      "step": 30928
    },
    {
      "epoch": 2.6610169491525424,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00044981591407057184,
      "loss": 1.2065,
      "step": 30929
    },
    {
      "epoch": 2.661102985459864,
      "grad_norm": 1.375,
      "learning_rate": 0.00044978900995427414,
      "loss": 1.4011,
      "step": 30930
    },
    {
      "epoch": 2.6611890217671856,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004497621059848338,
      "loss": 0.9489,
      "step": 30931
    },
    {
      "epoch": 2.6612750580745077,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004497352021623296,
      "loss": 1.368,
      "step": 30932
    },
    {
      "epoch": 2.661361094381829,
      "grad_norm": 1.171875,
      "learning_rate": 0.00044970829848684,
      "loss": 1.4854,
      "step": 30933
    },
    {
      "epoch": 2.661447130689151,
      "grad_norm": 0.875,
      "learning_rate": 0.000449681394958444,
      "loss": 0.6376,
      "step": 30934
    },
    {
      "epoch": 2.6615331669964726,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004496544915772201,
      "loss": 1.0444,
      "step": 30935
    },
    {
      "epoch": 2.661619203303794,
      "grad_norm": 1.125,
      "learning_rate": 0.0004496275883432469,
      "loss": 1.1255,
      "step": 30936
    },
    {
      "epoch": 2.661705239611116,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00044960068525660323,
      "loss": 1.1088,
      "step": 30937
    },
    {
      "epoch": 2.6617912759184374,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004495737823173678,
      "loss": 1.2947,
      "step": 30938
    },
    {
      "epoch": 2.6618773122257595,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00044954687952561927,
      "loss": 1.1704,
      "step": 30939
    },
    {
      "epoch": 2.6619633485330807,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00044951997688143637,
      "loss": 1.3002,
      "step": 30940
    },
    {
      "epoch": 2.6620493848404028,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004494930743848976,
      "loss": 0.9205,
      "step": 30941
    },
    {
      "epoch": 2.6621354211477244,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004494661720360818,
      "loss": 1.224,
      "step": 30942
    },
    {
      "epoch": 2.662221457455046,
      "grad_norm": 1.953125,
      "learning_rate": 0.00044943926983506775,
      "loss": 1.578,
      "step": 30943
    },
    {
      "epoch": 2.6623074937623676,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004494123677819339,
      "loss": 1.2028,
      "step": 30944
    },
    {
      "epoch": 2.6623935300696893,
      "grad_norm": 1.15625,
      "learning_rate": 0.00044938546587675894,
      "loss": 1.4694,
      "step": 30945
    },
    {
      "epoch": 2.6624795663770113,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004493585641196218,
      "loss": 0.9013,
      "step": 30946
    },
    {
      "epoch": 2.662565602684333,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000449331662510601,
      "loss": 0.8144,
      "step": 30947
    },
    {
      "epoch": 2.6626516389916546,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004493047610497751,
      "loss": 0.9044,
      "step": 30948
    },
    {
      "epoch": 2.662737675298976,
      "grad_norm": 2.921875,
      "learning_rate": 0.000449277859737223,
      "loss": 1.1744,
      "step": 30949
    },
    {
      "epoch": 2.662823711606298,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00044925095857302336,
      "loss": 1.0119,
      "step": 30950
    },
    {
      "epoch": 2.6629097479136195,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004492240575572547,
      "loss": 1.4115,
      "step": 30951
    },
    {
      "epoch": 2.662995784220941,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00044919715668999575,
      "loss": 1.2877,
      "step": 30952
    },
    {
      "epoch": 2.663081820528263,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004491702559713254,
      "loss": 1.4152,
      "step": 30953
    },
    {
      "epoch": 2.6631678568355848,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004491433554013221,
      "loss": 0.8935,
      "step": 30954
    },
    {
      "epoch": 2.6632538931429064,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004491164549800646,
      "loss": 1.2239,
      "step": 30955
    },
    {
      "epoch": 2.663339929450228,
      "grad_norm": 1.21875,
      "learning_rate": 0.00044908955470763145,
      "loss": 1.1545,
      "step": 30956
    },
    {
      "epoch": 2.6634259657575496,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00044906265458410157,
      "loss": 0.7076,
      "step": 30957
    },
    {
      "epoch": 2.6635120020648713,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004490357546095535,
      "loss": 1.7692,
      "step": 30958
    },
    {
      "epoch": 2.663598038372193,
      "grad_norm": 1.515625,
      "learning_rate": 0.0004490088547840659,
      "loss": 1.3638,
      "step": 30959
    },
    {
      "epoch": 2.663684074679515,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004489819551077175,
      "loss": 1.0231,
      "step": 30960
    },
    {
      "epoch": 2.6637701109868366,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000448955055580587,
      "loss": 1.8419,
      "step": 30961
    },
    {
      "epoch": 2.663856147294158,
      "grad_norm": 1.125,
      "learning_rate": 0.00044892815620275307,
      "loss": 1.6883,
      "step": 30962
    },
    {
      "epoch": 2.66394218360148,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004489012569742942,
      "loss": 1.2935,
      "step": 30963
    },
    {
      "epoch": 2.6640282199088015,
      "grad_norm": 1.109375,
      "learning_rate": 0.00044887435789528934,
      "loss": 1.4809,
      "step": 30964
    },
    {
      "epoch": 2.664114256216123,
      "grad_norm": 0.93359375,
      "learning_rate": 0.000448847458965817,
      "loss": 0.8575,
      "step": 30965
    },
    {
      "epoch": 2.6642002925234447,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00044882056018595584,
      "loss": 2.0155,
      "step": 30966
    },
    {
      "epoch": 2.6642863288307668,
      "grad_norm": 1.03125,
      "learning_rate": 0.00044879366155578467,
      "loss": 0.8403,
      "step": 30967
    },
    {
      "epoch": 2.6643723651380884,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004487667630753821,
      "loss": 0.958,
      "step": 30968
    },
    {
      "epoch": 2.66445840144541,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00044873986474482677,
      "loss": 1.0001,
      "step": 30969
    },
    {
      "epoch": 2.6645444377527316,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00044871296656419726,
      "loss": 1.5545,
      "step": 30970
    },
    {
      "epoch": 2.6646304740600533,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004486860685335725,
      "loss": 1.01,
      "step": 30971
    },
    {
      "epoch": 2.664716510367375,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004486591706530309,
      "loss": 1.2389,
      "step": 30972
    },
    {
      "epoch": 2.6648025466746965,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004486322729226513,
      "loss": 0.8235,
      "step": 30973
    },
    {
      "epoch": 2.6648885829820186,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004486053753425121,
      "loss": 1.6789,
      "step": 30974
    },
    {
      "epoch": 2.66497461928934,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004485784779126923,
      "loss": 0.9512,
      "step": 30975
    },
    {
      "epoch": 2.665060655596662,
      "grad_norm": 1.171875,
      "learning_rate": 0.00044855158063327057,
      "loss": 0.8949,
      "step": 30976
    },
    {
      "epoch": 2.6651466919039835,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004485246835043254,
      "loss": 0.8151,
      "step": 30977
    },
    {
      "epoch": 2.665232728211305,
      "grad_norm": 1.046875,
      "learning_rate": 0.00044849778652593554,
      "loss": 1.7196,
      "step": 30978
    },
    {
      "epoch": 2.6653187645186267,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00044847088969817957,
      "loss": 0.7635,
      "step": 30979
    },
    {
      "epoch": 2.6654048008259483,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004484439930211363,
      "loss": 0.5958,
      "step": 30980
    },
    {
      "epoch": 2.6654908371332704,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00044841709649488435,
      "loss": 0.7822,
      "step": 30981
    },
    {
      "epoch": 2.665576873440592,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004483902001195022,
      "loss": 0.8822,
      "step": 30982
    },
    {
      "epoch": 2.6656629097479136,
      "grad_norm": 1.078125,
      "learning_rate": 0.00044836330389506885,
      "loss": 0.6944,
      "step": 30983
    },
    {
      "epoch": 2.6657489460552353,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00044833640782166275,
      "loss": 1.5132,
      "step": 30984
    },
    {
      "epoch": 2.665834982362557,
      "grad_norm": 1.421875,
      "learning_rate": 0.00044830951189936267,
      "loss": 1.7262,
      "step": 30985
    },
    {
      "epoch": 2.665921018669879,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00044828261612824704,
      "loss": 1.3497,
      "step": 30986
    },
    {
      "epoch": 2.6660070549772,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004482557205083948,
      "loss": 1.8026,
      "step": 30987
    },
    {
      "epoch": 2.666093091284522,
      "grad_norm": 1.375,
      "learning_rate": 0.00044822882503988456,
      "loss": 0.628,
      "step": 30988
    },
    {
      "epoch": 2.666179127591844,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00044820192972279486,
      "loss": 1.1538,
      "step": 30989
    },
    {
      "epoch": 2.6662651638991655,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004481750345572044,
      "loss": 0.9628,
      "step": 30990
    },
    {
      "epoch": 2.666351200206487,
      "grad_norm": 1.203125,
      "learning_rate": 0.00044814813954319204,
      "loss": 1.1179,
      "step": 30991
    },
    {
      "epoch": 2.6664372365138087,
      "grad_norm": 0.90625,
      "learning_rate": 0.00044812124468083625,
      "loss": 0.8815,
      "step": 30992
    },
    {
      "epoch": 2.6665232728211308,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004480943499702156,
      "loss": 1.0481,
      "step": 30993
    },
    {
      "epoch": 2.666609309128452,
      "grad_norm": 1.015625,
      "learning_rate": 0.00044806745541140896,
      "loss": 1.3947,
      "step": 30994
    },
    {
      "epoch": 2.666695345435774,
      "grad_norm": 0.84375,
      "learning_rate": 0.00044804056100449495,
      "loss": 0.6167,
      "step": 30995
    },
    {
      "epoch": 2.6667813817430956,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004480136667495522,
      "loss": 1.1185,
      "step": 30996
    },
    {
      "epoch": 2.6668674180504173,
      "grad_norm": 0.890625,
      "learning_rate": 0.00044798677264665926,
      "loss": 1.6623,
      "step": 30997
    },
    {
      "epoch": 2.666953454357739,
      "grad_norm": 1.0625,
      "learning_rate": 0.000447959878695895,
      "loss": 1.289,
      "step": 30998
    },
    {
      "epoch": 2.6670394906650605,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004479329848973379,
      "loss": 1.3461,
      "step": 30999
    },
    {
      "epoch": 2.6671255269723826,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004479060912510667,
      "loss": 0.5575,
      "step": 31000
    },
    {
      "epoch": 2.6672115632797038,
      "grad_norm": 1.0,
      "learning_rate": 0.0004478791977571601,
      "loss": 1.2283,
      "step": 31001
    },
    {
      "epoch": 2.667297599587026,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004478523044156967,
      "loss": 1.2866,
      "step": 31002
    },
    {
      "epoch": 2.6673836358943475,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004478254112267552,
      "loss": 1.5537,
      "step": 31003
    },
    {
      "epoch": 2.667469672201669,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00044779851819041404,
      "loss": 1.2978,
      "step": 31004
    },
    {
      "epoch": 2.6675557085089907,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00044777162530675225,
      "loss": 1.4082,
      "step": 31005
    },
    {
      "epoch": 2.6676417448163123,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004477447325758483,
      "loss": 1.7082,
      "step": 31006
    },
    {
      "epoch": 2.6677277811236344,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00044771783999778073,
      "loss": 1.558,
      "step": 31007
    },
    {
      "epoch": 2.667813817430956,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004476909475726284,
      "loss": 0.6729,
      "step": 31008
    },
    {
      "epoch": 2.6678998537382776,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00044766405530046984,
      "loss": 1.6597,
      "step": 31009
    },
    {
      "epoch": 2.6679858900455993,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004476371631813838,
      "loss": 0.7285,
      "step": 31010
    },
    {
      "epoch": 2.668071926352921,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004476102712154486,
      "loss": 1.1723,
      "step": 31011
    },
    {
      "epoch": 2.6681579626602425,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004475833794027433,
      "loss": 0.976,
      "step": 31012
    },
    {
      "epoch": 2.668243998967564,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004475564877433466,
      "loss": 1.004,
      "step": 31013
    },
    {
      "epoch": 2.668330035274886,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00044752959623733685,
      "loss": 1.1659,
      "step": 31014
    },
    {
      "epoch": 2.668416071582208,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004475027048847929,
      "loss": 1.199,
      "step": 31015
    },
    {
      "epoch": 2.6685021078895295,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004474758136857932,
      "loss": 0.7173,
      "step": 31016
    },
    {
      "epoch": 2.668588144196851,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004474489226404166,
      "loss": 0.9385,
      "step": 31017
    },
    {
      "epoch": 2.6686741805041727,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0004474220317487418,
      "loss": 1.2217,
      "step": 31018
    },
    {
      "epoch": 2.6687602168114943,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004473951410108472,
      "loss": 1.133,
      "step": 31019
    },
    {
      "epoch": 2.668846253118816,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004473682504268116,
      "loss": 1.4983,
      "step": 31020
    },
    {
      "epoch": 2.668932289426138,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00044734135999671367,
      "loss": 1.437,
      "step": 31021
    },
    {
      "epoch": 2.6690183257334597,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000447314469720632,
      "loss": 0.8882,
      "step": 31022
    },
    {
      "epoch": 2.6691043620407813,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004472875795986451,
      "loss": 1.3432,
      "step": 31023
    },
    {
      "epoch": 2.669190398348103,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000447260689630832,
      "loss": 1.7428,
      "step": 31024
    },
    {
      "epoch": 2.6692764346554245,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00044723379981727107,
      "loss": 1.1436,
      "step": 31025
    },
    {
      "epoch": 2.669362470962746,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00044720691015804107,
      "loss": 1.0727,
      "step": 31026
    },
    {
      "epoch": 2.6694485072700678,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0004471800206532204,
      "loss": 0.6946,
      "step": 31027
    },
    {
      "epoch": 2.66953454357739,
      "grad_norm": 0.796875,
      "learning_rate": 0.00044715313130288804,
      "loss": 0.6377,
      "step": 31028
    },
    {
      "epoch": 2.6696205798847115,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004471262421071225,
      "loss": 1.0582,
      "step": 31029
    },
    {
      "epoch": 2.669706616192033,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004470993530660023,
      "loss": 1.1423,
      "step": 31030
    },
    {
      "epoch": 2.6697926524993547,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00044707246417960634,
      "loss": 0.9378,
      "step": 31031
    },
    {
      "epoch": 2.6698786888066763,
      "grad_norm": 1.390625,
      "learning_rate": 0.0004470455754480131,
      "loss": 1.5588,
      "step": 31032
    },
    {
      "epoch": 2.669964725113998,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00044701868687130125,
      "loss": 1.192,
      "step": 31033
    },
    {
      "epoch": 2.6700507614213196,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00044699179844954936,
      "loss": 1.0522,
      "step": 31034
    },
    {
      "epoch": 2.6701367977286417,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004469649101828362,
      "loss": 2.0558,
      "step": 31035
    },
    {
      "epoch": 2.6702228340359633,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004469380220712404,
      "loss": 1.2004,
      "step": 31036
    },
    {
      "epoch": 2.670308870343285,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00044691113411484047,
      "loss": 1.1601,
      "step": 31037
    },
    {
      "epoch": 2.6703949066506065,
      "grad_norm": 1.140625,
      "learning_rate": 0.00044688424631371523,
      "loss": 1.2166,
      "step": 31038
    },
    {
      "epoch": 2.670480942957928,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00044685735866794327,
      "loss": 1.5558,
      "step": 31039
    },
    {
      "epoch": 2.6705669792652498,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004468304711776032,
      "loss": 1.5074,
      "step": 31040
    },
    {
      "epoch": 2.6706530155725714,
      "grad_norm": 0.921875,
      "learning_rate": 0.00044680358384277346,
      "loss": 1.6106,
      "step": 31041
    },
    {
      "epoch": 2.6707390518798935,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004467766966635331,
      "loss": 1.4182,
      "step": 31042
    },
    {
      "epoch": 2.670825088187215,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004467498096399605,
      "loss": 1.4349,
      "step": 31043
    },
    {
      "epoch": 2.6709111244945367,
      "grad_norm": 0.875,
      "learning_rate": 0.0004467229227721344,
      "loss": 0.8509,
      "step": 31044
    },
    {
      "epoch": 2.6709971608018583,
      "grad_norm": 1.203125,
      "learning_rate": 0.00044669603606013324,
      "loss": 1.7318,
      "step": 31045
    },
    {
      "epoch": 2.67108319710918,
      "grad_norm": 1.484375,
      "learning_rate": 0.0004466691495040359,
      "loss": 1.7304,
      "step": 31046
    },
    {
      "epoch": 2.671169233416502,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00044664226310392073,
      "loss": 0.9765,
      "step": 31047
    },
    {
      "epoch": 2.671255269723823,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004466153768598668,
      "loss": 1.3576,
      "step": 31048
    },
    {
      "epoch": 2.6713413060311453,
      "grad_norm": 1.125,
      "learning_rate": 0.0004465884907719525,
      "loss": 1.2467,
      "step": 31049
    },
    {
      "epoch": 2.671427342338467,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004465616048402564,
      "loss": 0.6558,
      "step": 31050
    },
    {
      "epoch": 2.6715133786457885,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00044653471906485725,
      "loss": 1.3894,
      "step": 31051
    },
    {
      "epoch": 2.67159941495311,
      "grad_norm": 1.125,
      "learning_rate": 0.0004465078334458337,
      "loss": 1.3109,
      "step": 31052
    },
    {
      "epoch": 2.6716854512604318,
      "grad_norm": 1.28125,
      "learning_rate": 0.00044648094798326423,
      "loss": 1.0923,
      "step": 31053
    },
    {
      "epoch": 2.671771487567754,
      "grad_norm": 1.203125,
      "learning_rate": 0.00044645406267722766,
      "loss": 1.0011,
      "step": 31054
    },
    {
      "epoch": 2.671857523875075,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004464271775278025,
      "loss": 1.1704,
      "step": 31055
    },
    {
      "epoch": 2.671943560182397,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004464002925350675,
      "loss": 1.3965,
      "step": 31056
    },
    {
      "epoch": 2.6720295964897187,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004463734076991011,
      "loss": 0.7039,
      "step": 31057
    },
    {
      "epoch": 2.6721156327970403,
      "grad_norm": 1.25,
      "learning_rate": 0.0004463465230199821,
      "loss": 1.9195,
      "step": 31058
    },
    {
      "epoch": 2.672201669104362,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004463196384977892,
      "loss": 1.4488,
      "step": 31059
    },
    {
      "epoch": 2.6722877054116836,
      "grad_norm": 1.359375,
      "learning_rate": 0.00044629275413260086,
      "loss": 1.447,
      "step": 31060
    },
    {
      "epoch": 2.6723737417190057,
      "grad_norm": 1.09375,
      "learning_rate": 0.00044626586992449566,
      "loss": 1.4167,
      "step": 31061
    },
    {
      "epoch": 2.672459778026327,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004462389858735524,
      "loss": 1.5347,
      "step": 31062
    },
    {
      "epoch": 2.672545814333649,
      "grad_norm": 1.3125,
      "learning_rate": 0.00044621210197984967,
      "loss": 2.3513,
      "step": 31063
    },
    {
      "epoch": 2.6726318506409705,
      "grad_norm": 1.21875,
      "learning_rate": 0.00044618521824346604,
      "loss": 1.3574,
      "step": 31064
    },
    {
      "epoch": 2.672717886948292,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004461583346644802,
      "loss": 1.1515,
      "step": 31065
    },
    {
      "epoch": 2.672803923255614,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00044613145124297083,
      "loss": 0.6755,
      "step": 31066
    },
    {
      "epoch": 2.6728899595629354,
      "grad_norm": 2.1875,
      "learning_rate": 0.0004461045679790164,
      "loss": 1.9192,
      "step": 31067
    },
    {
      "epoch": 2.6729759958702575,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00044607768487269563,
      "loss": 1.0465,
      "step": 31068
    },
    {
      "epoch": 2.673062032177579,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004460508019240872,
      "loss": 1.4074,
      "step": 31069
    },
    {
      "epoch": 2.6731480684849007,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0004460239191332697,
      "loss": 1.2241,
      "step": 31070
    },
    {
      "epoch": 2.6732341047922223,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004459970365003216,
      "loss": 1.0189,
      "step": 31071
    },
    {
      "epoch": 2.673320141099544,
      "grad_norm": 1.046875,
      "learning_rate": 0.00044597015402532175,
      "loss": 1.41,
      "step": 31072
    },
    {
      "epoch": 2.6734061774068656,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00044594327170834877,
      "loss": 1.3456,
      "step": 31073
    },
    {
      "epoch": 2.673492213714187,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004459163895494811,
      "loss": 0.9496,
      "step": 31074
    },
    {
      "epoch": 2.6735782500215093,
      "grad_norm": 1.1875,
      "learning_rate": 0.00044588950754879746,
      "loss": 1.0686,
      "step": 31075
    },
    {
      "epoch": 2.673664286328831,
      "grad_norm": 1.125,
      "learning_rate": 0.0004458626257063765,
      "loss": 1.02,
      "step": 31076
    },
    {
      "epoch": 2.6737503226361525,
      "grad_norm": 1.046875,
      "learning_rate": 0.00044583574402229695,
      "loss": 1.0281,
      "step": 31077
    },
    {
      "epoch": 2.673836358943474,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00044580886249663716,
      "loss": 0.8279,
      "step": 31078
    },
    {
      "epoch": 2.673922395250796,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00044578198112947597,
      "loss": 1.3672,
      "step": 31079
    },
    {
      "epoch": 2.6740084315581174,
      "grad_norm": 1.03125,
      "learning_rate": 0.00044575509992089197,
      "loss": 1.6424,
      "step": 31080
    },
    {
      "epoch": 2.674094467865439,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004457282188709638,
      "loss": 1.0927,
      "step": 31081
    },
    {
      "epoch": 2.674180504172761,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004457013379797699,
      "loss": 0.9048,
      "step": 31082
    },
    {
      "epoch": 2.6742665404800827,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004456744572473891,
      "loss": 1.4337,
      "step": 31083
    },
    {
      "epoch": 2.6743525767874043,
      "grad_norm": 1.0625,
      "learning_rate": 0.00044564757667389974,
      "loss": 1.3043,
      "step": 31084
    },
    {
      "epoch": 2.674438613094726,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00044562069625938086,
      "loss": 0.4542,
      "step": 31085
    },
    {
      "epoch": 2.6745246494020476,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00044559381600391086,
      "loss": 0.6596,
      "step": 31086
    },
    {
      "epoch": 2.674610685709369,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004455669359075683,
      "loss": 1.4162,
      "step": 31087
    },
    {
      "epoch": 2.674696722016691,
      "grad_norm": 0.703125,
      "learning_rate": 0.000445540055970432,
      "loss": 0.5564,
      "step": 31088
    },
    {
      "epoch": 2.674782758324013,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00044551317619258036,
      "loss": 1.1689,
      "step": 31089
    },
    {
      "epoch": 2.6748687946313345,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00044548629657409214,
      "loss": 1.3051,
      "step": 31090
    },
    {
      "epoch": 2.674954830938656,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00044545941711504573,
      "loss": 1.1626,
      "step": 31091
    },
    {
      "epoch": 2.675040867245978,
      "grad_norm": 1.515625,
      "learning_rate": 0.0004454325378155201,
      "loss": 0.9516,
      "step": 31092
    },
    {
      "epoch": 2.6751269035532994,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004454056586755936,
      "loss": 1.8685,
      "step": 31093
    },
    {
      "epoch": 2.675212939860621,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004453787796953449,
      "loss": 1.3604,
      "step": 31094
    },
    {
      "epoch": 2.6752989761679427,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004453519008748527,
      "loss": 1.054,
      "step": 31095
    },
    {
      "epoch": 2.6753850124752647,
      "grad_norm": 1.890625,
      "learning_rate": 0.0004453250222141956,
      "loss": 1.1144,
      "step": 31096
    },
    {
      "epoch": 2.6754710487825863,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00044529814371345217,
      "loss": 0.7793,
      "step": 31097
    },
    {
      "epoch": 2.675557085089908,
      "grad_norm": 1.296875,
      "learning_rate": 0.00044527126537270093,
      "loss": 1.5063,
      "step": 31098
    },
    {
      "epoch": 2.6756431213972296,
      "grad_norm": 1.203125,
      "learning_rate": 0.00044524438719202065,
      "loss": 1.5209,
      "step": 31099
    },
    {
      "epoch": 2.675729157704551,
      "grad_norm": 1.34375,
      "learning_rate": 0.00044521750917148994,
      "loss": 1.0997,
      "step": 31100
    },
    {
      "epoch": 2.675815194011873,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00044519063131118723,
      "loss": 1.4456,
      "step": 31101
    },
    {
      "epoch": 2.6759012303191945,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004451637536111914,
      "loss": 0.9265,
      "step": 31102
    },
    {
      "epoch": 2.6759872666265165,
      "grad_norm": 1.125,
      "learning_rate": 0.00044513687607158095,
      "loss": 0.8553,
      "step": 31103
    },
    {
      "epoch": 2.676073302933838,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004451099986924344,
      "loss": 0.7155,
      "step": 31104
    },
    {
      "epoch": 2.67615933924116,
      "grad_norm": 0.875,
      "learning_rate": 0.0004450831214738303,
      "loss": 0.6992,
      "step": 31105
    },
    {
      "epoch": 2.6762453755484814,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004450562444158476,
      "loss": 1.1852,
      "step": 31106
    },
    {
      "epoch": 2.676331411855803,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004450293675185646,
      "loss": 1.7992,
      "step": 31107
    },
    {
      "epoch": 2.676417448163125,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00044500249078206007,
      "loss": 0.9002,
      "step": 31108
    },
    {
      "epoch": 2.6765034844704463,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004449756142064124,
      "loss": 1.937,
      "step": 31109
    },
    {
      "epoch": 2.6765895207777683,
      "grad_norm": 1.390625,
      "learning_rate": 0.00044494873779170053,
      "loss": 1.4981,
      "step": 31110
    },
    {
      "epoch": 2.67667555708509,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00044492186153800286,
      "loss": 1.3858,
      "step": 31111
    },
    {
      "epoch": 2.6767615933924116,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004448949854453979,
      "loss": 1.4967,
      "step": 31112
    },
    {
      "epoch": 2.676847629699733,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004448681095139645,
      "loss": 0.8421,
      "step": 31113
    },
    {
      "epoch": 2.676933666007055,
      "grad_norm": 1.015625,
      "learning_rate": 0.00044484123374378124,
      "loss": 1.2024,
      "step": 31114
    },
    {
      "epoch": 2.677019702314377,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00044481435813492653,
      "loss": 0.6671,
      "step": 31115
    },
    {
      "epoch": 2.677105738621698,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00044478748268747905,
      "loss": 1.4728,
      "step": 31116
    },
    {
      "epoch": 2.67719177492902,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00044476060740151753,
      "loss": 0.7339,
      "step": 31117
    },
    {
      "epoch": 2.677277811236342,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004447337322771205,
      "loss": 0.7898,
      "step": 31118
    },
    {
      "epoch": 2.6773638475436634,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004447068573143664,
      "loss": 0.9586,
      "step": 31119
    },
    {
      "epoch": 2.677449883850985,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004446799825133342,
      "loss": 0.7638,
      "step": 31120
    },
    {
      "epoch": 2.6775359201583067,
      "grad_norm": 1.453125,
      "learning_rate": 0.000444653107874102,
      "loss": 1.1403,
      "step": 31121
    },
    {
      "epoch": 2.6776219564656287,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00044462623339674897,
      "loss": 0.8015,
      "step": 31122
    },
    {
      "epoch": 2.67770799277295,
      "grad_norm": 0.890625,
      "learning_rate": 0.00044459935908135336,
      "loss": 0.6262,
      "step": 31123
    },
    {
      "epoch": 2.677794029080272,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00044457248492799377,
      "loss": 0.6115,
      "step": 31124
    },
    {
      "epoch": 2.6778800653875936,
      "grad_norm": 0.765625,
      "learning_rate": 0.000444545610936749,
      "loss": 0.8352,
      "step": 31125
    },
    {
      "epoch": 2.6779661016949152,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0004445187371076975,
      "loss": 0.5379,
      "step": 31126
    },
    {
      "epoch": 2.678052138002237,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000444491863440918,
      "loss": 1.0437,
      "step": 31127
    },
    {
      "epoch": 2.6781381743095585,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00044446498993648884,
      "loss": 0.9388,
      "step": 31128
    },
    {
      "epoch": 2.6782242106168805,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00044443811659448885,
      "loss": 1.7747,
      "step": 31129
    },
    {
      "epoch": 2.678310246924202,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00044441124341499657,
      "loss": 1.7048,
      "step": 31130
    },
    {
      "epoch": 2.678396283231524,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00044438437039809065,
      "loss": 1.5164,
      "step": 31131
    },
    {
      "epoch": 2.6784823195388454,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004443574975438495,
      "loss": 0.9152,
      "step": 31132
    },
    {
      "epoch": 2.678568355846167,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00044433062485235193,
      "loss": 1.0443,
      "step": 31133
    },
    {
      "epoch": 2.6786543921534887,
      "grad_norm": 0.921875,
      "learning_rate": 0.00044430375232367645,
      "loss": 1.2308,
      "step": 31134
    },
    {
      "epoch": 2.6787404284608103,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00044427687995790153,
      "loss": 0.6362,
      "step": 31135
    },
    {
      "epoch": 2.6788264647681324,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004442500077551061,
      "loss": 1.3997,
      "step": 31136
    },
    {
      "epoch": 2.678912501075454,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00044422313571536846,
      "loss": 0.5875,
      "step": 31137
    },
    {
      "epoch": 2.6789985373827756,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004441962638387674,
      "loss": 1.2989,
      "step": 31138
    },
    {
      "epoch": 2.6790845736900972,
      "grad_norm": 1.6875,
      "learning_rate": 0.0004441693921253812,
      "loss": 1.1451,
      "step": 31139
    },
    {
      "epoch": 2.679170609997419,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00044414252057528884,
      "loss": 2.1264,
      "step": 31140
    },
    {
      "epoch": 2.6792566463047405,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004441156491885687,
      "loss": 0.799,
      "step": 31141
    },
    {
      "epoch": 2.679342682612062,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004440887779652994,
      "loss": 0.9924,
      "step": 31142
    },
    {
      "epoch": 2.679428718919384,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004440619069055596,
      "loss": 0.7252,
      "step": 31143
    },
    {
      "epoch": 2.679514755226706,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004440350360094278,
      "loss": 1.9494,
      "step": 31144
    },
    {
      "epoch": 2.6796007915340274,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00044400816527698266,
      "loss": 1.2002,
      "step": 31145
    },
    {
      "epoch": 2.679686827841349,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00044398129470830266,
      "loss": 1.1713,
      "step": 31146
    },
    {
      "epoch": 2.6797728641486707,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004439544243034666,
      "loss": 0.6943,
      "step": 31147
    },
    {
      "epoch": 2.6798589004559923,
      "grad_norm": 0.953125,
      "learning_rate": 0.00044392755406255294,
      "loss": 1.1508,
      "step": 31148
    },
    {
      "epoch": 2.679944936763314,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00044390068398564026,
      "loss": 0.9278,
      "step": 31149
    },
    {
      "epoch": 2.680030973070636,
      "grad_norm": 1.125,
      "learning_rate": 0.00044387381407280706,
      "loss": 2.0885,
      "step": 31150
    },
    {
      "epoch": 2.6801170093779576,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004438469443241322,
      "loss": 1.2977,
      "step": 31151
    },
    {
      "epoch": 2.6802030456852792,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000443820074739694,
      "loss": 1.1936,
      "step": 31152
    },
    {
      "epoch": 2.680289081992601,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00044379320531957113,
      "loss": 0.9432,
      "step": 31153
    },
    {
      "epoch": 2.6803751182999225,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0004437663360638423,
      "loss": 0.9524,
      "step": 31154
    },
    {
      "epoch": 2.680461154607244,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000443739466972586,
      "loss": 1.0937,
      "step": 31155
    },
    {
      "epoch": 2.6805471909145657,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0004437125980458808,
      "loss": 0.862,
      "step": 31156
    },
    {
      "epoch": 2.680633227221888,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004436857292838052,
      "loss": 1.6305,
      "step": 31157
    },
    {
      "epoch": 2.6807192635292094,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004436588606864378,
      "loss": 1.9054,
      "step": 31158
    },
    {
      "epoch": 2.680805299836531,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00044363199225385747,
      "loss": 0.9359,
      "step": 31159
    },
    {
      "epoch": 2.6808913361438527,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0004436051239861426,
      "loss": 1.6398,
      "step": 31160
    },
    {
      "epoch": 2.6809773724511743,
      "grad_norm": 1.265625,
      "learning_rate": 0.00044357825588337176,
      "loss": 1.3969,
      "step": 31161
    },
    {
      "epoch": 2.6810634087584964,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004435513879456234,
      "loss": 1.3234,
      "step": 31162
    },
    {
      "epoch": 2.6811494450658175,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004435245201729765,
      "loss": 0.9623,
      "step": 31163
    },
    {
      "epoch": 2.6812354813731396,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004434976525655093,
      "loss": 1.5774,
      "step": 31164
    },
    {
      "epoch": 2.6813215176804612,
      "grad_norm": 0.875,
      "learning_rate": 0.0004434707851233004,
      "loss": 0.7053,
      "step": 31165
    },
    {
      "epoch": 2.681407553987783,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004434439178464285,
      "loss": 1.6052,
      "step": 31166
    },
    {
      "epoch": 2.6814935902951045,
      "grad_norm": 1.578125,
      "learning_rate": 0.00044341705073497225,
      "loss": 1.288,
      "step": 31167
    },
    {
      "epoch": 2.681579626602426,
      "grad_norm": 1.015625,
      "learning_rate": 0.00044339018378901005,
      "loss": 1.7856,
      "step": 31168
    },
    {
      "epoch": 2.681665662909748,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004433633170086204,
      "loss": 0.9527,
      "step": 31169
    },
    {
      "epoch": 2.6817516992170694,
      "grad_norm": 0.859375,
      "learning_rate": 0.00044333645039388225,
      "loss": 0.8038,
      "step": 31170
    },
    {
      "epoch": 2.6818377355243914,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004433095839448739,
      "loss": 1.5073,
      "step": 31171
    },
    {
      "epoch": 2.681923771831713,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000443282717661674,
      "loss": 1.3526,
      "step": 31172
    },
    {
      "epoch": 2.6820098081390347,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000443255851544361,
      "loss": 1.764,
      "step": 31173
    },
    {
      "epoch": 2.6820958444463563,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004432289855930137,
      "loss": 0.7801,
      "step": 31174
    },
    {
      "epoch": 2.682181880753678,
      "grad_norm": 1.265625,
      "learning_rate": 0.00044320211980771064,
      "loss": 1.0147,
      "step": 31175
    },
    {
      "epoch": 2.682267917061,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00044317525418853017,
      "loss": 1.2971,
      "step": 31176
    },
    {
      "epoch": 2.682353953368321,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004431483887355511,
      "loss": 0.7734,
      "step": 31177
    },
    {
      "epoch": 2.6824399896756432,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000443121523448852,
      "loss": 0.6759,
      "step": 31178
    },
    {
      "epoch": 2.682526025982965,
      "grad_norm": 1.75,
      "learning_rate": 0.0004430946583285114,
      "loss": 1.1149,
      "step": 31179
    },
    {
      "epoch": 2.6826120622902865,
      "grad_norm": 1.40625,
      "learning_rate": 0.00044306779337460766,
      "loss": 1.7802,
      "step": 31180
    },
    {
      "epoch": 2.682698098597608,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004430409285872197,
      "loss": 0.6305,
      "step": 31181
    },
    {
      "epoch": 2.6827841349049297,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000443014063966426,
      "loss": 0.6071,
      "step": 31182
    },
    {
      "epoch": 2.682870171212252,
      "grad_norm": 1.09375,
      "learning_rate": 0.00044298719951230494,
      "loss": 1.0925,
      "step": 31183
    },
    {
      "epoch": 2.6829562075195734,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004429603352249353,
      "loss": 0.9209,
      "step": 31184
    },
    {
      "epoch": 2.683042243826895,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00044293347110439563,
      "loss": 1.8033,
      "step": 31185
    },
    {
      "epoch": 2.6831282801342167,
      "grad_norm": 1.671875,
      "learning_rate": 0.00044290660715076447,
      "loss": 1.2849,
      "step": 31186
    },
    {
      "epoch": 2.6832143164415383,
      "grad_norm": 1.140625,
      "learning_rate": 0.00044287974336412023,
      "loss": 1.3833,
      "step": 31187
    },
    {
      "epoch": 2.68330035274886,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00044285287974454174,
      "loss": 1.0041,
      "step": 31188
    },
    {
      "epoch": 2.6833863890561815,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004428260162921075,
      "loss": 1.8702,
      "step": 31189
    },
    {
      "epoch": 2.6834724253635036,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00044279915300689596,
      "loss": 0.5801,
      "step": 31190
    },
    {
      "epoch": 2.6835584616708252,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004427722898889858,
      "loss": 1.4239,
      "step": 31191
    },
    {
      "epoch": 2.683644497978147,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004427454269384556,
      "loss": 1.1681,
      "step": 31192
    },
    {
      "epoch": 2.6837305342854685,
      "grad_norm": 1.3125,
      "learning_rate": 0.00044271856415538385,
      "loss": 1.6779,
      "step": 31193
    },
    {
      "epoch": 2.68381657059279,
      "grad_norm": 1.0,
      "learning_rate": 0.0004426917015398491,
      "loss": 1.0177,
      "step": 31194
    },
    {
      "epoch": 2.6839026069001117,
      "grad_norm": 1.453125,
      "learning_rate": 0.00044266483909193,
      "loss": 1.1495,
      "step": 31195
    },
    {
      "epoch": 2.6839886432074334,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00044263797681170505,
      "loss": 1.1345,
      "step": 31196
    },
    {
      "epoch": 2.6840746795147554,
      "grad_norm": 1.046875,
      "learning_rate": 0.00044261111469925295,
      "loss": 0.9047,
      "step": 31197
    },
    {
      "epoch": 2.684160715822077,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00044258425275465227,
      "loss": 1.231,
      "step": 31198
    },
    {
      "epoch": 2.6842467521293987,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00044255739097798127,
      "loss": 0.8053,
      "step": 31199
    },
    {
      "epoch": 2.6843327884367203,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004425305293693189,
      "loss": 0.9014,
      "step": 31200
    },
    {
      "epoch": 2.684418824744042,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004425036679287435,
      "loss": 1.3888,
      "step": 31201
    },
    {
      "epoch": 2.6845048610513635,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004424768066563337,
      "loss": 0.8358,
      "step": 31202
    },
    {
      "epoch": 2.684590897358685,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000442449945552168,
      "loss": 0.8902,
      "step": 31203
    },
    {
      "epoch": 2.6846769336660072,
      "grad_norm": 1.140625,
      "learning_rate": 0.00044242308461632507,
      "loss": 1.4209,
      "step": 31204
    },
    {
      "epoch": 2.684762969973329,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00044239622384888345,
      "loss": 1.2766,
      "step": 31205
    },
    {
      "epoch": 2.6848490062806505,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004423693632499216,
      "loss": 0.8014,
      "step": 31206
    },
    {
      "epoch": 2.684935042587972,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0004423425028195182,
      "loss": 1.1889,
      "step": 31207
    },
    {
      "epoch": 2.6850210788952937,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004423156425577518,
      "loss": 1.1202,
      "step": 31208
    },
    {
      "epoch": 2.6851071152026154,
      "grad_norm": 1.421875,
      "learning_rate": 0.0004422887824647009,
      "loss": 1.4553,
      "step": 31209
    },
    {
      "epoch": 2.685193151509937,
      "grad_norm": 1.125,
      "learning_rate": 0.000442261922540444,
      "loss": 1.1057,
      "step": 31210
    },
    {
      "epoch": 2.685279187817259,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00044223506278505984,
      "loss": 1.1617,
      "step": 31211
    },
    {
      "epoch": 2.6853652241245807,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00044220820319862685,
      "loss": 0.9004,
      "step": 31212
    },
    {
      "epoch": 2.6854512604319023,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004421813437812236,
      "loss": 1.1665,
      "step": 31213
    },
    {
      "epoch": 2.685537296739224,
      "grad_norm": 0.765625,
      "learning_rate": 0.00044215448453292875,
      "loss": 1.3014,
      "step": 31214
    },
    {
      "epoch": 2.6856233330465455,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00044212762545382074,
      "loss": 1.2479,
      "step": 31215
    },
    {
      "epoch": 2.685709369353867,
      "grad_norm": 1.15625,
      "learning_rate": 0.00044210076654397826,
      "loss": 1.7964,
      "step": 31216
    },
    {
      "epoch": 2.685795405661189,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00044207390780347954,
      "loss": 0.7462,
      "step": 31217
    },
    {
      "epoch": 2.685881441968511,
      "grad_norm": 1.140625,
      "learning_rate": 0.00044204704923240357,
      "loss": 2.1112,
      "step": 31218
    },
    {
      "epoch": 2.6859674782758325,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00044202019083082873,
      "loss": 1.2088,
      "step": 31219
    },
    {
      "epoch": 2.686053514583154,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004419933325988335,
      "loss": 1.3234,
      "step": 31220
    },
    {
      "epoch": 2.6861395508904757,
      "grad_norm": 1.125,
      "learning_rate": 0.00044196647453649643,
      "loss": 0.8864,
      "step": 31221
    },
    {
      "epoch": 2.6862255871977974,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004419396166438962,
      "loss": 1.4934,
      "step": 31222
    },
    {
      "epoch": 2.6863116235051194,
      "grad_norm": 1.390625,
      "learning_rate": 0.0004419127589211113,
      "loss": 1.8879,
      "step": 31223
    },
    {
      "epoch": 2.6863976598124406,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00044188590136822023,
      "loss": 1.5062,
      "step": 31224
    },
    {
      "epoch": 2.6864836961197627,
      "grad_norm": 1.671875,
      "learning_rate": 0.00044185904398530163,
      "loss": 1.0488,
      "step": 31225
    },
    {
      "epoch": 2.6865697324270843,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004418321867724341,
      "loss": 1.6971,
      "step": 31226
    },
    {
      "epoch": 2.686655768734406,
      "grad_norm": 1.125,
      "learning_rate": 0.000441805329729696,
      "loss": 1.3892,
      "step": 31227
    },
    {
      "epoch": 2.6867418050417275,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000441778472857166,
      "loss": 1.0131,
      "step": 31228
    },
    {
      "epoch": 2.686827841349049,
      "grad_norm": 1.5,
      "learning_rate": 0.00044175161615492265,
      "loss": 1.2569,
      "step": 31229
    },
    {
      "epoch": 2.6869138776563712,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00044172475962304453,
      "loss": 2.0026,
      "step": 31230
    },
    {
      "epoch": 2.6869999139636924,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004416979032616101,
      "loss": 1.4305,
      "step": 31231
    },
    {
      "epoch": 2.6870859502710145,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004416710470706979,
      "loss": 1.2374,
      "step": 31232
    },
    {
      "epoch": 2.687171986578336,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00044164419105038655,
      "loss": 1.1795,
      "step": 31233
    },
    {
      "epoch": 2.6872580228856577,
      "grad_norm": 1.109375,
      "learning_rate": 0.00044161733520075474,
      "loss": 1.3772,
      "step": 31234
    },
    {
      "epoch": 2.6873440591929794,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00044159047952188087,
      "loss": 0.4955,
      "step": 31235
    },
    {
      "epoch": 2.687430095500301,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00044156362401384343,
      "loss": 1.0014,
      "step": 31236
    },
    {
      "epoch": 2.687516131807623,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00044153676867672093,
      "loss": 1.0837,
      "step": 31237
    },
    {
      "epoch": 2.6876021681149442,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004415099135105922,
      "loss": 1.3937,
      "step": 31238
    },
    {
      "epoch": 2.6876882044222663,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004414830585155355,
      "loss": 0.8266,
      "step": 31239
    },
    {
      "epoch": 2.687774240729588,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00044145620369162944,
      "loss": 1.0219,
      "step": 31240
    },
    {
      "epoch": 2.6878602770369096,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004414293490389526,
      "loss": 1.9526,
      "step": 31241
    },
    {
      "epoch": 2.687946313344231,
      "grad_norm": 1.3125,
      "learning_rate": 0.00044140249455758366,
      "loss": 1.2724,
      "step": 31242
    },
    {
      "epoch": 2.688032349651553,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004413756402476009,
      "loss": 0.6755,
      "step": 31243
    },
    {
      "epoch": 2.688118385958875,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000441348786109083,
      "loss": 1.0202,
      "step": 31244
    },
    {
      "epoch": 2.6882044222661965,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00044132193214210845,
      "loss": 1.2908,
      "step": 31245
    },
    {
      "epoch": 2.688290458573518,
      "grad_norm": 1.34375,
      "learning_rate": 0.00044129507834675597,
      "loss": 1.5316,
      "step": 31246
    },
    {
      "epoch": 2.6883764948808397,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004412682247231038,
      "loss": 1.1504,
      "step": 31247
    },
    {
      "epoch": 2.6884625311881614,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00044124137127123076,
      "loss": 1.4952,
      "step": 31248
    },
    {
      "epoch": 2.688548567495483,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004412145179912153,
      "loss": 1.3551,
      "step": 31249
    },
    {
      "epoch": 2.6886346038028046,
      "grad_norm": 1.109375,
      "learning_rate": 0.00044118766488313596,
      "loss": 1.2925,
      "step": 31250
    },
    {
      "epoch": 2.6887206401101267,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004411608119470711,
      "loss": 1.2982,
      "step": 31251
    },
    {
      "epoch": 2.6888066764174483,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00044113395918309956,
      "loss": 1.0062,
      "step": 31252
    },
    {
      "epoch": 2.68889271272477,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00044110710659129974,
      "loss": 1.0609,
      "step": 31253
    },
    {
      "epoch": 2.6889787490320916,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004410802541717501,
      "loss": 0.548,
      "step": 31254
    },
    {
      "epoch": 2.689064785339413,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00044105340192452936,
      "loss": 1.2605,
      "step": 31255
    },
    {
      "epoch": 2.689150821646735,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004410265498497159,
      "loss": 1.1311,
      "step": 31256
    },
    {
      "epoch": 2.6892368579540564,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004409996979473884,
      "loss": 1.1639,
      "step": 31257
    },
    {
      "epoch": 2.6893228942613785,
      "grad_norm": 0.859375,
      "learning_rate": 0.00044097284621762514,
      "loss": 0.7599,
      "step": 31258
    },
    {
      "epoch": 2.6894089305687,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004409459946605049,
      "loss": 1.2281,
      "step": 31259
    },
    {
      "epoch": 2.6894949668760217,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00044091914327610616,
      "loss": 1.4712,
      "step": 31260
    },
    {
      "epoch": 2.6895810031833434,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004408922920645075,
      "loss": 1.0763,
      "step": 31261
    },
    {
      "epoch": 2.689667039490665,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004408654410257872,
      "loss": 1.6775,
      "step": 31262
    },
    {
      "epoch": 2.6897530757979866,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00044083859016002413,
      "loss": 1.0544,
      "step": 31263
    },
    {
      "epoch": 2.6898391121053082,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004408117394672967,
      "loss": 1.3694,
      "step": 31264
    },
    {
      "epoch": 2.6899251484126303,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00044078488894768326,
      "loss": 1.1968,
      "step": 31265
    },
    {
      "epoch": 2.690011184719952,
      "grad_norm": 0.890625,
      "learning_rate": 0.0004407580386012626,
      "loss": 0.5972,
      "step": 31266
    },
    {
      "epoch": 2.6900972210272736,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004407311884281132,
      "loss": 2.0364,
      "step": 31267
    },
    {
      "epoch": 2.690183257334595,
      "grad_norm": 1.46875,
      "learning_rate": 0.00044070433842831357,
      "loss": 1.6286,
      "step": 31268
    },
    {
      "epoch": 2.690269293641917,
      "grad_norm": 0.953125,
      "learning_rate": 0.00044067748860194194,
      "loss": 1.2252,
      "step": 31269
    },
    {
      "epoch": 2.6903553299492384,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004406506389490773,
      "loss": 0.9647,
      "step": 31270
    },
    {
      "epoch": 2.69044136625656,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004406237894697981,
      "loss": 1.8072,
      "step": 31271
    },
    {
      "epoch": 2.690527402563882,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004405969401641827,
      "loss": 1.0179,
      "step": 31272
    },
    {
      "epoch": 2.6906134388712037,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004405700910323097,
      "loss": 0.7241,
      "step": 31273
    },
    {
      "epoch": 2.6906994751785254,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00044054324207425756,
      "loss": 0.6153,
      "step": 31274
    },
    {
      "epoch": 2.690785511485847,
      "grad_norm": 1.015625,
      "learning_rate": 0.000440516393290105,
      "loss": 1.1992,
      "step": 31275
    },
    {
      "epoch": 2.6908715477931686,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00044048954467993035,
      "loss": 1.0387,
      "step": 31276
    },
    {
      "epoch": 2.6909575841004902,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00044046269624381214,
      "loss": 1.3089,
      "step": 31277
    },
    {
      "epoch": 2.691043620407812,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00044043584798182905,
      "loss": 1.4456,
      "step": 31278
    },
    {
      "epoch": 2.691129656715134,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004404089998940596,
      "loss": 1.3613,
      "step": 31279
    },
    {
      "epoch": 2.6912156930224556,
      "grad_norm": 1.375,
      "learning_rate": 0.0004403821519805821,
      "loss": 1.3886,
      "step": 31280
    },
    {
      "epoch": 2.691301729329777,
      "grad_norm": 1.46875,
      "learning_rate": 0.00044035530424147524,
      "loss": 1.3059,
      "step": 31281
    },
    {
      "epoch": 2.691387765637099,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00044032845667681754,
      "loss": 0.7615,
      "step": 31282
    },
    {
      "epoch": 2.6914738019444204,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004403016092866875,
      "loss": 1.5411,
      "step": 31283
    },
    {
      "epoch": 2.6915598382517425,
      "grad_norm": 1.125,
      "learning_rate": 0.0004402747620711637,
      "loss": 0.6258,
      "step": 31284
    },
    {
      "epoch": 2.6916458745590637,
      "grad_norm": 1.046875,
      "learning_rate": 0.00044024791503032446,
      "loss": 1.0388,
      "step": 31285
    },
    {
      "epoch": 2.6917319108663857,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0004402210681642486,
      "loss": 0.6073,
      "step": 31286
    },
    {
      "epoch": 2.6918179471737074,
      "grad_norm": 1.296875,
      "learning_rate": 0.00044019422147301433,
      "loss": 0.8856,
      "step": 31287
    },
    {
      "epoch": 2.691903983481029,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004401673749567004,
      "loss": 1.176,
      "step": 31288
    },
    {
      "epoch": 2.6919900197883506,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004401405286153853,
      "loss": 1.3213,
      "step": 31289
    },
    {
      "epoch": 2.6920760560956722,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004401136824491475,
      "loss": 1.5796,
      "step": 31290
    },
    {
      "epoch": 2.6921620924029943,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004400868364580656,
      "loss": 1.1335,
      "step": 31291
    },
    {
      "epoch": 2.6922481287103155,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00044005999064221786,
      "loss": 1.377,
      "step": 31292
    },
    {
      "epoch": 2.6923341650176376,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004400331450016831,
      "loss": 1.8266,
      "step": 31293
    },
    {
      "epoch": 2.692420201324959,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00044000629953653983,
      "loss": 0.8202,
      "step": 31294
    },
    {
      "epoch": 2.692506237632281,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00043997945424686626,
      "loss": 1.574,
      "step": 31295
    },
    {
      "epoch": 2.6925922739396024,
      "grad_norm": 0.953125,
      "learning_rate": 0.00043995260913274123,
      "loss": 1.2743,
      "step": 31296
    },
    {
      "epoch": 2.692678310246924,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004399257641942431,
      "loss": 1.7402,
      "step": 31297
    },
    {
      "epoch": 2.692764346554246,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00043989891943145055,
      "loss": 1.2507,
      "step": 31298
    },
    {
      "epoch": 2.6928503828615673,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004398720748444418,
      "loss": 1.1656,
      "step": 31299
    },
    {
      "epoch": 2.6929364191688894,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004398452304332956,
      "loss": 0.8818,
      "step": 31300
    },
    {
      "epoch": 2.693022455476211,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00043981838619809046,
      "loss": 1.1225,
      "step": 31301
    },
    {
      "epoch": 2.6931084917835326,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0004397915421389048,
      "loss": 1.3692,
      "step": 31302
    },
    {
      "epoch": 2.6931945280908542,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004397646982558171,
      "loss": 1.736,
      "step": 31303
    },
    {
      "epoch": 2.693280564398176,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000439737854548906,
      "loss": 0.7018,
      "step": 31304
    },
    {
      "epoch": 2.693366600705498,
      "grad_norm": 1.0625,
      "learning_rate": 0.00043971101101825,
      "loss": 1.6453,
      "step": 31305
    },
    {
      "epoch": 2.6934526370128196,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004396841676639273,
      "loss": 1.5425,
      "step": 31306
    },
    {
      "epoch": 2.693538673320141,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000439657324486017,
      "loss": 1.2603,
      "step": 31307
    },
    {
      "epoch": 2.693624709627463,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00043963048148459703,
      "loss": 0.6605,
      "step": 31308
    },
    {
      "epoch": 2.6937107459347844,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004396036386597464,
      "loss": 1.2519,
      "step": 31309
    },
    {
      "epoch": 2.693796782242106,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004395767960115433,
      "loss": 0.8888,
      "step": 31310
    },
    {
      "epoch": 2.6938828185494277,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00043954995354006626,
      "loss": 1.459,
      "step": 31311
    },
    {
      "epoch": 2.6939688548567498,
      "grad_norm": 0.703125,
      "learning_rate": 0.00043952311124539396,
      "loss": 0.7129,
      "step": 31312
    },
    {
      "epoch": 2.6940548911640714,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00043949626912760475,
      "loss": 0.8611,
      "step": 31313
    },
    {
      "epoch": 2.694140927471393,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004394694271867772,
      "loss": 0.7596,
      "step": 31314
    },
    {
      "epoch": 2.6942269637787146,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004394425854229898,
      "loss": 1.3137,
      "step": 31315
    },
    {
      "epoch": 2.6943130000860362,
      "grad_norm": 1.390625,
      "learning_rate": 0.00043941574383632106,
      "loss": 1.7742,
      "step": 31316
    },
    {
      "epoch": 2.694399036393358,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0004393889024268495,
      "loss": 0.5091,
      "step": 31317
    },
    {
      "epoch": 2.6944850727006795,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004393620611946535,
      "loss": 1.0318,
      "step": 31318
    },
    {
      "epoch": 2.6945711090080016,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00043933522013981177,
      "loss": 0.9721,
      "step": 31319
    },
    {
      "epoch": 2.694657145315323,
      "grad_norm": 1.203125,
      "learning_rate": 0.00043930837926240275,
      "loss": 1.4831,
      "step": 31320
    },
    {
      "epoch": 2.694743181622645,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004392815385625049,
      "loss": 1.4943,
      "step": 31321
    },
    {
      "epoch": 2.6948292179299664,
      "grad_norm": 1.09375,
      "learning_rate": 0.00043925469804019664,
      "loss": 1.631,
      "step": 31322
    },
    {
      "epoch": 2.694915254237288,
      "grad_norm": 1.203125,
      "learning_rate": 0.00043922785769555665,
      "loss": 1.1379,
      "step": 31323
    },
    {
      "epoch": 2.6950012905446097,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00043920101752866343,
      "loss": 1.6664,
      "step": 31324
    },
    {
      "epoch": 2.6950873268519313,
      "grad_norm": 1.0625,
      "learning_rate": 0.00043917417753959535,
      "loss": 1.0025,
      "step": 31325
    },
    {
      "epoch": 2.6951733631592534,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004391473377284309,
      "loss": 1.5221,
      "step": 31326
    },
    {
      "epoch": 2.695259399466575,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004391204980952488,
      "loss": 1.2982,
      "step": 31327
    },
    {
      "epoch": 2.6953454357738966,
      "grad_norm": 1.75,
      "learning_rate": 0.0004390936586401273,
      "loss": 1.3704,
      "step": 31328
    },
    {
      "epoch": 2.6954314720812182,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000439066819363145,
      "loss": 0.8795,
      "step": 31329
    },
    {
      "epoch": 2.69551750838854,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004390399802643804,
      "loss": 1.2387,
      "step": 31330
    },
    {
      "epoch": 2.6956035446958615,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004390131413439121,
      "loss": 1.4274,
      "step": 31331
    },
    {
      "epoch": 2.695689581003183,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00043898630260181843,
      "loss": 0.9671,
      "step": 31332
    },
    {
      "epoch": 2.695775617310505,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004389594640381779,
      "loss": 0.838,
      "step": 31333
    },
    {
      "epoch": 2.695861653617827,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004389326256530691,
      "loss": 2.2569,
      "step": 31334
    },
    {
      "epoch": 2.6959476899251484,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004389057874465706,
      "loss": 0.8578,
      "step": 31335
    },
    {
      "epoch": 2.69603372623247,
      "grad_norm": 0.84375,
      "learning_rate": 0.00043887894941876054,
      "loss": 0.9606,
      "step": 31336
    },
    {
      "epoch": 2.6961197625397917,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00043885211156971787,
      "loss": 0.6065,
      "step": 31337
    },
    {
      "epoch": 2.6962057988471138,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00043882527389952085,
      "loss": 1.1744,
      "step": 31338
    },
    {
      "epoch": 2.696291835154435,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004387984364082481,
      "loss": 1.5365,
      "step": 31339
    },
    {
      "epoch": 2.696377871461757,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004387715990959778,
      "loss": 1.4508,
      "step": 31340
    },
    {
      "epoch": 2.6964639077690786,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004387447619627888,
      "loss": 2.0068,
      "step": 31341
    },
    {
      "epoch": 2.6965499440764003,
      "grad_norm": 1.25,
      "learning_rate": 0.00043871792500875946,
      "loss": 1.0666,
      "step": 31342
    },
    {
      "epoch": 2.696635980383722,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004386910882339681,
      "loss": 1.3904,
      "step": 31343
    },
    {
      "epoch": 2.6967220166910435,
      "grad_norm": 1.34375,
      "learning_rate": 0.00043866425163849353,
      "loss": 1.1877,
      "step": 31344
    },
    {
      "epoch": 2.6968080529983656,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00043863741522241404,
      "loss": 1.3909,
      "step": 31345
    },
    {
      "epoch": 2.6968940893056867,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004386105789858083,
      "loss": 0.7825,
      "step": 31346
    },
    {
      "epoch": 2.696980125613009,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004385837429287546,
      "loss": 0.8321,
      "step": 31347
    },
    {
      "epoch": 2.6970661619203304,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004385569070513315,
      "loss": 1.3483,
      "step": 31348
    },
    {
      "epoch": 2.697152198227652,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004385300713536174,
      "loss": 0.6217,
      "step": 31349
    },
    {
      "epoch": 2.6972382345349737,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00043850323583569106,
      "loss": 0.6284,
      "step": 31350
    },
    {
      "epoch": 2.6973242708422953,
      "grad_norm": 1.125,
      "learning_rate": 0.00043847640049763075,
      "loss": 0.9899,
      "step": 31351
    },
    {
      "epoch": 2.6974103071496174,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000438449565339515,
      "loss": 1.2264,
      "step": 31352
    },
    {
      "epoch": 2.6974963434569386,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00043842273036142227,
      "loss": 1.5611,
      "step": 31353
    },
    {
      "epoch": 2.6975823797642606,
      "grad_norm": 1.109375,
      "learning_rate": 0.00043839589556343115,
      "loss": 1.3709,
      "step": 31354
    },
    {
      "epoch": 2.6976684160715823,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00043836906094562,
      "loss": 1.4304,
      "step": 31355
    },
    {
      "epoch": 2.697754452378904,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004383422265080673,
      "loss": 0.9939,
      "step": 31356
    },
    {
      "epoch": 2.6978404886862255,
      "grad_norm": 0.875,
      "learning_rate": 0.00043831539225085164,
      "loss": 0.8551,
      "step": 31357
    },
    {
      "epoch": 2.697926524993547,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004382885581740515,
      "loss": 1.4076,
      "step": 31358
    },
    {
      "epoch": 2.698012561300869,
      "grad_norm": 1.140625,
      "learning_rate": 0.00043826172427774527,
      "loss": 0.9301,
      "step": 31359
    },
    {
      "epoch": 2.698098597608191,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00043823489056201147,
      "loss": 1.3702,
      "step": 31360
    },
    {
      "epoch": 2.6981846339155124,
      "grad_norm": 1.015625,
      "learning_rate": 0.00043820805702692874,
      "loss": 0.5728,
      "step": 31361
    },
    {
      "epoch": 2.698270670222834,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00043818122367257536,
      "loss": 1.215,
      "step": 31362
    },
    {
      "epoch": 2.6983567065301557,
      "grad_norm": 1.125,
      "learning_rate": 0.00043815439049902977,
      "loss": 0.7872,
      "step": 31363
    },
    {
      "epoch": 2.6984427428374773,
      "grad_norm": 1.25,
      "learning_rate": 0.0004381275575063707,
      "loss": 1.5049,
      "step": 31364
    },
    {
      "epoch": 2.698528779144799,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004381007246946764,
      "loss": 2.1124,
      "step": 31365
    },
    {
      "epoch": 2.698614815452121,
      "grad_norm": 1.171875,
      "learning_rate": 0.00043807389206402544,
      "loss": 1.114,
      "step": 31366
    },
    {
      "epoch": 2.6987008517594426,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004380470596144964,
      "loss": 1.588,
      "step": 31367
    },
    {
      "epoch": 2.6987868880667643,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00043802022734616756,
      "loss": 1.8833,
      "step": 31368
    },
    {
      "epoch": 2.698872924374086,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00043799339525911756,
      "loss": 1.6203,
      "step": 31369
    },
    {
      "epoch": 2.6989589606814075,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00043796656335342476,
      "loss": 1.3223,
      "step": 31370
    },
    {
      "epoch": 2.699044996988729,
      "grad_norm": 1.0625,
      "learning_rate": 0.00043793973162916777,
      "loss": 1.9001,
      "step": 31371
    },
    {
      "epoch": 2.6991310332960508,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000437912900086425,
      "loss": 1.5625,
      "step": 31372
    },
    {
      "epoch": 2.699217069603373,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00043788606872527494,
      "loss": 1.1627,
      "step": 31373
    },
    {
      "epoch": 2.6993031059106944,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004378592375457959,
      "loss": 0.9686,
      "step": 31374
    },
    {
      "epoch": 2.699389142218016,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004378324065480666,
      "loss": 1.5779,
      "step": 31375
    },
    {
      "epoch": 2.6994751785253377,
      "grad_norm": 1.046875,
      "learning_rate": 0.00043780557573216543,
      "loss": 1.3217,
      "step": 31376
    },
    {
      "epoch": 2.6995612148326593,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00043777874509817077,
      "loss": 1.1608,
      "step": 31377
    },
    {
      "epoch": 2.699647251139981,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004377519146461613,
      "loss": 1.2545,
      "step": 31378
    },
    {
      "epoch": 2.6997332874473026,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00043772508437621535,
      "loss": 2.0341,
      "step": 31379
    },
    {
      "epoch": 2.6998193237546246,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004376982542884112,
      "loss": 1.082,
      "step": 31380
    },
    {
      "epoch": 2.6999053600619463,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00043767142438282775,
      "loss": 0.7366,
      "step": 31381
    },
    {
      "epoch": 2.699991396369268,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004376445946595431,
      "loss": 0.5838,
      "step": 31382
    },
    {
      "epoch": 2.7000774326765895,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004376177651186361,
      "loss": 1.4084,
      "step": 31383
    },
    {
      "epoch": 2.700163468983911,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000437590935760185,
      "loss": 0.7584,
      "step": 31384
    },
    {
      "epoch": 2.7002495052912328,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004375641065842682,
      "loss": 0.9799,
      "step": 31385
    },
    {
      "epoch": 2.7003355415985544,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00043753727759096415,
      "loss": 1.4292,
      "step": 31386
    },
    {
      "epoch": 2.7004215779058764,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004375104487803516,
      "loss": 1.3477,
      "step": 31387
    },
    {
      "epoch": 2.700507614213198,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00043748362015250873,
      "loss": 1.2146,
      "step": 31388
    },
    {
      "epoch": 2.7005936505205197,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0004374567917075141,
      "loss": 0.4833,
      "step": 31389
    },
    {
      "epoch": 2.7006796868278413,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00043742996344544623,
      "loss": 1.9821,
      "step": 31390
    },
    {
      "epoch": 2.700765723135163,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004374031353663835,
      "loss": 1.1089,
      "step": 31391
    },
    {
      "epoch": 2.7008517594424846,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00043737630747040456,
      "loss": 0.6504,
      "step": 31392
    },
    {
      "epoch": 2.700937795749806,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00043734947975758757,
      "loss": 0.9515,
      "step": 31393
    },
    {
      "epoch": 2.7010238320571283,
      "grad_norm": 1.09375,
      "learning_rate": 0.00043732265222801126,
      "loss": 1.2257,
      "step": 31394
    },
    {
      "epoch": 2.70110986836445,
      "grad_norm": 1.59375,
      "learning_rate": 0.00043729582488175403,
      "loss": 1.7605,
      "step": 31395
    },
    {
      "epoch": 2.7011959046717715,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00043726899771889425,
      "loss": 0.8899,
      "step": 31396
    },
    {
      "epoch": 2.701281940979093,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004372421707395104,
      "loss": 1.6403,
      "step": 31397
    },
    {
      "epoch": 2.7013679772864148,
      "grad_norm": 1.0,
      "learning_rate": 0.0004372153439436811,
      "loss": 1.6067,
      "step": 31398
    },
    {
      "epoch": 2.701454013593737,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004371885173314847,
      "loss": 1.5577,
      "step": 31399
    },
    {
      "epoch": 2.701540049901058,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00043716169090299956,
      "loss": 1.0294,
      "step": 31400
    },
    {
      "epoch": 2.70162608620838,
      "grad_norm": 1.015625,
      "learning_rate": 0.00043713486465830437,
      "loss": 2.2087,
      "step": 31401
    },
    {
      "epoch": 2.7017121225157017,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004371080385974775,
      "loss": 0.9443,
      "step": 31402
    },
    {
      "epoch": 2.7017981588230233,
      "grad_norm": 0.84375,
      "learning_rate": 0.0004370812127205973,
      "loss": 1.1493,
      "step": 31403
    },
    {
      "epoch": 2.701884195130345,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0004370543870277423,
      "loss": 0.7073,
      "step": 31404
    },
    {
      "epoch": 2.7019702314376666,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000437027561518991,
      "loss": 0.8644,
      "step": 31405
    },
    {
      "epoch": 2.7020562677449886,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00043700073619442194,
      "loss": 2.2315,
      "step": 31406
    },
    {
      "epoch": 2.70214230405231,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004369739110541133,
      "loss": 1.0939,
      "step": 31407
    },
    {
      "epoch": 2.702228340359632,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00043694708609814384,
      "loss": 0.7294,
      "step": 31408
    },
    {
      "epoch": 2.7023143766669535,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00043692026132659186,
      "loss": 1.4245,
      "step": 31409
    },
    {
      "epoch": 2.702400412974275,
      "grad_norm": 0.796875,
      "learning_rate": 0.00043689343673953586,
      "loss": 0.4904,
      "step": 31410
    },
    {
      "epoch": 2.7024864492815968,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00043686661233705416,
      "loss": 1.2212,
      "step": 31411
    },
    {
      "epoch": 2.7025724855889184,
      "grad_norm": 1.0,
      "learning_rate": 0.0004368397881192255,
      "loss": 1.3041,
      "step": 31412
    },
    {
      "epoch": 2.7026585218962405,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004368129640861281,
      "loss": 1.0282,
      "step": 31413
    },
    {
      "epoch": 2.7027445582035616,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004367861402378406,
      "loss": 1.1417,
      "step": 31414
    },
    {
      "epoch": 2.7028305945108837,
      "grad_norm": 0.890625,
      "learning_rate": 0.00043675931657444114,
      "loss": 1.3264,
      "step": 31415
    },
    {
      "epoch": 2.7029166308182053,
      "grad_norm": 1.28125,
      "learning_rate": 0.00043673249309600853,
      "loss": 0.8954,
      "step": 31416
    },
    {
      "epoch": 2.703002667125527,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004367056698026208,
      "loss": 0.8256,
      "step": 31417
    },
    {
      "epoch": 2.7030887034328486,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00043667884669435696,
      "loss": 1.4046,
      "step": 31418
    },
    {
      "epoch": 2.70317473974017,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004366520237712952,
      "loss": 1.3645,
      "step": 31419
    },
    {
      "epoch": 2.7032607760474923,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004366252010335138,
      "loss": 0.8304,
      "step": 31420
    },
    {
      "epoch": 2.703346812354814,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0004365983784810915,
      "loss": 0.7467,
      "step": 31421
    },
    {
      "epoch": 2.7034328486621355,
      "grad_norm": 1.140625,
      "learning_rate": 0.00043657155611410655,
      "loss": 1.2625,
      "step": 31422
    },
    {
      "epoch": 2.703518884969457,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004365447339326374,
      "loss": 1.0873,
      "step": 31423
    },
    {
      "epoch": 2.7036049212767788,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0004365179119367626,
      "loss": 1.8625,
      "step": 31424
    },
    {
      "epoch": 2.7036909575841004,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004364910901265606,
      "loss": 1.2218,
      "step": 31425
    },
    {
      "epoch": 2.703776993891422,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00043646426850210983,
      "loss": 0.9314,
      "step": 31426
    },
    {
      "epoch": 2.703863030198744,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00043643744706348863,
      "loss": 1.0594,
      "step": 31427
    },
    {
      "epoch": 2.7039490665060657,
      "grad_norm": 1.109375,
      "learning_rate": 0.00043641062581077563,
      "loss": 0.9935,
      "step": 31428
    },
    {
      "epoch": 2.7040351028133873,
      "grad_norm": 1.03125,
      "learning_rate": 0.00043638380474404916,
      "loss": 1.5868,
      "step": 31429
    },
    {
      "epoch": 2.704121139120709,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004363569838633876,
      "loss": 1.5809,
      "step": 31430
    },
    {
      "epoch": 2.7042071754280306,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004363301631688696,
      "loss": 0.7746,
      "step": 31431
    },
    {
      "epoch": 2.704293211735352,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004363033426605735,
      "loss": 1.6742,
      "step": 31432
    },
    {
      "epoch": 2.704379248042674,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004362765223385777,
      "loss": 0.6769,
      "step": 31433
    },
    {
      "epoch": 2.704465284349996,
      "grad_norm": 1.390625,
      "learning_rate": 0.00043624970220296054,
      "loss": 1.5789,
      "step": 31434
    },
    {
      "epoch": 2.7045513206573175,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004362228822538007,
      "loss": 0.9201,
      "step": 31435
    },
    {
      "epoch": 2.704637356964639,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004361960624911766,
      "loss": 1.3822,
      "step": 31436
    },
    {
      "epoch": 2.7047233932719608,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004361692429151666,
      "loss": 1.0728,
      "step": 31437
    },
    {
      "epoch": 2.7048094295792824,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00043614242352584903,
      "loss": 1.7473,
      "step": 31438
    },
    {
      "epoch": 2.704895465886604,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00043611560432330255,
      "loss": 0.7303,
      "step": 31439
    },
    {
      "epoch": 2.7049815021939256,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00043608878530760546,
      "loss": 0.6049,
      "step": 31440
    },
    {
      "epoch": 2.7050675385012477,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004360619664788362,
      "loss": 1.1058,
      "step": 31441
    },
    {
      "epoch": 2.7051535748085693,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004360351478370734,
      "loss": 1.0516,
      "step": 31442
    },
    {
      "epoch": 2.705239611115891,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00043600832938239526,
      "loss": 0.9977,
      "step": 31443
    },
    {
      "epoch": 2.7053256474232126,
      "grad_norm": 1.15625,
      "learning_rate": 0.00043598151111488034,
      "loss": 1.8738,
      "step": 31444
    },
    {
      "epoch": 2.705411683730534,
      "grad_norm": 2.546875,
      "learning_rate": 0.0004359546930346069,
      "loss": 1.3264,
      "step": 31445
    },
    {
      "epoch": 2.705497720037856,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004359278751416537,
      "loss": 1.0777,
      "step": 31446
    },
    {
      "epoch": 2.7055837563451774,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000435901057436099,
      "loss": 0.9713,
      "step": 31447
    },
    {
      "epoch": 2.7056697926524995,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004358742399180211,
      "loss": 1.0619,
      "step": 31448
    },
    {
      "epoch": 2.705755828959821,
      "grad_norm": 1.125,
      "learning_rate": 0.0004358474225874987,
      "loss": 1.0856,
      "step": 31449
    },
    {
      "epoch": 2.7058418652671428,
      "grad_norm": 1.234375,
      "learning_rate": 0.00043582060544461015,
      "loss": 2.2554,
      "step": 31450
    },
    {
      "epoch": 2.7059279015744644,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004357937884894338,
      "loss": 1.2104,
      "step": 31451
    },
    {
      "epoch": 2.706013937881786,
      "grad_norm": 1.28125,
      "learning_rate": 0.00043576697172204807,
      "loss": 1.0685,
      "step": 31452
    },
    {
      "epoch": 2.7060999741891076,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004357401551425315,
      "loss": 1.5085,
      "step": 31453
    },
    {
      "epoch": 2.7061860104964293,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004357133387509623,
      "loss": 1.6069,
      "step": 31454
    },
    {
      "epoch": 2.7062720468037513,
      "grad_norm": 1.125,
      "learning_rate": 0.00043568652254741925,
      "loss": 1.0242,
      "step": 31455
    },
    {
      "epoch": 2.706358083111073,
      "grad_norm": 1.65625,
      "learning_rate": 0.0004356597065319807,
      "loss": 1.2531,
      "step": 31456
    },
    {
      "epoch": 2.7064441194183946,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004356328907047248,
      "loss": 0.9412,
      "step": 31457
    },
    {
      "epoch": 2.706530155725716,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00043560607506573033,
      "loss": 1.0807,
      "step": 31458
    },
    {
      "epoch": 2.706616192033038,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00043557925961507554,
      "loss": 1.365,
      "step": 31459
    },
    {
      "epoch": 2.70670222834036,
      "grad_norm": 1.109375,
      "learning_rate": 0.00043555244435283894,
      "loss": 0.8271,
      "step": 31460
    },
    {
      "epoch": 2.706788264647681,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00043552562927909875,
      "loss": 0.7977,
      "step": 31461
    },
    {
      "epoch": 2.706874300955003,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004354988143939337,
      "loss": 1.2332,
      "step": 31462
    },
    {
      "epoch": 2.7069603372623248,
      "grad_norm": 1.046875,
      "learning_rate": 0.000435471999697422,
      "loss": 0.7471,
      "step": 31463
    },
    {
      "epoch": 2.7070463735696464,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004354451851896421,
      "loss": 1.3409,
      "step": 31464
    },
    {
      "epoch": 2.707132409876968,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004354183708706726,
      "loss": 1.1515,
      "step": 31465
    },
    {
      "epoch": 2.7072184461842896,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004353915567405917,
      "loss": 1.1639,
      "step": 31466
    },
    {
      "epoch": 2.7073044824916117,
      "grad_norm": 1.125,
      "learning_rate": 0.00043536474279947806,
      "loss": 1.5554,
      "step": 31467
    },
    {
      "epoch": 2.707390518798933,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004353379290474098,
      "loss": 1.0532,
      "step": 31468
    },
    {
      "epoch": 2.707476555106255,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00043531111548446565,
      "loss": 0.7371,
      "step": 31469
    },
    {
      "epoch": 2.7075625914135766,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004352843021107239,
      "loss": 1.1451,
      "step": 31470
    },
    {
      "epoch": 2.707648627720898,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004352574889262629,
      "loss": 0.6582,
      "step": 31471
    },
    {
      "epoch": 2.70773466402822,
      "grad_norm": 1.03125,
      "learning_rate": 0.00043523067593116124,
      "loss": 1.3988,
      "step": 31472
    },
    {
      "epoch": 2.7078207003355415,
      "grad_norm": 1.015625,
      "learning_rate": 0.00043520386312549724,
      "loss": 0.8339,
      "step": 31473
    },
    {
      "epoch": 2.7079067366428635,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004351770505093494,
      "loss": 0.8358,
      "step": 31474
    },
    {
      "epoch": 2.7079927729501847,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00043515023808279586,
      "loss": 0.8157,
      "step": 31475
    },
    {
      "epoch": 2.7080788092575068,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004351234258459155,
      "loss": 0.8262,
      "step": 31476
    },
    {
      "epoch": 2.7081648455648284,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004350966137987864,
      "loss": 1.0747,
      "step": 31477
    },
    {
      "epoch": 2.70825088187215,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00043506980194148707,
      "loss": 1.5728,
      "step": 31478
    },
    {
      "epoch": 2.7083369181794716,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00043504299027409587,
      "loss": 2.0025,
      "step": 31479
    },
    {
      "epoch": 2.7084229544867933,
      "grad_norm": 0.96875,
      "learning_rate": 0.00043501617879669147,
      "loss": 1.3342,
      "step": 31480
    },
    {
      "epoch": 2.7085089907941153,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00043498936750935204,
      "loss": 0.9563,
      "step": 31481
    },
    {
      "epoch": 2.708595027101437,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000434962556412156,
      "loss": 1.8048,
      "step": 31482
    },
    {
      "epoch": 2.7086810634087586,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004349357455051819,
      "loss": 1.4021,
      "step": 31483
    },
    {
      "epoch": 2.70876709971608,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00043490893478850807,
      "loss": 0.8458,
      "step": 31484
    },
    {
      "epoch": 2.708853136023402,
      "grad_norm": 1.546875,
      "learning_rate": 0.000434882124262213,
      "loss": 2.1575,
      "step": 31485
    },
    {
      "epoch": 2.7089391723307235,
      "grad_norm": 1.0,
      "learning_rate": 0.0004348553139263749,
      "loss": 0.7605,
      "step": 31486
    },
    {
      "epoch": 2.709025208638045,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004348285037810724,
      "loss": 1.3038,
      "step": 31487
    },
    {
      "epoch": 2.709111244945367,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00043480169382638395,
      "loss": 1.0516,
      "step": 31488
    },
    {
      "epoch": 2.7091972812526888,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004347748840623877,
      "loss": 1.3372,
      "step": 31489
    },
    {
      "epoch": 2.7092833175600104,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004347480744891624,
      "loss": 1.3242,
      "step": 31490
    },
    {
      "epoch": 2.709369353867332,
      "grad_norm": 1.09375,
      "learning_rate": 0.000434721265106786,
      "loss": 1.162,
      "step": 31491
    },
    {
      "epoch": 2.7094553901746536,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00043469445591533745,
      "loss": 1.2028,
      "step": 31492
    },
    {
      "epoch": 2.7095414264819753,
      "grad_norm": 1.1875,
      "learning_rate": 0.00043466764691489493,
      "loss": 1.8599,
      "step": 31493
    },
    {
      "epoch": 2.709627462789297,
      "grad_norm": 0.671875,
      "learning_rate": 0.00043464083810553666,
      "loss": 0.625,
      "step": 31494
    },
    {
      "epoch": 2.709713499096619,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004346140294873414,
      "loss": 0.9621,
      "step": 31495
    },
    {
      "epoch": 2.7097995354039406,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004345872210603874,
      "loss": 1.9455,
      "step": 31496
    },
    {
      "epoch": 2.709885571711262,
      "grad_norm": 1.078125,
      "learning_rate": 0.000434560412824753,
      "loss": 1.2908,
      "step": 31497
    },
    {
      "epoch": 2.709971608018584,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004345336047805166,
      "loss": 1.2361,
      "step": 31498
    },
    {
      "epoch": 2.7100576443259055,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004345067969277568,
      "loss": 0.7711,
      "step": 31499
    },
    {
      "epoch": 2.710143680633227,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00043447998926655185,
      "loss": 1.4935,
      "step": 31500
    },
    {
      "epoch": 2.7102297169405487,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004344531817969802,
      "loss": 0.6813,
      "step": 31501
    },
    {
      "epoch": 2.7103157532478708,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004344263745191201,
      "loss": 1.1639,
      "step": 31502
    },
    {
      "epoch": 2.7104017895551924,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004343995674330503,
      "loss": 1.4483,
      "step": 31503
    },
    {
      "epoch": 2.710487825862514,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004343727605388489,
      "loss": 1.4278,
      "step": 31504
    },
    {
      "epoch": 2.7105738621698356,
      "grad_norm": 1.125,
      "learning_rate": 0.0004343459538365944,
      "loss": 1.0377,
      "step": 31505
    },
    {
      "epoch": 2.7106598984771573,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00043431914732636526,
      "loss": 0.8815,
      "step": 31506
    },
    {
      "epoch": 2.710745934784479,
      "grad_norm": 1.84375,
      "learning_rate": 0.0004342923410082398,
      "loss": 1.9807,
      "step": 31507
    },
    {
      "epoch": 2.7108319710918005,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004342655348822965,
      "loss": 1.6594,
      "step": 31508
    },
    {
      "epoch": 2.7109180073991226,
      "grad_norm": 1.15625,
      "learning_rate": 0.00043423872894861364,
      "loss": 1.4994,
      "step": 31509
    },
    {
      "epoch": 2.711004043706444,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004342119232072698,
      "loss": 1.3361,
      "step": 31510
    },
    {
      "epoch": 2.711090080013766,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004341851176583433,
      "loss": 1.0041,
      "step": 31511
    },
    {
      "epoch": 2.7111761163210875,
      "grad_norm": 1.8125,
      "learning_rate": 0.00043415831230191244,
      "loss": 1.1573,
      "step": 31512
    },
    {
      "epoch": 2.711262152628409,
      "grad_norm": 1.25,
      "learning_rate": 0.0004341315071380558,
      "loss": 1.1101,
      "step": 31513
    },
    {
      "epoch": 2.711348188935731,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00043410470216685167,
      "loss": 0.591,
      "step": 31514
    },
    {
      "epoch": 2.7114342252430523,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004340778973883785,
      "loss": 1.5906,
      "step": 31515
    },
    {
      "epoch": 2.7115202615503744,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00043405109280271454,
      "loss": 1.0113,
      "step": 31516
    },
    {
      "epoch": 2.711606297857696,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00043402428840993843,
      "loss": 1.5079,
      "step": 31517
    },
    {
      "epoch": 2.7116923341650176,
      "grad_norm": 0.875,
      "learning_rate": 0.0004339974842101284,
      "loss": 1.4585,
      "step": 31518
    },
    {
      "epoch": 2.7117783704723393,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004339706802033628,
      "loss": 0.8903,
      "step": 31519
    },
    {
      "epoch": 2.711864406779661,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004339438763897203,
      "loss": 0.8045,
      "step": 31520
    },
    {
      "epoch": 2.711950443086983,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004339170727692791,
      "loss": 1.227,
      "step": 31521
    },
    {
      "epoch": 2.712036479394304,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004338902693421175,
      "loss": 1.0305,
      "step": 31522
    },
    {
      "epoch": 2.712122515701626,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00043386346610831394,
      "loss": 1.2128,
      "step": 31523
    },
    {
      "epoch": 2.712208552008948,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000433836663067947,
      "loss": 0.7731,
      "step": 31524
    },
    {
      "epoch": 2.7122945883162695,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000433809860221095,
      "loss": 1.526,
      "step": 31525
    },
    {
      "epoch": 2.712380624623591,
      "grad_norm": 1.25,
      "learning_rate": 0.0004337830575678362,
      "loss": 1.3245,
      "step": 31526
    },
    {
      "epoch": 2.7124666609309127,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00043375625510824907,
      "loss": 1.2305,
      "step": 31527
    },
    {
      "epoch": 2.7125526972382348,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00043372945284241184,
      "loss": 1.0845,
      "step": 31528
    },
    {
      "epoch": 2.712638733545556,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0004337026507704033,
      "loss": 1.2637,
      "step": 31529
    },
    {
      "epoch": 2.712724769852878,
      "grad_norm": 1.421875,
      "learning_rate": 0.0004336758488923016,
      "loss": 1.4255,
      "step": 31530
    },
    {
      "epoch": 2.7128108061601997,
      "grad_norm": 1.015625,
      "learning_rate": 0.00043364904720818515,
      "loss": 1.8211,
      "step": 31531
    },
    {
      "epoch": 2.7128968424675213,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00043362224571813225,
      "loss": 1.0634,
      "step": 31532
    },
    {
      "epoch": 2.712982878774843,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004335954444222214,
      "loss": 1.8366,
      "step": 31533
    },
    {
      "epoch": 2.7130689150821645,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000433568643320531,
      "loss": 0.9415,
      "step": 31534
    },
    {
      "epoch": 2.7131549513894866,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004335418424131393,
      "loss": 0.8578,
      "step": 31535
    },
    {
      "epoch": 2.713240987696808,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00043351504170012495,
      "loss": 0.9058,
      "step": 31536
    },
    {
      "epoch": 2.71332702400413,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004334882411815661,
      "loss": 1.0729,
      "step": 31537
    },
    {
      "epoch": 2.7134130603114515,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00043346144085754123,
      "loss": 1.652,
      "step": 31538
    },
    {
      "epoch": 2.713499096618773,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004334346407281286,
      "loss": 0.7084,
      "step": 31539
    },
    {
      "epoch": 2.7135851329260947,
      "grad_norm": 0.875,
      "learning_rate": 0.0004334078407934068,
      "loss": 1.2208,
      "step": 31540
    },
    {
      "epoch": 2.7136711692334163,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00043338104105345414,
      "loss": 1.4187,
      "step": 31541
    },
    {
      "epoch": 2.7137572055407384,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00043335424150834885,
      "loss": 1.3831,
      "step": 31542
    },
    {
      "epoch": 2.71384324184806,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00043332744215816957,
      "loss": 1.0436,
      "step": 31543
    },
    {
      "epoch": 2.7139292781553817,
      "grad_norm": 1.171875,
      "learning_rate": 0.00043330064300299456,
      "loss": 0.9348,
      "step": 31544
    },
    {
      "epoch": 2.7140153144627033,
      "grad_norm": 0.96875,
      "learning_rate": 0.00043327384404290216,
      "loss": 0.9782,
      "step": 31545
    },
    {
      "epoch": 2.714101350770025,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00043324704527797066,
      "loss": 1.284,
      "step": 31546
    },
    {
      "epoch": 2.7141873870773465,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004332202467082787,
      "loss": 1.2762,
      "step": 31547
    },
    {
      "epoch": 2.714273423384668,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00043319344833390457,
      "loss": 1.3041,
      "step": 31548
    },
    {
      "epoch": 2.71435945969199,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00043316665015492655,
      "loss": 1.0323,
      "step": 31549
    },
    {
      "epoch": 2.714445495999312,
      "grad_norm": 0.953125,
      "learning_rate": 0.000433139852171423,
      "loss": 1.3461,
      "step": 31550
    },
    {
      "epoch": 2.7145315323066335,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004331130543834725,
      "loss": 0.5604,
      "step": 31551
    },
    {
      "epoch": 2.714617568613955,
      "grad_norm": 0.953125,
      "learning_rate": 0.00043308625679115334,
      "loss": 0.9165,
      "step": 31552
    },
    {
      "epoch": 2.7147036049212767,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004330594593945437,
      "loss": 1.0235,
      "step": 31553
    },
    {
      "epoch": 2.7147896412285983,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004330326621937222,
      "loss": 1.0933,
      "step": 31554
    },
    {
      "epoch": 2.71487567753592,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004330058651887672,
      "loss": 1.5715,
      "step": 31555
    },
    {
      "epoch": 2.714961713843242,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000432979068379757,
      "loss": 0.9212,
      "step": 31556
    },
    {
      "epoch": 2.7150477501505637,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00043295227176676985,
      "loss": 1.0188,
      "step": 31557
    },
    {
      "epoch": 2.7151337864578853,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004329254753498844,
      "loss": 1.1312,
      "step": 31558
    },
    {
      "epoch": 2.715219822765207,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004328986791291789,
      "loss": 1.5351,
      "step": 31559
    },
    {
      "epoch": 2.7153058590725285,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004328718831047316,
      "loss": 1.2409,
      "step": 31560
    },
    {
      "epoch": 2.71539189537985,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004328450872766211,
      "loss": 0.6733,
      "step": 31561
    },
    {
      "epoch": 2.7154779316871718,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004328182916449256,
      "loss": 1.4746,
      "step": 31562
    },
    {
      "epoch": 2.715563967994494,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00043279149620972356,
      "loss": 1.3484,
      "step": 31563
    },
    {
      "epoch": 2.7156500043018155,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00043276470097109327,
      "loss": 1.6265,
      "step": 31564
    },
    {
      "epoch": 2.715736040609137,
      "grad_norm": 1.015625,
      "learning_rate": 0.000432737905929113,
      "loss": 1.0927,
      "step": 31565
    },
    {
      "epoch": 2.7158220769164587,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004327111110838615,
      "loss": 1.0082,
      "step": 31566
    },
    {
      "epoch": 2.7159081132237803,
      "grad_norm": 1.234375,
      "learning_rate": 0.00043268431643541695,
      "loss": 1.5907,
      "step": 31567
    },
    {
      "epoch": 2.715994149531102,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004326575219838576,
      "loss": 1.1856,
      "step": 31568
    },
    {
      "epoch": 2.7160801858384236,
      "grad_norm": 1.0,
      "learning_rate": 0.00043263072772926187,
      "loss": 1.7642,
      "step": 31569
    },
    {
      "epoch": 2.7161662221457457,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00043260393367170824,
      "loss": 1.0161,
      "step": 31570
    },
    {
      "epoch": 2.7162522584530673,
      "grad_norm": 1.109375,
      "learning_rate": 0.000432577139811275,
      "loss": 0.9222,
      "step": 31571
    },
    {
      "epoch": 2.716338294760389,
      "grad_norm": 0.953125,
      "learning_rate": 0.00043255034614804043,
      "loss": 0.8125,
      "step": 31572
    },
    {
      "epoch": 2.7164243310677105,
      "grad_norm": 1.296875,
      "learning_rate": 0.00043252355268208297,
      "loss": 1.1679,
      "step": 31573
    },
    {
      "epoch": 2.716510367375032,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004324967594134811,
      "loss": 1.4049,
      "step": 31574
    },
    {
      "epoch": 2.716596403682354,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004324699663423131,
      "loss": 1.0039,
      "step": 31575
    },
    {
      "epoch": 2.7166824399896754,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00043244317346865716,
      "loss": 1.3204,
      "step": 31576
    },
    {
      "epoch": 2.7167684762969975,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004324163807925919,
      "loss": 1.8018,
      "step": 31577
    },
    {
      "epoch": 2.716854512604319,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004323895883141956,
      "loss": 0.8225,
      "step": 31578
    },
    {
      "epoch": 2.7169405489116407,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004323627960335466,
      "loss": 1.2526,
      "step": 31579
    },
    {
      "epoch": 2.7170265852189623,
      "grad_norm": 1.375,
      "learning_rate": 0.00043233600395072307,
      "loss": 1.3012,
      "step": 31580
    },
    {
      "epoch": 2.717112621526284,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004323092120658038,
      "loss": 0.515,
      "step": 31581
    },
    {
      "epoch": 2.717198657833606,
      "grad_norm": 1.40625,
      "learning_rate": 0.0004322824203788669,
      "loss": 1.2062,
      "step": 31582
    },
    {
      "epoch": 2.717284694140927,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004322556288899906,
      "loss": 0.6284,
      "step": 31583
    },
    {
      "epoch": 2.7173707304482493,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004322288375992535,
      "loss": 1.0582,
      "step": 31584
    },
    {
      "epoch": 2.717456766755571,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004322020465067339,
      "loss": 1.5053,
      "step": 31585
    },
    {
      "epoch": 2.7175428030628925,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004321752556125102,
      "loss": 0.9713,
      "step": 31586
    },
    {
      "epoch": 2.717628839370214,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004321484649166605,
      "loss": 1.2406,
      "step": 31587
    },
    {
      "epoch": 2.717714875677536,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004321216744192635,
      "loss": 0.9755,
      "step": 31588
    },
    {
      "epoch": 2.717800911984858,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00043209488412039734,
      "loss": 1.098,
      "step": 31589
    },
    {
      "epoch": 2.717886948292179,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004320680940201405,
      "loss": 1.851,
      "step": 31590
    },
    {
      "epoch": 2.717972984599501,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0004320413041185711,
      "loss": 0.6119,
      "step": 31591
    },
    {
      "epoch": 2.7180590209068227,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004320145144157678,
      "loss": 0.9898,
      "step": 31592
    },
    {
      "epoch": 2.7181450572141443,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004319877249118088,
      "loss": 1.517,
      "step": 31593
    },
    {
      "epoch": 2.718231093521466,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004319609356067724,
      "loss": 2.1769,
      "step": 31594
    },
    {
      "epoch": 2.7183171298287876,
      "grad_norm": 1.0625,
      "learning_rate": 0.00043193414650073713,
      "loss": 1.2836,
      "step": 31595
    },
    {
      "epoch": 2.7184031661361097,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004319073575937813,
      "loss": 1.4997,
      "step": 31596
    },
    {
      "epoch": 2.7184892024434313,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004318805688859831,
      "loss": 1.9665,
      "step": 31597
    },
    {
      "epoch": 2.718575238750753,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00043185378037742094,
      "loss": 0.7359,
      "step": 31598
    },
    {
      "epoch": 2.7186612750580745,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004318269920681734,
      "loss": 1.7422,
      "step": 31599
    },
    {
      "epoch": 2.718747311365396,
      "grad_norm": 0.96875,
      "learning_rate": 0.00043180020395831853,
      "loss": 0.7281,
      "step": 31600
    },
    {
      "epoch": 2.718833347672718,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00043177341604793475,
      "loss": 1.0505,
      "step": 31601
    },
    {
      "epoch": 2.7189193839800394,
      "grad_norm": 1.375,
      "learning_rate": 0.00043174662833710045,
      "loss": 1.2553,
      "step": 31602
    },
    {
      "epoch": 2.7190054202873615,
      "grad_norm": 1.3125,
      "learning_rate": 0.000431719840825894,
      "loss": 1.9464,
      "step": 31603
    },
    {
      "epoch": 2.719091456594683,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004316930535143938,
      "loss": 1.1651,
      "step": 31604
    },
    {
      "epoch": 2.7191774929020047,
      "grad_norm": 1.078125,
      "learning_rate": 0.00043166626640267817,
      "loss": 0.7531,
      "step": 31605
    },
    {
      "epoch": 2.7192635292093263,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004316394794908253,
      "loss": 1.0562,
      "step": 31606
    },
    {
      "epoch": 2.719349565516648,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004316126927789137,
      "loss": 1.0509,
      "step": 31607
    },
    {
      "epoch": 2.7194356018239696,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004315859062670218,
      "loss": 1.8772,
      "step": 31608
    },
    {
      "epoch": 2.719521638131291,
      "grad_norm": 1.515625,
      "learning_rate": 0.00043155911995522775,
      "loss": 1.7758,
      "step": 31609
    },
    {
      "epoch": 2.7196076744386133,
      "grad_norm": 1.171875,
      "learning_rate": 0.00043153233384360986,
      "loss": 1.8203,
      "step": 31610
    },
    {
      "epoch": 2.719693710745935,
      "grad_norm": 1.234375,
      "learning_rate": 0.00043150554793224666,
      "loss": 1.5159,
      "step": 31611
    },
    {
      "epoch": 2.7197797470532565,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00043147876222121643,
      "loss": 0.8483,
      "step": 31612
    },
    {
      "epoch": 2.719865783360578,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00043145197671059756,
      "loss": 0.8063,
      "step": 31613
    },
    {
      "epoch": 2.7199518196679,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004314251914004681,
      "loss": 1.3427,
      "step": 31614
    },
    {
      "epoch": 2.7200378559752214,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00043139840629090683,
      "loss": 0.922,
      "step": 31615
    },
    {
      "epoch": 2.720123892282543,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0004313716213819918,
      "loss": 0.8542,
      "step": 31616
    },
    {
      "epoch": 2.720209928589865,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004313448366738014,
      "loss": 1.4965,
      "step": 31617
    },
    {
      "epoch": 2.7202959648971867,
      "grad_norm": 0.90625,
      "learning_rate": 0.00043131805216641407,
      "loss": 0.8937,
      "step": 31618
    },
    {
      "epoch": 2.7203820012045083,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000431291267859908,
      "loss": 1.3449,
      "step": 31619
    },
    {
      "epoch": 2.72046803751183,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004312644837543617,
      "loss": 0.8061,
      "step": 31620
    },
    {
      "epoch": 2.7205540738191516,
      "grad_norm": 2.109375,
      "learning_rate": 0.00043123769984985326,
      "loss": 1.028,
      "step": 31621
    },
    {
      "epoch": 2.720640110126473,
      "grad_norm": 1.015625,
      "learning_rate": 0.00043121091614646133,
      "loss": 0.7156,
      "step": 31622
    },
    {
      "epoch": 2.720726146433795,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00043118413264426415,
      "loss": 1.3168,
      "step": 31623
    },
    {
      "epoch": 2.720812182741117,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004311573493433398,
      "loss": 1.3734,
      "step": 31624
    },
    {
      "epoch": 2.7208982190484385,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004311305662437669,
      "loss": 1.0331,
      "step": 31625
    },
    {
      "epoch": 2.72098425535576,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00043110378334562373,
      "loss": 1.303,
      "step": 31626
    },
    {
      "epoch": 2.721070291663082,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00043107700064898865,
      "loss": 0.6847,
      "step": 31627
    },
    {
      "epoch": 2.7211563279704034,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004310502181539397,
      "loss": 1.263,
      "step": 31628
    },
    {
      "epoch": 2.721242364277725,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004310234358605557,
      "loss": 1.3239,
      "step": 31629
    },
    {
      "epoch": 2.7213284005850467,
      "grad_norm": 1.078125,
      "learning_rate": 0.00043099665376891466,
      "loss": 0.7632,
      "step": 31630
    },
    {
      "epoch": 2.7214144368923687,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004309698718790949,
      "loss": 1.2746,
      "step": 31631
    },
    {
      "epoch": 2.7215004731996904,
      "grad_norm": 1.0625,
      "learning_rate": 0.00043094309019117493,
      "loss": 0.839,
      "step": 31632
    },
    {
      "epoch": 2.721586509507012,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000430916308705233,
      "loss": 1.3871,
      "step": 31633
    },
    {
      "epoch": 2.7216725458143336,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004308895274213475,
      "loss": 1.3774,
      "step": 31634
    },
    {
      "epoch": 2.7217585821216552,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00043086274633959647,
      "loss": 1.0265,
      "step": 31635
    },
    {
      "epoch": 2.7218446184289773,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00043083596546005865,
      "loss": 0.7558,
      "step": 31636
    },
    {
      "epoch": 2.7219306547362985,
      "grad_norm": 0.78125,
      "learning_rate": 0.00043080918478281214,
      "loss": 1.0339,
      "step": 31637
    },
    {
      "epoch": 2.7220166910436205,
      "grad_norm": 1.375,
      "learning_rate": 0.0004307824043079353,
      "loss": 1.8129,
      "step": 31638
    },
    {
      "epoch": 2.722102727350942,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00043075562403550627,
      "loss": 1.656,
      "step": 31639
    },
    {
      "epoch": 2.722188763658264,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00043072884396560363,
      "loss": 1.5287,
      "step": 31640
    },
    {
      "epoch": 2.7222747999655854,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004307020640983058,
      "loss": 1.2994,
      "step": 31641
    },
    {
      "epoch": 2.722360836272907,
      "grad_norm": 1.171875,
      "learning_rate": 0.00043067528443369095,
      "loss": 1.0007,
      "step": 31642
    },
    {
      "epoch": 2.722446872580229,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00043064850497183746,
      "loss": 0.5729,
      "step": 31643
    },
    {
      "epoch": 2.7225329088875503,
      "grad_norm": 1.03125,
      "learning_rate": 0.00043062172571282345,
      "loss": 1.785,
      "step": 31644
    },
    {
      "epoch": 2.7226189451948724,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00043059494665672747,
      "loss": 1.0314,
      "step": 31645
    },
    {
      "epoch": 2.722704981502194,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00043056816780362786,
      "loss": 1.2038,
      "step": 31646
    },
    {
      "epoch": 2.7227910178095156,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004305413891536027,
      "loss": 0.787,
      "step": 31647
    },
    {
      "epoch": 2.7228770541168372,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004305146107067305,
      "loss": 0.776,
      "step": 31648
    },
    {
      "epoch": 2.722963090424159,
      "grad_norm": 0.921875,
      "learning_rate": 0.00043048783246308963,
      "loss": 1.0234,
      "step": 31649
    },
    {
      "epoch": 2.723049126731481,
      "grad_norm": 1.46875,
      "learning_rate": 0.00043046105442275833,
      "loss": 2.2153,
      "step": 31650
    },
    {
      "epoch": 2.723135163038802,
      "grad_norm": 1.25,
      "learning_rate": 0.00043043427658581484,
      "loss": 1.2313,
      "step": 31651
    },
    {
      "epoch": 2.723221199346124,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00043040749895233767,
      "loss": 0.9477,
      "step": 31652
    },
    {
      "epoch": 2.723307235653446,
      "grad_norm": 1.015625,
      "learning_rate": 0.000430380721522405,
      "loss": 1.7853,
      "step": 31653
    },
    {
      "epoch": 2.7233932719607674,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00043035394429609517,
      "loss": 0.865,
      "step": 31654
    },
    {
      "epoch": 2.723479308268089,
      "grad_norm": 1.703125,
      "learning_rate": 0.00043032716727348634,
      "loss": 1.622,
      "step": 31655
    },
    {
      "epoch": 2.7235653445754107,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004303003904546572,
      "loss": 1.9017,
      "step": 31656
    },
    {
      "epoch": 2.7236513808827327,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00043027361383968585,
      "loss": 1.4854,
      "step": 31657
    },
    {
      "epoch": 2.7237374171900544,
      "grad_norm": 1.3125,
      "learning_rate": 0.00043024683742865047,
      "loss": 1.1713,
      "step": 31658
    },
    {
      "epoch": 2.723823453497376,
      "grad_norm": 0.859375,
      "learning_rate": 0.00043022006122162964,
      "loss": 0.7055,
      "step": 31659
    },
    {
      "epoch": 2.7239094898046976,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004301932852187016,
      "loss": 1.5397,
      "step": 31660
    },
    {
      "epoch": 2.7239955261120192,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004301665094199446,
      "loss": 1.1739,
      "step": 31661
    },
    {
      "epoch": 2.724081562419341,
      "grad_norm": 1.078125,
      "learning_rate": 0.00043013973382543684,
      "loss": 0.8993,
      "step": 31662
    },
    {
      "epoch": 2.7241675987266625,
      "grad_norm": 0.84375,
      "learning_rate": 0.00043011295843525696,
      "loss": 0.8479,
      "step": 31663
    },
    {
      "epoch": 2.7242536350339845,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000430086183249483,
      "loss": 1.4707,
      "step": 31664
    },
    {
      "epoch": 2.724339671341306,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004300594082681933,
      "loss": 1.2353,
      "step": 31665
    },
    {
      "epoch": 2.724425707648628,
      "grad_norm": 1.15625,
      "learning_rate": 0.00043003263349146627,
      "loss": 1.6779,
      "step": 31666
    },
    {
      "epoch": 2.7245117439559494,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00043000585891938024,
      "loss": 1.7982,
      "step": 31667
    },
    {
      "epoch": 2.724597780263271,
      "grad_norm": 0.921875,
      "learning_rate": 0.00042997908455201343,
      "loss": 1.9366,
      "step": 31668
    },
    {
      "epoch": 2.7246838165705927,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004299523103894441,
      "loss": 0.8695,
      "step": 31669
    },
    {
      "epoch": 2.7247698528779143,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00042992553643175075,
      "loss": 1.5752,
      "step": 31670
    },
    {
      "epoch": 2.7248558891852364,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004298987626790116,
      "loss": 1.1285,
      "step": 31671
    },
    {
      "epoch": 2.724941925492558,
      "grad_norm": 1.328125,
      "learning_rate": 0.00042987198913130477,
      "loss": 1.0341,
      "step": 31672
    },
    {
      "epoch": 2.7250279617998796,
      "grad_norm": 1.1875,
      "learning_rate": 0.00042984521578870883,
      "loss": 0.9512,
      "step": 31673
    },
    {
      "epoch": 2.7251139981072012,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000429818442651302,
      "loss": 0.8575,
      "step": 31674
    },
    {
      "epoch": 2.725200034414523,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004297916697191626,
      "loss": 0.6025,
      "step": 31675
    },
    {
      "epoch": 2.7252860707218445,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004297648969923687,
      "loss": 1.647,
      "step": 31676
    },
    {
      "epoch": 2.725372107029166,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00042973812447099903,
      "loss": 0.74,
      "step": 31677
    },
    {
      "epoch": 2.725458143336488,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004297113521551316,
      "loss": 0.9702,
      "step": 31678
    },
    {
      "epoch": 2.72554417964381,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004296845800448448,
      "loss": 2.3144,
      "step": 31679
    },
    {
      "epoch": 2.7256302159511314,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00042965780814021703,
      "loss": 1.0238,
      "step": 31680
    },
    {
      "epoch": 2.725716252258453,
      "grad_norm": 1.015625,
      "learning_rate": 0.00042963103644132634,
      "loss": 1.0548,
      "step": 31681
    },
    {
      "epoch": 2.7258022885657747,
      "grad_norm": 1.34375,
      "learning_rate": 0.00042960426494825126,
      "loss": 0.8434,
      "step": 31682
    },
    {
      "epoch": 2.7258883248730963,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004295774936610701,
      "loss": 1.2772,
      "step": 31683
    },
    {
      "epoch": 2.725974361180418,
      "grad_norm": 1.03125,
      "learning_rate": 0.000429550722579861,
      "loss": 1.4818,
      "step": 31684
    },
    {
      "epoch": 2.72606039748774,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00042952395170470226,
      "loss": 1.1361,
      "step": 31685
    },
    {
      "epoch": 2.7261464337950616,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004294971810356724,
      "loss": 1.3419,
      "step": 31686
    },
    {
      "epoch": 2.7262324701023832,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004294704105728495,
      "loss": 0.8724,
      "step": 31687
    },
    {
      "epoch": 2.726318506409705,
      "grad_norm": 1.015625,
      "learning_rate": 0.00042944364031631186,
      "loss": 1.8238,
      "step": 31688
    },
    {
      "epoch": 2.7264045427170265,
      "grad_norm": 0.765625,
      "learning_rate": 0.0004294168702661379,
      "loss": 1.183,
      "step": 31689
    },
    {
      "epoch": 2.7264905790243485,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004293901004224059,
      "loss": 1.2596,
      "step": 31690
    },
    {
      "epoch": 2.7265766153316697,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004293633307851942,
      "loss": 0.9403,
      "step": 31691
    },
    {
      "epoch": 2.726662651638992,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004293365613545808,
      "loss": 1.5944,
      "step": 31692
    },
    {
      "epoch": 2.7267486879463134,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00042930979213064434,
      "loss": 1.331,
      "step": 31693
    },
    {
      "epoch": 2.726834724253635,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00042928302311346305,
      "loss": 0.7277,
      "step": 31694
    },
    {
      "epoch": 2.7269207605609567,
      "grad_norm": 1.28125,
      "learning_rate": 0.00042925625430311506,
      "loss": 1.4441,
      "step": 31695
    },
    {
      "epoch": 2.7270067968682783,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004292294856996788,
      "loss": 1.454,
      "step": 31696
    },
    {
      "epoch": 2.7270928331756004,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004292027173032326,
      "loss": 1.1339,
      "step": 31697
    },
    {
      "epoch": 2.7271788694829215,
      "grad_norm": 1.25,
      "learning_rate": 0.0004291759491138547,
      "loss": 1.815,
      "step": 31698
    },
    {
      "epoch": 2.7272649057902436,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00042914918113162314,
      "loss": 0.8368,
      "step": 31699
    },
    {
      "epoch": 2.7273509420975652,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00042912241335661666,
      "loss": 1.2489,
      "step": 31700
    },
    {
      "epoch": 2.727436978404887,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00042909564578891334,
      "loss": 0.8989,
      "step": 31701
    },
    {
      "epoch": 2.7275230147122085,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00042906887842859143,
      "loss": 0.8701,
      "step": 31702
    },
    {
      "epoch": 2.72760905101953,
      "grad_norm": 1.0625,
      "learning_rate": 0.00042904211127572917,
      "loss": 1.3702,
      "step": 31703
    },
    {
      "epoch": 2.727695087326852,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000429015344330405,
      "loss": 1.4906,
      "step": 31704
    },
    {
      "epoch": 2.7277811236341734,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00042898857759269713,
      "loss": 0.9108,
      "step": 31705
    },
    {
      "epoch": 2.7278671599414954,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004289618110626838,
      "loss": 1.2683,
      "step": 31706
    },
    {
      "epoch": 2.727953196248817,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004289350447404435,
      "loss": 1.0532,
      "step": 31707
    },
    {
      "epoch": 2.7280392325561387,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004289082786260543,
      "loss": 0.6579,
      "step": 31708
    },
    {
      "epoch": 2.7281252688634603,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00042888151271959463,
      "loss": 1.1006,
      "step": 31709
    },
    {
      "epoch": 2.728211305170782,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004288547470211426,
      "loss": 1.0218,
      "step": 31710
    },
    {
      "epoch": 2.728297341478104,
      "grad_norm": 1.125,
      "learning_rate": 0.00042882798153077664,
      "loss": 1.5806,
      "step": 31711
    },
    {
      "epoch": 2.7283833777854256,
      "grad_norm": 0.78125,
      "learning_rate": 0.000428801216248575,
      "loss": 0.4975,
      "step": 31712
    },
    {
      "epoch": 2.7284694140927472,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0004287744511746157,
      "loss": 1.0725,
      "step": 31713
    },
    {
      "epoch": 2.728555450400069,
      "grad_norm": 1.578125,
      "learning_rate": 0.0004287476863089775,
      "loss": 1.6687,
      "step": 31714
    },
    {
      "epoch": 2.7286414867073905,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004287209216517385,
      "loss": 1.558,
      "step": 31715
    },
    {
      "epoch": 2.728727523014712,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004286941572029769,
      "loss": 0.918,
      "step": 31716
    },
    {
      "epoch": 2.7288135593220337,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004286673929627711,
      "loss": 1.4732,
      "step": 31717
    },
    {
      "epoch": 2.728899595629356,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00042864062893119916,
      "loss": 1.0296,
      "step": 31718
    },
    {
      "epoch": 2.7289856319366774,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00042861386510833956,
      "loss": 1.6027,
      "step": 31719
    },
    {
      "epoch": 2.729071668243999,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004285871014942706,
      "loss": 0.9229,
      "step": 31720
    },
    {
      "epoch": 2.7291577045513207,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00042856033808907044,
      "loss": 1.3855,
      "step": 31721
    },
    {
      "epoch": 2.7292437408586423,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0004285335748928173,
      "loss": 0.709,
      "step": 31722
    },
    {
      "epoch": 2.729329777165964,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004285068119055897,
      "loss": 0.8568,
      "step": 31723
    },
    {
      "epoch": 2.7294158134732855,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004284800491274657,
      "loss": 1.0121,
      "step": 31724
    },
    {
      "epoch": 2.7295018497806076,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004284532865585237,
      "loss": 0.9237,
      "step": 31725
    },
    {
      "epoch": 2.7295878860879292,
      "grad_norm": 1.40625,
      "learning_rate": 0.00042842652419884176,
      "loss": 1.4485,
      "step": 31726
    },
    {
      "epoch": 2.729673922395251,
      "grad_norm": 1.234375,
      "learning_rate": 0.00042839976204849846,
      "loss": 0.9872,
      "step": 31727
    },
    {
      "epoch": 2.7297599587025725,
      "grad_norm": 0.9375,
      "learning_rate": 0.0004283730001075719,
      "loss": 1.1029,
      "step": 31728
    },
    {
      "epoch": 2.729845995009894,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00042834623837614035,
      "loss": 0.9058,
      "step": 31729
    },
    {
      "epoch": 2.7299320313172157,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004283194768542821,
      "loss": 0.7413,
      "step": 31730
    },
    {
      "epoch": 2.7300180676245374,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00042829271554207555,
      "loss": 0.9278,
      "step": 31731
    },
    {
      "epoch": 2.7301041039318594,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004282659544395988,
      "loss": 1.1815,
      "step": 31732
    },
    {
      "epoch": 2.730190140239181,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00042823919354693013,
      "loss": 1.5853,
      "step": 31733
    },
    {
      "epoch": 2.7302761765465027,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000428212432864148,
      "loss": 0.9836,
      "step": 31734
    },
    {
      "epoch": 2.7303622128538243,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004281856723913304,
      "loss": 0.7132,
      "step": 31735
    },
    {
      "epoch": 2.730448249161146,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004281589121285557,
      "loss": 2.1209,
      "step": 31736
    },
    {
      "epoch": 2.7305342854684675,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004281321520759024,
      "loss": 1.3841,
      "step": 31737
    },
    {
      "epoch": 2.730620321775789,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004281053922334485,
      "loss": 0.7338,
      "step": 31738
    },
    {
      "epoch": 2.7307063580831112,
      "grad_norm": 1.21875,
      "learning_rate": 0.00042807863260127234,
      "loss": 1.0347,
      "step": 31739
    },
    {
      "epoch": 2.730792394390433,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004280518731794521,
      "loss": 1.774,
      "step": 31740
    },
    {
      "epoch": 2.7308784306977545,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004280251139680663,
      "loss": 0.9287,
      "step": 31741
    },
    {
      "epoch": 2.730964467005076,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00042799835496719297,
      "loss": 1.6176,
      "step": 31742
    },
    {
      "epoch": 2.7310505033123977,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00042797159617691044,
      "loss": 1.597,
      "step": 31743
    },
    {
      "epoch": 2.7311365396197194,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004279448375972969,
      "loss": 0.6547,
      "step": 31744
    },
    {
      "epoch": 2.731222575927041,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00042791807922843084,
      "loss": 0.9352,
      "step": 31745
    },
    {
      "epoch": 2.731308612234363,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004278913210703903,
      "loss": 0.8656,
      "step": 31746
    },
    {
      "epoch": 2.7313946485416847,
      "grad_norm": 1.046875,
      "learning_rate": 0.00042786456312325356,
      "loss": 2.1024,
      "step": 31747
    },
    {
      "epoch": 2.7314806848490063,
      "grad_norm": 1.015625,
      "learning_rate": 0.00042783780538709906,
      "loss": 1.4533,
      "step": 31748
    },
    {
      "epoch": 2.731566721156328,
      "grad_norm": 1.03125,
      "learning_rate": 0.000427811047862005,
      "loss": 1.4434,
      "step": 31749
    },
    {
      "epoch": 2.7316527574636496,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004277842905480492,
      "loss": 1.158,
      "step": 31750
    },
    {
      "epoch": 2.7317387937709716,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004277575334453107,
      "loss": 1.4745,
      "step": 31751
    },
    {
      "epoch": 2.731824830078293,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004277307765538672,
      "loss": 1.4901,
      "step": 31752
    },
    {
      "epoch": 2.731910866385615,
      "grad_norm": 2.234375,
      "learning_rate": 0.00042770401987379716,
      "loss": 1.1149,
      "step": 31753
    },
    {
      "epoch": 2.7319969026929365,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00042767726340517886,
      "loss": 1.0568,
      "step": 31754
    },
    {
      "epoch": 2.732082939000258,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004276505071480905,
      "loss": 0.9824,
      "step": 31755
    },
    {
      "epoch": 2.7321689753075797,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004276237511026102,
      "loss": 1.2138,
      "step": 31756
    },
    {
      "epoch": 2.7322550116149014,
      "grad_norm": 1.09375,
      "learning_rate": 0.00042759699526881645,
      "loss": 1.0826,
      "step": 31757
    },
    {
      "epoch": 2.7323410479222234,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004275702396467875,
      "loss": 0.7668,
      "step": 31758
    },
    {
      "epoch": 2.7324270842295446,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004275434842366013,
      "loss": 0.9716,
      "step": 31759
    },
    {
      "epoch": 2.7325131205368667,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004275167290383365,
      "loss": 1.7938,
      "step": 31760
    },
    {
      "epoch": 2.7325991568441883,
      "grad_norm": 8.375,
      "learning_rate": 0.0004274899740520711,
      "loss": 1.2295,
      "step": 31761
    },
    {
      "epoch": 2.73268519315151,
      "grad_norm": 1.0625,
      "learning_rate": 0.00042746321927788345,
      "loss": 1.0389,
      "step": 31762
    },
    {
      "epoch": 2.7327712294588316,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0004274364647158517,
      "loss": 1.5191,
      "step": 31763
    },
    {
      "epoch": 2.732857265766153,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00042740971036605424,
      "loss": 1.1672,
      "step": 31764
    },
    {
      "epoch": 2.7329433020734752,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0004273829562285693,
      "loss": 1.4886,
      "step": 31765
    },
    {
      "epoch": 2.7330293383807964,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00042735620230347507,
      "loss": 1.5153,
      "step": 31766
    },
    {
      "epoch": 2.7331153746881185,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004273294485908497,
      "loss": 1.5185,
      "step": 31767
    },
    {
      "epoch": 2.73320141099544,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0004273026950907717,
      "loss": 1.0192,
      "step": 31768
    },
    {
      "epoch": 2.7332874473027617,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00042727594180331916,
      "loss": 0.714,
      "step": 31769
    },
    {
      "epoch": 2.7333734836100834,
      "grad_norm": 1.25,
      "learning_rate": 0.0004272491887285702,
      "loss": 1.3506,
      "step": 31770
    },
    {
      "epoch": 2.733459519917405,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004272224358666034,
      "loss": 0.8803,
      "step": 31771
    },
    {
      "epoch": 2.733545556224727,
      "grad_norm": 1.125,
      "learning_rate": 0.00042719568321749674,
      "loss": 1.3053,
      "step": 31772
    },
    {
      "epoch": 2.7336315925320487,
      "grad_norm": 1.234375,
      "learning_rate": 0.00042716893078132854,
      "loss": 1.6873,
      "step": 31773
    },
    {
      "epoch": 2.7337176288393703,
      "grad_norm": 3.734375,
      "learning_rate": 0.00042714217855817707,
      "loss": 1.4038,
      "step": 31774
    },
    {
      "epoch": 2.733803665146692,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004271154265481206,
      "loss": 1.1474,
      "step": 31775
    },
    {
      "epoch": 2.7338897014540136,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00042708867475123725,
      "loss": 1.4326,
      "step": 31776
    },
    {
      "epoch": 2.733975737761335,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00042706192316760536,
      "loss": 1.5383,
      "step": 31777
    },
    {
      "epoch": 2.734061774068657,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004270351717973032,
      "loss": 1.1962,
      "step": 31778
    },
    {
      "epoch": 2.734147810375979,
      "grad_norm": 1.46875,
      "learning_rate": 0.00042700842064040894,
      "loss": 1.3629,
      "step": 31779
    },
    {
      "epoch": 2.7342338466833005,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004269816696970009,
      "loss": 1.182,
      "step": 31780
    },
    {
      "epoch": 2.734319882990622,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004269549189671572,
      "loss": 0.9197,
      "step": 31781
    },
    {
      "epoch": 2.7344059192979437,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00042692816845095623,
      "loss": 0.7965,
      "step": 31782
    },
    {
      "epoch": 2.7344919556052654,
      "grad_norm": 8.6875,
      "learning_rate": 0.00042690141814847617,
      "loss": 1.6856,
      "step": 31783
    },
    {
      "epoch": 2.734577991912587,
      "grad_norm": 55.75,
      "learning_rate": 0.0004268746680597951,
      "loss": 0.7346,
      "step": 31784
    },
    {
      "epoch": 2.7346640282199086,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004268479181849915,
      "loss": 0.8321,
      "step": 31785
    },
    {
      "epoch": 2.7347500645272307,
      "grad_norm": 29.375,
      "learning_rate": 0.00042682116852414356,
      "loss": 0.9759,
      "step": 31786
    },
    {
      "epoch": 2.7348361008345523,
      "grad_norm": 1392.0,
      "learning_rate": 0.00042679441907732924,
      "loss": 0.9404,
      "step": 31787
    },
    {
      "epoch": 2.734922137141874,
      "grad_norm": 3856.0,
      "learning_rate": 0.00042676766984462724,
      "loss": 0.5969,
      "step": 31788
    },
    {
      "epoch": 2.7350081734491956,
      "grad_norm": 1012.0,
      "learning_rate": 0.00042674092082611555,
      "loss": 1.5184,
      "step": 31789
    },
    {
      "epoch": 2.735094209756517,
      "grad_norm": 5728.0,
      "learning_rate": 0.0004267141720218723,
      "loss": 1.4062,
      "step": 31790
    },
    {
      "epoch": 2.735180246063839,
      "grad_norm": 342.0,
      "learning_rate": 0.000426687423431976,
      "loss": 1.6093,
      "step": 31791
    },
    {
      "epoch": 2.7352662823711604,
      "grad_norm": 2008.0,
      "learning_rate": 0.00042666067505650466,
      "loss": 0.7903,
      "step": 31792
    },
    {
      "epoch": 2.7353523186784825,
      "grad_norm": 940.0,
      "learning_rate": 0.0004266339268955365,
      "loss": 0.9196,
      "step": 31793
    },
    {
      "epoch": 2.735438354985804,
      "grad_norm": 85.0,
      "learning_rate": 0.0004266071789491499,
      "loss": 0.7607,
      "step": 31794
    },
    {
      "epoch": 2.7355243912931257,
      "grad_norm": 8.625,
      "learning_rate": 0.00042658043121742307,
      "loss": 1.7754,
      "step": 31795
    },
    {
      "epoch": 2.7356104276004474,
      "grad_norm": 2.484375,
      "learning_rate": 0.00042655368370043425,
      "loss": 0.8766,
      "step": 31796
    },
    {
      "epoch": 2.735696463907769,
      "grad_norm": 140.0,
      "learning_rate": 0.0004265269363982614,
      "loss": 0.8806,
      "step": 31797
    },
    {
      "epoch": 2.7357825002150906,
      "grad_norm": 156.0,
      "learning_rate": 0.0004265001893109832,
      "loss": 1.1938,
      "step": 31798
    },
    {
      "epoch": 2.7358685365224122,
      "grad_norm": 2.21875,
      "learning_rate": 0.0004264734424386775,
      "loss": 1.1662,
      "step": 31799
    },
    {
      "epoch": 2.7359545728297343,
      "grad_norm": 1.375,
      "learning_rate": 0.0004264466957814227,
      "loss": 1.7818,
      "step": 31800
    },
    {
      "epoch": 2.736040609137056,
      "grad_norm": 1.203125,
      "learning_rate": 0.00042641994933929705,
      "loss": 1.2154,
      "step": 31801
    },
    {
      "epoch": 2.7361266454443776,
      "grad_norm": 1664.0,
      "learning_rate": 0.0004263932031123788,
      "loss": 1.7549,
      "step": 31802
    },
    {
      "epoch": 2.736212681751699,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00042636645710074605,
      "loss": 1.2928,
      "step": 31803
    },
    {
      "epoch": 2.736298718059021,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004263397113044771,
      "loss": 1.7032,
      "step": 31804
    },
    {
      "epoch": 2.7363847543663424,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004263129657236501,
      "loss": 0.9132,
      "step": 31805
    },
    {
      "epoch": 2.736470790673664,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00042628622035834345,
      "loss": 1.5343,
      "step": 31806
    },
    {
      "epoch": 2.736556826980986,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00042625947520863515,
      "loss": 1.2168,
      "step": 31807
    },
    {
      "epoch": 2.7366428632883077,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00042623273027460367,
      "loss": 0.9139,
      "step": 31808
    },
    {
      "epoch": 2.7367288995956294,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0004262059855563271,
      "loss": 1.7232,
      "step": 31809
    },
    {
      "epoch": 2.736814935902951,
      "grad_norm": 1.453125,
      "learning_rate": 0.00042617924105388367,
      "loss": 0.9942,
      "step": 31810
    },
    {
      "epoch": 2.7369009722102726,
      "grad_norm": 0.90625,
      "learning_rate": 0.00042615249676735144,
      "loss": 0.8858,
      "step": 31811
    },
    {
      "epoch": 2.7369870085175947,
      "grad_norm": 1.109375,
      "learning_rate": 0.000426125752696809,
      "loss": 1.3197,
      "step": 31812
    },
    {
      "epoch": 2.737073044824916,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00042609900884233424,
      "loss": 1.4825,
      "step": 31813
    },
    {
      "epoch": 2.737159081132238,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004260722652040056,
      "loss": 0.7198,
      "step": 31814
    },
    {
      "epoch": 2.7372451174395596,
      "grad_norm": 1.25,
      "learning_rate": 0.0004260455217819011,
      "loss": 0.9637,
      "step": 31815
    },
    {
      "epoch": 2.737331153746881,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00042601877857609917,
      "loss": 1.1265,
      "step": 31816
    },
    {
      "epoch": 2.737417190054203,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00042599203558667785,
      "loss": 0.8485,
      "step": 31817
    },
    {
      "epoch": 2.7375032263615244,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00042596529281371543,
      "loss": 1.1476,
      "step": 31818
    },
    {
      "epoch": 2.7375892626688465,
      "grad_norm": 1.453125,
      "learning_rate": 0.00042593855025729015,
      "loss": 1.7973,
      "step": 31819
    },
    {
      "epoch": 2.7376752989761677,
      "grad_norm": 1.0625,
      "learning_rate": 0.00042591180791748027,
      "loss": 0.8445,
      "step": 31820
    },
    {
      "epoch": 2.7377613352834897,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00042588506579436386,
      "loss": 0.8411,
      "step": 31821
    },
    {
      "epoch": 2.7378473715908114,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004258583238880192,
      "loss": 1.6565,
      "step": 31822
    },
    {
      "epoch": 2.737933407898133,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00042583158219852436,
      "loss": 0.6785,
      "step": 31823
    },
    {
      "epoch": 2.7380194442054546,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00042580484072595793,
      "loss": 1.4271,
      "step": 31824
    },
    {
      "epoch": 2.7381054805127762,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000425778099470398,
      "loss": 1.6751,
      "step": 31825
    },
    {
      "epoch": 2.7381915168200983,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004257513584319226,
      "loss": 1.4089,
      "step": 31826
    },
    {
      "epoch": 2.7382775531274195,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00042572461761060994,
      "loss": 0.8824,
      "step": 31827
    },
    {
      "epoch": 2.7383635894347416,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00042569787700653846,
      "loss": 1.1725,
      "step": 31828
    },
    {
      "epoch": 2.738449625742063,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004256711366197862,
      "loss": 1.2587,
      "step": 31829
    },
    {
      "epoch": 2.738535662049385,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00042564439645043144,
      "loss": 0.8959,
      "step": 31830
    },
    {
      "epoch": 2.7386216983567064,
      "grad_norm": 1.234375,
      "learning_rate": 0.00042561765649855223,
      "loss": 1.4613,
      "step": 31831
    },
    {
      "epoch": 2.738707734664028,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004255909167642271,
      "loss": 1.452,
      "step": 31832
    },
    {
      "epoch": 2.73879377097135,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00042556417724753394,
      "loss": 1.2155,
      "step": 31833
    },
    {
      "epoch": 2.7388798072786718,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00042553743794855104,
      "loss": 1.2172,
      "step": 31834
    },
    {
      "epoch": 2.7389658435859934,
      "grad_norm": 1.3125,
      "learning_rate": 0.00042551069886735676,
      "loss": 1.8723,
      "step": 31835
    },
    {
      "epoch": 2.739051879893315,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004254839600040292,
      "loss": 0.6406,
      "step": 31836
    },
    {
      "epoch": 2.7391379162006366,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00042545722135864654,
      "loss": 1.152,
      "step": 31837
    },
    {
      "epoch": 2.7392239525079582,
      "grad_norm": 1.328125,
      "learning_rate": 0.00042543048293128694,
      "loss": 1.7891,
      "step": 31838
    },
    {
      "epoch": 2.73930998881528,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00042540374472202876,
      "loss": 1.7491,
      "step": 31839
    },
    {
      "epoch": 2.739396025122602,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00042537700673095016,
      "loss": 1.4207,
      "step": 31840
    },
    {
      "epoch": 2.7394820614299236,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0004253502689581291,
      "loss": 0.7625,
      "step": 31841
    },
    {
      "epoch": 2.739568097737245,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00042532353140364414,
      "loss": 1.4627,
      "step": 31842
    },
    {
      "epoch": 2.739654134044567,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004252967940675734,
      "loss": 0.8126,
      "step": 31843
    },
    {
      "epoch": 2.7397401703518884,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00042527005694999496,
      "loss": 1.6219,
      "step": 31844
    },
    {
      "epoch": 2.73982620665921,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00042524332005098686,
      "loss": 1.6333,
      "step": 31845
    },
    {
      "epoch": 2.7399122429665317,
      "grad_norm": 1.375,
      "learning_rate": 0.00042521658337062775,
      "loss": 1.2335,
      "step": 31846
    },
    {
      "epoch": 2.7399982792738538,
      "grad_norm": 1.21875,
      "learning_rate": 0.00042518984690899554,
      "loss": 0.9561,
      "step": 31847
    },
    {
      "epoch": 2.7400843155811754,
      "grad_norm": 1.125,
      "learning_rate": 0.00042516311066616834,
      "loss": 1.3601,
      "step": 31848
    },
    {
      "epoch": 2.740170351888497,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00042513637464222464,
      "loss": 1.1388,
      "step": 31849
    },
    {
      "epoch": 2.7402563881958186,
      "grad_norm": 1.078125,
      "learning_rate": 0.00042510963883724245,
      "loss": 1.0075,
      "step": 31850
    },
    {
      "epoch": 2.7403424245031403,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00042508290325130006,
      "loss": 1.0837,
      "step": 31851
    },
    {
      "epoch": 2.740428460810462,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004250561678844754,
      "loss": 1.8163,
      "step": 31852
    },
    {
      "epoch": 2.7405144971177835,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00042502943273684706,
      "loss": 0.9945,
      "step": 31853
    },
    {
      "epoch": 2.7406005334251056,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000425002697808493,
      "loss": 1.2269,
      "step": 31854
    },
    {
      "epoch": 2.740686569732427,
      "grad_norm": 1.3125,
      "learning_rate": 0.00042497596309949156,
      "loss": 1.6885,
      "step": 31855
    },
    {
      "epoch": 2.740772606039749,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004249492286099207,
      "loss": 1.3432,
      "step": 31856
    },
    {
      "epoch": 2.7408586423470704,
      "grad_norm": 1.765625,
      "learning_rate": 0.00042492249433985876,
      "loss": 1.2222,
      "step": 31857
    },
    {
      "epoch": 2.740944678654392,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00042489576028938397,
      "loss": 1.1213,
      "step": 31858
    },
    {
      "epoch": 2.7410307149617137,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004248690264585744,
      "loss": 1.4053,
      "step": 31859
    },
    {
      "epoch": 2.7411167512690353,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004248422928475083,
      "loss": 0.9774,
      "step": 31860
    },
    {
      "epoch": 2.7412027875763574,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00042481555945626386,
      "loss": 0.7713,
      "step": 31861
    },
    {
      "epoch": 2.741288823883679,
      "grad_norm": 1.203125,
      "learning_rate": 0.00042478882628491945,
      "loss": 1.7094,
      "step": 31862
    },
    {
      "epoch": 2.7413748601910006,
      "grad_norm": 1.03125,
      "learning_rate": 0.000424762093333553,
      "loss": 1.016,
      "step": 31863
    },
    {
      "epoch": 2.7414608964983223,
      "grad_norm": 1.9375,
      "learning_rate": 0.0004247353606022427,
      "loss": 1.0652,
      "step": 31864
    },
    {
      "epoch": 2.741546932805644,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00042470862809106704,
      "loss": 1.138,
      "step": 31865
    },
    {
      "epoch": 2.741632969112966,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004246818958001039,
      "loss": 1.0953,
      "step": 31866
    },
    {
      "epoch": 2.741719005420287,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00042465516372943156,
      "loss": 0.8998,
      "step": 31867
    },
    {
      "epoch": 2.741805041727609,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004246284318791282,
      "loss": 1.7874,
      "step": 31868
    },
    {
      "epoch": 2.741891078034931,
      "grad_norm": 1.1875,
      "learning_rate": 0.00042460170024927207,
      "loss": 0.9265,
      "step": 31869
    },
    {
      "epoch": 2.7419771143422524,
      "grad_norm": 1.125,
      "learning_rate": 0.0004245749688399413,
      "loss": 1.4405,
      "step": 31870
    },
    {
      "epoch": 2.742063150649574,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000424548237651214,
      "loss": 1.5232,
      "step": 31871
    },
    {
      "epoch": 2.7421491869568957,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004245215066831685,
      "loss": 1.1531,
      "step": 31872
    },
    {
      "epoch": 2.7422352232642178,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0004244947759358829,
      "loss": 1.0625,
      "step": 31873
    },
    {
      "epoch": 2.742321259571539,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004244680454094354,
      "loss": 1.1406,
      "step": 31874
    },
    {
      "epoch": 2.742407295878861,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004244413151039041,
      "loss": 0.9399,
      "step": 31875
    },
    {
      "epoch": 2.7424933321861826,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004244145850193674,
      "loss": 0.9872,
      "step": 31876
    },
    {
      "epoch": 2.7425793684935043,
      "grad_norm": 0.78125,
      "learning_rate": 0.00042438785515590326,
      "loss": 0.6838,
      "step": 31877
    },
    {
      "epoch": 2.742665404800826,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00042436112551358997,
      "loss": 1.2696,
      "step": 31878
    },
    {
      "epoch": 2.7427514411081475,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004243343960925056,
      "loss": 1.4861,
      "step": 31879
    },
    {
      "epoch": 2.7428374774154696,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004243076668927285,
      "loss": 1.4439,
      "step": 31880
    },
    {
      "epoch": 2.7429235137227908,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004242809379143367,
      "loss": 1.4887,
      "step": 31881
    },
    {
      "epoch": 2.743009550030113,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004242542091574083,
      "loss": 1.0791,
      "step": 31882
    },
    {
      "epoch": 2.7430955863374344,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00042422748062202184,
      "loss": 1.4918,
      "step": 31883
    },
    {
      "epoch": 2.743181622644756,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00042420075230825515,
      "loss": 1.3118,
      "step": 31884
    },
    {
      "epoch": 2.7432676589520777,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00042417402421618655,
      "loss": 0.5406,
      "step": 31885
    },
    {
      "epoch": 2.7433536952593993,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004241472963458941,
      "loss": 1.6098,
      "step": 31886
    },
    {
      "epoch": 2.7434397315667214,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004241205686974562,
      "loss": 1.1565,
      "step": 31887
    },
    {
      "epoch": 2.743525767874043,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004240938412709508,
      "loss": 0.9244,
      "step": 31888
    },
    {
      "epoch": 2.7436118041813646,
      "grad_norm": 1.109375,
      "learning_rate": 0.000424067114066456,
      "loss": 1.5636,
      "step": 31889
    },
    {
      "epoch": 2.7436978404886863,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00042404038708405035,
      "loss": 1.9703,
      "step": 31890
    },
    {
      "epoch": 2.743783876796008,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004240136603238118,
      "loss": 0.6291,
      "step": 31891
    },
    {
      "epoch": 2.7438699131033295,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004239869337858184,
      "loss": 1.414,
      "step": 31892
    },
    {
      "epoch": 2.743955949410651,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004239602074701484,
      "loss": 1.6431,
      "step": 31893
    },
    {
      "epoch": 2.744041985717973,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00042393348137688007,
      "loss": 1.9894,
      "step": 31894
    },
    {
      "epoch": 2.744128022025295,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004239067555060916,
      "loss": 1.254,
      "step": 31895
    },
    {
      "epoch": 2.7442140583326164,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000423880029857861,
      "loss": 0.937,
      "step": 31896
    },
    {
      "epoch": 2.744300094639938,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004238533044322663,
      "loss": 1.6783,
      "step": 31897
    },
    {
      "epoch": 2.7443861309472597,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000423826579229386,
      "loss": 1.8291,
      "step": 31898
    },
    {
      "epoch": 2.7444721672545813,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00042379985424929824,
      "loss": 1.0645,
      "step": 31899
    },
    {
      "epoch": 2.744558203561903,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004237731294920811,
      "loss": 0.8984,
      "step": 31900
    },
    {
      "epoch": 2.744644239869225,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004237464049578127,
      "loss": 1.1876,
      "step": 31901
    },
    {
      "epoch": 2.7447302761765466,
      "grad_norm": 1.140625,
      "learning_rate": 0.00042371968064657114,
      "loss": 1.0925,
      "step": 31902
    },
    {
      "epoch": 2.7448163124838683,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00042369295655843476,
      "loss": 1.3253,
      "step": 31903
    },
    {
      "epoch": 2.74490234879119,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004236662326934817,
      "loss": 1.1225,
      "step": 31904
    },
    {
      "epoch": 2.7449883850985115,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00042363950905178994,
      "loss": 1.8054,
      "step": 31905
    },
    {
      "epoch": 2.745074421405833,
      "grad_norm": 1.4375,
      "learning_rate": 0.00042361278563343783,
      "loss": 1.8361,
      "step": 31906
    },
    {
      "epoch": 2.7451604577131548,
      "grad_norm": 1.0,
      "learning_rate": 0.0004235860624385035,
      "loss": 1.6206,
      "step": 31907
    },
    {
      "epoch": 2.745246494020477,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00042355933946706503,
      "loss": 1.4645,
      "step": 31908
    },
    {
      "epoch": 2.7453325303277984,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004235326167192006,
      "loss": 0.9246,
      "step": 31909
    },
    {
      "epoch": 2.74541856663512,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00042350589419498847,
      "loss": 1.3349,
      "step": 31910
    },
    {
      "epoch": 2.7455046029424417,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004234791718945067,
      "loss": 1.1432,
      "step": 31911
    },
    {
      "epoch": 2.7455906392497633,
      "grad_norm": 1.046875,
      "learning_rate": 0.00042345244981783337,
      "loss": 1.2894,
      "step": 31912
    },
    {
      "epoch": 2.745676675557085,
      "grad_norm": 1.53125,
      "learning_rate": 0.00042342572796504686,
      "loss": 0.9987,
      "step": 31913
    },
    {
      "epoch": 2.7457627118644066,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004233990063362252,
      "loss": 0.6885,
      "step": 31914
    },
    {
      "epoch": 2.7458487481717286,
      "grad_norm": 1.28125,
      "learning_rate": 0.00042337228493144653,
      "loss": 1.2758,
      "step": 31915
    },
    {
      "epoch": 2.7459347844790503,
      "grad_norm": 1.09375,
      "learning_rate": 0.00042334556375078894,
      "loss": 0.7987,
      "step": 31916
    },
    {
      "epoch": 2.746020820786372,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00042331884279433075,
      "loss": 1.4774,
      "step": 31917
    },
    {
      "epoch": 2.7461068570936935,
      "grad_norm": 0.984375,
      "learning_rate": 0.00042329212206215,
      "loss": 1.5631,
      "step": 31918
    },
    {
      "epoch": 2.746192893401015,
      "grad_norm": 1.03125,
      "learning_rate": 0.00042326540155432493,
      "loss": 1.8503,
      "step": 31919
    },
    {
      "epoch": 2.7462789297083368,
      "grad_norm": 1.515625,
      "learning_rate": 0.00042323868127093354,
      "loss": 0.7978,
      "step": 31920
    },
    {
      "epoch": 2.7463649660156584,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0004232119612120541,
      "loss": 0.9134,
      "step": 31921
    },
    {
      "epoch": 2.7464510023229805,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004231852413777648,
      "loss": 0.9414,
      "step": 31922
    },
    {
      "epoch": 2.746537038630302,
      "grad_norm": 1.34375,
      "learning_rate": 0.00042315852176814364,
      "loss": 1.2658,
      "step": 31923
    },
    {
      "epoch": 2.7466230749376237,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00042313180238326897,
      "loss": 0.7038,
      "step": 31924
    },
    {
      "epoch": 2.7467091112449453,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004231050832232188,
      "loss": 1.423,
      "step": 31925
    },
    {
      "epoch": 2.746795147552267,
      "grad_norm": 1.015625,
      "learning_rate": 0.00042307836428807135,
      "loss": 0.9547,
      "step": 31926
    },
    {
      "epoch": 2.746881183859589,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00042305164557790453,
      "loss": 1.5495,
      "step": 31927
    },
    {
      "epoch": 2.74696722016691,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004230249270927968,
      "loss": 0.8931,
      "step": 31928
    },
    {
      "epoch": 2.7470532564742323,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00042299820883282625,
      "loss": 1.3663,
      "step": 31929
    },
    {
      "epoch": 2.747139292781554,
      "grad_norm": 1.046875,
      "learning_rate": 0.00042297149079807087,
      "loss": 1.1897,
      "step": 31930
    },
    {
      "epoch": 2.7472253290888755,
      "grad_norm": 1.171875,
      "learning_rate": 0.00042294477298860896,
      "loss": 1.5115,
      "step": 31931
    },
    {
      "epoch": 2.747311365396197,
      "grad_norm": 1.15625,
      "learning_rate": 0.00042291805540451853,
      "loss": 0.9798,
      "step": 31932
    },
    {
      "epoch": 2.7473974017035188,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004228913380458779,
      "loss": 0.6251,
      "step": 31933
    },
    {
      "epoch": 2.747483438010841,
      "grad_norm": 1.0,
      "learning_rate": 0.0004228646209127649,
      "loss": 1.7376,
      "step": 31934
    },
    {
      "epoch": 2.747569474318162,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00042283790400525793,
      "loss": 0.5471,
      "step": 31935
    },
    {
      "epoch": 2.747655510625484,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004228111873234353,
      "loss": 1.1517,
      "step": 31936
    },
    {
      "epoch": 2.7477415469328057,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004227844708673748,
      "loss": 1.6244,
      "step": 31937
    },
    {
      "epoch": 2.7478275832401273,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004227577546371547,
      "loss": 1.2041,
      "step": 31938
    },
    {
      "epoch": 2.747913619547449,
      "grad_norm": 0.734375,
      "learning_rate": 0.00042273103863285313,
      "loss": 0.6305,
      "step": 31939
    },
    {
      "epoch": 2.7479996558547706,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004227043228545483,
      "loss": 1.2215,
      "step": 31940
    },
    {
      "epoch": 2.7480856921620926,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004226776073023183,
      "loss": 0.9306,
      "step": 31941
    },
    {
      "epoch": 2.748171728469414,
      "grad_norm": 1.125,
      "learning_rate": 0.0004226508919762413,
      "loss": 0.8515,
      "step": 31942
    },
    {
      "epoch": 2.748257764776736,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004226241768763953,
      "loss": 1.077,
      "step": 31943
    },
    {
      "epoch": 2.7483438010840575,
      "grad_norm": 1.1875,
      "learning_rate": 0.00042259746200285853,
      "loss": 1.9047,
      "step": 31944
    },
    {
      "epoch": 2.748429837391379,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004225707473557092,
      "loss": 1.1801,
      "step": 31945
    },
    {
      "epoch": 2.7485158736987008,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004225440329350253,
      "loss": 1.369,
      "step": 31946
    },
    {
      "epoch": 2.7486019100060224,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004225173187408851,
      "loss": 1.3541,
      "step": 31947
    },
    {
      "epoch": 2.7486879463133445,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004224906047733667,
      "loss": 0.7438,
      "step": 31948
    },
    {
      "epoch": 2.748773982620666,
      "grad_norm": 1.109375,
      "learning_rate": 0.00042246389103254824,
      "loss": 1.426,
      "step": 31949
    },
    {
      "epoch": 2.7488600189279877,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004224371775185076,
      "loss": 1.388,
      "step": 31950
    },
    {
      "epoch": 2.7489460552353093,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004224104642313233,
      "loss": 0.5901,
      "step": 31951
    },
    {
      "epoch": 2.749032091542631,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004223837511710733,
      "loss": 1.6705,
      "step": 31952
    },
    {
      "epoch": 2.7491181278499526,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00042235703833783567,
      "loss": 0.4737,
      "step": 31953
    },
    {
      "epoch": 2.749204164157274,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004223303257316887,
      "loss": 0.9782,
      "step": 31954
    },
    {
      "epoch": 2.7492902004645963,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004223036133527103,
      "loss": 1.4943,
      "step": 31955
    },
    {
      "epoch": 2.749376236771918,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004222769012009788,
      "loss": 1.0461,
      "step": 31956
    },
    {
      "epoch": 2.7494622730792395,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004222501892765722,
      "loss": 1.0467,
      "step": 31957
    },
    {
      "epoch": 2.749548309386561,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004222234775795688,
      "loss": 1.3957,
      "step": 31958
    },
    {
      "epoch": 2.7496343456938828,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004221967661100465,
      "loss": 1.2483,
      "step": 31959
    },
    {
      "epoch": 2.7497203820012044,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00042217005486808347,
      "loss": 1.5699,
      "step": 31960
    },
    {
      "epoch": 2.749806418308526,
      "grad_norm": 11.375,
      "learning_rate": 0.00042214334385375795,
      "loss": 1.7601,
      "step": 31961
    },
    {
      "epoch": 2.749892454615848,
      "grad_norm": 1.03125,
      "learning_rate": 0.000422116633067148,
      "loss": 1.4797,
      "step": 31962
    },
    {
      "epoch": 2.7499784909231697,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004220899225083318,
      "loss": 1.4826,
      "step": 31963
    },
    {
      "epoch": 2.7500645272304913,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00042206321217738734,
      "loss": 0.8728,
      "step": 31964
    },
    {
      "epoch": 2.750150563537813,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00042203650207439295,
      "loss": 1.1592,
      "step": 31965
    },
    {
      "epoch": 2.7502365998451346,
      "grad_norm": 1.1875,
      "learning_rate": 0.00042200979219942655,
      "loss": 0.833,
      "step": 31966
    },
    {
      "epoch": 2.7502365998451346,
      "eval_loss": 1.1257299184799194,
      "eval_runtime": 357.064,
      "eval_samples_per_second": 13.02,
      "eval_steps_per_second": 1.63,
      "step": 31966
    },
    {
      "epoch": 2.750322636152456,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00042198308255256636,
      "loss": 1.5913,
      "step": 31967
    },
    {
      "epoch": 2.750408672459778,
      "grad_norm": 0.796875,
      "learning_rate": 0.00042195637313389035,
      "loss": 1.5655,
      "step": 31968
    },
    {
      "epoch": 2.7504947087671,
      "grad_norm": 1.171875,
      "learning_rate": 0.00042192966394347694,
      "loss": 1.0308,
      "step": 31969
    },
    {
      "epoch": 2.7505807450744215,
      "grad_norm": 1.140625,
      "learning_rate": 0.00042190295498140405,
      "loss": 0.6412,
      "step": 31970
    },
    {
      "epoch": 2.750666781381743,
      "grad_norm": 2.015625,
      "learning_rate": 0.00042187624624774963,
      "loss": 1.7751,
      "step": 31971
    },
    {
      "epoch": 2.7507528176890648,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0004218495377425922,
      "loss": 2.0007,
      "step": 31972
    },
    {
      "epoch": 2.7508388539963864,
      "grad_norm": 1.25,
      "learning_rate": 0.0004218228294660095,
      "loss": 0.5678,
      "step": 31973
    },
    {
      "epoch": 2.750924890303708,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00042179612141808,
      "loss": 1.074,
      "step": 31974
    },
    {
      "epoch": 2.7510109266110296,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00042176941359888166,
      "loss": 0.8551,
      "step": 31975
    },
    {
      "epoch": 2.7510969629183517,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004217427060084924,
      "loss": 1.5713,
      "step": 31976
    },
    {
      "epoch": 2.7511829992256733,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004217159986469906,
      "loss": 1.4409,
      "step": 31977
    },
    {
      "epoch": 2.751269035532995,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00042168929151445433,
      "loss": 1.2309,
      "step": 31978
    },
    {
      "epoch": 2.7513550718403166,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00042166258461096164,
      "loss": 1.0232,
      "step": 31979
    },
    {
      "epoch": 2.751441108147638,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004216358779365905,
      "loss": 1.6564,
      "step": 31980
    },
    {
      "epoch": 2.75152714445496,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00042160917149141935,
      "loss": 1.0751,
      "step": 31981
    },
    {
      "epoch": 2.7516131807622815,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0004215824652755261,
      "loss": 0.9097,
      "step": 31982
    },
    {
      "epoch": 2.7516992170696035,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00042155575928898883,
      "loss": 2.127,
      "step": 31983
    },
    {
      "epoch": 2.751785253376925,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00042152905353188576,
      "loss": 1.5928,
      "step": 31984
    },
    {
      "epoch": 2.7518712896842468,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00042150234800429495,
      "loss": 0.782,
      "step": 31985
    },
    {
      "epoch": 2.7519573259915684,
      "grad_norm": 1.1875,
      "learning_rate": 0.00042147564270629453,
      "loss": 1.234,
      "step": 31986
    },
    {
      "epoch": 2.75204336229889,
      "grad_norm": 2.296875,
      "learning_rate": 0.00042144893763796244,
      "loss": 1.4097,
      "step": 31987
    },
    {
      "epoch": 2.752129398606212,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000421422232799377,
      "loss": 0.9481,
      "step": 31988
    },
    {
      "epoch": 2.7522154349135333,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004213955281906163,
      "loss": 1.2849,
      "step": 31989
    },
    {
      "epoch": 2.7523014712208553,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00042136882381175843,
      "loss": 0.6854,
      "step": 31990
    },
    {
      "epoch": 2.752387507528177,
      "grad_norm": 1.0625,
      "learning_rate": 0.00042134211966288133,
      "loss": 1.4461,
      "step": 31991
    },
    {
      "epoch": 2.7524735438354986,
      "grad_norm": 0.84375,
      "learning_rate": 0.00042131541574406327,
      "loss": 0.8462,
      "step": 31992
    },
    {
      "epoch": 2.75255958014282,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004212887120553824,
      "loss": 1.7401,
      "step": 31993
    },
    {
      "epoch": 2.752645616450142,
      "grad_norm": 1.390625,
      "learning_rate": 0.00042126200859691663,
      "loss": 1.7205,
      "step": 31994
    },
    {
      "epoch": 2.752731652757464,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00042123530536874424,
      "loss": 0.9211,
      "step": 31995
    },
    {
      "epoch": 2.752817689064785,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004212086023709433,
      "loss": 1.4333,
      "step": 31996
    },
    {
      "epoch": 2.752903725372107,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004211818996035918,
      "loss": 1.7016,
      "step": 31997
    },
    {
      "epoch": 2.7529897616794288,
      "grad_norm": 1.359375,
      "learning_rate": 0.00042115519706676784,
      "loss": 1.5146,
      "step": 31998
    },
    {
      "epoch": 2.7530757979867504,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004211284947605497,
      "loss": 1.3068,
      "step": 31999
    },
    {
      "epoch": 2.753161834294072,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004211017926850154,
      "loss": 1.0696,
      "step": 32000
    },
    {
      "epoch": 2.7532478706013936,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00042107509084024286,
      "loss": 1.0674,
      "step": 32001
    },
    {
      "epoch": 2.7533339069087157,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004210483892263105,
      "loss": 1.8261,
      "step": 32002
    },
    {
      "epoch": 2.753419943216037,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004210216878432962,
      "loss": 1.3679,
      "step": 32003
    },
    {
      "epoch": 2.753505979523359,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004209949866912781,
      "loss": 1.3862,
      "step": 32004
    },
    {
      "epoch": 2.7535920158306806,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004209682857703342,
      "loss": 0.9804,
      "step": 32005
    },
    {
      "epoch": 2.753678052138002,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004209415850805428,
      "loss": 1.9126,
      "step": 32006
    },
    {
      "epoch": 2.753764088445324,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004209148846219819,
      "loss": 1.5618,
      "step": 32007
    },
    {
      "epoch": 2.7538501247526455,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00042088818439472937,
      "loss": 1.1771,
      "step": 32008
    },
    {
      "epoch": 2.7539361610599675,
      "grad_norm": 1.03125,
      "learning_rate": 0.00042086148439886376,
      "loss": 0.8614,
      "step": 32009
    },
    {
      "epoch": 2.754022197367289,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00042083478463446274,
      "loss": 1.2836,
      "step": 32010
    },
    {
      "epoch": 2.7541082336746108,
      "grad_norm": 1.0,
      "learning_rate": 0.00042080808510160475,
      "loss": 1.1951,
      "step": 32011
    },
    {
      "epoch": 2.7541942699819324,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00042078138580036763,
      "loss": 0.5271,
      "step": 32012
    },
    {
      "epoch": 2.754280306289254,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004207546867308296,
      "loss": 3.1989,
      "step": 32013
    },
    {
      "epoch": 2.7543663425965756,
      "grad_norm": 1.703125,
      "learning_rate": 0.0004207279878930686,
      "loss": 1.7498,
      "step": 32014
    },
    {
      "epoch": 2.7544523789038973,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0004207012892871629,
      "loss": 1.1957,
      "step": 32015
    },
    {
      "epoch": 2.7545384152112193,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004206745909131905,
      "loss": 0.6984,
      "step": 32016
    },
    {
      "epoch": 2.754624451518541,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004206478927712294,
      "loss": 0.6713,
      "step": 32017
    },
    {
      "epoch": 2.7547104878258626,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00042062119486135794,
      "loss": 1.1153,
      "step": 32018
    },
    {
      "epoch": 2.754796524133184,
      "grad_norm": 1.09375,
      "learning_rate": 0.00042059449718365394,
      "loss": 1.0328,
      "step": 32019
    },
    {
      "epoch": 2.754882560440506,
      "grad_norm": 1.328125,
      "learning_rate": 0.00042056779973819564,
      "loss": 0.8139,
      "step": 32020
    },
    {
      "epoch": 2.7549685967478275,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000420541102525061,
      "loss": 1.6769,
      "step": 32021
    },
    {
      "epoch": 2.755054633055149,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004205144055443283,
      "loss": 0.7307,
      "step": 32022
    },
    {
      "epoch": 2.755140669362471,
      "grad_norm": 0.75,
      "learning_rate": 0.0004204877087960755,
      "loss": 1.7127,
      "step": 32023
    },
    {
      "epoch": 2.7552267056697928,
      "grad_norm": 1.375,
      "learning_rate": 0.00042046101228038045,
      "loss": 0.6288,
      "step": 32024
    },
    {
      "epoch": 2.7553127419771144,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00042043431599732175,
      "loss": 1.2975,
      "step": 32025
    },
    {
      "epoch": 2.755398778284436,
      "grad_norm": 5.03125,
      "learning_rate": 0.0004204076199469771,
      "loss": 1.6592,
      "step": 32026
    },
    {
      "epoch": 2.7554848145917576,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004203809241294247,
      "loss": 0.8994,
      "step": 32027
    },
    {
      "epoch": 2.7555708508990793,
      "grad_norm": 1.203125,
      "learning_rate": 0.00042035422854474255,
      "loss": 1.1739,
      "step": 32028
    },
    {
      "epoch": 2.755656887206401,
      "grad_norm": 1.46875,
      "learning_rate": 0.00042032753319300886,
      "loss": 0.8973,
      "step": 32029
    },
    {
      "epoch": 2.755742923513723,
      "grad_norm": 0.921875,
      "learning_rate": 0.00042030083807430166,
      "loss": 1.404,
      "step": 32030
    },
    {
      "epoch": 2.7558289598210446,
      "grad_norm": 1.140625,
      "learning_rate": 0.00042027414318869896,
      "loss": 0.9837,
      "step": 32031
    },
    {
      "epoch": 2.755914996128366,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004202474485362788,
      "loss": 0.986,
      "step": 32032
    },
    {
      "epoch": 2.756001032435688,
      "grad_norm": 1.140625,
      "learning_rate": 0.00042022075411711946,
      "loss": 0.9694,
      "step": 32033
    },
    {
      "epoch": 2.7560870687430095,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004201940599312989,
      "loss": 1.3265,
      "step": 32034
    },
    {
      "epoch": 2.756173105050331,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004201673659788951,
      "loss": 1.043,
      "step": 32035
    },
    {
      "epoch": 2.7562591413576527,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00042014067225998634,
      "loss": 0.9052,
      "step": 32036
    },
    {
      "epoch": 2.7563451776649748,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00042011397877465053,
      "loss": 0.8699,
      "step": 32037
    },
    {
      "epoch": 2.7564312139722964,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004200872855229658,
      "loss": 1.114,
      "step": 32038
    },
    {
      "epoch": 2.756517250279618,
      "grad_norm": 1.171875,
      "learning_rate": 0.00042006059250501016,
      "loss": 1.6589,
      "step": 32039
    },
    {
      "epoch": 2.7566032865869396,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00042003389972086183,
      "loss": 0.8172,
      "step": 32040
    },
    {
      "epoch": 2.7566893228942613,
      "grad_norm": 0.875,
      "learning_rate": 0.0004200072071705988,
      "loss": 0.9178,
      "step": 32041
    },
    {
      "epoch": 2.7567753592015833,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000419980514854299,
      "loss": 0.8324,
      "step": 32042
    },
    {
      "epoch": 2.7568613955089045,
      "grad_norm": 1.015625,
      "learning_rate": 0.00041995382277204075,
      "loss": 0.8845,
      "step": 32043
    },
    {
      "epoch": 2.7569474318162266,
      "grad_norm": 1.3125,
      "learning_rate": 0.000419927130923902,
      "loss": 1.3883,
      "step": 32044
    },
    {
      "epoch": 2.757033468123548,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004199004393099606,
      "loss": 1.1886,
      "step": 32045
    },
    {
      "epoch": 2.75711950443087,
      "grad_norm": 1.109375,
      "learning_rate": 0.00041987374793029503,
      "loss": 0.8716,
      "step": 32046
    },
    {
      "epoch": 2.7572055407381915,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00041984705678498307,
      "loss": 1.9861,
      "step": 32047
    },
    {
      "epoch": 2.757291577045513,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000419820365874103,
      "loss": 1.0766,
      "step": 32048
    },
    {
      "epoch": 2.757377613352835,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00041979367519773275,
      "loss": 0.9677,
      "step": 32049
    },
    {
      "epoch": 2.7574636496601563,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004197669847559504,
      "loss": 1.1543,
      "step": 32050
    },
    {
      "epoch": 2.7575496859674784,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00041974029454883393,
      "loss": 1.3363,
      "step": 32051
    },
    {
      "epoch": 2.7576357222748,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004197136045764616,
      "loss": 1.3666,
      "step": 32052
    },
    {
      "epoch": 2.7577217585821217,
      "grad_norm": 0.875,
      "learning_rate": 0.00041968691483891134,
      "loss": 0.7069,
      "step": 32053
    },
    {
      "epoch": 2.7578077948894433,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00041966022533626124,
      "loss": 1.3544,
      "step": 32054
    },
    {
      "epoch": 2.757893831196765,
      "grad_norm": 1.390625,
      "learning_rate": 0.00041963353606858924,
      "loss": 1.1301,
      "step": 32055
    },
    {
      "epoch": 2.757979867504087,
      "grad_norm": 1.484375,
      "learning_rate": 0.00041960684703597364,
      "loss": 1.7018,
      "step": 32056
    },
    {
      "epoch": 2.758065903811408,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00041958015823849237,
      "loss": 1.7613,
      "step": 32057
    },
    {
      "epoch": 2.75815194011873,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00041955346967622345,
      "loss": 0.7834,
      "step": 32058
    },
    {
      "epoch": 2.758237976426052,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000419526781349245,
      "loss": 1.5134,
      "step": 32059
    },
    {
      "epoch": 2.7583240127333735,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0004195000932576351,
      "loss": 1.2453,
      "step": 32060
    },
    {
      "epoch": 2.758410049040695,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004194734054014718,
      "loss": 1.4843,
      "step": 32061
    },
    {
      "epoch": 2.7584960853480167,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00041944671778083295,
      "loss": 1.0766,
      "step": 32062
    },
    {
      "epoch": 2.758582121655339,
      "grad_norm": 1.078125,
      "learning_rate": 0.00041942003039579693,
      "loss": 1.24,
      "step": 32063
    },
    {
      "epoch": 2.7586681579626604,
      "grad_norm": 1.078125,
      "learning_rate": 0.00041939334324644163,
      "loss": 1.2463,
      "step": 32064
    },
    {
      "epoch": 2.758754194269982,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000419366656332845,
      "loss": 0.7832,
      "step": 32065
    },
    {
      "epoch": 2.7588402305773037,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0004193399696550854,
      "loss": 1.4731,
      "step": 32066
    },
    {
      "epoch": 2.7589262668846253,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004193132832132407,
      "loss": 1.3264,
      "step": 32067
    },
    {
      "epoch": 2.759012303191947,
      "grad_norm": 1.390625,
      "learning_rate": 0.00041928659700738883,
      "loss": 0.8987,
      "step": 32068
    },
    {
      "epoch": 2.7590983394992685,
      "grad_norm": 1.421875,
      "learning_rate": 0.000419259911037608,
      "loss": 1.4072,
      "step": 32069
    },
    {
      "epoch": 2.7591843758065906,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004192332253039763,
      "loss": 1.4722,
      "step": 32070
    },
    {
      "epoch": 2.759270412113912,
      "grad_norm": 0.859375,
      "learning_rate": 0.00041920653980657166,
      "loss": 1.082,
      "step": 32071
    },
    {
      "epoch": 2.759356448421234,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0004191798545454722,
      "loss": 1.1702,
      "step": 32072
    },
    {
      "epoch": 2.7594424847285555,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00041915316952075585,
      "loss": 1.2377,
      "step": 32073
    },
    {
      "epoch": 2.759528521035877,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004191264847325008,
      "loss": 0.7988,
      "step": 32074
    },
    {
      "epoch": 2.7596145573431987,
      "grad_norm": 1.0625,
      "learning_rate": 0.00041909980018078516,
      "loss": 1.5367,
      "step": 32075
    },
    {
      "epoch": 2.7597005936505203,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004190731158656867,
      "loss": 1.3877,
      "step": 32076
    },
    {
      "epoch": 2.7597866299578424,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00041904643178728373,
      "loss": 0.8511,
      "step": 32077
    },
    {
      "epoch": 2.759872666265164,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004190197479456542,
      "loss": 0.7051,
      "step": 32078
    },
    {
      "epoch": 2.7599587025724857,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00041899306434087626,
      "loss": 0.6576,
      "step": 32079
    },
    {
      "epoch": 2.7600447388798073,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00041896638097302765,
      "loss": 0.7673,
      "step": 32080
    },
    {
      "epoch": 2.760130775187129,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0004189396978421867,
      "loss": 1.1,
      "step": 32081
    },
    {
      "epoch": 2.7602168114944505,
      "grad_norm": 1.03125,
      "learning_rate": 0.00041891301494843125,
      "loss": 1.4261,
      "step": 32082
    },
    {
      "epoch": 2.760302847801772,
      "grad_norm": 1.25,
      "learning_rate": 0.0004188863322918396,
      "loss": 2.195,
      "step": 32083
    },
    {
      "epoch": 2.760388884109094,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004188596498724897,
      "loss": 1.1988,
      "step": 32084
    },
    {
      "epoch": 2.760474920416416,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00041883296769045934,
      "loss": 0.8184,
      "step": 32085
    },
    {
      "epoch": 2.7605609567237375,
      "grad_norm": 1.078125,
      "learning_rate": 0.000418806285745827,
      "loss": 1.1116,
      "step": 32086
    },
    {
      "epoch": 2.760646993031059,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004187796040386704,
      "loss": 0.5853,
      "step": 32087
    },
    {
      "epoch": 2.7607330293383807,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00041875292256906763,
      "loss": 0.6098,
      "step": 32088
    },
    {
      "epoch": 2.7608190656457023,
      "grad_norm": 0.953125,
      "learning_rate": 0.00041872624133709675,
      "loss": 0.6521,
      "step": 32089
    },
    {
      "epoch": 2.760905101953024,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004186995603428359,
      "loss": 1.8908,
      "step": 32090
    },
    {
      "epoch": 2.760991138260346,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000418672879586363,
      "loss": 2.1319,
      "step": 32091
    },
    {
      "epoch": 2.7610771745676677,
      "grad_norm": 1.234375,
      "learning_rate": 0.000418646199067756,
      "loss": 1.7181,
      "step": 32092
    },
    {
      "epoch": 2.7611632108749893,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00041861951878709316,
      "loss": 1.3681,
      "step": 32093
    },
    {
      "epoch": 2.761249247182311,
      "grad_norm": 1.25,
      "learning_rate": 0.0004185928387444524,
      "loss": 1.8083,
      "step": 32094
    },
    {
      "epoch": 2.7613352834896325,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004185661589399117,
      "loss": 1.3917,
      "step": 32095
    },
    {
      "epoch": 2.761421319796954,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004185394793735491,
      "loss": 0.6045,
      "step": 32096
    },
    {
      "epoch": 2.761507356104276,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00041851280004544277,
      "loss": 1.623,
      "step": 32097
    },
    {
      "epoch": 2.761593392411598,
      "grad_norm": 1.34375,
      "learning_rate": 0.00041848612095567064,
      "loss": 1.1964,
      "step": 32098
    },
    {
      "epoch": 2.7616794287189195,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00041845944210431067,
      "loss": 1.6949,
      "step": 32099
    },
    {
      "epoch": 2.761765465026241,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00041843276349144107,
      "loss": 1.1833,
      "step": 32100
    },
    {
      "epoch": 2.7618515013335627,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00041840608511713975,
      "loss": 1.2903,
      "step": 32101
    },
    {
      "epoch": 2.7619375376408843,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004183794069814848,
      "loss": 1.6116,
      "step": 32102
    },
    {
      "epoch": 2.7620235739482064,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00041835272908455405,
      "loss": 0.9899,
      "step": 32103
    },
    {
      "epoch": 2.7621096102555276,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0004183260514264259,
      "loss": 0.8101,
      "step": 32104
    },
    {
      "epoch": 2.7621956465628497,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00041829937400717805,
      "loss": 1.0686,
      "step": 32105
    },
    {
      "epoch": 2.7622816828701713,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00041827269682688864,
      "loss": 2.4915,
      "step": 32106
    },
    {
      "epoch": 2.762367719177493,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004182460198856357,
      "loss": 1.4656,
      "step": 32107
    },
    {
      "epoch": 2.7624537554848145,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0004182193431834973,
      "loss": 0.6793,
      "step": 32108
    },
    {
      "epoch": 2.762539791792136,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004181926667205514,
      "loss": 1.7768,
      "step": 32109
    },
    {
      "epoch": 2.7626258280994582,
      "grad_norm": 0.96875,
      "learning_rate": 0.000418165990496876,
      "loss": 1.5667,
      "step": 32110
    },
    {
      "epoch": 2.7627118644067794,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004181393145125492,
      "loss": 1.447,
      "step": 32111
    },
    {
      "epoch": 2.7627979007141015,
      "grad_norm": 1.125,
      "learning_rate": 0.0004181126387676489,
      "loss": 1.4698,
      "step": 32112
    },
    {
      "epoch": 2.762883937021423,
      "grad_norm": 1.125,
      "learning_rate": 0.0004180859632622532,
      "loss": 1.4028,
      "step": 32113
    },
    {
      "epoch": 2.7629699733287447,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00041805928799644026,
      "loss": 0.7468,
      "step": 32114
    },
    {
      "epoch": 2.7630560096360663,
      "grad_norm": 0.875,
      "learning_rate": 0.00041803261297028796,
      "loss": 0.6199,
      "step": 32115
    },
    {
      "epoch": 2.763142045943388,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004180059381838743,
      "loss": 1.2977,
      "step": 32116
    },
    {
      "epoch": 2.76322808225071,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00041797926363727723,
      "loss": 1.7651,
      "step": 32117
    },
    {
      "epoch": 2.763314118558031,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004179525893305749,
      "loss": 1.3468,
      "step": 32118
    },
    {
      "epoch": 2.7634001548653533,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00041792591526384516,
      "loss": 1.1508,
      "step": 32119
    },
    {
      "epoch": 2.763486191172675,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004178992414371663,
      "loss": 1.8154,
      "step": 32120
    },
    {
      "epoch": 2.7635722274799965,
      "grad_norm": 1.15625,
      "learning_rate": 0.00041787256785061626,
      "loss": 1.3579,
      "step": 32121
    },
    {
      "epoch": 2.763658263787318,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00041784589450427287,
      "loss": 1.3383,
      "step": 32122
    },
    {
      "epoch": 2.76374430009464,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004178192213982143,
      "loss": 0.7487,
      "step": 32123
    },
    {
      "epoch": 2.763830336401962,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00041779254853251856,
      "loss": 1.5863,
      "step": 32124
    },
    {
      "epoch": 2.7639163727092835,
      "grad_norm": 0.96875,
      "learning_rate": 0.00041776587590726367,
      "loss": 1.1815,
      "step": 32125
    },
    {
      "epoch": 2.764002409016605,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004177392035225275,
      "loss": 0.9358,
      "step": 32126
    },
    {
      "epoch": 2.7640884453239267,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004177125313783882,
      "loss": 1.5298,
      "step": 32127
    },
    {
      "epoch": 2.7641744816312483,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004176858594749238,
      "loss": 0.847,
      "step": 32128
    },
    {
      "epoch": 2.76426051793857,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004176591878122121,
      "loss": 1.2912,
      "step": 32129
    },
    {
      "epoch": 2.7643465542458916,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004176325163903314,
      "loss": 0.9289,
      "step": 32130
    },
    {
      "epoch": 2.7644325905532137,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004176058452093595,
      "loss": 1.3998,
      "step": 32131
    },
    {
      "epoch": 2.7645186268605353,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00041757917426937455,
      "loss": 1.2271,
      "step": 32132
    },
    {
      "epoch": 2.764604663167857,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004175525035704544,
      "loss": 0.6867,
      "step": 32133
    },
    {
      "epoch": 2.7646906994751785,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004175258331126772,
      "loss": 1.8574,
      "step": 32134
    },
    {
      "epoch": 2.7647767357825,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0004174991628961209,
      "loss": 1.0802,
      "step": 32135
    },
    {
      "epoch": 2.764862772089822,
      "grad_norm": 0.828125,
      "learning_rate": 0.00041747249292086336,
      "loss": 0.9993,
      "step": 32136
    },
    {
      "epoch": 2.7649488083971434,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004174458231869829,
      "loss": 1.4284,
      "step": 32137
    },
    {
      "epoch": 2.7650348447044655,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004174191536945573,
      "loss": 1.9874,
      "step": 32138
    },
    {
      "epoch": 2.765120881011787,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004173924844436647,
      "loss": 0.7343,
      "step": 32139
    },
    {
      "epoch": 2.7652069173191087,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004173658154343828,
      "loss": 1.741,
      "step": 32140
    },
    {
      "epoch": 2.7652929536264304,
      "grad_norm": 1.203125,
      "learning_rate": 0.00041733914666679,
      "loss": 1.3336,
      "step": 32141
    },
    {
      "epoch": 2.765378989933752,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004173124781409641,
      "loss": 0.4948,
      "step": 32142
    },
    {
      "epoch": 2.7654650262410736,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004172858098569831,
      "loss": 0.769,
      "step": 32143
    },
    {
      "epoch": 2.7655510625483952,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000417259141814925,
      "loss": 0.784,
      "step": 32144
    },
    {
      "epoch": 2.7656370988557173,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004172324740148678,
      "loss": 1.3242,
      "step": 32145
    },
    {
      "epoch": 2.765723135163039,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004172058064568895,
      "loss": 1.2847,
      "step": 32146
    },
    {
      "epoch": 2.7658091714703605,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00041717913914106813,
      "loss": 1.3953,
      "step": 32147
    },
    {
      "epoch": 2.765895207777682,
      "grad_norm": 1.953125,
      "learning_rate": 0.00041715247206748174,
      "loss": 0.992,
      "step": 32148
    },
    {
      "epoch": 2.765981244085004,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00041712580523620824,
      "loss": 0.5799,
      "step": 32149
    },
    {
      "epoch": 2.7660672803923254,
      "grad_norm": 1.046875,
      "learning_rate": 0.00041709913864732565,
      "loss": 1.268,
      "step": 32150
    },
    {
      "epoch": 2.766153316699647,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0004170724723009118,
      "loss": 1.3135,
      "step": 32151
    },
    {
      "epoch": 2.766239353006969,
      "grad_norm": 0.859375,
      "learning_rate": 0.000417045806197045,
      "loss": 0.7592,
      "step": 32152
    },
    {
      "epoch": 2.7663253893142907,
      "grad_norm": 1.125,
      "learning_rate": 0.00041701914033580306,
      "loss": 1.0558,
      "step": 32153
    },
    {
      "epoch": 2.7664114256216124,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004169924747172639,
      "loss": 1.2428,
      "step": 32154
    },
    {
      "epoch": 2.766497461928934,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00041696580934150575,
      "loss": 1.0225,
      "step": 32155
    },
    {
      "epoch": 2.7665834982362556,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00041693914420860617,
      "loss": 0.955,
      "step": 32156
    },
    {
      "epoch": 2.7666695345435772,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004169124793186437,
      "loss": 0.7677,
      "step": 32157
    },
    {
      "epoch": 2.766755570850899,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004168858146716961,
      "loss": 1.3283,
      "step": 32158
    },
    {
      "epoch": 2.766841607158221,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00041685915026784114,
      "loss": 1.6766,
      "step": 32159
    },
    {
      "epoch": 2.7669276434655425,
      "grad_norm": 1.671875,
      "learning_rate": 0.00041683248610715714,
      "loss": 1.8379,
      "step": 32160
    },
    {
      "epoch": 2.767013679772864,
      "grad_norm": 1.234375,
      "learning_rate": 0.000416805822189722,
      "loss": 0.9451,
      "step": 32161
    },
    {
      "epoch": 2.767099716080186,
      "grad_norm": 1.03125,
      "learning_rate": 0.00041677915851561355,
      "loss": 2.0875,
      "step": 32162
    },
    {
      "epoch": 2.7671857523875074,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004167524950849098,
      "loss": 0.8855,
      "step": 32163
    },
    {
      "epoch": 2.7672717886948295,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004167258318976889,
      "loss": 0.455,
      "step": 32164
    },
    {
      "epoch": 2.7673578250021507,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004166991689540288,
      "loss": 0.9367,
      "step": 32165
    },
    {
      "epoch": 2.7674438613094727,
      "grad_norm": 1.015625,
      "learning_rate": 0.00041667250625400745,
      "loss": 1.0302,
      "step": 32166
    },
    {
      "epoch": 2.7675298976167944,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00041664584379770266,
      "loss": 1.2792,
      "step": 32167
    },
    {
      "epoch": 2.767615933924116,
      "grad_norm": 1.015625,
      "learning_rate": 0.00041661918158519265,
      "loss": 1.3529,
      "step": 32168
    },
    {
      "epoch": 2.7677019702314376,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004165925196165553,
      "loss": 1.1239,
      "step": 32169
    },
    {
      "epoch": 2.7677880065387592,
      "grad_norm": 1.296875,
      "learning_rate": 0.0004165658578918686,
      "loss": 1.3365,
      "step": 32170
    },
    {
      "epoch": 2.7678740428460813,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004165391964112106,
      "loss": 0.6873,
      "step": 32171
    },
    {
      "epoch": 2.7679600791534025,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004165125351746592,
      "loss": 0.9126,
      "step": 32172
    },
    {
      "epoch": 2.7680461154607245,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004164858741822924,
      "loss": 0.9611,
      "step": 32173
    },
    {
      "epoch": 2.768132151768046,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00041645921343418807,
      "loss": 0.742,
      "step": 32174
    },
    {
      "epoch": 2.768218188075368,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0004164325529304244,
      "loss": 0.758,
      "step": 32175
    },
    {
      "epoch": 2.7683042243826894,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00041640589267107926,
      "loss": 0.8668,
      "step": 32176
    },
    {
      "epoch": 2.768390260690011,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004163792326562305,
      "loss": 0.8793,
      "step": 32177
    },
    {
      "epoch": 2.768476296997333,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004163525728859563,
      "loss": 1.6714,
      "step": 32178
    },
    {
      "epoch": 2.7685623333046543,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004163259133603346,
      "loss": 1.0228,
      "step": 32179
    },
    {
      "epoch": 2.7686483696119764,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00041629925407944343,
      "loss": 1.2823,
      "step": 32180
    },
    {
      "epoch": 2.768734405919298,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004162725950433605,
      "loss": 1.6992,
      "step": 32181
    },
    {
      "epoch": 2.7688204422266196,
      "grad_norm": 0.734375,
      "learning_rate": 0.00041624593625216394,
      "loss": 0.5529,
      "step": 32182
    },
    {
      "epoch": 2.7689064785339412,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004162192777059318,
      "loss": 0.7099,
      "step": 32183
    },
    {
      "epoch": 2.768992514841263,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000416192619404742,
      "loss": 1.245,
      "step": 32184
    },
    {
      "epoch": 2.769078551148585,
      "grad_norm": 1.328125,
      "learning_rate": 0.00041616596134867235,
      "loss": 0.9875,
      "step": 32185
    },
    {
      "epoch": 2.7691645874559065,
      "grad_norm": 1.296875,
      "learning_rate": 0.00041613930353780106,
      "loss": 0.6231,
      "step": 32186
    },
    {
      "epoch": 2.769250623763228,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000416112645972206,
      "loss": 0.8209,
      "step": 32187
    },
    {
      "epoch": 2.76933666007055,
      "grad_norm": 1.09375,
      "learning_rate": 0.00041608598865196505,
      "loss": 1.3079,
      "step": 32188
    },
    {
      "epoch": 2.7694226963778714,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00041605933157715634,
      "loss": 1.4644,
      "step": 32189
    },
    {
      "epoch": 2.769508732685193,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00041603267474785784,
      "loss": 0.9622,
      "step": 32190
    },
    {
      "epoch": 2.7695947689925147,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00041600601816414734,
      "loss": 0.7701,
      "step": 32191
    },
    {
      "epoch": 2.7696808052998367,
      "grad_norm": 1.125,
      "learning_rate": 0.00041597936182610285,
      "loss": 1.4145,
      "step": 32192
    },
    {
      "epoch": 2.7697668416071584,
      "grad_norm": 1.46875,
      "learning_rate": 0.00041595270573380227,
      "loss": 1.0613,
      "step": 32193
    },
    {
      "epoch": 2.76985287791448,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004159260498873239,
      "loss": 1.2901,
      "step": 32194
    },
    {
      "epoch": 2.7699389142218016,
      "grad_norm": 1.234375,
      "learning_rate": 0.00041589939428674546,
      "loss": 1.2848,
      "step": 32195
    },
    {
      "epoch": 2.7700249505291232,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0004158727389321449,
      "loss": 1.1366,
      "step": 32196
    },
    {
      "epoch": 2.770110986836445,
      "grad_norm": 1.25,
      "learning_rate": 0.0004158460838236002,
      "loss": 1.3915,
      "step": 32197
    },
    {
      "epoch": 2.7701970231437665,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004158194289611894,
      "loss": 1.7754,
      "step": 32198
    },
    {
      "epoch": 2.7702830594510885,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00041579277434499037,
      "loss": 1.0156,
      "step": 32199
    },
    {
      "epoch": 2.77036909575841,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00041576611997508095,
      "loss": 1.5182,
      "step": 32200
    },
    {
      "epoch": 2.770455132065732,
      "grad_norm": 11.5625,
      "learning_rate": 0.00041573946585153946,
      "loss": 1.0572,
      "step": 32201
    },
    {
      "epoch": 2.7705411683730534,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004157128119744436,
      "loss": 0.9032,
      "step": 32202
    },
    {
      "epoch": 2.770627204680375,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00041568615834387134,
      "loss": 1.1661,
      "step": 32203
    },
    {
      "epoch": 2.7707132409876967,
      "grad_norm": 0.8125,
      "learning_rate": 0.00041565950495990063,
      "loss": 0.7633,
      "step": 32204
    },
    {
      "epoch": 2.7707992772950183,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0004156328518226095,
      "loss": 0.4082,
      "step": 32205
    },
    {
      "epoch": 2.7708853136023404,
      "grad_norm": 1.0625,
      "learning_rate": 0.00041560619893207595,
      "loss": 1.1687,
      "step": 32206
    },
    {
      "epoch": 2.770971349909662,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004155795462883778,
      "loss": 1.4583,
      "step": 32207
    },
    {
      "epoch": 2.7710573862169836,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00041555289389159293,
      "loss": 1.4663,
      "step": 32208
    },
    {
      "epoch": 2.7711434225243052,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004155262417417996,
      "loss": 1.7744,
      "step": 32209
    },
    {
      "epoch": 2.771229458831627,
      "grad_norm": 1.125,
      "learning_rate": 0.00041549958983907553,
      "loss": 1.1459,
      "step": 32210
    },
    {
      "epoch": 2.7713154951389485,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0004154729381834986,
      "loss": 1.147,
      "step": 32211
    },
    {
      "epoch": 2.77140153144627,
      "grad_norm": 1.15625,
      "learning_rate": 0.00041544628677514707,
      "loss": 1.7986,
      "step": 32212
    },
    {
      "epoch": 2.771487567753592,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00041541963561409867,
      "loss": 0.9013,
      "step": 32213
    },
    {
      "epoch": 2.771573604060914,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004153929847004314,
      "loss": 1.3065,
      "step": 32214
    },
    {
      "epoch": 2.7716596403682354,
      "grad_norm": 1.484375,
      "learning_rate": 0.00041536633403422305,
      "loss": 1.1188,
      "step": 32215
    },
    {
      "epoch": 2.771745676675557,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004153396836155518,
      "loss": 1.0446,
      "step": 32216
    },
    {
      "epoch": 2.7718317129828787,
      "grad_norm": 1.0,
      "learning_rate": 0.00041531303344449554,
      "loss": 1.5208,
      "step": 32217
    },
    {
      "epoch": 2.7719177492902007,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004152863835211321,
      "loss": 1.2972,
      "step": 32218
    },
    {
      "epoch": 2.772003785597522,
      "grad_norm": 1.09375,
      "learning_rate": 0.00041525973384553964,
      "loss": 1.7642,
      "step": 32219
    },
    {
      "epoch": 2.772089821904844,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004152330844177959,
      "loss": 1.1332,
      "step": 32220
    },
    {
      "epoch": 2.7721758582121656,
      "grad_norm": 1.25,
      "learning_rate": 0.00041520643523797895,
      "loss": 1.4349,
      "step": 32221
    },
    {
      "epoch": 2.7722618945194872,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004151797863061666,
      "loss": 1.7364,
      "step": 32222
    },
    {
      "epoch": 2.772347930826809,
      "grad_norm": 0.921875,
      "learning_rate": 0.0004151531376224369,
      "loss": 1.5369,
      "step": 32223
    },
    {
      "epoch": 2.7724339671341305,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00041512648918686786,
      "loss": 1.8381,
      "step": 32224
    },
    {
      "epoch": 2.7725200034414526,
      "grad_norm": 0.875,
      "learning_rate": 0.0004150998409995372,
      "loss": 0.9467,
      "step": 32225
    },
    {
      "epoch": 2.7726060397487737,
      "grad_norm": 1.234375,
      "learning_rate": 0.00041507319306052304,
      "loss": 1.2059,
      "step": 32226
    },
    {
      "epoch": 2.772692076056096,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004150465453699034,
      "loss": 0.5828,
      "step": 32227
    },
    {
      "epoch": 2.7727781123634174,
      "grad_norm": 1.171875,
      "learning_rate": 0.00041501989792775595,
      "loss": 1.2017,
      "step": 32228
    },
    {
      "epoch": 2.772864148670739,
      "grad_norm": 1.0625,
      "learning_rate": 0.00041499325073415875,
      "loss": 1.1065,
      "step": 32229
    },
    {
      "epoch": 2.7729501849780607,
      "grad_norm": 1.28125,
      "learning_rate": 0.00041496660378918977,
      "loss": 2.1676,
      "step": 32230
    },
    {
      "epoch": 2.7730362212853823,
      "grad_norm": 0.859375,
      "learning_rate": 0.00041493995709292685,
      "loss": 1.4954,
      "step": 32231
    },
    {
      "epoch": 2.7731222575927044,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00041491331064544827,
      "loss": 0.8753,
      "step": 32232
    },
    {
      "epoch": 2.7732082939000255,
      "grad_norm": 0.8125,
      "learning_rate": 0.00041488666444683154,
      "loss": 0.8429,
      "step": 32233
    },
    {
      "epoch": 2.7732943302073476,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004148600184971548,
      "loss": 0.8073,
      "step": 32234
    },
    {
      "epoch": 2.7733803665146692,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0004148333727964959,
      "loss": 0.9263,
      "step": 32235
    },
    {
      "epoch": 2.773466402821991,
      "grad_norm": 1.234375,
      "learning_rate": 0.00041480672734493297,
      "loss": 0.8161,
      "step": 32236
    },
    {
      "epoch": 2.7735524391293125,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004147800821425437,
      "loss": 0.7952,
      "step": 32237
    },
    {
      "epoch": 2.773638475436634,
      "grad_norm": 0.8125,
      "learning_rate": 0.00041475343718940606,
      "loss": 0.9584,
      "step": 32238
    },
    {
      "epoch": 2.773724511743956,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00041472679248559806,
      "loss": 1.6828,
      "step": 32239
    },
    {
      "epoch": 2.773810548051278,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004147001480311977,
      "loss": 1.4228,
      "step": 32240
    },
    {
      "epoch": 2.7738965843585994,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004146735038262827,
      "loss": 1.048,
      "step": 32241
    },
    {
      "epoch": 2.773982620665921,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00041464685987093115,
      "loss": 1.3547,
      "step": 32242
    },
    {
      "epoch": 2.7740686569732427,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000414620216165221,
      "loss": 1.0536,
      "step": 32243
    },
    {
      "epoch": 2.7741546932805643,
      "grad_norm": 1.4375,
      "learning_rate": 0.0004145935727092301,
      "loss": 2.0922,
      "step": 32244
    },
    {
      "epoch": 2.774240729587886,
      "grad_norm": 1.125,
      "learning_rate": 0.0004145669295030363,
      "loss": 1.0857,
      "step": 32245
    },
    {
      "epoch": 2.774326765895208,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004145402865467176,
      "loss": 1.9118,
      "step": 32246
    },
    {
      "epoch": 2.7744128022025296,
      "grad_norm": 1.125,
      "learning_rate": 0.00041451364384035205,
      "loss": 1.2834,
      "step": 32247
    },
    {
      "epoch": 2.7744988385098512,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00041448700138401747,
      "loss": 0.9132,
      "step": 32248
    },
    {
      "epoch": 2.774584874817173,
      "grad_norm": 31.25,
      "learning_rate": 0.00041446035917779163,
      "loss": 0.9336,
      "step": 32249
    },
    {
      "epoch": 2.7746709111244945,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004144337172217527,
      "loss": 1.0391,
      "step": 32250
    },
    {
      "epoch": 2.774756947431816,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00041440707551597855,
      "loss": 0.8373,
      "step": 32251
    },
    {
      "epoch": 2.7748429837391377,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000414380434060547,
      "loss": 0.5689,
      "step": 32252
    },
    {
      "epoch": 2.77492902004646,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00041435379285553606,
      "loss": 0.9853,
      "step": 32253
    },
    {
      "epoch": 2.7750150563537814,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004143271519010236,
      "loss": 1.4827,
      "step": 32254
    },
    {
      "epoch": 2.775101092661103,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00041430051119708763,
      "loss": 0.91,
      "step": 32255
    },
    {
      "epoch": 2.7751871289684247,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00041427387074380583,
      "loss": 1.171,
      "step": 32256
    },
    {
      "epoch": 2.7752731652757463,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004142472305412564,
      "loss": 1.0382,
      "step": 32257
    },
    {
      "epoch": 2.775359201583068,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00041422059058951715,
      "loss": 1.134,
      "step": 32258
    },
    {
      "epoch": 2.7754452378903895,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00041419395088866586,
      "loss": 1.2626,
      "step": 32259
    },
    {
      "epoch": 2.7755312741977116,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00041416731143878075,
      "loss": 0.7605,
      "step": 32260
    },
    {
      "epoch": 2.7756173105050332,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004141406722399395,
      "loss": 1.5279,
      "step": 32261
    },
    {
      "epoch": 2.775703346812355,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004141140332922201,
      "loss": 0.7993,
      "step": 32262
    },
    {
      "epoch": 2.7757893831196765,
      "grad_norm": 0.828125,
      "learning_rate": 0.00041408739459570035,
      "loss": 0.9077,
      "step": 32263
    },
    {
      "epoch": 2.775875419426998,
      "grad_norm": 1.03125,
      "learning_rate": 0.00041406075615045833,
      "loss": 1.1687,
      "step": 32264
    },
    {
      "epoch": 2.7759614557343197,
      "grad_norm": 1.2109375,
      "learning_rate": 0.000414034117956572,
      "loss": 1.4745,
      "step": 32265
    },
    {
      "epoch": 2.7760474920416414,
      "grad_norm": 1.125,
      "learning_rate": 0.00041400748001411893,
      "loss": 1.1588,
      "step": 32266
    },
    {
      "epoch": 2.7761335283489634,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004139808423231773,
      "loss": 1.6887,
      "step": 32267
    },
    {
      "epoch": 2.776219564656285,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000413954204883825,
      "loss": 0.9428,
      "step": 32268
    },
    {
      "epoch": 2.7763056009636067,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004139275676961399,
      "loss": 1.4409,
      "step": 32269
    },
    {
      "epoch": 2.7763916372709283,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00041390093076020006,
      "loss": 1.2078,
      "step": 32270
    },
    {
      "epoch": 2.77647767357825,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00041387429407608326,
      "loss": 1.347,
      "step": 32271
    },
    {
      "epoch": 2.7765637098855716,
      "grad_norm": 1.5,
      "learning_rate": 0.00041384765764386716,
      "loss": 0.9262,
      "step": 32272
    },
    {
      "epoch": 2.776649746192893,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004138210214636301,
      "loss": 1.3482,
      "step": 32273
    },
    {
      "epoch": 2.7767357825002152,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004137943855354498,
      "loss": 1.4097,
      "step": 32274
    },
    {
      "epoch": 2.776821818807537,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000413767749859404,
      "loss": 1.5558,
      "step": 32275
    },
    {
      "epoch": 2.7769078551148585,
      "grad_norm": 1.25,
      "learning_rate": 0.00041374111443557096,
      "loss": 1.1672,
      "step": 32276
    },
    {
      "epoch": 2.77699389142218,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004137144792640283,
      "loss": 1.3786,
      "step": 32277
    },
    {
      "epoch": 2.7770799277295017,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000413687844344854,
      "loss": 1.5199,
      "step": 32278
    },
    {
      "epoch": 2.777165964036824,
      "grad_norm": 0.640625,
      "learning_rate": 0.00041366120967812595,
      "loss": 0.4626,
      "step": 32279
    },
    {
      "epoch": 2.777252000344145,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00041363457526392213,
      "loss": 0.9083,
      "step": 32280
    },
    {
      "epoch": 2.777338036651467,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00041360794110232034,
      "loss": 0.733,
      "step": 32281
    },
    {
      "epoch": 2.7774240729587887,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00041358130719339843,
      "loss": 0.8053,
      "step": 32282
    },
    {
      "epoch": 2.7775101092661103,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004135546735372345,
      "loss": 1.6668,
      "step": 32283
    },
    {
      "epoch": 2.777596145573432,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0004135280401339064,
      "loss": 0.5235,
      "step": 32284
    },
    {
      "epoch": 2.7776821818807536,
      "grad_norm": 1.078125,
      "learning_rate": 0.00041350140698349187,
      "loss": 1.0201,
      "step": 32285
    },
    {
      "epoch": 2.7777682181880756,
      "grad_norm": 0.96875,
      "learning_rate": 0.00041347477408606884,
      "loss": 0.9664,
      "step": 32286
    },
    {
      "epoch": 2.777854254495397,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0004134481414417154,
      "loss": 1.0651,
      "step": 32287
    },
    {
      "epoch": 2.777940290802719,
      "grad_norm": 1.109375,
      "learning_rate": 0.00041342150905050924,
      "loss": 1.0684,
      "step": 32288
    },
    {
      "epoch": 2.7780263271100405,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004133948769125283,
      "loss": 1.2599,
      "step": 32289
    },
    {
      "epoch": 2.778112363417362,
      "grad_norm": 1.0,
      "learning_rate": 0.00041336824502785055,
      "loss": 1.0551,
      "step": 32290
    },
    {
      "epoch": 2.7781983997246837,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00041334161339655383,
      "loss": 1.152,
      "step": 32291
    },
    {
      "epoch": 2.7782844360320054,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004133149820187161,
      "loss": 1.1325,
      "step": 32292
    },
    {
      "epoch": 2.7783704723393274,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000413288350894415,
      "loss": 0.8444,
      "step": 32293
    },
    {
      "epoch": 2.7784565086466486,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004132617200237288,
      "loss": 0.8778,
      "step": 32294
    },
    {
      "epoch": 2.7785425449539707,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00041323508940673513,
      "loss": 0.8762,
      "step": 32295
    },
    {
      "epoch": 2.7786285812612923,
      "grad_norm": 1.0625,
      "learning_rate": 0.000413208459043512,
      "loss": 2.0121,
      "step": 32296
    },
    {
      "epoch": 2.778714617568614,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00041318182893413714,
      "loss": 0.8028,
      "step": 32297
    },
    {
      "epoch": 2.7788006538759356,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004131551990786886,
      "loss": 0.8635,
      "step": 32298
    },
    {
      "epoch": 2.778886690183257,
      "grad_norm": 1.125,
      "learning_rate": 0.00041312856947724424,
      "loss": 1.0195,
      "step": 32299
    },
    {
      "epoch": 2.7789727264905792,
      "grad_norm": 1.078125,
      "learning_rate": 0.00041310194012988185,
      "loss": 0.9894,
      "step": 32300
    },
    {
      "epoch": 2.779058762797901,
      "grad_norm": 1.15625,
      "learning_rate": 0.00041307531103667943,
      "loss": 1.1318,
      "step": 32301
    },
    {
      "epoch": 2.7791447991052225,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00041304868219771483,
      "loss": 0.8977,
      "step": 32302
    },
    {
      "epoch": 2.779230835412544,
      "grad_norm": 1.15625,
      "learning_rate": 0.00041302205361306595,
      "loss": 0.9454,
      "step": 32303
    },
    {
      "epoch": 2.7793168717198657,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004129954252828104,
      "loss": 1.574,
      "step": 32304
    },
    {
      "epoch": 2.7794029080271874,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00041296879720702646,
      "loss": 1.3025,
      "step": 32305
    },
    {
      "epoch": 2.779488944334509,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004129421693857919,
      "loss": 0.6569,
      "step": 32306
    },
    {
      "epoch": 2.779574980641831,
      "grad_norm": 1.421875,
      "learning_rate": 0.00041291554181918465,
      "loss": 1.7219,
      "step": 32307
    },
    {
      "epoch": 2.7796610169491527,
      "grad_norm": 0.96875,
      "learning_rate": 0.0004128889145072825,
      "loss": 0.8004,
      "step": 32308
    },
    {
      "epoch": 2.7797470532564743,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0004128622874501632,
      "loss": 1.5004,
      "step": 32309
    },
    {
      "epoch": 2.779833089563796,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004128356606479048,
      "loss": 0.698,
      "step": 32310
    },
    {
      "epoch": 2.7799191258711176,
      "grad_norm": 1.171875,
      "learning_rate": 0.00041280903410058524,
      "loss": 1.2248,
      "step": 32311
    },
    {
      "epoch": 2.780005162178439,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00041278240780828214,
      "loss": 1.0309,
      "step": 32312
    },
    {
      "epoch": 2.780091198485761,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00041275578177107366,
      "loss": 1.4007,
      "step": 32313
    },
    {
      "epoch": 2.780177234793083,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00041272915598903747,
      "loss": 0.9371,
      "step": 32314
    },
    {
      "epoch": 2.7802632711004045,
      "grad_norm": 1.40625,
      "learning_rate": 0.00041270253046225163,
      "loss": 0.848,
      "step": 32315
    },
    {
      "epoch": 2.780349307407726,
      "grad_norm": 1.25,
      "learning_rate": 0.0004126759051907938,
      "loss": 1.1117,
      "step": 32316
    },
    {
      "epoch": 2.7804353437150477,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00041264928017474203,
      "loss": 1.0085,
      "step": 32317
    },
    {
      "epoch": 2.7805213800223694,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00041262265541417407,
      "loss": 0.7497,
      "step": 32318
    },
    {
      "epoch": 2.780607416329691,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00041259603090916796,
      "loss": 0.8711,
      "step": 32319
    },
    {
      "epoch": 2.7806934526370126,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004125694066598013,
      "loss": 0.7393,
      "step": 32320
    },
    {
      "epoch": 2.7807794889443347,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004125427826661523,
      "loss": 1.3865,
      "step": 32321
    },
    {
      "epoch": 2.7808655252516563,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004125161589282985,
      "loss": 0.7398,
      "step": 32322
    },
    {
      "epoch": 2.780951561558978,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00041248953544631797,
      "loss": 0.7468,
      "step": 32323
    },
    {
      "epoch": 2.7810375978662996,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004124629122202885,
      "loss": 1.5242,
      "step": 32324
    },
    {
      "epoch": 2.781123634173621,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0004124362892502881,
      "loss": 1.1537,
      "step": 32325
    },
    {
      "epoch": 2.781209670480943,
      "grad_norm": 1.03125,
      "learning_rate": 0.00041240966653639447,
      "loss": 1.6774,
      "step": 32326
    },
    {
      "epoch": 2.7812957067882644,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004123830440786854,
      "loss": 1.0069,
      "step": 32327
    },
    {
      "epoch": 2.7813817430955865,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00041235642187723904,
      "loss": 1.4774,
      "step": 32328
    },
    {
      "epoch": 2.781467779402908,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004123297999321331,
      "loss": 1.0342,
      "step": 32329
    },
    {
      "epoch": 2.7815538157102297,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004123031782434453,
      "loss": 1.1088,
      "step": 32330
    },
    {
      "epoch": 2.7816398520175514,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00041227655681125373,
      "loss": 0.8005,
      "step": 32331
    },
    {
      "epoch": 2.781725888324873,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004122499356356363,
      "loss": 1.31,
      "step": 32332
    },
    {
      "epoch": 2.7818119246321946,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004122233147166706,
      "loss": 0.9066,
      "step": 32333
    },
    {
      "epoch": 2.7818979609395162,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004121966940544346,
      "loss": 1.9627,
      "step": 32334
    },
    {
      "epoch": 2.7819839972468383,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004121700736490063,
      "loss": 1.5787,
      "step": 32335
    },
    {
      "epoch": 2.78207003355416,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00041214345350046343,
      "loss": 1.2258,
      "step": 32336
    },
    {
      "epoch": 2.7821560698614816,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004121168336088839,
      "loss": 1.7163,
      "step": 32337
    },
    {
      "epoch": 2.782242106168803,
      "grad_norm": 1.703125,
      "learning_rate": 0.0004120902139743454,
      "loss": 0.668,
      "step": 32338
    },
    {
      "epoch": 2.782328142476125,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000412063594596926,
      "loss": 0.9827,
      "step": 32339
    },
    {
      "epoch": 2.782414178783447,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00041203697547670355,
      "loss": 0.7232,
      "step": 32340
    },
    {
      "epoch": 2.782500215090768,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00041201035661375565,
      "loss": 1.5786,
      "step": 32341
    },
    {
      "epoch": 2.78258625139809,
      "grad_norm": 1.125,
      "learning_rate": 0.0004119837380081605,
      "loss": 1.1331,
      "step": 32342
    },
    {
      "epoch": 2.7826722877054118,
      "grad_norm": 1.125,
      "learning_rate": 0.0004119571196599957,
      "loss": 1.3504,
      "step": 32343
    },
    {
      "epoch": 2.7827583240127334,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00041193050156933925,
      "loss": 0.9884,
      "step": 32344
    },
    {
      "epoch": 2.782844360320055,
      "grad_norm": 0.78125,
      "learning_rate": 0.00041190388373626904,
      "loss": 0.837,
      "step": 32345
    },
    {
      "epoch": 2.7829303966273766,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00041187726616086266,
      "loss": 1.2997,
      "step": 32346
    },
    {
      "epoch": 2.7830164329346987,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00041185064884319836,
      "loss": 0.8933,
      "step": 32347
    },
    {
      "epoch": 2.78310246924202,
      "grad_norm": 1.46875,
      "learning_rate": 0.00041182403178335365,
      "loss": 0.8975,
      "step": 32348
    },
    {
      "epoch": 2.783188505549342,
      "grad_norm": 1.34375,
      "learning_rate": 0.00041179741498140654,
      "loss": 1.0277,
      "step": 32349
    },
    {
      "epoch": 2.7832745418566636,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004117707984374347,
      "loss": 1.5954,
      "step": 32350
    },
    {
      "epoch": 2.783360578163985,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00041174418215151633,
      "loss": 0.876,
      "step": 32351
    },
    {
      "epoch": 2.783446614471307,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000411717566123729,
      "loss": 1.2871,
      "step": 32352
    },
    {
      "epoch": 2.7835326507786284,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00041169095035415045,
      "loss": 0.7239,
      "step": 32353
    },
    {
      "epoch": 2.7836186870859505,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004116643348428588,
      "loss": 1.0908,
      "step": 32354
    },
    {
      "epoch": 2.7837047233932717,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00041163771958993193,
      "loss": 1.2059,
      "step": 32355
    },
    {
      "epoch": 2.7837907597005938,
      "grad_norm": 1.921875,
      "learning_rate": 0.0004116111045954474,
      "loss": 1.7077,
      "step": 32356
    },
    {
      "epoch": 2.7838767960079154,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00041158448985948316,
      "loss": 1.1922,
      "step": 32357
    },
    {
      "epoch": 2.783962832315237,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004115578753821172,
      "loss": 0.6086,
      "step": 32358
    },
    {
      "epoch": 2.7840488686225586,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00041153126116342723,
      "loss": 0.5184,
      "step": 32359
    },
    {
      "epoch": 2.7841349049298802,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00041150464720349114,
      "loss": 1.8936,
      "step": 32360
    },
    {
      "epoch": 2.7842209412372023,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004114780335023866,
      "loss": 1.5593,
      "step": 32361
    },
    {
      "epoch": 2.784306977544524,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004114514200601917,
      "loss": 1.7217,
      "step": 32362
    },
    {
      "epoch": 2.7843930138518456,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00041142480687698425,
      "loss": 1.0107,
      "step": 32363
    },
    {
      "epoch": 2.784479050159167,
      "grad_norm": 1.0625,
      "learning_rate": 0.00041139819395284185,
      "loss": 0.6455,
      "step": 32364
    },
    {
      "epoch": 2.784565086466489,
      "grad_norm": 0.953125,
      "learning_rate": 0.00041137158128784257,
      "loss": 0.7527,
      "step": 32365
    },
    {
      "epoch": 2.7846511227738104,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004113449688820642,
      "loss": 0.9652,
      "step": 32366
    },
    {
      "epoch": 2.784737159081132,
      "grad_norm": 1.46875,
      "learning_rate": 0.0004113183567355846,
      "loss": 0.9599,
      "step": 32367
    },
    {
      "epoch": 2.784823195388454,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00041129174484848144,
      "loss": 0.9297,
      "step": 32368
    },
    {
      "epoch": 2.7849092316957758,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00041126513322083276,
      "loss": 1.2032,
      "step": 32369
    },
    {
      "epoch": 2.7849952680030974,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00041123852185271634,
      "loss": 0.9192,
      "step": 32370
    },
    {
      "epoch": 2.785081304310419,
      "grad_norm": 1.265625,
      "learning_rate": 0.00041121191074420984,
      "loss": 0.996,
      "step": 32371
    },
    {
      "epoch": 2.7851673406177406,
      "grad_norm": 0.90625,
      "learning_rate": 0.00041118529989539134,
      "loss": 0.7492,
      "step": 32372
    },
    {
      "epoch": 2.7852533769250623,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004111586893063386,
      "loss": 0.9631,
      "step": 32373
    },
    {
      "epoch": 2.785339413232384,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00041113207897712934,
      "loss": 1.1672,
      "step": 32374
    },
    {
      "epoch": 2.785425449539706,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004111054689078414,
      "loss": 1.0815,
      "step": 32375
    },
    {
      "epoch": 2.7855114858470276,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004110788590985528,
      "loss": 0.5931,
      "step": 32376
    },
    {
      "epoch": 2.785597522154349,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00041105224954934123,
      "loss": 1.0792,
      "step": 32377
    },
    {
      "epoch": 2.785683558461671,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00041102564026028433,
      "loss": 0.5683,
      "step": 32378
    },
    {
      "epoch": 2.7857695947689924,
      "grad_norm": 3.0,
      "learning_rate": 0.00041099903123146034,
      "loss": 2.0444,
      "step": 32379
    },
    {
      "epoch": 2.785855631076314,
      "grad_norm": 1.59375,
      "learning_rate": 0.00041097242246294674,
      "loss": 0.827,
      "step": 32380
    },
    {
      "epoch": 2.7859416673836357,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004109458139548216,
      "loss": 1.3071,
      "step": 32381
    },
    {
      "epoch": 2.7860277036909578,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004109192057071626,
      "loss": 0.9288,
      "step": 32382
    },
    {
      "epoch": 2.7861137399982794,
      "grad_norm": 1.25,
      "learning_rate": 0.00041089259772004766,
      "loss": 1.613,
      "step": 32383
    },
    {
      "epoch": 2.786199776305601,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004108659899935544,
      "loss": 1.1203,
      "step": 32384
    },
    {
      "epoch": 2.7862858126129226,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004108393825277609,
      "loss": 0.9129,
      "step": 32385
    },
    {
      "epoch": 2.7863718489202443,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004108127753227449,
      "loss": 1.5227,
      "step": 32386
    },
    {
      "epoch": 2.786457885227566,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004107861683785841,
      "loss": 1.1282,
      "step": 32387
    },
    {
      "epoch": 2.7865439215348875,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004107595616953564,
      "loss": 1.2195,
      "step": 32388
    },
    {
      "epoch": 2.7866299578422096,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004107329552731397,
      "loss": 1.3049,
      "step": 32389
    },
    {
      "epoch": 2.786715994149531,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00041070634911201175,
      "loss": 0.7868,
      "step": 32390
    },
    {
      "epoch": 2.786802030456853,
      "grad_norm": 1.125,
      "learning_rate": 0.0004106797432120503,
      "loss": 1.7265,
      "step": 32391
    },
    {
      "epoch": 2.7868880667641744,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00041065313757333323,
      "loss": 1.0123,
      "step": 32392
    },
    {
      "epoch": 2.786974103071496,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004106265321959384,
      "loss": 1.1182,
      "step": 32393
    },
    {
      "epoch": 2.787060139378818,
      "grad_norm": 1.078125,
      "learning_rate": 0.00041059992707994346,
      "loss": 1.3556,
      "step": 32394
    },
    {
      "epoch": 2.7871461756861393,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004105733222254265,
      "loss": 0.715,
      "step": 32395
    },
    {
      "epoch": 2.7872322119934614,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0004105467176324651,
      "loss": 0.6459,
      "step": 32396
    },
    {
      "epoch": 2.787318248300783,
      "grad_norm": 0.9375,
      "learning_rate": 0.00041052011330113724,
      "loss": 1.0017,
      "step": 32397
    },
    {
      "epoch": 2.7874042846081046,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004104935092315205,
      "loss": 1.5567,
      "step": 32398
    },
    {
      "epoch": 2.7874903209154263,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004104669054236929,
      "loss": 1.8462,
      "step": 32399
    },
    {
      "epoch": 2.787576357222748,
      "grad_norm": 1.609375,
      "learning_rate": 0.0004104403018777323,
      "loss": 1.3151,
      "step": 32400
    },
    {
      "epoch": 2.78766239353007,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0004104136985937162,
      "loss": 0.8182,
      "step": 32401
    },
    {
      "epoch": 2.787748429837391,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004103870955717227,
      "loss": 1.4906,
      "step": 32402
    },
    {
      "epoch": 2.787834466144713,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00041036049281182956,
      "loss": 0.7643,
      "step": 32403
    },
    {
      "epoch": 2.787920502452035,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0004103338903141146,
      "loss": 0.6685,
      "step": 32404
    },
    {
      "epoch": 2.7880065387593564,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0004103072880786554,
      "loss": 1.7636,
      "step": 32405
    },
    {
      "epoch": 2.788092575066678,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00041028068610553004,
      "loss": 1.4136,
      "step": 32406
    },
    {
      "epoch": 2.7881786113739997,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00041025408439481626,
      "loss": 0.6487,
      "step": 32407
    },
    {
      "epoch": 2.7882646476813218,
      "grad_norm": 1.34375,
      "learning_rate": 0.0004102274829465918,
      "loss": 1.1369,
      "step": 32408
    },
    {
      "epoch": 2.788350683988643,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004102008817609344,
      "loss": 0.8611,
      "step": 32409
    },
    {
      "epoch": 2.788436720295965,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000410174280837922,
      "loss": 1.0771,
      "step": 32410
    },
    {
      "epoch": 2.7885227566032866,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00041014768017763245,
      "loss": 1.0019,
      "step": 32411
    },
    {
      "epoch": 2.7886087929106083,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00041012107978014327,
      "loss": 1.0813,
      "step": 32412
    },
    {
      "epoch": 2.78869482921793,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00041009447964553266,
      "loss": 0.7498,
      "step": 32413
    },
    {
      "epoch": 2.7887808655252515,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0004100678797738781,
      "loss": 1.5811,
      "step": 32414
    },
    {
      "epoch": 2.7888669018325736,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004100412801652573,
      "loss": 1.2419,
      "step": 32415
    },
    {
      "epoch": 2.788952938139895,
      "grad_norm": 1.234375,
      "learning_rate": 0.00041001468081974855,
      "loss": 1.6938,
      "step": 32416
    },
    {
      "epoch": 2.789038974447217,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00040998808173742924,
      "loss": 0.6836,
      "step": 32417
    },
    {
      "epoch": 2.7891250107545384,
      "grad_norm": 0.828125,
      "learning_rate": 0.00040996148291837735,
      "loss": 0.5064,
      "step": 32418
    },
    {
      "epoch": 2.78921104706186,
      "grad_norm": 0.921875,
      "learning_rate": 0.00040993488436267064,
      "loss": 1.0882,
      "step": 32419
    },
    {
      "epoch": 2.7892970833691817,
      "grad_norm": 1.125,
      "learning_rate": 0.0004099082860703869,
      "loss": 0.9713,
      "step": 32420
    },
    {
      "epoch": 2.7893831196765033,
      "grad_norm": 1.109375,
      "learning_rate": 0.00040988168804160374,
      "loss": 1.1192,
      "step": 32421
    },
    {
      "epoch": 2.7894691559838254,
      "grad_norm": 0.984375,
      "learning_rate": 0.00040985509027639924,
      "loss": 1.5244,
      "step": 32422
    },
    {
      "epoch": 2.789555192291147,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0004098284927748511,
      "loss": 0.967,
      "step": 32423
    },
    {
      "epoch": 2.7896412285984686,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000409801895537037,
      "loss": 1.3437,
      "step": 32424
    },
    {
      "epoch": 2.7897272649057903,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00040977529856303473,
      "loss": 1.5724,
      "step": 32425
    },
    {
      "epoch": 2.789813301213112,
      "grad_norm": 1.140625,
      "learning_rate": 0.00040974870185292233,
      "loss": 0.9552,
      "step": 32426
    },
    {
      "epoch": 2.7898993375204335,
      "grad_norm": 3.578125,
      "learning_rate": 0.0004097221054067774,
      "loss": 0.6745,
      "step": 32427
    },
    {
      "epoch": 2.789985373827755,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00040969550922467763,
      "loss": 1.4132,
      "step": 32428
    },
    {
      "epoch": 2.790071410135077,
      "grad_norm": 0.9375,
      "learning_rate": 0.000409668913306701,
      "loss": 1.1114,
      "step": 32429
    },
    {
      "epoch": 2.790157446442399,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00040964231765292533,
      "loss": 1.4972,
      "step": 32430
    },
    {
      "epoch": 2.7902434827497204,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00040961572226342827,
      "loss": 1.5477,
      "step": 32431
    },
    {
      "epoch": 2.790329519057042,
      "grad_norm": 1.125,
      "learning_rate": 0.0004095891271382875,
      "loss": 0.9035,
      "step": 32432
    },
    {
      "epoch": 2.7904155553643637,
      "grad_norm": 1.796875,
      "learning_rate": 0.00040956253227758104,
      "loss": 1.6456,
      "step": 32433
    },
    {
      "epoch": 2.7905015916716853,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004095359376813866,
      "loss": 1.3225,
      "step": 32434
    },
    {
      "epoch": 2.790587627979007,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0004095093433497819,
      "loss": 1.032,
      "step": 32435
    },
    {
      "epoch": 2.790673664286329,
      "grad_norm": 1.390625,
      "learning_rate": 0.00040948274928284483,
      "loss": 1.3233,
      "step": 32436
    },
    {
      "epoch": 2.7907597005936506,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004094561554806532,
      "loss": 1.3738,
      "step": 32437
    },
    {
      "epoch": 2.7908457369009723,
      "grad_norm": 1.109375,
      "learning_rate": 0.00040942956194328455,
      "loss": 0.7379,
      "step": 32438
    },
    {
      "epoch": 2.790931773208294,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00040940296867081685,
      "loss": 2.0098,
      "step": 32439
    },
    {
      "epoch": 2.7910178095156155,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00040937637566332784,
      "loss": 0.867,
      "step": 32440
    },
    {
      "epoch": 2.791103845822937,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0004093497829208954,
      "loss": 1.2507,
      "step": 32441
    },
    {
      "epoch": 2.7911898821302588,
      "grad_norm": 1.203125,
      "learning_rate": 0.00040932319044359705,
      "loss": 1.4447,
      "step": 32442
    },
    {
      "epoch": 2.791275918437581,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004092965982315109,
      "loss": 1.1513,
      "step": 32443
    },
    {
      "epoch": 2.7913619547449025,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00040927000628471444,
      "loss": 0.9558,
      "step": 32444
    },
    {
      "epoch": 2.791447991052224,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004092434146032857,
      "loss": 1.6178,
      "step": 32445
    },
    {
      "epoch": 2.7915340273595457,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00040921682318730213,
      "loss": 0.6143,
      "step": 32446
    },
    {
      "epoch": 2.7916200636668673,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004091902320368418,
      "loss": 0.8335,
      "step": 32447
    },
    {
      "epoch": 2.791706099974189,
      "grad_norm": 0.84375,
      "learning_rate": 0.00040916364115198246,
      "loss": 0.8839,
      "step": 32448
    },
    {
      "epoch": 2.7917921362815106,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004091370505328017,
      "loss": 0.5964,
      "step": 32449
    },
    {
      "epoch": 2.7918781725888326,
      "grad_norm": 1.171875,
      "learning_rate": 0.00040911046017937737,
      "loss": 1.9062,
      "step": 32450
    },
    {
      "epoch": 2.7919642088961543,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00040908387009178733,
      "loss": 1.5534,
      "step": 32451
    },
    {
      "epoch": 2.792050245203476,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0004090572802701091,
      "loss": 1.7252,
      "step": 32452
    },
    {
      "epoch": 2.7921362815107975,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00040903069071442083,
      "loss": 0.5162,
      "step": 32453
    },
    {
      "epoch": 2.792222317818119,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004090041014248001,
      "loss": 1.0064,
      "step": 32454
    },
    {
      "epoch": 2.792308354125441,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004089775124013246,
      "loss": 1.0578,
      "step": 32455
    },
    {
      "epoch": 2.7923943904327624,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004089509236440722,
      "loss": 0.7023,
      "step": 32456
    },
    {
      "epoch": 2.7924804267400845,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004089243351531207,
      "loss": 1.1032,
      "step": 32457
    },
    {
      "epoch": 2.792566463047406,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004088977469285477,
      "loss": 1.5455,
      "step": 32458
    },
    {
      "epoch": 2.7926524993547277,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004088711589704312,
      "loss": 1.1274,
      "step": 32459
    },
    {
      "epoch": 2.7927385356620493,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004088445712788488,
      "loss": 1.7275,
      "step": 32460
    },
    {
      "epoch": 2.792824571969371,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0004088179838538783,
      "loss": 1.0886,
      "step": 32461
    },
    {
      "epoch": 2.792910608276693,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004087913966955974,
      "loss": 1.667,
      "step": 32462
    },
    {
      "epoch": 2.792996644584014,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00040876480980408397,
      "loss": 0.6516,
      "step": 32463
    },
    {
      "epoch": 2.7930826808913363,
      "grad_norm": 1.453125,
      "learning_rate": 0.0004087382231794158,
      "loss": 1.5209,
      "step": 32464
    },
    {
      "epoch": 2.793168717198658,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004087116368216704,
      "loss": 0.8209,
      "step": 32465
    },
    {
      "epoch": 2.7932547535059795,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004086850507309258,
      "loss": 0.8847,
      "step": 32466
    },
    {
      "epoch": 2.793340789813301,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004086584649072597,
      "loss": 1.7945,
      "step": 32467
    },
    {
      "epoch": 2.7934268261206228,
      "grad_norm": 1.046875,
      "learning_rate": 0.00040863187935074983,
      "loss": 1.4468,
      "step": 32468
    },
    {
      "epoch": 2.793512862427945,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00040860529406147383,
      "loss": 0.725,
      "step": 32469
    },
    {
      "epoch": 2.793598898735266,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004085787090395097,
      "loss": 1.4667,
      "step": 32470
    },
    {
      "epoch": 2.793684935042588,
      "grad_norm": 1.140625,
      "learning_rate": 0.00040855212428493506,
      "loss": 1.4291,
      "step": 32471
    },
    {
      "epoch": 2.7937709713499097,
      "grad_norm": 11.5625,
      "learning_rate": 0.0004085255397978277,
      "loss": 1.0947,
      "step": 32472
    },
    {
      "epoch": 2.7938570076572313,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00040849895557826515,
      "loss": 0.7641,
      "step": 32473
    },
    {
      "epoch": 2.793943043964553,
      "grad_norm": 1.5625,
      "learning_rate": 0.0004084723716263256,
      "loss": 0.591,
      "step": 32474
    },
    {
      "epoch": 2.7940290802718746,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004084457879420864,
      "loss": 1.2402,
      "step": 32475
    },
    {
      "epoch": 2.7941151165791966,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004084192045256254,
      "loss": 1.2893,
      "step": 32476
    },
    {
      "epoch": 2.7942011528865183,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00040839262137702063,
      "loss": 1.1058,
      "step": 32477
    },
    {
      "epoch": 2.79428718919384,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0004083660384963495,
      "loss": 0.6301,
      "step": 32478
    },
    {
      "epoch": 2.7943732255011615,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00040833945588369,
      "loss": 1.3221,
      "step": 32479
    },
    {
      "epoch": 2.794459261808483,
      "grad_norm": 1.25,
      "learning_rate": 0.00040831287353911954,
      "loss": 1.441,
      "step": 32480
    },
    {
      "epoch": 2.7945452981158048,
      "grad_norm": 1.28125,
      "learning_rate": 0.00040828629146271627,
      "loss": 1.2745,
      "step": 32481
    },
    {
      "epoch": 2.7946313344231264,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004082597096545578,
      "loss": 0.7919,
      "step": 32482
    },
    {
      "epoch": 2.7947173707304485,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004082331281147216,
      "loss": 1.5075,
      "step": 32483
    },
    {
      "epoch": 2.79480340703777,
      "grad_norm": 1.09375,
      "learning_rate": 0.00040820654684328586,
      "loss": 0.9604,
      "step": 32484
    },
    {
      "epoch": 2.7948894433450917,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0004081799658403281,
      "loss": 0.6388,
      "step": 32485
    },
    {
      "epoch": 2.7949754796524133,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00040815338510592606,
      "loss": 0.7359,
      "step": 32486
    },
    {
      "epoch": 2.795061515959735,
      "grad_norm": 3.34375,
      "learning_rate": 0.00040812680464015743,
      "loss": 1.1547,
      "step": 32487
    },
    {
      "epoch": 2.7951475522670566,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0004081002244431001,
      "loss": 1.5823,
      "step": 32488
    },
    {
      "epoch": 2.795233588574378,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00040807364451483154,
      "loss": 1.3278,
      "step": 32489
    },
    {
      "epoch": 2.7953196248817003,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0004080470648554299,
      "loss": 0.6958,
      "step": 32490
    },
    {
      "epoch": 2.795405661189022,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00040802048546497273,
      "loss": 0.4901,
      "step": 32491
    },
    {
      "epoch": 2.7954916974963435,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00040799390634353765,
      "loss": 1.3095,
      "step": 32492
    },
    {
      "epoch": 2.795577733803665,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004079673274912026,
      "loss": 1.9531,
      "step": 32493
    },
    {
      "epoch": 2.7956637701109868,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004079407489080452,
      "loss": 0.9796,
      "step": 32494
    },
    {
      "epoch": 2.7957498064183084,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004079141705941432,
      "loss": 1.1354,
      "step": 32495
    },
    {
      "epoch": 2.79583584272563,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00040788759254957425,
      "loss": 1.3902,
      "step": 32496
    },
    {
      "epoch": 2.795921879032952,
      "grad_norm": 1.5625,
      "learning_rate": 0.00040786101477441626,
      "loss": 0.7199,
      "step": 32497
    },
    {
      "epoch": 2.7960079153402737,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004078344372687469,
      "loss": 1.707,
      "step": 32498
    },
    {
      "epoch": 2.7960939516475953,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004078078600326438,
      "loss": 0.6898,
      "step": 32499
    },
    {
      "epoch": 2.796179987954917,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004077812830661849,
      "loss": 0.7505,
      "step": 32500
    },
    {
      "epoch": 2.7962660242622386,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00040775470636944777,
      "loss": 0.9315,
      "step": 32501
    },
    {
      "epoch": 2.79635206056956,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004077281299425102,
      "loss": 1.38,
      "step": 32502
    },
    {
      "epoch": 2.796438096876882,
      "grad_norm": 1.078125,
      "learning_rate": 0.00040770155378544974,
      "loss": 1.376,
      "step": 32503
    },
    {
      "epoch": 2.796524133184204,
      "grad_norm": 1.265625,
      "learning_rate": 0.0004076749778983445,
      "loss": 0.9114,
      "step": 32504
    },
    {
      "epoch": 2.7966101694915255,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004076484022812719,
      "loss": 1.8332,
      "step": 32505
    },
    {
      "epoch": 2.796696205798847,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0004076218269343097,
      "loss": 1.4861,
      "step": 32506
    },
    {
      "epoch": 2.7967822421061688,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00040759525185753577,
      "loss": 0.9464,
      "step": 32507
    },
    {
      "epoch": 2.7968682784134904,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004075686770510278,
      "loss": 1.3901,
      "step": 32508
    },
    {
      "epoch": 2.796954314720812,
      "grad_norm": 1.234375,
      "learning_rate": 0.00040754210251486346,
      "loss": 1.5403,
      "step": 32509
    },
    {
      "epoch": 2.7970403510281336,
      "grad_norm": 1.5,
      "learning_rate": 0.00040751552824912034,
      "loss": 0.9884,
      "step": 32510
    },
    {
      "epoch": 2.7971263873354557,
      "grad_norm": 1.4375,
      "learning_rate": 0.00040748895425387645,
      "loss": 1.8191,
      "step": 32511
    },
    {
      "epoch": 2.7972124236427773,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00040746238052920936,
      "loss": 1.0842,
      "step": 32512
    },
    {
      "epoch": 2.797298459950099,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00040743580707519687,
      "loss": 0.9291,
      "step": 32513
    },
    {
      "epoch": 2.7973844962574206,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004074092338919165,
      "loss": 1.3894,
      "step": 32514
    },
    {
      "epoch": 2.797470532564742,
      "grad_norm": 1.234375,
      "learning_rate": 0.00040738266097944617,
      "loss": 0.9497,
      "step": 32515
    },
    {
      "epoch": 2.7975565688720643,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004073560883378636,
      "loss": 1.4106,
      "step": 32516
    },
    {
      "epoch": 2.7976426051793855,
      "grad_norm": 1.015625,
      "learning_rate": 0.0004073295159672463,
      "loss": 1.4379,
      "step": 32517
    },
    {
      "epoch": 2.7977286414867075,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0004073029438676723,
      "loss": 0.8587,
      "step": 32518
    },
    {
      "epoch": 2.797814677794029,
      "grad_norm": 1.171875,
      "learning_rate": 0.00040727637203921913,
      "loss": 1.4193,
      "step": 32519
    },
    {
      "epoch": 2.7979007141013508,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0004072498004819645,
      "loss": 0.9584,
      "step": 32520
    },
    {
      "epoch": 2.7979867504086724,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00040722322919598615,
      "loss": 1.1031,
      "step": 32521
    },
    {
      "epoch": 2.798072786715994,
      "grad_norm": 2.03125,
      "learning_rate": 0.00040719665818136187,
      "loss": 1.4008,
      "step": 32522
    },
    {
      "epoch": 2.798158823023316,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00040717008743816933,
      "loss": 2.013,
      "step": 32523
    },
    {
      "epoch": 2.7982448593306373,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004071435169664861,
      "loss": 0.8523,
      "step": 32524
    },
    {
      "epoch": 2.7983308956379593,
      "grad_norm": 3.609375,
      "learning_rate": 0.0004071169467663901,
      "loss": 0.8096,
      "step": 32525
    },
    {
      "epoch": 2.798416931945281,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004070903768379588,
      "loss": 1.9022,
      "step": 32526
    },
    {
      "epoch": 2.7985029682526026,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00040706380718127027,
      "loss": 1.4563,
      "step": 32527
    },
    {
      "epoch": 2.798589004559924,
      "grad_norm": 2.140625,
      "learning_rate": 0.00040703723779640205,
      "loss": 1.1293,
      "step": 32528
    },
    {
      "epoch": 2.798675040867246,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004070106686834317,
      "loss": 1.3842,
      "step": 32529
    },
    {
      "epoch": 2.798761077174568,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004069840998424372,
      "loss": 1.8065,
      "step": 32530
    },
    {
      "epoch": 2.798847113481889,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004069575312734961,
      "loss": 1.1804,
      "step": 32531
    },
    {
      "epoch": 2.798933149789211,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00040693096297668614,
      "loss": 1.3501,
      "step": 32532
    },
    {
      "epoch": 2.7990191860965328,
      "grad_norm": 1.203125,
      "learning_rate": 0.00040690439495208485,
      "loss": 1.3904,
      "step": 32533
    },
    {
      "epoch": 2.7991052224038544,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004068778271997703,
      "loss": 1.4937,
      "step": 32534
    },
    {
      "epoch": 2.799191258711176,
      "grad_norm": 1.25,
      "learning_rate": 0.0004068512597198199,
      "loss": 1.7874,
      "step": 32535
    },
    {
      "epoch": 2.7992772950184976,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00040682469251231146,
      "loss": 1.1266,
      "step": 32536
    },
    {
      "epoch": 2.7993633313258197,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004067981255773226,
      "loss": 0.823,
      "step": 32537
    },
    {
      "epoch": 2.7994493676331413,
      "grad_norm": 1.09375,
      "learning_rate": 0.0004067715589149312,
      "loss": 1.5,
      "step": 32538
    },
    {
      "epoch": 2.799535403940463,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00040674499252521486,
      "loss": 1.8516,
      "step": 32539
    },
    {
      "epoch": 2.7996214402477846,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0004067184264082512,
      "loss": 1.111,
      "step": 32540
    },
    {
      "epoch": 2.799707476555106,
      "grad_norm": 4.625,
      "learning_rate": 0.00040669186056411805,
      "loss": 1.0243,
      "step": 32541
    },
    {
      "epoch": 2.799793512862428,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004066652949928931,
      "loss": 0.9605,
      "step": 32542
    },
    {
      "epoch": 2.7998795491697495,
      "grad_norm": 3.859375,
      "learning_rate": 0.000406638729694654,
      "loss": 1.3915,
      "step": 32543
    },
    {
      "epoch": 2.7999655854770715,
      "grad_norm": 1.65625,
      "learning_rate": 0.0004066121646694783,
      "loss": 1.5105,
      "step": 32544
    },
    {
      "epoch": 2.800051621784393,
      "grad_norm": 1.109375,
      "learning_rate": 0.00040658559991744406,
      "loss": 1.5093,
      "step": 32545
    },
    {
      "epoch": 2.8001376580917148,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004065590354386287,
      "loss": 0.9455,
      "step": 32546
    },
    {
      "epoch": 2.8002236943990364,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00040653247123310986,
      "loss": 0.7168,
      "step": 32547
    },
    {
      "epoch": 2.800309730706358,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0004065059073009655,
      "loss": 1.0966,
      "step": 32548
    },
    {
      "epoch": 2.8003957670136796,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00040647934364227323,
      "loss": 1.6453,
      "step": 32549
    },
    {
      "epoch": 2.8004818033210013,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0004064527802571106,
      "loss": 1.2526,
      "step": 32550
    },
    {
      "epoch": 2.8005678396283233,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004064262171455554,
      "loss": 0.8597,
      "step": 32551
    },
    {
      "epoch": 2.800653875935645,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00040639965430768534,
      "loss": 1.7172,
      "step": 32552
    },
    {
      "epoch": 2.8007399122429666,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004063730917435781,
      "loss": 0.8799,
      "step": 32553
    },
    {
      "epoch": 2.800825948550288,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00040634652945331125,
      "loss": 1.8788,
      "step": 32554
    },
    {
      "epoch": 2.80091198485761,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0004063199674369627,
      "loss": 1.2929,
      "step": 32555
    },
    {
      "epoch": 2.8009980211649315,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00040629340569461,
      "loss": 1.8553,
      "step": 32556
    },
    {
      "epoch": 2.801084057472253,
      "grad_norm": 1.09375,
      "learning_rate": 0.00040626684422633095,
      "loss": 1.0361,
      "step": 32557
    },
    {
      "epoch": 2.801170093779575,
      "grad_norm": 1.359375,
      "learning_rate": 0.00040624028303220297,
      "loss": 1.1637,
      "step": 32558
    },
    {
      "epoch": 2.801256130086897,
      "grad_norm": 1.109375,
      "learning_rate": 0.00040621372211230403,
      "loss": 0.8814,
      "step": 32559
    },
    {
      "epoch": 2.8013421663942184,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004061871614667117,
      "loss": 1.6506,
      "step": 32560
    },
    {
      "epoch": 2.80142820270154,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00040616060109550375,
      "loss": 0.9761,
      "step": 32561
    },
    {
      "epoch": 2.8015142390088617,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004061340409987576,
      "loss": 0.7844,
      "step": 32562
    },
    {
      "epoch": 2.8016002753161833,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00040610748117655103,
      "loss": 2.1754,
      "step": 32563
    },
    {
      "epoch": 2.801686311623505,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0004060809216289621,
      "loss": 1.2202,
      "step": 32564
    },
    {
      "epoch": 2.801772347930827,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004060543623560682,
      "loss": 1.7839,
      "step": 32565
    },
    {
      "epoch": 2.8018583842381486,
      "grad_norm": 1.171875,
      "learning_rate": 0.00040602780335794695,
      "loss": 1.3714,
      "step": 32566
    },
    {
      "epoch": 2.80194442054547,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00040600124463467603,
      "loss": 1.5246,
      "step": 32567
    },
    {
      "epoch": 2.802030456852792,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00040597468618633326,
      "loss": 0.7457,
      "step": 32568
    },
    {
      "epoch": 2.8021164931601135,
      "grad_norm": 1.484375,
      "learning_rate": 0.0004059481280129963,
      "loss": 0.8322,
      "step": 32569
    },
    {
      "epoch": 2.8022025294674355,
      "grad_norm": 1.203125,
      "learning_rate": 0.00040592157011474264,
      "loss": 1.6113,
      "step": 32570
    },
    {
      "epoch": 2.8022885657747567,
      "grad_norm": 1.25,
      "learning_rate": 0.00040589501249165016,
      "loss": 1.5013,
      "step": 32571
    },
    {
      "epoch": 2.802374602082079,
      "grad_norm": 0.953125,
      "learning_rate": 0.00040586845514379655,
      "loss": 1.0655,
      "step": 32572
    },
    {
      "epoch": 2.8024606383894004,
      "grad_norm": 1.21875,
      "learning_rate": 0.00040584189807125933,
      "loss": 1.6312,
      "step": 32573
    },
    {
      "epoch": 2.802546674696722,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004058153412741162,
      "loss": 1.3117,
      "step": 32574
    },
    {
      "epoch": 2.8026327110040437,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004057887847524449,
      "loss": 1.1812,
      "step": 32575
    },
    {
      "epoch": 2.8027187473113653,
      "grad_norm": 1.046875,
      "learning_rate": 0.0004057622285063232,
      "loss": 1.7524,
      "step": 32576
    },
    {
      "epoch": 2.8028047836186873,
      "grad_norm": 0.8125,
      "learning_rate": 0.00040573567253582857,
      "loss": 0.8617,
      "step": 32577
    },
    {
      "epoch": 2.8028908199260085,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0004057091168410386,
      "loss": 1.3783,
      "step": 32578
    },
    {
      "epoch": 2.8029768562333306,
      "grad_norm": 1.0,
      "learning_rate": 0.00040568256142203135,
      "loss": 0.9381,
      "step": 32579
    },
    {
      "epoch": 2.803062892540652,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004056560062788842,
      "loss": 1.2158,
      "step": 32580
    },
    {
      "epoch": 2.803148928847974,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004056294514116748,
      "loss": 1.2841,
      "step": 32581
    },
    {
      "epoch": 2.8032349651552955,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000405602896820481,
      "loss": 0.8221,
      "step": 32582
    },
    {
      "epoch": 2.803321001462617,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004055763425053804,
      "loss": 1.5765,
      "step": 32583
    },
    {
      "epoch": 2.803407037769939,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00040554978846645055,
      "loss": 1.175,
      "step": 32584
    },
    {
      "epoch": 2.8034930740772603,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0004055232347037692,
      "loss": 0.8279,
      "step": 32585
    },
    {
      "epoch": 2.8035791103845824,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00040549668121741406,
      "loss": 1.0075,
      "step": 32586
    },
    {
      "epoch": 2.803665146691904,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00040547012800746276,
      "loss": 1.1232,
      "step": 32587
    },
    {
      "epoch": 2.8037511829992257,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00040544357507399283,
      "loss": 1.4574,
      "step": 32588
    },
    {
      "epoch": 2.8038372193065473,
      "grad_norm": 1.09375,
      "learning_rate": 0.00040541702241708217,
      "loss": 0.708,
      "step": 32589
    },
    {
      "epoch": 2.803923255613869,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004053904700368083,
      "loss": 1.161,
      "step": 32590
    },
    {
      "epoch": 2.804009291921191,
      "grad_norm": 1.171875,
      "learning_rate": 0.00040536391793324894,
      "loss": 0.8602,
      "step": 32591
    },
    {
      "epoch": 2.8040953282285126,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0004053373661064816,
      "loss": 0.9425,
      "step": 32592
    },
    {
      "epoch": 2.804181364535834,
      "grad_norm": 1.125,
      "learning_rate": 0.00040531081455658416,
      "loss": 1.6769,
      "step": 32593
    },
    {
      "epoch": 2.804267400843156,
      "grad_norm": 1.03125,
      "learning_rate": 0.00040528426328363416,
      "loss": 1.3979,
      "step": 32594
    },
    {
      "epoch": 2.8043534371504775,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00040525771228770913,
      "loss": 1.1496,
      "step": 32595
    },
    {
      "epoch": 2.804439473457799,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00040523116156888705,
      "loss": 0.9932,
      "step": 32596
    },
    {
      "epoch": 2.8045255097651207,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004052046111272453,
      "loss": 1.8596,
      "step": 32597
    },
    {
      "epoch": 2.804611546072443,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0004051780609628617,
      "loss": 1.1666,
      "step": 32598
    },
    {
      "epoch": 2.8046975823797644,
      "grad_norm": 116.5,
      "learning_rate": 0.0004051515110758137,
      "loss": 1.2306,
      "step": 32599
    },
    {
      "epoch": 2.804783618687086,
      "grad_norm": 0.90625,
      "learning_rate": 0.00040512496146617906,
      "loss": 0.8476,
      "step": 32600
    },
    {
      "epoch": 2.8048696549944077,
      "grad_norm": 1.28125,
      "learning_rate": 0.0004050984121340355,
      "loss": 1.3126,
      "step": 32601
    },
    {
      "epoch": 2.8049556913017293,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004050718630794607,
      "loss": 1.3455,
      "step": 32602
    },
    {
      "epoch": 2.805041727609051,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0004050453143025323,
      "loss": 0.5325,
      "step": 32603
    },
    {
      "epoch": 2.8051277639163725,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004050187658033277,
      "loss": 0.9155,
      "step": 32604
    },
    {
      "epoch": 2.8052138002236946,
      "grad_norm": 2.078125,
      "learning_rate": 0.0004049922175819248,
      "loss": 1.4059,
      "step": 32605
    },
    {
      "epoch": 2.8052998365310162,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00040496566963840134,
      "loss": 0.9795,
      "step": 32606
    },
    {
      "epoch": 2.805385872838338,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0004049391219728347,
      "loss": 1.6373,
      "step": 32607
    },
    {
      "epoch": 2.8054719091456595,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004049125745853026,
      "loss": 0.824,
      "step": 32608
    },
    {
      "epoch": 2.805557945452981,
      "grad_norm": 5.71875,
      "learning_rate": 0.00040488602747588276,
      "loss": 1.5133,
      "step": 32609
    },
    {
      "epoch": 2.8056439817603027,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00040485948064465286,
      "loss": 1.2482,
      "step": 32610
    },
    {
      "epoch": 2.8057300180676243,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0004048329340916903,
      "loss": 1.0362,
      "step": 32611
    },
    {
      "epoch": 2.8058160543749464,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00040480638781707307,
      "loss": 1.0163,
      "step": 32612
    },
    {
      "epoch": 2.805902090682268,
      "grad_norm": 8.1875,
      "learning_rate": 0.0004047798418208786,
      "loss": 0.6963,
      "step": 32613
    },
    {
      "epoch": 2.8059881269895897,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004047532961031846,
      "loss": 1.2461,
      "step": 32614
    },
    {
      "epoch": 2.8060741632969113,
      "grad_norm": 1.203125,
      "learning_rate": 0.0004047267506640686,
      "loss": 0.8022,
      "step": 32615
    },
    {
      "epoch": 2.806160199604233,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00040470020550360835,
      "loss": 1.5489,
      "step": 32616
    },
    {
      "epoch": 2.8062462359115545,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00040467366062188156,
      "loss": 1.6354,
      "step": 32617
    },
    {
      "epoch": 2.806332272218876,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004046471160189656,
      "loss": 1.1887,
      "step": 32618
    },
    {
      "epoch": 2.8064183085261982,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00040462057169493836,
      "loss": 1.7924,
      "step": 32619
    },
    {
      "epoch": 2.80650434483352,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00040459402764987754,
      "loss": 1.0181,
      "step": 32620
    },
    {
      "epoch": 2.8065903811408415,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004045674838838605,
      "loss": 1.6461,
      "step": 32621
    },
    {
      "epoch": 2.806676417448163,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00040454094039696494,
      "loss": 1.1839,
      "step": 32622
    },
    {
      "epoch": 2.8067624537554847,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00040451439718926864,
      "loss": 1.2767,
      "step": 32623
    },
    {
      "epoch": 2.8068484900628063,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004044878542608492,
      "loss": 1.1063,
      "step": 32624
    },
    {
      "epoch": 2.806934526370128,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004044613116117842,
      "loss": 1.405,
      "step": 32625
    },
    {
      "epoch": 2.80702056267745,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00040443476924215116,
      "loss": 0.7199,
      "step": 32626
    },
    {
      "epoch": 2.8071065989847717,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00040440822715202796,
      "loss": 0.9071,
      "step": 32627
    },
    {
      "epoch": 2.8071926352920933,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004043816853414921,
      "loss": 1.3027,
      "step": 32628
    },
    {
      "epoch": 2.807278671599415,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004043551438106211,
      "loss": 1.5576,
      "step": 32629
    },
    {
      "epoch": 2.8073647079067365,
      "grad_norm": 1.125,
      "learning_rate": 0.0004043286025594928,
      "loss": 1.7345,
      "step": 32630
    },
    {
      "epoch": 2.8074507442140586,
      "grad_norm": 1.109375,
      "learning_rate": 0.00040430206158818474,
      "loss": 1.3134,
      "step": 32631
    },
    {
      "epoch": 2.80753678052138,
      "grad_norm": 1.390625,
      "learning_rate": 0.00040427552089677455,
      "loss": 0.9245,
      "step": 32632
    },
    {
      "epoch": 2.807622816828702,
      "grad_norm": 1.078125,
      "learning_rate": 0.00040424898048533974,
      "loss": 1.2424,
      "step": 32633
    },
    {
      "epoch": 2.8077088531360235,
      "grad_norm": 2.15625,
      "learning_rate": 0.0004042224403539581,
      "loss": 0.9627,
      "step": 32634
    },
    {
      "epoch": 2.807794889443345,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004041959005027073,
      "loss": 1.3227,
      "step": 32635
    },
    {
      "epoch": 2.8078809257506667,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0004041693609316645,
      "loss": 1.7942,
      "step": 32636
    },
    {
      "epoch": 2.8079669620579883,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000404142821640908,
      "loss": 1.6567,
      "step": 32637
    },
    {
      "epoch": 2.8080529983653104,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0004041162826305149,
      "loss": 1.5923,
      "step": 32638
    },
    {
      "epoch": 2.8081390346726316,
      "grad_norm": 1.125,
      "learning_rate": 0.0004040897439005632,
      "loss": 0.8899,
      "step": 32639
    },
    {
      "epoch": 2.8082250709799537,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004040632054511304,
      "loss": 0.6598,
      "step": 32640
    },
    {
      "epoch": 2.8083111072872753,
      "grad_norm": 1.25,
      "learning_rate": 0.0004040366672822939,
      "loss": 1.6021,
      "step": 32641
    },
    {
      "epoch": 2.808397143594597,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004040101293941316,
      "loss": 1.1624,
      "step": 32642
    },
    {
      "epoch": 2.8084831799019185,
      "grad_norm": 1.15625,
      "learning_rate": 0.00040398359178672093,
      "loss": 1.0225,
      "step": 32643
    },
    {
      "epoch": 2.80856921620924,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004039570544601397,
      "loss": 1.3963,
      "step": 32644
    },
    {
      "epoch": 2.8086552525165622,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004039305174144652,
      "loss": 1.3964,
      "step": 32645
    },
    {
      "epoch": 2.8087412888238834,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004039039806497754,
      "loss": 1.6679,
      "step": 32646
    },
    {
      "epoch": 2.8088273251312055,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004038774441661478,
      "loss": 0.6972,
      "step": 32647
    },
    {
      "epoch": 2.808913361438527,
      "grad_norm": 1.046875,
      "learning_rate": 0.00040385090796365996,
      "loss": 0.6834,
      "step": 32648
    },
    {
      "epoch": 2.8089993977458487,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0004038243720423894,
      "loss": 1.2978,
      "step": 32649
    },
    {
      "epoch": 2.8090854340531703,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00040379783640241397,
      "loss": 0.9641,
      "step": 32650
    },
    {
      "epoch": 2.809171470360492,
      "grad_norm": 1.6875,
      "learning_rate": 0.0004037713010438111,
      "loss": 1.28,
      "step": 32651
    },
    {
      "epoch": 2.809257506667814,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0004037447659666584,
      "loss": 0.9398,
      "step": 32652
    },
    {
      "epoch": 2.8093435429751357,
      "grad_norm": 0.859375,
      "learning_rate": 0.00040371823117103366,
      "loss": 0.5832,
      "step": 32653
    },
    {
      "epoch": 2.8094295792824573,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004036916966570143,
      "loss": 0.9932,
      "step": 32654
    },
    {
      "epoch": 2.809515615589779,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0004036651624246781,
      "loss": 0.8437,
      "step": 32655
    },
    {
      "epoch": 2.8096016518971005,
      "grad_norm": 1.0625,
      "learning_rate": 0.00040363862847410235,
      "loss": 1.112,
      "step": 32656
    },
    {
      "epoch": 2.809687688204422,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000403612094805365,
      "loss": 1.5116,
      "step": 32657
    },
    {
      "epoch": 2.809773724511744,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00040358556141854363,
      "loss": 0.7006,
      "step": 32658
    },
    {
      "epoch": 2.809859760819066,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004035590283137155,
      "loss": 1.5051,
      "step": 32659
    },
    {
      "epoch": 2.8099457971263875,
      "grad_norm": 1.03125,
      "learning_rate": 0.00040353249549095866,
      "loss": 1.5479,
      "step": 32660
    },
    {
      "epoch": 2.810031833433709,
      "grad_norm": 1.09375,
      "learning_rate": 0.00040350596295035045,
      "loss": 2.0261,
      "step": 32661
    },
    {
      "epoch": 2.8101178697410307,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00040347943069196857,
      "loss": 1.0091,
      "step": 32662
    },
    {
      "epoch": 2.8102039060483524,
      "grad_norm": 0.8125,
      "learning_rate": 0.0004034528987158904,
      "loss": 0.5998,
      "step": 32663
    },
    {
      "epoch": 2.810289942355674,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0004034263670221939,
      "loss": 0.8307,
      "step": 32664
    },
    {
      "epoch": 2.8103759786629956,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004033998356109565,
      "loss": 0.9562,
      "step": 32665
    },
    {
      "epoch": 2.8104620149703177,
      "grad_norm": 0.984375,
      "learning_rate": 0.00040337330448225576,
      "loss": 1.1438,
      "step": 32666
    },
    {
      "epoch": 2.8105480512776393,
      "grad_norm": 1.265625,
      "learning_rate": 0.00040334677363616924,
      "loss": 0.9044,
      "step": 32667
    },
    {
      "epoch": 2.810634087584961,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004033202430727747,
      "loss": 0.7952,
      "step": 32668
    },
    {
      "epoch": 2.8107201238922825,
      "grad_norm": 5.625,
      "learning_rate": 0.0004032937127921497,
      "loss": 1.4334,
      "step": 32669
    },
    {
      "epoch": 2.810806160199604,
      "grad_norm": 0.875,
      "learning_rate": 0.00040326718279437156,
      "loss": 0.984,
      "step": 32670
    },
    {
      "epoch": 2.810892196506926,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004032406530795182,
      "loss": 0.6901,
      "step": 32671
    },
    {
      "epoch": 2.8109782328142474,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004032141236476672,
      "loss": 1.4529,
      "step": 32672
    },
    {
      "epoch": 2.8110642691215695,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00040318759449889586,
      "loss": 1.5582,
      "step": 32673
    },
    {
      "epoch": 2.811150305428891,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00040316106563328215,
      "loss": 1.0165,
      "step": 32674
    },
    {
      "epoch": 2.8112363417362127,
      "grad_norm": 0.875,
      "learning_rate": 0.00040313453705090335,
      "loss": 1.2915,
      "step": 32675
    },
    {
      "epoch": 2.8113223780435344,
      "grad_norm": 1.171875,
      "learning_rate": 0.00040310800875183737,
      "loss": 1.1602,
      "step": 32676
    },
    {
      "epoch": 2.811408414350856,
      "grad_norm": 1.71875,
      "learning_rate": 0.00040308148073616156,
      "loss": 1.4664,
      "step": 32677
    },
    {
      "epoch": 2.8114944506581776,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00040305495300395355,
      "loss": 1.0283,
      "step": 32678
    },
    {
      "epoch": 2.8115804869654992,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0004030284255552909,
      "loss": 0.9803,
      "step": 32679
    },
    {
      "epoch": 2.8116665232728213,
      "grad_norm": 1.03125,
      "learning_rate": 0.00040300189839025135,
      "loss": 1.1019,
      "step": 32680
    },
    {
      "epoch": 2.811752559580143,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00040297537150891236,
      "loss": 1.6467,
      "step": 32681
    },
    {
      "epoch": 2.8118385958874645,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004029488449113514,
      "loss": 0.8746,
      "step": 32682
    },
    {
      "epoch": 2.811924632194786,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00040292231859764637,
      "loss": 1.5972,
      "step": 32683
    },
    {
      "epoch": 2.812010668502108,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004028957925678746,
      "loss": 1.4525,
      "step": 32684
    },
    {
      "epoch": 2.8120967048094294,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004028692668221138,
      "loss": 1.3115,
      "step": 32685
    },
    {
      "epoch": 2.812182741116751,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0004028427413604414,
      "loss": 0.7478,
      "step": 32686
    },
    {
      "epoch": 2.812268777424073,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0004028162161829352,
      "loss": 1.2411,
      "step": 32687
    },
    {
      "epoch": 2.8123548137313947,
      "grad_norm": 133.0,
      "learning_rate": 0.00040278969128967266,
      "loss": 1.2411,
      "step": 32688
    },
    {
      "epoch": 2.8124408500387164,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0004027631666807314,
      "loss": 1.2636,
      "step": 32689
    },
    {
      "epoch": 2.812526886346038,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0004027366423561888,
      "loss": 1.2649,
      "step": 32690
    },
    {
      "epoch": 2.8126129226533596,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004027101183161227,
      "loss": 1.0755,
      "step": 32691
    },
    {
      "epoch": 2.8126989589606817,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004026835945606106,
      "loss": 1.4991,
      "step": 32692
    },
    {
      "epoch": 2.812784995268003,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004026570710897299,
      "loss": 1.3759,
      "step": 32693
    },
    {
      "epoch": 2.812871031575325,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004026305479035586,
      "loss": 0.8412,
      "step": 32694
    },
    {
      "epoch": 2.8129570678826465,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00040260402500217387,
      "loss": 1.48,
      "step": 32695
    },
    {
      "epoch": 2.813043104189968,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00040257750238565347,
      "loss": 1.0315,
      "step": 32696
    },
    {
      "epoch": 2.81312914049729,
      "grad_norm": 1.109375,
      "learning_rate": 0.0004025509800540748,
      "loss": 1.2792,
      "step": 32697
    },
    {
      "epoch": 2.8132151768046114,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00040252445800751567,
      "loss": 1.2528,
      "step": 32698
    },
    {
      "epoch": 2.8133012131119335,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0004024979362460536,
      "loss": 1.0132,
      "step": 32699
    },
    {
      "epoch": 2.8133872494192547,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00040247141476976595,
      "loss": 1.9503,
      "step": 32700
    },
    {
      "epoch": 2.8134732857265767,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00040244489357873055,
      "loss": 1.6627,
      "step": 32701
    },
    {
      "epoch": 2.8135593220338984,
      "grad_norm": 0.90625,
      "learning_rate": 0.0004024183726730249,
      "loss": 0.6372,
      "step": 32702
    },
    {
      "epoch": 2.81364535834122,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004023918520527265,
      "loss": 1.1898,
      "step": 32703
    },
    {
      "epoch": 2.8137313946485416,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004023653317179128,
      "loss": 1.3942,
      "step": 32704
    },
    {
      "epoch": 2.8138174309558632,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004023388116686617,
      "loss": 1.1924,
      "step": 32705
    },
    {
      "epoch": 2.8139034672631853,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004023122919050505,
      "loss": 1.1657,
      "step": 32706
    },
    {
      "epoch": 2.8139895035705065,
      "grad_norm": 1.421875,
      "learning_rate": 0.0004022857724271568,
      "loss": 2.0372,
      "step": 32707
    },
    {
      "epoch": 2.8140755398778285,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0004022592532350583,
      "loss": 1.7174,
      "step": 32708
    },
    {
      "epoch": 2.81416157618515,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00040223273432883253,
      "loss": 1.1032,
      "step": 32709
    },
    {
      "epoch": 2.814247612492472,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0004022062157085567,
      "loss": 1.717,
      "step": 32710
    },
    {
      "epoch": 2.8143336487997934,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0004021796973743089,
      "loss": 1.0795,
      "step": 32711
    },
    {
      "epoch": 2.814419685107115,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004021531793261665,
      "loss": 1.1932,
      "step": 32712
    },
    {
      "epoch": 2.814505721414437,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00040212666156420684,
      "loss": 2.203,
      "step": 32713
    },
    {
      "epoch": 2.8145917577217587,
      "grad_norm": 1.0625,
      "learning_rate": 0.00040210014408850785,
      "loss": 1.774,
      "step": 32714
    },
    {
      "epoch": 2.8146777940290804,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0004020736268991468,
      "loss": 0.8146,
      "step": 32715
    },
    {
      "epoch": 2.814763830336402,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004020471099962014,
      "loss": 1.1803,
      "step": 32716
    },
    {
      "epoch": 2.8148498666437236,
      "grad_norm": 0.828125,
      "learning_rate": 0.0004020205933797491,
      "loss": 0.8712,
      "step": 32717
    },
    {
      "epoch": 2.8149359029510452,
      "grad_norm": 0.8125,
      "learning_rate": 0.00040199407704986754,
      "loss": 0.6433,
      "step": 32718
    },
    {
      "epoch": 2.815021939258367,
      "grad_norm": 1.71875,
      "learning_rate": 0.00040196756100663436,
      "loss": 1.3886,
      "step": 32719
    },
    {
      "epoch": 2.815107975565689,
      "grad_norm": 1.078125,
      "learning_rate": 0.00040194104525012673,
      "loss": 1.3209,
      "step": 32720
    },
    {
      "epoch": 2.8151940118730105,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004019145297804227,
      "loss": 1.0423,
      "step": 32721
    },
    {
      "epoch": 2.815280048180332,
      "grad_norm": 0.96875,
      "learning_rate": 0.00040188801459759954,
      "loss": 1.7697,
      "step": 32722
    },
    {
      "epoch": 2.815366084487654,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00040186149970173474,
      "loss": 0.8903,
      "step": 32723
    },
    {
      "epoch": 2.8154521207949754,
      "grad_norm": 1.15625,
      "learning_rate": 0.0004018349850929061,
      "loss": 1.2898,
      "step": 32724
    },
    {
      "epoch": 2.815538157102297,
      "grad_norm": 1.78125,
      "learning_rate": 0.00040180847077119105,
      "loss": 1.4424,
      "step": 32725
    },
    {
      "epoch": 2.8156241934096187,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0004017819567366671,
      "loss": 1.1013,
      "step": 32726
    },
    {
      "epoch": 2.8157102297169407,
      "grad_norm": 1.59375,
      "learning_rate": 0.0004017554429894117,
      "loss": 1.017,
      "step": 32727
    },
    {
      "epoch": 2.8157962660242624,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004017289295295027,
      "loss": 0.9111,
      "step": 32728
    },
    {
      "epoch": 2.815882302331584,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00040170241635701737,
      "loss": 1.1075,
      "step": 32729
    },
    {
      "epoch": 2.8159683386389056,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0004016759034720333,
      "loss": 1.1728,
      "step": 32730
    },
    {
      "epoch": 2.8160543749462272,
      "grad_norm": 1.140625,
      "learning_rate": 0.00040164939087462815,
      "loss": 1.1349,
      "step": 32731
    },
    {
      "epoch": 2.816140411253549,
      "grad_norm": 1.0,
      "learning_rate": 0.00040162287856487946,
      "loss": 0.8586,
      "step": 32732
    },
    {
      "epoch": 2.8162264475608705,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0004015963665428647,
      "loss": 1.3378,
      "step": 32733
    },
    {
      "epoch": 2.8163124838681926,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0004015698548086613,
      "loss": 1.201,
      "step": 32734
    },
    {
      "epoch": 2.816398520175514,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004015433433623471,
      "loss": 1.0216,
      "step": 32735
    },
    {
      "epoch": 2.816484556482836,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00040151683220399943,
      "loss": 1.0272,
      "step": 32736
    },
    {
      "epoch": 2.8165705927901574,
      "grad_norm": 1.390625,
      "learning_rate": 0.0004014903213336959,
      "loss": 1.5215,
      "step": 32737
    },
    {
      "epoch": 2.816656629097479,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00040146381075151385,
      "loss": 0.974,
      "step": 32738
    },
    {
      "epoch": 2.8167426654048007,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00040143730045753113,
      "loss": 1.3963,
      "step": 32739
    },
    {
      "epoch": 2.8168287017121223,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00040141079045182514,
      "loss": 1.0225,
      "step": 32740
    },
    {
      "epoch": 2.8169147380194444,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0004013842807344733,
      "loss": 0.5483,
      "step": 32741
    },
    {
      "epoch": 2.817000774326766,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004013577713055534,
      "loss": 1.0784,
      "step": 32742
    },
    {
      "epoch": 2.8170868106340876,
      "grad_norm": 1.25,
      "learning_rate": 0.00040133126216514284,
      "loss": 1.7196,
      "step": 32743
    },
    {
      "epoch": 2.8171728469414092,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00040130475331331906,
      "loss": 0.6347,
      "step": 32744
    },
    {
      "epoch": 2.817258883248731,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004012782447501596,
      "loss": 1.7457,
      "step": 32745
    },
    {
      "epoch": 2.817344919556053,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00040125173647574225,
      "loss": 0.6269,
      "step": 32746
    },
    {
      "epoch": 2.817430955863374,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004012252284901441,
      "loss": 1.0071,
      "step": 32747
    },
    {
      "epoch": 2.817516992170696,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00040119872079344323,
      "loss": 1.0592,
      "step": 32748
    },
    {
      "epoch": 2.817603028478018,
      "grad_norm": 1.25,
      "learning_rate": 0.0004011722133857168,
      "loss": 1.1133,
      "step": 32749
    },
    {
      "epoch": 2.8176890647853394,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00040114570626704235,
      "loss": 0.8457,
      "step": 32750
    },
    {
      "epoch": 2.817775101092661,
      "grad_norm": 1.1875,
      "learning_rate": 0.00040111919943749757,
      "loss": 1.6702,
      "step": 32751
    },
    {
      "epoch": 2.8178611373999827,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0004010926928971599,
      "loss": 1.9044,
      "step": 32752
    },
    {
      "epoch": 2.8179471737073047,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0004010661866461069,
      "loss": 1.3254,
      "step": 32753
    },
    {
      "epoch": 2.818033210014626,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00040103968068441596,
      "loss": 0.6996,
      "step": 32754
    },
    {
      "epoch": 2.818119246321948,
      "grad_norm": 1.03125,
      "learning_rate": 0.0004010131750121648,
      "loss": 0.8804,
      "step": 32755
    },
    {
      "epoch": 2.8182052826292696,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00040098666962943083,
      "loss": 1.2474,
      "step": 32756
    },
    {
      "epoch": 2.8182913189365912,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00040096016453629145,
      "loss": 1.9674,
      "step": 32757
    },
    {
      "epoch": 2.818377355243913,
      "grad_norm": 1.84375,
      "learning_rate": 0.00040093365973282454,
      "loss": 1.0642,
      "step": 32758
    },
    {
      "epoch": 2.8184633915512345,
      "grad_norm": 1.25,
      "learning_rate": 0.00040090715521910736,
      "loss": 1.6435,
      "step": 32759
    },
    {
      "epoch": 2.8185494278585566,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0004008806509952174,
      "loss": 1.4207,
      "step": 32760
    },
    {
      "epoch": 2.8186354641658777,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00040085414706123225,
      "loss": 0.9115,
      "step": 32761
    },
    {
      "epoch": 2.8187215004732,
      "grad_norm": 0.953125,
      "learning_rate": 0.0004008276434172296,
      "loss": 0.7343,
      "step": 32762
    },
    {
      "epoch": 2.8188075367805214,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00040080114006328665,
      "loss": 2.0582,
      "step": 32763
    },
    {
      "epoch": 2.818893573087843,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000400774636999481,
      "loss": 0.7605,
      "step": 32764
    },
    {
      "epoch": 2.8189796093951647,
      "grad_norm": 1.0,
      "learning_rate": 0.0004007481342258904,
      "loss": 1.5177,
      "step": 32765
    },
    {
      "epoch": 2.8190656457024863,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00040072163174259225,
      "loss": 1.142,
      "step": 32766
    },
    {
      "epoch": 2.8191516820098084,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004006951295496639,
      "loss": 1.1568,
      "step": 32767
    },
    {
      "epoch": 2.81923771831713,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00040066862764718295,
      "loss": 0.8738,
      "step": 32768
    },
    {
      "epoch": 2.8193237546244516,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00040064212603522706,
      "loss": 0.9283,
      "step": 32769
    },
    {
      "epoch": 2.8194097909317732,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00040061562471387357,
      "loss": 1.1965,
      "step": 32770
    },
    {
      "epoch": 2.819495827239095,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004005891236832,
      "loss": 1.2034,
      "step": 32771
    },
    {
      "epoch": 2.8195818635464165,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00040056262294328396,
      "loss": 1.171,
      "step": 32772
    },
    {
      "epoch": 2.819667899853738,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00040053612249420293,
      "loss": 1.4132,
      "step": 32773
    },
    {
      "epoch": 2.81975393616106,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004005096223360344,
      "loss": 1.2333,
      "step": 32774
    },
    {
      "epoch": 2.819839972468382,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004004831224688558,
      "loss": 1.144,
      "step": 32775
    },
    {
      "epoch": 2.8199260087757034,
      "grad_norm": 1.1875,
      "learning_rate": 0.00040045662289274475,
      "loss": 1.5661,
      "step": 32776
    },
    {
      "epoch": 2.820012045083025,
      "grad_norm": 1.0625,
      "learning_rate": 0.0004004301236077788,
      "loss": 0.8652,
      "step": 32777
    },
    {
      "epoch": 2.8200980813903467,
      "grad_norm": 1.078125,
      "learning_rate": 0.0004004036246140353,
      "loss": 1.0242,
      "step": 32778
    },
    {
      "epoch": 2.8201841176976683,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00040037712591159175,
      "loss": 1.052,
      "step": 32779
    },
    {
      "epoch": 2.82027015400499,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0004003506275005259,
      "loss": 1.0782,
      "step": 32780
    },
    {
      "epoch": 2.820356190312312,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000400324129380915,
      "loss": 0.9418,
      "step": 32781
    },
    {
      "epoch": 2.8204422266196336,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00040029763155283654,
      "loss": 1.4443,
      "step": 32782
    },
    {
      "epoch": 2.8205282629269552,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0004002711340163682,
      "loss": 0.7828,
      "step": 32783
    },
    {
      "epoch": 2.820614299234277,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00040024463677158724,
      "loss": 1.7778,
      "step": 32784
    },
    {
      "epoch": 2.8207003355415985,
      "grad_norm": 1.546875,
      "learning_rate": 0.0004002181398185715,
      "loss": 1.3789,
      "step": 32785
    },
    {
      "epoch": 2.82078637184892,
      "grad_norm": 0.984375,
      "learning_rate": 0.0004001916431573983,
      "loss": 1.4206,
      "step": 32786
    },
    {
      "epoch": 2.8208724081562417,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000400165146788145,
      "loss": 1.3884,
      "step": 32787
    },
    {
      "epoch": 2.820958444463564,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0004001386507108894,
      "loss": 1.3841,
      "step": 32788
    },
    {
      "epoch": 2.8210444807708854,
      "grad_norm": 1.328125,
      "learning_rate": 0.0004001121549257088,
      "loss": 1.1254,
      "step": 32789
    },
    {
      "epoch": 2.821130517078207,
      "grad_norm": 1.1875,
      "learning_rate": 0.0004000856594326807,
      "loss": 1.0754,
      "step": 32790
    },
    {
      "epoch": 2.8212165533855287,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0004000591642318825,
      "loss": 2.1192,
      "step": 32791
    },
    {
      "epoch": 2.8213025896928503,
      "grad_norm": 1.265625,
      "learning_rate": 0.00040003266932339194,
      "loss": 0.8658,
      "step": 32792
    },
    {
      "epoch": 2.821388626000172,
      "grad_norm": 3.640625,
      "learning_rate": 0.0004000061747072864,
      "loss": 1.1059,
      "step": 32793
    },
    {
      "epoch": 2.8214746623074936,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003999796803836432,
      "loss": 1.3915,
      "step": 32794
    },
    {
      "epoch": 2.8215606986148156,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003999531863525401,
      "loss": 1.5395,
      "step": 32795
    },
    {
      "epoch": 2.8216467349221372,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003999266926140545,
      "loss": 1.1878,
      "step": 32796
    },
    {
      "epoch": 2.821732771229459,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003999001991682638,
      "loss": 0.9229,
      "step": 32797
    },
    {
      "epoch": 2.8218188075367805,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003998737060152455,
      "loss": 1.4829,
      "step": 32798
    },
    {
      "epoch": 2.821904843844102,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00039984721315507724,
      "loss": 1.1247,
      "step": 32799
    },
    {
      "epoch": 2.8219908801514237,
      "grad_norm": 1.140625,
      "learning_rate": 0.00039982072058783637,
      "loss": 0.9007,
      "step": 32800
    },
    {
      "epoch": 2.8220769164587454,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003997942283136004,
      "loss": 0.8577,
      "step": 32801
    },
    {
      "epoch": 2.8221629527660674,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003997677363324467,
      "loss": 1.0358,
      "step": 32802
    },
    {
      "epoch": 2.822248989073389,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00039974124464445303,
      "loss": 1.4515,
      "step": 32803
    },
    {
      "epoch": 2.8223350253807107,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00039971475324969674,
      "loss": 0.8798,
      "step": 32804
    },
    {
      "epoch": 2.8224210616880323,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00039968826214825513,
      "loss": 0.5252,
      "step": 32805
    },
    {
      "epoch": 2.822507097995354,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000399661771340206,
      "loss": 1.692,
      "step": 32806
    },
    {
      "epoch": 2.822593134302676,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00039963528082562666,
      "loss": 1.3179,
      "step": 32807
    },
    {
      "epoch": 2.822679170609997,
      "grad_norm": 2.59375,
      "learning_rate": 0.0003996087906045946,
      "loss": 0.9825,
      "step": 32808
    },
    {
      "epoch": 2.8227652069173192,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0003995823006771872,
      "loss": 0.847,
      "step": 32809
    },
    {
      "epoch": 2.822851243224641,
      "grad_norm": 1.5625,
      "learning_rate": 0.00039955581104348216,
      "loss": 1.3429,
      "step": 32810
    },
    {
      "epoch": 2.8229372795319625,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00039952932170355684,
      "loss": 1.3126,
      "step": 32811
    },
    {
      "epoch": 2.823023315839284,
      "grad_norm": 1.03125,
      "learning_rate": 0.00039950283265748857,
      "loss": 1.421,
      "step": 32812
    },
    {
      "epoch": 2.8231093521466057,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00039947634390535505,
      "loss": 1.2066,
      "step": 32813
    },
    {
      "epoch": 2.823195388453928,
      "grad_norm": 1.625,
      "learning_rate": 0.00039944985544723374,
      "loss": 0.5351,
      "step": 32814
    },
    {
      "epoch": 2.823281424761249,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000399423367283202,
      "loss": 1.4113,
      "step": 32815
    },
    {
      "epoch": 2.823367461068571,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00039939687941333723,
      "loss": 1.2701,
      "step": 32816
    },
    {
      "epoch": 2.8234534973758927,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003993703918377172,
      "loss": 1.0988,
      "step": 32817
    },
    {
      "epoch": 2.8235395336832143,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00039934390455641916,
      "loss": 0.996,
      "step": 32818
    },
    {
      "epoch": 2.823625569990536,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003993174175695206,
      "loss": 0.9022,
      "step": 32819
    },
    {
      "epoch": 2.8237116062978576,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00039929093087709894,
      "loss": 1.1505,
      "step": 32820
    },
    {
      "epoch": 2.8237976426051796,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003992644444792316,
      "loss": 0.7118,
      "step": 32821
    },
    {
      "epoch": 2.823883678912501,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003992379583759964,
      "loss": 1.1987,
      "step": 32822
    },
    {
      "epoch": 2.823969715219823,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00039921147256747057,
      "loss": 1.5082,
      "step": 32823
    },
    {
      "epoch": 2.8240557515271445,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003991849870537316,
      "loss": 1.766,
      "step": 32824
    },
    {
      "epoch": 2.824141787834466,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00039915850183485675,
      "loss": 1.1157,
      "step": 32825
    },
    {
      "epoch": 2.8242278241417877,
      "grad_norm": 1.0625,
      "learning_rate": 0.00039913201691092385,
      "loss": 0.6887,
      "step": 32826
    },
    {
      "epoch": 2.8243138604491094,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00039910553228201016,
      "loss": 0.7079,
      "step": 32827
    },
    {
      "epoch": 2.8243998967564314,
      "grad_norm": 1.0,
      "learning_rate": 0.00039907904794819304,
      "loss": 1.4462,
      "step": 32828
    },
    {
      "epoch": 2.824485933063753,
      "grad_norm": 2.9375,
      "learning_rate": 0.0003990525639095502,
      "loss": 0.8633,
      "step": 32829
    },
    {
      "epoch": 2.8245719693710747,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000399026080166159,
      "loss": 1.0113,
      "step": 32830
    },
    {
      "epoch": 2.8246580056783963,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00039899959671809684,
      "loss": 0.8254,
      "step": 32831
    },
    {
      "epoch": 2.824744041985718,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003989731135654411,
      "loss": 1.8461,
      "step": 32832
    },
    {
      "epoch": 2.8248300782930396,
      "grad_norm": 1.046875,
      "learning_rate": 0.00039894663070826953,
      "loss": 1.7912,
      "step": 32833
    },
    {
      "epoch": 2.824916114600361,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003989201481466593,
      "loss": 1.1226,
      "step": 32834
    },
    {
      "epoch": 2.8250021509076833,
      "grad_norm": 1.3125,
      "learning_rate": 0.000398893665880688,
      "loss": 1.1895,
      "step": 32835
    },
    {
      "epoch": 2.825088187215005,
      "grad_norm": 1.0625,
      "learning_rate": 0.00039886718391043307,
      "loss": 1.2232,
      "step": 32836
    },
    {
      "epoch": 2.8251742235223265,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00039884070223597203,
      "loss": 1.1116,
      "step": 32837
    },
    {
      "epoch": 2.825260259829648,
      "grad_norm": 1.0,
      "learning_rate": 0.00039881422085738224,
      "loss": 1.8126,
      "step": 32838
    },
    {
      "epoch": 2.8253462961369697,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00039878773977474103,
      "loss": 1.8095,
      "step": 32839
    },
    {
      "epoch": 2.8254323324442914,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003987612589881261,
      "loss": 1.0336,
      "step": 32840
    },
    {
      "epoch": 2.825518368751613,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003987347784976149,
      "loss": 1.639,
      "step": 32841
    },
    {
      "epoch": 2.825604405058935,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00039870829830328464,
      "loss": 1.0536,
      "step": 32842
    },
    {
      "epoch": 2.8256904413662567,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003986818184052129,
      "loss": 1.7788,
      "step": 32843
    },
    {
      "epoch": 2.8257764776735783,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00039865533880347723,
      "loss": 1.0158,
      "step": 32844
    },
    {
      "epoch": 2.8258625139809,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00039862885949815493,
      "loss": 1.4505,
      "step": 32845
    },
    {
      "epoch": 2.8259485502882216,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0003986023804893234,
      "loss": 1.2806,
      "step": 32846
    },
    {
      "epoch": 2.826034586595543,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00039857590177706033,
      "loss": 0.7805,
      "step": 32847
    },
    {
      "epoch": 2.826120622902865,
      "grad_norm": 1.140625,
      "learning_rate": 0.000398549423361443,
      "loss": 0.7455,
      "step": 32848
    },
    {
      "epoch": 2.826206659210187,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003985229452425489,
      "loss": 1.2241,
      "step": 32849
    },
    {
      "epoch": 2.8262926955175085,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0003984964674204553,
      "loss": 0.9093,
      "step": 32850
    },
    {
      "epoch": 2.82637873182483,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00039846998989524,
      "loss": 0.5101,
      "step": 32851
    },
    {
      "epoch": 2.8264647681321518,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003984435126669801,
      "loss": 1.0942,
      "step": 32852
    },
    {
      "epoch": 2.8265508044394734,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003984170357357531,
      "loss": 1.092,
      "step": 32853
    },
    {
      "epoch": 2.826636840746795,
      "grad_norm": 1.5625,
      "learning_rate": 0.00039839055910163666,
      "loss": 0.9488,
      "step": 32854
    },
    {
      "epoch": 2.8267228770541166,
      "grad_norm": 1.25,
      "learning_rate": 0.000398364082764708,
      "loss": 1.7012,
      "step": 32855
    },
    {
      "epoch": 2.8268089133614387,
      "grad_norm": 1.71875,
      "learning_rate": 0.0003983376067250448,
      "loss": 1.6227,
      "step": 32856
    },
    {
      "epoch": 2.8268949496687603,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003983111309827241,
      "loss": 1.5494,
      "step": 32857
    },
    {
      "epoch": 2.826980985976082,
      "grad_norm": 1.328125,
      "learning_rate": 0.00039828465553782346,
      "loss": 2.0803,
      "step": 32858
    },
    {
      "epoch": 2.8270670222834036,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003982581803904206,
      "loss": 1.5153,
      "step": 32859
    },
    {
      "epoch": 2.827153058590725,
      "grad_norm": 1.109375,
      "learning_rate": 0.00039823170554059283,
      "loss": 1.0422,
      "step": 32860
    },
    {
      "epoch": 2.827239094898047,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003982052309884175,
      "loss": 1.3167,
      "step": 32861
    },
    {
      "epoch": 2.8273251312053684,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000398178756733972,
      "loss": 0.9692,
      "step": 32862
    },
    {
      "epoch": 2.8274111675126905,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003981522827773339,
      "loss": 1.0437,
      "step": 32863
    },
    {
      "epoch": 2.827497203820012,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003981258091185806,
      "loss": 0.8405,
      "step": 32864
    },
    {
      "epoch": 2.8275832401273338,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003980993357577895,
      "loss": 1.3502,
      "step": 32865
    },
    {
      "epoch": 2.8276692764346554,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003980728626950378,
      "loss": 0.9247,
      "step": 32866
    },
    {
      "epoch": 2.827755312741977,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00039804638993040344,
      "loss": 0.861,
      "step": 32867
    },
    {
      "epoch": 2.827841349049299,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00039801991746396347,
      "loss": 1.3349,
      "step": 32868
    },
    {
      "epoch": 2.8279273853566202,
      "grad_norm": 1.0,
      "learning_rate": 0.00039799344529579527,
      "loss": 0.8751,
      "step": 32869
    },
    {
      "epoch": 2.8280134216639423,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003979669734259766,
      "loss": 1.2097,
      "step": 32870
    },
    {
      "epoch": 2.828099457971264,
      "grad_norm": 1.078125,
      "learning_rate": 0.00039794050185458464,
      "loss": 1.254,
      "step": 32871
    },
    {
      "epoch": 2.8281854942785856,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003979140305816968,
      "loss": 0.9807,
      "step": 32872
    },
    {
      "epoch": 2.828271530585907,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003978875596073905,
      "loss": 0.7581,
      "step": 32873
    },
    {
      "epoch": 2.828357566893229,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003978610889317434,
      "loss": 1.3775,
      "step": 32874
    },
    {
      "epoch": 2.828443603200551,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003978346185548327,
      "loss": 0.7646,
      "step": 32875
    },
    {
      "epoch": 2.828529639507872,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003978081484767357,
      "loss": 0.9923,
      "step": 32876
    },
    {
      "epoch": 2.828615675815194,
      "grad_norm": 1.140625,
      "learning_rate": 0.00039778167869753024,
      "loss": 0.9103,
      "step": 32877
    },
    {
      "epoch": 2.8287017121225158,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003977552092172934,
      "loss": 0.5727,
      "step": 32878
    },
    {
      "epoch": 2.8287877484298374,
      "grad_norm": 1.25,
      "learning_rate": 0.00039772874003610273,
      "loss": 1.491,
      "step": 32879
    },
    {
      "epoch": 2.828873784737159,
      "grad_norm": 1.4375,
      "learning_rate": 0.00039770227115403547,
      "loss": 1.2715,
      "step": 32880
    },
    {
      "epoch": 2.8289598210444806,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00039767580257116925,
      "loss": 0.4424,
      "step": 32881
    },
    {
      "epoch": 2.8290458573518027,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003976493342875814,
      "loss": 1.0636,
      "step": 32882
    },
    {
      "epoch": 2.829131893659124,
      "grad_norm": 1.125,
      "learning_rate": 0.0003976228663033493,
      "loss": 1.1021,
      "step": 32883
    },
    {
      "epoch": 2.829217929966446,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00039759639861855045,
      "loss": 1.0964,
      "step": 32884
    },
    {
      "epoch": 2.8293039662737676,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003975699312332623,
      "loss": 0.9472,
      "step": 32885
    },
    {
      "epoch": 2.829390002581089,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003975434641475621,
      "loss": 1.2112,
      "step": 32886
    },
    {
      "epoch": 2.829476038888411,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00039751699736152724,
      "loss": 1.298,
      "step": 32887
    },
    {
      "epoch": 2.8295620751957324,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0003974905308752354,
      "loss": 0.5086,
      "step": 32888
    },
    {
      "epoch": 2.8296481115030545,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003974640646887637,
      "loss": 0.9583,
      "step": 32889
    },
    {
      "epoch": 2.829734147810376,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00039743759880218985,
      "loss": 1.4347,
      "step": 32890
    },
    {
      "epoch": 2.8298201841176978,
      "grad_norm": 0.84375,
      "learning_rate": 0.0003974111332155908,
      "loss": 0.5148,
      "step": 32891
    },
    {
      "epoch": 2.8299062204250194,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003973846679290444,
      "loss": 1.0341,
      "step": 32892
    },
    {
      "epoch": 2.829992256732341,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003973582029426279,
      "loss": 0.7429,
      "step": 32893
    },
    {
      "epoch": 2.8300782930396626,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003973317382564186,
      "loss": 1.1907,
      "step": 32894
    },
    {
      "epoch": 2.8301643293469843,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00039730527387049397,
      "loss": 1.1276,
      "step": 32895
    },
    {
      "epoch": 2.8302503656543063,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003972788097849315,
      "loss": 1.2003,
      "step": 32896
    },
    {
      "epoch": 2.830336401961628,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003972523459998085,
      "loss": 1.4628,
      "step": 32897
    },
    {
      "epoch": 2.8304224382689496,
      "grad_norm": 1.234375,
      "learning_rate": 0.00039722588251520254,
      "loss": 1.3792,
      "step": 32898
    },
    {
      "epoch": 2.830508474576271,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039719941933119075,
      "loss": 1.5193,
      "step": 32899
    },
    {
      "epoch": 2.830594510883593,
      "grad_norm": 1.125,
      "learning_rate": 0.0003971729564478507,
      "loss": 1.4103,
      "step": 32900
    },
    {
      "epoch": 2.8306805471909144,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00039714649386525985,
      "loss": 0.6384,
      "step": 32901
    },
    {
      "epoch": 2.830766583498236,
      "grad_norm": 1.1875,
      "learning_rate": 0.00039712003158349547,
      "loss": 2.447,
      "step": 32902
    },
    {
      "epoch": 2.830852619805558,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00039709356960263486,
      "loss": 0.8542,
      "step": 32903
    },
    {
      "epoch": 2.8309386561128798,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00039706710792275566,
      "loss": 0.807,
      "step": 32904
    },
    {
      "epoch": 2.8310246924202014,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003970406465439351,
      "loss": 1.7556,
      "step": 32905
    },
    {
      "epoch": 2.831110728727523,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003970141854662506,
      "loss": 1.0768,
      "step": 32906
    },
    {
      "epoch": 2.8311967650348446,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00039698772468977966,
      "loss": 1.7655,
      "step": 32907
    },
    {
      "epoch": 2.8312828013421663,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003969612642145996,
      "loss": 1.229,
      "step": 32908
    },
    {
      "epoch": 2.831368837649488,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003969348040407878,
      "loss": 1.392,
      "step": 32909
    },
    {
      "epoch": 2.83145487395681,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00039690834416842153,
      "loss": 0.922,
      "step": 32910
    },
    {
      "epoch": 2.8315409102641316,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003968818845975784,
      "loss": 1.2299,
      "step": 32911
    },
    {
      "epoch": 2.831626946571453,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00039685542532833575,
      "loss": 0.8849,
      "step": 32912
    },
    {
      "epoch": 2.831712982878775,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00039682896636077087,
      "loss": 1.0474,
      "step": 32913
    },
    {
      "epoch": 2.8317990191860964,
      "grad_norm": 1.1875,
      "learning_rate": 0.00039680250769496113,
      "loss": 1.7187,
      "step": 32914
    },
    {
      "epoch": 2.831885055493418,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00039677604933098406,
      "loss": 1.1589,
      "step": 32915
    },
    {
      "epoch": 2.8319710918007397,
      "grad_norm": 4.6875,
      "learning_rate": 0.00039674959126891697,
      "loss": 1.2855,
      "step": 32916
    },
    {
      "epoch": 2.8320571281080618,
      "grad_norm": 1.34375,
      "learning_rate": 0.00039672313350883715,
      "loss": 1.6719,
      "step": 32917
    },
    {
      "epoch": 2.8321431644153834,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00039669667605082213,
      "loss": 1.853,
      "step": 32918
    },
    {
      "epoch": 2.832229200722705,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003966702188949494,
      "loss": 0.7366,
      "step": 32919
    },
    {
      "epoch": 2.8323152370300266,
      "grad_norm": 1.03125,
      "learning_rate": 0.000396643762041296,
      "loss": 0.8491,
      "step": 32920
    },
    {
      "epoch": 2.8324012733373483,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003966173054899395,
      "loss": 1.4419,
      "step": 32921
    },
    {
      "epoch": 2.8324873096446703,
      "grad_norm": 1.890625,
      "learning_rate": 0.00039659084924095734,
      "loss": 2.0313,
      "step": 32922
    },
    {
      "epoch": 2.8325733459519915,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00039656439329442683,
      "loss": 0.971,
      "step": 32923
    },
    {
      "epoch": 2.8326593822593136,
      "grad_norm": 4.40625,
      "learning_rate": 0.0003965379376504252,
      "loss": 1.0117,
      "step": 32924
    },
    {
      "epoch": 2.832745418566635,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00039651148230903017,
      "loss": 1.129,
      "step": 32925
    },
    {
      "epoch": 2.832831454873957,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003964850272703189,
      "loss": 1.0057,
      "step": 32926
    },
    {
      "epoch": 2.8329174911812784,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003964585725343688,
      "loss": 1.1016,
      "step": 32927
    },
    {
      "epoch": 2.8330035274886,
      "grad_norm": 1.1875,
      "learning_rate": 0.00039643211810125713,
      "loss": 1.1415,
      "step": 32928
    },
    {
      "epoch": 2.833089563795922,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00039640566397106146,
      "loss": 0.8852,
      "step": 32929
    },
    {
      "epoch": 2.8331756001032433,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00039637921014385905,
      "loss": 1.304,
      "step": 32930
    },
    {
      "epoch": 2.8332616364105654,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039635275661972736,
      "loss": 0.5426,
      "step": 32931
    },
    {
      "epoch": 2.833347672717887,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003963263033987434,
      "loss": 0.9108,
      "step": 32932
    },
    {
      "epoch": 2.8334337090252086,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039629985048098503,
      "loss": 1.225,
      "step": 32933
    },
    {
      "epoch": 2.8335197453325303,
      "grad_norm": 1.21875,
      "learning_rate": 0.00039627339786652954,
      "loss": 0.8651,
      "step": 32934
    },
    {
      "epoch": 2.833605781639852,
      "grad_norm": 1.046875,
      "learning_rate": 0.00039624694555545414,
      "loss": 1.0418,
      "step": 32935
    },
    {
      "epoch": 2.833691817947174,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003962204935478363,
      "loss": 1.5474,
      "step": 32936
    },
    {
      "epoch": 2.833777854254495,
      "grad_norm": 6.5625,
      "learning_rate": 0.0003961940418437532,
      "loss": 1.0001,
      "step": 32937
    },
    {
      "epoch": 2.833863890561817,
      "grad_norm": 1.328125,
      "learning_rate": 0.00039616759044328243,
      "loss": 1.8912,
      "step": 32938
    },
    {
      "epoch": 2.833949926869139,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00039614113934650127,
      "loss": 0.6284,
      "step": 32939
    },
    {
      "epoch": 2.8340359631764604,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00039611468855348694,
      "loss": 0.684,
      "step": 32940
    },
    {
      "epoch": 2.834121999483782,
      "grad_norm": 1.125,
      "learning_rate": 0.000396088238064317,
      "loss": 1.0024,
      "step": 32941
    },
    {
      "epoch": 2.8342080357911037,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00039606178787906886,
      "loss": 1.1234,
      "step": 32942
    },
    {
      "epoch": 2.8342940720984258,
      "grad_norm": 0.859375,
      "learning_rate": 0.00039603533799781977,
      "loss": 1.4117,
      "step": 32943
    },
    {
      "epoch": 2.8343801084057474,
      "grad_norm": 1.28125,
      "learning_rate": 0.000396008888420647,
      "loss": 0.7316,
      "step": 32944
    },
    {
      "epoch": 2.834466144713069,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00039598243914762806,
      "loss": 0.8761,
      "step": 32945
    },
    {
      "epoch": 2.8345521810203906,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00039595599017884026,
      "loss": 1.3436,
      "step": 32946
    },
    {
      "epoch": 2.8346382173277123,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003959295415143609,
      "loss": 1.1806,
      "step": 32947
    },
    {
      "epoch": 2.834724253635034,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003959030931542674,
      "loss": 0.8058,
      "step": 32948
    },
    {
      "epoch": 2.8348102899423555,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0003958766450986372,
      "loss": 1.7629,
      "step": 32949
    },
    {
      "epoch": 2.8348963262496776,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00039585019734754754,
      "loss": 1.7473,
      "step": 32950
    },
    {
      "epoch": 2.834982362556999,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003958237499010756,
      "loss": 1.1154,
      "step": 32951
    },
    {
      "epoch": 2.835068398864321,
      "grad_norm": 0.953125,
      "learning_rate": 0.00039579730275929916,
      "loss": 1.341,
      "step": 32952
    },
    {
      "epoch": 2.8351544351716425,
      "grad_norm": 6.65625,
      "learning_rate": 0.00039577085592229527,
      "loss": 0.8213,
      "step": 32953
    },
    {
      "epoch": 2.835240471478964,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00039574440939014137,
      "loss": 0.9639,
      "step": 32954
    },
    {
      "epoch": 2.8353265077862857,
      "grad_norm": 1.015625,
      "learning_rate": 0.00039571796316291473,
      "loss": 1.0339,
      "step": 32955
    },
    {
      "epoch": 2.8354125440936073,
      "grad_norm": 2.515625,
      "learning_rate": 0.00039569151724069287,
      "loss": 1.5634,
      "step": 32956
    },
    {
      "epoch": 2.8354985804009294,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00039566507162355297,
      "loss": 1.3518,
      "step": 32957
    },
    {
      "epoch": 2.835584616708251,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003956386263115724,
      "loss": 0.7605,
      "step": 32958
    },
    {
      "epoch": 2.8356706530155726,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003956121813048286,
      "loss": 1.4803,
      "step": 32959
    },
    {
      "epoch": 2.8357566893228943,
      "grad_norm": 0.84375,
      "learning_rate": 0.00039558573660339894,
      "loss": 0.7907,
      "step": 32960
    },
    {
      "epoch": 2.835842725630216,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00039555929220736066,
      "loss": 1.316,
      "step": 32961
    },
    {
      "epoch": 2.8359287619375375,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000395532848116791,
      "loss": 0.757,
      "step": 32962
    },
    {
      "epoch": 2.836014798244859,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00039550640433176756,
      "loss": 1.3793,
      "step": 32963
    },
    {
      "epoch": 2.836100834552181,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003954799608523676,
      "loss": 1.5549,
      "step": 32964
    },
    {
      "epoch": 2.836186870859503,
      "grad_norm": 1.078125,
      "learning_rate": 0.00039545351767866825,
      "loss": 0.8812,
      "step": 32965
    },
    {
      "epoch": 2.8362729071668245,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00039542707481074723,
      "loss": 0.9558,
      "step": 32966
    },
    {
      "epoch": 2.836358943474146,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003954006322486816,
      "loss": 0.9377,
      "step": 32967
    },
    {
      "epoch": 2.8364449797814677,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003953741899925488,
      "loss": 0.5304,
      "step": 32968
    },
    {
      "epoch": 2.8365310160887893,
      "grad_norm": 1.7421875,
      "learning_rate": 0.00039534774804242596,
      "loss": 1.0047,
      "step": 32969
    },
    {
      "epoch": 2.836617052396111,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00039532130639839073,
      "loss": 1.57,
      "step": 32970
    },
    {
      "epoch": 2.836703088703433,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00039529486506052037,
      "loss": 1.4545,
      "step": 32971
    },
    {
      "epoch": 2.8367891250107546,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003952684240288922,
      "loss": 1.6118,
      "step": 32972
    },
    {
      "epoch": 2.8368751613180763,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003952419833035835,
      "loss": 0.4546,
      "step": 32973
    },
    {
      "epoch": 2.836961197625398,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003952155428846715,
      "loss": 1.4857,
      "step": 32974
    },
    {
      "epoch": 2.8370472339327195,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00039518910277223376,
      "loss": 1.8372,
      "step": 32975
    },
    {
      "epoch": 2.837133270240041,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003951626629663475,
      "loss": 1.5087,
      "step": 32976
    },
    {
      "epoch": 2.8372193065473628,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003951362234670901,
      "loss": 1.8644,
      "step": 32977
    },
    {
      "epoch": 2.837305342854685,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00039510978427453875,
      "loss": 1.6462,
      "step": 32978
    },
    {
      "epoch": 2.8373913791620065,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000395083345388771,
      "loss": 1.4644,
      "step": 32979
    },
    {
      "epoch": 2.837477415469328,
      "grad_norm": 1.0625,
      "learning_rate": 0.000395056906809864,
      "loss": 1.1048,
      "step": 32980
    },
    {
      "epoch": 2.8375634517766497,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003950304685378951,
      "loss": 1.4063,
      "step": 32981
    },
    {
      "epoch": 2.8376494880839713,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003950040305729417,
      "loss": 1.2914,
      "step": 32982
    },
    {
      "epoch": 2.8377355243912934,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00039497759291508115,
      "loss": 1.233,
      "step": 32983
    },
    {
      "epoch": 2.8378215606986146,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003949511555643907,
      "loss": 0.9119,
      "step": 32984
    },
    {
      "epoch": 2.8379075970059366,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003949247185209476,
      "loss": 1.1904,
      "step": 32985
    },
    {
      "epoch": 2.8379936333132583,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00039489828178482933,
      "loss": 1.649,
      "step": 32986
    },
    {
      "epoch": 2.83807966962058,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00039487184535611316,
      "loss": 0.9673,
      "step": 32987
    },
    {
      "epoch": 2.8381657059279015,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003948454092348763,
      "loss": 1.3989,
      "step": 32988
    },
    {
      "epoch": 2.838251742235223,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003948189734211962,
      "loss": 0.9089,
      "step": 32989
    },
    {
      "epoch": 2.838337778542545,
      "grad_norm": 1.015625,
      "learning_rate": 0.00039479253791515023,
      "loss": 1.4102,
      "step": 32990
    },
    {
      "epoch": 2.8384238148498664,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003947661027168156,
      "loss": 0.7467,
      "step": 32991
    },
    {
      "epoch": 2.8385098511571885,
      "grad_norm": 1.046875,
      "learning_rate": 0.00039473966782626956,
      "loss": 0.7329,
      "step": 32992
    },
    {
      "epoch": 2.83859588746451,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003947132332435896,
      "loss": 1.2891,
      "step": 32993
    },
    {
      "epoch": 2.8386819237718317,
      "grad_norm": 2.921875,
      "learning_rate": 0.00039468679896885293,
      "loss": 1.6806,
      "step": 32994
    },
    {
      "epoch": 2.8387679600791533,
      "grad_norm": 1.015625,
      "learning_rate": 0.000394660365002137,
      "loss": 0.8125,
      "step": 32995
    },
    {
      "epoch": 2.838853996386475,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00039463393134351883,
      "loss": 0.913,
      "step": 32996
    },
    {
      "epoch": 2.838940032693797,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00039460749799307603,
      "loss": 1.0357,
      "step": 32997
    },
    {
      "epoch": 2.839026069001118,
      "grad_norm": 1.484375,
      "learning_rate": 0.0003945810649508858,
      "loss": 1.3483,
      "step": 32998
    },
    {
      "epoch": 2.8391121053084403,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00039455463221702534,
      "loss": 1.577,
      "step": 32999
    },
    {
      "epoch": 2.839198141615762,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003945281997915722,
      "loss": 0.8854,
      "step": 33000
    },
    {
      "epoch": 2.8392841779230835,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00039450176767460353,
      "loss": 1.4675,
      "step": 33001
    },
    {
      "epoch": 2.839370214230405,
      "grad_norm": 3.796875,
      "learning_rate": 0.0003944753358661967,
      "loss": 1.6338,
      "step": 33002
    },
    {
      "epoch": 2.8394562505377268,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003944489043664289,
      "loss": 1.2563,
      "step": 33003
    },
    {
      "epoch": 2.839542286845049,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003944224731753776,
      "loss": 1.0898,
      "step": 33004
    },
    {
      "epoch": 2.8396283231523705,
      "grad_norm": 1.125,
      "learning_rate": 0.00039439604229312005,
      "loss": 1.0257,
      "step": 33005
    },
    {
      "epoch": 2.839714359459692,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003943696117197333,
      "loss": 1.4531,
      "step": 33006
    },
    {
      "epoch": 2.8398003957670137,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00039434318145529514,
      "loss": 1.2122,
      "step": 33007
    },
    {
      "epoch": 2.8398864320743353,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0003943167514998824,
      "loss": 0.5873,
      "step": 33008
    },
    {
      "epoch": 2.839972468381657,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00039429032185357287,
      "loss": 1.9036,
      "step": 33009
    },
    {
      "epoch": 2.8400585046889786,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039426389251644357,
      "loss": 0.9364,
      "step": 33010
    },
    {
      "epoch": 2.8401445409963006,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00039423746348857166,
      "loss": 1.0694,
      "step": 33011
    },
    {
      "epoch": 2.8402305773036223,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003942110347700347,
      "loss": 1.6245,
      "step": 33012
    },
    {
      "epoch": 2.840316613610944,
      "grad_norm": 0.875,
      "learning_rate": 0.0003941846063609099,
      "loss": 0.7774,
      "step": 33013
    },
    {
      "epoch": 2.8404026499182655,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003941581782612746,
      "loss": 1.0083,
      "step": 33014
    },
    {
      "epoch": 2.840488686225587,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003941317504712059,
      "loss": 0.7074,
      "step": 33015
    },
    {
      "epoch": 2.8405747225329088,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00039410532299078126,
      "loss": 0.7312,
      "step": 33016
    },
    {
      "epoch": 2.8406607588402304,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003940788958200781,
      "loss": 1.4709,
      "step": 33017
    },
    {
      "epoch": 2.8407467951475525,
      "grad_norm": 0.8125,
      "learning_rate": 0.00039405246895917346,
      "loss": 1.9797,
      "step": 33018
    },
    {
      "epoch": 2.840832831454874,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003940260424081447,
      "loss": 1.8039,
      "step": 33019
    },
    {
      "epoch": 2.8409188677621957,
      "grad_norm": 1.453125,
      "learning_rate": 0.00039399961616706926,
      "loss": 1.1703,
      "step": 33020
    },
    {
      "epoch": 2.8410049040695173,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0003939731902360243,
      "loss": 0.8241,
      "step": 33021
    },
    {
      "epoch": 2.841090940376839,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00039394676461508706,
      "loss": 0.6826,
      "step": 33022
    },
    {
      "epoch": 2.8411769766841606,
      "grad_norm": 0.984375,
      "learning_rate": 0.000393920339304335,
      "loss": 0.6829,
      "step": 33023
    },
    {
      "epoch": 2.841263012991482,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00039389391430384523,
      "loss": 0.7655,
      "step": 33024
    },
    {
      "epoch": 2.8413490492988043,
      "grad_norm": 1.15625,
      "learning_rate": 0.00039386748961369524,
      "loss": 0.9976,
      "step": 33025
    },
    {
      "epoch": 2.841435085606126,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003938410652339621,
      "loss": 1.1828,
      "step": 33026
    },
    {
      "epoch": 2.8415211219134475,
      "grad_norm": 0.765625,
      "learning_rate": 0.0003938146411647232,
      "loss": 0.812,
      "step": 33027
    },
    {
      "epoch": 2.841607158220769,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003937882174060559,
      "loss": 1.9407,
      "step": 33028
    },
    {
      "epoch": 2.8416931945280908,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003937617939580373,
      "loss": 0.9064,
      "step": 33029
    },
    {
      "epoch": 2.8417792308354124,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00039373537082074486,
      "loss": 1.0081,
      "step": 33030
    },
    {
      "epoch": 2.841865267142734,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00039370894799425584,
      "loss": 1.321,
      "step": 33031
    },
    {
      "epoch": 2.841951303450056,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003936825254786475,
      "loss": 1.3025,
      "step": 33032
    },
    {
      "epoch": 2.8420373397573777,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00039365610327399687,
      "loss": 0.7171,
      "step": 33033
    },
    {
      "epoch": 2.8421233760646993,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00039362968138038164,
      "loss": 1.2508,
      "step": 33034
    },
    {
      "epoch": 2.842209412372021,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003936032597978789,
      "loss": 1.2144,
      "step": 33035
    },
    {
      "epoch": 2.8422954486793426,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003935768385265658,
      "loss": 1.4049,
      "step": 33036
    },
    {
      "epoch": 2.842381484986664,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003935504175665199,
      "loss": 0.7493,
      "step": 33037
    },
    {
      "epoch": 2.842467521293986,
      "grad_norm": 1.328125,
      "learning_rate": 0.00039352399691781827,
      "loss": 1.5892,
      "step": 33038
    },
    {
      "epoch": 2.842553557601308,
      "grad_norm": 7.53125,
      "learning_rate": 0.0003934975765805383,
      "loss": 1.065,
      "step": 33039
    },
    {
      "epoch": 2.8426395939086295,
      "grad_norm": 1.28125,
      "learning_rate": 0.00039347115655475706,
      "loss": 0.7549,
      "step": 33040
    },
    {
      "epoch": 2.842725630215951,
      "grad_norm": 1.375,
      "learning_rate": 0.0003934447368405521,
      "loss": 1.3971,
      "step": 33041
    },
    {
      "epoch": 2.8428116665232728,
      "grad_norm": 0.796875,
      "learning_rate": 0.0003934183174380005,
      "loss": 1.0147,
      "step": 33042
    },
    {
      "epoch": 2.8428977028305944,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00039339189834717946,
      "loss": 1.1294,
      "step": 33043
    },
    {
      "epoch": 2.8429837391379165,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00039336547956816655,
      "loss": 1.376,
      "step": 33044
    },
    {
      "epoch": 2.8430697754452376,
      "grad_norm": 1.28125,
      "learning_rate": 0.00039333906110103876,
      "loss": 1.5991,
      "step": 33045
    },
    {
      "epoch": 2.8431558117525597,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003933126429458736,
      "loss": 1.1078,
      "step": 33046
    },
    {
      "epoch": 2.8432418480598813,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00039328622510274825,
      "loss": 1.2154,
      "step": 33047
    },
    {
      "epoch": 2.843327884367203,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00039325980757173986,
      "loss": 0.8426,
      "step": 33048
    },
    {
      "epoch": 2.8434139206745246,
      "grad_norm": 1.640625,
      "learning_rate": 0.0003932333903529257,
      "loss": 1.1096,
      "step": 33049
    },
    {
      "epoch": 2.843499956981846,
      "grad_norm": 1.40625,
      "learning_rate": 0.00039320697344638324,
      "loss": 1.2138,
      "step": 33050
    },
    {
      "epoch": 2.8435859932891683,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00039318055685218955,
      "loss": 1.4913,
      "step": 33051
    },
    {
      "epoch": 2.8436720295964895,
      "grad_norm": 1.125,
      "learning_rate": 0.0003931541405704219,
      "loss": 1.2137,
      "step": 33052
    },
    {
      "epoch": 2.8437580659038115,
      "grad_norm": 1.421875,
      "learning_rate": 0.00039312772460115767,
      "loss": 1.9542,
      "step": 33053
    },
    {
      "epoch": 2.843844102211133,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003931013089444741,
      "loss": 0.7871,
      "step": 33054
    },
    {
      "epoch": 2.8439301385184548,
      "grad_norm": 2.34375,
      "learning_rate": 0.00039307489360044845,
      "loss": 0.7615,
      "step": 33055
    },
    {
      "epoch": 2.8440161748257764,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003930484785691578,
      "loss": 0.5579,
      "step": 33056
    },
    {
      "epoch": 2.844102211133098,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003930220638506796,
      "loss": 1.2578,
      "step": 33057
    },
    {
      "epoch": 2.84418824744042,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00039299564944509106,
      "loss": 0.8193,
      "step": 33058
    },
    {
      "epoch": 2.8442742837477413,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003929692353524694,
      "loss": 1.8962,
      "step": 33059
    },
    {
      "epoch": 2.8443603200550633,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000392942821572892,
      "loss": 1.3623,
      "step": 33060
    },
    {
      "epoch": 2.844446356362385,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000392916408106436,
      "loss": 0.9058,
      "step": 33061
    },
    {
      "epoch": 2.8445323926697066,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00039288999495317866,
      "loss": 1.6405,
      "step": 33062
    },
    {
      "epoch": 2.844618428977028,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003928635821131972,
      "loss": 1.1239,
      "step": 33063
    },
    {
      "epoch": 2.84470446528435,
      "grad_norm": 1.7890625,
      "learning_rate": 0.000392837169586569,
      "loss": 1.0365,
      "step": 33064
    },
    {
      "epoch": 2.844790501591672,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00039281075737337127,
      "loss": 0.9503,
      "step": 33065
    },
    {
      "epoch": 2.8448765378989935,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003927843454736812,
      "loss": 0.8152,
      "step": 33066
    },
    {
      "epoch": 2.844962574206315,
      "grad_norm": 1.203125,
      "learning_rate": 0.000392757933887576,
      "loss": 1.5955,
      "step": 33067
    },
    {
      "epoch": 2.8450486105136368,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00039273152261513303,
      "loss": 1.82,
      "step": 33068
    },
    {
      "epoch": 2.8451346468209584,
      "grad_norm": 1.53125,
      "learning_rate": 0.00039270511165642947,
      "loss": 1.0373,
      "step": 33069
    },
    {
      "epoch": 2.84522068312828,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003926787010115426,
      "loss": 1.0804,
      "step": 33070
    },
    {
      "epoch": 2.8453067194356017,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003926522906805496,
      "loss": 1.3441,
      "step": 33071
    },
    {
      "epoch": 2.8453927557429237,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003926258806635279,
      "loss": 0.6848,
      "step": 33072
    },
    {
      "epoch": 2.8454787920502453,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003925994709605546,
      "loss": 1.0556,
      "step": 33073
    },
    {
      "epoch": 2.845564828357567,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003925730615717068,
      "loss": 0.8702,
      "step": 33074
    },
    {
      "epoch": 2.8456508646648886,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00039254665249706204,
      "loss": 1.1636,
      "step": 33075
    },
    {
      "epoch": 2.84573690097221,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003925202437366975,
      "loss": 1.1523,
      "step": 33076
    },
    {
      "epoch": 2.845822937279532,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00039249383529069013,
      "loss": 1.1603,
      "step": 33077
    },
    {
      "epoch": 2.8459089735868535,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00039246742715911756,
      "loss": 1.6394,
      "step": 33078
    },
    {
      "epoch": 2.8459950098941755,
      "grad_norm": 1.7421875,
      "learning_rate": 0.00039244101934205676,
      "loss": 1.9014,
      "step": 33079
    },
    {
      "epoch": 2.846081046201497,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003924146118395849,
      "loss": 0.7677,
      "step": 33080
    },
    {
      "epoch": 2.846167082508819,
      "grad_norm": 0.8125,
      "learning_rate": 0.0003923882046517796,
      "loss": 1.0997,
      "step": 33081
    },
    {
      "epoch": 2.8462531188161404,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00039236179777871774,
      "loss": 2.2355,
      "step": 33082
    },
    {
      "epoch": 2.846339155123462,
      "grad_norm": 1.109375,
      "learning_rate": 0.00039233539122047686,
      "loss": 0.6226,
      "step": 33083
    },
    {
      "epoch": 2.8464251914307837,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00039230898497713394,
      "loss": 1.003,
      "step": 33084
    },
    {
      "epoch": 2.8465112277381053,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003922825790487663,
      "loss": 1.2805,
      "step": 33085
    },
    {
      "epoch": 2.8465972640454273,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003922561734354511,
      "loss": 0.952,
      "step": 33086
    },
    {
      "epoch": 2.846683300352749,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003922297681372657,
      "loss": 0.8335,
      "step": 33087
    },
    {
      "epoch": 2.8467693366600706,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00039220336315428733,
      "loss": 0.5384,
      "step": 33088
    },
    {
      "epoch": 2.846855372967392,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00039217695848659315,
      "loss": 1.5588,
      "step": 33089
    },
    {
      "epoch": 2.846941409274714,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00039215055413426025,
      "loss": 1.2605,
      "step": 33090
    },
    {
      "epoch": 2.8470274455820355,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003921241500973661,
      "loss": 2.0855,
      "step": 33091
    },
    {
      "epoch": 2.847113481889357,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00039209774637598787,
      "loss": 1.0985,
      "step": 33092
    },
    {
      "epoch": 2.847199518196679,
      "grad_norm": 1.078125,
      "learning_rate": 0.00039207134297020266,
      "loss": 0.8013,
      "step": 33093
    },
    {
      "epoch": 2.847285554504001,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00039204493988008785,
      "loss": 0.8707,
      "step": 33094
    },
    {
      "epoch": 2.8473715908113224,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0003920185371057206,
      "loss": 0.7943,
      "step": 33095
    },
    {
      "epoch": 2.847457627118644,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0003919921346471781,
      "loss": 0.8991,
      "step": 33096
    },
    {
      "epoch": 2.8475436634259657,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003919657325045375,
      "loss": 1.6417,
      "step": 33097
    },
    {
      "epoch": 2.8476296997332877,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00039193933067787625,
      "loss": 0.7941,
      "step": 33098
    },
    {
      "epoch": 2.847715736040609,
      "grad_norm": 1.125,
      "learning_rate": 0.00039191292916727146,
      "loss": 2.1942,
      "step": 33099
    },
    {
      "epoch": 2.847801772347931,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00039188652797280017,
      "loss": 1.5954,
      "step": 33100
    },
    {
      "epoch": 2.8478878086552526,
      "grad_norm": 1.046875,
      "learning_rate": 0.00039186012709453987,
      "loss": 1.1752,
      "step": 33101
    },
    {
      "epoch": 2.847973844962574,
      "grad_norm": 0.96875,
      "learning_rate": 0.00039183372653256774,
      "loss": 0.8029,
      "step": 33102
    },
    {
      "epoch": 2.848059881269896,
      "grad_norm": 0.90625,
      "learning_rate": 0.00039180732628696087,
      "loss": 1.1292,
      "step": 33103
    },
    {
      "epoch": 2.8481459175772175,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003917809263577964,
      "loss": 1.4375,
      "step": 33104
    },
    {
      "epoch": 2.8482319538845395,
      "grad_norm": 1.109375,
      "learning_rate": 0.00039175452674515185,
      "loss": 2.0219,
      "step": 33105
    },
    {
      "epoch": 2.8483179901918607,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0003917281274491042,
      "loss": 1.4257,
      "step": 33106
    },
    {
      "epoch": 2.848404026499183,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003917017284697307,
      "loss": 1.0293,
      "step": 33107
    },
    {
      "epoch": 2.8484900628065044,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003916753298071085,
      "loss": 0.9275,
      "step": 33108
    },
    {
      "epoch": 2.848576099113826,
      "grad_norm": 1.234375,
      "learning_rate": 0.000391648931461315,
      "loss": 1.3408,
      "step": 33109
    },
    {
      "epoch": 2.8486621354211477,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003916225334324273,
      "loss": 1.3316,
      "step": 33110
    },
    {
      "epoch": 2.8487481717284693,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00039159613572052257,
      "loss": 0.5916,
      "step": 33111
    },
    {
      "epoch": 2.8488342080357913,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003915697383256781,
      "loss": 1.5158,
      "step": 33112
    },
    {
      "epoch": 2.8489202443431125,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000391543341247971,
      "loss": 0.5365,
      "step": 33113
    },
    {
      "epoch": 2.8490062806504346,
      "grad_norm": 1.0625,
      "learning_rate": 0.00039151694448747863,
      "loss": 1.1604,
      "step": 33114
    },
    {
      "epoch": 2.849092316957756,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00039149054804427796,
      "loss": 0.9207,
      "step": 33115
    },
    {
      "epoch": 2.849178353265078,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003914641519184464,
      "loss": 1.0945,
      "step": 33116
    },
    {
      "epoch": 2.8492643895723995,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003914377561100609,
      "loss": 2.123,
      "step": 33117
    },
    {
      "epoch": 2.849350425879721,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00039141136061919914,
      "loss": 2.0182,
      "step": 33118
    },
    {
      "epoch": 2.849436462187043,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003913849654459379,
      "loss": 0.8026,
      "step": 33119
    },
    {
      "epoch": 2.849522498494365,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003913585705903545,
      "loss": 1.8495,
      "step": 33120
    },
    {
      "epoch": 2.8496085348016864,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003913321760525262,
      "loss": 1.4726,
      "step": 33121
    },
    {
      "epoch": 2.849694571109008,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00039130578183253024,
      "loss": 1.3802,
      "step": 33122
    },
    {
      "epoch": 2.8497806074163297,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00039127938793044354,
      "loss": 0.7211,
      "step": 33123
    },
    {
      "epoch": 2.8498666437236513,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003912529943463436,
      "loss": 1.9782,
      "step": 33124
    },
    {
      "epoch": 2.849952680030973,
      "grad_norm": 1.125,
      "learning_rate": 0.00039122660108030754,
      "loss": 1.1643,
      "step": 33125
    },
    {
      "epoch": 2.850038716338295,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00039120020813241253,
      "loss": 1.5608,
      "step": 33126
    },
    {
      "epoch": 2.8501247526456166,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003911738155027356,
      "loss": 1.6412,
      "step": 33127
    },
    {
      "epoch": 2.8502107889529382,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003911474231913542,
      "loss": 1.3825,
      "step": 33128
    },
    {
      "epoch": 2.85029682526026,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003911210311983454,
      "loss": 1.5367,
      "step": 33129
    },
    {
      "epoch": 2.8503828615675815,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00039109463952378645,
      "loss": 0.7046,
      "step": 33130
    },
    {
      "epoch": 2.850468897874903,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00039106824816775436,
      "loss": 0.8748,
      "step": 33131
    },
    {
      "epoch": 2.8505549341822247,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0003910418571303266,
      "loss": 0.4487,
      "step": 33132
    },
    {
      "epoch": 2.850640970489547,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00039101546641158013,
      "loss": 1.0844,
      "step": 33133
    },
    {
      "epoch": 2.8507270067968684,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003909890760115922,
      "loss": 1.0106,
      "step": 33134
    },
    {
      "epoch": 2.85081304310419,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003909626859304402,
      "loss": 1.7598,
      "step": 33135
    },
    {
      "epoch": 2.8508990794115117,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00039093629616820095,
      "loss": 1.0961,
      "step": 33136
    },
    {
      "epoch": 2.8509851157188333,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00039090990672495195,
      "loss": 1.5272,
      "step": 33137
    },
    {
      "epoch": 2.851071152026155,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003908835176007701,
      "loss": 1.2697,
      "step": 33138
    },
    {
      "epoch": 2.8511571883334765,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00039085712879573286,
      "loss": 1.0052,
      "step": 33139
    },
    {
      "epoch": 2.8512432246407986,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003908307403099173,
      "loss": 1.7363,
      "step": 33140
    },
    {
      "epoch": 2.8513292609481202,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003908043521434005,
      "loss": 0.9008,
      "step": 33141
    },
    {
      "epoch": 2.851415297255442,
      "grad_norm": 1.234375,
      "learning_rate": 0.00039077796429625984,
      "loss": 1.9635,
      "step": 33142
    },
    {
      "epoch": 2.8515013335627635,
      "grad_norm": 0.875,
      "learning_rate": 0.00039075157676857236,
      "loss": 1.5322,
      "step": 33143
    },
    {
      "epoch": 2.851587369870085,
      "grad_norm": 1.359375,
      "learning_rate": 0.00039072518956041524,
      "loss": 1.9981,
      "step": 33144
    },
    {
      "epoch": 2.8516734061774067,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00039069880267186564,
      "loss": 0.9086,
      "step": 33145
    },
    {
      "epoch": 2.8517594424847283,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039067241610300085,
      "loss": 1.6907,
      "step": 33146
    },
    {
      "epoch": 2.8518454787920504,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003906460298538981,
      "loss": 0.5096,
      "step": 33147
    },
    {
      "epoch": 2.851931515099372,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003906196439246342,
      "loss": 0.7791,
      "step": 33148
    },
    {
      "epoch": 2.8520175514066937,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003905932583152867,
      "loss": 1.8919,
      "step": 33149
    },
    {
      "epoch": 2.8521035877140153,
      "grad_norm": 1.0625,
      "learning_rate": 0.00039056687302593265,
      "loss": 1.5292,
      "step": 33150
    },
    {
      "epoch": 2.852189624021337,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0003905404880566492,
      "loss": 1.4398,
      "step": 33151
    },
    {
      "epoch": 2.8522756603286585,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00039051410340751346,
      "loss": 0.8174,
      "step": 33152
    },
    {
      "epoch": 2.85236169663598,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039048771907860283,
      "loss": 1.3918,
      "step": 33153
    },
    {
      "epoch": 2.8524477329433022,
      "grad_norm": 1.015625,
      "learning_rate": 0.00039046133506999404,
      "loss": 0.9571,
      "step": 33154
    },
    {
      "epoch": 2.852533769250624,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0003904349513817647,
      "loss": 2.1948,
      "step": 33155
    },
    {
      "epoch": 2.8526198055579455,
      "grad_norm": 1.0,
      "learning_rate": 0.00039040856801399193,
      "loss": 1.0266,
      "step": 33156
    },
    {
      "epoch": 2.852705841865267,
      "grad_norm": 1.65625,
      "learning_rate": 0.00039038218496675257,
      "loss": 1.7999,
      "step": 33157
    },
    {
      "epoch": 2.8527918781725887,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00039035580224012416,
      "loss": 1.7642,
      "step": 33158
    },
    {
      "epoch": 2.852877914479911,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003903294198341837,
      "loss": 0.9527,
      "step": 33159
    },
    {
      "epoch": 2.852963950787232,
      "grad_norm": 1.3125,
      "learning_rate": 0.00039030303774900845,
      "loss": 1.4046,
      "step": 33160
    },
    {
      "epoch": 2.853049987094554,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00039027665598467527,
      "loss": 1.551,
      "step": 33161
    },
    {
      "epoch": 2.8531360234018757,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003902502745412616,
      "loss": 1.1923,
      "step": 33162
    },
    {
      "epoch": 2.8532220597091973,
      "grad_norm": 1.28125,
      "learning_rate": 0.00039022389341884464,
      "loss": 1.397,
      "step": 33163
    },
    {
      "epoch": 2.853308096016519,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00039019751261750124,
      "loss": 0.8955,
      "step": 33164
    },
    {
      "epoch": 2.8533941323238405,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00039017113213730894,
      "loss": 0.7443,
      "step": 33165
    },
    {
      "epoch": 2.8534801686311626,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00039014475197834465,
      "loss": 1.2052,
      "step": 33166
    },
    {
      "epoch": 2.853566204938484,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003901183721406857,
      "loss": 0.8205,
      "step": 33167
    },
    {
      "epoch": 2.853652241245806,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003900919926244089,
      "loss": 0.7793,
      "step": 33168
    },
    {
      "epoch": 2.8537382775531275,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00039006561342959183,
      "loss": 0.9208,
      "step": 33169
    },
    {
      "epoch": 2.853824313860449,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00039003923455631145,
      "loss": 1.6611,
      "step": 33170
    },
    {
      "epoch": 2.8539103501677707,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003900128560046449,
      "loss": 1.7175,
      "step": 33171
    },
    {
      "epoch": 2.8539963864750924,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003899864777746693,
      "loss": 1.3607,
      "step": 33172
    },
    {
      "epoch": 2.8540824227824144,
      "grad_norm": 1.125,
      "learning_rate": 0.0003899600998664619,
      "loss": 0.6563,
      "step": 33173
    },
    {
      "epoch": 2.8541684590897356,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003899337222800999,
      "loss": 1.7056,
      "step": 33174
    },
    {
      "epoch": 2.8542544953970577,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003899073450156602,
      "loss": 1.3999,
      "step": 33175
    },
    {
      "epoch": 2.8543405317043793,
      "grad_norm": 1.3125,
      "learning_rate": 0.00038988096807322017,
      "loss": 1.7238,
      "step": 33176
    },
    {
      "epoch": 2.854426568011701,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003898545914528569,
      "loss": 0.8682,
      "step": 33177
    },
    {
      "epoch": 2.8545126043190225,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00038982821515464757,
      "loss": 1.1729,
      "step": 33178
    },
    {
      "epoch": 2.854598640626344,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003898018391786691,
      "loss": 0.9257,
      "step": 33179
    },
    {
      "epoch": 2.8546846769336662,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00038977546352499907,
      "loss": 1.5358,
      "step": 33180
    },
    {
      "epoch": 2.854770713240988,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00038974908819371427,
      "loss": 1.5812,
      "step": 33181
    },
    {
      "epoch": 2.8548567495483095,
      "grad_norm": 1.28125,
      "learning_rate": 0.00038972271318489185,
      "loss": 1.9137,
      "step": 33182
    },
    {
      "epoch": 2.854942785855631,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00038969633849860906,
      "loss": 1.6614,
      "step": 33183
    },
    {
      "epoch": 2.8550288221629527,
      "grad_norm": 1.125,
      "learning_rate": 0.00038966996413494315,
      "loss": 1.7859,
      "step": 33184
    },
    {
      "epoch": 2.8551148584702744,
      "grad_norm": 0.8125,
      "learning_rate": 0.0003896435900939711,
      "loss": 0.8688,
      "step": 33185
    },
    {
      "epoch": 2.855200894777596,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00038961721637577,
      "loss": 1.3555,
      "step": 33186
    },
    {
      "epoch": 2.855286931084918,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003895908429804171,
      "loss": 0.7896,
      "step": 33187
    },
    {
      "epoch": 2.8553729673922397,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00038956446990798965,
      "loss": 1.1324,
      "step": 33188
    },
    {
      "epoch": 2.8554590036995613,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003895380971585645,
      "loss": 1.1545,
      "step": 33189
    },
    {
      "epoch": 2.855545040006883,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00038951172473221894,
      "loss": 0.7404,
      "step": 33190
    },
    {
      "epoch": 2.8556310763142045,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00038948535262902994,
      "loss": 2.2104,
      "step": 33191
    },
    {
      "epoch": 2.855717112621526,
      "grad_norm": 0.984375,
      "learning_rate": 0.00038945898084907505,
      "loss": 1.2533,
      "step": 33192
    },
    {
      "epoch": 2.855803148928848,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003894326093924311,
      "loss": 1.7195,
      "step": 33193
    },
    {
      "epoch": 2.85588918523617,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00038940623825917527,
      "loss": 0.7828,
      "step": 33194
    },
    {
      "epoch": 2.8559752215434915,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003893798674493846,
      "loss": 1.2998,
      "step": 33195
    },
    {
      "epoch": 2.856061257850813,
      "grad_norm": 1.09375,
      "learning_rate": 0.00038935349696313633,
      "loss": 0.8041,
      "step": 33196
    },
    {
      "epoch": 2.8561472941581347,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003893271268005076,
      "loss": 1.7796,
      "step": 33197
    },
    {
      "epoch": 2.8562333304654564,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00038930075696157543,
      "loss": 0.7146,
      "step": 33198
    },
    {
      "epoch": 2.856319366772778,
      "grad_norm": 1.09375,
      "learning_rate": 0.00038927438744641707,
      "loss": 1.6082,
      "step": 33199
    },
    {
      "epoch": 2.8564054030800996,
      "grad_norm": 0.828125,
      "learning_rate": 0.00038924801825510963,
      "loss": 0.8199,
      "step": 33200
    },
    {
      "epoch": 2.8564914393874217,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003892216493877302,
      "loss": 0.9108,
      "step": 33201
    },
    {
      "epoch": 2.8565774756947433,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003891952808443558,
      "loss": 1.2813,
      "step": 33202
    },
    {
      "epoch": 2.856663512002065,
      "grad_norm": 1.375,
      "learning_rate": 0.0003891689126250637,
      "loss": 1.8561,
      "step": 33203
    },
    {
      "epoch": 2.8567495483093865,
      "grad_norm": 1.171875,
      "learning_rate": 0.00038914254472993107,
      "loss": 1.2177,
      "step": 33204
    },
    {
      "epoch": 2.856835584616708,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00038911617715903474,
      "loss": 0.9576,
      "step": 33205
    },
    {
      "epoch": 2.85692162092403,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038908980991245217,
      "loss": 1.1881,
      "step": 33206
    },
    {
      "epoch": 2.8570076572313514,
      "grad_norm": 0.9375,
      "learning_rate": 0.00038906344299026034,
      "loss": 1.4855,
      "step": 33207
    },
    {
      "epoch": 2.8570936935386735,
      "grad_norm": 1.53125,
      "learning_rate": 0.0003890370763925364,
      "loss": 2.0501,
      "step": 33208
    },
    {
      "epoch": 2.857179729845995,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003890107101193572,
      "loss": 1.2207,
      "step": 33209
    },
    {
      "epoch": 2.8572657661533167,
      "grad_norm": 1.609375,
      "learning_rate": 0.0003889843441708003,
      "loss": 1.6301,
      "step": 33210
    },
    {
      "epoch": 2.8573518024606384,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003889579785469426,
      "loss": 0.9282,
      "step": 33211
    },
    {
      "epoch": 2.85743783876796,
      "grad_norm": 1.515625,
      "learning_rate": 0.0003889316132478611,
      "loss": 1.7738,
      "step": 33212
    },
    {
      "epoch": 2.8575238750752816,
      "grad_norm": 2.078125,
      "learning_rate": 0.0003889052482736331,
      "loss": 0.9577,
      "step": 33213
    },
    {
      "epoch": 2.8576099113826032,
      "grad_norm": 1.109375,
      "learning_rate": 0.00038887888362433565,
      "loss": 1.5076,
      "step": 33214
    },
    {
      "epoch": 2.8576959476899253,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003888525193000459,
      "loss": 0.5083,
      "step": 33215
    },
    {
      "epoch": 2.857781983997247,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038882615530084076,
      "loss": 1.5062,
      "step": 33216
    },
    {
      "epoch": 2.8578680203045685,
      "grad_norm": 1.140625,
      "learning_rate": 0.00038879979162679764,
      "loss": 0.8956,
      "step": 33217
    },
    {
      "epoch": 2.85795405661189,
      "grad_norm": 1.25,
      "learning_rate": 0.00038877342827799345,
      "loss": 1.2843,
      "step": 33218
    },
    {
      "epoch": 2.858040092919212,
      "grad_norm": 1.015625,
      "learning_rate": 0.00038874706525450535,
      "loss": 1.0177,
      "step": 33219
    },
    {
      "epoch": 2.858126129226534,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003887207025564104,
      "loss": 0.7159,
      "step": 33220
    },
    {
      "epoch": 2.858212165533855,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003886943401837858,
      "loss": 0.8911,
      "step": 33221
    },
    {
      "epoch": 2.858298201841177,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00038866797813670866,
      "loss": 1.1889,
      "step": 33222
    },
    {
      "epoch": 2.8583842381484987,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003886416164152559,
      "loss": 1.6513,
      "step": 33223
    },
    {
      "epoch": 2.8584702744558204,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00038861525501950484,
      "loss": 1.5796,
      "step": 33224
    },
    {
      "epoch": 2.858556310763142,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00038858889394953255,
      "loss": 1.0246,
      "step": 33225
    },
    {
      "epoch": 2.8586423470704636,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00038856253320541603,
      "loss": 0.8103,
      "step": 33226
    },
    {
      "epoch": 2.8587283833777857,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003885361727872323,
      "loss": 1.1397,
      "step": 33227
    },
    {
      "epoch": 2.858814419685107,
      "grad_norm": 0.9375,
      "learning_rate": 0.00038850981269505854,
      "loss": 1.2531,
      "step": 33228
    },
    {
      "epoch": 2.858900455992429,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00038848345292897207,
      "loss": 1.0865,
      "step": 33229
    },
    {
      "epoch": 2.8589864922997505,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003884570934890498,
      "loss": 1.0216,
      "step": 33230
    },
    {
      "epoch": 2.859072528607072,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003884307343753689,
      "loss": 0.9507,
      "step": 33231
    },
    {
      "epoch": 2.859158564914394,
      "grad_norm": 1.140625,
      "learning_rate": 0.00038840437558800624,
      "loss": 1.4262,
      "step": 33232
    },
    {
      "epoch": 2.8592446012217154,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003883780171270392,
      "loss": 1.4455,
      "step": 33233
    },
    {
      "epoch": 2.8593306375290375,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003883516589925448,
      "loss": 1.3207,
      "step": 33234
    },
    {
      "epoch": 2.8594166738363587,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00038832530118459985,
      "loss": 1.3411,
      "step": 33235
    },
    {
      "epoch": 2.8595027101436807,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003882989437032819,
      "loss": 1.5537,
      "step": 33236
    },
    {
      "epoch": 2.8595887464510024,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00038827258654866776,
      "loss": 1.2617,
      "step": 33237
    },
    {
      "epoch": 2.859674782758324,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00038824622972083455,
      "loss": 1.061,
      "step": 33238
    },
    {
      "epoch": 2.8597608190656456,
      "grad_norm": 0.953125,
      "learning_rate": 0.00038821987321985935,
      "loss": 1.1486,
      "step": 33239
    },
    {
      "epoch": 2.8598468553729672,
      "grad_norm": 1.15625,
      "learning_rate": 0.00038819351704581936,
      "loss": 1.291,
      "step": 33240
    },
    {
      "epoch": 2.8599328916802893,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003881671611987916,
      "loss": 0.8143,
      "step": 33241
    },
    {
      "epoch": 2.860018927987611,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003881408056788531,
      "loss": 0.8169,
      "step": 33242
    },
    {
      "epoch": 2.8601049642949326,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003881144504860809,
      "loss": 1.3348,
      "step": 33243
    },
    {
      "epoch": 2.860191000602254,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00038808809562055225,
      "loss": 0.9853,
      "step": 33244
    },
    {
      "epoch": 2.860277036909576,
      "grad_norm": 1.125,
      "learning_rate": 0.00038806174108234424,
      "loss": 1.1091,
      "step": 33245
    },
    {
      "epoch": 2.8603630732168974,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003880353868715337,
      "loss": 0.9692,
      "step": 33246
    },
    {
      "epoch": 2.860449109524219,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003880090329881979,
      "loss": 0.8909,
      "step": 33247
    },
    {
      "epoch": 2.860535145831541,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00038798267943241407,
      "loss": 1.1634,
      "step": 33248
    },
    {
      "epoch": 2.8606211821388627,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00038795632620425904,
      "loss": 0.6084,
      "step": 33249
    },
    {
      "epoch": 2.8607072184461844,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00038792997330380985,
      "loss": 1.6107,
      "step": 33250
    },
    {
      "epoch": 2.860793254753506,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003879036207311438,
      "loss": 1.6217,
      "step": 33251
    },
    {
      "epoch": 2.8608792910608276,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003878772684863379,
      "loss": 1.3844,
      "step": 33252
    },
    {
      "epoch": 2.8609653273681492,
      "grad_norm": 1.15625,
      "learning_rate": 0.00038785091656946895,
      "loss": 1.819,
      "step": 33253
    },
    {
      "epoch": 2.861051363675471,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003878245649806145,
      "loss": 0.9965,
      "step": 33254
    },
    {
      "epoch": 2.861137399982793,
      "grad_norm": 1.375,
      "learning_rate": 0.0003877982137198513,
      "loss": 1.1076,
      "step": 33255
    },
    {
      "epoch": 2.8612234362901146,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003877718627872565,
      "loss": 0.5647,
      "step": 33256
    },
    {
      "epoch": 2.861309472597436,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003877455121829071,
      "loss": 0.9923,
      "step": 33257
    },
    {
      "epoch": 2.861395508904758,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003877191619068804,
      "loss": 2.2296,
      "step": 33258
    },
    {
      "epoch": 2.8614815452120794,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003876928119592532,
      "loss": 1.7695,
      "step": 33259
    },
    {
      "epoch": 2.861567581519401,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003876664623401028,
      "loss": 1.3408,
      "step": 33260
    },
    {
      "epoch": 2.8616536178267227,
      "grad_norm": 1.1875,
      "learning_rate": 0.000387640113049506,
      "loss": 1.4626,
      "step": 33261
    },
    {
      "epoch": 2.8617396541340447,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003876137640875401,
      "loss": 1.9817,
      "step": 33262
    },
    {
      "epoch": 2.8618256904413664,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003875874154542821,
      "loss": 0.807,
      "step": 33263
    },
    {
      "epoch": 2.861911726748688,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00038756106714980896,
      "loss": 1.1512,
      "step": 33264
    },
    {
      "epoch": 2.8619977630560096,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00038753471917419777,
      "loss": 1.3905,
      "step": 33265
    },
    {
      "epoch": 2.8620837993633312,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00038750837152752565,
      "loss": 0.5249,
      "step": 33266
    },
    {
      "epoch": 2.862169835670653,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003874820242098698,
      "loss": 1.0722,
      "step": 33267
    },
    {
      "epoch": 2.8622558719779745,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003874556772213072,
      "loss": 1.4832,
      "step": 33268
    },
    {
      "epoch": 2.8623419082852966,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003874293305619147,
      "loss": 1.299,
      "step": 33269
    },
    {
      "epoch": 2.862427944592618,
      "grad_norm": 1.53125,
      "learning_rate": 0.0003874029842317696,
      "loss": 1.131,
      "step": 33270
    },
    {
      "epoch": 2.86251398089994,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003873766382309489,
      "loss": 1.6728,
      "step": 33271
    },
    {
      "epoch": 2.8626000172072614,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003873502925595296,
      "loss": 0.9667,
      "step": 33272
    },
    {
      "epoch": 2.862686053514583,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003873239472175888,
      "loss": 0.6832,
      "step": 33273
    },
    {
      "epoch": 2.862772089821905,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003872976022052035,
      "loss": 0.8239,
      "step": 33274
    },
    {
      "epoch": 2.8628581261292263,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038727125752245086,
      "loss": 1.4329,
      "step": 33275
    },
    {
      "epoch": 2.8629441624365484,
      "grad_norm": 0.65625,
      "learning_rate": 0.00038724491316940773,
      "loss": 1.1663,
      "step": 33276
    },
    {
      "epoch": 2.86303019874387,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003872185691461514,
      "loss": 0.753,
      "step": 33277
    },
    {
      "epoch": 2.8631162350511916,
      "grad_norm": 0.765625,
      "learning_rate": 0.0003871922254527589,
      "loss": 1.031,
      "step": 33278
    },
    {
      "epoch": 2.8632022713585132,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00038716588208930717,
      "loss": 0.8916,
      "step": 33279
    },
    {
      "epoch": 2.863288307665835,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003871395390558731,
      "loss": 1.1305,
      "step": 33280
    },
    {
      "epoch": 2.863374343973157,
      "grad_norm": 1.046875,
      "learning_rate": 0.00038711319635253413,
      "loss": 1.9787,
      "step": 33281
    },
    {
      "epoch": 2.863460380280478,
      "grad_norm": 0.921875,
      "learning_rate": 0.00038708685397936707,
      "loss": 1.0123,
      "step": 33282
    },
    {
      "epoch": 2.8635464165878,
      "grad_norm": 1.0625,
      "learning_rate": 0.00038706051193644896,
      "loss": 0.7633,
      "step": 33283
    },
    {
      "epoch": 2.863632452895122,
      "grad_norm": 1.25,
      "learning_rate": 0.00038703417022385684,
      "loss": 1.2861,
      "step": 33284
    },
    {
      "epoch": 2.8637184892024434,
      "grad_norm": 1.109375,
      "learning_rate": 0.00038700782884166786,
      "loss": 1.5807,
      "step": 33285
    },
    {
      "epoch": 2.863804525509765,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003869814877899591,
      "loss": 0.9888,
      "step": 33286
    },
    {
      "epoch": 2.8638905618170867,
      "grad_norm": 1.125,
      "learning_rate": 0.0003869551470688073,
      "loss": 1.2385,
      "step": 33287
    },
    {
      "epoch": 2.8639765981244087,
      "grad_norm": 0.96875,
      "learning_rate": 0.00038692880667828985,
      "loss": 1.5417,
      "step": 33288
    },
    {
      "epoch": 2.86406263443173,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003869024666184836,
      "loss": 1.1574,
      "step": 33289
    },
    {
      "epoch": 2.864148670739052,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003868761268894657,
      "loss": 0.6869,
      "step": 33290
    },
    {
      "epoch": 2.8642347070463736,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003868497874913131,
      "loss": 0.7487,
      "step": 33291
    },
    {
      "epoch": 2.8643207433536952,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00038682344842410276,
      "loss": 1.093,
      "step": 33292
    },
    {
      "epoch": 2.864406779661017,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000386797109687912,
      "loss": 0.5114,
      "step": 33293
    },
    {
      "epoch": 2.8644928159683385,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003867707712828175,
      "loss": 1.1167,
      "step": 33294
    },
    {
      "epoch": 2.8645788522756606,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00038674443320889655,
      "loss": 1.0179,
      "step": 33295
    },
    {
      "epoch": 2.864664888582982,
      "grad_norm": 1.328125,
      "learning_rate": 0.00038671809546622616,
      "loss": 1.3684,
      "step": 33296
    },
    {
      "epoch": 2.864750924890304,
      "grad_norm": 1.34375,
      "learning_rate": 0.00038669175805488334,
      "loss": 1.6402,
      "step": 33297
    },
    {
      "epoch": 2.8648369611976254,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00038666542097494497,
      "loss": 1.1279,
      "step": 33298
    },
    {
      "epoch": 2.864922997504947,
      "grad_norm": 1.03125,
      "learning_rate": 0.00038663908422648823,
      "loss": 1.2516,
      "step": 33299
    },
    {
      "epoch": 2.8650090338122687,
      "grad_norm": 1.25,
      "learning_rate": 0.0003866127478095902,
      "loss": 0.7032,
      "step": 33300
    },
    {
      "epoch": 2.8650950701195903,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00038658641172432766,
      "loss": 1.2789,
      "step": 33301
    },
    {
      "epoch": 2.8651811064269124,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003865600759707779,
      "loss": 1.1672,
      "step": 33302
    },
    {
      "epoch": 2.865267142734234,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003865337405490177,
      "loss": 1.168,
      "step": 33303
    },
    {
      "epoch": 2.8653531790415556,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0003865074054591245,
      "loss": 0.9194,
      "step": 33304
    },
    {
      "epoch": 2.8654392153488772,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000386481070701175,
      "loss": 1.1055,
      "step": 33305
    },
    {
      "epoch": 2.865525251656199,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003864547362752463,
      "loss": 1.703,
      "step": 33306
    },
    {
      "epoch": 2.8656112879635205,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003864284021814153,
      "loss": 1.5449,
      "step": 33307
    },
    {
      "epoch": 2.865697324270842,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003864020684197592,
      "loss": 1.8253,
      "step": 33308
    },
    {
      "epoch": 2.865783360578164,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00038637573499035504,
      "loss": 1.4311,
      "step": 33309
    },
    {
      "epoch": 2.865869396885486,
      "grad_norm": 1.125,
      "learning_rate": 0.0003863494018932796,
      "loss": 1.8098,
      "step": 33310
    },
    {
      "epoch": 2.8659554331928074,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00038632306912861017,
      "loss": 1.2522,
      "step": 33311
    },
    {
      "epoch": 2.866041469500129,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003862967366964237,
      "loss": 1.1734,
      "step": 33312
    },
    {
      "epoch": 2.8661275058074507,
      "grad_norm": 1.515625,
      "learning_rate": 0.00038627040459679706,
      "loss": 1.5299,
      "step": 33313
    },
    {
      "epoch": 2.8662135421147723,
      "grad_norm": 1.015625,
      "learning_rate": 0.00038624407282980734,
      "loss": 1.2942,
      "step": 33314
    },
    {
      "epoch": 2.866299578422094,
      "grad_norm": 0.9375,
      "learning_rate": 0.00038621774139553167,
      "loss": 1.6128,
      "step": 33315
    },
    {
      "epoch": 2.866385614729416,
      "grad_norm": 1.015625,
      "learning_rate": 0.000386191410294047,
      "loss": 2.2981,
      "step": 33316
    },
    {
      "epoch": 2.8664716510367376,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003861650795254302,
      "loss": 0.7566,
      "step": 33317
    },
    {
      "epoch": 2.8665576873440592,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003861387490897585,
      "loss": 0.6474,
      "step": 33318
    },
    {
      "epoch": 2.866643723651381,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003861124189871088,
      "loss": 1.4318,
      "step": 33319
    },
    {
      "epoch": 2.8667297599587025,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00038608608921755815,
      "loss": 1.2774,
      "step": 33320
    },
    {
      "epoch": 2.866815796266024,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003860597597811834,
      "loss": 1.8323,
      "step": 33321
    },
    {
      "epoch": 2.8669018325733457,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003860334306780618,
      "loss": 1.1161,
      "step": 33322
    },
    {
      "epoch": 2.866987868880668,
      "grad_norm": 0.984375,
      "learning_rate": 0.00038600710190827024,
      "loss": 0.8554,
      "step": 33323
    },
    {
      "epoch": 2.8670739051879894,
      "grad_norm": 1.21875,
      "learning_rate": 0.00038598077347188573,
      "loss": 1.9328,
      "step": 33324
    },
    {
      "epoch": 2.867159941495311,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003859544453689853,
      "loss": 0.8101,
      "step": 33325
    },
    {
      "epoch": 2.8672459778026327,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003859281175996459,
      "loss": 0.603,
      "step": 33326
    },
    {
      "epoch": 2.8673320141099543,
      "grad_norm": 1.21875,
      "learning_rate": 0.00038590179016394465,
      "loss": 1.209,
      "step": 33327
    },
    {
      "epoch": 2.867418050417276,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003858754630619583,
      "loss": 0.9384,
      "step": 33328
    },
    {
      "epoch": 2.8675040867245976,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00038584913629376416,
      "loss": 1.1327,
      "step": 33329
    },
    {
      "epoch": 2.8675901230319196,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00038582280985943913,
      "loss": 0.9285,
      "step": 33330
    },
    {
      "epoch": 2.8676761593392412,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003857964837590602,
      "loss": 0.6043,
      "step": 33331
    },
    {
      "epoch": 2.867762195646563,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00038577015799270417,
      "loss": 0.9266,
      "step": 33332
    },
    {
      "epoch": 2.8678482319538845,
      "grad_norm": 1.25,
      "learning_rate": 0.00038574383256044834,
      "loss": 1.7671,
      "step": 33333
    },
    {
      "epoch": 2.867934268261206,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00038571750746236954,
      "loss": 1.2204,
      "step": 33334
    },
    {
      "epoch": 2.868020304568528,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00038569118269854473,
      "loss": 0.8246,
      "step": 33335
    },
    {
      "epoch": 2.8681063408758494,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038566485826905107,
      "loss": 1.1495,
      "step": 33336
    },
    {
      "epoch": 2.8681923771831714,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003856385341739655,
      "loss": 1.596,
      "step": 33337
    },
    {
      "epoch": 2.868278413490493,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003856122104133649,
      "loss": 1.2775,
      "step": 33338
    },
    {
      "epoch": 2.8683644497978147,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00038558588698732623,
      "loss": 1.2228,
      "step": 33339
    },
    {
      "epoch": 2.8684504861051363,
      "grad_norm": 0.765625,
      "learning_rate": 0.0003855595638959266,
      "loss": 1.0327,
      "step": 33340
    },
    {
      "epoch": 2.868536522412458,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00038553324113924315,
      "loss": 0.6807,
      "step": 33341
    },
    {
      "epoch": 2.86862255871978,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003855069187173527,
      "loss": 0.7634,
      "step": 33342
    },
    {
      "epoch": 2.868708595027101,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00038548059663033227,
      "loss": 0.5836,
      "step": 33343
    },
    {
      "epoch": 2.8687946313344233,
      "grad_norm": 1.484375,
      "learning_rate": 0.0003854542748782587,
      "loss": 0.9409,
      "step": 33344
    },
    {
      "epoch": 2.868880667641745,
      "grad_norm": 1.265625,
      "learning_rate": 0.00038542795346120916,
      "loss": 1.644,
      "step": 33345
    },
    {
      "epoch": 2.8689667039490665,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003854016323792606,
      "loss": 1.1553,
      "step": 33346
    },
    {
      "epoch": 2.869052740256388,
      "grad_norm": 1.09375,
      "learning_rate": 0.00038537531163248997,
      "loss": 1.6607,
      "step": 33347
    },
    {
      "epoch": 2.8691387765637097,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00038534899122097415,
      "loss": 0.7498,
      "step": 33348
    },
    {
      "epoch": 2.869224812871032,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00038532267114479036,
      "loss": 0.5046,
      "step": 33349
    },
    {
      "epoch": 2.869310849178353,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003852963514040154,
      "loss": 0.903,
      "step": 33350
    },
    {
      "epoch": 2.869396885485675,
      "grad_norm": 1.234375,
      "learning_rate": 0.00038527003199872626,
      "loss": 1.4655,
      "step": 33351
    },
    {
      "epoch": 2.8694829217929967,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003852437129290001,
      "loss": 0.8186,
      "step": 33352
    },
    {
      "epoch": 2.8695689581003183,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00038521739419491367,
      "loss": 1.2726,
      "step": 33353
    },
    {
      "epoch": 2.86965499440764,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003851910757965441,
      "loss": 0.8058,
      "step": 33354
    },
    {
      "epoch": 2.8697410307149616,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003851647577339682,
      "loss": 0.8896,
      "step": 33355
    },
    {
      "epoch": 2.8698270670222836,
      "grad_norm": 1.125,
      "learning_rate": 0.00038513844000726314,
      "loss": 0.9421,
      "step": 33356
    },
    {
      "epoch": 2.8699131033296053,
      "grad_norm": 1.40625,
      "learning_rate": 0.00038511212261650577,
      "loss": 1.7199,
      "step": 33357
    },
    {
      "epoch": 2.869999139636927,
      "grad_norm": 1.9375,
      "learning_rate": 0.000385085805561773,
      "loss": 1.3983,
      "step": 33358
    },
    {
      "epoch": 2.8700851759442485,
      "grad_norm": 1.046875,
      "learning_rate": 0.000385059488843142,
      "loss": 0.9826,
      "step": 33359
    },
    {
      "epoch": 2.87017121225157,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003850331724606897,
      "loss": 1.3423,
      "step": 33360
    },
    {
      "epoch": 2.8702572485588917,
      "grad_norm": 0.671875,
      "learning_rate": 0.00038500685641449294,
      "loss": 0.7065,
      "step": 33361
    },
    {
      "epoch": 2.8703432848662134,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00038498054070462874,
      "loss": 0.9128,
      "step": 33362
    },
    {
      "epoch": 2.8704293211735354,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00038495422533117414,
      "loss": 0.764,
      "step": 33363
    },
    {
      "epoch": 2.870515357480857,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00038492791029420603,
      "loss": 0.7893,
      "step": 33364
    },
    {
      "epoch": 2.8706013937881787,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003849015955938013,
      "loss": 0.7443,
      "step": 33365
    },
    {
      "epoch": 2.8706874300955003,
      "grad_norm": 0.71875,
      "learning_rate": 0.00038487528123003716,
      "loss": 0.8767,
      "step": 33366
    },
    {
      "epoch": 2.870773466402822,
      "grad_norm": 1.078125,
      "learning_rate": 0.00038484896720299047,
      "loss": 1.0391,
      "step": 33367
    },
    {
      "epoch": 2.8708595027101436,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003848226535127381,
      "loss": 0.9364,
      "step": 33368
    },
    {
      "epoch": 2.870945539017465,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003847963401593569,
      "loss": 1.0696,
      "step": 33369
    },
    {
      "epoch": 2.8710315753247873,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003847700271429242,
      "loss": 1.0616,
      "step": 33370
    },
    {
      "epoch": 2.871117611632109,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0003847437144635167,
      "loss": 1.4907,
      "step": 33371
    },
    {
      "epoch": 2.8712036479394305,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003847174021212115,
      "loss": 0.7996,
      "step": 33372
    },
    {
      "epoch": 2.871289684246752,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003846910901160853,
      "loss": 1.667,
      "step": 33373
    },
    {
      "epoch": 2.8713757205540738,
      "grad_norm": 1.578125,
      "learning_rate": 0.0003846647784482154,
      "loss": 1.1107,
      "step": 33374
    },
    {
      "epoch": 2.8714617568613954,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00038463846711767855,
      "loss": 0.6827,
      "step": 33375
    },
    {
      "epoch": 2.871547793168717,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00038461215612455154,
      "loss": 1.3083,
      "step": 33376
    },
    {
      "epoch": 2.871633829476039,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00038458584546891174,
      "loss": 2.0138,
      "step": 33377
    },
    {
      "epoch": 2.8717198657833607,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00038455953515083573,
      "loss": 0.9791,
      "step": 33378
    },
    {
      "epoch": 2.8718059020906823,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003845332251704008,
      "loss": 1.184,
      "step": 33379
    },
    {
      "epoch": 2.871891938398004,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003845069155276838,
      "loss": 1.2224,
      "step": 33380
    },
    {
      "epoch": 2.8719779747053256,
      "grad_norm": 0.6875,
      "learning_rate": 0.00038448060622276137,
      "loss": 0.5019,
      "step": 33381
    },
    {
      "epoch": 2.872064011012647,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003844542972557109,
      "loss": 0.8629,
      "step": 33382
    },
    {
      "epoch": 2.872150047319969,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003844279886266091,
      "loss": 1.4865,
      "step": 33383
    },
    {
      "epoch": 2.872236083627291,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00038440168033553295,
      "loss": 0.4989,
      "step": 33384
    },
    {
      "epoch": 2.8723221199346125,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0003843753723825593,
      "loss": 0.931,
      "step": 33385
    },
    {
      "epoch": 2.872408156241934,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00038434906476776535,
      "loss": 1.2047,
      "step": 33386
    },
    {
      "epoch": 2.8724941925492558,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003843227574912278,
      "loss": 0.9174,
      "step": 33387
    },
    {
      "epoch": 2.8725802288565774,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00038429645055302366,
      "loss": 1.4536,
      "step": 33388
    },
    {
      "epoch": 2.872666265163899,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00038427014395323,
      "loss": 0.9243,
      "step": 33389
    },
    {
      "epoch": 2.8727523014712206,
      "grad_norm": 1.25,
      "learning_rate": 0.0003842438376919236,
      "loss": 1.5373,
      "step": 33390
    },
    {
      "epoch": 2.8728383377785427,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003842175317691815,
      "loss": 1.8047,
      "step": 33391
    },
    {
      "epoch": 2.8729243740858643,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003841912261850805,
      "loss": 1.406,
      "step": 33392
    },
    {
      "epoch": 2.873010410393186,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00038416492093969776,
      "loss": 1.2186,
      "step": 33393
    },
    {
      "epoch": 2.8730964467005076,
      "grad_norm": 1.5,
      "learning_rate": 0.00038413861603311006,
      "loss": 0.8621,
      "step": 33394
    },
    {
      "epoch": 2.873182483007829,
      "grad_norm": 1.1875,
      "learning_rate": 0.00038411231146539437,
      "loss": 1.7656,
      "step": 33395
    },
    {
      "epoch": 2.8732685193151513,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003840860072366276,
      "loss": 1.2052,
      "step": 33396
    },
    {
      "epoch": 2.8733545556224724,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00038405970334688677,
      "loss": 0.8393,
      "step": 33397
    },
    {
      "epoch": 2.8734405919297945,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003840333997962487,
      "loss": 1.1651,
      "step": 33398
    },
    {
      "epoch": 2.873526628237116,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00038400709658479036,
      "loss": 1.029,
      "step": 33399
    },
    {
      "epoch": 2.8736126645444378,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003839807937125888,
      "loss": 1.5322,
      "step": 33400
    },
    {
      "epoch": 2.8736987008517594,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003839544911797208,
      "loss": 0.953,
      "step": 33401
    },
    {
      "epoch": 2.873784737159081,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003839281889862634,
      "loss": 1.1946,
      "step": 33402
    },
    {
      "epoch": 2.873870773466403,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003839018871322933,
      "loss": 0.6854,
      "step": 33403
    },
    {
      "epoch": 2.8739568097737243,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00038387558561788783,
      "loss": 0.9671,
      "step": 33404
    },
    {
      "epoch": 2.8740428460810463,
      "grad_norm": 1.21875,
      "learning_rate": 0.00038384928444312353,
      "loss": 1.473,
      "step": 33405
    },
    {
      "epoch": 2.874128882388368,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0003838229836080775,
      "loss": 1.0067,
      "step": 33406
    },
    {
      "epoch": 2.8742149186956896,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00038379668311282666,
      "loss": 0.812,
      "step": 33407
    },
    {
      "epoch": 2.874300955003011,
      "grad_norm": 1.015625,
      "learning_rate": 0.00038377038295744805,
      "loss": 0.7542,
      "step": 33408
    },
    {
      "epoch": 2.874386991310333,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003837440831420184,
      "loss": 0.9047,
      "step": 33409
    },
    {
      "epoch": 2.874473027617655,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003837177836666145,
      "loss": 0.9405,
      "step": 33410
    },
    {
      "epoch": 2.874559063924976,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00038369148453131364,
      "loss": 0.8716,
      "step": 33411
    },
    {
      "epoch": 2.874645100232298,
      "grad_norm": 1.265625,
      "learning_rate": 0.00038366518573619237,
      "loss": 1.5167,
      "step": 33412
    },
    {
      "epoch": 2.8747311365396198,
      "grad_norm": 1.0,
      "learning_rate": 0.000383638887281328,
      "loss": 1.0685,
      "step": 33413
    },
    {
      "epoch": 2.8748171728469414,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0003836125891667973,
      "loss": 0.7081,
      "step": 33414
    },
    {
      "epoch": 2.874903209154263,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00038358629139267697,
      "loss": 1.0256,
      "step": 33415
    },
    {
      "epoch": 2.8749892454615846,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003835599939590443,
      "loss": 1.5481,
      "step": 33416
    },
    {
      "epoch": 2.8750752817689067,
      "grad_norm": 1.140625,
      "learning_rate": 0.00038353369686597587,
      "loss": 0.9625,
      "step": 33417
    },
    {
      "epoch": 2.8751613180762283,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003835074001135488,
      "loss": 1.5836,
      "step": 33418
    },
    {
      "epoch": 2.87524735438355,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003834811037018398,
      "loss": 1.866,
      "step": 33419
    },
    {
      "epoch": 2.8753333906908716,
      "grad_norm": 1.15625,
      "learning_rate": 0.00038345480763092604,
      "loss": 1.3281,
      "step": 33420
    },
    {
      "epoch": 2.875419426998193,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003834285119008843,
      "loss": 1.5417,
      "step": 33421
    },
    {
      "epoch": 2.875505463305515,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003834022165117914,
      "loss": 0.6378,
      "step": 33422
    },
    {
      "epoch": 2.8755914996128364,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00038337592146372437,
      "loss": 0.5724,
      "step": 33423
    },
    {
      "epoch": 2.8756775359201585,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00038334962675676016,
      "loss": 1.6033,
      "step": 33424
    },
    {
      "epoch": 2.87576357222748,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00038332333239097556,
      "loss": 0.8187,
      "step": 33425
    },
    {
      "epoch": 2.8758496085348018,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003832970383664475,
      "loss": 1.1844,
      "step": 33426
    },
    {
      "epoch": 2.8759356448421234,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003832707446832529,
      "loss": 1.7784,
      "step": 33427
    },
    {
      "epoch": 2.876021681149445,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003832444513414687,
      "loss": 0.6822,
      "step": 33428
    },
    {
      "epoch": 2.8761077174567666,
      "grad_norm": 1.09375,
      "learning_rate": 0.00038321815834117167,
      "loss": 0.798,
      "step": 33429
    },
    {
      "epoch": 2.8761937537640883,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003831918656824389,
      "loss": 1.2998,
      "step": 33430
    },
    {
      "epoch": 2.8762797900714103,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003831655733653473,
      "loss": 0.4084,
      "step": 33431
    },
    {
      "epoch": 2.876365826378732,
      "grad_norm": 1.125,
      "learning_rate": 0.00038313928138997356,
      "loss": 1.5901,
      "step": 33432
    },
    {
      "epoch": 2.8764518626860536,
      "grad_norm": 0.859375,
      "learning_rate": 0.00038311298975639464,
      "loss": 1.2467,
      "step": 33433
    },
    {
      "epoch": 2.876537898993375,
      "grad_norm": 0.953125,
      "learning_rate": 0.00038308669846468766,
      "loss": 0.9878,
      "step": 33434
    },
    {
      "epoch": 2.876623935300697,
      "grad_norm": 1.171875,
      "learning_rate": 0.00038306040751492925,
      "loss": 1.3464,
      "step": 33435
    },
    {
      "epoch": 2.8767099716080184,
      "grad_norm": 1.046875,
      "learning_rate": 0.00038303411690719647,
      "loss": 0.89,
      "step": 33436
    },
    {
      "epoch": 2.87679600791534,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038300782664156595,
      "loss": 0.9263,
      "step": 33437
    },
    {
      "epoch": 2.876882044222662,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00038298153671811495,
      "loss": 1.6219,
      "step": 33438
    },
    {
      "epoch": 2.8769680805299838,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00038295524713692023,
      "loss": 1.8028,
      "step": 33439
    },
    {
      "epoch": 2.8770541168373054,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0003829289578980585,
      "loss": 0.881,
      "step": 33440
    },
    {
      "epoch": 2.877140153144627,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0003829026690016069,
      "loss": 0.5809,
      "step": 33441
    },
    {
      "epoch": 2.8772261894519486,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003828763804476423,
      "loss": 1.9047,
      "step": 33442
    },
    {
      "epoch": 2.8773122257592703,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0003828500922362414,
      "loss": 0.9057,
      "step": 33443
    },
    {
      "epoch": 2.877398262066592,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00038282380436748117,
      "loss": 0.4719,
      "step": 33444
    },
    {
      "epoch": 2.877484298373914,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00038279751684143856,
      "loss": 0.983,
      "step": 33445
    },
    {
      "epoch": 2.8775703346812356,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003827712296581904,
      "loss": 2.0043,
      "step": 33446
    },
    {
      "epoch": 2.877656370988557,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003827449428178136,
      "loss": 0.9679,
      "step": 33447
    },
    {
      "epoch": 2.877742407295879,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003827186563203851,
      "loss": 1.2878,
      "step": 33448
    },
    {
      "epoch": 2.8778284436032004,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003826923701659815,
      "loss": 1.8058,
      "step": 33449
    },
    {
      "epoch": 2.8779144799105225,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00038266608435468007,
      "loss": 0.8875,
      "step": 33450
    },
    {
      "epoch": 2.8780005162178437,
      "grad_norm": 0.90625,
      "learning_rate": 0.00038263979888655754,
      "loss": 1.7054,
      "step": 33451
    },
    {
      "epoch": 2.8780865525251658,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003826135137616907,
      "loss": 1.9248,
      "step": 33452
    },
    {
      "epoch": 2.8781725888324874,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003825872289801566,
      "loss": 0.9881,
      "step": 33453
    },
    {
      "epoch": 2.878258625139809,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000382560944542032,
      "loss": 1.2187,
      "step": 33454
    },
    {
      "epoch": 2.8783446614471306,
      "grad_norm": 1.015625,
      "learning_rate": 0.00038253466044739384,
      "loss": 0.8002,
      "step": 33455
    },
    {
      "epoch": 2.8784306977544523,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00038250837669631877,
      "loss": 1.171,
      "step": 33456
    },
    {
      "epoch": 2.8785167340617743,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000382482093288884,
      "loss": 0.6452,
      "step": 33457
    },
    {
      "epoch": 2.8786027703690955,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003824558102251662,
      "loss": 1.0469,
      "step": 33458
    },
    {
      "epoch": 2.8786888066764176,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003824295275052424,
      "loss": 1.4616,
      "step": 33459
    },
    {
      "epoch": 2.878774842983739,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0003824032451291892,
      "loss": 0.5345,
      "step": 33460
    },
    {
      "epoch": 2.878860879291061,
      "grad_norm": 1.359375,
      "learning_rate": 0.00038237696309708377,
      "loss": 2.133,
      "step": 33461
    },
    {
      "epoch": 2.8789469155983824,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003823506814090028,
      "loss": 1.1166,
      "step": 33462
    },
    {
      "epoch": 2.879032951905704,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003823244000650231,
      "loss": 1.5285,
      "step": 33463
    },
    {
      "epoch": 2.879118988213026,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038229811906522176,
      "loss": 0.8057,
      "step": 33464
    },
    {
      "epoch": 2.8792050245203473,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00038227183840967556,
      "loss": 1.5602,
      "step": 33465
    },
    {
      "epoch": 2.8792910608276694,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003822455580984613,
      "loss": 0.4532,
      "step": 33466
    },
    {
      "epoch": 2.879377097134991,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003822192781316558,
      "loss": 0.7134,
      "step": 33467
    },
    {
      "epoch": 2.8794631334423126,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003821929985093361,
      "loss": 1.9177,
      "step": 33468
    },
    {
      "epoch": 2.8795491697496343,
      "grad_norm": 1.015625,
      "learning_rate": 0.000382166719231579,
      "loss": 1.0953,
      "step": 33469
    },
    {
      "epoch": 2.879635206056956,
      "grad_norm": 1.0625,
      "learning_rate": 0.00038214044029846115,
      "loss": 0.8602,
      "step": 33470
    },
    {
      "epoch": 2.879721242364278,
      "grad_norm": 1.375,
      "learning_rate": 0.00038211416171005975,
      "loss": 1.236,
      "step": 33471
    },
    {
      "epoch": 2.8798072786715996,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003820878834664515,
      "loss": 1.2827,
      "step": 33472
    },
    {
      "epoch": 2.879893314978921,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00038206160556771327,
      "loss": 0.7193,
      "step": 33473
    },
    {
      "epoch": 2.879979351286243,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003820353280139217,
      "loss": 1.1189,
      "step": 33474
    },
    {
      "epoch": 2.8800653875935645,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003820090508051541,
      "loss": 1.7078,
      "step": 33475
    },
    {
      "epoch": 2.880151423900886,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00038198277394148696,
      "loss": 1.0304,
      "step": 33476
    },
    {
      "epoch": 2.8802374602082077,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003819564974229972,
      "loss": 1.4317,
      "step": 33477
    },
    {
      "epoch": 2.8803234965155298,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00038193022124976184,
      "loss": 0.9195,
      "step": 33478
    },
    {
      "epoch": 2.8804095328228514,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003819039454218576,
      "loss": 1.0215,
      "step": 33479
    },
    {
      "epoch": 2.880495569130173,
      "grad_norm": 1.15625,
      "learning_rate": 0.00038187766993936134,
      "loss": 1.2322,
      "step": 33480
    },
    {
      "epoch": 2.8805816054374946,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003818513948023499,
      "loss": 1.5229,
      "step": 33481
    },
    {
      "epoch": 2.8806676417448163,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0003818251200109002,
      "loss": 0.7395,
      "step": 33482
    },
    {
      "epoch": 2.880753678052138,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00038179884556508906,
      "loss": 2.1386,
      "step": 33483
    },
    {
      "epoch": 2.8808397143594595,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003817725714649932,
      "loss": 1.4859,
      "step": 33484
    },
    {
      "epoch": 2.8809257506667816,
      "grad_norm": 1.390625,
      "learning_rate": 0.00038174629771068957,
      "loss": 1.1377,
      "step": 33485
    },
    {
      "epoch": 2.881011786974103,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003817200243022549,
      "loss": 1.6108,
      "step": 33486
    },
    {
      "epoch": 2.881097823281425,
      "grad_norm": 2.265625,
      "learning_rate": 0.00038169375123976636,
      "loss": 1.6367,
      "step": 33487
    },
    {
      "epoch": 2.8811838595887465,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00038166747852330066,
      "loss": 1.2922,
      "step": 33488
    },
    {
      "epoch": 2.881269895896068,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00038164120615293446,
      "loss": 1.7586,
      "step": 33489
    },
    {
      "epoch": 2.8813559322033897,
      "grad_norm": 1.203125,
      "learning_rate": 0.00038161493412874463,
      "loss": 0.6547,
      "step": 33490
    },
    {
      "epoch": 2.8814419685107113,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00038158866245080827,
      "loss": 1.3931,
      "step": 33491
    },
    {
      "epoch": 2.8815280048180334,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003815623911192019,
      "loss": 0.7123,
      "step": 33492
    },
    {
      "epoch": 2.881614041125355,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003815361201340025,
      "loss": 1.4772,
      "step": 33493
    },
    {
      "epoch": 2.8817000774326766,
      "grad_norm": 1.109375,
      "learning_rate": 0.000381509849495287,
      "loss": 1.1654,
      "step": 33494
    },
    {
      "epoch": 2.8817861137399983,
      "grad_norm": 1.03125,
      "learning_rate": 0.00038148357920313213,
      "loss": 0.923,
      "step": 33495
    },
    {
      "epoch": 2.88187215004732,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003814573092576148,
      "loss": 0.8519,
      "step": 33496
    },
    {
      "epoch": 2.8819581863546415,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00038143103965881166,
      "loss": 1.0324,
      "step": 33497
    },
    {
      "epoch": 2.882044222661963,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003814047704067997,
      "loss": 1.3908,
      "step": 33498
    },
    {
      "epoch": 2.882130258969285,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00038137850150165583,
      "loss": 1.1594,
      "step": 33499
    },
    {
      "epoch": 2.882216295276607,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003813522329434566,
      "loss": 1.3306,
      "step": 33500
    },
    {
      "epoch": 2.8823023315839285,
      "grad_norm": 1.671875,
      "learning_rate": 0.0003813259647322791,
      "loss": 1.5963,
      "step": 33501
    },
    {
      "epoch": 2.88238836789125,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003812996968682001,
      "loss": 1.1302,
      "step": 33502
    },
    {
      "epoch": 2.8824744041985717,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003812734293512964,
      "loss": 1.1008,
      "step": 33503
    },
    {
      "epoch": 2.8825604405058933,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003812471621816447,
      "loss": 1.3739,
      "step": 33504
    },
    {
      "epoch": 2.882646476813215,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0003812208953593221,
      "loss": 0.8477,
      "step": 33505
    },
    {
      "epoch": 2.882732513120537,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003811946288844053,
      "loss": 1.055,
      "step": 33506
    },
    {
      "epoch": 2.8828185494278586,
      "grad_norm": 1.0,
      "learning_rate": 0.00038116836275697103,
      "loss": 1.2313,
      "step": 33507
    },
    {
      "epoch": 2.8829045857351803,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003811420969770961,
      "loss": 1.0005,
      "step": 33508
    },
    {
      "epoch": 2.882990622042502,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00038111583154485756,
      "loss": 1.5068,
      "step": 33509
    },
    {
      "epoch": 2.8830766583498235,
      "grad_norm": 1.03125,
      "learning_rate": 0.00038108956646033213,
      "loss": 1.5815,
      "step": 33510
    },
    {
      "epoch": 2.8831626946571456,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00038106330172359636,
      "loss": 1.7365,
      "step": 33511
    },
    {
      "epoch": 2.8832487309644668,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003810370373347275,
      "loss": 0.8189,
      "step": 33512
    },
    {
      "epoch": 2.883334767271789,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00038101077329380213,
      "loss": 1.5293,
      "step": 33513
    },
    {
      "epoch": 2.8834208035791105,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00038098450960089714,
      "loss": 1.4019,
      "step": 33514
    },
    {
      "epoch": 2.883506839886432,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003809582462560892,
      "loss": 1.1881,
      "step": 33515
    },
    {
      "epoch": 2.8835928761937537,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00038093198325945534,
      "loss": 1.3146,
      "step": 33516
    },
    {
      "epoch": 2.8836789125010753,
      "grad_norm": 1.4375,
      "learning_rate": 0.0003809057206110722,
      "loss": 2.0181,
      "step": 33517
    },
    {
      "epoch": 2.8837649488083974,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00038087945831101667,
      "loss": 1.549,
      "step": 33518
    },
    {
      "epoch": 2.8838509851157186,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003808531963593656,
      "loss": 1.2292,
      "step": 33519
    },
    {
      "epoch": 2.8839370214230406,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003808269347561958,
      "loss": 0.9332,
      "step": 33520
    },
    {
      "epoch": 2.8840230577303623,
      "grad_norm": 1.09375,
      "learning_rate": 0.00038080067350158397,
      "loss": 1.4649,
      "step": 33521
    },
    {
      "epoch": 2.884109094037684,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003807744125956069,
      "loss": 1.3113,
      "step": 33522
    },
    {
      "epoch": 2.8841951303450055,
      "grad_norm": 1.09375,
      "learning_rate": 0.00038074815203834153,
      "loss": 0.9954,
      "step": 33523
    },
    {
      "epoch": 2.884281166652327,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003807218918298646,
      "loss": 1.5616,
      "step": 33524
    },
    {
      "epoch": 2.884367202959649,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00038069563197025305,
      "loss": 0.8072,
      "step": 33525
    },
    {
      "epoch": 2.8844532392669704,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00038066937245958356,
      "loss": 1.4697,
      "step": 33526
    },
    {
      "epoch": 2.8845392755742925,
      "grad_norm": 1.28125,
      "learning_rate": 0.00038064311329793284,
      "loss": 1.2731,
      "step": 33527
    },
    {
      "epoch": 2.884625311881614,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003806168544853779,
      "loss": 0.6599,
      "step": 33528
    },
    {
      "epoch": 2.8847113481889357,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00038059059602199547,
      "loss": 0.8281,
      "step": 33529
    },
    {
      "epoch": 2.8847973844962573,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003805643379078623,
      "loss": 0.8237,
      "step": 33530
    },
    {
      "epoch": 2.884883420803579,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00038053808014305515,
      "loss": 1.3281,
      "step": 33531
    },
    {
      "epoch": 2.884969457110901,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003805118227276509,
      "loss": 1.1916,
      "step": 33532
    },
    {
      "epoch": 2.8850554934182226,
      "grad_norm": 1.1875,
      "learning_rate": 0.00038048556566172635,
      "loss": 1.7152,
      "step": 33533
    },
    {
      "epoch": 2.8851415297255443,
      "grad_norm": 0.9375,
      "learning_rate": 0.00038045930894535827,
      "loss": 0.7863,
      "step": 33534
    },
    {
      "epoch": 2.885227566032866,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003804330525786235,
      "loss": 0.5896,
      "step": 33535
    },
    {
      "epoch": 2.8853136023401875,
      "grad_norm": 1.15625,
      "learning_rate": 0.00038040679656159876,
      "loss": 1.0926,
      "step": 33536
    },
    {
      "epoch": 2.885399638647509,
      "grad_norm": 1.46875,
      "learning_rate": 0.0003803805408943609,
      "loss": 1.6494,
      "step": 33537
    },
    {
      "epoch": 2.8854856749548308,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003803542855769866,
      "loss": 1.1072,
      "step": 33538
    },
    {
      "epoch": 2.885571711262153,
      "grad_norm": 0.78125,
      "learning_rate": 0.0003803280306095529,
      "loss": 0.7247,
      "step": 33539
    },
    {
      "epoch": 2.8856577475694745,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003803017759921363,
      "loss": 0.6621,
      "step": 33540
    },
    {
      "epoch": 2.885743783876796,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00038027552172481376,
      "loss": 1.3442,
      "step": 33541
    },
    {
      "epoch": 2.8858298201841177,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003802492678076621,
      "loss": 1.4973,
      "step": 33542
    },
    {
      "epoch": 2.8859158564914393,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000380223014240758,
      "loss": 1.0383,
      "step": 33543
    },
    {
      "epoch": 2.886001892798761,
      "grad_norm": 1.625,
      "learning_rate": 0.0003801967610241783,
      "loss": 0.8657,
      "step": 33544
    },
    {
      "epoch": 2.8860879291060826,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003801705081579997,
      "loss": 1.1555,
      "step": 33545
    },
    {
      "epoch": 2.8861739654134047,
      "grad_norm": 1.421875,
      "learning_rate": 0.00038014425564229917,
      "loss": 1.4486,
      "step": 33546
    },
    {
      "epoch": 2.8862600017207263,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003801180034771534,
      "loss": 0.6534,
      "step": 33547
    },
    {
      "epoch": 2.886346038028048,
      "grad_norm": 1.0,
      "learning_rate": 0.0003800917516626391,
      "loss": 0.979,
      "step": 33548
    },
    {
      "epoch": 2.8864320743353695,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000380065500198833,
      "loss": 1.5707,
      "step": 33549
    },
    {
      "epoch": 2.886518110642691,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00038003924908581213,
      "loss": 1.2299,
      "step": 33550
    },
    {
      "epoch": 2.8866041469500128,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00038001299832365307,
      "loss": 1.6569,
      "step": 33551
    },
    {
      "epoch": 2.8866901832573344,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003799867479124326,
      "loss": 0.9599,
      "step": 33552
    },
    {
      "epoch": 2.8867762195646565,
      "grad_norm": 1.625,
      "learning_rate": 0.00037996049785222764,
      "loss": 1.6116,
      "step": 33553
    },
    {
      "epoch": 2.886862255871978,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00037993424814311485,
      "loss": 1.3121,
      "step": 33554
    },
    {
      "epoch": 2.8869482921792997,
      "grad_norm": 1.203125,
      "learning_rate": 0.000379907998785171,
      "loss": 0.9579,
      "step": 33555
    },
    {
      "epoch": 2.8870343284866213,
      "grad_norm": 0.875,
      "learning_rate": 0.0003798817497784729,
      "loss": 0.9942,
      "step": 33556
    },
    {
      "epoch": 2.887120364793943,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00037985550112309735,
      "loss": 1.2831,
      "step": 33557
    },
    {
      "epoch": 2.8872064011012646,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003798292528191211,
      "loss": 0.7596,
      "step": 33558
    },
    {
      "epoch": 2.887292437408586,
      "grad_norm": 1.265625,
      "learning_rate": 0.00037980300486662084,
      "loss": 1.1691,
      "step": 33559
    },
    {
      "epoch": 2.8873784737159083,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003797767572656734,
      "loss": 1.3087,
      "step": 33560
    },
    {
      "epoch": 2.88746451002323,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003797505100163556,
      "loss": 1.2974,
      "step": 33561
    },
    {
      "epoch": 2.8875505463305515,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003797242631187442,
      "loss": 1.4423,
      "step": 33562
    },
    {
      "epoch": 2.887636582637873,
      "grad_norm": 0.8046875,
      "learning_rate": 0.000379698016572916,
      "loss": 1.0302,
      "step": 33563
    },
    {
      "epoch": 2.8877226189451948,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003796717703789476,
      "loss": 0.9425,
      "step": 33564
    },
    {
      "epoch": 2.8878086552525164,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003796455245369159,
      "loss": 1.0144,
      "step": 33565
    },
    {
      "epoch": 2.887894691559838,
      "grad_norm": 1.59375,
      "learning_rate": 0.0003796192790468977,
      "loss": 2.0595,
      "step": 33566
    },
    {
      "epoch": 2.88798072786716,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00037959303390896966,
      "loss": 0.6066,
      "step": 33567
    },
    {
      "epoch": 2.8880667641744817,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00037956678912320853,
      "loss": 1.0641,
      "step": 33568
    },
    {
      "epoch": 2.8881528004818033,
      "grad_norm": 1.484375,
      "learning_rate": 0.0003795405446896911,
      "loss": 1.5011,
      "step": 33569
    },
    {
      "epoch": 2.888238836789125,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00037951430060849427,
      "loss": 1.4949,
      "step": 33570
    },
    {
      "epoch": 2.8883248730964466,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003794880568796947,
      "loss": 1.6321,
      "step": 33571
    },
    {
      "epoch": 2.8884109094037687,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003794618135033689,
      "loss": 0.7364,
      "step": 33572
    },
    {
      "epoch": 2.88849694571109,
      "grad_norm": 1.046875,
      "learning_rate": 0.000379435570479594,
      "loss": 0.9364,
      "step": 33573
    },
    {
      "epoch": 2.888582982018412,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00037940932780844666,
      "loss": 0.6983,
      "step": 33574
    },
    {
      "epoch": 2.8886690183257335,
      "grad_norm": 1.0625,
      "learning_rate": 0.00037938308549000343,
      "loss": 1.2363,
      "step": 33575
    },
    {
      "epoch": 2.888755054633055,
      "grad_norm": 1.5,
      "learning_rate": 0.0003793568435243414,
      "loss": 1.3979,
      "step": 33576
    },
    {
      "epoch": 2.8888410909403768,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000379330601911537,
      "loss": 1.1264,
      "step": 33577
    },
    {
      "epoch": 2.8889271272476984,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003793043606516672,
      "loss": 0.9832,
      "step": 33578
    },
    {
      "epoch": 2.8890131635550205,
      "grad_norm": 1.03125,
      "learning_rate": 0.00037927811974480856,
      "loss": 1.2836,
      "step": 33579
    },
    {
      "epoch": 2.8890991998623416,
      "grad_norm": 1.0,
      "learning_rate": 0.00037925187919103805,
      "loss": 1.1761,
      "step": 33580
    },
    {
      "epoch": 2.8891852361696637,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00037922563899043225,
      "loss": 1.8994,
      "step": 33581
    },
    {
      "epoch": 2.8892712724769853,
      "grad_norm": 1.34375,
      "learning_rate": 0.000379199399143068,
      "loss": 1.4428,
      "step": 33582
    },
    {
      "epoch": 2.889357308784307,
      "grad_norm": 1.0,
      "learning_rate": 0.000379173159649022,
      "loss": 1.3902,
      "step": 33583
    },
    {
      "epoch": 2.8894433450916286,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003791469205083711,
      "loss": 1.4447,
      "step": 33584
    },
    {
      "epoch": 2.88952938139895,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00037912068172119186,
      "loss": 0.7555,
      "step": 33585
    },
    {
      "epoch": 2.8896154177062723,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000379094443287561,
      "loss": 0.8753,
      "step": 33586
    },
    {
      "epoch": 2.8897014540135935,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003790682052075556,
      "loss": 1.7202,
      "step": 33587
    },
    {
      "epoch": 2.8897874903209155,
      "grad_norm": 1.078125,
      "learning_rate": 0.00037904196748125206,
      "loss": 0.7973,
      "step": 33588
    },
    {
      "epoch": 2.889873526628237,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003790157301087273,
      "loss": 1.126,
      "step": 33589
    },
    {
      "epoch": 2.889959562935559,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00037898949309005787,
      "loss": 1.0107,
      "step": 33590
    },
    {
      "epoch": 2.8900455992428804,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00037896325642532077,
      "loss": 1.0413,
      "step": 33591
    },
    {
      "epoch": 2.890131635550202,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003789370201145926,
      "loss": 0.6529,
      "step": 33592
    },
    {
      "epoch": 2.890217671857524,
      "grad_norm": 1.203125,
      "learning_rate": 0.00037891078415795,
      "loss": 1.4194,
      "step": 33593
    },
    {
      "epoch": 2.8903037081648457,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0003788845485554699,
      "loss": 0.9816,
      "step": 33594
    },
    {
      "epoch": 2.8903897444721673,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00037885831330722894,
      "loss": 1.6588,
      "step": 33595
    },
    {
      "epoch": 2.890475780779489,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003788320784133038,
      "loss": 1.659,
      "step": 33596
    },
    {
      "epoch": 2.8905618170868106,
      "grad_norm": 0.96875,
      "learning_rate": 0.00037880584387377117,
      "loss": 1.682,
      "step": 33597
    },
    {
      "epoch": 2.890647853394132,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00037877960968870787,
      "loss": 1.1436,
      "step": 33598
    },
    {
      "epoch": 2.890733889701454,
      "grad_norm": 1.046875,
      "learning_rate": 0.00037875337585819083,
      "loss": 1.2915,
      "step": 33599
    },
    {
      "epoch": 2.890819926008776,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00037872714238229647,
      "loss": 0.9593,
      "step": 33600
    },
    {
      "epoch": 2.8909059623160975,
      "grad_norm": 1.5625,
      "learning_rate": 0.00037870090926110173,
      "loss": 1.8454,
      "step": 33601
    },
    {
      "epoch": 2.890991998623419,
      "grad_norm": 1.203125,
      "learning_rate": 0.00037867467649468315,
      "loss": 1.2926,
      "step": 33602
    },
    {
      "epoch": 2.891078034930741,
      "grad_norm": 1.6875,
      "learning_rate": 0.0003786484440831176,
      "loss": 1.4025,
      "step": 33603
    },
    {
      "epoch": 2.8911640712380624,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003786222120264817,
      "loss": 1.6917,
      "step": 33604
    },
    {
      "epoch": 2.891250107545384,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003785959803248522,
      "loss": 0.8337,
      "step": 33605
    },
    {
      "epoch": 2.8913361438527057,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003785697489783059,
      "loss": 1.6459,
      "step": 33606
    },
    {
      "epoch": 2.8914221801600277,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00037854351798691954,
      "loss": 1.1492,
      "step": 33607
    },
    {
      "epoch": 2.8915082164673493,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00037851728735076973,
      "loss": 0.6865,
      "step": 33608
    },
    {
      "epoch": 2.891594252774671,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003784910570699331,
      "loss": 1.27,
      "step": 33609
    },
    {
      "epoch": 2.8916802890819926,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003784648271444866,
      "loss": 1.1685,
      "step": 33610
    },
    {
      "epoch": 2.891766325389314,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003784385975745069,
      "loss": 1.6844,
      "step": 33611
    },
    {
      "epoch": 2.891852361696636,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003784123683600706,
      "loss": 1.4106,
      "step": 33612
    },
    {
      "epoch": 2.8919383980039575,
      "grad_norm": 1.15625,
      "learning_rate": 0.00037838613950125433,
      "loss": 0.9479,
      "step": 33613
    },
    {
      "epoch": 2.8920244343112795,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00037835991099813515,
      "loss": 2.1601,
      "step": 33614
    },
    {
      "epoch": 2.892110470618601,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003783336828507895,
      "loss": 1.0785,
      "step": 33615
    },
    {
      "epoch": 2.892196506925923,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00037830745505929416,
      "loss": 0.7921,
      "step": 33616
    },
    {
      "epoch": 2.8922825432332444,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00037828122762372586,
      "loss": 1.0345,
      "step": 33617
    },
    {
      "epoch": 2.892368579540566,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00037825500054416126,
      "loss": 1.2459,
      "step": 33618
    },
    {
      "epoch": 2.8924546158478877,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003782287738206772,
      "loss": 1.8243,
      "step": 33619
    },
    {
      "epoch": 2.8925406521552093,
      "grad_norm": 1.0,
      "learning_rate": 0.0003782025474533502,
      "loss": 1.5165,
      "step": 33620
    },
    {
      "epoch": 2.8926266884625313,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0003781763214422571,
      "loss": 0.8584,
      "step": 33621
    },
    {
      "epoch": 2.892712724769853,
      "grad_norm": 1.359375,
      "learning_rate": 0.00037815009578747463,
      "loss": 1.2935,
      "step": 33622
    },
    {
      "epoch": 2.8927987610771746,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00037812387048907927,
      "loss": 0.5456,
      "step": 33623
    },
    {
      "epoch": 2.892884797384496,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000378097645547148,
      "loss": 1.7613,
      "step": 33624
    },
    {
      "epoch": 2.892970833691818,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003780714209617574,
      "loss": 0.8549,
      "step": 33625
    },
    {
      "epoch": 2.89305686999914,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003780451967329842,
      "loss": 1.3077,
      "step": 33626
    },
    {
      "epoch": 2.893142906306461,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00037801897286090494,
      "loss": 1.1129,
      "step": 33627
    },
    {
      "epoch": 2.893228942613783,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003779927493455966,
      "loss": 1.1639,
      "step": 33628
    },
    {
      "epoch": 2.893314978921105,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003779665261871357,
      "loss": 1.293,
      "step": 33629
    },
    {
      "epoch": 2.8934010152284264,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00037794030338559893,
      "loss": 1.3101,
      "step": 33630
    },
    {
      "epoch": 2.893487051535748,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003779140809410631,
      "loss": 1.6921,
      "step": 33631
    },
    {
      "epoch": 2.8935730878430697,
      "grad_norm": 1.0,
      "learning_rate": 0.00037788785885360486,
      "loss": 1.5255,
      "step": 33632
    },
    {
      "epoch": 2.8936591241503917,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003778616371233009,
      "loss": 0.8008,
      "step": 33633
    },
    {
      "epoch": 2.893745160457713,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003778354157502276,
      "loss": 1.0584,
      "step": 33634
    },
    {
      "epoch": 2.893831196765035,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0003778091947344623,
      "loss": 1.0209,
      "step": 33635
    },
    {
      "epoch": 2.8939172330723566,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003777829740760811,
      "loss": 0.8929,
      "step": 33636
    },
    {
      "epoch": 2.8940032693796782,
      "grad_norm": 1.234375,
      "learning_rate": 0.00037775675377516114,
      "loss": 1.4333,
      "step": 33637
    },
    {
      "epoch": 2.894089305687,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003777305338317788,
      "loss": 0.727,
      "step": 33638
    },
    {
      "epoch": 2.8941753419943215,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003777043142460108,
      "loss": 1.111,
      "step": 33639
    },
    {
      "epoch": 2.8942613783016435,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00037767809501793406,
      "loss": 1.8399,
      "step": 33640
    },
    {
      "epoch": 2.8943474146089647,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00037765187614762505,
      "loss": 0.7673,
      "step": 33641
    },
    {
      "epoch": 2.894433450916287,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00037762565763516055,
      "loss": 1.4103,
      "step": 33642
    },
    {
      "epoch": 2.8945194872236084,
      "grad_norm": 1.1875,
      "learning_rate": 0.00037759943948061705,
      "loss": 1.2186,
      "step": 33643
    },
    {
      "epoch": 2.89460552353093,
      "grad_norm": 1.40625,
      "learning_rate": 0.00037757322168407146,
      "loss": 0.8014,
      "step": 33644
    },
    {
      "epoch": 2.8946915598382517,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00037754700424560045,
      "loss": 1.316,
      "step": 33645
    },
    {
      "epoch": 2.8947775961455733,
      "grad_norm": 1.125,
      "learning_rate": 0.0003775207871652805,
      "loss": 1.5127,
      "step": 33646
    },
    {
      "epoch": 2.8948636324528954,
      "grad_norm": 1.375,
      "learning_rate": 0.00037749457044318854,
      "loss": 1.5676,
      "step": 33647
    },
    {
      "epoch": 2.894949668760217,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003774683540794011,
      "loss": 1.8446,
      "step": 33648
    },
    {
      "epoch": 2.8950357050675386,
      "grad_norm": 0.703125,
      "learning_rate": 0.00037744213807399497,
      "loss": 1.4905,
      "step": 33649
    },
    {
      "epoch": 2.8951217413748602,
      "grad_norm": 1.171875,
      "learning_rate": 0.00037741592242704656,
      "loss": 0.9081,
      "step": 33650
    },
    {
      "epoch": 2.895207777682182,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003773897071386329,
      "loss": 0.761,
      "step": 33651
    },
    {
      "epoch": 2.8952938139895035,
      "grad_norm": 1.0,
      "learning_rate": 0.00037736349220883047,
      "loss": 1.2105,
      "step": 33652
    },
    {
      "epoch": 2.895379850296825,
      "grad_norm": 0.859375,
      "learning_rate": 0.00037733727763771593,
      "loss": 0.9921,
      "step": 33653
    },
    {
      "epoch": 2.895465886604147,
      "grad_norm": 1.453125,
      "learning_rate": 0.00037731106342536605,
      "loss": 1.3802,
      "step": 33654
    },
    {
      "epoch": 2.895551922911469,
      "grad_norm": 0.984375,
      "learning_rate": 0.00037728484957185745,
      "loss": 1.2784,
      "step": 33655
    },
    {
      "epoch": 2.8956379592187904,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003772586360772668,
      "loss": 1.7671,
      "step": 33656
    },
    {
      "epoch": 2.895723995526112,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00037723242294167064,
      "loss": 1.4846,
      "step": 33657
    },
    {
      "epoch": 2.8958100318334337,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003772062101651459,
      "loss": 1.8179,
      "step": 33658
    },
    {
      "epoch": 2.8958960681407553,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003771799977477691,
      "loss": 0.9464,
      "step": 33659
    },
    {
      "epoch": 2.895982104448077,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00037715378568961696,
      "loss": 1.1368,
      "step": 33660
    },
    {
      "epoch": 2.896068140755399,
      "grad_norm": 0.8125,
      "learning_rate": 0.00037712757399076594,
      "loss": 1.2207,
      "step": 33661
    },
    {
      "epoch": 2.8961541770627206,
      "grad_norm": 0.8125,
      "learning_rate": 0.000377101362651293,
      "loss": 0.7118,
      "step": 33662
    },
    {
      "epoch": 2.8962402133700422,
      "grad_norm": 1.234375,
      "learning_rate": 0.00037707515167127465,
      "loss": 1.5205,
      "step": 33663
    },
    {
      "epoch": 2.896326249677364,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00037704894105078753,
      "loss": 1.4124,
      "step": 33664
    },
    {
      "epoch": 2.8964122859846855,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003770227307899084,
      "loss": 1.2357,
      "step": 33665
    },
    {
      "epoch": 2.896498322292007,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00037699652088871383,
      "loss": 0.7473,
      "step": 33666
    },
    {
      "epoch": 2.8965843585993287,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00037697031134728055,
      "loss": 1.1223,
      "step": 33667
    },
    {
      "epoch": 2.896670394906651,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003769441021656851,
      "loss": 1.1722,
      "step": 33668
    },
    {
      "epoch": 2.8967564312139724,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00037691789334400434,
      "loss": 1.4426,
      "step": 33669
    },
    {
      "epoch": 2.896842467521294,
      "grad_norm": 1.046875,
      "learning_rate": 0.00037689168488231473,
      "loss": 0.898,
      "step": 33670
    },
    {
      "epoch": 2.8969285038286157,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003768654767806928,
      "loss": 1.7186,
      "step": 33671
    },
    {
      "epoch": 2.8970145401359373,
      "grad_norm": 1.015625,
      "learning_rate": 0.00037683926903921567,
      "loss": 0.9739,
      "step": 33672
    },
    {
      "epoch": 2.897100576443259,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0003768130616579596,
      "loss": 0.6606,
      "step": 33673
    },
    {
      "epoch": 2.8971866127505805,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00037678685463700146,
      "loss": 0.8232,
      "step": 33674
    },
    {
      "epoch": 2.8972726490579026,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0003767606479764178,
      "loss": 1.1739,
      "step": 33675
    },
    {
      "epoch": 2.8973586853652242,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003767344416762852,
      "loss": 0.9133,
      "step": 33676
    },
    {
      "epoch": 2.897444721672546,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00037670823573668046,
      "loss": 1.3181,
      "step": 33677
    },
    {
      "epoch": 2.8975307579798675,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00037668203015768016,
      "loss": 0.9478,
      "step": 33678
    },
    {
      "epoch": 2.897616794287189,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00037665582493936095,
      "loss": 0.937,
      "step": 33679
    },
    {
      "epoch": 2.8977028305945107,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00037662962008179935,
      "loss": 1.6286,
      "step": 33680
    },
    {
      "epoch": 2.8977888669018323,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003766034155850722,
      "loss": 1.3907,
      "step": 33681
    },
    {
      "epoch": 2.8978749032091544,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003765772114492561,
      "loss": 1.0858,
      "step": 33682
    },
    {
      "epoch": 2.897960939516476,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00037655100767442764,
      "loss": 1.1291,
      "step": 33683
    },
    {
      "epoch": 2.8980469758237977,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003765248042606633,
      "loss": 1.5875,
      "step": 33684
    },
    {
      "epoch": 2.8981330121311193,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00037649860120804007,
      "loss": 1.1846,
      "step": 33685
    },
    {
      "epoch": 2.898219048438441,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003764723985166344,
      "loss": 0.903,
      "step": 33686
    },
    {
      "epoch": 2.898305084745763,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00037644619618652287,
      "loss": 1.0238,
      "step": 33687
    },
    {
      "epoch": 2.898391121053084,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00037641999421778223,
      "loss": 0.7729,
      "step": 33688
    },
    {
      "epoch": 2.8984771573604062,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003763937926104891,
      "loss": 1.4523,
      "step": 33689
    },
    {
      "epoch": 2.898563193667728,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003763675913647201,
      "loss": 1.0583,
      "step": 33690
    },
    {
      "epoch": 2.8986492299750495,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00037634139048055174,
      "loss": 1.2307,
      "step": 33691
    },
    {
      "epoch": 2.898735266282371,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003763151899580608,
      "loss": 1.6496,
      "step": 33692
    },
    {
      "epoch": 2.8988213025896927,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000376288989797324,
      "loss": 1.3137,
      "step": 33693
    },
    {
      "epoch": 2.898907338897015,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00037626278999841763,
      "loss": 1.0172,
      "step": 33694
    },
    {
      "epoch": 2.898993375204336,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003762365905614187,
      "loss": 0.712,
      "step": 33695
    },
    {
      "epoch": 2.899079411511658,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003762103914864036,
      "loss": 0.715,
      "step": 33696
    },
    {
      "epoch": 2.8991654478189797,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003761841927734491,
      "loss": 1.2852,
      "step": 33697
    },
    {
      "epoch": 2.8992514841263013,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00037615799442263165,
      "loss": 1.5595,
      "step": 33698
    },
    {
      "epoch": 2.899337520433623,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003761317964340281,
      "loss": 1.6537,
      "step": 33699
    },
    {
      "epoch": 2.8994235567409445,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003761055988077149,
      "loss": 1.2232,
      "step": 33700
    },
    {
      "epoch": 2.8995095930482666,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00037607940154376875,
      "loss": 1.5216,
      "step": 33701
    },
    {
      "epoch": 2.899595629355588,
      "grad_norm": 1.234375,
      "learning_rate": 0.00037605320464226613,
      "loss": 1.2314,
      "step": 33702
    },
    {
      "epoch": 2.89968166566291,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003760270081032839,
      "loss": 1.5734,
      "step": 33703
    },
    {
      "epoch": 2.8997677019702315,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003760008119268985,
      "loss": 1.0307,
      "step": 33704
    },
    {
      "epoch": 2.899853738277553,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00037597461611318657,
      "loss": 0.8356,
      "step": 33705
    },
    {
      "epoch": 2.8999397745848747,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003759484206622249,
      "loss": 0.9389,
      "step": 33706
    },
    {
      "epoch": 2.9000258108921964,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003759222255740899,
      "loss": 0.8595,
      "step": 33707
    },
    {
      "epoch": 2.9001118471995184,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003758960308488581,
      "loss": 0.8619,
      "step": 33708
    },
    {
      "epoch": 2.90019788350684,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00037586983648660646,
      "loss": 1.3941,
      "step": 33709
    },
    {
      "epoch": 2.9002839198141617,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0003758436424874113,
      "loss": 0.5928,
      "step": 33710
    },
    {
      "epoch": 2.9003699561214833,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00037581744885134945,
      "loss": 0.5773,
      "step": 33711
    },
    {
      "epoch": 2.900455992428805,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00037579125557849743,
      "loss": 0.6328,
      "step": 33712
    },
    {
      "epoch": 2.9005420287361265,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003757650626689318,
      "loss": 1.2826,
      "step": 33713
    },
    {
      "epoch": 2.900628065043448,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00037573887012272905,
      "loss": 1.4591,
      "step": 33714
    },
    {
      "epoch": 2.9007141013507702,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0003757126779399661,
      "loss": 1.4614,
      "step": 33715
    },
    {
      "epoch": 2.900800137658092,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00037568648612071944,
      "loss": 1.2723,
      "step": 33716
    },
    {
      "epoch": 2.9008861739654135,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00037566029466506545,
      "loss": 1.2522,
      "step": 33717
    },
    {
      "epoch": 2.900972210272735,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000375634103573081,
      "loss": 0.7937,
      "step": 33718
    },
    {
      "epoch": 2.9010582465800567,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003756079128448427,
      "loss": 0.642,
      "step": 33719
    },
    {
      "epoch": 2.9011442828873784,
      "grad_norm": 1.15625,
      "learning_rate": 0.000375581722480427,
      "loss": 1.4507,
      "step": 33720
    },
    {
      "epoch": 2.9012303191947,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003755555324799105,
      "loss": 1.638,
      "step": 33721
    },
    {
      "epoch": 2.901316355502022,
      "grad_norm": 1.046875,
      "learning_rate": 0.00037552934284336995,
      "loss": 1.6205,
      "step": 33722
    },
    {
      "epoch": 2.9014023918093437,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00037550315357088183,
      "loss": 1.0186,
      "step": 33723
    },
    {
      "epoch": 2.9014884281166653,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003754769646625228,
      "loss": 2.1365,
      "step": 33724
    },
    {
      "epoch": 2.901574464423987,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003754507761183694,
      "loss": 0.8261,
      "step": 33725
    },
    {
      "epoch": 2.9016605007313085,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003754245879384982,
      "loss": 1.265,
      "step": 33726
    },
    {
      "epoch": 2.90174653703863,
      "grad_norm": 1.015625,
      "learning_rate": 0.000375398400122986,
      "loss": 0.8558,
      "step": 33727
    },
    {
      "epoch": 2.901832573345952,
      "grad_norm": 1.203125,
      "learning_rate": 0.00037537221267190907,
      "loss": 0.9952,
      "step": 33728
    },
    {
      "epoch": 2.901918609653274,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00037534602558534426,
      "loss": 1.2729,
      "step": 33729
    },
    {
      "epoch": 2.9020046459605955,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00037531983886336814,
      "loss": 1.6046,
      "step": 33730
    },
    {
      "epoch": 2.902090682267917,
      "grad_norm": 1.015625,
      "learning_rate": 0.00037529365250605723,
      "loss": 0.9968,
      "step": 33731
    },
    {
      "epoch": 2.9021767185752387,
      "grad_norm": 1.15625,
      "learning_rate": 0.000375267466513488,
      "loss": 0.9469,
      "step": 33732
    },
    {
      "epoch": 2.9022627548825604,
      "grad_norm": 1.015625,
      "learning_rate": 0.00037524128088573726,
      "loss": 1.3645,
      "step": 33733
    },
    {
      "epoch": 2.902348791189882,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003752150956228816,
      "loss": 0.8024,
      "step": 33734
    },
    {
      "epoch": 2.9024348274972036,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00037518891072499736,
      "loss": 0.7566,
      "step": 33735
    },
    {
      "epoch": 2.9025208638045257,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003751627261921614,
      "loss": 0.7932,
      "step": 33736
    },
    {
      "epoch": 2.9026069001118473,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00037513654202445015,
      "loss": 1.3343,
      "step": 33737
    },
    {
      "epoch": 2.902692936419169,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00037511035822194023,
      "loss": 0.8719,
      "step": 33738
    },
    {
      "epoch": 2.9027789727264905,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003750841747847081,
      "loss": 1.5511,
      "step": 33739
    },
    {
      "epoch": 2.902865009033812,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003750579917128306,
      "loss": 1.986,
      "step": 33740
    },
    {
      "epoch": 2.902951045341134,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00037503180900638424,
      "loss": 1.5795,
      "step": 33741
    },
    {
      "epoch": 2.9030370816484554,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00037500562666544537,
      "loss": 1.0154,
      "step": 33742
    },
    {
      "epoch": 2.9031231179557775,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003749794446900908,
      "loss": 1.5297,
      "step": 33743
    },
    {
      "epoch": 2.903209154263099,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00037495326308039707,
      "loss": 1.0279,
      "step": 33744
    },
    {
      "epoch": 2.9032951905704207,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003749270818364405,
      "loss": 1.4789,
      "step": 33745
    },
    {
      "epoch": 2.9033812268777424,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003749009009582981,
      "loss": 1.1721,
      "step": 33746
    },
    {
      "epoch": 2.903467263185064,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003748747204460463,
      "loss": 1.5495,
      "step": 33747
    },
    {
      "epoch": 2.903553299492386,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003748485402997615,
      "loss": 0.6437,
      "step": 33748
    },
    {
      "epoch": 2.9036393357997072,
      "grad_norm": 2.171875,
      "learning_rate": 0.00037482236051952035,
      "loss": 1.7509,
      "step": 33749
    },
    {
      "epoch": 2.9037253721070293,
      "grad_norm": 1.28125,
      "learning_rate": 0.00037479618110539955,
      "loss": 1.981,
      "step": 33750
    },
    {
      "epoch": 2.903811408414351,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00037477000205747547,
      "loss": 0.7728,
      "step": 33751
    },
    {
      "epoch": 2.9038974447216725,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00037474382337582485,
      "loss": 1.337,
      "step": 33752
    },
    {
      "epoch": 2.903983481028994,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0003747176450605242,
      "loss": 2.0211,
      "step": 33753
    },
    {
      "epoch": 2.904069517336316,
      "grad_norm": 0.90625,
      "learning_rate": 0.00037469146711165005,
      "loss": 1.1,
      "step": 33754
    },
    {
      "epoch": 2.904155553643638,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003746652895292789,
      "loss": 1.6793,
      "step": 33755
    },
    {
      "epoch": 2.904241589950959,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003746391123134876,
      "loss": 0.9704,
      "step": 33756
    },
    {
      "epoch": 2.904327626258281,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003746129354643524,
      "loss": 1.0171,
      "step": 33757
    },
    {
      "epoch": 2.9044136625656027,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0003745867589819498,
      "loss": 0.9029,
      "step": 33758
    },
    {
      "epoch": 2.9044996988729244,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003745605828663568,
      "loss": 1.3104,
      "step": 33759
    },
    {
      "epoch": 2.904585735180246,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00037453440711764964,
      "loss": 1.2123,
      "step": 33760
    },
    {
      "epoch": 2.9046717714875676,
      "grad_norm": 4.90625,
      "learning_rate": 0.000374508231735905,
      "loss": 1.5307,
      "step": 33761
    },
    {
      "epoch": 2.9047578077948897,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00037448205672119917,
      "loss": 1.0201,
      "step": 33762
    },
    {
      "epoch": 2.904843844102211,
      "grad_norm": 0.875,
      "learning_rate": 0.000374455882073609,
      "loss": 1.4126,
      "step": 33763
    },
    {
      "epoch": 2.904929880409533,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00037442970779321105,
      "loss": 1.0198,
      "step": 33764
    },
    {
      "epoch": 2.9050159167168546,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00037440353388008173,
      "loss": 1.119,
      "step": 33765
    },
    {
      "epoch": 2.905101953024176,
      "grad_norm": 1.046875,
      "learning_rate": 0.00037437736033429757,
      "loss": 1.0357,
      "step": 33766
    },
    {
      "epoch": 2.905187989331498,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003743511871559353,
      "loss": 1.0929,
      "step": 33767
    },
    {
      "epoch": 2.9052740256388194,
      "grad_norm": 1.203125,
      "learning_rate": 0.00037432501434507136,
      "loss": 1.1357,
      "step": 33768
    },
    {
      "epoch": 2.9053600619461415,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003742988419017822,
      "loss": 1.454,
      "step": 33769
    },
    {
      "epoch": 2.905446098253463,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003742726698261446,
      "loss": 1.297,
      "step": 33770
    },
    {
      "epoch": 2.9055321345607847,
      "grad_norm": 1.0,
      "learning_rate": 0.000374246498118235,
      "loss": 1.0691,
      "step": 33771
    },
    {
      "epoch": 2.9056181708681064,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00037422032677812987,
      "loss": 1.0507,
      "step": 33772
    },
    {
      "epoch": 2.905704207175428,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003741941558059058,
      "loss": 1.2716,
      "step": 33773
    },
    {
      "epoch": 2.9057902434827496,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003741679852016394,
      "loss": 1.3712,
      "step": 33774
    },
    {
      "epoch": 2.9058762797900712,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003741418149654072,
      "loss": 1.7682,
      "step": 33775
    },
    {
      "epoch": 2.9059623160973933,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003741156450972856,
      "loss": 1.266,
      "step": 33776
    },
    {
      "epoch": 2.906048352404715,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003740894755973514,
      "loss": 1.3425,
      "step": 33777
    },
    {
      "epoch": 2.9061343887120366,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00037406330646568096,
      "loss": 1.6905,
      "step": 33778
    },
    {
      "epoch": 2.906220425019358,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003740371377023508,
      "loss": 0.9927,
      "step": 33779
    },
    {
      "epoch": 2.90630646132668,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003740109693074375,
      "loss": 1.1286,
      "step": 33780
    },
    {
      "epoch": 2.9063924976340014,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003739848012810176,
      "loss": 0.8994,
      "step": 33781
    },
    {
      "epoch": 2.906478533941323,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003739586336231676,
      "loss": 1.8874,
      "step": 33782
    },
    {
      "epoch": 2.906564570248645,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00037393246633396417,
      "loss": 1.2935,
      "step": 33783
    },
    {
      "epoch": 2.9066506065559667,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00037390629941348383,
      "loss": 0.9474,
      "step": 33784
    },
    {
      "epoch": 2.9067366428632884,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003738801328618029,
      "loss": 0.9035,
      "step": 33785
    },
    {
      "epoch": 2.90682267917061,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003738539666789982,
      "loss": 1.0183,
      "step": 33786
    },
    {
      "epoch": 2.9069087154779316,
      "grad_norm": 0.953125,
      "learning_rate": 0.000373827800865146,
      "loss": 0.8279,
      "step": 33787
    },
    {
      "epoch": 2.9069947517852532,
      "grad_norm": 0.984375,
      "learning_rate": 0.000373801635420323,
      "loss": 0.9976,
      "step": 33788
    },
    {
      "epoch": 2.907080788092575,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003737754703446056,
      "loss": 1.2943,
      "step": 33789
    },
    {
      "epoch": 2.907166824399897,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003737493056380705,
      "loss": 0.8007,
      "step": 33790
    },
    {
      "epoch": 2.9072528607072186,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003737231413007941,
      "loss": 1.0167,
      "step": 33791
    },
    {
      "epoch": 2.90733889701454,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003736969773328529,
      "loss": 1.0465,
      "step": 33792
    },
    {
      "epoch": 2.907424933321862,
      "grad_norm": 1.421875,
      "learning_rate": 0.00037367081373432357,
      "loss": 1.4247,
      "step": 33793
    },
    {
      "epoch": 2.9075109696291834,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00037364465050528247,
      "loss": 0.7808,
      "step": 33794
    },
    {
      "epoch": 2.907597005936505,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00037361848764580626,
      "loss": 0.5015,
      "step": 33795
    },
    {
      "epoch": 2.9076830422438267,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003735923251559712,
      "loss": 0.6359,
      "step": 33796
    },
    {
      "epoch": 2.9077690785511487,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003735661630358542,
      "loss": 2.167,
      "step": 33797
    },
    {
      "epoch": 2.9078551148584704,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003735400012855316,
      "loss": 1.5324,
      "step": 33798
    },
    {
      "epoch": 2.907941151165792,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003735138399050797,
      "loss": 1.5817,
      "step": 33799
    },
    {
      "epoch": 2.9080271874731136,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003734876788945754,
      "loss": 1.3066,
      "step": 33800
    },
    {
      "epoch": 2.9081132237804352,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00037346151825409503,
      "loss": 1.5175,
      "step": 33801
    },
    {
      "epoch": 2.9081992600877573,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003734353579837151,
      "loss": 0.8312,
      "step": 33802
    },
    {
      "epoch": 2.9082852963950785,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000373409198083512,
      "loss": 0.549,
      "step": 33803
    },
    {
      "epoch": 2.9083713327024006,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003733830385535625,
      "loss": 1.189,
      "step": 33804
    },
    {
      "epoch": 2.908457369009722,
      "grad_norm": 1.015625,
      "learning_rate": 0.00037335687939394295,
      "loss": 1.0121,
      "step": 33805
    },
    {
      "epoch": 2.908543405317044,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003733307206047298,
      "loss": 0.6989,
      "step": 33806
    },
    {
      "epoch": 2.9086294416243654,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003733045621859998,
      "loss": 1.7894,
      "step": 33807
    },
    {
      "epoch": 2.908715477931687,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003732784041378293,
      "loss": 1.0988,
      "step": 33808
    },
    {
      "epoch": 2.908801514239009,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00037325224646029485,
      "loss": 0.9231,
      "step": 33809
    },
    {
      "epoch": 2.9088875505463303,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003732260891534728,
      "loss": 1.6404,
      "step": 33810
    },
    {
      "epoch": 2.9089735868536524,
      "grad_norm": 1.203125,
      "learning_rate": 0.00037319993221743985,
      "loss": 0.8807,
      "step": 33811
    },
    {
      "epoch": 2.909059623160974,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003731737756522725,
      "loss": 0.8863,
      "step": 33812
    },
    {
      "epoch": 2.9091456594682956,
      "grad_norm": 1.015625,
      "learning_rate": 0.00037314761945804714,
      "loss": 1.1698,
      "step": 33813
    },
    {
      "epoch": 2.9092316957756172,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00037312146363484023,
      "loss": 0.6169,
      "step": 33814
    },
    {
      "epoch": 2.909317732082939,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003730953081827285,
      "loss": 1.5616,
      "step": 33815
    },
    {
      "epoch": 2.909403768390261,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003730691531017883,
      "loss": 1.9969,
      "step": 33816
    },
    {
      "epoch": 2.909489804697582,
      "grad_norm": 0.875,
      "learning_rate": 0.00037304299839209596,
      "loss": 0.7191,
      "step": 33817
    },
    {
      "epoch": 2.909575841004904,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003730168440537284,
      "loss": 1.3086,
      "step": 33818
    },
    {
      "epoch": 2.909661877312226,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00037299069008676157,
      "loss": 0.8445,
      "step": 33819
    },
    {
      "epoch": 2.9097479136195474,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003729645364912725,
      "loss": 1.0104,
      "step": 33820
    },
    {
      "epoch": 2.909833949926869,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003729383832673374,
      "loss": 0.8857,
      "step": 33821
    },
    {
      "epoch": 2.9099199862341907,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003729122304150328,
      "loss": 1.5318,
      "step": 33822
    },
    {
      "epoch": 2.9100060225415127,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003728860779344353,
      "loss": 1.5078,
      "step": 33823
    },
    {
      "epoch": 2.9100920588488344,
      "grad_norm": 1.03125,
      "learning_rate": 0.00037285992582562126,
      "loss": 0.7044,
      "step": 33824
    },
    {
      "epoch": 2.910178095156156,
      "grad_norm": 1.15625,
      "learning_rate": 0.00037283377408866723,
      "loss": 0.9936,
      "step": 33825
    },
    {
      "epoch": 2.9102641314634776,
      "grad_norm": 0.828125,
      "learning_rate": 0.00037280762272364957,
      "loss": 0.9188,
      "step": 33826
    },
    {
      "epoch": 2.9103501677707992,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00037278147173064503,
      "loss": 1.4108,
      "step": 33827
    },
    {
      "epoch": 2.910436204078121,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00037275532110972985,
      "loss": 1.3463,
      "step": 33828
    },
    {
      "epoch": 2.9105222403854425,
      "grad_norm": 1.296875,
      "learning_rate": 0.00037272917086098056,
      "loss": 1.2715,
      "step": 33829
    },
    {
      "epoch": 2.9106082766927646,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00037270302098447376,
      "loss": 0.7402,
      "step": 33830
    },
    {
      "epoch": 2.910694313000086,
      "grad_norm": 1.109375,
      "learning_rate": 0.00037267687148028595,
      "loss": 1.6928,
      "step": 33831
    },
    {
      "epoch": 2.910780349307408,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003726507223484934,
      "loss": 1.0345,
      "step": 33832
    },
    {
      "epoch": 2.9108663856147294,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003726245735891727,
      "loss": 1.1456,
      "step": 33833
    },
    {
      "epoch": 2.910952421922051,
      "grad_norm": 1.171875,
      "learning_rate": 0.00037259842520240044,
      "loss": 1.1506,
      "step": 33834
    },
    {
      "epoch": 2.9110384582293727,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00037257227718825303,
      "loss": 1.2757,
      "step": 33835
    },
    {
      "epoch": 2.9111244945366943,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003725461295468069,
      "loss": 1.556,
      "step": 33836
    },
    {
      "epoch": 2.9112105308440164,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003725199822781384,
      "loss": 1.4886,
      "step": 33837
    },
    {
      "epoch": 2.911296567151338,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00037249383538232435,
      "loss": 1.2746,
      "step": 33838
    },
    {
      "epoch": 2.9113826034586596,
      "grad_norm": 1.03125,
      "learning_rate": 0.000372467688859441,
      "loss": 1.8101,
      "step": 33839
    },
    {
      "epoch": 2.9114686397659812,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003724415427095648,
      "loss": 1.0083,
      "step": 33840
    },
    {
      "epoch": 2.911554676073303,
      "grad_norm": 6.59375,
      "learning_rate": 0.00037241539693277226,
      "loss": 0.9694,
      "step": 33841
    },
    {
      "epoch": 2.9116407123806245,
      "grad_norm": 1.578125,
      "learning_rate": 0.00037238925152913995,
      "loss": 1.0249,
      "step": 33842
    },
    {
      "epoch": 2.911726748687946,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003723631064987443,
      "loss": 1.7273,
      "step": 33843
    },
    {
      "epoch": 2.911812784995268,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00037233696184166157,
      "loss": 1.0348,
      "step": 33844
    },
    {
      "epoch": 2.91189882130259,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00037231081755796847,
      "loss": 1.4195,
      "step": 33845
    },
    {
      "epoch": 2.9119848576099114,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00037228467364774145,
      "loss": 1.0283,
      "step": 33846
    },
    {
      "epoch": 2.912070893917233,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003722585301110568,
      "loss": 1.9734,
      "step": 33847
    },
    {
      "epoch": 2.9121569302245547,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00037223238694799124,
      "loss": 0.7978,
      "step": 33848
    },
    {
      "epoch": 2.9122429665318763,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00037220624415862104,
      "loss": 1.1473,
      "step": 33849
    },
    {
      "epoch": 2.912329002839198,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00037218010174302276,
      "loss": 1.2578,
      "step": 33850
    },
    {
      "epoch": 2.91241503914652,
      "grad_norm": 7.84375,
      "learning_rate": 0.00037215395970127273,
      "loss": 1.7995,
      "step": 33851
    },
    {
      "epoch": 2.9125010754538416,
      "grad_norm": 0.921875,
      "learning_rate": 0.00037212781803344766,
      "loss": 0.9318,
      "step": 33852
    },
    {
      "epoch": 2.9125871117611632,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0003721016767396237,
      "loss": 0.9746,
      "step": 33853
    },
    {
      "epoch": 2.912673148068485,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003720755358198776,
      "loss": 1.9241,
      "step": 33854
    },
    {
      "epoch": 2.9127591843758065,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00037204939527428547,
      "loss": 0.8822,
      "step": 33855
    },
    {
      "epoch": 2.912845220683128,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00037202325510292395,
      "loss": 0.8842,
      "step": 33856
    },
    {
      "epoch": 2.9129312569904497,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00037199711530586967,
      "loss": 1.2508,
      "step": 33857
    },
    {
      "epoch": 2.913017293297772,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00037197097588319896,
      "loss": 1.2637,
      "step": 33858
    },
    {
      "epoch": 2.9131033296050934,
      "grad_norm": 1.265625,
      "learning_rate": 0.00037194483683498827,
      "loss": 0.8266,
      "step": 33859
    },
    {
      "epoch": 2.913189365912415,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00037191869816131386,
      "loss": 1.3269,
      "step": 33860
    },
    {
      "epoch": 2.9132754022197367,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003718925598622525,
      "loss": 1.4332,
      "step": 33861
    },
    {
      "epoch": 2.9133614385270583,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003718664219378805,
      "loss": 0.9715,
      "step": 33862
    },
    {
      "epoch": 2.9134474748343804,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003718402843882742,
      "loss": 1.0461,
      "step": 33863
    },
    {
      "epoch": 2.9135335111417016,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00037181414721351014,
      "loss": 0.6132,
      "step": 33864
    },
    {
      "epoch": 2.9136195474490236,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003717880104136649,
      "loss": 1.1197,
      "step": 33865
    },
    {
      "epoch": 2.9137055837563453,
      "grad_norm": 1.125,
      "learning_rate": 0.0003717618739888147,
      "loss": 1.5387,
      "step": 33866
    },
    {
      "epoch": 2.913791620063667,
      "grad_norm": 1.46875,
      "learning_rate": 0.00037173573793903604,
      "loss": 1.6086,
      "step": 33867
    },
    {
      "epoch": 2.9138776563709885,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003717096022644054,
      "loss": 1.2263,
      "step": 33868
    },
    {
      "epoch": 2.91396369267831,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003716834669649993,
      "loss": 1.3618,
      "step": 33869
    },
    {
      "epoch": 2.914049728985632,
      "grad_norm": 1.046875,
      "learning_rate": 0.00037165733204089403,
      "loss": 2.021,
      "step": 33870
    },
    {
      "epoch": 2.9141357652929534,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0003716311974921662,
      "loss": 0.787,
      "step": 33871
    },
    {
      "epoch": 2.9142218016002754,
      "grad_norm": 58.25,
      "learning_rate": 0.0003716050633188921,
      "loss": 1.2051,
      "step": 33872
    },
    {
      "epoch": 2.914307837907597,
      "grad_norm": 3.546875,
      "learning_rate": 0.00037157892952114827,
      "loss": 1.0961,
      "step": 33873
    },
    {
      "epoch": 2.9143938742149187,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003715527960990109,
      "loss": 0.7173,
      "step": 33874
    },
    {
      "epoch": 2.9144799105222403,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003715266630525568,
      "loss": 1.2638,
      "step": 33875
    },
    {
      "epoch": 2.914565946829562,
      "grad_norm": 1.0,
      "learning_rate": 0.00037150053038186227,
      "loss": 1.3054,
      "step": 33876
    },
    {
      "epoch": 2.914651983136884,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00037147439808700363,
      "loss": 0.7821,
      "step": 33877
    },
    {
      "epoch": 2.914738019444205,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003714482661680573,
      "loss": 1.12,
      "step": 33878
    },
    {
      "epoch": 2.9148240557515273,
      "grad_norm": 0.96875,
      "learning_rate": 0.00037142213462509985,
      "loss": 1.0608,
      "step": 33879
    },
    {
      "epoch": 2.914910092058849,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003713960034582077,
      "loss": 1.1418,
      "step": 33880
    },
    {
      "epoch": 2.9149961283661705,
      "grad_norm": 1.015625,
      "learning_rate": 0.00037136987266745704,
      "loss": 0.9069,
      "step": 33881
    },
    {
      "epoch": 2.915082164673492,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00037134374225292465,
      "loss": 1.3853,
      "step": 33882
    },
    {
      "epoch": 2.9151682009808138,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003713176122146868,
      "loss": 1.2034,
      "step": 33883
    },
    {
      "epoch": 2.915254237288136,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003712914825528199,
      "loss": 1.2574,
      "step": 33884
    },
    {
      "epoch": 2.9153402735954574,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003712653532674003,
      "loss": 1.5454,
      "step": 33885
    },
    {
      "epoch": 2.915426309902779,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003712392243585045,
      "loss": 1.3714,
      "step": 33886
    },
    {
      "epoch": 2.9155123462101007,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003712130958262091,
      "loss": 1.1643,
      "step": 33887
    },
    {
      "epoch": 2.9155983825174223,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003711869676705901,
      "loss": 1.2547,
      "step": 33888
    },
    {
      "epoch": 2.915684418824744,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00037116083989172436,
      "loss": 1.4859,
      "step": 33889
    },
    {
      "epoch": 2.9157704551320656,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00037113471248968803,
      "loss": 1.5372,
      "step": 33890
    },
    {
      "epoch": 2.9158564914393876,
      "grad_norm": 8.8125,
      "learning_rate": 0.0003711085854645576,
      "loss": 0.9723,
      "step": 33891
    },
    {
      "epoch": 2.9159425277467093,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00037108245881640946,
      "loss": 0.8416,
      "step": 33892
    },
    {
      "epoch": 2.916028564054031,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003710563325453199,
      "loss": 1.1255,
      "step": 33893
    },
    {
      "epoch": 2.9161146003613525,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003710302066513658,
      "loss": 1.1388,
      "step": 33894
    },
    {
      "epoch": 2.916200636668674,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00037100408113462314,
      "loss": 1.1284,
      "step": 33895
    },
    {
      "epoch": 2.9162866729759958,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003709779559951685,
      "loss": 1.7827,
      "step": 33896
    },
    {
      "epoch": 2.9163727092833174,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003709518312330781,
      "loss": 1.9952,
      "step": 33897
    },
    {
      "epoch": 2.9164587455906394,
      "grad_norm": 1.46875,
      "learning_rate": 0.0003709257068484286,
      "loss": 0.8152,
      "step": 33898
    },
    {
      "epoch": 2.916544781897961,
      "grad_norm": 1.703125,
      "learning_rate": 0.00037089958284129633,
      "loss": 1.767,
      "step": 33899
    },
    {
      "epoch": 2.9166308182052827,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00037087345921175773,
      "loss": 1.064,
      "step": 33900
    },
    {
      "epoch": 2.9167168545126043,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00037084733595988897,
      "loss": 0.854,
      "step": 33901
    },
    {
      "epoch": 2.916802890819926,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00037082121308576677,
      "loss": 0.7625,
      "step": 33902
    },
    {
      "epoch": 2.9168889271272476,
      "grad_norm": 15.125,
      "learning_rate": 0.00037079509058946734,
      "loss": 1.4778,
      "step": 33903
    },
    {
      "epoch": 2.916974963434569,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003707689684710671,
      "loss": 1.0017,
      "step": 33904
    },
    {
      "epoch": 2.9170609997418913,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003707428467306426,
      "loss": 0.9386,
      "step": 33905
    },
    {
      "epoch": 2.917147036049213,
      "grad_norm": 2.078125,
      "learning_rate": 0.0003707167253682701,
      "loss": 1.3639,
      "step": 33906
    },
    {
      "epoch": 2.9172330723565345,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000370690604384026,
      "loss": 0.6947,
      "step": 33907
    },
    {
      "epoch": 2.917319108663856,
      "grad_norm": 1.71875,
      "learning_rate": 0.00037066448377798673,
      "loss": 1.1107,
      "step": 33908
    },
    {
      "epoch": 2.9174051449711778,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0003706383635502287,
      "loss": 1.4773,
      "step": 33909
    },
    {
      "epoch": 2.9174911812784994,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003706122437008284,
      "loss": 0.8542,
      "step": 33910
    },
    {
      "epoch": 2.917577217585821,
      "grad_norm": 15.3125,
      "learning_rate": 0.0003705861242298619,
      "loss": 1.2068,
      "step": 33911
    },
    {
      "epoch": 2.917663253893143,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000370560005137406,
      "loss": 0.4845,
      "step": 33912
    },
    {
      "epoch": 2.9177492902004647,
      "grad_norm": 1.15625,
      "learning_rate": 0.00037053388642353693,
      "loss": 1.3296,
      "step": 33913
    },
    {
      "epoch": 2.9178353265077863,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00037050776808833096,
      "loss": 0.8929,
      "step": 33914
    },
    {
      "epoch": 2.917921362815108,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00037048165013186455,
      "loss": 1.2316,
      "step": 33915
    },
    {
      "epoch": 2.9180073991224296,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00037045553255421427,
      "loss": 0.55,
      "step": 33916
    },
    {
      "epoch": 2.918093435429751,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003704294153554563,
      "loss": 1.0351,
      "step": 33917
    },
    {
      "epoch": 2.918179471737073,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000370403298535667,
      "loss": 1.3428,
      "step": 33918
    },
    {
      "epoch": 2.918265508044395,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003703771820949229,
      "loss": 1.1223,
      "step": 33919
    },
    {
      "epoch": 2.9183515443517165,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0003703510660333004,
      "loss": 1.2926,
      "step": 33920
    },
    {
      "epoch": 2.918437580659038,
      "grad_norm": 51.0,
      "learning_rate": 0.00037032495035087574,
      "loss": 1.0835,
      "step": 33921
    },
    {
      "epoch": 2.9185236169663598,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003702988350477254,
      "loss": 0.6896,
      "step": 33922
    },
    {
      "epoch": 2.9186096532736814,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00037027272012392576,
      "loss": 0.8339,
      "step": 33923
    },
    {
      "epoch": 2.9186956895810034,
      "grad_norm": 1.0,
      "learning_rate": 0.0003702466055795532,
      "loss": 1.2865,
      "step": 33924
    },
    {
      "epoch": 2.9187817258883246,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00037022049141468405,
      "loss": 0.8181,
      "step": 33925
    },
    {
      "epoch": 2.9188677621956467,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003701943776293947,
      "loss": 0.7444,
      "step": 33926
    },
    {
      "epoch": 2.9189537985029683,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003701682642237616,
      "loss": 1.3116,
      "step": 33927
    },
    {
      "epoch": 2.91903983481029,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003701421511978611,
      "loss": 2.1309,
      "step": 33928
    },
    {
      "epoch": 2.9191258711176116,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0003701160385517694,
      "loss": 0.805,
      "step": 33929
    },
    {
      "epoch": 2.919211907424933,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00037008992628556305,
      "loss": 2.1994,
      "step": 33930
    },
    {
      "epoch": 2.9192979437322553,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003700638143993184,
      "loss": 0.4751,
      "step": 33931
    },
    {
      "epoch": 2.9193839800395764,
      "grad_norm": 1.328125,
      "learning_rate": 0.00037003770289311197,
      "loss": 1.1991,
      "step": 33932
    },
    {
      "epoch": 2.9194700163468985,
      "grad_norm": 1.15625,
      "learning_rate": 0.00037001159176701994,
      "loss": 1.5809,
      "step": 33933
    },
    {
      "epoch": 2.91955605265422,
      "grad_norm": 1.09375,
      "learning_rate": 0.00036998548102111864,
      "loss": 1.2561,
      "step": 33934
    },
    {
      "epoch": 2.9196420889615418,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00036995937065548455,
      "loss": 1.7133,
      "step": 33935
    },
    {
      "epoch": 2.9197281252688634,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00036993326067019406,
      "loss": 1.0965,
      "step": 33936
    },
    {
      "epoch": 2.919814161576185,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00036990715106532355,
      "loss": 1.5633,
      "step": 33937
    },
    {
      "epoch": 2.919900197883507,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003698810418409491,
      "loss": 1.0579,
      "step": 33938
    },
    {
      "epoch": 2.9199862341908283,
      "grad_norm": 1.484375,
      "learning_rate": 0.00036985493299714745,
      "loss": 1.6127,
      "step": 33939
    },
    {
      "epoch": 2.9200722704981503,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00036982882453399484,
      "loss": 0.9822,
      "step": 33940
    },
    {
      "epoch": 2.920158306805472,
      "grad_norm": 3.6875,
      "learning_rate": 0.0003698027164515676,
      "loss": 1.2287,
      "step": 33941
    },
    {
      "epoch": 2.9202443431127936,
      "grad_norm": 0.921875,
      "learning_rate": 0.00036977660874994194,
      "loss": 1.4363,
      "step": 33942
    },
    {
      "epoch": 2.920330379420115,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003697505014291945,
      "loss": 1.337,
      "step": 33943
    },
    {
      "epoch": 2.920416415727437,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003697243944894015,
      "loss": 1.2082,
      "step": 33944
    },
    {
      "epoch": 2.920502452034759,
      "grad_norm": 1.171875,
      "learning_rate": 0.00036969828793063925,
      "loss": 1.3203,
      "step": 33945
    },
    {
      "epoch": 2.9205884883420805,
      "grad_norm": 1.25,
      "learning_rate": 0.0003696721817529842,
      "loss": 1.3944,
      "step": 33946
    },
    {
      "epoch": 2.920674524649402,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003696460759565128,
      "loss": 1.0655,
      "step": 33947
    },
    {
      "epoch": 2.9207605609567238,
      "grad_norm": 1.203125,
      "learning_rate": 0.00036961997054130115,
      "loss": 1.1031,
      "step": 33948
    },
    {
      "epoch": 2.9208465972640454,
      "grad_norm": 3.140625,
      "learning_rate": 0.0003695938655074257,
      "loss": 0.4629,
      "step": 33949
    },
    {
      "epoch": 2.920932633571367,
      "grad_norm": 1.140625,
      "learning_rate": 0.00036956776085496286,
      "loss": 2.1239,
      "step": 33950
    },
    {
      "epoch": 2.9210186698786886,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000369541656583989,
      "loss": 1.3361,
      "step": 33951
    },
    {
      "epoch": 2.9211047061860107,
      "grad_norm": 1.09375,
      "learning_rate": 0.00036951555269458026,
      "loss": 1.8637,
      "step": 33952
    },
    {
      "epoch": 2.9211907424933323,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00036948944918681337,
      "loss": 1.3815,
      "step": 33953
    },
    {
      "epoch": 2.921276778800654,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00036946334606076435,
      "loss": 0.7782,
      "step": 33954
    },
    {
      "epoch": 2.9213628151079756,
      "grad_norm": 0.890625,
      "learning_rate": 0.00036943724331650976,
      "loss": 0.8484,
      "step": 33955
    },
    {
      "epoch": 2.921448851415297,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00036941114095412564,
      "loss": 0.8118,
      "step": 33956
    },
    {
      "epoch": 2.921534887722619,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00036938503897368866,
      "loss": 1.0887,
      "step": 33957
    },
    {
      "epoch": 2.9216209240299404,
      "grad_norm": 51.0,
      "learning_rate": 0.000369358937375275,
      "loss": 1.4298,
      "step": 33958
    },
    {
      "epoch": 2.9217069603372625,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000369332836158961,
      "loss": 0.8326,
      "step": 33959
    },
    {
      "epoch": 2.921792996644584,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003693067353248232,
      "loss": 1.1925,
      "step": 33960
    },
    {
      "epoch": 2.9218790329519058,
      "grad_norm": 1.359375,
      "learning_rate": 0.00036928063487293763,
      "loss": 1.5679,
      "step": 33961
    },
    {
      "epoch": 2.9219650692592274,
      "grad_norm": 2.6875,
      "learning_rate": 0.0003692545348033809,
      "loss": 1.5632,
      "step": 33962
    },
    {
      "epoch": 2.922051105566549,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00036922843511622905,
      "loss": 1.3181,
      "step": 33963
    },
    {
      "epoch": 2.9221371418738706,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003692023358115587,
      "loss": 1.3952,
      "step": 33964
    },
    {
      "epoch": 2.9222231781811923,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00036917623688944613,
      "loss": 0.8177,
      "step": 33965
    },
    {
      "epoch": 2.9223092144885143,
      "grad_norm": 0.734375,
      "learning_rate": 0.0003691501383499676,
      "loss": 0.723,
      "step": 33966
    },
    {
      "epoch": 2.922395250795836,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0003691240401931992,
      "loss": 0.8409,
      "step": 33967
    },
    {
      "epoch": 2.9224812871031576,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003690979424192178,
      "loss": 0.5802,
      "step": 33968
    },
    {
      "epoch": 2.922567323410479,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003690718450280994,
      "loss": 1.3453,
      "step": 33969
    },
    {
      "epoch": 2.922653359717801,
      "grad_norm": 3.578125,
      "learning_rate": 0.0003690457480199204,
      "loss": 1.576,
      "step": 33970
    },
    {
      "epoch": 2.9227393960251224,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00036901965139475725,
      "loss": 0.9956,
      "step": 33971
    },
    {
      "epoch": 2.922825432332444,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00036899355515268594,
      "loss": 2.2137,
      "step": 33972
    },
    {
      "epoch": 2.922911468639766,
      "grad_norm": 0.78125,
      "learning_rate": 0.00036896745929378307,
      "loss": 0.6847,
      "step": 33973
    },
    {
      "epoch": 2.9229975049470878,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00036894136381812495,
      "loss": 1.2407,
      "step": 33974
    },
    {
      "epoch": 2.9230835412544094,
      "grad_norm": 51.75,
      "learning_rate": 0.00036891526872578776,
      "loss": 0.9373,
      "step": 33975
    },
    {
      "epoch": 2.923169577561731,
      "grad_norm": 0.890625,
      "learning_rate": 0.00036888917401684803,
      "loss": 1.6581,
      "step": 33976
    },
    {
      "epoch": 2.9232556138690526,
      "grad_norm": 2.0625,
      "learning_rate": 0.0003688630796913819,
      "loss": 1.0693,
      "step": 33977
    },
    {
      "epoch": 2.9233416501763747,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00036883698574946577,
      "loss": 1.3344,
      "step": 33978
    },
    {
      "epoch": 2.923427686483696,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00036881089219117593,
      "loss": 1.2698,
      "step": 33979
    },
    {
      "epoch": 2.923513722791018,
      "grad_norm": 0.65625,
      "learning_rate": 0.0003687847990165887,
      "loss": 0.4732,
      "step": 33980
    },
    {
      "epoch": 2.9235997590983396,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003687587062257805,
      "loss": 0.9392,
      "step": 33981
    },
    {
      "epoch": 2.923685795405661,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00036873261381882745,
      "loss": 1.4443,
      "step": 33982
    },
    {
      "epoch": 2.923771831712983,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00036870652179580606,
      "loss": 0.8908,
      "step": 33983
    },
    {
      "epoch": 2.9238578680203045,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00036868043015679254,
      "loss": 1.3018,
      "step": 33984
    },
    {
      "epoch": 2.9239439043276265,
      "grad_norm": 2.28125,
      "learning_rate": 0.0003686543389018632,
      "loss": 1.4529,
      "step": 33985
    },
    {
      "epoch": 2.9240299406349477,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003686282480310943,
      "loss": 1.1548,
      "step": 33986
    },
    {
      "epoch": 2.9241159769422698,
      "grad_norm": 55.25,
      "learning_rate": 0.0003686021575445624,
      "loss": 2.1803,
      "step": 33987
    },
    {
      "epoch": 2.9242020132495914,
      "grad_norm": 6.53125,
      "learning_rate": 0.0003685760674423435,
      "loss": 0.9745,
      "step": 33988
    },
    {
      "epoch": 2.924288049556913,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00036854997772451417,
      "loss": 1.0745,
      "step": 33989
    },
    {
      "epoch": 2.9243740858642346,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00036852388839115035,
      "loss": 1.2123,
      "step": 33990
    },
    {
      "epoch": 2.9244601221715563,
      "grad_norm": 103.0,
      "learning_rate": 0.0003684977994423288,
      "loss": 1.1453,
      "step": 33991
    },
    {
      "epoch": 2.9245461584788783,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003684717108781256,
      "loss": 0.7838,
      "step": 33992
    },
    {
      "epoch": 2.9246321947861995,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003684456226986169,
      "loss": 1.2302,
      "step": 33993
    },
    {
      "epoch": 2.9247182310935216,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00036841953490387937,
      "loss": 0.7078,
      "step": 33994
    },
    {
      "epoch": 2.924804267400843,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00036839344749398903,
      "loss": 0.6867,
      "step": 33995
    },
    {
      "epoch": 2.924890303708165,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00036836736046902223,
      "loss": 1.2367,
      "step": 33996
    },
    {
      "epoch": 2.9249763400154865,
      "grad_norm": 12.5,
      "learning_rate": 0.00036834127382905524,
      "loss": 1.529,
      "step": 33997
    },
    {
      "epoch": 2.925062376322808,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00036831518757416456,
      "loss": 0.7545,
      "step": 33998
    },
    {
      "epoch": 2.92514841263013,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003682891017044263,
      "loss": 1.237,
      "step": 33999
    },
    {
      "epoch": 2.9252344489374518,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003682630162199167,
      "loss": 0.65,
      "step": 34000
    },
    {
      "epoch": 2.9253204852447734,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00036823693112071225,
      "loss": 1.3227,
      "step": 34001
    },
    {
      "epoch": 2.925406521552095,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003682108464068892,
      "loss": 1.1694,
      "step": 34002
    },
    {
      "epoch": 2.9254925578594166,
      "grad_norm": 1.09375,
      "learning_rate": 0.00036818476207852374,
      "loss": 0.5902,
      "step": 34003
    },
    {
      "epoch": 2.9255785941667383,
      "grad_norm": 1.515625,
      "learning_rate": 0.000368158678135692,
      "loss": 1.1726,
      "step": 34004
    },
    {
      "epoch": 2.92566463047406,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00036813259457847057,
      "loss": 0.9126,
      "step": 34005
    },
    {
      "epoch": 2.925750666781382,
      "grad_norm": 2.5625,
      "learning_rate": 0.00036810651140693583,
      "loss": 1.5188,
      "step": 34006
    },
    {
      "epoch": 2.9258367030887036,
      "grad_norm": 1.171875,
      "learning_rate": 0.00036808042862116386,
      "loss": 1.2804,
      "step": 34007
    },
    {
      "epoch": 2.925922739396025,
      "grad_norm": 1.140625,
      "learning_rate": 0.00036805434622123103,
      "loss": 1.3016,
      "step": 34008
    },
    {
      "epoch": 2.926008775703347,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0003680282642072134,
      "loss": 1.0676,
      "step": 34009
    },
    {
      "epoch": 2.9260948120106685,
      "grad_norm": 1.0,
      "learning_rate": 0.0003680021825791876,
      "loss": 0.5652,
      "step": 34010
    },
    {
      "epoch": 2.92618084831799,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003679761013372297,
      "loss": 1.375,
      "step": 34011
    },
    {
      "epoch": 2.9262668846253117,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000367950020481416,
      "loss": 0.8584,
      "step": 34012
    },
    {
      "epoch": 2.9263529209326338,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003679239400118227,
      "loss": 1.1606,
      "step": 34013
    },
    {
      "epoch": 2.9264389572399554,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003678978599285264,
      "loss": 1.5315,
      "step": 34014
    },
    {
      "epoch": 2.926524993547277,
      "grad_norm": 1.1875,
      "learning_rate": 0.00036787178023160304,
      "loss": 1.2334,
      "step": 34015
    },
    {
      "epoch": 2.9266110298545986,
      "grad_norm": 1.03125,
      "learning_rate": 0.000367845700921129,
      "loss": 0.8144,
      "step": 34016
    },
    {
      "epoch": 2.9266970661619203,
      "grad_norm": 1.046875,
      "learning_rate": 0.00036781962199718067,
      "loss": 1.7724,
      "step": 34017
    },
    {
      "epoch": 2.926783102469242,
      "grad_norm": 2.234375,
      "learning_rate": 0.0003677935434598342,
      "loss": 1.7553,
      "step": 34018
    },
    {
      "epoch": 2.9268691387765635,
      "grad_norm": 1.21875,
      "learning_rate": 0.00036776746530916594,
      "loss": 1.0823,
      "step": 34019
    },
    {
      "epoch": 2.9269551750838856,
      "grad_norm": 0.90625,
      "learning_rate": 0.000367741387545252,
      "loss": 0.7003,
      "step": 34020
    },
    {
      "epoch": 2.927041211391207,
      "grad_norm": 2.25,
      "learning_rate": 0.0003677153101681689,
      "loss": 0.8557,
      "step": 34021
    },
    {
      "epoch": 2.927127247698529,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003676892331779928,
      "loss": 2.0749,
      "step": 34022
    },
    {
      "epoch": 2.9272132840058505,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00036766315657479984,
      "loss": 0.8492,
      "step": 34023
    },
    {
      "epoch": 2.927299320313172,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003676370803586665,
      "loss": 2.0913,
      "step": 34024
    },
    {
      "epoch": 2.9273853566204937,
      "grad_norm": 1.046875,
      "learning_rate": 0.000367611004529669,
      "loss": 1.4993,
      "step": 34025
    },
    {
      "epoch": 2.9274713929278153,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00036758492908788357,
      "loss": 1.1905,
      "step": 34026
    },
    {
      "epoch": 2.9275574292351374,
      "grad_norm": 1.515625,
      "learning_rate": 0.0003675588540333863,
      "loss": 1.5397,
      "step": 34027
    },
    {
      "epoch": 2.927643465542459,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003675327793662537,
      "loss": 1.0502,
      "step": 34028
    },
    {
      "epoch": 2.9277295018497806,
      "grad_norm": 1.609375,
      "learning_rate": 0.000367506705086562,
      "loss": 0.9317,
      "step": 34029
    },
    {
      "epoch": 2.9278155381571023,
      "grad_norm": 4.625,
      "learning_rate": 0.00036748063119438745,
      "loss": 1.3595,
      "step": 34030
    },
    {
      "epoch": 2.927901574464424,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003674545576898062,
      "loss": 0.6511,
      "step": 34031
    },
    {
      "epoch": 2.9279876107717455,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003674284845728946,
      "loss": 0.9336,
      "step": 34032
    },
    {
      "epoch": 2.928073647079067,
      "grad_norm": 3.109375,
      "learning_rate": 0.00036740241184372895,
      "loss": 1.5908,
      "step": 34033
    },
    {
      "epoch": 2.928159683386389,
      "grad_norm": 31.875,
      "learning_rate": 0.00036737633950238534,
      "loss": 1.3125,
      "step": 34034
    },
    {
      "epoch": 2.928245719693711,
      "grad_norm": 43.0,
      "learning_rate": 0.0003673502675489403,
      "loss": 1.3143,
      "step": 34035
    },
    {
      "epoch": 2.9283317560010325,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0003673241959834699,
      "loss": 1.0707,
      "step": 34036
    },
    {
      "epoch": 2.928417792308354,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00036729812480605043,
      "loss": 0.7346,
      "step": 34037
    },
    {
      "epoch": 2.9285038286156757,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000367272054016758,
      "loss": 1.4099,
      "step": 34038
    },
    {
      "epoch": 2.9285898649229978,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0003672459836156691,
      "loss": 1.1221,
      "step": 34039
    },
    {
      "epoch": 2.928675901230319,
      "grad_norm": 8128.0,
      "learning_rate": 0.0003672199136028599,
      "loss": 1.1452,
      "step": 34040
    },
    {
      "epoch": 2.928761937537641,
      "grad_norm": 67.0,
      "learning_rate": 0.0003671938439784064,
      "loss": 1.4884,
      "step": 34041
    },
    {
      "epoch": 2.9288479738449626,
      "grad_norm": 2.265625,
      "learning_rate": 0.0003671677747423853,
      "loss": 1.6846,
      "step": 34042
    },
    {
      "epoch": 2.9289340101522843,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003671417058948725,
      "loss": 1.1831,
      "step": 34043
    },
    {
      "epoch": 2.929020046459606,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0003671156374359446,
      "loss": 1.9427,
      "step": 34044
    },
    {
      "epoch": 2.9291060827669275,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00036708956936567746,
      "loss": 2.1916,
      "step": 34045
    },
    {
      "epoch": 2.9291921190742496,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003670635016841474,
      "loss": 0.8639,
      "step": 34046
    },
    {
      "epoch": 2.9292781553815708,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003670374343914309,
      "loss": 1.3668,
      "step": 34047
    },
    {
      "epoch": 2.929364191688893,
      "grad_norm": 1.234375,
      "learning_rate": 0.000367011367487604,
      "loss": 1.2133,
      "step": 34048
    },
    {
      "epoch": 2.9294502279962145,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000366985300972743,
      "loss": 1.6632,
      "step": 34049
    },
    {
      "epoch": 2.929536264303536,
      "grad_norm": 10.3125,
      "learning_rate": 0.00036695923484692406,
      "loss": 2.0043,
      "step": 34050
    },
    {
      "epoch": 2.9296223006108577,
      "grad_norm": 8.3125,
      "learning_rate": 0.00036693316911022355,
      "loss": 1.621,
      "step": 34051
    },
    {
      "epoch": 2.9297083369181793,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00036690710376271763,
      "loss": 1.0958,
      "step": 34052
    },
    {
      "epoch": 2.9297943732255014,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003668810388044825,
      "loss": 1.653,
      "step": 34053
    },
    {
      "epoch": 2.9298804095328226,
      "grad_norm": 1.09375,
      "learning_rate": 0.00036685497423559444,
      "loss": 2.1181,
      "step": 34054
    },
    {
      "epoch": 2.9299664458401447,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003668289100561297,
      "loss": 1.2992,
      "step": 34055
    },
    {
      "epoch": 2.9300524821474663,
      "grad_norm": 3.953125,
      "learning_rate": 0.00036680284626616457,
      "loss": 1.0505,
      "step": 34056
    },
    {
      "epoch": 2.930138518454788,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0003667767828657751,
      "loss": 0.9941,
      "step": 34057
    },
    {
      "epoch": 2.9302245547621095,
      "grad_norm": 3.1875,
      "learning_rate": 0.0003667507198550377,
      "loss": 1.119,
      "step": 34058
    },
    {
      "epoch": 2.930310591069431,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003667246572340286,
      "loss": 0.7583,
      "step": 34059
    },
    {
      "epoch": 2.930396627376753,
      "grad_norm": 1.265625,
      "learning_rate": 0.00036669859500282386,
      "loss": 1.0077,
      "step": 34060
    },
    {
      "epoch": 2.930482663684075,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003666725331614997,
      "loss": 1.5924,
      "step": 34061
    },
    {
      "epoch": 2.9305686999913965,
      "grad_norm": 6.0,
      "learning_rate": 0.00036664647171013264,
      "loss": 1.3633,
      "step": 34062
    },
    {
      "epoch": 2.930654736298718,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00036662041064879867,
      "loss": 0.852,
      "step": 34063
    },
    {
      "epoch": 2.9307407726060397,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003665943499775739,
      "loss": 2.6213,
      "step": 34064
    },
    {
      "epoch": 2.9308268089133613,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003665682896965349,
      "loss": 2.0638,
      "step": 34065
    },
    {
      "epoch": 2.930912845220683,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003665422298057577,
      "loss": 0.7013,
      "step": 34066
    },
    {
      "epoch": 2.930998881528005,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00036651617030531846,
      "loss": 0.785,
      "step": 34067
    },
    {
      "epoch": 2.9310849178353267,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00036649011119529334,
      "loss": 1.5855,
      "step": 34068
    },
    {
      "epoch": 2.9311709541426483,
      "grad_norm": 9.875,
      "learning_rate": 0.00036646405247575887,
      "loss": 1.149,
      "step": 34069
    },
    {
      "epoch": 2.93125699044997,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00036643799414679104,
      "loss": 1.3898,
      "step": 34070
    },
    {
      "epoch": 2.9313430267572915,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000366411936208466,
      "loss": 1.6942,
      "step": 34071
    },
    {
      "epoch": 2.931429063064613,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00036638587866086016,
      "loss": 1.2603,
      "step": 34072
    },
    {
      "epoch": 2.9315150993719348,
      "grad_norm": 0.734375,
      "learning_rate": 0.00036635982150404954,
      "loss": 0.8391,
      "step": 34073
    },
    {
      "epoch": 2.931601135679257,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00036633376473811066,
      "loss": 1.6974,
      "step": 34074
    },
    {
      "epoch": 2.9316871719865785,
      "grad_norm": 4.96875,
      "learning_rate": 0.0003663077083631192,
      "loss": 1.2071,
      "step": 34075
    },
    {
      "epoch": 2.9317732082939,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00036628165237915193,
      "loss": 0.9481,
      "step": 34076
    },
    {
      "epoch": 2.9318592446012217,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00036625559678628484,
      "loss": 1.2198,
      "step": 34077
    },
    {
      "epoch": 2.9319452809085433,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00036622954158459383,
      "loss": 1.0602,
      "step": 34078
    },
    {
      "epoch": 2.932031317215865,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003662034867741556,
      "loss": 0.8145,
      "step": 34079
    },
    {
      "epoch": 2.9321173535231866,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003661774323550461,
      "loss": 1.363,
      "step": 34080
    },
    {
      "epoch": 2.9322033898305087,
      "grad_norm": 1.4375,
      "learning_rate": 0.0003661513783273417,
      "loss": 1.6604,
      "step": 34081
    },
    {
      "epoch": 2.9322894261378303,
      "grad_norm": 2.421875,
      "learning_rate": 0.0003661253246911184,
      "loss": 1.0395,
      "step": 34082
    },
    {
      "epoch": 2.932375462445152,
      "grad_norm": 1.0625,
      "learning_rate": 0.00036609927144645255,
      "loss": 1.1159,
      "step": 34083
    },
    {
      "epoch": 2.9324614987524735,
      "grad_norm": 1.6875,
      "learning_rate": 0.00036607321859342013,
      "loss": 0.8402,
      "step": 34084
    },
    {
      "epoch": 2.932547535059795,
      "grad_norm": 1.0625,
      "learning_rate": 0.00036604716613209765,
      "loss": 0.931,
      "step": 34085
    },
    {
      "epoch": 2.9326335713671168,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003660211140625612,
      "loss": 1.6184,
      "step": 34086
    },
    {
      "epoch": 2.9327196076744384,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003659950623848868,
      "loss": 0.9975,
      "step": 34087
    },
    {
      "epoch": 2.9328056439817605,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003659690110991508,
      "loss": 1.4566,
      "step": 34088
    },
    {
      "epoch": 2.932891680289082,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003659429602054295,
      "loss": 0.9077,
      "step": 34089
    },
    {
      "epoch": 2.9329777165964037,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003659169097037989,
      "loss": 1.7177,
      "step": 34090
    },
    {
      "epoch": 2.9330637529037253,
      "grad_norm": 1.125,
      "learning_rate": 0.0003658908595943352,
      "loss": 1.0497,
      "step": 34091
    },
    {
      "epoch": 2.933149789211047,
      "grad_norm": 33.25,
      "learning_rate": 0.0003658648098771147,
      "loss": 1.6022,
      "step": 34092
    },
    {
      "epoch": 2.9332358255183686,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00036583876055221355,
      "loss": 0.9709,
      "step": 34093
    },
    {
      "epoch": 2.93332186182569,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00036581271161970785,
      "loss": 1.7354,
      "step": 34094
    },
    {
      "epoch": 2.9334078981330123,
      "grad_norm": 1.328125,
      "learning_rate": 0.00036578666307967396,
      "loss": 2.0563,
      "step": 34095
    },
    {
      "epoch": 2.933493934440334,
      "grad_norm": 1.390625,
      "learning_rate": 0.000365760614932188,
      "loss": 1.5778,
      "step": 34096
    },
    {
      "epoch": 2.9335799707476555,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003657345671773261,
      "loss": 1.5384,
      "step": 34097
    },
    {
      "epoch": 2.933666007054977,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003657085198151644,
      "loss": 1.2524,
      "step": 34098
    },
    {
      "epoch": 2.933752043362299,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003656824728457793,
      "loss": 1.0098,
      "step": 34099
    },
    {
      "epoch": 2.933838079669621,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003656564262692468,
      "loss": 1.7437,
      "step": 34100
    },
    {
      "epoch": 2.933924115976942,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003656303800856431,
      "loss": 1.5516,
      "step": 34101
    },
    {
      "epoch": 2.934010152284264,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003656043342950443,
      "loss": 1.2269,
      "step": 34102
    },
    {
      "epoch": 2.9340961885915857,
      "grad_norm": 22.375,
      "learning_rate": 0.00036557828889752686,
      "loss": 1.2814,
      "step": 34103
    },
    {
      "epoch": 2.9341822248989073,
      "grad_norm": 2.953125,
      "learning_rate": 0.00036555224389316673,
      "loss": 1.1245,
      "step": 34104
    },
    {
      "epoch": 2.934268261206229,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00036552619928204,
      "loss": 1.1105,
      "step": 34105
    },
    {
      "epoch": 2.9343542975135506,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003655001550642232,
      "loss": 1.4527,
      "step": 34106
    },
    {
      "epoch": 2.9344403338208727,
      "grad_norm": 1.40625,
      "learning_rate": 0.00036547411123979214,
      "loss": 1.3575,
      "step": 34107
    },
    {
      "epoch": 2.934526370128194,
      "grad_norm": 4.78125,
      "learning_rate": 0.0003654480678088232,
      "loss": 0.989,
      "step": 34108
    },
    {
      "epoch": 2.934612406435516,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003654220247713924,
      "loss": 1.0559,
      "step": 34109
    },
    {
      "epoch": 2.9346984427428375,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00036539598212757617,
      "loss": 1.3743,
      "step": 34110
    },
    {
      "epoch": 2.934784479050159,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003653699398774505,
      "loss": 1.0383,
      "step": 34111
    },
    {
      "epoch": 2.934870515357481,
      "grad_norm": 5.0,
      "learning_rate": 0.00036534389802109137,
      "loss": 0.7631,
      "step": 34112
    },
    {
      "epoch": 2.9349565516648024,
      "grad_norm": 8.1875,
      "learning_rate": 0.0003653178565585753,
      "loss": 0.9659,
      "step": 34113
    },
    {
      "epoch": 2.9350425879721245,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00036529181548997834,
      "loss": 1.1854,
      "step": 34114
    },
    {
      "epoch": 2.9351286242794457,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003652657748153764,
      "loss": 1.5909,
      "step": 34115
    },
    {
      "epoch": 2.9352146605867677,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00036523973453484604,
      "loss": 0.9936,
      "step": 34116
    },
    {
      "epoch": 2.9353006968940893,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003652136946484633,
      "loss": 0.5862,
      "step": 34117
    },
    {
      "epoch": 2.935386733201411,
      "grad_norm": 1.359375,
      "learning_rate": 0.00036518765515630415,
      "loss": 1.0274,
      "step": 34118
    },
    {
      "epoch": 2.9354727695087326,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00036516161605844495,
      "loss": 0.7079,
      "step": 34119
    },
    {
      "epoch": 2.935558805816054,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00036513557735496185,
      "loss": 1.5881,
      "step": 34120
    },
    {
      "epoch": 2.9356448421233763,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003651095390459308,
      "loss": 0.5997,
      "step": 34121
    },
    {
      "epoch": 2.935730878430698,
      "grad_norm": 27.25,
      "learning_rate": 0.0003650835011314283,
      "loss": 1.37,
      "step": 34122
    },
    {
      "epoch": 2.9358169147380195,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00036505746361153025,
      "loss": 1.0511,
      "step": 34123
    },
    {
      "epoch": 2.935902951045341,
      "grad_norm": 1.5625,
      "learning_rate": 0.00036503142648631294,
      "loss": 1.1311,
      "step": 34124
    },
    {
      "epoch": 2.935988987352663,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003650053897558523,
      "loss": 1.6322,
      "step": 34125
    },
    {
      "epoch": 2.9360750236599844,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00036497935342022476,
      "loss": 1.7342,
      "step": 34126
    },
    {
      "epoch": 2.936161059967306,
      "grad_norm": 1.484375,
      "learning_rate": 0.00036495331747950633,
      "loss": 1.1584,
      "step": 34127
    },
    {
      "epoch": 2.936247096274628,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003649272819337731,
      "loss": 0.8355,
      "step": 34128
    },
    {
      "epoch": 2.9363331325819497,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00036490124678310144,
      "loss": 1.1043,
      "step": 34129
    },
    {
      "epoch": 2.9364191688892713,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00036487521202756734,
      "loss": 1.7032,
      "step": 34130
    },
    {
      "epoch": 2.936505205196593,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003648491776672469,
      "loss": 0.9434,
      "step": 34131
    },
    {
      "epoch": 2.9365912415039146,
      "grad_norm": 2.25,
      "learning_rate": 0.0003648231437022163,
      "loss": 1.2384,
      "step": 34132
    },
    {
      "epoch": 2.936677277811236,
      "grad_norm": 0.8125,
      "learning_rate": 0.00036479711013255177,
      "loss": 0.9991,
      "step": 34133
    },
    {
      "epoch": 2.936763314118558,
      "grad_norm": 2.75,
      "learning_rate": 0.0003647710769583295,
      "loss": 1.0869,
      "step": 34134
    },
    {
      "epoch": 2.93684935042588,
      "grad_norm": 1.4375,
      "learning_rate": 0.00036474504417962527,
      "loss": 1.7,
      "step": 34135
    },
    {
      "epoch": 2.9369353867332015,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003647190117965157,
      "loss": 1.7383,
      "step": 34136
    },
    {
      "epoch": 2.937021423040523,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00036469297980907666,
      "loss": 1.5804,
      "step": 34137
    },
    {
      "epoch": 2.937107459347845,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003646669482173843,
      "loss": 1.0395,
      "step": 34138
    },
    {
      "epoch": 2.9371934956551664,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003646409170215148,
      "loss": 1.372,
      "step": 34139
    },
    {
      "epoch": 2.937279531962488,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003646148862215443,
      "loss": 0.8003,
      "step": 34140
    },
    {
      "epoch": 2.9373655682698097,
      "grad_norm": 592.0,
      "learning_rate": 0.000364588855817549,
      "loss": 1.1114,
      "step": 34141
    },
    {
      "epoch": 2.9374516045771317,
      "grad_norm": 0.984375,
      "learning_rate": 0.00036456282580960486,
      "loss": 1.4763,
      "step": 34142
    },
    {
      "epoch": 2.9375376408844533,
      "grad_norm": 0.796875,
      "learning_rate": 0.00036453679619778814,
      "loss": 1.3763,
      "step": 34143
    },
    {
      "epoch": 2.937623677191775,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00036451076698217495,
      "loss": 1.2404,
      "step": 34144
    },
    {
      "epoch": 2.9377097134990966,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00036448473816284144,
      "loss": 0.8205,
      "step": 34145
    },
    {
      "epoch": 2.937795749806418,
      "grad_norm": 16.0,
      "learning_rate": 0.0003644587097398636,
      "loss": 0.7386,
      "step": 34146
    },
    {
      "epoch": 2.93788178611374,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003644326817133178,
      "loss": 1.0161,
      "step": 34147
    },
    {
      "epoch": 2.9379678224210615,
      "grad_norm": 1.03125,
      "learning_rate": 0.00036440665408328,
      "loss": 0.9515,
      "step": 34148
    },
    {
      "epoch": 2.9380538587283835,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003643806268498264,
      "loss": 1.6656,
      "step": 34149
    },
    {
      "epoch": 2.938139895035705,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000364354600013033,
      "loss": 1.7255,
      "step": 34150
    },
    {
      "epoch": 2.938225931343027,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003643285735729761,
      "loss": 1.2107,
      "step": 34151
    },
    {
      "epoch": 2.9383119676503484,
      "grad_norm": 1.0625,
      "learning_rate": 0.00036430254752973157,
      "loss": 0.8134,
      "step": 34152
    },
    {
      "epoch": 2.93839800395767,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00036427652188337584,
      "loss": 1.107,
      "step": 34153
    },
    {
      "epoch": 2.938484040264992,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003642504966339849,
      "loss": 2.0048,
      "step": 34154
    },
    {
      "epoch": 2.9385700765723133,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00036422447178163477,
      "loss": 0.6211,
      "step": 34155
    },
    {
      "epoch": 2.9386561128796354,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003641984473264017,
      "loss": 0.929,
      "step": 34156
    },
    {
      "epoch": 2.938742149186957,
      "grad_norm": 5.0625,
      "learning_rate": 0.0003641724232683619,
      "loss": 1.5246,
      "step": 34157
    },
    {
      "epoch": 2.9388281854942786,
      "grad_norm": 106.5,
      "learning_rate": 0.0003641463996075911,
      "loss": 1.7693,
      "step": 34158
    },
    {
      "epoch": 2.9389142218016002,
      "grad_norm": 14.75,
      "learning_rate": 0.00036412037634416585,
      "loss": 1.2373,
      "step": 34159
    },
    {
      "epoch": 2.939000258108922,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000364094353478162,
      "loss": 0.8489,
      "step": 34160
    },
    {
      "epoch": 2.939086294416244,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00036406833100965577,
      "loss": 0.8052,
      "step": 34161
    },
    {
      "epoch": 2.939172330723565,
      "grad_norm": 1.15625,
      "learning_rate": 0.00036404230893872315,
      "loss": 1.431,
      "step": 34162
    },
    {
      "epoch": 2.939258367030887,
      "grad_norm": 9.8125,
      "learning_rate": 0.0003640162872654404,
      "loss": 0.9542,
      "step": 34163
    },
    {
      "epoch": 2.939344403338209,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003639902659898836,
      "loss": 1.1329,
      "step": 34164
    },
    {
      "epoch": 2.9394304396455304,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00036396424511212877,
      "loss": 0.971,
      "step": 34165
    },
    {
      "epoch": 2.939516475952852,
      "grad_norm": 0.875,
      "learning_rate": 0.000363938224632252,
      "loss": 0.8491,
      "step": 34166
    },
    {
      "epoch": 2.9396025122601737,
      "grad_norm": 6.75,
      "learning_rate": 0.00036391220455032956,
      "loss": 0.7167,
      "step": 34167
    },
    {
      "epoch": 2.9396885485674957,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00036388618486643743,
      "loss": 1.6828,
      "step": 34168
    },
    {
      "epoch": 2.939774584874817,
      "grad_norm": 1.109375,
      "learning_rate": 0.00036386016558065165,
      "loss": 1.2167,
      "step": 34169
    },
    {
      "epoch": 2.939860621182139,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0003638341466930485,
      "loss": 0.6094,
      "step": 34170
    },
    {
      "epoch": 2.9399466574894606,
      "grad_norm": 1.15625,
      "learning_rate": 0.000363808128203704,
      "loss": 0.6881,
      "step": 34171
    },
    {
      "epoch": 2.9400326937967822,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00036378211011269424,
      "loss": 1.1226,
      "step": 34172
    },
    {
      "epoch": 2.940118730104104,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003637560924200952,
      "loss": 0.8054,
      "step": 34173
    },
    {
      "epoch": 2.9402047664114255,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003637300751259832,
      "loss": 1.4049,
      "step": 34174
    },
    {
      "epoch": 2.9402908027187475,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00036370405823043427,
      "loss": 0.9562,
      "step": 34175
    },
    {
      "epoch": 2.940376839026069,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00036367804173352433,
      "loss": 1.7453,
      "step": 34176
    },
    {
      "epoch": 2.940462875333391,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00036365202563532975,
      "loss": 1.3449,
      "step": 34177
    },
    {
      "epoch": 2.9405489116407124,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00036362600993592635,
      "loss": 1.3593,
      "step": 34178
    },
    {
      "epoch": 2.940634947948034,
      "grad_norm": 1.8125,
      "learning_rate": 0.0003635999946353905,
      "loss": 0.7895,
      "step": 34179
    },
    {
      "epoch": 2.9407209842553557,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003635739797337979,
      "loss": 1.0256,
      "step": 34180
    },
    {
      "epoch": 2.9408070205626773,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000363547965231225,
      "loss": 1.6355,
      "step": 34181
    },
    {
      "epoch": 2.9408930568699994,
      "grad_norm": 1.171875,
      "learning_rate": 0.00036352195112774786,
      "loss": 0.9833,
      "step": 34182
    },
    {
      "epoch": 2.940979093177321,
      "grad_norm": 2.4375,
      "learning_rate": 0.0003634959374234424,
      "loss": 1.3551,
      "step": 34183
    },
    {
      "epoch": 2.9410651294846426,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003634699241183846,
      "loss": 0.8601,
      "step": 34184
    },
    {
      "epoch": 2.9411511657919642,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000363443911212651,
      "loss": 0.5686,
      "step": 34185
    },
    {
      "epoch": 2.941237202099286,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00036341789870631726,
      "loss": 1.5479,
      "step": 34186
    },
    {
      "epoch": 2.9413232384066075,
      "grad_norm": 0.640625,
      "learning_rate": 0.00036339188659945956,
      "loss": 0.6421,
      "step": 34187
    },
    {
      "epoch": 2.941409274713929,
      "grad_norm": 30.5,
      "learning_rate": 0.000363365874892154,
      "loss": 1.2338,
      "step": 34188
    },
    {
      "epoch": 2.941495311021251,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003633398635844767,
      "loss": 0.8342,
      "step": 34189
    },
    {
      "epoch": 2.941581347328573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003633138526765038,
      "loss": 0.8511,
      "step": 34190
    },
    {
      "epoch": 2.9416673836358944,
      "grad_norm": 1.9609375,
      "learning_rate": 0.00036328784216831133,
      "loss": 0.9881,
      "step": 34191
    },
    {
      "epoch": 2.941753419943216,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003632618320599752,
      "loss": 1.1186,
      "step": 34192
    },
    {
      "epoch": 2.9418394562505377,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003632358223515717,
      "loss": 1.8845,
      "step": 34193
    },
    {
      "epoch": 2.9419254925578593,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00036320981304317686,
      "loss": 0.9093,
      "step": 34194
    },
    {
      "epoch": 2.942011528865181,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0003631838041348667,
      "loss": 1.1425,
      "step": 34195
    },
    {
      "epoch": 2.942097565172503,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0003631577956267172,
      "loss": 1.0207,
      "step": 34196
    },
    {
      "epoch": 2.9421836014798246,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003631317875188046,
      "loss": 0.8489,
      "step": 34197
    },
    {
      "epoch": 2.9422696377871462,
      "grad_norm": 1.203125,
      "learning_rate": 0.00036310577981120494,
      "loss": 1.8023,
      "step": 34198
    },
    {
      "epoch": 2.942355674094468,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00036307977250399413,
      "loss": 1.3978,
      "step": 34199
    },
    {
      "epoch": 2.9424417104017895,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003630537655972486,
      "loss": 1.1147,
      "step": 34200
    },
    {
      "epoch": 2.942527746709111,
      "grad_norm": 1.5,
      "learning_rate": 0.00036302775909104396,
      "loss": 0.9199,
      "step": 34201
    },
    {
      "epoch": 2.9426137830164327,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00036300175298545654,
      "loss": 1.1354,
      "step": 34202
    },
    {
      "epoch": 2.942699819323755,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003629757472805623,
      "loss": 1.0727,
      "step": 34203
    },
    {
      "epoch": 2.9427858556310764,
      "grad_norm": 1.375,
      "learning_rate": 0.0003629497419764375,
      "loss": 1.0116,
      "step": 34204
    },
    {
      "epoch": 2.942871891938398,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00036292373707315796,
      "loss": 1.2775,
      "step": 34205
    },
    {
      "epoch": 2.9429579282457197,
      "grad_norm": 0.84375,
      "learning_rate": 0.00036289773257079986,
      "loss": 0.7999,
      "step": 34206
    },
    {
      "epoch": 2.9430439645530413,
      "grad_norm": 0.984375,
      "learning_rate": 0.00036287172846943903,
      "loss": 1.6885,
      "step": 34207
    },
    {
      "epoch": 2.943130000860363,
      "grad_norm": 10.4375,
      "learning_rate": 0.00036284572476915196,
      "loss": 0.5883,
      "step": 34208
    },
    {
      "epoch": 2.9432160371676845,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003628197214700144,
      "loss": 0.9718,
      "step": 34209
    },
    {
      "epoch": 2.9433020734750066,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003627937185721024,
      "loss": 1.7988,
      "step": 34210
    },
    {
      "epoch": 2.9433881097823282,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003627677160754922,
      "loss": 0.8883,
      "step": 34211
    },
    {
      "epoch": 2.94347414608965,
      "grad_norm": 1.1875,
      "learning_rate": 0.00036274171398025966,
      "loss": 1.2572,
      "step": 34212
    },
    {
      "epoch": 2.9435601823969715,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00036271571228648104,
      "loss": 1.3619,
      "step": 34213
    },
    {
      "epoch": 2.943646218704293,
      "grad_norm": 8.1875,
      "learning_rate": 0.0003626897109942321,
      "loss": 1.285,
      "step": 34214
    },
    {
      "epoch": 2.943732255011615,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003626637101035891,
      "loss": 0.825,
      "step": 34215
    },
    {
      "epoch": 2.9438182913189364,
      "grad_norm": 2.109375,
      "learning_rate": 0.00036263770961462804,
      "loss": 1.0818,
      "step": 34216
    },
    {
      "epoch": 2.9439043276262584,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000362611709527425,
      "loss": 0.9517,
      "step": 34217
    },
    {
      "epoch": 2.94399036393358,
      "grad_norm": 1.015625,
      "learning_rate": 0.00036258570984205597,
      "loss": 1.291,
      "step": 34218
    },
    {
      "epoch": 2.9440764002409017,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000362559710558597,
      "loss": 1.0578,
      "step": 34219
    },
    {
      "epoch": 2.9441624365482233,
      "grad_norm": 0.890625,
      "learning_rate": 0.00036253371167712425,
      "loss": 1.4126,
      "step": 34220
    },
    {
      "epoch": 2.944248472855545,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003625077131977135,
      "loss": 1.0579,
      "step": 34221
    },
    {
      "epoch": 2.944334509162867,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000362481715120441,
      "loss": 0.9036,
      "step": 34222
    },
    {
      "epoch": 2.944420545470188,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00036245571744538275,
      "loss": 1.1656,
      "step": 34223
    },
    {
      "epoch": 2.9445065817775102,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00036242972017261474,
      "loss": 1.1854,
      "step": 34224
    },
    {
      "epoch": 2.944592618084832,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00036240372330221296,
      "loss": 0.6315,
      "step": 34225
    },
    {
      "epoch": 2.9446786543921535,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0003623777268342536,
      "loss": 0.5577,
      "step": 34226
    },
    {
      "epoch": 2.944764690699475,
      "grad_norm": 1.125,
      "learning_rate": 0.00036235173076881266,
      "loss": 0.9432,
      "step": 34227
    },
    {
      "epoch": 2.9448507270067967,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003623257351059662,
      "loss": 1.0134,
      "step": 34228
    },
    {
      "epoch": 2.944936763314119,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003622997398457901,
      "loss": 1.307,
      "step": 34229
    },
    {
      "epoch": 2.94502279962144,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00036227374498836053,
      "loss": 0.992,
      "step": 34230
    },
    {
      "epoch": 2.945108835928762,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003622477505337535,
      "loss": 1.4261,
      "step": 34231
    },
    {
      "epoch": 2.9451948722360837,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003622217564820449,
      "loss": 0.7905,
      "step": 34232
    },
    {
      "epoch": 2.9452809085434053,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003621957628333109,
      "loss": 1.709,
      "step": 34233
    },
    {
      "epoch": 2.945366944850727,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00036216976958762753,
      "loss": 0.8133,
      "step": 34234
    },
    {
      "epoch": 2.9454529811580485,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003621437767450708,
      "loss": 1.2364,
      "step": 34235
    },
    {
      "epoch": 2.9455390174653706,
      "grad_norm": 0.96875,
      "learning_rate": 0.00036211778430571674,
      "loss": 0.9344,
      "step": 34236
    },
    {
      "epoch": 2.9456250537726922,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00036209179226964115,
      "loss": 0.9286,
      "step": 34237
    },
    {
      "epoch": 2.945711090080014,
      "grad_norm": 1.59375,
      "learning_rate": 0.0003620658006369204,
      "loss": 1.1373,
      "step": 34238
    },
    {
      "epoch": 2.9457971263873355,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00036203980940763036,
      "loss": 0.9791,
      "step": 34239
    },
    {
      "epoch": 2.945883162694657,
      "grad_norm": 0.953125,
      "learning_rate": 0.000362013818581847,
      "loss": 1.1742,
      "step": 34240
    },
    {
      "epoch": 2.9459691990019787,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003619878281596464,
      "loss": 0.9552,
      "step": 34241
    },
    {
      "epoch": 2.9460552353093004,
      "grad_norm": 0.953125,
      "learning_rate": 0.00036196183814110464,
      "loss": 0.8479,
      "step": 34242
    },
    {
      "epoch": 2.9461412716166224,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0003619358485262976,
      "loss": 1.0438,
      "step": 34243
    },
    {
      "epoch": 2.946227307923944,
      "grad_norm": 0.890625,
      "learning_rate": 0.00036190985931530126,
      "loss": 0.841,
      "step": 34244
    },
    {
      "epoch": 2.9463133442312657,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003618838705081918,
      "loss": 1.4603,
      "step": 34245
    },
    {
      "epoch": 2.9463993805385873,
      "grad_norm": 1.5,
      "learning_rate": 0.00036185788210504525,
      "loss": 1.341,
      "step": 34246
    },
    {
      "epoch": 2.946485416845909,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003618318941059373,
      "loss": 0.6417,
      "step": 34247
    },
    {
      "epoch": 2.9465714531532305,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00036180590651094436,
      "loss": 1.4676,
      "step": 34248
    },
    {
      "epoch": 2.946657489460552,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003617799193201422,
      "loss": 1.756,
      "step": 34249
    },
    {
      "epoch": 2.9467435257678742,
      "grad_norm": 1.03125,
      "learning_rate": 0.000361753932533607,
      "loss": 1.1516,
      "step": 34250
    },
    {
      "epoch": 2.946829562075196,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003617279461514145,
      "loss": 1.4983,
      "step": 34251
    },
    {
      "epoch": 2.9469155983825175,
      "grad_norm": 0.95703125,
      "learning_rate": 0.000361701960173641,
      "loss": 1.2499,
      "step": 34252
    },
    {
      "epoch": 2.947001634689839,
      "grad_norm": 1.015625,
      "learning_rate": 0.00036167597460036227,
      "loss": 1.0784,
      "step": 34253
    },
    {
      "epoch": 2.9470876709971607,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003616499894316545,
      "loss": 1.427,
      "step": 34254
    },
    {
      "epoch": 2.9471737073044824,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00036162400466759347,
      "loss": 1.0499,
      "step": 34255
    },
    {
      "epoch": 2.947259743611804,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00036159802030825543,
      "loss": 1.0531,
      "step": 34256
    },
    {
      "epoch": 2.947345779919126,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00036157203635371626,
      "loss": 0.8489,
      "step": 34257
    },
    {
      "epoch": 2.9474318162264477,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003615460528040519,
      "loss": 1.7415,
      "step": 34258
    },
    {
      "epoch": 2.9475178525337693,
      "grad_norm": 1.15625,
      "learning_rate": 0.00036152006965933846,
      "loss": 0.9962,
      "step": 34259
    },
    {
      "epoch": 2.947603888841091,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003614940869196519,
      "loss": 0.8036,
      "step": 34260
    },
    {
      "epoch": 2.9476899251484125,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003614681045850682,
      "loss": 0.6096,
      "step": 34261
    },
    {
      "epoch": 2.947775961455734,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00036144212265566315,
      "loss": 1.3408,
      "step": 34262
    },
    {
      "epoch": 2.947861997763056,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000361416141131513,
      "loss": 1.482,
      "step": 34263
    },
    {
      "epoch": 2.947948034070378,
      "grad_norm": 1.421875,
      "learning_rate": 0.00036139016001269386,
      "loss": 1.9747,
      "step": 34264
    },
    {
      "epoch": 2.9480340703776995,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00036136417929928145,
      "loss": 0.9417,
      "step": 34265
    },
    {
      "epoch": 2.948120106685021,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00036133819899135194,
      "loss": 1.3713,
      "step": 34266
    },
    {
      "epoch": 2.9482061429923427,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003613122190889811,
      "loss": 0.7589,
      "step": 34267
    },
    {
      "epoch": 2.9482921792996644,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003612862395922452,
      "loss": 0.9602,
      "step": 34268
    },
    {
      "epoch": 2.948378215606986,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00036126026050121995,
      "loss": 0.8386,
      "step": 34269
    },
    {
      "epoch": 2.9484642519143076,
      "grad_norm": 1.0,
      "learning_rate": 0.0003612342818159815,
      "loss": 1.2711,
      "step": 34270
    },
    {
      "epoch": 2.9485502882216297,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003612083035366057,
      "loss": 1.2305,
      "step": 34271
    },
    {
      "epoch": 2.9486363245289513,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003611823256631688,
      "loss": 1.1652,
      "step": 34272
    },
    {
      "epoch": 2.948722360836273,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003611563481957465,
      "loss": 1.222,
      "step": 34273
    },
    {
      "epoch": 2.9488083971435946,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003611303711344148,
      "loss": 1.0297,
      "step": 34274
    },
    {
      "epoch": 2.948894433450916,
      "grad_norm": 1.046875,
      "learning_rate": 0.00036110439447924983,
      "loss": 1.0799,
      "step": 34275
    },
    {
      "epoch": 2.9489804697582382,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0003610784182303275,
      "loss": 0.7137,
      "step": 34276
    },
    {
      "epoch": 2.9490665060655594,
      "grad_norm": 1.359375,
      "learning_rate": 0.00036105244238772384,
      "loss": 1.3822,
      "step": 34277
    },
    {
      "epoch": 2.9491525423728815,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003610264669515146,
      "loss": 1.4653,
      "step": 34278
    },
    {
      "epoch": 2.949238578680203,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00036100049192177604,
      "loss": 0.9809,
      "step": 34279
    },
    {
      "epoch": 2.9493246149875247,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000360974517298584,
      "loss": 1.2786,
      "step": 34280
    },
    {
      "epoch": 2.9494106512948464,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00036094854308201435,
      "loss": 0.7478,
      "step": 34281
    },
    {
      "epoch": 2.949496687602168,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00036092256927214327,
      "loss": 1.0211,
      "step": 34282
    },
    {
      "epoch": 2.94958272390949,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003608965958690467,
      "loss": 1.5315,
      "step": 34283
    },
    {
      "epoch": 2.9496687602168112,
      "grad_norm": 1.15625,
      "learning_rate": 0.00036087062287280044,
      "loss": 1.5433,
      "step": 34284
    },
    {
      "epoch": 2.9497547965241333,
      "grad_norm": 1.15625,
      "learning_rate": 0.00036084465028348047,
      "loss": 0.8525,
      "step": 34285
    },
    {
      "epoch": 2.949840832831455,
      "grad_norm": 1.171875,
      "learning_rate": 0.00036081867810116295,
      "loss": 1.5564,
      "step": 34286
    },
    {
      "epoch": 2.9499268691387766,
      "grad_norm": 8.875,
      "learning_rate": 0.00036079270632592366,
      "loss": 1.2717,
      "step": 34287
    },
    {
      "epoch": 2.950012905446098,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003607667349578386,
      "loss": 1.1406,
      "step": 34288
    },
    {
      "epoch": 2.95009894175342,
      "grad_norm": 1.125,
      "learning_rate": 0.00036074076399698386,
      "loss": 1.4862,
      "step": 34289
    },
    {
      "epoch": 2.950184978060742,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003607147934434353,
      "loss": 0.8246,
      "step": 34290
    },
    {
      "epoch": 2.950271014368063,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003606888232972688,
      "loss": 0.9774,
      "step": 34291
    },
    {
      "epoch": 2.950357050675385,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00036066285355856046,
      "loss": 0.7812,
      "step": 34292
    },
    {
      "epoch": 2.9504430869827067,
      "grad_norm": 1.671875,
      "learning_rate": 0.00036063688422738614,
      "loss": 1.5917,
      "step": 34293
    },
    {
      "epoch": 2.9505291232900284,
      "grad_norm": 1.1875,
      "learning_rate": 0.00036061091530382186,
      "loss": 0.7802,
      "step": 34294
    },
    {
      "epoch": 2.95061515959735,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003605849467879435,
      "loss": 1.0448,
      "step": 34295
    },
    {
      "epoch": 2.9507011959046716,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000360558978679827,
      "loss": 1.2434,
      "step": 34296
    },
    {
      "epoch": 2.9507872322119937,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003605330109795485,
      "loss": 1.0322,
      "step": 34297
    },
    {
      "epoch": 2.9508732685193153,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003605070436871838,
      "loss": 1.0518,
      "step": 34298
    },
    {
      "epoch": 2.950959304826637,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0003604810768028087,
      "loss": 1.2906,
      "step": 34299
    },
    {
      "epoch": 2.9510453411339586,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0003604551103264995,
      "loss": 0.761,
      "step": 34300
    },
    {
      "epoch": 2.95113137744128,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003604291442583318,
      "loss": 1.191,
      "step": 34301
    },
    {
      "epoch": 2.951217413748602,
      "grad_norm": 0.953125,
      "learning_rate": 0.00036040317859838177,
      "loss": 0.7342,
      "step": 34302
    },
    {
      "epoch": 2.9513034500559234,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003603772133467255,
      "loss": 1.9796,
      "step": 34303
    },
    {
      "epoch": 2.9513894863632455,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00036035124850343844,
      "loss": 1.4496,
      "step": 34304
    },
    {
      "epoch": 2.951475522670567,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003603252840685969,
      "loss": 1.3314,
      "step": 34305
    },
    {
      "epoch": 2.9515615589778887,
      "grad_norm": 0.796875,
      "learning_rate": 0.0003602993200422768,
      "loss": 1.1792,
      "step": 34306
    },
    {
      "epoch": 2.9516475952852104,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003602733564245541,
      "loss": 1.2565,
      "step": 34307
    },
    {
      "epoch": 2.951733631592532,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003602473932155045,
      "loss": 0.9641,
      "step": 34308
    },
    {
      "epoch": 2.9518196678998536,
      "grad_norm": 1.234375,
      "learning_rate": 0.00036022143041520416,
      "loss": 2.1534,
      "step": 34309
    },
    {
      "epoch": 2.9519057042071752,
      "grad_norm": 1.0,
      "learning_rate": 0.00036019546802372893,
      "loss": 1.3477,
      "step": 34310
    },
    {
      "epoch": 2.9519917405144973,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0003601695060411547,
      "loss": 1.5104,
      "step": 34311
    },
    {
      "epoch": 2.952077776821819,
      "grad_norm": 1.28125,
      "learning_rate": 0.00036014354446755757,
      "loss": 1.5598,
      "step": 34312
    },
    {
      "epoch": 2.9521638131291406,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00036011758330301344,
      "loss": 1.2246,
      "step": 34313
    },
    {
      "epoch": 2.952249849436462,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003600916225475981,
      "loss": 1.0778,
      "step": 34314
    },
    {
      "epoch": 2.952335885743784,
      "grad_norm": 1.25,
      "learning_rate": 0.0003600656622013875,
      "loss": 1.5245,
      "step": 34315
    },
    {
      "epoch": 2.9524219220511054,
      "grad_norm": 1.265625,
      "learning_rate": 0.00036003970226445767,
      "loss": 1.5569,
      "step": 34316
    },
    {
      "epoch": 2.952507958358427,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003600137427368845,
      "loss": 0.9812,
      "step": 34317
    },
    {
      "epoch": 2.952593994665749,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003599877836187439,
      "loss": 1.1758,
      "step": 34318
    },
    {
      "epoch": 2.9526800309730707,
      "grad_norm": 1.90625,
      "learning_rate": 0.0003599618249101117,
      "loss": 0.8842,
      "step": 34319
    },
    {
      "epoch": 2.9527660672803924,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000359935866611064,
      "loss": 0.8552,
      "step": 34320
    },
    {
      "epoch": 2.952852103587714,
      "grad_norm": 1.234375,
      "learning_rate": 0.00035990990872167674,
      "loss": 1.1663,
      "step": 34321
    },
    {
      "epoch": 2.9529381398950356,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003598839512420256,
      "loss": 0.5388,
      "step": 34322
    },
    {
      "epoch": 2.9530241762023572,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003598579941721868,
      "loss": 1.5405,
      "step": 34323
    },
    {
      "epoch": 2.953110212509679,
      "grad_norm": 0.90625,
      "learning_rate": 0.000359832037512236,
      "loss": 0.7842,
      "step": 34324
    },
    {
      "epoch": 2.953196248817001,
      "grad_norm": 2.453125,
      "learning_rate": 0.0003598060812622493,
      "loss": 1.1287,
      "step": 34325
    },
    {
      "epoch": 2.9532822851243226,
      "grad_norm": 10.875,
      "learning_rate": 0.00035978012542230245,
      "loss": 1.1816,
      "step": 34326
    },
    {
      "epoch": 2.953368321431644,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003597541699924716,
      "loss": 0.6622,
      "step": 34327
    },
    {
      "epoch": 2.953454357738966,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003597282149728324,
      "loss": 1.6695,
      "step": 34328
    },
    {
      "epoch": 2.9535403940462874,
      "grad_norm": 18.0,
      "learning_rate": 0.00035970226036346087,
      "loss": 1.3323,
      "step": 34329
    },
    {
      "epoch": 2.9536264303536095,
      "grad_norm": 1.015625,
      "learning_rate": 0.000359676306164433,
      "loss": 1.3173,
      "step": 34330
    },
    {
      "epoch": 2.9537124666609307,
      "grad_norm": 27.25,
      "learning_rate": 0.0003596503523758248,
      "loss": 1.4858,
      "step": 34331
    },
    {
      "epoch": 2.9537985029682527,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003596243989977119,
      "loss": 1.479,
      "step": 34332
    },
    {
      "epoch": 2.9538845392755744,
      "grad_norm": 1.140625,
      "learning_rate": 0.00035959844603017024,
      "loss": 1.0528,
      "step": 34333
    },
    {
      "epoch": 2.953970575582896,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00035957249347327587,
      "loss": 1.1874,
      "step": 34334
    },
    {
      "epoch": 2.9540566118902176,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00035954654132710475,
      "loss": 0.4077,
      "step": 34335
    },
    {
      "epoch": 2.9541426481975392,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00035952058959173237,
      "loss": 1.1196,
      "step": 34336
    },
    {
      "epoch": 2.9542286845048613,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00035949463826723515,
      "loss": 1.2138,
      "step": 34337
    },
    {
      "epoch": 2.9543147208121825,
      "grad_norm": 1.171875,
      "learning_rate": 0.00035946868735368877,
      "loss": 1.623,
      "step": 34338
    },
    {
      "epoch": 2.9544007571195046,
      "grad_norm": 2.203125,
      "learning_rate": 0.00035944273685116924,
      "loss": 0.7684,
      "step": 34339
    },
    {
      "epoch": 2.954486793426826,
      "grad_norm": 1.984375,
      "learning_rate": 0.00035941678675975233,
      "loss": 1.9819,
      "step": 34340
    },
    {
      "epoch": 2.954572829734148,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0003593908370795139,
      "loss": 2.0977,
      "step": 34341
    },
    {
      "epoch": 2.9546588660414694,
      "grad_norm": 2.171875,
      "learning_rate": 0.00035936488781052994,
      "loss": 1.5359,
      "step": 34342
    },
    {
      "epoch": 2.954744902348791,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003593389389528764,
      "loss": 1.638,
      "step": 34343
    },
    {
      "epoch": 2.954830938656113,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003593129905066291,
      "loss": 1.1173,
      "step": 34344
    },
    {
      "epoch": 2.9549169749634343,
      "grad_norm": 3.859375,
      "learning_rate": 0.0003592870424718638,
      "loss": 0.7823,
      "step": 34345
    },
    {
      "epoch": 2.9550030112707564,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00035926109484865675,
      "loss": 1.6176,
      "step": 34346
    },
    {
      "epoch": 2.955089047578078,
      "grad_norm": 1.0625,
      "learning_rate": 0.00035923514763708356,
      "loss": 1.2754,
      "step": 34347
    },
    {
      "epoch": 2.9551750838853996,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003592092008372202,
      "loss": 1.4818,
      "step": 34348
    },
    {
      "epoch": 2.9552611201927212,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003591832544491424,
      "loss": 1.4459,
      "step": 34349
    },
    {
      "epoch": 2.955347156500043,
      "grad_norm": 1.78125,
      "learning_rate": 0.0003591573084729263,
      "loss": 0.7893,
      "step": 34350
    },
    {
      "epoch": 2.955433192807365,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00035913136290864774,
      "loss": 1.102,
      "step": 34351
    },
    {
      "epoch": 2.9555192291146866,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003591054177563824,
      "loss": 1.7203,
      "step": 34352
    },
    {
      "epoch": 2.955605265422008,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003590794730162064,
      "loss": 0.5434,
      "step": 34353
    },
    {
      "epoch": 2.95569130172933,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003590535286881955,
      "loss": 1.2426,
      "step": 34354
    },
    {
      "epoch": 2.9557773380366514,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003590275847724257,
      "loss": 1.3654,
      "step": 34355
    },
    {
      "epoch": 2.955863374343973,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00035900164126897267,
      "loss": 0.7696,
      "step": 34356
    },
    {
      "epoch": 2.9559494106512947,
      "grad_norm": 1.640625,
      "learning_rate": 0.00035897569817791255,
      "loss": 1.079,
      "step": 34357
    },
    {
      "epoch": 2.9560354469586168,
      "grad_norm": 1.375,
      "learning_rate": 0.00035894975549932104,
      "loss": 0.9505,
      "step": 34358
    },
    {
      "epoch": 2.9561214832659384,
      "grad_norm": 2.015625,
      "learning_rate": 0.00035892381323327405,
      "loss": 0.8425,
      "step": 34359
    },
    {
      "epoch": 2.95620751957326,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00035889787137984735,
      "loss": 0.6302,
      "step": 34360
    },
    {
      "epoch": 2.9562935558805816,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003588719299391171,
      "loss": 2.3111,
      "step": 34361
    },
    {
      "epoch": 2.9563795921879032,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00035884598891115896,
      "loss": 1.2168,
      "step": 34362
    },
    {
      "epoch": 2.956465628495225,
      "grad_norm": 0.828125,
      "learning_rate": 0.00035882004829604873,
      "loss": 0.7265,
      "step": 34363
    },
    {
      "epoch": 2.9565516648025465,
      "grad_norm": 728.0,
      "learning_rate": 0.0003587941080938626,
      "loss": 0.9436,
      "step": 34364
    },
    {
      "epoch": 2.9566377011098686,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00035876816830467616,
      "loss": 0.3914,
      "step": 34365
    },
    {
      "epoch": 2.95672373741719,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003587422289285654,
      "loss": 1.5271,
      "step": 34366
    },
    {
      "epoch": 2.956809773724512,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000358716289965606,
      "loss": 0.7245,
      "step": 34367
    },
    {
      "epoch": 2.9568958100318334,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00035869035141587415,
      "loss": 1.2895,
      "step": 34368
    },
    {
      "epoch": 2.956981846339155,
      "grad_norm": 1.46875,
      "learning_rate": 0.0003586644132794454,
      "loss": 1.2375,
      "step": 34369
    },
    {
      "epoch": 2.9570678826464767,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003586384755563958,
      "loss": 0.9401,
      "step": 34370
    },
    {
      "epoch": 2.9571539189537983,
      "grad_norm": 3.375,
      "learning_rate": 0.0003586125382468012,
      "loss": 1.2143,
      "step": 34371
    },
    {
      "epoch": 2.9572399552611204,
      "grad_norm": 2.0,
      "learning_rate": 0.0003585866013507374,
      "loss": 1.6435,
      "step": 34372
    },
    {
      "epoch": 2.957325991568442,
      "grad_norm": 3.40625,
      "learning_rate": 0.00035856066486828007,
      "loss": 0.6197,
      "step": 34373
    },
    {
      "epoch": 2.9574120278757636,
      "grad_norm": 1.578125,
      "learning_rate": 0.0003585347287995056,
      "loss": 1.5548,
      "step": 34374
    },
    {
      "epoch": 2.9574980641830853,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003585087931444894,
      "loss": 2.2965,
      "step": 34375
    },
    {
      "epoch": 2.957584100490407,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003584828579033075,
      "loss": 0.9584,
      "step": 34376
    },
    {
      "epoch": 2.9576701367977285,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00035845692307603575,
      "loss": 1.5248,
      "step": 34377
    },
    {
      "epoch": 2.95775617310505,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00035843098866275,
      "loss": 0.7476,
      "step": 34378
    },
    {
      "epoch": 2.957842209412372,
      "grad_norm": 10.25,
      "learning_rate": 0.0003584050546635259,
      "loss": 1.4846,
      "step": 34379
    },
    {
      "epoch": 2.957928245719694,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003583791210784396,
      "loss": 1.6485,
      "step": 34380
    },
    {
      "epoch": 2.9580142820270154,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003583531879075669,
      "loss": 1.1547,
      "step": 34381
    },
    {
      "epoch": 2.958100318334337,
      "grad_norm": 1.1875,
      "learning_rate": 0.00035832725515098354,
      "loss": 1.3061,
      "step": 34382
    },
    {
      "epoch": 2.9581863546416587,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003583013228087653,
      "loss": 0.7797,
      "step": 34383
    },
    {
      "epoch": 2.9582723909489803,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003582753908809882,
      "loss": 1.4136,
      "step": 34384
    },
    {
      "epoch": 2.958358427256302,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00035824945936772804,
      "loss": 1.1574,
      "step": 34385
    },
    {
      "epoch": 2.958444463563624,
      "grad_norm": 5.90625,
      "learning_rate": 0.00035822352826906056,
      "loss": 0.6525,
      "step": 34386
    },
    {
      "epoch": 2.9585304998709456,
      "grad_norm": 28.625,
      "learning_rate": 0.00035819759758506177,
      "loss": 1.0971,
      "step": 34387
    },
    {
      "epoch": 2.9586165361782673,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00035817166731580744,
      "loss": 1.4809,
      "step": 34388
    },
    {
      "epoch": 2.958702572485589,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00035814573746137336,
      "loss": 1.9189,
      "step": 34389
    },
    {
      "epoch": 2.9587886087929105,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0003581198080218353,
      "loss": 0.8467,
      "step": 34390
    },
    {
      "epoch": 2.9588746451002326,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00035809387899726937,
      "loss": 1.5215,
      "step": 34391
    },
    {
      "epoch": 2.9589606814075537,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003580679503877512,
      "loss": 1.489,
      "step": 34392
    },
    {
      "epoch": 2.959046717714876,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00035804202219335657,
      "loss": 1.4689,
      "step": 34393
    },
    {
      "epoch": 2.9591327540221974,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0003580160944141616,
      "loss": 1.4176,
      "step": 34394
    },
    {
      "epoch": 2.959218790329519,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00035799016705024184,
      "loss": 1.0247,
      "step": 34395
    },
    {
      "epoch": 2.9593048266368407,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003579642401016733,
      "loss": 1.777,
      "step": 34396
    },
    {
      "epoch": 2.9593908629441623,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003579383135685316,
      "loss": 1.0157,
      "step": 34397
    },
    {
      "epoch": 2.9594768992514844,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003579123874508927,
      "loss": 1.4234,
      "step": 34398
    },
    {
      "epoch": 2.9595629355588056,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00035788646174883255,
      "loss": 1.137,
      "step": 34399
    },
    {
      "epoch": 2.9596489718661276,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003578605364624268,
      "loss": 1.1939,
      "step": 34400
    },
    {
      "epoch": 2.9597350081734493,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003578346115917513,
      "loss": 0.8648,
      "step": 34401
    },
    {
      "epoch": 2.959821044480771,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000357808687136882,
      "loss": 0.8824,
      "step": 34402
    },
    {
      "epoch": 2.9599070807880925,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003577827630978946,
      "loss": 1.6645,
      "step": 34403
    },
    {
      "epoch": 2.959993117095414,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00035775683947486493,
      "loss": 1.0503,
      "step": 34404
    },
    {
      "epoch": 2.960079153402736,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003577309162678689,
      "loss": 1.7869,
      "step": 34405
    },
    {
      "epoch": 2.9601651897100574,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003577049934769822,
      "loss": 1.3636,
      "step": 34406
    },
    {
      "epoch": 2.9602512260173794,
      "grad_norm": 2.8125,
      "learning_rate": 0.0003576790711022808,
      "loss": 1.055,
      "step": 34407
    },
    {
      "epoch": 2.960337262324701,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003576531491438403,
      "loss": 1.2215,
      "step": 34408
    },
    {
      "epoch": 2.9604232986320227,
      "grad_norm": 1.9375,
      "learning_rate": 0.00035762722760173673,
      "loss": 1.0824,
      "step": 34409
    },
    {
      "epoch": 2.9605093349393443,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003576013064760457,
      "loss": 0.8064,
      "step": 34410
    },
    {
      "epoch": 2.960595371246666,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003575753857668433,
      "loss": 0.6833,
      "step": 34411
    },
    {
      "epoch": 2.960681407553988,
      "grad_norm": 1.0,
      "learning_rate": 0.0003575494654742052,
      "loss": 1.8653,
      "step": 34412
    },
    {
      "epoch": 2.9607674438613096,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003575235455982071,
      "loss": 1.454,
      "step": 34413
    },
    {
      "epoch": 2.9608534801686313,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00035749762613892513,
      "loss": 1.4335,
      "step": 34414
    },
    {
      "epoch": 2.960939516475953,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00035747170709643473,
      "loss": 0.9842,
      "step": 34415
    },
    {
      "epoch": 2.9610255527832745,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003574457884708119,
      "loss": 1.703,
      "step": 34416
    },
    {
      "epoch": 2.961111589090596,
      "grad_norm": 1.46875,
      "learning_rate": 0.0003574198702621324,
      "loss": 1.1079,
      "step": 34417
    },
    {
      "epoch": 2.9611976253979178,
      "grad_norm": 1.21875,
      "learning_rate": 0.00035739395247047213,
      "loss": 1.4127,
      "step": 34418
    },
    {
      "epoch": 2.96128366170524,
      "grad_norm": 1.8984375,
      "learning_rate": 0.00035736803509590676,
      "loss": 1.0065,
      "step": 34419
    },
    {
      "epoch": 2.9613696980125614,
      "grad_norm": 1.0625,
      "learning_rate": 0.000357342118138512,
      "loss": 2.06,
      "step": 34420
    },
    {
      "epoch": 2.961455734319883,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000357316201598364,
      "loss": 1.0469,
      "step": 34421
    },
    {
      "epoch": 2.9615417706272047,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00035729028547553835,
      "loss": 1.3194,
      "step": 34422
    },
    {
      "epoch": 2.9616278069345263,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003572643697701108,
      "loss": 1.0018,
      "step": 34423
    },
    {
      "epoch": 2.961713843241848,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003572384544821572,
      "loss": 1.1336,
      "step": 34424
    },
    {
      "epoch": 2.9617998795491696,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003572125396117534,
      "loss": 1.0787,
      "step": 34425
    },
    {
      "epoch": 2.9618859158564916,
      "grad_norm": 1.375,
      "learning_rate": 0.00035718662515897516,
      "loss": 1.4929,
      "step": 34426
    },
    {
      "epoch": 2.9619719521638133,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00035716071112389813,
      "loss": 0.8862,
      "step": 34427
    },
    {
      "epoch": 2.962057988471135,
      "grad_norm": 3.40625,
      "learning_rate": 0.0003571347975065984,
      "loss": 1.0492,
      "step": 34428
    },
    {
      "epoch": 2.9621440247784565,
      "grad_norm": 3.296875,
      "learning_rate": 0.0003571088843071516,
      "loss": 1.2571,
      "step": 34429
    },
    {
      "epoch": 2.962230061085778,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0003570829715256335,
      "loss": 0.8579,
      "step": 34430
    },
    {
      "epoch": 2.9623160973930998,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00035705705916211974,
      "loss": 1.153,
      "step": 34431
    },
    {
      "epoch": 2.9624021337004214,
      "grad_norm": 4.21875,
      "learning_rate": 0.0003570311472166865,
      "loss": 1.0219,
      "step": 34432
    },
    {
      "epoch": 2.9624881700077434,
      "grad_norm": 1.03125,
      "learning_rate": 0.00035700523568940926,
      "loss": 1.9996,
      "step": 34433
    },
    {
      "epoch": 2.962574206315065,
      "grad_norm": 1.125,
      "learning_rate": 0.0003569793245803639,
      "loss": 1.2422,
      "step": 34434
    },
    {
      "epoch": 2.9626602426223867,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00035695341388962615,
      "loss": 1.4945,
      "step": 34435
    },
    {
      "epoch": 2.9627462789297083,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00035692750361727196,
      "loss": 0.8073,
      "step": 34436
    },
    {
      "epoch": 2.96283231523703,
      "grad_norm": 1.25,
      "learning_rate": 0.0003569015937633769,
      "loss": 1.7668,
      "step": 34437
    },
    {
      "epoch": 2.9629183515443516,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003568756843280169,
      "loss": 2.3944,
      "step": 34438
    },
    {
      "epoch": 2.963004387851673,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003568497753112676,
      "loss": 0.7861,
      "step": 34439
    },
    {
      "epoch": 2.9630904241589953,
      "grad_norm": 1.484375,
      "learning_rate": 0.000356823866713205,
      "loss": 1.235,
      "step": 34440
    },
    {
      "epoch": 2.963176460466317,
      "grad_norm": 12.5,
      "learning_rate": 0.0003567979585339046,
      "loss": 1.3974,
      "step": 34441
    },
    {
      "epoch": 2.9632624967736385,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00035677205077344246,
      "loss": 0.8081,
      "step": 34442
    },
    {
      "epoch": 2.96334853308096,
      "grad_norm": 2.40625,
      "learning_rate": 0.00035674614343189414,
      "loss": 1.5477,
      "step": 34443
    },
    {
      "epoch": 2.9634345693882818,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003567202365093355,
      "loss": 0.8861,
      "step": 34444
    },
    {
      "epoch": 2.9635206056956034,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003566943300058422,
      "loss": 1.3539,
      "step": 34445
    },
    {
      "epoch": 2.963606642002925,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003566684239214903,
      "loss": 1.1312,
      "step": 34446
    },
    {
      "epoch": 2.963692678310247,
      "grad_norm": 1.75,
      "learning_rate": 0.0003566425182563551,
      "loss": 1.2823,
      "step": 34447
    },
    {
      "epoch": 2.9637787146175687,
      "grad_norm": 2.65625,
      "learning_rate": 0.0003566166130105128,
      "loss": 0.705,
      "step": 34448
    },
    {
      "epoch": 2.9638647509248903,
      "grad_norm": 1.046875,
      "learning_rate": 0.00035659070818403913,
      "loss": 1.1884,
      "step": 34449
    },
    {
      "epoch": 2.963950787232212,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003565648037770096,
      "loss": 1.0646,
      "step": 34450
    },
    {
      "epoch": 2.9640368235395336,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0003565388997895003,
      "loss": 0.5579,
      "step": 34451
    },
    {
      "epoch": 2.9641228598468556,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00035651299622158674,
      "loss": 0.9006,
      "step": 34452
    },
    {
      "epoch": 2.964208896154177,
      "grad_norm": 4.25,
      "learning_rate": 0.0003564870930733448,
      "loss": 0.8273,
      "step": 34453
    },
    {
      "epoch": 2.964294932461499,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00035646119034485004,
      "loss": 1.6048,
      "step": 34454
    },
    {
      "epoch": 2.9643809687688205,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003564352880361785,
      "loss": 0.9582,
      "step": 34455
    },
    {
      "epoch": 2.964467005076142,
      "grad_norm": 3.15625,
      "learning_rate": 0.00035640938614740585,
      "loss": 1.1601,
      "step": 34456
    },
    {
      "epoch": 2.9645530413834638,
      "grad_norm": 4.6875,
      "learning_rate": 0.0003563834846786077,
      "loss": 1.6968,
      "step": 34457
    },
    {
      "epoch": 2.9646390776907854,
      "grad_norm": 1.296875,
      "learning_rate": 0.00035635758362986006,
      "loss": 1.5055,
      "step": 34458
    },
    {
      "epoch": 2.9647251139981075,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003563316830012385,
      "loss": 1.262,
      "step": 34459
    },
    {
      "epoch": 2.9648111503054286,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003563057827928189,
      "loss": 1.5703,
      "step": 34460
    },
    {
      "epoch": 2.9648971866127507,
      "grad_norm": 1.0625,
      "learning_rate": 0.00035627988300467674,
      "loss": 0.8081,
      "step": 34461
    },
    {
      "epoch": 2.9649832229200723,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003562539836368881,
      "loss": 0.8254,
      "step": 34462
    },
    {
      "epoch": 2.965069259227394,
      "grad_norm": 2.03125,
      "learning_rate": 0.0003562280846895286,
      "loss": 1.8478,
      "step": 34463
    },
    {
      "epoch": 2.9651552955347156,
      "grad_norm": 1.015625,
      "learning_rate": 0.00035620218616267394,
      "loss": 1.2965,
      "step": 34464
    },
    {
      "epoch": 2.965241331842037,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003561762880564,
      "loss": 1.8509,
      "step": 34465
    },
    {
      "epoch": 2.9653273681493593,
      "grad_norm": 8.625,
      "learning_rate": 0.0003561503903707825,
      "loss": 1.018,
      "step": 34466
    },
    {
      "epoch": 2.9654134044566804,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000356124493105897,
      "loss": 1.1868,
      "step": 34467
    },
    {
      "epoch": 2.9654994407640025,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003560985962618193,
      "loss": 1.1611,
      "step": 34468
    },
    {
      "epoch": 2.965585477071324,
      "grad_norm": 2.75,
      "learning_rate": 0.00035607269983862534,
      "loss": 1.6482,
      "step": 34469
    },
    {
      "epoch": 2.9656715133786458,
      "grad_norm": 41.0,
      "learning_rate": 0.0003560468038363908,
      "loss": 2.0368,
      "step": 34470
    },
    {
      "epoch": 2.9657575496859674,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003560209082551913,
      "loss": 0.7484,
      "step": 34471
    },
    {
      "epoch": 2.965843585993289,
      "grad_norm": 1.078125,
      "learning_rate": 0.00035599501309510254,
      "loss": 1.1417,
      "step": 34472
    },
    {
      "epoch": 2.965929622300611,
      "grad_norm": 0.75,
      "learning_rate": 0.00035596911835620047,
      "loss": 1.1018,
      "step": 34473
    },
    {
      "epoch": 2.9660156586079327,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003559432240385607,
      "loss": 1.2998,
      "step": 34474
    },
    {
      "epoch": 2.9661016949152543,
      "grad_norm": 1.078125,
      "learning_rate": 0.00035591733014225894,
      "loss": 1.4039,
      "step": 34475
    },
    {
      "epoch": 2.966187731222576,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000355891436667371,
      "loss": 0.6435,
      "step": 34476
    },
    {
      "epoch": 2.9662737675298976,
      "grad_norm": 1.234375,
      "learning_rate": 0.00035586554361397265,
      "loss": 1.2153,
      "step": 34477
    },
    {
      "epoch": 2.966359803837219,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00035583965098213955,
      "loss": 1.814,
      "step": 34478
    },
    {
      "epoch": 2.966445840144541,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00035581375877194725,
      "loss": 0.9199,
      "step": 34479
    },
    {
      "epoch": 2.966531876451863,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00035578786698347187,
      "loss": 1.1873,
      "step": 34480
    },
    {
      "epoch": 2.9666179127591845,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0003557619756167888,
      "loss": 1.4622,
      "step": 34481
    },
    {
      "epoch": 2.966703949066506,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003557360846719739,
      "loss": 0.7228,
      "step": 34482
    },
    {
      "epoch": 2.9667899853738278,
      "grad_norm": 1.03125,
      "learning_rate": 0.000355710194149103,
      "loss": 1.8872,
      "step": 34483
    },
    {
      "epoch": 2.9668760216811494,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00035568430404825147,
      "loss": 1.4239,
      "step": 34484
    },
    {
      "epoch": 2.966962057988471,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003556584143694955,
      "loss": 1.113,
      "step": 34485
    },
    {
      "epoch": 2.9670480942957926,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00035563252511291055,
      "loss": 0.7739,
      "step": 34486
    },
    {
      "epoch": 2.9671341306031147,
      "grad_norm": 7.875,
      "learning_rate": 0.0003556066362785724,
      "loss": 0.7854,
      "step": 34487
    },
    {
      "epoch": 2.9672201669104363,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003555807478665567,
      "loss": 1.4069,
      "step": 34488
    },
    {
      "epoch": 2.967306203217758,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00035555485987693935,
      "loss": 1.6678,
      "step": 34489
    },
    {
      "epoch": 2.9673922395250796,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00035552897230979586,
      "loss": 1.6622,
      "step": 34490
    },
    {
      "epoch": 2.967478275832401,
      "grad_norm": 7.375,
      "learning_rate": 0.000355503085165202,
      "loss": 1.1369,
      "step": 34491
    },
    {
      "epoch": 2.967564312139723,
      "grad_norm": 1.296875,
      "learning_rate": 0.00035547719844323356,
      "loss": 1.7967,
      "step": 34492
    },
    {
      "epoch": 2.9676503484470445,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00035545131214396623,
      "loss": 0.686,
      "step": 34493
    },
    {
      "epoch": 2.9677363847543665,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003554254262674758,
      "loss": 1.2827,
      "step": 34494
    },
    {
      "epoch": 2.967822421061688,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003553995408138376,
      "loss": 0.9895,
      "step": 34495
    },
    {
      "epoch": 2.9679084573690098,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003553736557831278,
      "loss": 1.0659,
      "step": 34496
    },
    {
      "epoch": 2.9679944936763314,
      "grad_norm": 1.90625,
      "learning_rate": 0.00035534777117542197,
      "loss": 1.2567,
      "step": 34497
    },
    {
      "epoch": 2.968080529983653,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003553218869907956,
      "loss": 1.7539,
      "step": 34498
    },
    {
      "epoch": 2.9681665662909746,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00035529600322932464,
      "loss": 0.9525,
      "step": 34499
    },
    {
      "epoch": 2.9682526025982963,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00035527011989108483,
      "loss": 1.9334,
      "step": 34500
    },
    {
      "epoch": 2.9683386389056183,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0003552442369761517,
      "loss": 2.0305,
      "step": 34501
    },
    {
      "epoch": 2.96842467521294,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000355218354484601,
      "loss": 1.7028,
      "step": 34502
    },
    {
      "epoch": 2.9685107115202616,
      "grad_norm": 1.84375,
      "learning_rate": 0.0003551924724165085,
      "loss": 1.1652,
      "step": 34503
    },
    {
      "epoch": 2.968596747827583,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003551665907719499,
      "loss": 1.0319,
      "step": 34504
    },
    {
      "epoch": 2.968682784134905,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0003551407095510007,
      "loss": 0.5012,
      "step": 34505
    },
    {
      "epoch": 2.968768820442227,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003551148287537369,
      "loss": 1.1685,
      "step": 34506
    },
    {
      "epoch": 2.968854856749548,
      "grad_norm": 33792.0,
      "learning_rate": 0.000355088948380234,
      "loss": 1.6615,
      "step": 34507
    },
    {
      "epoch": 2.96894089305687,
      "grad_norm": 5.125,
      "learning_rate": 0.0003550630684305677,
      "loss": 1.489,
      "step": 34508
    },
    {
      "epoch": 2.9690269293641918,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0003550371889048137,
      "loss": 0.6833,
      "step": 34509
    },
    {
      "epoch": 2.9691129656715134,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003550113098030479,
      "loss": 1.0695,
      "step": 34510
    },
    {
      "epoch": 2.969199001978835,
      "grad_norm": 5.34375,
      "learning_rate": 0.0003549854311253457,
      "loss": 1.3848,
      "step": 34511
    },
    {
      "epoch": 2.9692850382861566,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00035495955287178293,
      "loss": 1.9645,
      "step": 34512
    },
    {
      "epoch": 2.9693710745934787,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00035493367504243524,
      "loss": 0.5697,
      "step": 34513
    },
    {
      "epoch": 2.9694571109008,
      "grad_norm": 1.171875,
      "learning_rate": 0.00035490779763737836,
      "loss": 1.2779,
      "step": 34514
    },
    {
      "epoch": 2.969543147208122,
      "grad_norm": 5.53125,
      "learning_rate": 0.00035488192065668796,
      "loss": 0.9478,
      "step": 34515
    },
    {
      "epoch": 2.9696291835154436,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003548560441004396,
      "loss": 0.5231,
      "step": 34516
    },
    {
      "epoch": 2.969715219822765,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00035483016796870924,
      "loss": 0.9569,
      "step": 34517
    },
    {
      "epoch": 2.969801256130087,
      "grad_norm": 0.71875,
      "learning_rate": 0.0003548042922615724,
      "loss": 0.5222,
      "step": 34518
    },
    {
      "epoch": 2.9698872924374085,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003547784169791047,
      "loss": 0.8112,
      "step": 34519
    },
    {
      "epoch": 2.9699733287447305,
      "grad_norm": 1.25,
      "learning_rate": 0.00035475254212138186,
      "loss": 0.9873,
      "step": 34520
    },
    {
      "epoch": 2.9700593650520517,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00035472666768847946,
      "loss": 0.4954,
      "step": 34521
    },
    {
      "epoch": 2.9701454013593738,
      "grad_norm": 1.234375,
      "learning_rate": 0.00035470079368047345,
      "loss": 1.4912,
      "step": 34522
    },
    {
      "epoch": 2.9702314376666954,
      "grad_norm": 106.5,
      "learning_rate": 0.0003546749200974394,
      "loss": 1.2436,
      "step": 34523
    },
    {
      "epoch": 2.970317473974017,
      "grad_norm": 4.84375,
      "learning_rate": 0.000354649046939453,
      "loss": 0.8156,
      "step": 34524
    },
    {
      "epoch": 2.9704035102813386,
      "grad_norm": 4.9375,
      "learning_rate": 0.00035462317420658973,
      "loss": 0.9141,
      "step": 34525
    },
    {
      "epoch": 2.9704895465886603,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00035459730189892546,
      "loss": 1.6545,
      "step": 34526
    },
    {
      "epoch": 2.9705755828959823,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00035457143001653587,
      "loss": 1.2775,
      "step": 34527
    },
    {
      "epoch": 2.970661619203304,
      "grad_norm": 1.4375,
      "learning_rate": 0.00035454555855949645,
      "loss": 1.2172,
      "step": 34528
    },
    {
      "epoch": 2.9707476555106256,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000354519687527883,
      "loss": 1.8521,
      "step": 34529
    },
    {
      "epoch": 2.970833691817947,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003544938169217712,
      "loss": 1.2203,
      "step": 34530
    },
    {
      "epoch": 2.970919728125269,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00035446794674123675,
      "loss": 1.358,
      "step": 34531
    },
    {
      "epoch": 2.9710057644325905,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0003544420769863551,
      "loss": 1.5921,
      "step": 34532
    },
    {
      "epoch": 2.971091800739912,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003544162076572022,
      "loss": 1.3686,
      "step": 34533
    },
    {
      "epoch": 2.971177837047234,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003543903387538536,
      "loss": 1.5425,
      "step": 34534
    },
    {
      "epoch": 2.9712638733545558,
      "grad_norm": 210.0,
      "learning_rate": 0.00035436447027638486,
      "loss": 1.6739,
      "step": 34535
    },
    {
      "epoch": 2.9713499096618774,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003543386022248717,
      "loss": 0.79,
      "step": 34536
    },
    {
      "epoch": 2.971435945969199,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00035431273459938985,
      "loss": 0.6961,
      "step": 34537
    },
    {
      "epoch": 2.9715219822765206,
      "grad_norm": 4.40625,
      "learning_rate": 0.0003542868674000149,
      "loss": 1.0033,
      "step": 34538
    },
    {
      "epoch": 2.9716080185838423,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003542610006268224,
      "loss": 1.0349,
      "step": 34539
    },
    {
      "epoch": 2.971694054891164,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003542351342798883,
      "loss": 1.03,
      "step": 34540
    },
    {
      "epoch": 2.971780091198486,
      "grad_norm": 1.09375,
      "learning_rate": 0.00035420926835928805,
      "loss": 1.474,
      "step": 34541
    },
    {
      "epoch": 2.9718661275058076,
      "grad_norm": 1.0,
      "learning_rate": 0.0003541834028650974,
      "loss": 0.7998,
      "step": 34542
    },
    {
      "epoch": 2.971952163813129,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003541575377973918,
      "loss": 1.1436,
      "step": 34543
    },
    {
      "epoch": 2.972038200120451,
      "grad_norm": 0.828125,
      "learning_rate": 0.00035413167315624704,
      "loss": 0.6527,
      "step": 34544
    },
    {
      "epoch": 2.9721242364277725,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00035410580894173886,
      "loss": 1.6504,
      "step": 34545
    },
    {
      "epoch": 2.972210272735094,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003540799451539427,
      "loss": 0.9393,
      "step": 34546
    },
    {
      "epoch": 2.9722963090424157,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003540540817929344,
      "loss": 2.2653,
      "step": 34547
    },
    {
      "epoch": 2.9723823453497378,
      "grad_norm": 0.8125,
      "learning_rate": 0.00035402821885878954,
      "loss": 0.7205,
      "step": 34548
    },
    {
      "epoch": 2.9724683816570594,
      "grad_norm": 1.34375,
      "learning_rate": 0.00035400235635158384,
      "loss": 0.923,
      "step": 34549
    },
    {
      "epoch": 2.972554417964381,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003539764942713927,
      "loss": 1.4642,
      "step": 34550
    },
    {
      "epoch": 2.9726404542717026,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003539506326182919,
      "loss": 1.029,
      "step": 34551
    },
    {
      "epoch": 2.9727264905790243,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003539247713923572,
      "loss": 0.9871,
      "step": 34552
    },
    {
      "epoch": 2.972812526886346,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00035389891059366395,
      "loss": 1.4427,
      "step": 34553
    },
    {
      "epoch": 2.9728985631936675,
      "grad_norm": 1.15625,
      "learning_rate": 0.00035387305022228823,
      "loss": 1.0855,
      "step": 34554
    },
    {
      "epoch": 2.9729845995009896,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0003538471902783052,
      "loss": 0.959,
      "step": 34555
    },
    {
      "epoch": 2.973070635808311,
      "grad_norm": 2.625,
      "learning_rate": 0.0003538213307617909,
      "loss": 1.389,
      "step": 34556
    },
    {
      "epoch": 2.973156672115633,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003537954716728206,
      "loss": 0.8265,
      "step": 34557
    },
    {
      "epoch": 2.9732427084229545,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003537696130114701,
      "loss": 0.6766,
      "step": 34558
    },
    {
      "epoch": 2.973328744730276,
      "grad_norm": 1.015625,
      "learning_rate": 0.00035374375477781507,
      "loss": 1.1755,
      "step": 34559
    },
    {
      "epoch": 2.9734147810375977,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003537178969719312,
      "loss": 1.6923,
      "step": 34560
    },
    {
      "epoch": 2.9735008173449193,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00035369203959389404,
      "loss": 1.0206,
      "step": 34561
    },
    {
      "epoch": 2.9735868536522414,
      "grad_norm": 1.515625,
      "learning_rate": 0.00035366618264377915,
      "loss": 1.4127,
      "step": 34562
    },
    {
      "epoch": 2.973672889959563,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00035364032612166224,
      "loss": 0.9915,
      "step": 34563
    },
    {
      "epoch": 2.9737589262668847,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003536144700276189,
      "loss": 1.9573,
      "step": 34564
    },
    {
      "epoch": 2.9738449625742063,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00035358861436172485,
      "loss": 0.8571,
      "step": 34565
    },
    {
      "epoch": 2.973930998881528,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003535627591240554,
      "loss": 1.3981,
      "step": 34566
    },
    {
      "epoch": 2.97401703518885,
      "grad_norm": 0.859375,
      "learning_rate": 0.00035353690431468666,
      "loss": 1.0886,
      "step": 34567
    },
    {
      "epoch": 2.974103071496171,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003535110499336939,
      "loss": 0.6287,
      "step": 34568
    },
    {
      "epoch": 2.974189107803493,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00035348519598115276,
      "loss": 0.9739,
      "step": 34569
    },
    {
      "epoch": 2.974275144110815,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000353459342457139,
      "loss": 0.757,
      "step": 34570
    },
    {
      "epoch": 2.9743611804181365,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00035343348936172817,
      "loss": 1.3054,
      "step": 34571
    },
    {
      "epoch": 2.974447216725458,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00035340763669499585,
      "loss": 1.2007,
      "step": 34572
    },
    {
      "epoch": 2.9745332530327797,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003533817844570176,
      "loss": 0.7261,
      "step": 34573
    },
    {
      "epoch": 2.974619289340102,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003533559326478693,
      "loss": 1.3677,
      "step": 34574
    },
    {
      "epoch": 2.974705325647423,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003533300812676262,
      "loss": 1.0522,
      "step": 34575
    },
    {
      "epoch": 2.974791361954745,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00035330423031636416,
      "loss": 1.6268,
      "step": 34576
    },
    {
      "epoch": 2.9748773982620667,
      "grad_norm": 4.75,
      "learning_rate": 0.00035327837979415874,
      "loss": 0.5874,
      "step": 34577
    },
    {
      "epoch": 2.9749634345693883,
      "grad_norm": 0.875,
      "learning_rate": 0.00035325252970108554,
      "loss": 0.5602,
      "step": 34578
    },
    {
      "epoch": 2.97504947087671,
      "grad_norm": 1.25,
      "learning_rate": 0.00035322668003722016,
      "loss": 1.2372,
      "step": 34579
    },
    {
      "epoch": 2.9751355071840315,
      "grad_norm": 1.09375,
      "learning_rate": 0.00035320083080263797,
      "loss": 0.9038,
      "step": 34580
    },
    {
      "epoch": 2.9752215434913536,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00035317498199741507,
      "loss": 1.0225,
      "step": 34581
    },
    {
      "epoch": 2.9753075797986748,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003531491336216267,
      "loss": 0.9563,
      "step": 34582
    },
    {
      "epoch": 2.975393616105997,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00035312328567534856,
      "loss": 1.5224,
      "step": 34583
    },
    {
      "epoch": 2.9754796524133185,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00035309743815865614,
      "loss": 0.8956,
      "step": 34584
    },
    {
      "epoch": 2.97556568872064,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003530715910716253,
      "loss": 0.6572,
      "step": 34585
    },
    {
      "epoch": 2.9756517250279617,
      "grad_norm": 0.984375,
      "learning_rate": 0.00035304574441433147,
      "loss": 1.6207,
      "step": 34586
    },
    {
      "epoch": 2.9757377613352833,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003530198981868501,
      "loss": 1.7802,
      "step": 34587
    },
    {
      "epoch": 2.9758237976426054,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003529940523892571,
      "loss": 0.6278,
      "step": 34588
    },
    {
      "epoch": 2.975909833949927,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0003529682070216278,
      "loss": 1.7276,
      "step": 34589
    },
    {
      "epoch": 2.9759958702572487,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00035294236208403796,
      "loss": 1.0916,
      "step": 34590
    },
    {
      "epoch": 2.9760819065645703,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003529165175765631,
      "loss": 1.2537,
      "step": 34591
    },
    {
      "epoch": 2.976167942871892,
      "grad_norm": 0.78125,
      "learning_rate": 0.0003528906734992788,
      "loss": 0.632,
      "step": 34592
    },
    {
      "epoch": 2.9762539791792135,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003528648298522607,
      "loss": 1.1734,
      "step": 34593
    },
    {
      "epoch": 2.976340015486535,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003528389866355843,
      "loss": 0.8493,
      "step": 34594
    },
    {
      "epoch": 2.976426051793857,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003528131438493252,
      "loss": 1.4588,
      "step": 34595
    },
    {
      "epoch": 2.976512088101179,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000352787301493559,
      "loss": 1.1539,
      "step": 34596
    },
    {
      "epoch": 2.9765981244085005,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003527614595683615,
      "loss": 0.8792,
      "step": 34597
    },
    {
      "epoch": 2.976684160715822,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000352735618073808,
      "loss": 0.5887,
      "step": 34598
    },
    {
      "epoch": 2.9767701970231437,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00035270977700997414,
      "loss": 1.3031,
      "step": 34599
    },
    {
      "epoch": 2.9768562333304653,
      "grad_norm": 1.015625,
      "learning_rate": 0.00035268393637693564,
      "loss": 1.0902,
      "step": 34600
    },
    {
      "epoch": 2.976942269637787,
      "grad_norm": 0.765625,
      "learning_rate": 0.00035265809617476796,
      "loss": 1.0483,
      "step": 34601
    },
    {
      "epoch": 2.977028305945109,
      "grad_norm": 1.578125,
      "learning_rate": 0.0003526322564035467,
      "loss": 1.7108,
      "step": 34602
    },
    {
      "epoch": 2.9771143422524307,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0003526064170633473,
      "loss": 1.524,
      "step": 34603
    },
    {
      "epoch": 2.9772003785597523,
      "grad_norm": 1.0625,
      "learning_rate": 0.00035258057815424547,
      "loss": 0.9577,
      "step": 34604
    },
    {
      "epoch": 2.977286414867074,
      "grad_norm": 1.09375,
      "learning_rate": 0.000352554739676317,
      "loss": 0.9688,
      "step": 34605
    },
    {
      "epoch": 2.9773724511743955,
      "grad_norm": 0.8125,
      "learning_rate": 0.00035252890162963706,
      "loss": 0.5182,
      "step": 34606
    },
    {
      "epoch": 2.977458487481717,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00035250306401428134,
      "loss": 1.2488,
      "step": 34607
    },
    {
      "epoch": 2.9775445237890388,
      "grad_norm": 3.484375,
      "learning_rate": 0.00035247722683032565,
      "loss": 1.2857,
      "step": 34608
    },
    {
      "epoch": 2.977630560096361,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00035245139007784534,
      "loss": 1.7081,
      "step": 34609
    },
    {
      "epoch": 2.9777165964036825,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003524255537569159,
      "loss": 1.2463,
      "step": 34610
    },
    {
      "epoch": 2.977802632711004,
      "grad_norm": 1.96875,
      "learning_rate": 0.00035239971786761305,
      "loss": 0.8744,
      "step": 34611
    },
    {
      "epoch": 2.9778886690183257,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003523738824100125,
      "loss": 1.3162,
      "step": 34612
    },
    {
      "epoch": 2.9779747053256473,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00035234804738418945,
      "loss": 1.0133,
      "step": 34613
    },
    {
      "epoch": 2.978060741632969,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0003523222127902196,
      "loss": 0.7101,
      "step": 34614
    },
    {
      "epoch": 2.9781467779402906,
      "grad_norm": 1.234375,
      "learning_rate": 0.00035229637862817866,
      "loss": 1.0186,
      "step": 34615
    },
    {
      "epoch": 2.9782328142476127,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003522705448981421,
      "loss": 1.3209,
      "step": 34616
    },
    {
      "epoch": 2.9783188505549343,
      "grad_norm": 1.3125,
      "learning_rate": 0.00035224471160018534,
      "loss": 2.0438,
      "step": 34617
    },
    {
      "epoch": 2.978404886862256,
      "grad_norm": 3.078125,
      "learning_rate": 0.0003522188787343842,
      "loss": 0.8,
      "step": 34618
    },
    {
      "epoch": 2.9784909231695775,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00035219304630081403,
      "loss": 0.9718,
      "step": 34619
    },
    {
      "epoch": 2.978576959476899,
      "grad_norm": 1.140625,
      "learning_rate": 0.00035216721429955056,
      "loss": 1.8247,
      "step": 34620
    },
    {
      "epoch": 2.978662995784221,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003521413827306691,
      "loss": 1.319,
      "step": 34621
    },
    {
      "epoch": 2.9787490320915424,
      "grad_norm": 0.921875,
      "learning_rate": 0.00035211555159424534,
      "loss": 1.2052,
      "step": 34622
    },
    {
      "epoch": 2.9788350683988645,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00035208972089035493,
      "loss": 0.9281,
      "step": 34623
    },
    {
      "epoch": 2.978921104706186,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003520638906190733,
      "loss": 0.9564,
      "step": 34624
    },
    {
      "epoch": 2.9790071410135077,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003520380607804759,
      "loss": 0.6693,
      "step": 34625
    },
    {
      "epoch": 2.9790931773208293,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00035201223137463854,
      "loss": 0.5431,
      "step": 34626
    },
    {
      "epoch": 2.979179213628151,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003519864024016366,
      "loss": 1.161,
      "step": 34627
    },
    {
      "epoch": 2.979265249935473,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00035196057386154553,
      "loss": 1.1059,
      "step": 34628
    },
    {
      "epoch": 2.979351286242794,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003519347457544411,
      "loss": 0.9154,
      "step": 34629
    },
    {
      "epoch": 2.9794373225501163,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003519089180803987,
      "loss": 1.0028,
      "step": 34630
    },
    {
      "epoch": 2.979523358857438,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000351883090839494,
      "loss": 1.9019,
      "step": 34631
    },
    {
      "epoch": 2.9796093951647595,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0003518572640318022,
      "loss": 0.722,
      "step": 34632
    },
    {
      "epoch": 2.979695431472081,
      "grad_norm": 0.984375,
      "learning_rate": 0.00035183143765739913,
      "loss": 0.7005,
      "step": 34633
    },
    {
      "epoch": 2.979781467779403,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00035180561171636046,
      "loss": 1.004,
      "step": 34634
    },
    {
      "epoch": 2.979867504086725,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003517797862087615,
      "loss": 1.1653,
      "step": 34635
    },
    {
      "epoch": 2.979953540394046,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00035175396113467793,
      "loss": 0.5176,
      "step": 34636
    },
    {
      "epoch": 2.980039576701368,
      "grad_norm": 1.34375,
      "learning_rate": 0.000351728136494185,
      "loss": 1.4474,
      "step": 34637
    },
    {
      "epoch": 2.9801256130086897,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003517023122873586,
      "loss": 1.2493,
      "step": 34638
    },
    {
      "epoch": 2.9802116493160113,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00035167648851427405,
      "loss": 1.615,
      "step": 34639
    },
    {
      "epoch": 2.980297685623333,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003516506651750069,
      "loss": 1.8875,
      "step": 34640
    },
    {
      "epoch": 2.9803837219306546,
      "grad_norm": 1.125,
      "learning_rate": 0.0003516248422696327,
      "loss": 1.2613,
      "step": 34641
    },
    {
      "epoch": 2.9804697582379767,
      "grad_norm": 1.671875,
      "learning_rate": 0.00035159901979822704,
      "loss": 1.4041,
      "step": 34642
    },
    {
      "epoch": 2.980555794545298,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003515731977608654,
      "loss": 1.0426,
      "step": 34643
    },
    {
      "epoch": 2.98064183085262,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00035154737615762313,
      "loss": 1.1402,
      "step": 34644
    },
    {
      "epoch": 2.9807278671599415,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000351521554988576,
      "loss": 1.2793,
      "step": 34645
    },
    {
      "epoch": 2.980813903467263,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003514957342537995,
      "loss": 0.8254,
      "step": 34646
    },
    {
      "epoch": 2.980899939774585,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003514699139533691,
      "loss": 1.1095,
      "step": 34647
    },
    {
      "epoch": 2.9809859760819064,
      "grad_norm": 1.109375,
      "learning_rate": 0.00035144409408736015,
      "loss": 1.4057,
      "step": 34648
    },
    {
      "epoch": 2.9810720123892285,
      "grad_norm": 1.25,
      "learning_rate": 0.0003514182746558485,
      "loss": 1.5721,
      "step": 34649
    },
    {
      "epoch": 2.98115804869655,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003513924556589095,
      "loss": 1.2957,
      "step": 34650
    },
    {
      "epoch": 2.9812440850038717,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00035136663709661855,
      "loss": 1.4992,
      "step": 34651
    },
    {
      "epoch": 2.9813301213111933,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00035134081896905133,
      "loss": 1.0702,
      "step": 34652
    },
    {
      "epoch": 2.981416157618515,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003513150012762833,
      "loss": 1.3709,
      "step": 34653
    },
    {
      "epoch": 2.9815021939258366,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00035128918401839006,
      "loss": 1.059,
      "step": 34654
    },
    {
      "epoch": 2.981588230233158,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003512633671954469,
      "loss": 1.1989,
      "step": 34655
    },
    {
      "epoch": 2.9816742665404803,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003512375508075295,
      "loss": 1.7935,
      "step": 34656
    },
    {
      "epoch": 2.981760302847802,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003512117348547134,
      "loss": 1.6801,
      "step": 34657
    },
    {
      "epoch": 2.9818463391551235,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003511859193370739,
      "loss": 1.8417,
      "step": 34658
    },
    {
      "epoch": 2.981932375462445,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003511601042546868,
      "loss": 1.7545,
      "step": 34659
    },
    {
      "epoch": 2.982018411769767,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003511342896076274,
      "loss": 1.6712,
      "step": 34660
    },
    {
      "epoch": 2.9821044480770884,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003511084753959712,
      "loss": 1.3217,
      "step": 34661
    },
    {
      "epoch": 2.98219048438441,
      "grad_norm": 2.015625,
      "learning_rate": 0.00035108266161979374,
      "loss": 1.5414,
      "step": 34662
    },
    {
      "epoch": 2.982276520691732,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00035105684827917063,
      "loss": 0.8078,
      "step": 34663
    },
    {
      "epoch": 2.9823625569990537,
      "grad_norm": 1.46875,
      "learning_rate": 0.00035103103537417724,
      "loss": 1.1388,
      "step": 34664
    },
    {
      "epoch": 2.9824485933063754,
      "grad_norm": 0.828125,
      "learning_rate": 0.00035100522290488904,
      "loss": 0.5545,
      "step": 34665
    },
    {
      "epoch": 2.982534629613697,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00035097941087138164,
      "loss": 0.979,
      "step": 34666
    },
    {
      "epoch": 2.9826206659210186,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003509535992737305,
      "loss": 0.9203,
      "step": 34667
    },
    {
      "epoch": 2.9827067022283402,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003509277881120111,
      "loss": 0.9218,
      "step": 34668
    },
    {
      "epoch": 2.982792738535662,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003509019773862987,
      "loss": 0.6197,
      "step": 34669
    },
    {
      "epoch": 2.982878774842984,
      "grad_norm": 0.765625,
      "learning_rate": 0.0003508761670966692,
      "loss": 0.4442,
      "step": 34670
    },
    {
      "epoch": 2.9829648111503055,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003508503572431978,
      "loss": 0.7893,
      "step": 34671
    },
    {
      "epoch": 2.983050847457627,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003508245478259603,
      "loss": 0.6079,
      "step": 34672
    },
    {
      "epoch": 2.983136883764949,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003507987388450319,
      "loss": 1.9724,
      "step": 34673
    },
    {
      "epoch": 2.9832229200722704,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003507729303004881,
      "loss": 0.8315,
      "step": 34674
    },
    {
      "epoch": 2.983308956379592,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003507471221924045,
      "loss": 0.9241,
      "step": 34675
    },
    {
      "epoch": 2.9833949926869137,
      "grad_norm": 2.125,
      "learning_rate": 0.0003507213145208566,
      "loss": 1.3612,
      "step": 34676
    },
    {
      "epoch": 2.9834810289942357,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003506955072859198,
      "loss": 1.4071,
      "step": 34677
    },
    {
      "epoch": 2.9835670653015574,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00035066970048766956,
      "loss": 0.5433,
      "step": 34678
    },
    {
      "epoch": 2.983653101608879,
      "grad_norm": 1.09375,
      "learning_rate": 0.00035064389412618144,
      "loss": 1.493,
      "step": 34679
    },
    {
      "epoch": 2.9837391379162006,
      "grad_norm": 1.359375,
      "learning_rate": 0.00035061808820153085,
      "loss": 1.5227,
      "step": 34680
    },
    {
      "epoch": 2.9838251742235222,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0003505922827137933,
      "loss": 0.9017,
      "step": 34681
    },
    {
      "epoch": 2.9839112105308443,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003505664776630443,
      "loss": 0.8729,
      "step": 34682
    },
    {
      "epoch": 2.9839972468381655,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003505406730493593,
      "loss": 0.8703,
      "step": 34683
    },
    {
      "epoch": 2.9840832831454875,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003505148688728137,
      "loss": 0.7391,
      "step": 34684
    },
    {
      "epoch": 2.984169319452809,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000350489065133483,
      "loss": 0.5656,
      "step": 34685
    },
    {
      "epoch": 2.984255355760131,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00035046326183144283,
      "loss": 1.6064,
      "step": 34686
    },
    {
      "epoch": 2.9843413920674524,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003504374589667685,
      "loss": 0.9998,
      "step": 34687
    },
    {
      "epoch": 2.984427428374774,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00035041165653953554,
      "loss": 1.6757,
      "step": 34688
    },
    {
      "epoch": 2.984513464682096,
      "grad_norm": 0.875,
      "learning_rate": 0.00035038585454981924,
      "loss": 0.9313,
      "step": 34689
    },
    {
      "epoch": 2.9845995009894173,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00035036005299769534,
      "loss": 1.5811,
      "step": 34690
    },
    {
      "epoch": 2.9846855372967394,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003503342518832392,
      "loss": 1.052,
      "step": 34691
    },
    {
      "epoch": 2.984771573604061,
      "grad_norm": 1.046875,
      "learning_rate": 0.00035030845120652616,
      "loss": 0.8823,
      "step": 34692
    },
    {
      "epoch": 2.9848576099113826,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003502826509676319,
      "loss": 1.4184,
      "step": 34693
    },
    {
      "epoch": 2.9849436462187042,
      "grad_norm": 1.265625,
      "learning_rate": 0.00035025685116663177,
      "loss": 1.7874,
      "step": 34694
    },
    {
      "epoch": 2.985029682526026,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003502310518036012,
      "loss": 1.2092,
      "step": 34695
    },
    {
      "epoch": 2.985115718833348,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0003502052528786156,
      "loss": 0.8962,
      "step": 34696
    },
    {
      "epoch": 2.985201755140669,
      "grad_norm": 1.046875,
      "learning_rate": 0.00035017945439175057,
      "loss": 1.5661,
      "step": 34697
    },
    {
      "epoch": 2.985287791447991,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003501536563430815,
      "loss": 1.1979,
      "step": 34698
    },
    {
      "epoch": 2.985373827755313,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003501278587326837,
      "loss": 0.9342,
      "step": 34699
    },
    {
      "epoch": 2.9854598640626344,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000350102061560633,
      "loss": 1.3769,
      "step": 34700
    },
    {
      "epoch": 2.985545900369956,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0003500762648270045,
      "loss": 0.9843,
      "step": 34701
    },
    {
      "epoch": 2.9856319366772777,
      "grad_norm": 1.71875,
      "learning_rate": 0.00035005046853187386,
      "loss": 1.6871,
      "step": 34702
    },
    {
      "epoch": 2.9857179729845997,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0003500246726753162,
      "loss": 0.5681,
      "step": 34703
    },
    {
      "epoch": 2.9858040092919214,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003499988772574074,
      "loss": 0.7059,
      "step": 34704
    },
    {
      "epoch": 2.985890045599243,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00034997308227822265,
      "loss": 1.6565,
      "step": 34705
    },
    {
      "epoch": 2.9859760819065646,
      "grad_norm": 0.890625,
      "learning_rate": 0.00034994728773783733,
      "loss": 0.9321,
      "step": 34706
    },
    {
      "epoch": 2.9860621182138862,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003499214936363272,
      "loss": 1.4069,
      "step": 34707
    },
    {
      "epoch": 2.986148154521208,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00034989569997376734,
      "loss": 1.332,
      "step": 34708
    },
    {
      "epoch": 2.9862341908285295,
      "grad_norm": 1.109375,
      "learning_rate": 0.00034986990675023354,
      "loss": 0.8567,
      "step": 34709
    },
    {
      "epoch": 2.9863202271358515,
      "grad_norm": 1.609375,
      "learning_rate": 0.0003498441139658011,
      "loss": 1.374,
      "step": 34710
    },
    {
      "epoch": 2.986406263443173,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00034981832162054525,
      "loss": 2.1054,
      "step": 34711
    },
    {
      "epoch": 2.986492299750495,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003497925297145417,
      "loss": 1.8755,
      "step": 34712
    },
    {
      "epoch": 2.9865783360578164,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003497667382478658,
      "loss": 1.4945,
      "step": 34713
    },
    {
      "epoch": 2.986664372365138,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003497409472205929,
      "loss": 1.321,
      "step": 34714
    },
    {
      "epoch": 2.9867504086724597,
      "grad_norm": 1.171875,
      "learning_rate": 0.00034971515663279854,
      "loss": 1.6685,
      "step": 34715
    },
    {
      "epoch": 2.9868364449797813,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003496893664845581,
      "loss": 1.4868,
      "step": 34716
    },
    {
      "epoch": 2.9869224812871034,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003496635767759471,
      "loss": 0.7051,
      "step": 34717
    },
    {
      "epoch": 2.987008517594425,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0003496377875070409,
      "loss": 1.0926,
      "step": 34718
    },
    {
      "epoch": 2.9870945539017466,
      "grad_norm": 3.40625,
      "learning_rate": 0.00034961199867791483,
      "loss": 1.5148,
      "step": 34719
    },
    {
      "epoch": 2.9871805902090682,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00034958621028864445,
      "loss": 1.1026,
      "step": 34720
    },
    {
      "epoch": 2.98726662651639,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003495604223393052,
      "loss": 0.7927,
      "step": 34721
    },
    {
      "epoch": 2.9873526628237115,
      "grad_norm": 1.140625,
      "learning_rate": 0.00034953463482997237,
      "loss": 0.7239,
      "step": 34722
    },
    {
      "epoch": 2.987438699131033,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00034950884776072154,
      "loss": 0.697,
      "step": 34723
    },
    {
      "epoch": 2.987524735438355,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00034948306113162807,
      "loss": 1.6051,
      "step": 34724
    },
    {
      "epoch": 2.987610771745677,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00034945727494276735,
      "loss": 0.9564,
      "step": 34725
    },
    {
      "epoch": 2.9876968080529984,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00034943148919421474,
      "loss": 1.2347,
      "step": 34726
    },
    {
      "epoch": 2.98778284436032,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00034940570388604586,
      "loss": 0.8942,
      "step": 34727
    },
    {
      "epoch": 2.9878688806676417,
      "grad_norm": 0.875,
      "learning_rate": 0.00034937991901833596,
      "loss": 1.2516,
      "step": 34728
    },
    {
      "epoch": 2.9879549169749633,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003493541345911605,
      "loss": 0.8491,
      "step": 34729
    },
    {
      "epoch": 2.988040953282285,
      "grad_norm": 0.734375,
      "learning_rate": 0.00034932835060459493,
      "loss": 0.6921,
      "step": 34730
    },
    {
      "epoch": 2.988126989589607,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00034930256705871467,
      "loss": 2.2341,
      "step": 34731
    },
    {
      "epoch": 2.9882130258969286,
      "grad_norm": 1.28125,
      "learning_rate": 0.000349276783953595,
      "loss": 1.44,
      "step": 34732
    },
    {
      "epoch": 2.9882990622042502,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003492510012893114,
      "loss": 0.6646,
      "step": 34733
    },
    {
      "epoch": 2.988385098511572,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003492252190659394,
      "loss": 1.3146,
      "step": 34734
    },
    {
      "epoch": 2.9884711348188935,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0003491994372835543,
      "loss": 0.8124,
      "step": 34735
    },
    {
      "epoch": 2.988557171126215,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003491736559422315,
      "loss": 0.8617,
      "step": 34736
    },
    {
      "epoch": 2.9886432074335367,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003491478750420463,
      "loss": 1.2635,
      "step": 34737
    },
    {
      "epoch": 2.988729243740859,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003491220945830744,
      "loss": 0.9033,
      "step": 34738
    },
    {
      "epoch": 2.9888152800481804,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00034909631456539095,
      "loss": 1.1658,
      "step": 34739
    },
    {
      "epoch": 2.988901316355502,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00034907053498907137,
      "loss": 1.3125,
      "step": 34740
    },
    {
      "epoch": 2.9889873526628237,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00034904475585419126,
      "loss": 1.0497,
      "step": 34741
    },
    {
      "epoch": 2.9890733889701453,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003490189771608258,
      "loss": 1.539,
      "step": 34742
    },
    {
      "epoch": 2.9891594252774674,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00034899319890905033,
      "loss": 1.1328,
      "step": 34743
    },
    {
      "epoch": 2.9892454615847885,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00034896742109894054,
      "loss": 1.0065,
      "step": 34744
    },
    {
      "epoch": 2.9893314978921106,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003489416437305716,
      "loss": 1.8789,
      "step": 34745
    },
    {
      "epoch": 2.9894175341994322,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00034891586680401907,
      "loss": 1.3729,
      "step": 34746
    },
    {
      "epoch": 2.989503570506754,
      "grad_norm": 1.0625,
      "learning_rate": 0.00034889009031935825,
      "loss": 1.0293,
      "step": 34747
    },
    {
      "epoch": 2.9895896068140755,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003488643142766645,
      "loss": 1.5536,
      "step": 34748
    },
    {
      "epoch": 2.989675643121397,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0003488385386760131,
      "loss": 1.8077,
      "step": 34749
    },
    {
      "epoch": 2.989761679428719,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00034881276351747976,
      "loss": 0.7972,
      "step": 34750
    },
    {
      "epoch": 2.9898477157360404,
      "grad_norm": 1.1875,
      "learning_rate": 0.00034878698880113967,
      "loss": 1.5858,
      "step": 34751
    },
    {
      "epoch": 2.9899337520433624,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003487612145270681,
      "loss": 1.6109,
      "step": 34752
    },
    {
      "epoch": 2.990019788350684,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003487354406953407,
      "loss": 1.8609,
      "step": 34753
    },
    {
      "epoch": 2.9901058246580057,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003487096673060326,
      "loss": 1.2596,
      "step": 34754
    },
    {
      "epoch": 2.9901918609653273,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003486838943592194,
      "loss": 1.9283,
      "step": 34755
    },
    {
      "epoch": 2.990277897272649,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00034865812185497626,
      "loss": 0.7257,
      "step": 34756
    },
    {
      "epoch": 2.990363933579971,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003486323497933788,
      "loss": 1.4565,
      "step": 34757
    },
    {
      "epoch": 2.990449969887292,
      "grad_norm": 1.109375,
      "learning_rate": 0.00034860657817450226,
      "loss": 1.2497,
      "step": 34758
    },
    {
      "epoch": 2.9905360061946142,
      "grad_norm": 1.140625,
      "learning_rate": 0.00034858080699842205,
      "loss": 1.3344,
      "step": 34759
    },
    {
      "epoch": 2.990622042501936,
      "grad_norm": 1.578125,
      "learning_rate": 0.00034855503626521345,
      "loss": 1.128,
      "step": 34760
    },
    {
      "epoch": 2.9907080788092575,
      "grad_norm": 0.984375,
      "learning_rate": 0.00034852926597495195,
      "loss": 0.9545,
      "step": 34761
    },
    {
      "epoch": 2.990794115116579,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003485034961277129,
      "loss": 0.7811,
      "step": 34762
    },
    {
      "epoch": 2.9908801514239007,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003484777267235716,
      "loss": 1.1085,
      "step": 34763
    },
    {
      "epoch": 2.990966187731223,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00034845195776260355,
      "loss": 0.6553,
      "step": 34764
    },
    {
      "epoch": 2.9910522240385444,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003484261892448841,
      "loss": 1.7455,
      "step": 34765
    },
    {
      "epoch": 2.991138260345866,
      "grad_norm": 1.046875,
      "learning_rate": 0.00034840042117048847,
      "loss": 1.2109,
      "step": 34766
    },
    {
      "epoch": 2.9912242966531877,
      "grad_norm": 0.859375,
      "learning_rate": 0.00034837465353949204,
      "loss": 0.4774,
      "step": 34767
    },
    {
      "epoch": 2.9913103329605093,
      "grad_norm": 1.4375,
      "learning_rate": 0.00034834888635197045,
      "loss": 1.3869,
      "step": 34768
    },
    {
      "epoch": 2.991396369267831,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003483231196079988,
      "loss": 0.7348,
      "step": 34769
    },
    {
      "epoch": 2.9914824055751525,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00034829735330765246,
      "loss": 0.9839,
      "step": 34770
    },
    {
      "epoch": 2.9915684418824746,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00034827158745100693,
      "loss": 1.5343,
      "step": 34771
    },
    {
      "epoch": 2.9916544781897962,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003482458220381375,
      "loss": 1.8092,
      "step": 34772
    },
    {
      "epoch": 2.991740514497118,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003482200570691195,
      "loss": 0.9674,
      "step": 34773
    },
    {
      "epoch": 2.9918265508044395,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003481942925440283,
      "loss": 1.2734,
      "step": 34774
    },
    {
      "epoch": 2.991912587111761,
      "grad_norm": 1.171875,
      "learning_rate": 0.00034816852846293936,
      "loss": 0.8302,
      "step": 34775
    },
    {
      "epoch": 2.9919986234190827,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00034814276482592787,
      "loss": 0.7689,
      "step": 34776
    },
    {
      "epoch": 2.9920846597264044,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003481170016330693,
      "loss": 0.6815,
      "step": 34777
    },
    {
      "epoch": 2.9921706960337264,
      "grad_norm": 1.015625,
      "learning_rate": 0.00034809123888443884,
      "loss": 0.9505,
      "step": 34778
    },
    {
      "epoch": 2.992256732341048,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00034806547658011207,
      "loss": 1.7254,
      "step": 34779
    },
    {
      "epoch": 2.9923427686483697,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00034803971472016404,
      "loss": 0.8262,
      "step": 34780
    },
    {
      "epoch": 2.9924288049556913,
      "grad_norm": 1.28125,
      "learning_rate": 0.00034801395330467044,
      "loss": 1.127,
      "step": 34781
    },
    {
      "epoch": 2.992514841263013,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00034798819233370655,
      "loss": 1.8029,
      "step": 34782
    },
    {
      "epoch": 2.9926008775703345,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003479624318073475,
      "loss": 0.6746,
      "step": 34783
    },
    {
      "epoch": 2.992686913877656,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003479366717256688,
      "loss": 1.6869,
      "step": 34784
    },
    {
      "epoch": 2.9927729501849782,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003479109120887459,
      "loss": 0.9817,
      "step": 34785
    },
    {
      "epoch": 2.9928589864923,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003478851528966538,
      "loss": 1.5515,
      "step": 34786
    },
    {
      "epoch": 2.9929450227996215,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00034785939414946817,
      "loss": 0.9389,
      "step": 34787
    },
    {
      "epoch": 2.993031059106943,
      "grad_norm": 1.296875,
      "learning_rate": 0.00034783363584726425,
      "loss": 1.9395,
      "step": 34788
    },
    {
      "epoch": 2.9931170954142647,
      "grad_norm": 0.796875,
      "learning_rate": 0.00034780787799011724,
      "loss": 0.685,
      "step": 34789
    },
    {
      "epoch": 2.9932031317215864,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003477821205781026,
      "loss": 1.2543,
      "step": 34790
    },
    {
      "epoch": 2.993289168028908,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003477563636112957,
      "loss": 1.4389,
      "step": 34791
    },
    {
      "epoch": 2.99337520433623,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0003477306070897719,
      "loss": 0.6164,
      "step": 34792
    },
    {
      "epoch": 2.9934612406435517,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00034770485101360627,
      "loss": 0.6583,
      "step": 34793
    },
    {
      "epoch": 2.9935472769508733,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00034767909538287444,
      "loss": 1.4251,
      "step": 34794
    },
    {
      "epoch": 2.993633313258195,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003476533401976517,
      "loss": 1.2513,
      "step": 34795
    },
    {
      "epoch": 2.9937193495655166,
      "grad_norm": 0.9375,
      "learning_rate": 0.00034762758545801327,
      "loss": 1.4528,
      "step": 34796
    },
    {
      "epoch": 2.993805385872838,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003476018311640344,
      "loss": 1.2771,
      "step": 34797
    },
    {
      "epoch": 2.99389142218016,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00034757607731579065,
      "loss": 0.8031,
      "step": 34798
    },
    {
      "epoch": 2.993977458487482,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003475503239133572,
      "loss": 1.6943,
      "step": 34799
    },
    {
      "epoch": 2.9940634947948035,
      "grad_norm": 1.3125,
      "learning_rate": 0.00034752457095680943,
      "loss": 1.7799,
      "step": 34800
    },
    {
      "epoch": 2.994149531102125,
      "grad_norm": 1.4375,
      "learning_rate": 0.00034749881844622254,
      "loss": 1.6035,
      "step": 34801
    },
    {
      "epoch": 2.9942355674094467,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003474730663816721,
      "loss": 1.2803,
      "step": 34802
    },
    {
      "epoch": 2.9943216037167684,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003474473147632332,
      "loss": 1.9489,
      "step": 34803
    },
    {
      "epoch": 2.9944076400240904,
      "grad_norm": 1.015625,
      "learning_rate": 0.00034742156359098117,
      "loss": 0.7526,
      "step": 34804
    },
    {
      "epoch": 2.9944936763314116,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003473958128649914,
      "loss": 1.3496,
      "step": 34805
    },
    {
      "epoch": 2.9945797126387337,
      "grad_norm": 1.375,
      "learning_rate": 0.0003473700625853393,
      "loss": 1.54,
      "step": 34806
    },
    {
      "epoch": 2.9946657489460553,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003473443127521001,
      "loss": 1.5995,
      "step": 34807
    },
    {
      "epoch": 2.994751785253377,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00034731856336534894,
      "loss": 0.6898,
      "step": 34808
    },
    {
      "epoch": 2.9948378215606986,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00034729281442516134,
      "loss": 0.7253,
      "step": 34809
    },
    {
      "epoch": 2.99492385786802,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00034726706593161266,
      "loss": 0.7658,
      "step": 34810
    },
    {
      "epoch": 2.9950098941753422,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000347241317884778,
      "loss": 1.5352,
      "step": 34811
    },
    {
      "epoch": 2.9950959304826634,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00034721557028473285,
      "loss": 1.6589,
      "step": 34812
    },
    {
      "epoch": 2.9951819667899855,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003471898231315524,
      "loss": 0.5953,
      "step": 34813
    },
    {
      "epoch": 2.995268003097307,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000347164076425312,
      "loss": 0.7641,
      "step": 34814
    },
    {
      "epoch": 2.9953540394046287,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0003471383301660869,
      "loss": 0.4224,
      "step": 34815
    },
    {
      "epoch": 2.9954400757119504,
      "grad_norm": 1.109375,
      "learning_rate": 0.00034711258435395255,
      "loss": 1.2142,
      "step": 34816
    },
    {
      "epoch": 2.995526112019272,
      "grad_norm": 6.8125,
      "learning_rate": 0.0003470868389889839,
      "loss": 1.1522,
      "step": 34817
    },
    {
      "epoch": 2.995612148326594,
      "grad_norm": 0.6875,
      "learning_rate": 0.00034706109407125677,
      "loss": 0.8079,
      "step": 34818
    },
    {
      "epoch": 2.9956981846339152,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00034703534960084614,
      "loss": 1.7044,
      "step": 34819
    },
    {
      "epoch": 2.9957842209412373,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003470096055778273,
      "loss": 1.2221,
      "step": 34820
    },
    {
      "epoch": 2.995870257248559,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003469838620022757,
      "loss": 0.8697,
      "step": 34821
    },
    {
      "epoch": 2.9959562935558806,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003469581188742665,
      "loss": 1.4423,
      "step": 34822
    },
    {
      "epoch": 2.996042329863202,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00034693237619387514,
      "loss": 1.0122,
      "step": 34823
    },
    {
      "epoch": 2.996128366170524,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003469066339611766,
      "loss": 1.669,
      "step": 34824
    },
    {
      "epoch": 2.996214402477846,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003468808921762466,
      "loss": 1.401,
      "step": 34825
    },
    {
      "epoch": 2.9963004387851675,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0003468551508391601,
      "loss": 1.5346,
      "step": 34826
    },
    {
      "epoch": 2.996386475092489,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00034682940994999246,
      "loss": 0.5062,
      "step": 34827
    },
    {
      "epoch": 2.9964725113998107,
      "grad_norm": 1.359375,
      "learning_rate": 0.00034680366950881906,
      "loss": 1.943,
      "step": 34828
    },
    {
      "epoch": 2.9965585477071324,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003467779295157152,
      "loss": 1.56,
      "step": 34829
    },
    {
      "epoch": 2.996644584014454,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00034675218997075606,
      "loss": 1.4513,
      "step": 34830
    },
    {
      "epoch": 2.9967306203217756,
      "grad_norm": 1.25,
      "learning_rate": 0.00034672645087401683,
      "loss": 0.9102,
      "step": 34831
    },
    {
      "epoch": 2.9968166566290977,
      "grad_norm": 1.25,
      "learning_rate": 0.0003467007122255731,
      "loss": 1.5093,
      "step": 34832
    },
    {
      "epoch": 2.9969026929364193,
      "grad_norm": 1.5625,
      "learning_rate": 0.00034667497402549984,
      "loss": 1.4184,
      "step": 34833
    },
    {
      "epoch": 2.996988729243741,
      "grad_norm": 1.078125,
      "learning_rate": 0.00034664923627387246,
      "loss": 1.0238,
      "step": 34834
    },
    {
      "epoch": 2.9970747655510626,
      "grad_norm": 0.875,
      "learning_rate": 0.00034662349897076635,
      "loss": 1.2897,
      "step": 34835
    },
    {
      "epoch": 2.997160801858384,
      "grad_norm": 2.0,
      "learning_rate": 0.0003465977621162566,
      "loss": 1.4229,
      "step": 34836
    },
    {
      "epoch": 2.997246838165706,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0003465720257104186,
      "loss": 1.5372,
      "step": 34837
    },
    {
      "epoch": 2.9973328744730274,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003465462897533275,
      "loss": 1.6492,
      "step": 34838
    },
    {
      "epoch": 2.9974189107803495,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003465205542450588,
      "loss": 1.3982,
      "step": 34839
    },
    {
      "epoch": 2.997504947087671,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003464948191856875,
      "loss": 1.0544,
      "step": 34840
    },
    {
      "epoch": 2.9975909833949927,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000346469084575289,
      "loss": 1.1518,
      "step": 34841
    },
    {
      "epoch": 2.9976770197023144,
      "grad_norm": 3.9375,
      "learning_rate": 0.00034644335041393863,
      "loss": 1.6945,
      "step": 34842
    },
    {
      "epoch": 2.997763056009636,
      "grad_norm": 1.21875,
      "learning_rate": 0.00034641761670171163,
      "loss": 1.5232,
      "step": 34843
    },
    {
      "epoch": 2.9978490923169576,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003463918834386832,
      "loss": 1.6384,
      "step": 34844
    },
    {
      "epoch": 2.9979351286242792,
      "grad_norm": 1.375,
      "learning_rate": 0.00034636615062492857,
      "loss": 1.475,
      "step": 34845
    },
    {
      "epoch": 2.9980211649316013,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003463404182605231,
      "loss": 1.6094,
      "step": 34846
    },
    {
      "epoch": 2.998107201238923,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003463146863455421,
      "loss": 1.0727,
      "step": 34847
    },
    {
      "epoch": 2.9981932375462446,
      "grad_norm": 44.75,
      "learning_rate": 0.0003462889548800607,
      "loss": 1.4302,
      "step": 34848
    },
    {
      "epoch": 2.998279273853566,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0003462632238641541,
      "loss": 1.6202,
      "step": 34849
    },
    {
      "epoch": 2.998365310160888,
      "grad_norm": 1.125,
      "learning_rate": 0.00034623749329789775,
      "loss": 2.0728,
      "step": 34850
    },
    {
      "epoch": 2.9984513464682094,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003462117631813669,
      "loss": 0.9901,
      "step": 34851
    },
    {
      "epoch": 2.998537382775531,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0003461860335146366,
      "loss": 0.9761,
      "step": 34852
    },
    {
      "epoch": 2.998623419082853,
      "grad_norm": 1.15625,
      "learning_rate": 0.00034616030429778237,
      "loss": 2.0499,
      "step": 34853
    },
    {
      "epoch": 2.9987094553901747,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00034613457553087914,
      "loss": 0.8483,
      "step": 34854
    },
    {
      "epoch": 2.9987954916974964,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003461088472140025,
      "loss": 0.7905,
      "step": 34855
    },
    {
      "epoch": 2.998881528004818,
      "grad_norm": 0.765625,
      "learning_rate": 0.0003460831193472276,
      "loss": 1.2512,
      "step": 34856
    },
    {
      "epoch": 2.9989675643121396,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003460573919306295,
      "loss": 0.9318,
      "step": 34857
    },
    {
      "epoch": 2.9990536006194617,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003460316649642836,
      "loss": 0.7896,
      "step": 34858
    },
    {
      "epoch": 2.999139636926783,
      "grad_norm": 0.859375,
      "learning_rate": 0.00034600593844826526,
      "loss": 1.9609,
      "step": 34859
    },
    {
      "epoch": 2.999225673234105,
      "grad_norm": 2.90625,
      "learning_rate": 0.00034598021238264957,
      "loss": 1.2687,
      "step": 34860
    },
    {
      "epoch": 2.9993117095414266,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00034595448676751176,
      "loss": 0.7072,
      "step": 34861
    },
    {
      "epoch": 2.999397745848748,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003459287616029271,
      "loss": 0.6571,
      "step": 34862
    },
    {
      "epoch": 2.99948378215607,
      "grad_norm": 0.96875,
      "learning_rate": 0.00034590303688897093,
      "loss": 0.5045,
      "step": 34863
    },
    {
      "epoch": 2.9995698184633914,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003458773126257184,
      "loss": 1.6802,
      "step": 34864
    },
    {
      "epoch": 2.9996558547707135,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00034585158881324463,
      "loss": 1.2492,
      "step": 34865
    },
    {
      "epoch": 2.9997418910780347,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003458258654516251,
      "loss": 1.2955,
      "step": 34866
    },
    {
      "epoch": 2.9998279273853568,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003458001425409349,
      "loss": 0.6897,
      "step": 34867
    },
    {
      "epoch": 2.9999139636926784,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003457744200812492,
      "loss": 1.8963,
      "step": 34868
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003457486980726434,
      "loss": 1.3916,
      "step": 34869
    },
    {
      "epoch": 3.0000860363073216,
      "grad_norm": 3.21875,
      "learning_rate": 0.00034572297651519273,
      "loss": 1.3097,
      "step": 34870
    },
    {
      "epoch": 3.0001720726146432,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00034569725540897235,
      "loss": 0.6531,
      "step": 34871
    },
    {
      "epoch": 3.000258108921965,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003456715347540573,
      "loss": 0.9908,
      "step": 34872
    },
    {
      "epoch": 3.000258108921965,
      "eval_loss": 1.1076171398162842,
      "eval_runtime": 354.8838,
      "eval_samples_per_second": 13.1,
      "eval_steps_per_second": 1.64,
      "step": 34872
    },
    {
      "epoch": 3.000344145229287,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00034564581455052317,
      "loss": 0.8116,
      "step": 34873
    },
    {
      "epoch": 3.0004301815366086,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000345620094798445,
      "loss": 1.5021,
      "step": 34874
    },
    {
      "epoch": 3.00051621784393,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000345594375497898,
      "loss": 0.9937,
      "step": 34875
    },
    {
      "epoch": 3.000602254151252,
      "grad_norm": 1.03125,
      "learning_rate": 0.00034556865664895736,
      "loss": 1.2183,
      "step": 34876
    },
    {
      "epoch": 3.0006882904585734,
      "grad_norm": 1.390625,
      "learning_rate": 0.00034554293825169845,
      "loss": 1.3007,
      "step": 34877
    },
    {
      "epoch": 3.000774326765895,
      "grad_norm": 1.578125,
      "learning_rate": 0.00034551722030619635,
      "loss": 1.5928,
      "step": 34878
    },
    {
      "epoch": 3.0008603630732167,
      "grad_norm": 1.75,
      "learning_rate": 0.00034549150281252633,
      "loss": 1.2103,
      "step": 34879
    },
    {
      "epoch": 3.0009463993805388,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0003454657857707637,
      "loss": 0.6197,
      "step": 34880
    },
    {
      "epoch": 3.0010324356878604,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003454400691809835,
      "loss": 0.9271,
      "step": 34881
    },
    {
      "epoch": 3.001118471995182,
      "grad_norm": 1.203125,
      "learning_rate": 0.000345414353043261,
      "loss": 1.0833,
      "step": 34882
    },
    {
      "epoch": 3.0012045083025036,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00034538863735767154,
      "loss": 1.1249,
      "step": 34883
    },
    {
      "epoch": 3.0012905446098253,
      "grad_norm": 1.015625,
      "learning_rate": 0.00034536292212429033,
      "loss": 1.1531,
      "step": 34884
    },
    {
      "epoch": 3.001376580917147,
      "grad_norm": 1.125,
      "learning_rate": 0.0003453372073431924,
      "loss": 1.1185,
      "step": 34885
    },
    {
      "epoch": 3.001462617224469,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000345311493014453,
      "loss": 0.9483,
      "step": 34886
    },
    {
      "epoch": 3.0015486535317906,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003452857791381474,
      "loss": 0.7652,
      "step": 34887
    },
    {
      "epoch": 3.001634689839112,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003452600657143509,
      "loss": 0.5394,
      "step": 34888
    },
    {
      "epoch": 3.001720726146434,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00034523435274313866,
      "loss": 1.7307,
      "step": 34889
    },
    {
      "epoch": 3.0018067624537554,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00034520864022458565,
      "loss": 1.2802,
      "step": 34890
    },
    {
      "epoch": 3.001892798761077,
      "grad_norm": 1.203125,
      "learning_rate": 0.00034518292815876716,
      "loss": 1.2218,
      "step": 34891
    },
    {
      "epoch": 3.0019788350683987,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0003451572165457587,
      "loss": 0.917,
      "step": 34892
    },
    {
      "epoch": 3.0020648713757208,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003451315053856353,
      "loss": 1.366,
      "step": 34893
    },
    {
      "epoch": 3.0021509076830424,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003451057946784721,
      "loss": 1.0775,
      "step": 34894
    },
    {
      "epoch": 3.002236943990364,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00034508008442434425,
      "loss": 1.0754,
      "step": 34895
    },
    {
      "epoch": 3.0023229802976856,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0003450543746233271,
      "loss": 1.0206,
      "step": 34896
    },
    {
      "epoch": 3.0024090166050073,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003450286652754958,
      "loss": 2.2918,
      "step": 34897
    },
    {
      "epoch": 3.002495052912329,
      "grad_norm": 1064.0,
      "learning_rate": 0.00034500295638092537,
      "loss": 1.0396,
      "step": 34898
    },
    {
      "epoch": 3.0025810892196505,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0003449772479396913,
      "loss": 0.9756,
      "step": 34899
    },
    {
      "epoch": 3.0026671255269726,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003449515399518686,
      "loss": 1.1186,
      "step": 34900
    },
    {
      "epoch": 3.002753161834294,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003449258324175325,
      "loss": 0.931,
      "step": 34901
    },
    {
      "epoch": 3.002839198141616,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003449001253367581,
      "loss": 1.5263,
      "step": 34902
    },
    {
      "epoch": 3.0029252344489374,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00034487441870962075,
      "loss": 0.7086,
      "step": 34903
    },
    {
      "epoch": 3.003011270756259,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00034484871253619555,
      "loss": 1.3866,
      "step": 34904
    },
    {
      "epoch": 3.0030973070635807,
      "grad_norm": 0.96875,
      "learning_rate": 0.00034482300681655753,
      "loss": 0.8727,
      "step": 34905
    },
    {
      "epoch": 3.0031833433709023,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003447973015507822,
      "loss": 1.301,
      "step": 34906
    },
    {
      "epoch": 3.0032693796782244,
      "grad_norm": 4.21875,
      "learning_rate": 0.0003447715967389446,
      "loss": 1.5146,
      "step": 34907
    },
    {
      "epoch": 3.003355415985546,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0003447458923811199,
      "loss": 0.8787,
      "step": 34908
    },
    {
      "epoch": 3.0034414522928676,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003447201884773831,
      "loss": 1.2347,
      "step": 34909
    },
    {
      "epoch": 3.0035274886001893,
      "grad_norm": 1.71875,
      "learning_rate": 0.0003446944850278096,
      "loss": 0.9594,
      "step": 34910
    },
    {
      "epoch": 3.003613524907511,
      "grad_norm": 1.0625,
      "learning_rate": 0.00034466878203247465,
      "loss": 1.2759,
      "step": 34911
    },
    {
      "epoch": 3.0036995612148325,
      "grad_norm": 1.234375,
      "learning_rate": 0.00034464307949145324,
      "loss": 0.8182,
      "step": 34912
    },
    {
      "epoch": 3.003785597522154,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00034461737740482053,
      "loss": 0.6533,
      "step": 34913
    },
    {
      "epoch": 3.003871633829476,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003445916757726518,
      "loss": 0.4233,
      "step": 34914
    },
    {
      "epoch": 3.003957670136798,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003445659745950223,
      "loss": 0.7433,
      "step": 34915
    },
    {
      "epoch": 3.0040437064441194,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00034454027387200695,
      "loss": 1.1101,
      "step": 34916
    },
    {
      "epoch": 3.004129742751441,
      "grad_norm": 7.96875,
      "learning_rate": 0.00034451457360368124,
      "loss": 1.2239,
      "step": 34917
    },
    {
      "epoch": 3.0042157790587627,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00034448887379012007,
      "loss": 1.1554,
      "step": 34918
    },
    {
      "epoch": 3.0043018153660843,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0003444631744313986,
      "loss": 0.6936,
      "step": 34919
    },
    {
      "epoch": 3.0043878516734064,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003444374755275922,
      "loss": 0.7734,
      "step": 34920
    },
    {
      "epoch": 3.004473887980728,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003444117770787759,
      "loss": 1.2341,
      "step": 34921
    },
    {
      "epoch": 3.0045599242880496,
      "grad_norm": 1.078125,
      "learning_rate": 0.00034438607908502495,
      "loss": 1.2747,
      "step": 34922
    },
    {
      "epoch": 3.0046459605953713,
      "grad_norm": 1.375,
      "learning_rate": 0.00034436038154641435,
      "loss": 0.8685,
      "step": 34923
    },
    {
      "epoch": 3.004731996902693,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003443346844630194,
      "loss": 1.4721,
      "step": 34924
    },
    {
      "epoch": 3.0048180332100145,
      "grad_norm": 17.0,
      "learning_rate": 0.00034430898783491525,
      "loss": 1.1279,
      "step": 34925
    },
    {
      "epoch": 3.004904069517336,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00034428329166217707,
      "loss": 1.0107,
      "step": 34926
    },
    {
      "epoch": 3.004990105824658,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00034425759594487986,
      "loss": 1.2794,
      "step": 34927
    },
    {
      "epoch": 3.00507614213198,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00034423190068309876,
      "loss": 1.4643,
      "step": 34928
    },
    {
      "epoch": 3.0051621784393014,
      "grad_norm": 1.1875,
      "learning_rate": 0.00034420620587690933,
      "loss": 0.6202,
      "step": 34929
    },
    {
      "epoch": 3.005248214746623,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00034418051152638644,
      "loss": 1.4878,
      "step": 34930
    },
    {
      "epoch": 3.0053342510539447,
      "grad_norm": 1.484375,
      "learning_rate": 0.00034415481763160516,
      "loss": 1.6295,
      "step": 34931
    },
    {
      "epoch": 3.0054202873612663,
      "grad_norm": 1.328125,
      "learning_rate": 0.00034412912419264064,
      "loss": 1.1822,
      "step": 34932
    },
    {
      "epoch": 3.005506323668588,
      "grad_norm": 1.015625,
      "learning_rate": 0.00034410343120956833,
      "loss": 1.5081,
      "step": 34933
    },
    {
      "epoch": 3.00559235997591,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00034407773868246305,
      "loss": 1.5755,
      "step": 34934
    },
    {
      "epoch": 3.0056783962832316,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003440520466114001,
      "loss": 1.4997,
      "step": 34935
    },
    {
      "epoch": 3.0057644325905533,
      "grad_norm": 1.1875,
      "learning_rate": 0.00034402635499645453,
      "loss": 1.016,
      "step": 34936
    },
    {
      "epoch": 3.005850468897875,
      "grad_norm": 1.015625,
      "learning_rate": 0.00034400066383770156,
      "loss": 0.9109,
      "step": 34937
    },
    {
      "epoch": 3.0059365052051965,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003439749731352163,
      "loss": 0.8317,
      "step": 34938
    },
    {
      "epoch": 3.006022541512518,
      "grad_norm": 1.078125,
      "learning_rate": 0.00034394928288907384,
      "loss": 0.7336,
      "step": 34939
    },
    {
      "epoch": 3.0061085778198398,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003439235930993495,
      "loss": 1.0089,
      "step": 34940
    },
    {
      "epoch": 3.006194614127162,
      "grad_norm": 1.25,
      "learning_rate": 0.0003438979037661182,
      "loss": 1.2187,
      "step": 34941
    },
    {
      "epoch": 3.0062806504344834,
      "grad_norm": 16.25,
      "learning_rate": 0.0003438722148894553,
      "loss": 1.3902,
      "step": 34942
    },
    {
      "epoch": 3.006366686741805,
      "grad_norm": 1.25,
      "learning_rate": 0.0003438465264694356,
      "loss": 0.8569,
      "step": 34943
    },
    {
      "epoch": 3.0064527230491267,
      "grad_norm": 9.25,
      "learning_rate": 0.0003438208385061346,
      "loss": 1.7544,
      "step": 34944
    },
    {
      "epoch": 3.0065387593564483,
      "grad_norm": 1.1875,
      "learning_rate": 0.00034379515099962723,
      "loss": 1.5329,
      "step": 34945
    },
    {
      "epoch": 3.00662479566377,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003437694639499886,
      "loss": 1.3112,
      "step": 34946
    },
    {
      "epoch": 3.006710831971092,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000343743777357294,
      "loss": 1.1201,
      "step": 34947
    },
    {
      "epoch": 3.0067968682784136,
      "grad_norm": 1.578125,
      "learning_rate": 0.00034371809122161844,
      "loss": 0.5787,
      "step": 34948
    },
    {
      "epoch": 3.0068829045857353,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0003436924055430371,
      "loss": 1.351,
      "step": 34949
    },
    {
      "epoch": 3.006968940893057,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00034366672032162495,
      "loss": 1.187,
      "step": 34950
    },
    {
      "epoch": 3.0070549772003785,
      "grad_norm": 1.078125,
      "learning_rate": 0.00034364103555745737,
      "loss": 1.3153,
      "step": 34951
    },
    {
      "epoch": 3.0071410135077,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003436153512506093,
      "loss": 1.3321,
      "step": 34952
    },
    {
      "epoch": 3.0072270498150218,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003435896674011559,
      "loss": 1.3865,
      "step": 34953
    },
    {
      "epoch": 3.007313086122344,
      "grad_norm": 48.75,
      "learning_rate": 0.0003435639840091723,
      "loss": 1.1954,
      "step": 34954
    },
    {
      "epoch": 3.0073991224296655,
      "grad_norm": 44.75,
      "learning_rate": 0.0003435383010747336,
      "loss": 1.5773,
      "step": 34955
    },
    {
      "epoch": 3.007485158736987,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00034351261859791505,
      "loss": 1.76,
      "step": 34956
    },
    {
      "epoch": 3.0075711950443087,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00034348693657879146,
      "loss": 1.2619,
      "step": 34957
    },
    {
      "epoch": 3.0076572313516303,
      "grad_norm": 2.015625,
      "learning_rate": 0.0003434612550174383,
      "loss": 0.8773,
      "step": 34958
    },
    {
      "epoch": 3.007743267658952,
      "grad_norm": 1.3125,
      "learning_rate": 0.00034343557391393055,
      "loss": 1.2354,
      "step": 34959
    },
    {
      "epoch": 3.0078293039662736,
      "grad_norm": 1560.0,
      "learning_rate": 0.0003434098932683432,
      "loss": 0.6187,
      "step": 34960
    },
    {
      "epoch": 3.0079153402735956,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003433842130807514,
      "loss": 0.8951,
      "step": 34961
    },
    {
      "epoch": 3.0080013765809173,
      "grad_norm": 1.375,
      "learning_rate": 0.00034335853335123046,
      "loss": 1.0238,
      "step": 34962
    },
    {
      "epoch": 3.008087412888239,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003433328540798553,
      "loss": 1.5352,
      "step": 34963
    },
    {
      "epoch": 3.0081734491955605,
      "grad_norm": 1.21875,
      "learning_rate": 0.00034330717526670084,
      "loss": 1.2695,
      "step": 34964
    },
    {
      "epoch": 3.008259485502882,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00034328149691184275,
      "loss": 1.7217,
      "step": 34965
    },
    {
      "epoch": 3.0083455218102038,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003432558190153556,
      "loss": 0.9084,
      "step": 34966
    },
    {
      "epoch": 3.0084315581175254,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0003432301415773148,
      "loss": 1.0336,
      "step": 34967
    },
    {
      "epoch": 3.0085175944248475,
      "grad_norm": 1.0,
      "learning_rate": 0.0003432044645977953,
      "loss": 0.6536,
      "step": 34968
    },
    {
      "epoch": 3.008603630732169,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003431787880768722,
      "loss": 0.8059,
      "step": 34969
    },
    {
      "epoch": 3.0086896670394907,
      "grad_norm": 1.0,
      "learning_rate": 0.0003431531120146208,
      "loss": 1.2475,
      "step": 34970
    },
    {
      "epoch": 3.0087757033468123,
      "grad_norm": 0.96875,
      "learning_rate": 0.000343127436411116,
      "loss": 1.7569,
      "step": 34971
    },
    {
      "epoch": 3.008861739654134,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003431017612664329,
      "loss": 1.9022,
      "step": 34972
    },
    {
      "epoch": 3.0089477759614556,
      "grad_norm": 1.5625,
      "learning_rate": 0.00034307608658064656,
      "loss": 0.7035,
      "step": 34973
    },
    {
      "epoch": 3.0090338122687776,
      "grad_norm": 31.5,
      "learning_rate": 0.00034305041235383225,
      "loss": 1.0438,
      "step": 34974
    },
    {
      "epoch": 3.0091198485760993,
      "grad_norm": 1.1875,
      "learning_rate": 0.00034302473858606496,
      "loss": 1.2532,
      "step": 34975
    },
    {
      "epoch": 3.009205884883421,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00034299906527741973,
      "loss": 1.5954,
      "step": 34976
    },
    {
      "epoch": 3.0092919211907425,
      "grad_norm": 74.5,
      "learning_rate": 0.0003429733924279716,
      "loss": 0.8547,
      "step": 34977
    },
    {
      "epoch": 3.009377957498064,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00034294772003779596,
      "loss": 0.5912,
      "step": 34978
    },
    {
      "epoch": 3.0094639938053858,
      "grad_norm": 138.0,
      "learning_rate": 0.00034292204810696764,
      "loss": 1.1191,
      "step": 34979
    },
    {
      "epoch": 3.0095500301127074,
      "grad_norm": 1304.0,
      "learning_rate": 0.0003428963766355617,
      "loss": 1.2177,
      "step": 34980
    },
    {
      "epoch": 3.0096360664200295,
      "grad_norm": 4.75,
      "learning_rate": 0.00034287070562365337,
      "loss": 1.5677,
      "step": 34981
    },
    {
      "epoch": 3.009722102727351,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003428450350713177,
      "loss": 0.9093,
      "step": 34982
    },
    {
      "epoch": 3.0098081390346727,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00034281936497862963,
      "loss": 0.6885,
      "step": 34983
    },
    {
      "epoch": 3.0098941753419943,
      "grad_norm": 11.6875,
      "learning_rate": 0.0003427936953456644,
      "loss": 0.7055,
      "step": 34984
    },
    {
      "epoch": 3.009980211649316,
      "grad_norm": 1.6171875,
      "learning_rate": 0.000342768026172497,
      "loss": 1.3884,
      "step": 34985
    },
    {
      "epoch": 3.0100662479566376,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003427423574592027,
      "loss": 0.8135,
      "step": 34986
    },
    {
      "epoch": 3.010152284263959,
      "grad_norm": 4.15625,
      "learning_rate": 0.0003427166892058562,
      "loss": 1.8886,
      "step": 34987
    },
    {
      "epoch": 3.0102383205712813,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003426910214125329,
      "loss": 1.6193,
      "step": 34988
    },
    {
      "epoch": 3.010324356878603,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003426653540793078,
      "loss": 0.9616,
      "step": 34989
    },
    {
      "epoch": 3.0104103931859245,
      "grad_norm": 1.0,
      "learning_rate": 0.00034263968720625593,
      "loss": 0.8226,
      "step": 34990
    },
    {
      "epoch": 3.010496429493246,
      "grad_norm": 1.46875,
      "learning_rate": 0.0003426140207934523,
      "loss": 1.2581,
      "step": 34991
    },
    {
      "epoch": 3.0105824658005678,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00034258835484097204,
      "loss": 0.9615,
      "step": 34992
    },
    {
      "epoch": 3.0106685021078894,
      "grad_norm": 1.109375,
      "learning_rate": 0.00034256268934889034,
      "loss": 0.7201,
      "step": 34993
    },
    {
      "epoch": 3.010754538415211,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000342537024317282,
      "loss": 1.3225,
      "step": 34994
    },
    {
      "epoch": 3.010840574722533,
      "grad_norm": 41.75,
      "learning_rate": 0.0003425113597462224,
      "loss": 1.211,
      "step": 34995
    },
    {
      "epoch": 3.0109266110298547,
      "grad_norm": 62.75,
      "learning_rate": 0.00034248569563578634,
      "loss": 0.557,
      "step": 34996
    },
    {
      "epoch": 3.0110126473371763,
      "grad_norm": 62.25,
      "learning_rate": 0.00034246003198604906,
      "loss": 1.2164,
      "step": 34997
    },
    {
      "epoch": 3.011098683644498,
      "grad_norm": 1.171875,
      "learning_rate": 0.00034243436879708545,
      "loss": 0.7649,
      "step": 34998
    },
    {
      "epoch": 3.0111847199518196,
      "grad_norm": 1.171875,
      "learning_rate": 0.00034240870606897075,
      "loss": 1.1288,
      "step": 34999
    },
    {
      "epoch": 3.011270756259141,
      "grad_norm": 7.1875,
      "learning_rate": 0.00034238304380177994,
      "loss": 1.1006,
      "step": 35000
    },
    {
      "epoch": 3.0113567925664633,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00034235738199558783,
      "loss": 0.6837,
      "step": 35001
    },
    {
      "epoch": 3.011442828873785,
      "grad_norm": 3.203125,
      "learning_rate": 0.00034233172065047,
      "loss": 1.2643,
      "step": 35002
    },
    {
      "epoch": 3.0115288651811065,
      "grad_norm": 15.5,
      "learning_rate": 0.0003423060597665011,
      "loss": 1.2828,
      "step": 35003
    },
    {
      "epoch": 3.011614901488428,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00034228039934375634,
      "loss": 1.2706,
      "step": 35004
    },
    {
      "epoch": 3.0117009377957498,
      "grad_norm": 1.046875,
      "learning_rate": 0.00034225473938231075,
      "loss": 0.7717,
      "step": 35005
    },
    {
      "epoch": 3.0117869741030714,
      "grad_norm": 154.0,
      "learning_rate": 0.0003422290798822394,
      "loss": 1.0921,
      "step": 35006
    },
    {
      "epoch": 3.011873010410393,
      "grad_norm": 1.046875,
      "learning_rate": 0.00034220342084361724,
      "loss": 1.1133,
      "step": 35007
    },
    {
      "epoch": 3.011959046717715,
      "grad_norm": 41.75,
      "learning_rate": 0.0003421777622665194,
      "loss": 1.3683,
      "step": 35008
    },
    {
      "epoch": 3.0120450830250367,
      "grad_norm": 1.8984375,
      "learning_rate": 0.000342152104151021,
      "loss": 1.932,
      "step": 35009
    },
    {
      "epoch": 3.0121311193323583,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003421264464971968,
      "loss": 0.6413,
      "step": 35010
    },
    {
      "epoch": 3.01221715563968,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003421007893051222,
      "loss": 1.272,
      "step": 35011
    },
    {
      "epoch": 3.0123031919470016,
      "grad_norm": 3.015625,
      "learning_rate": 0.0003420751325748721,
      "loss": 0.8313,
      "step": 35012
    },
    {
      "epoch": 3.012389228254323,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003420494763065215,
      "loss": 0.5773,
      "step": 35013
    },
    {
      "epoch": 3.012475264561645,
      "grad_norm": 1.078125,
      "learning_rate": 0.00034202382050014537,
      "loss": 0.8747,
      "step": 35014
    },
    {
      "epoch": 3.012561300868967,
      "grad_norm": 5.65625,
      "learning_rate": 0.00034199816515581897,
      "loss": 0.8531,
      "step": 35015
    },
    {
      "epoch": 3.0126473371762885,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003419725102736171,
      "loss": 0.6846,
      "step": 35016
    },
    {
      "epoch": 3.01273337348361,
      "grad_norm": 17.5,
      "learning_rate": 0.00034194685585361485,
      "loss": 1.2336,
      "step": 35017
    },
    {
      "epoch": 3.0128194097909318,
      "grad_norm": 146.0,
      "learning_rate": 0.0003419212018958875,
      "loss": 1.6049,
      "step": 35018
    },
    {
      "epoch": 3.0129054460982534,
      "grad_norm": 1.375,
      "learning_rate": 0.0003418955484005098,
      "loss": 1.1796,
      "step": 35019
    },
    {
      "epoch": 3.012991482405575,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00034186989536755686,
      "loss": 0.2533,
      "step": 35020
    },
    {
      "epoch": 3.0130775187128966,
      "grad_norm": 45.75,
      "learning_rate": 0.0003418442427971037,
      "loss": 1.3088,
      "step": 35021
    },
    {
      "epoch": 3.0131635550202187,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00034181859068922537,
      "loss": 0.935,
      "step": 35022
    },
    {
      "epoch": 3.0132495913275403,
      "grad_norm": 2.515625,
      "learning_rate": 0.000341792939043997,
      "loss": 0.9,
      "step": 35023
    },
    {
      "epoch": 3.013335627634862,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003417672878614934,
      "loss": 1.4638,
      "step": 35024
    },
    {
      "epoch": 3.0134216639421836,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003417416371417897,
      "loss": 0.9816,
      "step": 35025
    },
    {
      "epoch": 3.013507700249505,
      "grad_norm": 72.5,
      "learning_rate": 0.000341715986884961,
      "loss": 1.4977,
      "step": 35026
    },
    {
      "epoch": 3.013593736556827,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0003416903370910822,
      "loss": 1.6497,
      "step": 35027
    },
    {
      "epoch": 3.0136797728641485,
      "grad_norm": 2.3125,
      "learning_rate": 0.0003416646877602284,
      "loss": 1.0258,
      "step": 35028
    },
    {
      "epoch": 3.0137658091714705,
      "grad_norm": 5.40625,
      "learning_rate": 0.0003416390388924746,
      "loss": 0.974,
      "step": 35029
    },
    {
      "epoch": 3.013851845478792,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00034161339048789587,
      "loss": 1.3601,
      "step": 35030
    },
    {
      "epoch": 3.0139378817861138,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003415877425465671,
      "loss": 0.6933,
      "step": 35031
    },
    {
      "epoch": 3.0140239180934354,
      "grad_norm": 0.921875,
      "learning_rate": 0.00034156209506856327,
      "loss": 1.07,
      "step": 35032
    },
    {
      "epoch": 3.014109954400757,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003415364480539597,
      "loss": 1.5451,
      "step": 35033
    },
    {
      "epoch": 3.0141959907080786,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00034151080150283107,
      "loss": 0.9267,
      "step": 35034
    },
    {
      "epoch": 3.0142820270154007,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003414851554152525,
      "loss": 0.6266,
      "step": 35035
    },
    {
      "epoch": 3.0143680633227223,
      "grad_norm": 24.625,
      "learning_rate": 0.00034145950979129914,
      "loss": 1.3117,
      "step": 35036
    },
    {
      "epoch": 3.014454099630044,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003414338646310459,
      "loss": 0.7584,
      "step": 35037
    },
    {
      "epoch": 3.0145401359373656,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00034140821993456743,
      "loss": 1.1082,
      "step": 35038
    },
    {
      "epoch": 3.014626172244687,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00034138257570193935,
      "loss": 1.6523,
      "step": 35039
    },
    {
      "epoch": 3.014712208552009,
      "grad_norm": 98.0,
      "learning_rate": 0.0003413569319332364,
      "loss": 1.25,
      "step": 35040
    },
    {
      "epoch": 3.0147982448593305,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00034133128862853346,
      "loss": 1.3605,
      "step": 35041
    },
    {
      "epoch": 3.0148842811666525,
      "grad_norm": 1.15625,
      "learning_rate": 0.00034130564578790576,
      "loss": 1.4142,
      "step": 35042
    },
    {
      "epoch": 3.014970317473974,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00034128000341142816,
      "loss": 1.0517,
      "step": 35043
    },
    {
      "epoch": 3.0150563537812958,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003412543614991756,
      "loss": 1.0453,
      "step": 35044
    },
    {
      "epoch": 3.0151423900886174,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003412287200512232,
      "loss": 0.8468,
      "step": 35045
    },
    {
      "epoch": 3.015228426395939,
      "grad_norm": 1.796875,
      "learning_rate": 0.000341203079067646,
      "loss": 1.5066,
      "step": 35046
    },
    {
      "epoch": 3.0153144627032606,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00034117743854851886,
      "loss": 0.8373,
      "step": 35047
    },
    {
      "epoch": 3.0154004990105823,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003411517984939167,
      "loss": 0.9645,
      "step": 35048
    },
    {
      "epoch": 3.0154865353179043,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00034112615890391474,
      "loss": 1.2287,
      "step": 35049
    },
    {
      "epoch": 3.015572571625226,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003411005197785879,
      "loss": 1.0804,
      "step": 35050
    },
    {
      "epoch": 3.0156586079325476,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00034107488111801104,
      "loss": 1.0564,
      "step": 35051
    },
    {
      "epoch": 3.015744644239869,
      "grad_norm": 0.765625,
      "learning_rate": 0.00034104924292225936,
      "loss": 0.7006,
      "step": 35052
    },
    {
      "epoch": 3.015830680547191,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0003410236051914077,
      "loss": 1.9358,
      "step": 35053
    },
    {
      "epoch": 3.0159167168545125,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00034099796792553107,
      "loss": 0.9923,
      "step": 35054
    },
    {
      "epoch": 3.016002753161834,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00034097233112470436,
      "loss": 0.7243,
      "step": 35055
    },
    {
      "epoch": 3.016088789469156,
      "grad_norm": 6.125,
      "learning_rate": 0.00034094669478900276,
      "loss": 1.3221,
      "step": 35056
    },
    {
      "epoch": 3.0161748257764778,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0003409210589185011,
      "loss": 1.514,
      "step": 35057
    },
    {
      "epoch": 3.0162608620837994,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003408954235132744,
      "loss": 1.8065,
      "step": 35058
    },
    {
      "epoch": 3.016346898391121,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00034086978857339765,
      "loss": 0.8401,
      "step": 35059
    },
    {
      "epoch": 3.0164329346984426,
      "grad_norm": 171.0,
      "learning_rate": 0.0003408441540989459,
      "loss": 1.3932,
      "step": 35060
    },
    {
      "epoch": 3.0165189710057643,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00034081852008999397,
      "loss": 1.3233,
      "step": 35061
    },
    {
      "epoch": 3.0166050073130863,
      "grad_norm": 1.59375,
      "learning_rate": 0.00034079288654661685,
      "loss": 1.2747,
      "step": 35062
    },
    {
      "epoch": 3.016691043620408,
      "grad_norm": 358.0,
      "learning_rate": 0.00034076725346888966,
      "loss": 0.6998,
      "step": 35063
    },
    {
      "epoch": 3.0167770799277296,
      "grad_norm": 4.375,
      "learning_rate": 0.0003407416208568873,
      "loss": 1.7202,
      "step": 35064
    },
    {
      "epoch": 3.016863116235051,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00034071598871068473,
      "loss": 0.3534,
      "step": 35065
    },
    {
      "epoch": 3.016949152542373,
      "grad_norm": 4.125,
      "learning_rate": 0.00034069035703035677,
      "loss": 1.2875,
      "step": 35066
    },
    {
      "epoch": 3.0170351888496945,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003406647258159786,
      "loss": 0.843,
      "step": 35067
    },
    {
      "epoch": 3.017121225157016,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00034063909506762527,
      "loss": 0.932,
      "step": 35068
    },
    {
      "epoch": 3.017207261464338,
      "grad_norm": 1.125,
      "learning_rate": 0.00034061346478537135,
      "loss": 0.7389,
      "step": 35069
    },
    {
      "epoch": 3.0172932977716598,
      "grad_norm": 2.953125,
      "learning_rate": 0.0003405878349692922,
      "loss": 1.3884,
      "step": 35070
    },
    {
      "epoch": 3.0173793340789814,
      "grad_norm": 1.59375,
      "learning_rate": 0.00034056220561946263,
      "loss": 1.1203,
      "step": 35071
    },
    {
      "epoch": 3.017465370386303,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003405365767359576,
      "loss": 1.6928,
      "step": 35072
    },
    {
      "epoch": 3.0175514066936246,
      "grad_norm": 1.09375,
      "learning_rate": 0.00034051094831885197,
      "loss": 1.604,
      "step": 35073
    },
    {
      "epoch": 3.0176374430009463,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003404853203682209,
      "loss": 0.8143,
      "step": 35074
    },
    {
      "epoch": 3.017723479308268,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00034045969288413905,
      "loss": 1.2936,
      "step": 35075
    },
    {
      "epoch": 3.01780951561559,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003404340658666817,
      "loss": 0.987,
      "step": 35076
    },
    {
      "epoch": 3.0178955519229116,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00034040843931592375,
      "loss": 0.7971,
      "step": 35077
    },
    {
      "epoch": 3.017981588230233,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00034038281323194,
      "loss": 0.8729,
      "step": 35078
    },
    {
      "epoch": 3.018067624537555,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003403571876148055,
      "loss": 1.9921,
      "step": 35079
    },
    {
      "epoch": 3.0181536608448765,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00034033156246459517,
      "loss": 1.2565,
      "step": 35080
    },
    {
      "epoch": 3.018239697152198,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00034030593778138396,
      "loss": 0.9014,
      "step": 35081
    },
    {
      "epoch": 3.0183257334595197,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003402803135652468,
      "loss": 0.7401,
      "step": 35082
    },
    {
      "epoch": 3.018411769766842,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00034025468981625876,
      "loss": 1.501,
      "step": 35083
    },
    {
      "epoch": 3.0184978060741634,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003402290665344947,
      "loss": 1.0826,
      "step": 35084
    },
    {
      "epoch": 3.018583842381485,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003402034437200294,
      "loss": 1.3327,
      "step": 35085
    },
    {
      "epoch": 3.0186698786888067,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00034017782137293803,
      "loss": 1.1215,
      "step": 35086
    },
    {
      "epoch": 3.0187559149961283,
      "grad_norm": 5.0625,
      "learning_rate": 0.00034015219949329545,
      "loss": 0.8591,
      "step": 35087
    },
    {
      "epoch": 3.01884195130345,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0003401265780811766,
      "loss": 0.566,
      "step": 35088
    },
    {
      "epoch": 3.0189279876107715,
      "grad_norm": 0.78125,
      "learning_rate": 0.00034010095713665636,
      "loss": 0.962,
      "step": 35089
    },
    {
      "epoch": 3.0190140239180936,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003400753366598098,
      "loss": 0.8803,
      "step": 35090
    },
    {
      "epoch": 3.019100060225415,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00034004971665071174,
      "loss": 0.749,
      "step": 35091
    },
    {
      "epoch": 3.019186096532737,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00034002409710943705,
      "loss": 0.8531,
      "step": 35092
    },
    {
      "epoch": 3.0192721328400585,
      "grad_norm": 37.25,
      "learning_rate": 0.00033999847803606087,
      "loss": 1.2227,
      "step": 35093
    },
    {
      "epoch": 3.01935816914738,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003399728594306581,
      "loss": 0.3672,
      "step": 35094
    },
    {
      "epoch": 3.0194442054547017,
      "grad_norm": 1.359375,
      "learning_rate": 0.00033994724129330354,
      "loss": 1.5708,
      "step": 35095
    },
    {
      "epoch": 3.019530241762024,
      "grad_norm": 18.875,
      "learning_rate": 0.000339921623624072,
      "loss": 0.6648,
      "step": 35096
    },
    {
      "epoch": 3.0196162780693454,
      "grad_norm": 3.296875,
      "learning_rate": 0.00033989600642303876,
      "loss": 0.8953,
      "step": 35097
    },
    {
      "epoch": 3.019702314376667,
      "grad_norm": 1.15625,
      "learning_rate": 0.00033987038969027856,
      "loss": 1.269,
      "step": 35098
    },
    {
      "epoch": 3.0197883506839887,
      "grad_norm": 2.09375,
      "learning_rate": 0.0003398447734258662,
      "loss": 1.28,
      "step": 35099
    },
    {
      "epoch": 3.0198743869913103,
      "grad_norm": 2.6875,
      "learning_rate": 0.00033981915762987686,
      "loss": 0.826,
      "step": 35100
    },
    {
      "epoch": 3.019960423298632,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00033979354230238536,
      "loss": 1.0285,
      "step": 35101
    },
    {
      "epoch": 3.0200464596059535,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003397679274434665,
      "loss": 0.7864,
      "step": 35102
    },
    {
      "epoch": 3.0201324959132756,
      "grad_norm": 1.09375,
      "learning_rate": 0.00033974231305319527,
      "loss": 0.8461,
      "step": 35103
    },
    {
      "epoch": 3.020218532220597,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003397166991316466,
      "loss": 0.5757,
      "step": 35104
    },
    {
      "epoch": 3.020304568527919,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003396910856788955,
      "loss": 0.7477,
      "step": 35105
    },
    {
      "epoch": 3.0203906048352405,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0003396654726950168,
      "loss": 0.6848,
      "step": 35106
    },
    {
      "epoch": 3.020476641142562,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003396398601800853,
      "loss": 1.8215,
      "step": 35107
    },
    {
      "epoch": 3.0205626774498837,
      "grad_norm": 8.375,
      "learning_rate": 0.0003396142481341761,
      "loss": 1.1829,
      "step": 35108
    },
    {
      "epoch": 3.0206487137572053,
      "grad_norm": 1.0,
      "learning_rate": 0.00033958863655736405,
      "loss": 0.8919,
      "step": 35109
    },
    {
      "epoch": 3.0207347500645274,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003395630254497239,
      "loss": 0.8488,
      "step": 35110
    },
    {
      "epoch": 3.020820786371849,
      "grad_norm": 17.75,
      "learning_rate": 0.00033953741481133085,
      "loss": 1.466,
      "step": 35111
    },
    {
      "epoch": 3.0209068226791707,
      "grad_norm": 1.3125,
      "learning_rate": 0.00033951180464225937,
      "loss": 1.4115,
      "step": 35112
    },
    {
      "epoch": 3.0209928589864923,
      "grad_norm": 2.234375,
      "learning_rate": 0.0003394861949425849,
      "loss": 0.7784,
      "step": 35113
    },
    {
      "epoch": 3.021078895293814,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00033946058571238217,
      "loss": 0.933,
      "step": 35114
    },
    {
      "epoch": 3.0211649316011355,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00033943497695172574,
      "loss": 1.6123,
      "step": 35115
    },
    {
      "epoch": 3.021250967908457,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000339409368660691,
      "loss": 1.0211,
      "step": 35116
    },
    {
      "epoch": 3.021337004215779,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003393837608393526,
      "loss": 1.2253,
      "step": 35117
    },
    {
      "epoch": 3.021423040523101,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003393581534877854,
      "loss": 0.9169,
      "step": 35118
    },
    {
      "epoch": 3.0215090768304225,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003393325466060643,
      "loss": 0.8005,
      "step": 35119
    },
    {
      "epoch": 3.021595113137744,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003393069401942644,
      "loss": 1.274,
      "step": 35120
    },
    {
      "epoch": 3.0216811494450657,
      "grad_norm": 3.6875,
      "learning_rate": 0.00033928133425246043,
      "loss": 1.7614,
      "step": 35121
    },
    {
      "epoch": 3.0217671857523873,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003392557287807271,
      "loss": 0.894,
      "step": 35122
    },
    {
      "epoch": 3.0218532220597094,
      "grad_norm": 1.1875,
      "learning_rate": 0.00033923012377913976,
      "loss": 1.2523,
      "step": 35123
    },
    {
      "epoch": 3.021939258367031,
      "grad_norm": 1.34375,
      "learning_rate": 0.00033920451924777294,
      "loss": 0.8083,
      "step": 35124
    },
    {
      "epoch": 3.0220252946743527,
      "grad_norm": 16.875,
      "learning_rate": 0.00033917891518670163,
      "loss": 1.5164,
      "step": 35125
    },
    {
      "epoch": 3.0221113309816743,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003391533115960006,
      "loss": 1.1684,
      "step": 35126
    },
    {
      "epoch": 3.022197367288996,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000339127708475745,
      "loss": 1.2646,
      "step": 35127
    },
    {
      "epoch": 3.0222834035963175,
      "grad_norm": 2.078125,
      "learning_rate": 0.00033910210582600956,
      "loss": 1.3381,
      "step": 35128
    },
    {
      "epoch": 3.022369439903639,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00033907650364686917,
      "loss": 0.6942,
      "step": 35129
    },
    {
      "epoch": 3.0224554762109612,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00033905090193839856,
      "loss": 1.3242,
      "step": 35130
    },
    {
      "epoch": 3.022541512518283,
      "grad_norm": 1.3125,
      "learning_rate": 0.00033902530070067287,
      "loss": 0.8303,
      "step": 35131
    },
    {
      "epoch": 3.0226275488256045,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003389996999337669,
      "loss": 0.5436,
      "step": 35132
    },
    {
      "epoch": 3.022713585132926,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003389740996377554,
      "loss": 0.7855,
      "step": 35133
    },
    {
      "epoch": 3.0227996214402477,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003389484998127134,
      "loss": 1.4416,
      "step": 35134
    },
    {
      "epoch": 3.0228856577475693,
      "grad_norm": 1.453125,
      "learning_rate": 0.00033892290045871575,
      "loss": 1.4323,
      "step": 35135
    },
    {
      "epoch": 3.022971694054891,
      "grad_norm": 2.40625,
      "learning_rate": 0.00033889730157583733,
      "loss": 0.7162,
      "step": 35136
    },
    {
      "epoch": 3.023057730362213,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003388717031641528,
      "loss": 0.8034,
      "step": 35137
    },
    {
      "epoch": 3.0231437666695347,
      "grad_norm": 2.828125,
      "learning_rate": 0.0003388461052237373,
      "loss": 1.0108,
      "step": 35138
    },
    {
      "epoch": 3.0232298029768563,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0003388205077546656,
      "loss": 0.6829,
      "step": 35139
    },
    {
      "epoch": 3.023315839284178,
      "grad_norm": 22.375,
      "learning_rate": 0.0003387949107570125,
      "loss": 0.8361,
      "step": 35140
    },
    {
      "epoch": 3.0234018755914995,
      "grad_norm": 1.359375,
      "learning_rate": 0.000338769314230853,
      "loss": 1.3821,
      "step": 35141
    },
    {
      "epoch": 3.023487911898821,
      "grad_norm": 1.4375,
      "learning_rate": 0.0003387437181762619,
      "loss": 0.8951,
      "step": 35142
    },
    {
      "epoch": 3.023573948206143,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00033871812259331414,
      "loss": 0.8815,
      "step": 35143
    },
    {
      "epoch": 3.023659984513465,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003386925274820843,
      "loss": 0.6579,
      "step": 35144
    },
    {
      "epoch": 3.0237460208207865,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003386669328426476,
      "loss": 0.7084,
      "step": 35145
    },
    {
      "epoch": 3.023832057128108,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00033864133867507874,
      "loss": 0.7663,
      "step": 35146
    },
    {
      "epoch": 3.0239180934354297,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003386157449794525,
      "loss": 1.5483,
      "step": 35147
    },
    {
      "epoch": 3.0240041297427513,
      "grad_norm": 1.015625,
      "learning_rate": 0.00033859015175584384,
      "loss": 1.9247,
      "step": 35148
    },
    {
      "epoch": 3.024090166050073,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003385645590043275,
      "loss": 1.5686,
      "step": 35149
    },
    {
      "epoch": 3.024176202357395,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00033853896672497854,
      "loss": 1.4534,
      "step": 35150
    },
    {
      "epoch": 3.0242622386647167,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00033851337491787173,
      "loss": 1.1796,
      "step": 35151
    },
    {
      "epoch": 3.0243482749720383,
      "grad_norm": 1.078125,
      "learning_rate": 0.00033848778358308183,
      "loss": 0.8411,
      "step": 35152
    },
    {
      "epoch": 3.02443431127936,
      "grad_norm": 0.984375,
      "learning_rate": 0.00033846219272068374,
      "loss": 0.9274,
      "step": 35153
    },
    {
      "epoch": 3.0245203475866815,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003384366023307524,
      "loss": 1.4089,
      "step": 35154
    },
    {
      "epoch": 3.024606383894003,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003384110124133626,
      "loss": 1.6132,
      "step": 35155
    },
    {
      "epoch": 3.024692420201325,
      "grad_norm": 1.203125,
      "learning_rate": 0.000338385422968589,
      "loss": 0.9734,
      "step": 35156
    },
    {
      "epoch": 3.024778456508647,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003383598339965067,
      "loss": 0.9925,
      "step": 35157
    },
    {
      "epoch": 3.0248644928159685,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003383342454971905,
      "loss": 0.6139,
      "step": 35158
    },
    {
      "epoch": 3.02495052912329,
      "grad_norm": 24.875,
      "learning_rate": 0.00033830865747071513,
      "loss": 0.9506,
      "step": 35159
    },
    {
      "epoch": 3.0250365654306117,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0003382830699171555,
      "loss": 1.5596,
      "step": 35160
    },
    {
      "epoch": 3.0251226017379333,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0003382574828365865,
      "loss": 0.6868,
      "step": 35161
    },
    {
      "epoch": 3.025208638045255,
      "grad_norm": 2.78125,
      "learning_rate": 0.00033823189622908284,
      "loss": 0.555,
      "step": 35162
    },
    {
      "epoch": 3.0252946743525766,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00033820631009471935,
      "loss": 0.9307,
      "step": 35163
    },
    {
      "epoch": 3.0253807106598987,
      "grad_norm": 1.1875,
      "learning_rate": 0.00033818072443357106,
      "loss": 0.5636,
      "step": 35164
    },
    {
      "epoch": 3.0254667469672203,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003381551392457126,
      "loss": 0.7346,
      "step": 35165
    },
    {
      "epoch": 3.025552783274542,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00033812955453121896,
      "loss": 0.9663,
      "step": 35166
    },
    {
      "epoch": 3.0256388195818635,
      "grad_norm": 7.53125,
      "learning_rate": 0.0003381039702901648,
      "loss": 0.8294,
      "step": 35167
    },
    {
      "epoch": 3.025724855889185,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0003380783865226251,
      "loss": 1.3203,
      "step": 35168
    },
    {
      "epoch": 3.025810892196507,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00033805280322867464,
      "loss": 1.1163,
      "step": 35169
    },
    {
      "epoch": 3.0258969285038284,
      "grad_norm": 3.359375,
      "learning_rate": 0.0003380272204083882,
      "loss": 0.8428,
      "step": 35170
    },
    {
      "epoch": 3.0259829648111505,
      "grad_norm": 1.53125,
      "learning_rate": 0.00033800163806184066,
      "loss": 1.4335,
      "step": 35171
    },
    {
      "epoch": 3.026069001118472,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00033797605618910684,
      "loss": 1.5842,
      "step": 35172
    },
    {
      "epoch": 3.0261550374257937,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003379504747902615,
      "loss": 1.3502,
      "step": 35173
    },
    {
      "epoch": 3.0262410737331153,
      "grad_norm": 205.0,
      "learning_rate": 0.00033792489386537947,
      "loss": 1.0613,
      "step": 35174
    },
    {
      "epoch": 3.026327110040437,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003378993134145356,
      "loss": 1.5132,
      "step": 35175
    },
    {
      "epoch": 3.0264131463477586,
      "grad_norm": 9.1875,
      "learning_rate": 0.00033787373343780483,
      "loss": 0.9438,
      "step": 35176
    },
    {
      "epoch": 3.0264991826550807,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003378481539352618,
      "loss": 0.8322,
      "step": 35177
    },
    {
      "epoch": 3.0265852189624023,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00033782257490698133,
      "loss": 0.6019,
      "step": 35178
    },
    {
      "epoch": 3.026671255269724,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00033779699635303834,
      "loss": 0.7416,
      "step": 35179
    },
    {
      "epoch": 3.0267572915770455,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0003377714182735077,
      "loss": 0.5441,
      "step": 35180
    },
    {
      "epoch": 3.026843327884367,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003377458406684638,
      "loss": 1.1632,
      "step": 35181
    },
    {
      "epoch": 3.026929364191689,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00033772026353798203,
      "loss": 0.8925,
      "step": 35182
    },
    {
      "epoch": 3.0270154004990104,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003376946868821368,
      "loss": 0.8576,
      "step": 35183
    },
    {
      "epoch": 3.0271014368063325,
      "grad_norm": 0.953125,
      "learning_rate": 0.00033766911070100313,
      "loss": 1.0813,
      "step": 35184
    },
    {
      "epoch": 3.027187473113654,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00033764353499465567,
      "loss": 1.61,
      "step": 35185
    },
    {
      "epoch": 3.0272735094209757,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00033761795976316913,
      "loss": 0.4189,
      "step": 35186
    },
    {
      "epoch": 3.0273595457282974,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0003375923850066187,
      "loss": 1.103,
      "step": 35187
    },
    {
      "epoch": 3.027445582035619,
      "grad_norm": 1.453125,
      "learning_rate": 0.00033756681072507895,
      "loss": 1.2877,
      "step": 35188
    },
    {
      "epoch": 3.0275316183429406,
      "grad_norm": 1.640625,
      "learning_rate": 0.0003375412369186247,
      "loss": 0.8545,
      "step": 35189
    },
    {
      "epoch": 3.0276176546502622,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003375156635873306,
      "loss": 1.0672,
      "step": 35190
    },
    {
      "epoch": 3.0277036909575843,
      "grad_norm": 3.765625,
      "learning_rate": 0.0003374900907312717,
      "loss": 1.1718,
      "step": 35191
    },
    {
      "epoch": 3.027789727264906,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003374645183505227,
      "loss": 1.0358,
      "step": 35192
    },
    {
      "epoch": 3.0278757635722275,
      "grad_norm": 1.21875,
      "learning_rate": 0.00033743894644515827,
      "loss": 0.9795,
      "step": 35193
    },
    {
      "epoch": 3.027961799879549,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003374133750152534,
      "loss": 1.4968,
      "step": 35194
    },
    {
      "epoch": 3.028047836186871,
      "grad_norm": 2.328125,
      "learning_rate": 0.0003373878040608828,
      "loss": 1.589,
      "step": 35195
    },
    {
      "epoch": 3.0281338724941924,
      "grad_norm": 2.6875,
      "learning_rate": 0.0003373622335821212,
      "loss": 1.4114,
      "step": 35196
    },
    {
      "epoch": 3.028219908801514,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00033733666357904337,
      "loss": 1.5843,
      "step": 35197
    },
    {
      "epoch": 3.028305945108836,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0003373110940517242,
      "loss": 1.7512,
      "step": 35198
    },
    {
      "epoch": 3.0283919814161577,
      "grad_norm": 0.84375,
      "learning_rate": 0.0003372855250002386,
      "loss": 0.9138,
      "step": 35199
    },
    {
      "epoch": 3.0284780177234794,
      "grad_norm": 3.890625,
      "learning_rate": 0.000337259956424661,
      "loss": 0.8935,
      "step": 35200
    },
    {
      "epoch": 3.028564054030801,
      "grad_norm": 4.375,
      "learning_rate": 0.0003372343883250664,
      "loss": 0.9177,
      "step": 35201
    },
    {
      "epoch": 3.0286500903381226,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003372088207015296,
      "loss": 0.5957,
      "step": 35202
    },
    {
      "epoch": 3.0287361266454442,
      "grad_norm": 0.8125,
      "learning_rate": 0.00033718325355412536,
      "loss": 0.791,
      "step": 35203
    },
    {
      "epoch": 3.028822162952766,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00033715768688292834,
      "loss": 0.8758,
      "step": 35204
    },
    {
      "epoch": 3.028908199260088,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003371321206880135,
      "loss": 1.5307,
      "step": 35205
    },
    {
      "epoch": 3.0289942355674095,
      "grad_norm": 1.0625,
      "learning_rate": 0.00033710655496945553,
      "loss": 1.5401,
      "step": 35206
    },
    {
      "epoch": 3.029080271874731,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00033708098972732925,
      "loss": 1.0538,
      "step": 35207
    },
    {
      "epoch": 3.029166308182053,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00033705542496170927,
      "loss": 1.351,
      "step": 35208
    },
    {
      "epoch": 3.0292523444893744,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00033702986067267053,
      "loss": 1.6398,
      "step": 35209
    },
    {
      "epoch": 3.029338380796696,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003370042968602878,
      "loss": 0.8196,
      "step": 35210
    },
    {
      "epoch": 3.029424417104018,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003369787335246357,
      "loss": 1.314,
      "step": 35211
    },
    {
      "epoch": 3.0295104534113397,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003369531706657892,
      "loss": 0.4952,
      "step": 35212
    },
    {
      "epoch": 3.0295964897186614,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00033692760828382294,
      "loss": 0.8683,
      "step": 35213
    },
    {
      "epoch": 3.029682526025983,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00033690204637881174,
      "loss": 0.9349,
      "step": 35214
    },
    {
      "epoch": 3.0297685623333046,
      "grad_norm": 1.21875,
      "learning_rate": 0.00033687648495083026,
      "loss": 2.0616,
      "step": 35215
    },
    {
      "epoch": 3.0298545986406262,
      "grad_norm": 5.875,
      "learning_rate": 0.0003368509239999533,
      "loss": 1.4729,
      "step": 35216
    },
    {
      "epoch": 3.029940634947948,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0003368253635262558,
      "loss": 0.8532,
      "step": 35217
    },
    {
      "epoch": 3.03002667125527,
      "grad_norm": 1.15625,
      "learning_rate": 0.00033679980352981233,
      "loss": 1.6453,
      "step": 35218
    },
    {
      "epoch": 3.0301127075625915,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003367742440106976,
      "loss": 0.7785,
      "step": 35219
    },
    {
      "epoch": 3.030198743869913,
      "grad_norm": 2.609375,
      "learning_rate": 0.0003367486849689865,
      "loss": 1.4848,
      "step": 35220
    },
    {
      "epoch": 3.030284780177235,
      "grad_norm": 1.5,
      "learning_rate": 0.0003367231264047538,
      "loss": 1.0387,
      "step": 35221
    },
    {
      "epoch": 3.0303708164845564,
      "grad_norm": 1.921875,
      "learning_rate": 0.0003366975683180741,
      "loss": 1.4292,
      "step": 35222
    },
    {
      "epoch": 3.030456852791878,
      "grad_norm": 1.328125,
      "learning_rate": 0.00033667201070902226,
      "loss": 1.2453,
      "step": 35223
    },
    {
      "epoch": 3.0305428890991997,
      "grad_norm": 1.1875,
      "learning_rate": 0.000336646453577673,
      "loss": 0.9113,
      "step": 35224
    },
    {
      "epoch": 3.0306289254065217,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003366208969241012,
      "loss": 1.1922,
      "step": 35225
    },
    {
      "epoch": 3.0307149617138434,
      "grad_norm": 2.78125,
      "learning_rate": 0.0003365953407483815,
      "loss": 1.4426,
      "step": 35226
    },
    {
      "epoch": 3.030800998021165,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00033656978505058855,
      "loss": 0.7419,
      "step": 35227
    },
    {
      "epoch": 3.0308870343284866,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003365442298307973,
      "loss": 0.8141,
      "step": 35228
    },
    {
      "epoch": 3.0309730706358082,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003365186750890824,
      "loss": 1.2267,
      "step": 35229
    },
    {
      "epoch": 3.03105910694313,
      "grad_norm": 1.015625,
      "learning_rate": 0.00033649312082551853,
      "loss": 1.2376,
      "step": 35230
    },
    {
      "epoch": 3.0311451432504515,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00033646756704018045,
      "loss": 1.5099,
      "step": 35231
    },
    {
      "epoch": 3.0312311795577735,
      "grad_norm": 1.140625,
      "learning_rate": 0.000336442013733143,
      "loss": 1.6352,
      "step": 35232
    },
    {
      "epoch": 3.031317215865095,
      "grad_norm": 0.96875,
      "learning_rate": 0.00033641646090448084,
      "loss": 1.3639,
      "step": 35233
    },
    {
      "epoch": 3.031403252172417,
      "grad_norm": 0.9375,
      "learning_rate": 0.00033639090855426855,
      "loss": 0.678,
      "step": 35234
    },
    {
      "epoch": 3.0314892884797384,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003363653566825813,
      "loss": 0.8485,
      "step": 35235
    },
    {
      "epoch": 3.03157532478706,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003363398052894934,
      "loss": 0.6184,
      "step": 35236
    },
    {
      "epoch": 3.0316613610943817,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0003363142543750798,
      "loss": 1.4266,
      "step": 35237
    },
    {
      "epoch": 3.0317473974017037,
      "grad_norm": 1.078125,
      "learning_rate": 0.00033628870393941513,
      "loss": 1.5461,
      "step": 35238
    },
    {
      "epoch": 3.0318334337090254,
      "grad_norm": 2.375,
      "learning_rate": 0.0003362631539825742,
      "loss": 0.8104,
      "step": 35239
    },
    {
      "epoch": 3.031919470016347,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003362376045046317,
      "loss": 0.7987,
      "step": 35240
    },
    {
      "epoch": 3.0320055063236686,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00033621205550566236,
      "loss": 1.5002,
      "step": 35241
    },
    {
      "epoch": 3.0320915426309902,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00033618650698574075,
      "loss": 1.1389,
      "step": 35242
    },
    {
      "epoch": 3.032177578938312,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003361609589449419,
      "loss": 0.9703,
      "step": 35243
    },
    {
      "epoch": 3.0322636152456335,
      "grad_norm": 1.046875,
      "learning_rate": 0.00033613541138334036,
      "loss": 0.7857,
      "step": 35244
    },
    {
      "epoch": 3.0323496515529555,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003361098643010108,
      "loss": 1.3361,
      "step": 35245
    },
    {
      "epoch": 3.032435687860277,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003360843176980281,
      "loss": 1.9556,
      "step": 35246
    },
    {
      "epoch": 3.032521724167599,
      "grad_norm": 1.4375,
      "learning_rate": 0.0003360587715744669,
      "loss": 1.1211,
      "step": 35247
    },
    {
      "epoch": 3.0326077604749204,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003360332259304019,
      "loss": 0.9497,
      "step": 35248
    },
    {
      "epoch": 3.032693796782242,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003360076807659077,
      "loss": 0.4609,
      "step": 35249
    },
    {
      "epoch": 3.0327798330895637,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003359821360810593,
      "loss": 1.7946,
      "step": 35250
    },
    {
      "epoch": 3.0328658693968853,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003359565918759312,
      "loss": 1.7921,
      "step": 35251
    },
    {
      "epoch": 3.0329519057042074,
      "grad_norm": 3.578125,
      "learning_rate": 0.0003359310481505981,
      "loss": 0.9051,
      "step": 35252
    },
    {
      "epoch": 3.033037942011529,
      "grad_norm": 1.703125,
      "learning_rate": 0.00033590550490513485,
      "loss": 1.7232,
      "step": 35253
    },
    {
      "epoch": 3.0331239783188506,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00033587996213961616,
      "loss": 1.5816,
      "step": 35254
    },
    {
      "epoch": 3.0332100146261722,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00033585441985411655,
      "loss": 1.2066,
      "step": 35255
    },
    {
      "epoch": 3.033296050933494,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00033582887804871076,
      "loss": 1.4372,
      "step": 35256
    },
    {
      "epoch": 3.0333820872408155,
      "grad_norm": 1.34375,
      "learning_rate": 0.00033580333672347367,
      "loss": 1.7357,
      "step": 35257
    },
    {
      "epoch": 3.033468123548137,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003357777958784799,
      "loss": 1.1212,
      "step": 35258
    },
    {
      "epoch": 3.033554159855459,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00033575225551380405,
      "loss": 0.7181,
      "step": 35259
    },
    {
      "epoch": 3.033640196162781,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00033572671562952087,
      "loss": 0.8987,
      "step": 35260
    },
    {
      "epoch": 3.0337262324701024,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003357011762257052,
      "loss": 0.5434,
      "step": 35261
    },
    {
      "epoch": 3.033812268777424,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00033567563730243165,
      "loss": 1.0613,
      "step": 35262
    },
    {
      "epoch": 3.0338983050847457,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0003356500988597749,
      "loss": 0.873,
      "step": 35263
    },
    {
      "epoch": 3.0339843413920673,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00033562456089780967,
      "loss": 0.7519,
      "step": 35264
    },
    {
      "epoch": 3.034070377699389,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003355990234166105,
      "loss": 1.4045,
      "step": 35265
    },
    {
      "epoch": 3.034156414006711,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00033557348641625236,
      "loss": 0.8371,
      "step": 35266
    },
    {
      "epoch": 3.0342424503140326,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00033554794989680976,
      "loss": 0.8257,
      "step": 35267
    },
    {
      "epoch": 3.0343284866213542,
      "grad_norm": 1.390625,
      "learning_rate": 0.00033552241385835735,
      "loss": 2.0801,
      "step": 35268
    },
    {
      "epoch": 3.034414522928676,
      "grad_norm": 0.984375,
      "learning_rate": 0.00033549687830097,
      "loss": 0.9885,
      "step": 35269
    },
    {
      "epoch": 3.0345005592359975,
      "grad_norm": 0.875,
      "learning_rate": 0.0003354713432247222,
      "loss": 0.7623,
      "step": 35270
    },
    {
      "epoch": 3.034586595543319,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003354458086296888,
      "loss": 1.8432,
      "step": 35271
    },
    {
      "epoch": 3.034672631850641,
      "grad_norm": 13.375,
      "learning_rate": 0.0003354202745159443,
      "loss": 0.9602,
      "step": 35272
    },
    {
      "epoch": 3.034758668157963,
      "grad_norm": 4.90625,
      "learning_rate": 0.0003353947408835635,
      "loss": 0.7956,
      "step": 35273
    },
    {
      "epoch": 3.0348447044652844,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00033536920773262116,
      "loss": 0.5834,
      "step": 35274
    },
    {
      "epoch": 3.034930740772606,
      "grad_norm": 196.0,
      "learning_rate": 0.00033534367506319175,
      "loss": 1.7497,
      "step": 35275
    },
    {
      "epoch": 3.0350167770799277,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003353181428753502,
      "loss": 0.8325,
      "step": 35276
    },
    {
      "epoch": 3.0351028133872493,
      "grad_norm": 2.28125,
      "learning_rate": 0.000335292611169171,
      "loss": 1.5235,
      "step": 35277
    },
    {
      "epoch": 3.035188849694571,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00033526707994472894,
      "loss": 0.5862,
      "step": 35278
    },
    {
      "epoch": 3.035274886001893,
      "grad_norm": 0.765625,
      "learning_rate": 0.00033524154920209847,
      "loss": 0.3347,
      "step": 35279
    },
    {
      "epoch": 3.0353609223092146,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00033521601894135456,
      "loss": 1.1905,
      "step": 35280
    },
    {
      "epoch": 3.0354469586165362,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003351904891625717,
      "loss": 0.7425,
      "step": 35281
    },
    {
      "epoch": 3.035532994923858,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0003351649598658246,
      "loss": 1.2953,
      "step": 35282
    },
    {
      "epoch": 3.0356190312311795,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00033513943105118794,
      "loss": 0.9616,
      "step": 35283
    },
    {
      "epoch": 3.035705067538501,
      "grad_norm": 1.703125,
      "learning_rate": 0.00033511390271873635,
      "loss": 0.5158,
      "step": 35284
    },
    {
      "epoch": 3.0357911038458227,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003350883748685446,
      "loss": 1.0086,
      "step": 35285
    },
    {
      "epoch": 3.035877140153145,
      "grad_norm": 0.921875,
      "learning_rate": 0.00033506284750068715,
      "loss": 0.6569,
      "step": 35286
    },
    {
      "epoch": 3.0359631764604664,
      "grad_norm": 1.5625,
      "learning_rate": 0.00033503732061523895,
      "loss": 1.6776,
      "step": 35287
    },
    {
      "epoch": 3.036049212767788,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003350117942122745,
      "loss": 0.934,
      "step": 35288
    },
    {
      "epoch": 3.0361352490751097,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0003349862682918684,
      "loss": 0.8006,
      "step": 35289
    },
    {
      "epoch": 3.0362212853824313,
      "grad_norm": 1.046875,
      "learning_rate": 0.00033496074285409527,
      "loss": 0.5813,
      "step": 35290
    },
    {
      "epoch": 3.036307321689753,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00033493521789903005,
      "loss": 1.2034,
      "step": 35291
    },
    {
      "epoch": 3.0363933579970745,
      "grad_norm": 1.171875,
      "learning_rate": 0.00033490969342674716,
      "loss": 0.8726,
      "step": 35292
    },
    {
      "epoch": 3.0364793943043966,
      "grad_norm": 0.859375,
      "learning_rate": 0.00033488416943732114,
      "loss": 0.7709,
      "step": 35293
    },
    {
      "epoch": 3.0365654306117182,
      "grad_norm": 88.5,
      "learning_rate": 0.00033485864593082705,
      "loss": 1.3911,
      "step": 35294
    },
    {
      "epoch": 3.03665146691904,
      "grad_norm": 2.203125,
      "learning_rate": 0.0003348331229073392,
      "loss": 0.7408,
      "step": 35295
    },
    {
      "epoch": 3.0367375032263615,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00033480760036693234,
      "loss": 0.7204,
      "step": 35296
    },
    {
      "epoch": 3.036823539533683,
      "grad_norm": 1.109375,
      "learning_rate": 0.000334782078309681,
      "loss": 1.2616,
      "step": 35297
    },
    {
      "epoch": 3.0369095758410047,
      "grad_norm": 0.9375,
      "learning_rate": 0.00033475655673566,
      "loss": 0.9831,
      "step": 35298
    },
    {
      "epoch": 3.036995612148327,
      "grad_norm": 29.625,
      "learning_rate": 0.000334731035644944,
      "loss": 1.2518,
      "step": 35299
    },
    {
      "epoch": 3.0370816484556484,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00033470551503760764,
      "loss": 1.5257,
      "step": 35300
    },
    {
      "epoch": 3.03716768476297,
      "grad_norm": 1.859375,
      "learning_rate": 0.00033467999491372546,
      "loss": 1.7458,
      "step": 35301
    },
    {
      "epoch": 3.0372537210702917,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00033465447527337213,
      "loss": 0.8011,
      "step": 35302
    },
    {
      "epoch": 3.0373397573776133,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00033462895611662224,
      "loss": 1.5683,
      "step": 35303
    },
    {
      "epoch": 3.037425793684935,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003346034374435506,
      "loss": 0.3416,
      "step": 35304
    },
    {
      "epoch": 3.0375118299922566,
      "grad_norm": 3744.0,
      "learning_rate": 0.0003345779192542318,
      "loss": 0.7853,
      "step": 35305
    },
    {
      "epoch": 3.0375978662995786,
      "grad_norm": 1.1875,
      "learning_rate": 0.00033455240154874017,
      "loss": 1.0485,
      "step": 35306
    },
    {
      "epoch": 3.0376839026069002,
      "grad_norm": 3.015625,
      "learning_rate": 0.0003345268843271507,
      "loss": 1.207,
      "step": 35307
    },
    {
      "epoch": 3.037769938914222,
      "grad_norm": 0.90625,
      "learning_rate": 0.000334501367589538,
      "loss": 1.1139,
      "step": 35308
    },
    {
      "epoch": 3.0378559752215435,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00033447585133597647,
      "loss": 1.0558,
      "step": 35309
    },
    {
      "epoch": 3.037942011528865,
      "grad_norm": 1.109375,
      "learning_rate": 0.00033445033556654105,
      "loss": 1.0225,
      "step": 35310
    },
    {
      "epoch": 3.0380280478361867,
      "grad_norm": 1.34375,
      "learning_rate": 0.00033442482028130604,
      "loss": 1.3163,
      "step": 35311
    },
    {
      "epoch": 3.0381140841435084,
      "grad_norm": 1.578125,
      "learning_rate": 0.00033439930548034635,
      "loss": 1.2521,
      "step": 35312
    },
    {
      "epoch": 3.0382001204508304,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00033437379116373636,
      "loss": 1.2887,
      "step": 35313
    },
    {
      "epoch": 3.038286156758152,
      "grad_norm": 0.734375,
      "learning_rate": 0.00033434827733155094,
      "loss": 0.6156,
      "step": 35314
    },
    {
      "epoch": 3.0383721930654737,
      "grad_norm": 1.25,
      "learning_rate": 0.0003343227639838645,
      "loss": 1.6744,
      "step": 35315
    },
    {
      "epoch": 3.0384582293727953,
      "grad_norm": 1.828125,
      "learning_rate": 0.00033429725112075174,
      "loss": 1.0901,
      "step": 35316
    },
    {
      "epoch": 3.038544265680117,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0003342717387422874,
      "loss": 0.6759,
      "step": 35317
    },
    {
      "epoch": 3.0386303019874386,
      "grad_norm": 1.25,
      "learning_rate": 0.000334246226848546,
      "loss": 0.8333,
      "step": 35318
    },
    {
      "epoch": 3.03871633829476,
      "grad_norm": 1.625,
      "learning_rate": 0.0003342207154396021,
      "loss": 0.9354,
      "step": 35319
    },
    {
      "epoch": 3.0388023746020822,
      "grad_norm": 1.578125,
      "learning_rate": 0.00033419520451553024,
      "loss": 0.8447,
      "step": 35320
    },
    {
      "epoch": 3.038888410909404,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003341696940764053,
      "loss": 0.7833,
      "step": 35321
    },
    {
      "epoch": 3.0389744472167255,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00033414418412230175,
      "loss": 1.2773,
      "step": 35322
    },
    {
      "epoch": 3.039060483524047,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00033411867465329417,
      "loss": 1.4016,
      "step": 35323
    },
    {
      "epoch": 3.0391465198313687,
      "grad_norm": 2.078125,
      "learning_rate": 0.0003340931656694572,
      "loss": 0.9639,
      "step": 35324
    },
    {
      "epoch": 3.0392325561386904,
      "grad_norm": 1.03125,
      "learning_rate": 0.00033406765717086553,
      "loss": 0.7728,
      "step": 35325
    },
    {
      "epoch": 3.0393185924460124,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003340421491575936,
      "loss": 0.7827,
      "step": 35326
    },
    {
      "epoch": 3.039404628753334,
      "grad_norm": 1.34375,
      "learning_rate": 0.00033401664162971607,
      "loss": 1.1565,
      "step": 35327
    },
    {
      "epoch": 3.0394906650606557,
      "grad_norm": 1.109375,
      "learning_rate": 0.00033399113458730767,
      "loss": 1.0127,
      "step": 35328
    },
    {
      "epoch": 3.0395767013679773,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00033396562803044295,
      "loss": 0.4912,
      "step": 35329
    },
    {
      "epoch": 3.039662737675299,
      "grad_norm": 16.5,
      "learning_rate": 0.00033394012195919645,
      "loss": 1.0593,
      "step": 35330
    },
    {
      "epoch": 3.0397487739826206,
      "grad_norm": 2.078125,
      "learning_rate": 0.0003339146163736426,
      "loss": 0.9722,
      "step": 35331
    },
    {
      "epoch": 3.039834810289942,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003338891112738565,
      "loss": 1.6546,
      "step": 35332
    },
    {
      "epoch": 3.0399208465972642,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003338636066599123,
      "loss": 0.823,
      "step": 35333
    },
    {
      "epoch": 3.040006882904586,
      "grad_norm": 1.328125,
      "learning_rate": 0.0003338381025318845,
      "loss": 0.6563,
      "step": 35334
    },
    {
      "epoch": 3.0400929192119075,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00033381259888984826,
      "loss": 1.0865,
      "step": 35335
    },
    {
      "epoch": 3.040178955519229,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003337870957338777,
      "loss": 0.9565,
      "step": 35336
    },
    {
      "epoch": 3.0402649918265507,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00033376159306404765,
      "loss": 1.3401,
      "step": 35337
    },
    {
      "epoch": 3.0403510281338724,
      "grad_norm": 1.4375,
      "learning_rate": 0.0003337360908804326,
      "loss": 1.2934,
      "step": 35338
    },
    {
      "epoch": 3.040437064441194,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00033371058918310706,
      "loss": 0.9855,
      "step": 35339
    },
    {
      "epoch": 3.040523100748516,
      "grad_norm": 14.25,
      "learning_rate": 0.00033368508797214575,
      "loss": 1.7039,
      "step": 35340
    },
    {
      "epoch": 3.0406091370558377,
      "grad_norm": 1.15625,
      "learning_rate": 0.00033365958724762324,
      "loss": 0.76,
      "step": 35341
    },
    {
      "epoch": 3.0406951733631593,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003336340870096141,
      "loss": 0.7247,
      "step": 35342
    },
    {
      "epoch": 3.040781209670481,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003336085872581928,
      "loss": 0.9243,
      "step": 35343
    },
    {
      "epoch": 3.0408672459778026,
      "grad_norm": 5.0,
      "learning_rate": 0.00033358308799343413,
      "loss": 1.4816,
      "step": 35344
    },
    {
      "epoch": 3.040953282285124,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003335575892154126,
      "loss": 1.0496,
      "step": 35345
    },
    {
      "epoch": 3.041039318592446,
      "grad_norm": 1.0625,
      "learning_rate": 0.00033353209092420257,
      "loss": 0.8352,
      "step": 35346
    },
    {
      "epoch": 3.041125354899768,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00033350659311987896,
      "loss": 1.4445,
      "step": 35347
    },
    {
      "epoch": 3.0412113912070895,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003334810958025161,
      "loss": 1.3532,
      "step": 35348
    },
    {
      "epoch": 3.041297427514411,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003334555989721887,
      "loss": 1.5696,
      "step": 35349
    },
    {
      "epoch": 3.0413834638217327,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0003334301026289712,
      "loss": 0.4918,
      "step": 35350
    },
    {
      "epoch": 3.0414695001290544,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00033340460677293835,
      "loss": 1.0407,
      "step": 35351
    },
    {
      "epoch": 3.041555536436376,
      "grad_norm": 0.96875,
      "learning_rate": 0.00033337911140416467,
      "loss": 0.627,
      "step": 35352
    },
    {
      "epoch": 3.041641572743698,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003333536165227246,
      "loss": 1.1516,
      "step": 35353
    },
    {
      "epoch": 3.0417276090510197,
      "grad_norm": 1.234375,
      "learning_rate": 0.00033332812212869266,
      "loss": 1.1513,
      "step": 35354
    },
    {
      "epoch": 3.0418136453583413,
      "grad_norm": 17.25,
      "learning_rate": 0.00033330262822214374,
      "loss": 1.03,
      "step": 35355
    },
    {
      "epoch": 3.041899681665663,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00033327713480315214,
      "loss": 0.9867,
      "step": 35356
    },
    {
      "epoch": 3.0419857179729846,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003332516418717925,
      "loss": 2.0484,
      "step": 35357
    },
    {
      "epoch": 3.042071754280306,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00033322614942813935,
      "loss": 0.7533,
      "step": 35358
    },
    {
      "epoch": 3.042157790587628,
      "grad_norm": 1.4375,
      "learning_rate": 0.00033320065747226735,
      "loss": 1.0046,
      "step": 35359
    },
    {
      "epoch": 3.04224382689495,
      "grad_norm": 1.59375,
      "learning_rate": 0.00033317516600425104,
      "loss": 1.6699,
      "step": 35360
    },
    {
      "epoch": 3.0423298632022715,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0003331496750241647,
      "loss": 1.1018,
      "step": 35361
    },
    {
      "epoch": 3.042415899509593,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00033312418453208326,
      "loss": 0.5545,
      "step": 35362
    },
    {
      "epoch": 3.0425019358169147,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003330986945280812,
      "loss": 0.8938,
      "step": 35363
    },
    {
      "epoch": 3.0425879721242364,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00033307320501223275,
      "loss": 1.3705,
      "step": 35364
    },
    {
      "epoch": 3.042674008431558,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003330477159846129,
      "loss": 1.4513,
      "step": 35365
    },
    {
      "epoch": 3.0427600447388796,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000333022227445296,
      "loss": 1.2861,
      "step": 35366
    },
    {
      "epoch": 3.0428460810462017,
      "grad_norm": 17.25,
      "learning_rate": 0.0003329967393943566,
      "loss": 0.9447,
      "step": 35367
    },
    {
      "epoch": 3.0429321173535233,
      "grad_norm": 10.375,
      "learning_rate": 0.00033297125183186915,
      "loss": 0.8204,
      "step": 35368
    },
    {
      "epoch": 3.043018153660845,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00033294576475790844,
      "loss": 0.9131,
      "step": 35369
    },
    {
      "epoch": 3.0431041899681666,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003329202781725489,
      "loss": 1.0242,
      "step": 35370
    },
    {
      "epoch": 3.043190226275488,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003328947920758648,
      "loss": 0.6063,
      "step": 35371
    },
    {
      "epoch": 3.04327626258281,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003328693064679311,
      "loss": 1.7809,
      "step": 35372
    },
    {
      "epoch": 3.0433622988901314,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00033284382134882206,
      "loss": 0.8335,
      "step": 35373
    },
    {
      "epoch": 3.0434483351974535,
      "grad_norm": 1.5625,
      "learning_rate": 0.0003328183367186125,
      "loss": 1.0935,
      "step": 35374
    },
    {
      "epoch": 3.043534371504775,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003327928525773768,
      "loss": 1.0231,
      "step": 35375
    },
    {
      "epoch": 3.0436204078120968,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003327673689251895,
      "loss": 0.9779,
      "step": 35376
    },
    {
      "epoch": 3.0437064441194184,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00033274188576212495,
      "loss": 1.2464,
      "step": 35377
    },
    {
      "epoch": 3.04379248042674,
      "grad_norm": 2.046875,
      "learning_rate": 0.000332716403088258,
      "loss": 1.503,
      "step": 35378
    },
    {
      "epoch": 3.0438785167340616,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000332690920903663,
      "loss": 1.4063,
      "step": 35379
    },
    {
      "epoch": 3.0439645530413832,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003326654392084145,
      "loss": 0.7628,
      "step": 35380
    },
    {
      "epoch": 3.0440505893487053,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00033263995800258706,
      "loss": 0.7468,
      "step": 35381
    },
    {
      "epoch": 3.044136625656027,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00033261447728625523,
      "loss": 1.695,
      "step": 35382
    },
    {
      "epoch": 3.0442226619633486,
      "grad_norm": 2.296875,
      "learning_rate": 0.0003325889970594935,
      "loss": 0.8718,
      "step": 35383
    },
    {
      "epoch": 3.04430869827067,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003325635173223762,
      "loss": 1.033,
      "step": 35384
    },
    {
      "epoch": 3.044394734577992,
      "grad_norm": 2.5,
      "learning_rate": 0.0003325380380749783,
      "loss": 0.9312,
      "step": 35385
    },
    {
      "epoch": 3.0444807708853134,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000332512559317374,
      "loss": 0.6592,
      "step": 35386
    },
    {
      "epoch": 3.0445668071926355,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0003324870810496378,
      "loss": 1.1267,
      "step": 35387
    },
    {
      "epoch": 3.044652843499957,
      "grad_norm": 1.203125,
      "learning_rate": 0.00033246160327184436,
      "loss": 1.9588,
      "step": 35388
    },
    {
      "epoch": 3.0447388798072788,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003324361259840682,
      "loss": 1.3608,
      "step": 35389
    },
    {
      "epoch": 3.0448249161146004,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00033241064918638374,
      "loss": 1.172,
      "step": 35390
    },
    {
      "epoch": 3.044910952421922,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0003323851728788655,
      "loss": 0.7707,
      "step": 35391
    },
    {
      "epoch": 3.0449969887292436,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003323596970615881,
      "loss": 1.5581,
      "step": 35392
    },
    {
      "epoch": 3.0450830250365652,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00033233422173462593,
      "loss": 0.9888,
      "step": 35393
    },
    {
      "epoch": 3.0451690613438873,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00033230874689805357,
      "loss": 2.0507,
      "step": 35394
    },
    {
      "epoch": 3.045255097651209,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003322832725519455,
      "loss": 1.0982,
      "step": 35395
    },
    {
      "epoch": 3.0453411339585306,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0003322577986963762,
      "loss": 1.1598,
      "step": 35396
    },
    {
      "epoch": 3.045427170265852,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00033223232533142035,
      "loss": 0.7512,
      "step": 35397
    },
    {
      "epoch": 3.045513206573174,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00033220685245715216,
      "loss": 0.8161,
      "step": 35398
    },
    {
      "epoch": 3.0455992428804954,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003321813800736464,
      "loss": 1.7971,
      "step": 35399
    },
    {
      "epoch": 3.045685279187817,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003321559081809774,
      "loss": 0.8327,
      "step": 35400
    },
    {
      "epoch": 3.045771315495139,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003321304367792198,
      "loss": 1.2676,
      "step": 35401
    },
    {
      "epoch": 3.0458573518024608,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003321049658684479,
      "loss": 1.0125,
      "step": 35402
    },
    {
      "epoch": 3.0459433881097824,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00033207949544873644,
      "loss": 0.6728,
      "step": 35403
    },
    {
      "epoch": 3.046029424417104,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003320540255201598,
      "loss": 0.9732,
      "step": 35404
    },
    {
      "epoch": 3.0461154607244256,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003320285560827924,
      "loss": 0.877,
      "step": 35405
    },
    {
      "epoch": 3.0462014970317473,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003320030871367088,
      "loss": 0.7427,
      "step": 35406
    },
    {
      "epoch": 3.046287533339069,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003319776186819836,
      "loss": 0.9004,
      "step": 35407
    },
    {
      "epoch": 3.046373569646391,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003319521507186909,
      "loss": 1.32,
      "step": 35408
    },
    {
      "epoch": 3.0464596059537126,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003319266832469057,
      "loss": 0.9453,
      "step": 35409
    },
    {
      "epoch": 3.046545642261034,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0003319012162667022,
      "loss": 1.1373,
      "step": 35410
    },
    {
      "epoch": 3.046631678568356,
      "grad_norm": 1.28125,
      "learning_rate": 0.00033187574977815507,
      "loss": 0.8968,
      "step": 35411
    },
    {
      "epoch": 3.0467177148756774,
      "grad_norm": 1.046875,
      "learning_rate": 0.00033185028378133856,
      "loss": 1.1777,
      "step": 35412
    },
    {
      "epoch": 3.046803751182999,
      "grad_norm": 1.125,
      "learning_rate": 0.0003318248182763274,
      "loss": 1.1323,
      "step": 35413
    },
    {
      "epoch": 3.046889787490321,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003317993532631958,
      "loss": 1.141,
      "step": 35414
    },
    {
      "epoch": 3.0469758237976428,
      "grad_norm": 2.46875,
      "learning_rate": 0.0003317738887420184,
      "loss": 0.7368,
      "step": 35415
    },
    {
      "epoch": 3.0470618601049644,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00033174842471286973,
      "loss": 0.5463,
      "step": 35416
    },
    {
      "epoch": 3.047147896412286,
      "grad_norm": 1.3125,
      "learning_rate": 0.00033172296117582416,
      "loss": 1.4213,
      "step": 35417
    },
    {
      "epoch": 3.0472339327196076,
      "grad_norm": 2.65625,
      "learning_rate": 0.0003316974981309561,
      "loss": 0.5573,
      "step": 35418
    },
    {
      "epoch": 3.0473199690269293,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003316720355783403,
      "loss": 0.8154,
      "step": 35419
    },
    {
      "epoch": 3.047406005334251,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003316465735180509,
      "loss": 1.6046,
      "step": 35420
    },
    {
      "epoch": 3.047492041641573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00033162111195016267,
      "loss": 0.9187,
      "step": 35421
    },
    {
      "epoch": 3.0475780779488946,
      "grad_norm": 244.0,
      "learning_rate": 0.00033159565087474983,
      "loss": 1.0593,
      "step": 35422
    },
    {
      "epoch": 3.047664114256216,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00033157019029188706,
      "loss": 0.8461,
      "step": 35423
    },
    {
      "epoch": 3.047750150563538,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0003315447302016487,
      "loss": 0.7024,
      "step": 35424
    },
    {
      "epoch": 3.0478361868708594,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003315192706041091,
      "loss": 0.8057,
      "step": 35425
    },
    {
      "epoch": 3.047922223178181,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000331493811499343,
      "loss": 1.0786,
      "step": 35426
    },
    {
      "epoch": 3.0480082594855027,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0003314683528874247,
      "loss": 0.9851,
      "step": 35427
    },
    {
      "epoch": 3.0480942957928248,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003314428947684287,
      "loss": 1.1244,
      "step": 35428
    },
    {
      "epoch": 3.0481803321001464,
      "grad_norm": 4.75,
      "learning_rate": 0.00033141743714242943,
      "loss": 1.6169,
      "step": 35429
    },
    {
      "epoch": 3.048266368407468,
      "grad_norm": 4.25,
      "learning_rate": 0.00033139198000950145,
      "loss": 1.5026,
      "step": 35430
    },
    {
      "epoch": 3.0483524047147896,
      "grad_norm": 181.0,
      "learning_rate": 0.0003313665233697191,
      "loss": 1.2966,
      "step": 35431
    },
    {
      "epoch": 3.0484384410221113,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003313410672231567,
      "loss": 1.1053,
      "step": 35432
    },
    {
      "epoch": 3.048524477329433,
      "grad_norm": 1.171875,
      "learning_rate": 0.00033131561156988907,
      "loss": 1.676,
      "step": 35433
    },
    {
      "epoch": 3.0486105136367545,
      "grad_norm": 1.125,
      "learning_rate": 0.0003312901564099905,
      "loss": 0.8612,
      "step": 35434
    },
    {
      "epoch": 3.0486965499440766,
      "grad_norm": 8.0625,
      "learning_rate": 0.0003312647017435352,
      "loss": 1.5472,
      "step": 35435
    },
    {
      "epoch": 3.048782586251398,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00033123924757059796,
      "loss": 1.0653,
      "step": 35436
    },
    {
      "epoch": 3.04886862255872,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00033121379389125315,
      "loss": 0.7834,
      "step": 35437
    },
    {
      "epoch": 3.0489546588660414,
      "grad_norm": 1.515625,
      "learning_rate": 0.00033118834070557514,
      "loss": 1.0322,
      "step": 35438
    },
    {
      "epoch": 3.049040695173363,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003311628880136383,
      "loss": 1.6168,
      "step": 35439
    },
    {
      "epoch": 3.0491267314806847,
      "grad_norm": 1.0,
      "learning_rate": 0.00033113743581551726,
      "loss": 1.2925,
      "step": 35440
    },
    {
      "epoch": 3.0492127677880063,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003311119841112864,
      "loss": 1.6796,
      "step": 35441
    },
    {
      "epoch": 3.0492988040953284,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00033108653290102015,
      "loss": 1.5068,
      "step": 35442
    },
    {
      "epoch": 3.04938484040265,
      "grad_norm": 1.359375,
      "learning_rate": 0.00033106108218479283,
      "loss": 1.1577,
      "step": 35443
    },
    {
      "epoch": 3.0494708767099716,
      "grad_norm": 1.109375,
      "learning_rate": 0.00033103563196267905,
      "loss": 1.1785,
      "step": 35444
    },
    {
      "epoch": 3.0495569130172933,
      "grad_norm": 1.34375,
      "learning_rate": 0.000331010182234753,
      "loss": 1.7634,
      "step": 35445
    },
    {
      "epoch": 3.049642949324615,
      "grad_norm": 1.328125,
      "learning_rate": 0.00033098473300108955,
      "loss": 1.4027,
      "step": 35446
    },
    {
      "epoch": 3.0497289856319365,
      "grad_norm": 132.0,
      "learning_rate": 0.00033095928426176286,
      "loss": 1.0257,
      "step": 35447
    },
    {
      "epoch": 3.0498150219392586,
      "grad_norm": 7.96875,
      "learning_rate": 0.0003309338360168472,
      "loss": 1.5503,
      "step": 35448
    },
    {
      "epoch": 3.04990105824658,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003309083882664174,
      "loss": 1.746,
      "step": 35449
    },
    {
      "epoch": 3.049987094553902,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00033088294101054765,
      "loss": 1.0428,
      "step": 35450
    },
    {
      "epoch": 3.0500731308612234,
      "grad_norm": 2.03125,
      "learning_rate": 0.0003308574942493123,
      "loss": 1.0383,
      "step": 35451
    },
    {
      "epoch": 3.050159167168545,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00033083204798278596,
      "loss": 0.6364,
      "step": 35452
    },
    {
      "epoch": 3.0502452034758667,
      "grad_norm": 16.375,
      "learning_rate": 0.0003308066022110431,
      "loss": 1.2619,
      "step": 35453
    },
    {
      "epoch": 3.0503312397831883,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003307811569341579,
      "loss": 1.3336,
      "step": 35454
    },
    {
      "epoch": 3.0504172760905104,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003307557121522048,
      "loss": 1.2655,
      "step": 35455
    },
    {
      "epoch": 3.050503312397832,
      "grad_norm": 1.421875,
      "learning_rate": 0.00033073026786525846,
      "loss": 0.9271,
      "step": 35456
    },
    {
      "epoch": 3.0505893487051536,
      "grad_norm": 56.75,
      "learning_rate": 0.00033070482407339315,
      "loss": 1.1343,
      "step": 35457
    },
    {
      "epoch": 3.0506753850124753,
      "grad_norm": 1.359375,
      "learning_rate": 0.0003306793807766833,
      "loss": 1.3702,
      "step": 35458
    },
    {
      "epoch": 3.050761421319797,
      "grad_norm": 199.0,
      "learning_rate": 0.0003306539379752033,
      "loss": 0.6145,
      "step": 35459
    },
    {
      "epoch": 3.0508474576271185,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003306284956690276,
      "loss": 1.3027,
      "step": 35460
    },
    {
      "epoch": 3.05093349393444,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00033060305385823065,
      "loss": 0.9238,
      "step": 35461
    },
    {
      "epoch": 3.051019530241762,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003305776125428867,
      "loss": 0.9548,
      "step": 35462
    },
    {
      "epoch": 3.051105566549084,
      "grad_norm": 1048.0,
      "learning_rate": 0.0003305521717230704,
      "loss": 1.5136,
      "step": 35463
    },
    {
      "epoch": 3.0511916028564054,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000330526731398856,
      "loss": 0.5713,
      "step": 35464
    },
    {
      "epoch": 3.051277639163727,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00033050129157031805,
      "loss": 0.6685,
      "step": 35465
    },
    {
      "epoch": 3.0513636754710487,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003304758522375307,
      "loss": 0.9535,
      "step": 35466
    },
    {
      "epoch": 3.0514497117783703,
      "grad_norm": 10.4375,
      "learning_rate": 0.0003304504134005686,
      "loss": 0.9793,
      "step": 35467
    },
    {
      "epoch": 3.051535748085692,
      "grad_norm": 1.21875,
      "learning_rate": 0.00033042497505950607,
      "loss": 0.7053,
      "step": 35468
    },
    {
      "epoch": 3.051621784393014,
      "grad_norm": 3.125,
      "learning_rate": 0.0003303995372144174,
      "loss": 0.827,
      "step": 35469
    },
    {
      "epoch": 3.0517078207003356,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00033037409986537723,
      "loss": 1.0699,
      "step": 35470
    },
    {
      "epoch": 3.0517938570076573,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003303486630124598,
      "loss": 1.4248,
      "step": 35471
    },
    {
      "epoch": 3.051879893314979,
      "grad_norm": 2016.0,
      "learning_rate": 0.0003303232266557396,
      "loss": 0.664,
      "step": 35472
    },
    {
      "epoch": 3.0519659296223005,
      "grad_norm": 1.390625,
      "learning_rate": 0.00033029779079529075,
      "loss": 0.8032,
      "step": 35473
    },
    {
      "epoch": 3.052051965929622,
      "grad_norm": 912.0,
      "learning_rate": 0.00033027235543118806,
      "loss": 1.3838,
      "step": 35474
    },
    {
      "epoch": 3.052138002236944,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00033024692056350566,
      "loss": 0.9416,
      "step": 35475
    },
    {
      "epoch": 3.052224038544266,
      "grad_norm": 1432.0,
      "learning_rate": 0.0003302214861923179,
      "loss": 0.9328,
      "step": 35476
    },
    {
      "epoch": 3.0523100748515875,
      "grad_norm": 46.5,
      "learning_rate": 0.0003301960523176994,
      "loss": 1.0622,
      "step": 35477
    },
    {
      "epoch": 3.052396111158909,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00033017061893972436,
      "loss": 0.3899,
      "step": 35478
    },
    {
      "epoch": 3.0524821474662307,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003301451860584673,
      "loss": 0.9064,
      "step": 35479
    },
    {
      "epoch": 3.0525681837735523,
      "grad_norm": 3.125,
      "learning_rate": 0.0003301197536740024,
      "loss": 0.614,
      "step": 35480
    },
    {
      "epoch": 3.052654220080874,
      "grad_norm": 792.0,
      "learning_rate": 0.0003300943217864043,
      "loss": 0.9243,
      "step": 35481
    },
    {
      "epoch": 3.052740256388196,
      "grad_norm": 9.6875,
      "learning_rate": 0.0003300688903957471,
      "loss": 1.8582,
      "step": 35482
    },
    {
      "epoch": 3.0528262926955176,
      "grad_norm": 45.0,
      "learning_rate": 0.00033004345950210544,
      "loss": 1.3315,
      "step": 35483
    },
    {
      "epoch": 3.0529123290028393,
      "grad_norm": 418.0,
      "learning_rate": 0.0003300180291055537,
      "loss": 1.6509,
      "step": 35484
    },
    {
      "epoch": 3.052998365310161,
      "grad_norm": 1.171875,
      "learning_rate": 0.00032999259920616606,
      "loss": 0.8384,
      "step": 35485
    },
    {
      "epoch": 3.0530844016174825,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003299671698040171,
      "loss": 1.7744,
      "step": 35486
    },
    {
      "epoch": 3.053170437924804,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00032994174089918103,
      "loss": 1.5829,
      "step": 35487
    },
    {
      "epoch": 3.0532564742321258,
      "grad_norm": 44.0,
      "learning_rate": 0.0003299163124917324,
      "loss": 0.6455,
      "step": 35488
    },
    {
      "epoch": 3.053342510539448,
      "grad_norm": 1.765625,
      "learning_rate": 0.00032989088458174526,
      "loss": 1.5627,
      "step": 35489
    },
    {
      "epoch": 3.0534285468467695,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0003298654571692944,
      "loss": 1.0059,
      "step": 35490
    },
    {
      "epoch": 3.053514583154091,
      "grad_norm": 880.0,
      "learning_rate": 0.0003298400302544539,
      "loss": 0.9687,
      "step": 35491
    },
    {
      "epoch": 3.0536006194614127,
      "grad_norm": 1.421875,
      "learning_rate": 0.00032981460383729825,
      "loss": 1.8865,
      "step": 35492
    },
    {
      "epoch": 3.0536866557687343,
      "grad_norm": 126.5,
      "learning_rate": 0.00032978917791790175,
      "loss": 1.3552,
      "step": 35493
    },
    {
      "epoch": 3.053772692076056,
      "grad_norm": 142.0,
      "learning_rate": 0.00032976375249633885,
      "loss": 1.3954,
      "step": 35494
    },
    {
      "epoch": 3.0538587283833776,
      "grad_norm": 10944.0,
      "learning_rate": 0.00032973832757268387,
      "loss": 0.8571,
      "step": 35495
    },
    {
      "epoch": 3.0539447646906996,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000329712903147011,
      "loss": 1.1462,
      "step": 35496
    },
    {
      "epoch": 3.0540308009980213,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00032968747921939496,
      "loss": 0.6895,
      "step": 35497
    },
    {
      "epoch": 3.054116837305343,
      "grad_norm": 1608.0,
      "learning_rate": 0.0003296620557899099,
      "loss": 0.7859,
      "step": 35498
    },
    {
      "epoch": 3.0542028736126645,
      "grad_norm": 24960.0,
      "learning_rate": 0.0003296366328586301,
      "loss": 1.0902,
      "step": 35499
    },
    {
      "epoch": 3.054288909919986,
      "grad_norm": 600.0,
      "learning_rate": 0.0003296112104256301,
      "loss": 1.7633,
      "step": 35500
    },
    {
      "epoch": 3.0543749462273078,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003295857884909841,
      "loss": 1.1429,
      "step": 35501
    },
    {
      "epoch": 3.05446098253463,
      "grad_norm": 23552.0,
      "learning_rate": 0.0003295603670547667,
      "loss": 0.9321,
      "step": 35502
    },
    {
      "epoch": 3.0545470188419515,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003295349461170518,
      "loss": 0.7063,
      "step": 35503
    },
    {
      "epoch": 3.054633055149273,
      "grad_norm": 1.7578125,
      "learning_rate": 0.00032950952567791425,
      "loss": 1.2831,
      "step": 35504
    },
    {
      "epoch": 3.0547190914565947,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0003294841057374281,
      "loss": 1.0383,
      "step": 35505
    },
    {
      "epoch": 3.0548051277639163,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00032945868629566784,
      "loss": 1.1045,
      "step": 35506
    },
    {
      "epoch": 3.054891164071238,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00032943326735270764,
      "loss": 0.8093,
      "step": 35507
    },
    {
      "epoch": 3.0549772003785596,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000329407848908622,
      "loss": 1.162,
      "step": 35508
    },
    {
      "epoch": 3.0550632366858816,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003293824309634853,
      "loss": 1.7807,
      "step": 35509
    },
    {
      "epoch": 3.0551492729932033,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003293570135173717,
      "loss": 1.2198,
      "step": 35510
    },
    {
      "epoch": 3.055235309300525,
      "grad_norm": 1.0,
      "learning_rate": 0.00032933159657035565,
      "loss": 0.8759,
      "step": 35511
    },
    {
      "epoch": 3.0553213456078465,
      "grad_norm": 29.375,
      "learning_rate": 0.00032930618012251155,
      "loss": 0.6442,
      "step": 35512
    },
    {
      "epoch": 3.055407381915168,
      "grad_norm": 27.25,
      "learning_rate": 0.0003292807641739137,
      "loss": 1.5526,
      "step": 35513
    },
    {
      "epoch": 3.0554934182224898,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00032925534872463633,
      "loss": 0.8462,
      "step": 35514
    },
    {
      "epoch": 3.0555794545298114,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00032922993377475386,
      "loss": 0.9628,
      "step": 35515
    },
    {
      "epoch": 3.0556654908371335,
      "grad_norm": 9.0625,
      "learning_rate": 0.0003292045193243407,
      "loss": 1.1508,
      "step": 35516
    },
    {
      "epoch": 3.055751527144455,
      "grad_norm": 6.75,
      "learning_rate": 0.000329179105373471,
      "loss": 1.6421,
      "step": 35517
    },
    {
      "epoch": 3.0558375634517767,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003291536919222193,
      "loss": 1.1525,
      "step": 35518
    },
    {
      "epoch": 3.0559235997590983,
      "grad_norm": 121.5,
      "learning_rate": 0.00032912827897065954,
      "loss": 0.8829,
      "step": 35519
    },
    {
      "epoch": 3.05600963606642,
      "grad_norm": 2.390625,
      "learning_rate": 0.0003291028665188666,
      "loss": 1.5856,
      "step": 35520
    },
    {
      "epoch": 3.0560956723737416,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003290774545669146,
      "loss": 0.4132,
      "step": 35521
    },
    {
      "epoch": 3.056181708681063,
      "grad_norm": 1776.0,
      "learning_rate": 0.0003290520431148776,
      "loss": 0.4819,
      "step": 35522
    },
    {
      "epoch": 3.0562677449883853,
      "grad_norm": 3.21875,
      "learning_rate": 0.0003290266321628302,
      "loss": 0.8338,
      "step": 35523
    },
    {
      "epoch": 3.056353781295707,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003290012217108468,
      "loss": 1.5365,
      "step": 35524
    },
    {
      "epoch": 3.0564398176030285,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00032897581175900145,
      "loss": 1.2482,
      "step": 35525
    },
    {
      "epoch": 3.05652585391035,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0003289504023073685,
      "loss": 0.9097,
      "step": 35526
    },
    {
      "epoch": 3.0566118902176718,
      "grad_norm": 1.5625,
      "learning_rate": 0.0003289249933560225,
      "loss": 1.0014,
      "step": 35527
    },
    {
      "epoch": 3.0566979265249934,
      "grad_norm": 1.359375,
      "learning_rate": 0.00032889958490503755,
      "loss": 1.3041,
      "step": 35528
    },
    {
      "epoch": 3.0567839628323155,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032887417695448806,
      "loss": 1.5212,
      "step": 35529
    },
    {
      "epoch": 3.056869999139637,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003288487695044483,
      "loss": 1.1674,
      "step": 35530
    },
    {
      "epoch": 3.0569560354469587,
      "grad_norm": 17.75,
      "learning_rate": 0.0003288233625549926,
      "loss": 1.0747,
      "step": 35531
    },
    {
      "epoch": 3.0570420717542803,
      "grad_norm": 11008.0,
      "learning_rate": 0.00032879795610619533,
      "loss": 1.1326,
      "step": 35532
    },
    {
      "epoch": 3.057128108061602,
      "grad_norm": 25728.0,
      "learning_rate": 0.00032877255015813066,
      "loss": 0.8768,
      "step": 35533
    },
    {
      "epoch": 3.0572141443689236,
      "grad_norm": 1152.0,
      "learning_rate": 0.000328747144710873,
      "loss": 0.8503,
      "step": 35534
    },
    {
      "epoch": 3.057300180676245,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0003287217397644967,
      "loss": 1.0975,
      "step": 35535
    },
    {
      "epoch": 3.0573862169835673,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000328696335319076,
      "loss": 0.7116,
      "step": 35536
    },
    {
      "epoch": 3.057472253290889,
      "grad_norm": 26.625,
      "learning_rate": 0.00032867093137468504,
      "loss": 1.8379,
      "step": 35537
    },
    {
      "epoch": 3.0575582895982105,
      "grad_norm": 2.453125,
      "learning_rate": 0.00032864552793139844,
      "loss": 1.3307,
      "step": 35538
    },
    {
      "epoch": 3.057644325905532,
      "grad_norm": 568.0,
      "learning_rate": 0.0003286201249892904,
      "loss": 0.8285,
      "step": 35539
    },
    {
      "epoch": 3.0577303622128538,
      "grad_norm": 23.125,
      "learning_rate": 0.0003285947225484349,
      "loss": 1.087,
      "step": 35540
    },
    {
      "epoch": 3.0578163985201754,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00032856932060890674,
      "loss": 1.86,
      "step": 35541
    },
    {
      "epoch": 3.057902434827497,
      "grad_norm": 125.5,
      "learning_rate": 0.00032854391917077995,
      "loss": 1.0565,
      "step": 35542
    },
    {
      "epoch": 3.057988471134819,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003285185182341288,
      "loss": 0.8549,
      "step": 35543
    },
    {
      "epoch": 3.0580745074421407,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003284931177990276,
      "loss": 0.7836,
      "step": 35544
    },
    {
      "epoch": 3.0581605437494623,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032846771786555073,
      "loss": 1.0217,
      "step": 35545
    },
    {
      "epoch": 3.058246580056784,
      "grad_norm": 35.0,
      "learning_rate": 0.00032844231843377243,
      "loss": 0.8701,
      "step": 35546
    },
    {
      "epoch": 3.0583326163641056,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00032841691950376696,
      "loss": 1.4713,
      "step": 35547
    },
    {
      "epoch": 3.058418652671427,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00032839152107560853,
      "loss": 0.7898,
      "step": 35548
    },
    {
      "epoch": 3.058504688978749,
      "grad_norm": 16.5,
      "learning_rate": 0.0003283661231493717,
      "loss": 0.3875,
      "step": 35549
    },
    {
      "epoch": 3.058590725286071,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00032834072572513053,
      "loss": 0.7415,
      "step": 35550
    },
    {
      "epoch": 3.0586767615933925,
      "grad_norm": 31.375,
      "learning_rate": 0.00032831532880295923,
      "loss": 1.0123,
      "step": 35551
    },
    {
      "epoch": 3.058762797900714,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00032828993238293236,
      "loss": 0.7155,
      "step": 35552
    },
    {
      "epoch": 3.0588488342080358,
      "grad_norm": 1.0625,
      "learning_rate": 0.000328264536465124,
      "loss": 0.7741,
      "step": 35553
    },
    {
      "epoch": 3.0589348705153574,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00032823914104960843,
      "loss": 0.5188,
      "step": 35554
    },
    {
      "epoch": 3.059020906822679,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00032821374613645996,
      "loss": 1.1957,
      "step": 35555
    },
    {
      "epoch": 3.0591069431300006,
      "grad_norm": 1.140625,
      "learning_rate": 0.00032818835172575267,
      "loss": 1.0418,
      "step": 35556
    },
    {
      "epoch": 3.0591929794373227,
      "grad_norm": 2.21875,
      "learning_rate": 0.00032816295781756133,
      "loss": 1.6759,
      "step": 35557
    },
    {
      "epoch": 3.0592790157446443,
      "grad_norm": 53.0,
      "learning_rate": 0.00032813756441195987,
      "loss": 1.6314,
      "step": 35558
    },
    {
      "epoch": 3.059365052051966,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003281121715090226,
      "loss": 1.1417,
      "step": 35559
    },
    {
      "epoch": 3.0594510883592876,
      "grad_norm": 7008.0,
      "learning_rate": 0.0003280867791088237,
      "loss": 1.3437,
      "step": 35560
    },
    {
      "epoch": 3.059537124666609,
      "grad_norm": 1.90625,
      "learning_rate": 0.00032806138721143765,
      "loss": 1.2929,
      "step": 35561
    },
    {
      "epoch": 3.059623160973931,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003280359958169386,
      "loss": 1.1029,
      "step": 35562
    },
    {
      "epoch": 3.059709197281253,
      "grad_norm": 360.0,
      "learning_rate": 0.0003280106049254007,
      "loss": 0.9698,
      "step": 35563
    },
    {
      "epoch": 3.0597952335885745,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00032798521453689843,
      "loss": 1.2534,
      "step": 35564
    },
    {
      "epoch": 3.059881269895896,
      "grad_norm": 2336.0,
      "learning_rate": 0.000327959824651506,
      "loss": 1.5015,
      "step": 35565
    },
    {
      "epoch": 3.0599673062032178,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003279344352692976,
      "loss": 0.9533,
      "step": 35566
    },
    {
      "epoch": 3.0600533425105394,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00032790904639034744,
      "loss": 1.431,
      "step": 35567
    },
    {
      "epoch": 3.060139378817861,
      "grad_norm": 13.75,
      "learning_rate": 0.00032788365801472984,
      "loss": 0.3128,
      "step": 35568
    },
    {
      "epoch": 3.0602254151251826,
      "grad_norm": 16768.0,
      "learning_rate": 0.0003278582701425192,
      "loss": 0.5854,
      "step": 35569
    },
    {
      "epoch": 3.0603114514325047,
      "grad_norm": 4.90625,
      "learning_rate": 0.0003278328827737895,
      "loss": 1.1448,
      "step": 35570
    },
    {
      "epoch": 3.0603974877398263,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003278074959086151,
      "loss": 1.2723,
      "step": 35571
    },
    {
      "epoch": 3.060483524047148,
      "grad_norm": 2.8125,
      "learning_rate": 0.00032778210954707046,
      "loss": 1.5318,
      "step": 35572
    },
    {
      "epoch": 3.0605695603544696,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003277567236892296,
      "loss": 0.6791,
      "step": 35573
    },
    {
      "epoch": 3.060655596661791,
      "grad_norm": 1.203125,
      "learning_rate": 0.00032773133833516674,
      "loss": 0.8022,
      "step": 35574
    },
    {
      "epoch": 3.060741632969113,
      "grad_norm": 1.453125,
      "learning_rate": 0.00032770595348495627,
      "loss": 1.3753,
      "step": 35575
    },
    {
      "epoch": 3.0608276692764345,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003276805691386724,
      "loss": 1.2628,
      "step": 35576
    },
    {
      "epoch": 3.0609137055837565,
      "grad_norm": 1.84375,
      "learning_rate": 0.0003276551852963894,
      "loss": 1.1589,
      "step": 35577
    },
    {
      "epoch": 3.060999741891078,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00032762980195818125,
      "loss": 1.5005,
      "step": 35578
    },
    {
      "epoch": 3.0610857781983998,
      "grad_norm": 580.0,
      "learning_rate": 0.0003276044191241225,
      "loss": 0.8622,
      "step": 35579
    },
    {
      "epoch": 3.0611718145057214,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00032757903679428744,
      "loss": 1.7978,
      "step": 35580
    },
    {
      "epoch": 3.061257850813043,
      "grad_norm": 1.125,
      "learning_rate": 0.00032755365496875,
      "loss": 1.7081,
      "step": 35581
    },
    {
      "epoch": 3.0613438871203646,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003275282736475846,
      "loss": 0.8372,
      "step": 35582
    },
    {
      "epoch": 3.0614299234276863,
      "grad_norm": 1.125,
      "learning_rate": 0.0003275028928308655,
      "loss": 1.9183,
      "step": 35583
    },
    {
      "epoch": 3.0615159597350083,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000327477512518667,
      "loss": 1.9522,
      "step": 35584
    },
    {
      "epoch": 3.06160199604233,
      "grad_norm": 1.015625,
      "learning_rate": 0.000327452132711063,
      "loss": 1.1432,
      "step": 35585
    },
    {
      "epoch": 3.0616880323496516,
      "grad_norm": 1080.0,
      "learning_rate": 0.0003274267534081281,
      "loss": 1.0844,
      "step": 35586
    },
    {
      "epoch": 3.061774068656973,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00032740137460993634,
      "loss": 1.4072,
      "step": 35587
    },
    {
      "epoch": 3.061860104964295,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003273759963165619,
      "loss": 1.5587,
      "step": 35588
    },
    {
      "epoch": 3.0619461412716165,
      "grad_norm": 36.75,
      "learning_rate": 0.00032735061852807925,
      "loss": 1.1829,
      "step": 35589
    },
    {
      "epoch": 3.0620321775789385,
      "grad_norm": 2.21875,
      "learning_rate": 0.0003273252412445624,
      "loss": 1.2734,
      "step": 35590
    },
    {
      "epoch": 3.06211821388626,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00032729986446608565,
      "loss": 0.9762,
      "step": 35591
    },
    {
      "epoch": 3.062204250193582,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00032727448819272313,
      "loss": 0.8735,
      "step": 35592
    },
    {
      "epoch": 3.0622902865009034,
      "grad_norm": 21.5,
      "learning_rate": 0.00032724911242454916,
      "loss": 0.826,
      "step": 35593
    },
    {
      "epoch": 3.062376322808225,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003272237371616379,
      "loss": 0.8299,
      "step": 35594
    },
    {
      "epoch": 3.0624623591155467,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003271983624040637,
      "loss": 1.1906,
      "step": 35595
    },
    {
      "epoch": 3.0625483954228683,
      "grad_norm": 3.8125,
      "learning_rate": 0.00032717298815190067,
      "loss": 1.1549,
      "step": 35596
    },
    {
      "epoch": 3.0626344317301903,
      "grad_norm": 1.0625,
      "learning_rate": 0.000327147614405223,
      "loss": 0.68,
      "step": 35597
    },
    {
      "epoch": 3.062720468037512,
      "grad_norm": 13.0,
      "learning_rate": 0.000327122241164105,
      "loss": 1.5925,
      "step": 35598
    },
    {
      "epoch": 3.0628065043448336,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00032709686842862087,
      "loss": 1.4582,
      "step": 35599
    },
    {
      "epoch": 3.062892540652155,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032707149619884467,
      "loss": 1.5034,
      "step": 35600
    },
    {
      "epoch": 3.062978576959477,
      "grad_norm": 23.5,
      "learning_rate": 0.00032704612447485073,
      "loss": 1.8649,
      "step": 35601
    },
    {
      "epoch": 3.0630646132667985,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003270207532567132,
      "loss": 1.2337,
      "step": 35602
    },
    {
      "epoch": 3.06315064957412,
      "grad_norm": 1.265625,
      "learning_rate": 0.00032699538254450644,
      "loss": 0.96,
      "step": 35603
    },
    {
      "epoch": 3.063236685881442,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00032697001233830436,
      "loss": 0.9829,
      "step": 35604
    },
    {
      "epoch": 3.063322722188764,
      "grad_norm": 4.0,
      "learning_rate": 0.00032694464263818146,
      "loss": 1.3717,
      "step": 35605
    },
    {
      "epoch": 3.0634087584960854,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003269192734442119,
      "loss": 1.0279,
      "step": 35606
    },
    {
      "epoch": 3.063494794803407,
      "grad_norm": 6.375,
      "learning_rate": 0.00032689390475646975,
      "loss": 0.8421,
      "step": 35607
    },
    {
      "epoch": 3.0635808311107287,
      "grad_norm": 3.390625,
      "learning_rate": 0.0003268685365750291,
      "loss": 1.0675,
      "step": 35608
    },
    {
      "epoch": 3.0636668674180503,
      "grad_norm": 286.0,
      "learning_rate": 0.00032684316889996444,
      "loss": 1.3141,
      "step": 35609
    },
    {
      "epoch": 3.063752903725372,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003268178017313499,
      "loss": 1.2374,
      "step": 35610
    },
    {
      "epoch": 3.063838940032694,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003267924350692595,
      "loss": 0.993,
      "step": 35611
    },
    {
      "epoch": 3.0639249763400156,
      "grad_norm": 1528.0,
      "learning_rate": 0.0003267670689137676,
      "loss": 0.6982,
      "step": 35612
    },
    {
      "epoch": 3.064011012647337,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00032674170326494836,
      "loss": 1.3125,
      "step": 35613
    },
    {
      "epoch": 3.064097048954659,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003267163381228759,
      "loss": 0.6248,
      "step": 35614
    },
    {
      "epoch": 3.0641830852619805,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003266909734876243,
      "loss": 1.246,
      "step": 35615
    },
    {
      "epoch": 3.064269121569302,
      "grad_norm": 25.75,
      "learning_rate": 0.00032666560935926814,
      "loss": 1.1257,
      "step": 35616
    },
    {
      "epoch": 3.0643551578766237,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003266402457378813,
      "loss": 1.0048,
      "step": 35617
    },
    {
      "epoch": 3.064441194183946,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000326614882623538,
      "loss": 1.3036,
      "step": 35618
    },
    {
      "epoch": 3.0645272304912674,
      "grad_norm": 1.671875,
      "learning_rate": 0.0003265895200163123,
      "loss": 1.223,
      "step": 35619
    },
    {
      "epoch": 3.064613266798589,
      "grad_norm": 4.09375,
      "learning_rate": 0.0003265641579162787,
      "loss": 1.0497,
      "step": 35620
    },
    {
      "epoch": 3.0646993031059107,
      "grad_norm": 1.265625,
      "learning_rate": 0.00032653879632351117,
      "loss": 1.411,
      "step": 35621
    },
    {
      "epoch": 3.0647853394132323,
      "grad_norm": 2592.0,
      "learning_rate": 0.00032651343523808386,
      "loss": 0.769,
      "step": 35622
    },
    {
      "epoch": 3.064871375720554,
      "grad_norm": 27.625,
      "learning_rate": 0.00032648807466007114,
      "loss": 1.0713,
      "step": 35623
    },
    {
      "epoch": 3.064957412027876,
      "grad_norm": 1.3515625,
      "learning_rate": 0.000326462714589547,
      "loss": 1.5011,
      "step": 35624
    },
    {
      "epoch": 3.0650434483351976,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00032643735502658566,
      "loss": 0.7449,
      "step": 35625
    },
    {
      "epoch": 3.065129484642519,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0003264119959712613,
      "loss": 2.2917,
      "step": 35626
    },
    {
      "epoch": 3.065215520949841,
      "grad_norm": 78.0,
      "learning_rate": 0.00032638663742364806,
      "loss": 0.9656,
      "step": 35627
    },
    {
      "epoch": 3.0653015572571625,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00032636127938382017,
      "loss": 1.1856,
      "step": 35628
    },
    {
      "epoch": 3.065387593564484,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00032633592185185166,
      "loss": 0.7059,
      "step": 35629
    },
    {
      "epoch": 3.0654736298718057,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00032631056482781685,
      "loss": 1.2521,
      "step": 35630
    },
    {
      "epoch": 3.065559666179128,
      "grad_norm": 1.40625,
      "learning_rate": 0.00032628520831178985,
      "loss": 1.3616,
      "step": 35631
    },
    {
      "epoch": 3.0656457024864494,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003262598523038449,
      "loss": 0.6749,
      "step": 35632
    },
    {
      "epoch": 3.065731738793771,
      "grad_norm": 2.65625,
      "learning_rate": 0.0003262344968040561,
      "loss": 1.0078,
      "step": 35633
    },
    {
      "epoch": 3.0658177751010927,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003262091418124976,
      "loss": 1.361,
      "step": 35634
    },
    {
      "epoch": 3.0659038114084143,
      "grad_norm": 2.5,
      "learning_rate": 0.0003261837873292435,
      "loss": 0.9457,
      "step": 35635
    },
    {
      "epoch": 3.065989847715736,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00032615843335436807,
      "loss": 0.9116,
      "step": 35636
    },
    {
      "epoch": 3.0660758840230575,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003261330798879454,
      "loss": 1.5791,
      "step": 35637
    },
    {
      "epoch": 3.0661619203303796,
      "grad_norm": 239616.0,
      "learning_rate": 0.00032610772693004964,
      "loss": 1.4443,
      "step": 35638
    },
    {
      "epoch": 3.066247956637701,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00032608237448075494,
      "loss": 0.6418,
      "step": 35639
    },
    {
      "epoch": 3.066333992945023,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003260570225401356,
      "loss": 1.2129,
      "step": 35640
    },
    {
      "epoch": 3.0664200292523445,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003260316711082656,
      "loss": 0.6672,
      "step": 35641
    },
    {
      "epoch": 3.066506065559666,
      "grad_norm": 1.859375,
      "learning_rate": 0.00032600632018521894,
      "loss": 1.142,
      "step": 35642
    },
    {
      "epoch": 3.0665921018669877,
      "grad_norm": 5.21875,
      "learning_rate": 0.0003259809697710701,
      "loss": 1.0693,
      "step": 35643
    },
    {
      "epoch": 3.06667813817431,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003259556198658931,
      "loss": 1.1147,
      "step": 35644
    },
    {
      "epoch": 3.0667641744816314,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000325930270469762,
      "loss": 1.1362,
      "step": 35645
    },
    {
      "epoch": 3.066850210788953,
      "grad_norm": 1.640625,
      "learning_rate": 0.00032590492158275105,
      "loss": 1.2208,
      "step": 35646
    },
    {
      "epoch": 3.0669362470962747,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003258795732049343,
      "loss": 0.8144,
      "step": 35647
    },
    {
      "epoch": 3.0670222834035963,
      "grad_norm": 1.5,
      "learning_rate": 0.000325854225336386,
      "loss": 1.6661,
      "step": 35648
    },
    {
      "epoch": 3.067108319710918,
      "grad_norm": 1.0625,
      "learning_rate": 0.00032582887797718006,
      "loss": 1.2558,
      "step": 35649
    },
    {
      "epoch": 3.0671943560182395,
      "grad_norm": 1.265625,
      "learning_rate": 0.00032580353112739103,
      "loss": 0.6669,
      "step": 35650
    },
    {
      "epoch": 3.0672803923255616,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003257781847870926,
      "loss": 0.7317,
      "step": 35651
    },
    {
      "epoch": 3.0673664286328832,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003257528389563591,
      "loss": 1.3658,
      "step": 35652
    },
    {
      "epoch": 3.067452464940205,
      "grad_norm": 5.71875,
      "learning_rate": 0.00032572749363526477,
      "loss": 1.6763,
      "step": 35653
    },
    {
      "epoch": 3.0675385012475265,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003257021488238836,
      "loss": 0.4962,
      "step": 35654
    },
    {
      "epoch": 3.067624537554848,
      "grad_norm": 8.625,
      "learning_rate": 0.0003256768045222898,
      "loss": 0.8391,
      "step": 35655
    },
    {
      "epoch": 3.0677105738621697,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032565146073055716,
      "loss": 1.4167,
      "step": 35656
    },
    {
      "epoch": 3.0677966101694913,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0003256261174487604,
      "loss": 0.8799,
      "step": 35657
    },
    {
      "epoch": 3.0678826464768134,
      "grad_norm": 1.015625,
      "learning_rate": 0.00032560077467697323,
      "loss": 1.4335,
      "step": 35658
    },
    {
      "epoch": 3.067968682784135,
      "grad_norm": 1.125,
      "learning_rate": 0.00032557543241526986,
      "loss": 1.7558,
      "step": 35659
    },
    {
      "epoch": 3.0680547190914567,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003255500906637244,
      "loss": 0.6519,
      "step": 35660
    },
    {
      "epoch": 3.0681407553987783,
      "grad_norm": 0.890625,
      "learning_rate": 0.00032552474942241106,
      "loss": 1.1438,
      "step": 35661
    },
    {
      "epoch": 3.0682267917061,
      "grad_norm": 119.0,
      "learning_rate": 0.0003254994086914039,
      "loss": 1.3452,
      "step": 35662
    },
    {
      "epoch": 3.0683128280134215,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00032547406847077687,
      "loss": 2.059,
      "step": 35663
    },
    {
      "epoch": 3.068398864320743,
      "grad_norm": 1.28125,
      "learning_rate": 0.00032544872876060437,
      "loss": 1.3806,
      "step": 35664
    },
    {
      "epoch": 3.0684849006280652,
      "grad_norm": 314.0,
      "learning_rate": 0.00032542338956096045,
      "loss": 0.7501,
      "step": 35665
    },
    {
      "epoch": 3.068570936935387,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003253980508719191,
      "loss": 1.0211,
      "step": 35666
    },
    {
      "epoch": 3.0686569732427085,
      "grad_norm": 4.96875,
      "learning_rate": 0.00032537271269355433,
      "loss": 1.5176,
      "step": 35667
    },
    {
      "epoch": 3.06874300955003,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0003253473750259405,
      "loss": 0.9904,
      "step": 35668
    },
    {
      "epoch": 3.0688290458573517,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003253220378691518,
      "loss": 1.5846,
      "step": 35669
    },
    {
      "epoch": 3.0689150821646733,
      "grad_norm": 15.6875,
      "learning_rate": 0.00032529670122326206,
      "loss": 0.848,
      "step": 35670
    },
    {
      "epoch": 3.069001118471995,
      "grad_norm": 1.109375,
      "learning_rate": 0.00032527136508834554,
      "loss": 0.8285,
      "step": 35671
    },
    {
      "epoch": 3.069087154779317,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00032524602946447617,
      "loss": 1.5115,
      "step": 35672
    },
    {
      "epoch": 3.0691731910866387,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003252206943517283,
      "loss": 1.681,
      "step": 35673
    },
    {
      "epoch": 3.0692592273939603,
      "grad_norm": 19.25,
      "learning_rate": 0.0003251953597501759,
      "loss": 0.9718,
      "step": 35674
    },
    {
      "epoch": 3.069345263701282,
      "grad_norm": 52.5,
      "learning_rate": 0.00032517002565989294,
      "loss": 1.2118,
      "step": 35675
    },
    {
      "epoch": 3.0694313000086035,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003251446920809538,
      "loss": 1.5218,
      "step": 35676
    },
    {
      "epoch": 3.069517336315925,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00032511935901343244,
      "loss": 1.8026,
      "step": 35677
    },
    {
      "epoch": 3.069603372623247,
      "grad_norm": 159.0,
      "learning_rate": 0.0003250940264574029,
      "loss": 1.8792,
      "step": 35678
    },
    {
      "epoch": 3.069689408930569,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003250686944129392,
      "loss": 0.9857,
      "step": 35679
    },
    {
      "epoch": 3.0697754452378905,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003250433628801157,
      "loss": 0.9164,
      "step": 35680
    },
    {
      "epoch": 3.069861481545212,
      "grad_norm": 2.078125,
      "learning_rate": 0.0003250180318590063,
      "loss": 0.7422,
      "step": 35681
    },
    {
      "epoch": 3.0699475178525337,
      "grad_norm": 1.0625,
      "learning_rate": 0.00032499270134968516,
      "loss": 1.3608,
      "step": 35682
    },
    {
      "epoch": 3.0700335541598553,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003249673713522262,
      "loss": 1.1613,
      "step": 35683
    },
    {
      "epoch": 3.070119590467177,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00032494204186670373,
      "loss": 0.8726,
      "step": 35684
    },
    {
      "epoch": 3.070205626774499,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003249167128931918,
      "loss": 0.5565,
      "step": 35685
    },
    {
      "epoch": 3.0702916630818207,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00032489138443176426,
      "loss": 1.0004,
      "step": 35686
    },
    {
      "epoch": 3.0703776993891423,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003248660564824955,
      "loss": 0.9819,
      "step": 35687
    },
    {
      "epoch": 3.070463735696464,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003248407290454595,
      "loss": 1.8296,
      "step": 35688
    },
    {
      "epoch": 3.0705497720037855,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00032481540212073027,
      "loss": 0.7206,
      "step": 35689
    },
    {
      "epoch": 3.070635808311107,
      "grad_norm": 1.015625,
      "learning_rate": 0.00032479007570838177,
      "loss": 1.3135,
      "step": 35690
    },
    {
      "epoch": 3.070721844618429,
      "grad_norm": 1.203125,
      "learning_rate": 0.00032476474980848837,
      "loss": 1.978,
      "step": 35691
    },
    {
      "epoch": 3.070807880925751,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003247394244211239,
      "loss": 0.9532,
      "step": 35692
    },
    {
      "epoch": 3.0708939172330725,
      "grad_norm": 1.0,
      "learning_rate": 0.00032471409954636256,
      "loss": 0.8892,
      "step": 35693
    },
    {
      "epoch": 3.070979953540394,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003246887751842784,
      "loss": 1.3598,
      "step": 35694
    },
    {
      "epoch": 3.0710659898477157,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003246634513349456,
      "loss": 0.5771,
      "step": 35695
    },
    {
      "epoch": 3.0711520261550374,
      "grad_norm": 1304.0,
      "learning_rate": 0.00032463812799843794,
      "loss": 0.8239,
      "step": 35696
    },
    {
      "epoch": 3.071238062462359,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00032461280517482964,
      "loss": 0.907,
      "step": 35697
    },
    {
      "epoch": 3.0713240987696806,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00032458748286419485,
      "loss": 0.7868,
      "step": 35698
    },
    {
      "epoch": 3.0714101350770027,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003245621610666076,
      "loss": 0.7293,
      "step": 35699
    },
    {
      "epoch": 3.0714961713843243,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003245368397821419,
      "loss": 0.9387,
      "step": 35700
    },
    {
      "epoch": 3.071582207691646,
      "grad_norm": 10.75,
      "learning_rate": 0.00032451151901087164,
      "loss": 0.9537,
      "step": 35701
    },
    {
      "epoch": 3.0716682439989675,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003244861987528712,
      "loss": 1.5762,
      "step": 35702
    },
    {
      "epoch": 3.071754280306289,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003244608790082145,
      "loss": 1.0596,
      "step": 35703
    },
    {
      "epoch": 3.071840316613611,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003244355597769754,
      "loss": 1.7646,
      "step": 35704
    },
    {
      "epoch": 3.071926352920933,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003244102410592283,
      "loss": 0.5541,
      "step": 35705
    },
    {
      "epoch": 3.0720123892282545,
      "grad_norm": 1.4375,
      "learning_rate": 0.00032438492285504704,
      "loss": 0.959,
      "step": 35706
    },
    {
      "epoch": 3.072098425535576,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00032435960516450583,
      "loss": 0.7865,
      "step": 35707
    },
    {
      "epoch": 3.0721844618428977,
      "grad_norm": 13568.0,
      "learning_rate": 0.0003243342879876786,
      "loss": 2.11,
      "step": 35708
    },
    {
      "epoch": 3.0722704981502194,
      "grad_norm": 494.0,
      "learning_rate": 0.0003243089713246393,
      "loss": 0.9283,
      "step": 35709
    },
    {
      "epoch": 3.072356534457541,
      "grad_norm": 12288.0,
      "learning_rate": 0.0003242836551754622,
      "loss": 0.9964,
      "step": 35710
    },
    {
      "epoch": 3.0724425707648626,
      "grad_norm": 18.875,
      "learning_rate": 0.0003242583395402213,
      "loss": 1.2571,
      "step": 35711
    },
    {
      "epoch": 3.0725286070721847,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003242330244189905,
      "loss": 1.2645,
      "step": 35712
    },
    {
      "epoch": 3.0726146433795063,
      "grad_norm": 16.375,
      "learning_rate": 0.0003242077098118438,
      "loss": 1.602,
      "step": 35713
    },
    {
      "epoch": 3.072700679686828,
      "grad_norm": 1.296875,
      "learning_rate": 0.00032418239571885553,
      "loss": 1.4376,
      "step": 35714
    },
    {
      "epoch": 3.0727867159941495,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003241570821400995,
      "loss": 1.5794,
      "step": 35715
    },
    {
      "epoch": 3.072872752301471,
      "grad_norm": 260.0,
      "learning_rate": 0.00032413176907564975,
      "loss": 1.067,
      "step": 35716
    },
    {
      "epoch": 3.072958788608793,
      "grad_norm": 0.84375,
      "learning_rate": 0.0003241064565255804,
      "loss": 0.6619,
      "step": 35717
    },
    {
      "epoch": 3.0730448249161144,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003240811444899655,
      "loss": 0.4866,
      "step": 35718
    },
    {
      "epoch": 3.0731308612234365,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003240558329688791,
      "loss": 0.9593,
      "step": 35719
    },
    {
      "epoch": 3.073216897530758,
      "grad_norm": 370.0,
      "learning_rate": 0.00032403052196239496,
      "loss": 1.4062,
      "step": 35720
    },
    {
      "epoch": 3.0733029338380797,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0003240052114705875,
      "loss": 0.6189,
      "step": 35721
    },
    {
      "epoch": 3.0733889701454014,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0003239799014935305,
      "loss": 0.4193,
      "step": 35722
    },
    {
      "epoch": 3.073475006452723,
      "grad_norm": 1.359375,
      "learning_rate": 0.00032395459203129806,
      "loss": 1.6656,
      "step": 35723
    },
    {
      "epoch": 3.0735610427600446,
      "grad_norm": 2.0625,
      "learning_rate": 0.0003239292830839642,
      "loss": 2.149,
      "step": 35724
    },
    {
      "epoch": 3.0736470790673662,
      "grad_norm": 154.0,
      "learning_rate": 0.0003239039746516029,
      "loss": 0.9469,
      "step": 35725
    },
    {
      "epoch": 3.0737331153746883,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0003238786667342883,
      "loss": 1.467,
      "step": 35726
    },
    {
      "epoch": 3.07381915168201,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003238533593320943,
      "loss": 0.8416,
      "step": 35727
    },
    {
      "epoch": 3.0739051879893315,
      "grad_norm": 5.0625,
      "learning_rate": 0.00032382805244509494,
      "loss": 1.1646,
      "step": 35728
    },
    {
      "epoch": 3.073991224296653,
      "grad_norm": 100.5,
      "learning_rate": 0.00032380274607336437,
      "loss": 1.3374,
      "step": 35729
    },
    {
      "epoch": 3.074077260603975,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003237774402169764,
      "loss": 1.8466,
      "step": 35730
    },
    {
      "epoch": 3.0741632969112964,
      "grad_norm": 1.65625,
      "learning_rate": 0.0003237521348760051,
      "loss": 1.1885,
      "step": 35731
    },
    {
      "epoch": 3.074249333218618,
      "grad_norm": 21.75,
      "learning_rate": 0.00032372683005052466,
      "loss": 0.7146,
      "step": 35732
    },
    {
      "epoch": 3.07433536952594,
      "grad_norm": 1.15625,
      "learning_rate": 0.00032370152574060895,
      "loss": 1.0453,
      "step": 35733
    },
    {
      "epoch": 3.0744214058332617,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003236762219463318,
      "loss": 1.2307,
      "step": 35734
    },
    {
      "epoch": 3.0745074421405834,
      "grad_norm": 2.09375,
      "learning_rate": 0.00032365091866776765,
      "loss": 0.6349,
      "step": 35735
    },
    {
      "epoch": 3.074593478447905,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00032362561590499017,
      "loss": 1.0202,
      "step": 35736
    },
    {
      "epoch": 3.0746795147552266,
      "grad_norm": 16.0,
      "learning_rate": 0.0003236003136580735,
      "loss": 1.0481,
      "step": 35737
    },
    {
      "epoch": 3.0747655510625482,
      "grad_norm": 88.0,
      "learning_rate": 0.00032357501192709147,
      "loss": 1.0494,
      "step": 35738
    },
    {
      "epoch": 3.0748515873698703,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00032354971071211835,
      "loss": 1.4565,
      "step": 35739
    },
    {
      "epoch": 3.074937623677192,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00032352441001322796,
      "loss": 1.1355,
      "step": 35740
    },
    {
      "epoch": 3.0750236599845135,
      "grad_norm": 3.3125,
      "learning_rate": 0.0003234991098304942,
      "loss": 1.2414,
      "step": 35741
    },
    {
      "epoch": 3.075109696291835,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00032347381016399135,
      "loss": 1.1244,
      "step": 35742
    },
    {
      "epoch": 3.075195732599157,
      "grad_norm": 80.5,
      "learning_rate": 0.00032344851101379324,
      "loss": 1.1196,
      "step": 35743
    },
    {
      "epoch": 3.0752817689064784,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00032342321237997394,
      "loss": 1.2633,
      "step": 35744
    },
    {
      "epoch": 3.0753678052138,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00032339791426260747,
      "loss": 1.7271,
      "step": 35745
    },
    {
      "epoch": 3.075453841521122,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00032337261666176776,
      "loss": 0.9424,
      "step": 35746
    },
    {
      "epoch": 3.0755398778284437,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003233473195775286,
      "loss": 1.8006,
      "step": 35747
    },
    {
      "epoch": 3.0756259141357654,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003233220230099644,
      "loss": 1.4627,
      "step": 35748
    },
    {
      "epoch": 3.075711950443087,
      "grad_norm": 20.5,
      "learning_rate": 0.0003232967269591488,
      "loss": 0.7426,
      "step": 35749
    },
    {
      "epoch": 3.0757979867504086,
      "grad_norm": 3.171875,
      "learning_rate": 0.0003232714314251559,
      "loss": 1.1812,
      "step": 35750
    },
    {
      "epoch": 3.0758840230577302,
      "grad_norm": 0.890625,
      "learning_rate": 0.00032324613640805967,
      "loss": 0.8324,
      "step": 35751
    },
    {
      "epoch": 3.075970059365052,
      "grad_norm": 16.25,
      "learning_rate": 0.0003232208419079342,
      "loss": 1.4279,
      "step": 35752
    },
    {
      "epoch": 3.076056095672374,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003231955479248534,
      "loss": 1.5753,
      "step": 35753
    },
    {
      "epoch": 3.0761421319796955,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00032317025445889114,
      "loss": 1.22,
      "step": 35754
    },
    {
      "epoch": 3.076228168287017,
      "grad_norm": 1.0,
      "learning_rate": 0.0003231449615101215,
      "loss": 1.7603,
      "step": 35755
    },
    {
      "epoch": 3.076314204594339,
      "grad_norm": 5.0625,
      "learning_rate": 0.00032311966907861854,
      "loss": 1.6958,
      "step": 35756
    },
    {
      "epoch": 3.0764002409016604,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000323094377164456,
      "loss": 0.9851,
      "step": 35757
    },
    {
      "epoch": 3.076486277208982,
      "grad_norm": 1.421875,
      "learning_rate": 0.00032306908576770814,
      "loss": 1.1291,
      "step": 35758
    },
    {
      "epoch": 3.0765723135163037,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0003230437948884488,
      "loss": 1.039,
      "step": 35759
    },
    {
      "epoch": 3.0766583498236257,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000323018504526752,
      "loss": 0.5317,
      "step": 35760
    },
    {
      "epoch": 3.0767443861309474,
      "grad_norm": 1.421875,
      "learning_rate": 0.00032299321468269137,
      "loss": 1.7172,
      "step": 35761
    },
    {
      "epoch": 3.076830422438269,
      "grad_norm": 217.0,
      "learning_rate": 0.00032296792535634137,
      "loss": 1.3723,
      "step": 35762
    },
    {
      "epoch": 3.0769164587455906,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003229426365477758,
      "loss": 1.0783,
      "step": 35763
    },
    {
      "epoch": 3.0770024950529122,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003229173482570684,
      "loss": 1.3179,
      "step": 35764
    },
    {
      "epoch": 3.077088531360234,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003228920604842935,
      "loss": 1.2548,
      "step": 35765
    },
    {
      "epoch": 3.077174567667556,
      "grad_norm": 11.8125,
      "learning_rate": 0.0003228667732295248,
      "loss": 0.9744,
      "step": 35766
    },
    {
      "epoch": 3.0772606039748776,
      "grad_norm": 1.125,
      "learning_rate": 0.0003228414864928364,
      "loss": 1.9143,
      "step": 35767
    },
    {
      "epoch": 3.077346640282199,
      "grad_norm": 1688.0,
      "learning_rate": 0.00032281620027430205,
      "loss": 0.6941,
      "step": 35768
    },
    {
      "epoch": 3.077432676589521,
      "grad_norm": 232.0,
      "learning_rate": 0.00032279091457399595,
      "loss": 1.5647,
      "step": 35769
    },
    {
      "epoch": 3.0775187128968424,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00032276562939199195,
      "loss": 1.3636,
      "step": 35770
    },
    {
      "epoch": 3.077604749204164,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003227403447283641,
      "loss": 0.9329,
      "step": 35771
    },
    {
      "epoch": 3.0776907855114857,
      "grad_norm": 2.078125,
      "learning_rate": 0.0003227150605831861,
      "loss": 1.6294,
      "step": 35772
    },
    {
      "epoch": 3.0777768218188077,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003226897769565321,
      "loss": 0.8356,
      "step": 35773
    },
    {
      "epoch": 3.0778628581261294,
      "grad_norm": 8.3125,
      "learning_rate": 0.0003226644938484762,
      "loss": 0.9622,
      "step": 35774
    },
    {
      "epoch": 3.077948894433451,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003226392112590919,
      "loss": 1.1774,
      "step": 35775
    },
    {
      "epoch": 3.0780349307407726,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003226139291884536,
      "loss": 1.9074,
      "step": 35776
    },
    {
      "epoch": 3.0781209670480942,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003225886476366348,
      "loss": 0.9556,
      "step": 35777
    },
    {
      "epoch": 3.078207003355416,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032256336660370995,
      "loss": 0.9713,
      "step": 35778
    },
    {
      "epoch": 3.0782930396627375,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003225380860897527,
      "loss": 1.3184,
      "step": 35779
    },
    {
      "epoch": 3.0783790759700596,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000322512806094837,
      "loss": 0.6989,
      "step": 35780
    },
    {
      "epoch": 3.078465112277381,
      "grad_norm": 1.953125,
      "learning_rate": 0.00032248752661903683,
      "loss": 1.4302,
      "step": 35781
    },
    {
      "epoch": 3.078551148584703,
      "grad_norm": 1.015625,
      "learning_rate": 0.00032246224766242613,
      "loss": 0.5999,
      "step": 35782
    },
    {
      "epoch": 3.0786371848920244,
      "grad_norm": 1.46875,
      "learning_rate": 0.00032243696922507894,
      "loss": 1.2864,
      "step": 35783
    },
    {
      "epoch": 3.078723221199346,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00032241169130706887,
      "loss": 1.3324,
      "step": 35784
    },
    {
      "epoch": 3.0788092575066677,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00032238641390847024,
      "loss": 1.2975,
      "step": 35785
    },
    {
      "epoch": 3.0788952938139893,
      "grad_norm": 1.59375,
      "learning_rate": 0.0003223611370293568,
      "loss": 1.5843,
      "step": 35786
    },
    {
      "epoch": 3.0789813301213114,
      "grad_norm": 3.453125,
      "learning_rate": 0.0003223358606698023,
      "loss": 1.2955,
      "step": 35787
    },
    {
      "epoch": 3.079067366428633,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00032231058482988094,
      "loss": 1.6768,
      "step": 35788
    },
    {
      "epoch": 3.0791534027359546,
      "grad_norm": 3.453125,
      "learning_rate": 0.0003222853095096667,
      "loss": 1.2565,
      "step": 35789
    },
    {
      "epoch": 3.0792394390432762,
      "grad_norm": 1.171875,
      "learning_rate": 0.00032226003470923326,
      "loss": 0.9538,
      "step": 35790
    },
    {
      "epoch": 3.079325475350598,
      "grad_norm": 7.5,
      "learning_rate": 0.00032223476042865463,
      "loss": 1.2485,
      "step": 35791
    },
    {
      "epoch": 3.0794115116579195,
      "grad_norm": 2.75,
      "learning_rate": 0.0003222094866680048,
      "loss": 1.486,
      "step": 35792
    },
    {
      "epoch": 3.079497547965241,
      "grad_norm": 9.0,
      "learning_rate": 0.0003221842134273576,
      "loss": 0.7454,
      "step": 35793
    },
    {
      "epoch": 3.079583584272563,
      "grad_norm": 8.875,
      "learning_rate": 0.00032215894070678706,
      "loss": 0.7323,
      "step": 35794
    },
    {
      "epoch": 3.079669620579885,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00032213366850636696,
      "loss": 0.7189,
      "step": 35795
    },
    {
      "epoch": 3.0797556568872064,
      "grad_norm": 1.546875,
      "learning_rate": 0.0003221083968261713,
      "loss": 1.7091,
      "step": 35796
    },
    {
      "epoch": 3.079841693194528,
      "grad_norm": 1.5625,
      "learning_rate": 0.000322083125666274,
      "loss": 1.1722,
      "step": 35797
    },
    {
      "epoch": 3.0799277295018497,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003220578550267489,
      "loss": 1.329,
      "step": 35798
    },
    {
      "epoch": 3.0800137658091713,
      "grad_norm": 1.015625,
      "learning_rate": 0.00032203258490767,
      "loss": 0.7527,
      "step": 35799
    },
    {
      "epoch": 3.0800998021164934,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003220073153091112,
      "loss": 1.5291,
      "step": 35800
    },
    {
      "epoch": 3.080185838423815,
      "grad_norm": 1.375,
      "learning_rate": 0.00032198204623114647,
      "loss": 1.7399,
      "step": 35801
    },
    {
      "epoch": 3.0802718747311366,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00032195677767384945,
      "loss": 0.787,
      "step": 35802
    },
    {
      "epoch": 3.0803579110384582,
      "grad_norm": 14.5,
      "learning_rate": 0.00032193150963729435,
      "loss": 0.5571,
      "step": 35803
    },
    {
      "epoch": 3.08044394734578,
      "grad_norm": 0.96875,
      "learning_rate": 0.00032190624212155496,
      "loss": 1.4233,
      "step": 35804
    },
    {
      "epoch": 3.0805299836531015,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000321880975126705,
      "loss": 0.8411,
      "step": 35805
    },
    {
      "epoch": 3.080616019960423,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00032185570865281873,
      "loss": 0.7799,
      "step": 35806
    },
    {
      "epoch": 3.080702056267745,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032183044269996984,
      "loss": 0.8535,
      "step": 35807
    },
    {
      "epoch": 3.080788092575067,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0003218051772682322,
      "loss": 0.7169,
      "step": 35808
    },
    {
      "epoch": 3.0808741288823884,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00032177991235767976,
      "loss": 1.4592,
      "step": 35809
    },
    {
      "epoch": 3.08096016518971,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00032175464796838646,
      "loss": 0.7282,
      "step": 35810
    },
    {
      "epoch": 3.0810462014970317,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003217293841004262,
      "loss": 1.5863,
      "step": 35811
    },
    {
      "epoch": 3.0811322378043533,
      "grad_norm": 11.625,
      "learning_rate": 0.0003217041207538727,
      "loss": 0.522,
      "step": 35812
    },
    {
      "epoch": 3.081218274111675,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00032167885792879993,
      "loss": 1.2363,
      "step": 35813
    },
    {
      "epoch": 3.081304310418997,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003216535956252817,
      "loss": 1.1472,
      "step": 35814
    },
    {
      "epoch": 3.0813903467263186,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003216283338433923,
      "loss": 0.9727,
      "step": 35815
    },
    {
      "epoch": 3.0814763830336402,
      "grad_norm": 14.875,
      "learning_rate": 0.0003216030725832053,
      "loss": 0.6216,
      "step": 35816
    },
    {
      "epoch": 3.081562419340962,
      "grad_norm": 1.25,
      "learning_rate": 0.00032157781184479454,
      "loss": 0.8243,
      "step": 35817
    },
    {
      "epoch": 3.0816484556482835,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00032155255162823394,
      "loss": 0.9399,
      "step": 35818
    },
    {
      "epoch": 3.081734491955605,
      "grad_norm": 5.0625,
      "learning_rate": 0.0003215272919335975,
      "loss": 1.6473,
      "step": 35819
    },
    {
      "epoch": 3.081820528262927,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003215020327609591,
      "loss": 1.3526,
      "step": 35820
    },
    {
      "epoch": 3.081906564570249,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00032147677411039233,
      "loss": 1.0037,
      "step": 35821
    },
    {
      "epoch": 3.0819926008775704,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00032145151598197145,
      "loss": 0.9253,
      "step": 35822
    },
    {
      "epoch": 3.082078637184892,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0003214262583757701,
      "loss": 1.339,
      "step": 35823
    },
    {
      "epoch": 3.0821646734922137,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003214010012918623,
      "loss": 0.6531,
      "step": 35824
    },
    {
      "epoch": 3.0822507097995353,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003213757447303216,
      "loss": 0.6865,
      "step": 35825
    },
    {
      "epoch": 3.082336746106857,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0003213504886912223,
      "loss": 1.4624,
      "step": 35826
    },
    {
      "epoch": 3.082422782414179,
      "grad_norm": 1.421875,
      "learning_rate": 0.00032132523317463814,
      "loss": 1.337,
      "step": 35827
    },
    {
      "epoch": 3.0825088187215006,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003212999781806427,
      "loss": 0.7853,
      "step": 35828
    },
    {
      "epoch": 3.0825948550288222,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003212747237093102,
      "loss": 0.9568,
      "step": 35829
    },
    {
      "epoch": 3.082680891336144,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003212494697607144,
      "loss": 0.6811,
      "step": 35830
    },
    {
      "epoch": 3.0827669276434655,
      "grad_norm": 1.515625,
      "learning_rate": 0.0003212242163349291,
      "loss": 1.0674,
      "step": 35831
    },
    {
      "epoch": 3.082852963950787,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00032119896343202814,
      "loss": 1.167,
      "step": 35832
    },
    {
      "epoch": 3.0829390002581087,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003211737110520855,
      "loss": 1.3555,
      "step": 35833
    },
    {
      "epoch": 3.083025036565431,
      "grad_norm": 1.078125,
      "learning_rate": 0.000321148459195175,
      "loss": 1.0535,
      "step": 35834
    },
    {
      "epoch": 3.0831110728727524,
      "grad_norm": 1.046875,
      "learning_rate": 0.00032112320786137046,
      "loss": 1.3192,
      "step": 35835
    },
    {
      "epoch": 3.083197109180074,
      "grad_norm": 1.25,
      "learning_rate": 0.0003210979570507456,
      "loss": 0.9637,
      "step": 35836
    },
    {
      "epoch": 3.0832831454873957,
      "grad_norm": 1.125,
      "learning_rate": 0.00032107270676337465,
      "loss": 0.596,
      "step": 35837
    },
    {
      "epoch": 3.0833691817947173,
      "grad_norm": 194.0,
      "learning_rate": 0.00032104745699933115,
      "loss": 0.5647,
      "step": 35838
    },
    {
      "epoch": 3.083455218102039,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003210222077586889,
      "loss": 1.3619,
      "step": 35839
    },
    {
      "epoch": 3.0835412544093606,
      "grad_norm": 1.4140625,
      "learning_rate": 0.000320996959041522,
      "loss": 0.8048,
      "step": 35840
    },
    {
      "epoch": 3.0836272907166826,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003209717108479042,
      "loss": 1.2,
      "step": 35841
    },
    {
      "epoch": 3.0837133270240042,
      "grad_norm": 8.4375,
      "learning_rate": 0.00032094646317790934,
      "loss": 1.2084,
      "step": 35842
    },
    {
      "epoch": 3.083799363331326,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00032092121603161116,
      "loss": 0.6834,
      "step": 35843
    },
    {
      "epoch": 3.0838853996386475,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003208959694090837,
      "loss": 1.157,
      "step": 35844
    },
    {
      "epoch": 3.083971435945969,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003208707233104006,
      "loss": 1.2053,
      "step": 35845
    },
    {
      "epoch": 3.0840574722532907,
      "grad_norm": 1.234375,
      "learning_rate": 0.00032084547773563577,
      "loss": 1.8356,
      "step": 35846
    },
    {
      "epoch": 3.0841435085606124,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00032082023268486317,
      "loss": 1.2303,
      "step": 35847
    },
    {
      "epoch": 3.0842295448679344,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00032079498815815654,
      "loss": 1.1603,
      "step": 35848
    },
    {
      "epoch": 3.084315581175256,
      "grad_norm": 0.859375,
      "learning_rate": 0.00032076974415558973,
      "loss": 0.8285,
      "step": 35849
    },
    {
      "epoch": 3.0844016174825777,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00032074450067723646,
      "loss": 0.9928,
      "step": 35850
    },
    {
      "epoch": 3.0844876537898993,
      "grad_norm": 2.84375,
      "learning_rate": 0.00032071925772317057,
      "loss": 1.5439,
      "step": 35851
    },
    {
      "epoch": 3.084573690097221,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003206940152934661,
      "loss": 1.0215,
      "step": 35852
    },
    {
      "epoch": 3.0846597264045426,
      "grad_norm": 3.4375,
      "learning_rate": 0.0003206687733881969,
      "loss": 1.5326,
      "step": 35853
    },
    {
      "epoch": 3.084745762711864,
      "grad_norm": 1.203125,
      "learning_rate": 0.00032064353200743654,
      "loss": 1.2406,
      "step": 35854
    },
    {
      "epoch": 3.0848317990191862,
      "grad_norm": 1.25,
      "learning_rate": 0.0003206182911512589,
      "loss": 1.1265,
      "step": 35855
    },
    {
      "epoch": 3.084917835326508,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000320593050819738,
      "loss": 1.7829,
      "step": 35856
    },
    {
      "epoch": 3.0850038716338295,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0003205678110129475,
      "loss": 1.8776,
      "step": 35857
    },
    {
      "epoch": 3.085089907941151,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00032054257173096123,
      "loss": 1.1582,
      "step": 35858
    },
    {
      "epoch": 3.0851759442484727,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000320517332973853,
      "loss": 1.1789,
      "step": 35859
    },
    {
      "epoch": 3.0852619805557944,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003204920947416967,
      "loss": 1.0835,
      "step": 35860
    },
    {
      "epoch": 3.0853480168631164,
      "grad_norm": 3.921875,
      "learning_rate": 0.0003204668570345662,
      "loss": 1.0801,
      "step": 35861
    },
    {
      "epoch": 3.085434053170438,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00032044161985253505,
      "loss": 1.1998,
      "step": 35862
    },
    {
      "epoch": 3.0855200894777597,
      "grad_norm": 4.4375,
      "learning_rate": 0.00032041638319567735,
      "loss": 1.6293,
      "step": 35863
    },
    {
      "epoch": 3.0856061257850813,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003203911470640668,
      "loss": 0.7397,
      "step": 35864
    },
    {
      "epoch": 3.085692162092403,
      "grad_norm": 0.78125,
      "learning_rate": 0.0003203659114577773,
      "loss": 0.8284,
      "step": 35865
    },
    {
      "epoch": 3.0857781983997246,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00032034067637688236,
      "loss": 1.1498,
      "step": 35866
    },
    {
      "epoch": 3.085864234707046,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003203154418214561,
      "loss": 1.2204,
      "step": 35867
    },
    {
      "epoch": 3.0859502710143683,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0003202902077915723,
      "loss": 1.2844,
      "step": 35868
    },
    {
      "epoch": 3.08603630732169,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003202649742873046,
      "loss": 0.7194,
      "step": 35869
    },
    {
      "epoch": 3.0861223436290115,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003202397413087269,
      "loss": 1.2045,
      "step": 35870
    },
    {
      "epoch": 3.086208379936333,
      "grad_norm": 1.203125,
      "learning_rate": 0.00032021450885591306,
      "loss": 1.6698,
      "step": 35871
    },
    {
      "epoch": 3.0862944162436547,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00032018927692893683,
      "loss": 0.5351,
      "step": 35872
    },
    {
      "epoch": 3.0863804525509764,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00032016404552787184,
      "loss": 1.2008,
      "step": 35873
    },
    {
      "epoch": 3.086466488858298,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00032013881465279214,
      "loss": 1.5853,
      "step": 35874
    },
    {
      "epoch": 3.08655252516562,
      "grad_norm": 1.125,
      "learning_rate": 0.00032011358430377145,
      "loss": 1.2339,
      "step": 35875
    },
    {
      "epoch": 3.0866385614729417,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00032008835448088356,
      "loss": 0.6945,
      "step": 35876
    },
    {
      "epoch": 3.0867245977802633,
      "grad_norm": 1.265625,
      "learning_rate": 0.00032006312518420213,
      "loss": 0.8616,
      "step": 35877
    },
    {
      "epoch": 3.086810634087585,
      "grad_norm": 1.7890625,
      "learning_rate": 0.00032003789641380116,
      "loss": 1.8208,
      "step": 35878
    },
    {
      "epoch": 3.0868966703949066,
      "grad_norm": 1.390625,
      "learning_rate": 0.00032001266816975433,
      "loss": 1.5035,
      "step": 35879
    },
    {
      "epoch": 3.086982706702228,
      "grad_norm": 1.140625,
      "learning_rate": 0.00031998744045213536,
      "loss": 1.5248,
      "step": 35880
    },
    {
      "epoch": 3.0870687430095503,
      "grad_norm": 1.25,
      "learning_rate": 0.0003199622132610182,
      "loss": 1.2865,
      "step": 35881
    },
    {
      "epoch": 3.087154779316872,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003199369865964765,
      "loss": 0.7957,
      "step": 35882
    },
    {
      "epoch": 3.0872408156241935,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003199117604585842,
      "loss": 1.5062,
      "step": 35883
    },
    {
      "epoch": 3.087326851931515,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003198865348474148,
      "loss": 0.5007,
      "step": 35884
    },
    {
      "epoch": 3.0874128882388368,
      "grad_norm": 2.59375,
      "learning_rate": 0.00031986130976304237,
      "loss": 1.1468,
      "step": 35885
    },
    {
      "epoch": 3.0874989245461584,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00031983608520554064,
      "loss": 1.1078,
      "step": 35886
    },
    {
      "epoch": 3.08758496085348,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003198108611749832,
      "loss": 0.8998,
      "step": 35887
    },
    {
      "epoch": 3.087670997160802,
      "grad_norm": 1.265625,
      "learning_rate": 0.00031978563767144386,
      "loss": 1.4143,
      "step": 35888
    },
    {
      "epoch": 3.0877570334681237,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003197604146949965,
      "loss": 1.8447,
      "step": 35889
    },
    {
      "epoch": 3.0878430697754453,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00031973519224571503,
      "loss": 1.1059,
      "step": 35890
    },
    {
      "epoch": 3.087929106082767,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000319709970323673,
      "loss": 1.1088,
      "step": 35891
    },
    {
      "epoch": 3.0880151423900886,
      "grad_norm": 1.03125,
      "learning_rate": 0.00031968474892894424,
      "loss": 0.6468,
      "step": 35892
    },
    {
      "epoch": 3.08810117869741,
      "grad_norm": 1.0625,
      "learning_rate": 0.00031965952806160255,
      "loss": 1.0722,
      "step": 35893
    },
    {
      "epoch": 3.088187215004732,
      "grad_norm": 1.046875,
      "learning_rate": 0.00031963430772172165,
      "loss": 1.2444,
      "step": 35894
    },
    {
      "epoch": 3.088273251312054,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00031960908790937536,
      "loss": 1.3064,
      "step": 35895
    },
    {
      "epoch": 3.0883592876193755,
      "grad_norm": 3.765625,
      "learning_rate": 0.0003195838686246373,
      "loss": 1.2149,
      "step": 35896
    },
    {
      "epoch": 3.088445323926697,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003195586498675814,
      "loss": 0.7922,
      "step": 35897
    },
    {
      "epoch": 3.0885313602340188,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003195334316382814,
      "loss": 0.7788,
      "step": 35898
    },
    {
      "epoch": 3.0886173965413404,
      "grad_norm": 1.0625,
      "learning_rate": 0.00031950821393681095,
      "loss": 1.3051,
      "step": 35899
    },
    {
      "epoch": 3.088703432848662,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003194829967632439,
      "loss": 1.9007,
      "step": 35900
    },
    {
      "epoch": 3.0887894691559836,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0003194577801176539,
      "loss": 0.6498,
      "step": 35901
    },
    {
      "epoch": 3.0888755054633057,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00031943256400011487,
      "loss": 1.6332,
      "step": 35902
    },
    {
      "epoch": 3.0889615417706273,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003194073484107004,
      "loss": 1.0056,
      "step": 35903
    },
    {
      "epoch": 3.089047578077949,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003193821333494844,
      "loss": 1.2828,
      "step": 35904
    },
    {
      "epoch": 3.0891336143852706,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003193569188165405,
      "loss": 1.6224,
      "step": 35905
    },
    {
      "epoch": 3.089219650692592,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003193317048119425,
      "loss": 1.2245,
      "step": 35906
    },
    {
      "epoch": 3.089305686999914,
      "grad_norm": 1.140625,
      "learning_rate": 0.00031930649133576406,
      "loss": 2.1369,
      "step": 35907
    },
    {
      "epoch": 3.0893917233072354,
      "grad_norm": 3.75,
      "learning_rate": 0.000319281278388079,
      "loss": 1.2112,
      "step": 35908
    },
    {
      "epoch": 3.0894777596145575,
      "grad_norm": 1.5625,
      "learning_rate": 0.0003192560659689611,
      "loss": 1.5476,
      "step": 35909
    },
    {
      "epoch": 3.089563795921879,
      "grad_norm": 1.40625,
      "learning_rate": 0.000319230854078484,
      "loss": 1.4995,
      "step": 35910
    },
    {
      "epoch": 3.0896498322292008,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003192056427167215,
      "loss": 0.9268,
      "step": 35911
    },
    {
      "epoch": 3.0897358685365224,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003191804318837475,
      "loss": 1.5685,
      "step": 35912
    },
    {
      "epoch": 3.089821904843844,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003191552215796354,
      "loss": 0.8086,
      "step": 35913
    },
    {
      "epoch": 3.0899079411511656,
      "grad_norm": 1.03125,
      "learning_rate": 0.000319130011804459,
      "loss": 0.6887,
      "step": 35914
    },
    {
      "epoch": 3.0899939774584877,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031910480255829235,
      "loss": 1.5747,
      "step": 35915
    },
    {
      "epoch": 3.0900800137658093,
      "grad_norm": 25.875,
      "learning_rate": 0.0003190795938412089,
      "loss": 1.1636,
      "step": 35916
    },
    {
      "epoch": 3.090166050073131,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00031905438565328235,
      "loss": 1.4678,
      "step": 35917
    },
    {
      "epoch": 3.0902520863804526,
      "grad_norm": 1.15625,
      "learning_rate": 0.00031902917799458665,
      "loss": 1.3439,
      "step": 35918
    },
    {
      "epoch": 3.090338122687774,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003190039708651954,
      "loss": 0.5594,
      "step": 35919
    },
    {
      "epoch": 3.090424158995096,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00031897876426518237,
      "loss": 0.8407,
      "step": 35920
    },
    {
      "epoch": 3.0905101953024174,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00031895355819462113,
      "loss": 1.1004,
      "step": 35921
    },
    {
      "epoch": 3.0905962316097395,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00031892835265358554,
      "loss": 0.7439,
      "step": 35922
    },
    {
      "epoch": 3.090682267917061,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003189031476421495,
      "loss": 0.592,
      "step": 35923
    },
    {
      "epoch": 3.0907683042243828,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00031887794316038634,
      "loss": 1.6132,
      "step": 35924
    },
    {
      "epoch": 3.0908543405317044,
      "grad_norm": 0.984375,
      "learning_rate": 0.00031885273920836976,
      "loss": 0.8406,
      "step": 35925
    },
    {
      "epoch": 3.090940376839026,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003188275357861739,
      "loss": 0.7449,
      "step": 35926
    },
    {
      "epoch": 3.0910264131463476,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003188023328938723,
      "loss": 0.9097,
      "step": 35927
    },
    {
      "epoch": 3.0911124494536693,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031877713053153866,
      "loss": 1.4794,
      "step": 35928
    },
    {
      "epoch": 3.0911984857609913,
      "grad_norm": 0.953125,
      "learning_rate": 0.00031875192869924667,
      "loss": 1.0103,
      "step": 35929
    },
    {
      "epoch": 3.091284522068313,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003187267273970699,
      "loss": 0.7589,
      "step": 35930
    },
    {
      "epoch": 3.0913705583756346,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00031870152662508237,
      "loss": 1.4505,
      "step": 35931
    },
    {
      "epoch": 3.091456594682956,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003186763263833575,
      "loss": 1.0175,
      "step": 35932
    },
    {
      "epoch": 3.091542630990278,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003186511266719692,
      "loss": 1.1535,
      "step": 35933
    },
    {
      "epoch": 3.0916286672975994,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000318625927490991,
      "loss": 1.2974,
      "step": 35934
    },
    {
      "epoch": 3.091714703604921,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00031860072884049674,
      "loss": 0.7937,
      "step": 35935
    },
    {
      "epoch": 3.091800739912243,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003185755307205601,
      "loss": 1.1118,
      "step": 35936
    },
    {
      "epoch": 3.0918867762195648,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003185503331312546,
      "loss": 1.7752,
      "step": 35937
    },
    {
      "epoch": 3.0919728125268864,
      "grad_norm": 1.734375,
      "learning_rate": 0.00031852513607265423,
      "loss": 0.9235,
      "step": 35938
    },
    {
      "epoch": 3.092058848834208,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00031849993954483253,
      "loss": 0.974,
      "step": 35939
    },
    {
      "epoch": 3.0921448851415296,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0003184747435478631,
      "loss": 1.2353,
      "step": 35940
    },
    {
      "epoch": 3.0922309214488513,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003184495480818198,
      "loss": 1.4257,
      "step": 35941
    },
    {
      "epoch": 3.0923169577561733,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003184243531467763,
      "loss": 1.9221,
      "step": 35942
    },
    {
      "epoch": 3.092402994063495,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00031839915874280636,
      "loss": 1.3025,
      "step": 35943
    },
    {
      "epoch": 3.0924890303708166,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003183739648699833,
      "loss": 1.0802,
      "step": 35944
    },
    {
      "epoch": 3.092575066678138,
      "grad_norm": 1.484375,
      "learning_rate": 0.00031834877152838126,
      "loss": 1.2318,
      "step": 35945
    },
    {
      "epoch": 3.09266110298546,
      "grad_norm": 0.953125,
      "learning_rate": 0.00031832357871807365,
      "loss": 1.3516,
      "step": 35946
    },
    {
      "epoch": 3.0927471392927814,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00031829838643913433,
      "loss": 1.237,
      "step": 35947
    },
    {
      "epoch": 3.092833175600103,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003182731946916367,
      "loss": 0.6759,
      "step": 35948
    },
    {
      "epoch": 3.092919211907425,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003182480034756548,
      "loss": 1.0391,
      "step": 35949
    },
    {
      "epoch": 3.0930052482147468,
      "grad_norm": 0.875,
      "learning_rate": 0.00031822281279126214,
      "loss": 1.2679,
      "step": 35950
    },
    {
      "epoch": 3.0930912845220684,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00031819762263853225,
      "loss": 0.9076,
      "step": 35951
    },
    {
      "epoch": 3.09317732082939,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003181724330175391,
      "loss": 1.0822,
      "step": 35952
    },
    {
      "epoch": 3.0932633571367116,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003181472439283562,
      "loss": 1.2474,
      "step": 35953
    },
    {
      "epoch": 3.0933493934440333,
      "grad_norm": 1.328125,
      "learning_rate": 0.00031812205537105724,
      "loss": 0.6781,
      "step": 35954
    },
    {
      "epoch": 3.093435429751355,
      "grad_norm": 123.5,
      "learning_rate": 0.00031809686734571584,
      "loss": 0.7631,
      "step": 35955
    },
    {
      "epoch": 3.093521466058677,
      "grad_norm": 2.734375,
      "learning_rate": 0.00031807167985240583,
      "loss": 1.1924,
      "step": 35956
    },
    {
      "epoch": 3.0936075023659986,
      "grad_norm": 1.3125,
      "learning_rate": 0.00031804649289120067,
      "loss": 1.2144,
      "step": 35957
    },
    {
      "epoch": 3.09369353867332,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003180213064621741,
      "loss": 1.1554,
      "step": 35958
    },
    {
      "epoch": 3.093779574980642,
      "grad_norm": 4.40625,
      "learning_rate": 0.0003179961205653999,
      "loss": 1.2161,
      "step": 35959
    },
    {
      "epoch": 3.0938656112879634,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00031797093520095167,
      "loss": 1.7251,
      "step": 35960
    },
    {
      "epoch": 3.093951647595285,
      "grad_norm": 0.890625,
      "learning_rate": 0.00031794575036890306,
      "loss": 0.5902,
      "step": 35961
    },
    {
      "epoch": 3.0940376839026067,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003179205660693275,
      "loss": 1.8552,
      "step": 35962
    },
    {
      "epoch": 3.0941237202099288,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0003178953823022991,
      "loss": 1.0737,
      "step": 35963
    },
    {
      "epoch": 3.0942097565172504,
      "grad_norm": 9.25,
      "learning_rate": 0.0003178701990678911,
      "loss": 1.2594,
      "step": 35964
    },
    {
      "epoch": 3.094295792824572,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031784501636617754,
      "loss": 0.6401,
      "step": 35965
    },
    {
      "epoch": 3.0943818291318936,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003178198341972318,
      "loss": 0.5369,
      "step": 35966
    },
    {
      "epoch": 3.0944678654392153,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003177946525611276,
      "loss": 1.0169,
      "step": 35967
    },
    {
      "epoch": 3.094553901746537,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003177694714579387,
      "loss": 0.9888,
      "step": 35968
    },
    {
      "epoch": 3.0946399380538585,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003177442908877386,
      "loss": 0.7173,
      "step": 35969
    },
    {
      "epoch": 3.0947259743611806,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000317719110850601,
      "loss": 0.8733,
      "step": 35970
    },
    {
      "epoch": 3.094812010668502,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003176939313465995,
      "loss": 1.6162,
      "step": 35971
    },
    {
      "epoch": 3.094898046975824,
      "grad_norm": 2.5625,
      "learning_rate": 0.0003176687523758078,
      "loss": 0.9297,
      "step": 35972
    },
    {
      "epoch": 3.0949840832831454,
      "grad_norm": 1.125,
      "learning_rate": 0.0003176435739382996,
      "loss": 0.8583,
      "step": 35973
    },
    {
      "epoch": 3.095070119590467,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00031761839603414844,
      "loss": 0.861,
      "step": 35974
    },
    {
      "epoch": 3.0951561558977887,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00031759321866342804,
      "loss": 1.8439,
      "step": 35975
    },
    {
      "epoch": 3.0952421922051108,
      "grad_norm": 1.296875,
      "learning_rate": 0.00031756804182621204,
      "loss": 1.5663,
      "step": 35976
    },
    {
      "epoch": 3.0953282285124324,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00031754286552257394,
      "loss": 1.2502,
      "step": 35977
    },
    {
      "epoch": 3.095414264819754,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0003175176897525874,
      "loss": 1.0068,
      "step": 35978
    },
    {
      "epoch": 3.0955003011270756,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003174925145163263,
      "loss": 1.2388,
      "step": 35979
    },
    {
      "epoch": 3.0955863374343973,
      "grad_norm": 142.0,
      "learning_rate": 0.00031746733981386405,
      "loss": 0.8492,
      "step": 35980
    },
    {
      "epoch": 3.095672373741719,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003174421656452742,
      "loss": 1.5133,
      "step": 35981
    },
    {
      "epoch": 3.0957584100490405,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003174169920106307,
      "loss": 0.9262,
      "step": 35982
    },
    {
      "epoch": 3.0958444463563626,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003173918189100069,
      "loss": 0.5768,
      "step": 35983
    },
    {
      "epoch": 3.095930482663684,
      "grad_norm": 1.0,
      "learning_rate": 0.00031736664634347654,
      "loss": 1.0035,
      "step": 35984
    },
    {
      "epoch": 3.096016518971006,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00031734147431111313,
      "loss": 0.9571,
      "step": 35985
    },
    {
      "epoch": 3.0961025552783275,
      "grad_norm": 28.875,
      "learning_rate": 0.0003173163028129905,
      "loss": 0.571,
      "step": 35986
    },
    {
      "epoch": 3.096188591585649,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0003172911318491821,
      "loss": 1.1734,
      "step": 35987
    },
    {
      "epoch": 3.0962746278929707,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00031726596141976173,
      "loss": 1.1283,
      "step": 35988
    },
    {
      "epoch": 3.0963606642002923,
      "grad_norm": 8.25,
      "learning_rate": 0.00031724079152480267,
      "loss": 1.6149,
      "step": 35989
    },
    {
      "epoch": 3.0964467005076144,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003172156221643788,
      "loss": 0.9233,
      "step": 35990
    },
    {
      "epoch": 3.096532736814936,
      "grad_norm": 1.203125,
      "learning_rate": 0.00031719045333856375,
      "loss": 1.5043,
      "step": 35991
    },
    {
      "epoch": 3.0966187731222576,
      "grad_norm": 127.5,
      "learning_rate": 0.00031716528504743103,
      "loss": 1.1839,
      "step": 35992
    },
    {
      "epoch": 3.0967048094295793,
      "grad_norm": 1984.0,
      "learning_rate": 0.00031714011729105435,
      "loss": 1.6131,
      "step": 35993
    },
    {
      "epoch": 3.096790845736901,
      "grad_norm": 0.96875,
      "learning_rate": 0.00031711495006950724,
      "loss": 1.0189,
      "step": 35994
    },
    {
      "epoch": 3.0968768820442225,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00031708978338286335,
      "loss": 1.2567,
      "step": 35995
    },
    {
      "epoch": 3.0969629183515446,
      "grad_norm": 1.171875,
      "learning_rate": 0.00031706461723119615,
      "loss": 1.2506,
      "step": 35996
    },
    {
      "epoch": 3.097048954658866,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00031703945161457947,
      "loss": 0.6283,
      "step": 35997
    },
    {
      "epoch": 3.097134990966188,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0003170142865330868,
      "loss": 0.7017,
      "step": 35998
    },
    {
      "epoch": 3.0972210272735095,
      "grad_norm": 3.75,
      "learning_rate": 0.00031698912198679165,
      "loss": 1.1954,
      "step": 35999
    },
    {
      "epoch": 3.097307063580831,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003169639579757678,
      "loss": 0.8281,
      "step": 36000
    },
    {
      "epoch": 3.0973930998881527,
      "grad_norm": 1.4375,
      "learning_rate": 0.00031693879450008877,
      "loss": 1.3784,
      "step": 36001
    },
    {
      "epoch": 3.0974791361954743,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00031691363155982824,
      "loss": 0.7922,
      "step": 36002
    },
    {
      "epoch": 3.0975651725027964,
      "grad_norm": 21.875,
      "learning_rate": 0.00031688846915505967,
      "loss": 1.0241,
      "step": 36003
    },
    {
      "epoch": 3.097651208810118,
      "grad_norm": 2.15625,
      "learning_rate": 0.00031686330728585666,
      "loss": 1.1736,
      "step": 36004
    },
    {
      "epoch": 3.0977372451174396,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00031683814595229295,
      "loss": 1.1932,
      "step": 36005
    },
    {
      "epoch": 3.0978232814247613,
      "grad_norm": 102.5,
      "learning_rate": 0.000316812985154442,
      "loss": 1.2009,
      "step": 36006
    },
    {
      "epoch": 3.097909317732083,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00031678782489237746,
      "loss": 1.0539,
      "step": 36007
    },
    {
      "epoch": 3.0979953540394045,
      "grad_norm": 1.28125,
      "learning_rate": 0.00031676266516617285,
      "loss": 1.1321,
      "step": 36008
    },
    {
      "epoch": 3.098081390346726,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0003167375059759019,
      "loss": 0.819,
      "step": 36009
    },
    {
      "epoch": 3.098167426654048,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00031671234732163806,
      "loss": 0.5213,
      "step": 36010
    },
    {
      "epoch": 3.09825346296137,
      "grad_norm": 1.6875,
      "learning_rate": 0.00031668718920345494,
      "loss": 1.297,
      "step": 36011
    },
    {
      "epoch": 3.0983394992686915,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00031666203162142607,
      "loss": 0.8779,
      "step": 36012
    },
    {
      "epoch": 3.098425535576013,
      "grad_norm": 1.59375,
      "learning_rate": 0.0003166368745756252,
      "loss": 1.3353,
      "step": 36013
    },
    {
      "epoch": 3.0985115718833347,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003166117180661259,
      "loss": 1.39,
      "step": 36014
    },
    {
      "epoch": 3.0985976081906563,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003165865620930014,
      "loss": 0.5587,
      "step": 36015
    },
    {
      "epoch": 3.098683644497978,
      "grad_norm": 52.25,
      "learning_rate": 0.0003165614066563257,
      "loss": 1.3889,
      "step": 36016
    },
    {
      "epoch": 3.0987696808053,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00031653625175617224,
      "loss": 1.6664,
      "step": 36017
    },
    {
      "epoch": 3.0988557171126216,
      "grad_norm": 161.0,
      "learning_rate": 0.00031651109739261454,
      "loss": 1.2376,
      "step": 36018
    },
    {
      "epoch": 3.0989417534199433,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00031648594356572607,
      "loss": 1.1418,
      "step": 36019
    },
    {
      "epoch": 3.099027789727265,
      "grad_norm": 274.0,
      "learning_rate": 0.0003164607902755806,
      "loss": 0.9618,
      "step": 36020
    },
    {
      "epoch": 3.0991138260345865,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003164356375222517,
      "loss": 1.4207,
      "step": 36021
    },
    {
      "epoch": 3.099199862341908,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003164104853058127,
      "loss": 1.0595,
      "step": 36022
    },
    {
      "epoch": 3.0992858986492298,
      "grad_norm": 1.109375,
      "learning_rate": 0.00031638533362633747,
      "loss": 1.0771,
      "step": 36023
    },
    {
      "epoch": 3.099371934956552,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003163601824838994,
      "loss": 1.4545,
      "step": 36024
    },
    {
      "epoch": 3.0994579712638735,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0003163350318785721,
      "loss": 1.81,
      "step": 36025
    },
    {
      "epoch": 3.099544007571195,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031630988181042895,
      "loss": 1.1655,
      "step": 36026
    },
    {
      "epoch": 3.0996300438785167,
      "grad_norm": 1.46875,
      "learning_rate": 0.00031628473227954387,
      "loss": 0.6987,
      "step": 36027
    },
    {
      "epoch": 3.0997160801858383,
      "grad_norm": 10.125,
      "learning_rate": 0.00031625958328599014,
      "loss": 0.7802,
      "step": 36028
    },
    {
      "epoch": 3.09980211649316,
      "grad_norm": 1.015625,
      "learning_rate": 0.00031623443482984144,
      "loss": 0.9745,
      "step": 36029
    },
    {
      "epoch": 3.0998881528004816,
      "grad_norm": 4.34375,
      "learning_rate": 0.0003162092869111711,
      "loss": 1.4504,
      "step": 36030
    },
    {
      "epoch": 3.0999741891078036,
      "grad_norm": 1.125,
      "learning_rate": 0.00031618413953005305,
      "loss": 1.3796,
      "step": 36031
    },
    {
      "epoch": 3.1000602254151253,
      "grad_norm": 1.125,
      "learning_rate": 0.0003161589926865606,
      "loss": 0.6149,
      "step": 36032
    },
    {
      "epoch": 3.100146261722447,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00031613384638076725,
      "loss": 0.9995,
      "step": 36033
    },
    {
      "epoch": 3.1002322980297685,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003161087006127467,
      "loss": 0.9402,
      "step": 36034
    },
    {
      "epoch": 3.10031833433709,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0003160835553825725,
      "loss": 0.6389,
      "step": 36035
    },
    {
      "epoch": 3.1004043706444118,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031605841069031783,
      "loss": 1.2468,
      "step": 36036
    },
    {
      "epoch": 3.100490406951734,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003160332665360568,
      "loss": 1.2146,
      "step": 36037
    },
    {
      "epoch": 3.1005764432590555,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0003160081229198626,
      "loss": 0.6704,
      "step": 36038
    },
    {
      "epoch": 3.100662479566377,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003159829798418089,
      "loss": 1.5706,
      "step": 36039
    },
    {
      "epoch": 3.1007485158736987,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00031595783730196925,
      "loss": 1.3078,
      "step": 36040
    },
    {
      "epoch": 3.1008345521810203,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003159326953004171,
      "loss": 0.9448,
      "step": 36041
    },
    {
      "epoch": 3.100920588488342,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00031590755383722595,
      "loss": 0.7663,
      "step": 36042
    },
    {
      "epoch": 3.1010066247956636,
      "grad_norm": 1.4375,
      "learning_rate": 0.00031588241291246954,
      "loss": 1.2055,
      "step": 36043
    },
    {
      "epoch": 3.1010926611029856,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003158572725262212,
      "loss": 0.8233,
      "step": 36044
    },
    {
      "epoch": 3.1011786974103073,
      "grad_norm": 2.125,
      "learning_rate": 0.0003158321326785544,
      "loss": 1.1277,
      "step": 36045
    },
    {
      "epoch": 3.101264733717629,
      "grad_norm": 1.046875,
      "learning_rate": 0.00031580699336954295,
      "loss": 1.0365,
      "step": 36046
    },
    {
      "epoch": 3.1013507700249505,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003157818545992602,
      "loss": 1.1519,
      "step": 36047
    },
    {
      "epoch": 3.101436806332272,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003157567163677797,
      "loss": 0.8747,
      "step": 36048
    },
    {
      "epoch": 3.1015228426395938,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003157315786751748,
      "loss": 0.7274,
      "step": 36049
    },
    {
      "epoch": 3.1016088789469154,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003157064415215194,
      "loss": 1.5402,
      "step": 36050
    },
    {
      "epoch": 3.1016949152542375,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00031568130490688674,
      "loss": 0.8878,
      "step": 36051
    },
    {
      "epoch": 3.101780951561559,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003156561688313505,
      "loss": 1.3163,
      "step": 36052
    },
    {
      "epoch": 3.1018669878688807,
      "grad_norm": 1.28125,
      "learning_rate": 0.00031563103329498393,
      "loss": 1.3708,
      "step": 36053
    },
    {
      "epoch": 3.1019530241762023,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003156058982978609,
      "loss": 1.1702,
      "step": 36054
    },
    {
      "epoch": 3.102039060483524,
      "grad_norm": 2.78125,
      "learning_rate": 0.00031558076384005467,
      "loss": 0.8674,
      "step": 36055
    },
    {
      "epoch": 3.1021250967908456,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003155556299216388,
      "loss": 1.2061,
      "step": 36056
    },
    {
      "epoch": 3.1022111330981677,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00031553049654268697,
      "loss": 1.2175,
      "step": 36057
    },
    {
      "epoch": 3.1022971694054893,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00031550536370327246,
      "loss": 0.8083,
      "step": 36058
    },
    {
      "epoch": 3.102383205712811,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003154802314034689,
      "loss": 1.0522,
      "step": 36059
    },
    {
      "epoch": 3.1024692420201325,
      "grad_norm": 12.4375,
      "learning_rate": 0.00031545509964334975,
      "loss": 1.2431,
      "step": 36060
    },
    {
      "epoch": 3.102555278327454,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0003154299684229886,
      "loss": 1.143,
      "step": 36061
    },
    {
      "epoch": 3.1026413146347758,
      "grad_norm": 1.25,
      "learning_rate": 0.0003154048377424589,
      "loss": 0.9894,
      "step": 36062
    },
    {
      "epoch": 3.1027273509420974,
      "grad_norm": 1.171875,
      "learning_rate": 0.000315379707601834,
      "loss": 0.9902,
      "step": 36063
    },
    {
      "epoch": 3.1028133872494195,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003153545780011877,
      "loss": 0.8856,
      "step": 36064
    },
    {
      "epoch": 3.102899423556741,
      "grad_norm": 7.90625,
      "learning_rate": 0.00031532944894059335,
      "loss": 1.6771,
      "step": 36065
    },
    {
      "epoch": 3.1029854598640627,
      "grad_norm": 1.0,
      "learning_rate": 0.0003153043204201245,
      "loss": 0.9944,
      "step": 36066
    },
    {
      "epoch": 3.1030714961713843,
      "grad_norm": 1.6875,
      "learning_rate": 0.0003152791924398544,
      "loss": 1.5648,
      "step": 36067
    },
    {
      "epoch": 3.103157532478706,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003152540649998569,
      "loss": 1.053,
      "step": 36068
    },
    {
      "epoch": 3.1032435687860276,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00031522893810020523,
      "loss": 1.6825,
      "step": 36069
    },
    {
      "epoch": 3.103329605093349,
      "grad_norm": 2.921875,
      "learning_rate": 0.00031520381174097295,
      "loss": 1.0727,
      "step": 36070
    },
    {
      "epoch": 3.1034156414006713,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003151786859222337,
      "loss": 1.211,
      "step": 36071
    },
    {
      "epoch": 3.103501677707993,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00031515356064406084,
      "loss": 1.1394,
      "step": 36072
    },
    {
      "epoch": 3.1035877140153145,
      "grad_norm": 1.140625,
      "learning_rate": 0.00031512843590652764,
      "loss": 0.7071,
      "step": 36073
    },
    {
      "epoch": 3.103673750322636,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00031510331170970797,
      "loss": 1.0851,
      "step": 36074
    },
    {
      "epoch": 3.1037597866299578,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003150781880536752,
      "loss": 1.5271,
      "step": 36075
    },
    {
      "epoch": 3.1038458229372794,
      "grad_norm": 1.765625,
      "learning_rate": 0.0003150530649385027,
      "loss": 1.1853,
      "step": 36076
    },
    {
      "epoch": 3.103931859244601,
      "grad_norm": 1.515625,
      "learning_rate": 0.000315027942364264,
      "loss": 1.5049,
      "step": 36077
    },
    {
      "epoch": 3.104017895551923,
      "grad_norm": 1.375,
      "learning_rate": 0.00031500282033103275,
      "loss": 1.2552,
      "step": 36078
    },
    {
      "epoch": 3.1041039318592447,
      "grad_norm": 1.015625,
      "learning_rate": 0.00031497769883888203,
      "loss": 1.163,
      "step": 36079
    },
    {
      "epoch": 3.1041899681665663,
      "grad_norm": 32.0,
      "learning_rate": 0.0003149525778878858,
      "loss": 1.0114,
      "step": 36080
    },
    {
      "epoch": 3.104276004473888,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00031492745747811713,
      "loss": 1.1119,
      "step": 36081
    },
    {
      "epoch": 3.1043620407812096,
      "grad_norm": 2.609375,
      "learning_rate": 0.00031490233760964975,
      "loss": 1.3081,
      "step": 36082
    },
    {
      "epoch": 3.104448077088531,
      "grad_norm": 1.5625,
      "learning_rate": 0.0003148772182825569,
      "loss": 1.2656,
      "step": 36083
    },
    {
      "epoch": 3.104534113395853,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003148520994969123,
      "loss": 1.2541,
      "step": 36084
    },
    {
      "epoch": 3.104620149703175,
      "grad_norm": 15.75,
      "learning_rate": 0.0003148269812527893,
      "loss": 1.2819,
      "step": 36085
    },
    {
      "epoch": 3.1047061860104965,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003148018635502613,
      "loss": 1.6297,
      "step": 36086
    },
    {
      "epoch": 3.104792222317818,
      "grad_norm": 1.078125,
      "learning_rate": 0.00031477674638940185,
      "loss": 1.4688,
      "step": 36087
    },
    {
      "epoch": 3.1048782586251398,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003147516297702844,
      "loss": 1.5128,
      "step": 36088
    },
    {
      "epoch": 3.1049642949324614,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00031472651369298245,
      "loss": 0.6535,
      "step": 36089
    },
    {
      "epoch": 3.105050331239783,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00031470139815756933,
      "loss": 1.8793,
      "step": 36090
    },
    {
      "epoch": 3.105136367547105,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0003146762831641187,
      "loss": 2.1014,
      "step": 36091
    },
    {
      "epoch": 3.1052224038544267,
      "grad_norm": 1.125,
      "learning_rate": 0.00031465116871270385,
      "loss": 0.8458,
      "step": 36092
    },
    {
      "epoch": 3.1053084401617483,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003146260548033981,
      "loss": 1.0524,
      "step": 36093
    },
    {
      "epoch": 3.10539447646907,
      "grad_norm": 11.6875,
      "learning_rate": 0.0003146009414362753,
      "loss": 1.5172,
      "step": 36094
    },
    {
      "epoch": 3.1054805127763916,
      "grad_norm": 1.46875,
      "learning_rate": 0.00031457582861140864,
      "loss": 1.4882,
      "step": 36095
    },
    {
      "epoch": 3.105566549083713,
      "grad_norm": 2.8125,
      "learning_rate": 0.0003145507163288717,
      "loss": 1.0277,
      "step": 36096
    },
    {
      "epoch": 3.105652585391035,
      "grad_norm": 5.15625,
      "learning_rate": 0.00031452560458873765,
      "loss": 1.3075,
      "step": 36097
    },
    {
      "epoch": 3.105738621698357,
      "grad_norm": 31.125,
      "learning_rate": 0.0003145004933910802,
      "loss": 1.1267,
      "step": 36098
    },
    {
      "epoch": 3.1058246580056785,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003144753827359729,
      "loss": 1.1451,
      "step": 36099
    },
    {
      "epoch": 3.105910694313,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003144502726234889,
      "loss": 0.6558,
      "step": 36100
    },
    {
      "epoch": 3.1059967306203218,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003144251630537017,
      "loss": 1.3739,
      "step": 36101
    },
    {
      "epoch": 3.1060827669276434,
      "grad_norm": 1.0,
      "learning_rate": 0.0003144000540266848,
      "loss": 0.7692,
      "step": 36102
    },
    {
      "epoch": 3.106168803234965,
      "grad_norm": 340.0,
      "learning_rate": 0.00031437494554251177,
      "loss": 0.8078,
      "step": 36103
    },
    {
      "epoch": 3.1062548395422866,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003143498376012558,
      "loss": 1.4037,
      "step": 36104
    },
    {
      "epoch": 3.1063408758496087,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00031432473020299057,
      "loss": 1.5461,
      "step": 36105
    },
    {
      "epoch": 3.1064269121569303,
      "grad_norm": 11.375,
      "learning_rate": 0.00031429962334778933,
      "loss": 0.5933,
      "step": 36106
    },
    {
      "epoch": 3.106512948464252,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003142745170357256,
      "loss": 0.7434,
      "step": 36107
    },
    {
      "epoch": 3.1065989847715736,
      "grad_norm": 1.375,
      "learning_rate": 0.00031424941126687267,
      "loss": 1.5203,
      "step": 36108
    },
    {
      "epoch": 3.106685021078895,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00031422430604130427,
      "loss": 1.4285,
      "step": 36109
    },
    {
      "epoch": 3.106771057386217,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003141992013590933,
      "loss": 1.5513,
      "step": 36110
    },
    {
      "epoch": 3.1068570936935385,
      "grad_norm": 11.25,
      "learning_rate": 0.00031417409722031385,
      "loss": 1.1089,
      "step": 36111
    },
    {
      "epoch": 3.1069431300008605,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00031414899362503895,
      "loss": 1.0004,
      "step": 36112
    },
    {
      "epoch": 3.107029166308182,
      "grad_norm": 1.4296875,
      "learning_rate": 0.000314123890573342,
      "loss": 1.1332,
      "step": 36113
    },
    {
      "epoch": 3.107115202615504,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003140987880652967,
      "loss": 1.6265,
      "step": 36114
    },
    {
      "epoch": 3.1072012389228254,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003140736861009762,
      "loss": 0.7984,
      "step": 36115
    },
    {
      "epoch": 3.107287275230147,
      "grad_norm": 1.21875,
      "learning_rate": 0.00031404858468045394,
      "loss": 0.7868,
      "step": 36116
    },
    {
      "epoch": 3.1073733115374687,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0003140234838038035,
      "loss": 1.0125,
      "step": 36117
    },
    {
      "epoch": 3.1074593478447907,
      "grad_norm": 3.328125,
      "learning_rate": 0.00031399838347109826,
      "loss": 1.1682,
      "step": 36118
    },
    {
      "epoch": 3.1075453841521123,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003139732836824115,
      "loss": 0.91,
      "step": 36119
    },
    {
      "epoch": 3.107631420459434,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0003139481844378166,
      "loss": 0.7368,
      "step": 36120
    },
    {
      "epoch": 3.1077174567667556,
      "grad_norm": 1.296875,
      "learning_rate": 0.00031392308573738727,
      "loss": 1.1417,
      "step": 36121
    },
    {
      "epoch": 3.107803493074077,
      "grad_norm": 1.046875,
      "learning_rate": 0.00031389798758119666,
      "loss": 1.554,
      "step": 36122
    },
    {
      "epoch": 3.107889529381399,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00031387288996931816,
      "loss": 1.1032,
      "step": 36123
    },
    {
      "epoch": 3.1079755656887205,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003138477929018253,
      "loss": 1.2452,
      "step": 36124
    },
    {
      "epoch": 3.1080616019960425,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031382269637879144,
      "loss": 1.4335,
      "step": 36125
    },
    {
      "epoch": 3.108147638303364,
      "grad_norm": 1.125,
      "learning_rate": 0.00031379760040029,
      "loss": 1.4179,
      "step": 36126
    },
    {
      "epoch": 3.108233674610686,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003137725049663943,
      "loss": 1.9102,
      "step": 36127
    },
    {
      "epoch": 3.1083197109180074,
      "grad_norm": 9.625,
      "learning_rate": 0.0003137474100771779,
      "loss": 1.2816,
      "step": 36128
    },
    {
      "epoch": 3.108405747225329,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003137223157327142,
      "loss": 1.0276,
      "step": 36129
    },
    {
      "epoch": 3.1084917835326507,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00031369722193307623,
      "loss": 1.3731,
      "step": 36130
    },
    {
      "epoch": 3.1085778198399723,
      "grad_norm": 0.953125,
      "learning_rate": 0.00031367212867833776,
      "loss": 1.045,
      "step": 36131
    },
    {
      "epoch": 3.1086638561472943,
      "grad_norm": 1.203125,
      "learning_rate": 0.00031364703596857216,
      "loss": 1.2482,
      "step": 36132
    },
    {
      "epoch": 3.108749892454616,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00031362194380385266,
      "loss": 0.7992,
      "step": 36133
    },
    {
      "epoch": 3.1088359287619376,
      "grad_norm": 2.375,
      "learning_rate": 0.0003135968521842527,
      "loss": 1.106,
      "step": 36134
    },
    {
      "epoch": 3.108921965069259,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00031357176110984573,
      "loss": 1.7312,
      "step": 36135
    },
    {
      "epoch": 3.109008001376581,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00031354667058070516,
      "loss": 1.5814,
      "step": 36136
    },
    {
      "epoch": 3.1090940376839025,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0003135215805969043,
      "loss": 0.5646,
      "step": 36137
    },
    {
      "epoch": 3.109180073991224,
      "grad_norm": 1.0625,
      "learning_rate": 0.00031349649115851644,
      "loss": 0.7592,
      "step": 36138
    },
    {
      "epoch": 3.109266110298546,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0003134714022656151,
      "loss": 2.2808,
      "step": 36139
    },
    {
      "epoch": 3.109352146605868,
      "grad_norm": 2.640625,
      "learning_rate": 0.0003134463139182737,
      "loss": 0.612,
      "step": 36140
    },
    {
      "epoch": 3.1094381829131894,
      "grad_norm": 16.5,
      "learning_rate": 0.00031342122611656556,
      "loss": 0.7214,
      "step": 36141
    },
    {
      "epoch": 3.109524219220511,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00031339613886056396,
      "loss": 0.8222,
      "step": 36142
    },
    {
      "epoch": 3.1096102555278327,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003133710521503424,
      "loss": 0.7615,
      "step": 36143
    },
    {
      "epoch": 3.1096962918351543,
      "grad_norm": 8.6875,
      "learning_rate": 0.00031334596598597424,
      "loss": 0.9908,
      "step": 36144
    },
    {
      "epoch": 3.109782328142476,
      "grad_norm": 0.875,
      "learning_rate": 0.00031332088036753274,
      "loss": 1.2813,
      "step": 36145
    },
    {
      "epoch": 3.109868364449798,
      "grad_norm": 1.375,
      "learning_rate": 0.00031329579529509145,
      "loss": 1.3516,
      "step": 36146
    },
    {
      "epoch": 3.1099544007571196,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0003132707107687235,
      "loss": 0.9545,
      "step": 36147
    },
    {
      "epoch": 3.110040437064441,
      "grad_norm": 1.296875,
      "learning_rate": 0.00031324562678850253,
      "loss": 1.4383,
      "step": 36148
    },
    {
      "epoch": 3.110126473371763,
      "grad_norm": 2.28125,
      "learning_rate": 0.00031322054335450176,
      "loss": 1.852,
      "step": 36149
    },
    {
      "epoch": 3.1102125096790845,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003131954604667946,
      "loss": 0.581,
      "step": 36150
    },
    {
      "epoch": 3.110298545986406,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003131703781254543,
      "loss": 1.3299,
      "step": 36151
    },
    {
      "epoch": 3.110384582293728,
      "grad_norm": 1.109375,
      "learning_rate": 0.00031314529633055447,
      "loss": 1.176,
      "step": 36152
    },
    {
      "epoch": 3.11047061860105,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00031312021508216823,
      "loss": 1.1024,
      "step": 36153
    },
    {
      "epoch": 3.1105566549083714,
      "grad_norm": 5.28125,
      "learning_rate": 0.00031309513438036897,
      "loss": 1.0724,
      "step": 36154
    },
    {
      "epoch": 3.110642691215693,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00031307005422523014,
      "loss": 1.9599,
      "step": 36155
    },
    {
      "epoch": 3.1107287275230147,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000313044974616825,
      "loss": 1.0484,
      "step": 36156
    },
    {
      "epoch": 3.1108147638303363,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000313019895555227,
      "loss": 1.4754,
      "step": 36157
    },
    {
      "epoch": 3.110900800137658,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003129948170405094,
      "loss": 1.1229,
      "step": 36158
    },
    {
      "epoch": 3.11098683644498,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003129697390727456,
      "loss": 0.947,
      "step": 36159
    },
    {
      "epoch": 3.1110728727523016,
      "grad_norm": 1.5,
      "learning_rate": 0.000312944661652009,
      "loss": 1.3951,
      "step": 36160
    },
    {
      "epoch": 3.1111589090596232,
      "grad_norm": 1.0,
      "learning_rate": 0.0003129195847783728,
      "loss": 1.2109,
      "step": 36161
    },
    {
      "epoch": 3.111244945366945,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0003128945084519105,
      "loss": 0.632,
      "step": 36162
    },
    {
      "epoch": 3.1113309816742665,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003128694326726954,
      "loss": 1.2274,
      "step": 36163
    },
    {
      "epoch": 3.111417017981588,
      "grad_norm": 1.25,
      "learning_rate": 0.00031284435744080076,
      "loss": 0.6415,
      "step": 36164
    },
    {
      "epoch": 3.1115030542889097,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00031281928275629986,
      "loss": 0.7005,
      "step": 36165
    },
    {
      "epoch": 3.111589090596232,
      "grad_norm": 1.296875,
      "learning_rate": 0.00031279420861926626,
      "loss": 1.7135,
      "step": 36166
    },
    {
      "epoch": 3.1116751269035534,
      "grad_norm": 2.21875,
      "learning_rate": 0.00031276913502977324,
      "loss": 0.8014,
      "step": 36167
    },
    {
      "epoch": 3.111761163210875,
      "grad_norm": 1.234375,
      "learning_rate": 0.000312744061987894,
      "loss": 1.0431,
      "step": 36168
    },
    {
      "epoch": 3.1118471995181967,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00031271898949370195,
      "loss": 1.0012,
      "step": 36169
    },
    {
      "epoch": 3.1119332358255183,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003126939175472705,
      "loss": 1.3892,
      "step": 36170
    },
    {
      "epoch": 3.11201927213284,
      "grad_norm": 1.0625,
      "learning_rate": 0.000312668846148673,
      "loss": 0.7765,
      "step": 36171
    },
    {
      "epoch": 3.112105308440162,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0003126437752979824,
      "loss": 0.6614,
      "step": 36172
    },
    {
      "epoch": 3.1121913447474836,
      "grad_norm": 2.875,
      "learning_rate": 0.0003126187049952726,
      "loss": 0.6084,
      "step": 36173
    },
    {
      "epoch": 3.1122773810548052,
      "grad_norm": 1.046875,
      "learning_rate": 0.00031259363524061657,
      "loss": 1.1304,
      "step": 36174
    },
    {
      "epoch": 3.112363417362127,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00031256856603408754,
      "loss": 1.1268,
      "step": 36175
    },
    {
      "epoch": 3.1124494536694485,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003125434973757592,
      "loss": 0.8897,
      "step": 36176
    },
    {
      "epoch": 3.11253548997677,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003125184292657046,
      "loss": 0.9092,
      "step": 36177
    },
    {
      "epoch": 3.1126215262840917,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003124933617039972,
      "loss": 0.8605,
      "step": 36178
    },
    {
      "epoch": 3.112707562591414,
      "grad_norm": 1.125,
      "learning_rate": 0.00031246829469071014,
      "loss": 1.0024,
      "step": 36179
    },
    {
      "epoch": 3.1127935988987354,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00031244322822591693,
      "loss": 0.8607,
      "step": 36180
    },
    {
      "epoch": 3.112879635206057,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00031241816230969077,
      "loss": 0.6946,
      "step": 36181
    },
    {
      "epoch": 3.1129656715133787,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031239309694210495,
      "loss": 1.451,
      "step": 36182
    },
    {
      "epoch": 3.1130517078207003,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003123680321232329,
      "loss": 1.1843,
      "step": 36183
    },
    {
      "epoch": 3.113137744128022,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00031234296785314763,
      "loss": 1.309,
      "step": 36184
    },
    {
      "epoch": 3.1132237804353435,
      "grad_norm": 502.0,
      "learning_rate": 0.000312317904131923,
      "loss": 1.3791,
      "step": 36185
    },
    {
      "epoch": 3.1133098167426656,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003122928409596319,
      "loss": 1.5417,
      "step": 36186
    },
    {
      "epoch": 3.1133958530499872,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00031226777833634776,
      "loss": 1.7169,
      "step": 36187
    },
    {
      "epoch": 3.113481889357309,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00031224271626214374,
      "loss": 1.5765,
      "step": 36188
    },
    {
      "epoch": 3.1135679256646305,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0003122176547370934,
      "loss": 1.2526,
      "step": 36189
    },
    {
      "epoch": 3.113653961971952,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00031219259376126984,
      "loss": 1.2218,
      "step": 36190
    },
    {
      "epoch": 3.1137399982792737,
      "grad_norm": 32512.0,
      "learning_rate": 0.00031216753333474634,
      "loss": 1.5809,
      "step": 36191
    },
    {
      "epoch": 3.1138260345865953,
      "grad_norm": 1.640625,
      "learning_rate": 0.0003121424734575964,
      "loss": 1.0127,
      "step": 36192
    },
    {
      "epoch": 3.1139120708939174,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00031211741412989313,
      "loss": 1.1328,
      "step": 36193
    },
    {
      "epoch": 3.113998107201239,
      "grad_norm": 1.390625,
      "learning_rate": 0.00031209235535171,
      "loss": 0.6689,
      "step": 36194
    },
    {
      "epoch": 3.1140841435085607,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00031206729712311996,
      "loss": 1.3329,
      "step": 36195
    },
    {
      "epoch": 3.1141701798158823,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003120422394441967,
      "loss": 1.6404,
      "step": 36196
    },
    {
      "epoch": 3.114256216123204,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003120171823150133,
      "loss": 0.9573,
      "step": 36197
    },
    {
      "epoch": 3.1143422524305255,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00031199212573564296,
      "loss": 1.4434,
      "step": 36198
    },
    {
      "epoch": 3.114428288737847,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00031196706970615916,
      "loss": 1.1412,
      "step": 36199
    },
    {
      "epoch": 3.1145143250451692,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0003119420142266353,
      "loss": 1.0428,
      "step": 36200
    },
    {
      "epoch": 3.114600361352491,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00031191695929714424,
      "loss": 0.6671,
      "step": 36201
    },
    {
      "epoch": 3.1146863976598125,
      "grad_norm": 15.4375,
      "learning_rate": 0.00031189190491775955,
      "loss": 1.9406,
      "step": 36202
    },
    {
      "epoch": 3.114772433967134,
      "grad_norm": 1.09375,
      "learning_rate": 0.00031186685108855445,
      "loss": 1.3639,
      "step": 36203
    },
    {
      "epoch": 3.1148584702744557,
      "grad_norm": 1.078125,
      "learning_rate": 0.00031184179780960233,
      "loss": 1.2077,
      "step": 36204
    },
    {
      "epoch": 3.1149445065817774,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003118167450809762,
      "loss": 0.7527,
      "step": 36205
    },
    {
      "epoch": 3.115030542889099,
      "grad_norm": 1.03125,
      "learning_rate": 0.00031179169290274956,
      "loss": 1.2523,
      "step": 36206
    },
    {
      "epoch": 3.115116579196421,
      "grad_norm": 2.71875,
      "learning_rate": 0.0003117666412749956,
      "loss": 1.6535,
      "step": 36207
    },
    {
      "epoch": 3.1152026155037427,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00031174159019778765,
      "loss": 0.9229,
      "step": 36208
    },
    {
      "epoch": 3.1152886518110643,
      "grad_norm": 1.078125,
      "learning_rate": 0.00031171653967119886,
      "loss": 1.1017,
      "step": 36209
    },
    {
      "epoch": 3.115374688118386,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0003116914896953026,
      "loss": 1.2675,
      "step": 36210
    },
    {
      "epoch": 3.1154607244257075,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003116664402701721,
      "loss": 1.0126,
      "step": 36211
    },
    {
      "epoch": 3.115546760733029,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003116413913958807,
      "loss": 1.0827,
      "step": 36212
    },
    {
      "epoch": 3.1156327970403512,
      "grad_norm": 1.125,
      "learning_rate": 0.0003116163430725015,
      "loss": 0.4653,
      "step": 36213
    },
    {
      "epoch": 3.115718833347673,
      "grad_norm": 1.078125,
      "learning_rate": 0.00031159129530010794,
      "loss": 1.2977,
      "step": 36214
    },
    {
      "epoch": 3.1158048696549945,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00031156624807877323,
      "loss": 0.8512,
      "step": 36215
    },
    {
      "epoch": 3.115890905962316,
      "grad_norm": 0.953125,
      "learning_rate": 0.00031154120140857046,
      "loss": 0.4276,
      "step": 36216
    },
    {
      "epoch": 3.1159769422696377,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003115161552895731,
      "loss": 1.4605,
      "step": 36217
    },
    {
      "epoch": 3.1160629785769594,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003114911097218543,
      "loss": 1.1967,
      "step": 36218
    },
    {
      "epoch": 3.116149014884281,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00031146606470548743,
      "loss": 1.2024,
      "step": 36219
    },
    {
      "epoch": 3.116235051191603,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003114410202405455,
      "loss": 0.864,
      "step": 36220
    },
    {
      "epoch": 3.1163210874989247,
      "grad_norm": 1.015625,
      "learning_rate": 0.00031141597632710185,
      "loss": 1.3464,
      "step": 36221
    },
    {
      "epoch": 3.1164071238062463,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003113909329652299,
      "loss": 1.0618,
      "step": 36222
    },
    {
      "epoch": 3.116493160113568,
      "grad_norm": 1.3125,
      "learning_rate": 0.0003113658901550029,
      "loss": 1.911,
      "step": 36223
    },
    {
      "epoch": 3.1165791964208895,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00031134084789649397,
      "loss": 0.7455,
      "step": 36224
    },
    {
      "epoch": 3.116665232728211,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0003113158061897762,
      "loss": 0.7249,
      "step": 36225
    },
    {
      "epoch": 3.116751269035533,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00031129076503492314,
      "loss": 0.9691,
      "step": 36226
    },
    {
      "epoch": 3.116837305342855,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003112657244320079,
      "loss": 1.5683,
      "step": 36227
    },
    {
      "epoch": 3.1169233416501765,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003112406843811037,
      "loss": 0.8663,
      "step": 36228
    },
    {
      "epoch": 3.117009377957498,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003112156448822837,
      "loss": 0.5797,
      "step": 36229
    },
    {
      "epoch": 3.1170954142648197,
      "grad_norm": 1.125,
      "learning_rate": 0.0003111906059356213,
      "loss": 1.5521,
      "step": 36230
    },
    {
      "epoch": 3.1171814505721414,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00031116556754118963,
      "loss": 0.6696,
      "step": 36231
    },
    {
      "epoch": 3.117267486879463,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003111405296990619,
      "loss": 0.8106,
      "step": 36232
    },
    {
      "epoch": 3.117353523186785,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00031111549240931145,
      "loss": 0.9097,
      "step": 36233
    },
    {
      "epoch": 3.1174395594941067,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003110904556720115,
      "loss": 0.7786,
      "step": 36234
    },
    {
      "epoch": 3.1175255958014283,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003110654194872352,
      "loss": 1.4967,
      "step": 36235
    },
    {
      "epoch": 3.11761163210875,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00031104038385505565,
      "loss": 1.8434,
      "step": 36236
    },
    {
      "epoch": 3.1176976684160715,
      "grad_norm": 540.0,
      "learning_rate": 0.00031101534877554636,
      "loss": 1.0112,
      "step": 36237
    },
    {
      "epoch": 3.117783704723393,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00031099031424878046,
      "loss": 1.4391,
      "step": 36238
    },
    {
      "epoch": 3.117869741030715,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00031096528027483103,
      "loss": 1.1163,
      "step": 36239
    },
    {
      "epoch": 3.117955777338037,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003109402468537714,
      "loss": 1.1861,
      "step": 36240
    },
    {
      "epoch": 3.1180418136453585,
      "grad_norm": 1.484375,
      "learning_rate": 0.0003109152139856749,
      "loss": 0.8036,
      "step": 36241
    },
    {
      "epoch": 3.11812784995268,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00031089018167061456,
      "loss": 1.4223,
      "step": 36242
    },
    {
      "epoch": 3.1182138862600017,
      "grad_norm": 1.3125,
      "learning_rate": 0.00031086514990866357,
      "loss": 1.2707,
      "step": 36243
    },
    {
      "epoch": 3.1182999225673234,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003108401186998953,
      "loss": 1.0866,
      "step": 36244
    },
    {
      "epoch": 3.118385958874645,
      "grad_norm": 0.875,
      "learning_rate": 0.00031081508804438297,
      "loss": 0.8088,
      "step": 36245
    },
    {
      "epoch": 3.1184719951819666,
      "grad_norm": 1.515625,
      "learning_rate": 0.0003107900579421995,
      "loss": 2.603,
      "step": 36246
    },
    {
      "epoch": 3.1185580314892887,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00031076502839341854,
      "loss": 0.9812,
      "step": 36247
    },
    {
      "epoch": 3.1186440677966103,
      "grad_norm": 1.125,
      "learning_rate": 0.00031073999939811297,
      "loss": 1.452,
      "step": 36248
    },
    {
      "epoch": 3.118730104103932,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00031071497095635614,
      "loss": 0.8421,
      "step": 36249
    },
    {
      "epoch": 3.1188161404112535,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00031068994306822107,
      "loss": 1.5322,
      "step": 36250
    },
    {
      "epoch": 3.118902176718575,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00031066491573378117,
      "loss": 0.8434,
      "step": 36251
    },
    {
      "epoch": 3.118988213025897,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00031063988895310967,
      "loss": 1.5143,
      "step": 36252
    },
    {
      "epoch": 3.1190742493332184,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003106148627262796,
      "loss": 1.4691,
      "step": 36253
    },
    {
      "epoch": 3.1191602856405405,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0003105898370533642,
      "loss": 0.9494,
      "step": 36254
    },
    {
      "epoch": 3.119246321947862,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00031056481193443667,
      "loss": 0.9935,
      "step": 36255
    },
    {
      "epoch": 3.1193323582551837,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003105397873695703,
      "loss": 1.7715,
      "step": 36256
    },
    {
      "epoch": 3.1194183945625054,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003105147633588381,
      "loss": 1.3844,
      "step": 36257
    },
    {
      "epoch": 3.119504430869827,
      "grad_norm": 2.3125,
      "learning_rate": 0.0003104897399023133,
      "loss": 0.9219,
      "step": 36258
    },
    {
      "epoch": 3.1195904671771486,
      "grad_norm": 0.9375,
      "learning_rate": 0.00031046471700006917,
      "loss": 0.75,
      "step": 36259
    },
    {
      "epoch": 3.1196765034844702,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000310439694652179,
      "loss": 1.8469,
      "step": 36260
    },
    {
      "epoch": 3.1197625397917923,
      "grad_norm": 1.640625,
      "learning_rate": 0.0003104146728587159,
      "loss": 1.9683,
      "step": 36261
    },
    {
      "epoch": 3.119848576099114,
      "grad_norm": 1.59375,
      "learning_rate": 0.00031038965161975284,
      "loss": 1.0612,
      "step": 36262
    },
    {
      "epoch": 3.1199346124064355,
      "grad_norm": 1.125,
      "learning_rate": 0.0003103646309353634,
      "loss": 1.014,
      "step": 36263
    },
    {
      "epoch": 3.120020648713757,
      "grad_norm": 1.53125,
      "learning_rate": 0.0003103396108056204,
      "loss": 1.4271,
      "step": 36264
    },
    {
      "epoch": 3.120106685021079,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003103145912305972,
      "loss": 0.419,
      "step": 36265
    },
    {
      "epoch": 3.1201927213284004,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003102895722103668,
      "loss": 0.9316,
      "step": 36266
    },
    {
      "epoch": 3.1202787576357225,
      "grad_norm": 1.0625,
      "learning_rate": 0.00031026455374500263,
      "loss": 1.3491,
      "step": 36267
    },
    {
      "epoch": 3.120364793943044,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00031023953583457774,
      "loss": 0.5647,
      "step": 36268
    },
    {
      "epoch": 3.1204508302503657,
      "grad_norm": 1.21875,
      "learning_rate": 0.00031021451847916515,
      "loss": 1.9183,
      "step": 36269
    },
    {
      "epoch": 3.1205368665576874,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003101895016788384,
      "loss": 0.7112,
      "step": 36270
    },
    {
      "epoch": 3.120622902865009,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003101644854336703,
      "loss": 1.5707,
      "step": 36271
    },
    {
      "epoch": 3.1207089391723306,
      "grad_norm": 111.5,
      "learning_rate": 0.0003101394697437343,
      "loss": 1.5563,
      "step": 36272
    },
    {
      "epoch": 3.1207949754796522,
      "grad_norm": 2.484375,
      "learning_rate": 0.0003101144546091032,
      "loss": 1.2921,
      "step": 36273
    },
    {
      "epoch": 3.1208810117869743,
      "grad_norm": 0.90625,
      "learning_rate": 0.0003100894400298505,
      "loss": 1.0121,
      "step": 36274
    },
    {
      "epoch": 3.120967048094296,
      "grad_norm": 1.4375,
      "learning_rate": 0.0003100644260060493,
      "loss": 0.8696,
      "step": 36275
    },
    {
      "epoch": 3.1210530844016176,
      "grad_norm": 6.4375,
      "learning_rate": 0.00031003941253777273,
      "loss": 1.2196,
      "step": 36276
    },
    {
      "epoch": 3.121139120708939,
      "grad_norm": 1.46875,
      "learning_rate": 0.0003100143996250938,
      "loss": 1.2639,
      "step": 36277
    },
    {
      "epoch": 3.121225157016261,
      "grad_norm": 2.421875,
      "learning_rate": 0.0003099893872680859,
      "loss": 1.3457,
      "step": 36278
    },
    {
      "epoch": 3.1213111933235824,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003099643754668221,
      "loss": 0.9691,
      "step": 36279
    },
    {
      "epoch": 3.121397229630904,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00030993936422137546,
      "loss": 0.9247,
      "step": 36280
    },
    {
      "epoch": 3.121483265938226,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00030991435353181927,
      "loss": 1.0205,
      "step": 36281
    },
    {
      "epoch": 3.1215693022455477,
      "grad_norm": 0.921875,
      "learning_rate": 0.00030988934339822667,
      "loss": 1.3089,
      "step": 36282
    },
    {
      "epoch": 3.1216553385528694,
      "grad_norm": 1.125,
      "learning_rate": 0.00030986433382067073,
      "loss": 0.9275,
      "step": 36283
    },
    {
      "epoch": 3.121741374860191,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003098393247992246,
      "loss": 0.5971,
      "step": 36284
    },
    {
      "epoch": 3.1218274111675126,
      "grad_norm": 2.640625,
      "learning_rate": 0.0003098143163339615,
      "loss": 0.8255,
      "step": 36285
    },
    {
      "epoch": 3.1219134474748342,
      "grad_norm": 1.296875,
      "learning_rate": 0.00030978930842495456,
      "loss": 1.2554,
      "step": 36286
    },
    {
      "epoch": 3.121999483782156,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003097643010722768,
      "loss": 1.2378,
      "step": 36287
    },
    {
      "epoch": 3.122085520089478,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003097392942760015,
      "loss": 1.3724,
      "step": 36288
    },
    {
      "epoch": 3.1221715563967996,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003097142880362018,
      "loss": 1.2512,
      "step": 36289
    },
    {
      "epoch": 3.122257592704121,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003096892823529508,
      "loss": 0.6279,
      "step": 36290
    },
    {
      "epoch": 3.122343629011443,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00030966427722632157,
      "loss": 1.3493,
      "step": 36291
    },
    {
      "epoch": 3.1224296653187644,
      "grad_norm": 27.5,
      "learning_rate": 0.0003096392726563874,
      "loss": 1.2322,
      "step": 36292
    },
    {
      "epoch": 3.122515701626086,
      "grad_norm": 21.75,
      "learning_rate": 0.00030961426864322127,
      "loss": 1.2125,
      "step": 36293
    },
    {
      "epoch": 3.122601737933408,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00030958926518689646,
      "loss": 0.7483,
      "step": 36294
    },
    {
      "epoch": 3.1226877742407297,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003095642622874858,
      "loss": 1.3971,
      "step": 36295
    },
    {
      "epoch": 3.1227738105480514,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003095392599450627,
      "loss": 0.8351,
      "step": 36296
    },
    {
      "epoch": 3.122859846855373,
      "grad_norm": 1.421875,
      "learning_rate": 0.0003095142581597003,
      "loss": 1.0204,
      "step": 36297
    },
    {
      "epoch": 3.1229458831626946,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003094892569314717,
      "loss": 0.7888,
      "step": 36298
    },
    {
      "epoch": 3.1230319194700162,
      "grad_norm": 1.0625,
      "learning_rate": 0.00030946425626044994,
      "loss": 0.8355,
      "step": 36299
    },
    {
      "epoch": 3.123117955777338,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00030943925614670807,
      "loss": 0.8702,
      "step": 36300
    },
    {
      "epoch": 3.12320399208466,
      "grad_norm": 0.96875,
      "learning_rate": 0.00030941425659031946,
      "loss": 0.9773,
      "step": 36301
    },
    {
      "epoch": 3.1232900283919816,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003093892575913571,
      "loss": 1.1141,
      "step": 36302
    },
    {
      "epoch": 3.123376064699303,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003093642591498939,
      "loss": 1.036,
      "step": 36303
    },
    {
      "epoch": 3.123462101006625,
      "grad_norm": 1.125,
      "learning_rate": 0.0003093392612660033,
      "loss": 1.4116,
      "step": 36304
    },
    {
      "epoch": 3.1235481373139464,
      "grad_norm": 1.53125,
      "learning_rate": 0.00030931426393975833,
      "loss": 1.272,
      "step": 36305
    },
    {
      "epoch": 3.123634173621268,
      "grad_norm": 1.140625,
      "learning_rate": 0.00030928926717123194,
      "loss": 1.1278,
      "step": 36306
    },
    {
      "epoch": 3.1237202099285897,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003092642709604974,
      "loss": 1.3168,
      "step": 36307
    },
    {
      "epoch": 3.1238062462359117,
      "grad_norm": 1.390625,
      "learning_rate": 0.00030923927530762777,
      "loss": 1.0951,
      "step": 36308
    },
    {
      "epoch": 3.1238922825432334,
      "grad_norm": 1.15625,
      "learning_rate": 0.00030921428021269613,
      "loss": 1.7557,
      "step": 36309
    },
    {
      "epoch": 3.123978318850555,
      "grad_norm": 1.40625,
      "learning_rate": 0.00030918928567577565,
      "loss": 1.8533,
      "step": 36310
    },
    {
      "epoch": 3.1240643551578766,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0003091642916969394,
      "loss": 1.6136,
      "step": 36311
    },
    {
      "epoch": 3.1241503914651982,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003091392982762605,
      "loss": 0.9059,
      "step": 36312
    },
    {
      "epoch": 3.12423642777252,
      "grad_norm": 1.234375,
      "learning_rate": 0.000309114305413812,
      "loss": 0.9565,
      "step": 36313
    },
    {
      "epoch": 3.1243224640798415,
      "grad_norm": 22.25,
      "learning_rate": 0.00030908931310966696,
      "loss": 1.1098,
      "step": 36314
    },
    {
      "epoch": 3.1244085003871636,
      "grad_norm": 1.328125,
      "learning_rate": 0.00030906432136389866,
      "loss": 1.211,
      "step": 36315
    },
    {
      "epoch": 3.124494536694485,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003090393301765801,
      "loss": 1.0087,
      "step": 36316
    },
    {
      "epoch": 3.124580573001807,
      "grad_norm": 1.046875,
      "learning_rate": 0.00030901433954778435,
      "loss": 0.6921,
      "step": 36317
    },
    {
      "epoch": 3.1246666093091284,
      "grad_norm": 1.796875,
      "learning_rate": 0.00030898934947758437,
      "loss": 1.3514,
      "step": 36318
    },
    {
      "epoch": 3.12475264561645,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00030896435996605355,
      "loss": 0.6954,
      "step": 36319
    },
    {
      "epoch": 3.1248386819237717,
      "grad_norm": 1.796875,
      "learning_rate": 0.00030893937101326475,
      "loss": 1.3856,
      "step": 36320
    },
    {
      "epoch": 3.1249247182310933,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003089143826192911,
      "loss": 1.1832,
      "step": 36321
    },
    {
      "epoch": 3.1250107545384154,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003088893947842058,
      "loss": 1.9019,
      "step": 36322
    },
    {
      "epoch": 3.125096790845737,
      "grad_norm": 1.546875,
      "learning_rate": 0.00030886440750808186,
      "loss": 1.4519,
      "step": 36323
    },
    {
      "epoch": 3.1251828271530586,
      "grad_norm": 4.03125,
      "learning_rate": 0.0003088394207909923,
      "loss": 0.95,
      "step": 36324
    },
    {
      "epoch": 3.1252688634603802,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00030881443463301017,
      "loss": 0.5221,
      "step": 36325
    },
    {
      "epoch": 3.125354899767702,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0003087894490342088,
      "loss": 1.0464,
      "step": 36326
    },
    {
      "epoch": 3.1254409360750235,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00030876446399466105,
      "loss": 1.469,
      "step": 36327
    },
    {
      "epoch": 3.125526972382345,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00030873947951443997,
      "loss": 0.5143,
      "step": 36328
    },
    {
      "epoch": 3.125613008689667,
      "grad_norm": 0.875,
      "learning_rate": 0.0003087144955936188,
      "loss": 0.4288,
      "step": 36329
    },
    {
      "epoch": 3.125699044996989,
      "grad_norm": 1.171875,
      "learning_rate": 0.00030868951223227055,
      "loss": 0.7241,
      "step": 36330
    },
    {
      "epoch": 3.1257850813043104,
      "grad_norm": 1.0625,
      "learning_rate": 0.00030866452943046827,
      "loss": 0.9279,
      "step": 36331
    },
    {
      "epoch": 3.125871117611632,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0003086395471882848,
      "loss": 0.7539,
      "step": 36332
    },
    {
      "epoch": 3.1259571539189537,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00030861456550579353,
      "loss": 1.9256,
      "step": 36333
    },
    {
      "epoch": 3.1260431902262753,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003085895843830675,
      "loss": 1.2649,
      "step": 36334
    },
    {
      "epoch": 3.1261292265335974,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00030856460382017983,
      "loss": 1.5432,
      "step": 36335
    },
    {
      "epoch": 3.126215262840919,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00030853962381720333,
      "loss": 1.4249,
      "step": 36336
    },
    {
      "epoch": 3.1263012991482406,
      "grad_norm": 1.5,
      "learning_rate": 0.0003085146443742112,
      "loss": 0.8668,
      "step": 36337
    },
    {
      "epoch": 3.1263873354555622,
      "grad_norm": 1.40625,
      "learning_rate": 0.00030848966549127657,
      "loss": 1.4515,
      "step": 36338
    },
    {
      "epoch": 3.126473371762884,
      "grad_norm": 6.4375,
      "learning_rate": 0.00030846468716847235,
      "loss": 1.185,
      "step": 36339
    },
    {
      "epoch": 3.1265594080702055,
      "grad_norm": 1.71875,
      "learning_rate": 0.00030843970940587175,
      "loss": 1.751,
      "step": 36340
    },
    {
      "epoch": 3.126645444377527,
      "grad_norm": 2.53125,
      "learning_rate": 0.0003084147322035476,
      "loss": 0.9717,
      "step": 36341
    },
    {
      "epoch": 3.126731480684849,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003083897555615732,
      "loss": 0.8991,
      "step": 36342
    },
    {
      "epoch": 3.126817516992171,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0003083647794800215,
      "loss": 1.062,
      "step": 36343
    },
    {
      "epoch": 3.1269035532994924,
      "grad_norm": 1.140625,
      "learning_rate": 0.00030833980395896546,
      "loss": 0.8346,
      "step": 36344
    },
    {
      "epoch": 3.126989589606814,
      "grad_norm": 1.578125,
      "learning_rate": 0.0003083148289984783,
      "loss": 1.4615,
      "step": 36345
    },
    {
      "epoch": 3.1270756259141357,
      "grad_norm": 1.6015625,
      "learning_rate": 0.000308289854598633,
      "loss": 1.3373,
      "step": 36346
    },
    {
      "epoch": 3.1271616622214573,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0003082648807595026,
      "loss": 1.3802,
      "step": 36347
    },
    {
      "epoch": 3.1272476985287794,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00030823990748116,
      "loss": 1.0776,
      "step": 36348
    },
    {
      "epoch": 3.127333734836101,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00030821493476367853,
      "loss": 1.7451,
      "step": 36349
    },
    {
      "epoch": 3.1274197711434226,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00030818996260713105,
      "loss": 1.2876,
      "step": 36350
    },
    {
      "epoch": 3.1275058074507442,
      "grad_norm": 1.203125,
      "learning_rate": 0.00030816499101159043,
      "loss": 0.8952,
      "step": 36351
    },
    {
      "epoch": 3.127591843758066,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0003081400199771301,
      "loss": 1.0311,
      "step": 36352
    },
    {
      "epoch": 3.1276778800653875,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00030811504950382294,
      "loss": 0.6072,
      "step": 36353
    },
    {
      "epoch": 3.127763916372709,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0003080900795917419,
      "loss": 1.0521,
      "step": 36354
    },
    {
      "epoch": 3.127849952680031,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00030806511024095997,
      "loss": 1.6711,
      "step": 36355
    },
    {
      "epoch": 3.127935988987353,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003080401414515503,
      "loss": 1.187,
      "step": 36356
    },
    {
      "epoch": 3.1280220252946744,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003080151732235859,
      "loss": 1.0213,
      "step": 36357
    },
    {
      "epoch": 3.128108061601996,
      "grad_norm": 0.890625,
      "learning_rate": 0.0003079902055571397,
      "loss": 0.9961,
      "step": 36358
    },
    {
      "epoch": 3.1281940979093177,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003079652384522849,
      "loss": 0.7652,
      "step": 36359
    },
    {
      "epoch": 3.1282801342166393,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003079402719090944,
      "loss": 1.5998,
      "step": 36360
    },
    {
      "epoch": 3.128366170523961,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00030791530592764126,
      "loss": 0.8948,
      "step": 36361
    },
    {
      "epoch": 3.128452206831283,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003078903405079985,
      "loss": 1.3893,
      "step": 36362
    },
    {
      "epoch": 3.1285382431386046,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003078653756502391,
      "loss": 0.7436,
      "step": 36363
    },
    {
      "epoch": 3.1286242794459262,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003078404113544361,
      "loss": 0.8105,
      "step": 36364
    },
    {
      "epoch": 3.128710315753248,
      "grad_norm": 7.25,
      "learning_rate": 0.0003078154476206626,
      "loss": 1.8309,
      "step": 36365
    },
    {
      "epoch": 3.1287963520605695,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003077904844489914,
      "loss": 0.7998,
      "step": 36366
    },
    {
      "epoch": 3.128882388367891,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0003077655218394958,
      "loss": 0.9072,
      "step": 36367
    },
    {
      "epoch": 3.1289684246752127,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00030774055979224856,
      "loss": 0.9491,
      "step": 36368
    },
    {
      "epoch": 3.129054460982535,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00030771559830732265,
      "loss": 1.2045,
      "step": 36369
    },
    {
      "epoch": 3.1291404972898564,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003076906373847915,
      "loss": 0.8359,
      "step": 36370
    },
    {
      "epoch": 3.129226533597178,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0003076656770247276,
      "loss": 0.9859,
      "step": 36371
    },
    {
      "epoch": 3.1293125699044997,
      "grad_norm": 0.8125,
      "learning_rate": 0.0003076407172272043,
      "loss": 0.4886,
      "step": 36372
    },
    {
      "epoch": 3.1293986062118213,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003076157579922946,
      "loss": 1.1813,
      "step": 36373
    },
    {
      "epoch": 3.129484642519143,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003075907993200712,
      "loss": 1.2775,
      "step": 36374
    },
    {
      "epoch": 3.1295706788264646,
      "grad_norm": 1.453125,
      "learning_rate": 0.0003075658412106074,
      "loss": 1.7414,
      "step": 36375
    },
    {
      "epoch": 3.1296567151337866,
      "grad_norm": 1.015625,
      "learning_rate": 0.00030754088366397617,
      "loss": 1.0509,
      "step": 36376
    },
    {
      "epoch": 3.1297427514411083,
      "grad_norm": 42.0,
      "learning_rate": 0.0003075159266802503,
      "loss": 1.427,
      "step": 36377
    },
    {
      "epoch": 3.12982878774843,
      "grad_norm": 1.640625,
      "learning_rate": 0.0003074909702595029,
      "loss": 1.5225,
      "step": 36378
    },
    {
      "epoch": 3.1299148240557515,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000307466014401807,
      "loss": 0.9241,
      "step": 36379
    },
    {
      "epoch": 3.130000860363073,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00030744105910723564,
      "loss": 1.2974,
      "step": 36380
    },
    {
      "epoch": 3.1300868966703947,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00030741610437586166,
      "loss": 1.2676,
      "step": 36381
    },
    {
      "epoch": 3.1301729329777164,
      "grad_norm": 1.34375,
      "learning_rate": 0.00030739115020775803,
      "loss": 2.0882,
      "step": 36382
    },
    {
      "epoch": 3.1302589692850384,
      "grad_norm": 1.1875,
      "learning_rate": 0.00030736619660299793,
      "loss": 1.2315,
      "step": 36383
    },
    {
      "epoch": 3.13034500559236,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0003073412435616543,
      "loss": 0.7568,
      "step": 36384
    },
    {
      "epoch": 3.1304310418996817,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0003073162910837999,
      "loss": 0.6999,
      "step": 36385
    },
    {
      "epoch": 3.1305170782070033,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00030729133916950797,
      "loss": 1.0305,
      "step": 36386
    },
    {
      "epoch": 3.130603114514325,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003072663878188514,
      "loss": 0.6306,
      "step": 36387
    },
    {
      "epoch": 3.1306891508216466,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003072414370319032,
      "loss": 1.6244,
      "step": 36388
    },
    {
      "epoch": 3.1307751871289686,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00030721648680873614,
      "loss": 0.6975,
      "step": 36389
    },
    {
      "epoch": 3.1308612234362903,
      "grad_norm": 1.703125,
      "learning_rate": 0.00030719153714942347,
      "loss": 1.65,
      "step": 36390
    },
    {
      "epoch": 3.130947259743612,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0003071665880540382,
      "loss": 1.2673,
      "step": 36391
    },
    {
      "epoch": 3.1310332960509335,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0003071416395226529,
      "loss": 1.2759,
      "step": 36392
    },
    {
      "epoch": 3.131119332358255,
      "grad_norm": 33.75,
      "learning_rate": 0.00030711669155534095,
      "loss": 0.8471,
      "step": 36393
    },
    {
      "epoch": 3.1312053686655767,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003070917441521751,
      "loss": 1.2073,
      "step": 36394
    },
    {
      "epoch": 3.1312914049728984,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00030706679731322845,
      "loss": 1.2628,
      "step": 36395
    },
    {
      "epoch": 3.1313774412802204,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003070418510385738,
      "loss": 1.1665,
      "step": 36396
    },
    {
      "epoch": 3.131463477587542,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003070169053282843,
      "loss": 0.756,
      "step": 36397
    },
    {
      "epoch": 3.1315495138948637,
      "grad_norm": 9.6875,
      "learning_rate": 0.0003069919601824328,
      "loss": 1.2441,
      "step": 36398
    },
    {
      "epoch": 3.1316355502021853,
      "grad_norm": 1.109375,
      "learning_rate": 0.00030696701560109214,
      "loss": 0.9883,
      "step": 36399
    },
    {
      "epoch": 3.131721586509507,
      "grad_norm": 1.0625,
      "learning_rate": 0.00030694207158433556,
      "loss": 0.8661,
      "step": 36400
    },
    {
      "epoch": 3.1318076228168286,
      "grad_norm": 1.25,
      "learning_rate": 0.00030691712813223586,
      "loss": 0.6813,
      "step": 36401
    },
    {
      "epoch": 3.13189365912415,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00030689218524486607,
      "loss": 1.899,
      "step": 36402
    },
    {
      "epoch": 3.1319796954314723,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00030686724292229887,
      "loss": 1.4082,
      "step": 36403
    },
    {
      "epoch": 3.132065731738794,
      "grad_norm": 5.59375,
      "learning_rate": 0.0003068423011646076,
      "loss": 1.6166,
      "step": 36404
    },
    {
      "epoch": 3.1321517680461155,
      "grad_norm": 3.40625,
      "learning_rate": 0.000306817359971865,
      "loss": 1.0231,
      "step": 36405
    },
    {
      "epoch": 3.132237804353437,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003067924193441439,
      "loss": 1.6908,
      "step": 36406
    },
    {
      "epoch": 3.1323238406607588,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00030676747928151747,
      "loss": 0.9521,
      "step": 36407
    },
    {
      "epoch": 3.1324098769680804,
      "grad_norm": 1.140625,
      "learning_rate": 0.00030674253978405864,
      "loss": 1.1261,
      "step": 36408
    },
    {
      "epoch": 3.1324959132754024,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003067176008518403,
      "loss": 1.0701,
      "step": 36409
    },
    {
      "epoch": 3.132581949582724,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003066926624849354,
      "loss": 1.1813,
      "step": 36410
    },
    {
      "epoch": 3.1326679858900457,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003066677246834169,
      "loss": 0.9736,
      "step": 36411
    },
    {
      "epoch": 3.1327540221973673,
      "grad_norm": 3.921875,
      "learning_rate": 0.0003066427874473576,
      "loss": 1.2556,
      "step": 36412
    },
    {
      "epoch": 3.132840058504689,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00030661785077683057,
      "loss": 1.5288,
      "step": 36413
    },
    {
      "epoch": 3.1329260948120106,
      "grad_norm": 32.5,
      "learning_rate": 0.0003065929146719088,
      "loss": 1.3376,
      "step": 36414
    },
    {
      "epoch": 3.133012131119332,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00030656797913266497,
      "loss": 0.9958,
      "step": 36415
    },
    {
      "epoch": 3.1330981674266543,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00030654304415917235,
      "loss": 1.6336,
      "step": 36416
    },
    {
      "epoch": 3.133184203733976,
      "grad_norm": 1.25,
      "learning_rate": 0.00030651810975150366,
      "loss": 1.1797,
      "step": 36417
    },
    {
      "epoch": 3.1332702400412975,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003064931759097319,
      "loss": 1.3081,
      "step": 36418
    },
    {
      "epoch": 3.133356276348619,
      "grad_norm": 143.0,
      "learning_rate": 0.00030646824263392983,
      "loss": 0.8867,
      "step": 36419
    },
    {
      "epoch": 3.1334423126559408,
      "grad_norm": 1.34375,
      "learning_rate": 0.00030644330992417063,
      "loss": 1.6577,
      "step": 36420
    },
    {
      "epoch": 3.1335283489632624,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003064183777805271,
      "loss": 1.0839,
      "step": 36421
    },
    {
      "epoch": 3.133614385270584,
      "grad_norm": 1.0,
      "learning_rate": 0.00030639344620307207,
      "loss": 1.204,
      "step": 36422
    },
    {
      "epoch": 3.133700421577906,
      "grad_norm": 1.015625,
      "learning_rate": 0.00030636851519187864,
      "loss": 1.2562,
      "step": 36423
    },
    {
      "epoch": 3.1337864578852277,
      "grad_norm": 1.25,
      "learning_rate": 0.0003063435847470196,
      "loss": 1.3913,
      "step": 36424
    },
    {
      "epoch": 3.1338724941925493,
      "grad_norm": 648.0,
      "learning_rate": 0.00030631865486856804,
      "loss": 1.1937,
      "step": 36425
    },
    {
      "epoch": 3.133958530499871,
      "grad_norm": 1.078125,
      "learning_rate": 0.00030629372555659654,
      "loss": 1.4142,
      "step": 36426
    },
    {
      "epoch": 3.1340445668071926,
      "grad_norm": 1.046875,
      "learning_rate": 0.00030626879681117835,
      "loss": 1.1184,
      "step": 36427
    },
    {
      "epoch": 3.134130603114514,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003062438686323862,
      "loss": 1.6963,
      "step": 36428
    },
    {
      "epoch": 3.134216639421836,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00030621894102029313,
      "loss": 0.6846,
      "step": 36429
    },
    {
      "epoch": 3.134302675729158,
      "grad_norm": 1.25,
      "learning_rate": 0.0003061940139749718,
      "loss": 1.2776,
      "step": 36430
    },
    {
      "epoch": 3.1343887120364795,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00030616908749649547,
      "loss": 1.6736,
      "step": 36431
    },
    {
      "epoch": 3.134474748343801,
      "grad_norm": 1.046875,
      "learning_rate": 0.00030614416158493684,
      "loss": 1.7126,
      "step": 36432
    },
    {
      "epoch": 3.1345607846511228,
      "grad_norm": 1.046875,
      "learning_rate": 0.00030611923624036867,
      "loss": 0.793,
      "step": 36433
    },
    {
      "epoch": 3.1346468209584444,
      "grad_norm": 1.15625,
      "learning_rate": 0.00030609431146286415,
      "loss": 0.9941,
      "step": 36434
    },
    {
      "epoch": 3.134732857265766,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003060693872524961,
      "loss": 0.9226,
      "step": 36435
    },
    {
      "epoch": 3.1348188935730876,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00030604446360933733,
      "loss": 1.1951,
      "step": 36436
    },
    {
      "epoch": 3.1349049298804097,
      "grad_norm": 3.703125,
      "learning_rate": 0.0003060195405334607,
      "loss": 1.2259,
      "step": 36437
    },
    {
      "epoch": 3.1349909661877313,
      "grad_norm": 6.90625,
      "learning_rate": 0.00030599461802493936,
      "loss": 1.5698,
      "step": 36438
    },
    {
      "epoch": 3.135077002495053,
      "grad_norm": 1.0,
      "learning_rate": 0.0003059696960838459,
      "loss": 0.5324,
      "step": 36439
    },
    {
      "epoch": 3.1351630388023746,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003059447747102533,
      "loss": 1.0324,
      "step": 36440
    },
    {
      "epoch": 3.135249075109696,
      "grad_norm": 7.28125,
      "learning_rate": 0.0003059198539042346,
      "loss": 1.5931,
      "step": 36441
    },
    {
      "epoch": 3.135335111417018,
      "grad_norm": 1.53125,
      "learning_rate": 0.00030589493366586253,
      "loss": 1.8049,
      "step": 36442
    },
    {
      "epoch": 3.1354211477243394,
      "grad_norm": 1.015625,
      "learning_rate": 0.00030587001399520985,
      "loss": 1.0155,
      "step": 36443
    },
    {
      "epoch": 3.1355071840316615,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0003058450948923499,
      "loss": 0.912,
      "step": 36444
    },
    {
      "epoch": 3.135593220338983,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003058201763573551,
      "loss": 1.0278,
      "step": 36445
    },
    {
      "epoch": 3.1356792566463048,
      "grad_norm": 1.1875,
      "learning_rate": 0.00030579525839029867,
      "loss": 1.5472,
      "step": 36446
    },
    {
      "epoch": 3.1357652929536264,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0003057703409912533,
      "loss": 0.958,
      "step": 36447
    },
    {
      "epoch": 3.135851329260948,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00030574542416029197,
      "loss": 1.6569,
      "step": 36448
    },
    {
      "epoch": 3.1359373655682696,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003057205078974873,
      "loss": 1.5954,
      "step": 36449
    },
    {
      "epoch": 3.1360234018755917,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0003056955922029125,
      "loss": 1.2981,
      "step": 36450
    },
    {
      "epoch": 3.1361094381829133,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003056706770766403,
      "loss": 1.1739,
      "step": 36451
    },
    {
      "epoch": 3.136195474490235,
      "grad_norm": 0.984375,
      "learning_rate": 0.0003056457625187436,
      "loss": 1.4029,
      "step": 36452
    },
    {
      "epoch": 3.1362815107975566,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003056208485292951,
      "loss": 2.02,
      "step": 36453
    },
    {
      "epoch": 3.136367547104878,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00030559593510836794,
      "loss": 1.0409,
      "step": 36454
    },
    {
      "epoch": 3.1364535834122,
      "grad_norm": 0.953125,
      "learning_rate": 0.00030557102225603483,
      "loss": 1.2078,
      "step": 36455
    },
    {
      "epoch": 3.1365396197195214,
      "grad_norm": 1.21875,
      "learning_rate": 0.00030554610997236863,
      "loss": 1.398,
      "step": 36456
    },
    {
      "epoch": 3.1366256560268435,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003055211982574423,
      "loss": 0.8365,
      "step": 36457
    },
    {
      "epoch": 3.136711692334165,
      "grad_norm": 0.828125,
      "learning_rate": 0.0003054962871113286,
      "loss": 0.6597,
      "step": 36458
    },
    {
      "epoch": 3.1367977286414868,
      "grad_norm": 47.5,
      "learning_rate": 0.00030547137653410053,
      "loss": 1.0358,
      "step": 36459
    },
    {
      "epoch": 3.1368837649488084,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003054464665258307,
      "loss": 1.3981,
      "step": 36460
    },
    {
      "epoch": 3.13696980125613,
      "grad_norm": 0.9375,
      "learning_rate": 0.00030542155708659224,
      "loss": 0.9,
      "step": 36461
    },
    {
      "epoch": 3.1370558375634516,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003053966482164579,
      "loss": 0.5371,
      "step": 36462
    },
    {
      "epoch": 3.1371418738707737,
      "grad_norm": 1.078125,
      "learning_rate": 0.00030537173991550026,
      "loss": 1.4625,
      "step": 36463
    },
    {
      "epoch": 3.1372279101780953,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00030534683218379265,
      "loss": 0.7985,
      "step": 36464
    },
    {
      "epoch": 3.137313946485417,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003053219250214077,
      "loss": 1.2844,
      "step": 36465
    },
    {
      "epoch": 3.1373999827927386,
      "grad_norm": 4.625,
      "learning_rate": 0.0003052970184284183,
      "loss": 1.2281,
      "step": 36466
    },
    {
      "epoch": 3.13748601910006,
      "grad_norm": 5.28125,
      "learning_rate": 0.00030527211240489706,
      "loss": 1.0252,
      "step": 36467
    },
    {
      "epoch": 3.137572055407382,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003052472069509171,
      "loss": 0.9455,
      "step": 36468
    },
    {
      "epoch": 3.1376580917147034,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0003052223020665513,
      "loss": 1.0902,
      "step": 36469
    },
    {
      "epoch": 3.1377441280220255,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00030519739775187234,
      "loss": 0.5933,
      "step": 36470
    },
    {
      "epoch": 3.137830164329347,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00030517249400695295,
      "loss": 1.3255,
      "step": 36471
    },
    {
      "epoch": 3.1379162006366688,
      "grad_norm": 0.890625,
      "learning_rate": 0.00030514759083186626,
      "loss": 0.8188,
      "step": 36472
    },
    {
      "epoch": 3.1380022369439904,
      "grad_norm": 2.71875,
      "learning_rate": 0.00030512268822668496,
      "loss": 1.3522,
      "step": 36473
    },
    {
      "epoch": 3.138088273251312,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003050977861914818,
      "loss": 1.8496,
      "step": 36474
    },
    {
      "epoch": 3.1381743095586336,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003050728847263299,
      "loss": 1.0946,
      "step": 36475
    },
    {
      "epoch": 3.1382603458659553,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003050479838313017,
      "loss": 1.4973,
      "step": 36476
    },
    {
      "epoch": 3.1383463821732773,
      "grad_norm": 9.25,
      "learning_rate": 0.0003050230835064704,
      "loss": 1.0719,
      "step": 36477
    },
    {
      "epoch": 3.138432418480599,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003049981837519085,
      "loss": 0.748,
      "step": 36478
    },
    {
      "epoch": 3.1385184547879206,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003049732845676891,
      "loss": 0.9055,
      "step": 36479
    },
    {
      "epoch": 3.138604491095242,
      "grad_norm": 1.296875,
      "learning_rate": 0.0003049483859538847,
      "loss": 0.9238,
      "step": 36480
    },
    {
      "epoch": 3.138690527402564,
      "grad_norm": 1.125,
      "learning_rate": 0.0003049234879105686,
      "loss": 1.2729,
      "step": 36481
    },
    {
      "epoch": 3.1387765637098854,
      "grad_norm": 3.40625,
      "learning_rate": 0.0003048985904378133,
      "loss": 1.3211,
      "step": 36482
    },
    {
      "epoch": 3.138862600017207,
      "grad_norm": 0.75,
      "learning_rate": 0.0003048736935356915,
      "loss": 0.7576,
      "step": 36483
    },
    {
      "epoch": 3.138948636324529,
      "grad_norm": 1.515625,
      "learning_rate": 0.00030484879720427637,
      "loss": 1.4417,
      "step": 36484
    },
    {
      "epoch": 3.1390346726318508,
      "grad_norm": 4.65625,
      "learning_rate": 0.00030482390144364056,
      "loss": 0.7312,
      "step": 36485
    },
    {
      "epoch": 3.1391207089391724,
      "grad_norm": 1.0,
      "learning_rate": 0.00030479900625385683,
      "loss": 0.7882,
      "step": 36486
    },
    {
      "epoch": 3.139206745246494,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00030477411163499807,
      "loss": 1.4258,
      "step": 36487
    },
    {
      "epoch": 3.1392927815538156,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003047492175871371,
      "loss": 1.1107,
      "step": 36488
    },
    {
      "epoch": 3.1393788178611373,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00030472432411034667,
      "loss": 1.5165,
      "step": 36489
    },
    {
      "epoch": 3.139464854168459,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0003046994312046995,
      "loss": 1.3132,
      "step": 36490
    },
    {
      "epoch": 3.139550890475781,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003046745388702686,
      "loss": 1.0142,
      "step": 36491
    },
    {
      "epoch": 3.1396369267831026,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003046496471071267,
      "loss": 0.4238,
      "step": 36492
    },
    {
      "epoch": 3.139722963090424,
      "grad_norm": 1.1875,
      "learning_rate": 0.0003046247559153467,
      "loss": 0.7109,
      "step": 36493
    },
    {
      "epoch": 3.139808999397746,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00030459986529500103,
      "loss": 1.2444,
      "step": 36494
    },
    {
      "epoch": 3.1398950357050674,
      "grad_norm": 1.109375,
      "learning_rate": 0.00030457497524616296,
      "loss": 0.6831,
      "step": 36495
    },
    {
      "epoch": 3.139981072012389,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00030455008576890495,
      "loss": 0.7419,
      "step": 36496
    },
    {
      "epoch": 3.1400671083197107,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00030452519686329996,
      "loss": 0.9475,
      "step": 36497
    },
    {
      "epoch": 3.1401531446270328,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0003045003085294208,
      "loss": 1.0311,
      "step": 36498
    },
    {
      "epoch": 3.1402391809343544,
      "grad_norm": 0.96875,
      "learning_rate": 0.00030447542076734026,
      "loss": 0.6274,
      "step": 36499
    },
    {
      "epoch": 3.140325217241676,
      "grad_norm": 1.4375,
      "learning_rate": 0.000304450533577131,
      "loss": 1.2246,
      "step": 36500
    },
    {
      "epoch": 3.1404112535489976,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0003044256469588658,
      "loss": 0.8949,
      "step": 36501
    },
    {
      "epoch": 3.1404972898563193,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0003044007609126177,
      "loss": 1.5401,
      "step": 36502
    },
    {
      "epoch": 3.140583326163641,
      "grad_norm": 2.140625,
      "learning_rate": 0.00030437587543845937,
      "loss": 1.9193,
      "step": 36503
    },
    {
      "epoch": 3.1406693624709625,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00030435099053646334,
      "loss": 1.0026,
      "step": 36504
    },
    {
      "epoch": 3.1407553987782846,
      "grad_norm": 5.34375,
      "learning_rate": 0.00030432610620670283,
      "loss": 0.651,
      "step": 36505
    },
    {
      "epoch": 3.140841435085606,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003043012224492503,
      "loss": 1.6244,
      "step": 36506
    },
    {
      "epoch": 3.140927471392928,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00030427633926417873,
      "loss": 1.8797,
      "step": 36507
    },
    {
      "epoch": 3.1410135077002495,
      "grad_norm": 16.75,
      "learning_rate": 0.0003042514566515606,
      "loss": 1.6187,
      "step": 36508
    },
    {
      "epoch": 3.141099544007571,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003042265746114691,
      "loss": 1.1415,
      "step": 36509
    },
    {
      "epoch": 3.1411855803148927,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003042016931439767,
      "loss": 1.1124,
      "step": 36510
    },
    {
      "epoch": 3.1412716166222148,
      "grad_norm": 1.09375,
      "learning_rate": 0.0003041768122491562,
      "loss": 0.8061,
      "step": 36511
    },
    {
      "epoch": 3.1413576529295364,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003041519319270806,
      "loss": 1.4583,
      "step": 36512
    },
    {
      "epoch": 3.141443689236858,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00030412705217782246,
      "loss": 1.382,
      "step": 36513
    },
    {
      "epoch": 3.1415297255441796,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003041021730014545,
      "loss": 1.7299,
      "step": 36514
    },
    {
      "epoch": 3.1416157618515013,
      "grad_norm": 1.15625,
      "learning_rate": 0.0003040772943980496,
      "loss": 0.7545,
      "step": 36515
    },
    {
      "epoch": 3.141701798158823,
      "grad_norm": 0.9375,
      "learning_rate": 0.00030405241636768054,
      "loss": 0.5965,
      "step": 36516
    },
    {
      "epoch": 3.1417878344661445,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0003040275389104199,
      "loss": 0.6766,
      "step": 36517
    },
    {
      "epoch": 3.1418738707734666,
      "grad_norm": 7.9375,
      "learning_rate": 0.0003040026620263408,
      "loss": 1.7247,
      "step": 36518
    },
    {
      "epoch": 3.141959907080788,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003039777857155158,
      "loss": 0.4193,
      "step": 36519
    },
    {
      "epoch": 3.14204594338811,
      "grad_norm": 1.109375,
      "learning_rate": 0.0003039529099780175,
      "loss": 1.0368,
      "step": 36520
    },
    {
      "epoch": 3.1421319796954315,
      "grad_norm": 1.28125,
      "learning_rate": 0.00030392803481391897,
      "loss": 1.3247,
      "step": 36521
    },
    {
      "epoch": 3.142218016002753,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003039031602232928,
      "loss": 0.9724,
      "step": 36522
    },
    {
      "epoch": 3.1423040523100747,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0003038782862062117,
      "loss": 1.0578,
      "step": 36523
    },
    {
      "epoch": 3.1423900886173968,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0003038534127627484,
      "loss": 1.3278,
      "step": 36524
    },
    {
      "epoch": 3.1424761249247184,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00030382853989297583,
      "loss": 1.022,
      "step": 36525
    },
    {
      "epoch": 3.14256216123204,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003038036675969666,
      "loss": 1.5194,
      "step": 36526
    },
    {
      "epoch": 3.1426481975393616,
      "grad_norm": 1.171875,
      "learning_rate": 0.0003037787958747934,
      "loss": 1.1821,
      "step": 36527
    },
    {
      "epoch": 3.1427342338466833,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003037539247265291,
      "loss": 0.5973,
      "step": 36528
    },
    {
      "epoch": 3.142820270154005,
      "grad_norm": 1.015625,
      "learning_rate": 0.0003037290541522465,
      "loss": 0.8292,
      "step": 36529
    },
    {
      "epoch": 3.1429063064613265,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0003037041841520182,
      "loss": 1.4246,
      "step": 36530
    },
    {
      "epoch": 3.1429923427686486,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0003036793147259169,
      "loss": 1.4001,
      "step": 36531
    },
    {
      "epoch": 3.14307837907597,
      "grad_norm": 1.0,
      "learning_rate": 0.0003036544458740155,
      "loss": 0.7956,
      "step": 36532
    },
    {
      "epoch": 3.143164415383292,
      "grad_norm": 1.125,
      "learning_rate": 0.00030362957759638666,
      "loss": 0.7774,
      "step": 36533
    },
    {
      "epoch": 3.1432504516906135,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00030360470989310303,
      "loss": 0.6421,
      "step": 36534
    },
    {
      "epoch": 3.143336487997935,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003035798427642375,
      "loss": 1.5468,
      "step": 36535
    },
    {
      "epoch": 3.1434225243052567,
      "grad_norm": 3.6875,
      "learning_rate": 0.0003035549762098627,
      "loss": 0.9826,
      "step": 36536
    },
    {
      "epoch": 3.1435085606125783,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00030353011023005153,
      "loss": 0.9012,
      "step": 36537
    },
    {
      "epoch": 3.1435945969199004,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003035052448248763,
      "loss": 1.4761,
      "step": 36538
    },
    {
      "epoch": 3.143680633227222,
      "grad_norm": 0.9375,
      "learning_rate": 0.00030348037999441025,
      "loss": 1.1713,
      "step": 36539
    },
    {
      "epoch": 3.1437666695345436,
      "grad_norm": 1.015625,
      "learning_rate": 0.00030345551573872576,
      "loss": 1.2443,
      "step": 36540
    },
    {
      "epoch": 3.1438527058418653,
      "grad_norm": 1.109375,
      "learning_rate": 0.00030343065205789573,
      "loss": 1.3885,
      "step": 36541
    },
    {
      "epoch": 3.143938742149187,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0003034057889519927,
      "loss": 1.0778,
      "step": 36542
    },
    {
      "epoch": 3.1440247784565085,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003033809264210896,
      "loss": 0.7243,
      "step": 36543
    },
    {
      "epoch": 3.14411081476383,
      "grad_norm": 0.859375,
      "learning_rate": 0.000303356064465259,
      "loss": 1.0103,
      "step": 36544
    },
    {
      "epoch": 3.144196851071152,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003033312030845736,
      "loss": 0.4021,
      "step": 36545
    },
    {
      "epoch": 3.144282887378474,
      "grad_norm": 1.234375,
      "learning_rate": 0.0003033063422791063,
      "loss": 0.8114,
      "step": 36546
    },
    {
      "epoch": 3.1443689236857955,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003032814820489297,
      "loss": 1.4097,
      "step": 36547
    },
    {
      "epoch": 3.144454959993117,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003032566223941165,
      "loss": 0.9907,
      "step": 36548
    },
    {
      "epoch": 3.1445409963004387,
      "grad_norm": 0.96875,
      "learning_rate": 0.00030323176331473927,
      "loss": 1.0563,
      "step": 36549
    },
    {
      "epoch": 3.1446270326077603,
      "grad_norm": 1.25,
      "learning_rate": 0.000303206904810871,
      "loss": 0.9805,
      "step": 36550
    },
    {
      "epoch": 3.144713068915082,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003031820468825843,
      "loss": 0.7006,
      "step": 36551
    },
    {
      "epoch": 3.144799105222404,
      "grad_norm": 1.453125,
      "learning_rate": 0.00030315718952995165,
      "loss": 0.8943,
      "step": 36552
    },
    {
      "epoch": 3.1448851415297256,
      "grad_norm": 1.234375,
      "learning_rate": 0.000303132332753046,
      "loss": 1.274,
      "step": 36553
    },
    {
      "epoch": 3.1449711778370473,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00030310747655193994,
      "loss": 1.8182,
      "step": 36554
    },
    {
      "epoch": 3.145057214144369,
      "grad_norm": 1.53125,
      "learning_rate": 0.00030308262092670636,
      "loss": 0.7867,
      "step": 36555
    },
    {
      "epoch": 3.1451432504516905,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0003030577658774178,
      "loss": 1.2095,
      "step": 36556
    },
    {
      "epoch": 3.145229286759012,
      "grad_norm": 6.4375,
      "learning_rate": 0.00030303291140414704,
      "loss": 1.2783,
      "step": 36557
    },
    {
      "epoch": 3.1453153230663338,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003030080575069665,
      "loss": 1.3545,
      "step": 36558
    },
    {
      "epoch": 3.145401359373656,
      "grad_norm": 0.78125,
      "learning_rate": 0.0003029832041859492,
      "loss": 0.7608,
      "step": 36559
    },
    {
      "epoch": 3.1454873956809775,
      "grad_norm": 1.359375,
      "learning_rate": 0.00030295835144116765,
      "loss": 0.8288,
      "step": 36560
    },
    {
      "epoch": 3.145573431988299,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00030293349927269466,
      "loss": 0.6326,
      "step": 36561
    },
    {
      "epoch": 3.1456594682956207,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0003029086476806029,
      "loss": 0.8351,
      "step": 36562
    },
    {
      "epoch": 3.1457455046029423,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0003028837966649649,
      "loss": 0.9382,
      "step": 36563
    },
    {
      "epoch": 3.145831540910264,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00030285894622585356,
      "loss": 1.3564,
      "step": 36564
    },
    {
      "epoch": 3.145917577217586,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0003028340963633414,
      "loss": 0.6499,
      "step": 36565
    },
    {
      "epoch": 3.1460036135249076,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00030280924707750117,
      "loss": 0.9238,
      "step": 36566
    },
    {
      "epoch": 3.1460896498322293,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003027843983684056,
      "loss": 1.1241,
      "step": 36567
    },
    {
      "epoch": 3.146175686139551,
      "grad_norm": 0.71875,
      "learning_rate": 0.00030275955023612716,
      "loss": 0.7157,
      "step": 36568
    },
    {
      "epoch": 3.1462617224468725,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00030273470268073877,
      "loss": 1.4168,
      "step": 36569
    },
    {
      "epoch": 3.146347758754194,
      "grad_norm": 1.09375,
      "learning_rate": 0.00030270985570231306,
      "loss": 0.8878,
      "step": 36570
    },
    {
      "epoch": 3.1464337950615158,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003026850093009227,
      "loss": 0.7455,
      "step": 36571
    },
    {
      "epoch": 3.146519831368838,
      "grad_norm": 1.40625,
      "learning_rate": 0.00030266016347664005,
      "loss": 1.1539,
      "step": 36572
    },
    {
      "epoch": 3.1466058676761595,
      "grad_norm": 1.6875,
      "learning_rate": 0.00030263531822953825,
      "loss": 0.9498,
      "step": 36573
    },
    {
      "epoch": 3.146691903983481,
      "grad_norm": 5.46875,
      "learning_rate": 0.00030261047355968976,
      "loss": 1.2392,
      "step": 36574
    },
    {
      "epoch": 3.1467779402908027,
      "grad_norm": 1.140625,
      "learning_rate": 0.00030258562946716716,
      "loss": 1.3792,
      "step": 36575
    },
    {
      "epoch": 3.1468639765981243,
      "grad_norm": 1.21875,
      "learning_rate": 0.00030256078595204325,
      "loss": 1.0669,
      "step": 36576
    },
    {
      "epoch": 3.146950012905446,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00030253594301439063,
      "loss": 0.9327,
      "step": 36577
    },
    {
      "epoch": 3.1470360492127676,
      "grad_norm": 2.515625,
      "learning_rate": 0.000302511100654282,
      "loss": 1.3333,
      "step": 36578
    },
    {
      "epoch": 3.1471220855200897,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003024862588717899,
      "loss": 0.978,
      "step": 36579
    },
    {
      "epoch": 3.1472081218274113,
      "grad_norm": 1.6875,
      "learning_rate": 0.0003024614176669872,
      "loss": 1.8769,
      "step": 36580
    },
    {
      "epoch": 3.147294158134733,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00030243657703994637,
      "loss": 0.6368,
      "step": 36581
    },
    {
      "epoch": 3.1473801944420545,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0003024117369907402,
      "loss": 0.6533,
      "step": 36582
    },
    {
      "epoch": 3.147466230749376,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003023868975194411,
      "loss": 0.9042,
      "step": 36583
    },
    {
      "epoch": 3.1475522670566978,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000302362058626122,
      "loss": 0.4887,
      "step": 36584
    },
    {
      "epoch": 3.14763830336402,
      "grad_norm": 11.0,
      "learning_rate": 0.00030233722031085546,
      "loss": 1.0845,
      "step": 36585
    },
    {
      "epoch": 3.1477243396713415,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003023123825737141,
      "loss": 0.5375,
      "step": 36586
    },
    {
      "epoch": 3.147810375978663,
      "grad_norm": 1.078125,
      "learning_rate": 0.00030228754541477057,
      "loss": 0.7669,
      "step": 36587
    },
    {
      "epoch": 3.1478964122859847,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00030226270883409746,
      "loss": 1.235,
      "step": 36588
    },
    {
      "epoch": 3.1479824485933063,
      "grad_norm": 1.875,
      "learning_rate": 0.00030223787283176754,
      "loss": 1.4326,
      "step": 36589
    },
    {
      "epoch": 3.148068484900628,
      "grad_norm": 0.953125,
      "learning_rate": 0.0003022130374078532,
      "loss": 0.8799,
      "step": 36590
    },
    {
      "epoch": 3.1481545212079496,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0003021882025624273,
      "loss": 0.8489,
      "step": 36591
    },
    {
      "epoch": 3.1482405575152717,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00030216336829556255,
      "loss": 1.191,
      "step": 36592
    },
    {
      "epoch": 3.1483265938225933,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003021385346073315,
      "loss": 0.9925,
      "step": 36593
    },
    {
      "epoch": 3.148412630129915,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0003021137014978068,
      "loss": 1.1295,
      "step": 36594
    },
    {
      "epoch": 3.1484986664372365,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0003020888689670609,
      "loss": 1.8448,
      "step": 36595
    },
    {
      "epoch": 3.148584702744558,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0003020640370151666,
      "loss": 1.3883,
      "step": 36596
    },
    {
      "epoch": 3.1486707390518798,
      "grad_norm": 1.1875,
      "learning_rate": 0.00030203920564219656,
      "loss": 1.4536,
      "step": 36597
    },
    {
      "epoch": 3.1487567753592014,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00030201437484822323,
      "loss": 1.2267,
      "step": 36598
    },
    {
      "epoch": 3.1488428116665235,
      "grad_norm": 1.3125,
      "learning_rate": 0.00030198954463331947,
      "loss": 1.6483,
      "step": 36599
    },
    {
      "epoch": 3.148928847973845,
      "grad_norm": 1.046875,
      "learning_rate": 0.00030196471499755774,
      "loss": 1.6493,
      "step": 36600
    },
    {
      "epoch": 3.1490148842811667,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00030193988594101075,
      "loss": 1.1394,
      "step": 36601
    },
    {
      "epoch": 3.1491009205884883,
      "grad_norm": 1.40625,
      "learning_rate": 0.0003019150574637509,
      "loss": 1.0453,
      "step": 36602
    },
    {
      "epoch": 3.14918695689581,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0003018902295658511,
      "loss": 1.1765,
      "step": 36603
    },
    {
      "epoch": 3.1492729932031316,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0003018654022473839,
      "loss": 0.4824,
      "step": 36604
    },
    {
      "epoch": 3.149359029510453,
      "grad_norm": 1.34375,
      "learning_rate": 0.0003018405755084218,
      "loss": 1.7263,
      "step": 36605
    },
    {
      "epoch": 3.1494450658177753,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00030181574934903745,
      "loss": 1.2688,
      "step": 36606
    },
    {
      "epoch": 3.149531102125097,
      "grad_norm": 1.390625,
      "learning_rate": 0.0003017909237693035,
      "loss": 1.7325,
      "step": 36607
    },
    {
      "epoch": 3.1496171384324185,
      "grad_norm": 0.953125,
      "learning_rate": 0.00030176609876929263,
      "loss": 1.1595,
      "step": 36608
    },
    {
      "epoch": 3.14970317473974,
      "grad_norm": 1.09375,
      "learning_rate": 0.00030174127434907727,
      "loss": 1.015,
      "step": 36609
    },
    {
      "epoch": 3.1497892110470618,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0003017164505087302,
      "loss": 0.9977,
      "step": 36610
    },
    {
      "epoch": 3.1498752473543834,
      "grad_norm": 1.046875,
      "learning_rate": 0.0003016916272483239,
      "loss": 0.5904,
      "step": 36611
    },
    {
      "epoch": 3.149961283661705,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00030166680456793107,
      "loss": 1.0681,
      "step": 36612
    },
    {
      "epoch": 3.150047319969027,
      "grad_norm": 1.25,
      "learning_rate": 0.00030164198246762414,
      "loss": 0.8417,
      "step": 36613
    },
    {
      "epoch": 3.1501333562763487,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00030161716094747595,
      "loss": 1.1798,
      "step": 36614
    },
    {
      "epoch": 3.1502193925836703,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000301592340007559,
      "loss": 1.2764,
      "step": 36615
    },
    {
      "epoch": 3.150305428890992,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003015675196479457,
      "loss": 1.8985,
      "step": 36616
    },
    {
      "epoch": 3.1503914651983136,
      "grad_norm": 1.484375,
      "learning_rate": 0.00030154269986870896,
      "loss": 1.0978,
      "step": 36617
    },
    {
      "epoch": 3.150477501505635,
      "grad_norm": 1.21875,
      "learning_rate": 0.00030151788066992127,
      "loss": 1.609,
      "step": 36618
    },
    {
      "epoch": 3.150563537812957,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00030149306205165506,
      "loss": 1.494,
      "step": 36619
    },
    {
      "epoch": 3.150649574120279,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000301468244013983,
      "loss": 1.0588,
      "step": 36620
    },
    {
      "epoch": 3.1507356104276005,
      "grad_norm": 1.234375,
      "learning_rate": 0.00030144342655697787,
      "loss": 1.3233,
      "step": 36621
    },
    {
      "epoch": 3.150821646734922,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000301418609680712,
      "loss": 0.5801,
      "step": 36622
    },
    {
      "epoch": 3.150907683042244,
      "grad_norm": 1.203125,
      "learning_rate": 0.0003013937933852581,
      "loss": 1.8274,
      "step": 36623
    },
    {
      "epoch": 3.1509937193495654,
      "grad_norm": 2.046875,
      "learning_rate": 0.0003013689776706887,
      "loss": 0.6403,
      "step": 36624
    },
    {
      "epoch": 3.151079755656887,
      "grad_norm": 1.609375,
      "learning_rate": 0.0003013441625370764,
      "loss": 2.2066,
      "step": 36625
    },
    {
      "epoch": 3.151165791964209,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0003013193479844939,
      "loss": 0.6916,
      "step": 36626
    },
    {
      "epoch": 3.1512518282715307,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00030129453401301335,
      "loss": 0.6668,
      "step": 36627
    },
    {
      "epoch": 3.1513378645788523,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00030126972062270794,
      "loss": 0.6495,
      "step": 36628
    },
    {
      "epoch": 3.151423900886174,
      "grad_norm": 0.84375,
      "learning_rate": 0.00030124490781364976,
      "loss": 1.0046,
      "step": 36629
    },
    {
      "epoch": 3.1515099371934956,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003012200955859118,
      "loss": 0.6576,
      "step": 36630
    },
    {
      "epoch": 3.151595973500817,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0003011952839395663,
      "loss": 1.2888,
      "step": 36631
    },
    {
      "epoch": 3.151682009808139,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00030117047287468577,
      "loss": 1.0444,
      "step": 36632
    },
    {
      "epoch": 3.151768046115461,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00030114566239134317,
      "loss": 0.5794,
      "step": 36633
    },
    {
      "epoch": 3.1518540824227825,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0003011208524896108,
      "loss": 1.1281,
      "step": 36634
    },
    {
      "epoch": 3.151940118730104,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0003010960431695612,
      "loss": 1.8027,
      "step": 36635
    },
    {
      "epoch": 3.152026155037426,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0003010712344312669,
      "loss": 1.5273,
      "step": 36636
    },
    {
      "epoch": 3.1521121913447474,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00030104642627480074,
      "loss": 1.537,
      "step": 36637
    },
    {
      "epoch": 3.152198227652069,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000301021618700235,
      "loss": 0.9563,
      "step": 36638
    },
    {
      "epoch": 3.152284263959391,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003009968117076423,
      "loss": 0.705,
      "step": 36639
    },
    {
      "epoch": 3.1523703002667127,
      "grad_norm": 1.0,
      "learning_rate": 0.0003009720052970952,
      "loss": 0.6796,
      "step": 36640
    },
    {
      "epoch": 3.1524563365740343,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00030094719946866633,
      "loss": 0.8891,
      "step": 36641
    },
    {
      "epoch": 3.152542372881356,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0003009223942224282,
      "loss": 1.6076,
      "step": 36642
    },
    {
      "epoch": 3.1526284091886776,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0003008975895584533,
      "loss": 1.13,
      "step": 36643
    },
    {
      "epoch": 3.152714445495999,
      "grad_norm": 1.28125,
      "learning_rate": 0.0003008727854768142,
      "loss": 1.9654,
      "step": 36644
    },
    {
      "epoch": 3.152800481803321,
      "grad_norm": 2.71875,
      "learning_rate": 0.0003008479819775836,
      "loss": 1.8396,
      "step": 36645
    },
    {
      "epoch": 3.152886518110643,
      "grad_norm": 1.03125,
      "learning_rate": 0.0003008231790608339,
      "loss": 0.8815,
      "step": 36646
    },
    {
      "epoch": 3.1529725544179645,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00030079837672663753,
      "loss": 1.2016,
      "step": 36647
    },
    {
      "epoch": 3.153058590725286,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00030077357497506723,
      "loss": 0.8549,
      "step": 36648
    },
    {
      "epoch": 3.153144627032608,
      "grad_norm": 1.140625,
      "learning_rate": 0.00030074877380619554,
      "loss": 0.8134,
      "step": 36649
    },
    {
      "epoch": 3.1532306633399294,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003007239732200948,
      "loss": 1.1861,
      "step": 36650
    },
    {
      "epoch": 3.153316699647251,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0003006991732168378,
      "loss": 1.5423,
      "step": 36651
    },
    {
      "epoch": 3.1534027359545727,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00030067437379649697,
      "loss": 1.1285,
      "step": 36652
    },
    {
      "epoch": 3.1534887722618947,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003006495749591448,
      "loss": 1.3685,
      "step": 36653
    },
    {
      "epoch": 3.1535748085692163,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0003006247767048538,
      "loss": 1.5686,
      "step": 36654
    },
    {
      "epoch": 3.153660844876538,
      "grad_norm": 1.125,
      "learning_rate": 0.0003005999790336966,
      "loss": 1.4222,
      "step": 36655
    },
    {
      "epoch": 3.1537468811838596,
      "grad_norm": 32.25,
      "learning_rate": 0.0003005751819457456,
      "loss": 1.3083,
      "step": 36656
    },
    {
      "epoch": 3.153832917491181,
      "grad_norm": 1.1875,
      "learning_rate": 0.00030055038544107344,
      "loss": 1.133,
      "step": 36657
    },
    {
      "epoch": 3.153918953798503,
      "grad_norm": 1.765625,
      "learning_rate": 0.0003005255895197527,
      "loss": 0.7056,
      "step": 36658
    },
    {
      "epoch": 3.1540049901058245,
      "grad_norm": 1.140625,
      "learning_rate": 0.0003005007941818558,
      "loss": 1.8574,
      "step": 36659
    },
    {
      "epoch": 3.1540910264131465,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00030047599942745526,
      "loss": 1.7443,
      "step": 36660
    },
    {
      "epoch": 3.154177062720468,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0003004512052566235,
      "loss": 1.3155,
      "step": 36661
    },
    {
      "epoch": 3.15426309902779,
      "grad_norm": 1.078125,
      "learning_rate": 0.0003004264116694333,
      "loss": 0.956,
      "step": 36662
    },
    {
      "epoch": 3.1543491353351114,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00030040161866595695,
      "loss": 0.7789,
      "step": 36663
    },
    {
      "epoch": 3.154435171642433,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0003003768262462669,
      "loss": 0.9271,
      "step": 36664
    },
    {
      "epoch": 3.1545212079497547,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000300352034410436,
      "loss": 0.8117,
      "step": 36665
    },
    {
      "epoch": 3.1546072442570763,
      "grad_norm": 1.046875,
      "learning_rate": 0.00030032724315853646,
      "loss": 1.1977,
      "step": 36666
    },
    {
      "epoch": 3.1546932805643983,
      "grad_norm": 4.875,
      "learning_rate": 0.00030030245249064105,
      "loss": 1.9946,
      "step": 36667
    },
    {
      "epoch": 3.15477931687172,
      "grad_norm": 0.921875,
      "learning_rate": 0.0003002776624068221,
      "loss": 1.0934,
      "step": 36668
    },
    {
      "epoch": 3.1548653531790416,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00030025287290715214,
      "loss": 1.4043,
      "step": 36669
    },
    {
      "epoch": 3.1549513894863632,
      "grad_norm": 1.5,
      "learning_rate": 0.0003002280839917036,
      "loss": 1.0848,
      "step": 36670
    },
    {
      "epoch": 3.155037425793685,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003002032956605492,
      "loss": 1.8289,
      "step": 36671
    },
    {
      "epoch": 3.1551234621010065,
      "grad_norm": 1.265625,
      "learning_rate": 0.0003001785079137612,
      "loss": 1.7434,
      "step": 36672
    },
    {
      "epoch": 3.155209498408328,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0003001537207514121,
      "loss": 1.4307,
      "step": 36673
    },
    {
      "epoch": 3.15529553471565,
      "grad_norm": 1.0625,
      "learning_rate": 0.0003001289341735747,
      "loss": 1.2344,
      "step": 36674
    },
    {
      "epoch": 3.155381571022972,
      "grad_norm": 1.375,
      "learning_rate": 0.0003001041481803212,
      "loss": 1.3154,
      "step": 36675
    },
    {
      "epoch": 3.1554676073302934,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0003000793627717242,
      "loss": 1.009,
      "step": 36676
    },
    {
      "epoch": 3.155553643637615,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00030005457794785614,
      "loss": 1.3831,
      "step": 36677
    },
    {
      "epoch": 3.1556396799449367,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00030002979370878955,
      "loss": 0.9735,
      "step": 36678
    },
    {
      "epoch": 3.1557257162522583,
      "grad_norm": 1.46875,
      "learning_rate": 0.000300005010054597,
      "loss": 0.5784,
      "step": 36679
    },
    {
      "epoch": 3.15581175255958,
      "grad_norm": 2240.0,
      "learning_rate": 0.0002999802269853508,
      "loss": 1.294,
      "step": 36680
    },
    {
      "epoch": 3.155897788866902,
      "grad_norm": 2.84375,
      "learning_rate": 0.00029995544450112365,
      "loss": 1.2314,
      "step": 36681
    },
    {
      "epoch": 3.1559838251742236,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002999306626019879,
      "loss": 1.4587,
      "step": 36682
    },
    {
      "epoch": 3.1560698614815452,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000299905881288016,
      "loss": 0.9737,
      "step": 36683
    },
    {
      "epoch": 3.156155897788867,
      "grad_norm": 1.828125,
      "learning_rate": 0.0002998811005592804,
      "loss": 1.6802,
      "step": 36684
    },
    {
      "epoch": 3.1562419340961885,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002998563204158538,
      "loss": 1.2987,
      "step": 36685
    },
    {
      "epoch": 3.15632797040351,
      "grad_norm": 0.875,
      "learning_rate": 0.0002998315408578085,
      "loss": 0.895,
      "step": 36686
    },
    {
      "epoch": 3.156414006710832,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002998067618852169,
      "loss": 1.1547,
      "step": 36687
    },
    {
      "epoch": 3.156500043018154,
      "grad_norm": 8.9375,
      "learning_rate": 0.0002997819834981518,
      "loss": 1.659,
      "step": 36688
    },
    {
      "epoch": 3.1565860793254754,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00029975720569668543,
      "loss": 1.8209,
      "step": 36689
    },
    {
      "epoch": 3.156672115632797,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002997324284808902,
      "loss": 1.5997,
      "step": 36690
    },
    {
      "epoch": 3.1567581519401187,
      "grad_norm": 3.984375,
      "learning_rate": 0.0002997076518508386,
      "loss": 1.4374,
      "step": 36691
    },
    {
      "epoch": 3.1568441882474403,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00029968287580660325,
      "loss": 1.0265,
      "step": 36692
    },
    {
      "epoch": 3.156930224554762,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002996581003482565,
      "loss": 1.4119,
      "step": 36693
    },
    {
      "epoch": 3.157016260862084,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002996333254758709,
      "loss": 1.4857,
      "step": 36694
    },
    {
      "epoch": 3.1571022971694056,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00029960855118951876,
      "loss": 1.0322,
      "step": 36695
    },
    {
      "epoch": 3.1571883334767272,
      "grad_norm": 2.03125,
      "learning_rate": 0.0002995837774892727,
      "loss": 1.3251,
      "step": 36696
    },
    {
      "epoch": 3.157274369784049,
      "grad_norm": 1.21875,
      "learning_rate": 0.00029955900437520514,
      "loss": 1.5306,
      "step": 36697
    },
    {
      "epoch": 3.1573604060913705,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00029953423184738836,
      "loss": 0.7929,
      "step": 36698
    },
    {
      "epoch": 3.157446442398692,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0002995094599058951,
      "loss": 1.1181,
      "step": 36699
    },
    {
      "epoch": 3.157532478706014,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002994846885507977,
      "loss": 1.2246,
      "step": 36700
    },
    {
      "epoch": 3.157618515013336,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00029945991778216837,
      "loss": 0.7379,
      "step": 36701
    },
    {
      "epoch": 3.1577045513206574,
      "grad_norm": 1.59375,
      "learning_rate": 0.00029943514760007997,
      "loss": 0.9679,
      "step": 36702
    },
    {
      "epoch": 3.157790587627979,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002994103780046046,
      "loss": 0.8655,
      "step": 36703
    },
    {
      "epoch": 3.1578766239353007,
      "grad_norm": 10.0,
      "learning_rate": 0.00029938560899581503,
      "loss": 1.016,
      "step": 36704
    },
    {
      "epoch": 3.1579626602426223,
      "grad_norm": 1.5,
      "learning_rate": 0.0002993608405737835,
      "loss": 1.2717,
      "step": 36705
    },
    {
      "epoch": 3.158048696549944,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00029933607273858255,
      "loss": 0.6851,
      "step": 36706
    },
    {
      "epoch": 3.158134732857266,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002993113054902844,
      "loss": 1.1948,
      "step": 36707
    },
    {
      "epoch": 3.1582207691645876,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002992865388289618,
      "loss": 1.9473,
      "step": 36708
    },
    {
      "epoch": 3.1583068054719092,
      "grad_norm": 1.0625,
      "learning_rate": 0.000299261772754687,
      "loss": 0.8959,
      "step": 36709
    },
    {
      "epoch": 3.158392841779231,
      "grad_norm": 1.578125,
      "learning_rate": 0.0002992370072675324,
      "loss": 1.3731,
      "step": 36710
    },
    {
      "epoch": 3.1584788780865525,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002992122423675706,
      "loss": 1.0692,
      "step": 36711
    },
    {
      "epoch": 3.158564914393874,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002991874780548739,
      "loss": 0.6553,
      "step": 36712
    },
    {
      "epoch": 3.1586509507011957,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002991627143295148,
      "loss": 0.835,
      "step": 36713
    },
    {
      "epoch": 3.158736987008518,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002991379511915656,
      "loss": 1.0403,
      "step": 36714
    },
    {
      "epoch": 3.1588230233158394,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00029911318864109894,
      "loss": 1.0592,
      "step": 36715
    },
    {
      "epoch": 3.158909059623161,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002990884266781871,
      "loss": 0.6377,
      "step": 36716
    },
    {
      "epoch": 3.1589950959304827,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002990636653029026,
      "loss": 0.464,
      "step": 36717
    },
    {
      "epoch": 3.1590811322378043,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002990389045153177,
      "loss": 1.3442,
      "step": 36718
    },
    {
      "epoch": 3.159167168545126,
      "grad_norm": 1.21875,
      "learning_rate": 0.00029901414431550487,
      "loss": 1.3265,
      "step": 36719
    },
    {
      "epoch": 3.1592532048524475,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00029898938470353676,
      "loss": 0.8941,
      "step": 36720
    },
    {
      "epoch": 3.1593392411597696,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002989646256794854,
      "loss": 1.3782,
      "step": 36721
    },
    {
      "epoch": 3.1594252774670912,
      "grad_norm": 2.046875,
      "learning_rate": 0.0002989398672434235,
      "loss": 1.3579,
      "step": 36722
    },
    {
      "epoch": 3.159511313774413,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00029891510939542346,
      "loss": 1.1008,
      "step": 36723
    },
    {
      "epoch": 3.1595973500817345,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002988903521355576,
      "loss": 0.7291,
      "step": 36724
    },
    {
      "epoch": 3.159683386389056,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00029886559546389823,
      "loss": 1.4313,
      "step": 36725
    },
    {
      "epoch": 3.1597694226963777,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000298840839380518,
      "loss": 1.7268,
      "step": 36726
    },
    {
      "epoch": 3.1598554590036994,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002988160838854891,
      "loss": 1.0791,
      "step": 36727
    },
    {
      "epoch": 3.1599414953110214,
      "grad_norm": 0.828125,
      "learning_rate": 0.000298791328978884,
      "loss": 0.6234,
      "step": 36728
    },
    {
      "epoch": 3.160027531618343,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00029876657466077524,
      "loss": 1.2,
      "step": 36729
    },
    {
      "epoch": 3.1601135679256647,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002987418209312351,
      "loss": 0.5464,
      "step": 36730
    },
    {
      "epoch": 3.1601996042329863,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00029871706779033595,
      "loss": 1.1451,
      "step": 36731
    },
    {
      "epoch": 3.160285640540308,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002986923152381502,
      "loss": 0.6881,
      "step": 36732
    },
    {
      "epoch": 3.1603716768476295,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002986675632747503,
      "loss": 1.62,
      "step": 36733
    },
    {
      "epoch": 3.160457713154951,
      "grad_norm": 1.453125,
      "learning_rate": 0.0002986428119002087,
      "loss": 1.1785,
      "step": 36734
    },
    {
      "epoch": 3.1605437494622732,
      "grad_norm": 1.015625,
      "learning_rate": 0.00029861806111459766,
      "loss": 0.714,
      "step": 36735
    },
    {
      "epoch": 3.160629785769595,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029859331091798953,
      "loss": 0.6553,
      "step": 36736
    },
    {
      "epoch": 3.1607158220769165,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002985685613104569,
      "loss": 0.7182,
      "step": 36737
    },
    {
      "epoch": 3.160801858384238,
      "grad_norm": 13.875,
      "learning_rate": 0.00029854381229207184,
      "loss": 0.9784,
      "step": 36738
    },
    {
      "epoch": 3.1608878946915597,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00029851906386290717,
      "loss": 0.6821,
      "step": 36739
    },
    {
      "epoch": 3.1609739309988814,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000298494316023035,
      "loss": 1.8368,
      "step": 36740
    },
    {
      "epoch": 3.1610599673062034,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002984695687725277,
      "loss": 1.1116,
      "step": 36741
    },
    {
      "epoch": 3.161146003613525,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002984448221114578,
      "loss": 0.7701,
      "step": 36742
    },
    {
      "epoch": 3.1612320399208467,
      "grad_norm": 3.546875,
      "learning_rate": 0.00029842007603989766,
      "loss": 1.5692,
      "step": 36743
    },
    {
      "epoch": 3.1613180762281683,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029839533055791943,
      "loss": 1.3147,
      "step": 36744
    },
    {
      "epoch": 3.16140411253549,
      "grad_norm": 1.546875,
      "learning_rate": 0.00029837058566559583,
      "loss": 1.0769,
      "step": 36745
    },
    {
      "epoch": 3.1614901488428115,
      "grad_norm": 260.0,
      "learning_rate": 0.00029834584136299896,
      "loss": 1.3333,
      "step": 36746
    },
    {
      "epoch": 3.161576185150133,
      "grad_norm": 1.125,
      "learning_rate": 0.00029832109765020134,
      "loss": 1.7916,
      "step": 36747
    },
    {
      "epoch": 3.1616622214574552,
      "grad_norm": 82.5,
      "learning_rate": 0.0002982963545272751,
      "loss": 1.2985,
      "step": 36748
    },
    {
      "epoch": 3.161748257764777,
      "grad_norm": 0.890625,
      "learning_rate": 0.000298271611994293,
      "loss": 0.9313,
      "step": 36749
    },
    {
      "epoch": 3.1618342940720985,
      "grad_norm": 4.65625,
      "learning_rate": 0.0002982468700513271,
      "loss": 1.7314,
      "step": 36750
    },
    {
      "epoch": 3.16192033037942,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002982221286984499,
      "loss": 0.6113,
      "step": 36751
    },
    {
      "epoch": 3.1620063666867417,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002981973879357337,
      "loss": 1.4248,
      "step": 36752
    },
    {
      "epoch": 3.1620924029940634,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000298172647763251,
      "loss": 0.9746,
      "step": 36753
    },
    {
      "epoch": 3.162178439301385,
      "grad_norm": 1.046875,
      "learning_rate": 0.000298147908181074,
      "loss": 1.5364,
      "step": 36754
    },
    {
      "epoch": 3.162264475608707,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00029812316918927497,
      "loss": 0.8682,
      "step": 36755
    },
    {
      "epoch": 3.1623505119160287,
      "grad_norm": 1.03125,
      "learning_rate": 0.00029809843078792655,
      "loss": 0.9932,
      "step": 36756
    },
    {
      "epoch": 3.1624365482233503,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000298073692977101,
      "loss": 0.5843,
      "step": 36757
    },
    {
      "epoch": 3.162522584530672,
      "grad_norm": 0.78125,
      "learning_rate": 0.00029804895575687054,
      "loss": 0.7606,
      "step": 36758
    },
    {
      "epoch": 3.1626086208379935,
      "grad_norm": 47.0,
      "learning_rate": 0.0002980242191273075,
      "loss": 0.7194,
      "step": 36759
    },
    {
      "epoch": 3.162694657145315,
      "grad_norm": 1.328125,
      "learning_rate": 0.00029799948308848443,
      "loss": 1.1662,
      "step": 36760
    },
    {
      "epoch": 3.1627806934526372,
      "grad_norm": 5.4375,
      "learning_rate": 0.00029797474764047373,
      "loss": 0.6147,
      "step": 36761
    },
    {
      "epoch": 3.162866729759959,
      "grad_norm": 1.609375,
      "learning_rate": 0.0002979500127833473,
      "loss": 1.2084,
      "step": 36762
    },
    {
      "epoch": 3.1629527660672805,
      "grad_norm": 1.15625,
      "learning_rate": 0.00029792527851717806,
      "loss": 0.6728,
      "step": 36763
    },
    {
      "epoch": 3.163038802374602,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000297900544842038,
      "loss": 0.8184,
      "step": 36764
    },
    {
      "epoch": 3.1631248386819237,
      "grad_norm": 65.5,
      "learning_rate": 0.00029787581175799954,
      "loss": 0.8051,
      "step": 36765
    },
    {
      "epoch": 3.1632108749892454,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002978510792651349,
      "loss": 1.529,
      "step": 36766
    },
    {
      "epoch": 3.163296911296567,
      "grad_norm": 30.75,
      "learning_rate": 0.00029782634736351664,
      "loss": 1.1803,
      "step": 36767
    },
    {
      "epoch": 3.163382947603889,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002978016160532171,
      "loss": 1.1452,
      "step": 36768
    },
    {
      "epoch": 3.1634689839112107,
      "grad_norm": 1.359375,
      "learning_rate": 0.00029777688533430823,
      "loss": 1.3671,
      "step": 36769
    },
    {
      "epoch": 3.1635550202185323,
      "grad_norm": 30.0,
      "learning_rate": 0.0002977521552068629,
      "loss": 1.327,
      "step": 36770
    },
    {
      "epoch": 3.163641056525854,
      "grad_norm": 1.125,
      "learning_rate": 0.00029772742567095315,
      "loss": 1.2524,
      "step": 36771
    },
    {
      "epoch": 3.1637270928331755,
      "grad_norm": 2.3125,
      "learning_rate": 0.0002977026967266513,
      "loss": 0.6981,
      "step": 36772
    },
    {
      "epoch": 3.163813129140497,
      "grad_norm": 84.5,
      "learning_rate": 0.0002976779683740296,
      "loss": 0.4898,
      "step": 36773
    },
    {
      "epoch": 3.163899165447819,
      "grad_norm": 0.71875,
      "learning_rate": 0.00029765324061316057,
      "loss": 0.9586,
      "step": 36774
    },
    {
      "epoch": 3.163985201755141,
      "grad_norm": 1.640625,
      "learning_rate": 0.00029762851344411636,
      "loss": 1.3,
      "step": 36775
    },
    {
      "epoch": 3.1640712380624625,
      "grad_norm": 1.140625,
      "learning_rate": 0.00029760378686696955,
      "loss": 0.674,
      "step": 36776
    },
    {
      "epoch": 3.164157274369784,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002975790608817922,
      "loss": 0.9561,
      "step": 36777
    },
    {
      "epoch": 3.1642433106771057,
      "grad_norm": 5.4375,
      "learning_rate": 0.0002975543354886567,
      "loss": 1.8994,
      "step": 36778
    },
    {
      "epoch": 3.1643293469844274,
      "grad_norm": 2.84375,
      "learning_rate": 0.00029752961068763554,
      "loss": 1.3653,
      "step": 36779
    },
    {
      "epoch": 3.164415383291749,
      "grad_norm": 162.0,
      "learning_rate": 0.0002975048864788009,
      "loss": 1.1328,
      "step": 36780
    },
    {
      "epoch": 3.1645014195990706,
      "grad_norm": 1.5078125,
      "learning_rate": 0.000297480162862225,
      "loss": 1.0687,
      "step": 36781
    },
    {
      "epoch": 3.1645874559063927,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002974554398379802,
      "loss": 1.353,
      "step": 36782
    },
    {
      "epoch": 3.1646734922137143,
      "grad_norm": 1.53125,
      "learning_rate": 0.0002974307174061389,
      "loss": 1.5859,
      "step": 36783
    },
    {
      "epoch": 3.164759528521036,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002974059955667734,
      "loss": 0.7885,
      "step": 36784
    },
    {
      "epoch": 3.1648455648283575,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002973812743199558,
      "loss": 0.6269,
      "step": 36785
    },
    {
      "epoch": 3.164931601135679,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002973565536657588,
      "loss": 0.9296,
      "step": 36786
    },
    {
      "epoch": 3.165017637443001,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002973318336042543,
      "loss": 1.8172,
      "step": 36787
    },
    {
      "epoch": 3.1651036737503224,
      "grad_norm": 5.53125,
      "learning_rate": 0.0002973071141355148,
      "loss": 0.6397,
      "step": 36788
    },
    {
      "epoch": 3.1651897100576445,
      "grad_norm": 1.0625,
      "learning_rate": 0.00029728239525961253,
      "loss": 1.0211,
      "step": 36789
    },
    {
      "epoch": 3.165275746364966,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002972576769766199,
      "loss": 1.6992,
      "step": 36790
    },
    {
      "epoch": 3.1653617826722877,
      "grad_norm": 3.890625,
      "learning_rate": 0.00029723295928660916,
      "loss": 0.9412,
      "step": 36791
    },
    {
      "epoch": 3.1654478189796094,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002972082421896525,
      "loss": 1.7896,
      "step": 36792
    },
    {
      "epoch": 3.165533855286931,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002971835256858223,
      "loss": 1.0869,
      "step": 36793
    },
    {
      "epoch": 3.1656198915942526,
      "grad_norm": 5.96875,
      "learning_rate": 0.00029715880977519086,
      "loss": 0.8497,
      "step": 36794
    },
    {
      "epoch": 3.1657059279015742,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002971340944578305,
      "loss": 0.9402,
      "step": 36795
    },
    {
      "epoch": 3.1657919642088963,
      "grad_norm": 2272.0,
      "learning_rate": 0.00029710937973381334,
      "loss": 1.3481,
      "step": 36796
    },
    {
      "epoch": 3.165878000516218,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002970846656032119,
      "loss": 1.1152,
      "step": 36797
    },
    {
      "epoch": 3.1659640368235396,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002970599520660984,
      "loss": 1.9622,
      "step": 36798
    },
    {
      "epoch": 3.166050073130861,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002970352391225449,
      "loss": 0.9753,
      "step": 36799
    },
    {
      "epoch": 3.166136109438183,
      "grad_norm": 1.0625,
      "learning_rate": 0.00029701052677262384,
      "loss": 0.6949,
      "step": 36800
    },
    {
      "epoch": 3.1662221457455044,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00029698581501640765,
      "loss": 1.2475,
      "step": 36801
    },
    {
      "epoch": 3.1663081820528265,
      "grad_norm": 1.640625,
      "learning_rate": 0.00029696110385396844,
      "loss": 1.9945,
      "step": 36802
    },
    {
      "epoch": 3.166394218360148,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00029693639328537847,
      "loss": 0.98,
      "step": 36803
    },
    {
      "epoch": 3.1664802546674697,
      "grad_norm": 1.703125,
      "learning_rate": 0.00029691168331071006,
      "loss": 1.5427,
      "step": 36804
    },
    {
      "epoch": 3.1665662909747914,
      "grad_norm": 21.625,
      "learning_rate": 0.00029688697393003547,
      "loss": 1.3034,
      "step": 36805
    },
    {
      "epoch": 3.166652327282113,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002968622651434271,
      "loss": 1.3195,
      "step": 36806
    },
    {
      "epoch": 3.1667383635894346,
      "grad_norm": 118.5,
      "learning_rate": 0.0002968375569509569,
      "loss": 1.4871,
      "step": 36807
    },
    {
      "epoch": 3.1668243998967562,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00029681284935269747,
      "loss": 1.3149,
      "step": 36808
    },
    {
      "epoch": 3.1669104362040783,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00029678814234872095,
      "loss": 1.0283,
      "step": 36809
    },
    {
      "epoch": 3.1669964725114,
      "grad_norm": 84.0,
      "learning_rate": 0.00029676343593909945,
      "loss": 0.9758,
      "step": 36810
    },
    {
      "epoch": 3.1670825088187216,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00029673873012390546,
      "loss": 1.2249,
      "step": 36811
    },
    {
      "epoch": 3.167168545126043,
      "grad_norm": 1.484375,
      "learning_rate": 0.00029671402490321096,
      "loss": 0.9121,
      "step": 36812
    },
    {
      "epoch": 3.167254581433365,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00029668932027708864,
      "loss": 1.3185,
      "step": 36813
    },
    {
      "epoch": 3.1673406177406864,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00029666461624561045,
      "loss": 0.7486,
      "step": 36814
    },
    {
      "epoch": 3.1674266540480085,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002966399128088487,
      "loss": 0.9933,
      "step": 36815
    },
    {
      "epoch": 3.16751269035533,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002966152099668757,
      "loss": 1.1221,
      "step": 36816
    },
    {
      "epoch": 3.1675987266626517,
      "grad_norm": 1.25,
      "learning_rate": 0.0002965905077197637,
      "loss": 1.6446,
      "step": 36817
    },
    {
      "epoch": 3.1676847629699734,
      "grad_norm": 1.21875,
      "learning_rate": 0.00029656580606758485,
      "loss": 0.9036,
      "step": 36818
    },
    {
      "epoch": 3.167770799277295,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002965411050104114,
      "loss": 0.5684,
      "step": 36819
    },
    {
      "epoch": 3.1678568355846166,
      "grad_norm": 1.0,
      "learning_rate": 0.00029651640454831567,
      "loss": 0.6393,
      "step": 36820
    },
    {
      "epoch": 3.1679428718919382,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002964917046813699,
      "loss": 0.9505,
      "step": 36821
    },
    {
      "epoch": 3.1680289081992603,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002964670054096464,
      "loss": 0.6749,
      "step": 36822
    },
    {
      "epoch": 3.168114944506582,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002964423067332172,
      "loss": 1.106,
      "step": 36823
    },
    {
      "epoch": 3.1682009808139036,
      "grad_norm": 62.5,
      "learning_rate": 0.00029641760865215466,
      "loss": 0.7305,
      "step": 36824
    },
    {
      "epoch": 3.168287017121225,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002963929111665311,
      "loss": 0.6219,
      "step": 36825
    },
    {
      "epoch": 3.168373053428547,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002963682142764186,
      "loss": 0.7799,
      "step": 36826
    },
    {
      "epoch": 3.1684590897358684,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0002963435179818895,
      "loss": 1.1846,
      "step": 36827
    },
    {
      "epoch": 3.16854512604319,
      "grad_norm": 0.96875,
      "learning_rate": 0.00029631882228301604,
      "loss": 0.4989,
      "step": 36828
    },
    {
      "epoch": 3.168631162350512,
      "grad_norm": 1.078125,
      "learning_rate": 0.00029629412717987043,
      "loss": 1.4787,
      "step": 36829
    },
    {
      "epoch": 3.1687171986578337,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002962694326725247,
      "loss": 1.3068,
      "step": 36830
    },
    {
      "epoch": 3.1688032349651554,
      "grad_norm": 69.5,
      "learning_rate": 0.00029624473876105144,
      "loss": 1.6758,
      "step": 36831
    },
    {
      "epoch": 3.168889271272477,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00029622004544552265,
      "loss": 1.353,
      "step": 36832
    },
    {
      "epoch": 3.1689753075797986,
      "grad_norm": 1.078125,
      "learning_rate": 0.00029619535272601053,
      "loss": 1.0923,
      "step": 36833
    },
    {
      "epoch": 3.1690613438871202,
      "grad_norm": 54.25,
      "learning_rate": 0.00029617066060258745,
      "loss": 1.3328,
      "step": 36834
    },
    {
      "epoch": 3.169147380194442,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00029614596907532556,
      "loss": 0.7387,
      "step": 36835
    },
    {
      "epoch": 3.169233416501764,
      "grad_norm": 1.03125,
      "learning_rate": 0.00029612127814429703,
      "loss": 0.9762,
      "step": 36836
    },
    {
      "epoch": 3.1693194528090856,
      "grad_norm": 1.484375,
      "learning_rate": 0.000296096587809574,
      "loss": 1.0321,
      "step": 36837
    },
    {
      "epoch": 3.169405489116407,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002960718980712289,
      "loss": 1.6414,
      "step": 36838
    },
    {
      "epoch": 3.169491525423729,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002960472089293339,
      "loss": 1.483,
      "step": 36839
    },
    {
      "epoch": 3.1695775617310504,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002960225203839609,
      "loss": 1.1945,
      "step": 36840
    },
    {
      "epoch": 3.169663598038372,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002959978324351826,
      "loss": 1.1281,
      "step": 36841
    },
    {
      "epoch": 3.1697496343456937,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00029597314508307093,
      "loss": 1.4877,
      "step": 36842
    },
    {
      "epoch": 3.1698356706530157,
      "grad_norm": 1.03125,
      "learning_rate": 0.00029594845832769815,
      "loss": 1.2327,
      "step": 36843
    },
    {
      "epoch": 3.1699217069603374,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002959237721691362,
      "loss": 1.5633,
      "step": 36844
    },
    {
      "epoch": 3.170007743267659,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00029589908660745777,
      "loss": 0.555,
      "step": 36845
    },
    {
      "epoch": 3.1700937795749806,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002958744016427348,
      "loss": 0.5327,
      "step": 36846
    },
    {
      "epoch": 3.1701798158823022,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002958497172750395,
      "loss": 0.8598,
      "step": 36847
    },
    {
      "epoch": 3.170265852189624,
      "grad_norm": 0.984375,
      "learning_rate": 0.000295825033504444,
      "loss": 0.7102,
      "step": 36848
    },
    {
      "epoch": 3.1703518884969455,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00029580035033102036,
      "loss": 1.4602,
      "step": 36849
    },
    {
      "epoch": 3.1704379248042676,
      "grad_norm": 2.40625,
      "learning_rate": 0.0002957756677548413,
      "loss": 1.8956,
      "step": 36850
    },
    {
      "epoch": 3.170523961111589,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002957509857759786,
      "loss": 0.7836,
      "step": 36851
    },
    {
      "epoch": 3.170609997418911,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002957263043945046,
      "loss": 0.6278,
      "step": 36852
    },
    {
      "epoch": 3.1706960337262324,
      "grad_norm": 696.0,
      "learning_rate": 0.00029570162361049136,
      "loss": 1.4055,
      "step": 36853
    },
    {
      "epoch": 3.170782070033554,
      "grad_norm": 61.75,
      "learning_rate": 0.00029567694342401115,
      "loss": 1.1963,
      "step": 36854
    },
    {
      "epoch": 3.1708681063408757,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00029565226383513623,
      "loss": 1.001,
      "step": 36855
    },
    {
      "epoch": 3.1709541426481973,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002956275848439385,
      "loss": 1.5207,
      "step": 36856
    },
    {
      "epoch": 3.1710401789555194,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00029560290645049047,
      "loss": 1.7273,
      "step": 36857
    },
    {
      "epoch": 3.171126215262841,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00029557822865486415,
      "loss": 1.3183,
      "step": 36858
    },
    {
      "epoch": 3.1712122515701626,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00029555355145713185,
      "loss": 1.3702,
      "step": 36859
    },
    {
      "epoch": 3.1712982878774842,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002955288748573655,
      "loss": 0.759,
      "step": 36860
    },
    {
      "epoch": 3.171384324184806,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002955041988556375,
      "loss": 0.9715,
      "step": 36861
    },
    {
      "epoch": 3.1714703604921275,
      "grad_norm": 0.953125,
      "learning_rate": 0.00029547952345202,
      "loss": 1.0369,
      "step": 36862
    },
    {
      "epoch": 3.1715563967994496,
      "grad_norm": 6.09375,
      "learning_rate": 0.000295454848646585,
      "loss": 0.6077,
      "step": 36863
    },
    {
      "epoch": 3.171642433106771,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002954301744394049,
      "loss": 1.5591,
      "step": 36864
    },
    {
      "epoch": 3.171728469414093,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00029540550083055167,
      "loss": 0.7584,
      "step": 36865
    },
    {
      "epoch": 3.1718145057214144,
      "grad_norm": 1.375,
      "learning_rate": 0.00029538082782009766,
      "loss": 1.2597,
      "step": 36866
    },
    {
      "epoch": 3.171900542028736,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002953561554081148,
      "loss": 1.8379,
      "step": 36867
    },
    {
      "epoch": 3.1719865783360577,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029533148359467545,
      "loss": 0.6627,
      "step": 36868
    },
    {
      "epoch": 3.1720726146433793,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0002953068123798517,
      "loss": 0.9061,
      "step": 36869
    },
    {
      "epoch": 3.1721586509507014,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00029528214176371574,
      "loss": 1.187,
      "step": 36870
    },
    {
      "epoch": 3.172244687258023,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002952574717463396,
      "loss": 1.4208,
      "step": 36871
    },
    {
      "epoch": 3.1723307235653446,
      "grad_norm": 1.125,
      "learning_rate": 0.00029523280232779565,
      "loss": 1.634,
      "step": 36872
    },
    {
      "epoch": 3.1724167598726662,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002952081335081559,
      "loss": 0.8901,
      "step": 36873
    },
    {
      "epoch": 3.172502796179988,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002951834652874925,
      "loss": 0.9891,
      "step": 36874
    },
    {
      "epoch": 3.1725888324873095,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002951587976658776,
      "loss": 1.643,
      "step": 36875
    },
    {
      "epoch": 3.1726748687946316,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00029513413064338344,
      "loss": 1.2791,
      "step": 36876
    },
    {
      "epoch": 3.172760905101953,
      "grad_norm": 21.125,
      "learning_rate": 0.00029510946422008213,
      "loss": 1.1024,
      "step": 36877
    },
    {
      "epoch": 3.172846941409275,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002950847983960457,
      "loss": 0.7719,
      "step": 36878
    },
    {
      "epoch": 3.1729329777165964,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00029506013317134644,
      "loss": 0.9549,
      "step": 36879
    },
    {
      "epoch": 3.173019014023918,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00029503546854605647,
      "loss": 1.4044,
      "step": 36880
    },
    {
      "epoch": 3.1731050503312397,
      "grad_norm": 3.46875,
      "learning_rate": 0.00029501080452024774,
      "loss": 0.9691,
      "step": 36881
    },
    {
      "epoch": 3.1731910866385613,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002949861410939927,
      "loss": 0.9484,
      "step": 36882
    },
    {
      "epoch": 3.1732771229458834,
      "grad_norm": 1.265625,
      "learning_rate": 0.00029496147826736326,
      "loss": 1.1447,
      "step": 36883
    },
    {
      "epoch": 3.173363159253205,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00029493681604043166,
      "loss": 0.9013,
      "step": 36884
    },
    {
      "epoch": 3.1734491955605266,
      "grad_norm": 1.5,
      "learning_rate": 0.00029491215441326986,
      "loss": 1.0809,
      "step": 36885
    },
    {
      "epoch": 3.1735352318678482,
      "grad_norm": 20.25,
      "learning_rate": 0.00029488749338595,
      "loss": 1.1024,
      "step": 36886
    },
    {
      "epoch": 3.17362126817517,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002948628329585446,
      "loss": 0.8571,
      "step": 36887
    },
    {
      "epoch": 3.1737073044824915,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002948381731311255,
      "loss": 0.9362,
      "step": 36888
    },
    {
      "epoch": 3.173793340789813,
      "grad_norm": 0.984375,
      "learning_rate": 0.00029481351390376476,
      "loss": 1.1271,
      "step": 36889
    },
    {
      "epoch": 3.173879377097135,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002947888552765346,
      "loss": 1.2086,
      "step": 36890
    },
    {
      "epoch": 3.173965413404457,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002947641972495072,
      "loss": 1.0054,
      "step": 36891
    },
    {
      "epoch": 3.1740514497117784,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002947395398227546,
      "loss": 1.2521,
      "step": 36892
    },
    {
      "epoch": 3.1741374860191,
      "grad_norm": 1.265625,
      "learning_rate": 0.00029471488299634885,
      "loss": 1.6863,
      "step": 36893
    },
    {
      "epoch": 3.1742235223264217,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002946902267703621,
      "loss": 0.7687,
      "step": 36894
    },
    {
      "epoch": 3.1743095586337433,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002946655711448666,
      "loss": 0.7977,
      "step": 36895
    },
    {
      "epoch": 3.174395594941065,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002946409161199344,
      "loss": 1.1278,
      "step": 36896
    },
    {
      "epoch": 3.174481631248387,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002946162616956375,
      "loss": 0.9549,
      "step": 36897
    },
    {
      "epoch": 3.1745676675557086,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0002945916078720481,
      "loss": 1.4649,
      "step": 36898
    },
    {
      "epoch": 3.1746537038630303,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002945669546492383,
      "loss": 0.6979,
      "step": 36899
    },
    {
      "epoch": 3.174739740170352,
      "grad_norm": 648.0,
      "learning_rate": 0.00029454230202728026,
      "loss": 1.06,
      "step": 36900
    },
    {
      "epoch": 3.1748257764776735,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000294517650006246,
      "loss": 1.5008,
      "step": 36901
    },
    {
      "epoch": 3.174911812784995,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002944929985862076,
      "loss": 1.0836,
      "step": 36902
    },
    {
      "epoch": 3.1749978490923167,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002944683477672373,
      "loss": 0.546,
      "step": 36903
    },
    {
      "epoch": 3.175083885399639,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000294443697549407,
      "loss": 0.6472,
      "step": 36904
    },
    {
      "epoch": 3.1751699217069604,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00029441904793278895,
      "loss": 0.9189,
      "step": 36905
    },
    {
      "epoch": 3.175255958014282,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00029439439891745524,
      "loss": 1.1758,
      "step": 36906
    },
    {
      "epoch": 3.1753419943216037,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00029436975050347794,
      "loss": 1.2611,
      "step": 36907
    },
    {
      "epoch": 3.1754280306289253,
      "grad_norm": 1.28125,
      "learning_rate": 0.000294345102690929,
      "loss": 1.3781,
      "step": 36908
    },
    {
      "epoch": 3.175514066936247,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00029432045547988074,
      "loss": 1.0211,
      "step": 36909
    },
    {
      "epoch": 3.1756001032435686,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002942958088704052,
      "loss": 0.743,
      "step": 36910
    },
    {
      "epoch": 3.1756861395508906,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002942711628625743,
      "loss": 1.8159,
      "step": 36911
    },
    {
      "epoch": 3.1757721758582123,
      "grad_norm": 1.609375,
      "learning_rate": 0.00029424651745646024,
      "loss": 1.4242,
      "step": 36912
    },
    {
      "epoch": 3.175858212165534,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002942218726521352,
      "loss": 0.9306,
      "step": 36913
    },
    {
      "epoch": 3.1759442484728555,
      "grad_norm": 6.21875,
      "learning_rate": 0.00029419722844967113,
      "loss": 1.096,
      "step": 36914
    },
    {
      "epoch": 3.176030284780177,
      "grad_norm": 1.203125,
      "learning_rate": 0.00029417258484914,
      "loss": 1.5492,
      "step": 36915
    },
    {
      "epoch": 3.1761163210874988,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00029414794185061413,
      "loss": 0.7721,
      "step": 36916
    },
    {
      "epoch": 3.176202357394821,
      "grad_norm": 0.875,
      "learning_rate": 0.00029412329945416554,
      "loss": 0.9051,
      "step": 36917
    },
    {
      "epoch": 3.1762883937021424,
      "grad_norm": 1.1875,
      "learning_rate": 0.00029409865765986625,
      "loss": 1.079,
      "step": 36918
    },
    {
      "epoch": 3.176374430009464,
      "grad_norm": 2.125,
      "learning_rate": 0.0002940740164677882,
      "loss": 1.2377,
      "step": 36919
    },
    {
      "epoch": 3.1764604663167857,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00029404937587800375,
      "loss": 1.2343,
      "step": 36920
    },
    {
      "epoch": 3.1765465026241073,
      "grad_norm": 0.90625,
      "learning_rate": 0.00029402473589058484,
      "loss": 0.5947,
      "step": 36921
    },
    {
      "epoch": 3.176632538931429,
      "grad_norm": 2.6875,
      "learning_rate": 0.0002940000965056034,
      "loss": 0.9417,
      "step": 36922
    },
    {
      "epoch": 3.1767185752387506,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00029397545772313156,
      "loss": 1.3654,
      "step": 36923
    },
    {
      "epoch": 3.1768046115460726,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002939508195432415,
      "loss": 1.3306,
      "step": 36924
    },
    {
      "epoch": 3.1768906478533943,
      "grad_norm": 7.84375,
      "learning_rate": 0.00029392618196600527,
      "loss": 1.0836,
      "step": 36925
    },
    {
      "epoch": 3.176976684160716,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002939015449914949,
      "loss": 1.3497,
      "step": 36926
    },
    {
      "epoch": 3.1770627204680375,
      "grad_norm": 1.65625,
      "learning_rate": 0.0002938769086197823,
      "loss": 0.8914,
      "step": 36927
    },
    {
      "epoch": 3.177148756775359,
      "grad_norm": 4.65625,
      "learning_rate": 0.00029385227285093986,
      "loss": 1.2631,
      "step": 36928
    },
    {
      "epoch": 3.1772347930826808,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002938276376850393,
      "loss": 0.8083,
      "step": 36929
    },
    {
      "epoch": 3.1773208293900024,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00029380300312215285,
      "loss": 0.8414,
      "step": 36930
    },
    {
      "epoch": 3.1774068656973244,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00029377836916235245,
      "loss": 0.9399,
      "step": 36931
    },
    {
      "epoch": 3.177492902004646,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00029375373580571026,
      "loss": 1.0852,
      "step": 36932
    },
    {
      "epoch": 3.1775789383119677,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00029372910305229825,
      "loss": 1.0919,
      "step": 36933
    },
    {
      "epoch": 3.1776649746192893,
      "grad_norm": 9.6875,
      "learning_rate": 0.00029370447090218855,
      "loss": 0.9077,
      "step": 36934
    },
    {
      "epoch": 3.177751010926611,
      "grad_norm": 1.28125,
      "learning_rate": 0.00029367983935545307,
      "loss": 1.1284,
      "step": 36935
    },
    {
      "epoch": 3.1778370472339326,
      "grad_norm": 1.28125,
      "learning_rate": 0.00029365520841216397,
      "loss": 1.7392,
      "step": 36936
    },
    {
      "epoch": 3.1779230835412546,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00029363057807239336,
      "loss": 0.9476,
      "step": 36937
    },
    {
      "epoch": 3.1780091198485763,
      "grad_norm": 1.140625,
      "learning_rate": 0.000293605948336213,
      "loss": 2.0793,
      "step": 36938
    },
    {
      "epoch": 3.178095156155898,
      "grad_norm": 1.3125,
      "learning_rate": 0.00029358131920369517,
      "loss": 1.1284,
      "step": 36939
    },
    {
      "epoch": 3.1781811924632195,
      "grad_norm": 1.453125,
      "learning_rate": 0.00029355669067491184,
      "loss": 0.9364,
      "step": 36940
    },
    {
      "epoch": 3.178267228770541,
      "grad_norm": 11264.0,
      "learning_rate": 0.0002935320627499351,
      "loss": 1.5553,
      "step": 36941
    },
    {
      "epoch": 3.1783532650778628,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002935074354288368,
      "loss": 0.5677,
      "step": 36942
    },
    {
      "epoch": 3.1784393013851844,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002934828087116892,
      "loss": 1.5941,
      "step": 36943
    },
    {
      "epoch": 3.1785253376925064,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00029345818259856417,
      "loss": 0.5977,
      "step": 36944
    },
    {
      "epoch": 3.178611373999828,
      "grad_norm": 1.5,
      "learning_rate": 0.0002934335570895337,
      "loss": 2.0501,
      "step": 36945
    },
    {
      "epoch": 3.1786974103071497,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00029340893218467,
      "loss": 1.181,
      "step": 36946
    },
    {
      "epoch": 3.1787834466144713,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000293384307884045,
      "loss": 1.0042,
      "step": 36947
    },
    {
      "epoch": 3.178869482921793,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00029335968418773076,
      "loss": 1.2302,
      "step": 36948
    },
    {
      "epoch": 3.1789555192291146,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002933350610957991,
      "loss": 1.1774,
      "step": 36949
    },
    {
      "epoch": 3.179041555536436,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002933104386083223,
      "loss": 1.1661,
      "step": 36950
    },
    {
      "epoch": 3.1791275918437583,
      "grad_norm": 1.140625,
      "learning_rate": 0.00029328581672537224,
      "loss": 0.6479,
      "step": 36951
    },
    {
      "epoch": 3.17921362815108,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002932611954470209,
      "loss": 0.8913,
      "step": 36952
    },
    {
      "epoch": 3.1792996644584015,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00029323657477334046,
      "loss": 1.3569,
      "step": 36953
    },
    {
      "epoch": 3.179385700765723,
      "grad_norm": 1.109375,
      "learning_rate": 0.00029321195470440286,
      "loss": 0.7595,
      "step": 36954
    },
    {
      "epoch": 3.1794717370730448,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00029318733524028,
      "loss": 1.2545,
      "step": 36955
    },
    {
      "epoch": 3.1795577733803664,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002931627163810439,
      "loss": 1.0435,
      "step": 36956
    },
    {
      "epoch": 3.179643809687688,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002931380981267667,
      "loss": 0.8487,
      "step": 36957
    },
    {
      "epoch": 3.17972984599501,
      "grad_norm": 2.96875,
      "learning_rate": 0.0002931134804775204,
      "loss": 0.8643,
      "step": 36958
    },
    {
      "epoch": 3.1798158823023317,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002930888634333769,
      "loss": 0.8903,
      "step": 36959
    },
    {
      "epoch": 3.1799019186096533,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00029306424699440805,
      "loss": 0.9128,
      "step": 36960
    },
    {
      "epoch": 3.179987954916975,
      "grad_norm": 1.96875,
      "learning_rate": 0.00029303963116068604,
      "loss": 0.4485,
      "step": 36961
    },
    {
      "epoch": 3.1800739912242966,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00029301501593228304,
      "loss": 1.3988,
      "step": 36962
    },
    {
      "epoch": 3.180160027531618,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002929904013092708,
      "loss": 0.9601,
      "step": 36963
    },
    {
      "epoch": 3.18024606383894,
      "grad_norm": 1.328125,
      "learning_rate": 0.0002929657872917214,
      "loss": 1.5336,
      "step": 36964
    },
    {
      "epoch": 3.180332100146262,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00029294117387970676,
      "loss": 0.7533,
      "step": 36965
    },
    {
      "epoch": 3.1804181364535835,
      "grad_norm": 1.65625,
      "learning_rate": 0.00029291656107329893,
      "loss": 1.4346,
      "step": 36966
    },
    {
      "epoch": 3.180504172760905,
      "grad_norm": 0.96875,
      "learning_rate": 0.00029289194887256997,
      "loss": 0.9787,
      "step": 36967
    },
    {
      "epoch": 3.1805902090682268,
      "grad_norm": 1.46875,
      "learning_rate": 0.00029286733727759164,
      "loss": 1.2597,
      "step": 36968
    },
    {
      "epoch": 3.1806762453755484,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002928427262884361,
      "loss": 1.2702,
      "step": 36969
    },
    {
      "epoch": 3.18076228168287,
      "grad_norm": 0.9375,
      "learning_rate": 0.00029281811590517535,
      "loss": 1.0029,
      "step": 36970
    },
    {
      "epoch": 3.1808483179901916,
      "grad_norm": 23936.0,
      "learning_rate": 0.00029279350612788137,
      "loss": 0.8506,
      "step": 36971
    },
    {
      "epoch": 3.1809343542975137,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00029276889695662594,
      "loss": 1.8842,
      "step": 36972
    },
    {
      "epoch": 3.1810203906048353,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00029274428839148127,
      "loss": 0.8826,
      "step": 36973
    },
    {
      "epoch": 3.181106426912157,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002927196804325193,
      "loss": 1.7593,
      "step": 36974
    },
    {
      "epoch": 3.1811924632194786,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00029269507307981194,
      "loss": 0.6629,
      "step": 36975
    },
    {
      "epoch": 3.1812784995268,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000292670466333431,
      "loss": 0.9356,
      "step": 36976
    },
    {
      "epoch": 3.181364535834122,
      "grad_norm": 1.296875,
      "learning_rate": 0.00029264586019344877,
      "loss": 0.7534,
      "step": 36977
    },
    {
      "epoch": 3.181450572141444,
      "grad_norm": 1.125,
      "learning_rate": 0.00029262125465993703,
      "loss": 1.4731,
      "step": 36978
    },
    {
      "epoch": 3.1815366084487655,
      "grad_norm": 3.359375,
      "learning_rate": 0.00029259664973296773,
      "loss": 1.1125,
      "step": 36979
    },
    {
      "epoch": 3.181622644756087,
      "grad_norm": 1.0625,
      "learning_rate": 0.00029257204541261296,
      "loss": 1.0884,
      "step": 36980
    },
    {
      "epoch": 3.1817086810634088,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00029254744169894466,
      "loss": 0.954,
      "step": 36981
    },
    {
      "epoch": 3.1817947173707304,
      "grad_norm": 4.84375,
      "learning_rate": 0.0002925228385920347,
      "loss": 1.0183,
      "step": 36982
    },
    {
      "epoch": 3.181880753678052,
      "grad_norm": 1.15625,
      "learning_rate": 0.00029249823609195504,
      "loss": 0.7103,
      "step": 36983
    },
    {
      "epoch": 3.1819667899853736,
      "grad_norm": 2.359375,
      "learning_rate": 0.0002924736341987777,
      "loss": 1.2843,
      "step": 36984
    },
    {
      "epoch": 3.1820528262926957,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00029244903291257476,
      "loss": 0.5683,
      "step": 36985
    },
    {
      "epoch": 3.1821388626000173,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002924244322334178,
      "loss": 0.5214,
      "step": 36986
    },
    {
      "epoch": 3.182224898907339,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00029239983216137913,
      "loss": 0.4552,
      "step": 36987
    },
    {
      "epoch": 3.1823109352146606,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002923752326965306,
      "loss": 1.5366,
      "step": 36988
    },
    {
      "epoch": 3.182396971521982,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002923506338389441,
      "loss": 0.7137,
      "step": 36989
    },
    {
      "epoch": 3.182483007829304,
      "grad_norm": 1.875,
      "learning_rate": 0.00029232603558869153,
      "loss": 1.1227,
      "step": 36990
    },
    {
      "epoch": 3.182569044136626,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002923014379458451,
      "loss": 0.8681,
      "step": 36991
    },
    {
      "epoch": 3.1826550804439475,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00029227684091047646,
      "loss": 0.9033,
      "step": 36992
    },
    {
      "epoch": 3.182741116751269,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002922522444826576,
      "loss": 1.4373,
      "step": 36993
    },
    {
      "epoch": 3.1828271530585908,
      "grad_norm": 32.5,
      "learning_rate": 0.00029222764866246054,
      "loss": 1.0101,
      "step": 36994
    },
    {
      "epoch": 3.1829131893659124,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00029220305344995734,
      "loss": 1.5805,
      "step": 36995
    },
    {
      "epoch": 3.182999225673234,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002921784588452197,
      "loss": 1.5897,
      "step": 36996
    },
    {
      "epoch": 3.1830852619805556,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00029215386484831944,
      "loss": 0.9213,
      "step": 36997
    },
    {
      "epoch": 3.1831712982878777,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000292129271459329,
      "loss": 0.9568,
      "step": 36998
    },
    {
      "epoch": 3.1832573345951993,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002921046786783198,
      "loss": 1.0864,
      "step": 36999
    },
    {
      "epoch": 3.183343370902521,
      "grad_norm": 0.890625,
      "learning_rate": 0.00029208008650536413,
      "loss": 0.7214,
      "step": 37000
    },
    {
      "epoch": 3.1834294072098426,
      "grad_norm": 12.9375,
      "learning_rate": 0.00029205549494053376,
      "loss": 0.9268,
      "step": 37001
    },
    {
      "epoch": 3.183515443517164,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002920309039839005,
      "loss": 1.6007,
      "step": 37002
    },
    {
      "epoch": 3.183601479824486,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002920063136355365,
      "loss": 1.9253,
      "step": 37003
    },
    {
      "epoch": 3.1836875161318074,
      "grad_norm": 2.359375,
      "learning_rate": 0.0002919817238955137,
      "loss": 0.9271,
      "step": 37004
    },
    {
      "epoch": 3.1837735524391295,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002919571347639038,
      "loss": 0.7351,
      "step": 37005
    },
    {
      "epoch": 3.183859588746451,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002919325462407787,
      "loss": 0.8186,
      "step": 37006
    },
    {
      "epoch": 3.1839456250537728,
      "grad_norm": 1.4375,
      "learning_rate": 0.0002919079583262105,
      "loss": 1.8132,
      "step": 37007
    },
    {
      "epoch": 3.1840316613610944,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002918833710202712,
      "loss": 1.1072,
      "step": 37008
    },
    {
      "epoch": 3.184117697668416,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002918587843230324,
      "loss": 0.7998,
      "step": 37009
    },
    {
      "epoch": 3.1842037339757376,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002918341982345663,
      "loss": 1.1709,
      "step": 37010
    },
    {
      "epoch": 3.1842897702830593,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002918096127549446,
      "loss": 0.603,
      "step": 37011
    },
    {
      "epoch": 3.1843758065903813,
      "grad_norm": 1.390625,
      "learning_rate": 0.00029178502788423916,
      "loss": 1.1607,
      "step": 37012
    },
    {
      "epoch": 3.184461842897703,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002917604436225223,
      "loss": 1.8829,
      "step": 37013
    },
    {
      "epoch": 3.1845478792050246,
      "grad_norm": 51.25,
      "learning_rate": 0.0002917358599698654,
      "loss": 0.91,
      "step": 37014
    },
    {
      "epoch": 3.184633915512346,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002917112769263407,
      "loss": 1.3436,
      "step": 37015
    },
    {
      "epoch": 3.184719951819668,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00029168669449202014,
      "loss": 1.3494,
      "step": 37016
    },
    {
      "epoch": 3.1848059881269895,
      "grad_norm": 5.21875,
      "learning_rate": 0.00029166211266697524,
      "loss": 1.5492,
      "step": 37017
    },
    {
      "epoch": 3.184892024434311,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00029163753145127814,
      "loss": 0.5717,
      "step": 37018
    },
    {
      "epoch": 3.184978060741633,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000291612950845001,
      "loss": 1.1525,
      "step": 37019
    },
    {
      "epoch": 3.1850640970489548,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002915883708482152,
      "loss": 0.9451,
      "step": 37020
    },
    {
      "epoch": 3.1851501333562764,
      "grad_norm": 4.40625,
      "learning_rate": 0.00029156379146099315,
      "loss": 0.6907,
      "step": 37021
    },
    {
      "epoch": 3.185236169663598,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002915392126834062,
      "loss": 1.4944,
      "step": 37022
    },
    {
      "epoch": 3.1853222059709196,
      "grad_norm": 1.0,
      "learning_rate": 0.0002915146345155266,
      "loss": 0.7328,
      "step": 37023
    },
    {
      "epoch": 3.1854082422782413,
      "grad_norm": 2.171875,
      "learning_rate": 0.00029149005695742623,
      "loss": 0.7613,
      "step": 37024
    },
    {
      "epoch": 3.185494278585563,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00029146548000917675,
      "loss": 1.5259,
      "step": 37025
    },
    {
      "epoch": 3.185580314892885,
      "grad_norm": 10.4375,
      "learning_rate": 0.0002914409036708502,
      "loss": 1.2141,
      "step": 37026
    },
    {
      "epoch": 3.1856663512002066,
      "grad_norm": 8.0,
      "learning_rate": 0.0002914163279425186,
      "loss": 1.3085,
      "step": 37027
    },
    {
      "epoch": 3.185752387507528,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002913917528242535,
      "loss": 2.0851,
      "step": 37028
    },
    {
      "epoch": 3.18583842381485,
      "grad_norm": 1.171875,
      "learning_rate": 0.00029136717831612695,
      "loss": 1.8149,
      "step": 37029
    },
    {
      "epoch": 3.1859244601221715,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00029134260441821105,
      "loss": 1.2785,
      "step": 37030
    },
    {
      "epoch": 3.186010496429493,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002913180311305772,
      "loss": 1.2969,
      "step": 37031
    },
    {
      "epoch": 3.1860965327368147,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002912934584532978,
      "loss": 0.831,
      "step": 37032
    },
    {
      "epoch": 3.1861825690441368,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002912688863864442,
      "loss": 1.3577,
      "step": 37033
    },
    {
      "epoch": 3.1862686053514584,
      "grad_norm": 12.5,
      "learning_rate": 0.0002912443149300883,
      "loss": 0.8843,
      "step": 37034
    },
    {
      "epoch": 3.18635464165878,
      "grad_norm": 1.078125,
      "learning_rate": 0.00029121974408430253,
      "loss": 0.5848,
      "step": 37035
    },
    {
      "epoch": 3.1864406779661016,
      "grad_norm": 7.5625,
      "learning_rate": 0.0002911951738491583,
      "loss": 1.2221,
      "step": 37036
    },
    {
      "epoch": 3.1865267142734233,
      "grad_norm": 17.625,
      "learning_rate": 0.0002911706042247277,
      "loss": 0.6003,
      "step": 37037
    },
    {
      "epoch": 3.186612750580745,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0002911460352110823,
      "loss": 0.5833,
      "step": 37038
    },
    {
      "epoch": 3.186698786888067,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002911214668082941,
      "loss": 0.8866,
      "step": 37039
    },
    {
      "epoch": 3.1867848231953886,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002910968990164352,
      "loss": 0.849,
      "step": 37040
    },
    {
      "epoch": 3.18687085950271,
      "grad_norm": 27.875,
      "learning_rate": 0.00029107233183557705,
      "loss": 1.3035,
      "step": 37041
    },
    {
      "epoch": 3.186956895810032,
      "grad_norm": 1.328125,
      "learning_rate": 0.00029104776526579166,
      "loss": 1.0894,
      "step": 37042
    },
    {
      "epoch": 3.1870429321173535,
      "grad_norm": 102.0,
      "learning_rate": 0.0002910231993071512,
      "loss": 1.7134,
      "step": 37043
    },
    {
      "epoch": 3.187128968424675,
      "grad_norm": 113.5,
      "learning_rate": 0.00029099863395972705,
      "loss": 0.8503,
      "step": 37044
    },
    {
      "epoch": 3.1872150047319967,
      "grad_norm": 26.125,
      "learning_rate": 0.00029097406922359116,
      "loss": 0.876,
      "step": 37045
    },
    {
      "epoch": 3.1873010410393188,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002909495050988157,
      "loss": 0.7829,
      "step": 37046
    },
    {
      "epoch": 3.1873870773466404,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000290924941585472,
      "loss": 0.8438,
      "step": 37047
    },
    {
      "epoch": 3.187473113653962,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00029090037868363227,
      "loss": 1.1072,
      "step": 37048
    },
    {
      "epoch": 3.1875591499612836,
      "grad_norm": 1.21875,
      "learning_rate": 0.00029087581639336844,
      "loss": 1.9213,
      "step": 37049
    },
    {
      "epoch": 3.1876451862686053,
      "grad_norm": 2.125,
      "learning_rate": 0.00029085125471475207,
      "loss": 1.3668,
      "step": 37050
    },
    {
      "epoch": 3.187731222575927,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002908266936478552,
      "loss": 1.4552,
      "step": 37051
    },
    {
      "epoch": 3.187817258883249,
      "grad_norm": 1.390625,
      "learning_rate": 0.00029080213319274936,
      "loss": 0.8257,
      "step": 37052
    },
    {
      "epoch": 3.1879032951905706,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002907775733495066,
      "loss": 1.9802,
      "step": 37053
    },
    {
      "epoch": 3.187989331497892,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000290753014118199,
      "loss": 1.271,
      "step": 37054
    },
    {
      "epoch": 3.188075367805214,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00029072845549889785,
      "loss": 1.6855,
      "step": 37055
    },
    {
      "epoch": 3.1881614041125355,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00029070389749167534,
      "loss": 0.9397,
      "step": 37056
    },
    {
      "epoch": 3.188247440419857,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002906793400966034,
      "loss": 1.3927,
      "step": 37057
    },
    {
      "epoch": 3.1883334767271787,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002906547833137535,
      "loss": 0.6751,
      "step": 37058
    },
    {
      "epoch": 3.1884195130345008,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00029063022714319764,
      "loss": 1.2695,
      "step": 37059
    },
    {
      "epoch": 3.1885055493418224,
      "grad_norm": 2.28125,
      "learning_rate": 0.00029060567158500786,
      "loss": 0.756,
      "step": 37060
    },
    {
      "epoch": 3.188591585649144,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002905811166392555,
      "loss": 0.8595,
      "step": 37061
    },
    {
      "epoch": 3.1886776219564656,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002905565623060129,
      "loss": 1.7209,
      "step": 37062
    },
    {
      "epoch": 3.1887636582637873,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00029053200858535137,
      "loss": 0.9547,
      "step": 37063
    },
    {
      "epoch": 3.188849694571109,
      "grad_norm": 0.953125,
      "learning_rate": 0.00029050745547734303,
      "loss": 1.0328,
      "step": 37064
    },
    {
      "epoch": 3.1889357308784305,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029048290298205984,
      "loss": 0.7861,
      "step": 37065
    },
    {
      "epoch": 3.1890217671857526,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002904583510995732,
      "loss": 0.6488,
      "step": 37066
    },
    {
      "epoch": 3.189107803493074,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002904337998299551,
      "loss": 0.8702,
      "step": 37067
    },
    {
      "epoch": 3.189193839800396,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002904092491732776,
      "loss": 0.8961,
      "step": 37068
    },
    {
      "epoch": 3.1892798761077175,
      "grad_norm": 3.015625,
      "learning_rate": 0.0002903846991296121,
      "loss": 0.6592,
      "step": 37069
    },
    {
      "epoch": 3.189365912415039,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002903601496990305,
      "loss": 1.6352,
      "step": 37070
    },
    {
      "epoch": 3.1894519487223607,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00029033560088160475,
      "loss": 1.5625,
      "step": 37071
    },
    {
      "epoch": 3.1895379850296823,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002903110526774066,
      "loss": 0.6662,
      "step": 37072
    },
    {
      "epoch": 3.1896240213370044,
      "grad_norm": 1.0,
      "learning_rate": 0.000290286505086508,
      "loss": 0.8333,
      "step": 37073
    },
    {
      "epoch": 3.189710057644326,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002902619581089805,
      "loss": 0.7457,
      "step": 37074
    },
    {
      "epoch": 3.1897960939516476,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002902374117448958,
      "loss": 0.7315,
      "step": 37075
    },
    {
      "epoch": 3.1898821302589693,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029021286599432625,
      "loss": 1.6947,
      "step": 37076
    },
    {
      "epoch": 3.189968166566291,
      "grad_norm": 3.875,
      "learning_rate": 0.00029018832085734295,
      "loss": 0.9491,
      "step": 37077
    },
    {
      "epoch": 3.1900542028736125,
      "grad_norm": 1.3125,
      "learning_rate": 0.00029016377633401825,
      "loss": 1.4201,
      "step": 37078
    },
    {
      "epoch": 3.190140239180934,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00029013923242442353,
      "loss": 0.8529,
      "step": 37079
    },
    {
      "epoch": 3.190226275488256,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002901146891286307,
      "loss": 1.1081,
      "step": 37080
    },
    {
      "epoch": 3.190312311795578,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002900901464467117,
      "loss": 1.1303,
      "step": 37081
    },
    {
      "epoch": 3.1903983481028995,
      "grad_norm": 1.203125,
      "learning_rate": 0.00029006560437873813,
      "loss": 1.0886,
      "step": 37082
    },
    {
      "epoch": 3.190484384410221,
      "grad_norm": 1.140625,
      "learning_rate": 0.00029004106292478176,
      "loss": 1.2135,
      "step": 37083
    },
    {
      "epoch": 3.1905704207175427,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00029001652208491467,
      "loss": 1.0363,
      "step": 37084
    },
    {
      "epoch": 3.1906564570248643,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002899919818592082,
      "loss": 1.3137,
      "step": 37085
    },
    {
      "epoch": 3.190742493332186,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002899674422477344,
      "loss": 0.9321,
      "step": 37086
    },
    {
      "epoch": 3.190828529639508,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002899429032505651,
      "loss": 0.7507,
      "step": 37087
    },
    {
      "epoch": 3.1909145659468297,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002899183648677718,
      "loss": 0.6412,
      "step": 37088
    },
    {
      "epoch": 3.1910006022541513,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00028989382709942643,
      "loss": 1.7043,
      "step": 37089
    },
    {
      "epoch": 3.191086638561473,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002898692899456009,
      "loss": 1.0226,
      "step": 37090
    },
    {
      "epoch": 3.1911726748687945,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002898447534063667,
      "loss": 1.5743,
      "step": 37091
    },
    {
      "epoch": 3.191258711176116,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002898202174817959,
      "loss": 1.6259,
      "step": 37092
    },
    {
      "epoch": 3.191344747483438,
      "grad_norm": 21.25,
      "learning_rate": 0.00028979568217195984,
      "loss": 1.465,
      "step": 37093
    },
    {
      "epoch": 3.19143078379076,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00028977114747693054,
      "loss": 1.9984,
      "step": 37094
    },
    {
      "epoch": 3.1915168200980815,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00028974661339678,
      "loss": 0.5184,
      "step": 37095
    },
    {
      "epoch": 3.191602856405403,
      "grad_norm": 36.25,
      "learning_rate": 0.0002897220799315795,
      "loss": 1.6122,
      "step": 37096
    },
    {
      "epoch": 3.1916888927127247,
      "grad_norm": 1.328125,
      "learning_rate": 0.000289697547081401,
      "loss": 1.9868,
      "step": 37097
    },
    {
      "epoch": 3.1917749290200463,
      "grad_norm": 1952.0,
      "learning_rate": 0.00028967301484631645,
      "loss": 1.2439,
      "step": 37098
    },
    {
      "epoch": 3.191860965327368,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00028964848322639725,
      "loss": 0.9632,
      "step": 37099
    },
    {
      "epoch": 3.19194700163469,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00028962395222171535,
      "loss": 0.8002,
      "step": 37100
    },
    {
      "epoch": 3.1920330379420117,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002895994218323427,
      "loss": 0.9178,
      "step": 37101
    },
    {
      "epoch": 3.1921190742493333,
      "grad_norm": 2.328125,
      "learning_rate": 0.00028957489205835054,
      "loss": 1.9403,
      "step": 37102
    },
    {
      "epoch": 3.192205110556655,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002895503628998112,
      "loss": 1.5765,
      "step": 37103
    },
    {
      "epoch": 3.1922911468639765,
      "grad_norm": 1.671875,
      "learning_rate": 0.0002895258343567958,
      "loss": 1.073,
      "step": 37104
    },
    {
      "epoch": 3.192377183171298,
      "grad_norm": 282.0,
      "learning_rate": 0.00028950130642937645,
      "loss": 1.3678,
      "step": 37105
    },
    {
      "epoch": 3.1924632194786198,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00028947677911762506,
      "loss": 0.5959,
      "step": 37106
    },
    {
      "epoch": 3.192549255785942,
      "grad_norm": 1.578125,
      "learning_rate": 0.0002894522524216129,
      "loss": 1.2723,
      "step": 37107
    },
    {
      "epoch": 3.1926352920932635,
      "grad_norm": 1.8359375,
      "learning_rate": 0.000289427726341412,
      "loss": 0.8773,
      "step": 37108
    },
    {
      "epoch": 3.192721328400585,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000289403200877094,
      "loss": 1.6323,
      "step": 37109
    },
    {
      "epoch": 3.1928073647079067,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002893786760287307,
      "loss": 0.6302,
      "step": 37110
    },
    {
      "epoch": 3.1928934010152283,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000289354151796394,
      "loss": 1.1586,
      "step": 37111
    },
    {
      "epoch": 3.19297943732255,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002893296281801553,
      "loss": 2.0893,
      "step": 37112
    },
    {
      "epoch": 3.193065473629872,
      "grad_norm": 1.890625,
      "learning_rate": 0.0002893051051800864,
      "loss": 1.9813,
      "step": 37113
    },
    {
      "epoch": 3.1931515099371937,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002892805827962593,
      "loss": 2.2637,
      "step": 37114
    },
    {
      "epoch": 3.1932375462445153,
      "grad_norm": 1.109375,
      "learning_rate": 0.00028925606102874537,
      "loss": 1.1843,
      "step": 37115
    },
    {
      "epoch": 3.193323582551837,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002892315398776164,
      "loss": 1.3712,
      "step": 37116
    },
    {
      "epoch": 3.1934096188591585,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00028920701934294433,
      "loss": 1.291,
      "step": 37117
    },
    {
      "epoch": 3.19349565516648,
      "grad_norm": 4.96875,
      "learning_rate": 0.0002891824994248007,
      "loss": 1.3768,
      "step": 37118
    },
    {
      "epoch": 3.1935816914738018,
      "grad_norm": 1.28125,
      "learning_rate": 0.00028915798012325736,
      "loss": 1.061,
      "step": 37119
    },
    {
      "epoch": 3.193667727781124,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002891334614383857,
      "loss": 1.0342,
      "step": 37120
    },
    {
      "epoch": 3.1937537640884455,
      "grad_norm": 0.859375,
      "learning_rate": 0.00028910894337025767,
      "loss": 0.6782,
      "step": 37121
    },
    {
      "epoch": 3.193839800395767,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002890844259189452,
      "loss": 0.7805,
      "step": 37122
    },
    {
      "epoch": 3.1939258367030887,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002890599090845195,
      "loss": 0.6783,
      "step": 37123
    },
    {
      "epoch": 3.1940118730104103,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002890353928670526,
      "loss": 0.8925,
      "step": 37124
    },
    {
      "epoch": 3.194097909317732,
      "grad_norm": 1.203125,
      "learning_rate": 0.00028901087726661636,
      "loss": 0.9472,
      "step": 37125
    },
    {
      "epoch": 3.1941839456250536,
      "grad_norm": 57.5,
      "learning_rate": 0.000288986362283282,
      "loss": 1.4141,
      "step": 37126
    },
    {
      "epoch": 3.1942699819323757,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00028896184791712157,
      "loss": 0.7586,
      "step": 37127
    },
    {
      "epoch": 3.1943560182396973,
      "grad_norm": 7.125,
      "learning_rate": 0.0002889373341682068,
      "loss": 1.4839,
      "step": 37128
    },
    {
      "epoch": 3.194442054547019,
      "grad_norm": 386.0,
      "learning_rate": 0.0002889128210366091,
      "loss": 1.0892,
      "step": 37129
    },
    {
      "epoch": 3.1945280908543405,
      "grad_norm": 6.34375,
      "learning_rate": 0.0002888883085224004,
      "loss": 1.7529,
      "step": 37130
    },
    {
      "epoch": 3.194614127161662,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002888637966256525,
      "loss": 0.6428,
      "step": 37131
    },
    {
      "epoch": 3.1947001634689838,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00028883928534643674,
      "loss": 0.8155,
      "step": 37132
    },
    {
      "epoch": 3.1947861997763054,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002888147746848252,
      "loss": 1.1471,
      "step": 37133
    },
    {
      "epoch": 3.1948722360836275,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00028879026464088915,
      "loss": 0.853,
      "step": 37134
    },
    {
      "epoch": 3.194958272390949,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002887657552147005,
      "loss": 1.2351,
      "step": 37135
    },
    {
      "epoch": 3.1950443086982707,
      "grad_norm": 11.3125,
      "learning_rate": 0.0002887412464063311,
      "loss": 0.9491,
      "step": 37136
    },
    {
      "epoch": 3.1951303450055923,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00028871673821585244,
      "loss": 1.0798,
      "step": 37137
    },
    {
      "epoch": 3.195216381312914,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028869223064333603,
      "loss": 1.0057,
      "step": 37138
    },
    {
      "epoch": 3.1953024176202356,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00028866772368885406,
      "loss": 1.7278,
      "step": 37139
    },
    {
      "epoch": 3.195388453927557,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002886432173524777,
      "loss": 1.0947,
      "step": 37140
    },
    {
      "epoch": 3.1954744902348793,
      "grad_norm": 9.5,
      "learning_rate": 0.0002886187116342787,
      "loss": 1.6552,
      "step": 37141
    },
    {
      "epoch": 3.195560526542201,
      "grad_norm": 1.234375,
      "learning_rate": 0.00028859420653432917,
      "loss": 1.3699,
      "step": 37142
    },
    {
      "epoch": 3.1956465628495225,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002885697020527003,
      "loss": 1.5412,
      "step": 37143
    },
    {
      "epoch": 3.195732599156844,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000288545198189464,
      "loss": 1.7554,
      "step": 37144
    },
    {
      "epoch": 3.195818635464166,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002885206949446916,
      "loss": 1.8872,
      "step": 37145
    },
    {
      "epoch": 3.1959046717714874,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0002884961923184551,
      "loss": 1.0352,
      "step": 37146
    },
    {
      "epoch": 3.195990708078809,
      "grad_norm": 8.1875,
      "learning_rate": 0.00028847169031082646,
      "loss": 1.3522,
      "step": 37147
    },
    {
      "epoch": 3.196076744386131,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002884471889218767,
      "loss": 0.6531,
      "step": 37148
    },
    {
      "epoch": 3.1961627806934527,
      "grad_norm": 9.5,
      "learning_rate": 0.00028842268815167793,
      "loss": 1.6592,
      "step": 37149
    },
    {
      "epoch": 3.1962488170007743,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002883981880003015,
      "loss": 1.6643,
      "step": 37150
    },
    {
      "epoch": 3.196334853308096,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002883736884678192,
      "loss": 0.8445,
      "step": 37151
    },
    {
      "epoch": 3.1964208896154176,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002883491895543029,
      "loss": 1.6246,
      "step": 37152
    },
    {
      "epoch": 3.196506925922739,
      "grad_norm": 0.9375,
      "learning_rate": 0.00028832469125982386,
      "loss": 0.9385,
      "step": 37153
    },
    {
      "epoch": 3.1965929622300613,
      "grad_norm": 5.0625,
      "learning_rate": 0.00028830019358445393,
      "loss": 1.5942,
      "step": 37154
    },
    {
      "epoch": 3.196678998537383,
      "grad_norm": 1.09375,
      "learning_rate": 0.00028827569652826497,
      "loss": 1.18,
      "step": 37155
    },
    {
      "epoch": 3.1967650348447045,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002882512000913282,
      "loss": 1.1341,
      "step": 37156
    },
    {
      "epoch": 3.196851071152026,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002882267042737155,
      "loss": 0.6381,
      "step": 37157
    },
    {
      "epoch": 3.196937107459348,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002882022090754987,
      "loss": 0.715,
      "step": 37158
    },
    {
      "epoch": 3.1970231437666694,
      "grad_norm": 0.921875,
      "learning_rate": 0.00028817771449674907,
      "loss": 0.9187,
      "step": 37159
    },
    {
      "epoch": 3.197109180073991,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002881532205375384,
      "loss": 1.169,
      "step": 37160
    },
    {
      "epoch": 3.197195216381313,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002881287271979386,
      "loss": 0.6188,
      "step": 37161
    },
    {
      "epoch": 3.1972812526886347,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002881042344780208,
      "loss": 1.2042,
      "step": 37162
    },
    {
      "epoch": 3.1973672889959563,
      "grad_norm": 5.0,
      "learning_rate": 0.0002880797423778571,
      "loss": 0.6252,
      "step": 37163
    },
    {
      "epoch": 3.197453325303278,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0002880552508975188,
      "loss": 0.9215,
      "step": 37164
    },
    {
      "epoch": 3.1975393616105996,
      "grad_norm": 1.09375,
      "learning_rate": 0.00028803076003707765,
      "loss": 1.3902,
      "step": 37165
    },
    {
      "epoch": 3.197625397917921,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002880062697966055,
      "loss": 1.7915,
      "step": 37166
    },
    {
      "epoch": 3.1977114342252433,
      "grad_norm": 2.15625,
      "learning_rate": 0.00028798178017617354,
      "loss": 0.8158,
      "step": 37167
    },
    {
      "epoch": 3.197797470532565,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002879572911758537,
      "loss": 1.1564,
      "step": 37168
    },
    {
      "epoch": 3.1978835068398865,
      "grad_norm": 1.015625,
      "learning_rate": 0.00028793280279571773,
      "loss": 1.584,
      "step": 37169
    },
    {
      "epoch": 3.197969543147208,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002879083150358368,
      "loss": 0.7375,
      "step": 37170
    },
    {
      "epoch": 3.19805557945453,
      "grad_norm": 1.234375,
      "learning_rate": 0.00028788382789628284,
      "loss": 1.737,
      "step": 37171
    },
    {
      "epoch": 3.1981416157618514,
      "grad_norm": 10.6875,
      "learning_rate": 0.00028785934137712764,
      "loss": 1.6053,
      "step": 37172
    },
    {
      "epoch": 3.198227652069173,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002878348554784425,
      "loss": 1.118,
      "step": 37173
    },
    {
      "epoch": 3.198313688376495,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00028781037020029923,
      "loss": 1.4484,
      "step": 37174
    },
    {
      "epoch": 3.1983997246838167,
      "grad_norm": 1.015625,
      "learning_rate": 0.00028778588554276914,
      "loss": 1.0079,
      "step": 37175
    },
    {
      "epoch": 3.1984857609911383,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002877614015059241,
      "loss": 1.1818,
      "step": 37176
    },
    {
      "epoch": 3.19857179729846,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002877369180898359,
      "loss": 0.6799,
      "step": 37177
    },
    {
      "epoch": 3.1986578336057816,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002877124352945757,
      "loss": 1.4982,
      "step": 37178
    },
    {
      "epoch": 3.198743869913103,
      "grad_norm": 1.078125,
      "learning_rate": 0.00028768795312021536,
      "loss": 0.483,
      "step": 37179
    },
    {
      "epoch": 3.198829906220425,
      "grad_norm": 1.109375,
      "learning_rate": 0.00028766347156682667,
      "loss": 1.6085,
      "step": 37180
    },
    {
      "epoch": 3.198915942527747,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002876389906344808,
      "loss": 1.0443,
      "step": 37181
    },
    {
      "epoch": 3.1990019788350685,
      "grad_norm": 1.125,
      "learning_rate": 0.0002876145103232496,
      "loss": 1.0333,
      "step": 37182
    },
    {
      "epoch": 3.19908801514239,
      "grad_norm": 2.359375,
      "learning_rate": 0.00028759003063320463,
      "loss": 1.262,
      "step": 37183
    },
    {
      "epoch": 3.199174051449712,
      "grad_norm": 2.71875,
      "learning_rate": 0.00028756555156441755,
      "loss": 1.9067,
      "step": 37184
    },
    {
      "epoch": 3.1992600877570334,
      "grad_norm": 1.34375,
      "learning_rate": 0.00028754107311696007,
      "loss": 0.7734,
      "step": 37185
    },
    {
      "epoch": 3.199346124064355,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00028751659529090346,
      "loss": 1.1022,
      "step": 37186
    },
    {
      "epoch": 3.1994321603716767,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00028749211808631945,
      "loss": 1.1439,
      "step": 37187
    },
    {
      "epoch": 3.1995181966789987,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00028746764150327993,
      "loss": 0.9852,
      "step": 37188
    },
    {
      "epoch": 3.1996042329863204,
      "grad_norm": 1.078125,
      "learning_rate": 0.00028744316554185596,
      "loss": 1.2771,
      "step": 37189
    },
    {
      "epoch": 3.199690269293642,
      "grad_norm": 1.09375,
      "learning_rate": 0.00028741869020211964,
      "loss": 0.9986,
      "step": 37190
    },
    {
      "epoch": 3.1997763056009636,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00028739421548414204,
      "loss": 1.1513,
      "step": 37191
    },
    {
      "epoch": 3.1998623419082852,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002873697413879951,
      "loss": 0.6441,
      "step": 37192
    },
    {
      "epoch": 3.199948378215607,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002873452679137504,
      "loss": 1.0493,
      "step": 37193
    },
    {
      "epoch": 3.2000344145229285,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00028732079506147936,
      "loss": 0.9861,
      "step": 37194
    },
    {
      "epoch": 3.2001204508302505,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002872963228312535,
      "loss": 0.4284,
      "step": 37195
    },
    {
      "epoch": 3.200206487137572,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00028727185122314484,
      "loss": 2.1003,
      "step": 37196
    },
    {
      "epoch": 3.200292523444894,
      "grad_norm": 2.421875,
      "learning_rate": 0.00028724738023722443,
      "loss": 1.0716,
      "step": 37197
    },
    {
      "epoch": 3.2003785597522154,
      "grad_norm": 1.21875,
      "learning_rate": 0.000287222909873564,
      "loss": 1.2418,
      "step": 37198
    },
    {
      "epoch": 3.200464596059537,
      "grad_norm": 1.84375,
      "learning_rate": 0.0002871984401322354,
      "loss": 1.775,
      "step": 37199
    },
    {
      "epoch": 3.2005506323668587,
      "grad_norm": 16.0,
      "learning_rate": 0.00028717397101330974,
      "loss": 1.875,
      "step": 37200
    },
    {
      "epoch": 3.2006366686741803,
      "grad_norm": 9.125,
      "learning_rate": 0.00028714950251685885,
      "loss": 1.2478,
      "step": 37201
    },
    {
      "epoch": 3.2007227049815024,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00028712503464295447,
      "loss": 1.1858,
      "step": 37202
    },
    {
      "epoch": 3.200808741288824,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002871005673916678,
      "loss": 0.5011,
      "step": 37203
    },
    {
      "epoch": 3.2008947775961456,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00028707610076307067,
      "loss": 0.7674,
      "step": 37204
    },
    {
      "epoch": 3.2009808139034672,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002870516347572344,
      "loss": 0.539,
      "step": 37205
    },
    {
      "epoch": 3.201066850210789,
      "grad_norm": 1.390625,
      "learning_rate": 0.00028702716937423065,
      "loss": 1.4534,
      "step": 37206
    },
    {
      "epoch": 3.2011528865181105,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00028700270461413127,
      "loss": 1.4786,
      "step": 37207
    },
    {
      "epoch": 3.201238922825432,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00028697824047700725,
      "loss": 0.9877,
      "step": 37208
    },
    {
      "epoch": 3.201324959132754,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00028695377696293057,
      "loss": 1.4033,
      "step": 37209
    },
    {
      "epoch": 3.201410995440076,
      "grad_norm": 0.9375,
      "learning_rate": 0.00028692931407197275,
      "loss": 1.3388,
      "step": 37210
    },
    {
      "epoch": 3.2014970317473974,
      "grad_norm": 2.09375,
      "learning_rate": 0.0002869048518042051,
      "loss": 0.8652,
      "step": 37211
    },
    {
      "epoch": 3.201583068054719,
      "grad_norm": 1.546875,
      "learning_rate": 0.00028688039015969935,
      "loss": 1.2248,
      "step": 37212
    },
    {
      "epoch": 3.2016691043620407,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00028685592913852716,
      "loss": 0.9655,
      "step": 37213
    },
    {
      "epoch": 3.2017551406693623,
      "grad_norm": 17.375,
      "learning_rate": 0.0002868314687407597,
      "loss": 1.1342,
      "step": 37214
    },
    {
      "epoch": 3.2018411769766844,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000286807008966469,
      "loss": 1.0942,
      "step": 37215
    },
    {
      "epoch": 3.201927213284006,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002867825498157261,
      "loss": 0.786,
      "step": 37216
    },
    {
      "epoch": 3.2020132495913276,
      "grad_norm": 1.234375,
      "learning_rate": 0.00028675809128860273,
      "loss": 0.8459,
      "step": 37217
    },
    {
      "epoch": 3.2020992858986492,
      "grad_norm": 1.328125,
      "learning_rate": 0.00028673363338517076,
      "loss": 1.7385,
      "step": 37218
    },
    {
      "epoch": 3.202185322205971,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00028670917610550096,
      "loss": 1.3113,
      "step": 37219
    },
    {
      "epoch": 3.2022713585132925,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002866847194496658,
      "loss": 1.3219,
      "step": 37220
    },
    {
      "epoch": 3.202357394820614,
      "grad_norm": 4.53125,
      "learning_rate": 0.0002866602634177361,
      "loss": 0.9987,
      "step": 37221
    },
    {
      "epoch": 3.202443431127936,
      "grad_norm": 1.4375,
      "learning_rate": 0.00028663580800978363,
      "loss": 1.4247,
      "step": 37222
    },
    {
      "epoch": 3.202529467435258,
      "grad_norm": 1.296875,
      "learning_rate": 0.00028661135322588017,
      "loss": 0.7338,
      "step": 37223
    },
    {
      "epoch": 3.2026155037425794,
      "grad_norm": 1.234375,
      "learning_rate": 0.00028658689906609677,
      "loss": 1.1211,
      "step": 37224
    },
    {
      "epoch": 3.202701540049901,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00028656244553050515,
      "loss": 0.7848,
      "step": 37225
    },
    {
      "epoch": 3.2027875763572227,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00028653799261917713,
      "loss": 1.347,
      "step": 37226
    },
    {
      "epoch": 3.2028736126645443,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00028651354033218374,
      "loss": 0.9804,
      "step": 37227
    },
    {
      "epoch": 3.2029596489718664,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00028648908866959674,
      "loss": 1.3331,
      "step": 37228
    },
    {
      "epoch": 3.203045685279188,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00028646463763148776,
      "loss": 1.1512,
      "step": 37229
    },
    {
      "epoch": 3.2031317215865096,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00028644018721792807,
      "loss": 1.0278,
      "step": 37230
    },
    {
      "epoch": 3.2032177578938312,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00028641573742898944,
      "loss": 0.6271,
      "step": 37231
    },
    {
      "epoch": 3.203303794201153,
      "grad_norm": 1.15625,
      "learning_rate": 0.00028639128826474307,
      "loss": 0.9088,
      "step": 37232
    },
    {
      "epoch": 3.2033898305084745,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00028636683972526066,
      "loss": 1.0259,
      "step": 37233
    },
    {
      "epoch": 3.203475866815796,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002863423918106138,
      "loss": 0.9817,
      "step": 37234
    },
    {
      "epoch": 3.203561903123118,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002863179445208738,
      "loss": 1.5106,
      "step": 37235
    },
    {
      "epoch": 3.20364793943044,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00028629349785611226,
      "loss": 1.8911,
      "step": 37236
    },
    {
      "epoch": 3.2037339757377614,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002862690518164008,
      "loss": 0.5749,
      "step": 37237
    },
    {
      "epoch": 3.203820012045083,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002862446064018106,
      "loss": 0.8038,
      "step": 37238
    },
    {
      "epoch": 3.2039060483524047,
      "grad_norm": 20.5,
      "learning_rate": 0.0002862201616124134,
      "loss": 0.7048,
      "step": 37239
    },
    {
      "epoch": 3.2039920846597263,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002861957174482808,
      "loss": 0.7674,
      "step": 37240
    },
    {
      "epoch": 3.204078120967048,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00028617127390948396,
      "loss": 0.56,
      "step": 37241
    },
    {
      "epoch": 3.20416415727437,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002861468309960945,
      "loss": 1.4785,
      "step": 37242
    },
    {
      "epoch": 3.2042501935816916,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002861223887081843,
      "loss": 0.9901,
      "step": 37243
    },
    {
      "epoch": 3.2043362298890132,
      "grad_norm": 1.3125,
      "learning_rate": 0.00028609794704582427,
      "loss": 1.8515,
      "step": 37244
    },
    {
      "epoch": 3.204422266196335,
      "grad_norm": 1.078125,
      "learning_rate": 0.00028607350600908627,
      "loss": 1.0292,
      "step": 37245
    },
    {
      "epoch": 3.2045083025036565,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002860490655980415,
      "loss": 1.7278,
      "step": 37246
    },
    {
      "epoch": 3.204594338810978,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002860246258127616,
      "loss": 0.8725,
      "step": 37247
    },
    {
      "epoch": 3.2046803751182997,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002860001866533183,
      "loss": 1.4849,
      "step": 37248
    },
    {
      "epoch": 3.204766411425622,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00028597574811978254,
      "loss": 0.6252,
      "step": 37249
    },
    {
      "epoch": 3.2048524477329434,
      "grad_norm": 1.125,
      "learning_rate": 0.0002859513102122262,
      "loss": 1.4593,
      "step": 37250
    },
    {
      "epoch": 3.204938484040265,
      "grad_norm": 1.78125,
      "learning_rate": 0.00028592687293072075,
      "loss": 0.9306,
      "step": 37251
    },
    {
      "epoch": 3.2050245203475867,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002859024362753374,
      "loss": 1.2548,
      "step": 37252
    },
    {
      "epoch": 3.2051105566549083,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002858780002461478,
      "loss": 0.8078,
      "step": 37253
    },
    {
      "epoch": 3.20519659296223,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002858535648432236,
      "loss": 1.0848,
      "step": 37254
    },
    {
      "epoch": 3.2052826292695515,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00028582913006663593,
      "loss": 0.3649,
      "step": 37255
    },
    {
      "epoch": 3.2053686655768736,
      "grad_norm": 1.375,
      "learning_rate": 0.0002858046959164563,
      "loss": 1.135,
      "step": 37256
    },
    {
      "epoch": 3.2054547018841952,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002857802623927563,
      "loss": 1.6677,
      "step": 37257
    },
    {
      "epoch": 3.205540738191517,
      "grad_norm": 1.0,
      "learning_rate": 0.0002857558294956074,
      "loss": 0.5701,
      "step": 37258
    },
    {
      "epoch": 3.2056267744988385,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002857313972250812,
      "loss": 0.5623,
      "step": 37259
    },
    {
      "epoch": 3.20571281080616,
      "grad_norm": 2.1875,
      "learning_rate": 0.00028570696558124874,
      "loss": 0.9478,
      "step": 37260
    },
    {
      "epoch": 3.2057988471134817,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00028568253456418203,
      "loss": 1.3001,
      "step": 37261
    },
    {
      "epoch": 3.2058848834208034,
      "grad_norm": 1.109375,
      "learning_rate": 0.000285658104173952,
      "loss": 1.279,
      "step": 37262
    },
    {
      "epoch": 3.2059709197281254,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0002856336744106303,
      "loss": 0.9397,
      "step": 37263
    },
    {
      "epoch": 3.206056956035447,
      "grad_norm": 2.84375,
      "learning_rate": 0.00028560924527428865,
      "loss": 1.1353,
      "step": 37264
    },
    {
      "epoch": 3.2061429923427687,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00028558481676499805,
      "loss": 0.7452,
      "step": 37265
    },
    {
      "epoch": 3.2062290286500903,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002855603888828301,
      "loss": 0.7119,
      "step": 37266
    },
    {
      "epoch": 3.206315064957412,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00028553596162785656,
      "loss": 1.8007,
      "step": 37267
    },
    {
      "epoch": 3.2064011012647335,
      "grad_norm": 1600.0,
      "learning_rate": 0.0002855115350001484,
      "loss": 1.5276,
      "step": 37268
    },
    {
      "epoch": 3.2064871375720556,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002854871089997773,
      "loss": 1.7214,
      "step": 37269
    },
    {
      "epoch": 3.2065731738793772,
      "grad_norm": 2.0,
      "learning_rate": 0.00028546268362681493,
      "loss": 1.2515,
      "step": 37270
    },
    {
      "epoch": 3.206659210186699,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002854382588813322,
      "loss": 0.873,
      "step": 37271
    },
    {
      "epoch": 3.2067452464940205,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00028541383476340113,
      "loss": 1.3332,
      "step": 37272
    },
    {
      "epoch": 3.206831282801342,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002853894112730926,
      "loss": 1.1699,
      "step": 37273
    },
    {
      "epoch": 3.2069173191086637,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002853649884104783,
      "loss": 1.5475,
      "step": 37274
    },
    {
      "epoch": 3.2070033554159854,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002853405661756298,
      "loss": 1.319,
      "step": 37275
    },
    {
      "epoch": 3.2070893917233074,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00028531614456861834,
      "loss": 1.1433,
      "step": 37276
    },
    {
      "epoch": 3.207175428030629,
      "grad_norm": 1.390625,
      "learning_rate": 0.00028529172358951525,
      "loss": 1.1756,
      "step": 37277
    },
    {
      "epoch": 3.2072614643379507,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00028526730323839237,
      "loss": 1.492,
      "step": 37278
    },
    {
      "epoch": 3.2073475006452723,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002852428835153207,
      "loss": 1.1065,
      "step": 37279
    },
    {
      "epoch": 3.207433536952594,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0002852184644203718,
      "loss": 1.7169,
      "step": 37280
    },
    {
      "epoch": 3.2075195732599155,
      "grad_norm": 0.953125,
      "learning_rate": 0.00028519404595361733,
      "loss": 1.7299,
      "step": 37281
    },
    {
      "epoch": 3.207605609567237,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002851696281151283,
      "loss": 1.2146,
      "step": 37282
    },
    {
      "epoch": 3.2076916458745592,
      "grad_norm": 1.359375,
      "learning_rate": 0.00028514521090497637,
      "loss": 2.3371,
      "step": 37283
    },
    {
      "epoch": 3.207777682181881,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00028512079432323304,
      "loss": 1.5461,
      "step": 37284
    },
    {
      "epoch": 3.2078637184892025,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002850963783699694,
      "loss": 1.1225,
      "step": 37285
    },
    {
      "epoch": 3.207949754796524,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002850719630452573,
      "loss": 1.4717,
      "step": 37286
    },
    {
      "epoch": 3.2080357911038457,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002850475483491677,
      "loss": 1.6056,
      "step": 37287
    },
    {
      "epoch": 3.2081218274111674,
      "grad_norm": 7.875,
      "learning_rate": 0.0002850231342817722,
      "loss": 1.1696,
      "step": 37288
    },
    {
      "epoch": 3.2082078637184894,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002849987208431424,
      "loss": 0.6611,
      "step": 37289
    },
    {
      "epoch": 3.208293900025811,
      "grad_norm": 1.125,
      "learning_rate": 0.00028497430803334933,
      "loss": 1.0194,
      "step": 37290
    },
    {
      "epoch": 3.2083799363331327,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002849498958524646,
      "loss": 0.8234,
      "step": 37291
    },
    {
      "epoch": 3.2084659726404543,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002849254843005598,
      "loss": 0.531,
      "step": 37292
    },
    {
      "epoch": 3.208552008947776,
      "grad_norm": 1.4375,
      "learning_rate": 0.0002849010733777057,
      "loss": 1.114,
      "step": 37293
    },
    {
      "epoch": 3.2086380452550975,
      "grad_norm": 12.0,
      "learning_rate": 0.00028487666308397454,
      "loss": 1.4175,
      "step": 37294
    },
    {
      "epoch": 3.208724081562419,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00028485225341943717,
      "loss": 1.7054,
      "step": 37295
    },
    {
      "epoch": 3.2088101178697412,
      "grad_norm": 2.125,
      "learning_rate": 0.00028482784438416505,
      "loss": 0.3835,
      "step": 37296
    },
    {
      "epoch": 3.208896154177063,
      "grad_norm": 1.234375,
      "learning_rate": 0.00028480343597822984,
      "loss": 1.8665,
      "step": 37297
    },
    {
      "epoch": 3.2089821904843845,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028477902820170256,
      "loss": 1.5183,
      "step": 37298
    },
    {
      "epoch": 3.209068226791706,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002847546210546547,
      "loss": 0.4979,
      "step": 37299
    },
    {
      "epoch": 3.2091542630990277,
      "grad_norm": 23.375,
      "learning_rate": 0.00028473021453715797,
      "loss": 1.4032,
      "step": 37300
    },
    {
      "epoch": 3.2092402994063494,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002847058086492833,
      "loss": 1.1133,
      "step": 37301
    },
    {
      "epoch": 3.209326335713671,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002846814033911024,
      "loss": 1.3604,
      "step": 37302
    },
    {
      "epoch": 3.209412372020993,
      "grad_norm": 1.28125,
      "learning_rate": 0.00028465699876268637,
      "loss": 1.542,
      "step": 37303
    },
    {
      "epoch": 3.2094984083283147,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002846325947641067,
      "loss": 1.671,
      "step": 37304
    },
    {
      "epoch": 3.2095844446356363,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00028460819139543503,
      "loss": 1.5484,
      "step": 37305
    },
    {
      "epoch": 3.209670480942958,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002845837886567423,
      "loss": 1.0665,
      "step": 37306
    },
    {
      "epoch": 3.2097565172502796,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002845593865481001,
      "loss": 1.9167,
      "step": 37307
    },
    {
      "epoch": 3.209842553557601,
      "grad_norm": 93.5,
      "learning_rate": 0.00028453498506958,
      "loss": 1.6527,
      "step": 37308
    },
    {
      "epoch": 3.209928589864923,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000284510584221253,
      "loss": 1.6227,
      "step": 37309
    },
    {
      "epoch": 3.210014626172245,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002844861840031906,
      "loss": 0.8167,
      "step": 37310
    },
    {
      "epoch": 3.2101006624795665,
      "grad_norm": 6.75,
      "learning_rate": 0.00028446178441546435,
      "loss": 1.797,
      "step": 37311
    },
    {
      "epoch": 3.210186698786888,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00028443738545814534,
      "loss": 0.9685,
      "step": 37312
    },
    {
      "epoch": 3.2102727350942097,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000284412987131305,
      "loss": 0.8232,
      "step": 37313
    },
    {
      "epoch": 3.2103587714015314,
      "grad_norm": 1.53125,
      "learning_rate": 0.00028438858943501496,
      "loss": 1.4345,
      "step": 37314
    },
    {
      "epoch": 3.210444807708853,
      "grad_norm": 169.0,
      "learning_rate": 0.00028436419236934615,
      "loss": 0.7333,
      "step": 37315
    },
    {
      "epoch": 3.2105308440161746,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002843397959343703,
      "loss": 0.6371,
      "step": 37316
    },
    {
      "epoch": 3.2106168803234967,
      "grad_norm": 0.875,
      "learning_rate": 0.00028431540013015844,
      "loss": 1.3452,
      "step": 37317
    },
    {
      "epoch": 3.2107029166308183,
      "grad_norm": 5.65625,
      "learning_rate": 0.0002842910049567821,
      "loss": 1.2572,
      "step": 37318
    },
    {
      "epoch": 3.21078895293814,
      "grad_norm": 81.5,
      "learning_rate": 0.00028426661041431277,
      "loss": 1.2475,
      "step": 37319
    },
    {
      "epoch": 3.2108749892454616,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00028424221650282156,
      "loss": 1.1152,
      "step": 37320
    },
    {
      "epoch": 3.210961025552783,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002842178232223798,
      "loss": 1.1641,
      "step": 37321
    },
    {
      "epoch": 3.211047061860105,
      "grad_norm": 0.953125,
      "learning_rate": 0.00028419343057305916,
      "loss": 1.5189,
      "step": 37322
    },
    {
      "epoch": 3.2111330981674264,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002841690385549306,
      "loss": 1.1254,
      "step": 37323
    },
    {
      "epoch": 3.2112191344747485,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002841446471680656,
      "loss": 0.9368,
      "step": 37324
    },
    {
      "epoch": 3.21130517078207,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002841202564125357,
      "loss": 1.4025,
      "step": 37325
    },
    {
      "epoch": 3.2113912070893917,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002840958662884119,
      "loss": 0.5508,
      "step": 37326
    },
    {
      "epoch": 3.2114772433967134,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002840714767957659,
      "loss": 1.3348,
      "step": 37327
    },
    {
      "epoch": 3.211563279704035,
      "grad_norm": 2.34375,
      "learning_rate": 0.0002840470879346686,
      "loss": 0.9204,
      "step": 37328
    },
    {
      "epoch": 3.2116493160113566,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002840226997051914,
      "loss": 0.8409,
      "step": 37329
    },
    {
      "epoch": 3.2117353523186787,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00028399831210740626,
      "loss": 1.2245,
      "step": 37330
    },
    {
      "epoch": 3.2118213886260003,
      "grad_norm": 5.90625,
      "learning_rate": 0.0002839739251413837,
      "loss": 1.786,
      "step": 37331
    },
    {
      "epoch": 3.211907424933322,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00028394953880719567,
      "loss": 1.0419,
      "step": 37332
    },
    {
      "epoch": 3.2119934612406436,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00028392515310491306,
      "loss": 1.1865,
      "step": 37333
    },
    {
      "epoch": 3.212079497547965,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002839007680346073,
      "loss": 0.6513,
      "step": 37334
    },
    {
      "epoch": 3.212165533855287,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002838763835963499,
      "loss": 0.771,
      "step": 37335
    },
    {
      "epoch": 3.2122515701626084,
      "grad_norm": 1.1875,
      "learning_rate": 0.000283851999790212,
      "loss": 1.0533,
      "step": 37336
    },
    {
      "epoch": 3.2123376064699305,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002838276166162649,
      "loss": 0.7018,
      "step": 37337
    },
    {
      "epoch": 3.212423642777252,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002838032340745801,
      "loss": 1.6357,
      "step": 37338
    },
    {
      "epoch": 3.2125096790845737,
      "grad_norm": 5.0,
      "learning_rate": 0.0002837788521652287,
      "loss": 1.8571,
      "step": 37339
    },
    {
      "epoch": 3.2125957153918954,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00028375447088828203,
      "loss": 1.6863,
      "step": 37340
    },
    {
      "epoch": 3.212681751699217,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002837300902438117,
      "loss": 0.7587,
      "step": 37341
    },
    {
      "epoch": 3.2127677880065386,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00028370571023188864,
      "loss": 1.0164,
      "step": 37342
    },
    {
      "epoch": 3.2128538243138607,
      "grad_norm": 3.0,
      "learning_rate": 0.0002836813308525844,
      "loss": 0.889,
      "step": 37343
    },
    {
      "epoch": 3.2129398606211823,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002836569521059701,
      "loss": 1.4059,
      "step": 37344
    },
    {
      "epoch": 3.213025896928504,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002836325739921171,
      "loss": 0.6727,
      "step": 37345
    },
    {
      "epoch": 3.2131119332358256,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002836081965110969,
      "loss": 0.7969,
      "step": 37346
    },
    {
      "epoch": 3.213197969543147,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002835838196629805,
      "loss": 1.3956,
      "step": 37347
    },
    {
      "epoch": 3.213284005850469,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00028355944344783934,
      "loss": 0.884,
      "step": 37348
    },
    {
      "epoch": 3.2133700421577904,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002835350678657449,
      "loss": 1.1052,
      "step": 37349
    },
    {
      "epoch": 3.2134560784651125,
      "grad_norm": 2.515625,
      "learning_rate": 0.00028351069291676807,
      "loss": 1.4153,
      "step": 37350
    },
    {
      "epoch": 3.213542114772434,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002834863186009804,
      "loss": 0.7791,
      "step": 37351
    },
    {
      "epoch": 3.2136281510797557,
      "grad_norm": 0.890625,
      "learning_rate": 0.00028346194491845326,
      "loss": 1.3213,
      "step": 37352
    },
    {
      "epoch": 3.2137141873870774,
      "grad_norm": 1.359375,
      "learning_rate": 0.00028343757186925767,
      "loss": 1.5683,
      "step": 37353
    },
    {
      "epoch": 3.213800223694399,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002834131994534651,
      "loss": 0.842,
      "step": 37354
    },
    {
      "epoch": 3.2138862600017206,
      "grad_norm": 1.421875,
      "learning_rate": 0.00028338882767114693,
      "loss": 1.4236,
      "step": 37355
    },
    {
      "epoch": 3.2139722963090422,
      "grad_norm": 1.0,
      "learning_rate": 0.0002833644565223741,
      "loss": 0.9888,
      "step": 37356
    },
    {
      "epoch": 3.2140583326163643,
      "grad_norm": 600.0,
      "learning_rate": 0.0002833400860072183,
      "loss": 1.1267,
      "step": 37357
    },
    {
      "epoch": 3.214144368923686,
      "grad_norm": 1.28125,
      "learning_rate": 0.00028331571612575046,
      "loss": 1.3073,
      "step": 37358
    },
    {
      "epoch": 3.2142304052310076,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00028329134687804205,
      "loss": 0.665,
      "step": 37359
    },
    {
      "epoch": 3.214316441538329,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002832669782641644,
      "loss": 1.0131,
      "step": 37360
    },
    {
      "epoch": 3.214402477845651,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028324261028418854,
      "loss": 0.9056,
      "step": 37361
    },
    {
      "epoch": 3.2144885141529724,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00028321824293818586,
      "loss": 0.6471,
      "step": 37362
    },
    {
      "epoch": 3.214574550460294,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00028319387622622797,
      "loss": 1.4132,
      "step": 37363
    },
    {
      "epoch": 3.214660586767616,
      "grad_norm": 1.328125,
      "learning_rate": 0.00028316951014838545,
      "loss": 1.4837,
      "step": 37364
    },
    {
      "epoch": 3.2147466230749377,
      "grad_norm": 22.5,
      "learning_rate": 0.0002831451447047301,
      "loss": 1.0578,
      "step": 37365
    },
    {
      "epoch": 3.2148326593822594,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002831207798953329,
      "loss": 1.0504,
      "step": 37366
    },
    {
      "epoch": 3.214918695689581,
      "grad_norm": 1.0,
      "learning_rate": 0.0002830964157202653,
      "loss": 1.2494,
      "step": 37367
    },
    {
      "epoch": 3.2150047319969026,
      "grad_norm": 1.296875,
      "learning_rate": 0.00028307205217959863,
      "loss": 1.0875,
      "step": 37368
    },
    {
      "epoch": 3.2150907683042242,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002830476892734039,
      "loss": 0.8377,
      "step": 37369
    },
    {
      "epoch": 3.215176804611546,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002830233270017525,
      "loss": 1.1208,
      "step": 37370
    },
    {
      "epoch": 3.215262840918868,
      "grad_norm": 3.34375,
      "learning_rate": 0.0002829989653647158,
      "loss": 1.2498,
      "step": 37371
    },
    {
      "epoch": 3.2153488772261896,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00028297460436236475,
      "loss": 1.4458,
      "step": 37372
    },
    {
      "epoch": 3.215434913533511,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028295024399477097,
      "loss": 0.7557,
      "step": 37373
    },
    {
      "epoch": 3.215520949840833,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00028292588426200526,
      "loss": 0.9049,
      "step": 37374
    },
    {
      "epoch": 3.2156069861481544,
      "grad_norm": 1.421875,
      "learning_rate": 0.00028290152516413923,
      "loss": 0.9722,
      "step": 37375
    },
    {
      "epoch": 3.215693022455476,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002828771667012442,
      "loss": 1.5438,
      "step": 37376
    },
    {
      "epoch": 3.2157790587627977,
      "grad_norm": 1.40625,
      "learning_rate": 0.00028285280887339095,
      "loss": 2.1357,
      "step": 37377
    },
    {
      "epoch": 3.2158650950701198,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00028282845168065105,
      "loss": 0.9897,
      "step": 37378
    },
    {
      "epoch": 3.2159511313774414,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002828040951230959,
      "loss": 0.7541,
      "step": 37379
    },
    {
      "epoch": 3.216037167684763,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00028277973920079626,
      "loss": 1.5639,
      "step": 37380
    },
    {
      "epoch": 3.2161232039920846,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00028275538391382373,
      "loss": 1.4728,
      "step": 37381
    },
    {
      "epoch": 3.2162092402994062,
      "grad_norm": 1.125,
      "learning_rate": 0.00028273102926224957,
      "loss": 0.6837,
      "step": 37382
    },
    {
      "epoch": 3.216295276606728,
      "grad_norm": 1.390625,
      "learning_rate": 0.0002827066752461447,
      "loss": 1.1352,
      "step": 37383
    },
    {
      "epoch": 3.2163813129140495,
      "grad_norm": 4.8125,
      "learning_rate": 0.0002826823218655809,
      "loss": 2.1403,
      "step": 37384
    },
    {
      "epoch": 3.2164673492213716,
      "grad_norm": 11.5,
      "learning_rate": 0.00028265796912062865,
      "loss": 1.5157,
      "step": 37385
    },
    {
      "epoch": 3.216553385528693,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002826336170113596,
      "loss": 1.5471,
      "step": 37386
    },
    {
      "epoch": 3.216639421836015,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00028260926553784515,
      "loss": 0.8962,
      "step": 37387
    },
    {
      "epoch": 3.2167254581433364,
      "grad_norm": 2.265625,
      "learning_rate": 0.0002825849147001562,
      "loss": 1.3193,
      "step": 37388
    },
    {
      "epoch": 3.216811494450658,
      "grad_norm": 66.5,
      "learning_rate": 0.000282560564498364,
      "loss": 1.1046,
      "step": 37389
    },
    {
      "epoch": 3.2168975307579797,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002825362149325401,
      "loss": 1.8988,
      "step": 37390
    },
    {
      "epoch": 3.2169835670653018,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002825118660027553,
      "loss": 1.3331,
      "step": 37391
    },
    {
      "epoch": 3.2170696033726234,
      "grad_norm": 1.7109375,
      "learning_rate": 0.000282487517709081,
      "loss": 1.4935,
      "step": 37392
    },
    {
      "epoch": 3.217155639679945,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002824631700515886,
      "loss": 0.5745,
      "step": 37393
    },
    {
      "epoch": 3.2172416759872666,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002824388230303488,
      "loss": 0.4723,
      "step": 37394
    },
    {
      "epoch": 3.2173277122945882,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002824144766454333,
      "loss": 1.0726,
      "step": 37395
    },
    {
      "epoch": 3.21741374860191,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00028239013089691325,
      "loss": 1.172,
      "step": 37396
    },
    {
      "epoch": 3.2174997849092315,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00028236578578485957,
      "loss": 0.6856,
      "step": 37397
    },
    {
      "epoch": 3.2175858212165536,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002823414413093438,
      "loss": 1.626,
      "step": 37398
    },
    {
      "epoch": 3.217671857523875,
      "grad_norm": 6.8125,
      "learning_rate": 0.00028231709747043675,
      "loss": 1.0503,
      "step": 37399
    },
    {
      "epoch": 3.217757893831197,
      "grad_norm": 0.84375,
      "learning_rate": 0.00028229275426820997,
      "loss": 1.0957,
      "step": 37400
    },
    {
      "epoch": 3.2178439301385184,
      "grad_norm": 1.09375,
      "learning_rate": 0.00028226841170273464,
      "loss": 0.6097,
      "step": 37401
    },
    {
      "epoch": 3.21792996644584,
      "grad_norm": 2.21875,
      "learning_rate": 0.0002822440697740817,
      "loss": 1.9631,
      "step": 37402
    },
    {
      "epoch": 3.2180160027531617,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00028221972848232254,
      "loss": 0.6926,
      "step": 37403
    },
    {
      "epoch": 3.2181020390604838,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002821953878275283,
      "loss": 1.8629,
      "step": 37404
    },
    {
      "epoch": 3.2181880753678054,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00028217104780977015,
      "loss": 1.307,
      "step": 37405
    },
    {
      "epoch": 3.218274111675127,
      "grad_norm": 3.3125,
      "learning_rate": 0.00028214670842911956,
      "loss": 0.9919,
      "step": 37406
    },
    {
      "epoch": 3.2183601479824486,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00028212236968564727,
      "loss": 1.2292,
      "step": 37407
    },
    {
      "epoch": 3.2184461842897703,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00028209803157942467,
      "loss": 0.8608,
      "step": 37408
    },
    {
      "epoch": 3.218532220597092,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0002820736941105231,
      "loss": 1.795,
      "step": 37409
    },
    {
      "epoch": 3.2186182569044135,
      "grad_norm": 25.875,
      "learning_rate": 0.0002820493572790135,
      "loss": 0.8353,
      "step": 37410
    },
    {
      "epoch": 3.2187042932117356,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00028202502108496704,
      "loss": 0.7993,
      "step": 37411
    },
    {
      "epoch": 3.218790329519057,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002820006855284552,
      "loss": 0.4469,
      "step": 37412
    },
    {
      "epoch": 3.218876365826379,
      "grad_norm": 2.78125,
      "learning_rate": 0.0002819763506095488,
      "loss": 1.0267,
      "step": 37413
    },
    {
      "epoch": 3.2189624021337004,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00028195201632831937,
      "loss": 0.7938,
      "step": 37414
    },
    {
      "epoch": 3.219048438441022,
      "grad_norm": 1.6875,
      "learning_rate": 0.0002819276826848377,
      "loss": 0.8086,
      "step": 37415
    },
    {
      "epoch": 3.2191344747483437,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00028190334967917506,
      "loss": 1.6166,
      "step": 37416
    },
    {
      "epoch": 3.2192205110556653,
      "grad_norm": 1.171875,
      "learning_rate": 0.000281879017311403,
      "loss": 0.993,
      "step": 37417
    },
    {
      "epoch": 3.2193065473629874,
      "grad_norm": 0.859375,
      "learning_rate": 0.00028185468558159215,
      "loss": 0.7125,
      "step": 37418
    },
    {
      "epoch": 3.219392583670309,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0002818303544898139,
      "loss": 1.0616,
      "step": 37419
    },
    {
      "epoch": 3.2194786199776306,
      "grad_norm": 772.0,
      "learning_rate": 0.0002818060240361397,
      "loss": 1.7228,
      "step": 37420
    },
    {
      "epoch": 3.2195646562849523,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00028178169422064017,
      "loss": 1.0492,
      "step": 37421
    },
    {
      "epoch": 3.219650692592274,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002817573650433868,
      "loss": 0.6004,
      "step": 37422
    },
    {
      "epoch": 3.2197367288995955,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002817330365044509,
      "loss": 1.1484,
      "step": 37423
    },
    {
      "epoch": 3.219822765206917,
      "grad_norm": 4.15625,
      "learning_rate": 0.00028170870860390313,
      "loss": 0.7687,
      "step": 37424
    },
    {
      "epoch": 3.219908801514239,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028168438134181493,
      "loss": 0.9666,
      "step": 37425
    },
    {
      "epoch": 3.219994837821561,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002816600547182578,
      "loss": 1.5462,
      "step": 37426
    },
    {
      "epoch": 3.2200808741288824,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00028163572873330224,
      "loss": 1.5847,
      "step": 37427
    },
    {
      "epoch": 3.220166910436204,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002816114033870199,
      "loss": 0.5488,
      "step": 37428
    },
    {
      "epoch": 3.2202529467435257,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002815870786794815,
      "loss": 0.7946,
      "step": 37429
    },
    {
      "epoch": 3.2203389830508473,
      "grad_norm": 0.828125,
      "learning_rate": 0.00028156275461075855,
      "loss": 1.0535,
      "step": 37430
    },
    {
      "epoch": 3.220425019358169,
      "grad_norm": 1.265625,
      "learning_rate": 0.00028153843118092214,
      "loss": 0.6085,
      "step": 37431
    },
    {
      "epoch": 3.220511055665491,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002815141083900432,
      "loss": 0.6872,
      "step": 37432
    },
    {
      "epoch": 3.2205970919728126,
      "grad_norm": 1.21875,
      "learning_rate": 0.00028148978623819293,
      "loss": 1.404,
      "step": 37433
    },
    {
      "epoch": 3.2206831282801343,
      "grad_norm": 2.40625,
      "learning_rate": 0.00028146546472544276,
      "loss": 1.2415,
      "step": 37434
    },
    {
      "epoch": 3.220769164587456,
      "grad_norm": 4.71875,
      "learning_rate": 0.0002814411438518634,
      "loss": 1.2895,
      "step": 37435
    },
    {
      "epoch": 3.2208552008947775,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002814168236175262,
      "loss": 1.3067,
      "step": 37436
    },
    {
      "epoch": 3.220941237202099,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002813925040225025,
      "loss": 1.8063,
      "step": 37437
    },
    {
      "epoch": 3.2210272735094208,
      "grad_norm": 1.796875,
      "learning_rate": 0.00028136818506686303,
      "loss": 1.4844,
      "step": 37438
    },
    {
      "epoch": 3.221113309816743,
      "grad_norm": 1.03125,
      "learning_rate": 0.00028134386675067924,
      "loss": 1.1323,
      "step": 37439
    },
    {
      "epoch": 3.2211993461240644,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002813195490740217,
      "loss": 1.3984,
      "step": 37440
    },
    {
      "epoch": 3.221285382431386,
      "grad_norm": 1.3125,
      "learning_rate": 0.00028129523203696216,
      "loss": 1.4971,
      "step": 37441
    },
    {
      "epoch": 3.2213714187387077,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00028127091563957173,
      "loss": 0.7858,
      "step": 37442
    },
    {
      "epoch": 3.2214574550460293,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002812465998819212,
      "loss": 1.3953,
      "step": 37443
    },
    {
      "epoch": 3.221543491353351,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00028122228476408197,
      "loss": 1.5977,
      "step": 37444
    },
    {
      "epoch": 3.221629527660673,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00028119797028612484,
      "loss": 0.7131,
      "step": 37445
    },
    {
      "epoch": 3.2217155639679946,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000281173656448121,
      "loss": 2.0616,
      "step": 37446
    },
    {
      "epoch": 3.2218016002753163,
      "grad_norm": 1.203125,
      "learning_rate": 0.000281149343250142,
      "loss": 1.7931,
      "step": 37447
    },
    {
      "epoch": 3.221887636582638,
      "grad_norm": 0.96875,
      "learning_rate": 0.00028112503069225827,
      "loss": 1.0035,
      "step": 37448
    },
    {
      "epoch": 3.2219736728899595,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002811007187745413,
      "loss": 1.896,
      "step": 37449
    },
    {
      "epoch": 3.222059709197281,
      "grad_norm": 13.625,
      "learning_rate": 0.00028107640749706236,
      "loss": 0.8088,
      "step": 37450
    },
    {
      "epoch": 3.2221457455046028,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002810520968598922,
      "loss": 1.019,
      "step": 37451
    },
    {
      "epoch": 3.222231781811925,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000281027786863102,
      "loss": 1.5461,
      "step": 37452
    },
    {
      "epoch": 3.2223178181192464,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002810034775067631,
      "loss": 0.9281,
      "step": 37453
    },
    {
      "epoch": 3.222403854426568,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00028097916879094633,
      "loss": 1.2667,
      "step": 37454
    },
    {
      "epoch": 3.2224898907338897,
      "grad_norm": 1.15625,
      "learning_rate": 0.00028095486071572307,
      "loss": 1.9574,
      "step": 37455
    },
    {
      "epoch": 3.2225759270412113,
      "grad_norm": 4.09375,
      "learning_rate": 0.00028093055328116403,
      "loss": 1.36,
      "step": 37456
    },
    {
      "epoch": 3.222661963348533,
      "grad_norm": 23.625,
      "learning_rate": 0.00028090624648734056,
      "loss": 1.7084,
      "step": 37457
    },
    {
      "epoch": 3.2227479996558546,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002808819403343238,
      "loss": 1.2544,
      "step": 37458
    },
    {
      "epoch": 3.2228340359631766,
      "grad_norm": 7.15625,
      "learning_rate": 0.00028085763482218465,
      "loss": 1.0107,
      "step": 37459
    },
    {
      "epoch": 3.2229200722704983,
      "grad_norm": 1.140625,
      "learning_rate": 0.00028083332995099423,
      "loss": 1.0232,
      "step": 37460
    },
    {
      "epoch": 3.22300610857782,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00028080902572082394,
      "loss": 0.9593,
      "step": 37461
    },
    {
      "epoch": 3.2230921448851415,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00028078472213174444,
      "loss": 0.7117,
      "step": 37462
    },
    {
      "epoch": 3.223178181192463,
      "grad_norm": 1.109375,
      "learning_rate": 0.00028076041918382693,
      "loss": 0.9896,
      "step": 37463
    },
    {
      "epoch": 3.2232642174997848,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002807361168771428,
      "loss": 0.8377,
      "step": 37464
    },
    {
      "epoch": 3.223350253807107,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002807118152117626,
      "loss": 1.5013,
      "step": 37465
    },
    {
      "epoch": 3.2234362901144284,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00028068751418775786,
      "loss": 0.8851,
      "step": 37466
    },
    {
      "epoch": 3.22352232642175,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00028066321380519957,
      "loss": 0.7982,
      "step": 37467
    },
    {
      "epoch": 3.2236083627290717,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002806389140641585,
      "loss": 0.4539,
      "step": 37468
    },
    {
      "epoch": 3.2236943990363933,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002806146149647062,
      "loss": 0.9104,
      "step": 37469
    },
    {
      "epoch": 3.223780435343715,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00028059031650691323,
      "loss": 1.2123,
      "step": 37470
    },
    {
      "epoch": 3.2238664716510366,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00028056601869085095,
      "loss": 1.0287,
      "step": 37471
    },
    {
      "epoch": 3.2239525079583586,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002805417215165906,
      "loss": 0.683,
      "step": 37472
    },
    {
      "epoch": 3.2240385442656803,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002805174249842028,
      "loss": 1.1827,
      "step": 37473
    },
    {
      "epoch": 3.224124580573002,
      "grad_norm": 3.015625,
      "learning_rate": 0.00028049312909375883,
      "loss": 1.0528,
      "step": 37474
    },
    {
      "epoch": 3.2242106168803235,
      "grad_norm": 2.140625,
      "learning_rate": 0.00028046883384533,
      "loss": 0.8336,
      "step": 37475
    },
    {
      "epoch": 3.224296653187645,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00028044453923898693,
      "loss": 0.6535,
      "step": 37476
    },
    {
      "epoch": 3.2243826894949668,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002804202452748009,
      "loss": 0.7575,
      "step": 37477
    },
    {
      "epoch": 3.2244687258022884,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002803959519528429,
      "loss": 0.7187,
      "step": 37478
    },
    {
      "epoch": 3.2245547621096105,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002803716592731841,
      "loss": 0.8529,
      "step": 37479
    },
    {
      "epoch": 3.224640798416932,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002803473672358956,
      "loss": 1.0881,
      "step": 37480
    },
    {
      "epoch": 3.2247268347242537,
      "grad_norm": 1.828125,
      "learning_rate": 0.00028032307584104814,
      "loss": 1.0964,
      "step": 37481
    },
    {
      "epoch": 3.2248128710315753,
      "grad_norm": 1.171875,
      "learning_rate": 0.00028029878508871296,
      "loss": 1.505,
      "step": 37482
    },
    {
      "epoch": 3.224898907338897,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002802744949789614,
      "loss": 0.7969,
      "step": 37483
    },
    {
      "epoch": 3.2249849436462186,
      "grad_norm": 1.0625,
      "learning_rate": 0.00028025020551186394,
      "loss": 1.8841,
      "step": 37484
    },
    {
      "epoch": 3.22507097995354,
      "grad_norm": 1.328125,
      "learning_rate": 0.0002802259166874921,
      "loss": 1.527,
      "step": 37485
    },
    {
      "epoch": 3.2251570162608623,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028020162850591647,
      "loss": 0.975,
      "step": 37486
    },
    {
      "epoch": 3.225243052568184,
      "grad_norm": 2.03125,
      "learning_rate": 0.00028017734096720835,
      "loss": 1.2059,
      "step": 37487
    },
    {
      "epoch": 3.2253290888755055,
      "grad_norm": 1.375,
      "learning_rate": 0.000280153054071439,
      "loss": 0.7349,
      "step": 37488
    },
    {
      "epoch": 3.225415125182827,
      "grad_norm": 119.5,
      "learning_rate": 0.0002801287678186789,
      "loss": 2.2318,
      "step": 37489
    },
    {
      "epoch": 3.2255011614901488,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002801044822089995,
      "loss": 0.9185,
      "step": 37490
    },
    {
      "epoch": 3.2255871977974704,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028008019724247186,
      "loss": 0.9887,
      "step": 37491
    },
    {
      "epoch": 3.225673234104792,
      "grad_norm": 1.390625,
      "learning_rate": 0.0002800559129191667,
      "loss": 1.2791,
      "step": 37492
    },
    {
      "epoch": 3.225759270412114,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002800316292391552,
      "loss": 0.4624,
      "step": 37493
    },
    {
      "epoch": 3.2258453067194357,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00028000734620250864,
      "loss": 2.1569,
      "step": 37494
    },
    {
      "epoch": 3.2259313430267573,
      "grad_norm": 1.515625,
      "learning_rate": 0.00027998306380929756,
      "loss": 1.696,
      "step": 37495
    },
    {
      "epoch": 3.226017379334079,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002799587820595935,
      "loss": 1.8733,
      "step": 37496
    },
    {
      "epoch": 3.2261034156414006,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00027993450095346693,
      "loss": 0.7823,
      "step": 37497
    },
    {
      "epoch": 3.226189451948722,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00027991022049098915,
      "loss": 1.506,
      "step": 37498
    },
    {
      "epoch": 3.226275488256044,
      "grad_norm": 0.984375,
      "learning_rate": 0.00027988594067223136,
      "loss": 1.058,
      "step": 37499
    },
    {
      "epoch": 3.226361524563366,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002798616614972642,
      "loss": 1.4719,
      "step": 37500
    },
    {
      "epoch": 3.2264475608706875,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002798373829661589,
      "loss": 0.8514,
      "step": 37501
    },
    {
      "epoch": 3.226533597178009,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002798131050789866,
      "loss": 0.8821,
      "step": 37502
    },
    {
      "epoch": 3.2266196334853308,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000279788827835818,
      "loss": 0.9622,
      "step": 37503
    },
    {
      "epoch": 3.2267056697926524,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002797645512367242,
      "loss": 1.2866,
      "step": 37504
    },
    {
      "epoch": 3.226791706099974,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002797402752817765,
      "loss": 1.8625,
      "step": 37505
    },
    {
      "epoch": 3.226877742407296,
      "grad_norm": 1.125,
      "learning_rate": 0.00027971599997104535,
      "loss": 0.6902,
      "step": 37506
    },
    {
      "epoch": 3.2269637787146177,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002796917253046022,
      "loss": 1.2011,
      "step": 37507
    },
    {
      "epoch": 3.2270498150219393,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000279667451282518,
      "loss": 0.9944,
      "step": 37508
    },
    {
      "epoch": 3.227135851329261,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00027964317790486347,
      "loss": 1.1987,
      "step": 37509
    },
    {
      "epoch": 3.2272218876365826,
      "grad_norm": 1.015625,
      "learning_rate": 0.00027961890517171,
      "loss": 1.397,
      "step": 37510
    },
    {
      "epoch": 3.227307923943904,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00027959463308312816,
      "loss": 0.8331,
      "step": 37511
    },
    {
      "epoch": 3.227393960251226,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00027957036163918917,
      "loss": 0.7555,
      "step": 37512
    },
    {
      "epoch": 3.227479996558548,
      "grad_norm": 82.0,
      "learning_rate": 0.00027954609083996417,
      "loss": 0.8908,
      "step": 37513
    },
    {
      "epoch": 3.2275660328658695,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002795218206855237,
      "loss": 0.9482,
      "step": 37514
    },
    {
      "epoch": 3.227652069173191,
      "grad_norm": 114.0,
      "learning_rate": 0.0002794975511759393,
      "loss": 0.9976,
      "step": 37515
    },
    {
      "epoch": 3.2277381054805128,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002794732823112814,
      "loss": 0.8076,
      "step": 37516
    },
    {
      "epoch": 3.2278241417878344,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00027944901409162145,
      "loss": 1.3589,
      "step": 37517
    },
    {
      "epoch": 3.227910178095156,
      "grad_norm": 1.328125,
      "learning_rate": 0.00027942474651703033,
      "loss": 2.1838,
      "step": 37518
    },
    {
      "epoch": 3.227996214402478,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00027940047958757876,
      "loss": 1.7397,
      "step": 37519
    },
    {
      "epoch": 3.2280822507097997,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00027937621330333785,
      "loss": 0.8975,
      "step": 37520
    },
    {
      "epoch": 3.2281682870171213,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002793519476643789,
      "loss": 1.3523,
      "step": 37521
    },
    {
      "epoch": 3.228254323324443,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00027932768267077224,
      "loss": 1.3706,
      "step": 37522
    },
    {
      "epoch": 3.2283403596317646,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002793034183225893,
      "loss": 0.7859,
      "step": 37523
    },
    {
      "epoch": 3.228426395939086,
      "grad_norm": 2.71875,
      "learning_rate": 0.0002792791546199011,
      "loss": 1.0293,
      "step": 37524
    },
    {
      "epoch": 3.228512432246408,
      "grad_norm": 1.125,
      "learning_rate": 0.0002792548915627783,
      "loss": 1.7283,
      "step": 37525
    },
    {
      "epoch": 3.22859846855373,
      "grad_norm": 6.9375,
      "learning_rate": 0.00027923062915129215,
      "loss": 1.5522,
      "step": 37526
    },
    {
      "epoch": 3.2286845048610515,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002792063673855133,
      "loss": 0.6124,
      "step": 37527
    },
    {
      "epoch": 3.228770541168373,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002791821062655129,
      "loss": 1.7927,
      "step": 37528
    },
    {
      "epoch": 3.2288565774756948,
      "grad_norm": 5.53125,
      "learning_rate": 0.000279157845791362,
      "loss": 1.274,
      "step": 37529
    },
    {
      "epoch": 3.2289426137830164,
      "grad_norm": 2.296875,
      "learning_rate": 0.0002791335859631313,
      "loss": 1.9277,
      "step": 37530
    },
    {
      "epoch": 3.229028650090338,
      "grad_norm": 1.015625,
      "learning_rate": 0.00027910932678089194,
      "loss": 1.3682,
      "step": 37531
    },
    {
      "epoch": 3.2291146863976596,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00027908506824471504,
      "loss": 1.5992,
      "step": 37532
    },
    {
      "epoch": 3.2292007227049817,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00027906081035467114,
      "loss": 0.6807,
      "step": 37533
    },
    {
      "epoch": 3.2292867590123033,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002790365531108313,
      "loss": 1.1732,
      "step": 37534
    },
    {
      "epoch": 3.229372795319625,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002790122965132669,
      "loss": 1.4383,
      "step": 37535
    },
    {
      "epoch": 3.2294588316269466,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002789880405620483,
      "loss": 1.3156,
      "step": 37536
    },
    {
      "epoch": 3.229544867934268,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002789637852572469,
      "loss": 0.8882,
      "step": 37537
    },
    {
      "epoch": 3.22963090424159,
      "grad_norm": 0.734375,
      "learning_rate": 0.00027893953059893325,
      "loss": 0.3782,
      "step": 37538
    },
    {
      "epoch": 3.2297169405489115,
      "grad_norm": 1.953125,
      "learning_rate": 0.0002789152765871784,
      "loss": 1.0765,
      "step": 37539
    },
    {
      "epoch": 3.2298029768562335,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0002788910232220537,
      "loss": 1.9483,
      "step": 37540
    },
    {
      "epoch": 3.229889013163555,
      "grad_norm": 0.828125,
      "learning_rate": 0.00027886677050362944,
      "loss": 0.8087,
      "step": 37541
    },
    {
      "epoch": 3.2299750494708768,
      "grad_norm": 1.1875,
      "learning_rate": 0.00027884251843197694,
      "loss": 1.0699,
      "step": 37542
    },
    {
      "epoch": 3.2300610857781984,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0002788182670071672,
      "loss": 1.0033,
      "step": 37543
    },
    {
      "epoch": 3.23014712208552,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002787940162292709,
      "loss": 0.5909,
      "step": 37544
    },
    {
      "epoch": 3.2302331583928416,
      "grad_norm": 1.171875,
      "learning_rate": 0.000278769766098359,
      "loss": 0.874,
      "step": 37545
    },
    {
      "epoch": 3.2303191947001633,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00027874551661450275,
      "loss": 1.5401,
      "step": 37546
    },
    {
      "epoch": 3.2304052310074853,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002787212677777726,
      "loss": 0.6258,
      "step": 37547
    },
    {
      "epoch": 3.230491267314807,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002786970195882398,
      "loss": 0.9769,
      "step": 37548
    },
    {
      "epoch": 3.2305773036221286,
      "grad_norm": 2.25,
      "learning_rate": 0.00027867277204597526,
      "loss": 1.0542,
      "step": 37549
    },
    {
      "epoch": 3.23066333992945,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00027864852515104965,
      "loss": 1.8232,
      "step": 37550
    },
    {
      "epoch": 3.230749376236772,
      "grad_norm": 1.5,
      "learning_rate": 0.00027862427890353405,
      "loss": 0.8156,
      "step": 37551
    },
    {
      "epoch": 3.2308354125440935,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002786000333034994,
      "loss": 1.9195,
      "step": 37552
    },
    {
      "epoch": 3.230921448851415,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002785757883510166,
      "loss": 1.0048,
      "step": 37553
    },
    {
      "epoch": 3.231007485158737,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00027855154404615674,
      "loss": 1.0421,
      "step": 37554
    },
    {
      "epoch": 3.2310935214660588,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002785273003889903,
      "loss": 0.7998,
      "step": 37555
    },
    {
      "epoch": 3.2311795577733804,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002785030573795887,
      "loss": 1.5786,
      "step": 37556
    },
    {
      "epoch": 3.231265594080702,
      "grad_norm": 1.359375,
      "learning_rate": 0.00027847881501802234,
      "loss": 1.0182,
      "step": 37557
    },
    {
      "epoch": 3.2313516303880236,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002784545733043624,
      "loss": 0.974,
      "step": 37558
    },
    {
      "epoch": 3.2314376666953453,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002784303322386799,
      "loss": 0.8166,
      "step": 37559
    },
    {
      "epoch": 3.231523703002667,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002784060918210454,
      "loss": 1.1663,
      "step": 37560
    },
    {
      "epoch": 3.231609739309989,
      "grad_norm": 1.078125,
      "learning_rate": 0.00027838185205153,
      "loss": 1.3571,
      "step": 37561
    },
    {
      "epoch": 3.2316957756173106,
      "grad_norm": 0.828125,
      "learning_rate": 0.00027835761293020476,
      "loss": 0.8542,
      "step": 37562
    },
    {
      "epoch": 3.231781811924632,
      "grad_norm": 40.0,
      "learning_rate": 0.00027833337445714023,
      "loss": 1.842,
      "step": 37563
    },
    {
      "epoch": 3.231867848231954,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002783091366324074,
      "loss": 0.9959,
      "step": 37564
    },
    {
      "epoch": 3.2319538845392755,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002782848994560775,
      "loss": 0.685,
      "step": 37565
    },
    {
      "epoch": 3.232039920846597,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00027826066292822095,
      "loss": 1.6712,
      "step": 37566
    },
    {
      "epoch": 3.232125957153919,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00027823642704890894,
      "loss": 1.4163,
      "step": 37567
    },
    {
      "epoch": 3.2322119934612408,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002782121918182121,
      "loss": 1.0579,
      "step": 37568
    },
    {
      "epoch": 3.2322980297685624,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002781879572362015,
      "loss": 1.1251,
      "step": 37569
    },
    {
      "epoch": 3.232384066075884,
      "grad_norm": 1.125,
      "learning_rate": 0.0002781637233029482,
      "loss": 1.1697,
      "step": 37570
    },
    {
      "epoch": 3.2324701023832056,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002781394900185226,
      "loss": 1.3912,
      "step": 37571
    },
    {
      "epoch": 3.2325561386905273,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00027811525738299594,
      "loss": 1.8074,
      "step": 37572
    },
    {
      "epoch": 3.232642174997849,
      "grad_norm": 32.5,
      "learning_rate": 0.000278091025396439,
      "loss": 0.8817,
      "step": 37573
    },
    {
      "epoch": 3.232728211305171,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00027806679405892266,
      "loss": 0.7721,
      "step": 37574
    },
    {
      "epoch": 3.2328142476124926,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002780425633705177,
      "loss": 1.3951,
      "step": 37575
    },
    {
      "epoch": 3.232900283919814,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002780183333312953,
      "loss": 1.9306,
      "step": 37576
    },
    {
      "epoch": 3.232986320227136,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00027799410394132586,
      "loss": 0.8525,
      "step": 37577
    },
    {
      "epoch": 3.2330723565344575,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002779698752006805,
      "loss": 1.0959,
      "step": 37578
    },
    {
      "epoch": 3.233158392841779,
      "grad_norm": 7.59375,
      "learning_rate": 0.00027794564710943026,
      "loss": 1.7823,
      "step": 37579
    },
    {
      "epoch": 3.233244429149101,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00027792141966764567,
      "loss": 1.6985,
      "step": 37580
    },
    {
      "epoch": 3.2333304654564228,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002778971928753979,
      "loss": 1.72,
      "step": 37581
    },
    {
      "epoch": 3.2334165017637444,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002778729667327574,
      "loss": 1.2315,
      "step": 37582
    },
    {
      "epoch": 3.233502538071066,
      "grad_norm": 1.125,
      "learning_rate": 0.0002778487412397953,
      "loss": 1.0483,
      "step": 37583
    },
    {
      "epoch": 3.2335885743783876,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002778245163965827,
      "loss": 1.5235,
      "step": 37584
    },
    {
      "epoch": 3.2336746106857093,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002778002922031899,
      "loss": 1.2933,
      "step": 37585
    },
    {
      "epoch": 3.233760646993031,
      "grad_norm": 2.578125,
      "learning_rate": 0.0002777760686596881,
      "loss": 1.4553,
      "step": 37586
    },
    {
      "epoch": 3.233846683300353,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002777518457661482,
      "loss": 0.9781,
      "step": 37587
    },
    {
      "epoch": 3.2339327196076746,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002777276235226406,
      "loss": 1.075,
      "step": 37588
    },
    {
      "epoch": 3.234018755914996,
      "grad_norm": 6.65625,
      "learning_rate": 0.00027770340192923677,
      "loss": 0.5911,
      "step": 37589
    },
    {
      "epoch": 3.234104792222318,
      "grad_norm": 23.5,
      "learning_rate": 0.0002776791809860071,
      "loss": 1.2666,
      "step": 37590
    },
    {
      "epoch": 3.2341908285296395,
      "grad_norm": 9.9375,
      "learning_rate": 0.00027765496069302265,
      "loss": 1.5822,
      "step": 37591
    },
    {
      "epoch": 3.234276864836961,
      "grad_norm": 1.25,
      "learning_rate": 0.00027763074105035434,
      "loss": 0.991,
      "step": 37592
    },
    {
      "epoch": 3.2343629011442827,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002776065220580726,
      "loss": 1.4792,
      "step": 37593
    },
    {
      "epoch": 3.2344489374516048,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00027758230371624867,
      "loss": 1.4807,
      "step": 37594
    },
    {
      "epoch": 3.2345349737589264,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00027755808602495334,
      "loss": 1.6344,
      "step": 37595
    },
    {
      "epoch": 3.234621010066248,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00027753386898425717,
      "loss": 1.0454,
      "step": 37596
    },
    {
      "epoch": 3.2347070463735696,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00027750965259423137,
      "loss": 1.5296,
      "step": 37597
    },
    {
      "epoch": 3.2347930826808913,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002774854368549464,
      "loss": 0.731,
      "step": 37598
    },
    {
      "epoch": 3.234879118988213,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002774612217664732,
      "loss": 0.6862,
      "step": 37599
    },
    {
      "epoch": 3.2349651552955345,
      "grad_norm": 1.125,
      "learning_rate": 0.0002774370073288829,
      "loss": 1.4352,
      "step": 37600
    },
    {
      "epoch": 3.2350511916028566,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002774127935422458,
      "loss": 0.6094,
      "step": 37601
    },
    {
      "epoch": 3.235137227910178,
      "grad_norm": 1.046875,
      "learning_rate": 0.00027738858040663307,
      "loss": 0.7281,
      "step": 37602
    },
    {
      "epoch": 3.2352232642175,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00027736436792211557,
      "loss": 1.1405,
      "step": 37603
    },
    {
      "epoch": 3.2353093005248215,
      "grad_norm": 11.6875,
      "learning_rate": 0.0002773401560887638,
      "loss": 0.7342,
      "step": 37604
    },
    {
      "epoch": 3.235395336832143,
      "grad_norm": 1.125,
      "learning_rate": 0.0002773159449066488,
      "loss": 1.3482,
      "step": 37605
    },
    {
      "epoch": 3.2354813731394647,
      "grad_norm": 1.03125,
      "learning_rate": 0.00027729173437584154,
      "loss": 0.9655,
      "step": 37606
    },
    {
      "epoch": 3.2355674094467863,
      "grad_norm": 1.375,
      "learning_rate": 0.00027726752449641243,
      "loss": 1.6839,
      "step": 37607
    },
    {
      "epoch": 3.2356534457541084,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002772433152684327,
      "loss": 0.8278,
      "step": 37608
    },
    {
      "epoch": 3.23573948206143,
      "grad_norm": 1.4375,
      "learning_rate": 0.00027721910669197274,
      "loss": 1.3152,
      "step": 37609
    },
    {
      "epoch": 3.2358255183687517,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002771948987671036,
      "loss": 0.7968,
      "step": 37610
    },
    {
      "epoch": 3.2359115546760733,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00027717069149389625,
      "loss": 1.2975,
      "step": 37611
    },
    {
      "epoch": 3.235997590983395,
      "grad_norm": 3.625,
      "learning_rate": 0.000277146484872421,
      "loss": 1.7079,
      "step": 37612
    },
    {
      "epoch": 3.2360836272907165,
      "grad_norm": 2.265625,
      "learning_rate": 0.000277122278902749,
      "loss": 1.3448,
      "step": 37613
    },
    {
      "epoch": 3.236169663598038,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002770980735849512,
      "loss": 0.9223,
      "step": 37614
    },
    {
      "epoch": 3.23625569990536,
      "grad_norm": 0.984375,
      "learning_rate": 0.00027707386891909805,
      "loss": 1.5193,
      "step": 37615
    },
    {
      "epoch": 3.236341736212682,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002770496649052603,
      "loss": 1.2611,
      "step": 37616
    },
    {
      "epoch": 3.2364277725200035,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002770254615435093,
      "loss": 1.3741,
      "step": 37617
    },
    {
      "epoch": 3.236513808827325,
      "grad_norm": 0.890625,
      "learning_rate": 0.00027700125883391516,
      "loss": 0.7136,
      "step": 37618
    },
    {
      "epoch": 3.2365998451346467,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000276977056776549,
      "loss": 0.9696,
      "step": 37619
    },
    {
      "epoch": 3.2366858814419683,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002769528553714817,
      "loss": 1.0079,
      "step": 37620
    },
    {
      "epoch": 3.2367719177492904,
      "grad_norm": 2.375,
      "learning_rate": 0.0002769286546187838,
      "loss": 1.0354,
      "step": 37621
    },
    {
      "epoch": 3.236857954056612,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002769044545185264,
      "loss": 0.872,
      "step": 37622
    },
    {
      "epoch": 3.2369439903639337,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00027688025507077984,
      "loss": 0.9419,
      "step": 37623
    },
    {
      "epoch": 3.2370300266712553,
      "grad_norm": 0.765625,
      "learning_rate": 0.00027685605627561516,
      "loss": 0.7177,
      "step": 37624
    },
    {
      "epoch": 3.237116062978577,
      "grad_norm": 2.0,
      "learning_rate": 0.00027683185813310316,
      "loss": 1.2819,
      "step": 37625
    },
    {
      "epoch": 3.2372020992858985,
      "grad_norm": 1.3125,
      "learning_rate": 0.00027680766064331453,
      "loss": 0.714,
      "step": 37626
    },
    {
      "epoch": 3.23728813559322,
      "grad_norm": 0.90625,
      "learning_rate": 0.00027678346380632024,
      "loss": 1.1467,
      "step": 37627
    },
    {
      "epoch": 3.237374171900542,
      "grad_norm": 1.03125,
      "learning_rate": 0.00027675926762219075,
      "loss": 0.6176,
      "step": 37628
    },
    {
      "epoch": 3.237460208207864,
      "grad_norm": 1.109375,
      "learning_rate": 0.000276735072090997,
      "loss": 1.1774,
      "step": 37629
    },
    {
      "epoch": 3.2375462445151855,
      "grad_norm": 3.765625,
      "learning_rate": 0.0002767108772128099,
      "loss": 0.8199,
      "step": 37630
    },
    {
      "epoch": 3.237632280822507,
      "grad_norm": 13.8125,
      "learning_rate": 0.0002766866829876998,
      "loss": 1.272,
      "step": 37631
    },
    {
      "epoch": 3.2377183171298287,
      "grad_norm": 14.0,
      "learning_rate": 0.0002766624894157379,
      "loss": 1.3655,
      "step": 37632
    },
    {
      "epoch": 3.2378043534371503,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002766382964969948,
      "loss": 0.905,
      "step": 37633
    },
    {
      "epoch": 3.237890389744472,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00027661410423154115,
      "loss": 0.9455,
      "step": 37634
    },
    {
      "epoch": 3.237976426051794,
      "grad_norm": 3.4375,
      "learning_rate": 0.0002765899126194478,
      "loss": 1.3294,
      "step": 37635
    },
    {
      "epoch": 3.2380624623591157,
      "grad_norm": 5.96875,
      "learning_rate": 0.00027656572166078565,
      "loss": 1.4545,
      "step": 37636
    },
    {
      "epoch": 3.2381484986664373,
      "grad_norm": 49.0,
      "learning_rate": 0.0002765415313556251,
      "loss": 1.5134,
      "step": 37637
    },
    {
      "epoch": 3.238234534973759,
      "grad_norm": 1.625,
      "learning_rate": 0.0002765173417040374,
      "loss": 1.9773,
      "step": 37638
    },
    {
      "epoch": 3.2383205712810805,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002764931527060928,
      "loss": 1.443,
      "step": 37639
    },
    {
      "epoch": 3.238406607588402,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00027646896436186216,
      "loss": 1.1553,
      "step": 37640
    },
    {
      "epoch": 3.238492643895724,
      "grad_norm": 1.109375,
      "learning_rate": 0.00027644477667141656,
      "loss": 1.0831,
      "step": 37641
    },
    {
      "epoch": 3.238578680203046,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002764205896348263,
      "loss": 1.1514,
      "step": 37642
    },
    {
      "epoch": 3.2386647165103675,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00027639640325216244,
      "loss": 0.5665,
      "step": 37643
    },
    {
      "epoch": 3.238750752817689,
      "grad_norm": 68.5,
      "learning_rate": 0.0002763722175234957,
      "loss": 1.7927,
      "step": 37644
    },
    {
      "epoch": 3.2388367891250107,
      "grad_norm": 12.5,
      "learning_rate": 0.0002763480324488965,
      "loss": 0.8463,
      "step": 37645
    },
    {
      "epoch": 3.2389228254323323,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002763238480284359,
      "loss": 1.1131,
      "step": 37646
    },
    {
      "epoch": 3.239008861739654,
      "grad_norm": 1.125,
      "learning_rate": 0.0002762996642621847,
      "loss": 0.8434,
      "step": 37647
    },
    {
      "epoch": 3.239094898046976,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002762754811502133,
      "loss": 1.0622,
      "step": 37648
    },
    {
      "epoch": 3.2391809343542977,
      "grad_norm": 1.5,
      "learning_rate": 0.0002762512986925928,
      "loss": 0.7158,
      "step": 37649
    },
    {
      "epoch": 3.2392669706616193,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0002762271168893935,
      "loss": 1.2374,
      "step": 37650
    },
    {
      "epoch": 3.239353006968941,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002762029357406864,
      "loss": 1.2431,
      "step": 37651
    },
    {
      "epoch": 3.2394390432762625,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00027617875524654236,
      "loss": 1.0779,
      "step": 37652
    },
    {
      "epoch": 3.239525079583584,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002761545754070318,
      "loss": 0.9947,
      "step": 37653
    },
    {
      "epoch": 3.239611115890906,
      "grad_norm": 1.03125,
      "learning_rate": 0.00027613039622222547,
      "loss": 1.2537,
      "step": 37654
    },
    {
      "epoch": 3.239697152198228,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002761062176921945,
      "loss": 0.6099,
      "step": 37655
    },
    {
      "epoch": 3.2397831885055495,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000276082039817009,
      "loss": 1.4068,
      "step": 37656
    },
    {
      "epoch": 3.239869224812871,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00027605786259674,
      "loss": 0.8699,
      "step": 37657
    },
    {
      "epoch": 3.2399552611201927,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002760336860314584,
      "loss": 1.0669,
      "step": 37658
    },
    {
      "epoch": 3.2400412974275143,
      "grad_norm": 1.484375,
      "learning_rate": 0.00027600951012123455,
      "loss": 1.3777,
      "step": 37659
    },
    {
      "epoch": 3.240127333734836,
      "grad_norm": 1.296875,
      "learning_rate": 0.00027598533486613923,
      "loss": 1.3963,
      "step": 37660
    },
    {
      "epoch": 3.2402133700421576,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002759611602662435,
      "loss": 0.9871,
      "step": 37661
    },
    {
      "epoch": 3.2402994063494797,
      "grad_norm": 46.5,
      "learning_rate": 0.00027593698632161743,
      "loss": 1.3687,
      "step": 37662
    },
    {
      "epoch": 3.2403854426568013,
      "grad_norm": 1.375,
      "learning_rate": 0.0002759128130323325,
      "loss": 1.7951,
      "step": 37663
    },
    {
      "epoch": 3.240471478964123,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002758886403984587,
      "loss": 1.9389,
      "step": 37664
    },
    {
      "epoch": 3.2405575152714445,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002758644684200672,
      "loss": 0.8581,
      "step": 37665
    },
    {
      "epoch": 3.240643551578766,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002758402970972286,
      "loss": 0.757,
      "step": 37666
    },
    {
      "epoch": 3.240729587886088,
      "grad_norm": 22.875,
      "learning_rate": 0.0002758161264300134,
      "loss": 1.1974,
      "step": 37667
    },
    {
      "epoch": 3.2408156241934094,
      "grad_norm": 1.078125,
      "learning_rate": 0.00027579195641849264,
      "loss": 0.999,
      "step": 37668
    },
    {
      "epoch": 3.2409016605007315,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002757677870627366,
      "loss": 0.8865,
      "step": 37669
    },
    {
      "epoch": 3.240987696808053,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00027574361836281614,
      "loss": 0.9023,
      "step": 37670
    },
    {
      "epoch": 3.2410737331153747,
      "grad_norm": 1.09375,
      "learning_rate": 0.00027571945031880217,
      "loss": 0.9223,
      "step": 37671
    },
    {
      "epoch": 3.2411597694226963,
      "grad_norm": 73.5,
      "learning_rate": 0.0002756952829307651,
      "loss": 1.093,
      "step": 37672
    },
    {
      "epoch": 3.241245805730018,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00027567111619877555,
      "loss": 0.9816,
      "step": 37673
    },
    {
      "epoch": 3.2413318420373396,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002756469501229046,
      "loss": 1.3203,
      "step": 37674
    },
    {
      "epoch": 3.241417878344661,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00027562278470322254,
      "loss": 0.5072,
      "step": 37675
    },
    {
      "epoch": 3.2415039146519833,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002755986199398002,
      "loss": 0.7096,
      "step": 37676
    },
    {
      "epoch": 3.241589950959305,
      "grad_norm": 6.0,
      "learning_rate": 0.00027557445583270833,
      "loss": 1.4785,
      "step": 37677
    },
    {
      "epoch": 3.2416759872666265,
      "grad_norm": 196.0,
      "learning_rate": 0.00027555029238201747,
      "loss": 1.6494,
      "step": 37678
    },
    {
      "epoch": 3.241762023573948,
      "grad_norm": 1.125,
      "learning_rate": 0.0002755261295877985,
      "loss": 0.6475,
      "step": 37679
    },
    {
      "epoch": 3.24184805988127,
      "grad_norm": 12.5,
      "learning_rate": 0.0002755019674501217,
      "loss": 1.0853,
      "step": 37680
    },
    {
      "epoch": 3.2419340961885914,
      "grad_norm": 10.375,
      "learning_rate": 0.00027547780596905803,
      "loss": 0.84,
      "step": 37681
    },
    {
      "epoch": 3.2420201324959135,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002754536451446783,
      "loss": 0.954,
      "step": 37682
    },
    {
      "epoch": 3.242106168803235,
      "grad_norm": 1.359375,
      "learning_rate": 0.00027542948497705276,
      "loss": 1.13,
      "step": 37683
    },
    {
      "epoch": 3.2421922051105567,
      "grad_norm": 9.75,
      "learning_rate": 0.0002754053254662524,
      "loss": 1.3051,
      "step": 37684
    },
    {
      "epoch": 3.2422782414178783,
      "grad_norm": 1.828125,
      "learning_rate": 0.0002753811666123479,
      "loss": 1.5812,
      "step": 37685
    },
    {
      "epoch": 3.2423642777252,
      "grad_norm": 2.625,
      "learning_rate": 0.00027535700841540954,
      "loss": 1.077,
      "step": 37686
    },
    {
      "epoch": 3.2424503140325216,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00027533285087550833,
      "loss": 1.4984,
      "step": 37687
    },
    {
      "epoch": 3.242536350339843,
      "grad_norm": 10.0,
      "learning_rate": 0.00027530869399271506,
      "loss": 1.3026,
      "step": 37688
    },
    {
      "epoch": 3.2426223866471653,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00027528453776709994,
      "loss": 1.2152,
      "step": 37689
    },
    {
      "epoch": 3.242708422954487,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00027526038219873397,
      "loss": 0.8475,
      "step": 37690
    },
    {
      "epoch": 3.2427944592618085,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002752362272876875,
      "loss": 1.3769,
      "step": 37691
    },
    {
      "epoch": 3.24288049556913,
      "grad_norm": 3.84375,
      "learning_rate": 0.0002752120730340314,
      "loss": 1.052,
      "step": 37692
    },
    {
      "epoch": 3.242966531876452,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002751879194378365,
      "loss": 1.3694,
      "step": 37693
    },
    {
      "epoch": 3.2430525681837734,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002751637664991729,
      "loss": 0.9172,
      "step": 37694
    },
    {
      "epoch": 3.2431386044910955,
      "grad_norm": 211.0,
      "learning_rate": 0.00027513961421811163,
      "loss": 1.361,
      "step": 37695
    },
    {
      "epoch": 3.243224640798417,
      "grad_norm": 12.5625,
      "learning_rate": 0.0002751154625947234,
      "loss": 0.9728,
      "step": 37696
    },
    {
      "epoch": 3.2433106771057387,
      "grad_norm": 1.1875,
      "learning_rate": 0.00027509131162907854,
      "loss": 1.7653,
      "step": 37697
    },
    {
      "epoch": 3.2433967134130604,
      "grad_norm": 0.921875,
      "learning_rate": 0.00027506716132124785,
      "loss": 1.0889,
      "step": 37698
    },
    {
      "epoch": 3.243482749720382,
      "grad_norm": 1.28125,
      "learning_rate": 0.000275043011671302,
      "loss": 1.433,
      "step": 37699
    },
    {
      "epoch": 3.2435687860277036,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00027501886267931153,
      "loss": 1.9564,
      "step": 37700
    },
    {
      "epoch": 3.2436548223350252,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002749947143453474,
      "loss": 1.0729,
      "step": 37701
    },
    {
      "epoch": 3.2437408586423473,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0002749705666694797,
      "loss": 1.0136,
      "step": 37702
    },
    {
      "epoch": 3.243826894949669,
      "grad_norm": 1.71875,
      "learning_rate": 0.00027494641965177934,
      "loss": 1.3243,
      "step": 37703
    },
    {
      "epoch": 3.2439129312569905,
      "grad_norm": 144.0,
      "learning_rate": 0.0002749222732923171,
      "loss": 1.3761,
      "step": 37704
    },
    {
      "epoch": 3.243998967564312,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00027489812759116326,
      "loss": 0.6432,
      "step": 37705
    },
    {
      "epoch": 3.244085003871634,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002748739825483887,
      "loss": 0.4651,
      "step": 37706
    },
    {
      "epoch": 3.2441710401789554,
      "grad_norm": 1.078125,
      "learning_rate": 0.00027484983816406406,
      "loss": 0.8684,
      "step": 37707
    },
    {
      "epoch": 3.244257076486277,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00027482569443825974,
      "loss": 1.2043,
      "step": 37708
    },
    {
      "epoch": 3.244343112793599,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00027480155137104656,
      "loss": 0.7739,
      "step": 37709
    },
    {
      "epoch": 3.2444291491009207,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002747774089624949,
      "loss": 1.7518,
      "step": 37710
    },
    {
      "epoch": 3.2445151854082424,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002747532672126756,
      "loss": 0.7976,
      "step": 37711
    },
    {
      "epoch": 3.244601221715564,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002747291261216592,
      "loss": 1.0405,
      "step": 37712
    },
    {
      "epoch": 3.2446872580228856,
      "grad_norm": 1.0,
      "learning_rate": 0.00027470498568951625,
      "loss": 0.6941,
      "step": 37713
    },
    {
      "epoch": 3.2447732943302072,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002746808459163173,
      "loss": 1.0373,
      "step": 37714
    },
    {
      "epoch": 3.244859330637529,
      "grad_norm": 9.25,
      "learning_rate": 0.0002746567068021333,
      "loss": 1.5444,
      "step": 37715
    },
    {
      "epoch": 3.244945366944851,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00027463256834703444,
      "loss": 1.5903,
      "step": 37716
    },
    {
      "epoch": 3.2450314032521725,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00027460843055109144,
      "loss": 1.2036,
      "step": 37717
    },
    {
      "epoch": 3.245117439559494,
      "grad_norm": 532.0,
      "learning_rate": 0.0002745842934143751,
      "loss": 1.3308,
      "step": 37718
    },
    {
      "epoch": 3.245203475866816,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002745601569369558,
      "loss": 0.7896,
      "step": 37719
    },
    {
      "epoch": 3.2452895121741374,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002745360211189043,
      "loss": 0.9986,
      "step": 37720
    },
    {
      "epoch": 3.245375548481459,
      "grad_norm": 43.0,
      "learning_rate": 0.0002745118859602909,
      "loss": 0.6836,
      "step": 37721
    },
    {
      "epoch": 3.2454615847887807,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002744877514611863,
      "loss": 0.5259,
      "step": 37722
    },
    {
      "epoch": 3.2455476210961027,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002744636176216615,
      "loss": 1.1578,
      "step": 37723
    },
    {
      "epoch": 3.2456336574034244,
      "grad_norm": 1.1875,
      "learning_rate": 0.00027443948444178646,
      "loss": 1.1214,
      "step": 37724
    },
    {
      "epoch": 3.245719693710746,
      "grad_norm": 2.28125,
      "learning_rate": 0.00027441535192163216,
      "loss": 1.4028,
      "step": 37725
    },
    {
      "epoch": 3.2458057300180676,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00027439122006126914,
      "loss": 1.1471,
      "step": 37726
    },
    {
      "epoch": 3.2458917663253892,
      "grad_norm": 1.1875,
      "learning_rate": 0.00027436708886076784,
      "loss": 0.9238,
      "step": 37727
    },
    {
      "epoch": 3.245977802632711,
      "grad_norm": 81.5,
      "learning_rate": 0.00027434295832019887,
      "loss": 1.51,
      "step": 37728
    },
    {
      "epoch": 3.2460638389400325,
      "grad_norm": 1.640625,
      "learning_rate": 0.0002743188284396331,
      "loss": 1.7431,
      "step": 37729
    },
    {
      "epoch": 3.2461498752473545,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002742946992191406,
      "loss": 0.9126,
      "step": 37730
    },
    {
      "epoch": 3.246235911554676,
      "grad_norm": 9.8125,
      "learning_rate": 0.00027427057065879226,
      "loss": 1.2956,
      "step": 37731
    },
    {
      "epoch": 3.246321947861998,
      "grad_norm": 1.484375,
      "learning_rate": 0.00027424644275865875,
      "loss": 2.009,
      "step": 37732
    },
    {
      "epoch": 3.2464079841693194,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002742223155188103,
      "loss": 0.622,
      "step": 37733
    },
    {
      "epoch": 3.246494020476641,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00027419818893931793,
      "loss": 0.9695,
      "step": 37734
    },
    {
      "epoch": 3.2465800567839627,
      "grad_norm": 374.0,
      "learning_rate": 0.00027417406302025174,
      "loss": 1.433,
      "step": 37735
    },
    {
      "epoch": 3.2466660930912843,
      "grad_norm": 10.25,
      "learning_rate": 0.0002741499377616824,
      "loss": 1.3581,
      "step": 37736
    },
    {
      "epoch": 3.2467521293986064,
      "grad_norm": 1.515625,
      "learning_rate": 0.0002741258131636806,
      "loss": 1.3323,
      "step": 37737
    },
    {
      "epoch": 3.246838165705928,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00027410168922631687,
      "loss": 1.103,
      "step": 37738
    },
    {
      "epoch": 3.2469242020132496,
      "grad_norm": 1.234375,
      "learning_rate": 0.00027407756594966204,
      "loss": 1.2841,
      "step": 37739
    },
    {
      "epoch": 3.2470102383205712,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00027405344333378615,
      "loss": 1.03,
      "step": 37740
    },
    {
      "epoch": 3.247096274627893,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00027402932137875994,
      "loss": 0.9062,
      "step": 37741
    },
    {
      "epoch": 3.2471823109352145,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002740052000846542,
      "loss": 1.5021,
      "step": 37742
    },
    {
      "epoch": 3.2472683472425365,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00027398107945153917,
      "loss": 0.973,
      "step": 37743
    },
    {
      "epoch": 3.247354383549858,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002739569594794855,
      "loss": 0.9669,
      "step": 37744
    },
    {
      "epoch": 3.24744041985718,
      "grad_norm": 1.34375,
      "learning_rate": 0.00027393284016856396,
      "loss": 1.5198,
      "step": 37745
    },
    {
      "epoch": 3.2475264561645014,
      "grad_norm": 3.296875,
      "learning_rate": 0.00027390872151884473,
      "loss": 1.3297,
      "step": 37746
    },
    {
      "epoch": 3.247612492471823,
      "grad_norm": 1.203125,
      "learning_rate": 0.00027388460353039846,
      "loss": 0.8273,
      "step": 37747
    },
    {
      "epoch": 3.2476985287791447,
      "grad_norm": 1.078125,
      "learning_rate": 0.00027386048620329595,
      "loss": 1.469,
      "step": 37748
    },
    {
      "epoch": 3.2477845650864663,
      "grad_norm": 5.75,
      "learning_rate": 0.0002738363695376074,
      "loss": 1.0665,
      "step": 37749
    },
    {
      "epoch": 3.2478706013937884,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002738122535334036,
      "loss": 0.7688,
      "step": 37750
    },
    {
      "epoch": 3.24795663770111,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002737881381907548,
      "loss": 1.4913,
      "step": 37751
    },
    {
      "epoch": 3.2480426740084316,
      "grad_norm": 5.1875,
      "learning_rate": 0.0002737640235097317,
      "loss": 0.4919,
      "step": 37752
    },
    {
      "epoch": 3.2481287103157532,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002737399094904051,
      "loss": 1.2084,
      "step": 37753
    },
    {
      "epoch": 3.248214746623075,
      "grad_norm": 148.0,
      "learning_rate": 0.000273715796132845,
      "loss": 1.0726,
      "step": 37754
    },
    {
      "epoch": 3.2483007829303965,
      "grad_norm": 2.03125,
      "learning_rate": 0.0002736916834371221,
      "loss": 0.9225,
      "step": 37755
    },
    {
      "epoch": 3.2483868192377185,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00027366757140330733,
      "loss": 1.3877,
      "step": 37756
    },
    {
      "epoch": 3.24847285554504,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00027364346003147066,
      "loss": 0.9601,
      "step": 37757
    },
    {
      "epoch": 3.248558891852362,
      "grad_norm": 6.0,
      "learning_rate": 0.00027361934932168286,
      "loss": 0.7782,
      "step": 37758
    },
    {
      "epoch": 3.2486449281596834,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002735952392740146,
      "loss": 0.6973,
      "step": 37759
    },
    {
      "epoch": 3.248730964467005,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00027357112988853603,
      "loss": 1.3359,
      "step": 37760
    },
    {
      "epoch": 3.2488170007743267,
      "grad_norm": 1.328125,
      "learning_rate": 0.00027354702116531816,
      "loss": 1.4911,
      "step": 37761
    },
    {
      "epoch": 3.2489030370816483,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002735229131044309,
      "loss": 0.685,
      "step": 37762
    },
    {
      "epoch": 3.2489890733889704,
      "grad_norm": 158.0,
      "learning_rate": 0.0002734988057059451,
      "loss": 0.8186,
      "step": 37763
    },
    {
      "epoch": 3.249075109696292,
      "grad_norm": 1.40625,
      "learning_rate": 0.00027347469896993147,
      "loss": 2.1397,
      "step": 37764
    },
    {
      "epoch": 3.2491611460036136,
      "grad_norm": 0.78125,
      "learning_rate": 0.00027345059289646007,
      "loss": 0.6881,
      "step": 37765
    },
    {
      "epoch": 3.2492471823109352,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0002734264874856016,
      "loss": 1.2876,
      "step": 37766
    },
    {
      "epoch": 3.249333218618257,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002734023827374268,
      "loss": 1.1941,
      "step": 37767
    },
    {
      "epoch": 3.2494192549255785,
      "grad_norm": 14.75,
      "learning_rate": 0.00027337827865200583,
      "loss": 0.8594,
      "step": 37768
    },
    {
      "epoch": 3.2495052912329,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0002733541752294092,
      "loss": 1.3617,
      "step": 37769
    },
    {
      "epoch": 3.249591327540222,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002733300724697078,
      "loss": 1.1686,
      "step": 37770
    },
    {
      "epoch": 3.249677363847544,
      "grad_norm": 1.9375,
      "learning_rate": 0.00027330597037297165,
      "loss": 0.9963,
      "step": 37771
    },
    {
      "epoch": 3.2497634001548654,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00027328186893927145,
      "loss": 1.1151,
      "step": 37772
    },
    {
      "epoch": 3.249849436462187,
      "grad_norm": 350.0,
      "learning_rate": 0.00027325776816867765,
      "loss": 1.3224,
      "step": 37773
    },
    {
      "epoch": 3.2499354727695087,
      "grad_norm": 0.90625,
      "learning_rate": 0.00027323366806126083,
      "loss": 0.8416,
      "step": 37774
    },
    {
      "epoch": 3.2500215090768303,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002732095686170916,
      "loss": 1.3038,
      "step": 37775
    },
    {
      "epoch": 3.250107545384152,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00027318546983624007,
      "loss": 0.5598,
      "step": 37776
    },
    {
      "epoch": 3.250193581691474,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002731613717187772,
      "loss": 0.8003,
      "step": 37777
    },
    {
      "epoch": 3.2502796179987956,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00027313727426477285,
      "loss": 1.3872,
      "step": 37778
    },
    {
      "epoch": 3.2502796179987956,
      "eval_loss": 1.0922720432281494,
      "eval_runtime": 398.5868,
      "eval_samples_per_second": 11.664,
      "eval_steps_per_second": 1.46,
      "step": 37778
    },
    {
      "epoch": 3.2503656543061172,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00027311317747429797,
      "loss": 1.127,
      "step": 37779
    },
    {
      "epoch": 3.250451690613439,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002730890813474231,
      "loss": 0.7004,
      "step": 37780
    },
    {
      "epoch": 3.2505377269207605,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00027306498588421834,
      "loss": 1.2824,
      "step": 37781
    },
    {
      "epoch": 3.250623763228082,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002730408910847543,
      "loss": 0.8216,
      "step": 37782
    },
    {
      "epoch": 3.2507097995354037,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00027301679694910177,
      "loss": 1.4832,
      "step": 37783
    },
    {
      "epoch": 3.250795835842726,
      "grad_norm": 1.4375,
      "learning_rate": 0.00027299270347733075,
      "loss": 0.5586,
      "step": 37784
    },
    {
      "epoch": 3.2508818721500474,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002729686106695119,
      "loss": 0.9753,
      "step": 37785
    },
    {
      "epoch": 3.250967908457369,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00027294451852571597,
      "loss": 1.1538,
      "step": 37786
    },
    {
      "epoch": 3.2510539447646907,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002729204270460129,
      "loss": 0.7459,
      "step": 37787
    },
    {
      "epoch": 3.2511399810720123,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00027289633623047343,
      "loss": 1.205,
      "step": 37788
    },
    {
      "epoch": 3.251226017379334,
      "grad_norm": 1.359375,
      "learning_rate": 0.00027287224607916823,
      "loss": 1.2137,
      "step": 37789
    },
    {
      "epoch": 3.2513120536866555,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002728481565921673,
      "loss": 1.5119,
      "step": 37790
    },
    {
      "epoch": 3.2513980899939776,
      "grad_norm": 1.171875,
      "learning_rate": 0.00027282406776954157,
      "loss": 1.1178,
      "step": 37791
    },
    {
      "epoch": 3.2514841263012992,
      "grad_norm": 1.21875,
      "learning_rate": 0.00027279997961136105,
      "loss": 1.3803,
      "step": 37792
    },
    {
      "epoch": 3.251570162608621,
      "grad_norm": 5.625,
      "learning_rate": 0.0002727758921176964,
      "loss": 1.2123,
      "step": 37793
    },
    {
      "epoch": 3.2516561989159425,
      "grad_norm": 3.609375,
      "learning_rate": 0.00027275180528861824,
      "loss": 1.5036,
      "step": 37794
    },
    {
      "epoch": 3.251742235223264,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002727277191241967,
      "loss": 0.9957,
      "step": 37795
    },
    {
      "epoch": 3.2518282715305857,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002727036336245023,
      "loss": 1.017,
      "step": 37796
    },
    {
      "epoch": 3.2519143078379074,
      "grad_norm": 29.5,
      "learning_rate": 0.0002726795487896059,
      "loss": 0.9511,
      "step": 37797
    },
    {
      "epoch": 3.2520003441452294,
      "grad_norm": 3.84375,
      "learning_rate": 0.0002726554646195773,
      "loss": 1.5823,
      "step": 37798
    },
    {
      "epoch": 3.252086380452551,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00027263138111448726,
      "loss": 0.8428,
      "step": 37799
    },
    {
      "epoch": 3.2521724167598727,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00027260729827440644,
      "loss": 0.7632,
      "step": 37800
    },
    {
      "epoch": 3.2522584530671943,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002725832160994048,
      "loss": 1.1723,
      "step": 37801
    },
    {
      "epoch": 3.252344489374516,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002725591345895533,
      "loss": 0.4797,
      "step": 37802
    },
    {
      "epoch": 3.2524305256818375,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002725350537449219,
      "loss": 0.8254,
      "step": 37803
    },
    {
      "epoch": 3.2525165619891596,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002725109735655812,
      "loss": 1.1505,
      "step": 37804
    },
    {
      "epoch": 3.2526025982964812,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00027248689405160175,
      "loss": 0.9365,
      "step": 37805
    },
    {
      "epoch": 3.252688634603803,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00027246281520305383,
      "loss": 0.8638,
      "step": 37806
    },
    {
      "epoch": 3.2527746709111245,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00027243873702000786,
      "loss": 1.3904,
      "step": 37807
    },
    {
      "epoch": 3.252860707218446,
      "grad_norm": 3.0625,
      "learning_rate": 0.0002724146595025345,
      "loss": 1.3601,
      "step": 37808
    },
    {
      "epoch": 3.2529467435257677,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002723905826507039,
      "loss": 0.8672,
      "step": 37809
    },
    {
      "epoch": 3.25303277983309,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002723665064645865,
      "loss": 1.1247,
      "step": 37810
    },
    {
      "epoch": 3.2531188161404114,
      "grad_norm": 1.328125,
      "learning_rate": 0.0002723424309442528,
      "loss": 1.4553,
      "step": 37811
    },
    {
      "epoch": 3.253204852447733,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00027231835608977325,
      "loss": 0.883,
      "step": 37812
    },
    {
      "epoch": 3.2532908887550547,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002722942819012184,
      "loss": 1.2377,
      "step": 37813
    },
    {
      "epoch": 3.2533769250623763,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002722702083786583,
      "loss": 0.7866,
      "step": 37814
    },
    {
      "epoch": 3.253462961369698,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0002722461355221635,
      "loss": 1.2294,
      "step": 37815
    },
    {
      "epoch": 3.2535489976770195,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002722220633318047,
      "loss": 1.1939,
      "step": 37816
    },
    {
      "epoch": 3.2536350339843416,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00027219799180765186,
      "loss": 0.537,
      "step": 37817
    },
    {
      "epoch": 3.2537210702916632,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002721739209497756,
      "loss": 1.0077,
      "step": 37818
    },
    {
      "epoch": 3.253807106598985,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002721498507582465,
      "loss": 0.8709,
      "step": 37819
    },
    {
      "epoch": 3.2538931429063065,
      "grad_norm": 0.828125,
      "learning_rate": 0.00027212578123313457,
      "loss": 1.2102,
      "step": 37820
    },
    {
      "epoch": 3.253979179213628,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00027210171237451063,
      "loss": 0.877,
      "step": 37821
    },
    {
      "epoch": 3.2540652155209497,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00027207764418244465,
      "loss": 0.6411,
      "step": 37822
    },
    {
      "epoch": 3.2541512518282714,
      "grad_norm": 1.140625,
      "learning_rate": 0.00027205357665700727,
      "loss": 1.2769,
      "step": 37823
    },
    {
      "epoch": 3.2542372881355934,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002720295097982689,
      "loss": 1.8399,
      "step": 37824
    },
    {
      "epoch": 3.254323324442915,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00027200544360629983,
      "loss": 1.4788,
      "step": 37825
    },
    {
      "epoch": 3.2544093607502367,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002719813780811704,
      "loss": 1.0368,
      "step": 37826
    },
    {
      "epoch": 3.2544953970575583,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00027195731322295134,
      "loss": 0.757,
      "step": 37827
    },
    {
      "epoch": 3.25458143336488,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002719332490317126,
      "loss": 0.6626,
      "step": 37828
    },
    {
      "epoch": 3.2546674696722016,
      "grad_norm": 7.96875,
      "learning_rate": 0.00027190918550752473,
      "loss": 1.0186,
      "step": 37829
    },
    {
      "epoch": 3.254753505979523,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00027188512265045834,
      "loss": 0.806,
      "step": 37830
    },
    {
      "epoch": 3.2548395422868452,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002718610604605834,
      "loss": 1.3939,
      "step": 37831
    },
    {
      "epoch": 3.254925578594167,
      "grad_norm": 1.375,
      "learning_rate": 0.00027183699893797066,
      "loss": 0.8491,
      "step": 37832
    },
    {
      "epoch": 3.2550116149014885,
      "grad_norm": 2.265625,
      "learning_rate": 0.00027181293808269016,
      "loss": 1.477,
      "step": 37833
    },
    {
      "epoch": 3.25509765120881,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002717888778948124,
      "loss": 1.5438,
      "step": 37834
    },
    {
      "epoch": 3.2551836875161317,
      "grad_norm": 1.28125,
      "learning_rate": 0.00027176481837440803,
      "loss": 1.0354,
      "step": 37835
    },
    {
      "epoch": 3.2552697238234534,
      "grad_norm": 1.734375,
      "learning_rate": 0.0002717407595215469,
      "loss": 0.7862,
      "step": 37836
    },
    {
      "epoch": 3.255355760130775,
      "grad_norm": 2.125,
      "learning_rate": 0.00027171670133629977,
      "loss": 1.6773,
      "step": 37837
    },
    {
      "epoch": 3.255441796438097,
      "grad_norm": 1.203125,
      "learning_rate": 0.000271692643818737,
      "loss": 1.4834,
      "step": 37838
    },
    {
      "epoch": 3.2555278327454187,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002716685869689287,
      "loss": 1.1788,
      "step": 37839
    },
    {
      "epoch": 3.2556138690527403,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002716445307869453,
      "loss": 1.1655,
      "step": 37840
    },
    {
      "epoch": 3.255699905360062,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002716204752728575,
      "loss": 0.7861,
      "step": 37841
    },
    {
      "epoch": 3.2557859416673836,
      "grad_norm": 13.6875,
      "learning_rate": 0.0002715964204267352,
      "loss": 0.9051,
      "step": 37842
    },
    {
      "epoch": 3.255871977974705,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002715723662486491,
      "loss": 1.2128,
      "step": 37843
    },
    {
      "epoch": 3.255958014282027,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002715483127386692,
      "loss": 1.1278,
      "step": 37844
    },
    {
      "epoch": 3.256044050589349,
      "grad_norm": 1.140625,
      "learning_rate": 0.00027152425989686606,
      "loss": 0.963,
      "step": 37845
    },
    {
      "epoch": 3.2561300868966705,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0002715002077233103,
      "loss": 0.7169,
      "step": 37846
    },
    {
      "epoch": 3.256216123203992,
      "grad_norm": 0.78125,
      "learning_rate": 0.00027147615621807156,
      "loss": 0.5761,
      "step": 37847
    },
    {
      "epoch": 3.2563021595113137,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002714521053812209,
      "loss": 1.1018,
      "step": 37848
    },
    {
      "epoch": 3.2563881958186354,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00027142805521282833,
      "loss": 0.6023,
      "step": 37849
    },
    {
      "epoch": 3.256474232125957,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00027140400571296416,
      "loss": 0.9002,
      "step": 37850
    },
    {
      "epoch": 3.2565602684332786,
      "grad_norm": 1.40625,
      "learning_rate": 0.000271379956881699,
      "loss": 1.1974,
      "step": 37851
    },
    {
      "epoch": 3.2566463047406007,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002713559087191029,
      "loss": 1.0449,
      "step": 37852
    },
    {
      "epoch": 3.2567323410479223,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002713318612252462,
      "loss": 1.1465,
      "step": 37853
    },
    {
      "epoch": 3.256818377355244,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002713078144001996,
      "loss": 0.9598,
      "step": 37854
    },
    {
      "epoch": 3.2569044136625656,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002712837682440329,
      "loss": 1.0157,
      "step": 37855
    },
    {
      "epoch": 3.256990449969887,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002712597227568168,
      "loss": 1.6162,
      "step": 37856
    },
    {
      "epoch": 3.257076486277209,
      "grad_norm": 1.078125,
      "learning_rate": 0.00027123567793862157,
      "loss": 1.238,
      "step": 37857
    },
    {
      "epoch": 3.2571625225845304,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002712116337895174,
      "loss": 0.8304,
      "step": 37858
    },
    {
      "epoch": 3.2572485588918525,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002711875903095747,
      "loss": 1.3077,
      "step": 37859
    },
    {
      "epoch": 3.257334595199174,
      "grad_norm": 23.75,
      "learning_rate": 0.00027116354749886395,
      "loss": 0.9155,
      "step": 37860
    },
    {
      "epoch": 3.2574206315064957,
      "grad_norm": 1.03125,
      "learning_rate": 0.00027113950535745515,
      "loss": 1.3571,
      "step": 37861
    },
    {
      "epoch": 3.2575066678138174,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00027111546388541895,
      "loss": 1.8613,
      "step": 37862
    },
    {
      "epoch": 3.257592704121139,
      "grad_norm": 1.109375,
      "learning_rate": 0.00027109142308282534,
      "loss": 0.8482,
      "step": 37863
    },
    {
      "epoch": 3.2576787404284606,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00027106738294974483,
      "loss": 0.9902,
      "step": 37864
    },
    {
      "epoch": 3.2577647767357827,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002710433434862478,
      "loss": 0.4608,
      "step": 37865
    },
    {
      "epoch": 3.2578508130431043,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00027101930469240434,
      "loss": 1.2597,
      "step": 37866
    },
    {
      "epoch": 3.257936849350426,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00027099526656828486,
      "loss": 0.7138,
      "step": 37867
    },
    {
      "epoch": 3.2580228856577476,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002709712291139599,
      "loss": 0.7012,
      "step": 37868
    },
    {
      "epoch": 3.258108921965069,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002709471923294994,
      "loss": 1.0261,
      "step": 37869
    },
    {
      "epoch": 3.258194958272391,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00027092315621497364,
      "loss": 1.3362,
      "step": 37870
    },
    {
      "epoch": 3.258280994579713,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00027089912077045343,
      "loss": 1.1251,
      "step": 37871
    },
    {
      "epoch": 3.2583670308870345,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002708750859960085,
      "loss": 1.3335,
      "step": 37872
    },
    {
      "epoch": 3.258453067194356,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002708510518917097,
      "loss": 1.8834,
      "step": 37873
    },
    {
      "epoch": 3.2585391035016777,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002708270184576266,
      "loss": 1.4443,
      "step": 37874
    },
    {
      "epoch": 3.2586251398089994,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00027080298569383,
      "loss": 0.778,
      "step": 37875
    },
    {
      "epoch": 3.258711176116321,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002707789536003903,
      "loss": 0.7236,
      "step": 37876
    },
    {
      "epoch": 3.2587972124236426,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002707549221773774,
      "loss": 1.1364,
      "step": 37877
    },
    {
      "epoch": 3.2588832487309647,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002707308914248617,
      "loss": 0.799,
      "step": 37878
    },
    {
      "epoch": 3.2589692850382863,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002707068613429138,
      "loss": 0.5394,
      "step": 37879
    },
    {
      "epoch": 3.259055321345608,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002706828319316035,
      "loss": 1.3388,
      "step": 37880
    },
    {
      "epoch": 3.2591413576529296,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00027065880319100135,
      "loss": 1.4078,
      "step": 37881
    },
    {
      "epoch": 3.259227393960251,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002706347751211778,
      "loss": 1.3033,
      "step": 37882
    },
    {
      "epoch": 3.259313430267573,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00027061074772220274,
      "loss": 1.5757,
      "step": 37883
    },
    {
      "epoch": 3.2593994665748944,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00027058672099414656,
      "loss": 1.3711,
      "step": 37884
    },
    {
      "epoch": 3.2594855028822165,
      "grad_norm": 1.0625,
      "learning_rate": 0.00027056269493707963,
      "loss": 1.1674,
      "step": 37885
    },
    {
      "epoch": 3.259571539189538,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002705386695510722,
      "loss": 2.109,
      "step": 37886
    },
    {
      "epoch": 3.2596575754968597,
      "grad_norm": 1.078125,
      "learning_rate": 0.00027051464483619477,
      "loss": 1.0297,
      "step": 37887
    },
    {
      "epoch": 3.2597436118041814,
      "grad_norm": 2.03125,
      "learning_rate": 0.00027049062079251713,
      "loss": 0.996,
      "step": 37888
    },
    {
      "epoch": 3.259829648111503,
      "grad_norm": 0.984375,
      "learning_rate": 0.00027046659742011,
      "loss": 0.8107,
      "step": 37889
    },
    {
      "epoch": 3.2599156844188246,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002704425747190432,
      "loss": 0.8021,
      "step": 37890
    },
    {
      "epoch": 3.2600017207261462,
      "grad_norm": 1.0,
      "learning_rate": 0.00027041855268938723,
      "loss": 1.0485,
      "step": 37891
    },
    {
      "epoch": 3.2600877570334683,
      "grad_norm": 1.625,
      "learning_rate": 0.0002703945313312126,
      "loss": 0.9443,
      "step": 37892
    },
    {
      "epoch": 3.26017379334079,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002703705106445891,
      "loss": 0.901,
      "step": 37893
    },
    {
      "epoch": 3.2602598296481116,
      "grad_norm": 1.21875,
      "learning_rate": 0.00027034649062958717,
      "loss": 0.6865,
      "step": 37894
    },
    {
      "epoch": 3.260345865955433,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00027032247128627723,
      "loss": 1.6457,
      "step": 37895
    },
    {
      "epoch": 3.260431902262755,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002702984526147293,
      "loss": 1.3148,
      "step": 37896
    },
    {
      "epoch": 3.2605179385700764,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00027027443461501366,
      "loss": 1.6751,
      "step": 37897
    },
    {
      "epoch": 3.260603974877398,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002702504172872008,
      "loss": 0.9932,
      "step": 37898
    },
    {
      "epoch": 3.26069001118472,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002702264006313606,
      "loss": 1.11,
      "step": 37899
    },
    {
      "epoch": 3.2607760474920418,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002702023846475635,
      "loss": 1.736,
      "step": 37900
    },
    {
      "epoch": 3.2608620837993634,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00027017836933587995,
      "loss": 1.5504,
      "step": 37901
    },
    {
      "epoch": 3.260948120106685,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002701543546963798,
      "loss": 0.8981,
      "step": 37902
    },
    {
      "epoch": 3.2610341564140066,
      "grad_norm": 1.0,
      "learning_rate": 0.00027013034072913354,
      "loss": 1.331,
      "step": 37903
    },
    {
      "epoch": 3.2611201927213282,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00027010632743421125,
      "loss": 0.9443,
      "step": 37904
    },
    {
      "epoch": 3.26120622902865,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002700823148116832,
      "loss": 1.3681,
      "step": 37905
    },
    {
      "epoch": 3.261292265335972,
      "grad_norm": 1.515625,
      "learning_rate": 0.00027005830286161984,
      "loss": 1.4871,
      "step": 37906
    },
    {
      "epoch": 3.2613783016432936,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00027003429158409107,
      "loss": 0.6998,
      "step": 37907
    },
    {
      "epoch": 3.261464337950615,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00027001028097916723,
      "loss": 0.8686,
      "step": 37908
    },
    {
      "epoch": 3.261550374257937,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002699862710469188,
      "loss": 0.6208,
      "step": 37909
    },
    {
      "epoch": 3.2616364105652584,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00026996226178741564,
      "loss": 1.7862,
      "step": 37910
    },
    {
      "epoch": 3.26172244687258,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002699382532007281,
      "loss": 0.6204,
      "step": 37911
    },
    {
      "epoch": 3.2618084831799017,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002699142452869265,
      "loss": 0.6512,
      "step": 37912
    },
    {
      "epoch": 3.2618945194872238,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002698902380460809,
      "loss": 1.0523,
      "step": 37913
    },
    {
      "epoch": 3.2619805557945454,
      "grad_norm": 0.90625,
      "learning_rate": 0.00026986623147826186,
      "loss": 1.2933,
      "step": 37914
    },
    {
      "epoch": 3.262066592101867,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000269842225583539,
      "loss": 0.4994,
      "step": 37915
    },
    {
      "epoch": 3.2621526284091886,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00026981822036198296,
      "loss": 1.3403,
      "step": 37916
    },
    {
      "epoch": 3.2622386647165103,
      "grad_norm": 0.84375,
      "learning_rate": 0.00026979421581366404,
      "loss": 0.7482,
      "step": 37917
    },
    {
      "epoch": 3.262324701023832,
      "grad_norm": 1.078125,
      "learning_rate": 0.000269770211938652,
      "loss": 1.7758,
      "step": 37918
    },
    {
      "epoch": 3.2624107373311535,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002697462087370174,
      "loss": 1.5252,
      "step": 37919
    },
    {
      "epoch": 3.2624967736384756,
      "grad_norm": 0.96875,
      "learning_rate": 0.00026972220620883046,
      "loss": 1.0954,
      "step": 37920
    },
    {
      "epoch": 3.262582809945797,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000269698204354161,
      "loss": 0.6687,
      "step": 37921
    },
    {
      "epoch": 3.262668846253119,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00026967420317307983,
      "loss": 2.1347,
      "step": 37922
    },
    {
      "epoch": 3.2627548825604404,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00026965020266565663,
      "loss": 0.7892,
      "step": 37923
    },
    {
      "epoch": 3.262840918867762,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002696262028319618,
      "loss": 1.2797,
      "step": 37924
    },
    {
      "epoch": 3.262926955175084,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002696022036720657,
      "loss": 0.824,
      "step": 37925
    },
    {
      "epoch": 3.2630129914824058,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00026957820518603816,
      "loss": 0.9328,
      "step": 37926
    },
    {
      "epoch": 3.2630990277897274,
      "grad_norm": 0.96875,
      "learning_rate": 0.00026955420737394953,
      "loss": 0.7282,
      "step": 37927
    },
    {
      "epoch": 3.263185064097049,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002695302102358702,
      "loss": 1.154,
      "step": 37928
    },
    {
      "epoch": 3.2632711004043706,
      "grad_norm": 4.25,
      "learning_rate": 0.0002695062137718701,
      "loss": 1.058,
      "step": 37929
    },
    {
      "epoch": 3.2633571367116923,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002694822179820194,
      "loss": 1.6517,
      "step": 37930
    },
    {
      "epoch": 3.263443173019014,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00026945822286638863,
      "loss": 0.7508,
      "step": 37931
    },
    {
      "epoch": 3.263529209326336,
      "grad_norm": 1.828125,
      "learning_rate": 0.00026943422842504747,
      "loss": 1.0119,
      "step": 37932
    },
    {
      "epoch": 3.2636152456336576,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00026941023465806656,
      "loss": 0.9196,
      "step": 37933
    },
    {
      "epoch": 3.263701281940979,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002693862415655157,
      "loss": 1.4373,
      "step": 37934
    },
    {
      "epoch": 3.263787318248301,
      "grad_norm": 1.15625,
      "learning_rate": 0.00026936224914746523,
      "loss": 1.1524,
      "step": 37935
    },
    {
      "epoch": 3.2638733545556224,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00026933825740398543,
      "loss": 0.8379,
      "step": 37936
    },
    {
      "epoch": 3.263959390862944,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002693142663351462,
      "loss": 0.7,
      "step": 37937
    },
    {
      "epoch": 3.2640454271702657,
      "grad_norm": 1.25,
      "learning_rate": 0.00026929027594101797,
      "loss": 1.3488,
      "step": 37938
    },
    {
      "epoch": 3.2641314634775878,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002692662862216709,
      "loss": 1.0125,
      "step": 37939
    },
    {
      "epoch": 3.2642174997849094,
      "grad_norm": 1.25,
      "learning_rate": 0.00026924229717717495,
      "loss": 1.3357,
      "step": 37940
    },
    {
      "epoch": 3.264303536092231,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002692183088076004,
      "loss": 1.2996,
      "step": 37941
    },
    {
      "epoch": 3.2643895723995526,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002691943211130176,
      "loss": 1.0507,
      "step": 37942
    },
    {
      "epoch": 3.2644756087068743,
      "grad_norm": 1.7578125,
      "learning_rate": 0.00026917033409349636,
      "loss": 1.0926,
      "step": 37943
    },
    {
      "epoch": 3.264561645014196,
      "grad_norm": 1.1875,
      "learning_rate": 0.00026914634774910707,
      "loss": 1.227,
      "step": 37944
    },
    {
      "epoch": 3.2646476813215175,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002691223620799197,
      "loss": 1.461,
      "step": 37945
    },
    {
      "epoch": 3.2647337176288396,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00026909837708600446,
      "loss": 0.7983,
      "step": 37946
    },
    {
      "epoch": 3.264819753936161,
      "grad_norm": 1.140625,
      "learning_rate": 0.00026907439276743187,
      "loss": 0.8543,
      "step": 37947
    },
    {
      "epoch": 3.264905790243483,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00026905040912427147,
      "loss": 1.3553,
      "step": 37948
    },
    {
      "epoch": 3.2649918265508044,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002690264261565938,
      "loss": 0.939,
      "step": 37949
    },
    {
      "epoch": 3.265077862858126,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00026900244386446906,
      "loss": 1.5368,
      "step": 37950
    },
    {
      "epoch": 3.2651638991654477,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000268978462247967,
      "loss": 0.8152,
      "step": 37951
    },
    {
      "epoch": 3.2652499354727693,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00026895448130715804,
      "loss": 0.539,
      "step": 37952
    },
    {
      "epoch": 3.2653359717800914,
      "grad_norm": 1.59375,
      "learning_rate": 0.0002689305010421125,
      "loss": 1.402,
      "step": 37953
    },
    {
      "epoch": 3.265422008087413,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002689065214529001,
      "loss": 1.2035,
      "step": 37954
    },
    {
      "epoch": 3.2655080443947346,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002688825425395914,
      "loss": 0.9438,
      "step": 37955
    },
    {
      "epoch": 3.2655940807020563,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002688585643022561,
      "loss": 1.0204,
      "step": 37956
    },
    {
      "epoch": 3.265680117009378,
      "grad_norm": 3.15625,
      "learning_rate": 0.0002688345867409645,
      "loss": 1.4701,
      "step": 37957
    },
    {
      "epoch": 3.2657661533166995,
      "grad_norm": 2.1875,
      "learning_rate": 0.0002688106098557868,
      "loss": 1.3847,
      "step": 37958
    },
    {
      "epoch": 3.265852189624021,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002687866336467931,
      "loss": 1.1735,
      "step": 37959
    },
    {
      "epoch": 3.265938225931343,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002687626581140538,
      "loss": 1.0445,
      "step": 37960
    },
    {
      "epoch": 3.266024262238665,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00026873868325763843,
      "loss": 1.1578,
      "step": 37961
    },
    {
      "epoch": 3.2661102985459864,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002687147090776175,
      "loss": 1.4428,
      "step": 37962
    },
    {
      "epoch": 3.266196334853308,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002686907355740613,
      "loss": 1.9,
      "step": 37963
    },
    {
      "epoch": 3.2662823711606297,
      "grad_norm": 1.546875,
      "learning_rate": 0.00026866676274703944,
      "loss": 1.2353,
      "step": 37964
    },
    {
      "epoch": 3.2663684074679513,
      "grad_norm": 1.046875,
      "learning_rate": 0.00026864279059662233,
      "loss": 0.6986,
      "step": 37965
    },
    {
      "epoch": 3.266454443775273,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002686188191228803,
      "loss": 0.7728,
      "step": 37966
    },
    {
      "epoch": 3.266540480082595,
      "grad_norm": 0.96875,
      "learning_rate": 0.000268594848325883,
      "loss": 1.3835,
      "step": 37967
    },
    {
      "epoch": 3.2666265163899166,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002685708782057007,
      "loss": 0.9925,
      "step": 37968
    },
    {
      "epoch": 3.2667125526972383,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00026854690876240386,
      "loss": 0.8449,
      "step": 37969
    },
    {
      "epoch": 3.26679858900456,
      "grad_norm": 1.40625,
      "learning_rate": 0.00026852293999606204,
      "loss": 1.4338,
      "step": 37970
    },
    {
      "epoch": 3.2668846253118815,
      "grad_norm": 1.015625,
      "learning_rate": 0.00026849897190674567,
      "loss": 0.6196,
      "step": 37971
    },
    {
      "epoch": 3.266970661619203,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0002684750044945249,
      "loss": 1.7785,
      "step": 37972
    },
    {
      "epoch": 3.2670566979265248,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00026845103775946955,
      "loss": 1.7439,
      "step": 37973
    },
    {
      "epoch": 3.267142734233847,
      "grad_norm": 2.1875,
      "learning_rate": 0.00026842707170165005,
      "loss": 0.9214,
      "step": 37974
    },
    {
      "epoch": 3.2672287705411684,
      "grad_norm": 2.28125,
      "learning_rate": 0.0002684031063211362,
      "loss": 0.8015,
      "step": 37975
    },
    {
      "epoch": 3.26731480684849,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00026837914161799816,
      "loss": 1.1788,
      "step": 37976
    },
    {
      "epoch": 3.2674008431558117,
      "grad_norm": 1.640625,
      "learning_rate": 0.0002683551775923063,
      "loss": 0.895,
      "step": 37977
    },
    {
      "epoch": 3.2674868794631333,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00026833121424413027,
      "loss": 1.1478,
      "step": 37978
    },
    {
      "epoch": 3.267572915770455,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002683072515735404,
      "loss": 1.7013,
      "step": 37979
    },
    {
      "epoch": 3.267658952077777,
      "grad_norm": 0.921875,
      "learning_rate": 0.000268283289580607,
      "loss": 0.5972,
      "step": 37980
    },
    {
      "epoch": 3.2677449883850986,
      "grad_norm": 2.328125,
      "learning_rate": 0.0002682593282653997,
      "loss": 1.071,
      "step": 37981
    },
    {
      "epoch": 3.2678310246924203,
      "grad_norm": 4.40625,
      "learning_rate": 0.00026823536762798873,
      "loss": 1.0905,
      "step": 37982
    },
    {
      "epoch": 3.267917060999742,
      "grad_norm": 1.234375,
      "learning_rate": 0.00026821140766844444,
      "loss": 1.6595,
      "step": 37983
    },
    {
      "epoch": 3.2680030973070635,
      "grad_norm": 2.046875,
      "learning_rate": 0.00026818744838683653,
      "loss": 1.4553,
      "step": 37984
    },
    {
      "epoch": 3.268089133614385,
      "grad_norm": 4.875,
      "learning_rate": 0.0002681634897832355,
      "loss": 0.9259,
      "step": 37985
    },
    {
      "epoch": 3.268175169921707,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002681395318577109,
      "loss": 1.1795,
      "step": 37986
    },
    {
      "epoch": 3.268261206229029,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002681155746103331,
      "loss": 1.5746,
      "step": 37987
    },
    {
      "epoch": 3.2683472425363504,
      "grad_norm": 6.78125,
      "learning_rate": 0.0002680916180411723,
      "loss": 0.8219,
      "step": 37988
    },
    {
      "epoch": 3.268433278843672,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002680676621502983,
      "loss": 0.6512,
      "step": 37989
    },
    {
      "epoch": 3.2685193151509937,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00026804370693778124,
      "loss": 1.3575,
      "step": 37990
    },
    {
      "epoch": 3.2686053514583153,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00026801975240369136,
      "loss": 1.1245,
      "step": 37991
    },
    {
      "epoch": 3.268691387765637,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00026799579854809845,
      "loss": 1.8796,
      "step": 37992
    },
    {
      "epoch": 3.268777424072959,
      "grad_norm": 1.25,
      "learning_rate": 0.0002679718453710727,
      "loss": 1.1371,
      "step": 37993
    },
    {
      "epoch": 3.2688634603802806,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002679478928726844,
      "loss": 1.2316,
      "step": 37994
    },
    {
      "epoch": 3.2689494966876023,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000267923941053003,
      "loss": 1.3616,
      "step": 37995
    },
    {
      "epoch": 3.269035532994924,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002678999899120993,
      "loss": 0.7192,
      "step": 37996
    },
    {
      "epoch": 3.2691215693022455,
      "grad_norm": 1.0625,
      "learning_rate": 0.00026787603945004274,
      "loss": 1.0366,
      "step": 37997
    },
    {
      "epoch": 3.269207605609567,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002678520896669037,
      "loss": 1.066,
      "step": 37998
    },
    {
      "epoch": 3.2692936419168888,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002678281405627523,
      "loss": 1.8524,
      "step": 37999
    },
    {
      "epoch": 3.269379678224211,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00026780419213765823,
      "loss": 0.7054,
      "step": 38000
    },
    {
      "epoch": 3.2694657145315325,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00026778024439169196,
      "loss": 1.1386,
      "step": 38001
    },
    {
      "epoch": 3.269551750838854,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002677562973249231,
      "loss": 1.2962,
      "step": 38002
    },
    {
      "epoch": 3.2696377871461757,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002677323509374219,
      "loss": 1.4778,
      "step": 38003
    },
    {
      "epoch": 3.2697238234534973,
      "grad_norm": 2.0625,
      "learning_rate": 0.0002677084052292586,
      "loss": 1.5845,
      "step": 38004
    },
    {
      "epoch": 3.269809859760819,
      "grad_norm": 1.15625,
      "learning_rate": 0.00026768446020050283,
      "loss": 1.4485,
      "step": 38005
    },
    {
      "epoch": 3.2698958960681406,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002676605158512249,
      "loss": 1.2969,
      "step": 38006
    },
    {
      "epoch": 3.2699819323754626,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002676365721814949,
      "loss": 1.1889,
      "step": 38007
    },
    {
      "epoch": 3.2700679686827843,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0002676126291913825,
      "loss": 0.6995,
      "step": 38008
    },
    {
      "epoch": 3.270154004990106,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00026758868688095806,
      "loss": 2.0326,
      "step": 38009
    },
    {
      "epoch": 3.2702400412974275,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00026756474525029163,
      "loss": 0.3666,
      "step": 38010
    },
    {
      "epoch": 3.270326077604749,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00026754080429945297,
      "loss": 0.6658,
      "step": 38011
    },
    {
      "epoch": 3.2704121139120708,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002675168640285123,
      "loss": 0.7485,
      "step": 38012
    },
    {
      "epoch": 3.2704981502193924,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002674929244375396,
      "loss": 0.9305,
      "step": 38013
    },
    {
      "epoch": 3.2705841865267145,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002674689855266048,
      "loss": 0.9099,
      "step": 38014
    },
    {
      "epoch": 3.270670222834036,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00026744504729577824,
      "loss": 1.4987,
      "step": 38015
    },
    {
      "epoch": 3.2707562591413577,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002674211097451294,
      "loss": 1.18,
      "step": 38016
    },
    {
      "epoch": 3.2708422954486793,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00026739717287472864,
      "loss": 1.4918,
      "step": 38017
    },
    {
      "epoch": 3.270928331756001,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026737323668464606,
      "loss": 1.917,
      "step": 38018
    },
    {
      "epoch": 3.2710143680633226,
      "grad_norm": 1.25,
      "learning_rate": 0.0002673493011749513,
      "loss": 1.0577,
      "step": 38019
    },
    {
      "epoch": 3.271100404370644,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002673253663457146,
      "loss": 1.3423,
      "step": 38020
    },
    {
      "epoch": 3.2711864406779663,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002673014321970062,
      "loss": 1.6273,
      "step": 38021
    },
    {
      "epoch": 3.271272476985288,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002672774987288956,
      "loss": 0.8879,
      "step": 38022
    },
    {
      "epoch": 3.2713585132926095,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002672535659414531,
      "loss": 1.4327,
      "step": 38023
    },
    {
      "epoch": 3.271444549599931,
      "grad_norm": 1.3125,
      "learning_rate": 0.00026722963383474886,
      "loss": 0.7077,
      "step": 38024
    },
    {
      "epoch": 3.2715305859072528,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00026720570240885245,
      "loss": 0.8778,
      "step": 38025
    },
    {
      "epoch": 3.2716166222145744,
      "grad_norm": 1.25,
      "learning_rate": 0.0002671817716638343,
      "loss": 0.9921,
      "step": 38026
    },
    {
      "epoch": 3.271702658521896,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00026715784159976396,
      "loss": 1.1288,
      "step": 38027
    },
    {
      "epoch": 3.271788694829218,
      "grad_norm": 1.46875,
      "learning_rate": 0.00026713391221671164,
      "loss": 1.4109,
      "step": 38028
    },
    {
      "epoch": 3.2718747311365397,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002671099835147476,
      "loss": 0.7533,
      "step": 38029
    },
    {
      "epoch": 3.2719607674438613,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002670860554939413,
      "loss": 1.5736,
      "step": 38030
    },
    {
      "epoch": 3.272046803751183,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000267062128154363,
      "loss": 1.4309,
      "step": 38031
    },
    {
      "epoch": 3.2721328400585046,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00026703820149608266,
      "loss": 0.8444,
      "step": 38032
    },
    {
      "epoch": 3.272218876365826,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00026701427551917024,
      "loss": 1.3956,
      "step": 38033
    },
    {
      "epoch": 3.272304912673148,
      "grad_norm": 1.0625,
      "learning_rate": 0.000266990350223696,
      "loss": 0.6948,
      "step": 38034
    },
    {
      "epoch": 3.27239094898047,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002669664256097295,
      "loss": 1.7764,
      "step": 38035
    },
    {
      "epoch": 3.2724769852877915,
      "grad_norm": 1.671875,
      "learning_rate": 0.00026694250167734083,
      "loss": 0.7273,
      "step": 38036
    },
    {
      "epoch": 3.272563021595113,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002669185784266003,
      "loss": 0.7885,
      "step": 38037
    },
    {
      "epoch": 3.2726490579024348,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002668946558575773,
      "loss": 1.2269,
      "step": 38038
    },
    {
      "epoch": 3.2727350942097564,
      "grad_norm": 1.03125,
      "learning_rate": 0.00026687073397034213,
      "loss": 0.7741,
      "step": 38039
    },
    {
      "epoch": 3.272821130517078,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002668468127649649,
      "loss": 1.1873,
      "step": 38040
    },
    {
      "epoch": 3.2729071668244,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002668228922415152,
      "loss": 1.1755,
      "step": 38041
    },
    {
      "epoch": 3.2729932031317217,
      "grad_norm": 1.625,
      "learning_rate": 0.0002667989724000634,
      "loss": 1.0277,
      "step": 38042
    },
    {
      "epoch": 3.2730792394390433,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002667750532406791,
      "loss": 0.9608,
      "step": 38043
    },
    {
      "epoch": 3.273165275746365,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002667511347634323,
      "loss": 0.9303,
      "step": 38044
    },
    {
      "epoch": 3.2732513120536866,
      "grad_norm": 1.15625,
      "learning_rate": 0.00026672721696839336,
      "loss": 1.0591,
      "step": 38045
    },
    {
      "epoch": 3.273337348361008,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00026670329985563167,
      "loss": 1.5594,
      "step": 38046
    },
    {
      "epoch": 3.2734233846683303,
      "grad_norm": 13.1875,
      "learning_rate": 0.00026667938342521755,
      "loss": 1.8283,
      "step": 38047
    },
    {
      "epoch": 3.273509420975652,
      "grad_norm": 1.109375,
      "learning_rate": 0.000266655467677221,
      "loss": 0.9149,
      "step": 38048
    },
    {
      "epoch": 3.2735954572829735,
      "grad_norm": 1.53125,
      "learning_rate": 0.0002666315526117116,
      "loss": 1.1599,
      "step": 38049
    },
    {
      "epoch": 3.273681493590295,
      "grad_norm": 1.25,
      "learning_rate": 0.00026660763822875953,
      "loss": 1.6145,
      "step": 38050
    },
    {
      "epoch": 3.2737675298976168,
      "grad_norm": 1.265625,
      "learning_rate": 0.00026658372452843495,
      "loss": 0.7556,
      "step": 38051
    },
    {
      "epoch": 3.2738535662049384,
      "grad_norm": 1.125,
      "learning_rate": 0.00026655981151080737,
      "loss": 1.4462,
      "step": 38052
    },
    {
      "epoch": 3.27393960251226,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00026653589917594695,
      "loss": 0.8223,
      "step": 38053
    },
    {
      "epoch": 3.274025638819582,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002665119875239238,
      "loss": 0.6546,
      "step": 38054
    },
    {
      "epoch": 3.2741116751269037,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0002664880765548075,
      "loss": 1.3046,
      "step": 38055
    },
    {
      "epoch": 3.2741977114342253,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002664641662686683,
      "loss": 1.6022,
      "step": 38056
    },
    {
      "epoch": 3.274283747741547,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00026644025666557584,
      "loss": 0.7164,
      "step": 38057
    },
    {
      "epoch": 3.2743697840488686,
      "grad_norm": 2.34375,
      "learning_rate": 0.00026641634774560017,
      "loss": 1.8208,
      "step": 38058
    },
    {
      "epoch": 3.27445582035619,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002663924395088115,
      "loss": 0.8514,
      "step": 38059
    },
    {
      "epoch": 3.274541856663512,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002663685319552793,
      "loss": 1.1121,
      "step": 38060
    },
    {
      "epoch": 3.274627892970834,
      "grad_norm": 1.0,
      "learning_rate": 0.0002663446250850737,
      "loss": 1.383,
      "step": 38061
    },
    {
      "epoch": 3.2747139292781555,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002663207188982649,
      "loss": 1.3314,
      "step": 38062
    },
    {
      "epoch": 3.274799965585477,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002662968133949223,
      "loss": 1.4393,
      "step": 38063
    },
    {
      "epoch": 3.2748860018927988,
      "grad_norm": 0.99609375,
      "learning_rate": 0.000266272908575116,
      "loss": 1.0937,
      "step": 38064
    },
    {
      "epoch": 3.2749720382001204,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00026624900443891617,
      "loss": 0.697,
      "step": 38065
    },
    {
      "epoch": 3.275058074507442,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00026622510098639246,
      "loss": 1.3168,
      "step": 38066
    },
    {
      "epoch": 3.2751441108147636,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000266201198217615,
      "loss": 1.4137,
      "step": 38067
    },
    {
      "epoch": 3.2752301471220857,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002661772961326534,
      "loss": 1.3089,
      "step": 38068
    },
    {
      "epoch": 3.2753161834294073,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00026615339473157754,
      "loss": 1.0321,
      "step": 38069
    },
    {
      "epoch": 3.275402219736729,
      "grad_norm": 0.875,
      "learning_rate": 0.00026612949401445785,
      "loss": 0.8946,
      "step": 38070
    },
    {
      "epoch": 3.2754882560440506,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002661055939813637,
      "loss": 1.0575,
      "step": 38071
    },
    {
      "epoch": 3.275574292351372,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002660816946323654,
      "loss": 0.8762,
      "step": 38072
    },
    {
      "epoch": 3.275660328658694,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026605779596753254,
      "loss": 0.8642,
      "step": 38073
    },
    {
      "epoch": 3.2757463649660155,
      "grad_norm": 1.8671875,
      "learning_rate": 0.0002660338979869351,
      "loss": 1.7221,
      "step": 38074
    },
    {
      "epoch": 3.2758324012733375,
      "grad_norm": 0.953125,
      "learning_rate": 0.00026601000069064314,
      "loss": 1.1598,
      "step": 38075
    },
    {
      "epoch": 3.275918437580659,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002659861040787263,
      "loss": 0.6329,
      "step": 38076
    },
    {
      "epoch": 3.2760044738879808,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002659622081512545,
      "loss": 0.8736,
      "step": 38077
    },
    {
      "epoch": 3.2760905101953024,
      "grad_norm": 1.3125,
      "learning_rate": 0.000265938312908298,
      "loss": 1.811,
      "step": 38078
    },
    {
      "epoch": 3.276176546502624,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002659144183499261,
      "loss": 0.821,
      "step": 38079
    },
    {
      "epoch": 3.2762625828099456,
      "grad_norm": 13.4375,
      "learning_rate": 0.0002658905244762091,
      "loss": 0.932,
      "step": 38080
    },
    {
      "epoch": 3.2763486191172673,
      "grad_norm": 1.171875,
      "learning_rate": 0.000265866631287217,
      "loss": 1.394,
      "step": 38081
    },
    {
      "epoch": 3.2764346554245893,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002658427387830192,
      "loss": 0.8713,
      "step": 38082
    },
    {
      "epoch": 3.276520691731911,
      "grad_norm": 3.203125,
      "learning_rate": 0.00026581884696368584,
      "loss": 1.2132,
      "step": 38083
    },
    {
      "epoch": 3.2766067280392326,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002657949558292871,
      "loss": 1.2971,
      "step": 38084
    },
    {
      "epoch": 3.276692764346554,
      "grad_norm": 41.0,
      "learning_rate": 0.0002657710653798923,
      "loss": 0.8324,
      "step": 38085
    },
    {
      "epoch": 3.276778800653876,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002657471756155718,
      "loss": 1.5351,
      "step": 38086
    },
    {
      "epoch": 3.2768648369611975,
      "grad_norm": 2.390625,
      "learning_rate": 0.000265723286536395,
      "loss": 1.5675,
      "step": 38087
    },
    {
      "epoch": 3.276950873268519,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002656993981424321,
      "loss": 1.4375,
      "step": 38088
    },
    {
      "epoch": 3.277036909575841,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00026567551043375304,
      "loss": 0.8991,
      "step": 38089
    },
    {
      "epoch": 3.2771229458831628,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002656516234104274,
      "loss": 0.9242,
      "step": 38090
    },
    {
      "epoch": 3.2772089821904844,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002656277370725251,
      "loss": 1.1604,
      "step": 38091
    },
    {
      "epoch": 3.277295018497806,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002656038514201163,
      "loss": 0.9438,
      "step": 38092
    },
    {
      "epoch": 3.2773810548051276,
      "grad_norm": 1.28125,
      "learning_rate": 0.00026557996645327055,
      "loss": 1.0379,
      "step": 38093
    },
    {
      "epoch": 3.2774670911124493,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002655560821720577,
      "loss": 1.0397,
      "step": 38094
    },
    {
      "epoch": 3.277553127419771,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002655321985765479,
      "loss": 1.0158,
      "step": 38095
    },
    {
      "epoch": 3.277639163727093,
      "grad_norm": 2.015625,
      "learning_rate": 0.00026550831566681064,
      "loss": 0.7729,
      "step": 38096
    },
    {
      "epoch": 3.2777252000344146,
      "grad_norm": 4.75,
      "learning_rate": 0.0002654844334429162,
      "loss": 1.9383,
      "step": 38097
    },
    {
      "epoch": 3.277811236341736,
      "grad_norm": 1.15625,
      "learning_rate": 0.00026546055190493394,
      "loss": 1.363,
      "step": 38098
    },
    {
      "epoch": 3.277897272649058,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00026543667105293395,
      "loss": 0.9467,
      "step": 38099
    },
    {
      "epoch": 3.2779833089563795,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00026541279088698635,
      "loss": 0.8752,
      "step": 38100
    },
    {
      "epoch": 3.2780693452637015,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002653889114071604,
      "loss": 0.7648,
      "step": 38101
    },
    {
      "epoch": 3.278155381571023,
      "grad_norm": 1.0,
      "learning_rate": 0.0002653650326135263,
      "loss": 1.5011,
      "step": 38102
    },
    {
      "epoch": 3.2782414178783448,
      "grad_norm": 1.6796875,
      "learning_rate": 0.000265341154506154,
      "loss": 0.9743,
      "step": 38103
    },
    {
      "epoch": 3.2783274541856664,
      "grad_norm": 1.125,
      "learning_rate": 0.0002653172770851131,
      "loss": 0.8862,
      "step": 38104
    },
    {
      "epoch": 3.278413490492988,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026529340035047347,
      "loss": 1.1354,
      "step": 38105
    },
    {
      "epoch": 3.2784995268003096,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00026526952430230497,
      "loss": 0.5558,
      "step": 38106
    },
    {
      "epoch": 3.2785855631076313,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002652456489406775,
      "loss": 0.6703,
      "step": 38107
    },
    {
      "epoch": 3.2786715994149533,
      "grad_norm": 1.3125,
      "learning_rate": 0.00026522177426566097,
      "loss": 1.7167,
      "step": 38108
    },
    {
      "epoch": 3.278757635722275,
      "grad_norm": 1.375,
      "learning_rate": 0.0002651979002773249,
      "loss": 1.5562,
      "step": 38109
    },
    {
      "epoch": 3.2788436720295966,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026517402697573937,
      "loss": 1.2161,
      "step": 38110
    },
    {
      "epoch": 3.278929708336918,
      "grad_norm": 1.171875,
      "learning_rate": 0.00026515015436097433,
      "loss": 1.6306,
      "step": 38111
    },
    {
      "epoch": 3.27901574464424,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00026512628243309923,
      "loss": 1.0087,
      "step": 38112
    },
    {
      "epoch": 3.2791017809515615,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002651024111921842,
      "loss": 0.8644,
      "step": 38113
    },
    {
      "epoch": 3.279187817258883,
      "grad_norm": 4.40625,
      "learning_rate": 0.00026507854063829875,
      "loss": 1.3713,
      "step": 38114
    },
    {
      "epoch": 3.279273853566205,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002650546707715129,
      "loss": 0.7954,
      "step": 38115
    },
    {
      "epoch": 3.279359889873527,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002650308015918966,
      "loss": 1.3789,
      "step": 38116
    },
    {
      "epoch": 3.2794459261808484,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002650069330995194,
      "loss": 0.832,
      "step": 38117
    },
    {
      "epoch": 3.27953196248817,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00026498306529445115,
      "loss": 0.7431,
      "step": 38118
    },
    {
      "epoch": 3.2796179987954917,
      "grad_norm": 1.03125,
      "learning_rate": 0.00026495919817676194,
      "loss": 0.7898,
      "step": 38119
    },
    {
      "epoch": 3.2797040351028133,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002649353317465212,
      "loss": 1.6832,
      "step": 38120
    },
    {
      "epoch": 3.279790071410135,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00026491146600379886,
      "loss": 0.9754,
      "step": 38121
    },
    {
      "epoch": 3.279876107717457,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002648876009486649,
      "loss": 1.3303,
      "step": 38122
    },
    {
      "epoch": 3.2799621440247786,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002648637365811889,
      "loss": 0.8317,
      "step": 38123
    },
    {
      "epoch": 3.2800481803321,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002648398729014407,
      "loss": 0.7513,
      "step": 38124
    },
    {
      "epoch": 3.280134216639422,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002648160099094903,
      "loss": 0.7652,
      "step": 38125
    },
    {
      "epoch": 3.2802202529467435,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00026479214760540714,
      "loss": 1.9917,
      "step": 38126
    },
    {
      "epoch": 3.280306289254065,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002647682859892614,
      "loss": 1.1022,
      "step": 38127
    },
    {
      "epoch": 3.2803923255613867,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002647444250611225,
      "loss": 1.0471,
      "step": 38128
    },
    {
      "epoch": 3.280478361868709,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002647205648210604,
      "loss": 0.9674,
      "step": 38129
    },
    {
      "epoch": 3.2805643981760304,
      "grad_norm": 2.125,
      "learning_rate": 0.00026469670526914503,
      "loss": 1.6563,
      "step": 38130
    },
    {
      "epoch": 3.280650434483352,
      "grad_norm": 1.28125,
      "learning_rate": 0.00026467284640544587,
      "loss": 1.5652,
      "step": 38131
    },
    {
      "epoch": 3.2807364707906737,
      "grad_norm": 1.265625,
      "learning_rate": 0.00026464898823003294,
      "loss": 1.2042,
      "step": 38132
    },
    {
      "epoch": 3.2808225070979953,
      "grad_norm": 1.609375,
      "learning_rate": 0.000264625130742976,
      "loss": 1.3171,
      "step": 38133
    },
    {
      "epoch": 3.280908543405317,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002646012739443447,
      "loss": 1.5936,
      "step": 38134
    },
    {
      "epoch": 3.2809945797126385,
      "grad_norm": 7840.0,
      "learning_rate": 0.00026457741783420885,
      "loss": 1.2337,
      "step": 38135
    },
    {
      "epoch": 3.2810806160199606,
      "grad_norm": 1.25,
      "learning_rate": 0.00026455356241263843,
      "loss": 1.5554,
      "step": 38136
    },
    {
      "epoch": 3.281166652327282,
      "grad_norm": 1.015625,
      "learning_rate": 0.00026452970767970287,
      "loss": 0.7634,
      "step": 38137
    },
    {
      "epoch": 3.281252688634604,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00026450585363547237,
      "loss": 1.4837,
      "step": 38138
    },
    {
      "epoch": 3.2813387249419255,
      "grad_norm": 1.046875,
      "learning_rate": 0.00026448200028001616,
      "loss": 0.7174,
      "step": 38139
    },
    {
      "epoch": 3.281424761249247,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002644581476134044,
      "loss": 0.9921,
      "step": 38140
    },
    {
      "epoch": 3.2815107975565687,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00026443429563570686,
      "loss": 0.7334,
      "step": 38141
    },
    {
      "epoch": 3.2815968338638903,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002644104443469928,
      "loss": 1.4937,
      "step": 38142
    },
    {
      "epoch": 3.2816828701712124,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00026438659374733274,
      "loss": 1.0151,
      "step": 38143
    },
    {
      "epoch": 3.281768906478534,
      "grad_norm": 1.359375,
      "learning_rate": 0.000264362743836796,
      "loss": 1.0476,
      "step": 38144
    },
    {
      "epoch": 3.2818549427858557,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002643388946154523,
      "loss": 0.7937,
      "step": 38145
    },
    {
      "epoch": 3.2819409790931773,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0002643150460833717,
      "loss": 0.8638,
      "step": 38146
    },
    {
      "epoch": 3.282027015400499,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00026429119824062356,
      "loss": 0.6403,
      "step": 38147
    },
    {
      "epoch": 3.2821130517078205,
      "grad_norm": 1.046875,
      "learning_rate": 0.00026426735108727784,
      "loss": 1.2612,
      "step": 38148
    },
    {
      "epoch": 3.282199088015142,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00026424350462340453,
      "loss": 1.3632,
      "step": 38149
    },
    {
      "epoch": 3.282285124322464,
      "grad_norm": 1.21875,
      "learning_rate": 0.00026421965884907285,
      "loss": 0.7683,
      "step": 38150
    },
    {
      "epoch": 3.282371160629786,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0002641958137643528,
      "loss": 1.2905,
      "step": 38151
    },
    {
      "epoch": 3.2824571969371075,
      "grad_norm": 5.8125,
      "learning_rate": 0.0002641719693693143,
      "loss": 0.9712,
      "step": 38152
    },
    {
      "epoch": 3.282543233244429,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002641481256640268,
      "loss": 1.1496,
      "step": 38153
    },
    {
      "epoch": 3.2826292695517507,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002641242826485603,
      "loss": 1.0031,
      "step": 38154
    },
    {
      "epoch": 3.2827153058590723,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00026410044032298426,
      "loss": 0.7552,
      "step": 38155
    },
    {
      "epoch": 3.2828013421663944,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00026407659868736853,
      "loss": 2.0781,
      "step": 38156
    },
    {
      "epoch": 3.282887378473716,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002640527577417831,
      "loss": 1.4297,
      "step": 38157
    },
    {
      "epoch": 3.2829734147810377,
      "grad_norm": 4.34375,
      "learning_rate": 0.00026402891748629714,
      "loss": 1.0556,
      "step": 38158
    },
    {
      "epoch": 3.2830594510883593,
      "grad_norm": 1.140625,
      "learning_rate": 0.00026400507792098086,
      "loss": 1.4433,
      "step": 38159
    },
    {
      "epoch": 3.283145487395681,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002639812390459039,
      "loss": 1.0571,
      "step": 38160
    },
    {
      "epoch": 3.2832315237030025,
      "grad_norm": 2.25,
      "learning_rate": 0.00026395740086113573,
      "loss": 0.8049,
      "step": 38161
    },
    {
      "epoch": 3.2833175600103246,
      "grad_norm": 1.203125,
      "learning_rate": 0.00026393356336674637,
      "loss": 1.2821,
      "step": 38162
    },
    {
      "epoch": 3.2834035963176462,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00026390972656280547,
      "loss": 0.9151,
      "step": 38163
    },
    {
      "epoch": 3.283489632624968,
      "grad_norm": 1.109375,
      "learning_rate": 0.00026388589044938263,
      "loss": 0.704,
      "step": 38164
    },
    {
      "epoch": 3.2835756689322895,
      "grad_norm": 1.25,
      "learning_rate": 0.00026386205502654757,
      "loss": 0.7887,
      "step": 38165
    },
    {
      "epoch": 3.283661705239611,
      "grad_norm": 93.5,
      "learning_rate": 0.0002638382202943703,
      "loss": 1.3772,
      "step": 38166
    },
    {
      "epoch": 3.2837477415469327,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002638143862529201,
      "loss": 1.0959,
      "step": 38167
    },
    {
      "epoch": 3.2838337778542543,
      "grad_norm": 1.390625,
      "learning_rate": 0.00026379055290226707,
      "loss": 0.7715,
      "step": 38168
    },
    {
      "epoch": 3.2839198141615764,
      "grad_norm": 1.875,
      "learning_rate": 0.0002637667202424805,
      "loss": 0.8208,
      "step": 38169
    },
    {
      "epoch": 3.284005850468898,
      "grad_norm": 3.25,
      "learning_rate": 0.00026374288827363035,
      "loss": 1.3578,
      "step": 38170
    },
    {
      "epoch": 3.2840918867762197,
      "grad_norm": 191.0,
      "learning_rate": 0.0002637190569957866,
      "loss": 1.7594,
      "step": 38171
    },
    {
      "epoch": 3.2841779230835413,
      "grad_norm": 88.0,
      "learning_rate": 0.0002636952264090183,
      "loss": 1.0884,
      "step": 38172
    },
    {
      "epoch": 3.284263959390863,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0002636713965133956,
      "loss": 1.0982,
      "step": 38173
    },
    {
      "epoch": 3.2843499956981845,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002636475673089883,
      "loss": 0.7438,
      "step": 38174
    },
    {
      "epoch": 3.284436032005506,
      "grad_norm": 4.875,
      "learning_rate": 0.0002636237387958657,
      "loss": 0.7304,
      "step": 38175
    },
    {
      "epoch": 3.2845220683128282,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026359991097409764,
      "loss": 1.3271,
      "step": 38176
    },
    {
      "epoch": 3.28460810462015,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002635760838437541,
      "loss": 0.5904,
      "step": 38177
    },
    {
      "epoch": 3.2846941409274715,
      "grad_norm": 118.0,
      "learning_rate": 0.0002635522574049043,
      "loss": 1.6,
      "step": 38178
    },
    {
      "epoch": 3.284780177234793,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002635284316576182,
      "loss": 0.9901,
      "step": 38179
    },
    {
      "epoch": 3.2848662135421147,
      "grad_norm": 1.40625,
      "learning_rate": 0.00026350460660196545,
      "loss": 1.3417,
      "step": 38180
    },
    {
      "epoch": 3.2849522498494363,
      "grad_norm": 0.890625,
      "learning_rate": 0.00026348078223801575,
      "loss": 0.999,
      "step": 38181
    },
    {
      "epoch": 3.285038286156758,
      "grad_norm": 1.0,
      "learning_rate": 0.0002634569585658389,
      "loss": 0.7085,
      "step": 38182
    },
    {
      "epoch": 3.28512432246408,
      "grad_norm": 1.046875,
      "learning_rate": 0.00026343313558550423,
      "loss": 0.5937,
      "step": 38183
    },
    {
      "epoch": 3.2852103587714017,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002634093132970819,
      "loss": 0.9989,
      "step": 38184
    },
    {
      "epoch": 3.2852963950787233,
      "grad_norm": 9.125,
      "learning_rate": 0.00026338549170064107,
      "loss": 0.4838,
      "step": 38185
    },
    {
      "epoch": 3.285382431386045,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00026336167079625165,
      "loss": 0.8937,
      "step": 38186
    },
    {
      "epoch": 3.2854684676933665,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002633378505839835,
      "loss": 1.3417,
      "step": 38187
    },
    {
      "epoch": 3.285554504000688,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000263314031063906,
      "loss": 1.2513,
      "step": 38188
    },
    {
      "epoch": 3.28564054030801,
      "grad_norm": 18.0,
      "learning_rate": 0.00026329021223608885,
      "loss": 1.4927,
      "step": 38189
    },
    {
      "epoch": 3.285726576615332,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00026326639410060204,
      "loss": 1.2367,
      "step": 38190
    },
    {
      "epoch": 3.2858126129226535,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002632425766575147,
      "loss": 1.3551,
      "step": 38191
    },
    {
      "epoch": 3.285898649229975,
      "grad_norm": 1.734375,
      "learning_rate": 0.0002632187599068969,
      "loss": 1.9949,
      "step": 38192
    },
    {
      "epoch": 3.2859846855372967,
      "grad_norm": 11.9375,
      "learning_rate": 0.00026319494384881824,
      "loss": 1.3561,
      "step": 38193
    },
    {
      "epoch": 3.2860707218446183,
      "grad_norm": 1.171875,
      "learning_rate": 0.00026317112848334823,
      "loss": 1.653,
      "step": 38194
    },
    {
      "epoch": 3.28615675815194,
      "grad_norm": 67.0,
      "learning_rate": 0.0002631473138105568,
      "loss": 1.0853,
      "step": 38195
    },
    {
      "epoch": 3.2862427944592616,
      "grad_norm": 0.859375,
      "learning_rate": 0.00026312349983051317,
      "loss": 0.911,
      "step": 38196
    },
    {
      "epoch": 3.2863288307665837,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002630996865432872,
      "loss": 0.9896,
      "step": 38197
    },
    {
      "epoch": 3.2864148670739053,
      "grad_norm": 1.109375,
      "learning_rate": 0.00026307587394894884,
      "loss": 1.2189,
      "step": 38198
    },
    {
      "epoch": 3.286500903381227,
      "grad_norm": 25.875,
      "learning_rate": 0.0002630520620475672,
      "loss": 1.369,
      "step": 38199
    },
    {
      "epoch": 3.2865869396885485,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002630282508392122,
      "loss": 1.1905,
      "step": 38200
    },
    {
      "epoch": 3.28667297599587,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002630044403239538,
      "loss": 0.8699,
      "step": 38201
    },
    {
      "epoch": 3.286759012303192,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000262980630501861,
      "loss": 0.8907,
      "step": 38202
    },
    {
      "epoch": 3.2868450486105134,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002629568213730038,
      "loss": 1.5445,
      "step": 38203
    },
    {
      "epoch": 3.2869310849178355,
      "grad_norm": 1.0625,
      "learning_rate": 0.00026293301293745197,
      "loss": 0.7318,
      "step": 38204
    },
    {
      "epoch": 3.287017121225157,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002629092051952748,
      "loss": 0.7061,
      "step": 38205
    },
    {
      "epoch": 3.2871031575324787,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002628853981465421,
      "loss": 0.7745,
      "step": 38206
    },
    {
      "epoch": 3.2871891938398003,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002628615917913236,
      "loss": 0.8343,
      "step": 38207
    },
    {
      "epoch": 3.287275230147122,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002628377861296887,
      "loss": 1.211,
      "step": 38208
    },
    {
      "epoch": 3.2873612664544436,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00026281398116170736,
      "loss": 1.8809,
      "step": 38209
    },
    {
      "epoch": 3.2874473027617652,
      "grad_norm": 111.5,
      "learning_rate": 0.00026279017688744875,
      "loss": 0.7784,
      "step": 38210
    },
    {
      "epoch": 3.2875333390690873,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002627663733069828,
      "loss": 1.418,
      "step": 38211
    },
    {
      "epoch": 3.287619375376409,
      "grad_norm": 51.0,
      "learning_rate": 0.00026274257042037917,
      "loss": 0.5275,
      "step": 38212
    },
    {
      "epoch": 3.2877054116837305,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002627187682277073,
      "loss": 1.1816,
      "step": 38213
    },
    {
      "epoch": 3.287791447991052,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00026269496672903683,
      "loss": 0.9696,
      "step": 38214
    },
    {
      "epoch": 3.287877484298374,
      "grad_norm": 221.0,
      "learning_rate": 0.0002626711659244376,
      "loss": 1.6184,
      "step": 38215
    },
    {
      "epoch": 3.2879635206056954,
      "grad_norm": 1.21875,
      "learning_rate": 0.00026264736581397874,
      "loss": 0.9551,
      "step": 38216
    },
    {
      "epoch": 3.2880495569130175,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00026262356639773044,
      "loss": 1.1878,
      "step": 38217
    },
    {
      "epoch": 3.288135593220339,
      "grad_norm": 7.0625,
      "learning_rate": 0.00026259976767576196,
      "loss": 1.1954,
      "step": 38218
    },
    {
      "epoch": 3.2882216295276607,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002625759696481429,
      "loss": 1.431,
      "step": 38219
    },
    {
      "epoch": 3.2883076658349824,
      "grad_norm": 1.921875,
      "learning_rate": 0.00026255217231494323,
      "loss": 0.7114,
      "step": 38220
    },
    {
      "epoch": 3.288393702142304,
      "grad_norm": 1.015625,
      "learning_rate": 0.000262528375676232,
      "loss": 0.7863,
      "step": 38221
    },
    {
      "epoch": 3.2884797384496256,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002625045797320791,
      "loss": 0.7203,
      "step": 38222
    },
    {
      "epoch": 3.2885657747569477,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00026248078448255443,
      "loss": 0.8338,
      "step": 38223
    },
    {
      "epoch": 3.2886518110642693,
      "grad_norm": 1.375,
      "learning_rate": 0.0002624569899277269,
      "loss": 1.4284,
      "step": 38224
    },
    {
      "epoch": 3.288737847371591,
      "grad_norm": 40.5,
      "learning_rate": 0.0002624331960676668,
      "loss": 1.7491,
      "step": 38225
    },
    {
      "epoch": 3.2888238836789125,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00026240940290244314,
      "loss": 0.5338,
      "step": 38226
    },
    {
      "epoch": 3.288909919986234,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00026238561043212576,
      "loss": 0.6224,
      "step": 38227
    },
    {
      "epoch": 3.288995956293556,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00026236181865678447,
      "loss": 0.9906,
      "step": 38228
    },
    {
      "epoch": 3.2890819926008774,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00026233802757648844,
      "loss": 1.0755,
      "step": 38229
    },
    {
      "epoch": 3.2891680289081995,
      "grad_norm": 10.875,
      "learning_rate": 0.0002623142371913075,
      "loss": 1.9442,
      "step": 38230
    },
    {
      "epoch": 3.289254065215521,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002622904475013114,
      "loss": 1.0832,
      "step": 38231
    },
    {
      "epoch": 3.2893401015228427,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002622666585065692,
      "loss": 1.0254,
      "step": 38232
    },
    {
      "epoch": 3.2894261378301644,
      "grad_norm": 7.0625,
      "learning_rate": 0.0002622428702071509,
      "loss": 1.3193,
      "step": 38233
    },
    {
      "epoch": 3.289512174137486,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00026221908260312606,
      "loss": 1.5192,
      "step": 38234
    },
    {
      "epoch": 3.2895982104448076,
      "grad_norm": 28.25,
      "learning_rate": 0.000262195295694564,
      "loss": 1.6679,
      "step": 38235
    },
    {
      "epoch": 3.2896842467521292,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00026217150948153444,
      "loss": 1.0785,
      "step": 38236
    },
    {
      "epoch": 3.2897702830594513,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002621477239641072,
      "loss": 1.22,
      "step": 38237
    },
    {
      "epoch": 3.289856319366773,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002621239391423514,
      "loss": 1.2946,
      "step": 38238
    },
    {
      "epoch": 3.2899423556740945,
      "grad_norm": 2.796875,
      "learning_rate": 0.000262100155016337,
      "loss": 0.9708,
      "step": 38239
    },
    {
      "epoch": 3.290028391981416,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002620763715861332,
      "loss": 1.0638,
      "step": 38240
    },
    {
      "epoch": 3.290114428288738,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002620525888518097,
      "loss": 0.6611,
      "step": 38241
    },
    {
      "epoch": 3.2902004645960594,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002620288068134363,
      "loss": 1.8571,
      "step": 38242
    },
    {
      "epoch": 3.290286500903381,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00026200502547108216,
      "loss": 0.9107,
      "step": 38243
    },
    {
      "epoch": 3.290372537210703,
      "grad_norm": 2.390625,
      "learning_rate": 0.000261981244824817,
      "loss": 1.643,
      "step": 38244
    },
    {
      "epoch": 3.2904585735180247,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0002619574648747107,
      "loss": 0.7144,
      "step": 38245
    },
    {
      "epoch": 3.2905446098253464,
      "grad_norm": 0.90625,
      "learning_rate": 0.00026193368562083224,
      "loss": 1.1602,
      "step": 38246
    },
    {
      "epoch": 3.290630646132668,
      "grad_norm": 42.0,
      "learning_rate": 0.00026190990706325144,
      "loss": 0.9593,
      "step": 38247
    },
    {
      "epoch": 3.2907166824399896,
      "grad_norm": 1.078125,
      "learning_rate": 0.00026188612920203814,
      "loss": 0.9694,
      "step": 38248
    },
    {
      "epoch": 3.2908027187473112,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002618623520372614,
      "loss": 1.591,
      "step": 38249
    },
    {
      "epoch": 3.290888755054633,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002618385755689911,
      "loss": 1.4182,
      "step": 38250
    },
    {
      "epoch": 3.290974791361955,
      "grad_norm": 35.25,
      "learning_rate": 0.0002618147997972965,
      "loss": 1.1226,
      "step": 38251
    },
    {
      "epoch": 3.2910608276692765,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00026179102472224724,
      "loss": 0.8723,
      "step": 38252
    },
    {
      "epoch": 3.291146863976598,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002617672503439129,
      "loss": 0.9084,
      "step": 38253
    },
    {
      "epoch": 3.29123290028392,
      "grad_norm": 1.0625,
      "learning_rate": 0.00026174347666236305,
      "loss": 0.6951,
      "step": 38254
    },
    {
      "epoch": 3.2913189365912414,
      "grad_norm": 94.0,
      "learning_rate": 0.0002617197036776674,
      "loss": 1.1894,
      "step": 38255
    },
    {
      "epoch": 3.291404972898563,
      "grad_norm": 37.75,
      "learning_rate": 0.000261695931389895,
      "loss": 0.5355,
      "step": 38256
    },
    {
      "epoch": 3.2914910092058847,
      "grad_norm": 224.0,
      "learning_rate": 0.00026167215979911574,
      "loss": 1.2206,
      "step": 38257
    },
    {
      "epoch": 3.2915770455132067,
      "grad_norm": 4.5625,
      "learning_rate": 0.0002616483889053992,
      "loss": 0.8757,
      "step": 38258
    },
    {
      "epoch": 3.2916630818205284,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0002616246187088146,
      "loss": 0.9074,
      "step": 38259
    },
    {
      "epoch": 3.29174911812785,
      "grad_norm": 0.9375,
      "learning_rate": 0.00026160084920943163,
      "loss": 1.1495,
      "step": 38260
    },
    {
      "epoch": 3.2918351544351716,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00026157708040732,
      "loss": 1.0046,
      "step": 38261
    },
    {
      "epoch": 3.2919211907424932,
      "grad_norm": 1.84375,
      "learning_rate": 0.0002615533123025489,
      "loss": 1.1073,
      "step": 38262
    },
    {
      "epoch": 3.292007227049815,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00026152954489518787,
      "loss": 0.8592,
      "step": 38263
    },
    {
      "epoch": 3.2920932633571365,
      "grad_norm": 6.0,
      "learning_rate": 0.0002615057781853068,
      "loss": 1.3372,
      "step": 38264
    },
    {
      "epoch": 3.2921792996644585,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00026148201217297476,
      "loss": 0.7515,
      "step": 38265
    },
    {
      "epoch": 3.29226533597178,
      "grad_norm": 3.8125,
      "learning_rate": 0.0002614582468582616,
      "loss": 1.4205,
      "step": 38266
    },
    {
      "epoch": 3.292351372279102,
      "grad_norm": 63.5,
      "learning_rate": 0.00026143448224123653,
      "loss": 1.6298,
      "step": 38267
    },
    {
      "epoch": 3.2924374085864234,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002614107183219692,
      "loss": 1.4923,
      "step": 38268
    },
    {
      "epoch": 3.292523444893745,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002613869551005292,
      "loss": 0.9715,
      "step": 38269
    },
    {
      "epoch": 3.2926094812010667,
      "grad_norm": 1.21875,
      "learning_rate": 0.00026136319257698585,
      "loss": 0.9184,
      "step": 38270
    },
    {
      "epoch": 3.2926955175083883,
      "grad_norm": 18.5,
      "learning_rate": 0.00026133943075140874,
      "loss": 0.8152,
      "step": 38271
    },
    {
      "epoch": 3.2927815538157104,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00026131566962386756,
      "loss": 1.2441,
      "step": 38272
    },
    {
      "epoch": 3.292867590123032,
      "grad_norm": 1.09375,
      "learning_rate": 0.00026129190919443147,
      "loss": 1.5413,
      "step": 38273
    },
    {
      "epoch": 3.2929536264303536,
      "grad_norm": 1.421875,
      "learning_rate": 0.00026126814946317013,
      "loss": 1.3519,
      "step": 38274
    },
    {
      "epoch": 3.2930396627376752,
      "grad_norm": 1.671875,
      "learning_rate": 0.00026124439043015315,
      "loss": 1.8776,
      "step": 38275
    },
    {
      "epoch": 3.293125699044997,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002612206320954497,
      "loss": 1.0518,
      "step": 38276
    },
    {
      "epoch": 3.293211735352319,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002611968744591295,
      "loss": 1.1027,
      "step": 38277
    },
    {
      "epoch": 3.2932977716596405,
      "grad_norm": 1.25,
      "learning_rate": 0.00026117311752126216,
      "loss": 1.8817,
      "step": 38278
    },
    {
      "epoch": 3.293383807966962,
      "grad_norm": 6.75,
      "learning_rate": 0.0002611493612819168,
      "loss": 1.1418,
      "step": 38279
    },
    {
      "epoch": 3.293469844274284,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002611256057411634,
      "loss": 0.6614,
      "step": 38280
    },
    {
      "epoch": 3.2935558805816054,
      "grad_norm": 2.734375,
      "learning_rate": 0.00026110185089907084,
      "loss": 1.1276,
      "step": 38281
    },
    {
      "epoch": 3.293641916888927,
      "grad_norm": 1.15625,
      "learning_rate": 0.00026107809675570895,
      "loss": 1.3094,
      "step": 38282
    },
    {
      "epoch": 3.2937279531962487,
      "grad_norm": 22.25,
      "learning_rate": 0.0002610543433111473,
      "loss": 1.2068,
      "step": 38283
    },
    {
      "epoch": 3.2938139895035707,
      "grad_norm": 112.5,
      "learning_rate": 0.0002610305905654551,
      "loss": 1.0453,
      "step": 38284
    },
    {
      "epoch": 3.2939000258108924,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0002610068385187019,
      "loss": 0.8164,
      "step": 38285
    },
    {
      "epoch": 3.293986062118214,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002609830871709574,
      "loss": 0.8534,
      "step": 38286
    },
    {
      "epoch": 3.2940720984255356,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002609593365222908,
      "loss": 0.6965,
      "step": 38287
    },
    {
      "epoch": 3.2941581347328572,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002609355865727715,
      "loss": 1.0279,
      "step": 38288
    },
    {
      "epoch": 3.294244171040179,
      "grad_norm": 1.03125,
      "learning_rate": 0.00026091183732246935,
      "loss": 1.3087,
      "step": 38289
    },
    {
      "epoch": 3.2943302073475005,
      "grad_norm": 1.453125,
      "learning_rate": 0.00026088808877145327,
      "loss": 1.069,
      "step": 38290
    },
    {
      "epoch": 3.2944162436548226,
      "grad_norm": 0.90625,
      "learning_rate": 0.00026086434091979335,
      "loss": 0.9724,
      "step": 38291
    },
    {
      "epoch": 3.294502279962144,
      "grad_norm": 39.5,
      "learning_rate": 0.00026084059376755853,
      "loss": 1.1344,
      "step": 38292
    },
    {
      "epoch": 3.294588316269466,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0002608168473148185,
      "loss": 1.6287,
      "step": 38293
    },
    {
      "epoch": 3.2946743525767874,
      "grad_norm": 1.90625,
      "learning_rate": 0.0002607931015616428,
      "loss": 1.1311,
      "step": 38294
    },
    {
      "epoch": 3.294760388884109,
      "grad_norm": 29.875,
      "learning_rate": 0.0002607693565081006,
      "loss": 1.1614,
      "step": 38295
    },
    {
      "epoch": 3.2948464251914307,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002607456121542617,
      "loss": 0.8468,
      "step": 38296
    },
    {
      "epoch": 3.2949324614987523,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00026072186850019516,
      "loss": 0.9834,
      "step": 38297
    },
    {
      "epoch": 3.2950184978060744,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002606981255459706,
      "loss": 1.5486,
      "step": 38298
    },
    {
      "epoch": 3.295104534113396,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00026067438329165765,
      "loss": 1.8127,
      "step": 38299
    },
    {
      "epoch": 3.2951905704207176,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00026065064173732546,
      "loss": 1.3718,
      "step": 38300
    },
    {
      "epoch": 3.2952766067280392,
      "grad_norm": 1.375,
      "learning_rate": 0.00026062690088304353,
      "loss": 0.7217,
      "step": 38301
    },
    {
      "epoch": 3.295362643035361,
      "grad_norm": 1.25,
      "learning_rate": 0.0002606031607288816,
      "loss": 1.3254,
      "step": 38302
    },
    {
      "epoch": 3.2954486793426825,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00026057942127490866,
      "loss": 0.9862,
      "step": 38303
    },
    {
      "epoch": 3.295534715650004,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002605556825211943,
      "loss": 1.251,
      "step": 38304
    },
    {
      "epoch": 3.295620751957326,
      "grad_norm": 2.84375,
      "learning_rate": 0.0002605319444678082,
      "loss": 0.9276,
      "step": 38305
    },
    {
      "epoch": 3.295706788264648,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00026050820711481944,
      "loss": 1.0132,
      "step": 38306
    },
    {
      "epoch": 3.2957928245719694,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002604844704622977,
      "loss": 1.349,
      "step": 38307
    },
    {
      "epoch": 3.295878860879291,
      "grad_norm": 85.0,
      "learning_rate": 0.0002604607345103123,
      "loss": 0.7654,
      "step": 38308
    },
    {
      "epoch": 3.2959648971866127,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002604369992589325,
      "loss": 1.3151,
      "step": 38309
    },
    {
      "epoch": 3.2960509334939343,
      "grad_norm": 1.78125,
      "learning_rate": 0.00026041326470822813,
      "loss": 1.9268,
      "step": 38310
    },
    {
      "epoch": 3.296136969801256,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00026038953085826813,
      "loss": 1.0196,
      "step": 38311
    },
    {
      "epoch": 3.296223006108578,
      "grad_norm": 3.28125,
      "learning_rate": 0.00026036579770912215,
      "loss": 0.9766,
      "step": 38312
    },
    {
      "epoch": 3.2963090424158996,
      "grad_norm": 1.015625,
      "learning_rate": 0.00026034206526085986,
      "loss": 0.5732,
      "step": 38313
    },
    {
      "epoch": 3.2963950787232212,
      "grad_norm": 1.53125,
      "learning_rate": 0.00026031833351355016,
      "loss": 1.456,
      "step": 38314
    },
    {
      "epoch": 3.296481115030543,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00026029460246726276,
      "loss": 1.3651,
      "step": 38315
    },
    {
      "epoch": 3.2965671513378645,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002602708721220672,
      "loss": 1.0603,
      "step": 38316
    },
    {
      "epoch": 3.296653187645186,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002602471424780326,
      "loss": 1.3948,
      "step": 38317
    },
    {
      "epoch": 3.2967392239525077,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002602234135352284,
      "loss": 1.3949,
      "step": 38318
    },
    {
      "epoch": 3.29682526025983,
      "grad_norm": 69.5,
      "learning_rate": 0.0002601996852937243,
      "loss": 1.579,
      "step": 38319
    },
    {
      "epoch": 3.2969112965671514,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00026017595775358924,
      "loss": 1.0527,
      "step": 38320
    },
    {
      "epoch": 3.296997332874473,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00026015223091489315,
      "loss": 0.8177,
      "step": 38321
    },
    {
      "epoch": 3.2970833691817947,
      "grad_norm": 0.875,
      "learning_rate": 0.0002601285047777048,
      "loss": 0.6186,
      "step": 38322
    },
    {
      "epoch": 3.2971694054891163,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002601047793420941,
      "loss": 1.7113,
      "step": 38323
    },
    {
      "epoch": 3.297255441796438,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002600810546081303,
      "loss": 1.4222,
      "step": 38324
    },
    {
      "epoch": 3.2973414781037595,
      "grad_norm": 1.09375,
      "learning_rate": 0.00026005733057588265,
      "loss": 0.6367,
      "step": 38325
    },
    {
      "epoch": 3.2974275144110816,
      "grad_norm": 1.328125,
      "learning_rate": 0.00026003360724542057,
      "loss": 1.0942,
      "step": 38326
    },
    {
      "epoch": 3.2975135507184032,
      "grad_norm": 28.625,
      "learning_rate": 0.0002600098846168136,
      "loss": 0.8688,
      "step": 38327
    },
    {
      "epoch": 3.297599587025725,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000259986162690131,
      "loss": 2.1678,
      "step": 38328
    },
    {
      "epoch": 3.2976856233330465,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002599624414654424,
      "loss": 0.7108,
      "step": 38329
    },
    {
      "epoch": 3.297771659640368,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00025993872094281676,
      "loss": 0.6826,
      "step": 38330
    },
    {
      "epoch": 3.2978576959476897,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002599150011223237,
      "loss": 0.8122,
      "step": 38331
    },
    {
      "epoch": 3.297943732255012,
      "grad_norm": 192.0,
      "learning_rate": 0.00025989128200403275,
      "loss": 0.8566,
      "step": 38332
    },
    {
      "epoch": 3.2980297685623334,
      "grad_norm": 1.890625,
      "learning_rate": 0.0002598675635880129,
      "loss": 1.7464,
      "step": 38333
    },
    {
      "epoch": 3.298115804869655,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00025984384587433374,
      "loss": 1.0089,
      "step": 38334
    },
    {
      "epoch": 3.2982018411769767,
      "grad_norm": 7.78125,
      "learning_rate": 0.00025982012886306477,
      "loss": 0.9059,
      "step": 38335
    },
    {
      "epoch": 3.2982878774842983,
      "grad_norm": 1.09375,
      "learning_rate": 0.00025979641255427505,
      "loss": 0.5096,
      "step": 38336
    },
    {
      "epoch": 3.29837391379162,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002597726969480343,
      "loss": 1.4501,
      "step": 38337
    },
    {
      "epoch": 3.298459950098942,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002597489820444115,
      "loss": 1.4987,
      "step": 38338
    },
    {
      "epoch": 3.2985459864062636,
      "grad_norm": 1.34375,
      "learning_rate": 0.00025972526784347617,
      "loss": 1.1751,
      "step": 38339
    },
    {
      "epoch": 3.2986320227135852,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00025970155434529795,
      "loss": 0.9365,
      "step": 38340
    },
    {
      "epoch": 3.298718059020907,
      "grad_norm": 6.03125,
      "learning_rate": 0.0002596778415499457,
      "loss": 1.6038,
      "step": 38341
    },
    {
      "epoch": 3.2988040953282285,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002596541294574891,
      "loss": 0.9408,
      "step": 38342
    },
    {
      "epoch": 3.29889013163555,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002596304180679976,
      "loss": 1.3274,
      "step": 38343
    },
    {
      "epoch": 3.2989761679428717,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002596067073815401,
      "loss": 1.1598,
      "step": 38344
    },
    {
      "epoch": 3.299062204250194,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002595829973981863,
      "loss": 1.0849,
      "step": 38345
    },
    {
      "epoch": 3.2991482405575154,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00025955928811800566,
      "loss": 1.3988,
      "step": 38346
    },
    {
      "epoch": 3.299234276864837,
      "grad_norm": 2.703125,
      "learning_rate": 0.00025953557954106713,
      "loss": 1.3041,
      "step": 38347
    },
    {
      "epoch": 3.2993203131721587,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002595118716674403,
      "loss": 0.5516,
      "step": 38348
    },
    {
      "epoch": 3.2994063494794803,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002594881644971946,
      "loss": 0.7501,
      "step": 38349
    },
    {
      "epoch": 3.299492385786802,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002594644580303991,
      "loss": 0.8848,
      "step": 38350
    },
    {
      "epoch": 3.2995784220941236,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00025944075226712343,
      "loss": 0.6915,
      "step": 38351
    },
    {
      "epoch": 3.2996644584014456,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00025941704720743665,
      "loss": 1.5473,
      "step": 38352
    },
    {
      "epoch": 3.2997504947087672,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00025939334285140814,
      "loss": 0.6176,
      "step": 38353
    },
    {
      "epoch": 3.299836531016089,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002593696391991075,
      "loss": 1.8132,
      "step": 38354
    },
    {
      "epoch": 3.2999225673234105,
      "grad_norm": 1.8125,
      "learning_rate": 0.0002593459362506037,
      "loss": 1.7858,
      "step": 38355
    },
    {
      "epoch": 3.300008603630732,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002593222340059662,
      "loss": 0.7694,
      "step": 38356
    },
    {
      "epoch": 3.3000946399380537,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002592985324652646,
      "loss": 1.9032,
      "step": 38357
    },
    {
      "epoch": 3.3001806762453754,
      "grad_norm": 3.015625,
      "learning_rate": 0.0002592748316285678,
      "loss": 1.1781,
      "step": 38358
    },
    {
      "epoch": 3.3002667125526974,
      "grad_norm": 1.484375,
      "learning_rate": 0.00025925113149594527,
      "loss": 0.7669,
      "step": 38359
    },
    {
      "epoch": 3.300352748860019,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00025922743206746655,
      "loss": 1.684,
      "step": 38360
    },
    {
      "epoch": 3.3004387851673407,
      "grad_norm": 2.71875,
      "learning_rate": 0.00025920373334320057,
      "loss": 0.9585,
      "step": 38361
    },
    {
      "epoch": 3.3005248214746623,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00025918003532321703,
      "loss": 1.6343,
      "step": 38362
    },
    {
      "epoch": 3.300610857781984,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002591563380075849,
      "loss": 1.4086,
      "step": 38363
    },
    {
      "epoch": 3.3006968940893056,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002591326413963734,
      "loss": 1.2205,
      "step": 38364
    },
    {
      "epoch": 3.300782930396627,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00025910894548965247,
      "loss": 0.9796,
      "step": 38365
    },
    {
      "epoch": 3.3008689667039492,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00025908525028749076,
      "loss": 0.8158,
      "step": 38366
    },
    {
      "epoch": 3.300955003011271,
      "grad_norm": 2.09375,
      "learning_rate": 0.00025906155578995816,
      "loss": 1.4643,
      "step": 38367
    },
    {
      "epoch": 3.3010410393185925,
      "grad_norm": 1.140625,
      "learning_rate": 0.00025903786199712336,
      "loss": 1.5605,
      "step": 38368
    },
    {
      "epoch": 3.301127075625914,
      "grad_norm": 223.0,
      "learning_rate": 0.00025901416890905597,
      "loss": 1.8437,
      "step": 38369
    },
    {
      "epoch": 3.3012131119332357,
      "grad_norm": 1296.0,
      "learning_rate": 0.00025899047652582554,
      "loss": 0.8328,
      "step": 38370
    },
    {
      "epoch": 3.3012991482405574,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002589667848475008,
      "loss": 0.8048,
      "step": 38371
    },
    {
      "epoch": 3.301385184547879,
      "grad_norm": 2.03125,
      "learning_rate": 0.0002589430938741514,
      "loss": 1.062,
      "step": 38372
    },
    {
      "epoch": 3.301471220855201,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00025891940360584664,
      "loss": 1.5993,
      "step": 38373
    },
    {
      "epoch": 3.3015572571625227,
      "grad_norm": 1.203125,
      "learning_rate": 0.00025889571404265566,
      "loss": 1.7586,
      "step": 38374
    },
    {
      "epoch": 3.3016432934698443,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002588720251846478,
      "loss": 1.1617,
      "step": 38375
    },
    {
      "epoch": 3.301729329777166,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002588483370318926,
      "loss": 1.1766,
      "step": 38376
    },
    {
      "epoch": 3.3018153660844876,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002588246495844588,
      "loss": 0.8153,
      "step": 38377
    },
    {
      "epoch": 3.301901402391809,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0002588009628424163,
      "loss": 1.0777,
      "step": 38378
    },
    {
      "epoch": 3.301987438699131,
      "grad_norm": 5.71875,
      "learning_rate": 0.00025877727680583386,
      "loss": 1.3406,
      "step": 38379
    },
    {
      "epoch": 3.302073475006453,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00025875359147478096,
      "loss": 0.6416,
      "step": 38380
    },
    {
      "epoch": 3.3021595113137745,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00025872990684932705,
      "loss": 0.9169,
      "step": 38381
    },
    {
      "epoch": 3.302245547621096,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002587062229295411,
      "loss": 1.2432,
      "step": 38382
    },
    {
      "epoch": 3.3023315839284177,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002586825397154925,
      "loss": 0.5624,
      "step": 38383
    },
    {
      "epoch": 3.3024176202357394,
      "grad_norm": 4.40625,
      "learning_rate": 0.0002586588572072507,
      "loss": 1.2884,
      "step": 38384
    },
    {
      "epoch": 3.302503656543061,
      "grad_norm": 0.875,
      "learning_rate": 0.0002586351754048847,
      "loss": 1.1348,
      "step": 38385
    },
    {
      "epoch": 3.3025896928503826,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00025861149430846374,
      "loss": 1.044,
      "step": 38386
    },
    {
      "epoch": 3.3026757291577047,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002585878139180575,
      "loss": 1.0375,
      "step": 38387
    },
    {
      "epoch": 3.3027617654650263,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002585641342337347,
      "loss": 1.3811,
      "step": 38388
    },
    {
      "epoch": 3.302847801772348,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00025854045525556494,
      "loss": 0.6383,
      "step": 38389
    },
    {
      "epoch": 3.3029338380796696,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002585167769836174,
      "loss": 1.0829,
      "step": 38390
    },
    {
      "epoch": 3.303019874386991,
      "grad_norm": 81.5,
      "learning_rate": 0.00025849309941796127,
      "loss": 0.7758,
      "step": 38391
    },
    {
      "epoch": 3.303105910694313,
      "grad_norm": 2.203125,
      "learning_rate": 0.000258469422558666,
      "loss": 2.0673,
      "step": 38392
    },
    {
      "epoch": 3.303191947001635,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00025844574640580055,
      "loss": 0.6733,
      "step": 38393
    },
    {
      "epoch": 3.3032779833089565,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00025842207095943425,
      "loss": 1.0067,
      "step": 38394
    },
    {
      "epoch": 3.303364019616278,
      "grad_norm": 1.34375,
      "learning_rate": 0.00025839839621963665,
      "loss": 1.3241,
      "step": 38395
    },
    {
      "epoch": 3.3034500559235997,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002583747221864766,
      "loss": 1.2756,
      "step": 38396
    },
    {
      "epoch": 3.3035360922309214,
      "grad_norm": 62.0,
      "learning_rate": 0.00025835104886002343,
      "loss": 1.431,
      "step": 38397
    },
    {
      "epoch": 3.303622128538243,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00025832737624034664,
      "loss": 0.8617,
      "step": 38398
    },
    {
      "epoch": 3.303708164845565,
      "grad_norm": 1.125,
      "learning_rate": 0.000258303704327515,
      "loss": 0.841,
      "step": 38399
    },
    {
      "epoch": 3.3037942011528867,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0002582800331215982,
      "loss": 1.3376,
      "step": 38400
    },
    {
      "epoch": 3.3038802374602083,
      "grad_norm": 2.609375,
      "learning_rate": 0.0002582563626226652,
      "loss": 1.453,
      "step": 38401
    },
    {
      "epoch": 3.30396627376753,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002582326928307853,
      "loss": 0.8526,
      "step": 38402
    },
    {
      "epoch": 3.3040523100748516,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00025820902374602795,
      "loss": 1.1795,
      "step": 38403
    },
    {
      "epoch": 3.304138346382173,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000258185355368462,
      "loss": 0.9272,
      "step": 38404
    },
    {
      "epoch": 3.304224382689495,
      "grad_norm": 1.15625,
      "learning_rate": 0.00025816168769815685,
      "loss": 1.503,
      "step": 38405
    },
    {
      "epoch": 3.304310418996817,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002581380207351819,
      "loss": 1.5336,
      "step": 38406
    },
    {
      "epoch": 3.3043964553041385,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002581143544796061,
      "loss": 0.7392,
      "step": 38407
    },
    {
      "epoch": 3.30448249161146,
      "grad_norm": 1.265625,
      "learning_rate": 0.00025809068893149886,
      "loss": 1.0022,
      "step": 38408
    },
    {
      "epoch": 3.3045685279187818,
      "grad_norm": 2.484375,
      "learning_rate": 0.00025806702409092913,
      "loss": 0.5547,
      "step": 38409
    },
    {
      "epoch": 3.3046545642261034,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00025804335995796633,
      "loss": 0.9324,
      "step": 38410
    },
    {
      "epoch": 3.304740600533425,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002580196965326798,
      "loss": 1.4077,
      "step": 38411
    },
    {
      "epoch": 3.3048266368407466,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002579960338151385,
      "loss": 1.1052,
      "step": 38412
    },
    {
      "epoch": 3.3049126731480687,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00025797237180541163,
      "loss": 1.4426,
      "step": 38413
    },
    {
      "epoch": 3.3049987094553903,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002579487105035687,
      "loss": 0.8089,
      "step": 38414
    },
    {
      "epoch": 3.305084745762712,
      "grad_norm": 2.546875,
      "learning_rate": 0.0002579250499096786,
      "loss": 1.2749,
      "step": 38415
    },
    {
      "epoch": 3.3051707820700336,
      "grad_norm": 11.9375,
      "learning_rate": 0.0002579013900238105,
      "loss": 1.1775,
      "step": 38416
    },
    {
      "epoch": 3.305256818377355,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002578777308460341,
      "loss": 1.0151,
      "step": 38417
    },
    {
      "epoch": 3.305342854684677,
      "grad_norm": 1.296875,
      "learning_rate": 0.000257854072376418,
      "loss": 1.2748,
      "step": 38418
    },
    {
      "epoch": 3.3054288909919984,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002578304146150318,
      "loss": 1.3109,
      "step": 38419
    },
    {
      "epoch": 3.3055149272993205,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0002578067575619444,
      "loss": 0.9715,
      "step": 38420
    },
    {
      "epoch": 3.305600963606642,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000257783101217225,
      "loss": 0.5183,
      "step": 38421
    },
    {
      "epoch": 3.3056869999139638,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002577594455809433,
      "loss": 0.6881,
      "step": 38422
    },
    {
      "epoch": 3.3057730362212854,
      "grad_norm": 1.234375,
      "learning_rate": 0.00025773579065316776,
      "loss": 1.1099,
      "step": 38423
    },
    {
      "epoch": 3.305859072528607,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002577121364339681,
      "loss": 1.1561,
      "step": 38424
    },
    {
      "epoch": 3.3059451088359286,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002576884829234133,
      "loss": 1.3464,
      "step": 38425
    },
    {
      "epoch": 3.3060311451432502,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002576648301215725,
      "loss": 1.9753,
      "step": 38426
    },
    {
      "epoch": 3.3061171814505723,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002576411780285149,
      "loss": 1.3583,
      "step": 38427
    },
    {
      "epoch": 3.306203217757894,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00025761752664430995,
      "loss": 0.8868,
      "step": 38428
    },
    {
      "epoch": 3.3062892540652156,
      "grad_norm": 1.140625,
      "learning_rate": 0.00025759387596902634,
      "loss": 1.3666,
      "step": 38429
    },
    {
      "epoch": 3.306375290372537,
      "grad_norm": 1.25,
      "learning_rate": 0.0002575702260027335,
      "loss": 1.5697,
      "step": 38430
    },
    {
      "epoch": 3.306461326679859,
      "grad_norm": 1.171875,
      "learning_rate": 0.00025754657674550085,
      "loss": 1.111,
      "step": 38431
    },
    {
      "epoch": 3.3065473629871804,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002575229281973971,
      "loss": 0.9663,
      "step": 38432
    },
    {
      "epoch": 3.306633399294502,
      "grad_norm": 140.0,
      "learning_rate": 0.0002574992803584918,
      "loss": 0.9561,
      "step": 38433
    },
    {
      "epoch": 3.306719435601824,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002574756332288537,
      "loss": 0.9763,
      "step": 38434
    },
    {
      "epoch": 3.3068054719091458,
      "grad_norm": 0.875,
      "learning_rate": 0.00025745198680855226,
      "loss": 0.6872,
      "step": 38435
    },
    {
      "epoch": 3.3068915082164674,
      "grad_norm": 2.90625,
      "learning_rate": 0.00025742834109765677,
      "loss": 1.3178,
      "step": 38436
    },
    {
      "epoch": 3.306977544523789,
      "grad_norm": 2.8125,
      "learning_rate": 0.000257404696096236,
      "loss": 0.9678,
      "step": 38437
    },
    {
      "epoch": 3.3070635808311106,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00025738105180435935,
      "loss": 1.1846,
      "step": 38438
    },
    {
      "epoch": 3.3071496171384323,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00025735740822209593,
      "loss": 1.3631,
      "step": 38439
    },
    {
      "epoch": 3.307235653445754,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002573337653495149,
      "loss": 0.6042,
      "step": 38440
    },
    {
      "epoch": 3.307321689753076,
      "grad_norm": 3.703125,
      "learning_rate": 0.0002573101231866856,
      "loss": 0.7478,
      "step": 38441
    },
    {
      "epoch": 3.3074077260603976,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00025728648173367673,
      "loss": 1.223,
      "step": 38442
    },
    {
      "epoch": 3.307493762367719,
      "grad_norm": 1.140625,
      "learning_rate": 0.00025726284099055776,
      "loss": 1.1079,
      "step": 38443
    },
    {
      "epoch": 3.307579798675041,
      "grad_norm": 1.0625,
      "learning_rate": 0.00025723920095739795,
      "loss": 1.1497,
      "step": 38444
    },
    {
      "epoch": 3.3076658349823624,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002572155616342661,
      "loss": 0.7459,
      "step": 38445
    },
    {
      "epoch": 3.307751871289684,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002571919230212314,
      "loss": 1.1417,
      "step": 38446
    },
    {
      "epoch": 3.3078379075970057,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00025716828511836344,
      "loss": 1.1188,
      "step": 38447
    },
    {
      "epoch": 3.3079239439043278,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002571446479257307,
      "loss": 0.8475,
      "step": 38448
    },
    {
      "epoch": 3.3080099802116494,
      "grad_norm": 1.15625,
      "learning_rate": 0.00025712101144340294,
      "loss": 1.0255,
      "step": 38449
    },
    {
      "epoch": 3.308096016518971,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002570973756714488,
      "loss": 1.4053,
      "step": 38450
    },
    {
      "epoch": 3.3081820528262926,
      "grad_norm": 0.96875,
      "learning_rate": 0.00025707374060993744,
      "loss": 1.3756,
      "step": 38451
    },
    {
      "epoch": 3.3082680891336143,
      "grad_norm": 5.28125,
      "learning_rate": 0.00025705010625893843,
      "loss": 1.0578,
      "step": 38452
    },
    {
      "epoch": 3.3083541254409363,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002570264726185204,
      "loss": 1.4697,
      "step": 38453
    },
    {
      "epoch": 3.308440161748258,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002570028396887527,
      "loss": 1.0517,
      "step": 38454
    },
    {
      "epoch": 3.3085261980555796,
      "grad_norm": 2704.0,
      "learning_rate": 0.0002569792074697046,
      "loss": 1.1454,
      "step": 38455
    },
    {
      "epoch": 3.308612234362901,
      "grad_norm": 416.0,
      "learning_rate": 0.00025695557596144494,
      "loss": 1.1027,
      "step": 38456
    },
    {
      "epoch": 3.308698270670223,
      "grad_norm": 1.03125,
      "learning_rate": 0.00025693194516404285,
      "loss": 1.2265,
      "step": 38457
    },
    {
      "epoch": 3.3087843069775444,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002569083150775678,
      "loss": 1.4067,
      "step": 38458
    },
    {
      "epoch": 3.308870343284866,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00025688468570208846,
      "loss": 1.3401,
      "step": 38459
    },
    {
      "epoch": 3.308956379592188,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00025686105703767427,
      "loss": 1.2735,
      "step": 38460
    },
    {
      "epoch": 3.3090424158995098,
      "grad_norm": 1.4375,
      "learning_rate": 0.00025683742908439404,
      "loss": 1.8631,
      "step": 38461
    },
    {
      "epoch": 3.3091284522068314,
      "grad_norm": 1.4375,
      "learning_rate": 0.00025681380184231704,
      "loss": 0.8506,
      "step": 38462
    },
    {
      "epoch": 3.309214488514153,
      "grad_norm": 1.828125,
      "learning_rate": 0.0002567901753115126,
      "loss": 1.2107,
      "step": 38463
    },
    {
      "epoch": 3.3093005248214746,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00025676654949204937,
      "loss": 1.881,
      "step": 38464
    },
    {
      "epoch": 3.3093865611287963,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002567429243839966,
      "loss": 0.7327,
      "step": 38465
    },
    {
      "epoch": 3.309472597436118,
      "grad_norm": 1.265625,
      "learning_rate": 0.00025671929998742377,
      "loss": 1.4218,
      "step": 38466
    },
    {
      "epoch": 3.30955863374344,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0002566956763023994,
      "loss": 1.3701,
      "step": 38467
    },
    {
      "epoch": 3.3096446700507616,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002566720533289929,
      "loss": 1.3612,
      "step": 38468
    },
    {
      "epoch": 3.309730706358083,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002566484310672735,
      "loss": 0.8474,
      "step": 38469
    },
    {
      "epoch": 3.309816742665405,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00025662480951730996,
      "loss": 0.8853,
      "step": 38470
    },
    {
      "epoch": 3.3099027789727264,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002566011886791715,
      "loss": 1.8545,
      "step": 38471
    },
    {
      "epoch": 3.309988815280048,
      "grad_norm": 1.03125,
      "learning_rate": 0.00025657756855292736,
      "loss": 0.7434,
      "step": 38472
    },
    {
      "epoch": 3.3100748515873697,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002565539491386464,
      "loss": 0.7451,
      "step": 38473
    },
    {
      "epoch": 3.3101608878946918,
      "grad_norm": 8.625,
      "learning_rate": 0.0002565303304363979,
      "loss": 1.4761,
      "step": 38474
    },
    {
      "epoch": 3.3102469242020134,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002565067124462505,
      "loss": 1.218,
      "step": 38475
    },
    {
      "epoch": 3.310332960509335,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002564830951682737,
      "loss": 0.9806,
      "step": 38476
    },
    {
      "epoch": 3.3104189968166566,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002564594786025368,
      "loss": 1.022,
      "step": 38477
    },
    {
      "epoch": 3.3105050331239783,
      "grad_norm": 2.28125,
      "learning_rate": 0.0002564358627491082,
      "loss": 0.6039,
      "step": 38478
    },
    {
      "epoch": 3.3105910694313,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002564122476080577,
      "loss": 1.2838,
      "step": 38479
    },
    {
      "epoch": 3.3106771057386215,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002563886331794538,
      "loss": 0.9424,
      "step": 38480
    },
    {
      "epoch": 3.3107631420459436,
      "grad_norm": 0.9375,
      "learning_rate": 0.00025636501946336564,
      "loss": 1.0931,
      "step": 38481
    },
    {
      "epoch": 3.310849178353265,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0002563414064598627,
      "loss": 0.8918,
      "step": 38482
    },
    {
      "epoch": 3.310935214660587,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002563177941690136,
      "loss": 1.7137,
      "step": 38483
    },
    {
      "epoch": 3.3110212509679084,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00025629418259088746,
      "loss": 1.1062,
      "step": 38484
    },
    {
      "epoch": 3.31110728727523,
      "grad_norm": 70.0,
      "learning_rate": 0.0002562705717255537,
      "loss": 1.4422,
      "step": 38485
    },
    {
      "epoch": 3.3111933235825517,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0002562469615730809,
      "loss": 1.407,
      "step": 38486
    },
    {
      "epoch": 3.3112793598898733,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002562233521335383,
      "loss": 1.2303,
      "step": 38487
    },
    {
      "epoch": 3.3113653961971954,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002561997434069952,
      "loss": 0.7574,
      "step": 38488
    },
    {
      "epoch": 3.311451432504517,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00025617613539352023,
      "loss": 1.3286,
      "step": 38489
    },
    {
      "epoch": 3.3115374688118386,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00025615252809318284,
      "loss": 1.0721,
      "step": 38490
    },
    {
      "epoch": 3.3116235051191603,
      "grad_norm": 4.21875,
      "learning_rate": 0.0002561289215060517,
      "loss": 0.8814,
      "step": 38491
    },
    {
      "epoch": 3.311709541426482,
      "grad_norm": 0.890625,
      "learning_rate": 0.000256105315632196,
      "loss": 1.2752,
      "step": 38492
    },
    {
      "epoch": 3.3117955777338035,
      "grad_norm": 3.03125,
      "learning_rate": 0.00025608171047168503,
      "loss": 1.7123,
      "step": 38493
    },
    {
      "epoch": 3.311881614041125,
      "grad_norm": 35.0,
      "learning_rate": 0.00025605810602458747,
      "loss": 1.5767,
      "step": 38494
    },
    {
      "epoch": 3.311967650348447,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00025603450229097247,
      "loss": 0.7815,
      "step": 38495
    },
    {
      "epoch": 3.312053686655769,
      "grad_norm": 1.046875,
      "learning_rate": 0.00025601089927090924,
      "loss": 1.0451,
      "step": 38496
    },
    {
      "epoch": 3.3121397229630904,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0002559872969644666,
      "loss": 1.021,
      "step": 38497
    },
    {
      "epoch": 3.312225759270412,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0002559636953717136,
      "loss": 0.9262,
      "step": 38498
    },
    {
      "epoch": 3.3123117955777337,
      "grad_norm": 8.6875,
      "learning_rate": 0.0002559400944927195,
      "loss": 0.8829,
      "step": 38499
    },
    {
      "epoch": 3.3123978318850553,
      "grad_norm": 1.265625,
      "learning_rate": 0.000255916494327553,
      "loss": 0.7275,
      "step": 38500
    },
    {
      "epoch": 3.312483868192377,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00025589289487628323,
      "loss": 0.8614,
      "step": 38501
    },
    {
      "epoch": 3.312569904499699,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002558692961389795,
      "loss": 0.6483,
      "step": 38502
    },
    {
      "epoch": 3.3126559408070206,
      "grad_norm": 5.25,
      "learning_rate": 0.0002558456981157104,
      "loss": 0.7225,
      "step": 38503
    },
    {
      "epoch": 3.3127419771143423,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00025582210080654535,
      "loss": 1.3836,
      "step": 38504
    },
    {
      "epoch": 3.312828013421664,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00025579850421155294,
      "loss": 0.437,
      "step": 38505
    },
    {
      "epoch": 3.3129140497289855,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002557749083308024,
      "loss": 1.0858,
      "step": 38506
    },
    {
      "epoch": 3.313000086036307,
      "grad_norm": 147.0,
      "learning_rate": 0.000255751313164363,
      "loss": 1.3799,
      "step": 38507
    },
    {
      "epoch": 3.313086122343629,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002557277187123033,
      "loss": 0.7961,
      "step": 38508
    },
    {
      "epoch": 3.313172158650951,
      "grad_norm": 4.375,
      "learning_rate": 0.00025570412497469253,
      "loss": 1.024,
      "step": 38509
    },
    {
      "epoch": 3.3132581949582725,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002556805319515998,
      "loss": 0.9267,
      "step": 38510
    },
    {
      "epoch": 3.313344231265594,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002556569396430939,
      "loss": 1.1501,
      "step": 38511
    },
    {
      "epoch": 3.3134302675729157,
      "grad_norm": 1.609375,
      "learning_rate": 0.0002556333480492439,
      "loss": 1.4661,
      "step": 38512
    },
    {
      "epoch": 3.3135163038802373,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002556097571701188,
      "loss": 0.9625,
      "step": 38513
    },
    {
      "epoch": 3.3136023401875594,
      "grad_norm": 1.453125,
      "learning_rate": 0.00025558616700578765,
      "loss": 1.596,
      "step": 38514
    },
    {
      "epoch": 3.313688376494881,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00025556257755631964,
      "loss": 0.8073,
      "step": 38515
    },
    {
      "epoch": 3.3137744128022026,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002555389888217833,
      "loss": 0.9762,
      "step": 38516
    },
    {
      "epoch": 3.3138604491095243,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002555154008022479,
      "loss": 0.4928,
      "step": 38517
    },
    {
      "epoch": 3.313946485416846,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00025549181349778253,
      "loss": 1.2179,
      "step": 38518
    },
    {
      "epoch": 3.3140325217241675,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00025546822690845596,
      "loss": 0.5179,
      "step": 38519
    },
    {
      "epoch": 3.314118558031489,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002554446410343374,
      "loss": 1.6337,
      "step": 38520
    },
    {
      "epoch": 3.314204594338811,
      "grad_norm": 1.4375,
      "learning_rate": 0.00025542105587549556,
      "loss": 1.1055,
      "step": 38521
    },
    {
      "epoch": 3.314290630646133,
      "grad_norm": 1.203125,
      "learning_rate": 0.00025539747143199954,
      "loss": 1.3381,
      "step": 38522
    },
    {
      "epoch": 3.3143766669534545,
      "grad_norm": 1.390625,
      "learning_rate": 0.00025537388770391855,
      "loss": 1.3894,
      "step": 38523
    },
    {
      "epoch": 3.314462703260776,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00025535030469132116,
      "loss": 0.7589,
      "step": 38524
    },
    {
      "epoch": 3.3145487395680977,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002553267223942765,
      "loss": 0.6644,
      "step": 38525
    },
    {
      "epoch": 3.3146347758754193,
      "grad_norm": 2.421875,
      "learning_rate": 0.00025530314081285385,
      "loss": 1.0082,
      "step": 38526
    },
    {
      "epoch": 3.314720812182741,
      "grad_norm": 1.640625,
      "learning_rate": 0.00025527955994712173,
      "loss": 1.2924,
      "step": 38527
    },
    {
      "epoch": 3.314806848490063,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002552559797971492,
      "loss": 1.3676,
      "step": 38528
    },
    {
      "epoch": 3.3148928847973846,
      "grad_norm": 7.71875,
      "learning_rate": 0.00025523240036300565,
      "loss": 0.6531,
      "step": 38529
    },
    {
      "epoch": 3.3149789211047063,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00025520882164475943,
      "loss": 1.0822,
      "step": 38530
    },
    {
      "epoch": 3.315064957412028,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00025518524364248006,
      "loss": 0.9188,
      "step": 38531
    },
    {
      "epoch": 3.3151509937193495,
      "grad_norm": 1.3125,
      "learning_rate": 0.000255161666356236,
      "loss": 1.7639,
      "step": 38532
    },
    {
      "epoch": 3.315237030026671,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002551380897860965,
      "loss": 1.4096,
      "step": 38533
    },
    {
      "epoch": 3.3153230663339928,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002551145139321306,
      "loss": 1.6657,
      "step": 38534
    },
    {
      "epoch": 3.315409102641315,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00025509093879440693,
      "loss": 1.432,
      "step": 38535
    },
    {
      "epoch": 3.3154951389486365,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002550673643729947,
      "loss": 0.7502,
      "step": 38536
    },
    {
      "epoch": 3.315581175255958,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002550437906679629,
      "loss": 1.0392,
      "step": 38537
    },
    {
      "epoch": 3.3156672115632797,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002550202176793802,
      "loss": 1.5022,
      "step": 38538
    },
    {
      "epoch": 3.3157532478706013,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002549966454073158,
      "loss": 0.8445,
      "step": 38539
    },
    {
      "epoch": 3.315839284177923,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002549730738518387,
      "loss": 0.425,
      "step": 38540
    },
    {
      "epoch": 3.3159253204852446,
      "grad_norm": 8.5625,
      "learning_rate": 0.00025494950301301744,
      "loss": 1.3621,
      "step": 38541
    },
    {
      "epoch": 3.3160113567925666,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002549259328909214,
      "loss": 0.5374,
      "step": 38542
    },
    {
      "epoch": 3.3160973930998883,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00025490236348561936,
      "loss": 0.9379,
      "step": 38543
    },
    {
      "epoch": 3.31618342940721,
      "grad_norm": 4.96875,
      "learning_rate": 0.0002548787947971801,
      "loss": 1.0528,
      "step": 38544
    },
    {
      "epoch": 3.3162694657145315,
      "grad_norm": 1.046875,
      "learning_rate": 0.00025485522682567284,
      "loss": 1.3547,
      "step": 38545
    },
    {
      "epoch": 3.316355502021853,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00025483165957116625,
      "loss": 1.2567,
      "step": 38546
    },
    {
      "epoch": 3.3164415383291748,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002548080930337293,
      "loss": 1.5677,
      "step": 38547
    },
    {
      "epoch": 3.3165275746364964,
      "grad_norm": 45.75,
      "learning_rate": 0.00025478452721343126,
      "loss": 0.7401,
      "step": 38548
    },
    {
      "epoch": 3.3166136109438185,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002547609621103404,
      "loss": 0.8498,
      "step": 38549
    },
    {
      "epoch": 3.31669964725114,
      "grad_norm": 3.375,
      "learning_rate": 0.0002547373977245264,
      "loss": 1.2984,
      "step": 38550
    },
    {
      "epoch": 3.3167856835584617,
      "grad_norm": 1.203125,
      "learning_rate": 0.00025471383405605756,
      "loss": 1.0597,
      "step": 38551
    },
    {
      "epoch": 3.3168717198657833,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002546902711050031,
      "loss": 1.0783,
      "step": 38552
    },
    {
      "epoch": 3.316957756173105,
      "grad_norm": 1.171875,
      "learning_rate": 0.00025466670887143206,
      "loss": 1.023,
      "step": 38553
    },
    {
      "epoch": 3.3170437924804266,
      "grad_norm": 1.09375,
      "learning_rate": 0.00025464314735541306,
      "loss": 0.8598,
      "step": 38554
    },
    {
      "epoch": 3.317129828787748,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000254619586557015,
      "loss": 0.6276,
      "step": 38555
    },
    {
      "epoch": 3.3172158650950703,
      "grad_norm": 1.0,
      "learning_rate": 0.0002545960264763072,
      "loss": 0.9196,
      "step": 38556
    },
    {
      "epoch": 3.317301901402392,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00025457246711335814,
      "loss": 0.8973,
      "step": 38557
    },
    {
      "epoch": 3.3173879377097135,
      "grad_norm": 17.75,
      "learning_rate": 0.0002545489084682368,
      "loss": 0.8204,
      "step": 38558
    },
    {
      "epoch": 3.317473974017035,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00025452535054101247,
      "loss": 1.1118,
      "step": 38559
    },
    {
      "epoch": 3.3175600103243568,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002545017933317534,
      "loss": 1.0516,
      "step": 38560
    },
    {
      "epoch": 3.3176460466316784,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002544782368405291,
      "loss": 1.2838,
      "step": 38561
    },
    {
      "epoch": 3.317732082939,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000254454681067408,
      "loss": 0.9859,
      "step": 38562
    },
    {
      "epoch": 3.317818119246322,
      "grad_norm": 11.9375,
      "learning_rate": 0.00025443112601245926,
      "loss": 1.5062,
      "step": 38563
    },
    {
      "epoch": 3.3179041555536437,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002544075716757519,
      "loss": 1.2909,
      "step": 38564
    },
    {
      "epoch": 3.3179901918609653,
      "grad_norm": 3.125,
      "learning_rate": 0.0002543840180573544,
      "loss": 0.9354,
      "step": 38565
    },
    {
      "epoch": 3.318076228168287,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002543604651573359,
      "loss": 0.8741,
      "step": 38566
    },
    {
      "epoch": 3.3181622644756086,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00025433691297576544,
      "loss": 1.2562,
      "step": 38567
    },
    {
      "epoch": 3.31824830078293,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002543133615127115,
      "loss": 1.9086,
      "step": 38568
    },
    {
      "epoch": 3.3183343370902523,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002542898107682433,
      "loss": 0.9155,
      "step": 38569
    },
    {
      "epoch": 3.318420373397574,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002542662607424297,
      "loss": 0.9095,
      "step": 38570
    },
    {
      "epoch": 3.3185064097048955,
      "grad_norm": 1.3125,
      "learning_rate": 0.00025424271143533936,
      "loss": 1.2004,
      "step": 38571
    },
    {
      "epoch": 3.318592446012217,
      "grad_norm": 1.015625,
      "learning_rate": 0.00025421916284704143,
      "loss": 0.7988,
      "step": 38572
    },
    {
      "epoch": 3.3186784823195388,
      "grad_norm": 0.875,
      "learning_rate": 0.0002541956149776045,
      "loss": 0.8078,
      "step": 38573
    },
    {
      "epoch": 3.3187645186268604,
      "grad_norm": 1.375,
      "learning_rate": 0.00025417206782709755,
      "loss": 0.8857,
      "step": 38574
    },
    {
      "epoch": 3.3188505549341825,
      "grad_norm": 19.75,
      "learning_rate": 0.00025414852139558973,
      "loss": 1.0224,
      "step": 38575
    },
    {
      "epoch": 3.318936591241504,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002541249756831495,
      "loss": 0.613,
      "step": 38576
    },
    {
      "epoch": 3.3190226275488257,
      "grad_norm": 8.0625,
      "learning_rate": 0.00025410143068984583,
      "loss": 0.997,
      "step": 38577
    },
    {
      "epoch": 3.3191086638561473,
      "grad_norm": 5760.0,
      "learning_rate": 0.0002540778864157479,
      "loss": 0.7597,
      "step": 38578
    },
    {
      "epoch": 3.319194700163469,
      "grad_norm": 1.65625,
      "learning_rate": 0.0002540543428609241,
      "loss": 1.4841,
      "step": 38579
    },
    {
      "epoch": 3.3192807364707906,
      "grad_norm": 1.453125,
      "learning_rate": 0.00025403080002544355,
      "loss": 0.9073,
      "step": 38580
    },
    {
      "epoch": 3.319366772778112,
      "grad_norm": 71.5,
      "learning_rate": 0.00025400725790937517,
      "loss": 0.9613,
      "step": 38581
    },
    {
      "epoch": 3.3194528090854343,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00025398371651278763,
      "loss": 1.2999,
      "step": 38582
    },
    {
      "epoch": 3.319538845392756,
      "grad_norm": 3.453125,
      "learning_rate": 0.00025396017583574984,
      "loss": 0.8038,
      "step": 38583
    },
    {
      "epoch": 3.3196248817000775,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002539366358783308,
      "loss": 0.5478,
      "step": 38584
    },
    {
      "epoch": 3.319710918007399,
      "grad_norm": 1.03125,
      "learning_rate": 0.00025391309664059915,
      "loss": 1.3606,
      "step": 38585
    },
    {
      "epoch": 3.3197969543147208,
      "grad_norm": 34.5,
      "learning_rate": 0.00025388955812262384,
      "loss": 1.4003,
      "step": 38586
    },
    {
      "epoch": 3.3198829906220424,
      "grad_norm": 153.0,
      "learning_rate": 0.0002538660203244736,
      "loss": 1.4505,
      "step": 38587
    },
    {
      "epoch": 3.319969026929364,
      "grad_norm": 3.40625,
      "learning_rate": 0.00025384248324621746,
      "loss": 0.5458,
      "step": 38588
    },
    {
      "epoch": 3.320055063236686,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00025381894688792435,
      "loss": 1.6741,
      "step": 38589
    },
    {
      "epoch": 3.3201410995440077,
      "grad_norm": 1.59375,
      "learning_rate": 0.00025379541124966276,
      "loss": 0.914,
      "step": 38590
    },
    {
      "epoch": 3.3202271358513293,
      "grad_norm": 2.59375,
      "learning_rate": 0.0002537718763315019,
      "loss": 1.2388,
      "step": 38591
    },
    {
      "epoch": 3.320313172158651,
      "grad_norm": 1.0,
      "learning_rate": 0.00025374834213351015,
      "loss": 0.7873,
      "step": 38592
    },
    {
      "epoch": 3.3203992084659726,
      "grad_norm": 11.625,
      "learning_rate": 0.00025372480865575676,
      "loss": 1.2792,
      "step": 38593
    },
    {
      "epoch": 3.320485244773294,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002537012758983105,
      "loss": 0.7792,
      "step": 38594
    },
    {
      "epoch": 3.320571281080616,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00025367774386123986,
      "loss": 0.6785,
      "step": 38595
    },
    {
      "epoch": 3.320657317387938,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00025365421254461404,
      "loss": 0.808,
      "step": 38596
    },
    {
      "epoch": 3.3207433536952595,
      "grad_norm": 2.828125,
      "learning_rate": 0.0002536306819485018,
      "loss": 1.1956,
      "step": 38597
    },
    {
      "epoch": 3.320829390002581,
      "grad_norm": 2.421875,
      "learning_rate": 0.0002536071520729718,
      "loss": 1.0054,
      "step": 38598
    },
    {
      "epoch": 3.3209154263099028,
      "grad_norm": 3.3125,
      "learning_rate": 0.00025358362291809296,
      "loss": 1.3471,
      "step": 38599
    },
    {
      "epoch": 3.3210014626172244,
      "grad_norm": 1.375,
      "learning_rate": 0.00025356009448393424,
      "loss": 0.784,
      "step": 38600
    },
    {
      "epoch": 3.321087498924546,
      "grad_norm": 1.03125,
      "learning_rate": 0.00025353656677056414,
      "loss": 0.9309,
      "step": 38601
    },
    {
      "epoch": 3.3211735352318676,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00025351303977805186,
      "loss": 1.6635,
      "step": 38602
    },
    {
      "epoch": 3.3212595715391897,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002534895135064658,
      "loss": 0.789,
      "step": 38603
    },
    {
      "epoch": 3.3213456078465113,
      "grad_norm": 1.390625,
      "learning_rate": 0.00025346598795587496,
      "loss": 1.0875,
      "step": 38604
    },
    {
      "epoch": 3.321431644153833,
      "grad_norm": 1.796875,
      "learning_rate": 0.00025344246312634835,
      "loss": 0.8879,
      "step": 38605
    },
    {
      "epoch": 3.3215176804611546,
      "grad_norm": 0.734375,
      "learning_rate": 0.00025341893901795437,
      "loss": 0.5226,
      "step": 38606
    },
    {
      "epoch": 3.321603716768476,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002533954156307621,
      "loss": 0.9916,
      "step": 38607
    },
    {
      "epoch": 3.321689753075798,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00025337189296484043,
      "loss": 0.9597,
      "step": 38608
    },
    {
      "epoch": 3.3217757893831195,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002533483710202578,
      "loss": 1.216,
      "step": 38609
    },
    {
      "epoch": 3.3218618256904415,
      "grad_norm": 1.046875,
      "learning_rate": 0.00025332484979708316,
      "loss": 0.6663,
      "step": 38610
    },
    {
      "epoch": 3.321947861997763,
      "grad_norm": 1712.0,
      "learning_rate": 0.0002533013292953856,
      "loss": 0.5531,
      "step": 38611
    },
    {
      "epoch": 3.3220338983050848,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002532778095152335,
      "loss": 1.6302,
      "step": 38612
    },
    {
      "epoch": 3.3221199346124064,
      "grad_norm": 5.78125,
      "learning_rate": 0.00025325429045669603,
      "loss": 0.6635,
      "step": 38613
    },
    {
      "epoch": 3.322205970919728,
      "grad_norm": 7.1875,
      "learning_rate": 0.0002532307721198416,
      "loss": 0.8856,
      "step": 38614
    },
    {
      "epoch": 3.3222920072270496,
      "grad_norm": 1.5625,
      "learning_rate": 0.0002532072545047391,
      "loss": 1.3861,
      "step": 38615
    },
    {
      "epoch": 3.3223780435343713,
      "grad_norm": 5.15625,
      "learning_rate": 0.00025318373761145757,
      "loss": 0.7163,
      "step": 38616
    },
    {
      "epoch": 3.3224640798416933,
      "grad_norm": 1.453125,
      "learning_rate": 0.00025316022144006544,
      "loss": 1.9365,
      "step": 38617
    },
    {
      "epoch": 3.322550116149015,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002531367059906317,
      "loss": 1.2466,
      "step": 38618
    },
    {
      "epoch": 3.3226361524563366,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00025311319126322516,
      "loss": 1.0272,
      "step": 38619
    },
    {
      "epoch": 3.322722188763658,
      "grad_norm": 2.328125,
      "learning_rate": 0.0002530896772579144,
      "loss": 0.9314,
      "step": 38620
    },
    {
      "epoch": 3.32280822507098,
      "grad_norm": 5.53125,
      "learning_rate": 0.0002530661639747683,
      "loss": 1.1842,
      "step": 38621
    },
    {
      "epoch": 3.3228942613783015,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002530426514138559,
      "loss": 1.246,
      "step": 38622
    },
    {
      "epoch": 3.322980297685623,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00025301913957524524,
      "loss": 0.6731,
      "step": 38623
    },
    {
      "epoch": 3.323066333992945,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000252995628459006,
      "loss": 0.6439,
      "step": 38624
    },
    {
      "epoch": 3.323152370300267,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00025297211806520634,
      "loss": 1.3445,
      "step": 38625
    },
    {
      "epoch": 3.3232384066075884,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00025294860839391507,
      "loss": 0.5932,
      "step": 38626
    },
    {
      "epoch": 3.32332444291491,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00025292509944520135,
      "loss": 1.5609,
      "step": 38627
    },
    {
      "epoch": 3.3234104792222317,
      "grad_norm": 1.984375,
      "learning_rate": 0.0002529015912191335,
      "loss": 1.1296,
      "step": 38628
    },
    {
      "epoch": 3.3234965155295537,
      "grad_norm": 1.140625,
      "learning_rate": 0.00025287808371578035,
      "loss": 0.7663,
      "step": 38629
    },
    {
      "epoch": 3.3235825518368753,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00025285457693521103,
      "loss": 1.2253,
      "step": 38630
    },
    {
      "epoch": 3.323668588144197,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002528310708774937,
      "loss": 1.1615,
      "step": 38631
    },
    {
      "epoch": 3.3237546244515186,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00025280756554269776,
      "loss": 1.3894,
      "step": 38632
    },
    {
      "epoch": 3.32384066075884,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002527840609308913,
      "loss": 1.0257,
      "step": 38633
    },
    {
      "epoch": 3.323926697066162,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002527605570421434,
      "loss": 0.7217,
      "step": 38634
    },
    {
      "epoch": 3.3240127333734835,
      "grad_norm": 0.828125,
      "learning_rate": 0.00025273705387652304,
      "loss": 0.8244,
      "step": 38635
    },
    {
      "epoch": 3.3240987696808055,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002527135514340985,
      "loss": 0.8535,
      "step": 38636
    },
    {
      "epoch": 3.324184805988127,
      "grad_norm": 1.515625,
      "learning_rate": 0.0002526900497149386,
      "loss": 0.9229,
      "step": 38637
    },
    {
      "epoch": 3.324270842295449,
      "grad_norm": 17.75,
      "learning_rate": 0.00025266654871911254,
      "loss": 0.9768,
      "step": 38638
    },
    {
      "epoch": 3.3243568786027704,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002526430484466885,
      "loss": 1.379,
      "step": 38639
    },
    {
      "epoch": 3.324442914910092,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002526195488977354,
      "loss": 0.6013,
      "step": 38640
    },
    {
      "epoch": 3.3245289512174137,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002525960500723222,
      "loss": 1.551,
      "step": 38641
    },
    {
      "epoch": 3.3246149875247353,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00025257255197051726,
      "loss": 1.1987,
      "step": 38642
    },
    {
      "epoch": 3.3247010238320573,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002525490545923897,
      "loss": 0.7958,
      "step": 38643
    },
    {
      "epoch": 3.324787060139379,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002525255579380078,
      "loss": 1.3649,
      "step": 38644
    },
    {
      "epoch": 3.3248730964467006,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0002525020620074405,
      "loss": 1.0264,
      "step": 38645
    },
    {
      "epoch": 3.324959132754022,
      "grad_norm": 14.0,
      "learning_rate": 0.00025247856680075676,
      "loss": 0.8255,
      "step": 38646
    },
    {
      "epoch": 3.325045169061344,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002524550723180249,
      "loss": 1.0915,
      "step": 38647
    },
    {
      "epoch": 3.3251312053686655,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00025243157855931375,
      "loss": 0.9928,
      "step": 38648
    },
    {
      "epoch": 3.325217241675987,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0002524080855246923,
      "loss": 0.962,
      "step": 38649
    },
    {
      "epoch": 3.325303277983309,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002523845932142289,
      "loss": 0.8209,
      "step": 38650
    },
    {
      "epoch": 3.325389314290631,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002523611016279923,
      "loss": 0.7954,
      "step": 38651
    },
    {
      "epoch": 3.3254753505979524,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002523376107660516,
      "loss": 1.026,
      "step": 38652
    },
    {
      "epoch": 3.325561386905274,
      "grad_norm": 1.515625,
      "learning_rate": 0.00025231412062847505,
      "loss": 1.926,
      "step": 38653
    },
    {
      "epoch": 3.3256474232125957,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002522906312153315,
      "loss": 1.6276,
      "step": 38654
    },
    {
      "epoch": 3.3257334595199173,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002522671425266898,
      "loss": 1.0332,
      "step": 38655
    },
    {
      "epoch": 3.325819495827239,
      "grad_norm": 1.203125,
      "learning_rate": 0.00025224365456261833,
      "loss": 1.3538,
      "step": 38656
    },
    {
      "epoch": 3.325905532134561,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002522201673231863,
      "loss": 0.8695,
      "step": 38657
    },
    {
      "epoch": 3.3259915684418826,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002521966808084618,
      "loss": 0.8493,
      "step": 38658
    },
    {
      "epoch": 3.326077604749204,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002521731950185138,
      "loss": 0.4352,
      "step": 38659
    },
    {
      "epoch": 3.326163641056526,
      "grad_norm": 1.015625,
      "learning_rate": 0.000252149709953411,
      "loss": 1.8451,
      "step": 38660
    },
    {
      "epoch": 3.3262496773638475,
      "grad_norm": 4.4375,
      "learning_rate": 0.0002521262256132222,
      "loss": 1.7142,
      "step": 38661
    },
    {
      "epoch": 3.326335713671169,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00025210274199801607,
      "loss": 0.5263,
      "step": 38662
    },
    {
      "epoch": 3.3264217499784907,
      "grad_norm": 76.0,
      "learning_rate": 0.00025207925910786114,
      "loss": 1.9304,
      "step": 38663
    },
    {
      "epoch": 3.326507786285813,
      "grad_norm": 0.8515625,
      "learning_rate": 0.000252055776942826,
      "loss": 0.6004,
      "step": 38664
    },
    {
      "epoch": 3.3265938225931344,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002520322955029798,
      "loss": 1.0961,
      "step": 38665
    },
    {
      "epoch": 3.326679858900456,
      "grad_norm": 1.046875,
      "learning_rate": 0.00025200881478839064,
      "loss": 0.6439,
      "step": 38666
    },
    {
      "epoch": 3.3267658952077777,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002519853347991275,
      "loss": 1.3324,
      "step": 38667
    },
    {
      "epoch": 3.3268519315150993,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002519618555352592,
      "loss": 1.2269,
      "step": 38668
    },
    {
      "epoch": 3.326937967822421,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002519383769968541,
      "loss": 0.6237,
      "step": 38669
    },
    {
      "epoch": 3.3270240041297425,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002519148991839809,
      "loss": 0.9932,
      "step": 38670
    },
    {
      "epoch": 3.3271100404370646,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00025189142209670857,
      "loss": 0.6562,
      "step": 38671
    },
    {
      "epoch": 3.327196076744386,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002518679457351055,
      "loss": 0.9686,
      "step": 38672
    },
    {
      "epoch": 3.327282113051708,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002518444700992405,
      "loss": 0.7693,
      "step": 38673
    },
    {
      "epoch": 3.3273681493590295,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000251820995189182,
      "loss": 0.8371,
      "step": 38674
    },
    {
      "epoch": 3.327454185666351,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00025179752100499886,
      "loss": 0.9052,
      "step": 38675
    },
    {
      "epoch": 3.3275402219736727,
      "grad_norm": 1.015625,
      "learning_rate": 0.00025177404754675984,
      "loss": 1.3794,
      "step": 38676
    },
    {
      "epoch": 3.3276262582809943,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00025175057481453334,
      "loss": 2.0292,
      "step": 38677
    },
    {
      "epoch": 3.3277122945883164,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002517271028083881,
      "loss": 0.7799,
      "step": 38678
    },
    {
      "epoch": 3.327798330895638,
      "grad_norm": 2.0,
      "learning_rate": 0.00025170363152839293,
      "loss": 1.5146,
      "step": 38679
    },
    {
      "epoch": 3.3278843672029597,
      "grad_norm": 1.1875,
      "learning_rate": 0.00025168016097461623,
      "loss": 1.122,
      "step": 38680
    },
    {
      "epoch": 3.3279704035102813,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002516566911471266,
      "loss": 1.1634,
      "step": 38681
    },
    {
      "epoch": 3.328056439817603,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00025163322204599327,
      "loss": 0.7103,
      "step": 38682
    },
    {
      "epoch": 3.3281424761249245,
      "grad_norm": 1.15625,
      "learning_rate": 0.00025160975367128415,
      "loss": 1.1943,
      "step": 38683
    },
    {
      "epoch": 3.3282285124322466,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002515862860230684,
      "loss": 1.6977,
      "step": 38684
    },
    {
      "epoch": 3.3283145487395682,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00025156281910141423,
      "loss": 0.6808,
      "step": 38685
    },
    {
      "epoch": 3.32840058504689,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00025153935290639053,
      "loss": 1.2434,
      "step": 38686
    },
    {
      "epoch": 3.3284866213542115,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002515158874380661,
      "loss": 1.7247,
      "step": 38687
    },
    {
      "epoch": 3.328572657661533,
      "grad_norm": 12.5,
      "learning_rate": 0.0002514924226965091,
      "loss": 1.3707,
      "step": 38688
    },
    {
      "epoch": 3.3286586939688547,
      "grad_norm": 304.0,
      "learning_rate": 0.00025146895868178855,
      "loss": 0.967,
      "step": 38689
    },
    {
      "epoch": 3.328744730276177,
      "grad_norm": 1.53125,
      "learning_rate": 0.0002514454953939731,
      "loss": 1.4318,
      "step": 38690
    },
    {
      "epoch": 3.3288307665834984,
      "grad_norm": 2.03125,
      "learning_rate": 0.000251422032833131,
      "loss": 1.3166,
      "step": 38691
    },
    {
      "epoch": 3.32891680289082,
      "grad_norm": 1.015625,
      "learning_rate": 0.00025139857099933116,
      "loss": 1.1158,
      "step": 38692
    },
    {
      "epoch": 3.3290028391981417,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002513751098926423,
      "loss": 1.4248,
      "step": 38693
    },
    {
      "epoch": 3.3290888755054633,
      "grad_norm": 1.546875,
      "learning_rate": 0.0002513516495131327,
      "loss": 1.2626,
      "step": 38694
    },
    {
      "epoch": 3.329174911812785,
      "grad_norm": 1.15625,
      "learning_rate": 0.00025132818986087126,
      "loss": 1.2108,
      "step": 38695
    },
    {
      "epoch": 3.3292609481201065,
      "grad_norm": 1.65625,
      "learning_rate": 0.0002513047309359266,
      "loss": 1.1729,
      "step": 38696
    },
    {
      "epoch": 3.3293469844274286,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002512812727383669,
      "loss": 1.2164,
      "step": 38697
    },
    {
      "epoch": 3.3294330207347502,
      "grad_norm": 1.34375,
      "learning_rate": 0.00025125781526826144,
      "loss": 1.0212,
      "step": 38698
    },
    {
      "epoch": 3.329519057042072,
      "grad_norm": 1.28125,
      "learning_rate": 0.00025123435852567836,
      "loss": 0.8497,
      "step": 38699
    },
    {
      "epoch": 3.3296050933493935,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00025121090251068633,
      "loss": 1.3051,
      "step": 38700
    },
    {
      "epoch": 3.329691129656715,
      "grad_norm": 1.4375,
      "learning_rate": 0.0002511874472233542,
      "loss": 1.5222,
      "step": 38701
    },
    {
      "epoch": 3.3297771659640367,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00025116399266375024,
      "loss": 0.7446,
      "step": 38702
    },
    {
      "epoch": 3.3298632022713583,
      "grad_norm": 1.75,
      "learning_rate": 0.00025114053883194343,
      "loss": 0.9684,
      "step": 38703
    },
    {
      "epoch": 3.3299492385786804,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000251117085728002,
      "loss": 1.0851,
      "step": 38704
    },
    {
      "epoch": 3.330035274886002,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002510936333519945,
      "loss": 0.9713,
      "step": 38705
    },
    {
      "epoch": 3.3301213111933237,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002510701817039901,
      "loss": 0.4654,
      "step": 38706
    },
    {
      "epoch": 3.3302073475006453,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002510467307840567,
      "loss": 1.4125,
      "step": 38707
    },
    {
      "epoch": 3.330293383807967,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002510232805922632,
      "loss": 0.814,
      "step": 38708
    },
    {
      "epoch": 3.3303794201152885,
      "grad_norm": 1.375,
      "learning_rate": 0.0002509998311286784,
      "loss": 1.102,
      "step": 38709
    },
    {
      "epoch": 3.33046545642261,
      "grad_norm": 23.625,
      "learning_rate": 0.0002509763823933705,
      "loss": 1.7419,
      "step": 38710
    },
    {
      "epoch": 3.3305514927299322,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00025095293438640823,
      "loss": 0.9128,
      "step": 38711
    },
    {
      "epoch": 3.330637529037254,
      "grad_norm": 26.625,
      "learning_rate": 0.00025092948710786034,
      "loss": 1.403,
      "step": 38712
    },
    {
      "epoch": 3.3307235653445755,
      "grad_norm": 3.671875,
      "learning_rate": 0.00025090604055779515,
      "loss": 1.7122,
      "step": 38713
    },
    {
      "epoch": 3.330809601651897,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002508825947362815,
      "loss": 1.281,
      "step": 38714
    },
    {
      "epoch": 3.3308956379592187,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00025085914964338756,
      "loss": 1.2429,
      "step": 38715
    },
    {
      "epoch": 3.3309816742665403,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002508357052791822,
      "loss": 0.5522,
      "step": 38716
    },
    {
      "epoch": 3.331067710573862,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002508122616437342,
      "loss": 0.8045,
      "step": 38717
    },
    {
      "epoch": 3.331153746881184,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0002507888187371116,
      "loss": 1.1275,
      "step": 38718
    },
    {
      "epoch": 3.3312397831885057,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00025076537655938325,
      "loss": 1.2563,
      "step": 38719
    },
    {
      "epoch": 3.3313258194958273,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002507419351106178,
      "loss": 1.7497,
      "step": 38720
    },
    {
      "epoch": 3.331411855803149,
      "grad_norm": 5.71875,
      "learning_rate": 0.0002507184943908837,
      "loss": 0.8549,
      "step": 38721
    },
    {
      "epoch": 3.3314978921104705,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00025069505440024944,
      "loss": 1.1777,
      "step": 38722
    },
    {
      "epoch": 3.331583928417792,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00025067161513878377,
      "loss": 0.6148,
      "step": 38723
    },
    {
      "epoch": 3.331669964725114,
      "grad_norm": 4.0625,
      "learning_rate": 0.000250648176606555,
      "loss": 0.9797,
      "step": 38724
    },
    {
      "epoch": 3.331756001032436,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00025062473880363206,
      "loss": 1.2163,
      "step": 38725
    },
    {
      "epoch": 3.3318420373397575,
      "grad_norm": 0.96875,
      "learning_rate": 0.000250601301730083,
      "loss": 1.3699,
      "step": 38726
    },
    {
      "epoch": 3.331928073647079,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00025057786538597673,
      "loss": 0.8585,
      "step": 38727
    },
    {
      "epoch": 3.3320141099544007,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0002505544297713818,
      "loss": 0.83,
      "step": 38728
    },
    {
      "epoch": 3.3321001462617224,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00025053099488636643,
      "loss": 0.6966,
      "step": 38729
    },
    {
      "epoch": 3.332186182569044,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00025050756073099947,
      "loss": 1.1532,
      "step": 38730
    },
    {
      "epoch": 3.3322722188763656,
      "grad_norm": 1.15625,
      "learning_rate": 0.00025048412730534946,
      "loss": 1.2909,
      "step": 38731
    },
    {
      "epoch": 3.3323582551836877,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002504606946094847,
      "loss": 2.3871,
      "step": 38732
    },
    {
      "epoch": 3.3324442914910093,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00025043726264347387,
      "loss": 1.7613,
      "step": 38733
    },
    {
      "epoch": 3.332530327798331,
      "grad_norm": 7.0625,
      "learning_rate": 0.0002504138314073856,
      "loss": 1.2568,
      "step": 38734
    },
    {
      "epoch": 3.3326163641056525,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002503904009012882,
      "loss": 1.1741,
      "step": 38735
    },
    {
      "epoch": 3.332702400412974,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00025036697112525066,
      "loss": 0.944,
      "step": 38736
    },
    {
      "epoch": 3.332788436720296,
      "grad_norm": 1.453125,
      "learning_rate": 0.00025034354207934087,
      "loss": 1.5436,
      "step": 38737
    },
    {
      "epoch": 3.3328744730276174,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002503201137636277,
      "loss": 1.2109,
      "step": 38738
    },
    {
      "epoch": 3.3329605093349395,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002502966861781798,
      "loss": 1.6379,
      "step": 38739
    },
    {
      "epoch": 3.333046545642261,
      "grad_norm": 4.09375,
      "learning_rate": 0.0002502732593230654,
      "loss": 1.34,
      "step": 38740
    },
    {
      "epoch": 3.3331325819495827,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002502498331983531,
      "loss": 0.9191,
      "step": 38741
    },
    {
      "epoch": 3.3332186182569044,
      "grad_norm": 1.203125,
      "learning_rate": 0.00025022640780411165,
      "loss": 1.3023,
      "step": 38742
    },
    {
      "epoch": 3.333304654564226,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00025020298314040924,
      "loss": 0.7526,
      "step": 38743
    },
    {
      "epoch": 3.3333906908715476,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002501795592073147,
      "loss": 0.735,
      "step": 38744
    },
    {
      "epoch": 3.3334767271788697,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0002501561360048961,
      "loss": 1.0383,
      "step": 38745
    },
    {
      "epoch": 3.3335627634861913,
      "grad_norm": 1.90625,
      "learning_rate": 0.0002501327135332223,
      "loss": 1.6359,
      "step": 38746
    },
    {
      "epoch": 3.333648799793513,
      "grad_norm": 1.0625,
      "learning_rate": 0.00025010929179236185,
      "loss": 0.8828,
      "step": 38747
    },
    {
      "epoch": 3.3337348361008345,
      "grad_norm": 0.875,
      "learning_rate": 0.00025008587078238293,
      "loss": 0.787,
      "step": 38748
    },
    {
      "epoch": 3.333820872408156,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00025006245050335424,
      "loss": 1.2785,
      "step": 38749
    },
    {
      "epoch": 3.333906908715478,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002500390309553445,
      "loss": 0.81,
      "step": 38750
    },
    {
      "epoch": 3.3339929450228,
      "grad_norm": 53.75,
      "learning_rate": 0.00025001561213842174,
      "loss": 1.1849,
      "step": 38751
    },
    {
      "epoch": 3.3340789813301215,
      "grad_norm": 1.015625,
      "learning_rate": 0.00024999219405265474,
      "loss": 1.0532,
      "step": 38752
    },
    {
      "epoch": 3.334165017637443,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00024996877669811207,
      "loss": 0.5608,
      "step": 38753
    },
    {
      "epoch": 3.3342510539447647,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00024994536007486197,
      "loss": 0.8906,
      "step": 38754
    },
    {
      "epoch": 3.3343370902520864,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00024992194418297324,
      "loss": 1.8509,
      "step": 38755
    },
    {
      "epoch": 3.334423126559408,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00024989852902251397,
      "loss": 0.7956,
      "step": 38756
    },
    {
      "epoch": 3.3345091628667296,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002498751145935529,
      "loss": 0.8873,
      "step": 38757
    },
    {
      "epoch": 3.3345951991740517,
      "grad_norm": 47.0,
      "learning_rate": 0.00024985170089615865,
      "loss": 1.4148,
      "step": 38758
    },
    {
      "epoch": 3.3346812354813733,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002498282879303993,
      "loss": 0.9505,
      "step": 38759
    },
    {
      "epoch": 3.334767271788695,
      "grad_norm": 14.9375,
      "learning_rate": 0.00024980487569634357,
      "loss": 1.1627,
      "step": 38760
    },
    {
      "epoch": 3.3348533080960165,
      "grad_norm": 1.859375,
      "learning_rate": 0.00024978146419406017,
      "loss": 1.1305,
      "step": 38761
    },
    {
      "epoch": 3.334939344403338,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002497580534236171,
      "loss": 0.9674,
      "step": 38762
    },
    {
      "epoch": 3.33502538071066,
      "grad_norm": 152.0,
      "learning_rate": 0.000249734643385083,
      "loss": 1.4439,
      "step": 38763
    },
    {
      "epoch": 3.3351114170179814,
      "grad_norm": 0.796875,
      "learning_rate": 0.00024971123407852664,
      "loss": 1.2356,
      "step": 38764
    },
    {
      "epoch": 3.3351974533253035,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000249687825504016,
      "loss": 0.5961,
      "step": 38765
    },
    {
      "epoch": 3.335283489632625,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00024966441766161983,
      "loss": 1.2407,
      "step": 38766
    },
    {
      "epoch": 3.3353695259399467,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002496410105514067,
      "loss": 0.9826,
      "step": 38767
    },
    {
      "epoch": 3.3354555622472684,
      "grad_norm": 31.5,
      "learning_rate": 0.0002496176041734448,
      "loss": 1.1713,
      "step": 38768
    },
    {
      "epoch": 3.33554159855459,
      "grad_norm": 2.9375,
      "learning_rate": 0.0002495941985278029,
      "loss": 0.7319,
      "step": 38769
    },
    {
      "epoch": 3.3356276348619116,
      "grad_norm": 1.25,
      "learning_rate": 0.000249570793614549,
      "loss": 1.4826,
      "step": 38770
    },
    {
      "epoch": 3.3357136711692332,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002495473894337518,
      "loss": 0.9082,
      "step": 38771
    },
    {
      "epoch": 3.3357997074765553,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00024952398598547987,
      "loss": 0.5095,
      "step": 38772
    },
    {
      "epoch": 3.335885743783877,
      "grad_norm": 12.4375,
      "learning_rate": 0.0002495005832698016,
      "loss": 1.1393,
      "step": 38773
    },
    {
      "epoch": 3.3359717800911985,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002494771812867855,
      "loss": 1.6397,
      "step": 38774
    },
    {
      "epoch": 3.33605781639852,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00024945378003649976,
      "loss": 1.2037,
      "step": 38775
    },
    {
      "epoch": 3.336143852705842,
      "grad_norm": 1.046875,
      "learning_rate": 0.000249430379519013,
      "loss": 0.7588,
      "step": 38776
    },
    {
      "epoch": 3.3362298890131634,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002494069797343938,
      "loss": 0.9234,
      "step": 38777
    },
    {
      "epoch": 3.336315925320485,
      "grad_norm": 6.21875,
      "learning_rate": 0.0002493835806827103,
      "loss": 0.9059,
      "step": 38778
    },
    {
      "epoch": 3.336401961627807,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000249360182364031,
      "loss": 1.2141,
      "step": 38779
    },
    {
      "epoch": 3.3364879979351287,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002493367847784247,
      "loss": 0.6703,
      "step": 38780
    },
    {
      "epoch": 3.3365740342424504,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002493133879259593,
      "loss": 0.7987,
      "step": 38781
    },
    {
      "epoch": 3.336660070549772,
      "grad_norm": 1.546875,
      "learning_rate": 0.00024928999180670353,
      "loss": 0.7484,
      "step": 38782
    },
    {
      "epoch": 3.3367461068570936,
      "grad_norm": 21.75,
      "learning_rate": 0.00024926659642072595,
      "loss": 1.604,
      "step": 38783
    },
    {
      "epoch": 3.3368321431644152,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00024924320176809465,
      "loss": 1.4041,
      "step": 38784
    },
    {
      "epoch": 3.336918179471737,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00024921980784887836,
      "loss": 1.5996,
      "step": 38785
    },
    {
      "epoch": 3.337004215779059,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002491964146631452,
      "loss": 0.6027,
      "step": 38786
    },
    {
      "epoch": 3.3370902520863805,
      "grad_norm": 86.5,
      "learning_rate": 0.00024917302221096375,
      "loss": 1.2335,
      "step": 38787
    },
    {
      "epoch": 3.337176288393702,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002491496304924026,
      "loss": 1.4504,
      "step": 38788
    },
    {
      "epoch": 3.337262324701024,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002491262395075298,
      "loss": 1.2585,
      "step": 38789
    },
    {
      "epoch": 3.3373483610083454,
      "grad_norm": 2624.0,
      "learning_rate": 0.00024910284925641396,
      "loss": 1.1803,
      "step": 38790
    },
    {
      "epoch": 3.337434397315667,
      "grad_norm": 13.4375,
      "learning_rate": 0.0002490794597391237,
      "loss": 0.824,
      "step": 38791
    },
    {
      "epoch": 3.3375204336229887,
      "grad_norm": 2.390625,
      "learning_rate": 0.000249056070955727,
      "loss": 0.6483,
      "step": 38792
    },
    {
      "epoch": 3.3376064699303107,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002490326829062925,
      "loss": 0.5883,
      "step": 38793
    },
    {
      "epoch": 3.3376925062376324,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002490092955908888,
      "loss": 1.7241,
      "step": 38794
    },
    {
      "epoch": 3.337778542544954,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002489859090095838,
      "loss": 1.1851,
      "step": 38795
    },
    {
      "epoch": 3.3378645788522756,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00024896252316244656,
      "loss": 0.6625,
      "step": 38796
    },
    {
      "epoch": 3.3379506151595972,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00024893913804954483,
      "loss": 1.4914,
      "step": 38797
    },
    {
      "epoch": 3.338036651466919,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00024891575367094726,
      "loss": 1.5329,
      "step": 38798
    },
    {
      "epoch": 3.3381226877742405,
      "grad_norm": 1.0625,
      "learning_rate": 0.00024889237002672255,
      "loss": 0.8595,
      "step": 38799
    },
    {
      "epoch": 3.3382087240815626,
      "grad_norm": 1.0625,
      "learning_rate": 0.00024886898711693855,
      "loss": 0.7834,
      "step": 38800
    },
    {
      "epoch": 3.338294760388884,
      "grad_norm": 0.96875,
      "learning_rate": 0.00024884560494166396,
      "loss": 0.6959,
      "step": 38801
    },
    {
      "epoch": 3.338380796696206,
      "grad_norm": 2.109375,
      "learning_rate": 0.00024882222350096735,
      "loss": 1.3661,
      "step": 38802
    },
    {
      "epoch": 3.3384668330035274,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00024879884279491665,
      "loss": 1.075,
      "step": 38803
    },
    {
      "epoch": 3.338552869310849,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002487754628235805,
      "loss": 1.4689,
      "step": 38804
    },
    {
      "epoch": 3.338638905618171,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00024875208358702747,
      "loss": 0.8218,
      "step": 38805
    },
    {
      "epoch": 3.3387249419254927,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00024872870508532545,
      "loss": 0.8151,
      "step": 38806
    },
    {
      "epoch": 3.3388109782328144,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002487053273185432,
      "loss": 1.0212,
      "step": 38807
    },
    {
      "epoch": 3.338897014540136,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002486819502867489,
      "loss": 0.9957,
      "step": 38808
    },
    {
      "epoch": 3.3389830508474576,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002486585739900111,
      "loss": 1.0391,
      "step": 38809
    },
    {
      "epoch": 3.3390690871547792,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00024863519842839826,
      "loss": 0.9307,
      "step": 38810
    },
    {
      "epoch": 3.339155123462101,
      "grad_norm": 52.5,
      "learning_rate": 0.0002486118236019784,
      "loss": 1.4066,
      "step": 38811
    },
    {
      "epoch": 3.339241159769423,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002485884495108202,
      "loss": 0.9793,
      "step": 38812
    },
    {
      "epoch": 3.3393271960767446,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002485650761549917,
      "loss": 1.4471,
      "step": 38813
    },
    {
      "epoch": 3.339413232384066,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002485417035345615,
      "loss": 1.8701,
      "step": 38814
    },
    {
      "epoch": 3.339499268691388,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00024851833164959806,
      "loss": 0.6888,
      "step": 38815
    },
    {
      "epoch": 3.3395853049987094,
      "grad_norm": 1.0625,
      "learning_rate": 0.00024849496050016945,
      "loss": 0.6996,
      "step": 38816
    },
    {
      "epoch": 3.339671341306031,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002484715900863441,
      "loss": 0.7686,
      "step": 38817
    },
    {
      "epoch": 3.3397573776133527,
      "grad_norm": 24.375,
      "learning_rate": 0.00024844822040819063,
      "loss": 0.9944,
      "step": 38818
    },
    {
      "epoch": 3.3398434139206747,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00024842485146577703,
      "loss": 1.5906,
      "step": 38819
    },
    {
      "epoch": 3.3399294502279964,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002484014832591718,
      "loss": 0.6881,
      "step": 38820
    },
    {
      "epoch": 3.340015486535318,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00024837811578844356,
      "loss": 1.0408,
      "step": 38821
    },
    {
      "epoch": 3.3401015228426396,
      "grad_norm": 0.78125,
      "learning_rate": 0.00024835474905366014,
      "loss": 0.6215,
      "step": 38822
    },
    {
      "epoch": 3.3401875591499612,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0002483313830548901,
      "loss": 0.7719,
      "step": 38823
    },
    {
      "epoch": 3.340273595457283,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00024830801779220205,
      "loss": 0.7391,
      "step": 38824
    },
    {
      "epoch": 3.3403596317646045,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00024828465326566393,
      "loss": 1.4994,
      "step": 38825
    },
    {
      "epoch": 3.3404456680719266,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00024826128947534436,
      "loss": 0.8157,
      "step": 38826
    },
    {
      "epoch": 3.340531704379248,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002482379264213114,
      "loss": 0.8458,
      "step": 38827
    },
    {
      "epoch": 3.34061774068657,
      "grad_norm": 1.109375,
      "learning_rate": 0.00024821456410363356,
      "loss": 1.4645,
      "step": 38828
    },
    {
      "epoch": 3.3407037769938914,
      "grad_norm": 1.4375,
      "learning_rate": 0.00024819120252237933,
      "loss": 1.609,
      "step": 38829
    },
    {
      "epoch": 3.340789813301213,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002481678416776167,
      "loss": 1.151,
      "step": 38830
    },
    {
      "epoch": 3.3408758496085347,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002481444815694141,
      "loss": 1.4648,
      "step": 38831
    },
    {
      "epoch": 3.3409618859158563,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00024812112219784014,
      "loss": 0.8244,
      "step": 38832
    },
    {
      "epoch": 3.3410479222231784,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00024809776356296266,
      "loss": 1.075,
      "step": 38833
    },
    {
      "epoch": 3.3411339585305,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002480744056648503,
      "loss": 0.6854,
      "step": 38834
    },
    {
      "epoch": 3.3412199948378216,
      "grad_norm": 10.5625,
      "learning_rate": 0.00024805104850357144,
      "loss": 0.8482,
      "step": 38835
    },
    {
      "epoch": 3.3413060311451432,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00024802769207919414,
      "loss": 1.0159,
      "step": 38836
    },
    {
      "epoch": 3.341392067452465,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00024800433639178695,
      "loss": 1.2182,
      "step": 38837
    },
    {
      "epoch": 3.3414781037597865,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002479809814414179,
      "loss": 1.156,
      "step": 38838
    },
    {
      "epoch": 3.341564140067108,
      "grad_norm": 1.609375,
      "learning_rate": 0.00024795762722815547,
      "loss": 0.9616,
      "step": 38839
    },
    {
      "epoch": 3.34165017637443,
      "grad_norm": 378.0,
      "learning_rate": 0.00024793427375206816,
      "loss": 1.0157,
      "step": 38840
    },
    {
      "epoch": 3.341736212681752,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00024791092101322386,
      "loss": 1.4751,
      "step": 38841
    },
    {
      "epoch": 3.3418222489890734,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00024788756901169114,
      "loss": 1.0355,
      "step": 38842
    },
    {
      "epoch": 3.341908285296395,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0002478642177475384,
      "loss": 1.814,
      "step": 38843
    },
    {
      "epoch": 3.3419943216037167,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002478408672208336,
      "loss": 1.0223,
      "step": 38844
    },
    {
      "epoch": 3.3420803579110383,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002478175174316452,
      "loss": 1.2656,
      "step": 38845
    },
    {
      "epoch": 3.34216639421836,
      "grad_norm": 1.046875,
      "learning_rate": 0.00024779416838004154,
      "loss": 1.3367,
      "step": 38846
    },
    {
      "epoch": 3.342252430525682,
      "grad_norm": 1.03125,
      "learning_rate": 0.00024777082006609085,
      "loss": 1.3862,
      "step": 38847
    },
    {
      "epoch": 3.3423384668330036,
      "grad_norm": 4.28125,
      "learning_rate": 0.00024774747248986173,
      "loss": 1.876,
      "step": 38848
    },
    {
      "epoch": 3.3424245031403252,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00024772412565142194,
      "loss": 1.1433,
      "step": 38849
    },
    {
      "epoch": 3.342510539447647,
      "grad_norm": 1.203125,
      "learning_rate": 0.00024770077955084,
      "loss": 1.292,
      "step": 38850
    },
    {
      "epoch": 3.3425965757549685,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00024767743418818445,
      "loss": 0.6831,
      "step": 38851
    },
    {
      "epoch": 3.34268261206229,
      "grad_norm": 1.40625,
      "learning_rate": 0.00024765408956352314,
      "loss": 1.4,
      "step": 38852
    },
    {
      "epoch": 3.3427686483696117,
      "grad_norm": 83.5,
      "learning_rate": 0.0002476307456769245,
      "loss": 0.9252,
      "step": 38853
    },
    {
      "epoch": 3.342854684676934,
      "grad_norm": 1.421875,
      "learning_rate": 0.000247607402528457,
      "loss": 1.2518,
      "step": 38854
    },
    {
      "epoch": 3.3429407209842554,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00024758406011818866,
      "loss": 0.605,
      "step": 38855
    },
    {
      "epoch": 3.343026757291577,
      "grad_norm": 3.671875,
      "learning_rate": 0.000247560718446188,
      "loss": 1.2092,
      "step": 38856
    },
    {
      "epoch": 3.3431127935988987,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00024753737751252296,
      "loss": 0.7476,
      "step": 38857
    },
    {
      "epoch": 3.3431988299062203,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000247514037317262,
      "loss": 1.3759,
      "step": 38858
    },
    {
      "epoch": 3.343284866213542,
      "grad_norm": 3.796875,
      "learning_rate": 0.00024749069786047357,
      "loss": 1.0485,
      "step": 38859
    },
    {
      "epoch": 3.343370902520864,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00024746735914222553,
      "loss": 1.4662,
      "step": 38860
    },
    {
      "epoch": 3.3434569388281856,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00024744402116258637,
      "loss": 0.8907,
      "step": 38861
    },
    {
      "epoch": 3.3435429751355072,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00024742068392162445,
      "loss": 0.5305,
      "step": 38862
    },
    {
      "epoch": 3.343629011442829,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002473973474194078,
      "loss": 1.1781,
      "step": 38863
    },
    {
      "epoch": 3.3437150477501505,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002473740116560047,
      "loss": 0.6282,
      "step": 38864
    },
    {
      "epoch": 3.343801084057472,
      "grad_norm": 1.5625,
      "learning_rate": 0.0002473506766314838,
      "loss": 1.2972,
      "step": 38865
    },
    {
      "epoch": 3.343887120364794,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00024732734234591267,
      "loss": 0.6241,
      "step": 38866
    },
    {
      "epoch": 3.343973156672116,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002473040087993602,
      "loss": 0.6958,
      "step": 38867
    },
    {
      "epoch": 3.3440591929794374,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002472806759918942,
      "loss": 1.0421,
      "step": 38868
    },
    {
      "epoch": 3.344145229286759,
      "grad_norm": 4.0,
      "learning_rate": 0.00024725734392358305,
      "loss": 0.5935,
      "step": 38869
    },
    {
      "epoch": 3.3442312655940807,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002472340125944952,
      "loss": 1.1569,
      "step": 38870
    },
    {
      "epoch": 3.3443173019014023,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00024721068200469847,
      "loss": 1.3433,
      "step": 38871
    },
    {
      "epoch": 3.344403338208724,
      "grad_norm": 1.71875,
      "learning_rate": 0.00024718735215426143,
      "loss": 1.7338,
      "step": 38872
    },
    {
      "epoch": 3.344489374516046,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00024716402304325233,
      "loss": 1.0156,
      "step": 38873
    },
    {
      "epoch": 3.3445754108233676,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002471406946717391,
      "loss": 1.344,
      "step": 38874
    },
    {
      "epoch": 3.3446614471306892,
      "grad_norm": 0.953125,
      "learning_rate": 0.00024711736703979015,
      "loss": 1.5805,
      "step": 38875
    },
    {
      "epoch": 3.344747483438011,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00024709404014747394,
      "loss": 0.9675,
      "step": 38876
    },
    {
      "epoch": 3.3448335197453325,
      "grad_norm": 1.0625,
      "learning_rate": 0.00024707071399485825,
      "loss": 0.6413,
      "step": 38877
    },
    {
      "epoch": 3.344919556052654,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00024704738858201175,
      "loss": 1.0619,
      "step": 38878
    },
    {
      "epoch": 3.3450055923599757,
      "grad_norm": 1.1875,
      "learning_rate": 0.00024702406390900226,
      "loss": 2.0113,
      "step": 38879
    },
    {
      "epoch": 3.345091628667298,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002470007399758982,
      "loss": 1.3764,
      "step": 38880
    },
    {
      "epoch": 3.3451776649746194,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002469774167827679,
      "loss": 0.587,
      "step": 38881
    },
    {
      "epoch": 3.345263701281941,
      "grad_norm": 1.109375,
      "learning_rate": 0.00024695409432967907,
      "loss": 1.1483,
      "step": 38882
    },
    {
      "epoch": 3.3453497375892627,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00024693077261670073,
      "loss": 1.9772,
      "step": 38883
    },
    {
      "epoch": 3.3454357738965843,
      "grad_norm": 1.359375,
      "learning_rate": 0.00024690745164390043,
      "loss": 1.2461,
      "step": 38884
    },
    {
      "epoch": 3.345521810203906,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00024688413141134665,
      "loss": 1.7183,
      "step": 38885
    },
    {
      "epoch": 3.3456078465112276,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002468608119191077,
      "loss": 0.9631,
      "step": 38886
    },
    {
      "epoch": 3.3456938828185496,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002468374931672515,
      "loss": 1.578,
      "step": 38887
    },
    {
      "epoch": 3.3457799191258712,
      "grad_norm": 6.53125,
      "learning_rate": 0.00024681417515584636,
      "loss": 0.6678,
      "step": 38888
    },
    {
      "epoch": 3.345865955433193,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002467908578849607,
      "loss": 1.0575,
      "step": 38889
    },
    {
      "epoch": 3.3459519917405145,
      "grad_norm": 117.0,
      "learning_rate": 0.0002467675413546624,
      "loss": 1.5121,
      "step": 38890
    },
    {
      "epoch": 3.346038028047836,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00024674422556501975,
      "loss": 0.6977,
      "step": 38891
    },
    {
      "epoch": 3.3461240643551577,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002467209105161011,
      "loss": 1.4486,
      "step": 38892
    },
    {
      "epoch": 3.3462101006624794,
      "grad_norm": 1.328125,
      "learning_rate": 0.0002466975962079744,
      "loss": 1.6226,
      "step": 38893
    },
    {
      "epoch": 3.3462961369698014,
      "grad_norm": 1.359375,
      "learning_rate": 0.000246674282640708,
      "loss": 0.9895,
      "step": 38894
    },
    {
      "epoch": 3.346382173277123,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00024665096981437025,
      "loss": 1.2861,
      "step": 38895
    },
    {
      "epoch": 3.3464682095844447,
      "grad_norm": 1.203125,
      "learning_rate": 0.000246627657729029,
      "loss": 1.0488,
      "step": 38896
    },
    {
      "epoch": 3.3465542458917663,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00024660434638475257,
      "loss": 0.6475,
      "step": 38897
    },
    {
      "epoch": 3.346640282199088,
      "grad_norm": 1.0,
      "learning_rate": 0.0002465810357816091,
      "loss": 1.4066,
      "step": 38898
    },
    {
      "epoch": 3.3467263185064096,
      "grad_norm": 1.125,
      "learning_rate": 0.0002465577259196668,
      "loss": 0.8552,
      "step": 38899
    },
    {
      "epoch": 3.346812354813731,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000246534416798994,
      "loss": 0.763,
      "step": 38900
    },
    {
      "epoch": 3.3468983911210533,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002465111084196586,
      "loss": 1.0335,
      "step": 38901
    },
    {
      "epoch": 3.346984427428375,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002464878007817289,
      "loss": 1.7262,
      "step": 38902
    },
    {
      "epoch": 3.3470704637356965,
      "grad_norm": 83.0,
      "learning_rate": 0.00024646449388527323,
      "loss": 1.8436,
      "step": 38903
    },
    {
      "epoch": 3.347156500043018,
      "grad_norm": 1.109375,
      "learning_rate": 0.00024644118773035944,
      "loss": 0.9511,
      "step": 38904
    },
    {
      "epoch": 3.3472425363503397,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00024641788231705587,
      "loss": 1.0712,
      "step": 38905
    },
    {
      "epoch": 3.3473285726576614,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002463945776454308,
      "loss": 1.1523,
      "step": 38906
    },
    {
      "epoch": 3.347414608964983,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002463712737155522,
      "loss": 1.2934,
      "step": 38907
    },
    {
      "epoch": 3.347500645272305,
      "grad_norm": 146.0,
      "learning_rate": 0.0002463479705274884,
      "loss": 1.7738,
      "step": 38908
    },
    {
      "epoch": 3.3475866815796267,
      "grad_norm": 2.46875,
      "learning_rate": 0.00024632466808130726,
      "loss": 1.1375,
      "step": 38909
    },
    {
      "epoch": 3.3476727178869483,
      "grad_norm": 1.125,
      "learning_rate": 0.00024630136637707715,
      "loss": 0.7373,
      "step": 38910
    },
    {
      "epoch": 3.34775875419427,
      "grad_norm": 14.9375,
      "learning_rate": 0.0002462780654148664,
      "loss": 1.0838,
      "step": 38911
    },
    {
      "epoch": 3.3478447905015916,
      "grad_norm": 1.171875,
      "learning_rate": 0.00024625476519474275,
      "loss": 0.8102,
      "step": 38912
    },
    {
      "epoch": 3.347930826808913,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002462314657167746,
      "loss": 1.7649,
      "step": 38913
    },
    {
      "epoch": 3.348016863116235,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002462081669810302,
      "loss": 1.5448,
      "step": 38914
    },
    {
      "epoch": 3.348102899423557,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002461848689875774,
      "loss": 1.3704,
      "step": 38915
    },
    {
      "epoch": 3.3481889357308785,
      "grad_norm": 436.0,
      "learning_rate": 0.00024616157173648444,
      "loss": 1.1476,
      "step": 38916
    },
    {
      "epoch": 3.3482749720382,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002461382752278198,
      "loss": 1.2286,
      "step": 38917
    },
    {
      "epoch": 3.3483610083455217,
      "grad_norm": 1.375,
      "learning_rate": 0.00024611497946165086,
      "loss": 1.972,
      "step": 38918
    },
    {
      "epoch": 3.3484470446528434,
      "grad_norm": 1.0625,
      "learning_rate": 0.00024609168443804664,
      "loss": 1.5708,
      "step": 38919
    },
    {
      "epoch": 3.348533080960165,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002460683901570747,
      "loss": 0.9162,
      "step": 38920
    },
    {
      "epoch": 3.348619117267487,
      "grad_norm": 6.0,
      "learning_rate": 0.00024604509661880326,
      "loss": 1.0054,
      "step": 38921
    },
    {
      "epoch": 3.3487051535748087,
      "grad_norm": 1.0,
      "learning_rate": 0.00024602180382330075,
      "loss": 1.414,
      "step": 38922
    },
    {
      "epoch": 3.3487911898821303,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00024599851177063486,
      "loss": 0.4454,
      "step": 38923
    },
    {
      "epoch": 3.348877226189452,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00024597522046087417,
      "loss": 1.4959,
      "step": 38924
    },
    {
      "epoch": 3.3489632624967736,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002459519298940863,
      "loss": 1.1264,
      "step": 38925
    },
    {
      "epoch": 3.349049298804095,
      "grad_norm": 1.03125,
      "learning_rate": 0.00024592864007033963,
      "loss": 0.659,
      "step": 38926
    },
    {
      "epoch": 3.3491353351114173,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002459053509897025,
      "loss": 1.21,
      "step": 38927
    },
    {
      "epoch": 3.349221371418739,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002458820626522426,
      "loss": 1.2489,
      "step": 38928
    },
    {
      "epoch": 3.3493074077260605,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002458587750580283,
      "loss": 1.3115,
      "step": 38929
    },
    {
      "epoch": 3.349393444033382,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00024583548820712774,
      "loss": 0.7302,
      "step": 38930
    },
    {
      "epoch": 3.3494794803407038,
      "grad_norm": 44.0,
      "learning_rate": 0.0002458122020996088,
      "loss": 0.7304,
      "step": 38931
    },
    {
      "epoch": 3.3495655166480254,
      "grad_norm": 2.03125,
      "learning_rate": 0.0002457889167355398,
      "loss": 0.9325,
      "step": 38932
    },
    {
      "epoch": 3.349651552955347,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002457656321149889,
      "loss": 1.1781,
      "step": 38933
    },
    {
      "epoch": 3.349737589262669,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00024574234823802383,
      "loss": 0.9695,
      "step": 38934
    },
    {
      "epoch": 3.3498236255699907,
      "grad_norm": 1.15625,
      "learning_rate": 0.00024571906510471305,
      "loss": 0.8436,
      "step": 38935
    },
    {
      "epoch": 3.3499096618773123,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002456957827151247,
      "loss": 0.9305,
      "step": 38936
    },
    {
      "epoch": 3.349995698184634,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002456725010693265,
      "loss": 0.9637,
      "step": 38937
    },
    {
      "epoch": 3.3500817344919556,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00024564922016738695,
      "loss": 1.318,
      "step": 38938
    },
    {
      "epoch": 3.350167770799277,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002456259400093738,
      "loss": 1.3871,
      "step": 38939
    },
    {
      "epoch": 3.350253807106599,
      "grad_norm": 64.0,
      "learning_rate": 0.00024560266059535537,
      "loss": 0.9338,
      "step": 38940
    },
    {
      "epoch": 3.350339843413921,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002455793819253998,
      "loss": 1.0417,
      "step": 38941
    },
    {
      "epoch": 3.3504258797212425,
      "grad_norm": 1.125,
      "learning_rate": 0.00024555610399957487,
      "loss": 1.3824,
      "step": 38942
    },
    {
      "epoch": 3.350511916028564,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00024553282681794887,
      "loss": 0.4911,
      "step": 38943
    },
    {
      "epoch": 3.3505979523358858,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00024550955038059004,
      "loss": 1.8384,
      "step": 38944
    },
    {
      "epoch": 3.3506839886432074,
      "grad_norm": 15.375,
      "learning_rate": 0.00024548627468756604,
      "loss": 1.2546,
      "step": 38945
    },
    {
      "epoch": 3.350770024950529,
      "grad_norm": 1.65625,
      "learning_rate": 0.0002454629997389453,
      "loss": 1.0876,
      "step": 38946
    },
    {
      "epoch": 3.3508560612578506,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002454397255347959,
      "loss": 1.2703,
      "step": 38947
    },
    {
      "epoch": 3.3509420975651727,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002454164520751856,
      "loss": 1.5752,
      "step": 38948
    },
    {
      "epoch": 3.3510281338724943,
      "grad_norm": 1.328125,
      "learning_rate": 0.00024539317936018285,
      "loss": 1.5939,
      "step": 38949
    },
    {
      "epoch": 3.351114170179816,
      "grad_norm": 21.5,
      "learning_rate": 0.0002453699073898553,
      "loss": 0.7389,
      "step": 38950
    },
    {
      "epoch": 3.3512002064871376,
      "grad_norm": 1.125,
      "learning_rate": 0.00024534663616427135,
      "loss": 0.9476,
      "step": 38951
    },
    {
      "epoch": 3.351286242794459,
      "grad_norm": 2.84375,
      "learning_rate": 0.0002453233656834991,
      "loss": 0.8671,
      "step": 38952
    },
    {
      "epoch": 3.351372279101781,
      "grad_norm": 123.0,
      "learning_rate": 0.00024530009594760625,
      "loss": 1.4412,
      "step": 38953
    },
    {
      "epoch": 3.3514583154091024,
      "grad_norm": 5.0,
      "learning_rate": 0.00024527682695666114,
      "loss": 0.8653,
      "step": 38954
    },
    {
      "epoch": 3.3515443517164245,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002452535587107317,
      "loss": 0.654,
      "step": 38955
    },
    {
      "epoch": 3.351630388023746,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00024523029120988616,
      "loss": 1.1779,
      "step": 38956
    },
    {
      "epoch": 3.3517164243310678,
      "grad_norm": 2.109375,
      "learning_rate": 0.00024520702445419263,
      "loss": 1.5919,
      "step": 38957
    },
    {
      "epoch": 3.3518024606383894,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00024518375844371875,
      "loss": 1.2479,
      "step": 38958
    },
    {
      "epoch": 3.351888496945711,
      "grad_norm": 1.515625,
      "learning_rate": 0.00024516049317853286,
      "loss": 1.7548,
      "step": 38959
    },
    {
      "epoch": 3.3519745332530326,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002451372286587031,
      "loss": 1.3739,
      "step": 38960
    },
    {
      "epoch": 3.3520605695603543,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002451139648842973,
      "loss": 0.6395,
      "step": 38961
    },
    {
      "epoch": 3.3521466058676763,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002450907018553835,
      "loss": 0.9599,
      "step": 38962
    },
    {
      "epoch": 3.352232642174998,
      "grad_norm": 3.125,
      "learning_rate": 0.00024506743957203005,
      "loss": 1.0055,
      "step": 38963
    },
    {
      "epoch": 3.3523186784823196,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002450441780343045,
      "loss": 0.9256,
      "step": 38964
    },
    {
      "epoch": 3.352404714789641,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002450209172422754,
      "loss": 0.6038,
      "step": 38965
    },
    {
      "epoch": 3.352490751096963,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002449976571960103,
      "loss": 1.2205,
      "step": 38966
    },
    {
      "epoch": 3.3525767874042844,
      "grad_norm": 17920.0,
      "learning_rate": 0.00024497439789557745,
      "loss": 1.1556,
      "step": 38967
    },
    {
      "epoch": 3.352662823711606,
      "grad_norm": 1.15625,
      "learning_rate": 0.00024495113934104514,
      "loss": 1.2034,
      "step": 38968
    },
    {
      "epoch": 3.352748860018928,
      "grad_norm": 1.609375,
      "learning_rate": 0.00024492788153248084,
      "loss": 0.8403,
      "step": 38969
    },
    {
      "epoch": 3.3528348963262498,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00024490462446995296,
      "loss": 1.5203,
      "step": 38970
    },
    {
      "epoch": 3.3529209326335714,
      "grad_norm": 5.8125,
      "learning_rate": 0.00024488136815352965,
      "loss": 1.5696,
      "step": 38971
    },
    {
      "epoch": 3.353006968940893,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0002448581125832785,
      "loss": 1.351,
      "step": 38972
    },
    {
      "epoch": 3.3530930052482146,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00024483485775926773,
      "loss": 1.4357,
      "step": 38973
    },
    {
      "epoch": 3.3531790415555363,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00024481160368156556,
      "loss": 1.5755,
      "step": 38974
    },
    {
      "epoch": 3.353265077862858,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00024478835035023967,
      "loss": 1.1212,
      "step": 38975
    },
    {
      "epoch": 3.35335111417018,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002447650977653582,
      "loss": 1.0187,
      "step": 38976
    },
    {
      "epoch": 3.3534371504775016,
      "grad_norm": 5.0,
      "learning_rate": 0.00024474184592698935,
      "loss": 0.967,
      "step": 38977
    },
    {
      "epoch": 3.353523186784823,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00024471859483520073,
      "loss": 0.9517,
      "step": 38978
    },
    {
      "epoch": 3.353609223092145,
      "grad_norm": 1.078125,
      "learning_rate": 0.00024469534449006084,
      "loss": 1.3529,
      "step": 38979
    },
    {
      "epoch": 3.3536952593994664,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002446720948916372,
      "loss": 0.9431,
      "step": 38980
    },
    {
      "epoch": 3.3537812957067885,
      "grad_norm": 0.796875,
      "learning_rate": 0.000244648846039998,
      "loss": 1.0923,
      "step": 38981
    },
    {
      "epoch": 3.35386733201411,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002446255979352115,
      "loss": 1.245,
      "step": 38982
    },
    {
      "epoch": 3.3539533683214318,
      "grad_norm": 1.03125,
      "learning_rate": 0.00024460235057734527,
      "loss": 1.4768,
      "step": 38983
    },
    {
      "epoch": 3.3540394046287534,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002445791039664674,
      "loss": 0.7646,
      "step": 38984
    },
    {
      "epoch": 3.354125440936075,
      "grad_norm": 0.984375,
      "learning_rate": 0.00024455585810264626,
      "loss": 0.8867,
      "step": 38985
    },
    {
      "epoch": 3.3542114772433966,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002445326129859493,
      "loss": 0.9594,
      "step": 38986
    },
    {
      "epoch": 3.3542975135507183,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002445093686164448,
      "loss": 0.8303,
      "step": 38987
    },
    {
      "epoch": 3.3543835498580403,
      "grad_norm": 4.28125,
      "learning_rate": 0.00024448612499420084,
      "loss": 0.6943,
      "step": 38988
    },
    {
      "epoch": 3.354469586165362,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002444628821192852,
      "loss": 1.4178,
      "step": 38989
    },
    {
      "epoch": 3.3545556224726836,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00024443963999176607,
      "loss": 0.6991,
      "step": 38990
    },
    {
      "epoch": 3.354641658780005,
      "grad_norm": 1.296875,
      "learning_rate": 0.00024441639861171103,
      "loss": 1.4255,
      "step": 38991
    },
    {
      "epoch": 3.354727695087327,
      "grad_norm": 1.125,
      "learning_rate": 0.00024439315797918814,
      "loss": 1.462,
      "step": 38992
    },
    {
      "epoch": 3.3548137313946484,
      "grad_norm": 1.21875,
      "learning_rate": 0.000244369918094266,
      "loss": 2.0287,
      "step": 38993
    },
    {
      "epoch": 3.35489976770197,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002443466789570119,
      "loss": 0.7679,
      "step": 38994
    },
    {
      "epoch": 3.354985804009292,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00024432344056749423,
      "loss": 0.8569,
      "step": 38995
    },
    {
      "epoch": 3.3550718403166138,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0002443002029257806,
      "loss": 1.2468,
      "step": 38996
    },
    {
      "epoch": 3.3551578766239354,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002442769660319391,
      "loss": 0.5869,
      "step": 38997
    },
    {
      "epoch": 3.355243912931257,
      "grad_norm": 1.875,
      "learning_rate": 0.000244253729886038,
      "loss": 1.5587,
      "step": 38998
    },
    {
      "epoch": 3.3553299492385786,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002442304944881448,
      "loss": 1.4986,
      "step": 38999
    },
    {
      "epoch": 3.3554159855459003,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002442072598383276,
      "loss": 0.7612,
      "step": 39000
    },
    {
      "epoch": 3.355502021853222,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002441840259366546,
      "loss": 1.6413,
      "step": 39001
    },
    {
      "epoch": 3.355588058160544,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002441607927831934,
      "loss": 1.601,
      "step": 39002
    },
    {
      "epoch": 3.3556740944678656,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00024413756037801215,
      "loss": 0.9226,
      "step": 39003
    },
    {
      "epoch": 3.355760130775187,
      "grad_norm": 1.1875,
      "learning_rate": 0.000244114328721179,
      "loss": 0.9414,
      "step": 39004
    },
    {
      "epoch": 3.355846167082509,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002440910978127614,
      "loss": 0.7597,
      "step": 39005
    },
    {
      "epoch": 3.3559322033898304,
      "grad_norm": 1.609375,
      "learning_rate": 0.00024406786765282762,
      "loss": 1.0369,
      "step": 39006
    },
    {
      "epoch": 3.356018239697152,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00024404463824144573,
      "loss": 1.0537,
      "step": 39007
    },
    {
      "epoch": 3.3561042760044737,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002440214095786833,
      "loss": 0.8819,
      "step": 39008
    },
    {
      "epoch": 3.3561903123117958,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00024399818166460864,
      "loss": 0.6527,
      "step": 39009
    },
    {
      "epoch": 3.3562763486191174,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00024397495449928936,
      "loss": 0.7637,
      "step": 39010
    },
    {
      "epoch": 3.356362384926439,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00024395172808279347,
      "loss": 0.7359,
      "step": 39011
    },
    {
      "epoch": 3.3564484212337606,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002439285024151892,
      "loss": 1.4207,
      "step": 39012
    },
    {
      "epoch": 3.3565344575410823,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00024390527749654413,
      "loss": 2.0735,
      "step": 39013
    },
    {
      "epoch": 3.356620493848404,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002438820533269262,
      "loss": 1.2954,
      "step": 39014
    },
    {
      "epoch": 3.3567065301557255,
      "grad_norm": 1.234375,
      "learning_rate": 0.00024385882990640373,
      "loss": 1.7535,
      "step": 39015
    },
    {
      "epoch": 3.3567925664630476,
      "grad_norm": 1.25,
      "learning_rate": 0.00024383560723504417,
      "loss": 0.9756,
      "step": 39016
    },
    {
      "epoch": 3.356878602770369,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002438123853129156,
      "loss": 1.1536,
      "step": 39017
    },
    {
      "epoch": 3.356964639077691,
      "grad_norm": 1.328125,
      "learning_rate": 0.00024378916414008616,
      "loss": 1.7009,
      "step": 39018
    },
    {
      "epoch": 3.3570506753850125,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002437659437166234,
      "loss": 0.7632,
      "step": 39019
    },
    {
      "epoch": 3.357136711692334,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00024374272404259568,
      "loss": 1.0954,
      "step": 39020
    },
    {
      "epoch": 3.3572227479996557,
      "grad_norm": 1.671875,
      "learning_rate": 0.0002437195051180704,
      "loss": 1.5684,
      "step": 39021
    },
    {
      "epoch": 3.3573087843069773,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00024369628694311573,
      "loss": 0.8868,
      "step": 39022
    },
    {
      "epoch": 3.3573948206142994,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00024367306951779977,
      "loss": 1.0393,
      "step": 39023
    },
    {
      "epoch": 3.357480856921621,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00024364985284219004,
      "loss": 0.5517,
      "step": 39024
    },
    {
      "epoch": 3.3575668932289426,
      "grad_norm": 32.25,
      "learning_rate": 0.0002436266369163546,
      "loss": 0.9151,
      "step": 39025
    },
    {
      "epoch": 3.3576529295362643,
      "grad_norm": 1.015625,
      "learning_rate": 0.00024360342174036166,
      "loss": 1.1883,
      "step": 39026
    },
    {
      "epoch": 3.357738965843586,
      "grad_norm": 0.84375,
      "learning_rate": 0.00024358020731427853,
      "loss": 0.515,
      "step": 39027
    },
    {
      "epoch": 3.3578250021509075,
      "grad_norm": 1.328125,
      "learning_rate": 0.00024355699363817353,
      "loss": 1.3293,
      "step": 39028
    },
    {
      "epoch": 3.357911038458229,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002435337807121144,
      "loss": 0.9102,
      "step": 39029
    },
    {
      "epoch": 3.357997074765551,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002435105685361691,
      "loss": 1.4909,
      "step": 39030
    },
    {
      "epoch": 3.358083111072873,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002434873571104057,
      "loss": 0.4339,
      "step": 39031
    },
    {
      "epoch": 3.3581691473801945,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00024346414643489168,
      "loss": 0.6528,
      "step": 39032
    },
    {
      "epoch": 3.358255183687516,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002434409365096951,
      "loss": 1.1908,
      "step": 39033
    },
    {
      "epoch": 3.3583412199948377,
      "grad_norm": 1.09375,
      "learning_rate": 0.00024341772733488417,
      "loss": 1.1121,
      "step": 39034
    },
    {
      "epoch": 3.3584272563021593,
      "grad_norm": 1.34375,
      "learning_rate": 0.00024339451891052622,
      "loss": 1.2583,
      "step": 39035
    },
    {
      "epoch": 3.3585132926094814,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0002433713112366896,
      "loss": 0.9828,
      "step": 39036
    },
    {
      "epoch": 3.358599328916803,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00024334810431344178,
      "loss": 1.1448,
      "step": 39037
    },
    {
      "epoch": 3.3586853652241246,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002433248981408509,
      "loss": 1.5834,
      "step": 39038
    },
    {
      "epoch": 3.3587714015314463,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002433016927189849,
      "loss": 0.49,
      "step": 39039
    },
    {
      "epoch": 3.358857437838768,
      "grad_norm": 2.46875,
      "learning_rate": 0.00024327848804791137,
      "loss": 0.6054,
      "step": 39040
    },
    {
      "epoch": 3.3589434741460895,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002432552841276983,
      "loss": 0.9564,
      "step": 39041
    },
    {
      "epoch": 3.3590295104534116,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00024323208095841382,
      "loss": 0.8228,
      "step": 39042
    },
    {
      "epoch": 3.359115546760733,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00024320887854012535,
      "loss": 1.1187,
      "step": 39043
    },
    {
      "epoch": 3.359201583068055,
      "grad_norm": 1.109375,
      "learning_rate": 0.00024318567687290095,
      "loss": 0.7528,
      "step": 39044
    },
    {
      "epoch": 3.3592876193753765,
      "grad_norm": 2.453125,
      "learning_rate": 0.00024316247595680874,
      "loss": 1.3697,
      "step": 39045
    },
    {
      "epoch": 3.359373655682698,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00024313927579191607,
      "loss": 1.495,
      "step": 39046
    },
    {
      "epoch": 3.3594596919900197,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00024311607637829115,
      "loss": 0.9767,
      "step": 39047
    },
    {
      "epoch": 3.3595457282973413,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00024309287771600186,
      "loss": 0.9069,
      "step": 39048
    },
    {
      "epoch": 3.3596317646046634,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002430696798051158,
      "loss": 1.2702,
      "step": 39049
    },
    {
      "epoch": 3.359717800911985,
      "grad_norm": 1.140625,
      "learning_rate": 0.00024304648264570112,
      "loss": 1.0999,
      "step": 39050
    },
    {
      "epoch": 3.3598038372193066,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00024302328623782533,
      "loss": 0.9868,
      "step": 39051
    },
    {
      "epoch": 3.3598898735266283,
      "grad_norm": 1.5,
      "learning_rate": 0.0002430000905815564,
      "loss": 0.945,
      "step": 39052
    },
    {
      "epoch": 3.35997590983395,
      "grad_norm": 4.625,
      "learning_rate": 0.00024297689567696247,
      "loss": 1.2233,
      "step": 39053
    },
    {
      "epoch": 3.3600619461412715,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002429537015241109,
      "loss": 1.0021,
      "step": 39054
    },
    {
      "epoch": 3.360147982448593,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00024293050812306978,
      "loss": 0.775,
      "step": 39055
    },
    {
      "epoch": 3.360234018755915,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00024290731547390709,
      "loss": 0.5322,
      "step": 39056
    },
    {
      "epoch": 3.360320055063237,
      "grad_norm": 5.6875,
      "learning_rate": 0.00024288412357669037,
      "loss": 1.0449,
      "step": 39057
    },
    {
      "epoch": 3.3604060913705585,
      "grad_norm": 0.828125,
      "learning_rate": 0.00024286093243148753,
      "loss": 0.5251,
      "step": 39058
    },
    {
      "epoch": 3.36049212767788,
      "grad_norm": 0.828125,
      "learning_rate": 0.00024283774203836666,
      "loss": 0.5233,
      "step": 39059
    },
    {
      "epoch": 3.3605781639852017,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00024281455239739518,
      "loss": 0.9245,
      "step": 39060
    },
    {
      "epoch": 3.3606642002925233,
      "grad_norm": 1.265625,
      "learning_rate": 0.00024279136350864133,
      "loss": 1.5003,
      "step": 39061
    },
    {
      "epoch": 3.360750236599845,
      "grad_norm": 1.375,
      "learning_rate": 0.0002427681753721725,
      "loss": 1.0281,
      "step": 39062
    },
    {
      "epoch": 3.360836272907167,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00024274498798805677,
      "loss": 1.1108,
      "step": 39063
    },
    {
      "epoch": 3.3609223092144886,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00024272180135636207,
      "loss": 1.6374,
      "step": 39064
    },
    {
      "epoch": 3.3610083455218103,
      "grad_norm": 30976.0,
      "learning_rate": 0.00024269861547715588,
      "loss": 1.3492,
      "step": 39065
    },
    {
      "epoch": 3.361094381829132,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002426754303505062,
      "loss": 0.9092,
      "step": 39066
    },
    {
      "epoch": 3.3611804181364535,
      "grad_norm": 5.1875,
      "learning_rate": 0.00024265224597648082,
      "loss": 1.1641,
      "step": 39067
    },
    {
      "epoch": 3.361266454443775,
      "grad_norm": 1.109375,
      "learning_rate": 0.00024262906235514759,
      "loss": 1.2811,
      "step": 39068
    },
    {
      "epoch": 3.3613524907510968,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00024260587948657448,
      "loss": 1.0405,
      "step": 39069
    },
    {
      "epoch": 3.361438527058419,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00024258269737082895,
      "loss": 1.0312,
      "step": 39070
    },
    {
      "epoch": 3.3615245633657405,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00024255951600797893,
      "loss": 0.7204,
      "step": 39071
    },
    {
      "epoch": 3.361610599673062,
      "grad_norm": 1.171875,
      "learning_rate": 0.00024253633539809245,
      "loss": 1.6802,
      "step": 39072
    },
    {
      "epoch": 3.3616966359803837,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00024251315554123693,
      "loss": 1.3197,
      "step": 39073
    },
    {
      "epoch": 3.3617826722877053,
      "grad_norm": 1.296875,
      "learning_rate": 0.00024248997643748033,
      "loss": 1.7643,
      "step": 39074
    },
    {
      "epoch": 3.361868708595027,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002424667980868907,
      "loss": 0.8616,
      "step": 39075
    },
    {
      "epoch": 3.3619547449023486,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00024244362048953539,
      "loss": 1.4866,
      "step": 39076
    },
    {
      "epoch": 3.3620407812096706,
      "grad_norm": 1.265625,
      "learning_rate": 0.00024242044364548255,
      "loss": 1.2594,
      "step": 39077
    },
    {
      "epoch": 3.3621268175169923,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00024239726755479968,
      "loss": 1.5676,
      "step": 39078
    },
    {
      "epoch": 3.362212853824314,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002423740922175547,
      "loss": 1.0428,
      "step": 39079
    },
    {
      "epoch": 3.3622988901316355,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00024235091763381557,
      "loss": 0.8152,
      "step": 39080
    },
    {
      "epoch": 3.362384926438957,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002423277438036497,
      "loss": 0.6546,
      "step": 39081
    },
    {
      "epoch": 3.3624709627462788,
      "grad_norm": 1.328125,
      "learning_rate": 0.00024230457072712503,
      "loss": 1.6202,
      "step": 39082
    },
    {
      "epoch": 3.3625569990536004,
      "grad_norm": 1.21875,
      "learning_rate": 0.00024228139840430963,
      "loss": 1.3605,
      "step": 39083
    },
    {
      "epoch": 3.3626430353609225,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00024225822683527078,
      "loss": 0.6463,
      "step": 39084
    },
    {
      "epoch": 3.362729071668244,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002422350560200765,
      "loss": 1.4595,
      "step": 39085
    },
    {
      "epoch": 3.3628151079755657,
      "grad_norm": 1.296875,
      "learning_rate": 0.00024221188595879473,
      "loss": 1.4628,
      "step": 39086
    },
    {
      "epoch": 3.3629011442828873,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002421887166514929,
      "loss": 1.5732,
      "step": 39087
    },
    {
      "epoch": 3.362987180590209,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00024216554809823894,
      "loss": 1.774,
      "step": 39088
    },
    {
      "epoch": 3.3630732168975306,
      "grad_norm": 1.296875,
      "learning_rate": 0.00024214238029910074,
      "loss": 0.7267,
      "step": 39089
    },
    {
      "epoch": 3.363159253204852,
      "grad_norm": 1.21875,
      "learning_rate": 0.00024211921325414576,
      "loss": 1.3441,
      "step": 39090
    },
    {
      "epoch": 3.3632452895121743,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002420960469634421,
      "loss": 1.7154,
      "step": 39091
    },
    {
      "epoch": 3.363331325819496,
      "grad_norm": 0.96875,
      "learning_rate": 0.00024207288142705718,
      "loss": 1.5551,
      "step": 39092
    },
    {
      "epoch": 3.3634173621268175,
      "grad_norm": 1.140625,
      "learning_rate": 0.00024204971664505893,
      "loss": 0.9707,
      "step": 39093
    },
    {
      "epoch": 3.363503398434139,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00024202655261751528,
      "loss": 1.1275,
      "step": 39094
    },
    {
      "epoch": 3.3635894347414608,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00024200338934449357,
      "loss": 1.1302,
      "step": 39095
    },
    {
      "epoch": 3.3636754710487824,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0002419802268260618,
      "loss": 1.1376,
      "step": 39096
    },
    {
      "epoch": 3.3637615073561045,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002419570650622878,
      "loss": 1.0613,
      "step": 39097
    },
    {
      "epoch": 3.363847543663426,
      "grad_norm": 1.0,
      "learning_rate": 0.00024193390405323912,
      "loss": 1.334,
      "step": 39098
    },
    {
      "epoch": 3.3639335799707477,
      "grad_norm": 1.125,
      "learning_rate": 0.00024191074379898347,
      "loss": 0.7709,
      "step": 39099
    },
    {
      "epoch": 3.3640196162780693,
      "grad_norm": 1.1875,
      "learning_rate": 0.00024188758429958895,
      "loss": 0.8777,
      "step": 39100
    },
    {
      "epoch": 3.364105652585391,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00024186442555512288,
      "loss": 1.2435,
      "step": 39101
    },
    {
      "epoch": 3.3641916888927126,
      "grad_norm": 0.9375,
      "learning_rate": 0.00024184126756565327,
      "loss": 1.0383,
      "step": 39102
    },
    {
      "epoch": 3.3642777252000347,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00024181811033124746,
      "loss": 1.0815,
      "step": 39103
    },
    {
      "epoch": 3.3643637615073563,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002417949538519736,
      "loss": 1.5976,
      "step": 39104
    },
    {
      "epoch": 3.364449797814678,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00024177179812789946,
      "loss": 1.3698,
      "step": 39105
    },
    {
      "epoch": 3.3645358341219995,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0002417486431590924,
      "loss": 1.229,
      "step": 39106
    },
    {
      "epoch": 3.364621870429321,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00024172548894562057,
      "loss": 0.5678,
      "step": 39107
    },
    {
      "epoch": 3.3647079067366428,
      "grad_norm": 1.015625,
      "learning_rate": 0.00024170233548755123,
      "loss": 1.8189,
      "step": 39108
    },
    {
      "epoch": 3.3647939430439644,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002416791827849523,
      "loss": 1.13,
      "step": 39109
    },
    {
      "epoch": 3.3648799793512865,
      "grad_norm": 1.78125,
      "learning_rate": 0.00024165603083789173,
      "loss": 1.5408,
      "step": 39110
    },
    {
      "epoch": 3.364966015658608,
      "grad_norm": 2704.0,
      "learning_rate": 0.0002416328796464369,
      "loss": 0.9472,
      "step": 39111
    },
    {
      "epoch": 3.3650520519659297,
      "grad_norm": 0.953125,
      "learning_rate": 0.00024160972921065556,
      "loss": 0.8923,
      "step": 39112
    },
    {
      "epoch": 3.3651380882732513,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002415865795306157,
      "loss": 0.7167,
      "step": 39113
    },
    {
      "epoch": 3.365224124580573,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00024156343060638465,
      "loss": 1.7807,
      "step": 39114
    },
    {
      "epoch": 3.3653101608878946,
      "grad_norm": 27.625,
      "learning_rate": 0.00024154028243803034,
      "loss": 1.3636,
      "step": 39115
    },
    {
      "epoch": 3.365396197195216,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002415171350256206,
      "loss": 1.1724,
      "step": 39116
    },
    {
      "epoch": 3.3654822335025383,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00024149398836922277,
      "loss": 1.2532,
      "step": 39117
    },
    {
      "epoch": 3.36556826980986,
      "grad_norm": 4.34375,
      "learning_rate": 0.00024147084246890478,
      "loss": 1.4545,
      "step": 39118
    },
    {
      "epoch": 3.3656543061171815,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002414476973247345,
      "loss": 1.507,
      "step": 39119
    },
    {
      "epoch": 3.365740342424503,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00024142455293677917,
      "loss": 1.6719,
      "step": 39120
    },
    {
      "epoch": 3.3658263787318248,
      "grad_norm": 1.1875,
      "learning_rate": 0.00024140140930510695,
      "loss": 1.1947,
      "step": 39121
    },
    {
      "epoch": 3.3659124150391464,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002413782664297851,
      "loss": 1.7379,
      "step": 39122
    },
    {
      "epoch": 3.365998451346468,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00024135512431088158,
      "loss": 0.9966,
      "step": 39123
    },
    {
      "epoch": 3.36608448765379,
      "grad_norm": 1.140625,
      "learning_rate": 0.00024133198294846414,
      "loss": 0.8733,
      "step": 39124
    },
    {
      "epoch": 3.3661705239611117,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00024130884234260019,
      "loss": 1.1348,
      "step": 39125
    },
    {
      "epoch": 3.3662565602684333,
      "grad_norm": 3.640625,
      "learning_rate": 0.00024128570249335758,
      "loss": 0.9863,
      "step": 39126
    },
    {
      "epoch": 3.366342596575755,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002412625634008041,
      "loss": 0.5869,
      "step": 39127
    },
    {
      "epoch": 3.3664286328830766,
      "grad_norm": 1.15625,
      "learning_rate": 0.00024123942506500717,
      "loss": 1.0206,
      "step": 39128
    },
    {
      "epoch": 3.366514669190398,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002412162874860346,
      "loss": 0.9931,
      "step": 39129
    },
    {
      "epoch": 3.36660070549772,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002411931506639542,
      "loss": 0.7579,
      "step": 39130
    },
    {
      "epoch": 3.366686741805042,
      "grad_norm": 187.0,
      "learning_rate": 0.00024117001459883332,
      "loss": 1.32,
      "step": 39131
    },
    {
      "epoch": 3.3667727781123635,
      "grad_norm": 1.390625,
      "learning_rate": 0.00024114687929074002,
      "loss": 1.5592,
      "step": 39132
    },
    {
      "epoch": 3.366858814419685,
      "grad_norm": 1.109375,
      "learning_rate": 0.00024112374473974157,
      "loss": 1.4557,
      "step": 39133
    },
    {
      "epoch": 3.3669448507270068,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002411006109459058,
      "loss": 1.9332,
      "step": 39134
    },
    {
      "epoch": 3.3670308870343284,
      "grad_norm": 0.984375,
      "learning_rate": 0.00024107747790930056,
      "loss": 0.9297,
      "step": 39135
    },
    {
      "epoch": 3.36711692334165,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00024105434562999317,
      "loss": 1.5807,
      "step": 39136
    },
    {
      "epoch": 3.3672029596489716,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00024103121410805147,
      "loss": 0.9971,
      "step": 39137
    },
    {
      "epoch": 3.3672889959562937,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002410080833435433,
      "loss": 0.8552,
      "step": 39138
    },
    {
      "epoch": 3.3673750322636153,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00024098495333653593,
      "loss": 1.2851,
      "step": 39139
    },
    {
      "epoch": 3.367461068570937,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00024096182408709716,
      "loss": 1.0793,
      "step": 39140
    },
    {
      "epoch": 3.3675471048782586,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002409386955952947,
      "loss": 0.7714,
      "step": 39141
    },
    {
      "epoch": 3.36763314118558,
      "grad_norm": 9.9375,
      "learning_rate": 0.00024091556786119613,
      "loss": 1.478,
      "step": 39142
    },
    {
      "epoch": 3.367719177492902,
      "grad_norm": 1.171875,
      "learning_rate": 0.00024089244088486934,
      "loss": 0.9888,
      "step": 39143
    },
    {
      "epoch": 3.3678052138002235,
      "grad_norm": 0.90625,
      "learning_rate": 0.00024086931466638157,
      "loss": 0.6673,
      "step": 39144
    },
    {
      "epoch": 3.3678912501075455,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002408461892058007,
      "loss": 0.529,
      "step": 39145
    },
    {
      "epoch": 3.367977286414867,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00024082306450319446,
      "loss": 0.839,
      "step": 39146
    },
    {
      "epoch": 3.368063322722189,
      "grad_norm": 446.0,
      "learning_rate": 0.0002407999405586302,
      "loss": 1.1428,
      "step": 39147
    },
    {
      "epoch": 3.3681493590295104,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00024077681737217588,
      "loss": 0.7261,
      "step": 39148
    },
    {
      "epoch": 3.368235395336832,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00024075369494389875,
      "loss": 0.5842,
      "step": 39149
    },
    {
      "epoch": 3.3683214316441537,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00024073057327386667,
      "loss": 1.1702,
      "step": 39150
    },
    {
      "epoch": 3.3684074679514753,
      "grad_norm": 2.515625,
      "learning_rate": 0.0002407074523621474,
      "loss": 1.2415,
      "step": 39151
    },
    {
      "epoch": 3.3684935042587973,
      "grad_norm": 1.015625,
      "learning_rate": 0.00024068433220880825,
      "loss": 1.5896,
      "step": 39152
    },
    {
      "epoch": 3.368579540566119,
      "grad_norm": 19.0,
      "learning_rate": 0.00024066121281391694,
      "loss": 0.8123,
      "step": 39153
    },
    {
      "epoch": 3.3686655768734406,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002406380941775414,
      "loss": 0.6123,
      "step": 39154
    },
    {
      "epoch": 3.368751613180762,
      "grad_norm": 1.28125,
      "learning_rate": 0.00024061497629974878,
      "loss": 1.5231,
      "step": 39155
    },
    {
      "epoch": 3.368837649488084,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00024059185918060688,
      "loss": 0.769,
      "step": 39156
    },
    {
      "epoch": 3.368923685795406,
      "grad_norm": 18.25,
      "learning_rate": 0.00024056874282018354,
      "loss": 1.6717,
      "step": 39157
    },
    {
      "epoch": 3.3690097221027275,
      "grad_norm": 56.25,
      "learning_rate": 0.000240545627218546,
      "loss": 0.7504,
      "step": 39158
    },
    {
      "epoch": 3.369095758410049,
      "grad_norm": 3.671875,
      "learning_rate": 0.000240522512375762,
      "loss": 0.813,
      "step": 39159
    },
    {
      "epoch": 3.369181794717371,
      "grad_norm": 1.328125,
      "learning_rate": 0.00024049939829189938,
      "loss": 1.1402,
      "step": 39160
    },
    {
      "epoch": 3.3692678310246924,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002404762849670254,
      "loss": 0.5033,
      "step": 39161
    },
    {
      "epoch": 3.369353867332014,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00024045317240120796,
      "loss": 0.7583,
      "step": 39162
    },
    {
      "epoch": 3.3694399036393357,
      "grad_norm": 1.25,
      "learning_rate": 0.00024043006059451432,
      "loss": 1.5554,
      "step": 39163
    },
    {
      "epoch": 3.3695259399466577,
      "grad_norm": 1.59375,
      "learning_rate": 0.0002404069495470123,
      "loss": 0.9206,
      "step": 39164
    },
    {
      "epoch": 3.3696119762539793,
      "grad_norm": 10.8125,
      "learning_rate": 0.00024038383925876968,
      "loss": 1.4011,
      "step": 39165
    },
    {
      "epoch": 3.369698012561301,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00024036072972985362,
      "loss": 1.2449,
      "step": 39166
    },
    {
      "epoch": 3.3697840488686226,
      "grad_norm": 39.0,
      "learning_rate": 0.0002403376209603319,
      "loss": 1.1923,
      "step": 39167
    },
    {
      "epoch": 3.369870085175944,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00024031451295027234,
      "loss": 0.7,
      "step": 39168
    },
    {
      "epoch": 3.369956121483266,
      "grad_norm": 0.953125,
      "learning_rate": 0.00024029140569974217,
      "loss": 0.5273,
      "step": 39169
    },
    {
      "epoch": 3.3700421577905875,
      "grad_norm": 1.0,
      "learning_rate": 0.0002402682992088091,
      "loss": 1.2633,
      "step": 39170
    },
    {
      "epoch": 3.3701281940979095,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00024024519347754092,
      "loss": 0.9967,
      "step": 39171
    },
    {
      "epoch": 3.370214230405231,
      "grad_norm": 0.953125,
      "learning_rate": 0.00024022208850600486,
      "loss": 0.868,
      "step": 39172
    },
    {
      "epoch": 3.370300266712553,
      "grad_norm": 1464.0,
      "learning_rate": 0.00024019898429426885,
      "loss": 0.6165,
      "step": 39173
    },
    {
      "epoch": 3.3703863030198744,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00024017588084240005,
      "loss": 1.0685,
      "step": 39174
    },
    {
      "epoch": 3.370472339327196,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00024015277815046633,
      "loss": 1.0793,
      "step": 39175
    },
    {
      "epoch": 3.3705583756345177,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00024012967621853538,
      "loss": 1.0004,
      "step": 39176
    },
    {
      "epoch": 3.3706444119418393,
      "grad_norm": 5.75,
      "learning_rate": 0.00024010657504667415,
      "loss": 1.2255,
      "step": 39177
    },
    {
      "epoch": 3.3707304482491613,
      "grad_norm": 1.375,
      "learning_rate": 0.00024008347463495106,
      "loss": 1.0351,
      "step": 39178
    },
    {
      "epoch": 3.370816484556483,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002400603749834331,
      "loss": 0.7913,
      "step": 39179
    },
    {
      "epoch": 3.3709025208638046,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002400372760921879,
      "loss": 1.0865,
      "step": 39180
    },
    {
      "epoch": 3.370988557171126,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002400141779612834,
      "loss": 0.645,
      "step": 39181
    },
    {
      "epoch": 3.371074593478448,
      "grad_norm": 442.0,
      "learning_rate": 0.00023999108059078662,
      "loss": 1.1425,
      "step": 39182
    },
    {
      "epoch": 3.3711606297857695,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00023996798398076536,
      "loss": 1.067,
      "step": 39183
    },
    {
      "epoch": 3.371246666093091,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00023994488813128734,
      "loss": 1.5945,
      "step": 39184
    },
    {
      "epoch": 3.371332702400413,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002399217930424198,
      "loss": 1.5513,
      "step": 39185
    },
    {
      "epoch": 3.371418738707735,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00023989869871423043,
      "loss": 1.3334,
      "step": 39186
    },
    {
      "epoch": 3.3715047750150564,
      "grad_norm": 3.84375,
      "learning_rate": 0.00023987560514678692,
      "loss": 0.9772,
      "step": 39187
    },
    {
      "epoch": 3.371590811322378,
      "grad_norm": 1.421875,
      "learning_rate": 0.00023985251234015653,
      "loss": 1.1753,
      "step": 39188
    },
    {
      "epoch": 3.3716768476296997,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00023982942029440713,
      "loss": 0.7051,
      "step": 39189
    },
    {
      "epoch": 3.3717628839370213,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00023980632900960587,
      "loss": 0.674,
      "step": 39190
    },
    {
      "epoch": 3.371848920244343,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002397832384858205,
      "loss": 1.5116,
      "step": 39191
    },
    {
      "epoch": 3.371934956551665,
      "grad_norm": 8.625,
      "learning_rate": 0.00023976014872311875,
      "loss": 1.8446,
      "step": 39192
    },
    {
      "epoch": 3.3720209928589866,
      "grad_norm": 1.265625,
      "learning_rate": 0.00023973705972156774,
      "loss": 1.1005,
      "step": 39193
    },
    {
      "epoch": 3.3721070291663082,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002397139714812352,
      "loss": 0.711,
      "step": 39194
    },
    {
      "epoch": 3.37219306547363,
      "grad_norm": 0.875,
      "learning_rate": 0.00023969088400218892,
      "loss": 1.409,
      "step": 39195
    },
    {
      "epoch": 3.3722791017809515,
      "grad_norm": 7872.0,
      "learning_rate": 0.0002396677972844959,
      "loss": 1.1504,
      "step": 39196
    },
    {
      "epoch": 3.372365138088273,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000239644711328224,
      "loss": 0.5383,
      "step": 39197
    },
    {
      "epoch": 3.3724511743955947,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00023962162613344086,
      "loss": 1.0988,
      "step": 39198
    },
    {
      "epoch": 3.372537210702917,
      "grad_norm": 1.0625,
      "learning_rate": 0.00023959854170021368,
      "loss": 1.337,
      "step": 39199
    },
    {
      "epoch": 3.3726232470102384,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00023957545802861003,
      "loss": 0.5982,
      "step": 39200
    },
    {
      "epoch": 3.37270928331756,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00023955237511869777,
      "loss": 0.8419,
      "step": 39201
    },
    {
      "epoch": 3.3727953196248817,
      "grad_norm": 1.5625,
      "learning_rate": 0.00023952929297054395,
      "loss": 1.0248,
      "step": 39202
    },
    {
      "epoch": 3.3728813559322033,
      "grad_norm": 5.3125,
      "learning_rate": 0.0002395062115842165,
      "loss": 0.5894,
      "step": 39203
    },
    {
      "epoch": 3.372967392239525,
      "grad_norm": 1.140625,
      "learning_rate": 0.00023948313095978246,
      "loss": 1.1281,
      "step": 39204
    },
    {
      "epoch": 3.3730534285468465,
      "grad_norm": 1.328125,
      "learning_rate": 0.00023946005109730967,
      "loss": 1.0415,
      "step": 39205
    },
    {
      "epoch": 3.3731394648541686,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00023943697199686572,
      "loss": 1.0196,
      "step": 39206
    },
    {
      "epoch": 3.3732255011614902,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00023941389365851774,
      "loss": 0.8138,
      "step": 39207
    },
    {
      "epoch": 3.373311537468812,
      "grad_norm": 18.375,
      "learning_rate": 0.0002393908160823335,
      "loss": 1.6999,
      "step": 39208
    },
    {
      "epoch": 3.3733975737761335,
      "grad_norm": 143.0,
      "learning_rate": 0.00023936773926838058,
      "loss": 1.2439,
      "step": 39209
    },
    {
      "epoch": 3.373483610083455,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00023934466321672614,
      "loss": 0.7982,
      "step": 39210
    },
    {
      "epoch": 3.3735696463907767,
      "grad_norm": 7.625,
      "learning_rate": 0.00023932158792743785,
      "loss": 1.1031,
      "step": 39211
    },
    {
      "epoch": 3.373655682698099,
      "grad_norm": 15.3125,
      "learning_rate": 0.00023929851340058345,
      "loss": 1.0035,
      "step": 39212
    },
    {
      "epoch": 3.3737417190054204,
      "grad_norm": 1.203125,
      "learning_rate": 0.00023927543963622998,
      "loss": 1.765,
      "step": 39213
    },
    {
      "epoch": 3.373827755312742,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00023925236663444515,
      "loss": 0.9953,
      "step": 39214
    },
    {
      "epoch": 3.3739137916200637,
      "grad_norm": 1.515625,
      "learning_rate": 0.00023922929439529645,
      "loss": 0.7999,
      "step": 39215
    },
    {
      "epoch": 3.3739998279273853,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002392062229188513,
      "loss": 2.0024,
      "step": 39216
    },
    {
      "epoch": 3.374085864234707,
      "grad_norm": 1.15625,
      "learning_rate": 0.00023918315220517743,
      "loss": 1.739,
      "step": 39217
    },
    {
      "epoch": 3.374171900542029,
      "grad_norm": 1.078125,
      "learning_rate": 0.000239160082254342,
      "loss": 1.0123,
      "step": 39218
    },
    {
      "epoch": 3.3742579368493506,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00023913701306641272,
      "loss": 0.6135,
      "step": 39219
    },
    {
      "epoch": 3.3743439731566722,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00023911394464145674,
      "loss": 1.4339,
      "step": 39220
    },
    {
      "epoch": 3.374430009463994,
      "grad_norm": 1.28125,
      "learning_rate": 0.00023909087697954174,
      "loss": 1.3553,
      "step": 39221
    },
    {
      "epoch": 3.3745160457713155,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002390678100807354,
      "loss": 1.2652,
      "step": 39222
    },
    {
      "epoch": 3.374602082078637,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002390447439451048,
      "loss": 0.7613,
      "step": 39223
    },
    {
      "epoch": 3.3746881183859587,
      "grad_norm": 2.28125,
      "learning_rate": 0.0002390216785727175,
      "loss": 1.19,
      "step": 39224
    },
    {
      "epoch": 3.374774154693281,
      "grad_norm": 1.28125,
      "learning_rate": 0.00023899861396364124,
      "loss": 0.9924,
      "step": 39225
    },
    {
      "epoch": 3.3748601910006024,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00023897555011794308,
      "loss": 0.5459,
      "step": 39226
    },
    {
      "epoch": 3.374946227307924,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00023895248703569073,
      "loss": 0.8188,
      "step": 39227
    },
    {
      "epoch": 3.3750322636152457,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00023892942471695171,
      "loss": 1.4574,
      "step": 39228
    },
    {
      "epoch": 3.3751182999225673,
      "grad_norm": 1.609375,
      "learning_rate": 0.00023890636316179314,
      "loss": 1.2169,
      "step": 39229
    },
    {
      "epoch": 3.375204336229889,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00023888330237028293,
      "loss": 1.0549,
      "step": 39230
    },
    {
      "epoch": 3.3752903725372105,
      "grad_norm": 2.15625,
      "learning_rate": 0.00023886024234248804,
      "loss": 1.0701,
      "step": 39231
    },
    {
      "epoch": 3.3753764088445326,
      "grad_norm": 1.125,
      "learning_rate": 0.00023883718307847617,
      "loss": 1.0641,
      "step": 39232
    },
    {
      "epoch": 3.3754624451518542,
      "grad_norm": 4.875,
      "learning_rate": 0.00023881412457831498,
      "loss": 0.8365,
      "step": 39233
    },
    {
      "epoch": 3.375548481459176,
      "grad_norm": 12.0625,
      "learning_rate": 0.00023879106684207143,
      "loss": 1.115,
      "step": 39234
    },
    {
      "epoch": 3.3756345177664975,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00023876800986981323,
      "loss": 1.0646,
      "step": 39235
    },
    {
      "epoch": 3.375720554073819,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00023874495366160798,
      "loss": 0.5328,
      "step": 39236
    },
    {
      "epoch": 3.3758065903811407,
      "grad_norm": 1.890625,
      "learning_rate": 0.00023872189821752272,
      "loss": 0.8219,
      "step": 39237
    },
    {
      "epoch": 3.3758926266884624,
      "grad_norm": 1.125,
      "learning_rate": 0.0002386988435376251,
      "loss": 1.5814,
      "step": 39238
    },
    {
      "epoch": 3.3759786629957844,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00023867578962198278,
      "loss": 1.0613,
      "step": 39239
    },
    {
      "epoch": 3.376064699303106,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00023865273647066273,
      "loss": 1.0679,
      "step": 39240
    },
    {
      "epoch": 3.3761507356104277,
      "grad_norm": 2.78125,
      "learning_rate": 0.00023862968408373258,
      "loss": 0.6859,
      "step": 39241
    },
    {
      "epoch": 3.3762367719177493,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00023860663246125997,
      "loss": 1.3447,
      "step": 39242
    },
    {
      "epoch": 3.376322808225071,
      "grad_norm": 1.75,
      "learning_rate": 0.0002385835816033119,
      "loss": 0.7207,
      "step": 39243
    },
    {
      "epoch": 3.3764088445323925,
      "grad_norm": 0.953125,
      "learning_rate": 0.00023856053150995627,
      "loss": 1.2055,
      "step": 39244
    },
    {
      "epoch": 3.376494880839714,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00023853748218126,
      "loss": 1.188,
      "step": 39245
    },
    {
      "epoch": 3.3765809171470362,
      "grad_norm": 1.046875,
      "learning_rate": 0.00023851443361729076,
      "loss": 0.8971,
      "step": 39246
    },
    {
      "epoch": 3.376666953454358,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00023849138581811613,
      "loss": 1.1472,
      "step": 39247
    },
    {
      "epoch": 3.3767529897616795,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00023846833878380313,
      "loss": 1.078,
      "step": 39248
    },
    {
      "epoch": 3.376839026069001,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00023844529251441937,
      "loss": 1.1597,
      "step": 39249
    },
    {
      "epoch": 3.3769250623763227,
      "grad_norm": 12.6875,
      "learning_rate": 0.0002384222470100325,
      "loss": 1.2699,
      "step": 39250
    },
    {
      "epoch": 3.3770110986836444,
      "grad_norm": 1.09375,
      "learning_rate": 0.00023839920227070928,
      "loss": 0.6894,
      "step": 39251
    },
    {
      "epoch": 3.377097134990966,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00023837615829651787,
      "loss": 1.1373,
      "step": 39252
    },
    {
      "epoch": 3.377183171298288,
      "grad_norm": 1.390625,
      "learning_rate": 0.00023835311508752516,
      "loss": 1.0719,
      "step": 39253
    },
    {
      "epoch": 3.3772692076056097,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002383300726437987,
      "loss": 1.3967,
      "step": 39254
    },
    {
      "epoch": 3.3773552439129313,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00023830703096540601,
      "loss": 0.8713,
      "step": 39255
    },
    {
      "epoch": 3.377441280220253,
      "grad_norm": 1.1875,
      "learning_rate": 0.00023828399005241423,
      "loss": 1.1386,
      "step": 39256
    },
    {
      "epoch": 3.3775273165275745,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002382609499048909,
      "loss": 1.1281,
      "step": 39257
    },
    {
      "epoch": 3.377613352834896,
      "grad_norm": 2.1875,
      "learning_rate": 0.0002382379105229035,
      "loss": 1.0246,
      "step": 39258
    },
    {
      "epoch": 3.377699389142218,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00023821487190651915,
      "loss": 0.583,
      "step": 39259
    },
    {
      "epoch": 3.37778542544954,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00023819183405580558,
      "loss": 1.2327,
      "step": 39260
    },
    {
      "epoch": 3.3778714617568615,
      "grad_norm": 1.0,
      "learning_rate": 0.00023816879697082977,
      "loss": 0.6021,
      "step": 39261
    },
    {
      "epoch": 3.377957498064183,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00023814576065165933,
      "loss": 1.1555,
      "step": 39262
    },
    {
      "epoch": 3.3780435343715047,
      "grad_norm": 1.390625,
      "learning_rate": 0.00023812272509836181,
      "loss": 1.6689,
      "step": 39263
    },
    {
      "epoch": 3.3781295706788264,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00023809969031100416,
      "loss": 0.6675,
      "step": 39264
    },
    {
      "epoch": 3.378215606986148,
      "grad_norm": 4.90625,
      "learning_rate": 0.000238076656289654,
      "loss": 0.9443,
      "step": 39265
    },
    {
      "epoch": 3.3783016432934696,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002380536230343789,
      "loss": 1.9339,
      "step": 39266
    },
    {
      "epoch": 3.3783876796007917,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00023803059054524573,
      "loss": 0.402,
      "step": 39267
    },
    {
      "epoch": 3.3784737159081133,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00023800755882232222,
      "loss": 1.2259,
      "step": 39268
    },
    {
      "epoch": 3.378559752215435,
      "grad_norm": 1.25,
      "learning_rate": 0.00023798452786567582,
      "loss": 0.7611,
      "step": 39269
    },
    {
      "epoch": 3.3786457885227565,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002379614976753735,
      "loss": 1.6644,
      "step": 39270
    },
    {
      "epoch": 3.378731824830078,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002379384682514828,
      "loss": 2.0684,
      "step": 39271
    },
    {
      "epoch": 3.3788178611374,
      "grad_norm": 1.015625,
      "learning_rate": 0.00023791543959407135,
      "loss": 0.7171,
      "step": 39272
    },
    {
      "epoch": 3.378903897444722,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00023789241170320608,
      "loss": 0.6278,
      "step": 39273
    },
    {
      "epoch": 3.3789899337520435,
      "grad_norm": 156.0,
      "learning_rate": 0.00023786938457895474,
      "loss": 1.0491,
      "step": 39274
    },
    {
      "epoch": 3.379075970059365,
      "grad_norm": 1.03125,
      "learning_rate": 0.00023784635822138422,
      "loss": 1.3514,
      "step": 39275
    },
    {
      "epoch": 3.3791620063666867,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00023782333263056216,
      "loss": 1.0639,
      "step": 39276
    },
    {
      "epoch": 3.3792480426740084,
      "grad_norm": 1.25,
      "learning_rate": 0.00023780030780655605,
      "loss": 1.5722,
      "step": 39277
    },
    {
      "epoch": 3.37933407898133,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002377772837494328,
      "loss": 1.2798,
      "step": 39278
    },
    {
      "epoch": 3.379420115288652,
      "grad_norm": 0.828125,
      "learning_rate": 0.00023775426045926008,
      "loss": 1.2837,
      "step": 39279
    },
    {
      "epoch": 3.3795061515959737,
      "grad_norm": 0.78125,
      "learning_rate": 0.00023773123793610528,
      "loss": 1.2165,
      "step": 39280
    },
    {
      "epoch": 3.3795921879032953,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002377082161800354,
      "loss": 1.4056,
      "step": 39281
    },
    {
      "epoch": 3.379678224210617,
      "grad_norm": 2.6875,
      "learning_rate": 0.00023768519519111803,
      "loss": 0.5064,
      "step": 39282
    },
    {
      "epoch": 3.3797642605179385,
      "grad_norm": 9.4375,
      "learning_rate": 0.0002376621749694206,
      "loss": 1.6883,
      "step": 39283
    },
    {
      "epoch": 3.37985029682526,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002376391555150101,
      "loss": 1.2097,
      "step": 39284
    },
    {
      "epoch": 3.379936333132582,
      "grad_norm": 1.234375,
      "learning_rate": 0.00023761613682795425,
      "loss": 1.435,
      "step": 39285
    },
    {
      "epoch": 3.380022369439904,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002375931189083199,
      "loss": 0.9258,
      "step": 39286
    },
    {
      "epoch": 3.3801084057472255,
      "grad_norm": 1.921875,
      "learning_rate": 0.0002375701017561747,
      "loss": 1.2943,
      "step": 39287
    },
    {
      "epoch": 3.380194442054547,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00023754708537158597,
      "loss": 1.7013,
      "step": 39288
    },
    {
      "epoch": 3.3802804783618687,
      "grad_norm": 1.515625,
      "learning_rate": 0.00023752406975462094,
      "loss": 1.4183,
      "step": 39289
    },
    {
      "epoch": 3.3803665146691904,
      "grad_norm": 0.96875,
      "learning_rate": 0.00023750105490534708,
      "loss": 0.9641,
      "step": 39290
    },
    {
      "epoch": 3.380452550976512,
      "grad_norm": 1.234375,
      "learning_rate": 0.00023747804082383146,
      "loss": 1.0623,
      "step": 39291
    },
    {
      "epoch": 3.3805385872838336,
      "grad_norm": 1.265625,
      "learning_rate": 0.00023745502751014149,
      "loss": 0.7012,
      "step": 39292
    },
    {
      "epoch": 3.3806246235911557,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002374320149643447,
      "loss": 1.0637,
      "step": 39293
    },
    {
      "epoch": 3.3807106598984773,
      "grad_norm": 1.78125,
      "learning_rate": 0.000237409003186508,
      "loss": 0.7529,
      "step": 39294
    },
    {
      "epoch": 3.380796696205799,
      "grad_norm": 1.390625,
      "learning_rate": 0.00023738599217669888,
      "loss": 1.7613,
      "step": 39295
    },
    {
      "epoch": 3.3808827325131205,
      "grad_norm": 3.59375,
      "learning_rate": 0.00023736298193498484,
      "loss": 1.0283,
      "step": 39296
    },
    {
      "epoch": 3.380968768820442,
      "grad_norm": 0.921875,
      "learning_rate": 0.00023733997246143284,
      "loss": 0.7592,
      "step": 39297
    },
    {
      "epoch": 3.381054805127764,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00023731696375611035,
      "loss": 1.3513,
      "step": 39298
    },
    {
      "epoch": 3.3811408414350854,
      "grad_norm": 0.921875,
      "learning_rate": 0.00023729395581908482,
      "loss": 0.4051,
      "step": 39299
    },
    {
      "epoch": 3.3812268777424075,
      "grad_norm": 1.546875,
      "learning_rate": 0.0002372709486504232,
      "loss": 0.9581,
      "step": 39300
    },
    {
      "epoch": 3.381312914049729,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00023724794225019307,
      "loss": 0.7435,
      "step": 39301
    },
    {
      "epoch": 3.3813989503570507,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00023722493661846152,
      "loss": 0.4652,
      "step": 39302
    },
    {
      "epoch": 3.3814849866643724,
      "grad_norm": 1.015625,
      "learning_rate": 0.00023720193175529587,
      "loss": 0.8716,
      "step": 39303
    },
    {
      "epoch": 3.381571022971694,
      "grad_norm": 1.296875,
      "learning_rate": 0.00023717892766076366,
      "loss": 0.8959,
      "step": 39304
    },
    {
      "epoch": 3.3816570592790156,
      "grad_norm": 7456.0,
      "learning_rate": 0.00023715592433493177,
      "loss": 0.6474,
      "step": 39305
    },
    {
      "epoch": 3.3817430955863372,
      "grad_norm": 14.625,
      "learning_rate": 0.00023713292177786772,
      "loss": 1.1512,
      "step": 39306
    },
    {
      "epoch": 3.3818291318936593,
      "grad_norm": 3.734375,
      "learning_rate": 0.00023710991998963888,
      "loss": 1.3453,
      "step": 39307
    },
    {
      "epoch": 3.381915168200981,
      "grad_norm": 1.109375,
      "learning_rate": 0.00023708691897031227,
      "loss": 1.3223,
      "step": 39308
    },
    {
      "epoch": 3.3820012045083025,
      "grad_norm": 19.25,
      "learning_rate": 0.00023706391871995526,
      "loss": 1.4393,
      "step": 39309
    },
    {
      "epoch": 3.382087240815624,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002370409192386353,
      "loss": 1.3837,
      "step": 39310
    },
    {
      "epoch": 3.382173277122946,
      "grad_norm": 1.046875,
      "learning_rate": 0.00023701792052641934,
      "loss": 1.5582,
      "step": 39311
    },
    {
      "epoch": 3.3822593134302674,
      "grad_norm": 12.9375,
      "learning_rate": 0.00023699492258337485,
      "loss": 1.5618,
      "step": 39312
    },
    {
      "epoch": 3.382345349737589,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00023697192540956925,
      "loss": 1.2008,
      "step": 39313
    },
    {
      "epoch": 3.382431386044911,
      "grad_norm": 186.0,
      "learning_rate": 0.00023694892900506936,
      "loss": 1.4431,
      "step": 39314
    },
    {
      "epoch": 3.3825174223522327,
      "grad_norm": 2.25,
      "learning_rate": 0.00023692593336994294,
      "loss": 0.9016,
      "step": 39315
    },
    {
      "epoch": 3.3826034586595544,
      "grad_norm": 39.5,
      "learning_rate": 0.00023690293850425683,
      "loss": 1.174,
      "step": 39316
    },
    {
      "epoch": 3.382689494966876,
      "grad_norm": 1.78125,
      "learning_rate": 0.00023687994440807836,
      "loss": 0.9539,
      "step": 39317
    },
    {
      "epoch": 3.3827755312741976,
      "grad_norm": 1.3125,
      "learning_rate": 0.00023685695108147516,
      "loss": 1.564,
      "step": 39318
    },
    {
      "epoch": 3.3828615675815192,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00023683395852451395,
      "loss": 1.4289,
      "step": 39319
    },
    {
      "epoch": 3.382947603888841,
      "grad_norm": 3.34375,
      "learning_rate": 0.00023681096673726227,
      "loss": 1.522,
      "step": 39320
    },
    {
      "epoch": 3.383033640196163,
      "grad_norm": 1.296875,
      "learning_rate": 0.00023678797571978749,
      "loss": 1.0965,
      "step": 39321
    },
    {
      "epoch": 3.3831196765034846,
      "grad_norm": 2.375,
      "learning_rate": 0.00023676498547215653,
      "loss": 1.0828,
      "step": 39322
    },
    {
      "epoch": 3.383205712810806,
      "grad_norm": 0.90625,
      "learning_rate": 0.00023674199599443675,
      "loss": 1.0704,
      "step": 39323
    },
    {
      "epoch": 3.383291749118128,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00023671900728669564,
      "loss": 0.9471,
      "step": 39324
    },
    {
      "epoch": 3.3833777854254494,
      "grad_norm": 9.6875,
      "learning_rate": 0.00023669601934899986,
      "loss": 1.3912,
      "step": 39325
    },
    {
      "epoch": 3.383463821732771,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00023667303218141732,
      "loss": 1.1868,
      "step": 39326
    },
    {
      "epoch": 3.3835498580400927,
      "grad_norm": 1.34375,
      "learning_rate": 0.00023665004578401482,
      "loss": 0.621,
      "step": 39327
    },
    {
      "epoch": 3.3836358943474147,
      "grad_norm": 0.921875,
      "learning_rate": 0.00023662706015685974,
      "loss": 1.1071,
      "step": 39328
    },
    {
      "epoch": 3.3837219306547364,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002366040753000195,
      "loss": 1.449,
      "step": 39329
    },
    {
      "epoch": 3.383807966962058,
      "grad_norm": 39.5,
      "learning_rate": 0.00023658109121356085,
      "loss": 1.7971,
      "step": 39330
    },
    {
      "epoch": 3.3838940032693796,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002365581078975515,
      "loss": 1.6232,
      "step": 39331
    },
    {
      "epoch": 3.3839800395767012,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00023653512535205823,
      "loss": 1.3781,
      "step": 39332
    },
    {
      "epoch": 3.3840660758840233,
      "grad_norm": 1.15625,
      "learning_rate": 0.00023651214357714852,
      "loss": 0.8171,
      "step": 39333
    },
    {
      "epoch": 3.384152112191345,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00023648916257288966,
      "loss": 1.0581,
      "step": 39334
    },
    {
      "epoch": 3.3842381484986666,
      "grad_norm": 33.0,
      "learning_rate": 0.00023646618233934863,
      "loss": 1.0357,
      "step": 39335
    },
    {
      "epoch": 3.384324184805988,
      "grad_norm": 1.203125,
      "learning_rate": 0.00023644320287659272,
      "loss": 1.1323,
      "step": 39336
    },
    {
      "epoch": 3.38441022111331,
      "grad_norm": 85.5,
      "learning_rate": 0.00023642022418468938,
      "loss": 1.0186,
      "step": 39337
    },
    {
      "epoch": 3.3844962574206314,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00023639724626370545,
      "loss": 1.1293,
      "step": 39338
    },
    {
      "epoch": 3.384582293727953,
      "grad_norm": 0.984375,
      "learning_rate": 0.00023637426911370828,
      "loss": 1.2348,
      "step": 39339
    },
    {
      "epoch": 3.384668330035275,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00023635129273476534,
      "loss": 1.2714,
      "step": 39340
    },
    {
      "epoch": 3.3847543663425967,
      "grad_norm": 1.296875,
      "learning_rate": 0.00023632831712694336,
      "loss": 0.753,
      "step": 39341
    },
    {
      "epoch": 3.3848404026499184,
      "grad_norm": 16064.0,
      "learning_rate": 0.00023630534229030997,
      "loss": 0.8878,
      "step": 39342
    },
    {
      "epoch": 3.38492643895724,
      "grad_norm": 1.375,
      "learning_rate": 0.00023628236822493198,
      "loss": 1.4808,
      "step": 39343
    },
    {
      "epoch": 3.3850124752645616,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00023625939493087677,
      "loss": 0.9713,
      "step": 39344
    },
    {
      "epoch": 3.3850985115718832,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00023623642240821175,
      "loss": 0.8214,
      "step": 39345
    },
    {
      "epoch": 3.385184547879205,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00023621345065700367,
      "loss": 1.0486,
      "step": 39346
    },
    {
      "epoch": 3.385270584186527,
      "grad_norm": 5.9375,
      "learning_rate": 0.00023619047967732,
      "loss": 0.5448,
      "step": 39347
    },
    {
      "epoch": 3.3853566204938486,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000236167509469228,
      "loss": 1.2326,
      "step": 39348
    },
    {
      "epoch": 3.38544265680117,
      "grad_norm": 40.5,
      "learning_rate": 0.00023614454003279456,
      "loss": 1.293,
      "step": 39349
    },
    {
      "epoch": 3.385528693108492,
      "grad_norm": 82.0,
      "learning_rate": 0.00023612157136808698,
      "loss": 1.1938,
      "step": 39350
    },
    {
      "epoch": 3.3856147294158134,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0002360986034751727,
      "loss": 0.8787,
      "step": 39351
    },
    {
      "epoch": 3.385700765723135,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002360756363541185,
      "loss": 1.2285,
      "step": 39352
    },
    {
      "epoch": 3.3857868020304567,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00023605267000499175,
      "loss": 1.6703,
      "step": 39353
    },
    {
      "epoch": 3.3858728383377787,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00023602970442785975,
      "loss": 1.2284,
      "step": 39354
    },
    {
      "epoch": 3.3859588746451004,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002360067396227894,
      "loss": 1.3157,
      "step": 39355
    },
    {
      "epoch": 3.386044910952422,
      "grad_norm": 30.625,
      "learning_rate": 0.00023598377558984812,
      "loss": 1.5292,
      "step": 39356
    },
    {
      "epoch": 3.3861309472597436,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00023596081232910276,
      "loss": 1.0763,
      "step": 39357
    },
    {
      "epoch": 3.3862169835670652,
      "grad_norm": 41.75,
      "learning_rate": 0.00023593784984062073,
      "loss": 1.4661,
      "step": 39358
    },
    {
      "epoch": 3.386303019874387,
      "grad_norm": 88.0,
      "learning_rate": 0.00023591488812446925,
      "loss": 1.3604,
      "step": 39359
    },
    {
      "epoch": 3.3863890561817085,
      "grad_norm": 2192.0,
      "learning_rate": 0.00023589192718071522,
      "loss": 1.5593,
      "step": 39360
    },
    {
      "epoch": 3.3864750924890306,
      "grad_norm": 1.125,
      "learning_rate": 0.0002358689670094259,
      "loss": 0.9237,
      "step": 39361
    },
    {
      "epoch": 3.386561128796352,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002358460076106685,
      "loss": 1.0074,
      "step": 39362
    },
    {
      "epoch": 3.386647165103674,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00023582304898451018,
      "loss": 1.5028,
      "step": 39363
    },
    {
      "epoch": 3.3867332014109954,
      "grad_norm": 9.9375,
      "learning_rate": 0.0002358000911310182,
      "loss": 0.856,
      "step": 39364
    },
    {
      "epoch": 3.386819237718317,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00023577713405025942,
      "loss": 1.7415,
      "step": 39365
    },
    {
      "epoch": 3.3869052740256387,
      "grad_norm": 1208.0,
      "learning_rate": 0.00023575417774230112,
      "loss": 1.5687,
      "step": 39366
    },
    {
      "epoch": 3.3869913103329603,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002357312222072106,
      "loss": 0.6858,
      "step": 39367
    },
    {
      "epoch": 3.3870773466402824,
      "grad_norm": 33.25,
      "learning_rate": 0.00023570826744505474,
      "loss": 0.935,
      "step": 39368
    },
    {
      "epoch": 3.387163382947604,
      "grad_norm": 1.09375,
      "learning_rate": 0.00023568531345590077,
      "loss": 0.6091,
      "step": 39369
    },
    {
      "epoch": 3.3872494192549256,
      "grad_norm": 4.65625,
      "learning_rate": 0.000235662360239816,
      "loss": 0.9836,
      "step": 39370
    },
    {
      "epoch": 3.3873354555622472,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00023563940779686732,
      "loss": 1.2379,
      "step": 39371
    },
    {
      "epoch": 3.387421491869569,
      "grad_norm": 6.0,
      "learning_rate": 0.00023561645612712206,
      "loss": 0.7729,
      "step": 39372
    },
    {
      "epoch": 3.3875075281768905,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002355935052306471,
      "loss": 0.7949,
      "step": 39373
    },
    {
      "epoch": 3.387593564484212,
      "grad_norm": 59.75,
      "learning_rate": 0.00023557055510750968,
      "loss": 1.0183,
      "step": 39374
    },
    {
      "epoch": 3.387679600791534,
      "grad_norm": 1.0625,
      "learning_rate": 0.00023554760575777718,
      "loss": 0.9222,
      "step": 39375
    },
    {
      "epoch": 3.387765637098856,
      "grad_norm": 1.328125,
      "learning_rate": 0.00023552465718151632,
      "loss": 1.4189,
      "step": 39376
    },
    {
      "epoch": 3.3878516734061774,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002355017093787944,
      "loss": 1.2175,
      "step": 39377
    },
    {
      "epoch": 3.387937709713499,
      "grad_norm": 1.40625,
      "learning_rate": 0.00023547876234967868,
      "loss": 1.7169,
      "step": 39378
    },
    {
      "epoch": 3.3880237460208207,
      "grad_norm": 1.09375,
      "learning_rate": 0.000235455816094236,
      "loss": 1.2449,
      "step": 39379
    },
    {
      "epoch": 3.3881097823281423,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00023543287061253355,
      "loss": 1.0913,
      "step": 39380
    },
    {
      "epoch": 3.388195818635464,
      "grad_norm": 40.5,
      "learning_rate": 0.00023540992590463872,
      "loss": 0.9232,
      "step": 39381
    },
    {
      "epoch": 3.388281854942786,
      "grad_norm": 2.515625,
      "learning_rate": 0.0002353869819706182,
      "loss": 0.8711,
      "step": 39382
    },
    {
      "epoch": 3.3883678912501076,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00023536403881053948,
      "loss": 1.0907,
      "step": 39383
    },
    {
      "epoch": 3.3884539275574292,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00023534109642446932,
      "loss": 0.7579,
      "step": 39384
    },
    {
      "epoch": 3.388539963864751,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00023531815481247492,
      "loss": 1.3012,
      "step": 39385
    },
    {
      "epoch": 3.3886260001720725,
      "grad_norm": 12.1875,
      "learning_rate": 0.00023529521397462367,
      "loss": 1.2398,
      "step": 39386
    },
    {
      "epoch": 3.388712036479394,
      "grad_norm": 1.0625,
      "learning_rate": 0.00023527227391098227,
      "loss": 0.8481,
      "step": 39387
    },
    {
      "epoch": 3.388798072786716,
      "grad_norm": 340.0,
      "learning_rate": 0.00023524933462161797,
      "loss": 1.4764,
      "step": 39388
    },
    {
      "epoch": 3.388884109094038,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00023522639610659802,
      "loss": 0.6579,
      "step": 39389
    },
    {
      "epoch": 3.3889701454013594,
      "grad_norm": 2.515625,
      "learning_rate": 0.00023520345836598923,
      "loss": 1.2415,
      "step": 39390
    },
    {
      "epoch": 3.389056181708681,
      "grad_norm": 2.265625,
      "learning_rate": 0.00023518052139985873,
      "loss": 1.3155,
      "step": 39391
    },
    {
      "epoch": 3.3891422180160027,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00023515758520827396,
      "loss": 0.9626,
      "step": 39392
    },
    {
      "epoch": 3.3892282543233243,
      "grad_norm": 1.59375,
      "learning_rate": 0.00023513464979130156,
      "loss": 0.9663,
      "step": 39393
    },
    {
      "epoch": 3.3893142906306464,
      "grad_norm": 1.0,
      "learning_rate": 0.0002351117151490088,
      "loss": 1.5955,
      "step": 39394
    },
    {
      "epoch": 3.389400326937968,
      "grad_norm": 664.0,
      "learning_rate": 0.00023508878128146294,
      "loss": 1.4246,
      "step": 39395
    },
    {
      "epoch": 3.3894863632452896,
      "grad_norm": 1.046875,
      "learning_rate": 0.00023506584818873066,
      "loss": 0.9842,
      "step": 39396
    },
    {
      "epoch": 3.3895723995526112,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00023504291587087944,
      "loss": 0.897,
      "step": 39397
    },
    {
      "epoch": 3.389658435859933,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000235019984327976,
      "loss": 1.1238,
      "step": 39398
    },
    {
      "epoch": 3.3897444721672545,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00023499705356008738,
      "loss": 1.69,
      "step": 39399
    },
    {
      "epoch": 3.389830508474576,
      "grad_norm": 14.875,
      "learning_rate": 0.0002349741235672812,
      "loss": 1.9499,
      "step": 39400
    },
    {
      "epoch": 3.389916544781898,
      "grad_norm": 1.25,
      "learning_rate": 0.00023495119434962397,
      "loss": 1.1991,
      "step": 39401
    },
    {
      "epoch": 3.39000258108922,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00023492826590718312,
      "loss": 0.5632,
      "step": 39402
    },
    {
      "epoch": 3.3900886173965414,
      "grad_norm": 3.53125,
      "learning_rate": 0.0002349053382400254,
      "loss": 1.7549,
      "step": 39403
    },
    {
      "epoch": 3.390174653703863,
      "grad_norm": 1.734375,
      "learning_rate": 0.00023488241134821797,
      "loss": 1.0637,
      "step": 39404
    },
    {
      "epoch": 3.3902606900111847,
      "grad_norm": 24.875,
      "learning_rate": 0.0002348594852318281,
      "loss": 1.2819,
      "step": 39405
    },
    {
      "epoch": 3.3903467263185063,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002348365598909225,
      "loss": 0.838,
      "step": 39406
    },
    {
      "epoch": 3.390432762625828,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00023481363532556844,
      "loss": 0.9268,
      "step": 39407
    },
    {
      "epoch": 3.39051879893315,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00023479071153583304,
      "loss": 0.8953,
      "step": 39408
    },
    {
      "epoch": 3.3906048352404716,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00023476778852178304,
      "loss": 1.6283,
      "step": 39409
    },
    {
      "epoch": 3.3906908715477933,
      "grad_norm": 1.859375,
      "learning_rate": 0.00023474486628348573,
      "loss": 1.6501,
      "step": 39410
    },
    {
      "epoch": 3.390776907855115,
      "grad_norm": 1.171875,
      "learning_rate": 0.00023472194482100823,
      "loss": 0.9497,
      "step": 39411
    },
    {
      "epoch": 3.3908629441624365,
      "grad_norm": 15808.0,
      "learning_rate": 0.00023469902413441728,
      "loss": 1.0047,
      "step": 39412
    },
    {
      "epoch": 3.390948980469758,
      "grad_norm": 2.03125,
      "learning_rate": 0.00023467610422378028,
      "loss": 1.3878,
      "step": 39413
    },
    {
      "epoch": 3.3910350167770797,
      "grad_norm": 1.734375,
      "learning_rate": 0.0002346531850891639,
      "loss": 1.6966,
      "step": 39414
    },
    {
      "epoch": 3.391121053084402,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00023463026673063532,
      "loss": 1.2688,
      "step": 39415
    },
    {
      "epoch": 3.3912070893917234,
      "grad_norm": 4.03125,
      "learning_rate": 0.00023460734914826177,
      "loss": 1.2713,
      "step": 39416
    },
    {
      "epoch": 3.391293125699045,
      "grad_norm": 1.125,
      "learning_rate": 0.00023458443234210998,
      "loss": 1.055,
      "step": 39417
    },
    {
      "epoch": 3.3913791620063667,
      "grad_norm": 3.734375,
      "learning_rate": 0.00023456151631224705,
      "loss": 1.2233,
      "step": 39418
    },
    {
      "epoch": 3.3914651983136883,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00023453860105874025,
      "loss": 0.9506,
      "step": 39419
    },
    {
      "epoch": 3.39155123462101,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00023451568658165622,
      "loss": 1.4888,
      "step": 39420
    },
    {
      "epoch": 3.3916372709283316,
      "grad_norm": 932.0,
      "learning_rate": 0.0002344927728810622,
      "loss": 0.5736,
      "step": 39421
    },
    {
      "epoch": 3.3917233072356536,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00023446985995702536,
      "loss": 1.3166,
      "step": 39422
    },
    {
      "epoch": 3.3918093435429753,
      "grad_norm": 1.125,
      "learning_rate": 0.00023444694780961234,
      "loss": 1.2965,
      "step": 39423
    },
    {
      "epoch": 3.391895379850297,
      "grad_norm": 78.5,
      "learning_rate": 0.00023442403643889037,
      "loss": 1.0184,
      "step": 39424
    },
    {
      "epoch": 3.3919814161576185,
      "grad_norm": 38.25,
      "learning_rate": 0.00023440112584492662,
      "loss": 1.2231,
      "step": 39425
    },
    {
      "epoch": 3.39206745246494,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00023437821602778774,
      "loss": 1.366,
      "step": 39426
    },
    {
      "epoch": 3.3921534887722617,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002343553069875411,
      "loss": 0.8265,
      "step": 39427
    },
    {
      "epoch": 3.3922395250795834,
      "grad_norm": 2.078125,
      "learning_rate": 0.0002343323987242533,
      "loss": 1.0747,
      "step": 39428
    },
    {
      "epoch": 3.3923255613869054,
      "grad_norm": 0.875,
      "learning_rate": 0.00023430949123799162,
      "loss": 0.6255,
      "step": 39429
    },
    {
      "epoch": 3.392411597694227,
      "grad_norm": 1.140625,
      "learning_rate": 0.00023428658452882318,
      "loss": 1.6092,
      "step": 39430
    },
    {
      "epoch": 3.3924976340015487,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00023426367859681458,
      "loss": 1.1882,
      "step": 39431
    },
    {
      "epoch": 3.3925836703088703,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00023424077344203308,
      "loss": 0.9284,
      "step": 39432
    },
    {
      "epoch": 3.392669706616192,
      "grad_norm": 1.0,
      "learning_rate": 0.00023421786906454578,
      "loss": 1.6096,
      "step": 39433
    },
    {
      "epoch": 3.3927557429235136,
      "grad_norm": 1.296875,
      "learning_rate": 0.00023419496546441936,
      "loss": 0.8602,
      "step": 39434
    },
    {
      "epoch": 3.392841779230835,
      "grad_norm": 1.28125,
      "learning_rate": 0.00023417206264172096,
      "loss": 1.4381,
      "step": 39435
    },
    {
      "epoch": 3.3929278155381573,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002341491605965176,
      "loss": 0.7553,
      "step": 39436
    },
    {
      "epoch": 3.393013851845479,
      "grad_norm": 1.1875,
      "learning_rate": 0.00023412625932887622,
      "loss": 1.2353,
      "step": 39437
    },
    {
      "epoch": 3.3930998881528005,
      "grad_norm": 61.0,
      "learning_rate": 0.00023410335883886397,
      "loss": 0.7993,
      "step": 39438
    },
    {
      "epoch": 3.393185924460122,
      "grad_norm": 4.625,
      "learning_rate": 0.00023408045912654753,
      "loss": 1.1303,
      "step": 39439
    },
    {
      "epoch": 3.3932719607674438,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00023405756019199403,
      "loss": 1.041,
      "step": 39440
    },
    {
      "epoch": 3.3933579970747654,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00023403466203527064,
      "loss": 0.544,
      "step": 39441
    },
    {
      "epoch": 3.393444033382087,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002340117646564439,
      "loss": 1.1986,
      "step": 39442
    },
    {
      "epoch": 3.393530069689409,
      "grad_norm": 107.5,
      "learning_rate": 0.00023398886805558127,
      "loss": 1.8898,
      "step": 39443
    },
    {
      "epoch": 3.3936161059967307,
      "grad_norm": 1.1875,
      "learning_rate": 0.00023396597223274924,
      "loss": 1.6414,
      "step": 39444
    },
    {
      "epoch": 3.3937021423040523,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00023394307718801506,
      "loss": 1.2887,
      "step": 39445
    },
    {
      "epoch": 3.393788178611374,
      "grad_norm": 24.75,
      "learning_rate": 0.00023392018292144574,
      "loss": 1.6916,
      "step": 39446
    },
    {
      "epoch": 3.3938742149186956,
      "grad_norm": 52.75,
      "learning_rate": 0.00023389728943310801,
      "loss": 0.8784,
      "step": 39447
    },
    {
      "epoch": 3.393960251226017,
      "grad_norm": 120.0,
      "learning_rate": 0.00023387439672306888,
      "loss": 0.4017,
      "step": 39448
    },
    {
      "epoch": 3.3940462875333393,
      "grad_norm": 1.484375,
      "learning_rate": 0.00023385150479139562,
      "loss": 0.9341,
      "step": 39449
    },
    {
      "epoch": 3.394132323840661,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00023382861363815477,
      "loss": 1.0555,
      "step": 39450
    },
    {
      "epoch": 3.3942183601479825,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00023380572326341342,
      "loss": 1.0507,
      "step": 39451
    },
    {
      "epoch": 3.394304396455304,
      "grad_norm": 1.390625,
      "learning_rate": 0.0002337828336672388,
      "loss": 1.825,
      "step": 39452
    },
    {
      "epoch": 3.3943904327626258,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00023375994484969737,
      "loss": 0.8713,
      "step": 39453
    },
    {
      "epoch": 3.3944764690699474,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002337370568108565,
      "loss": 1.1152,
      "step": 39454
    },
    {
      "epoch": 3.3945625053772694,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00023371416955078278,
      "loss": 1.1431,
      "step": 39455
    },
    {
      "epoch": 3.394648541684591,
      "grad_norm": 159.0,
      "learning_rate": 0.00023369128306954335,
      "loss": 1.2741,
      "step": 39456
    },
    {
      "epoch": 3.3947345779919127,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002336683973672053,
      "loss": 1.4114,
      "step": 39457
    },
    {
      "epoch": 3.3948206142992343,
      "grad_norm": 1.1875,
      "learning_rate": 0.00023364551244383519,
      "loss": 1.2874,
      "step": 39458
    },
    {
      "epoch": 3.394906650606556,
      "grad_norm": 1.65625,
      "learning_rate": 0.00023362262829950022,
      "loss": 0.8186,
      "step": 39459
    },
    {
      "epoch": 3.3949926869138776,
      "grad_norm": 1.625,
      "learning_rate": 0.0002335997449342674,
      "loss": 0.9214,
      "step": 39460
    },
    {
      "epoch": 3.395078723221199,
      "grad_norm": 5.5,
      "learning_rate": 0.0002335768623482033,
      "loss": 0.8837,
      "step": 39461
    },
    {
      "epoch": 3.3951647595285213,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00023355398054137504,
      "loss": 2.1559,
      "step": 39462
    },
    {
      "epoch": 3.395250795835843,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002335310995138498,
      "loss": 1.5673,
      "step": 39463
    },
    {
      "epoch": 3.3953368321431645,
      "grad_norm": 4.125,
      "learning_rate": 0.00023350821926569404,
      "loss": 1.2682,
      "step": 39464
    },
    {
      "epoch": 3.395422868450486,
      "grad_norm": 1.0,
      "learning_rate": 0.00023348533979697495,
      "loss": 0.9447,
      "step": 39465
    },
    {
      "epoch": 3.3955089047578078,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0002334624611077596,
      "loss": 1.216,
      "step": 39466
    },
    {
      "epoch": 3.3955949410651294,
      "grad_norm": 2.703125,
      "learning_rate": 0.00023343958319811447,
      "loss": 1.564,
      "step": 39467
    },
    {
      "epoch": 3.395680977372451,
      "grad_norm": 1.140625,
      "learning_rate": 0.000233416706068107,
      "loss": 1.8082,
      "step": 39468
    },
    {
      "epoch": 3.395767013679773,
      "grad_norm": 1.171875,
      "learning_rate": 0.00023339382971780353,
      "loss": 0.7067,
      "step": 39469
    },
    {
      "epoch": 3.3958530499870947,
      "grad_norm": 10.875,
      "learning_rate": 0.00023337095414727133,
      "loss": 1.2733,
      "step": 39470
    },
    {
      "epoch": 3.3959390862944163,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00023334807935657738,
      "loss": 1.7324,
      "step": 39471
    },
    {
      "epoch": 3.396025122601738,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00023332520534578822,
      "loss": 1.2148,
      "step": 39472
    },
    {
      "epoch": 3.3961111589090596,
      "grad_norm": 1.546875,
      "learning_rate": 0.00023330233211497103,
      "loss": 1.5562,
      "step": 39473
    },
    {
      "epoch": 3.396197195216381,
      "grad_norm": 2.625,
      "learning_rate": 0.00023327945966419261,
      "loss": 0.6128,
      "step": 39474
    },
    {
      "epoch": 3.396283231523703,
      "grad_norm": 1.515625,
      "learning_rate": 0.00023325658799351989,
      "loss": 0.9487,
      "step": 39475
    },
    {
      "epoch": 3.396369267831025,
      "grad_norm": 6336.0,
      "learning_rate": 0.00023323371710301999,
      "loss": 1.7822,
      "step": 39476
    },
    {
      "epoch": 3.3964553041383465,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002332108469927594,
      "loss": 0.5648,
      "step": 39477
    },
    {
      "epoch": 3.396541340445668,
      "grad_norm": 1752.0,
      "learning_rate": 0.0002331879776628052,
      "loss": 1.0391,
      "step": 39478
    },
    {
      "epoch": 3.3966273767529898,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00023316510911322442,
      "loss": 1.626,
      "step": 39479
    },
    {
      "epoch": 3.3967134130603114,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00023314224134408362,
      "loss": 0.7508,
      "step": 39480
    },
    {
      "epoch": 3.396799449367633,
      "grad_norm": 15.0625,
      "learning_rate": 0.00023311937435544988,
      "loss": 1.6841,
      "step": 39481
    },
    {
      "epoch": 3.3968854856749546,
      "grad_norm": 4.25,
      "learning_rate": 0.00023309650814739026,
      "loss": 1.7028,
      "step": 39482
    },
    {
      "epoch": 3.3969715219822767,
      "grad_norm": 2.03125,
      "learning_rate": 0.00023307364271997122,
      "loss": 2.2776,
      "step": 39483
    },
    {
      "epoch": 3.3970575582895983,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002330507780732601,
      "loss": 0.9558,
      "step": 39484
    },
    {
      "epoch": 3.39714359459692,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00023302791420732333,
      "loss": 1.1138,
      "step": 39485
    },
    {
      "epoch": 3.3972296309042416,
      "grad_norm": 3.4375,
      "learning_rate": 0.00023300505112222797,
      "loss": 0.958,
      "step": 39486
    },
    {
      "epoch": 3.397315667211563,
      "grad_norm": 2.203125,
      "learning_rate": 0.00023298218881804113,
      "loss": 1.256,
      "step": 39487
    },
    {
      "epoch": 3.397401703518885,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002329593272948292,
      "loss": 1.829,
      "step": 39488
    },
    {
      "epoch": 3.3974877398262064,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00023293646655265937,
      "loss": 0.6842,
      "step": 39489
    },
    {
      "epoch": 3.3975737761335285,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002329136065915986,
      "loss": 1.6478,
      "step": 39490
    },
    {
      "epoch": 3.39765981244085,
      "grad_norm": 1.03125,
      "learning_rate": 0.00023289074741171333,
      "loss": 0.4649,
      "step": 39491
    },
    {
      "epoch": 3.3977458487481718,
      "grad_norm": 7.75,
      "learning_rate": 0.00023286788901307078,
      "loss": 1.4492,
      "step": 39492
    },
    {
      "epoch": 3.3978318850554934,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00023284503139573781,
      "loss": 1.1125,
      "step": 39493
    },
    {
      "epoch": 3.397917921362815,
      "grad_norm": 1.671875,
      "learning_rate": 0.00023282217455978106,
      "loss": 2.0072,
      "step": 39494
    },
    {
      "epoch": 3.3980039576701366,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00023279931850526758,
      "loss": 1.5226,
      "step": 39495
    },
    {
      "epoch": 3.3980899939774583,
      "grad_norm": 1.828125,
      "learning_rate": 0.00023277646323226404,
      "loss": 0.9155,
      "step": 39496
    },
    {
      "epoch": 3.3981760302847803,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002327536087408373,
      "loss": 1.2312,
      "step": 39497
    },
    {
      "epoch": 3.398262066592102,
      "grad_norm": 1.125,
      "learning_rate": 0.00023273075503105445,
      "loss": 0.8958,
      "step": 39498
    },
    {
      "epoch": 3.3983481028994236,
      "grad_norm": 0.796875,
      "learning_rate": 0.00023270790210298204,
      "loss": 0.3883,
      "step": 39499
    },
    {
      "epoch": 3.398434139206745,
      "grad_norm": 1.125,
      "learning_rate": 0.00023268504995668698,
      "loss": 1.2224,
      "step": 39500
    },
    {
      "epoch": 3.398520175514067,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00023266219859223637,
      "loss": 1.1399,
      "step": 39501
    },
    {
      "epoch": 3.3986062118213884,
      "grad_norm": 1.25,
      "learning_rate": 0.00023263934800969661,
      "loss": 0.9865,
      "step": 39502
    },
    {
      "epoch": 3.39869224812871,
      "grad_norm": 1.296875,
      "learning_rate": 0.00023261649820913483,
      "loss": 1.4381,
      "step": 39503
    },
    {
      "epoch": 3.398778284436032,
      "grad_norm": 1.15625,
      "learning_rate": 0.00023259364919061792,
      "loss": 1.243,
      "step": 39504
    },
    {
      "epoch": 3.3988643207433538,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002325708009542124,
      "loss": 1.9992,
      "step": 39505
    },
    {
      "epoch": 3.3989503570506754,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00023254795349998526,
      "loss": 1.902,
      "step": 39506
    },
    {
      "epoch": 3.399036393357997,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00023252510682800353,
      "loss": 0.9563,
      "step": 39507
    },
    {
      "epoch": 3.3991224296653186,
      "grad_norm": 2.71875,
      "learning_rate": 0.0002325022609383337,
      "loss": 0.8401,
      "step": 39508
    },
    {
      "epoch": 3.3992084659726407,
      "grad_norm": 1.09375,
      "learning_rate": 0.00023247941583104287,
      "loss": 1.2572,
      "step": 39509
    },
    {
      "epoch": 3.3992945022799623,
      "grad_norm": 1.390625,
      "learning_rate": 0.00023245657150619737,
      "loss": 1.5215,
      "step": 39510
    },
    {
      "epoch": 3.399380538587284,
      "grad_norm": 29.875,
      "learning_rate": 0.00023243372796386458,
      "loss": 0.7983,
      "step": 39511
    },
    {
      "epoch": 3.3994665748946056,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00023241088520411124,
      "loss": 1.8985,
      "step": 39512
    },
    {
      "epoch": 3.399552611201927,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00023238804322700386,
      "loss": 0.8078,
      "step": 39513
    },
    {
      "epoch": 3.399638647509249,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00023236520203260958,
      "loss": 1.9162,
      "step": 39514
    },
    {
      "epoch": 3.3997246838165704,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002323423616209948,
      "loss": 1.8018,
      "step": 39515
    },
    {
      "epoch": 3.3998107201238925,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002323195219922266,
      "loss": 1.1976,
      "step": 39516
    },
    {
      "epoch": 3.399896756431214,
      "grad_norm": 54.0,
      "learning_rate": 0.00023229668314637194,
      "loss": 1.1329,
      "step": 39517
    },
    {
      "epoch": 3.3999827927385358,
      "grad_norm": 12.3125,
      "learning_rate": 0.0002322738450834972,
      "loss": 1.4211,
      "step": 39518
    },
    {
      "epoch": 3.4000688290458574,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00023225100780366936,
      "loss": 0.7825,
      "step": 39519
    },
    {
      "epoch": 3.400154865353179,
      "grad_norm": 1.53125,
      "learning_rate": 0.0002322281713069555,
      "loss": 0.9672,
      "step": 39520
    },
    {
      "epoch": 3.4002409016605006,
      "grad_norm": 13.3125,
      "learning_rate": 0.00023220533559342195,
      "loss": 1.3906,
      "step": 39521
    },
    {
      "epoch": 3.4003269379678223,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00023218250066313573,
      "loss": 0.8362,
      "step": 39522
    },
    {
      "epoch": 3.4004129742751443,
      "grad_norm": 1.03125,
      "learning_rate": 0.00023215966651616378,
      "loss": 1.6086,
      "step": 39523
    },
    {
      "epoch": 3.400499010582466,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00023213683315257251,
      "loss": 0.7912,
      "step": 39524
    },
    {
      "epoch": 3.4005850468897876,
      "grad_norm": 38.5,
      "learning_rate": 0.0002321140005724291,
      "loss": 0.9023,
      "step": 39525
    },
    {
      "epoch": 3.400671083197109,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00023209116877579995,
      "loss": 1.7066,
      "step": 39526
    },
    {
      "epoch": 3.400757119504431,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00023206833776275205,
      "loss": 0.9568,
      "step": 39527
    },
    {
      "epoch": 3.4008431558117524,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002320455075333523,
      "loss": 0.9335,
      "step": 39528
    },
    {
      "epoch": 3.400929192119074,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00023202267808766715,
      "loss": 1.8526,
      "step": 39529
    },
    {
      "epoch": 3.401015228426396,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00023199984942576357,
      "loss": 0.8723,
      "step": 39530
    },
    {
      "epoch": 3.4011012647337178,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0002319770215477085,
      "loss": 1.2713,
      "step": 39531
    },
    {
      "epoch": 3.4011873010410394,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002319541944535683,
      "loss": 0.7279,
      "step": 39532
    },
    {
      "epoch": 3.401273337348361,
      "grad_norm": 1.71875,
      "learning_rate": 0.00023193136814340993,
      "loss": 1.4076,
      "step": 39533
    },
    {
      "epoch": 3.4013593736556826,
      "grad_norm": 18.5,
      "learning_rate": 0.00023190854261730037,
      "loss": 0.4221,
      "step": 39534
    },
    {
      "epoch": 3.4014454099630043,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00023188571787530605,
      "loss": 1.2721,
      "step": 39535
    },
    {
      "epoch": 3.401531446270326,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00023186289391749398,
      "loss": 0.5338,
      "step": 39536
    },
    {
      "epoch": 3.401617482577648,
      "grad_norm": 1.6875,
      "learning_rate": 0.00023184007074393065,
      "loss": 1.3926,
      "step": 39537
    },
    {
      "epoch": 3.4017035188849696,
      "grad_norm": 36.0,
      "learning_rate": 0.00023181724835468292,
      "loss": 1.1238,
      "step": 39538
    },
    {
      "epoch": 3.401789555192291,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002317944267498178,
      "loss": 1.0794,
      "step": 39539
    },
    {
      "epoch": 3.401875591499613,
      "grad_norm": 1.1875,
      "learning_rate": 0.00023177160592940165,
      "loss": 1.3975,
      "step": 39540
    },
    {
      "epoch": 3.4019616278069345,
      "grad_norm": 2.046875,
      "learning_rate": 0.00023174878589350134,
      "loss": 1.1955,
      "step": 39541
    },
    {
      "epoch": 3.402047664114256,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00023172596664218387,
      "loss": 0.9326,
      "step": 39542
    },
    {
      "epoch": 3.4021337004215777,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00023170314817551557,
      "loss": 1.6731,
      "step": 39543
    },
    {
      "epoch": 3.4022197367288998,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002316803304935634,
      "loss": 1.5112,
      "step": 39544
    },
    {
      "epoch": 3.4023057730362214,
      "grad_norm": 1.046875,
      "learning_rate": 0.00023165751359639425,
      "loss": 1.4447,
      "step": 39545
    },
    {
      "epoch": 3.402391809343543,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002316346974840745,
      "loss": 1.0129,
      "step": 39546
    },
    {
      "epoch": 3.4024778456508646,
      "grad_norm": 48.75,
      "learning_rate": 0.00023161188215667106,
      "loss": 0.6496,
      "step": 39547
    },
    {
      "epoch": 3.4025638819581863,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002315890676142507,
      "loss": 1.6285,
      "step": 39548
    },
    {
      "epoch": 3.402649918265508,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00023156625385688007,
      "loss": 0.4596,
      "step": 39549
    },
    {
      "epoch": 3.4027359545728295,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00023154344088462613,
      "loss": 1.0327,
      "step": 39550
    },
    {
      "epoch": 3.4028219908801516,
      "grad_norm": 1.1875,
      "learning_rate": 0.00023152062869755526,
      "loss": 1.7841,
      "step": 39551
    },
    {
      "epoch": 3.402908027187473,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00023149781729573431,
      "loss": 0.5628,
      "step": 39552
    },
    {
      "epoch": 3.402994063494795,
      "grad_norm": 0.703125,
      "learning_rate": 0.00023147500667923022,
      "loss": 0.4687,
      "step": 39553
    },
    {
      "epoch": 3.4030800998021165,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002314521968481093,
      "loss": 0.6331,
      "step": 39554
    },
    {
      "epoch": 3.403166136109438,
      "grad_norm": 1.84375,
      "learning_rate": 0.00023142938780243872,
      "loss": 1.5232,
      "step": 39555
    },
    {
      "epoch": 3.4032521724167597,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002314065795422847,
      "loss": 1.6936,
      "step": 39556
    },
    {
      "epoch": 3.4033382087240813,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00023138377206771422,
      "loss": 0.7358,
      "step": 39557
    },
    {
      "epoch": 3.4034242450314034,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00023136096537879414,
      "loss": 0.7886,
      "step": 39558
    },
    {
      "epoch": 3.403510281338725,
      "grad_norm": 1.015625,
      "learning_rate": 0.00023133815947559084,
      "loss": 0.6662,
      "step": 39559
    },
    {
      "epoch": 3.4035963176460466,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00023131535435817115,
      "loss": 0.8721,
      "step": 39560
    },
    {
      "epoch": 3.4036823539533683,
      "grad_norm": 29.625,
      "learning_rate": 0.00023129255002660192,
      "loss": 1.2217,
      "step": 39561
    },
    {
      "epoch": 3.40376839026069,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00023126974648094963,
      "loss": 0.9856,
      "step": 39562
    },
    {
      "epoch": 3.4038544265680115,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00023124694372128103,
      "loss": 1.1317,
      "step": 39563
    },
    {
      "epoch": 3.4039404628753336,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00023122414174766299,
      "loss": 1.0964,
      "step": 39564
    },
    {
      "epoch": 3.404026499182655,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002312013405601619,
      "loss": 1.2228,
      "step": 39565
    },
    {
      "epoch": 3.404112535489977,
      "grad_norm": 31.875,
      "learning_rate": 0.0002311785401588448,
      "loss": 1.6491,
      "step": 39566
    },
    {
      "epoch": 3.4041985717972985,
      "grad_norm": 20.75,
      "learning_rate": 0.00023115574054377802,
      "loss": 0.9369,
      "step": 39567
    },
    {
      "epoch": 3.40428460810462,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002311329417150284,
      "loss": 0.9797,
      "step": 39568
    },
    {
      "epoch": 3.4043706444119417,
      "grad_norm": 0.859375,
      "learning_rate": 0.00023111014367266285,
      "loss": 0.7859,
      "step": 39569
    },
    {
      "epoch": 3.4044566807192638,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002310873464167476,
      "loss": 1.19,
      "step": 39570
    },
    {
      "epoch": 3.4045427170265854,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002310645499473496,
      "loss": 0.6397,
      "step": 39571
    },
    {
      "epoch": 3.404628753333907,
      "grad_norm": 1.046875,
      "learning_rate": 0.00023104175426453568,
      "loss": 1.1082,
      "step": 39572
    },
    {
      "epoch": 3.4047147896412286,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002310189593683721,
      "loss": 1.5208,
      "step": 39573
    },
    {
      "epoch": 3.4048008259485503,
      "grad_norm": 18.75,
      "learning_rate": 0.00023099616525892576,
      "loss": 0.9633,
      "step": 39574
    },
    {
      "epoch": 3.404886862255872,
      "grad_norm": 1.328125,
      "learning_rate": 0.00023097337193626343,
      "loss": 1.4873,
      "step": 39575
    },
    {
      "epoch": 3.4049728985631935,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002309505794004516,
      "loss": 0.5414,
      "step": 39576
    },
    {
      "epoch": 3.4050589348705156,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00023092778765155693,
      "loss": 1.2688,
      "step": 39577
    },
    {
      "epoch": 3.405144971177837,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00023090499668964638,
      "loss": 1.2762,
      "step": 39578
    },
    {
      "epoch": 3.405231007485159,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00023088220651478614,
      "loss": 1.2891,
      "step": 39579
    },
    {
      "epoch": 3.4053170437924805,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00023085941712704333,
      "loss": 1.1581,
      "step": 39580
    },
    {
      "epoch": 3.405403080099802,
      "grad_norm": 1.421875,
      "learning_rate": 0.0002308366285264842,
      "loss": 1.7488,
      "step": 39581
    },
    {
      "epoch": 3.4054891164071237,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00023081384071317552,
      "loss": 1.397,
      "step": 39582
    },
    {
      "epoch": 3.4055751527144453,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00023079105368718422,
      "loss": 1.4772,
      "step": 39583
    },
    {
      "epoch": 3.4056611890217674,
      "grad_norm": 1.0,
      "learning_rate": 0.00023076826744857637,
      "loss": 0.6749,
      "step": 39584
    },
    {
      "epoch": 3.405747225329089,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00023074548199741935,
      "loss": 1.1009,
      "step": 39585
    },
    {
      "epoch": 3.4058332616364106,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00023072269733377925,
      "loss": 1.5295,
      "step": 39586
    },
    {
      "epoch": 3.4059192979437323,
      "grad_norm": 14.0625,
      "learning_rate": 0.00023069991345772285,
      "loss": 2.0418,
      "step": 39587
    },
    {
      "epoch": 3.406005334251054,
      "grad_norm": 1.046875,
      "learning_rate": 0.00023067713036931703,
      "loss": 1.1581,
      "step": 39588
    },
    {
      "epoch": 3.4060913705583755,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00023065434806862807,
      "loss": 1.0106,
      "step": 39589
    },
    {
      "epoch": 3.406177406865697,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00023063156655572271,
      "loss": 1.1056,
      "step": 39590
    },
    {
      "epoch": 3.406263443173019,
      "grad_norm": 1.03125,
      "learning_rate": 0.00023060878583066786,
      "loss": 0.5592,
      "step": 39591
    },
    {
      "epoch": 3.406349479480341,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002305860058935297,
      "loss": 0.9717,
      "step": 39592
    },
    {
      "epoch": 3.4064355157876625,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00023056322674437507,
      "loss": 0.5254,
      "step": 39593
    },
    {
      "epoch": 3.406521552094984,
      "grad_norm": 1.546875,
      "learning_rate": 0.0002305404483832708,
      "loss": 0.9728,
      "step": 39594
    },
    {
      "epoch": 3.4066075884023057,
      "grad_norm": 2.5,
      "learning_rate": 0.00023051767081028312,
      "loss": 0.4954,
      "step": 39595
    },
    {
      "epoch": 3.4066936247096273,
      "grad_norm": 1.75,
      "learning_rate": 0.00023049489402547906,
      "loss": 1.471,
      "step": 39596
    },
    {
      "epoch": 3.406779661016949,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002304721180289248,
      "loss": 1.1943,
      "step": 39597
    },
    {
      "epoch": 3.406865697324271,
      "grad_norm": 1.015625,
      "learning_rate": 0.00023044934282068718,
      "loss": 1.2119,
      "step": 39598
    },
    {
      "epoch": 3.4069517336315926,
      "grad_norm": 1.140625,
      "learning_rate": 0.00023042656840083299,
      "loss": 0.6734,
      "step": 39599
    },
    {
      "epoch": 3.4070377699389143,
      "grad_norm": 4.5625,
      "learning_rate": 0.00023040379476942847,
      "loss": 0.9399,
      "step": 39600
    },
    {
      "epoch": 3.407123806246236,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00023038102192654043,
      "loss": 0.5938,
      "step": 39601
    },
    {
      "epoch": 3.4072098425535575,
      "grad_norm": 3.515625,
      "learning_rate": 0.00023035824987223565,
      "loss": 1.357,
      "step": 39602
    },
    {
      "epoch": 3.407295878860879,
      "grad_norm": 0.984375,
      "learning_rate": 0.00023033547860658033,
      "loss": 1.1151,
      "step": 39603
    },
    {
      "epoch": 3.4073819151682008,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00023031270812964134,
      "loss": 0.6882,
      "step": 39604
    },
    {
      "epoch": 3.407467951475523,
      "grad_norm": 1.09375,
      "learning_rate": 0.00023028993844148539,
      "loss": 1.765,
      "step": 39605
    },
    {
      "epoch": 3.4075539877828445,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002302671695421787,
      "loss": 1.3553,
      "step": 39606
    },
    {
      "epoch": 3.407640024090166,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00023024440143178826,
      "loss": 1.2966,
      "step": 39607
    },
    {
      "epoch": 3.4077260603974877,
      "grad_norm": 1.15625,
      "learning_rate": 0.00023022163411038032,
      "loss": 1.3858,
      "step": 39608
    },
    {
      "epoch": 3.4078120967048093,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00023019886757802167,
      "loss": 1.1425,
      "step": 39609
    },
    {
      "epoch": 3.407898133012131,
      "grad_norm": 9.3125,
      "learning_rate": 0.00023017610183477899,
      "loss": 1.239,
      "step": 39610
    },
    {
      "epoch": 3.4079841693194526,
      "grad_norm": 2.46875,
      "learning_rate": 0.0002301533368807186,
      "loss": 1.1423,
      "step": 39611
    },
    {
      "epoch": 3.4080702056267747,
      "grad_norm": 1.34375,
      "learning_rate": 0.00023013057271590714,
      "loss": 1.7274,
      "step": 39612
    },
    {
      "epoch": 3.4081562419340963,
      "grad_norm": 2.234375,
      "learning_rate": 0.00023010780934041154,
      "loss": 1.7668,
      "step": 39613
    },
    {
      "epoch": 3.408242278241418,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00023008504675429788,
      "loss": 0.5686,
      "step": 39614
    },
    {
      "epoch": 3.4083283145487395,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00023006228495763294,
      "loss": 1.2608,
      "step": 39615
    },
    {
      "epoch": 3.408414350856061,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002300395239504835,
      "loss": 1.2396,
      "step": 39616
    },
    {
      "epoch": 3.4085003871633828,
      "grad_norm": 4.71875,
      "learning_rate": 0.0002300167637329158,
      "loss": 1.3544,
      "step": 39617
    },
    {
      "epoch": 3.4085864234707044,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00022999400430499657,
      "loss": 1.2746,
      "step": 39618
    },
    {
      "epoch": 3.4086724597780265,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0002299712456667925,
      "loss": 1.2154,
      "step": 39619
    },
    {
      "epoch": 3.408758496085348,
      "grad_norm": 2.390625,
      "learning_rate": 0.00022994848781836985,
      "loss": 0.8371,
      "step": 39620
    },
    {
      "epoch": 3.4088445323926697,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00022992573075979539,
      "loss": 0.9126,
      "step": 39621
    },
    {
      "epoch": 3.4089305686999913,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00022990297449113555,
      "loss": 0.9433,
      "step": 39622
    },
    {
      "epoch": 3.409016605007313,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00022988021901245708,
      "loss": 0.3084,
      "step": 39623
    },
    {
      "epoch": 3.4091026413146346,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002298574643238266,
      "loss": 0.923,
      "step": 39624
    },
    {
      "epoch": 3.4091886776219567,
      "grad_norm": 378.0,
      "learning_rate": 0.00022983471042531024,
      "loss": 1.2622,
      "step": 39625
    },
    {
      "epoch": 3.4092747139292783,
      "grad_norm": 6.875,
      "learning_rate": 0.00022981195731697508,
      "loss": 1.628,
      "step": 39626
    },
    {
      "epoch": 3.4093607502366,
      "grad_norm": 0.84375,
      "learning_rate": 0.00022978920499888716,
      "loss": 1.5272,
      "step": 39627
    },
    {
      "epoch": 3.4094467865439215,
      "grad_norm": 1.5,
      "learning_rate": 0.00022976645347111323,
      "loss": 1.6783,
      "step": 39628
    },
    {
      "epoch": 3.409532822851243,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00022974370273372015,
      "loss": 1.1805,
      "step": 39629
    },
    {
      "epoch": 3.4096188591585648,
      "grad_norm": 1.125,
      "learning_rate": 0.00022972095278677392,
      "loss": 1.2786,
      "step": 39630
    },
    {
      "epoch": 3.409704895465887,
      "grad_norm": 60.25,
      "learning_rate": 0.00022969820363034132,
      "loss": 1.4381,
      "step": 39631
    },
    {
      "epoch": 3.4097909317732085,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00022967545526448912,
      "loss": 1.1015,
      "step": 39632
    },
    {
      "epoch": 3.40987696808053,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00022965270768928348,
      "loss": 1.4295,
      "step": 39633
    },
    {
      "epoch": 3.4099630043878517,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00022962996090479106,
      "loss": 1.4407,
      "step": 39634
    },
    {
      "epoch": 3.4100490406951733,
      "grad_norm": 52.5,
      "learning_rate": 0.0002296072149110786,
      "loss": 0.9059,
      "step": 39635
    },
    {
      "epoch": 3.410135077002495,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00022958446970821228,
      "loss": 0.9513,
      "step": 39636
    },
    {
      "epoch": 3.4102211133098166,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000229561725296259,
      "loss": 0.8833,
      "step": 39637
    },
    {
      "epoch": 3.4103071496171387,
      "grad_norm": 0.890625,
      "learning_rate": 0.00022953898167528487,
      "loss": 0.6461,
      "step": 39638
    },
    {
      "epoch": 3.4103931859244603,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002295162388453566,
      "loss": 0.6306,
      "step": 39639
    },
    {
      "epoch": 3.410479222231782,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002294934968065409,
      "loss": 1.642,
      "step": 39640
    },
    {
      "epoch": 3.4105652585391035,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00022947075555890395,
      "loss": 1.1209,
      "step": 39641
    },
    {
      "epoch": 3.410651294846425,
      "grad_norm": 5.25,
      "learning_rate": 0.00022944801510251246,
      "loss": 0.5488,
      "step": 39642
    },
    {
      "epoch": 3.4107373311537468,
      "grad_norm": 0.921875,
      "learning_rate": 0.000229425275437433,
      "loss": 1.1164,
      "step": 39643
    },
    {
      "epoch": 3.4108233674610684,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002294025365637319,
      "loss": 1.1309,
      "step": 39644
    },
    {
      "epoch": 3.4109094037683905,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00022937979848147572,
      "loss": 1.2451,
      "step": 39645
    },
    {
      "epoch": 3.410995440075712,
      "grad_norm": 1.484375,
      "learning_rate": 0.00022935706119073113,
      "loss": 1.497,
      "step": 39646
    },
    {
      "epoch": 3.4110814763830337,
      "grad_norm": 0.96875,
      "learning_rate": 0.00022933432469156434,
      "loss": 0.8525,
      "step": 39647
    },
    {
      "epoch": 3.4111675126903553,
      "grad_norm": 1.140625,
      "learning_rate": 0.00022931158898404226,
      "loss": 0.7638,
      "step": 39648
    },
    {
      "epoch": 3.411253548997677,
      "grad_norm": 1.21875,
      "learning_rate": 0.00022928885406823087,
      "loss": 1.2323,
      "step": 39649
    },
    {
      "epoch": 3.4113395853049986,
      "grad_norm": 1.015625,
      "learning_rate": 0.00022926611994419693,
      "loss": 1.114,
      "step": 39650
    },
    {
      "epoch": 3.41142562161232,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00022924338661200715,
      "loss": 1.0083,
      "step": 39651
    },
    {
      "epoch": 3.4115116579196423,
      "grad_norm": 43.75,
      "learning_rate": 0.00022922065407172766,
      "loss": 1.0344,
      "step": 39652
    },
    {
      "epoch": 3.411597694226964,
      "grad_norm": 27.375,
      "learning_rate": 0.000229197922323425,
      "loss": 0.9557,
      "step": 39653
    },
    {
      "epoch": 3.4116837305342855,
      "grad_norm": 1.125,
      "learning_rate": 0.00022917519136716604,
      "loss": 1.4666,
      "step": 39654
    },
    {
      "epoch": 3.411769766841607,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00022915246120301669,
      "loss": 1.0362,
      "step": 39655
    },
    {
      "epoch": 3.411855803148929,
      "grad_norm": 0.875,
      "learning_rate": 0.0002291297318310438,
      "loss": 0.7552,
      "step": 39656
    },
    {
      "epoch": 3.4119418394562504,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002291070032513139,
      "loss": 0.5152,
      "step": 39657
    },
    {
      "epoch": 3.412027875763572,
      "grad_norm": 0.90625,
      "learning_rate": 0.00022908427546389294,
      "loss": 1.0437,
      "step": 39658
    },
    {
      "epoch": 3.412113912070894,
      "grad_norm": 0.875,
      "learning_rate": 0.00022906154846884819,
      "loss": 1.0639,
      "step": 39659
    },
    {
      "epoch": 3.4121999483782157,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0002290388222662455,
      "loss": 1.9809,
      "step": 39660
    },
    {
      "epoch": 3.4122859846855373,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0002290160968561516,
      "loss": 0.8641,
      "step": 39661
    },
    {
      "epoch": 3.412372020992859,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002289933722386331,
      "loss": 0.6663,
      "step": 39662
    },
    {
      "epoch": 3.4124580573001806,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022897064841375604,
      "loss": 1.361,
      "step": 39663
    },
    {
      "epoch": 3.412544093607502,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00022894792538158716,
      "loss": 1.2468,
      "step": 39664
    },
    {
      "epoch": 3.412630129914824,
      "grad_norm": 4.25,
      "learning_rate": 0.00022892520314219306,
      "loss": 1.3969,
      "step": 39665
    },
    {
      "epoch": 3.412716166222146,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00022890248169563982,
      "loss": 0.6126,
      "step": 39666
    },
    {
      "epoch": 3.4128022025294675,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00022887976104199426,
      "loss": 1.15,
      "step": 39667
    },
    {
      "epoch": 3.412888238836789,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00022885704118132255,
      "loss": 1.26,
      "step": 39668
    },
    {
      "epoch": 3.412974275144111,
      "grad_norm": 1.765625,
      "learning_rate": 0.00022883432211369121,
      "loss": 1.2188,
      "step": 39669
    },
    {
      "epoch": 3.4130603114514324,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00022881160383916694,
      "loss": 0.9426,
      "step": 39670
    },
    {
      "epoch": 3.413146347758754,
      "grad_norm": 1.359375,
      "learning_rate": 0.00022878888635781582,
      "loss": 0.6207,
      "step": 39671
    },
    {
      "epoch": 3.4132323840660757,
      "grad_norm": 1960.0,
      "learning_rate": 0.00022876616966970443,
      "loss": 0.601,
      "step": 39672
    },
    {
      "epoch": 3.4133184203733977,
      "grad_norm": 3.859375,
      "learning_rate": 0.0002287434537748994,
      "loss": 0.5765,
      "step": 39673
    },
    {
      "epoch": 3.4134044566807193,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00022872073867346687,
      "loss": 0.6367,
      "step": 39674
    },
    {
      "epoch": 3.413490492988041,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00022869802436547342,
      "loss": 1.3176,
      "step": 39675
    },
    {
      "epoch": 3.4135765292953626,
      "grad_norm": 1.953125,
      "learning_rate": 0.00022867531085098565,
      "loss": 0.636,
      "step": 39676
    },
    {
      "epoch": 3.413662565602684,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00022865259813006968,
      "loss": 0.8041,
      "step": 39677
    },
    {
      "epoch": 3.413748601910006,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00022862988620279224,
      "loss": 1.3082,
      "step": 39678
    },
    {
      "epoch": 3.4138346382173275,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002286071750692194,
      "loss": 0.6673,
      "step": 39679
    },
    {
      "epoch": 3.4139206745246495,
      "grad_norm": 1.859375,
      "learning_rate": 0.0002285844647294178,
      "loss": 1.1791,
      "step": 39680
    },
    {
      "epoch": 3.414006710831971,
      "grad_norm": 0.984375,
      "learning_rate": 0.00022856175518345406,
      "loss": 0.6646,
      "step": 39681
    },
    {
      "epoch": 3.414092747139293,
      "grad_norm": 1.21875,
      "learning_rate": 0.00022853904643139417,
      "loss": 1.1178,
      "step": 39682
    },
    {
      "epoch": 3.4141787834466144,
      "grad_norm": 632.0,
      "learning_rate": 0.00022851633847330484,
      "loss": 0.9602,
      "step": 39683
    },
    {
      "epoch": 3.414264819753936,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00022849363130925254,
      "loss": 1.2849,
      "step": 39684
    },
    {
      "epoch": 3.414350856061258,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00022847092493930338,
      "loss": 0.8385,
      "step": 39685
    },
    {
      "epoch": 3.4144368923685797,
      "grad_norm": 580.0,
      "learning_rate": 0.000228448219363524,
      "loss": 1.0555,
      "step": 39686
    },
    {
      "epoch": 3.4145229286759013,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002284255145819809,
      "loss": 0.7111,
      "step": 39687
    },
    {
      "epoch": 3.414608964983223,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00022840281059474023,
      "loss": 0.8,
      "step": 39688
    },
    {
      "epoch": 3.4146950012905446,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0002283801074018685,
      "loss": 0.8565,
      "step": 39689
    },
    {
      "epoch": 3.414781037597866,
      "grad_norm": 1.109375,
      "learning_rate": 0.00022835740500343226,
      "loss": 1.1323,
      "step": 39690
    },
    {
      "epoch": 3.414867073905188,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00022833470339949762,
      "loss": 0.9166,
      "step": 39691
    },
    {
      "epoch": 3.41495311021251,
      "grad_norm": 147.0,
      "learning_rate": 0.00022831200259013136,
      "loss": 0.9152,
      "step": 39692
    },
    {
      "epoch": 3.4150391465198315,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022828930257539942,
      "loss": 1.2322,
      "step": 39693
    },
    {
      "epoch": 3.415125182827153,
      "grad_norm": 4.96875,
      "learning_rate": 0.00022826660335536846,
      "loss": 1.6874,
      "step": 39694
    },
    {
      "epoch": 3.415211219134475,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00022824390493010483,
      "loss": 0.7075,
      "step": 39695
    },
    {
      "epoch": 3.4152972554417964,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00022822120729967487,
      "loss": 0.4753,
      "step": 39696
    },
    {
      "epoch": 3.415383291749118,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00022819851046414525,
      "loss": 1.2104,
      "step": 39697
    },
    {
      "epoch": 3.4154693280564397,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00022817581442358194,
      "loss": 1.1347,
      "step": 39698
    },
    {
      "epoch": 3.4155553643637617,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00022815311917805155,
      "loss": 1.2725,
      "step": 39699
    },
    {
      "epoch": 3.4156414006710833,
      "grad_norm": 80.0,
      "learning_rate": 0.00022813042472762053,
      "loss": 1.1051,
      "step": 39700
    },
    {
      "epoch": 3.415727436978405,
      "grad_norm": 0.796875,
      "learning_rate": 0.000228107731072355,
      "loss": 0.983,
      "step": 39701
    },
    {
      "epoch": 3.4158134732857266,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002280850382123214,
      "loss": 0.7374,
      "step": 39702
    },
    {
      "epoch": 3.4158995095930482,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002280623461475864,
      "loss": 0.8312,
      "step": 39703
    },
    {
      "epoch": 3.41598554590037,
      "grad_norm": 1.28125,
      "learning_rate": 0.00022803965487821603,
      "loss": 0.9901,
      "step": 39704
    },
    {
      "epoch": 3.4160715822076915,
      "grad_norm": 66.5,
      "learning_rate": 0.00022801696440427673,
      "loss": 1.3767,
      "step": 39705
    },
    {
      "epoch": 3.4161576185150135,
      "grad_norm": 93.0,
      "learning_rate": 0.00022799427472583513,
      "loss": 0.8958,
      "step": 39706
    },
    {
      "epoch": 3.416243654822335,
      "grad_norm": 1.09375,
      "learning_rate": 0.00022797158584295714,
      "loss": 0.8616,
      "step": 39707
    },
    {
      "epoch": 3.416329691129657,
      "grad_norm": 348.0,
      "learning_rate": 0.00022794889775570954,
      "loss": 1.5135,
      "step": 39708
    },
    {
      "epoch": 3.4164157274369784,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022792621046415835,
      "loss": 0.9176,
      "step": 39709
    },
    {
      "epoch": 3.4165017637443,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00022790352396837005,
      "loss": 1.7015,
      "step": 39710
    },
    {
      "epoch": 3.4165878000516217,
      "grad_norm": 7.0,
      "learning_rate": 0.00022788083826841122,
      "loss": 0.9417,
      "step": 39711
    },
    {
      "epoch": 3.4166738363589433,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00022785815336434783,
      "loss": 0.9676,
      "step": 39712
    },
    {
      "epoch": 3.4167598726662654,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00022783546925624643,
      "loss": 1.423,
      "step": 39713
    },
    {
      "epoch": 3.416845908973587,
      "grad_norm": 596.0,
      "learning_rate": 0.0002278127859441735,
      "loss": 1.5858,
      "step": 39714
    },
    {
      "epoch": 3.4169319452809086,
      "grad_norm": 1.203125,
      "learning_rate": 0.00022779010342819506,
      "loss": 0.6314,
      "step": 39715
    },
    {
      "epoch": 3.4170179815882302,
      "grad_norm": 58.25,
      "learning_rate": 0.0002277674217083776,
      "loss": 1.496,
      "step": 39716
    },
    {
      "epoch": 3.417104017895552,
      "grad_norm": 1.46875,
      "learning_rate": 0.00022774474078478775,
      "loss": 1.3119,
      "step": 39717
    },
    {
      "epoch": 3.4171900542028735,
      "grad_norm": 1.28125,
      "learning_rate": 0.00022772206065749134,
      "loss": 0.5371,
      "step": 39718
    },
    {
      "epoch": 3.417276090510195,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002276993813265551,
      "loss": 0.553,
      "step": 39719
    },
    {
      "epoch": 3.417362126817517,
      "grad_norm": 16.75,
      "learning_rate": 0.0002276767027920451,
      "loss": 1.1217,
      "step": 39720
    },
    {
      "epoch": 3.417448163124839,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002276540250540277,
      "loss": 1.3415,
      "step": 39721
    },
    {
      "epoch": 3.4175341994321604,
      "grad_norm": 9792.0,
      "learning_rate": 0.0002276313481125696,
      "loss": 1.6535,
      "step": 39722
    },
    {
      "epoch": 3.417620235739482,
      "grad_norm": 1.09375,
      "learning_rate": 0.00022760867196773655,
      "loss": 0.8798,
      "step": 39723
    },
    {
      "epoch": 3.4177062720468037,
      "grad_norm": 1.609375,
      "learning_rate": 0.00022758599661959524,
      "loss": 1.4145,
      "step": 39724
    },
    {
      "epoch": 3.4177923083541253,
      "grad_norm": 0.890625,
      "learning_rate": 0.00022756332206821201,
      "loss": 0.8282,
      "step": 39725
    },
    {
      "epoch": 3.417878344661447,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00022754064831365296,
      "loss": 0.5971,
      "step": 39726
    },
    {
      "epoch": 3.417964380968769,
      "grad_norm": 2.75,
      "learning_rate": 0.00022751797535598445,
      "loss": 0.9072,
      "step": 39727
    },
    {
      "epoch": 3.4180504172760906,
      "grad_norm": 44.75,
      "learning_rate": 0.00022749530319527312,
      "loss": 1.282,
      "step": 39728
    },
    {
      "epoch": 3.4181364535834122,
      "grad_norm": 1.390625,
      "learning_rate": 0.00022747263183158485,
      "loss": 1.1605,
      "step": 39729
    },
    {
      "epoch": 3.418222489890734,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00022744996126498608,
      "loss": 1.2824,
      "step": 39730
    },
    {
      "epoch": 3.4183085261980555,
      "grad_norm": 1.015625,
      "learning_rate": 0.00022742729149554313,
      "loss": 1.145,
      "step": 39731
    },
    {
      "epoch": 3.418394562505377,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00022740462252332245,
      "loss": 0.9143,
      "step": 39732
    },
    {
      "epoch": 3.4184805988126987,
      "grad_norm": 1.140625,
      "learning_rate": 0.00022738195434839032,
      "loss": 1.7889,
      "step": 39733
    },
    {
      "epoch": 3.418566635120021,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0002273592869708128,
      "loss": 2.0632,
      "step": 39734
    },
    {
      "epoch": 3.4186526714273424,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00022733662039065656,
      "loss": 1.0454,
      "step": 39735
    },
    {
      "epoch": 3.418738707734664,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00022731395460798748,
      "loss": 0.6302,
      "step": 39736
    },
    {
      "epoch": 3.4188247440419857,
      "grad_norm": 3.484375,
      "learning_rate": 0.00022729128962287203,
      "loss": 0.7838,
      "step": 39737
    },
    {
      "epoch": 3.4189107803493073,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00022726862543537667,
      "loss": 2.0193,
      "step": 39738
    },
    {
      "epoch": 3.418996816656629,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022724596204556742,
      "loss": 1.0276,
      "step": 39739
    },
    {
      "epoch": 3.419082852963951,
      "grad_norm": 0.90625,
      "learning_rate": 0.00022722329945351066,
      "loss": 0.7449,
      "step": 39740
    },
    {
      "epoch": 3.4191688892712726,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00022720063765927284,
      "loss": 0.8857,
      "step": 39741
    },
    {
      "epoch": 3.4192549255785942,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00022717797666291996,
      "loss": 1.3637,
      "step": 39742
    },
    {
      "epoch": 3.419340961885916,
      "grad_norm": 1.328125,
      "learning_rate": 0.00022715531646451843,
      "loss": 1.2533,
      "step": 39743
    },
    {
      "epoch": 3.4194269981932375,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00022713265706413472,
      "loss": 1.0367,
      "step": 39744
    },
    {
      "epoch": 3.419513034500559,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00022710999846183472,
      "loss": 0.5566,
      "step": 39745
    },
    {
      "epoch": 3.419599070807881,
      "grad_norm": 20.875,
      "learning_rate": 0.00022708734065768487,
      "loss": 1.1503,
      "step": 39746
    },
    {
      "epoch": 3.419685107115203,
      "grad_norm": 24.5,
      "learning_rate": 0.0002270646836517517,
      "loss": 1.8079,
      "step": 39747
    },
    {
      "epoch": 3.4197711434225244,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00022704202744410108,
      "loss": 1.2166,
      "step": 39748
    },
    {
      "epoch": 3.419857179729846,
      "grad_norm": 25.375,
      "learning_rate": 0.0002270193720347996,
      "loss": 1.3047,
      "step": 39749
    },
    {
      "epoch": 3.4199432160371677,
      "grad_norm": 1.875,
      "learning_rate": 0.0002269967174239132,
      "loss": 1.1417,
      "step": 39750
    },
    {
      "epoch": 3.4200292523444893,
      "grad_norm": 1.234375,
      "learning_rate": 0.00022697406361150824,
      "loss": 1.3032,
      "step": 39751
    },
    {
      "epoch": 3.420115288651811,
      "grad_norm": 4.0,
      "learning_rate": 0.0002269514105976513,
      "loss": 1.6102,
      "step": 39752
    },
    {
      "epoch": 3.420201324959133,
      "grad_norm": 1.21875,
      "learning_rate": 0.00022692875838240816,
      "loss": 0.8604,
      "step": 39753
    },
    {
      "epoch": 3.4202873612664546,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002269061069658453,
      "loss": 1.838,
      "step": 39754
    },
    {
      "epoch": 3.4203733975737762,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00022688345634802915,
      "loss": 1.199,
      "step": 39755
    },
    {
      "epoch": 3.420459433881098,
      "grad_norm": 2.21875,
      "learning_rate": 0.00022686080652902557,
      "loss": 0.7397,
      "step": 39756
    },
    {
      "epoch": 3.4205454701884195,
      "grad_norm": 1.15625,
      "learning_rate": 0.00022683815750890104,
      "loss": 1.4213,
      "step": 39757
    },
    {
      "epoch": 3.420631506495741,
      "grad_norm": 1.125,
      "learning_rate": 0.00022681550928772194,
      "loss": 1.7184,
      "step": 39758
    },
    {
      "epoch": 3.4207175428030627,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00022679286186555415,
      "loss": 0.7854,
      "step": 39759
    },
    {
      "epoch": 3.420803579110385,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00022677021524246434,
      "loss": 1.0781,
      "step": 39760
    },
    {
      "epoch": 3.4208896154177064,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00022674756941851827,
      "loss": 0.6926,
      "step": 39761
    },
    {
      "epoch": 3.420975651725028,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00022672492439378245,
      "loss": 1.0736,
      "step": 39762
    },
    {
      "epoch": 3.4210616880323497,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00022670228016832323,
      "loss": 1.6062,
      "step": 39763
    },
    {
      "epoch": 3.4211477243396713,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0002266796367422065,
      "loss": 1.2896,
      "step": 39764
    },
    {
      "epoch": 3.421233760646993,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00022665699411549868,
      "loss": 1.2619,
      "step": 39765
    },
    {
      "epoch": 3.4213197969543145,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00022663435228826618,
      "loss": 0.728,
      "step": 39766
    },
    {
      "epoch": 3.4214058332616366,
      "grad_norm": 5.5,
      "learning_rate": 0.00022661171126057484,
      "loss": 1.3275,
      "step": 39767
    },
    {
      "epoch": 3.4214918695689582,
      "grad_norm": 2.203125,
      "learning_rate": 0.0002265890710324911,
      "loss": 2.1594,
      "step": 39768
    },
    {
      "epoch": 3.42157790587628,
      "grad_norm": 1.03125,
      "learning_rate": 0.00022656643160408113,
      "loss": 1.238,
      "step": 39769
    },
    {
      "epoch": 3.4216639421836015,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00022654379297541117,
      "loss": 0.8413,
      "step": 39770
    },
    {
      "epoch": 3.421749978490923,
      "grad_norm": 1.578125,
      "learning_rate": 0.00022652115514654764,
      "loss": 1.225,
      "step": 39771
    },
    {
      "epoch": 3.4218360147982447,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00022649851811755634,
      "loss": 1.5626,
      "step": 39772
    },
    {
      "epoch": 3.4219220511055664,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00022647588188850365,
      "loss": 1.2996,
      "step": 39773
    },
    {
      "epoch": 3.4220080874128884,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022645324645945603,
      "loss": 0.8437,
      "step": 39774
    },
    {
      "epoch": 3.42209412372021,
      "grad_norm": 1.5,
      "learning_rate": 0.00022643061183047925,
      "loss": 1.0089,
      "step": 39775
    },
    {
      "epoch": 3.4221801600275317,
      "grad_norm": 24.625,
      "learning_rate": 0.00022640797800163975,
      "loss": 1.4977,
      "step": 39776
    },
    {
      "epoch": 3.4222661963348533,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002263853449730039,
      "loss": 1.485,
      "step": 39777
    },
    {
      "epoch": 3.422352232642175,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022636271274463748,
      "loss": 1.3022,
      "step": 39778
    },
    {
      "epoch": 3.4224382689494965,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002263400813166071,
      "loss": 1.0988,
      "step": 39779
    },
    {
      "epoch": 3.422524305256818,
      "grad_norm": 3.15625,
      "learning_rate": 0.00022631745068897865,
      "loss": 1.3923,
      "step": 39780
    },
    {
      "epoch": 3.4226103415641402,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002262948208618183,
      "loss": 0.9535,
      "step": 39781
    },
    {
      "epoch": 3.422696377871462,
      "grad_norm": 2.78125,
      "learning_rate": 0.00022627219183519266,
      "loss": 1.0454,
      "step": 39782
    },
    {
      "epoch": 3.4227824141787835,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00022624956360916737,
      "loss": 1.3189,
      "step": 39783
    },
    {
      "epoch": 3.422868450486105,
      "grad_norm": 1.15625,
      "learning_rate": 0.00022622693618380886,
      "loss": 0.9914,
      "step": 39784
    },
    {
      "epoch": 3.4229544867934267,
      "grad_norm": 1.328125,
      "learning_rate": 0.00022620430955918352,
      "loss": 1.2604,
      "step": 39785
    },
    {
      "epoch": 3.4230405231007484,
      "grad_norm": 154.0,
      "learning_rate": 0.00022618168373535715,
      "loss": 1.4506,
      "step": 39786
    },
    {
      "epoch": 3.42312655940807,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002261590587123961,
      "loss": 1.6227,
      "step": 39787
    },
    {
      "epoch": 3.423212595715392,
      "grad_norm": 1.0,
      "learning_rate": 0.0002261364344903667,
      "loss": 1.5392,
      "step": 39788
    },
    {
      "epoch": 3.4232986320227137,
      "grad_norm": 0.90625,
      "learning_rate": 0.00022611381106933483,
      "loss": 0.9185,
      "step": 39789
    },
    {
      "epoch": 3.4233846683300353,
      "grad_norm": 11.1875,
      "learning_rate": 0.00022609118844936694,
      "loss": 1.221,
      "step": 39790
    },
    {
      "epoch": 3.423470704637357,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00022606856663052888,
      "loss": 1.4457,
      "step": 39791
    },
    {
      "epoch": 3.4235567409446785,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000226045945612887,
      "loss": 0.8328,
      "step": 39792
    },
    {
      "epoch": 3.423642777252,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00022602332539650762,
      "loss": 0.8107,
      "step": 39793
    },
    {
      "epoch": 3.423728813559322,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002260007059814565,
      "loss": 1.0486,
      "step": 39794
    },
    {
      "epoch": 3.423814849866644,
      "grad_norm": 81.5,
      "learning_rate": 0.00022597808736780006,
      "loss": 1.1701,
      "step": 39795
    },
    {
      "epoch": 3.4239008861739655,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022595546955560465,
      "loss": 1.2407,
      "step": 39796
    },
    {
      "epoch": 3.423986922481287,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00022593285254493597,
      "loss": 0.7363,
      "step": 39797
    },
    {
      "epoch": 3.4240729587886087,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002259102363358604,
      "loss": 0.9489,
      "step": 39798
    },
    {
      "epoch": 3.4241589950959304,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00022588762092844421,
      "loss": 1.2051,
      "step": 39799
    },
    {
      "epoch": 3.424245031403252,
      "grad_norm": 15.75,
      "learning_rate": 0.0002258650063227533,
      "loss": 1.1455,
      "step": 39800
    },
    {
      "epoch": 3.424331067710574,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00022584239251885414,
      "loss": 0.8373,
      "step": 39801
    },
    {
      "epoch": 3.4244171040178957,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00022581977951681236,
      "loss": 0.9766,
      "step": 39802
    },
    {
      "epoch": 3.4245031403252173,
      "grad_norm": 20.5,
      "learning_rate": 0.00022579716731669454,
      "loss": 1.413,
      "step": 39803
    },
    {
      "epoch": 3.424589176632539,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00022577455591856678,
      "loss": 1.2189,
      "step": 39804
    },
    {
      "epoch": 3.4246752129398605,
      "grad_norm": 15.25,
      "learning_rate": 0.0002257519453224947,
      "loss": 1.4644,
      "step": 39805
    },
    {
      "epoch": 3.424761249247182,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002257293355285453,
      "loss": 1.1077,
      "step": 39806
    },
    {
      "epoch": 3.4248472855545042,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00022570672653678405,
      "loss": 0.579,
      "step": 39807
    },
    {
      "epoch": 3.424933321861826,
      "grad_norm": 1.7890625,
      "learning_rate": 0.00022568411834727727,
      "loss": 1.1843,
      "step": 39808
    },
    {
      "epoch": 3.4250193581691475,
      "grad_norm": 2.015625,
      "learning_rate": 0.00022566151096009123,
      "loss": 1.9636,
      "step": 39809
    },
    {
      "epoch": 3.425105394476469,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002256389043752918,
      "loss": 1.2791,
      "step": 39810
    },
    {
      "epoch": 3.4251914307837907,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00022561629859294514,
      "loss": 0.8691,
      "step": 39811
    },
    {
      "epoch": 3.4252774670911124,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00022559369361311766,
      "loss": 1.6762,
      "step": 39812
    },
    {
      "epoch": 3.425363503398434,
      "grad_norm": 3.0,
      "learning_rate": 0.00022557108943587506,
      "loss": 0.6892,
      "step": 39813
    },
    {
      "epoch": 3.425449539705756,
      "grad_norm": 7.21875,
      "learning_rate": 0.0002255484860612837,
      "loss": 1.3907,
      "step": 39814
    },
    {
      "epoch": 3.4255355760130777,
      "grad_norm": 0.984375,
      "learning_rate": 0.00022552588348940977,
      "loss": 1.1795,
      "step": 39815
    },
    {
      "epoch": 3.4256216123203993,
      "grad_norm": 7.375,
      "learning_rate": 0.00022550328172031908,
      "loss": 1.2624,
      "step": 39816
    },
    {
      "epoch": 3.425707648627721,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00022548068075407796,
      "loss": 0.9534,
      "step": 39817
    },
    {
      "epoch": 3.4257936849350425,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00022545808059075257,
      "loss": 0.6631,
      "step": 39818
    },
    {
      "epoch": 3.425879721242364,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00022543548123040875,
      "loss": 1.25,
      "step": 39819
    },
    {
      "epoch": 3.425965757549686,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00022541288267311288,
      "loss": 0.7873,
      "step": 39820
    },
    {
      "epoch": 3.426051793857008,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00022539028491893082,
      "loss": 1.2979,
      "step": 39821
    },
    {
      "epoch": 3.4261378301643295,
      "grad_norm": 2.96875,
      "learning_rate": 0.00022536768796792872,
      "loss": 0.8787,
      "step": 39822
    },
    {
      "epoch": 3.426223866471651,
      "grad_norm": 1.953125,
      "learning_rate": 0.00022534509182017294,
      "loss": 2.2745,
      "step": 39823
    },
    {
      "epoch": 3.4263099027789727,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00022532249647572912,
      "loss": 0.7569,
      "step": 39824
    },
    {
      "epoch": 3.4263959390862944,
      "grad_norm": 1.125,
      "learning_rate": 0.0002252999019346636,
      "loss": 1.3977,
      "step": 39825
    },
    {
      "epoch": 3.426481975393616,
      "grad_norm": 1.65625,
      "learning_rate": 0.00022527730819704256,
      "loss": 1.1674,
      "step": 39826
    },
    {
      "epoch": 3.4265680117009376,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022525471526293185,
      "loss": 1.131,
      "step": 39827
    },
    {
      "epoch": 3.4266540480082597,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00022523212313239766,
      "loss": 0.5479,
      "step": 39828
    },
    {
      "epoch": 3.4267400843155813,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00022520953180550624,
      "loss": 1.5905,
      "step": 39829
    },
    {
      "epoch": 3.426826120622903,
      "grad_norm": 1.40625,
      "learning_rate": 0.00022518694128232332,
      "loss": 1.6154,
      "step": 39830
    },
    {
      "epoch": 3.4269121569302246,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002251643515629153,
      "loss": 0.7843,
      "step": 39831
    },
    {
      "epoch": 3.426998193237546,
      "grad_norm": 1.234375,
      "learning_rate": 0.000225141762647348,
      "loss": 1.3229,
      "step": 39832
    },
    {
      "epoch": 3.427084229544868,
      "grad_norm": 10.4375,
      "learning_rate": 0.00022511917453568747,
      "loss": 1.201,
      "step": 39833
    },
    {
      "epoch": 3.4271702658521894,
      "grad_norm": 1.78125,
      "learning_rate": 0.00022509658722800018,
      "loss": 1.3639,
      "step": 39834
    },
    {
      "epoch": 3.4272563021595115,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002250740007243517,
      "loss": 1.1606,
      "step": 39835
    },
    {
      "epoch": 3.427342338466833,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00022505141502480826,
      "loss": 1.2486,
      "step": 39836
    },
    {
      "epoch": 3.4274283747741547,
      "grad_norm": 1.390625,
      "learning_rate": 0.00022502883012943614,
      "loss": 1.4057,
      "step": 39837
    },
    {
      "epoch": 3.4275144110814764,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00022500624603830107,
      "loss": 0.5821,
      "step": 39838
    },
    {
      "epoch": 3.427600447388798,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002249836627514692,
      "loss": 1.0208,
      "step": 39839
    },
    {
      "epoch": 3.4276864836961196,
      "grad_norm": 2.1875,
      "learning_rate": 0.00022496108026900686,
      "loss": 0.9733,
      "step": 39840
    },
    {
      "epoch": 3.4277725200034412,
      "grad_norm": 12.9375,
      "learning_rate": 0.00022493849859097964,
      "loss": 1.4843,
      "step": 39841
    },
    {
      "epoch": 3.4278585563107633,
      "grad_norm": 1.25,
      "learning_rate": 0.00022491591771745385,
      "loss": 1.1914,
      "step": 39842
    },
    {
      "epoch": 3.427944592618085,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00022489333764849544,
      "loss": 1.0926,
      "step": 39843
    },
    {
      "epoch": 3.4280306289254066,
      "grad_norm": 22.5,
      "learning_rate": 0.0002248707583841706,
      "loss": 0.9306,
      "step": 39844
    },
    {
      "epoch": 3.428116665232728,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00022484817992454544,
      "loss": 1.2828,
      "step": 39845
    },
    {
      "epoch": 3.42820270154005,
      "grad_norm": 1.15625,
      "learning_rate": 0.00022482560226968562,
      "loss": 1.5509,
      "step": 39846
    },
    {
      "epoch": 3.4282887378473714,
      "grad_norm": 1.734375,
      "learning_rate": 0.00022480302541965757,
      "loss": 1.206,
      "step": 39847
    },
    {
      "epoch": 3.428374774154693,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000224780449374527,
      "loss": 1.1865,
      "step": 39848
    },
    {
      "epoch": 3.428460810462015,
      "grad_norm": 1.515625,
      "learning_rate": 0.0002247578741343601,
      "loss": 1.2588,
      "step": 39849
    },
    {
      "epoch": 3.4285468467693367,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000224735299699223,
      "loss": 0.4768,
      "step": 39850
    },
    {
      "epoch": 3.4286328830766584,
      "grad_norm": 77.0,
      "learning_rate": 0.00022471272606918152,
      "loss": 1.5027,
      "step": 39851
    },
    {
      "epoch": 3.42871891938398,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022469015324430176,
      "loss": 1.1283,
      "step": 39852
    },
    {
      "epoch": 3.4288049556913016,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00022466758122464985,
      "loss": 1.8785,
      "step": 39853
    },
    {
      "epoch": 3.4288909919986232,
      "grad_norm": 1.15625,
      "learning_rate": 0.00022464501001029163,
      "loss": 0.8082,
      "step": 39854
    },
    {
      "epoch": 3.428977028305945,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002246224396012932,
      "loss": 1.6931,
      "step": 39855
    },
    {
      "epoch": 3.429063064613267,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002245998699977207,
      "loss": 0.869,
      "step": 39856
    },
    {
      "epoch": 3.4291491009205886,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00022457730119963988,
      "loss": 0.8668,
      "step": 39857
    },
    {
      "epoch": 3.42923513722791,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00022455473320711684,
      "loss": 1.5586,
      "step": 39858
    },
    {
      "epoch": 3.429321173535232,
      "grad_norm": 7.4375,
      "learning_rate": 0.00022453216602021786,
      "loss": 0.994,
      "step": 39859
    },
    {
      "epoch": 3.4294072098425534,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022450959963900852,
      "loss": 1.4896,
      "step": 39860
    },
    {
      "epoch": 3.4294932461498755,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00022448703406355525,
      "loss": 0.6705,
      "step": 39861
    },
    {
      "epoch": 3.429579282457197,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00022446446929392356,
      "loss": 0.6,
      "step": 39862
    },
    {
      "epoch": 3.4296653187645187,
      "grad_norm": 4.65625,
      "learning_rate": 0.00022444190533017978,
      "loss": 1.4552,
      "step": 39863
    },
    {
      "epoch": 3.4297513550718404,
      "grad_norm": 1.46875,
      "learning_rate": 0.00022441934217238996,
      "loss": 1.1346,
      "step": 39864
    },
    {
      "epoch": 3.429837391379162,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00022439677982061978,
      "loss": 1.0876,
      "step": 39865
    },
    {
      "epoch": 3.4299234276864836,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00022437421827493548,
      "loss": 1.6615,
      "step": 39866
    },
    {
      "epoch": 3.4300094639938052,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0002243516575354031,
      "loss": 0.8284,
      "step": 39867
    },
    {
      "epoch": 3.4300955003011273,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022432909760208836,
      "loss": 0.8184,
      "step": 39868
    },
    {
      "epoch": 3.430181536608449,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002243065384750574,
      "loss": 1.3672,
      "step": 39869
    },
    {
      "epoch": 3.4302675729157706,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002242839801543764,
      "loss": 0.7233,
      "step": 39870
    },
    {
      "epoch": 3.430353609223092,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002242614226401109,
      "loss": 1.1215,
      "step": 39871
    },
    {
      "epoch": 3.430439645530414,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00022423886593232734,
      "loss": 1.2945,
      "step": 39872
    },
    {
      "epoch": 3.4305256818377354,
      "grad_norm": 1.234375,
      "learning_rate": 0.00022421631003109122,
      "loss": 1.3903,
      "step": 39873
    },
    {
      "epoch": 3.430611718145057,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0002241937549364688,
      "loss": 1.5862,
      "step": 39874
    },
    {
      "epoch": 3.430697754452379,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002241712006485262,
      "loss": 1.7,
      "step": 39875
    },
    {
      "epoch": 3.4307837907597007,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00022414864716732897,
      "loss": 1.226,
      "step": 39876
    },
    {
      "epoch": 3.4308698270670224,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022412609449294337,
      "loss": 1.437,
      "step": 39877
    },
    {
      "epoch": 3.430955863374344,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002241035426254354,
      "loss": 0.4622,
      "step": 39878
    },
    {
      "epoch": 3.4310418996816656,
      "grad_norm": 105984.0,
      "learning_rate": 0.00022408099156487056,
      "loss": 0.7364,
      "step": 39879
    },
    {
      "epoch": 3.4311279359889872,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00022405844131131552,
      "loss": 0.6598,
      "step": 39880
    },
    {
      "epoch": 3.431213972296309,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002240358918648357,
      "loss": 1.1634,
      "step": 39881
    },
    {
      "epoch": 3.431300008603631,
      "grad_norm": 1.15625,
      "learning_rate": 0.00022401334322549722,
      "loss": 1.18,
      "step": 39882
    },
    {
      "epoch": 3.4313860449109526,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002239907953933662,
      "loss": 1.0905,
      "step": 39883
    },
    {
      "epoch": 3.431472081218274,
      "grad_norm": 1.296875,
      "learning_rate": 0.00022396824836850828,
      "loss": 1.0025,
      "step": 39884
    },
    {
      "epoch": 3.431558117525596,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022394570215098953,
      "loss": 1.4258,
      "step": 39885
    },
    {
      "epoch": 3.4316441538329174,
      "grad_norm": 14.875,
      "learning_rate": 0.0002239231567408761,
      "loss": 0.8207,
      "step": 39886
    },
    {
      "epoch": 3.431730190140239,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00022390061213823354,
      "loss": 1.2806,
      "step": 39887
    },
    {
      "epoch": 3.4318162264475607,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002238780683431282,
      "loss": 0.9175,
      "step": 39888
    },
    {
      "epoch": 3.4319022627548827,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00022385552535562558,
      "loss": 1.3635,
      "step": 39889
    },
    {
      "epoch": 3.4319882990622044,
      "grad_norm": 1.0625,
      "learning_rate": 0.00022383298317579187,
      "loss": 0.9427,
      "step": 39890
    },
    {
      "epoch": 3.432074335369526,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00022381044180369314,
      "loss": 1.251,
      "step": 39891
    },
    {
      "epoch": 3.4321603716768476,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00022378790123939502,
      "loss": 1.2479,
      "step": 39892
    },
    {
      "epoch": 3.4322464079841692,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002237653614829635,
      "loss": 0.9016,
      "step": 39893
    },
    {
      "epoch": 3.432332444291491,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00022374282253446483,
      "loss": 0.9791,
      "step": 39894
    },
    {
      "epoch": 3.4324184805988125,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0002237202843939644,
      "loss": 0.5344,
      "step": 39895
    },
    {
      "epoch": 3.4325045169061346,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00022369774706152847,
      "loss": 1.2112,
      "step": 39896
    },
    {
      "epoch": 3.432590553213456,
      "grad_norm": 0.90625,
      "learning_rate": 0.00022367521053722307,
      "loss": 0.8828,
      "step": 39897
    },
    {
      "epoch": 3.432676589520778,
      "grad_norm": 3.015625,
      "learning_rate": 0.00022365267482111372,
      "loss": 1.4222,
      "step": 39898
    },
    {
      "epoch": 3.4327626258280994,
      "grad_norm": 1.34375,
      "learning_rate": 0.00022363013991326657,
      "loss": 1.8445,
      "step": 39899
    },
    {
      "epoch": 3.432848662135421,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022360760581374773,
      "loss": 1.2606,
      "step": 39900
    },
    {
      "epoch": 3.4329346984427427,
      "grad_norm": 1.25,
      "learning_rate": 0.00022358507252262267,
      "loss": 1.363,
      "step": 39901
    },
    {
      "epoch": 3.4330207347500643,
      "grad_norm": 1.625,
      "learning_rate": 0.0002235625400399577,
      "loss": 0.7343,
      "step": 39902
    },
    {
      "epoch": 3.4331067710573864,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022354000836581834,
      "loss": 1.1911,
      "step": 39903
    },
    {
      "epoch": 3.433192807364708,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002235174775002707,
      "loss": 1.3852,
      "step": 39904
    },
    {
      "epoch": 3.4332788436720296,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00022349494744338083,
      "loss": 0.5841,
      "step": 39905
    },
    {
      "epoch": 3.4333648799793512,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002234724181952143,
      "loss": 0.887,
      "step": 39906
    },
    {
      "epoch": 3.433450916286673,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002234498897558372,
      "loss": 0.9868,
      "step": 39907
    },
    {
      "epoch": 3.4335369525939945,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002234273621253155,
      "loss": 0.9945,
      "step": 39908
    },
    {
      "epoch": 3.433622988901316,
      "grad_norm": 2.375,
      "learning_rate": 0.0002234048353037148,
      "loss": 1.5598,
      "step": 39909
    },
    {
      "epoch": 3.433709025208638,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022338230929110122,
      "loss": 0.8733,
      "step": 39910
    },
    {
      "epoch": 3.43379506151596,
      "grad_norm": 47.0,
      "learning_rate": 0.0002233597840875407,
      "loss": 1.1703,
      "step": 39911
    },
    {
      "epoch": 3.4338810978232814,
      "grad_norm": 1.0625,
      "learning_rate": 0.00022333725969309886,
      "loss": 0.9871,
      "step": 39912
    },
    {
      "epoch": 3.433967134130603,
      "grad_norm": 2.625,
      "learning_rate": 0.00022331473610784193,
      "loss": 1.0326,
      "step": 39913
    },
    {
      "epoch": 3.4340531704379247,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00022329221333183537,
      "loss": 0.7747,
      "step": 39914
    },
    {
      "epoch": 3.4341392067452463,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002232696913651453,
      "loss": 1.6203,
      "step": 39915
    },
    {
      "epoch": 3.4342252430525684,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00022324717020783753,
      "loss": 0.8703,
      "step": 39916
    },
    {
      "epoch": 3.43431127935989,
      "grad_norm": 1.1875,
      "learning_rate": 0.00022322464985997798,
      "loss": 1.6174,
      "step": 39917
    },
    {
      "epoch": 3.4343973156672116,
      "grad_norm": 1.5625,
      "learning_rate": 0.00022320213032163267,
      "loss": 0.9176,
      "step": 39918
    },
    {
      "epoch": 3.4344833519745332,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002231796115928671,
      "loss": 1.1641,
      "step": 39919
    },
    {
      "epoch": 3.434569388281855,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00022315709367374737,
      "loss": 1.5701,
      "step": 39920
    },
    {
      "epoch": 3.4346554245891765,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00022313457656433945,
      "loss": 0.9177,
      "step": 39921
    },
    {
      "epoch": 3.4347414608964986,
      "grad_norm": 1.0625,
      "learning_rate": 0.00022311206026470886,
      "loss": 0.8807,
      "step": 39922
    },
    {
      "epoch": 3.43482749720382,
      "grad_norm": 1.40625,
      "learning_rate": 0.00022308954477492167,
      "loss": 1.7714,
      "step": 39923
    },
    {
      "epoch": 3.434913533511142,
      "grad_norm": 1.328125,
      "learning_rate": 0.00022306703009504386,
      "loss": 1.4455,
      "step": 39924
    },
    {
      "epoch": 3.4349995698184634,
      "grad_norm": 1.1875,
      "learning_rate": 0.00022304451622514094,
      "loss": 2.0451,
      "step": 39925
    },
    {
      "epoch": 3.435085606125785,
      "grad_norm": 1.140625,
      "learning_rate": 0.00022302200316527893,
      "loss": 0.9665,
      "step": 39926
    },
    {
      "epoch": 3.4351716424331067,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002229994909155239,
      "loss": 1.3327,
      "step": 39927
    },
    {
      "epoch": 3.4352576787404283,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0002229769794759413,
      "loss": 0.5967,
      "step": 39928
    },
    {
      "epoch": 3.4353437150477504,
      "grad_norm": 1.265625,
      "learning_rate": 0.00022295446884659721,
      "loss": 0.6823,
      "step": 39929
    },
    {
      "epoch": 3.435429751355072,
      "grad_norm": 0.96875,
      "learning_rate": 0.00022293195902755753,
      "loss": 1.2202,
      "step": 39930
    },
    {
      "epoch": 3.4355157876623936,
      "grad_norm": 1.203125,
      "learning_rate": 0.00022290945001888785,
      "loss": 0.7685,
      "step": 39931
    },
    {
      "epoch": 3.4356018239697153,
      "grad_norm": 2.234375,
      "learning_rate": 0.00022288694182065428,
      "loss": 1.1066,
      "step": 39932
    },
    {
      "epoch": 3.435687860277037,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0002228644344329223,
      "loss": 0.8077,
      "step": 39933
    },
    {
      "epoch": 3.4357738965843585,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000222841927855758,
      "loss": 1.4267,
      "step": 39934
    },
    {
      "epoch": 3.43585993289168,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00022281942208922733,
      "loss": 0.7377,
      "step": 39935
    },
    {
      "epoch": 3.435945969199002,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00022279691713339577,
      "loss": 0.7343,
      "step": 39936
    },
    {
      "epoch": 3.436032005506324,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002227744129883293,
      "loss": 1.46,
      "step": 39937
    },
    {
      "epoch": 3.4361180418136454,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00022275190965409392,
      "loss": 1.4691,
      "step": 39938
    },
    {
      "epoch": 3.436204078120967,
      "grad_norm": 6.6875,
      "learning_rate": 0.00022272940713075512,
      "loss": 1.4533,
      "step": 39939
    },
    {
      "epoch": 3.4362901144282887,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002227069054183789,
      "loss": 1.2914,
      "step": 39940
    },
    {
      "epoch": 3.4363761507356103,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002226844045170312,
      "loss": 0.9238,
      "step": 39941
    },
    {
      "epoch": 3.436462187042932,
      "grad_norm": 4.65625,
      "learning_rate": 0.00022266190442677753,
      "loss": 1.9468,
      "step": 39942
    },
    {
      "epoch": 3.436548223350254,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000222639405147684,
      "loss": 1.1833,
      "step": 39943
    },
    {
      "epoch": 3.4366342596575756,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002226169066798161,
      "loss": 2.051,
      "step": 39944
    },
    {
      "epoch": 3.4367202959648973,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002225944090232398,
      "loss": 0.9974,
      "step": 39945
    },
    {
      "epoch": 3.436806332272219,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00022257191217802102,
      "loss": 1.9465,
      "step": 39946
    },
    {
      "epoch": 3.4368923685795405,
      "grad_norm": 1.125,
      "learning_rate": 0.00022254941614422536,
      "loss": 0.9752,
      "step": 39947
    },
    {
      "epoch": 3.436978404886862,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00022252692092191863,
      "loss": 0.7448,
      "step": 39948
    },
    {
      "epoch": 3.4370644411941838,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00022250442651116693,
      "loss": 1.2439,
      "step": 39949
    },
    {
      "epoch": 3.437150477501506,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022248193291203556,
      "loss": 0.751,
      "step": 39950
    },
    {
      "epoch": 3.4372365138088274,
      "grad_norm": 0.921875,
      "learning_rate": 0.00022245944012459063,
      "loss": 0.8685,
      "step": 39951
    },
    {
      "epoch": 3.437322550116149,
      "grad_norm": 1.03125,
      "learning_rate": 0.00022243694814889797,
      "loss": 1.1795,
      "step": 39952
    },
    {
      "epoch": 3.4374085864234707,
      "grad_norm": 1.15625,
      "learning_rate": 0.00022241445698502295,
      "loss": 0.9252,
      "step": 39953
    },
    {
      "epoch": 3.4374946227307923,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000222391966633032,
      "loss": 0.7064,
      "step": 39954
    },
    {
      "epoch": 3.437580659038114,
      "grad_norm": 1.34375,
      "learning_rate": 0.00022236947709299037,
      "loss": 1.082,
      "step": 39955
    },
    {
      "epoch": 3.4376666953454356,
      "grad_norm": 1.015625,
      "learning_rate": 0.00022234698836496404,
      "loss": 0.9663,
      "step": 39956
    },
    {
      "epoch": 3.4377527316527576,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00022232450044901893,
      "loss": 1.009,
      "step": 39957
    },
    {
      "epoch": 3.4378387679600793,
      "grad_norm": 1.765625,
      "learning_rate": 0.00022230201334522048,
      "loss": 1.5312,
      "step": 39958
    },
    {
      "epoch": 3.437924804267401,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002222795270536348,
      "loss": 0.6682,
      "step": 39959
    },
    {
      "epoch": 3.4380108405747225,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002222570415743273,
      "loss": 1.489,
      "step": 39960
    },
    {
      "epoch": 3.438096876882044,
      "grad_norm": 4.78125,
      "learning_rate": 0.000222234556907364,
      "loss": 1.6595,
      "step": 39961
    },
    {
      "epoch": 3.4381829131893658,
      "grad_norm": 1.171875,
      "learning_rate": 0.00022221207305281071,
      "loss": 1.1983,
      "step": 39962
    },
    {
      "epoch": 3.4382689494966874,
      "grad_norm": 1.03125,
      "learning_rate": 0.00022218959001073292,
      "loss": 1.3415,
      "step": 39963
    },
    {
      "epoch": 3.4383549858040094,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00022216710778119654,
      "loss": 1.7984,
      "step": 39964
    },
    {
      "epoch": 3.438441022111331,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00022214462636426752,
      "loss": 0.7866,
      "step": 39965
    },
    {
      "epoch": 3.4385270584186527,
      "grad_norm": 0.921875,
      "learning_rate": 0.00022212214576001126,
      "loss": 1.1092,
      "step": 39966
    },
    {
      "epoch": 3.4386130947259743,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002220996659684937,
      "loss": 1.1249,
      "step": 39967
    },
    {
      "epoch": 3.438699131033296,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002220771869897807,
      "loss": 0.7286,
      "step": 39968
    },
    {
      "epoch": 3.4387851673406176,
      "grad_norm": 0.875,
      "learning_rate": 0.00022205470882393768,
      "loss": 0.76,
      "step": 39969
    },
    {
      "epoch": 3.438871203647939,
      "grad_norm": 9.0,
      "learning_rate": 0.00022203223147103062,
      "loss": 0.77,
      "step": 39970
    },
    {
      "epoch": 3.4389572399552613,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002220097549311254,
      "loss": 0.8123,
      "step": 39971
    },
    {
      "epoch": 3.439043276262583,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022198727920428742,
      "loss": 0.8121,
      "step": 39972
    },
    {
      "epoch": 3.4391293125699045,
      "grad_norm": 1.28125,
      "learning_rate": 0.00022196480429058273,
      "loss": 1.3217,
      "step": 39973
    },
    {
      "epoch": 3.439215348877226,
      "grad_norm": 0.875,
      "learning_rate": 0.00022194233019007676,
      "loss": 0.9425,
      "step": 39974
    },
    {
      "epoch": 3.4393013851845478,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00022191985690283534,
      "loss": 0.842,
      "step": 39975
    },
    {
      "epoch": 3.4393874214918694,
      "grad_norm": 1.703125,
      "learning_rate": 0.0002218973844289245,
      "loss": 1.0092,
      "step": 39976
    },
    {
      "epoch": 3.4394734577991914,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002218749127684095,
      "loss": 1.4818,
      "step": 39977
    },
    {
      "epoch": 3.439559494106513,
      "grad_norm": 1.125,
      "learning_rate": 0.00022185244192135628,
      "loss": 1.0492,
      "step": 39978
    },
    {
      "epoch": 3.4396455304138347,
      "grad_norm": 1.109375,
      "learning_rate": 0.00022182997188783076,
      "loss": 1.0248,
      "step": 39979
    },
    {
      "epoch": 3.4397315667211563,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002218075026678983,
      "loss": 1.1939,
      "step": 39980
    },
    {
      "epoch": 3.439817603028478,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00022178503426162477,
      "loss": 1.2577,
      "step": 39981
    },
    {
      "epoch": 3.4399036393357996,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00022176256666907613,
      "loss": 0.8195,
      "step": 39982
    },
    {
      "epoch": 3.4399896756431216,
      "grad_norm": 0.953125,
      "learning_rate": 0.00022174009989031767,
      "loss": 1.322,
      "step": 39983
    },
    {
      "epoch": 3.4400757119504433,
      "grad_norm": 1.34375,
      "learning_rate": 0.00022171763392541543,
      "loss": 0.7539,
      "step": 39984
    },
    {
      "epoch": 3.440161748257765,
      "grad_norm": 1.546875,
      "learning_rate": 0.00022169516877443485,
      "loss": 1.3093,
      "step": 39985
    },
    {
      "epoch": 3.4402477845650865,
      "grad_norm": 1.390625,
      "learning_rate": 0.0002216727044374418,
      "loss": 1.4754,
      "step": 39986
    },
    {
      "epoch": 3.440333820872408,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002216502409145021,
      "loss": 0.7012,
      "step": 39987
    },
    {
      "epoch": 3.4404198571797298,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00022162777820568114,
      "loss": 1.5501,
      "step": 39988
    },
    {
      "epoch": 3.4405058934870514,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002216053163110448,
      "loss": 0.4641,
      "step": 39989
    },
    {
      "epoch": 3.4405919297943734,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00022158285523065873,
      "loss": 0.9306,
      "step": 39990
    },
    {
      "epoch": 3.440677966101695,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002215603949645887,
      "loss": 1.1608,
      "step": 39991
    },
    {
      "epoch": 3.4407640024090167,
      "grad_norm": 1.53125,
      "learning_rate": 0.00022153793551290052,
      "loss": 1.5007,
      "step": 39992
    },
    {
      "epoch": 3.4408500387163383,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00022151547687565955,
      "loss": 1.2794,
      "step": 39993
    },
    {
      "epoch": 3.44093607502366,
      "grad_norm": 2.484375,
      "learning_rate": 0.0002214930190529316,
      "loss": 1.3153,
      "step": 39994
    },
    {
      "epoch": 3.4410221113309816,
      "grad_norm": 1.546875,
      "learning_rate": 0.00022147056204478266,
      "loss": 1.5133,
      "step": 39995
    },
    {
      "epoch": 3.441108147638303,
      "grad_norm": 0.96875,
      "learning_rate": 0.00022144810585127795,
      "loss": 0.551,
      "step": 39996
    },
    {
      "epoch": 3.4411941839456253,
      "grad_norm": 1.203125,
      "learning_rate": 0.00022142565047248336,
      "loss": 1.267,
      "step": 39997
    },
    {
      "epoch": 3.441280220252947,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002214031959084647,
      "loss": 0.9456,
      "step": 39998
    },
    {
      "epoch": 3.4413662565602685,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002213807421592874,
      "loss": 1.336,
      "step": 39999
    },
    {
      "epoch": 3.44145229286759,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00022135828922501734,
      "loss": 1.1227,
      "step": 40000
    },
    {
      "epoch": 3.4415383291749118,
      "grad_norm": 1.328125,
      "learning_rate": 0.00022133583710571997,
      "loss": 1.5723,
      "step": 40001
    },
    {
      "epoch": 3.4416243654822334,
      "grad_norm": 1.34375,
      "learning_rate": 0.00022131338580146104,
      "loss": 1.7284,
      "step": 40002
    },
    {
      "epoch": 3.441710401789555,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00022129093531230644,
      "loss": 0.5692,
      "step": 40003
    },
    {
      "epoch": 3.441796438096877,
      "grad_norm": 0.875,
      "learning_rate": 0.00022126848563832146,
      "loss": 1.0803,
      "step": 40004
    },
    {
      "epoch": 3.4418824744041987,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000221246036779572,
      "loss": 0.7274,
      "step": 40005
    },
    {
      "epoch": 3.4419685107115203,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002212235887361238,
      "loss": 1.5916,
      "step": 40006
    },
    {
      "epoch": 3.442054547018842,
      "grad_norm": 1.25,
      "learning_rate": 0.0002212011415080422,
      "loss": 1.216,
      "step": 40007
    },
    {
      "epoch": 3.4421405833261636,
      "grad_norm": 1.3125,
      "learning_rate": 0.00022117869509539307,
      "loss": 1.0763,
      "step": 40008
    },
    {
      "epoch": 3.442226619633485,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002211562494982422,
      "loss": 1.1678,
      "step": 40009
    },
    {
      "epoch": 3.442312655940807,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002211338047166549,
      "loss": 0.4056,
      "step": 40010
    },
    {
      "epoch": 3.442398692248129,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000221111360750697,
      "loss": 0.993,
      "step": 40011
    },
    {
      "epoch": 3.4424847285554505,
      "grad_norm": 1.34375,
      "learning_rate": 0.00022108891760043426,
      "loss": 1.2055,
      "step": 40012
    },
    {
      "epoch": 3.442570764862772,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00022106647526593204,
      "loss": 0.9493,
      "step": 40013
    },
    {
      "epoch": 3.4426568011700938,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00022104403374725628,
      "loss": 1.0329,
      "step": 40014
    },
    {
      "epoch": 3.4427428374774154,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00022102159304447235,
      "loss": 1.1828,
      "step": 40015
    },
    {
      "epoch": 3.442828873784737,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002209991531576459,
      "loss": 1.8389,
      "step": 40016
    },
    {
      "epoch": 3.4429149100920586,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00022097671408684294,
      "loss": 1.2899,
      "step": 40017
    },
    {
      "epoch": 3.4430009463993807,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022095427583212858,
      "loss": 1.3067,
      "step": 40018
    },
    {
      "epoch": 3.4430869827067023,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00022093183839356874,
      "loss": 1.1686,
      "step": 40019
    },
    {
      "epoch": 3.443173019014024,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00022090940177122914,
      "loss": 0.9889,
      "step": 40020
    },
    {
      "epoch": 3.4432590553213456,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0002208869659651751,
      "loss": 1.1105,
      "step": 40021
    },
    {
      "epoch": 3.443345091628667,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00022086453097547244,
      "loss": 1.1634,
      "step": 40022
    },
    {
      "epoch": 3.443431127935989,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00022084209680218686,
      "loss": 1.275,
      "step": 40023
    },
    {
      "epoch": 3.4435171642433104,
      "grad_norm": 1.40625,
      "learning_rate": 0.00022081966344538367,
      "loss": 1.5104,
      "step": 40024
    },
    {
      "epoch": 3.4436032005506325,
      "grad_norm": 1.140625,
      "learning_rate": 0.00022079723090512887,
      "loss": 1.1885,
      "step": 40025
    },
    {
      "epoch": 3.443689236857954,
      "grad_norm": 7.75,
      "learning_rate": 0.0002207747991814877,
      "loss": 1.946,
      "step": 40026
    },
    {
      "epoch": 3.4437752731652758,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002207523682745257,
      "loss": 1.1164,
      "step": 40027
    },
    {
      "epoch": 3.4438613094725974,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00022072993818430915,
      "loss": 1.1058,
      "step": 40028
    },
    {
      "epoch": 3.443947345779919,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00022070750891090297,
      "loss": 0.931,
      "step": 40029
    },
    {
      "epoch": 3.4440333820872406,
      "grad_norm": 1.046875,
      "learning_rate": 0.00022068508045437323,
      "loss": 0.7799,
      "step": 40030
    },
    {
      "epoch": 3.4441194183945623,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00022066265281478508,
      "loss": 0.9918,
      "step": 40031
    },
    {
      "epoch": 3.4442054547018843,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002206402259922044,
      "loss": 0.5795,
      "step": 40032
    },
    {
      "epoch": 3.444291491009206,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002206177999866969,
      "loss": 0.9449,
      "step": 40033
    },
    {
      "epoch": 3.4443775273165276,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00022059537479832776,
      "loss": 1.0174,
      "step": 40034
    },
    {
      "epoch": 3.444463563623849,
      "grad_norm": 94.0,
      "learning_rate": 0.00022057295042716285,
      "loss": 0.8791,
      "step": 40035
    },
    {
      "epoch": 3.444549599931171,
      "grad_norm": 31.25,
      "learning_rate": 0.00022055052687326788,
      "loss": 0.6572,
      "step": 40036
    },
    {
      "epoch": 3.444635636238493,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002205281041367081,
      "loss": 1.2739,
      "step": 40037
    },
    {
      "epoch": 3.4447216725458145,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002205056822175493,
      "loss": 0.8823,
      "step": 40038
    },
    {
      "epoch": 3.444807708853136,
      "grad_norm": 1.09375,
      "learning_rate": 0.00022048326111585715,
      "loss": 1.2591,
      "step": 40039
    },
    {
      "epoch": 3.4448937451604578,
      "grad_norm": 1.078125,
      "learning_rate": 0.00022046084083169693,
      "loss": 0.8782,
      "step": 40040
    },
    {
      "epoch": 3.4449797814677794,
      "grad_norm": 1.125,
      "learning_rate": 0.00022043842136513437,
      "loss": 1.6994,
      "step": 40041
    },
    {
      "epoch": 3.445065817775101,
      "grad_norm": 1.078125,
      "learning_rate": 0.00022041600271623525,
      "loss": 1.1599,
      "step": 40042
    },
    {
      "epoch": 3.4451518540824226,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00022039358488506477,
      "loss": 0.5835,
      "step": 40043
    },
    {
      "epoch": 3.4452378903897447,
      "grad_norm": 7.03125,
      "learning_rate": 0.00022037116787168882,
      "loss": 1.1905,
      "step": 40044
    },
    {
      "epoch": 3.4453239266970663,
      "grad_norm": 13.3125,
      "learning_rate": 0.00022034875167617268,
      "loss": 0.8494,
      "step": 40045
    },
    {
      "epoch": 3.445409963004388,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00022032633629858207,
      "loss": 1.3298,
      "step": 40046
    },
    {
      "epoch": 3.4454959993117096,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00022030392173898267,
      "loss": 0.7144,
      "step": 40047
    },
    {
      "epoch": 3.445582035619031,
      "grad_norm": 1.109375,
      "learning_rate": 0.00022028150799743973,
      "loss": 0.8261,
      "step": 40048
    },
    {
      "epoch": 3.445668071926353,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000220259095074019,
      "loss": 0.6541,
      "step": 40049
    },
    {
      "epoch": 3.4457541082336745,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00022023668296878614,
      "loss": 1.1684,
      "step": 40050
    },
    {
      "epoch": 3.4458401445409965,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002202142716818064,
      "loss": 0.8567,
      "step": 40051
    },
    {
      "epoch": 3.445926180848318,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00022019186121314556,
      "loss": 0.5575,
      "step": 40052
    },
    {
      "epoch": 3.4460122171556398,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00022016945156286922,
      "loss": 1.3496,
      "step": 40053
    },
    {
      "epoch": 3.4460982534629614,
      "grad_norm": 2.265625,
      "learning_rate": 0.00022014704273104268,
      "loss": 0.7328,
      "step": 40054
    },
    {
      "epoch": 3.446184289770283,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00022012463471773175,
      "loss": 0.6938,
      "step": 40055
    },
    {
      "epoch": 3.4462703260776046,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00022010222752300167,
      "loss": 1.0744,
      "step": 40056
    },
    {
      "epoch": 3.4463563623849263,
      "grad_norm": 9.5,
      "learning_rate": 0.00022007982114691815,
      "loss": 0.6737,
      "step": 40057
    },
    {
      "epoch": 3.4464423986922483,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002200574155895469,
      "loss": 1.0363,
      "step": 40058
    },
    {
      "epoch": 3.44652843499957,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00022003501085095302,
      "loss": 0.6349,
      "step": 40059
    },
    {
      "epoch": 3.4466144713068916,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00022001260693120235,
      "loss": 1.3303,
      "step": 40060
    },
    {
      "epoch": 3.446700507614213,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002199902038303605,
      "loss": 0.4764,
      "step": 40061
    },
    {
      "epoch": 3.446786543921535,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00021996780154849268,
      "loss": 1.0702,
      "step": 40062
    },
    {
      "epoch": 3.4468725802288565,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00021994540008566456,
      "loss": 1.2159,
      "step": 40063
    },
    {
      "epoch": 3.446958616536178,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00021992299944194167,
      "loss": 2.0275,
      "step": 40064
    },
    {
      "epoch": 3.4470446528435,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00021990059961738957,
      "loss": 1.2926,
      "step": 40065
    },
    {
      "epoch": 3.4471306891508218,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00021987820061207387,
      "loss": 1.7269,
      "step": 40066
    },
    {
      "epoch": 3.4472167254581434,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00021985580242605984,
      "loss": 0.9384,
      "step": 40067
    },
    {
      "epoch": 3.447302761765465,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00021983340505941302,
      "loss": 0.8467,
      "step": 40068
    },
    {
      "epoch": 3.4473887980727866,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00021981100851219926,
      "loss": 0.415,
      "step": 40069
    },
    {
      "epoch": 3.4474748343801083,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002197886127844836,
      "loss": 0.7743,
      "step": 40070
    },
    {
      "epoch": 3.44756087068743,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002197662178763319,
      "loss": 0.9583,
      "step": 40071
    },
    {
      "epoch": 3.447646906994752,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00021974382378780937,
      "loss": 0.8749,
      "step": 40072
    },
    {
      "epoch": 3.4477329433020736,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021972143051898164,
      "loss": 0.7682,
      "step": 40073
    },
    {
      "epoch": 3.447818979609395,
      "grad_norm": 356.0,
      "learning_rate": 0.00021969903806991437,
      "loss": 0.7513,
      "step": 40074
    },
    {
      "epoch": 3.447905015916717,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002196766464406727,
      "loss": 1.1494,
      "step": 40075
    },
    {
      "epoch": 3.4479910522240385,
      "grad_norm": 1.296875,
      "learning_rate": 0.00021965425563132235,
      "loss": 0.6627,
      "step": 40076
    },
    {
      "epoch": 3.44807708853136,
      "grad_norm": 1.0,
      "learning_rate": 0.00021963186564192899,
      "loss": 0.8878,
      "step": 40077
    },
    {
      "epoch": 3.4481631248386817,
      "grad_norm": 596.0,
      "learning_rate": 0.0002196094764725577,
      "loss": 1.2495,
      "step": 40078
    },
    {
      "epoch": 3.4482491611460038,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002195870881232741,
      "loss": 0.8629,
      "step": 40079
    },
    {
      "epoch": 3.4483351974533254,
      "grad_norm": 1.296875,
      "learning_rate": 0.00021956470059414395,
      "loss": 0.6067,
      "step": 40080
    },
    {
      "epoch": 3.448421233760647,
      "grad_norm": 0.875,
      "learning_rate": 0.00021954231388523237,
      "loss": 1.2704,
      "step": 40081
    },
    {
      "epoch": 3.4485072700679686,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00021951992799660493,
      "loss": 1.2825,
      "step": 40082
    },
    {
      "epoch": 3.4485933063752903,
      "grad_norm": 1.109375,
      "learning_rate": 0.00021949754292832735,
      "loss": 0.9,
      "step": 40083
    },
    {
      "epoch": 3.448679342682612,
      "grad_norm": 2.15625,
      "learning_rate": 0.0002194751586804647,
      "loss": 1.3614,
      "step": 40084
    },
    {
      "epoch": 3.4487653789899335,
      "grad_norm": 1.328125,
      "learning_rate": 0.00021945277525308284,
      "loss": 1.1439,
      "step": 40085
    },
    {
      "epoch": 3.4488514152972556,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00021943039264624687,
      "loss": 0.9559,
      "step": 40086
    },
    {
      "epoch": 3.448937451604577,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00021940801086002245,
      "loss": 0.8217,
      "step": 40087
    },
    {
      "epoch": 3.449023487911899,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002193856298944752,
      "loss": 1.0183,
      "step": 40088
    },
    {
      "epoch": 3.4491095242192205,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002193632497496702,
      "loss": 0.5813,
      "step": 40089
    },
    {
      "epoch": 3.449195560526542,
      "grad_norm": 7.78125,
      "learning_rate": 0.00021934087042567313,
      "loss": 0.9637,
      "step": 40090
    },
    {
      "epoch": 3.4492815968338637,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002193184919225496,
      "loss": 1.542,
      "step": 40091
    },
    {
      "epoch": 3.4493676331411858,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002192961142403647,
      "loss": 1.2353,
      "step": 40092
    },
    {
      "epoch": 3.4494536694485074,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00021927373737918416,
      "loss": 1.4498,
      "step": 40093
    },
    {
      "epoch": 3.449539705755829,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021925136133907341,
      "loss": 0.9606,
      "step": 40094
    },
    {
      "epoch": 3.4496257420631506,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021922898612009774,
      "loss": 1.0414,
      "step": 40095
    },
    {
      "epoch": 3.4497117783704723,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00021920661172232275,
      "loss": 1.3395,
      "step": 40096
    },
    {
      "epoch": 3.449797814677794,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021918423814581373,
      "loss": 1.2509,
      "step": 40097
    },
    {
      "epoch": 3.449883850985116,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00021916186539063616,
      "loss": 1.492,
      "step": 40098
    },
    {
      "epoch": 3.4499698872924376,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00021913949345685568,
      "loss": 1.1782,
      "step": 40099
    },
    {
      "epoch": 3.450055923599759,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00021911712234453734,
      "loss": 1.1021,
      "step": 40100
    },
    {
      "epoch": 3.450141959907081,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00021909475205374686,
      "loss": 1.3828,
      "step": 40101
    },
    {
      "epoch": 3.4502279962144025,
      "grad_norm": 2.640625,
      "learning_rate": 0.00021907238258454958,
      "loss": 1.6142,
      "step": 40102
    },
    {
      "epoch": 3.450314032521724,
      "grad_norm": 0.953125,
      "learning_rate": 0.0002190500139370109,
      "loss": 1.4989,
      "step": 40103
    },
    {
      "epoch": 3.4504000688290457,
      "grad_norm": 0.90625,
      "learning_rate": 0.00021902764611119653,
      "loss": 0.7665,
      "step": 40104
    },
    {
      "epoch": 3.4504861051363678,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00021900527910717144,
      "loss": 1.1229,
      "step": 40105
    },
    {
      "epoch": 3.4505721414436894,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00021898291292500127,
      "loss": 0.7385,
      "step": 40106
    },
    {
      "epoch": 3.450658177751011,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002189605475647516,
      "loss": 1.121,
      "step": 40107
    },
    {
      "epoch": 3.4507442140583326,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002189381830264875,
      "loss": 0.8393,
      "step": 40108
    },
    {
      "epoch": 3.4508302503656543,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002189158193102745,
      "loss": 1.2255,
      "step": 40109
    },
    {
      "epoch": 3.450916286672976,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002188934564161783,
      "loss": 0.592,
      "step": 40110
    },
    {
      "epoch": 3.4510023229802975,
      "grad_norm": 1.046875,
      "learning_rate": 0.00021887109434426383,
      "loss": 1.1168,
      "step": 40111
    },
    {
      "epoch": 3.4510883592876196,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021884873309459697,
      "loss": 0.7803,
      "step": 40112
    },
    {
      "epoch": 3.451174395594941,
      "grad_norm": 1.375,
      "learning_rate": 0.0002188263726672427,
      "loss": 0.8872,
      "step": 40113
    },
    {
      "epoch": 3.451260431902263,
      "grad_norm": 1.0,
      "learning_rate": 0.00021880401306226656,
      "loss": 1.0106,
      "step": 40114
    },
    {
      "epoch": 3.4513464682095845,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021878165427973418,
      "loss": 1.4944,
      "step": 40115
    },
    {
      "epoch": 3.451432504516906,
      "grad_norm": 1.15625,
      "learning_rate": 0.00021875929631971064,
      "loss": 1.2841,
      "step": 40116
    },
    {
      "epoch": 3.4515185408242277,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002187369391822614,
      "loss": 1.4111,
      "step": 40117
    },
    {
      "epoch": 3.4516045771315493,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00021871458286745212,
      "loss": 1.8314,
      "step": 40118
    },
    {
      "epoch": 3.4516906134388714,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00021869222737534778,
      "loss": 1.4951,
      "step": 40119
    },
    {
      "epoch": 3.451776649746193,
      "grad_norm": 2.65625,
      "learning_rate": 0.000218669872706014,
      "loss": 0.9969,
      "step": 40120
    },
    {
      "epoch": 3.4518626860535147,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002186475188595163,
      "loss": 1.2223,
      "step": 40121
    },
    {
      "epoch": 3.4519487223608363,
      "grad_norm": 3584.0,
      "learning_rate": 0.0002186251658359197,
      "loss": 1.5487,
      "step": 40122
    },
    {
      "epoch": 3.452034758668158,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002186028136352897,
      "loss": 0.8011,
      "step": 40123
    },
    {
      "epoch": 3.4521207949754795,
      "grad_norm": 1.203125,
      "learning_rate": 0.00021858046225769202,
      "loss": 0.6388,
      "step": 40124
    },
    {
      "epoch": 3.452206831282801,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0002185581117031915,
      "loss": 1.6554,
      "step": 40125
    },
    {
      "epoch": 3.452292867590123,
      "grad_norm": 4.65625,
      "learning_rate": 0.00021853576197185397,
      "loss": 1.1728,
      "step": 40126
    },
    {
      "epoch": 3.452378903897445,
      "grad_norm": 1.1875,
      "learning_rate": 0.00021851341306374444,
      "loss": 0.9935,
      "step": 40127
    },
    {
      "epoch": 3.4524649402047665,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002184910649789284,
      "loss": 1.3312,
      "step": 40128
    },
    {
      "epoch": 3.452550976512088,
      "grad_norm": 0.90625,
      "learning_rate": 0.00021846871771747146,
      "loss": 0.5737,
      "step": 40129
    },
    {
      "epoch": 3.4526370128194097,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021844637127943855,
      "loss": 0.8074,
      "step": 40130
    },
    {
      "epoch": 3.4527230491267313,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00021842402566489528,
      "loss": 0.9235,
      "step": 40131
    },
    {
      "epoch": 3.452809085434053,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002184016808739071,
      "loss": 1.1279,
      "step": 40132
    },
    {
      "epoch": 3.452895121741375,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00021837933690653912,
      "loss": 1.0836,
      "step": 40133
    },
    {
      "epoch": 3.4529811580486967,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002183569937628568,
      "loss": 0.9563,
      "step": 40134
    },
    {
      "epoch": 3.4530671943560183,
      "grad_norm": 1.8984375,
      "learning_rate": 0.00021833465144292563,
      "loss": 0.9665,
      "step": 40135
    },
    {
      "epoch": 3.45315323066334,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021831230994681067,
      "loss": 0.4854,
      "step": 40136
    },
    {
      "epoch": 3.4532392669706615,
      "grad_norm": 1.390625,
      "learning_rate": 0.00021828996927457755,
      "loss": 1.1063,
      "step": 40137
    },
    {
      "epoch": 3.453325303277983,
      "grad_norm": 4.40625,
      "learning_rate": 0.0002182676294262912,
      "loss": 1.1915,
      "step": 40138
    },
    {
      "epoch": 3.4534113395853048,
      "grad_norm": 4.0625,
      "learning_rate": 0.00021824529040201734,
      "loss": 1.2114,
      "step": 40139
    },
    {
      "epoch": 3.453497375892627,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00021822295220182147,
      "loss": 0.4278,
      "step": 40140
    },
    {
      "epoch": 3.4535834121999485,
      "grad_norm": 1.234375,
      "learning_rate": 0.0002182006148257684,
      "loss": 0.8142,
      "step": 40141
    },
    {
      "epoch": 3.45366944850727,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021817827827392395,
      "loss": 0.6218,
      "step": 40142
    },
    {
      "epoch": 3.4537554848145917,
      "grad_norm": 1.0625,
      "learning_rate": 0.00021815594254635307,
      "loss": 1.4719,
      "step": 40143
    },
    {
      "epoch": 3.4538415211219133,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00021813360764312117,
      "loss": 1.4745,
      "step": 40144
    },
    {
      "epoch": 3.453927557429235,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00021811127356429388,
      "loss": 1.5104,
      "step": 40145
    },
    {
      "epoch": 3.4540135937365566,
      "grad_norm": 1.765625,
      "learning_rate": 0.00021808894030993613,
      "loss": 1.6702,
      "step": 40146
    },
    {
      "epoch": 3.4540996300438787,
      "grad_norm": 1.484375,
      "learning_rate": 0.00021806660788011335,
      "loss": 1.1628,
      "step": 40147
    },
    {
      "epoch": 3.4541856663512003,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00021804427627489115,
      "loss": 0.8355,
      "step": 40148
    },
    {
      "epoch": 3.454271702658522,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00021802194549433435,
      "loss": 1.7629,
      "step": 40149
    },
    {
      "epoch": 3.4543577389658435,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00021799961553850856,
      "loss": 0.5914,
      "step": 40150
    },
    {
      "epoch": 3.454443775273165,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00021797728640747923,
      "loss": 1.3788,
      "step": 40151
    },
    {
      "epoch": 3.4545298115804868,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021795495810131134,
      "loss": 0.971,
      "step": 40152
    },
    {
      "epoch": 3.454615847887809,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021793263062007046,
      "loss": 0.9497,
      "step": 40153
    },
    {
      "epoch": 3.4547018841951305,
      "grad_norm": 1.09375,
      "learning_rate": 0.00021791030396382162,
      "loss": 0.9924,
      "step": 40154
    },
    {
      "epoch": 3.454787920502452,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00021788797813263028,
      "loss": 1.0652,
      "step": 40155
    },
    {
      "epoch": 3.4548739568097737,
      "grad_norm": 7.53125,
      "learning_rate": 0.00021786565312656192,
      "loss": 1.2775,
      "step": 40156
    },
    {
      "epoch": 3.4549599931170953,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00021784332894568144,
      "loss": 1.3682,
      "step": 40157
    },
    {
      "epoch": 3.455046029424417,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00021782100559005437,
      "loss": 0.832,
      "step": 40158
    },
    {
      "epoch": 3.455132065731739,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00021779868305974614,
      "loss": 1.2732,
      "step": 40159
    },
    {
      "epoch": 3.4552181020390607,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00021777636135482175,
      "loss": 1.0667,
      "step": 40160
    },
    {
      "epoch": 3.4553041383463823,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002177540404753466,
      "loss": 1.2474,
      "step": 40161
    },
    {
      "epoch": 3.455390174653704,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00021773172042138613,
      "loss": 0.5572,
      "step": 40162
    },
    {
      "epoch": 3.4554762109610255,
      "grad_norm": 2.8125,
      "learning_rate": 0.00021770940119300532,
      "loss": 0.7902,
      "step": 40163
    },
    {
      "epoch": 3.455562247268347,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002176870827902696,
      "loss": 0.5231,
      "step": 40164
    },
    {
      "epoch": 3.4556482835756688,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00021766476521324442,
      "loss": 0.7754,
      "step": 40165
    },
    {
      "epoch": 3.455734319882991,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00021764244846199477,
      "loss": 0.776,
      "step": 40166
    },
    {
      "epoch": 3.4558203561903125,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021762013253658614,
      "loss": 0.8936,
      "step": 40167
    },
    {
      "epoch": 3.455906392497634,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00021759781743708355,
      "loss": 0.7747,
      "step": 40168
    },
    {
      "epoch": 3.4559924288049557,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00021757550316355245,
      "loss": 0.8331,
      "step": 40169
    },
    {
      "epoch": 3.4560784651122773,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00021755318971605826,
      "loss": 1.6427,
      "step": 40170
    },
    {
      "epoch": 3.456164501419599,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021753087709466584,
      "loss": 0.9199,
      "step": 40171
    },
    {
      "epoch": 3.4562505377269206,
      "grad_norm": 1.1875,
      "learning_rate": 0.00021750856529944069,
      "loss": 1.2681,
      "step": 40172
    },
    {
      "epoch": 3.4563365740342427,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021748625433044827,
      "loss": 0.8752,
      "step": 40173
    },
    {
      "epoch": 3.4564226103415643,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00021746394418775333,
      "loss": 1.1939,
      "step": 40174
    },
    {
      "epoch": 3.456508646648886,
      "grad_norm": 1.53125,
      "learning_rate": 0.0002174416348714215,
      "loss": 0.9093,
      "step": 40175
    },
    {
      "epoch": 3.4565946829562075,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00021741932638151784,
      "loss": 2.1147,
      "step": 40176
    },
    {
      "epoch": 3.456680719263529,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021739701871810774,
      "loss": 0.8956,
      "step": 40177
    },
    {
      "epoch": 3.456766755570851,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00021737471188125663,
      "loss": 1.294,
      "step": 40178
    },
    {
      "epoch": 3.4568527918781724,
      "grad_norm": 1.09375,
      "learning_rate": 0.00021735240587102928,
      "loss": 1.3364,
      "step": 40179
    },
    {
      "epoch": 3.4569388281854945,
      "grad_norm": 1.015625,
      "learning_rate": 0.00021733010068749115,
      "loss": 0.8974,
      "step": 40180
    },
    {
      "epoch": 3.457024864492816,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00021730779633070775,
      "loss": 1.1068,
      "step": 40181
    },
    {
      "epoch": 3.4571109008001377,
      "grad_norm": 1.25,
      "learning_rate": 0.00021728549280074385,
      "loss": 1.0624,
      "step": 40182
    },
    {
      "epoch": 3.4571969371074593,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00021726319009766505,
      "loss": 1.6332,
      "step": 40183
    },
    {
      "epoch": 3.457282973414781,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00021724088822153625,
      "loss": 1.1703,
      "step": 40184
    },
    {
      "epoch": 3.4573690097221026,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00021721858717242289,
      "loss": 0.5131,
      "step": 40185
    },
    {
      "epoch": 3.457455046029424,
      "grad_norm": 1.546875,
      "learning_rate": 0.00021719628695039035,
      "loss": 1.3012,
      "step": 40186
    },
    {
      "epoch": 3.4575410823367463,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00021717398755550344,
      "loss": 1.4944,
      "step": 40187
    },
    {
      "epoch": 3.457627118644068,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00021715168898782766,
      "loss": 0.773,
      "step": 40188
    },
    {
      "epoch": 3.4577131549513895,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00021712939124742832,
      "loss": 1.6997,
      "step": 40189
    },
    {
      "epoch": 3.457799191258711,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00021710709433437036,
      "loss": 0.642,
      "step": 40190
    },
    {
      "epoch": 3.457885227566033,
      "grad_norm": 2.078125,
      "learning_rate": 0.00021708479824871906,
      "loss": 1.2906,
      "step": 40191
    },
    {
      "epoch": 3.4579712638733544,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00021706250299053997,
      "loss": 1.4439,
      "step": 40192
    },
    {
      "epoch": 3.458057300180676,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002170402085598978,
      "loss": 1.341,
      "step": 40193
    },
    {
      "epoch": 3.458143336487998,
      "grad_norm": 3.25,
      "learning_rate": 0.00021701791495685796,
      "loss": 0.9295,
      "step": 40194
    },
    {
      "epoch": 3.4582293727953197,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00021699562218148593,
      "loss": 0.6199,
      "step": 40195
    },
    {
      "epoch": 3.4583154091026413,
      "grad_norm": 1.15625,
      "learning_rate": 0.00021697333023384648,
      "loss": 0.9047,
      "step": 40196
    },
    {
      "epoch": 3.458401445409963,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002169510391140051,
      "loss": 1.2902,
      "step": 40197
    },
    {
      "epoch": 3.4584874817172846,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00021692874882202673,
      "loss": 1.3634,
      "step": 40198
    },
    {
      "epoch": 3.458573518024606,
      "grad_norm": 5.28125,
      "learning_rate": 0.00021690645935797676,
      "loss": 1.0496,
      "step": 40199
    },
    {
      "epoch": 3.458659554331928,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00021688417072192047,
      "loss": 1.283,
      "step": 40200
    },
    {
      "epoch": 3.45874559063925,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021686188291392277,
      "loss": 0.7523,
      "step": 40201
    },
    {
      "epoch": 3.4588316269465715,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00021683959593404896,
      "loss": 1.9697,
      "step": 40202
    },
    {
      "epoch": 3.458917663253893,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00021681730978236448,
      "loss": 0.912,
      "step": 40203
    },
    {
      "epoch": 3.459003699561215,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021679502445893408,
      "loss": 1.2185,
      "step": 40204
    },
    {
      "epoch": 3.4590897358685364,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00021677273996382318,
      "loss": 1.253,
      "step": 40205
    },
    {
      "epoch": 3.459175772175858,
      "grad_norm": 14.9375,
      "learning_rate": 0.00021675045629709712,
      "loss": 1.8121,
      "step": 40206
    },
    {
      "epoch": 3.4592618084831797,
      "grad_norm": 87.5,
      "learning_rate": 0.00021672817345882067,
      "loss": 0.987,
      "step": 40207
    },
    {
      "epoch": 3.4593478447905017,
      "grad_norm": 0.859375,
      "learning_rate": 0.00021670589144905938,
      "loss": 0.6665,
      "step": 40208
    },
    {
      "epoch": 3.4594338810978233,
      "grad_norm": 1.359375,
      "learning_rate": 0.00021668361026787813,
      "loss": 1.1482,
      "step": 40209
    },
    {
      "epoch": 3.459519917405145,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021666132991534222,
      "loss": 1.5734,
      "step": 40210
    },
    {
      "epoch": 3.4596059537124666,
      "grad_norm": 1.109375,
      "learning_rate": 0.00021663905039151694,
      "loss": 0.7421,
      "step": 40211
    },
    {
      "epoch": 3.459691990019788,
      "grad_norm": 30.0,
      "learning_rate": 0.00021661677169646705,
      "loss": 0.5899,
      "step": 40212
    },
    {
      "epoch": 3.4597780263271103,
      "grad_norm": 0.828125,
      "learning_rate": 0.00021659449383025825,
      "loss": 0.9476,
      "step": 40213
    },
    {
      "epoch": 3.459864062634432,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002165722167929553,
      "loss": 0.7306,
      "step": 40214
    },
    {
      "epoch": 3.4599500989417535,
      "grad_norm": 3.1875,
      "learning_rate": 0.00021654994058462352,
      "loss": 1.5366,
      "step": 40215
    },
    {
      "epoch": 3.460036135249075,
      "grad_norm": 1.078125,
      "learning_rate": 0.00021652766520532818,
      "loss": 1.4998,
      "step": 40216
    },
    {
      "epoch": 3.460122171556397,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00021650539065513414,
      "loss": 0.8196,
      "step": 40217
    },
    {
      "epoch": 3.4602082078637184,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00021648311693410667,
      "loss": 1.1326,
      "step": 40218
    },
    {
      "epoch": 3.46029424417104,
      "grad_norm": 1.109375,
      "learning_rate": 0.00021646084404231114,
      "loss": 1.3077,
      "step": 40219
    },
    {
      "epoch": 3.460380280478362,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002164385719798123,
      "loss": 1.0796,
      "step": 40220
    },
    {
      "epoch": 3.4604663167856837,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0002164163007466755,
      "loss": 1.0147,
      "step": 40221
    },
    {
      "epoch": 3.4605523530930054,
      "grad_norm": 1.28125,
      "learning_rate": 0.000216394030342966,
      "loss": 0.9998,
      "step": 40222
    },
    {
      "epoch": 3.460638389400327,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002163717607687487,
      "loss": 1.1191,
      "step": 40223
    },
    {
      "epoch": 3.4607244257076486,
      "grad_norm": 1.25,
      "learning_rate": 0.000216349492024089,
      "loss": 1.5969,
      "step": 40224
    },
    {
      "epoch": 3.4608104620149702,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021632722410905165,
      "loss": 1.1785,
      "step": 40225
    },
    {
      "epoch": 3.460896498322292,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00021630495702370202,
      "loss": 0.5986,
      "step": 40226
    },
    {
      "epoch": 3.460982534629614,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00021628269076810535,
      "loss": 0.9103,
      "step": 40227
    },
    {
      "epoch": 3.4610685709369355,
      "grad_norm": 1.296875,
      "learning_rate": 0.00021626042534232648,
      "loss": 1.5459,
      "step": 40228
    },
    {
      "epoch": 3.461154607244257,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002162381607464307,
      "loss": 1.0187,
      "step": 40229
    },
    {
      "epoch": 3.461240643551579,
      "grad_norm": 1.25,
      "learning_rate": 0.00021621589698048329,
      "loss": 1.5883,
      "step": 40230
    },
    {
      "epoch": 3.4613266798589004,
      "grad_norm": 25.5,
      "learning_rate": 0.000216193634044549,
      "loss": 1.0386,
      "step": 40231
    },
    {
      "epoch": 3.461412716166222,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00021617137193869312,
      "loss": 0.9452,
      "step": 40232
    },
    {
      "epoch": 3.4614987524735437,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00021614911066298098,
      "loss": 0.9122,
      "step": 40233
    },
    {
      "epoch": 3.4615847887808657,
      "grad_norm": 1.34375,
      "learning_rate": 0.0002161268502174773,
      "loss": 1.0912,
      "step": 40234
    },
    {
      "epoch": 3.4616708250881874,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021610459060224734,
      "loss": 0.8865,
      "step": 40235
    },
    {
      "epoch": 3.461756861395509,
      "grad_norm": 1.015625,
      "learning_rate": 0.00021608233181735648,
      "loss": 0.8563,
      "step": 40236
    },
    {
      "epoch": 3.4618428977028306,
      "grad_norm": 1.203125,
      "learning_rate": 0.00021606007386286937,
      "loss": 1.36,
      "step": 40237
    },
    {
      "epoch": 3.4619289340101522,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021603781673885147,
      "loss": 1.2698,
      "step": 40238
    },
    {
      "epoch": 3.462014970317474,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0002160155604453676,
      "loss": 1.1722,
      "step": 40239
    },
    {
      "epoch": 3.4621010066247955,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00021599330498248294,
      "loss": 0.7135,
      "step": 40240
    },
    {
      "epoch": 3.4621870429321175,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00021597105035026283,
      "loss": 0.8709,
      "step": 40241
    },
    {
      "epoch": 3.462273079239439,
      "grad_norm": 1.234375,
      "learning_rate": 0.00021594879654877197,
      "loss": 1.5501,
      "step": 40242
    },
    {
      "epoch": 3.462359115546761,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002159265435780756,
      "loss": 1.353,
      "step": 40243
    },
    {
      "epoch": 3.4624451518540824,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002159042914382391,
      "loss": 1.2728,
      "step": 40244
    },
    {
      "epoch": 3.462531188161404,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00021588204012932706,
      "loss": 1.0236,
      "step": 40245
    },
    {
      "epoch": 3.4626172244687257,
      "grad_norm": 1.578125,
      "learning_rate": 0.00021585978965140486,
      "loss": 1.6164,
      "step": 40246
    },
    {
      "epoch": 3.4627032607760473,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002158375400045376,
      "loss": 0.753,
      "step": 40247
    },
    {
      "epoch": 3.4627892970833694,
      "grad_norm": 1.53125,
      "learning_rate": 0.00021581529118879013,
      "loss": 1.1336,
      "step": 40248
    },
    {
      "epoch": 3.462875333390691,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00021579304320422767,
      "loss": 1.1055,
      "step": 40249
    },
    {
      "epoch": 3.4629613696980126,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00021577079605091527,
      "loss": 1.4891,
      "step": 40250
    },
    {
      "epoch": 3.4630474060053342,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000215748549728918,
      "loss": 0.6393,
      "step": 40251
    },
    {
      "epoch": 3.463133442312656,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00021572630423830115,
      "loss": 0.3813,
      "step": 40252
    },
    {
      "epoch": 3.4632194786199775,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00021570405957912936,
      "loss": 1.0171,
      "step": 40253
    },
    {
      "epoch": 3.463305514927299,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00021568181575146812,
      "loss": 1.3826,
      "step": 40254
    },
    {
      "epoch": 3.463391551234621,
      "grad_norm": 1.3125,
      "learning_rate": 0.00021565957275538207,
      "loss": 1.7982,
      "step": 40255
    },
    {
      "epoch": 3.463477587541943,
      "grad_norm": 0.875,
      "learning_rate": 0.00021563733059093648,
      "loss": 1.1747,
      "step": 40256
    },
    {
      "epoch": 3.4635636238492644,
      "grad_norm": 1.15625,
      "learning_rate": 0.00021561508925819657,
      "loss": 1.07,
      "step": 40257
    },
    {
      "epoch": 3.463649660156586,
      "grad_norm": 1.015625,
      "learning_rate": 0.000215592848757227,
      "loss": 1.4336,
      "step": 40258
    },
    {
      "epoch": 3.4637356964639077,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00021557060908809306,
      "loss": 0.6621,
      "step": 40259
    },
    {
      "epoch": 3.4638217327712293,
      "grad_norm": 1.609375,
      "learning_rate": 0.00021554837025086,
      "loss": 1.9766,
      "step": 40260
    },
    {
      "epoch": 3.463907769078551,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00021552613224559236,
      "loss": 0.5418,
      "step": 40261
    },
    {
      "epoch": 3.463993805385873,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002155038950723555,
      "loss": 0.8405,
      "step": 40262
    },
    {
      "epoch": 3.4640798416931946,
      "grad_norm": 2.453125,
      "learning_rate": 0.0002154816587312146,
      "loss": 1.793,
      "step": 40263
    },
    {
      "epoch": 3.4641658780005162,
      "grad_norm": 1.125,
      "learning_rate": 0.00021545942322223434,
      "loss": 0.8266,
      "step": 40264
    },
    {
      "epoch": 3.464251914307838,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002154371885454801,
      "loss": 0.6891,
      "step": 40265
    },
    {
      "epoch": 3.4643379506151595,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00021541495470101657,
      "loss": 1.3409,
      "step": 40266
    },
    {
      "epoch": 3.464423986922481,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002153927216889089,
      "loss": 1.5671,
      "step": 40267
    },
    {
      "epoch": 3.464510023229803,
      "grad_norm": 23.25,
      "learning_rate": 0.0002153704895092224,
      "loss": 1.4383,
      "step": 40268
    },
    {
      "epoch": 3.464596059537125,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00021534825816202163,
      "loss": 0.5209,
      "step": 40269
    },
    {
      "epoch": 3.4646820958444464,
      "grad_norm": 1.375,
      "learning_rate": 0.00021532602764737192,
      "loss": 2.016,
      "step": 40270
    },
    {
      "epoch": 3.464768132151768,
      "grad_norm": 35.25,
      "learning_rate": 0.00021530379796533833,
      "loss": 1.4046,
      "step": 40271
    },
    {
      "epoch": 3.4648541684590897,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00021528156911598562,
      "loss": 1.0547,
      "step": 40272
    },
    {
      "epoch": 3.4649402047664113,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002152593410993789,
      "loss": 0.8876,
      "step": 40273
    },
    {
      "epoch": 3.4650262410737334,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00021523711391558341,
      "loss": 1.3347,
      "step": 40274
    },
    {
      "epoch": 3.465112277381055,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021521488756466383,
      "loss": 0.2397,
      "step": 40275
    },
    {
      "epoch": 3.4651983136883766,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002151926620466853,
      "loss": 1.0396,
      "step": 40276
    },
    {
      "epoch": 3.4652843499956982,
      "grad_norm": 1.546875,
      "learning_rate": 0.00021517043736171305,
      "loss": 1.4875,
      "step": 40277
    },
    {
      "epoch": 3.46537038630302,
      "grad_norm": 0.765625,
      "learning_rate": 0.00021514821350981169,
      "loss": 0.5655,
      "step": 40278
    },
    {
      "epoch": 3.4654564226103415,
      "grad_norm": 1.734375,
      "learning_rate": 0.00021512599049104653,
      "loss": 0.9472,
      "step": 40279
    },
    {
      "epoch": 3.465542458917663,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00021510376830548235,
      "loss": 1.3489,
      "step": 40280
    },
    {
      "epoch": 3.465628495224985,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00021508154695318416,
      "loss": 0.923,
      "step": 40281
    },
    {
      "epoch": 3.465714531532307,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021505932643421722,
      "loss": 1.0639,
      "step": 40282
    },
    {
      "epoch": 3.4658005678396284,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021503710674864617,
      "loss": 1.4318,
      "step": 40283
    },
    {
      "epoch": 3.46588660414695,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00021501488789653616,
      "loss": 0.9842,
      "step": 40284
    },
    {
      "epoch": 3.4659726404542717,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002149926698779524,
      "loss": 1.8161,
      "step": 40285
    },
    {
      "epoch": 3.4660586767615933,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00021497045269295922,
      "loss": 0.8824,
      "step": 40286
    },
    {
      "epoch": 3.466144713068915,
      "grad_norm": 1.046875,
      "learning_rate": 0.00021494823634162242,
      "loss": 1.6436,
      "step": 40287
    },
    {
      "epoch": 3.466230749376237,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00021492602082400641,
      "loss": 0.7048,
      "step": 40288
    },
    {
      "epoch": 3.4663167856835586,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00021490380614017635,
      "loss": 0.737,
      "step": 40289
    },
    {
      "epoch": 3.4664028219908802,
      "grad_norm": 1.0625,
      "learning_rate": 0.00021488159229019737,
      "loss": 1.7055,
      "step": 40290
    },
    {
      "epoch": 3.466488858298202,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021485937927413412,
      "loss": 0.672,
      "step": 40291
    },
    {
      "epoch": 3.4665748946055235,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002148371670920517,
      "loss": 0.9915,
      "step": 40292
    },
    {
      "epoch": 3.466660930912845,
      "grad_norm": 6.15625,
      "learning_rate": 0.00021481495574401534,
      "loss": 1.2705,
      "step": 40293
    },
    {
      "epoch": 3.4667469672201667,
      "grad_norm": 0.96875,
      "learning_rate": 0.00021479274523008958,
      "loss": 1.1165,
      "step": 40294
    },
    {
      "epoch": 3.466833003527489,
      "grad_norm": 12.0,
      "learning_rate": 0.0002147705355503397,
      "loss": 1.4166,
      "step": 40295
    },
    {
      "epoch": 3.4669190398348104,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00021474832670483036,
      "loss": 0.681,
      "step": 40296
    },
    {
      "epoch": 3.467005076142132,
      "grad_norm": 1.125,
      "learning_rate": 0.00021472611869362675,
      "loss": 1.6981,
      "step": 40297
    },
    {
      "epoch": 3.4670911124494537,
      "grad_norm": 1.046875,
      "learning_rate": 0.00021470391151679386,
      "loss": 0.9188,
      "step": 40298
    },
    {
      "epoch": 3.4671771487567753,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002146817051743964,
      "loss": 1.0242,
      "step": 40299
    },
    {
      "epoch": 3.467263185064097,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00021465949966649944,
      "loss": 1.6986,
      "step": 40300
    },
    {
      "epoch": 3.4673492213714185,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00021463729499316813,
      "loss": 1.2156,
      "step": 40301
    },
    {
      "epoch": 3.4674352576787406,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002146150911544671,
      "loss": 0.5425,
      "step": 40302
    },
    {
      "epoch": 3.4675212939860622,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00021459288815046135,
      "loss": 0.9422,
      "step": 40303
    },
    {
      "epoch": 3.467607330293384,
      "grad_norm": 0.953125,
      "learning_rate": 0.00021457068598121614,
      "loss": 0.9522,
      "step": 40304
    },
    {
      "epoch": 3.4676933666007055,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00021454848464679593,
      "loss": 1.2869,
      "step": 40305
    },
    {
      "epoch": 3.467779402908027,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002145262841472661,
      "loss": 0.679,
      "step": 40306
    },
    {
      "epoch": 3.4678654392153487,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002145040844826912,
      "loss": 0.9325,
      "step": 40307
    },
    {
      "epoch": 3.4679514755226704,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021448188565313631,
      "loss": 1.3546,
      "step": 40308
    },
    {
      "epoch": 3.4680375118299924,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00021445968765866651,
      "loss": 0.8292,
      "step": 40309
    },
    {
      "epoch": 3.468123548137314,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00021443749049934646,
      "loss": 0.6397,
      "step": 40310
    },
    {
      "epoch": 3.4682095844446357,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002144152941752412,
      "loss": 0.825,
      "step": 40311
    },
    {
      "epoch": 3.4682956207519573,
      "grad_norm": 10.3125,
      "learning_rate": 0.00021439309868641582,
      "loss": 0.8471,
      "step": 40312
    },
    {
      "epoch": 3.468381657059279,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00021437090403293496,
      "loss": 1.0718,
      "step": 40313
    },
    {
      "epoch": 3.4684676933666005,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021434871021486357,
      "loss": 0.874,
      "step": 40314
    },
    {
      "epoch": 3.468553729673922,
      "grad_norm": 1.203125,
      "learning_rate": 0.00021432651723226687,
      "loss": 1.7062,
      "step": 40315
    },
    {
      "epoch": 3.4686397659812442,
      "grad_norm": 5.125,
      "learning_rate": 0.00021430432508520936,
      "loss": 1.2262,
      "step": 40316
    },
    {
      "epoch": 3.468725802288566,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00021428213377375617,
      "loss": 0.7105,
      "step": 40317
    },
    {
      "epoch": 3.4688118385958875,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021425994329797232,
      "loss": 0.634,
      "step": 40318
    },
    {
      "epoch": 3.468897874903209,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00021423775365792235,
      "loss": 0.9383,
      "step": 40319
    },
    {
      "epoch": 3.4689839112105307,
      "grad_norm": 1.25,
      "learning_rate": 0.00021421556485367162,
      "loss": 1.3578,
      "step": 40320
    },
    {
      "epoch": 3.4690699475178524,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00021419337688528456,
      "loss": 1.1739,
      "step": 40321
    },
    {
      "epoch": 3.469155983825174,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00021417118975282628,
      "loss": 0.7514,
      "step": 40322
    },
    {
      "epoch": 3.469242020132496,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00021414900345636167,
      "loss": 1.3756,
      "step": 40323
    },
    {
      "epoch": 3.4693280564398177,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00021412681799595567,
      "loss": 1.6334,
      "step": 40324
    },
    {
      "epoch": 3.4694140927471393,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002141046333716733,
      "loss": 0.9834,
      "step": 40325
    },
    {
      "epoch": 3.469500129054461,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002140824495835791,
      "loss": 0.8854,
      "step": 40326
    },
    {
      "epoch": 3.4695861653617825,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002140602666317381,
      "loss": 1.3127,
      "step": 40327
    },
    {
      "epoch": 3.469672201669104,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00021403808451621537,
      "loss": 1.1271,
      "step": 40328
    },
    {
      "epoch": 3.4697582379764262,
      "grad_norm": 1.078125,
      "learning_rate": 0.00021401590323707548,
      "loss": 0.7515,
      "step": 40329
    },
    {
      "epoch": 3.469844274283748,
      "grad_norm": 1.078125,
      "learning_rate": 0.00021399372279438345,
      "loss": 1.809,
      "step": 40330
    },
    {
      "epoch": 3.4699303105910695,
      "grad_norm": 6.40625,
      "learning_rate": 0.00021397154318820432,
      "loss": 1.0007,
      "step": 40331
    },
    {
      "epoch": 3.470016346898391,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021394936441860264,
      "loss": 1.3959,
      "step": 40332
    },
    {
      "epoch": 3.4701023832057127,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002139271864856434,
      "loss": 2.0256,
      "step": 40333
    },
    {
      "epoch": 3.4701884195130344,
      "grad_norm": 1.234375,
      "learning_rate": 0.00021390500938939174,
      "loss": 1.4863,
      "step": 40334
    },
    {
      "epoch": 3.4702744558203564,
      "grad_norm": 0.734375,
      "learning_rate": 0.00021388283312991207,
      "loss": 0.447,
      "step": 40335
    },
    {
      "epoch": 3.470360492127678,
      "grad_norm": 11.25,
      "learning_rate": 0.00021386065770726966,
      "loss": 1.2519,
      "step": 40336
    },
    {
      "epoch": 3.4704465284349997,
      "grad_norm": 1.53125,
      "learning_rate": 0.00021383848312152897,
      "loss": 1.4849,
      "step": 40337
    },
    {
      "epoch": 3.4705325647423213,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002138163093727551,
      "loss": 1.4025,
      "step": 40338
    },
    {
      "epoch": 3.470618601049643,
      "grad_norm": 1.15625,
      "learning_rate": 0.000213794136461013,
      "loss": 1.064,
      "step": 40339
    },
    {
      "epoch": 3.4707046373569646,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002137719643863672,
      "loss": 0.8391,
      "step": 40340
    },
    {
      "epoch": 3.470790673664286,
      "grad_norm": 0.796875,
      "learning_rate": 0.00021374979314888276,
      "loss": 1.1205,
      "step": 40341
    },
    {
      "epoch": 3.4708767099716082,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00021372762274862473,
      "loss": 1.7289,
      "step": 40342
    },
    {
      "epoch": 3.47096274627893,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00021370545318565747,
      "loss": 0.991,
      "step": 40343
    },
    {
      "epoch": 3.4710487825862515,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002136832844600461,
      "loss": 0.4844,
      "step": 40344
    },
    {
      "epoch": 3.471134818893573,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00021366111657185566,
      "loss": 0.6987,
      "step": 40345
    },
    {
      "epoch": 3.4712208552008947,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00021363894952115053,
      "loss": 0.8325,
      "step": 40346
    },
    {
      "epoch": 3.4713068915082164,
      "grad_norm": 1.1875,
      "learning_rate": 0.00021361678330799577,
      "loss": 1.0505,
      "step": 40347
    },
    {
      "epoch": 3.471392927815538,
      "grad_norm": 1.71875,
      "learning_rate": 0.00021359461793245643,
      "loss": 1.2965,
      "step": 40348
    },
    {
      "epoch": 3.47147896412286,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00021357245339459697,
      "loss": 1.2033,
      "step": 40349
    },
    {
      "epoch": 3.4715650004301817,
      "grad_norm": 0.953125,
      "learning_rate": 0.00021355028969448253,
      "loss": 0.9134,
      "step": 40350
    },
    {
      "epoch": 3.4716510367375033,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00021352812683217765,
      "loss": 1.262,
      "step": 40351
    },
    {
      "epoch": 3.471737073044825,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002135059648077472,
      "loss": 0.6061,
      "step": 40352
    },
    {
      "epoch": 3.4718231093521466,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002134838036212563,
      "loss": 1.2122,
      "step": 40353
    },
    {
      "epoch": 3.471909145659468,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002134616432727694,
      "loss": 1.1652,
      "step": 40354
    },
    {
      "epoch": 3.47199518196679,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00021343948376235144,
      "loss": 0.9895,
      "step": 40355
    },
    {
      "epoch": 3.472081218274112,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00021341732509006745,
      "loss": 1.2811,
      "step": 40356
    },
    {
      "epoch": 3.4721672545814335,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00021339516725598184,
      "loss": 0.5076,
      "step": 40357
    },
    {
      "epoch": 3.472253290888755,
      "grad_norm": 1.015625,
      "learning_rate": 0.00021337301026015966,
      "loss": 1.1116,
      "step": 40358
    },
    {
      "epoch": 3.4723393271960767,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002133508541026658,
      "loss": 1.083,
      "step": 40359
    },
    {
      "epoch": 3.4724253635033984,
      "grad_norm": 0.84375,
      "learning_rate": 0.00021332869878356464,
      "loss": 0.9487,
      "step": 40360
    },
    {
      "epoch": 3.47251139981072,
      "grad_norm": 0.921875,
      "learning_rate": 0.00021330654430292163,
      "loss": 1.0377,
      "step": 40361
    },
    {
      "epoch": 3.4725974361180416,
      "grad_norm": 1.375,
      "learning_rate": 0.00021328439066080102,
      "loss": 0.7926,
      "step": 40362
    },
    {
      "epoch": 3.4726834724253637,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00021326223785726777,
      "loss": 0.9247,
      "step": 40363
    },
    {
      "epoch": 3.4727695087326853,
      "grad_norm": 1.171875,
      "learning_rate": 0.00021324008589238696,
      "loss": 1.738,
      "step": 40364
    },
    {
      "epoch": 3.472855545040007,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00021321793476622292,
      "loss": 1.0048,
      "step": 40365
    },
    {
      "epoch": 3.4729415813473286,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002131957844788408,
      "loss": 0.7791,
      "step": 40366
    },
    {
      "epoch": 3.47302761765465,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00021317363503030507,
      "loss": 0.7178,
      "step": 40367
    },
    {
      "epoch": 3.473113653961972,
      "grad_norm": 1.59375,
      "learning_rate": 0.00021315148642068066,
      "loss": 1.3412,
      "step": 40368
    },
    {
      "epoch": 3.4731996902692934,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002131293386500326,
      "loss": 0.9158,
      "step": 40369
    },
    {
      "epoch": 3.4732857265766155,
      "grad_norm": 1.1875,
      "learning_rate": 0.00021310719171842518,
      "loss": 1.7082,
      "step": 40370
    },
    {
      "epoch": 3.473371762883937,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00021308504562592346,
      "loss": 1.2882,
      "step": 40371
    },
    {
      "epoch": 3.4734577991912587,
      "grad_norm": 1.625,
      "learning_rate": 0.00021306290037259235,
      "loss": 0.9459,
      "step": 40372
    },
    {
      "epoch": 3.4735438354985804,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00021304075595849625,
      "loss": 0.9537,
      "step": 40373
    },
    {
      "epoch": 3.473629871805902,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021301861238370019,
      "loss": 1.3103,
      "step": 40374
    },
    {
      "epoch": 3.4737159081132236,
      "grad_norm": 1.171875,
      "learning_rate": 0.00021299646964826902,
      "loss": 1.0624,
      "step": 40375
    },
    {
      "epoch": 3.4738019444205452,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00021297432775226716,
      "loss": 1.2804,
      "step": 40376
    },
    {
      "epoch": 3.4738879807278673,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021295218669575977,
      "loss": 0.8177,
      "step": 40377
    },
    {
      "epoch": 3.473974017035189,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002129300464788112,
      "loss": 0.4627,
      "step": 40378
    },
    {
      "epoch": 3.4740600533425106,
      "grad_norm": 1.125,
      "learning_rate": 0.00021290790710148645,
      "loss": 1.2787,
      "step": 40379
    },
    {
      "epoch": 3.474146089649832,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002128857685638504,
      "loss": 0.8725,
      "step": 40380
    },
    {
      "epoch": 3.474232125957154,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00021286363086596744,
      "loss": 0.7758,
      "step": 40381
    },
    {
      "epoch": 3.4743181622644754,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00021284149400790253,
      "loss": 1.4687,
      "step": 40382
    },
    {
      "epoch": 3.474404198571797,
      "grad_norm": 1.0,
      "learning_rate": 0.00021281935798972057,
      "loss": 1.6378,
      "step": 40383
    },
    {
      "epoch": 3.474490234879119,
      "grad_norm": 5.3125,
      "learning_rate": 0.000212797222811486,
      "loss": 1.2305,
      "step": 40384
    },
    {
      "epoch": 3.4745762711864407,
      "grad_norm": 1.21875,
      "learning_rate": 0.00021277508847326366,
      "loss": 0.7969,
      "step": 40385
    },
    {
      "epoch": 3.4746623074937624,
      "grad_norm": 1.40625,
      "learning_rate": 0.00021275295497511849,
      "loss": 1.2635,
      "step": 40386
    },
    {
      "epoch": 3.474748343801084,
      "grad_norm": 0.953125,
      "learning_rate": 0.00021273082231711488,
      "loss": 0.8169,
      "step": 40387
    },
    {
      "epoch": 3.4748343801084056,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00021270869049931773,
      "loss": 1.4361,
      "step": 40388
    },
    {
      "epoch": 3.4749204164157277,
      "grad_norm": 424.0,
      "learning_rate": 0.000212686559521792,
      "loss": 0.8655,
      "step": 40389
    },
    {
      "epoch": 3.4750064527230493,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000212664429384602,
      "loss": 1.1752,
      "step": 40390
    },
    {
      "epoch": 3.475092489030371,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002126423000878129,
      "loss": 0.8272,
      "step": 40391
    },
    {
      "epoch": 3.4751785253376926,
      "grad_norm": 1.3125,
      "learning_rate": 0.00021262017163148895,
      "loss": 1.0725,
      "step": 40392
    },
    {
      "epoch": 3.475264561645014,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002125980440156951,
      "loss": 1.2882,
      "step": 40393
    },
    {
      "epoch": 3.475350597952336,
      "grad_norm": 8.25,
      "learning_rate": 0.0002125759172404963,
      "loss": 1.0941,
      "step": 40394
    },
    {
      "epoch": 3.4754366342596574,
      "grad_norm": 1.25,
      "learning_rate": 0.0002125537913059568,
      "loss": 1.1412,
      "step": 40395
    },
    {
      "epoch": 3.4755226705669795,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002125316662121416,
      "loss": 0.8549,
      "step": 40396
    },
    {
      "epoch": 3.475608706874301,
      "grad_norm": 2.34375,
      "learning_rate": 0.00021250954195911538,
      "loss": 0.9984,
      "step": 40397
    },
    {
      "epoch": 3.4756947431816227,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021248741854694282,
      "loss": 1.4129,
      "step": 40398
    },
    {
      "epoch": 3.4757807794889444,
      "grad_norm": 1.15625,
      "learning_rate": 0.00021246529597568876,
      "loss": 0.8423,
      "step": 40399
    },
    {
      "epoch": 3.475866815796266,
      "grad_norm": 1.78125,
      "learning_rate": 0.0002124431742454177,
      "loss": 1.4607,
      "step": 40400
    },
    {
      "epoch": 3.4759528521035876,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00021242105335619437,
      "loss": 1.0175,
      "step": 40401
    },
    {
      "epoch": 3.4760388884109092,
      "grad_norm": 1904.0,
      "learning_rate": 0.00021239893330808374,
      "loss": 1.398,
      "step": 40402
    },
    {
      "epoch": 3.4761249247182313,
      "grad_norm": 1.328125,
      "learning_rate": 0.00021237681410115006,
      "loss": 1.0863,
      "step": 40403
    },
    {
      "epoch": 3.476210961025553,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00021235469573545836,
      "loss": 0.7225,
      "step": 40404
    },
    {
      "epoch": 3.4762969973328746,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002123325782110733,
      "loss": 1.199,
      "step": 40405
    },
    {
      "epoch": 3.476383033640196,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00021231046152805937,
      "loss": 1.3526,
      "step": 40406
    },
    {
      "epoch": 3.476469069947518,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002122883456864816,
      "loss": 1.2298,
      "step": 40407
    },
    {
      "epoch": 3.4765551062548394,
      "grad_norm": 1.484375,
      "learning_rate": 0.00021226623068640421,
      "loss": 0.5952,
      "step": 40408
    },
    {
      "epoch": 3.476641142562161,
      "grad_norm": 1.125,
      "learning_rate": 0.0002122441165278922,
      "loss": 1.78,
      "step": 40409
    },
    {
      "epoch": 3.476727178869483,
      "grad_norm": 1.109375,
      "learning_rate": 0.00021222200321101032,
      "loss": 1.4541,
      "step": 40410
    },
    {
      "epoch": 3.4768132151768047,
      "grad_norm": 1.265625,
      "learning_rate": 0.00021219989073582296,
      "loss": 1.1853,
      "step": 40411
    },
    {
      "epoch": 3.4768992514841264,
      "grad_norm": 1.96875,
      "learning_rate": 0.00021217777910239493,
      "loss": 0.7818,
      "step": 40412
    },
    {
      "epoch": 3.476985287791448,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00021215566831079115,
      "loss": 1.5766,
      "step": 40413
    },
    {
      "epoch": 3.4770713240987696,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00021213355836107578,
      "loss": 0.9816,
      "step": 40414
    },
    {
      "epoch": 3.4771573604060912,
      "grad_norm": 1.859375,
      "learning_rate": 0.00021211144925331382,
      "loss": 1.0158,
      "step": 40415
    },
    {
      "epoch": 3.477243396713413,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00021208934098757004,
      "loss": 0.7997,
      "step": 40416
    },
    {
      "epoch": 3.477329433020735,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00021206723356390877,
      "loss": 0.7601,
      "step": 40417
    },
    {
      "epoch": 3.4774154693280566,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002120451269823951,
      "loss": 1.7404,
      "step": 40418
    },
    {
      "epoch": 3.477501505635378,
      "grad_norm": 0.890625,
      "learning_rate": 0.00021202302124309313,
      "loss": 1.5253,
      "step": 40419
    },
    {
      "epoch": 3.4775875419427,
      "grad_norm": 1.46875,
      "learning_rate": 0.00021200091634606788,
      "loss": 1.463,
      "step": 40420
    },
    {
      "epoch": 3.4776735782500214,
      "grad_norm": 1.0,
      "learning_rate": 0.0002119788122913841,
      "loss": 1.6849,
      "step": 40421
    },
    {
      "epoch": 3.477759614557343,
      "grad_norm": 1.796875,
      "learning_rate": 0.00021195670907910607,
      "loss": 1.3786,
      "step": 40422
    },
    {
      "epoch": 3.4778456508646647,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002119346067092987,
      "loss": 1.2875,
      "step": 40423
    },
    {
      "epoch": 3.4779316871719868,
      "grad_norm": 1.75,
      "learning_rate": 0.00021191250518202672,
      "loss": 1.3518,
      "step": 40424
    },
    {
      "epoch": 3.4780177234793084,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00021189040449735448,
      "loss": 1.1752,
      "step": 40425
    },
    {
      "epoch": 3.47810375978663,
      "grad_norm": 0.796875,
      "learning_rate": 0.00021186830465534672,
      "loss": 1.2429,
      "step": 40426
    },
    {
      "epoch": 3.4781897960939516,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002118462056560684,
      "loss": 0.5404,
      "step": 40427
    },
    {
      "epoch": 3.4782758324012732,
      "grad_norm": 0.890625,
      "learning_rate": 0.00021182410749958364,
      "loss": 0.7064,
      "step": 40428
    },
    {
      "epoch": 3.478361868708595,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002118020101859573,
      "loss": 1.2026,
      "step": 40429
    },
    {
      "epoch": 3.4784479050159165,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00021177991371525425,
      "loss": 1.5636,
      "step": 40430
    },
    {
      "epoch": 3.4785339413232386,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00021175781808753868,
      "loss": 0.969,
      "step": 40431
    },
    {
      "epoch": 3.47861997763056,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00021173572330287566,
      "loss": 0.9876,
      "step": 40432
    },
    {
      "epoch": 3.478706013937882,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00021171362936132937,
      "loss": 0.8095,
      "step": 40433
    },
    {
      "epoch": 3.4787920502452034,
      "grad_norm": 1.125,
      "learning_rate": 0.0002116915362629645,
      "loss": 1.2807,
      "step": 40434
    },
    {
      "epoch": 3.478878086552525,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00021166944400784616,
      "loss": 0.873,
      "step": 40435
    },
    {
      "epoch": 3.4789641228598467,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002116473525960384,
      "loss": 0.7053,
      "step": 40436
    },
    {
      "epoch": 3.4790501591671683,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002116252620276063,
      "loss": 1.578,
      "step": 40437
    },
    {
      "epoch": 3.4791361954744904,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000211603172302614,
      "loss": 0.5989,
      "step": 40438
    },
    {
      "epoch": 3.479222231781812,
      "grad_norm": 1.125,
      "learning_rate": 0.00021158108342112636,
      "loss": 1.3284,
      "step": 40439
    },
    {
      "epoch": 3.4793082680891336,
      "grad_norm": 1.4375,
      "learning_rate": 0.0002115589953832081,
      "loss": 1.6164,
      "step": 40440
    },
    {
      "epoch": 3.4793943043964553,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00021153690818892352,
      "loss": 1.2776,
      "step": 40441
    },
    {
      "epoch": 3.479480340703777,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00021151482183833737,
      "loss": 1.0319,
      "step": 40442
    },
    {
      "epoch": 3.4795663770110985,
      "grad_norm": 1.203125,
      "learning_rate": 0.00021149273633151445,
      "loss": 1.1637,
      "step": 40443
    },
    {
      "epoch": 3.4796524133184206,
      "grad_norm": 1.078125,
      "learning_rate": 0.00021147065166851899,
      "loss": 1.2137,
      "step": 40444
    },
    {
      "epoch": 3.479738449625742,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00021144856784941575,
      "loss": 1.2346,
      "step": 40445
    },
    {
      "epoch": 3.479824485933064,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00021142648487426947,
      "loss": 1.1361,
      "step": 40446
    },
    {
      "epoch": 3.4799105222403854,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00021140440274314444,
      "loss": 1.6654,
      "step": 40447
    },
    {
      "epoch": 3.479996558547707,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021138232145610558,
      "loss": 0.6879,
      "step": 40448
    },
    {
      "epoch": 3.4800825948550287,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00021136024101321715,
      "loss": 1.1615,
      "step": 40449
    },
    {
      "epoch": 3.4801686311623508,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00021133816141454388,
      "loss": 1.2057,
      "step": 40450
    },
    {
      "epoch": 3.4802546674696724,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002113160826601505,
      "loss": 0.8414,
      "step": 40451
    },
    {
      "epoch": 3.480340703776994,
      "grad_norm": 8.0625,
      "learning_rate": 0.00021129400475010125,
      "loss": 0.699,
      "step": 40452
    },
    {
      "epoch": 3.4804267400843156,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002112719276844609,
      "loss": 1.3605,
      "step": 40453
    },
    {
      "epoch": 3.4805127763916373,
      "grad_norm": 1.015625,
      "learning_rate": 0.00021124985146329418,
      "loss": 1.2749,
      "step": 40454
    },
    {
      "epoch": 3.480598812698959,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002112277760866653,
      "loss": 0.44,
      "step": 40455
    },
    {
      "epoch": 3.4806848490062805,
      "grad_norm": 1.4375,
      "learning_rate": 0.00021120570155463902,
      "loss": 1.1562,
      "step": 40456
    },
    {
      "epoch": 3.4807708853136026,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00021118362786728007,
      "loss": 1.2166,
      "step": 40457
    },
    {
      "epoch": 3.480856921620924,
      "grad_norm": 11.625,
      "learning_rate": 0.00021116155502465267,
      "loss": 1.0238,
      "step": 40458
    },
    {
      "epoch": 3.480942957928246,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00021113948302682156,
      "loss": 1.7613,
      "step": 40459
    },
    {
      "epoch": 3.4810289942355674,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002111174118738514,
      "loss": 1.5095,
      "step": 40460
    },
    {
      "epoch": 3.481115030542889,
      "grad_norm": 1.203125,
      "learning_rate": 0.00021109534156580652,
      "loss": 1.5776,
      "step": 40461
    },
    {
      "epoch": 3.4812010668502107,
      "grad_norm": 2.03125,
      "learning_rate": 0.00021107327210275172,
      "loss": 1.0655,
      "step": 40462
    },
    {
      "epoch": 3.4812871031575323,
      "grad_norm": 1.328125,
      "learning_rate": 0.0002110512034847512,
      "loss": 1.6371,
      "step": 40463
    },
    {
      "epoch": 3.4813731394648544,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00021102913571186978,
      "loss": 1.3164,
      "step": 40464
    },
    {
      "epoch": 3.481459175772176,
      "grad_norm": 0.921875,
      "learning_rate": 0.00021100706878417202,
      "loss": 1.1771,
      "step": 40465
    },
    {
      "epoch": 3.4815452120794976,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00021098500270172227,
      "loss": 1.2655,
      "step": 40466
    },
    {
      "epoch": 3.4816312483868193,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00021096293746458516,
      "loss": 1.0453,
      "step": 40467
    },
    {
      "epoch": 3.481717284694141,
      "grad_norm": 1.53125,
      "learning_rate": 0.00021094087307282544,
      "loss": 0.9444,
      "step": 40468
    },
    {
      "epoch": 3.4818033210014625,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00021091880952650726,
      "loss": 1.2267,
      "step": 40469
    },
    {
      "epoch": 3.481889357308784,
      "grad_norm": 17.375,
      "learning_rate": 0.00021089674682569533,
      "loss": 0.6259,
      "step": 40470
    },
    {
      "epoch": 3.481975393616106,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002108746849704542,
      "loss": 1.1214,
      "step": 40471
    },
    {
      "epoch": 3.482061429923428,
      "grad_norm": 2.140625,
      "learning_rate": 0.00021085262396084836,
      "loss": 0.9649,
      "step": 40472
    },
    {
      "epoch": 3.4821474662307494,
      "grad_norm": 1.046875,
      "learning_rate": 0.00021083056379694255,
      "loss": 0.8531,
      "step": 40473
    },
    {
      "epoch": 3.482233502538071,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002108085044788009,
      "loss": 1.4839,
      "step": 40474
    },
    {
      "epoch": 3.4823195388453927,
      "grad_norm": 1.015625,
      "learning_rate": 0.00021078644600648817,
      "loss": 1.1881,
      "step": 40475
    },
    {
      "epoch": 3.4824055751527143,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000210764388380069,
      "loss": 0.5691,
      "step": 40476
    },
    {
      "epoch": 3.482491611460036,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00021074233159960754,
      "loss": 1.2781,
      "step": 40477
    },
    {
      "epoch": 3.482577647767358,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002107202756651687,
      "loss": 1.3907,
      "step": 40478
    },
    {
      "epoch": 3.4826636840746796,
      "grad_norm": 1.046875,
      "learning_rate": 0.00021069822057681665,
      "loss": 1.5648,
      "step": 40479
    },
    {
      "epoch": 3.4827497203820013,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000210676166334616,
      "loss": 0.8476,
      "step": 40480
    },
    {
      "epoch": 3.482835756689323,
      "grad_norm": 1.34375,
      "learning_rate": 0.00021065411293863146,
      "loss": 1.1143,
      "step": 40481
    },
    {
      "epoch": 3.4829217929966445,
      "grad_norm": 1.15625,
      "learning_rate": 0.00021063206038892724,
      "loss": 1.176,
      "step": 40482
    },
    {
      "epoch": 3.483007829303966,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002106100086855679,
      "loss": 1.2965,
      "step": 40483
    },
    {
      "epoch": 3.4830938656112878,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002105879578286182,
      "loss": 1.1526,
      "step": 40484
    },
    {
      "epoch": 3.48317990191861,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002105659078181423,
      "loss": 0.7188,
      "step": 40485
    },
    {
      "epoch": 3.4832659382259314,
      "grad_norm": 1.046875,
      "learning_rate": 0.00021054385865420478,
      "loss": 0.9728,
      "step": 40486
    },
    {
      "epoch": 3.483351974533253,
      "grad_norm": 1.0625,
      "learning_rate": 0.00021052181033687036,
      "loss": 0.7627,
      "step": 40487
    },
    {
      "epoch": 3.4834380108405747,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00021049976286620316,
      "loss": 0.7195,
      "step": 40488
    },
    {
      "epoch": 3.4835240471478963,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00021047771624226803,
      "loss": 1.2874,
      "step": 40489
    },
    {
      "epoch": 3.483610083455218,
      "grad_norm": 139.0,
      "learning_rate": 0.0002104556704651291,
      "loss": 1.3639,
      "step": 40490
    },
    {
      "epoch": 3.4836961197625396,
      "grad_norm": 1.125,
      "learning_rate": 0.00021043362553485102,
      "loss": 1.6436,
      "step": 40491
    },
    {
      "epoch": 3.4837821560698616,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00021041158145149847,
      "loss": 1.4891,
      "step": 40492
    },
    {
      "epoch": 3.4838681923771833,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002103895382151355,
      "loss": 1.1901,
      "step": 40493
    },
    {
      "epoch": 3.483954228684505,
      "grad_norm": 1.03125,
      "learning_rate": 0.00021036749582582686,
      "loss": 0.6213,
      "step": 40494
    },
    {
      "epoch": 3.4840402649918265,
      "grad_norm": 1.1875,
      "learning_rate": 0.0002103454542836371,
      "loss": 1.5386,
      "step": 40495
    },
    {
      "epoch": 3.484126301299148,
      "grad_norm": 27.0,
      "learning_rate": 0.00021032341358863038,
      "loss": 0.703,
      "step": 40496
    },
    {
      "epoch": 3.4842123376064698,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00021030137374087132,
      "loss": 0.5879,
      "step": 40497
    },
    {
      "epoch": 3.4842983739137914,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0002102793347404246,
      "loss": 0.7605,
      "step": 40498
    },
    {
      "epoch": 3.4843844102211134,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00021025729658735433,
      "loss": 0.9235,
      "step": 40499
    },
    {
      "epoch": 3.484470446528435,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00021023525928172505,
      "loss": 1.2506,
      "step": 40500
    },
    {
      "epoch": 3.4845564828357567,
      "grad_norm": 1.40625,
      "learning_rate": 0.0002102132228236015,
      "loss": 0.9399,
      "step": 40501
    },
    {
      "epoch": 3.4846425191430783,
      "grad_norm": 1.15625,
      "learning_rate": 0.00021019118721304775,
      "loss": 0.9555,
      "step": 40502
    },
    {
      "epoch": 3.4847285554504,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00021016915245012857,
      "loss": 1.6715,
      "step": 40503
    },
    {
      "epoch": 3.4848145917577216,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00021014711853490804,
      "loss": 0.4325,
      "step": 40504
    },
    {
      "epoch": 3.4849006280650436,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00021012508546745085,
      "loss": 0.988,
      "step": 40505
    },
    {
      "epoch": 3.4849866643723653,
      "grad_norm": 1.125,
      "learning_rate": 0.00021010305324782158,
      "loss": 1.1715,
      "step": 40506
    },
    {
      "epoch": 3.485072700679687,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00021008102187608412,
      "loss": 1.2482,
      "step": 40507
    },
    {
      "epoch": 3.4851587369870085,
      "grad_norm": 1.140625,
      "learning_rate": 0.00021005899135230366,
      "loss": 1.1185,
      "step": 40508
    },
    {
      "epoch": 3.48524477329433,
      "grad_norm": 1.5,
      "learning_rate": 0.00021003696167654412,
      "loss": 0.6709,
      "step": 40509
    },
    {
      "epoch": 3.4853308096016518,
      "grad_norm": 1.1875,
      "learning_rate": 0.00021001493284887002,
      "loss": 1.435,
      "step": 40510
    },
    {
      "epoch": 3.485416845908974,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020999290486934603,
      "loss": 1.2994,
      "step": 40511
    },
    {
      "epoch": 3.4855028822162955,
      "grad_norm": 5.53125,
      "learning_rate": 0.00020997087773803625,
      "loss": 1.1808,
      "step": 40512
    },
    {
      "epoch": 3.485588918523617,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002099488514550052,
      "loss": 0.3697,
      "step": 40513
    },
    {
      "epoch": 3.4856749548309387,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020992682602031753,
      "loss": 1.2832,
      "step": 40514
    },
    {
      "epoch": 3.4857609911382603,
      "grad_norm": 5.09375,
      "learning_rate": 0.00020990480143403734,
      "loss": 0.8977,
      "step": 40515
    },
    {
      "epoch": 3.485847027445582,
      "grad_norm": 1.015625,
      "learning_rate": 0.00020988277769622916,
      "loss": 1.0506,
      "step": 40516
    },
    {
      "epoch": 3.4859330637529036,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002098607548069576,
      "loss": 0.9946,
      "step": 40517
    },
    {
      "epoch": 3.4860191000602256,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020983873276628667,
      "loss": 0.8578,
      "step": 40518
    },
    {
      "epoch": 3.4861051363675473,
      "grad_norm": 1.015625,
      "learning_rate": 0.00020981671157428128,
      "loss": 0.9572,
      "step": 40519
    },
    {
      "epoch": 3.486191172674869,
      "grad_norm": 1.203125,
      "learning_rate": 0.00020979469123100537,
      "loss": 1.0667,
      "step": 40520
    },
    {
      "epoch": 3.4862772089821905,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00020977267173652353,
      "loss": 0.6685,
      "step": 40521
    },
    {
      "epoch": 3.486363245289512,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00020975065309090035,
      "loss": 0.8339,
      "step": 40522
    },
    {
      "epoch": 3.4864492815968338,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020972863529419993,
      "loss": 0.5606,
      "step": 40523
    },
    {
      "epoch": 3.4865353179041554,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020970661834648674,
      "loss": 0.7346,
      "step": 40524
    },
    {
      "epoch": 3.4866213542114775,
      "grad_norm": 1.09375,
      "learning_rate": 0.00020968460224782544,
      "loss": 0.8069,
      "step": 40525
    },
    {
      "epoch": 3.486707390518799,
      "grad_norm": 24.5,
      "learning_rate": 0.00020966258699828006,
      "loss": 0.929,
      "step": 40526
    },
    {
      "epoch": 3.4867934268261207,
      "grad_norm": 1.234375,
      "learning_rate": 0.00020964057259791508,
      "loss": 1.0846,
      "step": 40527
    },
    {
      "epoch": 3.4868794631334423,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00020961855904679517,
      "loss": 1.1425,
      "step": 40528
    },
    {
      "epoch": 3.486965499440764,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00020959654634498433,
      "loss": 0.9396,
      "step": 40529
    },
    {
      "epoch": 3.4870515357480856,
      "grad_norm": 1.375,
      "learning_rate": 0.00020957453449254728,
      "loss": 1.0331,
      "step": 40530
    },
    {
      "epoch": 3.487137572055407,
      "grad_norm": 5.90625,
      "learning_rate": 0.00020955252348954807,
      "loss": 1.161,
      "step": 40531
    },
    {
      "epoch": 3.4872236083627293,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00020953051333605122,
      "loss": 0.9943,
      "step": 40532
    },
    {
      "epoch": 3.487309644670051,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00020950850403212128,
      "loss": 0.9629,
      "step": 40533
    },
    {
      "epoch": 3.4873956809773725,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00020948649557782234,
      "loss": 0.7395,
      "step": 40534
    },
    {
      "epoch": 3.487481717284694,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00020946448797321888,
      "loss": 0.4121,
      "step": 40535
    },
    {
      "epoch": 3.4875677535920158,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00020944248121837546,
      "loss": 1.0358,
      "step": 40536
    },
    {
      "epoch": 3.4876537898993374,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00020942047531335607,
      "loss": 1.1093,
      "step": 40537
    },
    {
      "epoch": 3.487739826206659,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00020939847025822533,
      "loss": 0.7858,
      "step": 40538
    },
    {
      "epoch": 3.487825862513981,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00020937646605304766,
      "loss": 1.005,
      "step": 40539
    },
    {
      "epoch": 3.4879118988213027,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020935446269788717,
      "loss": 0.8384,
      "step": 40540
    },
    {
      "epoch": 3.4879979351286243,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00020933246019280832,
      "loss": 1.0902,
      "step": 40541
    },
    {
      "epoch": 3.488083971435946,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00020931045853787573,
      "loss": 0.4807,
      "step": 40542
    },
    {
      "epoch": 3.4881700077432676,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002092884577331533,
      "loss": 1.0882,
      "step": 40543
    },
    {
      "epoch": 3.488256044050589,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00020926645777870557,
      "loss": 0.8029,
      "step": 40544
    },
    {
      "epoch": 3.488342080357911,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0002092444586745969,
      "loss": 1.7199,
      "step": 40545
    },
    {
      "epoch": 3.488428116665233,
      "grad_norm": 1.65625,
      "learning_rate": 0.00020922246042089166,
      "loss": 1.9355,
      "step": 40546
    },
    {
      "epoch": 3.4885141529725545,
      "grad_norm": 1.0,
      "learning_rate": 0.0002092004630176544,
      "loss": 1.1317,
      "step": 40547
    },
    {
      "epoch": 3.488600189279876,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000209178466464949,
      "loss": 1.3664,
      "step": 40548
    },
    {
      "epoch": 3.4886862255871978,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00020915647076284017,
      "loss": 0.9862,
      "step": 40549
    },
    {
      "epoch": 3.4887722618945194,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00020913447591139195,
      "loss": 0.9407,
      "step": 40550
    },
    {
      "epoch": 3.488858298201841,
      "grad_norm": 1.234375,
      "learning_rate": 0.00020911248191066877,
      "loss": 0.8524,
      "step": 40551
    },
    {
      "epoch": 3.4889443345091626,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00020909048876073527,
      "loss": 1.2558,
      "step": 40552
    },
    {
      "epoch": 3.4890303708164847,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00020906849646165526,
      "loss": 0.7079,
      "step": 40553
    },
    {
      "epoch": 3.4891164071238063,
      "grad_norm": 1.03125,
      "learning_rate": 0.00020904650501349338,
      "loss": 0.7702,
      "step": 40554
    },
    {
      "epoch": 3.489202443431128,
      "grad_norm": 1.109375,
      "learning_rate": 0.00020902451441631403,
      "loss": 0.7264,
      "step": 40555
    },
    {
      "epoch": 3.4892884797384496,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002090025246701812,
      "loss": 0.9334,
      "step": 40556
    },
    {
      "epoch": 3.489374516045771,
      "grad_norm": 1.15625,
      "learning_rate": 0.00020898053577515942,
      "loss": 0.8111,
      "step": 40557
    },
    {
      "epoch": 3.489460552353093,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00020895854773131313,
      "loss": 0.8914,
      "step": 40558
    },
    {
      "epoch": 3.4895465886604145,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002089365605387063,
      "loss": 0.8618,
      "step": 40559
    },
    {
      "epoch": 3.4896326249677365,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002089145741974036,
      "loss": 1.751,
      "step": 40560
    },
    {
      "epoch": 3.489718661275058,
      "grad_norm": 4.21875,
      "learning_rate": 0.00020889258870746897,
      "loss": 0.7966,
      "step": 40561
    },
    {
      "epoch": 3.4898046975823798,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002088706040689669,
      "loss": 0.7661,
      "step": 40562
    },
    {
      "epoch": 3.4898907338897014,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0002088486202819619,
      "loss": 0.5341,
      "step": 40563
    },
    {
      "epoch": 3.489976770197023,
      "grad_norm": 2.078125,
      "learning_rate": 0.00020882663734651786,
      "loss": 0.8319,
      "step": 40564
    },
    {
      "epoch": 3.490062806504345,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002088046552626993,
      "loss": 1.2361,
      "step": 40565
    },
    {
      "epoch": 3.4901488428116667,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00020878267403057066,
      "loss": 0.9696,
      "step": 40566
    },
    {
      "epoch": 3.4902348791189883,
      "grad_norm": 1.328125,
      "learning_rate": 0.00020876069365019585,
      "loss": 1.5209,
      "step": 40567
    },
    {
      "epoch": 3.49032091542631,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002087387141216394,
      "loss": 1.3383,
      "step": 40568
    },
    {
      "epoch": 3.4904069517336316,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002087167354449657,
      "loss": 1.0238,
      "step": 40569
    },
    {
      "epoch": 3.490492988040953,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00020869475762023875,
      "loss": 1.6786,
      "step": 40570
    },
    {
      "epoch": 3.490579024348275,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002086727806475232,
      "loss": 1.5916,
      "step": 40571
    },
    {
      "epoch": 3.490665060655597,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00020865080452688283,
      "loss": 0.6321,
      "step": 40572
    },
    {
      "epoch": 3.4907510969629185,
      "grad_norm": 1.171875,
      "learning_rate": 0.00020862882925838227,
      "loss": 1.3176,
      "step": 40573
    },
    {
      "epoch": 3.49083713327024,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002086068548420859,
      "loss": 0.9566,
      "step": 40574
    },
    {
      "epoch": 3.4909231695775618,
      "grad_norm": 1.765625,
      "learning_rate": 0.00020858488127805758,
      "loss": 1.1832,
      "step": 40575
    },
    {
      "epoch": 3.4910092058848834,
      "grad_norm": 2.109375,
      "learning_rate": 0.00020856290856636185,
      "loss": 0.7947,
      "step": 40576
    },
    {
      "epoch": 3.491095242192205,
      "grad_norm": 1.109375,
      "learning_rate": 0.00020854093670706308,
      "loss": 0.8662,
      "step": 40577
    },
    {
      "epoch": 3.4911812784995266,
      "grad_norm": 6.59375,
      "learning_rate": 0.00020851896570022523,
      "loss": 1.337,
      "step": 40578
    },
    {
      "epoch": 3.4912673148068487,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020849699554591268,
      "loss": 0.5759,
      "step": 40579
    },
    {
      "epoch": 3.4913533511141703,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002084750262441899,
      "loss": 1.407,
      "step": 40580
    },
    {
      "epoch": 3.491439387421492,
      "grad_norm": 1.5,
      "learning_rate": 0.00020845305779512065,
      "loss": 1.2434,
      "step": 40581
    },
    {
      "epoch": 3.4915254237288136,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00020843109019876984,
      "loss": 0.7616,
      "step": 40582
    },
    {
      "epoch": 3.491611460036135,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002084091234552012,
      "loss": 0.8569,
      "step": 40583
    },
    {
      "epoch": 3.491697496343457,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0002083871575644792,
      "loss": 0.7325,
      "step": 40584
    },
    {
      "epoch": 3.4917835326507785,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00020836519252666813,
      "loss": 0.6435,
      "step": 40585
    },
    {
      "epoch": 3.4918695689581005,
      "grad_norm": 65.0,
      "learning_rate": 0.00020834322834183206,
      "loss": 1.1507,
      "step": 40586
    },
    {
      "epoch": 3.491955605265422,
      "grad_norm": 1.109375,
      "learning_rate": 0.0002083212650100353,
      "loss": 1.0591,
      "step": 40587
    },
    {
      "epoch": 3.4920416415727438,
      "grad_norm": 1.09375,
      "learning_rate": 0.00020829930253134227,
      "loss": 0.8411,
      "step": 40588
    },
    {
      "epoch": 3.4921276778800654,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002082773409058169,
      "loss": 1.1834,
      "step": 40589
    },
    {
      "epoch": 3.492213714187387,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00020825538013352374,
      "loss": 1.4954,
      "step": 40590
    },
    {
      "epoch": 3.4922997504947086,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002082334202145267,
      "loss": 1.2103,
      "step": 40591
    },
    {
      "epoch": 3.4923857868020303,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002082114611488901,
      "loss": 0.7416,
      "step": 40592
    },
    {
      "epoch": 3.4924718231093523,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00020818950293667848,
      "loss": 1.5177,
      "step": 40593
    },
    {
      "epoch": 3.492557859416674,
      "grad_norm": 1.03125,
      "learning_rate": 0.00020816754557795558,
      "loss": 1.1642,
      "step": 40594
    },
    {
      "epoch": 3.4926438957239956,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00020814558907278585,
      "loss": 1.7684,
      "step": 40595
    },
    {
      "epoch": 3.492729932031317,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002081236334212337,
      "loss": 0.7763,
      "step": 40596
    },
    {
      "epoch": 3.492815968338639,
      "grad_norm": 0.921875,
      "learning_rate": 0.000208101678623363,
      "loss": 0.7191,
      "step": 40597
    },
    {
      "epoch": 3.4929020046459605,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020807972467923807,
      "loss": 0.6326,
      "step": 40598
    },
    {
      "epoch": 3.492988040953282,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00020805777158892341,
      "loss": 0.6065,
      "step": 40599
    },
    {
      "epoch": 3.493074077260604,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00020803581935248278,
      "loss": 1.2356,
      "step": 40600
    },
    {
      "epoch": 3.4931601135679258,
      "grad_norm": 5.125,
      "learning_rate": 0.00020801386796998072,
      "loss": 1.3888,
      "step": 40601
    },
    {
      "epoch": 3.4932461498752474,
      "grad_norm": 1.25,
      "learning_rate": 0.00020799191744148122,
      "loss": 1.4833,
      "step": 40602
    },
    {
      "epoch": 3.493332186182569,
      "grad_norm": 2.203125,
      "learning_rate": 0.0002079699677670485,
      "loss": 1.3778,
      "step": 40603
    },
    {
      "epoch": 3.4934182224898906,
      "grad_norm": 1.046875,
      "learning_rate": 0.00020794801894674704,
      "loss": 0.7638,
      "step": 40604
    },
    {
      "epoch": 3.4935042587972123,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00020792607098064065,
      "loss": 0.9204,
      "step": 40605
    },
    {
      "epoch": 3.493590295104534,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00020790412386879366,
      "loss": 1.1069,
      "step": 40606
    },
    {
      "epoch": 3.493676331411856,
      "grad_norm": 1.03125,
      "learning_rate": 0.00020788217761127055,
      "loss": 0.5732,
      "step": 40607
    },
    {
      "epoch": 3.4937623677191776,
      "grad_norm": 1.03125,
      "learning_rate": 0.000207860232208135,
      "loss": 1.5532,
      "step": 40608
    },
    {
      "epoch": 3.493848404026499,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00020783828765945145,
      "loss": 1.3238,
      "step": 40609
    },
    {
      "epoch": 3.493934440333821,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0002078163439652843,
      "loss": 1.3672,
      "step": 40610
    },
    {
      "epoch": 3.4940204766411425,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00020779440112569735,
      "loss": 0.757,
      "step": 40611
    },
    {
      "epoch": 3.494106512948464,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00020777245914075493,
      "loss": 1.3841,
      "step": 40612
    },
    {
      "epoch": 3.4941925492557857,
      "grad_norm": 1.1875,
      "learning_rate": 0.00020775051801052142,
      "loss": 0.9994,
      "step": 40613
    },
    {
      "epoch": 3.4942785855631078,
      "grad_norm": 1.0,
      "learning_rate": 0.0002077285777350606,
      "loss": 0.8228,
      "step": 40614
    },
    {
      "epoch": 3.4943646218704294,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00020770663831443708,
      "loss": 1.2508,
      "step": 40615
    },
    {
      "epoch": 3.494450658177751,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00020768469974871456,
      "loss": 1.4165,
      "step": 40616
    },
    {
      "epoch": 3.4945366944850726,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00020766276203795748,
      "loss": 1.4142,
      "step": 40617
    },
    {
      "epoch": 3.4946227307923943,
      "grad_norm": 1.21875,
      "learning_rate": 0.00020764082518222993,
      "loss": 0.8777,
      "step": 40618
    },
    {
      "epoch": 3.494708767099716,
      "grad_norm": 1.375,
      "learning_rate": 0.0002076188891815961,
      "loss": 1.0409,
      "step": 40619
    },
    {
      "epoch": 3.494794803407038,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002075969540361204,
      "loss": 0.9375,
      "step": 40620
    },
    {
      "epoch": 3.4948808397143596,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002075750197458665,
      "loss": 1.5682,
      "step": 40621
    },
    {
      "epoch": 3.494966876021681,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002075530863108988,
      "loss": 0.97,
      "step": 40622
    },
    {
      "epoch": 3.495052912329003,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002075311537312816,
      "loss": 0.7391,
      "step": 40623
    },
    {
      "epoch": 3.4951389486363245,
      "grad_norm": 1.3125,
      "learning_rate": 0.00020750922200707877,
      "loss": 1.0578,
      "step": 40624
    },
    {
      "epoch": 3.495224984943646,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00020748729113835457,
      "loss": 0.6525,
      "step": 40625
    },
    {
      "epoch": 3.495311021250968,
      "grad_norm": 1.046875,
      "learning_rate": 0.00020746536112517332,
      "loss": 1.5159,
      "step": 40626
    },
    {
      "epoch": 3.4953970575582898,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002074434319675988,
      "loss": 1.6285,
      "step": 40627
    },
    {
      "epoch": 3.4954830938656114,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002074215036656954,
      "loss": 1.2283,
      "step": 40628
    },
    {
      "epoch": 3.495569130172933,
      "grad_norm": 1.921875,
      "learning_rate": 0.00020739957621952732,
      "loss": 1.2326,
      "step": 40629
    },
    {
      "epoch": 3.4956551664802546,
      "grad_norm": 17.5,
      "learning_rate": 0.0002073776496291584,
      "loss": 1.5451,
      "step": 40630
    },
    {
      "epoch": 3.4957412027875763,
      "grad_norm": 10.125,
      "learning_rate": 0.00020735572389465312,
      "loss": 0.7867,
      "step": 40631
    },
    {
      "epoch": 3.495827239094898,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00020733379901607523,
      "loss": 0.866,
      "step": 40632
    },
    {
      "epoch": 3.49591327540222,
      "grad_norm": 0.953125,
      "learning_rate": 0.00020731187499348913,
      "loss": 0.674,
      "step": 40633
    },
    {
      "epoch": 3.4959993117095416,
      "grad_norm": 1.125,
      "learning_rate": 0.00020728995182695898,
      "loss": 1.3797,
      "step": 40634
    },
    {
      "epoch": 3.496085348016863,
      "grad_norm": 1.078125,
      "learning_rate": 0.00020726802951654866,
      "loss": 0.849,
      "step": 40635
    },
    {
      "epoch": 3.496171384324185,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00020724610806232242,
      "loss": 1.2161,
      "step": 40636
    },
    {
      "epoch": 3.4962574206315065,
      "grad_norm": 1.03125,
      "learning_rate": 0.00020722418746434452,
      "loss": 0.5021,
      "step": 40637
    },
    {
      "epoch": 3.496343456938828,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00020720226772267874,
      "loss": 1.0089,
      "step": 40638
    },
    {
      "epoch": 3.4964294932461497,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00020718034883738945,
      "loss": 0.9407,
      "step": 40639
    },
    {
      "epoch": 3.496515529553472,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002071584308085408,
      "loss": 0.7973,
      "step": 40640
    },
    {
      "epoch": 3.4966015658607934,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00020713651363619663,
      "loss": 0.9198,
      "step": 40641
    },
    {
      "epoch": 3.496687602168115,
      "grad_norm": 1.078125,
      "learning_rate": 0.00020711459732042142,
      "loss": 1.5473,
      "step": 40642
    },
    {
      "epoch": 3.4967736384754367,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00020709268186127884,
      "loss": 1.2859,
      "step": 40643
    },
    {
      "epoch": 3.4968596747827583,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00020707076725883316,
      "loss": 1.2765,
      "step": 40644
    },
    {
      "epoch": 3.49694571109008,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00020704885351314873,
      "loss": 1.4726,
      "step": 40645
    },
    {
      "epoch": 3.4970317473974015,
      "grad_norm": 1.046875,
      "learning_rate": 0.00020702694062428923,
      "loss": 1.8987,
      "step": 40646
    },
    {
      "epoch": 3.4971177837047236,
      "grad_norm": 1.046875,
      "learning_rate": 0.00020700502859231896,
      "loss": 0.9916,
      "step": 40647
    },
    {
      "epoch": 3.497203820012045,
      "grad_norm": 6.40625,
      "learning_rate": 0.0002069831174173022,
      "loss": 1.2201,
      "step": 40648
    },
    {
      "epoch": 3.497289856319367,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002069612070993026,
      "loss": 1.0152,
      "step": 40649
    },
    {
      "epoch": 3.4973758926266885,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00020693929763838454,
      "loss": 1.0799,
      "step": 40650
    },
    {
      "epoch": 3.49746192893401,
      "grad_norm": 1.140625,
      "learning_rate": 0.00020691738903461215,
      "loss": 1.7455,
      "step": 40651
    },
    {
      "epoch": 3.4975479652413317,
      "grad_norm": 63.0,
      "learning_rate": 0.00020689548128804925,
      "loss": 1.4891,
      "step": 40652
    },
    {
      "epoch": 3.4976340015486533,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00020687357439876008,
      "loss": 0.7512,
      "step": 40653
    },
    {
      "epoch": 3.4977200378559754,
      "grad_norm": 1.234375,
      "learning_rate": 0.00020685166836680887,
      "loss": 1.1321,
      "step": 40654
    },
    {
      "epoch": 3.497806074163297,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020682976319225915,
      "loss": 1.2633,
      "step": 40655
    },
    {
      "epoch": 3.4978921104706187,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00020680785887517567,
      "loss": 1.0473,
      "step": 40656
    },
    {
      "epoch": 3.4979781467779403,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00020678595541562207,
      "loss": 1.5782,
      "step": 40657
    },
    {
      "epoch": 3.498064183085262,
      "grad_norm": 1.265625,
      "learning_rate": 0.00020676405281366266,
      "loss": 0.8885,
      "step": 40658
    },
    {
      "epoch": 3.4981502193925835,
      "grad_norm": 4.5625,
      "learning_rate": 0.00020674215106936117,
      "loss": 0.9374,
      "step": 40659
    },
    {
      "epoch": 3.498236255699905,
      "grad_norm": 4.125,
      "learning_rate": 0.0002067202501827819,
      "loss": 1.0076,
      "step": 40660
    },
    {
      "epoch": 3.498322292007227,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00020669835015398903,
      "loss": 0.7914,
      "step": 40661
    },
    {
      "epoch": 3.498408328314549,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00020667645098304622,
      "loss": 1.3319,
      "step": 40662
    },
    {
      "epoch": 3.4984943646218705,
      "grad_norm": 0.96875,
      "learning_rate": 0.00020665455267001776,
      "loss": 1.4297,
      "step": 40663
    },
    {
      "epoch": 3.498580400929192,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002066326552149679,
      "loss": 0.5022,
      "step": 40664
    },
    {
      "epoch": 3.4986664372365137,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00020661075861796025,
      "loss": 0.7848,
      "step": 40665
    },
    {
      "epoch": 3.4987524735438353,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002065888628790591,
      "loss": 1.4131,
      "step": 40666
    },
    {
      "epoch": 3.498838509851157,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002065669679983286,
      "loss": 1.0998,
      "step": 40667
    },
    {
      "epoch": 3.498924546158479,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002065450739758325,
      "loss": 1.564,
      "step": 40668
    },
    {
      "epoch": 3.4990105824658007,
      "grad_norm": 75.5,
      "learning_rate": 0.00020652318081163497,
      "loss": 1.0392,
      "step": 40669
    },
    {
      "epoch": 3.4990966187731223,
      "grad_norm": 1.21875,
      "learning_rate": 0.00020650128850580024,
      "loss": 1.0413,
      "step": 40670
    },
    {
      "epoch": 3.499182655080444,
      "grad_norm": 1.59375,
      "learning_rate": 0.00020647939705839197,
      "loss": 1.6609,
      "step": 40671
    },
    {
      "epoch": 3.4992686913877655,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00020645750646947452,
      "loss": 1.5371,
      "step": 40672
    },
    {
      "epoch": 3.499354727695087,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00020643561673911164,
      "loss": 0.5556,
      "step": 40673
    },
    {
      "epoch": 3.4994407640024088,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002064137278673675,
      "loss": 0.919,
      "step": 40674
    },
    {
      "epoch": 3.499526800309731,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020639183985430627,
      "loss": 0.7547,
      "step": 40675
    },
    {
      "epoch": 3.4996128366170525,
      "grad_norm": 1.109375,
      "learning_rate": 0.00020636995269999165,
      "loss": 0.5905,
      "step": 40676
    },
    {
      "epoch": 3.499698872924374,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020634806640448778,
      "loss": 0.716,
      "step": 40677
    },
    {
      "epoch": 3.4997849092316957,
      "grad_norm": 3.859375,
      "learning_rate": 0.00020632618096785887,
      "loss": 0.8267,
      "step": 40678
    },
    {
      "epoch": 3.4998709455390173,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00020630429639016863,
      "loss": 1.391,
      "step": 40679
    },
    {
      "epoch": 3.499956981846339,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00020628241267148118,
      "loss": 0.7492,
      "step": 40680
    },
    {
      "epoch": 3.5000430181536606,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00020626052981186072,
      "loss": 1.2967,
      "step": 40681
    },
    {
      "epoch": 3.5001290544609827,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00020623864781137093,
      "loss": 0.4351,
      "step": 40682
    },
    {
      "epoch": 3.5002150907683043,
      "grad_norm": 1.859375,
      "learning_rate": 0.00020621676667007612,
      "loss": 1.1941,
      "step": 40683
    },
    {
      "epoch": 3.500301127075626,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002061948863880399,
      "loss": 0.6839,
      "step": 40684
    },
    {
      "epoch": 3.500301127075626,
      "eval_loss": 1.0804526805877686,
      "eval_runtime": 353.8343,
      "eval_samples_per_second": 13.139,
      "eval_steps_per_second": 1.645,
      "step": 40684
    },
    {
      "epoch": 3.5003871633829475,
      "grad_norm": 0.890625,
      "learning_rate": 0.00020617300696532654,
      "loss": 0.6706,
      "step": 40685
    },
    {
      "epoch": 3.500473199690269,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00020615112840200017,
      "loss": 1.1563,
      "step": 40686
    },
    {
      "epoch": 3.5005592359975912,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020612925069812444,
      "loss": 0.3966,
      "step": 40687
    },
    {
      "epoch": 3.5006452723049124,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00020610737385376348,
      "loss": 0.7379,
      "step": 40688
    },
    {
      "epoch": 3.5007313086122345,
      "grad_norm": 1.546875,
      "learning_rate": 0.00020608549786898145,
      "loss": 1.6405,
      "step": 40689
    },
    {
      "epoch": 3.500817344919556,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00020606362274384204,
      "loss": 0.9898,
      "step": 40690
    },
    {
      "epoch": 3.5009033812268777,
      "grad_norm": 4.5,
      "learning_rate": 0.00020604174847840934,
      "loss": 1.5022,
      "step": 40691
    },
    {
      "epoch": 3.5009894175341993,
      "grad_norm": 1.015625,
      "learning_rate": 0.00020601987507274733,
      "loss": 1.261,
      "step": 40692
    },
    {
      "epoch": 3.501075453841521,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020599800252692001,
      "loss": 1.5322,
      "step": 40693
    },
    {
      "epoch": 3.501161490148843,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002059761308409916,
      "loss": 1.3794,
      "step": 40694
    },
    {
      "epoch": 3.5012475264561647,
      "grad_norm": 1.125,
      "learning_rate": 0.0002059542600150256,
      "loss": 0.9526,
      "step": 40695
    },
    {
      "epoch": 3.5013335627634863,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00020593239004908622,
      "loss": 0.8303,
      "step": 40696
    },
    {
      "epoch": 3.501419599070808,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00020591052094323753,
      "loss": 0.6819,
      "step": 40697
    },
    {
      "epoch": 3.5015056353781295,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00020588865269754325,
      "loss": 1.6265,
      "step": 40698
    },
    {
      "epoch": 3.501591671685451,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020586678531206743,
      "loss": 1.2751,
      "step": 40699
    },
    {
      "epoch": 3.501677707992773,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00020584491878687422,
      "loss": 1.0125,
      "step": 40700
    },
    {
      "epoch": 3.501763744300095,
      "grad_norm": 1.25,
      "learning_rate": 0.00020582305312202726,
      "loss": 1.004,
      "step": 40701
    },
    {
      "epoch": 3.5018497806074165,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020580118831759083,
      "loss": 1.6379,
      "step": 40702
    },
    {
      "epoch": 3.501935816914738,
      "grad_norm": 1.546875,
      "learning_rate": 0.00020577932437362855,
      "loss": 0.7372,
      "step": 40703
    },
    {
      "epoch": 3.5020218532220597,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0002057574612902045,
      "loss": 1.5509,
      "step": 40704
    },
    {
      "epoch": 3.5021078895293813,
      "grad_norm": 1.09375,
      "learning_rate": 0.00020573559906738283,
      "loss": 0.8261,
      "step": 40705
    },
    {
      "epoch": 3.502193925836703,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00020571373770522712,
      "loss": 0.6322,
      "step": 40706
    },
    {
      "epoch": 3.5022799621440246,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0002056918772038015,
      "loss": 0.769,
      "step": 40707
    },
    {
      "epoch": 3.5023659984513467,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020567001756317005,
      "loss": 1.2781,
      "step": 40708
    },
    {
      "epoch": 3.5024520347586683,
      "grad_norm": 1.15625,
      "learning_rate": 0.00020564815878339638,
      "loss": 0.6124,
      "step": 40709
    },
    {
      "epoch": 3.50253807106599,
      "grad_norm": 1.234375,
      "learning_rate": 0.00020562630086454464,
      "loss": 0.8029,
      "step": 40710
    },
    {
      "epoch": 3.5026241073733115,
      "grad_norm": 1.125,
      "learning_rate": 0.00020560444380667887,
      "loss": 0.9578,
      "step": 40711
    },
    {
      "epoch": 3.502710143680633,
      "grad_norm": 1.046875,
      "learning_rate": 0.00020558258760986265,
      "loss": 1.1387,
      "step": 40712
    },
    {
      "epoch": 3.502796179987955,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00020556073227416029,
      "loss": 1.0033,
      "step": 40713
    },
    {
      "epoch": 3.5028822162952764,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020553887779963537,
      "loss": 0.9169,
      "step": 40714
    },
    {
      "epoch": 3.5029682526025985,
      "grad_norm": 1944.0,
      "learning_rate": 0.00020551702418635193,
      "loss": 1.5942,
      "step": 40715
    },
    {
      "epoch": 3.50305428890992,
      "grad_norm": 0.9375,
      "learning_rate": 0.00020549517143437414,
      "loss": 0.9611,
      "step": 40716
    },
    {
      "epoch": 3.5031403252172417,
      "grad_norm": 0.796875,
      "learning_rate": 0.00020547331954376546,
      "loss": 1.0073,
      "step": 40717
    },
    {
      "epoch": 3.5032263615245633,
      "grad_norm": 1.0,
      "learning_rate": 0.00020545146851459007,
      "loss": 1.5763,
      "step": 40718
    },
    {
      "epoch": 3.503312397831885,
      "grad_norm": 0.953125,
      "learning_rate": 0.00020542961834691198,
      "loss": 0.6076,
      "step": 40719
    },
    {
      "epoch": 3.5033984341392066,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020540776904079484,
      "loss": 1.2235,
      "step": 40720
    },
    {
      "epoch": 3.503484470446528,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020538592059630268,
      "loss": 1.03,
      "step": 40721
    },
    {
      "epoch": 3.5035705067538503,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00020536407301349958,
      "loss": 1.2033,
      "step": 40722
    },
    {
      "epoch": 3.503656543061172,
      "grad_norm": 1.71875,
      "learning_rate": 0.0002053422262924491,
      "loss": 1.2477,
      "step": 40723
    },
    {
      "epoch": 3.5037425793684935,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00020532038043321544,
      "loss": 1.5841,
      "step": 40724
    },
    {
      "epoch": 3.503828615675815,
      "grad_norm": 1.796875,
      "learning_rate": 0.00020529853543586218,
      "loss": 0.698,
      "step": 40725
    },
    {
      "epoch": 3.503914651983137,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002052766913004534,
      "loss": 1.4056,
      "step": 40726
    },
    {
      "epoch": 3.5040006882904584,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00020525484802705313,
      "loss": 0.5844,
      "step": 40727
    },
    {
      "epoch": 3.50408672459778,
      "grad_norm": 7.75,
      "learning_rate": 0.000205233005615725,
      "loss": 1.0953,
      "step": 40728
    },
    {
      "epoch": 3.504172760905102,
      "grad_norm": 1.1875,
      "learning_rate": 0.00020521116406653294,
      "loss": 1.0144,
      "step": 40729
    },
    {
      "epoch": 3.5042587972124237,
      "grad_norm": 1.265625,
      "learning_rate": 0.00020518932337954088,
      "loss": 1.5599,
      "step": 40730
    },
    {
      "epoch": 3.5043448335197454,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00020516748355481275,
      "loss": 1.0416,
      "step": 40731
    },
    {
      "epoch": 3.504430869827067,
      "grad_norm": 0.953125,
      "learning_rate": 0.00020514564459241252,
      "loss": 1.0112,
      "step": 40732
    },
    {
      "epoch": 3.5045169061343886,
      "grad_norm": 1.0,
      "learning_rate": 0.00020512380649240382,
      "loss": 1.0344,
      "step": 40733
    },
    {
      "epoch": 3.5046029424417107,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00020510196925485057,
      "loss": 0.6993,
      "step": 40734
    },
    {
      "epoch": 3.504688978749032,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002050801328798169,
      "loss": 0.7518,
      "step": 40735
    },
    {
      "epoch": 3.504775015056354,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00020505829736736637,
      "loss": 1.036,
      "step": 40736
    },
    {
      "epoch": 3.5048610513636755,
      "grad_norm": 17.625,
      "learning_rate": 0.00020503646271756288,
      "loss": 1.0546,
      "step": 40737
    },
    {
      "epoch": 3.504947087670997,
      "grad_norm": 0.875,
      "learning_rate": 0.00020501462893047058,
      "loss": 0.6726,
      "step": 40738
    },
    {
      "epoch": 3.505033123978319,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00020499279600615295,
      "loss": 0.7323,
      "step": 40739
    },
    {
      "epoch": 3.5051191602856404,
      "grad_norm": 1.4375,
      "learning_rate": 0.000204970963944674,
      "loss": 0.492,
      "step": 40740
    },
    {
      "epoch": 3.5052051965929625,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00020494913274609778,
      "loss": 1.2308,
      "step": 40741
    },
    {
      "epoch": 3.5052912329002837,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002049273024104878,
      "loss": 1.1323,
      "step": 40742
    },
    {
      "epoch": 3.5053772692076057,
      "grad_norm": 1.015625,
      "learning_rate": 0.00020490547293790823,
      "loss": 1.283,
      "step": 40743
    },
    {
      "epoch": 3.5054633055149274,
      "grad_norm": 90.0,
      "learning_rate": 0.0002048836443284226,
      "loss": 1.2352,
      "step": 40744
    },
    {
      "epoch": 3.505549341822249,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00020486181658209486,
      "loss": 0.9145,
      "step": 40745
    },
    {
      "epoch": 3.5056353781295706,
      "grad_norm": 1.359375,
      "learning_rate": 0.00020483998969898914,
      "loss": 1.2561,
      "step": 40746
    },
    {
      "epoch": 3.5057214144368922,
      "grad_norm": 0.90625,
      "learning_rate": 0.00020481816367916883,
      "loss": 0.9683,
      "step": 40747
    },
    {
      "epoch": 3.5058074507442143,
      "grad_norm": 0.953125,
      "learning_rate": 0.00020479633852269803,
      "loss": 1.1745,
      "step": 40748
    },
    {
      "epoch": 3.5058934870515355,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00020477451422964065,
      "loss": 1.6793,
      "step": 40749
    },
    {
      "epoch": 3.5059795233588575,
      "grad_norm": 12.4375,
      "learning_rate": 0.00020475269080006027,
      "loss": 1.2736,
      "step": 40750
    },
    {
      "epoch": 3.506065559666179,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002047308682340208,
      "loss": 0.9241,
      "step": 40751
    },
    {
      "epoch": 3.506151595973501,
      "grad_norm": 27.25,
      "learning_rate": 0.00020470904653158633,
      "loss": 1.078,
      "step": 40752
    },
    {
      "epoch": 3.5062376322808224,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00020468722569282028,
      "loss": 1.1525,
      "step": 40753
    },
    {
      "epoch": 3.506323668588144,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0002046654057177868,
      "loss": 1.1403,
      "step": 40754
    },
    {
      "epoch": 3.506409704895466,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00020464358660654943,
      "loss": 0.4926,
      "step": 40755
    },
    {
      "epoch": 3.5064957412027877,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00020462176835917212,
      "loss": 1.0629,
      "step": 40756
    },
    {
      "epoch": 3.5065817775101094,
      "grad_norm": 95.0,
      "learning_rate": 0.00020459995097571888,
      "loss": 1.7661,
      "step": 40757
    },
    {
      "epoch": 3.506667813817431,
      "grad_norm": 164.0,
      "learning_rate": 0.00020457813445625312,
      "loss": 1.2375,
      "step": 40758
    },
    {
      "epoch": 3.5067538501247526,
      "grad_norm": 1.203125,
      "learning_rate": 0.00020455631880083886,
      "loss": 1.0874,
      "step": 40759
    },
    {
      "epoch": 3.5068398864320742,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00020453450400954016,
      "loss": 0.5078,
      "step": 40760
    },
    {
      "epoch": 3.506925922739396,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002045126900824203,
      "loss": 1.0078,
      "step": 40761
    },
    {
      "epoch": 3.507011959046718,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00020449087701954338,
      "loss": 1.4481,
      "step": 40762
    },
    {
      "epoch": 3.5070979953540395,
      "grad_norm": 0.984375,
      "learning_rate": 0.0002044690648209734,
      "loss": 1.2386,
      "step": 40763
    },
    {
      "epoch": 3.507184031661361,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002044472534867737,
      "loss": 1.287,
      "step": 40764
    },
    {
      "epoch": 3.507270067968683,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00020442544301700827,
      "loss": 1.4505,
      "step": 40765
    },
    {
      "epoch": 3.5073561042760044,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000204403633411741,
      "loss": 0.8555,
      "step": 40766
    },
    {
      "epoch": 3.507442140583326,
      "grad_norm": 1.125,
      "learning_rate": 0.00020438182467103554,
      "loss": 0.8065,
      "step": 40767
    },
    {
      "epoch": 3.5075281768906477,
      "grad_norm": 2.03125,
      "learning_rate": 0.00020436001679495603,
      "loss": 1.1673,
      "step": 40768
    },
    {
      "epoch": 3.5076142131979697,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00020433820978356572,
      "loss": 1.2013,
      "step": 40769
    },
    {
      "epoch": 3.5077002495052914,
      "grad_norm": 32.0,
      "learning_rate": 0.00020431640363692883,
      "loss": 0.8652,
      "step": 40770
    },
    {
      "epoch": 3.507786285812613,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00020429459835510878,
      "loss": 0.5425,
      "step": 40771
    },
    {
      "epoch": 3.5078723221199346,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002042727939381695,
      "loss": 1.7525,
      "step": 40772
    },
    {
      "epoch": 3.5079583584272562,
      "grad_norm": 26.875,
      "learning_rate": 0.000204250990386175,
      "loss": 1.146,
      "step": 40773
    },
    {
      "epoch": 3.508044394734578,
      "grad_norm": 1.34375,
      "learning_rate": 0.00020422918769918863,
      "loss": 0.7239,
      "step": 40774
    },
    {
      "epoch": 3.5081304310418995,
      "grad_norm": 2.421875,
      "learning_rate": 0.0002042073858772744,
      "loss": 0.7433,
      "step": 40775
    },
    {
      "epoch": 3.5082164673492215,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00020418558492049616,
      "loss": 0.9769,
      "step": 40776
    },
    {
      "epoch": 3.508302503656543,
      "grad_norm": 1.546875,
      "learning_rate": 0.0002041637848289174,
      "loss": 1.9309,
      "step": 40777
    },
    {
      "epoch": 3.508388539963865,
      "grad_norm": 1.15625,
      "learning_rate": 0.000204141985602602,
      "loss": 1.142,
      "step": 40778
    },
    {
      "epoch": 3.5084745762711864,
      "grad_norm": 1.125,
      "learning_rate": 0.00020412018724161397,
      "loss": 1.4686,
      "step": 40779
    },
    {
      "epoch": 3.508560612578508,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00020409838974601663,
      "loss": 1.0993,
      "step": 40780
    },
    {
      "epoch": 3.5086466488858297,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00020407659311587402,
      "loss": 1.3352,
      "step": 40781
    },
    {
      "epoch": 3.5087326851931513,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00020405479735124992,
      "loss": 1.2917,
      "step": 40782
    },
    {
      "epoch": 3.5088187215004734,
      "grad_norm": 0.953125,
      "learning_rate": 0.00020403300245220785,
      "loss": 1.6393,
      "step": 40783
    },
    {
      "epoch": 3.508904757807795,
      "grad_norm": 23.375,
      "learning_rate": 0.0002040112084188118,
      "loss": 1.205,
      "step": 40784
    },
    {
      "epoch": 3.5089907941151166,
      "grad_norm": 1.765625,
      "learning_rate": 0.0002039894152511253,
      "loss": 0.9826,
      "step": 40785
    },
    {
      "epoch": 3.5090768304224382,
      "grad_norm": 3.40625,
      "learning_rate": 0.00020396762294921217,
      "loss": 1.6362,
      "step": 40786
    },
    {
      "epoch": 3.50916286672976,
      "grad_norm": 1.296875,
      "learning_rate": 0.00020394583151313634,
      "loss": 1.3892,
      "step": 40787
    },
    {
      "epoch": 3.509248903037082,
      "grad_norm": 1.03125,
      "learning_rate": 0.00020392404094296117,
      "loss": 1.3971,
      "step": 40788
    },
    {
      "epoch": 3.509334939344403,
      "grad_norm": 1.515625,
      "learning_rate": 0.00020390225123875067,
      "loss": 1.5116,
      "step": 40789
    },
    {
      "epoch": 3.509420975651725,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00020388046240056863,
      "loss": 1.613,
      "step": 40790
    },
    {
      "epoch": 3.509507011959047,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00020385867442847844,
      "loss": 0.4777,
      "step": 40791
    },
    {
      "epoch": 3.5095930482663684,
      "grad_norm": 1.625,
      "learning_rate": 0.00020383688732254402,
      "loss": 0.5895,
      "step": 40792
    },
    {
      "epoch": 3.50967908457369,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020381510108282935,
      "loss": 1.1616,
      "step": 40793
    },
    {
      "epoch": 3.5097651208810117,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0002037933157093977,
      "loss": 1.353,
      "step": 40794
    },
    {
      "epoch": 3.5098511571883337,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00020377153120231312,
      "loss": 0.8144,
      "step": 40795
    },
    {
      "epoch": 3.509937193495655,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00020374974756163906,
      "loss": 1.1859,
      "step": 40796
    },
    {
      "epoch": 3.510023229802977,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00020372796478743937,
      "loss": 0.7693,
      "step": 40797
    },
    {
      "epoch": 3.5101092661102986,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00020370618287977792,
      "loss": 1.4478,
      "step": 40798
    },
    {
      "epoch": 3.5101953024176202,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00020368440183871813,
      "loss": 1.1997,
      "step": 40799
    },
    {
      "epoch": 3.510281338724942,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00020366262166432382,
      "loss": 0.6591,
      "step": 40800
    },
    {
      "epoch": 3.5103673750322635,
      "grad_norm": 0.8125,
      "learning_rate": 0.00020364084235665887,
      "loss": 0.6267,
      "step": 40801
    },
    {
      "epoch": 3.5104534113395855,
      "grad_norm": 5.8125,
      "learning_rate": 0.00020361906391578666,
      "loss": 0.7355,
      "step": 40802
    },
    {
      "epoch": 3.5105394476469067,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00020359728634177098,
      "loss": 0.4684,
      "step": 40803
    },
    {
      "epoch": 3.510625483954229,
      "grad_norm": 15.125,
      "learning_rate": 0.00020357550963467564,
      "loss": 1.2575,
      "step": 40804
    },
    {
      "epoch": 3.5107115202615504,
      "grad_norm": 1.125,
      "learning_rate": 0.00020355373379456432,
      "loss": 1.3914,
      "step": 40805
    },
    {
      "epoch": 3.510797556568872,
      "grad_norm": 25.125,
      "learning_rate": 0.00020353195882150078,
      "loss": 1.0469,
      "step": 40806
    },
    {
      "epoch": 3.5108835928761937,
      "grad_norm": 4.28125,
      "learning_rate": 0.00020351018471554845,
      "loss": 1.7468,
      "step": 40807
    },
    {
      "epoch": 3.5109696291835153,
      "grad_norm": 1.3125,
      "learning_rate": 0.00020348841147677115,
      "loss": 1.4164,
      "step": 40808
    },
    {
      "epoch": 3.5110556654908374,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020346663910523279,
      "loss": 1.1117,
      "step": 40809
    },
    {
      "epoch": 3.511141701798159,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002034448676009967,
      "loss": 0.7987,
      "step": 40810
    },
    {
      "epoch": 3.5112277381054806,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020342309696412658,
      "loss": 1.2986,
      "step": 40811
    },
    {
      "epoch": 3.5113137744128022,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00020340132719468646,
      "loss": 0.8014,
      "step": 40812
    },
    {
      "epoch": 3.511399810720124,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00020337955829273963,
      "loss": 0.9791,
      "step": 40813
    },
    {
      "epoch": 3.5114858470274455,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00020335779025835006,
      "loss": 1.2385,
      "step": 40814
    },
    {
      "epoch": 3.511571883334767,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00020333602309158106,
      "loss": 0.8358,
      "step": 40815
    },
    {
      "epoch": 3.511657919642089,
      "grad_norm": 20.125,
      "learning_rate": 0.00020331425679249649,
      "loss": 0.9582,
      "step": 40816
    },
    {
      "epoch": 3.511743955949411,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0002032924913611602,
      "loss": 0.9882,
      "step": 40817
    },
    {
      "epoch": 3.5118299922567324,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00020327072679763547,
      "loss": 1.0476,
      "step": 40818
    },
    {
      "epoch": 3.511916028564054,
      "grad_norm": 1.46875,
      "learning_rate": 0.0002032489631019862,
      "loss": 1.6384,
      "step": 40819
    },
    {
      "epoch": 3.5120020648713757,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00020322720027427616,
      "loss": 1.057,
      "step": 40820
    },
    {
      "epoch": 3.5120881011786973,
      "grad_norm": 8.0,
      "learning_rate": 0.00020320543831456862,
      "loss": 0.986,
      "step": 40821
    },
    {
      "epoch": 3.512174137486019,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002031836772229275,
      "loss": 0.8137,
      "step": 40822
    },
    {
      "epoch": 3.512260173793341,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00020316191699941655,
      "loss": 1.0583,
      "step": 40823
    },
    {
      "epoch": 3.5123462101006626,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002031401576440991,
      "loss": 0.6586,
      "step": 40824
    },
    {
      "epoch": 3.5124322464079842,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00020311839915703912,
      "loss": 1.7939,
      "step": 40825
    },
    {
      "epoch": 3.512518282715306,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00020309664153829994,
      "loss": 0.4168,
      "step": 40826
    },
    {
      "epoch": 3.5126043190226275,
      "grad_norm": 1.140625,
      "learning_rate": 0.00020307488478794528,
      "loss": 0.8471,
      "step": 40827
    },
    {
      "epoch": 3.512690355329949,
      "grad_norm": 1.5,
      "learning_rate": 0.00020305312890603905,
      "loss": 0.796,
      "step": 40828
    },
    {
      "epoch": 3.5127763916372707,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020303137389264447,
      "loss": 0.7941,
      "step": 40829
    },
    {
      "epoch": 3.512862427944593,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00020300961974782533,
      "loss": 0.7007,
      "step": 40830
    },
    {
      "epoch": 3.5129484642519144,
      "grad_norm": 1.34375,
      "learning_rate": 0.00020298786647164553,
      "loss": 1.5171,
      "step": 40831
    },
    {
      "epoch": 3.513034500559236,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00020296611406416825,
      "loss": 0.812,
      "step": 40832
    },
    {
      "epoch": 3.5131205368665577,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00020294436252545734,
      "loss": 0.6078,
      "step": 40833
    },
    {
      "epoch": 3.5132065731738793,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00020292261185557653,
      "loss": 1.3221,
      "step": 40834
    },
    {
      "epoch": 3.513292609481201,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00020290086205458914,
      "loss": 0.6301,
      "step": 40835
    },
    {
      "epoch": 3.5133786457885225,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00020287911312255914,
      "loss": 1.0532,
      "step": 40836
    },
    {
      "epoch": 3.5134646820958446,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00020285736505954978,
      "loss": 2.1395,
      "step": 40837
    },
    {
      "epoch": 3.5135507184031662,
      "grad_norm": 1.09375,
      "learning_rate": 0.00020283561786562483,
      "loss": 0.8547,
      "step": 40838
    },
    {
      "epoch": 3.513636754710488,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020281387154084807,
      "loss": 0.9338,
      "step": 40839
    },
    {
      "epoch": 3.5137227910178095,
      "grad_norm": 6.40625,
      "learning_rate": 0.00020279212608528257,
      "loss": 1.0089,
      "step": 40840
    },
    {
      "epoch": 3.513808827325131,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00020277038149899262,
      "loss": 0.7428,
      "step": 40841
    },
    {
      "epoch": 3.5138948636324527,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002027486377820414,
      "loss": 0.9978,
      "step": 40842
    },
    {
      "epoch": 3.5139808999397744,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00020272689493449253,
      "loss": 1.0001,
      "step": 40843
    },
    {
      "epoch": 3.5140669362470964,
      "grad_norm": 1.125,
      "learning_rate": 0.0002027051529564099,
      "loss": 2.4351,
      "step": 40844
    },
    {
      "epoch": 3.514152972554418,
      "grad_norm": 1.453125,
      "learning_rate": 0.0002026834118478567,
      "loss": 0.5318,
      "step": 40845
    },
    {
      "epoch": 3.5142390088617397,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00020266167160889665,
      "loss": 0.9554,
      "step": 40846
    },
    {
      "epoch": 3.5143250451690613,
      "grad_norm": 1.015625,
      "learning_rate": 0.00020263993223959358,
      "loss": 1.3628,
      "step": 40847
    },
    {
      "epoch": 3.514411081476383,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00020261819374001072,
      "loss": 1.7298,
      "step": 40848
    },
    {
      "epoch": 3.514497117783705,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002025964561102118,
      "loss": 1.0955,
      "step": 40849
    },
    {
      "epoch": 3.514583154091026,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00020257471935026055,
      "loss": 1.6197,
      "step": 40850
    },
    {
      "epoch": 3.5146691903983482,
      "grad_norm": 4.21875,
      "learning_rate": 0.00020255298346022026,
      "loss": 0.9693,
      "step": 40851
    },
    {
      "epoch": 3.51475522670567,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00020253124844015464,
      "loss": 0.6081,
      "step": 40852
    },
    {
      "epoch": 3.5148412630129915,
      "grad_norm": 1.484375,
      "learning_rate": 0.00020250951429012743,
      "loss": 1.2059,
      "step": 40853
    },
    {
      "epoch": 3.514927299320313,
      "grad_norm": 1.640625,
      "learning_rate": 0.00020248778101020188,
      "loss": 1.9213,
      "step": 40854
    },
    {
      "epoch": 3.5150133356276347,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002024660486004418,
      "loss": 1.296,
      "step": 40855
    },
    {
      "epoch": 3.515099371934957,
      "grad_norm": 1.5625,
      "learning_rate": 0.0002024443170609106,
      "loss": 1.7275,
      "step": 40856
    },
    {
      "epoch": 3.515185408242278,
      "grad_norm": 0.890625,
      "learning_rate": 0.00020242258639167183,
      "loss": 1.041,
      "step": 40857
    },
    {
      "epoch": 3.5152714445496,
      "grad_norm": 1.828125,
      "learning_rate": 0.00020240085659278934,
      "loss": 1.5202,
      "step": 40858
    },
    {
      "epoch": 3.5153574808569217,
      "grad_norm": 7.65625,
      "learning_rate": 0.00020237912766432625,
      "loss": 1.3219,
      "step": 40859
    },
    {
      "epoch": 3.5154435171642433,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002023573996063463,
      "loss": 1.4652,
      "step": 40860
    },
    {
      "epoch": 3.515529553471565,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00020233567241891332,
      "loss": 1.142,
      "step": 40861
    },
    {
      "epoch": 3.5156155897788866,
      "grad_norm": 1.28125,
      "learning_rate": 0.00020231394610209036,
      "loss": 0.8423,
      "step": 40862
    },
    {
      "epoch": 3.5157016260862086,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00020229222065594127,
      "loss": 0.6901,
      "step": 40863
    },
    {
      "epoch": 3.51578766239353,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00020227049608052965,
      "loss": 1.2915,
      "step": 40864
    },
    {
      "epoch": 3.515873698700852,
      "grad_norm": 0.984375,
      "learning_rate": 0.00020224877237591882,
      "loss": 1.0906,
      "step": 40865
    },
    {
      "epoch": 3.5159597350081735,
      "grad_norm": 7.75,
      "learning_rate": 0.00020222704954217253,
      "loss": 1.1965,
      "step": 40866
    },
    {
      "epoch": 3.516045771315495,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00020220532757935406,
      "loss": 1.0514,
      "step": 40867
    },
    {
      "epoch": 3.5161318076228167,
      "grad_norm": 2.140625,
      "learning_rate": 0.0002021836064875271,
      "loss": 1.3743,
      "step": 40868
    },
    {
      "epoch": 3.5162178439301384,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0002021618862667553,
      "loss": 1.0982,
      "step": 40869
    },
    {
      "epoch": 3.5163038802374604,
      "grad_norm": 1.140625,
      "learning_rate": 0.0002021401669171019,
      "loss": 1.6198,
      "step": 40870
    },
    {
      "epoch": 3.516389916544782,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00020211844843863065,
      "loss": 0.8227,
      "step": 40871
    },
    {
      "epoch": 3.5164759528521037,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00020209673083140508,
      "loss": 0.7237,
      "step": 40872
    },
    {
      "epoch": 3.5165619891594253,
      "grad_norm": 4.5,
      "learning_rate": 0.0002020750140954885,
      "loss": 2.2034,
      "step": 40873
    },
    {
      "epoch": 3.516648025466747,
      "grad_norm": 1.25,
      "learning_rate": 0.0002020532982309446,
      "loss": 1.1491,
      "step": 40874
    },
    {
      "epoch": 3.5167340617740686,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00020203158323783694,
      "loss": 0.7811,
      "step": 40875
    },
    {
      "epoch": 3.51682009808139,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002020098691162288,
      "loss": 1.1863,
      "step": 40876
    },
    {
      "epoch": 3.5169061343887122,
      "grad_norm": 1.3125,
      "learning_rate": 0.0002019881558661839,
      "loss": 1.3498,
      "step": 40877
    },
    {
      "epoch": 3.516992170696034,
      "grad_norm": 1.359375,
      "learning_rate": 0.00020196644348776567,
      "loss": 1.7777,
      "step": 40878
    },
    {
      "epoch": 3.5170782070033555,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00020194473198103757,
      "loss": 0.9307,
      "step": 40879
    },
    {
      "epoch": 3.517164243310677,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00020192302134606338,
      "loss": 1.3377,
      "step": 40880
    },
    {
      "epoch": 3.5172502796179987,
      "grad_norm": 12.875,
      "learning_rate": 0.0002019013115829062,
      "loss": 1.5812,
      "step": 40881
    },
    {
      "epoch": 3.5173363159253204,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00020187960269162986,
      "loss": 1.1498,
      "step": 40882
    },
    {
      "epoch": 3.517422352232642,
      "grad_norm": 0.8125,
      "learning_rate": 0.00020185789467229754,
      "loss": 1.4914,
      "step": 40883
    },
    {
      "epoch": 3.517508388539964,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002018361875249729,
      "loss": 1.2002,
      "step": 40884
    },
    {
      "epoch": 3.5175944248472857,
      "grad_norm": 0.921875,
      "learning_rate": 0.00020181448124971958,
      "loss": 0.7891,
      "step": 40885
    },
    {
      "epoch": 3.5176804611546073,
      "grad_norm": 3.859375,
      "learning_rate": 0.0002017927758466007,
      "loss": 1.0091,
      "step": 40886
    },
    {
      "epoch": 3.517766497461929,
      "grad_norm": 1.7421875,
      "learning_rate": 0.00020177107131568,
      "loss": 0.8924,
      "step": 40887
    },
    {
      "epoch": 3.5178525337692506,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00020174936765702113,
      "loss": 1.3052,
      "step": 40888
    },
    {
      "epoch": 3.517938570076572,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0002017276648706871,
      "loss": 1.1571,
      "step": 40889
    },
    {
      "epoch": 3.518024606383894,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002017059629567417,
      "loss": 1.4606,
      "step": 40890
    },
    {
      "epoch": 3.518110642691216,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00020168426191524846,
      "loss": 1.1131,
      "step": 40891
    },
    {
      "epoch": 3.5181966789985375,
      "grad_norm": 1.265625,
      "learning_rate": 0.00020166256174627062,
      "loss": 2.022,
      "step": 40892
    },
    {
      "epoch": 3.518282715305859,
      "grad_norm": 7.75,
      "learning_rate": 0.00020164086244987172,
      "loss": 1.1802,
      "step": 40893
    },
    {
      "epoch": 3.5183687516131807,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00020161916402611547,
      "loss": 1.4503,
      "step": 40894
    },
    {
      "epoch": 3.5184547879205024,
      "grad_norm": 4.1875,
      "learning_rate": 0.00020159746647506493,
      "loss": 1.2911,
      "step": 40895
    },
    {
      "epoch": 3.518540824227824,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000201575769796784,
      "loss": 0.8406,
      "step": 40896
    },
    {
      "epoch": 3.5186268605351456,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00020155407399133563,
      "loss": 1.2,
      "step": 40897
    },
    {
      "epoch": 3.5187128968424677,
      "grad_norm": 1.328125,
      "learning_rate": 0.00020153237905878364,
      "loss": 1.3233,
      "step": 40898
    },
    {
      "epoch": 3.5187989331497893,
      "grad_norm": 125.5,
      "learning_rate": 0.0002015106849991915,
      "loss": 0.9829,
      "step": 40899
    },
    {
      "epoch": 3.518884969457111,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00020148899181262242,
      "loss": 1.8982,
      "step": 40900
    },
    {
      "epoch": 3.5189710057644326,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00020146729949914,
      "loss": 0.8006,
      "step": 40901
    },
    {
      "epoch": 3.519057042071754,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002014456080588078,
      "loss": 2.0806,
      "step": 40902
    },
    {
      "epoch": 3.519143078379076,
      "grad_norm": 1.203125,
      "learning_rate": 0.00020142391749168899,
      "loss": 1.5989,
      "step": 40903
    },
    {
      "epoch": 3.5192291146863974,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020140222779784712,
      "loss": 1.1902,
      "step": 40904
    },
    {
      "epoch": 3.5193151509937195,
      "grad_norm": 1.078125,
      "learning_rate": 0.00020138053897734583,
      "loss": 0.7031,
      "step": 40905
    },
    {
      "epoch": 3.519401187301041,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002013588510302482,
      "loss": 1.0192,
      "step": 40906
    },
    {
      "epoch": 3.5194872236083627,
      "grad_norm": 1.546875,
      "learning_rate": 0.00020133716395661806,
      "loss": 1.1471,
      "step": 40907
    },
    {
      "epoch": 3.5195732599156844,
      "grad_norm": 3.265625,
      "learning_rate": 0.0002013154777565185,
      "loss": 1.0698,
      "step": 40908
    },
    {
      "epoch": 3.519659296223006,
      "grad_norm": 2.046875,
      "learning_rate": 0.000201293792430013,
      "loss": 1.0598,
      "step": 40909
    },
    {
      "epoch": 3.519745332530328,
      "grad_norm": 1.046875,
      "learning_rate": 0.00020127210797716522,
      "loss": 1.4275,
      "step": 40910
    },
    {
      "epoch": 3.5198313688376492,
      "grad_norm": 0.921875,
      "learning_rate": 0.00020125042439803832,
      "loss": 0.7671,
      "step": 40911
    },
    {
      "epoch": 3.5199174051449713,
      "grad_norm": 1.3125,
      "learning_rate": 0.00020122874169269578,
      "loss": 1.1094,
      "step": 40912
    },
    {
      "epoch": 3.520003441452293,
      "grad_norm": 0.640625,
      "learning_rate": 0.00020120705986120124,
      "loss": 0.4672,
      "step": 40913
    },
    {
      "epoch": 3.5200894777596146,
      "grad_norm": 1.203125,
      "learning_rate": 0.0002011853789036176,
      "loss": 0.7671,
      "step": 40914
    },
    {
      "epoch": 3.520175514066936,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00020116369882000897,
      "loss": 0.7301,
      "step": 40915
    },
    {
      "epoch": 3.520261550374258,
      "grad_norm": 58.25,
      "learning_rate": 0.0002011420196104382,
      "loss": 0.6707,
      "step": 40916
    },
    {
      "epoch": 3.52034758668158,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00020112034127496892,
      "loss": 0.6281,
      "step": 40917
    },
    {
      "epoch": 3.520433622988901,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00020109866381366466,
      "loss": 1.8077,
      "step": 40918
    },
    {
      "epoch": 3.520519659296223,
      "grad_norm": 1.109375,
      "learning_rate": 0.00020107698722658852,
      "loss": 0.6902,
      "step": 40919
    },
    {
      "epoch": 3.5206056956035447,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00020105531151380407,
      "loss": 0.9939,
      "step": 40920
    },
    {
      "epoch": 3.5206917319108664,
      "grad_norm": 1.21875,
      "learning_rate": 0.00020103363667537484,
      "loss": 0.8789,
      "step": 40921
    },
    {
      "epoch": 3.520777768218188,
      "grad_norm": 1.71875,
      "learning_rate": 0.0002010119627113639,
      "loss": 1.7389,
      "step": 40922
    },
    {
      "epoch": 3.5208638045255096,
      "grad_norm": 1.078125,
      "learning_rate": 0.00020099028962183502,
      "loss": 0.9902,
      "step": 40923
    },
    {
      "epoch": 3.5209498408328317,
      "grad_norm": 4.0,
      "learning_rate": 0.0002009686174068512,
      "loss": 1.0454,
      "step": 40924
    },
    {
      "epoch": 3.521035877140153,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000200946946066476,
      "loss": 1.5092,
      "step": 40925
    },
    {
      "epoch": 3.521121913447475,
      "grad_norm": 2.640625,
      "learning_rate": 0.00020092527560077302,
      "loss": 0.8368,
      "step": 40926
    },
    {
      "epoch": 3.5212079497547966,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0002009036060098053,
      "loss": 0.867,
      "step": 40927
    },
    {
      "epoch": 3.521293986062118,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002008819372936363,
      "loss": 0.6549,
      "step": 40928
    },
    {
      "epoch": 3.52138002236944,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00020086026945232967,
      "loss": 0.9276,
      "step": 40929
    },
    {
      "epoch": 3.5214660586767614,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00020083860248594837,
      "loss": 0.64,
      "step": 40930
    },
    {
      "epoch": 3.5215520949840835,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00020081693639455595,
      "loss": 0.8108,
      "step": 40931
    },
    {
      "epoch": 3.521638131291405,
      "grad_norm": 14.8125,
      "learning_rate": 0.00020079527117821606,
      "loss": 0.5633,
      "step": 40932
    },
    {
      "epoch": 3.5217241675987268,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00020077360683699158,
      "loss": 0.8898,
      "step": 40933
    },
    {
      "epoch": 3.5218102039060484,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002007519433709461,
      "loss": 1.4561,
      "step": 40934
    },
    {
      "epoch": 3.52189624021337,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00020073028078014315,
      "loss": 1.074,
      "step": 40935
    },
    {
      "epoch": 3.5219822765206916,
      "grad_norm": 183.0,
      "learning_rate": 0.00020070861906464578,
      "loss": 1.1953,
      "step": 40936
    },
    {
      "epoch": 3.5220683128280132,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00020068695822451766,
      "loss": 1.219,
      "step": 40937
    },
    {
      "epoch": 3.5221543491353353,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00020066529825982178,
      "loss": 0.8385,
      "step": 40938
    },
    {
      "epoch": 3.522240385442657,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00020064363917062173,
      "loss": 1.6426,
      "step": 40939
    },
    {
      "epoch": 3.5223264217499786,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00020062198095698093,
      "loss": 0.57,
      "step": 40940
    },
    {
      "epoch": 3.5224124580573,
      "grad_norm": 1.390625,
      "learning_rate": 0.00020060032361896246,
      "loss": 0.7564,
      "step": 40941
    },
    {
      "epoch": 3.522498494364622,
      "grad_norm": 1.28125,
      "learning_rate": 0.0002005786671566298,
      "loss": 0.7734,
      "step": 40942
    },
    {
      "epoch": 3.5225845306719434,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00020055701157004651,
      "loss": 2.032,
      "step": 40943
    },
    {
      "epoch": 3.522670566979265,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020053535685927553,
      "loss": 1.7811,
      "step": 40944
    },
    {
      "epoch": 3.522756603286587,
      "grad_norm": 1.21875,
      "learning_rate": 0.00020051370302438045,
      "loss": 1.3881,
      "step": 40945
    },
    {
      "epoch": 3.5228426395939088,
      "grad_norm": 1.09375,
      "learning_rate": 0.00020049205006542466,
      "loss": 1.3713,
      "step": 40946
    },
    {
      "epoch": 3.5229286759012304,
      "grad_norm": 1.484375,
      "learning_rate": 0.00020047039798247118,
      "loss": 1.3636,
      "step": 40947
    },
    {
      "epoch": 3.523014712208552,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00020044874677558372,
      "loss": 1.0731,
      "step": 40948
    },
    {
      "epoch": 3.5231007485158736,
      "grad_norm": 1.703125,
      "learning_rate": 0.0002004270964448252,
      "loss": 1.5501,
      "step": 40949
    },
    {
      "epoch": 3.5231867848231953,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00020040544699025925,
      "loss": 1.2981,
      "step": 40950
    },
    {
      "epoch": 3.523272821130517,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00020038379841194904,
      "loss": 1.1976,
      "step": 40951
    },
    {
      "epoch": 3.523358857437839,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00020036215070995796,
      "loss": 0.8066,
      "step": 40952
    },
    {
      "epoch": 3.5234448937451606,
      "grad_norm": 1.484375,
      "learning_rate": 0.0002003405038843495,
      "loss": 1.2017,
      "step": 40953
    },
    {
      "epoch": 3.523530930052482,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00020031885793518657,
      "loss": 1.1075,
      "step": 40954
    },
    {
      "epoch": 3.523616966359804,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00020029721286253272,
      "loss": 0.713,
      "step": 40955
    },
    {
      "epoch": 3.5237030026671254,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00020027556866645141,
      "loss": 0.9606,
      "step": 40956
    },
    {
      "epoch": 3.523789038974447,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002002539253470056,
      "loss": 0.6216,
      "step": 40957
    },
    {
      "epoch": 3.5238750752817687,
      "grad_norm": 0.890625,
      "learning_rate": 0.00020023228290425878,
      "loss": 0.8279,
      "step": 40958
    },
    {
      "epoch": 3.5239611115890908,
      "grad_norm": 1.140625,
      "learning_rate": 0.00020021064133827438,
      "loss": 1.1652,
      "step": 40959
    },
    {
      "epoch": 3.5240471478964124,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002001890006491154,
      "loss": 1.171,
      "step": 40960
    },
    {
      "epoch": 3.524133184203734,
      "grad_norm": 1.0625,
      "learning_rate": 0.00020016736083684522,
      "loss": 1.1506,
      "step": 40961
    },
    {
      "epoch": 3.5242192205110556,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00020014572190152742,
      "loss": 0.8288,
      "step": 40962
    },
    {
      "epoch": 3.5243052568183773,
      "grad_norm": 2.109375,
      "learning_rate": 0.00020012408384322488,
      "loss": 0.7015,
      "step": 40963
    },
    {
      "epoch": 3.5243912931256993,
      "grad_norm": 1.109375,
      "learning_rate": 0.00020010244666200105,
      "loss": 0.7015,
      "step": 40964
    },
    {
      "epoch": 3.5244773294330205,
      "grad_norm": 0.734375,
      "learning_rate": 0.00020008081035791946,
      "loss": 0.4557,
      "step": 40965
    },
    {
      "epoch": 3.5245633657403426,
      "grad_norm": 1.09375,
      "learning_rate": 0.00020005917493104298,
      "loss": 1.2833,
      "step": 40966
    },
    {
      "epoch": 3.524649402047664,
      "grad_norm": 1.3125,
      "learning_rate": 0.00020003754038143523,
      "loss": 1.2891,
      "step": 40967
    },
    {
      "epoch": 3.524735438354986,
      "grad_norm": 1.296875,
      "learning_rate": 0.00020001590670915914,
      "loss": 1.2167,
      "step": 40968
    },
    {
      "epoch": 3.5248214746623074,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019999427391427822,
      "loss": 0.9343,
      "step": 40969
    },
    {
      "epoch": 3.524907510969629,
      "grad_norm": 6.96875,
      "learning_rate": 0.00019997264199685588,
      "loss": 1.3036,
      "step": 40970
    },
    {
      "epoch": 3.524993547276951,
      "grad_norm": 2.734375,
      "learning_rate": 0.000199951010956955,
      "loss": 1.0432,
      "step": 40971
    },
    {
      "epoch": 3.5250795835842723,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019992938079463912,
      "loss": 1.1596,
      "step": 40972
    },
    {
      "epoch": 3.5251656198915944,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00019990775150997153,
      "loss": 1.4869,
      "step": 40973
    },
    {
      "epoch": 3.525251656198916,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001998861231030153,
      "loss": 0.8365,
      "step": 40974
    },
    {
      "epoch": 3.5253376925062376,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00019986449557383374,
      "loss": 1.5602,
      "step": 40975
    },
    {
      "epoch": 3.5254237288135593,
      "grad_norm": 14.5625,
      "learning_rate": 0.00019984286892249037,
      "loss": 1.7034,
      "step": 40976
    },
    {
      "epoch": 3.525509765120881,
      "grad_norm": 32.75,
      "learning_rate": 0.00019982124314904803,
      "loss": 1.0681,
      "step": 40977
    },
    {
      "epoch": 3.525595801428203,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019979961825357034,
      "loss": 1.2499,
      "step": 40978
    },
    {
      "epoch": 3.525681837735524,
      "grad_norm": 1.34375,
      "learning_rate": 0.00019977799423612024,
      "loss": 0.9616,
      "step": 40979
    },
    {
      "epoch": 3.525767874042846,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019975637109676108,
      "loss": 1.3584,
      "step": 40980
    },
    {
      "epoch": 3.525853910350168,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019973474883555636,
      "loss": 0.7041,
      "step": 40981
    },
    {
      "epoch": 3.5259399466574894,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019971312745256892,
      "loss": 0.4505,
      "step": 40982
    },
    {
      "epoch": 3.526025982964811,
      "grad_norm": 1.328125,
      "learning_rate": 0.00019969150694786216,
      "loss": 0.9196,
      "step": 40983
    },
    {
      "epoch": 3.5261120192721327,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019966988732149948,
      "loss": 0.6643,
      "step": 40984
    },
    {
      "epoch": 3.5261980555794548,
      "grad_norm": 1.359375,
      "learning_rate": 0.00019964826857354385,
      "loss": 1.7515,
      "step": 40985
    },
    {
      "epoch": 3.5262840918867764,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019962665070405862,
      "loss": 0.8474,
      "step": 40986
    },
    {
      "epoch": 3.526370128194098,
      "grad_norm": 1.703125,
      "learning_rate": 0.00019960503371310712,
      "loss": 1.4524,
      "step": 40987
    },
    {
      "epoch": 3.5264561645014196,
      "grad_norm": 1.25,
      "learning_rate": 0.00019958341760075216,
      "loss": 1.7101,
      "step": 40988
    },
    {
      "epoch": 3.5265422008087413,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019956180236705768,
      "loss": 1.0099,
      "step": 40989
    },
    {
      "epoch": 3.526628237116063,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019954018801208624,
      "loss": 1.581,
      "step": 40990
    },
    {
      "epoch": 3.5267142734233845,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001995185745359014,
      "loss": 1.2151,
      "step": 40991
    },
    {
      "epoch": 3.5268003097307066,
      "grad_norm": 1.25,
      "learning_rate": 0.00019949696193856636,
      "loss": 0.7819,
      "step": 40992
    },
    {
      "epoch": 3.526886346038028,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019947535022014418,
      "loss": 1.1743,
      "step": 40993
    },
    {
      "epoch": 3.52697238234535,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019945373938069828,
      "loss": 0.6282,
      "step": 40994
    },
    {
      "epoch": 3.5270584186526714,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019943212942029155,
      "loss": 0.8253,
      "step": 40995
    },
    {
      "epoch": 3.527144454959993,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019941052033898744,
      "loss": 1.1051,
      "step": 40996
    },
    {
      "epoch": 3.5272304912673147,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019938891213684923,
      "loss": 1.518,
      "step": 40997
    },
    {
      "epoch": 3.5273165275746363,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019936730481393979,
      "loss": 0.3955,
      "step": 40998
    },
    {
      "epoch": 3.5274025638819584,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019934569837032256,
      "loss": 0.6533,
      "step": 40999
    },
    {
      "epoch": 3.52748860018928,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019932409280606084,
      "loss": 1.3183,
      "step": 41000
    },
    {
      "epoch": 3.5275746364966016,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001993024881212175,
      "loss": 1.0336,
      "step": 41001
    },
    {
      "epoch": 3.5276606728039233,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001992808843158559,
      "loss": 1.8314,
      "step": 41002
    },
    {
      "epoch": 3.527746709111245,
      "grad_norm": 92.5,
      "learning_rate": 0.00019925928139003941,
      "loss": 0.848,
      "step": 41003
    },
    {
      "epoch": 3.5278327454185665,
      "grad_norm": 472.0,
      "learning_rate": 0.00019923767934383087,
      "loss": 1.1197,
      "step": 41004
    },
    {
      "epoch": 3.527918781725888,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001992160781772936,
      "loss": 0.5284,
      "step": 41005
    },
    {
      "epoch": 3.52800481803321,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00019919447789049094,
      "loss": 0.8917,
      "step": 41006
    },
    {
      "epoch": 3.528090854340532,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001991728784834858,
      "loss": 0.9661,
      "step": 41007
    },
    {
      "epoch": 3.5281768906478534,
      "grad_norm": 2.484375,
      "learning_rate": 0.00019915127995634164,
      "loss": 1.552,
      "step": 41008
    },
    {
      "epoch": 3.528262926955175,
      "grad_norm": 9.4375,
      "learning_rate": 0.00019912968230912132,
      "loss": 1.3688,
      "step": 41009
    },
    {
      "epoch": 3.5283489632624967,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019910808554188814,
      "loss": 0.8515,
      "step": 41010
    },
    {
      "epoch": 3.5284349995698183,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001990864896547055,
      "loss": 1.1448,
      "step": 41011
    },
    {
      "epoch": 3.52852103587714,
      "grad_norm": 2.359375,
      "learning_rate": 0.00019906489464763616,
      "loss": 1.9404,
      "step": 41012
    },
    {
      "epoch": 3.528607072184462,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019904330052074347,
      "loss": 1.2328,
      "step": 41013
    },
    {
      "epoch": 3.5286931084917836,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019902170727409074,
      "loss": 1.3147,
      "step": 41014
    },
    {
      "epoch": 3.5287791447991053,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019900011490774083,
      "loss": 0.5629,
      "step": 41015
    },
    {
      "epoch": 3.528865181106427,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001989785234217571,
      "loss": 0.5644,
      "step": 41016
    },
    {
      "epoch": 3.5289512174137485,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019895693281620275,
      "loss": 1.4453,
      "step": 41017
    },
    {
      "epoch": 3.52903725372107,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019893534309114064,
      "loss": 0.642,
      "step": 41018
    },
    {
      "epoch": 3.5291232900283918,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019891375424663433,
      "loss": 1.6453,
      "step": 41019
    },
    {
      "epoch": 3.529209326335714,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001988921662827465,
      "loss": 1.0457,
      "step": 41020
    },
    {
      "epoch": 3.5292953626430354,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019887057919954054,
      "loss": 0.8075,
      "step": 41021
    },
    {
      "epoch": 3.529381398950357,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019884899299707975,
      "loss": 1.232,
      "step": 41022
    },
    {
      "epoch": 3.5294674352576787,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001988274076754269,
      "loss": 1.5691,
      "step": 41023
    },
    {
      "epoch": 3.5295534715650003,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001988058232346453,
      "loss": 1.6053,
      "step": 41024
    },
    {
      "epoch": 3.5296395078723224,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00019878423967479807,
      "loss": 1.3133,
      "step": 41025
    },
    {
      "epoch": 3.5297255441796436,
      "grad_norm": 1.46875,
      "learning_rate": 0.00019876265699594843,
      "loss": 1.1251,
      "step": 41026
    },
    {
      "epoch": 3.5298115804869656,
      "grad_norm": 324.0,
      "learning_rate": 0.00019874107519815955,
      "loss": 1.4345,
      "step": 41027
    },
    {
      "epoch": 3.5298976167942873,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019871949428149427,
      "loss": 1.0379,
      "step": 41028
    },
    {
      "epoch": 3.529983653101609,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00019869791424601585,
      "loss": 1.1625,
      "step": 41029
    },
    {
      "epoch": 3.5300696894089305,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019867633509178768,
      "loss": 0.8917,
      "step": 41030
    },
    {
      "epoch": 3.530155725716252,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019865475681887234,
      "loss": 1.1858,
      "step": 41031
    },
    {
      "epoch": 3.530241762023574,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019863317942733332,
      "loss": 0.9892,
      "step": 41032
    },
    {
      "epoch": 3.5303277983308954,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019861160291723378,
      "loss": 1.304,
      "step": 41033
    },
    {
      "epoch": 3.5304138346382175,
      "grad_norm": 1.75,
      "learning_rate": 0.0001985900272886365,
      "loss": 1.511,
      "step": 41034
    },
    {
      "epoch": 3.530499870945539,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000198568452541605,
      "loss": 1.3168,
      "step": 41035
    },
    {
      "epoch": 3.5305859072528607,
      "grad_norm": 3.0625,
      "learning_rate": 0.00019854687867620191,
      "loss": 1.266,
      "step": 41036
    },
    {
      "epoch": 3.5306719435601823,
      "grad_norm": 2.203125,
      "learning_rate": 0.00019852530569249054,
      "loss": 1.0954,
      "step": 41037
    },
    {
      "epoch": 3.530757979867504,
      "grad_norm": 0.875,
      "learning_rate": 0.0001985037335905343,
      "loss": 1.144,
      "step": 41038
    },
    {
      "epoch": 3.530844016174826,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00019848216237039574,
      "loss": 1.8035,
      "step": 41039
    },
    {
      "epoch": 3.530930052482147,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019846059203213823,
      "loss": 0.7692,
      "step": 41040
    },
    {
      "epoch": 3.5310160887894693,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019843902257582507,
      "loss": 1.2532,
      "step": 41041
    },
    {
      "epoch": 3.531102125096791,
      "grad_norm": 1.296875,
      "learning_rate": 0.00019841745400151888,
      "loss": 1.3491,
      "step": 41042
    },
    {
      "epoch": 3.5311881614041125,
      "grad_norm": 1.0,
      "learning_rate": 0.00019839588630928302,
      "loss": 0.6424,
      "step": 41043
    },
    {
      "epoch": 3.531274197711434,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019837431949918071,
      "loss": 0.6942,
      "step": 41044
    },
    {
      "epoch": 3.5313602340187558,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019835275357127464,
      "loss": 1.4594,
      "step": 41045
    },
    {
      "epoch": 3.531446270326078,
      "grad_norm": 2.3125,
      "learning_rate": 0.00019833118852562816,
      "loss": 1.2476,
      "step": 41046
    },
    {
      "epoch": 3.5315323066333995,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019830962436230437,
      "loss": 1.3806,
      "step": 41047
    },
    {
      "epoch": 3.531618342940721,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019828806108136614,
      "loss": 1.301,
      "step": 41048
    },
    {
      "epoch": 3.5317043792480427,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00019826649868287681,
      "loss": 1.3636,
      "step": 41049
    },
    {
      "epoch": 3.5317904155553643,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019824493716689913,
      "loss": 1.179,
      "step": 41050
    },
    {
      "epoch": 3.531876451862686,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001982233765334963,
      "loss": 1.0395,
      "step": 41051
    },
    {
      "epoch": 3.5319624881700076,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019820181678273157,
      "loss": 1.0466,
      "step": 41052
    },
    {
      "epoch": 3.5320485244773296,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001981802579146677,
      "loss": 1.2023,
      "step": 41053
    },
    {
      "epoch": 3.5321345607846513,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001981586999293678,
      "loss": 1.3467,
      "step": 41054
    },
    {
      "epoch": 3.532220597091973,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001981371428268952,
      "loss": 1.0709,
      "step": 41055
    },
    {
      "epoch": 3.5323066333992945,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001981155866073126,
      "loss": 1.4351,
      "step": 41056
    },
    {
      "epoch": 3.532392669706616,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001980940312706832,
      "loss": 0.876,
      "step": 41057
    },
    {
      "epoch": 3.5324787060139378,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019807247681707019,
      "loss": 0.6027,
      "step": 41058
    },
    {
      "epoch": 3.5325647423212594,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00019805092324653634,
      "loss": 1.3437,
      "step": 41059
    },
    {
      "epoch": 3.5326507786285815,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019802937055914493,
      "loss": 1.002,
      "step": 41060
    },
    {
      "epoch": 3.532736814935903,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001980078187549587,
      "loss": 1.0731,
      "step": 41061
    },
    {
      "epoch": 3.5328228512432247,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001979862678340407,
      "loss": 0.7612,
      "step": 41062
    },
    {
      "epoch": 3.5329088875505463,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019796471779645454,
      "loss": 1.5717,
      "step": 41063
    },
    {
      "epoch": 3.532994923857868,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00019794316864226257,
      "loss": 1.4234,
      "step": 41064
    },
    {
      "epoch": 3.5330809601651896,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019792162037152834,
      "loss": 1.5519,
      "step": 41065
    },
    {
      "epoch": 3.533166996472511,
      "grad_norm": 1.65625,
      "learning_rate": 0.0001979000729843144,
      "loss": 1.2104,
      "step": 41066
    },
    {
      "epoch": 3.5332530327798333,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000197878526480684,
      "loss": 0.8123,
      "step": 41067
    },
    {
      "epoch": 3.533339069087155,
      "grad_norm": 1.125,
      "learning_rate": 0.00019785698086070036,
      "loss": 1.6016,
      "step": 41068
    },
    {
      "epoch": 3.5334251053944765,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019783543612442606,
      "loss": 1.7825,
      "step": 41069
    },
    {
      "epoch": 3.533511141701798,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019781389227192437,
      "loss": 0.8949,
      "step": 41070
    },
    {
      "epoch": 3.5335971780091198,
      "grad_norm": 123.0,
      "learning_rate": 0.0001977923493032584,
      "loss": 0.8666,
      "step": 41071
    },
    {
      "epoch": 3.5336832143164414,
      "grad_norm": 1.484375,
      "learning_rate": 0.00019777080721849094,
      "loss": 1.7075,
      "step": 41072
    },
    {
      "epoch": 3.533769250623763,
      "grad_norm": 3.140625,
      "learning_rate": 0.000197749266017685,
      "loss": 0.8645,
      "step": 41073
    },
    {
      "epoch": 3.533855286931085,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001977277257009039,
      "loss": 0.7371,
      "step": 41074
    },
    {
      "epoch": 3.5339413232384067,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00019770618626821024,
      "loss": 1.2633,
      "step": 41075
    },
    {
      "epoch": 3.5340273595457283,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019768464771966737,
      "loss": 1.2339,
      "step": 41076
    },
    {
      "epoch": 3.53411339585305,
      "grad_norm": 338.0,
      "learning_rate": 0.00019766311005533794,
      "loss": 1.0299,
      "step": 41077
    },
    {
      "epoch": 3.5341994321603716,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001976415732752851,
      "loss": 0.9409,
      "step": 41078
    },
    {
      "epoch": 3.534285468467693,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000197620037379572,
      "loss": 1.1859,
      "step": 41079
    },
    {
      "epoch": 3.534371504775015,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00019759850236826133,
      "loss": 0.953,
      "step": 41080
    },
    {
      "epoch": 3.534457541082337,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001975769682414162,
      "loss": 1.4244,
      "step": 41081
    },
    {
      "epoch": 3.5345435773896585,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019755543499909984,
      "loss": 1.9626,
      "step": 41082
    },
    {
      "epoch": 3.53462961369698,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019753390264137482,
      "loss": 0.5655,
      "step": 41083
    },
    {
      "epoch": 3.5347156500043018,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001975123711683043,
      "loss": 0.684,
      "step": 41084
    },
    {
      "epoch": 3.5348016863116234,
      "grad_norm": 3.9375,
      "learning_rate": 0.00019749084057995144,
      "loss": 1.1125,
      "step": 41085
    },
    {
      "epoch": 3.5348877226189455,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019746931087637887,
      "loss": 1.2535,
      "step": 41086
    },
    {
      "epoch": 3.5349737589262666,
      "grad_norm": 7.84375,
      "learning_rate": 0.0001974477820576497,
      "loss": 1.2438,
      "step": 41087
    },
    {
      "epoch": 3.5350597952335887,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019742625412382716,
      "loss": 0.6663,
      "step": 41088
    },
    {
      "epoch": 3.5351458315409103,
      "grad_norm": 11.6875,
      "learning_rate": 0.0001974047270749737,
      "loss": 1.0503,
      "step": 41089
    },
    {
      "epoch": 3.535231867848232,
      "grad_norm": 23.375,
      "learning_rate": 0.00019738320091115282,
      "loss": 0.6712,
      "step": 41090
    },
    {
      "epoch": 3.5353179041555536,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019736167563242702,
      "loss": 1.019,
      "step": 41091
    },
    {
      "epoch": 3.535403940462875,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019734015123885947,
      "loss": 0.9005,
      "step": 41092
    },
    {
      "epoch": 3.5354899767701973,
      "grad_norm": 1.453125,
      "learning_rate": 0.00019731862773051328,
      "loss": 0.9983,
      "step": 41093
    },
    {
      "epoch": 3.5355760130775185,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019729710510745107,
      "loss": 0.7037,
      "step": 41094
    },
    {
      "epoch": 3.5356620493848405,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019727558336973594,
      "loss": 0.8339,
      "step": 41095
    },
    {
      "epoch": 3.535748085692162,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00019725406251743104,
      "loss": 1.1615,
      "step": 41096
    },
    {
      "epoch": 3.5358341219994838,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019723254255059897,
      "loss": 1.2993,
      "step": 41097
    },
    {
      "epoch": 3.5359201583068054,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019721102346930281,
      "loss": 0.8375,
      "step": 41098
    },
    {
      "epoch": 3.536006194614127,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019718950527360552,
      "loss": 0.703,
      "step": 41099
    },
    {
      "epoch": 3.536092230921449,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019716798796357,
      "loss": 0.7283,
      "step": 41100
    },
    {
      "epoch": 3.5361782672287703,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00019714647153925942,
      "loss": 1.1924,
      "step": 41101
    },
    {
      "epoch": 3.5362643035360923,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019712495600073633,
      "loss": 1.0307,
      "step": 41102
    },
    {
      "epoch": 3.536350339843414,
      "grad_norm": 1.703125,
      "learning_rate": 0.00019710344134806385,
      "loss": 1.217,
      "step": 41103
    },
    {
      "epoch": 3.5364363761507356,
      "grad_norm": 1.28125,
      "learning_rate": 0.00019708192758130505,
      "loss": 0.6203,
      "step": 41104
    },
    {
      "epoch": 3.536522412458057,
      "grad_norm": 15.125,
      "learning_rate": 0.00019706041470052255,
      "loss": 1.7637,
      "step": 41105
    },
    {
      "epoch": 3.536608448765379,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00019703890270577956,
      "loss": 0.8921,
      "step": 41106
    },
    {
      "epoch": 3.536694485072701,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001970173915971387,
      "loss": 1.252,
      "step": 41107
    },
    {
      "epoch": 3.5367805213800225,
      "grad_norm": 25.375,
      "learning_rate": 0.00019699588137466308,
      "loss": 0.7953,
      "step": 41108
    },
    {
      "epoch": 3.536866557687344,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019697437203841566,
      "loss": 0.9398,
      "step": 41109
    },
    {
      "epoch": 3.5369525939946658,
      "grad_norm": 1.125,
      "learning_rate": 0.00019695286358845915,
      "loss": 0.6905,
      "step": 41110
    },
    {
      "epoch": 3.5370386303019874,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00019693135602485656,
      "loss": 1.4908,
      "step": 41111
    },
    {
      "epoch": 3.537124666609309,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019690984934767102,
      "loss": 0.419,
      "step": 41112
    },
    {
      "epoch": 3.5372107029166306,
      "grad_norm": 6.53125,
      "learning_rate": 0.000196888343556965,
      "loss": 0.9877,
      "step": 41113
    },
    {
      "epoch": 3.5372967392239527,
      "grad_norm": 5.5,
      "learning_rate": 0.00019686683865280163,
      "loss": 1.1672,
      "step": 41114
    },
    {
      "epoch": 3.5373827755312743,
      "grad_norm": 1.0625,
      "learning_rate": 0.000196845334635244,
      "loss": 0.7881,
      "step": 41115
    },
    {
      "epoch": 3.537468811838596,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019682383150435462,
      "loss": 1.3217,
      "step": 41116
    },
    {
      "epoch": 3.5375548481459176,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001968023292601966,
      "loss": 0.9949,
      "step": 41117
    },
    {
      "epoch": 3.537640884453239,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019678082790283293,
      "loss": 1.3818,
      "step": 41118
    },
    {
      "epoch": 3.537726920760561,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019675932743232623,
      "loss": 0.7583,
      "step": 41119
    },
    {
      "epoch": 3.5378129570678825,
      "grad_norm": 1.25,
      "learning_rate": 0.00019673782784873962,
      "loss": 0.9976,
      "step": 41120
    },
    {
      "epoch": 3.5378989933752045,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019671632915213577,
      "loss": 1.0071,
      "step": 41121
    },
    {
      "epoch": 3.537985029682526,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00019669483134257765,
      "loss": 0.9013,
      "step": 41122
    },
    {
      "epoch": 3.5380710659898478,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00019667333442012835,
      "loss": 0.9263,
      "step": 41123
    },
    {
      "epoch": 3.5381571022971694,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019665183838485034,
      "loss": 1.0723,
      "step": 41124
    },
    {
      "epoch": 3.538243138604491,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019663034323680678,
      "loss": 1.1662,
      "step": 41125
    },
    {
      "epoch": 3.5383291749118126,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019660884897606057,
      "loss": 0.6289,
      "step": 41126
    },
    {
      "epoch": 3.5384152112191343,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019658735560267433,
      "loss": 0.6193,
      "step": 41127
    },
    {
      "epoch": 3.5385012475264563,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019656586311671105,
      "loss": 0.9032,
      "step": 41128
    },
    {
      "epoch": 3.538587283833778,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001965443715182338,
      "loss": 0.9101,
      "step": 41129
    },
    {
      "epoch": 3.5386733201410996,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019652288080730502,
      "loss": 0.9613,
      "step": 41130
    },
    {
      "epoch": 3.538759356448421,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000196501390983988,
      "loss": 0.9796,
      "step": 41131
    },
    {
      "epoch": 3.538845392755743,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001964799020483452,
      "loss": 1.1082,
      "step": 41132
    },
    {
      "epoch": 3.5389314290630645,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019645841400043967,
      "loss": 0.9732,
      "step": 41133
    },
    {
      "epoch": 3.539017465370386,
      "grad_norm": 1.34375,
      "learning_rate": 0.00019643692684033444,
      "loss": 1.7223,
      "step": 41134
    },
    {
      "epoch": 3.539103501677708,
      "grad_norm": 1.25,
      "learning_rate": 0.00019641544056809196,
      "loss": 0.9071,
      "step": 41135
    },
    {
      "epoch": 3.5391895379850298,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019639395518377523,
      "loss": 1.0669,
      "step": 41136
    },
    {
      "epoch": 3.5392755742923514,
      "grad_norm": 1.375,
      "learning_rate": 0.00019637247068744719,
      "loss": 1.9443,
      "step": 41137
    },
    {
      "epoch": 3.539361610599673,
      "grad_norm": 2.125,
      "learning_rate": 0.0001963509870791706,
      "loss": 0.7385,
      "step": 41138
    },
    {
      "epoch": 3.5394476469069946,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0001963295043590085,
      "loss": 1.3393,
      "step": 41139
    },
    {
      "epoch": 3.5395336832143167,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019630802252702334,
      "loss": 1.1548,
      "step": 41140
    },
    {
      "epoch": 3.539619719521638,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001962865415832782,
      "loss": 0.6771,
      "step": 41141
    },
    {
      "epoch": 3.53970575582896,
      "grad_norm": 1.125,
      "learning_rate": 0.00019626506152783596,
      "loss": 0.9844,
      "step": 41142
    },
    {
      "epoch": 3.5397917921362816,
      "grad_norm": 7.46875,
      "learning_rate": 0.00019624358236075922,
      "loss": 1.1974,
      "step": 41143
    },
    {
      "epoch": 3.539877828443603,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001962221040821109,
      "loss": 0.5789,
      "step": 41144
    },
    {
      "epoch": 3.539963864750925,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019620062669195405,
      "loss": 0.9246,
      "step": 41145
    },
    {
      "epoch": 3.5400499010582465,
      "grad_norm": 1.375,
      "learning_rate": 0.00019617915019035103,
      "loss": 1.5305,
      "step": 41146
    },
    {
      "epoch": 3.5401359373655685,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019615767457736516,
      "loss": 0.6793,
      "step": 41147
    },
    {
      "epoch": 3.5402219736728897,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001961361998530588,
      "loss": 0.7685,
      "step": 41148
    },
    {
      "epoch": 3.540308009980212,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019611472601749498,
      "loss": 0.8001,
      "step": 41149
    },
    {
      "epoch": 3.5403940462875334,
      "grad_norm": 3.5,
      "learning_rate": 0.00019609325307073662,
      "loss": 1.3724,
      "step": 41150
    },
    {
      "epoch": 3.540480082594855,
      "grad_norm": 1.4375,
      "learning_rate": 0.00019607178101284622,
      "loss": 1.605,
      "step": 41151
    },
    {
      "epoch": 3.5405661189021767,
      "grad_norm": 6.15625,
      "learning_rate": 0.0001960503098438868,
      "loss": 0.8481,
      "step": 41152
    },
    {
      "epoch": 3.5406521552094983,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001960288395639212,
      "loss": 1.6921,
      "step": 41153
    },
    {
      "epoch": 3.5407381915168203,
      "grad_norm": 1.03125,
      "learning_rate": 0.000196007370173012,
      "loss": 0.9514,
      "step": 41154
    },
    {
      "epoch": 3.5408242278241415,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001959859016712221,
      "loss": 1.1952,
      "step": 41155
    },
    {
      "epoch": 3.5409102641314636,
      "grad_norm": 1.609375,
      "learning_rate": 0.00019596443405861451,
      "loss": 1.1972,
      "step": 41156
    },
    {
      "epoch": 3.540996300438785,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00019594296733525164,
      "loss": 1.2159,
      "step": 41157
    },
    {
      "epoch": 3.541082336746107,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019592150150119642,
      "loss": 1.1695,
      "step": 41158
    },
    {
      "epoch": 3.5411683730534285,
      "grad_norm": 1.375,
      "learning_rate": 0.0001959000365565119,
      "loss": 1.2674,
      "step": 41159
    },
    {
      "epoch": 3.54125440936075,
      "grad_norm": 1.546875,
      "learning_rate": 0.0001958785725012604,
      "loss": 1.6312,
      "step": 41160
    },
    {
      "epoch": 3.541340445668072,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001958571093355051,
      "loss": 1.224,
      "step": 41161
    },
    {
      "epoch": 3.541426481975394,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019583564705930845,
      "loss": 1.2077,
      "step": 41162
    },
    {
      "epoch": 3.5415125182827154,
      "grad_norm": 3.125,
      "learning_rate": 0.0001958141856727334,
      "loss": 1.5518,
      "step": 41163
    },
    {
      "epoch": 3.541598554590037,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00019579272517584285,
      "loss": 0.8411,
      "step": 41164
    },
    {
      "epoch": 3.5416845908973587,
      "grad_norm": 1.359375,
      "learning_rate": 0.00019577126556869924,
      "loss": 1.2527,
      "step": 41165
    },
    {
      "epoch": 3.5417706272046803,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001957498068513655,
      "loss": 1.4431,
      "step": 41166
    },
    {
      "epoch": 3.541856663512002,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019572834902390458,
      "loss": 1.5502,
      "step": 41167
    },
    {
      "epoch": 3.541942699819324,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019570689208637886,
      "loss": 1.1653,
      "step": 41168
    },
    {
      "epoch": 3.5420287361266456,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019568543603885137,
      "loss": 0.8766,
      "step": 41169
    },
    {
      "epoch": 3.542114772433967,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019566398088138488,
      "loss": 1.35,
      "step": 41170
    },
    {
      "epoch": 3.542200808741289,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019564252661404197,
      "loss": 1.1028,
      "step": 41171
    },
    {
      "epoch": 3.5422868450486105,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001956210732368856,
      "loss": 1.6779,
      "step": 41172
    },
    {
      "epoch": 3.542372881355932,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019559962074997805,
      "loss": 0.609,
      "step": 41173
    },
    {
      "epoch": 3.5424589176632537,
      "grad_norm": 1.453125,
      "learning_rate": 0.00019557816915338256,
      "loss": 1.8117,
      "step": 41174
    },
    {
      "epoch": 3.542544953970576,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019555671844716195,
      "loss": 0.6374,
      "step": 41175
    },
    {
      "epoch": 3.5426309902778974,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001955352686313785,
      "loss": 0.821,
      "step": 41176
    },
    {
      "epoch": 3.542717026585219,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00019551381970609534,
      "loss": 1.0833,
      "step": 41177
    },
    {
      "epoch": 3.5428030628925407,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001954923716713749,
      "loss": 0.7428,
      "step": 41178
    },
    {
      "epoch": 3.5428890991998623,
      "grad_norm": 1.59375,
      "learning_rate": 0.00019547092452728006,
      "loss": 1.7925,
      "step": 41179
    },
    {
      "epoch": 3.542975135507184,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019544947827387365,
      "loss": 0.9936,
      "step": 41180
    },
    {
      "epoch": 3.5430611718145055,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001954280329112182,
      "loss": 1.0054,
      "step": 41181
    },
    {
      "epoch": 3.5431472081218276,
      "grad_norm": 1.25,
      "learning_rate": 0.00019540658843937643,
      "loss": 1.16,
      "step": 41182
    },
    {
      "epoch": 3.543233244429149,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019538514485841136,
      "loss": 1.0663,
      "step": 41183
    },
    {
      "epoch": 3.543319280736471,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00019536370216838533,
      "loss": 0.6282,
      "step": 41184
    },
    {
      "epoch": 3.5434053170437925,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001953422603693612,
      "loss": 0.8874,
      "step": 41185
    },
    {
      "epoch": 3.543491353351114,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00019532081946140189,
      "loss": 1.1562,
      "step": 41186
    },
    {
      "epoch": 3.5435773896584357,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019529937944456977,
      "loss": 0.5465,
      "step": 41187
    },
    {
      "epoch": 3.5436634259657573,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019527794031892786,
      "loss": 1.0826,
      "step": 41188
    },
    {
      "epoch": 3.5437494622730794,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001952565020845386,
      "loss": 1.7555,
      "step": 41189
    },
    {
      "epoch": 3.543835498580401,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019523506474146475,
      "loss": 1.1853,
      "step": 41190
    },
    {
      "epoch": 3.5439215348877227,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019521362828976919,
      "loss": 0.8761,
      "step": 41191
    },
    {
      "epoch": 3.5440075711950443,
      "grad_norm": 1.28125,
      "learning_rate": 0.00019519219272951438,
      "loss": 1.0533,
      "step": 41192
    },
    {
      "epoch": 3.544093607502366,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019517075806076308,
      "loss": 0.7216,
      "step": 41193
    },
    {
      "epoch": 3.5441796438096875,
      "grad_norm": 0.875,
      "learning_rate": 0.00019514932428357825,
      "loss": 0.6552,
      "step": 41194
    },
    {
      "epoch": 3.544265680117009,
      "grad_norm": 16.25,
      "learning_rate": 0.00019512789139802212,
      "loss": 1.5104,
      "step": 41195
    },
    {
      "epoch": 3.5443517164243312,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019510645940415766,
      "loss": 0.9819,
      "step": 41196
    },
    {
      "epoch": 3.544437752731653,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019508502830204767,
      "loss": 0.7113,
      "step": 41197
    },
    {
      "epoch": 3.5445237890389745,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001950635980917545,
      "loss": 1.2794,
      "step": 41198
    },
    {
      "epoch": 3.544609825346296,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019504216877334096,
      "loss": 1.8236,
      "step": 41199
    },
    {
      "epoch": 3.5446958616536177,
      "grad_norm": 1.34375,
      "learning_rate": 0.00019502074034687,
      "loss": 1.1083,
      "step": 41200
    },
    {
      "epoch": 3.54478189796094,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019499931281240386,
      "loss": 0.9357,
      "step": 41201
    },
    {
      "epoch": 3.544867934268261,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00019497788617000555,
      "loss": 1.3327,
      "step": 41202
    },
    {
      "epoch": 3.544953970575583,
      "grad_norm": 1.0,
      "learning_rate": 0.00019495646041973747,
      "loss": 0.8366,
      "step": 41203
    },
    {
      "epoch": 3.5450400068829047,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019493503556166242,
      "loss": 0.813,
      "step": 41204
    },
    {
      "epoch": 3.5451260431902263,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001949136115958432,
      "loss": 0.8965,
      "step": 41205
    },
    {
      "epoch": 3.545212079497548,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001948921885223422,
      "loss": 1.3657,
      "step": 41206
    },
    {
      "epoch": 3.5452981158048695,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001948707663412222,
      "loss": 0.7304,
      "step": 41207
    },
    {
      "epoch": 3.5453841521121916,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00019484934505254597,
      "loss": 1.1713,
      "step": 41208
    },
    {
      "epoch": 3.545470188419513,
      "grad_norm": 86.0,
      "learning_rate": 0.00019482792465637595,
      "loss": 0.8636,
      "step": 41209
    },
    {
      "epoch": 3.545556224726835,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00019480650515277487,
      "loss": 0.6863,
      "step": 41210
    },
    {
      "epoch": 3.5456422610341565,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019478508654180543,
      "loss": 1.2191,
      "step": 41211
    },
    {
      "epoch": 3.545728297341478,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019476366882353026,
      "loss": 1.0275,
      "step": 41212
    },
    {
      "epoch": 3.5458143336487997,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019474225199801217,
      "loss": 1.3083,
      "step": 41213
    },
    {
      "epoch": 3.5459003699561213,
      "grad_norm": 1.390625,
      "learning_rate": 0.00019472083606531337,
      "loss": 1.1157,
      "step": 41214
    },
    {
      "epoch": 3.5459864062634434,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001946994210254968,
      "loss": 1.2977,
      "step": 41215
    },
    {
      "epoch": 3.5460724425707646,
      "grad_norm": 1.625,
      "learning_rate": 0.00019467800687862524,
      "loss": 2.0533,
      "step": 41216
    },
    {
      "epoch": 3.5461584788780867,
      "grad_norm": 116.0,
      "learning_rate": 0.00019465659362476095,
      "loss": 1.3036,
      "step": 41217
    },
    {
      "epoch": 3.5462445151854083,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00019463518126396689,
      "loss": 1.0985,
      "step": 41218
    },
    {
      "epoch": 3.54633055149273,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019461376979630534,
      "loss": 1.3848,
      "step": 41219
    },
    {
      "epoch": 3.5464165878000515,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019459235922183915,
      "loss": 1.2369,
      "step": 41220
    },
    {
      "epoch": 3.546502624107373,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001945709495406311,
      "loss": 1.6428,
      "step": 41221
    },
    {
      "epoch": 3.5465886604146952,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00019454954075274344,
      "loss": 1.0961,
      "step": 41222
    },
    {
      "epoch": 3.546674696722017,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019452813285823894,
      "loss": 1.1465,
      "step": 41223
    },
    {
      "epoch": 3.5467607330293385,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019450672585718043,
      "loss": 0.9177,
      "step": 41224
    },
    {
      "epoch": 3.54684676933666,
      "grad_norm": 2.078125,
      "learning_rate": 0.0001944853197496302,
      "loss": 0.9837,
      "step": 41225
    },
    {
      "epoch": 3.5469328056439817,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019446391453565093,
      "loss": 0.9339,
      "step": 41226
    },
    {
      "epoch": 3.5470188419513033,
      "grad_norm": 1.4375,
      "learning_rate": 0.00019444251021530552,
      "loss": 1.0567,
      "step": 41227
    },
    {
      "epoch": 3.547104878258625,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001944211067886561,
      "loss": 0.9034,
      "step": 41228
    },
    {
      "epoch": 3.547190914565947,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001943997042557657,
      "loss": 1.5884,
      "step": 41229
    },
    {
      "epoch": 3.5472769508732687,
      "grad_norm": 1704.0,
      "learning_rate": 0.00019437830261669658,
      "loss": 1.5684,
      "step": 41230
    },
    {
      "epoch": 3.5473629871805903,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019435690187151144,
      "loss": 0.9435,
      "step": 41231
    },
    {
      "epoch": 3.547449023487912,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019433550202027316,
      "loss": 1.1792,
      "step": 41232
    },
    {
      "epoch": 3.5475350597952335,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019431410306304387,
      "loss": 0.9191,
      "step": 41233
    },
    {
      "epoch": 3.547621096102555,
      "grad_norm": 26.25,
      "learning_rate": 0.0001942927049998864,
      "loss": 1.2328,
      "step": 41234
    },
    {
      "epoch": 3.547707132409877,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019427130783086343,
      "loss": 1.0314,
      "step": 41235
    },
    {
      "epoch": 3.547793168717199,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00019424991155603727,
      "loss": 1.9827,
      "step": 41236
    },
    {
      "epoch": 3.5478792050245205,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001942285161754707,
      "loss": 0.6859,
      "step": 41237
    },
    {
      "epoch": 3.547965241331842,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019420712168922638,
      "loss": 0.6892,
      "step": 41238
    },
    {
      "epoch": 3.5480512776391637,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019418572809736662,
      "loss": 1.345,
      "step": 41239
    },
    {
      "epoch": 3.5481373139464853,
      "grad_norm": 1.25,
      "learning_rate": 0.0001941643353999541,
      "loss": 1.3879,
      "step": 41240
    },
    {
      "epoch": 3.548223350253807,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019414294359705158,
      "loss": 0.8881,
      "step": 41241
    },
    {
      "epoch": 3.5483093865611286,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019412155268872135,
      "loss": 1.2898,
      "step": 41242
    },
    {
      "epoch": 3.5483954228684507,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019410016267502617,
      "loss": 0.6129,
      "step": 41243
    },
    {
      "epoch": 3.5484814591757723,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019407877355602843,
      "loss": 0.9865,
      "step": 41244
    },
    {
      "epoch": 3.548567495483094,
      "grad_norm": 1.3125,
      "learning_rate": 0.00019405738533179073,
      "loss": 1.5704,
      "step": 41245
    },
    {
      "epoch": 3.5486535317904155,
      "grad_norm": 1.65625,
      "learning_rate": 0.0001940359980023757,
      "loss": 1.5384,
      "step": 41246
    },
    {
      "epoch": 3.548739568097737,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019401461156784582,
      "loss": 1.6262,
      "step": 41247
    },
    {
      "epoch": 3.548825604405059,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019399322602826392,
      "loss": 0.651,
      "step": 41248
    },
    {
      "epoch": 3.5489116407123804,
      "grad_norm": 9.625,
      "learning_rate": 0.0001939718413836921,
      "loss": 1.5379,
      "step": 41249
    },
    {
      "epoch": 3.5489976770197025,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00019395045763419318,
      "loss": 1.4603,
      "step": 41250
    },
    {
      "epoch": 3.549083713327024,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019392907477982978,
      "loss": 1.5244,
      "step": 41251
    },
    {
      "epoch": 3.5491697496343457,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019390769282066417,
      "loss": 1.05,
      "step": 41252
    },
    {
      "epoch": 3.5492557859416674,
      "grad_norm": 4.125,
      "learning_rate": 0.00019388631175675902,
      "loss": 1.0003,
      "step": 41253
    },
    {
      "epoch": 3.549341822248989,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019386493158817704,
      "loss": 0.9279,
      "step": 41254
    },
    {
      "epoch": 3.5494278585563106,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019384355231498042,
      "loss": 1.2573,
      "step": 41255
    },
    {
      "epoch": 3.5495138948636322,
      "grad_norm": 23.25,
      "learning_rate": 0.00019382217393723185,
      "loss": 1.1953,
      "step": 41256
    },
    {
      "epoch": 3.5495999311709543,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019380079645499405,
      "loss": 1.0599,
      "step": 41257
    },
    {
      "epoch": 3.549685967478276,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001937794198683292,
      "loss": 1.5103,
      "step": 41258
    },
    {
      "epoch": 3.5497720037855975,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019375804417730015,
      "loss": 0.8736,
      "step": 41259
    },
    {
      "epoch": 3.549858040092919,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001937366693819691,
      "loss": 0.8076,
      "step": 41260
    },
    {
      "epoch": 3.549944076400241,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019371529548239873,
      "loss": 1.1122,
      "step": 41261
    },
    {
      "epoch": 3.550030112707563,
      "grad_norm": 1.484375,
      "learning_rate": 0.00019369392247865175,
      "loss": 1.3594,
      "step": 41262
    },
    {
      "epoch": 3.550116149014884,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00019367255037079024,
      "loss": 1.5706,
      "step": 41263
    },
    {
      "epoch": 3.550202185322206,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019365117915887698,
      "loss": 1.5474,
      "step": 41264
    },
    {
      "epoch": 3.5502882216295277,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001936298088429746,
      "loss": 1.3012,
      "step": 41265
    },
    {
      "epoch": 3.5503742579368494,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019360843942314526,
      "loss": 0.962,
      "step": 41266
    },
    {
      "epoch": 3.550460294244171,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001935870708994517,
      "loss": 0.6535,
      "step": 41267
    },
    {
      "epoch": 3.5505463305514926,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001935657032719565,
      "loss": 1.2018,
      "step": 41268
    },
    {
      "epoch": 3.5506323668588147,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019354433654072183,
      "loss": 1.1551,
      "step": 41269
    },
    {
      "epoch": 3.550718403166136,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019352297070581036,
      "loss": 1.0431,
      "step": 41270
    },
    {
      "epoch": 3.550804439473458,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001935016057672848,
      "loss": 1.5172,
      "step": 41271
    },
    {
      "epoch": 3.5508904757807795,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00019348024172520729,
      "loss": 1.4953,
      "step": 41272
    },
    {
      "epoch": 3.550976512088101,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019345887857964063,
      "loss": 0.5438,
      "step": 41273
    },
    {
      "epoch": 3.551062548395423,
      "grad_norm": 1.25,
      "learning_rate": 0.00019343751633064688,
      "loss": 1.2182,
      "step": 41274
    },
    {
      "epoch": 3.5511485847027444,
      "grad_norm": 1.359375,
      "learning_rate": 0.00019341615497828885,
      "loss": 1.203,
      "step": 41275
    },
    {
      "epoch": 3.5512346210100665,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019339479452262908,
      "loss": 1.0331,
      "step": 41276
    },
    {
      "epoch": 3.5513206573173877,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019337343496372971,
      "loss": 0.8212,
      "step": 41277
    },
    {
      "epoch": 3.5514066936247097,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001933520763016534,
      "loss": 0.9607,
      "step": 41278
    },
    {
      "epoch": 3.5514927299320314,
      "grad_norm": 2.078125,
      "learning_rate": 0.00019333071853646288,
      "loss": 0.5692,
      "step": 41279
    },
    {
      "epoch": 3.551578766239353,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001933093616682201,
      "loss": 0.9335,
      "step": 41280
    },
    {
      "epoch": 3.5516648025466746,
      "grad_norm": 88.0,
      "learning_rate": 0.00019328800569698784,
      "loss": 1.1686,
      "step": 41281
    },
    {
      "epoch": 3.5517508388539962,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019326665062282866,
      "loss": 1.1654,
      "step": 41282
    },
    {
      "epoch": 3.5518368751613183,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019324529644580457,
      "loss": 0.7477,
      "step": 41283
    },
    {
      "epoch": 3.55192291146864,
      "grad_norm": 1.484375,
      "learning_rate": 0.00019322394316597863,
      "loss": 0.919,
      "step": 41284
    },
    {
      "epoch": 3.5520089477759615,
      "grad_norm": 1.234375,
      "learning_rate": 0.00019320259078341284,
      "loss": 0.9435,
      "step": 41285
    },
    {
      "epoch": 3.552094984083283,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019318123929816977,
      "loss": 1.669,
      "step": 41286
    },
    {
      "epoch": 3.552181020390605,
      "grad_norm": 18.625,
      "learning_rate": 0.00019315988871031209,
      "loss": 1.5513,
      "step": 41287
    },
    {
      "epoch": 3.5522670566979264,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001931385390199019,
      "loss": 1.5267,
      "step": 41288
    },
    {
      "epoch": 3.552353093005248,
      "grad_norm": 43.25,
      "learning_rate": 0.00019311719022700196,
      "loss": 1.3067,
      "step": 41289
    },
    {
      "epoch": 3.55243912931257,
      "grad_norm": 12.6875,
      "learning_rate": 0.00019309584233167444,
      "loss": 1.4218,
      "step": 41290
    },
    {
      "epoch": 3.5525251656198917,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019307449533398181,
      "loss": 1.0537,
      "step": 41291
    },
    {
      "epoch": 3.5526112019272134,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019305314923398682,
      "loss": 0.5743,
      "step": 41292
    },
    {
      "epoch": 3.552697238234535,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019303180403175147,
      "loss": 0.8733,
      "step": 41293
    },
    {
      "epoch": 3.5527832745418566,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001930104597273384,
      "loss": 1.7323,
      "step": 41294
    },
    {
      "epoch": 3.5528693108491782,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019298911632081018,
      "loss": 1.5019,
      "step": 41295
    },
    {
      "epoch": 3.5529553471565,
      "grad_norm": 1.34375,
      "learning_rate": 0.00019296777381222892,
      "loss": 1.8055,
      "step": 41296
    },
    {
      "epoch": 3.553041383463822,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001929464322016572,
      "loss": 0.9596,
      "step": 41297
    },
    {
      "epoch": 3.5531274197711435,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001929250914891576,
      "loss": 1.3551,
      "step": 41298
    },
    {
      "epoch": 3.553213456078465,
      "grad_norm": 4.0625,
      "learning_rate": 0.00019290375167479228,
      "loss": 1.5991,
      "step": 41299
    },
    {
      "epoch": 3.553299492385787,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001928824127586239,
      "loss": 0.6876,
      "step": 41300
    },
    {
      "epoch": 3.5533855286931084,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019286107474071452,
      "loss": 1.1239,
      "step": 41301
    },
    {
      "epoch": 3.55347156500043,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019283973762112685,
      "loss": 0.7838,
      "step": 41302
    },
    {
      "epoch": 3.5535576013077517,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001928184013999233,
      "loss": 0.7751,
      "step": 41303
    },
    {
      "epoch": 3.5536436376150737,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019279706607716601,
      "loss": 1.2217,
      "step": 41304
    },
    {
      "epoch": 3.5537296739223954,
      "grad_norm": 4.53125,
      "learning_rate": 0.00019277573165291762,
      "loss": 1.2155,
      "step": 41305
    },
    {
      "epoch": 3.553815710229717,
      "grad_norm": 1.546875,
      "learning_rate": 0.00019275439812724054,
      "loss": 1.3878,
      "step": 41306
    },
    {
      "epoch": 3.5539017465370386,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019273306550019698,
      "loss": 0.6719,
      "step": 41307
    },
    {
      "epoch": 3.5539877828443602,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019271173377184947,
      "loss": 0.6792,
      "step": 41308
    },
    {
      "epoch": 3.554073819151682,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019269040294226048,
      "loss": 1.3475,
      "step": 41309
    },
    {
      "epoch": 3.5541598554590035,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019266907301149216,
      "loss": 1.1853,
      "step": 41310
    },
    {
      "epoch": 3.5542458917663255,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019264774397960706,
      "loss": 0.6066,
      "step": 41311
    },
    {
      "epoch": 3.554331928073647,
      "grad_norm": 1.125,
      "learning_rate": 0.00019262641584666762,
      "loss": 1.3068,
      "step": 41312
    },
    {
      "epoch": 3.554417964380969,
      "grad_norm": 1.125,
      "learning_rate": 0.000192605088612736,
      "loss": 1.2276,
      "step": 41313
    },
    {
      "epoch": 3.5545040006882904,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019258376227787493,
      "loss": 0.8826,
      "step": 41314
    },
    {
      "epoch": 3.554590036995612,
      "grad_norm": 1.328125,
      "learning_rate": 0.00019256243684214635,
      "loss": 1.2358,
      "step": 41315
    },
    {
      "epoch": 3.554676073302934,
      "grad_norm": 244.0,
      "learning_rate": 0.00019254111230561282,
      "loss": 0.7469,
      "step": 41316
    },
    {
      "epoch": 3.5547621096102553,
      "grad_norm": 34.75,
      "learning_rate": 0.00019251978866833698,
      "loss": 1.2056,
      "step": 41317
    },
    {
      "epoch": 3.5548481459175774,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019249846593038072,
      "loss": 1.8865,
      "step": 41318
    },
    {
      "epoch": 3.554934182224899,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019247714409180666,
      "loss": 0.8637,
      "step": 41319
    },
    {
      "epoch": 3.5550202185322206,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001924558231526772,
      "loss": 1.523,
      "step": 41320
    },
    {
      "epoch": 3.5551062548395422,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019243450311305456,
      "loss": 0.84,
      "step": 41321
    },
    {
      "epoch": 3.555192291146864,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019241318397300139,
      "loss": 0.7029,
      "step": 41322
    },
    {
      "epoch": 3.555278327454186,
      "grad_norm": 1.375,
      "learning_rate": 0.00019239186573257965,
      "loss": 0.7151,
      "step": 41323
    },
    {
      "epoch": 3.555364363761507,
      "grad_norm": 20.5,
      "learning_rate": 0.00019237054839185192,
      "loss": 1.3841,
      "step": 41324
    },
    {
      "epoch": 3.555450400068829,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019234923195088062,
      "loss": 1.4175,
      "step": 41325
    },
    {
      "epoch": 3.555536436376151,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019232791640972786,
      "loss": 1.2855,
      "step": 41326
    },
    {
      "epoch": 3.5556224726834724,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00019230660176845605,
      "loss": 1.7144,
      "step": 41327
    },
    {
      "epoch": 3.555708508990794,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001922852880271278,
      "loss": 0.9837,
      "step": 41328
    },
    {
      "epoch": 3.5557945452981157,
      "grad_norm": 1.96875,
      "learning_rate": 0.00019226397518580507,
      "loss": 1.0871,
      "step": 41329
    },
    {
      "epoch": 3.5558805816054377,
      "grad_norm": 1.234375,
      "learning_rate": 0.00019224266324455053,
      "loss": 1.6155,
      "step": 41330
    },
    {
      "epoch": 3.555966617912759,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019222135220342613,
      "loss": 1.2705,
      "step": 41331
    },
    {
      "epoch": 3.556052654220081,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019220004206249447,
      "loss": 0.745,
      "step": 41332
    },
    {
      "epoch": 3.5561386905274026,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019217873282181797,
      "loss": 0.8774,
      "step": 41333
    },
    {
      "epoch": 3.5562247268347242,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019215742448145862,
      "loss": 0.7391,
      "step": 41334
    },
    {
      "epoch": 3.556310763142046,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019213611704147892,
      "loss": 0.9014,
      "step": 41335
    },
    {
      "epoch": 3.5563967994493675,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001921148105019414,
      "loss": 1.0276,
      "step": 41336
    },
    {
      "epoch": 3.5564828357566896,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019209350486290805,
      "loss": 1.0322,
      "step": 41337
    },
    {
      "epoch": 3.556568872064011,
      "grad_norm": 3.6875,
      "learning_rate": 0.00019207220012444127,
      "loss": 1.0935,
      "step": 41338
    },
    {
      "epoch": 3.556654908371333,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019205089628660356,
      "loss": 0.9592,
      "step": 41339
    },
    {
      "epoch": 3.5567409446786544,
      "grad_norm": 1.0,
      "learning_rate": 0.00019202959334945697,
      "loss": 1.1364,
      "step": 41340
    },
    {
      "epoch": 3.556826980985976,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019200829131306403,
      "loss": 1.5083,
      "step": 41341
    },
    {
      "epoch": 3.5569130172932977,
      "grad_norm": 47.0,
      "learning_rate": 0.00019198699017748683,
      "loss": 0.9719,
      "step": 41342
    },
    {
      "epoch": 3.5569990536006193,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019196568994278774,
      "loss": 1.24,
      "step": 41343
    },
    {
      "epoch": 3.5570850899079414,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001919443906090293,
      "loss": 0.9868,
      "step": 41344
    },
    {
      "epoch": 3.557171126215263,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0001919230921762734,
      "loss": 0.9928,
      "step": 41345
    },
    {
      "epoch": 3.5572571625225846,
      "grad_norm": 12.375,
      "learning_rate": 0.00019190179464458252,
      "loss": 1.2856,
      "step": 41346
    },
    {
      "epoch": 3.5573431988299062,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001918804980140192,
      "loss": 1.1834,
      "step": 41347
    },
    {
      "epoch": 3.557429235137228,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019185920228464526,
      "loss": 0.917,
      "step": 41348
    },
    {
      "epoch": 3.5575152714445495,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001918379074565233,
      "loss": 1.1683,
      "step": 41349
    },
    {
      "epoch": 3.557601307751871,
      "grad_norm": 29.5,
      "learning_rate": 0.00019181661352971565,
      "loss": 1.2938,
      "step": 41350
    },
    {
      "epoch": 3.557687344059193,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00019179532050428427,
      "loss": 1.435,
      "step": 41351
    },
    {
      "epoch": 3.557773380366515,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019177402838029168,
      "loss": 0.6779,
      "step": 41352
    },
    {
      "epoch": 3.5578594166738364,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019175273715780027,
      "loss": 0.5408,
      "step": 41353
    },
    {
      "epoch": 3.557945452981158,
      "grad_norm": 1.1875,
      "learning_rate": 0.000191731446836872,
      "loss": 0.6373,
      "step": 41354
    },
    {
      "epoch": 3.5580314892884797,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019171015741756942,
      "loss": 1.0146,
      "step": 41355
    },
    {
      "epoch": 3.5581175255958013,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001916888688999545,
      "loss": 1.0971,
      "step": 41356
    },
    {
      "epoch": 3.558203561903123,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019166758128408957,
      "loss": 1.2415,
      "step": 41357
    },
    {
      "epoch": 3.558289598210445,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001916462945700373,
      "loss": 1.3792,
      "step": 41358
    },
    {
      "epoch": 3.5583756345177666,
      "grad_norm": 3.46875,
      "learning_rate": 0.0001916250087578595,
      "loss": 1.1012,
      "step": 41359
    },
    {
      "epoch": 3.5584616708250882,
      "grad_norm": 3.375,
      "learning_rate": 0.00019160372384761871,
      "loss": 1.1145,
      "step": 41360
    },
    {
      "epoch": 3.55854770713241,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019158243983937684,
      "loss": 0.8388,
      "step": 41361
    },
    {
      "epoch": 3.5586337434397315,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001915611567331964,
      "loss": 0.666,
      "step": 41362
    },
    {
      "epoch": 3.558719779747053,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001915398745291398,
      "loss": 1.0424,
      "step": 41363
    },
    {
      "epoch": 3.5588058160543747,
      "grad_norm": 0.875,
      "learning_rate": 0.00019151859322726883,
      "loss": 1.0038,
      "step": 41364
    },
    {
      "epoch": 3.558891852361697,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019149731282764603,
      "loss": 0.6465,
      "step": 41365
    },
    {
      "epoch": 3.5589778886690184,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019147603333033369,
      "loss": 0.6539,
      "step": 41366
    },
    {
      "epoch": 3.55906392497634,
      "grad_norm": 72.5,
      "learning_rate": 0.00019145475473539387,
      "loss": 1.3495,
      "step": 41367
    },
    {
      "epoch": 3.5591499612836617,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019143347704288883,
      "loss": 0.6525,
      "step": 41368
    },
    {
      "epoch": 3.5592359975909833,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019141220025288103,
      "loss": 1.2705,
      "step": 41369
    },
    {
      "epoch": 3.559322033898305,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019139092436543236,
      "loss": 1.0114,
      "step": 41370
    },
    {
      "epoch": 3.5594080702056266,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001913696493806054,
      "loss": 1.1656,
      "step": 41371
    },
    {
      "epoch": 3.5594941065129486,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000191348375298462,
      "loss": 1.6539,
      "step": 41372
    },
    {
      "epoch": 3.5595801428202702,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019132710211906457,
      "loss": 0.6087,
      "step": 41373
    },
    {
      "epoch": 3.559666179127592,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001913058298424755,
      "loss": 0.6533,
      "step": 41374
    },
    {
      "epoch": 3.5597522154349135,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001912845584687567,
      "loss": 1.1209,
      "step": 41375
    },
    {
      "epoch": 3.559838251742235,
      "grad_norm": 1.40625,
      "learning_rate": 0.00019126328799797055,
      "loss": 1.701,
      "step": 41376
    },
    {
      "epoch": 3.559924288049557,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019124201843017936,
      "loss": 1.503,
      "step": 41377
    },
    {
      "epoch": 3.5600103243568784,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019122074976544502,
      "loss": 1.6266,
      "step": 41378
    },
    {
      "epoch": 3.5600963606642004,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00019119948200382997,
      "loss": 1.5665,
      "step": 41379
    },
    {
      "epoch": 3.560182396971522,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019117821514539658,
      "loss": 0.4046,
      "step": 41380
    },
    {
      "epoch": 3.5602684332788437,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001911569491902066,
      "loss": 1.0561,
      "step": 41381
    },
    {
      "epoch": 3.5603544695861653,
      "grad_norm": 240.0,
      "learning_rate": 0.0001911356841383225,
      "loss": 0.7905,
      "step": 41382
    },
    {
      "epoch": 3.560440505893487,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019111441998980667,
      "loss": 1.078,
      "step": 41383
    },
    {
      "epoch": 3.560526542200809,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001910931567447209,
      "loss": 1.3026,
      "step": 41384
    },
    {
      "epoch": 3.56061257850813,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019107189440312772,
      "loss": 1.6147,
      "step": 41385
    },
    {
      "epoch": 3.5606986148154522,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00019105063296508896,
      "loss": 1.2283,
      "step": 41386
    },
    {
      "epoch": 3.560784651122774,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001910293724306671,
      "loss": 1.2239,
      "step": 41387
    },
    {
      "epoch": 3.5608706874300955,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001910081127999243,
      "loss": 0.8111,
      "step": 41388
    },
    {
      "epoch": 3.560956723737417,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001909868540729226,
      "loss": 1.64,
      "step": 41389
    },
    {
      "epoch": 3.5610427600447387,
      "grad_norm": 0.875,
      "learning_rate": 0.0001909655962497242,
      "loss": 0.5913,
      "step": 41390
    },
    {
      "epoch": 3.561128796352061,
      "grad_norm": 1.421875,
      "learning_rate": 0.00019094433933039152,
      "loss": 1.3477,
      "step": 41391
    },
    {
      "epoch": 3.561214832659382,
      "grad_norm": 2.734375,
      "learning_rate": 0.0001909230833149863,
      "loss": 0.9706,
      "step": 41392
    },
    {
      "epoch": 3.561300868966704,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019090182820357104,
      "loss": 1.8711,
      "step": 41393
    },
    {
      "epoch": 3.5613869052740257,
      "grad_norm": 4.28125,
      "learning_rate": 0.00019088057399620778,
      "loss": 1.4804,
      "step": 41394
    },
    {
      "epoch": 3.5614729415813473,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019085932069295874,
      "loss": 1.2304,
      "step": 41395
    },
    {
      "epoch": 3.561558977888669,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00019083806829388623,
      "loss": 0.7739,
      "step": 41396
    },
    {
      "epoch": 3.5616450141959906,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019081681679905204,
      "loss": 0.9431,
      "step": 41397
    },
    {
      "epoch": 3.5617310505033126,
      "grad_norm": 69.5,
      "learning_rate": 0.0001907955662085185,
      "loss": 1.0711,
      "step": 41398
    },
    {
      "epoch": 3.5618170868106342,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019077431652234805,
      "loss": 0.7657,
      "step": 41399
    },
    {
      "epoch": 3.561903123117956,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019075306774060235,
      "loss": 0.601,
      "step": 41400
    },
    {
      "epoch": 3.5619891594252775,
      "grad_norm": 1.453125,
      "learning_rate": 0.000190731819863344,
      "loss": 1.0546,
      "step": 41401
    },
    {
      "epoch": 3.562075195732599,
      "grad_norm": 29.125,
      "learning_rate": 0.0001907105728906347,
      "loss": 1.0638,
      "step": 41402
    },
    {
      "epoch": 3.5621612320399207,
      "grad_norm": 1.90625,
      "learning_rate": 0.00019068932682253688,
      "loss": 0.8716,
      "step": 41403
    },
    {
      "epoch": 3.5622472683472424,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019066808165911275,
      "loss": 0.9098,
      "step": 41404
    },
    {
      "epoch": 3.5623333046545644,
      "grad_norm": 172.0,
      "learning_rate": 0.00019064683740042417,
      "loss": 0.9643,
      "step": 41405
    },
    {
      "epoch": 3.562419340961886,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001906255940465334,
      "loss": 0.9024,
      "step": 41406
    },
    {
      "epoch": 3.5625053772692077,
      "grad_norm": 4.4375,
      "learning_rate": 0.00019060435159750277,
      "loss": 0.8341,
      "step": 41407
    },
    {
      "epoch": 3.5625914135765293,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019058311005339413,
      "loss": 0.663,
      "step": 41408
    },
    {
      "epoch": 3.562677449883851,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019056186941426962,
      "loss": 1.2945,
      "step": 41409
    },
    {
      "epoch": 3.5627634861911726,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001905406296801917,
      "loss": 0.8395,
      "step": 41410
    },
    {
      "epoch": 3.562849522498494,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00019051939085122204,
      "loss": 0.7317,
      "step": 41411
    },
    {
      "epoch": 3.5629355588058162,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019049815292742318,
      "loss": 1.0833,
      "step": 41412
    },
    {
      "epoch": 3.563021595113138,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019047691590885684,
      "loss": 1.1088,
      "step": 41413
    },
    {
      "epoch": 3.5631076314204595,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019045567979558532,
      "loss": 0.9509,
      "step": 41414
    },
    {
      "epoch": 3.563193667727781,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001904344445876709,
      "loss": 1.3252,
      "step": 41415
    },
    {
      "epoch": 3.5632797040351027,
      "grad_norm": 8.8125,
      "learning_rate": 0.00019041321028517532,
      "loss": 1.6122,
      "step": 41416
    },
    {
      "epoch": 3.5633657403424244,
      "grad_norm": 5.875,
      "learning_rate": 0.00019039197688816095,
      "loss": 0.9925,
      "step": 41417
    },
    {
      "epoch": 3.563451776649746,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019037074439669,
      "loss": 0.9892,
      "step": 41418
    },
    {
      "epoch": 3.563537812957068,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0001903495128108242,
      "loss": 1.9637,
      "step": 41419
    },
    {
      "epoch": 3.5636238492643897,
      "grad_norm": 1.25,
      "learning_rate": 0.00019032828213062585,
      "loss": 1.0291,
      "step": 41420
    },
    {
      "epoch": 3.5637098855717113,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001903070523561572,
      "loss": 1.1896,
      "step": 41421
    },
    {
      "epoch": 3.563795921879033,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019028582348748008,
      "loss": 0.8918,
      "step": 41422
    },
    {
      "epoch": 3.5638819581863546,
      "grad_norm": 108.0,
      "learning_rate": 0.00019026459552465663,
      "loss": 0.6016,
      "step": 41423
    },
    {
      "epoch": 3.563967994493676,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001902433684677492,
      "loss": 1.3905,
      "step": 41424
    },
    {
      "epoch": 3.564054030800998,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001902221423168195,
      "loss": 1.7473,
      "step": 41425
    },
    {
      "epoch": 3.56414006710832,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019020091707192994,
      "loss": 0.6377,
      "step": 41426
    },
    {
      "epoch": 3.5642261034156415,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001901796927331423,
      "loss": 0.4994,
      "step": 41427
    },
    {
      "epoch": 3.564312139722963,
      "grad_norm": 2.78125,
      "learning_rate": 0.00019015846930051878,
      "loss": 1.6184,
      "step": 41428
    },
    {
      "epoch": 3.5643981760302847,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001901372467741217,
      "loss": 1.2067,
      "step": 41429
    },
    {
      "epoch": 3.5644842123376064,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019011602515401266,
      "loss": 1.0331,
      "step": 41430
    },
    {
      "epoch": 3.564570248644928,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019009480444025402,
      "loss": 0.7748,
      "step": 41431
    },
    {
      "epoch": 3.5646562849522496,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001900735846329078,
      "loss": 0.7188,
      "step": 41432
    },
    {
      "epoch": 3.5647423212595717,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00019005236573203604,
      "loss": 2.0266,
      "step": 41433
    },
    {
      "epoch": 3.5648283575668933,
      "grad_norm": 33.0,
      "learning_rate": 0.00019003114773770102,
      "loss": 1.1972,
      "step": 41434
    },
    {
      "epoch": 3.564914393874215,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019000993064996447,
      "loss": 0.9839,
      "step": 41435
    },
    {
      "epoch": 3.5650004301815366,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00018998871446888855,
      "loss": 1.5323,
      "step": 41436
    },
    {
      "epoch": 3.565086466488858,
      "grad_norm": 5.25,
      "learning_rate": 0.00018996749919453549,
      "loss": 1.0216,
      "step": 41437
    },
    {
      "epoch": 3.5651725027961803,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018994628482696708,
      "loss": 1.4943,
      "step": 41438
    },
    {
      "epoch": 3.5652585391035014,
      "grad_norm": 13.0625,
      "learning_rate": 0.00018992507136624543,
      "loss": 0.8913,
      "step": 41439
    },
    {
      "epoch": 3.5653445754108235,
      "grad_norm": 1.515625,
      "learning_rate": 0.00018990385881243287,
      "loss": 1.675,
      "step": 41440
    },
    {
      "epoch": 3.565430611718145,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000189882647165591,
      "loss": 1.8147,
      "step": 41441
    },
    {
      "epoch": 3.5655166480254668,
      "grad_norm": 1.421875,
      "learning_rate": 0.00018986143642578225,
      "loss": 0.983,
      "step": 41442
    },
    {
      "epoch": 3.5656026843327884,
      "grad_norm": 1.234375,
      "learning_rate": 0.00018984022659306832,
      "loss": 1.2141,
      "step": 41443
    },
    {
      "epoch": 3.56568872064011,
      "grad_norm": 9.1875,
      "learning_rate": 0.00018981901766751142,
      "loss": 1.2651,
      "step": 41444
    },
    {
      "epoch": 3.565774756947432,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001897978096491737,
      "loss": 1.1374,
      "step": 41445
    },
    {
      "epoch": 3.5658607932547532,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001897766025381169,
      "loss": 0.7412,
      "step": 41446
    },
    {
      "epoch": 3.5659468295620753,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001897553963344032,
      "loss": 0.9557,
      "step": 41447
    },
    {
      "epoch": 3.566032865869397,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018973419103809482,
      "loss": 0.9168,
      "step": 41448
    },
    {
      "epoch": 3.5661189021767186,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001897129866492534,
      "loss": 1.537,
      "step": 41449
    },
    {
      "epoch": 3.56620493848404,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001896917831679411,
      "loss": 1.4851,
      "step": 41450
    },
    {
      "epoch": 3.566290974791362,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001896705805942202,
      "loss": 1.1444,
      "step": 41451
    },
    {
      "epoch": 3.566377011098684,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018964937892815238,
      "loss": 1.012,
      "step": 41452
    },
    {
      "epoch": 3.566463047406005,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018962817816979989,
      "loss": 0.644,
      "step": 41453
    },
    {
      "epoch": 3.566549083713327,
      "grad_norm": 1.125,
      "learning_rate": 0.00018960697831922442,
      "loss": 1.8796,
      "step": 41454
    },
    {
      "epoch": 3.5666351200206488,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001895857793764882,
      "loss": 2.2095,
      "step": 41455
    },
    {
      "epoch": 3.5667211563279704,
      "grad_norm": 0.875,
      "learning_rate": 0.00018956458134165334,
      "loss": 1.3553,
      "step": 41456
    },
    {
      "epoch": 3.566807192635292,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018954338421478152,
      "loss": 1.126,
      "step": 41457
    },
    {
      "epoch": 3.5668932289426136,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00018952218799593495,
      "loss": 1.3921,
      "step": 41458
    },
    {
      "epoch": 3.5669792652499357,
      "grad_norm": 23552.0,
      "learning_rate": 0.00018950099268517578,
      "loss": 0.9767,
      "step": 41459
    },
    {
      "epoch": 3.5670653015572573,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001894797982825656,
      "loss": 0.8618,
      "step": 41460
    },
    {
      "epoch": 3.567151337864579,
      "grad_norm": 4.0,
      "learning_rate": 0.00018945860478816666,
      "loss": 1.708,
      "step": 41461
    },
    {
      "epoch": 3.5672373741719006,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001894374122020411,
      "loss": 1.9228,
      "step": 41462
    },
    {
      "epoch": 3.567323410479222,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001894162205242505,
      "loss": 1.8656,
      "step": 41463
    },
    {
      "epoch": 3.567409446786544,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018939502975485705,
      "loss": 1.0052,
      "step": 41464
    },
    {
      "epoch": 3.5674954830938654,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018937383989392292,
      "loss": 2.0797,
      "step": 41465
    },
    {
      "epoch": 3.5675815194011875,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00018935265094150973,
      "loss": 0.8668,
      "step": 41466
    },
    {
      "epoch": 3.567667555708509,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018933146289767976,
      "loss": 0.9666,
      "step": 41467
    },
    {
      "epoch": 3.5677535920158308,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018931027576249448,
      "loss": 0.9154,
      "step": 41468
    },
    {
      "epoch": 3.5678396283231524,
      "grad_norm": 312.0,
      "learning_rate": 0.00018928908953601641,
      "loss": 1.1631,
      "step": 41469
    },
    {
      "epoch": 3.567925664630474,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00018926790421830753,
      "loss": 1.5432,
      "step": 41470
    },
    {
      "epoch": 3.5680117009377956,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018924671980942936,
      "loss": 0.8201,
      "step": 41471
    },
    {
      "epoch": 3.5680977372451173,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018922553630944434,
      "loss": 1.0438,
      "step": 41472
    },
    {
      "epoch": 3.5681837735524393,
      "grad_norm": 0.828125,
      "learning_rate": 0.00018920435371841393,
      "loss": 1.018,
      "step": 41473
    },
    {
      "epoch": 3.568269809859761,
      "grad_norm": 4.9375,
      "learning_rate": 0.0001891831720364004,
      "loss": 2.1911,
      "step": 41474
    },
    {
      "epoch": 3.5683558461670826,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001891619912634658,
      "loss": 1.0355,
      "step": 41475
    },
    {
      "epoch": 3.568441882474404,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018914081139967176,
      "loss": 0.6662,
      "step": 41476
    },
    {
      "epoch": 3.568527918781726,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018911963244508035,
      "loss": 1.1474,
      "step": 41477
    },
    {
      "epoch": 3.5686139550890474,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018909845439975376,
      "loss": 0.7239,
      "step": 41478
    },
    {
      "epoch": 3.568699991396369,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018907727726375352,
      "loss": 1.4483,
      "step": 41479
    },
    {
      "epoch": 3.568786027703691,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018905610103714182,
      "loss": 0.8898,
      "step": 41480
    },
    {
      "epoch": 3.5688720640110128,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018903492571998066,
      "loss": 1.6345,
      "step": 41481
    },
    {
      "epoch": 3.5689581003183344,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018901375131233174,
      "loss": 0.8211,
      "step": 41482
    },
    {
      "epoch": 3.569044136625656,
      "grad_norm": 1.453125,
      "learning_rate": 0.00018899257781425727,
      "loss": 1.2321,
      "step": 41483
    },
    {
      "epoch": 3.5691301729329776,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018897140522581885,
      "loss": 0.6839,
      "step": 41484
    },
    {
      "epoch": 3.5692162092402993,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00018895023354707858,
      "loss": 0.4835,
      "step": 41485
    },
    {
      "epoch": 3.569302245547621,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018892906277809856,
      "loss": 1.4005,
      "step": 41486
    },
    {
      "epoch": 3.569388281854943,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00018890789291894035,
      "loss": 0.8126,
      "step": 41487
    },
    {
      "epoch": 3.5694743181622646,
      "grad_norm": 1.5,
      "learning_rate": 0.00018888672396966604,
      "loss": 1.5392,
      "step": 41488
    },
    {
      "epoch": 3.569560354469586,
      "grad_norm": 1.125,
      "learning_rate": 0.0001888655559303377,
      "loss": 1.4871,
      "step": 41489
    },
    {
      "epoch": 3.569646390776908,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000188844388801017,
      "loss": 1.1391,
      "step": 41490
    },
    {
      "epoch": 3.5697324270842294,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018882322258176586,
      "loss": 0.8875,
      "step": 41491
    },
    {
      "epoch": 3.5698184633915515,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00018880205727264654,
      "loss": 1.7311,
      "step": 41492
    },
    {
      "epoch": 3.5699044996988727,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018878089287372045,
      "loss": 1.0094,
      "step": 41493
    },
    {
      "epoch": 3.5699905360061948,
      "grad_norm": 440.0,
      "learning_rate": 0.00018875972938504988,
      "loss": 0.8791,
      "step": 41494
    },
    {
      "epoch": 3.5700765723135164,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018873856680669642,
      "loss": 1.0743,
      "step": 41495
    },
    {
      "epoch": 3.570162608620838,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001887174051387221,
      "loss": 1.1931,
      "step": 41496
    },
    {
      "epoch": 3.5702486449281596,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00018869624438118903,
      "loss": 2.141,
      "step": 41497
    },
    {
      "epoch": 3.5703346812354813,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018867508453415872,
      "loss": 1.1586,
      "step": 41498
    },
    {
      "epoch": 3.5704207175428033,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018865392559769324,
      "loss": 0.9737,
      "step": 41499
    },
    {
      "epoch": 3.5705067538501245,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001886327675718546,
      "loss": 1.1147,
      "step": 41500
    },
    {
      "epoch": 3.5705927901574466,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018861161045670443,
      "loss": 0.6269,
      "step": 41501
    },
    {
      "epoch": 3.570678826464768,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00018859045425230475,
      "loss": 0.991,
      "step": 41502
    },
    {
      "epoch": 3.57076486277209,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018856929895871754,
      "loss": 1.1378,
      "step": 41503
    },
    {
      "epoch": 3.5708508990794114,
      "grad_norm": 776.0,
      "learning_rate": 0.00018854814457600443,
      "loss": 0.7473,
      "step": 41504
    },
    {
      "epoch": 3.570936935386733,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018852699110422745,
      "loss": 0.7211,
      "step": 41505
    },
    {
      "epoch": 3.571022971694055,
      "grad_norm": 1.515625,
      "learning_rate": 0.00018850583854344838,
      "loss": 0.9341,
      "step": 41506
    },
    {
      "epoch": 3.5711090080013763,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018848468689372917,
      "loss": 1.0957,
      "step": 41507
    },
    {
      "epoch": 3.5711950443086984,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018846353615513184,
      "loss": 0.6223,
      "step": 41508
    },
    {
      "epoch": 3.57128108061602,
      "grad_norm": 3.46875,
      "learning_rate": 0.00018844238632771793,
      "loss": 1.1073,
      "step": 41509
    },
    {
      "epoch": 3.5713671169233416,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018842123741154944,
      "loss": 0.5021,
      "step": 41510
    },
    {
      "epoch": 3.5714531532306633,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001884000894066884,
      "loss": 1.1903,
      "step": 41511
    },
    {
      "epoch": 3.571539189537985,
      "grad_norm": 1.125,
      "learning_rate": 0.0001883789423131963,
      "loss": 1.6222,
      "step": 41512
    },
    {
      "epoch": 3.571625225845307,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018835779613113535,
      "loss": 1.0482,
      "step": 41513
    },
    {
      "epoch": 3.5717112621526286,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018833665086056712,
      "loss": 0.5295,
      "step": 41514
    },
    {
      "epoch": 3.57179729845995,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001883155065015535,
      "loss": 1.3082,
      "step": 41515
    },
    {
      "epoch": 3.571883334767272,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00018829436305415665,
      "loss": 1.397,
      "step": 41516
    },
    {
      "epoch": 3.5719693710745934,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018827322051843793,
      "loss": 1.3644,
      "step": 41517
    },
    {
      "epoch": 3.572055407381915,
      "grad_norm": 3.296875,
      "learning_rate": 0.00018825207889445943,
      "loss": 0.859,
      "step": 41518
    },
    {
      "epoch": 3.5721414436892367,
      "grad_norm": 0.875,
      "learning_rate": 0.00018823093818228314,
      "loss": 0.9587,
      "step": 41519
    },
    {
      "epoch": 3.5722274799965588,
      "grad_norm": 1.3125,
      "learning_rate": 0.00018820979838197056,
      "loss": 1.0069,
      "step": 41520
    },
    {
      "epoch": 3.5723135163038804,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001881886594935837,
      "loss": 1.1314,
      "step": 41521
    },
    {
      "epoch": 3.572399552611202,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00018816752151718452,
      "loss": 1.3826,
      "step": 41522
    },
    {
      "epoch": 3.5724855889185236,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001881463844528345,
      "loss": 0.9367,
      "step": 41523
    },
    {
      "epoch": 3.5725716252258453,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018812524830059586,
      "loss": 1.0818,
      "step": 41524
    },
    {
      "epoch": 3.572657661533167,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018810411306053,
      "loss": 1.2416,
      "step": 41525
    },
    {
      "epoch": 3.5727436978404885,
      "grad_norm": 22.125,
      "learning_rate": 0.00018808297873269898,
      "loss": 0.8234,
      "step": 41526
    },
    {
      "epoch": 3.5728297341478106,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018806184531716474,
      "loss": 1.9914,
      "step": 41527
    },
    {
      "epoch": 3.572915770455132,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018804071281398877,
      "loss": 0.7955,
      "step": 41528
    },
    {
      "epoch": 3.573001806762454,
      "grad_norm": 2.46875,
      "learning_rate": 0.00018801958122323303,
      "loss": 1.0596,
      "step": 41529
    },
    {
      "epoch": 3.5730878430697754,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001879984505449595,
      "loss": 1.031,
      "step": 41530
    },
    {
      "epoch": 3.573173879377097,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018797732077922968,
      "loss": 0.668,
      "step": 41531
    },
    {
      "epoch": 3.5732599156844187,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018795619192610546,
      "loss": 0.7099,
      "step": 41532
    },
    {
      "epoch": 3.5733459519917403,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018793506398564891,
      "loss": 0.9192,
      "step": 41533
    },
    {
      "epoch": 3.5734319882990624,
      "grad_norm": 0.875,
      "learning_rate": 0.00018791393695792146,
      "loss": 0.5452,
      "step": 41534
    },
    {
      "epoch": 3.573518024606384,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00018789281084298498,
      "loss": 0.8613,
      "step": 41535
    },
    {
      "epoch": 3.5736040609137056,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018787168564090158,
      "loss": 1.0244,
      "step": 41536
    },
    {
      "epoch": 3.5736900972210273,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001878505613517325,
      "loss": 0.5735,
      "step": 41537
    },
    {
      "epoch": 3.573776133528349,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001878294379755401,
      "loss": 0.6946,
      "step": 41538
    },
    {
      "epoch": 3.5738621698356705,
      "grad_norm": 3.859375,
      "learning_rate": 0.00018780831551238564,
      "loss": 0.7987,
      "step": 41539
    },
    {
      "epoch": 3.573948206142992,
      "grad_norm": 1.25,
      "learning_rate": 0.0001877871939623312,
      "loss": 1.6457,
      "step": 41540
    },
    {
      "epoch": 3.574034242450314,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001877660733254386,
      "loss": 0.6491,
      "step": 41541
    },
    {
      "epoch": 3.574120278757636,
      "grad_norm": 1.390625,
      "learning_rate": 0.00018774495360176924,
      "loss": 1.7086,
      "step": 41542
    },
    {
      "epoch": 3.5742063150649575,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018772383479138545,
      "loss": 1.2527,
      "step": 41543
    },
    {
      "epoch": 3.574292351372279,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018770271689434853,
      "loss": 1.5522,
      "step": 41544
    },
    {
      "epoch": 3.5743783876796007,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018768159991072047,
      "loss": 0.8371,
      "step": 41545
    },
    {
      "epoch": 3.5744644239869223,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0001876604838405631,
      "loss": 2.0085,
      "step": 41546
    },
    {
      "epoch": 3.574550460294244,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000187639368683938,
      "loss": 1.4401,
      "step": 41547
    },
    {
      "epoch": 3.574636496601566,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018761825444090692,
      "loss": 2.0161,
      "step": 41548
    },
    {
      "epoch": 3.5747225329088876,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018759714111153187,
      "loss": 1.1547,
      "step": 41549
    },
    {
      "epoch": 3.5748085692162093,
      "grad_norm": 1.125,
      "learning_rate": 0.00018757602869587424,
      "loss": 0.7491,
      "step": 41550
    },
    {
      "epoch": 3.574894605523531,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018755491719399598,
      "loss": 1.7681,
      "step": 41551
    },
    {
      "epoch": 3.5749806418308525,
      "grad_norm": 1.453125,
      "learning_rate": 0.00018753380660595898,
      "loss": 1.5585,
      "step": 41552
    },
    {
      "epoch": 3.5750666781381746,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018751269693182465,
      "loss": 1.614,
      "step": 41553
    },
    {
      "epoch": 3.5751527144454958,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018749158817165507,
      "loss": 0.6402,
      "step": 41554
    },
    {
      "epoch": 3.575238750752818,
      "grad_norm": 1.671875,
      "learning_rate": 0.00018747048032551167,
      "loss": 0.845,
      "step": 41555
    },
    {
      "epoch": 3.5753247870601395,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018744937339345636,
      "loss": 0.9779,
      "step": 41556
    },
    {
      "epoch": 3.575410823367461,
      "grad_norm": 7.53125,
      "learning_rate": 0.00018742826737555096,
      "loss": 1.4033,
      "step": 41557
    },
    {
      "epoch": 3.5754968596747827,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018740716227185695,
      "loss": 1.0765,
      "step": 41558
    },
    {
      "epoch": 3.5755828959821043,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018738605808243613,
      "loss": 0.5507,
      "step": 41559
    },
    {
      "epoch": 3.5756689322894264,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00018736495480735057,
      "loss": 0.8169,
      "step": 41560
    },
    {
      "epoch": 3.5757549685967476,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018734385244666146,
      "loss": 1.052,
      "step": 41561
    },
    {
      "epoch": 3.5758410049040696,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018732275100043078,
      "loss": 0.4718,
      "step": 41562
    },
    {
      "epoch": 3.5759270412113913,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018730165046872043,
      "loss": 0.9084,
      "step": 41563
    },
    {
      "epoch": 3.576013077518713,
      "grad_norm": 46.75,
      "learning_rate": 0.00018728055085159173,
      "loss": 2.0505,
      "step": 41564
    },
    {
      "epoch": 3.5760991138260345,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018725945214910684,
      "loss": 0.89,
      "step": 41565
    },
    {
      "epoch": 3.576185150133356,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000187238354361327,
      "loss": 1.546,
      "step": 41566
    },
    {
      "epoch": 3.576271186440678,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001872172574883142,
      "loss": 0.9671,
      "step": 41567
    },
    {
      "epoch": 3.5763572227479994,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001871961615301302,
      "loss": 0.9139,
      "step": 41568
    },
    {
      "epoch": 3.5764432590553215,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00018717506648683642,
      "loss": 0.6745,
      "step": 41569
    },
    {
      "epoch": 3.576529295362643,
      "grad_norm": 58880.0,
      "learning_rate": 0.00018715397235849474,
      "loss": 1.2312,
      "step": 41570
    },
    {
      "epoch": 3.5766153316699647,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018713287914516708,
      "loss": 1.2817,
      "step": 41571
    },
    {
      "epoch": 3.5767013679772863,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018711178684691467,
      "loss": 1.5824,
      "step": 41572
    },
    {
      "epoch": 3.576787404284608,
      "grad_norm": 1.265625,
      "learning_rate": 0.00018709069546379942,
      "loss": 1.0597,
      "step": 41573
    },
    {
      "epoch": 3.57687344059193,
      "grad_norm": 100.0,
      "learning_rate": 0.00018706960499588322,
      "loss": 1.0551,
      "step": 41574
    },
    {
      "epoch": 3.5769594768992516,
      "grad_norm": 2.015625,
      "learning_rate": 0.00018704851544322737,
      "loss": 1.4558,
      "step": 41575
    },
    {
      "epoch": 3.5770455132065733,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001870274268058938,
      "loss": 1.3162,
      "step": 41576
    },
    {
      "epoch": 3.577131549513895,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018700633908394428,
      "loss": 0.9018,
      "step": 41577
    },
    {
      "epoch": 3.5772175858212165,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018698525227744012,
      "loss": 1.4402,
      "step": 41578
    },
    {
      "epoch": 3.577303622128538,
      "grad_norm": 2.09375,
      "learning_rate": 0.00018696416638644325,
      "loss": 1.2029,
      "step": 41579
    },
    {
      "epoch": 3.5773896584358598,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018694308141101534,
      "loss": 0.9374,
      "step": 41580
    },
    {
      "epoch": 3.577475694743182,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018692199735121824,
      "loss": 0.4668,
      "step": 41581
    },
    {
      "epoch": 3.5775617310505035,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00018690091420711313,
      "loss": 1.1944,
      "step": 41582
    },
    {
      "epoch": 3.577647767357825,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00018687983197876202,
      "loss": 1.9179,
      "step": 41583
    },
    {
      "epoch": 3.5777338036651467,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018685875066622664,
      "loss": 1.3385,
      "step": 41584
    },
    {
      "epoch": 3.5778198399724683,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001868376702695684,
      "loss": 1.3972,
      "step": 41585
    },
    {
      "epoch": 3.57790587627979,
      "grad_norm": 384.0,
      "learning_rate": 0.000186816590788849,
      "loss": 0.9879,
      "step": 41586
    },
    {
      "epoch": 3.5779919125871116,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018679551222413037,
      "loss": 0.6702,
      "step": 41587
    },
    {
      "epoch": 3.5780779488944336,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018677443457547376,
      "loss": 0.639,
      "step": 41588
    },
    {
      "epoch": 3.5781639852017553,
      "grad_norm": 1.078125,
      "learning_rate": 0.000186753357842941,
      "loss": 0.9449,
      "step": 41589
    },
    {
      "epoch": 3.578250021509077,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018673228202659397,
      "loss": 1.3236,
      "step": 41590
    },
    {
      "epoch": 3.5783360578163985,
      "grad_norm": 462.0,
      "learning_rate": 0.00018671120712649388,
      "loss": 1.2166,
      "step": 41591
    },
    {
      "epoch": 3.57842209412372,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001866901331427026,
      "loss": 1.0573,
      "step": 41592
    },
    {
      "epoch": 3.5785081304310418,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001866690600752819,
      "loss": 0.8105,
      "step": 41593
    },
    {
      "epoch": 3.5785941667383634,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018664798792429306,
      "loss": 1.4835,
      "step": 41594
    },
    {
      "epoch": 3.5786802030456855,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00018662691668979814,
      "loss": 1.3971,
      "step": 41595
    },
    {
      "epoch": 3.578766239353007,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018660584637185835,
      "loss": 0.7151,
      "step": 41596
    },
    {
      "epoch": 3.5788522756603287,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001865847769705355,
      "loss": 0.647,
      "step": 41597
    },
    {
      "epoch": 3.5789383119676503,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00018656370848589137,
      "loss": 1.6756,
      "step": 41598
    },
    {
      "epoch": 3.579024348274972,
      "grad_norm": 1.0,
      "learning_rate": 0.0001865426409179873,
      "loss": 1.2818,
      "step": 41599
    },
    {
      "epoch": 3.5791103845822936,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018652157426688498,
      "loss": 0.7327,
      "step": 41600
    },
    {
      "epoch": 3.579196420889615,
      "grad_norm": 253.0,
      "learning_rate": 0.00018650050853264634,
      "loss": 1.0233,
      "step": 41601
    },
    {
      "epoch": 3.5792824571969373,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001864794437153325,
      "loss": 0.5965,
      "step": 41602
    },
    {
      "epoch": 3.579368493504259,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018645837981500534,
      "loss": 1.0429,
      "step": 41603
    },
    {
      "epoch": 3.5794545298115805,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001864373168317266,
      "loss": 1.2062,
      "step": 41604
    },
    {
      "epoch": 3.579540566118902,
      "grad_norm": 1.34375,
      "learning_rate": 0.00018641625476555757,
      "loss": 1.966,
      "step": 41605
    },
    {
      "epoch": 3.5796266024262238,
      "grad_norm": 5.65625,
      "learning_rate": 0.0001863951936165602,
      "loss": 1.2445,
      "step": 41606
    },
    {
      "epoch": 3.5797126387335454,
      "grad_norm": 1.0,
      "learning_rate": 0.0001863741333847957,
      "loss": 1.2569,
      "step": 41607
    },
    {
      "epoch": 3.579798675040867,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00018635307407032582,
      "loss": 1.3802,
      "step": 41608
    },
    {
      "epoch": 3.579884711348189,
      "grad_norm": 2.828125,
      "learning_rate": 0.00018633201567321245,
      "loss": 1.2401,
      "step": 41609
    },
    {
      "epoch": 3.5799707476555107,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001863109581935167,
      "loss": 0.7786,
      "step": 41610
    },
    {
      "epoch": 3.5800567839628323,
      "grad_norm": 1.265625,
      "learning_rate": 0.00018628990163130037,
      "loss": 1.0475,
      "step": 41611
    },
    {
      "epoch": 3.580142820270154,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001862688459866253,
      "loss": 0.7759,
      "step": 41612
    },
    {
      "epoch": 3.5802288565774756,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001862477912595526,
      "loss": 0.6307,
      "step": 41613
    },
    {
      "epoch": 3.5803148928847977,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001862267374501441,
      "loss": 1.0697,
      "step": 41614
    },
    {
      "epoch": 3.580400929192119,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001862056845584616,
      "loss": 1.2154,
      "step": 41615
    },
    {
      "epoch": 3.580486965499441,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018618463258456607,
      "loss": 0.8772,
      "step": 41616
    },
    {
      "epoch": 3.5805730018067625,
      "grad_norm": 1.0,
      "learning_rate": 0.00018616358152851982,
      "loss": 0.854,
      "step": 41617
    },
    {
      "epoch": 3.580659038114084,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018614253139038385,
      "loss": 0.8547,
      "step": 41618
    },
    {
      "epoch": 3.5807450744214058,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001861214821702199,
      "loss": 1.0205,
      "step": 41619
    },
    {
      "epoch": 3.5808311107287274,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00018610043386808984,
      "loss": 1.9658,
      "step": 41620
    },
    {
      "epoch": 3.5809171470360495,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001860793864840547,
      "loss": 0.9335,
      "step": 41621
    },
    {
      "epoch": 3.5810031833433706,
      "grad_norm": 4.15625,
      "learning_rate": 0.00018605834001817633,
      "loss": 1.3714,
      "step": 41622
    },
    {
      "epoch": 3.5810892196506927,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018603729447051642,
      "loss": 0.7317,
      "step": 41623
    },
    {
      "epoch": 3.5811752559580143,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001860162498411362,
      "loss": 1.0428,
      "step": 41624
    },
    {
      "epoch": 3.581261292265336,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018599520613009758,
      "loss": 1.0025,
      "step": 41625
    },
    {
      "epoch": 3.5813473285726576,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001859741633374617,
      "loss": 1.333,
      "step": 41626
    },
    {
      "epoch": 3.581433364879979,
      "grad_norm": 1.234375,
      "learning_rate": 0.00018595312146329037,
      "loss": 1.6457,
      "step": 41627
    },
    {
      "epoch": 3.5815194011873013,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018593208050764515,
      "loss": 1.6132,
      "step": 41628
    },
    {
      "epoch": 3.5816054374946225,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018591104047058744,
      "loss": 0.7826,
      "step": 41629
    },
    {
      "epoch": 3.5816914738019445,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001858900013521788,
      "loss": 0.8529,
      "step": 41630
    },
    {
      "epoch": 3.581777510109266,
      "grad_norm": 1.40625,
      "learning_rate": 0.00018586896315248098,
      "loss": 1.415,
      "step": 41631
    },
    {
      "epoch": 3.5818635464165878,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001858479258715552,
      "loss": 0.8294,
      "step": 41632
    },
    {
      "epoch": 3.5819495827239094,
      "grad_norm": 1.375,
      "learning_rate": 0.00018582688950946315,
      "loss": 1.1565,
      "step": 41633
    },
    {
      "epoch": 3.582035619031231,
      "grad_norm": 30.75,
      "learning_rate": 0.00018580585406626648,
      "loss": 1.0079,
      "step": 41634
    },
    {
      "epoch": 3.582121655338553,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018578481954202648,
      "loss": 0.7741,
      "step": 41635
    },
    {
      "epoch": 3.5822076916458747,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00018576378593680487,
      "loss": 0.9421,
      "step": 41636
    },
    {
      "epoch": 3.5822937279531963,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018574275325066293,
      "loss": 0.3853,
      "step": 41637
    },
    {
      "epoch": 3.582379764260518,
      "grad_norm": 744.0,
      "learning_rate": 0.00018572172148366228,
      "loss": 1.0719,
      "step": 41638
    },
    {
      "epoch": 3.5824658005678396,
      "grad_norm": 6.0,
      "learning_rate": 0.0001857006906358647,
      "loss": 1.3943,
      "step": 41639
    },
    {
      "epoch": 3.582551836875161,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018567966070733128,
      "loss": 0.9486,
      "step": 41640
    },
    {
      "epoch": 3.582637873182483,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001856586316981237,
      "loss": 0.7164,
      "step": 41641
    },
    {
      "epoch": 3.582723909489805,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018563760360830366,
      "loss": 1.4549,
      "step": 41642
    },
    {
      "epoch": 3.5828099457971265,
      "grad_norm": 2.21875,
      "learning_rate": 0.00018561657643793235,
      "loss": 0.8805,
      "step": 41643
    },
    {
      "epoch": 3.582895982104448,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018559555018707137,
      "loss": 1.8756,
      "step": 41644
    },
    {
      "epoch": 3.5829820184117698,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001855745248557824,
      "loss": 0.766,
      "step": 41645
    },
    {
      "epoch": 3.5830680547190914,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018555350044412665,
      "loss": 1.3837,
      "step": 41646
    },
    {
      "epoch": 3.583154091026413,
      "grad_norm": 0.8125,
      "learning_rate": 0.00018553247695216597,
      "loss": 1.0291,
      "step": 41647
    },
    {
      "epoch": 3.5832401273337346,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018551145437996147,
      "loss": 0.894,
      "step": 41648
    },
    {
      "epoch": 3.5833261636410567,
      "grad_norm": 231.0,
      "learning_rate": 0.00018549043272757476,
      "loss": 0.6298,
      "step": 41649
    },
    {
      "epoch": 3.5834121999483783,
      "grad_norm": 3.765625,
      "learning_rate": 0.0001854694119950675,
      "loss": 1.433,
      "step": 41650
    },
    {
      "epoch": 3.5834982362557,
      "grad_norm": 1.234375,
      "learning_rate": 0.00018544839218250092,
      "loss": 1.1177,
      "step": 41651
    },
    {
      "epoch": 3.5835842725630216,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018542737328993664,
      "loss": 0.5345,
      "step": 41652
    },
    {
      "epoch": 3.583670308870343,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018540635531743606,
      "loss": 1.5164,
      "step": 41653
    },
    {
      "epoch": 3.583756345177665,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001853853382650607,
      "loss": 1.1966,
      "step": 41654
    },
    {
      "epoch": 3.5838423814849865,
      "grad_norm": 1.296875,
      "learning_rate": 0.00018536432213287225,
      "loss": 0.9941,
      "step": 41655
    },
    {
      "epoch": 3.5839284177923085,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00018534330692093176,
      "loss": 1.0446,
      "step": 41656
    },
    {
      "epoch": 3.58401445409963,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018532229262930094,
      "loss": 0.8732,
      "step": 41657
    },
    {
      "epoch": 3.5841004904069518,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018530127925804135,
      "loss": 1.4978,
      "step": 41658
    },
    {
      "epoch": 3.5841865267142734,
      "grad_norm": 14.8125,
      "learning_rate": 0.00018528026680721417,
      "loss": 1.0467,
      "step": 41659
    },
    {
      "epoch": 3.584272563021595,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018525925527688097,
      "loss": 0.6193,
      "step": 41660
    },
    {
      "epoch": 3.5843585993289167,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018523824466710342,
      "loss": 0.5567,
      "step": 41661
    },
    {
      "epoch": 3.5844446356362383,
      "grad_norm": 131.0,
      "learning_rate": 0.00018521723497794257,
      "loss": 1.3505,
      "step": 41662
    },
    {
      "epoch": 3.5845306719435603,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018519622620946015,
      "loss": 1.305,
      "step": 41663
    },
    {
      "epoch": 3.584616708250882,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001851752183617177,
      "loss": 0.5857,
      "step": 41664
    },
    {
      "epoch": 3.5847027445582036,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001851542114347763,
      "loss": 1.2443,
      "step": 41665
    },
    {
      "epoch": 3.584788780865525,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018513320542869782,
      "loss": 0.8822,
      "step": 41666
    },
    {
      "epoch": 3.584874817172847,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018511220034354327,
      "loss": 1.143,
      "step": 41667
    },
    {
      "epoch": 3.584960853480169,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001850911961793743,
      "loss": 0.6394,
      "step": 41668
    },
    {
      "epoch": 3.58504688978749,
      "grad_norm": 22.5,
      "learning_rate": 0.0001850701929362525,
      "loss": 1.3323,
      "step": 41669
    },
    {
      "epoch": 3.585132926094812,
      "grad_norm": 1.25,
      "learning_rate": 0.00018504919061423898,
      "loss": 1.8437,
      "step": 41670
    },
    {
      "epoch": 3.585218962402134,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001850281892133953,
      "loss": 1.6447,
      "step": 41671
    },
    {
      "epoch": 3.5853049987094554,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018500718873378308,
      "loss": 0.9969,
      "step": 41672
    },
    {
      "epoch": 3.585391035016777,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018498618917546346,
      "loss": 0.7669,
      "step": 41673
    },
    {
      "epoch": 3.5854770713240987,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018496519053849786,
      "loss": 0.5501,
      "step": 41674
    },
    {
      "epoch": 3.5855631076314207,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018494419282294805,
      "loss": 1.7168,
      "step": 41675
    },
    {
      "epoch": 3.585649143938742,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018492319602887502,
      "loss": 1.0769,
      "step": 41676
    },
    {
      "epoch": 3.585735180246064,
      "grad_norm": 1.125,
      "learning_rate": 0.00018490220015634058,
      "loss": 1.5735,
      "step": 41677
    },
    {
      "epoch": 3.5858212165533856,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001848812052054057,
      "loss": 0.6401,
      "step": 41678
    },
    {
      "epoch": 3.585907252860707,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000184860211176132,
      "loss": 0.8149,
      "step": 41679
    },
    {
      "epoch": 3.585993289168029,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018483921806858105,
      "loss": 1.6501,
      "step": 41680
    },
    {
      "epoch": 3.5860793254753505,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018481822588281395,
      "loss": 1.358,
      "step": 41681
    },
    {
      "epoch": 3.5861653617826725,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018479723461889224,
      "loss": 1.2501,
      "step": 41682
    },
    {
      "epoch": 3.5862513980899937,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018477624427687745,
      "loss": 0.9404,
      "step": 41683
    },
    {
      "epoch": 3.586337434397316,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001847552548568307,
      "loss": 1.1135,
      "step": 41684
    },
    {
      "epoch": 3.5864234707046374,
      "grad_norm": 1.5,
      "learning_rate": 0.00018473426635881352,
      "loss": 1.872,
      "step": 41685
    },
    {
      "epoch": 3.586509507011959,
      "grad_norm": 19.25,
      "learning_rate": 0.00018471327878288746,
      "loss": 1.1565,
      "step": 41686
    },
    {
      "epoch": 3.5865955433192807,
      "grad_norm": 2.03125,
      "learning_rate": 0.0001846922921291136,
      "loss": 1.2485,
      "step": 41687
    },
    {
      "epoch": 3.5866815796266023,
      "grad_norm": 1.640625,
      "learning_rate": 0.00018467130639755343,
      "loss": 1.634,
      "step": 41688
    },
    {
      "epoch": 3.5867676159339243,
      "grad_norm": 21.5,
      "learning_rate": 0.0001846503215882685,
      "loss": 0.7323,
      "step": 41689
    },
    {
      "epoch": 3.586853652241246,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018462933770131975,
      "loss": 0.6861,
      "step": 41690
    },
    {
      "epoch": 3.5869396885485676,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018460835473676918,
      "loss": 0.8063,
      "step": 41691
    },
    {
      "epoch": 3.587025724855889,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001845873726946777,
      "loss": 1.4292,
      "step": 41692
    },
    {
      "epoch": 3.587111761163211,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018456639157510691,
      "loss": 1.1188,
      "step": 41693
    },
    {
      "epoch": 3.5871977974705325,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018454541137811792,
      "loss": 0.9434,
      "step": 41694
    },
    {
      "epoch": 3.587283833777854,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001845244321037723,
      "loss": 0.9707,
      "step": 41695
    },
    {
      "epoch": 3.587369870085176,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00018450345375213147,
      "loss": 1.2207,
      "step": 41696
    },
    {
      "epoch": 3.587455906392498,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018448247632325655,
      "loss": 0.7447,
      "step": 41697
    },
    {
      "epoch": 3.5875419426998194,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000184461499817209,
      "loss": 0.7603,
      "step": 41698
    },
    {
      "epoch": 3.587627979007141,
      "grad_norm": 426.0,
      "learning_rate": 0.0001844405242340504,
      "loss": 0.6485,
      "step": 41699
    },
    {
      "epoch": 3.5877140153144627,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018441954957384166,
      "loss": 1.2821,
      "step": 41700
    },
    {
      "epoch": 3.5878000516217843,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018439857583664438,
      "loss": 0.9881,
      "step": 41701
    },
    {
      "epoch": 3.587886087929106,
      "grad_norm": 1.96875,
      "learning_rate": 0.0001843776030225201,
      "loss": 0.8481,
      "step": 41702
    },
    {
      "epoch": 3.587972124236428,
      "grad_norm": 91.0,
      "learning_rate": 0.00018435663113152974,
      "loss": 1.5239,
      "step": 41703
    },
    {
      "epoch": 3.5880581605437496,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001843356601637348,
      "loss": 0.5408,
      "step": 41704
    },
    {
      "epoch": 3.588144196851071,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018431469011919693,
      "loss": 0.5998,
      "step": 41705
    },
    {
      "epoch": 3.588230233158393,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018429372099797698,
      "loss": 1.2003,
      "step": 41706
    },
    {
      "epoch": 3.5883162694657145,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018427275280013665,
      "loss": 2.0655,
      "step": 41707
    },
    {
      "epoch": 3.588402305773036,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018425178552573695,
      "loss": 1.131,
      "step": 41708
    },
    {
      "epoch": 3.5884883420803577,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018423081917483936,
      "loss": 1.0219,
      "step": 41709
    },
    {
      "epoch": 3.58857437838768,
      "grad_norm": 1.375,
      "learning_rate": 0.00018420985374750537,
      "loss": 1.1242,
      "step": 41710
    },
    {
      "epoch": 3.5886604146950014,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000184188889243796,
      "loss": 1.3513,
      "step": 41711
    },
    {
      "epoch": 3.588746451002323,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001841679256637727,
      "loss": 1.0816,
      "step": 41712
    },
    {
      "epoch": 3.5888324873096447,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018414696300749696,
      "loss": 1.2554,
      "step": 41713
    },
    {
      "epoch": 3.5889185236169663,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018412600127502976,
      "loss": 1.3281,
      "step": 41714
    },
    {
      "epoch": 3.589004559924288,
      "grad_norm": 5.53125,
      "learning_rate": 0.00018410504046643256,
      "loss": 1.8486,
      "step": 41715
    },
    {
      "epoch": 3.5890905962316095,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018408408058176684,
      "loss": 1.2909,
      "step": 41716
    },
    {
      "epoch": 3.5891766325389316,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001840631216210936,
      "loss": 1.7502,
      "step": 41717
    },
    {
      "epoch": 3.5892626688462532,
      "grad_norm": 1.0,
      "learning_rate": 0.00018404216358447445,
      "loss": 1.156,
      "step": 41718
    },
    {
      "epoch": 3.589348705153575,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018402120647197036,
      "loss": 0.6297,
      "step": 41719
    },
    {
      "epoch": 3.5894347414608965,
      "grad_norm": 13.5,
      "learning_rate": 0.0001840002502836427,
      "loss": 1.7361,
      "step": 41720
    },
    {
      "epoch": 3.589520777768218,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018397929501955313,
      "loss": 0.9795,
      "step": 41721
    },
    {
      "epoch": 3.5896068140755397,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001839583406797624,
      "loss": 0.761,
      "step": 41722
    },
    {
      "epoch": 3.5896928503828613,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00018393738726433213,
      "loss": 0.9254,
      "step": 41723
    },
    {
      "epoch": 3.5897788866901834,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018391643477332366,
      "loss": 0.8142,
      "step": 41724
    },
    {
      "epoch": 3.589864922997505,
      "grad_norm": 1.53125,
      "learning_rate": 0.000183895483206798,
      "loss": 1.0291,
      "step": 41725
    },
    {
      "epoch": 3.5899509593048267,
      "grad_norm": 0.640625,
      "learning_rate": 0.00018387453256481657,
      "loss": 0.4374,
      "step": 41726
    },
    {
      "epoch": 3.5900369956121483,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018385358284744063,
      "loss": 1.5493,
      "step": 41727
    },
    {
      "epoch": 3.59012303191947,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00018383263405473145,
      "loss": 1.5239,
      "step": 41728
    },
    {
      "epoch": 3.590209068226792,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018381168618675055,
      "loss": 1.0612,
      "step": 41729
    },
    {
      "epoch": 3.590295104534113,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018379073924355876,
      "loss": 0.7954,
      "step": 41730
    },
    {
      "epoch": 3.5903811408414352,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018376979322521757,
      "loss": 0.5991,
      "step": 41731
    },
    {
      "epoch": 3.590467177148757,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018374884813178845,
      "loss": 1.4267,
      "step": 41732
    },
    {
      "epoch": 3.5905532134560785,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018372790396333223,
      "loss": 0.6753,
      "step": 41733
    },
    {
      "epoch": 3.5906392497634,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018370696071991034,
      "loss": 1.044,
      "step": 41734
    },
    {
      "epoch": 3.5907252860707217,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018368601840158428,
      "loss": 0.6692,
      "step": 41735
    },
    {
      "epoch": 3.590811322378044,
      "grad_norm": 1.78125,
      "learning_rate": 0.00018366507700841496,
      "loss": 1.3696,
      "step": 41736
    },
    {
      "epoch": 3.590897358685365,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018364413654046392,
      "loss": 0.8116,
      "step": 41737
    },
    {
      "epoch": 3.590983394992687,
      "grad_norm": 1.125,
      "learning_rate": 0.00018362319699779202,
      "loss": 1.0483,
      "step": 41738
    },
    {
      "epoch": 3.5910694313000087,
      "grad_norm": 1032.0,
      "learning_rate": 0.0001836022583804608,
      "loss": 0.9591,
      "step": 41739
    },
    {
      "epoch": 3.5911554676073303,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00018358132068853163,
      "loss": 1.3263,
      "step": 41740
    },
    {
      "epoch": 3.591241503914652,
      "grad_norm": 28.5,
      "learning_rate": 0.00018356038392206532,
      "loss": 1.1769,
      "step": 41741
    },
    {
      "epoch": 3.5913275402219735,
      "grad_norm": 1.9609375,
      "learning_rate": 0.00018353944808112337,
      "loss": 1.1576,
      "step": 41742
    },
    {
      "epoch": 3.5914135765292956,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018351851316576718,
      "loss": 1.2776,
      "step": 41743
    },
    {
      "epoch": 3.591499612836617,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00018349757917605759,
      "loss": 1.4067,
      "step": 41744
    },
    {
      "epoch": 3.591585649143939,
      "grad_norm": 2.25,
      "learning_rate": 0.000183476646112056,
      "loss": 0.7455,
      "step": 41745
    },
    {
      "epoch": 3.5916716854512605,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018345571397382387,
      "loss": 0.8708,
      "step": 41746
    },
    {
      "epoch": 3.591757721758582,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018343478276142207,
      "loss": 1.3102,
      "step": 41747
    },
    {
      "epoch": 3.5918437580659037,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00018341385247491206,
      "loss": 0.866,
      "step": 41748
    },
    {
      "epoch": 3.5919297943732253,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018339292311435484,
      "loss": 1.646,
      "step": 41749
    },
    {
      "epoch": 3.5920158306805474,
      "grad_norm": 2.25,
      "learning_rate": 0.0001833719946798117,
      "loss": 1.4389,
      "step": 41750
    },
    {
      "epoch": 3.592101866987869,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001833510671713441,
      "loss": 1.5057,
      "step": 41751
    },
    {
      "epoch": 3.5921879032951907,
      "grad_norm": 1.28125,
      "learning_rate": 0.00018333014058901287,
      "loss": 1.2454,
      "step": 41752
    },
    {
      "epoch": 3.5922739396025123,
      "grad_norm": 224.0,
      "learning_rate": 0.0001833092149328794,
      "loss": 0.6008,
      "step": 41753
    },
    {
      "epoch": 3.592359975909834,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018328829020300503,
      "loss": 0.6996,
      "step": 41754
    },
    {
      "epoch": 3.5924460122171555,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018326736639945063,
      "loss": 0.964,
      "step": 41755
    },
    {
      "epoch": 3.592532048524477,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001832464435222776,
      "loss": 1.4029,
      "step": 41756
    },
    {
      "epoch": 3.5926180848317992,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001832255215715473,
      "loss": 0.6456,
      "step": 41757
    },
    {
      "epoch": 3.592704121139121,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00018320460054732058,
      "loss": 1.4145,
      "step": 41758
    },
    {
      "epoch": 3.5927901574464425,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018318368044965895,
      "loss": 0.6755,
      "step": 41759
    },
    {
      "epoch": 3.592876193753764,
      "grad_norm": 20.125,
      "learning_rate": 0.00018316276127862324,
      "loss": 1.1277,
      "step": 41760
    },
    {
      "epoch": 3.5929622300610857,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018314184303427482,
      "loss": 1.5509,
      "step": 41761
    },
    {
      "epoch": 3.5930482663684074,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001831209257166751,
      "loss": 1.3078,
      "step": 41762
    },
    {
      "epoch": 3.593134302675729,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018310000932588488,
      "loss": 1.329,
      "step": 41763
    },
    {
      "epoch": 3.593220338983051,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00018307909386196548,
      "loss": 0.976,
      "step": 41764
    },
    {
      "epoch": 3.5933063752903727,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001830581793249781,
      "loss": 1.358,
      "step": 41765
    },
    {
      "epoch": 3.5933924115976943,
      "grad_norm": 1.71875,
      "learning_rate": 0.00018303726571498386,
      "loss": 1.4383,
      "step": 41766
    },
    {
      "epoch": 3.593478447905016,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018301635303204417,
      "loss": 1.1226,
      "step": 41767
    },
    {
      "epoch": 3.5935644842123375,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018299544127621986,
      "loss": 0.9198,
      "step": 41768
    },
    {
      "epoch": 3.593650520519659,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018297453044757218,
      "loss": 1.0104,
      "step": 41769
    },
    {
      "epoch": 3.593736556826981,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00018295362054616255,
      "loss": 1.0762,
      "step": 41770
    },
    {
      "epoch": 3.593822593134303,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001829327115720517,
      "loss": 1.8143,
      "step": 41771
    },
    {
      "epoch": 3.5939086294416245,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018291180352530108,
      "loss": 1.232,
      "step": 41772
    },
    {
      "epoch": 3.593994665748946,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001828908964059719,
      "loss": 1.2854,
      "step": 41773
    },
    {
      "epoch": 3.5940807020562677,
      "grad_norm": 418.0,
      "learning_rate": 0.00018286999021412497,
      "loss": 1.3402,
      "step": 41774
    },
    {
      "epoch": 3.5941667383635894,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018284908494982166,
      "loss": 1.0194,
      "step": 41775
    },
    {
      "epoch": 3.594252774670911,
      "grad_norm": 3.421875,
      "learning_rate": 0.00018282818061312329,
      "loss": 0.6115,
      "step": 41776
    },
    {
      "epoch": 3.5943388109782326,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018280727720409057,
      "loss": 1.4247,
      "step": 41777
    },
    {
      "epoch": 3.5944248472855547,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018278637472278508,
      "loss": 1.0324,
      "step": 41778
    },
    {
      "epoch": 3.5945108835928763,
      "grad_norm": 1.265625,
      "learning_rate": 0.00018276547316926762,
      "loss": 1.2321,
      "step": 41779
    },
    {
      "epoch": 3.594596919900198,
      "grad_norm": 1.421875,
      "learning_rate": 0.00018274457254359937,
      "loss": 0.8944,
      "step": 41780
    },
    {
      "epoch": 3.5946829562075195,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001827236728458418,
      "loss": 1.0343,
      "step": 41781
    },
    {
      "epoch": 3.594768992514841,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018270277407605556,
      "loss": 1.3027,
      "step": 41782
    },
    {
      "epoch": 3.594855028822163,
      "grad_norm": 3.59375,
      "learning_rate": 0.000182681876234302,
      "loss": 1.2981,
      "step": 41783
    },
    {
      "epoch": 3.5949410651294844,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018266097932064242,
      "loss": 1.0932,
      "step": 41784
    },
    {
      "epoch": 3.5950271014368065,
      "grad_norm": 31.125,
      "learning_rate": 0.0001826400833351376,
      "loss": 0.7804,
      "step": 41785
    },
    {
      "epoch": 3.595113137744128,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001826191882778488,
      "loss": 1.348,
      "step": 41786
    },
    {
      "epoch": 3.5951991740514497,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018259829414883728,
      "loss": 0.3506,
      "step": 41787
    },
    {
      "epoch": 3.5952852103587714,
      "grad_norm": 14.875,
      "learning_rate": 0.0001825774009481639,
      "loss": 1.5992,
      "step": 41788
    },
    {
      "epoch": 3.595371246666093,
      "grad_norm": 2.421875,
      "learning_rate": 0.00018255650867589002,
      "loss": 0.8533,
      "step": 41789
    },
    {
      "epoch": 3.595457282973415,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018253561733207646,
      "loss": 0.8625,
      "step": 41790
    },
    {
      "epoch": 3.5955433192807362,
      "grad_norm": 8.625,
      "learning_rate": 0.00018251472691678444,
      "loss": 1.4056,
      "step": 41791
    },
    {
      "epoch": 3.5956293555880583,
      "grad_norm": 5.96875,
      "learning_rate": 0.0001824938374300753,
      "loss": 1.2176,
      "step": 41792
    },
    {
      "epoch": 3.59571539189538,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018247294887200976,
      "loss": 0.6475,
      "step": 41793
    },
    {
      "epoch": 3.5958014282027015,
      "grad_norm": 1.84375,
      "learning_rate": 0.00018245206124264913,
      "loss": 1.1587,
      "step": 41794
    },
    {
      "epoch": 3.595887464510023,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00018243117454205454,
      "loss": 0.5686,
      "step": 41795
    },
    {
      "epoch": 3.595973500817345,
      "grad_norm": 1.25,
      "learning_rate": 0.00018241028877028694,
      "loss": 1.2804,
      "step": 41796
    },
    {
      "epoch": 3.596059537124667,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001823894039274074,
      "loss": 0.817,
      "step": 41797
    },
    {
      "epoch": 3.596145573431988,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018236852001347727,
      "loss": 0.6939,
      "step": 41798
    },
    {
      "epoch": 3.59623160973931,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00018234763702855729,
      "loss": 1.2422,
      "step": 41799
    },
    {
      "epoch": 3.5963176460466317,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001823267549727087,
      "loss": 0.8385,
      "step": 41800
    },
    {
      "epoch": 3.5964036823539534,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001823058738459925,
      "loss": 0.8112,
      "step": 41801
    },
    {
      "epoch": 3.596489718661275,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001822849936484699,
      "loss": 0.7592,
      "step": 41802
    },
    {
      "epoch": 3.5965757549685966,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018226411438020202,
      "loss": 0.5221,
      "step": 41803
    },
    {
      "epoch": 3.5966617912759187,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018224323604124965,
      "loss": 1.4905,
      "step": 41804
    },
    {
      "epoch": 3.59674782758324,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018222235863167425,
      "loss": 1.3543,
      "step": 41805
    },
    {
      "epoch": 3.596833863890562,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018220148215153641,
      "loss": 0.9444,
      "step": 41806
    },
    {
      "epoch": 3.5969199001978835,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001821806066008974,
      "loss": 1.2864,
      "step": 41807
    },
    {
      "epoch": 3.597005936505205,
      "grad_norm": 32.0,
      "learning_rate": 0.00018215973197981854,
      "loss": 1.9387,
      "step": 41808
    },
    {
      "epoch": 3.597091972812527,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018213885828836048,
      "loss": 1.1383,
      "step": 41809
    },
    {
      "epoch": 3.5971780091198484,
      "grad_norm": 24.0,
      "learning_rate": 0.00018211798552658436,
      "loss": 0.7853,
      "step": 41810
    },
    {
      "epoch": 3.5972640454271705,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018209711369455156,
      "loss": 0.5475,
      "step": 41811
    },
    {
      "epoch": 3.597350081734492,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018207624279232271,
      "loss": 1.1888,
      "step": 41812
    },
    {
      "epoch": 3.5974361180418137,
      "grad_norm": 28.75,
      "learning_rate": 0.00018205537281995894,
      "loss": 0.637,
      "step": 41813
    },
    {
      "epoch": 3.5975221543491354,
      "grad_norm": 2.796875,
      "learning_rate": 0.00018203450377752162,
      "loss": 1.2858,
      "step": 41814
    },
    {
      "epoch": 3.597608190656457,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018201363566507127,
      "loss": 0.8422,
      "step": 41815
    },
    {
      "epoch": 3.5976942269637786,
      "grad_norm": 940.0,
      "learning_rate": 0.00018199276848266928,
      "loss": 0.8866,
      "step": 41816
    },
    {
      "epoch": 3.5977802632711002,
      "grad_norm": 1.3125,
      "learning_rate": 0.00018197190223037674,
      "loss": 1.3531,
      "step": 41817
    },
    {
      "epoch": 3.5978662995784223,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00018195103690825433,
      "loss": 1.3061,
      "step": 41818
    },
    {
      "epoch": 3.597952335885744,
      "grad_norm": 17.0,
      "learning_rate": 0.00018193017251636346,
      "loss": 1.4874,
      "step": 41819
    },
    {
      "epoch": 3.5980383721930655,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018190930905476484,
      "loss": 1.1905,
      "step": 41820
    },
    {
      "epoch": 3.598124408500387,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001818884465235196,
      "loss": 1.149,
      "step": 41821
    },
    {
      "epoch": 3.598210444807709,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018186758492268895,
      "loss": 1.5593,
      "step": 41822
    },
    {
      "epoch": 3.5982964811150304,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018184672425233356,
      "loss": 1.4567,
      "step": 41823
    },
    {
      "epoch": 3.598382517422352,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018182586451251465,
      "loss": 1.0376,
      "step": 41824
    },
    {
      "epoch": 3.598468553729674,
      "grad_norm": 9.625,
      "learning_rate": 0.00018180500570329339,
      "loss": 1.3548,
      "step": 41825
    },
    {
      "epoch": 3.5985545900369957,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001817841478247304,
      "loss": 1.3538,
      "step": 41826
    },
    {
      "epoch": 3.5986406263443174,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001817632908768869,
      "loss": 0.7507,
      "step": 41827
    },
    {
      "epoch": 3.598726662651639,
      "grad_norm": 1.375,
      "learning_rate": 0.00018174243485982405,
      "loss": 1.4001,
      "step": 41828
    },
    {
      "epoch": 3.5988126989589606,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018172157977360248,
      "loss": 0.6343,
      "step": 41829
    },
    {
      "epoch": 3.5988987352662822,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001817007256182836,
      "loss": 2.2407,
      "step": 41830
    },
    {
      "epoch": 3.598984771573604,
      "grad_norm": 1.34375,
      "learning_rate": 0.00018167987239392803,
      "loss": 0.8318,
      "step": 41831
    },
    {
      "epoch": 3.599070807880926,
      "grad_norm": 1.265625,
      "learning_rate": 0.00018165902010059686,
      "loss": 1.145,
      "step": 41832
    },
    {
      "epoch": 3.5991568441882476,
      "grad_norm": 816.0,
      "learning_rate": 0.00018163816873835136,
      "loss": 1.2587,
      "step": 41833
    },
    {
      "epoch": 3.599242880495569,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018161731830725213,
      "loss": 1.4583,
      "step": 41834
    },
    {
      "epoch": 3.599328916802891,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018159646880736035,
      "loss": 0.733,
      "step": 41835
    },
    {
      "epoch": 3.5994149531102124,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001815756202387371,
      "loss": 1.1961,
      "step": 41836
    },
    {
      "epoch": 3.599500989417534,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018155477260144303,
      "loss": 0.8282,
      "step": 41837
    },
    {
      "epoch": 3.5995870257248557,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00018153392589553935,
      "loss": 1.0257,
      "step": 41838
    },
    {
      "epoch": 3.5996730620321777,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018151308012108702,
      "loss": 0.3458,
      "step": 41839
    },
    {
      "epoch": 3.5997590983394994,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00018149223527814696,
      "loss": 0.7303,
      "step": 41840
    },
    {
      "epoch": 3.599845134646821,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018147139136678037,
      "loss": 1.2236,
      "step": 41841
    },
    {
      "epoch": 3.5999311709541426,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00018145054838704784,
      "loss": 0.9276,
      "step": 41842
    },
    {
      "epoch": 3.6000172072614642,
      "grad_norm": 4.0625,
      "learning_rate": 0.00018142970633901046,
      "loss": 1.0221,
      "step": 41843
    },
    {
      "epoch": 3.6001032435687863,
      "grad_norm": 4.90625,
      "learning_rate": 0.00018140886522272944,
      "loss": 0.9703,
      "step": 41844
    },
    {
      "epoch": 3.6001892798761075,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001813880250382653,
      "loss": 1.1846,
      "step": 41845
    },
    {
      "epoch": 3.6002753161834296,
      "grad_norm": 1.0,
      "learning_rate": 0.0001813671857856794,
      "loss": 0.8748,
      "step": 41846
    },
    {
      "epoch": 3.600361352490751,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018134634746503237,
      "loss": 1.1353,
      "step": 41847
    },
    {
      "epoch": 3.600447388798073,
      "grad_norm": 4.75,
      "learning_rate": 0.00018132551007638532,
      "loss": 1.6341,
      "step": 41848
    },
    {
      "epoch": 3.6005334251053944,
      "grad_norm": 19.5,
      "learning_rate": 0.0001813046736197993,
      "loss": 1.6169,
      "step": 41849
    },
    {
      "epoch": 3.600619461412716,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018128383809533499,
      "loss": 0.7429,
      "step": 41850
    },
    {
      "epoch": 3.600705497720038,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018126300350305348,
      "loss": 1.2829,
      "step": 41851
    },
    {
      "epoch": 3.6007915340273593,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018124216984301584,
      "loss": 0.9759,
      "step": 41852
    },
    {
      "epoch": 3.6008775703346814,
      "grad_norm": 1.296875,
      "learning_rate": 0.00018122133711528265,
      "loss": 1.1247,
      "step": 41853
    },
    {
      "epoch": 3.600963606642003,
      "grad_norm": 2.125,
      "learning_rate": 0.0001812005053199151,
      "loss": 0.9332,
      "step": 41854
    },
    {
      "epoch": 3.6010496429493246,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001811796744569742,
      "loss": 0.8162,
      "step": 41855
    },
    {
      "epoch": 3.6011356792566462,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001811588445265206,
      "loss": 1.1592,
      "step": 41856
    },
    {
      "epoch": 3.601221715563968,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018113801552861535,
      "loss": 1.1628,
      "step": 41857
    },
    {
      "epoch": 3.60130775187129,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00018111718746331961,
      "loss": 2.0965,
      "step": 41858
    },
    {
      "epoch": 3.601393788178611,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001810963603306939,
      "loss": 1.6429,
      "step": 41859
    },
    {
      "epoch": 3.601479824485933,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001810755341307994,
      "loss": 0.6439,
      "step": 41860
    },
    {
      "epoch": 3.601565860793255,
      "grad_norm": 20.125,
      "learning_rate": 0.00018105470886369684,
      "loss": 0.6997,
      "step": 41861
    },
    {
      "epoch": 3.6016518971005764,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018103388452944718,
      "loss": 0.4259,
      "step": 41862
    },
    {
      "epoch": 3.601737933407898,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001810130611281116,
      "loss": 1.2357,
      "step": 41863
    },
    {
      "epoch": 3.6018239697152197,
      "grad_norm": 1.3125,
      "learning_rate": 0.00018099223865975056,
      "loss": 1.7956,
      "step": 41864
    },
    {
      "epoch": 3.6019100060225417,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001809714171244252,
      "loss": 0.924,
      "step": 41865
    },
    {
      "epoch": 3.6019960423298634,
      "grad_norm": 3.125,
      "learning_rate": 0.00018095059652219658,
      "loss": 1.3212,
      "step": 41866
    },
    {
      "epoch": 3.602082078637185,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018092977685312523,
      "loss": 1.469,
      "step": 41867
    },
    {
      "epoch": 3.6021681149445066,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001809089581172722,
      "loss": 1.0018,
      "step": 41868
    },
    {
      "epoch": 3.6022541512518282,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001808881403146986,
      "loss": 0.4061,
      "step": 41869
    },
    {
      "epoch": 3.60234018755915,
      "grad_norm": 20.125,
      "learning_rate": 0.00018086732344546497,
      "loss": 1.054,
      "step": 41870
    },
    {
      "epoch": 3.6024262238664715,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001808465075096325,
      "loss": 1.6614,
      "step": 41871
    },
    {
      "epoch": 3.6025122601737936,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018082569250726178,
      "loss": 1.3735,
      "step": 41872
    },
    {
      "epoch": 3.602598296481115,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018080487843841382,
      "loss": 0.859,
      "step": 41873
    },
    {
      "epoch": 3.602684332788437,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001807840653031496,
      "loss": 0.7176,
      "step": 41874
    },
    {
      "epoch": 3.6027703690957584,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001807632531015297,
      "loss": 0.6204,
      "step": 41875
    },
    {
      "epoch": 3.60285640540308,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018074244183361544,
      "loss": 1.0286,
      "step": 41876
    },
    {
      "epoch": 3.6029424417104017,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001807216314994673,
      "loss": 0.4987,
      "step": 41877
    },
    {
      "epoch": 3.6030284780177233,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018070082209914635,
      "loss": 0.6117,
      "step": 41878
    },
    {
      "epoch": 3.6031145143250454,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00018068001363271346,
      "loss": 0.6589,
      "step": 41879
    },
    {
      "epoch": 3.603200550632367,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018065920610022934,
      "loss": 0.6609,
      "step": 41880
    },
    {
      "epoch": 3.6032865869396886,
      "grad_norm": 35.25,
      "learning_rate": 0.00018063839950175492,
      "loss": 0.7298,
      "step": 41881
    },
    {
      "epoch": 3.6033726232470102,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00018061759383735122,
      "loss": 1.2352,
      "step": 41882
    },
    {
      "epoch": 3.603458659554332,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018059678910707877,
      "loss": 1.0979,
      "step": 41883
    },
    {
      "epoch": 3.6035446958616535,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001805759853109986,
      "loss": 0.8864,
      "step": 41884
    },
    {
      "epoch": 3.603630732168975,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00018055518244917173,
      "loss": 1.575,
      "step": 41885
    },
    {
      "epoch": 3.603716768476297,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018053438052165866,
      "loss": 1.5992,
      "step": 41886
    },
    {
      "epoch": 3.603802804783619,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018051357952852037,
      "loss": 1.3509,
      "step": 41887
    },
    {
      "epoch": 3.6038888410909404,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001804927794698179,
      "loss": 0.5236,
      "step": 41888
    },
    {
      "epoch": 3.603974877398262,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018047198034561174,
      "loss": 0.6433,
      "step": 41889
    },
    {
      "epoch": 3.6040609137055837,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000180451182155963,
      "loss": 0.8786,
      "step": 41890
    },
    {
      "epoch": 3.6041469500129053,
      "grad_norm": 1.59375,
      "learning_rate": 0.00018043038490093234,
      "loss": 1.5343,
      "step": 41891
    },
    {
      "epoch": 3.604232986320227,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001804095885805806,
      "loss": 0.9821,
      "step": 41892
    },
    {
      "epoch": 3.604319022627549,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018038879319496882,
      "loss": 0.7831,
      "step": 41893
    },
    {
      "epoch": 3.6044050589348706,
      "grad_norm": 1.0,
      "learning_rate": 0.0001803679987441575,
      "loss": 1.0145,
      "step": 41894
    },
    {
      "epoch": 3.6044910952421922,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018034720522820764,
      "loss": 1.0752,
      "step": 41895
    },
    {
      "epoch": 3.604577131549514,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001803264126471802,
      "loss": 0.8784,
      "step": 41896
    },
    {
      "epoch": 3.6046631678568355,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018030562100113562,
      "loss": 1.127,
      "step": 41897
    },
    {
      "epoch": 3.604749204164157,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018028483029013491,
      "loss": 1.0636,
      "step": 41898
    },
    {
      "epoch": 3.6048352404714787,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018026404051423912,
      "loss": 0.9352,
      "step": 41899
    },
    {
      "epoch": 3.604921276778801,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018024325167350863,
      "loss": 0.6784,
      "step": 41900
    },
    {
      "epoch": 3.6050073130861224,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018022246376800461,
      "loss": 1.0977,
      "step": 41901
    },
    {
      "epoch": 3.605093349393444,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018020167679778755,
      "loss": 1.2289,
      "step": 41902
    },
    {
      "epoch": 3.6051793857007657,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018018089076291831,
      "loss": 0.8911,
      "step": 41903
    },
    {
      "epoch": 3.6052654220080873,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018016010566345803,
      "loss": 0.6987,
      "step": 41904
    },
    {
      "epoch": 3.6053514583154094,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000180139321499467,
      "loss": 0.8726,
      "step": 41905
    },
    {
      "epoch": 3.6054374946227306,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018011853827100627,
      "loss": 0.562,
      "step": 41906
    },
    {
      "epoch": 3.6055235309300526,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018009775597813677,
      "loss": 1.0444,
      "step": 41907
    },
    {
      "epoch": 3.6056095672373742,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018007697462091892,
      "loss": 1.5015,
      "step": 41908
    },
    {
      "epoch": 3.605695603544696,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001800561941994137,
      "loss": 0.8646,
      "step": 41909
    },
    {
      "epoch": 3.6057816398520175,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018003541471368212,
      "loss": 1.8301,
      "step": 41910
    },
    {
      "epoch": 3.605867676159339,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0001800146361637845,
      "loss": 1.0577,
      "step": 41911
    },
    {
      "epoch": 3.605953712466661,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017999385854978184,
      "loss": 1.1061,
      "step": 41912
    },
    {
      "epoch": 3.6060397487739824,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00017997308187173494,
      "loss": 0.8679,
      "step": 41913
    },
    {
      "epoch": 3.6061257850813044,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017995230612970447,
      "loss": 1.0736,
      "step": 41914
    },
    {
      "epoch": 3.606211821388626,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017993153132375144,
      "loss": 1.0798,
      "step": 41915
    },
    {
      "epoch": 3.6062978576959477,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001799107574539363,
      "loss": 1.1659,
      "step": 41916
    },
    {
      "epoch": 3.6063838940032693,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017988998452032014,
      "loss": 0.9915,
      "step": 41917
    },
    {
      "epoch": 3.606469930310591,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001798692125229633,
      "loss": 1.4341,
      "step": 41918
    },
    {
      "epoch": 3.606555966617913,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017984844146192675,
      "loss": 0.9865,
      "step": 41919
    },
    {
      "epoch": 3.606642002925234,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00017982767133727142,
      "loss": 1.8544,
      "step": 41920
    },
    {
      "epoch": 3.6067280392325562,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001798069021490577,
      "loss": 0.8291,
      "step": 41921
    },
    {
      "epoch": 3.606814075539878,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001797861338973465,
      "loss": 1.4691,
      "step": 41922
    },
    {
      "epoch": 3.6069001118471995,
      "grad_norm": 1.265625,
      "learning_rate": 0.00017976536658219878,
      "loss": 1.4287,
      "step": 41923
    },
    {
      "epoch": 3.606986148154521,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001797446002036749,
      "loss": 0.9468,
      "step": 41924
    },
    {
      "epoch": 3.6070721844618427,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001797238347618358,
      "loss": 1.0519,
      "step": 41925
    },
    {
      "epoch": 3.607158220769165,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00017970307025674232,
      "loss": 1.2567,
      "step": 41926
    },
    {
      "epoch": 3.6072442570764864,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017968230668845492,
      "loss": 1.4651,
      "step": 41927
    },
    {
      "epoch": 3.607330293383808,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001796615440570345,
      "loss": 1.5714,
      "step": 41928
    },
    {
      "epoch": 3.6074163296911297,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001796407823625419,
      "loss": 0.5349,
      "step": 41929
    },
    {
      "epoch": 3.6075023659984513,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017962002160503755,
      "loss": 0.8094,
      "step": 41930
    },
    {
      "epoch": 3.607588402305773,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00017959926178458247,
      "loss": 0.6229,
      "step": 41931
    },
    {
      "epoch": 3.6076744386130946,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017957850290123713,
      "loss": 1.0059,
      "step": 41932
    },
    {
      "epoch": 3.6077604749204166,
      "grad_norm": 1.34375,
      "learning_rate": 0.00017955774495506227,
      "loss": 1.8059,
      "step": 41933
    },
    {
      "epoch": 3.6078465112277383,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001795369879461189,
      "loss": 1.6632,
      "step": 41934
    },
    {
      "epoch": 3.60793254753506,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001795162318744673,
      "loss": 0.792,
      "step": 41935
    },
    {
      "epoch": 3.6080185838423815,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017949547674016842,
      "loss": 0.9282,
      "step": 41936
    },
    {
      "epoch": 3.608104620149703,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017947472254328312,
      "loss": 1.2889,
      "step": 41937
    },
    {
      "epoch": 3.6081906564570247,
      "grad_norm": 3.515625,
      "learning_rate": 0.00017945396928387174,
      "loss": 1.1824,
      "step": 41938
    },
    {
      "epoch": 3.6082766927643464,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017943321696199516,
      "loss": 0.8838,
      "step": 41939
    },
    {
      "epoch": 3.6083627290716684,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00017941246557771424,
      "loss": 1.6774,
      "step": 41940
    },
    {
      "epoch": 3.60844876537899,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00017939171513108932,
      "loss": 0.9669,
      "step": 41941
    },
    {
      "epoch": 3.6085348016863117,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017937096562218146,
      "loss": 1.0343,
      "step": 41942
    },
    {
      "epoch": 3.6086208379936333,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017935021705105098,
      "loss": 0.6102,
      "step": 41943
    },
    {
      "epoch": 3.608706874300955,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001793294694177588,
      "loss": 0.9654,
      "step": 41944
    },
    {
      "epoch": 3.6087929106082766,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001793087227223657,
      "loss": 1.1402,
      "step": 41945
    },
    {
      "epoch": 3.608878946915598,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00017928797696493206,
      "loss": 0.967,
      "step": 41946
    },
    {
      "epoch": 3.6089649832229203,
      "grad_norm": 1.109375,
      "learning_rate": 0.00017926723214551865,
      "loss": 1.3028,
      "step": 41947
    },
    {
      "epoch": 3.609051019530242,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017924648826418645,
      "loss": 0.8907,
      "step": 41948
    },
    {
      "epoch": 3.6091370558375635,
      "grad_norm": 1.484375,
      "learning_rate": 0.0001792257453209955,
      "loss": 1.6692,
      "step": 41949
    },
    {
      "epoch": 3.609223092144885,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001792050033160072,
      "loss": 1.1322,
      "step": 41950
    },
    {
      "epoch": 3.6093091284522067,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017918426224928174,
      "loss": 0.9211,
      "step": 41951
    },
    {
      "epoch": 3.6093951647595284,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00017916352212087988,
      "loss": 0.701,
      "step": 41952
    },
    {
      "epoch": 3.60948120106685,
      "grad_norm": 10.1875,
      "learning_rate": 0.0001791427829308625,
      "loss": 1.263,
      "step": 41953
    },
    {
      "epoch": 3.609567237374172,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017912204467928996,
      "loss": 1.3151,
      "step": 41954
    },
    {
      "epoch": 3.6096532736814937,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000179101307366223,
      "loss": 0.656,
      "step": 41955
    },
    {
      "epoch": 3.6097393099888153,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017908057099172244,
      "loss": 1.0937,
      "step": 41956
    },
    {
      "epoch": 3.609825346296137,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017905983555584865,
      "loss": 0.6647,
      "step": 41957
    },
    {
      "epoch": 3.6099113826034586,
      "grad_norm": 1.640625,
      "learning_rate": 0.00017903910105866256,
      "loss": 2.104,
      "step": 41958
    },
    {
      "epoch": 3.60999741891078,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017901836750022453,
      "loss": 1.3642,
      "step": 41959
    },
    {
      "epoch": 3.610083455218102,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00017899763488059533,
      "loss": 1.2128,
      "step": 41960
    },
    {
      "epoch": 3.610169491525424,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017897690319983585,
      "loss": 1.5891,
      "step": 41961
    },
    {
      "epoch": 3.6102555278327455,
      "grad_norm": 1.21875,
      "learning_rate": 0.00017895617245800626,
      "loss": 1.3818,
      "step": 41962
    },
    {
      "epoch": 3.610341564140067,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017893544265516748,
      "loss": 0.8568,
      "step": 41963
    },
    {
      "epoch": 3.6104276004473888,
      "grad_norm": 2.609375,
      "learning_rate": 0.00017891471379138018,
      "loss": 1.1859,
      "step": 41964
    },
    {
      "epoch": 3.6105136367547104,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001788939858667048,
      "loss": 1.1302,
      "step": 41965
    },
    {
      "epoch": 3.6105996730620324,
      "grad_norm": 217.0,
      "learning_rate": 0.00017887325888120198,
      "loss": 1.3074,
      "step": 41966
    },
    {
      "epoch": 3.6106857093693536,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00017885253283493269,
      "loss": 0.4806,
      "step": 41967
    },
    {
      "epoch": 3.6107717456766757,
      "grad_norm": 3.921875,
      "learning_rate": 0.00017883180772795704,
      "loss": 1.484,
      "step": 41968
    },
    {
      "epoch": 3.6108577819839973,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00017881108356033592,
      "loss": 0.8145,
      "step": 41969
    },
    {
      "epoch": 3.610943818291319,
      "grad_norm": 1.265625,
      "learning_rate": 0.00017879036033213003,
      "loss": 0.8365,
      "step": 41970
    },
    {
      "epoch": 3.6110298545986406,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00017876963804339975,
      "loss": 1.2054,
      "step": 41971
    },
    {
      "epoch": 3.611115890905962,
      "grad_norm": 12.625,
      "learning_rate": 0.00017874891669420595,
      "loss": 0.8692,
      "step": 41972
    },
    {
      "epoch": 3.6112019272132843,
      "grad_norm": 1.390625,
      "learning_rate": 0.00017872819628460897,
      "loss": 1.405,
      "step": 41973
    },
    {
      "epoch": 3.6112879635206054,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001787074768146695,
      "loss": 1.0325,
      "step": 41974
    },
    {
      "epoch": 3.6113739998279275,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0001786867582844483,
      "loss": 1.4314,
      "step": 41975
    },
    {
      "epoch": 3.611460036135249,
      "grad_norm": 0.75,
      "learning_rate": 0.00017866604069400571,
      "loss": 0.5526,
      "step": 41976
    },
    {
      "epoch": 3.6115460724425708,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017864532404340244,
      "loss": 1.1008,
      "step": 41977
    },
    {
      "epoch": 3.6116321087498924,
      "grad_norm": 1.34375,
      "learning_rate": 0.00017862460833269927,
      "loss": 0.8924,
      "step": 41978
    },
    {
      "epoch": 3.611718145057214,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00017860389356195644,
      "loss": 1.4292,
      "step": 41979
    },
    {
      "epoch": 3.611804181364536,
      "grad_norm": 1.125,
      "learning_rate": 0.00017858317973123473,
      "loss": 1.5223,
      "step": 41980
    },
    {
      "epoch": 3.6118902176718573,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017856246684059486,
      "loss": 1.162,
      "step": 41981
    },
    {
      "epoch": 3.6119762539791793,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001785417548900971,
      "loss": 1.3313,
      "step": 41982
    },
    {
      "epoch": 3.612062290286501,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001785210438798024,
      "loss": 0.6012,
      "step": 41983
    },
    {
      "epoch": 3.6121483265938226,
      "grad_norm": 1.3125,
      "learning_rate": 0.00017850033380977087,
      "loss": 1.4737,
      "step": 41984
    },
    {
      "epoch": 3.612234362901144,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001784796246800634,
      "loss": 1.3653,
      "step": 41985
    },
    {
      "epoch": 3.612320399208466,
      "grad_norm": 1.265625,
      "learning_rate": 0.00017845891649074047,
      "loss": 0.9662,
      "step": 41986
    },
    {
      "epoch": 3.612406435515788,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017843820924186266,
      "loss": 0.6958,
      "step": 41987
    },
    {
      "epoch": 3.6124924718231095,
      "grad_norm": 1.796875,
      "learning_rate": 0.00017841750293349075,
      "loss": 1.0013,
      "step": 41988
    },
    {
      "epoch": 3.612578508130431,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001783967975656849,
      "loss": 0.4712,
      "step": 41989
    },
    {
      "epoch": 3.6126645444377528,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001783760931385059,
      "loss": 1.3902,
      "step": 41990
    },
    {
      "epoch": 3.6127505807450744,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017835538965201437,
      "loss": 0.7913,
      "step": 41991
    },
    {
      "epoch": 3.612836617052396,
      "grad_norm": 1.40625,
      "learning_rate": 0.00017833468710627066,
      "loss": 1.2947,
      "step": 41992
    },
    {
      "epoch": 3.6129226533597176,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00017831398550133542,
      "loss": 1.488,
      "step": 41993
    },
    {
      "epoch": 3.6130086896670397,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017829328483726935,
      "loss": 0.7471,
      "step": 41994
    },
    {
      "epoch": 3.6130947259743613,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017827258511413275,
      "loss": 1.5461,
      "step": 41995
    },
    {
      "epoch": 3.613180762281683,
      "grad_norm": 1.4375,
      "learning_rate": 0.00017825188633198625,
      "loss": 1.5131,
      "step": 41996
    },
    {
      "epoch": 3.6132667985890046,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00017823118849089053,
      "loss": 1.1742,
      "step": 41997
    },
    {
      "epoch": 3.613352834896326,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017821049159090586,
      "loss": 1.0578,
      "step": 41998
    },
    {
      "epoch": 3.613438871203648,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001781897956320931,
      "loss": 1.5303,
      "step": 41999
    },
    {
      "epoch": 3.6135249075109694,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001781691006145124,
      "loss": 0.7485,
      "step": 42000
    },
    {
      "epoch": 3.6136109438182915,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017814840653822451,
      "loss": 1.6491,
      "step": 42001
    },
    {
      "epoch": 3.613696980125613,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001781277134032901,
      "loss": 0.9139,
      "step": 42002
    },
    {
      "epoch": 3.6137830164329348,
      "grad_norm": 1.15625,
      "learning_rate": 0.00017810702120976934,
      "loss": 1.4416,
      "step": 42003
    },
    {
      "epoch": 3.6138690527402564,
      "grad_norm": 1.671875,
      "learning_rate": 0.00017808632995772294,
      "loss": 2.0046,
      "step": 42004
    },
    {
      "epoch": 3.613955089047578,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017806563964721162,
      "loss": 0.8412,
      "step": 42005
    },
    {
      "epoch": 3.6140411253548996,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017804495027829548,
      "loss": 0.875,
      "step": 42006
    },
    {
      "epoch": 3.6141271616622213,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001780242618510353,
      "loss": 0.7095,
      "step": 42007
    },
    {
      "epoch": 3.6142131979695433,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017800357436549165,
      "loss": 0.9049,
      "step": 42008
    },
    {
      "epoch": 3.614299234276865,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017798288782172478,
      "loss": 0.4786,
      "step": 42009
    },
    {
      "epoch": 3.6143852705841866,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001779622022197953,
      "loss": 1.2862,
      "step": 42010
    },
    {
      "epoch": 3.614471306891508,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00017794151755976396,
      "loss": 1.2877,
      "step": 42011
    },
    {
      "epoch": 3.61455734319883,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017792083384169082,
      "loss": 1.4214,
      "step": 42012
    },
    {
      "epoch": 3.6146433795061514,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00017790015106563685,
      "loss": 1.7393,
      "step": 42013
    },
    {
      "epoch": 3.614729415813473,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017787946923166203,
      "loss": 0.6109,
      "step": 42014
    },
    {
      "epoch": 3.614815452120795,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017785878833982715,
      "loss": 0.6167,
      "step": 42015
    },
    {
      "epoch": 3.6149014884281168,
      "grad_norm": 16.125,
      "learning_rate": 0.00017783810839019289,
      "loss": 1.5367,
      "step": 42016
    },
    {
      "epoch": 3.6149875247354384,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00017781742938281924,
      "loss": 1.5767,
      "step": 42017
    },
    {
      "epoch": 3.61507356104276,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000177796751317767,
      "loss": 0.7732,
      "step": 42018
    },
    {
      "epoch": 3.6151595973500816,
      "grad_norm": 1.3125,
      "learning_rate": 0.00017777607419509678,
      "loss": 1.1414,
      "step": 42019
    },
    {
      "epoch": 3.6152456336574037,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017775539801486868,
      "loss": 1.7869,
      "step": 42020
    },
    {
      "epoch": 3.615331669964725,
      "grad_norm": 7.0,
      "learning_rate": 0.00017773472277714337,
      "loss": 0.9577,
      "step": 42021
    },
    {
      "epoch": 3.615417706272047,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017771404848198152,
      "loss": 1.389,
      "step": 42022
    },
    {
      "epoch": 3.6155037425793686,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000177693375129443,
      "loss": 0.6292,
      "step": 42023
    },
    {
      "epoch": 3.61558977888669,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017767270271958906,
      "loss": 0.5094,
      "step": 42024
    },
    {
      "epoch": 3.615675815194012,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001776520312524796,
      "loss": 0.9279,
      "step": 42025
    },
    {
      "epoch": 3.6157618515013334,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001776313607281752,
      "loss": 0.8128,
      "step": 42026
    },
    {
      "epoch": 3.6158478878086555,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001776106911467366,
      "loss": 0.8182,
      "step": 42027
    },
    {
      "epoch": 3.6159339241159767,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017759002250822387,
      "loss": 1.0699,
      "step": 42028
    },
    {
      "epoch": 3.6160199604232988,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017756935481269776,
      "loss": 1.4453,
      "step": 42029
    },
    {
      "epoch": 3.6161059967306204,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001775486880602184,
      "loss": 1.3442,
      "step": 42030
    },
    {
      "epoch": 3.616192033037942,
      "grad_norm": 6.625,
      "learning_rate": 0.00017752802225084641,
      "loss": 1.6463,
      "step": 42031
    },
    {
      "epoch": 3.6162780693452636,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017750735738464242,
      "loss": 0.6502,
      "step": 42032
    },
    {
      "epoch": 3.6163641056525853,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001774866934616665,
      "loss": 0.8677,
      "step": 42033
    },
    {
      "epoch": 3.6164501419599073,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00017746603048197928,
      "loss": 1.0772,
      "step": 42034
    },
    {
      "epoch": 3.6165361782672285,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0001774453684456414,
      "loss": 1.0401,
      "step": 42035
    },
    {
      "epoch": 3.6166222145745506,
      "grad_norm": 8.3125,
      "learning_rate": 0.0001774247073527129,
      "loss": 0.6319,
      "step": 42036
    },
    {
      "epoch": 3.616708250881872,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017740404720325436,
      "loss": 0.8902,
      "step": 42037
    },
    {
      "epoch": 3.616794287189194,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017738338799732644,
      "loss": 0.9637,
      "step": 42038
    },
    {
      "epoch": 3.6168803234965154,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017736272973498918,
      "loss": 1.365,
      "step": 42039
    },
    {
      "epoch": 3.616966359803837,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017734207241630324,
      "loss": 0.6143,
      "step": 42040
    },
    {
      "epoch": 3.617052396111159,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017732141604132905,
      "loss": 1.3901,
      "step": 42041
    },
    {
      "epoch": 3.6171384324184808,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001773007606101269,
      "loss": 1.581,
      "step": 42042
    },
    {
      "epoch": 3.6172244687258024,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017728010612275735,
      "loss": 0.5463,
      "step": 42043
    },
    {
      "epoch": 3.617310505033124,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017725945257928062,
      "loss": 0.901,
      "step": 42044
    },
    {
      "epoch": 3.6173965413404456,
      "grad_norm": 3.34375,
      "learning_rate": 0.00017723879997975716,
      "loss": 1.3669,
      "step": 42045
    },
    {
      "epoch": 3.6174825776477673,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00017721814832424764,
      "loss": 0.7203,
      "step": 42046
    },
    {
      "epoch": 3.617568613955089,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017719749761281213,
      "loss": 1.4217,
      "step": 42047
    },
    {
      "epoch": 3.617654650262411,
      "grad_norm": 1.21875,
      "learning_rate": 0.00017717684784551108,
      "loss": 1.6274,
      "step": 42048
    },
    {
      "epoch": 3.6177406865697326,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017715619902240515,
      "loss": 0.746,
      "step": 42049
    },
    {
      "epoch": 3.617826722877054,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001771355511435544,
      "loss": 1.2514,
      "step": 42050
    },
    {
      "epoch": 3.617912759184376,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017711490420901937,
      "loss": 0.7245,
      "step": 42051
    },
    {
      "epoch": 3.6179987954916975,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0001770942582188606,
      "loss": 1.8474,
      "step": 42052
    },
    {
      "epoch": 3.618084831799019,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001770736131731382,
      "loss": 0.887,
      "step": 42053
    },
    {
      "epoch": 3.6181708681063407,
      "grad_norm": 1.15625,
      "learning_rate": 0.00017705296907191283,
      "loss": 1.204,
      "step": 42054
    },
    {
      "epoch": 3.6182569044136628,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017703232591524448,
      "loss": 0.8972,
      "step": 42055
    },
    {
      "epoch": 3.6183429407209844,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001770116837031938,
      "loss": 0.9961,
      "step": 42056
    },
    {
      "epoch": 3.618428977028306,
      "grad_norm": 16.75,
      "learning_rate": 0.0001769910424358213,
      "loss": 1.1591,
      "step": 42057
    },
    {
      "epoch": 3.6185150133356276,
      "grad_norm": 1.46875,
      "learning_rate": 0.00017697040211318705,
      "loss": 1.0619,
      "step": 42058
    },
    {
      "epoch": 3.6186010496429493,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017694976273535152,
      "loss": 1.22,
      "step": 42059
    },
    {
      "epoch": 3.618687085950271,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017692912430237507,
      "loss": 0.5794,
      "step": 42060
    },
    {
      "epoch": 3.6187731222575925,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017690848681431814,
      "loss": 0.7585,
      "step": 42061
    },
    {
      "epoch": 3.6188591585649146,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017688785027124122,
      "loss": 1.1351,
      "step": 42062
    },
    {
      "epoch": 3.618945194872236,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017686721467320427,
      "loss": 1.1225,
      "step": 42063
    },
    {
      "epoch": 3.619031231179558,
      "grad_norm": 3.140625,
      "learning_rate": 0.00017684658002026793,
      "loss": 1.336,
      "step": 42064
    },
    {
      "epoch": 3.6191172674868795,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017682594631249267,
      "loss": 0.747,
      "step": 42065
    },
    {
      "epoch": 3.619203303794201,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017680531354993844,
      "loss": 0.5733,
      "step": 42066
    },
    {
      "epoch": 3.6192893401015227,
      "grad_norm": 232.0,
      "learning_rate": 0.00017678468173266582,
      "loss": 1.1231,
      "step": 42067
    },
    {
      "epoch": 3.6193753764088443,
      "grad_norm": 1.46875,
      "learning_rate": 0.00017676405086073532,
      "loss": 1.6045,
      "step": 42068
    },
    {
      "epoch": 3.6194614127161664,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000176743420934207,
      "loss": 0.3986,
      "step": 42069
    },
    {
      "epoch": 3.619547449023488,
      "grad_norm": 1.53125,
      "learning_rate": 0.0001767227919531414,
      "loss": 1.3992,
      "step": 42070
    },
    {
      "epoch": 3.6196334853308096,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017670216391759864,
      "loss": 1.025,
      "step": 42071
    },
    {
      "epoch": 3.6197195216381313,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001766815368276391,
      "loss": 1.0808,
      "step": 42072
    },
    {
      "epoch": 3.619805557945453,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001766609106833234,
      "loss": 1.0349,
      "step": 42073
    },
    {
      "epoch": 3.6198915942527745,
      "grad_norm": 1.25,
      "learning_rate": 0.00017664028548471152,
      "loss": 1.5307,
      "step": 42074
    },
    {
      "epoch": 3.619977630560096,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017661966123186384,
      "loss": 1.2957,
      "step": 42075
    },
    {
      "epoch": 3.620063666867418,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017659903792484094,
      "loss": 1.1685,
      "step": 42076
    },
    {
      "epoch": 3.62014970317474,
      "grad_norm": 10.125,
      "learning_rate": 0.0001765784155637028,
      "loss": 1.0123,
      "step": 42077
    },
    {
      "epoch": 3.6202357394820615,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00017655779414850993,
      "loss": 1.0261,
      "step": 42078
    },
    {
      "epoch": 3.620321775789383,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001765371736793227,
      "loss": 1.0594,
      "step": 42079
    },
    {
      "epoch": 3.6204078120967047,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017651655415620122,
      "loss": 0.7323,
      "step": 42080
    },
    {
      "epoch": 3.6204938484040268,
      "grad_norm": 3.9375,
      "learning_rate": 0.00017649593557920591,
      "loss": 1.2855,
      "step": 42081
    },
    {
      "epoch": 3.620579884711348,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00017647531794839723,
      "loss": 1.4437,
      "step": 42082
    },
    {
      "epoch": 3.62066592101867,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001764547012638351,
      "loss": 0.8376,
      "step": 42083
    },
    {
      "epoch": 3.6207519573259916,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00017643408552558027,
      "loss": 0.9388,
      "step": 42084
    },
    {
      "epoch": 3.6208379936333133,
      "grad_norm": 16.25,
      "learning_rate": 0.00017641347073369263,
      "loss": 1.2337,
      "step": 42085
    },
    {
      "epoch": 3.620924029940635,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017639285688823264,
      "loss": 1.3529,
      "step": 42086
    },
    {
      "epoch": 3.6210100662479565,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017637224398926076,
      "loss": 1.0018,
      "step": 42087
    },
    {
      "epoch": 3.6210961025552786,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017635163203683695,
      "loss": 1.3648,
      "step": 42088
    },
    {
      "epoch": 3.6211821388625998,
      "grad_norm": 1.0,
      "learning_rate": 0.0001763310210310217,
      "loss": 1.0757,
      "step": 42089
    },
    {
      "epoch": 3.621268175169922,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017631041097187545,
      "loss": 1.5966,
      "step": 42090
    },
    {
      "epoch": 3.6213542114772435,
      "grad_norm": 3.703125,
      "learning_rate": 0.00017628980185945808,
      "loss": 1.8287,
      "step": 42091
    },
    {
      "epoch": 3.621440247784565,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001762691936938301,
      "loss": 1.1036,
      "step": 42092
    },
    {
      "epoch": 3.6215262840918867,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017624858647505198,
      "loss": 0.7063,
      "step": 42093
    },
    {
      "epoch": 3.6216123203992083,
      "grad_norm": 2.21875,
      "learning_rate": 0.00017622798020318353,
      "loss": 1.2318,
      "step": 42094
    },
    {
      "epoch": 3.6216983567065304,
      "grad_norm": 1.859375,
      "learning_rate": 0.00017620737487828548,
      "loss": 1.3367,
      "step": 42095
    },
    {
      "epoch": 3.6217843930138516,
      "grad_norm": 1.125,
      "learning_rate": 0.00017618677050041753,
      "loss": 1.0873,
      "step": 42096
    },
    {
      "epoch": 3.6218704293211736,
      "grad_norm": 4.71875,
      "learning_rate": 0.0001761661670696405,
      "loss": 0.9639,
      "step": 42097
    },
    {
      "epoch": 3.6219564656284953,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001761455645860146,
      "loss": 1.64,
      "step": 42098
    },
    {
      "epoch": 3.622042501935817,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017612496304959975,
      "loss": 1.8243,
      "step": 42099
    },
    {
      "epoch": 3.6221285382431385,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017610436246045657,
      "loss": 0.8154,
      "step": 42100
    },
    {
      "epoch": 3.62221457455046,
      "grad_norm": 1.234375,
      "learning_rate": 0.00017608376281864503,
      "loss": 1.7173,
      "step": 42101
    },
    {
      "epoch": 3.622300610857782,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001760631641242254,
      "loss": 0.9143,
      "step": 42102
    },
    {
      "epoch": 3.622386647165104,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001760425663772582,
      "loss": 1.6792,
      "step": 42103
    },
    {
      "epoch": 3.6224726834724255,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001760219695778033,
      "loss": 0.8054,
      "step": 42104
    },
    {
      "epoch": 3.622558719779747,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00017600137372592118,
      "loss": 1.2914,
      "step": 42105
    },
    {
      "epoch": 3.6226447560870687,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017598077882167217,
      "loss": 1.0156,
      "step": 42106
    },
    {
      "epoch": 3.6227307923943903,
      "grad_norm": 1.265625,
      "learning_rate": 0.00017596018486511623,
      "loss": 1.0192,
      "step": 42107
    },
    {
      "epoch": 3.622816828701712,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001759395918563137,
      "loss": 1.0928,
      "step": 42108
    },
    {
      "epoch": 3.622902865009034,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017591899979532499,
      "loss": 1.0683,
      "step": 42109
    },
    {
      "epoch": 3.6229889013163556,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017589840868221002,
      "loss": 0.6955,
      "step": 42110
    },
    {
      "epoch": 3.6230749376236773,
      "grad_norm": 1.59375,
      "learning_rate": 0.00017587781851702934,
      "loss": 0.9609,
      "step": 42111
    },
    {
      "epoch": 3.623160973930999,
      "grad_norm": 71.0,
      "learning_rate": 0.00017585722929984283,
      "loss": 1.1476,
      "step": 42112
    },
    {
      "epoch": 3.6232470102383205,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001758366410307109,
      "loss": 0.7886,
      "step": 42113
    },
    {
      "epoch": 3.623333046545642,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001758160537096939,
      "loss": 1.0565,
      "step": 42114
    },
    {
      "epoch": 3.6234190828529638,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001757954673368517,
      "loss": 0.8969,
      "step": 42115
    },
    {
      "epoch": 3.623505119160286,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017577488191224472,
      "loss": 1.0319,
      "step": 42116
    },
    {
      "epoch": 3.6235911554676075,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00017575429743593336,
      "loss": 1.3821,
      "step": 42117
    },
    {
      "epoch": 3.623677191774929,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017573371390797738,
      "loss": 0.7956,
      "step": 42118
    },
    {
      "epoch": 3.6237632280822507,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017571313132843725,
      "loss": 0.4695,
      "step": 42119
    },
    {
      "epoch": 3.6238492643895723,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017569254969737337,
      "loss": 1.5411,
      "step": 42120
    },
    {
      "epoch": 3.623935300696894,
      "grad_norm": 0.703125,
      "learning_rate": 0.00017567196901484543,
      "loss": 0.6659,
      "step": 42121
    },
    {
      "epoch": 3.6240213370042156,
      "grad_norm": 4.84375,
      "learning_rate": 0.000175651389280914,
      "loss": 0.9959,
      "step": 42122
    },
    {
      "epoch": 3.6241073733115376,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017563081049563927,
      "loss": 1.2682,
      "step": 42123
    },
    {
      "epoch": 3.6241934096188593,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017561023265908122,
      "loss": 1.5893,
      "step": 42124
    },
    {
      "epoch": 3.624279445926181,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017558965577130033,
      "loss": 1.396,
      "step": 42125
    },
    {
      "epoch": 3.6243654822335025,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017556907983235637,
      "loss": 1.5917,
      "step": 42126
    },
    {
      "epoch": 3.624451518540824,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017554850484230978,
      "loss": 0.9291,
      "step": 42127
    },
    {
      "epoch": 3.6245375548481458,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00017552793080122088,
      "loss": 1.0644,
      "step": 42128
    },
    {
      "epoch": 3.6246235911554674,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001755073577091495,
      "loss": 0.9989,
      "step": 42129
    },
    {
      "epoch": 3.6247096274627895,
      "grad_norm": 0.90625,
      "learning_rate": 0.000175486785566156,
      "loss": 1.407,
      "step": 42130
    },
    {
      "epoch": 3.624795663770111,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001754662143723007,
      "loss": 1.2759,
      "step": 42131
    },
    {
      "epoch": 3.6248817000774327,
      "grad_norm": 1.234375,
      "learning_rate": 0.00017544564412764346,
      "loss": 1.0825,
      "step": 42132
    },
    {
      "epoch": 3.6249677363847543,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017542507483224456,
      "loss": 0.5534,
      "step": 42133
    },
    {
      "epoch": 3.625053772692076,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001754045064861642,
      "loss": 0.9639,
      "step": 42134
    },
    {
      "epoch": 3.6251398089993976,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017538393908946248,
      "loss": 0.5764,
      "step": 42135
    },
    {
      "epoch": 3.625225845306719,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017536337264219986,
      "loss": 1.5074,
      "step": 42136
    },
    {
      "epoch": 3.6253118816140413,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017534280714443602,
      "loss": 1.3198,
      "step": 42137
    },
    {
      "epoch": 3.625397917921363,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001753222425962313,
      "loss": 0.8153,
      "step": 42138
    },
    {
      "epoch": 3.6254839542286845,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001753016789976461,
      "loss": 1.6041,
      "step": 42139
    },
    {
      "epoch": 3.625569990536006,
      "grad_norm": 29.625,
      "learning_rate": 0.00017528111634874016,
      "loss": 1.8351,
      "step": 42140
    },
    {
      "epoch": 3.6256560268433278,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017526055464957396,
      "loss": 1.1785,
      "step": 42141
    },
    {
      "epoch": 3.62574206315065,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00017523999390020727,
      "loss": 0.7183,
      "step": 42142
    },
    {
      "epoch": 3.625828099457971,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001752194341007005,
      "loss": 1.3841,
      "step": 42143
    },
    {
      "epoch": 3.625914135765293,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017519887525111383,
      "loss": 1.4849,
      "step": 42144
    },
    {
      "epoch": 3.6260001720726147,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001751783173515072,
      "loss": 1.535,
      "step": 42145
    },
    {
      "epoch": 3.6260862083799363,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001751577604019407,
      "loss": 0.9309,
      "step": 42146
    },
    {
      "epoch": 3.626172244687258,
      "grad_norm": 1.0,
      "learning_rate": 0.00017513720440247483,
      "loss": 0.6447,
      "step": 42147
    },
    {
      "epoch": 3.6262582809945796,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017511664935316923,
      "loss": 1.6327,
      "step": 42148
    },
    {
      "epoch": 3.6263443173019017,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017509609525408426,
      "loss": 0.9983,
      "step": 42149
    },
    {
      "epoch": 3.626430353609223,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001750755421052802,
      "loss": 1.7174,
      "step": 42150
    },
    {
      "epoch": 3.626516389916545,
      "grad_norm": 1.8125,
      "learning_rate": 0.00017505498990681684,
      "loss": 1.1954,
      "step": 42151
    },
    {
      "epoch": 3.6266024262238665,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001750344386587544,
      "loss": 1.0055,
      "step": 42152
    },
    {
      "epoch": 3.626688462531188,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00017501388836115323,
      "loss": 0.9226,
      "step": 42153
    },
    {
      "epoch": 3.6267744988385098,
      "grad_norm": 1.34375,
      "learning_rate": 0.000174993339014073,
      "loss": 1.4929,
      "step": 42154
    },
    {
      "epoch": 3.6268605351458314,
      "grad_norm": 7.875,
      "learning_rate": 0.00017497279061757428,
      "loss": 1.1853,
      "step": 42155
    },
    {
      "epoch": 3.6269465714531535,
      "grad_norm": 1.375,
      "learning_rate": 0.0001749522431717167,
      "loss": 1.3856,
      "step": 42156
    },
    {
      "epoch": 3.6270326077604746,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017493169667656062,
      "loss": 1.7797,
      "step": 42157
    },
    {
      "epoch": 3.6271186440677967,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017491115113216633,
      "loss": 1.7413,
      "step": 42158
    },
    {
      "epoch": 3.6272046803751183,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017489060653859347,
      "loss": 0.8344,
      "step": 42159
    },
    {
      "epoch": 3.62729071668244,
      "grad_norm": 1.78125,
      "learning_rate": 0.00017487006289590234,
      "loss": 1.5044,
      "step": 42160
    },
    {
      "epoch": 3.6273767529897616,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017484952020415322,
      "loss": 1.2445,
      "step": 42161
    },
    {
      "epoch": 3.627462789297083,
      "grad_norm": 1.40625,
      "learning_rate": 0.00017482897846340584,
      "loss": 2.1018,
      "step": 42162
    },
    {
      "epoch": 3.6275488256044053,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017480843767372052,
      "loss": 0.9941,
      "step": 42163
    },
    {
      "epoch": 3.627634861911727,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017478789783515735,
      "loss": 1.6416,
      "step": 42164
    },
    {
      "epoch": 3.6277208982190485,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017476735894777622,
      "loss": 0.9816,
      "step": 42165
    },
    {
      "epoch": 3.62780693452637,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017474682101163746,
      "loss": 0.9376,
      "step": 42166
    },
    {
      "epoch": 3.6278929708336918,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017472628402680085,
      "loss": 1.0552,
      "step": 42167
    },
    {
      "epoch": 3.6279790071410134,
      "grad_norm": 1.265625,
      "learning_rate": 0.00017470574799332655,
      "loss": 1.5869,
      "step": 42168
    },
    {
      "epoch": 3.628065043448335,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017468521291127487,
      "loss": 1.0252,
      "step": 42169
    },
    {
      "epoch": 3.628151079755657,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001746646787807053,
      "loss": 1.019,
      "step": 42170
    },
    {
      "epoch": 3.6282371160629787,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001746441456016787,
      "loss": 1.0053,
      "step": 42171
    },
    {
      "epoch": 3.6283231523703003,
      "grad_norm": 2.25,
      "learning_rate": 0.0001746236133742544,
      "loss": 1.0909,
      "step": 42172
    },
    {
      "epoch": 3.628409188677622,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017460308209849286,
      "loss": 0.775,
      "step": 42173
    },
    {
      "epoch": 3.6284952249849436,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001745825517744541,
      "loss": 0.4535,
      "step": 42174
    },
    {
      "epoch": 3.628581261292265,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017456202240219798,
      "loss": 1.2151,
      "step": 42175
    },
    {
      "epoch": 3.628667297599587,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001745414939817846,
      "loss": 0.9297,
      "step": 42176
    },
    {
      "epoch": 3.628753333906909,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00017452096651327426,
      "loss": 1.5212,
      "step": 42177
    },
    {
      "epoch": 3.6288393702142305,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001745004399967266,
      "loss": 1.4211,
      "step": 42178
    },
    {
      "epoch": 3.628925406521552,
      "grad_norm": 1.359375,
      "learning_rate": 0.00017447991443220184,
      "loss": 0.7683,
      "step": 42179
    },
    {
      "epoch": 3.629011442828874,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017445938981976023,
      "loss": 1.5329,
      "step": 42180
    },
    {
      "epoch": 3.6290974791361954,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00017443886615946136,
      "loss": 0.8668,
      "step": 42181
    },
    {
      "epoch": 3.629183515443517,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0001744183434513657,
      "loss": 1.0755,
      "step": 42182
    },
    {
      "epoch": 3.6292695517508387,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017439782169553292,
      "loss": 0.8333,
      "step": 42183
    },
    {
      "epoch": 3.6293555880581607,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017437730089202313,
      "loss": 1.7045,
      "step": 42184
    },
    {
      "epoch": 3.6294416243654823,
      "grad_norm": 1.0,
      "learning_rate": 0.00017435678104089659,
      "loss": 1.1222,
      "step": 42185
    },
    {
      "epoch": 3.629527660672804,
      "grad_norm": 1.3125,
      "learning_rate": 0.000174336262142213,
      "loss": 0.6883,
      "step": 42186
    },
    {
      "epoch": 3.6296136969801256,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017431574419603246,
      "loss": 0.81,
      "step": 42187
    },
    {
      "epoch": 3.629699733287447,
      "grad_norm": 5.5,
      "learning_rate": 0.0001742952272024152,
      "loss": 1.2882,
      "step": 42188
    },
    {
      "epoch": 3.629785769594769,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0001742747111614209,
      "loss": 1.4741,
      "step": 42189
    },
    {
      "epoch": 3.6298718059020905,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017425419607310972,
      "loss": 0.886,
      "step": 42190
    },
    {
      "epoch": 3.6299578422094125,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001742336819375418,
      "loss": 0.971,
      "step": 42191
    },
    {
      "epoch": 3.630043878516734,
      "grad_norm": 1.109375,
      "learning_rate": 0.00017421316875477688,
      "loss": 1.4444,
      "step": 42192
    },
    {
      "epoch": 3.630129914824056,
      "grad_norm": 3.78125,
      "learning_rate": 0.00017419265652487503,
      "loss": 1.5779,
      "step": 42193
    },
    {
      "epoch": 3.6302159511313774,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00017417214524789655,
      "loss": 1.3963,
      "step": 42194
    },
    {
      "epoch": 3.630301987438699,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001741516349239009,
      "loss": 1.0258,
      "step": 42195
    },
    {
      "epoch": 3.630388023746021,
      "grad_norm": 1.3125,
      "learning_rate": 0.00017413112555294858,
      "loss": 1.1873,
      "step": 42196
    },
    {
      "epoch": 3.6304740600533423,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017411061713509913,
      "loss": 1.0487,
      "step": 42197
    },
    {
      "epoch": 3.6305600963606643,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001740901096704128,
      "loss": 1.1899,
      "step": 42198
    },
    {
      "epoch": 3.630646132667986,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017406960315894966,
      "loss": 0.7323,
      "step": 42199
    },
    {
      "epoch": 3.6307321689753076,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001740490976007693,
      "loss": 1.4858,
      "step": 42200
    },
    {
      "epoch": 3.630818205282629,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00017402859299593204,
      "loss": 1.9765,
      "step": 42201
    },
    {
      "epoch": 3.630904241589951,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017400808934449786,
      "loss": 0.9113,
      "step": 42202
    },
    {
      "epoch": 3.630990277897273,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017398758664652648,
      "loss": 1.1817,
      "step": 42203
    },
    {
      "epoch": 3.631076314204594,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017396708490207796,
      "loss": 1.7038,
      "step": 42204
    },
    {
      "epoch": 3.631162350511916,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017394658411121251,
      "loss": 1.14,
      "step": 42205
    },
    {
      "epoch": 3.631248386819238,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001739260842739897,
      "loss": 0.9329,
      "step": 42206
    },
    {
      "epoch": 3.6313344231265594,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00017390558539046968,
      "loss": 1.0622,
      "step": 42207
    },
    {
      "epoch": 3.631420459433881,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001738850874607124,
      "loss": 1.0409,
      "step": 42208
    },
    {
      "epoch": 3.6315064957412027,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017386459048477783,
      "loss": 1.7832,
      "step": 42209
    },
    {
      "epoch": 3.6315925320485247,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017384409446272604,
      "loss": 0.9637,
      "step": 42210
    },
    {
      "epoch": 3.631678568355846,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017382359939461668,
      "loss": 1.4047,
      "step": 42211
    },
    {
      "epoch": 3.631764604663168,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00017380310528051002,
      "loss": 1.7908,
      "step": 42212
    },
    {
      "epoch": 3.6318506409704896,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001737826121204657,
      "loss": 0.6344,
      "step": 42213
    },
    {
      "epoch": 3.631936677277811,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017376211991454378,
      "loss": 1.2548,
      "step": 42214
    },
    {
      "epoch": 3.632022713585133,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017374162866280434,
      "loss": 1.0272,
      "step": 42215
    },
    {
      "epoch": 3.6321087498924545,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017372113836530706,
      "loss": 1.2765,
      "step": 42216
    },
    {
      "epoch": 3.6321947861997765,
      "grad_norm": 1.234375,
      "learning_rate": 0.00017370064902211198,
      "loss": 0.7984,
      "step": 42217
    },
    {
      "epoch": 3.632280822507098,
      "grad_norm": 0.875,
      "learning_rate": 0.00017368016063327917,
      "loss": 1.323,
      "step": 42218
    },
    {
      "epoch": 3.63236685881442,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017365967319886834,
      "loss": 1.549,
      "step": 42219
    },
    {
      "epoch": 3.6324528951217414,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00017363918671893942,
      "loss": 1.1535,
      "step": 42220
    },
    {
      "epoch": 3.632538931429063,
      "grad_norm": 7.625,
      "learning_rate": 0.0001736187011935526,
      "loss": 0.9112,
      "step": 42221
    },
    {
      "epoch": 3.6326249677363847,
      "grad_norm": 1.453125,
      "learning_rate": 0.00017359821662276743,
      "loss": 1.532,
      "step": 42222
    },
    {
      "epoch": 3.6327110040437063,
      "grad_norm": 1.9921875,
      "learning_rate": 0.00017357773300664413,
      "loss": 1.6334,
      "step": 42223
    },
    {
      "epoch": 3.6327970403510284,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017355725034524239,
      "loss": 0.9259,
      "step": 42224
    },
    {
      "epoch": 3.63288307665835,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001735367686386221,
      "loss": 1.8307,
      "step": 42225
    },
    {
      "epoch": 3.6329691129656716,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017351628788684354,
      "loss": 0.8301,
      "step": 42226
    },
    {
      "epoch": 3.6330551492729932,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017349580808996614,
      "loss": 0.5815,
      "step": 42227
    },
    {
      "epoch": 3.633141185580315,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017347532924805,
      "loss": 0.5846,
      "step": 42228
    },
    {
      "epoch": 3.6332272218876365,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017345485136115513,
      "loss": 1.1384,
      "step": 42229
    },
    {
      "epoch": 3.633313258194958,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001734343744293412,
      "loss": 0.8884,
      "step": 42230
    },
    {
      "epoch": 3.63339929450228,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001734138984526682,
      "loss": 1.4198,
      "step": 42231
    },
    {
      "epoch": 3.633485330809602,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017339342343119618,
      "loss": 0.5708,
      "step": 42232
    },
    {
      "epoch": 3.6335713671169234,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017337294936498472,
      "loss": 1.0631,
      "step": 42233
    },
    {
      "epoch": 3.633657403424245,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017335247625409385,
      "loss": 0.6746,
      "step": 42234
    },
    {
      "epoch": 3.6337434397315667,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017333200409858364,
      "loss": 1.6652,
      "step": 42235
    },
    {
      "epoch": 3.6338294760388883,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001733115328985136,
      "loss": 1.4976,
      "step": 42236
    },
    {
      "epoch": 3.63391551234621,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017329106265394394,
      "loss": 0.7039,
      "step": 42237
    },
    {
      "epoch": 3.634001548653532,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017327059336493423,
      "loss": 1.2152,
      "step": 42238
    },
    {
      "epoch": 3.6340875849608536,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017325012503154454,
      "loss": 0.4782,
      "step": 42239
    },
    {
      "epoch": 3.6341736212681752,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00017322965765383474,
      "loss": 1.0857,
      "step": 42240
    },
    {
      "epoch": 3.634259657575497,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017320919123186456,
      "loss": 0.7786,
      "step": 42241
    },
    {
      "epoch": 3.6343456938828185,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017318872576569394,
      "loss": 1.2321,
      "step": 42242
    },
    {
      "epoch": 3.63443173019014,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017316826125538288,
      "loss": 0.9855,
      "step": 42243
    },
    {
      "epoch": 3.6345177664974617,
      "grad_norm": 300.0,
      "learning_rate": 0.00017314779770099075,
      "loss": 1.2638,
      "step": 42244
    },
    {
      "epoch": 3.634603802804784,
      "grad_norm": 1.0,
      "learning_rate": 0.00017312733510257806,
      "loss": 0.6589,
      "step": 42245
    },
    {
      "epoch": 3.6346898391121054,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001731068734602042,
      "loss": 0.6722,
      "step": 42246
    },
    {
      "epoch": 3.634775875419427,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017308641277392916,
      "loss": 1.2082,
      "step": 42247
    },
    {
      "epoch": 3.6348619117267487,
      "grad_norm": 132.0,
      "learning_rate": 0.00017306595304381295,
      "loss": 0.9398,
      "step": 42248
    },
    {
      "epoch": 3.6349479480340703,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017304549426991507,
      "loss": 1.2871,
      "step": 42249
    },
    {
      "epoch": 3.635033984341392,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001730250364522955,
      "loss": 1.2782,
      "step": 42250
    },
    {
      "epoch": 3.6351200206487135,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00017300457959101436,
      "loss": 1.6181,
      "step": 42251
    },
    {
      "epoch": 3.6352060569560356,
      "grad_norm": 1.84375,
      "learning_rate": 0.00017298412368613097,
      "loss": 0.6872,
      "step": 42252
    },
    {
      "epoch": 3.6352920932633572,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017296366873770564,
      "loss": 0.7859,
      "step": 42253
    },
    {
      "epoch": 3.635378129570679,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017294321474579778,
      "loss": 0.6848,
      "step": 42254
    },
    {
      "epoch": 3.6354641658780005,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001729227617104674,
      "loss": 0.8773,
      "step": 42255
    },
    {
      "epoch": 3.635550202185322,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001729023096317745,
      "loss": 1.2869,
      "step": 42256
    },
    {
      "epoch": 3.635636238492644,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00017288185850977857,
      "loss": 1.0488,
      "step": 42257
    },
    {
      "epoch": 3.6357222747999653,
      "grad_norm": 10.125,
      "learning_rate": 0.00017286140834453956,
      "loss": 0.9116,
      "step": 42258
    },
    {
      "epoch": 3.6358083111072874,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00017284095913611754,
      "loss": 1.2722,
      "step": 42259
    },
    {
      "epoch": 3.635894347414609,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017282051088457186,
      "loss": 0.7631,
      "step": 42260
    },
    {
      "epoch": 3.6359803837219307,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017280006358996253,
      "loss": 0.8158,
      "step": 42261
    },
    {
      "epoch": 3.6360664200292523,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0001727796172523496,
      "loss": 1.7007,
      "step": 42262
    },
    {
      "epoch": 3.636152456336574,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00017275917187179246,
      "loss": 1.31,
      "step": 42263
    },
    {
      "epoch": 3.636238492643896,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017273872744835123,
      "loss": 1.2074,
      "step": 42264
    },
    {
      "epoch": 3.636324528951217,
      "grad_norm": 8.5625,
      "learning_rate": 0.00017271828398208544,
      "loss": 0.9811,
      "step": 42265
    },
    {
      "epoch": 3.6364105652585392,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00017269784147305506,
      "loss": 1.0108,
      "step": 42266
    },
    {
      "epoch": 3.636496601565861,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017267739992131992,
      "loss": 0.8966,
      "step": 42267
    },
    {
      "epoch": 3.6365826378731825,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017265695932693965,
      "loss": 1.0742,
      "step": 42268
    },
    {
      "epoch": 3.636668674180504,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00017263651968997402,
      "loss": 1.2475,
      "step": 42269
    },
    {
      "epoch": 3.6367547104878257,
      "grad_norm": 53.25,
      "learning_rate": 0.0001726160810104831,
      "loss": 1.0312,
      "step": 42270
    },
    {
      "epoch": 3.636840746795148,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017259564328852634,
      "loss": 0.9445,
      "step": 42271
    },
    {
      "epoch": 3.636926783102469,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017257520652416358,
      "loss": 0.8581,
      "step": 42272
    },
    {
      "epoch": 3.637012819409791,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001725547707174549,
      "loss": 1.2763,
      "step": 42273
    },
    {
      "epoch": 3.6370988557171127,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017253433586845963,
      "loss": 0.7486,
      "step": 42274
    },
    {
      "epoch": 3.6371848920244343,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001725139019772377,
      "loss": 1.0061,
      "step": 42275
    },
    {
      "epoch": 3.637270928331756,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017249346904384916,
      "loss": 1.583,
      "step": 42276
    },
    {
      "epoch": 3.6373569646390775,
      "grad_norm": 7.46875,
      "learning_rate": 0.00017247303706835332,
      "loss": 0.6661,
      "step": 42277
    },
    {
      "epoch": 3.6374430009463996,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001724526060508103,
      "loss": 1.0328,
      "step": 42278
    },
    {
      "epoch": 3.6375290372537212,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00017243217599127952,
      "loss": 1.7355,
      "step": 42279
    },
    {
      "epoch": 3.637615073561043,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000172411746889821,
      "loss": 0.92,
      "step": 42280
    },
    {
      "epoch": 3.6377011098683645,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017239131874649432,
      "loss": 0.7753,
      "step": 42281
    },
    {
      "epoch": 3.637787146175686,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00017237089156135933,
      "loss": 0.643,
      "step": 42282
    },
    {
      "epoch": 3.6378731824830077,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017235046533447596,
      "loss": 1.1224,
      "step": 42283
    },
    {
      "epoch": 3.6379592187903294,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017233004006590358,
      "loss": 0.5813,
      "step": 42284
    },
    {
      "epoch": 3.6380452550976514,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001723096157557021,
      "loss": 1.0034,
      "step": 42285
    },
    {
      "epoch": 3.638131291404973,
      "grad_norm": 85.5,
      "learning_rate": 0.00017228919240393144,
      "loss": 0.921,
      "step": 42286
    },
    {
      "epoch": 3.6382173277122947,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017226877001065094,
      "loss": 0.9573,
      "step": 42287
    },
    {
      "epoch": 3.6383033640196163,
      "grad_norm": 1.5,
      "learning_rate": 0.00017224834857592058,
      "loss": 1.3705,
      "step": 42288
    },
    {
      "epoch": 3.638389400326938,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017222792809980025,
      "loss": 1.5323,
      "step": 42289
    },
    {
      "epoch": 3.6384754366342595,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001722075085823493,
      "loss": 0.6265,
      "step": 42290
    },
    {
      "epoch": 3.638561472941581,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001721870900236276,
      "loss": 1.0132,
      "step": 42291
    },
    {
      "epoch": 3.6386475092489032,
      "grad_norm": 4.96875,
      "learning_rate": 0.0001721666724236951,
      "loss": 1.1495,
      "step": 42292
    },
    {
      "epoch": 3.638733545556225,
      "grad_norm": 5.65625,
      "learning_rate": 0.00017214625578261118,
      "loss": 1.7426,
      "step": 42293
    },
    {
      "epoch": 3.6388195818635465,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001721258401004358,
      "loss": 0.6985,
      "step": 42294
    },
    {
      "epoch": 3.638905618170868,
      "grad_norm": 1.109375,
      "learning_rate": 0.00017210542537722846,
      "loss": 0.9023,
      "step": 42295
    },
    {
      "epoch": 3.6389916544781897,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017208501161304895,
      "loss": 1.5861,
      "step": 42296
    },
    {
      "epoch": 3.6390776907855114,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017206459880795722,
      "loss": 0.5572,
      "step": 42297
    },
    {
      "epoch": 3.639163727092833,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017204418696201257,
      "loss": 1.0116,
      "step": 42298
    },
    {
      "epoch": 3.639249763400155,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017202377607527486,
      "loss": 0.5554,
      "step": 42299
    },
    {
      "epoch": 3.6393357997074767,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000172003366147804,
      "loss": 1.3976,
      "step": 42300
    },
    {
      "epoch": 3.6394218360147983,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017198295717965934,
      "loss": 0.7868,
      "step": 42301
    },
    {
      "epoch": 3.63950787232212,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001719625491709007,
      "loss": 1.6669,
      "step": 42302
    },
    {
      "epoch": 3.6395939086294415,
      "grad_norm": 1.421875,
      "learning_rate": 0.00017194214212158803,
      "loss": 1.1713,
      "step": 42303
    },
    {
      "epoch": 3.639679944936763,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017192173603178056,
      "loss": 0.8627,
      "step": 42304
    },
    {
      "epoch": 3.639765981244085,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017190133090153826,
      "loss": 1.0494,
      "step": 42305
    },
    {
      "epoch": 3.639852017551407,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00017188092673092089,
      "loss": 0.7102,
      "step": 42306
    },
    {
      "epoch": 3.6399380538587285,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017186052351998783,
      "loss": 0.612,
      "step": 42307
    },
    {
      "epoch": 3.64002409016605,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001718401212687991,
      "loss": 0.9994,
      "step": 42308
    },
    {
      "epoch": 3.6401101264733717,
      "grad_norm": 1.3125,
      "learning_rate": 0.00017181971997741403,
      "loss": 1.6131,
      "step": 42309
    },
    {
      "epoch": 3.6401961627806934,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00017179931964589245,
      "loss": 1.6114,
      "step": 42310
    },
    {
      "epoch": 3.640282199088015,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017177892027429414,
      "loss": 1.7033,
      "step": 42311
    },
    {
      "epoch": 3.6403682353953366,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017175852186267855,
      "loss": 0.698,
      "step": 42312
    },
    {
      "epoch": 3.6404542717026587,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00017173812441110542,
      "loss": 1.0921,
      "step": 42313
    },
    {
      "epoch": 3.6405403080099803,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00017171772791963464,
      "loss": 0.759,
      "step": 42314
    },
    {
      "epoch": 3.640626344317302,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017169733238832547,
      "loss": 0.8617,
      "step": 42315
    },
    {
      "epoch": 3.6407123806246235,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001716769378172377,
      "loss": 0.9365,
      "step": 42316
    },
    {
      "epoch": 3.640798416931945,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00017165654420643123,
      "loss": 1.3451,
      "step": 42317
    },
    {
      "epoch": 3.6408844532392672,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00017163615155596517,
      "loss": 1.4782,
      "step": 42318
    },
    {
      "epoch": 3.6409704895465884,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001716157598658999,
      "loss": 1.7852,
      "step": 42319
    },
    {
      "epoch": 3.6410565258539105,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017159536913629442,
      "loss": 0.6613,
      "step": 42320
    },
    {
      "epoch": 3.641142562161232,
      "grad_norm": 1.875,
      "learning_rate": 0.0001715749793672086,
      "loss": 0.9938,
      "step": 42321
    },
    {
      "epoch": 3.6412285984685537,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017155459055870231,
      "loss": 1.5196,
      "step": 42322
    },
    {
      "epoch": 3.6413146347758754,
      "grad_norm": 4352.0,
      "learning_rate": 0.0001715342027108348,
      "loss": 0.6461,
      "step": 42323
    },
    {
      "epoch": 3.641400671083197,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00017151381582366604,
      "loss": 1.5333,
      "step": 42324
    },
    {
      "epoch": 3.641486707390519,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001714934298972553,
      "loss": 0.6344,
      "step": 42325
    },
    {
      "epoch": 3.6415727436978402,
      "grad_norm": 1.34375,
      "learning_rate": 0.00017147304493166237,
      "loss": 1.0852,
      "step": 42326
    },
    {
      "epoch": 3.6416587800051623,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017145266092694707,
      "loss": 1.7817,
      "step": 42327
    },
    {
      "epoch": 3.641744816312484,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017143227788316863,
      "loss": 1.0704,
      "step": 42328
    },
    {
      "epoch": 3.6418308526198055,
      "grad_norm": 2.75,
      "learning_rate": 0.00017141189580038691,
      "loss": 1.1843,
      "step": 42329
    },
    {
      "epoch": 3.641916888927127,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017139151467866165,
      "loss": 1.0241,
      "step": 42330
    },
    {
      "epoch": 3.642002925234449,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00017137113451805213,
      "loss": 0.6316,
      "step": 42331
    },
    {
      "epoch": 3.642088961541771,
      "grad_norm": 1.125,
      "learning_rate": 0.00017135075531861816,
      "loss": 0.7824,
      "step": 42332
    },
    {
      "epoch": 3.642174997849092,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017133037708041943,
      "loss": 1.2199,
      "step": 42333
    },
    {
      "epoch": 3.642261034156414,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001713099998035153,
      "loss": 0.9554,
      "step": 42334
    },
    {
      "epoch": 3.6423470704637357,
      "grad_norm": 6.28125,
      "learning_rate": 0.0001712896234879656,
      "loss": 0.8826,
      "step": 42335
    },
    {
      "epoch": 3.6424331067710574,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017126924813382972,
      "loss": 0.7069,
      "step": 42336
    },
    {
      "epoch": 3.642519143078379,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017124887374116732,
      "loss": 0.8425,
      "step": 42337
    },
    {
      "epoch": 3.6426051793857006,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017122850031003824,
      "loss": 1.6511,
      "step": 42338
    },
    {
      "epoch": 3.6426912156930227,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017120812784050162,
      "loss": 0.5998,
      "step": 42339
    },
    {
      "epoch": 3.6427772520003443,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001711877563326173,
      "loss": 1.1162,
      "step": 42340
    },
    {
      "epoch": 3.642863288307666,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017116738578644504,
      "loss": 1.2874,
      "step": 42341
    },
    {
      "epoch": 3.6429493246149875,
      "grad_norm": 4.625,
      "learning_rate": 0.00017114701620204399,
      "loss": 1.2626,
      "step": 42342
    },
    {
      "epoch": 3.643035360922309,
      "grad_norm": 1.125,
      "learning_rate": 0.000171126647579474,
      "loss": 1.3571,
      "step": 42343
    },
    {
      "epoch": 3.643121397229631,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001711062799187948,
      "loss": 0.751,
      "step": 42344
    },
    {
      "epoch": 3.6432074335369524,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017108591322006556,
      "loss": 1.8384,
      "step": 42345
    },
    {
      "epoch": 3.6432934698442745,
      "grad_norm": 43.75,
      "learning_rate": 0.00017106554748334607,
      "loss": 0.9077,
      "step": 42346
    },
    {
      "epoch": 3.643379506151596,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000171045182708696,
      "loss": 1.1683,
      "step": 42347
    },
    {
      "epoch": 3.6434655424589177,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017102481889617473,
      "loss": 1.1302,
      "step": 42348
    },
    {
      "epoch": 3.6435515787662394,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00017100445604584192,
      "loss": 0.9759,
      "step": 42349
    },
    {
      "epoch": 3.643637615073561,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000170984094157757,
      "loss": 1.2023,
      "step": 42350
    },
    {
      "epoch": 3.6437236513808826,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001709637332319796,
      "loss": 1.9058,
      "step": 42351
    },
    {
      "epoch": 3.6438096876882042,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017094337326856946,
      "loss": 0.797,
      "step": 42352
    },
    {
      "epoch": 3.6438957239955263,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017092301426758577,
      "loss": 1.0603,
      "step": 42353
    },
    {
      "epoch": 3.643981760302848,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001709026562290883,
      "loss": 0.7728,
      "step": 42354
    },
    {
      "epoch": 3.6440677966101696,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017088229915313653,
      "loss": 1.1571,
      "step": 42355
    },
    {
      "epoch": 3.644153832917491,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017086194303979002,
      "loss": 1.079,
      "step": 42356
    },
    {
      "epoch": 3.644239869224813,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017084158788910846,
      "loss": 1.4986,
      "step": 42357
    },
    {
      "epoch": 3.6443259055321344,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001708212337011511,
      "loss": 0.6615,
      "step": 42358
    },
    {
      "epoch": 3.644411941839456,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017080088047597764,
      "loss": 0.8907,
      "step": 42359
    },
    {
      "epoch": 3.644497978146778,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001707805282136477,
      "loss": 1.5821,
      "step": 42360
    },
    {
      "epoch": 3.6445840144540997,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017076017691422052,
      "loss": 1.8337,
      "step": 42361
    },
    {
      "epoch": 3.6446700507614214,
      "grad_norm": 30.375,
      "learning_rate": 0.00017073982657775578,
      "loss": 0.6498,
      "step": 42362
    },
    {
      "epoch": 3.644756087068743,
      "grad_norm": 142.0,
      "learning_rate": 0.00017071947720431318,
      "loss": 1.5002,
      "step": 42363
    },
    {
      "epoch": 3.6448421233760646,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017069912879395194,
      "loss": 0.6563,
      "step": 42364
    },
    {
      "epoch": 3.6449281596833862,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017067878134673182,
      "loss": 0.8659,
      "step": 42365
    },
    {
      "epoch": 3.645014195990708,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001706584348627121,
      "loss": 0.9488,
      "step": 42366
    },
    {
      "epoch": 3.64510023229803,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017063808934195235,
      "loss": 1.1183,
      "step": 42367
    },
    {
      "epoch": 3.6451862686053516,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00017061774478451235,
      "loss": 1.2692,
      "step": 42368
    },
    {
      "epoch": 3.645272304912673,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001705974011904512,
      "loss": 1.2978,
      "step": 42369
    },
    {
      "epoch": 3.645358341219995,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017057705855982858,
      "loss": 0.7309,
      "step": 42370
    },
    {
      "epoch": 3.6454443775273164,
      "grad_norm": 1.234375,
      "learning_rate": 0.00017055671689270412,
      "loss": 1.9505,
      "step": 42371
    },
    {
      "epoch": 3.6455304138346385,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001705363761891371,
      "loss": 0.9628,
      "step": 42372
    },
    {
      "epoch": 3.6456164501419597,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017051603644918707,
      "loss": 0.7452,
      "step": 42373
    },
    {
      "epoch": 3.6457024864492817,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017049569767291372,
      "loss": 1.5216,
      "step": 42374
    },
    {
      "epoch": 3.6457885227566034,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017047535986037615,
      "loss": 0.975,
      "step": 42375
    },
    {
      "epoch": 3.645874559063925,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00017045502301163429,
      "loss": 1.4523,
      "step": 42376
    },
    {
      "epoch": 3.6459605953712466,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017043468712674725,
      "loss": 0.5087,
      "step": 42377
    },
    {
      "epoch": 3.6460466316785682,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001704143522057746,
      "loss": 1.3576,
      "step": 42378
    },
    {
      "epoch": 3.6461326679858903,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00017039401824877603,
      "loss": 0.736,
      "step": 42379
    },
    {
      "epoch": 3.6462187042932115,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017037368525581076,
      "loss": 0.874,
      "step": 42380
    },
    {
      "epoch": 3.6463047406005336,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001703533532269383,
      "loss": 0.9005,
      "step": 42381
    },
    {
      "epoch": 3.646390776907855,
      "grad_norm": 1.25,
      "learning_rate": 0.0001703330221622183,
      "loss": 1.3598,
      "step": 42382
    },
    {
      "epoch": 3.646476813215177,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017031269206170996,
      "loss": 0.8498,
      "step": 42383
    },
    {
      "epoch": 3.6465628495224984,
      "grad_norm": 1.734375,
      "learning_rate": 0.0001702923629254729,
      "loss": 1.3024,
      "step": 42384
    },
    {
      "epoch": 3.64664888582982,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00017027203475356674,
      "loss": 1.7452,
      "step": 42385
    },
    {
      "epoch": 3.646734922137142,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001702517075460505,
      "loss": 1.0073,
      "step": 42386
    },
    {
      "epoch": 3.6468209584444633,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00017023138130298394,
      "loss": 1.074,
      "step": 42387
    },
    {
      "epoch": 3.6469069947517854,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001702110560244266,
      "loss": 1.5648,
      "step": 42388
    },
    {
      "epoch": 3.646993031059107,
      "grad_norm": 1.125,
      "learning_rate": 0.00017019073171043765,
      "loss": 1.1311,
      "step": 42389
    },
    {
      "epoch": 3.6470790673664286,
      "grad_norm": 1.40625,
      "learning_rate": 0.00017017040836107678,
      "loss": 1.0908,
      "step": 42390
    },
    {
      "epoch": 3.6471651036737502,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017015008597640318,
      "loss": 1.3493,
      "step": 42391
    },
    {
      "epoch": 3.647251139981072,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00017012976455647622,
      "loss": 0.6633,
      "step": 42392
    },
    {
      "epoch": 3.647337176288394,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017010944410135592,
      "loss": 1.2638,
      "step": 42393
    },
    {
      "epoch": 3.6474232125957156,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017008912461110115,
      "loss": 1.8525,
      "step": 42394
    },
    {
      "epoch": 3.647509248903037,
      "grad_norm": 1.390625,
      "learning_rate": 0.00017006880608577175,
      "loss": 1.5519,
      "step": 42395
    },
    {
      "epoch": 3.647595285210359,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00017004848852542664,
      "loss": 0.9617,
      "step": 42396
    },
    {
      "epoch": 3.6476813215176804,
      "grad_norm": 440.0,
      "learning_rate": 0.00017002817193012565,
      "loss": 1.621,
      "step": 42397
    },
    {
      "epoch": 3.647767357825002,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00017000785629992814,
      "loss": 2.0669,
      "step": 42398
    },
    {
      "epoch": 3.6478533941323237,
      "grad_norm": 1.140625,
      "learning_rate": 0.00016998754163489337,
      "loss": 1.4442,
      "step": 42399
    },
    {
      "epoch": 3.6479394304396457,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00016996722793508084,
      "loss": 0.8495,
      "step": 42400
    },
    {
      "epoch": 3.6480254667469674,
      "grad_norm": 1.0,
      "learning_rate": 0.00016994691520055017,
      "loss": 0.7977,
      "step": 42401
    },
    {
      "epoch": 3.648111503054289,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001699266034313604,
      "loss": 1.1069,
      "step": 42402
    },
    {
      "epoch": 3.6481975393616106,
      "grad_norm": 20.375,
      "learning_rate": 0.00016990629262757112,
      "loss": 0.9281,
      "step": 42403
    },
    {
      "epoch": 3.6482835756689322,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016988598278924188,
      "loss": 0.8046,
      "step": 42404
    },
    {
      "epoch": 3.648369611976254,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016986567391643183,
      "loss": 1.2294,
      "step": 42405
    },
    {
      "epoch": 3.6484556482835755,
      "grad_norm": 1.203125,
      "learning_rate": 0.00016984536600920058,
      "loss": 1.0594,
      "step": 42406
    },
    {
      "epoch": 3.6485416845908976,
      "grad_norm": 1.5,
      "learning_rate": 0.00016982505906760725,
      "loss": 1.1322,
      "step": 42407
    },
    {
      "epoch": 3.648627720898219,
      "grad_norm": 3.0625,
      "learning_rate": 0.00016980475309171146,
      "loss": 1.9956,
      "step": 42408
    },
    {
      "epoch": 3.648713757205541,
      "grad_norm": 68.0,
      "learning_rate": 0.00016978444808157269,
      "loss": 0.9453,
      "step": 42409
    },
    {
      "epoch": 3.6487997935128624,
      "grad_norm": 1.03125,
      "learning_rate": 0.00016976414403725005,
      "loss": 1.5309,
      "step": 42410
    },
    {
      "epoch": 3.648885829820184,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016974384095880303,
      "loss": 0.5355,
      "step": 42411
    },
    {
      "epoch": 3.6489718661275057,
      "grad_norm": 7.25,
      "learning_rate": 0.00016972353884629117,
      "loss": 1.0429,
      "step": 42412
    },
    {
      "epoch": 3.6490579024348273,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00016970323769977357,
      "loss": 2.0728,
      "step": 42413
    },
    {
      "epoch": 3.6491439387421494,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001696829375193098,
      "loss": 1.5445,
      "step": 42414
    },
    {
      "epoch": 3.649229975049471,
      "grad_norm": 1.1875,
      "learning_rate": 0.00016966263830495937,
      "loss": 1.6654,
      "step": 42415
    },
    {
      "epoch": 3.6493160113567926,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001696423400567813,
      "loss": 1.2806,
      "step": 42416
    },
    {
      "epoch": 3.6494020476641142,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001696220427748353,
      "loss": 1.0757,
      "step": 42417
    },
    {
      "epoch": 3.649488083971436,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016960174645918037,
      "loss": 0.915,
      "step": 42418
    },
    {
      "epoch": 3.6495741202787575,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001695814511098761,
      "loss": 0.8694,
      "step": 42419
    },
    {
      "epoch": 3.649660156586079,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016956115672698196,
      "loss": 0.8204,
      "step": 42420
    },
    {
      "epoch": 3.649746192893401,
      "grad_norm": 1.359375,
      "learning_rate": 0.00016954086331055708,
      "loss": 1.8549,
      "step": 42421
    },
    {
      "epoch": 3.649832229200723,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016952057086066082,
      "loss": 0.7881,
      "step": 42422
    },
    {
      "epoch": 3.6499182655080444,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001695002793773528,
      "loss": 0.7463,
      "step": 42423
    },
    {
      "epoch": 3.650004301815366,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016947998886069206,
      "loss": 0.8567,
      "step": 42424
    },
    {
      "epoch": 3.6500903381226877,
      "grad_norm": 1.078125,
      "learning_rate": 0.000169459699310738,
      "loss": 1.0636,
      "step": 42425
    },
    {
      "epoch": 3.6501763744300093,
      "grad_norm": 422.0,
      "learning_rate": 0.00016943941072755027,
      "loss": 1.3242,
      "step": 42426
    },
    {
      "epoch": 3.650262410737331,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001694191231111878,
      "loss": 1.5967,
      "step": 42427
    },
    {
      "epoch": 3.650348447044653,
      "grad_norm": 1.28125,
      "learning_rate": 0.00016939883646171006,
      "loss": 1.0638,
      "step": 42428
    },
    {
      "epoch": 3.6504344833519746,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00016937855077917642,
      "loss": 1.1189,
      "step": 42429
    },
    {
      "epoch": 3.6505205196592962,
      "grad_norm": 23.25,
      "learning_rate": 0.00016935826606364623,
      "loss": 0.941,
      "step": 42430
    },
    {
      "epoch": 3.650606555966618,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000169337982315179,
      "loss": 0.9818,
      "step": 42431
    },
    {
      "epoch": 3.6506925922739395,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016931769953383368,
      "loss": 1.2541,
      "step": 42432
    },
    {
      "epoch": 3.6507786285812616,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00016929741771966978,
      "loss": 1.487,
      "step": 42433
    },
    {
      "epoch": 3.6508646648885827,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016927713687274676,
      "loss": 0.8648,
      "step": 42434
    },
    {
      "epoch": 3.650950701195905,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016925685699312366,
      "loss": 1.2836,
      "step": 42435
    },
    {
      "epoch": 3.6510367375032264,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016923657808086008,
      "loss": 0.9575,
      "step": 42436
    },
    {
      "epoch": 3.651122773810548,
      "grad_norm": 1.15625,
      "learning_rate": 0.00016921630013601497,
      "loss": 0.8639,
      "step": 42437
    },
    {
      "epoch": 3.6512088101178697,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00016919602315864792,
      "loss": 0.7908,
      "step": 42438
    },
    {
      "epoch": 3.6512948464251913,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00016917574714881827,
      "loss": 0.7884,
      "step": 42439
    },
    {
      "epoch": 3.6513808827325134,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016915547210658504,
      "loss": 1.0898,
      "step": 42440
    },
    {
      "epoch": 3.6514669190398346,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016913519803200778,
      "loss": 0.8929,
      "step": 42441
    },
    {
      "epoch": 3.6515529553471566,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016911492492514585,
      "loss": 0.9239,
      "step": 42442
    },
    {
      "epoch": 3.6516389916544783,
      "grad_norm": 1.234375,
      "learning_rate": 0.00016909465278605823,
      "loss": 1.0997,
      "step": 42443
    },
    {
      "epoch": 3.6517250279618,
      "grad_norm": 1.3125,
      "learning_rate": 0.00016907438161480436,
      "loss": 1.4166,
      "step": 42444
    },
    {
      "epoch": 3.6518110642691215,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00016905411141144384,
      "loss": 1.666,
      "step": 42445
    },
    {
      "epoch": 3.651897100576443,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001690338421760354,
      "loss": 0.9258,
      "step": 42446
    },
    {
      "epoch": 3.651983136883765,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001690135739086388,
      "loss": 0.9945,
      "step": 42447
    },
    {
      "epoch": 3.6520691731910864,
      "grad_norm": 1.203125,
      "learning_rate": 0.000168993306609313,
      "loss": 1.6243,
      "step": 42448
    },
    {
      "epoch": 3.6521552094984084,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016897304027811737,
      "loss": 0.5617,
      "step": 42449
    },
    {
      "epoch": 3.65224124580573,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001689527749151114,
      "loss": 0.8178,
      "step": 42450
    },
    {
      "epoch": 3.6523272821130517,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016893251052035397,
      "loss": 0.7246,
      "step": 42451
    },
    {
      "epoch": 3.6524133184203733,
      "grad_norm": 1.453125,
      "learning_rate": 0.00016891224709390457,
      "loss": 1.1695,
      "step": 42452
    },
    {
      "epoch": 3.652499354727695,
      "grad_norm": 1.25,
      "learning_rate": 0.00016889198463582267,
      "loss": 0.7529,
      "step": 42453
    },
    {
      "epoch": 3.652585391035017,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016887172314616712,
      "loss": 0.8263,
      "step": 42454
    },
    {
      "epoch": 3.6526714273423386,
      "grad_norm": 1.34375,
      "learning_rate": 0.00016885146262499734,
      "loss": 1.0812,
      "step": 42455
    },
    {
      "epoch": 3.6527574636496603,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001688312030723728,
      "loss": 0.8725,
      "step": 42456
    },
    {
      "epoch": 3.652843499956982,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001688109444883525,
      "loss": 1.1691,
      "step": 42457
    },
    {
      "epoch": 3.6529295362643035,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001687906868729957,
      "loss": 1.3453,
      "step": 42458
    },
    {
      "epoch": 3.653015572571625,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001687704302263619,
      "loss": 0.4839,
      "step": 42459
    },
    {
      "epoch": 3.6531016088789467,
      "grad_norm": 1.03125,
      "learning_rate": 0.00016875017454851004,
      "loss": 1.1634,
      "step": 42460
    },
    {
      "epoch": 3.653187645186269,
      "grad_norm": 5.3125,
      "learning_rate": 0.00016872991983949966,
      "loss": 0.6952,
      "step": 42461
    },
    {
      "epoch": 3.6532736814935904,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016870966609938958,
      "loss": 1.1348,
      "step": 42462
    },
    {
      "epoch": 3.653359717800912,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001686894133282394,
      "loss": 0.6595,
      "step": 42463
    },
    {
      "epoch": 3.6534457541082337,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00016866916152610833,
      "loss": 0.6908,
      "step": 42464
    },
    {
      "epoch": 3.6535317904155553,
      "grad_norm": 13.125,
      "learning_rate": 0.00016864891069305545,
      "loss": 0.8413,
      "step": 42465
    },
    {
      "epoch": 3.653617826722877,
      "grad_norm": 3.625,
      "learning_rate": 0.00016862866082913996,
      "loss": 1.3979,
      "step": 42466
    },
    {
      "epoch": 3.6537038630301986,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016860841193442128,
      "loss": 0.8527,
      "step": 42467
    },
    {
      "epoch": 3.6537898993375206,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001685881640089585,
      "loss": 1.7049,
      "step": 42468
    },
    {
      "epoch": 3.6538759356448423,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00016856791705281105,
      "loss": 1.2562,
      "step": 42469
    },
    {
      "epoch": 3.653961971952164,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001685476710660378,
      "loss": 1.2543,
      "step": 42470
    },
    {
      "epoch": 3.6540480082594855,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001685274260486982,
      "loss": 1.2273,
      "step": 42471
    },
    {
      "epoch": 3.654134044566807,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00016850718200085151,
      "loss": 1.5632,
      "step": 42472
    },
    {
      "epoch": 3.6542200808741288,
      "grad_norm": 10.125,
      "learning_rate": 0.00016848693892255673,
      "loss": 1.1778,
      "step": 42473
    },
    {
      "epoch": 3.6543061171814504,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00016846669681387316,
      "loss": 1.2309,
      "step": 42474
    },
    {
      "epoch": 3.6543921534887724,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016844645567486022,
      "loss": 0.9518,
      "step": 42475
    },
    {
      "epoch": 3.654478189796094,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016842621550557668,
      "loss": 0.9856,
      "step": 42476
    },
    {
      "epoch": 3.6545642261034157,
      "grad_norm": 2.15625,
      "learning_rate": 0.00016840597630608217,
      "loss": 1.7761,
      "step": 42477
    },
    {
      "epoch": 3.6546502624107373,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016838573807643554,
      "loss": 0.5956,
      "step": 42478
    },
    {
      "epoch": 3.654736298718059,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001683655008166961,
      "loss": 0.6616,
      "step": 42479
    },
    {
      "epoch": 3.6548223350253806,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001683452645269233,
      "loss": 0.9314,
      "step": 42480
    },
    {
      "epoch": 3.654908371332702,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00016832502920717586,
      "loss": 1.534,
      "step": 42481
    },
    {
      "epoch": 3.6549944076400243,
      "grad_norm": 4.09375,
      "learning_rate": 0.0001683047948575132,
      "loss": 1.2064,
      "step": 42482
    },
    {
      "epoch": 3.655080443947346,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001682845614779947,
      "loss": 1.3688,
      "step": 42483
    },
    {
      "epoch": 3.6551664802546675,
      "grad_norm": 1.484375,
      "learning_rate": 0.00016826432906867923,
      "loss": 1.2384,
      "step": 42484
    },
    {
      "epoch": 3.655252516561989,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000168244097629626,
      "loss": 0.933,
      "step": 42485
    },
    {
      "epoch": 3.6553385528693108,
      "grad_norm": 7.09375,
      "learning_rate": 0.00016822386716089445,
      "loss": 0.8012,
      "step": 42486
    },
    {
      "epoch": 3.6554245891766324,
      "grad_norm": 11.8125,
      "learning_rate": 0.00016820363766254336,
      "loss": 1.1299,
      "step": 42487
    },
    {
      "epoch": 3.655510625483954,
      "grad_norm": 11.125,
      "learning_rate": 0.00016818340913463231,
      "loss": 1.4369,
      "step": 42488
    },
    {
      "epoch": 3.655596661791276,
      "grad_norm": 17.625,
      "learning_rate": 0.0001681631815772201,
      "loss": 1.5485,
      "step": 42489
    },
    {
      "epoch": 3.6556826980985977,
      "grad_norm": 69120.0,
      "learning_rate": 0.00016814295499036603,
      "loss": 1.1325,
      "step": 42490
    },
    {
      "epoch": 3.6557687344059193,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001681227293741294,
      "loss": 1.0714,
      "step": 42491
    },
    {
      "epoch": 3.655854770713241,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016810250472856913,
      "loss": 0.8584,
      "step": 42492
    },
    {
      "epoch": 3.6559408070205626,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016808228105374447,
      "loss": 0.9989,
      "step": 42493
    },
    {
      "epoch": 3.6560268433278846,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00016806205834971473,
      "loss": 0.9403,
      "step": 42494
    },
    {
      "epoch": 3.656112879635206,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016804183661653875,
      "loss": 1.1023,
      "step": 42495
    },
    {
      "epoch": 3.656198915942528,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00016802161585427583,
      "loss": 0.6924,
      "step": 42496
    },
    {
      "epoch": 3.6562849522498495,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001680013960629853,
      "loss": 0.6971,
      "step": 42497
    },
    {
      "epoch": 3.656370988557171,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00016798117724272594,
      "loss": 1.8071,
      "step": 42498
    },
    {
      "epoch": 3.6564570248644928,
      "grad_norm": 1.984375,
      "learning_rate": 0.0001679609593935571,
      "loss": 0.8607,
      "step": 42499
    },
    {
      "epoch": 3.6565430611718144,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00016794074251553798,
      "loss": 1.988,
      "step": 42500
    },
    {
      "epoch": 3.6566290974791364,
      "grad_norm": 2.296875,
      "learning_rate": 0.00016792052660872747,
      "loss": 0.5285,
      "step": 42501
    },
    {
      "epoch": 3.6567151337864576,
      "grad_norm": 1.234375,
      "learning_rate": 0.00016790031167318497,
      "loss": 1.0811,
      "step": 42502
    },
    {
      "epoch": 3.6568011700937797,
      "grad_norm": 0.859375,
      "learning_rate": 0.00016788009770896917,
      "loss": 1.1157,
      "step": 42503
    },
    {
      "epoch": 3.6568872064011013,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016785988471613967,
      "loss": 1.063,
      "step": 42504
    },
    {
      "epoch": 3.656973242708423,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001678396726947556,
      "loss": 0.824,
      "step": 42505
    },
    {
      "epoch": 3.6570592790157446,
      "grad_norm": 99.5,
      "learning_rate": 0.00016781946164487566,
      "loss": 1.5099,
      "step": 42506
    },
    {
      "epoch": 3.657145315323066,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016779925156655933,
      "loss": 1.4681,
      "step": 42507
    },
    {
      "epoch": 3.6572313516303883,
      "grad_norm": 82.5,
      "learning_rate": 0.00016777904245986547,
      "loss": 0.7242,
      "step": 42508
    },
    {
      "epoch": 3.6573173879377094,
      "grad_norm": 83.0,
      "learning_rate": 0.0001677588343248533,
      "loss": 0.6417,
      "step": 42509
    },
    {
      "epoch": 3.6574034242450315,
      "grad_norm": 27.25,
      "learning_rate": 0.00016773862716158207,
      "loss": 0.7181,
      "step": 42510
    },
    {
      "epoch": 3.657489460552353,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016771842097011057,
      "loss": 1.2745,
      "step": 42511
    },
    {
      "epoch": 3.6575754968596748,
      "grad_norm": 1.796875,
      "learning_rate": 0.00016769821575049804,
      "loss": 0.8595,
      "step": 42512
    },
    {
      "epoch": 3.6576615331669964,
      "grad_norm": 1.25,
      "learning_rate": 0.00016767801150280383,
      "loss": 1.0504,
      "step": 42513
    },
    {
      "epoch": 3.657747569474318,
      "grad_norm": 204.0,
      "learning_rate": 0.00016765780822708653,
      "loss": 1.1882,
      "step": 42514
    },
    {
      "epoch": 3.65783360578164,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001676376059234056,
      "loss": 1.7412,
      "step": 42515
    },
    {
      "epoch": 3.6579196420889617,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00016761740459182018,
      "loss": 1.1527,
      "step": 42516
    },
    {
      "epoch": 3.6580056783962833,
      "grad_norm": 116.0,
      "learning_rate": 0.000167597204232389,
      "loss": 0.736,
      "step": 42517
    },
    {
      "epoch": 3.658091714703605,
      "grad_norm": 12.0625,
      "learning_rate": 0.00016757700484517157,
      "loss": 1.4302,
      "step": 42518
    },
    {
      "epoch": 3.6581777510109266,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016755680643022652,
      "loss": 0.547,
      "step": 42519
    },
    {
      "epoch": 3.658263787318248,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016753660898761324,
      "loss": 0.8753,
      "step": 42520
    },
    {
      "epoch": 3.65834982362557,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001675164125173908,
      "loss": 0.8682,
      "step": 42521
    },
    {
      "epoch": 3.658435859932892,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001674962170196181,
      "loss": 1.0206,
      "step": 42522
    },
    {
      "epoch": 3.6585218962402135,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00016747602249435423,
      "loss": 1.1839,
      "step": 42523
    },
    {
      "epoch": 3.658607932547535,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001674558289416585,
      "loss": 1.1874,
      "step": 42524
    },
    {
      "epoch": 3.6586939688548568,
      "grad_norm": 1.34375,
      "learning_rate": 0.00016743563636158966,
      "loss": 1.6111,
      "step": 42525
    },
    {
      "epoch": 3.6587800051621784,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001674154447542069,
      "loss": 0.9713,
      "step": 42526
    },
    {
      "epoch": 3.6588660414695,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016739525411956941,
      "loss": 0.9015,
      "step": 42527
    },
    {
      "epoch": 3.6589520777768216,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016737506445773599,
      "loss": 1.3464,
      "step": 42528
    },
    {
      "epoch": 3.6590381140841437,
      "grad_norm": 30.625,
      "learning_rate": 0.0001673548757687659,
      "loss": 1.1477,
      "step": 42529
    },
    {
      "epoch": 3.6591241503914653,
      "grad_norm": 31.375,
      "learning_rate": 0.000167334688052718,
      "loss": 0.7274,
      "step": 42530
    },
    {
      "epoch": 3.659210186698787,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00016731450130965143,
      "loss": 1.074,
      "step": 42531
    },
    {
      "epoch": 3.6592962230061086,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016729431553962542,
      "loss": 1.0077,
      "step": 42532
    },
    {
      "epoch": 3.65938225931343,
      "grad_norm": 0.75,
      "learning_rate": 0.00016727413074269865,
      "loss": 1.1798,
      "step": 42533
    },
    {
      "epoch": 3.659468295620752,
      "grad_norm": 1.078125,
      "learning_rate": 0.00016725394691893036,
      "loss": 0.7108,
      "step": 42534
    },
    {
      "epoch": 3.6595543319280734,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001672337640683797,
      "loss": 1.3367,
      "step": 42535
    },
    {
      "epoch": 3.6596403682353955,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001672135821911054,
      "loss": 0.8053,
      "step": 42536
    },
    {
      "epoch": 3.659726404542717,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016719340128716666,
      "loss": 1.0381,
      "step": 42537
    },
    {
      "epoch": 3.6598124408500388,
      "grad_norm": 1.125,
      "learning_rate": 0.0001671732213566226,
      "loss": 0.8307,
      "step": 42538
    },
    {
      "epoch": 3.6598984771573604,
      "grad_norm": 30.0,
      "learning_rate": 0.00016715304239953206,
      "loss": 0.9026,
      "step": 42539
    },
    {
      "epoch": 3.659984513464682,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00016713286441595406,
      "loss": 1.9048,
      "step": 42540
    },
    {
      "epoch": 3.6600705497720036,
      "grad_norm": 4.375,
      "learning_rate": 0.0001671126874059477,
      "loss": 0.9442,
      "step": 42541
    },
    {
      "epoch": 3.6601565860793253,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016709251136957205,
      "loss": 1.1867,
      "step": 42542
    },
    {
      "epoch": 3.6602426223866473,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016707233630688616,
      "loss": 1.2229,
      "step": 42543
    },
    {
      "epoch": 3.660328658693969,
      "grad_norm": 1.5625,
      "learning_rate": 0.00016705216221794884,
      "loss": 0.9128,
      "step": 42544
    },
    {
      "epoch": 3.6604146950012906,
      "grad_norm": 2.46875,
      "learning_rate": 0.00016703198910281915,
      "loss": 1.0525,
      "step": 42545
    },
    {
      "epoch": 3.660500731308612,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016701181696155627,
      "loss": 1.5496,
      "step": 42546
    },
    {
      "epoch": 3.660586767615934,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016699164579421887,
      "loss": 1.3518,
      "step": 42547
    },
    {
      "epoch": 3.660672803923256,
      "grad_norm": 2.984375,
      "learning_rate": 0.00016697147560086639,
      "loss": 0.6456,
      "step": 42548
    },
    {
      "epoch": 3.660758840230577,
      "grad_norm": 1.03125,
      "learning_rate": 0.00016695130638155737,
      "loss": 0.7788,
      "step": 42549
    },
    {
      "epoch": 3.660844876537899,
      "grad_norm": 237.0,
      "learning_rate": 0.00016693113813635104,
      "loss": 1.5004,
      "step": 42550
    },
    {
      "epoch": 3.6609309128452208,
      "grad_norm": 3456.0,
      "learning_rate": 0.00016691097086530643,
      "loss": 0.5691,
      "step": 42551
    },
    {
      "epoch": 3.6610169491525424,
      "grad_norm": 190.0,
      "learning_rate": 0.0001668908045684824,
      "loss": 1.1211,
      "step": 42552
    },
    {
      "epoch": 3.661102985459864,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016687063924593787,
      "loss": 0.6772,
      "step": 42553
    },
    {
      "epoch": 3.6611890217671856,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00016685047489773213,
      "loss": 1.2821,
      "step": 42554
    },
    {
      "epoch": 3.6612750580745077,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016683031152392387,
      "loss": 0.6079,
      "step": 42555
    },
    {
      "epoch": 3.661361094381829,
      "grad_norm": 1.46875,
      "learning_rate": 0.00016681014912457205,
      "loss": 1.1057,
      "step": 42556
    },
    {
      "epoch": 3.661447130689151,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016678998769973597,
      "loss": 0.602,
      "step": 42557
    },
    {
      "epoch": 3.6615331669964726,
      "grad_norm": 11.25,
      "learning_rate": 0.00016676982724947415,
      "loss": 1.1059,
      "step": 42558
    },
    {
      "epoch": 3.661619203303794,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016674966777384598,
      "loss": 1.038,
      "step": 42559
    },
    {
      "epoch": 3.661705239611116,
      "grad_norm": 1384.0,
      "learning_rate": 0.00016672950927291004,
      "loss": 0.8105,
      "step": 42560
    },
    {
      "epoch": 3.6617912759184374,
      "grad_norm": 31.25,
      "learning_rate": 0.00016670935174672546,
      "loss": 0.9358,
      "step": 42561
    },
    {
      "epoch": 3.6618773122257595,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016668919519535142,
      "loss": 0.9245,
      "step": 42562
    },
    {
      "epoch": 3.6619633485330807,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00016666903961884644,
      "loss": 1.1535,
      "step": 42563
    },
    {
      "epoch": 3.6620493848404028,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001666488850172697,
      "loss": 1.0627,
      "step": 42564
    },
    {
      "epoch": 3.6621354211477244,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001666287313906803,
      "loss": 1.1207,
      "step": 42565
    },
    {
      "epoch": 3.662221457455046,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00016660857873913688,
      "loss": 0.9127,
      "step": 42566
    },
    {
      "epoch": 3.6623074937623676,
      "grad_norm": 1.96875,
      "learning_rate": 0.0001665884270626985,
      "loss": 1.342,
      "step": 42567
    },
    {
      "epoch": 3.6623935300696893,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001665682763614243,
      "loss": 1.3863,
      "step": 42568
    },
    {
      "epoch": 3.6624795663770113,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00016654812663537287,
      "loss": 1.6317,
      "step": 42569
    },
    {
      "epoch": 3.662565602684333,
      "grad_norm": 6.28125,
      "learning_rate": 0.00016652797788460334,
      "loss": 1.3109,
      "step": 42570
    },
    {
      "epoch": 3.6626516389916546,
      "grad_norm": 1.859375,
      "learning_rate": 0.00016650783010917474,
      "loss": 0.8306,
      "step": 42571
    },
    {
      "epoch": 3.662737675298976,
      "grad_norm": 1.15625,
      "learning_rate": 0.00016648768330914578,
      "loss": 0.8836,
      "step": 42572
    },
    {
      "epoch": 3.662823711606298,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001664675374845756,
      "loss": 1.7957,
      "step": 42573
    },
    {
      "epoch": 3.6629097479136195,
      "grad_norm": 1.109375,
      "learning_rate": 0.00016644739263552282,
      "loss": 1.5447,
      "step": 42574
    },
    {
      "epoch": 3.662995784220941,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00016642724876204658,
      "loss": 1.1102,
      "step": 42575
    },
    {
      "epoch": 3.663081820528263,
      "grad_norm": 9.25,
      "learning_rate": 0.00016640710586420595,
      "loss": 1.0354,
      "step": 42576
    },
    {
      "epoch": 3.6631678568355848,
      "grad_norm": 2.65625,
      "learning_rate": 0.00016638696394205926,
      "loss": 1.3714,
      "step": 42577
    },
    {
      "epoch": 3.6632538931429064,
      "grad_norm": 1.359375,
      "learning_rate": 0.00016636682299566618,
      "loss": 1.2825,
      "step": 42578
    },
    {
      "epoch": 3.663339929450228,
      "grad_norm": 5.0,
      "learning_rate": 0.0001663466830250851,
      "loss": 1.2438,
      "step": 42579
    },
    {
      "epoch": 3.6634259657575496,
      "grad_norm": 2.421875,
      "learning_rate": 0.00016632654403037506,
      "loss": 1.7528,
      "step": 42580
    },
    {
      "epoch": 3.6635120020648713,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016630640601159518,
      "loss": 0.6421,
      "step": 42581
    },
    {
      "epoch": 3.663598038372193,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00016628626896880393,
      "loss": 0.9903,
      "step": 42582
    },
    {
      "epoch": 3.663684074679515,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001662661329020605,
      "loss": 1.4879,
      "step": 42583
    },
    {
      "epoch": 3.6637701109868366,
      "grad_norm": 63.5,
      "learning_rate": 0.0001662459978114239,
      "loss": 0.8014,
      "step": 42584
    },
    {
      "epoch": 3.663856147294158,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016622586369695265,
      "loss": 1.1277,
      "step": 42585
    },
    {
      "epoch": 3.66394218360148,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00016620573055870581,
      "loss": 1.1794,
      "step": 42586
    },
    {
      "epoch": 3.6640282199088015,
      "grad_norm": 332.0,
      "learning_rate": 0.0001661855983967425,
      "loss": 1.4669,
      "step": 42587
    },
    {
      "epoch": 3.664114256216123,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00016616546721112118,
      "loss": 1.2707,
      "step": 42588
    },
    {
      "epoch": 3.6642002925234447,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016614533700190116,
      "loss": 1.7271,
      "step": 42589
    },
    {
      "epoch": 3.6642863288307668,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001661252077691409,
      "loss": 0.62,
      "step": 42590
    },
    {
      "epoch": 3.6643723651380884,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0001661050795128995,
      "loss": 1.0183,
      "step": 42591
    },
    {
      "epoch": 3.66445840144541,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016608495223323594,
      "loss": 1.0719,
      "step": 42592
    },
    {
      "epoch": 3.6645444377527316,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016606482593020878,
      "loss": 1.8004,
      "step": 42593
    },
    {
      "epoch": 3.6646304740600533,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00016604470060387706,
      "loss": 0.6814,
      "step": 42594
    },
    {
      "epoch": 3.664716510367375,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00016602457625429983,
      "loss": 1.2193,
      "step": 42595
    },
    {
      "epoch": 3.6648025466746965,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00016600445288153558,
      "loss": 1.4973,
      "step": 42596
    },
    {
      "epoch": 3.6648885829820186,
      "grad_norm": 1.15625,
      "learning_rate": 0.00016598433048564332,
      "loss": 0.5645,
      "step": 42597
    },
    {
      "epoch": 3.66497461928934,
      "grad_norm": 0.984375,
      "learning_rate": 0.00016596420906668212,
      "loss": 0.6658,
      "step": 42598
    },
    {
      "epoch": 3.665060655596662,
      "grad_norm": 58.25,
      "learning_rate": 0.0001659440886247105,
      "loss": 1.4872,
      "step": 42599
    },
    {
      "epoch": 3.6651466919039835,
      "grad_norm": 1.25,
      "learning_rate": 0.00016592396915978758,
      "loss": 1.3302,
      "step": 42600
    },
    {
      "epoch": 3.665232728211305,
      "grad_norm": 1.46875,
      "learning_rate": 0.00016590385067197195,
      "loss": 1.6281,
      "step": 42601
    },
    {
      "epoch": 3.6653187645186267,
      "grad_norm": 17.5,
      "learning_rate": 0.00016588373316132256,
      "loss": 0.9233,
      "step": 42602
    },
    {
      "epoch": 3.6654048008259483,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001658636166278984,
      "loss": 1.3195,
      "step": 42603
    },
    {
      "epoch": 3.6654908371332704,
      "grad_norm": 1.78125,
      "learning_rate": 0.00016584350107175805,
      "loss": 1.3663,
      "step": 42604
    },
    {
      "epoch": 3.665576873440592,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016582338649296047,
      "loss": 0.5175,
      "step": 42605
    },
    {
      "epoch": 3.6656629097479136,
      "grad_norm": 1.15625,
      "learning_rate": 0.00016580327289156466,
      "loss": 1.1844,
      "step": 42606
    },
    {
      "epoch": 3.6657489460552353,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016578316026762908,
      "loss": 1.0632,
      "step": 42607
    },
    {
      "epoch": 3.665834982362557,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016576304862121277,
      "loss": 0.8459,
      "step": 42608
    },
    {
      "epoch": 3.665921018669879,
      "grad_norm": 952.0,
      "learning_rate": 0.0001657429379523747,
      "loss": 1.5524,
      "step": 42609
    },
    {
      "epoch": 3.6660070549772,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016572282826117336,
      "loss": 0.947,
      "step": 42610
    },
    {
      "epoch": 3.666093091284522,
      "grad_norm": 1.25,
      "learning_rate": 0.00016570271954766775,
      "loss": 1.7926,
      "step": 42611
    },
    {
      "epoch": 3.666179127591844,
      "grad_norm": 1.0625,
      "learning_rate": 0.00016568261181191686,
      "loss": 0.912,
      "step": 42612
    },
    {
      "epoch": 3.6662651638991655,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016566250505397907,
      "loss": 0.6742,
      "step": 42613
    },
    {
      "epoch": 3.666351200206487,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001656423992739135,
      "loss": 1.6398,
      "step": 42614
    },
    {
      "epoch": 3.6664372365138087,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016562229447177884,
      "loss": 0.3895,
      "step": 42615
    },
    {
      "epoch": 3.6665232728211308,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001656021906476341,
      "loss": 1.233,
      "step": 42616
    },
    {
      "epoch": 3.666609309128452,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00016558208780153777,
      "loss": 1.3812,
      "step": 42617
    },
    {
      "epoch": 3.666695345435774,
      "grad_norm": 1.109375,
      "learning_rate": 0.00016556198593354877,
      "loss": 1.5337,
      "step": 42618
    },
    {
      "epoch": 3.6667813817430956,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00016554188504372608,
      "loss": 0.7334,
      "step": 42619
    },
    {
      "epoch": 3.6668674180504173,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00016552178513212817,
      "loss": 1.3103,
      "step": 42620
    },
    {
      "epoch": 3.666953454357739,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000165501686198814,
      "loss": 0.5475,
      "step": 42621
    },
    {
      "epoch": 3.6670394906650605,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001654815882438425,
      "loss": 0.6961,
      "step": 42622
    },
    {
      "epoch": 3.6671255269723826,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00016546149126727216,
      "loss": 1.632,
      "step": 42623
    },
    {
      "epoch": 3.6672115632797038,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00016544139526916186,
      "loss": 1.0287,
      "step": 42624
    },
    {
      "epoch": 3.667297599587026,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016542130024957058,
      "loss": 1.4132,
      "step": 42625
    },
    {
      "epoch": 3.6673836358943475,
      "grad_norm": 1.078125,
      "learning_rate": 0.00016540120620855677,
      "loss": 1.4949,
      "step": 42626
    },
    {
      "epoch": 3.667469672201669,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001653811131461793,
      "loss": 0.6857,
      "step": 42627
    },
    {
      "epoch": 3.6675557085089907,
      "grad_norm": 4.0,
      "learning_rate": 0.00016536102106249722,
      "loss": 0.7142,
      "step": 42628
    },
    {
      "epoch": 3.6676417448163123,
      "grad_norm": 14.9375,
      "learning_rate": 0.0001653409299575689,
      "loss": 1.5983,
      "step": 42629
    },
    {
      "epoch": 3.6677277811236344,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001653208398314534,
      "loss": 0.6914,
      "step": 42630
    },
    {
      "epoch": 3.667813817430956,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00016530075068420925,
      "loss": 1.8515,
      "step": 42631
    },
    {
      "epoch": 3.6678998537382776,
      "grad_norm": 4.46875,
      "learning_rate": 0.00016528066251589524,
      "loss": 1.1738,
      "step": 42632
    },
    {
      "epoch": 3.6679858900455993,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001652605753265704,
      "loss": 0.7322,
      "step": 42633
    },
    {
      "epoch": 3.668071926352921,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016524048911629313,
      "loss": 0.9784,
      "step": 42634
    },
    {
      "epoch": 3.6681579626602425,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00016522040388512228,
      "loss": 1.5485,
      "step": 42635
    },
    {
      "epoch": 3.668243998967564,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016520031963311683,
      "loss": 1.3513,
      "step": 42636
    },
    {
      "epoch": 3.668330035274886,
      "grad_norm": 1.3125,
      "learning_rate": 0.00016518023636033513,
      "loss": 2.1542,
      "step": 42637
    },
    {
      "epoch": 3.668416071582208,
      "grad_norm": 1.0625,
      "learning_rate": 0.00016516015406683616,
      "loss": 1.1892,
      "step": 42638
    },
    {
      "epoch": 3.6685021078895295,
      "grad_norm": 1.203125,
      "learning_rate": 0.00016514007275267873,
      "loss": 0.8473,
      "step": 42639
    },
    {
      "epoch": 3.668588144196851,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001651199924179213,
      "loss": 0.7376,
      "step": 42640
    },
    {
      "epoch": 3.6686741805041727,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000165099913062623,
      "loss": 1.2736,
      "step": 42641
    },
    {
      "epoch": 3.6687602168114943,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00016507983468684207,
      "loss": 1.4645,
      "step": 42642
    },
    {
      "epoch": 3.668846253118816,
      "grad_norm": 4.1875,
      "learning_rate": 0.00016505975729063754,
      "loss": 2.0032,
      "step": 42643
    },
    {
      "epoch": 3.668932289426138,
      "grad_norm": 1224.0,
      "learning_rate": 0.00016503968087406824,
      "loss": 1.172,
      "step": 42644
    },
    {
      "epoch": 3.6690183257334597,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001650196054371925,
      "loss": 0.8861,
      "step": 42645
    },
    {
      "epoch": 3.6691043620407813,
      "grad_norm": 313344.0,
      "learning_rate": 0.00016499953098006936,
      "loss": 1.0802,
      "step": 42646
    },
    {
      "epoch": 3.669190398348103,
      "grad_norm": 14.0,
      "learning_rate": 0.00016497945750275757,
      "loss": 1.5784,
      "step": 42647
    },
    {
      "epoch": 3.6692764346554245,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016495938500531547,
      "loss": 1.6787,
      "step": 42648
    },
    {
      "epoch": 3.669362470962746,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001649393134878021,
      "loss": 0.9317,
      "step": 42649
    },
    {
      "epoch": 3.6694485072700678,
      "grad_norm": 3.546875,
      "learning_rate": 0.00016491924295027621,
      "loss": 1.592,
      "step": 42650
    },
    {
      "epoch": 3.66953454357739,
      "grad_norm": 101.5,
      "learning_rate": 0.00016489917339279603,
      "loss": 0.5913,
      "step": 42651
    },
    {
      "epoch": 3.6696205798847115,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001648791048154209,
      "loss": 0.9158,
      "step": 42652
    },
    {
      "epoch": 3.669706616192033,
      "grad_norm": 0.875,
      "learning_rate": 0.00016485903721820904,
      "loss": 0.6001,
      "step": 42653
    },
    {
      "epoch": 3.6697926524993547,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00016483897060121937,
      "loss": 0.7331,
      "step": 42654
    },
    {
      "epoch": 3.6698786888066763,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001648189049645107,
      "loss": 1.072,
      "step": 42655
    },
    {
      "epoch": 3.669964725113998,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00016479884030814128,
      "loss": 1.7721,
      "step": 42656
    },
    {
      "epoch": 3.6700507614213196,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00016477877663217017,
      "loss": 1.6582,
      "step": 42657
    },
    {
      "epoch": 3.6701367977286417,
      "grad_norm": 56.25,
      "learning_rate": 0.00016475871393665603,
      "loss": 1.1057,
      "step": 42658
    },
    {
      "epoch": 3.6702228340359633,
      "grad_norm": 1.0,
      "learning_rate": 0.00016473865222165735,
      "loss": 0.78,
      "step": 42659
    },
    {
      "epoch": 3.670308870343285,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000164718591487233,
      "loss": 0.7076,
      "step": 42660
    },
    {
      "epoch": 3.6703949066506065,
      "grad_norm": 6.28125,
      "learning_rate": 0.00016469853173344147,
      "loss": 0.732,
      "step": 42661
    },
    {
      "epoch": 3.670480942957928,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016467847296034143,
      "loss": 1.0206,
      "step": 42662
    },
    {
      "epoch": 3.6705669792652498,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001646584151679919,
      "loss": 1.6856,
      "step": 42663
    },
    {
      "epoch": 3.6706530155725714,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00016463835835645109,
      "loss": 1.8204,
      "step": 42664
    },
    {
      "epoch": 3.6707390518798935,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016461830252577781,
      "loss": 1.0782,
      "step": 42665
    },
    {
      "epoch": 3.670825088187215,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016459824767603098,
      "loss": 0.907,
      "step": 42666
    },
    {
      "epoch": 3.6709111244945367,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001645781938072689,
      "loss": 0.665,
      "step": 42667
    },
    {
      "epoch": 3.6709971608018583,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016455814091955029,
      "loss": 0.6692,
      "step": 42668
    },
    {
      "epoch": 3.67108319710918,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001645380890129341,
      "loss": 0.836,
      "step": 42669
    },
    {
      "epoch": 3.671169233416502,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016451803808747862,
      "loss": 0.5394,
      "step": 42670
    },
    {
      "epoch": 3.671255269723823,
      "grad_norm": 944.0,
      "learning_rate": 0.00016449798814324273,
      "loss": 0.7699,
      "step": 42671
    },
    {
      "epoch": 3.6713413060311453,
      "grad_norm": 1.75,
      "learning_rate": 0.00016447793918028482,
      "loss": 1.4287,
      "step": 42672
    },
    {
      "epoch": 3.671427342338467,
      "grad_norm": 272.0,
      "learning_rate": 0.00016445789119866374,
      "loss": 0.6764,
      "step": 42673
    },
    {
      "epoch": 3.6715133786457885,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00016443784419843815,
      "loss": 0.71,
      "step": 42674
    },
    {
      "epoch": 3.67159941495311,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016441779817966652,
      "loss": 1.4715,
      "step": 42675
    },
    {
      "epoch": 3.6716854512604318,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001643977531424075,
      "loss": 0.7919,
      "step": 42676
    },
    {
      "epoch": 3.671771487567754,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016437770908671995,
      "loss": 0.7068,
      "step": 42677
    },
    {
      "epoch": 3.671857523875075,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00016435766601266216,
      "loss": 0.9,
      "step": 42678
    },
    {
      "epoch": 3.671943560182397,
      "grad_norm": 3.3125,
      "learning_rate": 0.00016433762392029295,
      "loss": 1.4297,
      "step": 42679
    },
    {
      "epoch": 3.6720295964897187,
      "grad_norm": 1.609375,
      "learning_rate": 0.00016431758280967108,
      "loss": 1.3995,
      "step": 42680
    },
    {
      "epoch": 3.6721156327970403,
      "grad_norm": 1.359375,
      "learning_rate": 0.00016429754268085478,
      "loss": 1.4491,
      "step": 42681
    },
    {
      "epoch": 3.672201669104362,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016427750353390308,
      "loss": 0.8156,
      "step": 42682
    },
    {
      "epoch": 3.6722877054116836,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001642574653688742,
      "loss": 0.8782,
      "step": 42683
    },
    {
      "epoch": 3.6723737417190057,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016423742818582698,
      "loss": 0.4604,
      "step": 42684
    },
    {
      "epoch": 3.672459778026327,
      "grad_norm": 28.75,
      "learning_rate": 0.00016421739198482012,
      "loss": 1.6486,
      "step": 42685
    },
    {
      "epoch": 3.672545814333649,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001641973567659119,
      "loss": 0.9725,
      "step": 42686
    },
    {
      "epoch": 3.6726318506409705,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016417732252916106,
      "loss": 0.5977,
      "step": 42687
    },
    {
      "epoch": 3.672717886948292,
      "grad_norm": 2.171875,
      "learning_rate": 0.00016415728927462631,
      "loss": 1.5868,
      "step": 42688
    },
    {
      "epoch": 3.672803923255614,
      "grad_norm": 1.0,
      "learning_rate": 0.00016413725700236616,
      "loss": 0.8655,
      "step": 42689
    },
    {
      "epoch": 3.6728899595629354,
      "grad_norm": 1.078125,
      "learning_rate": 0.00016411722571243935,
      "loss": 1.1379,
      "step": 42690
    },
    {
      "epoch": 3.6729759958702575,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001640971954049042,
      "loss": 1.4411,
      "step": 42691
    },
    {
      "epoch": 3.673062032177579,
      "grad_norm": 1.140625,
      "learning_rate": 0.00016407716607981943,
      "loss": 0.7345,
      "step": 42692
    },
    {
      "epoch": 3.6731480684849007,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016405713773724373,
      "loss": 0.9326,
      "step": 42693
    },
    {
      "epoch": 3.6732341047922223,
      "grad_norm": 2.03125,
      "learning_rate": 0.0001640371103772354,
      "loss": 1.3045,
      "step": 42694
    },
    {
      "epoch": 3.673320141099544,
      "grad_norm": 42.75,
      "learning_rate": 0.00016401708399985321,
      "loss": 1.37,
      "step": 42695
    },
    {
      "epoch": 3.6734061774068656,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001639970586051559,
      "loss": 0.7494,
      "step": 42696
    },
    {
      "epoch": 3.673492213714187,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016397703419320165,
      "loss": 0.6344,
      "step": 42697
    },
    {
      "epoch": 3.6735782500215093,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001639570107640492,
      "loss": 1.3569,
      "step": 42698
    },
    {
      "epoch": 3.673664286328831,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016393698831775733,
      "loss": 1.0928,
      "step": 42699
    },
    {
      "epoch": 3.6737503226361525,
      "grad_norm": 5472.0,
      "learning_rate": 0.00016391696685438424,
      "loss": 1.2823,
      "step": 42700
    },
    {
      "epoch": 3.673836358943474,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016389694637398877,
      "loss": 0.8399,
      "step": 42701
    },
    {
      "epoch": 3.673922395250796,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016387692687662924,
      "loss": 0.6725,
      "step": 42702
    },
    {
      "epoch": 3.6740084315581174,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001638569083623644,
      "loss": 0.8382,
      "step": 42703
    },
    {
      "epoch": 3.674094467865439,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001638368908312528,
      "loss": 1.3783,
      "step": 42704
    },
    {
      "epoch": 3.674180504172761,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016381687428335272,
      "loss": 0.8566,
      "step": 42705
    },
    {
      "epoch": 3.6742665404800827,
      "grad_norm": 1.21875,
      "learning_rate": 0.00016379685871872296,
      "loss": 1.7616,
      "step": 42706
    },
    {
      "epoch": 3.6743525767874043,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00016377684413742215,
      "loss": 1.7285,
      "step": 42707
    },
    {
      "epoch": 3.674438613094726,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00016375683053950857,
      "loss": 1.9048,
      "step": 42708
    },
    {
      "epoch": 3.6745246494020476,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001637368179250408,
      "loss": 1.4988,
      "step": 42709
    },
    {
      "epoch": 3.674610685709369,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001637168062940776,
      "loss": 1.4365,
      "step": 42710
    },
    {
      "epoch": 3.674696722016691,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00016369679564667722,
      "loss": 1.0112,
      "step": 42711
    },
    {
      "epoch": 3.674782758324013,
      "grad_norm": 12.9375,
      "learning_rate": 0.00016367678598289842,
      "loss": 1.0984,
      "step": 42712
    },
    {
      "epoch": 3.6748687946313345,
      "grad_norm": 1.671875,
      "learning_rate": 0.0001636567773027995,
      "loss": 1.4353,
      "step": 42713
    },
    {
      "epoch": 3.674954830938656,
      "grad_norm": 20.625,
      "learning_rate": 0.00016363676960643903,
      "loss": 1.2671,
      "step": 42714
    },
    {
      "epoch": 3.675040867245978,
      "grad_norm": 1.46875,
      "learning_rate": 0.00016361676289387578,
      "loss": 1.1207,
      "step": 42715
    },
    {
      "epoch": 3.6751269035532994,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00016359675716516792,
      "loss": 1.832,
      "step": 42716
    },
    {
      "epoch": 3.675212939860621,
      "grad_norm": 60.75,
      "learning_rate": 0.00016357675242037413,
      "loss": 0.6828,
      "step": 42717
    },
    {
      "epoch": 3.6752989761679427,
      "grad_norm": 22.375,
      "learning_rate": 0.000163556748659553,
      "loss": 1.2553,
      "step": 42718
    },
    {
      "epoch": 3.6753850124752647,
      "grad_norm": 1.46875,
      "learning_rate": 0.00016353674588276285,
      "loss": 0.4628,
      "step": 42719
    },
    {
      "epoch": 3.6754710487825863,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001635167440900623,
      "loss": 0.8387,
      "step": 42720
    },
    {
      "epoch": 3.675557085089908,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016349674328150992,
      "loss": 0.5636,
      "step": 42721
    },
    {
      "epoch": 3.6756431213972296,
      "grad_norm": 1.390625,
      "learning_rate": 0.00016347674345716401,
      "loss": 1.0185,
      "step": 42722
    },
    {
      "epoch": 3.675729157704551,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016345674461708316,
      "loss": 0.8023,
      "step": 42723
    },
    {
      "epoch": 3.675815194011873,
      "grad_norm": 54.75,
      "learning_rate": 0.00016343674676132602,
      "loss": 1.2433,
      "step": 42724
    },
    {
      "epoch": 3.6759012303191945,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016341674988995064,
      "loss": 1.9829,
      "step": 42725
    },
    {
      "epoch": 3.6759872666265165,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016339675400301612,
      "loss": 0.6222,
      "step": 42726
    },
    {
      "epoch": 3.676073302933838,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016337675910058048,
      "loss": 1.3255,
      "step": 42727
    },
    {
      "epoch": 3.67615933924116,
      "grad_norm": 5.9375,
      "learning_rate": 0.00016335676518270255,
      "loss": 1.2608,
      "step": 42728
    },
    {
      "epoch": 3.6762453755484814,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016333677224944038,
      "loss": 0.8895,
      "step": 42729
    },
    {
      "epoch": 3.676331411855803,
      "grad_norm": 134.0,
      "learning_rate": 0.00016331678030085268,
      "loss": 1.1841,
      "step": 42730
    },
    {
      "epoch": 3.676417448163125,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001632967893369981,
      "loss": 1.4339,
      "step": 42731
    },
    {
      "epoch": 3.6765034844704463,
      "grad_norm": 6.375,
      "learning_rate": 0.00016327679935793476,
      "loss": 1.1882,
      "step": 42732
    },
    {
      "epoch": 3.6765895207777683,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016325681036372125,
      "loss": 0.8527,
      "step": 42733
    },
    {
      "epoch": 3.67667555708509,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001632368223544163,
      "loss": 1.6798,
      "step": 42734
    },
    {
      "epoch": 3.6767615933924116,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016321683533007797,
      "loss": 0.8054,
      "step": 42735
    },
    {
      "epoch": 3.676847629699733,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016319684929076489,
      "loss": 0.7858,
      "step": 42736
    },
    {
      "epoch": 3.676933666007055,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016317686423653567,
      "loss": 1.1169,
      "step": 42737
    },
    {
      "epoch": 3.677019702314377,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00016315688016744844,
      "loss": 1.003,
      "step": 42738
    },
    {
      "epoch": 3.677105738621698,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001631368970835619,
      "loss": 0.6669,
      "step": 42739
    },
    {
      "epoch": 3.67719177492902,
      "grad_norm": 0.984375,
      "learning_rate": 0.00016311691498493454,
      "loss": 0.8314,
      "step": 42740
    },
    {
      "epoch": 3.677277811236342,
      "grad_norm": 12608.0,
      "learning_rate": 0.0001630969338716245,
      "loss": 1.335,
      "step": 42741
    },
    {
      "epoch": 3.6773638475436634,
      "grad_norm": 1.328125,
      "learning_rate": 0.00016307695374369059,
      "loss": 1.5783,
      "step": 42742
    },
    {
      "epoch": 3.677449883850985,
      "grad_norm": 4.4375,
      "learning_rate": 0.00016305697460119084,
      "loss": 0.7993,
      "step": 42743
    },
    {
      "epoch": 3.6775359201583067,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00016303699644418395,
      "loss": 1.8326,
      "step": 42744
    },
    {
      "epoch": 3.6776219564656287,
      "grad_norm": 3.015625,
      "learning_rate": 0.0001630170192727285,
      "loss": 0.8735,
      "step": 42745
    },
    {
      "epoch": 3.67770799277295,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016299704308688257,
      "loss": 0.632,
      "step": 42746
    },
    {
      "epoch": 3.677794029080272,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001629770678867047,
      "loss": 0.4952,
      "step": 42747
    },
    {
      "epoch": 3.6778800653875936,
      "grad_norm": 4.8125,
      "learning_rate": 0.00016295709367225353,
      "loss": 1.4515,
      "step": 42748
    },
    {
      "epoch": 3.6779661016949152,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016293712044358717,
      "loss": 0.8364,
      "step": 42749
    },
    {
      "epoch": 3.678052138002237,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016291714820076413,
      "loss": 0.8943,
      "step": 42750
    },
    {
      "epoch": 3.6781381743095585,
      "grad_norm": 50.75,
      "learning_rate": 0.0001628971769438431,
      "loss": 1.3602,
      "step": 42751
    },
    {
      "epoch": 3.6782242106168805,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001628772066728821,
      "loss": 0.7501,
      "step": 42752
    },
    {
      "epoch": 3.678310246924202,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001628572373879398,
      "loss": 1.3705,
      "step": 42753
    },
    {
      "epoch": 3.678396283231524,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00016283726908907436,
      "loss": 0.8397,
      "step": 42754
    },
    {
      "epoch": 3.6784823195388454,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016281730177634442,
      "loss": 1.1091,
      "step": 42755
    },
    {
      "epoch": 3.678568355846167,
      "grad_norm": 1.203125,
      "learning_rate": 0.00016279733544980835,
      "loss": 1.5948,
      "step": 42756
    },
    {
      "epoch": 3.6786543921534887,
      "grad_norm": 1.359375,
      "learning_rate": 0.00016277737010952437,
      "loss": 2.042,
      "step": 42757
    },
    {
      "epoch": 3.6787404284608103,
      "grad_norm": 2.40625,
      "learning_rate": 0.00016275740575555104,
      "loss": 1.1967,
      "step": 42758
    },
    {
      "epoch": 3.6788264647681324,
      "grad_norm": 0.875,
      "learning_rate": 0.0001627374423879468,
      "loss": 1.0765,
      "step": 42759
    },
    {
      "epoch": 3.678912501075454,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016271748000676984,
      "loss": 0.5807,
      "step": 42760
    },
    {
      "epoch": 3.6789985373827756,
      "grad_norm": 1.125,
      "learning_rate": 0.0001626975186120786,
      "loss": 0.8694,
      "step": 42761
    },
    {
      "epoch": 3.6790845736900972,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001626775582039316,
      "loss": 0.7036,
      "step": 42762
    },
    {
      "epoch": 3.679170609997419,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00016265759878238706,
      "loss": 1.2896,
      "step": 42763
    },
    {
      "epoch": 3.6792566463047405,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001626376403475036,
      "loss": 1.4089,
      "step": 42764
    },
    {
      "epoch": 3.679342682612062,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001626176828993393,
      "loss": 0.8084,
      "step": 42765
    },
    {
      "epoch": 3.679428718919384,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00016259772643795263,
      "loss": 1.2871,
      "step": 42766
    },
    {
      "epoch": 3.679514755226706,
      "grad_norm": 29.625,
      "learning_rate": 0.0001625777709634022,
      "loss": 0.6928,
      "step": 42767
    },
    {
      "epoch": 3.6796007915340274,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00016255781647574598,
      "loss": 1.0097,
      "step": 42768
    },
    {
      "epoch": 3.679686827841349,
      "grad_norm": 159.0,
      "learning_rate": 0.00016253786297504264,
      "loss": 0.7713,
      "step": 42769
    },
    {
      "epoch": 3.6797728641486707,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0001625179104613503,
      "loss": 1.2331,
      "step": 42770
    },
    {
      "epoch": 3.6798589004559923,
      "grad_norm": 1.109375,
      "learning_rate": 0.00016249795893472741,
      "loss": 0.8626,
      "step": 42771
    },
    {
      "epoch": 3.679944936763314,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00016247800839523253,
      "loss": 1.0891,
      "step": 42772
    },
    {
      "epoch": 3.680030973070636,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00016245805884292365,
      "loss": 1.3348,
      "step": 42773
    },
    {
      "epoch": 3.6801170093779576,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016243811027785932,
      "loss": 0.7926,
      "step": 42774
    },
    {
      "epoch": 3.6802030456852792,
      "grad_norm": 1.453125,
      "learning_rate": 0.00016241816270009803,
      "loss": 1.7927,
      "step": 42775
    },
    {
      "epoch": 3.680289081992601,
      "grad_norm": 2.203125,
      "learning_rate": 0.0001623982161096978,
      "loss": 1.1113,
      "step": 42776
    },
    {
      "epoch": 3.6803751182999225,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016237827050671712,
      "loss": 1.168,
      "step": 42777
    },
    {
      "epoch": 3.680461154607244,
      "grad_norm": 1.265625,
      "learning_rate": 0.00016235832589121447,
      "loss": 0.9153,
      "step": 42778
    },
    {
      "epoch": 3.6805471909145657,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00016233838226324787,
      "loss": 0.606,
      "step": 42779
    },
    {
      "epoch": 3.680633227221888,
      "grad_norm": 206.0,
      "learning_rate": 0.0001623184396228759,
      "loss": 1.1339,
      "step": 42780
    },
    {
      "epoch": 3.6807192635292094,
      "grad_norm": 1.40625,
      "learning_rate": 0.00016229849797015695,
      "loss": 1.8095,
      "step": 42781
    },
    {
      "epoch": 3.680805299836531,
      "grad_norm": 45.75,
      "learning_rate": 0.00016227855730514902,
      "loss": 0.8377,
      "step": 42782
    },
    {
      "epoch": 3.6808913361438527,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016225861762791084,
      "loss": 1.0822,
      "step": 42783
    },
    {
      "epoch": 3.6809773724511743,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001622386789385003,
      "loss": 1.5306,
      "step": 42784
    },
    {
      "epoch": 3.6810634087584964,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016221874123697598,
      "loss": 0.5169,
      "step": 42785
    },
    {
      "epoch": 3.6811494450658175,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00016219880452339632,
      "loss": 0.48,
      "step": 42786
    },
    {
      "epoch": 3.6812354813731396,
      "grad_norm": 2.875,
      "learning_rate": 0.00016217886879781924,
      "loss": 1.128,
      "step": 42787
    },
    {
      "epoch": 3.6813215176804612,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0001621589340603033,
      "loss": 1.1183,
      "step": 42788
    },
    {
      "epoch": 3.681407553987783,
      "grad_norm": 1.234375,
      "learning_rate": 0.00016213900031090695,
      "loss": 0.8792,
      "step": 42789
    },
    {
      "epoch": 3.6814935902951045,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00016211906754968813,
      "loss": 1.4244,
      "step": 42790
    },
    {
      "epoch": 3.681579626602426,
      "grad_norm": 52.75,
      "learning_rate": 0.00016209913577670533,
      "loss": 1.3658,
      "step": 42791
    },
    {
      "epoch": 3.681665662909748,
      "grad_norm": 13.375,
      "learning_rate": 0.000162079204992017,
      "loss": 0.9125,
      "step": 42792
    },
    {
      "epoch": 3.6817516992170694,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001620592751956811,
      "loss": 0.4025,
      "step": 42793
    },
    {
      "epoch": 3.6818377355243914,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016203934638775626,
      "loss": 0.9611,
      "step": 42794
    },
    {
      "epoch": 3.681923771831713,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00016201941856830044,
      "loss": 0.7559,
      "step": 42795
    },
    {
      "epoch": 3.6820098081390347,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001619994917373721,
      "loss": 1.0934,
      "step": 42796
    },
    {
      "epoch": 3.6820958444463563,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00016197956589502965,
      "loss": 1.4445,
      "step": 42797
    },
    {
      "epoch": 3.682181880753678,
      "grad_norm": 13.25,
      "learning_rate": 0.0001619596410413311,
      "loss": 0.9784,
      "step": 42798
    },
    {
      "epoch": 3.682267917061,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016193971717633483,
      "loss": 1.2186,
      "step": 42799
    },
    {
      "epoch": 3.682353953368321,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016191979430009912,
      "loss": 0.6477,
      "step": 42800
    },
    {
      "epoch": 3.6824399896756432,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016189987241268234,
      "loss": 0.6307,
      "step": 42801
    },
    {
      "epoch": 3.682526025982965,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00016187995151414276,
      "loss": 0.5922,
      "step": 42802
    },
    {
      "epoch": 3.6826120622902865,
      "grad_norm": 11.0,
      "learning_rate": 0.00016186003160453838,
      "loss": 1.1505,
      "step": 42803
    },
    {
      "epoch": 3.682698098597608,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016184011268392774,
      "loss": 1.058,
      "step": 42804
    },
    {
      "epoch": 3.6827841349049297,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001618201947523691,
      "loss": 1.2836,
      "step": 42805
    },
    {
      "epoch": 3.682870171212252,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00016180027780992047,
      "loss": 0.7678,
      "step": 42806
    },
    {
      "epoch": 3.6829562075195734,
      "grad_norm": 248.0,
      "learning_rate": 0.00016178036185664018,
      "loss": 0.6255,
      "step": 42807
    },
    {
      "epoch": 3.683042243826895,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001617604468925868,
      "loss": 1.6033,
      "step": 42808
    },
    {
      "epoch": 3.6831282801342167,
      "grad_norm": 1.46875,
      "learning_rate": 0.00016174053291781814,
      "loss": 1.2799,
      "step": 42809
    },
    {
      "epoch": 3.6832143164415383,
      "grad_norm": 9.5,
      "learning_rate": 0.00016172061993239268,
      "loss": 1.2126,
      "step": 42810
    },
    {
      "epoch": 3.68330035274886,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001617007079363687,
      "loss": 1.1618,
      "step": 42811
    },
    {
      "epoch": 3.6833863890561815,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016168079692980425,
      "loss": 1.3882,
      "step": 42812
    },
    {
      "epoch": 3.6834724253635036,
      "grad_norm": 1.484375,
      "learning_rate": 0.00016166088691275778,
      "loss": 1.7033,
      "step": 42813
    },
    {
      "epoch": 3.6835584616708252,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016164097788528732,
      "loss": 1.1557,
      "step": 42814
    },
    {
      "epoch": 3.683644497978147,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00016162106984745112,
      "loss": 0.8064,
      "step": 42815
    },
    {
      "epoch": 3.6837305342854685,
      "grad_norm": 1.515625,
      "learning_rate": 0.0001616011627993077,
      "loss": 0.9508,
      "step": 42816
    },
    {
      "epoch": 3.68381657059279,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016158125674091485,
      "loss": 0.8516,
      "step": 42817
    },
    {
      "epoch": 3.6839026069001117,
      "grad_norm": 1.375,
      "learning_rate": 0.00016156135167233105,
      "loss": 0.8603,
      "step": 42818
    },
    {
      "epoch": 3.6839886432074334,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016154144759361461,
      "loss": 0.9128,
      "step": 42819
    },
    {
      "epoch": 3.6840746795147554,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016152154450482342,
      "loss": 1.5738,
      "step": 42820
    },
    {
      "epoch": 3.684160715822077,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00016150164240601595,
      "loss": 1.3402,
      "step": 42821
    },
    {
      "epoch": 3.6842467521293987,
      "grad_norm": 2944.0,
      "learning_rate": 0.00016148174129725047,
      "loss": 0.9512,
      "step": 42822
    },
    {
      "epoch": 3.6843327884367203,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016146184117858485,
      "loss": 1.0168,
      "step": 42823
    },
    {
      "epoch": 3.684418824744042,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00016144194205007772,
      "loss": 2.1408,
      "step": 42824
    },
    {
      "epoch": 3.6845048610513635,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001614220439117869,
      "loss": 1.7112,
      "step": 42825
    },
    {
      "epoch": 3.684590897358685,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00016140214676377073,
      "loss": 0.6885,
      "step": 42826
    },
    {
      "epoch": 3.6846769336660072,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00016138225060608762,
      "loss": 1.2584,
      "step": 42827
    },
    {
      "epoch": 3.684762969973329,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001613623554387954,
      "loss": 0.6271,
      "step": 42828
    },
    {
      "epoch": 3.6848490062806505,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001613424612619524,
      "loss": 1.5081,
      "step": 42829
    },
    {
      "epoch": 3.684935042587972,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00016132256807561702,
      "loss": 1.524,
      "step": 42830
    },
    {
      "epoch": 3.6850210788952937,
      "grad_norm": 4.5625,
      "learning_rate": 0.0001613026758798471,
      "loss": 1.086,
      "step": 42831
    },
    {
      "epoch": 3.6851071152026154,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00016128278467470098,
      "loss": 1.1334,
      "step": 42832
    },
    {
      "epoch": 3.685193151509937,
      "grad_norm": 1.125,
      "learning_rate": 0.00016126289446023696,
      "loss": 1.0034,
      "step": 42833
    },
    {
      "epoch": 3.685279187817259,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00016124300523651298,
      "loss": 1.137,
      "step": 42834
    },
    {
      "epoch": 3.6853652241245807,
      "grad_norm": 1.0625,
      "learning_rate": 0.00016122311700358732,
      "loss": 0.6314,
      "step": 42835
    },
    {
      "epoch": 3.6854512604319023,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00016120322976151814,
      "loss": 1.312,
      "step": 42836
    },
    {
      "epoch": 3.685537296739224,
      "grad_norm": 0.984375,
      "learning_rate": 0.00016118334351036362,
      "loss": 0.9333,
      "step": 42837
    },
    {
      "epoch": 3.6856233330465455,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00016116345825018208,
      "loss": 1.1586,
      "step": 42838
    },
    {
      "epoch": 3.685709369353867,
      "grad_norm": 1.3125,
      "learning_rate": 0.00016114357398103136,
      "loss": 1.4432,
      "step": 42839
    },
    {
      "epoch": 3.685795405661189,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00016112369070296994,
      "loss": 1.3941,
      "step": 42840
    },
    {
      "epoch": 3.685881441968511,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001611038084160557,
      "loss": 0.8611,
      "step": 42841
    },
    {
      "epoch": 3.6859674782758325,
      "grad_norm": 11.5,
      "learning_rate": 0.00016108392712034687,
      "loss": 1.3564,
      "step": 42842
    },
    {
      "epoch": 3.686053514583154,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001610640468159018,
      "loss": 1.3702,
      "step": 42843
    },
    {
      "epoch": 3.6861395508904757,
      "grad_norm": 11.6875,
      "learning_rate": 0.00016104416750277833,
      "loss": 1.2696,
      "step": 42844
    },
    {
      "epoch": 3.6862255871977974,
      "grad_norm": 1.4375,
      "learning_rate": 0.00016102428918103473,
      "loss": 1.1891,
      "step": 42845
    },
    {
      "epoch": 3.6863116235051194,
      "grad_norm": 5.59375,
      "learning_rate": 0.00016100441185072935,
      "loss": 1.9009,
      "step": 42846
    },
    {
      "epoch": 3.6863976598124406,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00016098453551191994,
      "loss": 1.0795,
      "step": 42847
    },
    {
      "epoch": 3.6864836961197627,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001609646601646648,
      "loss": 0.6944,
      "step": 42848
    },
    {
      "epoch": 3.6865697324270843,
      "grad_norm": 2.125,
      "learning_rate": 0.00016094478580902234,
      "loss": 1.409,
      "step": 42849
    },
    {
      "epoch": 3.686655768734406,
      "grad_norm": 22.125,
      "learning_rate": 0.0001609249124450502,
      "loss": 1.3958,
      "step": 42850
    },
    {
      "epoch": 3.6867418050417275,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001609050400728068,
      "loss": 0.9187,
      "step": 42851
    },
    {
      "epoch": 3.686827841349049,
      "grad_norm": 1.6875,
      "learning_rate": 0.0001608851686923503,
      "loss": 0.7725,
      "step": 42852
    },
    {
      "epoch": 3.6869138776563712,
      "grad_norm": 1.15625,
      "learning_rate": 0.00016086529830373864,
      "loss": 1.0843,
      "step": 42853
    },
    {
      "epoch": 3.6869999139636924,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00016084542890703013,
      "loss": 1.0884,
      "step": 42854
    },
    {
      "epoch": 3.6870859502710145,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00016082556050228265,
      "loss": 1.2993,
      "step": 42855
    },
    {
      "epoch": 3.687171986578336,
      "grad_norm": 1.390625,
      "learning_rate": 0.00016080569308955445,
      "loss": 1.5252,
      "step": 42856
    },
    {
      "epoch": 3.6872580228856577,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016078582666890378,
      "loss": 0.6708,
      "step": 42857
    },
    {
      "epoch": 3.6873440591929794,
      "grad_norm": 1.65625,
      "learning_rate": 0.0001607659612403884,
      "loss": 1.24,
      "step": 42858
    },
    {
      "epoch": 3.687430095500301,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00016074609680406666,
      "loss": 0.9103,
      "step": 42859
    },
    {
      "epoch": 3.687516131807623,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001607262333599967,
      "loss": 0.7863,
      "step": 42860
    },
    {
      "epoch": 3.6876021681149442,
      "grad_norm": 1.328125,
      "learning_rate": 0.00016070637090823643,
      "loss": 1.2499,
      "step": 42861
    },
    {
      "epoch": 3.6876882044222663,
      "grad_norm": 1.2421875,
      "learning_rate": 0.000160686509448844,
      "loss": 1.2076,
      "step": 42862
    },
    {
      "epoch": 3.687774240729588,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016066664898187766,
      "loss": 1.1457,
      "step": 42863
    },
    {
      "epoch": 3.6878602770369096,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016064678950739524,
      "loss": 0.8751,
      "step": 42864
    },
    {
      "epoch": 3.687946313344231,
      "grad_norm": 1.8984375,
      "learning_rate": 0.00016062693102545517,
      "loss": 1.187,
      "step": 42865
    },
    {
      "epoch": 3.688032349651553,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00016060707353611513,
      "loss": 0.9406,
      "step": 42866
    },
    {
      "epoch": 3.688118385958875,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016058721703943342,
      "loss": 1.6038,
      "step": 42867
    },
    {
      "epoch": 3.6882044222661965,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001605673615354682,
      "loss": 1.1041,
      "step": 42868
    },
    {
      "epoch": 3.688290458573518,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016054750702427728,
      "loss": 1.3928,
      "step": 42869
    },
    {
      "epoch": 3.6883764948808397,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00016052765350591886,
      "loss": 1.1474,
      "step": 42870
    },
    {
      "epoch": 3.6884625311881614,
      "grad_norm": 1.484375,
      "learning_rate": 0.00016050780098045126,
      "loss": 0.9046,
      "step": 42871
    },
    {
      "epoch": 3.688548567495483,
      "grad_norm": 22.125,
      "learning_rate": 0.00016048794944793193,
      "loss": 1.1588,
      "step": 42872
    },
    {
      "epoch": 3.6886346038028046,
      "grad_norm": 4.65625,
      "learning_rate": 0.00016046809890841969,
      "loss": 0.8157,
      "step": 42873
    },
    {
      "epoch": 3.6887206401101267,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000160448249361972,
      "loss": 1.6445,
      "step": 42874
    },
    {
      "epoch": 3.6888066764174483,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00016042840080864718,
      "loss": 1.17,
      "step": 42875
    },
    {
      "epoch": 3.68889271272477,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001604085532485034,
      "loss": 0.7254,
      "step": 42876
    },
    {
      "epoch": 3.6889787490320916,
      "grad_norm": 184.0,
      "learning_rate": 0.00016038870668159838,
      "loss": 1.2203,
      "step": 42877
    },
    {
      "epoch": 3.689064785339413,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00016036886110799037,
      "loss": 1.6165,
      "step": 42878
    },
    {
      "epoch": 3.689150821646735,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016034901652773754,
      "loss": 1.526,
      "step": 42879
    },
    {
      "epoch": 3.6892368579540564,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001603291729408976,
      "loss": 1.2447,
      "step": 42880
    },
    {
      "epoch": 3.6893228942613785,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016030933034752898,
      "loss": 1.2841,
      "step": 42881
    },
    {
      "epoch": 3.6894089305687,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00016028948874768928,
      "loss": 1.3723,
      "step": 42882
    },
    {
      "epoch": 3.6894949668760217,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001602696481414368,
      "loss": 1.8069,
      "step": 42883
    },
    {
      "epoch": 3.6895810031833434,
      "grad_norm": 10.9375,
      "learning_rate": 0.00016024980852882964,
      "loss": 1.1875,
      "step": 42884
    },
    {
      "epoch": 3.689667039490665,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001602299699099256,
      "loss": 1.0102,
      "step": 42885
    },
    {
      "epoch": 3.6897530757979866,
      "grad_norm": 0.75,
      "learning_rate": 0.00016021013228478275,
      "loss": 0.9497,
      "step": 42886
    },
    {
      "epoch": 3.6898391121053082,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001601902956534594,
      "loss": 1.2523,
      "step": 42887
    },
    {
      "epoch": 3.6899251484126303,
      "grad_norm": 1.109375,
      "learning_rate": 0.00016017046001601316,
      "loss": 1.0557,
      "step": 42888
    },
    {
      "epoch": 3.690011184719952,
      "grad_norm": 1.796875,
      "learning_rate": 0.00016015062537250225,
      "loss": 0.7134,
      "step": 42889
    },
    {
      "epoch": 3.6900972210272736,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016013079172298483,
      "loss": 0.9431,
      "step": 42890
    },
    {
      "epoch": 3.690183257334595,
      "grad_norm": 1.859375,
      "learning_rate": 0.00016011095906751855,
      "loss": 1.2123,
      "step": 42891
    },
    {
      "epoch": 3.690269293641917,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001600911274061616,
      "loss": 0.5887,
      "step": 42892
    },
    {
      "epoch": 3.6903553299492384,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001600712967389722,
      "loss": 2.048,
      "step": 42893
    },
    {
      "epoch": 3.69044136625656,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016005146706600797,
      "loss": 0.8183,
      "step": 42894
    },
    {
      "epoch": 3.690527402563882,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001600316383873272,
      "loss": 0.893,
      "step": 42895
    },
    {
      "epoch": 3.6906134388712037,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016001181070298765,
      "loss": 0.9327,
      "step": 42896
    },
    {
      "epoch": 3.6906994751785254,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00015999198401304743,
      "loss": 0.8947,
      "step": 42897
    },
    {
      "epoch": 3.690785511485847,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015997215831756468,
      "loss": 1.0517,
      "step": 42898
    },
    {
      "epoch": 3.6908715477931686,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015995233361659712,
      "loss": 0.688,
      "step": 42899
    },
    {
      "epoch": 3.6909575841004902,
      "grad_norm": 6.9375,
      "learning_rate": 0.0001599325099102028,
      "loss": 0.8115,
      "step": 42900
    },
    {
      "epoch": 3.691043620407812,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00015991268719843987,
      "loss": 1.456,
      "step": 42901
    },
    {
      "epoch": 3.691129656715134,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015989286548136612,
      "loss": 1.6574,
      "step": 42902
    },
    {
      "epoch": 3.6912156930224556,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001598730447590395,
      "loss": 1.0277,
      "step": 42903
    },
    {
      "epoch": 3.691301729329777,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00015985322503151827,
      "loss": 0.7765,
      "step": 42904
    },
    {
      "epoch": 3.691387765637099,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015983340629886007,
      "loss": 1.1185,
      "step": 42905
    },
    {
      "epoch": 3.6914738019444204,
      "grad_norm": 1.21875,
      "learning_rate": 0.00015981358856112315,
      "loss": 1.2183,
      "step": 42906
    },
    {
      "epoch": 3.6915598382517425,
      "grad_norm": 2.34375,
      "learning_rate": 0.00015979377181836514,
      "loss": 1.0624,
      "step": 42907
    },
    {
      "epoch": 3.6916458745590637,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00015977395607064417,
      "loss": 0.8719,
      "step": 42908
    },
    {
      "epoch": 3.6917319108663857,
      "grad_norm": 1.390625,
      "learning_rate": 0.00015975414131801823,
      "loss": 1.2987,
      "step": 42909
    },
    {
      "epoch": 3.6918179471737074,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00015973432756054523,
      "loss": 0.9139,
      "step": 42910
    },
    {
      "epoch": 3.691903983481029,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00015971451479828336,
      "loss": 1.1508,
      "step": 42911
    },
    {
      "epoch": 3.6919900197883506,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015969470303129013,
      "loss": 1.4548,
      "step": 42912
    },
    {
      "epoch": 3.6920760560956722,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015967489225962372,
      "loss": 0.8315,
      "step": 42913
    },
    {
      "epoch": 3.6921620924029943,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015965508248334226,
      "loss": 0.5085,
      "step": 42914
    },
    {
      "epoch": 3.6922481287103155,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001596352737025033,
      "loss": 1.1623,
      "step": 42915
    },
    {
      "epoch": 3.6923341650176376,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015961546591716498,
      "loss": 0.9377,
      "step": 42916
    },
    {
      "epoch": 3.692420201324959,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00015959565912738533,
      "loss": 0.6618,
      "step": 42917
    },
    {
      "epoch": 3.692506237632281,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015957585333322205,
      "loss": 1.5052,
      "step": 42918
    },
    {
      "epoch": 3.6925922739396024,
      "grad_norm": 1.140625,
      "learning_rate": 0.00015955604853473321,
      "loss": 1.6623,
      "step": 42919
    },
    {
      "epoch": 3.692678310246924,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015953624473197687,
      "loss": 0.664,
      "step": 42920
    },
    {
      "epoch": 3.692764346554246,
      "grad_norm": 1.578125,
      "learning_rate": 0.00015951644192501065,
      "loss": 1.0719,
      "step": 42921
    },
    {
      "epoch": 3.6928503828615673,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001594966401138927,
      "loss": 1.1091,
      "step": 42922
    },
    {
      "epoch": 3.6929364191688894,
      "grad_norm": 1376.0,
      "learning_rate": 0.0001594768392986808,
      "loss": 1.4644,
      "step": 42923
    },
    {
      "epoch": 3.693022455476211,
      "grad_norm": 1.109375,
      "learning_rate": 0.00015945703947943285,
      "loss": 0.9361,
      "step": 42924
    },
    {
      "epoch": 3.6931084917835326,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015943724065620697,
      "loss": 1.5209,
      "step": 42925
    },
    {
      "epoch": 3.6931945280908542,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001594174428290608,
      "loss": 1.2243,
      "step": 42926
    },
    {
      "epoch": 3.693280564398176,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001593976459980524,
      "loss": 1.6556,
      "step": 42927
    },
    {
      "epoch": 3.693366600705498,
      "grad_norm": 0.984375,
      "learning_rate": 0.00015937785016323974,
      "loss": 0.5532,
      "step": 42928
    },
    {
      "epoch": 3.6934526370128196,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001593580553246805,
      "loss": 1.1056,
      "step": 42929
    },
    {
      "epoch": 3.693538673320141,
      "grad_norm": 5.25,
      "learning_rate": 0.00015933826148243268,
      "loss": 0.6794,
      "step": 42930
    },
    {
      "epoch": 3.693624709627463,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00015931846863655435,
      "loss": 1.3622,
      "step": 42931
    },
    {
      "epoch": 3.6937107459347844,
      "grad_norm": 0.984375,
      "learning_rate": 0.00015929867678710308,
      "loss": 1.1776,
      "step": 42932
    },
    {
      "epoch": 3.693796782242106,
      "grad_norm": 1.5625,
      "learning_rate": 0.00015927888593413693,
      "loss": 1.592,
      "step": 42933
    },
    {
      "epoch": 3.6938828185494277,
      "grad_norm": 1.515625,
      "learning_rate": 0.00015925909607771393,
      "loss": 1.0796,
      "step": 42934
    },
    {
      "epoch": 3.6939688548567498,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015923930721789165,
      "loss": 1.5043,
      "step": 42935
    },
    {
      "epoch": 3.6940548911640714,
      "grad_norm": 2.90625,
      "learning_rate": 0.00015921951935472827,
      "loss": 0.9562,
      "step": 42936
    },
    {
      "epoch": 3.694140927471393,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00015919973248828135,
      "loss": 1.1478,
      "step": 42937
    },
    {
      "epoch": 3.6942269637787146,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00015917994661860895,
      "loss": 1.6469,
      "step": 42938
    },
    {
      "epoch": 3.6943130000860362,
      "grad_norm": 1.078125,
      "learning_rate": 0.00015916016174576908,
      "loss": 0.7903,
      "step": 42939
    },
    {
      "epoch": 3.694399036393358,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00015914037786981928,
      "loss": 1.2628,
      "step": 42940
    },
    {
      "epoch": 3.6944850727006795,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00015912059499081756,
      "loss": 1.6978,
      "step": 42941
    },
    {
      "epoch": 3.6945711090080016,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015910081310882196,
      "loss": 1.0872,
      "step": 42942
    },
    {
      "epoch": 3.694657145315323,
      "grad_norm": 1.984375,
      "learning_rate": 0.00015908103222389008,
      "loss": 1.0346,
      "step": 42943
    },
    {
      "epoch": 3.694743181622645,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00015906125233607982,
      "loss": 1.4276,
      "step": 42944
    },
    {
      "epoch": 3.6948292179299664,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00015904147344544928,
      "loss": 1.1789,
      "step": 42945
    },
    {
      "epoch": 3.694915254237288,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015902169555205575,
      "loss": 0.3609,
      "step": 42946
    },
    {
      "epoch": 3.6950012905446097,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015900191865595781,
      "loss": 1.026,
      "step": 42947
    },
    {
      "epoch": 3.6950873268519313,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001589821427572128,
      "loss": 0.8308,
      "step": 42948
    },
    {
      "epoch": 3.6951733631592534,
      "grad_norm": 1.03125,
      "learning_rate": 0.00015896236785587864,
      "loss": 1.0123,
      "step": 42949
    },
    {
      "epoch": 3.695259399466575,
      "grad_norm": 12.25,
      "learning_rate": 0.00015894259395201338,
      "loss": 0.8704,
      "step": 42950
    },
    {
      "epoch": 3.6953454357738966,
      "grad_norm": 1.0,
      "learning_rate": 0.00015892282104567457,
      "loss": 0.7024,
      "step": 42951
    },
    {
      "epoch": 3.6954314720812182,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015890304913692034,
      "loss": 0.8116,
      "step": 42952
    },
    {
      "epoch": 3.69551750838854,
      "grad_norm": 1.25,
      "learning_rate": 0.0001588832782258082,
      "loss": 2.0254,
      "step": 42953
    },
    {
      "epoch": 3.6956035446958615,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001588635083123961,
      "loss": 0.9302,
      "step": 42954
    },
    {
      "epoch": 3.695689581003183,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00015884373939674206,
      "loss": 1.4311,
      "step": 42955
    },
    {
      "epoch": 3.695775617310505,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00015882397147890358,
      "loss": 0.5673,
      "step": 42956
    },
    {
      "epoch": 3.695861653617827,
      "grad_norm": 1.15625,
      "learning_rate": 0.00015880420455893863,
      "loss": 0.9449,
      "step": 42957
    },
    {
      "epoch": 3.6959476899251484,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00015878443863690518,
      "loss": 1.6106,
      "step": 42958
    },
    {
      "epoch": 3.69603372623247,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015876467371286069,
      "loss": 1.1127,
      "step": 42959
    },
    {
      "epoch": 3.6961197625397917,
      "grad_norm": 0.90625,
      "learning_rate": 0.00015874490978686317,
      "loss": 1.2505,
      "step": 42960
    },
    {
      "epoch": 3.6962057988471138,
      "grad_norm": 1.375,
      "learning_rate": 0.00015872514685897067,
      "loss": 1.9321,
      "step": 42961
    },
    {
      "epoch": 3.696291835154435,
      "grad_norm": 1.4375,
      "learning_rate": 0.00015870538492924048,
      "loss": 0.8056,
      "step": 42962
    },
    {
      "epoch": 3.696377871461757,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00015868562399773067,
      "loss": 1.2132,
      "step": 42963
    },
    {
      "epoch": 3.6964639077690786,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015866586406449923,
      "loss": 1.5688,
      "step": 42964
    },
    {
      "epoch": 3.6965499440764003,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001586461051296036,
      "loss": 0.7289,
      "step": 42965
    },
    {
      "epoch": 3.696635980383722,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0001586263471931018,
      "loss": 1.7776,
      "step": 42966
    },
    {
      "epoch": 3.6967220166910435,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00015860659025505148,
      "loss": 1.3121,
      "step": 42967
    },
    {
      "epoch": 3.6968080529983656,
      "grad_norm": 1.75,
      "learning_rate": 0.0001585868343155104,
      "loss": 1.3363,
      "step": 42968
    },
    {
      "epoch": 3.6968940893056867,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015856707937453662,
      "loss": 0.6948,
      "step": 42969
    },
    {
      "epoch": 3.696980125613009,
      "grad_norm": 1.296875,
      "learning_rate": 0.00015854732543218758,
      "loss": 1.0551,
      "step": 42970
    },
    {
      "epoch": 3.6970661619203304,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00015852757248852124,
      "loss": 1.5546,
      "step": 42971
    },
    {
      "epoch": 3.697152198227652,
      "grad_norm": 1.125,
      "learning_rate": 0.00015850782054359548,
      "loss": 1.4589,
      "step": 42972
    },
    {
      "epoch": 3.6972382345349737,
      "grad_norm": 1.5,
      "learning_rate": 0.00015848806959746769,
      "loss": 1.6486,
      "step": 42973
    },
    {
      "epoch": 3.6973242708422953,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00015846831965019598,
      "loss": 1.4227,
      "step": 42974
    },
    {
      "epoch": 3.6974103071496174,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001584485707018381,
      "loss": 0.4698,
      "step": 42975
    },
    {
      "epoch": 3.6974963434569386,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015842882275245157,
      "loss": 1.4309,
      "step": 42976
    },
    {
      "epoch": 3.6975823797642606,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001584090758020945,
      "loss": 0.6294,
      "step": 42977
    },
    {
      "epoch": 3.6976684160715823,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00015838932985082427,
      "loss": 1.3701,
      "step": 42978
    },
    {
      "epoch": 3.697754452378904,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00015836958489869878,
      "loss": 1.4513,
      "step": 42979
    },
    {
      "epoch": 3.6978404886862255,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00015834984094577603,
      "loss": 0.8187,
      "step": 42980
    },
    {
      "epoch": 3.697926524993547,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00015833009799211335,
      "loss": 0.9347,
      "step": 42981
    },
    {
      "epoch": 3.698012561300869,
      "grad_norm": 47.75,
      "learning_rate": 0.00015831035603776868,
      "loss": 1.0136,
      "step": 42982
    },
    {
      "epoch": 3.698098597608191,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001582906150827998,
      "loss": 1.3411,
      "step": 42983
    },
    {
      "epoch": 3.6981846339155124,
      "grad_norm": 1.421875,
      "learning_rate": 0.00015827087512726434,
      "loss": 1.3738,
      "step": 42984
    },
    {
      "epoch": 3.698270670222834,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00015825113617122034,
      "loss": 1.0535,
      "step": 42985
    },
    {
      "epoch": 3.6983567065301557,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001582313982147251,
      "loss": 0.8638,
      "step": 42986
    },
    {
      "epoch": 3.6984427428374773,
      "grad_norm": 1.015625,
      "learning_rate": 0.00015821166125783653,
      "loss": 1.3384,
      "step": 42987
    },
    {
      "epoch": 3.698528779144799,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001581919253006126,
      "loss": 0.7392,
      "step": 42988
    },
    {
      "epoch": 3.698614815452121,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001581721903431106,
      "loss": 1.5258,
      "step": 42989
    },
    {
      "epoch": 3.6987008517594426,
      "grad_norm": 2.765625,
      "learning_rate": 0.0001581524563853885,
      "loss": 0.9444,
      "step": 42990
    },
    {
      "epoch": 3.6987868880667643,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001581327234275041,
      "loss": 1.1424,
      "step": 42991
    },
    {
      "epoch": 3.698872924374086,
      "grad_norm": 1.0,
      "learning_rate": 0.00015811299146951485,
      "loss": 0.9135,
      "step": 42992
    },
    {
      "epoch": 3.6989589606814075,
      "grad_norm": 8.0625,
      "learning_rate": 0.00015809326051147875,
      "loss": 1.6053,
      "step": 42993
    },
    {
      "epoch": 3.699044996988729,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015807353055345325,
      "loss": 0.7543,
      "step": 42994
    },
    {
      "epoch": 3.6991310332960508,
      "grad_norm": 1.03125,
      "learning_rate": 0.00015805380159549616,
      "loss": 1.2707,
      "step": 42995
    },
    {
      "epoch": 3.699217069603373,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00015803407363766537,
      "loss": 0.8476,
      "step": 42996
    },
    {
      "epoch": 3.6993031059106944,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015801434668001818,
      "loss": 1.016,
      "step": 42997
    },
    {
      "epoch": 3.699389142218016,
      "grad_norm": 6.21875,
      "learning_rate": 0.00015799462072261256,
      "loss": 1.4673,
      "step": 42998
    },
    {
      "epoch": 3.6994751785253377,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015797489576550633,
      "loss": 1.1591,
      "step": 42999
    },
    {
      "epoch": 3.6995612148326593,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001579551718087568,
      "loss": 0.5942,
      "step": 43000
    },
    {
      "epoch": 3.699647251139981,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001579354488524219,
      "loss": 0.7073,
      "step": 43001
    },
    {
      "epoch": 3.6997332874473026,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015791572689655947,
      "loss": 0.7993,
      "step": 43002
    },
    {
      "epoch": 3.6998193237546246,
      "grad_norm": 3.640625,
      "learning_rate": 0.00015789600594122682,
      "loss": 1.4386,
      "step": 43003
    },
    {
      "epoch": 3.6999053600619463,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001578762859864818,
      "loss": 1.7824,
      "step": 43004
    },
    {
      "epoch": 3.699991396369268,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015785656703238226,
      "loss": 1.3121,
      "step": 43005
    },
    {
      "epoch": 3.7000774326765895,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001578368490789856,
      "loss": 0.68,
      "step": 43006
    },
    {
      "epoch": 3.700163468983911,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00015781713212634975,
      "loss": 1.3612,
      "step": 43007
    },
    {
      "epoch": 3.7002495052912328,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00015779741617453209,
      "loss": 1.5657,
      "step": 43008
    },
    {
      "epoch": 3.7003355415985544,
      "grad_norm": 1.109375,
      "learning_rate": 0.00015777770122359043,
      "loss": 0.9605,
      "step": 43009
    },
    {
      "epoch": 3.7004215779058764,
      "grad_norm": 9.25,
      "learning_rate": 0.0001577579872735826,
      "loss": 1.1886,
      "step": 43010
    },
    {
      "epoch": 3.700507614213198,
      "grad_norm": 0.875,
      "learning_rate": 0.0001577382743245659,
      "loss": 0.6863,
      "step": 43011
    },
    {
      "epoch": 3.7005936505205197,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001577185623765982,
      "loss": 0.6293,
      "step": 43012
    },
    {
      "epoch": 3.7006796868278413,
      "grad_norm": 1.140625,
      "learning_rate": 0.00015769885142973727,
      "loss": 1.0879,
      "step": 43013
    },
    {
      "epoch": 3.700765723135163,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001576791414840405,
      "loss": 1.4083,
      "step": 43014
    },
    {
      "epoch": 3.7008517594424846,
      "grad_norm": 34.0,
      "learning_rate": 0.00015765943253956565,
      "loss": 1.0007,
      "step": 43015
    },
    {
      "epoch": 3.700937795749806,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015763972459637056,
      "loss": 1.2501,
      "step": 43016
    },
    {
      "epoch": 3.7010238320571283,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001576200176545125,
      "loss": 0.744,
      "step": 43017
    },
    {
      "epoch": 3.70110986836445,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001576003117140495,
      "loss": 1.0294,
      "step": 43018
    },
    {
      "epoch": 3.7011959046717715,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00015758060677503876,
      "loss": 1.7919,
      "step": 43019
    },
    {
      "epoch": 3.701281940979093,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015756090283753805,
      "loss": 0.7665,
      "step": 43020
    },
    {
      "epoch": 3.7013679772864148,
      "grad_norm": 172.0,
      "learning_rate": 0.00015754119990160542,
      "loss": 0.5891,
      "step": 43021
    },
    {
      "epoch": 3.701454013593737,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015752149796729797,
      "loss": 0.6222,
      "step": 43022
    },
    {
      "epoch": 3.701540049901058,
      "grad_norm": 1.125,
      "learning_rate": 0.00015750179703467377,
      "loss": 1.3712,
      "step": 43023
    },
    {
      "epoch": 3.70162608620838,
      "grad_norm": 1.765625,
      "learning_rate": 0.00015748209710378996,
      "loss": 1.7024,
      "step": 43024
    },
    {
      "epoch": 3.7017121225157017,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00015746239817470442,
      "loss": 1.3573,
      "step": 43025
    },
    {
      "epoch": 3.7017981588230233,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00015744270024747493,
      "loss": 1.3979,
      "step": 43026
    },
    {
      "epoch": 3.701884195130345,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00015742300332215875,
      "loss": 1.1403,
      "step": 43027
    },
    {
      "epoch": 3.7019702314376666,
      "grad_norm": 2.6875,
      "learning_rate": 0.00015740330739881365,
      "loss": 1.5566,
      "step": 43028
    },
    {
      "epoch": 3.7020562677449886,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015738361247749744,
      "loss": 0.7138,
      "step": 43029
    },
    {
      "epoch": 3.70214230405231,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001573639185582673,
      "loss": 1.1144,
      "step": 43030
    },
    {
      "epoch": 3.702228340359632,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001573442256411811,
      "loss": 1.1871,
      "step": 43031
    },
    {
      "epoch": 3.7023143766669535,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00015732453372629656,
      "loss": 1.0942,
      "step": 43032
    },
    {
      "epoch": 3.702400412974275,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015730484281367097,
      "loss": 0.9624,
      "step": 43033
    },
    {
      "epoch": 3.7024864492815968,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001572851529033622,
      "loss": 1.1756,
      "step": 43034
    },
    {
      "epoch": 3.7025724855889184,
      "grad_norm": 1.4375,
      "learning_rate": 0.00015726546399542752,
      "loss": 1.2507,
      "step": 43035
    },
    {
      "epoch": 3.7026585218962405,
      "grad_norm": 328.0,
      "learning_rate": 0.00015724577608992475,
      "loss": 0.8986,
      "step": 43036
    },
    {
      "epoch": 3.7027445582035616,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00015722608918691155,
      "loss": 1.0298,
      "step": 43037
    },
    {
      "epoch": 3.7028305945108837,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00015720640328644519,
      "loss": 0.503,
      "step": 43038
    },
    {
      "epoch": 3.7029166308182053,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015718671838858345,
      "loss": 0.7644,
      "step": 43039
    },
    {
      "epoch": 3.703002667125527,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001571670344933841,
      "loss": 0.8571,
      "step": 43040
    },
    {
      "epoch": 3.7030887034328486,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015714735160090427,
      "loss": 2.2023,
      "step": 43041
    },
    {
      "epoch": 3.70317473974017,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001571276697112018,
      "loss": 0.8763,
      "step": 43042
    },
    {
      "epoch": 3.7032607760474923,
      "grad_norm": 1.21875,
      "learning_rate": 0.00015710798882433432,
      "loss": 1.5,
      "step": 43043
    },
    {
      "epoch": 3.703346812354814,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00015708830894035913,
      "loss": 1.2476,
      "step": 43044
    },
    {
      "epoch": 3.7034328486621355,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015706863005933398,
      "loss": 0.8585,
      "step": 43045
    },
    {
      "epoch": 3.703518884969457,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015704895218131653,
      "loss": 0.7772,
      "step": 43046
    },
    {
      "epoch": 3.7036049212767788,
      "grad_norm": 3.1875,
      "learning_rate": 0.00015702927530636407,
      "loss": 1.433,
      "step": 43047
    },
    {
      "epoch": 3.7036909575841004,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001570095994345344,
      "loss": 1.4389,
      "step": 43048
    },
    {
      "epoch": 3.703776993891422,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001569899245658848,
      "loss": 0.6305,
      "step": 43049
    },
    {
      "epoch": 3.703863030198744,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00015697025070047295,
      "loss": 1.0743,
      "step": 43050
    },
    {
      "epoch": 3.7039490665060657,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015695057783835653,
      "loss": 1.0754,
      "step": 43051
    },
    {
      "epoch": 3.7040351028133873,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001569309059795928,
      "loss": 1.2607,
      "step": 43052
    },
    {
      "epoch": 3.704121139120709,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001569112351242395,
      "loss": 1.0531,
      "step": 43053
    },
    {
      "epoch": 3.7042071754280306,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00015689156527235422,
      "loss": 1.1259,
      "step": 43054
    },
    {
      "epoch": 3.704293211735352,
      "grad_norm": 2.015625,
      "learning_rate": 0.00015687189642399425,
      "loss": 1.7914,
      "step": 43055
    },
    {
      "epoch": 3.704379248042674,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001568522285792172,
      "loss": 0.8634,
      "step": 43056
    },
    {
      "epoch": 3.704465284349996,
      "grad_norm": 1.125,
      "learning_rate": 0.0001568325617380807,
      "loss": 1.2475,
      "step": 43057
    },
    {
      "epoch": 3.7045513206573175,
      "grad_norm": 6.875,
      "learning_rate": 0.00015681289590064218,
      "loss": 1.2191,
      "step": 43058
    },
    {
      "epoch": 3.704637356964639,
      "grad_norm": 1.0,
      "learning_rate": 0.00015679323106695942,
      "loss": 0.8714,
      "step": 43059
    },
    {
      "epoch": 3.7047233932719608,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00015677356723708946,
      "loss": 0.8874,
      "step": 43060
    },
    {
      "epoch": 3.7048094295792824,
      "grad_norm": 0.875,
      "learning_rate": 0.0001567539044110901,
      "loss": 0.7659,
      "step": 43061
    },
    {
      "epoch": 3.704895465886604,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015673424258901898,
      "loss": 0.6975,
      "step": 43062
    },
    {
      "epoch": 3.7049815021939256,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015671458177093328,
      "loss": 1.2612,
      "step": 43063
    },
    {
      "epoch": 3.7050675385012477,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00015669492195689077,
      "loss": 1.6345,
      "step": 43064
    },
    {
      "epoch": 3.7051535748085693,
      "grad_norm": 3.015625,
      "learning_rate": 0.00015667526314694873,
      "loss": 0.7524,
      "step": 43065
    },
    {
      "epoch": 3.705239611115891,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001566556053411648,
      "loss": 1.3772,
      "step": 43066
    },
    {
      "epoch": 3.7053256474232126,
      "grad_norm": 29.625,
      "learning_rate": 0.00015663594853959655,
      "loss": 1.2176,
      "step": 43067
    },
    {
      "epoch": 3.705411683730534,
      "grad_norm": 7.6875,
      "learning_rate": 0.00015661629274230126,
      "loss": 1.1042,
      "step": 43068
    },
    {
      "epoch": 3.705497720037856,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015659663794933648,
      "loss": 0.5086,
      "step": 43069
    },
    {
      "epoch": 3.7055837563451774,
      "grad_norm": 31.0,
      "learning_rate": 0.00015657698416075993,
      "loss": 1.5229,
      "step": 43070
    },
    {
      "epoch": 3.7056697926524995,
      "grad_norm": 13.8125,
      "learning_rate": 0.00015655733137662876,
      "loss": 1.0252,
      "step": 43071
    },
    {
      "epoch": 3.705755828959821,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001565376795970006,
      "loss": 1.856,
      "step": 43072
    },
    {
      "epoch": 3.7058418652671428,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00015651802882193306,
      "loss": 2.1342,
      "step": 43073
    },
    {
      "epoch": 3.7059279015744644,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00015649837905148335,
      "loss": 1.0785,
      "step": 43074
    },
    {
      "epoch": 3.706013937881786,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001564787302857091,
      "loss": 1.1222,
      "step": 43075
    },
    {
      "epoch": 3.7060999741891076,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015645908252466783,
      "loss": 0.8141,
      "step": 43076
    },
    {
      "epoch": 3.7061860104964293,
      "grad_norm": 1.34375,
      "learning_rate": 0.00015643943576841685,
      "loss": 0.9105,
      "step": 43077
    },
    {
      "epoch": 3.7062720468037513,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00015641979001701378,
      "loss": 1.7271,
      "step": 43078
    },
    {
      "epoch": 3.706358083111073,
      "grad_norm": 1.5625,
      "learning_rate": 0.00015640014527051595,
      "loss": 0.8554,
      "step": 43079
    },
    {
      "epoch": 3.7064441194183946,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00015638050152898074,
      "loss": 1.8376,
      "step": 43080
    },
    {
      "epoch": 3.706530155725716,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000156360858792466,
      "loss": 1.1839,
      "step": 43081
    },
    {
      "epoch": 3.706616192033038,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001563412170610287,
      "loss": 1.1643,
      "step": 43082
    },
    {
      "epoch": 3.70670222834036,
      "grad_norm": 0.96875,
      "learning_rate": 0.00015632157633472648,
      "loss": 0.7684,
      "step": 43083
    },
    {
      "epoch": 3.706788264647681,
      "grad_norm": 2.109375,
      "learning_rate": 0.00015630193661361703,
      "loss": 0.9609,
      "step": 43084
    },
    {
      "epoch": 3.706874300955003,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001562822978977574,
      "loss": 0.9246,
      "step": 43085
    },
    {
      "epoch": 3.7069603372623248,
      "grad_norm": 1.375,
      "learning_rate": 0.00015626266018720514,
      "loss": 1.1093,
      "step": 43086
    },
    {
      "epoch": 3.7070463735696464,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00015624302348201795,
      "loss": 1.0053,
      "step": 43087
    },
    {
      "epoch": 3.707132409876968,
      "grad_norm": 1.359375,
      "learning_rate": 0.00015622338778225293,
      "loss": 0.8159,
      "step": 43088
    },
    {
      "epoch": 3.7072184461842896,
      "grad_norm": 1.46875,
      "learning_rate": 0.00015620375308796774,
      "loss": 1.9318,
      "step": 43089
    },
    {
      "epoch": 3.7073044824916117,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00015618411939921962,
      "loss": 1.7464,
      "step": 43090
    },
    {
      "epoch": 3.707390518798933,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015616448671606603,
      "loss": 0.7062,
      "step": 43091
    },
    {
      "epoch": 3.707476555106255,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00015614485503856457,
      "loss": 0.7178,
      "step": 43092
    },
    {
      "epoch": 3.7075625914135766,
      "grad_norm": 1.265625,
      "learning_rate": 0.00015612522436677246,
      "loss": 1.0803,
      "step": 43093
    },
    {
      "epoch": 3.707648627720898,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015610559470074709,
      "loss": 0.5698,
      "step": 43094
    },
    {
      "epoch": 3.70773466402822,
      "grad_norm": 4.53125,
      "learning_rate": 0.00015608596604054604,
      "loss": 1.0336,
      "step": 43095
    },
    {
      "epoch": 3.7078207003355415,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015606633838622662,
      "loss": 0.4725,
      "step": 43096
    },
    {
      "epoch": 3.7079067366428635,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015604671173784646,
      "loss": 1.501,
      "step": 43097
    },
    {
      "epoch": 3.7079927729501847,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00015602708609546262,
      "loss": 0.5716,
      "step": 43098
    },
    {
      "epoch": 3.7080788092575068,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015600746145913265,
      "loss": 1.0348,
      "step": 43099
    },
    {
      "epoch": 3.7081648455648284,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015598783782891406,
      "loss": 0.7246,
      "step": 43100
    },
    {
      "epoch": 3.70825088187215,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00015596821520486403,
      "loss": 1.0126,
      "step": 43101
    },
    {
      "epoch": 3.7083369181794716,
      "grad_norm": 1.0,
      "learning_rate": 0.00015594859358704,
      "loss": 1.5025,
      "step": 43102
    },
    {
      "epoch": 3.7084229544867933,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015592897297549962,
      "loss": 0.7439,
      "step": 43103
    },
    {
      "epoch": 3.7085089907941153,
      "grad_norm": 1.078125,
      "learning_rate": 0.00015590935337029993,
      "loss": 1.1331,
      "step": 43104
    },
    {
      "epoch": 3.708595027101437,
      "grad_norm": 1.015625,
      "learning_rate": 0.00015588973477149865,
      "loss": 1.1648,
      "step": 43105
    },
    {
      "epoch": 3.7086810634087586,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015587011717915273,
      "loss": 0.5648,
      "step": 43106
    },
    {
      "epoch": 3.70876709971608,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00015585050059331986,
      "loss": 1.7898,
      "step": 43107
    },
    {
      "epoch": 3.708853136023402,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00015583088501405746,
      "loss": 1.0243,
      "step": 43108
    },
    {
      "epoch": 3.7089391723307235,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015581127044142262,
      "loss": 0.6809,
      "step": 43109
    },
    {
      "epoch": 3.709025208638045,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00015579165687547287,
      "loss": 0.9453,
      "step": 43110
    },
    {
      "epoch": 3.709111244945367,
      "grad_norm": 1.59375,
      "learning_rate": 0.00015577204431626573,
      "loss": 1.2409,
      "step": 43111
    },
    {
      "epoch": 3.7091972812526888,
      "grad_norm": 14.0625,
      "learning_rate": 0.0001557524327638583,
      "loss": 0.7577,
      "step": 43112
    },
    {
      "epoch": 3.7092833175600104,
      "grad_norm": 1.078125,
      "learning_rate": 0.00015573282221830804,
      "loss": 1.0305,
      "step": 43113
    },
    {
      "epoch": 3.709369353867332,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001557132126796725,
      "loss": 0.9979,
      "step": 43114
    },
    {
      "epoch": 3.7094553901746536,
      "grad_norm": 2976.0,
      "learning_rate": 0.0001556936041480086,
      "loss": 0.9396,
      "step": 43115
    },
    {
      "epoch": 3.7095414264819753,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000155673996623374,
      "loss": 0.8158,
      "step": 43116
    },
    {
      "epoch": 3.709627462789297,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00015565439010582617,
      "loss": 1.1099,
      "step": 43117
    },
    {
      "epoch": 3.709713499096619,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001556347845954221,
      "loss": 0.8607,
      "step": 43118
    },
    {
      "epoch": 3.7097995354039406,
      "grad_norm": 2.90625,
      "learning_rate": 0.00015561518009221946,
      "loss": 1.4376,
      "step": 43119
    },
    {
      "epoch": 3.709885571711262,
      "grad_norm": 418.0,
      "learning_rate": 0.00015559557659627533,
      "loss": 1.1781,
      "step": 43120
    },
    {
      "epoch": 3.709971608018584,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00015557597410764706,
      "loss": 0.8446,
      "step": 43121
    },
    {
      "epoch": 3.7100576443259055,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015555637262639233,
      "loss": 0.4836,
      "step": 43122
    },
    {
      "epoch": 3.710143680633227,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000155536772152568,
      "loss": 0.8862,
      "step": 43123
    },
    {
      "epoch": 3.7102297169405487,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015551717268623167,
      "loss": 1.0888,
      "step": 43124
    },
    {
      "epoch": 3.7103157532478708,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001554975742274407,
      "loss": 1.3231,
      "step": 43125
    },
    {
      "epoch": 3.7104017895551924,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015547797677625225,
      "loss": 0.9841,
      "step": 43126
    },
    {
      "epoch": 3.710487825862514,
      "grad_norm": 2208.0,
      "learning_rate": 0.00015545838033272365,
      "loss": 1.2106,
      "step": 43127
    },
    {
      "epoch": 3.7105738621698356,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00015543878489691242,
      "loss": 1.1482,
      "step": 43128
    },
    {
      "epoch": 3.7106598984771573,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00015541919046887565,
      "loss": 1.4471,
      "step": 43129
    },
    {
      "epoch": 3.710745934784479,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00015539959704867086,
      "loss": 1.0468,
      "step": 43130
    },
    {
      "epoch": 3.7108319710918005,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00015538000463635488,
      "loss": 1.5326,
      "step": 43131
    },
    {
      "epoch": 3.7109180073991226,
      "grad_norm": 1.53125,
      "learning_rate": 0.00015536041323198553,
      "loss": 1.3291,
      "step": 43132
    },
    {
      "epoch": 3.711004043706444,
      "grad_norm": 6.03125,
      "learning_rate": 0.00015534082283562013,
      "loss": 0.5474,
      "step": 43133
    },
    {
      "epoch": 3.711090080013766,
      "grad_norm": 1.1875,
      "learning_rate": 0.00015532123344731558,
      "loss": 1.0624,
      "step": 43134
    },
    {
      "epoch": 3.7111761163210875,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015530164506712956,
      "loss": 0.8543,
      "step": 43135
    },
    {
      "epoch": 3.711262152628409,
      "grad_norm": 26.75,
      "learning_rate": 0.00015528205769511905,
      "loss": 0.8378,
      "step": 43136
    },
    {
      "epoch": 3.711348188935731,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00015526247133134147,
      "loss": 0.6251,
      "step": 43137
    },
    {
      "epoch": 3.7114342252430523,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00015524288597585427,
      "loss": 1.5503,
      "step": 43138
    },
    {
      "epoch": 3.7115202615503744,
      "grad_norm": 6.75,
      "learning_rate": 0.00015522330162871445,
      "loss": 1.4529,
      "step": 43139
    },
    {
      "epoch": 3.711606297857696,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00015520371828997936,
      "loss": 1.8588,
      "step": 43140
    },
    {
      "epoch": 3.7116923341650176,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015518413595970654,
      "loss": 0.6241,
      "step": 43141
    },
    {
      "epoch": 3.7117783704723393,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00015516455463795287,
      "loss": 0.8545,
      "step": 43142
    },
    {
      "epoch": 3.711864406779661,
      "grad_norm": 1.25,
      "learning_rate": 0.00015514497432477588,
      "loss": 1.8883,
      "step": 43143
    },
    {
      "epoch": 3.711950443086983,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00015512539502023282,
      "loss": 1.2952,
      "step": 43144
    },
    {
      "epoch": 3.712036479394304,
      "grad_norm": 1.296875,
      "learning_rate": 0.00015510581672438083,
      "loss": 1.3245,
      "step": 43145
    },
    {
      "epoch": 3.712122515701626,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015508623943727735,
      "loss": 1.3123,
      "step": 43146
    },
    {
      "epoch": 3.712208552008948,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015506666315897943,
      "loss": 0.9572,
      "step": 43147
    },
    {
      "epoch": 3.7122945883162695,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015504708788954442,
      "loss": 1.4783,
      "step": 43148
    },
    {
      "epoch": 3.712380624623591,
      "grad_norm": 1.03125,
      "learning_rate": 0.00015502751362902973,
      "loss": 0.953,
      "step": 43149
    },
    {
      "epoch": 3.7124666609309127,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015500794037749232,
      "loss": 0.7493,
      "step": 43150
    },
    {
      "epoch": 3.7125526972382348,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00015498836813498956,
      "loss": 0.8758,
      "step": 43151
    },
    {
      "epoch": 3.712638733545556,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001549687969015789,
      "loss": 1.2225,
      "step": 43152
    },
    {
      "epoch": 3.712724769852878,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00015494922667731725,
      "loss": 0.9786,
      "step": 43153
    },
    {
      "epoch": 3.7128108061601997,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000154929657462262,
      "loss": 1.1606,
      "step": 43154
    },
    {
      "epoch": 3.7128968424675213,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001549100892564706,
      "loss": 0.5614,
      "step": 43155
    },
    {
      "epoch": 3.712982878774843,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001548905220599998,
      "loss": 0.9591,
      "step": 43156
    },
    {
      "epoch": 3.7130689150821645,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015487095587290723,
      "loss": 1.8633,
      "step": 43157
    },
    {
      "epoch": 3.7131549513894866,
      "grad_norm": 0.96875,
      "learning_rate": 0.00015485139069525006,
      "loss": 0.8998,
      "step": 43158
    },
    {
      "epoch": 3.713240987696808,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015483182652708532,
      "loss": 0.6651,
      "step": 43159
    },
    {
      "epoch": 3.71332702400413,
      "grad_norm": 1.265625,
      "learning_rate": 0.00015481226336847047,
      "loss": 0.7474,
      "step": 43160
    },
    {
      "epoch": 3.7134130603114515,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00015479270121946255,
      "loss": 0.7296,
      "step": 43161
    },
    {
      "epoch": 3.713499096618773,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00015477314008011879,
      "loss": 0.9804,
      "step": 43162
    },
    {
      "epoch": 3.7135851329260947,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001547535799504966,
      "loss": 0.6385,
      "step": 43163
    },
    {
      "epoch": 3.7136711692334163,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015473402083065286,
      "loss": 1.4335,
      "step": 43164
    },
    {
      "epoch": 3.7137572055407384,
      "grad_norm": 0.796875,
      "learning_rate": 0.000154714462720645,
      "loss": 1.179,
      "step": 43165
    },
    {
      "epoch": 3.71384324184806,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001546949056205303,
      "loss": 1.2107,
      "step": 43166
    },
    {
      "epoch": 3.7139292781553817,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00015467534953036573,
      "loss": 1.4602,
      "step": 43167
    },
    {
      "epoch": 3.7140153144627033,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00015465579445020854,
      "loss": 0.9162,
      "step": 43168
    },
    {
      "epoch": 3.714101350770025,
      "grad_norm": 1.15625,
      "learning_rate": 0.00015463624038011608,
      "loss": 0.7527,
      "step": 43169
    },
    {
      "epoch": 3.7141873870773465,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015461668732014538,
      "loss": 1.0877,
      "step": 43170
    },
    {
      "epoch": 3.714273423384668,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00015459713527035386,
      "loss": 1.0084,
      "step": 43171
    },
    {
      "epoch": 3.71435945969199,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00015457758423079843,
      "loss": 0.852,
      "step": 43172
    },
    {
      "epoch": 3.714445495999312,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0001545580342015363,
      "loss": 1.4652,
      "step": 43173
    },
    {
      "epoch": 3.7145315323066335,
      "grad_norm": 0.9375,
      "learning_rate": 0.00015453848518262497,
      "loss": 0.8943,
      "step": 43174
    },
    {
      "epoch": 3.714617568613955,
      "grad_norm": 1.859375,
      "learning_rate": 0.00015451893717412119,
      "loss": 0.9894,
      "step": 43175
    },
    {
      "epoch": 3.7147036049212767,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001544993901760825,
      "loss": 1.0035,
      "step": 43176
    },
    {
      "epoch": 3.7147896412285983,
      "grad_norm": 2.40625,
      "learning_rate": 0.0001544798441885657,
      "loss": 0.9752,
      "step": 43177
    },
    {
      "epoch": 3.71487567753592,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001544602992116282,
      "loss": 1.3209,
      "step": 43178
    },
    {
      "epoch": 3.714961713843242,
      "grad_norm": 0.96875,
      "learning_rate": 0.00015444075524532732,
      "loss": 1.0497,
      "step": 43179
    },
    {
      "epoch": 3.7150477501505637,
      "grad_norm": 1.515625,
      "learning_rate": 0.00015442121228971978,
      "loss": 1.1083,
      "step": 43180
    },
    {
      "epoch": 3.7151337864578853,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00015440167034486303,
      "loss": 1.6092,
      "step": 43181
    },
    {
      "epoch": 3.715219822765207,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001543821294108143,
      "loss": 0.8825,
      "step": 43182
    },
    {
      "epoch": 3.7153058590725285,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00015436258948763049,
      "loss": 0.4932,
      "step": 43183
    },
    {
      "epoch": 3.71539189537985,
      "grad_norm": 1.125,
      "learning_rate": 0.00015434305057536885,
      "loss": 0.9514,
      "step": 43184
    },
    {
      "epoch": 3.7154779316871718,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0001543235126740868,
      "loss": 1.7046,
      "step": 43185
    },
    {
      "epoch": 3.715563967994494,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00015430397578384097,
      "loss": 1.4187,
      "step": 43186
    },
    {
      "epoch": 3.7156500043018155,
      "grad_norm": 4.3125,
      "learning_rate": 0.00015428443990468904,
      "loss": 1.1457,
      "step": 43187
    },
    {
      "epoch": 3.715736040609137,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015426490503668767,
      "loss": 0.7479,
      "step": 43188
    },
    {
      "epoch": 3.7158220769164587,
      "grad_norm": 1.8984375,
      "learning_rate": 0.00015424537117989423,
      "loss": 1.2186,
      "step": 43189
    },
    {
      "epoch": 3.7159081132237803,
      "grad_norm": 125.0,
      "learning_rate": 0.00015422583833436598,
      "loss": 1.0543,
      "step": 43190
    },
    {
      "epoch": 3.715994149531102,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015420630650015977,
      "loss": 0.5057,
      "step": 43191
    },
    {
      "epoch": 3.7160801858384236,
      "grad_norm": 0.921875,
      "learning_rate": 0.00015418677567733286,
      "loss": 0.9624,
      "step": 43192
    },
    {
      "epoch": 3.7161662221457457,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001541672458659425,
      "loss": 1.8642,
      "step": 43193
    },
    {
      "epoch": 3.7162522584530673,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015414771706604553,
      "loss": 0.7327,
      "step": 43194
    },
    {
      "epoch": 3.716338294760389,
      "grad_norm": 1.46875,
      "learning_rate": 0.00015412818927769924,
      "loss": 1.7935,
      "step": 43195
    },
    {
      "epoch": 3.7164243310677105,
      "grad_norm": 1.0,
      "learning_rate": 0.0001541086625009609,
      "loss": 0.844,
      "step": 43196
    },
    {
      "epoch": 3.716510367375032,
      "grad_norm": 64.5,
      "learning_rate": 0.00015408913673588726,
      "loss": 1.5883,
      "step": 43197
    },
    {
      "epoch": 3.716596403682354,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00015406961198253562,
      "loss": 1.7811,
      "step": 43198
    },
    {
      "epoch": 3.7166824399896754,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001540500882409633,
      "loss": 1.1773,
      "step": 43199
    },
    {
      "epoch": 3.7167684762969975,
      "grad_norm": 0.890625,
      "learning_rate": 0.00015403056551122697,
      "loss": 0.6479,
      "step": 43200
    },
    {
      "epoch": 3.716854512604319,
      "grad_norm": 7.34375,
      "learning_rate": 0.0001540110437933841,
      "loss": 1.064,
      "step": 43201
    },
    {
      "epoch": 3.7169405489116407,
      "grad_norm": 1.296875,
      "learning_rate": 0.00015399152308749153,
      "loss": 0.9022,
      "step": 43202
    },
    {
      "epoch": 3.7170265852189623,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001539720033936064,
      "loss": 1.0281,
      "step": 43203
    },
    {
      "epoch": 3.717112621526284,
      "grad_norm": 73.0,
      "learning_rate": 0.00015395248471178607,
      "loss": 0.5452,
      "step": 43204
    },
    {
      "epoch": 3.717198657833606,
      "grad_norm": 1.125,
      "learning_rate": 0.0001539329670420871,
      "loss": 1.4509,
      "step": 43205
    },
    {
      "epoch": 3.717284694140927,
      "grad_norm": 1.28125,
      "learning_rate": 0.00015391345038456717,
      "loss": 1.0972,
      "step": 43206
    },
    {
      "epoch": 3.7173707304482493,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000153893934739283,
      "loss": 0.6761,
      "step": 43207
    },
    {
      "epoch": 3.717456766755571,
      "grad_norm": 1.75,
      "learning_rate": 0.00015387442010629166,
      "loss": 0.9748,
      "step": 43208
    },
    {
      "epoch": 3.7175428030628925,
      "grad_norm": 1.796875,
      "learning_rate": 0.00015385490648565055,
      "loss": 1.5878,
      "step": 43209
    },
    {
      "epoch": 3.717628839370214,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00015383539387741634,
      "loss": 1.6648,
      "step": 43210
    },
    {
      "epoch": 3.717714875677536,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015381588228164628,
      "loss": 0.915,
      "step": 43211
    },
    {
      "epoch": 3.717800911984858,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001537963716983976,
      "loss": 1.1261,
      "step": 43212
    },
    {
      "epoch": 3.717886948292179,
      "grad_norm": 264.0,
      "learning_rate": 0.000153776862127727,
      "loss": 0.7827,
      "step": 43213
    },
    {
      "epoch": 3.717972984599501,
      "grad_norm": 0.921875,
      "learning_rate": 0.00015375735356969178,
      "loss": 1.1454,
      "step": 43214
    },
    {
      "epoch": 3.7180590209068227,
      "grad_norm": 1.34375,
      "learning_rate": 0.00015373784602434914,
      "loss": 1.3451,
      "step": 43215
    },
    {
      "epoch": 3.7181450572141443,
      "grad_norm": 8.3125,
      "learning_rate": 0.0001537183394917558,
      "loss": 1.0099,
      "step": 43216
    },
    {
      "epoch": 3.718231093521466,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001536988339719691,
      "loss": 1.308,
      "step": 43217
    },
    {
      "epoch": 3.7183171298287876,
      "grad_norm": 1.125,
      "learning_rate": 0.00015367932946504576,
      "loss": 1.7466,
      "step": 43218
    },
    {
      "epoch": 3.7184031661361097,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00015365982597104305,
      "loss": 0.6305,
      "step": 43219
    },
    {
      "epoch": 3.7184892024434313,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015364032349001822,
      "loss": 1.113,
      "step": 43220
    },
    {
      "epoch": 3.718575238750753,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00015362082202202782,
      "loss": 1.7874,
      "step": 43221
    },
    {
      "epoch": 3.7186612750580745,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001536013215671292,
      "loss": 0.4318,
      "step": 43222
    },
    {
      "epoch": 3.718747311365396,
      "grad_norm": 1.78125,
      "learning_rate": 0.0001535818221253795,
      "loss": 1.0895,
      "step": 43223
    },
    {
      "epoch": 3.718833347672718,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001535623236968354,
      "loss": 1.0271,
      "step": 43224
    },
    {
      "epoch": 3.7189193839800394,
      "grad_norm": 1.0,
      "learning_rate": 0.0001535428262815542,
      "loss": 1.1717,
      "step": 43225
    },
    {
      "epoch": 3.7190054202873615,
      "grad_norm": 1.078125,
      "learning_rate": 0.000153523329879593,
      "loss": 1.5351,
      "step": 43226
    },
    {
      "epoch": 3.719091456594683,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015350383449100848,
      "loss": 0.7741,
      "step": 43227
    },
    {
      "epoch": 3.7191774929020047,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00015348434011585782,
      "loss": 1.1811,
      "step": 43228
    },
    {
      "epoch": 3.7192635292093263,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001534648467541983,
      "loss": 0.5626,
      "step": 43229
    },
    {
      "epoch": 3.719349565516648,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00015344535440608653,
      "loss": 1.369,
      "step": 43230
    },
    {
      "epoch": 3.7194356018239696,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0001534258630715798,
      "loss": 1.2262,
      "step": 43231
    },
    {
      "epoch": 3.719521638131291,
      "grad_norm": 1.0,
      "learning_rate": 0.00015340637275073493,
      "loss": 1.0776,
      "step": 43232
    },
    {
      "epoch": 3.7196076744386133,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00015338688344360897,
      "loss": 1.6351,
      "step": 43233
    },
    {
      "epoch": 3.719693710745935,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00015336739515025915,
      "loss": 0.961,
      "step": 43234
    },
    {
      "epoch": 3.7197797470532565,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001533479078707421,
      "loss": 0.7149,
      "step": 43235
    },
    {
      "epoch": 3.719865783360578,
      "grad_norm": 1.25,
      "learning_rate": 0.00015332842160511502,
      "loss": 1.4689,
      "step": 43236
    },
    {
      "epoch": 3.7199518196679,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000153308936353435,
      "loss": 1.358,
      "step": 43237
    },
    {
      "epoch": 3.7200378559752214,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001532894521157588,
      "loss": 0.7979,
      "step": 43238
    },
    {
      "epoch": 3.720123892282543,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001532699688921435,
      "loss": 0.6011,
      "step": 43239
    },
    {
      "epoch": 3.720209928589865,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015325048668264624,
      "loss": 2.0694,
      "step": 43240
    },
    {
      "epoch": 3.7202959648971867,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00015323100548732377,
      "loss": 1.0381,
      "step": 43241
    },
    {
      "epoch": 3.7203820012045083,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001532115253062331,
      "loss": 1.1614,
      "step": 43242
    },
    {
      "epoch": 3.72046803751183,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00015319204613943133,
      "loss": 1.1467,
      "step": 43243
    },
    {
      "epoch": 3.7205540738191516,
      "grad_norm": 0.890625,
      "learning_rate": 0.00015317256798697533,
      "loss": 1.1402,
      "step": 43244
    },
    {
      "epoch": 3.720640110126473,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00015315309084892233,
      "loss": 1.3836,
      "step": 43245
    },
    {
      "epoch": 3.720726146433795,
      "grad_norm": 1.3125,
      "learning_rate": 0.00015313361472532884,
      "loss": 1.0674,
      "step": 43246
    },
    {
      "epoch": 3.720812182741117,
      "grad_norm": 2.3125,
      "learning_rate": 0.0001531141396162523,
      "loss": 1.4316,
      "step": 43247
    },
    {
      "epoch": 3.7208982190484385,
      "grad_norm": 5.15625,
      "learning_rate": 0.00015309466552174922,
      "loss": 0.8111,
      "step": 43248
    },
    {
      "epoch": 3.72098425535576,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001530751924418768,
      "loss": 1.0586,
      "step": 43249
    },
    {
      "epoch": 3.721070291663082,
      "grad_norm": 110.5,
      "learning_rate": 0.00015305572037669214,
      "loss": 1.2893,
      "step": 43250
    },
    {
      "epoch": 3.7211563279704034,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015303624932625187,
      "loss": 0.7249,
      "step": 43251
    },
    {
      "epoch": 3.721242364277725,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015301677929061304,
      "loss": 1.2671,
      "step": 43252
    },
    {
      "epoch": 3.7213284005850467,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00015299731026983282,
      "loss": 1.0236,
      "step": 43253
    },
    {
      "epoch": 3.7214144368923687,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015297784226396788,
      "loss": 1.3868,
      "step": 43254
    },
    {
      "epoch": 3.7215004731996904,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015295837527307522,
      "loss": 1.5873,
      "step": 43255
    },
    {
      "epoch": 3.721586509507012,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015293890929721194,
      "loss": 1.0581,
      "step": 43256
    },
    {
      "epoch": 3.7216725458143336,
      "grad_norm": 0.9375,
      "learning_rate": 0.00015291944433643473,
      "loss": 1.3618,
      "step": 43257
    },
    {
      "epoch": 3.7217585821216552,
      "grad_norm": 1.46875,
      "learning_rate": 0.00015289998039080078,
      "loss": 1.0411,
      "step": 43258
    },
    {
      "epoch": 3.7218446184289773,
      "grad_norm": 2.625,
      "learning_rate": 0.00015288051746036675,
      "loss": 0.8834,
      "step": 43259
    },
    {
      "epoch": 3.7219306547362985,
      "grad_norm": 1.125,
      "learning_rate": 0.0001528610555451897,
      "loss": 0.929,
      "step": 43260
    },
    {
      "epoch": 3.7220166910436205,
      "grad_norm": 1.359375,
      "learning_rate": 0.00015284159464532665,
      "loss": 1.2486,
      "step": 43261
    },
    {
      "epoch": 3.722102727350942,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001528221347608343,
      "loss": 0.4262,
      "step": 43262
    },
    {
      "epoch": 3.722188763658264,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015280267589176967,
      "loss": 1.1368,
      "step": 43263
    },
    {
      "epoch": 3.7222747999655854,
      "grad_norm": 2.8125,
      "learning_rate": 0.00015278321803818983,
      "loss": 0.8119,
      "step": 43264
    },
    {
      "epoch": 3.722360836272907,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015276376120015146,
      "loss": 1.4536,
      "step": 43265
    },
    {
      "epoch": 3.722446872580229,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001527443053777115,
      "loss": 0.7162,
      "step": 43266
    },
    {
      "epoch": 3.7225329088875503,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00015272485057092706,
      "loss": 0.8329,
      "step": 43267
    },
    {
      "epoch": 3.7226189451948724,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015270539677985474,
      "loss": 1.177,
      "step": 43268
    },
    {
      "epoch": 3.722704981502194,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001526859440045516,
      "loss": 1.1582,
      "step": 43269
    },
    {
      "epoch": 3.7227910178095156,
      "grad_norm": 1.25,
      "learning_rate": 0.00015266649224507468,
      "loss": 1.1487,
      "step": 43270
    },
    {
      "epoch": 3.7228770541168372,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015264704150148056,
      "loss": 1.0893,
      "step": 43271
    },
    {
      "epoch": 3.722963090424159,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00015262759177382646,
      "loss": 1.3461,
      "step": 43272
    },
    {
      "epoch": 3.723049126731481,
      "grad_norm": 1.15625,
      "learning_rate": 0.00015260814306216892,
      "loss": 1.1227,
      "step": 43273
    },
    {
      "epoch": 3.723135163038802,
      "grad_norm": 0.953125,
      "learning_rate": 0.000152588695366565,
      "loss": 0.6,
      "step": 43274
    },
    {
      "epoch": 3.723221199346124,
      "grad_norm": 1.765625,
      "learning_rate": 0.00015256924868707172,
      "loss": 1.2894,
      "step": 43275
    },
    {
      "epoch": 3.723307235653446,
      "grad_norm": 1.296875,
      "learning_rate": 0.00015254980302374565,
      "loss": 0.9326,
      "step": 43276
    },
    {
      "epoch": 3.7233932719607674,
      "grad_norm": 1.453125,
      "learning_rate": 0.00015253035837664385,
      "loss": 0.7549,
      "step": 43277
    },
    {
      "epoch": 3.723479308268089,
      "grad_norm": 2.5625,
      "learning_rate": 0.00015251091474582335,
      "loss": 1.2663,
      "step": 43278
    },
    {
      "epoch": 3.7235653445754107,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015249147213134052,
      "loss": 1.0645,
      "step": 43279
    },
    {
      "epoch": 3.7236513808827327,
      "grad_norm": 0.984375,
      "learning_rate": 0.00015247203053325286,
      "loss": 1.4968,
      "step": 43280
    },
    {
      "epoch": 3.7237374171900544,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00015245258995161677,
      "loss": 0.7758,
      "step": 43281
    },
    {
      "epoch": 3.723823453497376,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001524331503864892,
      "loss": 1.4468,
      "step": 43282
    },
    {
      "epoch": 3.7239094898046976,
      "grad_norm": 5.8125,
      "learning_rate": 0.00015241371183792729,
      "loss": 0.9445,
      "step": 43283
    },
    {
      "epoch": 3.7239955261120192,
      "grad_norm": 1.125,
      "learning_rate": 0.00015239427430598752,
      "loss": 1.3159,
      "step": 43284
    },
    {
      "epoch": 3.724081562419341,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001523748377907268,
      "loss": 1.3901,
      "step": 43285
    },
    {
      "epoch": 3.7241675987266625,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015235540229220236,
      "loss": 1.547,
      "step": 43286
    },
    {
      "epoch": 3.7242536350339845,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001523359678104705,
      "loss": 1.2953,
      "step": 43287
    },
    {
      "epoch": 3.724339671341306,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001523165343455885,
      "loss": 1.4235,
      "step": 43288
    },
    {
      "epoch": 3.724425707648628,
      "grad_norm": 2.375,
      "learning_rate": 0.00015229710189761287,
      "loss": 1.7263,
      "step": 43289
    },
    {
      "epoch": 3.7245117439559494,
      "grad_norm": 3.1875,
      "learning_rate": 0.00015227767046660057,
      "loss": 0.9114,
      "step": 43290
    },
    {
      "epoch": 3.724597780263271,
      "grad_norm": 5.15625,
      "learning_rate": 0.00015225824005260863,
      "loss": 1.3999,
      "step": 43291
    },
    {
      "epoch": 3.7246838165705927,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00015223881065569355,
      "loss": 1.3902,
      "step": 43292
    },
    {
      "epoch": 3.7247698528779143,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001522193822759123,
      "loss": 1.4282,
      "step": 43293
    },
    {
      "epoch": 3.7248558891852364,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001521999549133219,
      "loss": 1.1499,
      "step": 43294
    },
    {
      "epoch": 3.724941925492558,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00015218052856797882,
      "loss": 1.3432,
      "step": 43295
    },
    {
      "epoch": 3.7250279617998796,
      "grad_norm": 3.9375,
      "learning_rate": 0.00015216110323994003,
      "loss": 1.5278,
      "step": 43296
    },
    {
      "epoch": 3.7251139981072012,
      "grad_norm": 145.0,
      "learning_rate": 0.00015214167892926256,
      "loss": 0.5758,
      "step": 43297
    },
    {
      "epoch": 3.725200034414523,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00015212225563600285,
      "loss": 0.8747,
      "step": 43298
    },
    {
      "epoch": 3.7252860707218445,
      "grad_norm": 19.0,
      "learning_rate": 0.000152102833360218,
      "loss": 0.9867,
      "step": 43299
    },
    {
      "epoch": 3.725372107029166,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00015208341210196457,
      "loss": 1.0878,
      "step": 43300
    },
    {
      "epoch": 3.725458143336488,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001520639918612995,
      "loss": 1.2279,
      "step": 43301
    },
    {
      "epoch": 3.72554417964381,
      "grad_norm": 2.421875,
      "learning_rate": 0.00015204457263827975,
      "loss": 1.2802,
      "step": 43302
    },
    {
      "epoch": 3.7256302159511314,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00015202515443296178,
      "loss": 0.7113,
      "step": 43303
    },
    {
      "epoch": 3.725716252258453,
      "grad_norm": 0.875,
      "learning_rate": 0.00015200573724540256,
      "loss": 1.3702,
      "step": 43304
    },
    {
      "epoch": 3.7258022885657747,
      "grad_norm": 1.375,
      "learning_rate": 0.000151986321075659,
      "loss": 1.1978,
      "step": 43305
    },
    {
      "epoch": 3.7258883248730963,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015196690592378763,
      "loss": 0.7905,
      "step": 43306
    },
    {
      "epoch": 3.725974361180418,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001519474917898454,
      "loss": 1.4086,
      "step": 43307
    },
    {
      "epoch": 3.72606039748774,
      "grad_norm": 142.0,
      "learning_rate": 0.00015192807867388925,
      "loss": 0.8606,
      "step": 43308
    },
    {
      "epoch": 3.7261464337950616,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015190866657597557,
      "loss": 0.4885,
      "step": 43309
    },
    {
      "epoch": 3.7262324701023832,
      "grad_norm": 1.0,
      "learning_rate": 0.00015188925549616133,
      "loss": 1.2419,
      "step": 43310
    },
    {
      "epoch": 3.726318506409705,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001518698454345035,
      "loss": 1.2276,
      "step": 43311
    },
    {
      "epoch": 3.7264045427170265,
      "grad_norm": 1.328125,
      "learning_rate": 0.00015185043639105844,
      "loss": 0.7733,
      "step": 43312
    },
    {
      "epoch": 3.7264905790243485,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00015183102836588335,
      "loss": 1.4995,
      "step": 43313
    },
    {
      "epoch": 3.7265766153316697,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00015181162135903463,
      "loss": 1.2907,
      "step": 43314
    },
    {
      "epoch": 3.726662651638992,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001517922153705692,
      "loss": 0.8933,
      "step": 43315
    },
    {
      "epoch": 3.7267486879463134,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015177281040054387,
      "loss": 0.9342,
      "step": 43316
    },
    {
      "epoch": 3.726834724253635,
      "grad_norm": 19.75,
      "learning_rate": 0.00015175340644901525,
      "loss": 0.9235,
      "step": 43317
    },
    {
      "epoch": 3.7269207605609567,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001517340035160404,
      "loss": 1.5053,
      "step": 43318
    },
    {
      "epoch": 3.7270067968682783,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00015171460160167565,
      "loss": 1.1973,
      "step": 43319
    },
    {
      "epoch": 3.7270928331756004,
      "grad_norm": 9.4375,
      "learning_rate": 0.00015169520070597796,
      "loss": 0.8333,
      "step": 43320
    },
    {
      "epoch": 3.7271788694829215,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015167580082900428,
      "loss": 1.5097,
      "step": 43321
    },
    {
      "epoch": 3.7272649057902436,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00015165640197081093,
      "loss": 0.7668,
      "step": 43322
    },
    {
      "epoch": 3.7273509420975652,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015163700413145487,
      "loss": 1.4632,
      "step": 43323
    },
    {
      "epoch": 3.727436978404887,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000151617607310993,
      "loss": 0.4645,
      "step": 43324
    },
    {
      "epoch": 3.7275230147122085,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015159821150948166,
      "loss": 0.7356,
      "step": 43325
    },
    {
      "epoch": 3.72760905101953,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015157881672697788,
      "loss": 2.1953,
      "step": 43326
    },
    {
      "epoch": 3.727695087326852,
      "grad_norm": 11.875,
      "learning_rate": 0.00015155942296353836,
      "loss": 0.8496,
      "step": 43327
    },
    {
      "epoch": 3.7277811236341734,
      "grad_norm": 5.09375,
      "learning_rate": 0.00015154003021921963,
      "loss": 1.0983,
      "step": 43328
    },
    {
      "epoch": 3.7278671599414954,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00015152063849407872,
      "loss": 1.4892,
      "step": 43329
    },
    {
      "epoch": 3.727953196248817,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00015150124778817205,
      "loss": 1.0524,
      "step": 43330
    },
    {
      "epoch": 3.7280392325561387,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00015148185810155636,
      "loss": 0.6866,
      "step": 43331
    },
    {
      "epoch": 3.7281252688634603,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015146246943428866,
      "loss": 1.3248,
      "step": 43332
    },
    {
      "epoch": 3.728211305170782,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001514430817864253,
      "loss": 0.6715,
      "step": 43333
    },
    {
      "epoch": 3.728297341478104,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00015142369515802305,
      "loss": 1.8563,
      "step": 43334
    },
    {
      "epoch": 3.7283833777854256,
      "grad_norm": 7.0,
      "learning_rate": 0.00015140430954913898,
      "loss": 1.2775,
      "step": 43335
    },
    {
      "epoch": 3.7284694140927472,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00015138492495982926,
      "loss": 1.8741,
      "step": 43336
    },
    {
      "epoch": 3.728555450400069,
      "grad_norm": 171.0,
      "learning_rate": 0.00015136554139015085,
      "loss": 0.9456,
      "step": 43337
    },
    {
      "epoch": 3.7286414867073905,
      "grad_norm": 9.5625,
      "learning_rate": 0.00015134615884016063,
      "loss": 0.8294,
      "step": 43338
    },
    {
      "epoch": 3.728727523014712,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001513267773099149,
      "loss": 0.8141,
      "step": 43339
    },
    {
      "epoch": 3.7288135593220337,
      "grad_norm": 1.109375,
      "learning_rate": 0.00015130739679947063,
      "loss": 1.1452,
      "step": 43340
    },
    {
      "epoch": 3.728899595629356,
      "grad_norm": 1.0,
      "learning_rate": 0.00015128801730888435,
      "loss": 1.0429,
      "step": 43341
    },
    {
      "epoch": 3.7289856319366774,
      "grad_norm": 8.0625,
      "learning_rate": 0.00015126863883821275,
      "loss": 0.9032,
      "step": 43342
    },
    {
      "epoch": 3.729071668243999,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001512492613875127,
      "loss": 1.4775,
      "step": 43343
    },
    {
      "epoch": 3.7291577045513207,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001512298849568406,
      "loss": 0.7778,
      "step": 43344
    },
    {
      "epoch": 3.7292437408586423,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015121050954625327,
      "loss": 1.9284,
      "step": 43345
    },
    {
      "epoch": 3.729329777165964,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015119113515580745,
      "loss": 0.9191,
      "step": 43346
    },
    {
      "epoch": 3.7294158134732855,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015117176178555965,
      "loss": 0.9146,
      "step": 43347
    },
    {
      "epoch": 3.7295018497806076,
      "grad_norm": 1.015625,
      "learning_rate": 0.00015115238943556652,
      "loss": 1.37,
      "step": 43348
    },
    {
      "epoch": 3.7295878860879292,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015113301810588503,
      "loss": 1.2423,
      "step": 43349
    },
    {
      "epoch": 3.729673922395251,
      "grad_norm": 2.1875,
      "learning_rate": 0.00015111364779657143,
      "loss": 1.0437,
      "step": 43350
    },
    {
      "epoch": 3.7297599587025725,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00015109427850768255,
      "loss": 1.0907,
      "step": 43351
    },
    {
      "epoch": 3.729845995009894,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001510749102392752,
      "loss": 0.7946,
      "step": 43352
    },
    {
      "epoch": 3.7299320313172157,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00015105554299140562,
      "loss": 1.4849,
      "step": 43353
    },
    {
      "epoch": 3.7300180676245374,
      "grad_norm": 1.234375,
      "learning_rate": 0.00015103617676413096,
      "loss": 1.1172,
      "step": 43354
    },
    {
      "epoch": 3.7301041039318594,
      "grad_norm": 0.890625,
      "learning_rate": 0.00015101681155750747,
      "loss": 0.7081,
      "step": 43355
    },
    {
      "epoch": 3.730190140239181,
      "grad_norm": 18.75,
      "learning_rate": 0.00015099744737159194,
      "loss": 0.7672,
      "step": 43356
    },
    {
      "epoch": 3.7302761765465027,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00015097808420644115,
      "loss": 0.4389,
      "step": 43357
    },
    {
      "epoch": 3.7303622128538243,
      "grad_norm": 11.375,
      "learning_rate": 0.0001509587220621114,
      "loss": 0.7435,
      "step": 43358
    },
    {
      "epoch": 3.730448249161146,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00015093936093865968,
      "loss": 1.2338,
      "step": 43359
    },
    {
      "epoch": 3.7305342854684675,
      "grad_norm": 1.59375,
      "learning_rate": 0.00015092000083614232,
      "loss": 1.2847,
      "step": 43360
    },
    {
      "epoch": 3.730620321775789,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015090064175461604,
      "loss": 1.3481,
      "step": 43361
    },
    {
      "epoch": 3.7307063580831112,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001508812836941376,
      "loss": 0.5331,
      "step": 43362
    },
    {
      "epoch": 3.730792394390433,
      "grad_norm": 74.0,
      "learning_rate": 0.00015086192665476338,
      "loss": 1.2174,
      "step": 43363
    },
    {
      "epoch": 3.7308784306977545,
      "grad_norm": 198.0,
      "learning_rate": 0.00015084257063655016,
      "loss": 0.9385,
      "step": 43364
    },
    {
      "epoch": 3.730964467005076,
      "grad_norm": 524.0,
      "learning_rate": 0.00015082321563955465,
      "loss": 0.7726,
      "step": 43365
    },
    {
      "epoch": 3.7310505033123977,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015080386166383314,
      "loss": 0.764,
      "step": 43366
    },
    {
      "epoch": 3.7311365396197194,
      "grad_norm": 1.078125,
      "learning_rate": 0.00015078450870944243,
      "loss": 1.7213,
      "step": 43367
    },
    {
      "epoch": 3.731222575927041,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00015076515677643927,
      "loss": 0.7519,
      "step": 43368
    },
    {
      "epoch": 3.731308612234363,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00015074580586487996,
      "loss": 1.614,
      "step": 43369
    },
    {
      "epoch": 3.7313946485416847,
      "grad_norm": 1.125,
      "learning_rate": 0.00015072645597482138,
      "loss": 1.927,
      "step": 43370
    },
    {
      "epoch": 3.7314806848490063,
      "grad_norm": 2.109375,
      "learning_rate": 0.00015070710710631986,
      "loss": 0.8989,
      "step": 43371
    },
    {
      "epoch": 3.731566721156328,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00015068775925943206,
      "loss": 1.3864,
      "step": 43372
    },
    {
      "epoch": 3.7316527574636496,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00015066841243421486,
      "loss": 1.2689,
      "step": 43373
    },
    {
      "epoch": 3.7317387937709716,
      "grad_norm": 1.125,
      "learning_rate": 0.00015064906663072442,
      "loss": 1.0973,
      "step": 43374
    },
    {
      "epoch": 3.731824830078293,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015062972184901751,
      "loss": 0.7043,
      "step": 43375
    },
    {
      "epoch": 3.731910866385615,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015061037808915086,
      "loss": 0.8866,
      "step": 43376
    },
    {
      "epoch": 3.7319969026929365,
      "grad_norm": 1.328125,
      "learning_rate": 0.00015059103535118075,
      "loss": 0.9239,
      "step": 43377
    },
    {
      "epoch": 3.732082939000258,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00015057169363516393,
      "loss": 0.9751,
      "step": 43378
    },
    {
      "epoch": 3.7321689753075797,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001505523529411571,
      "loss": 0.8832,
      "step": 43379
    },
    {
      "epoch": 3.7322550116149014,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00015053301326921649,
      "loss": 0.9845,
      "step": 43380
    },
    {
      "epoch": 3.7323410479222234,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00015051367461939885,
      "loss": 1.738,
      "step": 43381
    },
    {
      "epoch": 3.7324270842295446,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001504943369917609,
      "loss": 1.3466,
      "step": 43382
    },
    {
      "epoch": 3.7325131205368667,
      "grad_norm": 3.4375,
      "learning_rate": 0.00015047500038635893,
      "loss": 1.3905,
      "step": 43383
    },
    {
      "epoch": 3.7325991568441883,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015045566480324974,
      "loss": 0.7811,
      "step": 43384
    },
    {
      "epoch": 3.73268519315151,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00015043633024248954,
      "loss": 0.7781,
      "step": 43385
    },
    {
      "epoch": 3.7327712294588316,
      "grad_norm": 24.125,
      "learning_rate": 0.0001504169967041351,
      "loss": 1.0972,
      "step": 43386
    },
    {
      "epoch": 3.732857265766153,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001503976641882432,
      "loss": 1.3532,
      "step": 43387
    },
    {
      "epoch": 3.7329433020734752,
      "grad_norm": 2.109375,
      "learning_rate": 0.0001503783326948699,
      "loss": 1.557,
      "step": 43388
    },
    {
      "epoch": 3.7330293383807964,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00015035900222407195,
      "loss": 0.8275,
      "step": 43389
    },
    {
      "epoch": 3.7331153746881185,
      "grad_norm": 3.65625,
      "learning_rate": 0.000150339672775906,
      "loss": 1.6634,
      "step": 43390
    },
    {
      "epoch": 3.73320141099544,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00015032034435042847,
      "loss": 1.0173,
      "step": 43391
    },
    {
      "epoch": 3.7332874473027617,
      "grad_norm": 1.421875,
      "learning_rate": 0.00015030101694769605,
      "loss": 1.1801,
      "step": 43392
    },
    {
      "epoch": 3.7333734836100834,
      "grad_norm": 0.875,
      "learning_rate": 0.00015028169056776502,
      "loss": 1.3766,
      "step": 43393
    },
    {
      "epoch": 3.733459519917405,
      "grad_norm": 1.078125,
      "learning_rate": 0.00015026236521069203,
      "loss": 1.3604,
      "step": 43394
    },
    {
      "epoch": 3.733545556224727,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00015024304087653373,
      "loss": 0.9632,
      "step": 43395
    },
    {
      "epoch": 3.7336315925320487,
      "grad_norm": 102.0,
      "learning_rate": 0.00015022371756534636,
      "loss": 0.7368,
      "step": 43396
    },
    {
      "epoch": 3.7337176288393703,
      "grad_norm": 44.75,
      "learning_rate": 0.0001502043952771866,
      "loss": 1.0109,
      "step": 43397
    },
    {
      "epoch": 3.733803665146692,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001501850740121111,
      "loss": 0.929,
      "step": 43398
    },
    {
      "epoch": 3.7338897014540136,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015016575377017611,
      "loss": 0.9496,
      "step": 43399
    },
    {
      "epoch": 3.733975737761335,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00015014643455143833,
      "loss": 1.2447,
      "step": 43400
    },
    {
      "epoch": 3.734061774068657,
      "grad_norm": 4.25,
      "learning_rate": 0.0001501271163559541,
      "loss": 1.3233,
      "step": 43401
    },
    {
      "epoch": 3.734147810375979,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015010779918378,
      "loss": 0.996,
      "step": 43402
    },
    {
      "epoch": 3.7342338466833005,
      "grad_norm": 3.5625,
      "learning_rate": 0.0001500884830349727,
      "loss": 0.8276,
      "step": 43403
    },
    {
      "epoch": 3.734319882990622,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001500691679095883,
      "loss": 0.743,
      "step": 43404
    },
    {
      "epoch": 3.7344059192979437,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001500498538076836,
      "loss": 0.6283,
      "step": 43405
    },
    {
      "epoch": 3.7344919556052654,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015003054072931516,
      "loss": 0.9774,
      "step": 43406
    },
    {
      "epoch": 3.734577991912587,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00015001122867453915,
      "loss": 1.5278,
      "step": 43407
    },
    {
      "epoch": 3.7346640282199086,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014999191764341225,
      "loss": 0.585,
      "step": 43408
    },
    {
      "epoch": 3.7347500645272307,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000149972607635991,
      "loss": 0.3922,
      "step": 43409
    },
    {
      "epoch": 3.7348361008345523,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014995329865233177,
      "loss": 1.1195,
      "step": 43410
    },
    {
      "epoch": 3.734922137141874,
      "grad_norm": 1.1875,
      "learning_rate": 0.00014993399069249109,
      "loss": 0.8143,
      "step": 43411
    },
    {
      "epoch": 3.7350081734491956,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00014991468375652535,
      "loss": 0.8852,
      "step": 43412
    },
    {
      "epoch": 3.735094209756517,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00014989537784449098,
      "loss": 1.2398,
      "step": 43413
    },
    {
      "epoch": 3.735180246063839,
      "grad_norm": 1.359375,
      "learning_rate": 0.00014987607295644472,
      "loss": 1.7653,
      "step": 43414
    },
    {
      "epoch": 3.7352662823711604,
      "grad_norm": 11456.0,
      "learning_rate": 0.00014985676909244272,
      "loss": 0.8925,
      "step": 43415
    },
    {
      "epoch": 3.7353523186784825,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001498374662525416,
      "loss": 1.0088,
      "step": 43416
    },
    {
      "epoch": 3.735438354985804,
      "grad_norm": 7.84375,
      "learning_rate": 0.0001498181644367979,
      "loss": 0.7898,
      "step": 43417
    },
    {
      "epoch": 3.7355243912931257,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014979886364526785,
      "loss": 1.1466,
      "step": 43418
    },
    {
      "epoch": 3.7356104276004474,
      "grad_norm": 1.34375,
      "learning_rate": 0.00014977956387800796,
      "loss": 0.5401,
      "step": 43419
    },
    {
      "epoch": 3.735696463907769,
      "grad_norm": 8.1875,
      "learning_rate": 0.0001497602651350749,
      "loss": 1.5332,
      "step": 43420
    },
    {
      "epoch": 3.7357825002150906,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014974096741652483,
      "loss": 1.022,
      "step": 43421
    },
    {
      "epoch": 3.7358685365224122,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00014972167072241428,
      "loss": 1.492,
      "step": 43422
    },
    {
      "epoch": 3.7359545728297343,
      "grad_norm": 1.375,
      "learning_rate": 0.0001497023750527998,
      "loss": 1.4539,
      "step": 43423
    },
    {
      "epoch": 3.736040609137056,
      "grad_norm": 0.90625,
      "learning_rate": 0.00014968308040773772,
      "loss": 0.761,
      "step": 43424
    },
    {
      "epoch": 3.7361266454443776,
      "grad_norm": 9.1875,
      "learning_rate": 0.0001496637867872846,
      "loss": 1.3369,
      "step": 43425
    },
    {
      "epoch": 3.736212681751699,
      "grad_norm": 1.21875,
      "learning_rate": 0.00014964449419149656,
      "loss": 1.2923,
      "step": 43426
    },
    {
      "epoch": 3.736298718059021,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001496252026204301,
      "loss": 1.2951,
      "step": 43427
    },
    {
      "epoch": 3.7363847543663424,
      "grad_norm": 1.046875,
      "learning_rate": 0.00014960591207414215,
      "loss": 1.2695,
      "step": 43428
    },
    {
      "epoch": 3.736470790673664,
      "grad_norm": 1.0,
      "learning_rate": 0.00014958662255268857,
      "loss": 0.7476,
      "step": 43429
    },
    {
      "epoch": 3.736556826980986,
      "grad_norm": 18.5,
      "learning_rate": 0.0001495673340561261,
      "loss": 1.3087,
      "step": 43430
    },
    {
      "epoch": 3.7366428632883077,
      "grad_norm": 1.3125,
      "learning_rate": 0.00014954804658451088,
      "loss": 1.207,
      "step": 43431
    },
    {
      "epoch": 3.7367288995956294,
      "grad_norm": 2.375,
      "learning_rate": 0.00014952876013789947,
      "loss": 1.627,
      "step": 43432
    },
    {
      "epoch": 3.736814935902951,
      "grad_norm": 1424.0,
      "learning_rate": 0.00014950947471634846,
      "loss": 1.0876,
      "step": 43433
    },
    {
      "epoch": 3.7369009722102726,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014949019031991383,
      "loss": 1.6619,
      "step": 43434
    },
    {
      "epoch": 3.7369870085175947,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0001494709069486523,
      "loss": 1.1696,
      "step": 43435
    },
    {
      "epoch": 3.737073044824916,
      "grad_norm": 8.5,
      "learning_rate": 0.00014945162460262035,
      "loss": 1.7829,
      "step": 43436
    },
    {
      "epoch": 3.737159081132238,
      "grad_norm": 6.40625,
      "learning_rate": 0.000149432343281874,
      "loss": 0.9965,
      "step": 43437
    },
    {
      "epoch": 3.7372451174395596,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001494130629864699,
      "loss": 0.8703,
      "step": 43438
    },
    {
      "epoch": 3.737331153746881,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001493937837164646,
      "loss": 0.5501,
      "step": 43439
    },
    {
      "epoch": 3.737417190054203,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001493745054719141,
      "loss": 1.3557,
      "step": 43440
    },
    {
      "epoch": 3.7375032263615244,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014935522825287512,
      "loss": 1.2017,
      "step": 43441
    },
    {
      "epoch": 3.7375892626688465,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014933595205940376,
      "loss": 0.5182,
      "step": 43442
    },
    {
      "epoch": 3.7376752989761677,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014931667689155654,
      "loss": 1.6944,
      "step": 43443
    },
    {
      "epoch": 3.7377613352834897,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00014929740274939002,
      "loss": 1.0286,
      "step": 43444
    },
    {
      "epoch": 3.7378473715908114,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00014927812963296016,
      "loss": 1.1326,
      "step": 43445
    },
    {
      "epoch": 3.737933407898133,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014925885754232365,
      "loss": 0.6242,
      "step": 43446
    },
    {
      "epoch": 3.7380194442054546,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014923958647753688,
      "loss": 1.4806,
      "step": 43447
    },
    {
      "epoch": 3.7381054805127762,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00014922031643865592,
      "loss": 1.9455,
      "step": 43448
    },
    {
      "epoch": 3.7381915168200983,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001492010474257373,
      "loss": 1.3093,
      "step": 43449
    },
    {
      "epoch": 3.7382775531274195,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001491817794388376,
      "loss": 1.0752,
      "step": 43450
    },
    {
      "epoch": 3.7383635894347416,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00014916251247801282,
      "loss": 1.46,
      "step": 43451
    },
    {
      "epoch": 3.738449625742063,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014914324654331958,
      "loss": 0.3991,
      "step": 43452
    },
    {
      "epoch": 3.738535662049385,
      "grad_norm": 1.7890625,
      "learning_rate": 0.00014912398163481394,
      "loss": 1.6048,
      "step": 43453
    },
    {
      "epoch": 3.7386216983567064,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001491047177525524,
      "loss": 0.7804,
      "step": 43454
    },
    {
      "epoch": 3.738707734664028,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014908545489659153,
      "loss": 1.2029,
      "step": 43455
    },
    {
      "epoch": 3.73879377097135,
      "grad_norm": 4.78125,
      "learning_rate": 0.00014906619306698726,
      "loss": 1.7122,
      "step": 43456
    },
    {
      "epoch": 3.7388798072786718,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014904693226379612,
      "loss": 0.6973,
      "step": 43457
    },
    {
      "epoch": 3.7389658435859934,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0001490276724870746,
      "loss": 1.449,
      "step": 43458
    },
    {
      "epoch": 3.739051879893315,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014900841373687872,
      "loss": 0.498,
      "step": 43459
    },
    {
      "epoch": 3.7391379162006366,
      "grad_norm": 1.1875,
      "learning_rate": 0.00014898915601326495,
      "loss": 1.6253,
      "step": 43460
    },
    {
      "epoch": 3.7392239525079582,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00014896989931628986,
      "loss": 1.2003,
      "step": 43461
    },
    {
      "epoch": 3.73930998881528,
      "grad_norm": 1.03125,
      "learning_rate": 0.00014895064364600936,
      "loss": 1.4868,
      "step": 43462
    },
    {
      "epoch": 3.739396025122602,
      "grad_norm": 1.28125,
      "learning_rate": 0.00014893138900247987,
      "loss": 1.1494,
      "step": 43463
    },
    {
      "epoch": 3.7394820614299236,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00014891213538575788,
      "loss": 1.0885,
      "step": 43464
    },
    {
      "epoch": 3.739568097737245,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014889288279589958,
      "loss": 0.9693,
      "step": 43465
    },
    {
      "epoch": 3.739654134044567,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014887363123296149,
      "loss": 1.0408,
      "step": 43466
    },
    {
      "epoch": 3.7397401703518884,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00014885438069699953,
      "loss": 1.0369,
      "step": 43467
    },
    {
      "epoch": 3.73982620665921,
      "grad_norm": 1.0,
      "learning_rate": 0.00014883513118807025,
      "loss": 0.9753,
      "step": 43468
    },
    {
      "epoch": 3.7399122429665317,
      "grad_norm": 16.375,
      "learning_rate": 0.00014881588270623008,
      "loss": 0.8854,
      "step": 43469
    },
    {
      "epoch": 3.7399982792738538,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00014879663525153497,
      "loss": 1.4135,
      "step": 43470
    },
    {
      "epoch": 3.7400843155811754,
      "grad_norm": 24.375,
      "learning_rate": 0.00014877738882404157,
      "loss": 1.4242,
      "step": 43471
    },
    {
      "epoch": 3.740170351888497,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014875814342380585,
      "loss": 0.7232,
      "step": 43472
    },
    {
      "epoch": 3.7402563881958186,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00014873889905088418,
      "loss": 0.9026,
      "step": 43473
    },
    {
      "epoch": 3.7403424245031403,
      "grad_norm": 1.46875,
      "learning_rate": 0.00014871965570533314,
      "loss": 1.6344,
      "step": 43474
    },
    {
      "epoch": 3.740428460810462,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00014870041338720864,
      "loss": 1.2244,
      "step": 43475
    },
    {
      "epoch": 3.7405144971177835,
      "grad_norm": 32.5,
      "learning_rate": 0.00014868117209656707,
      "loss": 0.6925,
      "step": 43476
    },
    {
      "epoch": 3.7406005334251056,
      "grad_norm": 1.140625,
      "learning_rate": 0.00014866193183346487,
      "loss": 1.6372,
      "step": 43477
    },
    {
      "epoch": 3.740686569732427,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001486426925979581,
      "loss": 1.9137,
      "step": 43478
    },
    {
      "epoch": 3.740772606039749,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014862345439010306,
      "loss": 1.5552,
      "step": 43479
    },
    {
      "epoch": 3.7408586423470704,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014860421720995625,
      "loss": 1.6119,
      "step": 43480
    },
    {
      "epoch": 3.740944678654392,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001485849810575736,
      "loss": 1.1153,
      "step": 43481
    },
    {
      "epoch": 3.7410307149617137,
      "grad_norm": 2.703125,
      "learning_rate": 0.00014856574593301175,
      "loss": 0.9903,
      "step": 43482
    },
    {
      "epoch": 3.7411167512690353,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014854651183632655,
      "loss": 0.8679,
      "step": 43483
    },
    {
      "epoch": 3.7412027875763574,
      "grad_norm": 33.25,
      "learning_rate": 0.00014852727876757442,
      "loss": 1.2655,
      "step": 43484
    },
    {
      "epoch": 3.741288823883679,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014850804672681184,
      "loss": 0.8918,
      "step": 43485
    },
    {
      "epoch": 3.7413748601910006,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014848881571409466,
      "loss": 1.4943,
      "step": 43486
    },
    {
      "epoch": 3.7414608964983223,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014846958572947938,
      "loss": 1.3325,
      "step": 43487
    },
    {
      "epoch": 3.741546932805644,
      "grad_norm": 1.390625,
      "learning_rate": 0.00014845035677302232,
      "loss": 1.5622,
      "step": 43488
    },
    {
      "epoch": 3.741632969112966,
      "grad_norm": 2.375,
      "learning_rate": 0.00014843112884477944,
      "loss": 0.7869,
      "step": 43489
    },
    {
      "epoch": 3.741719005420287,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014841190194480713,
      "loss": 0.7117,
      "step": 43490
    },
    {
      "epoch": 3.741805041727609,
      "grad_norm": 29.375,
      "learning_rate": 0.00014839267607316182,
      "loss": 1.4477,
      "step": 43491
    },
    {
      "epoch": 3.741891078034931,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014837345122989938,
      "loss": 0.4898,
      "step": 43492
    },
    {
      "epoch": 3.7419771143422524,
      "grad_norm": 1.3125,
      "learning_rate": 0.00014835422741507615,
      "loss": 1.1113,
      "step": 43493
    },
    {
      "epoch": 3.742063150649574,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001483350046287486,
      "loss": 1.2405,
      "step": 43494
    },
    {
      "epoch": 3.7421491869568957,
      "grad_norm": 7.96875,
      "learning_rate": 0.00014831578287097263,
      "loss": 1.7023,
      "step": 43495
    },
    {
      "epoch": 3.7422352232642178,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014829656214180475,
      "loss": 1.1819,
      "step": 43496
    },
    {
      "epoch": 3.742321259571539,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014827734244130086,
      "loss": 1.5987,
      "step": 43497
    },
    {
      "epoch": 3.742407295878861,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001482581237695173,
      "loss": 1.0992,
      "step": 43498
    },
    {
      "epoch": 3.7424933321861826,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014823890612651052,
      "loss": 0.6179,
      "step": 43499
    },
    {
      "epoch": 3.7425793684935043,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00014821968951233639,
      "loss": 0.9661,
      "step": 43500
    },
    {
      "epoch": 3.742665404800826,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00014820047392705122,
      "loss": 0.9288,
      "step": 43501
    },
    {
      "epoch": 3.7427514411081475,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00014818125937071127,
      "loss": 1.2645,
      "step": 43502
    },
    {
      "epoch": 3.7428374774154696,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001481620458433726,
      "loss": 0.9008,
      "step": 43503
    },
    {
      "epoch": 3.7429235137227908,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00014814283334509183,
      "loss": 0.9851,
      "step": 43504
    },
    {
      "epoch": 3.743009550030113,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001481236218759246,
      "loss": 1.2744,
      "step": 43505
    },
    {
      "epoch": 3.7430955863374344,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014810441143592735,
      "loss": 1.3867,
      "step": 43506
    },
    {
      "epoch": 3.743181622644756,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001480852020251564,
      "loss": 0.7968,
      "step": 43507
    },
    {
      "epoch": 3.7432676589520777,
      "grad_norm": 1.28125,
      "learning_rate": 0.00014806599364366763,
      "loss": 1.1184,
      "step": 43508
    },
    {
      "epoch": 3.7433536952593993,
      "grad_norm": 1.375,
      "learning_rate": 0.00014804678629151742,
      "loss": 0.8553,
      "step": 43509
    },
    {
      "epoch": 3.7434397315667214,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00014802757996876205,
      "loss": 1.0201,
      "step": 43510
    },
    {
      "epoch": 3.743525767874043,
      "grad_norm": 8.75,
      "learning_rate": 0.00014800837467545736,
      "loss": 0.9426,
      "step": 43511
    },
    {
      "epoch": 3.7436118041813646,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014798917041165994,
      "loss": 0.9039,
      "step": 43512
    },
    {
      "epoch": 3.7436978404886863,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001479699671774255,
      "loss": 1.3846,
      "step": 43513
    },
    {
      "epoch": 3.743783876796008,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001479507649728105,
      "loss": 1.4948,
      "step": 43514
    },
    {
      "epoch": 3.7438699131033295,
      "grad_norm": 1.125,
      "learning_rate": 0.00014793156379787115,
      "loss": 1.0572,
      "step": 43515
    },
    {
      "epoch": 3.743955949410651,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00014791236365266342,
      "loss": 1.6559,
      "step": 43516
    },
    {
      "epoch": 3.744041985717973,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001478931645372435,
      "loss": 0.7403,
      "step": 43517
    },
    {
      "epoch": 3.744128022025295,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00014787396645166772,
      "loss": 0.7035,
      "step": 43518
    },
    {
      "epoch": 3.7442140583326164,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000147854769395992,
      "loss": 0.8184,
      "step": 43519
    },
    {
      "epoch": 3.744300094639938,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001478355733702726,
      "loss": 0.7487,
      "step": 43520
    },
    {
      "epoch": 3.7443861309472597,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014781637837456575,
      "loss": 1.2839,
      "step": 43521
    },
    {
      "epoch": 3.7444721672545813,
      "grad_norm": 1.390625,
      "learning_rate": 0.00014779718440892736,
      "loss": 1.3842,
      "step": 43522
    },
    {
      "epoch": 3.744558203561903,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014777799147341393,
      "loss": 1.4709,
      "step": 43523
    },
    {
      "epoch": 3.744644239869225,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014775879956808113,
      "loss": 1.1931,
      "step": 43524
    },
    {
      "epoch": 3.7447302761765466,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00014773960869298537,
      "loss": 0.692,
      "step": 43525
    },
    {
      "epoch": 3.7448163124838683,
      "grad_norm": 1.375,
      "learning_rate": 0.0001477204188481829,
      "loss": 1.002,
      "step": 43526
    },
    {
      "epoch": 3.74490234879119,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014770123003372953,
      "loss": 0.768,
      "step": 43527
    },
    {
      "epoch": 3.7449883850985115,
      "grad_norm": 3.96875,
      "learning_rate": 0.0001476820422496815,
      "loss": 0.7976,
      "step": 43528
    },
    {
      "epoch": 3.745074421405833,
      "grad_norm": 24.5,
      "learning_rate": 0.0001476628554960952,
      "loss": 1.7132,
      "step": 43529
    },
    {
      "epoch": 3.7451604577131548,
      "grad_norm": 1.171875,
      "learning_rate": 0.00014764366977302636,
      "loss": 1.3856,
      "step": 43530
    },
    {
      "epoch": 3.745246494020477,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014762448508053123,
      "loss": 1.384,
      "step": 43531
    },
    {
      "epoch": 3.7453325303277984,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014760530141866613,
      "loss": 0.3945,
      "step": 43532
    },
    {
      "epoch": 3.74541856663512,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001475861187874868,
      "loss": 0.7878,
      "step": 43533
    },
    {
      "epoch": 3.7455046029424417,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00014756693718704955,
      "loss": 1.1243,
      "step": 43534
    },
    {
      "epoch": 3.7455906392497633,
      "grad_norm": 1.0,
      "learning_rate": 0.0001475477566174106,
      "loss": 0.7981,
      "step": 43535
    },
    {
      "epoch": 3.745676675557085,
      "grad_norm": 7.03125,
      "learning_rate": 0.00014752857707862576,
      "loss": 1.7137,
      "step": 43536
    },
    {
      "epoch": 3.7457627118644066,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00014750939857075148,
      "loss": 1.0167,
      "step": 43537
    },
    {
      "epoch": 3.7458487481717286,
      "grad_norm": 76800.0,
      "learning_rate": 0.0001474902210938433,
      "loss": 1.1652,
      "step": 43538
    },
    {
      "epoch": 3.7459347844790503,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014747104464795798,
      "loss": 1.0655,
      "step": 43539
    },
    {
      "epoch": 3.746020820786372,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014745186923315114,
      "loss": 0.7593,
      "step": 43540
    },
    {
      "epoch": 3.7461068570936935,
      "grad_norm": 22.375,
      "learning_rate": 0.00014743269484947903,
      "loss": 1.3826,
      "step": 43541
    },
    {
      "epoch": 3.746192893401015,
      "grad_norm": 1.125,
      "learning_rate": 0.0001474135214969978,
      "loss": 0.917,
      "step": 43542
    },
    {
      "epoch": 3.7462789297083368,
      "grad_norm": 1.0625,
      "learning_rate": 0.00014739434917576338,
      "loss": 1.4035,
      "step": 43543
    },
    {
      "epoch": 3.7463649660156584,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014737517788583187,
      "loss": 1.1232,
      "step": 43544
    },
    {
      "epoch": 3.7464510023229805,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001473560076272596,
      "loss": 1.4701,
      "step": 43545
    },
    {
      "epoch": 3.746537038630302,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001473368384001022,
      "loss": 0.663,
      "step": 43546
    },
    {
      "epoch": 3.7466230749376237,
      "grad_norm": 0.875,
      "learning_rate": 0.00014731767020441594,
      "loss": 1.7539,
      "step": 43547
    },
    {
      "epoch": 3.7467091112449453,
      "grad_norm": 1.765625,
      "learning_rate": 0.0001472985030402571,
      "loss": 0.8207,
      "step": 43548
    },
    {
      "epoch": 3.746795147552267,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014727933690768136,
      "loss": 1.1548,
      "step": 43549
    },
    {
      "epoch": 3.746881183859589,
      "grad_norm": 1.0,
      "learning_rate": 0.00014726017180674502,
      "loss": 1.0173,
      "step": 43550
    },
    {
      "epoch": 3.74696722016691,
      "grad_norm": 1.03125,
      "learning_rate": 0.00014724100773750415,
      "loss": 1.2777,
      "step": 43551
    },
    {
      "epoch": 3.7470532564742323,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014722184470001458,
      "loss": 0.8141,
      "step": 43552
    },
    {
      "epoch": 3.747139292781554,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014720268269433274,
      "loss": 1.1042,
      "step": 43553
    },
    {
      "epoch": 3.7472253290888755,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014718352172051418,
      "loss": 0.6989,
      "step": 43554
    },
    {
      "epoch": 3.747311365396197,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00014716436177861526,
      "loss": 1.367,
      "step": 43555
    },
    {
      "epoch": 3.7473974017035188,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001471452028686921,
      "loss": 1.3368,
      "step": 43556
    },
    {
      "epoch": 3.747483438010841,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014712604499080045,
      "loss": 1.6516,
      "step": 43557
    },
    {
      "epoch": 3.747569474318162,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014710688814499644,
      "loss": 0.9048,
      "step": 43558
    },
    {
      "epoch": 3.747655510625484,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00014708773233133637,
      "loss": 1.2384,
      "step": 43559
    },
    {
      "epoch": 3.7477415469328057,
      "grad_norm": 1.40625,
      "learning_rate": 0.00014706857754987585,
      "loss": 1.3858,
      "step": 43560
    },
    {
      "epoch": 3.7478275832401273,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001470494238006711,
      "loss": 1.4153,
      "step": 43561
    },
    {
      "epoch": 3.747913619547449,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014703027108377832,
      "loss": 1.3165,
      "step": 43562
    },
    {
      "epoch": 3.7479996558547706,
      "grad_norm": 73.0,
      "learning_rate": 0.0001470111193992532,
      "loss": 1.8271,
      "step": 43563
    },
    {
      "epoch": 3.7480856921620926,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014699196874715204,
      "loss": 1.3553,
      "step": 43564
    },
    {
      "epoch": 3.748171728469414,
      "grad_norm": 1.265625,
      "learning_rate": 0.00014697281912753052,
      "loss": 1.6253,
      "step": 43565
    },
    {
      "epoch": 3.748257764776736,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001469536705404449,
      "loss": 0.7931,
      "step": 43566
    },
    {
      "epoch": 3.7483438010840575,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014693452298595127,
      "loss": 1.1068,
      "step": 43567
    },
    {
      "epoch": 3.748429837391379,
      "grad_norm": 1.328125,
      "learning_rate": 0.00014691537646410534,
      "loss": 0.7831,
      "step": 43568
    },
    {
      "epoch": 3.7485158736987008,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014689623097496323,
      "loss": 0.4926,
      "step": 43569
    },
    {
      "epoch": 3.7486019100060224,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00014687708651858118,
      "loss": 1.3374,
      "step": 43570
    },
    {
      "epoch": 3.7486879463133445,
      "grad_norm": 66.5,
      "learning_rate": 0.00014685794309501476,
      "loss": 0.5938,
      "step": 43571
    },
    {
      "epoch": 3.748773982620666,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001468388007043202,
      "loss": 1.3313,
      "step": 43572
    },
    {
      "epoch": 3.7488600189279877,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001468196593465536,
      "loss": 1.1906,
      "step": 43573
    },
    {
      "epoch": 3.7489460552353093,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014680051902177073,
      "loss": 0.8675,
      "step": 43574
    },
    {
      "epoch": 3.749032091542631,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014678137973002758,
      "loss": 1.3503,
      "step": 43575
    },
    {
      "epoch": 3.7491181278499526,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00014676224147138018,
      "loss": 1.717,
      "step": 43576
    },
    {
      "epoch": 3.749204164157274,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014674310424588456,
      "loss": 1.5189,
      "step": 43577
    },
    {
      "epoch": 3.7492902004645963,
      "grad_norm": 1.78125,
      "learning_rate": 0.00014672396805359684,
      "loss": 1.4901,
      "step": 43578
    },
    {
      "epoch": 3.749376236771918,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001467048328945726,
      "loss": 1.1076,
      "step": 43579
    },
    {
      "epoch": 3.7494622730792395,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000146685698768868,
      "loss": 1.2767,
      "step": 43580
    },
    {
      "epoch": 3.749548309386561,
      "grad_norm": 35.5,
      "learning_rate": 0.0001466665656765392,
      "loss": 0.8539,
      "step": 43581
    },
    {
      "epoch": 3.7496343456938828,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014664743361764181,
      "loss": 0.9467,
      "step": 43582
    },
    {
      "epoch": 3.7497203820012044,
      "grad_norm": 1.0,
      "learning_rate": 0.00014662830259223208,
      "loss": 0.8064,
      "step": 43583
    },
    {
      "epoch": 3.749806418308526,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014660917260036577,
      "loss": 0.6106,
      "step": 43584
    },
    {
      "epoch": 3.749892454615848,
      "grad_norm": 1.609375,
      "learning_rate": 0.00014659004364209882,
      "loss": 0.958,
      "step": 43585
    },
    {
      "epoch": 3.7499784909231697,
      "grad_norm": 1.8828125,
      "learning_rate": 0.00014657091571748743,
      "loss": 0.9595,
      "step": 43586
    },
    {
      "epoch": 3.7500645272304913,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001465517888265872,
      "loss": 1.1298,
      "step": 43587
    },
    {
      "epoch": 3.750150563537813,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001465326629694543,
      "loss": 1.2756,
      "step": 43588
    },
    {
      "epoch": 3.7502365998451346,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014651353814614478,
      "loss": 1.4153,
      "step": 43589
    },
    {
      "epoch": 3.750322636152456,
      "grad_norm": 5248.0,
      "learning_rate": 0.00014649441435671417,
      "loss": 1.3506,
      "step": 43590
    },
    {
      "epoch": 3.750322636152456,
      "eval_loss": 1.0737842321395874,
      "eval_runtime": 356.2732,
      "eval_samples_per_second": 13.049,
      "eval_steps_per_second": 1.634,
      "step": 43590
    },
    {
      "epoch": 3.750408672459778,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0001464752916012187,
      "loss": 1.812,
      "step": 43591
    },
    {
      "epoch": 3.7504947087671,
      "grad_norm": 5.78125,
      "learning_rate": 0.00014645616987971438,
      "loss": 1.345,
      "step": 43592
    },
    {
      "epoch": 3.7505807450744215,
      "grad_norm": 1.296875,
      "learning_rate": 0.00014643704919225687,
      "loss": 1.1533,
      "step": 43593
    },
    {
      "epoch": 3.750666781381743,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00014641792953890237,
      "loss": 1.5248,
      "step": 43594
    },
    {
      "epoch": 3.7507528176890648,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00014639881091970652,
      "loss": 1.654,
      "step": 43595
    },
    {
      "epoch": 3.7508388539963864,
      "grad_norm": 1.296875,
      "learning_rate": 0.00014637969333472533,
      "loss": 1.1033,
      "step": 43596
    },
    {
      "epoch": 3.750924890303708,
      "grad_norm": 1.234375,
      "learning_rate": 0.00014636057678401497,
      "loss": 1.5523,
      "step": 43597
    },
    {
      "epoch": 3.7510109266110296,
      "grad_norm": 1.140625,
      "learning_rate": 0.00014634146126763097,
      "loss": 1.8858,
      "step": 43598
    },
    {
      "epoch": 3.7510969629183517,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001463223467856294,
      "loss": 0.8709,
      "step": 43599
    },
    {
      "epoch": 3.7511829992256733,
      "grad_norm": 13120.0,
      "learning_rate": 0.0001463032333380664,
      "loss": 1.1104,
      "step": 43600
    },
    {
      "epoch": 3.751269035532995,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001462841209249974,
      "loss": 1.5982,
      "step": 43601
    },
    {
      "epoch": 3.7513550718403166,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001462650095464786,
      "loss": 1.3949,
      "step": 43602
    },
    {
      "epoch": 3.751441108147638,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00014624589920256598,
      "loss": 1.0965,
      "step": 43603
    },
    {
      "epoch": 3.75152714445496,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014622678989331518,
      "loss": 1.2895,
      "step": 43604
    },
    {
      "epoch": 3.7516131807622815,
      "grad_norm": 4320.0,
      "learning_rate": 0.00014620768161878234,
      "loss": 0.4949,
      "step": 43605
    },
    {
      "epoch": 3.7516992170696035,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001461885743790231,
      "loss": 1.0025,
      "step": 43606
    },
    {
      "epoch": 3.751785253376925,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014616946817409344,
      "loss": 1.0174,
      "step": 43607
    },
    {
      "epoch": 3.7518712896842468,
      "grad_norm": 5.9375,
      "learning_rate": 0.00014615036300404943,
      "loss": 1.1149,
      "step": 43608
    },
    {
      "epoch": 3.7519573259915684,
      "grad_norm": 1.265625,
      "learning_rate": 0.00014613125886894668,
      "loss": 1.0964,
      "step": 43609
    },
    {
      "epoch": 3.75204336229889,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014611215576884113,
      "loss": 0.9461,
      "step": 43610
    },
    {
      "epoch": 3.752129398606212,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00014609305370378867,
      "loss": 1.3115,
      "step": 43611
    },
    {
      "epoch": 3.7522154349135333,
      "grad_norm": 1.203125,
      "learning_rate": 0.00014607395267384526,
      "loss": 0.8987,
      "step": 43612
    },
    {
      "epoch": 3.7523014712208553,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014605485267906682,
      "loss": 0.8487,
      "step": 43613
    },
    {
      "epoch": 3.752387507528177,
      "grad_norm": 0.90625,
      "learning_rate": 0.00014603575371950896,
      "loss": 0.9279,
      "step": 43614
    },
    {
      "epoch": 3.7524735438354986,
      "grad_norm": 1.390625,
      "learning_rate": 0.00014601665579522767,
      "loss": 1.275,
      "step": 43615
    },
    {
      "epoch": 3.75255958014282,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014599755890627898,
      "loss": 1.1591,
      "step": 43616
    },
    {
      "epoch": 3.752645616450142,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014597846305271838,
      "loss": 1.1166,
      "step": 43617
    },
    {
      "epoch": 3.752731652757464,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014595936823460194,
      "loss": 0.9355,
      "step": 43618
    },
    {
      "epoch": 3.752817689064785,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014594027445198566,
      "loss": 0.7308,
      "step": 43619
    },
    {
      "epoch": 3.752903725372107,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001459211817049251,
      "loss": 1.7195,
      "step": 43620
    },
    {
      "epoch": 3.7529897616794288,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014590208999347616,
      "loss": 1.1829,
      "step": 43621
    },
    {
      "epoch": 3.7530757979867504,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00014588299931769493,
      "loss": 0.6121,
      "step": 43622
    },
    {
      "epoch": 3.753161834294072,
      "grad_norm": 9.8125,
      "learning_rate": 0.00014586390967763686,
      "loss": 0.8067,
      "step": 43623
    },
    {
      "epoch": 3.7532478706013936,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00014584482107335816,
      "loss": 1.1023,
      "step": 43624
    },
    {
      "epoch": 3.7533339069087157,
      "grad_norm": 12.125,
      "learning_rate": 0.00014582573350491434,
      "loss": 1.3433,
      "step": 43625
    },
    {
      "epoch": 3.753419943216037,
      "grad_norm": 1.171875,
      "learning_rate": 0.00014580664697236135,
      "loss": 1.462,
      "step": 43626
    },
    {
      "epoch": 3.753505979523359,
      "grad_norm": 1.140625,
      "learning_rate": 0.00014578756147575516,
      "loss": 0.96,
      "step": 43627
    },
    {
      "epoch": 3.7535920158306806,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014576847701515134,
      "loss": 0.8214,
      "step": 43628
    },
    {
      "epoch": 3.753678052138002,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014574939359060586,
      "loss": 1.2536,
      "step": 43629
    },
    {
      "epoch": 3.753764088445324,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00014573031120217463,
      "loss": 1.5599,
      "step": 43630
    },
    {
      "epoch": 3.7538501247526455,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014571122984991314,
      "loss": 1.4506,
      "step": 43631
    },
    {
      "epoch": 3.7539361610599675,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00014569214953387744,
      "loss": 1.4827,
      "step": 43632
    },
    {
      "epoch": 3.754022197367289,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014567307025412345,
      "loss": 0.9712,
      "step": 43633
    },
    {
      "epoch": 3.7541082336746108,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001456539920107066,
      "loss": 0.6353,
      "step": 43634
    },
    {
      "epoch": 3.7541942699819324,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014563491480368317,
      "loss": 1.0145,
      "step": 43635
    },
    {
      "epoch": 3.754280306289254,
      "grad_norm": 1.265625,
      "learning_rate": 0.00014561583863310847,
      "loss": 1.7848,
      "step": 43636
    },
    {
      "epoch": 3.7543663425965756,
      "grad_norm": 1.265625,
      "learning_rate": 0.00014559676349903855,
      "loss": 0.8886,
      "step": 43637
    },
    {
      "epoch": 3.7544523789038973,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00014557768940152928,
      "loss": 0.9798,
      "step": 43638
    },
    {
      "epoch": 3.7545384152112193,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00014555861634063622,
      "loss": 0.7022,
      "step": 43639
    },
    {
      "epoch": 3.754624451518541,
      "grad_norm": 74.0,
      "learning_rate": 0.00014553954431641524,
      "loss": 1.0412,
      "step": 43640
    },
    {
      "epoch": 3.7547104878258626,
      "grad_norm": 0.84375,
      "learning_rate": 0.00014552047332892236,
      "loss": 1.1,
      "step": 43641
    },
    {
      "epoch": 3.754796524133184,
      "grad_norm": 1.046875,
      "learning_rate": 0.000145501403378213,
      "loss": 0.6484,
      "step": 43642
    },
    {
      "epoch": 3.754882560440506,
      "grad_norm": 1.015625,
      "learning_rate": 0.000145482334464343,
      "loss": 1.5468,
      "step": 43643
    },
    {
      "epoch": 3.7549685967478275,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014546326658736847,
      "loss": 1.2094,
      "step": 43644
    },
    {
      "epoch": 3.755054633055149,
      "grad_norm": 1.25,
      "learning_rate": 0.00014544419974734475,
      "loss": 0.8648,
      "step": 43645
    },
    {
      "epoch": 3.755140669362471,
      "grad_norm": 67.5,
      "learning_rate": 0.00014542513394432777,
      "loss": 0.8248,
      "step": 43646
    },
    {
      "epoch": 3.7552267056697928,
      "grad_norm": 1.046875,
      "learning_rate": 0.00014540606917837345,
      "loss": 0.9214,
      "step": 43647
    },
    {
      "epoch": 3.7553127419771144,
      "grad_norm": 1.484375,
      "learning_rate": 0.00014538700544953714,
      "loss": 1.987,
      "step": 43648
    },
    {
      "epoch": 3.755398778284436,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001453679427578752,
      "loss": 1.0693,
      "step": 43649
    },
    {
      "epoch": 3.7554848145917576,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00014534888110344284,
      "loss": 0.9818,
      "step": 43650
    },
    {
      "epoch": 3.7555708508990793,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00014532982048629623,
      "loss": 0.5838,
      "step": 43651
    },
    {
      "epoch": 3.755656887206401,
      "grad_norm": 1.125,
      "learning_rate": 0.00014531076090649076,
      "loss": 1.7218,
      "step": 43652
    },
    {
      "epoch": 3.755742923513723,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0001452917023640823,
      "loss": 1.3841,
      "step": 43653
    },
    {
      "epoch": 3.7558289598210446,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001452726448591268,
      "loss": 0.9238,
      "step": 43654
    },
    {
      "epoch": 3.755914996128366,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014525358839167962,
      "loss": 1.4584,
      "step": 43655
    },
    {
      "epoch": 3.756001032435688,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001452345329617968,
      "loss": 0.9845,
      "step": 43656
    },
    {
      "epoch": 3.7560870687430095,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00014521547856953404,
      "loss": 0.7961,
      "step": 43657
    },
    {
      "epoch": 3.756173105050331,
      "grad_norm": 1.40625,
      "learning_rate": 0.00014519642521494685,
      "loss": 1.1822,
      "step": 43658
    },
    {
      "epoch": 3.7562591413576527,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014517737289809112,
      "loss": 0.7005,
      "step": 43659
    },
    {
      "epoch": 3.7563451776649748,
      "grad_norm": 0.9375,
      "learning_rate": 0.00014515832161902275,
      "loss": 0.9876,
      "step": 43660
    },
    {
      "epoch": 3.7564312139722964,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014513927137779715,
      "loss": 1.0731,
      "step": 43661
    },
    {
      "epoch": 3.756517250279618,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00014512022217447014,
      "loss": 1.2145,
      "step": 43662
    },
    {
      "epoch": 3.7566032865869396,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00014510117400909757,
      "loss": 0.8804,
      "step": 43663
    },
    {
      "epoch": 3.7566893228942613,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001450821268817349,
      "loss": 1.0871,
      "step": 43664
    },
    {
      "epoch": 3.7567753592015833,
      "grad_norm": 32.75,
      "learning_rate": 0.00014506308079243819,
      "loss": 1.0961,
      "step": 43665
    },
    {
      "epoch": 3.7568613955089045,
      "grad_norm": 1.125,
      "learning_rate": 0.00014504403574126275,
      "loss": 1.5063,
      "step": 43666
    },
    {
      "epoch": 3.7569474318162266,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014502499172826445,
      "loss": 1.9001,
      "step": 43667
    },
    {
      "epoch": 3.757033468123548,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014500594875349916,
      "loss": 0.8334,
      "step": 43668
    },
    {
      "epoch": 3.75711950443087,
      "grad_norm": 1.578125,
      "learning_rate": 0.00014498690681702232,
      "loss": 0.865,
      "step": 43669
    },
    {
      "epoch": 3.7572055407381915,
      "grad_norm": 1.03125,
      "learning_rate": 0.00014496786591888968,
      "loss": 1.2168,
      "step": 43670
    },
    {
      "epoch": 3.757291577045513,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014494882605915716,
      "loss": 0.5257,
      "step": 43671
    },
    {
      "epoch": 3.757377613352835,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014492978723788013,
      "loss": 1.6197,
      "step": 43672
    },
    {
      "epoch": 3.7574636496601563,
      "grad_norm": 132.0,
      "learning_rate": 0.0001449107494551144,
      "loss": 1.0649,
      "step": 43673
    },
    {
      "epoch": 3.7575496859674784,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014489171271091583,
      "loss": 0.4861,
      "step": 43674
    },
    {
      "epoch": 3.7576357222748,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014487267700533975,
      "loss": 0.7276,
      "step": 43675
    },
    {
      "epoch": 3.7577217585821217,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014485364233844222,
      "loss": 1.826,
      "step": 43676
    },
    {
      "epoch": 3.7578077948894433,
      "grad_norm": 278.0,
      "learning_rate": 0.00014483460871027854,
      "loss": 0.9041,
      "step": 43677
    },
    {
      "epoch": 3.757893831196765,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014481557612090458,
      "loss": 0.923,
      "step": 43678
    },
    {
      "epoch": 3.757979867504087,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014479654457037606,
      "loss": 1.4842,
      "step": 43679
    },
    {
      "epoch": 3.758065903811408,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00014477751405874846,
      "loss": 0.9993,
      "step": 43680
    },
    {
      "epoch": 3.75815194011873,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014475848458607755,
      "loss": 1.5369,
      "step": 43681
    },
    {
      "epoch": 3.758237976426052,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00014473945615241913,
      "loss": 0.6133,
      "step": 43682
    },
    {
      "epoch": 3.7583240127333735,
      "grad_norm": 41.25,
      "learning_rate": 0.00014472042875782854,
      "loss": 1.2366,
      "step": 43683
    },
    {
      "epoch": 3.758410049040695,
      "grad_norm": 1.046875,
      "learning_rate": 0.00014470140240236163,
      "loss": 0.8814,
      "step": 43684
    },
    {
      "epoch": 3.7584960853480167,
      "grad_norm": 1.296875,
      "learning_rate": 0.00014468237708607396,
      "loss": 1.0163,
      "step": 43685
    },
    {
      "epoch": 3.758582121655339,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00014466335280902126,
      "loss": 1.082,
      "step": 43686
    },
    {
      "epoch": 3.7586681579626604,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014464432957125934,
      "loss": 0.4129,
      "step": 43687
    },
    {
      "epoch": 3.758754194269982,
      "grad_norm": 5.15625,
      "learning_rate": 0.00014462530737284346,
      "loss": 0.8931,
      "step": 43688
    },
    {
      "epoch": 3.7588402305773037,
      "grad_norm": 38.75,
      "learning_rate": 0.00014460628621382944,
      "loss": 1.2097,
      "step": 43689
    },
    {
      "epoch": 3.7589262668846253,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00014458726609427303,
      "loss": 1.5252,
      "step": 43690
    },
    {
      "epoch": 3.759012303191947,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001445682470142296,
      "loss": 1.5018,
      "step": 43691
    },
    {
      "epoch": 3.7590983394992685,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001445492289737551,
      "loss": 1.2127,
      "step": 43692
    },
    {
      "epoch": 3.7591843758065906,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014453021197290483,
      "loss": 1.0585,
      "step": 43693
    },
    {
      "epoch": 3.759270412113912,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001445111960117345,
      "loss": 1.0397,
      "step": 43694
    },
    {
      "epoch": 3.759356448421234,
      "grad_norm": 1.3125,
      "learning_rate": 0.00014449218109029998,
      "loss": 1.5361,
      "step": 43695
    },
    {
      "epoch": 3.7594424847285555,
      "grad_norm": 1.171875,
      "learning_rate": 0.00014447316720865656,
      "loss": 0.9832,
      "step": 43696
    },
    {
      "epoch": 3.759528521035877,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014445415436685995,
      "loss": 1.0949,
      "step": 43697
    },
    {
      "epoch": 3.7596145573431987,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00014443514256496593,
      "loss": 0.8246,
      "step": 43698
    },
    {
      "epoch": 3.7597005936505203,
      "grad_norm": 1.171875,
      "learning_rate": 0.00014441613180302983,
      "loss": 1.5102,
      "step": 43699
    },
    {
      "epoch": 3.7597866299578424,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00014439712208110734,
      "loss": 1.554,
      "step": 43700
    },
    {
      "epoch": 3.759872666265164,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014437811339925433,
      "loss": 1.4484,
      "step": 43701
    },
    {
      "epoch": 3.7599587025724857,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00014435910575752596,
      "loss": 1.7711,
      "step": 43702
    },
    {
      "epoch": 3.7600447388798073,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014434009915597807,
      "loss": 1.4084,
      "step": 43703
    },
    {
      "epoch": 3.760130775187129,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014432109359466638,
      "loss": 0.9263,
      "step": 43704
    },
    {
      "epoch": 3.7602168114944505,
      "grad_norm": 1.25,
      "learning_rate": 0.00014430208907364618,
      "loss": 0.7565,
      "step": 43705
    },
    {
      "epoch": 3.760302847801772,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014428308559297326,
      "loss": 0.8844,
      "step": 43706
    },
    {
      "epoch": 3.760388884109094,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014426408315270305,
      "loss": 0.4542,
      "step": 43707
    },
    {
      "epoch": 3.760474920416416,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001442450817528912,
      "loss": 0.4966,
      "step": 43708
    },
    {
      "epoch": 3.7605609567237375,
      "grad_norm": 1.515625,
      "learning_rate": 0.00014422608139359346,
      "loss": 0.7355,
      "step": 43709
    },
    {
      "epoch": 3.760646993031059,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014420708207486504,
      "loss": 1.3907,
      "step": 43710
    },
    {
      "epoch": 3.7607330293383807,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014418808379676175,
      "loss": 0.9065,
      "step": 43711
    },
    {
      "epoch": 3.7608190656457023,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014416908655933924,
      "loss": 0.8499,
      "step": 43712
    },
    {
      "epoch": 3.760905101953024,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00014415009036265275,
      "loss": 0.7741,
      "step": 43713
    },
    {
      "epoch": 3.760991138260346,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001441310952067581,
      "loss": 0.6037,
      "step": 43714
    },
    {
      "epoch": 3.7610771745676677,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014411210109171096,
      "loss": 1.4415,
      "step": 43715
    },
    {
      "epoch": 3.7611632108749893,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00014409310801756647,
      "loss": 0.8717,
      "step": 43716
    },
    {
      "epoch": 3.761249247182311,
      "grad_norm": 1.296875,
      "learning_rate": 0.00014407411598438063,
      "loss": 0.9803,
      "step": 43717
    },
    {
      "epoch": 3.7613352834896325,
      "grad_norm": 23.0,
      "learning_rate": 0.00014405512499220864,
      "loss": 1.179,
      "step": 43718
    },
    {
      "epoch": 3.761421319796954,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00014403613504110612,
      "loss": 0.7206,
      "step": 43719
    },
    {
      "epoch": 3.761507356104276,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00014401714613112887,
      "loss": 0.8776,
      "step": 43720
    },
    {
      "epoch": 3.761593392411598,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00014399815826233209,
      "loss": 1.1666,
      "step": 43721
    },
    {
      "epoch": 3.7616794287189195,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00014397917143477146,
      "loss": 1.5073,
      "step": 43722
    },
    {
      "epoch": 3.761765465026241,
      "grad_norm": 0.890625,
      "learning_rate": 0.00014396018564850248,
      "loss": 1.38,
      "step": 43723
    },
    {
      "epoch": 3.7618515013335627,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00014394120090358071,
      "loss": 1.3826,
      "step": 43724
    },
    {
      "epoch": 3.7619375376408843,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014392221720006189,
      "loss": 1.018,
      "step": 43725
    },
    {
      "epoch": 3.7620235739482064,
      "grad_norm": 1.125,
      "learning_rate": 0.00014390323453800108,
      "loss": 1.6754,
      "step": 43726
    },
    {
      "epoch": 3.7621096102555276,
      "grad_norm": 3.1875,
      "learning_rate": 0.00014388425291745412,
      "loss": 0.6259,
      "step": 43727
    },
    {
      "epoch": 3.7621956465628497,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00014386527233847663,
      "loss": 0.9476,
      "step": 43728
    },
    {
      "epoch": 3.7622816828701713,
      "grad_norm": 2.09375,
      "learning_rate": 0.00014384629280112378,
      "loss": 1.0052,
      "step": 43729
    },
    {
      "epoch": 3.762367719177493,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001438273143054512,
      "loss": 1.0999,
      "step": 43730
    },
    {
      "epoch": 3.7624537554848145,
      "grad_norm": 1.375,
      "learning_rate": 0.0001438083368515147,
      "loss": 1.5305,
      "step": 43731
    },
    {
      "epoch": 3.762539791792136,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001437893604393693,
      "loss": 1.2037,
      "step": 43732
    },
    {
      "epoch": 3.7626258280994582,
      "grad_norm": 0.75,
      "learning_rate": 0.00014377038506907075,
      "loss": 0.4739,
      "step": 43733
    },
    {
      "epoch": 3.7627118644067794,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014375141074067478,
      "loss": 0.5208,
      "step": 43734
    },
    {
      "epoch": 3.7627979007141015,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014373243745423642,
      "loss": 0.7468,
      "step": 43735
    },
    {
      "epoch": 3.762883937021423,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00014371346520981155,
      "loss": 1.6953,
      "step": 43736
    },
    {
      "epoch": 3.7629699733287447,
      "grad_norm": 9.0625,
      "learning_rate": 0.00014369449400745537,
      "loss": 1.3986,
      "step": 43737
    },
    {
      "epoch": 3.7630560096360663,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014367552384722348,
      "loss": 0.9704,
      "step": 43738
    },
    {
      "epoch": 3.763142045943388,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00014365655472917154,
      "loss": 0.7762,
      "step": 43739
    },
    {
      "epoch": 3.76322808225071,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001436375866533547,
      "loss": 1.9624,
      "step": 43740
    },
    {
      "epoch": 3.763314118558031,
      "grad_norm": 1.140625,
      "learning_rate": 0.00014361861961982865,
      "loss": 0.9197,
      "step": 43741
    },
    {
      "epoch": 3.7634001548653533,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014359965362864896,
      "loss": 0.8236,
      "step": 43742
    },
    {
      "epoch": 3.763486191172675,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00014358068867987078,
      "loss": 0.6514,
      "step": 43743
    },
    {
      "epoch": 3.7635722274799965,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014356172477354983,
      "loss": 1.3435,
      "step": 43744
    },
    {
      "epoch": 3.763658263787318,
      "grad_norm": 0.984375,
      "learning_rate": 0.00014354276190974163,
      "loss": 0.8757,
      "step": 43745
    },
    {
      "epoch": 3.76374430009464,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00014352380008850137,
      "loss": 1.0076,
      "step": 43746
    },
    {
      "epoch": 3.763830336401962,
      "grad_norm": 10.8125,
      "learning_rate": 0.00014350483930988483,
      "loss": 0.819,
      "step": 43747
    },
    {
      "epoch": 3.7639163727092835,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014348587957394715,
      "loss": 1.1827,
      "step": 43748
    },
    {
      "epoch": 3.764002409016605,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00014346692088074392,
      "loss": 1.4937,
      "step": 43749
    },
    {
      "epoch": 3.7640884453239267,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00014344796323033073,
      "loss": 1.6875,
      "step": 43750
    },
    {
      "epoch": 3.7641744816312483,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014342900662276282,
      "loss": 1.143,
      "step": 43751
    },
    {
      "epoch": 3.76426051793857,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00014341005105809562,
      "loss": 1.0414,
      "step": 43752
    },
    {
      "epoch": 3.7643465542458916,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00014339109653638488,
      "loss": 1.3111,
      "step": 43753
    },
    {
      "epoch": 3.7644325905532137,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001433721430576857,
      "loss": 1.3114,
      "step": 43754
    },
    {
      "epoch": 3.7645186268605353,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001433531906220536,
      "loss": 0.6149,
      "step": 43755
    },
    {
      "epoch": 3.764604663167857,
      "grad_norm": 71.0,
      "learning_rate": 0.0001433342392295442,
      "loss": 1.0436,
      "step": 43756
    },
    {
      "epoch": 3.7646906994751785,
      "grad_norm": 1.734375,
      "learning_rate": 0.00014331528888021267,
      "loss": 1.7867,
      "step": 43757
    },
    {
      "epoch": 3.7647767357825,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001432963395741147,
      "loss": 0.8981,
      "step": 43758
    },
    {
      "epoch": 3.764862772089822,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001432773913113052,
      "loss": 1.2373,
      "step": 43759
    },
    {
      "epoch": 3.7649488083971434,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014325844409184013,
      "loss": 0.9411,
      "step": 43760
    },
    {
      "epoch": 3.7650348447044655,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014323949791577488,
      "loss": 0.9703,
      "step": 43761
    },
    {
      "epoch": 3.765120881011787,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001432205527831646,
      "loss": 1.351,
      "step": 43762
    },
    {
      "epoch": 3.7652069173191087,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014320160869406496,
      "loss": 0.606,
      "step": 43763
    },
    {
      "epoch": 3.7652929536264304,
      "grad_norm": 1.046875,
      "learning_rate": 0.000143182665648531,
      "loss": 1.541,
      "step": 43764
    },
    {
      "epoch": 3.765378989933752,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001431637236466184,
      "loss": 0.7523,
      "step": 43765
    },
    {
      "epoch": 3.7654650262410736,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014314478268838266,
      "loss": 1.6134,
      "step": 43766
    },
    {
      "epoch": 3.7655510625483952,
      "grad_norm": 3.515625,
      "learning_rate": 0.00014312584277387886,
      "loss": 1.6242,
      "step": 43767
    },
    {
      "epoch": 3.7656370988557173,
      "grad_norm": 1.328125,
      "learning_rate": 0.00014310690390316255,
      "loss": 0.8163,
      "step": 43768
    },
    {
      "epoch": 3.765723135163039,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014308796607628933,
      "loss": 0.5853,
      "step": 43769
    },
    {
      "epoch": 3.7658091714703605,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014306902929331423,
      "loss": 0.6133,
      "step": 43770
    },
    {
      "epoch": 3.765895207777682,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00014305009355429277,
      "loss": 1.4168,
      "step": 43771
    },
    {
      "epoch": 3.765981244085004,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00014303115885928053,
      "loss": 0.9665,
      "step": 43772
    },
    {
      "epoch": 3.7660672803923254,
      "grad_norm": 0.875,
      "learning_rate": 0.00014301222520833257,
      "loss": 0.6033,
      "step": 43773
    },
    {
      "epoch": 3.766153316699647,
      "grad_norm": 5.65625,
      "learning_rate": 0.0001429932926015044,
      "loss": 1.4427,
      "step": 43774
    },
    {
      "epoch": 3.766239353006969,
      "grad_norm": 2.015625,
      "learning_rate": 0.0001429743610388516,
      "loss": 1.1447,
      "step": 43775
    },
    {
      "epoch": 3.7663253893142907,
      "grad_norm": 48.5,
      "learning_rate": 0.0001429554305204292,
      "loss": 1.0113,
      "step": 43776
    },
    {
      "epoch": 3.7664114256216124,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014293650104629286,
      "loss": 1.0302,
      "step": 43777
    },
    {
      "epoch": 3.766497461928934,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00014291757261649768,
      "loss": 0.9187,
      "step": 43778
    },
    {
      "epoch": 3.7665834982362556,
      "grad_norm": 1.71875,
      "learning_rate": 0.00014289864523109913,
      "loss": 1.3695,
      "step": 43779
    },
    {
      "epoch": 3.7666695345435772,
      "grad_norm": 57.25,
      "learning_rate": 0.00014287971889015278,
      "loss": 1.3442,
      "step": 43780
    },
    {
      "epoch": 3.766755570850899,
      "grad_norm": 1.5,
      "learning_rate": 0.00014286079359371356,
      "loss": 1.2235,
      "step": 43781
    },
    {
      "epoch": 3.766841607158221,
      "grad_norm": 12.9375,
      "learning_rate": 0.00014284186934183712,
      "loss": 1.7598,
      "step": 43782
    },
    {
      "epoch": 3.7669276434655425,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001428229461345789,
      "loss": 0.6781,
      "step": 43783
    },
    {
      "epoch": 3.767013679772864,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00014280402397199394,
      "loss": 1.3445,
      "step": 43784
    },
    {
      "epoch": 3.767099716080186,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014278510285413776,
      "loss": 0.3624,
      "step": 43785
    },
    {
      "epoch": 3.7671857523875074,
      "grad_norm": 14336.0,
      "learning_rate": 0.0001427661827810658,
      "loss": 1.0403,
      "step": 43786
    },
    {
      "epoch": 3.7672717886948295,
      "grad_norm": 1.046875,
      "learning_rate": 0.00014274726375283314,
      "loss": 1.4037,
      "step": 43787
    },
    {
      "epoch": 3.7673578250021507,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001427283457694954,
      "loss": 1.5051,
      "step": 43788
    },
    {
      "epoch": 3.7674438613094727,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001427094288311076,
      "loss": 1.7821,
      "step": 43789
    },
    {
      "epoch": 3.7675298976167944,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00014269051293772523,
      "loss": 0.7819,
      "step": 43790
    },
    {
      "epoch": 3.767615933924116,
      "grad_norm": 1.40625,
      "learning_rate": 0.00014267159808940377,
      "loss": 1.1687,
      "step": 43791
    },
    {
      "epoch": 3.7677019702314376,
      "grad_norm": 1.484375,
      "learning_rate": 0.00014265268428619826,
      "loss": 1.3701,
      "step": 43792
    },
    {
      "epoch": 3.7677880065387592,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014263377152816414,
      "loss": 0.8006,
      "step": 43793
    },
    {
      "epoch": 3.7678740428460813,
      "grad_norm": 1.03125,
      "learning_rate": 0.00014261485981535688,
      "loss": 1.6366,
      "step": 43794
    },
    {
      "epoch": 3.7679600791534025,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014259594914783154,
      "loss": 1.2872,
      "step": 43795
    },
    {
      "epoch": 3.7680461154607245,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014257703952564343,
      "loss": 1.1466,
      "step": 43796
    },
    {
      "epoch": 3.768132151768046,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00014255813094884805,
      "loss": 2.2801,
      "step": 43797
    },
    {
      "epoch": 3.768218188075368,
      "grad_norm": 1.21875,
      "learning_rate": 0.00014253922341750058,
      "loss": 0.7349,
      "step": 43798
    },
    {
      "epoch": 3.7683042243826894,
      "grad_norm": 21.375,
      "learning_rate": 0.00014252031693165658,
      "loss": 1.0979,
      "step": 43799
    },
    {
      "epoch": 3.768390260690011,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001425014114913709,
      "loss": 0.6446,
      "step": 43800
    },
    {
      "epoch": 3.768476296997333,
      "grad_norm": 3.09375,
      "learning_rate": 0.00014248250709669909,
      "loss": 0.8462,
      "step": 43801
    },
    {
      "epoch": 3.7685623333046543,
      "grad_norm": 1328.0,
      "learning_rate": 0.00014246360374769657,
      "loss": 1.3468,
      "step": 43802
    },
    {
      "epoch": 3.7686483696119764,
      "grad_norm": 30976.0,
      "learning_rate": 0.0001424447014444183,
      "loss": 0.8248,
      "step": 43803
    },
    {
      "epoch": 3.768734405919298,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00014242580018691993,
      "loss": 0.4414,
      "step": 43804
    },
    {
      "epoch": 3.7688204422266196,
      "grad_norm": 2.25,
      "learning_rate": 0.00014240689997525635,
      "loss": 1.6257,
      "step": 43805
    },
    {
      "epoch": 3.7689064785339412,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014238800080948306,
      "loss": 1.7524,
      "step": 43806
    },
    {
      "epoch": 3.768992514841263,
      "grad_norm": 1.890625,
      "learning_rate": 0.00014236910268965546,
      "loss": 1.1751,
      "step": 43807
    },
    {
      "epoch": 3.769078551148585,
      "grad_norm": 26.625,
      "learning_rate": 0.0001423502056158285,
      "loss": 0.9402,
      "step": 43808
    },
    {
      "epoch": 3.7691645874559065,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00014233130958805768,
      "loss": 0.718,
      "step": 43809
    },
    {
      "epoch": 3.769250623763228,
      "grad_norm": 1.296875,
      "learning_rate": 0.00014231241460639838,
      "loss": 0.8449,
      "step": 43810
    },
    {
      "epoch": 3.76933666007055,
      "grad_norm": 74.0,
      "learning_rate": 0.00014229352067090552,
      "loss": 0.8173,
      "step": 43811
    },
    {
      "epoch": 3.7694226963778714,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014227462778163452,
      "loss": 1.437,
      "step": 43812
    },
    {
      "epoch": 3.769508732685193,
      "grad_norm": 0.875,
      "learning_rate": 0.00014225573593864082,
      "loss": 0.7565,
      "step": 43813
    },
    {
      "epoch": 3.7695947689925147,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001422368451419794,
      "loss": 1.4218,
      "step": 43814
    },
    {
      "epoch": 3.7696808052998367,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00014221795539170556,
      "loss": 0.6975,
      "step": 43815
    },
    {
      "epoch": 3.7697668416071584,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014219906668787487,
      "loss": 1.0108,
      "step": 43816
    },
    {
      "epoch": 3.76985287791448,
      "grad_norm": 15.4375,
      "learning_rate": 0.0001421801790305421,
      "loss": 1.0627,
      "step": 43817
    },
    {
      "epoch": 3.7699389142218016,
      "grad_norm": 1.25,
      "learning_rate": 0.00014216129241976288,
      "loss": 1.1415,
      "step": 43818
    },
    {
      "epoch": 3.7700249505291232,
      "grad_norm": 27904.0,
      "learning_rate": 0.0001421424068555921,
      "loss": 1.2933,
      "step": 43819
    },
    {
      "epoch": 3.770110986836445,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014212352233808524,
      "loss": 1.0207,
      "step": 43820
    },
    {
      "epoch": 3.7701970231437665,
      "grad_norm": 1.046875,
      "learning_rate": 0.00014210463886729757,
      "loss": 1.5535,
      "step": 43821
    },
    {
      "epoch": 3.7702830594510885,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001420857564432841,
      "loss": 1.5413,
      "step": 43822
    },
    {
      "epoch": 3.77036909575841,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00014206687506610017,
      "loss": 0.7189,
      "step": 43823
    },
    {
      "epoch": 3.770455132065732,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001420479947358012,
      "loss": 0.9653,
      "step": 43824
    },
    {
      "epoch": 3.7705411683730534,
      "grad_norm": 20.625,
      "learning_rate": 0.00014202911545244207,
      "loss": 1.131,
      "step": 43825
    },
    {
      "epoch": 3.770627204680375,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014201023721607814,
      "loss": 1.1132,
      "step": 43826
    },
    {
      "epoch": 3.7707132409876967,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014199136002676478,
      "loss": 0.6144,
      "step": 43827
    },
    {
      "epoch": 3.7707992772950183,
      "grad_norm": 1.375,
      "learning_rate": 0.0001419724838845569,
      "loss": 1.1703,
      "step": 43828
    },
    {
      "epoch": 3.7708853136023404,
      "grad_norm": 1.0625,
      "learning_rate": 0.00014195360878951007,
      "loss": 1.2423,
      "step": 43829
    },
    {
      "epoch": 3.770971349909662,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014193473474167912,
      "loss": 0.7654,
      "step": 43830
    },
    {
      "epoch": 3.7710573862169836,
      "grad_norm": 31.5,
      "learning_rate": 0.00014191586174111943,
      "loss": 0.7869,
      "step": 43831
    },
    {
      "epoch": 3.7711434225243052,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001418969897878864,
      "loss": 1.2639,
      "step": 43832
    },
    {
      "epoch": 3.771229458831627,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014187811888203465,
      "loss": 0.538,
      "step": 43833
    },
    {
      "epoch": 3.7713154951389485,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00014185924902362007,
      "loss": 1.4865,
      "step": 43834
    },
    {
      "epoch": 3.77140153144627,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001418403802126974,
      "loss": 1.1462,
      "step": 43835
    },
    {
      "epoch": 3.771487567753592,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014182151244932195,
      "loss": 0.7145,
      "step": 43836
    },
    {
      "epoch": 3.771573604060914,
      "grad_norm": 2.140625,
      "learning_rate": 0.0001418026457335491,
      "loss": 1.5793,
      "step": 43837
    },
    {
      "epoch": 3.7716596403682354,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001417837800654337,
      "loss": 1.2388,
      "step": 43838
    },
    {
      "epoch": 3.771745676675557,
      "grad_norm": 1.296875,
      "learning_rate": 0.000141764915445031,
      "loss": 0.6803,
      "step": 43839
    },
    {
      "epoch": 3.7718317129828787,
      "grad_norm": 1.6953125,
      "learning_rate": 0.00014174605187239648,
      "loss": 1.0679,
      "step": 43840
    },
    {
      "epoch": 3.7719177492902007,
      "grad_norm": 2.25,
      "learning_rate": 0.0001417271893475849,
      "loss": 1.2423,
      "step": 43841
    },
    {
      "epoch": 3.772003785597522,
      "grad_norm": 1288.0,
      "learning_rate": 0.00014170832787065157,
      "loss": 1.2028,
      "step": 43842
    },
    {
      "epoch": 3.772089821904844,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001416894674416519,
      "loss": 1.7829,
      "step": 43843
    },
    {
      "epoch": 3.7721758582121656,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014167060806064065,
      "loss": 0.438,
      "step": 43844
    },
    {
      "epoch": 3.7722618945194872,
      "grad_norm": 6.34375,
      "learning_rate": 0.0001416517497276732,
      "loss": 1.5145,
      "step": 43845
    },
    {
      "epoch": 3.772347930826809,
      "grad_norm": 1.125,
      "learning_rate": 0.00014163289244280487,
      "loss": 1.251,
      "step": 43846
    },
    {
      "epoch": 3.7724339671341305,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014161403620609048,
      "loss": 0.8967,
      "step": 43847
    },
    {
      "epoch": 3.7725200034414526,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014159518101758544,
      "loss": 0.9026,
      "step": 43848
    },
    {
      "epoch": 3.7726060397487737,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014157632687734466,
      "loss": 0.9917,
      "step": 43849
    },
    {
      "epoch": 3.772692076056096,
      "grad_norm": 1.0,
      "learning_rate": 0.00014155747378542344,
      "loss": 0.506,
      "step": 43850
    },
    {
      "epoch": 3.7727781123634174,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014153862174187698,
      "loss": 1.5287,
      "step": 43851
    },
    {
      "epoch": 3.772864148670739,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014151977074676025,
      "loss": 0.5267,
      "step": 43852
    },
    {
      "epoch": 3.7729501849780607,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00014150092080012844,
      "loss": 0.9396,
      "step": 43853
    },
    {
      "epoch": 3.7730362212853823,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014148207190203688,
      "loss": 1.0188,
      "step": 43854
    },
    {
      "epoch": 3.7731222575927044,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00014146322405254037,
      "loss": 1.3667,
      "step": 43855
    },
    {
      "epoch": 3.7732082939000255,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00014144437725169417,
      "loss": 1.1893,
      "step": 43856
    },
    {
      "epoch": 3.7732943302073476,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001414255314995536,
      "loss": 0.8954,
      "step": 43857
    },
    {
      "epoch": 3.7733803665146692,
      "grad_norm": 1.21875,
      "learning_rate": 0.00014140668679617352,
      "loss": 1.1717,
      "step": 43858
    },
    {
      "epoch": 3.773466402821991,
      "grad_norm": 2.296875,
      "learning_rate": 0.00014138784314160924,
      "loss": 0.8424,
      "step": 43859
    },
    {
      "epoch": 3.7735524391293125,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00014136900053591563,
      "loss": 0.9061,
      "step": 43860
    },
    {
      "epoch": 3.773638475436634,
      "grad_norm": 120.0,
      "learning_rate": 0.00014135015897914793,
      "loss": 0.7139,
      "step": 43861
    },
    {
      "epoch": 3.773724511743956,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00014133131847136144,
      "loss": 0.7516,
      "step": 43862
    },
    {
      "epoch": 3.773810548051278,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001413124790126109,
      "loss": 0.7038,
      "step": 43863
    },
    {
      "epoch": 3.7738965843585994,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001412936406029516,
      "loss": 0.8342,
      "step": 43864
    },
    {
      "epoch": 3.773982620665921,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001412748032424388,
      "loss": 0.6669,
      "step": 43865
    },
    {
      "epoch": 3.7740686569732427,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001412559669311273,
      "loss": 1.2243,
      "step": 43866
    },
    {
      "epoch": 3.7741546932805643,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00014123713166907227,
      "loss": 1.097,
      "step": 43867
    },
    {
      "epoch": 3.774240729587886,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014121829745632907,
      "loss": 0.822,
      "step": 43868
    },
    {
      "epoch": 3.774326765895208,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001411994642929524,
      "loss": 0.6971,
      "step": 43869
    },
    {
      "epoch": 3.7744128022025296,
      "grad_norm": 1.53125,
      "learning_rate": 0.00014118063217899746,
      "loss": 1.9943,
      "step": 43870
    },
    {
      "epoch": 3.7744988385098512,
      "grad_norm": 1.875,
      "learning_rate": 0.00014116180111451942,
      "loss": 1.0688,
      "step": 43871
    },
    {
      "epoch": 3.774584874817173,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00014114297109957337,
      "loss": 1.7313,
      "step": 43872
    },
    {
      "epoch": 3.7746709111244945,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00014112414213421443,
      "loss": 2.2037,
      "step": 43873
    },
    {
      "epoch": 3.774756947431816,
      "grad_norm": 5.34375,
      "learning_rate": 0.00014110531421849742,
      "loss": 1.5948,
      "step": 43874
    },
    {
      "epoch": 3.7748429837391377,
      "grad_norm": 1.171875,
      "learning_rate": 0.00014108648735247775,
      "loss": 1.0017,
      "step": 43875
    },
    {
      "epoch": 3.77492902004646,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014106766153621015,
      "loss": 1.3094,
      "step": 43876
    },
    {
      "epoch": 3.7750150563537814,
      "grad_norm": 1.234375,
      "learning_rate": 0.00014104883676974983,
      "loss": 0.9266,
      "step": 43877
    },
    {
      "epoch": 3.775101092661103,
      "grad_norm": 2.421875,
      "learning_rate": 0.000141030013053152,
      "loss": 0.6566,
      "step": 43878
    },
    {
      "epoch": 3.7751871289684247,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014101119038647142,
      "loss": 2.2956,
      "step": 43879
    },
    {
      "epoch": 3.7752731652757463,
      "grad_norm": 1.1875,
      "learning_rate": 0.00014099236876976323,
      "loss": 1.7659,
      "step": 43880
    },
    {
      "epoch": 3.775359201583068,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014097354820308278,
      "loss": 1.1513,
      "step": 43881
    },
    {
      "epoch": 3.7754452378903895,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014095472868648463,
      "loss": 0.8321,
      "step": 43882
    },
    {
      "epoch": 3.7755312741977116,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0001409359102200241,
      "loss": 1.1906,
      "step": 43883
    },
    {
      "epoch": 3.7756173105050332,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001409170928037564,
      "loss": 0.9959,
      "step": 43884
    },
    {
      "epoch": 3.775703346812355,
      "grad_norm": 1.28125,
      "learning_rate": 0.00014089827643773618,
      "loss": 1.0122,
      "step": 43885
    },
    {
      "epoch": 3.7757893831196765,
      "grad_norm": 21.5,
      "learning_rate": 0.00014087946112201867,
      "loss": 1.0665,
      "step": 43886
    },
    {
      "epoch": 3.775875419426998,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000140860646856659,
      "loss": 0.8789,
      "step": 43887
    },
    {
      "epoch": 3.7759614557343197,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014084183364171198,
      "loss": 0.6884,
      "step": 43888
    },
    {
      "epoch": 3.7760474920416414,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001408230214772329,
      "loss": 1.3263,
      "step": 43889
    },
    {
      "epoch": 3.7761335283489634,
      "grad_norm": 1.140625,
      "learning_rate": 0.00014080421036327646,
      "loss": 1.1684,
      "step": 43890
    },
    {
      "epoch": 3.776219564656285,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014078540029989784,
      "loss": 0.9891,
      "step": 43891
    },
    {
      "epoch": 3.7763056009636067,
      "grad_norm": 3.859375,
      "learning_rate": 0.00014076659128715226,
      "loss": 1.0908,
      "step": 43892
    },
    {
      "epoch": 3.7763916372709283,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001407477833250943,
      "loss": 1.3329,
      "step": 43893
    },
    {
      "epoch": 3.77647767357825,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00014072897641377924,
      "loss": 1.4952,
      "step": 43894
    },
    {
      "epoch": 3.7765637098855716,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001407101705532622,
      "loss": 0.978,
      "step": 43895
    },
    {
      "epoch": 3.776649746192893,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00014069136574359792,
      "loss": 1.0997,
      "step": 43896
    },
    {
      "epoch": 3.7767357825002152,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00014067256198484147,
      "loss": 0.9319,
      "step": 43897
    },
    {
      "epoch": 3.776821818807537,
      "grad_norm": 3.265625,
      "learning_rate": 0.00014065375927704805,
      "loss": 1.27,
      "step": 43898
    },
    {
      "epoch": 3.7769078551148585,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00014063495762027235,
      "loss": 0.5192,
      "step": 43899
    },
    {
      "epoch": 3.77699389142218,
      "grad_norm": 1.125,
      "learning_rate": 0.00014061615701456965,
      "loss": 1.1696,
      "step": 43900
    },
    {
      "epoch": 3.7770799277295017,
      "grad_norm": 1.375,
      "learning_rate": 0.00014059735745999464,
      "loss": 1.6257,
      "step": 43901
    },
    {
      "epoch": 3.777165964036824,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014057855895660242,
      "loss": 0.9424,
      "step": 43902
    },
    {
      "epoch": 3.777252000344145,
      "grad_norm": 1.8125,
      "learning_rate": 0.00014055976150444826,
      "loss": 1.0282,
      "step": 43903
    },
    {
      "epoch": 3.777338036651467,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014054096510358665,
      "loss": 1.0704,
      "step": 43904
    },
    {
      "epoch": 3.7774240729587887,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00014052216975407283,
      "loss": 1.0219,
      "step": 43905
    },
    {
      "epoch": 3.7775101092661103,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014050337545596191,
      "loss": 1.3949,
      "step": 43906
    },
    {
      "epoch": 3.777596145573432,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014048458220930843,
      "loss": 1.4531,
      "step": 43907
    },
    {
      "epoch": 3.7776821818807536,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00014046579001416793,
      "loss": 1.1702,
      "step": 43908
    },
    {
      "epoch": 3.7777682181880756,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00014044699887059487,
      "loss": 0.8729,
      "step": 43909
    },
    {
      "epoch": 3.777854254495397,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00014042820877864442,
      "loss": 0.6841,
      "step": 43910
    },
    {
      "epoch": 3.777940290802719,
      "grad_norm": 1.453125,
      "learning_rate": 0.00014040941973837167,
      "loss": 1.1338,
      "step": 43911
    },
    {
      "epoch": 3.7780263271100405,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001403906317498313,
      "loss": 0.7555,
      "step": 43912
    },
    {
      "epoch": 3.778112363417362,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00014037184481307835,
      "loss": 1.5451,
      "step": 43913
    },
    {
      "epoch": 3.7781983997246837,
      "grad_norm": 440.0,
      "learning_rate": 0.00014035305892816803,
      "loss": 1.5149,
      "step": 43914
    },
    {
      "epoch": 3.7782844360320054,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014033427409515487,
      "loss": 0.4203,
      "step": 43915
    },
    {
      "epoch": 3.7783704723393274,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001403154903140942,
      "loss": 0.8909,
      "step": 43916
    },
    {
      "epoch": 3.7784565086466486,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014029670758504054,
      "loss": 0.528,
      "step": 43917
    },
    {
      "epoch": 3.7785425449539707,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001402779259080491,
      "loss": 0.7107,
      "step": 43918
    },
    {
      "epoch": 3.7786285812612923,
      "grad_norm": 14.375,
      "learning_rate": 0.0001402591452831749,
      "loss": 1.3214,
      "step": 43919
    },
    {
      "epoch": 3.778714617568614,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014024036571047262,
      "loss": 1.0841,
      "step": 43920
    },
    {
      "epoch": 3.7788006538759356,
      "grad_norm": 0.96875,
      "learning_rate": 0.00014022158718999728,
      "loss": 0.9849,
      "step": 43921
    },
    {
      "epoch": 3.778886690183257,
      "grad_norm": 1.140625,
      "learning_rate": 0.00014020280972180393,
      "loss": 0.5791,
      "step": 43922
    },
    {
      "epoch": 3.7789727264905792,
      "grad_norm": 62.5,
      "learning_rate": 0.0001401840333059473,
      "loss": 1.3148,
      "step": 43923
    },
    {
      "epoch": 3.779058762797901,
      "grad_norm": 2.71875,
      "learning_rate": 0.00014016525794248235,
      "loss": 1.6409,
      "step": 43924
    },
    {
      "epoch": 3.7791447991052225,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001401464836314642,
      "loss": 0.9247,
      "step": 43925
    },
    {
      "epoch": 3.779230835412544,
      "grad_norm": 1.7890625,
      "learning_rate": 0.00014012771037294746,
      "loss": 0.9929,
      "step": 43926
    },
    {
      "epoch": 3.7793168717198657,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00014010893816698717,
      "loss": 0.8603,
      "step": 43927
    },
    {
      "epoch": 3.7794029080271874,
      "grad_norm": 68.0,
      "learning_rate": 0.00014009016701363835,
      "loss": 0.9728,
      "step": 43928
    },
    {
      "epoch": 3.779488944334509,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014007139691295573,
      "loss": 0.8293,
      "step": 43929
    },
    {
      "epoch": 3.779574980641831,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00014005262786499434,
      "loss": 1.1844,
      "step": 43930
    },
    {
      "epoch": 3.7796610169491527,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014003385986980887,
      "loss": 0.9426,
      "step": 43931
    },
    {
      "epoch": 3.7797470532564743,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001400150929274544,
      "loss": 1.3052,
      "step": 43932
    },
    {
      "epoch": 3.779833089563796,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013999632703798586,
      "loss": 1.3833,
      "step": 43933
    },
    {
      "epoch": 3.7799191258711176,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00013997756220145797,
      "loss": 1.3659,
      "step": 43934
    },
    {
      "epoch": 3.780005162178439,
      "grad_norm": 1.09375,
      "learning_rate": 0.00013995879841792564,
      "loss": 0.646,
      "step": 43935
    },
    {
      "epoch": 3.780091198485761,
      "grad_norm": 2.609375,
      "learning_rate": 0.00013994003568744396,
      "loss": 0.8917,
      "step": 43936
    },
    {
      "epoch": 3.780177234793083,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001399212740100675,
      "loss": 1.3346,
      "step": 43937
    },
    {
      "epoch": 3.7802632711004045,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00013990251338585124,
      "loss": 1.4399,
      "step": 43938
    },
    {
      "epoch": 3.780349307407726,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001398837538148503,
      "loss": 1.2788,
      "step": 43939
    },
    {
      "epoch": 3.7804353437150477,
      "grad_norm": 4.46875,
      "learning_rate": 0.00013986499529711915,
      "loss": 1.1199,
      "step": 43940
    },
    {
      "epoch": 3.7805213800223694,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00013984623783271305,
      "loss": 0.6463,
      "step": 43941
    },
    {
      "epoch": 3.780607416329691,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00013982748142168645,
      "loss": 0.8846,
      "step": 43942
    },
    {
      "epoch": 3.7806934526370126,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013980872606409445,
      "loss": 0.9166,
      "step": 43943
    },
    {
      "epoch": 3.7807794889443347,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00013978997175999186,
      "loss": 0.6752,
      "step": 43944
    },
    {
      "epoch": 3.7808655252516563,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013977121850943352,
      "loss": 1.299,
      "step": 43945
    },
    {
      "epoch": 3.780951561558978,
      "grad_norm": 5.84375,
      "learning_rate": 0.00013975246631247451,
      "loss": 0.4477,
      "step": 43946
    },
    {
      "epoch": 3.7810375978662996,
      "grad_norm": 1.0,
      "learning_rate": 0.00013973371516916934,
      "loss": 0.8452,
      "step": 43947
    },
    {
      "epoch": 3.781123634173621,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00013971496507957294,
      "loss": 1.0703,
      "step": 43948
    },
    {
      "epoch": 3.781209670480943,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00013969621604374032,
      "loss": 1.1046,
      "step": 43949
    },
    {
      "epoch": 3.7812957067882644,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00013967746806172604,
      "loss": 0.5793,
      "step": 43950
    },
    {
      "epoch": 3.7813817430955865,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00013965872113358513,
      "loss": 1.0746,
      "step": 43951
    },
    {
      "epoch": 3.781467779402908,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00013963997525937255,
      "loss": 1.0335,
      "step": 43952
    },
    {
      "epoch": 3.7815538157102297,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00013962123043914277,
      "loss": 1.1935,
      "step": 43953
    },
    {
      "epoch": 3.7816398520175514,
      "grad_norm": 0.875,
      "learning_rate": 0.00013960248667295078,
      "loss": 1.4715,
      "step": 43954
    },
    {
      "epoch": 3.781725888324873,
      "grad_norm": 1.015625,
      "learning_rate": 0.00013958374396085162,
      "loss": 0.9108,
      "step": 43955
    },
    {
      "epoch": 3.7818119246321946,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00013956500230289976,
      "loss": 0.8692,
      "step": 43956
    },
    {
      "epoch": 3.7818979609395162,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013954626169915025,
      "loss": 0.8096,
      "step": 43957
    },
    {
      "epoch": 3.7819839972468383,
      "grad_norm": 1.328125,
      "learning_rate": 0.00013952752214965776,
      "loss": 1.4236,
      "step": 43958
    },
    {
      "epoch": 3.78207003355416,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001395087836544771,
      "loss": 1.2679,
      "step": 43959
    },
    {
      "epoch": 3.7821560698614816,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001394900462136633,
      "loss": 1.6414,
      "step": 43960
    },
    {
      "epoch": 3.782242106168803,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00013947130982727086,
      "loss": 0.7509,
      "step": 43961
    },
    {
      "epoch": 3.782328142476125,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001394525744953547,
      "loss": 1.3502,
      "step": 43962
    },
    {
      "epoch": 3.782414178783447,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013943384021796978,
      "loss": 0.8048,
      "step": 43963
    },
    {
      "epoch": 3.782500215090768,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013941510699517064,
      "loss": 1.1584,
      "step": 43964
    },
    {
      "epoch": 3.78258625139809,
      "grad_norm": 544.0,
      "learning_rate": 0.00013939637482701217,
      "loss": 0.9075,
      "step": 43965
    },
    {
      "epoch": 3.7826722877054118,
      "grad_norm": 1.34375,
      "learning_rate": 0.00013937764371354933,
      "loss": 0.8729,
      "step": 43966
    },
    {
      "epoch": 3.7827583240127334,
      "grad_norm": 1.0,
      "learning_rate": 0.00013935891365483654,
      "loss": 0.8638,
      "step": 43967
    },
    {
      "epoch": 3.782844360320055,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00013934018465092886,
      "loss": 1.23,
      "step": 43968
    },
    {
      "epoch": 3.7829303966273766,
      "grad_norm": 1.0,
      "learning_rate": 0.00013932145670188107,
      "loss": 0.5202,
      "step": 43969
    },
    {
      "epoch": 3.7830164329346987,
      "grad_norm": 4.4375,
      "learning_rate": 0.00013930272980774777,
      "loss": 1.2364,
      "step": 43970
    },
    {
      "epoch": 3.78310246924202,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000139284003968584,
      "loss": 1.1554,
      "step": 43971
    },
    {
      "epoch": 3.783188505549342,
      "grad_norm": 2.859375,
      "learning_rate": 0.00013926527918444416,
      "loss": 1.4256,
      "step": 43972
    },
    {
      "epoch": 3.7832745418566636,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00013924655545538323,
      "loss": 0.9273,
      "step": 43973
    },
    {
      "epoch": 3.783360578163985,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013922783278145617,
      "loss": 1.6706,
      "step": 43974
    },
    {
      "epoch": 3.783446614471307,
      "grad_norm": 2.703125,
      "learning_rate": 0.00013920911116271735,
      "loss": 1.4879,
      "step": 43975
    },
    {
      "epoch": 3.7835326507786284,
      "grad_norm": 62.5,
      "learning_rate": 0.00013919039059922167,
      "loss": 1.2993,
      "step": 43976
    },
    {
      "epoch": 3.7836186870859505,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013917167109102414,
      "loss": 0.838,
      "step": 43977
    },
    {
      "epoch": 3.7837047233932717,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00013915295263817906,
      "loss": 0.7364,
      "step": 43978
    },
    {
      "epoch": 3.7837907597005938,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013913423524074148,
      "loss": 1.1764,
      "step": 43979
    },
    {
      "epoch": 3.7838767960079154,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00013911551889876623,
      "loss": 0.6147,
      "step": 43980
    },
    {
      "epoch": 3.783962832315237,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001390968036123076,
      "loss": 1.1405,
      "step": 43981
    },
    {
      "epoch": 3.7840488686225586,
      "grad_norm": 2.671875,
      "learning_rate": 0.00013907808938142086,
      "loss": 0.4832,
      "step": 43982
    },
    {
      "epoch": 3.7841349049298802,
      "grad_norm": 250.0,
      "learning_rate": 0.00013905937620616043,
      "loss": 0.8274,
      "step": 43983
    },
    {
      "epoch": 3.7842209412372023,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001390406640865811,
      "loss": 1.1794,
      "step": 43984
    },
    {
      "epoch": 3.784306977544524,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001390219530227378,
      "loss": 0.8554,
      "step": 43985
    },
    {
      "epoch": 3.7843930138518456,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001390032430146849,
      "loss": 0.9882,
      "step": 43986
    },
    {
      "epoch": 3.784479050159167,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00013898453406247747,
      "loss": 1.0319,
      "step": 43987
    },
    {
      "epoch": 3.784565086466489,
      "grad_norm": 4.375,
      "learning_rate": 0.00013896582616616994,
      "loss": 1.3541,
      "step": 43988
    },
    {
      "epoch": 3.7846511227738104,
      "grad_norm": 1.0,
      "learning_rate": 0.0001389471193258171,
      "loss": 1.2483,
      "step": 43989
    },
    {
      "epoch": 3.784737159081132,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001389284135414739,
      "loss": 1.1985,
      "step": 43990
    },
    {
      "epoch": 3.784823195388454,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001389097088131947,
      "loss": 1.2884,
      "step": 43991
    },
    {
      "epoch": 3.7849092316957758,
      "grad_norm": 9.9375,
      "learning_rate": 0.00013889100514103436,
      "loss": 1.1004,
      "step": 43992
    },
    {
      "epoch": 3.7849952680030974,
      "grad_norm": 2.53125,
      "learning_rate": 0.00013887230252504775,
      "loss": 0.8176,
      "step": 43993
    },
    {
      "epoch": 3.785081304310419,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001388536009652893,
      "loss": 2.0081,
      "step": 43994
    },
    {
      "epoch": 3.7851673406177406,
      "grad_norm": 8.4375,
      "learning_rate": 0.0001388349004618138,
      "loss": 1.0357,
      "step": 43995
    },
    {
      "epoch": 3.7852533769250623,
      "grad_norm": 1.109375,
      "learning_rate": 0.00013881620101467612,
      "loss": 0.8496,
      "step": 43996
    },
    {
      "epoch": 3.785339413232384,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001387975026239307,
      "loss": 1.7707,
      "step": 43997
    },
    {
      "epoch": 3.785425449539706,
      "grad_norm": 1.671875,
      "learning_rate": 0.00013877880528963228,
      "loss": 1.4838,
      "step": 43998
    },
    {
      "epoch": 3.7855114858470276,
      "grad_norm": 2.25,
      "learning_rate": 0.00013876010901183578,
      "loss": 1.2396,
      "step": 43999
    },
    {
      "epoch": 3.785597522154349,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00013874141379059557,
      "loss": 0.9915,
      "step": 44000
    },
    {
      "epoch": 3.785683558461671,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00013872271962596662,
      "loss": 0.6902,
      "step": 44001
    },
    {
      "epoch": 3.7857695947689924,
      "grad_norm": 12.25,
      "learning_rate": 0.00013870402651800323,
      "loss": 0.8664,
      "step": 44002
    },
    {
      "epoch": 3.785855631076314,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013868533446676035,
      "loss": 0.7839,
      "step": 44003
    },
    {
      "epoch": 3.7859416673836357,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00013866664347229275,
      "loss": 1.4383,
      "step": 44004
    },
    {
      "epoch": 3.7860277036909578,
      "grad_norm": 1.078125,
      "learning_rate": 0.00013864795353465476,
      "loss": 1.1186,
      "step": 44005
    },
    {
      "epoch": 3.7861137399982794,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00013862926465390126,
      "loss": 0.5534,
      "step": 44006
    },
    {
      "epoch": 3.786199776305601,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00013861057683008698,
      "loss": 1.4919,
      "step": 44007
    },
    {
      "epoch": 3.7862858126129226,
      "grad_norm": 55.25,
      "learning_rate": 0.0001385918900632664,
      "loss": 0.8131,
      "step": 44008
    },
    {
      "epoch": 3.7863718489202443,
      "grad_norm": 1.203125,
      "learning_rate": 0.00013857320435349419,
      "loss": 1.7532,
      "step": 44009
    },
    {
      "epoch": 3.786457885227566,
      "grad_norm": 50.75,
      "learning_rate": 0.00013855451970082522,
      "loss": 1.059,
      "step": 44010
    },
    {
      "epoch": 3.7865439215348875,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00013853583610531385,
      "loss": 0.7787,
      "step": 44011
    },
    {
      "epoch": 3.7866299578422096,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000138517153567015,
      "loss": 0.8706,
      "step": 44012
    },
    {
      "epoch": 3.786715994149531,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00013849847208598305,
      "loss": 1.678,
      "step": 44013
    },
    {
      "epoch": 3.786802030456853,
      "grad_norm": 0.921875,
      "learning_rate": 0.00013847979166227276,
      "loss": 1.2673,
      "step": 44014
    },
    {
      "epoch": 3.7868880667641744,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001384611122959389,
      "loss": 1.637,
      "step": 44015
    },
    {
      "epoch": 3.786974103071496,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00013844243398703578,
      "loss": 1.5806,
      "step": 44016
    },
    {
      "epoch": 3.787060139378818,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00013842375673561825,
      "loss": 1.0443,
      "step": 44017
    },
    {
      "epoch": 3.7871461756861393,
      "grad_norm": 1.1875,
      "learning_rate": 0.00013840508054174095,
      "loss": 0.8505,
      "step": 44018
    },
    {
      "epoch": 3.7872322119934614,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001383864054054584,
      "loss": 1.1292,
      "step": 44019
    },
    {
      "epoch": 3.787318248300783,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00013836773132682546,
      "loss": 1.1297,
      "step": 44020
    },
    {
      "epoch": 3.7874042846081046,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013834905830589645,
      "loss": 0.7764,
      "step": 44021
    },
    {
      "epoch": 3.7874903209154263,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00013833038634272605,
      "loss": 1.1878,
      "step": 44022
    },
    {
      "epoch": 3.787576357222748,
      "grad_norm": 81.5,
      "learning_rate": 0.00013831171543736914,
      "loss": 0.863,
      "step": 44023
    },
    {
      "epoch": 3.78766239353007,
      "grad_norm": 21.375,
      "learning_rate": 0.00013829304558987993,
      "loss": 0.9501,
      "step": 44024
    },
    {
      "epoch": 3.787748429837391,
      "grad_norm": 5.125,
      "learning_rate": 0.00013827437680031325,
      "loss": 1.0269,
      "step": 44025
    },
    {
      "epoch": 3.787834466144713,
      "grad_norm": 1.25,
      "learning_rate": 0.00013825570906872376,
      "loss": 1.0483,
      "step": 44026
    },
    {
      "epoch": 3.787920502452035,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001382370423951658,
      "loss": 0.9128,
      "step": 44027
    },
    {
      "epoch": 3.7880065387593564,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001382183767796944,
      "loss": 0.7185,
      "step": 44028
    },
    {
      "epoch": 3.788092575066678,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00013819971222236367,
      "loss": 1.0972,
      "step": 44029
    },
    {
      "epoch": 3.7881786113739997,
      "grad_norm": 1.296875,
      "learning_rate": 0.00013818104872322838,
      "loss": 1.3094,
      "step": 44030
    },
    {
      "epoch": 3.7882646476813218,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00013816238628234335,
      "loss": 1.9959,
      "step": 44031
    },
    {
      "epoch": 3.788350683988643,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0001381437248997628,
      "loss": 1.3542,
      "step": 44032
    },
    {
      "epoch": 3.788436720295965,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00013812506457554146,
      "loss": 0.9537,
      "step": 44033
    },
    {
      "epoch": 3.7885227566032866,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013810640530973405,
      "loss": 1.3244,
      "step": 44034
    },
    {
      "epoch": 3.7886087929106083,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001380877471023949,
      "loss": 0.6225,
      "step": 44035
    },
    {
      "epoch": 3.78869482921793,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013806908995357876,
      "loss": 0.7085,
      "step": 44036
    },
    {
      "epoch": 3.7887808655252515,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00013805043386334021,
      "loss": 0.5568,
      "step": 44037
    },
    {
      "epoch": 3.7888669018325736,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001380317788317336,
      "loss": 0.6539,
      "step": 44038
    },
    {
      "epoch": 3.788952938139895,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013801312485881373,
      "loss": 0.5651,
      "step": 44039
    },
    {
      "epoch": 3.789038974447217,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00013799447194463515,
      "loss": 0.5011,
      "step": 44040
    },
    {
      "epoch": 3.7891250107545384,
      "grad_norm": 3.0,
      "learning_rate": 0.00013797582008925218,
      "loss": 1.3038,
      "step": 44041
    },
    {
      "epoch": 3.78921104706186,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00013795716929271968,
      "loss": 1.0304,
      "step": 44042
    },
    {
      "epoch": 3.7892970833691817,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001379385195550919,
      "loss": 1.0125,
      "step": 44043
    },
    {
      "epoch": 3.7893831196765033,
      "grad_norm": 94.0,
      "learning_rate": 0.00013791987087642354,
      "loss": 0.6996,
      "step": 44044
    },
    {
      "epoch": 3.7894691559838254,
      "grad_norm": 1.375,
      "learning_rate": 0.00013790122325676928,
      "loss": 2.1547,
      "step": 44045
    },
    {
      "epoch": 3.789555192291147,
      "grad_norm": 81.5,
      "learning_rate": 0.00013788257669618338,
      "loss": 2.0671,
      "step": 44046
    },
    {
      "epoch": 3.7896412285984686,
      "grad_norm": 1.65625,
      "learning_rate": 0.00013786393119472046,
      "loss": 1.104,
      "step": 44047
    },
    {
      "epoch": 3.7897272649057903,
      "grad_norm": 2.453125,
      "learning_rate": 0.0001378452867524353,
      "loss": 1.0063,
      "step": 44048
    },
    {
      "epoch": 3.789813301213112,
      "grad_norm": 11.5,
      "learning_rate": 0.00013782664336938212,
      "loss": 1.898,
      "step": 44049
    },
    {
      "epoch": 3.7898993375204335,
      "grad_norm": 0.9375,
      "learning_rate": 0.00013780800104561553,
      "loss": 0.7179,
      "step": 44050
    },
    {
      "epoch": 3.789985373827755,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013778935978119028,
      "loss": 0.9958,
      "step": 44051
    },
    {
      "epoch": 3.790071410135077,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001377707195761605,
      "loss": 0.8452,
      "step": 44052
    },
    {
      "epoch": 3.790157446442399,
      "grad_norm": 2.03125,
      "learning_rate": 0.00013775208043058113,
      "loss": 0.9775,
      "step": 44053
    },
    {
      "epoch": 3.7902434827497204,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00013773344234450626,
      "loss": 0.5502,
      "step": 44054
    },
    {
      "epoch": 3.790329519057042,
      "grad_norm": 1.0,
      "learning_rate": 0.00013771480531799053,
      "loss": 1.6399,
      "step": 44055
    },
    {
      "epoch": 3.7904155553643637,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00013769616935108874,
      "loss": 1.0722,
      "step": 44056
    },
    {
      "epoch": 3.7905015916716853,
      "grad_norm": 39.75,
      "learning_rate": 0.00013767753444385511,
      "loss": 1.0722,
      "step": 44057
    },
    {
      "epoch": 3.790587627979007,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00013765890059634438,
      "loss": 1.4663,
      "step": 44058
    },
    {
      "epoch": 3.790673664286329,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013764026780861073,
      "loss": 0.7886,
      "step": 44059
    },
    {
      "epoch": 3.7907597005936506,
      "grad_norm": 1.25,
      "learning_rate": 0.00013762163608070882,
      "loss": 1.7464,
      "step": 44060
    },
    {
      "epoch": 3.7908457369009723,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00013760300541269327,
      "loss": 0.7433,
      "step": 44061
    },
    {
      "epoch": 3.790931773208294,
      "grad_norm": 42.0,
      "learning_rate": 0.00013758437580461836,
      "loss": 1.0733,
      "step": 44062
    },
    {
      "epoch": 3.7910178095156155,
      "grad_norm": 1.265625,
      "learning_rate": 0.00013756574725653858,
      "loss": 0.7647,
      "step": 44063
    },
    {
      "epoch": 3.791103845822937,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00013754711976850877,
      "loss": 1.4435,
      "step": 44064
    },
    {
      "epoch": 3.7911898821302588,
      "grad_norm": 1.21875,
      "learning_rate": 0.00013752849334058287,
      "loss": 1.3638,
      "step": 44065
    },
    {
      "epoch": 3.791275918437581,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00013750986797281568,
      "loss": 1.6493,
      "step": 44066
    },
    {
      "epoch": 3.7913619547449025,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013749124366526177,
      "loss": 0.5496,
      "step": 44067
    },
    {
      "epoch": 3.791447991052224,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00013747262041797532,
      "loss": 1.1341,
      "step": 44068
    },
    {
      "epoch": 3.7915340273595457,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00013745399823101112,
      "loss": 1.0793,
      "step": 44069
    },
    {
      "epoch": 3.7916200636668673,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013743537710442328,
      "loss": 1.5948,
      "step": 44070
    },
    {
      "epoch": 3.791706099974189,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013741675703826645,
      "loss": 0.5442,
      "step": 44071
    },
    {
      "epoch": 3.7917921362815106,
      "grad_norm": 1.109375,
      "learning_rate": 0.00013739813803259526,
      "loss": 0.8634,
      "step": 44072
    },
    {
      "epoch": 3.7918781725888326,
      "grad_norm": 1.25,
      "learning_rate": 0.00013737952008746385,
      "loss": 1.019,
      "step": 44073
    },
    {
      "epoch": 3.7919642088961543,
      "grad_norm": 2.65625,
      "learning_rate": 0.00013736090320292676,
      "loss": 1.0615,
      "step": 44074
    },
    {
      "epoch": 3.792050245203476,
      "grad_norm": 29.625,
      "learning_rate": 0.0001373422873790387,
      "loss": 1.2404,
      "step": 44075
    },
    {
      "epoch": 3.7921362815107975,
      "grad_norm": 1.0625,
      "learning_rate": 0.00013732367261585376,
      "loss": 1.2183,
      "step": 44076
    },
    {
      "epoch": 3.792222317818119,
      "grad_norm": 2.296875,
      "learning_rate": 0.00013730505891342655,
      "loss": 1.2388,
      "step": 44077
    },
    {
      "epoch": 3.792308354125441,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00013728644627181163,
      "loss": 1.6209,
      "step": 44078
    },
    {
      "epoch": 3.7923943904327624,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013726783469106317,
      "loss": 1.2363,
      "step": 44079
    },
    {
      "epoch": 3.7924804267400845,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00013724922417123576,
      "loss": 1.4373,
      "step": 44080
    },
    {
      "epoch": 3.792566463047406,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001372306147123839,
      "loss": 1.6103,
      "step": 44081
    },
    {
      "epoch": 3.7926524993547277,
      "grad_norm": 1.03125,
      "learning_rate": 0.00013721200631456188,
      "loss": 0.6172,
      "step": 44082
    },
    {
      "epoch": 3.7927385356620493,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001371933989778243,
      "loss": 0.6418,
      "step": 44083
    },
    {
      "epoch": 3.792824571969371,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00013717479270222528,
      "loss": 0.8954,
      "step": 44084
    },
    {
      "epoch": 3.792910608276693,
      "grad_norm": 50.0,
      "learning_rate": 0.00013715618748781944,
      "loss": 0.9995,
      "step": 44085
    },
    {
      "epoch": 3.792996644584014,
      "grad_norm": 2.09375,
      "learning_rate": 0.00013713758333466138,
      "loss": 1.1495,
      "step": 44086
    },
    {
      "epoch": 3.7930826808913363,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013711898024280512,
      "loss": 1.2593,
      "step": 44087
    },
    {
      "epoch": 3.793168717198658,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001371003782123053,
      "loss": 1.4475,
      "step": 44088
    },
    {
      "epoch": 3.7932547535059795,
      "grad_norm": 2.65625,
      "learning_rate": 0.0001370817772432164,
      "loss": 0.6552,
      "step": 44089
    },
    {
      "epoch": 3.793340789813301,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001370631773355926,
      "loss": 1.7808,
      "step": 44090
    },
    {
      "epoch": 3.7934268261206228,
      "grad_norm": 0.984375,
      "learning_rate": 0.00013704457848948842,
      "loss": 1.1459,
      "step": 44091
    },
    {
      "epoch": 3.793512862427945,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00013702598070495825,
      "loss": 1.0247,
      "step": 44092
    },
    {
      "epoch": 3.793598898735266,
      "grad_norm": 1.25,
      "learning_rate": 0.0001370073839820565,
      "loss": 1.5445,
      "step": 44093
    },
    {
      "epoch": 3.793684935042588,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00013698878832083767,
      "loss": 0.8851,
      "step": 44094
    },
    {
      "epoch": 3.7937709713499097,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001369701937213559,
      "loss": 1.3454,
      "step": 44095
    },
    {
      "epoch": 3.7938570076572313,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013695160018366572,
      "loss": 1.0968,
      "step": 44096
    },
    {
      "epoch": 3.793943043964553,
      "grad_norm": 1.140625,
      "learning_rate": 0.00013693300770782163,
      "loss": 1.0393,
      "step": 44097
    },
    {
      "epoch": 3.7940290802718746,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00013691441629387773,
      "loss": 0.9675,
      "step": 44098
    },
    {
      "epoch": 3.7941151165791966,
      "grad_norm": 0.984375,
      "learning_rate": 0.00013689582594188872,
      "loss": 1.2582,
      "step": 44099
    },
    {
      "epoch": 3.7942011528865183,
      "grad_norm": 1.1875,
      "learning_rate": 0.00013687723665190867,
      "loss": 1.205,
      "step": 44100
    },
    {
      "epoch": 3.79428718919384,
      "grad_norm": 1.359375,
      "learning_rate": 0.00013685864842399203,
      "loss": 1.3678,
      "step": 44101
    },
    {
      "epoch": 3.7943732255011615,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00013684006125819337,
      "loss": 0.676,
      "step": 44102
    },
    {
      "epoch": 3.794459261808483,
      "grad_norm": 119296.0,
      "learning_rate": 0.00013682147515456679,
      "loss": 1.634,
      "step": 44103
    },
    {
      "epoch": 3.7945452981158048,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001368028901131667,
      "loss": 0.6117,
      "step": 44104
    },
    {
      "epoch": 3.7946313344231264,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00013678430613404776,
      "loss": 1.3735,
      "step": 44105
    },
    {
      "epoch": 3.7947173707304485,
      "grad_norm": 1.25,
      "learning_rate": 0.00013676572321726384,
      "loss": 0.991,
      "step": 44106
    },
    {
      "epoch": 3.79480340703777,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001367471413628696,
      "loss": 1.3803,
      "step": 44107
    },
    {
      "epoch": 3.7948894433450917,
      "grad_norm": 7.53125,
      "learning_rate": 0.00013672856057091943,
      "loss": 0.8897,
      "step": 44108
    },
    {
      "epoch": 3.7949754796524133,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001367099808414674,
      "loss": 0.6821,
      "step": 44109
    },
    {
      "epoch": 3.795061515959735,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0001366914021745682,
      "loss": 1.0394,
      "step": 44110
    },
    {
      "epoch": 3.7951475522670566,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013667282457027586,
      "loss": 0.8175,
      "step": 44111
    },
    {
      "epoch": 3.795233588574378,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013665424802864485,
      "loss": 0.6651,
      "step": 44112
    },
    {
      "epoch": 3.7953196248817003,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00013663567254972958,
      "loss": 1.3859,
      "step": 44113
    },
    {
      "epoch": 3.795405661189022,
      "grad_norm": 1.125,
      "learning_rate": 0.0001366170981335842,
      "loss": 1.8734,
      "step": 44114
    },
    {
      "epoch": 3.7954916974963435,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00013659852478026314,
      "loss": 1.0062,
      "step": 44115
    },
    {
      "epoch": 3.795577733803665,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013657995248982086,
      "loss": 0.843,
      "step": 44116
    },
    {
      "epoch": 3.7956637701109868,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00013656138126231138,
      "loss": 1.6297,
      "step": 44117
    },
    {
      "epoch": 3.7957498064183084,
      "grad_norm": 1.0,
      "learning_rate": 0.00013654281109778916,
      "loss": 1.7215,
      "step": 44118
    },
    {
      "epoch": 3.79583584272563,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001365242419963087,
      "loss": 1.3626,
      "step": 44119
    },
    {
      "epoch": 3.795921879032952,
      "grad_norm": 1.625,
      "learning_rate": 0.00013650567395792396,
      "loss": 1.2541,
      "step": 44120
    },
    {
      "epoch": 3.7960079153402737,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013648710698268947,
      "loss": 1.0651,
      "step": 44121
    },
    {
      "epoch": 3.7960939516475953,
      "grad_norm": 24.75,
      "learning_rate": 0.00013646854107065965,
      "loss": 1.2144,
      "step": 44122
    },
    {
      "epoch": 3.796179987954917,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013644997622188842,
      "loss": 0.9596,
      "step": 44123
    },
    {
      "epoch": 3.7962660242622386,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00013643141243643053,
      "loss": 1.0211,
      "step": 44124
    },
    {
      "epoch": 3.79635206056956,
      "grad_norm": 1.296875,
      "learning_rate": 0.00013641284971433986,
      "loss": 1.2219,
      "step": 44125
    },
    {
      "epoch": 3.796438096876882,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0001363942880556709,
      "loss": 1.4148,
      "step": 44126
    },
    {
      "epoch": 3.796524133184204,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001363757274604781,
      "loss": 0.7955,
      "step": 44127
    },
    {
      "epoch": 3.7966101694915255,
      "grad_norm": 1.765625,
      "learning_rate": 0.00013635716792881542,
      "loss": 0.5873,
      "step": 44128
    },
    {
      "epoch": 3.796696205798847,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00013633860946073728,
      "loss": 0.7318,
      "step": 44129
    },
    {
      "epoch": 3.7967822421061688,
      "grad_norm": 0.9921875,
      "learning_rate": 0.000136320052056298,
      "loss": 1.4275,
      "step": 44130
    },
    {
      "epoch": 3.7968682784134904,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00013630149571555177,
      "loss": 0.96,
      "step": 44131
    },
    {
      "epoch": 3.796954314720812,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013628294043855317,
      "loss": 1.0347,
      "step": 44132
    },
    {
      "epoch": 3.7970403510281336,
      "grad_norm": 1.265625,
      "learning_rate": 0.00013626438622535604,
      "loss": 1.4572,
      "step": 44133
    },
    {
      "epoch": 3.7971263873354557,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013624583307601484,
      "loss": 0.5972,
      "step": 44134
    },
    {
      "epoch": 3.7972124236427773,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013622728099058396,
      "loss": 0.7968,
      "step": 44135
    },
    {
      "epoch": 3.797298459950099,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00013620872996911743,
      "loss": 1.2502,
      "step": 44136
    },
    {
      "epoch": 3.7973844962574206,
      "grad_norm": 1.09375,
      "learning_rate": 0.00013619018001166956,
      "loss": 1.4129,
      "step": 44137
    },
    {
      "epoch": 3.797470532564742,
      "grad_norm": 1.0,
      "learning_rate": 0.00013617163111829484,
      "loss": 1.2671,
      "step": 44138
    },
    {
      "epoch": 3.7975565688720643,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00013615308328904718,
      "loss": 0.6382,
      "step": 44139
    },
    {
      "epoch": 3.7976426051793855,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001361345365239811,
      "loss": 0.6548,
      "step": 44140
    },
    {
      "epoch": 3.7977286414867075,
      "grad_norm": 9.1875,
      "learning_rate": 0.00013611599082315068,
      "loss": 1.3681,
      "step": 44141
    },
    {
      "epoch": 3.797814677794029,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013609744618661013,
      "loss": 1.016,
      "step": 44142
    },
    {
      "epoch": 3.7979007141013508,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00013607890261441396,
      "loss": 0.4418,
      "step": 44143
    },
    {
      "epoch": 3.7979867504086724,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00013606036010661604,
      "loss": 1.5845,
      "step": 44144
    },
    {
      "epoch": 3.798072786715994,
      "grad_norm": 1.265625,
      "learning_rate": 0.00013604181866327082,
      "loss": 1.5491,
      "step": 44145
    },
    {
      "epoch": 3.798158823023316,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00013602327828443262,
      "loss": 1.7201,
      "step": 44146
    },
    {
      "epoch": 3.7982448593306373,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013600473897015542,
      "loss": 0.5747,
      "step": 44147
    },
    {
      "epoch": 3.7983308956379593,
      "grad_norm": 27.25,
      "learning_rate": 0.00013598620072049355,
      "loss": 1.5623,
      "step": 44148
    },
    {
      "epoch": 3.798416931945281,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00013596766353550139,
      "loss": 0.838,
      "step": 44149
    },
    {
      "epoch": 3.7985029682526026,
      "grad_norm": 4256.0,
      "learning_rate": 0.00013594912741523286,
      "loss": 1.1562,
      "step": 44150
    },
    {
      "epoch": 3.798589004559924,
      "grad_norm": 2.234375,
      "learning_rate": 0.00013593059235974232,
      "loss": 1.4869,
      "step": 44151
    },
    {
      "epoch": 3.798675040867246,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00013591205836908417,
      "loss": 1.4348,
      "step": 44152
    },
    {
      "epoch": 3.798761077174568,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013589352544331228,
      "loss": 1.4954,
      "step": 44153
    },
    {
      "epoch": 3.798847113481889,
      "grad_norm": 1.7578125,
      "learning_rate": 0.00013587499358248112,
      "loss": 1.7858,
      "step": 44154
    },
    {
      "epoch": 3.798933149789211,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001358564627866447,
      "loss": 0.9577,
      "step": 44155
    },
    {
      "epoch": 3.7990191860965328,
      "grad_norm": 1.015625,
      "learning_rate": 0.00013583793305585728,
      "loss": 0.8983,
      "step": 44156
    },
    {
      "epoch": 3.7991052224038544,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0001358194043901732,
      "loss": 1.4838,
      "step": 44157
    },
    {
      "epoch": 3.799191258711176,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00013580087678964641,
      "loss": 1.4182,
      "step": 44158
    },
    {
      "epoch": 3.7992772950184976,
      "grad_norm": 1.359375,
      "learning_rate": 0.00013578235025433122,
      "loss": 1.6983,
      "step": 44159
    },
    {
      "epoch": 3.7993633313258197,
      "grad_norm": 1.328125,
      "learning_rate": 0.000135763824784282,
      "loss": 1.14,
      "step": 44160
    },
    {
      "epoch": 3.7994493676331413,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00013574530037955258,
      "loss": 1.0172,
      "step": 44161
    },
    {
      "epoch": 3.799535403940463,
      "grad_norm": 3.625,
      "learning_rate": 0.00013572677704019732,
      "loss": 0.8829,
      "step": 44162
    },
    {
      "epoch": 3.7996214402477846,
      "grad_norm": 6.4375,
      "learning_rate": 0.00013570825476627053,
      "loss": 0.7294,
      "step": 44163
    },
    {
      "epoch": 3.799707476555106,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00013568973355782604,
      "loss": 0.5773,
      "step": 44164
    },
    {
      "epoch": 3.799793512862428,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013567121341491844,
      "loss": 0.6438,
      "step": 44165
    },
    {
      "epoch": 3.7998795491697495,
      "grad_norm": 20.5,
      "learning_rate": 0.00013565269433760135,
      "loss": 1.0512,
      "step": 44166
    },
    {
      "epoch": 3.7999655854770715,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00013563417632592945,
      "loss": 0.4952,
      "step": 44167
    },
    {
      "epoch": 3.800051621784393,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013561565937995684,
      "loss": 1.443,
      "step": 44168
    },
    {
      "epoch": 3.8001376580917148,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00013559714349973745,
      "loss": 1.4322,
      "step": 44169
    },
    {
      "epoch": 3.8002236943990364,
      "grad_norm": 1.109375,
      "learning_rate": 0.00013557862868532566,
      "loss": 0.7627,
      "step": 44170
    },
    {
      "epoch": 3.800309730706358,
      "grad_norm": 1.328125,
      "learning_rate": 0.00013556011493677533,
      "loss": 1.8546,
      "step": 44171
    },
    {
      "epoch": 3.8003957670136796,
      "grad_norm": 1.15625,
      "learning_rate": 0.00013554160225414085,
      "loss": 1.0186,
      "step": 44172
    },
    {
      "epoch": 3.8004818033210013,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00013552309063747642,
      "loss": 1.2363,
      "step": 44173
    },
    {
      "epoch": 3.8005678396283233,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013550458008683585,
      "loss": 0.9481,
      "step": 44174
    },
    {
      "epoch": 3.800653875935645,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00013548607060227358,
      "loss": 1.072,
      "step": 44175
    },
    {
      "epoch": 3.8007399122429666,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00013546756218384376,
      "loss": 0.9681,
      "step": 44176
    },
    {
      "epoch": 3.800825948550288,
      "grad_norm": 37.0,
      "learning_rate": 0.0001354490548316003,
      "loss": 0.9528,
      "step": 44177
    },
    {
      "epoch": 3.80091198485761,
      "grad_norm": 1.28125,
      "learning_rate": 0.00013543054854559745,
      "loss": 1.8212,
      "step": 44178
    },
    {
      "epoch": 3.8009980211649315,
      "grad_norm": 424.0,
      "learning_rate": 0.00013541204332588952,
      "loss": 1.1053,
      "step": 44179
    },
    {
      "epoch": 3.801084057472253,
      "grad_norm": 1.125,
      "learning_rate": 0.00013539353917253023,
      "loss": 1.0897,
      "step": 44180
    },
    {
      "epoch": 3.801170093779575,
      "grad_norm": 2.015625,
      "learning_rate": 0.00013537503608557411,
      "loss": 1.4026,
      "step": 44181
    },
    {
      "epoch": 3.801256130086897,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00013535653406507498,
      "loss": 1.1329,
      "step": 44182
    },
    {
      "epoch": 3.8013421663942184,
      "grad_norm": 1.203125,
      "learning_rate": 0.00013533803311108704,
      "loss": 0.6757,
      "step": 44183
    },
    {
      "epoch": 3.80142820270154,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001353195332236646,
      "loss": 1.7168,
      "step": 44184
    },
    {
      "epoch": 3.8015142390088617,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001353010344028614,
      "loss": 0.9087,
      "step": 44185
    },
    {
      "epoch": 3.8016002753161833,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001352825366487318,
      "loss": 1.0451,
      "step": 44186
    },
    {
      "epoch": 3.801686311623505,
      "grad_norm": 1.03125,
      "learning_rate": 0.00013526403996132997,
      "loss": 0.902,
      "step": 44187
    },
    {
      "epoch": 3.801772347930827,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00013524554434070972,
      "loss": 0.803,
      "step": 44188
    },
    {
      "epoch": 3.8018583842381486,
      "grad_norm": 8.0625,
      "learning_rate": 0.00013522704978692529,
      "loss": 1.3611,
      "step": 44189
    },
    {
      "epoch": 3.80194442054547,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00013520855630003105,
      "loss": 1.2869,
      "step": 44190
    },
    {
      "epoch": 3.802030456852792,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001351900638800806,
      "loss": 1.0719,
      "step": 44191
    },
    {
      "epoch": 3.8021164931601135,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00013517157252712826,
      "loss": 1.4939,
      "step": 44192
    },
    {
      "epoch": 3.8022025294674355,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00013515308224122828,
      "loss": 1.5259,
      "step": 44193
    },
    {
      "epoch": 3.8022885657747567,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001351345930224344,
      "loss": 1.573,
      "step": 44194
    },
    {
      "epoch": 3.802374602082079,
      "grad_norm": 972.0,
      "learning_rate": 0.00013511610487080105,
      "loss": 1.01,
      "step": 44195
    },
    {
      "epoch": 3.8024606383894004,
      "grad_norm": 1.015625,
      "learning_rate": 0.000135097617786382,
      "loss": 0.8574,
      "step": 44196
    },
    {
      "epoch": 3.802546674696722,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00013507913176923143,
      "loss": 2.0433,
      "step": 44197
    },
    {
      "epoch": 3.8026327110040437,
      "grad_norm": 1.078125,
      "learning_rate": 0.00013506064681940362,
      "loss": 1.0821,
      "step": 44198
    },
    {
      "epoch": 3.8027187473113653,
      "grad_norm": 15.5625,
      "learning_rate": 0.00013504216293695222,
      "loss": 0.8623,
      "step": 44199
    },
    {
      "epoch": 3.8028047836186873,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00013502368012193155,
      "loss": 1.8579,
      "step": 44200
    },
    {
      "epoch": 3.8028908199260085,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00013500519837439574,
      "loss": 0.9396,
      "step": 44201
    },
    {
      "epoch": 3.8029768562333306,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013498671769439862,
      "loss": 1.2788,
      "step": 44202
    },
    {
      "epoch": 3.803062892540652,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013496823808199437,
      "loss": 1.355,
      "step": 44203
    },
    {
      "epoch": 3.803148928847974,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00013494975953723704,
      "loss": 1.1152,
      "step": 44204
    },
    {
      "epoch": 3.8032349651552955,
      "grad_norm": 1.390625,
      "learning_rate": 0.00013493128206018063,
      "loss": 0.7582,
      "step": 44205
    },
    {
      "epoch": 3.803321001462617,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00013491280565087937,
      "loss": 1.4353,
      "step": 44206
    },
    {
      "epoch": 3.803407037769939,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013489433030938703,
      "loss": 0.9166,
      "step": 44207
    },
    {
      "epoch": 3.8034930740772603,
      "grad_norm": 6.3125,
      "learning_rate": 0.0001348758560357577,
      "loss": 1.6203,
      "step": 44208
    },
    {
      "epoch": 3.8035791103845824,
      "grad_norm": 1.15625,
      "learning_rate": 0.00013485738283004566,
      "loss": 1.2421,
      "step": 44209
    },
    {
      "epoch": 3.803665146691904,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00013483891069230463,
      "loss": 1.1318,
      "step": 44210
    },
    {
      "epoch": 3.8037511829992257,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00013482043962258894,
      "loss": 1.0485,
      "step": 44211
    },
    {
      "epoch": 3.8038372193065473,
      "grad_norm": 1.21875,
      "learning_rate": 0.00013480196962095225,
      "loss": 1.6408,
      "step": 44212
    },
    {
      "epoch": 3.803923255613869,
      "grad_norm": 1.59375,
      "learning_rate": 0.00013478350068744878,
      "loss": 1.5607,
      "step": 44213
    },
    {
      "epoch": 3.804009291921191,
      "grad_norm": 360.0,
      "learning_rate": 0.00013476503282213264,
      "loss": 0.6454,
      "step": 44214
    },
    {
      "epoch": 3.8040953282285126,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001347465660250577,
      "loss": 0.9317,
      "step": 44215
    },
    {
      "epoch": 3.804181364535834,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001347281002962779,
      "loss": 1.1874,
      "step": 44216
    },
    {
      "epoch": 3.804267400843156,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001347096356358476,
      "loss": 1.4432,
      "step": 44217
    },
    {
      "epoch": 3.8043534371504775,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00013469117204382042,
      "loss": 1.7336,
      "step": 44218
    },
    {
      "epoch": 3.804439473457799,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00013467270952025046,
      "loss": 1.2967,
      "step": 44219
    },
    {
      "epoch": 3.8045255097651207,
      "grad_norm": 187.0,
      "learning_rate": 0.00013465424806519195,
      "loss": 0.9847,
      "step": 44220
    },
    {
      "epoch": 3.804611546072443,
      "grad_norm": 52.75,
      "learning_rate": 0.00013463578767869855,
      "loss": 1.2778,
      "step": 44221
    },
    {
      "epoch": 3.8046975823797644,
      "grad_norm": 2.0625,
      "learning_rate": 0.00013461732836082458,
      "loss": 1.0325,
      "step": 44222
    },
    {
      "epoch": 3.804783618687086,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001345988701116237,
      "loss": 1.4918,
      "step": 44223
    },
    {
      "epoch": 3.8048696549944077,
      "grad_norm": 1104.0,
      "learning_rate": 0.00013458041293115008,
      "loss": 0.9898,
      "step": 44224
    },
    {
      "epoch": 3.8049556913017293,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001345619568194578,
      "loss": 1.6022,
      "step": 44225
    },
    {
      "epoch": 3.805041727609051,
      "grad_norm": 1.015625,
      "learning_rate": 0.00013454350177660056,
      "loss": 0.9513,
      "step": 44226
    },
    {
      "epoch": 3.8051277639163725,
      "grad_norm": 1376.0,
      "learning_rate": 0.00013452504780263257,
      "loss": 0.8381,
      "step": 44227
    },
    {
      "epoch": 3.8052138002236946,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001345065948976078,
      "loss": 1.8217,
      "step": 44228
    },
    {
      "epoch": 3.8052998365310162,
      "grad_norm": 11.0,
      "learning_rate": 0.00013448814306158007,
      "loss": 0.7502,
      "step": 44229
    },
    {
      "epoch": 3.805385872838338,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00013446969229460344,
      "loss": 0.8256,
      "step": 44230
    },
    {
      "epoch": 3.8054719091456595,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00013445124259673196,
      "loss": 0.9992,
      "step": 44231
    },
    {
      "epoch": 3.805557945452981,
      "grad_norm": 1.359375,
      "learning_rate": 0.00013443279396801937,
      "loss": 1.5406,
      "step": 44232
    },
    {
      "epoch": 3.8056439817603027,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001344143464085198,
      "loss": 1.8,
      "step": 44233
    },
    {
      "epoch": 3.8057300180676243,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013439589991828726,
      "loss": 1.0551,
      "step": 44234
    },
    {
      "epoch": 3.8058160543749464,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013437745449737548,
      "loss": 1.0063,
      "step": 44235
    },
    {
      "epoch": 3.805902090682268,
      "grad_norm": 1.25,
      "learning_rate": 0.00013435901014583867,
      "loss": 1.52,
      "step": 44236
    },
    {
      "epoch": 3.8059881269895897,
      "grad_norm": 3.75,
      "learning_rate": 0.00013434056686373053,
      "loss": 1.074,
      "step": 44237
    },
    {
      "epoch": 3.8060741632969113,
      "grad_norm": 1.03125,
      "learning_rate": 0.00013432212465110504,
      "loss": 1.2237,
      "step": 44238
    },
    {
      "epoch": 3.806160199604233,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001343036835080164,
      "loss": 1.4352,
      "step": 44239
    },
    {
      "epoch": 3.8062462359115545,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001342852434345181,
      "loss": 1.8481,
      "step": 44240
    },
    {
      "epoch": 3.806332272218876,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00013426680443066462,
      "loss": 1.4309,
      "step": 44241
    },
    {
      "epoch": 3.8064183085261982,
      "grad_norm": 2.875,
      "learning_rate": 0.00013424836649650944,
      "loss": 0.7574,
      "step": 44242
    },
    {
      "epoch": 3.80650434483352,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00013422992963210668,
      "loss": 1.3656,
      "step": 44243
    },
    {
      "epoch": 3.8065903811408415,
      "grad_norm": 1.578125,
      "learning_rate": 0.00013421149383751035,
      "loss": 2.1267,
      "step": 44244
    },
    {
      "epoch": 3.806676417448163,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00013419305911277412,
      "loss": 0.6057,
      "step": 44245
    },
    {
      "epoch": 3.8067624537554847,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00013417462545795207,
      "loss": 1.1151,
      "step": 44246
    },
    {
      "epoch": 3.8068484900628063,
      "grad_norm": 2.078125,
      "learning_rate": 0.0001341561928730982,
      "loss": 1.8579,
      "step": 44247
    },
    {
      "epoch": 3.806934526370128,
      "grad_norm": 1.75,
      "learning_rate": 0.00013413776135826622,
      "loss": 1.2729,
      "step": 44248
    },
    {
      "epoch": 3.80702056267745,
      "grad_norm": 2.125,
      "learning_rate": 0.0001341193309135101,
      "loss": 1.2732,
      "step": 44249
    },
    {
      "epoch": 3.8071065989847717,
      "grad_norm": 2.453125,
      "learning_rate": 0.00013410090153888394,
      "loss": 1.6529,
      "step": 44250
    },
    {
      "epoch": 3.8071926352920933,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001340824732344413,
      "loss": 0.7352,
      "step": 44251
    },
    {
      "epoch": 3.807278671599415,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00013406404600023643,
      "loss": 1.3702,
      "step": 44252
    },
    {
      "epoch": 3.8073647079067365,
      "grad_norm": 1.0,
      "learning_rate": 0.0001340456198363229,
      "loss": 1.2259,
      "step": 44253
    },
    {
      "epoch": 3.8074507442140586,
      "grad_norm": 1.015625,
      "learning_rate": 0.00013402719474275477,
      "loss": 1.1191,
      "step": 44254
    },
    {
      "epoch": 3.80753678052138,
      "grad_norm": 1.375,
      "learning_rate": 0.00013400877071958605,
      "loss": 1.5236,
      "step": 44255
    },
    {
      "epoch": 3.807622816828702,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00013399034776687035,
      "loss": 0.5318,
      "step": 44256
    },
    {
      "epoch": 3.8077088531360235,
      "grad_norm": 4.0625,
      "learning_rate": 0.00013397192588466167,
      "loss": 0.8564,
      "step": 44257
    },
    {
      "epoch": 3.807794889443345,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013395350507301412,
      "loss": 0.6233,
      "step": 44258
    },
    {
      "epoch": 3.8078809257506667,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013393508533198118,
      "loss": 0.9576,
      "step": 44259
    },
    {
      "epoch": 3.8079669620579883,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00013391666666161695,
      "loss": 0.8146,
      "step": 44260
    },
    {
      "epoch": 3.8080529983653104,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001338982490619754,
      "loss": 0.5833,
      "step": 44261
    },
    {
      "epoch": 3.8081390346726316,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00013387983253311016,
      "loss": 1.4878,
      "step": 44262
    },
    {
      "epoch": 3.8082250709799537,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013386141707507514,
      "loss": 1.4959,
      "step": 44263
    },
    {
      "epoch": 3.8083111072872753,
      "grad_norm": 1.34375,
      "learning_rate": 0.00013384300268792448,
      "loss": 1.4548,
      "step": 44264
    },
    {
      "epoch": 3.808397143594597,
      "grad_norm": 1.15625,
      "learning_rate": 0.00013382458937171167,
      "loss": 1.1269,
      "step": 44265
    },
    {
      "epoch": 3.8084831799019185,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001338061771264908,
      "loss": 0.8771,
      "step": 44266
    },
    {
      "epoch": 3.80856921620924,
      "grad_norm": 1.171875,
      "learning_rate": 0.00013378776595231556,
      "loss": 1.3721,
      "step": 44267
    },
    {
      "epoch": 3.8086552525165622,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00013376935584923983,
      "loss": 0.6029,
      "step": 44268
    },
    {
      "epoch": 3.8087412888238834,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001337509468173177,
      "loss": 1.5885,
      "step": 44269
    },
    {
      "epoch": 3.8088273251312055,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013373253885660264,
      "loss": 0.8736,
      "step": 44270
    },
    {
      "epoch": 3.808913361438527,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013371413196714866,
      "loss": 0.7261,
      "step": 44271
    },
    {
      "epoch": 3.8089993977458487,
      "grad_norm": 1.296875,
      "learning_rate": 0.00013369572614900976,
      "loss": 0.9168,
      "step": 44272
    },
    {
      "epoch": 3.8090854340531703,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00013367732140223947,
      "loss": 1.1289,
      "step": 44273
    },
    {
      "epoch": 3.809171470360492,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001336589177268918,
      "loss": 0.7122,
      "step": 44274
    },
    {
      "epoch": 3.809257506667814,
      "grad_norm": 1.46875,
      "learning_rate": 0.00013364051512302066,
      "loss": 0.9466,
      "step": 44275
    },
    {
      "epoch": 3.8093435429751357,
      "grad_norm": 3.078125,
      "learning_rate": 0.00013362211359067966,
      "loss": 1.477,
      "step": 44276
    },
    {
      "epoch": 3.8094295792824573,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001336037131299227,
      "loss": 1.1089,
      "step": 44277
    },
    {
      "epoch": 3.809515615589779,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00013358531374080358,
      "loss": 0.9279,
      "step": 44278
    },
    {
      "epoch": 3.8096016518971005,
      "grad_norm": 2.921875,
      "learning_rate": 0.00013356691542337618,
      "loss": 1.7827,
      "step": 44279
    },
    {
      "epoch": 3.809687688204422,
      "grad_norm": 10.0,
      "learning_rate": 0.00013354851817769443,
      "loss": 1.6978,
      "step": 44280
    },
    {
      "epoch": 3.809773724511744,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00013353012200381187,
      "loss": 0.3959,
      "step": 44281
    },
    {
      "epoch": 3.809859760819066,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00013351172690178253,
      "loss": 0.8285,
      "step": 44282
    },
    {
      "epoch": 3.8099457971263875,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00013349333287166,
      "loss": 0.6863,
      "step": 44283
    },
    {
      "epoch": 3.810031833433709,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00013347493991349818,
      "loss": 1.0119,
      "step": 44284
    },
    {
      "epoch": 3.8101178697410307,
      "grad_norm": 1.421875,
      "learning_rate": 0.00013345654802735102,
      "loss": 1.1979,
      "step": 44285
    },
    {
      "epoch": 3.8102039060483524,
      "grad_norm": 1.109375,
      "learning_rate": 0.00013343815721327202,
      "loss": 1.7339,
      "step": 44286
    },
    {
      "epoch": 3.810289942355674,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013341976747131511,
      "loss": 0.7394,
      "step": 44287
    },
    {
      "epoch": 3.8103759786629956,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00013340137880153425,
      "loss": 1.3627,
      "step": 44288
    },
    {
      "epoch": 3.8104620149703177,
      "grad_norm": 1.234375,
      "learning_rate": 0.00013338299120398288,
      "loss": 1.6044,
      "step": 44289
    },
    {
      "epoch": 3.8105480512776393,
      "grad_norm": 1.140625,
      "learning_rate": 0.00013336460467871502,
      "loss": 1.8553,
      "step": 44290
    },
    {
      "epoch": 3.810634087584961,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001333462192257845,
      "loss": 0.4252,
      "step": 44291
    },
    {
      "epoch": 3.8107201238922825,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001333278348452448,
      "loss": 0.6741,
      "step": 44292
    },
    {
      "epoch": 3.810806160199604,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00013330945153715008,
      "loss": 0.75,
      "step": 44293
    },
    {
      "epoch": 3.810892196506926,
      "grad_norm": 1.828125,
      "learning_rate": 0.0001332910693015537,
      "loss": 0.7249,
      "step": 44294
    },
    {
      "epoch": 3.8109782328142474,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00013327268813850962,
      "loss": 1.5365,
      "step": 44295
    },
    {
      "epoch": 3.8110642691215695,
      "grad_norm": 1.15625,
      "learning_rate": 0.00013325430804807175,
      "loss": 1.72,
      "step": 44296
    },
    {
      "epoch": 3.811150305428891,
      "grad_norm": 0.875,
      "learning_rate": 0.00013323592903029358,
      "loss": 0.6818,
      "step": 44297
    },
    {
      "epoch": 3.8112363417362127,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013321755108522897,
      "loss": 1.2264,
      "step": 44298
    },
    {
      "epoch": 3.8113223780435344,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001331991742129318,
      "loss": 1.6056,
      "step": 44299
    },
    {
      "epoch": 3.811408414350856,
      "grad_norm": 1.0625,
      "learning_rate": 0.00013318079841345565,
      "loss": 1.2328,
      "step": 44300
    },
    {
      "epoch": 3.8114944506581776,
      "grad_norm": 16.25,
      "learning_rate": 0.00013316242368685422,
      "loss": 1.7056,
      "step": 44301
    },
    {
      "epoch": 3.8115804869654992,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001331440500331816,
      "loss": 0.9749,
      "step": 44302
    },
    {
      "epoch": 3.8116665232728213,
      "grad_norm": 1.25,
      "learning_rate": 0.00013312567745249105,
      "loss": 1.758,
      "step": 44303
    },
    {
      "epoch": 3.811752559580143,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013310730594483662,
      "loss": 0.5563,
      "step": 44304
    },
    {
      "epoch": 3.8118385958874645,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00013308893551027202,
      "loss": 1.1143,
      "step": 44305
    },
    {
      "epoch": 3.811924632194786,
      "grad_norm": 1.296875,
      "learning_rate": 0.00013307056614885083,
      "loss": 1.4784,
      "step": 44306
    },
    {
      "epoch": 3.812010668502108,
      "grad_norm": 1.21875,
      "learning_rate": 0.000133052197860627,
      "loss": 0.9346,
      "step": 44307
    },
    {
      "epoch": 3.8120967048094294,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000133033830645654,
      "loss": 0.7623,
      "step": 44308
    },
    {
      "epoch": 3.812182741116751,
      "grad_norm": 25.25,
      "learning_rate": 0.00013301546450398565,
      "loss": 1.6641,
      "step": 44309
    },
    {
      "epoch": 3.812268777424073,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001329970994356759,
      "loss": 1.156,
      "step": 44310
    },
    {
      "epoch": 3.8123548137313947,
      "grad_norm": 1.21875,
      "learning_rate": 0.00013297873544077803,
      "loss": 0.94,
      "step": 44311
    },
    {
      "epoch": 3.8124408500387164,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00013296037251934602,
      "loss": 1.9676,
      "step": 44312
    },
    {
      "epoch": 3.812526886346038,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00013294201067143364,
      "loss": 0.5691,
      "step": 44313
    },
    {
      "epoch": 3.8126129226533596,
      "grad_norm": 1.1875,
      "learning_rate": 0.00013292364989709422,
      "loss": 0.6799,
      "step": 44314
    },
    {
      "epoch": 3.8126989589606817,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00013290529019638204,
      "loss": 0.7133,
      "step": 44315
    },
    {
      "epoch": 3.812784995268003,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013288693156935033,
      "loss": 1.3499,
      "step": 44316
    },
    {
      "epoch": 3.812871031575325,
      "grad_norm": 4.78125,
      "learning_rate": 0.00013286857401605297,
      "loss": 1.3132,
      "step": 44317
    },
    {
      "epoch": 3.8129570678826465,
      "grad_norm": 2.796875,
      "learning_rate": 0.00013285021753654374,
      "loss": 0.4403,
      "step": 44318
    },
    {
      "epoch": 3.813043104189968,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001328318621308761,
      "loss": 1.4651,
      "step": 44319
    },
    {
      "epoch": 3.81312914049729,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013281350779910384,
      "loss": 0.5694,
      "step": 44320
    },
    {
      "epoch": 3.8132151768046114,
      "grad_norm": 2.09375,
      "learning_rate": 0.00013279515454128076,
      "loss": 1.6199,
      "step": 44321
    },
    {
      "epoch": 3.8133012131119335,
      "grad_norm": 82.5,
      "learning_rate": 0.00013277680235746037,
      "loss": 0.8349,
      "step": 44322
    },
    {
      "epoch": 3.8133872494192547,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00013275845124769654,
      "loss": 1.2464,
      "step": 44323
    },
    {
      "epoch": 3.8134732857265767,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001327401012120426,
      "loss": 1.7452,
      "step": 44324
    },
    {
      "epoch": 3.8135593220338984,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001327217522505525,
      "loss": 0.9523,
      "step": 44325
    },
    {
      "epoch": 3.81364535834122,
      "grad_norm": 1.4375,
      "learning_rate": 0.00013270340436328,
      "loss": 1.6885,
      "step": 44326
    },
    {
      "epoch": 3.8137313946485416,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00013268505755027838,
      "loss": 1.8684,
      "step": 44327
    },
    {
      "epoch": 3.8138174309558632,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001326667118116016,
      "loss": 0.688,
      "step": 44328
    },
    {
      "epoch": 3.8139034672631853,
      "grad_norm": 3.625,
      "learning_rate": 0.00013264836714730339,
      "loss": 1.3071,
      "step": 44329
    },
    {
      "epoch": 3.8139895035705065,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001326300235574371,
      "loss": 1.1149,
      "step": 44330
    },
    {
      "epoch": 3.8140755398778285,
      "grad_norm": 544.0,
      "learning_rate": 0.0001326116810420565,
      "loss": 1.3203,
      "step": 44331
    },
    {
      "epoch": 3.81416157618515,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001325933396012155,
      "loss": 0.7235,
      "step": 44332
    },
    {
      "epoch": 3.814247612492472,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0001325749992349673,
      "loss": 0.7808,
      "step": 44333
    },
    {
      "epoch": 3.8143336487997934,
      "grad_norm": 1.25,
      "learning_rate": 0.00013255665994336597,
      "loss": 1.5539,
      "step": 44334
    },
    {
      "epoch": 3.814419685107115,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001325383217264648,
      "loss": 0.901,
      "step": 44335
    },
    {
      "epoch": 3.814505721414437,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013251998458431758,
      "loss": 0.4381,
      "step": 44336
    },
    {
      "epoch": 3.8145917577217587,
      "grad_norm": 1.328125,
      "learning_rate": 0.00013250164851697806,
      "loss": 1.591,
      "step": 44337
    },
    {
      "epoch": 3.8146777940290804,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00013248331352449955,
      "loss": 2.1506,
      "step": 44338
    },
    {
      "epoch": 3.814763830336402,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00013246497960693593,
      "loss": 1.7238,
      "step": 44339
    },
    {
      "epoch": 3.8148498666437236,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00013244664676434088,
      "loss": 0.6732,
      "step": 44340
    },
    {
      "epoch": 3.8149359029510452,
      "grad_norm": 1.140625,
      "learning_rate": 0.00013242831499676776,
      "loss": 1.5715,
      "step": 44341
    },
    {
      "epoch": 3.815021939258367,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00013240998430427036,
      "loss": 0.6936,
      "step": 44342
    },
    {
      "epoch": 3.815107975565689,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001323916546869024,
      "loss": 1.7873,
      "step": 44343
    },
    {
      "epoch": 3.8151940118730105,
      "grad_norm": 10.875,
      "learning_rate": 0.0001323733261447172,
      "loss": 1.427,
      "step": 44344
    },
    {
      "epoch": 3.815280048180332,
      "grad_norm": 1.203125,
      "learning_rate": 0.00013235499867776852,
      "loss": 2.1577,
      "step": 44345
    },
    {
      "epoch": 3.815366084487654,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00013233667228611012,
      "loss": 0.8571,
      "step": 44346
    },
    {
      "epoch": 3.8154521207949754,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001323183469697953,
      "loss": 0.7137,
      "step": 44347
    },
    {
      "epoch": 3.815538157102297,
      "grad_norm": 1.6875,
      "learning_rate": 0.00013230002272887793,
      "loss": 1.1106,
      "step": 44348
    },
    {
      "epoch": 3.8156241934096187,
      "grad_norm": 1.0625,
      "learning_rate": 0.00013228169956341134,
      "loss": 1.2726,
      "step": 44349
    },
    {
      "epoch": 3.8157102297169407,
      "grad_norm": 1.03125,
      "learning_rate": 0.00013226337747344928,
      "loss": 1.4183,
      "step": 44350
    },
    {
      "epoch": 3.8157962660242624,
      "grad_norm": 2.390625,
      "learning_rate": 0.00013224505645904533,
      "loss": 1.6292,
      "step": 44351
    },
    {
      "epoch": 3.815882302331584,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013222673652025303,
      "loss": 1.3811,
      "step": 44352
    },
    {
      "epoch": 3.8159683386389056,
      "grad_norm": 11.3125,
      "learning_rate": 0.00013220841765712615,
      "loss": 1.2223,
      "step": 44353
    },
    {
      "epoch": 3.8160543749462272,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00013219009986971798,
      "loss": 0.6843,
      "step": 44354
    },
    {
      "epoch": 3.816140411253549,
      "grad_norm": 1.125,
      "learning_rate": 0.00013217178315808221,
      "loss": 0.9299,
      "step": 44355
    },
    {
      "epoch": 3.8162264475608705,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00013215346752227264,
      "loss": 0.4393,
      "step": 44356
    },
    {
      "epoch": 3.8163124838681926,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00013213515296234247,
      "loss": 1.1269,
      "step": 44357
    },
    {
      "epoch": 3.816398520175514,
      "grad_norm": 1.46875,
      "learning_rate": 0.00013211683947834542,
      "loss": 1.5711,
      "step": 44358
    },
    {
      "epoch": 3.816484556482836,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00013209852707033514,
      "loss": 0.6596,
      "step": 44359
    },
    {
      "epoch": 3.8165705927901574,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001320802157383651,
      "loss": 1.6893,
      "step": 44360
    },
    {
      "epoch": 3.816656629097479,
      "grad_norm": 1.140625,
      "learning_rate": 0.00013206190548248876,
      "loss": 1.5753,
      "step": 44361
    },
    {
      "epoch": 3.8167426654048007,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00013204359630276003,
      "loss": 0.912,
      "step": 44362
    },
    {
      "epoch": 3.8168287017121223,
      "grad_norm": 1.515625,
      "learning_rate": 0.00013202528819923197,
      "loss": 1.6576,
      "step": 44363
    },
    {
      "epoch": 3.8169147380194444,
      "grad_norm": 1.25,
      "learning_rate": 0.0001320069811719586,
      "loss": 0.9896,
      "step": 44364
    },
    {
      "epoch": 3.817000774326766,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00013198867522099306,
      "loss": 1.5818,
      "step": 44365
    },
    {
      "epoch": 3.8170868106340876,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013197037034638908,
      "loss": 0.7424,
      "step": 44366
    },
    {
      "epoch": 3.8171728469414092,
      "grad_norm": 1.21875,
      "learning_rate": 0.00013195206654820037,
      "loss": 0.7914,
      "step": 44367
    },
    {
      "epoch": 3.817258883248731,
      "grad_norm": 7.4375,
      "learning_rate": 0.00013193376382648002,
      "loss": 1.4519,
      "step": 44368
    },
    {
      "epoch": 3.817344919556053,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001319154621812819,
      "loss": 0.8188,
      "step": 44369
    },
    {
      "epoch": 3.817430955863374,
      "grad_norm": 1.0,
      "learning_rate": 0.00013189716161265968,
      "loss": 1.4671,
      "step": 44370
    },
    {
      "epoch": 3.817516992170696,
      "grad_norm": 1.34375,
      "learning_rate": 0.00013187886212066646,
      "loss": 1.8301,
      "step": 44371
    },
    {
      "epoch": 3.817603028478018,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000131860563705356,
      "loss": 1.7802,
      "step": 44372
    },
    {
      "epoch": 3.8176890647853394,
      "grad_norm": 748.0,
      "learning_rate": 0.00013184226636678186,
      "loss": 1.1301,
      "step": 44373
    },
    {
      "epoch": 3.817775101092661,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001318239701049974,
      "loss": 1.1136,
      "step": 44374
    },
    {
      "epoch": 3.8178611373999827,
      "grad_norm": 1.1875,
      "learning_rate": 0.00013180567492005634,
      "loss": 0.8318,
      "step": 44375
    },
    {
      "epoch": 3.8179471737073047,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00013178738081201196,
      "loss": 1.6312,
      "step": 44376
    },
    {
      "epoch": 3.818033210014626,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001317690877809179,
      "loss": 1.9172,
      "step": 44377
    },
    {
      "epoch": 3.818119246321948,
      "grad_norm": 1.0,
      "learning_rate": 0.00013175079582682775,
      "loss": 0.3808,
      "step": 44378
    },
    {
      "epoch": 3.8182052826292696,
      "grad_norm": 10.125,
      "learning_rate": 0.00013173250494979472,
      "loss": 1.0147,
      "step": 44379
    },
    {
      "epoch": 3.8182913189365912,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013171421514987253,
      "loss": 0.5824,
      "step": 44380
    },
    {
      "epoch": 3.818377355243913,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001316959264271148,
      "loss": 0.3614,
      "step": 44381
    },
    {
      "epoch": 3.8184633915512345,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001316776387815747,
      "loss": 1.0842,
      "step": 44382
    },
    {
      "epoch": 3.8185494278585566,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00013165935221330584,
      "loss": 1.4608,
      "step": 44383
    },
    {
      "epoch": 3.8186354641658777,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013164106672236187,
      "loss": 0.6471,
      "step": 44384
    },
    {
      "epoch": 3.8187215004732,
      "grad_norm": 1.390625,
      "learning_rate": 0.00013162278230879605,
      "loss": 0.9085,
      "step": 44385
    },
    {
      "epoch": 3.8188075367805214,
      "grad_norm": 4.6875,
      "learning_rate": 0.00013160449897266192,
      "loss": 0.8475,
      "step": 44386
    },
    {
      "epoch": 3.818893573087843,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00013158621671401294,
      "loss": 0.8843,
      "step": 44387
    },
    {
      "epoch": 3.8189796093951647,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001315679355329027,
      "loss": 1.4535,
      "step": 44388
    },
    {
      "epoch": 3.8190656457024863,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00013154965542938469,
      "loss": 0.8939,
      "step": 44389
    },
    {
      "epoch": 3.8191516820098084,
      "grad_norm": 95.0,
      "learning_rate": 0.00013153137640351214,
      "loss": 1.0073,
      "step": 44390
    },
    {
      "epoch": 3.81923771831713,
      "grad_norm": 1.0625,
      "learning_rate": 0.00013151309845533865,
      "loss": 1.4388,
      "step": 44391
    },
    {
      "epoch": 3.8193237546244516,
      "grad_norm": 1.96875,
      "learning_rate": 0.0001314948215849178,
      "loss": 1.2596,
      "step": 44392
    },
    {
      "epoch": 3.8194097909317732,
      "grad_norm": 0.984375,
      "learning_rate": 0.00013147654579230285,
      "loss": 1.0018,
      "step": 44393
    },
    {
      "epoch": 3.819495827239095,
      "grad_norm": 1.234375,
      "learning_rate": 0.00013145827107754744,
      "loss": 1.0721,
      "step": 44394
    },
    {
      "epoch": 3.8195818635464165,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00013143999744070472,
      "loss": 1.3416,
      "step": 44395
    },
    {
      "epoch": 3.819667899853738,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013142172488182836,
      "loss": 0.945,
      "step": 44396
    },
    {
      "epoch": 3.81975393616106,
      "grad_norm": 0.859375,
      "learning_rate": 0.00013140345340097192,
      "loss": 1.0991,
      "step": 44397
    },
    {
      "epoch": 3.819839972468382,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00013138518299818857,
      "loss": 0.8399,
      "step": 44398
    },
    {
      "epoch": 3.8199260087757034,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00013136691367353183,
      "loss": 0.8519,
      "step": 44399
    },
    {
      "epoch": 3.820012045083025,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00013134864542705532,
      "loss": 1.3018,
      "step": 44400
    },
    {
      "epoch": 3.8200980813903467,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013133037825881222,
      "loss": 0.9386,
      "step": 44401
    },
    {
      "epoch": 3.8201841176976683,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00013131211216885604,
      "loss": 1.124,
      "step": 44402
    },
    {
      "epoch": 3.82027015400499,
      "grad_norm": 3.875,
      "learning_rate": 0.00013129384715724035,
      "loss": 1.1759,
      "step": 44403
    },
    {
      "epoch": 3.820356190312312,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00013127558322401827,
      "loss": 0.783,
      "step": 44404
    },
    {
      "epoch": 3.8204422266196336,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00013125732036924366,
      "loss": 1.5964,
      "step": 44405
    },
    {
      "epoch": 3.8205282629269552,
      "grad_norm": 2.484375,
      "learning_rate": 0.0001312390585929694,
      "loss": 1.2173,
      "step": 44406
    },
    {
      "epoch": 3.820614299234277,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013122079789524926,
      "loss": 0.7531,
      "step": 44407
    },
    {
      "epoch": 3.8207003355415985,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013120253827613666,
      "loss": 0.5318,
      "step": 44408
    },
    {
      "epoch": 3.82078637184892,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013118427973568474,
      "loss": 0.9416,
      "step": 44409
    },
    {
      "epoch": 3.8208724081562417,
      "grad_norm": 26.75,
      "learning_rate": 0.00013116602227394714,
      "loss": 1.024,
      "step": 44410
    },
    {
      "epoch": 3.820958444463564,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013114776589097732,
      "loss": 1.1593,
      "step": 44411
    },
    {
      "epoch": 3.8210444807708854,
      "grad_norm": 416.0,
      "learning_rate": 0.00013112951058682836,
      "loss": 1.5132,
      "step": 44412
    },
    {
      "epoch": 3.821130517078207,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00013111125636155386,
      "loss": 1.0871,
      "step": 44413
    },
    {
      "epoch": 3.8212165533855287,
      "grad_norm": 1.15625,
      "learning_rate": 0.00013109300321520733,
      "loss": 1.7995,
      "step": 44414
    },
    {
      "epoch": 3.8213025896928503,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001310747511478419,
      "loss": 0.491,
      "step": 44415
    },
    {
      "epoch": 3.821388626000172,
      "grad_norm": 1.703125,
      "learning_rate": 0.00013105650015951108,
      "loss": 1.1229,
      "step": 44416
    },
    {
      "epoch": 3.8214746623074936,
      "grad_norm": 1.0625,
      "learning_rate": 0.00013103825025026833,
      "loss": 0.7432,
      "step": 44417
    },
    {
      "epoch": 3.8215606986148156,
      "grad_norm": 1.421875,
      "learning_rate": 0.00013102000142016686,
      "loss": 0.58,
      "step": 44418
    },
    {
      "epoch": 3.8216467349221372,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001310017536692602,
      "loss": 1.0547,
      "step": 44419
    },
    {
      "epoch": 3.821732771229459,
      "grad_norm": 176.0,
      "learning_rate": 0.00013098350699760152,
      "loss": 0.9263,
      "step": 44420
    },
    {
      "epoch": 3.8218188075367805,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00013096526140524435,
      "loss": 0.6009,
      "step": 44421
    },
    {
      "epoch": 3.821904843844102,
      "grad_norm": 1.234375,
      "learning_rate": 0.00013094701689224215,
      "loss": 0.9039,
      "step": 44422
    },
    {
      "epoch": 3.8219908801514237,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000130928773458648,
      "loss": 1.0341,
      "step": 44423
    },
    {
      "epoch": 3.8220769164587454,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001309105311045154,
      "loss": 1.4899,
      "step": 44424
    },
    {
      "epoch": 3.8221629527660674,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001308922898298977,
      "loss": 0.8499,
      "step": 44425
    },
    {
      "epoch": 3.822248989073389,
      "grad_norm": 61.0,
      "learning_rate": 0.00013087404963484827,
      "loss": 1.1865,
      "step": 44426
    },
    {
      "epoch": 3.8223350253807107,
      "grad_norm": 205.0,
      "learning_rate": 0.00013085581051942058,
      "loss": 1.4407,
      "step": 44427
    },
    {
      "epoch": 3.8224210616880323,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013083757248366772,
      "loss": 0.7964,
      "step": 44428
    },
    {
      "epoch": 3.822507097995354,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001308193355276432,
      "loss": 1.3814,
      "step": 44429
    },
    {
      "epoch": 3.822593134302676,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00013080109965140046,
      "loss": 0.8896,
      "step": 44430
    },
    {
      "epoch": 3.822679170609997,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00013078286485499252,
      "loss": 1.6304,
      "step": 44431
    },
    {
      "epoch": 3.8227652069173192,
      "grad_norm": 8.8125,
      "learning_rate": 0.00013076463113847293,
      "loss": 0.7401,
      "step": 44432
    },
    {
      "epoch": 3.822851243224641,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00013074639850189508,
      "loss": 0.7645,
      "step": 44433
    },
    {
      "epoch": 3.8229372795319625,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00013072816694531208,
      "loss": 1.3225,
      "step": 44434
    },
    {
      "epoch": 3.823023315839284,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00013070993646877748,
      "loss": 0.9436,
      "step": 44435
    },
    {
      "epoch": 3.8231093521466057,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001306917070723444,
      "loss": 1.3537,
      "step": 44436
    },
    {
      "epoch": 3.823195388453928,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00013067347875606622,
      "loss": 1.0431,
      "step": 44437
    },
    {
      "epoch": 3.823281424761249,
      "grad_norm": 43.25,
      "learning_rate": 0.00013065525151999646,
      "loss": 0.9513,
      "step": 44438
    },
    {
      "epoch": 3.823367461068571,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001306370253641881,
      "loss": 0.9648,
      "step": 44439
    },
    {
      "epoch": 3.8234534973758927,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00013061880028869454,
      "loss": 1.0782,
      "step": 44440
    },
    {
      "epoch": 3.8235395336832143,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013060057629356937,
      "loss": 1.002,
      "step": 44441
    },
    {
      "epoch": 3.823625569990536,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001305823533788655,
      "loss": 1.148,
      "step": 44442
    },
    {
      "epoch": 3.8237116062978576,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001305641315446364,
      "loss": 1.0556,
      "step": 44443
    },
    {
      "epoch": 3.8237976426051796,
      "grad_norm": 7.34375,
      "learning_rate": 0.00013054591079093554,
      "loss": 0.9228,
      "step": 44444
    },
    {
      "epoch": 3.823883678912501,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013052769111781583,
      "loss": 1.1255,
      "step": 44445
    },
    {
      "epoch": 3.823969715219823,
      "grad_norm": 1.359375,
      "learning_rate": 0.00013050947252533096,
      "loss": 1.3744,
      "step": 44446
    },
    {
      "epoch": 3.8240557515271445,
      "grad_norm": 9.625,
      "learning_rate": 0.0001304912550135338,
      "loss": 1.0368,
      "step": 44447
    },
    {
      "epoch": 3.824141787834466,
      "grad_norm": 5.6875,
      "learning_rate": 0.00013047303858247793,
      "loss": 0.999,
      "step": 44448
    },
    {
      "epoch": 3.8242278241417877,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001304548232322167,
      "loss": 0.7635,
      "step": 44449
    },
    {
      "epoch": 3.8243138604491094,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00013043660896280306,
      "loss": 1.0821,
      "step": 44450
    },
    {
      "epoch": 3.8243998967564314,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001304183957742905,
      "loss": 1.1769,
      "step": 44451
    },
    {
      "epoch": 3.824485933063753,
      "grad_norm": 129.0,
      "learning_rate": 0.00013040018366673235,
      "loss": 1.5672,
      "step": 44452
    },
    {
      "epoch": 3.8245719693710747,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013038197264018164,
      "loss": 0.658,
      "step": 44453
    },
    {
      "epoch": 3.8246580056783963,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001303637626946918,
      "loss": 0.8724,
      "step": 44454
    },
    {
      "epoch": 3.824744041985718,
      "grad_norm": 1.375,
      "learning_rate": 0.00013034555383031622,
      "loss": 1.4528,
      "step": 44455
    },
    {
      "epoch": 3.8248300782930396,
      "grad_norm": 0.921875,
      "learning_rate": 0.00013032734604710784,
      "loss": 0.8953,
      "step": 44456
    },
    {
      "epoch": 3.824916114600361,
      "grad_norm": 2.78125,
      "learning_rate": 0.00013030913934512005,
      "loss": 1.0026,
      "step": 44457
    },
    {
      "epoch": 3.8250021509076833,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001302909337244063,
      "loss": 1.4407,
      "step": 44458
    },
    {
      "epoch": 3.825088187215005,
      "grad_norm": 1.0,
      "learning_rate": 0.00013027272918501954,
      "loss": 1.2759,
      "step": 44459
    },
    {
      "epoch": 3.8251742235223265,
      "grad_norm": 1952.0,
      "learning_rate": 0.00013025452572701325,
      "loss": 1.1665,
      "step": 44460
    },
    {
      "epoch": 3.825260259829648,
      "grad_norm": 1.265625,
      "learning_rate": 0.00013023632335044027,
      "loss": 1.0085,
      "step": 44461
    },
    {
      "epoch": 3.8253462961369697,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00013021812205535444,
      "loss": 1.6723,
      "step": 44462
    },
    {
      "epoch": 3.8254323324442914,
      "grad_norm": 163.0,
      "learning_rate": 0.0001301999218418085,
      "loss": 0.7144,
      "step": 44463
    },
    {
      "epoch": 3.825518368751613,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001301817227098559,
      "loss": 0.75,
      "step": 44464
    },
    {
      "epoch": 3.825604405058935,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013016352465954994,
      "loss": 0.8702,
      "step": 44465
    },
    {
      "epoch": 3.8256904413662567,
      "grad_norm": 2.21875,
      "learning_rate": 0.00013014532769094368,
      "loss": 0.5808,
      "step": 44466
    },
    {
      "epoch": 3.8257764776735783,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001301271318040903,
      "loss": 1.0814,
      "step": 44467
    },
    {
      "epoch": 3.8258625139809,
      "grad_norm": 2.875,
      "learning_rate": 0.00013010893699904336,
      "loss": 1.093,
      "step": 44468
    },
    {
      "epoch": 3.8259485502882216,
      "grad_norm": 29.875,
      "learning_rate": 0.00013009074327585563,
      "loss": 1.0714,
      "step": 44469
    },
    {
      "epoch": 3.826034586595543,
      "grad_norm": 4.4375,
      "learning_rate": 0.00013007255063458056,
      "loss": 1.1036,
      "step": 44470
    },
    {
      "epoch": 3.826120622902865,
      "grad_norm": 1.09375,
      "learning_rate": 0.00013005435907527147,
      "loss": 0.6674,
      "step": 44471
    },
    {
      "epoch": 3.826206659210187,
      "grad_norm": 1.5,
      "learning_rate": 0.0001300361685979813,
      "loss": 1.3696,
      "step": 44472
    },
    {
      "epoch": 3.8262926955175085,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00013001797920276337,
      "loss": 1.5939,
      "step": 44473
    },
    {
      "epoch": 3.82637873182483,
      "grad_norm": 1.296875,
      "learning_rate": 0.00012999979088967106,
      "loss": 1.0598,
      "step": 44474
    },
    {
      "epoch": 3.8264647681321518,
      "grad_norm": 1.09375,
      "learning_rate": 0.00012998160365875722,
      "loss": 1.2006,
      "step": 44475
    },
    {
      "epoch": 3.8265508044394734,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001299634175100754,
      "loss": 1.0855,
      "step": 44476
    },
    {
      "epoch": 3.826636840746795,
      "grad_norm": 1.375,
      "learning_rate": 0.00012994523244367844,
      "loss": 0.8689,
      "step": 44477
    },
    {
      "epoch": 3.8267228770541166,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00012992704845961973,
      "loss": 1.0624,
      "step": 44478
    },
    {
      "epoch": 3.8268089133614387,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012990886555795257,
      "loss": 1.3376,
      "step": 44479
    },
    {
      "epoch": 3.8268949496687603,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001298906837387298,
      "loss": 1.0237,
      "step": 44480
    },
    {
      "epoch": 3.826980985976082,
      "grad_norm": 292.0,
      "learning_rate": 0.00012987250300200488,
      "loss": 1.2839,
      "step": 44481
    },
    {
      "epoch": 3.8270670222834036,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00012985432334783098,
      "loss": 1.5295,
      "step": 44482
    },
    {
      "epoch": 3.827153058590725,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001298361447762611,
      "loss": 1.871,
      "step": 44483
    },
    {
      "epoch": 3.827239094898047,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001298179672873485,
      "loss": 1.5055,
      "step": 44484
    },
    {
      "epoch": 3.8273251312053684,
      "grad_norm": 1.9375,
      "learning_rate": 0.00012979979088114646,
      "loss": 0.7728,
      "step": 44485
    },
    {
      "epoch": 3.8274111675126905,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001297816155577079,
      "loss": 0.6387,
      "step": 44486
    },
    {
      "epoch": 3.827497203820012,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00012976344131708628,
      "loss": 0.67,
      "step": 44487
    },
    {
      "epoch": 3.8275832401273338,
      "grad_norm": 1.25,
      "learning_rate": 0.00012974526815933446,
      "loss": 0.7104,
      "step": 44488
    },
    {
      "epoch": 3.8276692764346554,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00012972709608450567,
      "loss": 1.0384,
      "step": 44489
    },
    {
      "epoch": 3.827755312741977,
      "grad_norm": 1.140625,
      "learning_rate": 0.00012970892509265325,
      "loss": 1.1142,
      "step": 44490
    },
    {
      "epoch": 3.827841349049299,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00012969075518383012,
      "loss": 0.9157,
      "step": 44491
    },
    {
      "epoch": 3.8279273853566202,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00012967258635808942,
      "loss": 1.3492,
      "step": 44492
    },
    {
      "epoch": 3.8280134216639423,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001296544186154846,
      "loss": 0.9118,
      "step": 44493
    },
    {
      "epoch": 3.828099457971264,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00012963625195606843,
      "loss": 0.4998,
      "step": 44494
    },
    {
      "epoch": 3.8281854942785856,
      "grad_norm": 1.0625,
      "learning_rate": 0.00012961808637989414,
      "loss": 1.0548,
      "step": 44495
    },
    {
      "epoch": 3.828271530585907,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0001295999218870151,
      "loss": 1.4281,
      "step": 44496
    },
    {
      "epoch": 3.828357566893229,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001295817584774841,
      "loss": 1.3534,
      "step": 44497
    },
    {
      "epoch": 3.828443603200551,
      "grad_norm": 1.359375,
      "learning_rate": 0.00012956359615135438,
      "loss": 0.6579,
      "step": 44498
    },
    {
      "epoch": 3.828529639507872,
      "grad_norm": 1.171875,
      "learning_rate": 0.00012954543490867915,
      "loss": 1.3671,
      "step": 44499
    },
    {
      "epoch": 3.828615675815194,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001295272747495114,
      "loss": 0.6266,
      "step": 44500
    },
    {
      "epoch": 3.8287017121225158,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012950911567390455,
      "loss": 0.5088,
      "step": 44501
    },
    {
      "epoch": 3.8287877484298374,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00012949095768191127,
      "loss": 0.975,
      "step": 44502
    },
    {
      "epoch": 3.828873784737159,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00012947280077358492,
      "loss": 0.6155,
      "step": 44503
    },
    {
      "epoch": 3.8289598210444806,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00012945464494897867,
      "loss": 0.9106,
      "step": 44504
    },
    {
      "epoch": 3.8290458573518027,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001294364902081454,
      "loss": 1.8486,
      "step": 44505
    },
    {
      "epoch": 3.829131893659124,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001294183365511385,
      "loss": 1.1355,
      "step": 44506
    },
    {
      "epoch": 3.829217929966446,
      "grad_norm": 136.0,
      "learning_rate": 0.00012940018397801067,
      "loss": 1.0005,
      "step": 44507
    },
    {
      "epoch": 3.8293039662737676,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001293820324888153,
      "loss": 1.054,
      "step": 44508
    },
    {
      "epoch": 3.829390002581089,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001293638820836055,
      "loss": 0.8007,
      "step": 44509
    },
    {
      "epoch": 3.829476038888411,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00012934573276243412,
      "loss": 1.6078,
      "step": 44510
    },
    {
      "epoch": 3.8295620751957324,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012932758452535443,
      "loss": 1.5748,
      "step": 44511
    },
    {
      "epoch": 3.8296481115030545,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00012930943737241956,
      "loss": 0.5389,
      "step": 44512
    },
    {
      "epoch": 3.829734147810376,
      "grad_norm": 0.953125,
      "learning_rate": 0.00012929129130368234,
      "loss": 0.8661,
      "step": 44513
    },
    {
      "epoch": 3.8298201841176978,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00012927314631919602,
      "loss": 1.5365,
      "step": 44514
    },
    {
      "epoch": 3.8299062204250194,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012925500241901377,
      "loss": 0.7927,
      "step": 44515
    },
    {
      "epoch": 3.829992256732341,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012923685960318837,
      "loss": 1.4773,
      "step": 44516
    },
    {
      "epoch": 3.8300782930396626,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012921871787177324,
      "loss": 0.5326,
      "step": 44517
    },
    {
      "epoch": 3.8301643293469843,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012920057722482108,
      "loss": 1.111,
      "step": 44518
    },
    {
      "epoch": 3.8302503656543063,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00012918243766238513,
      "loss": 0.7811,
      "step": 44519
    },
    {
      "epoch": 3.830336401961628,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00012916429918451855,
      "loss": 1.6159,
      "step": 44520
    },
    {
      "epoch": 3.8304224382689496,
      "grad_norm": 1.03125,
      "learning_rate": 0.00012914616179127413,
      "loss": 1.0026,
      "step": 44521
    },
    {
      "epoch": 3.830508474576271,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012912802548270509,
      "loss": 0.7979,
      "step": 44522
    },
    {
      "epoch": 3.830594510883593,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012910989025886454,
      "loss": 1.1377,
      "step": 44523
    },
    {
      "epoch": 3.8306805471909144,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012909175611980535,
      "loss": 1.1806,
      "step": 44524
    },
    {
      "epoch": 3.830766583498236,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012907362306558058,
      "loss": 1.4523,
      "step": 44525
    },
    {
      "epoch": 3.830852619805558,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001290554910962435,
      "loss": 1.5987,
      "step": 44526
    },
    {
      "epoch": 3.8309386561128798,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012903736021184676,
      "loss": 0.7925,
      "step": 44527
    },
    {
      "epoch": 3.8310246924202014,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012901923041244378,
      "loss": 1.0451,
      "step": 44528
    },
    {
      "epoch": 3.831110728727523,
      "grad_norm": 1.3125,
      "learning_rate": 0.00012900110169808727,
      "loss": 0.7727,
      "step": 44529
    },
    {
      "epoch": 3.8311967650348446,
      "grad_norm": 1.046875,
      "learning_rate": 0.00012898297406883035,
      "loss": 0.752,
      "step": 44530
    },
    {
      "epoch": 3.8312828013421663,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001289648475247262,
      "loss": 0.7916,
      "step": 44531
    },
    {
      "epoch": 3.831368837649488,
      "grad_norm": 1.953125,
      "learning_rate": 0.00012894672206582763,
      "loss": 0.7783,
      "step": 44532
    },
    {
      "epoch": 3.83145487395681,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012892859769218768,
      "loss": 1.573,
      "step": 44533
    },
    {
      "epoch": 3.8315409102641316,
      "grad_norm": 1168.0,
      "learning_rate": 0.00012891047440385955,
      "loss": 0.4648,
      "step": 44534
    },
    {
      "epoch": 3.831626946571453,
      "grad_norm": 5056.0,
      "learning_rate": 0.00012889235220089583,
      "loss": 1.1731,
      "step": 44535
    },
    {
      "epoch": 3.831712982878775,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00012887423108335011,
      "loss": 0.8191,
      "step": 44536
    },
    {
      "epoch": 3.8317990191860964,
      "grad_norm": 1.65625,
      "learning_rate": 0.00012885611105127497,
      "loss": 1.0494,
      "step": 44537
    },
    {
      "epoch": 3.831885055493418,
      "grad_norm": 233.0,
      "learning_rate": 0.00012883799210472346,
      "loss": 1.7098,
      "step": 44538
    },
    {
      "epoch": 3.8319710918007397,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001288198742437488,
      "loss": 1.1962,
      "step": 44539
    },
    {
      "epoch": 3.8320571281080618,
      "grad_norm": 1.203125,
      "learning_rate": 0.00012880175746840373,
      "loss": 0.8862,
      "step": 44540
    },
    {
      "epoch": 3.8321431644153834,
      "grad_norm": 1.4375,
      "learning_rate": 0.00012878364177874125,
      "loss": 1.655,
      "step": 44541
    },
    {
      "epoch": 3.832229200722705,
      "grad_norm": 1496.0,
      "learning_rate": 0.0001287655271748146,
      "loss": 0.8562,
      "step": 44542
    },
    {
      "epoch": 3.8323152370300266,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00012874741365667642,
      "loss": 0.9703,
      "step": 44543
    },
    {
      "epoch": 3.8324012733373483,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00012872930122437986,
      "loss": 0.7284,
      "step": 44544
    },
    {
      "epoch": 3.8324873096446703,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012871118987797803,
      "loss": 0.6725,
      "step": 44545
    },
    {
      "epoch": 3.8325733459519915,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012869307961752363,
      "loss": 0.6542,
      "step": 44546
    },
    {
      "epoch": 3.8326593822593136,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012867497044306985,
      "loss": 1.4143,
      "step": 44547
    },
    {
      "epoch": 3.832745418566635,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012865686235466944,
      "loss": 1.3211,
      "step": 44548
    },
    {
      "epoch": 3.832831454873957,
      "grad_norm": 15.9375,
      "learning_rate": 0.00012863875535237552,
      "loss": 1.3544,
      "step": 44549
    },
    {
      "epoch": 3.8329174911812784,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012862064943624114,
      "loss": 0.679,
      "step": 44550
    },
    {
      "epoch": 3.8330035274886,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012860254460631899,
      "loss": 1.2717,
      "step": 44551
    },
    {
      "epoch": 3.833089563795922,
      "grad_norm": 1.25,
      "learning_rate": 0.00012858444086266213,
      "loss": 0.6681,
      "step": 44552
    },
    {
      "epoch": 3.8331756001032433,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00012856633820532376,
      "loss": 1.202,
      "step": 44553
    },
    {
      "epoch": 3.8332616364105654,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012854823663435637,
      "loss": 0.5937,
      "step": 44554
    },
    {
      "epoch": 3.833347672717887,
      "grad_norm": 2.421875,
      "learning_rate": 0.00012853013614981323,
      "loss": 1.3107,
      "step": 44555
    },
    {
      "epoch": 3.8334337090252086,
      "grad_norm": 0.96875,
      "learning_rate": 0.00012851203675174732,
      "loss": 1.5856,
      "step": 44556
    },
    {
      "epoch": 3.8335197453325303,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001284939384402113,
      "loss": 0.8246,
      "step": 44557
    },
    {
      "epoch": 3.833605781639852,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00012847584121525836,
      "loss": 0.9646,
      "step": 44558
    },
    {
      "epoch": 3.833691817947174,
      "grad_norm": 1.125,
      "learning_rate": 0.00012845774507694126,
      "loss": 1.0663,
      "step": 44559
    },
    {
      "epoch": 3.833777854254495,
      "grad_norm": 2.703125,
      "learning_rate": 0.00012843965002531295,
      "loss": 0.7456,
      "step": 44560
    },
    {
      "epoch": 3.833863890561817,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001284215560604265,
      "loss": 0.7555,
      "step": 44561
    },
    {
      "epoch": 3.833949926869139,
      "grad_norm": 41.0,
      "learning_rate": 0.00012840346318233463,
      "loss": 1.0244,
      "step": 44562
    },
    {
      "epoch": 3.8340359631764604,
      "grad_norm": 121.0,
      "learning_rate": 0.00012838537139109036,
      "loss": 0.671,
      "step": 44563
    },
    {
      "epoch": 3.834121999483782,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00012836728068674674,
      "loss": 0.8173,
      "step": 44564
    },
    {
      "epoch": 3.8342080357911037,
      "grad_norm": 1.234375,
      "learning_rate": 0.00012834919106935645,
      "loss": 1.3072,
      "step": 44565
    },
    {
      "epoch": 3.8342940720984258,
      "grad_norm": 24.5,
      "learning_rate": 0.00012833110253897246,
      "loss": 1.1749,
      "step": 44566
    },
    {
      "epoch": 3.8343801084057474,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001283130150956478,
      "loss": 0.8601,
      "step": 44567
    },
    {
      "epoch": 3.834466144713069,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001282949287394352,
      "loss": 1.3017,
      "step": 44568
    },
    {
      "epoch": 3.8345521810203906,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00012827684347038764,
      "loss": 1.5131,
      "step": 44569
    },
    {
      "epoch": 3.8346382173277123,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00012825875928855813,
      "loss": 0.3786,
      "step": 44570
    },
    {
      "epoch": 3.834724253635034,
      "grad_norm": 1.375,
      "learning_rate": 0.00012824067619399932,
      "loss": 1.6439,
      "step": 44571
    },
    {
      "epoch": 3.8348102899423555,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012822259418676424,
      "loss": 0.6333,
      "step": 44572
    },
    {
      "epoch": 3.8348963262496776,
      "grad_norm": 1.921875,
      "learning_rate": 0.00012820451326690574,
      "loss": 1.2106,
      "step": 44573
    },
    {
      "epoch": 3.834982362556999,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001281864334344769,
      "loss": 1.2224,
      "step": 44574
    },
    {
      "epoch": 3.835068398864321,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012816835468953025,
      "loss": 1.0377,
      "step": 44575
    },
    {
      "epoch": 3.8351544351716425,
      "grad_norm": 10.4375,
      "learning_rate": 0.0001281502770321189,
      "loss": 1.0723,
      "step": 44576
    },
    {
      "epoch": 3.835240471478964,
      "grad_norm": 1.328125,
      "learning_rate": 0.00012813220046229578,
      "loss": 1.4291,
      "step": 44577
    },
    {
      "epoch": 3.8353265077862857,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0001281141249801135,
      "loss": 1.3582,
      "step": 44578
    },
    {
      "epoch": 3.8354125440936073,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012809605058562506,
      "loss": 1.0175,
      "step": 44579
    },
    {
      "epoch": 3.8354985804009294,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012807797727888354,
      "loss": 1.0796,
      "step": 44580
    },
    {
      "epoch": 3.835584616708251,
      "grad_norm": 21.625,
      "learning_rate": 0.0001280599050599414,
      "loss": 1.2517,
      "step": 44581
    },
    {
      "epoch": 3.8356706530155726,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012804183392885172,
      "loss": 0.9797,
      "step": 44582
    },
    {
      "epoch": 3.8357566893228943,
      "grad_norm": 1.90625,
      "learning_rate": 0.0001280237638856675,
      "loss": 0.78,
      "step": 44583
    },
    {
      "epoch": 3.835842725630216,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00012800569493044128,
      "loss": 1.2155,
      "step": 44584
    },
    {
      "epoch": 3.8359287619375375,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00012798762706322603,
      "loss": 1.1097,
      "step": 44585
    },
    {
      "epoch": 3.836014798244859,
      "grad_norm": 5.03125,
      "learning_rate": 0.00012796956028407474,
      "loss": 0.6111,
      "step": 44586
    },
    {
      "epoch": 3.836100834552181,
      "grad_norm": 1.25,
      "learning_rate": 0.00012795149459304006,
      "loss": 1.4811,
      "step": 44587
    },
    {
      "epoch": 3.836186870859503,
      "grad_norm": 288.0,
      "learning_rate": 0.000127933429990175,
      "loss": 0.4406,
      "step": 44588
    },
    {
      "epoch": 3.8362729071668245,
      "grad_norm": 0.953125,
      "learning_rate": 0.00012791536647553214,
      "loss": 1.0432,
      "step": 44589
    },
    {
      "epoch": 3.836358943474146,
      "grad_norm": 49.75,
      "learning_rate": 0.00012789730404916445,
      "loss": 1.2019,
      "step": 44590
    },
    {
      "epoch": 3.8364449797814677,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00012787924271112495,
      "loss": 0.9976,
      "step": 44591
    },
    {
      "epoch": 3.8365310160887893,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00012786118246146616,
      "loss": 1.06,
      "step": 44592
    },
    {
      "epoch": 3.836617052396111,
      "grad_norm": 0.875,
      "learning_rate": 0.000127843123300241,
      "loss": 0.8162,
      "step": 44593
    },
    {
      "epoch": 3.836703088703433,
      "grad_norm": 34.0,
      "learning_rate": 0.00012782506522750247,
      "loss": 1.3853,
      "step": 44594
    },
    {
      "epoch": 3.8367891250107546,
      "grad_norm": 47.5,
      "learning_rate": 0.00012780700824330317,
      "loss": 1.421,
      "step": 44595
    },
    {
      "epoch": 3.8368751613180763,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001277889523476959,
      "loss": 0.7864,
      "step": 44596
    },
    {
      "epoch": 3.836961197625398,
      "grad_norm": 0.984375,
      "learning_rate": 0.00012777089754073374,
      "loss": 0.9322,
      "step": 44597
    },
    {
      "epoch": 3.8370472339327195,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001277528438224691,
      "loss": 1.137,
      "step": 44598
    },
    {
      "epoch": 3.837133270240041,
      "grad_norm": 31.375,
      "learning_rate": 0.00012773479119295523,
      "loss": 1.8389,
      "step": 44599
    },
    {
      "epoch": 3.8372193065473628,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001277167396522445,
      "loss": 1.1075,
      "step": 44600
    },
    {
      "epoch": 3.837305342854685,
      "grad_norm": 2.046875,
      "learning_rate": 0.0001276986892003899,
      "loss": 1.2339,
      "step": 44601
    },
    {
      "epoch": 3.8373913791620065,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001276806398374444,
      "loss": 0.7772,
      "step": 44602
    },
    {
      "epoch": 3.837477415469328,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00012766259156346045,
      "loss": 1.43,
      "step": 44603
    },
    {
      "epoch": 3.8375634517766497,
      "grad_norm": 1.7265625,
      "learning_rate": 0.000127644544378491,
      "loss": 0.9948,
      "step": 44604
    },
    {
      "epoch": 3.8376494880839713,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00012762649828258892,
      "loss": 1.4461,
      "step": 44605
    },
    {
      "epoch": 3.8377355243912934,
      "grad_norm": 27.25,
      "learning_rate": 0.0001276084532758069,
      "loss": 0.7477,
      "step": 44606
    },
    {
      "epoch": 3.8378215606986146,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001275904093581976,
      "loss": 0.4661,
      "step": 44607
    },
    {
      "epoch": 3.8379075970059366,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012757236652981413,
      "loss": 1.6116,
      "step": 44608
    },
    {
      "epoch": 3.8379936333132583,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012755432479070878,
      "loss": 0.7739,
      "step": 44609
    },
    {
      "epoch": 3.83807966962058,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00012753628414093488,
      "loss": 1.3895,
      "step": 44610
    },
    {
      "epoch": 3.8381657059279015,
      "grad_norm": 1.40625,
      "learning_rate": 0.00012751824458054472,
      "loss": 1.1972,
      "step": 44611
    },
    {
      "epoch": 3.838251742235223,
      "grad_norm": 109.0,
      "learning_rate": 0.0001275002061095913,
      "loss": 1.0851,
      "step": 44612
    },
    {
      "epoch": 3.838337778542545,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00012748216872812747,
      "loss": 1.5303,
      "step": 44613
    },
    {
      "epoch": 3.8384238148498664,
      "grad_norm": 1.84375,
      "learning_rate": 0.0001274641324362057,
      "loss": 0.8672,
      "step": 44614
    },
    {
      "epoch": 3.8385098511571885,
      "grad_norm": 1.484375,
      "learning_rate": 0.00012744609723387902,
      "loss": 1.6675,
      "step": 44615
    },
    {
      "epoch": 3.83859588746451,
      "grad_norm": 1.03125,
      "learning_rate": 0.00012742806312119988,
      "loss": 1.0553,
      "step": 44616
    },
    {
      "epoch": 3.8386819237718317,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00012741003009822123,
      "loss": 1.2025,
      "step": 44617
    },
    {
      "epoch": 3.8387679600791533,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00012739199816499592,
      "loss": 1.0694,
      "step": 44618
    },
    {
      "epoch": 3.838853996386475,
      "grad_norm": 1.015625,
      "learning_rate": 0.00012737396732157634,
      "loss": 1.0402,
      "step": 44619
    },
    {
      "epoch": 3.838940032693797,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001273559375680155,
      "loss": 1.4571,
      "step": 44620
    },
    {
      "epoch": 3.839026069001118,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001273379089043662,
      "loss": 1.2847,
      "step": 44621
    },
    {
      "epoch": 3.8391121053084403,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012731988133068089,
      "loss": 1.1392,
      "step": 44622
    },
    {
      "epoch": 3.839198141615762,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012730185484701245,
      "loss": 1.1486,
      "step": 44623
    },
    {
      "epoch": 3.8392841779230835,
      "grad_norm": 260.0,
      "learning_rate": 0.0001272838294534137,
      "loss": 0.8139,
      "step": 44624
    },
    {
      "epoch": 3.839370214230405,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001272658051499372,
      "loss": 0.5792,
      "step": 44625
    },
    {
      "epoch": 3.8394562505377268,
      "grad_norm": 0.984375,
      "learning_rate": 0.00012724778193663568,
      "loss": 1.2837,
      "step": 44626
    },
    {
      "epoch": 3.839542286845049,
      "grad_norm": 4.375,
      "learning_rate": 0.00012722975981356205,
      "loss": 0.8484,
      "step": 44627
    },
    {
      "epoch": 3.8396283231523705,
      "grad_norm": 11.1875,
      "learning_rate": 0.0001272117387807687,
      "loss": 1.1181,
      "step": 44628
    },
    {
      "epoch": 3.839714359459692,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012719371883830865,
      "loss": 0.8266,
      "step": 44629
    },
    {
      "epoch": 3.8398003957670137,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001271756999862344,
      "loss": 1.5883,
      "step": 44630
    },
    {
      "epoch": 3.8398864320743353,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00012715768222459867,
      "loss": 0.5715,
      "step": 44631
    },
    {
      "epoch": 3.839972468381657,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00012713966555345435,
      "loss": 0.6604,
      "step": 44632
    },
    {
      "epoch": 3.8400585046889786,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001271216499728538,
      "loss": 1.348,
      "step": 44633
    },
    {
      "epoch": 3.8401445409963006,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00012710363548284998,
      "loss": 0.9554,
      "step": 44634
    },
    {
      "epoch": 3.8402305773036223,
      "grad_norm": 1.09375,
      "learning_rate": 0.00012708562208349562,
      "loss": 0.6741,
      "step": 44635
    },
    {
      "epoch": 3.840316613610944,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001270676097748431,
      "loss": 1.0864,
      "step": 44636
    },
    {
      "epoch": 3.8404026499182655,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00012704959855694536,
      "loss": 0.8501,
      "step": 44637
    },
    {
      "epoch": 3.840488686225587,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012703158842985513,
      "loss": 1.4338,
      "step": 44638
    },
    {
      "epoch": 3.8405747225329088,
      "grad_norm": 1.15625,
      "learning_rate": 0.00012701357939362478,
      "loss": 1.0314,
      "step": 44639
    },
    {
      "epoch": 3.8406607588402304,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00012699557144830737,
      "loss": 1.4248,
      "step": 44640
    },
    {
      "epoch": 3.8407467951475525,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001269775645939552,
      "loss": 1.3544,
      "step": 44641
    },
    {
      "epoch": 3.840832831454874,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001269595588306211,
      "loss": 1.2125,
      "step": 44642
    },
    {
      "epoch": 3.8409188677621957,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0001269415541583579,
      "loss": 1.3999,
      "step": 44643
    },
    {
      "epoch": 3.8410049040695173,
      "grad_norm": 1.328125,
      "learning_rate": 0.00012692355057721794,
      "loss": 0.7512,
      "step": 44644
    },
    {
      "epoch": 3.841090940376839,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00012690554808725401,
      "loss": 1.1128,
      "step": 44645
    },
    {
      "epoch": 3.8411769766841606,
      "grad_norm": 1.3125,
      "learning_rate": 0.00012688754668851882,
      "loss": 0.9041,
      "step": 44646
    },
    {
      "epoch": 3.841263012991482,
      "grad_norm": 1.21875,
      "learning_rate": 0.00012686954638106495,
      "loss": 1.1441,
      "step": 44647
    },
    {
      "epoch": 3.8413490492988043,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00012685154716494528,
      "loss": 0.9011,
      "step": 44648
    },
    {
      "epoch": 3.841435085606126,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001268335490402121,
      "loss": 1.5209,
      "step": 44649
    },
    {
      "epoch": 3.8415211219134475,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00012681555200691824,
      "loss": 1.1646,
      "step": 44650
    },
    {
      "epoch": 3.841607158220769,
      "grad_norm": 5.1875,
      "learning_rate": 0.0001267975560651164,
      "loss": 1.5099,
      "step": 44651
    },
    {
      "epoch": 3.8416931945280908,
      "grad_norm": 0.9375,
      "learning_rate": 0.000126779561214859,
      "loss": 1.3275,
      "step": 44652
    },
    {
      "epoch": 3.8417792308354124,
      "grad_norm": 3.453125,
      "learning_rate": 0.0001267615674561988,
      "loss": 1.3944,
      "step": 44653
    },
    {
      "epoch": 3.841865267142734,
      "grad_norm": 3.09375,
      "learning_rate": 0.00012674357478918857,
      "loss": 1.6237,
      "step": 44654
    },
    {
      "epoch": 3.841951303450056,
      "grad_norm": 588.0,
      "learning_rate": 0.00012672558321388068,
      "loss": 0.6235,
      "step": 44655
    },
    {
      "epoch": 3.8420373397573777,
      "grad_norm": 1.0625,
      "learning_rate": 0.00012670759273032784,
      "loss": 1.7235,
      "step": 44656
    },
    {
      "epoch": 3.8421233760646993,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00012668960333858282,
      "loss": 0.8208,
      "step": 44657
    },
    {
      "epoch": 3.842209412372021,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012667161503869794,
      "loss": 0.4036,
      "step": 44658
    },
    {
      "epoch": 3.8422954486793426,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00012665362783072614,
      "loss": 1.1806,
      "step": 44659
    },
    {
      "epoch": 3.842381484986664,
      "grad_norm": 1.359375,
      "learning_rate": 0.00012663564171471974,
      "loss": 0.9441,
      "step": 44660
    },
    {
      "epoch": 3.842467521293986,
      "grad_norm": 1368.0,
      "learning_rate": 0.0001266176566907315,
      "loss": 1.1533,
      "step": 44661
    },
    {
      "epoch": 3.842553557601308,
      "grad_norm": 1.28125,
      "learning_rate": 0.00012659967275881407,
      "loss": 1.872,
      "step": 44662
    },
    {
      "epoch": 3.8426395939086295,
      "grad_norm": 1.0,
      "learning_rate": 0.00012658168991901985,
      "loss": 1.7003,
      "step": 44663
    },
    {
      "epoch": 3.842725630215951,
      "grad_norm": 4.3125,
      "learning_rate": 0.00012656370817140162,
      "loss": 2.0623,
      "step": 44664
    },
    {
      "epoch": 3.8428116665232728,
      "grad_norm": 1.640625,
      "learning_rate": 0.00012654572751601202,
      "loss": 1.7691,
      "step": 44665
    },
    {
      "epoch": 3.8428977028305944,
      "grad_norm": 5.03125,
      "learning_rate": 0.00012652774795290334,
      "loss": 1.2282,
      "step": 44666
    },
    {
      "epoch": 3.8429837391379165,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001265097694821284,
      "loss": 1.1209,
      "step": 44667
    },
    {
      "epoch": 3.8430697754452376,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001264917921037399,
      "loss": 0.9126,
      "step": 44668
    },
    {
      "epoch": 3.8431558117525597,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001264738158177901,
      "loss": 0.9751,
      "step": 44669
    },
    {
      "epoch": 3.8432418480598813,
      "grad_norm": 20.875,
      "learning_rate": 0.0001264558406243319,
      "loss": 1.2523,
      "step": 44670
    },
    {
      "epoch": 3.843327884367203,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001264378665234176,
      "loss": 0.6946,
      "step": 44671
    },
    {
      "epoch": 3.8434139206745246,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00012641989351509986,
      "loss": 0.5881,
      "step": 44672
    },
    {
      "epoch": 3.843499956981846,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00012640192159943137,
      "loss": 1.4119,
      "step": 44673
    },
    {
      "epoch": 3.8435859932891683,
      "grad_norm": 45.25,
      "learning_rate": 0.00012638395077646452,
      "loss": 0.7964,
      "step": 44674
    },
    {
      "epoch": 3.8436720295964895,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001263659810462519,
      "loss": 1.1075,
      "step": 44675
    },
    {
      "epoch": 3.8437580659038115,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001263480124088463,
      "loss": 0.7684,
      "step": 44676
    },
    {
      "epoch": 3.843844102211133,
      "grad_norm": 0.921875,
      "learning_rate": 0.00012633004486429993,
      "loss": 0.9202,
      "step": 44677
    },
    {
      "epoch": 3.8439301385184548,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001263120784126655,
      "loss": 0.9429,
      "step": 44678
    },
    {
      "epoch": 3.8440161748257764,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00012629411305399574,
      "loss": 1.6775,
      "step": 44679
    },
    {
      "epoch": 3.844102211133098,
      "grad_norm": 1.015625,
      "learning_rate": 0.00012627614878834287,
      "loss": 1.1168,
      "step": 44680
    },
    {
      "epoch": 3.84418824744042,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00012625818561575968,
      "loss": 1.172,
      "step": 44681
    },
    {
      "epoch": 3.8442742837477413,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001262402235362985,
      "loss": 1.0886,
      "step": 44682
    },
    {
      "epoch": 3.8443603200550633,
      "grad_norm": 1.265625,
      "learning_rate": 0.00012622226255001184,
      "loss": 1.4477,
      "step": 44683
    },
    {
      "epoch": 3.844446356362385,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012620430265695265,
      "loss": 1.119,
      "step": 44684
    },
    {
      "epoch": 3.8445323926697066,
      "grad_norm": 2.21875,
      "learning_rate": 0.000126186343857173,
      "loss": 0.8246,
      "step": 44685
    },
    {
      "epoch": 3.844618428977028,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00012616838615072572,
      "loss": 1.2333,
      "step": 44686
    },
    {
      "epoch": 3.84470446528435,
      "grad_norm": 1.03125,
      "learning_rate": 0.00012615042953766315,
      "loss": 0.745,
      "step": 44687
    },
    {
      "epoch": 3.844790501591672,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00012613247401803778,
      "loss": 0.6191,
      "step": 44688
    },
    {
      "epoch": 3.8448765378989935,
      "grad_norm": 29.125,
      "learning_rate": 0.0001261145195919024,
      "loss": 0.7854,
      "step": 44689
    },
    {
      "epoch": 3.844962574206315,
      "grad_norm": 1.28125,
      "learning_rate": 0.00012609656625930916,
      "loss": 1.2271,
      "step": 44690
    },
    {
      "epoch": 3.8450486105136368,
      "grad_norm": 608.0,
      "learning_rate": 0.00012607861402031073,
      "loss": 1.083,
      "step": 44691
    },
    {
      "epoch": 3.8451346468209584,
      "grad_norm": 2.359375,
      "learning_rate": 0.0001260606628749598,
      "loss": 1.4238,
      "step": 44692
    },
    {
      "epoch": 3.84522068312828,
      "grad_norm": 1.109375,
      "learning_rate": 0.00012604271282330853,
      "loss": 1.5526,
      "step": 44693
    },
    {
      "epoch": 3.8453067194356017,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012602476386540967,
      "loss": 1.1016,
      "step": 44694
    },
    {
      "epoch": 3.8453927557429237,
      "grad_norm": 1.15625,
      "learning_rate": 0.00012600681600131574,
      "loss": 1.2063,
      "step": 44695
    },
    {
      "epoch": 3.8454787920502453,
      "grad_norm": 0.859375,
      "learning_rate": 0.000125988869231079,
      "loss": 0.9599,
      "step": 44696
    },
    {
      "epoch": 3.845564828357567,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012597092355475203,
      "loss": 0.9309,
      "step": 44697
    },
    {
      "epoch": 3.8456508646648886,
      "grad_norm": 1.203125,
      "learning_rate": 0.00012595297897238756,
      "loss": 0.7775,
      "step": 44698
    },
    {
      "epoch": 3.84573690097221,
      "grad_norm": 1.0,
      "learning_rate": 0.00012593503548403773,
      "loss": 1.6135,
      "step": 44699
    },
    {
      "epoch": 3.845822937279532,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001259170930897553,
      "loss": 1.7325,
      "step": 44700
    },
    {
      "epoch": 3.8459089735868535,
      "grad_norm": 1.65625,
      "learning_rate": 0.00012589915178959243,
      "loss": 2.2156,
      "step": 44701
    },
    {
      "epoch": 3.8459950098941755,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00012588121158360182,
      "loss": 1.3164,
      "step": 44702
    },
    {
      "epoch": 3.846081046201497,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001258632724718361,
      "loss": 1.722,
      "step": 44703
    },
    {
      "epoch": 3.846167082508819,
      "grad_norm": 548.0,
      "learning_rate": 0.00012584533445434733,
      "loss": 0.7827,
      "step": 44704
    },
    {
      "epoch": 3.8462531188161404,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001258273975311882,
      "loss": 0.9082,
      "step": 44705
    },
    {
      "epoch": 3.846339155123462,
      "grad_norm": 712.0,
      "learning_rate": 0.0001258094617024113,
      "loss": 0.8237,
      "step": 44706
    },
    {
      "epoch": 3.8464251914307837,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00012579152696806883,
      "loss": 0.7673,
      "step": 44707
    },
    {
      "epoch": 3.8465112277381053,
      "grad_norm": 11.375,
      "learning_rate": 0.00012577359332821336,
      "loss": 1.1311,
      "step": 44708
    },
    {
      "epoch": 3.8465972640454273,
      "grad_norm": 1.046875,
      "learning_rate": 0.00012575566078289752,
      "loss": 0.703,
      "step": 44709
    },
    {
      "epoch": 3.846683300352749,
      "grad_norm": 1.140625,
      "learning_rate": 0.00012573772933217336,
      "loss": 1.6272,
      "step": 44710
    },
    {
      "epoch": 3.8467693366600706,
      "grad_norm": 1848.0,
      "learning_rate": 0.00012571979897609375,
      "loss": 1.5021,
      "step": 44711
    },
    {
      "epoch": 3.846855372967392,
      "grad_norm": 0.90625,
      "learning_rate": 0.00012570186971471082,
      "loss": 0.7937,
      "step": 44712
    },
    {
      "epoch": 3.846941409274714,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012568394154807705,
      "loss": 0.8419,
      "step": 44713
    },
    {
      "epoch": 3.8470274455820355,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00012566601447624515,
      "loss": 1.0034,
      "step": 44714
    },
    {
      "epoch": 3.847113481889357,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00012564808849926717,
      "loss": 0.8025,
      "step": 44715
    },
    {
      "epoch": 3.847199518196679,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00012563016361719575,
      "loss": 1.1072,
      "step": 44716
    },
    {
      "epoch": 3.847285554504001,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00012561223983008346,
      "loss": 0.9755,
      "step": 44717
    },
    {
      "epoch": 3.8473715908113224,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012559431713798236,
      "loss": 0.958,
      "step": 44718
    },
    {
      "epoch": 3.847457627118644,
      "grad_norm": 1.359375,
      "learning_rate": 0.00012557639554094507,
      "loss": 1.4825,
      "step": 44719
    },
    {
      "epoch": 3.8475436634259657,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00012555847503902402,
      "loss": 0.7788,
      "step": 44720
    },
    {
      "epoch": 3.8476296997332877,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00012554055563227163,
      "loss": 0.9198,
      "step": 44721
    },
    {
      "epoch": 3.847715736040609,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012552263732074038,
      "loss": 0.5032,
      "step": 44722
    },
    {
      "epoch": 3.847801772347931,
      "grad_norm": 0.9375,
      "learning_rate": 0.00012550472010448244,
      "loss": 0.9061,
      "step": 44723
    },
    {
      "epoch": 3.8478878086552526,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001254868039835504,
      "loss": 0.8333,
      "step": 44724
    },
    {
      "epoch": 3.847973844962574,
      "grad_norm": 1.046875,
      "learning_rate": 0.00012546888895799673,
      "loss": 0.7764,
      "step": 44725
    },
    {
      "epoch": 3.848059881269896,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001254509750278736,
      "loss": 0.5694,
      "step": 44726
    },
    {
      "epoch": 3.8481459175772175,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00012543306219323363,
      "loss": 0.7371,
      "step": 44727
    },
    {
      "epoch": 3.8482319538845395,
      "grad_norm": 1.1875,
      "learning_rate": 0.00012541515045412898,
      "loss": 1.6186,
      "step": 44728
    },
    {
      "epoch": 3.8483179901918607,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001253972398106122,
      "loss": 1.7572,
      "step": 44729
    },
    {
      "epoch": 3.848404026499183,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00012537933026273573,
      "loss": 2.0088,
      "step": 44730
    },
    {
      "epoch": 3.8484900628065044,
      "grad_norm": 154.0,
      "learning_rate": 0.0001253614218105517,
      "loss": 1.4992,
      "step": 44731
    },
    {
      "epoch": 3.848576099113826,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012534351445411273,
      "loss": 1.0514,
      "step": 44732
    },
    {
      "epoch": 3.8486621354211477,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012532560819347122,
      "loss": 1.0846,
      "step": 44733
    },
    {
      "epoch": 3.8487481717284693,
      "grad_norm": 1.125,
      "learning_rate": 0.00012530770302867928,
      "loss": 0.9506,
      "step": 44734
    },
    {
      "epoch": 3.8488342080357913,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001252897989597894,
      "loss": 1.3417,
      "step": 44735
    },
    {
      "epoch": 3.8489202443431125,
      "grad_norm": 14.0625,
      "learning_rate": 0.00012527189598685417,
      "loss": 0.7365,
      "step": 44736
    },
    {
      "epoch": 3.8490062806504346,
      "grad_norm": 14.875,
      "learning_rate": 0.00012525399410992562,
      "loss": 0.5115,
      "step": 44737
    },
    {
      "epoch": 3.849092316957756,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001252360933290562,
      "loss": 0.8677,
      "step": 44738
    },
    {
      "epoch": 3.849178353265078,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00012521819364429853,
      "loss": 1.1192,
      "step": 44739
    },
    {
      "epoch": 3.8492643895723995,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00012520029505570457,
      "loss": 1.4447,
      "step": 44740
    },
    {
      "epoch": 3.849350425879721,
      "grad_norm": 1.1875,
      "learning_rate": 0.000125182397563327,
      "loss": 1.4683,
      "step": 44741
    },
    {
      "epoch": 3.849436462187043,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001251645011672179,
      "loss": 1.377,
      "step": 44742
    },
    {
      "epoch": 3.849522498494365,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001251466058674297,
      "loss": 1.7585,
      "step": 44743
    },
    {
      "epoch": 3.8496085348016864,
      "grad_norm": 1.625,
      "learning_rate": 0.00012512871166401495,
      "loss": 1.3226,
      "step": 44744
    },
    {
      "epoch": 3.849694571109008,
      "grad_norm": 1.15625,
      "learning_rate": 0.00012511081855702566,
      "loss": 0.8941,
      "step": 44745
    },
    {
      "epoch": 3.8497806074163297,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012509292654651433,
      "loss": 0.665,
      "step": 44746
    },
    {
      "epoch": 3.8498666437236513,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00012507503563253337,
      "loss": 1.1524,
      "step": 44747
    },
    {
      "epoch": 3.849952680030973,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012505714581513487,
      "loss": 1.237,
      "step": 44748
    },
    {
      "epoch": 3.850038716338295,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012503925709437137,
      "loss": 1.4283,
      "step": 44749
    },
    {
      "epoch": 3.8501247526456166,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001250213694702952,
      "loss": 1.5684,
      "step": 44750
    },
    {
      "epoch": 3.8502107889529382,
      "grad_norm": 1.09375,
      "learning_rate": 0.00012500348294295843,
      "loss": 1.5297,
      "step": 44751
    },
    {
      "epoch": 3.85029682526026,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00012498559751241372,
      "loss": 1.4971,
      "step": 44752
    },
    {
      "epoch": 3.8503828615675815,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012496771317871302,
      "loss": 0.7378,
      "step": 44753
    },
    {
      "epoch": 3.850468897874903,
      "grad_norm": 1.546875,
      "learning_rate": 0.0001249498299419089,
      "loss": 0.5949,
      "step": 44754
    },
    {
      "epoch": 3.8505549341822247,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012493194780205364,
      "loss": 1.0697,
      "step": 44755
    },
    {
      "epoch": 3.850640970489547,
      "grad_norm": 1.0,
      "learning_rate": 0.0001249140667591994,
      "loss": 0.5633,
      "step": 44756
    },
    {
      "epoch": 3.8507270067968684,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012489618681339854,
      "loss": 1.1958,
      "step": 44757
    },
    {
      "epoch": 3.85081304310419,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001248783079647034,
      "loss": 0.8143,
      "step": 44758
    },
    {
      "epoch": 3.8508990794115117,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00012486043021316622,
      "loss": 1.2652,
      "step": 44759
    },
    {
      "epoch": 3.8509851157188333,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001248425535588395,
      "loss": 1.2669,
      "step": 44760
    },
    {
      "epoch": 3.851071152026155,
      "grad_norm": 1.375,
      "learning_rate": 0.00012482467800177516,
      "loss": 0.7236,
      "step": 44761
    },
    {
      "epoch": 3.8511571883334765,
      "grad_norm": 1.125,
      "learning_rate": 0.00012480680354202568,
      "loss": 1.4792,
      "step": 44762
    },
    {
      "epoch": 3.8512432246407986,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001247889301796435,
      "loss": 1.4494,
      "step": 44763
    },
    {
      "epoch": 3.8513292609481202,
      "grad_norm": 2.875,
      "learning_rate": 0.00012477105791468057,
      "loss": 1.0773,
      "step": 44764
    },
    {
      "epoch": 3.851415297255442,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00012475318674718932,
      "loss": 1.1357,
      "step": 44765
    },
    {
      "epoch": 3.8515013335627635,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012473531667722216,
      "loss": 0.7179,
      "step": 44766
    },
    {
      "epoch": 3.851587369870085,
      "grad_norm": 1.171875,
      "learning_rate": 0.00012471744770483107,
      "loss": 0.8978,
      "step": 44767
    },
    {
      "epoch": 3.8516734061774067,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001246995798300684,
      "loss": 1.0641,
      "step": 44768
    },
    {
      "epoch": 3.8517594424847283,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0001246817130529867,
      "loss": 1.5261,
      "step": 44769
    },
    {
      "epoch": 3.8518454787920504,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001246638473736378,
      "loss": 1.1809,
      "step": 44770
    },
    {
      "epoch": 3.851931515099372,
      "grad_norm": 1.09375,
      "learning_rate": 0.00012464598279207434,
      "loss": 1.1731,
      "step": 44771
    },
    {
      "epoch": 3.8520175514066937,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00012462811930834817,
      "loss": 0.941,
      "step": 44772
    },
    {
      "epoch": 3.8521035877140153,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00012461025692251176,
      "loss": 0.6871,
      "step": 44773
    },
    {
      "epoch": 3.852189624021337,
      "grad_norm": 47.75,
      "learning_rate": 0.0001245923956346175,
      "loss": 1.3129,
      "step": 44774
    },
    {
      "epoch": 3.8522756603286585,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012457453544471736,
      "loss": 1.1121,
      "step": 44775
    },
    {
      "epoch": 3.85236169663598,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001245566763528636,
      "loss": 1.043,
      "step": 44776
    },
    {
      "epoch": 3.8524477329433022,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00012453881835910874,
      "loss": 0.5229,
      "step": 44777
    },
    {
      "epoch": 3.852533769250624,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00012452096146350466,
      "loss": 0.7567,
      "step": 44778
    },
    {
      "epoch": 3.8526198055579455,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012450310566610378,
      "loss": 0.9638,
      "step": 44779
    },
    {
      "epoch": 3.852705841865267,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001244852509669584,
      "loss": 1.7634,
      "step": 44780
    },
    {
      "epoch": 3.8527918781725887,
      "grad_norm": 1.109375,
      "learning_rate": 0.00012446739736612046,
      "loss": 1.0032,
      "step": 44781
    },
    {
      "epoch": 3.852877914479911,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001244495448636425,
      "loss": 1.2236,
      "step": 44782
    },
    {
      "epoch": 3.852963950787232,
      "grad_norm": 3.609375,
      "learning_rate": 0.00012443169345957645,
      "loss": 1.1767,
      "step": 44783
    },
    {
      "epoch": 3.853049987094554,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0001244138431539747,
      "loss": 1.4553,
      "step": 44784
    },
    {
      "epoch": 3.8531360234018757,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001243959939468895,
      "loss": 1.1216,
      "step": 44785
    },
    {
      "epoch": 3.8532220597091973,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001243781458383728,
      "loss": 1.2364,
      "step": 44786
    },
    {
      "epoch": 3.853308096016519,
      "grad_norm": 1.09375,
      "learning_rate": 0.000124360298828477,
      "loss": 1.4241,
      "step": 44787
    },
    {
      "epoch": 3.8533941323238405,
      "grad_norm": 1.328125,
      "learning_rate": 0.00012434245291725442,
      "loss": 1.0078,
      "step": 44788
    },
    {
      "epoch": 3.8534801686311626,
      "grad_norm": 8.625,
      "learning_rate": 0.00012432460810475694,
      "loss": 1.3335,
      "step": 44789
    },
    {
      "epoch": 3.853566204938484,
      "grad_norm": 3.796875,
      "learning_rate": 0.00012430676439103693,
      "loss": 0.9228,
      "step": 44790
    },
    {
      "epoch": 3.853652241245806,
      "grad_norm": 62.75,
      "learning_rate": 0.0001242889217761467,
      "loss": 1.2117,
      "step": 44791
    },
    {
      "epoch": 3.8537382775531275,
      "grad_norm": 1.125,
      "learning_rate": 0.00012427108026013818,
      "loss": 1.4431,
      "step": 44792
    },
    {
      "epoch": 3.853824313860449,
      "grad_norm": 1.390625,
      "learning_rate": 0.00012425323984306382,
      "loss": 1.1021,
      "step": 44793
    },
    {
      "epoch": 3.8539103501677707,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001242354005249753,
      "loss": 1.4294,
      "step": 44794
    },
    {
      "epoch": 3.8539963864750924,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00012421756230592535,
      "loss": 0.9971,
      "step": 44795
    },
    {
      "epoch": 3.8540824227824144,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00012419972518596607,
      "loss": 0.7153,
      "step": 44796
    },
    {
      "epoch": 3.8541684590897356,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001241818891651494,
      "loss": 1.0123,
      "step": 44797
    },
    {
      "epoch": 3.8542544953970577,
      "grad_norm": 29.75,
      "learning_rate": 0.00012416405424352773,
      "loss": 1.4871,
      "step": 44798
    },
    {
      "epoch": 3.8543405317043793,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012414622042115293,
      "loss": 0.9949,
      "step": 44799
    },
    {
      "epoch": 3.854426568011701,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00012412838769807738,
      "loss": 0.6999,
      "step": 44800
    },
    {
      "epoch": 3.8545126043190225,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00012411055607435334,
      "loss": 1.3959,
      "step": 44801
    },
    {
      "epoch": 3.854598640626344,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012409272555003264,
      "loss": 0.699,
      "step": 44802
    },
    {
      "epoch": 3.8546846769336662,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00012407489612516765,
      "loss": 0.6303,
      "step": 44803
    },
    {
      "epoch": 3.854770713240988,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001240570677998106,
      "loss": 0.9,
      "step": 44804
    },
    {
      "epoch": 3.8548567495483095,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012403924057401334,
      "loss": 1.3135,
      "step": 44805
    },
    {
      "epoch": 3.854942785855631,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012402141444782817,
      "loss": 0.7048,
      "step": 44806
    },
    {
      "epoch": 3.8550288221629527,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012400358942130741,
      "loss": 1.202,
      "step": 44807
    },
    {
      "epoch": 3.8551148584702744,
      "grad_norm": 6.28125,
      "learning_rate": 0.00012398576549450292,
      "loss": 1.1274,
      "step": 44808
    },
    {
      "epoch": 3.855200894777596,
      "grad_norm": 0.921875,
      "learning_rate": 0.00012396794266746687,
      "loss": 0.8935,
      "step": 44809
    },
    {
      "epoch": 3.855286931084918,
      "grad_norm": 278.0,
      "learning_rate": 0.00012395012094025166,
      "loss": 1.3588,
      "step": 44810
    },
    {
      "epoch": 3.8553729673922397,
      "grad_norm": 1.828125,
      "learning_rate": 0.00012393230031290903,
      "loss": 1.63,
      "step": 44811
    },
    {
      "epoch": 3.8554590036995613,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001239144807854914,
      "loss": 0.8079,
      "step": 44812
    },
    {
      "epoch": 3.855545040006883,
      "grad_norm": 1.234375,
      "learning_rate": 0.00012389666235805064,
      "loss": 1.7234,
      "step": 44813
    },
    {
      "epoch": 3.8556310763142045,
      "grad_norm": 1.234375,
      "learning_rate": 0.000123878845030639,
      "loss": 1.779,
      "step": 44814
    },
    {
      "epoch": 3.855717112621526,
      "grad_norm": 0.75,
      "learning_rate": 0.00012386102880330874,
      "loss": 0.9255,
      "step": 44815
    },
    {
      "epoch": 3.855803148928848,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001238432136761117,
      "loss": 0.8621,
      "step": 44816
    },
    {
      "epoch": 3.85588918523617,
      "grad_norm": 1.90625,
      "learning_rate": 0.00012382539964910007,
      "loss": 1.0416,
      "step": 44817
    },
    {
      "epoch": 3.8559752215434915,
      "grad_norm": 1.125,
      "learning_rate": 0.0001238075867223261,
      "loss": 0.9906,
      "step": 44818
    },
    {
      "epoch": 3.856061257850813,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00012378977489584166,
      "loss": 1.652,
      "step": 44819
    },
    {
      "epoch": 3.8561472941581347,
      "grad_norm": 1.0,
      "learning_rate": 0.00012377196416969895,
      "loss": 1.0832,
      "step": 44820
    },
    {
      "epoch": 3.8562333304654564,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012375415454395024,
      "loss": 0.9291,
      "step": 44821
    },
    {
      "epoch": 3.856319366772778,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00012373634601864724,
      "loss": 1.2599,
      "step": 44822
    },
    {
      "epoch": 3.8564054030800996,
      "grad_norm": 1.171875,
      "learning_rate": 0.00012371853859384247,
      "loss": 1.5219,
      "step": 44823
    },
    {
      "epoch": 3.8564914393874217,
      "grad_norm": 7.21875,
      "learning_rate": 0.0001237007322695876,
      "loss": 1.7304,
      "step": 44824
    },
    {
      "epoch": 3.8565774756947433,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00012368292704593493,
      "loss": 1.8406,
      "step": 44825
    },
    {
      "epoch": 3.856663512002065,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012366512292293663,
      "loss": 1.1548,
      "step": 44826
    },
    {
      "epoch": 3.8567495483093865,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012364731990064455,
      "loss": 0.7572,
      "step": 44827
    },
    {
      "epoch": 3.856835584616708,
      "grad_norm": 1.46875,
      "learning_rate": 0.00012362951797911082,
      "loss": 1.3701,
      "step": 44828
    },
    {
      "epoch": 3.85692162092403,
      "grad_norm": 1.3125,
      "learning_rate": 0.00012361171715838775,
      "loss": 1.0614,
      "step": 44829
    },
    {
      "epoch": 3.8570076572313514,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000123593917438527,
      "loss": 1.0429,
      "step": 44830
    },
    {
      "epoch": 3.8570936935386735,
      "grad_norm": 2.703125,
      "learning_rate": 0.00012357611881958085,
      "loss": 1.1103,
      "step": 44831
    },
    {
      "epoch": 3.857179729845995,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00012355832130160133,
      "loss": 0.5808,
      "step": 44832
    },
    {
      "epoch": 3.8572657661533167,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012354052488464058,
      "loss": 0.7773,
      "step": 44833
    },
    {
      "epoch": 3.8573518024606384,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012352272956875067,
      "loss": 0.9777,
      "step": 44834
    },
    {
      "epoch": 3.85743783876796,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001235049353539834,
      "loss": 0.8547,
      "step": 44835
    },
    {
      "epoch": 3.8575238750752816,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00012348714224039094,
      "loss": 0.8745,
      "step": 44836
    },
    {
      "epoch": 3.8576099113826032,
      "grad_norm": 2.03125,
      "learning_rate": 0.00012346935022802553,
      "loss": 1.1912,
      "step": 44837
    },
    {
      "epoch": 3.8576959476899253,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001234515593169389,
      "loss": 1.5454,
      "step": 44838
    },
    {
      "epoch": 3.857781983997247,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001234337695071834,
      "loss": 0.9454,
      "step": 44839
    },
    {
      "epoch": 3.8578680203045685,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00012341598079881072,
      "loss": 1.2882,
      "step": 44840
    },
    {
      "epoch": 3.85795405661189,
      "grad_norm": 5.78125,
      "learning_rate": 0.00012339819319187308,
      "loss": 1.6286,
      "step": 44841
    },
    {
      "epoch": 3.858040092919212,
      "grad_norm": 15.1875,
      "learning_rate": 0.00012338040668642252,
      "loss": 1.2973,
      "step": 44842
    },
    {
      "epoch": 3.858126129226534,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00012336262128251097,
      "loss": 0.9826,
      "step": 44843
    },
    {
      "epoch": 3.858212165533855,
      "grad_norm": 7.75,
      "learning_rate": 0.0001233448369801905,
      "loss": 0.8329,
      "step": 44844
    },
    {
      "epoch": 3.858298201841177,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00012332705377951327,
      "loss": 1.1324,
      "step": 44845
    },
    {
      "epoch": 3.8583842381484987,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00012330927168053096,
      "loss": 0.6222,
      "step": 44846
    },
    {
      "epoch": 3.8584702744558204,
      "grad_norm": 36.5,
      "learning_rate": 0.00012329149068329582,
      "loss": 1.5291,
      "step": 44847
    },
    {
      "epoch": 3.858556310763142,
      "grad_norm": 0.90625,
      "learning_rate": 0.00012327371078785994,
      "loss": 0.6607,
      "step": 44848
    },
    {
      "epoch": 3.8586423470704636,
      "grad_norm": 1.40625,
      "learning_rate": 0.00012325593199427498,
      "loss": 0.8824,
      "step": 44849
    },
    {
      "epoch": 3.8587283833777857,
      "grad_norm": 68.5,
      "learning_rate": 0.00012323815430259323,
      "loss": 1.0014,
      "step": 44850
    },
    {
      "epoch": 3.858814419685107,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001232203777128667,
      "loss": 1.0516,
      "step": 44851
    },
    {
      "epoch": 3.858900455992429,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012320260222514718,
      "loss": 0.9819,
      "step": 44852
    },
    {
      "epoch": 3.8589864922997505,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012318482783948688,
      "loss": 1.0708,
      "step": 44853
    },
    {
      "epoch": 3.859072528607072,
      "grad_norm": 4.46875,
      "learning_rate": 0.00012316705455593752,
      "loss": 1.557,
      "step": 44854
    },
    {
      "epoch": 3.859158564914394,
      "grad_norm": 157.0,
      "learning_rate": 0.00012314928237455125,
      "loss": 0.7507,
      "step": 44855
    },
    {
      "epoch": 3.8592446012217154,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012313151129538015,
      "loss": 1.0132,
      "step": 44856
    },
    {
      "epoch": 3.8593306375290375,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00012311374131847598,
      "loss": 0.6982,
      "step": 44857
    },
    {
      "epoch": 3.8594166738363587,
      "grad_norm": 252.0,
      "learning_rate": 0.0001230959724438908,
      "loss": 1.0709,
      "step": 44858
    },
    {
      "epoch": 3.8595027101436807,
      "grad_norm": 1.453125,
      "learning_rate": 0.00012307820467167673,
      "loss": 1.2749,
      "step": 44859
    },
    {
      "epoch": 3.8595887464510024,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001230604380018855,
      "loss": 0.9878,
      "step": 44860
    },
    {
      "epoch": 3.859674782758324,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00012304267243456912,
      "loss": 0.8087,
      "step": 44861
    },
    {
      "epoch": 3.8597608190656456,
      "grad_norm": 1.328125,
      "learning_rate": 0.00012302490796977977,
      "loss": 1.208,
      "step": 44862
    },
    {
      "epoch": 3.8598468553729672,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012300714460756912,
      "loss": 1.199,
      "step": 44863
    },
    {
      "epoch": 3.8599328916802893,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001229893823479894,
      "loss": 1.2126,
      "step": 44864
    },
    {
      "epoch": 3.860018927987611,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00012297162119109228,
      "loss": 0.8602,
      "step": 44865
    },
    {
      "epoch": 3.8601049642949326,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001229538611369298,
      "loss": 0.7662,
      "step": 44866
    },
    {
      "epoch": 3.860191000602254,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00012293610218555413,
      "loss": 0.3989,
      "step": 44867
    },
    {
      "epoch": 3.860277036909576,
      "grad_norm": 10.25,
      "learning_rate": 0.0001229183443370167,
      "loss": 1.2091,
      "step": 44868
    },
    {
      "epoch": 3.8603630732168974,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00012290058759137007,
      "loss": 0.7612,
      "step": 44869
    },
    {
      "epoch": 3.860449109524219,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00012288283194866574,
      "loss": 0.8483,
      "step": 44870
    },
    {
      "epoch": 3.860535145831541,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00012286507740895582,
      "loss": 0.6148,
      "step": 44871
    },
    {
      "epoch": 3.8606211821388627,
      "grad_norm": 1.03125,
      "learning_rate": 0.00012284732397229226,
      "loss": 0.8137,
      "step": 44872
    },
    {
      "epoch": 3.8607072184461844,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00012282957163872683,
      "loss": 0.9329,
      "step": 44873
    },
    {
      "epoch": 3.860793254753506,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012281182040831156,
      "loss": 1.4473,
      "step": 44874
    },
    {
      "epoch": 3.8608792910608276,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001227940702810985,
      "loss": 1.7587,
      "step": 44875
    },
    {
      "epoch": 3.8609653273681492,
      "grad_norm": 1.109375,
      "learning_rate": 0.00012277632125713928,
      "loss": 1.0898,
      "step": 44876
    },
    {
      "epoch": 3.861051363675471,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00012275857333648595,
      "loss": 1.1471,
      "step": 44877
    },
    {
      "epoch": 3.861137399982793,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001227408265191905,
      "loss": 1.7944,
      "step": 44878
    },
    {
      "epoch": 3.8612234362901146,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001227230808053047,
      "loss": 1.2175,
      "step": 44879
    },
    {
      "epoch": 3.861309472597436,
      "grad_norm": 5.40625,
      "learning_rate": 0.0001227053361948806,
      "loss": 1.4483,
      "step": 44880
    },
    {
      "epoch": 3.861395508904758,
      "grad_norm": 3.953125,
      "learning_rate": 0.00012268759268796992,
      "loss": 1.1592,
      "step": 44881
    },
    {
      "epoch": 3.8614815452120794,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00012266985028462464,
      "loss": 0.7456,
      "step": 44882
    },
    {
      "epoch": 3.861567581519401,
      "grad_norm": 274.0,
      "learning_rate": 0.00012265210898489677,
      "loss": 0.8654,
      "step": 44883
    },
    {
      "epoch": 3.8616536178267227,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000122634368788838,
      "loss": 0.9715,
      "step": 44884
    },
    {
      "epoch": 3.8617396541340447,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00012261662969650022,
      "loss": 0.8861,
      "step": 44885
    },
    {
      "epoch": 3.8618256904413664,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00012259889170793565,
      "loss": 1.197,
      "step": 44886
    },
    {
      "epoch": 3.861911726748688,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001225811548231957,
      "loss": 1.1464,
      "step": 44887
    },
    {
      "epoch": 3.8619977630560096,
      "grad_norm": 264.0,
      "learning_rate": 0.00012256341904233248,
      "loss": 0.8542,
      "step": 44888
    },
    {
      "epoch": 3.8620837993633312,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00012254568436539804,
      "loss": 1.8403,
      "step": 44889
    },
    {
      "epoch": 3.862169835670653,
      "grad_norm": 4.15625,
      "learning_rate": 0.0001225279507924439,
      "loss": 1.4062,
      "step": 44890
    },
    {
      "epoch": 3.8622558719779745,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00012251021832352204,
      "loss": 1.103,
      "step": 44891
    },
    {
      "epoch": 3.8623419082852966,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001224924869586846,
      "loss": 0.643,
      "step": 44892
    },
    {
      "epoch": 3.862427944592618,
      "grad_norm": 1.5625,
      "learning_rate": 0.00012247475669798302,
      "loss": 0.7393,
      "step": 44893
    },
    {
      "epoch": 3.86251398089994,
      "grad_norm": 1.25,
      "learning_rate": 0.00012245702754146948,
      "loss": 1.4586,
      "step": 44894
    },
    {
      "epoch": 3.8626000172072614,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00012243929948919558,
      "loss": 1.3327,
      "step": 44895
    },
    {
      "epoch": 3.862686053514583,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00012242157254121333,
      "loss": 1.1457,
      "step": 44896
    },
    {
      "epoch": 3.862772089821905,
      "grad_norm": 23.75,
      "learning_rate": 0.00012240384669757464,
      "loss": 1.0189,
      "step": 44897
    },
    {
      "epoch": 3.8628581261292263,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00012238612195833115,
      "loss": 0.7879,
      "step": 44898
    },
    {
      "epoch": 3.8629441624365484,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001223683983235348,
      "loss": 0.9521,
      "step": 44899
    },
    {
      "epoch": 3.86303019874387,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001223506757932376,
      "loss": 0.532,
      "step": 44900
    },
    {
      "epoch": 3.8631162350511916,
      "grad_norm": 608.0,
      "learning_rate": 0.00012233295436749102,
      "loss": 0.6389,
      "step": 44901
    },
    {
      "epoch": 3.8632022713585132,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0001223152340463472,
      "loss": 1.7637,
      "step": 44902
    },
    {
      "epoch": 3.863288307665835,
      "grad_norm": 704.0,
      "learning_rate": 0.00012229751482985797,
      "loss": 0.6122,
      "step": 44903
    },
    {
      "epoch": 3.863374343973157,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012227979671807483,
      "loss": 1.0935,
      "step": 44904
    },
    {
      "epoch": 3.863460380280478,
      "grad_norm": 1.046875,
      "learning_rate": 0.00012226207971104992,
      "loss": 1.0796,
      "step": 44905
    },
    {
      "epoch": 3.8635464165878,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00012224436380883492,
      "loss": 1.4454,
      "step": 44906
    },
    {
      "epoch": 3.863632452895122,
      "grad_norm": 1.0625,
      "learning_rate": 0.00012222664901148167,
      "loss": 0.8023,
      "step": 44907
    },
    {
      "epoch": 3.8637184892024434,
      "grad_norm": 1.53125,
      "learning_rate": 0.00012220893531904214,
      "loss": 1.6119,
      "step": 44908
    },
    {
      "epoch": 3.863804525509765,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001221912227315679,
      "loss": 1.6223,
      "step": 44909
    },
    {
      "epoch": 3.8638905618170867,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012217351124911097,
      "loss": 1.2225,
      "step": 44910
    },
    {
      "epoch": 3.8639765981244087,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001221558008717229,
      "loss": 1.5787,
      "step": 44911
    },
    {
      "epoch": 3.86406263443173,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001221380915994556,
      "loss": 1.1165,
      "step": 44912
    },
    {
      "epoch": 3.864148670739052,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012212038343236103,
      "loss": 0.8732,
      "step": 44913
    },
    {
      "epoch": 3.8642347070463736,
      "grad_norm": 2.234375,
      "learning_rate": 0.00012210267637049072,
      "loss": 1.2238,
      "step": 44914
    },
    {
      "epoch": 3.8643207433536952,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001220849704138966,
      "loss": 0.4506,
      "step": 44915
    },
    {
      "epoch": 3.864406779661017,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001220672655626306,
      "loss": 0.8564,
      "step": 44916
    },
    {
      "epoch": 3.8644928159683385,
      "grad_norm": 1.1875,
      "learning_rate": 0.00012204956181674415,
      "loss": 1.1458,
      "step": 44917
    },
    {
      "epoch": 3.8645788522756606,
      "grad_norm": 1.21875,
      "learning_rate": 0.00012203185917628929,
      "loss": 1.7451,
      "step": 44918
    },
    {
      "epoch": 3.864664888582982,
      "grad_norm": 3.28125,
      "learning_rate": 0.00012201415764131779,
      "loss": 1.0389,
      "step": 44919
    },
    {
      "epoch": 3.864750924890304,
      "grad_norm": 81.5,
      "learning_rate": 0.0001219964572118813,
      "loss": 1.0125,
      "step": 44920
    },
    {
      "epoch": 3.8648369611976254,
      "grad_norm": 75.5,
      "learning_rate": 0.00012197875788803159,
      "loss": 1.2178,
      "step": 44921
    },
    {
      "epoch": 3.864922997504947,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00012196105966982069,
      "loss": 1.1204,
      "step": 44922
    },
    {
      "epoch": 3.8650090338122687,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00012194336255729999,
      "loss": 0.7248,
      "step": 44923
    },
    {
      "epoch": 3.8650950701195903,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00012192566655052156,
      "loss": 0.6049,
      "step": 44924
    },
    {
      "epoch": 3.8651811064269124,
      "grad_norm": 233.0,
      "learning_rate": 0.00012190797164953694,
      "loss": 1.128,
      "step": 44925
    },
    {
      "epoch": 3.865267142734234,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00012189027785439794,
      "loss": 1.3494,
      "step": 44926
    },
    {
      "epoch": 3.8653531790415556,
      "grad_norm": 13.5,
      "learning_rate": 0.00012187258516515643,
      "loss": 0.8636,
      "step": 44927
    },
    {
      "epoch": 3.8654392153488772,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00012185489358186397,
      "loss": 1.1021,
      "step": 44928
    },
    {
      "epoch": 3.865525251656199,
      "grad_norm": 140.0,
      "learning_rate": 0.00012183720310457242,
      "loss": 1.0412,
      "step": 44929
    },
    {
      "epoch": 3.8656112879635205,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001218195137333336,
      "loss": 1.4341,
      "step": 44930
    },
    {
      "epoch": 3.865697324270842,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00012180182546819901,
      "loss": 0.8341,
      "step": 44931
    },
    {
      "epoch": 3.865783360578164,
      "grad_norm": 3.703125,
      "learning_rate": 0.00012178413830922059,
      "loss": 1.3851,
      "step": 44932
    },
    {
      "epoch": 3.865869396885486,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012176645225645005,
      "loss": 1.1256,
      "step": 44933
    },
    {
      "epoch": 3.8659554331928074,
      "grad_norm": 0.96875,
      "learning_rate": 0.000121748767309939,
      "loss": 0.6806,
      "step": 44934
    },
    {
      "epoch": 3.866041469500129,
      "grad_norm": 20.25,
      "learning_rate": 0.00012173108346973938,
      "loss": 1.1202,
      "step": 44935
    },
    {
      "epoch": 3.8661275058074507,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001217134007359026,
      "loss": 1.9006,
      "step": 44936
    },
    {
      "epoch": 3.8662135421147723,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001216957191084806,
      "loss": 1.4751,
      "step": 44937
    },
    {
      "epoch": 3.866299578422094,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00012167803858752513,
      "loss": 1.0459,
      "step": 44938
    },
    {
      "epoch": 3.866385614729416,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00012166035917308771,
      "loss": 1.7837,
      "step": 44939
    },
    {
      "epoch": 3.8664716510367376,
      "grad_norm": 40.75,
      "learning_rate": 0.00012164268086522012,
      "loss": 1.7541,
      "step": 44940
    },
    {
      "epoch": 3.8665576873440592,
      "grad_norm": 0.96875,
      "learning_rate": 0.00012162500366397428,
      "loss": 0.8594,
      "step": 44941
    },
    {
      "epoch": 3.866643723651381,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001216073275694014,
      "loss": 0.6789,
      "step": 44942
    },
    {
      "epoch": 3.8667297599587025,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001215896525815538,
      "loss": 0.7509,
      "step": 44943
    },
    {
      "epoch": 3.866815796266024,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00012157197870048275,
      "loss": 0.855,
      "step": 44944
    },
    {
      "epoch": 3.8669018325733457,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012155430592624001,
      "loss": 0.6345,
      "step": 44945
    },
    {
      "epoch": 3.866987868880668,
      "grad_norm": 2.90625,
      "learning_rate": 0.00012153663425887751,
      "loss": 1.1064,
      "step": 44946
    },
    {
      "epoch": 3.8670739051879894,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012151896369844656,
      "loss": 0.5189,
      "step": 44947
    },
    {
      "epoch": 3.867159941495311,
      "grad_norm": 1.125,
      "learning_rate": 0.00012150129424499907,
      "loss": 1.1239,
      "step": 44948
    },
    {
      "epoch": 3.8672459778026327,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001214836258985868,
      "loss": 0.9165,
      "step": 44949
    },
    {
      "epoch": 3.8673320141099543,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00012146595865926119,
      "loss": 0.7041,
      "step": 44950
    },
    {
      "epoch": 3.867418050417276,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012144829252707417,
      "loss": 0.7392,
      "step": 44951
    },
    {
      "epoch": 3.8675040867245976,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00012143062750207712,
      "loss": 0.9387,
      "step": 44952
    },
    {
      "epoch": 3.8675901230319196,
      "grad_norm": 3.4375,
      "learning_rate": 0.00012141296358432185,
      "loss": 0.8559,
      "step": 44953
    },
    {
      "epoch": 3.8676761593392412,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00012139530077386019,
      "loss": 0.8472,
      "step": 44954
    },
    {
      "epoch": 3.867762195646563,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012137763907074345,
      "loss": 0.6582,
      "step": 44955
    },
    {
      "epoch": 3.8678482319538845,
      "grad_norm": 124.5,
      "learning_rate": 0.0001213599784750235,
      "loss": 1.4461,
      "step": 44956
    },
    {
      "epoch": 3.867934268261206,
      "grad_norm": 1.671875,
      "learning_rate": 0.00012134231898675214,
      "loss": 1.3987,
      "step": 44957
    },
    {
      "epoch": 3.868020304568528,
      "grad_norm": 24.5,
      "learning_rate": 0.00012132466060598069,
      "loss": 1.3631,
      "step": 44958
    },
    {
      "epoch": 3.8681063408758494,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00012130700333276096,
      "loss": 1.3047,
      "step": 44959
    },
    {
      "epoch": 3.8681923771831714,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00012128934716714473,
      "loss": 1.4579,
      "step": 44960
    },
    {
      "epoch": 3.868278413490493,
      "grad_norm": 2096.0,
      "learning_rate": 0.00012127169210918337,
      "loss": 1.1524,
      "step": 44961
    },
    {
      "epoch": 3.8683644497978147,
      "grad_norm": 1.0625,
      "learning_rate": 0.00012125403815892866,
      "loss": 1.5271,
      "step": 44962
    },
    {
      "epoch": 3.8684504861051363,
      "grad_norm": 1.296875,
      "learning_rate": 0.00012123638531643238,
      "loss": 1.6423,
      "step": 44963
    },
    {
      "epoch": 3.868536522412458,
      "grad_norm": 1.03125,
      "learning_rate": 0.00012121873358174585,
      "loss": 0.7227,
      "step": 44964
    },
    {
      "epoch": 3.86862255871978,
      "grad_norm": 0.9375,
      "learning_rate": 0.00012120108295492099,
      "loss": 0.7936,
      "step": 44965
    },
    {
      "epoch": 3.868708595027101,
      "grad_norm": 1.984375,
      "learning_rate": 0.00012118343343600912,
      "loss": 1.0466,
      "step": 44966
    },
    {
      "epoch": 3.8687946313344233,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001211657850250621,
      "loss": 1.3757,
      "step": 44967
    },
    {
      "epoch": 3.868880667641745,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012114813772213156,
      "loss": 1.0106,
      "step": 44968
    },
    {
      "epoch": 3.8689667039490665,
      "grad_norm": 3.28125,
      "learning_rate": 0.0001211304915272689,
      "loss": 1.29,
      "step": 44969
    },
    {
      "epoch": 3.869052740256388,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001211128464405259,
      "loss": 1.2554,
      "step": 44970
    },
    {
      "epoch": 3.8691387765637097,
      "grad_norm": 1.328125,
      "learning_rate": 0.00012109520246195427,
      "loss": 1.4392,
      "step": 44971
    },
    {
      "epoch": 3.869224812871032,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00012107755959160532,
      "loss": 1.0349,
      "step": 44972
    },
    {
      "epoch": 3.869310849178353,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00012105991782953085,
      "loss": 0.7732,
      "step": 44973
    },
    {
      "epoch": 3.869396885485675,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00012104227717578253,
      "loss": 0.7941,
      "step": 44974
    },
    {
      "epoch": 3.8694829217929967,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00012102463763041171,
      "loss": 1.5606,
      "step": 44975
    },
    {
      "epoch": 3.8695689581003183,
      "grad_norm": 1.15625,
      "learning_rate": 0.00012100699919347025,
      "loss": 1.4677,
      "step": 44976
    },
    {
      "epoch": 3.86965499440764,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012098936186500948,
      "loss": 0.6532,
      "step": 44977
    },
    {
      "epoch": 3.8697410307149616,
      "grad_norm": 1.09375,
      "learning_rate": 0.00012097172564508108,
      "loss": 1.4838,
      "step": 44978
    },
    {
      "epoch": 3.8698270670222836,
      "grad_norm": 1.15625,
      "learning_rate": 0.00012095409053373668,
      "loss": 1.0254,
      "step": 44979
    },
    {
      "epoch": 3.8699131033296053,
      "grad_norm": 1064.0,
      "learning_rate": 0.00012093645653102786,
      "loss": 1.013,
      "step": 44980
    },
    {
      "epoch": 3.869999139636927,
      "grad_norm": 3.25,
      "learning_rate": 0.00012091882363700635,
      "loss": 1.0161,
      "step": 44981
    },
    {
      "epoch": 3.8700851759442485,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012090119185172338,
      "loss": 0.6725,
      "step": 44982
    },
    {
      "epoch": 3.87017121225157,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012088356117523069,
      "loss": 0.986,
      "step": 44983
    },
    {
      "epoch": 3.8702572485588917,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00012086593160757997,
      "loss": 1.7209,
      "step": 44984
    },
    {
      "epoch": 3.8703432848662134,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012084830314882255,
      "loss": 0.7249,
      "step": 44985
    },
    {
      "epoch": 3.8704293211735354,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012083067579901014,
      "loss": 1.3763,
      "step": 44986
    },
    {
      "epoch": 3.870515357480857,
      "grad_norm": 1.53125,
      "learning_rate": 0.00012081304955819434,
      "loss": 0.8257,
      "step": 44987
    },
    {
      "epoch": 3.8706013937881787,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001207954244264265,
      "loss": 1.5562,
      "step": 44988
    },
    {
      "epoch": 3.8706874300955003,
      "grad_norm": 1.125,
      "learning_rate": 0.0001207778004037583,
      "loss": 1.3517,
      "step": 44989
    },
    {
      "epoch": 3.870773466402822,
      "grad_norm": 1.28125,
      "learning_rate": 0.00012076017749024143,
      "loss": 1.0464,
      "step": 44990
    },
    {
      "epoch": 3.8708595027101436,
      "grad_norm": 6.21875,
      "learning_rate": 0.00012074255568592707,
      "loss": 1.3739,
      "step": 44991
    },
    {
      "epoch": 3.870945539017465,
      "grad_norm": 1.125,
      "learning_rate": 0.00012072493499086718,
      "loss": 0.9551,
      "step": 44992
    },
    {
      "epoch": 3.8710315753247873,
      "grad_norm": 1.375,
      "learning_rate": 0.00012070731540511292,
      "loss": 1.1299,
      "step": 44993
    },
    {
      "epoch": 3.871117611632109,
      "grad_norm": 1.328125,
      "learning_rate": 0.00012068969692871607,
      "loss": 1.9739,
      "step": 44994
    },
    {
      "epoch": 3.8712036479394305,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00012067207956172815,
      "loss": 1.4251,
      "step": 44995
    },
    {
      "epoch": 3.871289684246752,
      "grad_norm": 1.234375,
      "learning_rate": 0.00012065446330420049,
      "loss": 1.3144,
      "step": 44996
    },
    {
      "epoch": 3.8713757205540738,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00012063684815618476,
      "loss": 1.1243,
      "step": 44997
    },
    {
      "epoch": 3.8714617568613954,
      "grad_norm": 1.203125,
      "learning_rate": 0.00012061923411773262,
      "loss": 0.7613,
      "step": 44998
    },
    {
      "epoch": 3.871547793168717,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012060162118889528,
      "loss": 0.9538,
      "step": 44999
    },
    {
      "epoch": 3.871633829476039,
      "grad_norm": 3.53125,
      "learning_rate": 0.00012058400936972435,
      "loss": 1.5193,
      "step": 45000
    },
    {
      "epoch": 3.8717198657833607,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012056639866027163,
      "loss": 0.9368,
      "step": 45001
    },
    {
      "epoch": 3.8718059020906823,
      "grad_norm": 0.90625,
      "learning_rate": 0.00012054878906058824,
      "loss": 1.0795,
      "step": 45002
    },
    {
      "epoch": 3.871891938398004,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012053118057072581,
      "loss": 0.8087,
      "step": 45003
    },
    {
      "epoch": 3.8719779747053256,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012051357319073602,
      "loss": 0.5848,
      "step": 45004
    },
    {
      "epoch": 3.872064011012647,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012049596692067011,
      "loss": 0.6337,
      "step": 45005
    },
    {
      "epoch": 3.872150047319969,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001204783617605798,
      "loss": 1.4094,
      "step": 45006
    },
    {
      "epoch": 3.872236083627291,
      "grad_norm": 19456.0,
      "learning_rate": 0.00012046075771051634,
      "loss": 1.5096,
      "step": 45007
    },
    {
      "epoch": 3.8723221199346125,
      "grad_norm": 1.375,
      "learning_rate": 0.00012044315477053136,
      "loss": 1.1052,
      "step": 45008
    },
    {
      "epoch": 3.872408156241934,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012042555294067648,
      "loss": 1.3622,
      "step": 45009
    },
    {
      "epoch": 3.8724941925492558,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00012040795222100287,
      "loss": 0.8143,
      "step": 45010
    },
    {
      "epoch": 3.8725802288565774,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00012039035261156217,
      "loss": 1.0889,
      "step": 45011
    },
    {
      "epoch": 3.872666265163899,
      "grad_norm": 1.734375,
      "learning_rate": 0.000120372754112406,
      "loss": 0.8392,
      "step": 45012
    },
    {
      "epoch": 3.8727523014712206,
      "grad_norm": 1.09375,
      "learning_rate": 0.00012035515672358555,
      "loss": 1.2352,
      "step": 45013
    },
    {
      "epoch": 3.8728383377785427,
      "grad_norm": 2.25,
      "learning_rate": 0.00012033756044515243,
      "loss": 1.1319,
      "step": 45014
    },
    {
      "epoch": 3.8729243740858643,
      "grad_norm": 1.03125,
      "learning_rate": 0.00012031996527715828,
      "loss": 0.7976,
      "step": 45015
    },
    {
      "epoch": 3.873010410393186,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00012030237121965404,
      "loss": 1.5471,
      "step": 45016
    },
    {
      "epoch": 3.8730964467005076,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00012028477827269185,
      "loss": 1.0081,
      "step": 45017
    },
    {
      "epoch": 3.873182483007829,
      "grad_norm": 1.125,
      "learning_rate": 0.00012026718643632267,
      "loss": 0.8196,
      "step": 45018
    },
    {
      "epoch": 3.8732685193151513,
      "grad_norm": 2.296875,
      "learning_rate": 0.00012024959571059813,
      "loss": 0.7882,
      "step": 45019
    },
    {
      "epoch": 3.8733545556224724,
      "grad_norm": 1.0,
      "learning_rate": 0.00012023200609556983,
      "loss": 1.169,
      "step": 45020
    },
    {
      "epoch": 3.8734405919297945,
      "grad_norm": 420.0,
      "learning_rate": 0.00012021441759128887,
      "loss": 1.1552,
      "step": 45021
    },
    {
      "epoch": 3.873526628237116,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012019683019780702,
      "loss": 0.8938,
      "step": 45022
    },
    {
      "epoch": 3.8736126645444378,
      "grad_norm": 2.0625,
      "learning_rate": 0.00012017924391517548,
      "loss": 2.0708,
      "step": 45023
    },
    {
      "epoch": 3.8736987008517594,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001201616587434457,
      "loss": 0.9339,
      "step": 45024
    },
    {
      "epoch": 3.873784737159081,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00012014407468266941,
      "loss": 1.5977,
      "step": 45025
    },
    {
      "epoch": 3.873870773466403,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012012649173289764,
      "loss": 0.5455,
      "step": 45026
    },
    {
      "epoch": 3.8739568097737243,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000120108909894182,
      "loss": 0.6927,
      "step": 45027
    },
    {
      "epoch": 3.8740428460810463,
      "grad_norm": 492.0,
      "learning_rate": 0.00012009132916657406,
      "loss": 1.5991,
      "step": 45028
    },
    {
      "epoch": 3.874128882388368,
      "grad_norm": 26.875,
      "learning_rate": 0.00012007374955012496,
      "loss": 1.5276,
      "step": 45029
    },
    {
      "epoch": 3.8742149186956896,
      "grad_norm": 2096.0,
      "learning_rate": 0.00012005617104488625,
      "loss": 0.9325,
      "step": 45030
    },
    {
      "epoch": 3.874300955003011,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00012003859365090947,
      "loss": 1.9231,
      "step": 45031
    },
    {
      "epoch": 3.874386991310333,
      "grad_norm": 1.359375,
      "learning_rate": 0.00012002101736824578,
      "loss": 0.9795,
      "step": 45032
    },
    {
      "epoch": 3.874473027617655,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012000344219694686,
      "loss": 0.8343,
      "step": 45033
    },
    {
      "epoch": 3.874559063924976,
      "grad_norm": 1.28125,
      "learning_rate": 0.00011998586813706381,
      "loss": 1.4409,
      "step": 45034
    },
    {
      "epoch": 3.874645100232298,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011996829518864821,
      "loss": 1.1078,
      "step": 45035
    },
    {
      "epoch": 3.8747311365396198,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011995072335175156,
      "loss": 0.5061,
      "step": 45036
    },
    {
      "epoch": 3.8748171728469414,
      "grad_norm": 1.171875,
      "learning_rate": 0.00011993315262642501,
      "loss": 1.5813,
      "step": 45037
    },
    {
      "epoch": 3.874903209154263,
      "grad_norm": 1.28125,
      "learning_rate": 0.00011991558301272004,
      "loss": 0.9771,
      "step": 45038
    },
    {
      "epoch": 3.8749892454615846,
      "grad_norm": 1.078125,
      "learning_rate": 0.00011989801451068822,
      "loss": 0.6105,
      "step": 45039
    },
    {
      "epoch": 3.8750752817689067,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001198804471203806,
      "loss": 2.006,
      "step": 45040
    },
    {
      "epoch": 3.8751613180762283,
      "grad_norm": 6.28125,
      "learning_rate": 0.00011986288084184876,
      "loss": 1.3437,
      "step": 45041
    },
    {
      "epoch": 3.87524735438355,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011984531567514423,
      "loss": 1.2139,
      "step": 45042
    },
    {
      "epoch": 3.8753333906908716,
      "grad_norm": 1.3125,
      "learning_rate": 0.00011982775162031811,
      "loss": 1.3009,
      "step": 45043
    },
    {
      "epoch": 3.875419426998193,
      "grad_norm": 14.875,
      "learning_rate": 0.00011981018867742182,
      "loss": 0.7275,
      "step": 45044
    },
    {
      "epoch": 3.875505463305515,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011979262684650693,
      "loss": 0.7957,
      "step": 45045
    },
    {
      "epoch": 3.8755914996128364,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001197750661276245,
      "loss": 0.7321,
      "step": 45046
    },
    {
      "epoch": 3.8756775359201585,
      "grad_norm": 54.5,
      "learning_rate": 0.00011975750652082623,
      "loss": 0.8294,
      "step": 45047
    },
    {
      "epoch": 3.87576357222748,
      "grad_norm": 1.359375,
      "learning_rate": 0.00011973994802616317,
      "loss": 1.2491,
      "step": 45048
    },
    {
      "epoch": 3.8758496085348018,
      "grad_norm": 868.0,
      "learning_rate": 0.00011972239064368678,
      "loss": 0.7364,
      "step": 45049
    },
    {
      "epoch": 3.8759356448421234,
      "grad_norm": 2.671875,
      "learning_rate": 0.00011970483437344859,
      "loss": 1.1128,
      "step": 45050
    },
    {
      "epoch": 3.876021681149445,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00011968727921549966,
      "loss": 1.423,
      "step": 45051
    },
    {
      "epoch": 3.8761077174567666,
      "grad_norm": 1.25,
      "learning_rate": 0.00011966972516989144,
      "loss": 1.2999,
      "step": 45052
    },
    {
      "epoch": 3.8761937537640883,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011965217223667535,
      "loss": 1.0613,
      "step": 45053
    },
    {
      "epoch": 3.8762797900714103,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011963462041590261,
      "loss": 1.0242,
      "step": 45054
    },
    {
      "epoch": 3.876365826378732,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011961706970762488,
      "loss": 0.6131,
      "step": 45055
    },
    {
      "epoch": 3.8764518626860536,
      "grad_norm": 1.34375,
      "learning_rate": 0.00011959952011189301,
      "loss": 1.4585,
      "step": 45056
    },
    {
      "epoch": 3.876537898993375,
      "grad_norm": 2.453125,
      "learning_rate": 0.00011958197162875856,
      "loss": 1.2663,
      "step": 45057
    },
    {
      "epoch": 3.876623935300697,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011956442425827308,
      "loss": 0.7605,
      "step": 45058
    },
    {
      "epoch": 3.8767099716080184,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00011954687800048742,
      "loss": 1.4459,
      "step": 45059
    },
    {
      "epoch": 3.87679600791534,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001195293328554532,
      "loss": 2.1631,
      "step": 45060
    },
    {
      "epoch": 3.876882044222662,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011951178882322183,
      "loss": 0.734,
      "step": 45061
    },
    {
      "epoch": 3.8769680805299838,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011949424590384433,
      "loss": 1.6711,
      "step": 45062
    },
    {
      "epoch": 3.8770541168373054,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011947670409737232,
      "loss": 1.9494,
      "step": 45063
    },
    {
      "epoch": 3.877140153144627,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00011945916340385677,
      "loss": 1.3429,
      "step": 45064
    },
    {
      "epoch": 3.8772261894519486,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00011944162382334916,
      "loss": 0.8606,
      "step": 45065
    },
    {
      "epoch": 3.8773122257592703,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00011942408535590094,
      "loss": 1.6252,
      "step": 45066
    },
    {
      "epoch": 3.877398262066592,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00011940654800156309,
      "loss": 1.2999,
      "step": 45067
    },
    {
      "epoch": 3.877484298373914,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011938901176038708,
      "loss": 1.4928,
      "step": 45068
    },
    {
      "epoch": 3.8775703346812356,
      "grad_norm": 1.0625,
      "learning_rate": 0.00011937147663242432,
      "loss": 1.8734,
      "step": 45069
    },
    {
      "epoch": 3.877656370988557,
      "grad_norm": 18.625,
      "learning_rate": 0.00011935394261772587,
      "loss": 1.2993,
      "step": 45070
    },
    {
      "epoch": 3.877742407295879,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00011933640971634302,
      "loss": 0.6485,
      "step": 45071
    },
    {
      "epoch": 3.8778284436032004,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011931887792832735,
      "loss": 1.0125,
      "step": 45072
    },
    {
      "epoch": 3.8779144799105225,
      "grad_norm": 230.0,
      "learning_rate": 0.0001193013472537297,
      "loss": 1.0896,
      "step": 45073
    },
    {
      "epoch": 3.8780005162178437,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011928381769260166,
      "loss": 0.6496,
      "step": 45074
    },
    {
      "epoch": 3.8780865525251658,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001192662892449945,
      "loss": 0.7381,
      "step": 45075
    },
    {
      "epoch": 3.8781725888324874,
      "grad_norm": 0.984375,
      "learning_rate": 0.00011924876191095929,
      "loss": 0.7639,
      "step": 45076
    },
    {
      "epoch": 3.878258625139809,
      "grad_norm": 1.125,
      "learning_rate": 0.00011923123569054751,
      "loss": 1.7164,
      "step": 45077
    },
    {
      "epoch": 3.8783446614471306,
      "grad_norm": 1.453125,
      "learning_rate": 0.00011921371058381015,
      "loss": 1.3,
      "step": 45078
    },
    {
      "epoch": 3.8784306977544523,
      "grad_norm": 1.25,
      "learning_rate": 0.00011919618659079867,
      "loss": 1.1472,
      "step": 45079
    },
    {
      "epoch": 3.8785167340617743,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00011917866371156439,
      "loss": 1.1157,
      "step": 45080
    },
    {
      "epoch": 3.8786027703690955,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011916114194615835,
      "loss": 1.3179,
      "step": 45081
    },
    {
      "epoch": 3.8786888066764176,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011914362129463185,
      "loss": 0.7807,
      "step": 45082
    },
    {
      "epoch": 3.878774842983739,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011912610175703636,
      "loss": 0.8431,
      "step": 45083
    },
    {
      "epoch": 3.878860879291061,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00011910858333342278,
      "loss": 0.6735,
      "step": 45084
    },
    {
      "epoch": 3.8789469155983824,
      "grad_norm": 9.3125,
      "learning_rate": 0.00011909106602384256,
      "loss": 0.768,
      "step": 45085
    },
    {
      "epoch": 3.879032951905704,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000119073549828347,
      "loss": 1.0064,
      "step": 45086
    },
    {
      "epoch": 3.879118988213026,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011905603474698706,
      "loss": 0.9993,
      "step": 45087
    },
    {
      "epoch": 3.8792050245203473,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00011903852077981431,
      "loss": 0.6453,
      "step": 45088
    },
    {
      "epoch": 3.8792910608276694,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011902100792687964,
      "loss": 0.932,
      "step": 45089
    },
    {
      "epoch": 3.879377097134991,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011900349618823431,
      "loss": 1.1433,
      "step": 45090
    },
    {
      "epoch": 3.8794631334423126,
      "grad_norm": 1.0,
      "learning_rate": 0.00011898598556392987,
      "loss": 1.5462,
      "step": 45091
    },
    {
      "epoch": 3.8795491697496343,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011896847605401723,
      "loss": 1.12,
      "step": 45092
    },
    {
      "epoch": 3.879635206056956,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00011895096765854779,
      "loss": 0.498,
      "step": 45093
    },
    {
      "epoch": 3.879721242364278,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00011893346037757258,
      "loss": 2.2675,
      "step": 45094
    },
    {
      "epoch": 3.8798072786715996,
      "grad_norm": 1.3125,
      "learning_rate": 0.00011891595421114282,
      "loss": 0.8484,
      "step": 45095
    },
    {
      "epoch": 3.879893314978921,
      "grad_norm": 1.546875,
      "learning_rate": 0.00011889844915930998,
      "loss": 0.6268,
      "step": 45096
    },
    {
      "epoch": 3.879979351286243,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001188809452221249,
      "loss": 0.8436,
      "step": 45097
    },
    {
      "epoch": 3.8800653875935645,
      "grad_norm": 1.125,
      "learning_rate": 0.00011886344239963887,
      "loss": 1.4548,
      "step": 45098
    },
    {
      "epoch": 3.880151423900886,
      "grad_norm": 50.75,
      "learning_rate": 0.00011884594069190335,
      "loss": 1.3519,
      "step": 45099
    },
    {
      "epoch": 3.8802374602082077,
      "grad_norm": 1.03125,
      "learning_rate": 0.00011882844009896914,
      "loss": 1.5243,
      "step": 45100
    },
    {
      "epoch": 3.8803234965155298,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011881094062088765,
      "loss": 1.0331,
      "step": 45101
    },
    {
      "epoch": 3.8804095328228514,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011879344225771016,
      "loss": 1.3774,
      "step": 45102
    },
    {
      "epoch": 3.880495569130173,
      "grad_norm": 2.15625,
      "learning_rate": 0.00011877594500948752,
      "loss": 1.3173,
      "step": 45103
    },
    {
      "epoch": 3.8805816054374946,
      "grad_norm": 63.0,
      "learning_rate": 0.0001187584488762713,
      "loss": 1.0514,
      "step": 45104
    },
    {
      "epoch": 3.8806676417448163,
      "grad_norm": 10.9375,
      "learning_rate": 0.00011874095385811229,
      "loss": 1.8386,
      "step": 45105
    },
    {
      "epoch": 3.880753678052138,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011872345995506184,
      "loss": 0.8546,
      "step": 45106
    },
    {
      "epoch": 3.8808397143594595,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001187059671671713,
      "loss": 1.0721,
      "step": 45107
    },
    {
      "epoch": 3.8809257506667816,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00011868847549449146,
      "loss": 0.9352,
      "step": 45108
    },
    {
      "epoch": 3.881011786974103,
      "grad_norm": 1544.0,
      "learning_rate": 0.00011867098493707367,
      "loss": 1.2599,
      "step": 45109
    },
    {
      "epoch": 3.881097823281425,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00011865349549496924,
      "loss": 1.3888,
      "step": 45110
    },
    {
      "epoch": 3.8811838595887465,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00011863600716822903,
      "loss": 0.8902,
      "step": 45111
    },
    {
      "epoch": 3.881269895896068,
      "grad_norm": 1.125,
      "learning_rate": 0.00011861851995690431,
      "loss": 0.9986,
      "step": 45112
    },
    {
      "epoch": 3.8813559322033897,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00011860103386104642,
      "loss": 1.106,
      "step": 45113
    },
    {
      "epoch": 3.8814419685107113,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00011858354888070616,
      "loss": 1.6283,
      "step": 45114
    },
    {
      "epoch": 3.8815280048180334,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001185660650159348,
      "loss": 0.8466,
      "step": 45115
    },
    {
      "epoch": 3.881614041125355,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011854858226678372,
      "loss": 0.6657,
      "step": 45116
    },
    {
      "epoch": 3.8817000774326766,
      "grad_norm": 1.296875,
      "learning_rate": 0.00011853110063330369,
      "loss": 1.6339,
      "step": 45117
    },
    {
      "epoch": 3.8817861137399983,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00011851362011554617,
      "loss": 1.2361,
      "step": 45118
    },
    {
      "epoch": 3.88187215004732,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00011849614071356191,
      "loss": 1.4289,
      "step": 45119
    },
    {
      "epoch": 3.8819581863546415,
      "grad_norm": 0.953125,
      "learning_rate": 0.00011847866242740229,
      "loss": 0.7563,
      "step": 45120
    },
    {
      "epoch": 3.882044222661963,
      "grad_norm": 3.734375,
      "learning_rate": 0.00011846118525711851,
      "loss": 0.5878,
      "step": 45121
    },
    {
      "epoch": 3.882130258969285,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011844370920276137,
      "loss": 0.8507,
      "step": 45122
    },
    {
      "epoch": 3.882216295276607,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011842623426438226,
      "loss": 1.0762,
      "step": 45123
    },
    {
      "epoch": 3.8823023315839285,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011840876044203224,
      "loss": 0.6637,
      "step": 45124
    },
    {
      "epoch": 3.88238836789125,
      "grad_norm": 9.1875,
      "learning_rate": 0.00011839128773576235,
      "loss": 0.7371,
      "step": 45125
    },
    {
      "epoch": 3.8824744041985717,
      "grad_norm": 1.1875,
      "learning_rate": 0.00011837381614562364,
      "loss": 1.2427,
      "step": 45126
    },
    {
      "epoch": 3.8825604405058933,
      "grad_norm": 38.25,
      "learning_rate": 0.0001183563456716673,
      "loss": 0.9227,
      "step": 45127
    },
    {
      "epoch": 3.882646476813215,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011833887631394447,
      "loss": 1.1709,
      "step": 45128
    },
    {
      "epoch": 3.882732513120537,
      "grad_norm": 1.328125,
      "learning_rate": 0.00011832140807250635,
      "loss": 1.5061,
      "step": 45129
    },
    {
      "epoch": 3.8828185494278586,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00011830394094740372,
      "loss": 0.6201,
      "step": 45130
    },
    {
      "epoch": 3.8829045857351803,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00011828647493868777,
      "loss": 1.6329,
      "step": 45131
    },
    {
      "epoch": 3.882990622042502,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011826901004640983,
      "loss": 1.3658,
      "step": 45132
    },
    {
      "epoch": 3.8830766583498235,
      "grad_norm": 1.421875,
      "learning_rate": 0.00011825154627062062,
      "loss": 1.1341,
      "step": 45133
    },
    {
      "epoch": 3.8831626946571456,
      "grad_norm": 1584.0,
      "learning_rate": 0.00011823408361137162,
      "loss": 1.2028,
      "step": 45134
    },
    {
      "epoch": 3.8832487309644668,
      "grad_norm": 10.5625,
      "learning_rate": 0.00011821662206871342,
      "loss": 1.3963,
      "step": 45135
    },
    {
      "epoch": 3.883334767271789,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011819916164269745,
      "loss": 1.155,
      "step": 45136
    },
    {
      "epoch": 3.8834208035791105,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011818170233337478,
      "loss": 0.9776,
      "step": 45137
    },
    {
      "epoch": 3.883506839886432,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00011816424414079618,
      "loss": 0.6685,
      "step": 45138
    },
    {
      "epoch": 3.8835928761937537,
      "grad_norm": 33.0,
      "learning_rate": 0.00011814678706501297,
      "loss": 1.2755,
      "step": 45139
    },
    {
      "epoch": 3.8836789125010753,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00011812933110607621,
      "loss": 0.8656,
      "step": 45140
    },
    {
      "epoch": 3.8837649488083974,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001181118762640368,
      "loss": 0.9761,
      "step": 45141
    },
    {
      "epoch": 3.8838509851157186,
      "grad_norm": 740.0,
      "learning_rate": 0.00011809442253894586,
      "loss": 0.5602,
      "step": 45142
    },
    {
      "epoch": 3.8839370214230406,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011807696993085465,
      "loss": 1.0525,
      "step": 45143
    },
    {
      "epoch": 3.8840230577303623,
      "grad_norm": 1.265625,
      "learning_rate": 0.00011805951843981383,
      "loss": 1.5308,
      "step": 45144
    },
    {
      "epoch": 3.884109094037684,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011804206806587481,
      "loss": 0.977,
      "step": 45145
    },
    {
      "epoch": 3.8841951303450055,
      "grad_norm": 2.03125,
      "learning_rate": 0.00011802461880908827,
      "loss": 0.6009,
      "step": 45146
    },
    {
      "epoch": 3.884281166652327,
      "grad_norm": 2.078125,
      "learning_rate": 0.00011800717066950545,
      "loss": 1.0253,
      "step": 45147
    },
    {
      "epoch": 3.884367202959649,
      "grad_norm": 1.625,
      "learning_rate": 0.00011798972364717747,
      "loss": 1.6719,
      "step": 45148
    },
    {
      "epoch": 3.8844532392669704,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011797227774215519,
      "loss": 0.8044,
      "step": 45149
    },
    {
      "epoch": 3.8845392755742925,
      "grad_norm": 7.21875,
      "learning_rate": 0.0001179548329544896,
      "loss": 1.5126,
      "step": 45150
    },
    {
      "epoch": 3.884625311881614,
      "grad_norm": 1.8125,
      "learning_rate": 0.00011793738928423204,
      "loss": 1.246,
      "step": 45151
    },
    {
      "epoch": 3.8847113481889357,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00011791994673143314,
      "loss": 0.9499,
      "step": 45152
    },
    {
      "epoch": 3.8847973844962573,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001179025052961441,
      "loss": 0.6764,
      "step": 45153
    },
    {
      "epoch": 3.884883420803579,
      "grad_norm": 1.15625,
      "learning_rate": 0.000117885064978416,
      "loss": 0.9972,
      "step": 45154
    },
    {
      "epoch": 3.884969457110901,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011786762577829968,
      "loss": 1.7083,
      "step": 45155
    },
    {
      "epoch": 3.8850554934182226,
      "grad_norm": 8.5,
      "learning_rate": 0.00011785018769584627,
      "loss": 0.7873,
      "step": 45156
    },
    {
      "epoch": 3.8851415297255443,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00011783275073110683,
      "loss": 0.8446,
      "step": 45157
    },
    {
      "epoch": 3.885227566032866,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011781531488413211,
      "loss": 0.7337,
      "step": 45158
    },
    {
      "epoch": 3.8853136023401875,
      "grad_norm": 1.46875,
      "learning_rate": 0.00011779788015497345,
      "loss": 1.1011,
      "step": 45159
    },
    {
      "epoch": 3.885399638647509,
      "grad_norm": 2.171875,
      "learning_rate": 0.00011778044654368148,
      "loss": 0.6264,
      "step": 45160
    },
    {
      "epoch": 3.8854856749548308,
      "grad_norm": 2.21875,
      "learning_rate": 0.00011776301405030738,
      "loss": 0.7868,
      "step": 45161
    },
    {
      "epoch": 3.885571711262153,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001177455826749022,
      "loss": 1.3485,
      "step": 45162
    },
    {
      "epoch": 3.8856577475694745,
      "grad_norm": 1.171875,
      "learning_rate": 0.00011772815241751666,
      "loss": 1.66,
      "step": 45163
    },
    {
      "epoch": 3.885743783876796,
      "grad_norm": 1.203125,
      "learning_rate": 0.00011771072327820215,
      "loss": 0.9248,
      "step": 45164
    },
    {
      "epoch": 3.8858298201841177,
      "grad_norm": 4.40625,
      "learning_rate": 0.00011769329525700934,
      "loss": 0.6427,
      "step": 45165
    },
    {
      "epoch": 3.8859158564914393,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00011767586835398925,
      "loss": 1.1493,
      "step": 45166
    },
    {
      "epoch": 3.886001892798761,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011765844256919301,
      "loss": 1.4425,
      "step": 45167
    },
    {
      "epoch": 3.8860879291060826,
      "grad_norm": 12.375,
      "learning_rate": 0.00011764101790267134,
      "loss": 1.5361,
      "step": 45168
    },
    {
      "epoch": 3.8861739654134047,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011762359435447534,
      "loss": 0.717,
      "step": 45169
    },
    {
      "epoch": 3.8862600017207263,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011760617192465606,
      "loss": 1.0569,
      "step": 45170
    },
    {
      "epoch": 3.886346038028048,
      "grad_norm": 6.625,
      "learning_rate": 0.00011758875061326424,
      "loss": 0.6444,
      "step": 45171
    },
    {
      "epoch": 3.8864320743353695,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011757133042035095,
      "loss": 1.2802,
      "step": 45172
    },
    {
      "epoch": 3.886518110642691,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011755391134596732,
      "loss": 0.9798,
      "step": 45173
    },
    {
      "epoch": 3.8866041469500128,
      "grad_norm": 25.0,
      "learning_rate": 0.00011753649339016392,
      "loss": 1.2821,
      "step": 45174
    },
    {
      "epoch": 3.8866901832573344,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00011751907655299204,
      "loss": 1.0631,
      "step": 45175
    },
    {
      "epoch": 3.8867762195646565,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00011750166083450237,
      "loss": 1.0139,
      "step": 45176
    },
    {
      "epoch": 3.886862255871978,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001174842462347459,
      "loss": 1.2351,
      "step": 45177
    },
    {
      "epoch": 3.8869482921792997,
      "grad_norm": 1.28125,
      "learning_rate": 0.00011746683275377379,
      "loss": 0.9918,
      "step": 45178
    },
    {
      "epoch": 3.8870343284866213,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011744942039163664,
      "loss": 1.5161,
      "step": 45179
    },
    {
      "epoch": 3.887120364793943,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00011743200914838554,
      "loss": 1.3005,
      "step": 45180
    },
    {
      "epoch": 3.8872064011012646,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001174145990240716,
      "loss": 0.9898,
      "step": 45181
    },
    {
      "epoch": 3.887292437408586,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011739719001874533,
      "loss": 0.8233,
      "step": 45182
    },
    {
      "epoch": 3.8873784737159083,
      "grad_norm": 1.203125,
      "learning_rate": 0.00011737978213245787,
      "loss": 1.1279,
      "step": 45183
    },
    {
      "epoch": 3.88746451002323,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00011736237536526028,
      "loss": 0.957,
      "step": 45184
    },
    {
      "epoch": 3.8875505463305515,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001173449697172032,
      "loss": 1.306,
      "step": 45185
    },
    {
      "epoch": 3.887636582637873,
      "grad_norm": 1.328125,
      "learning_rate": 0.00011732756518833771,
      "loss": 1.2117,
      "step": 45186
    },
    {
      "epoch": 3.8877226189451948,
      "grad_norm": 0.90625,
      "learning_rate": 0.00011731016177871473,
      "loss": 0.8187,
      "step": 45187
    },
    {
      "epoch": 3.8878086552525164,
      "grad_norm": 1.078125,
      "learning_rate": 0.00011729275948838497,
      "loss": 1.0849,
      "step": 45188
    },
    {
      "epoch": 3.887894691559838,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001172753583173996,
      "loss": 0.4107,
      "step": 45189
    },
    {
      "epoch": 3.88798072786716,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00011725795826580927,
      "loss": 1.5594,
      "step": 45190
    },
    {
      "epoch": 3.8880667641744817,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00011724055933366495,
      "loss": 0.9623,
      "step": 45191
    },
    {
      "epoch": 3.8881528004818033,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001172231615210177,
      "loss": 1.32,
      "step": 45192
    },
    {
      "epoch": 3.888238836789125,
      "grad_norm": 1.390625,
      "learning_rate": 0.00011720576482791807,
      "loss": 1.3432,
      "step": 45193
    },
    {
      "epoch": 3.8883248730964466,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011718836925441717,
      "loss": 0.9722,
      "step": 45194
    },
    {
      "epoch": 3.8884109094037687,
      "grad_norm": 0.875,
      "learning_rate": 0.00011717097480056598,
      "loss": 0.9803,
      "step": 45195
    },
    {
      "epoch": 3.88849694571109,
      "grad_norm": 1.25,
      "learning_rate": 0.00011715358146641508,
      "loss": 1.0608,
      "step": 45196
    },
    {
      "epoch": 3.888582982018412,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00011713618925201546,
      "loss": 1.7514,
      "step": 45197
    },
    {
      "epoch": 3.8886690183257335,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011711879815741822,
      "loss": 1.3624,
      "step": 45198
    },
    {
      "epoch": 3.888755054633055,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001171014081826739,
      "loss": 1.3265,
      "step": 45199
    },
    {
      "epoch": 3.8888410909403768,
      "grad_norm": 11.75,
      "learning_rate": 0.00011708401932783341,
      "loss": 1.3583,
      "step": 45200
    },
    {
      "epoch": 3.8889271272476984,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00011706663159294778,
      "loss": 0.8546,
      "step": 45201
    },
    {
      "epoch": 3.8890131635550205,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00011704924497806774,
      "loss": 1.5314,
      "step": 45202
    },
    {
      "epoch": 3.8890991998623416,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001170318594832443,
      "loss": 1.3868,
      "step": 45203
    },
    {
      "epoch": 3.8891852361696637,
      "grad_norm": 1.234375,
      "learning_rate": 0.00011701447510852803,
      "loss": 1.12,
      "step": 45204
    },
    {
      "epoch": 3.8892712724769853,
      "grad_norm": 1.4375,
      "learning_rate": 0.00011699709185397011,
      "loss": 1.0255,
      "step": 45205
    },
    {
      "epoch": 3.889357308784307,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011697970971962108,
      "loss": 0.9873,
      "step": 45206
    },
    {
      "epoch": 3.8894433450916286,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00011696232870553186,
      "loss": 0.7523,
      "step": 45207
    },
    {
      "epoch": 3.88952938139895,
      "grad_norm": 16.5,
      "learning_rate": 0.00011694494881175355,
      "loss": 1.0503,
      "step": 45208
    },
    {
      "epoch": 3.8896154177062723,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011692757003833653,
      "loss": 1.7634,
      "step": 45209
    },
    {
      "epoch": 3.8897014540135935,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001169101923853319,
      "loss": 0.5557,
      "step": 45210
    },
    {
      "epoch": 3.8897874903209155,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001168928158527906,
      "loss": 1.557,
      "step": 45211
    },
    {
      "epoch": 3.889873526628237,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001168754404407632,
      "loss": 0.7485,
      "step": 45212
    },
    {
      "epoch": 3.889959562935559,
      "grad_norm": 6.6875,
      "learning_rate": 0.00011685806614930055,
      "loss": 0.9633,
      "step": 45213
    },
    {
      "epoch": 3.8900455992428804,
      "grad_norm": 1.640625,
      "learning_rate": 0.00011684069297845373,
      "loss": 1.9671,
      "step": 45214
    },
    {
      "epoch": 3.890131635550202,
      "grad_norm": 1.46875,
      "learning_rate": 0.00011682332092827319,
      "loss": 2.0285,
      "step": 45215
    },
    {
      "epoch": 3.890217671857524,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00011680594999881012,
      "loss": 0.6551,
      "step": 45216
    },
    {
      "epoch": 3.8903037081648457,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001167885801901149,
      "loss": 1.5344,
      "step": 45217
    },
    {
      "epoch": 3.8903897444721673,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011677121150223857,
      "loss": 1.7807,
      "step": 45218
    },
    {
      "epoch": 3.890475780779489,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011675384393523208,
      "loss": 1.1756,
      "step": 45219
    },
    {
      "epoch": 3.8905618170868106,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001167364774891459,
      "loss": 1.1847,
      "step": 45220
    },
    {
      "epoch": 3.890647853394132,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011671911216403097,
      "loss": 1.9756,
      "step": 45221
    },
    {
      "epoch": 3.890733889701454,
      "grad_norm": 214.0,
      "learning_rate": 0.00011670174795993827,
      "loss": 1.2446,
      "step": 45222
    },
    {
      "epoch": 3.890819926008776,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011668438487691824,
      "loss": 1.195,
      "step": 45223
    },
    {
      "epoch": 3.8909059623160975,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011666702291502185,
      "loss": 0.6359,
      "step": 45224
    },
    {
      "epoch": 3.890991998623419,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001166496620743,
      "loss": 0.8673,
      "step": 45225
    },
    {
      "epoch": 3.891078034930741,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00011663230235480321,
      "loss": 0.9742,
      "step": 45226
    },
    {
      "epoch": 3.8911640712380624,
      "grad_norm": 1.140625,
      "learning_rate": 0.00011661494375658233,
      "loss": 0.4307,
      "step": 45227
    },
    {
      "epoch": 3.891250107545384,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011659758627968837,
      "loss": 0.7638,
      "step": 45228
    },
    {
      "epoch": 3.8913361438527057,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00011658022992417173,
      "loss": 1.2842,
      "step": 45229
    },
    {
      "epoch": 3.8914221801600277,
      "grad_norm": 2.015625,
      "learning_rate": 0.00011656287469008348,
      "loss": 0.9281,
      "step": 45230
    },
    {
      "epoch": 3.8915082164673493,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011654552057747409,
      "loss": 1.617,
      "step": 45231
    },
    {
      "epoch": 3.891594252774671,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001165281675863945,
      "loss": 1.317,
      "step": 45232
    },
    {
      "epoch": 3.8916802890819926,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001165108157168956,
      "loss": 0.6948,
      "step": 45233
    },
    {
      "epoch": 3.891766325389314,
      "grad_norm": 188.0,
      "learning_rate": 0.0001164934649690278,
      "loss": 1.1102,
      "step": 45234
    },
    {
      "epoch": 3.891852361696636,
      "grad_norm": 350.0,
      "learning_rate": 0.00011647611534284202,
      "loss": 0.6357,
      "step": 45235
    },
    {
      "epoch": 3.8919383980039575,
      "grad_norm": 1.421875,
      "learning_rate": 0.00011645876683838919,
      "loss": 1.3545,
      "step": 45236
    },
    {
      "epoch": 3.8920244343112795,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011644141945571957,
      "loss": 0.9314,
      "step": 45237
    },
    {
      "epoch": 3.892110470618601,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011642407319488446,
      "loss": 0.5898,
      "step": 45238
    },
    {
      "epoch": 3.892196506925923,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00011640672805593422,
      "loss": 1.3605,
      "step": 45239
    },
    {
      "epoch": 3.8922825432332444,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00011638938403891974,
      "loss": 1.1736,
      "step": 45240
    },
    {
      "epoch": 3.892368579540566,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00011637204114389178,
      "loss": 1.6539,
      "step": 45241
    },
    {
      "epoch": 3.8924546158478877,
      "grad_norm": 2.375,
      "learning_rate": 0.00011635469937090088,
      "loss": 1.499,
      "step": 45242
    },
    {
      "epoch": 3.8925406521552093,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00011633735871999785,
      "loss": 1.3994,
      "step": 45243
    },
    {
      "epoch": 3.8926266884625313,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011632001919123359,
      "loss": 0.5811,
      "step": 45244
    },
    {
      "epoch": 3.892712724769853,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001163026807846585,
      "loss": 0.8608,
      "step": 45245
    },
    {
      "epoch": 3.8927987610771746,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011628534350032355,
      "loss": 0.6044,
      "step": 45246
    },
    {
      "epoch": 3.892884797384496,
      "grad_norm": 1.609375,
      "learning_rate": 0.00011626800733827924,
      "loss": 1.3355,
      "step": 45247
    },
    {
      "epoch": 3.892970833691818,
      "grad_norm": 1.21875,
      "learning_rate": 0.00011625067229857638,
      "loss": 1.1897,
      "step": 45248
    },
    {
      "epoch": 3.89305686999914,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00011623333838126582,
      "loss": 1.5141,
      "step": 45249
    },
    {
      "epoch": 3.893142906306461,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011621600558639795,
      "loss": 0.6835,
      "step": 45250
    },
    {
      "epoch": 3.893228942613783,
      "grad_norm": 15.6875,
      "learning_rate": 0.00011619867391402361,
      "loss": 1.4512,
      "step": 45251
    },
    {
      "epoch": 3.893314978921105,
      "grad_norm": 1.140625,
      "learning_rate": 0.00011618134336419361,
      "loss": 1.5587,
      "step": 45252
    },
    {
      "epoch": 3.8934010152284264,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011616401393695847,
      "loss": 0.8361,
      "step": 45253
    },
    {
      "epoch": 3.893487051535748,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011614668563236891,
      "loss": 1.2207,
      "step": 45254
    },
    {
      "epoch": 3.8935730878430697,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00011612935845047573,
      "loss": 1.3787,
      "step": 45255
    },
    {
      "epoch": 3.8936591241503917,
      "grad_norm": 2.5625,
      "learning_rate": 0.00011611203239132945,
      "loss": 1.4235,
      "step": 45256
    },
    {
      "epoch": 3.893745160457713,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00011609470745498091,
      "loss": 0.6949,
      "step": 45257
    },
    {
      "epoch": 3.893831196765035,
      "grad_norm": 1.0,
      "learning_rate": 0.00011607738364148057,
      "loss": 1.3747,
      "step": 45258
    },
    {
      "epoch": 3.8939172330723566,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011606006095087917,
      "loss": 0.865,
      "step": 45259
    },
    {
      "epoch": 3.8940032693796782,
      "grad_norm": 1.53125,
      "learning_rate": 0.00011604273938322757,
      "loss": 1.4899,
      "step": 45260
    },
    {
      "epoch": 3.894089305687,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011602541893857621,
      "loss": 1.1923,
      "step": 45261
    },
    {
      "epoch": 3.8941753419943215,
      "grad_norm": 1.25,
      "learning_rate": 0.00011600809961697572,
      "loss": 1.5064,
      "step": 45262
    },
    {
      "epoch": 3.8942613783016435,
      "grad_norm": 2.03125,
      "learning_rate": 0.00011599078141847708,
      "loss": 1.7728,
      "step": 45263
    },
    {
      "epoch": 3.8943474146089647,
      "grad_norm": 0.953125,
      "learning_rate": 0.00011597346434313055,
      "loss": 1.5619,
      "step": 45264
    },
    {
      "epoch": 3.894433450916287,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00011595614839098694,
      "loss": 1.3605,
      "step": 45265
    },
    {
      "epoch": 3.8945194872236084,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00011593883356209706,
      "loss": 0.8312,
      "step": 45266
    },
    {
      "epoch": 3.89460552353093,
      "grad_norm": 1.28125,
      "learning_rate": 0.00011592151985651122,
      "loss": 1.0898,
      "step": 45267
    },
    {
      "epoch": 3.8946915598382517,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011590420727428025,
      "loss": 0.7561,
      "step": 45268
    },
    {
      "epoch": 3.8947775961455733,
      "grad_norm": 1.359375,
      "learning_rate": 0.00011588689581545492,
      "loss": 0.8204,
      "step": 45269
    },
    {
      "epoch": 3.8948636324528954,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011586958548008558,
      "loss": 1.29,
      "step": 45270
    },
    {
      "epoch": 3.894949668760217,
      "grad_norm": 2.109375,
      "learning_rate": 0.00011585227626822314,
      "loss": 1.5407,
      "step": 45271
    },
    {
      "epoch": 3.8950357050675386,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011583496817991795,
      "loss": 1.0531,
      "step": 45272
    },
    {
      "epoch": 3.8951217413748602,
      "grad_norm": 35.25,
      "learning_rate": 0.0001158176612152207,
      "loss": 1.0344,
      "step": 45273
    },
    {
      "epoch": 3.895207777682182,
      "grad_norm": 2.296875,
      "learning_rate": 0.00011580035537418216,
      "loss": 0.9866,
      "step": 45274
    },
    {
      "epoch": 3.8952938139895035,
      "grad_norm": 1.21875,
      "learning_rate": 0.00011578305065685279,
      "loss": 1.0641,
      "step": 45275
    },
    {
      "epoch": 3.895379850296825,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011576574706328341,
      "loss": 1.5319,
      "step": 45276
    },
    {
      "epoch": 3.895465886604147,
      "grad_norm": 0.953125,
      "learning_rate": 0.00011574844459352435,
      "loss": 1.1672,
      "step": 45277
    },
    {
      "epoch": 3.895551922911469,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001157311432476264,
      "loss": 1.0527,
      "step": 45278
    },
    {
      "epoch": 3.8956379592187904,
      "grad_norm": 1.375,
      "learning_rate": 0.00011571384302564026,
      "loss": 1.4838,
      "step": 45279
    },
    {
      "epoch": 3.895723995526112,
      "grad_norm": 1.296875,
      "learning_rate": 0.00011569654392761619,
      "loss": 1.0527,
      "step": 45280
    },
    {
      "epoch": 3.8958100318334337,
      "grad_norm": 1.03125,
      "learning_rate": 0.00011567924595360502,
      "loss": 1.0031,
      "step": 45281
    },
    {
      "epoch": 3.8958960681407553,
      "grad_norm": 364.0,
      "learning_rate": 0.00011566194910365746,
      "loss": 1.1498,
      "step": 45282
    },
    {
      "epoch": 3.895982104448077,
      "grad_norm": 3.734375,
      "learning_rate": 0.00011564465337782382,
      "loss": 0.9773,
      "step": 45283
    },
    {
      "epoch": 3.896068140755399,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011562735877615476,
      "loss": 0.8825,
      "step": 45284
    },
    {
      "epoch": 3.8961541770627206,
      "grad_norm": 8.1875,
      "learning_rate": 0.00011561006529870111,
      "loss": 1.4627,
      "step": 45285
    },
    {
      "epoch": 3.8962402133700422,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00011559277294551312,
      "loss": 1.4128,
      "step": 45286
    },
    {
      "epoch": 3.896326249677364,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00011557548171664162,
      "loss": 1.5426,
      "step": 45287
    },
    {
      "epoch": 3.8964122859846855,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011555819161213693,
      "loss": 1.1328,
      "step": 45288
    },
    {
      "epoch": 3.896498322292007,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001155409026320498,
      "loss": 0.6169,
      "step": 45289
    },
    {
      "epoch": 3.8965843585993287,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011552361477643087,
      "loss": 1.8886,
      "step": 45290
    },
    {
      "epoch": 3.896670394906651,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011550632804533046,
      "loss": 1.5599,
      "step": 45291
    },
    {
      "epoch": 3.8967564312139724,
      "grad_norm": 1.375,
      "learning_rate": 0.00011548904243879921,
      "loss": 1.5892,
      "step": 45292
    },
    {
      "epoch": 3.896842467521294,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001154717579568879,
      "loss": 1.0259,
      "step": 45293
    },
    {
      "epoch": 3.8969285038286157,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011545447459964681,
      "loss": 1.4331,
      "step": 45294
    },
    {
      "epoch": 3.8970145401359373,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011543719236712651,
      "loss": 0.8432,
      "step": 45295
    },
    {
      "epoch": 3.897100576443259,
      "grad_norm": 1.125,
      "learning_rate": 0.00011541991125937784,
      "loss": 1.0907,
      "step": 45296
    },
    {
      "epoch": 3.8971866127505805,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011540263127645101,
      "loss": 0.8129,
      "step": 45297
    },
    {
      "epoch": 3.8972726490579026,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00011538535241839676,
      "loss": 1.4397,
      "step": 45298
    },
    {
      "epoch": 3.8973586853652242,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00011536807468526544,
      "loss": 0.9951,
      "step": 45299
    },
    {
      "epoch": 3.897444721672546,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001153507980771077,
      "loss": 0.8384,
      "step": 45300
    },
    {
      "epoch": 3.8975307579798675,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011533352259397417,
      "loss": 1.3985,
      "step": 45301
    },
    {
      "epoch": 3.897616794287189,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00011531624823591519,
      "loss": 1.0876,
      "step": 45302
    },
    {
      "epoch": 3.8977028305945107,
      "grad_norm": 0.921875,
      "learning_rate": 0.00011529897500298131,
      "loss": 0.6981,
      "step": 45303
    },
    {
      "epoch": 3.8977888669018323,
      "grad_norm": 0.96875,
      "learning_rate": 0.00011528170289522328,
      "loss": 1.6617,
      "step": 45304
    },
    {
      "epoch": 3.8978749032091544,
      "grad_norm": 4.5,
      "learning_rate": 0.00011526443191269131,
      "loss": 1.399,
      "step": 45305
    },
    {
      "epoch": 3.897960939516476,
      "grad_norm": 53.25,
      "learning_rate": 0.00011524716205543601,
      "loss": 0.9428,
      "step": 45306
    },
    {
      "epoch": 3.8980469758237977,
      "grad_norm": 2.703125,
      "learning_rate": 0.00011522989332350808,
      "loss": 1.1115,
      "step": 45307
    },
    {
      "epoch": 3.8981330121311193,
      "grad_norm": 1.390625,
      "learning_rate": 0.00011521262571695779,
      "loss": 0.6141,
      "step": 45308
    },
    {
      "epoch": 3.898219048438441,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00011519535923583568,
      "loss": 0.9553,
      "step": 45309
    },
    {
      "epoch": 3.898305084745763,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011517809388019245,
      "loss": 0.661,
      "step": 45310
    },
    {
      "epoch": 3.898391121053084,
      "grad_norm": 0.921875,
      "learning_rate": 0.00011516082965007824,
      "loss": 1.4922,
      "step": 45311
    },
    {
      "epoch": 3.8984771573604062,
      "grad_norm": 5.25,
      "learning_rate": 0.00011514356654554398,
      "loss": 0.8512,
      "step": 45312
    },
    {
      "epoch": 3.898563193667728,
      "grad_norm": 22.0,
      "learning_rate": 0.00011512630456663975,
      "loss": 1.3086,
      "step": 45313
    },
    {
      "epoch": 3.8986492299750495,
      "grad_norm": 157.0,
      "learning_rate": 0.0001151090437134163,
      "loss": 0.6763,
      "step": 45314
    },
    {
      "epoch": 3.898735266282371,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00011509178398592418,
      "loss": 0.7677,
      "step": 45315
    },
    {
      "epoch": 3.8988213025896927,
      "grad_norm": 1.59375,
      "learning_rate": 0.00011507452538421354,
      "loss": 1.6098,
      "step": 45316
    },
    {
      "epoch": 3.898907338897015,
      "grad_norm": 27.5,
      "learning_rate": 0.00011505726790833521,
      "loss": 1.3781,
      "step": 45317
    },
    {
      "epoch": 3.898993375204336,
      "grad_norm": 3.125,
      "learning_rate": 0.00011504001155833938,
      "loss": 1.4281,
      "step": 45318
    },
    {
      "epoch": 3.899079411511658,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011502275633427662,
      "loss": 1.3854,
      "step": 45319
    },
    {
      "epoch": 3.8991654478189797,
      "grad_norm": 2.453125,
      "learning_rate": 0.0001150055022361976,
      "loss": 1.1149,
      "step": 45320
    },
    {
      "epoch": 3.8992514841263013,
      "grad_norm": 4.4375,
      "learning_rate": 0.00011498824926415247,
      "loss": 1.0247,
      "step": 45321
    },
    {
      "epoch": 3.899337520433623,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00011497099741819179,
      "loss": 0.8977,
      "step": 45322
    },
    {
      "epoch": 3.8994235567409445,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011495374669836622,
      "loss": 1.2978,
      "step": 45323
    },
    {
      "epoch": 3.8995095930482666,
      "grad_norm": 21.75,
      "learning_rate": 0.00011493649710472587,
      "loss": 0.8356,
      "step": 45324
    },
    {
      "epoch": 3.899595629355588,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00011491924863732145,
      "loss": 1.4758,
      "step": 45325
    },
    {
      "epoch": 3.89968166566291,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011490200129620343,
      "loss": 0.6956,
      "step": 45326
    },
    {
      "epoch": 3.8997677019702315,
      "grad_norm": 1.359375,
      "learning_rate": 0.00011488475508142204,
      "loss": 1.2173,
      "step": 45327
    },
    {
      "epoch": 3.899853738277553,
      "grad_norm": 1.1875,
      "learning_rate": 0.00011486750999302798,
      "loss": 1.0989,
      "step": 45328
    },
    {
      "epoch": 3.8999397745848747,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011485026603107134,
      "loss": 0.6585,
      "step": 45329
    },
    {
      "epoch": 3.9000258108921964,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00011483302319560284,
      "loss": 0.9683,
      "step": 45330
    },
    {
      "epoch": 3.9001118471995184,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00011481578148667299,
      "loss": 1.3572,
      "step": 45331
    },
    {
      "epoch": 3.90019788350684,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001147985409043319,
      "loss": 0.9404,
      "step": 45332
    },
    {
      "epoch": 3.9002839198141617,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011478130144863013,
      "loss": 0.8105,
      "step": 45333
    },
    {
      "epoch": 3.9003699561214833,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00011476406311961828,
      "loss": 0.6564,
      "step": 45334
    },
    {
      "epoch": 3.900455992428805,
      "grad_norm": 1.328125,
      "learning_rate": 0.00011474682591734653,
      "loss": 0.72,
      "step": 45335
    },
    {
      "epoch": 3.9005420287361265,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00011472958984186532,
      "loss": 0.6764,
      "step": 45336
    },
    {
      "epoch": 3.900628065043448,
      "grad_norm": 5.34375,
      "learning_rate": 0.00011471235489322534,
      "loss": 1.0607,
      "step": 45337
    },
    {
      "epoch": 3.9007141013507702,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011469512107147661,
      "loss": 0.701,
      "step": 45338
    },
    {
      "epoch": 3.900800137658092,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001146778883766697,
      "loss": 0.8031,
      "step": 45339
    },
    {
      "epoch": 3.9008861739654135,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011466065680885518,
      "loss": 0.9042,
      "step": 45340
    },
    {
      "epoch": 3.900972210272735,
      "grad_norm": 33.25,
      "learning_rate": 0.00011464342636808317,
      "loss": 1.3105,
      "step": 45341
    },
    {
      "epoch": 3.9010582465800567,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001146261970544043,
      "loss": 0.8919,
      "step": 45342
    },
    {
      "epoch": 3.9011442828873784,
      "grad_norm": 15.5625,
      "learning_rate": 0.0001146089688678687,
      "loss": 1.0038,
      "step": 45343
    },
    {
      "epoch": 3.9012303191947,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011459174180852693,
      "loss": 1.6113,
      "step": 45344
    },
    {
      "epoch": 3.901316355502022,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00011457451587642952,
      "loss": 0.9195,
      "step": 45345
    },
    {
      "epoch": 3.9014023918093437,
      "grad_norm": 1.140625,
      "learning_rate": 0.00011455729107162654,
      "loss": 1.0114,
      "step": 45346
    },
    {
      "epoch": 3.9014884281166653,
      "grad_norm": 1.6875,
      "learning_rate": 0.0001145400673941685,
      "loss": 1.5858,
      "step": 45347
    },
    {
      "epoch": 3.901574464423987,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001145228448441058,
      "loss": 1.6541,
      "step": 45348
    },
    {
      "epoch": 3.9016605007313085,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00011450562342148885,
      "loss": 0.5774,
      "step": 45349
    },
    {
      "epoch": 3.90174653703863,
      "grad_norm": 1.125,
      "learning_rate": 0.00011448840312636811,
      "loss": 0.8332,
      "step": 45350
    },
    {
      "epoch": 3.901832573345952,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011447118395879363,
      "loss": 0.8986,
      "step": 45351
    },
    {
      "epoch": 3.901918609653274,
      "grad_norm": 78.5,
      "learning_rate": 0.00011445396591881596,
      "loss": 1.0338,
      "step": 45352
    },
    {
      "epoch": 3.9020046459605955,
      "grad_norm": 1.21875,
      "learning_rate": 0.00011443674900648559,
      "loss": 0.8538,
      "step": 45353
    },
    {
      "epoch": 3.902090682267917,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011441953322185261,
      "loss": 1.0972,
      "step": 45354
    },
    {
      "epoch": 3.9021767185752387,
      "grad_norm": 2.09375,
      "learning_rate": 0.00011440231856496752,
      "loss": 0.9051,
      "step": 45355
    },
    {
      "epoch": 3.9022627548825604,
      "grad_norm": 1.125,
      "learning_rate": 0.00011438510503588079,
      "loss": 1.5716,
      "step": 45356
    },
    {
      "epoch": 3.902348791189882,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00011436789263464248,
      "loss": 1.2782,
      "step": 45357
    },
    {
      "epoch": 3.9024348274972036,
      "grad_norm": 3.640625,
      "learning_rate": 0.00011435068136130328,
      "loss": 1.836,
      "step": 45358
    },
    {
      "epoch": 3.9025208638045257,
      "grad_norm": 1.46875,
      "learning_rate": 0.00011433347121591308,
      "loss": 1.2898,
      "step": 45359
    },
    {
      "epoch": 3.9026069001118473,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011431626219852253,
      "loss": 1.3974,
      "step": 45360
    },
    {
      "epoch": 3.902692936419169,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00011429905430918208,
      "loss": 1.9437,
      "step": 45361
    },
    {
      "epoch": 3.9027789727264905,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011428184754794169,
      "loss": 0.699,
      "step": 45362
    },
    {
      "epoch": 3.902865009033812,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00011426464191485192,
      "loss": 1.169,
      "step": 45363
    },
    {
      "epoch": 3.902951045341134,
      "grad_norm": 20.875,
      "learning_rate": 0.00011424743740996313,
      "loss": 1.2937,
      "step": 45364
    },
    {
      "epoch": 3.9030370816484554,
      "grad_norm": 852.0,
      "learning_rate": 0.00011423023403332544,
      "loss": 0.9252,
      "step": 45365
    },
    {
      "epoch": 3.9031231179557775,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011421303178498931,
      "loss": 1.0235,
      "step": 45366
    },
    {
      "epoch": 3.903209154263099,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011419583066500516,
      "loss": 0.6893,
      "step": 45367
    },
    {
      "epoch": 3.9032951905704207,
      "grad_norm": 1.03125,
      "learning_rate": 0.00011417863067342305,
      "loss": 0.6614,
      "step": 45368
    },
    {
      "epoch": 3.9033812268777424,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00011416143181029354,
      "loss": 0.8666,
      "step": 45369
    },
    {
      "epoch": 3.903467263185064,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00011414423407566666,
      "loss": 0.6078,
      "step": 45370
    },
    {
      "epoch": 3.903553299492386,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001141270374695928,
      "loss": 1.4897,
      "step": 45371
    },
    {
      "epoch": 3.9036393357997072,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011410984199212248,
      "loss": 1.9684,
      "step": 45372
    },
    {
      "epoch": 3.9037253721070293,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011409264764330573,
      "loss": 0.5108,
      "step": 45373
    },
    {
      "epoch": 3.903811408414351,
      "grad_norm": 1.671875,
      "learning_rate": 0.00011407545442319289,
      "loss": 1.1607,
      "step": 45374
    },
    {
      "epoch": 3.9038974447216725,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00011405826233183442,
      "loss": 0.9559,
      "step": 45375
    },
    {
      "epoch": 3.903983481028994,
      "grad_norm": 1.234375,
      "learning_rate": 0.00011404107136928032,
      "loss": 1.6809,
      "step": 45376
    },
    {
      "epoch": 3.904069517336316,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011402388153558108,
      "loss": 0.7651,
      "step": 45377
    },
    {
      "epoch": 3.904155553643638,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011400669283078696,
      "loss": 0.929,
      "step": 45378
    },
    {
      "epoch": 3.904241589950959,
      "grad_norm": 2.78125,
      "learning_rate": 0.0001139895052549481,
      "loss": 0.497,
      "step": 45379
    },
    {
      "epoch": 3.904327626258281,
      "grad_norm": 205.0,
      "learning_rate": 0.00011397231880811493,
      "loss": 1.3573,
      "step": 45380
    },
    {
      "epoch": 3.9044136625656027,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011395513349033771,
      "loss": 1.0996,
      "step": 45381
    },
    {
      "epoch": 3.9044996988729244,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011393794930166651,
      "loss": 0.8583,
      "step": 45382
    },
    {
      "epoch": 3.904585735180246,
      "grad_norm": 39.75,
      "learning_rate": 0.00011392076624215186,
      "loss": 1.0021,
      "step": 45383
    },
    {
      "epoch": 3.9046717714875676,
      "grad_norm": 2912.0,
      "learning_rate": 0.00011390358431184378,
      "loss": 1.8056,
      "step": 45384
    },
    {
      "epoch": 3.9047578077948897,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001138864035107926,
      "loss": 1.651,
      "step": 45385
    },
    {
      "epoch": 3.904843844102211,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00011386922383904863,
      "loss": 0.8741,
      "step": 45386
    },
    {
      "epoch": 3.904929880409533,
      "grad_norm": 1.0625,
      "learning_rate": 0.00011385204529666204,
      "loss": 1.8697,
      "step": 45387
    },
    {
      "epoch": 3.9050159167168546,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011383486788368325,
      "loss": 0.5176,
      "step": 45388
    },
    {
      "epoch": 3.905101953024176,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00011381769160016226,
      "loss": 1.7802,
      "step": 45389
    },
    {
      "epoch": 3.905187989331498,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011380051644614936,
      "loss": 0.8907,
      "step": 45390
    },
    {
      "epoch": 3.9052740256388194,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00011378334242169502,
      "loss": 0.8506,
      "step": 45391
    },
    {
      "epoch": 3.9053600619461415,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011376616952684915,
      "loss": 1.3924,
      "step": 45392
    },
    {
      "epoch": 3.905446098253463,
      "grad_norm": 1.0625,
      "learning_rate": 0.00011374899776166209,
      "loss": 1.2485,
      "step": 45393
    },
    {
      "epoch": 3.9055321345607847,
      "grad_norm": 2.265625,
      "learning_rate": 0.00011373182712618423,
      "loss": 1.0822,
      "step": 45394
    },
    {
      "epoch": 3.9056181708681064,
      "grad_norm": 5216.0,
      "learning_rate": 0.00011371465762046557,
      "loss": 0.7325,
      "step": 45395
    },
    {
      "epoch": 3.905704207175428,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00011369748924455636,
      "loss": 0.9642,
      "step": 45396
    },
    {
      "epoch": 3.9057902434827496,
      "grad_norm": 47.5,
      "learning_rate": 0.00011368032199850704,
      "loss": 0.7106,
      "step": 45397
    },
    {
      "epoch": 3.9058762797900712,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011366315588236742,
      "loss": 0.9825,
      "step": 45398
    },
    {
      "epoch": 3.9059623160973933,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011364599089618815,
      "loss": 1.7728,
      "step": 45399
    },
    {
      "epoch": 3.906048352404715,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011362882704001903,
      "loss": 1.2067,
      "step": 45400
    },
    {
      "epoch": 3.9061343887120366,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00011361166431391046,
      "loss": 0.5542,
      "step": 45401
    },
    {
      "epoch": 3.906220425019358,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011359450271791277,
      "loss": 1.5924,
      "step": 45402
    },
    {
      "epoch": 3.90630646132668,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011357734225207588,
      "loss": 1.0688,
      "step": 45403
    },
    {
      "epoch": 3.9063924976340014,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001135601829164501,
      "loss": 0.708,
      "step": 45404
    },
    {
      "epoch": 3.906478533941323,
      "grad_norm": 1.078125,
      "learning_rate": 0.00011354302471108574,
      "loss": 1.1395,
      "step": 45405
    },
    {
      "epoch": 3.906564570248645,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00011352586763603273,
      "loss": 2.0124,
      "step": 45406
    },
    {
      "epoch": 3.9066506065559667,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00011350871169134141,
      "loss": 0.9868,
      "step": 45407
    },
    {
      "epoch": 3.9067366428632884,
      "grad_norm": 1.640625,
      "learning_rate": 0.00011349155687706208,
      "loss": 0.8604,
      "step": 45408
    },
    {
      "epoch": 3.90682267917061,
      "grad_norm": 1.515625,
      "learning_rate": 0.00011347440319324465,
      "loss": 1.0019,
      "step": 45409
    },
    {
      "epoch": 3.9069087154779316,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00011345725063993956,
      "loss": 1.4837,
      "step": 45410
    },
    {
      "epoch": 3.9069947517852532,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00011344009921719667,
      "loss": 0.5098,
      "step": 45411
    },
    {
      "epoch": 3.907080788092575,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011342294892506633,
      "loss": 1.0649,
      "step": 45412
    },
    {
      "epoch": 3.907166824399897,
      "grad_norm": 272.0,
      "learning_rate": 0.00011340579976359877,
      "loss": 0.9027,
      "step": 45413
    },
    {
      "epoch": 3.9072528607072186,
      "grad_norm": 0.90625,
      "learning_rate": 0.00011338865173284397,
      "loss": 0.6352,
      "step": 45414
    },
    {
      "epoch": 3.90733889701454,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00011337150483285214,
      "loss": 0.8548,
      "step": 45415
    },
    {
      "epoch": 3.907424933321862,
      "grad_norm": 2.390625,
      "learning_rate": 0.00011335435906367364,
      "loss": 1.6521,
      "step": 45416
    },
    {
      "epoch": 3.9075109696291834,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011333721442535828,
      "loss": 1.0585,
      "step": 45417
    },
    {
      "epoch": 3.907597005936505,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00011332007091795631,
      "loss": 1.1154,
      "step": 45418
    },
    {
      "epoch": 3.9076830422438267,
      "grad_norm": 1.546875,
      "learning_rate": 0.00011330292854151814,
      "loss": 1.5513,
      "step": 45419
    },
    {
      "epoch": 3.9077690785511487,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00011328578729609356,
      "loss": 0.9234,
      "step": 45420
    },
    {
      "epoch": 3.9078551148584704,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011326864718173279,
      "loss": 0.7208,
      "step": 45421
    },
    {
      "epoch": 3.907941151165792,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011325150819848606,
      "loss": 0.7996,
      "step": 45422
    },
    {
      "epoch": 3.9080271874731136,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011323437034640344,
      "loss": 0.9256,
      "step": 45423
    },
    {
      "epoch": 3.9081132237804352,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00011321723362553516,
      "loss": 1.0091,
      "step": 45424
    },
    {
      "epoch": 3.9081992600877573,
      "grad_norm": 1.34375,
      "learning_rate": 0.00011320009803593118,
      "loss": 1.3811,
      "step": 45425
    },
    {
      "epoch": 3.9082852963950785,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00011318296357764163,
      "loss": 0.5232,
      "step": 45426
    },
    {
      "epoch": 3.9083713327024006,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001131658302507168,
      "loss": 1.1719,
      "step": 45427
    },
    {
      "epoch": 3.908457369009722,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011314869805520661,
      "loss": 0.9223,
      "step": 45428
    },
    {
      "epoch": 3.908543405317044,
      "grad_norm": 0.953125,
      "learning_rate": 0.00011313156699116134,
      "loss": 1.4205,
      "step": 45429
    },
    {
      "epoch": 3.9086294416243654,
      "grad_norm": 1.265625,
      "learning_rate": 0.00011311443705863084,
      "loss": 1.0038,
      "step": 45430
    },
    {
      "epoch": 3.908715477931687,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001130973082576654,
      "loss": 1.209,
      "step": 45431
    },
    {
      "epoch": 3.908801514239009,
      "grad_norm": 4.15625,
      "learning_rate": 0.00011308018058831521,
      "loss": 0.5483,
      "step": 45432
    },
    {
      "epoch": 3.9088875505463303,
      "grad_norm": 2.234375,
      "learning_rate": 0.00011306305405063011,
      "loss": 0.76,
      "step": 45433
    },
    {
      "epoch": 3.9089735868536524,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011304592864466034,
      "loss": 1.3702,
      "step": 45434
    },
    {
      "epoch": 3.909059623160974,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011302880437045604,
      "loss": 1.3348,
      "step": 45435
    },
    {
      "epoch": 3.9091456594682956,
      "grad_norm": 1.890625,
      "learning_rate": 0.00011301168122806715,
      "loss": 1.7279,
      "step": 45436
    },
    {
      "epoch": 3.9092316957756172,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011299455921754381,
      "loss": 1.038,
      "step": 45437
    },
    {
      "epoch": 3.909317732082939,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011297743833893626,
      "loss": 1.9544,
      "step": 45438
    },
    {
      "epoch": 3.909403768390261,
      "grad_norm": 1.171875,
      "learning_rate": 0.00011296031859229427,
      "loss": 0.961,
      "step": 45439
    },
    {
      "epoch": 3.909489804697582,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011294319997766822,
      "loss": 0.731,
      "step": 45440
    },
    {
      "epoch": 3.909575841004904,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011292608249510794,
      "loss": 1.0673,
      "step": 45441
    },
    {
      "epoch": 3.909661877312226,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011290896614466351,
      "loss": 1.0846,
      "step": 45442
    },
    {
      "epoch": 3.9097479136195474,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00011289185092638526,
      "loss": 1.2687,
      "step": 45443
    },
    {
      "epoch": 3.909833949926869,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001128747368403229,
      "loss": 1.1931,
      "step": 45444
    },
    {
      "epoch": 3.9099199862341907,
      "grad_norm": 2.453125,
      "learning_rate": 0.00011285762388652665,
      "loss": 1.2892,
      "step": 45445
    },
    {
      "epoch": 3.9100060225415127,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001128405120650467,
      "loss": 0.9118,
      "step": 45446
    },
    {
      "epoch": 3.9100920588488344,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00011282340137593278,
      "loss": 1.278,
      "step": 45447
    },
    {
      "epoch": 3.910178095156156,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011280629181923513,
      "loss": 1.2853,
      "step": 45448
    },
    {
      "epoch": 3.9102641314634776,
      "grad_norm": 1.140625,
      "learning_rate": 0.00011278918339500393,
      "loss": 0.9864,
      "step": 45449
    },
    {
      "epoch": 3.9103501677707992,
      "grad_norm": 426.0,
      "learning_rate": 0.00011277207610328888,
      "loss": 0.803,
      "step": 45450
    },
    {
      "epoch": 3.910436204078121,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00011275496994414036,
      "loss": 0.9721,
      "step": 45451
    },
    {
      "epoch": 3.9105222403854425,
      "grad_norm": 3.9375,
      "learning_rate": 0.00011273786491760813,
      "loss": 1.226,
      "step": 45452
    },
    {
      "epoch": 3.9106082766927646,
      "grad_norm": 306.0,
      "learning_rate": 0.00011272076102374229,
      "loss": 1.2782,
      "step": 45453
    },
    {
      "epoch": 3.910694313000086,
      "grad_norm": 1288.0,
      "learning_rate": 0.00011270365826259304,
      "loss": 1.3051,
      "step": 45454
    },
    {
      "epoch": 3.910780349307408,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011268655663421012,
      "loss": 1.1234,
      "step": 45455
    },
    {
      "epoch": 3.9108663856147294,
      "grad_norm": 1.140625,
      "learning_rate": 0.00011266945613864372,
      "loss": 1.1968,
      "step": 45456
    },
    {
      "epoch": 3.910952421922051,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00011265235677594398,
      "loss": 0.7564,
      "step": 45457
    },
    {
      "epoch": 3.9110384582293727,
      "grad_norm": 77.0,
      "learning_rate": 0.00011263525854616058,
      "loss": 1.6913,
      "step": 45458
    },
    {
      "epoch": 3.9111244945366943,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00011261816144934372,
      "loss": 1.0556,
      "step": 45459
    },
    {
      "epoch": 3.9112105308440164,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011260106548554344,
      "loss": 1.1231,
      "step": 45460
    },
    {
      "epoch": 3.911296567151338,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011258397065480964,
      "loss": 1.0083,
      "step": 45461
    },
    {
      "epoch": 3.9113826034586596,
      "grad_norm": 1.1875,
      "learning_rate": 0.00011256687695719254,
      "loss": 1.6014,
      "step": 45462
    },
    {
      "epoch": 3.9114686397659812,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011254978439274182,
      "loss": 1.152,
      "step": 45463
    },
    {
      "epoch": 3.911554676073303,
      "grad_norm": 3.640625,
      "learning_rate": 0.00011253269296150765,
      "loss": 1.0972,
      "step": 45464
    },
    {
      "epoch": 3.9116407123806245,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001125156026635401,
      "loss": 0.9742,
      "step": 45465
    },
    {
      "epoch": 3.911726748687946,
      "grad_norm": 1.453125,
      "learning_rate": 0.00011249851349888895,
      "loss": 1.436,
      "step": 45466
    },
    {
      "epoch": 3.911812784995268,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00011248142546760426,
      "loss": 1.5403,
      "step": 45467
    },
    {
      "epoch": 3.91189882130259,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001124643385697362,
      "loss": 1.0832,
      "step": 45468
    },
    {
      "epoch": 3.9119848576099114,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00011244725280533441,
      "loss": 1.5525,
      "step": 45469
    },
    {
      "epoch": 3.912070893917233,
      "grad_norm": 21.5,
      "learning_rate": 0.00011243016817444917,
      "loss": 1.5891,
      "step": 45470
    },
    {
      "epoch": 3.9121569302245547,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011241308467713019,
      "loss": 0.7488,
      "step": 45471
    },
    {
      "epoch": 3.9122429665318763,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011239600231342757,
      "loss": 0.8525,
      "step": 45472
    },
    {
      "epoch": 3.912329002839198,
      "grad_norm": 115.0,
      "learning_rate": 0.00011237892108339138,
      "loss": 0.8239,
      "step": 45473
    },
    {
      "epoch": 3.91241503914652,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001123618409870713,
      "loss": 1.4335,
      "step": 45474
    },
    {
      "epoch": 3.9125010754538416,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011234476202451743,
      "loss": 0.6586,
      "step": 45475
    },
    {
      "epoch": 3.9125871117611632,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011232768419577993,
      "loss": 1.5244,
      "step": 45476
    },
    {
      "epoch": 3.912673148068485,
      "grad_norm": 12032.0,
      "learning_rate": 0.00011231060750090838,
      "loss": 1.1463,
      "step": 45477
    },
    {
      "epoch": 3.9127591843758065,
      "grad_norm": 1.5625,
      "learning_rate": 0.00011229353193995289,
      "loss": 1.6046,
      "step": 45478
    },
    {
      "epoch": 3.912845220683128,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011227645751296356,
      "loss": 1.4847,
      "step": 45479
    },
    {
      "epoch": 3.9129312569904497,
      "grad_norm": 9.625,
      "learning_rate": 0.00011225938421999005,
      "loss": 1.3202,
      "step": 45480
    },
    {
      "epoch": 3.913017293297772,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001122423120610826,
      "loss": 1.4013,
      "step": 45481
    },
    {
      "epoch": 3.9131033296050934,
      "grad_norm": 316.0,
      "learning_rate": 0.00011222524103629078,
      "loss": 1.125,
      "step": 45482
    },
    {
      "epoch": 3.913189365912415,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011220817114566479,
      "loss": 0.7154,
      "step": 45483
    },
    {
      "epoch": 3.9132754022197367,
      "grad_norm": 1.296875,
      "learning_rate": 0.00011219110238925456,
      "loss": 0.883,
      "step": 45484
    },
    {
      "epoch": 3.9133614385270583,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001121740347671098,
      "loss": 1.0461,
      "step": 45485
    },
    {
      "epoch": 3.9134474748343804,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011215696827928057,
      "loss": 1.4287,
      "step": 45486
    },
    {
      "epoch": 3.9135335111417016,
      "grad_norm": 5.4375,
      "learning_rate": 0.00011213990292581688,
      "loss": 1.8204,
      "step": 45487
    },
    {
      "epoch": 3.9136195474490236,
      "grad_norm": 1.0625,
      "learning_rate": 0.00011212283870676843,
      "loss": 1.2549,
      "step": 45488
    },
    {
      "epoch": 3.9137055837563453,
      "grad_norm": 0.96875,
      "learning_rate": 0.00011210577562218521,
      "loss": 0.7689,
      "step": 45489
    },
    {
      "epoch": 3.913791620063667,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00011208871367211738,
      "loss": 0.825,
      "step": 45490
    },
    {
      "epoch": 3.9138776563709885,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00011207165285661436,
      "loss": 1.2529,
      "step": 45491
    },
    {
      "epoch": 3.91396369267831,
      "grad_norm": 1.4375,
      "learning_rate": 0.00011205459317572636,
      "loss": 1.1491,
      "step": 45492
    },
    {
      "epoch": 3.914049728985632,
      "grad_norm": 1.640625,
      "learning_rate": 0.00011203753462950339,
      "loss": 0.738,
      "step": 45493
    },
    {
      "epoch": 3.9141357652929534,
      "grad_norm": 3.578125,
      "learning_rate": 0.00011202047721799502,
      "loss": 1.1186,
      "step": 45494
    },
    {
      "epoch": 3.9142218016002754,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011200342094125144,
      "loss": 1.1763,
      "step": 45495
    },
    {
      "epoch": 3.914307837907597,
      "grad_norm": 1.75,
      "learning_rate": 0.00011198636579932209,
      "loss": 0.7567,
      "step": 45496
    },
    {
      "epoch": 3.9143938742149187,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011196931179225744,
      "loss": 0.9113,
      "step": 45497
    },
    {
      "epoch": 3.9144799105222403,
      "grad_norm": 9.5625,
      "learning_rate": 0.00011195225892010697,
      "loss": 1.3746,
      "step": 45498
    },
    {
      "epoch": 3.914565946829562,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00011193520718292066,
      "loss": 1.0261,
      "step": 45499
    },
    {
      "epoch": 3.914651983136884,
      "grad_norm": 45.5,
      "learning_rate": 0.00011191815658074855,
      "loss": 0.6754,
      "step": 45500
    },
    {
      "epoch": 3.914738019444205,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011190110711364015,
      "loss": 0.7705,
      "step": 45501
    },
    {
      "epoch": 3.9148240557515273,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001118840587816456,
      "loss": 1.156,
      "step": 45502
    },
    {
      "epoch": 3.914910092058849,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00011186701158481483,
      "loss": 1.4905,
      "step": 45503
    },
    {
      "epoch": 3.9149961283661705,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011184996552319737,
      "loss": 1.5827,
      "step": 45504
    },
    {
      "epoch": 3.915082164673492,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001118329205968433,
      "loss": 1.3406,
      "step": 45505
    },
    {
      "epoch": 3.9151682009808138,
      "grad_norm": 2672.0,
      "learning_rate": 0.00011181587680580263,
      "loss": 1.0664,
      "step": 45506
    },
    {
      "epoch": 3.915254237288136,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00011179883415012482,
      "loss": 1.1092,
      "step": 45507
    },
    {
      "epoch": 3.9153402735954574,
      "grad_norm": 4.90625,
      "learning_rate": 0.00011178179262985993,
      "loss": 1.0874,
      "step": 45508
    },
    {
      "epoch": 3.915426309902779,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011176475224505795,
      "loss": 0.5603,
      "step": 45509
    },
    {
      "epoch": 3.9155123462101007,
      "grad_norm": 282.0,
      "learning_rate": 0.00011174771299576841,
      "loss": 0.8788,
      "step": 45510
    },
    {
      "epoch": 3.9155983825174223,
      "grad_norm": 4.09375,
      "learning_rate": 0.00011173067488204142,
      "loss": 1.0551,
      "step": 45511
    },
    {
      "epoch": 3.915684418824744,
      "grad_norm": 1.109375,
      "learning_rate": 0.00011171363790392657,
      "loss": 0.6796,
      "step": 45512
    },
    {
      "epoch": 3.9157704551320656,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001116966020614738,
      "loss": 0.6981,
      "step": 45513
    },
    {
      "epoch": 3.9158564914393876,
      "grad_norm": 42.25,
      "learning_rate": 0.00011167956735473311,
      "loss": 1.0351,
      "step": 45514
    },
    {
      "epoch": 3.9159425277467093,
      "grad_norm": 1.4375,
      "learning_rate": 0.00011166253378375402,
      "loss": 1.5689,
      "step": 45515
    },
    {
      "epoch": 3.916028564054031,
      "grad_norm": 2.703125,
      "learning_rate": 0.00011164550134858647,
      "loss": 2.0164,
      "step": 45516
    },
    {
      "epoch": 3.9161146003613525,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001116284700492805,
      "loss": 1.5818,
      "step": 45517
    },
    {
      "epoch": 3.916200636668674,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00011161143988588552,
      "loss": 0.926,
      "step": 45518
    },
    {
      "epoch": 3.9162866729759958,
      "grad_norm": 0.96875,
      "learning_rate": 0.00011159441085845157,
      "loss": 0.9232,
      "step": 45519
    },
    {
      "epoch": 3.9163727092833174,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00011157738296702852,
      "loss": 0.7892,
      "step": 45520
    },
    {
      "epoch": 3.9164587455906394,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000111560356211666,
      "loss": 1.6506,
      "step": 45521
    },
    {
      "epoch": 3.916544781897961,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011154333059241401,
      "loss": 1.4087,
      "step": 45522
    },
    {
      "epoch": 3.9166308182052827,
      "grad_norm": 1.234375,
      "learning_rate": 0.00011152630610932207,
      "loss": 1.0035,
      "step": 45523
    },
    {
      "epoch": 3.9167168545126043,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011150928276244015,
      "loss": 1.6334,
      "step": 45524
    },
    {
      "epoch": 3.916802890819926,
      "grad_norm": 2.984375,
      "learning_rate": 0.00011149226055181815,
      "loss": 1.0252,
      "step": 45525
    },
    {
      "epoch": 3.9168889271272476,
      "grad_norm": 472.0,
      "learning_rate": 0.0001114752394775056,
      "loss": 1.3882,
      "step": 45526
    },
    {
      "epoch": 3.916974963434569,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00011145821953955238,
      "loss": 1.1179,
      "step": 45527
    },
    {
      "epoch": 3.9170609997418913,
      "grad_norm": 22.625,
      "learning_rate": 0.00011144120073800846,
      "loss": 1.0205,
      "step": 45528
    },
    {
      "epoch": 3.917147036049213,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00011142418307292334,
      "loss": 1.4081,
      "step": 45529
    },
    {
      "epoch": 3.9172330723565345,
      "grad_norm": 1.03125,
      "learning_rate": 0.00011140716654434685,
      "loss": 0.9883,
      "step": 45530
    },
    {
      "epoch": 3.917319108663856,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011139015115232903,
      "loss": 0.7053,
      "step": 45531
    },
    {
      "epoch": 3.9174051449711778,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00011137313689691925,
      "loss": 1.8522,
      "step": 45532
    },
    {
      "epoch": 3.9174911812784994,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011135612377816751,
      "loss": 1.2384,
      "step": 45533
    },
    {
      "epoch": 3.917577217585821,
      "grad_norm": 0.96875,
      "learning_rate": 0.00011133911179612344,
      "loss": 0.9262,
      "step": 45534
    },
    {
      "epoch": 3.917663253893143,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00011132210095083695,
      "loss": 0.7648,
      "step": 45535
    },
    {
      "epoch": 3.9177492902004647,
      "grad_norm": 3.28125,
      "learning_rate": 0.00011130509124235783,
      "loss": 1.3183,
      "step": 45536
    },
    {
      "epoch": 3.9178353265077863,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001112880826707356,
      "loss": 1.1873,
      "step": 45537
    },
    {
      "epoch": 3.917921362815108,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011127107523602009,
      "loss": 1.2848,
      "step": 45538
    },
    {
      "epoch": 3.9180073991224296,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011125406893826128,
      "loss": 0.9224,
      "step": 45539
    },
    {
      "epoch": 3.918093435429751,
      "grad_norm": 1.296875,
      "learning_rate": 0.00011123706377750854,
      "loss": 1.1032,
      "step": 45540
    },
    {
      "epoch": 3.918179471737073,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011122005975381189,
      "loss": 0.6165,
      "step": 45541
    },
    {
      "epoch": 3.918265508044395,
      "grad_norm": 1.296875,
      "learning_rate": 0.00011120305686722088,
      "loss": 0.7196,
      "step": 45542
    },
    {
      "epoch": 3.9183515443517165,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00011118605511778523,
      "loss": 0.874,
      "step": 45543
    },
    {
      "epoch": 3.918437580659038,
      "grad_norm": 30.375,
      "learning_rate": 0.00011116905450555498,
      "loss": 0.5507,
      "step": 45544
    },
    {
      "epoch": 3.9185236169663598,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001111520550305794,
      "loss": 1.377,
      "step": 45545
    },
    {
      "epoch": 3.9186096532736814,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00011113505669290846,
      "loss": 1.061,
      "step": 45546
    },
    {
      "epoch": 3.9186956895810034,
      "grad_norm": 1.0625,
      "learning_rate": 0.00011111805949259202,
      "loss": 1.3499,
      "step": 45547
    },
    {
      "epoch": 3.9187817258883246,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011110106342967946,
      "loss": 0.9872,
      "step": 45548
    },
    {
      "epoch": 3.9188677621956467,
      "grad_norm": 1.234375,
      "learning_rate": 0.00011108406850422064,
      "loss": 1.4229,
      "step": 45549
    },
    {
      "epoch": 3.9189537985029683,
      "grad_norm": 76.0,
      "learning_rate": 0.00011106707471626543,
      "loss": 1.155,
      "step": 45550
    },
    {
      "epoch": 3.91903983481029,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00011105008206586326,
      "loss": 1.7368,
      "step": 45551
    },
    {
      "epoch": 3.9191258711176116,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011103309055306404,
      "loss": 1.0921,
      "step": 45552
    },
    {
      "epoch": 3.919211907424933,
      "grad_norm": 32.75,
      "learning_rate": 0.0001110161001779173,
      "loss": 0.4839,
      "step": 45553
    },
    {
      "epoch": 3.9192979437322553,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011099911094047276,
      "loss": 1.3157,
      "step": 45554
    },
    {
      "epoch": 3.9193839800395764,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011098212284078036,
      "loss": 1.257,
      "step": 45555
    },
    {
      "epoch": 3.9194700163468985,
      "grad_norm": 1.125,
      "learning_rate": 0.0001109651358788894,
      "loss": 1.4458,
      "step": 45556
    },
    {
      "epoch": 3.91955605265422,
      "grad_norm": 1.765625,
      "learning_rate": 0.00011094815005484977,
      "loss": 0.9392,
      "step": 45557
    },
    {
      "epoch": 3.9196420889615418,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011093116536871129,
      "loss": 1.1041,
      "step": 45558
    },
    {
      "epoch": 3.9197281252688634,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00011091418182052332,
      "loss": 1.2673,
      "step": 45559
    },
    {
      "epoch": 3.919814161576185,
      "grad_norm": 2.1875,
      "learning_rate": 0.00011089719941033572,
      "loss": 0.5548,
      "step": 45560
    },
    {
      "epoch": 3.919900197883507,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011088021813819827,
      "loss": 0.4774,
      "step": 45561
    },
    {
      "epoch": 3.9199862341908283,
      "grad_norm": 10.4375,
      "learning_rate": 0.00011086323800416036,
      "loss": 0.5137,
      "step": 45562
    },
    {
      "epoch": 3.9200722704981503,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00011084625900827194,
      "loss": 0.752,
      "step": 45563
    },
    {
      "epoch": 3.920158306805472,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011082928115058238,
      "loss": 1.5256,
      "step": 45564
    },
    {
      "epoch": 3.9202443431127936,
      "grad_norm": 2.1875,
      "learning_rate": 0.00011081230443114149,
      "loss": 1.6378,
      "step": 45565
    },
    {
      "epoch": 3.920330379420115,
      "grad_norm": 13.5,
      "learning_rate": 0.00011079532884999904,
      "loss": 1.1156,
      "step": 45566
    },
    {
      "epoch": 3.920416415727437,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0001107783544072044,
      "loss": 1.6982,
      "step": 45567
    },
    {
      "epoch": 3.920502452034759,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00011076138110280743,
      "loss": 0.9495,
      "step": 45568
    },
    {
      "epoch": 3.9205884883420805,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00011074440893685783,
      "loss": 1.3268,
      "step": 45569
    },
    {
      "epoch": 3.920674524649402,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011072743790940488,
      "loss": 1.524,
      "step": 45570
    },
    {
      "epoch": 3.9207605609567238,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011071046802049866,
      "loss": 1.6293,
      "step": 45571
    },
    {
      "epoch": 3.9208465972640454,
      "grad_norm": 1.734375,
      "learning_rate": 0.00011069349927018857,
      "loss": 1.3104,
      "step": 45572
    },
    {
      "epoch": 3.920932633571367,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00011067653165852421,
      "loss": 1.0783,
      "step": 45573
    },
    {
      "epoch": 3.9210186698786886,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011065956518555542,
      "loss": 0.5873,
      "step": 45574
    },
    {
      "epoch": 3.9211047061860107,
      "grad_norm": 0.921875,
      "learning_rate": 0.00011064259985133157,
      "loss": 1.1214,
      "step": 45575
    },
    {
      "epoch": 3.9211907424933323,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011062563565590234,
      "loss": 0.7488,
      "step": 45576
    },
    {
      "epoch": 3.921276778800654,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011060867259931761,
      "loss": 0.574,
      "step": 45577
    },
    {
      "epoch": 3.9213628151079756,
      "grad_norm": 2.3125,
      "learning_rate": 0.0001105917106816266,
      "loss": 1.2944,
      "step": 45578
    },
    {
      "epoch": 3.921448851415297,
      "grad_norm": 50.75,
      "learning_rate": 0.00011057474990287908,
      "loss": 1.2451,
      "step": 45579
    },
    {
      "epoch": 3.921534887722619,
      "grad_norm": 2.796875,
      "learning_rate": 0.00011055779026312484,
      "loss": 0.9482,
      "step": 45580
    },
    {
      "epoch": 3.9216209240299404,
      "grad_norm": 2.671875,
      "learning_rate": 0.0001105408317624132,
      "loss": 0.9899,
      "step": 45581
    },
    {
      "epoch": 3.9217069603372625,
      "grad_norm": 19.75,
      "learning_rate": 0.00011052387440079398,
      "loss": 0.8625,
      "step": 45582
    },
    {
      "epoch": 3.921792996644584,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00011050691817831654,
      "loss": 1.1943,
      "step": 45583
    },
    {
      "epoch": 3.9218790329519058,
      "grad_norm": 1.03125,
      "learning_rate": 0.00011048996309503063,
      "loss": 1.3837,
      "step": 45584
    },
    {
      "epoch": 3.9219650692592274,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011047300915098596,
      "loss": 0.4964,
      "step": 45585
    },
    {
      "epoch": 3.922051105566549,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011045605634623185,
      "loss": 0.9033,
      "step": 45586
    },
    {
      "epoch": 3.9221371418738706,
      "grad_norm": 1.28125,
      "learning_rate": 0.00011043910468081803,
      "loss": 1.124,
      "step": 45587
    },
    {
      "epoch": 3.9222231781811923,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00011042215415479418,
      "loss": 1.0643,
      "step": 45588
    },
    {
      "epoch": 3.9223092144885143,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011040520476820959,
      "loss": 1.0078,
      "step": 45589
    },
    {
      "epoch": 3.922395250795836,
      "grad_norm": 1.375,
      "learning_rate": 0.00011038825652111406,
      "loss": 1.5248,
      "step": 45590
    },
    {
      "epoch": 3.9224812871031576,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011037130941355723,
      "loss": 0.7711,
      "step": 45591
    },
    {
      "epoch": 3.922567323410479,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00011035436344558836,
      "loss": 0.5682,
      "step": 45592
    },
    {
      "epoch": 3.922653359717801,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011033741861725738,
      "loss": 0.7801,
      "step": 45593
    },
    {
      "epoch": 3.9227393960251224,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00011032047492861353,
      "loss": 1.0281,
      "step": 45594
    },
    {
      "epoch": 3.922825432332444,
      "grad_norm": 1.0,
      "learning_rate": 0.00011030353237970647,
      "loss": 0.9918,
      "step": 45595
    },
    {
      "epoch": 3.922911468639766,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011028659097058591,
      "loss": 1.0669,
      "step": 45596
    },
    {
      "epoch": 3.9229975049470878,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00011026965070130119,
      "loss": 0.6005,
      "step": 45597
    },
    {
      "epoch": 3.9230835412544094,
      "grad_norm": 334.0,
      "learning_rate": 0.00011025271157190192,
      "loss": 1.3132,
      "step": 45598
    },
    {
      "epoch": 3.923169577561731,
      "grad_norm": 44.5,
      "learning_rate": 0.0001102357735824378,
      "loss": 1.4832,
      "step": 45599
    },
    {
      "epoch": 3.9232556138690526,
      "grad_norm": 22.125,
      "learning_rate": 0.00011021883673295807,
      "loss": 1.2112,
      "step": 45600
    },
    {
      "epoch": 3.9233416501763747,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00011020190102351251,
      "loss": 1.7957,
      "step": 45601
    },
    {
      "epoch": 3.923427686483696,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011018496645415066,
      "loss": 2.0578,
      "step": 45602
    },
    {
      "epoch": 3.923513722791018,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011016803302492185,
      "loss": 0.9314,
      "step": 45603
    },
    {
      "epoch": 3.9235997590983396,
      "grad_norm": 1.296875,
      "learning_rate": 0.00011015110073587569,
      "loss": 0.882,
      "step": 45604
    },
    {
      "epoch": 3.923685795405661,
      "grad_norm": 1.046875,
      "learning_rate": 0.00011013416958706191,
      "loss": 0.9962,
      "step": 45605
    },
    {
      "epoch": 3.923771831712983,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001101172395785297,
      "loss": 2.2542,
      "step": 45606
    },
    {
      "epoch": 3.9238578680203045,
      "grad_norm": 3.203125,
      "learning_rate": 0.00011010031071032878,
      "loss": 1.0346,
      "step": 45607
    },
    {
      "epoch": 3.9239439043276265,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00011008338298250859,
      "loss": 1.5987,
      "step": 45608
    },
    {
      "epoch": 3.9240299406349477,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00011006645639511881,
      "loss": 1.0428,
      "step": 45609
    },
    {
      "epoch": 3.9241159769422698,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00011004953094820869,
      "loss": 0.8211,
      "step": 45610
    },
    {
      "epoch": 3.9242020132495914,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00011003260664182784,
      "loss": 2.052,
      "step": 45611
    },
    {
      "epoch": 3.924288049556913,
      "grad_norm": 1.09375,
      "learning_rate": 0.00011001568347602587,
      "loss": 0.5008,
      "step": 45612
    },
    {
      "epoch": 3.9243740858642346,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00010999876145085208,
      "loss": 1.0117,
      "step": 45613
    },
    {
      "epoch": 3.9244601221715563,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010998184056635608,
      "loss": 1.4818,
      "step": 45614
    },
    {
      "epoch": 3.9245461584788783,
      "grad_norm": 1.328125,
      "learning_rate": 0.00010996492082258742,
      "loss": 1.2033,
      "step": 45615
    },
    {
      "epoch": 3.9246321947861995,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00010994800221959538,
      "loss": 0.8451,
      "step": 45616
    },
    {
      "epoch": 3.9247182310935216,
      "grad_norm": 4096.0,
      "learning_rate": 0.00010993108475742963,
      "loss": 1.175,
      "step": 45617
    },
    {
      "epoch": 3.924804267400843,
      "grad_norm": 13.375,
      "learning_rate": 0.00010991416843613972,
      "loss": 1.3218,
      "step": 45618
    },
    {
      "epoch": 3.924890303708165,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010989725325577483,
      "loss": 0.5852,
      "step": 45619
    },
    {
      "epoch": 3.9249763400154865,
      "grad_norm": 1.9453125,
      "learning_rate": 0.00010988033921638457,
      "loss": 1.3662,
      "step": 45620
    },
    {
      "epoch": 3.925062376322808,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00010986342631801865,
      "loss": 1.2739,
      "step": 45621
    },
    {
      "epoch": 3.92514841263013,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00010984651456072615,
      "loss": 0.7731,
      "step": 45622
    },
    {
      "epoch": 3.9252344489374518,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010982960394455688,
      "loss": 0.7718,
      "step": 45623
    },
    {
      "epoch": 3.9253204852447734,
      "grad_norm": 1.296875,
      "learning_rate": 0.00010981269446956,
      "loss": 0.8267,
      "step": 45624
    },
    {
      "epoch": 3.925406521552095,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010979578613578507,
      "loss": 1.6152,
      "step": 45625
    },
    {
      "epoch": 3.9254925578594166,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010977887894328176,
      "loss": 1.4546,
      "step": 45626
    },
    {
      "epoch": 3.9255785941667383,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010976197289209917,
      "loss": 1.3801,
      "step": 45627
    },
    {
      "epoch": 3.92566463047406,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001097450679822869,
      "loss": 0.9105,
      "step": 45628
    },
    {
      "epoch": 3.925750666781382,
      "grad_norm": 6.15625,
      "learning_rate": 0.00010972816421389453,
      "loss": 1.383,
      "step": 45629
    },
    {
      "epoch": 3.9258367030887036,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010971126158697126,
      "loss": 1.2672,
      "step": 45630
    },
    {
      "epoch": 3.925922739396025,
      "grad_norm": 1.75,
      "learning_rate": 0.00010969436010156663,
      "loss": 1.2583,
      "step": 45631
    },
    {
      "epoch": 3.926008775703347,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00010967745975773025,
      "loss": 1.0566,
      "step": 45632
    },
    {
      "epoch": 3.9260948120106685,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010966056055551127,
      "loss": 0.7753,
      "step": 45633
    },
    {
      "epoch": 3.92618084831799,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010964366249495933,
      "loss": 0.7211,
      "step": 45634
    },
    {
      "epoch": 3.9262668846253117,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00010962676557612361,
      "loss": 1.4699,
      "step": 45635
    },
    {
      "epoch": 3.9263529209326338,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00010960986979905369,
      "loss": 1.7287,
      "step": 45636
    },
    {
      "epoch": 3.9264389572399554,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00010959297516379912,
      "loss": 1.14,
      "step": 45637
    },
    {
      "epoch": 3.926524993547277,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010957608167040905,
      "loss": 1.3655,
      "step": 45638
    },
    {
      "epoch": 3.9266110298545986,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010955918931893294,
      "loss": 1.7946,
      "step": 45639
    },
    {
      "epoch": 3.9266970661619203,
      "grad_norm": 3.390625,
      "learning_rate": 0.00010954229810942045,
      "loss": 0.8615,
      "step": 45640
    },
    {
      "epoch": 3.926783102469242,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00010952540804192068,
      "loss": 0.864,
      "step": 45641
    },
    {
      "epoch": 3.9268691387765635,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010950851911648313,
      "loss": 1.4444,
      "step": 45642
    },
    {
      "epoch": 3.9269551750838856,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010949163133315737,
      "loss": 1.3998,
      "step": 45643
    },
    {
      "epoch": 3.927041211391207,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001094747446919923,
      "loss": 1.0391,
      "step": 45644
    },
    {
      "epoch": 3.927127247698529,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010945785919303797,
      "loss": 0.9965,
      "step": 45645
    },
    {
      "epoch": 3.9272132840058505,
      "grad_norm": 10.375,
      "learning_rate": 0.0001094409748363433,
      "loss": 0.7964,
      "step": 45646
    },
    {
      "epoch": 3.927299320313172,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010942409162195782,
      "loss": 1.0489,
      "step": 45647
    },
    {
      "epoch": 3.9273853566204937,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00010940720954993106,
      "loss": 1.1778,
      "step": 45648
    },
    {
      "epoch": 3.9274713929278153,
      "grad_norm": 12.0625,
      "learning_rate": 0.00010939032862031212,
      "loss": 0.6894,
      "step": 45649
    },
    {
      "epoch": 3.9275574292351374,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00010937344883315059,
      "loss": 0.7908,
      "step": 45650
    },
    {
      "epoch": 3.927643465542459,
      "grad_norm": 2.734375,
      "learning_rate": 0.00010935657018849565,
      "loss": 0.8987,
      "step": 45651
    },
    {
      "epoch": 3.9277295018497806,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010933969268639676,
      "loss": 0.6701,
      "step": 45652
    },
    {
      "epoch": 3.9278155381571023,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010932281632690343,
      "loss": 0.8696,
      "step": 45653
    },
    {
      "epoch": 3.927901574464424,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001093059411100647,
      "loss": 1.5505,
      "step": 45654
    },
    {
      "epoch": 3.9279876107717455,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010928906703593017,
      "loss": 0.7836,
      "step": 45655
    },
    {
      "epoch": 3.928073647079067,
      "grad_norm": 1.03125,
      "learning_rate": 0.00010927219410454925,
      "loss": 1.2005,
      "step": 45656
    },
    {
      "epoch": 3.928159683386389,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00010925532231597102,
      "loss": 0.7177,
      "step": 45657
    },
    {
      "epoch": 3.928245719693711,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00010923845167024493,
      "loss": 0.9508,
      "step": 45658
    },
    {
      "epoch": 3.9283317560010325,
      "grad_norm": 1.34375,
      "learning_rate": 0.00010922158216742056,
      "loss": 1.2559,
      "step": 45659
    },
    {
      "epoch": 3.928417792308354,
      "grad_norm": 1352.0,
      "learning_rate": 0.00010920471380754693,
      "loss": 1.2689,
      "step": 45660
    },
    {
      "epoch": 3.9285038286156757,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010918784659067348,
      "loss": 1.3612,
      "step": 45661
    },
    {
      "epoch": 3.9285898649229978,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00010917098051684976,
      "loss": 1.0568,
      "step": 45662
    },
    {
      "epoch": 3.928675901230319,
      "grad_norm": 0.875,
      "learning_rate": 0.0001091541155861247,
      "loss": 0.969,
      "step": 45663
    },
    {
      "epoch": 3.928761937537641,
      "grad_norm": 1.40625,
      "learning_rate": 0.000109137251798548,
      "loss": 1.2631,
      "step": 45664
    },
    {
      "epoch": 3.9288479738449626,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010912038915416871,
      "loss": 1.6112,
      "step": 45665
    },
    {
      "epoch": 3.9289340101522843,
      "grad_norm": 12.8125,
      "learning_rate": 0.00010910352765303622,
      "loss": 1.1981,
      "step": 45666
    },
    {
      "epoch": 3.929020046459606,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010908666729520006,
      "loss": 1.5242,
      "step": 45667
    },
    {
      "epoch": 3.9291060827669275,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00010906980808070921,
      "loss": 0.7173,
      "step": 45668
    },
    {
      "epoch": 3.9291921190742496,
      "grad_norm": 128.0,
      "learning_rate": 0.00010905295000961313,
      "loss": 1.2664,
      "step": 45669
    },
    {
      "epoch": 3.9292781553815708,
      "grad_norm": 9280.0,
      "learning_rate": 0.00010903609308196133,
      "loss": 0.9978,
      "step": 45670
    },
    {
      "epoch": 3.929364191688893,
      "grad_norm": 0.96875,
      "learning_rate": 0.00010901923729780273,
      "loss": 0.4772,
      "step": 45671
    },
    {
      "epoch": 3.9294502279962145,
      "grad_norm": 1.25,
      "learning_rate": 0.00010900238265718682,
      "loss": 1.7887,
      "step": 45672
    },
    {
      "epoch": 3.929536264303536,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00010898552916016302,
      "loss": 0.7565,
      "step": 45673
    },
    {
      "epoch": 3.9296223006108577,
      "grad_norm": 212.0,
      "learning_rate": 0.00010896867680678041,
      "loss": 0.864,
      "step": 45674
    },
    {
      "epoch": 3.9297083369181793,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010895182559708849,
      "loss": 1.3044,
      "step": 45675
    },
    {
      "epoch": 3.9297943732255014,
      "grad_norm": 10.375,
      "learning_rate": 0.0001089349755311363,
      "loss": 0.5975,
      "step": 45676
    },
    {
      "epoch": 3.9298804095328226,
      "grad_norm": 2.375,
      "learning_rate": 0.00010891812660897321,
      "loss": 1.1106,
      "step": 45677
    },
    {
      "epoch": 3.9299664458401447,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010890127883064865,
      "loss": 1.3775,
      "step": 45678
    },
    {
      "epoch": 3.9300524821474663,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010888443219621169,
      "loss": 0.9428,
      "step": 45679
    },
    {
      "epoch": 3.930138518454788,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00010886758670571172,
      "loss": 1.3107,
      "step": 45680
    },
    {
      "epoch": 3.9302245547621095,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00010885074235919795,
      "loss": 1.1144,
      "step": 45681
    },
    {
      "epoch": 3.930310591069431,
      "grad_norm": 1.453125,
      "learning_rate": 0.00010883389915671971,
      "loss": 1.6941,
      "step": 45682
    },
    {
      "epoch": 3.930396627376753,
      "grad_norm": 72.0,
      "learning_rate": 0.0001088170570983264,
      "loss": 1.4349,
      "step": 45683
    },
    {
      "epoch": 3.930482663684075,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010880021618406693,
      "loss": 0.6749,
      "step": 45684
    },
    {
      "epoch": 3.9305686999913965,
      "grad_norm": 544.0,
      "learning_rate": 0.00010878337641399078,
      "loss": 0.7674,
      "step": 45685
    },
    {
      "epoch": 3.930654736298718,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010876653778814728,
      "loss": 1.0734,
      "step": 45686
    },
    {
      "epoch": 3.9307407726060397,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00010874970030658543,
      "loss": 0.7588,
      "step": 45687
    },
    {
      "epoch": 3.9308268089133613,
      "grad_norm": 38.0,
      "learning_rate": 0.00010873286396935461,
      "loss": 1.0638,
      "step": 45688
    },
    {
      "epoch": 3.930912845220683,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00010871602877650422,
      "loss": 0.9158,
      "step": 45689
    },
    {
      "epoch": 3.930998881528005,
      "grad_norm": 0.984375,
      "learning_rate": 0.00010869919472808316,
      "loss": 0.8872,
      "step": 45690
    },
    {
      "epoch": 3.9310849178353267,
      "grad_norm": 1.7890625,
      "learning_rate": 0.00010868236182414088,
      "loss": 1.4484,
      "step": 45691
    },
    {
      "epoch": 3.9311709541426483,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00010866553006472673,
      "loss": 1.427,
      "step": 45692
    },
    {
      "epoch": 3.93125699044997,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001086486994498897,
      "loss": 1.085,
      "step": 45693
    },
    {
      "epoch": 3.9313430267572915,
      "grad_norm": 1.359375,
      "learning_rate": 0.00010863186997967917,
      "loss": 0.9439,
      "step": 45694
    },
    {
      "epoch": 3.931429063064613,
      "grad_norm": 2.21875,
      "learning_rate": 0.00010861504165414421,
      "loss": 1.4438,
      "step": 45695
    },
    {
      "epoch": 3.9315150993719348,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010859821447333412,
      "loss": 1.4637,
      "step": 45696
    },
    {
      "epoch": 3.931601135679257,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010858138843729826,
      "loss": 0.8263,
      "step": 45697
    },
    {
      "epoch": 3.9316871719865785,
      "grad_norm": 0.984375,
      "learning_rate": 0.00010856456354608562,
      "loss": 1.0275,
      "step": 45698
    },
    {
      "epoch": 3.9317732082939,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010854773979974547,
      "loss": 0.5308,
      "step": 45699
    },
    {
      "epoch": 3.9318592446012217,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00010853091719832714,
      "loss": 1.0721,
      "step": 45700
    },
    {
      "epoch": 3.9319452809085433,
      "grad_norm": 3.765625,
      "learning_rate": 0.00010851409574187965,
      "loss": 0.9003,
      "step": 45701
    },
    {
      "epoch": 3.932031317215865,
      "grad_norm": 59.5,
      "learning_rate": 0.00010849727543045229,
      "loss": 1.3573,
      "step": 45702
    },
    {
      "epoch": 3.9321173535231866,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010848045626409442,
      "loss": 1.0415,
      "step": 45703
    },
    {
      "epoch": 3.9322033898305087,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00010846363824285488,
      "loss": 0.7685,
      "step": 45704
    },
    {
      "epoch": 3.9322894261378303,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001084468213667832,
      "loss": 0.5463,
      "step": 45705
    },
    {
      "epoch": 3.932375462445152,
      "grad_norm": 1.375,
      "learning_rate": 0.0001084300056359283,
      "loss": 1.7046,
      "step": 45706
    },
    {
      "epoch": 3.9324614987524735,
      "grad_norm": 1.3125,
      "learning_rate": 0.00010841319105033941,
      "loss": 1.0505,
      "step": 45707
    },
    {
      "epoch": 3.932547535059795,
      "grad_norm": 30.75,
      "learning_rate": 0.00010839637761006599,
      "loss": 1.638,
      "step": 45708
    },
    {
      "epoch": 3.9326335713671168,
      "grad_norm": 284.0,
      "learning_rate": 0.00010837956531515686,
      "loss": 1.1083,
      "step": 45709
    },
    {
      "epoch": 3.9327196076744384,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00010836275416566132,
      "loss": 1.2514,
      "step": 45710
    },
    {
      "epoch": 3.9328056439817605,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00010834594416162869,
      "loss": 0.9632,
      "step": 45711
    },
    {
      "epoch": 3.932891680289082,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00010832913530310784,
      "loss": 1.2122,
      "step": 45712
    },
    {
      "epoch": 3.9329777165964037,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010831232759014814,
      "loss": 0.6251,
      "step": 45713
    },
    {
      "epoch": 3.9330637529037253,
      "grad_norm": 1.03125,
      "learning_rate": 0.00010829552102279878,
      "loss": 0.8031,
      "step": 45714
    },
    {
      "epoch": 3.933149789211047,
      "grad_norm": 1.421875,
      "learning_rate": 0.00010827871560110875,
      "loss": 1.0161,
      "step": 45715
    },
    {
      "epoch": 3.9332358255183686,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010826191132512742,
      "loss": 0.8369,
      "step": 45716
    },
    {
      "epoch": 3.93332186182569,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00010824510819490368,
      "loss": 0.6791,
      "step": 45717
    },
    {
      "epoch": 3.9334078981330123,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010822830621048662,
      "loss": 0.7642,
      "step": 45718
    },
    {
      "epoch": 3.933493934440334,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010821150537192592,
      "loss": 0.9116,
      "step": 45719
    },
    {
      "epoch": 3.9335799707476555,
      "grad_norm": 2.6875,
      "learning_rate": 0.0001081947056792702,
      "loss": 1.1243,
      "step": 45720
    },
    {
      "epoch": 3.933666007054977,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010817790713256887,
      "loss": 0.7865,
      "step": 45721
    },
    {
      "epoch": 3.933752043362299,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00010816110973187082,
      "loss": 0.7516,
      "step": 45722
    },
    {
      "epoch": 3.933838079669621,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00010814431347722537,
      "loss": 0.7677,
      "step": 45723
    },
    {
      "epoch": 3.933924115976942,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010812751836868168,
      "loss": 0.9694,
      "step": 45724
    },
    {
      "epoch": 3.934010152284264,
      "grad_norm": 1.828125,
      "learning_rate": 0.00010811072440628866,
      "loss": 1.4588,
      "step": 45725
    },
    {
      "epoch": 3.9340961885915857,
      "grad_norm": 3.828125,
      "learning_rate": 0.00010809393159009556,
      "loss": 1.1855,
      "step": 45726
    },
    {
      "epoch": 3.9341822248989073,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010807713992015161,
      "loss": 1.5269,
      "step": 45727
    },
    {
      "epoch": 3.934268261206229,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00010806034939650572,
      "loss": 0.6322,
      "step": 45728
    },
    {
      "epoch": 3.9343542975135506,
      "grad_norm": 7.25,
      "learning_rate": 0.00010804356001920701,
      "loss": 1.5554,
      "step": 45729
    },
    {
      "epoch": 3.9344403338208727,
      "grad_norm": 7.03125,
      "learning_rate": 0.00010802677178830484,
      "loss": 1.0172,
      "step": 45730
    },
    {
      "epoch": 3.934526370128194,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00010800998470384792,
      "loss": 1.2743,
      "step": 45731
    },
    {
      "epoch": 3.934612406435516,
      "grad_norm": 3104.0,
      "learning_rate": 0.00010799319876588565,
      "loss": 0.9799,
      "step": 45732
    },
    {
      "epoch": 3.9346984427428375,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010797641397446712,
      "loss": 1.0252,
      "step": 45733
    },
    {
      "epoch": 3.934784479050159,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010795963032964119,
      "loss": 0.9653,
      "step": 45734
    },
    {
      "epoch": 3.934870515357481,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00010794284783145725,
      "loss": 1.521,
      "step": 45735
    },
    {
      "epoch": 3.9349565516648024,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00010792606647996405,
      "loss": 0.7451,
      "step": 45736
    },
    {
      "epoch": 3.9350425879721245,
      "grad_norm": 2.21875,
      "learning_rate": 0.00010790928627521091,
      "loss": 0.9689,
      "step": 45737
    },
    {
      "epoch": 3.9351286242794457,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010789250721724697,
      "loss": 1.1155,
      "step": 45738
    },
    {
      "epoch": 3.9352146605867677,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00010787572930612105,
      "loss": 0.745,
      "step": 45739
    },
    {
      "epoch": 3.9353006968940893,
      "grad_norm": 1.21875,
      "learning_rate": 0.00010785895254188232,
      "loss": 1.4389,
      "step": 45740
    },
    {
      "epoch": 3.935386733201411,
      "grad_norm": 7.84375,
      "learning_rate": 0.00010784217692458009,
      "loss": 1.4146,
      "step": 45741
    },
    {
      "epoch": 3.9354727695087326,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010782540245426304,
      "loss": 0.85,
      "step": 45742
    },
    {
      "epoch": 3.935558805816054,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010780862913098043,
      "loss": 1.1626,
      "step": 45743
    },
    {
      "epoch": 3.9356448421233763,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010779185695478144,
      "loss": 1.1997,
      "step": 45744
    },
    {
      "epoch": 3.935730878430698,
      "grad_norm": 1.25,
      "learning_rate": 0.00010777508592571483,
      "loss": 0.81,
      "step": 45745
    },
    {
      "epoch": 3.9358169147380195,
      "grad_norm": 1.203125,
      "learning_rate": 0.00010775831604382996,
      "loss": 1.5338,
      "step": 45746
    },
    {
      "epoch": 3.935902951045341,
      "grad_norm": 1.4375,
      "learning_rate": 0.00010774154730917563,
      "loss": 0.8157,
      "step": 45747
    },
    {
      "epoch": 3.935988987352663,
      "grad_norm": 155.0,
      "learning_rate": 0.00010772477972180095,
      "loss": 1.9309,
      "step": 45748
    },
    {
      "epoch": 3.9360750236599844,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00010770801328175511,
      "loss": 1.3388,
      "step": 45749
    },
    {
      "epoch": 3.936161059967306,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010769124798908692,
      "loss": 1.6701,
      "step": 45750
    },
    {
      "epoch": 3.936247096274628,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010767448384384554,
      "loss": 0.9795,
      "step": 45751
    },
    {
      "epoch": 3.9363331325819497,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010765772084608011,
      "loss": 1.3687,
      "step": 45752
    },
    {
      "epoch": 3.9364191688892713,
      "grad_norm": 1.515625,
      "learning_rate": 0.00010764095899583942,
      "loss": 0.7349,
      "step": 45753
    },
    {
      "epoch": 3.936505205196593,
      "grad_norm": 4.28125,
      "learning_rate": 0.0001076241982931726,
      "loss": 1.5477,
      "step": 45754
    },
    {
      "epoch": 3.9365912415039146,
      "grad_norm": 13.4375,
      "learning_rate": 0.00010760743873812867,
      "loss": 1.7951,
      "step": 45755
    },
    {
      "epoch": 3.936677277811236,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001075906803307567,
      "loss": 1.3587,
      "step": 45756
    },
    {
      "epoch": 3.936763314118558,
      "grad_norm": 1.21875,
      "learning_rate": 0.00010757392307110581,
      "loss": 1.2311,
      "step": 45757
    },
    {
      "epoch": 3.93684935042588,
      "grad_norm": 105.0,
      "learning_rate": 0.00010755716695922468,
      "loss": 1.0993,
      "step": 45758
    },
    {
      "epoch": 3.9369353867332015,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010754041199516257,
      "loss": 1.2253,
      "step": 45759
    },
    {
      "epoch": 3.937021423040523,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00010752365817896853,
      "loss": 1.0708,
      "step": 45760
    },
    {
      "epoch": 3.937107459347845,
      "grad_norm": 1.21875,
      "learning_rate": 0.00010750690551069137,
      "loss": 1.0508,
      "step": 45761
    },
    {
      "epoch": 3.9371934956551664,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010749015399038025,
      "loss": 1.5336,
      "step": 45762
    },
    {
      "epoch": 3.937279531962488,
      "grad_norm": 1.109375,
      "learning_rate": 0.000107473403618084,
      "loss": 1.0652,
      "step": 45763
    },
    {
      "epoch": 3.9373655682698097,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010745665439385166,
      "loss": 0.964,
      "step": 45764
    },
    {
      "epoch": 3.9374516045771317,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010743990631773242,
      "loss": 0.9987,
      "step": 45765
    },
    {
      "epoch": 3.9375376408844533,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00010742315938977498,
      "loss": 1.539,
      "step": 45766
    },
    {
      "epoch": 3.937623677191775,
      "grad_norm": 2.578125,
      "learning_rate": 0.00010740641361002845,
      "loss": 1.4566,
      "step": 45767
    },
    {
      "epoch": 3.9377097134990966,
      "grad_norm": 5.375,
      "learning_rate": 0.00010738966897854191,
      "loss": 0.9229,
      "step": 45768
    },
    {
      "epoch": 3.937795749806418,
      "grad_norm": 1.4375,
      "learning_rate": 0.00010737292549536414,
      "loss": 1.4544,
      "step": 45769
    },
    {
      "epoch": 3.93788178611374,
      "grad_norm": 1.203125,
      "learning_rate": 0.00010735618316054418,
      "loss": 1.3841,
      "step": 45770
    },
    {
      "epoch": 3.9379678224210615,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010733944197413115,
      "loss": 0.7558,
      "step": 45771
    },
    {
      "epoch": 3.9380538587283835,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00010732270193617372,
      "loss": 1.0687,
      "step": 45772
    },
    {
      "epoch": 3.938139895035705,
      "grad_norm": 1.140625,
      "learning_rate": 0.00010730596304672103,
      "loss": 0.7771,
      "step": 45773
    },
    {
      "epoch": 3.938225931343027,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010728922530582219,
      "loss": 1.2946,
      "step": 45774
    },
    {
      "epoch": 3.9383119676503484,
      "grad_norm": 1.46875,
      "learning_rate": 0.00010727248871352585,
      "loss": 2.0738,
      "step": 45775
    },
    {
      "epoch": 3.93839800395767,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010725575326988118,
      "loss": 0.7957,
      "step": 45776
    },
    {
      "epoch": 3.938484040264992,
      "grad_norm": 1.0,
      "learning_rate": 0.00010723901897493699,
      "loss": 1.0391,
      "step": 45777
    },
    {
      "epoch": 3.9385700765723133,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010722228582874221,
      "loss": 0.6825,
      "step": 45778
    },
    {
      "epoch": 3.9386561128796354,
      "grad_norm": 1.015625,
      "learning_rate": 0.000107205553831346,
      "loss": 1.0325,
      "step": 45779
    },
    {
      "epoch": 3.938742149186957,
      "grad_norm": 1.6953125,
      "learning_rate": 0.00010718882298279708,
      "loss": 1.0229,
      "step": 45780
    },
    {
      "epoch": 3.9388281854942786,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010717209328314436,
      "loss": 1.1279,
      "step": 45781
    },
    {
      "epoch": 3.9389142218016002,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010715536473243708,
      "loss": 1.2416,
      "step": 45782
    },
    {
      "epoch": 3.939000258108922,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010713863733072376,
      "loss": 0.8702,
      "step": 45783
    },
    {
      "epoch": 3.939086294416244,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010712191107805358,
      "loss": 0.9365,
      "step": 45784
    },
    {
      "epoch": 3.939172330723565,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001071051859744755,
      "loss": 0.992,
      "step": 45785
    },
    {
      "epoch": 3.939258367030887,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001070884620200382,
      "loss": 1.4141,
      "step": 45786
    },
    {
      "epoch": 3.939344403338209,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00010707173921479085,
      "loss": 1.2776,
      "step": 45787
    },
    {
      "epoch": 3.9394304396455304,
      "grad_norm": 366.0,
      "learning_rate": 0.00010705501755878216,
      "loss": 0.8824,
      "step": 45788
    },
    {
      "epoch": 3.939516475952852,
      "grad_norm": 24.5,
      "learning_rate": 0.00010703829705206109,
      "loss": 2.1739,
      "step": 45789
    },
    {
      "epoch": 3.9396025122601737,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010702157769467669,
      "loss": 1.6786,
      "step": 45790
    },
    {
      "epoch": 3.9396885485674957,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00010700485948667765,
      "loss": 1.3542,
      "step": 45791
    },
    {
      "epoch": 3.939774584874817,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010698814242811295,
      "loss": 0.5895,
      "step": 45792
    },
    {
      "epoch": 3.939860621182139,
      "grad_norm": 1.421875,
      "learning_rate": 0.00010697142651903147,
      "loss": 1.6622,
      "step": 45793
    },
    {
      "epoch": 3.9399466574894606,
      "grad_norm": 1.15625,
      "learning_rate": 0.00010695471175948213,
      "loss": 1.5324,
      "step": 45794
    },
    {
      "epoch": 3.9400326937967822,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010693799814951399,
      "loss": 0.6256,
      "step": 45795
    },
    {
      "epoch": 3.940118730104104,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010692128568917558,
      "loss": 0.8646,
      "step": 45796
    },
    {
      "epoch": 3.9402047664114255,
      "grad_norm": 1.7734375,
      "learning_rate": 0.00010690457437851597,
      "loss": 1.1076,
      "step": 45797
    },
    {
      "epoch": 3.9402908027187475,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010688786421758417,
      "loss": 0.9451,
      "step": 45798
    },
    {
      "epoch": 3.940376839026069,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00010687115520642882,
      "loss": 1.7357,
      "step": 45799
    },
    {
      "epoch": 3.940462875333391,
      "grad_norm": 13.4375,
      "learning_rate": 0.00010685444734509881,
      "loss": 1.0126,
      "step": 45800
    },
    {
      "epoch": 3.9405489116407124,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010683774063364326,
      "loss": 0.6053,
      "step": 45801
    },
    {
      "epoch": 3.940634947948034,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010682103507211072,
      "loss": 1.1881,
      "step": 45802
    },
    {
      "epoch": 3.9407209842553557,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00010680433066055028,
      "loss": 1.0821,
      "step": 45803
    },
    {
      "epoch": 3.9408070205626773,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001067876273990106,
      "loss": 1.3547,
      "step": 45804
    },
    {
      "epoch": 3.9408930568699994,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010677092528754062,
      "loss": 1.0649,
      "step": 45805
    },
    {
      "epoch": 3.940979093177321,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00010675422432618931,
      "loss": 1.399,
      "step": 45806
    },
    {
      "epoch": 3.9410651294846426,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00010673752451500534,
      "loss": 1.4198,
      "step": 45807
    },
    {
      "epoch": 3.9411511657919642,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010672082585403759,
      "loss": 0.5158,
      "step": 45808
    },
    {
      "epoch": 3.941237202099286,
      "grad_norm": 462.0,
      "learning_rate": 0.00010670412834333504,
      "loss": 0.9049,
      "step": 45809
    },
    {
      "epoch": 3.9413232384066075,
      "grad_norm": 1.171875,
      "learning_rate": 0.00010668743198294634,
      "loss": 1.0443,
      "step": 45810
    },
    {
      "epoch": 3.941409274713929,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010667073677292039,
      "loss": 0.8056,
      "step": 45811
    },
    {
      "epoch": 3.941495311021251,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00010665404271330614,
      "loss": 1.3038,
      "step": 45812
    },
    {
      "epoch": 3.941581347328573,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00010663734980415224,
      "loss": 1.3989,
      "step": 45813
    },
    {
      "epoch": 3.9416673836358944,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010662065804550752,
      "loss": 0.6127,
      "step": 45814
    },
    {
      "epoch": 3.941753419943216,
      "grad_norm": 1.7578125,
      "learning_rate": 0.00010660396743742106,
      "loss": 0.6229,
      "step": 45815
    },
    {
      "epoch": 3.9418394562505377,
      "grad_norm": 1.25,
      "learning_rate": 0.00010658727797994133,
      "loss": 0.875,
      "step": 45816
    },
    {
      "epoch": 3.9419254925578593,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00010657058967311744,
      "loss": 1.8943,
      "step": 45817
    },
    {
      "epoch": 3.942011528865181,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010655390251699793,
      "loss": 0.9431,
      "step": 45818
    },
    {
      "epoch": 3.942097565172503,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010653721651163168,
      "loss": 1.2073,
      "step": 45819
    },
    {
      "epoch": 3.9421836014798246,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010652053165706771,
      "loss": 1.7703,
      "step": 45820
    },
    {
      "epoch": 3.9422696377871462,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010650384795335455,
      "loss": 1.2121,
      "step": 45821
    },
    {
      "epoch": 3.942355674094468,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010648716540054104,
      "loss": 1.6826,
      "step": 45822
    },
    {
      "epoch": 3.9424417104017895,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00010647048399867626,
      "loss": 1.2994,
      "step": 45823
    },
    {
      "epoch": 3.942527746709111,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001064538037478086,
      "loss": 0.7378,
      "step": 45824
    },
    {
      "epoch": 3.9426137830164327,
      "grad_norm": 1.65625,
      "learning_rate": 0.000106437124647987,
      "loss": 0.971,
      "step": 45825
    },
    {
      "epoch": 3.942699819323755,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010642044669926043,
      "loss": 1.291,
      "step": 45826
    },
    {
      "epoch": 3.9427858556310764,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010640376990167738,
      "loss": 1.1526,
      "step": 45827
    },
    {
      "epoch": 3.942871891938398,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00010638709425528687,
      "loss": 0.9066,
      "step": 45828
    },
    {
      "epoch": 3.9429579282457197,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010637041976013728,
      "loss": 1.2832,
      "step": 45829
    },
    {
      "epoch": 3.9430439645530413,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010635374641627776,
      "loss": 0.7501,
      "step": 45830
    },
    {
      "epoch": 3.943130000860363,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010633707422375715,
      "loss": 1.985,
      "step": 45831
    },
    {
      "epoch": 3.9432160371676845,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00010632040318262392,
      "loss": 1.64,
      "step": 45832
    },
    {
      "epoch": 3.9433020734750066,
      "grad_norm": 1.203125,
      "learning_rate": 0.00010630373329292703,
      "loss": 1.1555,
      "step": 45833
    },
    {
      "epoch": 3.9433881097823282,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010628706455471504,
      "loss": 1.7306,
      "step": 45834
    },
    {
      "epoch": 3.94347414608965,
      "grad_norm": 1.453125,
      "learning_rate": 0.00010627039696803681,
      "loss": 1.7754,
      "step": 45835
    },
    {
      "epoch": 3.9435601823969715,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00010625373053294124,
      "loss": 1.4376,
      "step": 45836
    },
    {
      "epoch": 3.943646218704293,
      "grad_norm": 1.25,
      "learning_rate": 0.0001062370652494768,
      "loss": 0.9615,
      "step": 45837
    },
    {
      "epoch": 3.943732255011615,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00010622040111769232,
      "loss": 0.9171,
      "step": 45838
    },
    {
      "epoch": 3.9438182913189364,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010620373813763679,
      "loss": 1.1962,
      "step": 45839
    },
    {
      "epoch": 3.9439043276262584,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00010618707630935854,
      "loss": 1.7702,
      "step": 45840
    },
    {
      "epoch": 3.94399036393358,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010617041563290646,
      "loss": 0.8544,
      "step": 45841
    },
    {
      "epoch": 3.9440764002409017,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010615375610832956,
      "loss": 1.2639,
      "step": 45842
    },
    {
      "epoch": 3.9441624365482233,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001061370977356761,
      "loss": 1.3771,
      "step": 45843
    },
    {
      "epoch": 3.944248472855545,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010612044051499508,
      "loss": 0.4646,
      "step": 45844
    },
    {
      "epoch": 3.944334509162867,
      "grad_norm": 1.03125,
      "learning_rate": 0.00010610378444633528,
      "loss": 0.9146,
      "step": 45845
    },
    {
      "epoch": 3.944420545470188,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001060871295297452,
      "loss": 1.734,
      "step": 45846
    },
    {
      "epoch": 3.9445065817775102,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010607047576527373,
      "loss": 1.7277,
      "step": 45847
    },
    {
      "epoch": 3.944592618084832,
      "grad_norm": 13.0,
      "learning_rate": 0.00010605382315296941,
      "loss": 1.7883,
      "step": 45848
    },
    {
      "epoch": 3.9446786543921535,
      "grad_norm": 1.40625,
      "learning_rate": 0.00010603717169288102,
      "loss": 0.7821,
      "step": 45849
    },
    {
      "epoch": 3.944764690699475,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001060205213850574,
      "loss": 1.5748,
      "step": 45850
    },
    {
      "epoch": 3.9448507270067967,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00010600387222954705,
      "loss": 1.4459,
      "step": 45851
    },
    {
      "epoch": 3.944936763314119,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00010598722422639872,
      "loss": 1.8335,
      "step": 45852
    },
    {
      "epoch": 3.94502279962144,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010597057737566124,
      "loss": 1.0164,
      "step": 45853
    },
    {
      "epoch": 3.945108835928762,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001059539316773831,
      "loss": 1.1571,
      "step": 45854
    },
    {
      "epoch": 3.9451948722360837,
      "grad_norm": 1.21875,
      "learning_rate": 0.00010593728713161299,
      "loss": 0.8095,
      "step": 45855
    },
    {
      "epoch": 3.9452809085434053,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010592064373839988,
      "loss": 1.1317,
      "step": 45856
    },
    {
      "epoch": 3.945366944850727,
      "grad_norm": 1.671875,
      "learning_rate": 0.00010590400149779212,
      "loss": 1.7477,
      "step": 45857
    },
    {
      "epoch": 3.9454529811580485,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001058873604098386,
      "loss": 1.0489,
      "step": 45858
    },
    {
      "epoch": 3.9455390174653706,
      "grad_norm": 2.546875,
      "learning_rate": 0.00010587072047458773,
      "loss": 0.7093,
      "step": 45859
    },
    {
      "epoch": 3.9456250537726922,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00010585408169208843,
      "loss": 0.8643,
      "step": 45860
    },
    {
      "epoch": 3.945711090080014,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010583744406238938,
      "loss": 1.0665,
      "step": 45861
    },
    {
      "epoch": 3.9457971263873355,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00010582080758553902,
      "loss": 1.305,
      "step": 45862
    },
    {
      "epoch": 3.945883162694657,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010580417226158612,
      "loss": 0.9107,
      "step": 45863
    },
    {
      "epoch": 3.9459691990019787,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010578753809057951,
      "loss": 1.1725,
      "step": 45864
    },
    {
      "epoch": 3.9460552353093004,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010577090507256754,
      "loss": 1.2182,
      "step": 45865
    },
    {
      "epoch": 3.9461412716166224,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00010575427320759895,
      "loss": 0.9646,
      "step": 45866
    },
    {
      "epoch": 3.946227307923944,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010573764249572249,
      "loss": 0.8498,
      "step": 45867
    },
    {
      "epoch": 3.9463133442312657,
      "grad_norm": 1.28125,
      "learning_rate": 0.00010572101293698671,
      "loss": 1.9861,
      "step": 45868
    },
    {
      "epoch": 3.9463993805385873,
      "grad_norm": 1.890625,
      "learning_rate": 0.00010570438453144044,
      "loss": 1.8019,
      "step": 45869
    },
    {
      "epoch": 3.946485416845909,
      "grad_norm": 1.28125,
      "learning_rate": 0.00010568775727913204,
      "loss": 1.5199,
      "step": 45870
    },
    {
      "epoch": 3.9465714531532305,
      "grad_norm": 1.203125,
      "learning_rate": 0.00010567113118011023,
      "loss": 1.3342,
      "step": 45871
    },
    {
      "epoch": 3.946657489460552,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001056545062344238,
      "loss": 0.9488,
      "step": 45872
    },
    {
      "epoch": 3.9467435257678742,
      "grad_norm": 1.8515625,
      "learning_rate": 0.00010563788244212113,
      "loss": 0.6456,
      "step": 45873
    },
    {
      "epoch": 3.946829562075196,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010562125980325111,
      "loss": 1.4597,
      "step": 45874
    },
    {
      "epoch": 3.9469155983825175,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010560463831786204,
      "loss": 0.9081,
      "step": 45875
    },
    {
      "epoch": 3.947001634689839,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010558801798600271,
      "loss": 1.1974,
      "step": 45876
    },
    {
      "epoch": 3.9470876709971607,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010557139880772181,
      "loss": 0.5648,
      "step": 45877
    },
    {
      "epoch": 3.9471737073044824,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001055547807830678,
      "loss": 0.7502,
      "step": 45878
    },
    {
      "epoch": 3.947259743611804,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00010553816391208926,
      "loss": 0.7408,
      "step": 45879
    },
    {
      "epoch": 3.947345779919126,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010552154819483506,
      "loss": 1.434,
      "step": 45880
    },
    {
      "epoch": 3.9474318162264477,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00010550493363135343,
      "loss": 1.2865,
      "step": 45881
    },
    {
      "epoch": 3.9475178525337693,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00010548832022169319,
      "loss": 1.2939,
      "step": 45882
    },
    {
      "epoch": 3.947603888841091,
      "grad_norm": 7.28125,
      "learning_rate": 0.00010547170796590299,
      "loss": 0.6316,
      "step": 45883
    },
    {
      "epoch": 3.9476899251484125,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010545509686403121,
      "loss": 0.9101,
      "step": 45884
    },
    {
      "epoch": 3.947775961455734,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010543848691612651,
      "loss": 1.4048,
      "step": 45885
    },
    {
      "epoch": 3.947861997763056,
      "grad_norm": 2.15625,
      "learning_rate": 0.00010542187812223764,
      "loss": 2.2071,
      "step": 45886
    },
    {
      "epoch": 3.947948034070378,
      "grad_norm": 1.3125,
      "learning_rate": 0.00010540527048241289,
      "loss": 1.1659,
      "step": 45887
    },
    {
      "epoch": 3.9480340703776995,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001053886639967011,
      "loss": 0.8355,
      "step": 45888
    },
    {
      "epoch": 3.948120106685021,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010537205866515065,
      "loss": 0.6013,
      "step": 45889
    },
    {
      "epoch": 3.9482061429923427,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010535545448781008,
      "loss": 1.135,
      "step": 45890
    },
    {
      "epoch": 3.9482921792996644,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010533885146472827,
      "loss": 1.0561,
      "step": 45891
    },
    {
      "epoch": 3.948378215606986,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00010532224959595338,
      "loss": 1.1601,
      "step": 45892
    },
    {
      "epoch": 3.9484642519143076,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001053056488815342,
      "loss": 1.3716,
      "step": 45893
    },
    {
      "epoch": 3.9485502882216297,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010528904932151934,
      "loss": 0.6652,
      "step": 45894
    },
    {
      "epoch": 3.9486363245289513,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00010527245091595711,
      "loss": 1.601,
      "step": 45895
    },
    {
      "epoch": 3.948722360836273,
      "grad_norm": 3.359375,
      "learning_rate": 0.00010525585366489614,
      "loss": 1.3213,
      "step": 45896
    },
    {
      "epoch": 3.9488083971435946,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010523925756838527,
      "loss": 0.6679,
      "step": 45897
    },
    {
      "epoch": 3.948894433450916,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00010522266262647256,
      "loss": 1.4552,
      "step": 45898
    },
    {
      "epoch": 3.9489804697582382,
      "grad_norm": 0.96875,
      "learning_rate": 0.00010520606883920697,
      "loss": 1.1097,
      "step": 45899
    },
    {
      "epoch": 3.9490665060655594,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010518947620663666,
      "loss": 1.3754,
      "step": 45900
    },
    {
      "epoch": 3.9491525423728815,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010517288472881037,
      "loss": 0.7029,
      "step": 45901
    },
    {
      "epoch": 3.949238578680203,
      "grad_norm": 86.5,
      "learning_rate": 0.00010515629440577678,
      "loss": 1.9916,
      "step": 45902
    },
    {
      "epoch": 3.9493246149875247,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010513970523758393,
      "loss": 1.3204,
      "step": 45903
    },
    {
      "epoch": 3.9494106512948464,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010512311722428092,
      "loss": 0.9809,
      "step": 45904
    },
    {
      "epoch": 3.949496687602168,
      "grad_norm": 1.421875,
      "learning_rate": 0.00010510653036591584,
      "loss": 1.9897,
      "step": 45905
    },
    {
      "epoch": 3.94958272390949,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010508994466253735,
      "loss": 1.2109,
      "step": 45906
    },
    {
      "epoch": 3.9496687602168112,
      "grad_norm": 1.359375,
      "learning_rate": 0.00010507336011419416,
      "loss": 0.7322,
      "step": 45907
    },
    {
      "epoch": 3.9497547965241333,
      "grad_norm": 1.375,
      "learning_rate": 0.00010505677672093439,
      "loss": 0.7258,
      "step": 45908
    },
    {
      "epoch": 3.949840832831455,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010504019448280677,
      "loss": 0.7351,
      "step": 45909
    },
    {
      "epoch": 3.9499268691387766,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010502361339985989,
      "loss": 0.7583,
      "step": 45910
    },
    {
      "epoch": 3.950012905446098,
      "grad_norm": 1.9140625,
      "learning_rate": 0.00010500703347214196,
      "loss": 0.9171,
      "step": 45911
    },
    {
      "epoch": 3.95009894175342,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001049904546997017,
      "loss": 1.4884,
      "step": 45912
    },
    {
      "epoch": 3.950184978060742,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001049738770825876,
      "loss": 0.8636,
      "step": 45913
    },
    {
      "epoch": 3.950271014368063,
      "grad_norm": 1.328125,
      "learning_rate": 0.00010495730062084802,
      "loss": 1.0249,
      "step": 45914
    },
    {
      "epoch": 3.950357050675385,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010494072531453159,
      "loss": 0.9331,
      "step": 45915
    },
    {
      "epoch": 3.9504430869827067,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010492415116368654,
      "loss": 0.3814,
      "step": 45916
    },
    {
      "epoch": 3.9505291232900284,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010490757816836155,
      "loss": 1.0848,
      "step": 45917
    },
    {
      "epoch": 3.95061515959735,
      "grad_norm": 1.28125,
      "learning_rate": 0.00010489100632860521,
      "loss": 1.5602,
      "step": 45918
    },
    {
      "epoch": 3.9507011959046716,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00010487443564446564,
      "loss": 0.8629,
      "step": 45919
    },
    {
      "epoch": 3.9507872322119937,
      "grad_norm": 2.46875,
      "learning_rate": 0.00010485786611599152,
      "loss": 1.1684,
      "step": 45920
    },
    {
      "epoch": 3.9508732685193153,
      "grad_norm": 1.296875,
      "learning_rate": 0.00010484129774323142,
      "loss": 0.6224,
      "step": 45921
    },
    {
      "epoch": 3.950959304826637,
      "grad_norm": 31.0,
      "learning_rate": 0.00010482473052623359,
      "loss": 1.3772,
      "step": 45922
    },
    {
      "epoch": 3.9510453411339586,
      "grad_norm": 1.453125,
      "learning_rate": 0.00010480816446504648,
      "loss": 1.3293,
      "step": 45923
    },
    {
      "epoch": 3.95113137744128,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010479159955971884,
      "loss": 1.2723,
      "step": 45924
    },
    {
      "epoch": 3.951217413748602,
      "grad_norm": 5.0625,
      "learning_rate": 0.00010477503581029874,
      "loss": 1.662,
      "step": 45925
    },
    {
      "epoch": 3.9513034500559234,
      "grad_norm": 23.75,
      "learning_rate": 0.00010475847321683479,
      "loss": 0.9593,
      "step": 45926
    },
    {
      "epoch": 3.9513894863632455,
      "grad_norm": 4.40625,
      "learning_rate": 0.00010474191177937559,
      "loss": 1.5316,
      "step": 45927
    },
    {
      "epoch": 3.951475522670567,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010472535149796925,
      "loss": 1.3973,
      "step": 45928
    },
    {
      "epoch": 3.9515615589778887,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010470879237266451,
      "loss": 1.1245,
      "step": 45929
    },
    {
      "epoch": 3.9516475952852104,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00010469223440350962,
      "loss": 1.2881,
      "step": 45930
    },
    {
      "epoch": 3.951733631592532,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00010467567759055296,
      "loss": 0.656,
      "step": 45931
    },
    {
      "epoch": 3.9518196678998536,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010465912193384325,
      "loss": 1.2081,
      "step": 45932
    },
    {
      "epoch": 3.9519057042071752,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010464256743342859,
      "loss": 1.2602,
      "step": 45933
    },
    {
      "epoch": 3.9519917405144973,
      "grad_norm": 1.25,
      "learning_rate": 0.00010462601408935751,
      "loss": 0.8034,
      "step": 45934
    },
    {
      "epoch": 3.952077776821819,
      "grad_norm": 84.0,
      "learning_rate": 0.00010460946190167858,
      "loss": 0.7207,
      "step": 45935
    },
    {
      "epoch": 3.9521638131291406,
      "grad_norm": 31.5,
      "learning_rate": 0.00010459291087043992,
      "loss": 1.1988,
      "step": 45936
    },
    {
      "epoch": 3.952249849436462,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00010457636099569012,
      "loss": 1.8312,
      "step": 45937
    },
    {
      "epoch": 3.952335885743784,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00010455981227747769,
      "loss": 0.6331,
      "step": 45938
    },
    {
      "epoch": 3.9524219220511054,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010454326471585074,
      "loss": 1.3342,
      "step": 45939
    },
    {
      "epoch": 3.952507958358427,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010452671831085787,
      "loss": 0.9859,
      "step": 45940
    },
    {
      "epoch": 3.952593994665749,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001045101730625474,
      "loss": 0.4439,
      "step": 45941
    },
    {
      "epoch": 3.9526800309730707,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010449362897096776,
      "loss": 0.9864,
      "step": 45942
    },
    {
      "epoch": 3.9527660672803924,
      "grad_norm": 1.359375,
      "learning_rate": 0.00010447708603616746,
      "loss": 1.264,
      "step": 45943
    },
    {
      "epoch": 3.952852103587714,
      "grad_norm": 1.03125,
      "learning_rate": 0.00010446054425819468,
      "loss": 1.7314,
      "step": 45944
    },
    {
      "epoch": 3.9529381398950356,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00010444400363709794,
      "loss": 0.6318,
      "step": 45945
    },
    {
      "epoch": 3.9530241762023572,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010442746417292547,
      "loss": 1.7171,
      "step": 45946
    },
    {
      "epoch": 3.953110212509679,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001044109258657257,
      "loss": 1.5605,
      "step": 45947
    },
    {
      "epoch": 3.953196248817001,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010439438871554718,
      "loss": 2.3001,
      "step": 45948
    },
    {
      "epoch": 3.9532822851243226,
      "grad_norm": 2.796875,
      "learning_rate": 0.00010437785272243805,
      "loss": 1.3937,
      "step": 45949
    },
    {
      "epoch": 3.953368321431644,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010436131788644671,
      "loss": 1.013,
      "step": 45950
    },
    {
      "epoch": 3.953454357738966,
      "grad_norm": 2.1875,
      "learning_rate": 0.00010434478420762167,
      "loss": 1.2216,
      "step": 45951
    },
    {
      "epoch": 3.9535403940462874,
      "grad_norm": 1.15625,
      "learning_rate": 0.00010432825168601112,
      "loss": 1.2637,
      "step": 45952
    },
    {
      "epoch": 3.9536264303536095,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010431172032166341,
      "loss": 0.6639,
      "step": 45953
    },
    {
      "epoch": 3.9537124666609307,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00010429519011462713,
      "loss": 0.6916,
      "step": 45954
    },
    {
      "epoch": 3.9537985029682527,
      "grad_norm": 1.359375,
      "learning_rate": 0.00010427866106495032,
      "loss": 0.9182,
      "step": 45955
    },
    {
      "epoch": 3.9538845392755744,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010426213317268146,
      "loss": 1.1544,
      "step": 45956
    },
    {
      "epoch": 3.953970575582896,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00010424560643786907,
      "loss": 1.6638,
      "step": 45957
    },
    {
      "epoch": 3.9540566118902176,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010422908086056115,
      "loss": 1.3385,
      "step": 45958
    },
    {
      "epoch": 3.9541426481975392,
      "grad_norm": 3.5,
      "learning_rate": 0.0001042125564408063,
      "loss": 0.7812,
      "step": 45959
    },
    {
      "epoch": 3.9542286845048613,
      "grad_norm": 1.25,
      "learning_rate": 0.00010419603317865262,
      "loss": 1.4331,
      "step": 45960
    },
    {
      "epoch": 3.9543147208121825,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010417951107414858,
      "loss": 2.171,
      "step": 45961
    },
    {
      "epoch": 3.9544007571195046,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010416299012734265,
      "loss": 1.7031,
      "step": 45962
    },
    {
      "epoch": 3.954486793426826,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010414647033828278,
      "loss": 1.8608,
      "step": 45963
    },
    {
      "epoch": 3.954572829734148,
      "grad_norm": 1.125,
      "learning_rate": 0.00010412995170701756,
      "loss": 1.4231,
      "step": 45964
    },
    {
      "epoch": 3.9546588660414694,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010411343423359537,
      "loss": 1.3686,
      "step": 45965
    },
    {
      "epoch": 3.954744902348791,
      "grad_norm": 1.21875,
      "learning_rate": 0.00010409691791806419,
      "loss": 1.5849,
      "step": 45966
    },
    {
      "epoch": 3.954830938656113,
      "grad_norm": 1.453125,
      "learning_rate": 0.00010408040276047259,
      "loss": 0.9216,
      "step": 45967
    },
    {
      "epoch": 3.9549169749634343,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00010406388876086892,
      "loss": 1.2009,
      "step": 45968
    },
    {
      "epoch": 3.9550030112707564,
      "grad_norm": 39.25,
      "learning_rate": 0.00010404737591930125,
      "loss": 1.0709,
      "step": 45969
    },
    {
      "epoch": 3.955089047578078,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010403086423581814,
      "loss": 0.9228,
      "step": 45970
    },
    {
      "epoch": 3.9551750838853996,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010401435371046758,
      "loss": 1.3129,
      "step": 45971
    },
    {
      "epoch": 3.9552611201927212,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00010399784434329801,
      "loss": 0.7677,
      "step": 45972
    },
    {
      "epoch": 3.955347156500043,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010398133613435789,
      "loss": 1.2709,
      "step": 45973
    },
    {
      "epoch": 3.955433192807365,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001039648290836952,
      "loss": 0.7009,
      "step": 45974
    },
    {
      "epoch": 3.9555192291146866,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010394832319135839,
      "loss": 0.4691,
      "step": 45975
    },
    {
      "epoch": 3.955605265422008,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010393181845739564,
      "loss": 0.7254,
      "step": 45976
    },
    {
      "epoch": 3.95569130172933,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010391531488185536,
      "loss": 0.7092,
      "step": 45977
    },
    {
      "epoch": 3.9557773380366514,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00010389881246478582,
      "loss": 1.0867,
      "step": 45978
    },
    {
      "epoch": 3.955863374343973,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00010388231120623509,
      "loss": 0.8135,
      "step": 45979
    },
    {
      "epoch": 3.9559494106512947,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010386581110625159,
      "loss": 1.1007,
      "step": 45980
    },
    {
      "epoch": 3.9560354469586168,
      "grad_norm": 0.96875,
      "learning_rate": 0.00010384931216488369,
      "loss": 0.9888,
      "step": 45981
    },
    {
      "epoch": 3.9561214832659384,
      "grad_norm": 1.203125,
      "learning_rate": 0.00010383281438217935,
      "loss": 0.9187,
      "step": 45982
    },
    {
      "epoch": 3.95620751957326,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010381631775818695,
      "loss": 0.5839,
      "step": 45983
    },
    {
      "epoch": 3.9562935558805816,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010379982229295493,
      "loss": 0.9768,
      "step": 45984
    },
    {
      "epoch": 3.9563795921879032,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00010378332798653128,
      "loss": 0.9705,
      "step": 45985
    },
    {
      "epoch": 3.956465628495225,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00010376683483896443,
      "loss": 1.2851,
      "step": 45986
    },
    {
      "epoch": 3.9565516648025465,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00010375034285030238,
      "loss": 1.0672,
      "step": 45987
    },
    {
      "epoch": 3.9566377011098686,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010373385202059349,
      "loss": 0.8873,
      "step": 45988
    },
    {
      "epoch": 3.95672373741719,
      "grad_norm": 3584.0,
      "learning_rate": 0.00010371736234988622,
      "loss": 1.6519,
      "step": 45989
    },
    {
      "epoch": 3.956809773724512,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010370087383822841,
      "loss": 1.8045,
      "step": 45990
    },
    {
      "epoch": 3.9568958100318334,
      "grad_norm": 1.25,
      "learning_rate": 0.00010368438648566847,
      "loss": 1.5735,
      "step": 45991
    },
    {
      "epoch": 3.956981846339155,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010366790029225476,
      "loss": 0.8133,
      "step": 45992
    },
    {
      "epoch": 3.9570678826464767,
      "grad_norm": 9.875,
      "learning_rate": 0.00010365141525803523,
      "loss": 0.7675,
      "step": 45993
    },
    {
      "epoch": 3.9571539189537983,
      "grad_norm": 248.0,
      "learning_rate": 0.00010363493138305818,
      "loss": 0.6209,
      "step": 45994
    },
    {
      "epoch": 3.9572399552611204,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010361844866737208,
      "loss": 0.4821,
      "step": 45995
    },
    {
      "epoch": 3.957325991568442,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010360196711102477,
      "loss": 1.3058,
      "step": 45996
    },
    {
      "epoch": 3.9574120278757636,
      "grad_norm": 4.0625,
      "learning_rate": 0.00010358548671406459,
      "loss": 1.2508,
      "step": 45997
    },
    {
      "epoch": 3.9574980641830853,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010356900747653985,
      "loss": 1.7063,
      "step": 45998
    },
    {
      "epoch": 3.957584100490407,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00010355252939849863,
      "loss": 1.1145,
      "step": 45999
    },
    {
      "epoch": 3.9576701367977285,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001035360524799892,
      "loss": 0.7688,
      "step": 46000
    },
    {
      "epoch": 3.95775617310505,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001035195767210596,
      "loss": 0.8118,
      "step": 46001
    },
    {
      "epoch": 3.957842209412372,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010350310212175818,
      "loss": 1.2128,
      "step": 46002
    },
    {
      "epoch": 3.957928245719694,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010348662868213316,
      "loss": 1.018,
      "step": 46003
    },
    {
      "epoch": 3.9580142820270154,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010347015640223251,
      "loss": 0.8613,
      "step": 46004
    },
    {
      "epoch": 3.958100318334337,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010345368528210453,
      "loss": 0.8554,
      "step": 46005
    },
    {
      "epoch": 3.9581863546416587,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010343721532179751,
      "loss": 2.0435,
      "step": 46006
    },
    {
      "epoch": 3.9582723909489803,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010342074652135941,
      "loss": 1.3983,
      "step": 46007
    },
    {
      "epoch": 3.958358427256302,
      "grad_norm": 1.96875,
      "learning_rate": 0.00010340427888083848,
      "loss": 0.9224,
      "step": 46008
    },
    {
      "epoch": 3.958444463563624,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00010338781240028305,
      "loss": 1.1208,
      "step": 46009
    },
    {
      "epoch": 3.9585304998709456,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00010337134707974093,
      "loss": 1.127,
      "step": 46010
    },
    {
      "epoch": 3.9586165361782673,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001033548829192607,
      "loss": 0.8203,
      "step": 46011
    },
    {
      "epoch": 3.958702572485589,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001033384199188901,
      "loss": 0.7228,
      "step": 46012
    },
    {
      "epoch": 3.9587886087929105,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001033219580786774,
      "loss": 0.5011,
      "step": 46013
    },
    {
      "epoch": 3.9588746451002326,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010330549739867107,
      "loss": 0.9728,
      "step": 46014
    },
    {
      "epoch": 3.9589606814075537,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00010328903787891885,
      "loss": 1.3466,
      "step": 46015
    },
    {
      "epoch": 3.959046717714876,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010327257951946917,
      "loss": 1.0183,
      "step": 46016
    },
    {
      "epoch": 3.9591327540221974,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001032561223203699,
      "loss": 0.9178,
      "step": 46017
    },
    {
      "epoch": 3.959218790329519,
      "grad_norm": 1.71875,
      "learning_rate": 0.00010323966628166931,
      "loss": 1.1375,
      "step": 46018
    },
    {
      "epoch": 3.9593048266368407,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00010322321140341573,
      "loss": 1.3337,
      "step": 46019
    },
    {
      "epoch": 3.9593908629441623,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001032067576856569,
      "loss": 1.1258,
      "step": 46020
    },
    {
      "epoch": 3.9594768992514844,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010319030512844113,
      "loss": 1.0445,
      "step": 46021
    },
    {
      "epoch": 3.9595629355588056,
      "grad_norm": 1.125,
      "learning_rate": 0.00010317385373181664,
      "loss": 1.0367,
      "step": 46022
    },
    {
      "epoch": 3.9596489718661276,
      "grad_norm": 5.46875,
      "learning_rate": 0.00010315740349583135,
      "loss": 0.9009,
      "step": 46023
    },
    {
      "epoch": 3.9597350081734493,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00010314095442053351,
      "loss": 1.5853,
      "step": 46024
    },
    {
      "epoch": 3.959821044480771,
      "grad_norm": 6.25,
      "learning_rate": 0.00010312450650597128,
      "loss": 1.1486,
      "step": 46025
    },
    {
      "epoch": 3.9599070807880925,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00010310805975219256,
      "loss": 1.3246,
      "step": 46026
    },
    {
      "epoch": 3.959993117095414,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010309161415924572,
      "loss": 1.5641,
      "step": 46027
    },
    {
      "epoch": 3.960079153402736,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00010307516972717856,
      "loss": 1.4279,
      "step": 46028
    },
    {
      "epoch": 3.9601651897100574,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010305872645603937,
      "loss": 1.1538,
      "step": 46029
    },
    {
      "epoch": 3.9602512260173794,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00010304228434587632,
      "loss": 0.5668,
      "step": 46030
    },
    {
      "epoch": 3.960337262324701,
      "grad_norm": 3.03125,
      "learning_rate": 0.00010302584339673721,
      "loss": 1.3995,
      "step": 46031
    },
    {
      "epoch": 3.9604232986320227,
      "grad_norm": 13.5,
      "learning_rate": 0.00010300940360867033,
      "loss": 1.6353,
      "step": 46032
    },
    {
      "epoch": 3.9605093349393443,
      "grad_norm": 1.375,
      "learning_rate": 0.00010299296498172389,
      "loss": 1.6645,
      "step": 46033
    },
    {
      "epoch": 3.960595371246666,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00010297652751594566,
      "loss": 1.8974,
      "step": 46034
    },
    {
      "epoch": 3.960681407553988,
      "grad_norm": 920.0,
      "learning_rate": 0.0001029600912113839,
      "loss": 1.5361,
      "step": 46035
    },
    {
      "epoch": 3.9607674438613096,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00010294365606808676,
      "loss": 1.1243,
      "step": 46036
    },
    {
      "epoch": 3.9608534801686313,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010292722208610206,
      "loss": 1.8106,
      "step": 46037
    },
    {
      "epoch": 3.960939516475953,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010291078926547798,
      "loss": 0.9226,
      "step": 46038
    },
    {
      "epoch": 3.9610255527832745,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001028943576062628,
      "loss": 1.0016,
      "step": 46039
    },
    {
      "epoch": 3.961111589090596,
      "grad_norm": 1.15625,
      "learning_rate": 0.00010287792710850425,
      "loss": 1.0911,
      "step": 46040
    },
    {
      "epoch": 3.9611976253979178,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00010286149777225063,
      "loss": 1.7851,
      "step": 46041
    },
    {
      "epoch": 3.96128366170524,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001028450695975498,
      "loss": 0.8169,
      "step": 46042
    },
    {
      "epoch": 3.9613696980125614,
      "grad_norm": 1.4375,
      "learning_rate": 0.00010282864258444986,
      "loss": 1.0141,
      "step": 46043
    },
    {
      "epoch": 3.961455734319883,
      "grad_norm": 60.25,
      "learning_rate": 0.00010281221673299901,
      "loss": 1.1371,
      "step": 46044
    },
    {
      "epoch": 3.9615417706272047,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00010279579204324513,
      "loss": 1.2358,
      "step": 46045
    },
    {
      "epoch": 3.9616278069345263,
      "grad_norm": 3.78125,
      "learning_rate": 0.00010277936851523623,
      "loss": 1.0616,
      "step": 46046
    },
    {
      "epoch": 3.961713843241848,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001027629461490206,
      "loss": 0.771,
      "step": 46047
    },
    {
      "epoch": 3.9617998795491696,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010274652494464592,
      "loss": 1.1883,
      "step": 46048
    },
    {
      "epoch": 3.9618859158564916,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010273010490216039,
      "loss": 0.6473,
      "step": 46049
    },
    {
      "epoch": 3.9619719521638133,
      "grad_norm": 1.125,
      "learning_rate": 0.00010271368602161208,
      "loss": 1.1413,
      "step": 46050
    },
    {
      "epoch": 3.962057988471135,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010269726830304892,
      "loss": 0.483,
      "step": 46051
    },
    {
      "epoch": 3.9621440247784565,
      "grad_norm": 1480.0,
      "learning_rate": 0.0001026808517465192,
      "loss": 0.8868,
      "step": 46052
    },
    {
      "epoch": 3.962230061085778,
      "grad_norm": 1.7421875,
      "learning_rate": 0.00010266443635207052,
      "loss": 1.0732,
      "step": 46053
    },
    {
      "epoch": 3.9623160973930998,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010264802211975106,
      "loss": 0.7298,
      "step": 46054
    },
    {
      "epoch": 3.9624021337004214,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010263160904960906,
      "loss": 0.6961,
      "step": 46055
    },
    {
      "epoch": 3.9624881700077434,
      "grad_norm": 1.53125,
      "learning_rate": 0.0001026151971416921,
      "loss": 1.9312,
      "step": 46056
    },
    {
      "epoch": 3.962574206315065,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010259878639604858,
      "loss": 1.3992,
      "step": 46057
    },
    {
      "epoch": 3.9626602426223867,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00010258237681272619,
      "loss": 0.9522,
      "step": 46058
    },
    {
      "epoch": 3.9627462789297083,
      "grad_norm": 1.171875,
      "learning_rate": 0.00010256596839177307,
      "loss": 1.6701,
      "step": 46059
    },
    {
      "epoch": 3.96283231523703,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00010254956113323733,
      "loss": 1.1273,
      "step": 46060
    },
    {
      "epoch": 3.9629183515443516,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010253315503716665,
      "loss": 0.7455,
      "step": 46061
    },
    {
      "epoch": 3.963004387851673,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010251675010360922,
      "loss": 1.6302,
      "step": 46062
    },
    {
      "epoch": 3.9630904241589953,
      "grad_norm": 1.171875,
      "learning_rate": 0.00010250034633261307,
      "loss": 0.7084,
      "step": 46063
    },
    {
      "epoch": 3.963176460466317,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010248394372422603,
      "loss": 1.3037,
      "step": 46064
    },
    {
      "epoch": 3.9632624967736385,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010246754227849613,
      "loss": 1.4746,
      "step": 46065
    },
    {
      "epoch": 3.96334853308096,
      "grad_norm": 1.234375,
      "learning_rate": 0.00010245114199547145,
      "loss": 1.1455,
      "step": 46066
    },
    {
      "epoch": 3.9634345693882818,
      "grad_norm": 1.140625,
      "learning_rate": 0.00010243474287519977,
      "loss": 1.0033,
      "step": 46067
    },
    {
      "epoch": 3.9635206056956034,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010241834491772922,
      "loss": 0.7121,
      "step": 46068
    },
    {
      "epoch": 3.963606642002925,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001024019481231076,
      "loss": 0.804,
      "step": 46069
    },
    {
      "epoch": 3.963692678310247,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010238555249138298,
      "loss": 0.8695,
      "step": 46070
    },
    {
      "epoch": 3.9637787146175687,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00010236915802260338,
      "loss": 0.9368,
      "step": 46071
    },
    {
      "epoch": 3.9638647509248903,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010235276471681653,
      "loss": 1.0226,
      "step": 46072
    },
    {
      "epoch": 3.963950787232212,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010233637257407053,
      "loss": 0.7048,
      "step": 46073
    },
    {
      "epoch": 3.9640368235395336,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00010231998159441336,
      "loss": 1.3441,
      "step": 46074
    },
    {
      "epoch": 3.9641228598468556,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010230359177789283,
      "loss": 1.5094,
      "step": 46075
    },
    {
      "epoch": 3.964208896154177,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010228720312455692,
      "loss": 0.8303,
      "step": 46076
    },
    {
      "epoch": 3.964294932461499,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00010227081563445373,
      "loss": 1.4243,
      "step": 46077
    },
    {
      "epoch": 3.9643809687688205,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010225442930763096,
      "loss": 1.3297,
      "step": 46078
    },
    {
      "epoch": 3.964467005076142,
      "grad_norm": 4.375,
      "learning_rate": 0.00010223804414413657,
      "loss": 1.5826,
      "step": 46079
    },
    {
      "epoch": 3.9645530413834638,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010222166014401868,
      "loss": 1.1164,
      "step": 46080
    },
    {
      "epoch": 3.9646390776907854,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010220527730732493,
      "loss": 1.1175,
      "step": 46081
    },
    {
      "epoch": 3.9647251139981075,
      "grad_norm": 1.375,
      "learning_rate": 0.00010218889563410355,
      "loss": 1.5527,
      "step": 46082
    },
    {
      "epoch": 3.9648111503054286,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010217251512440212,
      "loss": 0.5852,
      "step": 46083
    },
    {
      "epoch": 3.9648971866127507,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010215613577826872,
      "loss": 0.6124,
      "step": 46084
    },
    {
      "epoch": 3.9649832229200723,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001021397575957514,
      "loss": 1.199,
      "step": 46085
    },
    {
      "epoch": 3.965069259227394,
      "grad_norm": 13.8125,
      "learning_rate": 0.00010212338057689774,
      "loss": 1.2932,
      "step": 46086
    },
    {
      "epoch": 3.9651552955347156,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010210700472175583,
      "loss": 1.186,
      "step": 46087
    },
    {
      "epoch": 3.965241331842037,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001020906300303735,
      "loss": 0.5286,
      "step": 46088
    },
    {
      "epoch": 3.9653273681493593,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010207425650279872,
      "loss": 1.155,
      "step": 46089
    },
    {
      "epoch": 3.9654134044566804,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010205788413907951,
      "loss": 1.4599,
      "step": 46090
    },
    {
      "epoch": 3.9654994407640025,
      "grad_norm": 14.8125,
      "learning_rate": 0.00010204151293926345,
      "loss": 1.4556,
      "step": 46091
    },
    {
      "epoch": 3.965585477071324,
      "grad_norm": 14.375,
      "learning_rate": 0.00010202514290339853,
      "loss": 1.1549,
      "step": 46092
    },
    {
      "epoch": 3.9656715133786458,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00010200877403153286,
      "loss": 1.3567,
      "step": 46093
    },
    {
      "epoch": 3.9657575496859674,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010199240632371393,
      "loss": 1.6195,
      "step": 46094
    },
    {
      "epoch": 3.965843585993289,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00010197603977998981,
      "loss": 1.0664,
      "step": 46095
    },
    {
      "epoch": 3.965929622300611,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00010195967440040849,
      "loss": 1.9185,
      "step": 46096
    },
    {
      "epoch": 3.9660156586079327,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00010194331018501763,
      "loss": 0.8125,
      "step": 46097
    },
    {
      "epoch": 3.9661016949152543,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010192694713386525,
      "loss": 0.7916,
      "step": 46098
    },
    {
      "epoch": 3.966187731222576,
      "grad_norm": 13.5625,
      "learning_rate": 0.00010191058524699898,
      "loss": 1.2956,
      "step": 46099
    },
    {
      "epoch": 3.9662737675298976,
      "grad_norm": 2.203125,
      "learning_rate": 0.0001018942245244669,
      "loss": 1.0769,
      "step": 46100
    },
    {
      "epoch": 3.966359803837219,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010187786496631685,
      "loss": 0.699,
      "step": 46101
    },
    {
      "epoch": 3.966445840144541,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010186150657259651,
      "loss": 1.1256,
      "step": 46102
    },
    {
      "epoch": 3.966531876451863,
      "grad_norm": 3.078125,
      "learning_rate": 0.00010184514934335382,
      "loss": 1.2155,
      "step": 46103
    },
    {
      "epoch": 3.9666179127591845,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00010182879327863675,
      "loss": 1.0747,
      "step": 46104
    },
    {
      "epoch": 3.966703949066506,
      "grad_norm": 69.5,
      "learning_rate": 0.00010181243837849291,
      "loss": 0.8376,
      "step": 46105
    },
    {
      "epoch": 3.9667899853738278,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00010179608464297024,
      "loss": 1.3717,
      "step": 46106
    },
    {
      "epoch": 3.9668760216811494,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010177973207211672,
      "loss": 1.792,
      "step": 46107
    },
    {
      "epoch": 3.966962057988471,
      "grad_norm": 1.484375,
      "learning_rate": 0.00010176338066597989,
      "loss": 1.7792,
      "step": 46108
    },
    {
      "epoch": 3.9670480942957926,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0001017470304246077,
      "loss": 1.2235,
      "step": 46109
    },
    {
      "epoch": 3.9671341306031147,
      "grad_norm": 7.4375,
      "learning_rate": 0.00010173068134804813,
      "loss": 1.1545,
      "step": 46110
    },
    {
      "epoch": 3.9672201669104363,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00010171433343634879,
      "loss": 0.9899,
      "step": 46111
    },
    {
      "epoch": 3.967306203217758,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001016979866895576,
      "loss": 0.6059,
      "step": 46112
    },
    {
      "epoch": 3.9673922395250796,
      "grad_norm": 1.09375,
      "learning_rate": 0.00010168164110772227,
      "loss": 1.0497,
      "step": 46113
    },
    {
      "epoch": 3.967478275832401,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010166529669089064,
      "loss": 0.5605,
      "step": 46114
    },
    {
      "epoch": 3.967564312139723,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010164895343911068,
      "loss": 1.4017,
      "step": 46115
    },
    {
      "epoch": 3.9676503484470445,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010163261135242991,
      "loss": 1.2402,
      "step": 46116
    },
    {
      "epoch": 3.9677363847543665,
      "grad_norm": 7.6875,
      "learning_rate": 0.00010161627043089627,
      "loss": 0.5291,
      "step": 46117
    },
    {
      "epoch": 3.967822421061688,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010159993067455775,
      "loss": 1.3972,
      "step": 46118
    },
    {
      "epoch": 3.9679084573690098,
      "grad_norm": 70.0,
      "learning_rate": 0.00010158359208346175,
      "loss": 1.6662,
      "step": 46119
    },
    {
      "epoch": 3.9679944936763314,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010156725465765632,
      "loss": 0.8695,
      "step": 46120
    },
    {
      "epoch": 3.968080529983653,
      "grad_norm": 23.625,
      "learning_rate": 0.00010155091839718927,
      "loss": 1.08,
      "step": 46121
    },
    {
      "epoch": 3.9681665662909746,
      "grad_norm": 1.171875,
      "learning_rate": 0.00010153458330210813,
      "loss": 1.1921,
      "step": 46122
    },
    {
      "epoch": 3.9682526025982963,
      "grad_norm": 1.375,
      "learning_rate": 0.00010151824937246101,
      "loss": 1.4214,
      "step": 46123
    },
    {
      "epoch": 3.9683386389056183,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010150191660829522,
      "loss": 1.1425,
      "step": 46124
    },
    {
      "epoch": 3.96842467521294,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010148558500965898,
      "loss": 1.5653,
      "step": 46125
    },
    {
      "epoch": 3.9685107115202616,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001014692545766,
      "loss": 1.6443,
      "step": 46126
    },
    {
      "epoch": 3.968596747827583,
      "grad_norm": 0.984375,
      "learning_rate": 0.00010145292530916583,
      "loss": 0.842,
      "step": 46127
    },
    {
      "epoch": 3.968682784134905,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00010143659720740444,
      "loss": 1.7873,
      "step": 46128
    },
    {
      "epoch": 3.968768820442227,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00010142027027136341,
      "loss": 1.0807,
      "step": 46129
    },
    {
      "epoch": 3.968854856749548,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010140394450109053,
      "loss": 1.1588,
      "step": 46130
    },
    {
      "epoch": 3.96894089305687,
      "grad_norm": 1.21875,
      "learning_rate": 0.00010138761989663375,
      "loss": 1.4105,
      "step": 46131
    },
    {
      "epoch": 3.9690269293641918,
      "grad_norm": 496.0,
      "learning_rate": 0.00010137129645804049,
      "loss": 0.5684,
      "step": 46132
    },
    {
      "epoch": 3.9691129656715134,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00010135497418535872,
      "loss": 1.7611,
      "step": 46133
    },
    {
      "epoch": 3.969199001978835,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010133865307863616,
      "loss": 1.0962,
      "step": 46134
    },
    {
      "epoch": 3.9692850382861566,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00010132233313792044,
      "loss": 1.374,
      "step": 46135
    },
    {
      "epoch": 3.9693710745934787,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010130601436325932,
      "loss": 0.9911,
      "step": 46136
    },
    {
      "epoch": 3.9694571109008,
      "grad_norm": 1.046875,
      "learning_rate": 0.00010128969675470073,
      "loss": 1.1771,
      "step": 46137
    },
    {
      "epoch": 3.969543147208122,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001012733803122921,
      "loss": 1.1361,
      "step": 46138
    },
    {
      "epoch": 3.9696291835154436,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001012570650360814,
      "loss": 0.4894,
      "step": 46139
    },
    {
      "epoch": 3.969715219822765,
      "grad_norm": 17.625,
      "learning_rate": 0.00010124075092611606,
      "loss": 1.0846,
      "step": 46140
    },
    {
      "epoch": 3.969801256130087,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010122443798244402,
      "loss": 0.6719,
      "step": 46141
    },
    {
      "epoch": 3.9698872924374085,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001012081262051131,
      "loss": 1.0375,
      "step": 46142
    },
    {
      "epoch": 3.9699733287447305,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010119181559417068,
      "loss": 1.5061,
      "step": 46143
    },
    {
      "epoch": 3.9700593650520517,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00010117550614966464,
      "loss": 1.289,
      "step": 46144
    },
    {
      "epoch": 3.9701454013593738,
      "grad_norm": 2.5,
      "learning_rate": 0.00010115919787164279,
      "loss": 1.8346,
      "step": 46145
    },
    {
      "epoch": 3.9702314376666954,
      "grad_norm": 1.5,
      "learning_rate": 0.0001011428907601526,
      "loss": 0.9499,
      "step": 46146
    },
    {
      "epoch": 3.970317473974017,
      "grad_norm": 22.0,
      "learning_rate": 0.0001011265848152419,
      "loss": 1.9738,
      "step": 46147
    },
    {
      "epoch": 3.9704035102813386,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00010111028003695844,
      "loss": 0.8959,
      "step": 46148
    },
    {
      "epoch": 3.9704895465886603,
      "grad_norm": 1.4375,
      "learning_rate": 0.00010109397642534973,
      "loss": 1.75,
      "step": 46149
    },
    {
      "epoch": 3.9705755828959823,
      "grad_norm": 1.8046875,
      "learning_rate": 0.00010107767398046358,
      "loss": 1.3013,
      "step": 46150
    },
    {
      "epoch": 3.970661619203304,
      "grad_norm": 1.3125,
      "learning_rate": 0.00010106137270234777,
      "loss": 1.1939,
      "step": 46151
    },
    {
      "epoch": 3.9707476555106256,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00010104507259104973,
      "loss": 0.8621,
      "step": 46152
    },
    {
      "epoch": 3.970833691817947,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010102877364661739,
      "loss": 1.0094,
      "step": 46153
    },
    {
      "epoch": 3.970919728125269,
      "grad_norm": 76.0,
      "learning_rate": 0.00010101247586909812,
      "loss": 1.4255,
      "step": 46154
    },
    {
      "epoch": 3.9710057644325905,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010099617925853977,
      "loss": 1.0843,
      "step": 46155
    },
    {
      "epoch": 3.971091800739912,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00010097988381499018,
      "loss": 0.7188,
      "step": 46156
    },
    {
      "epoch": 3.971177837047234,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010096358953849666,
      "loss": 1.2961,
      "step": 46157
    },
    {
      "epoch": 3.9712638733545558,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00010094729642910699,
      "loss": 1.0932,
      "step": 46158
    },
    {
      "epoch": 3.9713499096618774,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010093100448686903,
      "loss": 0.5628,
      "step": 46159
    },
    {
      "epoch": 3.971435945969199,
      "grad_norm": 1.578125,
      "learning_rate": 0.00010091471371183014,
      "loss": 1.6167,
      "step": 46160
    },
    {
      "epoch": 3.9715219822765206,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001008984241040381,
      "loss": 0.9236,
      "step": 46161
    },
    {
      "epoch": 3.9716080185838423,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00010088213566354054,
      "loss": 1.6579,
      "step": 46162
    },
    {
      "epoch": 3.971694054891164,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00010086584839038516,
      "loss": 0.8393,
      "step": 46163
    },
    {
      "epoch": 3.971780091198486,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00010084956228461962,
      "loss": 0.9689,
      "step": 46164
    },
    {
      "epoch": 3.9718661275058076,
      "grad_norm": 6.15625,
      "learning_rate": 0.00010083327734629133,
      "loss": 1.4061,
      "step": 46165
    },
    {
      "epoch": 3.971952163813129,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001008169935754481,
      "loss": 0.7878,
      "step": 46166
    },
    {
      "epoch": 3.972038200120451,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00010080071097213766,
      "loss": 1.495,
      "step": 46167
    },
    {
      "epoch": 3.9721242364277725,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00010078442953640737,
      "loss": 1.5156,
      "step": 46168
    },
    {
      "epoch": 3.972210272735094,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00010076814926830514,
      "loss": 0.8383,
      "step": 46169
    },
    {
      "epoch": 3.9722963090424157,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010075187016787824,
      "loss": 0.6495,
      "step": 46170
    },
    {
      "epoch": 3.9723823453497378,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00010073559223517453,
      "loss": 0.6712,
      "step": 46171
    },
    {
      "epoch": 3.9724683816570594,
      "grad_norm": 1.328125,
      "learning_rate": 0.00010071931547024165,
      "loss": 1.3448,
      "step": 46172
    },
    {
      "epoch": 3.972554417964381,
      "grad_norm": 43.75,
      "learning_rate": 0.00010070303987312695,
      "loss": 1.1547,
      "step": 46173
    },
    {
      "epoch": 3.9726404542717026,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010068676544387828,
      "loss": 0.5627,
      "step": 46174
    },
    {
      "epoch": 3.9727264905790243,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00010067049218254327,
      "loss": 1.4545,
      "step": 46175
    },
    {
      "epoch": 3.972812526886346,
      "grad_norm": 1.296875,
      "learning_rate": 0.00010065422008916925,
      "loss": 1.7126,
      "step": 46176
    },
    {
      "epoch": 3.9728985631936675,
      "grad_norm": 62.0,
      "learning_rate": 0.00010063794916380392,
      "loss": 0.3847,
      "step": 46177
    },
    {
      "epoch": 3.9729845995009896,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00010062167940649513,
      "loss": 0.4606,
      "step": 46178
    },
    {
      "epoch": 3.973070635808311,
      "grad_norm": 1.171875,
      "learning_rate": 0.00010060541081729007,
      "loss": 1.14,
      "step": 46179
    },
    {
      "epoch": 3.973156672115633,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001005891433962367,
      "loss": 0.6929,
      "step": 46180
    },
    {
      "epoch": 3.9732427084229545,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001005728771433822,
      "loss": 0.6735,
      "step": 46181
    },
    {
      "epoch": 3.973328744730276,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010055661205877442,
      "loss": 1.7558,
      "step": 46182
    },
    {
      "epoch": 3.9734147810375977,
      "grad_norm": 12.125,
      "learning_rate": 0.00010054034814246094,
      "loss": 0.9288,
      "step": 46183
    },
    {
      "epoch": 3.9735008173449193,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010052408539448916,
      "loss": 0.5957,
      "step": 46184
    },
    {
      "epoch": 3.9735868536522414,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00010050782381490669,
      "loss": 1.645,
      "step": 46185
    },
    {
      "epoch": 3.973672889959563,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001004915634037613,
      "loss": 1.3747,
      "step": 46186
    },
    {
      "epoch": 3.9737589262668847,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010047530416110024,
      "loss": 0.8916,
      "step": 46187
    },
    {
      "epoch": 3.9738449625742063,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001004590460869712,
      "loss": 1.5657,
      "step": 46188
    },
    {
      "epoch": 3.973930998881528,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00010044278918142191,
      "loss": 0.8466,
      "step": 46189
    },
    {
      "epoch": 3.97401703518885,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001004265334444996,
      "loss": 0.6076,
      "step": 46190
    },
    {
      "epoch": 3.974103071496171,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010041027887625198,
      "loss": 1.5539,
      "step": 46191
    },
    {
      "epoch": 3.974189107803493,
      "grad_norm": 2.078125,
      "learning_rate": 0.00010039402547672671,
      "loss": 1.7359,
      "step": 46192
    },
    {
      "epoch": 3.974275144110815,
      "grad_norm": 1.171875,
      "learning_rate": 0.00010037777324597103,
      "loss": 1.0609,
      "step": 46193
    },
    {
      "epoch": 3.9743611804181365,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010036152218403283,
      "loss": 0.9323,
      "step": 46194
    },
    {
      "epoch": 3.974447216725458,
      "grad_norm": 0.875,
      "learning_rate": 0.0001003452722909593,
      "loss": 0.9581,
      "step": 46195
    },
    {
      "epoch": 3.9745332530327797,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00010032902356679807,
      "loss": 0.8786,
      "step": 46196
    },
    {
      "epoch": 3.974619289340102,
      "grad_norm": 1.140625,
      "learning_rate": 0.00010031277601159694,
      "loss": 0.7574,
      "step": 46197
    },
    {
      "epoch": 3.974705325647423,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010029652962540286,
      "loss": 1.5497,
      "step": 46198
    },
    {
      "epoch": 3.974791361954745,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010028028440826398,
      "loss": 0.9064,
      "step": 46199
    },
    {
      "epoch": 3.9748773982620667,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00010026404036022741,
      "loss": 0.9598,
      "step": 46200
    },
    {
      "epoch": 3.9749634345693883,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010024779748134077,
      "loss": 1.0205,
      "step": 46201
    },
    {
      "epoch": 3.97504947087671,
      "grad_norm": 1.0,
      "learning_rate": 0.00010023155577165172,
      "loss": 1.195,
      "step": 46202
    },
    {
      "epoch": 3.9751355071840315,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010021531523120747,
      "loss": 0.7163,
      "step": 46203
    },
    {
      "epoch": 3.9752215434913536,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010019907586005567,
      "loss": 0.6757,
      "step": 46204
    },
    {
      "epoch": 3.9753075797986748,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010018283765824393,
      "loss": 0.9454,
      "step": 46205
    },
    {
      "epoch": 3.975393616105997,
      "grad_norm": 1.109375,
      "learning_rate": 0.00010016660062581951,
      "loss": 0.7938,
      "step": 46206
    },
    {
      "epoch": 3.9754796524133185,
      "grad_norm": 1.84375,
      "learning_rate": 0.00010015036476283002,
      "loss": 1.6929,
      "step": 46207
    },
    {
      "epoch": 3.97556568872064,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001001341300693231,
      "loss": 0.8876,
      "step": 46208
    },
    {
      "epoch": 3.9756517250279617,
      "grad_norm": 5.28125,
      "learning_rate": 0.00010011789654534592,
      "loss": 1.2713,
      "step": 46209
    },
    {
      "epoch": 3.9757377613352833,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001001016641909463,
      "loss": 0.9666,
      "step": 46210
    },
    {
      "epoch": 3.9758237976426054,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00010008543300617134,
      "loss": 1.423,
      "step": 46211
    },
    {
      "epoch": 3.975909833949927,
      "grad_norm": 94.0,
      "learning_rate": 0.00010006920299106876,
      "loss": 1.1119,
      "step": 46212
    },
    {
      "epoch": 3.9759958702572487,
      "grad_norm": 5.34375,
      "learning_rate": 0.0001000529741456861,
      "loss": 1.1284,
      "step": 46213
    },
    {
      "epoch": 3.9760819065645703,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00010003674647007055,
      "loss": 1.365,
      "step": 46214
    },
    {
      "epoch": 3.976167942871892,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00010002051996426976,
      "loss": 1.6484,
      "step": 46215
    },
    {
      "epoch": 3.9762539791792135,
      "grad_norm": 1.3125,
      "learning_rate": 0.00010000429462833121,
      "loss": 0.6566,
      "step": 46216
    },
    {
      "epoch": 3.976340015486535,
      "grad_norm": 0.8828125,
      "learning_rate": 9.998807046230224e-05,
      "loss": 0.5825,
      "step": 46217
    },
    {
      "epoch": 3.976426051793857,
      "grad_norm": 1.1640625,
      "learning_rate": 9.997184746623039e-05,
      "loss": 1.1417,
      "step": 46218
    },
    {
      "epoch": 3.976512088101179,
      "grad_norm": 1.1640625,
      "learning_rate": 9.99556256401632e-05,
      "loss": 0.7127,
      "step": 46219
    },
    {
      "epoch": 3.9765981244085005,
      "grad_norm": 1.046875,
      "learning_rate": 9.993940498414783e-05,
      "loss": 0.9186,
      "step": 46220
    },
    {
      "epoch": 3.976684160715822,
      "grad_norm": 1.296875,
      "learning_rate": 9.992318549823203e-05,
      "loss": 1.0919,
      "step": 46221
    },
    {
      "epoch": 3.9767701970231437,
      "grad_norm": 0.98828125,
      "learning_rate": 9.9906967182463e-05,
      "loss": 1.1489,
      "step": 46222
    },
    {
      "epoch": 3.9768562333304653,
      "grad_norm": 1.0703125,
      "learning_rate": 9.989075003688825e-05,
      "loss": 0.8482,
      "step": 46223
    },
    {
      "epoch": 3.976942269637787,
      "grad_norm": 1.171875,
      "learning_rate": 9.987453406155538e-05,
      "loss": 1.4038,
      "step": 46224
    },
    {
      "epoch": 3.977028305945109,
      "grad_norm": 1.109375,
      "learning_rate": 9.985831925651156e-05,
      "loss": 1.0484,
      "step": 46225
    },
    {
      "epoch": 3.9771143422524307,
      "grad_norm": 1.3671875,
      "learning_rate": 9.984210562180435e-05,
      "loss": 1.247,
      "step": 46226
    },
    {
      "epoch": 3.9772003785597523,
      "grad_norm": 0.921875,
      "learning_rate": 9.982589315748126e-05,
      "loss": 1.3706,
      "step": 46227
    },
    {
      "epoch": 3.977286414867074,
      "grad_norm": 1.1796875,
      "learning_rate": 9.980968186358947e-05,
      "loss": 1.6835,
      "step": 46228
    },
    {
      "epoch": 3.9773724511743955,
      "grad_norm": 0.78515625,
      "learning_rate": 9.979347174017656e-05,
      "loss": 1.1275,
      "step": 46229
    },
    {
      "epoch": 3.977458487481717,
      "grad_norm": 102.5,
      "learning_rate": 9.977726278728999e-05,
      "loss": 1.2059,
      "step": 46230
    },
    {
      "epoch": 3.9775445237890388,
      "grad_norm": 0.94140625,
      "learning_rate": 9.976105500497696e-05,
      "loss": 1.2367,
      "step": 46231
    },
    {
      "epoch": 3.977630560096361,
      "grad_norm": 1.2421875,
      "learning_rate": 9.974484839328501e-05,
      "loss": 1.5724,
      "step": 46232
    },
    {
      "epoch": 3.9777165964036825,
      "grad_norm": 1.5546875,
      "learning_rate": 9.97286429522617e-05,
      "loss": 1.5208,
      "step": 46233
    },
    {
      "epoch": 3.977802632711004,
      "grad_norm": 1.3203125,
      "learning_rate": 9.971243868195412e-05,
      "loss": 1.7161,
      "step": 46234
    },
    {
      "epoch": 3.9778886690183257,
      "grad_norm": 1.1796875,
      "learning_rate": 9.969623558240976e-05,
      "loss": 0.9032,
      "step": 46235
    },
    {
      "epoch": 3.9779747053256473,
      "grad_norm": 1.1328125,
      "learning_rate": 9.968003365367606e-05,
      "loss": 0.6894,
      "step": 46236
    },
    {
      "epoch": 3.978060741632969,
      "grad_norm": 2.25,
      "learning_rate": 9.966383289580045e-05,
      "loss": 1.1318,
      "step": 46237
    },
    {
      "epoch": 3.9781467779402906,
      "grad_norm": 1.2734375,
      "learning_rate": 9.964763330883036e-05,
      "loss": 1.3567,
      "step": 46238
    },
    {
      "epoch": 3.9782328142476127,
      "grad_norm": 1.2421875,
      "learning_rate": 9.963143489281301e-05,
      "loss": 0.9992,
      "step": 46239
    },
    {
      "epoch": 3.9783188505549343,
      "grad_norm": 2.21875,
      "learning_rate": 9.961523764779589e-05,
      "loss": 0.9866,
      "step": 46240
    },
    {
      "epoch": 3.978404886862256,
      "grad_norm": 0.90234375,
      "learning_rate": 9.959904157382627e-05,
      "loss": 0.8995,
      "step": 46241
    },
    {
      "epoch": 3.9784909231695775,
      "grad_norm": 1.6484375,
      "learning_rate": 9.958284667095153e-05,
      "loss": 1.0146,
      "step": 46242
    },
    {
      "epoch": 3.978576959476899,
      "grad_norm": 0.75390625,
      "learning_rate": 9.956665293921924e-05,
      "loss": 0.5765,
      "step": 46243
    },
    {
      "epoch": 3.978662995784221,
      "grad_norm": 1.859375,
      "learning_rate": 9.955046037867649e-05,
      "loss": 1.3562,
      "step": 46244
    },
    {
      "epoch": 3.9787490320915424,
      "grad_norm": 1.25,
      "learning_rate": 9.953426898937073e-05,
      "loss": 0.8859,
      "step": 46245
    },
    {
      "epoch": 3.9788350683988645,
      "grad_norm": 1.1484375,
      "learning_rate": 9.95180787713495e-05,
      "loss": 0.9461,
      "step": 46246
    },
    {
      "epoch": 3.978921104706186,
      "grad_norm": 1.15625,
      "learning_rate": 9.950188972465984e-05,
      "loss": 1.7716,
      "step": 46247
    },
    {
      "epoch": 3.9790071410135077,
      "grad_norm": 1.0078125,
      "learning_rate": 9.94857018493493e-05,
      "loss": 1.004,
      "step": 46248
    },
    {
      "epoch": 3.9790931773208293,
      "grad_norm": 1.21875,
      "learning_rate": 9.946951514546531e-05,
      "loss": 1.2725,
      "step": 46249
    },
    {
      "epoch": 3.979179213628151,
      "grad_norm": 1.0234375,
      "learning_rate": 9.945332961305497e-05,
      "loss": 1.1415,
      "step": 46250
    },
    {
      "epoch": 3.979265249935473,
      "grad_norm": 1.75,
      "learning_rate": 9.943714525216585e-05,
      "loss": 0.9076,
      "step": 46251
    },
    {
      "epoch": 3.979351286242794,
      "grad_norm": 1.359375,
      "learning_rate": 9.9420962062845e-05,
      "loss": 1.5702,
      "step": 46252
    },
    {
      "epoch": 3.9794373225501163,
      "grad_norm": 2.859375,
      "learning_rate": 9.940478004513997e-05,
      "loss": 1.8543,
      "step": 46253
    },
    {
      "epoch": 3.979523358857438,
      "grad_norm": 1.2265625,
      "learning_rate": 9.938859919909821e-05,
      "loss": 1.4454,
      "step": 46254
    },
    {
      "epoch": 3.9796093951647595,
      "grad_norm": 0.85546875,
      "learning_rate": 9.937241952476672e-05,
      "loss": 0.716,
      "step": 46255
    },
    {
      "epoch": 3.979695431472081,
      "grad_norm": 1.3203125,
      "learning_rate": 9.935624102219298e-05,
      "loss": 1.5623,
      "step": 46256
    },
    {
      "epoch": 3.979781467779403,
      "grad_norm": 1.2265625,
      "learning_rate": 9.934006369142446e-05,
      "loss": 0.8175,
      "step": 46257
    },
    {
      "epoch": 3.979867504086725,
      "grad_norm": 1.1171875,
      "learning_rate": 9.932388753250815e-05,
      "loss": 0.7941,
      "step": 46258
    },
    {
      "epoch": 3.979953540394046,
      "grad_norm": 1.0859375,
      "learning_rate": 9.93077125454916e-05,
      "loss": 0.7145,
      "step": 46259
    },
    {
      "epoch": 3.980039576701368,
      "grad_norm": 1.3203125,
      "learning_rate": 9.929153873042218e-05,
      "loss": 1.2452,
      "step": 46260
    },
    {
      "epoch": 3.9801256130086897,
      "grad_norm": 1.4296875,
      "learning_rate": 9.927536608734695e-05,
      "loss": 1.011,
      "step": 46261
    },
    {
      "epoch": 3.9802116493160113,
      "grad_norm": 1.2890625,
      "learning_rate": 9.925919461631333e-05,
      "loss": 0.8803,
      "step": 46262
    },
    {
      "epoch": 3.980297685623333,
      "grad_norm": 1.125,
      "learning_rate": 9.924302431736875e-05,
      "loss": 1.8194,
      "step": 46263
    },
    {
      "epoch": 3.9803837219306546,
      "grad_norm": 1.3125,
      "learning_rate": 9.922685519056024e-05,
      "loss": 1.3599,
      "step": 46264
    },
    {
      "epoch": 3.9804697582379767,
      "grad_norm": 1.1328125,
      "learning_rate": 9.921068723593535e-05,
      "loss": 1.2416,
      "step": 46265
    },
    {
      "epoch": 3.980555794545298,
      "grad_norm": 1.0078125,
      "learning_rate": 9.919452045354116e-05,
      "loss": 0.9695,
      "step": 46266
    },
    {
      "epoch": 3.98064183085262,
      "grad_norm": 1.109375,
      "learning_rate": 9.917835484342502e-05,
      "loss": 1.1945,
      "step": 46267
    },
    {
      "epoch": 3.9807278671599415,
      "grad_norm": 7.6875,
      "learning_rate": 9.91621904056344e-05,
      "loss": 0.8158,
      "step": 46268
    },
    {
      "epoch": 3.980813903467263,
      "grad_norm": 1.0546875,
      "learning_rate": 9.914602714021625e-05,
      "loss": 0.8677,
      "step": 46269
    },
    {
      "epoch": 3.980899939774585,
      "grad_norm": 1212416.0,
      "learning_rate": 9.912986504721805e-05,
      "loss": 1.3912,
      "step": 46270
    },
    {
      "epoch": 3.9809859760819064,
      "grad_norm": 0.9140625,
      "learning_rate": 9.911370412668713e-05,
      "loss": 1.1055,
      "step": 46271
    },
    {
      "epoch": 3.9810720123892285,
      "grad_norm": 1.484375,
      "learning_rate": 9.909754437867036e-05,
      "loss": 1.0834,
      "step": 46272
    },
    {
      "epoch": 3.98115804869655,
      "grad_norm": 0.828125,
      "learning_rate": 9.908138580321562e-05,
      "loss": 0.8083,
      "step": 46273
    },
    {
      "epoch": 3.9812440850038717,
      "grad_norm": 1.2265625,
      "learning_rate": 9.906522840036969e-05,
      "loss": 1.0652,
      "step": 46274
    },
    {
      "epoch": 3.9813301213111933,
      "grad_norm": 2.15625,
      "learning_rate": 9.904907217018e-05,
      "loss": 0.7968,
      "step": 46275
    },
    {
      "epoch": 3.981416157618515,
      "grad_norm": 1.4765625,
      "learning_rate": 9.90329171126939e-05,
      "loss": 0.9747,
      "step": 46276
    },
    {
      "epoch": 3.9815021939258366,
      "grad_norm": 1.359375,
      "learning_rate": 9.901676322795844e-05,
      "loss": 1.8113,
      "step": 46277
    },
    {
      "epoch": 3.981588230233158,
      "grad_norm": 1.375,
      "learning_rate": 9.900061051602094e-05,
      "loss": 1.3688,
      "step": 46278
    },
    {
      "epoch": 3.9816742665404803,
      "grad_norm": 0.94140625,
      "learning_rate": 9.898445897692881e-05,
      "loss": 1.2222,
      "step": 46279
    },
    {
      "epoch": 3.981760302847802,
      "grad_norm": 1.0,
      "learning_rate": 9.896830861072903e-05,
      "loss": 1.0389,
      "step": 46280
    },
    {
      "epoch": 3.9818463391551235,
      "grad_norm": 1.7890625,
      "learning_rate": 9.895215941746905e-05,
      "loss": 1.783,
      "step": 46281
    },
    {
      "epoch": 3.981932375462445,
      "grad_norm": 0.859375,
      "learning_rate": 9.893601139719588e-05,
      "loss": 0.5266,
      "step": 46282
    },
    {
      "epoch": 3.982018411769767,
      "grad_norm": 1.109375,
      "learning_rate": 9.891986454995689e-05,
      "loss": 1.017,
      "step": 46283
    },
    {
      "epoch": 3.9821044480770884,
      "grad_norm": 1.1484375,
      "learning_rate": 9.890371887579941e-05,
      "loss": 1.3403,
      "step": 46284
    },
    {
      "epoch": 3.98219048438441,
      "grad_norm": 3.625,
      "learning_rate": 9.888757437477042e-05,
      "loss": 0.6349,
      "step": 46285
    },
    {
      "epoch": 3.982276520691732,
      "grad_norm": 2.40625,
      "learning_rate": 9.88714310469172e-05,
      "loss": 0.6303,
      "step": 46286
    },
    {
      "epoch": 3.9823625569990537,
      "grad_norm": 1.015625,
      "learning_rate": 9.885528889228724e-05,
      "loss": 0.6914,
      "step": 46287
    },
    {
      "epoch": 3.9824485933063754,
      "grad_norm": 1.8828125,
      "learning_rate": 9.883914791092741e-05,
      "loss": 1.619,
      "step": 46288
    },
    {
      "epoch": 3.982534629613697,
      "grad_norm": 1.1484375,
      "learning_rate": 9.882300810288497e-05,
      "loss": 0.9223,
      "step": 46289
    },
    {
      "epoch": 3.9826206659210186,
      "grad_norm": 1.4765625,
      "learning_rate": 9.880686946820744e-05,
      "loss": 1.1768,
      "step": 46290
    },
    {
      "epoch": 3.9827067022283402,
      "grad_norm": 1.203125,
      "learning_rate": 9.879073200694161e-05,
      "loss": 1.6053,
      "step": 46291
    },
    {
      "epoch": 3.982792738535662,
      "grad_norm": 5.96875,
      "learning_rate": 9.877459571913494e-05,
      "loss": 1.0793,
      "step": 46292
    },
    {
      "epoch": 3.982878774842984,
      "grad_norm": 1.0625,
      "learning_rate": 9.875846060483451e-05,
      "loss": 1.1878,
      "step": 46293
    },
    {
      "epoch": 3.9829648111503055,
      "grad_norm": 0.703125,
      "learning_rate": 9.874232666408745e-05,
      "loss": 0.2986,
      "step": 46294
    },
    {
      "epoch": 3.983050847457627,
      "grad_norm": 0.7734375,
      "learning_rate": 9.872619389694126e-05,
      "loss": 1.5198,
      "step": 46295
    },
    {
      "epoch": 3.983136883764949,
      "grad_norm": 1.0390625,
      "learning_rate": 9.87100623034427e-05,
      "loss": 0.6528,
      "step": 46296
    },
    {
      "epoch": 3.9832229200722704,
      "grad_norm": 0.97265625,
      "learning_rate": 9.869393188363924e-05,
      "loss": 1.0257,
      "step": 46297
    },
    {
      "epoch": 3.983308956379592,
      "grad_norm": 1.0546875,
      "learning_rate": 9.867780263757808e-05,
      "loss": 1.0821,
      "step": 46298
    },
    {
      "epoch": 3.9833949926869137,
      "grad_norm": 1.1328125,
      "learning_rate": 9.866167456530612e-05,
      "loss": 1.6331,
      "step": 46299
    },
    {
      "epoch": 3.9834810289942357,
      "grad_norm": 1.1328125,
      "learning_rate": 9.864554766687073e-05,
      "loss": 1.3704,
      "step": 46300
    },
    {
      "epoch": 3.9835670653015574,
      "grad_norm": 1.2265625,
      "learning_rate": 9.862942194231923e-05,
      "loss": 1.0571,
      "step": 46301
    },
    {
      "epoch": 3.983653101608879,
      "grad_norm": 1.234375,
      "learning_rate": 9.861329739169844e-05,
      "loss": 1.8508,
      "step": 46302
    },
    {
      "epoch": 3.9837391379162006,
      "grad_norm": 2512.0,
      "learning_rate": 9.859717401505569e-05,
      "loss": 1.2408,
      "step": 46303
    },
    {
      "epoch": 3.9838251742235222,
      "grad_norm": 2.046875,
      "learning_rate": 9.858105181243826e-05,
      "loss": 1.6795,
      "step": 46304
    },
    {
      "epoch": 3.9839112105308443,
      "grad_norm": 1.328125,
      "learning_rate": 9.856493078389306e-05,
      "loss": 0.857,
      "step": 46305
    },
    {
      "epoch": 3.9839972468381655,
      "grad_norm": 27.125,
      "learning_rate": 9.854881092946749e-05,
      "loss": 0.6266,
      "step": 46306
    },
    {
      "epoch": 3.9840832831454875,
      "grad_norm": 0.67578125,
      "learning_rate": 9.853269224920846e-05,
      "loss": 0.3982,
      "step": 46307
    },
    {
      "epoch": 3.984169319452809,
      "grad_norm": 1.2890625,
      "learning_rate": 9.851657474316322e-05,
      "loss": 1.7758,
      "step": 46308
    },
    {
      "epoch": 3.984255355760131,
      "grad_norm": 1.1875,
      "learning_rate": 9.850045841137889e-05,
      "loss": 0.889,
      "step": 46309
    },
    {
      "epoch": 3.9843413920674524,
      "grad_norm": 0.98828125,
      "learning_rate": 9.848434325390265e-05,
      "loss": 0.701,
      "step": 46310
    },
    {
      "epoch": 3.984427428374774,
      "grad_norm": 1.546875,
      "learning_rate": 9.846822927078176e-05,
      "loss": 1.2597,
      "step": 46311
    },
    {
      "epoch": 3.984513464682096,
      "grad_norm": 1.3203125,
      "learning_rate": 9.845211646206304e-05,
      "loss": 0.9701,
      "step": 46312
    },
    {
      "epoch": 3.9845995009894173,
      "grad_norm": 1.2890625,
      "learning_rate": 9.84360048277938e-05,
      "loss": 1.1999,
      "step": 46313
    },
    {
      "epoch": 3.9846855372967394,
      "grad_norm": 1.0625,
      "learning_rate": 9.841989436802129e-05,
      "loss": 1.4007,
      "step": 46314
    },
    {
      "epoch": 3.984771573604061,
      "grad_norm": 1.109375,
      "learning_rate": 9.840378508279235e-05,
      "loss": 0.8691,
      "step": 46315
    },
    {
      "epoch": 3.9848576099113826,
      "grad_norm": 1.25,
      "learning_rate": 9.838767697215423e-05,
      "loss": 1.27,
      "step": 46316
    },
    {
      "epoch": 3.9849436462187042,
      "grad_norm": 1.21875,
      "learning_rate": 9.837157003615415e-05,
      "loss": 0.8423,
      "step": 46317
    },
    {
      "epoch": 3.985029682526026,
      "grad_norm": 1.234375,
      "learning_rate": 9.8355464274839e-05,
      "loss": 1.6251,
      "step": 46318
    },
    {
      "epoch": 3.985115718833348,
      "grad_norm": 0.9296875,
      "learning_rate": 9.833935968825602e-05,
      "loss": 0.7965,
      "step": 46319
    },
    {
      "epoch": 3.985201755140669,
      "grad_norm": 21.375,
      "learning_rate": 9.832325627645238e-05,
      "loss": 1.0382,
      "step": 46320
    },
    {
      "epoch": 3.985287791447991,
      "grad_norm": 2.46875,
      "learning_rate": 9.830715403947504e-05,
      "loss": 1.2281,
      "step": 46321
    },
    {
      "epoch": 3.985373827755313,
      "grad_norm": 1.3515625,
      "learning_rate": 9.829105297737123e-05,
      "loss": 1.4685,
      "step": 46322
    },
    {
      "epoch": 3.9854598640626344,
      "grad_norm": 0.8359375,
      "learning_rate": 9.82749530901878e-05,
      "loss": 0.4832,
      "step": 46323
    },
    {
      "epoch": 3.985545900369956,
      "grad_norm": 1.0078125,
      "learning_rate": 9.825885437797205e-05,
      "loss": 1.1453,
      "step": 46324
    },
    {
      "epoch": 3.9856319366772777,
      "grad_norm": 1.3828125,
      "learning_rate": 9.824275684077117e-05,
      "loss": 1.3052,
      "step": 46325
    },
    {
      "epoch": 3.9857179729845997,
      "grad_norm": 1.140625,
      "learning_rate": 9.822666047863187e-05,
      "loss": 0.891,
      "step": 46326
    },
    {
      "epoch": 3.9858040092919214,
      "grad_norm": 0.9453125,
      "learning_rate": 9.821056529160156e-05,
      "loss": 0.775,
      "step": 46327
    },
    {
      "epoch": 3.985890045599243,
      "grad_norm": 7.3125,
      "learning_rate": 9.819447127972725e-05,
      "loss": 0.9176,
      "step": 46328
    },
    {
      "epoch": 3.9859760819065646,
      "grad_norm": 0.85546875,
      "learning_rate": 9.817837844305583e-05,
      "loss": 0.4772,
      "step": 46329
    },
    {
      "epoch": 3.9860621182138862,
      "grad_norm": 0.90625,
      "learning_rate": 9.816228678163452e-05,
      "loss": 0.9252,
      "step": 46330
    },
    {
      "epoch": 3.986148154521208,
      "grad_norm": 1.3046875,
      "learning_rate": 9.814619629551053e-05,
      "loss": 1.4278,
      "step": 46331
    },
    {
      "epoch": 3.9862341908285295,
      "grad_norm": 0.90234375,
      "learning_rate": 9.813010698473063e-05,
      "loss": 1.0544,
      "step": 46332
    },
    {
      "epoch": 3.9863202271358515,
      "grad_norm": 1.609375,
      "learning_rate": 9.811401884934212e-05,
      "loss": 2.1278,
      "step": 46333
    },
    {
      "epoch": 3.986406263443173,
      "grad_norm": 0.98828125,
      "learning_rate": 9.809793188939182e-05,
      "loss": 1.0562,
      "step": 46334
    },
    {
      "epoch": 3.986492299750495,
      "grad_norm": 0.94140625,
      "learning_rate": 9.808184610492687e-05,
      "loss": 0.9994,
      "step": 46335
    },
    {
      "epoch": 3.9865783360578164,
      "grad_norm": 2.0,
      "learning_rate": 9.806576149599449e-05,
      "loss": 0.8154,
      "step": 46336
    },
    {
      "epoch": 3.986664372365138,
      "grad_norm": 8.25,
      "learning_rate": 9.804967806264149e-05,
      "loss": 1.4648,
      "step": 46337
    },
    {
      "epoch": 3.9867504086724597,
      "grad_norm": 1.390625,
      "learning_rate": 9.8033595804915e-05,
      "loss": 0.918,
      "step": 46338
    },
    {
      "epoch": 3.9868364449797813,
      "grad_norm": 1.2421875,
      "learning_rate": 9.801751472286213e-05,
      "loss": 1.1316,
      "step": 46339
    },
    {
      "epoch": 3.9869224812871034,
      "grad_norm": 0.9609375,
      "learning_rate": 9.800143481652979e-05,
      "loss": 0.9603,
      "step": 46340
    },
    {
      "epoch": 3.987008517594425,
      "grad_norm": 1.0,
      "learning_rate": 9.798535608596504e-05,
      "loss": 1.0164,
      "step": 46341
    },
    {
      "epoch": 3.9870945539017466,
      "grad_norm": 0.9609375,
      "learning_rate": 9.796927853121507e-05,
      "loss": 1.2658,
      "step": 46342
    },
    {
      "epoch": 3.9871805902090682,
      "grad_norm": 1.4375,
      "learning_rate": 9.795320215232667e-05,
      "loss": 0.9728,
      "step": 46343
    },
    {
      "epoch": 3.98726662651639,
      "grad_norm": 0.93359375,
      "learning_rate": 9.793712694934692e-05,
      "loss": 1.1597,
      "step": 46344
    },
    {
      "epoch": 3.9873526628237115,
      "grad_norm": 1.546875,
      "learning_rate": 9.7921052922323e-05,
      "loss": 1.4204,
      "step": 46345
    },
    {
      "epoch": 3.987438699131033,
      "grad_norm": 1.015625,
      "learning_rate": 9.790498007130156e-05,
      "loss": 0.4037,
      "step": 46346
    },
    {
      "epoch": 3.987524735438355,
      "grad_norm": 0.91015625,
      "learning_rate": 9.788890839633013e-05,
      "loss": 2.2022,
      "step": 46347
    },
    {
      "epoch": 3.987610771745677,
      "grad_norm": 1.0703125,
      "learning_rate": 9.787283789745526e-05,
      "loss": 1.2828,
      "step": 46348
    },
    {
      "epoch": 3.9876968080529984,
      "grad_norm": 1.796875,
      "learning_rate": 9.78567685747242e-05,
      "loss": 0.9637,
      "step": 46349
    },
    {
      "epoch": 3.98778284436032,
      "grad_norm": 1.3203125,
      "learning_rate": 9.784070042818394e-05,
      "loss": 1.6327,
      "step": 46350
    },
    {
      "epoch": 3.9878688806676417,
      "grad_norm": 0.90234375,
      "learning_rate": 9.782463345788134e-05,
      "loss": 0.9002,
      "step": 46351
    },
    {
      "epoch": 3.9879549169749633,
      "grad_norm": 1.109375,
      "learning_rate": 9.780856766386353e-05,
      "loss": 1.8558,
      "step": 46352
    },
    {
      "epoch": 3.988040953282285,
      "grad_norm": 57.0,
      "learning_rate": 9.779250304617737e-05,
      "loss": 1.3121,
      "step": 46353
    },
    {
      "epoch": 3.988126989589607,
      "grad_norm": 0.97265625,
      "learning_rate": 9.777643960486987e-05,
      "loss": 0.797,
      "step": 46354
    },
    {
      "epoch": 3.9882130258969286,
      "grad_norm": 1.1875,
      "learning_rate": 9.776037733998822e-05,
      "loss": 1.4085,
      "step": 46355
    },
    {
      "epoch": 3.9882990622042502,
      "grad_norm": 1.0390625,
      "learning_rate": 9.774431625157909e-05,
      "loss": 1.1083,
      "step": 46356
    },
    {
      "epoch": 3.988385098511572,
      "grad_norm": 1.4453125,
      "learning_rate": 9.772825633968962e-05,
      "loss": 1.376,
      "step": 46357
    },
    {
      "epoch": 3.9884711348188935,
      "grad_norm": 4.75,
      "learning_rate": 9.771219760436683e-05,
      "loss": 1.0534,
      "step": 46358
    },
    {
      "epoch": 3.988557171126215,
      "grad_norm": 0.98828125,
      "learning_rate": 9.769614004565757e-05,
      "loss": 0.9759,
      "step": 46359
    },
    {
      "epoch": 3.9886432074335367,
      "grad_norm": 1.5078125,
      "learning_rate": 9.768008366360881e-05,
      "loss": 1.1929,
      "step": 46360
    },
    {
      "epoch": 3.988729243740859,
      "grad_norm": 1.71875,
      "learning_rate": 9.766402845826772e-05,
      "loss": 1.196,
      "step": 46361
    },
    {
      "epoch": 3.9888152800481804,
      "grad_norm": 1.21875,
      "learning_rate": 9.764797442968093e-05,
      "loss": 1.2681,
      "step": 46362
    },
    {
      "epoch": 3.988901316355502,
      "grad_norm": 1.0859375,
      "learning_rate": 9.763192157789574e-05,
      "loss": 1.1589,
      "step": 46363
    },
    {
      "epoch": 3.9889873526628237,
      "grad_norm": 1.5,
      "learning_rate": 9.761586990295873e-05,
      "loss": 1.3789,
      "step": 46364
    },
    {
      "epoch": 3.9890733889701453,
      "grad_norm": 1.234375,
      "learning_rate": 9.759981940491707e-05,
      "loss": 1.4962,
      "step": 46365
    },
    {
      "epoch": 3.9891594252774674,
      "grad_norm": 0.91015625,
      "learning_rate": 9.75837700838178e-05,
      "loss": 1.1523,
      "step": 46366
    },
    {
      "epoch": 3.9892454615847885,
      "grad_norm": 2.921875,
      "learning_rate": 9.756772193970764e-05,
      "loss": 0.646,
      "step": 46367
    },
    {
      "epoch": 3.9893314978921106,
      "grad_norm": 0.875,
      "learning_rate": 9.755167497263361e-05,
      "loss": 0.9175,
      "step": 46368
    },
    {
      "epoch": 3.9894175341994322,
      "grad_norm": 1.3671875,
      "learning_rate": 9.753562918264275e-05,
      "loss": 1.133,
      "step": 46369
    },
    {
      "epoch": 3.989503570506754,
      "grad_norm": 1.3359375,
      "learning_rate": 9.751958456978182e-05,
      "loss": 1.4087,
      "step": 46370
    },
    {
      "epoch": 3.9895896068140755,
      "grad_norm": 1.2109375,
      "learning_rate": 9.750354113409782e-05,
      "loss": 0.8352,
      "step": 46371
    },
    {
      "epoch": 3.989675643121397,
      "grad_norm": 1.46875,
      "learning_rate": 9.748749887563779e-05,
      "loss": 1.108,
      "step": 46372
    },
    {
      "epoch": 3.989761679428719,
      "grad_norm": 1.3984375,
      "learning_rate": 9.747145779444844e-05,
      "loss": 1.4375,
      "step": 46373
    },
    {
      "epoch": 3.9898477157360404,
      "grad_norm": 1.203125,
      "learning_rate": 9.745541789057689e-05,
      "loss": 1.3098,
      "step": 46374
    },
    {
      "epoch": 3.9899337520433624,
      "grad_norm": 0.90625,
      "learning_rate": 9.743937916406981e-05,
      "loss": 1.1077,
      "step": 46375
    },
    {
      "epoch": 3.990019788350684,
      "grad_norm": 1.515625,
      "learning_rate": 9.74233416149743e-05,
      "loss": 1.2955,
      "step": 46376
    },
    {
      "epoch": 3.9901058246580057,
      "grad_norm": 1.09375,
      "learning_rate": 9.740730524333729e-05,
      "loss": 0.9903,
      "step": 46377
    },
    {
      "epoch": 3.9901918609653273,
      "grad_norm": 1.046875,
      "learning_rate": 9.739127004920551e-05,
      "loss": 1.0475,
      "step": 46378
    },
    {
      "epoch": 3.990277897272649,
      "grad_norm": 1.03125,
      "learning_rate": 9.737523603262599e-05,
      "loss": 1.6709,
      "step": 46379
    },
    {
      "epoch": 3.990363933579971,
      "grad_norm": 1.1796875,
      "learning_rate": 9.735920319364567e-05,
      "loss": 0.9763,
      "step": 46380
    },
    {
      "epoch": 3.990449969887292,
      "grad_norm": 1.25,
      "learning_rate": 9.734317153231131e-05,
      "loss": 1.0148,
      "step": 46381
    },
    {
      "epoch": 3.9905360061946142,
      "grad_norm": 1.0390625,
      "learning_rate": 9.73271410486698e-05,
      "loss": 1.1103,
      "step": 46382
    },
    {
      "epoch": 3.990622042501936,
      "grad_norm": 1.1484375,
      "learning_rate": 9.731111174276813e-05,
      "loss": 1.2337,
      "step": 46383
    },
    {
      "epoch": 3.9907080788092575,
      "grad_norm": 1.1328125,
      "learning_rate": 9.729508361465311e-05,
      "loss": 1.6679,
      "step": 46384
    },
    {
      "epoch": 3.990794115116579,
      "grad_norm": 1.15625,
      "learning_rate": 9.727905666437182e-05,
      "loss": 0.8698,
      "step": 46385
    },
    {
      "epoch": 3.9908801514239007,
      "grad_norm": 0.9453125,
      "learning_rate": 9.72630308919708e-05,
      "loss": 1.1877,
      "step": 46386
    },
    {
      "epoch": 3.990966187731223,
      "grad_norm": 1.4375,
      "learning_rate": 9.724700629749706e-05,
      "loss": 0.81,
      "step": 46387
    },
    {
      "epoch": 3.9910522240385444,
      "grad_norm": 0.78515625,
      "learning_rate": 9.723098288099769e-05,
      "loss": 0.6813,
      "step": 46388
    },
    {
      "epoch": 3.991138260345866,
      "grad_norm": 1.3046875,
      "learning_rate": 9.721496064251917e-05,
      "loss": 1.4151,
      "step": 46389
    },
    {
      "epoch": 3.9912242966531877,
      "grad_norm": 1.0546875,
      "learning_rate": 9.719893958210857e-05,
      "loss": 0.9828,
      "step": 46390
    },
    {
      "epoch": 3.9913103329605093,
      "grad_norm": 1.0859375,
      "learning_rate": 9.718291969981291e-05,
      "loss": 1.7212,
      "step": 46391
    },
    {
      "epoch": 3.991396369267831,
      "grad_norm": 1.6796875,
      "learning_rate": 9.716690099567866e-05,
      "loss": 1.6301,
      "step": 46392
    },
    {
      "epoch": 3.9914824055751525,
      "grad_norm": 0.83984375,
      "learning_rate": 9.715088346975303e-05,
      "loss": 0.9785,
      "step": 46393
    },
    {
      "epoch": 3.9915684418824746,
      "grad_norm": 0.8984375,
      "learning_rate": 9.713486712208264e-05,
      "loss": 0.6014,
      "step": 46394
    },
    {
      "epoch": 3.9916544781897962,
      "grad_norm": 24.5,
      "learning_rate": 9.711885195271436e-05,
      "loss": 1.3019,
      "step": 46395
    },
    {
      "epoch": 3.991740514497118,
      "grad_norm": 0.85546875,
      "learning_rate": 9.710283796169522e-05,
      "loss": 0.6245,
      "step": 46396
    },
    {
      "epoch": 3.9918265508044395,
      "grad_norm": 0.9765625,
      "learning_rate": 9.70868251490718e-05,
      "loss": 0.785,
      "step": 46397
    },
    {
      "epoch": 3.991912587111761,
      "grad_norm": 1.609375,
      "learning_rate": 9.707081351489105e-05,
      "loss": 1.3711,
      "step": 46398
    },
    {
      "epoch": 3.9919986234190827,
      "grad_norm": 0.97265625,
      "learning_rate": 9.705480305919995e-05,
      "loss": 1.1701,
      "step": 46399
    },
    {
      "epoch": 3.9920846597264044,
      "grad_norm": 2.921875,
      "learning_rate": 9.703879378204506e-05,
      "loss": 1.6392,
      "step": 46400
    },
    {
      "epoch": 3.9921706960337264,
      "grad_norm": 12.25,
      "learning_rate": 9.702278568347329e-05,
      "loss": 1.0768,
      "step": 46401
    },
    {
      "epoch": 3.992256732341048,
      "grad_norm": 0.83203125,
      "learning_rate": 9.700677876353165e-05,
      "loss": 0.8417,
      "step": 46402
    },
    {
      "epoch": 3.9923427686483697,
      "grad_norm": 4.75,
      "learning_rate": 9.699077302226666e-05,
      "loss": 0.9655,
      "step": 46403
    },
    {
      "epoch": 3.9924288049556913,
      "grad_norm": 2.359375,
      "learning_rate": 9.697476845972542e-05,
      "loss": 0.7109,
      "step": 46404
    },
    {
      "epoch": 3.992514841263013,
      "grad_norm": 14.1875,
      "learning_rate": 9.695876507595442e-05,
      "loss": 1.4169,
      "step": 46405
    },
    {
      "epoch": 3.9926008775703345,
      "grad_norm": 1.2421875,
      "learning_rate": 9.694276287100067e-05,
      "loss": 1.6063,
      "step": 46406
    },
    {
      "epoch": 3.992686913877656,
      "grad_norm": 0.89453125,
      "learning_rate": 9.692676184491112e-05,
      "loss": 0.5451,
      "step": 46407
    },
    {
      "epoch": 3.9927729501849782,
      "grad_norm": 2.5625,
      "learning_rate": 9.691076199773224e-05,
      "loss": 1.2061,
      "step": 46408
    },
    {
      "epoch": 3.9928589864923,
      "grad_norm": 42.25,
      "learning_rate": 9.689476332951097e-05,
      "loss": 1.3296,
      "step": 46409
    },
    {
      "epoch": 3.9929450227996215,
      "grad_norm": 0.76953125,
      "learning_rate": 9.687876584029431e-05,
      "loss": 0.5791,
      "step": 46410
    },
    {
      "epoch": 3.993031059106943,
      "grad_norm": 2.171875,
      "learning_rate": 9.686276953012867e-05,
      "loss": 0.7859,
      "step": 46411
    },
    {
      "epoch": 3.9931170954142647,
      "grad_norm": 1.1875,
      "learning_rate": 9.684677439906103e-05,
      "loss": 1.111,
      "step": 46412
    },
    {
      "epoch": 3.9932031317215864,
      "grad_norm": 1.0,
      "learning_rate": 9.683078044713828e-05,
      "loss": 1.58,
      "step": 46413
    },
    {
      "epoch": 3.993289168028908,
      "grad_norm": 1.1171875,
      "learning_rate": 9.6814787674407e-05,
      "loss": 1.4704,
      "step": 46414
    },
    {
      "epoch": 3.99337520433623,
      "grad_norm": 7.0,
      "learning_rate": 9.679879608091402e-05,
      "loss": 1.4706,
      "step": 46415
    },
    {
      "epoch": 3.9934612406435517,
      "grad_norm": 0.95703125,
      "learning_rate": 9.678280566670628e-05,
      "loss": 0.8753,
      "step": 46416
    },
    {
      "epoch": 3.9935472769508733,
      "grad_norm": 1.5,
      "learning_rate": 9.676681643183028e-05,
      "loss": 0.9403,
      "step": 46417
    },
    {
      "epoch": 3.993633313258195,
      "grad_norm": 1.609375,
      "learning_rate": 9.675082837633303e-05,
      "loss": 0.8069,
      "step": 46418
    },
    {
      "epoch": 3.9937193495655166,
      "grad_norm": 0.84375,
      "learning_rate": 9.673484150026107e-05,
      "loss": 0.7912,
      "step": 46419
    },
    {
      "epoch": 3.993805385872838,
      "grad_norm": 0.98046875,
      "learning_rate": 9.67188558036613e-05,
      "loss": 1.2358,
      "step": 46420
    },
    {
      "epoch": 3.99389142218016,
      "grad_norm": 1.15625,
      "learning_rate": 9.67028712865804e-05,
      "loss": 1.0325,
      "step": 46421
    },
    {
      "epoch": 3.993977458487482,
      "grad_norm": 1.1484375,
      "learning_rate": 9.668688794906516e-05,
      "loss": 2.0407,
      "step": 46422
    },
    {
      "epoch": 3.9940634947948035,
      "grad_norm": 1.2421875,
      "learning_rate": 9.66709057911625e-05,
      "loss": 1.8474,
      "step": 46423
    },
    {
      "epoch": 3.994149531102125,
      "grad_norm": 0.9765625,
      "learning_rate": 9.665492481291882e-05,
      "loss": 0.9525,
      "step": 46424
    },
    {
      "epoch": 3.9942355674094467,
      "grad_norm": 1.0390625,
      "learning_rate": 9.663894501438109e-05,
      "loss": 1.0793,
      "step": 46425
    },
    {
      "epoch": 3.9943216037167684,
      "grad_norm": 0.81640625,
      "learning_rate": 9.662296639559609e-05,
      "loss": 0.8195,
      "step": 46426
    },
    {
      "epoch": 3.9944076400240904,
      "grad_norm": 1.109375,
      "learning_rate": 9.660698895661036e-05,
      "loss": 1.5316,
      "step": 46427
    },
    {
      "epoch": 3.9944936763314116,
      "grad_norm": 1.3828125,
      "learning_rate": 9.659101269747068e-05,
      "loss": 1.7554,
      "step": 46428
    },
    {
      "epoch": 3.9945797126387337,
      "grad_norm": 1.0234375,
      "learning_rate": 9.657503761822401e-05,
      "loss": 0.7235,
      "step": 46429
    },
    {
      "epoch": 3.9946657489460553,
      "grad_norm": 0.99609375,
      "learning_rate": 9.655906371891671e-05,
      "loss": 1.6212,
      "step": 46430
    },
    {
      "epoch": 3.994751785253377,
      "grad_norm": 0.94921875,
      "learning_rate": 9.654309099959574e-05,
      "loss": 1.0241,
      "step": 46431
    },
    {
      "epoch": 3.9948378215606986,
      "grad_norm": 107.5,
      "learning_rate": 9.652711946030785e-05,
      "loss": 1.2916,
      "step": 46432
    },
    {
      "epoch": 3.99492385786802,
      "grad_norm": 0.984375,
      "learning_rate": 9.651114910109959e-05,
      "loss": 1.2755,
      "step": 46433
    },
    {
      "epoch": 3.9950098941753422,
      "grad_norm": 0.95703125,
      "learning_rate": 9.649517992201784e-05,
      "loss": 1.1054,
      "step": 46434
    },
    {
      "epoch": 3.9950959304826634,
      "grad_norm": 3.890625,
      "learning_rate": 9.647921192310905e-05,
      "loss": 1.1434,
      "step": 46435
    },
    {
      "epoch": 3.9951819667899855,
      "grad_norm": 454.0,
      "learning_rate": 9.646324510442012e-05,
      "loss": 1.5117,
      "step": 46436
    },
    {
      "epoch": 3.995268003097307,
      "grad_norm": 0.94140625,
      "learning_rate": 9.644727946599785e-05,
      "loss": 0.8498,
      "step": 46437
    },
    {
      "epoch": 3.9953540394046287,
      "grad_norm": 1.03125,
      "learning_rate": 9.643131500788866e-05,
      "loss": 0.7754,
      "step": 46438
    },
    {
      "epoch": 3.9954400757119504,
      "grad_norm": 3.015625,
      "learning_rate": 9.641535173013938e-05,
      "loss": 1.0664,
      "step": 46439
    },
    {
      "epoch": 3.995526112019272,
      "grad_norm": 1.125,
      "learning_rate": 9.639938963279682e-05,
      "loss": 1.0222,
      "step": 46440
    },
    {
      "epoch": 3.995612148326594,
      "grad_norm": 1.1875,
      "learning_rate": 9.638342871590744e-05,
      "loss": 0.8204,
      "step": 46441
    },
    {
      "epoch": 3.9956981846339152,
      "grad_norm": 0.85546875,
      "learning_rate": 9.636746897951804e-05,
      "loss": 0.676,
      "step": 46442
    },
    {
      "epoch": 3.9957842209412373,
      "grad_norm": 0.70703125,
      "learning_rate": 9.63515104236754e-05,
      "loss": 0.7124,
      "step": 46443
    },
    {
      "epoch": 3.995870257248559,
      "grad_norm": 0.890625,
      "learning_rate": 9.633555304842595e-05,
      "loss": 1.1389,
      "step": 46444
    },
    {
      "epoch": 3.9959562935558806,
      "grad_norm": 1.140625,
      "learning_rate": 9.631959685381669e-05,
      "loss": 1.313,
      "step": 46445
    },
    {
      "epoch": 3.996042329863202,
      "grad_norm": 1.1484375,
      "learning_rate": 9.630364183989387e-05,
      "loss": 0.6279,
      "step": 46446
    },
    {
      "epoch": 3.996128366170524,
      "grad_norm": 1.2109375,
      "learning_rate": 9.628768800670446e-05,
      "loss": 0.881,
      "step": 46447
    },
    {
      "epoch": 3.996214402477846,
      "grad_norm": 7.4375,
      "learning_rate": 9.627173535429517e-05,
      "loss": 1.4757,
      "step": 46448
    },
    {
      "epoch": 3.9963004387851675,
      "grad_norm": 1.1484375,
      "learning_rate": 9.62557838827124e-05,
      "loss": 1.3861,
      "step": 46449
    },
    {
      "epoch": 3.996386475092489,
      "grad_norm": 0.9140625,
      "learning_rate": 9.623983359200289e-05,
      "loss": 0.9183,
      "step": 46450
    },
    {
      "epoch": 3.9964725113998107,
      "grad_norm": 95.5,
      "learning_rate": 9.622388448221348e-05,
      "loss": 1.028,
      "step": 46451
    },
    {
      "epoch": 3.9965585477071324,
      "grad_norm": 1.0859375,
      "learning_rate": 9.620793655339061e-05,
      "loss": 1.2904,
      "step": 46452
    },
    {
      "epoch": 3.996644584014454,
      "grad_norm": 0.76953125,
      "learning_rate": 9.619198980558091e-05,
      "loss": 1.3065,
      "step": 46453
    },
    {
      "epoch": 3.9967306203217756,
      "grad_norm": 1.578125,
      "learning_rate": 9.61760442388313e-05,
      "loss": 1.6566,
      "step": 46454
    },
    {
      "epoch": 3.9968166566290977,
      "grad_norm": 1.5703125,
      "learning_rate": 9.616009985318808e-05,
      "loss": 1.0984,
      "step": 46455
    },
    {
      "epoch": 3.9969026929364193,
      "grad_norm": 0.96875,
      "learning_rate": 9.614415664869797e-05,
      "loss": 1.5502,
      "step": 46456
    },
    {
      "epoch": 3.996988729243741,
      "grad_norm": 1.4921875,
      "learning_rate": 9.612821462540772e-05,
      "loss": 1.3172,
      "step": 46457
    },
    {
      "epoch": 3.9970747655510626,
      "grad_norm": 1.1796875,
      "learning_rate": 9.611227378336385e-05,
      "loss": 0.9266,
      "step": 46458
    },
    {
      "epoch": 3.997160801858384,
      "grad_norm": 1.015625,
      "learning_rate": 9.60963341226132e-05,
      "loss": 1.1548,
      "step": 46459
    },
    {
      "epoch": 3.997246838165706,
      "grad_norm": 2.0625,
      "learning_rate": 9.60803956432021e-05,
      "loss": 1.1995,
      "step": 46460
    },
    {
      "epoch": 3.9973328744730274,
      "grad_norm": 0.9765625,
      "learning_rate": 9.606445834517724e-05,
      "loss": 1.5314,
      "step": 46461
    },
    {
      "epoch": 3.9974189107803495,
      "grad_norm": 0.8515625,
      "learning_rate": 9.60485222285854e-05,
      "loss": 0.9663,
      "step": 46462
    },
    {
      "epoch": 3.997504947087671,
      "grad_norm": 1.046875,
      "learning_rate": 9.603258729347297e-05,
      "loss": 0.6855,
      "step": 46463
    },
    {
      "epoch": 3.9975909833949927,
      "grad_norm": 1.265625,
      "learning_rate": 9.601665353988676e-05,
      "loss": 1.2528,
      "step": 46464
    },
    {
      "epoch": 3.9976770197023144,
      "grad_norm": 1.0859375,
      "learning_rate": 9.600072096787316e-05,
      "loss": 0.9442,
      "step": 46465
    },
    {
      "epoch": 3.997763056009636,
      "grad_norm": 1.3203125,
      "learning_rate": 9.598478957747886e-05,
      "loss": 1.5244,
      "step": 46466
    },
    {
      "epoch": 3.9978490923169576,
      "grad_norm": 1.0234375,
      "learning_rate": 9.59688593687506e-05,
      "loss": 0.9817,
      "step": 46467
    },
    {
      "epoch": 3.9979351286242792,
      "grad_norm": 1.234375,
      "learning_rate": 9.595293034173474e-05,
      "loss": 1.6785,
      "step": 46468
    },
    {
      "epoch": 3.9980211649316013,
      "grad_norm": 1.265625,
      "learning_rate": 9.593700249647796e-05,
      "loss": 2.0318,
      "step": 46469
    },
    {
      "epoch": 3.998107201238923,
      "grad_norm": 1.5546875,
      "learning_rate": 9.592107583302701e-05,
      "loss": 1.1283,
      "step": 46470
    },
    {
      "epoch": 3.9981932375462446,
      "grad_norm": 0.9140625,
      "learning_rate": 9.590515035142816e-05,
      "loss": 1.3116,
      "step": 46471
    },
    {
      "epoch": 3.998279273853566,
      "grad_norm": 1.8359375,
      "learning_rate": 9.588922605172817e-05,
      "loss": 1.0245,
      "step": 46472
    },
    {
      "epoch": 3.998365310160888,
      "grad_norm": 1.2578125,
      "learning_rate": 9.587330293397373e-05,
      "loss": 1.3701,
      "step": 46473
    },
    {
      "epoch": 3.9984513464682094,
      "grad_norm": 0.9921875,
      "learning_rate": 9.585738099821117e-05,
      "loss": 1.1551,
      "step": 46474
    },
    {
      "epoch": 3.998537382775531,
      "grad_norm": 1.1953125,
      "learning_rate": 9.584146024448732e-05,
      "loss": 0.6489,
      "step": 46475
    },
    {
      "epoch": 3.998623419082853,
      "grad_norm": 1.2734375,
      "learning_rate": 9.582554067284838e-05,
      "loss": 0.8822,
      "step": 46476
    },
    {
      "epoch": 3.9987094553901747,
      "grad_norm": 1.09375,
      "learning_rate": 9.580962228334117e-05,
      "loss": 1.4364,
      "step": 46477
    },
    {
      "epoch": 3.9987954916974964,
      "grad_norm": 1.734375,
      "learning_rate": 9.579370507601237e-05,
      "loss": 1.368,
      "step": 46478
    },
    {
      "epoch": 3.998881528004818,
      "grad_norm": 1.40625,
      "learning_rate": 9.577778905090818e-05,
      "loss": 0.9449,
      "step": 46479
    },
    {
      "epoch": 3.9989675643121396,
      "grad_norm": 1.109375,
      "learning_rate": 9.576187420807536e-05,
      "loss": 1.1254,
      "step": 46480
    },
    {
      "epoch": 3.9990536006194617,
      "grad_norm": 1.0546875,
      "learning_rate": 9.574596054756057e-05,
      "loss": 0.9344,
      "step": 46481
    },
    {
      "epoch": 3.999139636926783,
      "grad_norm": 4.71875,
      "learning_rate": 9.57300480694101e-05,
      "loss": 0.9866,
      "step": 46482
    },
    {
      "epoch": 3.999225673234105,
      "grad_norm": 2.953125,
      "learning_rate": 9.571413677367064e-05,
      "loss": 1.033,
      "step": 46483
    },
    {
      "epoch": 3.9993117095414266,
      "grad_norm": 1.078125,
      "learning_rate": 9.56982266603888e-05,
      "loss": 1.6282,
      "step": 46484
    },
    {
      "epoch": 3.999397745848748,
      "grad_norm": 0.87890625,
      "learning_rate": 9.568231772961089e-05,
      "loss": 0.7244,
      "step": 46485
    },
    {
      "epoch": 3.99948378215607,
      "grad_norm": 1.59375,
      "learning_rate": 9.566640998138366e-05,
      "loss": 1.2726,
      "step": 46486
    },
    {
      "epoch": 3.9995698184633914,
      "grad_norm": 1.0625,
      "learning_rate": 9.565050341575343e-05,
      "loss": 1.755,
      "step": 46487
    },
    {
      "epoch": 3.9996558547707135,
      "grad_norm": 0.96484375,
      "learning_rate": 9.563459803276686e-05,
      "loss": 1.4281,
      "step": 46488
    },
    {
      "epoch": 3.9997418910780347,
      "grad_norm": 1.1640625,
      "learning_rate": 9.561869383247057e-05,
      "loss": 0.6879,
      "step": 46489
    },
    {
      "epoch": 3.9998279273853568,
      "grad_norm": 1.6015625,
      "learning_rate": 9.56027908149108e-05,
      "loss": 1.5974,
      "step": 46490
    },
    {
      "epoch": 3.9999139636926784,
      "grad_norm": 0.91796875,
      "learning_rate": 9.558688898013423e-05,
      "loss": 0.7321,
      "step": 46491
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.171875,
      "learning_rate": 9.55709883281875e-05,
      "loss": 0.6379,
      "step": 46492
    },
    {
      "epoch": 4.000086036307322,
      "grad_norm": 0.921875,
      "learning_rate": 9.555508885911684e-05,
      "loss": 0.8548,
      "step": 46493
    },
    {
      "epoch": 4.000172072614643,
      "grad_norm": 0.9296875,
      "learning_rate": 9.55391905729689e-05,
      "loss": 0.7436,
      "step": 46494
    },
    {
      "epoch": 4.000258108921965,
      "grad_norm": 358.0,
      "learning_rate": 9.552329346979011e-05,
      "loss": 1.2713,
      "step": 46495
    },
    {
      "epoch": 4.0003441452292865,
      "grad_norm": 1.1328125,
      "learning_rate": 9.550739754962707e-05,
      "loss": 1.0676,
      "step": 46496
    },
    {
      "epoch": 4.0003441452292865,
      "eval_loss": 1.0701700448989868,
      "eval_runtime": 357.4411,
      "eval_samples_per_second": 13.006,
      "eval_steps_per_second": 1.628,
      "step": 46496
    },
    {
      "epoch": 4.000430181536609,
      "grad_norm": 1.3125,
      "learning_rate": 9.549150281252633e-05,
      "loss": 1.0497,
      "step": 46497
    },
    {
      "epoch": 4.00051621784393,
      "grad_norm": 1.171875,
      "learning_rate": 9.547560925853416e-05,
      "loss": 0.73,
      "step": 46498
    },
    {
      "epoch": 4.000602254151252,
      "grad_norm": 0.79296875,
      "learning_rate": 9.545971688769711e-05,
      "loss": 0.8834,
      "step": 46499
    },
    {
      "epoch": 4.000688290458574,
      "grad_norm": 0.765625,
      "learning_rate": 9.544382570006189e-05,
      "loss": 0.9694,
      "step": 46500
    },
    {
      "epoch": 4.000774326765895,
      "grad_norm": 0.99609375,
      "learning_rate": 9.542793569567465e-05,
      "loss": 1.3287,
      "step": 46501
    },
    {
      "epoch": 4.000860363073217,
      "grad_norm": 2.0625,
      "learning_rate": 9.541204687458199e-05,
      "loss": 1.2109,
      "step": 46502
    },
    {
      "epoch": 4.000946399380538,
      "grad_norm": 0.96484375,
      "learning_rate": 9.539615923683054e-05,
      "loss": 0.7892,
      "step": 46503
    },
    {
      "epoch": 4.00103243568786,
      "grad_norm": 0.9765625,
      "learning_rate": 9.538027278246652e-05,
      "loss": 1.2395,
      "step": 46504
    },
    {
      "epoch": 4.001118471995182,
      "grad_norm": 0.8828125,
      "learning_rate": 9.536438751153664e-05,
      "loss": 0.7805,
      "step": 46505
    },
    {
      "epoch": 4.001204508302504,
      "grad_norm": 0.94140625,
      "learning_rate": 9.534850342408714e-05,
      "loss": 0.7639,
      "step": 46506
    },
    {
      "epoch": 4.001290544609826,
      "grad_norm": 0.890625,
      "learning_rate": 9.533262052016451e-05,
      "loss": 0.6755,
      "step": 46507
    },
    {
      "epoch": 4.001376580917147,
      "grad_norm": 0.90625,
      "learning_rate": 9.53167387998154e-05,
      "loss": 0.7141,
      "step": 46508
    },
    {
      "epoch": 4.001462617224469,
      "grad_norm": 25.75,
      "learning_rate": 9.5300858263086e-05,
      "loss": 0.6894,
      "step": 46509
    },
    {
      "epoch": 4.00154865353179,
      "grad_norm": 0.96484375,
      "learning_rate": 9.528497891002286e-05,
      "loss": 1.5908,
      "step": 46510
    },
    {
      "epoch": 4.001634689839112,
      "grad_norm": 1.5078125,
      "learning_rate": 9.526910074067263e-05,
      "loss": 0.9492,
      "step": 46511
    },
    {
      "epoch": 4.001720726146433,
      "grad_norm": 1.109375,
      "learning_rate": 9.525322375508139e-05,
      "loss": 0.6303,
      "step": 46512
    },
    {
      "epoch": 4.001806762453755,
      "grad_norm": 0.8515625,
      "learning_rate": 9.523734795329574e-05,
      "loss": 1.1607,
      "step": 46513
    },
    {
      "epoch": 4.0018927987610775,
      "grad_norm": 1.078125,
      "learning_rate": 9.52214733353623e-05,
      "loss": 1.2377,
      "step": 46514
    },
    {
      "epoch": 4.001978835068399,
      "grad_norm": 1.2109375,
      "learning_rate": 9.520559990132716e-05,
      "loss": 1.0002,
      "step": 46515
    },
    {
      "epoch": 4.002064871375721,
      "grad_norm": 1.0078125,
      "learning_rate": 9.518972765123706e-05,
      "loss": 0.8688,
      "step": 46516
    },
    {
      "epoch": 4.002150907683042,
      "grad_norm": 1.1328125,
      "learning_rate": 9.517385658513811e-05,
      "loss": 0.9172,
      "step": 46517
    },
    {
      "epoch": 4.002236943990364,
      "grad_norm": 1.125,
      "learning_rate": 9.515798670307696e-05,
      "loss": 0.8442,
      "step": 46518
    },
    {
      "epoch": 4.002322980297685,
      "grad_norm": 0.9140625,
      "learning_rate": 9.514211800510003e-05,
      "loss": 1.3034,
      "step": 46519
    },
    {
      "epoch": 4.002409016605007,
      "grad_norm": 1.46875,
      "learning_rate": 9.512625049125356e-05,
      "loss": 1.8729,
      "step": 46520
    },
    {
      "epoch": 4.002495052912329,
      "grad_norm": 0.97265625,
      "learning_rate": 9.511038416158407e-05,
      "loss": 1.1949,
      "step": 46521
    },
    {
      "epoch": 4.0025810892196505,
      "grad_norm": 2.640625,
      "learning_rate": 9.509451901613809e-05,
      "loss": 0.6648,
      "step": 46522
    },
    {
      "epoch": 4.002667125526973,
      "grad_norm": 1.2890625,
      "learning_rate": 9.507865505496172e-05,
      "loss": 1.6266,
      "step": 46523
    },
    {
      "epoch": 4.002753161834294,
      "grad_norm": 1.046875,
      "learning_rate": 9.506279227810156e-05,
      "loss": 0.9337,
      "step": 46524
    },
    {
      "epoch": 4.002839198141616,
      "grad_norm": 1.1796875,
      "learning_rate": 9.50469306856041e-05,
      "loss": 1.5033,
      "step": 46525
    },
    {
      "epoch": 4.002925234448938,
      "grad_norm": 1.1796875,
      "learning_rate": 9.50310702775155e-05,
      "loss": 1.8268,
      "step": 46526
    },
    {
      "epoch": 4.003011270756259,
      "grad_norm": 0.94140625,
      "learning_rate": 9.501521105388228e-05,
      "loss": 1.1881,
      "step": 46527
    },
    {
      "epoch": 4.003097307063581,
      "grad_norm": 1.125,
      "learning_rate": 9.499935301475088e-05,
      "loss": 0.7288,
      "step": 46528
    },
    {
      "epoch": 4.003183343370902,
      "grad_norm": 1.078125,
      "learning_rate": 9.49834961601675e-05,
      "loss": 1.026,
      "step": 46529
    },
    {
      "epoch": 4.003269379678224,
      "grad_norm": 0.88671875,
      "learning_rate": 9.496764049017876e-05,
      "loss": 0.7068,
      "step": 46530
    },
    {
      "epoch": 4.003355415985546,
      "grad_norm": 0.92578125,
      "learning_rate": 9.495178600483062e-05,
      "loss": 1.4005,
      "step": 46531
    },
    {
      "epoch": 4.003441452292868,
      "grad_norm": 0.91796875,
      "learning_rate": 9.493593270417e-05,
      "loss": 0.7333,
      "step": 46532
    },
    {
      "epoch": 4.00352748860019,
      "grad_norm": 1.4765625,
      "learning_rate": 9.492008058824286e-05,
      "loss": 1.0567,
      "step": 46533
    },
    {
      "epoch": 4.003613524907511,
      "grad_norm": 3.3125,
      "learning_rate": 9.490422965709567e-05,
      "loss": 1.4432,
      "step": 46534
    },
    {
      "epoch": 4.003699561214833,
      "grad_norm": 0.765625,
      "learning_rate": 9.488837991077498e-05,
      "loss": 0.7312,
      "step": 46535
    },
    {
      "epoch": 4.003785597522154,
      "grad_norm": 1.2421875,
      "learning_rate": 9.487253134932688e-05,
      "loss": 1.9476,
      "step": 46536
    },
    {
      "epoch": 4.003871633829476,
      "grad_norm": 1.2421875,
      "learning_rate": 9.485668397279778e-05,
      "loss": 1.1077,
      "step": 46537
    },
    {
      "epoch": 4.003957670136797,
      "grad_norm": 1.2578125,
      "learning_rate": 9.484083778123426e-05,
      "loss": 1.1943,
      "step": 46538
    },
    {
      "epoch": 4.004043706444119,
      "grad_norm": 1.2734375,
      "learning_rate": 9.482499277468237e-05,
      "loss": 1.2734,
      "step": 46539
    },
    {
      "epoch": 4.0041297427514415,
      "grad_norm": 0.80859375,
      "learning_rate": 9.480914895318854e-05,
      "loss": 1.1489,
      "step": 46540
    },
    {
      "epoch": 4.004215779058763,
      "grad_norm": 2.4375,
      "learning_rate": 9.479330631679928e-05,
      "loss": 0.6537,
      "step": 46541
    },
    {
      "epoch": 4.004301815366085,
      "grad_norm": 0.9609375,
      "learning_rate": 9.477746486556065e-05,
      "loss": 1.2774,
      "step": 46542
    },
    {
      "epoch": 4.004387851673406,
      "grad_norm": 1.1875,
      "learning_rate": 9.476162459951914e-05,
      "loss": 1.043,
      "step": 46543
    },
    {
      "epoch": 4.004473887980728,
      "grad_norm": 1.59375,
      "learning_rate": 9.474578551872121e-05,
      "loss": 1.3708,
      "step": 46544
    },
    {
      "epoch": 4.004559924288049,
      "grad_norm": 1.390625,
      "learning_rate": 9.472994762321286e-05,
      "loss": 1.049,
      "step": 46545
    },
    {
      "epoch": 4.004645960595371,
      "grad_norm": 1.09375,
      "learning_rate": 9.471411091304077e-05,
      "loss": 0.9134,
      "step": 46546
    },
    {
      "epoch": 4.004731996902693,
      "grad_norm": 1.171875,
      "learning_rate": 9.469827538825093e-05,
      "loss": 0.9165,
      "step": 46547
    },
    {
      "epoch": 4.0048180332100145,
      "grad_norm": 0.6171875,
      "learning_rate": 9.468244104888985e-05,
      "loss": 0.6427,
      "step": 46548
    },
    {
      "epoch": 4.004904069517337,
      "grad_norm": 1.2109375,
      "learning_rate": 9.466660789500387e-05,
      "loss": 1.4366,
      "step": 46549
    },
    {
      "epoch": 4.004990105824658,
      "grad_norm": 1.1953125,
      "learning_rate": 9.465077592663913e-05,
      "loss": 1.0071,
      "step": 46550
    },
    {
      "epoch": 4.00507614213198,
      "grad_norm": 1.0625,
      "learning_rate": 9.463494514384202e-05,
      "loss": 0.6853,
      "step": 46551
    },
    {
      "epoch": 4.005162178439301,
      "grad_norm": 0.875,
      "learning_rate": 9.461911554665898e-05,
      "loss": 0.471,
      "step": 46552
    },
    {
      "epoch": 4.005248214746623,
      "grad_norm": 1.1171875,
      "learning_rate": 9.46032871351361e-05,
      "loss": 0.9587,
      "step": 46553
    },
    {
      "epoch": 4.005334251053945,
      "grad_norm": 2.984375,
      "learning_rate": 9.458745990931971e-05,
      "loss": 0.9346,
      "step": 46554
    },
    {
      "epoch": 4.005420287361266,
      "grad_norm": 1.484375,
      "learning_rate": 9.457163386925632e-05,
      "loss": 0.9471,
      "step": 46555
    },
    {
      "epoch": 4.005506323668588,
      "grad_norm": 0.96484375,
      "learning_rate": 9.455580901499189e-05,
      "loss": 1.4285,
      "step": 46556
    },
    {
      "epoch": 4.00559235997591,
      "grad_norm": 1.3203125,
      "learning_rate": 9.453998534657298e-05,
      "loss": 0.8846,
      "step": 46557
    },
    {
      "epoch": 4.005678396283232,
      "grad_norm": 1.0234375,
      "learning_rate": 9.452416286404563e-05,
      "loss": 2.1262,
      "step": 46558
    },
    {
      "epoch": 4.005764432590553,
      "grad_norm": 19.625,
      "learning_rate": 9.45083415674563e-05,
      "loss": 1.7821,
      "step": 46559
    },
    {
      "epoch": 4.005850468897875,
      "grad_norm": 6.6875,
      "learning_rate": 9.449252145685128e-05,
      "loss": 0.642,
      "step": 46560
    },
    {
      "epoch": 4.005936505205197,
      "grad_norm": 0.9296875,
      "learning_rate": 9.447670253227664e-05,
      "loss": 0.8627,
      "step": 46561
    },
    {
      "epoch": 4.006022541512518,
      "grad_norm": 0.95703125,
      "learning_rate": 9.446088479377878e-05,
      "loss": 1.092,
      "step": 46562
    },
    {
      "epoch": 4.00610857781984,
      "grad_norm": 0.953125,
      "learning_rate": 9.444506824140414e-05,
      "loss": 0.6697,
      "step": 46563
    },
    {
      "epoch": 4.006194614127161,
      "grad_norm": 1.6484375,
      "learning_rate": 9.44292528751986e-05,
      "loss": 1.1706,
      "step": 46564
    },
    {
      "epoch": 4.0062806504344834,
      "grad_norm": 0.91015625,
      "learning_rate": 9.441343869520869e-05,
      "loss": 1.4118,
      "step": 46565
    },
    {
      "epoch": 4.006366686741805,
      "grad_norm": 0.99609375,
      "learning_rate": 9.439762570148064e-05,
      "loss": 1.2204,
      "step": 46566
    },
    {
      "epoch": 4.006452723049127,
      "grad_norm": 2.25,
      "learning_rate": 9.438181389406059e-05,
      "loss": 1.0196,
      "step": 46567
    },
    {
      "epoch": 4.006538759356449,
      "grad_norm": 1.0,
      "learning_rate": 9.436600327299477e-05,
      "loss": 2.157,
      "step": 46568
    },
    {
      "epoch": 4.00662479566377,
      "grad_norm": 1.0390625,
      "learning_rate": 9.435019383832955e-05,
      "loss": 0.7006,
      "step": 46569
    },
    {
      "epoch": 4.006710831971092,
      "grad_norm": 1.2578125,
      "learning_rate": 9.433438559011109e-05,
      "loss": 1.3028,
      "step": 46570
    },
    {
      "epoch": 4.006796868278413,
      "grad_norm": 0.921875,
      "learning_rate": 9.431857852838583e-05,
      "loss": 0.7565,
      "step": 46571
    },
    {
      "epoch": 4.006882904585735,
      "grad_norm": 0.96875,
      "learning_rate": 9.43027726531997e-05,
      "loss": 1.0907,
      "step": 46572
    },
    {
      "epoch": 4.006968940893056,
      "grad_norm": 0.98046875,
      "learning_rate": 9.428696796459912e-05,
      "loss": 0.7578,
      "step": 46573
    },
    {
      "epoch": 4.0070549772003785,
      "grad_norm": 11.125,
      "learning_rate": 9.427116446263018e-05,
      "loss": 0.7663,
      "step": 46574
    },
    {
      "epoch": 4.007141013507701,
      "grad_norm": 1.484375,
      "learning_rate": 9.425536214733915e-05,
      "loss": 1.6986,
      "step": 46575
    },
    {
      "epoch": 4.007227049815022,
      "grad_norm": 0.828125,
      "learning_rate": 9.42395610187724e-05,
      "loss": 0.7473,
      "step": 46576
    },
    {
      "epoch": 4.007313086122344,
      "grad_norm": 1.296875,
      "learning_rate": 9.422376107697589e-05,
      "loss": 0.9142,
      "step": 46577
    },
    {
      "epoch": 4.007399122429665,
      "grad_norm": 0.9296875,
      "learning_rate": 9.420796232199596e-05,
      "loss": 0.8151,
      "step": 46578
    },
    {
      "epoch": 4.007485158736987,
      "grad_norm": 0.8203125,
      "learning_rate": 9.419216475387892e-05,
      "loss": 0.9338,
      "step": 46579
    },
    {
      "epoch": 4.007571195044308,
      "grad_norm": 1.234375,
      "learning_rate": 9.41763683726708e-05,
      "loss": 0.8509,
      "step": 46580
    },
    {
      "epoch": 4.00765723135163,
      "grad_norm": 1.8828125,
      "learning_rate": 9.416057317841786e-05,
      "loss": 1.4914,
      "step": 46581
    },
    {
      "epoch": 4.007743267658952,
      "grad_norm": 1.6328125,
      "learning_rate": 9.414477917116643e-05,
      "loss": 1.4262,
      "step": 46582
    },
    {
      "epoch": 4.007829303966274,
      "grad_norm": 1.25,
      "learning_rate": 9.412898635096246e-05,
      "loss": 0.723,
      "step": 46583
    },
    {
      "epoch": 4.007915340273596,
      "grad_norm": 0.9140625,
      "learning_rate": 9.411319471785223e-05,
      "loss": 1.135,
      "step": 46584
    },
    {
      "epoch": 4.008001376580917,
      "grad_norm": 1.21875,
      "learning_rate": 9.409740427188218e-05,
      "loss": 1.3644,
      "step": 46585
    },
    {
      "epoch": 4.008087412888239,
      "grad_norm": 0.875,
      "learning_rate": 9.408161501309809e-05,
      "loss": 0.805,
      "step": 46586
    },
    {
      "epoch": 4.008173449195561,
      "grad_norm": 1.046875,
      "learning_rate": 9.406582694154648e-05,
      "loss": 0.6438,
      "step": 46587
    },
    {
      "epoch": 4.008259485502882,
      "grad_norm": 1.0078125,
      "learning_rate": 9.40500400572732e-05,
      "loss": 0.6937,
      "step": 46588
    },
    {
      "epoch": 4.008345521810204,
      "grad_norm": 1.0078125,
      "learning_rate": 9.403425436032465e-05,
      "loss": 0.9065,
      "step": 46589
    },
    {
      "epoch": 4.008431558117525,
      "grad_norm": 43.5,
      "learning_rate": 9.401846985074702e-05,
      "loss": 0.835,
      "step": 46590
    },
    {
      "epoch": 4.0085175944248475,
      "grad_norm": 0.9921875,
      "learning_rate": 9.400268652858634e-05,
      "loss": 0.681,
      "step": 46591
    },
    {
      "epoch": 4.008603630732169,
      "grad_norm": 0.8515625,
      "learning_rate": 9.398690439388884e-05,
      "loss": 0.8,
      "step": 46592
    },
    {
      "epoch": 4.008689667039491,
      "grad_norm": 1.15625,
      "learning_rate": 9.397112344670078e-05,
      "loss": 1.3154,
      "step": 46593
    },
    {
      "epoch": 4.008775703346813,
      "grad_norm": 0.84765625,
      "learning_rate": 9.39553436870681e-05,
      "loss": 1.4987,
      "step": 46594
    },
    {
      "epoch": 4.008861739654134,
      "grad_norm": 0.9609375,
      "learning_rate": 9.393956511503704e-05,
      "loss": 0.6688,
      "step": 46595
    },
    {
      "epoch": 4.008947775961456,
      "grad_norm": 1.171875,
      "learning_rate": 9.392378773065397e-05,
      "loss": 0.6468,
      "step": 46596
    },
    {
      "epoch": 4.009033812268777,
      "grad_norm": 1.4453125,
      "learning_rate": 9.390801153396467e-05,
      "loss": 0.8893,
      "step": 46597
    },
    {
      "epoch": 4.009119848576099,
      "grad_norm": 0.83984375,
      "learning_rate": 9.389223652501554e-05,
      "loss": 0.8303,
      "step": 46598
    },
    {
      "epoch": 4.0092058848834204,
      "grad_norm": 54.0,
      "learning_rate": 9.387646270385259e-05,
      "loss": 1.0673,
      "step": 46599
    },
    {
      "epoch": 4.0092919211907425,
      "grad_norm": 0.79296875,
      "learning_rate": 9.386069007052195e-05,
      "loss": 0.5544,
      "step": 46600
    },
    {
      "epoch": 4.009377957498065,
      "grad_norm": 0.73046875,
      "learning_rate": 9.384491862506994e-05,
      "loss": 0.6617,
      "step": 46601
    },
    {
      "epoch": 4.009463993805386,
      "grad_norm": 1.0859375,
      "learning_rate": 9.382914836754241e-05,
      "loss": 0.6336,
      "step": 46602
    },
    {
      "epoch": 4.009550030112708,
      "grad_norm": 1.3359375,
      "learning_rate": 9.381337929798567e-05,
      "loss": 1.116,
      "step": 46603
    },
    {
      "epoch": 4.009636066420029,
      "grad_norm": 0.6796875,
      "learning_rate": 9.379761141644588e-05,
      "loss": 0.5617,
      "step": 46604
    },
    {
      "epoch": 4.009722102727351,
      "grad_norm": 1.0234375,
      "learning_rate": 9.378184472296886e-05,
      "loss": 0.7072,
      "step": 46605
    },
    {
      "epoch": 4.009808139034672,
      "grad_norm": 1.046875,
      "learning_rate": 9.376607921760116e-05,
      "loss": 0.6962,
      "step": 46606
    },
    {
      "epoch": 4.009894175341994,
      "grad_norm": 0.91015625,
      "learning_rate": 9.375031490038855e-05,
      "loss": 0.8788,
      "step": 46607
    },
    {
      "epoch": 4.009980211649316,
      "grad_norm": 1.0390625,
      "learning_rate": 9.37345517713773e-05,
      "loss": 0.599,
      "step": 46608
    },
    {
      "epoch": 4.010066247956638,
      "grad_norm": 0.6640625,
      "learning_rate": 9.371878983061355e-05,
      "loss": 0.3734,
      "step": 46609
    },
    {
      "epoch": 4.01015228426396,
      "grad_norm": 1.3515625,
      "learning_rate": 9.370302907814321e-05,
      "loss": 1.4962,
      "step": 46610
    },
    {
      "epoch": 4.010238320571281,
      "grad_norm": 1.5625,
      "learning_rate": 9.368726951401247e-05,
      "loss": 0.8353,
      "step": 46611
    },
    {
      "epoch": 4.010324356878603,
      "grad_norm": 10.0625,
      "learning_rate": 9.367151113826755e-05,
      "loss": 0.5287,
      "step": 46612
    },
    {
      "epoch": 4.010410393185924,
      "grad_norm": 3.796875,
      "learning_rate": 9.365575395095439e-05,
      "loss": 0.9437,
      "step": 46613
    },
    {
      "epoch": 4.010496429493246,
      "grad_norm": 0.8984375,
      "learning_rate": 9.363999795211903e-05,
      "loss": 1.2278,
      "step": 46614
    },
    {
      "epoch": 4.010582465800568,
      "grad_norm": 46.5,
      "learning_rate": 9.362424314180785e-05,
      "loss": 1.3342,
      "step": 46615
    },
    {
      "epoch": 4.010668502107889,
      "grad_norm": 1.3828125,
      "learning_rate": 9.360848952006656e-05,
      "loss": 0.748,
      "step": 46616
    },
    {
      "epoch": 4.0107545384152115,
      "grad_norm": 1.3671875,
      "learning_rate": 9.35927370869415e-05,
      "loss": 1.6444,
      "step": 46617
    },
    {
      "epoch": 4.010840574722533,
      "grad_norm": 1.21875,
      "learning_rate": 9.357698584247853e-05,
      "loss": 1.6906,
      "step": 46618
    },
    {
      "epoch": 4.010926611029855,
      "grad_norm": 1.5390625,
      "learning_rate": 9.356123578672388e-05,
      "loss": 1.0162,
      "step": 46619
    },
    {
      "epoch": 4.011012647337176,
      "grad_norm": 1.3984375,
      "learning_rate": 9.354548691972364e-05,
      "loss": 0.8036,
      "step": 46620
    },
    {
      "epoch": 4.011098683644498,
      "grad_norm": 1.265625,
      "learning_rate": 9.352973924152369e-05,
      "loss": 0.8741,
      "step": 46621
    },
    {
      "epoch": 4.01118471995182,
      "grad_norm": 0.671875,
      "learning_rate": 9.351399275217021e-05,
      "loss": 0.6464,
      "step": 46622
    },
    {
      "epoch": 4.011270756259141,
      "grad_norm": 4.21875,
      "learning_rate": 9.349824745170932e-05,
      "loss": 1.1814,
      "step": 46623
    },
    {
      "epoch": 4.011356792566463,
      "grad_norm": 4.9375,
      "learning_rate": 9.348250334018694e-05,
      "loss": 0.9953,
      "step": 46624
    },
    {
      "epoch": 4.0114428288737844,
      "grad_norm": 1.109375,
      "learning_rate": 9.346676041764913e-05,
      "loss": 0.7484,
      "step": 46625
    },
    {
      "epoch": 4.0115288651811065,
      "grad_norm": 1.015625,
      "learning_rate": 9.345101868414213e-05,
      "loss": 0.8265,
      "step": 46626
    },
    {
      "epoch": 4.011614901488428,
      "grad_norm": 1.3671875,
      "learning_rate": 9.34352781397117e-05,
      "loss": 1.289,
      "step": 46627
    },
    {
      "epoch": 4.01170093779575,
      "grad_norm": 1.109375,
      "learning_rate": 9.341953878440412e-05,
      "loss": 1.8158,
      "step": 46628
    },
    {
      "epoch": 4.011786974103072,
      "grad_norm": 1.0625,
      "learning_rate": 9.34038006182652e-05,
      "loss": 1.476,
      "step": 46629
    },
    {
      "epoch": 4.011873010410393,
      "grad_norm": 1.09375,
      "learning_rate": 9.338806364134107e-05,
      "loss": 1.042,
      "step": 46630
    },
    {
      "epoch": 4.011959046717715,
      "grad_norm": 1.1171875,
      "learning_rate": 9.337232785367788e-05,
      "loss": 0.7419,
      "step": 46631
    },
    {
      "epoch": 4.012045083025036,
      "grad_norm": 95.5,
      "learning_rate": 9.335659325532147e-05,
      "loss": 0.8931,
      "step": 46632
    },
    {
      "epoch": 4.012131119332358,
      "grad_norm": 1.1796875,
      "learning_rate": 9.334085984631791e-05,
      "loss": 1.8017,
      "step": 46633
    },
    {
      "epoch": 4.0122171556396795,
      "grad_norm": 1.046875,
      "learning_rate": 9.332512762671336e-05,
      "loss": 1.3071,
      "step": 46634
    },
    {
      "epoch": 4.012303191947002,
      "grad_norm": 3.046875,
      "learning_rate": 9.330939659655364e-05,
      "loss": 0.9761,
      "step": 46635
    },
    {
      "epoch": 4.012389228254324,
      "grad_norm": 1.71875,
      "learning_rate": 9.32936667558848e-05,
      "loss": 1.3361,
      "step": 46636
    },
    {
      "epoch": 4.012475264561645,
      "grad_norm": 0.9765625,
      "learning_rate": 9.327793810475299e-05,
      "loss": 1.7338,
      "step": 46637
    },
    {
      "epoch": 4.012561300868967,
      "grad_norm": 2.125,
      "learning_rate": 9.3262210643204e-05,
      "loss": 1.4536,
      "step": 46638
    },
    {
      "epoch": 4.012647337176288,
      "grad_norm": 1.1796875,
      "learning_rate": 9.324648437128408e-05,
      "loss": 1.5459,
      "step": 46639
    },
    {
      "epoch": 4.01273337348361,
      "grad_norm": 1.7890625,
      "learning_rate": 9.323075928903895e-05,
      "loss": 1.6495,
      "step": 46640
    },
    {
      "epoch": 4.012819409790932,
      "grad_norm": 1.328125,
      "learning_rate": 9.321503539651472e-05,
      "loss": 1.9412,
      "step": 46641
    },
    {
      "epoch": 4.012905446098253,
      "grad_norm": 0.83984375,
      "learning_rate": 9.319931269375742e-05,
      "loss": 0.4679,
      "step": 46642
    },
    {
      "epoch": 4.0129914824055755,
      "grad_norm": 1.078125,
      "learning_rate": 9.318359118081298e-05,
      "loss": 1.0124,
      "step": 46643
    },
    {
      "epoch": 4.013077518712897,
      "grad_norm": 1.0078125,
      "learning_rate": 9.31678708577276e-05,
      "loss": 1.1,
      "step": 46644
    },
    {
      "epoch": 4.013163555020219,
      "grad_norm": 1.140625,
      "learning_rate": 9.315215172454689e-05,
      "loss": 0.7577,
      "step": 46645
    },
    {
      "epoch": 4.01324959132754,
      "grad_norm": 18.0,
      "learning_rate": 9.313643378131703e-05,
      "loss": 1.0513,
      "step": 46646
    },
    {
      "epoch": 4.013335627634862,
      "grad_norm": 1.265625,
      "learning_rate": 9.312071702808411e-05,
      "loss": 1.0358,
      "step": 46647
    },
    {
      "epoch": 4.013421663942184,
      "grad_norm": 0.7734375,
      "learning_rate": 9.310500146489381e-05,
      "loss": 0.7761,
      "step": 46648
    },
    {
      "epoch": 4.013507700249505,
      "grad_norm": 1.1328125,
      "learning_rate": 9.308928709179226e-05,
      "loss": 0.8863,
      "step": 46649
    },
    {
      "epoch": 4.013593736556827,
      "grad_norm": 1.140625,
      "learning_rate": 9.307357390882554e-05,
      "loss": 1.4236,
      "step": 46650
    },
    {
      "epoch": 4.0136797728641485,
      "grad_norm": 76.0,
      "learning_rate": 9.305786191603933e-05,
      "loss": 1.0864,
      "step": 46651
    },
    {
      "epoch": 4.0137658091714705,
      "grad_norm": 1.078125,
      "learning_rate": 9.304215111347974e-05,
      "loss": 0.975,
      "step": 46652
    },
    {
      "epoch": 4.013851845478792,
      "grad_norm": 0.9296875,
      "learning_rate": 9.302644150119282e-05,
      "loss": 0.818,
      "step": 46653
    },
    {
      "epoch": 4.013937881786114,
      "grad_norm": 1.1953125,
      "learning_rate": 9.301073307922431e-05,
      "loss": 1.3933,
      "step": 46654
    },
    {
      "epoch": 4.014023918093436,
      "grad_norm": 1.203125,
      "learning_rate": 9.299502584762021e-05,
      "loss": 0.8428,
      "step": 46655
    },
    {
      "epoch": 4.014109954400757,
      "grad_norm": 1.0390625,
      "learning_rate": 9.297931980642665e-05,
      "loss": 1.4615,
      "step": 46656
    },
    {
      "epoch": 4.014195990708079,
      "grad_norm": 1.0390625,
      "learning_rate": 9.296361495568927e-05,
      "loss": 1.0551,
      "step": 46657
    },
    {
      "epoch": 4.0142820270154,
      "grad_norm": 2.09375,
      "learning_rate": 9.294791129545433e-05,
      "loss": 1.8998,
      "step": 46658
    },
    {
      "epoch": 4.014368063322722,
      "grad_norm": 1.4296875,
      "learning_rate": 9.293220882576741e-05,
      "loss": 1.3003,
      "step": 46659
    },
    {
      "epoch": 4.0144540996300435,
      "grad_norm": 0.90625,
      "learning_rate": 9.291650754667457e-05,
      "loss": 0.8197,
      "step": 46660
    },
    {
      "epoch": 4.014540135937366,
      "grad_norm": 3.9375,
      "learning_rate": 9.2900807458222e-05,
      "loss": 1.2826,
      "step": 46661
    },
    {
      "epoch": 4.014626172244688,
      "grad_norm": 61.0,
      "learning_rate": 9.288510856045518e-05,
      "loss": 1.3226,
      "step": 46662
    },
    {
      "epoch": 4.014712208552009,
      "grad_norm": 482.0,
      "learning_rate": 9.286941085342026e-05,
      "loss": 0.6444,
      "step": 46663
    },
    {
      "epoch": 4.014798244859331,
      "grad_norm": 0.875,
      "learning_rate": 9.285371433716327e-05,
      "loss": 0.4899,
      "step": 46664
    },
    {
      "epoch": 4.014884281166652,
      "grad_norm": 1.40625,
      "learning_rate": 9.283801901172989e-05,
      "loss": 1.0665,
      "step": 46665
    },
    {
      "epoch": 4.014970317473974,
      "grad_norm": 7.34375,
      "learning_rate": 9.282232487716608e-05,
      "loss": 0.8201,
      "step": 46666
    },
    {
      "epoch": 4.015056353781295,
      "grad_norm": 1.0625,
      "learning_rate": 9.28066319335179e-05,
      "loss": 1.234,
      "step": 46667
    },
    {
      "epoch": 4.015142390088617,
      "grad_norm": 0.87109375,
      "learning_rate": 9.279094018083101e-05,
      "loss": 1.3208,
      "step": 46668
    },
    {
      "epoch": 4.0152284263959395,
      "grad_norm": 19.5,
      "learning_rate": 9.277524961915151e-05,
      "loss": 0.9312,
      "step": 46669
    },
    {
      "epoch": 4.015314462703261,
      "grad_norm": 1.0625,
      "learning_rate": 9.275956024852511e-05,
      "loss": 1.1459,
      "step": 46670
    },
    {
      "epoch": 4.015400499010583,
      "grad_norm": 1.265625,
      "learning_rate": 9.274387206899781e-05,
      "loss": 1.0977,
      "step": 46671
    },
    {
      "epoch": 4.015486535317904,
      "grad_norm": 15.4375,
      "learning_rate": 9.272818508061558e-05,
      "loss": 0.8495,
      "step": 46672
    },
    {
      "epoch": 4.015572571625226,
      "grad_norm": 8.4375,
      "learning_rate": 9.271249928342407e-05,
      "loss": 0.8977,
      "step": 46673
    },
    {
      "epoch": 4.015658607932547,
      "grad_norm": 0.86328125,
      "learning_rate": 9.269681467746932e-05,
      "loss": 0.9865,
      "step": 46674
    },
    {
      "epoch": 4.015744644239869,
      "grad_norm": 564.0,
      "learning_rate": 9.268113126279725e-05,
      "loss": 1.8129,
      "step": 46675
    },
    {
      "epoch": 4.015830680547191,
      "grad_norm": 1.109375,
      "learning_rate": 9.266544903945356e-05,
      "loss": 0.7805,
      "step": 46676
    },
    {
      "epoch": 4.0159167168545125,
      "grad_norm": 1.078125,
      "learning_rate": 9.26497680074842e-05,
      "loss": 0.8545,
      "step": 46677
    },
    {
      "epoch": 4.0160027531618345,
      "grad_norm": 1.3515625,
      "learning_rate": 9.26340881669352e-05,
      "loss": 1.2172,
      "step": 46678
    },
    {
      "epoch": 4.016088789469156,
      "grad_norm": 1.46875,
      "learning_rate": 9.261840951785195e-05,
      "loss": 1.385,
      "step": 46679
    },
    {
      "epoch": 4.016174825776478,
      "grad_norm": 6.03125,
      "learning_rate": 9.260273206028092e-05,
      "loss": 1.3292,
      "step": 46680
    },
    {
      "epoch": 4.016260862083799,
      "grad_norm": 1.4375,
      "learning_rate": 9.258705579426752e-05,
      "loss": 1.7766,
      "step": 46681
    },
    {
      "epoch": 4.016346898391121,
      "grad_norm": 0.98046875,
      "learning_rate": 9.257138071985772e-05,
      "loss": 1.1946,
      "step": 46682
    },
    {
      "epoch": 4.016432934698443,
      "grad_norm": 0.87890625,
      "learning_rate": 9.255570683709752e-05,
      "loss": 0.826,
      "step": 46683
    },
    {
      "epoch": 4.016518971005764,
      "grad_norm": 1.109375,
      "learning_rate": 9.254003414603256e-05,
      "loss": 1.1831,
      "step": 46684
    },
    {
      "epoch": 4.016605007313086,
      "grad_norm": 1.3515625,
      "learning_rate": 9.252436264670883e-05,
      "loss": 1.1792,
      "step": 46685
    },
    {
      "epoch": 4.0166910436204075,
      "grad_norm": 308.0,
      "learning_rate": 9.250869233917197e-05,
      "loss": 1.4721,
      "step": 46686
    },
    {
      "epoch": 4.01677707992773,
      "grad_norm": 0.85546875,
      "learning_rate": 9.249302322346797e-05,
      "loss": 0.4826,
      "step": 46687
    },
    {
      "epoch": 4.016863116235051,
      "grad_norm": 1.1953125,
      "learning_rate": 9.247735529964274e-05,
      "loss": 0.8535,
      "step": 46688
    },
    {
      "epoch": 4.016949152542373,
      "grad_norm": 2.515625,
      "learning_rate": 9.246168856774184e-05,
      "loss": 1.6271,
      "step": 46689
    },
    {
      "epoch": 4.017035188849695,
      "grad_norm": 0.84375,
      "learning_rate": 9.244602302781124e-05,
      "loss": 0.6947,
      "step": 46690
    },
    {
      "epoch": 4.017121225157016,
      "grad_norm": 0.90625,
      "learning_rate": 9.243035867989685e-05,
      "loss": 0.633,
      "step": 46691
    },
    {
      "epoch": 4.017207261464338,
      "grad_norm": 0.75390625,
      "learning_rate": 9.24146955240443e-05,
      "loss": 0.3258,
      "step": 46692
    },
    {
      "epoch": 4.017293297771659,
      "grad_norm": 1.5703125,
      "learning_rate": 9.23990335602995e-05,
      "loss": 1.6248,
      "step": 46693
    },
    {
      "epoch": 4.017379334078981,
      "grad_norm": 1.015625,
      "learning_rate": 9.23833727887084e-05,
      "loss": 1.1681,
      "step": 46694
    },
    {
      "epoch": 4.017465370386303,
      "grad_norm": 2.984375,
      "learning_rate": 9.236771320931647e-05,
      "loss": 0.835,
      "step": 46695
    },
    {
      "epoch": 4.017551406693625,
      "grad_norm": 1.15625,
      "learning_rate": 9.235205482216974e-05,
      "loss": 0.7016,
      "step": 46696
    },
    {
      "epoch": 4.017637443000947,
      "grad_norm": 1.0078125,
      "learning_rate": 9.233639762731411e-05,
      "loss": 1.2487,
      "step": 46697
    },
    {
      "epoch": 4.017723479308268,
      "grad_norm": 1.0234375,
      "learning_rate": 9.232074162479509e-05,
      "loss": 0.7273,
      "step": 46698
    },
    {
      "epoch": 4.01780951561559,
      "grad_norm": 5.03125,
      "learning_rate": 9.230508681465871e-05,
      "loss": 0.9183,
      "step": 46699
    },
    {
      "epoch": 4.017895551922911,
      "grad_norm": 1.6484375,
      "learning_rate": 9.228943319695054e-05,
      "loss": 1.7167,
      "step": 46700
    },
    {
      "epoch": 4.017981588230233,
      "grad_norm": 3.40625,
      "learning_rate": 9.22737807717165e-05,
      "loss": 0.4646,
      "step": 46701
    },
    {
      "epoch": 4.018067624537555,
      "grad_norm": 0.76171875,
      "learning_rate": 9.225812953900247e-05,
      "loss": 0.7257,
      "step": 46702
    },
    {
      "epoch": 4.0181536608448765,
      "grad_norm": 0.81640625,
      "learning_rate": 9.224247949885401e-05,
      "loss": 0.675,
      "step": 46703
    },
    {
      "epoch": 4.0182396971521985,
      "grad_norm": 0.8515625,
      "learning_rate": 9.222683065131698e-05,
      "loss": 0.6356,
      "step": 46704
    },
    {
      "epoch": 4.01832573345952,
      "grad_norm": 1.2421875,
      "learning_rate": 9.221118299643727e-05,
      "loss": 1.0117,
      "step": 46705
    },
    {
      "epoch": 4.018411769766842,
      "grad_norm": 0.9609375,
      "learning_rate": 9.219553653426044e-05,
      "loss": 0.9707,
      "step": 46706
    },
    {
      "epoch": 4.018497806074163,
      "grad_norm": 18.5,
      "learning_rate": 9.217989126483234e-05,
      "loss": 1.2704,
      "step": 46707
    },
    {
      "epoch": 4.018583842381485,
      "grad_norm": 1.3203125,
      "learning_rate": 9.216424718819889e-05,
      "loss": 1.8067,
      "step": 46708
    },
    {
      "epoch": 4.018669878688807,
      "grad_norm": 1.1796875,
      "learning_rate": 9.214860430440552e-05,
      "loss": 1.6313,
      "step": 46709
    },
    {
      "epoch": 4.018755914996128,
      "grad_norm": 1.0546875,
      "learning_rate": 9.213296261349829e-05,
      "loss": 0.9243,
      "step": 46710
    },
    {
      "epoch": 4.01884195130345,
      "grad_norm": 0.90625,
      "learning_rate": 9.21173221155227e-05,
      "loss": 0.8058,
      "step": 46711
    },
    {
      "epoch": 4.0189279876107715,
      "grad_norm": 0.91015625,
      "learning_rate": 9.210168281052467e-05,
      "loss": 0.6336,
      "step": 46712
    },
    {
      "epoch": 4.019014023918094,
      "grad_norm": 0.8671875,
      "learning_rate": 9.208604469854998e-05,
      "loss": 0.884,
      "step": 46713
    },
    {
      "epoch": 4.019100060225415,
      "grad_norm": 0.96484375,
      "learning_rate": 9.20704077796441e-05,
      "loss": 0.6148,
      "step": 46714
    },
    {
      "epoch": 4.019186096532737,
      "grad_norm": 1.015625,
      "learning_rate": 9.205477205385299e-05,
      "loss": 1.4453,
      "step": 46715
    },
    {
      "epoch": 4.019272132840059,
      "grad_norm": 1.5234375,
      "learning_rate": 9.203913752122233e-05,
      "loss": 0.9528,
      "step": 46716
    },
    {
      "epoch": 4.01935816914738,
      "grad_norm": 1.796875,
      "learning_rate": 9.202350418179783e-05,
      "loss": 1.8146,
      "step": 46717
    },
    {
      "epoch": 4.019444205454702,
      "grad_norm": 0.84765625,
      "learning_rate": 9.200787203562538e-05,
      "loss": 1.0901,
      "step": 46718
    },
    {
      "epoch": 4.019530241762023,
      "grad_norm": 2.328125,
      "learning_rate": 9.19922410827504e-05,
      "loss": 1.6026,
      "step": 46719
    },
    {
      "epoch": 4.019616278069345,
      "grad_norm": 1.2109375,
      "learning_rate": 9.197661132321877e-05,
      "loss": 1.3855,
      "step": 46720
    },
    {
      "epoch": 4.019702314376667,
      "grad_norm": 150.0,
      "learning_rate": 9.196098275707631e-05,
      "loss": 0.8954,
      "step": 46721
    },
    {
      "epoch": 4.019788350683989,
      "grad_norm": 1.1015625,
      "learning_rate": 9.194535538436854e-05,
      "loss": 1.1923,
      "step": 46722
    },
    {
      "epoch": 4.019874386991311,
      "grad_norm": 0.85546875,
      "learning_rate": 9.19297292051412e-05,
      "loss": 1.0582,
      "step": 46723
    },
    {
      "epoch": 4.019960423298632,
      "grad_norm": 1.25,
      "learning_rate": 9.191410421944018e-05,
      "loss": 1.6076,
      "step": 46724
    },
    {
      "epoch": 4.020046459605954,
      "grad_norm": 1.53125,
      "learning_rate": 9.189848042731091e-05,
      "loss": 0.9917,
      "step": 46725
    },
    {
      "epoch": 4.020132495913275,
      "grad_norm": 0.890625,
      "learning_rate": 9.18828578287993e-05,
      "loss": 1.134,
      "step": 46726
    },
    {
      "epoch": 4.020218532220597,
      "grad_norm": 0.91015625,
      "learning_rate": 9.18672364239509e-05,
      "loss": 1.7572,
      "step": 46727
    },
    {
      "epoch": 4.020304568527918,
      "grad_norm": 1.2578125,
      "learning_rate": 9.185161621281141e-05,
      "loss": 1.0333,
      "step": 46728
    },
    {
      "epoch": 4.0203906048352405,
      "grad_norm": 1.0078125,
      "learning_rate": 9.183599719542667e-05,
      "loss": 0.8741,
      "step": 46729
    },
    {
      "epoch": 4.0204766411425625,
      "grad_norm": 0.9140625,
      "learning_rate": 9.182037937184218e-05,
      "loss": 0.871,
      "step": 46730
    },
    {
      "epoch": 4.020562677449884,
      "grad_norm": 1.1171875,
      "learning_rate": 9.180476274210364e-05,
      "loss": 0.6905,
      "step": 46731
    },
    {
      "epoch": 4.020648713757206,
      "grad_norm": 2.90625,
      "learning_rate": 9.178914730625698e-05,
      "loss": 0.4591,
      "step": 46732
    },
    {
      "epoch": 4.020734750064527,
      "grad_norm": 628.0,
      "learning_rate": 9.17735330643475e-05,
      "loss": 1.3324,
      "step": 46733
    },
    {
      "epoch": 4.020820786371849,
      "grad_norm": 0.83984375,
      "learning_rate": 9.175792001642102e-05,
      "loss": 0.6405,
      "step": 46734
    },
    {
      "epoch": 4.02090682267917,
      "grad_norm": 0.86328125,
      "learning_rate": 9.174230816252338e-05,
      "loss": 0.9915,
      "step": 46735
    },
    {
      "epoch": 4.020992858986492,
      "grad_norm": 0.921875,
      "learning_rate": 9.172669750269996e-05,
      "loss": 0.8235,
      "step": 46736
    },
    {
      "epoch": 4.021078895293814,
      "grad_norm": 1.140625,
      "learning_rate": 9.171108803699652e-05,
      "loss": 1.6968,
      "step": 46737
    },
    {
      "epoch": 4.0211649316011355,
      "grad_norm": 1.3046875,
      "learning_rate": 9.169547976545888e-05,
      "loss": 1.0599,
      "step": 46738
    },
    {
      "epoch": 4.021250967908458,
      "grad_norm": 52.0,
      "learning_rate": 9.167987268813244e-05,
      "loss": 1.2378,
      "step": 46739
    },
    {
      "epoch": 4.021337004215779,
      "grad_norm": 0.76171875,
      "learning_rate": 9.166426680506306e-05,
      "loss": 0.7973,
      "step": 46740
    },
    {
      "epoch": 4.021423040523101,
      "grad_norm": 1.0546875,
      "learning_rate": 9.164866211629619e-05,
      "loss": 1.3917,
      "step": 46741
    },
    {
      "epoch": 4.021509076830422,
      "grad_norm": 0.81640625,
      "learning_rate": 9.163305862187754e-05,
      "loss": 0.857,
      "step": 46742
    },
    {
      "epoch": 4.021595113137744,
      "grad_norm": 1.0859375,
      "learning_rate": 9.161745632185292e-05,
      "loss": 1.2696,
      "step": 46743
    },
    {
      "epoch": 4.021681149445066,
      "grad_norm": 0.83203125,
      "learning_rate": 9.160185521626768e-05,
      "loss": 0.752,
      "step": 46744
    },
    {
      "epoch": 4.021767185752387,
      "grad_norm": 1.078125,
      "learning_rate": 9.158625530516756e-05,
      "loss": 1.0221,
      "step": 46745
    },
    {
      "epoch": 4.021853222059709,
      "grad_norm": 1.140625,
      "learning_rate": 9.157065658859837e-05,
      "loss": 1.0924,
      "step": 46746
    },
    {
      "epoch": 4.021939258367031,
      "grad_norm": 1.0625,
      "learning_rate": 9.155505906660538e-05,
      "loss": 0.731,
      "step": 46747
    },
    {
      "epoch": 4.022025294674353,
      "grad_norm": 1.1328125,
      "learning_rate": 9.153946273923447e-05,
      "loss": 0.7462,
      "step": 46748
    },
    {
      "epoch": 4.022111330981674,
      "grad_norm": 1.21875,
      "learning_rate": 9.152386760653131e-05,
      "loss": 1.4044,
      "step": 46749
    },
    {
      "epoch": 4.022197367288996,
      "grad_norm": 1.390625,
      "learning_rate": 9.150827366854126e-05,
      "loss": 1.3407,
      "step": 46750
    },
    {
      "epoch": 4.022283403596318,
      "grad_norm": 1.40625,
      "learning_rate": 9.149268092531015e-05,
      "loss": 1.0998,
      "step": 46751
    },
    {
      "epoch": 4.022369439903639,
      "grad_norm": 1.234375,
      "learning_rate": 9.147708937688331e-05,
      "loss": 1.2577,
      "step": 46752
    },
    {
      "epoch": 4.022455476210961,
      "grad_norm": 1.296875,
      "learning_rate": 9.146149902330664e-05,
      "loss": 1.9584,
      "step": 46753
    },
    {
      "epoch": 4.022541512518282,
      "grad_norm": 1.4375,
      "learning_rate": 9.14459098646257e-05,
      "loss": 2.5092,
      "step": 46754
    },
    {
      "epoch": 4.0226275488256045,
      "grad_norm": 1.1953125,
      "learning_rate": 9.143032190088596e-05,
      "loss": 1.081,
      "step": 46755
    },
    {
      "epoch": 4.0227135851329265,
      "grad_norm": 26.0,
      "learning_rate": 9.141473513213316e-05,
      "loss": 0.7564,
      "step": 46756
    },
    {
      "epoch": 4.022799621440248,
      "grad_norm": 0.9453125,
      "learning_rate": 9.139914955841266e-05,
      "loss": 0.4331,
      "step": 46757
    },
    {
      "epoch": 4.02288565774757,
      "grad_norm": 0.90625,
      "learning_rate": 9.13835651797702e-05,
      "loss": 1.5122,
      "step": 46758
    },
    {
      "epoch": 4.022971694054891,
      "grad_norm": 1.0234375,
      "learning_rate": 9.136798199625145e-05,
      "loss": 0.706,
      "step": 46759
    },
    {
      "epoch": 4.023057730362213,
      "grad_norm": 1.2734375,
      "learning_rate": 9.135240000790174e-05,
      "loss": 1.4786,
      "step": 46760
    },
    {
      "epoch": 4.023143766669534,
      "grad_norm": 1.46875,
      "learning_rate": 9.133681921476683e-05,
      "loss": 1.1827,
      "step": 46761
    },
    {
      "epoch": 4.023229802976856,
      "grad_norm": 11.875,
      "learning_rate": 9.132123961689232e-05,
      "loss": 0.864,
      "step": 46762
    },
    {
      "epoch": 4.023315839284178,
      "grad_norm": 1.125,
      "learning_rate": 9.13056612143236e-05,
      "loss": 0.6661,
      "step": 46763
    },
    {
      "epoch": 4.0234018755914995,
      "grad_norm": 1.078125,
      "learning_rate": 9.129008400710631e-05,
      "loss": 1.8096,
      "step": 46764
    },
    {
      "epoch": 4.023487911898822,
      "grad_norm": 0.80078125,
      "learning_rate": 9.12745079952862e-05,
      "loss": 0.6458,
      "step": 46765
    },
    {
      "epoch": 4.023573948206143,
      "grad_norm": 1.2109375,
      "learning_rate": 9.125893317890855e-05,
      "loss": 1.0519,
      "step": 46766
    },
    {
      "epoch": 4.023659984513465,
      "grad_norm": 1.21875,
      "learning_rate": 9.124335955801893e-05,
      "loss": 1.642,
      "step": 46767
    },
    {
      "epoch": 4.023746020820786,
      "grad_norm": 1.078125,
      "learning_rate": 9.122778713266322e-05,
      "loss": 1.5043,
      "step": 46768
    },
    {
      "epoch": 4.023832057128108,
      "grad_norm": 1.0859375,
      "learning_rate": 9.121221590288653e-05,
      "loss": 1.1241,
      "step": 46769
    },
    {
      "epoch": 4.02391809343543,
      "grad_norm": 0.68359375,
      "learning_rate": 9.119664586873477e-05,
      "loss": 0.4433,
      "step": 46770
    },
    {
      "epoch": 4.024004129742751,
      "grad_norm": 1.0234375,
      "learning_rate": 9.118107703025314e-05,
      "loss": 1.2964,
      "step": 46771
    },
    {
      "epoch": 4.024090166050073,
      "grad_norm": 1.0234375,
      "learning_rate": 9.116550938748742e-05,
      "loss": 1.0239,
      "step": 46772
    },
    {
      "epoch": 4.024176202357395,
      "grad_norm": 78.0,
      "learning_rate": 9.114994294048312e-05,
      "loss": 1.0123,
      "step": 46773
    },
    {
      "epoch": 4.024262238664717,
      "grad_norm": 1.0546875,
      "learning_rate": 9.113437768928567e-05,
      "loss": 0.7649,
      "step": 46774
    },
    {
      "epoch": 4.024348274972038,
      "grad_norm": 0.90625,
      "learning_rate": 9.11188136339406e-05,
      "loss": 1.022,
      "step": 46775
    },
    {
      "epoch": 4.02443431127936,
      "grad_norm": 1.5546875,
      "learning_rate": 9.11032507744936e-05,
      "loss": 1.425,
      "step": 46776
    },
    {
      "epoch": 4.024520347586682,
      "grad_norm": 0.765625,
      "learning_rate": 9.108768911098997e-05,
      "loss": 0.6256,
      "step": 46777
    },
    {
      "epoch": 4.024606383894003,
      "grad_norm": 1.015625,
      "learning_rate": 9.107212864347525e-05,
      "loss": 1.0407,
      "step": 46778
    },
    {
      "epoch": 4.024692420201325,
      "grad_norm": 0.890625,
      "learning_rate": 9.105656937199525e-05,
      "loss": 0.7547,
      "step": 46779
    },
    {
      "epoch": 4.024778456508646,
      "grad_norm": 0.9140625,
      "learning_rate": 9.104101129659503e-05,
      "loss": 1.2101,
      "step": 46780
    },
    {
      "epoch": 4.0248644928159685,
      "grad_norm": 1.2265625,
      "learning_rate": 9.102545441732052e-05,
      "loss": 1.3791,
      "step": 46781
    },
    {
      "epoch": 4.02495052912329,
      "grad_norm": 1.328125,
      "learning_rate": 9.100989873421683e-05,
      "loss": 0.9561,
      "step": 46782
    },
    {
      "epoch": 4.025036565430612,
      "grad_norm": 1.1015625,
      "learning_rate": 9.099434424732966e-05,
      "loss": 1.5733,
      "step": 46783
    },
    {
      "epoch": 4.025122601737934,
      "grad_norm": 0.98828125,
      "learning_rate": 9.097879095670464e-05,
      "loss": 1.5398,
      "step": 46784
    },
    {
      "epoch": 4.025208638045255,
      "grad_norm": 0.79296875,
      "learning_rate": 9.096323886238694e-05,
      "loss": 0.7334,
      "step": 46785
    },
    {
      "epoch": 4.025294674352577,
      "grad_norm": 1.6796875,
      "learning_rate": 9.094768796442221e-05,
      "loss": 1.8199,
      "step": 46786
    },
    {
      "epoch": 4.025380710659898,
      "grad_norm": 1.2265625,
      "learning_rate": 9.093213826285607e-05,
      "loss": 1.3058,
      "step": 46787
    },
    {
      "epoch": 4.02546674696722,
      "grad_norm": 1.1796875,
      "learning_rate": 9.091658975773376e-05,
      "loss": 1.3265,
      "step": 46788
    },
    {
      "epoch": 4.0255527832745415,
      "grad_norm": 0.87109375,
      "learning_rate": 9.090104244910085e-05,
      "loss": 0.7392,
      "step": 46789
    },
    {
      "epoch": 4.0256388195818635,
      "grad_norm": 0.91796875,
      "learning_rate": 9.088549633700282e-05,
      "loss": 0.6829,
      "step": 46790
    },
    {
      "epoch": 4.025724855889186,
      "grad_norm": 1.1328125,
      "learning_rate": 9.086995142148513e-05,
      "loss": 1.4451,
      "step": 46791
    },
    {
      "epoch": 4.025810892196507,
      "grad_norm": 1.328125,
      "learning_rate": 9.085440770259345e-05,
      "loss": 1.2026,
      "step": 46792
    },
    {
      "epoch": 4.025896928503829,
      "grad_norm": 0.78125,
      "learning_rate": 9.083886518037287e-05,
      "loss": 1.1056,
      "step": 46793
    },
    {
      "epoch": 4.02598296481115,
      "grad_norm": 0.7890625,
      "learning_rate": 9.082332385486908e-05,
      "loss": 0.473,
      "step": 46794
    },
    {
      "epoch": 4.026069001118472,
      "grad_norm": 1.0703125,
      "learning_rate": 9.080778372612757e-05,
      "loss": 0.982,
      "step": 46795
    },
    {
      "epoch": 4.026155037425793,
      "grad_norm": 1.0703125,
      "learning_rate": 9.079224479419356e-05,
      "loss": 0.5983,
      "step": 46796
    },
    {
      "epoch": 4.026241073733115,
      "grad_norm": 1.09375,
      "learning_rate": 9.077670705911284e-05,
      "loss": 0.9922,
      "step": 46797
    },
    {
      "epoch": 4.026327110040437,
      "grad_norm": 0.98046875,
      "learning_rate": 9.076117052093047e-05,
      "loss": 1.262,
      "step": 46798
    },
    {
      "epoch": 4.026413146347759,
      "grad_norm": 1.1484375,
      "learning_rate": 9.074563517969214e-05,
      "loss": 1.0123,
      "step": 46799
    },
    {
      "epoch": 4.026499182655081,
      "grad_norm": 1.2109375,
      "learning_rate": 9.073010103544332e-05,
      "loss": 0.9455,
      "step": 46800
    },
    {
      "epoch": 4.026585218962402,
      "grad_norm": 1.125,
      "learning_rate": 9.071456808822925e-05,
      "loss": 0.5992,
      "step": 46801
    },
    {
      "epoch": 4.026671255269724,
      "grad_norm": 1.1875,
      "learning_rate": 9.069903633809545e-05,
      "loss": 1.3177,
      "step": 46802
    },
    {
      "epoch": 4.026757291577045,
      "grad_norm": 0.609375,
      "learning_rate": 9.068350578508754e-05,
      "loss": 0.4375,
      "step": 46803
    },
    {
      "epoch": 4.026843327884367,
      "grad_norm": 0.87890625,
      "learning_rate": 9.066797642925057e-05,
      "loss": 0.5123,
      "step": 46804
    },
    {
      "epoch": 4.026929364191689,
      "grad_norm": 17536.0,
      "learning_rate": 9.065244827063018e-05,
      "loss": 0.9104,
      "step": 46805
    },
    {
      "epoch": 4.02701540049901,
      "grad_norm": 1.109375,
      "learning_rate": 9.063692130927192e-05,
      "loss": 1.3209,
      "step": 46806
    },
    {
      "epoch": 4.0271014368063325,
      "grad_norm": 3.6875,
      "learning_rate": 9.062139554522092e-05,
      "loss": 0.6883,
      "step": 46807
    },
    {
      "epoch": 4.027187473113654,
      "grad_norm": 0.984375,
      "learning_rate": 9.060587097852268e-05,
      "loss": 0.9802,
      "step": 46808
    },
    {
      "epoch": 4.027273509420976,
      "grad_norm": 0.96875,
      "learning_rate": 9.05903476092228e-05,
      "loss": 1.3097,
      "step": 46809
    },
    {
      "epoch": 4.027359545728297,
      "grad_norm": 1.171875,
      "learning_rate": 9.05748254373664e-05,
      "loss": 1.6676,
      "step": 46810
    },
    {
      "epoch": 4.027445582035619,
      "grad_norm": 1.15625,
      "learning_rate": 9.055930446299915e-05,
      "loss": 1.3769,
      "step": 46811
    },
    {
      "epoch": 4.027531618342941,
      "grad_norm": 240.0,
      "learning_rate": 9.054378468616615e-05,
      "loss": 0.5238,
      "step": 46812
    },
    {
      "epoch": 4.027617654650262,
      "grad_norm": 1208.0,
      "learning_rate": 9.052826610691295e-05,
      "loss": 0.9567,
      "step": 46813
    },
    {
      "epoch": 4.027703690957584,
      "grad_norm": 165.0,
      "learning_rate": 9.051274872528504e-05,
      "loss": 1.3393,
      "step": 46814
    },
    {
      "epoch": 4.0277897272649055,
      "grad_norm": 0.9609375,
      "learning_rate": 9.049723254132764e-05,
      "loss": 0.9841,
      "step": 46815
    },
    {
      "epoch": 4.0278757635722275,
      "grad_norm": 1.4609375,
      "learning_rate": 9.048171755508611e-05,
      "loss": 1.4374,
      "step": 46816
    },
    {
      "epoch": 4.02796179987955,
      "grad_norm": 0.984375,
      "learning_rate": 9.046620376660608e-05,
      "loss": 1.059,
      "step": 46817
    },
    {
      "epoch": 4.028047836186871,
      "grad_norm": 1.2421875,
      "learning_rate": 9.045069117593263e-05,
      "loss": 0.8085,
      "step": 46818
    },
    {
      "epoch": 4.028133872494193,
      "grad_norm": 1.2734375,
      "learning_rate": 9.043517978311122e-05,
      "loss": 1.0913,
      "step": 46819
    },
    {
      "epoch": 4.028219908801514,
      "grad_norm": 1.1484375,
      "learning_rate": 9.041966958818743e-05,
      "loss": 1.0745,
      "step": 46820
    },
    {
      "epoch": 4.028305945108836,
      "grad_norm": 1.2578125,
      "learning_rate": 9.040416059120626e-05,
      "loss": 1.0746,
      "step": 46821
    },
    {
      "epoch": 4.028391981416157,
      "grad_norm": 1.1484375,
      "learning_rate": 9.038865279221342e-05,
      "loss": 1.1306,
      "step": 46822
    },
    {
      "epoch": 4.028478017723479,
      "grad_norm": 1.0078125,
      "learning_rate": 9.037314619125398e-05,
      "loss": 1.3799,
      "step": 46823
    },
    {
      "epoch": 4.028564054030801,
      "grad_norm": 0.85546875,
      "learning_rate": 9.035764078837344e-05,
      "loss": 0.9256,
      "step": 46824
    },
    {
      "epoch": 4.028650090338123,
      "grad_norm": 2.875,
      "learning_rate": 9.034213658361723e-05,
      "loss": 1.3204,
      "step": 46825
    },
    {
      "epoch": 4.028736126645445,
      "grad_norm": 8.125,
      "learning_rate": 9.032663357703036e-05,
      "loss": 1.073,
      "step": 46826
    },
    {
      "epoch": 4.028822162952766,
      "grad_norm": 1.15625,
      "learning_rate": 9.031113176865863e-05,
      "loss": 0.9268,
      "step": 46827
    },
    {
      "epoch": 4.028908199260088,
      "grad_norm": 1.2265625,
      "learning_rate": 9.029563115854706e-05,
      "loss": 0.9521,
      "step": 46828
    },
    {
      "epoch": 4.028994235567409,
      "grad_norm": 1.078125,
      "learning_rate": 9.028013174674105e-05,
      "loss": 1.4382,
      "step": 46829
    },
    {
      "epoch": 4.029080271874731,
      "grad_norm": 1.2421875,
      "learning_rate": 9.026463353328612e-05,
      "loss": 1.6552,
      "step": 46830
    },
    {
      "epoch": 4.029166308182053,
      "grad_norm": 0.8359375,
      "learning_rate": 9.024913651822736e-05,
      "loss": 0.755,
      "step": 46831
    },
    {
      "epoch": 4.029252344489374,
      "grad_norm": 4.09375,
      "learning_rate": 9.023364070161011e-05,
      "loss": 1.8362,
      "step": 46832
    },
    {
      "epoch": 4.0293383807966965,
      "grad_norm": 1.2578125,
      "learning_rate": 9.021814608347994e-05,
      "loss": 1.0631,
      "step": 46833
    },
    {
      "epoch": 4.029424417104018,
      "grad_norm": 1.171875,
      "learning_rate": 9.020265266388188e-05,
      "loss": 0.6693,
      "step": 46834
    },
    {
      "epoch": 4.02951045341134,
      "grad_norm": 1.2734375,
      "learning_rate": 9.018716044286134e-05,
      "loss": 0.8993,
      "step": 46835
    },
    {
      "epoch": 4.029596489718661,
      "grad_norm": 1.359375,
      "learning_rate": 9.017166942046379e-05,
      "loss": 0.941,
      "step": 46836
    },
    {
      "epoch": 4.029682526025983,
      "grad_norm": 0.98046875,
      "learning_rate": 9.015617959673422e-05,
      "loss": 0.6397,
      "step": 46837
    },
    {
      "epoch": 4.029768562333305,
      "grad_norm": 1.3046875,
      "learning_rate": 9.014069097171829e-05,
      "loss": 0.9497,
      "step": 46838
    },
    {
      "epoch": 4.029854598640626,
      "grad_norm": 0.8515625,
      "learning_rate": 9.0125203545461e-05,
      "loss": 1.3127,
      "step": 46839
    },
    {
      "epoch": 4.029940634947948,
      "grad_norm": 0.9296875,
      "learning_rate": 9.010971731800777e-05,
      "loss": 0.8779,
      "step": 46840
    },
    {
      "epoch": 4.0300266712552695,
      "grad_norm": 1.3203125,
      "learning_rate": 9.0094232289404e-05,
      "loss": 1.0958,
      "step": 46841
    },
    {
      "epoch": 4.0301127075625915,
      "grad_norm": 1.25,
      "learning_rate": 9.007874845969478e-05,
      "loss": 1.6247,
      "step": 46842
    },
    {
      "epoch": 4.030198743869913,
      "grad_norm": 6.53125,
      "learning_rate": 9.006326582892549e-05,
      "loss": 1.325,
      "step": 46843
    },
    {
      "epoch": 4.030284780177235,
      "grad_norm": 0.9296875,
      "learning_rate": 9.00477843971415e-05,
      "loss": 1.5357,
      "step": 46844
    },
    {
      "epoch": 4.030370816484557,
      "grad_norm": 0.69921875,
      "learning_rate": 9.003230416438795e-05,
      "loss": 0.5285,
      "step": 46845
    },
    {
      "epoch": 4.030456852791878,
      "grad_norm": 1.1953125,
      "learning_rate": 9.001682513071014e-05,
      "loss": 0.9577,
      "step": 46846
    },
    {
      "epoch": 4.0305428890992,
      "grad_norm": 0.7890625,
      "learning_rate": 9.00013472961535e-05,
      "loss": 0.8395,
      "step": 46847
    },
    {
      "epoch": 4.030628925406521,
      "grad_norm": 9.125,
      "learning_rate": 8.998587066076308e-05,
      "loss": 1.3304,
      "step": 46848
    },
    {
      "epoch": 4.030714961713843,
      "grad_norm": 1.03125,
      "learning_rate": 8.997039522458423e-05,
      "loss": 0.4841,
      "step": 46849
    },
    {
      "epoch": 4.0308009980211645,
      "grad_norm": 1.09375,
      "learning_rate": 8.99549209876624e-05,
      "loss": 0.555,
      "step": 46850
    },
    {
      "epoch": 4.030887034328487,
      "grad_norm": 0.75,
      "learning_rate": 8.993944795004249e-05,
      "loss": 0.8296,
      "step": 46851
    },
    {
      "epoch": 4.030973070635809,
      "grad_norm": 0.98046875,
      "learning_rate": 8.992397611177005e-05,
      "loss": 0.5957,
      "step": 46852
    },
    {
      "epoch": 4.03105910694313,
      "grad_norm": 0.9765625,
      "learning_rate": 8.990850547289014e-05,
      "loss": 0.7118,
      "step": 46853
    },
    {
      "epoch": 4.031145143250452,
      "grad_norm": 0.9140625,
      "learning_rate": 8.989303603344812e-05,
      "loss": 0.4977,
      "step": 46854
    },
    {
      "epoch": 4.031231179557773,
      "grad_norm": 1.5546875,
      "learning_rate": 8.987756779348926e-05,
      "loss": 1.1541,
      "step": 46855
    },
    {
      "epoch": 4.031317215865095,
      "grad_norm": 0.8984375,
      "learning_rate": 8.986210075305868e-05,
      "loss": 0.3923,
      "step": 46856
    },
    {
      "epoch": 4.031403252172416,
      "grad_norm": 0.9765625,
      "learning_rate": 8.984663491220169e-05,
      "loss": 0.9619,
      "step": 46857
    },
    {
      "epoch": 4.031489288479738,
      "grad_norm": 181.0,
      "learning_rate": 8.98311702709636e-05,
      "loss": 0.6947,
      "step": 46858
    },
    {
      "epoch": 4.0315753247870605,
      "grad_norm": 3.65625,
      "learning_rate": 8.98157068293895e-05,
      "loss": 0.825,
      "step": 46859
    },
    {
      "epoch": 4.031661361094382,
      "grad_norm": 0.71484375,
      "learning_rate": 8.980024458752462e-05,
      "loss": 0.4029,
      "step": 46860
    },
    {
      "epoch": 4.031747397401704,
      "grad_norm": 1.4140625,
      "learning_rate": 8.978478354541442e-05,
      "loss": 1.0602,
      "step": 46861
    },
    {
      "epoch": 4.031833433709025,
      "grad_norm": 2.84375,
      "learning_rate": 8.97693237031038e-05,
      "loss": 1.261,
      "step": 46862
    },
    {
      "epoch": 4.031919470016347,
      "grad_norm": 1.09375,
      "learning_rate": 8.975386506063809e-05,
      "loss": 1.0189,
      "step": 46863
    },
    {
      "epoch": 4.032005506323668,
      "grad_norm": 62976.0,
      "learning_rate": 8.973840761806262e-05,
      "loss": 1.2951,
      "step": 46864
    },
    {
      "epoch": 4.03209154263099,
      "grad_norm": 1.09375,
      "learning_rate": 8.972295137542247e-05,
      "loss": 1.0321,
      "step": 46865
    },
    {
      "epoch": 4.032177578938312,
      "grad_norm": 1.109375,
      "learning_rate": 8.970749633276299e-05,
      "loss": 0.9828,
      "step": 46866
    },
    {
      "epoch": 4.0322636152456335,
      "grad_norm": 0.859375,
      "learning_rate": 8.96920424901292e-05,
      "loss": 0.7205,
      "step": 46867
    },
    {
      "epoch": 4.0323496515529555,
      "grad_norm": 1.015625,
      "learning_rate": 8.967658984756649e-05,
      "loss": 1.1022,
      "step": 46868
    },
    {
      "epoch": 4.032435687860277,
      "grad_norm": 1.203125,
      "learning_rate": 8.96611384051198e-05,
      "loss": 1.4265,
      "step": 46869
    },
    {
      "epoch": 4.032521724167599,
      "grad_norm": 24.0,
      "learning_rate": 8.964568816283452e-05,
      "loss": 1.3113,
      "step": 46870
    },
    {
      "epoch": 4.03260776047492,
      "grad_norm": 1.8125,
      "learning_rate": 8.96302391207559e-05,
      "loss": 1.069,
      "step": 46871
    },
    {
      "epoch": 4.032693796782242,
      "grad_norm": 0.9609375,
      "learning_rate": 8.961479127892891e-05,
      "loss": 0.9766,
      "step": 46872
    },
    {
      "epoch": 4.032779833089564,
      "grad_norm": 0.921875,
      "learning_rate": 8.959934463739883e-05,
      "loss": 0.5515,
      "step": 46873
    },
    {
      "epoch": 4.032865869396885,
      "grad_norm": 0.7109375,
      "learning_rate": 8.958389919621097e-05,
      "loss": 0.3793,
      "step": 46874
    },
    {
      "epoch": 4.032951905704207,
      "grad_norm": 0.97265625,
      "learning_rate": 8.956845495541021e-05,
      "loss": 0.9575,
      "step": 46875
    },
    {
      "epoch": 4.0330379420115285,
      "grad_norm": 1.2265625,
      "learning_rate": 8.955301191504195e-05,
      "loss": 1.3447,
      "step": 46876
    },
    {
      "epoch": 4.033123978318851,
      "grad_norm": 2.4375,
      "learning_rate": 8.953757007515145e-05,
      "loss": 1.4871,
      "step": 46877
    },
    {
      "epoch": 4.033210014626173,
      "grad_norm": 0.953125,
      "learning_rate": 8.952212943578358e-05,
      "loss": 0.958,
      "step": 46878
    },
    {
      "epoch": 4.033296050933494,
      "grad_norm": 1.09375,
      "learning_rate": 8.950668999698364e-05,
      "loss": 1.4942,
      "step": 46879
    },
    {
      "epoch": 4.033382087240816,
      "grad_norm": 1.15625,
      "learning_rate": 8.949125175879692e-05,
      "loss": 0.6131,
      "step": 46880
    },
    {
      "epoch": 4.033468123548137,
      "grad_norm": 1.171875,
      "learning_rate": 8.947581472126831e-05,
      "loss": 1.9021,
      "step": 46881
    },
    {
      "epoch": 4.033554159855459,
      "grad_norm": 1.3359375,
      "learning_rate": 8.946037888444326e-05,
      "loss": 1.2395,
      "step": 46882
    },
    {
      "epoch": 4.03364019616278,
      "grad_norm": 1.7578125,
      "learning_rate": 8.944494424836658e-05,
      "loss": 1.1437,
      "step": 46883
    },
    {
      "epoch": 4.033726232470102,
      "grad_norm": 1.53125,
      "learning_rate": 8.942951081308365e-05,
      "loss": 1.2601,
      "step": 46884
    },
    {
      "epoch": 4.0338122687774245,
      "grad_norm": 0.9140625,
      "learning_rate": 8.941407857863959e-05,
      "loss": 0.6509,
      "step": 46885
    },
    {
      "epoch": 4.033898305084746,
      "grad_norm": 0.984375,
      "learning_rate": 8.939864754507942e-05,
      "loss": 1.1696,
      "step": 46886
    },
    {
      "epoch": 4.033984341392068,
      "grad_norm": 14.75,
      "learning_rate": 8.938321771244839e-05,
      "loss": 0.8057,
      "step": 46887
    },
    {
      "epoch": 4.034070377699389,
      "grad_norm": 0.984375,
      "learning_rate": 8.936778908079169e-05,
      "loss": 1.0847,
      "step": 46888
    },
    {
      "epoch": 4.034156414006711,
      "grad_norm": 1.2890625,
      "learning_rate": 8.935236165015421e-05,
      "loss": 0.6985,
      "step": 46889
    },
    {
      "epoch": 4.034242450314032,
      "grad_norm": 1.203125,
      "learning_rate": 8.933693542058119e-05,
      "loss": 1.1383,
      "step": 46890
    },
    {
      "epoch": 4.034328486621354,
      "grad_norm": 1.0859375,
      "learning_rate": 8.932151039211789e-05,
      "loss": 0.7738,
      "step": 46891
    },
    {
      "epoch": 4.034414522928676,
      "grad_norm": 1.0703125,
      "learning_rate": 8.930608656480916e-05,
      "loss": 0.6479,
      "step": 46892
    },
    {
      "epoch": 4.0345005592359975,
      "grad_norm": 4.4375,
      "learning_rate": 8.929066393870039e-05,
      "loss": 1.2393,
      "step": 46893
    },
    {
      "epoch": 4.0345865955433196,
      "grad_norm": 1.0703125,
      "learning_rate": 8.927524251383645e-05,
      "loss": 1.1187,
      "step": 46894
    },
    {
      "epoch": 4.034672631850641,
      "grad_norm": 1.421875,
      "learning_rate": 8.925982229026253e-05,
      "loss": 1.1342,
      "step": 46895
    },
    {
      "epoch": 4.034758668157963,
      "grad_norm": 3.21875,
      "learning_rate": 8.924440326802386e-05,
      "loss": 0.4948,
      "step": 46896
    },
    {
      "epoch": 4.034844704465284,
      "grad_norm": 0.76953125,
      "learning_rate": 8.92289854471653e-05,
      "loss": 0.8138,
      "step": 46897
    },
    {
      "epoch": 4.034930740772606,
      "grad_norm": 1.75,
      "learning_rate": 8.921356882773207e-05,
      "loss": 1.0733,
      "step": 46898
    },
    {
      "epoch": 4.035016777079928,
      "grad_norm": 1.0546875,
      "learning_rate": 8.91981534097694e-05,
      "loss": 0.8424,
      "step": 46899
    },
    {
      "epoch": 4.035102813387249,
      "grad_norm": 4.625,
      "learning_rate": 8.918273919332198e-05,
      "loss": 0.7607,
      "step": 46900
    },
    {
      "epoch": 4.035188849694571,
      "grad_norm": 0.84375,
      "learning_rate": 8.916732617843537e-05,
      "loss": 0.934,
      "step": 46901
    },
    {
      "epoch": 4.0352748860018925,
      "grad_norm": 1.1484375,
      "learning_rate": 8.915191436515435e-05,
      "loss": 1.0436,
      "step": 46902
    },
    {
      "epoch": 4.035360922309215,
      "grad_norm": 1.0859375,
      "learning_rate": 8.9136503753524e-05,
      "loss": 1.2709,
      "step": 46903
    },
    {
      "epoch": 4.035446958616536,
      "grad_norm": 1.1328125,
      "learning_rate": 8.912109434358967e-05,
      "loss": 1.4396,
      "step": 46904
    },
    {
      "epoch": 4.035532994923858,
      "grad_norm": 1.296875,
      "learning_rate": 8.910568613539605e-05,
      "loss": 1.9537,
      "step": 46905
    },
    {
      "epoch": 4.03561903123118,
      "grad_norm": 0.984375,
      "learning_rate": 8.909027912898843e-05,
      "loss": 1.0056,
      "step": 46906
    },
    {
      "epoch": 4.035705067538501,
      "grad_norm": 0.90625,
      "learning_rate": 8.907487332441194e-05,
      "loss": 0.9024,
      "step": 46907
    },
    {
      "epoch": 4.035791103845823,
      "grad_norm": 1.84375,
      "learning_rate": 8.905946872171139e-05,
      "loss": 1.3141,
      "step": 46908
    },
    {
      "epoch": 4.035877140153144,
      "grad_norm": 1.3046875,
      "learning_rate": 8.904406532093206e-05,
      "loss": 0.5689,
      "step": 46909
    },
    {
      "epoch": 4.035963176460466,
      "grad_norm": 1.2265625,
      "learning_rate": 8.902866312211882e-05,
      "loss": 0.9373,
      "step": 46910
    },
    {
      "epoch": 4.036049212767788,
      "grad_norm": 1.4296875,
      "learning_rate": 8.90132621253168e-05,
      "loss": 0.688,
      "step": 46911
    },
    {
      "epoch": 4.03613524907511,
      "grad_norm": 0.94921875,
      "learning_rate": 8.899786233057122e-05,
      "loss": 0.5779,
      "step": 46912
    },
    {
      "epoch": 4.036221285382432,
      "grad_norm": 1.3125,
      "learning_rate": 8.898246373792679e-05,
      "loss": 0.5471,
      "step": 46913
    },
    {
      "epoch": 4.036307321689753,
      "grad_norm": 1.4453125,
      "learning_rate": 8.896706634742873e-05,
      "loss": 0.8781,
      "step": 46914
    },
    {
      "epoch": 4.036393357997075,
      "grad_norm": 0.8046875,
      "learning_rate": 8.89516701591222e-05,
      "loss": 0.642,
      "step": 46915
    },
    {
      "epoch": 4.036479394304396,
      "grad_norm": 1.4921875,
      "learning_rate": 8.893627517305192e-05,
      "loss": 1.7507,
      "step": 46916
    },
    {
      "epoch": 4.036565430611718,
      "grad_norm": 1.21875,
      "learning_rate": 8.892088138926312e-05,
      "loss": 1.297,
      "step": 46917
    },
    {
      "epoch": 4.036651466919039,
      "grad_norm": 14.5625,
      "learning_rate": 8.890548880780092e-05,
      "loss": 0.7913,
      "step": 46918
    },
    {
      "epoch": 4.0367375032263615,
      "grad_norm": 5.09375,
      "learning_rate": 8.889009742871008e-05,
      "loss": 0.9835,
      "step": 46919
    },
    {
      "epoch": 4.036823539533684,
      "grad_norm": 1.1640625,
      "learning_rate": 8.887470725203572e-05,
      "loss": 1.3259,
      "step": 46920
    },
    {
      "epoch": 4.036909575841005,
      "grad_norm": 0.921875,
      "learning_rate": 8.885931827782307e-05,
      "loss": 1.3303,
      "step": 46921
    },
    {
      "epoch": 4.036995612148327,
      "grad_norm": 0.98046875,
      "learning_rate": 8.884393050611678e-05,
      "loss": 1.4446,
      "step": 46922
    },
    {
      "epoch": 4.037081648455648,
      "grad_norm": 0.8515625,
      "learning_rate": 8.882854393696216e-05,
      "loss": 0.7037,
      "step": 46923
    },
    {
      "epoch": 4.03716768476297,
      "grad_norm": 1.3046875,
      "learning_rate": 8.881315857040395e-05,
      "loss": 0.8133,
      "step": 46924
    },
    {
      "epoch": 4.037253721070291,
      "grad_norm": 0.83203125,
      "learning_rate": 8.879777440648728e-05,
      "loss": 0.7079,
      "step": 46925
    },
    {
      "epoch": 4.037339757377613,
      "grad_norm": 2.046875,
      "learning_rate": 8.878239144525729e-05,
      "loss": 0.6872,
      "step": 46926
    },
    {
      "epoch": 4.037425793684935,
      "grad_norm": 1.1875,
      "learning_rate": 8.876700968675872e-05,
      "loss": 1.3629,
      "step": 46927
    },
    {
      "epoch": 4.0375118299922566,
      "grad_norm": 0.9921875,
      "learning_rate": 8.875162913103663e-05,
      "loss": 0.8517,
      "step": 46928
    },
    {
      "epoch": 4.037597866299579,
      "grad_norm": 1.0078125,
      "learning_rate": 8.873624977813616e-05,
      "loss": 0.5566,
      "step": 46929
    },
    {
      "epoch": 4.0376839026069,
      "grad_norm": 1.0546875,
      "learning_rate": 8.872087162810206e-05,
      "loss": 0.8292,
      "step": 46930
    },
    {
      "epoch": 4.037769938914222,
      "grad_norm": 1.1640625,
      "learning_rate": 8.870549468097943e-05,
      "loss": 2.0114,
      "step": 46931
    },
    {
      "epoch": 4.037855975221543,
      "grad_norm": 1.1015625,
      "learning_rate": 8.86901189368134e-05,
      "loss": 0.8213,
      "step": 46932
    },
    {
      "epoch": 4.037942011528865,
      "grad_norm": 0.99609375,
      "learning_rate": 8.86747443956486e-05,
      "loss": 1.7561,
      "step": 46933
    },
    {
      "epoch": 4.038028047836187,
      "grad_norm": 1.1171875,
      "learning_rate": 8.86593710575303e-05,
      "loss": 1.1692,
      "step": 46934
    },
    {
      "epoch": 4.038114084143508,
      "grad_norm": 1.3046875,
      "learning_rate": 8.864399892250318e-05,
      "loss": 1.6131,
      "step": 46935
    },
    {
      "epoch": 4.03820012045083,
      "grad_norm": 1.4921875,
      "learning_rate": 8.862862799061244e-05,
      "loss": 0.8092,
      "step": 46936
    },
    {
      "epoch": 4.038286156758152,
      "grad_norm": 1.046875,
      "learning_rate": 8.861325826190291e-05,
      "loss": 1.2142,
      "step": 46937
    },
    {
      "epoch": 4.038372193065474,
      "grad_norm": 51.75,
      "learning_rate": 8.859788973641958e-05,
      "loss": 0.5607,
      "step": 46938
    },
    {
      "epoch": 4.038458229372796,
      "grad_norm": 0.90625,
      "learning_rate": 8.858252241420756e-05,
      "loss": 1.217,
      "step": 46939
    },
    {
      "epoch": 4.038544265680117,
      "grad_norm": 1.1171875,
      "learning_rate": 8.856715629531149e-05,
      "loss": 1.5703,
      "step": 46940
    },
    {
      "epoch": 4.038630301987439,
      "grad_norm": 1.2265625,
      "learning_rate": 8.855179137977648e-05,
      "loss": 1.137,
      "step": 46941
    },
    {
      "epoch": 4.03871633829476,
      "grad_norm": 1.59375,
      "learning_rate": 8.853642766764757e-05,
      "loss": 1.0058,
      "step": 46942
    },
    {
      "epoch": 4.038802374602082,
      "grad_norm": 3.59375,
      "learning_rate": 8.852106515896952e-05,
      "loss": 0.8886,
      "step": 46943
    },
    {
      "epoch": 4.038888410909403,
      "grad_norm": 1.296875,
      "learning_rate": 8.850570385378726e-05,
      "loss": 1.7377,
      "step": 46944
    },
    {
      "epoch": 4.0389744472167255,
      "grad_norm": 1.0703125,
      "learning_rate": 8.849034375214593e-05,
      "loss": 1.111,
      "step": 46945
    },
    {
      "epoch": 4.039060483524048,
      "grad_norm": 1.3515625,
      "learning_rate": 8.847498485409023e-05,
      "loss": 1.7781,
      "step": 46946
    },
    {
      "epoch": 4.039146519831369,
      "grad_norm": 0.98046875,
      "learning_rate": 8.84596271596651e-05,
      "loss": 0.9359,
      "step": 46947
    },
    {
      "epoch": 4.039232556138691,
      "grad_norm": 1.265625,
      "learning_rate": 8.844427066891569e-05,
      "loss": 1.2451,
      "step": 46948
    },
    {
      "epoch": 4.039318592446012,
      "grad_norm": 1.0703125,
      "learning_rate": 8.842891538188657e-05,
      "loss": 0.7593,
      "step": 46949
    },
    {
      "epoch": 4.039404628753334,
      "grad_norm": 1.015625,
      "learning_rate": 8.841356129862299e-05,
      "loss": 1.3923,
      "step": 46950
    },
    {
      "epoch": 4.039490665060655,
      "grad_norm": 1.21875,
      "learning_rate": 8.839820841916957e-05,
      "loss": 1.5708,
      "step": 46951
    },
    {
      "epoch": 4.039576701367977,
      "grad_norm": 1.0546875,
      "learning_rate": 8.838285674357133e-05,
      "loss": 1.2687,
      "step": 46952
    },
    {
      "epoch": 4.039662737675299,
      "grad_norm": 0.8515625,
      "learning_rate": 8.836750627187329e-05,
      "loss": 1.3604,
      "step": 46953
    },
    {
      "epoch": 4.0397487739826206,
      "grad_norm": 1.0390625,
      "learning_rate": 8.835215700412014e-05,
      "loss": 1.1756,
      "step": 46954
    },
    {
      "epoch": 4.039834810289943,
      "grad_norm": 1.1484375,
      "learning_rate": 8.833680894035684e-05,
      "loss": 0.8377,
      "step": 46955
    },
    {
      "epoch": 4.039920846597264,
      "grad_norm": 1.0390625,
      "learning_rate": 8.832146208062841e-05,
      "loss": 1.552,
      "step": 46956
    },
    {
      "epoch": 4.040006882904586,
      "grad_norm": 0.97265625,
      "learning_rate": 8.830611642497954e-05,
      "loss": 0.7439,
      "step": 46957
    },
    {
      "epoch": 4.040092919211907,
      "grad_norm": 0.9921875,
      "learning_rate": 8.829077197345525e-05,
      "loss": 1.1261,
      "step": 46958
    },
    {
      "epoch": 4.040178955519229,
      "grad_norm": 0.88671875,
      "learning_rate": 8.827542872610044e-05,
      "loss": 1.4452,
      "step": 46959
    },
    {
      "epoch": 4.040264991826551,
      "grad_norm": 1.2890625,
      "learning_rate": 8.826008668295987e-05,
      "loss": 1.2403,
      "step": 46960
    },
    {
      "epoch": 4.040351028133872,
      "grad_norm": 0.84765625,
      "learning_rate": 8.824474584407844e-05,
      "loss": 0.9465,
      "step": 46961
    },
    {
      "epoch": 4.040437064441194,
      "grad_norm": 4.25,
      "learning_rate": 8.822940620950115e-05,
      "loss": 0.9463,
      "step": 46962
    },
    {
      "epoch": 4.040523100748516,
      "grad_norm": 1.5546875,
      "learning_rate": 8.821406777927265e-05,
      "loss": 1.2085,
      "step": 46963
    },
    {
      "epoch": 4.040609137055838,
      "grad_norm": 1.859375,
      "learning_rate": 8.819873055343803e-05,
      "loss": 1.2611,
      "step": 46964
    },
    {
      "epoch": 4.040695173363159,
      "grad_norm": 3.203125,
      "learning_rate": 8.818339453204189e-05,
      "loss": 1.0651,
      "step": 46965
    },
    {
      "epoch": 4.040781209670481,
      "grad_norm": 0.87109375,
      "learning_rate": 8.816805971512926e-05,
      "loss": 0.6987,
      "step": 46966
    },
    {
      "epoch": 4.040867245977803,
      "grad_norm": 1.2890625,
      "learning_rate": 8.815272610274505e-05,
      "loss": 1.8049,
      "step": 46967
    },
    {
      "epoch": 4.040953282285124,
      "grad_norm": 1.125,
      "learning_rate": 8.813739369493395e-05,
      "loss": 1.697,
      "step": 46968
    },
    {
      "epoch": 4.041039318592446,
      "grad_norm": 1.25,
      "learning_rate": 8.812206249174082e-05,
      "loss": 1.1428,
      "step": 46969
    },
    {
      "epoch": 4.041125354899767,
      "grad_norm": 1.328125,
      "learning_rate": 8.810673249321071e-05,
      "loss": 1.4494,
      "step": 46970
    },
    {
      "epoch": 4.0412113912070895,
      "grad_norm": 1.3359375,
      "learning_rate": 8.809140369938812e-05,
      "loss": 1.3556,
      "step": 46971
    },
    {
      "epoch": 4.041297427514411,
      "grad_norm": 0.82421875,
      "learning_rate": 8.80760761103181e-05,
      "loss": 0.8517,
      "step": 46972
    },
    {
      "epoch": 4.041383463821733,
      "grad_norm": 1.2734375,
      "learning_rate": 8.806074972604555e-05,
      "loss": 1.6432,
      "step": 46973
    },
    {
      "epoch": 4.041469500129055,
      "grad_norm": 0.87109375,
      "learning_rate": 8.804542454661496e-05,
      "loss": 0.9321,
      "step": 46974
    },
    {
      "epoch": 4.041555536436376,
      "grad_norm": 2.6875,
      "learning_rate": 8.803010057207168e-05,
      "loss": 0.9792,
      "step": 46975
    },
    {
      "epoch": 4.041641572743698,
      "grad_norm": 1.375,
      "learning_rate": 8.801477780246003e-05,
      "loss": 0.95,
      "step": 46976
    },
    {
      "epoch": 4.041727609051019,
      "grad_norm": 1.0,
      "learning_rate": 8.799945623782508e-05,
      "loss": 1.2406,
      "step": 46977
    },
    {
      "epoch": 4.041813645358341,
      "grad_norm": 1.1640625,
      "learning_rate": 8.798413587821164e-05,
      "loss": 0.7899,
      "step": 46978
    },
    {
      "epoch": 4.0418996816656625,
      "grad_norm": 1.234375,
      "learning_rate": 8.796881672366442e-05,
      "loss": 1.5498,
      "step": 46979
    },
    {
      "epoch": 4.041985717972985,
      "grad_norm": 1.265625,
      "learning_rate": 8.795349877422837e-05,
      "loss": 0.5804,
      "step": 46980
    },
    {
      "epoch": 4.042071754280307,
      "grad_norm": 0.92578125,
      "learning_rate": 8.793818202994808e-05,
      "loss": 0.8242,
      "step": 46981
    },
    {
      "epoch": 4.042157790587628,
      "grad_norm": 0.66796875,
      "learning_rate": 8.792286649086845e-05,
      "loss": 0.398,
      "step": 46982
    },
    {
      "epoch": 4.04224382689495,
      "grad_norm": 0.98828125,
      "learning_rate": 8.790755215703445e-05,
      "loss": 0.97,
      "step": 46983
    },
    {
      "epoch": 4.042329863202271,
      "grad_norm": 1.0546875,
      "learning_rate": 8.789223902849053e-05,
      "loss": 0.8437,
      "step": 46984
    },
    {
      "epoch": 4.042415899509593,
      "grad_norm": 0.88671875,
      "learning_rate": 8.787692710528173e-05,
      "loss": 0.854,
      "step": 46985
    },
    {
      "epoch": 4.042501935816914,
      "grad_norm": 1.1328125,
      "learning_rate": 8.786161638745283e-05,
      "loss": 1.0457,
      "step": 46986
    },
    {
      "epoch": 4.042587972124236,
      "grad_norm": 1.71875,
      "learning_rate": 8.784630687504847e-05,
      "loss": 0.9251,
      "step": 46987
    },
    {
      "epoch": 4.042674008431558,
      "grad_norm": 0.9375,
      "learning_rate": 8.783099856811344e-05,
      "loss": 0.9966,
      "step": 46988
    },
    {
      "epoch": 4.04276004473888,
      "grad_norm": 0.87109375,
      "learning_rate": 8.781569146669277e-05,
      "loss": 1.6431,
      "step": 46989
    },
    {
      "epoch": 4.042846081046202,
      "grad_norm": 1.1171875,
      "learning_rate": 8.780038557083087e-05,
      "loss": 1.7181,
      "step": 46990
    },
    {
      "epoch": 4.042932117353523,
      "grad_norm": 14.1875,
      "learning_rate": 8.77850808805728e-05,
      "loss": 1.1118,
      "step": 46991
    },
    {
      "epoch": 4.043018153660845,
      "grad_norm": 63.5,
      "learning_rate": 8.776977739596309e-05,
      "loss": 0.6999,
      "step": 46992
    },
    {
      "epoch": 4.043104189968167,
      "grad_norm": 0.96484375,
      "learning_rate": 8.775447511704659e-05,
      "loss": 0.7546,
      "step": 46993
    },
    {
      "epoch": 4.043190226275488,
      "grad_norm": 0.9765625,
      "learning_rate": 8.773917404386822e-05,
      "loss": 1.2275,
      "step": 46994
    },
    {
      "epoch": 4.04327626258281,
      "grad_norm": 0.89453125,
      "learning_rate": 8.772387417647243e-05,
      "loss": 0.8714,
      "step": 46995
    },
    {
      "epoch": 4.043362298890131,
      "grad_norm": 1.2890625,
      "learning_rate": 8.770857551490419e-05,
      "loss": 0.8679,
      "step": 46996
    },
    {
      "epoch": 4.0434483351974535,
      "grad_norm": 1.65625,
      "learning_rate": 8.769327805920824e-05,
      "loss": 0.8982,
      "step": 46997
    },
    {
      "epoch": 4.043534371504775,
      "grad_norm": 0.921875,
      "learning_rate": 8.767798180942915e-05,
      "loss": 1.2694,
      "step": 46998
    },
    {
      "epoch": 4.043620407812097,
      "grad_norm": 1.3125,
      "learning_rate": 8.766268676561179e-05,
      "loss": 0.9808,
      "step": 46999
    },
    {
      "epoch": 4.043706444119419,
      "grad_norm": 1.1015625,
      "learning_rate": 8.7647392927801e-05,
      "loss": 1.0264,
      "step": 47000
    },
    {
      "epoch": 4.04379248042674,
      "grad_norm": 0.82421875,
      "learning_rate": 8.763210029604124e-05,
      "loss": 0.3828,
      "step": 47001
    },
    {
      "epoch": 4.043878516734062,
      "grad_norm": 1.0703125,
      "learning_rate": 8.761680887037737e-05,
      "loss": 1.061,
      "step": 47002
    },
    {
      "epoch": 4.043964553041383,
      "grad_norm": 0.921875,
      "learning_rate": 8.760151865085431e-05,
      "loss": 0.8211,
      "step": 47003
    },
    {
      "epoch": 4.044050589348705,
      "grad_norm": 1.1328125,
      "learning_rate": 8.758622963751645e-05,
      "loss": 1.0678,
      "step": 47004
    },
    {
      "epoch": 4.0441366256560265,
      "grad_norm": 1.40625,
      "learning_rate": 8.757094183040875e-05,
      "loss": 1.0777,
      "step": 47005
    },
    {
      "epoch": 4.044222661963349,
      "grad_norm": 1.3046875,
      "learning_rate": 8.755565522957575e-05,
      "loss": 1.2946,
      "step": 47006
    },
    {
      "epoch": 4.044308698270671,
      "grad_norm": 1.234375,
      "learning_rate": 8.754036983506225e-05,
      "loss": 1.374,
      "step": 47007
    },
    {
      "epoch": 4.044394734577992,
      "grad_norm": 1.40625,
      "learning_rate": 8.752508564691302e-05,
      "loss": 1.8179,
      "step": 47008
    },
    {
      "epoch": 4.044480770885314,
      "grad_norm": 0.96875,
      "learning_rate": 8.750980266517261e-05,
      "loss": 1.3626,
      "step": 47009
    },
    {
      "epoch": 4.044566807192635,
      "grad_norm": 1.0390625,
      "learning_rate": 8.74945208898858e-05,
      "loss": 0.7123,
      "step": 47010
    },
    {
      "epoch": 4.044652843499957,
      "grad_norm": 0.88671875,
      "learning_rate": 8.747924032109727e-05,
      "loss": 1.3174,
      "step": 47011
    },
    {
      "epoch": 4.044738879807278,
      "grad_norm": 0.73046875,
      "learning_rate": 8.746396095885178e-05,
      "loss": 0.6276,
      "step": 47012
    },
    {
      "epoch": 4.0448249161146,
      "grad_norm": 0.98046875,
      "learning_rate": 8.744868280319407e-05,
      "loss": 1.2455,
      "step": 47013
    },
    {
      "epoch": 4.0449109524219224,
      "grad_norm": 1096.0,
      "learning_rate": 8.743340585416859e-05,
      "loss": 0.8696,
      "step": 47014
    },
    {
      "epoch": 4.044996988729244,
      "grad_norm": 1.671875,
      "learning_rate": 8.741813011182014e-05,
      "loss": 1.4335,
      "step": 47015
    },
    {
      "epoch": 4.045083025036566,
      "grad_norm": 22.0,
      "learning_rate": 8.740285557619354e-05,
      "loss": 0.9662,
      "step": 47016
    },
    {
      "epoch": 4.045169061343887,
      "grad_norm": 1.1328125,
      "learning_rate": 8.738758224733323e-05,
      "loss": 0.7434,
      "step": 47017
    },
    {
      "epoch": 4.045255097651209,
      "grad_norm": 1.1640625,
      "learning_rate": 8.737231012528402e-05,
      "loss": 1.6017,
      "step": 47018
    },
    {
      "epoch": 4.04534113395853,
      "grad_norm": 12.4375,
      "learning_rate": 8.735703921009064e-05,
      "loss": 1.0482,
      "step": 47019
    },
    {
      "epoch": 4.045427170265852,
      "grad_norm": 1.3828125,
      "learning_rate": 8.734176950179756e-05,
      "loss": 1.522,
      "step": 47020
    },
    {
      "epoch": 4.045513206573174,
      "grad_norm": 1.1953125,
      "learning_rate": 8.73265010004497e-05,
      "loss": 1.0076,
      "step": 47021
    },
    {
      "epoch": 4.045599242880495,
      "grad_norm": 1.234375,
      "learning_rate": 8.73112337060914e-05,
      "loss": 1.2947,
      "step": 47022
    },
    {
      "epoch": 4.0456852791878175,
      "grad_norm": 1.1875,
      "learning_rate": 8.729596761876746e-05,
      "loss": 1.5255,
      "step": 47023
    },
    {
      "epoch": 4.045771315495139,
      "grad_norm": 1.0234375,
      "learning_rate": 8.728070273852273e-05,
      "loss": 0.7661,
      "step": 47024
    },
    {
      "epoch": 4.045857351802461,
      "grad_norm": 0.9921875,
      "learning_rate": 8.726543906540152e-05,
      "loss": 1.3752,
      "step": 47025
    },
    {
      "epoch": 4.045943388109782,
      "grad_norm": 36.25,
      "learning_rate": 8.725017659944861e-05,
      "loss": 0.8263,
      "step": 47026
    },
    {
      "epoch": 4.046029424417104,
      "grad_norm": 1.046875,
      "learning_rate": 8.723491534070883e-05,
      "loss": 1.0379,
      "step": 47027
    },
    {
      "epoch": 4.046115460724426,
      "grad_norm": 0.859375,
      "learning_rate": 8.721965528922648e-05,
      "loss": 0.9972,
      "step": 47028
    },
    {
      "epoch": 4.046201497031747,
      "grad_norm": 1.0390625,
      "learning_rate": 8.720439644504635e-05,
      "loss": 0.9995,
      "step": 47029
    },
    {
      "epoch": 4.046287533339069,
      "grad_norm": 736.0,
      "learning_rate": 8.718913880821322e-05,
      "loss": 1.1449,
      "step": 47030
    },
    {
      "epoch": 4.0463735696463905,
      "grad_norm": 1.3125,
      "learning_rate": 8.717388237877144e-05,
      "loss": 1.6101,
      "step": 47031
    },
    {
      "epoch": 4.046459605953713,
      "grad_norm": 1.21875,
      "learning_rate": 8.715862715676575e-05,
      "loss": 0.8878,
      "step": 47032
    },
    {
      "epoch": 4.046545642261034,
      "grad_norm": 1.0546875,
      "learning_rate": 8.714337314224097e-05,
      "loss": 0.7396,
      "step": 47033
    },
    {
      "epoch": 4.046631678568356,
      "grad_norm": 7.375,
      "learning_rate": 8.712812033524131e-05,
      "loss": 1.2785,
      "step": 47034
    },
    {
      "epoch": 4.046717714875678,
      "grad_norm": 1.21875,
      "learning_rate": 8.711286873581182e-05,
      "loss": 0.7314,
      "step": 47035
    },
    {
      "epoch": 4.046803751182999,
      "grad_norm": 1.234375,
      "learning_rate": 8.709761834399672e-05,
      "loss": 0.6067,
      "step": 47036
    },
    {
      "epoch": 4.046889787490321,
      "grad_norm": 1.078125,
      "learning_rate": 8.708236915984075e-05,
      "loss": 0.8615,
      "step": 47037
    },
    {
      "epoch": 4.046975823797642,
      "grad_norm": 1.34375,
      "learning_rate": 8.706712118338867e-05,
      "loss": 0.837,
      "step": 47038
    },
    {
      "epoch": 4.047061860104964,
      "grad_norm": 1.3359375,
      "learning_rate": 8.705187441468487e-05,
      "loss": 1.0993,
      "step": 47039
    },
    {
      "epoch": 4.047147896412286,
      "grad_norm": 1.0859375,
      "learning_rate": 8.7036628853774e-05,
      "loss": 0.9769,
      "step": 47040
    },
    {
      "epoch": 4.047233932719608,
      "grad_norm": 1.4375,
      "learning_rate": 8.70213845007008e-05,
      "loss": 1.5394,
      "step": 47041
    },
    {
      "epoch": 4.04731996902693,
      "grad_norm": 1.2109375,
      "learning_rate": 8.70061413555096e-05,
      "loss": 1.3433,
      "step": 47042
    },
    {
      "epoch": 4.047406005334251,
      "grad_norm": 1.171875,
      "learning_rate": 8.699089941824511e-05,
      "loss": 1.362,
      "step": 47043
    },
    {
      "epoch": 4.047492041641573,
      "grad_norm": 1.3046875,
      "learning_rate": 8.697565868895207e-05,
      "loss": 1.7835,
      "step": 47044
    },
    {
      "epoch": 4.047578077948894,
      "grad_norm": 18.25,
      "learning_rate": 8.696041916767473e-05,
      "loss": 1.6116,
      "step": 47045
    },
    {
      "epoch": 4.047664114256216,
      "grad_norm": 1.4375,
      "learning_rate": 8.694518085445796e-05,
      "loss": 0.6852,
      "step": 47046
    },
    {
      "epoch": 4.047750150563537,
      "grad_norm": 1.2265625,
      "learning_rate": 8.69299437493461e-05,
      "loss": 1.5047,
      "step": 47047
    },
    {
      "epoch": 4.047836186870859,
      "grad_norm": 1.3203125,
      "learning_rate": 8.69147078523837e-05,
      "loss": 0.9061,
      "step": 47048
    },
    {
      "epoch": 4.0479222231781815,
      "grad_norm": 0.890625,
      "learning_rate": 8.689947316361568e-05,
      "loss": 1.2169,
      "step": 47049
    },
    {
      "epoch": 4.048008259485503,
      "grad_norm": 1.5546875,
      "learning_rate": 8.688423968308623e-05,
      "loss": 1.3398,
      "step": 47050
    },
    {
      "epoch": 4.048094295792825,
      "grad_norm": 0.9140625,
      "learning_rate": 8.686900741084014e-05,
      "loss": 1.4853,
      "step": 47051
    },
    {
      "epoch": 4.048180332100146,
      "grad_norm": 1.1484375,
      "learning_rate": 8.685377634692177e-05,
      "loss": 1.4471,
      "step": 47052
    },
    {
      "epoch": 4.048266368407468,
      "grad_norm": 1.2890625,
      "learning_rate": 8.683854649137574e-05,
      "loss": 0.932,
      "step": 47053
    },
    {
      "epoch": 4.04835240471479,
      "grad_norm": 0.97265625,
      "learning_rate": 8.682331784424668e-05,
      "loss": 1.0385,
      "step": 47054
    },
    {
      "epoch": 4.048438441022111,
      "grad_norm": 0.76953125,
      "learning_rate": 8.680809040557896e-05,
      "loss": 0.5854,
      "step": 47055
    },
    {
      "epoch": 4.048524477329433,
      "grad_norm": 7.8125,
      "learning_rate": 8.679286417541727e-05,
      "loss": 1.0669,
      "step": 47056
    },
    {
      "epoch": 4.0486105136367545,
      "grad_norm": 43.5,
      "learning_rate": 8.677763915380615e-05,
      "loss": 1.5152,
      "step": 47057
    },
    {
      "epoch": 4.048696549944077,
      "grad_norm": 1.21875,
      "learning_rate": 8.676241534078994e-05,
      "loss": 1.2679,
      "step": 47058
    },
    {
      "epoch": 4.048782586251398,
      "grad_norm": 1.0703125,
      "learning_rate": 8.67471927364133e-05,
      "loss": 1.4773,
      "step": 47059
    },
    {
      "epoch": 4.04886862255872,
      "grad_norm": 1.0625,
      "learning_rate": 8.673197134072092e-05,
      "loss": 0.8526,
      "step": 47060
    },
    {
      "epoch": 4.048954658866042,
      "grad_norm": 0.75390625,
      "learning_rate": 8.6716751153757e-05,
      "loss": 0.2808,
      "step": 47061
    },
    {
      "epoch": 4.049040695173363,
      "grad_norm": 1.1015625,
      "learning_rate": 8.67015321755663e-05,
      "loss": 0.7547,
      "step": 47062
    },
    {
      "epoch": 4.049126731480685,
      "grad_norm": 1.234375,
      "learning_rate": 8.668631440619313e-05,
      "loss": 0.7905,
      "step": 47063
    },
    {
      "epoch": 4.049212767788006,
      "grad_norm": 1.09375,
      "learning_rate": 8.667109784568212e-05,
      "loss": 0.6983,
      "step": 47064
    },
    {
      "epoch": 4.049298804095328,
      "grad_norm": 0.9765625,
      "learning_rate": 8.665588249407786e-05,
      "loss": 1.1529,
      "step": 47065
    },
    {
      "epoch": 4.04938484040265,
      "grad_norm": 374.0,
      "learning_rate": 8.664066835142464e-05,
      "loss": 1.51,
      "step": 47066
    },
    {
      "epoch": 4.049470876709972,
      "grad_norm": 1.140625,
      "learning_rate": 8.662545541776706e-05,
      "loss": 0.8517,
      "step": 47067
    },
    {
      "epoch": 4.049556913017294,
      "grad_norm": 3.828125,
      "learning_rate": 8.661024369314974e-05,
      "loss": 1.4347,
      "step": 47068
    },
    {
      "epoch": 4.049642949324615,
      "grad_norm": 15.75,
      "learning_rate": 8.659503317761696e-05,
      "loss": 0.919,
      "step": 47069
    },
    {
      "epoch": 4.049728985631937,
      "grad_norm": 1.078125,
      "learning_rate": 8.657982387121333e-05,
      "loss": 1.1729,
      "step": 47070
    },
    {
      "epoch": 4.049815021939258,
      "grad_norm": 1.140625,
      "learning_rate": 8.656461577398339e-05,
      "loss": 1.2805,
      "step": 47071
    },
    {
      "epoch": 4.04990105824658,
      "grad_norm": 0.8515625,
      "learning_rate": 8.65494088859714e-05,
      "loss": 0.6377,
      "step": 47072
    },
    {
      "epoch": 4.049987094553901,
      "grad_norm": 1.2734375,
      "learning_rate": 8.6534203207222e-05,
      "loss": 1.5621,
      "step": 47073
    },
    {
      "epoch": 4.0500731308612234,
      "grad_norm": 1.0078125,
      "learning_rate": 8.65189987377798e-05,
      "loss": 1.1601,
      "step": 47074
    },
    {
      "epoch": 4.0501591671685455,
      "grad_norm": 1.0546875,
      "learning_rate": 8.650379547768894e-05,
      "loss": 0.5519,
      "step": 47075
    },
    {
      "epoch": 4.050245203475867,
      "grad_norm": 0.80078125,
      "learning_rate": 8.648859342699422e-05,
      "loss": 0.7464,
      "step": 47076
    },
    {
      "epoch": 4.050331239783189,
      "grad_norm": 0.9375,
      "learning_rate": 8.647339258573978e-05,
      "loss": 1.0031,
      "step": 47077
    },
    {
      "epoch": 4.05041727609051,
      "grad_norm": 0.80859375,
      "learning_rate": 8.645819295397023e-05,
      "loss": 0.7158,
      "step": 47078
    },
    {
      "epoch": 4.050503312397832,
      "grad_norm": 63.75,
      "learning_rate": 8.644299453173021e-05,
      "loss": 1.9337,
      "step": 47079
    },
    {
      "epoch": 4.050589348705153,
      "grad_norm": 1.21875,
      "learning_rate": 8.642779731906386e-05,
      "loss": 1.9241,
      "step": 47080
    },
    {
      "epoch": 4.050675385012475,
      "grad_norm": 1.4296875,
      "learning_rate": 8.641260131601575e-05,
      "loss": 1.3923,
      "step": 47081
    },
    {
      "epoch": 4.050761421319797,
      "grad_norm": 1.0546875,
      "learning_rate": 8.639740652263045e-05,
      "loss": 0.7704,
      "step": 47082
    },
    {
      "epoch": 4.0508474576271185,
      "grad_norm": 56.5,
      "learning_rate": 8.638221293895221e-05,
      "loss": 0.9399,
      "step": 47083
    },
    {
      "epoch": 4.050933493934441,
      "grad_norm": 1.2578125,
      "learning_rate": 8.636702056502555e-05,
      "loss": 1.2071,
      "step": 47084
    },
    {
      "epoch": 4.051019530241762,
      "grad_norm": 1.40625,
      "learning_rate": 8.635182940089486e-05,
      "loss": 0.5416,
      "step": 47085
    },
    {
      "epoch": 4.051105566549084,
      "grad_norm": 1.625,
      "learning_rate": 8.63366394466047e-05,
      "loss": 0.764,
      "step": 47086
    },
    {
      "epoch": 4.051191602856405,
      "grad_norm": 1.0078125,
      "learning_rate": 8.63214507021995e-05,
      "loss": 1.6596,
      "step": 47087
    },
    {
      "epoch": 4.051277639163727,
      "grad_norm": 1.0234375,
      "learning_rate": 8.630626316772344e-05,
      "loss": 0.7199,
      "step": 47088
    },
    {
      "epoch": 4.051363675471049,
      "grad_norm": 1.0703125,
      "learning_rate": 8.629107684322112e-05,
      "loss": 1.8915,
      "step": 47089
    },
    {
      "epoch": 4.05144971177837,
      "grad_norm": 1.0078125,
      "learning_rate": 8.627589172873713e-05,
      "loss": 1.4863,
      "step": 47090
    },
    {
      "epoch": 4.051535748085692,
      "grad_norm": 1.390625,
      "learning_rate": 8.62607078243155e-05,
      "loss": 1.1083,
      "step": 47091
    },
    {
      "epoch": 4.051621784393014,
      "grad_norm": 0.9453125,
      "learning_rate": 8.624552513000095e-05,
      "loss": 0.8381,
      "step": 47092
    },
    {
      "epoch": 4.051707820700336,
      "grad_norm": 0.953125,
      "learning_rate": 8.623034364583765e-05,
      "loss": 0.8775,
      "step": 47093
    },
    {
      "epoch": 4.051793857007657,
      "grad_norm": 0.88671875,
      "learning_rate": 8.621516337187014e-05,
      "loss": 0.7696,
      "step": 47094
    },
    {
      "epoch": 4.051879893314979,
      "grad_norm": 0.80859375,
      "learning_rate": 8.61999843081429e-05,
      "loss": 1.3216,
      "step": 47095
    },
    {
      "epoch": 4.051965929622301,
      "grad_norm": 1.0703125,
      "learning_rate": 8.618480645470017e-05,
      "loss": 1.0561,
      "step": 47096
    },
    {
      "epoch": 4.052051965929622,
      "grad_norm": 7.84375,
      "learning_rate": 8.616962981158633e-05,
      "loss": 1.4156,
      "step": 47097
    },
    {
      "epoch": 4.052138002236944,
      "grad_norm": 1.734375,
      "learning_rate": 8.6154454378846e-05,
      "loss": 1.1635,
      "step": 47098
    },
    {
      "epoch": 4.052224038544265,
      "grad_norm": 0.95703125,
      "learning_rate": 8.613928015652322e-05,
      "loss": 0.7356,
      "step": 47099
    },
    {
      "epoch": 4.0523100748515875,
      "grad_norm": 1.2578125,
      "learning_rate": 8.612410714466263e-05,
      "loss": 1.1313,
      "step": 47100
    },
    {
      "epoch": 4.052396111158909,
      "grad_norm": 1.2578125,
      "learning_rate": 8.610893534330861e-05,
      "loss": 0.958,
      "step": 47101
    },
    {
      "epoch": 4.052482147466231,
      "grad_norm": 22.875,
      "learning_rate": 8.609376475250536e-05,
      "loss": 1.0683,
      "step": 47102
    },
    {
      "epoch": 4.052568183773553,
      "grad_norm": 17.375,
      "learning_rate": 8.607859537229745e-05,
      "loss": 0.8389,
      "step": 47103
    },
    {
      "epoch": 4.052654220080874,
      "grad_norm": 7.1875,
      "learning_rate": 8.606342720272902e-05,
      "loss": 1.0624,
      "step": 47104
    },
    {
      "epoch": 4.052740256388196,
      "grad_norm": 0.890625,
      "learning_rate": 8.604826024384454e-05,
      "loss": 0.9861,
      "step": 47105
    },
    {
      "epoch": 4.052826292695517,
      "grad_norm": 0.9921875,
      "learning_rate": 8.603309449568853e-05,
      "loss": 0.6785,
      "step": 47106
    },
    {
      "epoch": 4.052912329002839,
      "grad_norm": 1.25,
      "learning_rate": 8.601792995830509e-05,
      "loss": 1.9815,
      "step": 47107
    },
    {
      "epoch": 4.052998365310161,
      "grad_norm": 1.125,
      "learning_rate": 8.600276663173873e-05,
      "loss": 0.8162,
      "step": 47108
    },
    {
      "epoch": 4.0530844016174825,
      "grad_norm": 11.125,
      "learning_rate": 8.598760451603382e-05,
      "loss": 0.6263,
      "step": 47109
    },
    {
      "epoch": 4.053170437924805,
      "grad_norm": 0.84765625,
      "learning_rate": 8.597244361123458e-05,
      "loss": 0.4902,
      "step": 47110
    },
    {
      "epoch": 4.053256474232126,
      "grad_norm": 1.015625,
      "learning_rate": 8.595728391738533e-05,
      "loss": 1.3423,
      "step": 47111
    },
    {
      "epoch": 4.053342510539448,
      "grad_norm": 0.92578125,
      "learning_rate": 8.594212543453072e-05,
      "loss": 1.0382,
      "step": 47112
    },
    {
      "epoch": 4.053428546846769,
      "grad_norm": 2.0625,
      "learning_rate": 8.592696816271468e-05,
      "loss": 0.9202,
      "step": 47113
    },
    {
      "epoch": 4.053514583154091,
      "grad_norm": 4.90625,
      "learning_rate": 8.591181210198173e-05,
      "loss": 1.6167,
      "step": 47114
    },
    {
      "epoch": 4.053600619461413,
      "grad_norm": 1.0,
      "learning_rate": 8.589665725237634e-05,
      "loss": 0.7353,
      "step": 47115
    },
    {
      "epoch": 4.053686655768734,
      "grad_norm": 1.09375,
      "learning_rate": 8.588150361394254e-05,
      "loss": 1.2112,
      "step": 47116
    },
    {
      "epoch": 4.053772692076056,
      "grad_norm": 1.1875,
      "learning_rate": 8.586635118672497e-05,
      "loss": 1.0411,
      "step": 47117
    },
    {
      "epoch": 4.053858728383378,
      "grad_norm": 0.8125,
      "learning_rate": 8.585119997076762e-05,
      "loss": 0.652,
      "step": 47118
    },
    {
      "epoch": 4.0539447646907,
      "grad_norm": 1.25,
      "learning_rate": 8.583604996611494e-05,
      "loss": 1.8969,
      "step": 47119
    },
    {
      "epoch": 4.054030800998021,
      "grad_norm": 1.203125,
      "learning_rate": 8.582090117281144e-05,
      "loss": 1.0202,
      "step": 47120
    },
    {
      "epoch": 4.054116837305343,
      "grad_norm": 0.98046875,
      "learning_rate": 8.580575359090109e-05,
      "loss": 0.5704,
      "step": 47121
    },
    {
      "epoch": 4.054202873612665,
      "grad_norm": 1.1640625,
      "learning_rate": 8.579060722042836e-05,
      "loss": 1.3001,
      "step": 47122
    },
    {
      "epoch": 4.054288909919986,
      "grad_norm": 0.98828125,
      "learning_rate": 8.577546206143754e-05,
      "loss": 1.0419,
      "step": 47123
    },
    {
      "epoch": 4.054374946227308,
      "grad_norm": 1.125,
      "learning_rate": 8.576031811397294e-05,
      "loss": 0.7368,
      "step": 47124
    },
    {
      "epoch": 4.054460982534629,
      "grad_norm": 0.9609375,
      "learning_rate": 8.574517537807896e-05,
      "loss": 0.9559,
      "step": 47125
    },
    {
      "epoch": 4.0545470188419515,
      "grad_norm": 1.3828125,
      "learning_rate": 8.573003385379968e-05,
      "loss": 1.2789,
      "step": 47126
    },
    {
      "epoch": 4.054633055149273,
      "grad_norm": 460.0,
      "learning_rate": 8.571489354117946e-05,
      "loss": 1.7216,
      "step": 47127
    },
    {
      "epoch": 4.054719091456595,
      "grad_norm": 0.9609375,
      "learning_rate": 8.569975444026274e-05,
      "loss": 1.0058,
      "step": 47128
    },
    {
      "epoch": 4.054805127763917,
      "grad_norm": 1.4296875,
      "learning_rate": 8.568461655109349e-05,
      "loss": 1.0877,
      "step": 47129
    },
    {
      "epoch": 4.054891164071238,
      "grad_norm": 1.265625,
      "learning_rate": 8.56694798737162e-05,
      "loss": 0.9999,
      "step": 47130
    },
    {
      "epoch": 4.05497720037856,
      "grad_norm": 1.28125,
      "learning_rate": 8.565434440817521e-05,
      "loss": 1.6465,
      "step": 47131
    },
    {
      "epoch": 4.055063236685881,
      "grad_norm": 2.3125,
      "learning_rate": 8.563921015451453e-05,
      "loss": 0.9664,
      "step": 47132
    },
    {
      "epoch": 4.055149272993203,
      "grad_norm": 1.421875,
      "learning_rate": 8.562407711277875e-05,
      "loss": 1.4557,
      "step": 47133
    },
    {
      "epoch": 4.0552353093005244,
      "grad_norm": 1.3203125,
      "learning_rate": 8.560894528301178e-05,
      "loss": 1.6265,
      "step": 47134
    },
    {
      "epoch": 4.0553213456078465,
      "grad_norm": 1.203125,
      "learning_rate": 8.559381466525807e-05,
      "loss": 1.143,
      "step": 47135
    },
    {
      "epoch": 4.055407381915169,
      "grad_norm": 0.96484375,
      "learning_rate": 8.557868525956197e-05,
      "loss": 1.1966,
      "step": 47136
    },
    {
      "epoch": 4.05549341822249,
      "grad_norm": 0.8203125,
      "learning_rate": 8.556355706596753e-05,
      "loss": 0.8674,
      "step": 47137
    },
    {
      "epoch": 4.055579454529812,
      "grad_norm": 0.87109375,
      "learning_rate": 8.554843008451901e-05,
      "loss": 0.7608,
      "step": 47138
    },
    {
      "epoch": 4.055665490837133,
      "grad_norm": 0.95703125,
      "learning_rate": 8.553330431526091e-05,
      "loss": 0.7,
      "step": 47139
    },
    {
      "epoch": 4.055751527144455,
      "grad_norm": 0.89453125,
      "learning_rate": 8.551817975823711e-05,
      "loss": 1.0117,
      "step": 47140
    },
    {
      "epoch": 4.055837563451776,
      "grad_norm": 1.1328125,
      "learning_rate": 8.55030564134921e-05,
      "loss": 1.1426,
      "step": 47141
    },
    {
      "epoch": 4.055923599759098,
      "grad_norm": 37.75,
      "learning_rate": 8.54879342810701e-05,
      "loss": 0.6457,
      "step": 47142
    },
    {
      "epoch": 4.05600963606642,
      "grad_norm": 0.88671875,
      "learning_rate": 8.547281336101515e-05,
      "loss": 0.5558,
      "step": 47143
    },
    {
      "epoch": 4.056095672373742,
      "grad_norm": 1.28125,
      "learning_rate": 8.545769365337175e-05,
      "loss": 1.9027,
      "step": 47144
    },
    {
      "epoch": 4.056181708681064,
      "grad_norm": 0.93359375,
      "learning_rate": 8.544257515818382e-05,
      "loss": 0.7071,
      "step": 47145
    },
    {
      "epoch": 4.056267744988385,
      "grad_norm": 0.80078125,
      "learning_rate": 8.542745787549577e-05,
      "loss": 1.0971,
      "step": 47146
    },
    {
      "epoch": 4.056353781295707,
      "grad_norm": 1.046875,
      "learning_rate": 8.541234180535185e-05,
      "loss": 1.5226,
      "step": 47147
    },
    {
      "epoch": 4.056439817603028,
      "grad_norm": 1.140625,
      "learning_rate": 8.539722694779611e-05,
      "loss": 1.3862,
      "step": 47148
    },
    {
      "epoch": 4.05652585391035,
      "grad_norm": 1.03125,
      "learning_rate": 8.538211330287282e-05,
      "loss": 1.3171,
      "step": 47149
    },
    {
      "epoch": 4.056611890217672,
      "grad_norm": 1.1953125,
      "learning_rate": 8.536700087062638e-05,
      "loss": 0.7856,
      "step": 47150
    },
    {
      "epoch": 4.056697926524993,
      "grad_norm": 84.5,
      "learning_rate": 8.535188965110064e-05,
      "loss": 1.079,
      "step": 47151
    },
    {
      "epoch": 4.0567839628323155,
      "grad_norm": 1.0859375,
      "learning_rate": 8.533677964434e-05,
      "loss": 1.0846,
      "step": 47152
    },
    {
      "epoch": 4.056869999139637,
      "grad_norm": 1.3203125,
      "learning_rate": 8.532167085038878e-05,
      "loss": 1.2548,
      "step": 47153
    },
    {
      "epoch": 4.056956035446959,
      "grad_norm": 1.46875,
      "learning_rate": 8.530656326929087e-05,
      "loss": 0.9518,
      "step": 47154
    },
    {
      "epoch": 4.05704207175428,
      "grad_norm": 18.375,
      "learning_rate": 8.529145690109058e-05,
      "loss": 1.465,
      "step": 47155
    },
    {
      "epoch": 4.057128108061602,
      "grad_norm": 1.015625,
      "learning_rate": 8.527635174583232e-05,
      "loss": 1.1507,
      "step": 47156
    },
    {
      "epoch": 4.057214144368924,
      "grad_norm": 1.3671875,
      "learning_rate": 8.526124780355987e-05,
      "loss": 1.0043,
      "step": 47157
    },
    {
      "epoch": 4.057300180676245,
      "grad_norm": 1.4453125,
      "learning_rate": 8.524614507431777e-05,
      "loss": 1.5549,
      "step": 47158
    },
    {
      "epoch": 4.057386216983567,
      "grad_norm": 1.1328125,
      "learning_rate": 8.523104355814981e-05,
      "loss": 0.7507,
      "step": 47159
    },
    {
      "epoch": 4.0574722532908885,
      "grad_norm": 2.578125,
      "learning_rate": 8.521594325510045e-05,
      "loss": 1.4746,
      "step": 47160
    },
    {
      "epoch": 4.0575582895982105,
      "grad_norm": 1.078125,
      "learning_rate": 8.520084416521395e-05,
      "loss": 1.4584,
      "step": 47161
    },
    {
      "epoch": 4.057644325905532,
      "grad_norm": 1.0078125,
      "learning_rate": 8.51857462885342e-05,
      "loss": 1.0109,
      "step": 47162
    },
    {
      "epoch": 4.057730362212854,
      "grad_norm": 1.3671875,
      "learning_rate": 8.517064962510551e-05,
      "loss": 0.5954,
      "step": 47163
    },
    {
      "epoch": 4.057816398520176,
      "grad_norm": 0.8984375,
      "learning_rate": 8.515555417497195e-05,
      "loss": 0.7289,
      "step": 47164
    },
    {
      "epoch": 4.057902434827497,
      "grad_norm": 1.2421875,
      "learning_rate": 8.514045993817765e-05,
      "loss": 0.8222,
      "step": 47165
    },
    {
      "epoch": 4.057988471134819,
      "grad_norm": 1.1171875,
      "learning_rate": 8.512536691476697e-05,
      "loss": 2.0747,
      "step": 47166
    },
    {
      "epoch": 4.05807450744214,
      "grad_norm": 1.265625,
      "learning_rate": 8.51102751047838e-05,
      "loss": 0.7386,
      "step": 47167
    },
    {
      "epoch": 4.058160543749462,
      "grad_norm": 0.75390625,
      "learning_rate": 8.509518450827236e-05,
      "loss": 1.0868,
      "step": 47168
    },
    {
      "epoch": 4.058246580056784,
      "grad_norm": 0.796875,
      "learning_rate": 8.508009512527692e-05,
      "loss": 0.4475,
      "step": 47169
    },
    {
      "epoch": 4.058332616364106,
      "grad_norm": 11.625,
      "learning_rate": 8.50650069558414e-05,
      "loss": 1.5286,
      "step": 47170
    },
    {
      "epoch": 4.058418652671428,
      "grad_norm": 1.234375,
      "learning_rate": 8.504992000001004e-05,
      "loss": 0.768,
      "step": 47171
    },
    {
      "epoch": 4.058504688978749,
      "grad_norm": 1.0234375,
      "learning_rate": 8.50348342578271e-05,
      "loss": 1.155,
      "step": 47172
    },
    {
      "epoch": 4.058590725286071,
      "grad_norm": 0.81640625,
      "learning_rate": 8.50197497293364e-05,
      "loss": 0.5949,
      "step": 47173
    },
    {
      "epoch": 4.058676761593392,
      "grad_norm": 37.75,
      "learning_rate": 8.500466641458237e-05,
      "loss": 0.6418,
      "step": 47174
    },
    {
      "epoch": 4.058762797900714,
      "grad_norm": 0.7578125,
      "learning_rate": 8.498958431360882e-05,
      "loss": 1.0376,
      "step": 47175
    },
    {
      "epoch": 4.058848834208036,
      "grad_norm": 1.1171875,
      "learning_rate": 8.497450342646007e-05,
      "loss": 1.3605,
      "step": 47176
    },
    {
      "epoch": 4.058934870515357,
      "grad_norm": 1.03125,
      "learning_rate": 8.495942375318028e-05,
      "loss": 1.3567,
      "step": 47177
    },
    {
      "epoch": 4.0590209068226795,
      "grad_norm": 1.0703125,
      "learning_rate": 8.494434529381334e-05,
      "loss": 0.9793,
      "step": 47178
    },
    {
      "epoch": 4.059106943130001,
      "grad_norm": 0.7109375,
      "learning_rate": 8.492926804840345e-05,
      "loss": 0.3834,
      "step": 47179
    },
    {
      "epoch": 4.059192979437323,
      "grad_norm": 1.03125,
      "learning_rate": 8.491419201699485e-05,
      "loss": 0.8584,
      "step": 47180
    },
    {
      "epoch": 4.059279015744644,
      "grad_norm": 1.1484375,
      "learning_rate": 8.489911719963139e-05,
      "loss": 0.8771,
      "step": 47181
    },
    {
      "epoch": 4.059365052051966,
      "grad_norm": 0.95703125,
      "learning_rate": 8.488404359635726e-05,
      "loss": 1.3336,
      "step": 47182
    },
    {
      "epoch": 4.059451088359288,
      "grad_norm": 1.1640625,
      "learning_rate": 8.48689712072167e-05,
      "loss": 1.504,
      "step": 47183
    },
    {
      "epoch": 4.059537124666609,
      "grad_norm": 1.15625,
      "learning_rate": 8.485390003225357e-05,
      "loss": 1.0786,
      "step": 47184
    },
    {
      "epoch": 4.059623160973931,
      "grad_norm": 1.296875,
      "learning_rate": 8.483883007151206e-05,
      "loss": 0.9967,
      "step": 47185
    },
    {
      "epoch": 4.0597091972812525,
      "grad_norm": 1.15625,
      "learning_rate": 8.482376132503627e-05,
      "loss": 1.5156,
      "step": 47186
    },
    {
      "epoch": 4.0597952335885745,
      "grad_norm": 1.2578125,
      "learning_rate": 8.480869379287015e-05,
      "loss": 1.0946,
      "step": 47187
    },
    {
      "epoch": 4.059881269895896,
      "grad_norm": 1.2421875,
      "learning_rate": 8.479362747505798e-05,
      "loss": 0.7783,
      "step": 47188
    },
    {
      "epoch": 4.059967306203218,
      "grad_norm": 3.96875,
      "learning_rate": 8.477856237164355e-05,
      "loss": 1.0593,
      "step": 47189
    },
    {
      "epoch": 4.06005334251054,
      "grad_norm": 1.296875,
      "learning_rate": 8.476349848267112e-05,
      "loss": 0.8525,
      "step": 47190
    },
    {
      "epoch": 4.060139378817861,
      "grad_norm": 1.6796875,
      "learning_rate": 8.474843580818476e-05,
      "loss": 0.8063,
      "step": 47191
    },
    {
      "epoch": 4.060225415125183,
      "grad_norm": 0.7578125,
      "learning_rate": 8.473337434822836e-05,
      "loss": 1.2733,
      "step": 47192
    },
    {
      "epoch": 4.060311451432504,
      "grad_norm": 1.265625,
      "learning_rate": 8.471831410284608e-05,
      "loss": 0.9221,
      "step": 47193
    },
    {
      "epoch": 4.060397487739826,
      "grad_norm": 1.2109375,
      "learning_rate": 8.47032550720821e-05,
      "loss": 1.8374,
      "step": 47194
    },
    {
      "epoch": 4.0604835240471475,
      "grad_norm": 1.0859375,
      "learning_rate": 8.46881972559802e-05,
      "loss": 0.5769,
      "step": 47195
    },
    {
      "epoch": 4.06056956035447,
      "grad_norm": 0.9609375,
      "learning_rate": 8.467314065458453e-05,
      "loss": 0.9415,
      "step": 47196
    },
    {
      "epoch": 4.060655596661792,
      "grad_norm": 1.5859375,
      "learning_rate": 8.465808526793917e-05,
      "loss": 1.3581,
      "step": 47197
    },
    {
      "epoch": 4.060741632969113,
      "grad_norm": 1.078125,
      "learning_rate": 8.464303109608812e-05,
      "loss": 0.6468,
      "step": 47198
    },
    {
      "epoch": 4.060827669276435,
      "grad_norm": 1.8203125,
      "learning_rate": 8.462797813907553e-05,
      "loss": 1.1198,
      "step": 47199
    },
    {
      "epoch": 4.060913705583756,
      "grad_norm": 1.125,
      "learning_rate": 8.461292639694518e-05,
      "loss": 0.614,
      "step": 47200
    },
    {
      "epoch": 4.060999741891078,
      "grad_norm": 0.84765625,
      "learning_rate": 8.459787586974127e-05,
      "loss": 0.8996,
      "step": 47201
    },
    {
      "epoch": 4.061085778198399,
      "grad_norm": 1.296875,
      "learning_rate": 8.458282655750793e-05,
      "loss": 0.8699,
      "step": 47202
    },
    {
      "epoch": 4.061171814505721,
      "grad_norm": 1.0234375,
      "learning_rate": 8.456777846028884e-05,
      "loss": 0.9691,
      "step": 47203
    },
    {
      "epoch": 4.0612578508130435,
      "grad_norm": 1.1953125,
      "learning_rate": 8.455273157812837e-05,
      "loss": 1.3609,
      "step": 47204
    },
    {
      "epoch": 4.061343887120365,
      "grad_norm": 1.125,
      "learning_rate": 8.453768591107019e-05,
      "loss": 0.8141,
      "step": 47205
    },
    {
      "epoch": 4.061429923427687,
      "grad_norm": 1.4921875,
      "learning_rate": 8.452264145915855e-05,
      "loss": 1.0983,
      "step": 47206
    },
    {
      "epoch": 4.061515959735008,
      "grad_norm": 0.69140625,
      "learning_rate": 8.450759822243742e-05,
      "loss": 0.4969,
      "step": 47207
    },
    {
      "epoch": 4.06160199604233,
      "grad_norm": 1.0,
      "learning_rate": 8.44925562009507e-05,
      "loss": 0.8908,
      "step": 47208
    },
    {
      "epoch": 4.061688032349651,
      "grad_norm": 0.68359375,
      "learning_rate": 8.447751539474241e-05,
      "loss": 0.4245,
      "step": 47209
    },
    {
      "epoch": 4.061774068656973,
      "grad_norm": 1.09375,
      "learning_rate": 8.446247580385669e-05,
      "loss": 1.1972,
      "step": 47210
    },
    {
      "epoch": 4.061860104964295,
      "grad_norm": 0.9296875,
      "learning_rate": 8.444743742833733e-05,
      "loss": 0.7875,
      "step": 47211
    },
    {
      "epoch": 4.0619461412716165,
      "grad_norm": 1.2265625,
      "learning_rate": 8.443240026822835e-05,
      "loss": 0.8149,
      "step": 47212
    },
    {
      "epoch": 4.0620321775789385,
      "grad_norm": 0.98046875,
      "learning_rate": 8.441736432357389e-05,
      "loss": 1.0486,
      "step": 47213
    },
    {
      "epoch": 4.06211821388626,
      "grad_norm": 1.2421875,
      "learning_rate": 8.440232959441774e-05,
      "loss": 1.5268,
      "step": 47214
    },
    {
      "epoch": 4.062204250193582,
      "grad_norm": 1.0703125,
      "learning_rate": 8.438729608080403e-05,
      "loss": 1.8803,
      "step": 47215
    },
    {
      "epoch": 4.062290286500903,
      "grad_norm": 0.95703125,
      "learning_rate": 8.437226378277652e-05,
      "loss": 0.9042,
      "step": 47216
    },
    {
      "epoch": 4.062376322808225,
      "grad_norm": 0.85546875,
      "learning_rate": 8.435723270037932e-05,
      "loss": 0.6316,
      "step": 47217
    },
    {
      "epoch": 4.062462359115547,
      "grad_norm": 1.1875,
      "learning_rate": 8.434220283365651e-05,
      "loss": 1.2425,
      "step": 47218
    },
    {
      "epoch": 4.062548395422868,
      "grad_norm": 3.1875,
      "learning_rate": 8.432717418265179e-05,
      "loss": 1.5489,
      "step": 47219
    },
    {
      "epoch": 4.06263443173019,
      "grad_norm": 1.0859375,
      "learning_rate": 8.431214674740923e-05,
      "loss": 1.0794,
      "step": 47220
    },
    {
      "epoch": 4.0627204680375115,
      "grad_norm": 4.5,
      "learning_rate": 8.429712052797289e-05,
      "loss": 0.9512,
      "step": 47221
    },
    {
      "epoch": 4.062806504344834,
      "grad_norm": 0.9765625,
      "learning_rate": 8.428209552438654e-05,
      "loss": 0.8185,
      "step": 47222
    },
    {
      "epoch": 4.062892540652156,
      "grad_norm": 1.5546875,
      "learning_rate": 8.426707173669418e-05,
      "loss": 1.239,
      "step": 47223
    },
    {
      "epoch": 4.062978576959477,
      "grad_norm": 37.5,
      "learning_rate": 8.425204916493995e-05,
      "loss": 0.6148,
      "step": 47224
    },
    {
      "epoch": 4.063064613266799,
      "grad_norm": 1.6640625,
      "learning_rate": 8.423702780916742e-05,
      "loss": 1.4082,
      "step": 47225
    },
    {
      "epoch": 4.06315064957412,
      "grad_norm": 41.0,
      "learning_rate": 8.422200766942078e-05,
      "loss": 1.471,
      "step": 47226
    },
    {
      "epoch": 4.063236685881442,
      "grad_norm": 1.5390625,
      "learning_rate": 8.420698874574401e-05,
      "loss": 1.7874,
      "step": 47227
    },
    {
      "epoch": 4.063322722188763,
      "grad_norm": 1.4296875,
      "learning_rate": 8.419197103818082e-05,
      "loss": 0.8905,
      "step": 47228
    },
    {
      "epoch": 4.063408758496085,
      "grad_norm": 1.1640625,
      "learning_rate": 8.417695454677538e-05,
      "loss": 0.6683,
      "step": 47229
    },
    {
      "epoch": 4.0634947948034075,
      "grad_norm": 1.3046875,
      "learning_rate": 8.416193927157128e-05,
      "loss": 0.7089,
      "step": 47230
    },
    {
      "epoch": 4.063580831110729,
      "grad_norm": 1.0234375,
      "learning_rate": 8.414692521261269e-05,
      "loss": 1.1687,
      "step": 47231
    },
    {
      "epoch": 4.063666867418051,
      "grad_norm": 1.1328125,
      "learning_rate": 8.41319123699436e-05,
      "loss": 1.1027,
      "step": 47232
    },
    {
      "epoch": 4.063752903725372,
      "grad_norm": 1.3359375,
      "learning_rate": 8.41169007436075e-05,
      "loss": 0.8394,
      "step": 47233
    },
    {
      "epoch": 4.063838940032694,
      "grad_norm": 1.078125,
      "learning_rate": 8.410189033364885e-05,
      "loss": 0.7494,
      "step": 47234
    },
    {
      "epoch": 4.063924976340015,
      "grad_norm": 1.203125,
      "learning_rate": 8.408688114011115e-05,
      "loss": 1.1404,
      "step": 47235
    },
    {
      "epoch": 4.064011012647337,
      "grad_norm": 2.4375,
      "learning_rate": 8.407187316303844e-05,
      "loss": 1.452,
      "step": 47236
    },
    {
      "epoch": 4.064097048954659,
      "grad_norm": 0.96875,
      "learning_rate": 8.405686640247473e-05,
      "loss": 1.3224,
      "step": 47237
    },
    {
      "epoch": 4.0641830852619805,
      "grad_norm": 0.859375,
      "learning_rate": 8.404186085846366e-05,
      "loss": 0.7472,
      "step": 47238
    },
    {
      "epoch": 4.0642691215693025,
      "grad_norm": 0.75,
      "learning_rate": 8.402685653104925e-05,
      "loss": 0.474,
      "step": 47239
    },
    {
      "epoch": 4.064355157876624,
      "grad_norm": 1.578125,
      "learning_rate": 8.401185342027551e-05,
      "loss": 1.1523,
      "step": 47240
    },
    {
      "epoch": 4.064441194183946,
      "grad_norm": 0.97265625,
      "learning_rate": 8.399685152618603e-05,
      "loss": 0.4896,
      "step": 47241
    },
    {
      "epoch": 4.064527230491267,
      "grad_norm": 5.15625,
      "learning_rate": 8.398185084882493e-05,
      "loss": 0.9428,
      "step": 47242
    },
    {
      "epoch": 4.064613266798589,
      "grad_norm": 1.1171875,
      "learning_rate": 8.396685138823606e-05,
      "loss": 1.1854,
      "step": 47243
    },
    {
      "epoch": 4.064699303105911,
      "grad_norm": 1.3984375,
      "learning_rate": 8.395185314446313e-05,
      "loss": 1.2742,
      "step": 47244
    },
    {
      "epoch": 4.064785339413232,
      "grad_norm": 1.0546875,
      "learning_rate": 8.393685611755026e-05,
      "loss": 1.4806,
      "step": 47245
    },
    {
      "epoch": 4.064871375720554,
      "grad_norm": 1.09375,
      "learning_rate": 8.392186030754101e-05,
      "loss": 1.4809,
      "step": 47246
    },
    {
      "epoch": 4.0649574120278755,
      "grad_norm": 0.921875,
      "learning_rate": 8.390686571447943e-05,
      "loss": 1.0449,
      "step": 47247
    },
    {
      "epoch": 4.065043448335198,
      "grad_norm": 1.015625,
      "learning_rate": 8.389187233840945e-05,
      "loss": 1.2045,
      "step": 47248
    },
    {
      "epoch": 4.065129484642519,
      "grad_norm": 2.234375,
      "learning_rate": 8.387688017937466e-05,
      "loss": 0.8823,
      "step": 47249
    },
    {
      "epoch": 4.065215520949841,
      "grad_norm": 17.0,
      "learning_rate": 8.386188923741911e-05,
      "loss": 1.4174,
      "step": 47250
    },
    {
      "epoch": 4.065301557257163,
      "grad_norm": 1.0078125,
      "learning_rate": 8.384689951258673e-05,
      "loss": 1.2683,
      "step": 47251
    },
    {
      "epoch": 4.065387593564484,
      "grad_norm": 1.296875,
      "learning_rate": 8.383191100492104e-05,
      "loss": 0.9463,
      "step": 47252
    },
    {
      "epoch": 4.065473629871806,
      "grad_norm": 1.03125,
      "learning_rate": 8.381692371446615e-05,
      "loss": 1.3202,
      "step": 47253
    },
    {
      "epoch": 4.065559666179127,
      "grad_norm": 1.46875,
      "learning_rate": 8.380193764126587e-05,
      "loss": 1.3128,
      "step": 47254
    },
    {
      "epoch": 4.065645702486449,
      "grad_norm": 1.0625,
      "learning_rate": 8.378695278536391e-05,
      "loss": 0.9987,
      "step": 47255
    },
    {
      "epoch": 4.065731738793771,
      "grad_norm": 1.0546875,
      "learning_rate": 8.377196914680424e-05,
      "loss": 1.0781,
      "step": 47256
    },
    {
      "epoch": 4.065817775101093,
      "grad_norm": 1.65625,
      "learning_rate": 8.375698672563054e-05,
      "loss": 1.3409,
      "step": 47257
    },
    {
      "epoch": 4.065903811408415,
      "grad_norm": 0.89453125,
      "learning_rate": 8.374200552188666e-05,
      "loss": 0.8587,
      "step": 47258
    },
    {
      "epoch": 4.065989847715736,
      "grad_norm": 1.0625,
      "learning_rate": 8.372702553561657e-05,
      "loss": 1.218,
      "step": 47259
    },
    {
      "epoch": 4.066075884023058,
      "grad_norm": 1.75,
      "learning_rate": 8.371204676686389e-05,
      "loss": 1.7704,
      "step": 47260
    },
    {
      "epoch": 4.066161920330379,
      "grad_norm": 3.953125,
      "learning_rate": 8.369706921567244e-05,
      "loss": 1.0967,
      "step": 47261
    },
    {
      "epoch": 4.066247956637701,
      "grad_norm": 0.8671875,
      "learning_rate": 8.368209288208628e-05,
      "loss": 0.7334,
      "step": 47262
    },
    {
      "epoch": 4.066333992945022,
      "grad_norm": 2.46875,
      "learning_rate": 8.366711776614888e-05,
      "loss": 1.7598,
      "step": 47263
    },
    {
      "epoch": 4.0664200292523445,
      "grad_norm": 1.1796875,
      "learning_rate": 8.365214386790421e-05,
      "loss": 1.2728,
      "step": 47264
    },
    {
      "epoch": 4.0665060655596665,
      "grad_norm": 1.34375,
      "learning_rate": 8.363717118739617e-05,
      "loss": 1.1882,
      "step": 47265
    },
    {
      "epoch": 4.066592101866988,
      "grad_norm": 0.74609375,
      "learning_rate": 8.362219972466827e-05,
      "loss": 0.4005,
      "step": 47266
    },
    {
      "epoch": 4.06667813817431,
      "grad_norm": 1.03125,
      "learning_rate": 8.360722947976446e-05,
      "loss": 0.6303,
      "step": 47267
    },
    {
      "epoch": 4.066764174481631,
      "grad_norm": 1.2578125,
      "learning_rate": 8.359226045272866e-05,
      "loss": 1.213,
      "step": 47268
    },
    {
      "epoch": 4.066850210788953,
      "grad_norm": 0.89453125,
      "learning_rate": 8.357729264360437e-05,
      "loss": 0.7538,
      "step": 47269
    },
    {
      "epoch": 4.066936247096274,
      "grad_norm": 2.546875,
      "learning_rate": 8.356232605243552e-05,
      "loss": 1.3033,
      "step": 47270
    },
    {
      "epoch": 4.067022283403596,
      "grad_norm": 0.84375,
      "learning_rate": 8.354736067926588e-05,
      "loss": 0.5722,
      "step": 47271
    },
    {
      "epoch": 4.067108319710918,
      "grad_norm": 1.0546875,
      "learning_rate": 8.353239652413925e-05,
      "loss": 1.2127,
      "step": 47272
    },
    {
      "epoch": 4.0671943560182395,
      "grad_norm": 1.59375,
      "learning_rate": 8.35174335870994e-05,
      "loss": 0.8009,
      "step": 47273
    },
    {
      "epoch": 4.067280392325562,
      "grad_norm": 1.828125,
      "learning_rate": 8.350247186818999e-05,
      "loss": 1.4059,
      "step": 47274
    },
    {
      "epoch": 4.067366428632883,
      "grad_norm": 1.234375,
      "learning_rate": 8.348751136745497e-05,
      "loss": 0.7671,
      "step": 47275
    },
    {
      "epoch": 4.067452464940205,
      "grad_norm": 2.765625,
      "learning_rate": 8.347255208493781e-05,
      "loss": 0.7064,
      "step": 47276
    },
    {
      "epoch": 4.067538501247526,
      "grad_norm": 1.0,
      "learning_rate": 8.345759402068248e-05,
      "loss": 0.8977,
      "step": 47277
    },
    {
      "epoch": 4.067624537554848,
      "grad_norm": 1.1953125,
      "learning_rate": 8.344263717473277e-05,
      "loss": 1.1425,
      "step": 47278
    },
    {
      "epoch": 4.06771057386217,
      "grad_norm": 268.0,
      "learning_rate": 8.34276815471322e-05,
      "loss": 0.7766,
      "step": 47279
    },
    {
      "epoch": 4.067796610169491,
      "grad_norm": 1.046875,
      "learning_rate": 8.341272713792464e-05,
      "loss": 0.8709,
      "step": 47280
    },
    {
      "epoch": 4.067882646476813,
      "grad_norm": 1.1328125,
      "learning_rate": 8.339777394715398e-05,
      "loss": 1.0702,
      "step": 47281
    },
    {
      "epoch": 4.067968682784135,
      "grad_norm": 1.9609375,
      "learning_rate": 8.338282197486363e-05,
      "loss": 1.515,
      "step": 47282
    },
    {
      "epoch": 4.068054719091457,
      "grad_norm": 1.2421875,
      "learning_rate": 8.336787122109757e-05,
      "loss": 0.9251,
      "step": 47283
    },
    {
      "epoch": 4.068140755398778,
      "grad_norm": 3.765625,
      "learning_rate": 8.335292168589953e-05,
      "loss": 1.2945,
      "step": 47284
    },
    {
      "epoch": 4.0682267917061,
      "grad_norm": 0.921875,
      "learning_rate": 8.333797336931304e-05,
      "loss": 0.7054,
      "step": 47285
    },
    {
      "epoch": 4.068312828013422,
      "grad_norm": 0.765625,
      "learning_rate": 8.332302627138206e-05,
      "loss": 1.0919,
      "step": 47286
    },
    {
      "epoch": 4.068398864320743,
      "grad_norm": 1.0546875,
      "learning_rate": 8.33080803921501e-05,
      "loss": 0.6714,
      "step": 47287
    },
    {
      "epoch": 4.068484900628065,
      "grad_norm": 4.4375,
      "learning_rate": 8.329313573166091e-05,
      "loss": 1.135,
      "step": 47288
    },
    {
      "epoch": 4.068570936935386,
      "grad_norm": 20.125,
      "learning_rate": 8.327819228995842e-05,
      "loss": 0.5916,
      "step": 47289
    },
    {
      "epoch": 4.0686569732427085,
      "grad_norm": 1.125,
      "learning_rate": 8.326325006708608e-05,
      "loss": 1.7673,
      "step": 47290
    },
    {
      "epoch": 4.0687430095500305,
      "grad_norm": 0.99609375,
      "learning_rate": 8.324830906308761e-05,
      "loss": 0.8131,
      "step": 47291
    },
    {
      "epoch": 4.068829045857352,
      "grad_norm": 1.15625,
      "learning_rate": 8.323336927800696e-05,
      "loss": 1.3751,
      "step": 47292
    },
    {
      "epoch": 4.068915082164674,
      "grad_norm": 1.0078125,
      "learning_rate": 8.321843071188745e-05,
      "loss": 1.1852,
      "step": 47293
    },
    {
      "epoch": 4.069001118471995,
      "grad_norm": 1.46875,
      "learning_rate": 8.320349336477306e-05,
      "loss": 1.5422,
      "step": 47294
    },
    {
      "epoch": 4.069087154779317,
      "grad_norm": 1.3125,
      "learning_rate": 8.318855723670742e-05,
      "loss": 1.1249,
      "step": 47295
    },
    {
      "epoch": 4.069173191086638,
      "grad_norm": 1.25,
      "learning_rate": 8.317362232773412e-05,
      "loss": 0.7027,
      "step": 47296
    },
    {
      "epoch": 4.06925922739396,
      "grad_norm": 114.0,
      "learning_rate": 8.315868863789694e-05,
      "loss": 1.6504,
      "step": 47297
    },
    {
      "epoch": 4.069345263701282,
      "grad_norm": 4.5625,
      "learning_rate": 8.314375616723957e-05,
      "loss": 1.1661,
      "step": 47298
    },
    {
      "epoch": 4.0694313000086035,
      "grad_norm": 1.53125,
      "learning_rate": 8.31288249158056e-05,
      "loss": 1.6587,
      "step": 47299
    },
    {
      "epoch": 4.069517336315926,
      "grad_norm": 0.984375,
      "learning_rate": 8.311389488363879e-05,
      "loss": 1.0125,
      "step": 47300
    },
    {
      "epoch": 4.069603372623247,
      "grad_norm": 1.0703125,
      "learning_rate": 8.309896607078265e-05,
      "loss": 0.5351,
      "step": 47301
    },
    {
      "epoch": 4.069689408930569,
      "grad_norm": 1.296875,
      "learning_rate": 8.308403847728096e-05,
      "loss": 1.1229,
      "step": 47302
    },
    {
      "epoch": 4.06977544523789,
      "grad_norm": 1.2109375,
      "learning_rate": 8.306911210317753e-05,
      "loss": 1.5071,
      "step": 47303
    },
    {
      "epoch": 4.069861481545212,
      "grad_norm": 0.73828125,
      "learning_rate": 8.305418694851569e-05,
      "loss": 0.5455,
      "step": 47304
    },
    {
      "epoch": 4.069947517852534,
      "grad_norm": 1.0625,
      "learning_rate": 8.30392630133393e-05,
      "loss": 1.4618,
      "step": 47305
    },
    {
      "epoch": 4.070033554159855,
      "grad_norm": 1.1171875,
      "learning_rate": 8.302434029769207e-05,
      "loss": 1.121,
      "step": 47306
    },
    {
      "epoch": 4.070119590467177,
      "grad_norm": 2.53125,
      "learning_rate": 8.300941880161728e-05,
      "loss": 1.4741,
      "step": 47307
    },
    {
      "epoch": 4.070205626774499,
      "grad_norm": 1.2890625,
      "learning_rate": 8.299449852515911e-05,
      "loss": 1.6586,
      "step": 47308
    },
    {
      "epoch": 4.070291663081821,
      "grad_norm": 1.234375,
      "learning_rate": 8.297957946836082e-05,
      "loss": 1.434,
      "step": 47309
    },
    {
      "epoch": 4.070377699389142,
      "grad_norm": 1.09375,
      "learning_rate": 8.296466163126609e-05,
      "loss": 0.641,
      "step": 47310
    },
    {
      "epoch": 4.070463735696464,
      "grad_norm": 0.92578125,
      "learning_rate": 8.294974501391884e-05,
      "loss": 0.9104,
      "step": 47311
    },
    {
      "epoch": 4.070549772003786,
      "grad_norm": 1.0,
      "learning_rate": 8.293482961636228e-05,
      "loss": 1.8072,
      "step": 47312
    },
    {
      "epoch": 4.070635808311107,
      "grad_norm": 13.0,
      "learning_rate": 8.291991543864025e-05,
      "loss": 0.9448,
      "step": 47313
    },
    {
      "epoch": 4.070721844618429,
      "grad_norm": 1.15625,
      "learning_rate": 8.290500248079646e-05,
      "loss": 1.0283,
      "step": 47314
    },
    {
      "epoch": 4.07080788092575,
      "grad_norm": 0.76171875,
      "learning_rate": 8.289009074287429e-05,
      "loss": 0.3557,
      "step": 47315
    },
    {
      "epoch": 4.0708939172330725,
      "grad_norm": 1.0703125,
      "learning_rate": 8.287518022491764e-05,
      "loss": 1.0632,
      "step": 47316
    },
    {
      "epoch": 4.070979953540394,
      "grad_norm": 1.2890625,
      "learning_rate": 8.28602709269698e-05,
      "loss": 1.2142,
      "step": 47317
    },
    {
      "epoch": 4.071065989847716,
      "grad_norm": 1.4140625,
      "learning_rate": 8.284536284907457e-05,
      "loss": 0.9622,
      "step": 47318
    },
    {
      "epoch": 4.071152026155038,
      "grad_norm": 1.171875,
      "learning_rate": 8.283045599127559e-05,
      "loss": 1.2923,
      "step": 47319
    },
    {
      "epoch": 4.071238062462359,
      "grad_norm": 0.8203125,
      "learning_rate": 8.281555035361632e-05,
      "loss": 1.0845,
      "step": 47320
    },
    {
      "epoch": 4.071324098769681,
      "grad_norm": 3.078125,
      "learning_rate": 8.28006459361404e-05,
      "loss": 0.6678,
      "step": 47321
    },
    {
      "epoch": 4.071410135077002,
      "grad_norm": 9664.0,
      "learning_rate": 8.278574273889162e-05,
      "loss": 1.0314,
      "step": 47322
    },
    {
      "epoch": 4.071496171384324,
      "grad_norm": 1.09375,
      "learning_rate": 8.277084076191321e-05,
      "loss": 0.5946,
      "step": 47323
    },
    {
      "epoch": 4.0715822076916455,
      "grad_norm": 1.2265625,
      "learning_rate": 8.2755940005249e-05,
      "loss": 1.0055,
      "step": 47324
    },
    {
      "epoch": 4.0716682439989675,
      "grad_norm": 1.2734375,
      "learning_rate": 8.27410404689426e-05,
      "loss": 1.4651,
      "step": 47325
    },
    {
      "epoch": 4.07175428030629,
      "grad_norm": 1.2734375,
      "learning_rate": 8.272614215303742e-05,
      "loss": 0.8164,
      "step": 47326
    },
    {
      "epoch": 4.071840316613611,
      "grad_norm": 1.2265625,
      "learning_rate": 8.27112450575772e-05,
      "loss": 1.3786,
      "step": 47327
    },
    {
      "epoch": 4.071926352920933,
      "grad_norm": 0.97265625,
      "learning_rate": 8.269634918260532e-05,
      "loss": 0.9311,
      "step": 47328
    },
    {
      "epoch": 4.072012389228254,
      "grad_norm": 0.8671875,
      "learning_rate": 8.26814545281655e-05,
      "loss": 0.7973,
      "step": 47329
    },
    {
      "epoch": 4.072098425535576,
      "grad_norm": 0.984375,
      "learning_rate": 8.266656109430132e-05,
      "loss": 0.5772,
      "step": 47330
    },
    {
      "epoch": 4.072184461842897,
      "grad_norm": 1.4921875,
      "learning_rate": 8.265166888105618e-05,
      "loss": 1.3192,
      "step": 47331
    },
    {
      "epoch": 4.072270498150219,
      "grad_norm": 0.81640625,
      "learning_rate": 8.263677788847374e-05,
      "loss": 0.9976,
      "step": 47332
    },
    {
      "epoch": 4.072356534457541,
      "grad_norm": 2.015625,
      "learning_rate": 8.262188811659771e-05,
      "loss": 1.0547,
      "step": 47333
    },
    {
      "epoch": 4.072442570764863,
      "grad_norm": 1.765625,
      "learning_rate": 8.26069995654713e-05,
      "loss": 1.2268,
      "step": 47334
    },
    {
      "epoch": 4.072528607072185,
      "grad_norm": 0.9609375,
      "learning_rate": 8.25921122351383e-05,
      "loss": 1.1885,
      "step": 47335
    },
    {
      "epoch": 4.072614643379506,
      "grad_norm": 6.375,
      "learning_rate": 8.257722612564228e-05,
      "loss": 1.1826,
      "step": 47336
    },
    {
      "epoch": 4.072700679686828,
      "grad_norm": 1.21875,
      "learning_rate": 8.256234123702655e-05,
      "loss": 1.1728,
      "step": 47337
    },
    {
      "epoch": 4.072786715994149,
      "grad_norm": 44.25,
      "learning_rate": 8.254745756933479e-05,
      "loss": 0.923,
      "step": 47338
    },
    {
      "epoch": 4.072872752301471,
      "grad_norm": 0.86328125,
      "learning_rate": 8.25325751226107e-05,
      "loss": 0.8856,
      "step": 47339
    },
    {
      "epoch": 4.072958788608793,
      "grad_norm": 0.984375,
      "learning_rate": 8.251769389689744e-05,
      "loss": 0.7289,
      "step": 47340
    },
    {
      "epoch": 4.073044824916114,
      "grad_norm": 24.75,
      "learning_rate": 8.25028138922389e-05,
      "loss": 0.7466,
      "step": 47341
    },
    {
      "epoch": 4.0731308612234365,
      "grad_norm": 0.8125,
      "learning_rate": 8.24879351086783e-05,
      "loss": 0.4732,
      "step": 47342
    },
    {
      "epoch": 4.073216897530758,
      "grad_norm": 1.34375,
      "learning_rate": 8.247305754625928e-05,
      "loss": 1.6343,
      "step": 47343
    },
    {
      "epoch": 4.07330293383808,
      "grad_norm": 1.4453125,
      "learning_rate": 8.245818120502535e-05,
      "loss": 0.5543,
      "step": 47344
    },
    {
      "epoch": 4.073388970145402,
      "grad_norm": 0.8125,
      "learning_rate": 8.244330608502004e-05,
      "loss": 0.6697,
      "step": 47345
    },
    {
      "epoch": 4.073475006452723,
      "grad_norm": 1.4375,
      "learning_rate": 8.242843218628698e-05,
      "loss": 1.6291,
      "step": 47346
    },
    {
      "epoch": 4.073561042760045,
      "grad_norm": 1.015625,
      "learning_rate": 8.241355950886942e-05,
      "loss": 0.7493,
      "step": 47347
    },
    {
      "epoch": 4.073647079067366,
      "grad_norm": 1.5390625,
      "learning_rate": 8.239868805281098e-05,
      "loss": 1.064,
      "step": 47348
    },
    {
      "epoch": 4.073733115374688,
      "grad_norm": 0.98828125,
      "learning_rate": 8.238381781815529e-05,
      "loss": 0.5016,
      "step": 47349
    },
    {
      "epoch": 4.0738191516820095,
      "grad_norm": 1.453125,
      "learning_rate": 8.236894880494555e-05,
      "loss": 1.5261,
      "step": 47350
    },
    {
      "epoch": 4.0739051879893315,
      "grad_norm": 1.1640625,
      "learning_rate": 8.235408101322544e-05,
      "loss": 1.1654,
      "step": 47351
    },
    {
      "epoch": 4.073991224296654,
      "grad_norm": 171.0,
      "learning_rate": 8.233921444303855e-05,
      "loss": 0.6387,
      "step": 47352
    },
    {
      "epoch": 4.074077260603975,
      "grad_norm": 1.3046875,
      "learning_rate": 8.23243490944281e-05,
      "loss": 1.5427,
      "step": 47353
    },
    {
      "epoch": 4.074163296911297,
      "grad_norm": 572.0,
      "learning_rate": 8.230948496743767e-05,
      "loss": 0.8361,
      "step": 47354
    },
    {
      "epoch": 4.074249333218618,
      "grad_norm": 0.96875,
      "learning_rate": 8.229462206211091e-05,
      "loss": 1.525,
      "step": 47355
    },
    {
      "epoch": 4.07433536952594,
      "grad_norm": 1.0,
      "learning_rate": 8.227976037849105e-05,
      "loss": 1.2594,
      "step": 47356
    },
    {
      "epoch": 4.074421405833261,
      "grad_norm": 1.09375,
      "learning_rate": 8.226489991662173e-05,
      "loss": 1.3052,
      "step": 47357
    },
    {
      "epoch": 4.074507442140583,
      "grad_norm": 1.109375,
      "learning_rate": 8.225004067654618e-05,
      "loss": 0.7269,
      "step": 47358
    },
    {
      "epoch": 4.074593478447905,
      "grad_norm": 1.0546875,
      "learning_rate": 8.223518265830804e-05,
      "loss": 1.1664,
      "step": 47359
    },
    {
      "epoch": 4.074679514755227,
      "grad_norm": 1.2578125,
      "learning_rate": 8.222032586195089e-05,
      "loss": 1.0084,
      "step": 47360
    },
    {
      "epoch": 4.074765551062549,
      "grad_norm": 0.84765625,
      "learning_rate": 8.22054702875179e-05,
      "loss": 0.7308,
      "step": 47361
    },
    {
      "epoch": 4.07485158736987,
      "grad_norm": 0.88671875,
      "learning_rate": 8.219061593505262e-05,
      "loss": 0.5693,
      "step": 47362
    },
    {
      "epoch": 4.074937623677192,
      "grad_norm": 0.9765625,
      "learning_rate": 8.217576280459865e-05,
      "loss": 1.1165,
      "step": 47363
    },
    {
      "epoch": 4.075023659984513,
      "grad_norm": 1.4765625,
      "learning_rate": 8.216091089619926e-05,
      "loss": 1.2694,
      "step": 47364
    },
    {
      "epoch": 4.075109696291835,
      "grad_norm": 1.2265625,
      "learning_rate": 8.214606020989784e-05,
      "loss": 1.2496,
      "step": 47365
    },
    {
      "epoch": 4.075195732599157,
      "grad_norm": 3.046875,
      "learning_rate": 8.213121074573815e-05,
      "loss": 1.4564,
      "step": 47366
    },
    {
      "epoch": 4.075281768906478,
      "grad_norm": 1.09375,
      "learning_rate": 8.21163625037632e-05,
      "loss": 0.736,
      "step": 47367
    },
    {
      "epoch": 4.0753678052138005,
      "grad_norm": 0.953125,
      "learning_rate": 8.210151548401679e-05,
      "loss": 0.6584,
      "step": 47368
    },
    {
      "epoch": 4.075453841521122,
      "grad_norm": 1.125,
      "learning_rate": 8.208666968654199e-05,
      "loss": 1.136,
      "step": 47369
    },
    {
      "epoch": 4.075539877828444,
      "grad_norm": 8.25,
      "learning_rate": 8.207182511138245e-05,
      "loss": 0.786,
      "step": 47370
    },
    {
      "epoch": 4.075625914135765,
      "grad_norm": 0.78125,
      "learning_rate": 8.205698175858162e-05,
      "loss": 0.7372,
      "step": 47371
    },
    {
      "epoch": 4.075711950443087,
      "grad_norm": 9.875,
      "learning_rate": 8.204213962818274e-05,
      "loss": 0.5646,
      "step": 47372
    },
    {
      "epoch": 4.075797986750409,
      "grad_norm": 1.203125,
      "learning_rate": 8.202729872022929e-05,
      "loss": 1.1912,
      "step": 47373
    },
    {
      "epoch": 4.07588402305773,
      "grad_norm": 1.2265625,
      "learning_rate": 8.20124590347649e-05,
      "loss": 1.6141,
      "step": 47374
    },
    {
      "epoch": 4.075970059365052,
      "grad_norm": 1.53125,
      "learning_rate": 8.199762057183258e-05,
      "loss": 0.9905,
      "step": 47375
    },
    {
      "epoch": 4.0760560956723735,
      "grad_norm": 1.09375,
      "learning_rate": 8.198278333147591e-05,
      "loss": 0.6857,
      "step": 47376
    },
    {
      "epoch": 4.0761421319796955,
      "grad_norm": 0.953125,
      "learning_rate": 8.196794731373846e-05,
      "loss": 1.3658,
      "step": 47377
    },
    {
      "epoch": 4.076228168287017,
      "grad_norm": 120.5,
      "learning_rate": 8.195311251866332e-05,
      "loss": 2.0482,
      "step": 47378
    },
    {
      "epoch": 4.076314204594339,
      "grad_norm": 0.96875,
      "learning_rate": 8.193827894629402e-05,
      "loss": 0.859,
      "step": 47379
    },
    {
      "epoch": 4.076400240901661,
      "grad_norm": 1.4140625,
      "learning_rate": 8.192344659667405e-05,
      "loss": 0.8409,
      "step": 47380
    },
    {
      "epoch": 4.076486277208982,
      "grad_norm": 1.3125,
      "learning_rate": 8.190861546984652e-05,
      "loss": 0.8721,
      "step": 47381
    },
    {
      "epoch": 4.076572313516304,
      "grad_norm": 1.2578125,
      "learning_rate": 8.189378556585513e-05,
      "loss": 1.5619,
      "step": 47382
    },
    {
      "epoch": 4.076658349823625,
      "grad_norm": 1.1640625,
      "learning_rate": 8.187895688474301e-05,
      "loss": 0.7912,
      "step": 47383
    },
    {
      "epoch": 4.076744386130947,
      "grad_norm": 5.875,
      "learning_rate": 8.186412942655364e-05,
      "loss": 1.5544,
      "step": 47384
    },
    {
      "epoch": 4.0768304224382685,
      "grad_norm": 1.2734375,
      "learning_rate": 8.184930319133049e-05,
      "loss": 1.0367,
      "step": 47385
    },
    {
      "epoch": 4.076916458745591,
      "grad_norm": 1.5234375,
      "learning_rate": 8.183447817911665e-05,
      "loss": 1.2869,
      "step": 47386
    },
    {
      "epoch": 4.077002495052913,
      "grad_norm": 0.984375,
      "learning_rate": 8.181965438995575e-05,
      "loss": 1.1743,
      "step": 47387
    },
    {
      "epoch": 4.077088531360234,
      "grad_norm": 1.078125,
      "learning_rate": 8.180483182389093e-05,
      "loss": 1.0657,
      "step": 47388
    },
    {
      "epoch": 4.077174567667556,
      "grad_norm": 1.09375,
      "learning_rate": 8.179001048096562e-05,
      "loss": 1.1226,
      "step": 47389
    },
    {
      "epoch": 4.077260603974877,
      "grad_norm": 1.0546875,
      "learning_rate": 8.177519036122339e-05,
      "loss": 0.8126,
      "step": 47390
    },
    {
      "epoch": 4.077346640282199,
      "grad_norm": 1.046875,
      "learning_rate": 8.176037146470716e-05,
      "loss": 0.8404,
      "step": 47391
    },
    {
      "epoch": 4.07743267658952,
      "grad_norm": 1.1953125,
      "learning_rate": 8.174555379146059e-05,
      "loss": 0.9907,
      "step": 47392
    },
    {
      "epoch": 4.077518712896842,
      "grad_norm": 0.83203125,
      "learning_rate": 8.173073734152698e-05,
      "loss": 1.1363,
      "step": 47393
    },
    {
      "epoch": 4.0776047492041645,
      "grad_norm": 1.1953125,
      "learning_rate": 8.171592211494954e-05,
      "loss": 1.2577,
      "step": 47394
    },
    {
      "epoch": 4.077690785511486,
      "grad_norm": 988.0,
      "learning_rate": 8.170110811177167e-05,
      "loss": 0.6178,
      "step": 47395
    },
    {
      "epoch": 4.077776821818808,
      "grad_norm": 0.86328125,
      "learning_rate": 8.168629533203681e-05,
      "loss": 0.6306,
      "step": 47396
    },
    {
      "epoch": 4.077862858126129,
      "grad_norm": 0.984375,
      "learning_rate": 8.167148377578804e-05,
      "loss": 1.0452,
      "step": 47397
    },
    {
      "epoch": 4.077948894433451,
      "grad_norm": 4.03125,
      "learning_rate": 8.165667344306899e-05,
      "loss": 1.2872,
      "step": 47398
    },
    {
      "epoch": 4.078034930740772,
      "grad_norm": 0.921875,
      "learning_rate": 8.164186433392267e-05,
      "loss": 0.8424,
      "step": 47399
    },
    {
      "epoch": 4.078120967048094,
      "grad_norm": 0.97265625,
      "learning_rate": 8.162705644839252e-05,
      "loss": 1.4246,
      "step": 47400
    },
    {
      "epoch": 4.078207003355416,
      "grad_norm": 1.0703125,
      "learning_rate": 8.161224978652198e-05,
      "loss": 0.7231,
      "step": 47401
    },
    {
      "epoch": 4.0782930396627375,
      "grad_norm": 1.3828125,
      "learning_rate": 8.159744434835409e-05,
      "loss": 1.0195,
      "step": 47402
    },
    {
      "epoch": 4.0783790759700596,
      "grad_norm": 1.109375,
      "learning_rate": 8.158264013393234e-05,
      "loss": 0.9054,
      "step": 47403
    },
    {
      "epoch": 4.078465112277381,
      "grad_norm": 1.109375,
      "learning_rate": 8.156783714330007e-05,
      "loss": 1.7361,
      "step": 47404
    },
    {
      "epoch": 4.078551148584703,
      "grad_norm": 1.0703125,
      "learning_rate": 8.155303537650045e-05,
      "loss": 1.7516,
      "step": 47405
    },
    {
      "epoch": 4.078637184892025,
      "grad_norm": 276.0,
      "learning_rate": 8.153823483357676e-05,
      "loss": 1.2339,
      "step": 47406
    },
    {
      "epoch": 4.078723221199346,
      "grad_norm": 0.921875,
      "learning_rate": 8.152343551457247e-05,
      "loss": 0.8522,
      "step": 47407
    },
    {
      "epoch": 4.078809257506668,
      "grad_norm": 0.9765625,
      "learning_rate": 8.150863741953063e-05,
      "loss": 1.6069,
      "step": 47408
    },
    {
      "epoch": 4.078895293813989,
      "grad_norm": 0.84765625,
      "learning_rate": 8.14938405484948e-05,
      "loss": 0.4991,
      "step": 47409
    },
    {
      "epoch": 4.078981330121311,
      "grad_norm": 1.3984375,
      "learning_rate": 8.147904490150787e-05,
      "loss": 1.1697,
      "step": 47410
    },
    {
      "epoch": 4.0790673664286325,
      "grad_norm": 1.1875,
      "learning_rate": 8.146425047861339e-05,
      "loss": 0.9411,
      "step": 47411
    },
    {
      "epoch": 4.079153402735955,
      "grad_norm": 6.65625,
      "learning_rate": 8.14494572798547e-05,
      "loss": 0.8706,
      "step": 47412
    },
    {
      "epoch": 4.079239439043277,
      "grad_norm": 1.203125,
      "learning_rate": 8.143466530527483e-05,
      "loss": 0.8288,
      "step": 47413
    },
    {
      "epoch": 4.079325475350598,
      "grad_norm": 1.3515625,
      "learning_rate": 8.141987455491717e-05,
      "loss": 1.1091,
      "step": 47414
    },
    {
      "epoch": 4.07941151165792,
      "grad_norm": 0.890625,
      "learning_rate": 8.140508502882504e-05,
      "loss": 1.0031,
      "step": 47415
    },
    {
      "epoch": 4.079497547965241,
      "grad_norm": 1.2265625,
      "learning_rate": 8.139029672704157e-05,
      "loss": 1.0545,
      "step": 47416
    },
    {
      "epoch": 4.079583584272563,
      "grad_norm": 1.4296875,
      "learning_rate": 8.137550964960999e-05,
      "loss": 1.4976,
      "step": 47417
    },
    {
      "epoch": 4.079669620579884,
      "grad_norm": 1.078125,
      "learning_rate": 8.136072379657367e-05,
      "loss": 1.07,
      "step": 47418
    },
    {
      "epoch": 4.079755656887206,
      "grad_norm": 6.6875,
      "learning_rate": 8.134593916797584e-05,
      "loss": 0.8487,
      "step": 47419
    },
    {
      "epoch": 4.0798416931945285,
      "grad_norm": 1.0625,
      "learning_rate": 8.13311557638598e-05,
      "loss": 0.6576,
      "step": 47420
    },
    {
      "epoch": 4.07992772950185,
      "grad_norm": 1.3203125,
      "learning_rate": 8.131637358426857e-05,
      "loss": 0.8576,
      "step": 47421
    },
    {
      "epoch": 4.080013765809172,
      "grad_norm": 1.203125,
      "learning_rate": 8.130159262924552e-05,
      "loss": 1.5513,
      "step": 47422
    },
    {
      "epoch": 4.080099802116493,
      "grad_norm": 1.125,
      "learning_rate": 8.128681289883406e-05,
      "loss": 0.858,
      "step": 47423
    },
    {
      "epoch": 4.080185838423815,
      "grad_norm": 1.3203125,
      "learning_rate": 8.127203439307701e-05,
      "loss": 1.4008,
      "step": 47424
    },
    {
      "epoch": 4.080271874731136,
      "grad_norm": 1.0,
      "learning_rate": 8.125725711201793e-05,
      "loss": 0.8402,
      "step": 47425
    },
    {
      "epoch": 4.080357911038458,
      "grad_norm": 1.296875,
      "learning_rate": 8.124248105569997e-05,
      "loss": 1.5682,
      "step": 47426
    },
    {
      "epoch": 4.08044394734578,
      "grad_norm": 1.1328125,
      "learning_rate": 8.122770622416625e-05,
      "loss": 1.3527,
      "step": 47427
    },
    {
      "epoch": 4.0805299836531015,
      "grad_norm": 1.2265625,
      "learning_rate": 8.12129326174601e-05,
      "loss": 0.8649,
      "step": 47428
    },
    {
      "epoch": 4.080616019960424,
      "grad_norm": 16.0,
      "learning_rate": 8.119816023562459e-05,
      "loss": 0.7427,
      "step": 47429
    },
    {
      "epoch": 4.080702056267745,
      "grad_norm": 0.90625,
      "learning_rate": 8.118338907870304e-05,
      "loss": 0.6892,
      "step": 47430
    },
    {
      "epoch": 4.080788092575067,
      "grad_norm": 1.1640625,
      "learning_rate": 8.116861914673867e-05,
      "loss": 1.4697,
      "step": 47431
    },
    {
      "epoch": 4.080874128882388,
      "grad_norm": 0.8828125,
      "learning_rate": 8.115385043977458e-05,
      "loss": 0.7873,
      "step": 47432
    },
    {
      "epoch": 4.08096016518971,
      "grad_norm": 1.3828125,
      "learning_rate": 8.113908295785393e-05,
      "loss": 1.248,
      "step": 47433
    },
    {
      "epoch": 4.081046201497032,
      "grad_norm": 1.0,
      "learning_rate": 8.112431670102022e-05,
      "loss": 1.5072,
      "step": 47434
    },
    {
      "epoch": 4.081132237804353,
      "grad_norm": 1.0546875,
      "learning_rate": 8.110955166931621e-05,
      "loss": 1.0859,
      "step": 47435
    },
    {
      "epoch": 4.081218274111675,
      "grad_norm": 1.0234375,
      "learning_rate": 8.109478786278534e-05,
      "loss": 0.938,
      "step": 47436
    },
    {
      "epoch": 4.0813043104189966,
      "grad_norm": 0.92578125,
      "learning_rate": 8.108002528147085e-05,
      "loss": 0.9296,
      "step": 47437
    },
    {
      "epoch": 4.081390346726319,
      "grad_norm": 0.94921875,
      "learning_rate": 8.106526392541569e-05,
      "loss": 0.7194,
      "step": 47438
    },
    {
      "epoch": 4.08147638303364,
      "grad_norm": 138.0,
      "learning_rate": 8.105050379466333e-05,
      "loss": 0.7368,
      "step": 47439
    },
    {
      "epoch": 4.081562419340962,
      "grad_norm": 1.296875,
      "learning_rate": 8.103574488925659e-05,
      "loss": 0.7492,
      "step": 47440
    },
    {
      "epoch": 4.081648455648284,
      "grad_norm": 26.5,
      "learning_rate": 8.102098720923879e-05,
      "loss": 1.9293,
      "step": 47441
    },
    {
      "epoch": 4.081734491955605,
      "grad_norm": 0.96484375,
      "learning_rate": 8.100623075465325e-05,
      "loss": 1.6757,
      "step": 47442
    },
    {
      "epoch": 4.081820528262927,
      "grad_norm": 1.1328125,
      "learning_rate": 8.099147552554292e-05,
      "loss": 0.6029,
      "step": 47443
    },
    {
      "epoch": 4.081906564570248,
      "grad_norm": 1.09375,
      "learning_rate": 8.097672152195096e-05,
      "loss": 1.1212,
      "step": 47444
    },
    {
      "epoch": 4.08199260087757,
      "grad_norm": 1.1875,
      "learning_rate": 8.096196874392076e-05,
      "loss": 1.0549,
      "step": 47445
    },
    {
      "epoch": 4.082078637184892,
      "grad_norm": 1.1171875,
      "learning_rate": 8.094721719149517e-05,
      "loss": 1.9646,
      "step": 47446
    },
    {
      "epoch": 4.082164673492214,
      "grad_norm": 1.5,
      "learning_rate": 8.09324668647175e-05,
      "loss": 0.7671,
      "step": 47447
    },
    {
      "epoch": 4.082250709799536,
      "grad_norm": 1.2109375,
      "learning_rate": 8.091771776363093e-05,
      "loss": 1.2421,
      "step": 47448
    },
    {
      "epoch": 4.082336746106857,
      "grad_norm": 35.75,
      "learning_rate": 8.090296988827845e-05,
      "loss": 0.9413,
      "step": 47449
    },
    {
      "epoch": 4.082422782414179,
      "grad_norm": 21.0,
      "learning_rate": 8.088822323870327e-05,
      "loss": 1.816,
      "step": 47450
    },
    {
      "epoch": 4.0825088187215,
      "grad_norm": 1.4609375,
      "learning_rate": 8.087347781494864e-05,
      "loss": 0.9411,
      "step": 47451
    },
    {
      "epoch": 4.082594855028822,
      "grad_norm": 1.71875,
      "learning_rate": 8.085873361705742e-05,
      "loss": 1.3566,
      "step": 47452
    },
    {
      "epoch": 4.082680891336143,
      "grad_norm": 1.0703125,
      "learning_rate": 8.084399064507302e-05,
      "loss": 1.2771,
      "step": 47453
    },
    {
      "epoch": 4.0827669276434655,
      "grad_norm": 1.046875,
      "learning_rate": 8.082924889903831e-05,
      "loss": 0.7272,
      "step": 47454
    },
    {
      "epoch": 4.082852963950788,
      "grad_norm": 1.4453125,
      "learning_rate": 8.081450837899656e-05,
      "loss": 0.9988,
      "step": 47455
    },
    {
      "epoch": 4.082939000258109,
      "grad_norm": 1.3046875,
      "learning_rate": 8.079976908499081e-05,
      "loss": 0.8147,
      "step": 47456
    },
    {
      "epoch": 4.083025036565431,
      "grad_norm": 0.7265625,
      "learning_rate": 8.07850310170642e-05,
      "loss": 0.5872,
      "step": 47457
    },
    {
      "epoch": 4.083111072872752,
      "grad_norm": 0.83203125,
      "learning_rate": 8.077029417525994e-05,
      "loss": 0.9177,
      "step": 47458
    },
    {
      "epoch": 4.083197109180074,
      "grad_norm": 1.0546875,
      "learning_rate": 8.075555855962097e-05,
      "loss": 1.3823,
      "step": 47459
    },
    {
      "epoch": 4.083283145487396,
      "grad_norm": 0.90234375,
      "learning_rate": 8.07408241701904e-05,
      "loss": 0.6531,
      "step": 47460
    },
    {
      "epoch": 4.083369181794717,
      "grad_norm": 6.25,
      "learning_rate": 8.072609100701156e-05,
      "loss": 0.5288,
      "step": 47461
    },
    {
      "epoch": 4.083455218102039,
      "grad_norm": 1.421875,
      "learning_rate": 8.071135907012722e-05,
      "loss": 1.6318,
      "step": 47462
    },
    {
      "epoch": 4.0835412544093606,
      "grad_norm": 1.046875,
      "learning_rate": 8.069662835958058e-05,
      "loss": 0.9559,
      "step": 47463
    },
    {
      "epoch": 4.083627290716683,
      "grad_norm": 27392.0,
      "learning_rate": 8.068189887541488e-05,
      "loss": 0.4798,
      "step": 47464
    },
    {
      "epoch": 4.083713327024004,
      "grad_norm": 3.40625,
      "learning_rate": 8.066717061767292e-05,
      "loss": 0.5979,
      "step": 47465
    },
    {
      "epoch": 4.083799363331326,
      "grad_norm": 1.171875,
      "learning_rate": 8.0652443586398e-05,
      "loss": 1.1338,
      "step": 47466
    },
    {
      "epoch": 4.083885399638648,
      "grad_norm": 0.93359375,
      "learning_rate": 8.06377177816332e-05,
      "loss": 0.7149,
      "step": 47467
    },
    {
      "epoch": 4.083971435945969,
      "grad_norm": 1.0859375,
      "learning_rate": 8.062299320342142e-05,
      "loss": 0.8584,
      "step": 47468
    },
    {
      "epoch": 4.084057472253291,
      "grad_norm": 56.5,
      "learning_rate": 8.060826985180591e-05,
      "loss": 1.2319,
      "step": 47469
    },
    {
      "epoch": 4.084143508560612,
      "grad_norm": 1.109375,
      "learning_rate": 8.059354772682953e-05,
      "loss": 0.8673,
      "step": 47470
    },
    {
      "epoch": 4.084229544867934,
      "grad_norm": 1.1484375,
      "learning_rate": 8.057882682853546e-05,
      "loss": 0.9225,
      "step": 47471
    },
    {
      "epoch": 4.084315581175256,
      "grad_norm": 1.5546875,
      "learning_rate": 8.056410715696688e-05,
      "loss": 1.5483,
      "step": 47472
    },
    {
      "epoch": 4.084401617482578,
      "grad_norm": 0.9140625,
      "learning_rate": 8.05493887121666e-05,
      "loss": 1.1979,
      "step": 47473
    },
    {
      "epoch": 4.0844876537899,
      "grad_norm": 1.8671875,
      "learning_rate": 8.053467149417775e-05,
      "loss": 0.81,
      "step": 47474
    },
    {
      "epoch": 4.084573690097221,
      "grad_norm": 0.9921875,
      "learning_rate": 8.051995550304353e-05,
      "loss": 1.0686,
      "step": 47475
    },
    {
      "epoch": 4.084659726404543,
      "grad_norm": 0.88671875,
      "learning_rate": 8.050524073880671e-05,
      "loss": 1.3251,
      "step": 47476
    },
    {
      "epoch": 4.084745762711864,
      "grad_norm": 1.109375,
      "learning_rate": 8.049052720151051e-05,
      "loss": 0.7276,
      "step": 47477
    },
    {
      "epoch": 4.084831799019186,
      "grad_norm": 1.140625,
      "learning_rate": 8.047581489119809e-05,
      "loss": 1.2144,
      "step": 47478
    },
    {
      "epoch": 4.084917835326507,
      "grad_norm": 1.3359375,
      "learning_rate": 8.046110380791211e-05,
      "loss": 1.7459,
      "step": 47479
    },
    {
      "epoch": 4.0850038716338295,
      "grad_norm": 1.3046875,
      "learning_rate": 8.044639395169601e-05,
      "loss": 1.4742,
      "step": 47480
    },
    {
      "epoch": 4.085089907941152,
      "grad_norm": 0.8984375,
      "learning_rate": 8.043168532259243e-05,
      "loss": 0.6846,
      "step": 47481
    },
    {
      "epoch": 4.085175944248473,
      "grad_norm": 1.0390625,
      "learning_rate": 8.041697792064457e-05,
      "loss": 0.6341,
      "step": 47482
    },
    {
      "epoch": 4.085261980555795,
      "grad_norm": 0.89453125,
      "learning_rate": 8.04022717458956e-05,
      "loss": 0.7268,
      "step": 47483
    },
    {
      "epoch": 4.085348016863116,
      "grad_norm": 46848.0,
      "learning_rate": 8.038756679838821e-05,
      "loss": 1.4684,
      "step": 47484
    },
    {
      "epoch": 4.085434053170438,
      "grad_norm": 20.5,
      "learning_rate": 8.03728630781656e-05,
      "loss": 0.8796,
      "step": 47485
    },
    {
      "epoch": 4.085520089477759,
      "grad_norm": 0.984375,
      "learning_rate": 8.03581605852709e-05,
      "loss": 1.4309,
      "step": 47486
    },
    {
      "epoch": 4.085606125785081,
      "grad_norm": 0.94140625,
      "learning_rate": 8.034345931974679e-05,
      "loss": 0.6576,
      "step": 47487
    },
    {
      "epoch": 4.085692162092403,
      "grad_norm": 0.9921875,
      "learning_rate": 8.032875928163653e-05,
      "loss": 0.8715,
      "step": 47488
    },
    {
      "epoch": 4.085778198399725,
      "grad_norm": 1.0703125,
      "learning_rate": 8.031406047098306e-05,
      "loss": 1.5632,
      "step": 47489
    },
    {
      "epoch": 4.085864234707047,
      "grad_norm": 1.09375,
      "learning_rate": 8.029936288782924e-05,
      "loss": 0.8688,
      "step": 47490
    },
    {
      "epoch": 4.085950271014368,
      "grad_norm": 6.15625,
      "learning_rate": 8.028466653221816e-05,
      "loss": 1.7576,
      "step": 47491
    },
    {
      "epoch": 4.08603630732169,
      "grad_norm": 1.375,
      "learning_rate": 8.026997140419284e-05,
      "loss": 0.7841,
      "step": 47492
    },
    {
      "epoch": 4.086122343629011,
      "grad_norm": 1.1328125,
      "learning_rate": 8.025527750379618e-05,
      "loss": 1.3444,
      "step": 47493
    },
    {
      "epoch": 4.086208379936333,
      "grad_norm": 1.046875,
      "learning_rate": 8.024058483107132e-05,
      "loss": 0.8789,
      "step": 47494
    },
    {
      "epoch": 4.086294416243655,
      "grad_norm": 1.1875,
      "learning_rate": 8.022589338606101e-05,
      "loss": 0.8407,
      "step": 47495
    },
    {
      "epoch": 4.086380452550976,
      "grad_norm": 0.97265625,
      "learning_rate": 8.021120316880842e-05,
      "loss": 1.3623,
      "step": 47496
    },
    {
      "epoch": 4.086466488858298,
      "grad_norm": 0.7734375,
      "learning_rate": 8.019651417935631e-05,
      "loss": 0.5249,
      "step": 47497
    },
    {
      "epoch": 4.08655252516562,
      "grad_norm": 1.046875,
      "learning_rate": 8.018182641774774e-05,
      "loss": 1.0393,
      "step": 47498
    },
    {
      "epoch": 4.086638561472942,
      "grad_norm": 12.375,
      "learning_rate": 8.01671398840258e-05,
      "loss": 0.4483,
      "step": 47499
    },
    {
      "epoch": 4.086724597780263,
      "grad_norm": 1.140625,
      "learning_rate": 8.015245457823317e-05,
      "loss": 1.073,
      "step": 47500
    },
    {
      "epoch": 4.086810634087585,
      "grad_norm": 1.78125,
      "learning_rate": 8.013777050041299e-05,
      "loss": 1.3484,
      "step": 47501
    },
    {
      "epoch": 4.086896670394907,
      "grad_norm": 7.09375,
      "learning_rate": 8.012308765060827e-05,
      "loss": 1.1004,
      "step": 47502
    },
    {
      "epoch": 4.086982706702228,
      "grad_norm": 1.1015625,
      "learning_rate": 8.010840602886176e-05,
      "loss": 0.798,
      "step": 47503
    },
    {
      "epoch": 4.08706874300955,
      "grad_norm": 1.1875,
      "learning_rate": 8.00937256352165e-05,
      "loss": 1.0585,
      "step": 47504
    },
    {
      "epoch": 4.087154779316871,
      "grad_norm": 1.125,
      "learning_rate": 8.007904646971553e-05,
      "loss": 0.6193,
      "step": 47505
    },
    {
      "epoch": 4.0872408156241935,
      "grad_norm": 61.25,
      "learning_rate": 8.006436853240161e-05,
      "loss": 0.9753,
      "step": 47506
    },
    {
      "epoch": 4.087326851931515,
      "grad_norm": 218.0,
      "learning_rate": 8.004969182331767e-05,
      "loss": 0.899,
      "step": 47507
    },
    {
      "epoch": 4.087412888238837,
      "grad_norm": 1.1875,
      "learning_rate": 8.003501634250687e-05,
      "loss": 1.1635,
      "step": 47508
    },
    {
      "epoch": 4.087498924546159,
      "grad_norm": 1.09375,
      "learning_rate": 8.002034209001186e-05,
      "loss": 1.2792,
      "step": 47509
    },
    {
      "epoch": 4.08758496085348,
      "grad_norm": 0.9765625,
      "learning_rate": 8.000566906587576e-05,
      "loss": 1.6216,
      "step": 47510
    },
    {
      "epoch": 4.087670997160802,
      "grad_norm": 1.0234375,
      "learning_rate": 7.999099727014125e-05,
      "loss": 0.8949,
      "step": 47511
    },
    {
      "epoch": 4.087757033468123,
      "grad_norm": 2.046875,
      "learning_rate": 7.997632670285137e-05,
      "loss": 0.7807,
      "step": 47512
    },
    {
      "epoch": 4.087843069775445,
      "grad_norm": 0.94140625,
      "learning_rate": 7.996165736404925e-05,
      "loss": 1.6568,
      "step": 47513
    },
    {
      "epoch": 4.0879291060827665,
      "grad_norm": 0.9609375,
      "learning_rate": 7.994698925377742e-05,
      "loss": 1.3604,
      "step": 47514
    },
    {
      "epoch": 4.088015142390089,
      "grad_norm": 1.0703125,
      "learning_rate": 7.993232237207892e-05,
      "loss": 1.089,
      "step": 47515
    },
    {
      "epoch": 4.088101178697411,
      "grad_norm": 0.94140625,
      "learning_rate": 7.991765671899687e-05,
      "loss": 0.8592,
      "step": 47516
    },
    {
      "epoch": 4.088187215004732,
      "grad_norm": 1.21875,
      "learning_rate": 7.990299229457382e-05,
      "loss": 2.074,
      "step": 47517
    },
    {
      "epoch": 4.088273251312054,
      "grad_norm": 0.87890625,
      "learning_rate": 7.98883290988528e-05,
      "loss": 0.5518,
      "step": 47518
    },
    {
      "epoch": 4.088359287619375,
      "grad_norm": 5.09375,
      "learning_rate": 7.987366713187683e-05,
      "loss": 1.0011,
      "step": 47519
    },
    {
      "epoch": 4.088445323926697,
      "grad_norm": 1.3203125,
      "learning_rate": 7.985900639368854e-05,
      "loss": 2.057,
      "step": 47520
    },
    {
      "epoch": 4.088531360234018,
      "grad_norm": 0.8984375,
      "learning_rate": 7.984434688433106e-05,
      "loss": 1.0441,
      "step": 47521
    },
    {
      "epoch": 4.08861739654134,
      "grad_norm": 0.91796875,
      "learning_rate": 7.982968860384709e-05,
      "loss": 0.9969,
      "step": 47522
    },
    {
      "epoch": 4.0887034328486624,
      "grad_norm": 1.0703125,
      "learning_rate": 7.981503155227954e-05,
      "loss": 1.2867,
      "step": 47523
    },
    {
      "epoch": 4.088789469155984,
      "grad_norm": 1.140625,
      "learning_rate": 7.980037572967142e-05,
      "loss": 1.2855,
      "step": 47524
    },
    {
      "epoch": 4.088875505463306,
      "grad_norm": 0.99609375,
      "learning_rate": 7.97857211360653e-05,
      "loss": 0.804,
      "step": 47525
    },
    {
      "epoch": 4.088961541770627,
      "grad_norm": 0.87890625,
      "learning_rate": 7.977106777150428e-05,
      "loss": 0.7144,
      "step": 47526
    },
    {
      "epoch": 4.089047578077949,
      "grad_norm": 0.953125,
      "learning_rate": 7.975641563603125e-05,
      "loss": 0.7091,
      "step": 47527
    },
    {
      "epoch": 4.089133614385271,
      "grad_norm": 4.03125,
      "learning_rate": 7.974176472968875e-05,
      "loss": 0.6407,
      "step": 47528
    },
    {
      "epoch": 4.089219650692592,
      "grad_norm": 1.1953125,
      "learning_rate": 7.972711505252006e-05,
      "loss": 0.8018,
      "step": 47529
    },
    {
      "epoch": 4.089305686999914,
      "grad_norm": 0.68359375,
      "learning_rate": 7.971246660456777e-05,
      "loss": 0.6853,
      "step": 47530
    },
    {
      "epoch": 4.089391723307235,
      "grad_norm": 0.9140625,
      "learning_rate": 7.969781938587467e-05,
      "loss": 0.5852,
      "step": 47531
    },
    {
      "epoch": 4.0894777596145575,
      "grad_norm": 0.9609375,
      "learning_rate": 7.96831733964839e-05,
      "loss": 1.2075,
      "step": 47532
    },
    {
      "epoch": 4.089563795921879,
      "grad_norm": 1.1171875,
      "learning_rate": 7.966852863643798e-05,
      "loss": 0.8803,
      "step": 47533
    },
    {
      "epoch": 4.089649832229201,
      "grad_norm": 1.1640625,
      "learning_rate": 7.965388510577981e-05,
      "loss": 0.7967,
      "step": 47534
    },
    {
      "epoch": 4.089735868536523,
      "grad_norm": 1.140625,
      "learning_rate": 7.963924280455248e-05,
      "loss": 1.3237,
      "step": 47535
    },
    {
      "epoch": 4.089821904843844,
      "grad_norm": 1.234375,
      "learning_rate": 7.962460173279839e-05,
      "loss": 0.7464,
      "step": 47536
    },
    {
      "epoch": 4.089907941151166,
      "grad_norm": 1.4296875,
      "learning_rate": 7.960996189056063e-05,
      "loss": 1.6653,
      "step": 47537
    },
    {
      "epoch": 4.089993977458487,
      "grad_norm": 1.109375,
      "learning_rate": 7.95953232778821e-05,
      "loss": 0.7399,
      "step": 47538
    },
    {
      "epoch": 4.090080013765809,
      "grad_norm": 1.34375,
      "learning_rate": 7.958068589480538e-05,
      "loss": 0.8547,
      "step": 47539
    },
    {
      "epoch": 4.0901660500731305,
      "grad_norm": 1.0390625,
      "learning_rate": 7.956604974137349e-05,
      "loss": 0.8065,
      "step": 47540
    },
    {
      "epoch": 4.090252086380453,
      "grad_norm": 2.984375,
      "learning_rate": 7.955141481762902e-05,
      "loss": 0.9474,
      "step": 47541
    },
    {
      "epoch": 4.090338122687775,
      "grad_norm": 1.0078125,
      "learning_rate": 7.953678112361484e-05,
      "loss": 1.1367,
      "step": 47542
    },
    {
      "epoch": 4.090424158995096,
      "grad_norm": 1.2265625,
      "learning_rate": 7.952214865937396e-05,
      "loss": 1.1211,
      "step": 47543
    },
    {
      "epoch": 4.090510195302418,
      "grad_norm": 1.203125,
      "learning_rate": 7.950751742494888e-05,
      "loss": 0.8212,
      "step": 47544
    },
    {
      "epoch": 4.090596231609739,
      "grad_norm": 1.546875,
      "learning_rate": 7.949288742038252e-05,
      "loss": 1.7605,
      "step": 47545
    },
    {
      "epoch": 4.090682267917061,
      "grad_norm": 0.9375,
      "learning_rate": 7.947825864571784e-05,
      "loss": 0.8787,
      "step": 47546
    },
    {
      "epoch": 4.090768304224382,
      "grad_norm": 0.9140625,
      "learning_rate": 7.946363110099736e-05,
      "loss": 0.8526,
      "step": 47547
    },
    {
      "epoch": 4.090854340531704,
      "grad_norm": 7.6875,
      "learning_rate": 7.944900478626393e-05,
      "loss": 1.7361,
      "step": 47548
    },
    {
      "epoch": 4.0909403768390264,
      "grad_norm": 52.75,
      "learning_rate": 7.943437970156048e-05,
      "loss": 0.8543,
      "step": 47549
    },
    {
      "epoch": 4.091026413146348,
      "grad_norm": 0.92578125,
      "learning_rate": 7.941975584692957e-05,
      "loss": 0.6231,
      "step": 47550
    },
    {
      "epoch": 4.09111244945367,
      "grad_norm": 1.015625,
      "learning_rate": 7.940513322241422e-05,
      "loss": 0.6129,
      "step": 47551
    },
    {
      "epoch": 4.091198485760991,
      "grad_norm": 1.3671875,
      "learning_rate": 7.93905118280569e-05,
      "loss": 1.2757,
      "step": 47552
    },
    {
      "epoch": 4.091284522068313,
      "grad_norm": 1.125,
      "learning_rate": 7.937589166390053e-05,
      "loss": 1.6705,
      "step": 47553
    },
    {
      "epoch": 4.091370558375634,
      "grad_norm": 1.46875,
      "learning_rate": 7.9361272729988e-05,
      "loss": 0.9396,
      "step": 47554
    },
    {
      "epoch": 4.091456594682956,
      "grad_norm": 1.140625,
      "learning_rate": 7.934665502636179e-05,
      "loss": 0.8183,
      "step": 47555
    },
    {
      "epoch": 4.091542630990278,
      "grad_norm": 1.4609375,
      "learning_rate": 7.933203855306487e-05,
      "loss": 1.1592,
      "step": 47556
    },
    {
      "epoch": 4.091628667297599,
      "grad_norm": 0.91796875,
      "learning_rate": 7.931742331013997e-05,
      "loss": 0.554,
      "step": 47557
    },
    {
      "epoch": 4.0917147036049215,
      "grad_norm": 1.140625,
      "learning_rate": 7.930280929762973e-05,
      "loss": 1.5319,
      "step": 47558
    },
    {
      "epoch": 4.091800739912243,
      "grad_norm": 1.3046875,
      "learning_rate": 7.92881965155769e-05,
      "loss": 1.1193,
      "step": 47559
    },
    {
      "epoch": 4.091886776219565,
      "grad_norm": 1.1015625,
      "learning_rate": 7.927358496402443e-05,
      "loss": 0.6574,
      "step": 47560
    },
    {
      "epoch": 4.091972812526886,
      "grad_norm": 1.1640625,
      "learning_rate": 7.925897464301474e-05,
      "loss": 0.8832,
      "step": 47561
    },
    {
      "epoch": 4.092058848834208,
      "grad_norm": 1.2421875,
      "learning_rate": 7.924436555259084e-05,
      "loss": 1.1356,
      "step": 47562
    },
    {
      "epoch": 4.09214488514153,
      "grad_norm": 1.78125,
      "learning_rate": 7.922975769279522e-05,
      "loss": 1.2202,
      "step": 47563
    },
    {
      "epoch": 4.092230921448851,
      "grad_norm": 1.2265625,
      "learning_rate": 7.921515106367077e-05,
      "loss": 0.7962,
      "step": 47564
    },
    {
      "epoch": 4.092316957756173,
      "grad_norm": 3.421875,
      "learning_rate": 7.920054566526014e-05,
      "loss": 1.1675,
      "step": 47565
    },
    {
      "epoch": 4.0924029940634945,
      "grad_norm": 1.1796875,
      "learning_rate": 7.918594149760605e-05,
      "loss": 1.5014,
      "step": 47566
    },
    {
      "epoch": 4.092489030370817,
      "grad_norm": 1.671875,
      "learning_rate": 7.917133856075137e-05,
      "loss": 1.9211,
      "step": 47567
    },
    {
      "epoch": 4.092575066678138,
      "grad_norm": 0.8359375,
      "learning_rate": 7.915673685473856e-05,
      "loss": 0.9477,
      "step": 47568
    },
    {
      "epoch": 4.09266110298546,
      "grad_norm": 0.91015625,
      "learning_rate": 7.914213637961048e-05,
      "loss": 1.0043,
      "step": 47569
    },
    {
      "epoch": 4.092747139292782,
      "grad_norm": 0.97265625,
      "learning_rate": 7.912753713540987e-05,
      "loss": 0.9305,
      "step": 47570
    },
    {
      "epoch": 4.092833175600103,
      "grad_norm": 354.0,
      "learning_rate": 7.911293912217926e-05,
      "loss": 1.1068,
      "step": 47571
    },
    {
      "epoch": 4.092919211907425,
      "grad_norm": 1.078125,
      "learning_rate": 7.909834233996144e-05,
      "loss": 1.691,
      "step": 47572
    },
    {
      "epoch": 4.093005248214746,
      "grad_norm": 0.96875,
      "learning_rate": 7.90837467887992e-05,
      "loss": 1.0203,
      "step": 47573
    },
    {
      "epoch": 4.093091284522068,
      "grad_norm": 0.9765625,
      "learning_rate": 7.906915246873508e-05,
      "loss": 1.3592,
      "step": 47574
    },
    {
      "epoch": 4.0931773208293905,
      "grad_norm": 1.09375,
      "learning_rate": 7.905455937981176e-05,
      "loss": 0.9063,
      "step": 47575
    },
    {
      "epoch": 4.093263357136712,
      "grad_norm": 4.53125,
      "learning_rate": 7.903996752207216e-05,
      "loss": 0.8485,
      "step": 47576
    },
    {
      "epoch": 4.093349393444034,
      "grad_norm": 1.375,
      "learning_rate": 7.902537689555861e-05,
      "loss": 1.0123,
      "step": 47577
    },
    {
      "epoch": 4.093435429751355,
      "grad_norm": 1.5703125,
      "learning_rate": 7.901078750031399e-05,
      "loss": 1.6919,
      "step": 47578
    },
    {
      "epoch": 4.093521466058677,
      "grad_norm": 1.2265625,
      "learning_rate": 7.899619933638107e-05,
      "loss": 1.068,
      "step": 47579
    },
    {
      "epoch": 4.093607502365998,
      "grad_norm": 1.0625,
      "learning_rate": 7.898161240380225e-05,
      "loss": 1.3241,
      "step": 47580
    },
    {
      "epoch": 4.09369353867332,
      "grad_norm": 1.125,
      "learning_rate": 7.896702670262045e-05,
      "loss": 0.8336,
      "step": 47581
    },
    {
      "epoch": 4.093779574980642,
      "grad_norm": 1.109375,
      "learning_rate": 7.895244223287807e-05,
      "loss": 1.3806,
      "step": 47582
    },
    {
      "epoch": 4.0938656112879634,
      "grad_norm": 1.9765625,
      "learning_rate": 7.893785899461792e-05,
      "loss": 1.0662,
      "step": 47583
    },
    {
      "epoch": 4.0939516475952855,
      "grad_norm": 0.92578125,
      "learning_rate": 7.892327698788276e-05,
      "loss": 0.5842,
      "step": 47584
    },
    {
      "epoch": 4.094037683902607,
      "grad_norm": 0.9375,
      "learning_rate": 7.890869621271501e-05,
      "loss": 0.5003,
      "step": 47585
    },
    {
      "epoch": 4.094123720209929,
      "grad_norm": 1.2578125,
      "learning_rate": 7.889411666915747e-05,
      "loss": 1.5059,
      "step": 47586
    },
    {
      "epoch": 4.09420975651725,
      "grad_norm": 1.4375,
      "learning_rate": 7.887953835725281e-05,
      "loss": 1.2821,
      "step": 47587
    },
    {
      "epoch": 4.094295792824572,
      "grad_norm": 2.515625,
      "learning_rate": 7.886496127704346e-05,
      "loss": 1.192,
      "step": 47588
    },
    {
      "epoch": 4.094381829131894,
      "grad_norm": 1.3203125,
      "learning_rate": 7.88503854285722e-05,
      "loss": 1.5787,
      "step": 47589
    },
    {
      "epoch": 4.094467865439215,
      "grad_norm": 1.546875,
      "learning_rate": 7.883581081188179e-05,
      "loss": 0.9017,
      "step": 47590
    },
    {
      "epoch": 4.094553901746537,
      "grad_norm": 1.28125,
      "learning_rate": 7.882123742701458e-05,
      "loss": 1.4418,
      "step": 47591
    },
    {
      "epoch": 4.0946399380538585,
      "grad_norm": 1.0078125,
      "learning_rate": 7.880666527401348e-05,
      "loss": 0.6319,
      "step": 47592
    },
    {
      "epoch": 4.094725974361181,
      "grad_norm": 1.2109375,
      "learning_rate": 7.879209435292084e-05,
      "loss": 1.4624,
      "step": 47593
    },
    {
      "epoch": 4.094812010668502,
      "grad_norm": 1.046875,
      "learning_rate": 7.87775246637794e-05,
      "loss": 0.9389,
      "step": 47594
    },
    {
      "epoch": 4.094898046975824,
      "grad_norm": 2992.0,
      "learning_rate": 7.876295620663188e-05,
      "loss": 1.366,
      "step": 47595
    },
    {
      "epoch": 4.094984083283146,
      "grad_norm": 1.15625,
      "learning_rate": 7.87483889815207e-05,
      "loss": 1.4157,
      "step": 47596
    },
    {
      "epoch": 4.095070119590467,
      "grad_norm": 0.90234375,
      "learning_rate": 7.873382298848852e-05,
      "loss": 0.6249,
      "step": 47597
    },
    {
      "epoch": 4.095156155897789,
      "grad_norm": 1.203125,
      "learning_rate": 7.871925822757814e-05,
      "loss": 1.2037,
      "step": 47598
    },
    {
      "epoch": 4.09524219220511,
      "grad_norm": 0.8671875,
      "learning_rate": 7.870469469883185e-05,
      "loss": 0.5656,
      "step": 47599
    },
    {
      "epoch": 4.095328228512432,
      "grad_norm": 396.0,
      "learning_rate": 7.86901324022924e-05,
      "loss": 1.3631,
      "step": 47600
    },
    {
      "epoch": 4.095414264819754,
      "grad_norm": 2.140625,
      "learning_rate": 7.867557133800246e-05,
      "loss": 1.1862,
      "step": 47601
    },
    {
      "epoch": 4.095500301127076,
      "grad_norm": 1.171875,
      "learning_rate": 7.86610115060043e-05,
      "loss": 1.4872,
      "step": 47602
    },
    {
      "epoch": 4.095586337434398,
      "grad_norm": 0.89453125,
      "learning_rate": 7.864645290634098e-05,
      "loss": 0.9331,
      "step": 47603
    },
    {
      "epoch": 4.095672373741719,
      "grad_norm": 1.1953125,
      "learning_rate": 7.863189553905475e-05,
      "loss": 1.3445,
      "step": 47604
    },
    {
      "epoch": 4.095758410049041,
      "grad_norm": 0.98828125,
      "learning_rate": 7.861733940418825e-05,
      "loss": 1.4527,
      "step": 47605
    },
    {
      "epoch": 4.095844446356362,
      "grad_norm": 1.140625,
      "learning_rate": 7.86027845017842e-05,
      "loss": 1.3176,
      "step": 47606
    },
    {
      "epoch": 4.095930482663684,
      "grad_norm": 1.046875,
      "learning_rate": 7.858823083188494e-05,
      "loss": 1.1273,
      "step": 47607
    },
    {
      "epoch": 4.096016518971005,
      "grad_norm": 1.0078125,
      "learning_rate": 7.857367839453322e-05,
      "loss": 0.6847,
      "step": 47608
    },
    {
      "epoch": 4.0961025552783275,
      "grad_norm": 1.1953125,
      "learning_rate": 7.855912718977143e-05,
      "loss": 1.3517,
      "step": 47609
    },
    {
      "epoch": 4.0961885915856495,
      "grad_norm": 1.09375,
      "learning_rate": 7.854457721764224e-05,
      "loss": 0.689,
      "step": 47610
    },
    {
      "epoch": 4.096274627892971,
      "grad_norm": 1.25,
      "learning_rate": 7.85300284781883e-05,
      "loss": 0.941,
      "step": 47611
    },
    {
      "epoch": 4.096360664200293,
      "grad_norm": 0.890625,
      "learning_rate": 7.851548097145195e-05,
      "loss": 0.6126,
      "step": 47612
    },
    {
      "epoch": 4.096446700507614,
      "grad_norm": 1.0390625,
      "learning_rate": 7.850093469747583e-05,
      "loss": 0.5645,
      "step": 47613
    },
    {
      "epoch": 4.096532736814936,
      "grad_norm": 1.234375,
      "learning_rate": 7.848638965630262e-05,
      "loss": 1.6119,
      "step": 47614
    },
    {
      "epoch": 4.096618773122257,
      "grad_norm": 0.9765625,
      "learning_rate": 7.847184584797462e-05,
      "loss": 1.1285,
      "step": 47615
    },
    {
      "epoch": 4.096704809429579,
      "grad_norm": 1.390625,
      "learning_rate": 7.845730327253452e-05,
      "loss": 1.5325,
      "step": 47616
    },
    {
      "epoch": 4.096790845736901,
      "grad_norm": 1.828125,
      "learning_rate": 7.844276193002492e-05,
      "loss": 1.3148,
      "step": 47617
    },
    {
      "epoch": 4.0968768820442225,
      "grad_norm": 5.96875,
      "learning_rate": 7.842822182048814e-05,
      "loss": 1.0849,
      "step": 47618
    },
    {
      "epoch": 4.096962918351545,
      "grad_norm": 0.99609375,
      "learning_rate": 7.841368294396678e-05,
      "loss": 0.6729,
      "step": 47619
    },
    {
      "epoch": 4.097048954658866,
      "grad_norm": 1.1015625,
      "learning_rate": 7.839914530050357e-05,
      "loss": 0.8504,
      "step": 47620
    },
    {
      "epoch": 4.097134990966188,
      "grad_norm": 9.6875,
      "learning_rate": 7.838460889014071e-05,
      "loss": 0.9799,
      "step": 47621
    },
    {
      "epoch": 4.097221027273509,
      "grad_norm": 1.2265625,
      "learning_rate": 7.837007371292104e-05,
      "loss": 1.1025,
      "step": 47622
    },
    {
      "epoch": 4.097307063580831,
      "grad_norm": 0.94140625,
      "learning_rate": 7.835553976888671e-05,
      "loss": 1.0223,
      "step": 47623
    },
    {
      "epoch": 4.097393099888153,
      "grad_norm": 1.0234375,
      "learning_rate": 7.834100705808045e-05,
      "loss": 0.6629,
      "step": 47624
    },
    {
      "epoch": 4.097479136195474,
      "grad_norm": 1.1484375,
      "learning_rate": 7.83264755805449e-05,
      "loss": 1.5201,
      "step": 47625
    },
    {
      "epoch": 4.097565172502796,
      "grad_norm": 1.0234375,
      "learning_rate": 7.831194533632225e-05,
      "loss": 0.9498,
      "step": 47626
    },
    {
      "epoch": 4.097651208810118,
      "grad_norm": 1.0078125,
      "learning_rate": 7.829741632545512e-05,
      "loss": 0.7064,
      "step": 47627
    },
    {
      "epoch": 4.09773724511744,
      "grad_norm": 1.2578125,
      "learning_rate": 7.828288854798616e-05,
      "loss": 1.2886,
      "step": 47628
    },
    {
      "epoch": 4.097823281424761,
      "grad_norm": 8.125,
      "learning_rate": 7.826836200395764e-05,
      "loss": 1.2794,
      "step": 47629
    },
    {
      "epoch": 4.097909317732083,
      "grad_norm": 1.1484375,
      "learning_rate": 7.825383669341213e-05,
      "loss": 1.4974,
      "step": 47630
    },
    {
      "epoch": 4.097995354039405,
      "grad_norm": 1.53125,
      "learning_rate": 7.823931261639223e-05,
      "loss": 1.0241,
      "step": 47631
    },
    {
      "epoch": 4.098081390346726,
      "grad_norm": 0.90625,
      "learning_rate": 7.822478977294018e-05,
      "loss": 0.729,
      "step": 47632
    },
    {
      "epoch": 4.098167426654048,
      "grad_norm": 1.03125,
      "learning_rate": 7.821026816309873e-05,
      "loss": 1.5712,
      "step": 47633
    },
    {
      "epoch": 4.098253462961369,
      "grad_norm": 0.83984375,
      "learning_rate": 7.819574778691008e-05,
      "loss": 1.1064,
      "step": 47634
    },
    {
      "epoch": 4.0983394992686915,
      "grad_norm": 1.046875,
      "learning_rate": 7.818122864441685e-05,
      "loss": 0.7179,
      "step": 47635
    },
    {
      "epoch": 4.098425535576013,
      "grad_norm": 7.5625,
      "learning_rate": 7.816671073566156e-05,
      "loss": 1.5441,
      "step": 47636
    },
    {
      "epoch": 4.098511571883335,
      "grad_norm": 13.625,
      "learning_rate": 7.815219406068653e-05,
      "loss": 0.6401,
      "step": 47637
    },
    {
      "epoch": 4.098597608190657,
      "grad_norm": 0.953125,
      "learning_rate": 7.81376786195343e-05,
      "loss": 1.5431,
      "step": 47638
    },
    {
      "epoch": 4.098683644497978,
      "grad_norm": 1.0,
      "learning_rate": 7.812316441224726e-05,
      "loss": 1.0831,
      "step": 47639
    },
    {
      "epoch": 4.0987696808053,
      "grad_norm": 14.625,
      "learning_rate": 7.810865143886791e-05,
      "loss": 1.0872,
      "step": 47640
    },
    {
      "epoch": 4.098855717112621,
      "grad_norm": 1.0625,
      "learning_rate": 7.809413969943891e-05,
      "loss": 1.1382,
      "step": 47641
    },
    {
      "epoch": 4.098941753419943,
      "grad_norm": 0.99609375,
      "learning_rate": 7.80796291940023e-05,
      "loss": 1.1356,
      "step": 47642
    },
    {
      "epoch": 4.099027789727265,
      "grad_norm": 1.28125,
      "learning_rate": 7.806511992260073e-05,
      "loss": 0.9112,
      "step": 47643
    },
    {
      "epoch": 4.0991138260345865,
      "grad_norm": 1.328125,
      "learning_rate": 7.805061188527674e-05,
      "loss": 1.3857,
      "step": 47644
    },
    {
      "epoch": 4.099199862341909,
      "grad_norm": 1.046875,
      "learning_rate": 7.803610508207249e-05,
      "loss": 0.8361,
      "step": 47645
    },
    {
      "epoch": 4.09928589864923,
      "grad_norm": 0.984375,
      "learning_rate": 7.802159951303061e-05,
      "loss": 1.7617,
      "step": 47646
    },
    {
      "epoch": 4.099371934956552,
      "grad_norm": 1.1328125,
      "learning_rate": 7.80070951781936e-05,
      "loss": 1.1291,
      "step": 47647
    },
    {
      "epoch": 4.099457971263873,
      "grad_norm": 0.796875,
      "learning_rate": 7.799259207760361e-05,
      "loss": 0.5491,
      "step": 47648
    },
    {
      "epoch": 4.099544007571195,
      "grad_norm": 0.921875,
      "learning_rate": 7.797809021130325e-05,
      "loss": 1.1959,
      "step": 47649
    },
    {
      "epoch": 4.099630043878517,
      "grad_norm": 1.015625,
      "learning_rate": 7.796358957933503e-05,
      "loss": 1.23,
      "step": 47650
    },
    {
      "epoch": 4.099716080185838,
      "grad_norm": 0.80859375,
      "learning_rate": 7.794909018174107e-05,
      "loss": 0.9467,
      "step": 47651
    },
    {
      "epoch": 4.09980211649316,
      "grad_norm": 1.171875,
      "learning_rate": 7.793459201856407e-05,
      "loss": 1.3757,
      "step": 47652
    },
    {
      "epoch": 4.099888152800482,
      "grad_norm": 1.078125,
      "learning_rate": 7.792009508984621e-05,
      "loss": 1.0911,
      "step": 47653
    },
    {
      "epoch": 4.099974189107804,
      "grad_norm": 0.8515625,
      "learning_rate": 7.790559939562997e-05,
      "loss": 0.7975,
      "step": 47654
    },
    {
      "epoch": 4.100060225415125,
      "grad_norm": 154.0,
      "learning_rate": 7.789110493595786e-05,
      "loss": 1.0035,
      "step": 47655
    },
    {
      "epoch": 4.100146261722447,
      "grad_norm": 1.0390625,
      "learning_rate": 7.787661171087202e-05,
      "loss": 1.0778,
      "step": 47656
    },
    {
      "epoch": 4.100232298029769,
      "grad_norm": 0.9375,
      "learning_rate": 7.786211972041507e-05,
      "loss": 1.1725,
      "step": 47657
    },
    {
      "epoch": 4.10031833433709,
      "grad_norm": 8.5625,
      "learning_rate": 7.784762896462938e-05,
      "loss": 0.8489,
      "step": 47658
    },
    {
      "epoch": 4.100404370644412,
      "grad_norm": 0.86328125,
      "learning_rate": 7.783313944355719e-05,
      "loss": 0.5911,
      "step": 47659
    },
    {
      "epoch": 4.100490406951733,
      "grad_norm": 0.9296875,
      "learning_rate": 7.78186511572409e-05,
      "loss": 1.1421,
      "step": 47660
    },
    {
      "epoch": 4.1005764432590555,
      "grad_norm": 0.8984375,
      "learning_rate": 7.780416410572311e-05,
      "loss": 0.6241,
      "step": 47661
    },
    {
      "epoch": 4.100662479566377,
      "grad_norm": 0.81640625,
      "learning_rate": 7.778967828904593e-05,
      "loss": 0.7186,
      "step": 47662
    },
    {
      "epoch": 4.100748515873699,
      "grad_norm": 0.8046875,
      "learning_rate": 7.777519370725189e-05,
      "loss": 1.1951,
      "step": 47663
    },
    {
      "epoch": 4.100834552181021,
      "grad_norm": 0.8671875,
      "learning_rate": 7.776071036038324e-05,
      "loss": 0.4861,
      "step": 47664
    },
    {
      "epoch": 4.100920588488342,
      "grad_norm": 1.109375,
      "learning_rate": 7.774622824848231e-05,
      "loss": 0.7442,
      "step": 47665
    },
    {
      "epoch": 4.101006624795664,
      "grad_norm": 1.109375,
      "learning_rate": 7.773174737159172e-05,
      "loss": 1.5842,
      "step": 47666
    },
    {
      "epoch": 4.101092661102985,
      "grad_norm": 1.6328125,
      "learning_rate": 7.771726772975352e-05,
      "loss": 0.7377,
      "step": 47667
    },
    {
      "epoch": 4.101178697410307,
      "grad_norm": 1.0859375,
      "learning_rate": 7.770278932301012e-05,
      "loss": 1.0024,
      "step": 47668
    },
    {
      "epoch": 4.1012647337176285,
      "grad_norm": 1.421875,
      "learning_rate": 7.76883121514041e-05,
      "loss": 1.3884,
      "step": 47669
    },
    {
      "epoch": 4.1013507700249505,
      "grad_norm": 1.2890625,
      "learning_rate": 7.767383621497754e-05,
      "loss": 1.3419,
      "step": 47670
    },
    {
      "epoch": 4.101436806332273,
      "grad_norm": 0.9921875,
      "learning_rate": 7.765936151377278e-05,
      "loss": 1.4785,
      "step": 47671
    },
    {
      "epoch": 4.101522842639594,
      "grad_norm": 0.86328125,
      "learning_rate": 7.764488804783243e-05,
      "loss": 1.33,
      "step": 47672
    },
    {
      "epoch": 4.101608878946916,
      "grad_norm": 1.4375,
      "learning_rate": 7.763041581719855e-05,
      "loss": 0.9657,
      "step": 47673
    },
    {
      "epoch": 4.101694915254237,
      "grad_norm": 0.90625,
      "learning_rate": 7.761594482191359e-05,
      "loss": 0.5919,
      "step": 47674
    },
    {
      "epoch": 4.101780951561559,
      "grad_norm": 0.9296875,
      "learning_rate": 7.760147506201981e-05,
      "loss": 1.154,
      "step": 47675
    },
    {
      "epoch": 4.10186698786888,
      "grad_norm": 1.1484375,
      "learning_rate": 7.758700653755941e-05,
      "loss": 1.2942,
      "step": 47676
    },
    {
      "epoch": 4.101953024176202,
      "grad_norm": 0.70703125,
      "learning_rate": 7.757253924857504e-05,
      "loss": 0.7087,
      "step": 47677
    },
    {
      "epoch": 4.102039060483524,
      "grad_norm": 1.140625,
      "learning_rate": 7.755807319510882e-05,
      "loss": 1.0319,
      "step": 47678
    },
    {
      "epoch": 4.102125096790846,
      "grad_norm": 1.203125,
      "learning_rate": 7.754360837720315e-05,
      "loss": 1.6574,
      "step": 47679
    },
    {
      "epoch": 4.102211133098168,
      "grad_norm": 57.75,
      "learning_rate": 7.752914479490015e-05,
      "loss": 1.4013,
      "step": 47680
    },
    {
      "epoch": 4.102297169405489,
      "grad_norm": 38.75,
      "learning_rate": 7.751468244824217e-05,
      "loss": 0.8348,
      "step": 47681
    },
    {
      "epoch": 4.102383205712811,
      "grad_norm": 0.6796875,
      "learning_rate": 7.750022133727175e-05,
      "loss": 0.9179,
      "step": 47682
    },
    {
      "epoch": 4.102469242020132,
      "grad_norm": 1.1171875,
      "learning_rate": 7.74857614620309e-05,
      "loss": 1.1554,
      "step": 47683
    },
    {
      "epoch": 4.102555278327454,
      "grad_norm": 1.2265625,
      "learning_rate": 7.747130282256203e-05,
      "loss": 1.2888,
      "step": 47684
    },
    {
      "epoch": 4.102641314634776,
      "grad_norm": 1.0625,
      "learning_rate": 7.745684541890751e-05,
      "loss": 1.2441,
      "step": 47685
    },
    {
      "epoch": 4.102727350942097,
      "grad_norm": 1.1015625,
      "learning_rate": 7.744238925110941e-05,
      "loss": 1.405,
      "step": 47686
    },
    {
      "epoch": 4.1028133872494195,
      "grad_norm": 1.1328125,
      "learning_rate": 7.742793431921019e-05,
      "loss": 1.2262,
      "step": 47687
    },
    {
      "epoch": 4.102899423556741,
      "grad_norm": 1.0546875,
      "learning_rate": 7.741348062325215e-05,
      "loss": 1.5004,
      "step": 47688
    },
    {
      "epoch": 4.102985459864063,
      "grad_norm": 1.2734375,
      "learning_rate": 7.739902816327737e-05,
      "loss": 1.7248,
      "step": 47689
    },
    {
      "epoch": 4.103071496171384,
      "grad_norm": 1.140625,
      "learning_rate": 7.738457693932826e-05,
      "loss": 1.632,
      "step": 47690
    },
    {
      "epoch": 4.103157532478706,
      "grad_norm": 1.0234375,
      "learning_rate": 7.737012695144724e-05,
      "loss": 0.9119,
      "step": 47691
    },
    {
      "epoch": 4.103243568786028,
      "grad_norm": 1.1484375,
      "learning_rate": 7.735567819967621e-05,
      "loss": 1.0321,
      "step": 47692
    },
    {
      "epoch": 4.103329605093349,
      "grad_norm": 0.9921875,
      "learning_rate": 7.734123068405774e-05,
      "loss": 0.8234,
      "step": 47693
    },
    {
      "epoch": 4.103415641400671,
      "grad_norm": 1.171875,
      "learning_rate": 7.732678440463387e-05,
      "loss": 1.1472,
      "step": 47694
    },
    {
      "epoch": 4.1035016777079925,
      "grad_norm": 3.671875,
      "learning_rate": 7.731233936144694e-05,
      "loss": 0.7967,
      "step": 47695
    },
    {
      "epoch": 4.1035877140153145,
      "grad_norm": 2144.0,
      "learning_rate": 7.729789555453937e-05,
      "loss": 1.0347,
      "step": 47696
    },
    {
      "epoch": 4.103673750322637,
      "grad_norm": 1.0390625,
      "learning_rate": 7.728345298395307e-05,
      "loss": 1.2076,
      "step": 47697
    },
    {
      "epoch": 4.103759786629958,
      "grad_norm": 1.125,
      "learning_rate": 7.726901164973049e-05,
      "loss": 0.8139,
      "step": 47698
    },
    {
      "epoch": 4.10384582293728,
      "grad_norm": 0.921875,
      "learning_rate": 7.725457155191395e-05,
      "loss": 0.9018,
      "step": 47699
    },
    {
      "epoch": 4.103931859244601,
      "grad_norm": 1.3984375,
      "learning_rate": 7.724013269054541e-05,
      "loss": 0.6437,
      "step": 47700
    },
    {
      "epoch": 4.104017895551923,
      "grad_norm": 1.078125,
      "learning_rate": 7.722569506566729e-05,
      "loss": 1.1837,
      "step": 47701
    },
    {
      "epoch": 4.104103931859244,
      "grad_norm": 1.1015625,
      "learning_rate": 7.721125867732193e-05,
      "loss": 1.4072,
      "step": 47702
    },
    {
      "epoch": 4.104189968166566,
      "grad_norm": 0.96484375,
      "learning_rate": 7.719682352555124e-05,
      "loss": 1.1865,
      "step": 47703
    },
    {
      "epoch": 4.104276004473888,
      "grad_norm": 1.0859375,
      "learning_rate": 7.718238961039781e-05,
      "loss": 1.8892,
      "step": 47704
    },
    {
      "epoch": 4.10436204078121,
      "grad_norm": 25.25,
      "learning_rate": 7.716795693190348e-05,
      "loss": 0.7119,
      "step": 47705
    },
    {
      "epoch": 4.104448077088532,
      "grad_norm": 1.1015625,
      "learning_rate": 7.715352549011062e-05,
      "loss": 0.992,
      "step": 47706
    },
    {
      "epoch": 4.104534113395853,
      "grad_norm": 1.15625,
      "learning_rate": 7.713909528506163e-05,
      "loss": 1.5162,
      "step": 47707
    },
    {
      "epoch": 4.104620149703175,
      "grad_norm": 0.859375,
      "learning_rate": 7.712466631679837e-05,
      "loss": 0.7325,
      "step": 47708
    },
    {
      "epoch": 4.104706186010496,
      "grad_norm": 1.2109375,
      "learning_rate": 7.711023858536326e-05,
      "loss": 1.4692,
      "step": 47709
    },
    {
      "epoch": 4.104792222317818,
      "grad_norm": 2.6875,
      "learning_rate": 7.709581209079852e-05,
      "loss": 1.9315,
      "step": 47710
    },
    {
      "epoch": 4.10487825862514,
      "grad_norm": 49.25,
      "learning_rate": 7.708138683314625e-05,
      "loss": 1.564,
      "step": 47711
    },
    {
      "epoch": 4.104964294932461,
      "grad_norm": 3.515625,
      "learning_rate": 7.706696281244857e-05,
      "loss": 1.6541,
      "step": 47712
    },
    {
      "epoch": 4.1050503312397835,
      "grad_norm": 1.015625,
      "learning_rate": 7.705254002874784e-05,
      "loss": 0.6311,
      "step": 47713
    },
    {
      "epoch": 4.105136367547105,
      "grad_norm": 1.140625,
      "learning_rate": 7.703811848208614e-05,
      "loss": 1.138,
      "step": 47714
    },
    {
      "epoch": 4.105222403854427,
      "grad_norm": 0.9921875,
      "learning_rate": 7.702369817250582e-05,
      "loss": 0.8118,
      "step": 47715
    },
    {
      "epoch": 4.105308440161748,
      "grad_norm": 1.0703125,
      "learning_rate": 7.700927910004878e-05,
      "loss": 1.161,
      "step": 47716
    },
    {
      "epoch": 4.10539447646907,
      "grad_norm": 2.328125,
      "learning_rate": 7.699486126475735e-05,
      "loss": 0.6675,
      "step": 47717
    },
    {
      "epoch": 4.105480512776392,
      "grad_norm": 0.86328125,
      "learning_rate": 7.698044466667381e-05,
      "loss": 0.5831,
      "step": 47718
    },
    {
      "epoch": 4.105566549083713,
      "grad_norm": 1.1484375,
      "learning_rate": 7.696602930584002e-05,
      "loss": 1.0798,
      "step": 47719
    },
    {
      "epoch": 4.105652585391035,
      "grad_norm": 1.15625,
      "learning_rate": 7.695161518229848e-05,
      "loss": 0.9135,
      "step": 47720
    },
    {
      "epoch": 4.1057386216983565,
      "grad_norm": 1.0234375,
      "learning_rate": 7.693720229609108e-05,
      "loss": 1.1264,
      "step": 47721
    },
    {
      "epoch": 4.1058246580056785,
      "grad_norm": 9.6875,
      "learning_rate": 7.692279064726004e-05,
      "loss": 0.4756,
      "step": 47722
    },
    {
      "epoch": 4.105910694313,
      "grad_norm": 0.89453125,
      "learning_rate": 7.690838023584773e-05,
      "loss": 0.9355,
      "step": 47723
    },
    {
      "epoch": 4.105996730620322,
      "grad_norm": 1.0390625,
      "learning_rate": 7.689397106189594e-05,
      "loss": 1.3409,
      "step": 47724
    },
    {
      "epoch": 4.106082766927644,
      "grad_norm": 1.1640625,
      "learning_rate": 7.687956312544702e-05,
      "loss": 0.8703,
      "step": 47725
    },
    {
      "epoch": 4.106168803234965,
      "grad_norm": 1.2578125,
      "learning_rate": 7.686515642654323e-05,
      "loss": 1.0,
      "step": 47726
    },
    {
      "epoch": 4.106254839542287,
      "grad_norm": 0.94921875,
      "learning_rate": 7.685075096522643e-05,
      "loss": 0.7725,
      "step": 47727
    },
    {
      "epoch": 4.106340875849608,
      "grad_norm": 0.859375,
      "learning_rate": 7.683634674153889e-05,
      "loss": 0.7948,
      "step": 47728
    },
    {
      "epoch": 4.10642691215693,
      "grad_norm": 2.875,
      "learning_rate": 7.682194375552282e-05,
      "loss": 1.2596,
      "step": 47729
    },
    {
      "epoch": 4.1065129484642515,
      "grad_norm": 1.0859375,
      "learning_rate": 7.680754200722018e-05,
      "loss": 1.7734,
      "step": 47730
    },
    {
      "epoch": 4.106598984771574,
      "grad_norm": 1.1640625,
      "learning_rate": 7.679314149667315e-05,
      "loss": 1.5336,
      "step": 47731
    },
    {
      "epoch": 4.106685021078896,
      "grad_norm": 1.3125,
      "learning_rate": 7.677874222392401e-05,
      "loss": 0.6543,
      "step": 47732
    },
    {
      "epoch": 4.106771057386217,
      "grad_norm": 1.3515625,
      "learning_rate": 7.676434418901462e-05,
      "loss": 1.8274,
      "step": 47733
    },
    {
      "epoch": 4.106857093693539,
      "grad_norm": 1.234375,
      "learning_rate": 7.674994739198732e-05,
      "loss": 0.7117,
      "step": 47734
    },
    {
      "epoch": 4.10694313000086,
      "grad_norm": 1.0234375,
      "learning_rate": 7.673555183288394e-05,
      "loss": 0.4553,
      "step": 47735
    },
    {
      "epoch": 4.107029166308182,
      "grad_norm": 1.3671875,
      "learning_rate": 7.67211575117468e-05,
      "loss": 1.255,
      "step": 47736
    },
    {
      "epoch": 4.107115202615503,
      "grad_norm": 37.0,
      "learning_rate": 7.670676442861807e-05,
      "loss": 1.4102,
      "step": 47737
    },
    {
      "epoch": 4.107201238922825,
      "grad_norm": 0.94140625,
      "learning_rate": 7.669237258353961e-05,
      "loss": 0.6918,
      "step": 47738
    },
    {
      "epoch": 4.1072872752301475,
      "grad_norm": 1.015625,
      "learning_rate": 7.667798197655362e-05,
      "loss": 0.7551,
      "step": 47739
    },
    {
      "epoch": 4.107373311537469,
      "grad_norm": 1.21875,
      "learning_rate": 7.666359260770234e-05,
      "loss": 1.1553,
      "step": 47740
    },
    {
      "epoch": 4.107459347844791,
      "grad_norm": 25.25,
      "learning_rate": 7.664920447702756e-05,
      "loss": 0.7114,
      "step": 47741
    },
    {
      "epoch": 4.107545384152112,
      "grad_norm": 0.796875,
      "learning_rate": 7.663481758457152e-05,
      "loss": 0.8167,
      "step": 47742
    },
    {
      "epoch": 4.107631420459434,
      "grad_norm": 0.890625,
      "learning_rate": 7.662043193037648e-05,
      "loss": 1.0498,
      "step": 47743
    },
    {
      "epoch": 4.107717456766755,
      "grad_norm": 1.125,
      "learning_rate": 7.660604751448413e-05,
      "loss": 1.7303,
      "step": 47744
    },
    {
      "epoch": 4.107803493074077,
      "grad_norm": 36.0,
      "learning_rate": 7.659166433693693e-05,
      "loss": 0.9214,
      "step": 47745
    },
    {
      "epoch": 4.107889529381399,
      "grad_norm": 1.1953125,
      "learning_rate": 7.657728239777661e-05,
      "loss": 1.1977,
      "step": 47746
    },
    {
      "epoch": 4.1079755656887205,
      "grad_norm": 1.109375,
      "learning_rate": 7.656290169704544e-05,
      "loss": 1.7011,
      "step": 47747
    },
    {
      "epoch": 4.1080616019960425,
      "grad_norm": 1.421875,
      "learning_rate": 7.654852223478547e-05,
      "loss": 1.2745,
      "step": 47748
    },
    {
      "epoch": 4.108147638303364,
      "grad_norm": 1.6171875,
      "learning_rate": 7.653414401103865e-05,
      "loss": 1.135,
      "step": 47749
    },
    {
      "epoch": 4.108233674610686,
      "grad_norm": 12.5625,
      "learning_rate": 7.651976702584707e-05,
      "loss": 1.0473,
      "step": 47750
    },
    {
      "epoch": 4.108319710918007,
      "grad_norm": 1.59375,
      "learning_rate": 7.650539127925283e-05,
      "loss": 0.9767,
      "step": 47751
    },
    {
      "epoch": 4.108405747225329,
      "grad_norm": 1.2890625,
      "learning_rate": 7.649101677129794e-05,
      "loss": 1.184,
      "step": 47752
    },
    {
      "epoch": 4.108491783532651,
      "grad_norm": 1.390625,
      "learning_rate": 7.64766435020246e-05,
      "loss": 1.3515,
      "step": 47753
    },
    {
      "epoch": 4.108577819839972,
      "grad_norm": 0.75390625,
      "learning_rate": 7.646227147147456e-05,
      "loss": 0.5126,
      "step": 47754
    },
    {
      "epoch": 4.108663856147294,
      "grad_norm": 0.98828125,
      "learning_rate": 7.644790067969005e-05,
      "loss": 1.3564,
      "step": 47755
    },
    {
      "epoch": 4.1087498924546155,
      "grad_norm": 12.5,
      "learning_rate": 7.643353112671314e-05,
      "loss": 0.8621,
      "step": 47756
    },
    {
      "epoch": 4.108835928761938,
      "grad_norm": 1.09375,
      "learning_rate": 7.641916281258565e-05,
      "loss": 1.3938,
      "step": 47757
    },
    {
      "epoch": 4.10892196506926,
      "grad_norm": 0.890625,
      "learning_rate": 7.64047957373497e-05,
      "loss": 0.6121,
      "step": 47758
    },
    {
      "epoch": 4.109008001376581,
      "grad_norm": 0.97265625,
      "learning_rate": 7.639042990104749e-05,
      "loss": 0.7847,
      "step": 47759
    },
    {
      "epoch": 4.109094037683903,
      "grad_norm": 1.1171875,
      "learning_rate": 7.637606530372077e-05,
      "loss": 1.2494,
      "step": 47760
    },
    {
      "epoch": 4.109180073991224,
      "grad_norm": 1.1875,
      "learning_rate": 7.636170194541176e-05,
      "loss": 1.4424,
      "step": 47761
    },
    {
      "epoch": 4.109266110298546,
      "grad_norm": 1.1328125,
      "learning_rate": 7.634733982616232e-05,
      "loss": 1.5868,
      "step": 47762
    },
    {
      "epoch": 4.109352146605867,
      "grad_norm": 1.21875,
      "learning_rate": 7.633297894601443e-05,
      "loss": 1.4773,
      "step": 47763
    },
    {
      "epoch": 4.109438182913189,
      "grad_norm": 1.125,
      "learning_rate": 7.631861930501039e-05,
      "loss": 1.3233,
      "step": 47764
    },
    {
      "epoch": 4.1095242192205115,
      "grad_norm": 0.95703125,
      "learning_rate": 7.630426090319181e-05,
      "loss": 1.0057,
      "step": 47765
    },
    {
      "epoch": 4.109610255527833,
      "grad_norm": 1.140625,
      "learning_rate": 7.628990374060086e-05,
      "loss": 0.9883,
      "step": 47766
    },
    {
      "epoch": 4.109696291835155,
      "grad_norm": 1.0234375,
      "learning_rate": 7.62755478172797e-05,
      "loss": 1.153,
      "step": 47767
    },
    {
      "epoch": 4.109782328142476,
      "grad_norm": 1.21875,
      "learning_rate": 7.626119313326996e-05,
      "loss": 1.3688,
      "step": 47768
    },
    {
      "epoch": 4.109868364449798,
      "grad_norm": 0.98046875,
      "learning_rate": 7.624683968861385e-05,
      "loss": 1.26,
      "step": 47769
    },
    {
      "epoch": 4.109954400757119,
      "grad_norm": 4.59375,
      "learning_rate": 7.623248748335343e-05,
      "loss": 1.173,
      "step": 47770
    },
    {
      "epoch": 4.110040437064441,
      "grad_norm": 0.76953125,
      "learning_rate": 7.621813651753051e-05,
      "loss": 0.3999,
      "step": 47771
    },
    {
      "epoch": 4.110126473371763,
      "grad_norm": 1.359375,
      "learning_rate": 7.620378679118705e-05,
      "loss": 1.7118,
      "step": 47772
    },
    {
      "epoch": 4.1102125096790845,
      "grad_norm": 3.328125,
      "learning_rate": 7.61894383043652e-05,
      "loss": 1.3206,
      "step": 47773
    },
    {
      "epoch": 4.1102985459864065,
      "grad_norm": 63.75,
      "learning_rate": 7.617509105710674e-05,
      "loss": 0.8628,
      "step": 47774
    },
    {
      "epoch": 4.110384582293728,
      "grad_norm": 1.015625,
      "learning_rate": 7.61607450494538e-05,
      "loss": 1.3201,
      "step": 47775
    },
    {
      "epoch": 4.11047061860105,
      "grad_norm": 19.375,
      "learning_rate": 7.614640028144815e-05,
      "loss": 0.9033,
      "step": 47776
    },
    {
      "epoch": 4.110556654908371,
      "grad_norm": 0.85546875,
      "learning_rate": 7.613205675313184e-05,
      "loss": 0.6419,
      "step": 47777
    },
    {
      "epoch": 4.110642691215693,
      "grad_norm": 1.1875,
      "learning_rate": 7.611771446454696e-05,
      "loss": 1.3483,
      "step": 47778
    },
    {
      "epoch": 4.110728727523015,
      "grad_norm": 0.9453125,
      "learning_rate": 7.61033734157352e-05,
      "loss": 1.1137,
      "step": 47779
    },
    {
      "epoch": 4.110814763830336,
      "grad_norm": 0.90625,
      "learning_rate": 7.608903360673863e-05,
      "loss": 0.9608,
      "step": 47780
    },
    {
      "epoch": 4.110900800137658,
      "grad_norm": 1.3515625,
      "learning_rate": 7.607469503759928e-05,
      "loss": 1.6905,
      "step": 47781
    },
    {
      "epoch": 4.1109868364449795,
      "grad_norm": 0.609375,
      "learning_rate": 7.606035770835889e-05,
      "loss": 0.322,
      "step": 47782
    },
    {
      "epoch": 4.111072872752302,
      "grad_norm": 1.46875,
      "learning_rate": 7.604602161905955e-05,
      "loss": 1.8385,
      "step": 47783
    },
    {
      "epoch": 4.111158909059623,
      "grad_norm": 0.87890625,
      "learning_rate": 7.60316867697432e-05,
      "loss": 0.6992,
      "step": 47784
    },
    {
      "epoch": 4.111244945366945,
      "grad_norm": 1.3515625,
      "learning_rate": 7.601735316045161e-05,
      "loss": 2.0507,
      "step": 47785
    },
    {
      "epoch": 4.111330981674267,
      "grad_norm": 4.0625,
      "learning_rate": 7.600302079122689e-05,
      "loss": 0.5433,
      "step": 47786
    },
    {
      "epoch": 4.111417017981588,
      "grad_norm": 1.109375,
      "learning_rate": 7.59886896621107e-05,
      "loss": 0.8269,
      "step": 47787
    },
    {
      "epoch": 4.11150305428891,
      "grad_norm": 1.0234375,
      "learning_rate": 7.597435977314521e-05,
      "loss": 1.0563,
      "step": 47788
    },
    {
      "epoch": 4.111589090596231,
      "grad_norm": 0.95703125,
      "learning_rate": 7.596003112437239e-05,
      "loss": 0.5647,
      "step": 47789
    },
    {
      "epoch": 4.111675126903553,
      "grad_norm": 1.1328125,
      "learning_rate": 7.594570371583387e-05,
      "loss": 0.5849,
      "step": 47790
    },
    {
      "epoch": 4.111761163210875,
      "grad_norm": 1.09375,
      "learning_rate": 7.59313775475718e-05,
      "loss": 0.7434,
      "step": 47791
    },
    {
      "epoch": 4.111847199518197,
      "grad_norm": 0.9765625,
      "learning_rate": 7.591705261962784e-05,
      "loss": 0.8584,
      "step": 47792
    },
    {
      "epoch": 4.111933235825519,
      "grad_norm": 0.86328125,
      "learning_rate": 7.590272893204403e-05,
      "loss": 1.343,
      "step": 47793
    },
    {
      "epoch": 4.11201927213284,
      "grad_norm": 0.890625,
      "learning_rate": 7.588840648486239e-05,
      "loss": 0.5343,
      "step": 47794
    },
    {
      "epoch": 4.112105308440162,
      "grad_norm": 1.203125,
      "learning_rate": 7.587408527812455e-05,
      "loss": 0.9324,
      "step": 47795
    },
    {
      "epoch": 4.112191344747483,
      "grad_norm": 0.875,
      "learning_rate": 7.585976531187256e-05,
      "loss": 1.0552,
      "step": 47796
    },
    {
      "epoch": 4.112277381054805,
      "grad_norm": 8.875,
      "learning_rate": 7.584544658614834e-05,
      "loss": 1.0405,
      "step": 47797
    },
    {
      "epoch": 4.112363417362126,
      "grad_norm": 1.359375,
      "learning_rate": 7.583112910099361e-05,
      "loss": 1.325,
      "step": 47798
    },
    {
      "epoch": 4.1124494536694485,
      "grad_norm": 3.15625,
      "learning_rate": 7.581681285645032e-05,
      "loss": 1.0739,
      "step": 47799
    },
    {
      "epoch": 4.1125354899767705,
      "grad_norm": 0.81640625,
      "learning_rate": 7.580249785256044e-05,
      "loss": 0.9147,
      "step": 47800
    },
    {
      "epoch": 4.112621526284092,
      "grad_norm": 3.9375,
      "learning_rate": 7.578818408936566e-05,
      "loss": 0.9231,
      "step": 47801
    },
    {
      "epoch": 4.112707562591414,
      "grad_norm": 1.3203125,
      "learning_rate": 7.577387156690795e-05,
      "loss": 0.8523,
      "step": 47802
    },
    {
      "epoch": 4.112793598898735,
      "grad_norm": 3.234375,
      "learning_rate": 7.57595602852293e-05,
      "loss": 0.4981,
      "step": 47803
    },
    {
      "epoch": 4.112879635206057,
      "grad_norm": 2.859375,
      "learning_rate": 7.574525024437123e-05,
      "loss": 0.8532,
      "step": 47804
    },
    {
      "epoch": 4.112965671513378,
      "grad_norm": 0.84765625,
      "learning_rate": 7.573094144437598e-05,
      "loss": 0.6092,
      "step": 47805
    },
    {
      "epoch": 4.1130517078207,
      "grad_norm": 1.0859375,
      "learning_rate": 7.571663388528505e-05,
      "loss": 1.0662,
      "step": 47806
    },
    {
      "epoch": 4.113137744128022,
      "grad_norm": 1.0625,
      "learning_rate": 7.570232756714046e-05,
      "loss": 1.3287,
      "step": 47807
    },
    {
      "epoch": 4.1132237804353435,
      "grad_norm": 5.75,
      "learning_rate": 7.568802248998413e-05,
      "loss": 1.2788,
      "step": 47808
    },
    {
      "epoch": 4.113309816742666,
      "grad_norm": 0.83203125,
      "learning_rate": 7.567371865385774e-05,
      "loss": 1.0042,
      "step": 47809
    },
    {
      "epoch": 4.113395853049987,
      "grad_norm": 1.4140625,
      "learning_rate": 7.565941605880316e-05,
      "loss": 1.8694,
      "step": 47810
    },
    {
      "epoch": 4.113481889357309,
      "grad_norm": 1.0625,
      "learning_rate": 7.56451147048624e-05,
      "loss": 1.3562,
      "step": 47811
    },
    {
      "epoch": 4.113567925664631,
      "grad_norm": 2.46875,
      "learning_rate": 7.563081459207698e-05,
      "loss": 1.2531,
      "step": 47812
    },
    {
      "epoch": 4.113653961971952,
      "grad_norm": 1.359375,
      "learning_rate": 7.561651572048894e-05,
      "loss": 0.7352,
      "step": 47813
    },
    {
      "epoch": 4.113739998279274,
      "grad_norm": 1.125,
      "learning_rate": 7.560221809014017e-05,
      "loss": 0.9264,
      "step": 47814
    },
    {
      "epoch": 4.113826034586595,
      "grad_norm": 1.0859375,
      "learning_rate": 7.558792170107226e-05,
      "loss": 1.0771,
      "step": 47815
    },
    {
      "epoch": 4.113912070893917,
      "grad_norm": 1.5078125,
      "learning_rate": 7.557362655332722e-05,
      "loss": 1.2959,
      "step": 47816
    },
    {
      "epoch": 4.113998107201239,
      "grad_norm": 44544.0,
      "learning_rate": 7.55593326469467e-05,
      "loss": 1.2918,
      "step": 47817
    },
    {
      "epoch": 4.114084143508561,
      "grad_norm": 1.21875,
      "learning_rate": 7.554503998197254e-05,
      "loss": 0.8298,
      "step": 47818
    },
    {
      "epoch": 4.114170179815883,
      "grad_norm": 6.34375,
      "learning_rate": 7.553074855844677e-05,
      "loss": 1.3339,
      "step": 47819
    },
    {
      "epoch": 4.114256216123204,
      "grad_norm": 1.0078125,
      "learning_rate": 7.55164583764108e-05,
      "loss": 1.3328,
      "step": 47820
    },
    {
      "epoch": 4.114342252430526,
      "grad_norm": 1.671875,
      "learning_rate": 7.550216943590671e-05,
      "loss": 1.4887,
      "step": 47821
    },
    {
      "epoch": 4.114428288737847,
      "grad_norm": 0.91796875,
      "learning_rate": 7.54878817369763e-05,
      "loss": 0.8698,
      "step": 47822
    },
    {
      "epoch": 4.114514325045169,
      "grad_norm": 4.84375,
      "learning_rate": 7.547359527966114e-05,
      "loss": 0.6932,
      "step": 47823
    },
    {
      "epoch": 4.11460036135249,
      "grad_norm": 0.9609375,
      "learning_rate": 7.545931006400314e-05,
      "loss": 0.8334,
      "step": 47824
    },
    {
      "epoch": 4.1146863976598125,
      "grad_norm": 25.125,
      "learning_rate": 7.544502609004411e-05,
      "loss": 0.9168,
      "step": 47825
    },
    {
      "epoch": 4.1147724339671345,
      "grad_norm": 0.921875,
      "learning_rate": 7.543074335782579e-05,
      "loss": 0.7189,
      "step": 47826
    },
    {
      "epoch": 4.114858470274456,
      "grad_norm": 1.2421875,
      "learning_rate": 7.541646186739016e-05,
      "loss": 0.5237,
      "step": 47827
    },
    {
      "epoch": 4.114944506581778,
      "grad_norm": 1.4375,
      "learning_rate": 7.54021816187786e-05,
      "loss": 0.8408,
      "step": 47828
    },
    {
      "epoch": 4.115030542889099,
      "grad_norm": 1.15625,
      "learning_rate": 7.53879026120331e-05,
      "loss": 0.7765,
      "step": 47829
    },
    {
      "epoch": 4.115116579196421,
      "grad_norm": 1.3359375,
      "learning_rate": 7.537362484719556e-05,
      "loss": 1.3853,
      "step": 47830
    },
    {
      "epoch": 4.115202615503742,
      "grad_norm": 1.1171875,
      "learning_rate": 7.535934832430741e-05,
      "loss": 1.3895,
      "step": 47831
    },
    {
      "epoch": 4.115288651811064,
      "grad_norm": 728.0,
      "learning_rate": 7.534507304341071e-05,
      "loss": 0.6554,
      "step": 47832
    },
    {
      "epoch": 4.115374688118386,
      "grad_norm": 2.015625,
      "learning_rate": 7.533079900454699e-05,
      "loss": 1.1272,
      "step": 47833
    },
    {
      "epoch": 4.1154607244257075,
      "grad_norm": 1.03125,
      "learning_rate": 7.531652620775808e-05,
      "loss": 0.6132,
      "step": 47834
    },
    {
      "epoch": 4.11554676073303,
      "grad_norm": 2.40625,
      "learning_rate": 7.530225465308583e-05,
      "loss": 1.3679,
      "step": 47835
    },
    {
      "epoch": 4.115632797040351,
      "grad_norm": 102.0,
      "learning_rate": 7.528798434057177e-05,
      "loss": 0.7398,
      "step": 47836
    },
    {
      "epoch": 4.115718833347673,
      "grad_norm": 0.875,
      "learning_rate": 7.527371527025779e-05,
      "loss": 0.848,
      "step": 47837
    },
    {
      "epoch": 4.115804869654994,
      "grad_norm": 1.265625,
      "learning_rate": 7.525944744218566e-05,
      "loss": 1.1836,
      "step": 47838
    },
    {
      "epoch": 4.115890905962316,
      "grad_norm": 1.25,
      "learning_rate": 7.524518085639697e-05,
      "loss": 0.7383,
      "step": 47839
    },
    {
      "epoch": 4.115976942269638,
      "grad_norm": 1.046875,
      "learning_rate": 7.523091551293349e-05,
      "loss": 0.9772,
      "step": 47840
    },
    {
      "epoch": 4.116062978576959,
      "grad_norm": 4.5625,
      "learning_rate": 7.521665141183709e-05,
      "loss": 1.1724,
      "step": 47841
    },
    {
      "epoch": 4.116149014884281,
      "grad_norm": 352.0,
      "learning_rate": 7.52023885531492e-05,
      "loss": 0.879,
      "step": 47842
    },
    {
      "epoch": 4.116235051191603,
      "grad_norm": 1.234375,
      "learning_rate": 7.51881269369118e-05,
      "loss": 1.2226,
      "step": 47843
    },
    {
      "epoch": 4.116321087498925,
      "grad_norm": 2.265625,
      "learning_rate": 7.517386656316655e-05,
      "loss": 0.7471,
      "step": 47844
    },
    {
      "epoch": 4.116407123806246,
      "grad_norm": 0.84375,
      "learning_rate": 7.515960743195504e-05,
      "loss": 0.7097,
      "step": 47845
    },
    {
      "epoch": 4.116493160113568,
      "grad_norm": 1.1640625,
      "learning_rate": 7.514534954331915e-05,
      "loss": 1.3588,
      "step": 47846
    },
    {
      "epoch": 4.11657919642089,
      "grad_norm": 1.140625,
      "learning_rate": 7.513109289730036e-05,
      "loss": 0.8098,
      "step": 47847
    },
    {
      "epoch": 4.116665232728211,
      "grad_norm": 11.875,
      "learning_rate": 7.51168374939405e-05,
      "loss": 0.9108,
      "step": 47848
    },
    {
      "epoch": 4.116751269035533,
      "grad_norm": 152.0,
      "learning_rate": 7.510258333328135e-05,
      "loss": 0.6512,
      "step": 47849
    },
    {
      "epoch": 4.116837305342854,
      "grad_norm": 0.98046875,
      "learning_rate": 7.508833041536444e-05,
      "loss": 0.7004,
      "step": 47850
    },
    {
      "epoch": 4.1169233416501765,
      "grad_norm": 0.83984375,
      "learning_rate": 7.507407874023148e-05,
      "loss": 0.8374,
      "step": 47851
    },
    {
      "epoch": 4.117009377957498,
      "grad_norm": 1.0625,
      "learning_rate": 7.505982830792435e-05,
      "loss": 1.4538,
      "step": 47852
    },
    {
      "epoch": 4.11709541426482,
      "grad_norm": 0.984375,
      "learning_rate": 7.504557911848442e-05,
      "loss": 0.9706,
      "step": 47853
    },
    {
      "epoch": 4.117181450572142,
      "grad_norm": 1.0234375,
      "learning_rate": 7.503133117195354e-05,
      "loss": 0.7445,
      "step": 47854
    },
    {
      "epoch": 4.117267486879463,
      "grad_norm": 0.921875,
      "learning_rate": 7.501708446837346e-05,
      "loss": 0.7219,
      "step": 47855
    },
    {
      "epoch": 4.117353523186785,
      "grad_norm": 1.296875,
      "learning_rate": 7.500283900778565e-05,
      "loss": 1.4246,
      "step": 47856
    },
    {
      "epoch": 4.117439559494106,
      "grad_norm": 1.40625,
      "learning_rate": 7.498859479023201e-05,
      "loss": 0.9257,
      "step": 47857
    },
    {
      "epoch": 4.117525595801428,
      "grad_norm": 59.25,
      "learning_rate": 7.497435181575396e-05,
      "loss": 0.8355,
      "step": 47858
    },
    {
      "epoch": 4.1176116321087495,
      "grad_norm": 8.5,
      "learning_rate": 7.496011008439323e-05,
      "loss": 1.0677,
      "step": 47859
    },
    {
      "epoch": 4.1176976684160715,
      "grad_norm": 18.625,
      "learning_rate": 7.494586959619165e-05,
      "loss": 1.128,
      "step": 47860
    },
    {
      "epoch": 4.117783704723394,
      "grad_norm": 1.2109375,
      "learning_rate": 7.493163035119055e-05,
      "loss": 0.9294,
      "step": 47861
    },
    {
      "epoch": 4.117869741030715,
      "grad_norm": 0.984375,
      "learning_rate": 7.49173923494319e-05,
      "loss": 1.3221,
      "step": 47862
    },
    {
      "epoch": 4.117955777338037,
      "grad_norm": 0.83203125,
      "learning_rate": 7.490315559095718e-05,
      "loss": 0.9021,
      "step": 47863
    },
    {
      "epoch": 4.118041813645358,
      "grad_norm": 0.90625,
      "learning_rate": 7.4888920075808e-05,
      "loss": 0.6941,
      "step": 47864
    },
    {
      "epoch": 4.11812784995268,
      "grad_norm": 1.8125,
      "learning_rate": 7.487468580402617e-05,
      "loss": 0.8977,
      "step": 47865
    },
    {
      "epoch": 4.118213886260001,
      "grad_norm": 0.8515625,
      "learning_rate": 7.486045277565307e-05,
      "loss": 0.8854,
      "step": 47866
    },
    {
      "epoch": 4.118299922567323,
      "grad_norm": 7.3125,
      "learning_rate": 7.484622099073046e-05,
      "loss": 0.8048,
      "step": 47867
    },
    {
      "epoch": 4.118385958874645,
      "grad_norm": 3.078125,
      "learning_rate": 7.483199044930012e-05,
      "loss": 1.6197,
      "step": 47868
    },
    {
      "epoch": 4.118471995181967,
      "grad_norm": 1.6015625,
      "learning_rate": 7.481776115140338e-05,
      "loss": 1.0316,
      "step": 47869
    },
    {
      "epoch": 4.118558031489289,
      "grad_norm": 0.9453125,
      "learning_rate": 7.480353309708199e-05,
      "loss": 1.4551,
      "step": 47870
    },
    {
      "epoch": 4.11864406779661,
      "grad_norm": 0.90625,
      "learning_rate": 7.47893062863777e-05,
      "loss": 0.5092,
      "step": 47871
    },
    {
      "epoch": 4.118730104103932,
      "grad_norm": 1.9296875,
      "learning_rate": 7.477508071933187e-05,
      "loss": 1.2192,
      "step": 47872
    },
    {
      "epoch": 4.118816140411253,
      "grad_norm": 1.3125,
      "learning_rate": 7.476085639598635e-05,
      "loss": 1.7298,
      "step": 47873
    },
    {
      "epoch": 4.118902176718575,
      "grad_norm": 1.1484375,
      "learning_rate": 7.474663331638248e-05,
      "loss": 1.5627,
      "step": 47874
    },
    {
      "epoch": 4.118988213025897,
      "grad_norm": 36.75,
      "learning_rate": 7.473241148056197e-05,
      "loss": 0.8548,
      "step": 47875
    },
    {
      "epoch": 4.119074249333218,
      "grad_norm": 0.92578125,
      "learning_rate": 7.471819088856663e-05,
      "loss": 1.0453,
      "step": 47876
    },
    {
      "epoch": 4.1191602856405405,
      "grad_norm": 1.140625,
      "learning_rate": 7.470397154043773e-05,
      "loss": 1.4251,
      "step": 47877
    },
    {
      "epoch": 4.119246321947862,
      "grad_norm": 1.0703125,
      "learning_rate": 7.468975343621698e-05,
      "loss": 0.8141,
      "step": 47878
    },
    {
      "epoch": 4.119332358255184,
      "grad_norm": 0.92578125,
      "learning_rate": 7.467553657594611e-05,
      "loss": 1.4465,
      "step": 47879
    },
    {
      "epoch": 4.119418394562506,
      "grad_norm": 4.59375,
      "learning_rate": 7.466132095966649e-05,
      "loss": 1.8065,
      "step": 47880
    },
    {
      "epoch": 4.119504430869827,
      "grad_norm": 1.8984375,
      "learning_rate": 7.464710658741975e-05,
      "loss": 1.1621,
      "step": 47881
    },
    {
      "epoch": 4.119590467177149,
      "grad_norm": 1.703125,
      "learning_rate": 7.463289345924763e-05,
      "loss": 1.2223,
      "step": 47882
    },
    {
      "epoch": 4.11967650348447,
      "grad_norm": 1.0546875,
      "learning_rate": 7.461868157519141e-05,
      "loss": 0.6546,
      "step": 47883
    },
    {
      "epoch": 4.119762539791792,
      "grad_norm": 13.375,
      "learning_rate": 7.460447093529288e-05,
      "loss": 1.4407,
      "step": 47884
    },
    {
      "epoch": 4.1198485760991135,
      "grad_norm": 1.1328125,
      "learning_rate": 7.459026153959359e-05,
      "loss": 1.6931,
      "step": 47885
    },
    {
      "epoch": 4.1199346124064355,
      "grad_norm": 1.4609375,
      "learning_rate": 7.457605338813495e-05,
      "loss": 1.4121,
      "step": 47886
    },
    {
      "epoch": 4.120020648713758,
      "grad_norm": 1.2578125,
      "learning_rate": 7.456184648095877e-05,
      "loss": 1.9644,
      "step": 47887
    },
    {
      "epoch": 4.120106685021079,
      "grad_norm": 0.7265625,
      "learning_rate": 7.454764081810628e-05,
      "loss": 0.3729,
      "step": 47888
    },
    {
      "epoch": 4.120192721328401,
      "grad_norm": 1.1484375,
      "learning_rate": 7.453343639961924e-05,
      "loss": 0.8223,
      "step": 47889
    },
    {
      "epoch": 4.120278757635722,
      "grad_norm": 1.4453125,
      "learning_rate": 7.451923322553922e-05,
      "loss": 1.329,
      "step": 47890
    },
    {
      "epoch": 4.120364793943044,
      "grad_norm": 1.046875,
      "learning_rate": 7.450503129590758e-05,
      "loss": 0.4329,
      "step": 47891
    },
    {
      "epoch": 4.120450830250365,
      "grad_norm": 3.171875,
      "learning_rate": 7.4490830610766e-05,
      "loss": 1.2652,
      "step": 47892
    },
    {
      "epoch": 4.120536866557687,
      "grad_norm": 0.921875,
      "learning_rate": 7.447663117015607e-05,
      "loss": 0.93,
      "step": 47893
    },
    {
      "epoch": 4.120622902865009,
      "grad_norm": 0.90234375,
      "learning_rate": 7.446243297411914e-05,
      "loss": 0.9747,
      "step": 47894
    },
    {
      "epoch": 4.120708939172331,
      "grad_norm": 1.203125,
      "learning_rate": 7.444823602269679e-05,
      "loss": 0.7985,
      "step": 47895
    },
    {
      "epoch": 4.120794975479653,
      "grad_norm": 1.5859375,
      "learning_rate": 7.443404031593077e-05,
      "loss": 1.4532,
      "step": 47896
    },
    {
      "epoch": 4.120881011786974,
      "grad_norm": 1.5546875,
      "learning_rate": 7.441984585386225e-05,
      "loss": 1.1055,
      "step": 47897
    },
    {
      "epoch": 4.120967048094296,
      "grad_norm": 1.3828125,
      "learning_rate": 7.440565263653293e-05,
      "loss": 0.9769,
      "step": 47898
    },
    {
      "epoch": 4.121053084401617,
      "grad_norm": 1.3671875,
      "learning_rate": 7.439146066398428e-05,
      "loss": 0.817,
      "step": 47899
    },
    {
      "epoch": 4.121139120708939,
      "grad_norm": 2.203125,
      "learning_rate": 7.437726993625787e-05,
      "loss": 1.493,
      "step": 47900
    },
    {
      "epoch": 4.121225157016261,
      "grad_norm": 2.296875,
      "learning_rate": 7.436308045339524e-05,
      "loss": 1.6392,
      "step": 47901
    },
    {
      "epoch": 4.121311193323582,
      "grad_norm": 556.0,
      "learning_rate": 7.434889221543772e-05,
      "loss": 1.447,
      "step": 47902
    },
    {
      "epoch": 4.1213972296309045,
      "grad_norm": 1.2734375,
      "learning_rate": 7.433470522242703e-05,
      "loss": 1.5283,
      "step": 47903
    },
    {
      "epoch": 4.121483265938226,
      "grad_norm": 0.86328125,
      "learning_rate": 7.432051947440438e-05,
      "loss": 0.4683,
      "step": 47904
    },
    {
      "epoch": 4.121569302245548,
      "grad_norm": 1.3125,
      "learning_rate": 7.430633497141148e-05,
      "loss": 1.353,
      "step": 47905
    },
    {
      "epoch": 4.121655338552869,
      "grad_norm": 1.25,
      "learning_rate": 7.429215171348985e-05,
      "loss": 1.5322,
      "step": 47906
    },
    {
      "epoch": 4.121741374860191,
      "grad_norm": 1.2578125,
      "learning_rate": 7.427796970068074e-05,
      "loss": 0.9258,
      "step": 47907
    },
    {
      "epoch": 4.121827411167513,
      "grad_norm": 1.296875,
      "learning_rate": 7.426378893302582e-05,
      "loss": 1.6391,
      "step": 47908
    },
    {
      "epoch": 4.121913447474834,
      "grad_norm": 1.109375,
      "learning_rate": 7.424960941056658e-05,
      "loss": 1.5502,
      "step": 47909
    },
    {
      "epoch": 4.121999483782156,
      "grad_norm": 1.546875,
      "learning_rate": 7.423543113334435e-05,
      "loss": 2.2117,
      "step": 47910
    },
    {
      "epoch": 4.1220855200894775,
      "grad_norm": 1.109375,
      "learning_rate": 7.422125410140068e-05,
      "loss": 1.3034,
      "step": 47911
    },
    {
      "epoch": 4.1221715563967996,
      "grad_norm": 0.99609375,
      "learning_rate": 7.420707831477714e-05,
      "loss": 1.1849,
      "step": 47912
    },
    {
      "epoch": 4.122257592704121,
      "grad_norm": 0.89453125,
      "learning_rate": 7.419290377351495e-05,
      "loss": 0.7104,
      "step": 47913
    },
    {
      "epoch": 4.122343629011443,
      "grad_norm": 1.046875,
      "learning_rate": 7.417873047765583e-05,
      "loss": 0.8714,
      "step": 47914
    },
    {
      "epoch": 4.122429665318765,
      "grad_norm": 1.03125,
      "learning_rate": 7.416455842724096e-05,
      "loss": 0.8891,
      "step": 47915
    },
    {
      "epoch": 4.122515701626086,
      "grad_norm": 4.625,
      "learning_rate": 7.415038762231196e-05,
      "loss": 1.1281,
      "step": 47916
    },
    {
      "epoch": 4.122601737933408,
      "grad_norm": 1.109375,
      "learning_rate": 7.413621806291043e-05,
      "loss": 1.5002,
      "step": 47917
    },
    {
      "epoch": 4.122687774240729,
      "grad_norm": 1.234375,
      "learning_rate": 7.412204974907744e-05,
      "loss": 1.32,
      "step": 47918
    },
    {
      "epoch": 4.122773810548051,
      "grad_norm": 2.53125,
      "learning_rate": 7.410788268085466e-05,
      "loss": 1.4691,
      "step": 47919
    },
    {
      "epoch": 4.1228598468553725,
      "grad_norm": 0.8515625,
      "learning_rate": 7.409371685828359e-05,
      "loss": 0.3689,
      "step": 47920
    },
    {
      "epoch": 4.122945883162695,
      "grad_norm": 0.9140625,
      "learning_rate": 7.407955228140544e-05,
      "loss": 0.5318,
      "step": 47921
    },
    {
      "epoch": 4.123031919470017,
      "grad_norm": 1.078125,
      "learning_rate": 7.406538895026177e-05,
      "loss": 1.4777,
      "step": 47922
    },
    {
      "epoch": 4.123117955777338,
      "grad_norm": 1.0390625,
      "learning_rate": 7.405122686489413e-05,
      "loss": 1.0872,
      "step": 47923
    },
    {
      "epoch": 4.12320399208466,
      "grad_norm": 1.2578125,
      "learning_rate": 7.403706602534372e-05,
      "loss": 0.9494,
      "step": 47924
    },
    {
      "epoch": 4.123290028391981,
      "grad_norm": 1.4140625,
      "learning_rate": 7.402290643165199e-05,
      "loss": 0.7507,
      "step": 47925
    },
    {
      "epoch": 4.123376064699303,
      "grad_norm": 0.9375,
      "learning_rate": 7.400874808386055e-05,
      "loss": 1.4515,
      "step": 47926
    },
    {
      "epoch": 4.123462101006625,
      "grad_norm": 1.1015625,
      "learning_rate": 7.399459098201056e-05,
      "loss": 1.4648,
      "step": 47927
    },
    {
      "epoch": 4.123548137313946,
      "grad_norm": 15.1875,
      "learning_rate": 7.398043512614366e-05,
      "loss": 0.8326,
      "step": 47928
    },
    {
      "epoch": 4.1236341736212685,
      "grad_norm": 1.734375,
      "learning_rate": 7.3966280516301e-05,
      "loss": 1.1001,
      "step": 47929
    },
    {
      "epoch": 4.12372020992859,
      "grad_norm": 1.1015625,
      "learning_rate": 7.395212715252409e-05,
      "loss": 1.6632,
      "step": 47930
    },
    {
      "epoch": 4.123806246235912,
      "grad_norm": 1.1484375,
      "learning_rate": 7.39379750348545e-05,
      "loss": 1.4528,
      "step": 47931
    },
    {
      "epoch": 4.123892282543233,
      "grad_norm": 1.3203125,
      "learning_rate": 7.392382416333331e-05,
      "loss": 1.1501,
      "step": 47932
    },
    {
      "epoch": 4.123978318850555,
      "grad_norm": 1.078125,
      "learning_rate": 7.390967453800213e-05,
      "loss": 0.8112,
      "step": 47933
    },
    {
      "epoch": 4.124064355157877,
      "grad_norm": 1.1640625,
      "learning_rate": 7.389552615890233e-05,
      "loss": 1.1073,
      "step": 47934
    },
    {
      "epoch": 4.124150391465198,
      "grad_norm": 1.9140625,
      "learning_rate": 7.388137902607506e-05,
      "loss": 0.552,
      "step": 47935
    },
    {
      "epoch": 4.12423642777252,
      "grad_norm": 0.97265625,
      "learning_rate": 7.386723313956212e-05,
      "loss": 1.2475,
      "step": 47936
    },
    {
      "epoch": 4.1243224640798415,
      "grad_norm": 1.7421875,
      "learning_rate": 7.385308849940447e-05,
      "loss": 1.6951,
      "step": 47937
    },
    {
      "epoch": 4.124408500387164,
      "grad_norm": 0.8984375,
      "learning_rate": 7.383894510564371e-05,
      "loss": 0.7515,
      "step": 47938
    },
    {
      "epoch": 4.124494536694485,
      "grad_norm": 0.81640625,
      "learning_rate": 7.382480295832123e-05,
      "loss": 0.7445,
      "step": 47939
    },
    {
      "epoch": 4.124580573001807,
      "grad_norm": 70.5,
      "learning_rate": 7.381066205747822e-05,
      "loss": 1.0082,
      "step": 47940
    },
    {
      "epoch": 4.124666609309129,
      "grad_norm": 1.1171875,
      "learning_rate": 7.379652240315616e-05,
      "loss": 0.9185,
      "step": 47941
    },
    {
      "epoch": 4.12475264561645,
      "grad_norm": 1.078125,
      "learning_rate": 7.378238399539644e-05,
      "loss": 1.0502,
      "step": 47942
    },
    {
      "epoch": 4.124838681923772,
      "grad_norm": 0.8515625,
      "learning_rate": 7.376824683424026e-05,
      "loss": 1.2168,
      "step": 47943
    },
    {
      "epoch": 4.124924718231093,
      "grad_norm": 1.1796875,
      "learning_rate": 7.375411091972922e-05,
      "loss": 1.0142,
      "step": 47944
    },
    {
      "epoch": 4.125010754538415,
      "grad_norm": 1.484375,
      "learning_rate": 7.373997625190432e-05,
      "loss": 0.7466,
      "step": 47945
    },
    {
      "epoch": 4.1250967908457365,
      "grad_norm": 1.140625,
      "learning_rate": 7.372584283080713e-05,
      "loss": 0.8332,
      "step": 47946
    },
    {
      "epoch": 4.125182827153059,
      "grad_norm": 0.94140625,
      "learning_rate": 7.371171065647903e-05,
      "loss": 0.9258,
      "step": 47947
    },
    {
      "epoch": 4.125268863460381,
      "grad_norm": 1.1796875,
      "learning_rate": 7.369757972896119e-05,
      "loss": 1.5413,
      "step": 47948
    },
    {
      "epoch": 4.125354899767702,
      "grad_norm": 1.1328125,
      "learning_rate": 7.368345004829497e-05,
      "loss": 0.589,
      "step": 47949
    },
    {
      "epoch": 4.125440936075024,
      "grad_norm": 1.1640625,
      "learning_rate": 7.36693216145219e-05,
      "loss": 1.8424,
      "step": 47950
    },
    {
      "epoch": 4.125526972382345,
      "grad_norm": 0.96484375,
      "learning_rate": 7.365519442768303e-05,
      "loss": 0.6025,
      "step": 47951
    },
    {
      "epoch": 4.125613008689667,
      "grad_norm": 67.5,
      "learning_rate": 7.364106848781977e-05,
      "loss": 1.8111,
      "step": 47952
    },
    {
      "epoch": 4.125699044996988,
      "grad_norm": 1.1875,
      "learning_rate": 7.362694379497359e-05,
      "loss": 1.0362,
      "step": 47953
    },
    {
      "epoch": 4.12578508130431,
      "grad_norm": 1.109375,
      "learning_rate": 7.36128203491856e-05,
      "loss": 1.4472,
      "step": 47954
    },
    {
      "epoch": 4.1258711176116325,
      "grad_norm": 0.8359375,
      "learning_rate": 7.359869815049714e-05,
      "loss": 0.5898,
      "step": 47955
    },
    {
      "epoch": 4.125957153918954,
      "grad_norm": 0.90234375,
      "learning_rate": 7.358457719894968e-05,
      "loss": 1.1974,
      "step": 47956
    },
    {
      "epoch": 4.126043190226276,
      "grad_norm": 4.875,
      "learning_rate": 7.357045749458424e-05,
      "loss": 0.8215,
      "step": 47957
    },
    {
      "epoch": 4.126129226533597,
      "grad_norm": 0.6796875,
      "learning_rate": 7.355633903744246e-05,
      "loss": 0.4544,
      "step": 47958
    },
    {
      "epoch": 4.126215262840919,
      "grad_norm": 2.078125,
      "learning_rate": 7.354222182756532e-05,
      "loss": 1.7044,
      "step": 47959
    },
    {
      "epoch": 4.12630129914824,
      "grad_norm": 13.375,
      "learning_rate": 7.35281058649942e-05,
      "loss": 0.681,
      "step": 47960
    },
    {
      "epoch": 4.126387335455562,
      "grad_norm": 1.328125,
      "learning_rate": 7.351399114977059e-05,
      "loss": 1.02,
      "step": 47961
    },
    {
      "epoch": 4.126473371762884,
      "grad_norm": 1.078125,
      "learning_rate": 7.349987768193544e-05,
      "loss": 1.5734,
      "step": 47962
    },
    {
      "epoch": 4.1265594080702055,
      "grad_norm": 0.859375,
      "learning_rate": 7.348576546153024e-05,
      "loss": 1.1504,
      "step": 47963
    },
    {
      "epoch": 4.126645444377528,
      "grad_norm": 1.0859375,
      "learning_rate": 7.347165448859632e-05,
      "loss": 1.0276,
      "step": 47964
    },
    {
      "epoch": 4.126731480684849,
      "grad_norm": 1.6171875,
      "learning_rate": 7.34575447631748e-05,
      "loss": 1.1295,
      "step": 47965
    },
    {
      "epoch": 4.126817516992171,
      "grad_norm": 1.1015625,
      "learning_rate": 7.344343628530692e-05,
      "loss": 1.9431,
      "step": 47966
    },
    {
      "epoch": 4.126903553299492,
      "grad_norm": 56.0,
      "learning_rate": 7.342932905503419e-05,
      "loss": 0.8643,
      "step": 47967
    },
    {
      "epoch": 4.126989589606814,
      "grad_norm": 1.609375,
      "learning_rate": 7.341522307239756e-05,
      "loss": 0.9485,
      "step": 47968
    },
    {
      "epoch": 4.127075625914136,
      "grad_norm": 1.4140625,
      "learning_rate": 7.340111833743856e-05,
      "loss": 1.266,
      "step": 47969
    },
    {
      "epoch": 4.127161662221457,
      "grad_norm": 0.9453125,
      "learning_rate": 7.338701485019821e-05,
      "loss": 0.7445,
      "step": 47970
    },
    {
      "epoch": 4.127247698528779,
      "grad_norm": 1.09375,
      "learning_rate": 7.337291261071787e-05,
      "loss": 1.49,
      "step": 47971
    },
    {
      "epoch": 4.1273337348361006,
      "grad_norm": 0.91015625,
      "learning_rate": 7.335881161903885e-05,
      "loss": 1.1752,
      "step": 47972
    },
    {
      "epoch": 4.127419771143423,
      "grad_norm": 27.125,
      "learning_rate": 7.334471187520225e-05,
      "loss": 0.7897,
      "step": 47973
    },
    {
      "epoch": 4.127505807450744,
      "grad_norm": 0.75390625,
      "learning_rate": 7.333061337924951e-05,
      "loss": 0.5292,
      "step": 47974
    },
    {
      "epoch": 4.127591843758066,
      "grad_norm": 1.1484375,
      "learning_rate": 7.331651613122164e-05,
      "loss": 1.242,
      "step": 47975
    },
    {
      "epoch": 4.127677880065388,
      "grad_norm": 1.0703125,
      "learning_rate": 7.330242013116e-05,
      "loss": 0.7005,
      "step": 47976
    },
    {
      "epoch": 4.127763916372709,
      "grad_norm": 0.85546875,
      "learning_rate": 7.328832537910584e-05,
      "loss": 0.934,
      "step": 47977
    },
    {
      "epoch": 4.127849952680031,
      "grad_norm": 1.0546875,
      "learning_rate": 7.32742318751003e-05,
      "loss": 1.1993,
      "step": 47978
    },
    {
      "epoch": 4.127935988987352,
      "grad_norm": 12.625,
      "learning_rate": 7.32601396191846e-05,
      "loss": 1.4519,
      "step": 47979
    },
    {
      "epoch": 4.128022025294674,
      "grad_norm": 1.328125,
      "learning_rate": 7.324604861140011e-05,
      "loss": 1.2912,
      "step": 47980
    },
    {
      "epoch": 4.128108061601996,
      "grad_norm": 1.1171875,
      "learning_rate": 7.323195885178785e-05,
      "loss": 1.288,
      "step": 47981
    },
    {
      "epoch": 4.128194097909318,
      "grad_norm": 0.73046875,
      "learning_rate": 7.321787034038907e-05,
      "loss": 0.7566,
      "step": 47982
    },
    {
      "epoch": 4.12828013421664,
      "grad_norm": 1.4765625,
      "learning_rate": 7.320378307724512e-05,
      "loss": 1.0117,
      "step": 47983
    },
    {
      "epoch": 4.128366170523961,
      "grad_norm": 2.5625,
      "learning_rate": 7.318969706239703e-05,
      "loss": 1.3224,
      "step": 47984
    },
    {
      "epoch": 4.128452206831283,
      "grad_norm": 1.1015625,
      "learning_rate": 7.317561229588616e-05,
      "loss": 1.0671,
      "step": 47985
    },
    {
      "epoch": 4.128538243138604,
      "grad_norm": 1.0625,
      "learning_rate": 7.316152877775345e-05,
      "loss": 0.8708,
      "step": 47986
    },
    {
      "epoch": 4.128624279445926,
      "grad_norm": 1.5390625,
      "learning_rate": 7.314744650804028e-05,
      "loss": 1.4372,
      "step": 47987
    },
    {
      "epoch": 4.128710315753247,
      "grad_norm": 1.4921875,
      "learning_rate": 7.313336548678795e-05,
      "loss": 1.0354,
      "step": 47988
    },
    {
      "epoch": 4.1287963520605695,
      "grad_norm": 0.859375,
      "learning_rate": 7.311928571403737e-05,
      "loss": 1.1605,
      "step": 47989
    },
    {
      "epoch": 4.128882388367892,
      "grad_norm": 1.4296875,
      "learning_rate": 7.310520718982988e-05,
      "loss": 1.0597,
      "step": 47990
    },
    {
      "epoch": 4.128968424675213,
      "grad_norm": 1.2578125,
      "learning_rate": 7.309112991420675e-05,
      "loss": 1.7215,
      "step": 47991
    },
    {
      "epoch": 4.129054460982535,
      "grad_norm": 1.109375,
      "learning_rate": 7.307705388720887e-05,
      "loss": 0.8359,
      "step": 47992
    },
    {
      "epoch": 4.129140497289856,
      "grad_norm": 101.5,
      "learning_rate": 7.30629791088776e-05,
      "loss": 0.8989,
      "step": 47993
    },
    {
      "epoch": 4.129226533597178,
      "grad_norm": 1.3125,
      "learning_rate": 7.30489055792542e-05,
      "loss": 1.1113,
      "step": 47994
    },
    {
      "epoch": 4.129312569904499,
      "grad_norm": 1.0234375,
      "learning_rate": 7.303483329837962e-05,
      "loss": 0.7951,
      "step": 47995
    },
    {
      "epoch": 4.129398606211821,
      "grad_norm": 0.91015625,
      "learning_rate": 7.302076226629512e-05,
      "loss": 0.8837,
      "step": 47996
    },
    {
      "epoch": 4.129484642519143,
      "grad_norm": 131.0,
      "learning_rate": 7.300669248304192e-05,
      "loss": 0.7574,
      "step": 47997
    },
    {
      "epoch": 4.129570678826465,
      "grad_norm": 67.5,
      "learning_rate": 7.299262394866096e-05,
      "loss": 0.8926,
      "step": 47998
    },
    {
      "epoch": 4.129656715133787,
      "grad_norm": 0.85546875,
      "learning_rate": 7.29785566631937e-05,
      "loss": 0.5561,
      "step": 47999
    },
    {
      "epoch": 4.129742751441108,
      "grad_norm": 1.5,
      "learning_rate": 7.296449062668098e-05,
      "loss": 1.2933,
      "step": 48000
    },
    {
      "epoch": 4.12982878774843,
      "grad_norm": 1.1328125,
      "learning_rate": 7.295042583916406e-05,
      "loss": 1.3186,
      "step": 48001
    },
    {
      "epoch": 4.129914824055752,
      "grad_norm": 1.15625,
      "learning_rate": 7.293636230068424e-05,
      "loss": 1.462,
      "step": 48002
    },
    {
      "epoch": 4.130000860363073,
      "grad_norm": 1.2109375,
      "learning_rate": 7.292230001128235e-05,
      "loss": 0.893,
      "step": 48003
    },
    {
      "epoch": 4.130086896670395,
      "grad_norm": 1.2265625,
      "learning_rate": 7.290823897099969e-05,
      "loss": 1.4951,
      "step": 48004
    },
    {
      "epoch": 4.130172932977716,
      "grad_norm": 1.1171875,
      "learning_rate": 7.289417917987745e-05,
      "loss": 1.9539,
      "step": 48005
    },
    {
      "epoch": 4.130258969285038,
      "grad_norm": 1.6953125,
      "learning_rate": 7.288012063795657e-05,
      "loss": 1.1934,
      "step": 48006
    },
    {
      "epoch": 4.13034500559236,
      "grad_norm": 0.7890625,
      "learning_rate": 7.286606334527829e-05,
      "loss": 0.636,
      "step": 48007
    },
    {
      "epoch": 4.130431041899682,
      "grad_norm": 0.91796875,
      "learning_rate": 7.285200730188379e-05,
      "loss": 0.7458,
      "step": 48008
    },
    {
      "epoch": 4.130517078207004,
      "grad_norm": 1.2421875,
      "learning_rate": 7.283795250781388e-05,
      "loss": 1.5949,
      "step": 48009
    },
    {
      "epoch": 4.130603114514325,
      "grad_norm": 1.4140625,
      "learning_rate": 7.282389896311009e-05,
      "loss": 1.4156,
      "step": 48010
    },
    {
      "epoch": 4.130689150821647,
      "grad_norm": 0.98046875,
      "learning_rate": 7.280984666781321e-05,
      "loss": 0.4604,
      "step": 48011
    },
    {
      "epoch": 4.130775187128968,
      "grad_norm": 1.953125,
      "learning_rate": 7.279579562196443e-05,
      "loss": 1.7166,
      "step": 48012
    },
    {
      "epoch": 4.13086122343629,
      "grad_norm": 1.546875,
      "learning_rate": 7.278174582560499e-05,
      "loss": 1.1711,
      "step": 48013
    },
    {
      "epoch": 4.130947259743611,
      "grad_norm": 1.4765625,
      "learning_rate": 7.276769727877575e-05,
      "loss": 1.5946,
      "step": 48014
    },
    {
      "epoch": 4.1310332960509335,
      "grad_norm": 1.5390625,
      "learning_rate": 7.275364998151802e-05,
      "loss": 1.389,
      "step": 48015
    },
    {
      "epoch": 4.131119332358256,
      "grad_norm": 1.4765625,
      "learning_rate": 7.273960393387263e-05,
      "loss": 1.0342,
      "step": 48016
    },
    {
      "epoch": 4.131205368665577,
      "grad_norm": 1.1171875,
      "learning_rate": 7.27255591358808e-05,
      "loss": 0.9364,
      "step": 48017
    },
    {
      "epoch": 4.131291404972899,
      "grad_norm": 0.97265625,
      "learning_rate": 7.271151558758376e-05,
      "loss": 0.6891,
      "step": 48018
    },
    {
      "epoch": 4.13137744128022,
      "grad_norm": 1.34375,
      "learning_rate": 7.269747328902227e-05,
      "loss": 1.2482,
      "step": 48019
    },
    {
      "epoch": 4.131463477587542,
      "grad_norm": 1.3125,
      "learning_rate": 7.26834322402376e-05,
      "loss": 0.6123,
      "step": 48020
    },
    {
      "epoch": 4.131549513894863,
      "grad_norm": 0.9140625,
      "learning_rate": 7.26693924412709e-05,
      "loss": 0.8643,
      "step": 48021
    },
    {
      "epoch": 4.131635550202185,
      "grad_norm": 1.546875,
      "learning_rate": 7.265535389216299e-05,
      "loss": 1.2102,
      "step": 48022
    },
    {
      "epoch": 4.131721586509507,
      "grad_norm": 1.2734375,
      "learning_rate": 7.264131659295503e-05,
      "loss": 0.909,
      "step": 48023
    },
    {
      "epoch": 4.131807622816829,
      "grad_norm": 1.234375,
      "learning_rate": 7.262728054368822e-05,
      "loss": 1.0533,
      "step": 48024
    },
    {
      "epoch": 4.131893659124151,
      "grad_norm": 1.078125,
      "learning_rate": 7.261324574440343e-05,
      "loss": 1.5144,
      "step": 48025
    },
    {
      "epoch": 4.131979695431472,
      "grad_norm": 1.2890625,
      "learning_rate": 7.259921219514187e-05,
      "loss": 1.4138,
      "step": 48026
    },
    {
      "epoch": 4.132065731738794,
      "grad_norm": 34.5,
      "learning_rate": 7.258517989594437e-05,
      "loss": 1.3814,
      "step": 48027
    },
    {
      "epoch": 4.132151768046115,
      "grad_norm": 1.609375,
      "learning_rate": 7.257114884685206e-05,
      "loss": 1.0298,
      "step": 48028
    },
    {
      "epoch": 4.132237804353437,
      "grad_norm": 8.25,
      "learning_rate": 7.255711904790618e-05,
      "loss": 1.1576,
      "step": 48029
    },
    {
      "epoch": 4.132323840660759,
      "grad_norm": 1.0078125,
      "learning_rate": 7.254309049914743e-05,
      "loss": 1.2649,
      "step": 48030
    },
    {
      "epoch": 4.13240987696808,
      "grad_norm": 1.0390625,
      "learning_rate": 7.252906320061703e-05,
      "loss": 0.9008,
      "step": 48031
    },
    {
      "epoch": 4.132495913275402,
      "grad_norm": 1.1328125,
      "learning_rate": 7.251503715235608e-05,
      "loss": 0.7159,
      "step": 48032
    },
    {
      "epoch": 4.132581949582724,
      "grad_norm": 0.921875,
      "learning_rate": 7.250101235440542e-05,
      "loss": 0.5776,
      "step": 48033
    },
    {
      "epoch": 4.132667985890046,
      "grad_norm": 2.90625,
      "learning_rate": 7.248698880680615e-05,
      "loss": 1.7694,
      "step": 48034
    },
    {
      "epoch": 4.132754022197367,
      "grad_norm": 9.5625,
      "learning_rate": 7.24729665095994e-05,
      "loss": 0.8055,
      "step": 48035
    },
    {
      "epoch": 4.132840058504689,
      "grad_norm": 1.0546875,
      "learning_rate": 7.245894546282594e-05,
      "loss": 0.7272,
      "step": 48036
    },
    {
      "epoch": 4.132926094812011,
      "grad_norm": 8.4375,
      "learning_rate": 7.244492566652688e-05,
      "loss": 0.9793,
      "step": 48037
    },
    {
      "epoch": 4.133012131119332,
      "grad_norm": 1.234375,
      "learning_rate": 7.243090712074347e-05,
      "loss": 1.0852,
      "step": 48038
    },
    {
      "epoch": 4.133098167426654,
      "grad_norm": 0.92578125,
      "learning_rate": 7.241688982551631e-05,
      "loss": 0.9975,
      "step": 48039
    },
    {
      "epoch": 4.133184203733975,
      "grad_norm": 0.82421875,
      "learning_rate": 7.240287378088672e-05,
      "loss": 0.7772,
      "step": 48040
    },
    {
      "epoch": 4.1332702400412975,
      "grad_norm": 1.2890625,
      "learning_rate": 7.238885898689546e-05,
      "loss": 1.2865,
      "step": 48041
    },
    {
      "epoch": 4.13335627634862,
      "grad_norm": 1.2265625,
      "learning_rate": 7.237484544358359e-05,
      "loss": 1.3718,
      "step": 48042
    },
    {
      "epoch": 4.133442312655941,
      "grad_norm": 1.2734375,
      "learning_rate": 7.236083315099224e-05,
      "loss": 1.4202,
      "step": 48043
    },
    {
      "epoch": 4.133528348963263,
      "grad_norm": 0.87109375,
      "learning_rate": 7.234682210916221e-05,
      "loss": 0.9582,
      "step": 48044
    },
    {
      "epoch": 4.133614385270584,
      "grad_norm": 1.0078125,
      "learning_rate": 7.233281231813455e-05,
      "loss": 1.4188,
      "step": 48045
    },
    {
      "epoch": 4.133700421577906,
      "grad_norm": 1.8359375,
      "learning_rate": 7.23188037779502e-05,
      "loss": 0.9575,
      "step": 48046
    },
    {
      "epoch": 4.133786457885227,
      "grad_norm": 0.9609375,
      "learning_rate": 7.230479648865019e-05,
      "loss": 0.7985,
      "step": 48047
    },
    {
      "epoch": 4.133872494192549,
      "grad_norm": 1.109375,
      "learning_rate": 7.229079045027559e-05,
      "loss": 1.1319,
      "step": 48048
    },
    {
      "epoch": 4.133958530499871,
      "grad_norm": 1.1875,
      "learning_rate": 7.22767856628671e-05,
      "loss": 1.5922,
      "step": 48049
    },
    {
      "epoch": 4.134044566807193,
      "grad_norm": 1.265625,
      "learning_rate": 7.226278212646587e-05,
      "loss": 1.1671,
      "step": 48050
    },
    {
      "epoch": 4.134130603114515,
      "grad_norm": 1.28125,
      "learning_rate": 7.224877984111289e-05,
      "loss": 1.3601,
      "step": 48051
    },
    {
      "epoch": 4.134216639421836,
      "grad_norm": 1.5,
      "learning_rate": 7.223477880684893e-05,
      "loss": 1.0971,
      "step": 48052
    },
    {
      "epoch": 4.134302675729158,
      "grad_norm": 1.234375,
      "learning_rate": 7.222077902371504e-05,
      "loss": 1.4081,
      "step": 48053
    },
    {
      "epoch": 4.134388712036479,
      "grad_norm": 1.3046875,
      "learning_rate": 7.220678049175228e-05,
      "loss": 1.2935,
      "step": 48054
    },
    {
      "epoch": 4.134474748343801,
      "grad_norm": 11.375,
      "learning_rate": 7.219278321100143e-05,
      "loss": 0.9105,
      "step": 48055
    },
    {
      "epoch": 4.134560784651123,
      "grad_norm": 0.828125,
      "learning_rate": 7.21787871815035e-05,
      "loss": 0.7992,
      "step": 48056
    },
    {
      "epoch": 4.134646820958444,
      "grad_norm": 1.1796875,
      "learning_rate": 7.216479240329937e-05,
      "loss": 1.6977,
      "step": 48057
    },
    {
      "epoch": 4.1347328572657664,
      "grad_norm": 0.84375,
      "learning_rate": 7.215079887643e-05,
      "loss": 0.7836,
      "step": 48058
    },
    {
      "epoch": 4.134818893573088,
      "grad_norm": 0.8828125,
      "learning_rate": 7.213680660093646e-05,
      "loss": 1.2954,
      "step": 48059
    },
    {
      "epoch": 4.13490492988041,
      "grad_norm": 1.15625,
      "learning_rate": 7.21228155768594e-05,
      "loss": 1.4027,
      "step": 48060
    },
    {
      "epoch": 4.134990966187731,
      "grad_norm": 1.2265625,
      "learning_rate": 7.210882580423989e-05,
      "loss": 0.7727,
      "step": 48061
    },
    {
      "epoch": 4.135077002495053,
      "grad_norm": 1.1484375,
      "learning_rate": 7.2094837283119e-05,
      "loss": 1.0867,
      "step": 48062
    },
    {
      "epoch": 4.135163038802375,
      "grad_norm": 1.6796875,
      "learning_rate": 7.208085001353731e-05,
      "loss": 1.0321,
      "step": 48063
    },
    {
      "epoch": 4.135249075109696,
      "grad_norm": 0.90625,
      "learning_rate": 7.206686399553597e-05,
      "loss": 0.4282,
      "step": 48064
    },
    {
      "epoch": 4.135335111417018,
      "grad_norm": 1.0,
      "learning_rate": 7.205287922915588e-05,
      "loss": 1.1917,
      "step": 48065
    },
    {
      "epoch": 4.135421147724339,
      "grad_norm": 0.953125,
      "learning_rate": 7.203889571443783e-05,
      "loss": 1.5811,
      "step": 48066
    },
    {
      "epoch": 4.1355071840316615,
      "grad_norm": 0.921875,
      "learning_rate": 7.202491345142287e-05,
      "loss": 0.7143,
      "step": 48067
    },
    {
      "epoch": 4.135593220338983,
      "grad_norm": 0.91015625,
      "learning_rate": 7.201093244015167e-05,
      "loss": 1.0616,
      "step": 48068
    },
    {
      "epoch": 4.135679256646305,
      "grad_norm": 1.1015625,
      "learning_rate": 7.199695268066525e-05,
      "loss": 1.5621,
      "step": 48069
    },
    {
      "epoch": 4.135765292953627,
      "grad_norm": 20.75,
      "learning_rate": 7.198297417300465e-05,
      "loss": 1.2657,
      "step": 48070
    },
    {
      "epoch": 4.135851329260948,
      "grad_norm": 1.25,
      "learning_rate": 7.196899691721043e-05,
      "loss": 1.0629,
      "step": 48071
    },
    {
      "epoch": 4.13593736556827,
      "grad_norm": 1.2578125,
      "learning_rate": 7.195502091332368e-05,
      "loss": 0.9296,
      "step": 48072
    },
    {
      "epoch": 4.136023401875591,
      "grad_norm": 0.88671875,
      "learning_rate": 7.194104616138542e-05,
      "loss": 0.6714,
      "step": 48073
    },
    {
      "epoch": 4.136109438182913,
      "grad_norm": 1.4921875,
      "learning_rate": 7.192707266143617e-05,
      "loss": 0.9214,
      "step": 48074
    },
    {
      "epoch": 4.1361954744902345,
      "grad_norm": 1.1484375,
      "learning_rate": 7.191310041351695e-05,
      "loss": 0.9767,
      "step": 48075
    },
    {
      "epoch": 4.136281510797557,
      "grad_norm": 1.1796875,
      "learning_rate": 7.18991294176688e-05,
      "loss": 1.7215,
      "step": 48076
    },
    {
      "epoch": 4.136367547104879,
      "grad_norm": 0.99609375,
      "learning_rate": 7.188515967393233e-05,
      "loss": 0.6702,
      "step": 48077
    },
    {
      "epoch": 4.1364535834122,
      "grad_norm": 1.1015625,
      "learning_rate": 7.187119118234853e-05,
      "loss": 1.1917,
      "step": 48078
    },
    {
      "epoch": 4.136539619719522,
      "grad_norm": 0.8984375,
      "learning_rate": 7.18572239429583e-05,
      "loss": 1.0121,
      "step": 48079
    },
    {
      "epoch": 4.136625656026843,
      "grad_norm": 8.875,
      "learning_rate": 7.184325795580232e-05,
      "loss": 0.7843,
      "step": 48080
    },
    {
      "epoch": 4.136711692334165,
      "grad_norm": 0.9296875,
      "learning_rate": 7.182929322092163e-05,
      "loss": 0.5666,
      "step": 48081
    },
    {
      "epoch": 4.136797728641486,
      "grad_norm": 1.171875,
      "learning_rate": 7.181532973835692e-05,
      "loss": 0.9755,
      "step": 48082
    },
    {
      "epoch": 4.136883764948808,
      "grad_norm": 0.8984375,
      "learning_rate": 7.180136750814891e-05,
      "loss": 0.5375,
      "step": 48083
    },
    {
      "epoch": 4.1369698012561305,
      "grad_norm": 0.98828125,
      "learning_rate": 7.178740653033889e-05,
      "loss": 0.6655,
      "step": 48084
    },
    {
      "epoch": 4.137055837563452,
      "grad_norm": 2.453125,
      "learning_rate": 7.177344680496734e-05,
      "loss": 1.0079,
      "step": 48085
    },
    {
      "epoch": 4.137141873870774,
      "grad_norm": 0.73828125,
      "learning_rate": 7.175948833207525e-05,
      "loss": 0.4138,
      "step": 48086
    },
    {
      "epoch": 4.137227910178095,
      "grad_norm": 1.0390625,
      "learning_rate": 7.174553111170323e-05,
      "loss": 1.9095,
      "step": 48087
    },
    {
      "epoch": 4.137313946485417,
      "grad_norm": 1.5390625,
      "learning_rate": 7.173157514389228e-05,
      "loss": 1.3056,
      "step": 48088
    },
    {
      "epoch": 4.137399982792738,
      "grad_norm": 10.0,
      "learning_rate": 7.171762042868323e-05,
      "loss": 0.9381,
      "step": 48089
    },
    {
      "epoch": 4.13748601910006,
      "grad_norm": 0.96875,
      "learning_rate": 7.17036669661168e-05,
      "loss": 0.7103,
      "step": 48090
    },
    {
      "epoch": 4.137572055407382,
      "grad_norm": 1.4609375,
      "learning_rate": 7.168971475623381e-05,
      "loss": 1.4767,
      "step": 48091
    },
    {
      "epoch": 4.1376580917147034,
      "grad_norm": 1.421875,
      "learning_rate": 7.167576379907525e-05,
      "loss": 0.8902,
      "step": 48092
    },
    {
      "epoch": 4.1377441280220255,
      "grad_norm": 1.0703125,
      "learning_rate": 7.166181409468164e-05,
      "loss": 1.0674,
      "step": 48093
    },
    {
      "epoch": 4.137830164329347,
      "grad_norm": 0.9609375,
      "learning_rate": 7.164786564309395e-05,
      "loss": 0.9164,
      "step": 48094
    },
    {
      "epoch": 4.137916200636669,
      "grad_norm": 1.25,
      "learning_rate": 7.163391844435302e-05,
      "loss": 0.5877,
      "step": 48095
    },
    {
      "epoch": 4.13800223694399,
      "grad_norm": 1.0234375,
      "learning_rate": 7.161997249849945e-05,
      "loss": 1.044,
      "step": 48096
    },
    {
      "epoch": 4.138088273251312,
      "grad_norm": 1.5234375,
      "learning_rate": 7.16060278055743e-05,
      "loss": 1.5936,
      "step": 48097
    },
    {
      "epoch": 4.138174309558634,
      "grad_norm": 1.0390625,
      "learning_rate": 7.159208436561804e-05,
      "loss": 1.1563,
      "step": 48098
    },
    {
      "epoch": 4.138260345865955,
      "grad_norm": 1.265625,
      "learning_rate": 7.157814217867158e-05,
      "loss": 1.9751,
      "step": 48099
    },
    {
      "epoch": 4.138346382173277,
      "grad_norm": 1.1796875,
      "learning_rate": 7.156420124477591e-05,
      "loss": 1.61,
      "step": 48100
    },
    {
      "epoch": 4.1384324184805985,
      "grad_norm": 1.140625,
      "learning_rate": 7.155026156397148e-05,
      "loss": 1.1007,
      "step": 48101
    },
    {
      "epoch": 4.138518454787921,
      "grad_norm": 1.2578125,
      "learning_rate": 7.153632313629927e-05,
      "loss": 1.1832,
      "step": 48102
    },
    {
      "epoch": 4.138604491095242,
      "grad_norm": 0.98046875,
      "learning_rate": 7.152238596180005e-05,
      "loss": 1.0744,
      "step": 48103
    },
    {
      "epoch": 4.138690527402564,
      "grad_norm": 0.6875,
      "learning_rate": 7.150845004051437e-05,
      "loss": 0.7663,
      "step": 48104
    },
    {
      "epoch": 4.138776563709886,
      "grad_norm": 1.6875,
      "learning_rate": 7.149451537248319e-05,
      "loss": 1.4913,
      "step": 48105
    },
    {
      "epoch": 4.138862600017207,
      "grad_norm": 1.34375,
      "learning_rate": 7.148058195774732e-05,
      "loss": 1.1206,
      "step": 48106
    },
    {
      "epoch": 4.138948636324529,
      "grad_norm": 182.0,
      "learning_rate": 7.146664979634726e-05,
      "loss": 1.2613,
      "step": 48107
    },
    {
      "epoch": 4.13903467263185,
      "grad_norm": 6.375,
      "learning_rate": 7.145271888832389e-05,
      "loss": 1.1864,
      "step": 48108
    },
    {
      "epoch": 4.139120708939172,
      "grad_norm": 1.3125,
      "learning_rate": 7.143878923371811e-05,
      "loss": 0.6754,
      "step": 48109
    },
    {
      "epoch": 4.139206745246494,
      "grad_norm": 1.84375,
      "learning_rate": 7.142486083257044e-05,
      "loss": 1.4278,
      "step": 48110
    },
    {
      "epoch": 4.139292781553816,
      "grad_norm": 1.140625,
      "learning_rate": 7.141093368492175e-05,
      "loss": 1.1194,
      "step": 48111
    },
    {
      "epoch": 4.139378817861138,
      "grad_norm": 0.76171875,
      "learning_rate": 7.139700779081265e-05,
      "loss": 0.5776,
      "step": 48112
    },
    {
      "epoch": 4.139464854168459,
      "grad_norm": 1.171875,
      "learning_rate": 7.138308315028397e-05,
      "loss": 1.7371,
      "step": 48113
    },
    {
      "epoch": 4.139550890475781,
      "grad_norm": 1.1328125,
      "learning_rate": 7.136915976337649e-05,
      "loss": 1.1698,
      "step": 48114
    },
    {
      "epoch": 4.139636926783102,
      "grad_norm": 1.1640625,
      "learning_rate": 7.135523763013068e-05,
      "loss": 0.8792,
      "step": 48115
    },
    {
      "epoch": 4.139722963090424,
      "grad_norm": 1.5703125,
      "learning_rate": 7.134131675058753e-05,
      "loss": 0.875,
      "step": 48116
    },
    {
      "epoch": 4.139808999397746,
      "grad_norm": 1.3046875,
      "learning_rate": 7.132739712478776e-05,
      "loss": 1.5401,
      "step": 48117
    },
    {
      "epoch": 4.1398950357050674,
      "grad_norm": 5024.0,
      "learning_rate": 7.131347875277183e-05,
      "loss": 0.7468,
      "step": 48118
    },
    {
      "epoch": 4.1399810720123895,
      "grad_norm": 1.2890625,
      "learning_rate": 7.129956163458057e-05,
      "loss": 0.9139,
      "step": 48119
    },
    {
      "epoch": 4.140067108319711,
      "grad_norm": 38.25,
      "learning_rate": 7.128564577025475e-05,
      "loss": 1.1158,
      "step": 48120
    },
    {
      "epoch": 4.140153144627033,
      "grad_norm": 1.1328125,
      "learning_rate": 7.12717311598351e-05,
      "loss": 0.9799,
      "step": 48121
    },
    {
      "epoch": 4.140239180934354,
      "grad_norm": 1.0390625,
      "learning_rate": 7.125781780336227e-05,
      "loss": 0.7108,
      "step": 48122
    },
    {
      "epoch": 4.140325217241676,
      "grad_norm": 8.875,
      "learning_rate": 7.124390570087685e-05,
      "loss": 1.1301,
      "step": 48123
    },
    {
      "epoch": 4.140411253548998,
      "grad_norm": 1.6484375,
      "learning_rate": 7.122999485241966e-05,
      "loss": 1.3692,
      "step": 48124
    },
    {
      "epoch": 4.140497289856319,
      "grad_norm": 1.21875,
      "learning_rate": 7.121608525803142e-05,
      "loss": 1.0994,
      "step": 48125
    },
    {
      "epoch": 4.140583326163641,
      "grad_norm": 0.73828125,
      "learning_rate": 7.120217691775266e-05,
      "loss": 1.198,
      "step": 48126
    },
    {
      "epoch": 4.1406693624709625,
      "grad_norm": 1.09375,
      "learning_rate": 7.118826983162419e-05,
      "loss": 1.2437,
      "step": 48127
    },
    {
      "epoch": 4.140755398778285,
      "grad_norm": 1.1328125,
      "learning_rate": 7.117436399968657e-05,
      "loss": 0.6842,
      "step": 48128
    },
    {
      "epoch": 4.140841435085606,
      "grad_norm": 0.89453125,
      "learning_rate": 7.116045942198051e-05,
      "loss": 1.4953,
      "step": 48129
    },
    {
      "epoch": 4.140927471392928,
      "grad_norm": 1200.0,
      "learning_rate": 7.114655609854681e-05,
      "loss": 1.3958,
      "step": 48130
    },
    {
      "epoch": 4.14101350770025,
      "grad_norm": 0.8515625,
      "learning_rate": 7.113265402942592e-05,
      "loss": 0.5779,
      "step": 48131
    },
    {
      "epoch": 4.141099544007571,
      "grad_norm": 1.21875,
      "learning_rate": 7.11187532146586e-05,
      "loss": 1.2921,
      "step": 48132
    },
    {
      "epoch": 4.141185580314893,
      "grad_norm": 37.25,
      "learning_rate": 7.110485365428565e-05,
      "loss": 0.549,
      "step": 48133
    },
    {
      "epoch": 4.141271616622214,
      "grad_norm": 1.1015625,
      "learning_rate": 7.109095534834741e-05,
      "loss": 1.4681,
      "step": 48134
    },
    {
      "epoch": 4.141357652929536,
      "grad_norm": 0.9296875,
      "learning_rate": 7.107705829688477e-05,
      "loss": 0.4675,
      "step": 48135
    },
    {
      "epoch": 4.141443689236858,
      "grad_norm": 1.09375,
      "learning_rate": 7.106316249993844e-05,
      "loss": 0.8362,
      "step": 48136
    },
    {
      "epoch": 4.14152972554418,
      "grad_norm": 1.4453125,
      "learning_rate": 7.104926795754879e-05,
      "loss": 0.9296,
      "step": 48137
    },
    {
      "epoch": 4.141615761851502,
      "grad_norm": 1.265625,
      "learning_rate": 7.103537466975674e-05,
      "loss": 0.8473,
      "step": 48138
    },
    {
      "epoch": 4.141701798158823,
      "grad_norm": 4.375,
      "learning_rate": 7.102148263660264e-05,
      "loss": 1.2773,
      "step": 48139
    },
    {
      "epoch": 4.141787834466145,
      "grad_norm": 1.3359375,
      "learning_rate": 7.100759185812728e-05,
      "loss": 0.8587,
      "step": 48140
    },
    {
      "epoch": 4.141873870773466,
      "grad_norm": 30.375,
      "learning_rate": 7.099370233437141e-05,
      "loss": 1.3831,
      "step": 48141
    },
    {
      "epoch": 4.141959907080788,
      "grad_norm": 0.9921875,
      "learning_rate": 7.09798140653754e-05,
      "loss": 1.5308,
      "step": 48142
    },
    {
      "epoch": 4.142045943388109,
      "grad_norm": 1.1953125,
      "learning_rate": 7.096592705117999e-05,
      "loss": 1.2254,
      "step": 48143
    },
    {
      "epoch": 4.1421319796954315,
      "grad_norm": 2960.0,
      "learning_rate": 7.095204129182592e-05,
      "loss": 0.7183,
      "step": 48144
    },
    {
      "epoch": 4.1422180160027535,
      "grad_norm": 1.21875,
      "learning_rate": 7.09381567873536e-05,
      "loss": 1.4411,
      "step": 48145
    },
    {
      "epoch": 4.142304052310075,
      "grad_norm": 1.1796875,
      "learning_rate": 7.092427353780367e-05,
      "loss": 0.638,
      "step": 48146
    },
    {
      "epoch": 4.142390088617397,
      "grad_norm": 6.34375,
      "learning_rate": 7.091039154321694e-05,
      "loss": 1.0924,
      "step": 48147
    },
    {
      "epoch": 4.142476124924718,
      "grad_norm": 1.0078125,
      "learning_rate": 7.089651080363375e-05,
      "loss": 1.1809,
      "step": 48148
    },
    {
      "epoch": 4.14256216123204,
      "grad_norm": 0.8671875,
      "learning_rate": 7.088263131909478e-05,
      "loss": 1.7999,
      "step": 48149
    },
    {
      "epoch": 4.142648197539361,
      "grad_norm": 2.640625,
      "learning_rate": 7.08687530896408e-05,
      "loss": 0.6907,
      "step": 48150
    },
    {
      "epoch": 4.142734233846683,
      "grad_norm": 1.328125,
      "learning_rate": 7.08548761153121e-05,
      "loss": 1.1168,
      "step": 48151
    },
    {
      "epoch": 4.142820270154005,
      "grad_norm": 13.0625,
      "learning_rate": 7.08410003961496e-05,
      "loss": 1.6783,
      "step": 48152
    },
    {
      "epoch": 4.1429063064613265,
      "grad_norm": 0.984375,
      "learning_rate": 7.082712593219354e-05,
      "loss": 0.7956,
      "step": 48153
    },
    {
      "epoch": 4.142992342768649,
      "grad_norm": 1.390625,
      "learning_rate": 7.08132527234847e-05,
      "loss": 1.2271,
      "step": 48154
    },
    {
      "epoch": 4.14307837907597,
      "grad_norm": 1.125,
      "learning_rate": 7.07993807700637e-05,
      "loss": 0.8994,
      "step": 48155
    },
    {
      "epoch": 4.143164415383292,
      "grad_norm": 1.2890625,
      "learning_rate": 7.078551007197098e-05,
      "loss": 0.8122,
      "step": 48156
    },
    {
      "epoch": 4.143250451690613,
      "grad_norm": 1.1953125,
      "learning_rate": 7.077164062924712e-05,
      "loss": 1.0915,
      "step": 48157
    },
    {
      "epoch": 4.143336487997935,
      "grad_norm": 1544.0,
      "learning_rate": 7.075777244193276e-05,
      "loss": 1.1074,
      "step": 48158
    },
    {
      "epoch": 4.143422524305257,
      "grad_norm": 0.86328125,
      "learning_rate": 7.074390551006837e-05,
      "loss": 0.497,
      "step": 48159
    },
    {
      "epoch": 4.143508560612578,
      "grad_norm": 1.3984375,
      "learning_rate": 7.073003983369474e-05,
      "loss": 0.8306,
      "step": 48160
    },
    {
      "epoch": 4.1435945969199,
      "grad_norm": 0.97265625,
      "learning_rate": 7.071617541285214e-05,
      "loss": 0.754,
      "step": 48161
    },
    {
      "epoch": 4.143680633227222,
      "grad_norm": 1.5703125,
      "learning_rate": 7.070231224758123e-05,
      "loss": 0.89,
      "step": 48162
    },
    {
      "epoch": 4.143766669534544,
      "grad_norm": 1.0234375,
      "learning_rate": 7.068845033792264e-05,
      "loss": 0.9549,
      "step": 48163
    },
    {
      "epoch": 4.143852705841866,
      "grad_norm": 1.1953125,
      "learning_rate": 7.067458968391676e-05,
      "loss": 1.2426,
      "step": 48164
    },
    {
      "epoch": 4.143938742149187,
      "grad_norm": 1.2109375,
      "learning_rate": 7.066073028560421e-05,
      "loss": 1.5697,
      "step": 48165
    },
    {
      "epoch": 4.144024778456509,
      "grad_norm": 1.375,
      "learning_rate": 7.064687214302562e-05,
      "loss": 1.8395,
      "step": 48166
    },
    {
      "epoch": 4.14411081476383,
      "grad_norm": 1.328125,
      "learning_rate": 7.063301525622135e-05,
      "loss": 1.41,
      "step": 48167
    },
    {
      "epoch": 4.144196851071152,
      "grad_norm": 5024.0,
      "learning_rate": 7.061915962523208e-05,
      "loss": 0.5828,
      "step": 48168
    },
    {
      "epoch": 4.144282887378473,
      "grad_norm": 1.3828125,
      "learning_rate": 7.060530525009817e-05,
      "loss": 0.7307,
      "step": 48169
    },
    {
      "epoch": 4.1443689236857955,
      "grad_norm": 0.859375,
      "learning_rate": 7.05914521308602e-05,
      "loss": 0.9536,
      "step": 48170
    },
    {
      "epoch": 4.1444549599931175,
      "grad_norm": 1.375,
      "learning_rate": 7.057760026755883e-05,
      "loss": 2.0822,
      "step": 48171
    },
    {
      "epoch": 4.144540996300439,
      "grad_norm": 1.2109375,
      "learning_rate": 7.056374966023438e-05,
      "loss": 1.5267,
      "step": 48172
    },
    {
      "epoch": 4.144627032607761,
      "grad_norm": 1.09375,
      "learning_rate": 7.054990030892739e-05,
      "loss": 0.858,
      "step": 48173
    },
    {
      "epoch": 4.144713068915082,
      "grad_norm": 1.4765625,
      "learning_rate": 7.053605221367853e-05,
      "loss": 1.6485,
      "step": 48174
    },
    {
      "epoch": 4.144799105222404,
      "grad_norm": 1.3515625,
      "learning_rate": 7.052220537452813e-05,
      "loss": 1.2586,
      "step": 48175
    },
    {
      "epoch": 4.144885141529725,
      "grad_norm": 0.9765625,
      "learning_rate": 7.050835979151677e-05,
      "loss": 0.8257,
      "step": 48176
    },
    {
      "epoch": 4.144971177837047,
      "grad_norm": 1.359375,
      "learning_rate": 7.049451546468494e-05,
      "loss": 1.0119,
      "step": 48177
    },
    {
      "epoch": 4.145057214144369,
      "grad_norm": 0.9375,
      "learning_rate": 7.048067239407308e-05,
      "loss": 0.5297,
      "step": 48178
    },
    {
      "epoch": 4.1451432504516905,
      "grad_norm": 0.73046875,
      "learning_rate": 7.046683057972181e-05,
      "loss": 0.5882,
      "step": 48179
    },
    {
      "epoch": 4.145229286759013,
      "grad_norm": 1.09375,
      "learning_rate": 7.045299002167144e-05,
      "loss": 1.2764,
      "step": 48180
    },
    {
      "epoch": 4.145315323066334,
      "grad_norm": 0.921875,
      "learning_rate": 7.043915071996249e-05,
      "loss": 1.2102,
      "step": 48181
    },
    {
      "epoch": 4.145401359373656,
      "grad_norm": 1.015625,
      "learning_rate": 7.042531267463559e-05,
      "loss": 1.0917,
      "step": 48182
    },
    {
      "epoch": 4.145487395680977,
      "grad_norm": 1.4453125,
      "learning_rate": 7.0411475885731e-05,
      "loss": 1.3131,
      "step": 48183
    },
    {
      "epoch": 4.145573431988299,
      "grad_norm": 0.79296875,
      "learning_rate": 7.039764035328933e-05,
      "loss": 0.9282,
      "step": 48184
    },
    {
      "epoch": 4.145659468295621,
      "grad_norm": 1.078125,
      "learning_rate": 7.038380607735106e-05,
      "loss": 1.9318,
      "step": 48185
    },
    {
      "epoch": 4.145745504602942,
      "grad_norm": 1.3515625,
      "learning_rate": 7.03699730579565e-05,
      "loss": 1.161,
      "step": 48186
    },
    {
      "epoch": 4.145831540910264,
      "grad_norm": 1.203125,
      "learning_rate": 7.035614129514623e-05,
      "loss": 0.9964,
      "step": 48187
    },
    {
      "epoch": 4.145917577217586,
      "grad_norm": 1.3046875,
      "learning_rate": 7.034231078896081e-05,
      "loss": 0.6998,
      "step": 48188
    },
    {
      "epoch": 4.146003613524908,
      "grad_norm": 1.0234375,
      "learning_rate": 7.032848153944044e-05,
      "loss": 1.1523,
      "step": 48189
    },
    {
      "epoch": 4.146089649832229,
      "grad_norm": 1.2109375,
      "learning_rate": 7.031465354662569e-05,
      "loss": 0.4137,
      "step": 48190
    },
    {
      "epoch": 4.146175686139551,
      "grad_norm": 1.3671875,
      "learning_rate": 7.030082681055715e-05,
      "loss": 0.7622,
      "step": 48191
    },
    {
      "epoch": 4.146261722446873,
      "grad_norm": 1.1640625,
      "learning_rate": 7.028700133127497e-05,
      "loss": 1.1277,
      "step": 48192
    },
    {
      "epoch": 4.146347758754194,
      "grad_norm": 1.078125,
      "learning_rate": 7.027317710881986e-05,
      "loss": 1.6444,
      "step": 48193
    },
    {
      "epoch": 4.146433795061516,
      "grad_norm": 2.734375,
      "learning_rate": 7.025935414323192e-05,
      "loss": 1.0248,
      "step": 48194
    },
    {
      "epoch": 4.146519831368837,
      "grad_norm": 1.53125,
      "learning_rate": 7.02455324345519e-05,
      "loss": 1.1098,
      "step": 48195
    },
    {
      "epoch": 4.1466058676761595,
      "grad_norm": 1.21875,
      "learning_rate": 7.023171198282019e-05,
      "loss": 1.7612,
      "step": 48196
    },
    {
      "epoch": 4.146691903983481,
      "grad_norm": 1.125,
      "learning_rate": 7.021789278807705e-05,
      "loss": 1.3309,
      "step": 48197
    },
    {
      "epoch": 4.146777940290803,
      "grad_norm": 1.2421875,
      "learning_rate": 7.020407485036312e-05,
      "loss": 0.5837,
      "step": 48198
    },
    {
      "epoch": 4.146863976598125,
      "grad_norm": 0.80078125,
      "learning_rate": 7.019025816971852e-05,
      "loss": 1.2383,
      "step": 48199
    },
    {
      "epoch": 4.146950012905446,
      "grad_norm": 1.078125,
      "learning_rate": 7.017644274618385e-05,
      "loss": 1.2776,
      "step": 48200
    },
    {
      "epoch": 4.147036049212768,
      "grad_norm": 1.671875,
      "learning_rate": 7.01626285797996e-05,
      "loss": 1.3508,
      "step": 48201
    },
    {
      "epoch": 4.147122085520089,
      "grad_norm": 1.2578125,
      "learning_rate": 7.014881567060599e-05,
      "loss": 1.4382,
      "step": 48202
    },
    {
      "epoch": 4.147208121827411,
      "grad_norm": 1.09375,
      "learning_rate": 7.013500401864343e-05,
      "loss": 0.7445,
      "step": 48203
    },
    {
      "epoch": 4.1472941581347325,
      "grad_norm": 1.359375,
      "learning_rate": 7.012119362395253e-05,
      "loss": 1.3023,
      "step": 48204
    },
    {
      "epoch": 4.1473801944420545,
      "grad_norm": 1.0078125,
      "learning_rate": 7.010738448657338e-05,
      "loss": 0.7987,
      "step": 48205
    },
    {
      "epoch": 4.147466230749377,
      "grad_norm": 596.0,
      "learning_rate": 7.009357660654659e-05,
      "loss": 0.8754,
      "step": 48206
    },
    {
      "epoch": 4.147552267056698,
      "grad_norm": 15.75,
      "learning_rate": 7.007976998391252e-05,
      "loss": 1.7157,
      "step": 48207
    },
    {
      "epoch": 4.14763830336402,
      "grad_norm": 12.125,
      "learning_rate": 7.006596461871139e-05,
      "loss": 1.074,
      "step": 48208
    },
    {
      "epoch": 4.147724339671341,
      "grad_norm": 0.96484375,
      "learning_rate": 7.005216051098384e-05,
      "loss": 1.2413,
      "step": 48209
    },
    {
      "epoch": 4.147810375978663,
      "grad_norm": 1.0546875,
      "learning_rate": 7.003835766076999e-05,
      "loss": 0.8658,
      "step": 48210
    },
    {
      "epoch": 4.147896412285984,
      "grad_norm": 1.109375,
      "learning_rate": 7.00245560681103e-05,
      "loss": 0.7141,
      "step": 48211
    },
    {
      "epoch": 4.147982448593306,
      "grad_norm": 2.34375,
      "learning_rate": 7.001075573304533e-05,
      "loss": 1.0282,
      "step": 48212
    },
    {
      "epoch": 4.148068484900628,
      "grad_norm": 6.1875,
      "learning_rate": 6.999695665561512e-05,
      "loss": 1.7469,
      "step": 48213
    },
    {
      "epoch": 4.14815452120795,
      "grad_norm": 1.0546875,
      "learning_rate": 6.998315883586015e-05,
      "loss": 1.7443,
      "step": 48214
    },
    {
      "epoch": 4.148240557515272,
      "grad_norm": 1.09375,
      "learning_rate": 6.996936227382095e-05,
      "loss": 1.0742,
      "step": 48215
    },
    {
      "epoch": 4.148326593822593,
      "grad_norm": 34.0,
      "learning_rate": 6.995556696953765e-05,
      "loss": 1.0472,
      "step": 48216
    },
    {
      "epoch": 4.148412630129915,
      "grad_norm": 1.6484375,
      "learning_rate": 6.99417729230506e-05,
      "loss": 1.6196,
      "step": 48217
    },
    {
      "epoch": 4.148498666437236,
      "grad_norm": 4.8125,
      "learning_rate": 6.992798013440038e-05,
      "loss": 1.2986,
      "step": 48218
    },
    {
      "epoch": 4.148584702744558,
      "grad_norm": 1.671875,
      "learning_rate": 6.991418860362708e-05,
      "loss": 1.5391,
      "step": 48219
    },
    {
      "epoch": 4.14867073905188,
      "grad_norm": 1.2109375,
      "learning_rate": 6.990039833077111e-05,
      "loss": 1.2146,
      "step": 48220
    },
    {
      "epoch": 4.148756775359201,
      "grad_norm": 0.828125,
      "learning_rate": 6.988660931587293e-05,
      "loss": 0.6767,
      "step": 48221
    },
    {
      "epoch": 4.1488428116665235,
      "grad_norm": 1.046875,
      "learning_rate": 6.987282155897262e-05,
      "loss": 1.6481,
      "step": 48222
    },
    {
      "epoch": 4.148928847973845,
      "grad_norm": 1.078125,
      "learning_rate": 6.985903506011082e-05,
      "loss": 1.3949,
      "step": 48223
    },
    {
      "epoch": 4.149014884281167,
      "grad_norm": 1.28125,
      "learning_rate": 6.984524981932754e-05,
      "loss": 1.0724,
      "step": 48224
    },
    {
      "epoch": 4.149100920588488,
      "grad_norm": 1.109375,
      "learning_rate": 6.983146583666327e-05,
      "loss": 1.3329,
      "step": 48225
    },
    {
      "epoch": 4.14918695689581,
      "grad_norm": 0.8984375,
      "learning_rate": 6.981768311215842e-05,
      "loss": 1.1335,
      "step": 48226
    },
    {
      "epoch": 4.149272993203132,
      "grad_norm": 0.9453125,
      "learning_rate": 6.980390164585304e-05,
      "loss": 0.4911,
      "step": 48227
    },
    {
      "epoch": 4.149359029510453,
      "grad_norm": 1.71875,
      "learning_rate": 6.979012143778757e-05,
      "loss": 0.9196,
      "step": 48228
    },
    {
      "epoch": 4.149445065817775,
      "grad_norm": 1.3125,
      "learning_rate": 6.977634248800247e-05,
      "loss": 1.8091,
      "step": 48229
    },
    {
      "epoch": 4.1495311021250965,
      "grad_norm": 0.94140625,
      "learning_rate": 6.976256479653776e-05,
      "loss": 0.7512,
      "step": 48230
    },
    {
      "epoch": 4.1496171384324185,
      "grad_norm": 3104.0,
      "learning_rate": 6.97487883634339e-05,
      "loss": 1.2018,
      "step": 48231
    },
    {
      "epoch": 4.149703174739741,
      "grad_norm": 1.390625,
      "learning_rate": 6.97350131887311e-05,
      "loss": 0.6983,
      "step": 48232
    },
    {
      "epoch": 4.149789211047062,
      "grad_norm": 29.625,
      "learning_rate": 6.972123927246976e-05,
      "loss": 1.2701,
      "step": 48233
    },
    {
      "epoch": 4.149875247354384,
      "grad_norm": 1.1953125,
      "learning_rate": 6.970746661469018e-05,
      "loss": 1.0777,
      "step": 48234
    },
    {
      "epoch": 4.149961283661705,
      "grad_norm": 1.3046875,
      "learning_rate": 6.96936952154325e-05,
      "loss": 0.8173,
      "step": 48235
    },
    {
      "epoch": 4.150047319969027,
      "grad_norm": 3.40625,
      "learning_rate": 6.967992507473702e-05,
      "loss": 0.6793,
      "step": 48236
    },
    {
      "epoch": 4.150133356276348,
      "grad_norm": 1.4296875,
      "learning_rate": 6.966615619264422e-05,
      "loss": 1.0434,
      "step": 48237
    },
    {
      "epoch": 4.15021939258367,
      "grad_norm": 0.91015625,
      "learning_rate": 6.965238856919409e-05,
      "loss": 1.2617,
      "step": 48238
    },
    {
      "epoch": 4.150305428890992,
      "grad_norm": 1.3671875,
      "learning_rate": 6.963862220442718e-05,
      "loss": 0.8832,
      "step": 48239
    },
    {
      "epoch": 4.150391465198314,
      "grad_norm": 1.265625,
      "learning_rate": 6.962485709838346e-05,
      "loss": 1.391,
      "step": 48240
    },
    {
      "epoch": 4.150477501505636,
      "grad_norm": 0.796875,
      "learning_rate": 6.961109325110331e-05,
      "loss": 0.9801,
      "step": 48241
    },
    {
      "epoch": 4.150563537812957,
      "grad_norm": 260096.0,
      "learning_rate": 6.959733066262714e-05,
      "loss": 1.4028,
      "step": 48242
    },
    {
      "epoch": 4.150649574120279,
      "grad_norm": 1.2734375,
      "learning_rate": 6.958356933299492e-05,
      "loss": 0.9494,
      "step": 48243
    },
    {
      "epoch": 4.1507356104276,
      "grad_norm": 9.25,
      "learning_rate": 6.95698092622471e-05,
      "loss": 1.7331,
      "step": 48244
    },
    {
      "epoch": 4.150821646734922,
      "grad_norm": 1.390625,
      "learning_rate": 6.955605045042396e-05,
      "loss": 0.7164,
      "step": 48245
    },
    {
      "epoch": 4.150907683042244,
      "grad_norm": 1.109375,
      "learning_rate": 6.954229289756553e-05,
      "loss": 0.919,
      "step": 48246
    },
    {
      "epoch": 4.150993719349565,
      "grad_norm": 1.125,
      "learning_rate": 6.952853660371216e-05,
      "loss": 0.8708,
      "step": 48247
    },
    {
      "epoch": 4.1510797556568875,
      "grad_norm": 2.546875,
      "learning_rate": 6.951478156890423e-05,
      "loss": 0.9641,
      "step": 48248
    },
    {
      "epoch": 4.151165791964209,
      "grad_norm": 1.21875,
      "learning_rate": 6.950102779318173e-05,
      "loss": 0.9418,
      "step": 48249
    },
    {
      "epoch": 4.151251828271531,
      "grad_norm": 0.8203125,
      "learning_rate": 6.948727527658516e-05,
      "loss": 0.8776,
      "step": 48250
    },
    {
      "epoch": 4.151337864578852,
      "grad_norm": 1.0,
      "learning_rate": 6.947352401915441e-05,
      "loss": 0.9666,
      "step": 48251
    },
    {
      "epoch": 4.151423900886174,
      "grad_norm": 1.046875,
      "learning_rate": 6.94597740209299e-05,
      "loss": 0.7202,
      "step": 48252
    },
    {
      "epoch": 4.151509937193496,
      "grad_norm": 0.765625,
      "learning_rate": 6.944602528195188e-05,
      "loss": 0.9359,
      "step": 48253
    },
    {
      "epoch": 4.151595973500817,
      "grad_norm": 1.03125,
      "learning_rate": 6.943227780226041e-05,
      "loss": 0.7453,
      "step": 48254
    },
    {
      "epoch": 4.151682009808139,
      "grad_norm": 0.9765625,
      "learning_rate": 6.941853158189582e-05,
      "loss": 1.2984,
      "step": 48255
    },
    {
      "epoch": 4.1517680461154605,
      "grad_norm": 1.1015625,
      "learning_rate": 6.940478662089833e-05,
      "loss": 1.5258,
      "step": 48256
    },
    {
      "epoch": 4.1518540824227825,
      "grad_norm": 1.390625,
      "learning_rate": 6.939104291930803e-05,
      "loss": 0.9562,
      "step": 48257
    },
    {
      "epoch": 4.151940118730104,
      "grad_norm": 1.0625,
      "learning_rate": 6.937730047716517e-05,
      "loss": 0.6831,
      "step": 48258
    },
    {
      "epoch": 4.152026155037426,
      "grad_norm": 1.2109375,
      "learning_rate": 6.936355929451004e-05,
      "loss": 1.1948,
      "step": 48259
    },
    {
      "epoch": 4.152112191344748,
      "grad_norm": 2.734375,
      "learning_rate": 6.934981937138268e-05,
      "loss": 1.187,
      "step": 48260
    },
    {
      "epoch": 4.152198227652069,
      "grad_norm": 1.0625,
      "learning_rate": 6.933608070782332e-05,
      "loss": 1.3901,
      "step": 48261
    },
    {
      "epoch": 4.152284263959391,
      "grad_norm": 1.3046875,
      "learning_rate": 6.932234330387227e-05,
      "loss": 1.0903,
      "step": 48262
    },
    {
      "epoch": 4.152370300266712,
      "grad_norm": 1.3046875,
      "learning_rate": 6.93086071595695e-05,
      "loss": 1.3665,
      "step": 48263
    },
    {
      "epoch": 4.152456336574034,
      "grad_norm": 1.1875,
      "learning_rate": 6.92948722749554e-05,
      "loss": 1.3785,
      "step": 48264
    },
    {
      "epoch": 4.1525423728813555,
      "grad_norm": 35.75,
      "learning_rate": 6.92811386500699e-05,
      "loss": 0.9216,
      "step": 48265
    },
    {
      "epoch": 4.152628409188678,
      "grad_norm": 1.140625,
      "learning_rate": 6.926740628495331e-05,
      "loss": 1.3459,
      "step": 48266
    },
    {
      "epoch": 4.152714445496,
      "grad_norm": 1.7578125,
      "learning_rate": 6.925367517964593e-05,
      "loss": 1.3632,
      "step": 48267
    },
    {
      "epoch": 4.152800481803321,
      "grad_norm": 1.2421875,
      "learning_rate": 6.923994533418748e-05,
      "loss": 1.0455,
      "step": 48268
    },
    {
      "epoch": 4.152886518110643,
      "grad_norm": 0.84765625,
      "learning_rate": 6.922621674861868e-05,
      "loss": 0.9384,
      "step": 48269
    },
    {
      "epoch": 4.152972554417964,
      "grad_norm": 1.171875,
      "learning_rate": 6.92124894229793e-05,
      "loss": 1.1721,
      "step": 48270
    },
    {
      "epoch": 4.153058590725286,
      "grad_norm": 1.4296875,
      "learning_rate": 6.919876335730962e-05,
      "loss": 0.9156,
      "step": 48271
    },
    {
      "epoch": 4.153144627032607,
      "grad_norm": 1.3828125,
      "learning_rate": 6.918503855164986e-05,
      "loss": 1.1757,
      "step": 48272
    },
    {
      "epoch": 4.153230663339929,
      "grad_norm": 0.9296875,
      "learning_rate": 6.917131500603996e-05,
      "loss": 1.5843,
      "step": 48273
    },
    {
      "epoch": 4.1533166996472515,
      "grad_norm": 1.234375,
      "learning_rate": 6.91575927205202e-05,
      "loss": 1.4891,
      "step": 48274
    },
    {
      "epoch": 4.153402735954573,
      "grad_norm": 1.8671875,
      "learning_rate": 6.914387169513082e-05,
      "loss": 0.9802,
      "step": 48275
    },
    {
      "epoch": 4.153488772261895,
      "grad_norm": 1.2265625,
      "learning_rate": 6.913015192991163e-05,
      "loss": 1.8161,
      "step": 48276
    },
    {
      "epoch": 4.153574808569216,
      "grad_norm": 0.97265625,
      "learning_rate": 6.911643342490298e-05,
      "loss": 1.1192,
      "step": 48277
    },
    {
      "epoch": 4.153660844876538,
      "grad_norm": 148.0,
      "learning_rate": 6.910271618014507e-05,
      "loss": 1.0268,
      "step": 48278
    },
    {
      "epoch": 4.15374688118386,
      "grad_norm": 1.28125,
      "learning_rate": 6.908900019567783e-05,
      "loss": 1.4919,
      "step": 48279
    },
    {
      "epoch": 4.153832917491181,
      "grad_norm": 1.171875,
      "learning_rate": 6.907528547154152e-05,
      "loss": 1.4318,
      "step": 48280
    },
    {
      "epoch": 4.153918953798503,
      "grad_norm": 1.1796875,
      "learning_rate": 6.906157200777613e-05,
      "loss": 1.3161,
      "step": 48281
    },
    {
      "epoch": 4.1540049901058245,
      "grad_norm": 1.1171875,
      "learning_rate": 6.90478598044218e-05,
      "loss": 1.2891,
      "step": 48282
    },
    {
      "epoch": 4.1540910264131465,
      "grad_norm": 1.0078125,
      "learning_rate": 6.903414886151876e-05,
      "loss": 0.7145,
      "step": 48283
    },
    {
      "epoch": 4.154177062720468,
      "grad_norm": 1.78125,
      "learning_rate": 6.902043917910694e-05,
      "loss": 1.352,
      "step": 48284
    },
    {
      "epoch": 4.15426309902779,
      "grad_norm": 0.8984375,
      "learning_rate": 6.900673075722647e-05,
      "loss": 0.8024,
      "step": 48285
    },
    {
      "epoch": 4.154349135335112,
      "grad_norm": 0.7890625,
      "learning_rate": 6.899302359591764e-05,
      "loss": 0.8102,
      "step": 48286
    },
    {
      "epoch": 4.154435171642433,
      "grad_norm": 1.2265625,
      "learning_rate": 6.897931769522026e-05,
      "loss": 0.4886,
      "step": 48287
    },
    {
      "epoch": 4.154521207949755,
      "grad_norm": 0.8203125,
      "learning_rate": 6.896561305517451e-05,
      "loss": 0.6699,
      "step": 48288
    },
    {
      "epoch": 4.154607244257076,
      "grad_norm": 1.09375,
      "learning_rate": 6.895190967582071e-05,
      "loss": 1.3097,
      "step": 48289
    },
    {
      "epoch": 4.154693280564398,
      "grad_norm": 1.3359375,
      "learning_rate": 6.893820755719854e-05,
      "loss": 1.8472,
      "step": 48290
    },
    {
      "epoch": 4.1547793168717195,
      "grad_norm": 0.91015625,
      "learning_rate": 6.892450669934846e-05,
      "loss": 0.5171,
      "step": 48291
    },
    {
      "epoch": 4.154865353179042,
      "grad_norm": 140.0,
      "learning_rate": 6.891080710231019e-05,
      "loss": 0.8852,
      "step": 48292
    },
    {
      "epoch": 4.154951389486364,
      "grad_norm": 0.8125,
      "learning_rate": 6.889710876612398e-05,
      "loss": 0.8248,
      "step": 48293
    },
    {
      "epoch": 4.155037425793685,
      "grad_norm": 3.984375,
      "learning_rate": 6.888341169083001e-05,
      "loss": 0.9853,
      "step": 48294
    },
    {
      "epoch": 4.155123462101007,
      "grad_norm": 1.125,
      "learning_rate": 6.886971587646812e-05,
      "loss": 1.248,
      "step": 48295
    },
    {
      "epoch": 4.155209498408328,
      "grad_norm": 1.5,
      "learning_rate": 6.885602132307844e-05,
      "loss": 0.8717,
      "step": 48296
    },
    {
      "epoch": 4.15529553471565,
      "grad_norm": 94.5,
      "learning_rate": 6.884232803070117e-05,
      "loss": 0.6974,
      "step": 48297
    },
    {
      "epoch": 4.155381571022971,
      "grad_norm": 0.9453125,
      "learning_rate": 6.882863599937611e-05,
      "loss": 1.3432,
      "step": 48298
    },
    {
      "epoch": 4.155467607330293,
      "grad_norm": 1.0625,
      "learning_rate": 6.881494522914339e-05,
      "loss": 0.8767,
      "step": 48299
    },
    {
      "epoch": 4.1555536436376155,
      "grad_norm": 2.5,
      "learning_rate": 6.880125572004325e-05,
      "loss": 1.0369,
      "step": 48300
    },
    {
      "epoch": 4.155639679944937,
      "grad_norm": 1.2265625,
      "learning_rate": 6.878756747211546e-05,
      "loss": 0.7379,
      "step": 48301
    },
    {
      "epoch": 4.155725716252259,
      "grad_norm": 1.1484375,
      "learning_rate": 6.877388048540018e-05,
      "loss": 1.275,
      "step": 48302
    },
    {
      "epoch": 4.15581175255958,
      "grad_norm": 1.7578125,
      "learning_rate": 6.87601947599375e-05,
      "loss": 1.3443,
      "step": 48303
    },
    {
      "epoch": 4.155897788866902,
      "grad_norm": 1.109375,
      "learning_rate": 6.874651029576723e-05,
      "loss": 0.5803,
      "step": 48304
    },
    {
      "epoch": 4.155983825174223,
      "grad_norm": 1.03125,
      "learning_rate": 6.873282709292971e-05,
      "loss": 1.1317,
      "step": 48305
    },
    {
      "epoch": 4.156069861481545,
      "grad_norm": 1.2421875,
      "learning_rate": 6.871914515146466e-05,
      "loss": 1.2355,
      "step": 48306
    },
    {
      "epoch": 4.156155897788867,
      "grad_norm": 0.73046875,
      "learning_rate": 6.870546447141229e-05,
      "loss": 0.5203,
      "step": 48307
    },
    {
      "epoch": 4.1562419340961885,
      "grad_norm": 2752.0,
      "learning_rate": 6.869178505281264e-05,
      "loss": 0.8727,
      "step": 48308
    },
    {
      "epoch": 4.1563279704035105,
      "grad_norm": 225.0,
      "learning_rate": 6.867810689570553e-05,
      "loss": 0.7588,
      "step": 48309
    },
    {
      "epoch": 4.156414006710832,
      "grad_norm": 1.0234375,
      "learning_rate": 6.866443000013117e-05,
      "loss": 0.7747,
      "step": 48310
    },
    {
      "epoch": 4.156500043018154,
      "grad_norm": 0.75390625,
      "learning_rate": 6.865075436612932e-05,
      "loss": 0.5523,
      "step": 48311
    },
    {
      "epoch": 4.156586079325475,
      "grad_norm": 0.91796875,
      "learning_rate": 6.863707999374019e-05,
      "loss": 1.1929,
      "step": 48312
    },
    {
      "epoch": 4.156672115632797,
      "grad_norm": 0.91015625,
      "learning_rate": 6.862340688300378e-05,
      "loss": 0.9374,
      "step": 48313
    },
    {
      "epoch": 4.156758151940119,
      "grad_norm": 1.2578125,
      "learning_rate": 6.860973503395989e-05,
      "loss": 1.6744,
      "step": 48314
    },
    {
      "epoch": 4.15684418824744,
      "grad_norm": 1.3203125,
      "learning_rate": 6.859606444664863e-05,
      "loss": 1.0905,
      "step": 48315
    },
    {
      "epoch": 4.156930224554762,
      "grad_norm": 1.4453125,
      "learning_rate": 6.858239512111008e-05,
      "loss": 0.8245,
      "step": 48316
    },
    {
      "epoch": 4.1570162608620835,
      "grad_norm": 1.375,
      "learning_rate": 6.856872705738404e-05,
      "loss": 0.7697,
      "step": 48317
    },
    {
      "epoch": 4.157102297169406,
      "grad_norm": 1.1171875,
      "learning_rate": 6.855506025551057e-05,
      "loss": 0.8858,
      "step": 48318
    },
    {
      "epoch": 4.157188333476727,
      "grad_norm": 19.125,
      "learning_rate": 6.854139471552966e-05,
      "loss": 0.8409,
      "step": 48319
    },
    {
      "epoch": 4.157274369784049,
      "grad_norm": 1.078125,
      "learning_rate": 6.852773043748123e-05,
      "loss": 0.8249,
      "step": 48320
    },
    {
      "epoch": 4.157360406091371,
      "grad_norm": 1.21875,
      "learning_rate": 6.851406742140537e-05,
      "loss": 0.9571,
      "step": 48321
    },
    {
      "epoch": 4.157446442398692,
      "grad_norm": 0.765625,
      "learning_rate": 6.850040566734184e-05,
      "loss": 0.6196,
      "step": 48322
    },
    {
      "epoch": 4.157532478706014,
      "grad_norm": 0.8828125,
      "learning_rate": 6.848674517533066e-05,
      "loss": 0.6859,
      "step": 48323
    },
    {
      "epoch": 4.157618515013335,
      "grad_norm": 1.6171875,
      "learning_rate": 6.847308594541197e-05,
      "loss": 1.2902,
      "step": 48324
    },
    {
      "epoch": 4.157704551320657,
      "grad_norm": 1.0625,
      "learning_rate": 6.845942797762544e-05,
      "loss": 1.3543,
      "step": 48325
    },
    {
      "epoch": 4.157790587627979,
      "grad_norm": 1.1640625,
      "learning_rate": 6.844577127201118e-05,
      "loss": 0.7672,
      "step": 48326
    },
    {
      "epoch": 4.157876623935301,
      "grad_norm": 1.203125,
      "learning_rate": 6.84321158286092e-05,
      "loss": 0.9077,
      "step": 48327
    },
    {
      "epoch": 4.157962660242623,
      "grad_norm": 9.875,
      "learning_rate": 6.841846164745924e-05,
      "loss": 1.4451,
      "step": 48328
    },
    {
      "epoch": 4.158048696549944,
      "grad_norm": 1.0,
      "learning_rate": 6.840480872860133e-05,
      "loss": 0.8656,
      "step": 48329
    },
    {
      "epoch": 4.158134732857266,
      "grad_norm": 1.140625,
      "learning_rate": 6.839115707207555e-05,
      "loss": 0.6751,
      "step": 48330
    },
    {
      "epoch": 4.158220769164587,
      "grad_norm": 0.98828125,
      "learning_rate": 6.837750667792158e-05,
      "loss": 0.7334,
      "step": 48331
    },
    {
      "epoch": 4.158306805471909,
      "grad_norm": 1.1015625,
      "learning_rate": 6.836385754617952e-05,
      "loss": 0.9623,
      "step": 48332
    },
    {
      "epoch": 4.15839284177923,
      "grad_norm": 1.328125,
      "learning_rate": 6.835020967688915e-05,
      "loss": 1.3015,
      "step": 48333
    },
    {
      "epoch": 4.1584788780865525,
      "grad_norm": 1.015625,
      "learning_rate": 6.83365630700905e-05,
      "loss": 1.2421,
      "step": 48334
    },
    {
      "epoch": 4.1585649143938745,
      "grad_norm": 13.25,
      "learning_rate": 6.832291772582349e-05,
      "loss": 0.8897,
      "step": 48335
    },
    {
      "epoch": 4.158650950701196,
      "grad_norm": 1.0625,
      "learning_rate": 6.830927364412792e-05,
      "loss": 0.6583,
      "step": 48336
    },
    {
      "epoch": 4.158736987008518,
      "grad_norm": 1.4453125,
      "learning_rate": 6.829563082504381e-05,
      "loss": 1.4395,
      "step": 48337
    },
    {
      "epoch": 4.158823023315839,
      "grad_norm": 1.015625,
      "learning_rate": 6.828198926861107e-05,
      "loss": 1.6398,
      "step": 48338
    },
    {
      "epoch": 4.158909059623161,
      "grad_norm": 2864.0,
      "learning_rate": 6.826834897486945e-05,
      "loss": 0.9778,
      "step": 48339
    },
    {
      "epoch": 4.158995095930482,
      "grad_norm": 1.1875,
      "learning_rate": 6.825470994385891e-05,
      "loss": 1.2826,
      "step": 48340
    },
    {
      "epoch": 4.159081132237804,
      "grad_norm": 3.5625,
      "learning_rate": 6.824107217561942e-05,
      "loss": 1.0401,
      "step": 48341
    },
    {
      "epoch": 4.159167168545126,
      "grad_norm": 1.09375,
      "learning_rate": 6.82274356701908e-05,
      "loss": 0.9314,
      "step": 48342
    },
    {
      "epoch": 4.1592532048524475,
      "grad_norm": 2.984375,
      "learning_rate": 6.821380042761304e-05,
      "loss": 1.0392,
      "step": 48343
    },
    {
      "epoch": 4.15933924115977,
      "grad_norm": 1.15625,
      "learning_rate": 6.820016644792587e-05,
      "loss": 1.2314,
      "step": 48344
    },
    {
      "epoch": 4.159425277467091,
      "grad_norm": 1.1171875,
      "learning_rate": 6.818653373116923e-05,
      "loss": 1.383,
      "step": 48345
    },
    {
      "epoch": 4.159511313774413,
      "grad_norm": 1.078125,
      "learning_rate": 6.81729022773831e-05,
      "loss": 1.3006,
      "step": 48346
    },
    {
      "epoch": 4.159597350081734,
      "grad_norm": 1.0,
      "learning_rate": 6.81592720866071e-05,
      "loss": 0.9009,
      "step": 48347
    },
    {
      "epoch": 4.159683386389056,
      "grad_norm": 1.078125,
      "learning_rate": 6.814564315888127e-05,
      "loss": 0.9929,
      "step": 48348
    },
    {
      "epoch": 4.159769422696378,
      "grad_norm": 1.359375,
      "learning_rate": 6.813201549424558e-05,
      "loss": 1.7831,
      "step": 48349
    },
    {
      "epoch": 4.159855459003699,
      "grad_norm": 1.3515625,
      "learning_rate": 6.81183890927396e-05,
      "loss": 1.5767,
      "step": 48350
    },
    {
      "epoch": 4.159941495311021,
      "grad_norm": 0.92578125,
      "learning_rate": 6.810476395440346e-05,
      "loss": 1.0804,
      "step": 48351
    },
    {
      "epoch": 4.160027531618343,
      "grad_norm": 0.8984375,
      "learning_rate": 6.80911400792768e-05,
      "loss": 1.1251,
      "step": 48352
    },
    {
      "epoch": 4.160113567925665,
      "grad_norm": 1.171875,
      "learning_rate": 6.807751746739954e-05,
      "loss": 1.4029,
      "step": 48353
    },
    {
      "epoch": 4.160199604232987,
      "grad_norm": 4.875,
      "learning_rate": 6.806389611881164e-05,
      "loss": 0.7285,
      "step": 48354
    },
    {
      "epoch": 4.160285640540308,
      "grad_norm": 1.2265625,
      "learning_rate": 6.805027603355268e-05,
      "loss": 0.9169,
      "step": 48355
    },
    {
      "epoch": 4.16037167684763,
      "grad_norm": 0.9609375,
      "learning_rate": 6.803665721166275e-05,
      "loss": 0.9535,
      "step": 48356
    },
    {
      "epoch": 4.160457713154951,
      "grad_norm": 1.0546875,
      "learning_rate": 6.802303965318163e-05,
      "loss": 1.5987,
      "step": 48357
    },
    {
      "epoch": 4.160543749462273,
      "grad_norm": 96.0,
      "learning_rate": 6.800942335814897e-05,
      "loss": 0.982,
      "step": 48358
    },
    {
      "epoch": 4.160629785769594,
      "grad_norm": 0.96484375,
      "learning_rate": 6.799580832660479e-05,
      "loss": 0.9452,
      "step": 48359
    },
    {
      "epoch": 4.1607158220769165,
      "grad_norm": 1.1875,
      "learning_rate": 6.798219455858895e-05,
      "loss": 1.8583,
      "step": 48360
    },
    {
      "epoch": 4.1608018583842385,
      "grad_norm": 1.3125,
      "learning_rate": 6.796858205414102e-05,
      "loss": 1.3106,
      "step": 48361
    },
    {
      "epoch": 4.16088789469156,
      "grad_norm": 1.359375,
      "learning_rate": 6.79549708133011e-05,
      "loss": 1.2227,
      "step": 48362
    },
    {
      "epoch": 4.160973930998882,
      "grad_norm": 0.94921875,
      "learning_rate": 6.794136083610875e-05,
      "loss": 0.7266,
      "step": 48363
    },
    {
      "epoch": 4.161059967306203,
      "grad_norm": 1.84375,
      "learning_rate": 6.792775212260388e-05,
      "loss": 1.2084,
      "step": 48364
    },
    {
      "epoch": 4.161146003613525,
      "grad_norm": 1.015625,
      "learning_rate": 6.791414467282641e-05,
      "loss": 1.5214,
      "step": 48365
    },
    {
      "epoch": 4.161232039920846,
      "grad_norm": 1.1796875,
      "learning_rate": 6.790053848681594e-05,
      "loss": 1.2023,
      "step": 48366
    },
    {
      "epoch": 4.161318076228168,
      "grad_norm": 0.953125,
      "learning_rate": 6.788693356461234e-05,
      "loss": 0.371,
      "step": 48367
    },
    {
      "epoch": 4.16140411253549,
      "grad_norm": 0.91796875,
      "learning_rate": 6.787332990625556e-05,
      "loss": 1.2356,
      "step": 48368
    },
    {
      "epoch": 4.1614901488428115,
      "grad_norm": 1.359375,
      "learning_rate": 6.78597275117851e-05,
      "loss": 1.4925,
      "step": 48369
    },
    {
      "epoch": 4.161576185150134,
      "grad_norm": 1.0234375,
      "learning_rate": 6.784612638124088e-05,
      "loss": 0.7354,
      "step": 48370
    },
    {
      "epoch": 4.161662221457455,
      "grad_norm": 1.0625,
      "learning_rate": 6.783252651466281e-05,
      "loss": 0.7838,
      "step": 48371
    },
    {
      "epoch": 4.161748257764777,
      "grad_norm": 0.94921875,
      "learning_rate": 6.781892791209049e-05,
      "loss": 1.3145,
      "step": 48372
    },
    {
      "epoch": 4.161834294072098,
      "grad_norm": 1.0234375,
      "learning_rate": 6.78053305735637e-05,
      "loss": 0.7243,
      "step": 48373
    },
    {
      "epoch": 4.16192033037942,
      "grad_norm": 1.03125,
      "learning_rate": 6.77917344991224e-05,
      "loss": 0.6809,
      "step": 48374
    },
    {
      "epoch": 4.162006366686742,
      "grad_norm": 0.87890625,
      "learning_rate": 6.777813968880608e-05,
      "loss": 0.8113,
      "step": 48375
    },
    {
      "epoch": 4.162092402994063,
      "grad_norm": 1.015625,
      "learning_rate": 6.776454614265477e-05,
      "loss": 0.5624,
      "step": 48376
    },
    {
      "epoch": 4.162178439301385,
      "grad_norm": 1.3828125,
      "learning_rate": 6.775095386070796e-05,
      "loss": 1.0409,
      "step": 48377
    },
    {
      "epoch": 4.162264475608707,
      "grad_norm": 1.1484375,
      "learning_rate": 6.773736284300558e-05,
      "loss": 0.9997,
      "step": 48378
    },
    {
      "epoch": 4.162350511916029,
      "grad_norm": 1.46875,
      "learning_rate": 6.772377308958728e-05,
      "loss": 1.485,
      "step": 48379
    },
    {
      "epoch": 4.16243654822335,
      "grad_norm": 1.03125,
      "learning_rate": 6.771018460049295e-05,
      "loss": 0.6906,
      "step": 48380
    },
    {
      "epoch": 4.162522584530672,
      "grad_norm": 1.078125,
      "learning_rate": 6.769659737576228e-05,
      "loss": 1.6027,
      "step": 48381
    },
    {
      "epoch": 4.162608620837994,
      "grad_norm": 1.3203125,
      "learning_rate": 6.768301141543493e-05,
      "loss": 0.9403,
      "step": 48382
    },
    {
      "epoch": 4.162694657145315,
      "grad_norm": 11.9375,
      "learning_rate": 6.766942671955068e-05,
      "loss": 0.9205,
      "step": 48383
    },
    {
      "epoch": 4.162780693452637,
      "grad_norm": 0.99609375,
      "learning_rate": 6.765584328814938e-05,
      "loss": 1.3456,
      "step": 48384
    },
    {
      "epoch": 4.162866729759958,
      "grad_norm": 1.3203125,
      "learning_rate": 6.764226112127059e-05,
      "loss": 1.0232,
      "step": 48385
    },
    {
      "epoch": 4.1629527660672805,
      "grad_norm": 0.9296875,
      "learning_rate": 6.762868021895402e-05,
      "loss": 1.0386,
      "step": 48386
    },
    {
      "epoch": 4.163038802374602,
      "grad_norm": 6.0625,
      "learning_rate": 6.76151005812396e-05,
      "loss": 1.2064,
      "step": 48387
    },
    {
      "epoch": 4.163124838681924,
      "grad_norm": 9.5625,
      "learning_rate": 6.760152220816684e-05,
      "loss": 1.291,
      "step": 48388
    },
    {
      "epoch": 4.163210874989246,
      "grad_norm": 1.0546875,
      "learning_rate": 6.75879450997755e-05,
      "loss": 0.6085,
      "step": 48389
    },
    {
      "epoch": 4.163296911296567,
      "grad_norm": 1.1953125,
      "learning_rate": 6.757436925610544e-05,
      "loss": 0.999,
      "step": 48390
    },
    {
      "epoch": 4.163382947603889,
      "grad_norm": 1.359375,
      "learning_rate": 6.756079467719611e-05,
      "loss": 1.2621,
      "step": 48391
    },
    {
      "epoch": 4.16346898391121,
      "grad_norm": 0.8515625,
      "learning_rate": 6.75472213630875e-05,
      "loss": 0.3068,
      "step": 48392
    },
    {
      "epoch": 4.163555020218532,
      "grad_norm": 1.3203125,
      "learning_rate": 6.753364931381905e-05,
      "loss": 1.7445,
      "step": 48393
    },
    {
      "epoch": 4.163641056525854,
      "grad_norm": 0.93359375,
      "learning_rate": 6.752007852943053e-05,
      "loss": 0.607,
      "step": 48394
    },
    {
      "epoch": 4.1637270928331755,
      "grad_norm": 1.1953125,
      "learning_rate": 6.750650900996185e-05,
      "loss": 0.8498,
      "step": 48395
    },
    {
      "epoch": 4.163813129140498,
      "grad_norm": 0.98828125,
      "learning_rate": 6.749294075545231e-05,
      "loss": 0.6209,
      "step": 48396
    },
    {
      "epoch": 4.163899165447819,
      "grad_norm": 28.625,
      "learning_rate": 6.747937376594187e-05,
      "loss": 1.2126,
      "step": 48397
    },
    {
      "epoch": 4.163985201755141,
      "grad_norm": 1.2578125,
      "learning_rate": 6.746580804147023e-05,
      "loss": 0.9638,
      "step": 48398
    },
    {
      "epoch": 4.164071238062462,
      "grad_norm": 1.1640625,
      "learning_rate": 6.745224358207685e-05,
      "loss": 0.7847,
      "step": 48399
    },
    {
      "epoch": 4.164157274369784,
      "grad_norm": 1.453125,
      "learning_rate": 6.74386803878016e-05,
      "loss": 1.0445,
      "step": 48400
    },
    {
      "epoch": 4.164243310677106,
      "grad_norm": 1.3125,
      "learning_rate": 6.74251184586841e-05,
      "loss": 0.9725,
      "step": 48401
    },
    {
      "epoch": 4.164329346984427,
      "grad_norm": 1.109375,
      "learning_rate": 6.741155779476393e-05,
      "loss": 0.9044,
      "step": 48402
    },
    {
      "epoch": 4.164415383291749,
      "grad_norm": 0.73828125,
      "learning_rate": 6.739799839608096e-05,
      "loss": 0.814,
      "step": 48403
    },
    {
      "epoch": 4.164501419599071,
      "grad_norm": 0.96484375,
      "learning_rate": 6.738444026267453e-05,
      "loss": 0.5821,
      "step": 48404
    },
    {
      "epoch": 4.164587455906393,
      "grad_norm": 1.1171875,
      "learning_rate": 6.737088339458452e-05,
      "loss": 0.909,
      "step": 48405
    },
    {
      "epoch": 4.164673492213714,
      "grad_norm": 0.99609375,
      "learning_rate": 6.735732779185067e-05,
      "loss": 0.8781,
      "step": 48406
    },
    {
      "epoch": 4.164759528521036,
      "grad_norm": 1.015625,
      "learning_rate": 6.734377345451232e-05,
      "loss": 1.3145,
      "step": 48407
    },
    {
      "epoch": 4.164845564828358,
      "grad_norm": 46.25,
      "learning_rate": 6.733022038260933e-05,
      "loss": 0.8458,
      "step": 48408
    },
    {
      "epoch": 4.164931601135679,
      "grad_norm": 1744.0,
      "learning_rate": 6.731666857618135e-05,
      "loss": 1.289,
      "step": 48409
    },
    {
      "epoch": 4.165017637443001,
      "grad_norm": 0.91015625,
      "learning_rate": 6.73031180352679e-05,
      "loss": 1.1174,
      "step": 48410
    },
    {
      "epoch": 4.165103673750322,
      "grad_norm": 1.328125,
      "learning_rate": 6.728956875990866e-05,
      "loss": 1.5318,
      "step": 48411
    },
    {
      "epoch": 4.1651897100576445,
      "grad_norm": 1.7734375,
      "learning_rate": 6.727602075014338e-05,
      "loss": 1.257,
      "step": 48412
    },
    {
      "epoch": 4.165275746364966,
      "grad_norm": 1.0,
      "learning_rate": 6.726247400601149e-05,
      "loss": 0.7699,
      "step": 48413
    },
    {
      "epoch": 4.165361782672288,
      "grad_norm": 1.21875,
      "learning_rate": 6.724892852755272e-05,
      "loss": 1.5878,
      "step": 48414
    },
    {
      "epoch": 4.16544781897961,
      "grad_norm": 1.921875,
      "learning_rate": 6.723538431480658e-05,
      "loss": 1.3238,
      "step": 48415
    },
    {
      "epoch": 4.165533855286931,
      "grad_norm": 2.203125,
      "learning_rate": 6.722184136781284e-05,
      "loss": 1.4939,
      "step": 48416
    },
    {
      "epoch": 4.165619891594253,
      "grad_norm": 1.1171875,
      "learning_rate": 6.720829968661112e-05,
      "loss": 1.3035,
      "step": 48417
    },
    {
      "epoch": 4.165705927901574,
      "grad_norm": 1.28125,
      "learning_rate": 6.719475927124091e-05,
      "loss": 1.9365,
      "step": 48418
    },
    {
      "epoch": 4.165791964208896,
      "grad_norm": 1.4609375,
      "learning_rate": 6.718122012174189e-05,
      "loss": 1.3627,
      "step": 48419
    },
    {
      "epoch": 4.1658780005162175,
      "grad_norm": 1.1484375,
      "learning_rate": 6.716768223815356e-05,
      "loss": 1.4542,
      "step": 48420
    },
    {
      "epoch": 4.1659640368235396,
      "grad_norm": 1.1171875,
      "learning_rate": 6.715414562051552e-05,
      "loss": 1.0668,
      "step": 48421
    },
    {
      "epoch": 4.166050073130862,
      "grad_norm": 1.390625,
      "learning_rate": 6.714061026886758e-05,
      "loss": 1.5155,
      "step": 48422
    },
    {
      "epoch": 4.166136109438183,
      "grad_norm": 1.109375,
      "learning_rate": 6.712707618324904e-05,
      "loss": 1.3118,
      "step": 48423
    },
    {
      "epoch": 4.166222145745505,
      "grad_norm": 110592.0,
      "learning_rate": 6.711354336369962e-05,
      "loss": 1.2604,
      "step": 48424
    },
    {
      "epoch": 4.166308182052826,
      "grad_norm": 180.0,
      "learning_rate": 6.710001181025905e-05,
      "loss": 1.7,
      "step": 48425
    },
    {
      "epoch": 4.166394218360148,
      "grad_norm": 1.265625,
      "learning_rate": 6.708648152296653e-05,
      "loss": 1.1812,
      "step": 48426
    },
    {
      "epoch": 4.166480254667469,
      "grad_norm": 0.86328125,
      "learning_rate": 6.707295250186197e-05,
      "loss": 0.5253,
      "step": 48427
    },
    {
      "epoch": 4.166566290974791,
      "grad_norm": 0.94140625,
      "learning_rate": 6.705942474698484e-05,
      "loss": 0.5974,
      "step": 48428
    },
    {
      "epoch": 4.166652327282113,
      "grad_norm": 1.046875,
      "learning_rate": 6.704589825837464e-05,
      "loss": 1.2479,
      "step": 48429
    },
    {
      "epoch": 4.166738363589435,
      "grad_norm": 9.125,
      "learning_rate": 6.703237303607096e-05,
      "loss": 1.8185,
      "step": 48430
    },
    {
      "epoch": 4.166824399896757,
      "grad_norm": 0.828125,
      "learning_rate": 6.701884908011352e-05,
      "loss": 0.8037,
      "step": 48431
    },
    {
      "epoch": 4.166910436204078,
      "grad_norm": 402.0,
      "learning_rate": 6.700532639054157e-05,
      "loss": 1.4147,
      "step": 48432
    },
    {
      "epoch": 4.1669964725114,
      "grad_norm": 0.92578125,
      "learning_rate": 6.699180496739498e-05,
      "loss": 0.642,
      "step": 48433
    },
    {
      "epoch": 4.167082508818721,
      "grad_norm": 1.2265625,
      "learning_rate": 6.697828481071305e-05,
      "loss": 1.6988,
      "step": 48434
    },
    {
      "epoch": 4.167168545126043,
      "grad_norm": 1.34375,
      "learning_rate": 6.696476592053535e-05,
      "loss": 1.1016,
      "step": 48435
    },
    {
      "epoch": 4.167254581433365,
      "grad_norm": 2.140625,
      "learning_rate": 6.695124829690169e-05,
      "loss": 1.4331,
      "step": 48436
    },
    {
      "epoch": 4.167340617740686,
      "grad_norm": 0.9453125,
      "learning_rate": 6.69377319398512e-05,
      "loss": 1.3804,
      "step": 48437
    },
    {
      "epoch": 4.1674266540480085,
      "grad_norm": 1.203125,
      "learning_rate": 6.692421684942368e-05,
      "loss": 1.3363,
      "step": 48438
    },
    {
      "epoch": 4.16751269035533,
      "grad_norm": 20.0,
      "learning_rate": 6.691070302565872e-05,
      "loss": 1.4514,
      "step": 48439
    },
    {
      "epoch": 4.167598726662652,
      "grad_norm": 1.7578125,
      "learning_rate": 6.689719046859555e-05,
      "loss": 2.0085,
      "step": 48440
    },
    {
      "epoch": 4.167684762969973,
      "grad_norm": 1.3359375,
      "learning_rate": 6.688367917827393e-05,
      "loss": 2.0518,
      "step": 48441
    },
    {
      "epoch": 4.167770799277295,
      "grad_norm": 0.8828125,
      "learning_rate": 6.687016915473337e-05,
      "loss": 0.8762,
      "step": 48442
    },
    {
      "epoch": 4.167856835584617,
      "grad_norm": 0.94140625,
      "learning_rate": 6.685666039801325e-05,
      "loss": 0.7912,
      "step": 48443
    },
    {
      "epoch": 4.167942871891938,
      "grad_norm": 1.0546875,
      "learning_rate": 6.684315290815323e-05,
      "loss": 1.2319,
      "step": 48444
    },
    {
      "epoch": 4.16802890819926,
      "grad_norm": 1.2578125,
      "learning_rate": 6.682964668519271e-05,
      "loss": 1.0789,
      "step": 48445
    },
    {
      "epoch": 4.1681149445065815,
      "grad_norm": 1.2109375,
      "learning_rate": 6.681614172917117e-05,
      "loss": 0.6648,
      "step": 48446
    },
    {
      "epoch": 4.1682009808139036,
      "grad_norm": 1.0546875,
      "learning_rate": 6.680263804012826e-05,
      "loss": 1.2454,
      "step": 48447
    },
    {
      "epoch": 4.168287017121225,
      "grad_norm": 1.078125,
      "learning_rate": 6.67891356181033e-05,
      "loss": 0.9842,
      "step": 48448
    },
    {
      "epoch": 4.168373053428547,
      "grad_norm": 1.09375,
      "learning_rate": 6.677563446313583e-05,
      "loss": 0.9971,
      "step": 48449
    },
    {
      "epoch": 4.168459089735869,
      "grad_norm": 1.0078125,
      "learning_rate": 6.676213457526553e-05,
      "loss": 0.7601,
      "step": 48450
    },
    {
      "epoch": 4.16854512604319,
      "grad_norm": 1.109375,
      "learning_rate": 6.674863595453157e-05,
      "loss": 0.8174,
      "step": 48451
    },
    {
      "epoch": 4.168631162350512,
      "grad_norm": 1.515625,
      "learning_rate": 6.673513860097364e-05,
      "loss": 1.196,
      "step": 48452
    },
    {
      "epoch": 4.168717198657833,
      "grad_norm": 1.109375,
      "learning_rate": 6.672164251463114e-05,
      "loss": 0.8959,
      "step": 48453
    },
    {
      "epoch": 4.168803234965155,
      "grad_norm": 0.95703125,
      "learning_rate": 6.670814769554357e-05,
      "loss": 0.8779,
      "step": 48454
    },
    {
      "epoch": 4.1688892712724765,
      "grad_norm": 1.3515625,
      "learning_rate": 6.669465414375047e-05,
      "loss": 1.3924,
      "step": 48455
    },
    {
      "epoch": 4.168975307579799,
      "grad_norm": 1.0625,
      "learning_rate": 6.668116185929113e-05,
      "loss": 1.2385,
      "step": 48456
    },
    {
      "epoch": 4.169061343887121,
      "grad_norm": 0.91015625,
      "learning_rate": 6.666767084220515e-05,
      "loss": 1.4948,
      "step": 48457
    },
    {
      "epoch": 4.169147380194442,
      "grad_norm": 1.1328125,
      "learning_rate": 6.665418109253207e-05,
      "loss": 0.7794,
      "step": 48458
    },
    {
      "epoch": 4.169233416501764,
      "grad_norm": 1.125,
      "learning_rate": 6.664069261031108e-05,
      "loss": 1.3019,
      "step": 48459
    },
    {
      "epoch": 4.169319452809085,
      "grad_norm": 1.359375,
      "learning_rate": 6.662720539558182e-05,
      "loss": 1.6383,
      "step": 48460
    },
    {
      "epoch": 4.169405489116407,
      "grad_norm": 1.2890625,
      "learning_rate": 6.66137194483838e-05,
      "loss": 1.8383,
      "step": 48461
    },
    {
      "epoch": 4.169491525423728,
      "grad_norm": 0.99609375,
      "learning_rate": 6.660023476875621e-05,
      "loss": 1.1657,
      "step": 48462
    },
    {
      "epoch": 4.16957756173105,
      "grad_norm": 0.9921875,
      "learning_rate": 6.65867513567388e-05,
      "loss": 0.6572,
      "step": 48463
    },
    {
      "epoch": 4.1696635980383725,
      "grad_norm": 0.828125,
      "learning_rate": 6.657326921237073e-05,
      "loss": 0.8021,
      "step": 48464
    },
    {
      "epoch": 4.169749634345694,
      "grad_norm": 23.75,
      "learning_rate": 6.655978833569154e-05,
      "loss": 0.8028,
      "step": 48465
    },
    {
      "epoch": 4.169835670653016,
      "grad_norm": 13.375,
      "learning_rate": 6.654630872674078e-05,
      "loss": 1.5084,
      "step": 48466
    },
    {
      "epoch": 4.169921706960337,
      "grad_norm": 0.8203125,
      "learning_rate": 6.653283038555768e-05,
      "loss": 0.9125,
      "step": 48467
    },
    {
      "epoch": 4.170007743267659,
      "grad_norm": 1.59375,
      "learning_rate": 6.651935331218173e-05,
      "loss": 1.4586,
      "step": 48468
    },
    {
      "epoch": 4.170093779574981,
      "grad_norm": 0.96875,
      "learning_rate": 6.650587750665244e-05,
      "loss": 0.6997,
      "step": 48469
    },
    {
      "epoch": 4.170179815882302,
      "grad_norm": 1184.0,
      "learning_rate": 6.649240296900905e-05,
      "loss": 1.0631,
      "step": 48470
    },
    {
      "epoch": 4.170265852189624,
      "grad_norm": 0.890625,
      "learning_rate": 6.647892969929109e-05,
      "loss": 0.7636,
      "step": 48471
    },
    {
      "epoch": 4.1703518884969455,
      "grad_norm": 13.75,
      "learning_rate": 6.64654576975381e-05,
      "loss": 0.9779,
      "step": 48472
    },
    {
      "epoch": 4.170437924804268,
      "grad_norm": 0.99609375,
      "learning_rate": 6.645198696378913e-05,
      "loss": 0.8482,
      "step": 48473
    },
    {
      "epoch": 4.170523961111589,
      "grad_norm": 5.40625,
      "learning_rate": 6.643851749808394e-05,
      "loss": 0.9239,
      "step": 48474
    },
    {
      "epoch": 4.170609997418911,
      "grad_norm": 0.8515625,
      "learning_rate": 6.642504930046161e-05,
      "loss": 0.4911,
      "step": 48475
    },
    {
      "epoch": 4.170696033726233,
      "grad_norm": 0.8984375,
      "learning_rate": 6.641158237096173e-05,
      "loss": 0.7925,
      "step": 48476
    },
    {
      "epoch": 4.170782070033554,
      "grad_norm": 24.625,
      "learning_rate": 6.639811670962376e-05,
      "loss": 0.8785,
      "step": 48477
    },
    {
      "epoch": 4.170868106340876,
      "grad_norm": 1.0390625,
      "learning_rate": 6.638465231648683e-05,
      "loss": 1.3562,
      "step": 48478
    },
    {
      "epoch": 4.170954142648197,
      "grad_norm": 1.2890625,
      "learning_rate": 6.637118919159046e-05,
      "loss": 1.3105,
      "step": 48479
    },
    {
      "epoch": 4.171040178955519,
      "grad_norm": 1.046875,
      "learning_rate": 6.635772733497414e-05,
      "loss": 1.3583,
      "step": 48480
    },
    {
      "epoch": 4.1711262152628406,
      "grad_norm": 0.9375,
      "learning_rate": 6.634426674667703e-05,
      "loss": 1.1627,
      "step": 48481
    },
    {
      "epoch": 4.171212251570163,
      "grad_norm": 0.8203125,
      "learning_rate": 6.633080742673858e-05,
      "loss": 1.3594,
      "step": 48482
    },
    {
      "epoch": 4.171298287877485,
      "grad_norm": 0.765625,
      "learning_rate": 6.631734937519828e-05,
      "loss": 0.3674,
      "step": 48483
    },
    {
      "epoch": 4.171384324184806,
      "grad_norm": 0.921875,
      "learning_rate": 6.63038925920953e-05,
      "loss": 1.0782,
      "step": 48484
    },
    {
      "epoch": 4.171470360492128,
      "grad_norm": 1.453125,
      "learning_rate": 6.629043707746918e-05,
      "loss": 1.6851,
      "step": 48485
    },
    {
      "epoch": 4.171556396799449,
      "grad_norm": 510.0,
      "learning_rate": 6.627698283135908e-05,
      "loss": 1.1473,
      "step": 48486
    },
    {
      "epoch": 4.171642433106771,
      "grad_norm": 1.09375,
      "learning_rate": 6.626352985380447e-05,
      "loss": 0.9524,
      "step": 48487
    },
    {
      "epoch": 4.171728469414092,
      "grad_norm": 7.125,
      "learning_rate": 6.625007814484473e-05,
      "loss": 0.5956,
      "step": 48488
    },
    {
      "epoch": 4.171814505721414,
      "grad_norm": 1.421875,
      "learning_rate": 6.623662770451893e-05,
      "loss": 1.7797,
      "step": 48489
    },
    {
      "epoch": 4.1719005420287365,
      "grad_norm": 236.0,
      "learning_rate": 6.622317853286691e-05,
      "loss": 0.6397,
      "step": 48490
    },
    {
      "epoch": 4.171986578336058,
      "grad_norm": 1.265625,
      "learning_rate": 6.620973062992757e-05,
      "loss": 1.2093,
      "step": 48491
    },
    {
      "epoch": 4.17207261464338,
      "grad_norm": 1.2265625,
      "learning_rate": 6.619628399574046e-05,
      "loss": 1.6515,
      "step": 48492
    },
    {
      "epoch": 4.172158650950701,
      "grad_norm": 1.0859375,
      "learning_rate": 6.618283863034491e-05,
      "loss": 1.2161,
      "step": 48493
    },
    {
      "epoch": 4.172244687258023,
      "grad_norm": 59.25,
      "learning_rate": 6.616939453378007e-05,
      "loss": 1.0492,
      "step": 48494
    },
    {
      "epoch": 4.172330723565344,
      "grad_norm": 1.234375,
      "learning_rate": 6.61559517060854e-05,
      "loss": 1.4047,
      "step": 48495
    },
    {
      "epoch": 4.172416759872666,
      "grad_norm": 0.9921875,
      "learning_rate": 6.614251014730033e-05,
      "loss": 0.9782,
      "step": 48496
    },
    {
      "epoch": 4.172502796179988,
      "grad_norm": 1.46875,
      "learning_rate": 6.612906985746387e-05,
      "loss": 1.3047,
      "step": 48497
    },
    {
      "epoch": 4.1725888324873095,
      "grad_norm": 1.640625,
      "learning_rate": 6.611563083661554e-05,
      "loss": 0.9757,
      "step": 48498
    },
    {
      "epoch": 4.172674868794632,
      "grad_norm": 8.9375,
      "learning_rate": 6.610219308479465e-05,
      "loss": 0.8092,
      "step": 48499
    },
    {
      "epoch": 4.172760905101953,
      "grad_norm": 1.1328125,
      "learning_rate": 6.608875660204039e-05,
      "loss": 1.1766,
      "step": 48500
    },
    {
      "epoch": 4.172846941409275,
      "grad_norm": 0.69140625,
      "learning_rate": 6.607532138839217e-05,
      "loss": 0.782,
      "step": 48501
    },
    {
      "epoch": 4.172932977716596,
      "grad_norm": 1.1953125,
      "learning_rate": 6.60618874438893e-05,
      "loss": 1.3946,
      "step": 48502
    },
    {
      "epoch": 4.173019014023918,
      "grad_norm": 1.2734375,
      "learning_rate": 6.604845476857091e-05,
      "loss": 1.6645,
      "step": 48503
    },
    {
      "epoch": 4.17310505033124,
      "grad_norm": 1.125,
      "learning_rate": 6.603502336247646e-05,
      "loss": 0.5055,
      "step": 48504
    },
    {
      "epoch": 4.173191086638561,
      "grad_norm": 1.0390625,
      "learning_rate": 6.60215932256451e-05,
      "loss": 0.7903,
      "step": 48505
    },
    {
      "epoch": 4.173277122945883,
      "grad_norm": 0.890625,
      "learning_rate": 6.600816435811618e-05,
      "loss": 0.5359,
      "step": 48506
    },
    {
      "epoch": 4.173363159253205,
      "grad_norm": 2.0,
      "learning_rate": 6.599473675992913e-05,
      "loss": 1.2151,
      "step": 48507
    },
    {
      "epoch": 4.173449195560527,
      "grad_norm": 1.203125,
      "learning_rate": 6.598131043112288e-05,
      "loss": 0.8433,
      "step": 48508
    },
    {
      "epoch": 4.173535231867848,
      "grad_norm": 1.046875,
      "learning_rate": 6.596788537173686e-05,
      "loss": 1.8008,
      "step": 48509
    },
    {
      "epoch": 4.17362126817517,
      "grad_norm": 56.5,
      "learning_rate": 6.595446158181056e-05,
      "loss": 0.962,
      "step": 48510
    },
    {
      "epoch": 4.173707304482492,
      "grad_norm": 1.1640625,
      "learning_rate": 6.594103906138288e-05,
      "loss": 1.0353,
      "step": 48511
    },
    {
      "epoch": 4.173793340789813,
      "grad_norm": 1.0859375,
      "learning_rate": 6.592761781049323e-05,
      "loss": 1.1698,
      "step": 48512
    },
    {
      "epoch": 4.173879377097135,
      "grad_norm": 1.1171875,
      "learning_rate": 6.591419782918101e-05,
      "loss": 0.6551,
      "step": 48513
    },
    {
      "epoch": 4.173965413404456,
      "grad_norm": 1.2734375,
      "learning_rate": 6.590077911748527e-05,
      "loss": 1.3853,
      "step": 48514
    },
    {
      "epoch": 4.174051449711778,
      "grad_norm": 20.375,
      "learning_rate": 6.588736167544534e-05,
      "loss": 0.9866,
      "step": 48515
    },
    {
      "epoch": 4.1741374860191005,
      "grad_norm": 10.1875,
      "learning_rate": 6.587394550310039e-05,
      "loss": 1.4371,
      "step": 48516
    },
    {
      "epoch": 4.174223522326422,
      "grad_norm": 3.15625,
      "learning_rate": 6.586053060048969e-05,
      "loss": 0.7198,
      "step": 48517
    },
    {
      "epoch": 4.174309558633744,
      "grad_norm": 1.0859375,
      "learning_rate": 6.584711696765267e-05,
      "loss": 1.7798,
      "step": 48518
    },
    {
      "epoch": 4.174395594941065,
      "grad_norm": 1.796875,
      "learning_rate": 6.583370460462823e-05,
      "loss": 1.0815,
      "step": 48519
    },
    {
      "epoch": 4.174481631248387,
      "grad_norm": 0.8515625,
      "learning_rate": 6.58202935114558e-05,
      "loss": 0.8903,
      "step": 48520
    },
    {
      "epoch": 4.174567667555708,
      "grad_norm": 2.171875,
      "learning_rate": 6.580688368817466e-05,
      "loss": 1.4626,
      "step": 48521
    },
    {
      "epoch": 4.17465370386303,
      "grad_norm": 0.97265625,
      "learning_rate": 6.57934751348238e-05,
      "loss": 1.3114,
      "step": 48522
    },
    {
      "epoch": 4.174739740170352,
      "grad_norm": 1.3984375,
      "learning_rate": 6.578006785144264e-05,
      "loss": 1.6704,
      "step": 48523
    },
    {
      "epoch": 4.1748257764776735,
      "grad_norm": 0.81640625,
      "learning_rate": 6.57666618380704e-05,
      "loss": 0.9069,
      "step": 48524
    },
    {
      "epoch": 4.174911812784996,
      "grad_norm": 39.5,
      "learning_rate": 6.57532570947461e-05,
      "loss": 1.6668,
      "step": 48525
    },
    {
      "epoch": 4.174997849092317,
      "grad_norm": 0.82421875,
      "learning_rate": 6.573985362150908e-05,
      "loss": 1.5613,
      "step": 48526
    },
    {
      "epoch": 4.175083885399639,
      "grad_norm": 1.046875,
      "learning_rate": 6.57264514183985e-05,
      "loss": 0.9748,
      "step": 48527
    },
    {
      "epoch": 4.17516992170696,
      "grad_norm": 1.4453125,
      "learning_rate": 6.571305048545361e-05,
      "loss": 0.9741,
      "step": 48528
    },
    {
      "epoch": 4.175255958014282,
      "grad_norm": 0.95703125,
      "learning_rate": 6.569965082271372e-05,
      "loss": 0.7974,
      "step": 48529
    },
    {
      "epoch": 4.175341994321604,
      "grad_norm": 1.171875,
      "learning_rate": 6.568625243021775e-05,
      "loss": 1.7625,
      "step": 48530
    },
    {
      "epoch": 4.175428030628925,
      "grad_norm": 2.015625,
      "learning_rate": 6.567285530800516e-05,
      "loss": 1.0473,
      "step": 48531
    },
    {
      "epoch": 4.175514066936247,
      "grad_norm": 0.83984375,
      "learning_rate": 6.565945945611484e-05,
      "loss": 0.9051,
      "step": 48532
    },
    {
      "epoch": 4.175600103243569,
      "grad_norm": 0.9375,
      "learning_rate": 6.564606487458619e-05,
      "loss": 0.8987,
      "step": 48533
    },
    {
      "epoch": 4.175686139550891,
      "grad_norm": 1.03125,
      "learning_rate": 6.563267156345836e-05,
      "loss": 1.0974,
      "step": 48534
    },
    {
      "epoch": 4.175772175858212,
      "grad_norm": 1.046875,
      "learning_rate": 6.561927952277042e-05,
      "loss": 1.0137,
      "step": 48535
    },
    {
      "epoch": 4.175858212165534,
      "grad_norm": 1.40625,
      "learning_rate": 6.560588875256162e-05,
      "loss": 1.1864,
      "step": 48536
    },
    {
      "epoch": 4.175944248472856,
      "grad_norm": 1.078125,
      "learning_rate": 6.55924992528712e-05,
      "loss": 0.8162,
      "step": 48537
    },
    {
      "epoch": 4.176030284780177,
      "grad_norm": 0.82421875,
      "learning_rate": 6.557911102373809e-05,
      "loss": 0.9652,
      "step": 48538
    },
    {
      "epoch": 4.176116321087499,
      "grad_norm": 164.0,
      "learning_rate": 6.556572406520162e-05,
      "loss": 1.4994,
      "step": 48539
    },
    {
      "epoch": 4.17620235739482,
      "grad_norm": 0.953125,
      "learning_rate": 6.555233837730102e-05,
      "loss": 0.6222,
      "step": 48540
    },
    {
      "epoch": 4.176288393702142,
      "grad_norm": 2.15625,
      "learning_rate": 6.553895396007525e-05,
      "loss": 0.7887,
      "step": 48541
    },
    {
      "epoch": 4.176374430009464,
      "grad_norm": 374.0,
      "learning_rate": 6.552557081356352e-05,
      "loss": 1.2608,
      "step": 48542
    },
    {
      "epoch": 4.176460466316786,
      "grad_norm": 0.91796875,
      "learning_rate": 6.551218893780509e-05,
      "loss": 1.2184,
      "step": 48543
    },
    {
      "epoch": 4.176546502624108,
      "grad_norm": 1.1328125,
      "learning_rate": 6.549880833283889e-05,
      "loss": 0.9709,
      "step": 48544
    },
    {
      "epoch": 4.176632538931429,
      "grad_norm": 0.93359375,
      "learning_rate": 6.54854289987043e-05,
      "loss": 0.7971,
      "step": 48545
    },
    {
      "epoch": 4.176718575238751,
      "grad_norm": 24.0,
      "learning_rate": 6.547205093544023e-05,
      "loss": 0.688,
      "step": 48546
    },
    {
      "epoch": 4.176804611546072,
      "grad_norm": 1.28125,
      "learning_rate": 6.545867414308587e-05,
      "loss": 1.2397,
      "step": 48547
    },
    {
      "epoch": 4.176890647853394,
      "grad_norm": 1.3515625,
      "learning_rate": 6.544529862168053e-05,
      "loss": 1.7225,
      "step": 48548
    },
    {
      "epoch": 4.176976684160715,
      "grad_norm": 1.0546875,
      "learning_rate": 6.5431924371263e-05,
      "loss": 1.0558,
      "step": 48549
    },
    {
      "epoch": 4.1770627204680375,
      "grad_norm": 1.46875,
      "learning_rate": 6.541855139187258e-05,
      "loss": 1.0159,
      "step": 48550
    },
    {
      "epoch": 4.17714875677536,
      "grad_norm": 1.0390625,
      "learning_rate": 6.540517968354853e-05,
      "loss": 1.4206,
      "step": 48551
    },
    {
      "epoch": 4.177234793082681,
      "grad_norm": 1.1484375,
      "learning_rate": 6.539180924632965e-05,
      "loss": 1.2448,
      "step": 48552
    },
    {
      "epoch": 4.177320829390003,
      "grad_norm": 1.1328125,
      "learning_rate": 6.537844008025523e-05,
      "loss": 1.1675,
      "step": 48553
    },
    {
      "epoch": 4.177406865697324,
      "grad_norm": 1.203125,
      "learning_rate": 6.536507218536447e-05,
      "loss": 1.3628,
      "step": 48554
    },
    {
      "epoch": 4.177492902004646,
      "grad_norm": 1.046875,
      "learning_rate": 6.535170556169623e-05,
      "loss": 0.8663,
      "step": 48555
    },
    {
      "epoch": 4.177578938311967,
      "grad_norm": 1.453125,
      "learning_rate": 6.533834020928981e-05,
      "loss": 1.5549,
      "step": 48556
    },
    {
      "epoch": 4.177664974619289,
      "grad_norm": 16.5,
      "learning_rate": 6.532497612818406e-05,
      "loss": 0.8507,
      "step": 48557
    },
    {
      "epoch": 4.177751010926611,
      "grad_norm": 1.0625,
      "learning_rate": 6.531161331841828e-05,
      "loss": 0.5456,
      "step": 48558
    },
    {
      "epoch": 4.177837047233933,
      "grad_norm": 0.94140625,
      "learning_rate": 6.529825178003157e-05,
      "loss": 1.6011,
      "step": 48559
    },
    {
      "epoch": 4.177923083541255,
      "grad_norm": 1.1015625,
      "learning_rate": 6.528489151306278e-05,
      "loss": 0.7295,
      "step": 48560
    },
    {
      "epoch": 4.178009119848576,
      "grad_norm": 0.953125,
      "learning_rate": 6.527153251755119e-05,
      "loss": 1.0025,
      "step": 48561
    },
    {
      "epoch": 4.178095156155898,
      "grad_norm": 1.1171875,
      "learning_rate": 6.52581747935359e-05,
      "loss": 1.735,
      "step": 48562
    },
    {
      "epoch": 4.178181192463219,
      "grad_norm": 0.9296875,
      "learning_rate": 6.524481834105567e-05,
      "loss": 0.6069,
      "step": 48563
    },
    {
      "epoch": 4.178267228770541,
      "grad_norm": 1.2890625,
      "learning_rate": 6.523146316015006e-05,
      "loss": 1.1266,
      "step": 48564
    },
    {
      "epoch": 4.178353265077863,
      "grad_norm": 3.46875,
      "learning_rate": 6.52181092508577e-05,
      "loss": 1.016,
      "step": 48565
    },
    {
      "epoch": 4.178439301385184,
      "grad_norm": 0.9453125,
      "learning_rate": 6.520475661321785e-05,
      "loss": 0.7394,
      "step": 48566
    },
    {
      "epoch": 4.1785253376925064,
      "grad_norm": 1.046875,
      "learning_rate": 6.519140524726963e-05,
      "loss": 1.3973,
      "step": 48567
    },
    {
      "epoch": 4.178611373999828,
      "grad_norm": 1.3046875,
      "learning_rate": 6.517805515305181e-05,
      "loss": 1.8585,
      "step": 48568
    },
    {
      "epoch": 4.17869741030715,
      "grad_norm": 1.0625,
      "learning_rate": 6.516470633060367e-05,
      "loss": 0.8336,
      "step": 48569
    },
    {
      "epoch": 4.178783446614471,
      "grad_norm": 1.4453125,
      "learning_rate": 6.515135877996432e-05,
      "loss": 0.7496,
      "step": 48570
    },
    {
      "epoch": 4.178869482921793,
      "grad_norm": 1.1640625,
      "learning_rate": 6.513801250117251e-05,
      "loss": 1.0872,
      "step": 48571
    },
    {
      "epoch": 4.178955519229115,
      "grad_norm": 1.3515625,
      "learning_rate": 6.512466749426748e-05,
      "loss": 1.6113,
      "step": 48572
    },
    {
      "epoch": 4.179041555536436,
      "grad_norm": 8.75,
      "learning_rate": 6.51113237592883e-05,
      "loss": 1.1096,
      "step": 48573
    },
    {
      "epoch": 4.179127591843758,
      "grad_norm": 1.1328125,
      "learning_rate": 6.509798129627382e-05,
      "loss": 1.6989,
      "step": 48574
    },
    {
      "epoch": 4.179213628151079,
      "grad_norm": 1.359375,
      "learning_rate": 6.508464010526327e-05,
      "loss": 1.1546,
      "step": 48575
    },
    {
      "epoch": 4.1792996644584015,
      "grad_norm": 10.0,
      "learning_rate": 6.507130018629542e-05,
      "loss": 1.3235,
      "step": 48576
    },
    {
      "epoch": 4.179385700765723,
      "grad_norm": 1.5234375,
      "learning_rate": 6.505796153940947e-05,
      "loss": 1.6183,
      "step": 48577
    },
    {
      "epoch": 4.179471737073045,
      "grad_norm": 1.1953125,
      "learning_rate": 6.504462416464451e-05,
      "loss": 1.4047,
      "step": 48578
    },
    {
      "epoch": 4.179557773380367,
      "grad_norm": 1.109375,
      "learning_rate": 6.503128806203929e-05,
      "loss": 1.9029,
      "step": 48579
    },
    {
      "epoch": 4.179643809687688,
      "grad_norm": 1.03125,
      "learning_rate": 6.501795323163295e-05,
      "loss": 0.9145,
      "step": 48580
    },
    {
      "epoch": 4.17972984599501,
      "grad_norm": 2.421875,
      "learning_rate": 6.50046196734646e-05,
      "loss": 1.116,
      "step": 48581
    },
    {
      "epoch": 4.179815882302331,
      "grad_norm": 1.046875,
      "learning_rate": 6.499128738757309e-05,
      "loss": 2.2489,
      "step": 48582
    },
    {
      "epoch": 4.179901918609653,
      "grad_norm": 1.1328125,
      "learning_rate": 6.49779563739974e-05,
      "loss": 0.6807,
      "step": 48583
    },
    {
      "epoch": 4.179987954916975,
      "grad_norm": 1.578125,
      "learning_rate": 6.49646266327767e-05,
      "loss": 0.972,
      "step": 48584
    },
    {
      "epoch": 4.180073991224297,
      "grad_norm": 1.5390625,
      "learning_rate": 6.495129816394973e-05,
      "loss": 1.6435,
      "step": 48585
    },
    {
      "epoch": 4.180160027531619,
      "grad_norm": 0.9375,
      "learning_rate": 6.493797096755577e-05,
      "loss": 0.8567,
      "step": 48586
    },
    {
      "epoch": 4.18024606383894,
      "grad_norm": 14.625,
      "learning_rate": 6.492464504363349e-05,
      "loss": 1.7825,
      "step": 48587
    },
    {
      "epoch": 4.180332100146262,
      "grad_norm": 1.5078125,
      "learning_rate": 6.491132039222198e-05,
      "loss": 1.0978,
      "step": 48588
    },
    {
      "epoch": 4.180418136453583,
      "grad_norm": 1.4140625,
      "learning_rate": 6.489799701336035e-05,
      "loss": 1.0698,
      "step": 48589
    },
    {
      "epoch": 4.180504172760905,
      "grad_norm": 7.9375,
      "learning_rate": 6.488467490708738e-05,
      "loss": 1.0465,
      "step": 48590
    },
    {
      "epoch": 4.180590209068227,
      "grad_norm": 0.921875,
      "learning_rate": 6.487135407344214e-05,
      "loss": 0.8797,
      "step": 48591
    },
    {
      "epoch": 4.180676245375548,
      "grad_norm": 1.078125,
      "learning_rate": 6.485803451246359e-05,
      "loss": 1.4258,
      "step": 48592
    },
    {
      "epoch": 4.1807622816828705,
      "grad_norm": 1.2265625,
      "learning_rate": 6.484471622419058e-05,
      "loss": 1.0323,
      "step": 48593
    },
    {
      "epoch": 4.180848317990192,
      "grad_norm": 1.2421875,
      "learning_rate": 6.483139920866216e-05,
      "loss": 0.712,
      "step": 48594
    },
    {
      "epoch": 4.180934354297514,
      "grad_norm": 2.234375,
      "learning_rate": 6.48180834659174e-05,
      "loss": 0.773,
      "step": 48595
    },
    {
      "epoch": 4.181020390604835,
      "grad_norm": 1.21875,
      "learning_rate": 6.48047689959949e-05,
      "loss": 1.1583,
      "step": 48596
    },
    {
      "epoch": 4.181106426912157,
      "grad_norm": 0.82421875,
      "learning_rate": 6.479145579893398e-05,
      "loss": 0.8224,
      "step": 48597
    },
    {
      "epoch": 4.181192463219479,
      "grad_norm": 1.2578125,
      "learning_rate": 6.477814387477332e-05,
      "loss": 0.9218,
      "step": 48598
    },
    {
      "epoch": 4.1812784995268,
      "grad_norm": 0.88671875,
      "learning_rate": 6.476483322355193e-05,
      "loss": 0.9879,
      "step": 48599
    },
    {
      "epoch": 4.181364535834122,
      "grad_norm": 1.3671875,
      "learning_rate": 6.475152384530874e-05,
      "loss": 1.229,
      "step": 48600
    },
    {
      "epoch": 4.1814505721414434,
      "grad_norm": 1.5078125,
      "learning_rate": 6.47382157400827e-05,
      "loss": 2.1327,
      "step": 48601
    },
    {
      "epoch": 4.1815366084487655,
      "grad_norm": 0.8828125,
      "learning_rate": 6.472490890791283e-05,
      "loss": 0.8711,
      "step": 48602
    },
    {
      "epoch": 4.181622644756087,
      "grad_norm": 1.125,
      "learning_rate": 6.47116033488378e-05,
      "loss": 1.4049,
      "step": 48603
    },
    {
      "epoch": 4.181708681063409,
      "grad_norm": 12.3125,
      "learning_rate": 6.46982990628967e-05,
      "loss": 1.3202,
      "step": 48604
    },
    {
      "epoch": 4.181794717370731,
      "grad_norm": 1.046875,
      "learning_rate": 6.468499605012856e-05,
      "loss": 0.4685,
      "step": 48605
    },
    {
      "epoch": 4.181880753678052,
      "grad_norm": 1.0234375,
      "learning_rate": 6.467169431057201e-05,
      "loss": 1.0599,
      "step": 48606
    },
    {
      "epoch": 4.181966789985374,
      "grad_norm": 0.671875,
      "learning_rate": 6.465839384426608e-05,
      "loss": 0.471,
      "step": 48607
    },
    {
      "epoch": 4.182052826292695,
      "grad_norm": 0.83984375,
      "learning_rate": 6.46450946512498e-05,
      "loss": 1.0821,
      "step": 48608
    },
    {
      "epoch": 4.182138862600017,
      "grad_norm": 1.0625,
      "learning_rate": 6.463179673156177e-05,
      "loss": 0.9281,
      "step": 48609
    },
    {
      "epoch": 4.1822248989073385,
      "grad_norm": 1.4140625,
      "learning_rate": 6.461850008524112e-05,
      "loss": 1.113,
      "step": 48610
    },
    {
      "epoch": 4.182310935214661,
      "grad_norm": 0.828125,
      "learning_rate": 6.460520471232678e-05,
      "loss": 0.4591,
      "step": 48611
    },
    {
      "epoch": 4.182396971521983,
      "grad_norm": 1.5,
      "learning_rate": 6.459191061285746e-05,
      "loss": 1.4565,
      "step": 48612
    },
    {
      "epoch": 4.182483007829304,
      "grad_norm": 1.1328125,
      "learning_rate": 6.457861778687208e-05,
      "loss": 1.5114,
      "step": 48613
    },
    {
      "epoch": 4.182569044136626,
      "grad_norm": 0.9140625,
      "learning_rate": 6.456532623440969e-05,
      "loss": 0.6777,
      "step": 48614
    },
    {
      "epoch": 4.182655080443947,
      "grad_norm": 1.2734375,
      "learning_rate": 6.45520359555089e-05,
      "loss": 2.0585,
      "step": 48615
    },
    {
      "epoch": 4.182741116751269,
      "grad_norm": 0.765625,
      "learning_rate": 6.453874695020884e-05,
      "loss": 0.9781,
      "step": 48616
    },
    {
      "epoch": 4.18282715305859,
      "grad_norm": 1.0703125,
      "learning_rate": 6.452545921854813e-05,
      "loss": 1.3187,
      "step": 48617
    },
    {
      "epoch": 4.182913189365912,
      "grad_norm": 3.53125,
      "learning_rate": 6.451217276056582e-05,
      "loss": 1.0237,
      "step": 48618
    },
    {
      "epoch": 4.1829992256732345,
      "grad_norm": 0.90234375,
      "learning_rate": 6.449888757630074e-05,
      "loss": 0.6631,
      "step": 48619
    },
    {
      "epoch": 4.183085261980556,
      "grad_norm": 1.5546875,
      "learning_rate": 6.448560366579159e-05,
      "loss": 1.0729,
      "step": 48620
    },
    {
      "epoch": 4.183171298287878,
      "grad_norm": 1.3671875,
      "learning_rate": 6.44723210290774e-05,
      "loss": 1.4572,
      "step": 48621
    },
    {
      "epoch": 4.183257334595199,
      "grad_norm": 1.0703125,
      "learning_rate": 6.44590396661971e-05,
      "loss": 0.8222,
      "step": 48622
    },
    {
      "epoch": 4.183343370902521,
      "grad_norm": 1.046875,
      "learning_rate": 6.444575957718923e-05,
      "loss": 0.7042,
      "step": 48623
    },
    {
      "epoch": 4.183429407209842,
      "grad_norm": 1.1953125,
      "learning_rate": 6.44324807620928e-05,
      "loss": 1.0398,
      "step": 48624
    },
    {
      "epoch": 4.183515443517164,
      "grad_norm": 0.99609375,
      "learning_rate": 6.441920322094679e-05,
      "loss": 1.2656,
      "step": 48625
    },
    {
      "epoch": 4.183601479824486,
      "grad_norm": 0.85546875,
      "learning_rate": 6.440592695378977e-05,
      "loss": 0.9724,
      "step": 48626
    },
    {
      "epoch": 4.1836875161318074,
      "grad_norm": 0.96875,
      "learning_rate": 6.439265196066085e-05,
      "loss": 1.0857,
      "step": 48627
    },
    {
      "epoch": 4.1837735524391295,
      "grad_norm": 0.9765625,
      "learning_rate": 6.437937824159857e-05,
      "loss": 0.8269,
      "step": 48628
    },
    {
      "epoch": 4.183859588746451,
      "grad_norm": 0.9765625,
      "learning_rate": 6.436610579664182e-05,
      "loss": 1.1876,
      "step": 48629
    },
    {
      "epoch": 4.183945625053773,
      "grad_norm": 330.0,
      "learning_rate": 6.435283462582964e-05,
      "loss": 0.5581,
      "step": 48630
    },
    {
      "epoch": 4.184031661361095,
      "grad_norm": 0.76171875,
      "learning_rate": 6.43395647292006e-05,
      "loss": 0.5027,
      "step": 48631
    },
    {
      "epoch": 4.184117697668416,
      "grad_norm": 30.125,
      "learning_rate": 6.432629610679358e-05,
      "loss": 0.7305,
      "step": 48632
    },
    {
      "epoch": 4.184203733975738,
      "grad_norm": 1.140625,
      "learning_rate": 6.431302875864753e-05,
      "loss": 1.0651,
      "step": 48633
    },
    {
      "epoch": 4.184289770283059,
      "grad_norm": 113.0,
      "learning_rate": 6.429976268480103e-05,
      "loss": 0.9663,
      "step": 48634
    },
    {
      "epoch": 4.184375806590381,
      "grad_norm": 1.1015625,
      "learning_rate": 6.4286497885293e-05,
      "loss": 0.7902,
      "step": 48635
    },
    {
      "epoch": 4.1844618428977025,
      "grad_norm": 1.203125,
      "learning_rate": 6.42732343601623e-05,
      "loss": 1.2699,
      "step": 48636
    },
    {
      "epoch": 4.184547879205025,
      "grad_norm": 64.5,
      "learning_rate": 6.425997210944745e-05,
      "loss": 0.9778,
      "step": 48637
    },
    {
      "epoch": 4.184633915512347,
      "grad_norm": 0.8828125,
      "learning_rate": 6.424671113318764e-05,
      "loss": 0.9007,
      "step": 48638
    },
    {
      "epoch": 4.184719951819668,
      "grad_norm": 1.1171875,
      "learning_rate": 6.423345143142134e-05,
      "loss": 0.9109,
      "step": 48639
    },
    {
      "epoch": 4.18480598812699,
      "grad_norm": 1.0234375,
      "learning_rate": 6.422019300418741e-05,
      "loss": 1.3572,
      "step": 48640
    },
    {
      "epoch": 4.184892024434311,
      "grad_norm": 1.1796875,
      "learning_rate": 6.420693585152482e-05,
      "loss": 1.3178,
      "step": 48641
    },
    {
      "epoch": 4.184978060741633,
      "grad_norm": 3.75,
      "learning_rate": 6.419367997347203e-05,
      "loss": 0.417,
      "step": 48642
    },
    {
      "epoch": 4.185064097048954,
      "grad_norm": 0.86328125,
      "learning_rate": 6.418042537006813e-05,
      "loss": 0.5559,
      "step": 48643
    },
    {
      "epoch": 4.185150133356276,
      "grad_norm": 1.140625,
      "learning_rate": 6.416717204135158e-05,
      "loss": 0.7676,
      "step": 48644
    },
    {
      "epoch": 4.1852361696635985,
      "grad_norm": 1.0703125,
      "learning_rate": 6.415391998736131e-05,
      "loss": 0.8934,
      "step": 48645
    },
    {
      "epoch": 4.18532220597092,
      "grad_norm": 1.3984375,
      "learning_rate": 6.414066920813616e-05,
      "loss": 1.5731,
      "step": 48646
    },
    {
      "epoch": 4.185408242278242,
      "grad_norm": 1.171875,
      "learning_rate": 6.412741970371466e-05,
      "loss": 0.6817,
      "step": 48647
    },
    {
      "epoch": 4.185494278585563,
      "grad_norm": 1.671875,
      "learning_rate": 6.411417147413568e-05,
      "loss": 1.1889,
      "step": 48648
    },
    {
      "epoch": 4.185580314892885,
      "grad_norm": 1.3046875,
      "learning_rate": 6.410092451943811e-05,
      "loss": 0.6092,
      "step": 48649
    },
    {
      "epoch": 4.185666351200206,
      "grad_norm": 1.2109375,
      "learning_rate": 6.408767883966044e-05,
      "loss": 1.0465,
      "step": 48650
    },
    {
      "epoch": 4.185752387507528,
      "grad_norm": 0.76171875,
      "learning_rate": 6.40744344348415e-05,
      "loss": 0.756,
      "step": 48651
    },
    {
      "epoch": 4.18583842381485,
      "grad_norm": 1.203125,
      "learning_rate": 6.40611913050202e-05,
      "loss": 1.2392,
      "step": 48652
    },
    {
      "epoch": 4.1859244601221715,
      "grad_norm": 1.21875,
      "learning_rate": 6.4047949450235e-05,
      "loss": 1.7418,
      "step": 48653
    },
    {
      "epoch": 4.1860104964294935,
      "grad_norm": 1.2265625,
      "learning_rate": 6.403470887052476e-05,
      "loss": 1.1798,
      "step": 48654
    },
    {
      "epoch": 4.186096532736815,
      "grad_norm": 2.890625,
      "learning_rate": 6.40214695659283e-05,
      "loss": 1.1136,
      "step": 48655
    },
    {
      "epoch": 4.186182569044137,
      "grad_norm": 0.96484375,
      "learning_rate": 6.400823153648416e-05,
      "loss": 0.7733,
      "step": 48656
    },
    {
      "epoch": 4.186268605351458,
      "grad_norm": 0.95703125,
      "learning_rate": 6.39949947822313e-05,
      "loss": 0.7184,
      "step": 48657
    },
    {
      "epoch": 4.18635464165878,
      "grad_norm": 1.1953125,
      "learning_rate": 6.39817593032081e-05,
      "loss": 0.8847,
      "step": 48658
    },
    {
      "epoch": 4.186440677966102,
      "grad_norm": 1.234375,
      "learning_rate": 6.39685250994535e-05,
      "loss": 1.5529,
      "step": 48659
    },
    {
      "epoch": 4.186526714273423,
      "grad_norm": 1.5078125,
      "learning_rate": 6.395529217100627e-05,
      "loss": 1.0151,
      "step": 48660
    },
    {
      "epoch": 4.186612750580745,
      "grad_norm": 0.87890625,
      "learning_rate": 6.394206051790485e-05,
      "loss": 0.9158,
      "step": 48661
    },
    {
      "epoch": 4.1866987868880665,
      "grad_norm": 0.9140625,
      "learning_rate": 6.392883014018813e-05,
      "loss": 0.5169,
      "step": 48662
    },
    {
      "epoch": 4.186784823195389,
      "grad_norm": 25.625,
      "learning_rate": 6.39156010378949e-05,
      "loss": 0.8263,
      "step": 48663
    },
    {
      "epoch": 4.18687085950271,
      "grad_norm": 1.0234375,
      "learning_rate": 6.390237321106357e-05,
      "loss": 0.8708,
      "step": 48664
    },
    {
      "epoch": 4.186956895810032,
      "grad_norm": 1.21875,
      "learning_rate": 6.3889146659733e-05,
      "loss": 1.4795,
      "step": 48665
    },
    {
      "epoch": 4.187042932117354,
      "grad_norm": 1.9140625,
      "learning_rate": 6.387592138394194e-05,
      "loss": 1.9494,
      "step": 48666
    },
    {
      "epoch": 4.187128968424675,
      "grad_norm": 1.7578125,
      "learning_rate": 6.386269738372891e-05,
      "loss": 0.5446,
      "step": 48667
    },
    {
      "epoch": 4.187215004731997,
      "grad_norm": 1.25,
      "learning_rate": 6.384947465913276e-05,
      "loss": 1.1961,
      "step": 48668
    },
    {
      "epoch": 4.187301041039318,
      "grad_norm": 0.9296875,
      "learning_rate": 6.383625321019198e-05,
      "loss": 1.5463,
      "step": 48669
    },
    {
      "epoch": 4.18738707734664,
      "grad_norm": 0.86328125,
      "learning_rate": 6.382303303694536e-05,
      "loss": 0.3988,
      "step": 48670
    },
    {
      "epoch": 4.187473113653962,
      "grad_norm": 1.109375,
      "learning_rate": 6.38098141394316e-05,
      "loss": 0.932,
      "step": 48671
    },
    {
      "epoch": 4.187559149961284,
      "grad_norm": 3.84375,
      "learning_rate": 6.37965965176892e-05,
      "loss": 1.4411,
      "step": 48672
    },
    {
      "epoch": 4.187645186268606,
      "grad_norm": 1.171875,
      "learning_rate": 6.378338017175689e-05,
      "loss": 1.0738,
      "step": 48673
    },
    {
      "epoch": 4.187731222575927,
      "grad_norm": 1.0546875,
      "learning_rate": 6.377016510167339e-05,
      "loss": 1.0361,
      "step": 48674
    },
    {
      "epoch": 4.187817258883249,
      "grad_norm": 0.94140625,
      "learning_rate": 6.375695130747732e-05,
      "loss": 0.6973,
      "step": 48675
    },
    {
      "epoch": 4.18790329519057,
      "grad_norm": 1.390625,
      "learning_rate": 6.374373878920741e-05,
      "loss": 1.6296,
      "step": 48676
    },
    {
      "epoch": 4.187989331497892,
      "grad_norm": 1.3046875,
      "learning_rate": 6.373052754690212e-05,
      "loss": 1.2169,
      "step": 48677
    },
    {
      "epoch": 4.188075367805213,
      "grad_norm": 1.296875,
      "learning_rate": 6.371731758060018e-05,
      "loss": 1.1994,
      "step": 48678
    },
    {
      "epoch": 4.1881614041125355,
      "grad_norm": 1.1796875,
      "learning_rate": 6.370410889034033e-05,
      "loss": 1.2613,
      "step": 48679
    },
    {
      "epoch": 4.1882474404198575,
      "grad_norm": 1.2578125,
      "learning_rate": 6.369090147616102e-05,
      "loss": 1.3191,
      "step": 48680
    },
    {
      "epoch": 4.188333476727179,
      "grad_norm": 4.75,
      "learning_rate": 6.367769533810097e-05,
      "loss": 1.7858,
      "step": 48681
    },
    {
      "epoch": 4.188419513034501,
      "grad_norm": 0.98046875,
      "learning_rate": 6.366449047619887e-05,
      "loss": 0.7893,
      "step": 48682
    },
    {
      "epoch": 4.188505549341822,
      "grad_norm": 0.796875,
      "learning_rate": 6.36512868904932e-05,
      "loss": 0.8356,
      "step": 48683
    },
    {
      "epoch": 4.188591585649144,
      "grad_norm": 1.2265625,
      "learning_rate": 6.363808458102277e-05,
      "loss": 1.1758,
      "step": 48684
    },
    {
      "epoch": 4.188677621956465,
      "grad_norm": 0.71484375,
      "learning_rate": 6.362488354782591e-05,
      "loss": 0.6215,
      "step": 48685
    },
    {
      "epoch": 4.188763658263787,
      "grad_norm": 0.94921875,
      "learning_rate": 6.361168379094146e-05,
      "loss": 0.649,
      "step": 48686
    },
    {
      "epoch": 4.188849694571109,
      "grad_norm": 0.96484375,
      "learning_rate": 6.359848531040801e-05,
      "loss": 0.6792,
      "step": 48687
    },
    {
      "epoch": 4.1889357308784305,
      "grad_norm": 0.875,
      "learning_rate": 6.358528810626407e-05,
      "loss": 1.2681,
      "step": 48688
    },
    {
      "epoch": 4.189021767185753,
      "grad_norm": 544.0,
      "learning_rate": 6.357209217854826e-05,
      "loss": 1.1442,
      "step": 48689
    },
    {
      "epoch": 4.189107803493074,
      "grad_norm": 1.1328125,
      "learning_rate": 6.355889752729933e-05,
      "loss": 1.247,
      "step": 48690
    },
    {
      "epoch": 4.189193839800396,
      "grad_norm": 1.0078125,
      "learning_rate": 6.354570415255556e-05,
      "loss": 0.9877,
      "step": 48691
    },
    {
      "epoch": 4.189279876107717,
      "grad_norm": 8.5625,
      "learning_rate": 6.353251205435579e-05,
      "loss": 1.1898,
      "step": 48692
    },
    {
      "epoch": 4.189365912415039,
      "grad_norm": 22.0,
      "learning_rate": 6.351932123273862e-05,
      "loss": 0.9028,
      "step": 48693
    },
    {
      "epoch": 4.189451948722361,
      "grad_norm": 1.03125,
      "learning_rate": 6.350613168774244e-05,
      "loss": 0.7804,
      "step": 48694
    },
    {
      "epoch": 4.189537985029682,
      "grad_norm": 1.03125,
      "learning_rate": 6.349294341940592e-05,
      "loss": 0.7007,
      "step": 48695
    },
    {
      "epoch": 4.189624021337004,
      "grad_norm": 0.96484375,
      "learning_rate": 6.347975642776776e-05,
      "loss": 1.6171,
      "step": 48696
    },
    {
      "epoch": 4.189710057644326,
      "grad_norm": 1.0,
      "learning_rate": 6.346657071286627e-05,
      "loss": 0.5519,
      "step": 48697
    },
    {
      "epoch": 4.189796093951648,
      "grad_norm": 0.96875,
      "learning_rate": 6.345338627474035e-05,
      "loss": 1.2513,
      "step": 48698
    },
    {
      "epoch": 4.189882130258969,
      "grad_norm": 0.86328125,
      "learning_rate": 6.344020311342818e-05,
      "loss": 0.72,
      "step": 48699
    },
    {
      "epoch": 4.189968166566291,
      "grad_norm": 1.0234375,
      "learning_rate": 6.342702122896854e-05,
      "loss": 0.8227,
      "step": 48700
    },
    {
      "epoch": 4.190054202873613,
      "grad_norm": 1.140625,
      "learning_rate": 6.34138406214001e-05,
      "loss": 1.1072,
      "step": 48701
    },
    {
      "epoch": 4.190140239180934,
      "grad_norm": 0.9453125,
      "learning_rate": 6.34006612907611e-05,
      "loss": 0.433,
      "step": 48702
    },
    {
      "epoch": 4.190226275488256,
      "grad_norm": 1.1484375,
      "learning_rate": 6.338748323709026e-05,
      "loss": 1.1962,
      "step": 48703
    },
    {
      "epoch": 4.190312311795577,
      "grad_norm": 11.8125,
      "learning_rate": 6.337430646042624e-05,
      "loss": 0.9624,
      "step": 48704
    },
    {
      "epoch": 4.1903983481028995,
      "grad_norm": 1.7421875,
      "learning_rate": 6.336113096080736e-05,
      "loss": 1.5853,
      "step": 48705
    },
    {
      "epoch": 4.1904843844102215,
      "grad_norm": 1.0625,
      "learning_rate": 6.334795673827227e-05,
      "loss": 0.7158,
      "step": 48706
    },
    {
      "epoch": 4.190570420717543,
      "grad_norm": 15.9375,
      "learning_rate": 6.33347837928595e-05,
      "loss": 0.7427,
      "step": 48707
    },
    {
      "epoch": 4.190656457024865,
      "grad_norm": 1.3984375,
      "learning_rate": 6.332161212460752e-05,
      "loss": 1.0448,
      "step": 48708
    },
    {
      "epoch": 4.190742493332186,
      "grad_norm": 1.609375,
      "learning_rate": 6.330844173355499e-05,
      "loss": 1.2688,
      "step": 48709
    },
    {
      "epoch": 4.190828529639508,
      "grad_norm": 150.0,
      "learning_rate": 6.329527261974027e-05,
      "loss": 0.9091,
      "step": 48710
    },
    {
      "epoch": 4.190914565946829,
      "grad_norm": 1.1328125,
      "learning_rate": 6.328210478320179e-05,
      "loss": 0.8587,
      "step": 48711
    },
    {
      "epoch": 4.191000602254151,
      "grad_norm": 272.0,
      "learning_rate": 6.326893822397844e-05,
      "loss": 1.3631,
      "step": 48712
    },
    {
      "epoch": 4.191086638561473,
      "grad_norm": 1.234375,
      "learning_rate": 6.325577294210838e-05,
      "loss": 0.3894,
      "step": 48713
    },
    {
      "epoch": 4.1911726748687945,
      "grad_norm": 1.234375,
      "learning_rate": 6.324260893763034e-05,
      "loss": 0.7168,
      "step": 48714
    },
    {
      "epoch": 4.191258711176117,
      "grad_norm": 1.03125,
      "learning_rate": 6.322944621058268e-05,
      "loss": 0.9877,
      "step": 48715
    },
    {
      "epoch": 4.191344747483438,
      "grad_norm": 3.015625,
      "learning_rate": 6.321628476100388e-05,
      "loss": 0.7812,
      "step": 48716
    },
    {
      "epoch": 4.19143078379076,
      "grad_norm": 0.98828125,
      "learning_rate": 6.320312458893262e-05,
      "loss": 0.5939,
      "step": 48717
    },
    {
      "epoch": 4.191516820098081,
      "grad_norm": 1.984375,
      "learning_rate": 6.318996569440722e-05,
      "loss": 1.3359,
      "step": 48718
    },
    {
      "epoch": 4.191602856405403,
      "grad_norm": 0.734375,
      "learning_rate": 6.317680807746611e-05,
      "loss": 0.4369,
      "step": 48719
    },
    {
      "epoch": 4.191688892712725,
      "grad_norm": 1.015625,
      "learning_rate": 6.316365173814809e-05,
      "loss": 1.6406,
      "step": 48720
    },
    {
      "epoch": 4.191774929020046,
      "grad_norm": 1024.0,
      "learning_rate": 6.315049667649125e-05,
      "loss": 1.7515,
      "step": 48721
    },
    {
      "epoch": 4.191860965327368,
      "grad_norm": 1.40625,
      "learning_rate": 6.313734289253426e-05,
      "loss": 1.5615,
      "step": 48722
    },
    {
      "epoch": 4.19194700163469,
      "grad_norm": 1.109375,
      "learning_rate": 6.312419038631572e-05,
      "loss": 1.125,
      "step": 48723
    },
    {
      "epoch": 4.192033037942012,
      "grad_norm": 1.0234375,
      "learning_rate": 6.31110391578738e-05,
      "loss": 0.5614,
      "step": 48724
    },
    {
      "epoch": 4.192119074249333,
      "grad_norm": 1.0,
      "learning_rate": 6.309788920724718e-05,
      "loss": 0.8982,
      "step": 48725
    },
    {
      "epoch": 4.192205110556655,
      "grad_norm": 1.2578125,
      "learning_rate": 6.308474053447433e-05,
      "loss": 1.1456,
      "step": 48726
    },
    {
      "epoch": 4.192291146863977,
      "grad_norm": 1.1953125,
      "learning_rate": 6.307159313959349e-05,
      "loss": 1.8623,
      "step": 48727
    },
    {
      "epoch": 4.192377183171298,
      "grad_norm": 1.265625,
      "learning_rate": 6.305844702264346e-05,
      "loss": 1.6309,
      "step": 48728
    },
    {
      "epoch": 4.19246321947862,
      "grad_norm": 1.140625,
      "learning_rate": 6.304530218366233e-05,
      "loss": 1.0524,
      "step": 48729
    },
    {
      "epoch": 4.192549255785941,
      "grad_norm": 1.1015625,
      "learning_rate": 6.303215862268869e-05,
      "loss": 0.5194,
      "step": 48730
    },
    {
      "epoch": 4.1926352920932635,
      "grad_norm": 0.65234375,
      "learning_rate": 6.301901633976114e-05,
      "loss": 0.5598,
      "step": 48731
    },
    {
      "epoch": 4.192721328400585,
      "grad_norm": 1.171875,
      "learning_rate": 6.300587533491785e-05,
      "loss": 0.9218,
      "step": 48732
    },
    {
      "epoch": 4.192807364707907,
      "grad_norm": 1.140625,
      "learning_rate": 6.29927356081974e-05,
      "loss": 1.2144,
      "step": 48733
    },
    {
      "epoch": 4.192893401015229,
      "grad_norm": 5.03125,
      "learning_rate": 6.29795971596383e-05,
      "loss": 1.687,
      "step": 48734
    },
    {
      "epoch": 4.19297943732255,
      "grad_norm": 0.93359375,
      "learning_rate": 6.29664599892788e-05,
      "loss": 0.591,
      "step": 48735
    },
    {
      "epoch": 4.193065473629872,
      "grad_norm": 1.28125,
      "learning_rate": 6.295332409715737e-05,
      "loss": 0.8545,
      "step": 48736
    },
    {
      "epoch": 4.193151509937193,
      "grad_norm": 1.046875,
      "learning_rate": 6.294018948331264e-05,
      "loss": 1.0871,
      "step": 48737
    },
    {
      "epoch": 4.193237546244515,
      "grad_norm": 1.296875,
      "learning_rate": 6.292705614778266e-05,
      "loss": 0.8457,
      "step": 48738
    },
    {
      "epoch": 4.1933235825518365,
      "grad_norm": 0.953125,
      "learning_rate": 6.29139240906062e-05,
      "loss": 1.6305,
      "step": 48739
    },
    {
      "epoch": 4.1934096188591585,
      "grad_norm": 15.75,
      "learning_rate": 6.290079331182136e-05,
      "loss": 0.7633,
      "step": 48740
    },
    {
      "epoch": 4.193495655166481,
      "grad_norm": 1.2109375,
      "learning_rate": 6.288766381146671e-05,
      "loss": 0.7065,
      "step": 48741
    },
    {
      "epoch": 4.193581691473802,
      "grad_norm": 1.140625,
      "learning_rate": 6.287453558958078e-05,
      "loss": 0.8954,
      "step": 48742
    },
    {
      "epoch": 4.193667727781124,
      "grad_norm": 1.765625,
      "learning_rate": 6.286140864620166e-05,
      "loss": 1.0084,
      "step": 48743
    },
    {
      "epoch": 4.193753764088445,
      "grad_norm": 1.5234375,
      "learning_rate": 6.28482829813679e-05,
      "loss": 1.6145,
      "step": 48744
    },
    {
      "epoch": 4.193839800395767,
      "grad_norm": 1.1953125,
      "learning_rate": 6.283515859511801e-05,
      "loss": 0.7423,
      "step": 48745
    },
    {
      "epoch": 4.193925836703089,
      "grad_norm": 1.8828125,
      "learning_rate": 6.28220354874901e-05,
      "loss": 1.5073,
      "step": 48746
    },
    {
      "epoch": 4.19401187301041,
      "grad_norm": 3.15625,
      "learning_rate": 6.280891365852275e-05,
      "loss": 0.7005,
      "step": 48747
    },
    {
      "epoch": 4.194097909317732,
      "grad_norm": 0.9296875,
      "learning_rate": 6.279579310825433e-05,
      "loss": 0.8293,
      "step": 48748
    },
    {
      "epoch": 4.194183945625054,
      "grad_norm": 1.0078125,
      "learning_rate": 6.278267383672315e-05,
      "loss": 0.6747,
      "step": 48749
    },
    {
      "epoch": 4.194269981932376,
      "grad_norm": 1.171875,
      "learning_rate": 6.27695558439677e-05,
      "loss": 1.2279,
      "step": 48750
    },
    {
      "epoch": 4.194356018239697,
      "grad_norm": 1.046875,
      "learning_rate": 6.275643913002621e-05,
      "loss": 1.1466,
      "step": 48751
    },
    {
      "epoch": 4.194442054547019,
      "grad_norm": 1.3828125,
      "learning_rate": 6.274332369493707e-05,
      "loss": 0.8208,
      "step": 48752
    },
    {
      "epoch": 4.194528090854341,
      "grad_norm": 1.1640625,
      "learning_rate": 6.273020953873875e-05,
      "loss": 1.3101,
      "step": 48753
    },
    {
      "epoch": 4.194614127161662,
      "grad_norm": 1.1015625,
      "learning_rate": 6.271709666146947e-05,
      "loss": 0.9728,
      "step": 48754
    },
    {
      "epoch": 4.194700163468984,
      "grad_norm": 0.96484375,
      "learning_rate": 6.270398506316771e-05,
      "loss": 1.2787,
      "step": 48755
    },
    {
      "epoch": 4.194786199776305,
      "grad_norm": 1.0546875,
      "learning_rate": 6.269087474387164e-05,
      "loss": 1.6305,
      "step": 48756
    },
    {
      "epoch": 4.1948722360836275,
      "grad_norm": 1.1484375,
      "learning_rate": 6.267776570361971e-05,
      "loss": 1.0879,
      "step": 48757
    },
    {
      "epoch": 4.194958272390949,
      "grad_norm": 1.0234375,
      "learning_rate": 6.266465794245035e-05,
      "loss": 1.6969,
      "step": 48758
    },
    {
      "epoch": 4.195044308698271,
      "grad_norm": 2.578125,
      "learning_rate": 6.265155146040175e-05,
      "loss": 0.9572,
      "step": 48759
    },
    {
      "epoch": 4.195130345005593,
      "grad_norm": 0.953125,
      "learning_rate": 6.263844625751225e-05,
      "loss": 1.4756,
      "step": 48760
    },
    {
      "epoch": 4.195216381312914,
      "grad_norm": 1.25,
      "learning_rate": 6.262534233382034e-05,
      "loss": 1.0968,
      "step": 48761
    },
    {
      "epoch": 4.195302417620236,
      "grad_norm": 216.0,
      "learning_rate": 6.26122396893642e-05,
      "loss": 1.0153,
      "step": 48762
    },
    {
      "epoch": 4.195388453927557,
      "grad_norm": 1.5625,
      "learning_rate": 6.259913832418212e-05,
      "loss": 1.3324,
      "step": 48763
    },
    {
      "epoch": 4.195474490234879,
      "grad_norm": 1.21875,
      "learning_rate": 6.258603823831261e-05,
      "loss": 0.9414,
      "step": 48764
    },
    {
      "epoch": 4.1955605265422005,
      "grad_norm": 20.625,
      "learning_rate": 6.257293943179376e-05,
      "loss": 1.7263,
      "step": 48765
    },
    {
      "epoch": 4.1956465628495225,
      "grad_norm": 1.40625,
      "learning_rate": 6.255984190466402e-05,
      "loss": 1.2572,
      "step": 48766
    },
    {
      "epoch": 4.195732599156845,
      "grad_norm": 0.7890625,
      "learning_rate": 6.254674565696172e-05,
      "loss": 0.6962,
      "step": 48767
    },
    {
      "epoch": 4.195818635464166,
      "grad_norm": 238.0,
      "learning_rate": 6.253365068872502e-05,
      "loss": 0.8744,
      "step": 48768
    },
    {
      "epoch": 4.195904671771488,
      "grad_norm": 1.0,
      "learning_rate": 6.252055699999243e-05,
      "loss": 0.4927,
      "step": 48769
    },
    {
      "epoch": 4.195990708078809,
      "grad_norm": 1.3828125,
      "learning_rate": 6.250746459080203e-05,
      "loss": 1.2,
      "step": 48770
    },
    {
      "epoch": 4.196076744386131,
      "grad_norm": 1.0234375,
      "learning_rate": 6.249437346119219e-05,
      "loss": 0.5662,
      "step": 48771
    },
    {
      "epoch": 4.196162780693452,
      "grad_norm": 4.3125,
      "learning_rate": 6.248128361120136e-05,
      "loss": 1.0244,
      "step": 48772
    },
    {
      "epoch": 4.196248817000774,
      "grad_norm": 1.0078125,
      "learning_rate": 6.246819504086754e-05,
      "loss": 0.6025,
      "step": 48773
    },
    {
      "epoch": 4.196334853308096,
      "grad_norm": 0.94140625,
      "learning_rate": 6.245510775022917e-05,
      "loss": 1.2281,
      "step": 48774
    },
    {
      "epoch": 4.196420889615418,
      "grad_norm": 1.1875,
      "learning_rate": 6.244202173932456e-05,
      "loss": 1.0794,
      "step": 48775
    },
    {
      "epoch": 4.19650692592274,
      "grad_norm": 1.25,
      "learning_rate": 6.242893700819191e-05,
      "loss": 0.9895,
      "step": 48776
    },
    {
      "epoch": 4.196592962230061,
      "grad_norm": 1.1953125,
      "learning_rate": 6.241585355686946e-05,
      "loss": 1.8859,
      "step": 48777
    },
    {
      "epoch": 4.196678998537383,
      "grad_norm": 0.94921875,
      "learning_rate": 6.240277138539568e-05,
      "loss": 1.3207,
      "step": 48778
    },
    {
      "epoch": 4.196765034844704,
      "grad_norm": 1.03125,
      "learning_rate": 6.238969049380855e-05,
      "loss": 1.0336,
      "step": 48779
    },
    {
      "epoch": 4.196851071152026,
      "grad_norm": 1704.0,
      "learning_rate": 6.237661088214657e-05,
      "loss": 0.936,
      "step": 48780
    },
    {
      "epoch": 4.196937107459348,
      "grad_norm": 0.91796875,
      "learning_rate": 6.23635325504478e-05,
      "loss": 0.7696,
      "step": 48781
    },
    {
      "epoch": 4.197023143766669,
      "grad_norm": 1.0703125,
      "learning_rate": 6.235045549875063e-05,
      "loss": 0.7578,
      "step": 48782
    },
    {
      "epoch": 4.1971091800739915,
      "grad_norm": 1.140625,
      "learning_rate": 6.23373797270933e-05,
      "loss": 0.7598,
      "step": 48783
    },
    {
      "epoch": 4.197195216381313,
      "grad_norm": 1.109375,
      "learning_rate": 6.232430523551392e-05,
      "loss": 0.7623,
      "step": 48784
    },
    {
      "epoch": 4.197281252688635,
      "grad_norm": 1.0859375,
      "learning_rate": 6.231123202405082e-05,
      "loss": 0.7959,
      "step": 48785
    },
    {
      "epoch": 4.197367288995956,
      "grad_norm": 1.0390625,
      "learning_rate": 6.229816009274225e-05,
      "loss": 1.2439,
      "step": 48786
    },
    {
      "epoch": 4.197453325303278,
      "grad_norm": 1.2265625,
      "learning_rate": 6.228508944162642e-05,
      "loss": 0.8406,
      "step": 48787
    },
    {
      "epoch": 4.1975393616106,
      "grad_norm": 14.125,
      "learning_rate": 6.227202007074168e-05,
      "loss": 1.3119,
      "step": 48788
    },
    {
      "epoch": 4.197625397917921,
      "grad_norm": 15.3125,
      "learning_rate": 6.225895198012605e-05,
      "loss": 1.4058,
      "step": 48789
    },
    {
      "epoch": 4.197711434225243,
      "grad_norm": 1.296875,
      "learning_rate": 6.224588516981788e-05,
      "loss": 1.2713,
      "step": 48790
    },
    {
      "epoch": 4.1977974705325645,
      "grad_norm": 16.375,
      "learning_rate": 6.223281963985539e-05,
      "loss": 0.7715,
      "step": 48791
    },
    {
      "epoch": 4.1978835068398865,
      "grad_norm": 10.0625,
      "learning_rate": 6.221975539027675e-05,
      "loss": 1.5648,
      "step": 48792
    },
    {
      "epoch": 4.197969543147208,
      "grad_norm": 0.96484375,
      "learning_rate": 6.22066924211201e-05,
      "loss": 1.2012,
      "step": 48793
    },
    {
      "epoch": 4.19805557945453,
      "grad_norm": 1.578125,
      "learning_rate": 6.219363073242384e-05,
      "loss": 1.0787,
      "step": 48794
    },
    {
      "epoch": 4.198141615761852,
      "grad_norm": 0.8125,
      "learning_rate": 6.218057032422603e-05,
      "loss": 1.2002,
      "step": 48795
    },
    {
      "epoch": 4.198227652069173,
      "grad_norm": 0.921875,
      "learning_rate": 6.216751119656494e-05,
      "loss": 1.9427,
      "step": 48796
    },
    {
      "epoch": 4.198313688376495,
      "grad_norm": 12.6875,
      "learning_rate": 6.215445334947862e-05,
      "loss": 1.3923,
      "step": 48797
    },
    {
      "epoch": 4.198399724683816,
      "grad_norm": 1.3125,
      "learning_rate": 6.21413967830054e-05,
      "loss": 1.2768,
      "step": 48798
    },
    {
      "epoch": 4.198485760991138,
      "grad_norm": 36.0,
      "learning_rate": 6.212834149718354e-05,
      "loss": 0.9103,
      "step": 48799
    },
    {
      "epoch": 4.1985717972984595,
      "grad_norm": 9.875,
      "learning_rate": 6.211528749205098e-05,
      "loss": 0.7946,
      "step": 48800
    },
    {
      "epoch": 4.198657833605782,
      "grad_norm": 1.1015625,
      "learning_rate": 6.210223476764609e-05,
      "loss": 0.9267,
      "step": 48801
    },
    {
      "epoch": 4.198743869913104,
      "grad_norm": 0.9375,
      "learning_rate": 6.208918332400704e-05,
      "loss": 1.0366,
      "step": 48802
    },
    {
      "epoch": 4.198829906220425,
      "grad_norm": 1.265625,
      "learning_rate": 6.207613316117189e-05,
      "loss": 0.9024,
      "step": 48803
    },
    {
      "epoch": 4.198915942527747,
      "grad_norm": 1.2421875,
      "learning_rate": 6.20630842791789e-05,
      "loss": 1.8349,
      "step": 48804
    },
    {
      "epoch": 4.199001978835068,
      "grad_norm": 1.4609375,
      "learning_rate": 6.20500366780663e-05,
      "loss": 1.0689,
      "step": 48805
    },
    {
      "epoch": 4.19908801514239,
      "grad_norm": 0.75390625,
      "learning_rate": 6.203699035787203e-05,
      "loss": 0.3457,
      "step": 48806
    },
    {
      "epoch": 4.199174051449711,
      "grad_norm": 2.015625,
      "learning_rate": 6.202394531863442e-05,
      "loss": 1.1308,
      "step": 48807
    },
    {
      "epoch": 4.199260087757033,
      "grad_norm": 0.8984375,
      "learning_rate": 6.201090156039169e-05,
      "loss": 0.7666,
      "step": 48808
    },
    {
      "epoch": 4.1993461240643555,
      "grad_norm": 1.1875,
      "learning_rate": 6.199785908318179e-05,
      "loss": 1.2682,
      "step": 48809
    },
    {
      "epoch": 4.199432160371677,
      "grad_norm": 194.0,
      "learning_rate": 6.198481788704308e-05,
      "loss": 0.88,
      "step": 48810
    },
    {
      "epoch": 4.199518196678999,
      "grad_norm": 1.6171875,
      "learning_rate": 6.197177797201342e-05,
      "loss": 1.6793,
      "step": 48811
    },
    {
      "epoch": 4.19960423298632,
      "grad_norm": 1.1171875,
      "learning_rate": 6.195873933813118e-05,
      "loss": 1.0851,
      "step": 48812
    },
    {
      "epoch": 4.199690269293642,
      "grad_norm": 0.94140625,
      "learning_rate": 6.194570198543453e-05,
      "loss": 1.8021,
      "step": 48813
    },
    {
      "epoch": 4.199776305600963,
      "grad_norm": 0.84765625,
      "learning_rate": 6.193266591396135e-05,
      "loss": 0.408,
      "step": 48814
    },
    {
      "epoch": 4.199862341908285,
      "grad_norm": 1.203125,
      "learning_rate": 6.191963112374998e-05,
      "loss": 1.67,
      "step": 48815
    },
    {
      "epoch": 4.199948378215607,
      "grad_norm": 1.09375,
      "learning_rate": 6.190659761483858e-05,
      "loss": 0.7207,
      "step": 48816
    },
    {
      "epoch": 4.2000344145229285,
      "grad_norm": 1.0546875,
      "learning_rate": 6.189356538726504e-05,
      "loss": 1.4679,
      "step": 48817
    },
    {
      "epoch": 4.2001204508302505,
      "grad_norm": 1.046875,
      "learning_rate": 6.188053444106766e-05,
      "loss": 1.1789,
      "step": 48818
    },
    {
      "epoch": 4.200206487137572,
      "grad_norm": 1.09375,
      "learning_rate": 6.186750477628461e-05,
      "loss": 1.4557,
      "step": 48819
    },
    {
      "epoch": 4.200292523444894,
      "grad_norm": 1.078125,
      "learning_rate": 6.185447639295378e-05,
      "loss": 0.5725,
      "step": 48820
    },
    {
      "epoch": 4.200378559752216,
      "grad_norm": 1.2421875,
      "learning_rate": 6.184144929111346e-05,
      "loss": 1.1488,
      "step": 48821
    },
    {
      "epoch": 4.200464596059537,
      "grad_norm": 27.0,
      "learning_rate": 6.182842347080153e-05,
      "loss": 1.5466,
      "step": 48822
    },
    {
      "epoch": 4.200550632366859,
      "grad_norm": 0.96484375,
      "learning_rate": 6.181539893205628e-05,
      "loss": 0.6763,
      "step": 48823
    },
    {
      "epoch": 4.20063666867418,
      "grad_norm": 1.1171875,
      "learning_rate": 6.180237567491593e-05,
      "loss": 1.4627,
      "step": 48824
    },
    {
      "epoch": 4.200722704981502,
      "grad_norm": 1.046875,
      "learning_rate": 6.178935369941829e-05,
      "loss": 1.5032,
      "step": 48825
    },
    {
      "epoch": 4.2008087412888235,
      "grad_norm": 0.97265625,
      "learning_rate": 6.177633300560165e-05,
      "loss": 0.6105,
      "step": 48826
    },
    {
      "epoch": 4.200894777596146,
      "grad_norm": 1.03125,
      "learning_rate": 6.176331359350396e-05,
      "loss": 1.0857,
      "step": 48827
    },
    {
      "epoch": 4.200980813903468,
      "grad_norm": 1.03125,
      "learning_rate": 6.175029546316324e-05,
      "loss": 0.8787,
      "step": 48828
    },
    {
      "epoch": 4.201066850210789,
      "grad_norm": 1.078125,
      "learning_rate": 6.173727861461786e-05,
      "loss": 0.6855,
      "step": 48829
    },
    {
      "epoch": 4.201152886518111,
      "grad_norm": 1.1953125,
      "learning_rate": 6.172426304790557e-05,
      "loss": 0.9557,
      "step": 48830
    },
    {
      "epoch": 4.201238922825432,
      "grad_norm": 1.0078125,
      "learning_rate": 6.171124876306456e-05,
      "loss": 1.5623,
      "step": 48831
    },
    {
      "epoch": 4.201324959132754,
      "grad_norm": 1.1015625,
      "learning_rate": 6.169823576013306e-05,
      "loss": 1.906,
      "step": 48832
    },
    {
      "epoch": 4.201410995440075,
      "grad_norm": 1.3203125,
      "learning_rate": 6.168522403914884e-05,
      "loss": 0.9983,
      "step": 48833
    },
    {
      "epoch": 4.201497031747397,
      "grad_norm": 1.140625,
      "learning_rate": 6.167221360015007e-05,
      "loss": 1.3053,
      "step": 48834
    },
    {
      "epoch": 4.2015830680547195,
      "grad_norm": 0.8984375,
      "learning_rate": 6.165920444317491e-05,
      "loss": 0.673,
      "step": 48835
    },
    {
      "epoch": 4.201669104362041,
      "grad_norm": 1.3046875,
      "learning_rate": 6.164619656826126e-05,
      "loss": 0.7918,
      "step": 48836
    },
    {
      "epoch": 4.201755140669363,
      "grad_norm": 20.125,
      "learning_rate": 6.163318997544737e-05,
      "loss": 1.8889,
      "step": 48837
    },
    {
      "epoch": 4.201841176976684,
      "grad_norm": 1.0390625,
      "learning_rate": 6.162018466477098e-05,
      "loss": 0.7672,
      "step": 48838
    },
    {
      "epoch": 4.201927213284006,
      "grad_norm": 0.890625,
      "learning_rate": 6.16071806362703e-05,
      "loss": 0.6245,
      "step": 48839
    },
    {
      "epoch": 4.202013249591327,
      "grad_norm": 1.109375,
      "learning_rate": 6.159417788998344e-05,
      "loss": 1.0977,
      "step": 48840
    },
    {
      "epoch": 4.202099285898649,
      "grad_norm": 1.0390625,
      "learning_rate": 6.158117642594829e-05,
      "loss": 0.7974,
      "step": 48841
    },
    {
      "epoch": 4.202185322205971,
      "grad_norm": 15.5,
      "learning_rate": 6.156817624420291e-05,
      "loss": 1.0364,
      "step": 48842
    },
    {
      "epoch": 4.2022713585132925,
      "grad_norm": 0.97265625,
      "learning_rate": 6.155517734478544e-05,
      "loss": 1.1723,
      "step": 48843
    },
    {
      "epoch": 4.2023573948206145,
      "grad_norm": 0.91015625,
      "learning_rate": 6.154217972773369e-05,
      "loss": 0.6387,
      "step": 48844
    },
    {
      "epoch": 4.202443431127936,
      "grad_norm": 1.2421875,
      "learning_rate": 6.152918339308583e-05,
      "loss": 0.58,
      "step": 48845
    },
    {
      "epoch": 4.202529467435258,
      "grad_norm": 1.296875,
      "learning_rate": 6.15161883408799e-05,
      "loss": 0.9524,
      "step": 48846
    },
    {
      "epoch": 4.202615503742579,
      "grad_norm": 1.03125,
      "learning_rate": 6.150319457115371e-05,
      "loss": 1.1617,
      "step": 48847
    },
    {
      "epoch": 4.202701540049901,
      "grad_norm": 0.73828125,
      "learning_rate": 6.149020208394545e-05,
      "loss": 0.6859,
      "step": 48848
    },
    {
      "epoch": 4.202787576357223,
      "grad_norm": 1.1015625,
      "learning_rate": 6.147721087929315e-05,
      "loss": 0.9009,
      "step": 48849
    },
    {
      "epoch": 4.202873612664544,
      "grad_norm": 0.76953125,
      "learning_rate": 6.146422095723458e-05,
      "loss": 0.7531,
      "step": 48850
    },
    {
      "epoch": 4.202959648971866,
      "grad_norm": 0.84765625,
      "learning_rate": 6.145123231780802e-05,
      "loss": 0.5049,
      "step": 48851
    },
    {
      "epoch": 4.2030456852791875,
      "grad_norm": 1.265625,
      "learning_rate": 6.143824496105122e-05,
      "loss": 1.1497,
      "step": 48852
    },
    {
      "epoch": 4.20313172158651,
      "grad_norm": 1.140625,
      "learning_rate": 6.142525888700224e-05,
      "loss": 1.3226,
      "step": 48853
    },
    {
      "epoch": 4.203217757893831,
      "grad_norm": 1.0859375,
      "learning_rate": 6.141227409569915e-05,
      "loss": 0.9808,
      "step": 48854
    },
    {
      "epoch": 4.203303794201153,
      "grad_norm": 1.0234375,
      "learning_rate": 6.13992905871798e-05,
      "loss": 1.0351,
      "step": 48855
    },
    {
      "epoch": 4.203389830508475,
      "grad_norm": 1.1015625,
      "learning_rate": 6.138630836148218e-05,
      "loss": 0.7262,
      "step": 48856
    },
    {
      "epoch": 4.203475866815796,
      "grad_norm": 1.125,
      "learning_rate": 6.137332741864443e-05,
      "loss": 0.5358,
      "step": 48857
    },
    {
      "epoch": 4.203561903123118,
      "grad_norm": 0.9765625,
      "learning_rate": 6.13603477587043e-05,
      "loss": 0.8289,
      "step": 48858
    },
    {
      "epoch": 4.203647939430439,
      "grad_norm": 1.1796875,
      "learning_rate": 6.134736938169982e-05,
      "loss": 1.7462,
      "step": 48859
    },
    {
      "epoch": 4.203733975737761,
      "grad_norm": 0.89453125,
      "learning_rate": 6.133439228766896e-05,
      "loss": 0.7352,
      "step": 48860
    },
    {
      "epoch": 4.203820012045083,
      "grad_norm": 1.2734375,
      "learning_rate": 6.132141647664973e-05,
      "loss": 0.9067,
      "step": 48861
    },
    {
      "epoch": 4.203906048352405,
      "grad_norm": 1.0703125,
      "learning_rate": 6.130844194868013e-05,
      "loss": 0.9241,
      "step": 48862
    },
    {
      "epoch": 4.203992084659727,
      "grad_norm": 7.15625,
      "learning_rate": 6.129546870379788e-05,
      "loss": 0.6685,
      "step": 48863
    },
    {
      "epoch": 4.204078120967048,
      "grad_norm": 1.71875,
      "learning_rate": 6.128249674204106e-05,
      "loss": 1.7151,
      "step": 48864
    },
    {
      "epoch": 4.20416415727437,
      "grad_norm": 1.2109375,
      "learning_rate": 6.126952606344777e-05,
      "loss": 0.9066,
      "step": 48865
    },
    {
      "epoch": 4.204250193581691,
      "grad_norm": 191.0,
      "learning_rate": 6.125655666805563e-05,
      "loss": 0.9323,
      "step": 48866
    },
    {
      "epoch": 4.204336229889013,
      "grad_norm": 0.8671875,
      "learning_rate": 6.124358855590284e-05,
      "loss": 1.0218,
      "step": 48867
    },
    {
      "epoch": 4.204422266196335,
      "grad_norm": 1.1484375,
      "learning_rate": 6.123062172702714e-05,
      "loss": 0.7184,
      "step": 48868
    },
    {
      "epoch": 4.2045083025036565,
      "grad_norm": 1.25,
      "learning_rate": 6.121765618146652e-05,
      "loss": 1.7522,
      "step": 48869
    },
    {
      "epoch": 4.2045943388109785,
      "grad_norm": 1.1875,
      "learning_rate": 6.120469191925904e-05,
      "loss": 0.9303,
      "step": 48870
    },
    {
      "epoch": 4.2046803751183,
      "grad_norm": 31.625,
      "learning_rate": 6.11917289404424e-05,
      "loss": 0.8596,
      "step": 48871
    },
    {
      "epoch": 4.204766411425622,
      "grad_norm": 0.99609375,
      "learning_rate": 6.117876724505456e-05,
      "loss": 0.6271,
      "step": 48872
    },
    {
      "epoch": 4.204852447732943,
      "grad_norm": 1.1640625,
      "learning_rate": 6.11658068331336e-05,
      "loss": 1.6428,
      "step": 48873
    },
    {
      "epoch": 4.204938484040265,
      "grad_norm": 1.265625,
      "learning_rate": 6.115284770471724e-05,
      "loss": 1.0034,
      "step": 48874
    },
    {
      "epoch": 4.205024520347587,
      "grad_norm": 0.984375,
      "learning_rate": 6.113988985984342e-05,
      "loss": 1.6423,
      "step": 48875
    },
    {
      "epoch": 4.205110556654908,
      "grad_norm": 1.125,
      "learning_rate": 6.112693329855018e-05,
      "loss": 1.1666,
      "step": 48876
    },
    {
      "epoch": 4.20519659296223,
      "grad_norm": 120.0,
      "learning_rate": 6.111397802087515e-05,
      "loss": 0.9096,
      "step": 48877
    },
    {
      "epoch": 4.2052826292695515,
      "grad_norm": 1.25,
      "learning_rate": 6.110102402685647e-05,
      "loss": 1.1967,
      "step": 48878
    },
    {
      "epoch": 4.205368665576874,
      "grad_norm": 1.1875,
      "learning_rate": 6.1088071316532e-05,
      "loss": 1.6433,
      "step": 48879
    },
    {
      "epoch": 4.205454701884195,
      "grad_norm": 1.2578125,
      "learning_rate": 6.107511988993941e-05,
      "loss": 0.8738,
      "step": 48880
    },
    {
      "epoch": 4.205540738191517,
      "grad_norm": 1.109375,
      "learning_rate": 6.106216974711687e-05,
      "loss": 0.8271,
      "step": 48881
    },
    {
      "epoch": 4.205626774498839,
      "grad_norm": 3.9375,
      "learning_rate": 6.1049220888102e-05,
      "loss": 1.0257,
      "step": 48882
    },
    {
      "epoch": 4.20571281080616,
      "grad_norm": 1.203125,
      "learning_rate": 6.103627331293277e-05,
      "loss": 0.5613,
      "step": 48883
    },
    {
      "epoch": 4.205798847113482,
      "grad_norm": 1.6171875,
      "learning_rate": 6.1023327021647204e-05,
      "loss": 1.3732,
      "step": 48884
    },
    {
      "epoch": 4.205884883420803,
      "grad_norm": 0.8984375,
      "learning_rate": 6.101038201428294e-05,
      "loss": 0.9914,
      "step": 48885
    },
    {
      "epoch": 4.205970919728125,
      "grad_norm": 1.0234375,
      "learning_rate": 6.0997438290877894e-05,
      "loss": 0.8884,
      "step": 48886
    },
    {
      "epoch": 4.206056956035447,
      "grad_norm": 1.078125,
      "learning_rate": 6.098449585147009e-05,
      "loss": 0.9269,
      "step": 48887
    },
    {
      "epoch": 4.206142992342769,
      "grad_norm": 1.609375,
      "learning_rate": 6.097155469609711e-05,
      "loss": 1.5233,
      "step": 48888
    },
    {
      "epoch": 4.206229028650091,
      "grad_norm": 2040.0,
      "learning_rate": 6.095861482479698e-05,
      "loss": 1.2384,
      "step": 48889
    },
    {
      "epoch": 4.206315064957412,
      "grad_norm": 0.92578125,
      "learning_rate": 6.0945676237607605e-05,
      "loss": 0.6442,
      "step": 48890
    },
    {
      "epoch": 4.206401101264734,
      "grad_norm": 0.828125,
      "learning_rate": 6.093273893456663e-05,
      "loss": 0.6667,
      "step": 48891
    },
    {
      "epoch": 4.206487137572055,
      "grad_norm": 0.94921875,
      "learning_rate": 6.0919802915712084e-05,
      "loss": 0.6097,
      "step": 48892
    },
    {
      "epoch": 4.206573173879377,
      "grad_norm": 0.8671875,
      "learning_rate": 6.0906868181081655e-05,
      "loss": 0.8863,
      "step": 48893
    },
    {
      "epoch": 4.206659210186698,
      "grad_norm": 0.9609375,
      "learning_rate": 6.089393473071325e-05,
      "loss": 1.4037,
      "step": 48894
    },
    {
      "epoch": 4.2067452464940205,
      "grad_norm": 0.6328125,
      "learning_rate": 6.088100256464474e-05,
      "loss": 0.4332,
      "step": 48895
    },
    {
      "epoch": 4.2068312828013426,
      "grad_norm": 1.0078125,
      "learning_rate": 6.0868071682913695e-05,
      "loss": 0.7761,
      "step": 48896
    },
    {
      "epoch": 4.206917319108664,
      "grad_norm": 454.0,
      "learning_rate": 6.0855142085558365e-05,
      "loss": 1.1589,
      "step": 48897
    },
    {
      "epoch": 4.207003355415986,
      "grad_norm": 0.91015625,
      "learning_rate": 6.084221377261623e-05,
      "loss": 0.818,
      "step": 48898
    },
    {
      "epoch": 4.207089391723307,
      "grad_norm": 0.99609375,
      "learning_rate": 6.082928674412519e-05,
      "loss": 0.7352,
      "step": 48899
    },
    {
      "epoch": 4.207175428030629,
      "grad_norm": 1.0234375,
      "learning_rate": 6.081636100012317e-05,
      "loss": 0.8013,
      "step": 48900
    },
    {
      "epoch": 4.20726146433795,
      "grad_norm": 0.96875,
      "learning_rate": 6.0803436540647794e-05,
      "loss": 1.1726,
      "step": 48901
    },
    {
      "epoch": 4.207347500645272,
      "grad_norm": 1.078125,
      "learning_rate": 6.079051336573693e-05,
      "loss": 0.6957,
      "step": 48902
    },
    {
      "epoch": 4.207433536952594,
      "grad_norm": 0.92578125,
      "learning_rate": 6.0777591475428495e-05,
      "loss": 0.5878,
      "step": 48903
    },
    {
      "epoch": 4.2075195732599155,
      "grad_norm": 1.21875,
      "learning_rate": 6.076467086976012e-05,
      "loss": 1.0522,
      "step": 48904
    },
    {
      "epoch": 4.207605609567238,
      "grad_norm": 1.234375,
      "learning_rate": 6.075175154876961e-05,
      "loss": 1.7452,
      "step": 48905
    },
    {
      "epoch": 4.207691645874559,
      "grad_norm": 0.90234375,
      "learning_rate": 6.0738833512494925e-05,
      "loss": 1.1978,
      "step": 48906
    },
    {
      "epoch": 4.207777682181881,
      "grad_norm": 1.125,
      "learning_rate": 6.072591676097361e-05,
      "loss": 0.7043,
      "step": 48907
    },
    {
      "epoch": 4.207863718489202,
      "grad_norm": 1.3359375,
      "learning_rate": 6.071300129424362e-05,
      "loss": 0.7718,
      "step": 48908
    },
    {
      "epoch": 4.207949754796524,
      "grad_norm": 1.0234375,
      "learning_rate": 6.070008711234259e-05,
      "loss": 1.2561,
      "step": 48909
    },
    {
      "epoch": 4.208035791103846,
      "grad_norm": 2.3125,
      "learning_rate": 6.0687174215308395e-05,
      "loss": 1.1198,
      "step": 48910
    },
    {
      "epoch": 4.208121827411167,
      "grad_norm": 1.171875,
      "learning_rate": 6.067426260317882e-05,
      "loss": 1.2976,
      "step": 48911
    },
    {
      "epoch": 4.208207863718489,
      "grad_norm": 0.859375,
      "learning_rate": 6.066135227599157e-05,
      "loss": 0.7173,
      "step": 48912
    },
    {
      "epoch": 4.208293900025811,
      "grad_norm": 1.3125,
      "learning_rate": 6.0648443233784324e-05,
      "loss": 1.6453,
      "step": 48913
    },
    {
      "epoch": 4.208379936333133,
      "grad_norm": 0.6328125,
      "learning_rate": 6.063553547659506e-05,
      "loss": 0.6348,
      "step": 48914
    },
    {
      "epoch": 4.208465972640454,
      "grad_norm": 0.90625,
      "learning_rate": 6.062262900446136e-05,
      "loss": 0.7225,
      "step": 48915
    },
    {
      "epoch": 4.208552008947776,
      "grad_norm": 0.94140625,
      "learning_rate": 6.060972381742091e-05,
      "loss": 0.8037,
      "step": 48916
    },
    {
      "epoch": 4.208638045255098,
      "grad_norm": 1.1484375,
      "learning_rate": 6.059681991551175e-05,
      "loss": 0.4976,
      "step": 48917
    },
    {
      "epoch": 4.208724081562419,
      "grad_norm": 1.2109375,
      "learning_rate": 6.058391729877133e-05,
      "loss": 1.4908,
      "step": 48918
    },
    {
      "epoch": 4.208810117869741,
      "grad_norm": 1.1328125,
      "learning_rate": 6.057101596723746e-05,
      "loss": 1.5541,
      "step": 48919
    },
    {
      "epoch": 4.208896154177062,
      "grad_norm": 6.1875,
      "learning_rate": 6.055811592094801e-05,
      "loss": 1.3959,
      "step": 48920
    },
    {
      "epoch": 4.2089821904843845,
      "grad_norm": 0.97265625,
      "learning_rate": 6.054521715994049e-05,
      "loss": 0.9294,
      "step": 48921
    },
    {
      "epoch": 4.209068226791706,
      "grad_norm": 1.1796875,
      "learning_rate": 6.053231968425282e-05,
      "loss": 0.8321,
      "step": 48922
    },
    {
      "epoch": 4.209154263099028,
      "grad_norm": 0.98828125,
      "learning_rate": 6.051942349392259e-05,
      "loss": 0.9292,
      "step": 48923
    },
    {
      "epoch": 4.20924029940635,
      "grad_norm": 0.875,
      "learning_rate": 6.050652858898758e-05,
      "loss": 1.0426,
      "step": 48924
    },
    {
      "epoch": 4.209326335713671,
      "grad_norm": 0.76953125,
      "learning_rate": 6.0493634969485566e-05,
      "loss": 0.5265,
      "step": 48925
    },
    {
      "epoch": 4.209412372020993,
      "grad_norm": 1.453125,
      "learning_rate": 6.048074263545411e-05,
      "loss": 0.7018,
      "step": 48926
    },
    {
      "epoch": 4.209498408328314,
      "grad_norm": 1.5859375,
      "learning_rate": 6.0467851586930966e-05,
      "loss": 0.5011,
      "step": 48927
    },
    {
      "epoch": 4.209584444635636,
      "grad_norm": 2.921875,
      "learning_rate": 6.045496182395399e-05,
      "loss": 1.2552,
      "step": 48928
    },
    {
      "epoch": 4.2096704809429575,
      "grad_norm": 1.109375,
      "learning_rate": 6.044207334656071e-05,
      "loss": 1.2574,
      "step": 48929
    },
    {
      "epoch": 4.2097565172502796,
      "grad_norm": 0.78515625,
      "learning_rate": 6.0429186154788816e-05,
      "loss": 0.6429,
      "step": 48930
    },
    {
      "epoch": 4.209842553557602,
      "grad_norm": 1.015625,
      "learning_rate": 6.041630024867617e-05,
      "loss": 1.5204,
      "step": 48931
    },
    {
      "epoch": 4.209928589864923,
      "grad_norm": 166.0,
      "learning_rate": 6.0403415628260294e-05,
      "loss": 1.1162,
      "step": 48932
    },
    {
      "epoch": 4.210014626172245,
      "grad_norm": 0.8515625,
      "learning_rate": 6.039053229357888e-05,
      "loss": 1.0202,
      "step": 48933
    },
    {
      "epoch": 4.210100662479566,
      "grad_norm": 1.234375,
      "learning_rate": 6.037765024466968e-05,
      "loss": 0.9544,
      "step": 48934
    },
    {
      "epoch": 4.210186698786888,
      "grad_norm": 1.2890625,
      "learning_rate": 6.036476948157033e-05,
      "loss": 1.0708,
      "step": 48935
    },
    {
      "epoch": 4.21027273509421,
      "grad_norm": 1.46875,
      "learning_rate": 6.0351890004318675e-05,
      "loss": 1.3171,
      "step": 48936
    },
    {
      "epoch": 4.210358771401531,
      "grad_norm": 476.0,
      "learning_rate": 6.033901181295209e-05,
      "loss": 1.5674,
      "step": 48937
    },
    {
      "epoch": 4.210444807708853,
      "grad_norm": 0.859375,
      "learning_rate": 6.032613490750849e-05,
      "loss": 1.2523,
      "step": 48938
    },
    {
      "epoch": 4.210530844016175,
      "grad_norm": 1.1796875,
      "learning_rate": 6.031325928802533e-05,
      "loss": 1.4889,
      "step": 48939
    },
    {
      "epoch": 4.210616880323497,
      "grad_norm": 1.03125,
      "learning_rate": 6.0300384954540374e-05,
      "loss": 1.6766,
      "step": 48940
    },
    {
      "epoch": 4.210702916630818,
      "grad_norm": 1.1328125,
      "learning_rate": 6.028751190709142e-05,
      "loss": 1.0086,
      "step": 48941
    },
    {
      "epoch": 4.21078895293814,
      "grad_norm": 1.28125,
      "learning_rate": 6.027464014571582e-05,
      "loss": 0.6926,
      "step": 48942
    },
    {
      "epoch": 4.210874989245462,
      "grad_norm": 1.265625,
      "learning_rate": 6.026176967045144e-05,
      "loss": 1.8968,
      "step": 48943
    },
    {
      "epoch": 4.210961025552783,
      "grad_norm": 1.7421875,
      "learning_rate": 6.024890048133591e-05,
      "loss": 1.3228,
      "step": 48944
    },
    {
      "epoch": 4.211047061860105,
      "grad_norm": 0.75390625,
      "learning_rate": 6.0236032578406705e-05,
      "loss": 0.4725,
      "step": 48945
    },
    {
      "epoch": 4.211133098167426,
      "grad_norm": 48128.0,
      "learning_rate": 6.022316596170158e-05,
      "loss": 1.1735,
      "step": 48946
    },
    {
      "epoch": 4.2112191344747485,
      "grad_norm": 0.96875,
      "learning_rate": 6.021030063125832e-05,
      "loss": 0.9832,
      "step": 48947
    },
    {
      "epoch": 4.21130517078207,
      "grad_norm": 16.125,
      "learning_rate": 6.019743658711424e-05,
      "loss": 0.8513,
      "step": 48948
    },
    {
      "epoch": 4.211391207089392,
      "grad_norm": 0.99609375,
      "learning_rate": 6.018457382930726e-05,
      "loss": 1.0672,
      "step": 48949
    },
    {
      "epoch": 4.211477243396714,
      "grad_norm": 1.140625,
      "learning_rate": 6.017171235787472e-05,
      "loss": 1.3268,
      "step": 48950
    },
    {
      "epoch": 4.211563279704035,
      "grad_norm": 1.0625,
      "learning_rate": 6.0158852172854386e-05,
      "loss": 0.5016,
      "step": 48951
    },
    {
      "epoch": 4.211649316011357,
      "grad_norm": 1.0546875,
      "learning_rate": 6.0145993274284e-05,
      "loss": 1.3117,
      "step": 48952
    },
    {
      "epoch": 4.211735352318678,
      "grad_norm": 0.9765625,
      "learning_rate": 6.013313566220091e-05,
      "loss": 0.7887,
      "step": 48953
    },
    {
      "epoch": 4.211821388626,
      "grad_norm": 1.109375,
      "learning_rate": 6.012027933664283e-05,
      "loss": 0.6309,
      "step": 48954
    },
    {
      "epoch": 4.2119074249333215,
      "grad_norm": 1.21875,
      "learning_rate": 6.0107424297647485e-05,
      "loss": 0.6915,
      "step": 48955
    },
    {
      "epoch": 4.2119934612406436,
      "grad_norm": 1.2109375,
      "learning_rate": 6.00945705452523e-05,
      "loss": 0.6973,
      "step": 48956
    },
    {
      "epoch": 4.212079497547966,
      "grad_norm": 1.09375,
      "learning_rate": 6.0081718079494917e-05,
      "loss": 1.2804,
      "step": 48957
    },
    {
      "epoch": 4.212165533855287,
      "grad_norm": 1.171875,
      "learning_rate": 6.0068866900413076e-05,
      "loss": 1.3432,
      "step": 48958
    },
    {
      "epoch": 4.212251570162609,
      "grad_norm": 1.6953125,
      "learning_rate": 6.005601700804414e-05,
      "loss": 0.7413,
      "step": 48959
    },
    {
      "epoch": 4.21233760646993,
      "grad_norm": 11.125,
      "learning_rate": 6.004316840242574e-05,
      "loss": 1.5433,
      "step": 48960
    },
    {
      "epoch": 4.212423642777252,
      "grad_norm": 1.578125,
      "learning_rate": 6.003032108359563e-05,
      "loss": 1.8196,
      "step": 48961
    },
    {
      "epoch": 4.212509679084573,
      "grad_norm": 0.9609375,
      "learning_rate": 6.001747505159116e-05,
      "loss": 1.1042,
      "step": 48962
    },
    {
      "epoch": 4.212595715391895,
      "grad_norm": 1.1875,
      "learning_rate": 6.000463030645015e-05,
      "loss": 0.8706,
      "step": 48963
    },
    {
      "epoch": 4.212681751699217,
      "grad_norm": 1.328125,
      "learning_rate": 5.999178684820988e-05,
      "loss": 1.1857,
      "step": 48964
    },
    {
      "epoch": 4.212767788006539,
      "grad_norm": 0.98046875,
      "learning_rate": 5.997894467690806e-05,
      "loss": 0.7924,
      "step": 48965
    },
    {
      "epoch": 4.212853824313861,
      "grad_norm": 1.046875,
      "learning_rate": 5.9966103792582325e-05,
      "loss": 1.6442,
      "step": 48966
    },
    {
      "epoch": 4.212939860621182,
      "grad_norm": 2.5,
      "learning_rate": 5.9953264195270086e-05,
      "loss": 1.1347,
      "step": 48967
    },
    {
      "epoch": 4.213025896928504,
      "grad_norm": 1.140625,
      "learning_rate": 5.994042588500897e-05,
      "loss": 1.4576,
      "step": 48968
    },
    {
      "epoch": 4.213111933235825,
      "grad_norm": 0.77734375,
      "learning_rate": 5.9927588861836636e-05,
      "loss": 0.5791,
      "step": 48969
    },
    {
      "epoch": 4.213197969543147,
      "grad_norm": 51.75,
      "learning_rate": 5.991475312579026e-05,
      "loss": 0.631,
      "step": 48970
    },
    {
      "epoch": 4.213284005850469,
      "grad_norm": 1.4765625,
      "learning_rate": 5.990191867690786e-05,
      "loss": 0.9239,
      "step": 48971
    },
    {
      "epoch": 4.21337004215779,
      "grad_norm": 1.2578125,
      "learning_rate": 5.988908551522665e-05,
      "loss": 1.3343,
      "step": 48972
    },
    {
      "epoch": 4.2134560784651125,
      "grad_norm": 0.90234375,
      "learning_rate": 5.987625364078431e-05,
      "loss": 0.8121,
      "step": 48973
    },
    {
      "epoch": 4.213542114772434,
      "grad_norm": 1.171875,
      "learning_rate": 5.986342305361836e-05,
      "loss": 1.4042,
      "step": 48974
    },
    {
      "epoch": 4.213628151079756,
      "grad_norm": 1.421875,
      "learning_rate": 5.9850593753766224e-05,
      "loss": 1.5437,
      "step": 48975
    },
    {
      "epoch": 4.213714187387077,
      "grad_norm": 1.203125,
      "learning_rate": 5.983776574126554e-05,
      "loss": 0.936,
      "step": 48976
    },
    {
      "epoch": 4.213800223694399,
      "grad_norm": 0.9921875,
      "learning_rate": 5.9824939016153823e-05,
      "loss": 1.3796,
      "step": 48977
    },
    {
      "epoch": 4.213886260001721,
      "grad_norm": 10.5,
      "learning_rate": 5.98121135784685e-05,
      "loss": 0.8039,
      "step": 48978
    },
    {
      "epoch": 4.213972296309042,
      "grad_norm": 1.140625,
      "learning_rate": 5.979928942824719e-05,
      "loss": 1.0304,
      "step": 48979
    },
    {
      "epoch": 4.214058332616364,
      "grad_norm": 3.609375,
      "learning_rate": 5.9786466565527274e-05,
      "loss": 1.521,
      "step": 48980
    },
    {
      "epoch": 4.2141443689236855,
      "grad_norm": 1.1015625,
      "learning_rate": 5.977364499034632e-05,
      "loss": 1.2796,
      "step": 48981
    },
    {
      "epoch": 4.214230405231008,
      "grad_norm": 1.1640625,
      "learning_rate": 5.976082470274191e-05,
      "loss": 0.7695,
      "step": 48982
    },
    {
      "epoch": 4.21431644153833,
      "grad_norm": 1.296875,
      "learning_rate": 5.9748005702751406e-05,
      "loss": 1.1412,
      "step": 48983
    },
    {
      "epoch": 4.214402477845651,
      "grad_norm": 1.125,
      "learning_rate": 5.973518799041233e-05,
      "loss": 0.7213,
      "step": 48984
    },
    {
      "epoch": 4.214488514152973,
      "grad_norm": 1.0234375,
      "learning_rate": 5.9722371565762325e-05,
      "loss": 1.7666,
      "step": 48985
    },
    {
      "epoch": 4.214574550460294,
      "grad_norm": 1.234375,
      "learning_rate": 5.970955642883863e-05,
      "loss": 1.2907,
      "step": 48986
    },
    {
      "epoch": 4.214660586767616,
      "grad_norm": 1.4140625,
      "learning_rate": 5.9696742579678885e-05,
      "loss": 1.5229,
      "step": 48987
    },
    {
      "epoch": 4.214746623074937,
      "grad_norm": 1.4375,
      "learning_rate": 5.968393001832062e-05,
      "loss": 1.6215,
      "step": 48988
    },
    {
      "epoch": 4.214832659382259,
      "grad_norm": 1560.0,
      "learning_rate": 5.967111874480108e-05,
      "loss": 1.6609,
      "step": 48989
    },
    {
      "epoch": 4.214918695689581,
      "grad_norm": 1.21875,
      "learning_rate": 5.9658308759157954e-05,
      "loss": 1.3922,
      "step": 48990
    },
    {
      "epoch": 4.215004731996903,
      "grad_norm": 4.5,
      "learning_rate": 5.964550006142866e-05,
      "loss": 0.7443,
      "step": 48991
    },
    {
      "epoch": 4.215090768304225,
      "grad_norm": 1.015625,
      "learning_rate": 5.963269265165061e-05,
      "loss": 1.0019,
      "step": 48992
    },
    {
      "epoch": 4.215176804611546,
      "grad_norm": 2.546875,
      "learning_rate": 5.9619886529861335e-05,
      "loss": 0.8387,
      "step": 48993
    },
    {
      "epoch": 4.215262840918868,
      "grad_norm": 0.99609375,
      "learning_rate": 5.960708169609813e-05,
      "loss": 1.0392,
      "step": 48994
    },
    {
      "epoch": 4.215348877226189,
      "grad_norm": 1.1015625,
      "learning_rate": 5.959427815039859e-05,
      "loss": 1.3698,
      "step": 48995
    },
    {
      "epoch": 4.215434913533511,
      "grad_norm": 0.8046875,
      "learning_rate": 5.958147589280022e-05,
      "loss": 0.4883,
      "step": 48996
    },
    {
      "epoch": 4.215520949840833,
      "grad_norm": 1.28125,
      "learning_rate": 5.956867492334028e-05,
      "loss": 0.9355,
      "step": 48997
    },
    {
      "epoch": 4.215606986148154,
      "grad_norm": 1.03125,
      "learning_rate": 5.95558752420563e-05,
      "loss": 0.7302,
      "step": 48998
    },
    {
      "epoch": 4.2156930224554765,
      "grad_norm": 0.8671875,
      "learning_rate": 5.9543076848985856e-05,
      "loss": 0.9318,
      "step": 48999
    },
    {
      "epoch": 4.215779058762798,
      "grad_norm": 1.2265625,
      "learning_rate": 5.953027974416614e-05,
      "loss": 0.7229,
      "step": 49000
    },
    {
      "epoch": 4.21586509507012,
      "grad_norm": 1.3671875,
      "learning_rate": 5.951748392763462e-05,
      "loss": 0.658,
      "step": 49001
    },
    {
      "epoch": 4.215951131377441,
      "grad_norm": 1.5234375,
      "learning_rate": 5.950468939942899e-05,
      "loss": 1.1271,
      "step": 49002
    },
    {
      "epoch": 4.216037167684763,
      "grad_norm": 0.8984375,
      "learning_rate": 5.949189615958628e-05,
      "loss": 1.4057,
      "step": 49003
    },
    {
      "epoch": 4.216123203992085,
      "grad_norm": 1.265625,
      "learning_rate": 5.947910420814429e-05,
      "loss": 1.0443,
      "step": 49004
    },
    {
      "epoch": 4.216209240299406,
      "grad_norm": 0.984375,
      "learning_rate": 5.94663135451401e-05,
      "loss": 0.8815,
      "step": 49005
    },
    {
      "epoch": 4.216295276606728,
      "grad_norm": 1.3359375,
      "learning_rate": 5.945352417061123e-05,
      "loss": 0.9568,
      "step": 49006
    },
    {
      "epoch": 4.2163813129140495,
      "grad_norm": 1.390625,
      "learning_rate": 5.944073608459516e-05,
      "loss": 0.7666,
      "step": 49007
    },
    {
      "epoch": 4.216467349221372,
      "grad_norm": 1.0390625,
      "learning_rate": 5.942794928712925e-05,
      "loss": 0.9052,
      "step": 49008
    },
    {
      "epoch": 4.216553385528693,
      "grad_norm": 0.875,
      "learning_rate": 5.941516377825101e-05,
      "loss": 1.052,
      "step": 49009
    },
    {
      "epoch": 4.216639421836015,
      "grad_norm": 1.0234375,
      "learning_rate": 5.94023795579976e-05,
      "loss": 1.5803,
      "step": 49010
    },
    {
      "epoch": 4.216725458143337,
      "grad_norm": 1.15625,
      "learning_rate": 5.938959662640653e-05,
      "loss": 1.1543,
      "step": 49011
    },
    {
      "epoch": 4.216811494450658,
      "grad_norm": 1.0625,
      "learning_rate": 5.937681498351533e-05,
      "loss": 0.9932,
      "step": 49012
    },
    {
      "epoch": 4.21689753075798,
      "grad_norm": 310.0,
      "learning_rate": 5.936403462936113e-05,
      "loss": 0.6796,
      "step": 49013
    },
    {
      "epoch": 4.216983567065301,
      "grad_norm": 1.3359375,
      "learning_rate": 5.9351255563981464e-05,
      "loss": 1.5378,
      "step": 49014
    },
    {
      "epoch": 4.217069603372623,
      "grad_norm": 0.8515625,
      "learning_rate": 5.9338477787413745e-05,
      "loss": 0.5984,
      "step": 49015
    },
    {
      "epoch": 4.217155639679945,
      "grad_norm": 0.64453125,
      "learning_rate": 5.932570129969517e-05,
      "loss": 0.4756,
      "step": 49016
    },
    {
      "epoch": 4.217241675987267,
      "grad_norm": 0.8359375,
      "learning_rate": 5.9312926100863196e-05,
      "loss": 0.4543,
      "step": 49017
    },
    {
      "epoch": 4.217327712294589,
      "grad_norm": 290.0,
      "learning_rate": 5.930015219095536e-05,
      "loss": 1.47,
      "step": 49018
    },
    {
      "epoch": 4.21741374860191,
      "grad_norm": 1.2421875,
      "learning_rate": 5.9287379570008735e-05,
      "loss": 0.9226,
      "step": 49019
    },
    {
      "epoch": 4.217499784909232,
      "grad_norm": 1.1875,
      "learning_rate": 5.927460823806091e-05,
      "loss": 1.2749,
      "step": 49020
    },
    {
      "epoch": 4.217585821216553,
      "grad_norm": 0.890625,
      "learning_rate": 5.926183819514902e-05,
      "loss": 0.7209,
      "step": 49021
    },
    {
      "epoch": 4.217671857523875,
      "grad_norm": 1.0859375,
      "learning_rate": 5.924906944131053e-05,
      "loss": 1.0953,
      "step": 49022
    },
    {
      "epoch": 4.217757893831196,
      "grad_norm": 1.5859375,
      "learning_rate": 5.923630197658292e-05,
      "loss": 1.3958,
      "step": 49023
    },
    {
      "epoch": 4.217843930138518,
      "grad_norm": 1.25,
      "learning_rate": 5.9223535801003316e-05,
      "loss": 1.3379,
      "step": 49024
    },
    {
      "epoch": 4.2179299664458405,
      "grad_norm": 0.92578125,
      "learning_rate": 5.921077091460914e-05,
      "loss": 1.2124,
      "step": 49025
    },
    {
      "epoch": 4.218016002753162,
      "grad_norm": 1.203125,
      "learning_rate": 5.91980073174378e-05,
      "loss": 1.436,
      "step": 49026
    },
    {
      "epoch": 4.218102039060484,
      "grad_norm": 0.9765625,
      "learning_rate": 5.918524500952649e-05,
      "loss": 1.3039,
      "step": 49027
    },
    {
      "epoch": 4.218188075367805,
      "grad_norm": 1.078125,
      "learning_rate": 5.9172483990912576e-05,
      "loss": 0.9136,
      "step": 49028
    },
    {
      "epoch": 4.218274111675127,
      "grad_norm": 0.99609375,
      "learning_rate": 5.9159724261633517e-05,
      "loss": 0.941,
      "step": 49029
    },
    {
      "epoch": 4.218360147982448,
      "grad_norm": 2.71875,
      "learning_rate": 5.914696582172641e-05,
      "loss": 1.3514,
      "step": 49030
    },
    {
      "epoch": 4.21844618428977,
      "grad_norm": 1.265625,
      "learning_rate": 5.913420867122871e-05,
      "loss": 0.9335,
      "step": 49031
    },
    {
      "epoch": 4.218532220597092,
      "grad_norm": 1.4375,
      "learning_rate": 5.9121452810177834e-05,
      "loss": 1.0233,
      "step": 49032
    },
    {
      "epoch": 4.2186182569044135,
      "grad_norm": 1.234375,
      "learning_rate": 5.9108698238610814e-05,
      "loss": 1.242,
      "step": 49033
    },
    {
      "epoch": 4.218704293211736,
      "grad_norm": 0.83203125,
      "learning_rate": 5.909594495656523e-05,
      "loss": 0.7347,
      "step": 49034
    },
    {
      "epoch": 4.218790329519057,
      "grad_norm": 1.1015625,
      "learning_rate": 5.908319296407816e-05,
      "loss": 1.2404,
      "step": 49035
    },
    {
      "epoch": 4.218876365826379,
      "grad_norm": 4.1875,
      "learning_rate": 5.9070442261186964e-05,
      "loss": 1.2505,
      "step": 49036
    },
    {
      "epoch": 4.2189624021337,
      "grad_norm": 1.4453125,
      "learning_rate": 5.9057692847929115e-05,
      "loss": 1.0712,
      "step": 49037
    },
    {
      "epoch": 4.219048438441022,
      "grad_norm": 1.03125,
      "learning_rate": 5.904494472434163e-05,
      "loss": 0.707,
      "step": 49038
    },
    {
      "epoch": 4.219134474748344,
      "grad_norm": 0.8125,
      "learning_rate": 5.9032197890461884e-05,
      "loss": 1.1828,
      "step": 49039
    },
    {
      "epoch": 4.219220511055665,
      "grad_norm": 1.3046875,
      "learning_rate": 5.901945234632733e-05,
      "loss": 0.918,
      "step": 49040
    },
    {
      "epoch": 4.219306547362987,
      "grad_norm": 7.375,
      "learning_rate": 5.9006708091975005e-05,
      "loss": 1.2973,
      "step": 49041
    },
    {
      "epoch": 4.219392583670309,
      "grad_norm": 1.03125,
      "learning_rate": 5.8993965127442316e-05,
      "loss": 0.8407,
      "step": 49042
    },
    {
      "epoch": 4.219478619977631,
      "grad_norm": 0.91015625,
      "learning_rate": 5.8981223452766574e-05,
      "loss": 0.6493,
      "step": 49043
    },
    {
      "epoch": 4.219564656284952,
      "grad_norm": 1.359375,
      "learning_rate": 5.89684830679848e-05,
      "loss": 1.0352,
      "step": 49044
    },
    {
      "epoch": 4.219650692592274,
      "grad_norm": 1.328125,
      "learning_rate": 5.895574397313463e-05,
      "loss": 1.6252,
      "step": 49045
    },
    {
      "epoch": 4.219736728899596,
      "grad_norm": 3.1875,
      "learning_rate": 5.8943006168253043e-05,
      "loss": 1.4496,
      "step": 49046
    },
    {
      "epoch": 4.219822765206917,
      "grad_norm": 0.96484375,
      "learning_rate": 5.8930269653377386e-05,
      "loss": 0.8436,
      "step": 49047
    },
    {
      "epoch": 4.219908801514239,
      "grad_norm": 1.0390625,
      "learning_rate": 5.8917534428544975e-05,
      "loss": 1.6523,
      "step": 49048
    },
    {
      "epoch": 4.21999483782156,
      "grad_norm": 1.0078125,
      "learning_rate": 5.890480049379293e-05,
      "loss": 0.7735,
      "step": 49049
    },
    {
      "epoch": 4.220080874128882,
      "grad_norm": 0.92578125,
      "learning_rate": 5.889206784915863e-05,
      "loss": 0.9404,
      "step": 49050
    },
    {
      "epoch": 4.220166910436204,
      "grad_norm": 1.140625,
      "learning_rate": 5.8879336494679136e-05,
      "loss": 0.9315,
      "step": 49051
    },
    {
      "epoch": 4.220252946743526,
      "grad_norm": 1.0546875,
      "learning_rate": 5.886660643039182e-05,
      "loss": 1.0505,
      "step": 49052
    },
    {
      "epoch": 4.220338983050848,
      "grad_norm": 0.7578125,
      "learning_rate": 5.8853877656333934e-05,
      "loss": 0.5394,
      "step": 49053
    },
    {
      "epoch": 4.220425019358169,
      "grad_norm": 6.5,
      "learning_rate": 5.88411501725426e-05,
      "loss": 0.9559,
      "step": 49054
    },
    {
      "epoch": 4.220511055665491,
      "grad_norm": 1.1484375,
      "learning_rate": 5.882842397905508e-05,
      "loss": 1.6792,
      "step": 49055
    },
    {
      "epoch": 4.220597091972812,
      "grad_norm": 0.94140625,
      "learning_rate": 5.8815699075908725e-05,
      "loss": 0.8875,
      "step": 49056
    },
    {
      "epoch": 4.220683128280134,
      "grad_norm": 1.0625,
      "learning_rate": 5.8802975463140564e-05,
      "loss": 1.3145,
      "step": 49057
    },
    {
      "epoch": 4.220769164587456,
      "grad_norm": 0.921875,
      "learning_rate": 5.87902531407879e-05,
      "loss": 0.6757,
      "step": 49058
    },
    {
      "epoch": 4.2208552008947775,
      "grad_norm": 0.8984375,
      "learning_rate": 5.8777532108887974e-05,
      "loss": 0.8281,
      "step": 49059
    },
    {
      "epoch": 4.2209412372021,
      "grad_norm": 4.96875,
      "learning_rate": 5.876481236747788e-05,
      "loss": 1.0754,
      "step": 49060
    },
    {
      "epoch": 4.221027273509421,
      "grad_norm": 11.25,
      "learning_rate": 5.875209391659503e-05,
      "loss": 1.2745,
      "step": 49061
    },
    {
      "epoch": 4.221113309816743,
      "grad_norm": 0.8125,
      "learning_rate": 5.873937675627633e-05,
      "loss": 0.7071,
      "step": 49062
    },
    {
      "epoch": 4.221199346124064,
      "grad_norm": 1.171875,
      "learning_rate": 5.87266608865592e-05,
      "loss": 1.9672,
      "step": 49063
    },
    {
      "epoch": 4.221285382431386,
      "grad_norm": 20.75,
      "learning_rate": 5.871394630748078e-05,
      "loss": 0.6305,
      "step": 49064
    },
    {
      "epoch": 4.221371418738708,
      "grad_norm": 1.046875,
      "learning_rate": 5.87012330190782e-05,
      "loss": 1.593,
      "step": 49065
    },
    {
      "epoch": 4.221457455046029,
      "grad_norm": 1.1796875,
      "learning_rate": 5.868852102138866e-05,
      "loss": 1.198,
      "step": 49066
    },
    {
      "epoch": 4.221543491353351,
      "grad_norm": 1.265625,
      "learning_rate": 5.8675810314449515e-05,
      "loss": 1.9009,
      "step": 49067
    },
    {
      "epoch": 4.221629527660673,
      "grad_norm": 175.0,
      "learning_rate": 5.866310089829763e-05,
      "loss": 0.9144,
      "step": 49068
    },
    {
      "epoch": 4.221715563967995,
      "grad_norm": 0.953125,
      "learning_rate": 5.865039277297035e-05,
      "loss": 0.9008,
      "step": 49069
    },
    {
      "epoch": 4.221801600275316,
      "grad_norm": 1.09375,
      "learning_rate": 5.863768593850499e-05,
      "loss": 1.3212,
      "step": 49070
    },
    {
      "epoch": 4.221887636582638,
      "grad_norm": 1.078125,
      "learning_rate": 5.862498039493841e-05,
      "loss": 1.3198,
      "step": 49071
    },
    {
      "epoch": 4.22197367288996,
      "grad_norm": 0.94140625,
      "learning_rate": 5.861227614230791e-05,
      "loss": 0.808,
      "step": 49072
    },
    {
      "epoch": 4.222059709197281,
      "grad_norm": 1.1640625,
      "learning_rate": 5.859957318065079e-05,
      "loss": 1.0289,
      "step": 49073
    },
    {
      "epoch": 4.222145745504603,
      "grad_norm": 1.0390625,
      "learning_rate": 5.858687151000397e-05,
      "loss": 0.4889,
      "step": 49074
    },
    {
      "epoch": 4.222231781811924,
      "grad_norm": 0.91015625,
      "learning_rate": 5.8574171130404754e-05,
      "loss": 0.5215,
      "step": 49075
    },
    {
      "epoch": 4.2223178181192464,
      "grad_norm": 117.5,
      "learning_rate": 5.8561472041890165e-05,
      "loss": 0.6912,
      "step": 49076
    },
    {
      "epoch": 4.222403854426568,
      "grad_norm": 0.87109375,
      "learning_rate": 5.854877424449745e-05,
      "loss": 1.0326,
      "step": 49077
    },
    {
      "epoch": 4.22248989073389,
      "grad_norm": 1.09375,
      "learning_rate": 5.8536077738263746e-05,
      "loss": 1.0067,
      "step": 49078
    },
    {
      "epoch": 4.222575927041212,
      "grad_norm": 0.99609375,
      "learning_rate": 5.852338252322609e-05,
      "loss": 1.4411,
      "step": 49079
    },
    {
      "epoch": 4.222661963348533,
      "grad_norm": 1.0703125,
      "learning_rate": 5.851068859942166e-05,
      "loss": 1.166,
      "step": 49080
    },
    {
      "epoch": 4.222747999655855,
      "grad_norm": 1.4140625,
      "learning_rate": 5.849799596688765e-05,
      "loss": 1.1321,
      "step": 49081
    },
    {
      "epoch": 4.222834035963176,
      "grad_norm": 1.125,
      "learning_rate": 5.8485304625661095e-05,
      "loss": 0.9909,
      "step": 49082
    },
    {
      "epoch": 4.222920072270498,
      "grad_norm": 5.0625,
      "learning_rate": 5.8472614575779236e-05,
      "loss": 1.1093,
      "step": 49083
    },
    {
      "epoch": 4.223006108577819,
      "grad_norm": 1.234375,
      "learning_rate": 5.8459925817279046e-05,
      "loss": 0.6636,
      "step": 49084
    },
    {
      "epoch": 4.2230921448851415,
      "grad_norm": 1.0625,
      "learning_rate": 5.844723835019766e-05,
      "loss": 0.7735,
      "step": 49085
    },
    {
      "epoch": 4.223178181192464,
      "grad_norm": 1.2890625,
      "learning_rate": 5.8434552174572386e-05,
      "loss": 1.1947,
      "step": 49086
    },
    {
      "epoch": 4.223264217499785,
      "grad_norm": 1.296875,
      "learning_rate": 5.842186729044002e-05,
      "loss": 1.2975,
      "step": 49087
    },
    {
      "epoch": 4.223350253807107,
      "grad_norm": 1.6953125,
      "learning_rate": 5.840918369783782e-05,
      "loss": 1.324,
      "step": 49088
    },
    {
      "epoch": 4.223436290114428,
      "grad_norm": 1.234375,
      "learning_rate": 5.839650139680297e-05,
      "loss": 0.9501,
      "step": 49089
    },
    {
      "epoch": 4.22352232642175,
      "grad_norm": 1.015625,
      "learning_rate": 5.8383820387372334e-05,
      "loss": 1.1582,
      "step": 49090
    },
    {
      "epoch": 4.223608362729071,
      "grad_norm": 1.390625,
      "learning_rate": 5.837114066958327e-05,
      "loss": 0.6317,
      "step": 49091
    },
    {
      "epoch": 4.223694399036393,
      "grad_norm": 1.3125,
      "learning_rate": 5.835846224347263e-05,
      "loss": 0.9901,
      "step": 49092
    },
    {
      "epoch": 4.223780435343715,
      "grad_norm": 1.34375,
      "learning_rate": 5.834578510907762e-05,
      "loss": 1.3892,
      "step": 49093
    },
    {
      "epoch": 4.223866471651037,
      "grad_norm": 0.8671875,
      "learning_rate": 5.833310926643537e-05,
      "loss": 1.7402,
      "step": 49094
    },
    {
      "epoch": 4.223952507958359,
      "grad_norm": 0.9140625,
      "learning_rate": 5.8320434715582735e-05,
      "loss": 0.9239,
      "step": 49095
    },
    {
      "epoch": 4.22403854426568,
      "grad_norm": 1.28125,
      "learning_rate": 5.830776145655698e-05,
      "loss": 1.1164,
      "step": 49096
    },
    {
      "epoch": 4.224124580573002,
      "grad_norm": 0.921875,
      "learning_rate": 5.829508948939516e-05,
      "loss": 1.9208,
      "step": 49097
    },
    {
      "epoch": 4.224210616880324,
      "grad_norm": 1.1484375,
      "learning_rate": 5.828241881413426e-05,
      "loss": 1.0685,
      "step": 49098
    },
    {
      "epoch": 4.224296653187645,
      "grad_norm": 1.1328125,
      "learning_rate": 5.8269749430811314e-05,
      "loss": 1.4887,
      "step": 49099
    },
    {
      "epoch": 4.224382689494967,
      "grad_norm": 7.75,
      "learning_rate": 5.8257081339463615e-05,
      "loss": 1.5067,
      "step": 49100
    },
    {
      "epoch": 4.224468725802288,
      "grad_norm": 1.1015625,
      "learning_rate": 5.824441454012791e-05,
      "loss": 1.4225,
      "step": 49101
    },
    {
      "epoch": 4.2245547621096105,
      "grad_norm": 19.875,
      "learning_rate": 5.8231749032841454e-05,
      "loss": 1.7108,
      "step": 49102
    },
    {
      "epoch": 4.224640798416932,
      "grad_norm": 1.0234375,
      "learning_rate": 5.8219084817641096e-05,
      "loss": 0.9759,
      "step": 49103
    },
    {
      "epoch": 4.224726834724254,
      "grad_norm": 1.328125,
      "learning_rate": 5.820642189456404e-05,
      "loss": 1.7104,
      "step": 49104
    },
    {
      "epoch": 4.224812871031576,
      "grad_norm": 1.2421875,
      "learning_rate": 5.819376026364737e-05,
      "loss": 1.0698,
      "step": 49105
    },
    {
      "epoch": 4.224898907338897,
      "grad_norm": 0.9140625,
      "learning_rate": 5.8181099924927874e-05,
      "loss": 0.6856,
      "step": 49106
    },
    {
      "epoch": 4.224984943646219,
      "grad_norm": 0.91015625,
      "learning_rate": 5.816844087844275e-05,
      "loss": 0.8405,
      "step": 49107
    },
    {
      "epoch": 4.22507097995354,
      "grad_norm": 2.984375,
      "learning_rate": 5.815578312422914e-05,
      "loss": 0.9406,
      "step": 49108
    },
    {
      "epoch": 4.225157016260862,
      "grad_norm": 1.109375,
      "learning_rate": 5.81431266623238e-05,
      "loss": 0.9572,
      "step": 49109
    },
    {
      "epoch": 4.225243052568183,
      "grad_norm": 0.8515625,
      "learning_rate": 5.8130471492763907e-05,
      "loss": 0.7366,
      "step": 49110
    },
    {
      "epoch": 4.2253290888755055,
      "grad_norm": 1.2578125,
      "learning_rate": 5.811781761558654e-05,
      "loss": 0.8548,
      "step": 49111
    },
    {
      "epoch": 4.225415125182828,
      "grad_norm": 0.91015625,
      "learning_rate": 5.810516503082852e-05,
      "loss": 1.0343,
      "step": 49112
    },
    {
      "epoch": 4.225501161490149,
      "grad_norm": 37.75,
      "learning_rate": 5.809251373852692e-05,
      "loss": 1.0708,
      "step": 49113
    },
    {
      "epoch": 4.225587197797471,
      "grad_norm": 1.0390625,
      "learning_rate": 5.807986373871887e-05,
      "loss": 1.9179,
      "step": 49114
    },
    {
      "epoch": 4.225673234104792,
      "grad_norm": 0.87890625,
      "learning_rate": 5.806721503144119e-05,
      "loss": 0.4591,
      "step": 49115
    },
    {
      "epoch": 4.225759270412114,
      "grad_norm": 1.0859375,
      "learning_rate": 5.805456761673106e-05,
      "loss": 1.4421,
      "step": 49116
    },
    {
      "epoch": 4.225845306719435,
      "grad_norm": 9.25,
      "learning_rate": 5.804192149462517e-05,
      "loss": 0.8543,
      "step": 49117
    },
    {
      "epoch": 4.225931343026757,
      "grad_norm": 0.94140625,
      "learning_rate": 5.802927666516078e-05,
      "loss": 1.5935,
      "step": 49118
    },
    {
      "epoch": 4.226017379334079,
      "grad_norm": 1.015625,
      "learning_rate": 5.8016633128374904e-05,
      "loss": 0.622,
      "step": 49119
    },
    {
      "epoch": 4.226103415641401,
      "grad_norm": 0.97265625,
      "learning_rate": 5.80039908843043e-05,
      "loss": 1.2001,
      "step": 49120
    },
    {
      "epoch": 4.226189451948723,
      "grad_norm": 1.265625,
      "learning_rate": 5.799134993298616e-05,
      "loss": 1.1215,
      "step": 49121
    },
    {
      "epoch": 4.226275488256044,
      "grad_norm": 7.625,
      "learning_rate": 5.797871027445728e-05,
      "loss": 1.1332,
      "step": 49122
    },
    {
      "epoch": 4.226361524563366,
      "grad_norm": 2.6875,
      "learning_rate": 5.7966071908754694e-05,
      "loss": 1.2653,
      "step": 49123
    },
    {
      "epoch": 4.226447560870687,
      "grad_norm": 0.92578125,
      "learning_rate": 5.795343483591548e-05,
      "loss": 0.5983,
      "step": 49124
    },
    {
      "epoch": 4.226533597178009,
      "grad_norm": 1.0703125,
      "learning_rate": 5.794079905597643e-05,
      "loss": 0.9482,
      "step": 49125
    },
    {
      "epoch": 4.226619633485331,
      "grad_norm": 6.4375,
      "learning_rate": 5.792816456897448e-05,
      "loss": 0.8379,
      "step": 49126
    },
    {
      "epoch": 4.226705669792652,
      "grad_norm": 0.921875,
      "learning_rate": 5.7915531374946875e-05,
      "loss": 1.3965,
      "step": 49127
    },
    {
      "epoch": 4.2267917060999745,
      "grad_norm": 1.2890625,
      "learning_rate": 5.790289947393018e-05,
      "loss": 1.3827,
      "step": 49128
    },
    {
      "epoch": 4.226877742407296,
      "grad_norm": 1.296875,
      "learning_rate": 5.789026886596155e-05,
      "loss": 1.2378,
      "step": 49129
    },
    {
      "epoch": 4.226963778714618,
      "grad_norm": 1.1328125,
      "learning_rate": 5.7877639551078e-05,
      "loss": 0.6981,
      "step": 49130
    },
    {
      "epoch": 4.227049815021939,
      "grad_norm": 0.82421875,
      "learning_rate": 5.786501152931628e-05,
      "loss": 0.5277,
      "step": 49131
    },
    {
      "epoch": 4.227135851329261,
      "grad_norm": 9.75,
      "learning_rate": 5.785238480071348e-05,
      "loss": 0.9749,
      "step": 49132
    },
    {
      "epoch": 4.227221887636583,
      "grad_norm": 1.09375,
      "learning_rate": 5.783975936530639e-05,
      "loss": 0.5363,
      "step": 49133
    },
    {
      "epoch": 4.227307923943904,
      "grad_norm": 3.59375,
      "learning_rate": 5.782713522313199e-05,
      "loss": 1.6384,
      "step": 49134
    },
    {
      "epoch": 4.227393960251226,
      "grad_norm": 1.40625,
      "learning_rate": 5.78145123742273e-05,
      "loss": 0.9594,
      "step": 49135
    },
    {
      "epoch": 4.2274799965585474,
      "grad_norm": 1.125,
      "learning_rate": 5.780189081862908e-05,
      "loss": 1.5035,
      "step": 49136
    },
    {
      "epoch": 4.2275660328658695,
      "grad_norm": 1.0234375,
      "learning_rate": 5.778927055637434e-05,
      "loss": 1.2333,
      "step": 49137
    },
    {
      "epoch": 4.227652069173191,
      "grad_norm": 1.15625,
      "learning_rate": 5.7776651587500065e-05,
      "loss": 0.88,
      "step": 49138
    },
    {
      "epoch": 4.227738105480513,
      "grad_norm": 1.09375,
      "learning_rate": 5.776403391204299e-05,
      "loss": 0.7043,
      "step": 49139
    },
    {
      "epoch": 4.227824141787835,
      "grad_norm": 1.2109375,
      "learning_rate": 5.775141753004009e-05,
      "loss": 1.4967,
      "step": 49140
    },
    {
      "epoch": 4.227910178095156,
      "grad_norm": 2.546875,
      "learning_rate": 5.77388024415284e-05,
      "loss": 1.6632,
      "step": 49141
    },
    {
      "epoch": 4.227996214402478,
      "grad_norm": 0.98046875,
      "learning_rate": 5.772618864654461e-05,
      "loss": 1.3507,
      "step": 49142
    },
    {
      "epoch": 4.228082250709799,
      "grad_norm": 5.6875,
      "learning_rate": 5.771357614512562e-05,
      "loss": 1.4667,
      "step": 49143
    },
    {
      "epoch": 4.228168287017121,
      "grad_norm": 0.796875,
      "learning_rate": 5.7700964937308584e-05,
      "loss": 0.4087,
      "step": 49144
    },
    {
      "epoch": 4.2282543233244425,
      "grad_norm": 1.125,
      "learning_rate": 5.7688355023130076e-05,
      "loss": 1.4073,
      "step": 49145
    },
    {
      "epoch": 4.228340359631765,
      "grad_norm": 1.0078125,
      "learning_rate": 5.767574640262724e-05,
      "loss": 1.2327,
      "step": 49146
    },
    {
      "epoch": 4.228426395939087,
      "grad_norm": 1.046875,
      "learning_rate": 5.766313907583665e-05,
      "loss": 0.6375,
      "step": 49147
    },
    {
      "epoch": 4.228512432246408,
      "grad_norm": 1.1171875,
      "learning_rate": 5.765053304279538e-05,
      "loss": 1.2832,
      "step": 49148
    },
    {
      "epoch": 4.22859846855373,
      "grad_norm": 1.5234375,
      "learning_rate": 5.7637928303540365e-05,
      "loss": 1.2627,
      "step": 49149
    },
    {
      "epoch": 4.228684504861051,
      "grad_norm": 1.2109375,
      "learning_rate": 5.762532485810829e-05,
      "loss": 0.531,
      "step": 49150
    },
    {
      "epoch": 4.228770541168373,
      "grad_norm": 48.5,
      "learning_rate": 5.761272270653611e-05,
      "loss": 1.2271,
      "step": 49151
    },
    {
      "epoch": 4.228856577475694,
      "grad_norm": 0.92578125,
      "learning_rate": 5.7600121848860765e-05,
      "loss": 0.6879,
      "step": 49152
    },
    {
      "epoch": 4.228942613783016,
      "grad_norm": 2.890625,
      "learning_rate": 5.758752228511888e-05,
      "loss": 1.0436,
      "step": 49153
    },
    {
      "epoch": 4.2290286500903385,
      "grad_norm": 1.25,
      "learning_rate": 5.7574924015347474e-05,
      "loss": 0.9394,
      "step": 49154
    },
    {
      "epoch": 4.22911468639766,
      "grad_norm": 7.125,
      "learning_rate": 5.756232703958342e-05,
      "loss": 1.4281,
      "step": 49155
    },
    {
      "epoch": 4.229200722704982,
      "grad_norm": 0.80078125,
      "learning_rate": 5.754973135786346e-05,
      "loss": 0.5349,
      "step": 49156
    },
    {
      "epoch": 4.229286759012303,
      "grad_norm": 1.3359375,
      "learning_rate": 5.753713697022461e-05,
      "loss": 2.0197,
      "step": 49157
    },
    {
      "epoch": 4.229372795319625,
      "grad_norm": 2.8125,
      "learning_rate": 5.752454387670342e-05,
      "loss": 0.6625,
      "step": 49158
    },
    {
      "epoch": 4.229458831626946,
      "grad_norm": 1.3046875,
      "learning_rate": 5.751195207733695e-05,
      "loss": 1.102,
      "step": 49159
    },
    {
      "epoch": 4.229544867934268,
      "grad_norm": 1.1640625,
      "learning_rate": 5.7499361572162015e-05,
      "loss": 1.2634,
      "step": 49160
    },
    {
      "epoch": 4.22963090424159,
      "grad_norm": 2.6875,
      "learning_rate": 5.74867723612153e-05,
      "loss": 0.8898,
      "step": 49161
    },
    {
      "epoch": 4.2297169405489115,
      "grad_norm": 1.234375,
      "learning_rate": 5.747418444453384e-05,
      "loss": 0.716,
      "step": 49162
    },
    {
      "epoch": 4.2298029768562335,
      "grad_norm": 1.5625,
      "learning_rate": 5.746159782215421e-05,
      "loss": 1.0527,
      "step": 49163
    },
    {
      "epoch": 4.229889013163555,
      "grad_norm": 1.03125,
      "learning_rate": 5.744901249411333e-05,
      "loss": 0.6287,
      "step": 49164
    },
    {
      "epoch": 4.229975049470877,
      "grad_norm": 1.21875,
      "learning_rate": 5.743642846044811e-05,
      "loss": 1.2471,
      "step": 49165
    },
    {
      "epoch": 4.230061085778198,
      "grad_norm": 1.203125,
      "learning_rate": 5.742384572119519e-05,
      "loss": 1.4282,
      "step": 49166
    },
    {
      "epoch": 4.23014712208552,
      "grad_norm": 1.703125,
      "learning_rate": 5.741126427639143e-05,
      "loss": 1.0036,
      "step": 49167
    },
    {
      "epoch": 4.230233158392842,
      "grad_norm": 1.3046875,
      "learning_rate": 5.739868412607374e-05,
      "loss": 1.0872,
      "step": 49168
    },
    {
      "epoch": 4.230319194700163,
      "grad_norm": 21.875,
      "learning_rate": 5.738610527027871e-05,
      "loss": 1.1786,
      "step": 49169
    },
    {
      "epoch": 4.230405231007485,
      "grad_norm": 1.1640625,
      "learning_rate": 5.7373527709043246e-05,
      "loss": 0.7606,
      "step": 49170
    },
    {
      "epoch": 4.2304912673148065,
      "grad_norm": 326.0,
      "learning_rate": 5.736095144240422e-05,
      "loss": 1.3399,
      "step": 49171
    },
    {
      "epoch": 4.230577303622129,
      "grad_norm": 1.484375,
      "learning_rate": 5.73483764703982e-05,
      "loss": 1.2364,
      "step": 49172
    },
    {
      "epoch": 4.230663339929451,
      "grad_norm": 1.0859375,
      "learning_rate": 5.7335802793062217e-05,
      "loss": 1.4129,
      "step": 49173
    },
    {
      "epoch": 4.230749376236772,
      "grad_norm": 0.94140625,
      "learning_rate": 5.732323041043286e-05,
      "loss": 0.7222,
      "step": 49174
    },
    {
      "epoch": 4.230835412544094,
      "grad_norm": 2.546875,
      "learning_rate": 5.731065932254687e-05,
      "loss": 1.6348,
      "step": 49175
    },
    {
      "epoch": 4.230921448851415,
      "grad_norm": 0.93359375,
      "learning_rate": 5.729808952944127e-05,
      "loss": 0.9267,
      "step": 49176
    },
    {
      "epoch": 4.231007485158737,
      "grad_norm": 0.98828125,
      "learning_rate": 5.728552103115253e-05,
      "loss": 1.0408,
      "step": 49177
    },
    {
      "epoch": 4.231093521466058,
      "grad_norm": 1.0234375,
      "learning_rate": 5.727295382771752e-05,
      "loss": 1.5072,
      "step": 49178
    },
    {
      "epoch": 4.23117955777338,
      "grad_norm": 8.375,
      "learning_rate": 5.7260387919173154e-05,
      "loss": 1.4419,
      "step": 49179
    },
    {
      "epoch": 4.2312655940807025,
      "grad_norm": 57.25,
      "learning_rate": 5.724782330555595e-05,
      "loss": 1.1707,
      "step": 49180
    },
    {
      "epoch": 4.231351630388024,
      "grad_norm": 40.5,
      "learning_rate": 5.723525998690271e-05,
      "loss": 1.4463,
      "step": 49181
    },
    {
      "epoch": 4.231437666695346,
      "grad_norm": 1.0703125,
      "learning_rate": 5.7222697963250356e-05,
      "loss": 0.8097,
      "step": 49182
    },
    {
      "epoch": 4.231523703002667,
      "grad_norm": 1.171875,
      "learning_rate": 5.721013723463536e-05,
      "loss": 1.7865,
      "step": 49183
    },
    {
      "epoch": 4.231609739309989,
      "grad_norm": 0.8203125,
      "learning_rate": 5.719757780109464e-05,
      "loss": 0.9263,
      "step": 49184
    },
    {
      "epoch": 4.23169577561731,
      "grad_norm": 0.83203125,
      "learning_rate": 5.718501966266493e-05,
      "loss": 0.8916,
      "step": 49185
    },
    {
      "epoch": 4.231781811924632,
      "grad_norm": 1.046875,
      "learning_rate": 5.7172462819382876e-05,
      "loss": 1.382,
      "step": 49186
    },
    {
      "epoch": 4.231867848231954,
      "grad_norm": 1.3203125,
      "learning_rate": 5.715990727128528e-05,
      "loss": 1.2081,
      "step": 49187
    },
    {
      "epoch": 4.2319538845392755,
      "grad_norm": 0.7890625,
      "learning_rate": 5.714735301840879e-05,
      "loss": 0.5849,
      "step": 49188
    },
    {
      "epoch": 4.2320399208465975,
      "grad_norm": 1.0546875,
      "learning_rate": 5.713480006079014e-05,
      "loss": 0.6527,
      "step": 49189
    },
    {
      "epoch": 4.232125957153919,
      "grad_norm": 1.078125,
      "learning_rate": 5.7122248398466136e-05,
      "loss": 1.2296,
      "step": 49190
    },
    {
      "epoch": 4.232211993461241,
      "grad_norm": 1.5,
      "learning_rate": 5.710969803147326e-05,
      "loss": 1.5687,
      "step": 49191
    },
    {
      "epoch": 4.232298029768562,
      "grad_norm": 1.1796875,
      "learning_rate": 5.709714895984858e-05,
      "loss": 1.5605,
      "step": 49192
    },
    {
      "epoch": 4.232384066075884,
      "grad_norm": 1.1640625,
      "learning_rate": 5.708460118362846e-05,
      "loss": 0.9595,
      "step": 49193
    },
    {
      "epoch": 4.232470102383206,
      "grad_norm": 1.0,
      "learning_rate": 5.7072054702849765e-05,
      "loss": 0.7784,
      "step": 49194
    },
    {
      "epoch": 4.232556138690527,
      "grad_norm": 1.109375,
      "learning_rate": 5.705950951754929e-05,
      "loss": 1.6203,
      "step": 49195
    },
    {
      "epoch": 4.232642174997849,
      "grad_norm": 1.09375,
      "learning_rate": 5.704696562776346e-05,
      "loss": 0.7746,
      "step": 49196
    },
    {
      "epoch": 4.2327282113051705,
      "grad_norm": 1.140625,
      "learning_rate": 5.703442303352913e-05,
      "loss": 0.9783,
      "step": 49197
    },
    {
      "epoch": 4.232814247612493,
      "grad_norm": 1.1796875,
      "learning_rate": 5.702188173488304e-05,
      "loss": 1.3273,
      "step": 49198
    },
    {
      "epoch": 4.232900283919814,
      "grad_norm": 1.1171875,
      "learning_rate": 5.7009341731861676e-05,
      "loss": 1.3581,
      "step": 49199
    },
    {
      "epoch": 4.232986320227136,
      "grad_norm": 0.98046875,
      "learning_rate": 5.699680302450189e-05,
      "loss": 0.5401,
      "step": 49200
    },
    {
      "epoch": 4.233072356534458,
      "grad_norm": 1.140625,
      "learning_rate": 5.698426561284037e-05,
      "loss": 0.8357,
      "step": 49201
    },
    {
      "epoch": 4.233158392841779,
      "grad_norm": 1.2578125,
      "learning_rate": 5.6971729496913594e-05,
      "loss": 1.3216,
      "step": 49202
    },
    {
      "epoch": 4.233244429149101,
      "grad_norm": 1.234375,
      "learning_rate": 5.695919467675847e-05,
      "loss": 0.5165,
      "step": 49203
    },
    {
      "epoch": 4.233330465456422,
      "grad_norm": 1.0546875,
      "learning_rate": 5.694666115241143e-05,
      "loss": 1.1923,
      "step": 49204
    },
    {
      "epoch": 4.233416501763744,
      "grad_norm": 3.984375,
      "learning_rate": 5.69341289239092e-05,
      "loss": 1.4432,
      "step": 49205
    },
    {
      "epoch": 4.233502538071066,
      "grad_norm": 1.0859375,
      "learning_rate": 5.69215979912886e-05,
      "loss": 0.6835,
      "step": 49206
    },
    {
      "epoch": 4.233588574378388,
      "grad_norm": 1.1171875,
      "learning_rate": 5.6909068354586036e-05,
      "loss": 0.7737,
      "step": 49207
    },
    {
      "epoch": 4.23367461068571,
      "grad_norm": 2.015625,
      "learning_rate": 5.689654001383832e-05,
      "loss": 1.0841,
      "step": 49208
    },
    {
      "epoch": 4.233760646993031,
      "grad_norm": 0.8984375,
      "learning_rate": 5.688401296908208e-05,
      "loss": 1.4149,
      "step": 49209
    },
    {
      "epoch": 4.233846683300353,
      "grad_norm": 1.078125,
      "learning_rate": 5.6871487220353904e-05,
      "loss": 1.4322,
      "step": 49210
    },
    {
      "epoch": 4.233932719607674,
      "grad_norm": 1.0625,
      "learning_rate": 5.685896276769037e-05,
      "loss": 1.1731,
      "step": 49211
    },
    {
      "epoch": 4.234018755914996,
      "grad_norm": 0.97265625,
      "learning_rate": 5.6846439611128286e-05,
      "loss": 0.4205,
      "step": 49212
    },
    {
      "epoch": 4.234104792222317,
      "grad_norm": 1.265625,
      "learning_rate": 5.683391775070412e-05,
      "loss": 1.4336,
      "step": 49213
    },
    {
      "epoch": 4.2341908285296395,
      "grad_norm": 1.1484375,
      "learning_rate": 5.682139718645463e-05,
      "loss": 1.2956,
      "step": 49214
    },
    {
      "epoch": 4.2342768648369615,
      "grad_norm": 1.0546875,
      "learning_rate": 5.680887791841627e-05,
      "loss": 1.5012,
      "step": 49215
    },
    {
      "epoch": 4.234362901144283,
      "grad_norm": 1.015625,
      "learning_rate": 5.6796359946625745e-05,
      "loss": 1.5466,
      "step": 49216
    },
    {
      "epoch": 4.234448937451605,
      "grad_norm": 2.34375,
      "learning_rate": 5.6783843271119737e-05,
      "loss": 1.3148,
      "step": 49217
    },
    {
      "epoch": 4.234534973758926,
      "grad_norm": 1.140625,
      "learning_rate": 5.677132789193473e-05,
      "loss": 0.919,
      "step": 49218
    },
    {
      "epoch": 4.234621010066248,
      "grad_norm": 1.5625,
      "learning_rate": 5.675881380910736e-05,
      "loss": 1.2542,
      "step": 49219
    },
    {
      "epoch": 4.23470704637357,
      "grad_norm": 1.1640625,
      "learning_rate": 5.6746301022674366e-05,
      "loss": 0.545,
      "step": 49220
    },
    {
      "epoch": 4.234793082680891,
      "grad_norm": 2.078125,
      "learning_rate": 5.673378953267211e-05,
      "loss": 1.4985,
      "step": 49221
    },
    {
      "epoch": 4.234879118988213,
      "grad_norm": 0.76171875,
      "learning_rate": 5.6721279339137346e-05,
      "loss": 0.9392,
      "step": 49222
    },
    {
      "epoch": 4.2349651552955345,
      "grad_norm": 0.81640625,
      "learning_rate": 5.6708770442106705e-05,
      "loss": 0.7212,
      "step": 49223
    },
    {
      "epoch": 4.235051191602857,
      "grad_norm": 0.88671875,
      "learning_rate": 5.669626284161661e-05,
      "loss": 0.6335,
      "step": 49224
    },
    {
      "epoch": 4.235137227910178,
      "grad_norm": 1.703125,
      "learning_rate": 5.668375653770369e-05,
      "loss": 0.9171,
      "step": 49225
    },
    {
      "epoch": 4.2352232642175,
      "grad_norm": 1.5625,
      "learning_rate": 5.66712515304047e-05,
      "loss": 1.4725,
      "step": 49226
    },
    {
      "epoch": 4.235309300524822,
      "grad_norm": 5.09375,
      "learning_rate": 5.665874781975594e-05,
      "loss": 1.7772,
      "step": 49227
    },
    {
      "epoch": 4.235395336832143,
      "grad_norm": 1.3046875,
      "learning_rate": 5.664624540579416e-05,
      "loss": 0.9876,
      "step": 49228
    },
    {
      "epoch": 4.235481373139465,
      "grad_norm": 1.3828125,
      "learning_rate": 5.663374428855589e-05,
      "loss": 1.2956,
      "step": 49229
    },
    {
      "epoch": 4.235567409446786,
      "grad_norm": 0.796875,
      "learning_rate": 5.662124446807765e-05,
      "loss": 0.8317,
      "step": 49230
    },
    {
      "epoch": 4.235653445754108,
      "grad_norm": 1.1484375,
      "learning_rate": 5.6608745944396134e-05,
      "loss": 1.3514,
      "step": 49231
    },
    {
      "epoch": 4.23573948206143,
      "grad_norm": 1.203125,
      "learning_rate": 5.659624871754776e-05,
      "loss": 1.4658,
      "step": 49232
    },
    {
      "epoch": 4.235825518368752,
      "grad_norm": 1.203125,
      "learning_rate": 5.658375278756917e-05,
      "loss": 1.9864,
      "step": 49233
    },
    {
      "epoch": 4.235911554676074,
      "grad_norm": 1.0703125,
      "learning_rate": 5.657125815449676e-05,
      "loss": 0.8558,
      "step": 49234
    },
    {
      "epoch": 4.235997590983395,
      "grad_norm": 1.1796875,
      "learning_rate": 5.655876481836719e-05,
      "loss": 0.7328,
      "step": 49235
    },
    {
      "epoch": 4.236083627290717,
      "grad_norm": 1.21875,
      "learning_rate": 5.654627277921709e-05,
      "loss": 1.4927,
      "step": 49236
    },
    {
      "epoch": 4.236169663598038,
      "grad_norm": 0.859375,
      "learning_rate": 5.653378203708281e-05,
      "loss": 0.6359,
      "step": 49237
    },
    {
      "epoch": 4.23625569990536,
      "grad_norm": 1.2265625,
      "learning_rate": 5.652129259200095e-05,
      "loss": 1.523,
      "step": 49238
    },
    {
      "epoch": 4.236341736212681,
      "grad_norm": 1.3046875,
      "learning_rate": 5.650880444400819e-05,
      "loss": 1.2836,
      "step": 49239
    },
    {
      "epoch": 4.2364277725200035,
      "grad_norm": 1.8203125,
      "learning_rate": 5.649631759314078e-05,
      "loss": 1.0595,
      "step": 49240
    },
    {
      "epoch": 4.2365138088273255,
      "grad_norm": 0.97265625,
      "learning_rate": 5.648383203943536e-05,
      "loss": 0.6294,
      "step": 49241
    },
    {
      "epoch": 4.236599845134647,
      "grad_norm": 1.484375,
      "learning_rate": 5.647134778292862e-05,
      "loss": 1.4507,
      "step": 49242
    },
    {
      "epoch": 4.236685881441969,
      "grad_norm": 1.3125,
      "learning_rate": 5.645886482365681e-05,
      "loss": 1.9734,
      "step": 49243
    },
    {
      "epoch": 4.23677191774929,
      "grad_norm": 1.5703125,
      "learning_rate": 5.644638316165668e-05,
      "loss": 1.9636,
      "step": 49244
    },
    {
      "epoch": 4.236857954056612,
      "grad_norm": 1.40625,
      "learning_rate": 5.643390279696453e-05,
      "loss": 1.5068,
      "step": 49245
    },
    {
      "epoch": 4.236943990363933,
      "grad_norm": 1.328125,
      "learning_rate": 5.642142372961689e-05,
      "loss": 1.126,
      "step": 49246
    },
    {
      "epoch": 4.237030026671255,
      "grad_norm": 0.9765625,
      "learning_rate": 5.640894595965046e-05,
      "loss": 1.4127,
      "step": 49247
    },
    {
      "epoch": 4.237116062978577,
      "grad_norm": 1.1875,
      "learning_rate": 5.639646948710142e-05,
      "loss": 1.8385,
      "step": 49248
    },
    {
      "epoch": 4.2372020992858985,
      "grad_norm": 1.2109375,
      "learning_rate": 5.638399431200647e-05,
      "loss": 1.1187,
      "step": 49249
    },
    {
      "epoch": 4.237288135593221,
      "grad_norm": 1.0,
      "learning_rate": 5.637152043440219e-05,
      "loss": 0.7784,
      "step": 49250
    },
    {
      "epoch": 4.237374171900542,
      "grad_norm": 1.28125,
      "learning_rate": 5.635904785432483e-05,
      "loss": 0.9552,
      "step": 49251
    },
    {
      "epoch": 4.237460208207864,
      "grad_norm": 0.97265625,
      "learning_rate": 5.634657657181092e-05,
      "loss": 1.032,
      "step": 49252
    },
    {
      "epoch": 4.237546244515185,
      "grad_norm": 1.5234375,
      "learning_rate": 5.633410658689714e-05,
      "loss": 1.2726,
      "step": 49253
    },
    {
      "epoch": 4.237632280822507,
      "grad_norm": 0.921875,
      "learning_rate": 5.6321637899619694e-05,
      "loss": 1.0923,
      "step": 49254
    },
    {
      "epoch": 4.237718317129829,
      "grad_norm": 1.2890625,
      "learning_rate": 5.630917051001522e-05,
      "loss": 1.2354,
      "step": 49255
    },
    {
      "epoch": 4.23780435343715,
      "grad_norm": 0.93359375,
      "learning_rate": 5.629670441812007e-05,
      "loss": 1.1501,
      "step": 49256
    },
    {
      "epoch": 4.237890389744472,
      "grad_norm": 6.09375,
      "learning_rate": 5.628423962397078e-05,
      "loss": 0.6912,
      "step": 49257
    },
    {
      "epoch": 4.237976426051794,
      "grad_norm": 416.0,
      "learning_rate": 5.627177612760387e-05,
      "loss": 1.1867,
      "step": 49258
    },
    {
      "epoch": 4.238062462359116,
      "grad_norm": 1.625,
      "learning_rate": 5.625931392905559e-05,
      "loss": 0.661,
      "step": 49259
    },
    {
      "epoch": 4.238148498666437,
      "grad_norm": 1.203125,
      "learning_rate": 5.6246853028362576e-05,
      "loss": 0.9968,
      "step": 49260
    },
    {
      "epoch": 4.238234534973759,
      "grad_norm": 10.125,
      "learning_rate": 5.623439342556125e-05,
      "loss": 1.4376,
      "step": 49261
    },
    {
      "epoch": 4.238320571281081,
      "grad_norm": 38.25,
      "learning_rate": 5.622193512068796e-05,
      "loss": 1.7259,
      "step": 49262
    },
    {
      "epoch": 4.238406607588402,
      "grad_norm": 1.125,
      "learning_rate": 5.620947811377919e-05,
      "loss": 1.1908,
      "step": 49263
    },
    {
      "epoch": 4.238492643895724,
      "grad_norm": 1.078125,
      "learning_rate": 5.619702240487145e-05,
      "loss": 0.9642,
      "step": 49264
    },
    {
      "epoch": 4.238578680203045,
      "grad_norm": 1.75,
      "learning_rate": 5.618456799400096e-05,
      "loss": 1.1178,
      "step": 49265
    },
    {
      "epoch": 4.2386647165103675,
      "grad_norm": 0.90625,
      "learning_rate": 5.617211488120444e-05,
      "loss": 1.6096,
      "step": 49266
    },
    {
      "epoch": 4.238750752817689,
      "grad_norm": 1.296875,
      "learning_rate": 5.6159663066518096e-05,
      "loss": 0.8451,
      "step": 49267
    },
    {
      "epoch": 4.238836789125011,
      "grad_norm": 1.0625,
      "learning_rate": 5.61472125499784e-05,
      "loss": 1.616,
      "step": 49268
    },
    {
      "epoch": 4.238922825432333,
      "grad_norm": 1.375,
      "learning_rate": 5.613476333162187e-05,
      "loss": 1.1115,
      "step": 49269
    },
    {
      "epoch": 4.239008861739654,
      "grad_norm": 7.125,
      "learning_rate": 5.612231541148477e-05,
      "loss": 0.8853,
      "step": 49270
    },
    {
      "epoch": 4.239094898046976,
      "grad_norm": 1.3203125,
      "learning_rate": 5.610986878960356e-05,
      "loss": 1.4142,
      "step": 49271
    },
    {
      "epoch": 4.239180934354297,
      "grad_norm": 2.328125,
      "learning_rate": 5.609742346601471e-05,
      "loss": 1.0853,
      "step": 49272
    },
    {
      "epoch": 4.239266970661619,
      "grad_norm": 1.0390625,
      "learning_rate": 5.608497944075453e-05,
      "loss": 1.2063,
      "step": 49273
    },
    {
      "epoch": 4.2393530069689405,
      "grad_norm": 1.0859375,
      "learning_rate": 5.607253671385948e-05,
      "loss": 0.9512,
      "step": 49274
    },
    {
      "epoch": 4.2394390432762625,
      "grad_norm": 1.3671875,
      "learning_rate": 5.6060095285365876e-05,
      "loss": 1.2636,
      "step": 49275
    },
    {
      "epoch": 4.239525079583585,
      "grad_norm": 450.0,
      "learning_rate": 5.6047655155310076e-05,
      "loss": 1.2339,
      "step": 49276
    },
    {
      "epoch": 4.239611115890906,
      "grad_norm": 0.9453125,
      "learning_rate": 5.6035216323728714e-05,
      "loss": 0.756,
      "step": 49277
    },
    {
      "epoch": 4.239697152198228,
      "grad_norm": 1.484375,
      "learning_rate": 5.602277879065787e-05,
      "loss": 1.8677,
      "step": 49278
    },
    {
      "epoch": 4.239783188505549,
      "grad_norm": 1.0625,
      "learning_rate": 5.6010342556134074e-05,
      "loss": 1.3338,
      "step": 49279
    },
    {
      "epoch": 4.239869224812871,
      "grad_norm": 1.2109375,
      "learning_rate": 5.599790762019374e-05,
      "loss": 1.5956,
      "step": 49280
    },
    {
      "epoch": 4.239955261120192,
      "grad_norm": 1.2890625,
      "learning_rate": 5.598547398287307e-05,
      "loss": 0.9414,
      "step": 49281
    },
    {
      "epoch": 4.240041297427514,
      "grad_norm": 4.125,
      "learning_rate": 5.597304164420858e-05,
      "loss": 0.8968,
      "step": 49282
    },
    {
      "epoch": 4.240127333734836,
      "grad_norm": 1.1328125,
      "learning_rate": 5.5960610604236626e-05,
      "loss": 1.2697,
      "step": 49283
    },
    {
      "epoch": 4.240213370042158,
      "grad_norm": 0.984375,
      "learning_rate": 5.5948180862993466e-05,
      "loss": 1.5427,
      "step": 49284
    },
    {
      "epoch": 4.24029940634948,
      "grad_norm": 1.0234375,
      "learning_rate": 5.593575242051563e-05,
      "loss": 1.4312,
      "step": 49285
    },
    {
      "epoch": 4.240385442656801,
      "grad_norm": 4.28125,
      "learning_rate": 5.592332527683919e-05,
      "loss": 1.3301,
      "step": 49286
    },
    {
      "epoch": 4.240471478964123,
      "grad_norm": 1.1640625,
      "learning_rate": 5.591089943200067e-05,
      "loss": 0.7955,
      "step": 49287
    },
    {
      "epoch": 4.240557515271445,
      "grad_norm": 1.0234375,
      "learning_rate": 5.5898474886036554e-05,
      "loss": 0.6966,
      "step": 49288
    },
    {
      "epoch": 4.240643551578766,
      "grad_norm": 1.0859375,
      "learning_rate": 5.588605163898286e-05,
      "loss": 1.3106,
      "step": 49289
    },
    {
      "epoch": 4.240729587886088,
      "grad_norm": 1.8515625,
      "learning_rate": 5.5873629690876114e-05,
      "loss": 1.0613,
      "step": 49290
    },
    {
      "epoch": 4.240815624193409,
      "grad_norm": 0.77734375,
      "learning_rate": 5.586120904175274e-05,
      "loss": 0.5678,
      "step": 49291
    },
    {
      "epoch": 4.2409016605007315,
      "grad_norm": 1.0546875,
      "learning_rate": 5.584878969164886e-05,
      "loss": 1.0535,
      "step": 49292
    },
    {
      "epoch": 4.240987696808053,
      "grad_norm": 1.0625,
      "learning_rate": 5.583637164060085e-05,
      "loss": 0.875,
      "step": 49293
    },
    {
      "epoch": 4.241073733115375,
      "grad_norm": 1.1875,
      "learning_rate": 5.582395488864517e-05,
      "loss": 1.5485,
      "step": 49294
    },
    {
      "epoch": 4.241159769422697,
      "grad_norm": 0.9453125,
      "learning_rate": 5.581153943581796e-05,
      "loss": 0.631,
      "step": 49295
    },
    {
      "epoch": 4.241245805730018,
      "grad_norm": 1.078125,
      "learning_rate": 5.579912528215558e-05,
      "loss": 1.281,
      "step": 49296
    },
    {
      "epoch": 4.24133184203734,
      "grad_norm": 0.93359375,
      "learning_rate": 5.5786712427694506e-05,
      "loss": 1.2635,
      "step": 49297
    },
    {
      "epoch": 4.241417878344661,
      "grad_norm": 1.7734375,
      "learning_rate": 5.577430087247076e-05,
      "loss": 1.1828,
      "step": 49298
    },
    {
      "epoch": 4.241503914651983,
      "grad_norm": 1.7890625,
      "learning_rate": 5.5761890616520925e-05,
      "loss": 1.3989,
      "step": 49299
    },
    {
      "epoch": 4.2415899509593045,
      "grad_norm": 3.84375,
      "learning_rate": 5.574948165988103e-05,
      "loss": 0.9953,
      "step": 49300
    },
    {
      "epoch": 4.2416759872666265,
      "grad_norm": 0.87890625,
      "learning_rate": 5.573707400258754e-05,
      "loss": 0.7356,
      "step": 49301
    },
    {
      "epoch": 4.241762023573949,
      "grad_norm": 1.1015625,
      "learning_rate": 5.572466764467665e-05,
      "loss": 0.7627,
      "step": 49302
    },
    {
      "epoch": 4.24184805988127,
      "grad_norm": 0.984375,
      "learning_rate": 5.571226258618473e-05,
      "loss": 0.7604,
      "step": 49303
    },
    {
      "epoch": 4.241934096188592,
      "grad_norm": 1.1796875,
      "learning_rate": 5.569985882714812e-05,
      "loss": 0.4829,
      "step": 49304
    },
    {
      "epoch": 4.242020132495913,
      "grad_norm": 0.9921875,
      "learning_rate": 5.568745636760292e-05,
      "loss": 0.6483,
      "step": 49305
    },
    {
      "epoch": 4.242106168803235,
      "grad_norm": 1.0,
      "learning_rate": 5.5675055207585477e-05,
      "loss": 1.2688,
      "step": 49306
    },
    {
      "epoch": 4.242192205110556,
      "grad_norm": 1.109375,
      "learning_rate": 5.5662655347132217e-05,
      "loss": 1.1665,
      "step": 49307
    },
    {
      "epoch": 4.242278241417878,
      "grad_norm": 1.140625,
      "learning_rate": 5.565025678627916e-05,
      "loss": 0.7325,
      "step": 49308
    },
    {
      "epoch": 4.2423642777252,
      "grad_norm": 0.98046875,
      "learning_rate": 5.563785952506267e-05,
      "loss": 0.5729,
      "step": 49309
    },
    {
      "epoch": 4.242450314032522,
      "grad_norm": 7.0,
      "learning_rate": 5.562546356351911e-05,
      "loss": 0.8579,
      "step": 49310
    },
    {
      "epoch": 4.242536350339844,
      "grad_norm": 1.2578125,
      "learning_rate": 5.561306890168455e-05,
      "loss": 0.6182,
      "step": 49311
    },
    {
      "epoch": 4.242622386647165,
      "grad_norm": 27.125,
      "learning_rate": 5.5600675539595355e-05,
      "loss": 1.1976,
      "step": 49312
    },
    {
      "epoch": 4.242708422954487,
      "grad_norm": 4.125,
      "learning_rate": 5.558828347728784e-05,
      "loss": 0.6971,
      "step": 49313
    },
    {
      "epoch": 4.242794459261808,
      "grad_norm": 1.0078125,
      "learning_rate": 5.557589271479807e-05,
      "loss": 1.0645,
      "step": 49314
    },
    {
      "epoch": 4.24288049556913,
      "grad_norm": 1.0,
      "learning_rate": 5.556350325216247e-05,
      "loss": 1.1708,
      "step": 49315
    },
    {
      "epoch": 4.242966531876452,
      "grad_norm": 0.9921875,
      "learning_rate": 5.555111508941707e-05,
      "loss": 0.9835,
      "step": 49316
    },
    {
      "epoch": 4.243052568183773,
      "grad_norm": 0.7109375,
      "learning_rate": 5.553872822659822e-05,
      "loss": 0.7448,
      "step": 49317
    },
    {
      "epoch": 4.2431386044910955,
      "grad_norm": 11.9375,
      "learning_rate": 5.5526342663742294e-05,
      "loss": 1.6281,
      "step": 49318
    },
    {
      "epoch": 4.243224640798417,
      "grad_norm": 0.9375,
      "learning_rate": 5.551395840088519e-05,
      "loss": 0.8546,
      "step": 49319
    },
    {
      "epoch": 4.243310677105739,
      "grad_norm": 1.265625,
      "learning_rate": 5.550157543806339e-05,
      "loss": 1.3197,
      "step": 49320
    },
    {
      "epoch": 4.24339671341306,
      "grad_norm": 2.3125,
      "learning_rate": 5.548919377531309e-05,
      "loss": 0.8198,
      "step": 49321
    },
    {
      "epoch": 4.243482749720382,
      "grad_norm": 1.34375,
      "learning_rate": 5.547681341267036e-05,
      "loss": 1.079,
      "step": 49322
    },
    {
      "epoch": 4.243568786027704,
      "grad_norm": 1.1953125,
      "learning_rate": 5.5464434350171454e-05,
      "loss": 0.7165,
      "step": 49323
    },
    {
      "epoch": 4.243654822335025,
      "grad_norm": 1.0390625,
      "learning_rate": 5.545205658785274e-05,
      "loss": 1.8653,
      "step": 49324
    },
    {
      "epoch": 4.243740858642347,
      "grad_norm": 1.6484375,
      "learning_rate": 5.5439680125750236e-05,
      "loss": 1.305,
      "step": 49325
    },
    {
      "epoch": 4.2438268949496685,
      "grad_norm": 1.2265625,
      "learning_rate": 5.542730496390031e-05,
      "loss": 1.2399,
      "step": 49326
    },
    {
      "epoch": 4.2439129312569905,
      "grad_norm": 1.0078125,
      "learning_rate": 5.5414931102338916e-05,
      "loss": 0.994,
      "step": 49327
    },
    {
      "epoch": 4.243998967564312,
      "grad_norm": 0.69921875,
      "learning_rate": 5.5402558541102376e-05,
      "loss": 0.4798,
      "step": 49328
    },
    {
      "epoch": 4.244085003871634,
      "grad_norm": 1.4140625,
      "learning_rate": 5.539018728022699e-05,
      "loss": 1.2783,
      "step": 49329
    },
    {
      "epoch": 4.244171040178956,
      "grad_norm": 1.1640625,
      "learning_rate": 5.537781731974873e-05,
      "loss": 1.5813,
      "step": 49330
    },
    {
      "epoch": 4.244257076486277,
      "grad_norm": 1.5078125,
      "learning_rate": 5.536544865970389e-05,
      "loss": 1.3864,
      "step": 49331
    },
    {
      "epoch": 4.244343112793599,
      "grad_norm": 1.1484375,
      "learning_rate": 5.535308130012878e-05,
      "loss": 0.9197,
      "step": 49332
    },
    {
      "epoch": 4.24442914910092,
      "grad_norm": 1.1796875,
      "learning_rate": 5.534071524105927e-05,
      "loss": 1.1451,
      "step": 49333
    },
    {
      "epoch": 4.244515185408242,
      "grad_norm": 1.15625,
      "learning_rate": 5.5328350482531706e-05,
      "loss": 0.9605,
      "step": 49334
    },
    {
      "epoch": 4.244601221715564,
      "grad_norm": 0.75390625,
      "learning_rate": 5.531598702458235e-05,
      "loss": 0.7516,
      "step": 49335
    },
    {
      "epoch": 4.244687258022886,
      "grad_norm": 1.1875,
      "learning_rate": 5.5303624867247106e-05,
      "loss": 1.1292,
      "step": 49336
    },
    {
      "epoch": 4.244773294330208,
      "grad_norm": 1.2421875,
      "learning_rate": 5.529126401056228e-05,
      "loss": 1.4498,
      "step": 49337
    },
    {
      "epoch": 4.244859330637529,
      "grad_norm": 1.2578125,
      "learning_rate": 5.527890445456413e-05,
      "loss": 1.3391,
      "step": 49338
    },
    {
      "epoch": 4.244945366944851,
      "grad_norm": 0.77734375,
      "learning_rate": 5.5266546199288505e-05,
      "loss": 0.6158,
      "step": 49339
    },
    {
      "epoch": 4.245031403252172,
      "grad_norm": 1.09375,
      "learning_rate": 5.525418924477188e-05,
      "loss": 0.9266,
      "step": 49340
    },
    {
      "epoch": 4.245117439559494,
      "grad_norm": 1.1953125,
      "learning_rate": 5.5241833591050226e-05,
      "loss": 1.4725,
      "step": 49341
    },
    {
      "epoch": 4.245203475866816,
      "grad_norm": 1.9609375,
      "learning_rate": 5.5229479238159684e-05,
      "loss": 1.4257,
      "step": 49342
    },
    {
      "epoch": 4.245289512174137,
      "grad_norm": 1.2578125,
      "learning_rate": 5.52171261861365e-05,
      "loss": 1.4246,
      "step": 49343
    },
    {
      "epoch": 4.2453755484814595,
      "grad_norm": 0.97265625,
      "learning_rate": 5.520477443501659e-05,
      "loss": 0.5835,
      "step": 49344
    },
    {
      "epoch": 4.245461584788781,
      "grad_norm": 1.2265625,
      "learning_rate": 5.519242398483637e-05,
      "loss": 0.736,
      "step": 49345
    },
    {
      "epoch": 4.245547621096103,
      "grad_norm": 1.1640625,
      "learning_rate": 5.5180074835631646e-05,
      "loss": 1.0164,
      "step": 49346
    },
    {
      "epoch": 4.245633657403424,
      "grad_norm": 0.86328125,
      "learning_rate": 5.5167726987438716e-05,
      "loss": 0.6067,
      "step": 49347
    },
    {
      "epoch": 4.245719693710746,
      "grad_norm": 46.0,
      "learning_rate": 5.515538044029378e-05,
      "loss": 1.1061,
      "step": 49348
    },
    {
      "epoch": 4.245805730018068,
      "grad_norm": 2.0625,
      "learning_rate": 5.514303519423269e-05,
      "loss": 1.6852,
      "step": 49349
    },
    {
      "epoch": 4.245891766325389,
      "grad_norm": 1.0234375,
      "learning_rate": 5.5130691249291764e-05,
      "loss": 1.389,
      "step": 49350
    },
    {
      "epoch": 4.245977802632711,
      "grad_norm": 0.99609375,
      "learning_rate": 5.5118348605507074e-05,
      "loss": 1.1975,
      "step": 49351
    },
    {
      "epoch": 4.2460638389400325,
      "grad_norm": 0.87890625,
      "learning_rate": 5.5106007262914646e-05,
      "loss": 1.1985,
      "step": 49352
    },
    {
      "epoch": 4.2461498752473545,
      "grad_norm": 4.0,
      "learning_rate": 5.509366722155062e-05,
      "loss": 1.0504,
      "step": 49353
    },
    {
      "epoch": 4.246235911554676,
      "grad_norm": 1.0,
      "learning_rate": 5.508132848145114e-05,
      "loss": 0.6876,
      "step": 49354
    },
    {
      "epoch": 4.246321947861998,
      "grad_norm": 1.6171875,
      "learning_rate": 5.5068991042652174e-05,
      "loss": 1.2093,
      "step": 49355
    },
    {
      "epoch": 4.24640798416932,
      "grad_norm": 1.0078125,
      "learning_rate": 5.5056654905189964e-05,
      "loss": 1.1405,
      "step": 49356
    },
    {
      "epoch": 4.246494020476641,
      "grad_norm": 1.4609375,
      "learning_rate": 5.504432006910043e-05,
      "loss": 1.4078,
      "step": 49357
    },
    {
      "epoch": 4.246580056783963,
      "grad_norm": 79.5,
      "learning_rate": 5.503198653441971e-05,
      "loss": 1.3258,
      "step": 49358
    },
    {
      "epoch": 4.246666093091284,
      "grad_norm": 1.234375,
      "learning_rate": 5.5019654301183996e-05,
      "loss": 1.3273,
      "step": 49359
    },
    {
      "epoch": 4.246752129398606,
      "grad_norm": 6.09375,
      "learning_rate": 5.50073233694291e-05,
      "loss": 1.0555,
      "step": 49360
    },
    {
      "epoch": 4.2468381657059275,
      "grad_norm": 1.421875,
      "learning_rate": 5.499499373919131e-05,
      "loss": 0.8944,
      "step": 49361
    },
    {
      "epoch": 4.24692420201325,
      "grad_norm": 0.9375,
      "learning_rate": 5.498266541050667e-05,
      "loss": 1.3506,
      "step": 49362
    },
    {
      "epoch": 4.247010238320572,
      "grad_norm": 1.2109375,
      "learning_rate": 5.497033838341109e-05,
      "loss": 1.2523,
      "step": 49363
    },
    {
      "epoch": 4.247096274627893,
      "grad_norm": 1.1640625,
      "learning_rate": 5.4958012657940706e-05,
      "loss": 1.663,
      "step": 49364
    },
    {
      "epoch": 4.247182310935215,
      "grad_norm": 1.359375,
      "learning_rate": 5.4945688234131655e-05,
      "loss": 1.8035,
      "step": 49365
    },
    {
      "epoch": 4.247268347242536,
      "grad_norm": 1.140625,
      "learning_rate": 5.493336511201985e-05,
      "loss": 0.8252,
      "step": 49366
    },
    {
      "epoch": 4.247354383549858,
      "grad_norm": 63.25,
      "learning_rate": 5.492104329164149e-05,
      "loss": 0.8527,
      "step": 49367
    },
    {
      "epoch": 4.247440419857179,
      "grad_norm": 1.0625,
      "learning_rate": 5.490872277303238e-05,
      "loss": 0.538,
      "step": 49368
    },
    {
      "epoch": 4.247526456164501,
      "grad_norm": 0.9921875,
      "learning_rate": 5.489640355622877e-05,
      "loss": 0.8089,
      "step": 49369
    },
    {
      "epoch": 4.2476124924718235,
      "grad_norm": 0.79296875,
      "learning_rate": 5.4884085641266626e-05,
      "loss": 0.4605,
      "step": 49370
    },
    {
      "epoch": 4.247698528779145,
      "grad_norm": 1.2890625,
      "learning_rate": 5.487176902818192e-05,
      "loss": 0.9752,
      "step": 49371
    },
    {
      "epoch": 4.247784565086467,
      "grad_norm": 1.40625,
      "learning_rate": 5.4859453717010676e-05,
      "loss": 1.715,
      "step": 49372
    },
    {
      "epoch": 4.247870601393788,
      "grad_norm": 1.15625,
      "learning_rate": 5.48471397077891e-05,
      "loss": 1.6133,
      "step": 49373
    },
    {
      "epoch": 4.24795663770111,
      "grad_norm": 1.5234375,
      "learning_rate": 5.483482700055292e-05,
      "loss": 0.9439,
      "step": 49374
    },
    {
      "epoch": 4.248042674008431,
      "grad_norm": 1.5390625,
      "learning_rate": 5.482251559533829e-05,
      "loss": 1.081,
      "step": 49375
    },
    {
      "epoch": 4.248128710315753,
      "grad_norm": 1.09375,
      "learning_rate": 5.481020549218124e-05,
      "loss": 1.0574,
      "step": 49376
    },
    {
      "epoch": 4.248214746623075,
      "grad_norm": 1488.0,
      "learning_rate": 5.479789669111773e-05,
      "loss": 0.978,
      "step": 49377
    },
    {
      "epoch": 4.2483007829303965,
      "grad_norm": 0.85546875,
      "learning_rate": 5.4785589192183904e-05,
      "loss": 0.7066,
      "step": 49378
    },
    {
      "epoch": 4.2483868192377185,
      "grad_norm": 1.09375,
      "learning_rate": 5.4773282995415566e-05,
      "loss": 1.1504,
      "step": 49379
    },
    {
      "epoch": 4.24847285554504,
      "grad_norm": 1.484375,
      "learning_rate": 5.476097810084874e-05,
      "loss": 0.9667,
      "step": 49380
    },
    {
      "epoch": 4.248558891852362,
      "grad_norm": 1.5546875,
      "learning_rate": 5.474867450851956e-05,
      "loss": 1.5014,
      "step": 49381
    },
    {
      "epoch": 4.248644928159683,
      "grad_norm": 1.3984375,
      "learning_rate": 5.4736372218463834e-05,
      "loss": 0.8426,
      "step": 49382
    },
    {
      "epoch": 4.248730964467005,
      "grad_norm": 1.0703125,
      "learning_rate": 5.47240712307176e-05,
      "loss": 2.045,
      "step": 49383
    },
    {
      "epoch": 4.248817000774327,
      "grad_norm": 0.96875,
      "learning_rate": 5.471177154531698e-05,
      "loss": 1.1852,
      "step": 49384
    },
    {
      "epoch": 4.248903037081648,
      "grad_norm": 1.3828125,
      "learning_rate": 5.469947316229773e-05,
      "loss": 1.0347,
      "step": 49385
    },
    {
      "epoch": 4.24898907338897,
      "grad_norm": 0.97265625,
      "learning_rate": 5.468717608169599e-05,
      "loss": 1.6156,
      "step": 49386
    },
    {
      "epoch": 4.2490751096962915,
      "grad_norm": 1.4609375,
      "learning_rate": 5.467488030354756e-05,
      "loss": 1.6265,
      "step": 49387
    },
    {
      "epoch": 4.249161146003614,
      "grad_norm": 3.921875,
      "learning_rate": 5.4662585827888525e-05,
      "loss": 1.7011,
      "step": 49388
    },
    {
      "epoch": 4.249247182310935,
      "grad_norm": 1.1953125,
      "learning_rate": 5.4650292654754854e-05,
      "loss": 1.6628,
      "step": 49389
    },
    {
      "epoch": 4.249333218618257,
      "grad_norm": 1.125,
      "learning_rate": 5.4638000784182406e-05,
      "loss": 0.7914,
      "step": 49390
    },
    {
      "epoch": 4.249419254925579,
      "grad_norm": 1.484375,
      "learning_rate": 5.462571021620716e-05,
      "loss": 1.4725,
      "step": 49391
    },
    {
      "epoch": 4.2495052912329,
      "grad_norm": 1.2578125,
      "learning_rate": 5.461342095086519e-05,
      "loss": 1.3686,
      "step": 49392
    },
    {
      "epoch": 4.249591327540222,
      "grad_norm": 207.0,
      "learning_rate": 5.460113298819225e-05,
      "loss": 1.2773,
      "step": 49393
    },
    {
      "epoch": 4.249677363847543,
      "grad_norm": 107.5,
      "learning_rate": 5.458884632822436e-05,
      "loss": 1.0172,
      "step": 49394
    },
    {
      "epoch": 4.249763400154865,
      "grad_norm": 1.1015625,
      "learning_rate": 5.4576560970997614e-05,
      "loss": 0.5976,
      "step": 49395
    },
    {
      "epoch": 4.249849436462187,
      "grad_norm": 1.015625,
      "learning_rate": 5.4564276916547636e-05,
      "loss": 0.8399,
      "step": 49396
    },
    {
      "epoch": 4.249935472769509,
      "grad_norm": 1.3046875,
      "learning_rate": 5.455199416491063e-05,
      "loss": 1.3491,
      "step": 49397
    },
    {
      "epoch": 4.250021509076831,
      "grad_norm": 1.2109375,
      "learning_rate": 5.453971271612235e-05,
      "loss": 1.3931,
      "step": 49398
    },
    {
      "epoch": 4.250107545384152,
      "grad_norm": 1.015625,
      "learning_rate": 5.452743257021875e-05,
      "loss": 1.5136,
      "step": 49399
    },
    {
      "epoch": 4.250193581691474,
      "grad_norm": 1.0390625,
      "learning_rate": 5.4515153727235865e-05,
      "loss": 1.0769,
      "step": 49400
    },
    {
      "epoch": 4.250279617998795,
      "grad_norm": 0.94140625,
      "learning_rate": 5.4502876187209396e-05,
      "loss": 0.7476,
      "step": 49401
    },
    {
      "epoch": 4.250365654306117,
      "grad_norm": 0.84375,
      "learning_rate": 5.449059995017541e-05,
      "loss": 1.1697,
      "step": 49402
    },
    {
      "epoch": 4.250365654306117,
      "eval_loss": 1.067977786064148,
      "eval_runtime": 378.3211,
      "eval_samples_per_second": 12.289,
      "eval_steps_per_second": 1.538,
      "step": 49402
    },
    {
      "epoch": 4.250451690613438,
      "grad_norm": 47.75,
      "learning_rate": 5.447832501616984e-05,
      "loss": 0.8901,
      "step": 49403
    },
    {
      "epoch": 4.2505377269207605,
      "grad_norm": 1.09375,
      "learning_rate": 5.446605138522842e-05,
      "loss": 0.6387,
      "step": 49404
    },
    {
      "epoch": 4.2506237632280826,
      "grad_norm": 0.75,
      "learning_rate": 5.445377905738713e-05,
      "loss": 0.5273,
      "step": 49405
    },
    {
      "epoch": 4.250709799535404,
      "grad_norm": 2.5625,
      "learning_rate": 5.4441508032682054e-05,
      "loss": 1.4753,
      "step": 49406
    },
    {
      "epoch": 4.250795835842726,
      "grad_norm": 1.1328125,
      "learning_rate": 5.442923831114871e-05,
      "loss": 1.715,
      "step": 49407
    },
    {
      "epoch": 4.250881872150047,
      "grad_norm": 1.375,
      "learning_rate": 5.441696989282324e-05,
      "loss": 1.3659,
      "step": 49408
    },
    {
      "epoch": 4.250967908457369,
      "grad_norm": 1.1171875,
      "learning_rate": 5.4404702777741624e-05,
      "loss": 1.1385,
      "step": 49409
    },
    {
      "epoch": 4.25105394476469,
      "grad_norm": 0.9140625,
      "learning_rate": 5.439243696593943e-05,
      "loss": 0.6569,
      "step": 49410
    },
    {
      "epoch": 4.251139981072012,
      "grad_norm": 1.2421875,
      "learning_rate": 5.43801724574528e-05,
      "loss": 1.4539,
      "step": 49411
    },
    {
      "epoch": 4.251226017379334,
      "grad_norm": 0.95703125,
      "learning_rate": 5.436790925231738e-05,
      "loss": 0.7445,
      "step": 49412
    },
    {
      "epoch": 4.2513120536866555,
      "grad_norm": 0.87109375,
      "learning_rate": 5.4355647350569136e-05,
      "loss": 0.5032,
      "step": 49413
    },
    {
      "epoch": 4.251398089993978,
      "grad_norm": 0.75390625,
      "learning_rate": 5.4343386752243984e-05,
      "loss": 0.7788,
      "step": 49414
    },
    {
      "epoch": 4.251484126301299,
      "grad_norm": 1.2421875,
      "learning_rate": 5.433112745737778e-05,
      "loss": 1.0013,
      "step": 49415
    },
    {
      "epoch": 4.251570162608621,
      "grad_norm": 2.703125,
      "learning_rate": 5.431886946600639e-05,
      "loss": 1.1593,
      "step": 49416
    },
    {
      "epoch": 4.251656198915943,
      "grad_norm": 1.28125,
      "learning_rate": 5.430661277816556e-05,
      "loss": 1.4393,
      "step": 49417
    },
    {
      "epoch": 4.251742235223264,
      "grad_norm": 0.90234375,
      "learning_rate": 5.4294357393891214e-05,
      "loss": 1.1403,
      "step": 49418
    },
    {
      "epoch": 4.251828271530586,
      "grad_norm": 1.0625,
      "learning_rate": 5.4282103313219314e-05,
      "loss": 1.4419,
      "step": 49419
    },
    {
      "epoch": 4.251914307837907,
      "grad_norm": 0.96484375,
      "learning_rate": 5.426985053618544e-05,
      "loss": 0.6582,
      "step": 49420
    },
    {
      "epoch": 4.252000344145229,
      "grad_norm": 1.1875,
      "learning_rate": 5.4257599062825516e-05,
      "loss": 0.7336,
      "step": 49421
    },
    {
      "epoch": 4.252086380452551,
      "grad_norm": 0.91796875,
      "learning_rate": 5.4245348893175614e-05,
      "loss": 0.721,
      "step": 49422
    },
    {
      "epoch": 4.252172416759873,
      "grad_norm": 0.8125,
      "learning_rate": 5.423310002727122e-05,
      "loss": 0.7417,
      "step": 49423
    },
    {
      "epoch": 4.252258453067195,
      "grad_norm": 1.546875,
      "learning_rate": 5.4220852465148295e-05,
      "loss": 1.2459,
      "step": 49424
    },
    {
      "epoch": 4.252344489374516,
      "grad_norm": 1.5234375,
      "learning_rate": 5.4208606206842814e-05,
      "loss": 0.803,
      "step": 49425
    },
    {
      "epoch": 4.252430525681838,
      "grad_norm": 1800.0,
      "learning_rate": 5.419636125239036e-05,
      "loss": 0.9974,
      "step": 49426
    },
    {
      "epoch": 4.252516561989159,
      "grad_norm": 1.296875,
      "learning_rate": 5.41841176018269e-05,
      "loss": 1.8409,
      "step": 49427
    },
    {
      "epoch": 4.252602598296481,
      "grad_norm": 0.9296875,
      "learning_rate": 5.417187525518813e-05,
      "loss": 1.3539,
      "step": 49428
    },
    {
      "epoch": 4.252688634603802,
      "grad_norm": 1.0703125,
      "learning_rate": 5.415963421250991e-05,
      "loss": 1.5505,
      "step": 49429
    },
    {
      "epoch": 4.2527746709111245,
      "grad_norm": 1.15625,
      "learning_rate": 5.41473944738281e-05,
      "loss": 0.9757,
      "step": 49430
    },
    {
      "epoch": 4.252860707218447,
      "grad_norm": 1.1484375,
      "learning_rate": 5.41351560391784e-05,
      "loss": 1.2363,
      "step": 49431
    },
    {
      "epoch": 4.252946743525768,
      "grad_norm": 1.234375,
      "learning_rate": 5.412291890859666e-05,
      "loss": 0.9048,
      "step": 49432
    },
    {
      "epoch": 4.25303277983309,
      "grad_norm": 1.3125,
      "learning_rate": 5.411068308211875e-05,
      "loss": 1.4693,
      "step": 49433
    },
    {
      "epoch": 4.253118816140411,
      "grad_norm": 1.2578125,
      "learning_rate": 5.4098448559780244e-05,
      "loss": 1.6133,
      "step": 49434
    },
    {
      "epoch": 4.253204852447733,
      "grad_norm": 0.94921875,
      "learning_rate": 5.408621534161712e-05,
      "loss": 0.4359,
      "step": 49435
    },
    {
      "epoch": 4.253290888755054,
      "grad_norm": 1.0546875,
      "learning_rate": 5.407398342766512e-05,
      "loss": 1.3274,
      "step": 49436
    },
    {
      "epoch": 4.253376925062376,
      "grad_norm": 1.46875,
      "learning_rate": 5.4061752817959944e-05,
      "loss": 1.367,
      "step": 49437
    },
    {
      "epoch": 4.253462961369698,
      "grad_norm": 1.0546875,
      "learning_rate": 5.40495235125375e-05,
      "loss": 1.0095,
      "step": 49438
    },
    {
      "epoch": 4.2535489976770195,
      "grad_norm": 1.4921875,
      "learning_rate": 5.403729551143333e-05,
      "loss": 0.856,
      "step": 49439
    },
    {
      "epoch": 4.253635033984342,
      "grad_norm": 1.078125,
      "learning_rate": 5.4025068814683334e-05,
      "loss": 0.7135,
      "step": 49440
    },
    {
      "epoch": 4.253721070291663,
      "grad_norm": 0.9921875,
      "learning_rate": 5.4012843422323434e-05,
      "loss": 1.7405,
      "step": 49441
    },
    {
      "epoch": 4.253807106598985,
      "grad_norm": 1.40625,
      "learning_rate": 5.4000619334389103e-05,
      "loss": 1.4952,
      "step": 49442
    },
    {
      "epoch": 4.253893142906306,
      "grad_norm": 1.015625,
      "learning_rate": 5.3988396550916194e-05,
      "loss": 0.6568,
      "step": 49443
    },
    {
      "epoch": 4.253979179213628,
      "grad_norm": 1.203125,
      "learning_rate": 5.397617507194058e-05,
      "loss": 1.3783,
      "step": 49444
    },
    {
      "epoch": 4.25406521552095,
      "grad_norm": 4.0625,
      "learning_rate": 5.3963954897497837e-05,
      "loss": 0.9692,
      "step": 49445
    },
    {
      "epoch": 4.254151251828271,
      "grad_norm": 0.94921875,
      "learning_rate": 5.395173602762376e-05,
      "loss": 1.5087,
      "step": 49446
    },
    {
      "epoch": 4.254237288135593,
      "grad_norm": 1.046875,
      "learning_rate": 5.393951846235418e-05,
      "loss": 1.13,
      "step": 49447
    },
    {
      "epoch": 4.254323324442915,
      "grad_norm": 1.1875,
      "learning_rate": 5.3927302201724705e-05,
      "loss": 1.3302,
      "step": 49448
    },
    {
      "epoch": 4.254409360750237,
      "grad_norm": 1.3828125,
      "learning_rate": 5.39150872457711e-05,
      "loss": 0.8515,
      "step": 49449
    },
    {
      "epoch": 4.254495397057559,
      "grad_norm": 1.171875,
      "learning_rate": 5.390287359452906e-05,
      "loss": 1.6913,
      "step": 49450
    },
    {
      "epoch": 4.25458143336488,
      "grad_norm": 1.0078125,
      "learning_rate": 5.3890661248034377e-05,
      "loss": 1.5451,
      "step": 49451
    },
    {
      "epoch": 4.254667469672202,
      "grad_norm": 1.0546875,
      "learning_rate": 5.387845020632287e-05,
      "loss": 0.7979,
      "step": 49452
    },
    {
      "epoch": 4.254753505979523,
      "grad_norm": 1.25,
      "learning_rate": 5.386624046943006e-05,
      "loss": 1.1424,
      "step": 49453
    },
    {
      "epoch": 4.254839542286845,
      "grad_norm": 1.0,
      "learning_rate": 5.385403203739175e-05,
      "loss": 0.7182,
      "step": 49454
    },
    {
      "epoch": 4.254925578594166,
      "grad_norm": 0.875,
      "learning_rate": 5.384182491024359e-05,
      "loss": 0.5643,
      "step": 49455
    },
    {
      "epoch": 4.2550116149014885,
      "grad_norm": 0.96484375,
      "learning_rate": 5.3829619088021254e-05,
      "loss": 0.5979,
      "step": 49456
    },
    {
      "epoch": 4.255097651208811,
      "grad_norm": 0.91015625,
      "learning_rate": 5.381741457076067e-05,
      "loss": 0.8806,
      "step": 49457
    },
    {
      "epoch": 4.255183687516132,
      "grad_norm": 1.2734375,
      "learning_rate": 5.380521135849725e-05,
      "loss": 1.5333,
      "step": 49458
    },
    {
      "epoch": 4.255269723823454,
      "grad_norm": 1.2734375,
      "learning_rate": 5.379300945126681e-05,
      "loss": 0.8744,
      "step": 49459
    },
    {
      "epoch": 4.255355760130775,
      "grad_norm": 1.234375,
      "learning_rate": 5.3780808849105136e-05,
      "loss": 1.1143,
      "step": 49460
    },
    {
      "epoch": 4.255441796438097,
      "grad_norm": 1.578125,
      "learning_rate": 5.3768609552047655e-05,
      "loss": 1.3863,
      "step": 49461
    },
    {
      "epoch": 4.255527832745418,
      "grad_norm": 3.125,
      "learning_rate": 5.375641156013028e-05,
      "loss": 1.191,
      "step": 49462
    },
    {
      "epoch": 4.25561386905274,
      "grad_norm": 0.90625,
      "learning_rate": 5.374421487338865e-05,
      "loss": 0.407,
      "step": 49463
    },
    {
      "epoch": 4.255699905360062,
      "grad_norm": 2.59375,
      "learning_rate": 5.373201949185835e-05,
      "loss": 1.6566,
      "step": 49464
    },
    {
      "epoch": 4.2557859416673836,
      "grad_norm": 1.03125,
      "learning_rate": 5.3719825415575074e-05,
      "loss": 1.2161,
      "step": 49465
    },
    {
      "epoch": 4.255871977974706,
      "grad_norm": 0.97265625,
      "learning_rate": 5.3707632644574625e-05,
      "loss": 1.5235,
      "step": 49466
    },
    {
      "epoch": 4.255958014282027,
      "grad_norm": 0.8984375,
      "learning_rate": 5.369544117889247e-05,
      "loss": 1.4644,
      "step": 49467
    },
    {
      "epoch": 4.256044050589349,
      "grad_norm": 1.09375,
      "learning_rate": 5.3683251018564415e-05,
      "loss": 1.036,
      "step": 49468
    },
    {
      "epoch": 4.25613008689667,
      "grad_norm": 1.0234375,
      "learning_rate": 5.367106216362594e-05,
      "loss": 1.1274,
      "step": 49469
    },
    {
      "epoch": 4.256216123203992,
      "grad_norm": 0.96484375,
      "learning_rate": 5.3658874614112844e-05,
      "loss": 0.7429,
      "step": 49470
    },
    {
      "epoch": 4.256302159511314,
      "grad_norm": 1.0,
      "learning_rate": 5.364668837006082e-05,
      "loss": 1.5653,
      "step": 49471
    },
    {
      "epoch": 4.256388195818635,
      "grad_norm": 1.3125,
      "learning_rate": 5.363450343150528e-05,
      "loss": 1.1991,
      "step": 49472
    },
    {
      "epoch": 4.256474232125957,
      "grad_norm": 1.28125,
      "learning_rate": 5.362231979848203e-05,
      "loss": 1.1881,
      "step": 49473
    },
    {
      "epoch": 4.256560268433279,
      "grad_norm": 1.265625,
      "learning_rate": 5.361013747102677e-05,
      "loss": 1.0451,
      "step": 49474
    },
    {
      "epoch": 4.256646304740601,
      "grad_norm": 1.0703125,
      "learning_rate": 5.3597956449174976e-05,
      "loss": 1.6394,
      "step": 49475
    },
    {
      "epoch": 4.256732341047922,
      "grad_norm": 1.1875,
      "learning_rate": 5.3585776732962274e-05,
      "loss": 1.261,
      "step": 49476
    },
    {
      "epoch": 4.256818377355244,
      "grad_norm": 0.87109375,
      "learning_rate": 5.3573598322424534e-05,
      "loss": 1.2159,
      "step": 49477
    },
    {
      "epoch": 4.256904413662566,
      "grad_norm": 15.1875,
      "learning_rate": 5.356142121759705e-05,
      "loss": 0.9237,
      "step": 49478
    },
    {
      "epoch": 4.256990449969887,
      "grad_norm": 32.25,
      "learning_rate": 5.3549245418515634e-05,
      "loss": 0.7949,
      "step": 49479
    },
    {
      "epoch": 4.257076486277209,
      "grad_norm": 1.03125,
      "learning_rate": 5.3537070925215814e-05,
      "loss": 1.7806,
      "step": 49480
    },
    {
      "epoch": 4.25716252258453,
      "grad_norm": 1.3671875,
      "learning_rate": 5.3524897737733223e-05,
      "loss": 0.8489,
      "step": 49481
    },
    {
      "epoch": 4.2572485588918525,
      "grad_norm": 1.3203125,
      "learning_rate": 5.351272585610356e-05,
      "loss": 1.2289,
      "step": 49482
    },
    {
      "epoch": 4.257334595199174,
      "grad_norm": 1.1640625,
      "learning_rate": 5.350055528036224e-05,
      "loss": 1.2103,
      "step": 49483
    },
    {
      "epoch": 4.257420631506496,
      "grad_norm": 2.3125,
      "learning_rate": 5.348838601054495e-05,
      "loss": 1.0155,
      "step": 49484
    },
    {
      "epoch": 4.257506667813818,
      "grad_norm": 1.0859375,
      "learning_rate": 5.3476218046687396e-05,
      "loss": 1.3266,
      "step": 49485
    },
    {
      "epoch": 4.257592704121139,
      "grad_norm": 1.21875,
      "learning_rate": 5.346405138882493e-05,
      "loss": 1.6688,
      "step": 49486
    },
    {
      "epoch": 4.257678740428461,
      "grad_norm": 1.1796875,
      "learning_rate": 5.34518860369933e-05,
      "loss": 0.7219,
      "step": 49487
    },
    {
      "epoch": 4.257764776735782,
      "grad_norm": 211.0,
      "learning_rate": 5.343972199122804e-05,
      "loss": 1.3862,
      "step": 49488
    },
    {
      "epoch": 4.257850813043104,
      "grad_norm": 1.03125,
      "learning_rate": 5.342755925156473e-05,
      "loss": 1.4344,
      "step": 49489
    },
    {
      "epoch": 4.2579368493504255,
      "grad_norm": 1.1796875,
      "learning_rate": 5.341539781803906e-05,
      "loss": 0.9884,
      "step": 49490
    },
    {
      "epoch": 4.258022885657748,
      "grad_norm": 0.9140625,
      "learning_rate": 5.3403237690686446e-05,
      "loss": 1.1568,
      "step": 49491
    },
    {
      "epoch": 4.25810892196507,
      "grad_norm": 0.796875,
      "learning_rate": 5.339107886954247e-05,
      "loss": 1.3893,
      "step": 49492
    },
    {
      "epoch": 4.258194958272391,
      "grad_norm": 1.0078125,
      "learning_rate": 5.3378921354642784e-05,
      "loss": 1.8736,
      "step": 49493
    },
    {
      "epoch": 4.258280994579713,
      "grad_norm": 0.63671875,
      "learning_rate": 5.3366765146022845e-05,
      "loss": 0.3856,
      "step": 49494
    },
    {
      "epoch": 4.258367030887034,
      "grad_norm": 1.1015625,
      "learning_rate": 5.335461024371824e-05,
      "loss": 1.5584,
      "step": 49495
    },
    {
      "epoch": 4.258453067194356,
      "grad_norm": 1.140625,
      "learning_rate": 5.334245664776466e-05,
      "loss": 0.7139,
      "step": 49496
    },
    {
      "epoch": 4.258539103501677,
      "grad_norm": 1.1328125,
      "learning_rate": 5.3330304358197356e-05,
      "loss": 1.336,
      "step": 49497
    },
    {
      "epoch": 4.258625139808999,
      "grad_norm": 1.0390625,
      "learning_rate": 5.3318153375052195e-05,
      "loss": 1.3988,
      "step": 49498
    },
    {
      "epoch": 4.258711176116321,
      "grad_norm": 3.953125,
      "learning_rate": 5.330600369836442e-05,
      "loss": 0.5477,
      "step": 49499
    },
    {
      "epoch": 4.258797212423643,
      "grad_norm": 1.0234375,
      "learning_rate": 5.3293855328169775e-05,
      "loss": 1.3986,
      "step": 49500
    },
    {
      "epoch": 4.258883248730965,
      "grad_norm": 1.265625,
      "learning_rate": 5.3281708264503734e-05,
      "loss": 1.0863,
      "step": 49501
    },
    {
      "epoch": 4.258969285038286,
      "grad_norm": 1.90625,
      "learning_rate": 5.3269562507401784e-05,
      "loss": 0.7114,
      "step": 49502
    },
    {
      "epoch": 4.259055321345608,
      "grad_norm": 50.0,
      "learning_rate": 5.325741805689943e-05,
      "loss": 1.4169,
      "step": 49503
    },
    {
      "epoch": 4.259141357652929,
      "grad_norm": 0.84765625,
      "learning_rate": 5.3245274913032384e-05,
      "loss": 0.556,
      "step": 49504
    },
    {
      "epoch": 4.259227393960251,
      "grad_norm": 1.296875,
      "learning_rate": 5.3233133075835884e-05,
      "loss": 1.3408,
      "step": 49505
    },
    {
      "epoch": 4.259313430267573,
      "grad_norm": 1.078125,
      "learning_rate": 5.3220992545345625e-05,
      "loss": 0.8777,
      "step": 49506
    },
    {
      "epoch": 4.259399466574894,
      "grad_norm": 1.0234375,
      "learning_rate": 5.320885332159714e-05,
      "loss": 1.121,
      "step": 49507
    },
    {
      "epoch": 4.2594855028822165,
      "grad_norm": 1.234375,
      "learning_rate": 5.319671540462573e-05,
      "loss": 1.1206,
      "step": 49508
    },
    {
      "epoch": 4.259571539189538,
      "grad_norm": 1.484375,
      "learning_rate": 5.31845787944672e-05,
      "loss": 1.1565,
      "step": 49509
    },
    {
      "epoch": 4.25965757549686,
      "grad_norm": 1.0703125,
      "learning_rate": 5.317244349115674e-05,
      "loss": 1.0831,
      "step": 49510
    },
    {
      "epoch": 4.259743611804181,
      "grad_norm": 84.0,
      "learning_rate": 5.316030949472994e-05,
      "loss": 0.9385,
      "step": 49511
    },
    {
      "epoch": 4.259829648111503,
      "grad_norm": 22.625,
      "learning_rate": 5.3148176805222494e-05,
      "loss": 1.4204,
      "step": 49512
    },
    {
      "epoch": 4.259915684418825,
      "grad_norm": 1.21875,
      "learning_rate": 5.313604542266959e-05,
      "loss": 1.088,
      "step": 49513
    },
    {
      "epoch": 4.260001720726146,
      "grad_norm": 1.6796875,
      "learning_rate": 5.312391534710687e-05,
      "loss": 1.1783,
      "step": 49514
    },
    {
      "epoch": 4.260087757033468,
      "grad_norm": 1.1953125,
      "learning_rate": 5.3111786578569856e-05,
      "loss": 0.9248,
      "step": 49515
    },
    {
      "epoch": 4.2601737933407895,
      "grad_norm": 1.3359375,
      "learning_rate": 5.309965911709386e-05,
      "loss": 1.7753,
      "step": 49516
    },
    {
      "epoch": 4.260259829648112,
      "grad_norm": 0.8125,
      "learning_rate": 5.3087532962714456e-05,
      "loss": 0.5789,
      "step": 49517
    },
    {
      "epoch": 4.260345865955433,
      "grad_norm": 0.9375,
      "learning_rate": 5.3075408115467184e-05,
      "loss": 1.4198,
      "step": 49518
    },
    {
      "epoch": 4.260431902262755,
      "grad_norm": 1.671875,
      "learning_rate": 5.3063284575387286e-05,
      "loss": 1.14,
      "step": 49519
    },
    {
      "epoch": 4.260517938570077,
      "grad_norm": 1.234375,
      "learning_rate": 5.3051162342510505e-05,
      "loss": 1.6697,
      "step": 49520
    },
    {
      "epoch": 4.260603974877398,
      "grad_norm": 1.2734375,
      "learning_rate": 5.3039041416872046e-05,
      "loss": 1.002,
      "step": 49521
    },
    {
      "epoch": 4.26069001118472,
      "grad_norm": 1.0546875,
      "learning_rate": 5.3026921798507436e-05,
      "loss": 0.678,
      "step": 49522
    },
    {
      "epoch": 4.260776047492041,
      "grad_norm": 0.98046875,
      "learning_rate": 5.3014803487452304e-05,
      "loss": 1.643,
      "step": 49523
    },
    {
      "epoch": 4.260862083799363,
      "grad_norm": 1.1484375,
      "learning_rate": 5.300268648374168e-05,
      "loss": 0.805,
      "step": 49524
    },
    {
      "epoch": 4.2609481201066846,
      "grad_norm": 1.1875,
      "learning_rate": 5.299057078741148e-05,
      "loss": 1.4761,
      "step": 49525
    },
    {
      "epoch": 4.261034156414007,
      "grad_norm": 8.6875,
      "learning_rate": 5.29784563984968e-05,
      "loss": 1.3444,
      "step": 49526
    },
    {
      "epoch": 4.261120192721329,
      "grad_norm": 6.09375,
      "learning_rate": 5.29663433170332e-05,
      "loss": 0.7777,
      "step": 49527
    },
    {
      "epoch": 4.26120622902865,
      "grad_norm": 1.3984375,
      "learning_rate": 5.295423154305623e-05,
      "loss": 1.2886,
      "step": 49528
    },
    {
      "epoch": 4.261292265335972,
      "grad_norm": 3.09375,
      "learning_rate": 5.2942121076601065e-05,
      "loss": 1.393,
      "step": 49529
    },
    {
      "epoch": 4.261378301643293,
      "grad_norm": 4.03125,
      "learning_rate": 5.2930011917703305e-05,
      "loss": 0.6718,
      "step": 49530
    },
    {
      "epoch": 4.261464337950615,
      "grad_norm": 1.796875,
      "learning_rate": 5.2917904066398345e-05,
      "loss": 1.1832,
      "step": 49531
    },
    {
      "epoch": 4.261550374257937,
      "grad_norm": 0.921875,
      "learning_rate": 5.2905797522721456e-05,
      "loss": 0.7695,
      "step": 49532
    },
    {
      "epoch": 4.261636410565258,
      "grad_norm": 0.9765625,
      "learning_rate": 5.289369228670821e-05,
      "loss": 0.4561,
      "step": 49533
    },
    {
      "epoch": 4.2617224468725805,
      "grad_norm": 1.1171875,
      "learning_rate": 5.288158835839402e-05,
      "loss": 1.2148,
      "step": 49534
    },
    {
      "epoch": 4.261808483179902,
      "grad_norm": 0.88671875,
      "learning_rate": 5.286948573781414e-05,
      "loss": 1.1821,
      "step": 49535
    },
    {
      "epoch": 4.261894519487224,
      "grad_norm": 1.15625,
      "learning_rate": 5.285738442500404e-05,
      "loss": 1.235,
      "step": 49536
    },
    {
      "epoch": 4.261980555794545,
      "grad_norm": 1.2265625,
      "learning_rate": 5.2845284419999196e-05,
      "loss": 0.6938,
      "step": 49537
    },
    {
      "epoch": 4.262066592101867,
      "grad_norm": 1.0390625,
      "learning_rate": 5.283318572283491e-05,
      "loss": 1.5904,
      "step": 49538
    },
    {
      "epoch": 4.262152628409189,
      "grad_norm": 1.46875,
      "learning_rate": 5.2821088333546594e-05,
      "loss": 0.8372,
      "step": 49539
    },
    {
      "epoch": 4.26223866471651,
      "grad_norm": 5.65625,
      "learning_rate": 5.280899225216956e-05,
      "loss": 1.1642,
      "step": 49540
    },
    {
      "epoch": 4.262324701023832,
      "grad_norm": 1.3203125,
      "learning_rate": 5.279689747873928e-05,
      "loss": 1.1625,
      "step": 49541
    },
    {
      "epoch": 4.2624107373311535,
      "grad_norm": 5.28125,
      "learning_rate": 5.278480401329117e-05,
      "loss": 1.1439,
      "step": 49542
    },
    {
      "epoch": 4.262496773638476,
      "grad_norm": 0.875,
      "learning_rate": 5.277271185586047e-05,
      "loss": 1.6055,
      "step": 49543
    },
    {
      "epoch": 4.262582809945797,
      "grad_norm": 1.3359375,
      "learning_rate": 5.276062100648255e-05,
      "loss": 0.6958,
      "step": 49544
    },
    {
      "epoch": 4.262668846253119,
      "grad_norm": 1.453125,
      "learning_rate": 5.274853146519293e-05,
      "loss": 1.3648,
      "step": 49545
    },
    {
      "epoch": 4.262754882560441,
      "grad_norm": 1.4609375,
      "learning_rate": 5.273644323202681e-05,
      "loss": 0.4832,
      "step": 49546
    },
    {
      "epoch": 4.262840918867762,
      "grad_norm": 2.3125,
      "learning_rate": 5.272435630701961e-05,
      "loss": 1.3365,
      "step": 49547
    },
    {
      "epoch": 4.262926955175084,
      "grad_norm": 1.03125,
      "learning_rate": 5.271227069020673e-05,
      "loss": 0.9027,
      "step": 49548
    },
    {
      "epoch": 4.263012991482405,
      "grad_norm": 1.1171875,
      "learning_rate": 5.2700186381623385e-05,
      "loss": 0.9734,
      "step": 49549
    },
    {
      "epoch": 4.263099027789727,
      "grad_norm": 1.1328125,
      "learning_rate": 5.2688103381305084e-05,
      "loss": 0.8609,
      "step": 49550
    },
    {
      "epoch": 4.263185064097049,
      "grad_norm": 1.046875,
      "learning_rate": 5.267602168928703e-05,
      "loss": 0.8142,
      "step": 49551
    },
    {
      "epoch": 4.263271100404371,
      "grad_norm": 1.046875,
      "learning_rate": 5.266394130560459e-05,
      "loss": 0.8883,
      "step": 49552
    },
    {
      "epoch": 4.263357136711693,
      "grad_norm": 0.79296875,
      "learning_rate": 5.265186223029322e-05,
      "loss": 0.753,
      "step": 49553
    },
    {
      "epoch": 4.263443173019014,
      "grad_norm": 1.0546875,
      "learning_rate": 5.2639784463388015e-05,
      "loss": 1.2487,
      "step": 49554
    },
    {
      "epoch": 4.263529209326336,
      "grad_norm": 0.84375,
      "learning_rate": 5.26277080049245e-05,
      "loss": 0.7245,
      "step": 49555
    },
    {
      "epoch": 4.263615245633657,
      "grad_norm": 1.1953125,
      "learning_rate": 5.261563285493798e-05,
      "loss": 1.742,
      "step": 49556
    },
    {
      "epoch": 4.263701281940979,
      "grad_norm": 1.1328125,
      "learning_rate": 5.260355901346364e-05,
      "loss": 0.6092,
      "step": 49557
    },
    {
      "epoch": 4.2637873182483,
      "grad_norm": 0.83984375,
      "learning_rate": 5.2591486480536856e-05,
      "loss": 0.72,
      "step": 49558
    },
    {
      "epoch": 4.263873354555622,
      "grad_norm": 1.09375,
      "learning_rate": 5.2579415256193085e-05,
      "loss": 0.9069,
      "step": 49559
    },
    {
      "epoch": 4.2639593908629445,
      "grad_norm": 0.92578125,
      "learning_rate": 5.256734534046742e-05,
      "loss": 0.9755,
      "step": 49560
    },
    {
      "epoch": 4.264045427170266,
      "grad_norm": 1.40625,
      "learning_rate": 5.255527673339522e-05,
      "loss": 0.7164,
      "step": 49561
    },
    {
      "epoch": 4.264131463477588,
      "grad_norm": 0.97265625,
      "learning_rate": 5.254320943501184e-05,
      "loss": 1.4909,
      "step": 49562
    },
    {
      "epoch": 4.264217499784909,
      "grad_norm": 1.0703125,
      "learning_rate": 5.253114344535248e-05,
      "loss": 1.3305,
      "step": 49563
    },
    {
      "epoch": 4.264303536092231,
      "grad_norm": 1.5625,
      "learning_rate": 5.2519078764452666e-05,
      "loss": 1.3782,
      "step": 49564
    },
    {
      "epoch": 4.264389572399553,
      "grad_norm": 6.4375,
      "learning_rate": 5.250701539234737e-05,
      "loss": 0.9433,
      "step": 49565
    },
    {
      "epoch": 4.264475608706874,
      "grad_norm": 1.0234375,
      "learning_rate": 5.249495332907217e-05,
      "loss": 1.3025,
      "step": 49566
    },
    {
      "epoch": 4.264561645014196,
      "grad_norm": 0.86328125,
      "learning_rate": 5.2482892574662046e-05,
      "loss": 1.126,
      "step": 49567
    },
    {
      "epoch": 4.2646476813215175,
      "grad_norm": 1.1484375,
      "learning_rate": 5.247083312915246e-05,
      "loss": 1.1537,
      "step": 49568
    },
    {
      "epoch": 4.26473371762884,
      "grad_norm": 1.1015625,
      "learning_rate": 5.245877499257873e-05,
      "loss": 0.516,
      "step": 49569
    },
    {
      "epoch": 4.264819753936161,
      "grad_norm": 54.25,
      "learning_rate": 5.244671816497598e-05,
      "loss": 1.3416,
      "step": 49570
    },
    {
      "epoch": 4.264905790243483,
      "grad_norm": 1.3671875,
      "learning_rate": 5.243466264637947e-05,
      "loss": 1.0629,
      "step": 49571
    },
    {
      "epoch": 4.264991826550805,
      "grad_norm": 0.95703125,
      "learning_rate": 5.242260843682467e-05,
      "loss": 1.0084,
      "step": 49572
    },
    {
      "epoch": 4.265077862858126,
      "grad_norm": 1.1484375,
      "learning_rate": 5.241055553634655e-05,
      "loss": 0.7183,
      "step": 49573
    },
    {
      "epoch": 4.265163899165448,
      "grad_norm": 0.80859375,
      "learning_rate": 5.239850394498058e-05,
      "loss": 0.5482,
      "step": 49574
    },
    {
      "epoch": 4.265249935472769,
      "grad_norm": 0.81640625,
      "learning_rate": 5.2386453662761955e-05,
      "loss": 0.6405,
      "step": 49575
    },
    {
      "epoch": 4.265335971780091,
      "grad_norm": 1.046875,
      "learning_rate": 5.237440468972582e-05,
      "loss": 0.4801,
      "step": 49576
    },
    {
      "epoch": 4.265422008087413,
      "grad_norm": 1.0078125,
      "learning_rate": 5.236235702590753e-05,
      "loss": 0.6245,
      "step": 49577
    },
    {
      "epoch": 4.265508044394735,
      "grad_norm": 1.171875,
      "learning_rate": 5.235031067134233e-05,
      "loss": 0.9247,
      "step": 49578
    },
    {
      "epoch": 4.265594080702057,
      "grad_norm": 0.734375,
      "learning_rate": 5.233826562606536e-05,
      "loss": 0.5617,
      "step": 49579
    },
    {
      "epoch": 4.265680117009378,
      "grad_norm": 1.09375,
      "learning_rate": 5.2326221890111936e-05,
      "loss": 0.9697,
      "step": 49580
    },
    {
      "epoch": 4.2657661533167,
      "grad_norm": 1.046875,
      "learning_rate": 5.231417946351719e-05,
      "loss": 1.3555,
      "step": 49581
    },
    {
      "epoch": 4.265852189624021,
      "grad_norm": 1.1484375,
      "learning_rate": 5.230213834631642e-05,
      "loss": 1.0019,
      "step": 49582
    },
    {
      "epoch": 4.265938225931343,
      "grad_norm": 0.94921875,
      "learning_rate": 5.229009853854489e-05,
      "loss": 1.0452,
      "step": 49583
    },
    {
      "epoch": 4.266024262238664,
      "grad_norm": 5.875,
      "learning_rate": 5.227806004023761e-05,
      "loss": 0.8857,
      "step": 49584
    },
    {
      "epoch": 4.2661102985459864,
      "grad_norm": 1.2734375,
      "learning_rate": 5.2266022851429954e-05,
      "loss": 1.6584,
      "step": 49585
    },
    {
      "epoch": 4.2661963348533085,
      "grad_norm": 0.859375,
      "learning_rate": 5.2253986972157216e-05,
      "loss": 0.4373,
      "step": 49586
    },
    {
      "epoch": 4.26628237116063,
      "grad_norm": 0.99609375,
      "learning_rate": 5.2241952402454384e-05,
      "loss": 0.7642,
      "step": 49587
    },
    {
      "epoch": 4.266368407467952,
      "grad_norm": 1.2265625,
      "learning_rate": 5.222991914235675e-05,
      "loss": 1.0666,
      "step": 49588
    },
    {
      "epoch": 4.266454443775273,
      "grad_norm": 1.0,
      "learning_rate": 5.221788719189963e-05,
      "loss": 0.6884,
      "step": 49589
    },
    {
      "epoch": 4.266540480082595,
      "grad_norm": 1.390625,
      "learning_rate": 5.220585655111798e-05,
      "loss": 1.0553,
      "step": 49590
    },
    {
      "epoch": 4.266626516389916,
      "grad_norm": 0.92578125,
      "learning_rate": 5.219382722004723e-05,
      "loss": 1.046,
      "step": 49591
    },
    {
      "epoch": 4.266712552697238,
      "grad_norm": 1.046875,
      "learning_rate": 5.2181799198722345e-05,
      "loss": 0.8663,
      "step": 49592
    },
    {
      "epoch": 4.26679858900456,
      "grad_norm": 1.1015625,
      "learning_rate": 5.216977248717858e-05,
      "loss": 1.2467,
      "step": 49593
    },
    {
      "epoch": 4.2668846253118815,
      "grad_norm": 1.0546875,
      "learning_rate": 5.215774708545124e-05,
      "loss": 1.5059,
      "step": 49594
    },
    {
      "epoch": 4.266970661619204,
      "grad_norm": 1.140625,
      "learning_rate": 5.214572299357528e-05,
      "loss": 1.0365,
      "step": 49595
    },
    {
      "epoch": 4.267056697926525,
      "grad_norm": 1.0546875,
      "learning_rate": 5.213370021158603e-05,
      "loss": 1.8164,
      "step": 49596
    },
    {
      "epoch": 4.267142734233847,
      "grad_norm": 1.0625,
      "learning_rate": 5.212167873951862e-05,
      "loss": 1.0399,
      "step": 49597
    },
    {
      "epoch": 4.267228770541168,
      "grad_norm": 1.1875,
      "learning_rate": 5.210965857740807e-05,
      "loss": 1.2797,
      "step": 49598
    },
    {
      "epoch": 4.26731480684849,
      "grad_norm": 1.2421875,
      "learning_rate": 5.2097639725289804e-05,
      "loss": 1.6192,
      "step": 49599
    },
    {
      "epoch": 4.267400843155812,
      "grad_norm": 1.3203125,
      "learning_rate": 5.2085622183198734e-05,
      "loss": 0.7933,
      "step": 49600
    },
    {
      "epoch": 4.267486879463133,
      "grad_norm": 1.1171875,
      "learning_rate": 5.207360595117011e-05,
      "loss": 1.4639,
      "step": 49601
    },
    {
      "epoch": 4.267572915770455,
      "grad_norm": 1.1484375,
      "learning_rate": 5.206159102923918e-05,
      "loss": 0.6698,
      "step": 49602
    },
    {
      "epoch": 4.267658952077777,
      "grad_norm": 0.84765625,
      "learning_rate": 5.204957741744088e-05,
      "loss": 0.7735,
      "step": 49603
    },
    {
      "epoch": 4.267744988385099,
      "grad_norm": 1.1875,
      "learning_rate": 5.203756511581048e-05,
      "loss": 0.9928,
      "step": 49604
    },
    {
      "epoch": 4.26783102469242,
      "grad_norm": 1.1953125,
      "learning_rate": 5.202555412438309e-05,
      "loss": 0.9215,
      "step": 49605
    },
    {
      "epoch": 4.267917060999742,
      "grad_norm": 1.3203125,
      "learning_rate": 5.201354444319378e-05,
      "loss": 1.7422,
      "step": 49606
    },
    {
      "epoch": 4.268003097307064,
      "grad_norm": 0.9453125,
      "learning_rate": 5.200153607227787e-05,
      "loss": 0.4959,
      "step": 49607
    },
    {
      "epoch": 4.268089133614385,
      "grad_norm": 1.4609375,
      "learning_rate": 5.198952901167014e-05,
      "loss": 1.2424,
      "step": 49608
    },
    {
      "epoch": 4.268175169921707,
      "grad_norm": 0.9609375,
      "learning_rate": 5.197752326140598e-05,
      "loss": 1.0237,
      "step": 49609
    },
    {
      "epoch": 4.268261206229028,
      "grad_norm": 0.78125,
      "learning_rate": 5.196551882152051e-05,
      "loss": 1.1478,
      "step": 49610
    },
    {
      "epoch": 4.2683472425363504,
      "grad_norm": 1.4765625,
      "learning_rate": 5.195351569204865e-05,
      "loss": 0.5483,
      "step": 49611
    },
    {
      "epoch": 4.268433278843672,
      "grad_norm": 2.515625,
      "learning_rate": 5.194151387302559e-05,
      "loss": 0.8243,
      "step": 49612
    },
    {
      "epoch": 4.268519315150994,
      "grad_norm": 1.328125,
      "learning_rate": 5.192951336448659e-05,
      "loss": 1.2874,
      "step": 49613
    },
    {
      "epoch": 4.268605351458316,
      "grad_norm": 0.9296875,
      "learning_rate": 5.191751416646651e-05,
      "loss": 1.3932,
      "step": 49614
    },
    {
      "epoch": 4.268691387765637,
      "grad_norm": 0.77734375,
      "learning_rate": 5.190551627900059e-05,
      "loss": 0.8242,
      "step": 49615
    },
    {
      "epoch": 4.268777424072959,
      "grad_norm": 0.9765625,
      "learning_rate": 5.189351970212391e-05,
      "loss": 0.8368,
      "step": 49616
    },
    {
      "epoch": 4.26886346038028,
      "grad_norm": 1.0078125,
      "learning_rate": 5.188152443587152e-05,
      "loss": 0.9178,
      "step": 49617
    },
    {
      "epoch": 4.268949496687602,
      "grad_norm": 1.0390625,
      "learning_rate": 5.1869530480278426e-05,
      "loss": 0.6931,
      "step": 49618
    },
    {
      "epoch": 4.269035532994923,
      "grad_norm": 1.234375,
      "learning_rate": 5.185753783537994e-05,
      "loss": 1.0569,
      "step": 49619
    },
    {
      "epoch": 4.2691215693022455,
      "grad_norm": 0.85546875,
      "learning_rate": 5.1845546501210926e-05,
      "loss": 0.7302,
      "step": 49620
    },
    {
      "epoch": 4.269207605609568,
      "grad_norm": 0.8046875,
      "learning_rate": 5.1833556477806635e-05,
      "loss": 0.5856,
      "step": 49621
    },
    {
      "epoch": 4.269293641916889,
      "grad_norm": 11.4375,
      "learning_rate": 5.182156776520186e-05,
      "loss": 1.3601,
      "step": 49622
    },
    {
      "epoch": 4.269379678224211,
      "grad_norm": 1.109375,
      "learning_rate": 5.1809580363431864e-05,
      "loss": 1.4712,
      "step": 49623
    },
    {
      "epoch": 4.269465714531532,
      "grad_norm": 2.703125,
      "learning_rate": 5.179759427253178e-05,
      "loss": 0.7042,
      "step": 49624
    },
    {
      "epoch": 4.269551750838854,
      "grad_norm": 1.15625,
      "learning_rate": 5.1785609492536464e-05,
      "loss": 1.7346,
      "step": 49625
    },
    {
      "epoch": 4.269637787146175,
      "grad_norm": 1.2578125,
      "learning_rate": 5.177362602348107e-05,
      "loss": 0.8015,
      "step": 49626
    },
    {
      "epoch": 4.269723823453497,
      "grad_norm": 1.25,
      "learning_rate": 5.1761643865400776e-05,
      "loss": 1.5183,
      "step": 49627
    },
    {
      "epoch": 4.269809859760819,
      "grad_norm": 1.4375,
      "learning_rate": 5.174966301833034e-05,
      "loss": 1.0881,
      "step": 49628
    },
    {
      "epoch": 4.269895896068141,
      "grad_norm": 1.3046875,
      "learning_rate": 5.173768348230501e-05,
      "loss": 1.9082,
      "step": 49629
    },
    {
      "epoch": 4.269981932375463,
      "grad_norm": 24.875,
      "learning_rate": 5.172570525735981e-05,
      "loss": 1.4509,
      "step": 49630
    },
    {
      "epoch": 4.270067968682784,
      "grad_norm": 3.546875,
      "learning_rate": 5.1713728343529664e-05,
      "loss": 0.8609,
      "step": 49631
    },
    {
      "epoch": 4.270154004990106,
      "grad_norm": 0.99609375,
      "learning_rate": 5.1701752740849814e-05,
      "loss": 0.6605,
      "step": 49632
    },
    {
      "epoch": 4.270240041297427,
      "grad_norm": 1.03125,
      "learning_rate": 5.168977844935502e-05,
      "loss": 1.1526,
      "step": 49633
    },
    {
      "epoch": 4.270326077604749,
      "grad_norm": 1.1640625,
      "learning_rate": 5.167780546908041e-05,
      "loss": 1.0682,
      "step": 49634
    },
    {
      "epoch": 4.270412113912071,
      "grad_norm": 1.421875,
      "learning_rate": 5.166583380006107e-05,
      "loss": 1.2585,
      "step": 49635
    },
    {
      "epoch": 4.270498150219392,
      "grad_norm": 1.2265625,
      "learning_rate": 5.165386344233197e-05,
      "loss": 1.7541,
      "step": 49636
    },
    {
      "epoch": 4.2705841865267145,
      "grad_norm": 0.74609375,
      "learning_rate": 5.16418943959282e-05,
      "loss": 0.5342,
      "step": 49637
    },
    {
      "epoch": 4.270670222834036,
      "grad_norm": 1.4609375,
      "learning_rate": 5.1629926660884565e-05,
      "loss": 1.3329,
      "step": 49638
    },
    {
      "epoch": 4.270756259141358,
      "grad_norm": 4.25,
      "learning_rate": 5.161796023723619e-05,
      "loss": 1.3588,
      "step": 49639
    },
    {
      "epoch": 4.270842295448679,
      "grad_norm": 1.0625,
      "learning_rate": 5.1605995125018234e-05,
      "loss": 0.8453,
      "step": 49640
    },
    {
      "epoch": 4.270928331756001,
      "grad_norm": 0.765625,
      "learning_rate": 5.159403132426538e-05,
      "loss": 0.6332,
      "step": 49641
    },
    {
      "epoch": 4.271014368063323,
      "grad_norm": 1.25,
      "learning_rate": 5.158206883501282e-05,
      "loss": 0.947,
      "step": 49642
    },
    {
      "epoch": 4.271100404370644,
      "grad_norm": 0.96484375,
      "learning_rate": 5.157010765729553e-05,
      "loss": 1.1548,
      "step": 49643
    },
    {
      "epoch": 4.271186440677966,
      "grad_norm": 1.15625,
      "learning_rate": 5.155814779114842e-05,
      "loss": 0.9465,
      "step": 49644
    },
    {
      "epoch": 4.2712724769852874,
      "grad_norm": 0.8828125,
      "learning_rate": 5.154618923660648e-05,
      "loss": 1.0708,
      "step": 49645
    },
    {
      "epoch": 4.2713585132926095,
      "grad_norm": 1.234375,
      "learning_rate": 5.153423199370483e-05,
      "loss": 1.702,
      "step": 49646
    },
    {
      "epoch": 4.271444549599932,
      "grad_norm": 1.3515625,
      "learning_rate": 5.152227606247822e-05,
      "loss": 1.0101,
      "step": 49647
    },
    {
      "epoch": 4.271530585907253,
      "grad_norm": 1.015625,
      "learning_rate": 5.151032144296175e-05,
      "loss": 0.679,
      "step": 49648
    },
    {
      "epoch": 4.271616622214575,
      "grad_norm": 5.25,
      "learning_rate": 5.149836813519043e-05,
      "loss": 1.326,
      "step": 49649
    },
    {
      "epoch": 4.271702658521896,
      "grad_norm": 0.8203125,
      "learning_rate": 5.1486416139199076e-05,
      "loss": 0.7662,
      "step": 49650
    },
    {
      "epoch": 4.271788694829218,
      "grad_norm": 1.28125,
      "learning_rate": 5.147446545502282e-05,
      "loss": 0.8138,
      "step": 49651
    },
    {
      "epoch": 4.271874731136539,
      "grad_norm": 1.3125,
      "learning_rate": 5.146251608269642e-05,
      "loss": 0.7002,
      "step": 49652
    },
    {
      "epoch": 4.271960767443861,
      "grad_norm": 0.95703125,
      "learning_rate": 5.145056802225495e-05,
      "loss": 0.6985,
      "step": 49653
    },
    {
      "epoch": 4.272046803751183,
      "grad_norm": 1.046875,
      "learning_rate": 5.143862127373339e-05,
      "loss": 0.981,
      "step": 49654
    },
    {
      "epoch": 4.272132840058505,
      "grad_norm": 1.9921875,
      "learning_rate": 5.1426675837166536e-05,
      "loss": 1.0119,
      "step": 49655
    },
    {
      "epoch": 4.272218876365827,
      "grad_norm": 1.25,
      "learning_rate": 5.1414731712589424e-05,
      "loss": 0.6858,
      "step": 49656
    },
    {
      "epoch": 4.272304912673148,
      "grad_norm": 1.0703125,
      "learning_rate": 5.140278890003708e-05,
      "loss": 1.3942,
      "step": 49657
    },
    {
      "epoch": 4.27239094898047,
      "grad_norm": 0.890625,
      "learning_rate": 5.139084739954419e-05,
      "loss": 0.733,
      "step": 49658
    },
    {
      "epoch": 4.272476985287791,
      "grad_norm": 1.1640625,
      "learning_rate": 5.1378907211145855e-05,
      "loss": 1.3404,
      "step": 49659
    },
    {
      "epoch": 4.272563021595113,
      "grad_norm": 0.9609375,
      "learning_rate": 5.136696833487708e-05,
      "loss": 0.8899,
      "step": 49660
    },
    {
      "epoch": 4.272649057902435,
      "grad_norm": 1.0390625,
      "learning_rate": 5.1355030770772525e-05,
      "loss": 1.3123,
      "step": 49661
    },
    {
      "epoch": 4.272735094209756,
      "grad_norm": 1.046875,
      "learning_rate": 5.1343094518867376e-05,
      "loss": 0.8616,
      "step": 49662
    },
    {
      "epoch": 4.2728211305170785,
      "grad_norm": 0.88671875,
      "learning_rate": 5.133115957919632e-05,
      "loss": 1.0794,
      "step": 49663
    },
    {
      "epoch": 4.2729071668244,
      "grad_norm": 1.3671875,
      "learning_rate": 5.131922595179439e-05,
      "loss": 1.5036,
      "step": 49664
    },
    {
      "epoch": 4.272993203131722,
      "grad_norm": 1.046875,
      "learning_rate": 5.13072936366965e-05,
      "loss": 1.4255,
      "step": 49665
    },
    {
      "epoch": 4.273079239439043,
      "grad_norm": 1.0078125,
      "learning_rate": 5.1295362633937414e-05,
      "loss": 0.6008,
      "step": 49666
    },
    {
      "epoch": 4.273165275746365,
      "grad_norm": 1.1875,
      "learning_rate": 5.128343294355215e-05,
      "loss": 1.381,
      "step": 49667
    },
    {
      "epoch": 4.273251312053687,
      "grad_norm": 0.6484375,
      "learning_rate": 5.1271504565575665e-05,
      "loss": 0.2855,
      "step": 49668
    },
    {
      "epoch": 4.273337348361008,
      "grad_norm": 1.171875,
      "learning_rate": 5.125957750004268e-05,
      "loss": 1.4066,
      "step": 49669
    },
    {
      "epoch": 4.27342338466833,
      "grad_norm": 1.1171875,
      "learning_rate": 5.1247651746988144e-05,
      "loss": 1.1986,
      "step": 49670
    },
    {
      "epoch": 4.2735094209756515,
      "grad_norm": 0.8671875,
      "learning_rate": 5.123572730644704e-05,
      "loss": 0.8695,
      "step": 49671
    },
    {
      "epoch": 4.2735954572829735,
      "grad_norm": 2.453125,
      "learning_rate": 5.122380417845396e-05,
      "loss": 1.0739,
      "step": 49672
    },
    {
      "epoch": 4.273681493590295,
      "grad_norm": 29568.0,
      "learning_rate": 5.1211882363044136e-05,
      "loss": 0.8934,
      "step": 49673
    },
    {
      "epoch": 4.273767529897617,
      "grad_norm": 0.88671875,
      "learning_rate": 5.119996186025222e-05,
      "loss": 0.7143,
      "step": 49674
    },
    {
      "epoch": 4.273853566204939,
      "grad_norm": 0.98828125,
      "learning_rate": 5.118804267011312e-05,
      "loss": 1.0533,
      "step": 49675
    },
    {
      "epoch": 4.27393960251226,
      "grad_norm": 14.625,
      "learning_rate": 5.117612479266176e-05,
      "loss": 0.9332,
      "step": 49676
    },
    {
      "epoch": 4.274025638819582,
      "grad_norm": 1.09375,
      "learning_rate": 5.116420822793288e-05,
      "loss": 1.3481,
      "step": 49677
    },
    {
      "epoch": 4.274111675126903,
      "grad_norm": 27.0,
      "learning_rate": 5.1152292975961526e-05,
      "loss": 1.0853,
      "step": 49678
    },
    {
      "epoch": 4.274197711434225,
      "grad_norm": 0.8828125,
      "learning_rate": 5.114037903678226e-05,
      "loss": 0.921,
      "step": 49679
    },
    {
      "epoch": 4.274283747741547,
      "grad_norm": 0.97265625,
      "learning_rate": 5.112846641043012e-05,
      "loss": 0.5833,
      "step": 49680
    },
    {
      "epoch": 4.274369784048869,
      "grad_norm": 1.6953125,
      "learning_rate": 5.111655509694002e-05,
      "loss": 0.8557,
      "step": 49681
    },
    {
      "epoch": 4.274455820356191,
      "grad_norm": 1.1015625,
      "learning_rate": 5.1104645096346605e-05,
      "loss": 1.3617,
      "step": 49682
    },
    {
      "epoch": 4.274541856663512,
      "grad_norm": 192.0,
      "learning_rate": 5.109273640868478e-05,
      "loss": 1.2222,
      "step": 49683
    },
    {
      "epoch": 4.274627892970834,
      "grad_norm": 1.015625,
      "learning_rate": 5.1080829033989474e-05,
      "loss": 0.9414,
      "step": 49684
    },
    {
      "epoch": 4.274713929278155,
      "grad_norm": 0.96484375,
      "learning_rate": 5.106892297229537e-05,
      "loss": 0.4745,
      "step": 49685
    },
    {
      "epoch": 4.274799965585477,
      "grad_norm": 0.98046875,
      "learning_rate": 5.105701822363728e-05,
      "loss": 1.4476,
      "step": 49686
    },
    {
      "epoch": 4.274886001892799,
      "grad_norm": 27.0,
      "learning_rate": 5.10451147880503e-05,
      "loss": 1.406,
      "step": 49687
    },
    {
      "epoch": 4.27497203820012,
      "grad_norm": 1.234375,
      "learning_rate": 5.1033212665568875e-05,
      "loss": 1.0874,
      "step": 49688
    },
    {
      "epoch": 4.2750580745074425,
      "grad_norm": 3.328125,
      "learning_rate": 5.1021311856227994e-05,
      "loss": 1.0797,
      "step": 49689
    },
    {
      "epoch": 4.275144110814764,
      "grad_norm": 1.25,
      "learning_rate": 5.1009412360062565e-05,
      "loss": 1.0324,
      "step": 49690
    },
    {
      "epoch": 4.275230147122086,
      "grad_norm": 1.2109375,
      "learning_rate": 5.099751417710718e-05,
      "loss": 0.9886,
      "step": 49691
    },
    {
      "epoch": 4.275316183429407,
      "grad_norm": 4608.0,
      "learning_rate": 5.098561730739682e-05,
      "loss": 0.706,
      "step": 49692
    },
    {
      "epoch": 4.275402219736729,
      "grad_norm": 1.2734375,
      "learning_rate": 5.097372175096615e-05,
      "loss": 1.5823,
      "step": 49693
    },
    {
      "epoch": 4.275488256044051,
      "grad_norm": 1.75,
      "learning_rate": 5.096182750785e-05,
      "loss": 1.2325,
      "step": 49694
    },
    {
      "epoch": 4.275574292351372,
      "grad_norm": 88.0,
      "learning_rate": 5.094993457808323e-05,
      "loss": 1.5313,
      "step": 49695
    },
    {
      "epoch": 4.275660328658694,
      "grad_norm": 0.8828125,
      "learning_rate": 5.093804296170046e-05,
      "loss": 1.0016,
      "step": 49696
    },
    {
      "epoch": 4.2757463649660155,
      "grad_norm": 1.0859375,
      "learning_rate": 5.0926152658736624e-05,
      "loss": 0.5555,
      "step": 49697
    },
    {
      "epoch": 4.2758324012733375,
      "grad_norm": 1.078125,
      "learning_rate": 5.091426366922652e-05,
      "loss": 0.9838,
      "step": 49698
    },
    {
      "epoch": 4.275918437580659,
      "grad_norm": 0.98046875,
      "learning_rate": 5.09023759932048e-05,
      "loss": 0.7004,
      "step": 49699
    },
    {
      "epoch": 4.276004473887981,
      "grad_norm": 40.5,
      "learning_rate": 5.089048963070625e-05,
      "loss": 0.8193,
      "step": 49700
    },
    {
      "epoch": 4.276090510195303,
      "grad_norm": 2.203125,
      "learning_rate": 5.0878604581765744e-05,
      "loss": 0.9458,
      "step": 49701
    },
    {
      "epoch": 4.276176546502624,
      "grad_norm": 1.375,
      "learning_rate": 5.086672084641791e-05,
      "loss": 0.9558,
      "step": 49702
    },
    {
      "epoch": 4.276262582809946,
      "grad_norm": 1.125,
      "learning_rate": 5.085483842469768e-05,
      "loss": 1.296,
      "step": 49703
    },
    {
      "epoch": 4.276348619117267,
      "grad_norm": 7.875,
      "learning_rate": 5.0842957316639574e-05,
      "loss": 0.9701,
      "step": 49704
    },
    {
      "epoch": 4.276434655424589,
      "grad_norm": 1.1640625,
      "learning_rate": 5.083107752227845e-05,
      "loss": 1.5758,
      "step": 49705
    },
    {
      "epoch": 4.2765206917319105,
      "grad_norm": 1.859375,
      "learning_rate": 5.0819199041649166e-05,
      "loss": 1.1717,
      "step": 49706
    },
    {
      "epoch": 4.276606728039233,
      "grad_norm": 1.2109375,
      "learning_rate": 5.080732187478626e-05,
      "loss": 1.8135,
      "step": 49707
    },
    {
      "epoch": 4.276692764346555,
      "grad_norm": 1.15625,
      "learning_rate": 5.079544602172459e-05,
      "loss": 1.0327,
      "step": 49708
    },
    {
      "epoch": 4.276778800653876,
      "grad_norm": 1.2734375,
      "learning_rate": 5.078357148249885e-05,
      "loss": 1.7784,
      "step": 49709
    },
    {
      "epoch": 4.276864836961198,
      "grad_norm": 2.25,
      "learning_rate": 5.0771698257143836e-05,
      "loss": 1.0469,
      "step": 49710
    },
    {
      "epoch": 4.276950873268519,
      "grad_norm": 1.1328125,
      "learning_rate": 5.075982634569432e-05,
      "loss": 0.9025,
      "step": 49711
    },
    {
      "epoch": 4.277036909575841,
      "grad_norm": 1.2265625,
      "learning_rate": 5.0747955748184805e-05,
      "loss": 1.6978,
      "step": 49712
    },
    {
      "epoch": 4.277122945883162,
      "grad_norm": 3.8125,
      "learning_rate": 5.073608646465017e-05,
      "loss": 1.3832,
      "step": 49713
    },
    {
      "epoch": 4.277208982190484,
      "grad_norm": 1.40625,
      "learning_rate": 5.072421849512521e-05,
      "loss": 1.2716,
      "step": 49714
    },
    {
      "epoch": 4.2772950184978065,
      "grad_norm": 2.734375,
      "learning_rate": 5.0712351839644453e-05,
      "loss": 0.9852,
      "step": 49715
    },
    {
      "epoch": 4.277381054805128,
      "grad_norm": 2.765625,
      "learning_rate": 5.070048649824266e-05,
      "loss": 1.4821,
      "step": 49716
    },
    {
      "epoch": 4.27746709111245,
      "grad_norm": 0.87109375,
      "learning_rate": 5.068862247095462e-05,
      "loss": 0.5365,
      "step": 49717
    },
    {
      "epoch": 4.277553127419771,
      "grad_norm": 1.1875,
      "learning_rate": 5.067675975781494e-05,
      "loss": 1.0026,
      "step": 49718
    },
    {
      "epoch": 4.277639163727093,
      "grad_norm": 1.1953125,
      "learning_rate": 5.06648983588584e-05,
      "loss": 0.8215,
      "step": 49719
    },
    {
      "epoch": 4.277725200034414,
      "grad_norm": 0.828125,
      "learning_rate": 5.06530382741196e-05,
      "loss": 0.5267,
      "step": 49720
    },
    {
      "epoch": 4.277811236341736,
      "grad_norm": 1.1640625,
      "learning_rate": 5.064117950363323e-05,
      "loss": 1.7297,
      "step": 49721
    },
    {
      "epoch": 4.277897272649058,
      "grad_norm": 11.875,
      "learning_rate": 5.062932204743414e-05,
      "loss": 0.9223,
      "step": 49722
    },
    {
      "epoch": 4.2779833089563795,
      "grad_norm": 0.8671875,
      "learning_rate": 5.061746590555677e-05,
      "loss": 0.69,
      "step": 49723
    },
    {
      "epoch": 4.2780693452637015,
      "grad_norm": 0.9921875,
      "learning_rate": 5.0605611078035904e-05,
      "loss": 0.8476,
      "step": 49724
    },
    {
      "epoch": 4.278155381571023,
      "grad_norm": 1.140625,
      "learning_rate": 5.059375756490631e-05,
      "loss": 1.4269,
      "step": 49725
    },
    {
      "epoch": 4.278241417878345,
      "grad_norm": 1.5546875,
      "learning_rate": 5.0581905366202495e-05,
      "loss": 0.7312,
      "step": 49726
    },
    {
      "epoch": 4.278327454185666,
      "grad_norm": 0.8828125,
      "learning_rate": 5.057005448195917e-05,
      "loss": 0.8318,
      "step": 49727
    },
    {
      "epoch": 4.278413490492988,
      "grad_norm": 1.25,
      "learning_rate": 5.0558204912211134e-05,
      "loss": 1.4821,
      "step": 49728
    },
    {
      "epoch": 4.27849952680031,
      "grad_norm": 0.8984375,
      "learning_rate": 5.0546356656992866e-05,
      "loss": 0.5379,
      "step": 49729
    },
    {
      "epoch": 4.278585563107631,
      "grad_norm": 446.0,
      "learning_rate": 5.0534509716339104e-05,
      "loss": 0.9757,
      "step": 49730
    },
    {
      "epoch": 4.278671599414953,
      "grad_norm": 1.3125,
      "learning_rate": 5.0522664090284556e-05,
      "loss": 1.3107,
      "step": 49731
    },
    {
      "epoch": 4.2787576357222745,
      "grad_norm": 0.9453125,
      "learning_rate": 5.0510819778863684e-05,
      "loss": 0.7425,
      "step": 49732
    },
    {
      "epoch": 4.278843672029597,
      "grad_norm": 1.1484375,
      "learning_rate": 5.049897678211135e-05,
      "loss": 1.2733,
      "step": 49733
    },
    {
      "epoch": 4.278929708336918,
      "grad_norm": 1.0546875,
      "learning_rate": 5.0487135100061976e-05,
      "loss": 1.3149,
      "step": 49734
    },
    {
      "epoch": 4.27901574464424,
      "grad_norm": 1.0390625,
      "learning_rate": 5.0475294732750365e-05,
      "loss": 1.5637,
      "step": 49735
    },
    {
      "epoch": 4.279101780951562,
      "grad_norm": 1.59375,
      "learning_rate": 5.04634556802111e-05,
      "loss": 1.2292,
      "step": 49736
    },
    {
      "epoch": 4.279187817258883,
      "grad_norm": 1.1171875,
      "learning_rate": 5.0451617942478764e-05,
      "loss": 1.3251,
      "step": 49737
    },
    {
      "epoch": 4.279273853566205,
      "grad_norm": 15.5625,
      "learning_rate": 5.0439781519588e-05,
      "loss": 1.893,
      "step": 49738
    },
    {
      "epoch": 4.279359889873526,
      "grad_norm": 0.8984375,
      "learning_rate": 5.042794641157356e-05,
      "loss": 0.9649,
      "step": 49739
    },
    {
      "epoch": 4.279445926180848,
      "grad_norm": 1.2265625,
      "learning_rate": 5.0416112618469786e-05,
      "loss": 1.135,
      "step": 49740
    },
    {
      "epoch": 4.27953196248817,
      "grad_norm": 12.125,
      "learning_rate": 5.0404280140311455e-05,
      "loss": 1.3379,
      "step": 49741
    },
    {
      "epoch": 4.279617998795492,
      "grad_norm": 0.90234375,
      "learning_rate": 5.03924489771333e-05,
      "loss": 0.7806,
      "step": 49742
    },
    {
      "epoch": 4.279704035102814,
      "grad_norm": 1.28125,
      "learning_rate": 5.0380619128969686e-05,
      "loss": 1.0731,
      "step": 49743
    },
    {
      "epoch": 4.279790071410135,
      "grad_norm": 1.0390625,
      "learning_rate": 5.036879059585536e-05,
      "loss": 1.0116,
      "step": 49744
    },
    {
      "epoch": 4.279876107717457,
      "grad_norm": 1.09375,
      "learning_rate": 5.0356963377824804e-05,
      "loss": 0.9864,
      "step": 49745
    },
    {
      "epoch": 4.279962144024778,
      "grad_norm": 1.421875,
      "learning_rate": 5.03451374749126e-05,
      "loss": 1.7323,
      "step": 49746
    },
    {
      "epoch": 4.2800481803321,
      "grad_norm": 5.46875,
      "learning_rate": 5.033331288715359e-05,
      "loss": 0.9449,
      "step": 49747
    },
    {
      "epoch": 4.280134216639421,
      "grad_norm": 1.375,
      "learning_rate": 5.032148961458211e-05,
      "loss": 0.7323,
      "step": 49748
    },
    {
      "epoch": 4.2802202529467435,
      "grad_norm": 0.96875,
      "learning_rate": 5.030966765723288e-05,
      "loss": 0.7132,
      "step": 49749
    },
    {
      "epoch": 4.2803062892540655,
      "grad_norm": 1.5390625,
      "learning_rate": 5.029784701514034e-05,
      "loss": 0.6487,
      "step": 49750
    },
    {
      "epoch": 4.280392325561387,
      "grad_norm": 1.21875,
      "learning_rate": 5.028602768833912e-05,
      "loss": 1.1316,
      "step": 49751
    },
    {
      "epoch": 4.280478361868709,
      "grad_norm": 1.171875,
      "learning_rate": 5.027420967686391e-05,
      "loss": 0.5527,
      "step": 49752
    },
    {
      "epoch": 4.28056439817603,
      "grad_norm": 1.03125,
      "learning_rate": 5.0262392980749085e-05,
      "loss": 1.8233,
      "step": 49753
    },
    {
      "epoch": 4.280650434483352,
      "grad_norm": 2.25,
      "learning_rate": 5.025057760002927e-05,
      "loss": 0.8753,
      "step": 49754
    },
    {
      "epoch": 4.280736470790673,
      "grad_norm": 1.078125,
      "learning_rate": 5.023876353473916e-05,
      "loss": 1.0552,
      "step": 49755
    },
    {
      "epoch": 4.280822507097995,
      "grad_norm": 0.984375,
      "learning_rate": 5.0226950784913125e-05,
      "loss": 1.1787,
      "step": 49756
    },
    {
      "epoch": 4.280908543405317,
      "grad_norm": 0.875,
      "learning_rate": 5.021513935058575e-05,
      "loss": 0.6541,
      "step": 49757
    },
    {
      "epoch": 4.2809945797126385,
      "grad_norm": 1.3984375,
      "learning_rate": 5.0203329231791664e-05,
      "loss": 0.951,
      "step": 49758
    },
    {
      "epoch": 4.281080616019961,
      "grad_norm": 4.46875,
      "learning_rate": 5.019152042856534e-05,
      "loss": 0.9448,
      "step": 49759
    },
    {
      "epoch": 4.281166652327282,
      "grad_norm": 1.03125,
      "learning_rate": 5.017971294094137e-05,
      "loss": 1.3072,
      "step": 49760
    },
    {
      "epoch": 4.281252688634604,
      "grad_norm": 1.109375,
      "learning_rate": 5.016790676895416e-05,
      "loss": 0.7597,
      "step": 49761
    },
    {
      "epoch": 4.281338724941925,
      "grad_norm": 1.328125,
      "learning_rate": 5.015610191263836e-05,
      "loss": 1.6438,
      "step": 49762
    },
    {
      "epoch": 4.281424761249247,
      "grad_norm": 0.859375,
      "learning_rate": 5.014429837202861e-05,
      "loss": 0.9725,
      "step": 49763
    },
    {
      "epoch": 4.281510797556569,
      "grad_norm": 1.2578125,
      "learning_rate": 5.0132496147159144e-05,
      "loss": 0.6554,
      "step": 49764
    },
    {
      "epoch": 4.28159683386389,
      "grad_norm": 1.1875,
      "learning_rate": 5.012069523806467e-05,
      "loss": 0.7221,
      "step": 49765
    },
    {
      "epoch": 4.281682870171212,
      "grad_norm": 0.98046875,
      "learning_rate": 5.010889564477972e-05,
      "loss": 1.2627,
      "step": 49766
    },
    {
      "epoch": 4.281768906478534,
      "grad_norm": 29.0,
      "learning_rate": 5.00970973673387e-05,
      "loss": 1.5983,
      "step": 49767
    },
    {
      "epoch": 4.281854942785856,
      "grad_norm": 0.96484375,
      "learning_rate": 5.0085300405776134e-05,
      "loss": 0.6911,
      "step": 49768
    },
    {
      "epoch": 4.281940979093178,
      "grad_norm": 1.1171875,
      "learning_rate": 5.007350476012662e-05,
      "loss": 0.7248,
      "step": 49769
    },
    {
      "epoch": 4.282027015400499,
      "grad_norm": 1.1171875,
      "learning_rate": 5.0061710430424565e-05,
      "loss": 0.5655,
      "step": 49770
    },
    {
      "epoch": 4.282113051707821,
      "grad_norm": 1.15625,
      "learning_rate": 5.00499174167045e-05,
      "loss": 0.9606,
      "step": 49771
    },
    {
      "epoch": 4.282199088015142,
      "grad_norm": 3.1875,
      "learning_rate": 5.003812571900096e-05,
      "loss": 0.7683,
      "step": 49772
    },
    {
      "epoch": 4.282285124322464,
      "grad_norm": 0.6953125,
      "learning_rate": 5.002633533734835e-05,
      "loss": 0.5636,
      "step": 49773
    },
    {
      "epoch": 4.282371160629785,
      "grad_norm": 1.0625,
      "learning_rate": 5.001454627178126e-05,
      "loss": 1.4838,
      "step": 49774
    },
    {
      "epoch": 4.2824571969371075,
      "grad_norm": 1.5234375,
      "learning_rate": 5.0002758522333994e-05,
      "loss": 0.7613,
      "step": 49775
    },
    {
      "epoch": 4.2825432332444295,
      "grad_norm": 1.0625,
      "learning_rate": 4.999097208904119e-05,
      "loss": 0.7201,
      "step": 49776
    },
    {
      "epoch": 4.282629269551751,
      "grad_norm": 0.96484375,
      "learning_rate": 4.997918697193732e-05,
      "loss": 0.781,
      "step": 49777
    },
    {
      "epoch": 4.282715305859073,
      "grad_norm": 1.3984375,
      "learning_rate": 4.9967403171056745e-05,
      "loss": 0.9667,
      "step": 49778
    },
    {
      "epoch": 4.282801342166394,
      "grad_norm": 1.140625,
      "learning_rate": 4.995562068643394e-05,
      "loss": 1.5355,
      "step": 49779
    },
    {
      "epoch": 4.282887378473716,
      "grad_norm": 11.5625,
      "learning_rate": 4.994383951810355e-05,
      "loss": 0.7841,
      "step": 49780
    },
    {
      "epoch": 4.282973414781037,
      "grad_norm": 0.87109375,
      "learning_rate": 4.9932059666099815e-05,
      "loss": 0.8515,
      "step": 49781
    },
    {
      "epoch": 4.283059451088359,
      "grad_norm": 1.5,
      "learning_rate": 4.992028113045727e-05,
      "loss": 1.8091,
      "step": 49782
    },
    {
      "epoch": 4.283145487395681,
      "grad_norm": 1.65625,
      "learning_rate": 4.990850391121032e-05,
      "loss": 1.4481,
      "step": 49783
    },
    {
      "epoch": 4.2832315237030025,
      "grad_norm": 1.078125,
      "learning_rate": 4.989672800839351e-05,
      "loss": 1.3203,
      "step": 49784
    },
    {
      "epoch": 4.283317560010325,
      "grad_norm": 1.078125,
      "learning_rate": 4.98849534220413e-05,
      "loss": 0.9633,
      "step": 49785
    },
    {
      "epoch": 4.283403596317646,
      "grad_norm": 1.5234375,
      "learning_rate": 4.987318015218795e-05,
      "loss": 1.0648,
      "step": 49786
    },
    {
      "epoch": 4.283489632624968,
      "grad_norm": 1.3984375,
      "learning_rate": 4.986140819886803e-05,
      "loss": 0.9841,
      "step": 49787
    },
    {
      "epoch": 4.283575668932289,
      "grad_norm": 1.109375,
      "learning_rate": 4.984963756211603e-05,
      "loss": 1.6069,
      "step": 49788
    },
    {
      "epoch": 4.283661705239611,
      "grad_norm": 1.171875,
      "learning_rate": 4.9837868241966175e-05,
      "loss": 1.5057,
      "step": 49789
    },
    {
      "epoch": 4.283747741546933,
      "grad_norm": 1.453125,
      "learning_rate": 4.982610023845313e-05,
      "loss": 1.866,
      "step": 49790
    },
    {
      "epoch": 4.283833777854254,
      "grad_norm": 3.3125,
      "learning_rate": 4.981433355161108e-05,
      "loss": 0.3176,
      "step": 49791
    },
    {
      "epoch": 4.283919814161576,
      "grad_norm": 0.94921875,
      "learning_rate": 4.980256818147455e-05,
      "loss": 0.5599,
      "step": 49792
    },
    {
      "epoch": 4.284005850468898,
      "grad_norm": 1.3359375,
      "learning_rate": 4.979080412807802e-05,
      "loss": 1.4676,
      "step": 49793
    },
    {
      "epoch": 4.28409188677622,
      "grad_norm": 5.40625,
      "learning_rate": 4.977904139145578e-05,
      "loss": 1.0423,
      "step": 49794
    },
    {
      "epoch": 4.284177923083541,
      "grad_norm": 0.96875,
      "learning_rate": 4.9767279971642264e-05,
      "loss": 0.6909,
      "step": 49795
    },
    {
      "epoch": 4.284263959390863,
      "grad_norm": 1.328125,
      "learning_rate": 4.975551986867194e-05,
      "loss": 1.2597,
      "step": 49796
    },
    {
      "epoch": 4.284349995698185,
      "grad_norm": 0.828125,
      "learning_rate": 4.974376108257911e-05,
      "loss": 0.6914,
      "step": 49797
    },
    {
      "epoch": 4.284436032005506,
      "grad_norm": 0.96484375,
      "learning_rate": 4.973200361339819e-05,
      "loss": 1.0797,
      "step": 49798
    },
    {
      "epoch": 4.284522068312828,
      "grad_norm": 1.0234375,
      "learning_rate": 4.972024746116366e-05,
      "loss": 0.9731,
      "step": 49799
    },
    {
      "epoch": 4.284608104620149,
      "grad_norm": 1.4765625,
      "learning_rate": 4.970849262590976e-05,
      "loss": 1.4007,
      "step": 49800
    },
    {
      "epoch": 4.2846941409274715,
      "grad_norm": 1.5390625,
      "learning_rate": 4.9696739107670974e-05,
      "loss": 0.9718,
      "step": 49801
    },
    {
      "epoch": 4.2847801772347935,
      "grad_norm": 0.9375,
      "learning_rate": 4.968498690648171e-05,
      "loss": 1.1954,
      "step": 49802
    },
    {
      "epoch": 4.284866213542115,
      "grad_norm": 1.1875,
      "learning_rate": 4.967323602237617e-05,
      "loss": 1.5281,
      "step": 49803
    },
    {
      "epoch": 4.284952249849437,
      "grad_norm": 1.1328125,
      "learning_rate": 4.966148645538893e-05,
      "loss": 1.1508,
      "step": 49804
    },
    {
      "epoch": 4.285038286156758,
      "grad_norm": 3.34375,
      "learning_rate": 4.9649738205554185e-05,
      "loss": 1.3861,
      "step": 49805
    },
    {
      "epoch": 4.28512432246408,
      "grad_norm": 1.09375,
      "learning_rate": 4.9637991272906355e-05,
      "loss": 1.0151,
      "step": 49806
    },
    {
      "epoch": 4.285210358771401,
      "grad_norm": 1.0078125,
      "learning_rate": 4.9626245657479915e-05,
      "loss": 0.8374,
      "step": 49807
    },
    {
      "epoch": 4.285296395078723,
      "grad_norm": 0.875,
      "learning_rate": 4.9614501359309005e-05,
      "loss": 0.6364,
      "step": 49808
    },
    {
      "epoch": 4.285382431386045,
      "grad_norm": 1.8046875,
      "learning_rate": 4.9602758378428094e-05,
      "loss": 1.6341,
      "step": 49809
    },
    {
      "epoch": 4.2854684676933665,
      "grad_norm": 1.46875,
      "learning_rate": 4.959101671487159e-05,
      "loss": 1.7041,
      "step": 49810
    },
    {
      "epoch": 4.285554504000689,
      "grad_norm": 2.15625,
      "learning_rate": 4.9579276368673654e-05,
      "loss": 1.6481,
      "step": 49811
    },
    {
      "epoch": 4.28564054030801,
      "grad_norm": 1.375,
      "learning_rate": 4.9567537339868796e-05,
      "loss": 0.7364,
      "step": 49812
    },
    {
      "epoch": 4.285726576615332,
      "grad_norm": 219.0,
      "learning_rate": 4.9555799628491326e-05,
      "loss": 1.5258,
      "step": 49813
    },
    {
      "epoch": 4.285812612922653,
      "grad_norm": 1.234375,
      "learning_rate": 4.9544063234575445e-05,
      "loss": 1.5664,
      "step": 49814
    },
    {
      "epoch": 4.285898649229975,
      "grad_norm": 1.0703125,
      "learning_rate": 4.9532328158155725e-05,
      "loss": 1.5083,
      "step": 49815
    },
    {
      "epoch": 4.285984685537297,
      "grad_norm": 23.375,
      "learning_rate": 4.952059439926615e-05,
      "loss": 1.1884,
      "step": 49816
    },
    {
      "epoch": 4.286070721844618,
      "grad_norm": 1.09375,
      "learning_rate": 4.9508861957941296e-05,
      "loss": 1.2617,
      "step": 49817
    },
    {
      "epoch": 4.28615675815194,
      "grad_norm": 5.0625,
      "learning_rate": 4.9497130834215475e-05,
      "loss": 0.8376,
      "step": 49818
    },
    {
      "epoch": 4.286242794459262,
      "grad_norm": 1.1796875,
      "learning_rate": 4.9485401028122824e-05,
      "loss": 0.8973,
      "step": 49819
    },
    {
      "epoch": 4.286328830766584,
      "grad_norm": 1.0703125,
      "learning_rate": 4.947367253969776e-05,
      "loss": 0.6883,
      "step": 49820
    },
    {
      "epoch": 4.286414867073905,
      "grad_norm": 2.25,
      "learning_rate": 4.9461945368974584e-05,
      "loss": 0.7788,
      "step": 49821
    },
    {
      "epoch": 4.286500903381227,
      "grad_norm": 1.0859375,
      "learning_rate": 4.9450219515987604e-05,
      "loss": 1.2443,
      "step": 49822
    },
    {
      "epoch": 4.286586939688549,
      "grad_norm": 0.79296875,
      "learning_rate": 4.9438494980771184e-05,
      "loss": 0.4737,
      "step": 49823
    },
    {
      "epoch": 4.28667297599587,
      "grad_norm": 1.34375,
      "learning_rate": 4.9426771763359414e-05,
      "loss": 2.0933,
      "step": 49824
    },
    {
      "epoch": 4.286759012303192,
      "grad_norm": 1.03125,
      "learning_rate": 4.941504986378675e-05,
      "loss": 0.8597,
      "step": 49825
    },
    {
      "epoch": 4.286845048610513,
      "grad_norm": 5.8125,
      "learning_rate": 4.940332928208752e-05,
      "loss": 0.8865,
      "step": 49826
    },
    {
      "epoch": 4.2869310849178355,
      "grad_norm": 1.984375,
      "learning_rate": 4.939161001829579e-05,
      "loss": 1.1112,
      "step": 49827
    },
    {
      "epoch": 4.287017121225157,
      "grad_norm": 1.328125,
      "learning_rate": 4.937989207244598e-05,
      "loss": 1.4467,
      "step": 49828
    },
    {
      "epoch": 4.287103157532479,
      "grad_norm": 1.6328125,
      "learning_rate": 4.936817544457245e-05,
      "loss": 1.1377,
      "step": 49829
    },
    {
      "epoch": 4.287189193839801,
      "grad_norm": 4.75,
      "learning_rate": 4.935646013470929e-05,
      "loss": 0.9428,
      "step": 49830
    },
    {
      "epoch": 4.287275230147122,
      "grad_norm": 1.2890625,
      "learning_rate": 4.9344746142890915e-05,
      "loss": 1.8038,
      "step": 49831
    },
    {
      "epoch": 4.287361266454444,
      "grad_norm": 0.9296875,
      "learning_rate": 4.933303346915141e-05,
      "loss": 0.6894,
      "step": 49832
    },
    {
      "epoch": 4.287447302761765,
      "grad_norm": 1.1328125,
      "learning_rate": 4.932132211352508e-05,
      "loss": 0.6195,
      "step": 49833
    },
    {
      "epoch": 4.287533339069087,
      "grad_norm": 1.1796875,
      "learning_rate": 4.9309612076046394e-05,
      "loss": 1.6134,
      "step": 49834
    },
    {
      "epoch": 4.2876193753764085,
      "grad_norm": 1.1875,
      "learning_rate": 4.929790335674933e-05,
      "loss": 0.9171,
      "step": 49835
    },
    {
      "epoch": 4.2877054116837305,
      "grad_norm": 1.90625,
      "learning_rate": 4.9286195955668245e-05,
      "loss": 1.6014,
      "step": 49836
    },
    {
      "epoch": 4.287791447991053,
      "grad_norm": 1.3359375,
      "learning_rate": 4.927448987283745e-05,
      "loss": 1.477,
      "step": 49837
    },
    {
      "epoch": 4.287877484298374,
      "grad_norm": 3.53125,
      "learning_rate": 4.926278510829102e-05,
      "loss": 1.559,
      "step": 49838
    },
    {
      "epoch": 4.287963520605696,
      "grad_norm": 0.85546875,
      "learning_rate": 4.925108166206332e-05,
      "loss": 0.4205,
      "step": 49839
    },
    {
      "epoch": 4.288049556913017,
      "grad_norm": 1.3125,
      "learning_rate": 4.923937953418861e-05,
      "loss": 1.0358,
      "step": 49840
    },
    {
      "epoch": 4.288135593220339,
      "grad_norm": 1.265625,
      "learning_rate": 4.922767872470091e-05,
      "loss": 1.4051,
      "step": 49841
    },
    {
      "epoch": 4.28822162952766,
      "grad_norm": 1.21875,
      "learning_rate": 4.921597923363463e-05,
      "loss": 1.5709,
      "step": 49842
    },
    {
      "epoch": 4.288307665834982,
      "grad_norm": 21632.0,
      "learning_rate": 4.9204281061024035e-05,
      "loss": 0.8876,
      "step": 49843
    },
    {
      "epoch": 4.288393702142304,
      "grad_norm": 1.015625,
      "learning_rate": 4.919258420690315e-05,
      "loss": 0.8101,
      "step": 49844
    },
    {
      "epoch": 4.288479738449626,
      "grad_norm": 1.2421875,
      "learning_rate": 4.918088867130632e-05,
      "loss": 1.0363,
      "step": 49845
    },
    {
      "epoch": 4.288565774756948,
      "grad_norm": 1.09375,
      "learning_rate": 4.916919445426765e-05,
      "loss": 1.2364,
      "step": 49846
    },
    {
      "epoch": 4.288651811064269,
      "grad_norm": 0.66015625,
      "learning_rate": 4.9157501555821435e-05,
      "loss": 0.5533,
      "step": 49847
    },
    {
      "epoch": 4.288737847371591,
      "grad_norm": 1.0234375,
      "learning_rate": 4.9145809976001876e-05,
      "loss": 1.746,
      "step": 49848
    },
    {
      "epoch": 4.288823883678912,
      "grad_norm": 1.453125,
      "learning_rate": 4.913411971484311e-05,
      "loss": 1.9,
      "step": 49849
    },
    {
      "epoch": 4.288909919986234,
      "grad_norm": 2.171875,
      "learning_rate": 4.9122430772379324e-05,
      "loss": 1.2749,
      "step": 49850
    },
    {
      "epoch": 4.288995956293556,
      "grad_norm": 1.015625,
      "learning_rate": 4.911074314864483e-05,
      "loss": 1.7294,
      "step": 49851
    },
    {
      "epoch": 4.289081992600877,
      "grad_norm": 2.0,
      "learning_rate": 4.9099056843673606e-05,
      "loss": 1.0891,
      "step": 49852
    },
    {
      "epoch": 4.2891680289081995,
      "grad_norm": 1.203125,
      "learning_rate": 4.908737185749995e-05,
      "loss": 0.7771,
      "step": 49853
    },
    {
      "epoch": 4.289254065215521,
      "grad_norm": 1.21875,
      "learning_rate": 4.9075688190158174e-05,
      "loss": 1.2959,
      "step": 49854
    },
    {
      "epoch": 4.289340101522843,
      "grad_norm": 2.484375,
      "learning_rate": 4.906400584168219e-05,
      "loss": 1.6071,
      "step": 49855
    },
    {
      "epoch": 4.289426137830164,
      "grad_norm": 1.125,
      "learning_rate": 4.905232481210636e-05,
      "loss": 0.6165,
      "step": 49856
    },
    {
      "epoch": 4.289512174137486,
      "grad_norm": 1.078125,
      "learning_rate": 4.904064510146461e-05,
      "loss": 0.8759,
      "step": 49857
    },
    {
      "epoch": 4.289598210444808,
      "grad_norm": 1.4375,
      "learning_rate": 4.9028966709791344e-05,
      "loss": 0.9749,
      "step": 49858
    },
    {
      "epoch": 4.289684246752129,
      "grad_norm": 2.71875,
      "learning_rate": 4.901728963712071e-05,
      "loss": 0.8789,
      "step": 49859
    },
    {
      "epoch": 4.289770283059451,
      "grad_norm": 1.2265625,
      "learning_rate": 4.900561388348673e-05,
      "loss": 1.7538,
      "step": 49860
    },
    {
      "epoch": 4.2898563193667725,
      "grad_norm": 1.4375,
      "learning_rate": 4.899393944892372e-05,
      "loss": 1.9987,
      "step": 49861
    },
    {
      "epoch": 4.2899423556740945,
      "grad_norm": 17.875,
      "learning_rate": 4.8982266333465644e-05,
      "loss": 0.9791,
      "step": 49862
    },
    {
      "epoch": 4.290028391981416,
      "grad_norm": 0.8984375,
      "learning_rate": 4.8970594537146694e-05,
      "loss": 1.2097,
      "step": 49863
    },
    {
      "epoch": 4.290114428288738,
      "grad_norm": 1.1171875,
      "learning_rate": 4.895892406000113e-05,
      "loss": 1.6094,
      "step": 49864
    },
    {
      "epoch": 4.29020046459606,
      "grad_norm": 1.1171875,
      "learning_rate": 4.894725490206287e-05,
      "loss": 0.8931,
      "step": 49865
    },
    {
      "epoch": 4.290286500903381,
      "grad_norm": 1.0625,
      "learning_rate": 4.893558706336621e-05,
      "loss": 1.5895,
      "step": 49866
    },
    {
      "epoch": 4.290372537210703,
      "grad_norm": 0.9921875,
      "learning_rate": 4.892392054394529e-05,
      "loss": 1.431,
      "step": 49867
    },
    {
      "epoch": 4.290458573518024,
      "grad_norm": 1.03125,
      "learning_rate": 4.891225534383409e-05,
      "loss": 1.0668,
      "step": 49868
    },
    {
      "epoch": 4.290544609825346,
      "grad_norm": 1.1796875,
      "learning_rate": 4.8900591463066854e-05,
      "loss": 1.0117,
      "step": 49869
    },
    {
      "epoch": 4.2906306461326675,
      "grad_norm": 9.75,
      "learning_rate": 4.8888928901677674e-05,
      "loss": 1.5677,
      "step": 49870
    },
    {
      "epoch": 4.29071668243999,
      "grad_norm": 1.3203125,
      "learning_rate": 4.887726765970063e-05,
      "loss": 1.7858,
      "step": 49871
    },
    {
      "epoch": 4.290802718747312,
      "grad_norm": 0.90234375,
      "learning_rate": 4.886560773716986e-05,
      "loss": 0.7539,
      "step": 49872
    },
    {
      "epoch": 4.290888755054633,
      "grad_norm": 1.0546875,
      "learning_rate": 4.885394913411939e-05,
      "loss": 1.2966,
      "step": 49873
    },
    {
      "epoch": 4.290974791361955,
      "grad_norm": 1.0234375,
      "learning_rate": 4.8842291850583364e-05,
      "loss": 0.7463,
      "step": 49874
    },
    {
      "epoch": 4.291060827669276,
      "grad_norm": 1.8125,
      "learning_rate": 4.8830635886595975e-05,
      "loss": 0.7509,
      "step": 49875
    },
    {
      "epoch": 4.291146863976598,
      "grad_norm": 7.0625,
      "learning_rate": 4.881898124219114e-05,
      "loss": 0.9534,
      "step": 49876
    },
    {
      "epoch": 4.291232900283919,
      "grad_norm": 1.296875,
      "learning_rate": 4.880732791740305e-05,
      "loss": 1.0055,
      "step": 49877
    },
    {
      "epoch": 4.291318936591241,
      "grad_norm": 1.6171875,
      "learning_rate": 4.879567591226586e-05,
      "loss": 1.3842,
      "step": 49878
    },
    {
      "epoch": 4.2914049728985635,
      "grad_norm": 0.96875,
      "learning_rate": 4.878402522681347e-05,
      "loss": 1.1273,
      "step": 49879
    },
    {
      "epoch": 4.291491009205885,
      "grad_norm": 1.1171875,
      "learning_rate": 4.877237586108002e-05,
      "loss": 0.9434,
      "step": 49880
    },
    {
      "epoch": 4.291577045513207,
      "grad_norm": 1.0234375,
      "learning_rate": 4.876072781509977e-05,
      "loss": 0.777,
      "step": 49881
    },
    {
      "epoch": 4.291663081820528,
      "grad_norm": 1.578125,
      "learning_rate": 4.874908108890647e-05,
      "loss": 1.2201,
      "step": 49882
    },
    {
      "epoch": 4.29174911812785,
      "grad_norm": 0.95703125,
      "learning_rate": 4.873743568253436e-05,
      "loss": 0.6364,
      "step": 49883
    },
    {
      "epoch": 4.291835154435172,
      "grad_norm": 1.609375,
      "learning_rate": 4.872579159601759e-05,
      "loss": 0.8299,
      "step": 49884
    },
    {
      "epoch": 4.291921190742493,
      "grad_norm": 1.375,
      "learning_rate": 4.8714148829390016e-05,
      "loss": 1.3854,
      "step": 49885
    },
    {
      "epoch": 4.292007227049815,
      "grad_norm": 0.828125,
      "learning_rate": 4.870250738268589e-05,
      "loss": 0.4568,
      "step": 49886
    },
    {
      "epoch": 4.2920932633571365,
      "grad_norm": 1.4140625,
      "learning_rate": 4.869086725593908e-05,
      "loss": 1.3905,
      "step": 49887
    },
    {
      "epoch": 4.2921792996644585,
      "grad_norm": 1.1796875,
      "learning_rate": 4.8679228449183655e-05,
      "loss": 1.3481,
      "step": 49888
    },
    {
      "epoch": 4.29226533597178,
      "grad_norm": 1.15625,
      "learning_rate": 4.866759096245388e-05,
      "loss": 1.0371,
      "step": 49889
    },
    {
      "epoch": 4.292351372279102,
      "grad_norm": 1.1328125,
      "learning_rate": 4.865595479578344e-05,
      "loss": 1.2295,
      "step": 49890
    },
    {
      "epoch": 4.292437408586424,
      "grad_norm": 2.046875,
      "learning_rate": 4.86443199492066e-05,
      "loss": 1.4282,
      "step": 49891
    },
    {
      "epoch": 4.292523444893745,
      "grad_norm": 1.3125,
      "learning_rate": 4.8632686422757425e-05,
      "loss": 1.3442,
      "step": 49892
    },
    {
      "epoch": 4.292609481201067,
      "grad_norm": 0.87890625,
      "learning_rate": 4.862105421646967e-05,
      "loss": 1.0558,
      "step": 49893
    },
    {
      "epoch": 4.292695517508388,
      "grad_norm": 0.94921875,
      "learning_rate": 4.8609423330377766e-05,
      "loss": 0.7414,
      "step": 49894
    },
    {
      "epoch": 4.29278155381571,
      "grad_norm": 50.75,
      "learning_rate": 4.859779376451534e-05,
      "loss": 1.0818,
      "step": 49895
    },
    {
      "epoch": 4.2928675901230315,
      "grad_norm": 478.0,
      "learning_rate": 4.8586165518916646e-05,
      "loss": 0.8252,
      "step": 49896
    },
    {
      "epoch": 4.292953626430354,
      "grad_norm": 1.3984375,
      "learning_rate": 4.8574538593615656e-05,
      "loss": 0.8874,
      "step": 49897
    },
    {
      "epoch": 4.293039662737676,
      "grad_norm": 1.40625,
      "learning_rate": 4.8562912988646294e-05,
      "loss": 1.4999,
      "step": 49898
    },
    {
      "epoch": 4.293125699044997,
      "grad_norm": 0.890625,
      "learning_rate": 4.8551288704042575e-05,
      "loss": 0.7905,
      "step": 49899
    },
    {
      "epoch": 4.293211735352319,
      "grad_norm": 51.75,
      "learning_rate": 4.853966573983865e-05,
      "loss": 1.1698,
      "step": 49900
    },
    {
      "epoch": 4.29329777165964,
      "grad_norm": 0.84375,
      "learning_rate": 4.8528044096068316e-05,
      "loss": 0.4823,
      "step": 49901
    },
    {
      "epoch": 4.293383807966962,
      "grad_norm": 29.75,
      "learning_rate": 4.851642377276571e-05,
      "loss": 0.6637,
      "step": 49902
    },
    {
      "epoch": 4.293469844274283,
      "grad_norm": 1.796875,
      "learning_rate": 4.8504804769964714e-05,
      "loss": 1.5855,
      "step": 49903
    },
    {
      "epoch": 4.293555880581605,
      "grad_norm": 0.921875,
      "learning_rate": 4.849318708769929e-05,
      "loss": 1.1278,
      "step": 49904
    },
    {
      "epoch": 4.2936419168889275,
      "grad_norm": 17.25,
      "learning_rate": 4.848157072600362e-05,
      "loss": 1.1246,
      "step": 49905
    },
    {
      "epoch": 4.293727953196249,
      "grad_norm": 1.0703125,
      "learning_rate": 4.8469955684911416e-05,
      "loss": 0.8787,
      "step": 49906
    },
    {
      "epoch": 4.293813989503571,
      "grad_norm": 1.140625,
      "learning_rate": 4.845834196445681e-05,
      "loss": 1.1418,
      "step": 49907
    },
    {
      "epoch": 4.293900025810892,
      "grad_norm": 1.203125,
      "learning_rate": 4.844672956467377e-05,
      "loss": 0.7377,
      "step": 49908
    },
    {
      "epoch": 4.293986062118214,
      "grad_norm": 1.0234375,
      "learning_rate": 4.843511848559617e-05,
      "loss": 1.8309,
      "step": 49909
    },
    {
      "epoch": 4.294072098425535,
      "grad_norm": 1.03125,
      "learning_rate": 4.842350872725804e-05,
      "loss": 1.4305,
      "step": 49910
    },
    {
      "epoch": 4.294158134732857,
      "grad_norm": 1.7734375,
      "learning_rate": 4.841190028969339e-05,
      "loss": 1.2369,
      "step": 49911
    },
    {
      "epoch": 4.294244171040179,
      "grad_norm": 2.453125,
      "learning_rate": 4.8400293172935985e-05,
      "loss": 1.3971,
      "step": 49912
    },
    {
      "epoch": 4.2943302073475005,
      "grad_norm": 1.09375,
      "learning_rate": 4.838868737701996e-05,
      "loss": 1.036,
      "step": 49913
    },
    {
      "epoch": 4.2944162436548226,
      "grad_norm": 123.0,
      "learning_rate": 4.837708290197923e-05,
      "loss": 1.2545,
      "step": 49914
    },
    {
      "epoch": 4.294502279962144,
      "grad_norm": 5.71875,
      "learning_rate": 4.8365479747847604e-05,
      "loss": 1.3717,
      "step": 49915
    },
    {
      "epoch": 4.294588316269466,
      "grad_norm": 0.98828125,
      "learning_rate": 4.8353877914659226e-05,
      "loss": 0.5164,
      "step": 49916
    },
    {
      "epoch": 4.294674352576788,
      "grad_norm": 26.875,
      "learning_rate": 4.8342277402447786e-05,
      "loss": 0.8167,
      "step": 49917
    },
    {
      "epoch": 4.294760388884109,
      "grad_norm": 9.0625,
      "learning_rate": 4.833067821124737e-05,
      "loss": 1.0426,
      "step": 49918
    },
    {
      "epoch": 4.294846425191431,
      "grad_norm": 1.5546875,
      "learning_rate": 4.8319080341092004e-05,
      "loss": 1.0431,
      "step": 49919
    },
    {
      "epoch": 4.294932461498752,
      "grad_norm": 1.390625,
      "learning_rate": 4.830748379201533e-05,
      "loss": 1.3874,
      "step": 49920
    },
    {
      "epoch": 4.295018497806074,
      "grad_norm": 1.3515625,
      "learning_rate": 4.829588856405143e-05,
      "loss": 0.9919,
      "step": 49921
    },
    {
      "epoch": 4.2951045341133955,
      "grad_norm": 1.4765625,
      "learning_rate": 4.828429465723433e-05,
      "loss": 1.4599,
      "step": 49922
    },
    {
      "epoch": 4.295190570420718,
      "grad_norm": 3.5,
      "learning_rate": 4.827270207159767e-05,
      "loss": 1.518,
      "step": 49923
    },
    {
      "epoch": 4.29527660672804,
      "grad_norm": 1.0625,
      "learning_rate": 4.8261110807175546e-05,
      "loss": 1.8446,
      "step": 49924
    },
    {
      "epoch": 4.295362643035361,
      "grad_norm": 1.1015625,
      "learning_rate": 4.8249520864001864e-05,
      "loss": 0.8751,
      "step": 49925
    },
    {
      "epoch": 4.295448679342683,
      "grad_norm": 1.2109375,
      "learning_rate": 4.8237932242110424e-05,
      "loss": 0.6419,
      "step": 49926
    },
    {
      "epoch": 4.295534715650004,
      "grad_norm": 0.88671875,
      "learning_rate": 4.822634494153522e-05,
      "loss": 1.3264,
      "step": 49927
    },
    {
      "epoch": 4.295620751957326,
      "grad_norm": 1.34375,
      "learning_rate": 4.8214758962310035e-05,
      "loss": 1.0153,
      "step": 49928
    },
    {
      "epoch": 4.295706788264647,
      "grad_norm": 1.2578125,
      "learning_rate": 4.82031743044688e-05,
      "loss": 1.591,
      "step": 49929
    },
    {
      "epoch": 4.295792824571969,
      "grad_norm": 1.03125,
      "learning_rate": 4.819159096804543e-05,
      "loss": 1.171,
      "step": 49930
    },
    {
      "epoch": 4.2958788608792915,
      "grad_norm": 1.3984375,
      "learning_rate": 4.818000895307373e-05,
      "loss": 1.1487,
      "step": 49931
    },
    {
      "epoch": 4.295964897186613,
      "grad_norm": 6.3125,
      "learning_rate": 4.816842825958778e-05,
      "loss": 0.789,
      "step": 49932
    },
    {
      "epoch": 4.296050933493935,
      "grad_norm": 1.234375,
      "learning_rate": 4.815684888762117e-05,
      "loss": 1.5723,
      "step": 49933
    },
    {
      "epoch": 4.296136969801256,
      "grad_norm": 1.453125,
      "learning_rate": 4.8145270837207934e-05,
      "loss": 1.322,
      "step": 49934
    },
    {
      "epoch": 4.296223006108578,
      "grad_norm": 1.0,
      "learning_rate": 4.813369410838198e-05,
      "loss": 0.9334,
      "step": 49935
    },
    {
      "epoch": 4.296309042415899,
      "grad_norm": 1.234375,
      "learning_rate": 4.8122118701176944e-05,
      "loss": 1.2479,
      "step": 49936
    },
    {
      "epoch": 4.296395078723221,
      "grad_norm": 0.84765625,
      "learning_rate": 4.811054461562686e-05,
      "loss": 1.0992,
      "step": 49937
    },
    {
      "epoch": 4.296481115030543,
      "grad_norm": 1.6953125,
      "learning_rate": 4.809897185176565e-05,
      "loss": 1.8078,
      "step": 49938
    },
    {
      "epoch": 4.2965671513378645,
      "grad_norm": 0.86328125,
      "learning_rate": 4.8087400409627e-05,
      "loss": 0.8071,
      "step": 49939
    },
    {
      "epoch": 4.2966531876451866,
      "grad_norm": 1.359375,
      "learning_rate": 4.807583028924473e-05,
      "loss": 0.7891,
      "step": 49940
    },
    {
      "epoch": 4.296739223952508,
      "grad_norm": 1.390625,
      "learning_rate": 4.806426149065291e-05,
      "loss": 1.2633,
      "step": 49941
    },
    {
      "epoch": 4.29682526025983,
      "grad_norm": 0.9140625,
      "learning_rate": 4.8052694013885125e-05,
      "loss": 1.0748,
      "step": 49942
    },
    {
      "epoch": 4.296911296567151,
      "grad_norm": 1.265625,
      "learning_rate": 4.804112785897541e-05,
      "loss": 1.9235,
      "step": 49943
    },
    {
      "epoch": 4.296997332874473,
      "grad_norm": 4.59375,
      "learning_rate": 4.80295630259574e-05,
      "loss": 1.5602,
      "step": 49944
    },
    {
      "epoch": 4.297083369181795,
      "grad_norm": 1.4140625,
      "learning_rate": 4.801799951486502e-05,
      "loss": 1.9298,
      "step": 49945
    },
    {
      "epoch": 4.297169405489116,
      "grad_norm": 0.7734375,
      "learning_rate": 4.8006437325732174e-05,
      "loss": 0.898,
      "step": 49946
    },
    {
      "epoch": 4.297255441796438,
      "grad_norm": 1.1015625,
      "learning_rate": 4.7994876458592465e-05,
      "loss": 1.0794,
      "step": 49947
    },
    {
      "epoch": 4.2973414781037595,
      "grad_norm": 1.5,
      "learning_rate": 4.7983316913479845e-05,
      "loss": 1.0958,
      "step": 49948
    },
    {
      "epoch": 4.297427514411082,
      "grad_norm": 1.8984375,
      "learning_rate": 4.797175869042825e-05,
      "loss": 0.8548,
      "step": 49949
    },
    {
      "epoch": 4.297513550718403,
      "grad_norm": 1.0390625,
      "learning_rate": 4.7960201789471195e-05,
      "loss": 1.1262,
      "step": 49950
    },
    {
      "epoch": 4.297599587025725,
      "grad_norm": 1.1015625,
      "learning_rate": 4.794864621064266e-05,
      "loss": 0.6204,
      "step": 49951
    },
    {
      "epoch": 4.297685623333047,
      "grad_norm": 1.3203125,
      "learning_rate": 4.7937091953976496e-05,
      "loss": 1.68,
      "step": 49952
    },
    {
      "epoch": 4.297771659640368,
      "grad_norm": 1.046875,
      "learning_rate": 4.7925539019506356e-05,
      "loss": 0.5895,
      "step": 49953
    },
    {
      "epoch": 4.29785769594769,
      "grad_norm": 1.2109375,
      "learning_rate": 4.7913987407266045e-05,
      "loss": 2.0143,
      "step": 49954
    },
    {
      "epoch": 4.297943732255011,
      "grad_norm": 1.0,
      "learning_rate": 4.790243711728953e-05,
      "loss": 1.2198,
      "step": 49955
    },
    {
      "epoch": 4.298029768562333,
      "grad_norm": 0.8359375,
      "learning_rate": 4.7890888149610345e-05,
      "loss": 0.7765,
      "step": 49956
    },
    {
      "epoch": 4.298115804869655,
      "grad_norm": 1.1875,
      "learning_rate": 4.787934050426246e-05,
      "loss": 1.1893,
      "step": 49957
    },
    {
      "epoch": 4.298201841176977,
      "grad_norm": 1.015625,
      "learning_rate": 4.786779418127951e-05,
      "loss": 0.8578,
      "step": 49958
    },
    {
      "epoch": 4.298287877484299,
      "grad_norm": 1.0234375,
      "learning_rate": 4.785624918069531e-05,
      "loss": 0.8348,
      "step": 49959
    },
    {
      "epoch": 4.29837391379162,
      "grad_norm": 1.1640625,
      "learning_rate": 4.784470550254377e-05,
      "loss": 0.9257,
      "step": 49960
    },
    {
      "epoch": 4.298459950098942,
      "grad_norm": 0.7265625,
      "learning_rate": 4.783316314685837e-05,
      "loss": 0.3153,
      "step": 49961
    },
    {
      "epoch": 4.298545986406263,
      "grad_norm": 1.078125,
      "learning_rate": 4.782162211367308e-05,
      "loss": 1.136,
      "step": 49962
    },
    {
      "epoch": 4.298632022713585,
      "grad_norm": 1.140625,
      "learning_rate": 4.7810082403021646e-05,
      "loss": 1.7383,
      "step": 49963
    },
    {
      "epoch": 4.298718059020906,
      "grad_norm": 1.4921875,
      "learning_rate": 4.779854401493772e-05,
      "loss": 0.879,
      "step": 49964
    },
    {
      "epoch": 4.2988040953282285,
      "grad_norm": 0.9140625,
      "learning_rate": 4.778700694945509e-05,
      "loss": 1.4246,
      "step": 49965
    },
    {
      "epoch": 4.298890131635551,
      "grad_norm": 2.140625,
      "learning_rate": 4.777547120660764e-05,
      "loss": 0.7104,
      "step": 49966
    },
    {
      "epoch": 4.298976167942872,
      "grad_norm": 1.296875,
      "learning_rate": 4.776393678642876e-05,
      "loss": 1.3796,
      "step": 49967
    },
    {
      "epoch": 4.299062204250194,
      "grad_norm": 1.078125,
      "learning_rate": 4.775240368895262e-05,
      "loss": 0.8942,
      "step": 49968
    },
    {
      "epoch": 4.299148240557515,
      "grad_norm": 1.015625,
      "learning_rate": 4.774087191421261e-05,
      "loss": 1.1846,
      "step": 49969
    },
    {
      "epoch": 4.299234276864837,
      "grad_norm": 1.21875,
      "learning_rate": 4.772934146224261e-05,
      "loss": 0.8835,
      "step": 49970
    },
    {
      "epoch": 4.299320313172158,
      "grad_norm": 1.2109375,
      "learning_rate": 4.7717812333076427e-05,
      "loss": 0.5072,
      "step": 49971
    },
    {
      "epoch": 4.29940634947948,
      "grad_norm": 1.1328125,
      "learning_rate": 4.770628452674758e-05,
      "loss": 1.6999,
      "step": 49972
    },
    {
      "epoch": 4.299492385786802,
      "grad_norm": 0.6875,
      "learning_rate": 4.7694758043289985e-05,
      "loss": 0.9849,
      "step": 49973
    },
    {
      "epoch": 4.2995784220941236,
      "grad_norm": 1.1484375,
      "learning_rate": 4.768323288273718e-05,
      "loss": 0.9776,
      "step": 49974
    },
    {
      "epoch": 4.299664458401446,
      "grad_norm": 1.375,
      "learning_rate": 4.767170904512291e-05,
      "loss": 1.9126,
      "step": 49975
    },
    {
      "epoch": 4.299750494708767,
      "grad_norm": 0.94921875,
      "learning_rate": 4.7660186530481044e-05,
      "loss": 0.8404,
      "step": 49976
    },
    {
      "epoch": 4.299836531016089,
      "grad_norm": 1.0234375,
      "learning_rate": 4.764866533884504e-05,
      "loss": 0.7393,
      "step": 49977
    },
    {
      "epoch": 4.29992256732341,
      "grad_norm": 0.6484375,
      "learning_rate": 4.763714547024872e-05,
      "loss": 0.4819,
      "step": 49978
    },
    {
      "epoch": 4.300008603630732,
      "grad_norm": 1.1015625,
      "learning_rate": 4.762562692472583e-05,
      "loss": 1.847,
      "step": 49979
    },
    {
      "epoch": 4.300094639938054,
      "grad_norm": 1.4765625,
      "learning_rate": 4.761410970230995e-05,
      "loss": 1.7719,
      "step": 49980
    },
    {
      "epoch": 4.300180676245375,
      "grad_norm": 1.015625,
      "learning_rate": 4.7602593803034774e-05,
      "loss": 0.9309,
      "step": 49981
    },
    {
      "epoch": 4.300266712552697,
      "grad_norm": 0.71484375,
      "learning_rate": 4.7591079226934175e-05,
      "loss": 0.7223,
      "step": 49982
    },
    {
      "epoch": 4.300352748860019,
      "grad_norm": 1.5078125,
      "learning_rate": 4.757956597404156e-05,
      "loss": 1.0653,
      "step": 49983
    },
    {
      "epoch": 4.300438785167341,
      "grad_norm": 1.0390625,
      "learning_rate": 4.75680540443908e-05,
      "loss": 1.03,
      "step": 49984
    },
    {
      "epoch": 4.300524821474662,
      "grad_norm": 0.890625,
      "learning_rate": 4.755654343801541e-05,
      "loss": 0.9249,
      "step": 49985
    },
    {
      "epoch": 4.300610857781984,
      "grad_norm": 1.1484375,
      "learning_rate": 4.754503415494915e-05,
      "loss": 0.89,
      "step": 49986
    },
    {
      "epoch": 4.300696894089306,
      "grad_norm": 0.91015625,
      "learning_rate": 4.753352619522572e-05,
      "loss": 0.9459,
      "step": 49987
    },
    {
      "epoch": 4.300782930396627,
      "grad_norm": 1.3046875,
      "learning_rate": 4.752201955887869e-05,
      "loss": 0.9581,
      "step": 49988
    },
    {
      "epoch": 4.300868966703949,
      "grad_norm": 1.5703125,
      "learning_rate": 4.751051424594172e-05,
      "loss": 1.2861,
      "step": 49989
    },
    {
      "epoch": 4.30095500301127,
      "grad_norm": 1.2265625,
      "learning_rate": 4.7499010256448536e-05,
      "loss": 1.4437,
      "step": 49990
    },
    {
      "epoch": 4.3010410393185925,
      "grad_norm": 1.5390625,
      "learning_rate": 4.748750759043274e-05,
      "loss": 2.2294,
      "step": 49991
    },
    {
      "epoch": 4.301127075625914,
      "grad_norm": 59.75,
      "learning_rate": 4.747600624792792e-05,
      "loss": 1.0132,
      "step": 49992
    },
    {
      "epoch": 4.301213111933236,
      "grad_norm": 1.0859375,
      "learning_rate": 4.746450622896786e-05,
      "loss": 1.1381,
      "step": 49993
    },
    {
      "epoch": 4.301299148240558,
      "grad_norm": 1.21875,
      "learning_rate": 4.7453007533586054e-05,
      "loss": 1.1961,
      "step": 49994
    },
    {
      "epoch": 4.301385184547879,
      "grad_norm": 6.0625,
      "learning_rate": 4.7441510161816134e-05,
      "loss": 1.0311,
      "step": 49995
    },
    {
      "epoch": 4.301471220855201,
      "grad_norm": 1.328125,
      "learning_rate": 4.743001411369191e-05,
      "loss": 1.3136,
      "step": 49996
    },
    {
      "epoch": 4.301557257162522,
      "grad_norm": 3.78125,
      "learning_rate": 4.741851938924674e-05,
      "loss": 0.9617,
      "step": 49997
    },
    {
      "epoch": 4.301643293469844,
      "grad_norm": 1.1484375,
      "learning_rate": 4.740702598851454e-05,
      "loss": 1.2705,
      "step": 49998
    },
    {
      "epoch": 4.301729329777166,
      "grad_norm": 1.203125,
      "learning_rate": 4.739553391152862e-05,
      "loss": 1.2756,
      "step": 49999
    },
    {
      "epoch": 4.301815366084488,
      "grad_norm": 1.3125,
      "learning_rate": 4.738404315832273e-05,
      "loss": 1.8619,
      "step": 50000
    },
    {
      "epoch": 4.30190140239181,
      "grad_norm": 1.25,
      "learning_rate": 4.737255372893062e-05,
      "loss": 1.5124,
      "step": 50001
    },
    {
      "epoch": 4.301987438699131,
      "grad_norm": 1.0625,
      "learning_rate": 4.73610656233856e-05,
      "loss": 2.0306,
      "step": 50002
    },
    {
      "epoch": 4.302073475006453,
      "grad_norm": 0.94921875,
      "learning_rate": 4.734957884172147e-05,
      "loss": 1.1818,
      "step": 50003
    },
    {
      "epoch": 4.302159511313774,
      "grad_norm": 1.78125,
      "learning_rate": 4.7338093383971827e-05,
      "loss": 1.2437,
      "step": 50004
    },
    {
      "epoch": 4.302245547621096,
      "grad_norm": 1.65625,
      "learning_rate": 4.7326609250170185e-05,
      "loss": 1.7147,
      "step": 50005
    },
    {
      "epoch": 4.302331583928418,
      "grad_norm": 1.4296875,
      "learning_rate": 4.7315126440350245e-05,
      "loss": 0.676,
      "step": 50006
    },
    {
      "epoch": 4.302417620235739,
      "grad_norm": 1.0703125,
      "learning_rate": 4.730364495454542e-05,
      "loss": 0.818,
      "step": 50007
    },
    {
      "epoch": 4.302503656543061,
      "grad_norm": 1.046875,
      "learning_rate": 4.7292164792789414e-05,
      "loss": 1.3771,
      "step": 50008
    },
    {
      "epoch": 4.302589692850383,
      "grad_norm": 3.90625,
      "learning_rate": 4.7280685955115864e-05,
      "loss": 1.3104,
      "step": 50009
    },
    {
      "epoch": 4.302675729157705,
      "grad_norm": 0.96875,
      "learning_rate": 4.726920844155819e-05,
      "loss": 1.3763,
      "step": 50010
    },
    {
      "epoch": 4.302761765465026,
      "grad_norm": 1.2265625,
      "learning_rate": 4.725773225215002e-05,
      "loss": 1.6952,
      "step": 50011
    },
    {
      "epoch": 4.302847801772348,
      "grad_norm": 0.921875,
      "learning_rate": 4.724625738692501e-05,
      "loss": 1.3218,
      "step": 50012
    },
    {
      "epoch": 4.30293383807967,
      "grad_norm": 1.234375,
      "learning_rate": 4.723478384591656e-05,
      "loss": 1.4304,
      "step": 50013
    },
    {
      "epoch": 4.303019874386991,
      "grad_norm": 1.109375,
      "learning_rate": 4.722331162915838e-05,
      "loss": 0.7815,
      "step": 50014
    },
    {
      "epoch": 4.303105910694313,
      "grad_norm": 1.1875,
      "learning_rate": 4.7211840736683884e-05,
      "loss": 0.7049,
      "step": 50015
    },
    {
      "epoch": 4.303191947001634,
      "grad_norm": 1.1640625,
      "learning_rate": 4.7200371168526655e-05,
      "loss": 0.936,
      "step": 50016
    },
    {
      "epoch": 4.3032779833089565,
      "grad_norm": 1.578125,
      "learning_rate": 4.718890292472045e-05,
      "loss": 1.0626,
      "step": 50017
    },
    {
      "epoch": 4.303364019616278,
      "grad_norm": 1.3125,
      "learning_rate": 4.717743600529845e-05,
      "loss": 1.5214,
      "step": 50018
    },
    {
      "epoch": 4.3034500559236,
      "grad_norm": 0.8125,
      "learning_rate": 4.716597041029441e-05,
      "loss": 0.8757,
      "step": 50019
    },
    {
      "epoch": 4.303536092230922,
      "grad_norm": 1.359375,
      "learning_rate": 4.7154506139741985e-05,
      "loss": 1.1918,
      "step": 50020
    },
    {
      "epoch": 4.303622128538243,
      "grad_norm": 1.3828125,
      "learning_rate": 4.7143043193674405e-05,
      "loss": 1.0014,
      "step": 50021
    },
    {
      "epoch": 4.303708164845565,
      "grad_norm": 1.109375,
      "learning_rate": 4.7131581572125324e-05,
      "loss": 0.8438,
      "step": 50022
    },
    {
      "epoch": 4.303794201152886,
      "grad_norm": 1.0703125,
      "learning_rate": 4.712012127512844e-05,
      "loss": 0.9164,
      "step": 50023
    },
    {
      "epoch": 4.303880237460208,
      "grad_norm": 0.765625,
      "learning_rate": 4.7108662302716986e-05,
      "loss": 0.646,
      "step": 50024
    },
    {
      "epoch": 4.3039662737675295,
      "grad_norm": 1.2890625,
      "learning_rate": 4.709720465492473e-05,
      "loss": 1.2278,
      "step": 50025
    },
    {
      "epoch": 4.304052310074852,
      "grad_norm": 0.88671875,
      "learning_rate": 4.7085748331784974e-05,
      "loss": 0.9156,
      "step": 50026
    },
    {
      "epoch": 4.304138346382174,
      "grad_norm": 2.21875,
      "learning_rate": 4.7074293333331295e-05,
      "loss": 0.8326,
      "step": 50027
    },
    {
      "epoch": 4.304224382689495,
      "grad_norm": 1.4453125,
      "learning_rate": 4.706283965959734e-05,
      "loss": 1.2086,
      "step": 50028
    },
    {
      "epoch": 4.304310418996817,
      "grad_norm": 30.875,
      "learning_rate": 4.705138731061642e-05,
      "loss": 1.4286,
      "step": 50029
    },
    {
      "epoch": 4.304396455304138,
      "grad_norm": 0.91796875,
      "learning_rate": 4.703993628642206e-05,
      "loss": 0.5428,
      "step": 50030
    },
    {
      "epoch": 4.30448249161146,
      "grad_norm": 4.78125,
      "learning_rate": 4.702848658704783e-05,
      "loss": 1.3901,
      "step": 50031
    },
    {
      "epoch": 4.304568527918782,
      "grad_norm": 0.9609375,
      "learning_rate": 4.701703821252717e-05,
      "loss": 0.9811,
      "step": 50032
    },
    {
      "epoch": 4.304654564226103,
      "grad_norm": 1.125,
      "learning_rate": 4.700559116289355e-05,
      "loss": 1.0014,
      "step": 50033
    },
    {
      "epoch": 4.304740600533425,
      "grad_norm": 1.234375,
      "learning_rate": 4.699414543818053e-05,
      "loss": 1.1926,
      "step": 50034
    },
    {
      "epoch": 4.304826636840747,
      "grad_norm": 109056.0,
      "learning_rate": 4.6982701038421505e-05,
      "loss": 1.2664,
      "step": 50035
    },
    {
      "epoch": 4.304912673148069,
      "grad_norm": 0.94921875,
      "learning_rate": 4.697125796364993e-05,
      "loss": 0.688,
      "step": 50036
    },
    {
      "epoch": 4.30499870945539,
      "grad_norm": 0.8203125,
      "learning_rate": 4.695981621389939e-05,
      "loss": 0.3517,
      "step": 50037
    },
    {
      "epoch": 4.305084745762712,
      "grad_norm": 0.984375,
      "learning_rate": 4.694837578920319e-05,
      "loss": 1.1111,
      "step": 50038
    },
    {
      "epoch": 4.305170782070034,
      "grad_norm": 0.828125,
      "learning_rate": 4.6936936689595035e-05,
      "loss": 0.9146,
      "step": 50039
    },
    {
      "epoch": 4.305256818377355,
      "grad_norm": 1.0546875,
      "learning_rate": 4.6925498915108055e-05,
      "loss": 1.3463,
      "step": 50040
    },
    {
      "epoch": 4.305342854684677,
      "grad_norm": 0.80078125,
      "learning_rate": 4.691406246577579e-05,
      "loss": 0.5474,
      "step": 50041
    },
    {
      "epoch": 4.305428890991998,
      "grad_norm": 0.9140625,
      "learning_rate": 4.690262734163198e-05,
      "loss": 1.0332,
      "step": 50042
    },
    {
      "epoch": 4.3055149272993205,
      "grad_norm": 1.15625,
      "learning_rate": 4.689119354270977e-05,
      "loss": 0.4962,
      "step": 50043
    },
    {
      "epoch": 4.305600963606642,
      "grad_norm": 0.89453125,
      "learning_rate": 4.687976106904279e-05,
      "loss": 0.7441,
      "step": 50044
    },
    {
      "epoch": 4.305686999913964,
      "grad_norm": 1.65625,
      "learning_rate": 4.686832992066425e-05,
      "loss": 0.8714,
      "step": 50045
    },
    {
      "epoch": 4.305773036221286,
      "grad_norm": 64.5,
      "learning_rate": 4.685690009760779e-05,
      "loss": 1.405,
      "step": 50046
    },
    {
      "epoch": 4.305859072528607,
      "grad_norm": 0.8203125,
      "learning_rate": 4.684547159990682e-05,
      "loss": 1.1306,
      "step": 50047
    },
    {
      "epoch": 4.305945108835929,
      "grad_norm": 0.921875,
      "learning_rate": 4.683404442759459e-05,
      "loss": 1.1517,
      "step": 50048
    },
    {
      "epoch": 4.30603114514325,
      "grad_norm": 0.96484375,
      "learning_rate": 4.682261858070469e-05,
      "loss": 1.1567,
      "step": 50049
    },
    {
      "epoch": 4.306117181450572,
      "grad_norm": 55552.0,
      "learning_rate": 4.6811194059270534e-05,
      "loss": 1.5636,
      "step": 50050
    },
    {
      "epoch": 4.3062032177578935,
      "grad_norm": 1.2421875,
      "learning_rate": 4.679977086332543e-05,
      "loss": 1.3362,
      "step": 50051
    },
    {
      "epoch": 4.306289254065216,
      "grad_norm": 1.1015625,
      "learning_rate": 4.6788348992902855e-05,
      "loss": 1.1626,
      "step": 50052
    },
    {
      "epoch": 4.306375290372538,
      "grad_norm": 1.078125,
      "learning_rate": 4.677692844803633e-05,
      "loss": 1.4143,
      "step": 50053
    },
    {
      "epoch": 4.306461326679859,
      "grad_norm": 2.359375,
      "learning_rate": 4.6765509228759e-05,
      "loss": 1.09,
      "step": 50054
    },
    {
      "epoch": 4.306547362987181,
      "grad_norm": 1.2109375,
      "learning_rate": 4.675409133510455e-05,
      "loss": 1.0946,
      "step": 50055
    },
    {
      "epoch": 4.306633399294502,
      "grad_norm": 0.859375,
      "learning_rate": 4.674267476710608e-05,
      "loss": 0.6296,
      "step": 50056
    },
    {
      "epoch": 4.306719435601824,
      "grad_norm": 1.6796875,
      "learning_rate": 4.673125952479718e-05,
      "loss": 1.2537,
      "step": 50057
    },
    {
      "epoch": 4.306805471909145,
      "grad_norm": 1.46875,
      "learning_rate": 4.67198456082113e-05,
      "loss": 2.1995,
      "step": 50058
    },
    {
      "epoch": 4.306891508216467,
      "grad_norm": 1.0078125,
      "learning_rate": 4.670843301738159e-05,
      "loss": 0.6796,
      "step": 50059
    },
    {
      "epoch": 4.3069775445237894,
      "grad_norm": 0.9140625,
      "learning_rate": 4.669702175234153e-05,
      "loss": 1.3709,
      "step": 50060
    },
    {
      "epoch": 4.307063580831111,
      "grad_norm": 1.1796875,
      "learning_rate": 4.668561181312464e-05,
      "loss": 0.8596,
      "step": 50061
    },
    {
      "epoch": 4.307149617138433,
      "grad_norm": 3.46875,
      "learning_rate": 4.667420319976406e-05,
      "loss": 1.2722,
      "step": 50062
    },
    {
      "epoch": 4.307235653445754,
      "grad_norm": 93.0,
      "learning_rate": 4.6662795912293274e-05,
      "loss": 0.8603,
      "step": 50063
    },
    {
      "epoch": 4.307321689753076,
      "grad_norm": 8.3125,
      "learning_rate": 4.665138995074575e-05,
      "loss": 1.4996,
      "step": 50064
    },
    {
      "epoch": 4.307407726060397,
      "grad_norm": 1.0234375,
      "learning_rate": 4.6639985315154674e-05,
      "loss": 0.7536,
      "step": 50065
    },
    {
      "epoch": 4.307493762367719,
      "grad_norm": 1.234375,
      "learning_rate": 4.662858200555342e-05,
      "loss": 1.3489,
      "step": 50066
    },
    {
      "epoch": 4.307579798675041,
      "grad_norm": 1.0078125,
      "learning_rate": 4.661718002197546e-05,
      "loss": 0.573,
      "step": 50067
    },
    {
      "epoch": 4.307665834982362,
      "grad_norm": 1.203125,
      "learning_rate": 4.6605779364454035e-05,
      "loss": 1.5567,
      "step": 50068
    },
    {
      "epoch": 4.3077518712896845,
      "grad_norm": 1.140625,
      "learning_rate": 4.6594380033022577e-05,
      "loss": 0.5856,
      "step": 50069
    },
    {
      "epoch": 4.307837907597006,
      "grad_norm": 0.8984375,
      "learning_rate": 4.6582982027714324e-05,
      "loss": 0.6982,
      "step": 50070
    },
    {
      "epoch": 4.307923943904328,
      "grad_norm": 1.203125,
      "learning_rate": 4.6571585348562595e-05,
      "loss": 2.156,
      "step": 50071
    },
    {
      "epoch": 4.308009980211649,
      "grad_norm": 1.0078125,
      "learning_rate": 4.6560189995600966e-05,
      "loss": 0.9236,
      "step": 50072
    },
    {
      "epoch": 4.308096016518971,
      "grad_norm": 1.0546875,
      "learning_rate": 4.6548795968862465e-05,
      "loss": 0.9351,
      "step": 50073
    },
    {
      "epoch": 4.308182052826293,
      "grad_norm": 1.234375,
      "learning_rate": 4.653740326838052e-05,
      "loss": 1.3202,
      "step": 50074
    },
    {
      "epoch": 4.308268089133614,
      "grad_norm": 1.390625,
      "learning_rate": 4.652601189418859e-05,
      "loss": 1.1356,
      "step": 50075
    },
    {
      "epoch": 4.308354125440936,
      "grad_norm": 228.0,
      "learning_rate": 4.6514621846319774e-05,
      "loss": 0.7476,
      "step": 50076
    },
    {
      "epoch": 4.3084401617482575,
      "grad_norm": 1.0234375,
      "learning_rate": 4.650323312480753e-05,
      "loss": 1.8896,
      "step": 50077
    },
    {
      "epoch": 4.30852619805558,
      "grad_norm": 1.328125,
      "learning_rate": 4.649184572968507e-05,
      "loss": 0.8396,
      "step": 50078
    },
    {
      "epoch": 4.308612234362901,
      "grad_norm": 1.171875,
      "learning_rate": 4.64804596609858e-05,
      "loss": 1.3021,
      "step": 50079
    },
    {
      "epoch": 4.308698270670223,
      "grad_norm": 1.109375,
      "learning_rate": 4.646907491874308e-05,
      "loss": 0.9691,
      "step": 50080
    },
    {
      "epoch": 4.308784306977545,
      "grad_norm": 1.2265625,
      "learning_rate": 4.645769150299001e-05,
      "loss": 1.2359,
      "step": 50081
    },
    {
      "epoch": 4.308870343284866,
      "grad_norm": 0.7265625,
      "learning_rate": 4.6446309413759936e-05,
      "loss": 0.6049,
      "step": 50082
    },
    {
      "epoch": 4.308956379592188,
      "grad_norm": 3.09375,
      "learning_rate": 4.643492865108634e-05,
      "loss": 1.4751,
      "step": 50083
    },
    {
      "epoch": 4.309042415899509,
      "grad_norm": 1.4296875,
      "learning_rate": 4.6423549215002245e-05,
      "loss": 1.2863,
      "step": 50084
    },
    {
      "epoch": 4.309128452206831,
      "grad_norm": 1.4375,
      "learning_rate": 4.641217110554119e-05,
      "loss": 1.3686,
      "step": 50085
    },
    {
      "epoch": 4.309214488514153,
      "grad_norm": 1.0390625,
      "learning_rate": 4.640079432273614e-05,
      "loss": 0.8745,
      "step": 50086
    },
    {
      "epoch": 4.309300524821475,
      "grad_norm": 1.0,
      "learning_rate": 4.638941886662063e-05,
      "loss": 1.1979,
      "step": 50087
    },
    {
      "epoch": 4.309386561128797,
      "grad_norm": 2.59375,
      "learning_rate": 4.6378044737227895e-05,
      "loss": 1.2888,
      "step": 50088
    },
    {
      "epoch": 4.309472597436118,
      "grad_norm": 1.0390625,
      "learning_rate": 4.63666719345911e-05,
      "loss": 1.3515,
      "step": 50089
    },
    {
      "epoch": 4.30955863374344,
      "grad_norm": 1.2578125,
      "learning_rate": 4.635530045874353e-05,
      "loss": 1.0456,
      "step": 50090
    },
    {
      "epoch": 4.309644670050761,
      "grad_norm": 1.28125,
      "learning_rate": 4.6343930309718563e-05,
      "loss": 0.7408,
      "step": 50091
    },
    {
      "epoch": 4.309730706358083,
      "grad_norm": 3.890625,
      "learning_rate": 4.6332561487549275e-05,
      "loss": 1.4,
      "step": 50092
    },
    {
      "epoch": 4.309816742665404,
      "grad_norm": 1.328125,
      "learning_rate": 4.6321193992269025e-05,
      "loss": 0.9799,
      "step": 50093
    },
    {
      "epoch": 4.3099027789727264,
      "grad_norm": 1.046875,
      "learning_rate": 4.630982782391113e-05,
      "loss": 0.628,
      "step": 50094
    },
    {
      "epoch": 4.3099888152800485,
      "grad_norm": 80.5,
      "learning_rate": 4.6298462982508725e-05,
      "loss": 1.2899,
      "step": 50095
    },
    {
      "epoch": 4.31007485158737,
      "grad_norm": 1.5703125,
      "learning_rate": 4.628709946809506e-05,
      "loss": 1.5654,
      "step": 50096
    },
    {
      "epoch": 4.310160887894692,
      "grad_norm": 1.5859375,
      "learning_rate": 4.627573728070339e-05,
      "loss": 1.7363,
      "step": 50097
    },
    {
      "epoch": 4.310246924202013,
      "grad_norm": 1.3359375,
      "learning_rate": 4.6264376420366905e-05,
      "loss": 1.4443,
      "step": 50098
    },
    {
      "epoch": 4.310332960509335,
      "grad_norm": 1.0,
      "learning_rate": 4.625301688711891e-05,
      "loss": 1.3,
      "step": 50099
    },
    {
      "epoch": 4.310418996816656,
      "grad_norm": 1.390625,
      "learning_rate": 4.624165868099256e-05,
      "loss": 1.1798,
      "step": 50100
    },
    {
      "epoch": 4.310505033123978,
      "grad_norm": 0.9453125,
      "learning_rate": 4.623030180202109e-05,
      "loss": 1.1122,
      "step": 50101
    },
    {
      "epoch": 4.3105910694313,
      "grad_norm": 0.7734375,
      "learning_rate": 4.621894625023787e-05,
      "loss": 0.8246,
      "step": 50102
    },
    {
      "epoch": 4.3106771057386215,
      "grad_norm": 280.0,
      "learning_rate": 4.620759202567582e-05,
      "loss": 1.4511,
      "step": 50103
    },
    {
      "epoch": 4.310763142045944,
      "grad_norm": 0.84765625,
      "learning_rate": 4.619623912836835e-05,
      "loss": 1.2498,
      "step": 50104
    },
    {
      "epoch": 4.310849178353265,
      "grad_norm": 0.921875,
      "learning_rate": 4.6184887558348706e-05,
      "loss": 1.0447,
      "step": 50105
    },
    {
      "epoch": 4.310935214660587,
      "grad_norm": 1.7265625,
      "learning_rate": 4.6173537315649886e-05,
      "loss": 1.2939,
      "step": 50106
    },
    {
      "epoch": 4.311021250967908,
      "grad_norm": 0.890625,
      "learning_rate": 4.616218840030523e-05,
      "loss": 1.3013,
      "step": 50107
    },
    {
      "epoch": 4.31110728727523,
      "grad_norm": 0.90234375,
      "learning_rate": 4.615084081234799e-05,
      "loss": 0.6769,
      "step": 50108
    },
    {
      "epoch": 4.311193323582552,
      "grad_norm": 2.765625,
      "learning_rate": 4.613949455181116e-05,
      "loss": 0.7107,
      "step": 50109
    },
    {
      "epoch": 4.311279359889873,
      "grad_norm": 1.15625,
      "learning_rate": 4.612814961872819e-05,
      "loss": 0.92,
      "step": 50110
    },
    {
      "epoch": 4.311365396197195,
      "grad_norm": 1.03125,
      "learning_rate": 4.611680601313195e-05,
      "loss": 1.0139,
      "step": 50111
    },
    {
      "epoch": 4.311451432504517,
      "grad_norm": 1.109375,
      "learning_rate": 4.610546373505581e-05,
      "loss": 0.7419,
      "step": 50112
    },
    {
      "epoch": 4.311537468811839,
      "grad_norm": 1.34375,
      "learning_rate": 4.609412278453301e-05,
      "loss": 0.6777,
      "step": 50113
    },
    {
      "epoch": 4.31162350511916,
      "grad_norm": 1.3203125,
      "learning_rate": 4.608278316159648e-05,
      "loss": 1.1687,
      "step": 50114
    },
    {
      "epoch": 4.311709541426482,
      "grad_norm": 0.890625,
      "learning_rate": 4.607144486627956e-05,
      "loss": 1.0038,
      "step": 50115
    },
    {
      "epoch": 4.311795577733804,
      "grad_norm": 0.74609375,
      "learning_rate": 4.606010789861542e-05,
      "loss": 0.6028,
      "step": 50116
    },
    {
      "epoch": 4.311881614041125,
      "grad_norm": 0.69921875,
      "learning_rate": 4.604877225863713e-05,
      "loss": 0.4252,
      "step": 50117
    },
    {
      "epoch": 4.311967650348447,
      "grad_norm": 0.890625,
      "learning_rate": 4.6037437946377994e-05,
      "loss": 0.5977,
      "step": 50118
    },
    {
      "epoch": 4.312053686655768,
      "grad_norm": 0.87109375,
      "learning_rate": 4.602610496187093e-05,
      "loss": 0.3891,
      "step": 50119
    },
    {
      "epoch": 4.3121397229630904,
      "grad_norm": 1.390625,
      "learning_rate": 4.60147733051493e-05,
      "loss": 1.3127,
      "step": 50120
    },
    {
      "epoch": 4.3122257592704125,
      "grad_norm": 0.99609375,
      "learning_rate": 4.60034429762462e-05,
      "loss": 0.714,
      "step": 50121
    },
    {
      "epoch": 4.312311795577734,
      "grad_norm": 1.4765625,
      "learning_rate": 4.599211397519465e-05,
      "loss": 1.6695,
      "step": 50122
    },
    {
      "epoch": 4.312397831885056,
      "grad_norm": 1.140625,
      "learning_rate": 4.5980786302027846e-05,
      "loss": 0.5978,
      "step": 50123
    },
    {
      "epoch": 4.312483868192377,
      "grad_norm": 0.96484375,
      "learning_rate": 4.596945995677904e-05,
      "loss": 0.9012,
      "step": 50124
    },
    {
      "epoch": 4.312569904499699,
      "grad_norm": 1.2734375,
      "learning_rate": 4.59581349394812e-05,
      "loss": 0.8124,
      "step": 50125
    },
    {
      "epoch": 4.31265594080702,
      "grad_norm": 1.3984375,
      "learning_rate": 4.5946811250167585e-05,
      "loss": 0.8005,
      "step": 50126
    },
    {
      "epoch": 4.312741977114342,
      "grad_norm": 1.15625,
      "learning_rate": 4.5935488888871165e-05,
      "loss": 1.205,
      "step": 50127
    },
    {
      "epoch": 4.312828013421664,
      "grad_norm": 7.0,
      "learning_rate": 4.5924167855625086e-05,
      "loss": 0.8566,
      "step": 50128
    },
    {
      "epoch": 4.3129140497289855,
      "grad_norm": 0.9140625,
      "learning_rate": 4.591284815046259e-05,
      "loss": 0.9435,
      "step": 50129
    },
    {
      "epoch": 4.313000086036308,
      "grad_norm": 1.09375,
      "learning_rate": 4.5901529773416663e-05,
      "loss": 0.8719,
      "step": 50130
    },
    {
      "epoch": 4.313086122343629,
      "grad_norm": 1.1953125,
      "learning_rate": 4.589021272452043e-05,
      "loss": 1.0175,
      "step": 50131
    },
    {
      "epoch": 4.313172158650951,
      "grad_norm": 0.8203125,
      "learning_rate": 4.587889700380704e-05,
      "loss": 0.7964,
      "step": 50132
    },
    {
      "epoch": 4.313258194958272,
      "grad_norm": 1.0,
      "learning_rate": 4.5867582611309524e-05,
      "loss": 0.7013,
      "step": 50133
    },
    {
      "epoch": 4.313344231265594,
      "grad_norm": 1.078125,
      "learning_rate": 4.585626954706096e-05,
      "loss": 0.9372,
      "step": 50134
    },
    {
      "epoch": 4.313430267572916,
      "grad_norm": 1.4453125,
      "learning_rate": 4.58449578110946e-05,
      "loss": 1.3788,
      "step": 50135
    },
    {
      "epoch": 4.313516303880237,
      "grad_norm": 1.0,
      "learning_rate": 4.5833647403443314e-05,
      "loss": 0.6849,
      "step": 50136
    },
    {
      "epoch": 4.313602340187559,
      "grad_norm": 1.1875,
      "learning_rate": 4.58223383241404e-05,
      "loss": 0.6556,
      "step": 50137
    },
    {
      "epoch": 4.313688376494881,
      "grad_norm": 1.0078125,
      "learning_rate": 4.581103057321867e-05,
      "loss": 0.7698,
      "step": 50138
    },
    {
      "epoch": 4.313774412802203,
      "grad_norm": 1.2890625,
      "learning_rate": 4.579972415071137e-05,
      "loss": 1.1377,
      "step": 50139
    },
    {
      "epoch": 4.313860449109524,
      "grad_norm": 1.1640625,
      "learning_rate": 4.578841905665165e-05,
      "loss": 1.1813,
      "step": 50140
    },
    {
      "epoch": 4.313946485416846,
      "grad_norm": 1.3828125,
      "learning_rate": 4.577711529107237e-05,
      "loss": 1.6947,
      "step": 50141
    },
    {
      "epoch": 4.314032521724168,
      "grad_norm": 0.7109375,
      "learning_rate": 4.576581285400666e-05,
      "loss": 0.8607,
      "step": 50142
    },
    {
      "epoch": 4.314118558031489,
      "grad_norm": 11.8125,
      "learning_rate": 4.575451174548767e-05,
      "loss": 0.5525,
      "step": 50143
    },
    {
      "epoch": 4.314204594338811,
      "grad_norm": 1.3828125,
      "learning_rate": 4.574321196554837e-05,
      "loss": 1.819,
      "step": 50144
    },
    {
      "epoch": 4.314290630646132,
      "grad_norm": 0.921875,
      "learning_rate": 4.573191351422173e-05,
      "loss": 0.7019,
      "step": 50145
    },
    {
      "epoch": 4.3143766669534545,
      "grad_norm": 0.90234375,
      "learning_rate": 4.572061639154107e-05,
      "loss": 1.1794,
      "step": 50146
    },
    {
      "epoch": 4.314462703260776,
      "grad_norm": 2.375,
      "learning_rate": 4.570932059753913e-05,
      "loss": 0.6611,
      "step": 50147
    },
    {
      "epoch": 4.314548739568098,
      "grad_norm": 1.171875,
      "learning_rate": 4.5698026132249104e-05,
      "loss": 1.4896,
      "step": 50148
    },
    {
      "epoch": 4.31463477587542,
      "grad_norm": 1.359375,
      "learning_rate": 4.5686732995704026e-05,
      "loss": 1.5716,
      "step": 50149
    },
    {
      "epoch": 4.314720812182741,
      "grad_norm": 1.109375,
      "learning_rate": 4.5675441187936816e-05,
      "loss": 1.2054,
      "step": 50150
    },
    {
      "epoch": 4.314806848490063,
      "grad_norm": 1.1171875,
      "learning_rate": 4.5664150708980725e-05,
      "loss": 1.1663,
      "step": 50151
    },
    {
      "epoch": 4.314892884797384,
      "grad_norm": 1.1796875,
      "learning_rate": 4.5652861558868386e-05,
      "loss": 1.1621,
      "step": 50152
    },
    {
      "epoch": 4.314978921104706,
      "grad_norm": 0.99609375,
      "learning_rate": 4.5641573737633167e-05,
      "loss": 0.7318,
      "step": 50153
    },
    {
      "epoch": 4.315064957412028,
      "grad_norm": 3.921875,
      "learning_rate": 4.56302872453081e-05,
      "loss": 1.3031,
      "step": 50154
    },
    {
      "epoch": 4.3151509937193495,
      "grad_norm": 1.3984375,
      "learning_rate": 4.5619002081925986e-05,
      "loss": 0.734,
      "step": 50155
    },
    {
      "epoch": 4.315237030026672,
      "grad_norm": 1.265625,
      "learning_rate": 4.560771824751997e-05,
      "loss": 0.7234,
      "step": 50156
    },
    {
      "epoch": 4.315323066333993,
      "grad_norm": 1.203125,
      "learning_rate": 4.559643574212291e-05,
      "loss": 1.2302,
      "step": 50157
    },
    {
      "epoch": 4.315409102641315,
      "grad_norm": 1.40625,
      "learning_rate": 4.558515456576795e-05,
      "loss": 0.781,
      "step": 50158
    },
    {
      "epoch": 4.315495138948636,
      "grad_norm": 1.046875,
      "learning_rate": 4.5573874718488064e-05,
      "loss": 1.0463,
      "step": 50159
    },
    {
      "epoch": 4.315581175255958,
      "grad_norm": 1.0859375,
      "learning_rate": 4.556259620031616e-05,
      "loss": 0.6848,
      "step": 50160
    },
    {
      "epoch": 4.31566721156328,
      "grad_norm": 84.0,
      "learning_rate": 4.555131901128529e-05,
      "loss": 1.3738,
      "step": 50161
    },
    {
      "epoch": 4.315753247870601,
      "grad_norm": 1.1171875,
      "learning_rate": 4.5540043151428456e-05,
      "loss": 1.3181,
      "step": 50162
    },
    {
      "epoch": 4.315839284177923,
      "grad_norm": 1.046875,
      "learning_rate": 4.552876862077859e-05,
      "loss": 0.7245,
      "step": 50163
    },
    {
      "epoch": 4.315925320485245,
      "grad_norm": 1.3671875,
      "learning_rate": 4.551749541936861e-05,
      "loss": 1.6252,
      "step": 50164
    },
    {
      "epoch": 4.316011356792567,
      "grad_norm": 0.6171875,
      "learning_rate": 4.550622354723172e-05,
      "loss": 0.2976,
      "step": 50165
    },
    {
      "epoch": 4.316097393099888,
      "grad_norm": 0.85546875,
      "learning_rate": 4.5494953004400596e-05,
      "loss": 0.9551,
      "step": 50166
    },
    {
      "epoch": 4.31618342940721,
      "grad_norm": 1.171875,
      "learning_rate": 4.54836837909085e-05,
      "loss": 0.9189,
      "step": 50167
    },
    {
      "epoch": 4.316269465714532,
      "grad_norm": 1.25,
      "learning_rate": 4.5472415906788075e-05,
      "loss": 0.9298,
      "step": 50168
    },
    {
      "epoch": 4.316355502021853,
      "grad_norm": 1.3203125,
      "learning_rate": 4.546114935207246e-05,
      "loss": 1.1579,
      "step": 50169
    },
    {
      "epoch": 4.316441538329175,
      "grad_norm": 1.0703125,
      "learning_rate": 4.544988412679463e-05,
      "loss": 0.9436,
      "step": 50170
    },
    {
      "epoch": 4.316527574636496,
      "grad_norm": 1.3515625,
      "learning_rate": 4.543862023098744e-05,
      "loss": 1.3688,
      "step": 50171
    },
    {
      "epoch": 4.3166136109438185,
      "grad_norm": 95.5,
      "learning_rate": 4.5427357664683864e-05,
      "loss": 0.8223,
      "step": 50172
    },
    {
      "epoch": 4.31669964725114,
      "grad_norm": 1.046875,
      "learning_rate": 4.541609642791694e-05,
      "loss": 0.9177,
      "step": 50173
    },
    {
      "epoch": 4.316785683558462,
      "grad_norm": 1.078125,
      "learning_rate": 4.5404836520719414e-05,
      "loss": 0.9034,
      "step": 50174
    },
    {
      "epoch": 4.316871719865784,
      "grad_norm": 5.28125,
      "learning_rate": 4.539357794312438e-05,
      "loss": 1.1426,
      "step": 50175
    },
    {
      "epoch": 4.316957756173105,
      "grad_norm": 1.109375,
      "learning_rate": 4.5382320695164794e-05,
      "loss": 1.7722,
      "step": 50176
    },
    {
      "epoch": 4.317043792480427,
      "grad_norm": 772.0,
      "learning_rate": 4.537106477687336e-05,
      "loss": 0.8768,
      "step": 50177
    },
    {
      "epoch": 4.317129828787748,
      "grad_norm": 1.0390625,
      "learning_rate": 4.535981018828328e-05,
      "loss": 1.0702,
      "step": 50178
    },
    {
      "epoch": 4.31721586509507,
      "grad_norm": 7.375,
      "learning_rate": 4.5348556929427245e-05,
      "loss": 1.2672,
      "step": 50179
    },
    {
      "epoch": 4.3173019014023915,
      "grad_norm": 0.94921875,
      "learning_rate": 4.533730500033823e-05,
      "loss": 0.8388,
      "step": 50180
    },
    {
      "epoch": 4.3173879377097135,
      "grad_norm": 1.3515625,
      "learning_rate": 4.532605440104926e-05,
      "loss": 0.8844,
      "step": 50181
    },
    {
      "epoch": 4.317473974017036,
      "grad_norm": 1.25,
      "learning_rate": 4.531480513159308e-05,
      "loss": 1.3475,
      "step": 50182
    },
    {
      "epoch": 4.317560010324357,
      "grad_norm": 1.1015625,
      "learning_rate": 4.530355719200263e-05,
      "loss": 1.5058,
      "step": 50183
    },
    {
      "epoch": 4.317646046631679,
      "grad_norm": 1.21875,
      "learning_rate": 4.529231058231098e-05,
      "loss": 0.7521,
      "step": 50184
    },
    {
      "epoch": 4.317732082939,
      "grad_norm": 0.9375,
      "learning_rate": 4.528106530255077e-05,
      "loss": 0.7968,
      "step": 50185
    },
    {
      "epoch": 4.317818119246322,
      "grad_norm": 1.1484375,
      "learning_rate": 4.526982135275498e-05,
      "loss": 0.7028,
      "step": 50186
    },
    {
      "epoch": 4.317904155553643,
      "grad_norm": 0.96484375,
      "learning_rate": 4.525857873295669e-05,
      "loss": 1.3258,
      "step": 50187
    },
    {
      "epoch": 4.317990191860965,
      "grad_norm": 2.640625,
      "learning_rate": 4.524733744318843e-05,
      "loss": 0.814,
      "step": 50188
    },
    {
      "epoch": 4.318076228168287,
      "grad_norm": 0.921875,
      "learning_rate": 4.523609748348334e-05,
      "loss": 0.9247,
      "step": 50189
    },
    {
      "epoch": 4.318162264475609,
      "grad_norm": 12.5625,
      "learning_rate": 4.5224858853874175e-05,
      "loss": 0.9312,
      "step": 50190
    },
    {
      "epoch": 4.318248300782931,
      "grad_norm": 1.0546875,
      "learning_rate": 4.5213621554393846e-05,
      "loss": 2.0657,
      "step": 50191
    },
    {
      "epoch": 4.318334337090252,
      "grad_norm": 0.96484375,
      "learning_rate": 4.5202385585075335e-05,
      "loss": 1.3295,
      "step": 50192
    },
    {
      "epoch": 4.318420373397574,
      "grad_norm": 1.359375,
      "learning_rate": 4.519115094595133e-05,
      "loss": 1.8003,
      "step": 50193
    },
    {
      "epoch": 4.318506409704895,
      "grad_norm": 1.3125,
      "learning_rate": 4.5179917637054703e-05,
      "loss": 1.7185,
      "step": 50194
    },
    {
      "epoch": 4.318592446012217,
      "grad_norm": 1.6171875,
      "learning_rate": 4.5168685658418475e-05,
      "loss": 0.7166,
      "step": 50195
    },
    {
      "epoch": 4.318678482319539,
      "grad_norm": 1.0703125,
      "learning_rate": 4.515745501007529e-05,
      "loss": 1.3965,
      "step": 50196
    },
    {
      "epoch": 4.31876451862686,
      "grad_norm": 1.234375,
      "learning_rate": 4.514622569205817e-05,
      "loss": 0.8748,
      "step": 50197
    },
    {
      "epoch": 4.3188505549341825,
      "grad_norm": 1.0234375,
      "learning_rate": 4.5134997704399774e-05,
      "loss": 0.5698,
      "step": 50198
    },
    {
      "epoch": 4.318936591241504,
      "grad_norm": 1.3359375,
      "learning_rate": 4.512377104713306e-05,
      "loss": 1.1073,
      "step": 50199
    },
    {
      "epoch": 4.319022627548826,
      "grad_norm": 0.98046875,
      "learning_rate": 4.511254572029089e-05,
      "loss": 1.0082,
      "step": 50200
    },
    {
      "epoch": 4.319108663856147,
      "grad_norm": 0.90234375,
      "learning_rate": 4.510132172390602e-05,
      "loss": 0.6876,
      "step": 50201
    },
    {
      "epoch": 4.319194700163469,
      "grad_norm": 1.859375,
      "learning_rate": 4.5090099058011316e-05,
      "loss": 1.1098,
      "step": 50202
    },
    {
      "epoch": 4.319280736470791,
      "grad_norm": 1.1953125,
      "learning_rate": 4.507887772263963e-05,
      "loss": 0.8511,
      "step": 50203
    },
    {
      "epoch": 4.319366772778112,
      "grad_norm": 1.21875,
      "learning_rate": 4.506765771782373e-05,
      "loss": 1.4095,
      "step": 50204
    },
    {
      "epoch": 4.319452809085434,
      "grad_norm": 1.03125,
      "learning_rate": 4.505643904359641e-05,
      "loss": 1.413,
      "step": 50205
    },
    {
      "epoch": 4.3195388453927555,
      "grad_norm": 1.0234375,
      "learning_rate": 4.504522169999065e-05,
      "loss": 1.1737,
      "step": 50206
    },
    {
      "epoch": 4.3196248817000775,
      "grad_norm": 0.83984375,
      "learning_rate": 4.503400568703897e-05,
      "loss": 0.5662,
      "step": 50207
    },
    {
      "epoch": 4.319710918007399,
      "grad_norm": 1.3828125,
      "learning_rate": 4.502279100477447e-05,
      "loss": 1.0669,
      "step": 50208
    },
    {
      "epoch": 4.319796954314721,
      "grad_norm": 1.6796875,
      "learning_rate": 4.5011577653229774e-05,
      "loss": 1.6598,
      "step": 50209
    },
    {
      "epoch": 4.319882990622043,
      "grad_norm": 1.265625,
      "learning_rate": 4.5000365632437646e-05,
      "loss": 1.6427,
      "step": 50210
    },
    {
      "epoch": 4.319969026929364,
      "grad_norm": 8.5625,
      "learning_rate": 4.498915494243111e-05,
      "loss": 0.7426,
      "step": 50211
    },
    {
      "epoch": 4.320055063236686,
      "grad_norm": 0.8203125,
      "learning_rate": 4.497794558324264e-05,
      "loss": 0.7157,
      "step": 50212
    },
    {
      "epoch": 4.320141099544007,
      "grad_norm": 1.109375,
      "learning_rate": 4.4966737554905215e-05,
      "loss": 1.129,
      "step": 50213
    },
    {
      "epoch": 4.320227135851329,
      "grad_norm": 2.15625,
      "learning_rate": 4.4955530857451685e-05,
      "loss": 1.0733,
      "step": 50214
    },
    {
      "epoch": 4.3203131721586505,
      "grad_norm": 2.15625,
      "learning_rate": 4.494432549091465e-05,
      "loss": 1.5333,
      "step": 50215
    },
    {
      "epoch": 4.320399208465973,
      "grad_norm": 1.4765625,
      "learning_rate": 4.493312145532696e-05,
      "loss": 0.6856,
      "step": 50216
    },
    {
      "epoch": 4.320485244773295,
      "grad_norm": 1.34375,
      "learning_rate": 4.4921918750721436e-05,
      "loss": 0.7949,
      "step": 50217
    },
    {
      "epoch": 4.320571281080616,
      "grad_norm": 1.2578125,
      "learning_rate": 4.49107173771307e-05,
      "loss": 1.0305,
      "step": 50218
    },
    {
      "epoch": 4.320657317387938,
      "grad_norm": 1.6640625,
      "learning_rate": 4.4899517334587636e-05,
      "loss": 1.4036,
      "step": 50219
    },
    {
      "epoch": 4.320743353695259,
      "grad_norm": 0.85546875,
      "learning_rate": 4.488831862312509e-05,
      "loss": 1.0908,
      "step": 50220
    },
    {
      "epoch": 4.320829390002581,
      "grad_norm": 1.265625,
      "learning_rate": 4.487712124277554e-05,
      "loss": 1.4369,
      "step": 50221
    },
    {
      "epoch": 4.320915426309902,
      "grad_norm": 3.1875,
      "learning_rate": 4.486592519357202e-05,
      "loss": 1.0327,
      "step": 50222
    },
    {
      "epoch": 4.321001462617224,
      "grad_norm": 0.92578125,
      "learning_rate": 4.4854730475547056e-05,
      "loss": 0.9579,
      "step": 50223
    },
    {
      "epoch": 4.3210874989245465,
      "grad_norm": 1.21875,
      "learning_rate": 4.484353708873351e-05,
      "loss": 1.255,
      "step": 50224
    },
    {
      "epoch": 4.321173535231868,
      "grad_norm": 1.296875,
      "learning_rate": 4.4832345033164136e-05,
      "loss": 0.5616,
      "step": 50225
    },
    {
      "epoch": 4.32125957153919,
      "grad_norm": 2.921875,
      "learning_rate": 4.4821154308871457e-05,
      "loss": 1.4641,
      "step": 50226
    },
    {
      "epoch": 4.321345607846511,
      "grad_norm": 0.890625,
      "learning_rate": 4.480996491588857e-05,
      "loss": 1.0121,
      "step": 50227
    },
    {
      "epoch": 4.321431644153833,
      "grad_norm": 0.72265625,
      "learning_rate": 4.4798776854247936e-05,
      "loss": 0.787,
      "step": 50228
    },
    {
      "epoch": 4.321517680461154,
      "grad_norm": 21.875,
      "learning_rate": 4.478759012398231e-05,
      "loss": 1.4354,
      "step": 50229
    },
    {
      "epoch": 4.321603716768476,
      "grad_norm": 0.9296875,
      "learning_rate": 4.477640472512456e-05,
      "loss": 0.7582,
      "step": 50230
    },
    {
      "epoch": 4.321689753075798,
      "grad_norm": 1.0859375,
      "learning_rate": 4.476522065770722e-05,
      "loss": 0.973,
      "step": 50231
    },
    {
      "epoch": 4.3217757893831195,
      "grad_norm": 29.0,
      "learning_rate": 4.4754037921763025e-05,
      "loss": 1.3286,
      "step": 50232
    },
    {
      "epoch": 4.3218618256904415,
      "grad_norm": 0.7421875,
      "learning_rate": 4.4742856517324845e-05,
      "loss": 0.543,
      "step": 50233
    },
    {
      "epoch": 4.321947861997763,
      "grad_norm": 0.90234375,
      "learning_rate": 4.473167644442516e-05,
      "loss": 0.9437,
      "step": 50234
    },
    {
      "epoch": 4.322033898305085,
      "grad_norm": 0.8984375,
      "learning_rate": 4.472049770309677e-05,
      "loss": 0.6947,
      "step": 50235
    },
    {
      "epoch": 4.322119934612407,
      "grad_norm": 0.71484375,
      "learning_rate": 4.470932029337249e-05,
      "loss": 0.6887,
      "step": 50236
    },
    {
      "epoch": 4.322205970919728,
      "grad_norm": 0.98046875,
      "learning_rate": 4.469814421528484e-05,
      "loss": 0.5582,
      "step": 50237
    },
    {
      "epoch": 4.32229200722705,
      "grad_norm": 1.453125,
      "learning_rate": 4.468696946886658e-05,
      "loss": 0.754,
      "step": 50238
    },
    {
      "epoch": 4.322378043534371,
      "grad_norm": 4192.0,
      "learning_rate": 4.467579605415034e-05,
      "loss": 1.4902,
      "step": 50239
    },
    {
      "epoch": 4.322464079841693,
      "grad_norm": 1.3125,
      "learning_rate": 4.4664623971168826e-05,
      "loss": 0.9337,
      "step": 50240
    },
    {
      "epoch": 4.3225501161490145,
      "grad_norm": 0.890625,
      "learning_rate": 4.4653453219954846e-05,
      "loss": 1.1219,
      "step": 50241
    },
    {
      "epoch": 4.322636152456337,
      "grad_norm": 15.875,
      "learning_rate": 4.464228380054086e-05,
      "loss": 0.6623,
      "step": 50242
    },
    {
      "epoch": 4.322722188763659,
      "grad_norm": 9.5625,
      "learning_rate": 4.463111571295958e-05,
      "loss": 0.8131,
      "step": 50243
    },
    {
      "epoch": 4.32280822507098,
      "grad_norm": 1.03125,
      "learning_rate": 4.461994895724386e-05,
      "loss": 0.7874,
      "step": 50244
    },
    {
      "epoch": 4.322894261378302,
      "grad_norm": 1.0234375,
      "learning_rate": 4.460878353342612e-05,
      "loss": 1.1741,
      "step": 50245
    },
    {
      "epoch": 4.322980297685623,
      "grad_norm": 3.6875,
      "learning_rate": 4.459761944153912e-05,
      "loss": 1.1183,
      "step": 50246
    },
    {
      "epoch": 4.323066333992945,
      "grad_norm": 1.34375,
      "learning_rate": 4.4586456681615595e-05,
      "loss": 0.7769,
      "step": 50247
    },
    {
      "epoch": 4.323152370300266,
      "grad_norm": 1.125,
      "learning_rate": 4.457529525368803e-05,
      "loss": 1.0519,
      "step": 50248
    },
    {
      "epoch": 4.323238406607588,
      "grad_norm": 1.53125,
      "learning_rate": 4.456413515778923e-05,
      "loss": 0.6004,
      "step": 50249
    },
    {
      "epoch": 4.3233244429149105,
      "grad_norm": 0.875,
      "learning_rate": 4.455297639395173e-05,
      "loss": 1.6223,
      "step": 50250
    },
    {
      "epoch": 4.323410479222232,
      "grad_norm": 1.09375,
      "learning_rate": 4.45418189622081e-05,
      "loss": 0.7724,
      "step": 50251
    },
    {
      "epoch": 4.323496515529554,
      "grad_norm": 1.34375,
      "learning_rate": 4.453066286259122e-05,
      "loss": 1.4123,
      "step": 50252
    },
    {
      "epoch": 4.323582551836875,
      "grad_norm": 1.25,
      "learning_rate": 4.451950809513344e-05,
      "loss": 1.6704,
      "step": 50253
    },
    {
      "epoch": 4.323668588144197,
      "grad_norm": 2.734375,
      "learning_rate": 4.450835465986752e-05,
      "loss": 0.6141,
      "step": 50254
    },
    {
      "epoch": 4.323754624451518,
      "grad_norm": 2.078125,
      "learning_rate": 4.449720255682621e-05,
      "loss": 1.5675,
      "step": 50255
    },
    {
      "epoch": 4.32384066075884,
      "grad_norm": 1.4453125,
      "learning_rate": 4.448605178604187e-05,
      "loss": 1.1932,
      "step": 50256
    },
    {
      "epoch": 4.323926697066162,
      "grad_norm": 372.0,
      "learning_rate": 4.4474902347547195e-05,
      "loss": 0.9725,
      "step": 50257
    },
    {
      "epoch": 4.3240127333734835,
      "grad_norm": 57.25,
      "learning_rate": 4.4463754241375e-05,
      "loss": 0.7658,
      "step": 50258
    },
    {
      "epoch": 4.3240987696808055,
      "grad_norm": 1.078125,
      "learning_rate": 4.445260746755758e-05,
      "loss": 0.762,
      "step": 50259
    },
    {
      "epoch": 4.324184805988127,
      "grad_norm": 1.15625,
      "learning_rate": 4.44414620261277e-05,
      "loss": 1.4515,
      "step": 50260
    },
    {
      "epoch": 4.324270842295449,
      "grad_norm": 1.3828125,
      "learning_rate": 4.4430317917118046e-05,
      "loss": 1.1136,
      "step": 50261
    },
    {
      "epoch": 4.32435687860277,
      "grad_norm": 2.078125,
      "learning_rate": 4.441917514056104e-05,
      "loss": 1.2011,
      "step": 50262
    },
    {
      "epoch": 4.324442914910092,
      "grad_norm": 0.8984375,
      "learning_rate": 4.440803369648927e-05,
      "loss": 0.5365,
      "step": 50263
    },
    {
      "epoch": 4.324528951217414,
      "grad_norm": 1.1953125,
      "learning_rate": 4.439689358493543e-05,
      "loss": 0.9129,
      "step": 50264
    },
    {
      "epoch": 4.324614987524735,
      "grad_norm": 0.875,
      "learning_rate": 4.43857548059321e-05,
      "loss": 0.6301,
      "step": 50265
    },
    {
      "epoch": 4.324701023832057,
      "grad_norm": 1.84375,
      "learning_rate": 4.437461735951187e-05,
      "loss": 0.8121,
      "step": 50266
    },
    {
      "epoch": 4.3247870601393785,
      "grad_norm": 0.7421875,
      "learning_rate": 4.436348124570721e-05,
      "loss": 0.5386,
      "step": 50267
    },
    {
      "epoch": 4.324873096446701,
      "grad_norm": 1.3125,
      "learning_rate": 4.435234646455083e-05,
      "loss": 1.2632,
      "step": 50268
    },
    {
      "epoch": 4.324959132754023,
      "grad_norm": 1.125,
      "learning_rate": 4.434121301607513e-05,
      "loss": 1.1231,
      "step": 50269
    },
    {
      "epoch": 4.325045169061344,
      "grad_norm": 0.92578125,
      "learning_rate": 4.4330080900312766e-05,
      "loss": 0.8174,
      "step": 50270
    },
    {
      "epoch": 4.325131205368666,
      "grad_norm": 1.859375,
      "learning_rate": 4.431895011729636e-05,
      "loss": 0.7487,
      "step": 50271
    },
    {
      "epoch": 4.325217241675987,
      "grad_norm": 3.046875,
      "learning_rate": 4.43078206670583e-05,
      "loss": 1.8696,
      "step": 50272
    },
    {
      "epoch": 4.325303277983309,
      "grad_norm": 6.28125,
      "learning_rate": 4.4296692549631254e-05,
      "loss": 1.0623,
      "step": 50273
    },
    {
      "epoch": 4.32538931429063,
      "grad_norm": 11.25,
      "learning_rate": 4.428556576504789e-05,
      "loss": 1.4746,
      "step": 50274
    },
    {
      "epoch": 4.325475350597952,
      "grad_norm": 0.9296875,
      "learning_rate": 4.427444031334049e-05,
      "loss": 1.309,
      "step": 50275
    },
    {
      "epoch": 4.3255613869052745,
      "grad_norm": 0.97265625,
      "learning_rate": 4.4263316194541713e-05,
      "loss": 0.8866,
      "step": 50276
    },
    {
      "epoch": 4.325647423212596,
      "grad_norm": 0.9609375,
      "learning_rate": 4.425219340868414e-05,
      "loss": 0.9343,
      "step": 50277
    },
    {
      "epoch": 4.325733459519918,
      "grad_norm": 1.203125,
      "learning_rate": 4.424107195580024e-05,
      "loss": 0.9963,
      "step": 50278
    },
    {
      "epoch": 4.325819495827239,
      "grad_norm": 1.3046875,
      "learning_rate": 4.4229951835922654e-05,
      "loss": 1.5295,
      "step": 50279
    },
    {
      "epoch": 4.325905532134561,
      "grad_norm": 1.0,
      "learning_rate": 4.42188330490837e-05,
      "loss": 0.8246,
      "step": 50280
    },
    {
      "epoch": 4.325991568441882,
      "grad_norm": 1.046875,
      "learning_rate": 4.4207715595316e-05,
      "loss": 0.6624,
      "step": 50281
    },
    {
      "epoch": 4.326077604749204,
      "grad_norm": 1.5625,
      "learning_rate": 4.419659947465221e-05,
      "loss": 1.3309,
      "step": 50282
    },
    {
      "epoch": 4.326163641056526,
      "grad_norm": 1.171875,
      "learning_rate": 4.4185484687124635e-05,
      "loss": 1.2787,
      "step": 50283
    },
    {
      "epoch": 4.3262496773638475,
      "grad_norm": 2880.0,
      "learning_rate": 4.4174371232765856e-05,
      "loss": 1.053,
      "step": 50284
    },
    {
      "epoch": 4.3263357136711695,
      "grad_norm": 0.87109375,
      "learning_rate": 4.4163259111608455e-05,
      "loss": 0.7412,
      "step": 50285
    },
    {
      "epoch": 4.326421749978491,
      "grad_norm": 1.59375,
      "learning_rate": 4.41521483236848e-05,
      "loss": 0.7066,
      "step": 50286
    },
    {
      "epoch": 4.326507786285813,
      "grad_norm": 1.0625,
      "learning_rate": 4.4141038869027425e-05,
      "loss": 1.0404,
      "step": 50287
    },
    {
      "epoch": 4.326593822593134,
      "grad_norm": 0.73828125,
      "learning_rate": 4.412993074766897e-05,
      "loss": 0.4594,
      "step": 50288
    },
    {
      "epoch": 4.326679858900456,
      "grad_norm": 0.984375,
      "learning_rate": 4.411882395964173e-05,
      "loss": 0.8741,
      "step": 50289
    },
    {
      "epoch": 4.326765895207778,
      "grad_norm": 45.5,
      "learning_rate": 4.41077185049783e-05,
      "loss": 1.2985,
      "step": 50290
    },
    {
      "epoch": 4.326851931515099,
      "grad_norm": 0.6796875,
      "learning_rate": 4.40966143837111e-05,
      "loss": 0.5627,
      "step": 50291
    },
    {
      "epoch": 4.326937967822421,
      "grad_norm": 1.1875,
      "learning_rate": 4.4085511595872595e-05,
      "loss": 0.8676,
      "step": 50292
    },
    {
      "epoch": 4.3270240041297425,
      "grad_norm": 1.84375,
      "learning_rate": 4.4074410141495383e-05,
      "loss": 1.1454,
      "step": 50293
    },
    {
      "epoch": 4.327110040437065,
      "grad_norm": 16.625,
      "learning_rate": 4.406331002061181e-05,
      "loss": 0.7638,
      "step": 50294
    },
    {
      "epoch": 4.327196076744386,
      "grad_norm": 0.9765625,
      "learning_rate": 4.4052211233254315e-05,
      "loss": 0.7369,
      "step": 50295
    },
    {
      "epoch": 4.327282113051708,
      "grad_norm": 1.3671875,
      "learning_rate": 4.404111377945558e-05,
      "loss": 1.303,
      "step": 50296
    },
    {
      "epoch": 4.32736814935903,
      "grad_norm": 5.40625,
      "learning_rate": 4.4030017659247745e-05,
      "loss": 0.7961,
      "step": 50297
    },
    {
      "epoch": 4.327454185666351,
      "grad_norm": 107.5,
      "learning_rate": 4.401892287266346e-05,
      "loss": 0.7831,
      "step": 50298
    },
    {
      "epoch": 4.327540221973673,
      "grad_norm": 1.1875,
      "learning_rate": 4.4007829419735246e-05,
      "loss": 1.2436,
      "step": 50299
    },
    {
      "epoch": 4.327626258280994,
      "grad_norm": 10.4375,
      "learning_rate": 4.399673730049525e-05,
      "loss": 1.3545,
      "step": 50300
    },
    {
      "epoch": 4.327712294588316,
      "grad_norm": 1.0859375,
      "learning_rate": 4.398564651497633e-05,
      "loss": 1.1844,
      "step": 50301
    },
    {
      "epoch": 4.327798330895638,
      "grad_norm": 0.9609375,
      "learning_rate": 4.397455706321057e-05,
      "loss": 0.9115,
      "step": 50302
    },
    {
      "epoch": 4.32788436720296,
      "grad_norm": 1.53125,
      "learning_rate": 4.396346894523051e-05,
      "loss": 0.9361,
      "step": 50303
    },
    {
      "epoch": 4.327970403510282,
      "grad_norm": 22.75,
      "learning_rate": 4.3952382161068784e-05,
      "loss": 1.1278,
      "step": 50304
    },
    {
      "epoch": 4.328056439817603,
      "grad_norm": 0.98046875,
      "learning_rate": 4.394129671075747e-05,
      "loss": 1.4495,
      "step": 50305
    },
    {
      "epoch": 4.328142476124925,
      "grad_norm": 244.0,
      "learning_rate": 4.393021259432922e-05,
      "loss": 1.0973,
      "step": 50306
    },
    {
      "epoch": 4.328228512432246,
      "grad_norm": 70.5,
      "learning_rate": 4.391912981181645e-05,
      "loss": 1.122,
      "step": 50307
    },
    {
      "epoch": 4.328314548739568,
      "grad_norm": 1.0078125,
      "learning_rate": 4.390804836325146e-05,
      "loss": 1.2177,
      "step": 50308
    },
    {
      "epoch": 4.328400585046889,
      "grad_norm": 9.75,
      "learning_rate": 4.389696824866679e-05,
      "loss": 1.1345,
      "step": 50309
    },
    {
      "epoch": 4.3284866213542115,
      "grad_norm": 0.8671875,
      "learning_rate": 4.388588946809474e-05,
      "loss": 0.6499,
      "step": 50310
    },
    {
      "epoch": 4.3285726576615335,
      "grad_norm": 1.2890625,
      "learning_rate": 4.387481202156773e-05,
      "loss": 0.9798,
      "step": 50311
    },
    {
      "epoch": 4.328658693968855,
      "grad_norm": 1.046875,
      "learning_rate": 4.386373590911824e-05,
      "loss": 0.5277,
      "step": 50312
    },
    {
      "epoch": 4.328744730276177,
      "grad_norm": 0.8828125,
      "learning_rate": 4.385266113077857e-05,
      "loss": 1.024,
      "step": 50313
    },
    {
      "epoch": 4.328830766583498,
      "grad_norm": 1.0390625,
      "learning_rate": 4.384158768658108e-05,
      "loss": 0.4976,
      "step": 50314
    },
    {
      "epoch": 4.32891680289082,
      "grad_norm": 1.4296875,
      "learning_rate": 4.3830515576558374e-05,
      "loss": 2.0264,
      "step": 50315
    },
    {
      "epoch": 4.329002839198141,
      "grad_norm": 1.015625,
      "learning_rate": 4.381944480074257e-05,
      "loss": 0.9199,
      "step": 50316
    },
    {
      "epoch": 4.329088875505463,
      "grad_norm": 1.015625,
      "learning_rate": 4.3808375359166205e-05,
      "loss": 0.9737,
      "step": 50317
    },
    {
      "epoch": 4.329174911812785,
      "grad_norm": 0.94921875,
      "learning_rate": 4.3797307251861704e-05,
      "loss": 0.6284,
      "step": 50318
    },
    {
      "epoch": 4.3292609481201065,
      "grad_norm": 11.125,
      "learning_rate": 4.378624047886126e-05,
      "loss": 0.4772,
      "step": 50319
    },
    {
      "epoch": 4.329346984427429,
      "grad_norm": 2.03125,
      "learning_rate": 4.3775175040197404e-05,
      "loss": 0.9851,
      "step": 50320
    },
    {
      "epoch": 4.32943302073475,
      "grad_norm": 1.09375,
      "learning_rate": 4.376411093590233e-05,
      "loss": 0.8062,
      "step": 50321
    },
    {
      "epoch": 4.329519057042072,
      "grad_norm": 1.375,
      "learning_rate": 4.37530481660085e-05,
      "loss": 0.9445,
      "step": 50322
    },
    {
      "epoch": 4.329605093349393,
      "grad_norm": 1.09375,
      "learning_rate": 4.374198673054841e-05,
      "loss": 1.3116,
      "step": 50323
    },
    {
      "epoch": 4.329691129656715,
      "grad_norm": 1.5078125,
      "learning_rate": 4.373092662955414e-05,
      "loss": 1.2027,
      "step": 50324
    },
    {
      "epoch": 4.329777165964037,
      "grad_norm": 2.0,
      "learning_rate": 4.371986786305815e-05,
      "loss": 0.9822,
      "step": 50325
    },
    {
      "epoch": 4.329863202271358,
      "grad_norm": 0.890625,
      "learning_rate": 4.3708810431092924e-05,
      "loss": 0.3609,
      "step": 50326
    },
    {
      "epoch": 4.32994923857868,
      "grad_norm": 0.94921875,
      "learning_rate": 4.369775433369061e-05,
      "loss": 1.3847,
      "step": 50327
    },
    {
      "epoch": 4.330035274886002,
      "grad_norm": 1.078125,
      "learning_rate": 4.368669957088356e-05,
      "loss": 1.5475,
      "step": 50328
    },
    {
      "epoch": 4.330121311193324,
      "grad_norm": 1.1328125,
      "learning_rate": 4.367564614270431e-05,
      "loss": 1.0978,
      "step": 50329
    },
    {
      "epoch": 4.330207347500645,
      "grad_norm": 1.3125,
      "learning_rate": 4.366459404918494e-05,
      "loss": 1.0665,
      "step": 50330
    },
    {
      "epoch": 4.330293383807967,
      "grad_norm": 1.578125,
      "learning_rate": 4.365354329035787e-05,
      "loss": 1.2587,
      "step": 50331
    },
    {
      "epoch": 4.330379420115289,
      "grad_norm": 0.74609375,
      "learning_rate": 4.3642493866255576e-05,
      "loss": 1.1186,
      "step": 50332
    },
    {
      "epoch": 4.33046545642261,
      "grad_norm": 1.5390625,
      "learning_rate": 4.3631445776910085e-05,
      "loss": 1.4353,
      "step": 50333
    },
    {
      "epoch": 4.330551492729932,
      "grad_norm": 0.87109375,
      "learning_rate": 4.362039902235393e-05,
      "loss": 0.9969,
      "step": 50334
    },
    {
      "epoch": 4.330637529037253,
      "grad_norm": 1.0234375,
      "learning_rate": 4.360935360261931e-05,
      "loss": 1.5142,
      "step": 50335
    },
    {
      "epoch": 4.3307235653445755,
      "grad_norm": 1.8515625,
      "learning_rate": 4.359830951773852e-05,
      "loss": 1.1471,
      "step": 50336
    },
    {
      "epoch": 4.330809601651897,
      "grad_norm": 1.1171875,
      "learning_rate": 4.358726676774394e-05,
      "loss": 1.0987,
      "step": 50337
    },
    {
      "epoch": 4.330895637959219,
      "grad_norm": 1.4765625,
      "learning_rate": 4.357622535266781e-05,
      "loss": 1.2649,
      "step": 50338
    },
    {
      "epoch": 4.330981674266541,
      "grad_norm": 0.84375,
      "learning_rate": 4.3565185272542554e-05,
      "loss": 1.2128,
      "step": 50339
    },
    {
      "epoch": 4.331067710573862,
      "grad_norm": 1.3125,
      "learning_rate": 4.3554146527400304e-05,
      "loss": 1.3119,
      "step": 50340
    },
    {
      "epoch": 4.331153746881184,
      "grad_norm": 2768.0,
      "learning_rate": 4.354310911727338e-05,
      "loss": 1.6222,
      "step": 50341
    },
    {
      "epoch": 4.331239783188505,
      "grad_norm": 1.2734375,
      "learning_rate": 4.353207304219414e-05,
      "loss": 2.194,
      "step": 50342
    },
    {
      "epoch": 4.331325819495827,
      "grad_norm": 0.90234375,
      "learning_rate": 4.352103830219478e-05,
      "loss": 0.8016,
      "step": 50343
    },
    {
      "epoch": 4.3314118558031485,
      "grad_norm": 0.94140625,
      "learning_rate": 4.351000489730755e-05,
      "loss": 0.9349,
      "step": 50344
    },
    {
      "epoch": 4.3314978921104705,
      "grad_norm": 0.86328125,
      "learning_rate": 4.349897282756487e-05,
      "loss": 0.9629,
      "step": 50345
    },
    {
      "epoch": 4.331583928417793,
      "grad_norm": 0.88671875,
      "learning_rate": 4.348794209299883e-05,
      "loss": 1.2706,
      "step": 50346
    },
    {
      "epoch": 4.331669964725114,
      "grad_norm": 16.75,
      "learning_rate": 4.347691269364179e-05,
      "loss": 1.6455,
      "step": 50347
    },
    {
      "epoch": 4.331756001032436,
      "grad_norm": 1.1328125,
      "learning_rate": 4.346588462952606e-05,
      "loss": 0.9225,
      "step": 50348
    },
    {
      "epoch": 4.331842037339757,
      "grad_norm": 115.0,
      "learning_rate": 4.345485790068376e-05,
      "loss": 1.3095,
      "step": 50349
    },
    {
      "epoch": 4.331928073647079,
      "grad_norm": 1.2734375,
      "learning_rate": 4.344383250714728e-05,
      "loss": 1.7046,
      "step": 50350
    },
    {
      "epoch": 4.332014109954401,
      "grad_norm": 2.109375,
      "learning_rate": 4.34328084489487e-05,
      "loss": 1.6096,
      "step": 50351
    },
    {
      "epoch": 4.332100146261722,
      "grad_norm": 1.203125,
      "learning_rate": 4.342178572612038e-05,
      "loss": 1.1881,
      "step": 50352
    },
    {
      "epoch": 4.332186182569044,
      "grad_norm": 1.5625,
      "learning_rate": 4.341076433869462e-05,
      "loss": 1.1309,
      "step": 50353
    },
    {
      "epoch": 4.332272218876366,
      "grad_norm": 1.09375,
      "learning_rate": 4.339974428670346e-05,
      "loss": 0.8869,
      "step": 50354
    },
    {
      "epoch": 4.332358255183688,
      "grad_norm": 0.890625,
      "learning_rate": 4.3388725570179255e-05,
      "loss": 0.7466,
      "step": 50355
    },
    {
      "epoch": 4.332444291491009,
      "grad_norm": 1.2265625,
      "learning_rate": 4.337770818915432e-05,
      "loss": 1.3704,
      "step": 50356
    },
    {
      "epoch": 4.332530327798331,
      "grad_norm": 1.296875,
      "learning_rate": 4.336669214366068e-05,
      "loss": 0.9389,
      "step": 50357
    },
    {
      "epoch": 4.332616364105653,
      "grad_norm": 1.0625,
      "learning_rate": 4.335567743373065e-05,
      "loss": 0.9619,
      "step": 50358
    },
    {
      "epoch": 4.332702400412974,
      "grad_norm": 0.7265625,
      "learning_rate": 4.334466405939658e-05,
      "loss": 0.6163,
      "step": 50359
    },
    {
      "epoch": 4.332788436720296,
      "grad_norm": 1.0078125,
      "learning_rate": 4.3333652020690396e-05,
      "loss": 0.9339,
      "step": 50360
    },
    {
      "epoch": 4.332874473027617,
      "grad_norm": 112.0,
      "learning_rate": 4.332264131764463e-05,
      "loss": 1.9694,
      "step": 50361
    },
    {
      "epoch": 4.3329605093349395,
      "grad_norm": 1.2734375,
      "learning_rate": 4.3311631950291195e-05,
      "loss": 1.9198,
      "step": 50362
    },
    {
      "epoch": 4.333046545642261,
      "grad_norm": 0.72265625,
      "learning_rate": 4.330062391866241e-05,
      "loss": 0.4093,
      "step": 50363
    },
    {
      "epoch": 4.333132581949583,
      "grad_norm": 1.1015625,
      "learning_rate": 4.328961722279057e-05,
      "loss": 0.9093,
      "step": 50364
    },
    {
      "epoch": 4.333218618256905,
      "grad_norm": 0.98046875,
      "learning_rate": 4.327861186270765e-05,
      "loss": 0.6203,
      "step": 50365
    },
    {
      "epoch": 4.333304654564226,
      "grad_norm": 20.25,
      "learning_rate": 4.3267607838446023e-05,
      "loss": 0.6158,
      "step": 50366
    },
    {
      "epoch": 4.333390690871548,
      "grad_norm": 1.375,
      "learning_rate": 4.3256605150037876e-05,
      "loss": 0.6125,
      "step": 50367
    },
    {
      "epoch": 4.333476727178869,
      "grad_norm": 0.94140625,
      "learning_rate": 4.3245603797515245e-05,
      "loss": 0.6182,
      "step": 50368
    },
    {
      "epoch": 4.333562763486191,
      "grad_norm": 1.34375,
      "learning_rate": 4.323460378091037e-05,
      "loss": 0.943,
      "step": 50369
    },
    {
      "epoch": 4.3336487997935125,
      "grad_norm": 1.0625,
      "learning_rate": 4.322360510025552e-05,
      "loss": 0.8821,
      "step": 50370
    },
    {
      "epoch": 4.3337348361008345,
      "grad_norm": 1.0625,
      "learning_rate": 4.321260775558272e-05,
      "loss": 0.9334,
      "step": 50371
    },
    {
      "epoch": 4.333820872408157,
      "grad_norm": 0.9609375,
      "learning_rate": 4.320161174692422e-05,
      "loss": 0.6715,
      "step": 50372
    },
    {
      "epoch": 4.333906908715478,
      "grad_norm": 1.046875,
      "learning_rate": 4.319061707431221e-05,
      "loss": 1.1392,
      "step": 50373
    },
    {
      "epoch": 4.3339929450228,
      "grad_norm": 47.25,
      "learning_rate": 4.317962373777867e-05,
      "loss": 0.7867,
      "step": 50374
    },
    {
      "epoch": 4.334078981330121,
      "grad_norm": 1.203125,
      "learning_rate": 4.316863173735602e-05,
      "loss": 1.4907,
      "step": 50375
    },
    {
      "epoch": 4.334165017637443,
      "grad_norm": 360.0,
      "learning_rate": 4.3157641073076237e-05,
      "loss": 0.8324,
      "step": 50376
    },
    {
      "epoch": 4.334251053944764,
      "grad_norm": 32.0,
      "learning_rate": 4.3146651744971566e-05,
      "loss": 0.8696,
      "step": 50377
    },
    {
      "epoch": 4.334337090252086,
      "grad_norm": 0.96484375,
      "learning_rate": 4.313566375307398e-05,
      "loss": 1.0619,
      "step": 50378
    },
    {
      "epoch": 4.334423126559408,
      "grad_norm": 26.875,
      "learning_rate": 4.312467709741574e-05,
      "loss": 0.8575,
      "step": 50379
    },
    {
      "epoch": 4.33450916286673,
      "grad_norm": 1.3359375,
      "learning_rate": 4.311369177802904e-05,
      "loss": 0.9527,
      "step": 50380
    },
    {
      "epoch": 4.334595199174052,
      "grad_norm": 1.21875,
      "learning_rate": 4.310270779494585e-05,
      "loss": 0.9298,
      "step": 50381
    },
    {
      "epoch": 4.334681235481373,
      "grad_norm": 1.2109375,
      "learning_rate": 4.309172514819837e-05,
      "loss": 0.6913,
      "step": 50382
    },
    {
      "epoch": 4.334767271788695,
      "grad_norm": 1.3671875,
      "learning_rate": 4.308074383781885e-05,
      "loss": 1.1491,
      "step": 50383
    },
    {
      "epoch": 4.334853308096017,
      "grad_norm": 1.1015625,
      "learning_rate": 4.306976386383915e-05,
      "loss": 1.3279,
      "step": 50384
    },
    {
      "epoch": 4.334939344403338,
      "grad_norm": 1.2265625,
      "learning_rate": 4.305878522629153e-05,
      "loss": 0.6233,
      "step": 50385
    },
    {
      "epoch": 4.33502538071066,
      "grad_norm": 34.0,
      "learning_rate": 4.304780792520824e-05,
      "loss": 0.6277,
      "step": 50386
    },
    {
      "epoch": 4.335111417017981,
      "grad_norm": 1.2890625,
      "learning_rate": 4.303683196062108e-05,
      "loss": 1.5028,
      "step": 50387
    },
    {
      "epoch": 4.3351974533253035,
      "grad_norm": 1.2890625,
      "learning_rate": 4.302585733256237e-05,
      "loss": 1.2507,
      "step": 50388
    },
    {
      "epoch": 4.335283489632625,
      "grad_norm": 692.0,
      "learning_rate": 4.301488404106418e-05,
      "loss": 1.34,
      "step": 50389
    },
    {
      "epoch": 4.335369525939947,
      "grad_norm": 1.4375,
      "learning_rate": 4.30039120861585e-05,
      "loss": 1.1698,
      "step": 50390
    },
    {
      "epoch": 4.335455562247269,
      "grad_norm": 0.99609375,
      "learning_rate": 4.2992941467877634e-05,
      "loss": 0.7973,
      "step": 50391
    },
    {
      "epoch": 4.33554159855459,
      "grad_norm": 0.75390625,
      "learning_rate": 4.298197218625338e-05,
      "loss": 0.5933,
      "step": 50392
    },
    {
      "epoch": 4.335627634861912,
      "grad_norm": 3.03125,
      "learning_rate": 4.2971004241318e-05,
      "loss": 0.5382,
      "step": 50393
    },
    {
      "epoch": 4.335713671169233,
      "grad_norm": 1.390625,
      "learning_rate": 4.296003763310369e-05,
      "loss": 0.9197,
      "step": 50394
    },
    {
      "epoch": 4.335799707476555,
      "grad_norm": 1.3515625,
      "learning_rate": 4.2949072361642196e-05,
      "loss": 1.5053,
      "step": 50395
    },
    {
      "epoch": 4.3358857437838765,
      "grad_norm": 4.09375,
      "learning_rate": 4.2938108426965826e-05,
      "loss": 1.2655,
      "step": 50396
    },
    {
      "epoch": 4.3359717800911985,
      "grad_norm": 1.1171875,
      "learning_rate": 4.2927145829106675e-05,
      "loss": 0.8524,
      "step": 50397
    },
    {
      "epoch": 4.336057816398521,
      "grad_norm": 1.2109375,
      "learning_rate": 4.291618456809665e-05,
      "loss": 1.6078,
      "step": 50398
    },
    {
      "epoch": 4.336143852705842,
      "grad_norm": 0.7265625,
      "learning_rate": 4.2905224643967846e-05,
      "loss": 0.6107,
      "step": 50399
    },
    {
      "epoch": 4.336229889013164,
      "grad_norm": 1.3515625,
      "learning_rate": 4.289426605675251e-05,
      "loss": 1.795,
      "step": 50400
    },
    {
      "epoch": 4.336315925320485,
      "grad_norm": 1.2734375,
      "learning_rate": 4.288330880648239e-05,
      "loss": 1.3736,
      "step": 50401
    },
    {
      "epoch": 4.336401961627807,
      "grad_norm": 1.25,
      "learning_rate": 4.287235289318986e-05,
      "loss": 1.3162,
      "step": 50402
    },
    {
      "epoch": 4.336487997935128,
      "grad_norm": 4.46875,
      "learning_rate": 4.286139831690661e-05,
      "loss": 0.8979,
      "step": 50403
    },
    {
      "epoch": 4.33657403424245,
      "grad_norm": 22.125,
      "learning_rate": 4.285044507766495e-05,
      "loss": 1.3393,
      "step": 50404
    },
    {
      "epoch": 4.336660070549772,
      "grad_norm": 245.0,
      "learning_rate": 4.283949317549685e-05,
      "loss": 1.336,
      "step": 50405
    },
    {
      "epoch": 4.336746106857094,
      "grad_norm": 1.0625,
      "learning_rate": 4.282854261043423e-05,
      "loss": 0.4796,
      "step": 50406
    },
    {
      "epoch": 4.336832143164416,
      "grad_norm": 1.1953125,
      "learning_rate": 4.281759338250923e-05,
      "loss": 1.2533,
      "step": 50407
    },
    {
      "epoch": 4.336918179471737,
      "grad_norm": 7.65625,
      "learning_rate": 4.280664549175395e-05,
      "loss": 0.878,
      "step": 50408
    },
    {
      "epoch": 4.337004215779059,
      "grad_norm": 1.34375,
      "learning_rate": 4.279569893820023e-05,
      "loss": 1.9432,
      "step": 50409
    },
    {
      "epoch": 4.33709025208638,
      "grad_norm": 1.5546875,
      "learning_rate": 4.278475372188012e-05,
      "loss": 1.2885,
      "step": 50410
    },
    {
      "epoch": 4.337176288393702,
      "grad_norm": 29.125,
      "learning_rate": 4.277380984282575e-05,
      "loss": 0.7238,
      "step": 50411
    },
    {
      "epoch": 4.337262324701024,
      "grad_norm": 1.1015625,
      "learning_rate": 4.2762867301068984e-05,
      "loss": 1.008,
      "step": 50412
    },
    {
      "epoch": 4.337348361008345,
      "grad_norm": 1.5234375,
      "learning_rate": 4.275192609664208e-05,
      "loss": 1.4215,
      "step": 50413
    },
    {
      "epoch": 4.3374343973156675,
      "grad_norm": 0.87109375,
      "learning_rate": 4.274098622957673e-05,
      "loss": 0.7518,
      "step": 50414
    },
    {
      "epoch": 4.337520433622989,
      "grad_norm": 1.3359375,
      "learning_rate": 4.273004769990502e-05,
      "loss": 1.3006,
      "step": 50415
    },
    {
      "epoch": 4.337606469930311,
      "grad_norm": 1.0546875,
      "learning_rate": 4.2719110507659144e-05,
      "loss": 0.9702,
      "step": 50416
    },
    {
      "epoch": 4.337692506237632,
      "grad_norm": 0.92578125,
      "learning_rate": 4.2708174652870805e-05,
      "loss": 1.5074,
      "step": 50417
    },
    {
      "epoch": 4.337778542544954,
      "grad_norm": 1.0234375,
      "learning_rate": 4.2697240135572134e-05,
      "loss": 0.9082,
      "step": 50418
    },
    {
      "epoch": 4.337864578852276,
      "grad_norm": 0.89453125,
      "learning_rate": 4.268630695579517e-05,
      "loss": 0.7268,
      "step": 50419
    },
    {
      "epoch": 4.337950615159597,
      "grad_norm": 608.0,
      "learning_rate": 4.267537511357172e-05,
      "loss": 0.9304,
      "step": 50420
    },
    {
      "epoch": 4.338036651466919,
      "grad_norm": 0.953125,
      "learning_rate": 4.266444460893398e-05,
      "loss": 0.8561,
      "step": 50421
    },
    {
      "epoch": 4.3381226877742405,
      "grad_norm": 1.0703125,
      "learning_rate": 4.265351544191365e-05,
      "loss": 0.7157,
      "step": 50422
    },
    {
      "epoch": 4.3382087240815626,
      "grad_norm": 1.203125,
      "learning_rate": 4.2642587612542916e-05,
      "loss": 1.2802,
      "step": 50423
    },
    {
      "epoch": 4.338294760388884,
      "grad_norm": 1.296875,
      "learning_rate": 4.2631661120853706e-05,
      "loss": 1.1118,
      "step": 50424
    },
    {
      "epoch": 4.338380796696206,
      "grad_norm": 0.9921875,
      "learning_rate": 4.262073596687782e-05,
      "loss": 1.1315,
      "step": 50425
    },
    {
      "epoch": 4.338466833003528,
      "grad_norm": 0.98828125,
      "learning_rate": 4.260981215064735e-05,
      "loss": 0.7426,
      "step": 50426
    },
    {
      "epoch": 4.338552869310849,
      "grad_norm": 1.1640625,
      "learning_rate": 4.259888967219433e-05,
      "loss": 1.051,
      "step": 50427
    },
    {
      "epoch": 4.338638905618171,
      "grad_norm": 1.5234375,
      "learning_rate": 4.258796853155045e-05,
      "loss": 0.7662,
      "step": 50428
    },
    {
      "epoch": 4.338724941925492,
      "grad_norm": 1.171875,
      "learning_rate": 4.2577048728747846e-05,
      "loss": 1.0608,
      "step": 50429
    },
    {
      "epoch": 4.338810978232814,
      "grad_norm": 1.6015625,
      "learning_rate": 4.25661302638185e-05,
      "loss": 0.9246,
      "step": 50430
    },
    {
      "epoch": 4.3388970145401355,
      "grad_norm": 1.1640625,
      "learning_rate": 4.255521313679417e-05,
      "loss": 0.5631,
      "step": 50431
    },
    {
      "epoch": 4.338983050847458,
      "grad_norm": 1.0703125,
      "learning_rate": 4.254429734770693e-05,
      "loss": 1.4802,
      "step": 50432
    },
    {
      "epoch": 4.33906908715478,
      "grad_norm": 1.4375,
      "learning_rate": 4.253338289658859e-05,
      "loss": 1.1309,
      "step": 50433
    },
    {
      "epoch": 4.339155123462101,
      "grad_norm": 1.015625,
      "learning_rate": 4.252246978347113e-05,
      "loss": 0.9109,
      "step": 50434
    },
    {
      "epoch": 4.339241159769423,
      "grad_norm": 1.3046875,
      "learning_rate": 4.2511558008386575e-05,
      "loss": 1.115,
      "step": 50435
    },
    {
      "epoch": 4.339327196076744,
      "grad_norm": 34.0,
      "learning_rate": 4.250064757136662e-05,
      "loss": 1.0179,
      "step": 50436
    },
    {
      "epoch": 4.339413232384066,
      "grad_norm": 1.1640625,
      "learning_rate": 4.2489738472443304e-05,
      "loss": 1.3702,
      "step": 50437
    },
    {
      "epoch": 4.339499268691387,
      "grad_norm": 1.1171875,
      "learning_rate": 4.2478830711648594e-05,
      "loss": 1.408,
      "step": 50438
    },
    {
      "epoch": 4.339585304998709,
      "grad_norm": 1.1796875,
      "learning_rate": 4.246792428901425e-05,
      "loss": 1.6172,
      "step": 50439
    },
    {
      "epoch": 4.3396713413060315,
      "grad_norm": 0.95703125,
      "learning_rate": 4.2457019204572235e-05,
      "loss": 0.658,
      "step": 50440
    },
    {
      "epoch": 4.339757377613353,
      "grad_norm": 2.90625,
      "learning_rate": 4.244611545835458e-05,
      "loss": 0.8654,
      "step": 50441
    },
    {
      "epoch": 4.339843413920675,
      "grad_norm": 0.890625,
      "learning_rate": 4.243521305039288e-05,
      "loss": 1.043,
      "step": 50442
    },
    {
      "epoch": 4.339929450227996,
      "grad_norm": 0.91015625,
      "learning_rate": 4.2424311980719334e-05,
      "loss": 1.1933,
      "step": 50443
    },
    {
      "epoch": 4.340015486535318,
      "grad_norm": 6.125,
      "learning_rate": 4.241341224936562e-05,
      "loss": 1.2575,
      "step": 50444
    },
    {
      "epoch": 4.340101522842639,
      "grad_norm": 1.375,
      "learning_rate": 4.240251385636362e-05,
      "loss": 1.0803,
      "step": 50445
    },
    {
      "epoch": 4.340187559149961,
      "grad_norm": 0.87890625,
      "learning_rate": 4.239161680174536e-05,
      "loss": 0.4441,
      "step": 50446
    },
    {
      "epoch": 4.340273595457283,
      "grad_norm": 556.0,
      "learning_rate": 4.2380721085542586e-05,
      "loss": 1.4682,
      "step": 50447
    },
    {
      "epoch": 4.3403596317646045,
      "grad_norm": 1.2421875,
      "learning_rate": 4.236982670778716e-05,
      "loss": 1.1244,
      "step": 50448
    },
    {
      "epoch": 4.3404456680719266,
      "grad_norm": 0.9453125,
      "learning_rate": 4.235893366851101e-05,
      "loss": 0.8597,
      "step": 50449
    },
    {
      "epoch": 4.340531704379248,
      "grad_norm": 0.81640625,
      "learning_rate": 4.2348041967745944e-05,
      "loss": 0.9259,
      "step": 50450
    },
    {
      "epoch": 4.34061774068657,
      "grad_norm": 1.1171875,
      "learning_rate": 4.233715160552393e-05,
      "loss": 1.0578,
      "step": 50451
    },
    {
      "epoch": 4.340703776993891,
      "grad_norm": 1.1015625,
      "learning_rate": 4.232626258187666e-05,
      "loss": 1.1293,
      "step": 50452
    },
    {
      "epoch": 4.340789813301213,
      "grad_norm": 1.0078125,
      "learning_rate": 4.2315374896836066e-05,
      "loss": 1.4093,
      "step": 50453
    },
    {
      "epoch": 4.340875849608535,
      "grad_norm": 1.0234375,
      "learning_rate": 4.230448855043412e-05,
      "loss": 0.5939,
      "step": 50454
    },
    {
      "epoch": 4.340961885915856,
      "grad_norm": 1.125,
      "learning_rate": 4.22936035427024e-05,
      "loss": 0.733,
      "step": 50455
    },
    {
      "epoch": 4.341047922223178,
      "grad_norm": 1.1015625,
      "learning_rate": 4.2282719873672824e-05,
      "loss": 1.1132,
      "step": 50456
    },
    {
      "epoch": 4.3411339585304995,
      "grad_norm": 1.0078125,
      "learning_rate": 4.2271837543377435e-05,
      "loss": 0.8643,
      "step": 50457
    },
    {
      "epoch": 4.341219994837822,
      "grad_norm": 1.5234375,
      "learning_rate": 4.226095655184775e-05,
      "loss": 1.5592,
      "step": 50458
    },
    {
      "epoch": 4.341306031145143,
      "grad_norm": 1.15625,
      "learning_rate": 4.2250076899115755e-05,
      "loss": 1.9633,
      "step": 50459
    },
    {
      "epoch": 4.341392067452465,
      "grad_norm": 1.5625,
      "learning_rate": 4.223919858521336e-05,
      "loss": 1.4025,
      "step": 50460
    },
    {
      "epoch": 4.341478103759787,
      "grad_norm": 3.875,
      "learning_rate": 4.2228321610172206e-05,
      "loss": 1.1095,
      "step": 50461
    },
    {
      "epoch": 4.341564140067108,
      "grad_norm": 8.6875,
      "learning_rate": 4.221744597402427e-05,
      "loss": 0.5289,
      "step": 50462
    },
    {
      "epoch": 4.34165017637443,
      "grad_norm": 1.4453125,
      "learning_rate": 4.220657167680114e-05,
      "loss": 1.3532,
      "step": 50463
    },
    {
      "epoch": 4.341736212681751,
      "grad_norm": 1.4140625,
      "learning_rate": 4.219569871853479e-05,
      "loss": 1.1811,
      "step": 50464
    },
    {
      "epoch": 4.341822248989073,
      "grad_norm": 0.94140625,
      "learning_rate": 4.2184827099257025e-05,
      "loss": 0.9346,
      "step": 50465
    },
    {
      "epoch": 4.341908285296395,
      "grad_norm": 1.34375,
      "learning_rate": 4.217395681899955e-05,
      "loss": 1.1833,
      "step": 50466
    },
    {
      "epoch": 4.341994321603717,
      "grad_norm": 2912.0,
      "learning_rate": 4.2163087877794217e-05,
      "loss": 1.2314,
      "step": 50467
    },
    {
      "epoch": 4.342080357911039,
      "grad_norm": 1.1796875,
      "learning_rate": 4.215222027567289e-05,
      "loss": 1.2479,
      "step": 50468
    },
    {
      "epoch": 4.34216639421836,
      "grad_norm": 1.140625,
      "learning_rate": 4.2141354012667165e-05,
      "loss": 0.8875,
      "step": 50469
    },
    {
      "epoch": 4.342252430525682,
      "grad_norm": 1.1640625,
      "learning_rate": 4.2130489088808954e-05,
      "loss": 1.2019,
      "step": 50470
    },
    {
      "epoch": 4.342338466833003,
      "grad_norm": 1.4609375,
      "learning_rate": 4.211962550413007e-05,
      "loss": 1.5508,
      "step": 50471
    },
    {
      "epoch": 4.342424503140325,
      "grad_norm": 0.90234375,
      "learning_rate": 4.210876325866214e-05,
      "loss": 0.9146,
      "step": 50472
    },
    {
      "epoch": 4.342510539447647,
      "grad_norm": 1.0234375,
      "learning_rate": 4.2097902352437156e-05,
      "loss": 0.9269,
      "step": 50473
    },
    {
      "epoch": 4.3425965757549685,
      "grad_norm": 1.4609375,
      "learning_rate": 4.208704278548664e-05,
      "loss": 1.6624,
      "step": 50474
    },
    {
      "epoch": 4.342682612062291,
      "grad_norm": 1.453125,
      "learning_rate": 4.207618455784245e-05,
      "loss": 1.1093,
      "step": 50475
    },
    {
      "epoch": 4.342768648369612,
      "grad_norm": 1.6875,
      "learning_rate": 4.206532766953647e-05,
      "loss": 1.3597,
      "step": 50476
    },
    {
      "epoch": 4.342854684676934,
      "grad_norm": 0.98046875,
      "learning_rate": 4.2054472120600265e-05,
      "loss": 0.9916,
      "step": 50477
    },
    {
      "epoch": 4.342940720984255,
      "grad_norm": 1.21875,
      "learning_rate": 4.204361791106559e-05,
      "loss": 1.4095,
      "step": 50478
    },
    {
      "epoch": 4.343026757291577,
      "grad_norm": 1.4453125,
      "learning_rate": 4.2032765040964424e-05,
      "loss": 1.5641,
      "step": 50479
    },
    {
      "epoch": 4.343112793598899,
      "grad_norm": 1.0625,
      "learning_rate": 4.2021913510328244e-05,
      "loss": 0.8639,
      "step": 50480
    },
    {
      "epoch": 4.34319882990622,
      "grad_norm": 1.1484375,
      "learning_rate": 4.201106331918892e-05,
      "loss": 1.0847,
      "step": 50481
    },
    {
      "epoch": 4.343284866213542,
      "grad_norm": 1.2578125,
      "learning_rate": 4.200021446757818e-05,
      "loss": 0.7263,
      "step": 50482
    },
    {
      "epoch": 4.3433709025208636,
      "grad_norm": 23.0,
      "learning_rate": 4.1989366955527685e-05,
      "loss": 1.276,
      "step": 50483
    },
    {
      "epoch": 4.343456938828186,
      "grad_norm": 1.546875,
      "learning_rate": 4.197852078306924e-05,
      "loss": 1.2422,
      "step": 50484
    },
    {
      "epoch": 4.343542975135507,
      "grad_norm": 0.8984375,
      "learning_rate": 4.196767595023454e-05,
      "loss": 0.4986,
      "step": 50485
    },
    {
      "epoch": 4.343629011442829,
      "grad_norm": 0.85546875,
      "learning_rate": 4.1956832457055284e-05,
      "loss": 1.0102,
      "step": 50486
    },
    {
      "epoch": 4.343715047750151,
      "grad_norm": 1.0078125,
      "learning_rate": 4.194599030356327e-05,
      "loss": 0.7786,
      "step": 50487
    },
    {
      "epoch": 4.343801084057472,
      "grad_norm": 133.0,
      "learning_rate": 4.1935149489790105e-05,
      "loss": 1.497,
      "step": 50488
    },
    {
      "epoch": 4.343887120364794,
      "grad_norm": 81.0,
      "learning_rate": 4.192431001576763e-05,
      "loss": 1.341,
      "step": 50489
    },
    {
      "epoch": 4.343973156672115,
      "grad_norm": 2.125,
      "learning_rate": 4.191347188152733e-05,
      "loss": 1.5306,
      "step": 50490
    },
    {
      "epoch": 4.344059192979437,
      "grad_norm": 1.0078125,
      "learning_rate": 4.190263508710107e-05,
      "loss": 0.9284,
      "step": 50491
    },
    {
      "epoch": 4.344145229286759,
      "grad_norm": 6.4375,
      "learning_rate": 4.18917996325206e-05,
      "loss": 1.5409,
      "step": 50492
    },
    {
      "epoch": 4.344231265594081,
      "grad_norm": 1.6953125,
      "learning_rate": 4.1880965517817396e-05,
      "loss": 1.1995,
      "step": 50493
    },
    {
      "epoch": 4.344317301901403,
      "grad_norm": 1.1640625,
      "learning_rate": 4.1870132743023256e-05,
      "loss": 1.0152,
      "step": 50494
    },
    {
      "epoch": 4.344403338208724,
      "grad_norm": 1.0390625,
      "learning_rate": 4.1859301308170004e-05,
      "loss": 0.9875,
      "step": 50495
    },
    {
      "epoch": 4.344489374516046,
      "grad_norm": 1.75,
      "learning_rate": 4.184847121328911e-05,
      "loss": 1.5547,
      "step": 50496
    },
    {
      "epoch": 4.344575410823367,
      "grad_norm": 1.03125,
      "learning_rate": 4.1837642458412316e-05,
      "loss": 1.5389,
      "step": 50497
    },
    {
      "epoch": 4.344661447130689,
      "grad_norm": 1.1640625,
      "learning_rate": 4.182681504357139e-05,
      "loss": 1.1604,
      "step": 50498
    },
    {
      "epoch": 4.34474748343801,
      "grad_norm": 1.390625,
      "learning_rate": 4.1815988968797845e-05,
      "loss": 0.9444,
      "step": 50499
    },
    {
      "epoch": 4.3448335197453325,
      "grad_norm": 184.0,
      "learning_rate": 4.1805164234123395e-05,
      "loss": 0.8818,
      "step": 50500
    },
    {
      "epoch": 4.344919556052655,
      "grad_norm": 7.84375,
      "learning_rate": 4.179434083957983e-05,
      "loss": 1.5252,
      "step": 50501
    },
    {
      "epoch": 4.345005592359976,
      "grad_norm": 1.1328125,
      "learning_rate": 4.178351878519859e-05,
      "loss": 1.2835,
      "step": 50502
    },
    {
      "epoch": 4.345091628667298,
      "grad_norm": 0.9296875,
      "learning_rate": 4.1772698071011585e-05,
      "loss": 1.3798,
      "step": 50503
    },
    {
      "epoch": 4.345177664974619,
      "grad_norm": 1.40625,
      "learning_rate": 4.176187869705017e-05,
      "loss": 1.1471,
      "step": 50504
    },
    {
      "epoch": 4.345263701281941,
      "grad_norm": 1.3046875,
      "learning_rate": 4.175106066334622e-05,
      "loss": 0.7107,
      "step": 50505
    },
    {
      "epoch": 4.345349737589263,
      "grad_norm": 0.94140625,
      "learning_rate": 4.174024396993131e-05,
      "loss": 0.7549,
      "step": 50506
    },
    {
      "epoch": 4.345435773896584,
      "grad_norm": 1.0390625,
      "learning_rate": 4.172942861683698e-05,
      "loss": 0.5196,
      "step": 50507
    },
    {
      "epoch": 4.345521810203906,
      "grad_norm": 1.2890625,
      "learning_rate": 4.171861460409493e-05,
      "loss": 1.2768,
      "step": 50508
    },
    {
      "epoch": 4.345607846511228,
      "grad_norm": 0.94140625,
      "learning_rate": 4.170780193173696e-05,
      "loss": 0.7019,
      "step": 50509
    },
    {
      "epoch": 4.34569388281855,
      "grad_norm": 0.890625,
      "learning_rate": 4.1696990599794374e-05,
      "loss": 0.5823,
      "step": 50510
    },
    {
      "epoch": 4.345779919125871,
      "grad_norm": 16.25,
      "learning_rate": 4.168618060829899e-05,
      "loss": 0.9043,
      "step": 50511
    },
    {
      "epoch": 4.345865955433193,
      "grad_norm": 1.1484375,
      "learning_rate": 4.1675371957282435e-05,
      "loss": 1.1052,
      "step": 50512
    },
    {
      "epoch": 4.345951991740515,
      "grad_norm": 30.125,
      "learning_rate": 4.166456464677626e-05,
      "loss": 1.6574,
      "step": 50513
    },
    {
      "epoch": 4.346038028047836,
      "grad_norm": 1.2421875,
      "learning_rate": 4.1653758676812145e-05,
      "loss": 1.3084,
      "step": 50514
    },
    {
      "epoch": 4.346124064355158,
      "grad_norm": 0.875,
      "learning_rate": 4.164295404742152e-05,
      "loss": 1.0386,
      "step": 50515
    },
    {
      "epoch": 4.346210100662479,
      "grad_norm": 0.9296875,
      "learning_rate": 4.1632150758636136e-05,
      "loss": 0.9539,
      "step": 50516
    },
    {
      "epoch": 4.346296136969801,
      "grad_norm": 0.921875,
      "learning_rate": 4.162134881048768e-05,
      "loss": 1.0134,
      "step": 50517
    },
    {
      "epoch": 4.346382173277123,
      "grad_norm": 1.4140625,
      "learning_rate": 4.161054820300753e-05,
      "loss": 1.7831,
      "step": 50518
    },
    {
      "epoch": 4.346468209584445,
      "grad_norm": 0.94921875,
      "learning_rate": 4.1599748936227365e-05,
      "loss": 0.9144,
      "step": 50519
    },
    {
      "epoch": 4.346554245891767,
      "grad_norm": 1.21875,
      "learning_rate": 4.1588951010178846e-05,
      "loss": 2.1704,
      "step": 50520
    },
    {
      "epoch": 4.346640282199088,
      "grad_norm": 6880.0,
      "learning_rate": 4.157815442489343e-05,
      "loss": 0.6416,
      "step": 50521
    },
    {
      "epoch": 4.34672631850641,
      "grad_norm": 0.93359375,
      "learning_rate": 4.156735918040278e-05,
      "loss": 0.9751,
      "step": 50522
    },
    {
      "epoch": 4.346812354813731,
      "grad_norm": 1.3828125,
      "learning_rate": 4.15565652767384e-05,
      "loss": 1.8448,
      "step": 50523
    },
    {
      "epoch": 4.346898391121053,
      "grad_norm": 1.515625,
      "learning_rate": 4.154577271393189e-05,
      "loss": 1.6266,
      "step": 50524
    },
    {
      "epoch": 4.346984427428374,
      "grad_norm": 1.1875,
      "learning_rate": 4.153498149201495e-05,
      "loss": 1.2627,
      "step": 50525
    },
    {
      "epoch": 4.3470704637356965,
      "grad_norm": 1.4296875,
      "learning_rate": 4.1524191611018946e-05,
      "loss": 1.531,
      "step": 50526
    },
    {
      "epoch": 4.347156500043019,
      "grad_norm": 1.3828125,
      "learning_rate": 4.1513403070975455e-05,
      "loss": 1.1267,
      "step": 50527
    },
    {
      "epoch": 4.34724253635034,
      "grad_norm": 1.015625,
      "learning_rate": 4.1502615871916226e-05,
      "loss": 1.169,
      "step": 50528
    },
    {
      "epoch": 4.347328572657662,
      "grad_norm": 1.1953125,
      "learning_rate": 4.149183001387258e-05,
      "loss": 1.2009,
      "step": 50529
    },
    {
      "epoch": 4.347414608964983,
      "grad_norm": 1.1640625,
      "learning_rate": 4.148104549687626e-05,
      "loss": 1.855,
      "step": 50530
    },
    {
      "epoch": 4.347500645272305,
      "grad_norm": 1.1875,
      "learning_rate": 4.1470262320958576e-05,
      "loss": 1.2198,
      "step": 50531
    },
    {
      "epoch": 4.347586681579626,
      "grad_norm": 1.109375,
      "learning_rate": 4.145948048615122e-05,
      "loss": 1.3664,
      "step": 50532
    },
    {
      "epoch": 4.347672717886948,
      "grad_norm": 2.5,
      "learning_rate": 4.1448699992485796e-05,
      "loss": 1.5212,
      "step": 50533
    },
    {
      "epoch": 4.34775875419427,
      "grad_norm": 1.2421875,
      "learning_rate": 4.143792083999365e-05,
      "loss": 1.0557,
      "step": 50534
    },
    {
      "epoch": 4.347844790501592,
      "grad_norm": 1.140625,
      "learning_rate": 4.142714302870637e-05,
      "loss": 1.0746,
      "step": 50535
    },
    {
      "epoch": 4.347930826808914,
      "grad_norm": 1.34375,
      "learning_rate": 4.141636655865566e-05,
      "loss": 0.9022,
      "step": 50536
    },
    {
      "epoch": 4.348016863116235,
      "grad_norm": 1.078125,
      "learning_rate": 4.1405591429872715e-05,
      "loss": 0.7215,
      "step": 50537
    },
    {
      "epoch": 4.348102899423557,
      "grad_norm": 1.171875,
      "learning_rate": 4.1394817642389284e-05,
      "loss": 1.1124,
      "step": 50538
    },
    {
      "epoch": 4.348188935730878,
      "grad_norm": 1.21875,
      "learning_rate": 4.13840451962369e-05,
      "loss": 0.7972,
      "step": 50539
    },
    {
      "epoch": 4.3482749720382,
      "grad_norm": 1.3125,
      "learning_rate": 4.1373274091446864e-05,
      "loss": 1.1092,
      "step": 50540
    },
    {
      "epoch": 4.348361008345522,
      "grad_norm": 251.0,
      "learning_rate": 4.136250432805089e-05,
      "loss": 1.8212,
      "step": 50541
    },
    {
      "epoch": 4.348447044652843,
      "grad_norm": 0.875,
      "learning_rate": 4.135173590608038e-05,
      "loss": 0.4043,
      "step": 50542
    },
    {
      "epoch": 4.348533080960165,
      "grad_norm": 1.15625,
      "learning_rate": 4.134096882556682e-05,
      "loss": 1.1523,
      "step": 50543
    },
    {
      "epoch": 4.348619117267487,
      "grad_norm": 1.1015625,
      "learning_rate": 4.1330203086541784e-05,
      "loss": 0.4599,
      "step": 50544
    },
    {
      "epoch": 4.348705153574809,
      "grad_norm": 0.7578125,
      "learning_rate": 4.131943868903659e-05,
      "loss": 0.6201,
      "step": 50545
    },
    {
      "epoch": 4.34879118988213,
      "grad_norm": 0.96875,
      "learning_rate": 4.1308675633082824e-05,
      "loss": 0.847,
      "step": 50546
    },
    {
      "epoch": 4.348877226189452,
      "grad_norm": 1.109375,
      "learning_rate": 4.1297913918712127e-05,
      "loss": 0.8477,
      "step": 50547
    },
    {
      "epoch": 4.348963262496774,
      "grad_norm": 3.578125,
      "learning_rate": 4.1287153545955694e-05,
      "loss": 1.0415,
      "step": 50548
    },
    {
      "epoch": 4.349049298804095,
      "grad_norm": 39.5,
      "learning_rate": 4.1276394514845105e-05,
      "loss": 1.1266,
      "step": 50549
    },
    {
      "epoch": 4.349135335111417,
      "grad_norm": 1.109375,
      "learning_rate": 4.12656368254119e-05,
      "loss": 1.6517,
      "step": 50550
    },
    {
      "epoch": 4.349221371418738,
      "grad_norm": 1.4296875,
      "learning_rate": 4.125488047768749e-05,
      "loss": 1.6993,
      "step": 50551
    },
    {
      "epoch": 4.3493074077260605,
      "grad_norm": 0.88671875,
      "learning_rate": 4.124412547170325e-05,
      "loss": 1.0122,
      "step": 50552
    },
    {
      "epoch": 4.349393444033382,
      "grad_norm": 1.2109375,
      "learning_rate": 4.123337180749087e-05,
      "loss": 0.5903,
      "step": 50553
    },
    {
      "epoch": 4.349479480340704,
      "grad_norm": 58.75,
      "learning_rate": 4.1222619485081545e-05,
      "loss": 0.936,
      "step": 50554
    },
    {
      "epoch": 4.349565516648026,
      "grad_norm": 1.3359375,
      "learning_rate": 4.121186850450692e-05,
      "loss": 1.4211,
      "step": 50555
    },
    {
      "epoch": 4.349651552955347,
      "grad_norm": 1.03125,
      "learning_rate": 4.120111886579825e-05,
      "loss": 0.9474,
      "step": 50556
    },
    {
      "epoch": 4.349737589262669,
      "grad_norm": 0.8984375,
      "learning_rate": 4.119037056898706e-05,
      "loss": 0.589,
      "step": 50557
    },
    {
      "epoch": 4.34982362556999,
      "grad_norm": 0.9375,
      "learning_rate": 4.1179623614104935e-05,
      "loss": 0.7174,
      "step": 50558
    },
    {
      "epoch": 4.349909661877312,
      "grad_norm": 1.0625,
      "learning_rate": 4.1168878001182964e-05,
      "loss": 1.325,
      "step": 50559
    },
    {
      "epoch": 4.3499956981846335,
      "grad_norm": 0.95703125,
      "learning_rate": 4.1158133730252954e-05,
      "loss": 0.716,
      "step": 50560
    },
    {
      "epoch": 4.350081734491956,
      "grad_norm": 0.91015625,
      "learning_rate": 4.114739080134611e-05,
      "loss": 0.889,
      "step": 50561
    },
    {
      "epoch": 4.350167770799278,
      "grad_norm": 1.1875,
      "learning_rate": 4.113664921449389e-05,
      "loss": 0.9997,
      "step": 50562
    },
    {
      "epoch": 4.350253807106599,
      "grad_norm": 1.7890625,
      "learning_rate": 4.1125908969727845e-05,
      "loss": 1.6954,
      "step": 50563
    },
    {
      "epoch": 4.350339843413921,
      "grad_norm": 0.9453125,
      "learning_rate": 4.1115170067079155e-05,
      "loss": 1.4285,
      "step": 50564
    },
    {
      "epoch": 4.350425879721242,
      "grad_norm": 1.7734375,
      "learning_rate": 4.11044325065793e-05,
      "loss": 1.3897,
      "step": 50565
    },
    {
      "epoch": 4.350511916028564,
      "grad_norm": 1.0390625,
      "learning_rate": 4.109369628825987e-05,
      "loss": 1.1326,
      "step": 50566
    },
    {
      "epoch": 4.350597952335885,
      "grad_norm": 1.328125,
      "learning_rate": 4.108296141215206e-05,
      "loss": 1.0879,
      "step": 50567
    },
    {
      "epoch": 4.350683988643207,
      "grad_norm": 0.9140625,
      "learning_rate": 4.1072227878287294e-05,
      "loss": 0.8902,
      "step": 50568
    },
    {
      "epoch": 4.3507700249505294,
      "grad_norm": 1.0546875,
      "learning_rate": 4.106149568669709e-05,
      "loss": 0.7363,
      "step": 50569
    },
    {
      "epoch": 4.350856061257851,
      "grad_norm": 1.109375,
      "learning_rate": 4.105076483741266e-05,
      "loss": 1.1033,
      "step": 50570
    },
    {
      "epoch": 4.350942097565173,
      "grad_norm": 1.0625,
      "learning_rate": 4.104003533046552e-05,
      "loss": 1.4256,
      "step": 50571
    },
    {
      "epoch": 4.351028133872494,
      "grad_norm": 1.046875,
      "learning_rate": 4.1029307165887096e-05,
      "loss": 1.0691,
      "step": 50572
    },
    {
      "epoch": 4.351114170179816,
      "grad_norm": 1.046875,
      "learning_rate": 4.101858034370859e-05,
      "loss": 0.7414,
      "step": 50573
    },
    {
      "epoch": 4.351200206487137,
      "grad_norm": 0.97265625,
      "learning_rate": 4.100785486396152e-05,
      "loss": 1.2888,
      "step": 50574
    },
    {
      "epoch": 4.351286242794459,
      "grad_norm": 0.890625,
      "learning_rate": 4.099713072667716e-05,
      "loss": 0.5701,
      "step": 50575
    },
    {
      "epoch": 4.351372279101781,
      "grad_norm": 0.94140625,
      "learning_rate": 4.09864079318869e-05,
      "loss": 0.7894,
      "step": 50576
    },
    {
      "epoch": 4.351458315409102,
      "grad_norm": 0.93359375,
      "learning_rate": 4.0975686479622184e-05,
      "loss": 1.2481,
      "step": 50577
    },
    {
      "epoch": 4.3515443517164245,
      "grad_norm": 1.2890625,
      "learning_rate": 4.0964966369914255e-05,
      "loss": 1.56,
      "step": 50578
    },
    {
      "epoch": 4.351630388023746,
      "grad_norm": 0.7578125,
      "learning_rate": 4.095424760279454e-05,
      "loss": 0.7428,
      "step": 50579
    },
    {
      "epoch": 4.351716424331068,
      "grad_norm": 1.1875,
      "learning_rate": 4.094353017829444e-05,
      "loss": 0.9579,
      "step": 50580
    },
    {
      "epoch": 4.351802460638389,
      "grad_norm": 4.15625,
      "learning_rate": 4.093281409644517e-05,
      "loss": 0.7999,
      "step": 50581
    },
    {
      "epoch": 4.351888496945711,
      "grad_norm": 1.328125,
      "learning_rate": 4.092209935727809e-05,
      "loss": 0.8688,
      "step": 50582
    },
    {
      "epoch": 4.351974533253033,
      "grad_norm": 0.89453125,
      "learning_rate": 4.0911385960824724e-05,
      "loss": 0.7905,
      "step": 50583
    },
    {
      "epoch": 4.352060569560354,
      "grad_norm": 1.1015625,
      "learning_rate": 4.090067390711616e-05,
      "loss": 1.0191,
      "step": 50584
    },
    {
      "epoch": 4.352146605867676,
      "grad_norm": 0.984375,
      "learning_rate": 4.088996319618388e-05,
      "loss": 1.426,
      "step": 50585
    },
    {
      "epoch": 4.3522326421749975,
      "grad_norm": 1.125,
      "learning_rate": 4.087925382805913e-05,
      "loss": 0.606,
      "step": 50586
    },
    {
      "epoch": 4.35231867848232,
      "grad_norm": 1.7109375,
      "learning_rate": 4.086854580277322e-05,
      "loss": 0.6704,
      "step": 50587
    },
    {
      "epoch": 4.352404714789642,
      "grad_norm": 0.93359375,
      "learning_rate": 4.085783912035762e-05,
      "loss": 1.4846,
      "step": 50588
    },
    {
      "epoch": 4.352490751096963,
      "grad_norm": 15.125,
      "learning_rate": 4.084713378084348e-05,
      "loss": 1.4082,
      "step": 50589
    },
    {
      "epoch": 4.352576787404285,
      "grad_norm": 0.87109375,
      "learning_rate": 4.083642978426216e-05,
      "loss": 0.6572,
      "step": 50590
    },
    {
      "epoch": 4.352662823711606,
      "grad_norm": 0.9921875,
      "learning_rate": 4.0825727130645086e-05,
      "loss": 0.6315,
      "step": 50591
    },
    {
      "epoch": 4.352748860018928,
      "grad_norm": 18.0,
      "learning_rate": 4.081502582002333e-05,
      "loss": 1.4539,
      "step": 50592
    },
    {
      "epoch": 4.352834896326249,
      "grad_norm": 1.34375,
      "learning_rate": 4.0804325852428385e-05,
      "loss": 1.1251,
      "step": 50593
    },
    {
      "epoch": 4.352920932633571,
      "grad_norm": 4.03125,
      "learning_rate": 4.0793627227891486e-05,
      "loss": 1.4214,
      "step": 50594
    },
    {
      "epoch": 4.3530069689408935,
      "grad_norm": 0.87890625,
      "learning_rate": 4.078292994644389e-05,
      "loss": 0.9993,
      "step": 50595
    },
    {
      "epoch": 4.353093005248215,
      "grad_norm": 1.15625,
      "learning_rate": 4.077223400811686e-05,
      "loss": 1.0597,
      "step": 50596
    },
    {
      "epoch": 4.353179041555537,
      "grad_norm": 1.5859375,
      "learning_rate": 4.076153941294175e-05,
      "loss": 1.5792,
      "step": 50597
    },
    {
      "epoch": 4.353265077862858,
      "grad_norm": 1.296875,
      "learning_rate": 4.0750846160949815e-05,
      "loss": 1.5515,
      "step": 50598
    },
    {
      "epoch": 4.35335111417018,
      "grad_norm": 1.171875,
      "learning_rate": 4.074015425217242e-05,
      "loss": 0.9652,
      "step": 50599
    },
    {
      "epoch": 4.353437150477501,
      "grad_norm": 0.9765625,
      "learning_rate": 4.072946368664071e-05,
      "loss": 1.0432,
      "step": 50600
    },
    {
      "epoch": 4.353523186784823,
      "grad_norm": 2368.0,
      "learning_rate": 4.0718774464386046e-05,
      "loss": 0.9141,
      "step": 50601
    },
    {
      "epoch": 4.353609223092145,
      "grad_norm": 0.87890625,
      "learning_rate": 4.070808658543956e-05,
      "loss": 0.633,
      "step": 50602
    },
    {
      "epoch": 4.353695259399466,
      "grad_norm": 0.87109375,
      "learning_rate": 4.0697400049832576e-05,
      "loss": 0.7379,
      "step": 50603
    },
    {
      "epoch": 4.3537812957067885,
      "grad_norm": 1.1796875,
      "learning_rate": 4.068671485759651e-05,
      "loss": 1.5022,
      "step": 50604
    },
    {
      "epoch": 4.35386733201411,
      "grad_norm": 1.1328125,
      "learning_rate": 4.0676031008762325e-05,
      "loss": 1.5126,
      "step": 50605
    },
    {
      "epoch": 4.353953368321432,
      "grad_norm": 1.171875,
      "learning_rate": 4.0665348503361455e-05,
      "loss": 1.1857,
      "step": 50606
    },
    {
      "epoch": 4.354039404628753,
      "grad_norm": 1.21875,
      "learning_rate": 4.065466734142514e-05,
      "loss": 0.608,
      "step": 50607
    },
    {
      "epoch": 4.354125440936075,
      "grad_norm": 1.203125,
      "learning_rate": 4.064398752298454e-05,
      "loss": 1.4151,
      "step": 50608
    },
    {
      "epoch": 4.354211477243397,
      "grad_norm": 1.0,
      "learning_rate": 4.063330904807094e-05,
      "loss": 0.4261,
      "step": 50609
    },
    {
      "epoch": 4.354297513550718,
      "grad_norm": 1.2421875,
      "learning_rate": 4.0622631916715615e-05,
      "loss": 1.7732,
      "step": 50610
    },
    {
      "epoch": 4.35438354985804,
      "grad_norm": 0.91796875,
      "learning_rate": 4.061195612894969e-05,
      "loss": 0.8362,
      "step": 50611
    },
    {
      "epoch": 4.3544695861653615,
      "grad_norm": 0.9765625,
      "learning_rate": 4.060128168480448e-05,
      "loss": 0.7316,
      "step": 50612
    },
    {
      "epoch": 4.354555622472684,
      "grad_norm": 103.5,
      "learning_rate": 4.0590608584311183e-05,
      "loss": 1.7541,
      "step": 50613
    },
    {
      "epoch": 4.354641658780005,
      "grad_norm": 1.1484375,
      "learning_rate": 4.0579936827501e-05,
      "loss": 0.9601,
      "step": 50614
    },
    {
      "epoch": 4.354727695087327,
      "grad_norm": 1.40625,
      "learning_rate": 4.056926641440517e-05,
      "loss": 1.1163,
      "step": 50615
    },
    {
      "epoch": 4.354813731394649,
      "grad_norm": 1.171875,
      "learning_rate": 4.055859734505485e-05,
      "loss": 1.0799,
      "step": 50616
    },
    {
      "epoch": 4.35489976770197,
      "grad_norm": 1.203125,
      "learning_rate": 4.054792961948123e-05,
      "loss": 1.528,
      "step": 50617
    },
    {
      "epoch": 4.354985804009292,
      "grad_norm": 1.3828125,
      "learning_rate": 4.053726323771567e-05,
      "loss": 1.4833,
      "step": 50618
    },
    {
      "epoch": 4.355071840316613,
      "grad_norm": 1.1328125,
      "learning_rate": 4.0526598199789153e-05,
      "loss": 1.4486,
      "step": 50619
    },
    {
      "epoch": 4.355157876623935,
      "grad_norm": 0.91796875,
      "learning_rate": 4.051593450573299e-05,
      "loss": 0.5872,
      "step": 50620
    },
    {
      "epoch": 4.3552439129312575,
      "grad_norm": 0.9765625,
      "learning_rate": 4.050527215557842e-05,
      "loss": 1.0415,
      "step": 50621
    },
    {
      "epoch": 4.355329949238579,
      "grad_norm": 1.765625,
      "learning_rate": 4.0494611149356486e-05,
      "loss": 1.1441,
      "step": 50622
    },
    {
      "epoch": 4.355415985545901,
      "grad_norm": 496.0,
      "learning_rate": 4.0483951487098494e-05,
      "loss": 1.1525,
      "step": 50623
    },
    {
      "epoch": 4.355502021853222,
      "grad_norm": 5.40625,
      "learning_rate": 4.047329316883558e-05,
      "loss": 0.8729,
      "step": 50624
    },
    {
      "epoch": 4.355588058160544,
      "grad_norm": 5.375,
      "learning_rate": 4.046263619459889e-05,
      "loss": 0.8416,
      "step": 50625
    },
    {
      "epoch": 4.355674094467865,
      "grad_norm": 1.203125,
      "learning_rate": 4.0451980564419676e-05,
      "loss": 1.0392,
      "step": 50626
    },
    {
      "epoch": 4.355760130775187,
      "grad_norm": 1.109375,
      "learning_rate": 4.0441326278328914e-05,
      "loss": 0.9142,
      "step": 50627
    },
    {
      "epoch": 4.355846167082509,
      "grad_norm": 1.015625,
      "learning_rate": 4.0430673336357913e-05,
      "loss": 1.0658,
      "step": 50628
    },
    {
      "epoch": 4.3559322033898304,
      "grad_norm": 1.5078125,
      "learning_rate": 4.042002173853793e-05,
      "loss": 1.2444,
      "step": 50629
    },
    {
      "epoch": 4.3560182396971525,
      "grad_norm": 0.9609375,
      "learning_rate": 4.040937148489987e-05,
      "loss": 1.4594,
      "step": 50630
    },
    {
      "epoch": 4.356104276004474,
      "grad_norm": 1.078125,
      "learning_rate": 4.0398722575475055e-05,
      "loss": 1.8196,
      "step": 50631
    },
    {
      "epoch": 4.356190312311796,
      "grad_norm": 1.1015625,
      "learning_rate": 4.038807501029468e-05,
      "loss": 1.0767,
      "step": 50632
    },
    {
      "epoch": 4.356276348619117,
      "grad_norm": 0.79296875,
      "learning_rate": 4.0377428789389605e-05,
      "loss": 0.4228,
      "step": 50633
    },
    {
      "epoch": 4.356362384926439,
      "grad_norm": 1.0234375,
      "learning_rate": 4.036678391279131e-05,
      "loss": 1.0287,
      "step": 50634
    },
    {
      "epoch": 4.356448421233761,
      "grad_norm": 1.265625,
      "learning_rate": 4.0356140380530704e-05,
      "loss": 0.697,
      "step": 50635
    },
    {
      "epoch": 4.356534457541082,
      "grad_norm": 1.234375,
      "learning_rate": 4.0345498192639e-05,
      "loss": 1.5399,
      "step": 50636
    },
    {
      "epoch": 4.356620493848404,
      "grad_norm": 0.953125,
      "learning_rate": 4.0334857349147436e-05,
      "loss": 0.5501,
      "step": 50637
    },
    {
      "epoch": 4.3567065301557255,
      "grad_norm": 0.94140625,
      "learning_rate": 4.032421785008688e-05,
      "loss": 1.1329,
      "step": 50638
    },
    {
      "epoch": 4.356792566463048,
      "grad_norm": 0.93359375,
      "learning_rate": 4.03135796954886e-05,
      "loss": 1.2006,
      "step": 50639
    },
    {
      "epoch": 4.356878602770369,
      "grad_norm": 1.1015625,
      "learning_rate": 4.030294288538378e-05,
      "loss": 0.8001,
      "step": 50640
    },
    {
      "epoch": 4.356964639077691,
      "grad_norm": 1.1171875,
      "learning_rate": 4.029230741980333e-05,
      "loss": 1.6063,
      "step": 50641
    },
    {
      "epoch": 4.357050675385013,
      "grad_norm": 5.5,
      "learning_rate": 4.028167329877858e-05,
      "loss": 1.4671,
      "step": 50642
    },
    {
      "epoch": 4.357136711692334,
      "grad_norm": 1.1484375,
      "learning_rate": 4.027104052234048e-05,
      "loss": 0.5431,
      "step": 50643
    },
    {
      "epoch": 4.357222747999656,
      "grad_norm": 1.390625,
      "learning_rate": 4.0260409090520144e-05,
      "loss": 1.9967,
      "step": 50644
    },
    {
      "epoch": 4.357308784306977,
      "grad_norm": 1.3671875,
      "learning_rate": 4.024977900334875e-05,
      "loss": 1.5219,
      "step": 50645
    },
    {
      "epoch": 4.357394820614299,
      "grad_norm": 1.609375,
      "learning_rate": 4.023915026085728e-05,
      "loss": 1.9148,
      "step": 50646
    },
    {
      "epoch": 4.357480856921621,
      "grad_norm": 1.234375,
      "learning_rate": 4.022852286307688e-05,
      "loss": 1.4082,
      "step": 50647
    },
    {
      "epoch": 4.357566893228943,
      "grad_norm": 1.3515625,
      "learning_rate": 4.021789681003873e-05,
      "loss": 0.9771,
      "step": 50648
    },
    {
      "epoch": 4.357652929536265,
      "grad_norm": 0.98828125,
      "learning_rate": 4.020727210177372e-05,
      "loss": 0.8614,
      "step": 50649
    },
    {
      "epoch": 4.357738965843586,
      "grad_norm": 1.03125,
      "learning_rate": 4.019664873831297e-05,
      "loss": 1.1557,
      "step": 50650
    },
    {
      "epoch": 4.357825002150908,
      "grad_norm": 56.0,
      "learning_rate": 4.018602671968774e-05,
      "loss": 1.0317,
      "step": 50651
    },
    {
      "epoch": 4.357911038458229,
      "grad_norm": 2.328125,
      "learning_rate": 4.017540604592884e-05,
      "loss": 1.1367,
      "step": 50652
    },
    {
      "epoch": 4.357997074765551,
      "grad_norm": 0.97265625,
      "learning_rate": 4.016478671706742e-05,
      "loss": 1.26,
      "step": 50653
    },
    {
      "epoch": 4.358083111072872,
      "grad_norm": 1.4375,
      "learning_rate": 4.015416873313471e-05,
      "loss": 1.4223,
      "step": 50654
    },
    {
      "epoch": 4.3581691473801945,
      "grad_norm": 7.8125,
      "learning_rate": 4.014355209416148e-05,
      "loss": 0.9196,
      "step": 50655
    },
    {
      "epoch": 4.3582551836875165,
      "grad_norm": 1.2890625,
      "learning_rate": 4.0132936800179044e-05,
      "loss": 1.5983,
      "step": 50656
    },
    {
      "epoch": 4.358341219994838,
      "grad_norm": 1.171875,
      "learning_rate": 4.01223228512182e-05,
      "loss": 1.1414,
      "step": 50657
    },
    {
      "epoch": 4.35842725630216,
      "grad_norm": 1.4765625,
      "learning_rate": 4.011171024731014e-05,
      "loss": 0.87,
      "step": 50658
    },
    {
      "epoch": 4.358513292609481,
      "grad_norm": 1.328125,
      "learning_rate": 4.0101098988486015e-05,
      "loss": 0.8657,
      "step": 50659
    },
    {
      "epoch": 4.358599328916803,
      "grad_norm": 0.99609375,
      "learning_rate": 4.009048907477658e-05,
      "loss": 0.5898,
      "step": 50660
    },
    {
      "epoch": 4.358685365224124,
      "grad_norm": 1.0859375,
      "learning_rate": 4.007988050621309e-05,
      "loss": 1.2387,
      "step": 50661
    },
    {
      "epoch": 4.358771401531446,
      "grad_norm": 0.76171875,
      "learning_rate": 4.00692732828265e-05,
      "loss": 0.6365,
      "step": 50662
    },
    {
      "epoch": 4.358857437838768,
      "grad_norm": 4.0625,
      "learning_rate": 4.005866740464781e-05,
      "loss": 1.1951,
      "step": 50663
    },
    {
      "epoch": 4.3589434741460895,
      "grad_norm": 1.4609375,
      "learning_rate": 4.0048062871708037e-05,
      "loss": 1.1895,
      "step": 50664
    },
    {
      "epoch": 4.359029510453412,
      "grad_norm": 1.1796875,
      "learning_rate": 4.003745968403832e-05,
      "loss": 1.8871,
      "step": 50665
    },
    {
      "epoch": 4.359115546760733,
      "grad_norm": 5.65625,
      "learning_rate": 4.0026857841669484e-05,
      "loss": 0.4626,
      "step": 50666
    },
    {
      "epoch": 4.359201583068055,
      "grad_norm": 1.1875,
      "learning_rate": 4.001625734463271e-05,
      "loss": 0.684,
      "step": 50667
    },
    {
      "epoch": 4.359287619375376,
      "grad_norm": 1.2890625,
      "learning_rate": 4.000565819295887e-05,
      "loss": 1.2244,
      "step": 50668
    },
    {
      "epoch": 4.359373655682698,
      "grad_norm": 0.96875,
      "learning_rate": 3.9995060386678885e-05,
      "loss": 1.0253,
      "step": 50669
    },
    {
      "epoch": 4.35945969199002,
      "grad_norm": 1.234375,
      "learning_rate": 3.998446392582411e-05,
      "loss": 1.2739,
      "step": 50670
    },
    {
      "epoch": 4.359545728297341,
      "grad_norm": 1.09375,
      "learning_rate": 3.99738688104252e-05,
      "loss": 1.2839,
      "step": 50671
    },
    {
      "epoch": 4.359631764604663,
      "grad_norm": 0.90234375,
      "learning_rate": 3.996327504051328e-05,
      "loss": 1.0611,
      "step": 50672
    },
    {
      "epoch": 4.359717800911985,
      "grad_norm": 4160.0,
      "learning_rate": 3.995268261611928e-05,
      "loss": 1.158,
      "step": 50673
    },
    {
      "epoch": 4.359803837219307,
      "grad_norm": 1.0859375,
      "learning_rate": 3.9942091537274185e-05,
      "loss": 0.7875,
      "step": 50674
    },
    {
      "epoch": 4.359889873526628,
      "grad_norm": 0.8359375,
      "learning_rate": 3.993150180400912e-05,
      "loss": 0.7653,
      "step": 50675
    },
    {
      "epoch": 4.35997590983395,
      "grad_norm": 1.171875,
      "learning_rate": 3.9920913416354845e-05,
      "loss": 0.9361,
      "step": 50676
    },
    {
      "epoch": 4.360061946141272,
      "grad_norm": 75.5,
      "learning_rate": 3.991032637434239e-05,
      "loss": 0.6037,
      "step": 50677
    },
    {
      "epoch": 4.360147982448593,
      "grad_norm": 0.94140625,
      "learning_rate": 3.9899740678002836e-05,
      "loss": 1.7936,
      "step": 50678
    },
    {
      "epoch": 4.360234018755915,
      "grad_norm": 0.9375,
      "learning_rate": 3.988915632736695e-05,
      "loss": 0.7846,
      "step": 50679
    },
    {
      "epoch": 4.360320055063236,
      "grad_norm": 1.296875,
      "learning_rate": 3.98785733224658e-05,
      "loss": 0.9184,
      "step": 50680
    },
    {
      "epoch": 4.3604060913705585,
      "grad_norm": 14.4375,
      "learning_rate": 3.9867991663330495e-05,
      "loss": 1.5664,
      "step": 50681
    },
    {
      "epoch": 4.36049212767788,
      "grad_norm": 0.921875,
      "learning_rate": 3.9857411349991655e-05,
      "loss": 0.6265,
      "step": 50682
    },
    {
      "epoch": 4.360578163985202,
      "grad_norm": 1.1328125,
      "learning_rate": 3.984683238248044e-05,
      "loss": 1.3636,
      "step": 50683
    },
    {
      "epoch": 4.360664200292524,
      "grad_norm": 1.15625,
      "learning_rate": 3.9836254760827806e-05,
      "loss": 1.6133,
      "step": 50684
    },
    {
      "epoch": 4.360750236599845,
      "grad_norm": 1.21875,
      "learning_rate": 3.982567848506452e-05,
      "loss": 1.1408,
      "step": 50685
    },
    {
      "epoch": 4.360836272907167,
      "grad_norm": 2.0,
      "learning_rate": 3.9815103555221775e-05,
      "loss": 1.727,
      "step": 50686
    },
    {
      "epoch": 4.360922309214488,
      "grad_norm": 1.15625,
      "learning_rate": 3.9804529971330215e-05,
      "loss": 1.7538,
      "step": 50687
    },
    {
      "epoch": 4.36100834552181,
      "grad_norm": 4.4375,
      "learning_rate": 3.979395773342093e-05,
      "loss": 1.2671,
      "step": 50688
    },
    {
      "epoch": 4.3610943818291314,
      "grad_norm": 1.0546875,
      "learning_rate": 3.978338684152488e-05,
      "loss": 0.4907,
      "step": 50689
    },
    {
      "epoch": 4.3611804181364535,
      "grad_norm": 2.4375,
      "learning_rate": 3.9772817295672835e-05,
      "loss": 0.5677,
      "step": 50690
    },
    {
      "epoch": 4.361266454443776,
      "grad_norm": 1.203125,
      "learning_rate": 3.9762249095895816e-05,
      "loss": 1.5915,
      "step": 50691
    },
    {
      "epoch": 4.361352490751097,
      "grad_norm": 0.93359375,
      "learning_rate": 3.975168224222475e-05,
      "loss": 0.8272,
      "step": 50692
    },
    {
      "epoch": 4.361438527058419,
      "grad_norm": 12.625,
      "learning_rate": 3.974111673469044e-05,
      "loss": 1.6946,
      "step": 50693
    },
    {
      "epoch": 4.36152456336574,
      "grad_norm": 1.3671875,
      "learning_rate": 3.973055257332386e-05,
      "loss": 0.8035,
      "step": 50694
    },
    {
      "epoch": 4.361610599673062,
      "grad_norm": 1.3515625,
      "learning_rate": 3.971998975815594e-05,
      "loss": 0.7375,
      "step": 50695
    },
    {
      "epoch": 4.361696635980383,
      "grad_norm": 18.75,
      "learning_rate": 3.9709428289217475e-05,
      "loss": 0.5667,
      "step": 50696
    },
    {
      "epoch": 4.361782672287705,
      "grad_norm": 1.046875,
      "learning_rate": 3.9698868166539504e-05,
      "loss": 0.7735,
      "step": 50697
    },
    {
      "epoch": 4.361868708595027,
      "grad_norm": 1.1015625,
      "learning_rate": 3.968830939015272e-05,
      "loss": 0.9665,
      "step": 50698
    },
    {
      "epoch": 4.361954744902349,
      "grad_norm": 1.0078125,
      "learning_rate": 3.96777519600881e-05,
      "loss": 0.943,
      "step": 50699
    },
    {
      "epoch": 4.362040781209671,
      "grad_norm": 2.453125,
      "learning_rate": 3.9667195876376627e-05,
      "loss": 1.7493,
      "step": 50700
    },
    {
      "epoch": 4.362126817516992,
      "grad_norm": 1.4453125,
      "learning_rate": 3.965664113904899e-05,
      "loss": 1.3494,
      "step": 50701
    },
    {
      "epoch": 4.362212853824314,
      "grad_norm": 1.6875,
      "learning_rate": 3.964608774813616e-05,
      "loss": 0.8187,
      "step": 50702
    },
    {
      "epoch": 4.362298890131636,
      "grad_norm": 1.0859375,
      "learning_rate": 3.9635535703669015e-05,
      "loss": 1.1188,
      "step": 50703
    },
    {
      "epoch": 4.362384926438957,
      "grad_norm": 282.0,
      "learning_rate": 3.9624985005678346e-05,
      "loss": 0.7574,
      "step": 50704
    },
    {
      "epoch": 4.362470962746279,
      "grad_norm": 16.125,
      "learning_rate": 3.961443565419509e-05,
      "loss": 0.779,
      "step": 50705
    },
    {
      "epoch": 4.3625569990536,
      "grad_norm": 1.921875,
      "learning_rate": 3.960388764925005e-05,
      "loss": 1.5295,
      "step": 50706
    },
    {
      "epoch": 4.3626430353609225,
      "grad_norm": 0.96875,
      "learning_rate": 3.9593340990874085e-05,
      "loss": 0.5701,
      "step": 50707
    },
    {
      "epoch": 4.362729071668244,
      "grad_norm": 1.0390625,
      "learning_rate": 3.958279567909812e-05,
      "loss": 0.9566,
      "step": 50708
    },
    {
      "epoch": 4.362815107975566,
      "grad_norm": 1.1953125,
      "learning_rate": 3.957225171395284e-05,
      "loss": 0.7068,
      "step": 50709
    },
    {
      "epoch": 4.362901144282888,
      "grad_norm": 1.0703125,
      "learning_rate": 3.956170909546924e-05,
      "loss": 1.0488,
      "step": 50710
    },
    {
      "epoch": 4.362987180590209,
      "grad_norm": 1.265625,
      "learning_rate": 3.9551167823678126e-05,
      "loss": 1.2811,
      "step": 50711
    },
    {
      "epoch": 4.363073216897531,
      "grad_norm": 1.125,
      "learning_rate": 3.954062789861024e-05,
      "loss": 0.7989,
      "step": 50712
    },
    {
      "epoch": 4.363159253204852,
      "grad_norm": 3.0625,
      "learning_rate": 3.95300893202965e-05,
      "loss": 1.6833,
      "step": 50713
    },
    {
      "epoch": 4.363245289512174,
      "grad_norm": 1.53125,
      "learning_rate": 3.951955208876762e-05,
      "loss": 1.7237,
      "step": 50714
    },
    {
      "epoch": 4.3633313258194955,
      "grad_norm": 4.65625,
      "learning_rate": 3.950901620405451e-05,
      "loss": 1.6402,
      "step": 50715
    },
    {
      "epoch": 4.3634173621268175,
      "grad_norm": 1.4609375,
      "learning_rate": 3.949848166618808e-05,
      "loss": 0.7873,
      "step": 50716
    },
    {
      "epoch": 4.36350339843414,
      "grad_norm": 1.578125,
      "learning_rate": 3.9487948475198876e-05,
      "loss": 1.2436,
      "step": 50717
    },
    {
      "epoch": 4.363589434741461,
      "grad_norm": 1.2109375,
      "learning_rate": 3.9477416631117926e-05,
      "loss": 1.0167,
      "step": 50718
    },
    {
      "epoch": 4.363675471048783,
      "grad_norm": 1.390625,
      "learning_rate": 3.9466886133975975e-05,
      "loss": 1.9057,
      "step": 50719
    },
    {
      "epoch": 4.363761507356104,
      "grad_norm": 1.046875,
      "learning_rate": 3.945635698380379e-05,
      "loss": 0.951,
      "step": 50720
    },
    {
      "epoch": 4.363847543663426,
      "grad_norm": 1.0703125,
      "learning_rate": 3.944582918063216e-05,
      "loss": 0.9065,
      "step": 50721
    },
    {
      "epoch": 4.363933579970747,
      "grad_norm": 1.5625,
      "learning_rate": 3.943530272449203e-05,
      "loss": 1.5123,
      "step": 50722
    },
    {
      "epoch": 4.364019616278069,
      "grad_norm": 1.1484375,
      "learning_rate": 3.942477761541396e-05,
      "loss": 0.6966,
      "step": 50723
    },
    {
      "epoch": 4.364105652585391,
      "grad_norm": 2.171875,
      "learning_rate": 3.9414253853428837e-05,
      "loss": 1.5148,
      "step": 50724
    },
    {
      "epoch": 4.364191688892713,
      "grad_norm": 1.25,
      "learning_rate": 3.940373143856751e-05,
      "loss": 0.9635,
      "step": 50725
    },
    {
      "epoch": 4.364277725200035,
      "grad_norm": 3.734375,
      "learning_rate": 3.939321037086063e-05,
      "loss": 0.9721,
      "step": 50726
    },
    {
      "epoch": 4.364363761507356,
      "grad_norm": 5.71875,
      "learning_rate": 3.938269065033906e-05,
      "loss": 0.8353,
      "step": 50727
    },
    {
      "epoch": 4.364449797814678,
      "grad_norm": 0.6953125,
      "learning_rate": 3.937217227703349e-05,
      "loss": 0.3686,
      "step": 50728
    },
    {
      "epoch": 4.364535834121999,
      "grad_norm": 1.3984375,
      "learning_rate": 3.9361655250974734e-05,
      "loss": 0.7153,
      "step": 50729
    },
    {
      "epoch": 4.364621870429321,
      "grad_norm": 1.0390625,
      "learning_rate": 3.935113957219361e-05,
      "loss": 1.0064,
      "step": 50730
    },
    {
      "epoch": 4.364707906736643,
      "grad_norm": 1.3515625,
      "learning_rate": 3.934062524072074e-05,
      "loss": 0.9614,
      "step": 50731
    },
    {
      "epoch": 4.364793943043964,
      "grad_norm": 1.5390625,
      "learning_rate": 3.9330112256586945e-05,
      "loss": 0.8783,
      "step": 50732
    },
    {
      "epoch": 4.3648799793512865,
      "grad_norm": 1.625,
      "learning_rate": 3.931960061982309e-05,
      "loss": 0.8136,
      "step": 50733
    },
    {
      "epoch": 4.364966015658608,
      "grad_norm": 11.5625,
      "learning_rate": 3.9309090330459705e-05,
      "loss": 0.9524,
      "step": 50734
    },
    {
      "epoch": 4.36505205196593,
      "grad_norm": 1.078125,
      "learning_rate": 3.92985813885276e-05,
      "loss": 1.9842,
      "step": 50735
    },
    {
      "epoch": 4.365138088273252,
      "grad_norm": 1.0390625,
      "learning_rate": 3.928807379405763e-05,
      "loss": 0.6934,
      "step": 50736
    },
    {
      "epoch": 4.365224124580573,
      "grad_norm": 1.7890625,
      "learning_rate": 3.9277567547080394e-05,
      "loss": 1.3537,
      "step": 50737
    },
    {
      "epoch": 4.365310160887895,
      "grad_norm": 0.87109375,
      "learning_rate": 3.926706264762675e-05,
      "loss": 0.4568,
      "step": 50738
    },
    {
      "epoch": 4.365396197195216,
      "grad_norm": 266.0,
      "learning_rate": 3.9256559095727226e-05,
      "loss": 1.0235,
      "step": 50739
    },
    {
      "epoch": 4.365482233502538,
      "grad_norm": 1.3203125,
      "learning_rate": 3.924605689141269e-05,
      "loss": 1.8117,
      "step": 50740
    },
    {
      "epoch": 4.3655682698098595,
      "grad_norm": 0.94140625,
      "learning_rate": 3.923555603471385e-05,
      "loss": 1.1001,
      "step": 50741
    },
    {
      "epoch": 4.3656543061171815,
      "grad_norm": 1.28125,
      "learning_rate": 3.922505652566139e-05,
      "loss": 1.163,
      "step": 50742
    },
    {
      "epoch": 4.365740342424504,
      "grad_norm": 1.1171875,
      "learning_rate": 3.921455836428595e-05,
      "loss": 0.7323,
      "step": 50743
    },
    {
      "epoch": 4.365826378731825,
      "grad_norm": 1.1015625,
      "learning_rate": 3.920406155061834e-05,
      "loss": 1.2982,
      "step": 50744
    },
    {
      "epoch": 4.365912415039147,
      "grad_norm": 0.94921875,
      "learning_rate": 3.919356608468927e-05,
      "loss": 0.8927,
      "step": 50745
    },
    {
      "epoch": 4.365998451346468,
      "grad_norm": 42.25,
      "learning_rate": 3.918307196652943e-05,
      "loss": 0.9975,
      "step": 50746
    },
    {
      "epoch": 4.36608448765379,
      "grad_norm": 0.9375,
      "learning_rate": 3.91725791961694e-05,
      "loss": 1.3316,
      "step": 50747
    },
    {
      "epoch": 4.366170523961111,
      "grad_norm": 1.515625,
      "learning_rate": 3.916208777363994e-05,
      "loss": 1.7468,
      "step": 50748
    },
    {
      "epoch": 4.366256560268433,
      "grad_norm": 8.0,
      "learning_rate": 3.9151597698971864e-05,
      "loss": 1.3494,
      "step": 50749
    },
    {
      "epoch": 4.366342596575755,
      "grad_norm": 1.203125,
      "learning_rate": 3.914110897219564e-05,
      "loss": 0.8651,
      "step": 50750
    },
    {
      "epoch": 4.366428632883077,
      "grad_norm": 0.98828125,
      "learning_rate": 3.913062159334207e-05,
      "loss": 0.5277,
      "step": 50751
    },
    {
      "epoch": 4.366514669190399,
      "grad_norm": 1.171875,
      "learning_rate": 3.912013556244182e-05,
      "loss": 0.885,
      "step": 50752
    },
    {
      "epoch": 4.36660070549772,
      "grad_norm": 1.28125,
      "learning_rate": 3.91096508795255e-05,
      "loss": 1.2411,
      "step": 50753
    },
    {
      "epoch": 4.366686741805042,
      "grad_norm": 1.2265625,
      "learning_rate": 3.909916754462389e-05,
      "loss": 1.176,
      "step": 50754
    },
    {
      "epoch": 4.366772778112363,
      "grad_norm": 302.0,
      "learning_rate": 3.908868555776746e-05,
      "loss": 1.08,
      "step": 50755
    },
    {
      "epoch": 4.366858814419685,
      "grad_norm": 0.875,
      "learning_rate": 3.9078204918987006e-05,
      "loss": 0.6287,
      "step": 50756
    },
    {
      "epoch": 4.366944850727007,
      "grad_norm": 0.94140625,
      "learning_rate": 3.906772562831323e-05,
      "loss": 0.7599,
      "step": 50757
    },
    {
      "epoch": 4.367030887034328,
      "grad_norm": 141.0,
      "learning_rate": 3.905724768577662e-05,
      "loss": 1.4608,
      "step": 50758
    },
    {
      "epoch": 4.3671169233416505,
      "grad_norm": 1.140625,
      "learning_rate": 3.9046771091407975e-05,
      "loss": 1.6828,
      "step": 50759
    },
    {
      "epoch": 4.367202959648972,
      "grad_norm": 0.8359375,
      "learning_rate": 3.903629584523788e-05,
      "loss": 0.8299,
      "step": 50760
    },
    {
      "epoch": 4.367288995956294,
      "grad_norm": 1.53125,
      "learning_rate": 3.902582194729693e-05,
      "loss": 1.3928,
      "step": 50761
    },
    {
      "epoch": 4.367375032263615,
      "grad_norm": 1.359375,
      "learning_rate": 3.901534939761575e-05,
      "loss": 1.2421,
      "step": 50762
    },
    {
      "epoch": 4.367461068570937,
      "grad_norm": 4.71875,
      "learning_rate": 3.900487819622517e-05,
      "loss": 1.1933,
      "step": 50763
    },
    {
      "epoch": 4.367547104878259,
      "grad_norm": 76.5,
      "learning_rate": 3.899440834315554e-05,
      "loss": 0.621,
      "step": 50764
    },
    {
      "epoch": 4.36763314118558,
      "grad_norm": 106.0,
      "learning_rate": 3.8983939838437567e-05,
      "loss": 1.1246,
      "step": 50765
    },
    {
      "epoch": 4.367719177492902,
      "grad_norm": 1.0859375,
      "learning_rate": 3.897347268210205e-05,
      "loss": 1.244,
      "step": 50766
    },
    {
      "epoch": 4.3678052138002235,
      "grad_norm": 1.09375,
      "learning_rate": 3.896300687417936e-05,
      "loss": 1.103,
      "step": 50767
    },
    {
      "epoch": 4.3678912501075455,
      "grad_norm": 1.4765625,
      "learning_rate": 3.8952542414700255e-05,
      "loss": 0.7977,
      "step": 50768
    },
    {
      "epoch": 4.367977286414867,
      "grad_norm": 2304.0,
      "learning_rate": 3.894207930369526e-05,
      "loss": 1.4504,
      "step": 50769
    },
    {
      "epoch": 4.368063322722189,
      "grad_norm": 52.25,
      "learning_rate": 3.893161754119501e-05,
      "loss": 1.0877,
      "step": 50770
    },
    {
      "epoch": 4.368149359029511,
      "grad_norm": 1.0078125,
      "learning_rate": 3.8921157127230154e-05,
      "loss": 0.7988,
      "step": 50771
    },
    {
      "epoch": 4.368235395336832,
      "grad_norm": 0.984375,
      "learning_rate": 3.891069806183117e-05,
      "loss": 1.0889,
      "step": 50772
    },
    {
      "epoch": 4.368321431644154,
      "grad_norm": 0.89453125,
      "learning_rate": 3.890024034502876e-05,
      "loss": 1.2078,
      "step": 50773
    },
    {
      "epoch": 4.368407467951475,
      "grad_norm": 0.9765625,
      "learning_rate": 3.88897839768535e-05,
      "loss": 1.0442,
      "step": 50774
    },
    {
      "epoch": 4.368493504258797,
      "grad_norm": 0.78125,
      "learning_rate": 3.887932895733587e-05,
      "loss": 0.607,
      "step": 50775
    },
    {
      "epoch": 4.3685795405661185,
      "grad_norm": 1.4140625,
      "learning_rate": 3.886887528650651e-05,
      "loss": 0.8209,
      "step": 50776
    },
    {
      "epoch": 4.368665576873441,
      "grad_norm": 1.25,
      "learning_rate": 3.8858422964396126e-05,
      "loss": 1.4784,
      "step": 50777
    },
    {
      "epoch": 4.368751613180763,
      "grad_norm": 9.0625,
      "learning_rate": 3.884797199103501e-05,
      "loss": 0.7355,
      "step": 50778
    },
    {
      "epoch": 4.368837649488084,
      "grad_norm": 1.203125,
      "learning_rate": 3.883752236645405e-05,
      "loss": 0.9569,
      "step": 50779
    },
    {
      "epoch": 4.368923685795406,
      "grad_norm": 141.0,
      "learning_rate": 3.8827074090683424e-05,
      "loss": 1.1655,
      "step": 50780
    },
    {
      "epoch": 4.369009722102727,
      "grad_norm": 2.671875,
      "learning_rate": 3.881662716375406e-05,
      "loss": 1.699,
      "step": 50781
    },
    {
      "epoch": 4.369095758410049,
      "grad_norm": 0.86328125,
      "learning_rate": 3.880618158569638e-05,
      "loss": 1.048,
      "step": 50782
    },
    {
      "epoch": 4.36918179471737,
      "grad_norm": 0.9375,
      "learning_rate": 3.879573735654085e-05,
      "loss": 0.7671,
      "step": 50783
    },
    {
      "epoch": 4.369267831024692,
      "grad_norm": 1.609375,
      "learning_rate": 3.878529447631823e-05,
      "loss": 0.8546,
      "step": 50784
    },
    {
      "epoch": 4.3693538673320145,
      "grad_norm": 0.93359375,
      "learning_rate": 3.877485294505878e-05,
      "loss": 1.6932,
      "step": 50785
    },
    {
      "epoch": 4.369439903639336,
      "grad_norm": 1.0546875,
      "learning_rate": 3.876441276279319e-05,
      "loss": 1.91,
      "step": 50786
    },
    {
      "epoch": 4.369525939946658,
      "grad_norm": 1.53125,
      "learning_rate": 3.87539739295521e-05,
      "loss": 1.0913,
      "step": 50787
    },
    {
      "epoch": 4.369611976253979,
      "grad_norm": 1.078125,
      "learning_rate": 3.874353644536582e-05,
      "loss": 1.4276,
      "step": 50788
    },
    {
      "epoch": 4.369698012561301,
      "grad_norm": 1.0078125,
      "learning_rate": 3.8733100310265e-05,
      "loss": 0.8567,
      "step": 50789
    },
    {
      "epoch": 4.369784048868622,
      "grad_norm": 0.953125,
      "learning_rate": 3.872266552428022e-05,
      "loss": 0.798,
      "step": 50790
    },
    {
      "epoch": 4.369870085175944,
      "grad_norm": 15.9375,
      "learning_rate": 3.8712232087441846e-05,
      "loss": 1.1904,
      "step": 50791
    },
    {
      "epoch": 4.369956121483266,
      "grad_norm": 1.375,
      "learning_rate": 3.870179999978052e-05,
      "loss": 1.9075,
      "step": 50792
    },
    {
      "epoch": 4.3700421577905875,
      "grad_norm": 0.91796875,
      "learning_rate": 3.8691369261326715e-05,
      "loss": 0.4985,
      "step": 50793
    },
    {
      "epoch": 4.3701281940979095,
      "grad_norm": 31.25,
      "learning_rate": 3.868093987211091e-05,
      "loss": 1.131,
      "step": 50794
    },
    {
      "epoch": 4.370214230405231,
      "grad_norm": 1.4921875,
      "learning_rate": 3.867051183216369e-05,
      "loss": 0.9333,
      "step": 50795
    },
    {
      "epoch": 4.370300266712553,
      "grad_norm": 1.2578125,
      "learning_rate": 3.866008514151548e-05,
      "loss": 0.8676,
      "step": 50796
    },
    {
      "epoch": 4.370386303019874,
      "grad_norm": 1.1328125,
      "learning_rate": 3.8649659800196746e-05,
      "loss": 1.1447,
      "step": 50797
    },
    {
      "epoch": 4.370472339327196,
      "grad_norm": 1.109375,
      "learning_rate": 3.863923580823808e-05,
      "loss": 0.7381,
      "step": 50798
    },
    {
      "epoch": 4.370558375634518,
      "grad_norm": 0.8984375,
      "learning_rate": 3.8628813165669854e-05,
      "loss": 0.5336,
      "step": 50799
    },
    {
      "epoch": 4.370644411941839,
      "grad_norm": 1.203125,
      "learning_rate": 3.861839187252264e-05,
      "loss": 0.9087,
      "step": 50800
    },
    {
      "epoch": 4.370730448249161,
      "grad_norm": 0.88671875,
      "learning_rate": 3.8607971928826976e-05,
      "loss": 0.7315,
      "step": 50801
    },
    {
      "epoch": 4.3708164845564825,
      "grad_norm": 1.2109375,
      "learning_rate": 3.859755333461318e-05,
      "loss": 0.6711,
      "step": 50802
    },
    {
      "epoch": 4.370902520863805,
      "grad_norm": 1.2265625,
      "learning_rate": 3.858713608991182e-05,
      "loss": 0.8273,
      "step": 50803
    },
    {
      "epoch": 4.370988557171126,
      "grad_norm": 1.140625,
      "learning_rate": 3.857672019475339e-05,
      "loss": 0.7092,
      "step": 50804
    },
    {
      "epoch": 4.371074593478448,
      "grad_norm": 0.828125,
      "learning_rate": 3.8566305649168244e-05,
      "loss": 0.7563,
      "step": 50805
    },
    {
      "epoch": 4.37116062978577,
      "grad_norm": 14.8125,
      "learning_rate": 3.855589245318691e-05,
      "loss": 1.0901,
      "step": 50806
    },
    {
      "epoch": 4.371246666093091,
      "grad_norm": 1.2265625,
      "learning_rate": 3.8545480606839934e-05,
      "loss": 0.5493,
      "step": 50807
    },
    {
      "epoch": 4.371332702400413,
      "grad_norm": 1.0546875,
      "learning_rate": 3.853507011015756e-05,
      "loss": 1.4101,
      "step": 50808
    },
    {
      "epoch": 4.371418738707734,
      "grad_norm": 1.2265625,
      "learning_rate": 3.852466096317048e-05,
      "loss": 0.9425,
      "step": 50809
    },
    {
      "epoch": 4.371504775015056,
      "grad_norm": 0.984375,
      "learning_rate": 3.8514253165908895e-05,
      "loss": 0.6973,
      "step": 50810
    },
    {
      "epoch": 4.371590811322378,
      "grad_norm": 9.625,
      "learning_rate": 3.85038467184034e-05,
      "loss": 0.7618,
      "step": 50811
    },
    {
      "epoch": 4.3716768476297,
      "grad_norm": 1.34375,
      "learning_rate": 3.8493441620684465e-05,
      "loss": 0.7027,
      "step": 50812
    },
    {
      "epoch": 4.371762883937022,
      "grad_norm": 1.03125,
      "learning_rate": 3.84830378727824e-05,
      "loss": 0.8263,
      "step": 50813
    },
    {
      "epoch": 4.371848920244343,
      "grad_norm": 1.109375,
      "learning_rate": 3.8472635474727676e-05,
      "loss": 0.8519,
      "step": 50814
    },
    {
      "epoch": 4.371934956551665,
      "grad_norm": 1.0390625,
      "learning_rate": 3.846223442655078e-05,
      "loss": 1.0758,
      "step": 50815
    },
    {
      "epoch": 4.372020992858986,
      "grad_norm": 1.7578125,
      "learning_rate": 3.845183472828206e-05,
      "loss": 1.3317,
      "step": 50816
    },
    {
      "epoch": 4.372107029166308,
      "grad_norm": 1.171875,
      "learning_rate": 3.84414363799519e-05,
      "loss": 1.0804,
      "step": 50817
    },
    {
      "epoch": 4.372193065473629,
      "grad_norm": 1.265625,
      "learning_rate": 3.8431039381590815e-05,
      "loss": 1.6242,
      "step": 50818
    },
    {
      "epoch": 4.3722791017809515,
      "grad_norm": 1.0859375,
      "learning_rate": 3.8420643733229124e-05,
      "loss": 0.837,
      "step": 50819
    },
    {
      "epoch": 4.3723651380882735,
      "grad_norm": 1.3203125,
      "learning_rate": 3.8410249434897406e-05,
      "loss": 1.2976,
      "step": 50820
    },
    {
      "epoch": 4.372451174395595,
      "grad_norm": 0.80859375,
      "learning_rate": 3.8399856486625806e-05,
      "loss": 0.9499,
      "step": 50821
    },
    {
      "epoch": 4.372537210702917,
      "grad_norm": 1.09375,
      "learning_rate": 3.838946488844486e-05,
      "loss": 0.6274,
      "step": 50822
    },
    {
      "epoch": 4.372623247010238,
      "grad_norm": 26.5,
      "learning_rate": 3.837907464038509e-05,
      "loss": 0.8472,
      "step": 50823
    },
    {
      "epoch": 4.37270928331756,
      "grad_norm": 2.21875,
      "learning_rate": 3.836868574247659e-05,
      "loss": 0.8885,
      "step": 50824
    },
    {
      "epoch": 4.372795319624882,
      "grad_norm": 1.1953125,
      "learning_rate": 3.835829819475001e-05,
      "loss": 1.0665,
      "step": 50825
    },
    {
      "epoch": 4.372881355932203,
      "grad_norm": 1.1640625,
      "learning_rate": 3.834791199723559e-05,
      "loss": 1.7264,
      "step": 50826
    },
    {
      "epoch": 4.372967392239525,
      "grad_norm": 1.1484375,
      "learning_rate": 3.833752714996369e-05,
      "loss": 1.4521,
      "step": 50827
    },
    {
      "epoch": 4.3730534285468465,
      "grad_norm": 1.171875,
      "learning_rate": 3.83271436529648e-05,
      "loss": 1.9857,
      "step": 50828
    },
    {
      "epoch": 4.373139464854169,
      "grad_norm": 0.953125,
      "learning_rate": 3.831676150626917e-05,
      "loss": 0.9533,
      "step": 50829
    },
    {
      "epoch": 4.37322550116149,
      "grad_norm": 5.34375,
      "learning_rate": 3.830638070990722e-05,
      "loss": 1.3807,
      "step": 50830
    },
    {
      "epoch": 4.373311537468812,
      "grad_norm": 1.4921875,
      "learning_rate": 3.829600126390936e-05,
      "loss": 1.2672,
      "step": 50831
    },
    {
      "epoch": 4.373397573776134,
      "grad_norm": 1.4921875,
      "learning_rate": 3.828562316830586e-05,
      "loss": 1.0957,
      "step": 50832
    },
    {
      "epoch": 4.373483610083455,
      "grad_norm": 1.1015625,
      "learning_rate": 3.827524642312713e-05,
      "loss": 0.9042,
      "step": 50833
    },
    {
      "epoch": 4.373569646390777,
      "grad_norm": 8.875,
      "learning_rate": 3.8264871028403536e-05,
      "loss": 0.6966,
      "step": 50834
    },
    {
      "epoch": 4.373655682698098,
      "grad_norm": 1.1015625,
      "learning_rate": 3.825449698416533e-05,
      "loss": 0.9841,
      "step": 50835
    },
    {
      "epoch": 4.37374171900542,
      "grad_norm": 0.9453125,
      "learning_rate": 3.8244124290442937e-05,
      "loss": 0.8036,
      "step": 50836
    },
    {
      "epoch": 4.373827755312742,
      "grad_norm": 1.1875,
      "learning_rate": 3.823375294726672e-05,
      "loss": 0.9213,
      "step": 50837
    },
    {
      "epoch": 4.373913791620064,
      "grad_norm": 1.1875,
      "learning_rate": 3.822338295466693e-05,
      "loss": 1.1508,
      "step": 50838
    },
    {
      "epoch": 4.373999827927386,
      "grad_norm": 2.25,
      "learning_rate": 3.8213014312673986e-05,
      "loss": 1.2245,
      "step": 50839
    },
    {
      "epoch": 4.374085864234707,
      "grad_norm": 1.3203125,
      "learning_rate": 3.820264702131809e-05,
      "loss": 1.1238,
      "step": 50840
    },
    {
      "epoch": 4.374171900542029,
      "grad_norm": 1.6953125,
      "learning_rate": 3.8192281080629607e-05,
      "loss": 1.4909,
      "step": 50841
    },
    {
      "epoch": 4.37425793684935,
      "grad_norm": 1.1953125,
      "learning_rate": 3.8181916490639015e-05,
      "loss": 0.8416,
      "step": 50842
    },
    {
      "epoch": 4.374343973156672,
      "grad_norm": 1.3984375,
      "learning_rate": 3.817155325137639e-05,
      "loss": 1.776,
      "step": 50843
    },
    {
      "epoch": 4.374430009463993,
      "grad_norm": 1.59375,
      "learning_rate": 3.8161191362872105e-05,
      "loss": 1.3887,
      "step": 50844
    },
    {
      "epoch": 4.3745160457713155,
      "grad_norm": 2.71875,
      "learning_rate": 3.815083082515669e-05,
      "loss": 1.1657,
      "step": 50845
    },
    {
      "epoch": 4.3746020820786375,
      "grad_norm": 62.5,
      "learning_rate": 3.814047163826012e-05,
      "loss": 0.868,
      "step": 50846
    },
    {
      "epoch": 4.374688118385959,
      "grad_norm": 0.9609375,
      "learning_rate": 3.8130113802212875e-05,
      "loss": 0.6314,
      "step": 50847
    },
    {
      "epoch": 4.374774154693281,
      "grad_norm": 1.171875,
      "learning_rate": 3.8119757317045255e-05,
      "loss": 0.7449,
      "step": 50848
    },
    {
      "epoch": 4.374860191000602,
      "grad_norm": 0.82421875,
      "learning_rate": 3.8109402182787464e-05,
      "loss": 0.4094,
      "step": 50849
    },
    {
      "epoch": 4.374946227307924,
      "grad_norm": 0.98828125,
      "learning_rate": 3.809904839946993e-05,
      "loss": 1.2043,
      "step": 50850
    },
    {
      "epoch": 4.375032263615245,
      "grad_norm": 2.4375,
      "learning_rate": 3.808869596712278e-05,
      "loss": 0.8382,
      "step": 50851
    },
    {
      "epoch": 4.375118299922567,
      "grad_norm": 38.0,
      "learning_rate": 3.807834488577633e-05,
      "loss": 1.0697,
      "step": 50852
    },
    {
      "epoch": 4.375204336229889,
      "grad_norm": 0.9453125,
      "learning_rate": 3.8067995155460946e-05,
      "loss": 0.6255,
      "step": 50853
    },
    {
      "epoch": 4.3752903725372105,
      "grad_norm": 1.0234375,
      "learning_rate": 3.8057646776206714e-05,
      "loss": 0.7818,
      "step": 50854
    },
    {
      "epoch": 4.375376408844533,
      "grad_norm": 1.4921875,
      "learning_rate": 3.804729974804416e-05,
      "loss": 1.1445,
      "step": 50855
    },
    {
      "epoch": 4.375462445151854,
      "grad_norm": 1.125,
      "learning_rate": 3.8036954071003326e-05,
      "loss": 0.9153,
      "step": 50856
    },
    {
      "epoch": 4.375548481459176,
      "grad_norm": 15.125,
      "learning_rate": 3.802660974511457e-05,
      "loss": 1.1571,
      "step": 50857
    },
    {
      "epoch": 4.375634517766498,
      "grad_norm": 1.046875,
      "learning_rate": 3.80162667704082e-05,
      "loss": 0.6501,
      "step": 50858
    },
    {
      "epoch": 4.375720554073819,
      "grad_norm": 1.25,
      "learning_rate": 3.800592514691431e-05,
      "loss": 1.4768,
      "step": 50859
    },
    {
      "epoch": 4.375806590381141,
      "grad_norm": 1.046875,
      "learning_rate": 3.799558487466326e-05,
      "loss": 1.1934,
      "step": 50860
    },
    {
      "epoch": 4.375892626688462,
      "grad_norm": 1.0546875,
      "learning_rate": 3.798524595368535e-05,
      "loss": 0.7914,
      "step": 50861
    },
    {
      "epoch": 4.375978662995784,
      "grad_norm": 0.87109375,
      "learning_rate": 3.797490838401063e-05,
      "loss": 1.0276,
      "step": 50862
    },
    {
      "epoch": 4.376064699303106,
      "grad_norm": 0.796875,
      "learning_rate": 3.7964572165669455e-05,
      "loss": 0.5303,
      "step": 50863
    },
    {
      "epoch": 4.376150735610428,
      "grad_norm": 1.203125,
      "learning_rate": 3.7954237298692126e-05,
      "loss": 1.5645,
      "step": 50864
    },
    {
      "epoch": 4.37623677191775,
      "grad_norm": 1.3359375,
      "learning_rate": 3.794390378310875e-05,
      "loss": 1.4661,
      "step": 50865
    },
    {
      "epoch": 4.376322808225071,
      "grad_norm": 1.0859375,
      "learning_rate": 3.7933571618949614e-05,
      "loss": 0.8984,
      "step": 50866
    },
    {
      "epoch": 4.376408844532393,
      "grad_norm": 1.2734375,
      "learning_rate": 3.7923240806244874e-05,
      "loss": 1.4889,
      "step": 50867
    },
    {
      "epoch": 4.376494880839714,
      "grad_norm": 1.0625,
      "learning_rate": 3.791291134502478e-05,
      "loss": 0.7972,
      "step": 50868
    },
    {
      "epoch": 4.376580917147036,
      "grad_norm": 1.1875,
      "learning_rate": 3.790258323531959e-05,
      "loss": 0.7714,
      "step": 50869
    },
    {
      "epoch": 4.376666953454357,
      "grad_norm": 1.046875,
      "learning_rate": 3.789225647715944e-05,
      "loss": 1.9726,
      "step": 50870
    },
    {
      "epoch": 4.3767529897616795,
      "grad_norm": 0.9609375,
      "learning_rate": 3.7881931070574526e-05,
      "loss": 1.0422,
      "step": 50871
    },
    {
      "epoch": 4.3768390260690015,
      "grad_norm": 1.125,
      "learning_rate": 3.787160701559517e-05,
      "loss": 1.2332,
      "step": 50872
    },
    {
      "epoch": 4.376925062376323,
      "grad_norm": 1.4296875,
      "learning_rate": 3.7861284312251444e-05,
      "loss": 1.0118,
      "step": 50873
    },
    {
      "epoch": 4.377011098683645,
      "grad_norm": 1.28125,
      "learning_rate": 3.7850962960573557e-05,
      "loss": 1.1348,
      "step": 50874
    },
    {
      "epoch": 4.377097134990966,
      "grad_norm": 1.0625,
      "learning_rate": 3.784064296059181e-05,
      "loss": 1.0057,
      "step": 50875
    },
    {
      "epoch": 4.377183171298288,
      "grad_norm": 9.375,
      "learning_rate": 3.7830324312336184e-05,
      "loss": 0.9727,
      "step": 50876
    },
    {
      "epoch": 4.377269207605609,
      "grad_norm": 0.98828125,
      "learning_rate": 3.782000701583699e-05,
      "loss": 1.3423,
      "step": 50877
    },
    {
      "epoch": 4.377355243912931,
      "grad_norm": 1.5859375,
      "learning_rate": 3.7809691071124476e-05,
      "loss": 1.239,
      "step": 50878
    },
    {
      "epoch": 4.377441280220253,
      "grad_norm": 1.2578125,
      "learning_rate": 3.779937647822862e-05,
      "loss": 1.7306,
      "step": 50879
    },
    {
      "epoch": 4.3775273165275745,
      "grad_norm": 1.390625,
      "learning_rate": 3.778906323717979e-05,
      "loss": 1.3334,
      "step": 50880
    },
    {
      "epoch": 4.377613352834897,
      "grad_norm": 1.1640625,
      "learning_rate": 3.777875134800801e-05,
      "loss": 1.0321,
      "step": 50881
    },
    {
      "epoch": 4.377699389142218,
      "grad_norm": 2.234375,
      "learning_rate": 3.776844081074343e-05,
      "loss": 1.1431,
      "step": 50882
    },
    {
      "epoch": 4.37778542544954,
      "grad_norm": 1.328125,
      "learning_rate": 3.7758131625416357e-05,
      "loss": 1.8062,
      "step": 50883
    },
    {
      "epoch": 4.377871461756861,
      "grad_norm": 1.421875,
      "learning_rate": 3.774782379205677e-05,
      "loss": 1.0576,
      "step": 50884
    },
    {
      "epoch": 4.377957498064183,
      "grad_norm": 1.1796875,
      "learning_rate": 3.773751731069491e-05,
      "loss": 1.2481,
      "step": 50885
    },
    {
      "epoch": 4.378043534371505,
      "grad_norm": 0.84375,
      "learning_rate": 3.7727212181360996e-05,
      "loss": 0.7902,
      "step": 50886
    },
    {
      "epoch": 4.378129570678826,
      "grad_norm": 2.234375,
      "learning_rate": 3.771690840408498e-05,
      "loss": 0.7194,
      "step": 50887
    },
    {
      "epoch": 4.378215606986148,
      "grad_norm": 3.8125,
      "learning_rate": 3.770660597889708e-05,
      "loss": 0.9748,
      "step": 50888
    },
    {
      "epoch": 4.37830164329347,
      "grad_norm": 1.1640625,
      "learning_rate": 3.769630490582754e-05,
      "loss": 1.2537,
      "step": 50889
    },
    {
      "epoch": 4.378387679600792,
      "grad_norm": 1.5859375,
      "learning_rate": 3.7686005184906335e-05,
      "loss": 1.1087,
      "step": 50890
    },
    {
      "epoch": 4.378473715908113,
      "grad_norm": 1.0859375,
      "learning_rate": 3.767570681616361e-05,
      "loss": 1.6148,
      "step": 50891
    },
    {
      "epoch": 4.378559752215435,
      "grad_norm": 1.0390625,
      "learning_rate": 3.7665409799629566e-05,
      "loss": 1.1514,
      "step": 50892
    },
    {
      "epoch": 4.378645788522757,
      "grad_norm": 1.0859375,
      "learning_rate": 3.765511413533429e-05,
      "loss": 1.5432,
      "step": 50893
    },
    {
      "epoch": 4.378731824830078,
      "grad_norm": 0.9375,
      "learning_rate": 3.764481982330792e-05,
      "loss": 0.5146,
      "step": 50894
    },
    {
      "epoch": 4.3788178611374,
      "grad_norm": 0.98046875,
      "learning_rate": 3.7634526863580486e-05,
      "loss": 0.8435,
      "step": 50895
    },
    {
      "epoch": 4.378903897444721,
      "grad_norm": 1.59375,
      "learning_rate": 3.7624235256182195e-05,
      "loss": 1.2137,
      "step": 50896
    },
    {
      "epoch": 4.3789899337520435,
      "grad_norm": 2.859375,
      "learning_rate": 3.761394500114301e-05,
      "loss": 1.5278,
      "step": 50897
    },
    {
      "epoch": 4.379075970059365,
      "grad_norm": 1.453125,
      "learning_rate": 3.7603656098493086e-05,
      "loss": 1.0811,
      "step": 50898
    },
    {
      "epoch": 4.379162006366687,
      "grad_norm": 1.375,
      "learning_rate": 3.759336854826267e-05,
      "loss": 1.6961,
      "step": 50899
    },
    {
      "epoch": 4.379248042674009,
      "grad_norm": 0.85546875,
      "learning_rate": 3.758308235048158e-05,
      "loss": 1.2116,
      "step": 50900
    },
    {
      "epoch": 4.37933407898133,
      "grad_norm": 1.0234375,
      "learning_rate": 3.757279750518006e-05,
      "loss": 0.8518,
      "step": 50901
    },
    {
      "epoch": 4.379420115288652,
      "grad_norm": 1.1484375,
      "learning_rate": 3.7562514012388246e-05,
      "loss": 0.8983,
      "step": 50902
    },
    {
      "epoch": 4.379506151595973,
      "grad_norm": 1.0234375,
      "learning_rate": 3.755223187213608e-05,
      "loss": 1.0261,
      "step": 50903
    },
    {
      "epoch": 4.379592187903295,
      "grad_norm": 1.6796875,
      "learning_rate": 3.7541951084453684e-05,
      "loss": 1.0479,
      "step": 50904
    },
    {
      "epoch": 4.3796782242106165,
      "grad_norm": 1.1953125,
      "learning_rate": 3.7531671649371215e-05,
      "loss": 1.1359,
      "step": 50905
    },
    {
      "epoch": 4.3797642605179385,
      "grad_norm": 671744.0,
      "learning_rate": 3.752139356691858e-05,
      "loss": 1.2738,
      "step": 50906
    },
    {
      "epoch": 4.379850296825261,
      "grad_norm": 1.4921875,
      "learning_rate": 3.751111683712605e-05,
      "loss": 1.5805,
      "step": 50907
    },
    {
      "epoch": 4.379936333132582,
      "grad_norm": 1.2578125,
      "learning_rate": 3.7500841460023415e-05,
      "loss": 1.0954,
      "step": 50908
    },
    {
      "epoch": 4.380022369439904,
      "grad_norm": 1.4140625,
      "learning_rate": 3.749056743564089e-05,
      "loss": 1.8282,
      "step": 50909
    },
    {
      "epoch": 4.380108405747225,
      "grad_norm": 3.4375,
      "learning_rate": 3.748029476400855e-05,
      "loss": 0.5849,
      "step": 50910
    },
    {
      "epoch": 4.380194442054547,
      "grad_norm": 1.1328125,
      "learning_rate": 3.747002344515632e-05,
      "loss": 0.8061,
      "step": 50911
    },
    {
      "epoch": 4.380280478361868,
      "grad_norm": 31.5,
      "learning_rate": 3.745975347911434e-05,
      "loss": 1.9255,
      "step": 50912
    },
    {
      "epoch": 4.38036651466919,
      "grad_norm": 1.1015625,
      "learning_rate": 3.7449484865912706e-05,
      "loss": 1.078,
      "step": 50913
    },
    {
      "epoch": 4.380452550976512,
      "grad_norm": 0.96484375,
      "learning_rate": 3.743921760558128e-05,
      "loss": 1.2469,
      "step": 50914
    },
    {
      "epoch": 4.380538587283834,
      "grad_norm": 1.359375,
      "learning_rate": 3.742895169815014e-05,
      "loss": 1.583,
      "step": 50915
    },
    {
      "epoch": 4.380624623591156,
      "grad_norm": 36.75,
      "learning_rate": 3.741868714364949e-05,
      "loss": 0.7196,
      "step": 50916
    },
    {
      "epoch": 4.380710659898477,
      "grad_norm": 3.46875,
      "learning_rate": 3.740842394210914e-05,
      "loss": 2.0571,
      "step": 50917
    },
    {
      "epoch": 4.380796696205799,
      "grad_norm": 1.4296875,
      "learning_rate": 3.739816209355912e-05,
      "loss": 1.4111,
      "step": 50918
    },
    {
      "epoch": 4.38088273251312,
      "grad_norm": 7.0625,
      "learning_rate": 3.7387901598029626e-05,
      "loss": 0.9541,
      "step": 50919
    },
    {
      "epoch": 4.380968768820442,
      "grad_norm": 1.5,
      "learning_rate": 3.737764245555042e-05,
      "loss": 1.7523,
      "step": 50920
    },
    {
      "epoch": 4.381054805127764,
      "grad_norm": 8.25,
      "learning_rate": 3.736738466615175e-05,
      "loss": 0.9233,
      "step": 50921
    },
    {
      "epoch": 4.381140841435085,
      "grad_norm": 1.03125,
      "learning_rate": 3.735712822986342e-05,
      "loss": 1.1104,
      "step": 50922
    },
    {
      "epoch": 4.3812268777424075,
      "grad_norm": 1.1015625,
      "learning_rate": 3.734687314671553e-05,
      "loss": 0.6677,
      "step": 50923
    },
    {
      "epoch": 4.381312914049729,
      "grad_norm": 4.34375,
      "learning_rate": 3.7336619416738106e-05,
      "loss": 1.8451,
      "step": 50924
    },
    {
      "epoch": 4.381398950357051,
      "grad_norm": 1.3515625,
      "learning_rate": 3.732636703996106e-05,
      "loss": 1.2651,
      "step": 50925
    },
    {
      "epoch": 4.381484986664372,
      "grad_norm": 1.5703125,
      "learning_rate": 3.731611601641432e-05,
      "loss": 0.9383,
      "step": 50926
    },
    {
      "epoch": 4.381571022971694,
      "grad_norm": 1.296875,
      "learning_rate": 3.730586634612809e-05,
      "loss": 1.3667,
      "step": 50927
    },
    {
      "epoch": 4.381657059279016,
      "grad_norm": 1.0703125,
      "learning_rate": 3.7295618029132064e-05,
      "loss": 1.254,
      "step": 50928
    },
    {
      "epoch": 4.381743095586337,
      "grad_norm": 0.90625,
      "learning_rate": 3.7285371065456485e-05,
      "loss": 1.2161,
      "step": 50929
    },
    {
      "epoch": 4.381829131893659,
      "grad_norm": 1.140625,
      "learning_rate": 3.727512545513112e-05,
      "loss": 1.4563,
      "step": 50930
    },
    {
      "epoch": 4.3819151682009805,
      "grad_norm": 1.1953125,
      "learning_rate": 3.726488119818605e-05,
      "loss": 1.5457,
      "step": 50931
    },
    {
      "epoch": 4.3820012045083025,
      "grad_norm": 1.1171875,
      "learning_rate": 3.725463829465131e-05,
      "loss": 1.3829,
      "step": 50932
    },
    {
      "epoch": 4.382087240815624,
      "grad_norm": 1.0859375,
      "learning_rate": 3.72443967445566e-05,
      "loss": 1.7291,
      "step": 50933
    },
    {
      "epoch": 4.382173277122946,
      "grad_norm": 1.0703125,
      "learning_rate": 3.723415654793205e-05,
      "loss": 0.9046,
      "step": 50934
    },
    {
      "epoch": 4.382259313430268,
      "grad_norm": 12.9375,
      "learning_rate": 3.722391770480771e-05,
      "loss": 1.0578,
      "step": 50935
    },
    {
      "epoch": 4.382345349737589,
      "grad_norm": 1.0703125,
      "learning_rate": 3.7213680215213316e-05,
      "loss": 1.3367,
      "step": 50936
    },
    {
      "epoch": 4.382431386044911,
      "grad_norm": 2.453125,
      "learning_rate": 3.720344407917897e-05,
      "loss": 1.2744,
      "step": 50937
    },
    {
      "epoch": 4.382517422352232,
      "grad_norm": 1.1484375,
      "learning_rate": 3.719320929673442e-05,
      "loss": 0.973,
      "step": 50938
    },
    {
      "epoch": 4.382603458659554,
      "grad_norm": 0.94140625,
      "learning_rate": 3.718297586790975e-05,
      "loss": 0.96,
      "step": 50939
    },
    {
      "epoch": 4.382689494966876,
      "grad_norm": 1.28125,
      "learning_rate": 3.7172743792734945e-05,
      "loss": 1.9181,
      "step": 50940
    },
    {
      "epoch": 4.382775531274198,
      "grad_norm": 0.98828125,
      "learning_rate": 3.71625130712398e-05,
      "loss": 0.4295,
      "step": 50941
    },
    {
      "epoch": 4.38286156758152,
      "grad_norm": 1.21875,
      "learning_rate": 3.715228370345425e-05,
      "loss": 1.035,
      "step": 50942
    },
    {
      "epoch": 4.382947603888841,
      "grad_norm": 1.484375,
      "learning_rate": 3.714205568940837e-05,
      "loss": 0.8805,
      "step": 50943
    },
    {
      "epoch": 4.383033640196163,
      "grad_norm": 1.140625,
      "learning_rate": 3.7131829029131816e-05,
      "loss": 1.2835,
      "step": 50944
    },
    {
      "epoch": 4.383119676503484,
      "grad_norm": 1.2734375,
      "learning_rate": 3.7121603722654665e-05,
      "loss": 1.6536,
      "step": 50945
    },
    {
      "epoch": 4.383205712810806,
      "grad_norm": 0.8515625,
      "learning_rate": 3.711137977000689e-05,
      "loss": 0.5792,
      "step": 50946
    },
    {
      "epoch": 4.383291749118128,
      "grad_norm": 1.15625,
      "learning_rate": 3.71011571712182e-05,
      "loss": 1.6984,
      "step": 50947
    },
    {
      "epoch": 4.383377785425449,
      "grad_norm": 1.40625,
      "learning_rate": 3.709093592631874e-05,
      "loss": 1.1778,
      "step": 50948
    },
    {
      "epoch": 4.3834638217327715,
      "grad_norm": 1.2421875,
      "learning_rate": 3.7080716035338136e-05,
      "loss": 0.636,
      "step": 50949
    },
    {
      "epoch": 4.383549858040093,
      "grad_norm": 0.9609375,
      "learning_rate": 3.7070497498306375e-05,
      "loss": 1.5031,
      "step": 50950
    },
    {
      "epoch": 4.383635894347415,
      "grad_norm": 1.046875,
      "learning_rate": 3.706028031525349e-05,
      "loss": 1.6548,
      "step": 50951
    },
    {
      "epoch": 4.383721930654736,
      "grad_norm": 1.0546875,
      "learning_rate": 3.705006448620918e-05,
      "loss": 0.6278,
      "step": 50952
    },
    {
      "epoch": 4.383807966962058,
      "grad_norm": 1.28125,
      "learning_rate": 3.703985001120336e-05,
      "loss": 1.9374,
      "step": 50953
    },
    {
      "epoch": 4.38389400326938,
      "grad_norm": 1.4375,
      "learning_rate": 3.702963689026606e-05,
      "loss": 1.1295,
      "step": 50954
    },
    {
      "epoch": 4.383980039576701,
      "grad_norm": 1.0390625,
      "learning_rate": 3.701942512342688e-05,
      "loss": 1.2984,
      "step": 50955
    },
    {
      "epoch": 4.384066075884023,
      "grad_norm": 0.87890625,
      "learning_rate": 3.7009214710715886e-05,
      "loss": 1.0415,
      "step": 50956
    },
    {
      "epoch": 4.3841521121913445,
      "grad_norm": 0.9375,
      "learning_rate": 3.6999005652162965e-05,
      "loss": 0.9164,
      "step": 50957
    },
    {
      "epoch": 4.3842381484986666,
      "grad_norm": 0.8515625,
      "learning_rate": 3.6988797947797856e-05,
      "loss": 0.5723,
      "step": 50958
    },
    {
      "epoch": 4.384324184805988,
      "grad_norm": 0.93359375,
      "learning_rate": 3.697859159765038e-05,
      "loss": 0.9626,
      "step": 50959
    },
    {
      "epoch": 4.38441022111331,
      "grad_norm": 1.203125,
      "learning_rate": 3.6968386601750615e-05,
      "loss": 0.9973,
      "step": 50960
    },
    {
      "epoch": 4.384496257420632,
      "grad_norm": 1.3828125,
      "learning_rate": 3.695818296012821e-05,
      "loss": 0.5832,
      "step": 50961
    },
    {
      "epoch": 4.384582293727953,
      "grad_norm": 2.265625,
      "learning_rate": 3.6947980672813074e-05,
      "loss": 0.9124,
      "step": 50962
    },
    {
      "epoch": 4.384668330035275,
      "grad_norm": 0.9609375,
      "learning_rate": 3.693777973983498e-05,
      "loss": 1.2606,
      "step": 50963
    },
    {
      "epoch": 4.384754366342596,
      "grad_norm": 1.84375,
      "learning_rate": 3.6927580161223825e-05,
      "loss": 1.1637,
      "step": 50964
    },
    {
      "epoch": 4.384840402649918,
      "grad_norm": 1.078125,
      "learning_rate": 3.6917381937009444e-05,
      "loss": 1.2056,
      "step": 50965
    },
    {
      "epoch": 4.3849264389572395,
      "grad_norm": 1.234375,
      "learning_rate": 3.690718506722163e-05,
      "loss": 1.1602,
      "step": 50966
    },
    {
      "epoch": 4.385012475264562,
      "grad_norm": 0.890625,
      "learning_rate": 3.689698955189036e-05,
      "loss": 0.6964,
      "step": 50967
    },
    {
      "epoch": 4.385098511571884,
      "grad_norm": 51.5,
      "learning_rate": 3.688679539104522e-05,
      "loss": 1.0447,
      "step": 50968
    },
    {
      "epoch": 4.385184547879205,
      "grad_norm": 90.0,
      "learning_rate": 3.687660258471609e-05,
      "loss": 1.6224,
      "step": 50969
    },
    {
      "epoch": 4.385270584186527,
      "grad_norm": 0.98046875,
      "learning_rate": 3.6866411132932985e-05,
      "loss": 0.8029,
      "step": 50970
    },
    {
      "epoch": 4.385356620493848,
      "grad_norm": 1.8359375,
      "learning_rate": 3.685622103572539e-05,
      "loss": 1.4537,
      "step": 50971
    },
    {
      "epoch": 4.38544265680117,
      "grad_norm": 29.0,
      "learning_rate": 3.6846032293123334e-05,
      "loss": 1.138,
      "step": 50972
    },
    {
      "epoch": 4.385528693108492,
      "grad_norm": 1.1953125,
      "learning_rate": 3.6835844905156624e-05,
      "loss": 1.0879,
      "step": 50973
    },
    {
      "epoch": 4.385614729415813,
      "grad_norm": 1.046875,
      "learning_rate": 3.68256588718549e-05,
      "loss": 1.5482,
      "step": 50974
    },
    {
      "epoch": 4.3857007657231355,
      "grad_norm": 1.109375,
      "learning_rate": 3.681547419324804e-05,
      "loss": 0.7076,
      "step": 50975
    },
    {
      "epoch": 4.385786802030457,
      "grad_norm": 6.4375,
      "learning_rate": 3.6805290869365903e-05,
      "loss": 1.1085,
      "step": 50976
    },
    {
      "epoch": 4.385872838337779,
      "grad_norm": 1.109375,
      "learning_rate": 3.679510890023813e-05,
      "loss": 0.8423,
      "step": 50977
    },
    {
      "epoch": 4.3859588746451,
      "grad_norm": 1.0859375,
      "learning_rate": 3.678492828589469e-05,
      "loss": 1.916,
      "step": 50978
    },
    {
      "epoch": 4.386044910952422,
      "grad_norm": 0.8671875,
      "learning_rate": 3.6774749026365126e-05,
      "loss": 1.4362,
      "step": 50979
    },
    {
      "epoch": 4.386130947259744,
      "grad_norm": 0.90234375,
      "learning_rate": 3.676457112167936e-05,
      "loss": 1.0315,
      "step": 50980
    },
    {
      "epoch": 4.386216983567065,
      "grad_norm": 1.1875,
      "learning_rate": 3.675439457186719e-05,
      "loss": 0.8082,
      "step": 50981
    },
    {
      "epoch": 4.386303019874387,
      "grad_norm": 1.4453125,
      "learning_rate": 3.674421937695821e-05,
      "loss": 1.1609,
      "step": 50982
    },
    {
      "epoch": 4.3863890561817085,
      "grad_norm": 23.375,
      "learning_rate": 3.673404553698234e-05,
      "loss": 0.8944,
      "step": 50983
    },
    {
      "epoch": 4.386475092489031,
      "grad_norm": 1.3515625,
      "learning_rate": 3.6723873051969336e-05,
      "loss": 0.9806,
      "step": 50984
    },
    {
      "epoch": 4.386561128796352,
      "grad_norm": 1.1640625,
      "learning_rate": 3.6713701921948786e-05,
      "loss": 1.7737,
      "step": 50985
    },
    {
      "epoch": 4.386647165103674,
      "grad_norm": 0.91796875,
      "learning_rate": 3.67035321469506e-05,
      "loss": 0.447,
      "step": 50986
    },
    {
      "epoch": 4.386733201410996,
      "grad_norm": 1.078125,
      "learning_rate": 3.6693363727004545e-05,
      "loss": 0.6462,
      "step": 50987
    },
    {
      "epoch": 4.386819237718317,
      "grad_norm": 1.4375,
      "learning_rate": 3.668319666214026e-05,
      "loss": 1.34,
      "step": 50988
    },
    {
      "epoch": 4.386905274025639,
      "grad_norm": 1.234375,
      "learning_rate": 3.667303095238744e-05,
      "loss": 1.0439,
      "step": 50989
    },
    {
      "epoch": 4.38699131033296,
      "grad_norm": 33.75,
      "learning_rate": 3.666286659777601e-05,
      "loss": 1.0308,
      "step": 50990
    },
    {
      "epoch": 4.387077346640282,
      "grad_norm": 0.859375,
      "learning_rate": 3.66527035983355e-05,
      "loss": 1.0741,
      "step": 50991
    },
    {
      "epoch": 4.3871633829476036,
      "grad_norm": 1.09375,
      "learning_rate": 3.664254195409583e-05,
      "loss": 0.9259,
      "step": 50992
    },
    {
      "epoch": 4.387249419254926,
      "grad_norm": 1.2734375,
      "learning_rate": 3.6632381665086475e-05,
      "loss": 1.011,
      "step": 50993
    },
    {
      "epoch": 4.387335455562248,
      "grad_norm": 10.3125,
      "learning_rate": 3.662222273133731e-05,
      "loss": 0.9191,
      "step": 50994
    },
    {
      "epoch": 4.387421491869569,
      "grad_norm": 1.0390625,
      "learning_rate": 3.661206515287807e-05,
      "loss": 0.6122,
      "step": 50995
    },
    {
      "epoch": 4.387507528176891,
      "grad_norm": 1.1015625,
      "learning_rate": 3.660190892973836e-05,
      "loss": 0.9952,
      "step": 50996
    },
    {
      "epoch": 4.387593564484212,
      "grad_norm": 1.125,
      "learning_rate": 3.659175406194792e-05,
      "loss": 1.0981,
      "step": 50997
    },
    {
      "epoch": 4.387679600791534,
      "grad_norm": 1.09375,
      "learning_rate": 3.6581600549536574e-05,
      "loss": 1.3289,
      "step": 50998
    },
    {
      "epoch": 4.387765637098855,
      "grad_norm": 1.375,
      "learning_rate": 3.657144839253384e-05,
      "loss": 0.7018,
      "step": 50999
    },
    {
      "epoch": 4.387851673406177,
      "grad_norm": 1.109375,
      "learning_rate": 3.656129759096949e-05,
      "loss": 1.568,
      "step": 51000
    },
    {
      "epoch": 4.3879377097134995,
      "grad_norm": 1.0625,
      "learning_rate": 3.655114814487326e-05,
      "loss": 0.699,
      "step": 51001
    },
    {
      "epoch": 4.388023746020821,
      "grad_norm": 0.82421875,
      "learning_rate": 3.654100005427469e-05,
      "loss": 0.8515,
      "step": 51002
    },
    {
      "epoch": 4.388109782328143,
      "grad_norm": 1.0859375,
      "learning_rate": 3.653085331920364e-05,
      "loss": 1.2076,
      "step": 51003
    },
    {
      "epoch": 4.388195818635464,
      "grad_norm": 1.34375,
      "learning_rate": 3.652070793968965e-05,
      "loss": 1.6049,
      "step": 51004
    },
    {
      "epoch": 4.388281854942786,
      "grad_norm": 0.734375,
      "learning_rate": 3.651056391576246e-05,
      "loss": 0.4268,
      "step": 51005
    },
    {
      "epoch": 4.388367891250107,
      "grad_norm": 2.875,
      "learning_rate": 3.650042124745179e-05,
      "loss": 0.6605,
      "step": 51006
    },
    {
      "epoch": 4.388453927557429,
      "grad_norm": 76.5,
      "learning_rate": 3.649027993478721e-05,
      "loss": 0.6202,
      "step": 51007
    },
    {
      "epoch": 4.388539963864751,
      "grad_norm": 1.296875,
      "learning_rate": 3.648013997779842e-05,
      "loss": 0.6795,
      "step": 51008
    },
    {
      "epoch": 4.3886260001720725,
      "grad_norm": 3.296875,
      "learning_rate": 3.647000137651502e-05,
      "loss": 1.0368,
      "step": 51009
    },
    {
      "epoch": 4.388712036479395,
      "grad_norm": 1.3359375,
      "learning_rate": 3.6459864130966745e-05,
      "loss": 0.8443,
      "step": 51010
    },
    {
      "epoch": 4.388798072786716,
      "grad_norm": 0.95703125,
      "learning_rate": 3.6449728241183254e-05,
      "loss": 0.5919,
      "step": 51011
    },
    {
      "epoch": 4.388884109094038,
      "grad_norm": 1.4921875,
      "learning_rate": 3.6439593707194127e-05,
      "loss": 1.509,
      "step": 51012
    },
    {
      "epoch": 4.388970145401359,
      "grad_norm": 1.0390625,
      "learning_rate": 3.642946052902896e-05,
      "loss": 0.6865,
      "step": 51013
    },
    {
      "epoch": 4.389056181708681,
      "grad_norm": 0.85546875,
      "learning_rate": 3.6419328706717605e-05,
      "loss": 1.0169,
      "step": 51014
    },
    {
      "epoch": 4.389142218016003,
      "grad_norm": 0.97265625,
      "learning_rate": 3.640919824028943e-05,
      "loss": 1.229,
      "step": 51015
    },
    {
      "epoch": 4.389228254323324,
      "grad_norm": 1.2421875,
      "learning_rate": 3.63990691297742e-05,
      "loss": 1.379,
      "step": 51016
    },
    {
      "epoch": 4.389314290630646,
      "grad_norm": 0.75390625,
      "learning_rate": 3.63889413752016e-05,
      "loss": 0.594,
      "step": 51017
    },
    {
      "epoch": 4.3894003269379676,
      "grad_norm": 1.171875,
      "learning_rate": 3.6378814976601114e-05,
      "loss": 1.4872,
      "step": 51018
    },
    {
      "epoch": 4.38948636324529,
      "grad_norm": 1.0859375,
      "learning_rate": 3.6368689934002495e-05,
      "loss": 0.5097,
      "step": 51019
    },
    {
      "epoch": 4.389572399552611,
      "grad_norm": 1.296875,
      "learning_rate": 3.635856624743522e-05,
      "loss": 0.9276,
      "step": 51020
    },
    {
      "epoch": 4.389658435859933,
      "grad_norm": 1.21875,
      "learning_rate": 3.634844391692893e-05,
      "loss": 0.8298,
      "step": 51021
    },
    {
      "epoch": 4.389744472167255,
      "grad_norm": 1.0390625,
      "learning_rate": 3.633832294251338e-05,
      "loss": 0.5074,
      "step": 51022
    },
    {
      "epoch": 4.389830508474576,
      "grad_norm": 5.3125,
      "learning_rate": 3.632820332421794e-05,
      "loss": 0.3807,
      "step": 51023
    },
    {
      "epoch": 4.389916544781898,
      "grad_norm": 1.046875,
      "learning_rate": 3.631808506207235e-05,
      "loss": 1.0425,
      "step": 51024
    },
    {
      "epoch": 4.390002581089219,
      "grad_norm": 1.9296875,
      "learning_rate": 3.6307968156106266e-05,
      "loss": 1.2198,
      "step": 51025
    },
    {
      "epoch": 4.390088617396541,
      "grad_norm": 1.03125,
      "learning_rate": 3.62978526063491e-05,
      "loss": 1.2378,
      "step": 51026
    },
    {
      "epoch": 4.390174653703863,
      "grad_norm": 1.203125,
      "learning_rate": 3.628773841283056e-05,
      "loss": 1.6014,
      "step": 51027
    },
    {
      "epoch": 4.390260690011185,
      "grad_norm": 1.3125,
      "learning_rate": 3.627762557558029e-05,
      "loss": 1.5193,
      "step": 51028
    },
    {
      "epoch": 4.390346726318507,
      "grad_norm": 45.0,
      "learning_rate": 3.626751409462764e-05,
      "loss": 0.746,
      "step": 51029
    },
    {
      "epoch": 4.390432762625828,
      "grad_norm": 0.8203125,
      "learning_rate": 3.625740397000232e-05,
      "loss": 0.6683,
      "step": 51030
    },
    {
      "epoch": 4.39051879893315,
      "grad_norm": 1.40625,
      "learning_rate": 3.624729520173403e-05,
      "loss": 1.2222,
      "step": 51031
    },
    {
      "epoch": 4.390604835240471,
      "grad_norm": 1.953125,
      "learning_rate": 3.6237187789852074e-05,
      "loss": 1.0234,
      "step": 51032
    },
    {
      "epoch": 4.390690871547793,
      "grad_norm": 1.296875,
      "learning_rate": 3.622708173438627e-05,
      "loss": 1.1442,
      "step": 51033
    },
    {
      "epoch": 4.390776907855114,
      "grad_norm": 1.1875,
      "learning_rate": 3.621697703536597e-05,
      "loss": 1.2715,
      "step": 51034
    },
    {
      "epoch": 4.3908629441624365,
      "grad_norm": 0.9765625,
      "learning_rate": 3.6206873692820774e-05,
      "loss": 1.196,
      "step": 51035
    },
    {
      "epoch": 4.390948980469759,
      "grad_norm": 1.1875,
      "learning_rate": 3.619677170678037e-05,
      "loss": 1.8774,
      "step": 51036
    },
    {
      "epoch": 4.39103501677708,
      "grad_norm": 1.4765625,
      "learning_rate": 3.6186671077274136e-05,
      "loss": 1.1638,
      "step": 51037
    },
    {
      "epoch": 4.391121053084402,
      "grad_norm": 22.625,
      "learning_rate": 3.61765718043317e-05,
      "loss": 0.7309,
      "step": 51038
    },
    {
      "epoch": 4.391207089391723,
      "grad_norm": 0.8359375,
      "learning_rate": 3.616647388798255e-05,
      "loss": 0.8762,
      "step": 51039
    },
    {
      "epoch": 4.391293125699045,
      "grad_norm": 1.1953125,
      "learning_rate": 3.615637732825627e-05,
      "loss": 1.4548,
      "step": 51040
    },
    {
      "epoch": 4.391379162006366,
      "grad_norm": 1.7109375,
      "learning_rate": 3.614628212518245e-05,
      "loss": 1.4331,
      "step": 51041
    },
    {
      "epoch": 4.391465198313688,
      "grad_norm": 44.75,
      "learning_rate": 3.613618827879045e-05,
      "loss": 0.9588,
      "step": 51042
    },
    {
      "epoch": 4.39155123462101,
      "grad_norm": 0.96484375,
      "learning_rate": 3.612609578910986e-05,
      "loss": 0.6074,
      "step": 51043
    },
    {
      "epoch": 4.391637270928332,
      "grad_norm": 1.109375,
      "learning_rate": 3.611600465617032e-05,
      "loss": 1.1546,
      "step": 51044
    },
    {
      "epoch": 4.391723307235654,
      "grad_norm": 1.1796875,
      "learning_rate": 3.61059148800012e-05,
      "loss": 1.0195,
      "step": 51045
    },
    {
      "epoch": 4.391809343542975,
      "grad_norm": 1.1953125,
      "learning_rate": 3.609582646063203e-05,
      "loss": 1.0395,
      "step": 51046
    },
    {
      "epoch": 4.391895379850297,
      "grad_norm": 1.2109375,
      "learning_rate": 3.60857393980924e-05,
      "loss": 1.8565,
      "step": 51047
    },
    {
      "epoch": 4.391981416157618,
      "grad_norm": 1.53125,
      "learning_rate": 3.607565369241173e-05,
      "loss": 0.7456,
      "step": 51048
    },
    {
      "epoch": 4.39206745246494,
      "grad_norm": 1.6015625,
      "learning_rate": 3.60655693436196e-05,
      "loss": 1.0159,
      "step": 51049
    },
    {
      "epoch": 4.392153488772262,
      "grad_norm": 84.0,
      "learning_rate": 3.605548635174533e-05,
      "loss": 1.1276,
      "step": 51050
    },
    {
      "epoch": 4.392239525079583,
      "grad_norm": 1.0703125,
      "learning_rate": 3.604540471681861e-05,
      "loss": 1.021,
      "step": 51051
    },
    {
      "epoch": 4.392325561386905,
      "grad_norm": 1.1484375,
      "learning_rate": 3.603532443886887e-05,
      "loss": 1.4572,
      "step": 51052
    },
    {
      "epoch": 4.392411597694227,
      "grad_norm": 0.91015625,
      "learning_rate": 3.602524551792546e-05,
      "loss": 1.2208,
      "step": 51053
    },
    {
      "epoch": 4.392497634001549,
      "grad_norm": 1.1640625,
      "learning_rate": 3.6015167954018043e-05,
      "loss": 1.0,
      "step": 51054
    },
    {
      "epoch": 4.392583670308871,
      "grad_norm": 1.03125,
      "learning_rate": 3.600509174717609e-05,
      "loss": 1.6356,
      "step": 51055
    },
    {
      "epoch": 4.392669706616192,
      "grad_norm": 235.0,
      "learning_rate": 3.59950168974289e-05,
      "loss": 1.1589,
      "step": 51056
    },
    {
      "epoch": 4.392755742923514,
      "grad_norm": 1.078125,
      "learning_rate": 3.5984943404806017e-05,
      "loss": 1.1003,
      "step": 51057
    },
    {
      "epoch": 4.392841779230835,
      "grad_norm": 4.3125,
      "learning_rate": 3.5974871269337017e-05,
      "loss": 1.3192,
      "step": 51058
    },
    {
      "epoch": 4.392927815538157,
      "grad_norm": 0.859375,
      "learning_rate": 3.596480049105122e-05,
      "loss": 0.6089,
      "step": 51059
    },
    {
      "epoch": 4.393013851845478,
      "grad_norm": 1.0390625,
      "learning_rate": 3.595473106997821e-05,
      "loss": 1.4342,
      "step": 51060
    },
    {
      "epoch": 4.3930998881528005,
      "grad_norm": 1.1015625,
      "learning_rate": 3.594466300614724e-05,
      "loss": 1.2917,
      "step": 51061
    },
    {
      "epoch": 4.393185924460123,
      "grad_norm": 1.265625,
      "learning_rate": 3.59345962995879e-05,
      "loss": 1.6204,
      "step": 51062
    },
    {
      "epoch": 4.393271960767444,
      "grad_norm": 15.0,
      "learning_rate": 3.592453095032971e-05,
      "loss": 0.9921,
      "step": 51063
    },
    {
      "epoch": 4.393357997074766,
      "grad_norm": 1.0078125,
      "learning_rate": 3.591446695840189e-05,
      "loss": 1.0367,
      "step": 51064
    },
    {
      "epoch": 4.393444033382087,
      "grad_norm": 67.0,
      "learning_rate": 3.590440432383402e-05,
      "loss": 1.2572,
      "step": 51065
    },
    {
      "epoch": 4.393530069689409,
      "grad_norm": 0.94921875,
      "learning_rate": 3.589434304665562e-05,
      "loss": 1.1682,
      "step": 51066
    },
    {
      "epoch": 4.39361610599673,
      "grad_norm": 40.5,
      "learning_rate": 3.58842831268959e-05,
      "loss": 1.6993,
      "step": 51067
    },
    {
      "epoch": 4.393702142304052,
      "grad_norm": 1.34375,
      "learning_rate": 3.587422456458434e-05,
      "loss": 1.0313,
      "step": 51068
    },
    {
      "epoch": 4.393788178611374,
      "grad_norm": 1.1875,
      "learning_rate": 3.586416735975051e-05,
      "loss": 0.7515,
      "step": 51069
    },
    {
      "epoch": 4.393874214918696,
      "grad_norm": 1.015625,
      "learning_rate": 3.5854111512423684e-05,
      "loss": 1.2228,
      "step": 51070
    },
    {
      "epoch": 4.393960251226018,
      "grad_norm": 1.609375,
      "learning_rate": 3.584405702263327e-05,
      "loss": 1.6538,
      "step": 51071
    },
    {
      "epoch": 4.394046287533339,
      "grad_norm": 1.4765625,
      "learning_rate": 3.5834003890408804e-05,
      "loss": 1.4769,
      "step": 51072
    },
    {
      "epoch": 4.394132323840661,
      "grad_norm": 0.89453125,
      "learning_rate": 3.582395211577949e-05,
      "loss": 0.9726,
      "step": 51073
    },
    {
      "epoch": 4.394218360147982,
      "grad_norm": 0.96484375,
      "learning_rate": 3.5813901698774963e-05,
      "loss": 0.6547,
      "step": 51074
    },
    {
      "epoch": 4.394304396455304,
      "grad_norm": 1.2421875,
      "learning_rate": 3.580385263942437e-05,
      "loss": 0.9675,
      "step": 51075
    },
    {
      "epoch": 4.394390432762626,
      "grad_norm": 1.8203125,
      "learning_rate": 3.579380493775719e-05,
      "loss": 1.2963,
      "step": 51076
    },
    {
      "epoch": 4.394476469069947,
      "grad_norm": 18.375,
      "learning_rate": 3.578375859380301e-05,
      "loss": 1.3855,
      "step": 51077
    },
    {
      "epoch": 4.3945625053772694,
      "grad_norm": 1.2421875,
      "learning_rate": 3.5773713607590965e-05,
      "loss": 1.3004,
      "step": 51078
    },
    {
      "epoch": 4.394648541684591,
      "grad_norm": 5.375,
      "learning_rate": 3.576366997915059e-05,
      "loss": 1.284,
      "step": 51079
    },
    {
      "epoch": 4.394734577991913,
      "grad_norm": 0.9609375,
      "learning_rate": 3.575362770851109e-05,
      "loss": 0.8768,
      "step": 51080
    },
    {
      "epoch": 4.394820614299234,
      "grad_norm": 1.1171875,
      "learning_rate": 3.574358679570194e-05,
      "loss": 0.9106,
      "step": 51081
    },
    {
      "epoch": 4.394906650606556,
      "grad_norm": 1.375,
      "learning_rate": 3.5733547240752616e-05,
      "loss": 1.043,
      "step": 51082
    },
    {
      "epoch": 4.394992686913878,
      "grad_norm": 1.2734375,
      "learning_rate": 3.572350904369231e-05,
      "loss": 0.9641,
      "step": 51083
    },
    {
      "epoch": 4.395078723221199,
      "grad_norm": 0.98046875,
      "learning_rate": 3.57134722045504e-05,
      "loss": 0.6691,
      "step": 51084
    },
    {
      "epoch": 4.395164759528521,
      "grad_norm": 1.5625,
      "learning_rate": 3.5703436723356406e-05,
      "loss": 0.6798,
      "step": 51085
    },
    {
      "epoch": 4.395250795835842,
      "grad_norm": 0.8359375,
      "learning_rate": 3.569340260013942e-05,
      "loss": 0.831,
      "step": 51086
    },
    {
      "epoch": 4.3953368321431645,
      "grad_norm": 1.0546875,
      "learning_rate": 3.568336983492898e-05,
      "loss": 0.6423,
      "step": 51087
    },
    {
      "epoch": 4.395422868450487,
      "grad_norm": 1.7890625,
      "learning_rate": 3.5673338427754444e-05,
      "loss": 1.8623,
      "step": 51088
    },
    {
      "epoch": 4.395508904757808,
      "grad_norm": 0.9375,
      "learning_rate": 3.566330837864501e-05,
      "loss": 0.509,
      "step": 51089
    },
    {
      "epoch": 4.39559494106513,
      "grad_norm": 4.65625,
      "learning_rate": 3.565327968763016e-05,
      "loss": 0.9805,
      "step": 51090
    },
    {
      "epoch": 4.395680977372451,
      "grad_norm": 26.0,
      "learning_rate": 3.5643252354739086e-05,
      "loss": 1.2824,
      "step": 51091
    },
    {
      "epoch": 4.395767013679773,
      "grad_norm": 0.859375,
      "learning_rate": 3.563322638000116e-05,
      "loss": 0.9082,
      "step": 51092
    },
    {
      "epoch": 4.395853049987094,
      "grad_norm": 0.984375,
      "learning_rate": 3.5623201763445856e-05,
      "loss": 1.024,
      "step": 51093
    },
    {
      "epoch": 4.395939086294416,
      "grad_norm": 1.984375,
      "learning_rate": 3.561317850510226e-05,
      "loss": 1.0287,
      "step": 51094
    },
    {
      "epoch": 4.396025122601738,
      "grad_norm": 1.03125,
      "learning_rate": 3.56031566049998e-05,
      "loss": 0.9043,
      "step": 51095
    },
    {
      "epoch": 4.39611115890906,
      "grad_norm": 0.96484375,
      "learning_rate": 3.5593136063167883e-05,
      "loss": 1.0298,
      "step": 51096
    },
    {
      "epoch": 4.396197195216382,
      "grad_norm": 0.80078125,
      "learning_rate": 3.558311687963561e-05,
      "loss": 0.7303,
      "step": 51097
    },
    {
      "epoch": 4.396283231523703,
      "grad_norm": 11.625,
      "learning_rate": 3.55730990544324e-05,
      "loss": 1.3,
      "step": 51098
    },
    {
      "epoch": 4.396369267831025,
      "grad_norm": 0.984375,
      "learning_rate": 3.556308258758767e-05,
      "loss": 1.0118,
      "step": 51099
    },
    {
      "epoch": 4.396455304138346,
      "grad_norm": 25.0,
      "learning_rate": 3.555306747913051e-05,
      "loss": 1.025,
      "step": 51100
    },
    {
      "epoch": 4.396541340445668,
      "grad_norm": 1.5859375,
      "learning_rate": 3.554305372909022e-05,
      "loss": 1.3003,
      "step": 51101
    },
    {
      "epoch": 4.39662737675299,
      "grad_norm": 10.625,
      "learning_rate": 3.553304133749635e-05,
      "loss": 0.9354,
      "step": 51102
    },
    {
      "epoch": 4.396713413060311,
      "grad_norm": 540.0,
      "learning_rate": 3.552303030437787e-05,
      "loss": 1.0838,
      "step": 51103
    },
    {
      "epoch": 4.3967994493676334,
      "grad_norm": 1.09375,
      "learning_rate": 3.5513020629764244e-05,
      "loss": 1.5237,
      "step": 51104
    },
    {
      "epoch": 4.396885485674955,
      "grad_norm": 1.3125,
      "learning_rate": 3.550301231368469e-05,
      "loss": 1.2206,
      "step": 51105
    },
    {
      "epoch": 4.396971521982277,
      "grad_norm": 1.1953125,
      "learning_rate": 3.54930053561684e-05,
      "loss": 0.7131,
      "step": 51106
    },
    {
      "epoch": 4.397057558289598,
      "grad_norm": 1.0,
      "learning_rate": 3.5482999757244896e-05,
      "loss": 1.2577,
      "step": 51107
    },
    {
      "epoch": 4.39714359459692,
      "grad_norm": 248.0,
      "learning_rate": 3.547299551694311e-05,
      "loss": 1.0201,
      "step": 51108
    },
    {
      "epoch": 4.397229630904242,
      "grad_norm": 1.15625,
      "learning_rate": 3.546299263529251e-05,
      "loss": 1.4122,
      "step": 51109
    },
    {
      "epoch": 4.397315667211563,
      "grad_norm": 1.0390625,
      "learning_rate": 3.5452991112322365e-05,
      "loss": 1.2546,
      "step": 51110
    },
    {
      "epoch": 4.397401703518885,
      "grad_norm": 0.90625,
      "learning_rate": 3.544299094806186e-05,
      "loss": 0.5315,
      "step": 51111
    },
    {
      "epoch": 4.397487739826206,
      "grad_norm": 1.46875,
      "learning_rate": 3.54329921425402e-05,
      "loss": 1.7232,
      "step": 51112
    },
    {
      "epoch": 4.3975737761335285,
      "grad_norm": 1.5234375,
      "learning_rate": 3.542299469578669e-05,
      "loss": 1.0903,
      "step": 51113
    },
    {
      "epoch": 4.39765981244085,
      "grad_norm": 1.265625,
      "learning_rate": 3.541299860783054e-05,
      "loss": 1.4967,
      "step": 51114
    },
    {
      "epoch": 4.397745848748172,
      "grad_norm": 0.91015625,
      "learning_rate": 3.54030038787011e-05,
      "loss": 0.6768,
      "step": 51115
    },
    {
      "epoch": 4.397831885055494,
      "grad_norm": 2.046875,
      "learning_rate": 3.539301050842747e-05,
      "loss": 0.8103,
      "step": 51116
    },
    {
      "epoch": 4.397917921362815,
      "grad_norm": 0.95703125,
      "learning_rate": 3.53830184970389e-05,
      "loss": 1.3238,
      "step": 51117
    },
    {
      "epoch": 4.398003957670137,
      "grad_norm": 2.78125,
      "learning_rate": 3.5373027844564756e-05,
      "loss": 1.7683,
      "step": 51118
    },
    {
      "epoch": 4.398089993977458,
      "grad_norm": 0.98046875,
      "learning_rate": 3.536303855103401e-05,
      "loss": 1.0695,
      "step": 51119
    },
    {
      "epoch": 4.39817603028478,
      "grad_norm": 1.546875,
      "learning_rate": 3.535305061647609e-05,
      "loss": 1.5777,
      "step": 51120
    },
    {
      "epoch": 4.3982620665921015,
      "grad_norm": 1.2890625,
      "learning_rate": 3.534306404092008e-05,
      "loss": 1.5045,
      "step": 51121
    },
    {
      "epoch": 4.398348102899424,
      "grad_norm": 0.75,
      "learning_rate": 3.5333078824395235e-05,
      "loss": 0.4982,
      "step": 51122
    },
    {
      "epoch": 4.398434139206746,
      "grad_norm": 1.0703125,
      "learning_rate": 3.532309496693081e-05,
      "loss": 0.7761,
      "step": 51123
    },
    {
      "epoch": 4.398520175514067,
      "grad_norm": 1.4140625,
      "learning_rate": 3.53131124685559e-05,
      "loss": 1.4275,
      "step": 51124
    },
    {
      "epoch": 4.398606211821389,
      "grad_norm": 1.1015625,
      "learning_rate": 3.530313132929974e-05,
      "loss": 1.2705,
      "step": 51125
    },
    {
      "epoch": 4.39869224812871,
      "grad_norm": 0.98828125,
      "learning_rate": 3.529315154919166e-05,
      "loss": 0.7572,
      "step": 51126
    },
    {
      "epoch": 4.398778284436032,
      "grad_norm": 1.1796875,
      "learning_rate": 3.528317312826062e-05,
      "loss": 1.8721,
      "step": 51127
    },
    {
      "epoch": 4.398864320743353,
      "grad_norm": 0.87109375,
      "learning_rate": 3.52731960665359e-05,
      "loss": 0.4483,
      "step": 51128
    },
    {
      "epoch": 4.398950357050675,
      "grad_norm": 1.6875,
      "learning_rate": 3.526322036404683e-05,
      "loss": 1.1304,
      "step": 51129
    },
    {
      "epoch": 4.3990363933579975,
      "grad_norm": 1.0546875,
      "learning_rate": 3.525324602082236e-05,
      "loss": 0.4979,
      "step": 51130
    },
    {
      "epoch": 4.399122429665319,
      "grad_norm": 1.28125,
      "learning_rate": 3.524327303689179e-05,
      "loss": 1.299,
      "step": 51131
    },
    {
      "epoch": 4.399208465972641,
      "grad_norm": 10.0625,
      "learning_rate": 3.52333014122842e-05,
      "loss": 0.7872,
      "step": 51132
    },
    {
      "epoch": 4.399294502279962,
      "grad_norm": 348.0,
      "learning_rate": 3.522333114702886e-05,
      "loss": 1.1489,
      "step": 51133
    },
    {
      "epoch": 4.399380538587284,
      "grad_norm": 1.578125,
      "learning_rate": 3.52133622411549e-05,
      "loss": 1.5822,
      "step": 51134
    },
    {
      "epoch": 4.399466574894605,
      "grad_norm": 0.9921875,
      "learning_rate": 3.5203394694691415e-05,
      "loss": 0.8799,
      "step": 51135
    },
    {
      "epoch": 4.399552611201927,
      "grad_norm": 11.5625,
      "learning_rate": 3.51934285076676e-05,
      "loss": 0.9267,
      "step": 51136
    },
    {
      "epoch": 4.399638647509249,
      "grad_norm": 0.9296875,
      "learning_rate": 3.518346368011266e-05,
      "loss": 0.7476,
      "step": 51137
    },
    {
      "epoch": 4.3997246838165704,
      "grad_norm": 1.59375,
      "learning_rate": 3.517350021205562e-05,
      "loss": 0.9288,
      "step": 51138
    },
    {
      "epoch": 4.3998107201238925,
      "grad_norm": 2.578125,
      "learning_rate": 3.516353810352568e-05,
      "loss": 0.9333,
      "step": 51139
    },
    {
      "epoch": 4.399896756431214,
      "grad_norm": 2.921875,
      "learning_rate": 3.51535773545521e-05,
      "loss": 1.3487,
      "step": 51140
    },
    {
      "epoch": 4.399982792738536,
      "grad_norm": 1.3515625,
      "learning_rate": 3.51436179651638e-05,
      "loss": 1.0813,
      "step": 51141
    },
    {
      "epoch": 4.400068829045857,
      "grad_norm": 1.03125,
      "learning_rate": 3.513365993538997e-05,
      "loss": 1.505,
      "step": 51142
    },
    {
      "epoch": 4.400154865353179,
      "grad_norm": 0.98046875,
      "learning_rate": 3.512370326525993e-05,
      "loss": 1.1439,
      "step": 51143
    },
    {
      "epoch": 4.400240901660501,
      "grad_norm": 5.46875,
      "learning_rate": 3.511374795480249e-05,
      "loss": 1.2082,
      "step": 51144
    },
    {
      "epoch": 4.400326937967822,
      "grad_norm": 1.421875,
      "learning_rate": 3.510379400404706e-05,
      "loss": 2.1779,
      "step": 51145
    },
    {
      "epoch": 4.400412974275144,
      "grad_norm": 1.6484375,
      "learning_rate": 3.509384141302252e-05,
      "loss": 1.3603,
      "step": 51146
    },
    {
      "epoch": 4.4004990105824655,
      "grad_norm": 1.359375,
      "learning_rate": 3.508389018175806e-05,
      "loss": 1.21,
      "step": 51147
    },
    {
      "epoch": 4.400585046889788,
      "grad_norm": 0.93359375,
      "learning_rate": 3.507394031028288e-05,
      "loss": 1.034,
      "step": 51148
    },
    {
      "epoch": 4.400671083197109,
      "grad_norm": 0.8203125,
      "learning_rate": 3.5063991798625906e-05,
      "loss": 0.6995,
      "step": 51149
    },
    {
      "epoch": 4.400757119504431,
      "grad_norm": 1.46875,
      "learning_rate": 3.505404464681633e-05,
      "loss": 1.2513,
      "step": 51150
    },
    {
      "epoch": 4.400843155811753,
      "grad_norm": 0.8984375,
      "learning_rate": 3.50440988548833e-05,
      "loss": 0.7061,
      "step": 51151
    },
    {
      "epoch": 4.400929192119074,
      "grad_norm": 14.5,
      "learning_rate": 3.503415442285579e-05,
      "loss": 1.2843,
      "step": 51152
    },
    {
      "epoch": 4.401015228426396,
      "grad_norm": 0.96875,
      "learning_rate": 3.502421135076306e-05,
      "loss": 0.9009,
      "step": 51153
    },
    {
      "epoch": 4.401101264733717,
      "grad_norm": 1.6328125,
      "learning_rate": 3.501426963863402e-05,
      "loss": 1.5449,
      "step": 51154
    },
    {
      "epoch": 4.401187301041039,
      "grad_norm": 1.109375,
      "learning_rate": 3.5004329286497773e-05,
      "loss": 1.5731,
      "step": 51155
    },
    {
      "epoch": 4.401273337348361,
      "grad_norm": 41.0,
      "learning_rate": 3.4994390294383507e-05,
      "loss": 0.7936,
      "step": 51156
    },
    {
      "epoch": 4.401359373655683,
      "grad_norm": 0.9140625,
      "learning_rate": 3.4984452662320146e-05,
      "loss": 0.6384,
      "step": 51157
    },
    {
      "epoch": 4.401445409963005,
      "grad_norm": 0.98046875,
      "learning_rate": 3.497451639033683e-05,
      "loss": 1.0148,
      "step": 51158
    },
    {
      "epoch": 4.401531446270326,
      "grad_norm": 0.9296875,
      "learning_rate": 3.4964581478462656e-05,
      "loss": 1.0718,
      "step": 51159
    },
    {
      "epoch": 4.401617482577648,
      "grad_norm": 1.671875,
      "learning_rate": 3.495464792672659e-05,
      "loss": 1.3018,
      "step": 51160
    },
    {
      "epoch": 4.401703518884969,
      "grad_norm": 6.5,
      "learning_rate": 3.494471573515778e-05,
      "loss": 1.6312,
      "step": 51161
    },
    {
      "epoch": 4.401789555192291,
      "grad_norm": 1.2265625,
      "learning_rate": 3.493478490378516e-05,
      "loss": 0.7718,
      "step": 51162
    },
    {
      "epoch": 4.401875591499612,
      "grad_norm": 0.890625,
      "learning_rate": 3.4924855432637845e-05,
      "loss": 0.5842,
      "step": 51163
    },
    {
      "epoch": 4.4019616278069345,
      "grad_norm": 6.8125,
      "learning_rate": 3.491492732174495e-05,
      "loss": 0.5439,
      "step": 51164
    },
    {
      "epoch": 4.4020476641142565,
      "grad_norm": 0.8828125,
      "learning_rate": 3.490500057113538e-05,
      "loss": 0.5397,
      "step": 51165
    },
    {
      "epoch": 4.402133700421578,
      "grad_norm": 0.93359375,
      "learning_rate": 3.489507518083823e-05,
      "loss": 1.5622,
      "step": 51166
    },
    {
      "epoch": 4.4022197367289,
      "grad_norm": 1.5078125,
      "learning_rate": 3.488515115088259e-05,
      "loss": 1.3406,
      "step": 51167
    },
    {
      "epoch": 4.402305773036221,
      "grad_norm": 122.0,
      "learning_rate": 3.487522848129737e-05,
      "loss": 0.8024,
      "step": 51168
    },
    {
      "epoch": 4.402391809343543,
      "grad_norm": 1.3515625,
      "learning_rate": 3.486530717211162e-05,
      "loss": 1.7494,
      "step": 51169
    },
    {
      "epoch": 4.402477845650864,
      "grad_norm": 1.0234375,
      "learning_rate": 3.485538722335446e-05,
      "loss": 0.9136,
      "step": 51170
    },
    {
      "epoch": 4.402563881958186,
      "grad_norm": 1.4609375,
      "learning_rate": 3.484546863505472e-05,
      "loss": 1.2465,
      "step": 51171
    },
    {
      "epoch": 4.402649918265508,
      "grad_norm": 1.0234375,
      "learning_rate": 3.483555140724165e-05,
      "loss": 1.2915,
      "step": 51172
    },
    {
      "epoch": 4.4027359545728295,
      "grad_norm": 1.171875,
      "learning_rate": 3.4825635539944e-05,
      "loss": 0.6309,
      "step": 51173
    },
    {
      "epoch": 4.402821990880152,
      "grad_norm": 1.0546875,
      "learning_rate": 3.4815721033190915e-05,
      "loss": 1.2526,
      "step": 51174
    },
    {
      "epoch": 4.402908027187473,
      "grad_norm": 18.25,
      "learning_rate": 3.480580788701149e-05,
      "loss": 0.9169,
      "step": 51175
    },
    {
      "epoch": 4.402994063494795,
      "grad_norm": 292.0,
      "learning_rate": 3.4795896101434465e-05,
      "loss": 1.1791,
      "step": 51176
    },
    {
      "epoch": 4.403080099802117,
      "grad_norm": 44.5,
      "learning_rate": 3.4785985676489e-05,
      "loss": 1.0313,
      "step": 51177
    },
    {
      "epoch": 4.403166136109438,
      "grad_norm": 1.5234375,
      "learning_rate": 3.477607661220411e-05,
      "loss": 1.2754,
      "step": 51178
    },
    {
      "epoch": 4.40325217241676,
      "grad_norm": 0.9765625,
      "learning_rate": 3.476616890860862e-05,
      "loss": 0.9039,
      "step": 51179
    },
    {
      "epoch": 4.403338208724081,
      "grad_norm": 0.76171875,
      "learning_rate": 3.475626256573161e-05,
      "loss": 0.8728,
      "step": 51180
    },
    {
      "epoch": 4.403424245031403,
      "grad_norm": 1.2109375,
      "learning_rate": 3.4746357583602176e-05,
      "loss": 1.1238,
      "step": 51181
    },
    {
      "epoch": 4.403510281338725,
      "grad_norm": 20.625,
      "learning_rate": 3.4736453962249005e-05,
      "loss": 1.4069,
      "step": 51182
    },
    {
      "epoch": 4.403596317646047,
      "grad_norm": 0.86328125,
      "learning_rate": 3.472655170170125e-05,
      "loss": 0.9172,
      "step": 51183
    },
    {
      "epoch": 4.403682353953369,
      "grad_norm": 1.8515625,
      "learning_rate": 3.4716650801987934e-05,
      "loss": 0.9825,
      "step": 51184
    },
    {
      "epoch": 4.40376839026069,
      "grad_norm": 1.0625,
      "learning_rate": 3.470675126313782e-05,
      "loss": 1.0696,
      "step": 51185
    },
    {
      "epoch": 4.403854426568012,
      "grad_norm": 1.2734375,
      "learning_rate": 3.469685308518006e-05,
      "loss": 0.9313,
      "step": 51186
    },
    {
      "epoch": 4.403940462875333,
      "grad_norm": 1.21875,
      "learning_rate": 3.4686956268143385e-05,
      "loss": 0.6173,
      "step": 51187
    },
    {
      "epoch": 4.404026499182655,
      "grad_norm": 24.375,
      "learning_rate": 3.46770608120569e-05,
      "loss": 0.9118,
      "step": 51188
    },
    {
      "epoch": 4.404112535489976,
      "grad_norm": 0.99609375,
      "learning_rate": 3.4667166716949584e-05,
      "loss": 1.0838,
      "step": 51189
    },
    {
      "epoch": 4.4041985717972985,
      "grad_norm": 20.25,
      "learning_rate": 3.465727398285023e-05,
      "loss": 1.0275,
      "step": 51190
    },
    {
      "epoch": 4.4042846081046205,
      "grad_norm": 1.28125,
      "learning_rate": 3.464738260978795e-05,
      "loss": 1.2949,
      "step": 51191
    },
    {
      "epoch": 4.404370644411942,
      "grad_norm": 27.25,
      "learning_rate": 3.4637492597791476e-05,
      "loss": 0.7302,
      "step": 51192
    },
    {
      "epoch": 4.404456680719264,
      "grad_norm": 1.375,
      "learning_rate": 3.46276039468898e-05,
      "loss": 1.801,
      "step": 51193
    },
    {
      "epoch": 4.404542717026585,
      "grad_norm": 1.15625,
      "learning_rate": 3.4617716657111995e-05,
      "loss": 1.7098,
      "step": 51194
    },
    {
      "epoch": 4.404628753333907,
      "grad_norm": 1.015625,
      "learning_rate": 3.460783072848672e-05,
      "loss": 1.0402,
      "step": 51195
    },
    {
      "epoch": 4.404714789641228,
      "grad_norm": 1.578125,
      "learning_rate": 3.4597946161043056e-05,
      "loss": 1.7413,
      "step": 51196
    },
    {
      "epoch": 4.40480082594855,
      "grad_norm": 1.390625,
      "learning_rate": 3.4588062954809986e-05,
      "loss": 1.6288,
      "step": 51197
    },
    {
      "epoch": 4.404886862255872,
      "grad_norm": 486.0,
      "learning_rate": 3.457818110981625e-05,
      "loss": 0.937,
      "step": 51198
    },
    {
      "epoch": 4.4049728985631935,
      "grad_norm": 1.09375,
      "learning_rate": 3.4568300626090785e-05,
      "loss": 0.5974,
      "step": 51199
    },
    {
      "epoch": 4.405058934870516,
      "grad_norm": 1.0546875,
      "learning_rate": 3.4558421503662564e-05,
      "loss": 0.8671,
      "step": 51200
    },
    {
      "epoch": 4.405144971177837,
      "grad_norm": 14.3125,
      "learning_rate": 3.454854374256039e-05,
      "loss": 1.4169,
      "step": 51201
    },
    {
      "epoch": 4.405231007485159,
      "grad_norm": 1.03125,
      "learning_rate": 3.45386673428133e-05,
      "loss": 0.6641,
      "step": 51202
    },
    {
      "epoch": 4.40531704379248,
      "grad_norm": 1.0703125,
      "learning_rate": 3.452879230445e-05,
      "loss": 2.009,
      "step": 51203
    },
    {
      "epoch": 4.405403080099802,
      "grad_norm": 1.1328125,
      "learning_rate": 3.45189186274994e-05,
      "loss": 1.2474,
      "step": 51204
    },
    {
      "epoch": 4.405489116407124,
      "grad_norm": 6.28125,
      "learning_rate": 3.450904631199059e-05,
      "loss": 0.758,
      "step": 51205
    },
    {
      "epoch": 4.405575152714445,
      "grad_norm": 1.2109375,
      "learning_rate": 3.449917535795216e-05,
      "loss": 0.9347,
      "step": 51206
    },
    {
      "epoch": 4.405661189021767,
      "grad_norm": 1.28125,
      "learning_rate": 3.448930576541309e-05,
      "loss": 1.16,
      "step": 51207
    },
    {
      "epoch": 4.405747225329089,
      "grad_norm": 12.9375,
      "learning_rate": 3.447943753440236e-05,
      "loss": 0.884,
      "step": 51208
    },
    {
      "epoch": 4.405833261636411,
      "grad_norm": 1.1875,
      "learning_rate": 3.446957066494866e-05,
      "loss": 0.6301,
      "step": 51209
    },
    {
      "epoch": 4.405919297943733,
      "grad_norm": 1.234375,
      "learning_rate": 3.445970515708091e-05,
      "loss": 1.4251,
      "step": 51210
    },
    {
      "epoch": 4.406005334251054,
      "grad_norm": 19.25,
      "learning_rate": 3.44498410108281e-05,
      "loss": 1.938,
      "step": 51211
    },
    {
      "epoch": 4.406091370558376,
      "grad_norm": 0.953125,
      "learning_rate": 3.443997822621886e-05,
      "loss": 0.7883,
      "step": 51212
    },
    {
      "epoch": 4.406177406865697,
      "grad_norm": 1.0859375,
      "learning_rate": 3.443011680328223e-05,
      "loss": 1.3272,
      "step": 51213
    },
    {
      "epoch": 4.406263443173019,
      "grad_norm": 1.0625,
      "learning_rate": 3.442025674204685e-05,
      "loss": 0.8314,
      "step": 51214
    },
    {
      "epoch": 4.40634947948034,
      "grad_norm": 1.0625,
      "learning_rate": 3.44103980425417e-05,
      "loss": 0.8034,
      "step": 51215
    },
    {
      "epoch": 4.4064355157876625,
      "grad_norm": 0.93359375,
      "learning_rate": 3.4400540704795644e-05,
      "loss": 0.8948,
      "step": 51216
    },
    {
      "epoch": 4.4065215520949845,
      "grad_norm": 1.3125,
      "learning_rate": 3.4390684728837386e-05,
      "loss": 0.9765,
      "step": 51217
    },
    {
      "epoch": 4.406607588402306,
      "grad_norm": 1.328125,
      "learning_rate": 3.438083011469578e-05,
      "loss": 0.9191,
      "step": 51218
    },
    {
      "epoch": 4.406693624709628,
      "grad_norm": 1.1875,
      "learning_rate": 3.437097686239981e-05,
      "loss": 0.8998,
      "step": 51219
    },
    {
      "epoch": 4.406779661016949,
      "grad_norm": 1.203125,
      "learning_rate": 3.436112497197808e-05,
      "loss": 1.3487,
      "step": 51220
    },
    {
      "epoch": 4.406865697324271,
      "grad_norm": 1.0,
      "learning_rate": 3.435127444345948e-05,
      "loss": 0.8444,
      "step": 51221
    },
    {
      "epoch": 4.406951733631592,
      "grad_norm": 1.0390625,
      "learning_rate": 3.434142527687295e-05,
      "loss": 1.0837,
      "step": 51222
    },
    {
      "epoch": 4.407037769938914,
      "grad_norm": 1.15625,
      "learning_rate": 3.433157747224708e-05,
      "loss": 0.7058,
      "step": 51223
    },
    {
      "epoch": 4.407123806246236,
      "grad_norm": 1.140625,
      "learning_rate": 3.432173102961078e-05,
      "loss": 0.7366,
      "step": 51224
    },
    {
      "epoch": 4.4072098425535575,
      "grad_norm": 0.94921875,
      "learning_rate": 3.431188594899287e-05,
      "loss": 0.8671,
      "step": 51225
    },
    {
      "epoch": 4.40729587886088,
      "grad_norm": 1.1796875,
      "learning_rate": 3.4302042230422106e-05,
      "loss": 0.7347,
      "step": 51226
    },
    {
      "epoch": 4.407381915168201,
      "grad_norm": 1.3671875,
      "learning_rate": 3.4292199873927396e-05,
      "loss": 1.6308,
      "step": 51227
    },
    {
      "epoch": 4.407467951475523,
      "grad_norm": 1.5546875,
      "learning_rate": 3.428235887953729e-05,
      "loss": 0.9436,
      "step": 51228
    },
    {
      "epoch": 4.407553987782844,
      "grad_norm": 1.1328125,
      "learning_rate": 3.427251924728075e-05,
      "loss": 1.1787,
      "step": 51229
    },
    {
      "epoch": 4.407640024090166,
      "grad_norm": 1.515625,
      "learning_rate": 3.42626809771866e-05,
      "loss": 1.2024,
      "step": 51230
    },
    {
      "epoch": 4.407726060397488,
      "grad_norm": 0.92578125,
      "learning_rate": 3.425284406928342e-05,
      "loss": 0.866,
      "step": 51231
    },
    {
      "epoch": 4.407812096704809,
      "grad_norm": 1.1796875,
      "learning_rate": 3.4243008523600185e-05,
      "loss": 1.1049,
      "step": 51232
    },
    {
      "epoch": 4.407898133012131,
      "grad_norm": 135.0,
      "learning_rate": 3.423317434016548e-05,
      "loss": 1.6836,
      "step": 51233
    },
    {
      "epoch": 4.407984169319453,
      "grad_norm": 1.5390625,
      "learning_rate": 3.422334151900819e-05,
      "loss": 0.7476,
      "step": 51234
    },
    {
      "epoch": 4.408070205626775,
      "grad_norm": 17.375,
      "learning_rate": 3.421351006015705e-05,
      "loss": 1.1556,
      "step": 51235
    },
    {
      "epoch": 4.408156241934096,
      "grad_norm": 1.0703125,
      "learning_rate": 3.420367996364076e-05,
      "loss": 1.5681,
      "step": 51236
    },
    {
      "epoch": 4.408242278241418,
      "grad_norm": 1.53125,
      "learning_rate": 3.419385122948815e-05,
      "loss": 0.9253,
      "step": 51237
    },
    {
      "epoch": 4.40832831454874,
      "grad_norm": 1.296875,
      "learning_rate": 3.418402385772795e-05,
      "loss": 1.3026,
      "step": 51238
    },
    {
      "epoch": 4.408414350856061,
      "grad_norm": 1.8515625,
      "learning_rate": 3.4174197848388824e-05,
      "loss": 1.3355,
      "step": 51239
    },
    {
      "epoch": 4.408500387163383,
      "grad_norm": 71.0,
      "learning_rate": 3.4164373201499564e-05,
      "loss": 1.3179,
      "step": 51240
    },
    {
      "epoch": 4.408586423470704,
      "grad_norm": 1.15625,
      "learning_rate": 3.4154549917088996e-05,
      "loss": 1.7196,
      "step": 51241
    },
    {
      "epoch": 4.4086724597780265,
      "grad_norm": 1.40625,
      "learning_rate": 3.41447279951857e-05,
      "loss": 1.1998,
      "step": 51242
    },
    {
      "epoch": 4.408758496085348,
      "grad_norm": 1.7265625,
      "learning_rate": 3.413490743581854e-05,
      "loss": 1.4688,
      "step": 51243
    },
    {
      "epoch": 4.40884453239267,
      "grad_norm": 1.203125,
      "learning_rate": 3.412508823901611e-05,
      "loss": 1.572,
      "step": 51244
    },
    {
      "epoch": 4.408930568699992,
      "grad_norm": 1.5546875,
      "learning_rate": 3.411527040480716e-05,
      "loss": 1.6702,
      "step": 51245
    },
    {
      "epoch": 4.409016605007313,
      "grad_norm": 1.8046875,
      "learning_rate": 3.410545393322057e-05,
      "loss": 1.641,
      "step": 51246
    },
    {
      "epoch": 4.409102641314635,
      "grad_norm": 13.75,
      "learning_rate": 3.409563882428479e-05,
      "loss": 1.4683,
      "step": 51247
    },
    {
      "epoch": 4.409188677621956,
      "grad_norm": 1.2890625,
      "learning_rate": 3.4085825078028655e-05,
      "loss": 1.7151,
      "step": 51248
    },
    {
      "epoch": 4.409274713929278,
      "grad_norm": 0.91015625,
      "learning_rate": 3.407601269448091e-05,
      "loss": 0.8908,
      "step": 51249
    },
    {
      "epoch": 4.4093607502365995,
      "grad_norm": 282.0,
      "learning_rate": 3.406620167367019e-05,
      "loss": 0.8134,
      "step": 51250
    },
    {
      "epoch": 4.4094467865439215,
      "grad_norm": 1.28125,
      "learning_rate": 3.4056392015625216e-05,
      "loss": 0.7378,
      "step": 51251
    },
    {
      "epoch": 4.409532822851244,
      "grad_norm": 1.28125,
      "learning_rate": 3.404658372037472e-05,
      "loss": 1.4224,
      "step": 51252
    },
    {
      "epoch": 4.409618859158565,
      "grad_norm": 1.0546875,
      "learning_rate": 3.4036776787947256e-05,
      "loss": 0.9641,
      "step": 51253
    },
    {
      "epoch": 4.409704895465887,
      "grad_norm": 0.79296875,
      "learning_rate": 3.402697121837162e-05,
      "loss": 1.1149,
      "step": 51254
    },
    {
      "epoch": 4.409790931773208,
      "grad_norm": 20.375,
      "learning_rate": 3.401716701167651e-05,
      "loss": 1.0559,
      "step": 51255
    },
    {
      "epoch": 4.40987696808053,
      "grad_norm": 17.875,
      "learning_rate": 3.400736416789052e-05,
      "loss": 0.5487,
      "step": 51256
    },
    {
      "epoch": 4.409963004387851,
      "grad_norm": 3.3125,
      "learning_rate": 3.399756268704241e-05,
      "loss": 0.6315,
      "step": 51257
    },
    {
      "epoch": 4.410049040695173,
      "grad_norm": 7.21875,
      "learning_rate": 3.398776256916075e-05,
      "loss": 0.7172,
      "step": 51258
    },
    {
      "epoch": 4.410135077002495,
      "grad_norm": 1.03125,
      "learning_rate": 3.39779638142742e-05,
      "loss": 0.6371,
      "step": 51259
    },
    {
      "epoch": 4.410221113309817,
      "grad_norm": 1.4140625,
      "learning_rate": 3.396816642241163e-05,
      "loss": 1.0032,
      "step": 51260
    },
    {
      "epoch": 4.410307149617139,
      "grad_norm": 1.0078125,
      "learning_rate": 3.395837039360128e-05,
      "loss": 0.9149,
      "step": 51261
    },
    {
      "epoch": 4.41039318592446,
      "grad_norm": 0.74609375,
      "learning_rate": 3.39485757278723e-05,
      "loss": 0.9301,
      "step": 51262
    },
    {
      "epoch": 4.410479222231782,
      "grad_norm": 1.296875,
      "learning_rate": 3.3938782425252944e-05,
      "loss": 1.7442,
      "step": 51263
    },
    {
      "epoch": 4.410565258539103,
      "grad_norm": 1.1484375,
      "learning_rate": 3.3928990485772026e-05,
      "loss": 1.3599,
      "step": 51264
    },
    {
      "epoch": 4.410651294846425,
      "grad_norm": 1.4609375,
      "learning_rate": 3.3919199909458296e-05,
      "loss": 1.4251,
      "step": 51265
    },
    {
      "epoch": 4.410737331153747,
      "grad_norm": 0.7734375,
      "learning_rate": 3.3909410696340126e-05,
      "loss": 0.538,
      "step": 51266
    },
    {
      "epoch": 4.410823367461068,
      "grad_norm": 1.3125,
      "learning_rate": 3.389962284644632e-05,
      "loss": 0.7448,
      "step": 51267
    },
    {
      "epoch": 4.4109094037683905,
      "grad_norm": 6.09375,
      "learning_rate": 3.388983635980547e-05,
      "loss": 1.1825,
      "step": 51268
    },
    {
      "epoch": 4.410995440075712,
      "grad_norm": 0.98828125,
      "learning_rate": 3.3880051236446164e-05,
      "loss": 0.8954,
      "step": 51269
    },
    {
      "epoch": 4.411081476383034,
      "grad_norm": 1.3984375,
      "learning_rate": 3.3870267476397096e-05,
      "loss": 0.9133,
      "step": 51270
    },
    {
      "epoch": 4.411167512690355,
      "grad_norm": 0.88671875,
      "learning_rate": 3.3860485079686865e-05,
      "loss": 0.8014,
      "step": 51271
    },
    {
      "epoch": 4.411253548997677,
      "grad_norm": 660.0,
      "learning_rate": 3.3850704046343993e-05,
      "loss": 1.1167,
      "step": 51272
    },
    {
      "epoch": 4.411339585304999,
      "grad_norm": 0.984375,
      "learning_rate": 3.3840924376397185e-05,
      "loss": 0.9274,
      "step": 51273
    },
    {
      "epoch": 4.41142562161232,
      "grad_norm": 1.4375,
      "learning_rate": 3.3831146069874975e-05,
      "loss": 1.0746,
      "step": 51274
    },
    {
      "epoch": 4.411511657919642,
      "grad_norm": 0.90234375,
      "learning_rate": 3.3821369126806e-05,
      "loss": 0.6912,
      "step": 51275
    },
    {
      "epoch": 4.4115976942269635,
      "grad_norm": 1.203125,
      "learning_rate": 3.381159354721897e-05,
      "loss": 2.1777,
      "step": 51276
    },
    {
      "epoch": 4.4116837305342855,
      "grad_norm": 1.015625,
      "learning_rate": 3.380181933114218e-05,
      "loss": 0.7306,
      "step": 51277
    },
    {
      "epoch": 4.411769766841607,
      "grad_norm": 0.99609375,
      "learning_rate": 3.379204647860446e-05,
      "loss": 0.989,
      "step": 51278
    },
    {
      "epoch": 4.411855803148929,
      "grad_norm": 0.8203125,
      "learning_rate": 3.378227498963443e-05,
      "loss": 0.8305,
      "step": 51279
    },
    {
      "epoch": 4.411941839456251,
      "grad_norm": 1.328125,
      "learning_rate": 3.377250486426048e-05,
      "loss": 0.8125,
      "step": 51280
    },
    {
      "epoch": 4.412027875763572,
      "grad_norm": 0.953125,
      "learning_rate": 3.376273610251124e-05,
      "loss": 1.0228,
      "step": 51281
    },
    {
      "epoch": 4.412113912070894,
      "grad_norm": 1.046875,
      "learning_rate": 3.375296870441541e-05,
      "loss": 1.9778,
      "step": 51282
    },
    {
      "epoch": 4.412199948378215,
      "grad_norm": 1.4609375,
      "learning_rate": 3.374320267000142e-05,
      "loss": 1.0809,
      "step": 51283
    },
    {
      "epoch": 4.412285984685537,
      "grad_norm": 1.21875,
      "learning_rate": 3.3733437999297954e-05,
      "loss": 1.5762,
      "step": 51284
    },
    {
      "epoch": 4.4123720209928585,
      "grad_norm": 1.171875,
      "learning_rate": 3.37236746923334e-05,
      "loss": 1.0235,
      "step": 51285
    },
    {
      "epoch": 4.412458057300181,
      "grad_norm": 0.9296875,
      "learning_rate": 3.371391274913643e-05,
      "loss": 0.5817,
      "step": 51286
    },
    {
      "epoch": 4.412544093607503,
      "grad_norm": 1.0859375,
      "learning_rate": 3.37041521697356e-05,
      "loss": 0.9193,
      "step": 51287
    },
    {
      "epoch": 4.412630129914824,
      "grad_norm": 0.85546875,
      "learning_rate": 3.369439295415944e-05,
      "loss": 1.0513,
      "step": 51288
    },
    {
      "epoch": 4.412716166222146,
      "grad_norm": 4.40625,
      "learning_rate": 3.368463510243641e-05,
      "loss": 1.4271,
      "step": 51289
    },
    {
      "epoch": 4.412802202529467,
      "grad_norm": 1.0546875,
      "learning_rate": 3.367487861459528e-05,
      "loss": 1.4832,
      "step": 51290
    },
    {
      "epoch": 4.412888238836789,
      "grad_norm": 1.09375,
      "learning_rate": 3.3665123490664306e-05,
      "loss": 1.6771,
      "step": 51291
    },
    {
      "epoch": 4.412974275144111,
      "grad_norm": 1.640625,
      "learning_rate": 3.365536973067218e-05,
      "loss": 0.8413,
      "step": 51292
    },
    {
      "epoch": 4.413060311451432,
      "grad_norm": 3.75,
      "learning_rate": 3.364561733464744e-05,
      "loss": 2.2014,
      "step": 51293
    },
    {
      "epoch": 4.4131463477587545,
      "grad_norm": 1.21875,
      "learning_rate": 3.36358663026185e-05,
      "loss": 1.2599,
      "step": 51294
    },
    {
      "epoch": 4.413232384066076,
      "grad_norm": 1.0625,
      "learning_rate": 3.3626116634613954e-05,
      "loss": 0.7676,
      "step": 51295
    },
    {
      "epoch": 4.413318420373398,
      "grad_norm": 1.078125,
      "learning_rate": 3.3616368330662396e-05,
      "loss": 1.3378,
      "step": 51296
    },
    {
      "epoch": 4.413404456680719,
      "grad_norm": 1.1328125,
      "learning_rate": 3.360662139079218e-05,
      "loss": 0.9626,
      "step": 51297
    },
    {
      "epoch": 4.413490492988041,
      "grad_norm": 1.0078125,
      "learning_rate": 3.35968758150319e-05,
      "loss": 0.9848,
      "step": 51298
    },
    {
      "epoch": 4.413576529295363,
      "grad_norm": 1.3671875,
      "learning_rate": 3.3587131603410036e-05,
      "loss": 0.6947,
      "step": 51299
    },
    {
      "epoch": 4.413662565602684,
      "grad_norm": 1.078125,
      "learning_rate": 3.3577388755955176e-05,
      "loss": 0.5568,
      "step": 51300
    },
    {
      "epoch": 4.413748601910006,
      "grad_norm": 1.078125,
      "learning_rate": 3.356764727269568e-05,
      "loss": 1.4663,
      "step": 51301
    },
    {
      "epoch": 4.4138346382173275,
      "grad_norm": 1.0234375,
      "learning_rate": 3.355790715366008e-05,
      "loss": 0.4179,
      "step": 51302
    },
    {
      "epoch": 4.4139206745246495,
      "grad_norm": 0.93359375,
      "learning_rate": 3.354816839887692e-05,
      "loss": 0.9217,
      "step": 51303
    },
    {
      "epoch": 4.414006710831971,
      "grad_norm": 1.1875,
      "learning_rate": 3.353843100837462e-05,
      "loss": 1.382,
      "step": 51304
    },
    {
      "epoch": 4.414092747139293,
      "grad_norm": 0.828125,
      "learning_rate": 3.3528694982181696e-05,
      "loss": 0.9095,
      "step": 51305
    },
    {
      "epoch": 4.414178783446615,
      "grad_norm": 0.91796875,
      "learning_rate": 3.3518960320326644e-05,
      "loss": 0.8498,
      "step": 51306
    },
    {
      "epoch": 4.414264819753936,
      "grad_norm": 1.234375,
      "learning_rate": 3.350922702283787e-05,
      "loss": 1.9415,
      "step": 51307
    },
    {
      "epoch": 4.414350856061258,
      "grad_norm": 0.7734375,
      "learning_rate": 3.349949508974387e-05,
      "loss": 0.5618,
      "step": 51308
    },
    {
      "epoch": 4.414436892368579,
      "grad_norm": 312.0,
      "learning_rate": 3.3489764521073216e-05,
      "loss": 0.5512,
      "step": 51309
    },
    {
      "epoch": 4.414522928675901,
      "grad_norm": 0.9921875,
      "learning_rate": 3.3480035316854175e-05,
      "loss": 0.6473,
      "step": 51310
    },
    {
      "epoch": 4.4146089649832225,
      "grad_norm": 0.90625,
      "learning_rate": 3.347030747711527e-05,
      "loss": 1.0967,
      "step": 51311
    },
    {
      "epoch": 4.414695001290545,
      "grad_norm": 0.859375,
      "learning_rate": 3.346058100188509e-05,
      "loss": 0.9548,
      "step": 51312
    },
    {
      "epoch": 4.414781037597867,
      "grad_norm": 1.15625,
      "learning_rate": 3.34508558911919e-05,
      "loss": 1.1888,
      "step": 51313
    },
    {
      "epoch": 4.414867073905188,
      "grad_norm": 1.2578125,
      "learning_rate": 3.344113214506428e-05,
      "loss": 0.717,
      "step": 51314
    },
    {
      "epoch": 4.41495311021251,
      "grad_norm": 9.1875,
      "learning_rate": 3.343140976353054e-05,
      "loss": 1.2942,
      "step": 51315
    },
    {
      "epoch": 4.415039146519831,
      "grad_norm": 1.046875,
      "learning_rate": 3.342168874661916e-05,
      "loss": 1.1295,
      "step": 51316
    },
    {
      "epoch": 4.415125182827153,
      "grad_norm": 1.1953125,
      "learning_rate": 3.341196909435873e-05,
      "loss": 0.7727,
      "step": 51317
    },
    {
      "epoch": 4.415211219134474,
      "grad_norm": 1.125,
      "learning_rate": 3.340225080677739e-05,
      "loss": 1.3395,
      "step": 51318
    },
    {
      "epoch": 4.415297255441796,
      "grad_norm": 1.140625,
      "learning_rate": 3.339253388390379e-05,
      "loss": 1.1393,
      "step": 51319
    },
    {
      "epoch": 4.4153832917491185,
      "grad_norm": 1.234375,
      "learning_rate": 3.338281832576629e-05,
      "loss": 1.6569,
      "step": 51320
    },
    {
      "epoch": 4.41546932805644,
      "grad_norm": 1.9140625,
      "learning_rate": 3.3373104132393254e-05,
      "loss": 1.0389,
      "step": 51321
    },
    {
      "epoch": 4.415555364363762,
      "grad_norm": 1.3203125,
      "learning_rate": 3.336339130381316e-05,
      "loss": 1.4793,
      "step": 51322
    },
    {
      "epoch": 4.415641400671083,
      "grad_norm": 0.79296875,
      "learning_rate": 3.335367984005444e-05,
      "loss": 0.4207,
      "step": 51323
    },
    {
      "epoch": 4.415727436978405,
      "grad_norm": 0.8828125,
      "learning_rate": 3.334396974114534e-05,
      "loss": 0.524,
      "step": 51324
    },
    {
      "epoch": 4.415813473285727,
      "grad_norm": 1.1171875,
      "learning_rate": 3.333426100711451e-05,
      "loss": 0.922,
      "step": 51325
    },
    {
      "epoch": 4.415899509593048,
      "grad_norm": 2.046875,
      "learning_rate": 3.332455363799008e-05,
      "loss": 0.904,
      "step": 51326
    },
    {
      "epoch": 4.41598554590037,
      "grad_norm": 1.4140625,
      "learning_rate": 3.331484763380055e-05,
      "loss": 1.2165,
      "step": 51327
    },
    {
      "epoch": 4.4160715822076915,
      "grad_norm": 1.3984375,
      "learning_rate": 3.3305142994574434e-05,
      "loss": 1.4936,
      "step": 51328
    },
    {
      "epoch": 4.4161576185150135,
      "grad_norm": 1.1171875,
      "learning_rate": 3.329543972033994e-05,
      "loss": 0.9824,
      "step": 51329
    },
    {
      "epoch": 4.416243654822335,
      "grad_norm": 1.3203125,
      "learning_rate": 3.3285737811125486e-05,
      "loss": 1.0592,
      "step": 51330
    },
    {
      "epoch": 4.416329691129657,
      "grad_norm": 1.2734375,
      "learning_rate": 3.3276037266959603e-05,
      "loss": 1.5118,
      "step": 51331
    },
    {
      "epoch": 4.416415727436979,
      "grad_norm": 1.5,
      "learning_rate": 3.326633808787044e-05,
      "loss": 1.8923,
      "step": 51332
    },
    {
      "epoch": 4.4165017637443,
      "grad_norm": 1.140625,
      "learning_rate": 3.3256640273886415e-05,
      "loss": 1.0485,
      "step": 51333
    },
    {
      "epoch": 4.416587800051622,
      "grad_norm": 1.078125,
      "learning_rate": 3.324694382503607e-05,
      "loss": 1.2613,
      "step": 51334
    },
    {
      "epoch": 4.416673836358943,
      "grad_norm": 0.984375,
      "learning_rate": 3.323724874134748e-05,
      "loss": 1.0716,
      "step": 51335
    },
    {
      "epoch": 4.416759872666265,
      "grad_norm": 0.8515625,
      "learning_rate": 3.32275550228493e-05,
      "loss": 0.8327,
      "step": 51336
    },
    {
      "epoch": 4.4168459089735865,
      "grad_norm": 1.1796875,
      "learning_rate": 3.321786266956967e-05,
      "loss": 0.86,
      "step": 51337
    },
    {
      "epoch": 4.416931945280909,
      "grad_norm": 2.734375,
      "learning_rate": 3.3208171681537e-05,
      "loss": 1.1274,
      "step": 51338
    },
    {
      "epoch": 4.417017981588231,
      "grad_norm": 1.3125,
      "learning_rate": 3.319848205877973e-05,
      "loss": 1.2129,
      "step": 51339
    },
    {
      "epoch": 4.417104017895552,
      "grad_norm": 0.85546875,
      "learning_rate": 3.3188793801326e-05,
      "loss": 1.213,
      "step": 51340
    },
    {
      "epoch": 4.417190054202874,
      "grad_norm": 1.2734375,
      "learning_rate": 3.317910690920428e-05,
      "loss": 1.4992,
      "step": 51341
    },
    {
      "epoch": 4.417276090510195,
      "grad_norm": 41.5,
      "learning_rate": 3.3169421382443e-05,
      "loss": 1.2791,
      "step": 51342
    },
    {
      "epoch": 4.417362126817517,
      "grad_norm": 1.015625,
      "learning_rate": 3.315973722107024e-05,
      "loss": 0.6524,
      "step": 51343
    },
    {
      "epoch": 4.417448163124838,
      "grad_norm": 1.0234375,
      "learning_rate": 3.315005442511454e-05,
      "loss": 1.3863,
      "step": 51344
    },
    {
      "epoch": 4.41753419943216,
      "grad_norm": 1.0625,
      "learning_rate": 3.314037299460404e-05,
      "loss": 0.644,
      "step": 51345
    },
    {
      "epoch": 4.4176202357394825,
      "grad_norm": 1.1171875,
      "learning_rate": 3.313069292956716e-05,
      "loss": 1.3755,
      "step": 51346
    },
    {
      "epoch": 4.417706272046804,
      "grad_norm": 1.21875,
      "learning_rate": 3.312101423003233e-05,
      "loss": 1.09,
      "step": 51347
    },
    {
      "epoch": 4.417792308354126,
      "grad_norm": 0.9921875,
      "learning_rate": 3.311133689602758e-05,
      "loss": 1.1213,
      "step": 51348
    },
    {
      "epoch": 4.417878344661447,
      "grad_norm": 1.15625,
      "learning_rate": 3.3101660927581436e-05,
      "loss": 1.2693,
      "step": 51349
    },
    {
      "epoch": 4.417964380968769,
      "grad_norm": 1.1328125,
      "learning_rate": 3.309198632472216e-05,
      "loss": 1.4481,
      "step": 51350
    },
    {
      "epoch": 4.41805041727609,
      "grad_norm": 0.91796875,
      "learning_rate": 3.3082313087477947e-05,
      "loss": 0.6341,
      "step": 51351
    },
    {
      "epoch": 4.418136453583412,
      "grad_norm": 0.91796875,
      "learning_rate": 3.3072641215877166e-05,
      "loss": 1.0269,
      "step": 51352
    },
    {
      "epoch": 4.418222489890734,
      "grad_norm": 2.375,
      "learning_rate": 3.306297070994813e-05,
      "loss": 1.1528,
      "step": 51353
    },
    {
      "epoch": 4.4183085261980555,
      "grad_norm": 27.875,
      "learning_rate": 3.305330156971903e-05,
      "loss": 1.5865,
      "step": 51354
    },
    {
      "epoch": 4.4183945625053775,
      "grad_norm": 1.1484375,
      "learning_rate": 3.3043633795218356e-05,
      "loss": 1.1238,
      "step": 51355
    },
    {
      "epoch": 4.418480598812699,
      "grad_norm": 1.2109375,
      "learning_rate": 3.3033967386474075e-05,
      "loss": 1.1881,
      "step": 51356
    },
    {
      "epoch": 4.418566635120021,
      "grad_norm": 1.046875,
      "learning_rate": 3.3024302343514666e-05,
      "loss": 1.0451,
      "step": 51357
    },
    {
      "epoch": 4.418652671427342,
      "grad_norm": 1.015625,
      "learning_rate": 3.301463866636839e-05,
      "loss": 1.3298,
      "step": 51358
    },
    {
      "epoch": 4.418738707734664,
      "grad_norm": 1.0859375,
      "learning_rate": 3.3004976355063385e-05,
      "loss": 1.5923,
      "step": 51359
    },
    {
      "epoch": 4.418824744041986,
      "grad_norm": 1.125,
      "learning_rate": 3.299531540962803e-05,
      "loss": 0.8209,
      "step": 51360
    },
    {
      "epoch": 4.418910780349307,
      "grad_norm": 1.3359375,
      "learning_rate": 3.2985655830090664e-05,
      "loss": 1.9103,
      "step": 51361
    },
    {
      "epoch": 4.418996816656629,
      "grad_norm": 1.015625,
      "learning_rate": 3.297599761647929e-05,
      "loss": 1.1571,
      "step": 51362
    },
    {
      "epoch": 4.4190828529639505,
      "grad_norm": 1.2421875,
      "learning_rate": 3.296634076882232e-05,
      "loss": 1.3661,
      "step": 51363
    },
    {
      "epoch": 4.419168889271273,
      "grad_norm": 1.0,
      "learning_rate": 3.295668528714801e-05,
      "loss": 0.7749,
      "step": 51364
    },
    {
      "epoch": 4.419254925578594,
      "grad_norm": 1.25,
      "learning_rate": 3.294703117148451e-05,
      "loss": 1.2535,
      "step": 51365
    },
    {
      "epoch": 4.419340961885916,
      "grad_norm": 5.5,
      "learning_rate": 3.2937378421860233e-05,
      "loss": 0.9104,
      "step": 51366
    },
    {
      "epoch": 4.419426998193238,
      "grad_norm": 1.125,
      "learning_rate": 3.2927727038303156e-05,
      "loss": 0.8013,
      "step": 51367
    },
    {
      "epoch": 4.419513034500559,
      "grad_norm": 83.5,
      "learning_rate": 3.2918077020841594e-05,
      "loss": 1.1878,
      "step": 51368
    },
    {
      "epoch": 4.419599070807881,
      "grad_norm": 0.8359375,
      "learning_rate": 3.2908428369503976e-05,
      "loss": 0.5785,
      "step": 51369
    },
    {
      "epoch": 4.419685107115202,
      "grad_norm": 1.125,
      "learning_rate": 3.2898781084318266e-05,
      "loss": 0.6779,
      "step": 51370
    },
    {
      "epoch": 4.419771143422524,
      "grad_norm": 1.265625,
      "learning_rate": 3.288913516531272e-05,
      "loss": 1.2751,
      "step": 51371
    },
    {
      "epoch": 4.419857179729846,
      "grad_norm": 1.4375,
      "learning_rate": 3.287949061251566e-05,
      "loss": 1.32,
      "step": 51372
    },
    {
      "epoch": 4.419943216037168,
      "grad_norm": 5.96875,
      "learning_rate": 3.286984742595522e-05,
      "loss": 1.2852,
      "step": 51373
    },
    {
      "epoch": 4.42002925234449,
      "grad_norm": 0.8828125,
      "learning_rate": 3.286020560565972e-05,
      "loss": 0.5448,
      "step": 51374
    },
    {
      "epoch": 4.420115288651811,
      "grad_norm": 1.3828125,
      "learning_rate": 3.285056515165713e-05,
      "loss": 1.4627,
      "step": 51375
    },
    {
      "epoch": 4.420201324959133,
      "grad_norm": 2.734375,
      "learning_rate": 3.284092606397582e-05,
      "loss": 1.3237,
      "step": 51376
    },
    {
      "epoch": 4.420287361266454,
      "grad_norm": 1.2890625,
      "learning_rate": 3.283128834264404e-05,
      "loss": 1.3807,
      "step": 51377
    },
    {
      "epoch": 4.420373397573776,
      "grad_norm": 1.3984375,
      "learning_rate": 3.282165198768977e-05,
      "loss": 1.9484,
      "step": 51378
    },
    {
      "epoch": 4.420459433881097,
      "grad_norm": 0.9140625,
      "learning_rate": 3.281201699914132e-05,
      "loss": 0.4984,
      "step": 51379
    },
    {
      "epoch": 4.4205454701884195,
      "grad_norm": 1.125,
      "learning_rate": 3.280238337702696e-05,
      "loss": 1.4625,
      "step": 51380
    },
    {
      "epoch": 4.4206315064957415,
      "grad_norm": 1.2734375,
      "learning_rate": 3.27927511213747e-05,
      "loss": 1.0146,
      "step": 51381
    },
    {
      "epoch": 4.420717542803063,
      "grad_norm": 1.3203125,
      "learning_rate": 3.2783120232212696e-05,
      "loss": 1.8124,
      "step": 51382
    },
    {
      "epoch": 4.420803579110385,
      "grad_norm": 1304.0,
      "learning_rate": 3.2773490709569366e-05,
      "loss": 1.2435,
      "step": 51383
    },
    {
      "epoch": 4.420889615417706,
      "grad_norm": 1.328125,
      "learning_rate": 3.276386255347258e-05,
      "loss": 1.007,
      "step": 51384
    },
    {
      "epoch": 4.420975651725028,
      "grad_norm": 0.95703125,
      "learning_rate": 3.27542357639507e-05,
      "loss": 0.9119,
      "step": 51385
    },
    {
      "epoch": 4.421061688032349,
      "grad_norm": 95.5,
      "learning_rate": 3.274461034103171e-05,
      "loss": 1.3993,
      "step": 51386
    },
    {
      "epoch": 4.421147724339671,
      "grad_norm": 5.65625,
      "learning_rate": 3.273498628474386e-05,
      "loss": 1.2823,
      "step": 51387
    },
    {
      "epoch": 4.421233760646993,
      "grad_norm": 11.0625,
      "learning_rate": 3.272536359511541e-05,
      "loss": 0.9509,
      "step": 51388
    },
    {
      "epoch": 4.4213197969543145,
      "grad_norm": 1.4453125,
      "learning_rate": 3.271574227217433e-05,
      "loss": 1.1434,
      "step": 51389
    },
    {
      "epoch": 4.421405833261637,
      "grad_norm": 1.28125,
      "learning_rate": 3.270612231594877e-05,
      "loss": 1.5074,
      "step": 51390
    },
    {
      "epoch": 4.421491869568958,
      "grad_norm": 0.81640625,
      "learning_rate": 3.269650372646704e-05,
      "loss": 0.5355,
      "step": 51391
    },
    {
      "epoch": 4.42157790587628,
      "grad_norm": 1.3125,
      "learning_rate": 3.2686886503757005e-05,
      "loss": 0.7179,
      "step": 51392
    },
    {
      "epoch": 4.421663942183601,
      "grad_norm": 0.91796875,
      "learning_rate": 3.2677270647847035e-05,
      "loss": 1.0207,
      "step": 51393
    },
    {
      "epoch": 4.421749978490923,
      "grad_norm": 0.82421875,
      "learning_rate": 3.266765615876516e-05,
      "loss": 0.7024,
      "step": 51394
    },
    {
      "epoch": 4.421836014798245,
      "grad_norm": 1.1015625,
      "learning_rate": 3.2658043036539466e-05,
      "loss": 1.3538,
      "step": 51395
    },
    {
      "epoch": 4.421922051105566,
      "grad_norm": 1.0078125,
      "learning_rate": 3.2648431281198156e-05,
      "loss": 1.0666,
      "step": 51396
    },
    {
      "epoch": 4.422008087412888,
      "grad_norm": 1.2109375,
      "learning_rate": 3.263882089276921e-05,
      "loss": 1.0458,
      "step": 51397
    },
    {
      "epoch": 4.42209412372021,
      "grad_norm": 1.3515625,
      "learning_rate": 3.2629211871280816e-05,
      "loss": 1.2373,
      "step": 51398
    },
    {
      "epoch": 4.422180160027532,
      "grad_norm": 1.2890625,
      "learning_rate": 3.261960421676119e-05,
      "loss": 0.6239,
      "step": 51399
    },
    {
      "epoch": 4.422266196334853,
      "grad_norm": 1.046875,
      "learning_rate": 3.2609997929238247e-05,
      "loss": 1.4498,
      "step": 51400
    },
    {
      "epoch": 4.422352232642175,
      "grad_norm": 1.78125,
      "learning_rate": 3.260039300874013e-05,
      "loss": 1.3292,
      "step": 51401
    },
    {
      "epoch": 4.422438268949497,
      "grad_norm": 1.1484375,
      "learning_rate": 3.259078945529503e-05,
      "loss": 0.6058,
      "step": 51402
    },
    {
      "epoch": 4.422524305256818,
      "grad_norm": 1.125,
      "learning_rate": 3.258118726893089e-05,
      "loss": 1.0663,
      "step": 51403
    },
    {
      "epoch": 4.42261034156414,
      "grad_norm": 1.109375,
      "learning_rate": 3.257158644967584e-05,
      "loss": 1.1131,
      "step": 51404
    },
    {
      "epoch": 4.422696377871461,
      "grad_norm": 1.328125,
      "learning_rate": 3.2561986997558135e-05,
      "loss": 1.6462,
      "step": 51405
    },
    {
      "epoch": 4.4227824141787835,
      "grad_norm": 1.2734375,
      "learning_rate": 3.2552388912605594e-05,
      "loss": 1.3562,
      "step": 51406
    },
    {
      "epoch": 4.4228684504861056,
      "grad_norm": 1.046875,
      "learning_rate": 3.2542792194846405e-05,
      "loss": 1.1425,
      "step": 51407
    },
    {
      "epoch": 4.422954486793427,
      "grad_norm": 0.9453125,
      "learning_rate": 3.2533196844308725e-05,
      "loss": 1.3349,
      "step": 51408
    },
    {
      "epoch": 4.423040523100749,
      "grad_norm": 0.7578125,
      "learning_rate": 3.2523602861020355e-05,
      "loss": 0.4578,
      "step": 51409
    },
    {
      "epoch": 4.42312655940807,
      "grad_norm": 1.0078125,
      "learning_rate": 3.251401024500966e-05,
      "loss": 0.5477,
      "step": 51410
    },
    {
      "epoch": 4.423212595715392,
      "grad_norm": 0.8359375,
      "learning_rate": 3.2504418996304516e-05,
      "loss": 1.345,
      "step": 51411
    },
    {
      "epoch": 4.423298632022713,
      "grad_norm": 1.203125,
      "learning_rate": 3.2494829114933066e-05,
      "loss": 1.1411,
      "step": 51412
    },
    {
      "epoch": 4.423384668330035,
      "grad_norm": 1.0859375,
      "learning_rate": 3.2485240600923284e-05,
      "loss": 0.758,
      "step": 51413
    },
    {
      "epoch": 4.423470704637357,
      "grad_norm": 0.93359375,
      "learning_rate": 3.247565345430326e-05,
      "loss": 1.0407,
      "step": 51414
    },
    {
      "epoch": 4.4235567409446785,
      "grad_norm": 1.5625,
      "learning_rate": 3.2466067675101075e-05,
      "loss": 0.7773,
      "step": 51415
    },
    {
      "epoch": 4.423642777252001,
      "grad_norm": 219.0,
      "learning_rate": 3.245648326334461e-05,
      "loss": 1.1735,
      "step": 51416
    },
    {
      "epoch": 4.423728813559322,
      "grad_norm": 1.4609375,
      "learning_rate": 3.2446900219061995e-05,
      "loss": 2.104,
      "step": 51417
    },
    {
      "epoch": 4.423814849866644,
      "grad_norm": 0.984375,
      "learning_rate": 3.243731854228138e-05,
      "loss": 1.748,
      "step": 51418
    },
    {
      "epoch": 4.423900886173965,
      "grad_norm": 3.03125,
      "learning_rate": 3.242773823303058e-05,
      "loss": 1.3469,
      "step": 51419
    },
    {
      "epoch": 4.423986922481287,
      "grad_norm": 1.4609375,
      "learning_rate": 3.241815929133768e-05,
      "loss": 0.8068,
      "step": 51420
    },
    {
      "epoch": 4.424072958788609,
      "grad_norm": 1.640625,
      "learning_rate": 3.2408581717230875e-05,
      "loss": 1.5402,
      "step": 51421
    },
    {
      "epoch": 4.42415899509593,
      "grad_norm": 0.97265625,
      "learning_rate": 3.2399005510737875e-05,
      "loss": 1.0474,
      "step": 51422
    },
    {
      "epoch": 4.424245031403252,
      "grad_norm": 1.1953125,
      "learning_rate": 3.238943067188688e-05,
      "loss": 1.8718,
      "step": 51423
    },
    {
      "epoch": 4.424331067710574,
      "grad_norm": 1.046875,
      "learning_rate": 3.237985720070591e-05,
      "loss": 1.1198,
      "step": 51424
    },
    {
      "epoch": 4.424417104017896,
      "grad_norm": 249.0,
      "learning_rate": 3.237028509722284e-05,
      "loss": 0.9245,
      "step": 51425
    },
    {
      "epoch": 4.424503140325217,
      "grad_norm": 0.9296875,
      "learning_rate": 3.2360714361465816e-05,
      "loss": 1.1662,
      "step": 51426
    },
    {
      "epoch": 4.424589176632539,
      "grad_norm": 1.140625,
      "learning_rate": 3.23511449934627e-05,
      "loss": 1.7201,
      "step": 51427
    },
    {
      "epoch": 4.424675212939861,
      "grad_norm": 50.25,
      "learning_rate": 3.2341576993241476e-05,
      "loss": 0.9716,
      "step": 51428
    },
    {
      "epoch": 4.424761249247182,
      "grad_norm": 1.09375,
      "learning_rate": 3.233201036083033e-05,
      "loss": 1.8586,
      "step": 51429
    },
    {
      "epoch": 4.424847285554504,
      "grad_norm": 1.4765625,
      "learning_rate": 3.232244509625698e-05,
      "loss": 0.6707,
      "step": 51430
    },
    {
      "epoch": 4.424933321861825,
      "grad_norm": 1.265625,
      "learning_rate": 3.2312881199549505e-05,
      "loss": 0.6151,
      "step": 51431
    },
    {
      "epoch": 4.4250193581691475,
      "grad_norm": 1.0703125,
      "learning_rate": 3.230331867073599e-05,
      "loss": 0.8495,
      "step": 51432
    },
    {
      "epoch": 4.425105394476469,
      "grad_norm": 0.9765625,
      "learning_rate": 3.229375750984426e-05,
      "loss": 0.8406,
      "step": 51433
    },
    {
      "epoch": 4.425191430783791,
      "grad_norm": 1.03125,
      "learning_rate": 3.2284197716902276e-05,
      "loss": 0.6289,
      "step": 51434
    },
    {
      "epoch": 4.425277467091113,
      "grad_norm": 0.953125,
      "learning_rate": 3.2274639291938136e-05,
      "loss": 0.6148,
      "step": 51435
    },
    {
      "epoch": 4.425363503398434,
      "grad_norm": 5.9375,
      "learning_rate": 3.2265082234979646e-05,
      "loss": 0.7542,
      "step": 51436
    },
    {
      "epoch": 4.425449539705756,
      "grad_norm": 5.4375,
      "learning_rate": 3.22555265460549e-05,
      "loss": 1.3402,
      "step": 51437
    },
    {
      "epoch": 4.425535576013077,
      "grad_norm": 1.5078125,
      "learning_rate": 3.224597222519166e-05,
      "loss": 1.4331,
      "step": 51438
    },
    {
      "epoch": 4.425621612320399,
      "grad_norm": 4.46875,
      "learning_rate": 3.2236419272417994e-05,
      "loss": 1.2273,
      "step": 51439
    },
    {
      "epoch": 4.425707648627721,
      "grad_norm": 1.15625,
      "learning_rate": 3.222686768776195e-05,
      "loss": 0.9437,
      "step": 51440
    },
    {
      "epoch": 4.4257936849350425,
      "grad_norm": 2.28125,
      "learning_rate": 3.221731747125117e-05,
      "loss": 0.9004,
      "step": 51441
    },
    {
      "epoch": 4.425879721242365,
      "grad_norm": 1.4140625,
      "learning_rate": 3.2207768622913856e-05,
      "loss": 1.2159,
      "step": 51442
    },
    {
      "epoch": 4.425965757549686,
      "grad_norm": 1.15625,
      "learning_rate": 3.219822114277787e-05,
      "loss": 0.7744,
      "step": 51443
    },
    {
      "epoch": 4.426051793857008,
      "grad_norm": 0.94921875,
      "learning_rate": 3.218867503087103e-05,
      "loss": 1.214,
      "step": 51444
    },
    {
      "epoch": 4.426137830164329,
      "grad_norm": 1.0703125,
      "learning_rate": 3.2179130287221304e-05,
      "loss": 1.0013,
      "step": 51445
    },
    {
      "epoch": 4.426223866471651,
      "grad_norm": 1.1796875,
      "learning_rate": 3.2169586911856676e-05,
      "loss": 0.9201,
      "step": 51446
    },
    {
      "epoch": 4.426309902778973,
      "grad_norm": 1.2421875,
      "learning_rate": 3.216004490480501e-05,
      "loss": 0.894,
      "step": 51447
    },
    {
      "epoch": 4.426395939086294,
      "grad_norm": 0.88671875,
      "learning_rate": 3.215050426609428e-05,
      "loss": 0.7539,
      "step": 51448
    },
    {
      "epoch": 4.426481975393616,
      "grad_norm": 0.99609375,
      "learning_rate": 3.214096499575225e-05,
      "loss": 1.4062,
      "step": 51449
    },
    {
      "epoch": 4.426568011700938,
      "grad_norm": 1.171875,
      "learning_rate": 3.213142709380695e-05,
      "loss": 1.5921,
      "step": 51450
    },
    {
      "epoch": 4.42665404800826,
      "grad_norm": 17.25,
      "learning_rate": 3.212189056028625e-05,
      "loss": 1.1,
      "step": 51451
    },
    {
      "epoch": 4.426740084315581,
      "grad_norm": 1.0859375,
      "learning_rate": 3.211235539521801e-05,
      "loss": 1.1452,
      "step": 51452
    },
    {
      "epoch": 4.426826120622903,
      "grad_norm": 0.90625,
      "learning_rate": 3.210282159863015e-05,
      "loss": 0.818,
      "step": 51453
    },
    {
      "epoch": 4.426912156930225,
      "grad_norm": 1.1015625,
      "learning_rate": 3.209328917055049e-05,
      "loss": 1.8106,
      "step": 51454
    },
    {
      "epoch": 4.426998193237546,
      "grad_norm": 1.0234375,
      "learning_rate": 3.2083758111006945e-05,
      "loss": 1.571,
      "step": 51455
    },
    {
      "epoch": 4.427084229544868,
      "grad_norm": 1.234375,
      "learning_rate": 3.207422842002749e-05,
      "loss": 0.6074,
      "step": 51456
    },
    {
      "epoch": 4.427170265852189,
      "grad_norm": 5.6875,
      "learning_rate": 3.2064700097639844e-05,
      "loss": 1.5257,
      "step": 51457
    },
    {
      "epoch": 4.4272563021595115,
      "grad_norm": 1.3359375,
      "learning_rate": 3.2055173143871904e-05,
      "loss": 1.3934,
      "step": 51458
    },
    {
      "epoch": 4.427342338466833,
      "grad_norm": 1.0703125,
      "learning_rate": 3.2045647558751666e-05,
      "loss": 1.2215,
      "step": 51459
    },
    {
      "epoch": 4.427428374774155,
      "grad_norm": 1.3359375,
      "learning_rate": 3.203612334230682e-05,
      "loss": 1.2431,
      "step": 51460
    },
    {
      "epoch": 4.427514411081477,
      "grad_norm": 0.85546875,
      "learning_rate": 3.202660049456535e-05,
      "loss": 0.7066,
      "step": 51461
    },
    {
      "epoch": 4.427600447388798,
      "grad_norm": 1.078125,
      "learning_rate": 3.201707901555506e-05,
      "loss": 1.0539,
      "step": 51462
    },
    {
      "epoch": 4.42768648369612,
      "grad_norm": 1.078125,
      "learning_rate": 3.2007558905303766e-05,
      "loss": 0.7394,
      "step": 51463
    },
    {
      "epoch": 4.427772520003441,
      "grad_norm": 0.94140625,
      "learning_rate": 3.199804016383928e-05,
      "loss": 1.817,
      "step": 51464
    },
    {
      "epoch": 4.427858556310763,
      "grad_norm": 1.8828125,
      "learning_rate": 3.198852279118963e-05,
      "loss": 1.149,
      "step": 51465
    },
    {
      "epoch": 4.4279445926180845,
      "grad_norm": 0.98046875,
      "learning_rate": 3.197900678738247e-05,
      "loss": 1.1976,
      "step": 51466
    },
    {
      "epoch": 4.4280306289254066,
      "grad_norm": 1.1953125,
      "learning_rate": 3.196949215244577e-05,
      "loss": 1.0776,
      "step": 51467
    },
    {
      "epoch": 4.428116665232729,
      "grad_norm": 1.0078125,
      "learning_rate": 3.195997888640717e-05,
      "loss": 0.8628,
      "step": 51468
    },
    {
      "epoch": 4.42820270154005,
      "grad_norm": 1.2890625,
      "learning_rate": 3.19504669892946e-05,
      "loss": 0.6794,
      "step": 51469
    },
    {
      "epoch": 4.428288737847372,
      "grad_norm": 1.2890625,
      "learning_rate": 3.194095646113598e-05,
      "loss": 0.9536,
      "step": 51470
    },
    {
      "epoch": 4.428374774154693,
      "grad_norm": 1.1171875,
      "learning_rate": 3.19314473019589e-05,
      "loss": 0.7169,
      "step": 51471
    },
    {
      "epoch": 4.428460810462015,
      "grad_norm": 1.21875,
      "learning_rate": 3.19219395117914e-05,
      "loss": 1.6135,
      "step": 51472
    },
    {
      "epoch": 4.428546846769336,
      "grad_norm": 22.125,
      "learning_rate": 3.191243309066116e-05,
      "loss": 0.8861,
      "step": 51473
    },
    {
      "epoch": 4.428632883076658,
      "grad_norm": 1.4765625,
      "learning_rate": 3.190292803859601e-05,
      "loss": 1.4722,
      "step": 51474
    },
    {
      "epoch": 4.42871891938398,
      "grad_norm": 1.34375,
      "learning_rate": 3.1893424355623755e-05,
      "loss": 1.5906,
      "step": 51475
    },
    {
      "epoch": 4.428804955691302,
      "grad_norm": 1.1171875,
      "learning_rate": 3.1883922041772265e-05,
      "loss": 0.8257,
      "step": 51476
    },
    {
      "epoch": 4.428890991998624,
      "grad_norm": 1.0234375,
      "learning_rate": 3.1874421097069175e-05,
      "loss": 0.9745,
      "step": 51477
    },
    {
      "epoch": 4.428977028305945,
      "grad_norm": 1.2890625,
      "learning_rate": 3.186492152154241e-05,
      "loss": 1.4229,
      "step": 51478
    },
    {
      "epoch": 4.429063064613267,
      "grad_norm": 1.1171875,
      "learning_rate": 3.185542331521962e-05,
      "loss": 0.876,
      "step": 51479
    },
    {
      "epoch": 4.429149100920588,
      "grad_norm": 1.046875,
      "learning_rate": 3.184592647812873e-05,
      "loss": 0.5708,
      "step": 51480
    },
    {
      "epoch": 4.42923513722791,
      "grad_norm": 0.8984375,
      "learning_rate": 3.1836431010297485e-05,
      "loss": 1.3114,
      "step": 51481
    },
    {
      "epoch": 4.429321173535232,
      "grad_norm": 0.94921875,
      "learning_rate": 3.182693691175348e-05,
      "loss": 1.5485,
      "step": 51482
    },
    {
      "epoch": 4.429407209842553,
      "grad_norm": 2.265625,
      "learning_rate": 3.181744418252474e-05,
      "loss": 1.587,
      "step": 51483
    },
    {
      "epoch": 4.4294932461498755,
      "grad_norm": 1.2109375,
      "learning_rate": 3.1807952822638933e-05,
      "loss": 1.0637,
      "step": 51484
    },
    {
      "epoch": 4.429579282457197,
      "grad_norm": 1.1015625,
      "learning_rate": 3.179846283212373e-05,
      "loss": 1.0455,
      "step": 51485
    },
    {
      "epoch": 4.429665318764519,
      "grad_norm": 1.1640625,
      "learning_rate": 3.178897421100707e-05,
      "loss": 0.6573,
      "step": 51486
    },
    {
      "epoch": 4.42975135507184,
      "grad_norm": 0.8515625,
      "learning_rate": 3.177948695931654e-05,
      "loss": 1.0488,
      "step": 51487
    },
    {
      "epoch": 4.429837391379162,
      "grad_norm": 1.2890625,
      "learning_rate": 3.177000107707989e-05,
      "loss": 1.0219,
      "step": 51488
    },
    {
      "epoch": 4.429923427686484,
      "grad_norm": 1.1953125,
      "learning_rate": 3.176051656432505e-05,
      "loss": 1.7088,
      "step": 51489
    },
    {
      "epoch": 4.430009463993805,
      "grad_norm": 1.2421875,
      "learning_rate": 3.1751033421079545e-05,
      "loss": 0.8766,
      "step": 51490
    },
    {
      "epoch": 4.430095500301127,
      "grad_norm": 8.1875,
      "learning_rate": 3.1741551647371195e-05,
      "loss": 1.0768,
      "step": 51491
    },
    {
      "epoch": 4.4301815366084485,
      "grad_norm": 24.5,
      "learning_rate": 3.17320712432278e-05,
      "loss": 1.2928,
      "step": 51492
    },
    {
      "epoch": 4.430267572915771,
      "grad_norm": 1.1328125,
      "learning_rate": 3.172259220867696e-05,
      "loss": 0.8545,
      "step": 51493
    },
    {
      "epoch": 4.430353609223092,
      "grad_norm": 1.1484375,
      "learning_rate": 3.171311454374648e-05,
      "loss": 0.552,
      "step": 51494
    },
    {
      "epoch": 4.430439645530414,
      "grad_norm": 0.93359375,
      "learning_rate": 3.1703638248464115e-05,
      "loss": 1.2778,
      "step": 51495
    },
    {
      "epoch": 4.430525681837736,
      "grad_norm": 1.046875,
      "learning_rate": 3.169416332285746e-05,
      "loss": 1.8904,
      "step": 51496
    },
    {
      "epoch": 4.430611718145057,
      "grad_norm": 1.2109375,
      "learning_rate": 3.168468976695438e-05,
      "loss": 0.8086,
      "step": 51497
    },
    {
      "epoch": 4.430697754452379,
      "grad_norm": 1.1484375,
      "learning_rate": 3.16752175807824e-05,
      "loss": 1.1905,
      "step": 51498
    },
    {
      "epoch": 4.4307837907597,
      "grad_norm": 2.03125,
      "learning_rate": 3.166574676436934e-05,
      "loss": 0.6259,
      "step": 51499
    },
    {
      "epoch": 4.430869827067022,
      "grad_norm": 0.8828125,
      "learning_rate": 3.165627731774296e-05,
      "loss": 0.6464,
      "step": 51500
    },
    {
      "epoch": 4.4309558633743436,
      "grad_norm": 0.93359375,
      "learning_rate": 3.164680924093077e-05,
      "loss": 0.5291,
      "step": 51501
    },
    {
      "epoch": 4.431041899681666,
      "grad_norm": 2.34375,
      "learning_rate": 3.163734253396061e-05,
      "loss": 1.2754,
      "step": 51502
    },
    {
      "epoch": 4.431127935988988,
      "grad_norm": 252.0,
      "learning_rate": 3.162787719686022e-05,
      "loss": 1.5578,
      "step": 51503
    },
    {
      "epoch": 4.431213972296309,
      "grad_norm": 1.2421875,
      "learning_rate": 3.161841322965714e-05,
      "loss": 1.316,
      "step": 51504
    },
    {
      "epoch": 4.431300008603631,
      "grad_norm": 1.0859375,
      "learning_rate": 3.160895063237906e-05,
      "loss": 1.0167,
      "step": 51505
    },
    {
      "epoch": 4.431386044910952,
      "grad_norm": 1.6875,
      "learning_rate": 3.15994894050538e-05,
      "loss": 1.6164,
      "step": 51506
    },
    {
      "epoch": 4.431472081218274,
      "grad_norm": 1.1953125,
      "learning_rate": 3.1590029547708844e-05,
      "loss": 1.3785,
      "step": 51507
    },
    {
      "epoch": 4.431558117525595,
      "grad_norm": 1.34375,
      "learning_rate": 3.1580571060372044e-05,
      "loss": 1.6232,
      "step": 51508
    },
    {
      "epoch": 4.431644153832917,
      "grad_norm": 0.82421875,
      "learning_rate": 3.157111394307089e-05,
      "loss": 0.6526,
      "step": 51509
    },
    {
      "epoch": 4.4317301901402395,
      "grad_norm": 1.5,
      "learning_rate": 3.156165819583307e-05,
      "loss": 1.5834,
      "step": 51510
    },
    {
      "epoch": 4.431816226447561,
      "grad_norm": 1.640625,
      "learning_rate": 3.1552203818686455e-05,
      "loss": 0.8455,
      "step": 51511
    },
    {
      "epoch": 4.431902262754883,
      "grad_norm": 1.0859375,
      "learning_rate": 3.1542750811658414e-05,
      "loss": 1.114,
      "step": 51512
    },
    {
      "epoch": 4.431988299062204,
      "grad_norm": 1.3203125,
      "learning_rate": 3.1533299174776706e-05,
      "loss": 0.8177,
      "step": 51513
    },
    {
      "epoch": 4.432074335369526,
      "grad_norm": 1.5078125,
      "learning_rate": 3.152384890806909e-05,
      "loss": 1.1658,
      "step": 51514
    },
    {
      "epoch": 4.432160371676847,
      "grad_norm": 0.95703125,
      "learning_rate": 3.151440001156303e-05,
      "loss": 1.6889,
      "step": 51515
    },
    {
      "epoch": 4.432246407984169,
      "grad_norm": 556.0,
      "learning_rate": 3.1504952485286175e-05,
      "loss": 1.1183,
      "step": 51516
    },
    {
      "epoch": 4.432332444291491,
      "grad_norm": 0.9921875,
      "learning_rate": 3.1495506329266345e-05,
      "loss": 1.2525,
      "step": 51517
    },
    {
      "epoch": 4.4324184805988125,
      "grad_norm": 1.171875,
      "learning_rate": 3.1486061543530956e-05,
      "loss": 0.5376,
      "step": 51518
    },
    {
      "epoch": 4.432504516906135,
      "grad_norm": 0.99609375,
      "learning_rate": 3.14766181281077e-05,
      "loss": 0.6662,
      "step": 51519
    },
    {
      "epoch": 4.432590553213456,
      "grad_norm": 0.90625,
      "learning_rate": 3.146717608302424e-05,
      "loss": 0.9838,
      "step": 51520
    },
    {
      "epoch": 4.432676589520778,
      "grad_norm": 1.2109375,
      "learning_rate": 3.145773540830815e-05,
      "loss": 0.9766,
      "step": 51521
    },
    {
      "epoch": 4.432762625828099,
      "grad_norm": 1.328125,
      "learning_rate": 3.144829610398714e-05,
      "loss": 0.9303,
      "step": 51522
    },
    {
      "epoch": 4.432848662135421,
      "grad_norm": 1.3125,
      "learning_rate": 3.143885817008868e-05,
      "loss": 1.9103,
      "step": 51523
    },
    {
      "epoch": 4.432934698442743,
      "grad_norm": 1.0859375,
      "learning_rate": 3.1429421606640475e-05,
      "loss": 1.5615,
      "step": 51524
    },
    {
      "epoch": 4.433020734750064,
      "grad_norm": 1.046875,
      "learning_rate": 3.1419986413669996e-05,
      "loss": 1.0625,
      "step": 51525
    },
    {
      "epoch": 4.433106771057386,
      "grad_norm": 1.46875,
      "learning_rate": 3.141055259120496e-05,
      "loss": 1.1122,
      "step": 51526
    },
    {
      "epoch": 4.4331928073647076,
      "grad_norm": 1.0234375,
      "learning_rate": 3.140112013927299e-05,
      "loss": 1.0526,
      "step": 51527
    },
    {
      "epoch": 4.43327884367203,
      "grad_norm": 113.5,
      "learning_rate": 3.1391689057901515e-05,
      "loss": 0.8409,
      "step": 51528
    },
    {
      "epoch": 4.433364879979352,
      "grad_norm": 1.375,
      "learning_rate": 3.138225934711819e-05,
      "loss": 1.2242,
      "step": 51529
    },
    {
      "epoch": 4.433450916286673,
      "grad_norm": 0.921875,
      "learning_rate": 3.137283100695076e-05,
      "loss": 1.2556,
      "step": 51530
    },
    {
      "epoch": 4.433536952593995,
      "grad_norm": 24.0,
      "learning_rate": 3.136340403742649e-05,
      "loss": 0.8846,
      "step": 51531
    },
    {
      "epoch": 4.433622988901316,
      "grad_norm": 1.2109375,
      "learning_rate": 3.135397843857318e-05,
      "loss": 1.4232,
      "step": 51532
    },
    {
      "epoch": 4.433709025208638,
      "grad_norm": 6.96875,
      "learning_rate": 3.134455421041843e-05,
      "loss": 1.1495,
      "step": 51533
    },
    {
      "epoch": 4.433795061515959,
      "grad_norm": 1.2421875,
      "learning_rate": 3.13351313529896e-05,
      "loss": 1.1044,
      "step": 51534
    },
    {
      "epoch": 4.433881097823281,
      "grad_norm": 1.6796875,
      "learning_rate": 3.132570986631433e-05,
      "loss": 1.338,
      "step": 51535
    },
    {
      "epoch": 4.4339671341306035,
      "grad_norm": 1.1328125,
      "learning_rate": 3.1316289750420336e-05,
      "loss": 1.6775,
      "step": 51536
    },
    {
      "epoch": 4.434053170437925,
      "grad_norm": 540.0,
      "learning_rate": 3.1306871005334916e-05,
      "loss": 0.8636,
      "step": 51537
    },
    {
      "epoch": 4.434139206745247,
      "grad_norm": 2.125,
      "learning_rate": 3.129745363108588e-05,
      "loss": 1.6019,
      "step": 51538
    },
    {
      "epoch": 4.434225243052568,
      "grad_norm": 1.375,
      "learning_rate": 3.128803762770049e-05,
      "loss": 1.2151,
      "step": 51539
    },
    {
      "epoch": 4.43431127935989,
      "grad_norm": 1.1015625,
      "learning_rate": 3.127862299520645e-05,
      "loss": 0.9512,
      "step": 51540
    },
    {
      "epoch": 4.434397315667211,
      "grad_norm": 0.984375,
      "learning_rate": 3.12692097336314e-05,
      "loss": 0.8013,
      "step": 51541
    },
    {
      "epoch": 4.434483351974533,
      "grad_norm": 0.890625,
      "learning_rate": 3.125979784300265e-05,
      "loss": 0.9344,
      "step": 51542
    },
    {
      "epoch": 4.434569388281855,
      "grad_norm": 1.0078125,
      "learning_rate": 3.1250387323347786e-05,
      "loss": 0.8885,
      "step": 51543
    },
    {
      "epoch": 4.4346554245891765,
      "grad_norm": 1.2421875,
      "learning_rate": 3.124097817469451e-05,
      "loss": 1.4365,
      "step": 51544
    },
    {
      "epoch": 4.434741460896499,
      "grad_norm": 1.2109375,
      "learning_rate": 3.1231570397070085e-05,
      "loss": 0.8383,
      "step": 51545
    },
    {
      "epoch": 4.43482749720382,
      "grad_norm": 1.140625,
      "learning_rate": 3.122216399050215e-05,
      "loss": 0.5321,
      "step": 51546
    },
    {
      "epoch": 4.434913533511142,
      "grad_norm": 3040.0,
      "learning_rate": 3.1212758955018297e-05,
      "loss": 1.11,
      "step": 51547
    },
    {
      "epoch": 4.434999569818463,
      "grad_norm": 1.21875,
      "learning_rate": 3.120335529064589e-05,
      "loss": 1.1102,
      "step": 51548
    },
    {
      "epoch": 4.435085606125785,
      "grad_norm": 17.125,
      "learning_rate": 3.119395299741251e-05,
      "loss": 0.9013,
      "step": 51549
    },
    {
      "epoch": 4.435171642433107,
      "grad_norm": 0.8203125,
      "learning_rate": 3.118455207534565e-05,
      "loss": 0.5211,
      "step": 51550
    },
    {
      "epoch": 4.435257678740428,
      "grad_norm": 0.95703125,
      "learning_rate": 3.1175152524472726e-05,
      "loss": 0.7307,
      "step": 51551
    },
    {
      "epoch": 4.43534371504775,
      "grad_norm": 0.94140625,
      "learning_rate": 3.116575434482139e-05,
      "loss": 0.5254,
      "step": 51552
    },
    {
      "epoch": 4.435429751355072,
      "grad_norm": 1.078125,
      "learning_rate": 3.115635753641899e-05,
      "loss": 1.207,
      "step": 51553
    },
    {
      "epoch": 4.435515787662394,
      "grad_norm": 0.96484375,
      "learning_rate": 3.114696209929302e-05,
      "loss": 1.2906,
      "step": 51554
    },
    {
      "epoch": 4.435601823969715,
      "grad_norm": 1.34375,
      "learning_rate": 3.113756803347112e-05,
      "loss": 1.2126,
      "step": 51555
    },
    {
      "epoch": 4.435687860277037,
      "grad_norm": 1.046875,
      "learning_rate": 3.1128175338980495e-05,
      "loss": 1.5185,
      "step": 51556
    },
    {
      "epoch": 4.435773896584359,
      "grad_norm": 1.1640625,
      "learning_rate": 3.111878401584889e-05,
      "loss": 1.0731,
      "step": 51557
    },
    {
      "epoch": 4.43585993289168,
      "grad_norm": 1.375,
      "learning_rate": 3.110939406410357e-05,
      "loss": 0.7081,
      "step": 51558
    },
    {
      "epoch": 4.435945969199002,
      "grad_norm": 1.140625,
      "learning_rate": 3.1100005483772064e-05,
      "loss": 1.4425,
      "step": 51559
    },
    {
      "epoch": 4.436032005506323,
      "grad_norm": 1.609375,
      "learning_rate": 3.1090618274881964e-05,
      "loss": 1.0108,
      "step": 51560
    },
    {
      "epoch": 4.436118041813645,
      "grad_norm": 1.34375,
      "learning_rate": 3.1081232437460524e-05,
      "loss": 1.195,
      "step": 51561
    },
    {
      "epoch": 4.4362040781209675,
      "grad_norm": 0.9296875,
      "learning_rate": 3.107184797153523e-05,
      "loss": 0.8678,
      "step": 51562
    },
    {
      "epoch": 4.436290114428289,
      "grad_norm": 1.0546875,
      "learning_rate": 3.106246487713371e-05,
      "loss": 1.2699,
      "step": 51563
    },
    {
      "epoch": 4.436376150735611,
      "grad_norm": 109.0,
      "learning_rate": 3.105308315428318e-05,
      "loss": 1.1217,
      "step": 51564
    },
    {
      "epoch": 4.436462187042932,
      "grad_norm": 1.09375,
      "learning_rate": 3.104370280301128e-05,
      "loss": 0.6391,
      "step": 51565
    },
    {
      "epoch": 4.436548223350254,
      "grad_norm": 1.15625,
      "learning_rate": 3.1034323823345256e-05,
      "loss": 1.3601,
      "step": 51566
    },
    {
      "epoch": 4.436634259657575,
      "grad_norm": 1.109375,
      "learning_rate": 3.10249462153126e-05,
      "loss": 1.4557,
      "step": 51567
    },
    {
      "epoch": 4.436720295964897,
      "grad_norm": 1.2109375,
      "learning_rate": 3.1015569978940836e-05,
      "loss": 1.0093,
      "step": 51568
    },
    {
      "epoch": 4.436806332272219,
      "grad_norm": 1.265625,
      "learning_rate": 3.1006195114257285e-05,
      "loss": 1.1188,
      "step": 51569
    },
    {
      "epoch": 4.4368923685795405,
      "grad_norm": 1.0234375,
      "learning_rate": 3.099682162128936e-05,
      "loss": 1.2975,
      "step": 51570
    },
    {
      "epoch": 4.436978404886863,
      "grad_norm": 1.0390625,
      "learning_rate": 3.0987449500064604e-05,
      "loss": 1.1378,
      "step": 51571
    },
    {
      "epoch": 4.437064441194184,
      "grad_norm": 1.6015625,
      "learning_rate": 3.097807875061026e-05,
      "loss": 0.7704,
      "step": 51572
    },
    {
      "epoch": 4.437150477501506,
      "grad_norm": 1.5703125,
      "learning_rate": 3.0968709372953816e-05,
      "loss": 0.8198,
      "step": 51573
    },
    {
      "epoch": 4.437236513808827,
      "grad_norm": 1.2421875,
      "learning_rate": 3.0959341367122754e-05,
      "loss": 0.7246,
      "step": 51574
    },
    {
      "epoch": 4.437322550116149,
      "grad_norm": 1.3046875,
      "learning_rate": 3.0949974733144315e-05,
      "loss": 0.9042,
      "step": 51575
    },
    {
      "epoch": 4.437408586423471,
      "grad_norm": 1.078125,
      "learning_rate": 3.094060947104599e-05,
      "loss": 1.2072,
      "step": 51576
    },
    {
      "epoch": 4.437494622730792,
      "grad_norm": 1.109375,
      "learning_rate": 3.093124558085525e-05,
      "loss": 1.2191,
      "step": 51577
    },
    {
      "epoch": 4.437580659038114,
      "grad_norm": 1.359375,
      "learning_rate": 3.09218830625993e-05,
      "loss": 1.2206,
      "step": 51578
    },
    {
      "epoch": 4.437666695345436,
      "grad_norm": 2.15625,
      "learning_rate": 3.091252191630567e-05,
      "loss": 1.6574,
      "step": 51579
    },
    {
      "epoch": 4.437752731652758,
      "grad_norm": 1.1484375,
      "learning_rate": 3.090316214200156e-05,
      "loss": 0.6663,
      "step": 51580
    },
    {
      "epoch": 4.437838767960079,
      "grad_norm": 1.03125,
      "learning_rate": 3.089380373971451e-05,
      "loss": 1.2948,
      "step": 51581
    },
    {
      "epoch": 4.437924804267401,
      "grad_norm": 1.46875,
      "learning_rate": 3.0884446709471935e-05,
      "loss": 1.0843,
      "step": 51582
    },
    {
      "epoch": 4.438010840574723,
      "grad_norm": 0.79296875,
      "learning_rate": 3.087509105130104e-05,
      "loss": 0.5873,
      "step": 51583
    },
    {
      "epoch": 4.438096876882044,
      "grad_norm": 0.88671875,
      "learning_rate": 3.0865736765229246e-05,
      "loss": 1.2421,
      "step": 51584
    },
    {
      "epoch": 4.438182913189366,
      "grad_norm": 0.8671875,
      "learning_rate": 3.085638385128403e-05,
      "loss": 0.9542,
      "step": 51585
    },
    {
      "epoch": 4.438268949496687,
      "grad_norm": 1.3828125,
      "learning_rate": 3.08470323094926e-05,
      "loss": 1.085,
      "step": 51586
    },
    {
      "epoch": 4.4383549858040094,
      "grad_norm": 0.92578125,
      "learning_rate": 3.08376821398823e-05,
      "loss": 0.8896,
      "step": 51587
    },
    {
      "epoch": 4.438441022111331,
      "grad_norm": 1.390625,
      "learning_rate": 3.082833334248059e-05,
      "loss": 1.4239,
      "step": 51588
    },
    {
      "epoch": 4.438527058418653,
      "grad_norm": 1.4296875,
      "learning_rate": 3.0818985917314744e-05,
      "loss": 1.5373,
      "step": 51589
    },
    {
      "epoch": 4.438613094725975,
      "grad_norm": 0.8203125,
      "learning_rate": 3.0809639864412156e-05,
      "loss": 0.6249,
      "step": 51590
    },
    {
      "epoch": 4.438699131033296,
      "grad_norm": 1.0,
      "learning_rate": 3.080029518380001e-05,
      "loss": 1.2918,
      "step": 51591
    },
    {
      "epoch": 4.438785167340618,
      "grad_norm": 712.0,
      "learning_rate": 3.0790951875505794e-05,
      "loss": 1.5008,
      "step": 51592
    },
    {
      "epoch": 4.438871203647939,
      "grad_norm": 0.98046875,
      "learning_rate": 3.078160993955681e-05,
      "loss": 0.9984,
      "step": 51593
    },
    {
      "epoch": 4.438957239955261,
      "grad_norm": 1.2890625,
      "learning_rate": 3.077226937598032e-05,
      "loss": 1.6813,
      "step": 51594
    },
    {
      "epoch": 4.439043276262582,
      "grad_norm": 1.5859375,
      "learning_rate": 3.076293018480375e-05,
      "loss": 1.8117,
      "step": 51595
    },
    {
      "epoch": 4.4391293125699045,
      "grad_norm": 8.375,
      "learning_rate": 3.0753592366054295e-05,
      "loss": 0.999,
      "step": 51596
    },
    {
      "epoch": 4.439215348877227,
      "grad_norm": 1.078125,
      "learning_rate": 3.074425591975927e-05,
      "loss": 0.9202,
      "step": 51597
    },
    {
      "epoch": 4.439301385184548,
      "grad_norm": 5.8125,
      "learning_rate": 3.073492084594615e-05,
      "loss": 0.9769,
      "step": 51598
    },
    {
      "epoch": 4.43938742149187,
      "grad_norm": 1.28125,
      "learning_rate": 3.072558714464208e-05,
      "loss": 1.78,
      "step": 51599
    },
    {
      "epoch": 4.439473457799191,
      "grad_norm": 1.0234375,
      "learning_rate": 3.071625481587431e-05,
      "loss": 1.7779,
      "step": 51600
    },
    {
      "epoch": 4.439559494106513,
      "grad_norm": 1.4296875,
      "learning_rate": 3.0706923859670386e-05,
      "loss": 1.6058,
      "step": 51601
    },
    {
      "epoch": 4.439645530413834,
      "grad_norm": 1.0625,
      "learning_rate": 3.069759427605734e-05,
      "loss": 1.0312,
      "step": 51602
    },
    {
      "epoch": 4.439731566721156,
      "grad_norm": 5.875,
      "learning_rate": 3.0688266065062535e-05,
      "loss": 1.3218,
      "step": 51603
    },
    {
      "epoch": 4.439817603028478,
      "grad_norm": 1.15625,
      "learning_rate": 3.067893922671333e-05,
      "loss": 1.1665,
      "step": 51604
    },
    {
      "epoch": 4.4399036393358,
      "grad_norm": 2.5625,
      "learning_rate": 3.066961376103689e-05,
      "loss": 1.4705,
      "step": 51605
    },
    {
      "epoch": 4.439989675643122,
      "grad_norm": 1.40625,
      "learning_rate": 3.066028966806056e-05,
      "loss": 0.7735,
      "step": 51606
    },
    {
      "epoch": 4.440075711950443,
      "grad_norm": 0.8671875,
      "learning_rate": 3.065096694781166e-05,
      "loss": 0.4142,
      "step": 51607
    },
    {
      "epoch": 4.440161748257765,
      "grad_norm": 1.4609375,
      "learning_rate": 3.064164560031735e-05,
      "loss": 1.0635,
      "step": 51608
    },
    {
      "epoch": 4.440247784565086,
      "grad_norm": 1.6015625,
      "learning_rate": 3.0632325625605026e-05,
      "loss": 1.1567,
      "step": 51609
    },
    {
      "epoch": 4.440333820872408,
      "grad_norm": 1.3125,
      "learning_rate": 3.06230070237018e-05,
      "loss": 1.1244,
      "step": 51610
    },
    {
      "epoch": 4.44041985717973,
      "grad_norm": 1.140625,
      "learning_rate": 3.061368979463491e-05,
      "loss": 0.76,
      "step": 51611
    },
    {
      "epoch": 4.440505893487051,
      "grad_norm": 1.09375,
      "learning_rate": 3.060437393843185e-05,
      "loss": 1.8429,
      "step": 51612
    },
    {
      "epoch": 4.4405919297943734,
      "grad_norm": 1.21875,
      "learning_rate": 3.059505945511959e-05,
      "loss": 1.6746,
      "step": 51613
    },
    {
      "epoch": 4.440677966101695,
      "grad_norm": 1.578125,
      "learning_rate": 3.058574634472544e-05,
      "loss": 1.7692,
      "step": 51614
    },
    {
      "epoch": 4.440764002409017,
      "grad_norm": 0.87109375,
      "learning_rate": 3.057643460727682e-05,
      "loss": 0.9053,
      "step": 51615
    },
    {
      "epoch": 4.440850038716338,
      "grad_norm": 0.76171875,
      "learning_rate": 3.056712424280073e-05,
      "loss": 0.6394,
      "step": 51616
    },
    {
      "epoch": 4.44093607502366,
      "grad_norm": 1.0859375,
      "learning_rate": 3.0557815251324564e-05,
      "loss": 0.823,
      "step": 51617
    },
    {
      "epoch": 4.441022111330982,
      "grad_norm": 5.1875,
      "learning_rate": 3.054850763287548e-05,
      "loss": 0.6941,
      "step": 51618
    },
    {
      "epoch": 4.441108147638303,
      "grad_norm": 1.2109375,
      "learning_rate": 3.053920138748067e-05,
      "loss": 1.5726,
      "step": 51619
    },
    {
      "epoch": 4.441194183945625,
      "grad_norm": 0.86328125,
      "learning_rate": 3.052989651516747e-05,
      "loss": 0.8422,
      "step": 51620
    },
    {
      "epoch": 4.441280220252946,
      "grad_norm": 1.8671875,
      "learning_rate": 3.052059301596294e-05,
      "loss": 1.7073,
      "step": 51621
    },
    {
      "epoch": 4.4413662565602685,
      "grad_norm": 0.8515625,
      "learning_rate": 3.051129088989435e-05,
      "loss": 0.5327,
      "step": 51622
    },
    {
      "epoch": 4.44145229286759,
      "grad_norm": 1.234375,
      "learning_rate": 3.0501990136989012e-05,
      "loss": 1.2135,
      "step": 51623
    },
    {
      "epoch": 4.441538329174912,
      "grad_norm": 1.234375,
      "learning_rate": 3.0492690757273956e-05,
      "loss": 1.3798,
      "step": 51624
    },
    {
      "epoch": 4.441624365482234,
      "grad_norm": 1.1875,
      "learning_rate": 3.048339275077644e-05,
      "loss": 0.9397,
      "step": 51625
    },
    {
      "epoch": 4.441710401789555,
      "grad_norm": 18.0,
      "learning_rate": 3.0474096117523832e-05,
      "loss": 1.0953,
      "step": 51626
    },
    {
      "epoch": 4.441796438096877,
      "grad_norm": 2.84375,
      "learning_rate": 3.046480085754305e-05,
      "loss": 1.5404,
      "step": 51627
    },
    {
      "epoch": 4.441882474404198,
      "grad_norm": 1.3515625,
      "learning_rate": 3.0455506970861412e-05,
      "loss": 1.1138,
      "step": 51628
    },
    {
      "epoch": 4.44196851071152,
      "grad_norm": 0.86328125,
      "learning_rate": 3.0446214457506173e-05,
      "loss": 1.3436,
      "step": 51629
    },
    {
      "epoch": 4.4420545470188415,
      "grad_norm": 1.0703125,
      "learning_rate": 3.0436923317504252e-05,
      "loss": 1.6017,
      "step": 51630
    },
    {
      "epoch": 4.442140583326164,
      "grad_norm": 1.0625,
      "learning_rate": 3.042763355088318e-05,
      "loss": 1.3959,
      "step": 51631
    },
    {
      "epoch": 4.442226619633486,
      "grad_norm": 0.84765625,
      "learning_rate": 3.041834515766989e-05,
      "loss": 1.3025,
      "step": 51632
    },
    {
      "epoch": 4.442312655940807,
      "grad_norm": 0.93359375,
      "learning_rate": 3.0409058137891633e-05,
      "loss": 0.9365,
      "step": 51633
    },
    {
      "epoch": 4.442398692248129,
      "grad_norm": 1.234375,
      "learning_rate": 3.0399772491575605e-05,
      "loss": 1.1545,
      "step": 51634
    },
    {
      "epoch": 4.44248472855545,
      "grad_norm": 1.1015625,
      "learning_rate": 3.039048821874879e-05,
      "loss": 0.6,
      "step": 51635
    },
    {
      "epoch": 4.442570764862772,
      "grad_norm": 0.796875,
      "learning_rate": 3.0381205319438608e-05,
      "loss": 0.7571,
      "step": 51636
    },
    {
      "epoch": 4.442656801170093,
      "grad_norm": 6976.0,
      "learning_rate": 3.037192379367193e-05,
      "loss": 0.8832,
      "step": 51637
    },
    {
      "epoch": 4.442742837477415,
      "grad_norm": 1.1171875,
      "learning_rate": 3.0362643641476063e-05,
      "loss": 0.4983,
      "step": 51638
    },
    {
      "epoch": 4.4428288737847375,
      "grad_norm": 1.1875,
      "learning_rate": 3.0353364862878208e-05,
      "loss": 0.9697,
      "step": 51639
    },
    {
      "epoch": 4.442914910092059,
      "grad_norm": 30.25,
      "learning_rate": 3.0344087457905345e-05,
      "loss": 1.1391,
      "step": 51640
    },
    {
      "epoch": 4.443000946399381,
      "grad_norm": 1.2890625,
      "learning_rate": 3.0334811426584674e-05,
      "loss": 0.8801,
      "step": 51641
    },
    {
      "epoch": 4.443086982706702,
      "grad_norm": 0.9296875,
      "learning_rate": 3.0325536768943396e-05,
      "loss": 0.7476,
      "step": 51642
    },
    {
      "epoch": 4.443173019014024,
      "grad_norm": 1.234375,
      "learning_rate": 3.0316263485008543e-05,
      "loss": 1.0082,
      "step": 51643
    },
    {
      "epoch": 4.443259055321346,
      "grad_norm": 1.171875,
      "learning_rate": 3.0306991574807263e-05,
      "loss": 0.6546,
      "step": 51644
    },
    {
      "epoch": 4.443345091628667,
      "grad_norm": 0.94921875,
      "learning_rate": 3.02977210383667e-05,
      "loss": 0.8559,
      "step": 51645
    },
    {
      "epoch": 4.443431127935989,
      "grad_norm": 0.953125,
      "learning_rate": 3.028845187571394e-05,
      "loss": 0.5801,
      "step": 51646
    },
    {
      "epoch": 4.4435171642433104,
      "grad_norm": 1.03125,
      "learning_rate": 3.0279184086876076e-05,
      "loss": 0.6355,
      "step": 51647
    },
    {
      "epoch": 4.4436032005506325,
      "grad_norm": 0.9765625,
      "learning_rate": 3.026991767188031e-05,
      "loss": 1.3623,
      "step": 51648
    },
    {
      "epoch": 4.443689236857954,
      "grad_norm": 1.1328125,
      "learning_rate": 3.0260652630753616e-05,
      "loss": 1.2623,
      "step": 51649
    },
    {
      "epoch": 4.443775273165276,
      "grad_norm": 1.328125,
      "learning_rate": 3.025138896352325e-05,
      "loss": 0.5431,
      "step": 51650
    },
    {
      "epoch": 4.443861309472598,
      "grad_norm": 1.1015625,
      "learning_rate": 3.0242126670216084e-05,
      "loss": 0.657,
      "step": 51651
    },
    {
      "epoch": 4.443947345779919,
      "grad_norm": 1.0859375,
      "learning_rate": 3.0232865750859374e-05,
      "loss": 1.0183,
      "step": 51652
    },
    {
      "epoch": 4.444033382087241,
      "grad_norm": 8.0,
      "learning_rate": 3.0223606205480204e-05,
      "loss": 1.1076,
      "step": 51653
    },
    {
      "epoch": 4.444119418394562,
      "grad_norm": 7.0625,
      "learning_rate": 3.021434803410561e-05,
      "loss": 1.1824,
      "step": 51654
    },
    {
      "epoch": 4.444205454701884,
      "grad_norm": 0.953125,
      "learning_rate": 3.0205091236762628e-05,
      "loss": 1.1449,
      "step": 51655
    },
    {
      "epoch": 4.4442914910092055,
      "grad_norm": 207.0,
      "learning_rate": 3.0195835813478458e-05,
      "loss": 0.9305,
      "step": 51656
    },
    {
      "epoch": 4.444377527316528,
      "grad_norm": 1.2109375,
      "learning_rate": 3.018658176428002e-05,
      "loss": 1.2788,
      "step": 51657
    },
    {
      "epoch": 4.44446356362385,
      "grad_norm": 1.140625,
      "learning_rate": 3.0177329089194517e-05,
      "loss": 1.7612,
      "step": 51658
    },
    {
      "epoch": 4.444549599931171,
      "grad_norm": 1.0703125,
      "learning_rate": 3.0168077788248983e-05,
      "loss": 0.8806,
      "step": 51659
    },
    {
      "epoch": 4.444635636238493,
      "grad_norm": 22.25,
      "learning_rate": 3.015882786147034e-05,
      "loss": 1.5029,
      "step": 51660
    },
    {
      "epoch": 4.444721672545814,
      "grad_norm": 1.0078125,
      "learning_rate": 3.0149579308885844e-05,
      "loss": 0.9461,
      "step": 51661
    },
    {
      "epoch": 4.444807708853136,
      "grad_norm": 1.015625,
      "learning_rate": 3.0140332130522364e-05,
      "loss": 1.2238,
      "step": 51662
    },
    {
      "epoch": 4.444893745160457,
      "grad_norm": 1.234375,
      "learning_rate": 3.013108632640704e-05,
      "loss": 0.9414,
      "step": 51663
    },
    {
      "epoch": 4.444979781467779,
      "grad_norm": 0.8828125,
      "learning_rate": 3.0121841896567026e-05,
      "loss": 0.7108,
      "step": 51664
    },
    {
      "epoch": 4.4450658177751015,
      "grad_norm": 1.0546875,
      "learning_rate": 3.011259884102907e-05,
      "loss": 0.6203,
      "step": 51665
    },
    {
      "epoch": 4.445151854082423,
      "grad_norm": 1.1953125,
      "learning_rate": 3.0103357159820432e-05,
      "loss": 1.0826,
      "step": 51666
    },
    {
      "epoch": 4.445237890389745,
      "grad_norm": 1.65625,
      "learning_rate": 3.009411685296809e-05,
      "loss": 1.5819,
      "step": 51667
    },
    {
      "epoch": 4.445323926697066,
      "grad_norm": 1.0546875,
      "learning_rate": 3.0084877920499077e-05,
      "loss": 1.2084,
      "step": 51668
    },
    {
      "epoch": 4.445409963004388,
      "grad_norm": 0.90234375,
      "learning_rate": 3.0075640362440428e-05,
      "loss": 0.7938,
      "step": 51669
    },
    {
      "epoch": 4.445495999311709,
      "grad_norm": 1.3125,
      "learning_rate": 3.0066404178819062e-05,
      "loss": 1.7064,
      "step": 51670
    },
    {
      "epoch": 4.445582035619031,
      "grad_norm": 1.2421875,
      "learning_rate": 3.005716936966213e-05,
      "loss": 1.1799,
      "step": 51671
    },
    {
      "epoch": 4.445668071926353,
      "grad_norm": 1.109375,
      "learning_rate": 3.0047935934996607e-05,
      "loss": 1.6384,
      "step": 51672
    },
    {
      "epoch": 4.4457541082336745,
      "grad_norm": 0.99609375,
      "learning_rate": 3.003870387484936e-05,
      "loss": 0.9039,
      "step": 51673
    },
    {
      "epoch": 4.4458401445409965,
      "grad_norm": 0.796875,
      "learning_rate": 3.002947318924759e-05,
      "loss": 0.9846,
      "step": 51674
    },
    {
      "epoch": 4.445926180848318,
      "grad_norm": 1.0390625,
      "learning_rate": 3.0020243878218222e-05,
      "loss": 0.7685,
      "step": 51675
    },
    {
      "epoch": 4.44601221715564,
      "grad_norm": 0.84375,
      "learning_rate": 3.0011015941788177e-05,
      "loss": 1.1113,
      "step": 51676
    },
    {
      "epoch": 4.446098253462962,
      "grad_norm": 0.90625,
      "learning_rate": 3.0001789379984545e-05,
      "loss": 0.6536,
      "step": 51677
    },
    {
      "epoch": 4.446184289770283,
      "grad_norm": 1.1796875,
      "learning_rate": 2.999256419283425e-05,
      "loss": 1.0906,
      "step": 51678
    },
    {
      "epoch": 4.446270326077605,
      "grad_norm": 1.1953125,
      "learning_rate": 2.9983340380364266e-05,
      "loss": 1.1837,
      "step": 51679
    },
    {
      "epoch": 4.446356362384926,
      "grad_norm": 0.98046875,
      "learning_rate": 2.9974117942601687e-05,
      "loss": 1.3181,
      "step": 51680
    },
    {
      "epoch": 4.446442398692248,
      "grad_norm": 1.7421875,
      "learning_rate": 2.9964896879573267e-05,
      "loss": 1.44,
      "step": 51681
    },
    {
      "epoch": 4.4465284349995695,
      "grad_norm": 1.0625,
      "learning_rate": 2.995567719130615e-05,
      "loss": 0.993,
      "step": 51682
    },
    {
      "epoch": 4.446614471306892,
      "grad_norm": 1.15625,
      "learning_rate": 2.9946458877827322e-05,
      "loss": 1.7123,
      "step": 51683
    },
    {
      "epoch": 4.446700507614214,
      "grad_norm": 0.87109375,
      "learning_rate": 2.993724193916364e-05,
      "loss": 0.5102,
      "step": 51684
    },
    {
      "epoch": 4.446786543921535,
      "grad_norm": 436.0,
      "learning_rate": 2.9928026375342088e-05,
      "loss": 1.3714,
      "step": 51685
    },
    {
      "epoch": 4.446872580228857,
      "grad_norm": 1.046875,
      "learning_rate": 2.99188121863897e-05,
      "loss": 0.9463,
      "step": 51686
    },
    {
      "epoch": 4.446958616536178,
      "grad_norm": 1.109375,
      "learning_rate": 2.9909599372333284e-05,
      "loss": 1.2551,
      "step": 51687
    },
    {
      "epoch": 4.4470446528435,
      "grad_norm": 0.8671875,
      "learning_rate": 2.9900387933199878e-05,
      "loss": 0.6297,
      "step": 51688
    },
    {
      "epoch": 4.447130689150821,
      "grad_norm": 0.88671875,
      "learning_rate": 2.989117786901646e-05,
      "loss": 0.9383,
      "step": 51689
    },
    {
      "epoch": 4.447216725458143,
      "grad_norm": 1.21875,
      "learning_rate": 2.9881969179809843e-05,
      "loss": 1.0815,
      "step": 51690
    },
    {
      "epoch": 4.4473027617654655,
      "grad_norm": 1.28125,
      "learning_rate": 2.9872761865607112e-05,
      "loss": 2.0908,
      "step": 51691
    },
    {
      "epoch": 4.447388798072787,
      "grad_norm": 127.0,
      "learning_rate": 2.9863555926435083e-05,
      "loss": 1.1052,
      "step": 51692
    },
    {
      "epoch": 4.447474834380109,
      "grad_norm": 1.578125,
      "learning_rate": 2.9854351362320677e-05,
      "loss": 1.6906,
      "step": 51693
    },
    {
      "epoch": 4.44756087068743,
      "grad_norm": 1.8515625,
      "learning_rate": 2.9845148173290925e-05,
      "loss": 1.2171,
      "step": 51694
    },
    {
      "epoch": 4.447646906994752,
      "grad_norm": 0.90234375,
      "learning_rate": 2.983594635937259e-05,
      "loss": 0.7941,
      "step": 51695
    },
    {
      "epoch": 4.447732943302073,
      "grad_norm": 1.296875,
      "learning_rate": 2.98267459205927e-05,
      "loss": 0.9044,
      "step": 51696
    },
    {
      "epoch": 4.447818979609395,
      "grad_norm": 2.421875,
      "learning_rate": 2.9817546856978184e-05,
      "loss": 1.2849,
      "step": 51697
    },
    {
      "epoch": 4.447905015916717,
      "grad_norm": 1.0625,
      "learning_rate": 2.9808349168555848e-05,
      "loss": 1.4576,
      "step": 51698
    },
    {
      "epoch": 4.4479910522240385,
      "grad_norm": 100.0,
      "learning_rate": 2.979915285535262e-05,
      "loss": 0.7759,
      "step": 51699
    },
    {
      "epoch": 4.4480770885313605,
      "grad_norm": 0.9140625,
      "learning_rate": 2.9789957917395527e-05,
      "loss": 0.6999,
      "step": 51700
    },
    {
      "epoch": 4.448163124838682,
      "grad_norm": 1.0,
      "learning_rate": 2.9780764354711275e-05,
      "loss": 1.3701,
      "step": 51701
    },
    {
      "epoch": 4.448249161146004,
      "grad_norm": 1.0078125,
      "learning_rate": 2.9771572167326898e-05,
      "loss": 1.4065,
      "step": 51702
    },
    {
      "epoch": 4.448335197453325,
      "grad_norm": 0.9296875,
      "learning_rate": 2.976238135526915e-05,
      "loss": 0.4439,
      "step": 51703
    },
    {
      "epoch": 4.448421233760647,
      "grad_norm": 1.09375,
      "learning_rate": 2.975319191856485e-05,
      "loss": 1.7789,
      "step": 51704
    },
    {
      "epoch": 4.448507270067969,
      "grad_norm": 1.7734375,
      "learning_rate": 2.9744003857241243e-05,
      "loss": 1.1042,
      "step": 51705
    },
    {
      "epoch": 4.44859330637529,
      "grad_norm": 1.28125,
      "learning_rate": 2.9734817171324813e-05,
      "loss": 1.6552,
      "step": 51706
    },
    {
      "epoch": 4.448679342682612,
      "grad_norm": 1.28125,
      "learning_rate": 2.9725631860842705e-05,
      "loss": 1.1825,
      "step": 51707
    },
    {
      "epoch": 4.4487653789899335,
      "grad_norm": 0.84765625,
      "learning_rate": 2.9716447925821564e-05,
      "loss": 1.1588,
      "step": 51708
    },
    {
      "epoch": 4.448851415297256,
      "grad_norm": 1.1484375,
      "learning_rate": 2.9707265366288373e-05,
      "loss": 1.2422,
      "step": 51709
    },
    {
      "epoch": 4.448937451604577,
      "grad_norm": 3936.0,
      "learning_rate": 2.9698084182270048e-05,
      "loss": 0.8527,
      "step": 51710
    },
    {
      "epoch": 4.449023487911899,
      "grad_norm": 1.0234375,
      "learning_rate": 2.9688904373793235e-05,
      "loss": 0.6967,
      "step": 51711
    },
    {
      "epoch": 4.449109524219221,
      "grad_norm": 46.5,
      "learning_rate": 2.9679725940884918e-05,
      "loss": 1.8157,
      "step": 51712
    },
    {
      "epoch": 4.449195560526542,
      "grad_norm": 1.234375,
      "learning_rate": 2.9670548883572012e-05,
      "loss": 1.397,
      "step": 51713
    },
    {
      "epoch": 4.449281596833864,
      "grad_norm": 1.015625,
      "learning_rate": 2.9661373201881226e-05,
      "loss": 1.0686,
      "step": 51714
    },
    {
      "epoch": 4.449367633141185,
      "grad_norm": 1.5390625,
      "learning_rate": 2.965219889583942e-05,
      "loss": 1.3606,
      "step": 51715
    },
    {
      "epoch": 4.449453669448507,
      "grad_norm": 1.2265625,
      "learning_rate": 2.964302596547358e-05,
      "loss": 1.3227,
      "step": 51716
    },
    {
      "epoch": 4.449539705755829,
      "grad_norm": 1.1484375,
      "learning_rate": 2.963385441081029e-05,
      "loss": 0.4543,
      "step": 51717
    },
    {
      "epoch": 4.449625742063151,
      "grad_norm": 2.875,
      "learning_rate": 2.962468423187653e-05,
      "loss": 0.6981,
      "step": 51718
    },
    {
      "epoch": 4.449711778370473,
      "grad_norm": 1.6171875,
      "learning_rate": 2.9615515428699057e-05,
      "loss": 0.6591,
      "step": 51719
    },
    {
      "epoch": 4.449797814677794,
      "grad_norm": 1.0,
      "learning_rate": 2.960634800130474e-05,
      "loss": 0.7177,
      "step": 51720
    },
    {
      "epoch": 4.449883850985116,
      "grad_norm": 1.3671875,
      "learning_rate": 2.9597181949720388e-05,
      "loss": 1.5171,
      "step": 51721
    },
    {
      "epoch": 4.449969887292437,
      "grad_norm": 0.9375,
      "learning_rate": 2.9588017273972754e-05,
      "loss": 1.5862,
      "step": 51722
    },
    {
      "epoch": 4.450055923599759,
      "grad_norm": 1.296875,
      "learning_rate": 2.957885397408866e-05,
      "loss": 1.6376,
      "step": 51723
    },
    {
      "epoch": 4.45014195990708,
      "grad_norm": 1.0546875,
      "learning_rate": 2.956969205009502e-05,
      "loss": 0.5299,
      "step": 51724
    },
    {
      "epoch": 4.4502279962144025,
      "grad_norm": 0.98046875,
      "learning_rate": 2.9560531502018485e-05,
      "loss": 0.7095,
      "step": 51725
    },
    {
      "epoch": 4.4503140325217245,
      "grad_norm": 1.0078125,
      "learning_rate": 2.955137232988586e-05,
      "loss": 0.9035,
      "step": 51726
    },
    {
      "epoch": 4.450400068829046,
      "grad_norm": 0.953125,
      "learning_rate": 2.954221453372402e-05,
      "loss": 0.4768,
      "step": 51727
    },
    {
      "epoch": 4.450486105136368,
      "grad_norm": 1.1171875,
      "learning_rate": 2.9533058113559663e-05,
      "loss": 2.023,
      "step": 51728
    },
    {
      "epoch": 4.450572141443689,
      "grad_norm": 1.734375,
      "learning_rate": 2.9523903069419654e-05,
      "loss": 1.2691,
      "step": 51729
    },
    {
      "epoch": 4.450658177751011,
      "grad_norm": 1.7578125,
      "learning_rate": 2.9514749401330754e-05,
      "loss": 1.0492,
      "step": 51730
    },
    {
      "epoch": 4.450744214058332,
      "grad_norm": 0.9453125,
      "learning_rate": 2.9505597109319603e-05,
      "loss": 0.9833,
      "step": 51731
    },
    {
      "epoch": 4.450830250365654,
      "grad_norm": 1.21875,
      "learning_rate": 2.9496446193413184e-05,
      "loss": 0.9777,
      "step": 51732
    },
    {
      "epoch": 4.450916286672976,
      "grad_norm": 1.5234375,
      "learning_rate": 2.9487296653638084e-05,
      "loss": 0.875,
      "step": 51733
    },
    {
      "epoch": 4.4510023229802975,
      "grad_norm": 1.703125,
      "learning_rate": 2.9478148490021174e-05,
      "loss": 1.9556,
      "step": 51734
    },
    {
      "epoch": 4.45108835928762,
      "grad_norm": 1.21875,
      "learning_rate": 2.946900170258915e-05,
      "loss": 1.2366,
      "step": 51735
    },
    {
      "epoch": 4.451174395594941,
      "grad_norm": 1.140625,
      "learning_rate": 2.9459856291368768e-05,
      "loss": 0.6798,
      "step": 51736
    },
    {
      "epoch": 4.451260431902263,
      "grad_norm": 1.375,
      "learning_rate": 2.9450712256386792e-05,
      "loss": 1.2102,
      "step": 51737
    },
    {
      "epoch": 4.451346468209584,
      "grad_norm": 1.140625,
      "learning_rate": 2.9441569597670027e-05,
      "loss": 1.1408,
      "step": 51738
    },
    {
      "epoch": 4.451432504516906,
      "grad_norm": 1.15625,
      "learning_rate": 2.9432428315245063e-05,
      "loss": 0.8162,
      "step": 51739
    },
    {
      "epoch": 4.451518540824228,
      "grad_norm": 0.890625,
      "learning_rate": 2.9423288409138717e-05,
      "loss": 0.9563,
      "step": 51740
    },
    {
      "epoch": 4.451604577131549,
      "grad_norm": 1.2421875,
      "learning_rate": 2.9414149879377792e-05,
      "loss": 1.0286,
      "step": 51741
    },
    {
      "epoch": 4.451690613438871,
      "grad_norm": 832.0,
      "learning_rate": 2.940501272598889e-05,
      "loss": 0.7463,
      "step": 51742
    },
    {
      "epoch": 4.451776649746193,
      "grad_norm": 1.0625,
      "learning_rate": 2.9395876948998924e-05,
      "loss": 1.015,
      "step": 51743
    },
    {
      "epoch": 4.451862686053515,
      "grad_norm": 0.91015625,
      "learning_rate": 2.9386742548434375e-05,
      "loss": 0.5284,
      "step": 51744
    },
    {
      "epoch": 4.451948722360836,
      "grad_norm": 60.5,
      "learning_rate": 2.9377609524322057e-05,
      "loss": 1.8071,
      "step": 51745
    },
    {
      "epoch": 4.452034758668158,
      "grad_norm": 1.078125,
      "learning_rate": 2.9368477876688837e-05,
      "loss": 0.9725,
      "step": 51746
    },
    {
      "epoch": 4.45212079497548,
      "grad_norm": 0.93359375,
      "learning_rate": 2.935934760556119e-05,
      "loss": 0.9336,
      "step": 51747
    },
    {
      "epoch": 4.452206831282801,
      "grad_norm": 71.0,
      "learning_rate": 2.9350218710965993e-05,
      "loss": 0.7764,
      "step": 51748
    },
    {
      "epoch": 4.452292867590123,
      "grad_norm": 1.2421875,
      "learning_rate": 2.934109119292977e-05,
      "loss": 1.2361,
      "step": 51749
    },
    {
      "epoch": 4.452378903897444,
      "grad_norm": 1.640625,
      "learning_rate": 2.9331965051479336e-05,
      "loss": 1.3882,
      "step": 51750
    },
    {
      "epoch": 4.4524649402047665,
      "grad_norm": 1.1640625,
      "learning_rate": 2.932284028664145e-05,
      "loss": 1.2202,
      "step": 51751
    },
    {
      "epoch": 4.452550976512088,
      "grad_norm": 2.515625,
      "learning_rate": 2.931371689844259e-05,
      "loss": 1.8945,
      "step": 51752
    },
    {
      "epoch": 4.45263701281941,
      "grad_norm": 1.203125,
      "learning_rate": 2.930459488690962e-05,
      "loss": 1.1083,
      "step": 51753
    },
    {
      "epoch": 4.452723049126732,
      "grad_norm": 1.1953125,
      "learning_rate": 2.9295474252069243e-05,
      "loss": 1.1002,
      "step": 51754
    },
    {
      "epoch": 4.452809085434053,
      "grad_norm": 1.15625,
      "learning_rate": 2.9286354993947938e-05,
      "loss": 0.7492,
      "step": 51755
    },
    {
      "epoch": 4.452895121741375,
      "grad_norm": 1.7421875,
      "learning_rate": 2.9277237112572518e-05,
      "loss": 0.7259,
      "step": 51756
    },
    {
      "epoch": 4.452981158048696,
      "grad_norm": 0.8984375,
      "learning_rate": 2.9268120607969685e-05,
      "loss": 1.0212,
      "step": 51757
    },
    {
      "epoch": 4.453067194356018,
      "grad_norm": 1.046875,
      "learning_rate": 2.9259005480166023e-05,
      "loss": 1.4947,
      "step": 51758
    },
    {
      "epoch": 4.45315323066334,
      "grad_norm": 1.1875,
      "learning_rate": 2.9249891729188182e-05,
      "loss": 1.0617,
      "step": 51759
    },
    {
      "epoch": 4.4532392669706615,
      "grad_norm": 1.453125,
      "learning_rate": 2.924077935506292e-05,
      "loss": 1.007,
      "step": 51760
    },
    {
      "epoch": 4.453325303277984,
      "grad_norm": 1.015625,
      "learning_rate": 2.9231668357816765e-05,
      "loss": 1.0032,
      "step": 51761
    },
    {
      "epoch": 4.453411339585305,
      "grad_norm": 1.171875,
      "learning_rate": 2.9222558737476478e-05,
      "loss": 1.052,
      "step": 51762
    },
    {
      "epoch": 4.453497375892627,
      "grad_norm": 0.9375,
      "learning_rate": 2.9213450494068593e-05,
      "loss": 0.6896,
      "step": 51763
    },
    {
      "epoch": 4.453583412199948,
      "grad_norm": 0.96484375,
      "learning_rate": 2.9204343627619757e-05,
      "loss": 1.577,
      "step": 51764
    },
    {
      "epoch": 4.45366944850727,
      "grad_norm": 1.6953125,
      "learning_rate": 2.9195238138156778e-05,
      "loss": 1.0491,
      "step": 51765
    },
    {
      "epoch": 4.453755484814592,
      "grad_norm": 1.03125,
      "learning_rate": 2.918613402570608e-05,
      "loss": 1.2443,
      "step": 51766
    },
    {
      "epoch": 4.453841521121913,
      "grad_norm": 0.859375,
      "learning_rate": 2.917703129029431e-05,
      "loss": 1.7734,
      "step": 51767
    },
    {
      "epoch": 4.453927557429235,
      "grad_norm": 0.921875,
      "learning_rate": 2.916792993194828e-05,
      "loss": 1.0997,
      "step": 51768
    },
    {
      "epoch": 4.454013593736557,
      "grad_norm": 1.03125,
      "learning_rate": 2.915882995069441e-05,
      "loss": 1.5385,
      "step": 51769
    },
    {
      "epoch": 4.454099630043879,
      "grad_norm": 1.2734375,
      "learning_rate": 2.9149731346559405e-05,
      "loss": 0.8411,
      "step": 51770
    },
    {
      "epoch": 4.4541856663512,
      "grad_norm": 1.0,
      "learning_rate": 2.9140634119569908e-05,
      "loss": 0.8778,
      "step": 51771
    },
    {
      "epoch": 4.454271702658522,
      "grad_norm": 1.4296875,
      "learning_rate": 2.91315382697524e-05,
      "loss": 0.9907,
      "step": 51772
    },
    {
      "epoch": 4.454357738965844,
      "grad_norm": 19.5,
      "learning_rate": 2.9122443797133635e-05,
      "loss": 1.0398,
      "step": 51773
    },
    {
      "epoch": 4.454443775273165,
      "grad_norm": 1.0859375,
      "learning_rate": 2.9113350701740093e-05,
      "loss": 0.9661,
      "step": 51774
    },
    {
      "epoch": 4.454529811580487,
      "grad_norm": 1.1796875,
      "learning_rate": 2.910425898359842e-05,
      "loss": 0.7246,
      "step": 51775
    },
    {
      "epoch": 4.454615847887808,
      "grad_norm": 0.96875,
      "learning_rate": 2.909516864273526e-05,
      "loss": 0.6961,
      "step": 51776
    },
    {
      "epoch": 4.4547018841951305,
      "grad_norm": 1.0546875,
      "learning_rate": 2.9086079679177036e-05,
      "loss": 0.7532,
      "step": 51777
    },
    {
      "epoch": 4.454787920502452,
      "grad_norm": 96.0,
      "learning_rate": 2.9076992092950507e-05,
      "loss": 1.3121,
      "step": 51778
    },
    {
      "epoch": 4.454873956809774,
      "grad_norm": 1.0703125,
      "learning_rate": 2.9067905884082145e-05,
      "loss": 1.5919,
      "step": 51779
    },
    {
      "epoch": 4.454959993117096,
      "grad_norm": 1.1796875,
      "learning_rate": 2.9058821052598606e-05,
      "loss": 1.2394,
      "step": 51780
    },
    {
      "epoch": 4.455046029424417,
      "grad_norm": 1.1875,
      "learning_rate": 2.9049737598526473e-05,
      "loss": 1.0357,
      "step": 51781
    },
    {
      "epoch": 4.455132065731739,
      "grad_norm": 928.0,
      "learning_rate": 2.904065552189217e-05,
      "loss": 1.1348,
      "step": 51782
    },
    {
      "epoch": 4.45521810203906,
      "grad_norm": 1.8203125,
      "learning_rate": 2.903157482272234e-05,
      "loss": 1.7489,
      "step": 51783
    },
    {
      "epoch": 4.455304138346382,
      "grad_norm": 1296.0,
      "learning_rate": 2.902249550104369e-05,
      "loss": 1.4684,
      "step": 51784
    },
    {
      "epoch": 4.4553901746537035,
      "grad_norm": 1.09375,
      "learning_rate": 2.9013417556882525e-05,
      "loss": 0.7287,
      "step": 51785
    },
    {
      "epoch": 4.4554762109610255,
      "grad_norm": 1.125,
      "learning_rate": 2.9004340990265553e-05,
      "loss": 0.7759,
      "step": 51786
    },
    {
      "epoch": 4.455562247268348,
      "grad_norm": 8.0625,
      "learning_rate": 2.8995265801219304e-05,
      "loss": 0.4075,
      "step": 51787
    },
    {
      "epoch": 4.455648283575669,
      "grad_norm": 1.40625,
      "learning_rate": 2.898619198977026e-05,
      "loss": 0.9862,
      "step": 51788
    },
    {
      "epoch": 4.455734319882991,
      "grad_norm": 5.8125,
      "learning_rate": 2.897711955594501e-05,
      "loss": 0.9905,
      "step": 51789
    },
    {
      "epoch": 4.455820356190312,
      "grad_norm": 1.0546875,
      "learning_rate": 2.8968048499770082e-05,
      "loss": 1.1691,
      "step": 51790
    },
    {
      "epoch": 4.455906392497634,
      "grad_norm": 1.265625,
      "learning_rate": 2.8958978821271963e-05,
      "loss": 1.3539,
      "step": 51791
    },
    {
      "epoch": 4.455992428804956,
      "grad_norm": 1.0234375,
      "learning_rate": 2.8949910520477297e-05,
      "loss": 1.1323,
      "step": 51792
    },
    {
      "epoch": 4.456078465112277,
      "grad_norm": 0.88671875,
      "learning_rate": 2.8940843597412446e-05,
      "loss": 0.8398,
      "step": 51793
    },
    {
      "epoch": 4.456164501419599,
      "grad_norm": 1.4296875,
      "learning_rate": 2.8931778052104008e-05,
      "loss": 1.1294,
      "step": 51794
    },
    {
      "epoch": 4.456250537726921,
      "grad_norm": 1.671875,
      "learning_rate": 2.8922713884578623e-05,
      "loss": 1.7028,
      "step": 51795
    },
    {
      "epoch": 4.456336574034243,
      "grad_norm": 1.5,
      "learning_rate": 2.8913651094862547e-05,
      "loss": 0.6959,
      "step": 51796
    },
    {
      "epoch": 4.456422610341564,
      "grad_norm": 1.0625,
      "learning_rate": 2.890458968298243e-05,
      "loss": 1.5371,
      "step": 51797
    },
    {
      "epoch": 4.456508646648886,
      "grad_norm": 3.859375,
      "learning_rate": 2.8895529648964857e-05,
      "loss": 1.0163,
      "step": 51798
    },
    {
      "epoch": 4.456594682956208,
      "grad_norm": 1.0625,
      "learning_rate": 2.8886470992836144e-05,
      "loss": 0.7529,
      "step": 51799
    },
    {
      "epoch": 4.456680719263529,
      "grad_norm": 3.921875,
      "learning_rate": 2.8877413714622935e-05,
      "loss": 1.2761,
      "step": 51800
    },
    {
      "epoch": 4.456766755570851,
      "grad_norm": 0.921875,
      "learning_rate": 2.8868357814351653e-05,
      "loss": 0.8388,
      "step": 51801
    },
    {
      "epoch": 4.456852791878172,
      "grad_norm": 21.375,
      "learning_rate": 2.8859303292048776e-05,
      "loss": 1.0776,
      "step": 51802
    },
    {
      "epoch": 4.4569388281854945,
      "grad_norm": 1.0625,
      "learning_rate": 2.8850250147740842e-05,
      "loss": 0.5025,
      "step": 51803
    },
    {
      "epoch": 4.457024864492816,
      "grad_norm": 65.0,
      "learning_rate": 2.884119838145427e-05,
      "loss": 1.4331,
      "step": 51804
    },
    {
      "epoch": 4.457110900800138,
      "grad_norm": 0.9296875,
      "learning_rate": 2.883214799321554e-05,
      "loss": 1.1708,
      "step": 51805
    },
    {
      "epoch": 4.45719693710746,
      "grad_norm": 0.91796875,
      "learning_rate": 2.882309898305119e-05,
      "loss": 0.761,
      "step": 51806
    },
    {
      "epoch": 4.457282973414781,
      "grad_norm": 0.96875,
      "learning_rate": 2.881405135098758e-05,
      "loss": 0.7888,
      "step": 51807
    },
    {
      "epoch": 4.457369009722103,
      "grad_norm": 0.953125,
      "learning_rate": 2.8805005097051195e-05,
      "loss": 0.9935,
      "step": 51808
    },
    {
      "epoch": 4.457455046029424,
      "grad_norm": 418.0,
      "learning_rate": 2.8795960221268625e-05,
      "loss": 0.6813,
      "step": 51809
    },
    {
      "epoch": 4.457541082336746,
      "grad_norm": 1.03125,
      "learning_rate": 2.8786916723666177e-05,
      "loss": 0.8518,
      "step": 51810
    },
    {
      "epoch": 4.4576271186440675,
      "grad_norm": 1.0703125,
      "learning_rate": 2.8777874604270336e-05,
      "loss": 1.3694,
      "step": 51811
    },
    {
      "epoch": 4.4577131549513895,
      "grad_norm": 0.98828125,
      "learning_rate": 2.8768833863107635e-05,
      "loss": 0.614,
      "step": 51812
    },
    {
      "epoch": 4.457799191258712,
      "grad_norm": 1.171875,
      "learning_rate": 2.875979450020433e-05,
      "loss": 1.3806,
      "step": 51813
    },
    {
      "epoch": 4.457885227566033,
      "grad_norm": 1.140625,
      "learning_rate": 2.8750756515587117e-05,
      "loss": 0.9706,
      "step": 51814
    },
    {
      "epoch": 4.457971263873355,
      "grad_norm": 1.1953125,
      "learning_rate": 2.8741719909282095e-05,
      "loss": 1.2423,
      "step": 51815
    },
    {
      "epoch": 4.458057300180676,
      "grad_norm": 1.546875,
      "learning_rate": 2.8732684681316013e-05,
      "loss": 0.7407,
      "step": 51816
    },
    {
      "epoch": 4.458143336487998,
      "grad_norm": 1.3046875,
      "learning_rate": 2.8723650831715187e-05,
      "loss": 1.2425,
      "step": 51817
    },
    {
      "epoch": 4.458229372795319,
      "grad_norm": 1.015625,
      "learning_rate": 2.8714618360505984e-05,
      "loss": 1.1401,
      "step": 51818
    },
    {
      "epoch": 4.458315409102641,
      "grad_norm": 1.15625,
      "learning_rate": 2.870558726771494e-05,
      "loss": 0.85,
      "step": 51819
    },
    {
      "epoch": 4.458401445409963,
      "grad_norm": 0.71875,
      "learning_rate": 2.8696557553368306e-05,
      "loss": 0.4796,
      "step": 51820
    },
    {
      "epoch": 4.458487481717285,
      "grad_norm": 1.4296875,
      "learning_rate": 2.868752921749257e-05,
      "loss": 1.129,
      "step": 51821
    },
    {
      "epoch": 4.458573518024607,
      "grad_norm": 1.203125,
      "learning_rate": 2.8678502260114205e-05,
      "loss": 1.5782,
      "step": 51822
    },
    {
      "epoch": 4.458659554331928,
      "grad_norm": 1.390625,
      "learning_rate": 2.866947668125952e-05,
      "loss": 0.5743,
      "step": 51823
    },
    {
      "epoch": 4.45874559063925,
      "grad_norm": 1.421875,
      "learning_rate": 2.8660452480954946e-05,
      "loss": 1.892,
      "step": 51824
    },
    {
      "epoch": 4.458831626946571,
      "grad_norm": 10.9375,
      "learning_rate": 2.8651429659226902e-05,
      "loss": 0.6606,
      "step": 51825
    },
    {
      "epoch": 4.458917663253893,
      "grad_norm": 1.328125,
      "learning_rate": 2.86424082161017e-05,
      "loss": 1.3594,
      "step": 51826
    },
    {
      "epoch": 4.459003699561215,
      "grad_norm": 0.8671875,
      "learning_rate": 2.8633388151605823e-05,
      "loss": 0.874,
      "step": 51827
    },
    {
      "epoch": 4.459089735868536,
      "grad_norm": 0.86328125,
      "learning_rate": 2.8624369465765632e-05,
      "loss": 1.351,
      "step": 51828
    },
    {
      "epoch": 4.4591757721758585,
      "grad_norm": 1.0234375,
      "learning_rate": 2.8615352158607444e-05,
      "loss": 1.0121,
      "step": 51829
    },
    {
      "epoch": 4.45926180848318,
      "grad_norm": 3.8125,
      "learning_rate": 2.8606336230157737e-05,
      "loss": 1.6648,
      "step": 51830
    },
    {
      "epoch": 4.459347844790502,
      "grad_norm": 0.859375,
      "learning_rate": 2.859732168044271e-05,
      "loss": 0.4451,
      "step": 51831
    },
    {
      "epoch": 4.459433881097823,
      "grad_norm": 1.1875,
      "learning_rate": 2.8588308509488902e-05,
      "loss": 1.5016,
      "step": 51832
    },
    {
      "epoch": 4.459519917405145,
      "grad_norm": 0.93359375,
      "learning_rate": 2.8579296717322623e-05,
      "loss": 0.5624,
      "step": 51833
    },
    {
      "epoch": 4.459605953712467,
      "grad_norm": 1.484375,
      "learning_rate": 2.8570286303970128e-05,
      "loss": 0.964,
      "step": 51834
    },
    {
      "epoch": 4.459691990019788,
      "grad_norm": 0.81640625,
      "learning_rate": 2.85612772694579e-05,
      "loss": 1.0455,
      "step": 51835
    },
    {
      "epoch": 4.45977802632711,
      "grad_norm": 1.1171875,
      "learning_rate": 2.8552269613812298e-05,
      "loss": 1.7541,
      "step": 51836
    },
    {
      "epoch": 4.4598640626344315,
      "grad_norm": 1.3203125,
      "learning_rate": 2.8543263337059587e-05,
      "loss": 1.4445,
      "step": 51837
    },
    {
      "epoch": 4.4599500989417535,
      "grad_norm": 0.98046875,
      "learning_rate": 2.853425843922608e-05,
      "loss": 1.0398,
      "step": 51838
    },
    {
      "epoch": 4.460036135249075,
      "grad_norm": 1.1015625,
      "learning_rate": 2.8525254920338306e-05,
      "loss": 0.9597,
      "step": 51839
    },
    {
      "epoch": 4.460122171556397,
      "grad_norm": 0.953125,
      "learning_rate": 2.8516252780422302e-05,
      "loss": 0.5841,
      "step": 51840
    },
    {
      "epoch": 4.460208207863719,
      "grad_norm": 1.1171875,
      "learning_rate": 2.850725201950466e-05,
      "loss": 0.9416,
      "step": 51841
    },
    {
      "epoch": 4.46029424417104,
      "grad_norm": 6.125,
      "learning_rate": 2.8498252637611634e-05,
      "loss": 0.7322,
      "step": 51842
    },
    {
      "epoch": 4.460380280478362,
      "grad_norm": 1.125,
      "learning_rate": 2.8489254634769424e-05,
      "loss": 0.9336,
      "step": 51843
    },
    {
      "epoch": 4.460466316785683,
      "grad_norm": 1.125,
      "learning_rate": 2.848025801100451e-05,
      "loss": 0.8455,
      "step": 51844
    },
    {
      "epoch": 4.460552353093005,
      "grad_norm": 4.25,
      "learning_rate": 2.8471262766343097e-05,
      "loss": 1.0248,
      "step": 51845
    },
    {
      "epoch": 4.4606383894003265,
      "grad_norm": 1.03125,
      "learning_rate": 2.8462268900811495e-05,
      "loss": 1.6841,
      "step": 51846
    },
    {
      "epoch": 4.460724425707649,
      "grad_norm": 1.203125,
      "learning_rate": 2.845327641443618e-05,
      "loss": 0.8834,
      "step": 51847
    },
    {
      "epoch": 4.460810462014971,
      "grad_norm": 1.4453125,
      "learning_rate": 2.844428530724319e-05,
      "loss": 1.123,
      "step": 51848
    },
    {
      "epoch": 4.460896498322292,
      "grad_norm": 1.140625,
      "learning_rate": 2.8435295579258946e-05,
      "loss": 1.1509,
      "step": 51849
    },
    {
      "epoch": 4.460982534629614,
      "grad_norm": 1.140625,
      "learning_rate": 2.8426307230509874e-05,
      "loss": 1.678,
      "step": 51850
    },
    {
      "epoch": 4.461068570936935,
      "grad_norm": 1.0234375,
      "learning_rate": 2.841732026102206e-05,
      "loss": 1.084,
      "step": 51851
    },
    {
      "epoch": 4.461154607244257,
      "grad_norm": 1.2578125,
      "learning_rate": 2.8408334670821824e-05,
      "loss": 1.2124,
      "step": 51852
    },
    {
      "epoch": 4.461240643551578,
      "grad_norm": 1.109375,
      "learning_rate": 2.8399350459935524e-05,
      "loss": 0.993,
      "step": 51853
    },
    {
      "epoch": 4.4613266798589,
      "grad_norm": 1.2421875,
      "learning_rate": 2.8390367628389367e-05,
      "loss": 1.6699,
      "step": 51854
    },
    {
      "epoch": 4.4614127161662225,
      "grad_norm": 1.625,
      "learning_rate": 2.838138617620978e-05,
      "loss": 1.2297,
      "step": 51855
    },
    {
      "epoch": 4.461498752473544,
      "grad_norm": 14.0,
      "learning_rate": 2.8372406103422787e-05,
      "loss": 1.2019,
      "step": 51856
    },
    {
      "epoch": 4.461584788780866,
      "grad_norm": 0.8984375,
      "learning_rate": 2.836342741005482e-05,
      "loss": 0.899,
      "step": 51857
    },
    {
      "epoch": 4.461670825088187,
      "grad_norm": 0.78515625,
      "learning_rate": 2.835445009613219e-05,
      "loss": 1.0705,
      "step": 51858
    },
    {
      "epoch": 4.461756861395509,
      "grad_norm": 0.96484375,
      "learning_rate": 2.8345474161680985e-05,
      "loss": 0.4026,
      "step": 51859
    },
    {
      "epoch": 4.46184289770283,
      "grad_norm": 1.4296875,
      "learning_rate": 2.8336499606727572e-05,
      "loss": 1.471,
      "step": 51860
    },
    {
      "epoch": 4.461928934010152,
      "grad_norm": 1.1875,
      "learning_rate": 2.83275264312981e-05,
      "loss": 1.0107,
      "step": 51861
    },
    {
      "epoch": 4.462014970317474,
      "grad_norm": 1.234375,
      "learning_rate": 2.831855463541888e-05,
      "loss": 0.6723,
      "step": 51862
    },
    {
      "epoch": 4.4621010066247955,
      "grad_norm": 1.1953125,
      "learning_rate": 2.8309584219116226e-05,
      "loss": 1.8965,
      "step": 51863
    },
    {
      "epoch": 4.4621870429321175,
      "grad_norm": 40.0,
      "learning_rate": 2.8300615182416222e-05,
      "loss": 0.8276,
      "step": 51864
    },
    {
      "epoch": 4.462273079239439,
      "grad_norm": 1.03125,
      "learning_rate": 2.8291647525345186e-05,
      "loss": 1.1795,
      "step": 51865
    },
    {
      "epoch": 4.462359115546761,
      "grad_norm": 0.91015625,
      "learning_rate": 2.828268124792943e-05,
      "loss": 0.9355,
      "step": 51866
    },
    {
      "epoch": 4.462445151854082,
      "grad_norm": 0.7890625,
      "learning_rate": 2.8273716350194988e-05,
      "loss": 0.4772,
      "step": 51867
    },
    {
      "epoch": 4.462531188161404,
      "grad_norm": 1.140625,
      "learning_rate": 2.8264752832168172e-05,
      "loss": 1.651,
      "step": 51868
    },
    {
      "epoch": 4.462617224468726,
      "grad_norm": 0.7890625,
      "learning_rate": 2.825579069387535e-05,
      "loss": 0.967,
      "step": 51869
    },
    {
      "epoch": 4.462703260776047,
      "grad_norm": 0.76953125,
      "learning_rate": 2.8246829935342445e-05,
      "loss": 0.6835,
      "step": 51870
    },
    {
      "epoch": 4.462789297083369,
      "grad_norm": 1.1875,
      "learning_rate": 2.823787055659588e-05,
      "loss": 1.2402,
      "step": 51871
    },
    {
      "epoch": 4.4628753333906905,
      "grad_norm": 0.98046875,
      "learning_rate": 2.8228912557661745e-05,
      "loss": 1.0668,
      "step": 51872
    },
    {
      "epoch": 4.462961369698013,
      "grad_norm": 1.1328125,
      "learning_rate": 2.82199559385663e-05,
      "loss": 1.283,
      "step": 51873
    },
    {
      "epoch": 4.463047406005334,
      "grad_norm": 0.96484375,
      "learning_rate": 2.8211000699335797e-05,
      "loss": 1.2126,
      "step": 51874
    },
    {
      "epoch": 4.463133442312656,
      "grad_norm": 41.25,
      "learning_rate": 2.8202046839996276e-05,
      "loss": 0.9665,
      "step": 51875
    },
    {
      "epoch": 4.463219478619978,
      "grad_norm": 1.1171875,
      "learning_rate": 2.819309436057399e-05,
      "loss": 1.173,
      "step": 51876
    },
    {
      "epoch": 4.463305514927299,
      "grad_norm": 0.94140625,
      "learning_rate": 2.818414326109525e-05,
      "loss": 0.6843,
      "step": 51877
    },
    {
      "epoch": 4.463391551234621,
      "grad_norm": 199.0,
      "learning_rate": 2.8175193541585987e-05,
      "loss": 0.7605,
      "step": 51878
    },
    {
      "epoch": 4.463477587541942,
      "grad_norm": 1.125,
      "learning_rate": 2.8166245202072616e-05,
      "loss": 1.5523,
      "step": 51879
    },
    {
      "epoch": 4.463563623849264,
      "grad_norm": 1.1875,
      "learning_rate": 2.815729824258123e-05,
      "loss": 1.4376,
      "step": 51880
    },
    {
      "epoch": 4.4636496601565865,
      "grad_norm": 1.0703125,
      "learning_rate": 2.8148352663137868e-05,
      "loss": 1.1223,
      "step": 51881
    },
    {
      "epoch": 4.463735696463908,
      "grad_norm": 1.0546875,
      "learning_rate": 2.813940846376889e-05,
      "loss": 0.9238,
      "step": 51882
    },
    {
      "epoch": 4.46382173277123,
      "grad_norm": 1.15625,
      "learning_rate": 2.8130465644500392e-05,
      "loss": 1.0618,
      "step": 51883
    },
    {
      "epoch": 4.463907769078551,
      "grad_norm": 1.34375,
      "learning_rate": 2.8121524205358406e-05,
      "loss": 1.8283,
      "step": 51884
    },
    {
      "epoch": 4.463993805385873,
      "grad_norm": 1.328125,
      "learning_rate": 2.81125841463693e-05,
      "loss": 1.4136,
      "step": 51885
    },
    {
      "epoch": 4.464079841693194,
      "grad_norm": 1.1484375,
      "learning_rate": 2.810364546755906e-05,
      "loss": 1.5232,
      "step": 51886
    },
    {
      "epoch": 4.464165878000516,
      "grad_norm": 2.640625,
      "learning_rate": 2.809470816895382e-05,
      "loss": 1.1516,
      "step": 51887
    },
    {
      "epoch": 4.464251914307838,
      "grad_norm": 1.5625,
      "learning_rate": 2.80857722505799e-05,
      "loss": 1.0715,
      "step": 51888
    },
    {
      "epoch": 4.4643379506151595,
      "grad_norm": 1.0234375,
      "learning_rate": 2.8076837712463166e-05,
      "loss": 0.6287,
      "step": 51889
    },
    {
      "epoch": 4.4644239869224815,
      "grad_norm": 1.2734375,
      "learning_rate": 2.8067904554629985e-05,
      "loss": 0.6608,
      "step": 51890
    },
    {
      "epoch": 4.464510023229803,
      "grad_norm": 0.7890625,
      "learning_rate": 2.8058972777106397e-05,
      "loss": 0.5115,
      "step": 51891
    },
    {
      "epoch": 4.464596059537125,
      "grad_norm": 0.69140625,
      "learning_rate": 2.8050042379918484e-05,
      "loss": 0.6102,
      "step": 51892
    },
    {
      "epoch": 4.464682095844446,
      "grad_norm": 1.03125,
      "learning_rate": 2.8041113363092508e-05,
      "loss": 0.8973,
      "step": 51893
    },
    {
      "epoch": 4.464768132151768,
      "grad_norm": 342.0,
      "learning_rate": 2.803218572665439e-05,
      "loss": 1.2715,
      "step": 51894
    },
    {
      "epoch": 4.46485416845909,
      "grad_norm": 0.84765625,
      "learning_rate": 2.8023259470630334e-05,
      "loss": 1.7372,
      "step": 51895
    },
    {
      "epoch": 4.464940204766411,
      "grad_norm": 1.0,
      "learning_rate": 2.801433459504654e-05,
      "loss": 1.5619,
      "step": 51896
    },
    {
      "epoch": 4.465026241073733,
      "grad_norm": 2.15625,
      "learning_rate": 2.8005411099928934e-05,
      "loss": 1.4014,
      "step": 51897
    },
    {
      "epoch": 4.4651122773810545,
      "grad_norm": 1.15625,
      "learning_rate": 2.7996488985303713e-05,
      "loss": 1.2851,
      "step": 51898
    },
    {
      "epoch": 4.465198313688377,
      "grad_norm": 0.9609375,
      "learning_rate": 2.7987568251197028e-05,
      "loss": 0.6999,
      "step": 51899
    },
    {
      "epoch": 4.465284349995698,
      "grad_norm": 1.046875,
      "learning_rate": 2.7978648897634795e-05,
      "loss": 0.9448,
      "step": 51900
    },
    {
      "epoch": 4.46537038630302,
      "grad_norm": 1.046875,
      "learning_rate": 2.7969730924643333e-05,
      "loss": 1.1193,
      "step": 51901
    },
    {
      "epoch": 4.465456422610342,
      "grad_norm": 0.84375,
      "learning_rate": 2.7960814332248508e-05,
      "loss": 0.4003,
      "step": 51902
    },
    {
      "epoch": 4.465542458917663,
      "grad_norm": 0.9375,
      "learning_rate": 2.795189912047652e-05,
      "loss": 0.7225,
      "step": 51903
    },
    {
      "epoch": 4.465628495224985,
      "grad_norm": 1.1953125,
      "learning_rate": 2.794298528935346e-05,
      "loss": 1.295,
      "step": 51904
    },
    {
      "epoch": 4.465714531532306,
      "grad_norm": 1.2890625,
      "learning_rate": 2.793407283890531e-05,
      "loss": 0.751,
      "step": 51905
    },
    {
      "epoch": 4.465800567839628,
      "grad_norm": 109.0,
      "learning_rate": 2.7925161769158213e-05,
      "loss": 0.8008,
      "step": 51906
    },
    {
      "epoch": 4.46588660414695,
      "grad_norm": 0.8828125,
      "learning_rate": 2.7916252080138205e-05,
      "loss": 0.7642,
      "step": 51907
    },
    {
      "epoch": 4.465972640454272,
      "grad_norm": 0.75,
      "learning_rate": 2.790734377187132e-05,
      "loss": 1.5373,
      "step": 51908
    },
    {
      "epoch": 4.466058676761594,
      "grad_norm": 0.9453125,
      "learning_rate": 2.7898436844383644e-05,
      "loss": 0.6707,
      "step": 51909
    },
    {
      "epoch": 4.466144713068915,
      "grad_norm": 1.0625,
      "learning_rate": 2.7889531297701275e-05,
      "loss": 1.2389,
      "step": 51910
    },
    {
      "epoch": 4.466230749376237,
      "grad_norm": 0.77734375,
      "learning_rate": 2.7880627131850134e-05,
      "loss": 0.9891,
      "step": 51911
    },
    {
      "epoch": 4.466316785683558,
      "grad_norm": 0.984375,
      "learning_rate": 2.7871724346856307e-05,
      "loss": 0.871,
      "step": 51912
    },
    {
      "epoch": 4.46640282199088,
      "grad_norm": 1.34375,
      "learning_rate": 2.7862822942746004e-05,
      "loss": 1.1296,
      "step": 51913
    },
    {
      "epoch": 4.466488858298202,
      "grad_norm": 2.171875,
      "learning_rate": 2.7853922919544973e-05,
      "loss": 1.3031,
      "step": 51914
    },
    {
      "epoch": 4.4665748946055235,
      "grad_norm": 988.0,
      "learning_rate": 2.784502427727953e-05,
      "loss": 1.142,
      "step": 51915
    },
    {
      "epoch": 4.4666609309128456,
      "grad_norm": 1.2421875,
      "learning_rate": 2.783612701597543e-05,
      "loss": 0.8753,
      "step": 51916
    },
    {
      "epoch": 4.466746967220167,
      "grad_norm": 1.4375,
      "learning_rate": 2.782723113565888e-05,
      "loss": 1.0836,
      "step": 51917
    },
    {
      "epoch": 4.466833003527489,
      "grad_norm": 0.83203125,
      "learning_rate": 2.7818336636355856e-05,
      "loss": 0.8795,
      "step": 51918
    },
    {
      "epoch": 4.46691903983481,
      "grad_norm": 2.515625,
      "learning_rate": 2.7809443518092337e-05,
      "loss": 1.591,
      "step": 51919
    },
    {
      "epoch": 4.467005076142132,
      "grad_norm": 1.453125,
      "learning_rate": 2.7800551780894357e-05,
      "loss": 0.5811,
      "step": 51920
    },
    {
      "epoch": 4.467091112449454,
      "grad_norm": 0.875,
      "learning_rate": 2.7791661424788006e-05,
      "loss": 0.5114,
      "step": 51921
    },
    {
      "epoch": 4.467177148756775,
      "grad_norm": 1.15625,
      "learning_rate": 2.7782772449799097e-05,
      "loss": 1.2015,
      "step": 51922
    },
    {
      "epoch": 4.467263185064097,
      "grad_norm": 1.65625,
      "learning_rate": 2.7773884855953778e-05,
      "loss": 1.413,
      "step": 51923
    },
    {
      "epoch": 4.4673492213714185,
      "grad_norm": 2.328125,
      "learning_rate": 2.7764998643278027e-05,
      "loss": 1.2319,
      "step": 51924
    },
    {
      "epoch": 4.467435257678741,
      "grad_norm": 1.4375,
      "learning_rate": 2.7756113811797768e-05,
      "loss": 1.068,
      "step": 51925
    },
    {
      "epoch": 4.467521293986062,
      "grad_norm": 3.875,
      "learning_rate": 2.7747230361539032e-05,
      "loss": 0.6916,
      "step": 51926
    },
    {
      "epoch": 4.467607330293384,
      "grad_norm": 0.890625,
      "learning_rate": 2.7738348292527805e-05,
      "loss": 1.0518,
      "step": 51927
    },
    {
      "epoch": 4.467693366600706,
      "grad_norm": 0.79296875,
      "learning_rate": 2.7729467604790058e-05,
      "loss": 0.3967,
      "step": 51928
    },
    {
      "epoch": 4.467779402908027,
      "grad_norm": 1.3515625,
      "learning_rate": 2.7720588298351835e-05,
      "loss": 1.4046,
      "step": 51929
    },
    {
      "epoch": 4.467865439215349,
      "grad_norm": 1.4609375,
      "learning_rate": 2.771171037323894e-05,
      "loss": 0.7126,
      "step": 51930
    },
    {
      "epoch": 4.46795147552267,
      "grad_norm": 224.0,
      "learning_rate": 2.770283382947758e-05,
      "loss": 0.9857,
      "step": 51931
    },
    {
      "epoch": 4.468037511829992,
      "grad_norm": 0.88671875,
      "learning_rate": 2.7693958667093454e-05,
      "loss": 1.5423,
      "step": 51932
    },
    {
      "epoch": 4.468123548137314,
      "grad_norm": 20.5,
      "learning_rate": 2.7685084886112598e-05,
      "loss": 0.783,
      "step": 51933
    },
    {
      "epoch": 4.468209584444636,
      "grad_norm": 0.9140625,
      "learning_rate": 2.767621248656116e-05,
      "loss": 0.9433,
      "step": 51934
    },
    {
      "epoch": 4.468295620751958,
      "grad_norm": 1.0625,
      "learning_rate": 2.7667341468464833e-05,
      "loss": 1.0999,
      "step": 51935
    },
    {
      "epoch": 4.468381657059279,
      "grad_norm": 1.0703125,
      "learning_rate": 2.765847183184966e-05,
      "loss": 1.3257,
      "step": 51936
    },
    {
      "epoch": 4.468467693366601,
      "grad_norm": 6.1875,
      "learning_rate": 2.7649603576741676e-05,
      "loss": 1.3526,
      "step": 51937
    },
    {
      "epoch": 4.468553729673922,
      "grad_norm": 1.109375,
      "learning_rate": 2.7640736703166637e-05,
      "loss": 0.5148,
      "step": 51938
    },
    {
      "epoch": 4.468639765981244,
      "grad_norm": 1.2265625,
      "learning_rate": 2.763187121115057e-05,
      "loss": 0.9302,
      "step": 51939
    },
    {
      "epoch": 4.468725802288565,
      "grad_norm": 1.1328125,
      "learning_rate": 2.762300710071952e-05,
      "loss": 0.7095,
      "step": 51940
    },
    {
      "epoch": 4.4688118385958875,
      "grad_norm": 1.0078125,
      "learning_rate": 2.7614144371899242e-05,
      "loss": 1.1718,
      "step": 51941
    },
    {
      "epoch": 4.4688978749032096,
      "grad_norm": 0.73046875,
      "learning_rate": 2.760528302471571e-05,
      "loss": 0.6675,
      "step": 51942
    },
    {
      "epoch": 4.468983911210531,
      "grad_norm": 1.0703125,
      "learning_rate": 2.7596423059194852e-05,
      "loss": 1.3102,
      "step": 51943
    },
    {
      "epoch": 4.469069947517853,
      "grad_norm": 1.28125,
      "learning_rate": 2.7587564475362535e-05,
      "loss": 1.1501,
      "step": 51944
    },
    {
      "epoch": 4.469155983825174,
      "grad_norm": 576.0,
      "learning_rate": 2.7578707273244853e-05,
      "loss": 1.053,
      "step": 51945
    },
    {
      "epoch": 4.469242020132496,
      "grad_norm": 2.90625,
      "learning_rate": 2.7569851452867444e-05,
      "loss": 0.8681,
      "step": 51946
    },
    {
      "epoch": 4.469328056439817,
      "grad_norm": 0.796875,
      "learning_rate": 2.756099701425635e-05,
      "loss": 0.4836,
      "step": 51947
    },
    {
      "epoch": 4.469414092747139,
      "grad_norm": 0.99609375,
      "learning_rate": 2.7552143957437493e-05,
      "loss": 0.7956,
      "step": 51948
    },
    {
      "epoch": 4.469500129054461,
      "grad_norm": 422.0,
      "learning_rate": 2.7543292282436684e-05,
      "loss": 0.89,
      "step": 51949
    },
    {
      "epoch": 4.4695861653617825,
      "grad_norm": 1.125,
      "learning_rate": 2.7534441989279846e-05,
      "loss": 0.9814,
      "step": 51950
    },
    {
      "epoch": 4.469672201669105,
      "grad_norm": 0.8984375,
      "learning_rate": 2.752559307799296e-05,
      "loss": 0.8695,
      "step": 51951
    },
    {
      "epoch": 4.469758237976426,
      "grad_norm": 700.0,
      "learning_rate": 2.7516745548601786e-05,
      "loss": 1.0096,
      "step": 51952
    },
    {
      "epoch": 4.469844274283748,
      "grad_norm": 3.0625,
      "learning_rate": 2.7507899401132187e-05,
      "loss": 1.0132,
      "step": 51953
    },
    {
      "epoch": 4.469930310591069,
      "grad_norm": 1.28125,
      "learning_rate": 2.74990546356102e-05,
      "loss": 1.7681,
      "step": 51954
    },
    {
      "epoch": 4.470016346898391,
      "grad_norm": 1.1953125,
      "learning_rate": 2.749021125206147e-05,
      "loss": 1.2684,
      "step": 51955
    },
    {
      "epoch": 4.470102383205713,
      "grad_norm": 1.0703125,
      "learning_rate": 2.748136925051209e-05,
      "loss": 1.875,
      "step": 51956
    },
    {
      "epoch": 4.470188419513034,
      "grad_norm": 0.98828125,
      "learning_rate": 2.7472528630987702e-05,
      "loss": 1.3755,
      "step": 51957
    },
    {
      "epoch": 4.470274455820356,
      "grad_norm": 1.3828125,
      "learning_rate": 2.7463689393514234e-05,
      "loss": 1.4494,
      "step": 51958
    },
    {
      "epoch": 4.470360492127678,
      "grad_norm": 1.1171875,
      "learning_rate": 2.745485153811772e-05,
      "loss": 0.7836,
      "step": 51959
    },
    {
      "epoch": 4.470446528435,
      "grad_norm": 0.8359375,
      "learning_rate": 2.744601506482375e-05,
      "loss": 1.1316,
      "step": 51960
    },
    {
      "epoch": 4.470532564742321,
      "grad_norm": 0.98828125,
      "learning_rate": 2.7437179973658245e-05,
      "loss": 0.9427,
      "step": 51961
    },
    {
      "epoch": 4.470618601049643,
      "grad_norm": 1.140625,
      "learning_rate": 2.7428346264647186e-05,
      "loss": 1.1311,
      "step": 51962
    },
    {
      "epoch": 4.470704637356965,
      "grad_norm": 1.078125,
      "learning_rate": 2.7419513937816166e-05,
      "loss": 0.6512,
      "step": 51963
    },
    {
      "epoch": 4.470790673664286,
      "grad_norm": 1.046875,
      "learning_rate": 2.7410682993191273e-05,
      "loss": 0.7277,
      "step": 51964
    },
    {
      "epoch": 4.470876709971608,
      "grad_norm": 0.8984375,
      "learning_rate": 2.7401853430798207e-05,
      "loss": 0.7185,
      "step": 51965
    },
    {
      "epoch": 4.470962746278929,
      "grad_norm": 1.0546875,
      "learning_rate": 2.739302525066273e-05,
      "loss": 2.0249,
      "step": 51966
    },
    {
      "epoch": 4.4710487825862515,
      "grad_norm": 1.0859375,
      "learning_rate": 2.738419845281087e-05,
      "loss": 1.1828,
      "step": 51967
    },
    {
      "epoch": 4.471134818893573,
      "grad_norm": 0.83984375,
      "learning_rate": 2.7375373037268225e-05,
      "loss": 0.7568,
      "step": 51968
    },
    {
      "epoch": 4.471220855200895,
      "grad_norm": 1.1640625,
      "learning_rate": 2.736654900406066e-05,
      "loss": 1.1991,
      "step": 51969
    },
    {
      "epoch": 4.471306891508217,
      "grad_norm": 2.015625,
      "learning_rate": 2.7357726353214096e-05,
      "loss": 0.9122,
      "step": 51970
    },
    {
      "epoch": 4.471392927815538,
      "grad_norm": 1.2265625,
      "learning_rate": 2.7348905084754183e-05,
      "loss": 1.4906,
      "step": 51971
    },
    {
      "epoch": 4.47147896412286,
      "grad_norm": 1.765625,
      "learning_rate": 2.7340085198706897e-05,
      "loss": 0.7488,
      "step": 51972
    },
    {
      "epoch": 4.471565000430181,
      "grad_norm": 1.078125,
      "learning_rate": 2.733126669509789e-05,
      "loss": 1.1161,
      "step": 51973
    },
    {
      "epoch": 4.471651036737503,
      "grad_norm": 1.296875,
      "learning_rate": 2.7322449573952912e-05,
      "loss": 1.4678,
      "step": 51974
    },
    {
      "epoch": 4.4717370730448245,
      "grad_norm": 1.1875,
      "learning_rate": 2.7313633835298003e-05,
      "loss": 1.3036,
      "step": 51975
    },
    {
      "epoch": 4.4718231093521466,
      "grad_norm": 1.8125,
      "learning_rate": 2.730481947915864e-05,
      "loss": 0.8104,
      "step": 51976
    },
    {
      "epoch": 4.471909145659469,
      "grad_norm": 1.3046875,
      "learning_rate": 2.72960065055608e-05,
      "loss": 0.8717,
      "step": 51977
    },
    {
      "epoch": 4.47199518196679,
      "grad_norm": 0.97265625,
      "learning_rate": 2.7287194914530244e-05,
      "loss": 1.2924,
      "step": 51978
    },
    {
      "epoch": 4.472081218274112,
      "grad_norm": 1.015625,
      "learning_rate": 2.7278384706092617e-05,
      "loss": 1.0599,
      "step": 51979
    },
    {
      "epoch": 4.472167254581433,
      "grad_norm": 1.34375,
      "learning_rate": 2.726957588027379e-05,
      "loss": 1.8835,
      "step": 51980
    },
    {
      "epoch": 4.472253290888755,
      "grad_norm": 1.171875,
      "learning_rate": 2.7260768437099627e-05,
      "loss": 1.4323,
      "step": 51981
    },
    {
      "epoch": 4.472339327196076,
      "grad_norm": 1.078125,
      "learning_rate": 2.725196237659566e-05,
      "loss": 0.6912,
      "step": 51982
    },
    {
      "epoch": 4.472425363503398,
      "grad_norm": 12.6875,
      "learning_rate": 2.7243157698787825e-05,
      "loss": 0.8747,
      "step": 51983
    },
    {
      "epoch": 4.47251139981072,
      "grad_norm": 1.0,
      "learning_rate": 2.7234354403701756e-05,
      "loss": 0.7839,
      "step": 51984
    },
    {
      "epoch": 4.472597436118042,
      "grad_norm": 1.4296875,
      "learning_rate": 2.7225552491363214e-05,
      "loss": 1.3876,
      "step": 51985
    },
    {
      "epoch": 4.472683472425364,
      "grad_norm": 1.2578125,
      "learning_rate": 2.7216751961798125e-05,
      "loss": 1.2942,
      "step": 51986
    },
    {
      "epoch": 4.472769508732685,
      "grad_norm": 1.4375,
      "learning_rate": 2.7207952815031968e-05,
      "loss": 1.4011,
      "step": 51987
    },
    {
      "epoch": 4.472855545040007,
      "grad_norm": 1.46875,
      "learning_rate": 2.7199155051090607e-05,
      "loss": 1.2783,
      "step": 51988
    },
    {
      "epoch": 4.472941581347328,
      "grad_norm": 129.0,
      "learning_rate": 2.7190358669999803e-05,
      "loss": 0.9991,
      "step": 51989
    },
    {
      "epoch": 4.47302761765465,
      "grad_norm": 51.5,
      "learning_rate": 2.7181563671785205e-05,
      "loss": 1.061,
      "step": 51990
    },
    {
      "epoch": 4.473113653961972,
      "grad_norm": 5.875,
      "learning_rate": 2.7172770056472564e-05,
      "loss": 0.9241,
      "step": 51991
    },
    {
      "epoch": 4.473199690269293,
      "grad_norm": 1.46875,
      "learning_rate": 2.7163977824087692e-05,
      "loss": 0.76,
      "step": 51992
    },
    {
      "epoch": 4.4732857265766155,
      "grad_norm": 1.2578125,
      "learning_rate": 2.7155186974656132e-05,
      "loss": 0.6142,
      "step": 51993
    },
    {
      "epoch": 4.473371762883937,
      "grad_norm": 0.88671875,
      "learning_rate": 2.714639750820369e-05,
      "loss": 0.6985,
      "step": 51994
    },
    {
      "epoch": 4.473457799191259,
      "grad_norm": 1.1640625,
      "learning_rate": 2.713760942475618e-05,
      "loss": 0.6812,
      "step": 51995
    },
    {
      "epoch": 4.473543835498581,
      "grad_norm": 1.0703125,
      "learning_rate": 2.7128822724339087e-05,
      "loss": 2.039,
      "step": 51996
    },
    {
      "epoch": 4.473629871805902,
      "grad_norm": 2.296875,
      "learning_rate": 2.712003740697827e-05,
      "loss": 1.1512,
      "step": 51997
    },
    {
      "epoch": 4.473715908113224,
      "grad_norm": 1.109375,
      "learning_rate": 2.711125347269938e-05,
      "loss": 0.5146,
      "step": 51998
    },
    {
      "epoch": 4.473801944420545,
      "grad_norm": 1.015625,
      "learning_rate": 2.7102470921528012e-05,
      "loss": 1.0626,
      "step": 51999
    },
    {
      "epoch": 4.473887980727867,
      "grad_norm": 1.09375,
      "learning_rate": 2.7093689753490025e-05,
      "loss": 0.8212,
      "step": 52000
    },
    {
      "epoch": 4.4739740170351885,
      "grad_norm": 0.9609375,
      "learning_rate": 2.708490996861096e-05,
      "loss": 0.8013,
      "step": 52001
    },
    {
      "epoch": 4.474060053342511,
      "grad_norm": 1.234375,
      "learning_rate": 2.707613156691663e-05,
      "loss": 1.0963,
      "step": 52002
    },
    {
      "epoch": 4.474146089649833,
      "grad_norm": 1.0078125,
      "learning_rate": 2.7067354548432622e-05,
      "loss": 1.39,
      "step": 52003
    },
    {
      "epoch": 4.474232125957154,
      "grad_norm": 1.21875,
      "learning_rate": 2.7058578913184583e-05,
      "loss": 0.6782,
      "step": 52004
    },
    {
      "epoch": 4.474318162264476,
      "grad_norm": 1.1328125,
      "learning_rate": 2.7049804661198273e-05,
      "loss": 1.1865,
      "step": 52005
    },
    {
      "epoch": 4.474404198571797,
      "grad_norm": 1.2265625,
      "learning_rate": 2.7041031792499227e-05,
      "loss": 0.8105,
      "step": 52006
    },
    {
      "epoch": 4.474490234879119,
      "grad_norm": 1.1640625,
      "learning_rate": 2.70322603071132e-05,
      "loss": 1.566,
      "step": 52007
    },
    {
      "epoch": 4.47457627118644,
      "grad_norm": 0.9453125,
      "learning_rate": 2.7023490205065893e-05,
      "loss": 1.0913,
      "step": 52008
    },
    {
      "epoch": 4.474662307493762,
      "grad_norm": 114.0,
      "learning_rate": 2.7014721486382844e-05,
      "loss": 1.2929,
      "step": 52009
    },
    {
      "epoch": 4.474748343801084,
      "grad_norm": 1.0625,
      "learning_rate": 2.7005954151089695e-05,
      "loss": 0.5776,
      "step": 52010
    },
    {
      "epoch": 4.474834380108406,
      "grad_norm": 1.1171875,
      "learning_rate": 2.699718819921221e-05,
      "loss": 1.5397,
      "step": 52011
    },
    {
      "epoch": 4.474920416415728,
      "grad_norm": 1.296875,
      "learning_rate": 2.6988423630775915e-05,
      "loss": 1.3233,
      "step": 52012
    },
    {
      "epoch": 4.475006452723049,
      "grad_norm": 1.0625,
      "learning_rate": 2.6979660445806575e-05,
      "loss": 0.9615,
      "step": 52013
    },
    {
      "epoch": 4.475092489030371,
      "grad_norm": 0.8359375,
      "learning_rate": 2.6970898644329612e-05,
      "loss": 0.4556,
      "step": 52014
    },
    {
      "epoch": 4.475178525337692,
      "grad_norm": 3.4375,
      "learning_rate": 2.696213822637078e-05,
      "loss": 0.6631,
      "step": 52015
    },
    {
      "epoch": 4.475264561645014,
      "grad_norm": 651264.0,
      "learning_rate": 2.6953379191955783e-05,
      "loss": 0.976,
      "step": 52016
    },
    {
      "epoch": 4.475350597952336,
      "grad_norm": 12.5,
      "learning_rate": 2.6944621541110048e-05,
      "loss": 1.1902,
      "step": 52017
    },
    {
      "epoch": 4.475436634259657,
      "grad_norm": 1.0859375,
      "learning_rate": 2.6935865273859326e-05,
      "loss": 0.9255,
      "step": 52018
    },
    {
      "epoch": 4.4755226705669795,
      "grad_norm": 1.0078125,
      "learning_rate": 2.692711039022927e-05,
      "loss": 0.935,
      "step": 52019
    },
    {
      "epoch": 4.475608706874301,
      "grad_norm": 1.3046875,
      "learning_rate": 2.6918356890245298e-05,
      "loss": 0.7721,
      "step": 52020
    },
    {
      "epoch": 4.475694743181623,
      "grad_norm": 2.765625,
      "learning_rate": 2.6909604773933172e-05,
      "loss": 0.9334,
      "step": 52021
    },
    {
      "epoch": 4.475780779488944,
      "grad_norm": 1.1796875,
      "learning_rate": 2.690085404131848e-05,
      "loss": 1.4575,
      "step": 52022
    },
    {
      "epoch": 4.475866815796266,
      "grad_norm": 0.91796875,
      "learning_rate": 2.68921046924267e-05,
      "loss": 0.4965,
      "step": 52023
    },
    {
      "epoch": 4.475952852103588,
      "grad_norm": 1.2578125,
      "learning_rate": 2.6883356727283482e-05,
      "loss": 1.7297,
      "step": 52024
    },
    {
      "epoch": 4.476038888410909,
      "grad_norm": 1.0625,
      "learning_rate": 2.6874610145914523e-05,
      "loss": 1.9128,
      "step": 52025
    },
    {
      "epoch": 4.476124924718231,
      "grad_norm": 1.1015625,
      "learning_rate": 2.6865864948345255e-05,
      "loss": 1.508,
      "step": 52026
    },
    {
      "epoch": 4.4762109610255525,
      "grad_norm": 1.03125,
      "learning_rate": 2.685712113460137e-05,
      "loss": 0.7115,
      "step": 52027
    },
    {
      "epoch": 4.476296997332875,
      "grad_norm": 1.15625,
      "learning_rate": 2.6848378704708354e-05,
      "loss": 1.8662,
      "step": 52028
    },
    {
      "epoch": 4.476383033640197,
      "grad_norm": 1.2265625,
      "learning_rate": 2.683963765869174e-05,
      "loss": 1.4448,
      "step": 52029
    },
    {
      "epoch": 4.476469069947518,
      "grad_norm": 1064.0,
      "learning_rate": 2.6830897996577284e-05,
      "loss": 1.101,
      "step": 52030
    },
    {
      "epoch": 4.47655510625484,
      "grad_norm": 1.421875,
      "learning_rate": 2.6822159718390303e-05,
      "loss": 1.5868,
      "step": 52031
    },
    {
      "epoch": 4.476641142562161,
      "grad_norm": 1.3515625,
      "learning_rate": 2.6813422824156498e-05,
      "loss": 1.3628,
      "step": 52032
    },
    {
      "epoch": 4.476727178869483,
      "grad_norm": 1.296875,
      "learning_rate": 2.6804687313901512e-05,
      "loss": 1.2011,
      "step": 52033
    },
    {
      "epoch": 4.476813215176804,
      "grad_norm": 41.75,
      "learning_rate": 2.6795953187650656e-05,
      "loss": 1.3222,
      "step": 52034
    },
    {
      "epoch": 4.476899251484126,
      "grad_norm": 1.484375,
      "learning_rate": 2.6787220445429637e-05,
      "loss": 1.9396,
      "step": 52035
    },
    {
      "epoch": 4.476985287791448,
      "grad_norm": 30.75,
      "learning_rate": 2.6778489087264046e-05,
      "loss": 1.5804,
      "step": 52036
    },
    {
      "epoch": 4.47707132409877,
      "grad_norm": 1.171875,
      "learning_rate": 2.6769759113179137e-05,
      "loss": 1.4527,
      "step": 52037
    },
    {
      "epoch": 4.477157360406092,
      "grad_norm": 5.15625,
      "learning_rate": 2.6761030523200836e-05,
      "loss": 0.9115,
      "step": 52038
    },
    {
      "epoch": 4.477243396713413,
      "grad_norm": 34.75,
      "learning_rate": 2.6752303317354398e-05,
      "loss": 1.1511,
      "step": 52039
    },
    {
      "epoch": 4.477329433020735,
      "grad_norm": 1.8671875,
      "learning_rate": 2.6743577495665415e-05,
      "loss": 1.5217,
      "step": 52040
    },
    {
      "epoch": 4.477415469328056,
      "grad_norm": 7.34375,
      "learning_rate": 2.6734853058159537e-05,
      "loss": 1.1444,
      "step": 52041
    },
    {
      "epoch": 4.477501505635378,
      "grad_norm": 0.984375,
      "learning_rate": 2.672613000486207e-05,
      "loss": 0.5647,
      "step": 52042
    },
    {
      "epoch": 4.4775875419427,
      "grad_norm": 1.015625,
      "learning_rate": 2.6717408335798722e-05,
      "loss": 0.963,
      "step": 52043
    },
    {
      "epoch": 4.477673578250021,
      "grad_norm": 1.0703125,
      "learning_rate": 2.67086880509948e-05,
      "loss": 0.8518,
      "step": 52044
    },
    {
      "epoch": 4.4777596145573435,
      "grad_norm": 1.1640625,
      "learning_rate": 2.6699969150475956e-05,
      "loss": 1.4008,
      "step": 52045
    },
    {
      "epoch": 4.477845650864665,
      "grad_norm": 1.4375,
      "learning_rate": 2.6691251634267665e-05,
      "loss": 1.0856,
      "step": 52046
    },
    {
      "epoch": 4.477931687171987,
      "grad_norm": 1.203125,
      "learning_rate": 2.6682535502395356e-05,
      "loss": 0.8542,
      "step": 52047
    },
    {
      "epoch": 4.478017723479308,
      "grad_norm": 4.75,
      "learning_rate": 2.6673820754884616e-05,
      "loss": 0.7124,
      "step": 52048
    },
    {
      "epoch": 4.47810375978663,
      "grad_norm": 1.1953125,
      "learning_rate": 2.6665107391760923e-05,
      "loss": 0.9778,
      "step": 52049
    },
    {
      "epoch": 4.478189796093952,
      "grad_norm": 0.984375,
      "learning_rate": 2.6656395413049705e-05,
      "loss": 1.2415,
      "step": 52050
    },
    {
      "epoch": 4.478275832401273,
      "grad_norm": 0.81640625,
      "learning_rate": 2.664768481877644e-05,
      "loss": 0.7338,
      "step": 52051
    },
    {
      "epoch": 4.478361868708595,
      "grad_norm": 1.046875,
      "learning_rate": 2.6638975608966665e-05,
      "loss": 0.594,
      "step": 52052
    },
    {
      "epoch": 4.4784479050159165,
      "grad_norm": 1.140625,
      "learning_rate": 2.66302677836458e-05,
      "loss": 0.5579,
      "step": 52053
    },
    {
      "epoch": 4.478533941323239,
      "grad_norm": 0.96875,
      "learning_rate": 2.6621561342839383e-05,
      "loss": 0.9381,
      "step": 52054
    },
    {
      "epoch": 4.47861997763056,
      "grad_norm": 1.734375,
      "learning_rate": 2.6612856286572785e-05,
      "loss": 1.3384,
      "step": 52055
    },
    {
      "epoch": 4.478706013937882,
      "grad_norm": 1.140625,
      "learning_rate": 2.6604152614871534e-05,
      "loss": 0.8927,
      "step": 52056
    },
    {
      "epoch": 4.478792050245204,
      "grad_norm": 1.203125,
      "learning_rate": 2.6595450327761116e-05,
      "loss": 1.5465,
      "step": 52057
    },
    {
      "epoch": 4.478878086552525,
      "grad_norm": 1.34375,
      "learning_rate": 2.658674942526684e-05,
      "loss": 1.4337,
      "step": 52058
    },
    {
      "epoch": 4.478964122859847,
      "grad_norm": 1.1796875,
      "learning_rate": 2.657804990741425e-05,
      "loss": 1.2585,
      "step": 52059
    },
    {
      "epoch": 4.479050159167168,
      "grad_norm": 0.87890625,
      "learning_rate": 2.6569351774228866e-05,
      "loss": 0.4175,
      "step": 52060
    },
    {
      "epoch": 4.47913619547449,
      "grad_norm": 1.1484375,
      "learning_rate": 2.6560655025736013e-05,
      "loss": 0.6941,
      "step": 52061
    },
    {
      "epoch": 4.479222231781812,
      "grad_norm": 1.03125,
      "learning_rate": 2.6551959661961168e-05,
      "loss": 1.1579,
      "step": 52062
    },
    {
      "epoch": 4.479308268089134,
      "grad_norm": 372.0,
      "learning_rate": 2.6543265682929805e-05,
      "loss": 0.9526,
      "step": 52063
    },
    {
      "epoch": 4.479394304396456,
      "grad_norm": 38.25,
      "learning_rate": 2.653457308866719e-05,
      "loss": 0.7187,
      "step": 52064
    },
    {
      "epoch": 4.479480340703777,
      "grad_norm": 0.94140625,
      "learning_rate": 2.652588187919891e-05,
      "loss": 0.5982,
      "step": 52065
    },
    {
      "epoch": 4.479566377011099,
      "grad_norm": 1.1328125,
      "learning_rate": 2.6517192054550442e-05,
      "loss": 0.988,
      "step": 52066
    },
    {
      "epoch": 4.47965241331842,
      "grad_norm": 1.1875,
      "learning_rate": 2.6508503614746993e-05,
      "loss": 0.7542,
      "step": 52067
    },
    {
      "epoch": 4.479738449625742,
      "grad_norm": 7744.0,
      "learning_rate": 2.6499816559814148e-05,
      "loss": 0.9742,
      "step": 52068
    },
    {
      "epoch": 4.479824485933063,
      "grad_norm": 1.546875,
      "learning_rate": 2.649113088977717e-05,
      "loss": 1.3256,
      "step": 52069
    },
    {
      "epoch": 4.479910522240385,
      "grad_norm": 1.8046875,
      "learning_rate": 2.6482446604661593e-05,
      "loss": 1.3941,
      "step": 52070
    },
    {
      "epoch": 4.4799965585477075,
      "grad_norm": 0.9453125,
      "learning_rate": 2.6473763704492782e-05,
      "loss": 0.5524,
      "step": 52071
    },
    {
      "epoch": 4.480082594855029,
      "grad_norm": 1.0390625,
      "learning_rate": 2.6465082189296108e-05,
      "loss": 1.4418,
      "step": 52072
    },
    {
      "epoch": 4.480168631162351,
      "grad_norm": 1.03125,
      "learning_rate": 2.6456402059096884e-05,
      "loss": 0.9553,
      "step": 52073
    },
    {
      "epoch": 4.480254667469672,
      "grad_norm": 1.125,
      "learning_rate": 2.6447723313920646e-05,
      "loss": 0.8975,
      "step": 52074
    },
    {
      "epoch": 4.480340703776994,
      "grad_norm": 1.1328125,
      "learning_rate": 2.6439045953792707e-05,
      "loss": 1.3829,
      "step": 52075
    },
    {
      "epoch": 4.480426740084315,
      "grad_norm": 1.6015625,
      "learning_rate": 2.6430369978738543e-05,
      "loss": 1.3452,
      "step": 52076
    },
    {
      "epoch": 4.480512776391637,
      "grad_norm": 1.2734375,
      "learning_rate": 2.6421695388783306e-05,
      "loss": 1.0441,
      "step": 52077
    },
    {
      "epoch": 4.480598812698959,
      "grad_norm": 266.0,
      "learning_rate": 2.6413022183952584e-05,
      "loss": 0.8468,
      "step": 52078
    },
    {
      "epoch": 4.4806848490062805,
      "grad_norm": 1.546875,
      "learning_rate": 2.640435036427169e-05,
      "loss": 1.0915,
      "step": 52079
    },
    {
      "epoch": 4.480770885313603,
      "grad_norm": 1.3984375,
      "learning_rate": 2.6395679929765882e-05,
      "loss": 1.5796,
      "step": 52080
    },
    {
      "epoch": 4.480856921620924,
      "grad_norm": 0.796875,
      "learning_rate": 2.638701088046064e-05,
      "loss": 1.0082,
      "step": 52081
    },
    {
      "epoch": 4.480942957928246,
      "grad_norm": 6.15625,
      "learning_rate": 2.6378343216381328e-05,
      "loss": 1.1358,
      "step": 52082
    },
    {
      "epoch": 4.481028994235567,
      "grad_norm": 2.828125,
      "learning_rate": 2.6369676937553156e-05,
      "loss": 1.3907,
      "step": 52083
    },
    {
      "epoch": 4.481115030542889,
      "grad_norm": 1.40625,
      "learning_rate": 2.6361012044001654e-05,
      "loss": 1.4308,
      "step": 52084
    },
    {
      "epoch": 4.481201066850211,
      "grad_norm": 43.5,
      "learning_rate": 2.6352348535752025e-05,
      "loss": 0.5498,
      "step": 52085
    },
    {
      "epoch": 4.481287103157532,
      "grad_norm": 5.875,
      "learning_rate": 2.6343686412829637e-05,
      "loss": 0.744,
      "step": 52086
    },
    {
      "epoch": 4.481373139464854,
      "grad_norm": 1.125,
      "learning_rate": 2.6335025675259917e-05,
      "loss": 1.5271,
      "step": 52087
    },
    {
      "epoch": 4.481459175772176,
      "grad_norm": 0.8671875,
      "learning_rate": 2.6326366323068062e-05,
      "loss": 0.5816,
      "step": 52088
    },
    {
      "epoch": 4.481545212079498,
      "grad_norm": 1.0234375,
      "learning_rate": 2.6317708356279444e-05,
      "loss": 0.7909,
      "step": 52089
    },
    {
      "epoch": 4.481631248386819,
      "grad_norm": 744.0,
      "learning_rate": 2.630905177491949e-05,
      "loss": 1.1565,
      "step": 52090
    },
    {
      "epoch": 4.481717284694141,
      "grad_norm": 1984.0,
      "learning_rate": 2.630039657901334e-05,
      "loss": 1.4837,
      "step": 52091
    },
    {
      "epoch": 4.481803321001463,
      "grad_norm": 3.734375,
      "learning_rate": 2.629174276858637e-05,
      "loss": 1.0418,
      "step": 52092
    },
    {
      "epoch": 4.481889357308784,
      "grad_norm": 1.578125,
      "learning_rate": 2.628309034366405e-05,
      "loss": 1.393,
      "step": 52093
    },
    {
      "epoch": 4.481975393616106,
      "grad_norm": 0.9296875,
      "learning_rate": 2.627443930427148e-05,
      "loss": 0.6863,
      "step": 52094
    },
    {
      "epoch": 4.482061429923427,
      "grad_norm": 1.640625,
      "learning_rate": 2.626578965043408e-05,
      "loss": 1.1166,
      "step": 52095
    },
    {
      "epoch": 4.4821474662307494,
      "grad_norm": 1.359375,
      "learning_rate": 2.6257141382177107e-05,
      "loss": 1.6026,
      "step": 52096
    },
    {
      "epoch": 4.482233502538071,
      "grad_norm": 1.40625,
      "learning_rate": 2.6248494499525766e-05,
      "loss": 1.1886,
      "step": 52097
    },
    {
      "epoch": 4.482319538845393,
      "grad_norm": 0.9609375,
      "learning_rate": 2.6239849002505588e-05,
      "loss": 1.1464,
      "step": 52098
    },
    {
      "epoch": 4.482405575152715,
      "grad_norm": 1.21875,
      "learning_rate": 2.6231204891141613e-05,
      "loss": 0.8921,
      "step": 52099
    },
    {
      "epoch": 4.482491611460036,
      "grad_norm": 1.1875,
      "learning_rate": 2.6222562165459207e-05,
      "loss": 0.8795,
      "step": 52100
    },
    {
      "epoch": 4.482577647767358,
      "grad_norm": 0.78515625,
      "learning_rate": 2.621392082548374e-05,
      "loss": 1.4683,
      "step": 52101
    },
    {
      "epoch": 4.482663684074679,
      "grad_norm": 2.0,
      "learning_rate": 2.6205280871240302e-05,
      "loss": 0.7325,
      "step": 52102
    },
    {
      "epoch": 4.482749720382001,
      "grad_norm": 1.25,
      "learning_rate": 2.6196642302754313e-05,
      "loss": 1.2912,
      "step": 52103
    },
    {
      "epoch": 4.482835756689322,
      "grad_norm": 0.91015625,
      "learning_rate": 2.6188005120051095e-05,
      "loss": 1.1941,
      "step": 52104
    },
    {
      "epoch": 4.4829217929966445,
      "grad_norm": 0.7890625,
      "learning_rate": 2.6179369323155677e-05,
      "loss": 0.7244,
      "step": 52105
    },
    {
      "epoch": 4.483007829303967,
      "grad_norm": 6.625,
      "learning_rate": 2.6170734912093487e-05,
      "loss": 1.0053,
      "step": 52106
    },
    {
      "epoch": 4.483093865611288,
      "grad_norm": 1.0078125,
      "learning_rate": 2.616210188688978e-05,
      "loss": 0.9542,
      "step": 52107
    },
    {
      "epoch": 4.48317990191861,
      "grad_norm": 1.0625,
      "learning_rate": 2.6153470247569754e-05,
      "loss": 0.8106,
      "step": 52108
    },
    {
      "epoch": 4.483265938225931,
      "grad_norm": 1.6015625,
      "learning_rate": 2.6144839994158675e-05,
      "loss": 0.9656,
      "step": 52109
    },
    {
      "epoch": 4.483351974533253,
      "grad_norm": 1.40625,
      "learning_rate": 2.6136211126681742e-05,
      "loss": 1.4491,
      "step": 52110
    },
    {
      "epoch": 4.483438010840575,
      "grad_norm": 1.21875,
      "learning_rate": 2.6127583645164156e-05,
      "loss": 1.4385,
      "step": 52111
    },
    {
      "epoch": 4.483524047147896,
      "grad_norm": 1.6953125,
      "learning_rate": 2.6118957549631396e-05,
      "loss": 1.8206,
      "step": 52112
    },
    {
      "epoch": 4.483610083455218,
      "grad_norm": 0.98828125,
      "learning_rate": 2.6110332840108386e-05,
      "loss": 0.7249,
      "step": 52113
    },
    {
      "epoch": 4.48369611976254,
      "grad_norm": 1.15625,
      "learning_rate": 2.610170951662061e-05,
      "loss": 1.3622,
      "step": 52114
    },
    {
      "epoch": 4.483782156069862,
      "grad_norm": 1.4375,
      "learning_rate": 2.6093087579193096e-05,
      "loss": 1.2868,
      "step": 52115
    },
    {
      "epoch": 4.483868192377183,
      "grad_norm": 1.2421875,
      "learning_rate": 2.608446702785111e-05,
      "loss": 1.6979,
      "step": 52116
    },
    {
      "epoch": 4.483954228684505,
      "grad_norm": 1.671875,
      "learning_rate": 2.607584786261996e-05,
      "loss": 1.2133,
      "step": 52117
    },
    {
      "epoch": 4.484040264991827,
      "grad_norm": 1.296875,
      "learning_rate": 2.6067230083524684e-05,
      "loss": 1.2324,
      "step": 52118
    },
    {
      "epoch": 4.484126301299148,
      "grad_norm": 7.5,
      "learning_rate": 2.6058613690590648e-05,
      "loss": 1.2899,
      "step": 52119
    },
    {
      "epoch": 4.48421233760647,
      "grad_norm": 1.3671875,
      "learning_rate": 2.6049998683843e-05,
      "loss": 0.8569,
      "step": 52120
    },
    {
      "epoch": 4.484298373913791,
      "grad_norm": 1.453125,
      "learning_rate": 2.6041385063306887e-05,
      "loss": 1.4841,
      "step": 52121
    },
    {
      "epoch": 4.4843844102211134,
      "grad_norm": 1.3515625,
      "learning_rate": 2.6032772829007567e-05,
      "loss": 1.8989,
      "step": 52122
    },
    {
      "epoch": 4.484470446528435,
      "grad_norm": 1.09375,
      "learning_rate": 2.602416198097024e-05,
      "loss": 0.9555,
      "step": 52123
    },
    {
      "epoch": 4.484556482835757,
      "grad_norm": 1.1484375,
      "learning_rate": 2.601555251922e-05,
      "loss": 1.509,
      "step": 52124
    },
    {
      "epoch": 4.484642519143079,
      "grad_norm": 1.3984375,
      "learning_rate": 2.60069444437821e-05,
      "loss": 0.9919,
      "step": 52125
    },
    {
      "epoch": 4.4847285554504,
      "grad_norm": 1.21875,
      "learning_rate": 2.599833775468169e-05,
      "loss": 0.7134,
      "step": 52126
    },
    {
      "epoch": 4.484814591757722,
      "grad_norm": 86.5,
      "learning_rate": 2.598973245194397e-05,
      "loss": 1.1409,
      "step": 52127
    },
    {
      "epoch": 4.484900628065043,
      "grad_norm": 1.1875,
      "learning_rate": 2.598112853559409e-05,
      "loss": 1.2708,
      "step": 52128
    },
    {
      "epoch": 4.484986664372365,
      "grad_norm": 1.0234375,
      "learning_rate": 2.5972526005657194e-05,
      "loss": 1.0174,
      "step": 52129
    },
    {
      "epoch": 4.485072700679686,
      "grad_norm": 106.0,
      "learning_rate": 2.5963924862158482e-05,
      "loss": 0.7779,
      "step": 52130
    },
    {
      "epoch": 4.4851587369870085,
      "grad_norm": 1.421875,
      "learning_rate": 2.5955325105123162e-05,
      "loss": 1.005,
      "step": 52131
    },
    {
      "epoch": 4.485244773294331,
      "grad_norm": 1.8359375,
      "learning_rate": 2.594672673457621e-05,
      "loss": 1.2878,
      "step": 52132
    },
    {
      "epoch": 4.485330809601652,
      "grad_norm": 0.78125,
      "learning_rate": 2.593812975054294e-05,
      "loss": 0.5369,
      "step": 52133
    },
    {
      "epoch": 4.485416845908974,
      "grad_norm": 1.0546875,
      "learning_rate": 2.592953415304844e-05,
      "loss": 1.1619,
      "step": 52134
    },
    {
      "epoch": 4.485502882216295,
      "grad_norm": 207.0,
      "learning_rate": 2.5920939942117862e-05,
      "loss": 1.5816,
      "step": 52135
    },
    {
      "epoch": 4.485588918523617,
      "grad_norm": 1.9296875,
      "learning_rate": 2.5912347117776348e-05,
      "loss": 0.8579,
      "step": 52136
    },
    {
      "epoch": 4.485674954830938,
      "grad_norm": 2.5625,
      "learning_rate": 2.5903755680048934e-05,
      "loss": 1.1564,
      "step": 52137
    },
    {
      "epoch": 4.48576099113826,
      "grad_norm": 1.421875,
      "learning_rate": 2.5895165628960883e-05,
      "loss": 1.2063,
      "step": 52138
    },
    {
      "epoch": 4.485847027445582,
      "grad_norm": 1.171875,
      "learning_rate": 2.588657696453728e-05,
      "loss": 1.6054,
      "step": 52139
    },
    {
      "epoch": 4.485933063752904,
      "grad_norm": 0.9453125,
      "learning_rate": 2.5877989686803162e-05,
      "loss": 0.7744,
      "step": 52140
    },
    {
      "epoch": 4.486019100060226,
      "grad_norm": 17.5,
      "learning_rate": 2.586940379578373e-05,
      "loss": 0.6701,
      "step": 52141
    },
    {
      "epoch": 4.486105136367547,
      "grad_norm": 0.84375,
      "learning_rate": 2.5860819291504132e-05,
      "loss": 0.8833,
      "step": 52142
    },
    {
      "epoch": 4.486191172674869,
      "grad_norm": 1.0,
      "learning_rate": 2.585223617398941e-05,
      "loss": 1.2626,
      "step": 52143
    },
    {
      "epoch": 4.486277208982191,
      "grad_norm": 1.40625,
      "learning_rate": 2.5843654443264642e-05,
      "loss": 1.1487,
      "step": 52144
    },
    {
      "epoch": 4.486363245289512,
      "grad_norm": 1.34375,
      "learning_rate": 2.583507409935504e-05,
      "loss": 1.2745,
      "step": 52145
    },
    {
      "epoch": 4.486449281596834,
      "grad_norm": 1.2109375,
      "learning_rate": 2.5826495142285523e-05,
      "loss": 1.3012,
      "step": 52146
    },
    {
      "epoch": 4.486535317904155,
      "grad_norm": 0.9921875,
      "learning_rate": 2.5817917572081294e-05,
      "loss": 0.8393,
      "step": 52147
    },
    {
      "epoch": 4.4866213542114775,
      "grad_norm": 47.25,
      "learning_rate": 2.5809341388767448e-05,
      "loss": 1.9167,
      "step": 52148
    },
    {
      "epoch": 4.486707390518799,
      "grad_norm": 0.94140625,
      "learning_rate": 2.5800766592369075e-05,
      "loss": 0.4159,
      "step": 52149
    },
    {
      "epoch": 4.486793426826121,
      "grad_norm": 1.0390625,
      "learning_rate": 2.5792193182911318e-05,
      "loss": 1.0923,
      "step": 52150
    },
    {
      "epoch": 4.486879463133443,
      "grad_norm": 1.1953125,
      "learning_rate": 2.5783621160419047e-05,
      "loss": 0.8632,
      "step": 52151
    },
    {
      "epoch": 4.486965499440764,
      "grad_norm": 0.9296875,
      "learning_rate": 2.5775050524917464e-05,
      "loss": 0.9877,
      "step": 52152
    },
    {
      "epoch": 4.487051535748086,
      "grad_norm": 0.94140625,
      "learning_rate": 2.576648127643172e-05,
      "loss": 1.0278,
      "step": 52153
    },
    {
      "epoch": 4.487137572055407,
      "grad_norm": 1.3046875,
      "learning_rate": 2.5757913414986677e-05,
      "loss": 1.7745,
      "step": 52154
    },
    {
      "epoch": 4.487223608362729,
      "grad_norm": 1.21875,
      "learning_rate": 2.5749346940607598e-05,
      "loss": 1.0842,
      "step": 52155
    },
    {
      "epoch": 4.4873096446700504,
      "grad_norm": 1.015625,
      "learning_rate": 2.5740781853319406e-05,
      "loss": 1.0906,
      "step": 52156
    },
    {
      "epoch": 4.4873956809773725,
      "grad_norm": 22.875,
      "learning_rate": 2.5732218153147135e-05,
      "loss": 0.5331,
      "step": 52157
    },
    {
      "epoch": 4.487481717284695,
      "grad_norm": 1.2734375,
      "learning_rate": 2.572365584011599e-05,
      "loss": 1.0985,
      "step": 52158
    },
    {
      "epoch": 4.487567753592016,
      "grad_norm": 0.87890625,
      "learning_rate": 2.571509491425078e-05,
      "loss": 0.9074,
      "step": 52159
    },
    {
      "epoch": 4.487653789899338,
      "grad_norm": 376.0,
      "learning_rate": 2.570653537557671e-05,
      "loss": 0.9848,
      "step": 52160
    },
    {
      "epoch": 4.487739826206659,
      "grad_norm": 0.77734375,
      "learning_rate": 2.5697977224118874e-05,
      "loss": 0.4023,
      "step": 52161
    },
    {
      "epoch": 4.487825862513981,
      "grad_norm": 2.625,
      "learning_rate": 2.5689420459902136e-05,
      "loss": 0.9296,
      "step": 52162
    },
    {
      "epoch": 4.487911898821302,
      "grad_norm": 1.3046875,
      "learning_rate": 2.5680865082951588e-05,
      "loss": 1.2876,
      "step": 52163
    },
    {
      "epoch": 4.487997935128624,
      "grad_norm": 1.2265625,
      "learning_rate": 2.5672311093292322e-05,
      "loss": 0.9315,
      "step": 52164
    },
    {
      "epoch": 4.488083971435946,
      "grad_norm": 0.90625,
      "learning_rate": 2.566375849094921e-05,
      "loss": 1.2082,
      "step": 52165
    },
    {
      "epoch": 4.488170007743268,
      "grad_norm": 0.7890625,
      "learning_rate": 2.5655207275947444e-05,
      "loss": 0.863,
      "step": 52166
    },
    {
      "epoch": 4.48825604405059,
      "grad_norm": 1.046875,
      "learning_rate": 2.564665744831185e-05,
      "loss": 1.7705,
      "step": 52167
    },
    {
      "epoch": 4.488342080357911,
      "grad_norm": 11.0625,
      "learning_rate": 2.5638109008067568e-05,
      "loss": 0.8679,
      "step": 52168
    },
    {
      "epoch": 4.488428116665233,
      "grad_norm": 1.3125,
      "learning_rate": 2.5629561955239576e-05,
      "loss": 1.0332,
      "step": 52169
    },
    {
      "epoch": 4.488514152972554,
      "grad_norm": 145.0,
      "learning_rate": 2.5621016289852806e-05,
      "loss": 0.6207,
      "step": 52170
    },
    {
      "epoch": 4.488600189279876,
      "grad_norm": 1.0546875,
      "learning_rate": 2.5612472011932288e-05,
      "loss": 0.7442,
      "step": 52171
    },
    {
      "epoch": 4.488686225587198,
      "grad_norm": 0.93359375,
      "learning_rate": 2.5603929121503112e-05,
      "loss": 1.1671,
      "step": 52172
    },
    {
      "epoch": 4.488772261894519,
      "grad_norm": 0.79296875,
      "learning_rate": 2.559538761859015e-05,
      "loss": 0.4441,
      "step": 52173
    },
    {
      "epoch": 4.4888582982018415,
      "grad_norm": 202.0,
      "learning_rate": 2.5586847503218326e-05,
      "loss": 1.3232,
      "step": 52174
    },
    {
      "epoch": 4.488944334509163,
      "grad_norm": 1.328125,
      "learning_rate": 2.557830877541284e-05,
      "loss": 1.0963,
      "step": 52175
    },
    {
      "epoch": 4.489030370816485,
      "grad_norm": 0.921875,
      "learning_rate": 2.556977143519845e-05,
      "loss": 1.046,
      "step": 52176
    },
    {
      "epoch": 4.489116407123806,
      "grad_norm": 0.890625,
      "learning_rate": 2.5561235482600197e-05,
      "loss": 0.5272,
      "step": 52177
    },
    {
      "epoch": 4.489202443431128,
      "grad_norm": 1.328125,
      "learning_rate": 2.555270091764311e-05,
      "loss": 1.3666,
      "step": 52178
    },
    {
      "epoch": 4.48928847973845,
      "grad_norm": 0.91015625,
      "learning_rate": 2.5544167740352063e-05,
      "loss": 0.5514,
      "step": 52179
    },
    {
      "epoch": 4.489374516045771,
      "grad_norm": 0.984375,
      "learning_rate": 2.553563595075209e-05,
      "loss": 0.7123,
      "step": 52180
    },
    {
      "epoch": 4.489460552353093,
      "grad_norm": 0.99609375,
      "learning_rate": 2.5527105548868058e-05,
      "loss": 1.5946,
      "step": 52181
    },
    {
      "epoch": 4.4895465886604145,
      "grad_norm": 3.3125,
      "learning_rate": 2.5518576534724948e-05,
      "loss": 0.7518,
      "step": 52182
    },
    {
      "epoch": 4.4896326249677365,
      "grad_norm": 1.5,
      "learning_rate": 2.5510048908347795e-05,
      "loss": 0.9206,
      "step": 52183
    },
    {
      "epoch": 4.489718661275058,
      "grad_norm": 0.953125,
      "learning_rate": 2.5501522669761413e-05,
      "loss": 1.6559,
      "step": 52184
    },
    {
      "epoch": 4.48980469758238,
      "grad_norm": 0.9765625,
      "learning_rate": 2.5492997818990783e-05,
      "loss": 1.6037,
      "step": 52185
    },
    {
      "epoch": 4.489890733889702,
      "grad_norm": 1.1484375,
      "learning_rate": 2.5484474356060828e-05,
      "loss": 1.0965,
      "step": 52186
    },
    {
      "epoch": 4.489976770197023,
      "grad_norm": 1.2109375,
      "learning_rate": 2.5475952280996472e-05,
      "loss": 1.1271,
      "step": 52187
    },
    {
      "epoch": 4.490062806504345,
      "grad_norm": 1.1484375,
      "learning_rate": 2.546743159382281e-05,
      "loss": 1.0055,
      "step": 52188
    },
    {
      "epoch": 4.490148842811666,
      "grad_norm": 0.99609375,
      "learning_rate": 2.5458912294564483e-05,
      "loss": 1.0701,
      "step": 52189
    },
    {
      "epoch": 4.490234879118988,
      "grad_norm": 1.1640625,
      "learning_rate": 2.545039438324659e-05,
      "loss": 1.3927,
      "step": 52190
    },
    {
      "epoch": 4.4903209154263095,
      "grad_norm": 0.93359375,
      "learning_rate": 2.5441877859894046e-05,
      "loss": 0.9874,
      "step": 52191
    },
    {
      "epoch": 4.490406951733632,
      "grad_norm": 1.2109375,
      "learning_rate": 2.5433362724531616e-05,
      "loss": 1.4265,
      "step": 52192
    },
    {
      "epoch": 4.490492988040954,
      "grad_norm": 1.0,
      "learning_rate": 2.5424848977184332e-05,
      "loss": 0.7807,
      "step": 52193
    },
    {
      "epoch": 4.490579024348275,
      "grad_norm": 0.82421875,
      "learning_rate": 2.5416336617877177e-05,
      "loss": 0.4239,
      "step": 52194
    },
    {
      "epoch": 4.490665060655597,
      "grad_norm": 1.4296875,
      "learning_rate": 2.5407825646634797e-05,
      "loss": 0.9876,
      "step": 52195
    },
    {
      "epoch": 4.490751096962918,
      "grad_norm": 1.8125,
      "learning_rate": 2.5399316063482337e-05,
      "loss": 0.8415,
      "step": 52196
    },
    {
      "epoch": 4.49083713327024,
      "grad_norm": 0.92578125,
      "learning_rate": 2.53908078684445e-05,
      "loss": 1.3791,
      "step": 52197
    },
    {
      "epoch": 4.490923169577561,
      "grad_norm": 2.046875,
      "learning_rate": 2.538230106154621e-05,
      "loss": 0.9779,
      "step": 52198
    },
    {
      "epoch": 4.491009205884883,
      "grad_norm": 1.125,
      "learning_rate": 2.5373795642812447e-05,
      "loss": 0.9092,
      "step": 52199
    },
    {
      "epoch": 4.4910952421922055,
      "grad_norm": 1.34375,
      "learning_rate": 2.5365291612268026e-05,
      "loss": 1.5404,
      "step": 52200
    },
    {
      "epoch": 4.491181278499527,
      "grad_norm": 1.8125,
      "learning_rate": 2.535678896993776e-05,
      "loss": 0.5711,
      "step": 52201
    },
    {
      "epoch": 4.491267314806849,
      "grad_norm": 4.78125,
      "learning_rate": 2.534828771584663e-05,
      "loss": 0.8224,
      "step": 52202
    },
    {
      "epoch": 4.49135335111417,
      "grad_norm": 1.0546875,
      "learning_rate": 2.533978785001939e-05,
      "loss": 1.2166,
      "step": 52203
    },
    {
      "epoch": 4.491439387421492,
      "grad_norm": 4.0,
      "learning_rate": 2.533128937248097e-05,
      "loss": 1.7947,
      "step": 52204
    },
    {
      "epoch": 4.491525423728813,
      "grad_norm": 0.90625,
      "learning_rate": 2.532279228325629e-05,
      "loss": 0.8786,
      "step": 52205
    },
    {
      "epoch": 4.491611460036135,
      "grad_norm": 1.3125,
      "learning_rate": 2.531429658237e-05,
      "loss": 0.5699,
      "step": 52206
    },
    {
      "epoch": 4.491697496343457,
      "grad_norm": 1.8046875,
      "learning_rate": 2.530580226984719e-05,
      "loss": 0.824,
      "step": 52207
    },
    {
      "epoch": 4.4917835326507785,
      "grad_norm": 1.0390625,
      "learning_rate": 2.5297309345712503e-05,
      "loss": 1.2926,
      "step": 52208
    },
    {
      "epoch": 4.4918695689581005,
      "grad_norm": 1.4375,
      "learning_rate": 2.5288817809990815e-05,
      "loss": 0.8208,
      "step": 52209
    },
    {
      "epoch": 4.491955605265422,
      "grad_norm": 0.734375,
      "learning_rate": 2.528032766270716e-05,
      "loss": 0.5004,
      "step": 52210
    },
    {
      "epoch": 4.492041641572744,
      "grad_norm": 1.90625,
      "learning_rate": 2.5271838903886068e-05,
      "loss": 1.0625,
      "step": 52211
    },
    {
      "epoch": 4.492127677880065,
      "grad_norm": 0.9453125,
      "learning_rate": 2.526335153355258e-05,
      "loss": 0.5859,
      "step": 52212
    },
    {
      "epoch": 4.492213714187387,
      "grad_norm": 1.1328125,
      "learning_rate": 2.525486555173151e-05,
      "loss": 1.6305,
      "step": 52213
    },
    {
      "epoch": 4.492299750494709,
      "grad_norm": 1.53125,
      "learning_rate": 2.5246380958447556e-05,
      "loss": 0.948,
      "step": 52214
    },
    {
      "epoch": 4.49238578680203,
      "grad_norm": 1.1171875,
      "learning_rate": 2.5237897753725593e-05,
      "loss": 1.6294,
      "step": 52215
    },
    {
      "epoch": 4.492471823109352,
      "grad_norm": 1.0,
      "learning_rate": 2.5229415937590538e-05,
      "loss": 1.2312,
      "step": 52216
    },
    {
      "epoch": 4.4925578594166735,
      "grad_norm": 2.15625,
      "learning_rate": 2.5220935510066988e-05,
      "loss": 0.8314,
      "step": 52217
    },
    {
      "epoch": 4.492643895723996,
      "grad_norm": 1.140625,
      "learning_rate": 2.5212456471179925e-05,
      "loss": 1.115,
      "step": 52218
    },
    {
      "epoch": 4.492729932031317,
      "grad_norm": 166.0,
      "learning_rate": 2.5203978820954098e-05,
      "loss": 1.5714,
      "step": 52219
    },
    {
      "epoch": 4.492815968338639,
      "grad_norm": 14.8125,
      "learning_rate": 2.5195502559414218e-05,
      "loss": 1.0777,
      "step": 52220
    },
    {
      "epoch": 4.492902004645961,
      "grad_norm": 16.25,
      "learning_rate": 2.518702768658526e-05,
      "loss": 1.1975,
      "step": 52221
    },
    {
      "epoch": 4.492988040953282,
      "grad_norm": 1.484375,
      "learning_rate": 2.5178554202491765e-05,
      "loss": 1.6549,
      "step": 52222
    },
    {
      "epoch": 4.493074077260604,
      "grad_norm": 1.0625,
      "learning_rate": 2.5170082107158764e-05,
      "loss": 1.0725,
      "step": 52223
    },
    {
      "epoch": 4.493160113567925,
      "grad_norm": 107.5,
      "learning_rate": 2.516161140061085e-05,
      "loss": 0.5535,
      "step": 52224
    },
    {
      "epoch": 4.493246149875247,
      "grad_norm": 11.4375,
      "learning_rate": 2.5153142082872893e-05,
      "loss": 1.1095,
      "step": 52225
    },
    {
      "epoch": 4.493332186182569,
      "grad_norm": 1.1015625,
      "learning_rate": 2.5144674153969704e-05,
      "loss": 0.4744,
      "step": 52226
    },
    {
      "epoch": 4.493418222489891,
      "grad_norm": 229.0,
      "learning_rate": 2.5136207613925876e-05,
      "loss": 0.9303,
      "step": 52227
    },
    {
      "epoch": 4.493504258797213,
      "grad_norm": 1.109375,
      "learning_rate": 2.5127742462766335e-05,
      "loss": 0.9475,
      "step": 52228
    },
    {
      "epoch": 4.493590295104534,
      "grad_norm": 0.8984375,
      "learning_rate": 2.5119278700515836e-05,
      "loss": 0.9782,
      "step": 52229
    },
    {
      "epoch": 4.493676331411856,
      "grad_norm": 1.140625,
      "learning_rate": 2.511081632719908e-05,
      "loss": 0.9927,
      "step": 52230
    },
    {
      "epoch": 4.493762367719177,
      "grad_norm": 1.203125,
      "learning_rate": 2.5102355342840775e-05,
      "loss": 0.9376,
      "step": 52231
    },
    {
      "epoch": 4.493848404026499,
      "grad_norm": 1.09375,
      "learning_rate": 2.509389574746579e-05,
      "loss": 1.1174,
      "step": 52232
    },
    {
      "epoch": 4.493934440333821,
      "grad_norm": 1.125,
      "learning_rate": 2.508543754109871e-05,
      "loss": 1.541,
      "step": 52233
    },
    {
      "epoch": 4.4940204766411425,
      "grad_norm": 33.5,
      "learning_rate": 2.5076980723764353e-05,
      "loss": 1.0199,
      "step": 52234
    },
    {
      "epoch": 4.4941065129484645,
      "grad_norm": 0.671875,
      "learning_rate": 2.5068525295487587e-05,
      "loss": 0.4507,
      "step": 52235
    },
    {
      "epoch": 4.494192549255786,
      "grad_norm": 2192.0,
      "learning_rate": 2.5060071256292893e-05,
      "loss": 0.9235,
      "step": 52236
    },
    {
      "epoch": 4.494278585563108,
      "grad_norm": 1.71875,
      "learning_rate": 2.5051618606205195e-05,
      "loss": 0.5123,
      "step": 52237
    },
    {
      "epoch": 4.494364621870429,
      "grad_norm": 0.9296875,
      "learning_rate": 2.5043167345249085e-05,
      "loss": 0.7567,
      "step": 52238
    },
    {
      "epoch": 4.494450658177751,
      "grad_norm": 1.1953125,
      "learning_rate": 2.503471747344932e-05,
      "loss": 1.2363,
      "step": 52239
    },
    {
      "epoch": 4.494536694485073,
      "grad_norm": 250.0,
      "learning_rate": 2.5026268990830712e-05,
      "loss": 1.0889,
      "step": 52240
    },
    {
      "epoch": 4.494622730792394,
      "grad_norm": 1.25,
      "learning_rate": 2.501782189741786e-05,
      "loss": 1.0841,
      "step": 52241
    },
    {
      "epoch": 4.494708767099716,
      "grad_norm": 0.875,
      "learning_rate": 2.5009376193235455e-05,
      "loss": 0.7748,
      "step": 52242
    },
    {
      "epoch": 4.4947948034070375,
      "grad_norm": 1.078125,
      "learning_rate": 2.500093187830832e-05,
      "loss": 1.6288,
      "step": 52243
    },
    {
      "epoch": 4.49488083971436,
      "grad_norm": 1.046875,
      "learning_rate": 2.4992488952660987e-05,
      "loss": 0.7732,
      "step": 52244
    },
    {
      "epoch": 4.494966876021681,
      "grad_norm": 0.98046875,
      "learning_rate": 2.4984047416318268e-05,
      "loss": 1.8659,
      "step": 52245
    },
    {
      "epoch": 4.495052912329003,
      "grad_norm": 1.1796875,
      "learning_rate": 2.4975607269304867e-05,
      "loss": 0.7599,
      "step": 52246
    },
    {
      "epoch": 4.495138948636325,
      "grad_norm": 1.0625,
      "learning_rate": 2.4967168511645376e-05,
      "loss": 0.7146,
      "step": 52247
    },
    {
      "epoch": 4.495224984943646,
      "grad_norm": 0.921875,
      "learning_rate": 2.495873114336461e-05,
      "loss": 0.6265,
      "step": 52248
    },
    {
      "epoch": 4.495311021250968,
      "grad_norm": 1.2265625,
      "learning_rate": 2.4950295164487104e-05,
      "loss": 1.0484,
      "step": 52249
    },
    {
      "epoch": 4.495397057558289,
      "grad_norm": 34.0,
      "learning_rate": 2.4941860575037556e-05,
      "loss": 1.0944,
      "step": 52250
    },
    {
      "epoch": 4.495483093865611,
      "grad_norm": 1.21875,
      "learning_rate": 2.4933427375040786e-05,
      "loss": 1.3689,
      "step": 52251
    },
    {
      "epoch": 4.495569130172933,
      "grad_norm": 1.09375,
      "learning_rate": 2.4924995564521214e-05,
      "loss": 0.8967,
      "step": 52252
    },
    {
      "epoch": 4.495655166480255,
      "grad_norm": 1.03125,
      "learning_rate": 2.491656514350371e-05,
      "loss": 0.9871,
      "step": 52253
    },
    {
      "epoch": 4.495741202787577,
      "grad_norm": 1.25,
      "learning_rate": 2.4908136112012868e-05,
      "loss": 1.3104,
      "step": 52254
    },
    {
      "epoch": 4.495827239094898,
      "grad_norm": 25.75,
      "learning_rate": 2.489970847007328e-05,
      "loss": 0.4693,
      "step": 52255
    },
    {
      "epoch": 4.49591327540222,
      "grad_norm": 1.3359375,
      "learning_rate": 2.4891282217709644e-05,
      "loss": 1.1956,
      "step": 52256
    },
    {
      "epoch": 4.495999311709541,
      "grad_norm": 13.6875,
      "learning_rate": 2.4882857354946664e-05,
      "loss": 1.5462,
      "step": 52257
    },
    {
      "epoch": 4.496085348016863,
      "grad_norm": 1.125,
      "learning_rate": 2.4874433881808766e-05,
      "loss": 1.3782,
      "step": 52258
    },
    {
      "epoch": 4.496171384324184,
      "grad_norm": 1.15625,
      "learning_rate": 2.4866011798320874e-05,
      "loss": 1.0421,
      "step": 52259
    },
    {
      "epoch": 4.4962574206315065,
      "grad_norm": 22.5,
      "learning_rate": 2.485759110450747e-05,
      "loss": 1.0732,
      "step": 52260
    },
    {
      "epoch": 4.4963434569388285,
      "grad_norm": 1.1640625,
      "learning_rate": 2.48491718003932e-05,
      "loss": 1.5087,
      "step": 52261
    },
    {
      "epoch": 4.49642949324615,
      "grad_norm": 3.609375,
      "learning_rate": 2.484075388600271e-05,
      "loss": 1.5751,
      "step": 52262
    },
    {
      "epoch": 4.496515529553472,
      "grad_norm": 1.6171875,
      "learning_rate": 2.483233736136059e-05,
      "loss": 1.6131,
      "step": 52263
    },
    {
      "epoch": 4.496601565860793,
      "grad_norm": 0.94921875,
      "learning_rate": 2.4823922226491435e-05,
      "loss": 1.093,
      "step": 52264
    },
    {
      "epoch": 4.496687602168115,
      "grad_norm": 2.3125,
      "learning_rate": 2.4815508481419946e-05,
      "loss": 1.5691,
      "step": 52265
    },
    {
      "epoch": 4.496773638475437,
      "grad_norm": 1.234375,
      "learning_rate": 2.4807096126170603e-05,
      "loss": 1.1293,
      "step": 52266
    },
    {
      "epoch": 4.496859674782758,
      "grad_norm": 1.328125,
      "learning_rate": 2.4798685160768164e-05,
      "loss": 1.6469,
      "step": 52267
    },
    {
      "epoch": 4.49694571109008,
      "grad_norm": 1.125,
      "learning_rate": 2.479027558523711e-05,
      "loss": 0.9238,
      "step": 52268
    },
    {
      "epoch": 4.4970317473974015,
      "grad_norm": 1.0859375,
      "learning_rate": 2.478186739960203e-05,
      "loss": 1.28,
      "step": 52269
    },
    {
      "epoch": 4.497117783704724,
      "grad_norm": 1.1796875,
      "learning_rate": 2.4773460603887688e-05,
      "loss": 1.2284,
      "step": 52270
    },
    {
      "epoch": 4.497203820012045,
      "grad_norm": 1.4453125,
      "learning_rate": 2.4765055198118446e-05,
      "loss": 1.5275,
      "step": 52271
    },
    {
      "epoch": 4.497289856319367,
      "grad_norm": 0.92578125,
      "learning_rate": 2.4756651182319013e-05,
      "loss": 0.4851,
      "step": 52272
    },
    {
      "epoch": 4.497375892626689,
      "grad_norm": 1.234375,
      "learning_rate": 2.4748248556513975e-05,
      "loss": 1.5129,
      "step": 52273
    },
    {
      "epoch": 4.49746192893401,
      "grad_norm": 1.5390625,
      "learning_rate": 2.4739847320727814e-05,
      "loss": 1.7614,
      "step": 52274
    },
    {
      "epoch": 4.497547965241332,
      "grad_norm": 1.3203125,
      "learning_rate": 2.473144747498518e-05,
      "loss": 1.1864,
      "step": 52275
    },
    {
      "epoch": 4.497634001548653,
      "grad_norm": 0.8984375,
      "learning_rate": 2.4723049019310716e-05,
      "loss": 0.8023,
      "step": 52276
    },
    {
      "epoch": 4.497720037855975,
      "grad_norm": 9.5625,
      "learning_rate": 2.4714651953728793e-05,
      "loss": 1.0603,
      "step": 52277
    },
    {
      "epoch": 4.497806074163297,
      "grad_norm": 2.703125,
      "learning_rate": 2.470625627826417e-05,
      "loss": 1.0955,
      "step": 52278
    },
    {
      "epoch": 4.497892110470619,
      "grad_norm": 1.125,
      "learning_rate": 2.4697861992941216e-05,
      "loss": 0.756,
      "step": 52279
    },
    {
      "epoch": 4.497978146777941,
      "grad_norm": 0.828125,
      "learning_rate": 2.4689469097784577e-05,
      "loss": 0.9685,
      "step": 52280
    },
    {
      "epoch": 4.498064183085262,
      "grad_norm": 1344.0,
      "learning_rate": 2.4681077592818902e-05,
      "loss": 1.3733,
      "step": 52281
    },
    {
      "epoch": 4.498150219392584,
      "grad_norm": 1.296875,
      "learning_rate": 2.4672687478068556e-05,
      "loss": 1.5028,
      "step": 52282
    },
    {
      "epoch": 4.498236255699905,
      "grad_norm": 82.0,
      "learning_rate": 2.4664298753558133e-05,
      "loss": 1.0925,
      "step": 52283
    },
    {
      "epoch": 4.498322292007227,
      "grad_norm": 0.82421875,
      "learning_rate": 2.4655911419312228e-05,
      "loss": 0.4573,
      "step": 52284
    },
    {
      "epoch": 4.498408328314548,
      "grad_norm": 1.34375,
      "learning_rate": 2.4647525475355316e-05,
      "loss": 2.3215,
      "step": 52285
    },
    {
      "epoch": 4.4984943646218705,
      "grad_norm": 1.265625,
      "learning_rate": 2.4639140921711934e-05,
      "loss": 1.0184,
      "step": 52286
    },
    {
      "epoch": 4.4985804009291925,
      "grad_norm": 1.0703125,
      "learning_rate": 2.4630757758406674e-05,
      "loss": 1.7017,
      "step": 52287
    },
    {
      "epoch": 4.498666437236514,
      "grad_norm": 1.0703125,
      "learning_rate": 2.4622375985463906e-05,
      "loss": 1.7684,
      "step": 52288
    },
    {
      "epoch": 4.498752473543836,
      "grad_norm": 1.15625,
      "learning_rate": 2.461399560290828e-05,
      "loss": 1.369,
      "step": 52289
    },
    {
      "epoch": 4.498838509851157,
      "grad_norm": 1.3125,
      "learning_rate": 2.460561661076427e-05,
      "loss": 0.7762,
      "step": 52290
    },
    {
      "epoch": 4.498924546158479,
      "grad_norm": 1.2265625,
      "learning_rate": 2.4597239009056306e-05,
      "loss": 1.6031,
      "step": 52291
    },
    {
      "epoch": 4.4990105824658,
      "grad_norm": 1.390625,
      "learning_rate": 2.458886279780903e-05,
      "loss": 0.8294,
      "step": 52292
    },
    {
      "epoch": 4.499096618773122,
      "grad_norm": 1.21875,
      "learning_rate": 2.4580487977046817e-05,
      "loss": 1.6599,
      "step": 52293
    },
    {
      "epoch": 4.499182655080444,
      "grad_norm": 1.0390625,
      "learning_rate": 2.45721145467942e-05,
      "loss": 0.9192,
      "step": 52294
    },
    {
      "epoch": 4.4992686913877655,
      "grad_norm": 1.015625,
      "learning_rate": 2.4563742507075716e-05,
      "loss": 0.872,
      "step": 52295
    },
    {
      "epoch": 4.499354727695088,
      "grad_norm": 0.96484375,
      "learning_rate": 2.455537185791579e-05,
      "loss": 1.1359,
      "step": 52296
    },
    {
      "epoch": 4.499440764002409,
      "grad_norm": 1.453125,
      "learning_rate": 2.454700259933901e-05,
      "loss": 1.5212,
      "step": 52297
    },
    {
      "epoch": 4.499526800309731,
      "grad_norm": 1.046875,
      "learning_rate": 2.4538634731369692e-05,
      "loss": 1.7116,
      "step": 52298
    },
    {
      "epoch": 4.499612836617052,
      "grad_norm": 1.046875,
      "learning_rate": 2.453026825403243e-05,
      "loss": 0.5331,
      "step": 52299
    },
    {
      "epoch": 4.499698872924374,
      "grad_norm": 0.67578125,
      "learning_rate": 2.45219031673517e-05,
      "loss": 0.5357,
      "step": 52300
    },
    {
      "epoch": 4.499784909231696,
      "grad_norm": 0.94921875,
      "learning_rate": 2.4513539471351875e-05,
      "loss": 0.8204,
      "step": 52301
    },
    {
      "epoch": 4.499870945539017,
      "grad_norm": 123.0,
      "learning_rate": 2.4505177166057436e-05,
      "loss": 0.6719,
      "step": 52302
    },
    {
      "epoch": 4.499956981846339,
      "grad_norm": 1.15625,
      "learning_rate": 2.4496816251492972e-05,
      "loss": 1.1717,
      "step": 52303
    },
    {
      "epoch": 4.500043018153661,
      "grad_norm": 2256.0,
      "learning_rate": 2.4488456727682794e-05,
      "loss": 1.6925,
      "step": 52304
    },
    {
      "epoch": 4.500129054460983,
      "grad_norm": 1.4140625,
      "learning_rate": 2.4480098594651333e-05,
      "loss": 1.1349,
      "step": 52305
    },
    {
      "epoch": 4.500215090768304,
      "grad_norm": 1.1640625,
      "learning_rate": 2.4471741852423235e-05,
      "loss": 1.9715,
      "step": 52306
    },
    {
      "epoch": 4.500301127075626,
      "grad_norm": 0.90625,
      "learning_rate": 2.44633865010227e-05,
      "loss": 0.9486,
      "step": 52307
    },
    {
      "epoch": 4.500387163382948,
      "grad_norm": 302.0,
      "learning_rate": 2.4455032540474376e-05,
      "loss": 1.0678,
      "step": 52308
    },
    {
      "epoch": 4.500387163382948,
      "eval_loss": 1.0676006078720093,
      "eval_runtime": 370.2563,
      "eval_samples_per_second": 12.556,
      "eval_steps_per_second": 1.572,
      "step": 52308
    },
    {
      "epoch": 4.500473199690269,
      "grad_norm": 0.875,
      "learning_rate": 2.4446679970802466e-05,
      "loss": 0.9493,
      "step": 52309
    },
    {
      "epoch": 4.500559235997591,
      "grad_norm": 0.75,
      "learning_rate": 2.443832879203156e-05,
      "loss": 0.9357,
      "step": 52310
    },
    {
      "epoch": 4.500645272304912,
      "grad_norm": 2.5,
      "learning_rate": 2.4429979004186143e-05,
      "loss": 0.8032,
      "step": 52311
    },
    {
      "epoch": 4.5007313086122345,
      "grad_norm": 1.0625,
      "learning_rate": 2.442163060729047e-05,
      "loss": 0.8595,
      "step": 52312
    },
    {
      "epoch": 4.500817344919556,
      "grad_norm": 1.0078125,
      "learning_rate": 2.4413283601369017e-05,
      "loss": 0.7223,
      "step": 52313
    },
    {
      "epoch": 4.500903381226878,
      "grad_norm": 1.0625,
      "learning_rate": 2.4404937986446275e-05,
      "loss": 0.604,
      "step": 52314
    },
    {
      "epoch": 4.5009894175342,
      "grad_norm": 10.0,
      "learning_rate": 2.4396593762546548e-05,
      "loss": 0.7985,
      "step": 52315
    },
    {
      "epoch": 4.501075453841521,
      "grad_norm": 0.8515625,
      "learning_rate": 2.438825092969421e-05,
      "loss": 0.3996,
      "step": 52316
    },
    {
      "epoch": 4.501161490148843,
      "grad_norm": 1.375,
      "learning_rate": 2.437990948791391e-05,
      "loss": 0.9564,
      "step": 52317
    },
    {
      "epoch": 4.501247526456164,
      "grad_norm": 1.1953125,
      "learning_rate": 2.4371569437229735e-05,
      "loss": 1.3525,
      "step": 52318
    },
    {
      "epoch": 4.501333562763486,
      "grad_norm": 1.2890625,
      "learning_rate": 2.4363230777666278e-05,
      "loss": 1.5286,
      "step": 52319
    },
    {
      "epoch": 4.5014195990708075,
      "grad_norm": 1.09375,
      "learning_rate": 2.4354893509247857e-05,
      "loss": 1.0903,
      "step": 52320
    },
    {
      "epoch": 4.5015056353781295,
      "grad_norm": 1.1875,
      "learning_rate": 2.4346557631998835e-05,
      "loss": 1.1048,
      "step": 52321
    },
    {
      "epoch": 4.501591671685452,
      "grad_norm": 1.3125,
      "learning_rate": 2.4338223145943638e-05,
      "loss": 1.4589,
      "step": 52322
    },
    {
      "epoch": 4.501677707992773,
      "grad_norm": 1.203125,
      "learning_rate": 2.432989005110664e-05,
      "loss": 1.8711,
      "step": 52323
    },
    {
      "epoch": 4.501763744300095,
      "grad_norm": 5.75,
      "learning_rate": 2.4321558347512153e-05,
      "loss": 0.9596,
      "step": 52324
    },
    {
      "epoch": 4.501849780607416,
      "grad_norm": 1.1640625,
      "learning_rate": 2.4313228035184654e-05,
      "loss": 1.1403,
      "step": 52325
    },
    {
      "epoch": 4.501935816914738,
      "grad_norm": 1.2890625,
      "learning_rate": 2.4304899114148405e-05,
      "loss": 0.7854,
      "step": 52326
    },
    {
      "epoch": 4.502021853222059,
      "grad_norm": 0.86328125,
      "learning_rate": 2.429657158442783e-05,
      "loss": 0.9209,
      "step": 52327
    },
    {
      "epoch": 4.502107889529381,
      "grad_norm": 1.2265625,
      "learning_rate": 2.4288245446047298e-05,
      "loss": 0.6675,
      "step": 52328
    },
    {
      "epoch": 4.502193925836703,
      "grad_norm": 1.1171875,
      "learning_rate": 2.4279920699031066e-05,
      "loss": 1.2882,
      "step": 52329
    },
    {
      "epoch": 4.502279962144025,
      "grad_norm": 1.09375,
      "learning_rate": 2.427159734340356e-05,
      "loss": 1.3097,
      "step": 52330
    },
    {
      "epoch": 4.502365998451347,
      "grad_norm": 2.453125,
      "learning_rate": 2.4263275379189153e-05,
      "loss": 1.2263,
      "step": 52331
    },
    {
      "epoch": 4.502452034758668,
      "grad_norm": 1.140625,
      "learning_rate": 2.425495480641199e-05,
      "loss": 0.9192,
      "step": 52332
    },
    {
      "epoch": 4.50253807106599,
      "grad_norm": 1.40625,
      "learning_rate": 2.4246635625096713e-05,
      "loss": 1.1328,
      "step": 52333
    },
    {
      "epoch": 4.502624107373311,
      "grad_norm": 1.6953125,
      "learning_rate": 2.423831783526742e-05,
      "loss": 1.0697,
      "step": 52334
    },
    {
      "epoch": 4.502710143680633,
      "grad_norm": 1.046875,
      "learning_rate": 2.423000143694859e-05,
      "loss": 1.1306,
      "step": 52335
    },
    {
      "epoch": 4.502796179987955,
      "grad_norm": 1.125,
      "learning_rate": 2.4221686430164426e-05,
      "loss": 0.8833,
      "step": 52336
    },
    {
      "epoch": 4.502882216295276,
      "grad_norm": 1.2734375,
      "learning_rate": 2.4213372814939293e-05,
      "loss": 0.7194,
      "step": 52337
    },
    {
      "epoch": 4.5029682526025985,
      "grad_norm": 1.1328125,
      "learning_rate": 2.4205060591297566e-05,
      "loss": 1.4178,
      "step": 52338
    },
    {
      "epoch": 4.50305428890992,
      "grad_norm": 0.90234375,
      "learning_rate": 2.4196749759263446e-05,
      "loss": 1.426,
      "step": 52339
    },
    {
      "epoch": 4.503140325217242,
      "grad_norm": 0.98828125,
      "learning_rate": 2.418844031886125e-05,
      "loss": 1.4769,
      "step": 52340
    },
    {
      "epoch": 4.503226361524563,
      "grad_norm": 8.1875,
      "learning_rate": 2.4180132270115453e-05,
      "loss": 1.1077,
      "step": 52341
    },
    {
      "epoch": 4.503312397831885,
      "grad_norm": 33.25,
      "learning_rate": 2.417182561305009e-05,
      "loss": 1.3598,
      "step": 52342
    },
    {
      "epoch": 4.503398434139207,
      "grad_norm": 1.0078125,
      "learning_rate": 2.416352034768965e-05,
      "loss": 0.8825,
      "step": 52343
    },
    {
      "epoch": 4.503484470446528,
      "grad_norm": 1.03125,
      "learning_rate": 2.4155216474058437e-05,
      "loss": 1.1644,
      "step": 52344
    },
    {
      "epoch": 4.50357050675385,
      "grad_norm": 1.4296875,
      "learning_rate": 2.4146913992180553e-05,
      "loss": 1.1879,
      "step": 52345
    },
    {
      "epoch": 4.5036565430611715,
      "grad_norm": 1.140625,
      "learning_rate": 2.4138612902080415e-05,
      "loss": 1.4893,
      "step": 52346
    },
    {
      "epoch": 4.5037425793684935,
      "grad_norm": 1.171875,
      "learning_rate": 2.4130313203782396e-05,
      "loss": 1.0777,
      "step": 52347
    },
    {
      "epoch": 4.503828615675815,
      "grad_norm": 0.8359375,
      "learning_rate": 2.4122014897310528e-05,
      "loss": 0.8252,
      "step": 52348
    },
    {
      "epoch": 4.503914651983137,
      "grad_norm": 0.953125,
      "learning_rate": 2.4113717982689355e-05,
      "loss": 0.7571,
      "step": 52349
    },
    {
      "epoch": 4.504000688290459,
      "grad_norm": 1.25,
      "learning_rate": 2.410542245994285e-05,
      "loss": 0.8,
      "step": 52350
    },
    {
      "epoch": 4.50408672459778,
      "grad_norm": 0.875,
      "learning_rate": 2.4097128329095495e-05,
      "loss": 0.5602,
      "step": 52351
    },
    {
      "epoch": 4.504172760905102,
      "grad_norm": 1.6171875,
      "learning_rate": 2.4088835590171498e-05,
      "loss": 0.8613,
      "step": 52352
    },
    {
      "epoch": 4.504258797212423,
      "grad_norm": 1.4921875,
      "learning_rate": 2.4080544243195056e-05,
      "loss": 1.0268,
      "step": 52353
    },
    {
      "epoch": 4.504344833519745,
      "grad_norm": 1.21875,
      "learning_rate": 2.4072254288190486e-05,
      "loss": 1.1797,
      "step": 52354
    },
    {
      "epoch": 4.5044308698270665,
      "grad_norm": 0.9296875,
      "learning_rate": 2.4063965725182046e-05,
      "loss": 0.5305,
      "step": 52355
    },
    {
      "epoch": 4.504516906134389,
      "grad_norm": 1.1796875,
      "learning_rate": 2.405567855419388e-05,
      "loss": 1.8444,
      "step": 52356
    },
    {
      "epoch": 4.504602942441711,
      "grad_norm": 1.0234375,
      "learning_rate": 2.404739277525031e-05,
      "loss": 1.9006,
      "step": 52357
    },
    {
      "epoch": 4.504688978749032,
      "grad_norm": 1.3125,
      "learning_rate": 2.4039108388375584e-05,
      "loss": 1.3729,
      "step": 52358
    },
    {
      "epoch": 4.504775015056354,
      "grad_norm": 1.3359375,
      "learning_rate": 2.4030825393593804e-05,
      "loss": 0.6558,
      "step": 52359
    },
    {
      "epoch": 4.504861051363675,
      "grad_norm": 1.0625,
      "learning_rate": 2.4022543790929385e-05,
      "loss": 1.0869,
      "step": 52360
    },
    {
      "epoch": 4.504947087670997,
      "grad_norm": 1.1484375,
      "learning_rate": 2.4014263580406427e-05,
      "loss": 0.663,
      "step": 52361
    },
    {
      "epoch": 4.505033123978319,
      "grad_norm": 1.078125,
      "learning_rate": 2.4005984762049125e-05,
      "loss": 1.2461,
      "step": 52362
    },
    {
      "epoch": 4.50511916028564,
      "grad_norm": 1.578125,
      "learning_rate": 2.3997707335881803e-05,
      "loss": 1.7842,
      "step": 52363
    },
    {
      "epoch": 4.5052051965929625,
      "grad_norm": 0.90234375,
      "learning_rate": 2.39894313019286e-05,
      "loss": 1.194,
      "step": 52364
    },
    {
      "epoch": 4.505291232900284,
      "grad_norm": 1.2578125,
      "learning_rate": 2.3981156660213664e-05,
      "loss": 1.0339,
      "step": 52365
    },
    {
      "epoch": 4.505377269207606,
      "grad_norm": 1.28125,
      "learning_rate": 2.397288341076137e-05,
      "loss": 0.9081,
      "step": 52366
    },
    {
      "epoch": 4.505463305514928,
      "grad_norm": 1.1953125,
      "learning_rate": 2.396461155359575e-05,
      "loss": 1.1174,
      "step": 52367
    },
    {
      "epoch": 4.505549341822249,
      "grad_norm": 1.84375,
      "learning_rate": 2.3956341088741008e-05,
      "loss": 1.24,
      "step": 52368
    },
    {
      "epoch": 4.505635378129571,
      "grad_norm": 1.328125,
      "learning_rate": 2.3948072016221456e-05,
      "loss": 1.103,
      "step": 52369
    },
    {
      "epoch": 4.505721414436892,
      "grad_norm": 1.046875,
      "learning_rate": 2.3939804336061133e-05,
      "loss": 1.4447,
      "step": 52370
    },
    {
      "epoch": 4.505807450744214,
      "grad_norm": 222.0,
      "learning_rate": 2.393153804828441e-05,
      "loss": 1.0815,
      "step": 52371
    },
    {
      "epoch": 4.5058934870515355,
      "grad_norm": 1.4140625,
      "learning_rate": 2.392327315291526e-05,
      "loss": 1.908,
      "step": 52372
    },
    {
      "epoch": 4.5059795233588575,
      "grad_norm": 5.40625,
      "learning_rate": 2.3915009649977947e-05,
      "loss": 0.903,
      "step": 52373
    },
    {
      "epoch": 4.50606555966618,
      "grad_norm": 1.609375,
      "learning_rate": 2.3906747539496732e-05,
      "loss": 1.0833,
      "step": 52374
    },
    {
      "epoch": 4.506151595973501,
      "grad_norm": 5.6875,
      "learning_rate": 2.389848682149559e-05,
      "loss": 0.7231,
      "step": 52375
    },
    {
      "epoch": 4.506237632280823,
      "grad_norm": 0.89453125,
      "learning_rate": 2.3890227495998785e-05,
      "loss": 1.3982,
      "step": 52376
    },
    {
      "epoch": 4.506323668588144,
      "grad_norm": 1.5390625,
      "learning_rate": 2.3881969563030516e-05,
      "loss": 1.6284,
      "step": 52377
    },
    {
      "epoch": 4.506409704895466,
      "grad_norm": 1.25,
      "learning_rate": 2.3873713022614818e-05,
      "loss": 1.4363,
      "step": 52378
    },
    {
      "epoch": 4.506495741202787,
      "grad_norm": 0.828125,
      "learning_rate": 2.386545787477601e-05,
      "loss": 0.7339,
      "step": 52379
    },
    {
      "epoch": 4.506581777510109,
      "grad_norm": 1.25,
      "learning_rate": 2.3857204119538013e-05,
      "loss": 1.1744,
      "step": 52380
    },
    {
      "epoch": 4.506667813817431,
      "grad_norm": 2.0,
      "learning_rate": 2.3848951756925143e-05,
      "loss": 1.7522,
      "step": 52381
    },
    {
      "epoch": 4.506753850124753,
      "grad_norm": 0.796875,
      "learning_rate": 2.3840700786961544e-05,
      "loss": 0.627,
      "step": 52382
    },
    {
      "epoch": 4.506839886432075,
      "grad_norm": 1.828125,
      "learning_rate": 2.38324512096712e-05,
      "loss": 1.1928,
      "step": 52383
    },
    {
      "epoch": 4.506925922739396,
      "grad_norm": 0.890625,
      "learning_rate": 2.3824203025078317e-05,
      "loss": 0.8636,
      "step": 52384
    },
    {
      "epoch": 4.507011959046718,
      "grad_norm": 1.5,
      "learning_rate": 2.3815956233207147e-05,
      "loss": 1.2339,
      "step": 52385
    },
    {
      "epoch": 4.507097995354039,
      "grad_norm": 3.5625,
      "learning_rate": 2.3807710834081563e-05,
      "loss": 1.1681,
      "step": 52386
    },
    {
      "epoch": 4.507184031661361,
      "grad_norm": 1.0703125,
      "learning_rate": 2.3799466827725826e-05,
      "loss": 1.4176,
      "step": 52387
    },
    {
      "epoch": 4.507270067968683,
      "grad_norm": 0.9609375,
      "learning_rate": 2.379122421416413e-05,
      "loss": 1.1069,
      "step": 52388
    },
    {
      "epoch": 4.507356104276004,
      "grad_norm": 4.25,
      "learning_rate": 2.378298299342041e-05,
      "loss": 0.7558,
      "step": 52389
    },
    {
      "epoch": 4.5074421405833265,
      "grad_norm": 1.171875,
      "learning_rate": 2.377474316551892e-05,
      "loss": 1.1727,
      "step": 52390
    },
    {
      "epoch": 4.507528176890648,
      "grad_norm": 4.09375,
      "learning_rate": 2.3766504730483583e-05,
      "loss": 1.2052,
      "step": 52391
    },
    {
      "epoch": 4.50761421319797,
      "grad_norm": 9.375,
      "learning_rate": 2.375826768833861e-05,
      "loss": 1.3465,
      "step": 52392
    },
    {
      "epoch": 4.507700249505291,
      "grad_norm": 1032.0,
      "learning_rate": 2.3750032039108137e-05,
      "loss": 1.0132,
      "step": 52393
    },
    {
      "epoch": 4.507786285812613,
      "grad_norm": 1.6171875,
      "learning_rate": 2.3741797782816156e-05,
      "loss": 1.3622,
      "step": 52394
    },
    {
      "epoch": 4.507872322119935,
      "grad_norm": 1.3671875,
      "learning_rate": 2.3733564919486805e-05,
      "loss": 1.1301,
      "step": 52395
    },
    {
      "epoch": 4.507958358427256,
      "grad_norm": 1.265625,
      "learning_rate": 2.3725333449144182e-05,
      "loss": 1.6102,
      "step": 52396
    },
    {
      "epoch": 4.508044394734578,
      "grad_norm": 0.8671875,
      "learning_rate": 2.3717103371812266e-05,
      "loss": 1.0879,
      "step": 52397
    },
    {
      "epoch": 4.5081304310418995,
      "grad_norm": 1.1796875,
      "learning_rate": 2.37088746875152e-05,
      "loss": 1.5081,
      "step": 52398
    },
    {
      "epoch": 4.5082164673492215,
      "grad_norm": 1.109375,
      "learning_rate": 2.3700647396277087e-05,
      "loss": 0.7118,
      "step": 52399
    },
    {
      "epoch": 4.508302503656543,
      "grad_norm": 155.0,
      "learning_rate": 2.3692421498121897e-05,
      "loss": 0.6804,
      "step": 52400
    },
    {
      "epoch": 4.508388539963865,
      "grad_norm": 1.0078125,
      "learning_rate": 2.3684196993073782e-05,
      "loss": 1.1105,
      "step": 52401
    },
    {
      "epoch": 4.508474576271187,
      "grad_norm": 1.3671875,
      "learning_rate": 2.367597388115672e-05,
      "loss": 0.7208,
      "step": 52402
    },
    {
      "epoch": 4.508560612578508,
      "grad_norm": 1.2890625,
      "learning_rate": 2.3667752162394752e-05,
      "loss": 1.4661,
      "step": 52403
    },
    {
      "epoch": 4.50864664888583,
      "grad_norm": 1.0390625,
      "learning_rate": 2.3659531836812076e-05,
      "loss": 1.2904,
      "step": 52404
    },
    {
      "epoch": 4.508732685193151,
      "grad_norm": 0.8515625,
      "learning_rate": 2.3651312904432566e-05,
      "loss": 0.8212,
      "step": 52405
    },
    {
      "epoch": 4.508818721500473,
      "grad_norm": 1.578125,
      "learning_rate": 2.3643095365280255e-05,
      "loss": 1.2532,
      "step": 52406
    },
    {
      "epoch": 4.5089047578077945,
      "grad_norm": 0.83203125,
      "learning_rate": 2.363487921937929e-05,
      "loss": 0.4295,
      "step": 52407
    },
    {
      "epoch": 4.508990794115117,
      "grad_norm": 1.109375,
      "learning_rate": 2.3626664466753657e-05,
      "loss": 0.7009,
      "step": 52408
    },
    {
      "epoch": 4.509076830422439,
      "grad_norm": 48.0,
      "learning_rate": 2.361845110742744e-05,
      "loss": 1.7299,
      "step": 52409
    },
    {
      "epoch": 4.50916286672976,
      "grad_norm": 0.7109375,
      "learning_rate": 2.361023914142457e-05,
      "loss": 0.5261,
      "step": 52410
    },
    {
      "epoch": 4.509248903037082,
      "grad_norm": 2.875,
      "learning_rate": 2.360202856876903e-05,
      "loss": 1.3452,
      "step": 52411
    },
    {
      "epoch": 4.509334939344403,
      "grad_norm": 1.03125,
      "learning_rate": 2.3593819389485017e-05,
      "loss": 1.1457,
      "step": 52412
    },
    {
      "epoch": 4.509420975651725,
      "grad_norm": 1.21875,
      "learning_rate": 2.3585611603596346e-05,
      "loss": 1.3859,
      "step": 52413
    },
    {
      "epoch": 4.509507011959046,
      "grad_norm": 1.0546875,
      "learning_rate": 2.3577405211127112e-05,
      "loss": 1.3295,
      "step": 52414
    },
    {
      "epoch": 4.509593048266368,
      "grad_norm": 94.0,
      "learning_rate": 2.356920021210135e-05,
      "loss": 0.7586,
      "step": 52415
    },
    {
      "epoch": 4.5096790845736905,
      "grad_norm": 1.203125,
      "learning_rate": 2.3560996606542983e-05,
      "loss": 0.6749,
      "step": 52416
    },
    {
      "epoch": 4.509765120881012,
      "grad_norm": 1.265625,
      "learning_rate": 2.355279439447605e-05,
      "loss": 1.3283,
      "step": 52417
    },
    {
      "epoch": 4.509851157188334,
      "grad_norm": 0.85546875,
      "learning_rate": 2.3544593575924534e-05,
      "loss": 0.5087,
      "step": 52418
    },
    {
      "epoch": 4.509937193495655,
      "grad_norm": 1.0546875,
      "learning_rate": 2.3536394150912412e-05,
      "loss": 1.5295,
      "step": 52419
    },
    {
      "epoch": 4.510023229802977,
      "grad_norm": 0.71875,
      "learning_rate": 2.352819611946372e-05,
      "loss": 0.6548,
      "step": 52420
    },
    {
      "epoch": 4.510109266110298,
      "grad_norm": 4.6875,
      "learning_rate": 2.3519999481602326e-05,
      "loss": 1.4493,
      "step": 52421
    },
    {
      "epoch": 4.51019530241762,
      "grad_norm": 1.21875,
      "learning_rate": 2.351180423735222e-05,
      "loss": 0.6089,
      "step": 52422
    },
    {
      "epoch": 4.510281338724942,
      "grad_norm": 48.25,
      "learning_rate": 2.350361038673754e-05,
      "loss": 1.0925,
      "step": 52423
    },
    {
      "epoch": 4.5103673750322635,
      "grad_norm": 1.46875,
      "learning_rate": 2.3495417929781993e-05,
      "loss": 1.5012,
      "step": 52424
    },
    {
      "epoch": 4.5104534113395855,
      "grad_norm": 0.83984375,
      "learning_rate": 2.3487226866509727e-05,
      "loss": 1.2739,
      "step": 52425
    },
    {
      "epoch": 4.510539447646907,
      "grad_norm": 1.3125,
      "learning_rate": 2.3479037196944665e-05,
      "loss": 1.8905,
      "step": 52426
    },
    {
      "epoch": 4.510625483954229,
      "grad_norm": 27.0,
      "learning_rate": 2.3470848921110732e-05,
      "loss": 0.3841,
      "step": 52427
    },
    {
      "epoch": 4.51071152026155,
      "grad_norm": 1.0390625,
      "learning_rate": 2.346266203903186e-05,
      "loss": 2.0088,
      "step": 52428
    },
    {
      "epoch": 4.510797556568872,
      "grad_norm": 1.7265625,
      "learning_rate": 2.3454476550732072e-05,
      "loss": 0.8243,
      "step": 52429
    },
    {
      "epoch": 4.510883592876194,
      "grad_norm": 1.28125,
      "learning_rate": 2.3446292456235196e-05,
      "loss": 1.0193,
      "step": 52430
    },
    {
      "epoch": 4.510969629183515,
      "grad_norm": 1.3125,
      "learning_rate": 2.3438109755565318e-05,
      "loss": 1.033,
      "step": 52431
    },
    {
      "epoch": 4.511055665490837,
      "grad_norm": 1.078125,
      "learning_rate": 2.3429928448746198e-05,
      "loss": 0.3753,
      "step": 52432
    },
    {
      "epoch": 4.5111417017981585,
      "grad_norm": 1.0234375,
      "learning_rate": 2.342174853580187e-05,
      "loss": 0.8318,
      "step": 52433
    },
    {
      "epoch": 4.511227738105481,
      "grad_norm": 0.91015625,
      "learning_rate": 2.3413570016756257e-05,
      "loss": 0.6502,
      "step": 52434
    },
    {
      "epoch": 4.511313774412802,
      "grad_norm": 1.2109375,
      "learning_rate": 2.3405392891633236e-05,
      "loss": 0.7507,
      "step": 52435
    },
    {
      "epoch": 4.511399810720124,
      "grad_norm": 1.2421875,
      "learning_rate": 2.339721716045673e-05,
      "loss": 0.8224,
      "step": 52436
    },
    {
      "epoch": 4.511485847027446,
      "grad_norm": 1.125,
      "learning_rate": 2.3389042823250773e-05,
      "loss": 0.5034,
      "step": 52437
    },
    {
      "epoch": 4.511571883334767,
      "grad_norm": 1.1953125,
      "learning_rate": 2.3380869880039067e-05,
      "loss": 1.2627,
      "step": 52438
    },
    {
      "epoch": 4.511657919642089,
      "grad_norm": 1.5703125,
      "learning_rate": 2.33726983308456e-05,
      "loss": 1.9805,
      "step": 52439
    },
    {
      "epoch": 4.51174395594941,
      "grad_norm": 0.9609375,
      "learning_rate": 2.33645281756944e-05,
      "loss": 1.4793,
      "step": 52440
    },
    {
      "epoch": 4.511829992256732,
      "grad_norm": 1.0390625,
      "learning_rate": 2.3356359414609173e-05,
      "loss": 0.7539,
      "step": 52441
    },
    {
      "epoch": 4.511916028564054,
      "grad_norm": 1.2421875,
      "learning_rate": 2.3348192047613903e-05,
      "loss": 1.8755,
      "step": 52442
    },
    {
      "epoch": 4.512002064871376,
      "grad_norm": 1.1796875,
      "learning_rate": 2.3340026074732458e-05,
      "loss": 1.4358,
      "step": 52443
    },
    {
      "epoch": 4.512088101178698,
      "grad_norm": 0.91796875,
      "learning_rate": 2.3331861495988703e-05,
      "loss": 1.3724,
      "step": 52444
    },
    {
      "epoch": 4.512174137486019,
      "grad_norm": 1.3359375,
      "learning_rate": 2.3323698311406682e-05,
      "loss": 1.786,
      "step": 52445
    },
    {
      "epoch": 4.512260173793341,
      "grad_norm": 1.15625,
      "learning_rate": 2.331553652101004e-05,
      "loss": 1.0352,
      "step": 52446
    },
    {
      "epoch": 4.512346210100662,
      "grad_norm": 1.2109375,
      "learning_rate": 2.3307376124822754e-05,
      "loss": 1.4969,
      "step": 52447
    },
    {
      "epoch": 4.512432246407984,
      "grad_norm": 1.2265625,
      "learning_rate": 2.3299217122868697e-05,
      "loss": 1.6398,
      "step": 52448
    },
    {
      "epoch": 4.512518282715305,
      "grad_norm": 0.96484375,
      "learning_rate": 2.3291059515171685e-05,
      "loss": 1.071,
      "step": 52449
    },
    {
      "epoch": 4.5126043190226275,
      "grad_norm": 1.0625,
      "learning_rate": 2.3282903301755642e-05,
      "loss": 1.2536,
      "step": 52450
    },
    {
      "epoch": 4.5126903553299496,
      "grad_norm": 1.1640625,
      "learning_rate": 2.3274748482644324e-05,
      "loss": 1.0232,
      "step": 52451
    },
    {
      "epoch": 4.512776391637271,
      "grad_norm": 3.125,
      "learning_rate": 2.3266595057861717e-05,
      "loss": 0.7155,
      "step": 52452
    },
    {
      "epoch": 4.512862427944593,
      "grad_norm": 28.875,
      "learning_rate": 2.325844302743163e-05,
      "loss": 1.1871,
      "step": 52453
    },
    {
      "epoch": 4.512948464251914,
      "grad_norm": 1.328125,
      "learning_rate": 2.3250292391377768e-05,
      "loss": 0.939,
      "step": 52454
    },
    {
      "epoch": 4.513034500559236,
      "grad_norm": 0.83984375,
      "learning_rate": 2.3242143149724115e-05,
      "loss": 1.0016,
      "step": 52455
    },
    {
      "epoch": 4.513120536866557,
      "grad_norm": 1.0390625,
      "learning_rate": 2.323399530249454e-05,
      "loss": 1.4677,
      "step": 52456
    },
    {
      "epoch": 4.513206573173879,
      "grad_norm": 1.4765625,
      "learning_rate": 2.3225848849712738e-05,
      "loss": 1.6189,
      "step": 52457
    },
    {
      "epoch": 4.513292609481201,
      "grad_norm": 0.8046875,
      "learning_rate": 2.321770379140259e-05,
      "loss": 0.8249,
      "step": 52458
    },
    {
      "epoch": 4.5133786457885225,
      "grad_norm": 1.046875,
      "learning_rate": 2.3209560127588013e-05,
      "loss": 0.8496,
      "step": 52459
    },
    {
      "epoch": 4.513464682095845,
      "grad_norm": 2.828125,
      "learning_rate": 2.3201417858292662e-05,
      "loss": 1.422,
      "step": 52460
    },
    {
      "epoch": 4.513550718403166,
      "grad_norm": 1.4296875,
      "learning_rate": 2.319327698354051e-05,
      "loss": 1.2228,
      "step": 52461
    },
    {
      "epoch": 4.513636754710488,
      "grad_norm": 1.265625,
      "learning_rate": 2.3185137503355215e-05,
      "loss": 1.6213,
      "step": 52462
    },
    {
      "epoch": 4.513722791017809,
      "grad_norm": 3.40625,
      "learning_rate": 2.3176999417760637e-05,
      "loss": 1.5368,
      "step": 52463
    },
    {
      "epoch": 4.513808827325131,
      "grad_norm": 3.046875,
      "learning_rate": 2.3168862726780703e-05,
      "loss": 1.6224,
      "step": 52464
    },
    {
      "epoch": 4.513894863632453,
      "grad_norm": 0.8203125,
      "learning_rate": 2.316072743043901e-05,
      "loss": 0.8087,
      "step": 52465
    },
    {
      "epoch": 4.513980899939774,
      "grad_norm": 1.3203125,
      "learning_rate": 2.3152593528759426e-05,
      "loss": 1.4287,
      "step": 52466
    },
    {
      "epoch": 4.514066936247096,
      "grad_norm": 1.359375,
      "learning_rate": 2.3144461021765872e-05,
      "loss": 1.6797,
      "step": 52467
    },
    {
      "epoch": 4.514152972554418,
      "grad_norm": 1.3515625,
      "learning_rate": 2.3136329909481946e-05,
      "loss": 1.0959,
      "step": 52468
    },
    {
      "epoch": 4.51423900886174,
      "grad_norm": 1.1484375,
      "learning_rate": 2.3128200191931513e-05,
      "loss": 0.9633,
      "step": 52469
    },
    {
      "epoch": 4.514325045169061,
      "grad_norm": 1.015625,
      "learning_rate": 2.3120071869138447e-05,
      "loss": 1.141,
      "step": 52470
    },
    {
      "epoch": 4.514411081476383,
      "grad_norm": 1.25,
      "learning_rate": 2.311194494112634e-05,
      "loss": 1.7487,
      "step": 52471
    },
    {
      "epoch": 4.514497117783705,
      "grad_norm": 1.1796875,
      "learning_rate": 2.310381940791906e-05,
      "loss": 0.5905,
      "step": 52472
    },
    {
      "epoch": 4.514583154091026,
      "grad_norm": 1.1796875,
      "learning_rate": 2.309569526954036e-05,
      "loss": 1.0266,
      "step": 52473
    },
    {
      "epoch": 4.514669190398348,
      "grad_norm": 1.4140625,
      "learning_rate": 2.3087572526013954e-05,
      "loss": 0.558,
      "step": 52474
    },
    {
      "epoch": 4.514755226705669,
      "grad_norm": 4.5625,
      "learning_rate": 2.307945117736371e-05,
      "loss": 1.3275,
      "step": 52475
    },
    {
      "epoch": 4.5148412630129915,
      "grad_norm": 0.7578125,
      "learning_rate": 2.3071331223613268e-05,
      "loss": 0.4922,
      "step": 52476
    },
    {
      "epoch": 4.514927299320314,
      "grad_norm": 1.1484375,
      "learning_rate": 2.306321266478645e-05,
      "loss": 0.8017,
      "step": 52477
    },
    {
      "epoch": 4.515013335627635,
      "grad_norm": 1.078125,
      "learning_rate": 2.305509550090701e-05,
      "loss": 0.6828,
      "step": 52478
    },
    {
      "epoch": 4.515099371934957,
      "grad_norm": 6.875,
      "learning_rate": 2.3046979731998542e-05,
      "loss": 1.0771,
      "step": 52479
    },
    {
      "epoch": 4.515185408242278,
      "grad_norm": 1.1328125,
      "learning_rate": 2.303886535808497e-05,
      "loss": 1.5427,
      "step": 52480
    },
    {
      "epoch": 4.5152714445496,
      "grad_norm": 1.546875,
      "learning_rate": 2.3030752379189944e-05,
      "loss": 1.5675,
      "step": 52481
    },
    {
      "epoch": 4.515357480856921,
      "grad_norm": 1.1875,
      "learning_rate": 2.302264079533717e-05,
      "loss": 1.2719,
      "step": 52482
    },
    {
      "epoch": 4.515443517164243,
      "grad_norm": 1.6171875,
      "learning_rate": 2.3014530606550454e-05,
      "loss": 0.964,
      "step": 52483
    },
    {
      "epoch": 4.515529553471565,
      "grad_norm": 1.15625,
      "learning_rate": 2.300642181285345e-05,
      "loss": 1.1725,
      "step": 52484
    },
    {
      "epoch": 4.5156155897788866,
      "grad_norm": 96.0,
      "learning_rate": 2.299831441426986e-05,
      "loss": 1.6604,
      "step": 52485
    },
    {
      "epoch": 4.515701626086209,
      "grad_norm": 9.0625,
      "learning_rate": 2.2990208410823554e-05,
      "loss": 0.8305,
      "step": 52486
    },
    {
      "epoch": 4.51578766239353,
      "grad_norm": 1.21875,
      "learning_rate": 2.2982103802537957e-05,
      "loss": 0.9813,
      "step": 52487
    },
    {
      "epoch": 4.515873698700852,
      "grad_norm": 1.125,
      "learning_rate": 2.297400058943705e-05,
      "loss": 1.8578,
      "step": 52488
    },
    {
      "epoch": 4.515959735008174,
      "grad_norm": 1.5,
      "learning_rate": 2.296589877154437e-05,
      "loss": 0.7853,
      "step": 52489
    },
    {
      "epoch": 4.516045771315495,
      "grad_norm": 1.0859375,
      "learning_rate": 2.2957798348883618e-05,
      "loss": 1.2192,
      "step": 52490
    },
    {
      "epoch": 4.516131807622817,
      "grad_norm": 0.8359375,
      "learning_rate": 2.294969932147861e-05,
      "loss": 0.5488,
      "step": 52491
    },
    {
      "epoch": 4.516217843930138,
      "grad_norm": 0.84765625,
      "learning_rate": 2.2941601689352885e-05,
      "loss": 0.7092,
      "step": 52492
    },
    {
      "epoch": 4.51630388023746,
      "grad_norm": 33.25,
      "learning_rate": 2.2933505452530202e-05,
      "loss": 0.9342,
      "step": 52493
    },
    {
      "epoch": 4.516389916544782,
      "grad_norm": 105.5,
      "learning_rate": 2.2925410611034257e-05,
      "loss": 1.0962,
      "step": 52494
    },
    {
      "epoch": 4.516475952852104,
      "grad_norm": 35.25,
      "learning_rate": 2.2917317164888653e-05,
      "loss": 0.9659,
      "step": 52495
    },
    {
      "epoch": 4.516561989159426,
      "grad_norm": 2.71875,
      "learning_rate": 2.290922511411714e-05,
      "loss": 1.1562,
      "step": 52496
    },
    {
      "epoch": 4.516648025466747,
      "grad_norm": 1.015625,
      "learning_rate": 2.2901134458743367e-05,
      "loss": 1.0246,
      "step": 52497
    },
    {
      "epoch": 4.516734061774069,
      "grad_norm": 0.921875,
      "learning_rate": 2.2893045198790985e-05,
      "loss": 1.4652,
      "step": 52498
    },
    {
      "epoch": 4.51682009808139,
      "grad_norm": 1.359375,
      "learning_rate": 2.2884957334283587e-05,
      "loss": 1.5803,
      "step": 52499
    },
    {
      "epoch": 4.516906134388712,
      "grad_norm": 1.1328125,
      "learning_rate": 2.287687086524498e-05,
      "loss": 0.7914,
      "step": 52500
    },
    {
      "epoch": 4.516992170696033,
      "grad_norm": 7.0625,
      "learning_rate": 2.286878579169871e-05,
      "loss": 1.1172,
      "step": 52501
    },
    {
      "epoch": 4.5170782070033555,
      "grad_norm": 1.484375,
      "learning_rate": 2.2860702113668476e-05,
      "loss": 0.8009,
      "step": 52502
    },
    {
      "epoch": 4.517164243310678,
      "grad_norm": 1.1171875,
      "learning_rate": 2.2852619831177813e-05,
      "loss": 0.9734,
      "step": 52503
    },
    {
      "epoch": 4.517250279617999,
      "grad_norm": 2512.0,
      "learning_rate": 2.284453894425048e-05,
      "loss": 1.0824,
      "step": 52504
    },
    {
      "epoch": 4.517336315925321,
      "grad_norm": 0.98828125,
      "learning_rate": 2.2836459452910074e-05,
      "loss": 1.4436,
      "step": 52505
    },
    {
      "epoch": 4.517422352232642,
      "grad_norm": 1.03125,
      "learning_rate": 2.2828381357180185e-05,
      "loss": 1.3966,
      "step": 52506
    },
    {
      "epoch": 4.517508388539964,
      "grad_norm": 0.86328125,
      "learning_rate": 2.2820304657084457e-05,
      "loss": 0.6659,
      "step": 52507
    },
    {
      "epoch": 4.517594424847285,
      "grad_norm": 1.8125,
      "learning_rate": 2.281222935264665e-05,
      "loss": 1.0105,
      "step": 52508
    },
    {
      "epoch": 4.517680461154607,
      "grad_norm": 1.03125,
      "learning_rate": 2.280415544389014e-05,
      "loss": 1.0862,
      "step": 52509
    },
    {
      "epoch": 4.517766497461929,
      "grad_norm": 0.95703125,
      "learning_rate": 2.279608293083868e-05,
      "loss": 1.3975,
      "step": 52510
    },
    {
      "epoch": 4.5178525337692506,
      "grad_norm": 1.1875,
      "learning_rate": 2.2788011813515975e-05,
      "loss": 1.6954,
      "step": 52511
    },
    {
      "epoch": 4.517938570076573,
      "grad_norm": 1.5703125,
      "learning_rate": 2.2779942091945393e-05,
      "loss": 1.4865,
      "step": 52512
    },
    {
      "epoch": 4.518024606383894,
      "grad_norm": 1.046875,
      "learning_rate": 2.277187376615081e-05,
      "loss": 0.7252,
      "step": 52513
    },
    {
      "epoch": 4.518110642691216,
      "grad_norm": 1.125,
      "learning_rate": 2.2763806836155533e-05,
      "loss": 0.9253,
      "step": 52514
    },
    {
      "epoch": 4.518196678998537,
      "grad_norm": 1.0625,
      "learning_rate": 2.2755741301983324e-05,
      "loss": 1.5117,
      "step": 52515
    },
    {
      "epoch": 4.518282715305859,
      "grad_norm": 57.5,
      "learning_rate": 2.274767716365783e-05,
      "loss": 0.8257,
      "step": 52516
    },
    {
      "epoch": 4.518368751613181,
      "grad_norm": 1.84375,
      "learning_rate": 2.273961442120248e-05,
      "loss": 1.8364,
      "step": 52517
    },
    {
      "epoch": 4.518454787920502,
      "grad_norm": 0.625,
      "learning_rate": 2.2731553074640976e-05,
      "loss": 0.5396,
      "step": 52518
    },
    {
      "epoch": 4.518540824227824,
      "grad_norm": 1.2109375,
      "learning_rate": 2.2723493123996853e-05,
      "loss": 1.629,
      "step": 52519
    },
    {
      "epoch": 4.518626860535146,
      "grad_norm": 7.75,
      "learning_rate": 2.2715434569293648e-05,
      "loss": 0.5736,
      "step": 52520
    },
    {
      "epoch": 4.518712896842468,
      "grad_norm": 1.1171875,
      "learning_rate": 2.2707377410555063e-05,
      "loss": 1.3315,
      "step": 52521
    },
    {
      "epoch": 4.518798933149789,
      "grad_norm": 1.1875,
      "learning_rate": 2.269932164780453e-05,
      "loss": 1.3675,
      "step": 52522
    },
    {
      "epoch": 4.518884969457111,
      "grad_norm": 9.625,
      "learning_rate": 2.2691267281065577e-05,
      "loss": 1.7948,
      "step": 52523
    },
    {
      "epoch": 4.518971005764433,
      "grad_norm": 190.0,
      "learning_rate": 2.2683214310361967e-05,
      "loss": 1.0748,
      "step": 52524
    },
    {
      "epoch": 4.519057042071754,
      "grad_norm": 1.546875,
      "learning_rate": 2.2675162735717014e-05,
      "loss": 1.7705,
      "step": 52525
    },
    {
      "epoch": 4.519143078379076,
      "grad_norm": 3.140625,
      "learning_rate": 2.2667112557154425e-05,
      "loss": 0.9495,
      "step": 52526
    },
    {
      "epoch": 4.519229114686397,
      "grad_norm": 1.1171875,
      "learning_rate": 2.265906377469773e-05,
      "loss": 1.3194,
      "step": 52527
    },
    {
      "epoch": 4.5193151509937195,
      "grad_norm": 4.15625,
      "learning_rate": 2.265101638837036e-05,
      "loss": 1.3185,
      "step": 52528
    },
    {
      "epoch": 4.519401187301041,
      "grad_norm": 0.921875,
      "learning_rate": 2.2642970398195905e-05,
      "loss": 1.0262,
      "step": 52529
    },
    {
      "epoch": 4.519487223608363,
      "grad_norm": 1.6875,
      "learning_rate": 2.263492580419807e-05,
      "loss": 1.4532,
      "step": 52530
    },
    {
      "epoch": 4.519573259915685,
      "grad_norm": 1.5078125,
      "learning_rate": 2.2626882606400112e-05,
      "loss": 0.7541,
      "step": 52531
    },
    {
      "epoch": 4.519659296223006,
      "grad_norm": 1.046875,
      "learning_rate": 2.2618840804825736e-05,
      "loss": 1.0995,
      "step": 52532
    },
    {
      "epoch": 4.519745332530328,
      "grad_norm": 15.1875,
      "learning_rate": 2.2610800399498367e-05,
      "loss": 1.4782,
      "step": 52533
    },
    {
      "epoch": 4.519831368837649,
      "grad_norm": 68.0,
      "learning_rate": 2.260276139044154e-05,
      "loss": 1.5743,
      "step": 52534
    },
    {
      "epoch": 4.519917405144971,
      "grad_norm": 1.1953125,
      "learning_rate": 2.259472377767885e-05,
      "loss": 1.5121,
      "step": 52535
    },
    {
      "epoch": 4.5200034414522925,
      "grad_norm": 1.5625,
      "learning_rate": 2.2586687561233666e-05,
      "loss": 1.3607,
      "step": 52536
    },
    {
      "epoch": 4.520089477759615,
      "grad_norm": 1.1953125,
      "learning_rate": 2.2578652741129634e-05,
      "loss": 1.1128,
      "step": 52537
    },
    {
      "epoch": 4.520175514066937,
      "grad_norm": 41.25,
      "learning_rate": 2.2570619317390183e-05,
      "loss": 0.9707,
      "step": 52538
    },
    {
      "epoch": 4.520261550374258,
      "grad_norm": 19.375,
      "learning_rate": 2.2562587290038793e-05,
      "loss": 0.9082,
      "step": 52539
    },
    {
      "epoch": 4.52034758668158,
      "grad_norm": 1.953125,
      "learning_rate": 2.2554556659098947e-05,
      "loss": 1.4925,
      "step": 52540
    },
    {
      "epoch": 4.520433622988901,
      "grad_norm": 0.64453125,
      "learning_rate": 2.254652742459423e-05,
      "loss": 0.2956,
      "step": 52541
    },
    {
      "epoch": 4.520519659296223,
      "grad_norm": 1.2421875,
      "learning_rate": 2.2538499586548024e-05,
      "loss": 1.3671,
      "step": 52542
    },
    {
      "epoch": 4.520605695603544,
      "grad_norm": 1.1328125,
      "learning_rate": 2.253047314498391e-05,
      "loss": 0.8471,
      "step": 52543
    },
    {
      "epoch": 4.520691731910866,
      "grad_norm": 4576.0,
      "learning_rate": 2.2522448099925206e-05,
      "loss": 1.4394,
      "step": 52544
    },
    {
      "epoch": 4.520777768218188,
      "grad_norm": 13.3125,
      "learning_rate": 2.251442445139551e-05,
      "loss": 0.8447,
      "step": 52545
    },
    {
      "epoch": 4.52086380452551,
      "grad_norm": 0.8984375,
      "learning_rate": 2.2506402199418296e-05,
      "loss": 0.7964,
      "step": 52546
    },
    {
      "epoch": 4.520949840832832,
      "grad_norm": 1.09375,
      "learning_rate": 2.249838134401694e-05,
      "loss": 0.5543,
      "step": 52547
    },
    {
      "epoch": 4.521035877140153,
      "grad_norm": 1.28125,
      "learning_rate": 2.249036188521497e-05,
      "loss": 1.2023,
      "step": 52548
    },
    {
      "epoch": 4.521121913447475,
      "grad_norm": 132.0,
      "learning_rate": 2.2482343823035824e-05,
      "loss": 0.9117,
      "step": 52549
    },
    {
      "epoch": 4.521207949754796,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2474327157502917e-05,
      "loss": 1.6675,
      "step": 52550
    },
    {
      "epoch": 4.521293986062118,
      "grad_norm": 2.0,
      "learning_rate": 2.2466311888639733e-05,
      "loss": 0.9646,
      "step": 52551
    },
    {
      "epoch": 4.52138002236944,
      "grad_norm": 0.87109375,
      "learning_rate": 2.2458298016469813e-05,
      "loss": 0.9973,
      "step": 52552
    },
    {
      "epoch": 4.521466058676761,
      "grad_norm": 1.1484375,
      "learning_rate": 2.245028554101636e-05,
      "loss": 1.3965,
      "step": 52553
    },
    {
      "epoch": 4.5215520949840835,
      "grad_norm": 0.78515625,
      "learning_rate": 2.2442274462303013e-05,
      "loss": 0.6686,
      "step": 52554
    },
    {
      "epoch": 4.521638131291405,
      "grad_norm": 1.1640625,
      "learning_rate": 2.24342647803531e-05,
      "loss": 1.3261,
      "step": 52555
    },
    {
      "epoch": 4.521724167598727,
      "grad_norm": 1.03125,
      "learning_rate": 2.2426256495190035e-05,
      "loss": 0.8845,
      "step": 52556
    },
    {
      "epoch": 4.521810203906048,
      "grad_norm": 1.34375,
      "learning_rate": 2.2418249606837415e-05,
      "loss": 1.3806,
      "step": 52557
    },
    {
      "epoch": 4.52189624021337,
      "grad_norm": 1.2109375,
      "learning_rate": 2.2410244115318445e-05,
      "loss": 1.2527,
      "step": 52558
    },
    {
      "epoch": 4.521982276520692,
      "grad_norm": 1.3828125,
      "learning_rate": 2.2402240020656718e-05,
      "loss": 0.9619,
      "step": 52559
    },
    {
      "epoch": 4.522068312828013,
      "grad_norm": 0.859375,
      "learning_rate": 2.2394237322875432e-05,
      "loss": 0.8624,
      "step": 52560
    },
    {
      "epoch": 4.522154349135335,
      "grad_norm": 0.9140625,
      "learning_rate": 2.238623602199813e-05,
      "loss": 0.6924,
      "step": 52561
    },
    {
      "epoch": 4.5222403854426565,
      "grad_norm": 1.2578125,
      "learning_rate": 2.237823611804829e-05,
      "loss": 0.9548,
      "step": 52562
    },
    {
      "epoch": 4.522326421749979,
      "grad_norm": 1.3125,
      "learning_rate": 2.2370237611049117e-05,
      "loss": 1.7124,
      "step": 52563
    },
    {
      "epoch": 4.5224124580573,
      "grad_norm": 1.0,
      "learning_rate": 2.2362240501024146e-05,
      "loss": 1.1934,
      "step": 52564
    },
    {
      "epoch": 4.522498494364622,
      "grad_norm": 0.9296875,
      "learning_rate": 2.2354244787996747e-05,
      "loss": 1.1651,
      "step": 52565
    },
    {
      "epoch": 4.522584530671944,
      "grad_norm": 0.984375,
      "learning_rate": 2.2346250471990238e-05,
      "loss": 0.6419,
      "step": 52566
    },
    {
      "epoch": 4.522670566979265,
      "grad_norm": 1.6484375,
      "learning_rate": 2.233825755302804e-05,
      "loss": 1.5334,
      "step": 52567
    },
    {
      "epoch": 4.522756603286587,
      "grad_norm": 0.92578125,
      "learning_rate": 2.233026603113364e-05,
      "loss": 0.6241,
      "step": 52568
    },
    {
      "epoch": 4.522842639593908,
      "grad_norm": 1.1171875,
      "learning_rate": 2.2322275906330237e-05,
      "loss": 1.0742,
      "step": 52569
    },
    {
      "epoch": 4.52292867590123,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2314287178641202e-05,
      "loss": 1.1528,
      "step": 52570
    },
    {
      "epoch": 4.523014712208552,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2306299848090127e-05,
      "loss": 0.9838,
      "step": 52571
    },
    {
      "epoch": 4.523100748515874,
      "grad_norm": 1.0703125,
      "learning_rate": 2.2298313914700107e-05,
      "loss": 1.1993,
      "step": 52572
    },
    {
      "epoch": 4.523186784823196,
      "grad_norm": 1.46875,
      "learning_rate": 2.229032937849468e-05,
      "loss": 1.5367,
      "step": 52573
    },
    {
      "epoch": 4.523272821130517,
      "grad_norm": 1.171875,
      "learning_rate": 2.2282346239497096e-05,
      "loss": 1.51,
      "step": 52574
    },
    {
      "epoch": 4.523358857437839,
      "grad_norm": 1.265625,
      "learning_rate": 2.227436449773068e-05,
      "loss": 1.5262,
      "step": 52575
    },
    {
      "epoch": 4.52344489374516,
      "grad_norm": 1.3046875,
      "learning_rate": 2.2266384153218967e-05,
      "loss": 0.7588,
      "step": 52576
    },
    {
      "epoch": 4.523530930052482,
      "grad_norm": 1.359375,
      "learning_rate": 2.2258405205985044e-05,
      "loss": 1.6167,
      "step": 52577
    },
    {
      "epoch": 4.523616966359803,
      "grad_norm": 7.78125,
      "learning_rate": 2.2250427656052398e-05,
      "loss": 1.2151,
      "step": 52578
    },
    {
      "epoch": 4.523703002667125,
      "grad_norm": 1.265625,
      "learning_rate": 2.2242451503444395e-05,
      "loss": 1.5485,
      "step": 52579
    },
    {
      "epoch": 4.5237890389744475,
      "grad_norm": 1.296875,
      "learning_rate": 2.223447674818424e-05,
      "loss": 1.7196,
      "step": 52580
    },
    {
      "epoch": 4.523875075281769,
      "grad_norm": 1.109375,
      "learning_rate": 2.2226503390295306e-05,
      "loss": 0.6648,
      "step": 52581
    },
    {
      "epoch": 4.523961111589091,
      "grad_norm": 0.66015625,
      "learning_rate": 2.2218531429801013e-05,
      "loss": 0.5414,
      "step": 52582
    },
    {
      "epoch": 4.524047147896412,
      "grad_norm": 1.3984375,
      "learning_rate": 2.2210560866724516e-05,
      "loss": 0.9059,
      "step": 52583
    },
    {
      "epoch": 4.524133184203734,
      "grad_norm": 1.0546875,
      "learning_rate": 2.220259170108929e-05,
      "loss": 0.8598,
      "step": 52584
    },
    {
      "epoch": 4.524219220511055,
      "grad_norm": 1.109375,
      "learning_rate": 2.2194623932918435e-05,
      "loss": 1.7127,
      "step": 52585
    },
    {
      "epoch": 4.524305256818377,
      "grad_norm": 1.4140625,
      "learning_rate": 2.2186657562235423e-05,
      "loss": 1.9006,
      "step": 52586
    },
    {
      "epoch": 4.524391293125699,
      "grad_norm": 22.75,
      "learning_rate": 2.2178692589063577e-05,
      "loss": 1.6552,
      "step": 52587
    },
    {
      "epoch": 4.5244773294330205,
      "grad_norm": 0.97265625,
      "learning_rate": 2.217072901342604e-05,
      "loss": 0.7406,
      "step": 52588
    },
    {
      "epoch": 4.524563365740343,
      "grad_norm": 1.140625,
      "learning_rate": 2.2162766835346238e-05,
      "loss": 0.949,
      "step": 52589
    },
    {
      "epoch": 4.524649402047664,
      "grad_norm": 1.046875,
      "learning_rate": 2.2154806054847433e-05,
      "loss": 1.0923,
      "step": 52590
    },
    {
      "epoch": 4.524735438354986,
      "grad_norm": 1.0,
      "learning_rate": 2.2146846671952713e-05,
      "loss": 0.7236,
      "step": 52591
    },
    {
      "epoch": 4.524821474662307,
      "grad_norm": 3.0,
      "learning_rate": 2.213888868668573e-05,
      "loss": 1.5664,
      "step": 52592
    },
    {
      "epoch": 4.524907510969629,
      "grad_norm": 0.94921875,
      "learning_rate": 2.2130932099069468e-05,
      "loss": 0.7448,
      "step": 52593
    },
    {
      "epoch": 4.524993547276951,
      "grad_norm": 16.625,
      "learning_rate": 2.2122976909127234e-05,
      "loss": 1.073,
      "step": 52594
    },
    {
      "epoch": 4.525079583584272,
      "grad_norm": 1.0,
      "learning_rate": 2.2115023116882514e-05,
      "loss": 0.998,
      "step": 52595
    },
    {
      "epoch": 4.525165619891594,
      "grad_norm": 0.79296875,
      "learning_rate": 2.210707072235829e-05,
      "loss": 0.466,
      "step": 52596
    },
    {
      "epoch": 4.525251656198916,
      "grad_norm": 14.8125,
      "learning_rate": 2.209911972557793e-05,
      "loss": 0.8923,
      "step": 52597
    },
    {
      "epoch": 4.525337692506238,
      "grad_norm": 1.1171875,
      "learning_rate": 2.209117012656475e-05,
      "loss": 1.0414,
      "step": 52598
    },
    {
      "epoch": 4.52542372881356,
      "grad_norm": 1.1015625,
      "learning_rate": 2.2083221925341956e-05,
      "loss": 1.2537,
      "step": 52599
    },
    {
      "epoch": 4.525509765120881,
      "grad_norm": 1.2109375,
      "learning_rate": 2.2075275121932802e-05,
      "loss": 1.2324,
      "step": 52600
    },
    {
      "epoch": 4.525595801428203,
      "grad_norm": 1.0859375,
      "learning_rate": 2.206732971636044e-05,
      "loss": 0.912,
      "step": 52601
    },
    {
      "epoch": 4.525681837735524,
      "grad_norm": 1.2578125,
      "learning_rate": 2.2059385708648184e-05,
      "loss": 0.879,
      "step": 52602
    },
    {
      "epoch": 4.525767874042846,
      "grad_norm": 5.3125,
      "learning_rate": 2.20514430988194e-05,
      "loss": 1.0873,
      "step": 52603
    },
    {
      "epoch": 4.525853910350168,
      "grad_norm": 1.1328125,
      "learning_rate": 2.204350188689708e-05,
      "loss": 1.3389,
      "step": 52604
    },
    {
      "epoch": 4.525939946657489,
      "grad_norm": 1.53125,
      "learning_rate": 2.2035562072904525e-05,
      "loss": 1.1107,
      "step": 52605
    },
    {
      "epoch": 4.5260259829648115,
      "grad_norm": 1.1484375,
      "learning_rate": 2.2027623656865115e-05,
      "loss": 0.8544,
      "step": 52606
    },
    {
      "epoch": 4.526112019272133,
      "grad_norm": 1.0703125,
      "learning_rate": 2.2019686638801827e-05,
      "loss": 0.7486,
      "step": 52607
    },
    {
      "epoch": 4.526198055579455,
      "grad_norm": 1.03125,
      "learning_rate": 2.2011751018738037e-05,
      "loss": 0.8751,
      "step": 52608
    },
    {
      "epoch": 4.526284091886776,
      "grad_norm": 1.0234375,
      "learning_rate": 2.2003816796696996e-05,
      "loss": 1.073,
      "step": 52609
    },
    {
      "epoch": 4.526370128194098,
      "grad_norm": 1.265625,
      "learning_rate": 2.199588397270169e-05,
      "loss": 1.0935,
      "step": 52610
    },
    {
      "epoch": 4.52645616450142,
      "grad_norm": 1.0390625,
      "learning_rate": 2.198795254677549e-05,
      "loss": 1.0445,
      "step": 52611
    },
    {
      "epoch": 4.526542200808741,
      "grad_norm": 1.125,
      "learning_rate": 2.1980022518941657e-05,
      "loss": 1.1432,
      "step": 52612
    },
    {
      "epoch": 4.526628237116063,
      "grad_norm": 1.1015625,
      "learning_rate": 2.1972093889223167e-05,
      "loss": 0.6638,
      "step": 52613
    },
    {
      "epoch": 4.5267142734233845,
      "grad_norm": 1.4140625,
      "learning_rate": 2.1964166657643447e-05,
      "loss": 1.3728,
      "step": 52614
    },
    {
      "epoch": 4.526800309730707,
      "grad_norm": 1.15625,
      "learning_rate": 2.195624082422548e-05,
      "loss": 1.3012,
      "step": 52615
    },
    {
      "epoch": 4.526886346038028,
      "grad_norm": 0.984375,
      "learning_rate": 2.1948316388992527e-05,
      "loss": 0.9679,
      "step": 52616
    },
    {
      "epoch": 4.52697238234535,
      "grad_norm": 40.0,
      "learning_rate": 2.1940393351967846e-05,
      "loss": 1.3627,
      "step": 52617
    },
    {
      "epoch": 4.527058418652672,
      "grad_norm": 6912.0,
      "learning_rate": 2.1932471713174472e-05,
      "loss": 1.0112,
      "step": 52618
    },
    {
      "epoch": 4.527144454959993,
      "grad_norm": 1.484375,
      "learning_rate": 2.1924551472635666e-05,
      "loss": 1.2235,
      "step": 52619
    },
    {
      "epoch": 4.527230491267315,
      "grad_norm": 1.296875,
      "learning_rate": 2.1916632630374578e-05,
      "loss": 1.6076,
      "step": 52620
    },
    {
      "epoch": 4.527316527574636,
      "grad_norm": 1.015625,
      "learning_rate": 2.1908715186414296e-05,
      "loss": 0.6587,
      "step": 52621
    },
    {
      "epoch": 4.527402563881958,
      "grad_norm": 0.73046875,
      "learning_rate": 2.190079914077808e-05,
      "loss": 0.6889,
      "step": 52622
    },
    {
      "epoch": 4.52748860018928,
      "grad_norm": 1.1953125,
      "learning_rate": 2.1892884493489084e-05,
      "loss": 0.617,
      "step": 52623
    },
    {
      "epoch": 4.527574636496602,
      "grad_norm": 2.859375,
      "learning_rate": 2.188497124457034e-05,
      "loss": 1.4671,
      "step": 52624
    },
    {
      "epoch": 4.527660672803924,
      "grad_norm": 0.953125,
      "learning_rate": 2.1877059394045164e-05,
      "loss": 0.7059,
      "step": 52625
    },
    {
      "epoch": 4.527746709111245,
      "grad_norm": 1.09375,
      "learning_rate": 2.1869148941936533e-05,
      "loss": 1.4267,
      "step": 52626
    },
    {
      "epoch": 4.527832745418567,
      "grad_norm": 1.25,
      "learning_rate": 2.1861239888267602e-05,
      "loss": 0.9146,
      "step": 52627
    },
    {
      "epoch": 4.527918781725888,
      "grad_norm": 0.78515625,
      "learning_rate": 2.1853332233061573e-05,
      "loss": 0.6389,
      "step": 52628
    },
    {
      "epoch": 4.52800481803321,
      "grad_norm": 1.2421875,
      "learning_rate": 2.1845425976341594e-05,
      "loss": 1.7151,
      "step": 52629
    },
    {
      "epoch": 4.528090854340531,
      "grad_norm": 1112.0,
      "learning_rate": 2.1837521118130755e-05,
      "loss": 0.7154,
      "step": 52630
    },
    {
      "epoch": 4.5281768906478534,
      "grad_norm": 37.75,
      "learning_rate": 2.182961765845215e-05,
      "loss": 1.366,
      "step": 52631
    },
    {
      "epoch": 4.5282629269551755,
      "grad_norm": 2.15625,
      "learning_rate": 2.1821715597328874e-05,
      "loss": 1.4201,
      "step": 52632
    },
    {
      "epoch": 4.528348963262497,
      "grad_norm": 1.1484375,
      "learning_rate": 2.181381493478418e-05,
      "loss": 1.4232,
      "step": 52633
    },
    {
      "epoch": 4.528434999569819,
      "grad_norm": 1.2109375,
      "learning_rate": 2.1805915670841002e-05,
      "loss": 0.8753,
      "step": 52634
    },
    {
      "epoch": 4.52852103587714,
      "grad_norm": 0.90234375,
      "learning_rate": 2.1798017805522485e-05,
      "loss": 0.5935,
      "step": 52635
    },
    {
      "epoch": 4.528607072184462,
      "grad_norm": 1.4140625,
      "learning_rate": 2.1790121338851886e-05,
      "loss": 1.2132,
      "step": 52636
    },
    {
      "epoch": 4.528693108491783,
      "grad_norm": 1.265625,
      "learning_rate": 2.178222627085208e-05,
      "loss": 1.2325,
      "step": 52637
    },
    {
      "epoch": 4.528779144799105,
      "grad_norm": 13.4375,
      "learning_rate": 2.177433260154621e-05,
      "loss": 1.0185,
      "step": 52638
    },
    {
      "epoch": 4.528865181106427,
      "grad_norm": 1.5625,
      "learning_rate": 2.1766440330957536e-05,
      "loss": 0.9513,
      "step": 52639
    },
    {
      "epoch": 4.5289512174137485,
      "grad_norm": 1.0703125,
      "learning_rate": 2.1758549459108933e-05,
      "loss": 0.5761,
      "step": 52640
    },
    {
      "epoch": 4.529037253721071,
      "grad_norm": 1.1171875,
      "learning_rate": 2.1750659986023603e-05,
      "loss": 0.7665,
      "step": 52641
    },
    {
      "epoch": 4.529123290028392,
      "grad_norm": 1.390625,
      "learning_rate": 2.1742771911724525e-05,
      "loss": 1.5478,
      "step": 52642
    },
    {
      "epoch": 4.529209326335714,
      "grad_norm": 0.921875,
      "learning_rate": 2.1734885236234792e-05,
      "loss": 1.2362,
      "step": 52643
    },
    {
      "epoch": 4.529295362643035,
      "grad_norm": 0.94140625,
      "learning_rate": 2.172699995957761e-05,
      "loss": 0.8941,
      "step": 52644
    },
    {
      "epoch": 4.529381398950357,
      "grad_norm": 1.0625,
      "learning_rate": 2.1719116081775848e-05,
      "loss": 0.6079,
      "step": 52645
    },
    {
      "epoch": 4.529467435257679,
      "grad_norm": 1.6015625,
      "learning_rate": 2.171123360285271e-05,
      "loss": 0.9133,
      "step": 52646
    },
    {
      "epoch": 4.529553471565,
      "grad_norm": 0.70703125,
      "learning_rate": 2.1703352522831176e-05,
      "loss": 0.3378,
      "step": 52647
    },
    {
      "epoch": 4.529639507872322,
      "grad_norm": 1.5078125,
      "learning_rate": 2.1695472841734288e-05,
      "loss": 1.362,
      "step": 52648
    },
    {
      "epoch": 4.529725544179644,
      "grad_norm": 1.0390625,
      "learning_rate": 2.1687594559585133e-05,
      "loss": 1.0663,
      "step": 52649
    },
    {
      "epoch": 4.529811580486966,
      "grad_norm": 1.3984375,
      "learning_rate": 2.167971767640675e-05,
      "loss": 0.8053,
      "step": 52650
    },
    {
      "epoch": 4.529897616794287,
      "grad_norm": 1.21875,
      "learning_rate": 2.1671842192222123e-05,
      "loss": 1.1331,
      "step": 52651
    },
    {
      "epoch": 4.529983653101609,
      "grad_norm": 1.25,
      "learning_rate": 2.1663968107054343e-05,
      "loss": 1.011,
      "step": 52652
    },
    {
      "epoch": 4.530069689408931,
      "grad_norm": 1.03125,
      "learning_rate": 2.1656095420926447e-05,
      "loss": 1.679,
      "step": 52653
    },
    {
      "epoch": 4.530155725716252,
      "grad_norm": 1.109375,
      "learning_rate": 2.1648224133861417e-05,
      "loss": 0.576,
      "step": 52654
    },
    {
      "epoch": 4.530241762023574,
      "grad_norm": 55.5,
      "learning_rate": 2.1640354245882343e-05,
      "loss": 1.7172,
      "step": 52655
    },
    {
      "epoch": 4.530327798330895,
      "grad_norm": 0.95703125,
      "learning_rate": 2.1632485757012155e-05,
      "loss": 0.6422,
      "step": 52656
    },
    {
      "epoch": 4.5304138346382175,
      "grad_norm": 0.890625,
      "learning_rate": 2.1624618667273886e-05,
      "loss": 0.9469,
      "step": 52657
    },
    {
      "epoch": 4.530499870945539,
      "grad_norm": 0.86328125,
      "learning_rate": 2.1616752976690633e-05,
      "loss": 1.0278,
      "step": 52658
    },
    {
      "epoch": 4.530585907252861,
      "grad_norm": 31.75,
      "learning_rate": 2.160888868528532e-05,
      "loss": 2.2478,
      "step": 52659
    },
    {
      "epoch": 4.530671943560183,
      "grad_norm": 1.0234375,
      "learning_rate": 2.160102579308093e-05,
      "loss": 0.8669,
      "step": 52660
    },
    {
      "epoch": 4.530757979867504,
      "grad_norm": 1.2890625,
      "learning_rate": 2.1593164300100555e-05,
      "loss": 0.9701,
      "step": 52661
    },
    {
      "epoch": 4.530844016174826,
      "grad_norm": 103.5,
      "learning_rate": 2.158530420636706e-05,
      "loss": 1.0313,
      "step": 52662
    },
    {
      "epoch": 4.530930052482147,
      "grad_norm": 0.88671875,
      "learning_rate": 2.1577445511903492e-05,
      "loss": 0.8544,
      "step": 52663
    },
    {
      "epoch": 4.531016088789469,
      "grad_norm": 1.171875,
      "learning_rate": 2.1569588216732937e-05,
      "loss": 0.9318,
      "step": 52664
    },
    {
      "epoch": 4.5311021250967904,
      "grad_norm": 1.1171875,
      "learning_rate": 2.156173232087816e-05,
      "loss": 0.6582,
      "step": 52665
    },
    {
      "epoch": 4.5311881614041125,
      "grad_norm": 1.0859375,
      "learning_rate": 2.1553877824362355e-05,
      "loss": 1.2179,
      "step": 52666
    },
    {
      "epoch": 4.531274197711435,
      "grad_norm": 1.4921875,
      "learning_rate": 2.1546024727208403e-05,
      "loss": 0.8556,
      "step": 52667
    },
    {
      "epoch": 4.531360234018756,
      "grad_norm": 1.6328125,
      "learning_rate": 2.1538173029439222e-05,
      "loss": 1.1284,
      "step": 52668
    },
    {
      "epoch": 4.531446270326078,
      "grad_norm": 1.2421875,
      "learning_rate": 2.153032273107791e-05,
      "loss": 0.67,
      "step": 52669
    },
    {
      "epoch": 4.531532306633399,
      "grad_norm": 1.296875,
      "learning_rate": 2.1522473832147284e-05,
      "loss": 0.7649,
      "step": 52670
    },
    {
      "epoch": 4.531618342940721,
      "grad_norm": 1.0625,
      "learning_rate": 2.151462633267043e-05,
      "loss": 1.001,
      "step": 52671
    },
    {
      "epoch": 4.531704379248042,
      "grad_norm": 17.75,
      "learning_rate": 2.1506780232670163e-05,
      "loss": 1.2997,
      "step": 52672
    },
    {
      "epoch": 4.531790415555364,
      "grad_norm": 1.125,
      "learning_rate": 2.1498935532169528e-05,
      "loss": 0.675,
      "step": 52673
    },
    {
      "epoch": 4.531876451862686,
      "grad_norm": 1.09375,
      "learning_rate": 2.14910922311915e-05,
      "loss": 0.7743,
      "step": 52674
    },
    {
      "epoch": 4.531962488170008,
      "grad_norm": 1.1015625,
      "learning_rate": 2.1483250329758896e-05,
      "loss": 1.294,
      "step": 52675
    },
    {
      "epoch": 4.53204852447733,
      "grad_norm": 0.7421875,
      "learning_rate": 2.1475409827894698e-05,
      "loss": 0.5045,
      "step": 52676
    },
    {
      "epoch": 4.532134560784651,
      "grad_norm": 0.94921875,
      "learning_rate": 2.1467570725621944e-05,
      "loss": 0.4844,
      "step": 52677
    },
    {
      "epoch": 4.532220597091973,
      "grad_norm": 3.765625,
      "learning_rate": 2.145973302296339e-05,
      "loss": 0.9585,
      "step": 52678
    },
    {
      "epoch": 4.532306633399294,
      "grad_norm": 0.8671875,
      "learning_rate": 2.1451896719942077e-05,
      "loss": 1.1226,
      "step": 52679
    },
    {
      "epoch": 4.532392669706616,
      "grad_norm": 1.171875,
      "learning_rate": 2.1444061816580927e-05,
      "loss": 0.9528,
      "step": 52680
    },
    {
      "epoch": 4.532478706013938,
      "grad_norm": 1.140625,
      "learning_rate": 2.1436228312902763e-05,
      "loss": 1.1897,
      "step": 52681
    },
    {
      "epoch": 4.532564742321259,
      "grad_norm": 1.2421875,
      "learning_rate": 2.1428396208930555e-05,
      "loss": 0.8935,
      "step": 52682
    },
    {
      "epoch": 4.5326507786285815,
      "grad_norm": 1.2421875,
      "learning_rate": 2.1420565504687296e-05,
      "loss": 1.1657,
      "step": 52683
    },
    {
      "epoch": 4.532736814935903,
      "grad_norm": 1.234375,
      "learning_rate": 2.1412736200195737e-05,
      "loss": 1.6861,
      "step": 52684
    },
    {
      "epoch": 4.532822851243225,
      "grad_norm": 0.98828125,
      "learning_rate": 2.1404908295478866e-05,
      "loss": 0.708,
      "step": 52685
    },
    {
      "epoch": 4.532908887550546,
      "grad_norm": 93.5,
      "learning_rate": 2.1397081790559546e-05,
      "loss": 1.1485,
      "step": 52686
    },
    {
      "epoch": 4.532994923857868,
      "grad_norm": 1.140625,
      "learning_rate": 2.1389256685460655e-05,
      "loss": 1.599,
      "step": 52687
    },
    {
      "epoch": 4.53308096016519,
      "grad_norm": 1.046875,
      "learning_rate": 2.1381432980205174e-05,
      "loss": 1.2159,
      "step": 52688
    },
    {
      "epoch": 4.533166996472511,
      "grad_norm": 1.1328125,
      "learning_rate": 2.1373610674815857e-05,
      "loss": 1.7005,
      "step": 52689
    },
    {
      "epoch": 4.533253032779833,
      "grad_norm": 0.796875,
      "learning_rate": 2.1365789769315582e-05,
      "loss": 0.705,
      "step": 52690
    },
    {
      "epoch": 4.5333390690871544,
      "grad_norm": 1.5078125,
      "learning_rate": 2.1357970263727434e-05,
      "loss": 2.0261,
      "step": 52691
    },
    {
      "epoch": 4.5334251053944765,
      "grad_norm": 0.9609375,
      "learning_rate": 2.135015215807401e-05,
      "loss": 0.9097,
      "step": 52692
    },
    {
      "epoch": 4.533511141701798,
      "grad_norm": 1.109375,
      "learning_rate": 2.134233545237835e-05,
      "loss": 1.9547,
      "step": 52693
    },
    {
      "epoch": 4.53359717800912,
      "grad_norm": 1.5078125,
      "learning_rate": 2.1334520146663262e-05,
      "loss": 1.6676,
      "step": 52694
    },
    {
      "epoch": 4.533683214316442,
      "grad_norm": 1.1015625,
      "learning_rate": 2.1326706240951566e-05,
      "loss": 1.0256,
      "step": 52695
    },
    {
      "epoch": 4.533769250623763,
      "grad_norm": 0.96484375,
      "learning_rate": 2.1318893735266243e-05,
      "loss": 1.6184,
      "step": 52696
    },
    {
      "epoch": 4.533855286931085,
      "grad_norm": 1.90625,
      "learning_rate": 2.1311082629629996e-05,
      "loss": 0.926,
      "step": 52697
    },
    {
      "epoch": 4.533941323238406,
      "grad_norm": 840.0,
      "learning_rate": 2.1303272924065752e-05,
      "loss": 1.1135,
      "step": 52698
    },
    {
      "epoch": 4.534027359545728,
      "grad_norm": 0.89453125,
      "learning_rate": 2.1295464618596328e-05,
      "loss": 0.8996,
      "step": 52699
    },
    {
      "epoch": 4.5341133958530495,
      "grad_norm": 1.3125,
      "learning_rate": 2.1287657713244536e-05,
      "loss": 0.7932,
      "step": 52700
    },
    {
      "epoch": 4.534199432160372,
      "grad_norm": 1.59375,
      "learning_rate": 2.1279852208033245e-05,
      "loss": 1.2465,
      "step": 52701
    },
    {
      "epoch": 4.534285468467694,
      "grad_norm": 3.640625,
      "learning_rate": 2.1272048102985332e-05,
      "loss": 0.8614,
      "step": 52702
    },
    {
      "epoch": 4.534371504775015,
      "grad_norm": 1.125,
      "learning_rate": 2.1264245398123495e-05,
      "loss": 1.7014,
      "step": 52703
    },
    {
      "epoch": 4.534457541082337,
      "grad_norm": 61.0,
      "learning_rate": 2.1256444093470718e-05,
      "loss": 0.6091,
      "step": 52704
    },
    {
      "epoch": 4.534543577389658,
      "grad_norm": 0.953125,
      "learning_rate": 2.1248644189049648e-05,
      "loss": 0.8284,
      "step": 52705
    },
    {
      "epoch": 4.53462961369698,
      "grad_norm": 0.9375,
      "learning_rate": 2.1240845684883216e-05,
      "loss": 1.4391,
      "step": 52706
    },
    {
      "epoch": 4.534715650004301,
      "grad_norm": 1.109375,
      "learning_rate": 2.1233048580994284e-05,
      "loss": 0.7588,
      "step": 52707
    },
    {
      "epoch": 4.534801686311623,
      "grad_norm": 47.75,
      "learning_rate": 2.1225252877405455e-05,
      "loss": 0.7118,
      "step": 52708
    },
    {
      "epoch": 4.5348877226189455,
      "grad_norm": 0.88671875,
      "learning_rate": 2.1217458574139647e-05,
      "loss": 1.5598,
      "step": 52709
    },
    {
      "epoch": 4.534973758926267,
      "grad_norm": 1.3515625,
      "learning_rate": 2.1209665671219735e-05,
      "loss": 0.8628,
      "step": 52710
    },
    {
      "epoch": 4.535059795233589,
      "grad_norm": 20.5,
      "learning_rate": 2.1201874168668366e-05,
      "loss": 1.3987,
      "step": 52711
    },
    {
      "epoch": 4.53514583154091,
      "grad_norm": 0.9375,
      "learning_rate": 2.119408406650847e-05,
      "loss": 1.2377,
      "step": 52712
    },
    {
      "epoch": 4.535231867848232,
      "grad_norm": 1.1796875,
      "learning_rate": 2.1186295364762688e-05,
      "loss": 1.4014,
      "step": 52713
    },
    {
      "epoch": 4.535317904155554,
      "grad_norm": 55.75,
      "learning_rate": 2.117850806345384e-05,
      "loss": 1.7404,
      "step": 52714
    },
    {
      "epoch": 4.535403940462875,
      "grad_norm": 5.75,
      "learning_rate": 2.1170722162604794e-05,
      "loss": 1.3757,
      "step": 52715
    },
    {
      "epoch": 4.535489976770197,
      "grad_norm": 0.7890625,
      "learning_rate": 2.1162937662238258e-05,
      "loss": 0.7548,
      "step": 52716
    },
    {
      "epoch": 4.5355760130775185,
      "grad_norm": 1.0546875,
      "learning_rate": 2.1155154562376932e-05,
      "loss": 1.0516,
      "step": 52717
    },
    {
      "epoch": 4.5356620493848405,
      "grad_norm": 1.5859375,
      "learning_rate": 2.1147372863043747e-05,
      "loss": 0.5074,
      "step": 52718
    },
    {
      "epoch": 4.535748085692163,
      "grad_norm": 1.265625,
      "learning_rate": 2.113959256426129e-05,
      "loss": 1.5062,
      "step": 52719
    },
    {
      "epoch": 4.535834121999484,
      "grad_norm": 1.4375,
      "learning_rate": 2.113181366605238e-05,
      "loss": 1.1856,
      "step": 52720
    },
    {
      "epoch": 4.535920158306806,
      "grad_norm": 1.2265625,
      "learning_rate": 2.1124036168439886e-05,
      "loss": 1.8281,
      "step": 52721
    },
    {
      "epoch": 4.536006194614127,
      "grad_norm": 0.98046875,
      "learning_rate": 2.111626007144635e-05,
      "loss": 0.4506,
      "step": 52722
    },
    {
      "epoch": 4.536092230921449,
      "grad_norm": 0.9453125,
      "learning_rate": 2.110848537509463e-05,
      "loss": 1.4601,
      "step": 52723
    },
    {
      "epoch": 4.53617826722877,
      "grad_norm": 1.171875,
      "learning_rate": 2.11007120794075e-05,
      "loss": 0.5183,
      "step": 52724
    },
    {
      "epoch": 4.536264303536092,
      "grad_norm": 71.5,
      "learning_rate": 2.109294018440755e-05,
      "loss": 1.127,
      "step": 52725
    },
    {
      "epoch": 4.536350339843414,
      "grad_norm": 0.9921875,
      "learning_rate": 2.1085169690117757e-05,
      "loss": 1.6555,
      "step": 52726
    },
    {
      "epoch": 4.536436376150736,
      "grad_norm": 0.98046875,
      "learning_rate": 2.1077400596560547e-05,
      "loss": 0.5093,
      "step": 52727
    },
    {
      "epoch": 4.536522412458058,
      "grad_norm": 1.4609375,
      "learning_rate": 2.1069632903758852e-05,
      "loss": 1.4839,
      "step": 52728
    },
    {
      "epoch": 4.536608448765379,
      "grad_norm": 0.93359375,
      "learning_rate": 2.106186661173537e-05,
      "loss": 0.6478,
      "step": 52729
    },
    {
      "epoch": 4.536694485072701,
      "grad_norm": 2.328125,
      "learning_rate": 2.1054101720512752e-05,
      "loss": 0.9466,
      "step": 52730
    },
    {
      "epoch": 4.536780521380022,
      "grad_norm": 11.125,
      "learning_rate": 2.1046338230113704e-05,
      "loss": 1.1245,
      "step": 52731
    },
    {
      "epoch": 4.536866557687344,
      "grad_norm": 1.09375,
      "learning_rate": 2.1038576140561038e-05,
      "loss": 0.9869,
      "step": 52732
    },
    {
      "epoch": 4.536952593994666,
      "grad_norm": 1.328125,
      "learning_rate": 2.103081545187735e-05,
      "loss": 1.4786,
      "step": 52733
    },
    {
      "epoch": 4.537038630301987,
      "grad_norm": 552.0,
      "learning_rate": 2.102305616408534e-05,
      "loss": 1.2123,
      "step": 52734
    },
    {
      "epoch": 4.5371246666093095,
      "grad_norm": 1.5546875,
      "learning_rate": 2.1015298277207774e-05,
      "loss": 0.9134,
      "step": 52735
    },
    {
      "epoch": 4.537210702916631,
      "grad_norm": 1.0390625,
      "learning_rate": 2.1007541791267294e-05,
      "loss": 0.8879,
      "step": 52736
    },
    {
      "epoch": 4.537296739223953,
      "grad_norm": 3.140625,
      "learning_rate": 2.0999786706286605e-05,
      "loss": 1.8841,
      "step": 52737
    },
    {
      "epoch": 4.537382775531274,
      "grad_norm": 1968.0,
      "learning_rate": 2.099203302228836e-05,
      "loss": 0.794,
      "step": 52738
    },
    {
      "epoch": 4.537468811838596,
      "grad_norm": 1.03125,
      "learning_rate": 2.0984280739295147e-05,
      "loss": 1.104,
      "step": 52739
    },
    {
      "epoch": 4.537554848145918,
      "grad_norm": 1.03125,
      "learning_rate": 2.097652985732995e-05,
      "loss": 0.9295,
      "step": 52740
    },
    {
      "epoch": 4.537640884453239,
      "grad_norm": 3.90625,
      "learning_rate": 2.0968780376415086e-05,
      "loss": 1.1921,
      "step": 52741
    },
    {
      "epoch": 4.537726920760561,
      "grad_norm": 1.2578125,
      "learning_rate": 2.0961032296573478e-05,
      "loss": 1.3112,
      "step": 52742
    },
    {
      "epoch": 4.5378129570678825,
      "grad_norm": 1.2890625,
      "learning_rate": 2.0953285617827665e-05,
      "loss": 1.4547,
      "step": 52743
    },
    {
      "epoch": 4.5378989933752045,
      "grad_norm": 0.9765625,
      "learning_rate": 2.0945540340200244e-05,
      "loss": 1.2916,
      "step": 52744
    },
    {
      "epoch": 4.537985029682526,
      "grad_norm": 1.171875,
      "learning_rate": 2.0937796463714076e-05,
      "loss": 0.9758,
      "step": 52745
    },
    {
      "epoch": 4.538071065989848,
      "grad_norm": 15.125,
      "learning_rate": 2.0930053988391595e-05,
      "loss": 0.7388,
      "step": 52746
    },
    {
      "epoch": 4.53815710229717,
      "grad_norm": 1.0,
      "learning_rate": 2.0922312914255502e-05,
      "loss": 1.0075,
      "step": 52747
    },
    {
      "epoch": 4.538243138604491,
      "grad_norm": 1.5,
      "learning_rate": 2.091457324132856e-05,
      "loss": 1.5331,
      "step": 52748
    },
    {
      "epoch": 4.538329174911813,
      "grad_norm": 11.25,
      "learning_rate": 2.090683496963325e-05,
      "loss": 0.9199,
      "step": 52749
    },
    {
      "epoch": 4.538415211219134,
      "grad_norm": 1.09375,
      "learning_rate": 2.089909809919227e-05,
      "loss": 0.7718,
      "step": 52750
    },
    {
      "epoch": 4.538501247526456,
      "grad_norm": 1.2734375,
      "learning_rate": 2.0891362630028277e-05,
      "loss": 1.1998,
      "step": 52751
    },
    {
      "epoch": 4.5385872838337775,
      "grad_norm": 1.140625,
      "learning_rate": 2.088362856216386e-05,
      "loss": 1.5975,
      "step": 52752
    },
    {
      "epoch": 4.5386733201411,
      "grad_norm": 10.0,
      "learning_rate": 2.0875895895621665e-05,
      "loss": 0.8271,
      "step": 52753
    },
    {
      "epoch": 4.538759356448422,
      "grad_norm": 0.9296875,
      "learning_rate": 2.0868164630424237e-05,
      "loss": 0.7848,
      "step": 52754
    },
    {
      "epoch": 4.538845392755743,
      "grad_norm": 0.9375,
      "learning_rate": 2.086043476659427e-05,
      "loss": 0.4982,
      "step": 52755
    },
    {
      "epoch": 4.538931429063065,
      "grad_norm": 0.984375,
      "learning_rate": 2.0852706304154367e-05,
      "loss": 0.6195,
      "step": 52756
    },
    {
      "epoch": 4.539017465370386,
      "grad_norm": 1.296875,
      "learning_rate": 2.084497924312706e-05,
      "loss": 1.389,
      "step": 52757
    },
    {
      "epoch": 4.539103501677708,
      "grad_norm": 0.984375,
      "learning_rate": 2.0837253583535e-05,
      "loss": 0.941,
      "step": 52758
    },
    {
      "epoch": 4.539189537985029,
      "grad_norm": 107.0,
      "learning_rate": 2.0829529325400833e-05,
      "loss": 1.2074,
      "step": 52759
    },
    {
      "epoch": 4.539275574292351,
      "grad_norm": 1.328125,
      "learning_rate": 2.0821806468747095e-05,
      "loss": 0.9829,
      "step": 52760
    },
    {
      "epoch": 4.5393616105996735,
      "grad_norm": 1.125,
      "learning_rate": 2.081408501359633e-05,
      "loss": 1.2215,
      "step": 52761
    },
    {
      "epoch": 4.539447646906995,
      "grad_norm": 1.1484375,
      "learning_rate": 2.0806364959971234e-05,
      "loss": 1.8593,
      "step": 52762
    },
    {
      "epoch": 4.539533683214317,
      "grad_norm": 1.921875,
      "learning_rate": 2.0798646307894244e-05,
      "loss": 1.3734,
      "step": 52763
    },
    {
      "epoch": 4.539619719521638,
      "grad_norm": 1.1328125,
      "learning_rate": 2.079092905738805e-05,
      "loss": 0.6266,
      "step": 52764
    },
    {
      "epoch": 4.53970575582896,
      "grad_norm": 118.0,
      "learning_rate": 2.0783213208475205e-05,
      "loss": 1.1736,
      "step": 52765
    },
    {
      "epoch": 4.539791792136281,
      "grad_norm": 1.265625,
      "learning_rate": 2.0775498761178236e-05,
      "loss": 1.3217,
      "step": 52766
    },
    {
      "epoch": 4.539877828443603,
      "grad_norm": 1.453125,
      "learning_rate": 2.0767785715519793e-05,
      "loss": 0.9375,
      "step": 52767
    },
    {
      "epoch": 4.539963864750925,
      "grad_norm": 8.6875,
      "learning_rate": 2.076007407152225e-05,
      "loss": 0.8083,
      "step": 52768
    },
    {
      "epoch": 4.5400499010582465,
      "grad_norm": 1.1796875,
      "learning_rate": 2.0752363829208366e-05,
      "loss": 1.4446,
      "step": 52769
    },
    {
      "epoch": 4.5401359373655685,
      "grad_norm": 1.2421875,
      "learning_rate": 2.074465498860062e-05,
      "loss": 1.4036,
      "step": 52770
    },
    {
      "epoch": 4.54022197367289,
      "grad_norm": 1.0234375,
      "learning_rate": 2.0736947549721496e-05,
      "loss": 1.2851,
      "step": 52771
    },
    {
      "epoch": 4.540308009980212,
      "grad_norm": 1.296875,
      "learning_rate": 2.072924151259359e-05,
      "loss": 1.3968,
      "step": 52772
    },
    {
      "epoch": 4.540394046287533,
      "grad_norm": 1.1328125,
      "learning_rate": 2.0721536877239487e-05,
      "loss": 0.5759,
      "step": 52773
    },
    {
      "epoch": 4.540480082594855,
      "grad_norm": 1.203125,
      "learning_rate": 2.0713833643681623e-05,
      "loss": 1.7346,
      "step": 52774
    },
    {
      "epoch": 4.540566118902177,
      "grad_norm": 1.09375,
      "learning_rate": 2.0706131811942584e-05,
      "loss": 0.7732,
      "step": 52775
    },
    {
      "epoch": 4.540652155209498,
      "grad_norm": 0.90234375,
      "learning_rate": 2.0698431382044914e-05,
      "loss": 1.2084,
      "step": 52776
    },
    {
      "epoch": 4.54073819151682,
      "grad_norm": 0.9609375,
      "learning_rate": 2.0690732354011087e-05,
      "loss": 0.7436,
      "step": 52777
    },
    {
      "epoch": 4.5408242278241415,
      "grad_norm": 1.1875,
      "learning_rate": 2.0683034727863704e-05,
      "loss": 1.4126,
      "step": 52778
    },
    {
      "epoch": 4.540910264131464,
      "grad_norm": 0.9765625,
      "learning_rate": 2.0675338503625186e-05,
      "loss": 0.5152,
      "step": 52779
    },
    {
      "epoch": 4.540996300438785,
      "grad_norm": 0.875,
      "learning_rate": 2.0667643681318072e-05,
      "loss": 0.4724,
      "step": 52780
    },
    {
      "epoch": 4.541082336746107,
      "grad_norm": 1.1875,
      "learning_rate": 2.0659950260964954e-05,
      "loss": 1.0933,
      "step": 52781
    },
    {
      "epoch": 4.541168373053429,
      "grad_norm": 0.81640625,
      "learning_rate": 2.065225824258815e-05,
      "loss": 1.034,
      "step": 52782
    },
    {
      "epoch": 4.54125440936075,
      "grad_norm": 0.8984375,
      "learning_rate": 2.0644567626210308e-05,
      "loss": 1.0479,
      "step": 52783
    },
    {
      "epoch": 4.541340445668072,
      "grad_norm": 1.1796875,
      "learning_rate": 2.0636878411853855e-05,
      "loss": 1.3891,
      "step": 52784
    },
    {
      "epoch": 4.541426481975393,
      "grad_norm": 1.1875,
      "learning_rate": 2.0629190599541326e-05,
      "loss": 0.9538,
      "step": 52785
    },
    {
      "epoch": 4.541512518282715,
      "grad_norm": 0.9296875,
      "learning_rate": 2.0621504189295204e-05,
      "loss": 1.0167,
      "step": 52786
    },
    {
      "epoch": 4.541598554590037,
      "grad_norm": 1.0625,
      "learning_rate": 2.0613819181137916e-05,
      "loss": 0.9984,
      "step": 52787
    },
    {
      "epoch": 4.541684590897359,
      "grad_norm": 1.046875,
      "learning_rate": 2.0606135575091944e-05,
      "loss": 1.0534,
      "step": 52788
    },
    {
      "epoch": 4.541770627204681,
      "grad_norm": 7.1875,
      "learning_rate": 2.0598453371179827e-05,
      "loss": 0.7315,
      "step": 52789
    },
    {
      "epoch": 4.541856663512002,
      "grad_norm": 17.375,
      "learning_rate": 2.0590772569423987e-05,
      "loss": 0.6747,
      "step": 52790
    },
    {
      "epoch": 4.541942699819324,
      "grad_norm": 1.1328125,
      "learning_rate": 2.0583093169846857e-05,
      "loss": 0.6424,
      "step": 52791
    },
    {
      "epoch": 4.542028736126645,
      "grad_norm": 3.796875,
      "learning_rate": 2.0575415172471023e-05,
      "loss": 1.0324,
      "step": 52792
    },
    {
      "epoch": 4.542114772433967,
      "grad_norm": 6.15625,
      "learning_rate": 2.0567738577318806e-05,
      "loss": 1.4781,
      "step": 52793
    },
    {
      "epoch": 4.542200808741288,
      "grad_norm": 43.0,
      "learning_rate": 2.0560063384412685e-05,
      "loss": 0.6893,
      "step": 52794
    },
    {
      "epoch": 4.5422868450486105,
      "grad_norm": 656.0,
      "learning_rate": 2.0552389593775202e-05,
      "loss": 1.4783,
      "step": 52795
    },
    {
      "epoch": 4.5423728813559325,
      "grad_norm": 0.9140625,
      "learning_rate": 2.0544717205428664e-05,
      "loss": 0.8272,
      "step": 52796
    },
    {
      "epoch": 4.542458917663254,
      "grad_norm": 0.84765625,
      "learning_rate": 2.053704621939567e-05,
      "loss": 0.6435,
      "step": 52797
    },
    {
      "epoch": 4.542544953970576,
      "grad_norm": 1.4375,
      "learning_rate": 2.0529376635698482e-05,
      "loss": 1.1696,
      "step": 52798
    },
    {
      "epoch": 4.542630990277897,
      "grad_norm": 1.2109375,
      "learning_rate": 2.0521708454359632e-05,
      "loss": 0.8778,
      "step": 52799
    },
    {
      "epoch": 4.542717026585219,
      "grad_norm": 1.0703125,
      "learning_rate": 2.051404167540155e-05,
      "loss": 1.0594,
      "step": 52800
    },
    {
      "epoch": 4.54280306289254,
      "grad_norm": 22.125,
      "learning_rate": 2.050637629884661e-05,
      "loss": 0.9506,
      "step": 52801
    },
    {
      "epoch": 4.542889099199862,
      "grad_norm": 1.015625,
      "learning_rate": 2.0498712324717284e-05,
      "loss": 0.7042,
      "step": 52802
    },
    {
      "epoch": 4.542975135507184,
      "grad_norm": 1.3046875,
      "learning_rate": 2.0491049753036005e-05,
      "loss": 0.8658,
      "step": 52803
    },
    {
      "epoch": 4.5430611718145055,
      "grad_norm": 1.078125,
      "learning_rate": 2.0483388583825036e-05,
      "loss": 1.4012,
      "step": 52804
    },
    {
      "epoch": 4.543147208121828,
      "grad_norm": 0.9609375,
      "learning_rate": 2.0475728817106966e-05,
      "loss": 0.3967,
      "step": 52805
    },
    {
      "epoch": 4.543233244429149,
      "grad_norm": 0.94921875,
      "learning_rate": 2.0468070452904162e-05,
      "loss": 1.1607,
      "step": 52806
    },
    {
      "epoch": 4.543319280736471,
      "grad_norm": 1.2578125,
      "learning_rate": 2.046041349123895e-05,
      "loss": 1.2265,
      "step": 52807
    },
    {
      "epoch": 4.543405317043792,
      "grad_norm": 1.2265625,
      "learning_rate": 2.0452757932133803e-05,
      "loss": 1.1633,
      "step": 52808
    },
    {
      "epoch": 4.543491353351114,
      "grad_norm": 1.484375,
      "learning_rate": 2.0445103775610983e-05,
      "loss": 0.9792,
      "step": 52809
    },
    {
      "epoch": 4.543577389658436,
      "grad_norm": 0.7265625,
      "learning_rate": 2.0437451021692976e-05,
      "loss": 0.9695,
      "step": 52810
    },
    {
      "epoch": 4.543663425965757,
      "grad_norm": 1.203125,
      "learning_rate": 2.042979967040226e-05,
      "loss": 0.8893,
      "step": 52811
    },
    {
      "epoch": 4.543749462273079,
      "grad_norm": 1.328125,
      "learning_rate": 2.042214972176104e-05,
      "loss": 1.7785,
      "step": 52812
    },
    {
      "epoch": 4.543835498580401,
      "grad_norm": 1.015625,
      "learning_rate": 2.0414501175791746e-05,
      "loss": 0.7365,
      "step": 52813
    },
    {
      "epoch": 4.543921534887723,
      "grad_norm": 132.0,
      "learning_rate": 2.040685403251674e-05,
      "loss": 1.2841,
      "step": 52814
    },
    {
      "epoch": 4.544007571195044,
      "grad_norm": 48.0,
      "learning_rate": 2.0399208291958404e-05,
      "loss": 0.8783,
      "step": 52815
    },
    {
      "epoch": 4.544093607502366,
      "grad_norm": 1.1640625,
      "learning_rate": 2.0391563954139215e-05,
      "loss": 0.6742,
      "step": 52816
    },
    {
      "epoch": 4.544179643809688,
      "grad_norm": 1.6875,
      "learning_rate": 2.038392101908132e-05,
      "loss": 1.0297,
      "step": 52817
    },
    {
      "epoch": 4.544265680117009,
      "grad_norm": 0.99609375,
      "learning_rate": 2.03762794868072e-05,
      "loss": 0.9843,
      "step": 52818
    },
    {
      "epoch": 4.544351716424331,
      "grad_norm": 0.9453125,
      "learning_rate": 2.0368639357339236e-05,
      "loss": 1.1314,
      "step": 52819
    },
    {
      "epoch": 4.544437752731652,
      "grad_norm": 1.1796875,
      "learning_rate": 2.036100063069968e-05,
      "loss": 0.8086,
      "step": 52820
    },
    {
      "epoch": 4.5445237890389745,
      "grad_norm": 1.1953125,
      "learning_rate": 2.0353363306910844e-05,
      "loss": 1.0139,
      "step": 52821
    },
    {
      "epoch": 4.544609825346296,
      "grad_norm": 1.90625,
      "learning_rate": 2.0345727385995273e-05,
      "loss": 1.7531,
      "step": 52822
    },
    {
      "epoch": 4.544695861653618,
      "grad_norm": 1.484375,
      "learning_rate": 2.0338092867975057e-05,
      "loss": 1.4123,
      "step": 52823
    },
    {
      "epoch": 4.54478189796094,
      "grad_norm": 1.6640625,
      "learning_rate": 2.0330459752872732e-05,
      "loss": 0.7657,
      "step": 52824
    },
    {
      "epoch": 4.544867934268261,
      "grad_norm": 1.1015625,
      "learning_rate": 2.0322828040710394e-05,
      "loss": 0.7493,
      "step": 52825
    },
    {
      "epoch": 4.544953970575583,
      "grad_norm": 0.828125,
      "learning_rate": 2.0315197731510525e-05,
      "loss": 1.02,
      "step": 52826
    },
    {
      "epoch": 4.545040006882904,
      "grad_norm": 6.34375,
      "learning_rate": 2.0307568825295498e-05,
      "loss": 0.9049,
      "step": 52827
    },
    {
      "epoch": 4.545126043190226,
      "grad_norm": 1.140625,
      "learning_rate": 2.0299941322087457e-05,
      "loss": 1.3318,
      "step": 52828
    },
    {
      "epoch": 4.545212079497548,
      "grad_norm": 5.25,
      "learning_rate": 2.0292315221908775e-05,
      "loss": 1.2978,
      "step": 52829
    },
    {
      "epoch": 4.5452981158048695,
      "grad_norm": 1.5078125,
      "learning_rate": 2.0284690524781822e-05,
      "loss": 1.4349,
      "step": 52830
    },
    {
      "epoch": 4.545384152112192,
      "grad_norm": 1.1953125,
      "learning_rate": 2.0277067230728864e-05,
      "loss": 0.87,
      "step": 52831
    },
    {
      "epoch": 4.545470188419513,
      "grad_norm": 1.453125,
      "learning_rate": 2.0269445339772097e-05,
      "loss": 1.0214,
      "step": 52832
    },
    {
      "epoch": 4.545556224726835,
      "grad_norm": 0.9296875,
      "learning_rate": 2.0261824851934007e-05,
      "loss": 1.1093,
      "step": 52833
    },
    {
      "epoch": 4.545642261034156,
      "grad_norm": 1.2421875,
      "learning_rate": 2.0254205767236688e-05,
      "loss": 1.4735,
      "step": 52834
    },
    {
      "epoch": 4.545728297341478,
      "grad_norm": 1.1640625,
      "learning_rate": 2.0246588085702567e-05,
      "loss": 0.5677,
      "step": 52835
    },
    {
      "epoch": 4.5458143336488,
      "grad_norm": 4.21875,
      "learning_rate": 2.0238971807353846e-05,
      "loss": 0.662,
      "step": 52836
    },
    {
      "epoch": 4.545900369956121,
      "grad_norm": 1.5546875,
      "learning_rate": 2.023135693221284e-05,
      "loss": 0.7556,
      "step": 52837
    },
    {
      "epoch": 4.545986406263443,
      "grad_norm": 1.3359375,
      "learning_rate": 2.0223743460301812e-05,
      "loss": 2.0357,
      "step": 52838
    },
    {
      "epoch": 4.546072442570765,
      "grad_norm": 0.921875,
      "learning_rate": 2.0216131391642965e-05,
      "loss": 1.1916,
      "step": 52839
    },
    {
      "epoch": 4.546158478878087,
      "grad_norm": 0.859375,
      "learning_rate": 2.0208520726258673e-05,
      "loss": 0.6695,
      "step": 52840
    },
    {
      "epoch": 4.546244515185409,
      "grad_norm": 1.0234375,
      "learning_rate": 2.020091146417119e-05,
      "loss": 1.319,
      "step": 52841
    },
    {
      "epoch": 4.54633055149273,
      "grad_norm": 0.9765625,
      "learning_rate": 2.0193303605402613e-05,
      "loss": 1.468,
      "step": 52842
    },
    {
      "epoch": 4.546416587800052,
      "grad_norm": 0.91015625,
      "learning_rate": 2.018569714997537e-05,
      "loss": 0.8037,
      "step": 52843
    },
    {
      "epoch": 4.546502624107373,
      "grad_norm": 1.0234375,
      "learning_rate": 2.017809209791166e-05,
      "loss": 1.2025,
      "step": 52844
    },
    {
      "epoch": 4.546588660414695,
      "grad_norm": 1.1875,
      "learning_rate": 2.0170488449233637e-05,
      "loss": 0.7491,
      "step": 52845
    },
    {
      "epoch": 4.546674696722016,
      "grad_norm": 79.5,
      "learning_rate": 2.016288620396367e-05,
      "loss": 1.4176,
      "step": 52846
    },
    {
      "epoch": 4.5467607330293385,
      "grad_norm": 0.984375,
      "learning_rate": 2.0155285362123965e-05,
      "loss": 1.0653,
      "step": 52847
    },
    {
      "epoch": 4.5468467693366605,
      "grad_norm": 1.015625,
      "learning_rate": 2.014768592373667e-05,
      "loss": 0.7463,
      "step": 52848
    },
    {
      "epoch": 4.546932805643982,
      "grad_norm": 1.296875,
      "learning_rate": 2.0140087888824098e-05,
      "loss": 1.1094,
      "step": 52849
    },
    {
      "epoch": 4.547018841951304,
      "grad_norm": 1.0078125,
      "learning_rate": 2.013249125740835e-05,
      "loss": 1.2611,
      "step": 52850
    },
    {
      "epoch": 4.547104878258625,
      "grad_norm": 2.234375,
      "learning_rate": 2.0124896029511787e-05,
      "loss": 0.6148,
      "step": 52851
    },
    {
      "epoch": 4.547190914565947,
      "grad_norm": 0.9453125,
      "learning_rate": 2.0117302205156672e-05,
      "loss": 1.2752,
      "step": 52852
    },
    {
      "epoch": 4.547276950873268,
      "grad_norm": 1.2265625,
      "learning_rate": 2.0109709784364994e-05,
      "loss": 0.7581,
      "step": 52853
    },
    {
      "epoch": 4.54736298718059,
      "grad_norm": 1.3984375,
      "learning_rate": 2.010211876715917e-05,
      "loss": 1.2891,
      "step": 52854
    },
    {
      "epoch": 4.547449023487912,
      "grad_norm": 1.25,
      "learning_rate": 2.00945291535613e-05,
      "loss": 1.6425,
      "step": 52855
    },
    {
      "epoch": 4.5475350597952335,
      "grad_norm": 1.4921875,
      "learning_rate": 2.0086940943593536e-05,
      "loss": 1.2581,
      "step": 52856
    },
    {
      "epoch": 4.547621096102556,
      "grad_norm": 1.0234375,
      "learning_rate": 2.0079354137278183e-05,
      "loss": 1.604,
      "step": 52857
    },
    {
      "epoch": 4.547707132409877,
      "grad_norm": 1.203125,
      "learning_rate": 2.0071768734637342e-05,
      "loss": 1.7514,
      "step": 52858
    },
    {
      "epoch": 4.547793168717199,
      "grad_norm": 1.25,
      "learning_rate": 2.0064184735693268e-05,
      "loss": 1.1103,
      "step": 52859
    },
    {
      "epoch": 4.54787920502452,
      "grad_norm": 0.93359375,
      "learning_rate": 2.0056602140468118e-05,
      "loss": 1.1728,
      "step": 52860
    },
    {
      "epoch": 4.547965241331842,
      "grad_norm": 9.875,
      "learning_rate": 2.004902094898403e-05,
      "loss": 1.1273,
      "step": 52861
    },
    {
      "epoch": 4.548051277639164,
      "grad_norm": 1.0390625,
      "learning_rate": 2.0041441161263218e-05,
      "loss": 1.0123,
      "step": 52862
    },
    {
      "epoch": 4.548137313946485,
      "grad_norm": 0.91796875,
      "learning_rate": 2.0033862777327883e-05,
      "loss": 1.1646,
      "step": 52863
    },
    {
      "epoch": 4.548223350253807,
      "grad_norm": 1.0859375,
      "learning_rate": 2.0026285797200116e-05,
      "loss": 1.3373,
      "step": 52864
    },
    {
      "epoch": 4.548309386561129,
      "grad_norm": 1.2109375,
      "learning_rate": 2.0018710220902126e-05,
      "loss": 0.8244,
      "step": 52865
    },
    {
      "epoch": 4.548395422868451,
      "grad_norm": 1.0859375,
      "learning_rate": 2.001113604845606e-05,
      "loss": 1.5476,
      "step": 52866
    },
    {
      "epoch": 4.548481459175772,
      "grad_norm": 223.0,
      "learning_rate": 2.0003563279884007e-05,
      "loss": 1.1641,
      "step": 52867
    },
    {
      "epoch": 4.548567495483094,
      "grad_norm": 0.86328125,
      "learning_rate": 1.999599191520829e-05,
      "loss": 0.4404,
      "step": 52868
    },
    {
      "epoch": 4.548653531790416,
      "grad_norm": 8.3125,
      "learning_rate": 1.9988421954450885e-05,
      "loss": 1.2931,
      "step": 52869
    },
    {
      "epoch": 4.548739568097737,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9980853397633947e-05,
      "loss": 0.8599,
      "step": 52870
    },
    {
      "epoch": 4.548825604405059,
      "grad_norm": 0.96875,
      "learning_rate": 1.997328624477973e-05,
      "loss": 1.6949,
      "step": 52871
    },
    {
      "epoch": 4.54891164071238,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9965720495910224e-05,
      "loss": 0.5315,
      "step": 52872
    },
    {
      "epoch": 4.5489976770197025,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9958156151047622e-05,
      "loss": 0.9814,
      "step": 52873
    },
    {
      "epoch": 4.549083713327024,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9950593210214142e-05,
      "loss": 0.6907,
      "step": 52874
    },
    {
      "epoch": 4.549169749634346,
      "grad_norm": 1.34375,
      "learning_rate": 1.9943031673431698e-05,
      "loss": 1.0842,
      "step": 52875
    },
    {
      "epoch": 4.549255785941668,
      "grad_norm": 1.3671875,
      "learning_rate": 1.993547154072256e-05,
      "loss": 1.5332,
      "step": 52876
    },
    {
      "epoch": 4.549341822248989,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9927912812108873e-05,
      "loss": 1.0626,
      "step": 52877
    },
    {
      "epoch": 4.549427858556311,
      "grad_norm": 1.1328125,
      "learning_rate": 1.992035548761256e-05,
      "loss": 0.7201,
      "step": 52878
    },
    {
      "epoch": 4.549513894863632,
      "grad_norm": 1.296875,
      "learning_rate": 1.9912799567255946e-05,
      "loss": 1.5317,
      "step": 52879
    },
    {
      "epoch": 4.549599931170954,
      "grad_norm": 0.85546875,
      "learning_rate": 1.990524505106095e-05,
      "loss": 0.9491,
      "step": 52880
    },
    {
      "epoch": 4.5496859674782755,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9897691939049724e-05,
      "loss": 1.5845,
      "step": 52881
    },
    {
      "epoch": 4.5497720037855975,
      "grad_norm": 1.125,
      "learning_rate": 1.9890140231244468e-05,
      "loss": 0.9683,
      "step": 52882
    },
    {
      "epoch": 4.54985804009292,
      "grad_norm": 1.1328125,
      "learning_rate": 1.988258992766706e-05,
      "loss": 0.9085,
      "step": 52883
    },
    {
      "epoch": 4.549944076400241,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9875041028339758e-05,
      "loss": 1.1137,
      "step": 52884
    },
    {
      "epoch": 4.550030112707563,
      "grad_norm": 45.75,
      "learning_rate": 1.9867493533284598e-05,
      "loss": 1.2476,
      "step": 52885
    },
    {
      "epoch": 4.550116149014884,
      "grad_norm": 1.1875,
      "learning_rate": 1.985994744252362e-05,
      "loss": 1.7434,
      "step": 52886
    },
    {
      "epoch": 4.550202185322206,
      "grad_norm": 832.0,
      "learning_rate": 1.9852402756078856e-05,
      "loss": 0.6441,
      "step": 52887
    },
    {
      "epoch": 4.550288221629527,
      "grad_norm": 1.2109375,
      "learning_rate": 1.9844859473972466e-05,
      "loss": 1.5539,
      "step": 52888
    },
    {
      "epoch": 4.550374257936849,
      "grad_norm": 1.140625,
      "learning_rate": 1.983731759622648e-05,
      "loss": 1.4119,
      "step": 52889
    },
    {
      "epoch": 4.550460294244171,
      "grad_norm": 1.0625,
      "learning_rate": 1.982977712286299e-05,
      "loss": 0.7456,
      "step": 52890
    },
    {
      "epoch": 4.550546330551493,
      "grad_norm": 1.0625,
      "learning_rate": 1.9822238053903986e-05,
      "loss": 0.8117,
      "step": 52891
    },
    {
      "epoch": 4.550632366858815,
      "grad_norm": 1.2421875,
      "learning_rate": 1.9814700389371554e-05,
      "loss": 1.742,
      "step": 52892
    },
    {
      "epoch": 4.550718403166136,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9807164129287735e-05,
      "loss": 1.1602,
      "step": 52893
    },
    {
      "epoch": 4.550804439473458,
      "grad_norm": 1.125,
      "learning_rate": 1.979962927367457e-05,
      "loss": 1.754,
      "step": 52894
    },
    {
      "epoch": 4.550890475780779,
      "grad_norm": 31.375,
      "learning_rate": 1.9792095822554145e-05,
      "loss": 0.6751,
      "step": 52895
    },
    {
      "epoch": 4.550976512088101,
      "grad_norm": 1.359375,
      "learning_rate": 1.978456377594834e-05,
      "loss": 1.4736,
      "step": 52896
    },
    {
      "epoch": 4.551062548395423,
      "grad_norm": 1.421875,
      "learning_rate": 1.9777033133879298e-05,
      "loss": 1.5311,
      "step": 52897
    },
    {
      "epoch": 4.551148584702744,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9769503896369167e-05,
      "loss": 1.0758,
      "step": 52898
    },
    {
      "epoch": 4.5512346210100665,
      "grad_norm": 24.875,
      "learning_rate": 1.9761976063439714e-05,
      "loss": 1.7307,
      "step": 52899
    },
    {
      "epoch": 4.551320657317388,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9754449635113082e-05,
      "loss": 1.5587,
      "step": 52900
    },
    {
      "epoch": 4.55140669362471,
      "grad_norm": 1.296875,
      "learning_rate": 1.9746924611411364e-05,
      "loss": 1.7696,
      "step": 52901
    },
    {
      "epoch": 4.551492729932031,
      "grad_norm": 8.125,
      "learning_rate": 1.9739400992356384e-05,
      "loss": 1.1344,
      "step": 52902
    },
    {
      "epoch": 4.551578766239353,
      "grad_norm": 2.125,
      "learning_rate": 1.973187877797028e-05,
      "loss": 1.389,
      "step": 52903
    },
    {
      "epoch": 4.551664802546675,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9724357968275098e-05,
      "loss": 0.8946,
      "step": 52904
    },
    {
      "epoch": 4.551750838853996,
      "grad_norm": 1.078125,
      "learning_rate": 1.9716838563292703e-05,
      "loss": 0.6158,
      "step": 52905
    },
    {
      "epoch": 4.551836875161318,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9709320563045198e-05,
      "loss": 0.9211,
      "step": 52906
    },
    {
      "epoch": 4.5519229114686395,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9701803967554443e-05,
      "loss": 1.6096,
      "step": 52907
    },
    {
      "epoch": 4.5520089477759615,
      "grad_norm": 1.109375,
      "learning_rate": 1.9694288776842484e-05,
      "loss": 0.9944,
      "step": 52908
    },
    {
      "epoch": 4.552094984083283,
      "grad_norm": 1.9140625,
      "learning_rate": 1.9686774990931412e-05,
      "loss": 0.9892,
      "step": 52909
    },
    {
      "epoch": 4.552181020390605,
      "grad_norm": 1.15625,
      "learning_rate": 1.967926260984304e-05,
      "loss": 0.8545,
      "step": 52910
    },
    {
      "epoch": 4.552267056697927,
      "grad_norm": 5.9375,
      "learning_rate": 1.967175163359941e-05,
      "loss": 1.7504,
      "step": 52911
    },
    {
      "epoch": 4.552353093005248,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9664242062222558e-05,
      "loss": 0.7198,
      "step": 52912
    },
    {
      "epoch": 4.55243912931257,
      "grad_norm": 1.25,
      "learning_rate": 1.9656733895734246e-05,
      "loss": 1.08,
      "step": 52913
    },
    {
      "epoch": 4.552525165619891,
      "grad_norm": 1.4453125,
      "learning_rate": 1.964922713415662e-05,
      "loss": 1.0667,
      "step": 52914
    },
    {
      "epoch": 4.552611201927213,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9641721777511667e-05,
      "loss": 0.7428,
      "step": 52915
    },
    {
      "epoch": 4.5526972382345345,
      "grad_norm": 1.0,
      "learning_rate": 1.9634217825821144e-05,
      "loss": 1.3973,
      "step": 52916
    },
    {
      "epoch": 4.552783274541857,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9626715279107145e-05,
      "loss": 1.3965,
      "step": 52917
    },
    {
      "epoch": 4.552869310849179,
      "grad_norm": 1.0,
      "learning_rate": 1.961921413739165e-05,
      "loss": 0.9176,
      "step": 52918
    },
    {
      "epoch": 4.5529553471565,
      "grad_norm": 1.171875,
      "learning_rate": 1.9611714400696422e-05,
      "loss": 0.5847,
      "step": 52919
    },
    {
      "epoch": 4.553041383463822,
      "grad_norm": 320.0,
      "learning_rate": 1.9604216069043558e-05,
      "loss": 1.1197,
      "step": 52920
    },
    {
      "epoch": 4.553127419771143,
      "grad_norm": 1.375,
      "learning_rate": 1.9596719142454923e-05,
      "loss": 0.9361,
      "step": 52921
    },
    {
      "epoch": 4.553213456078465,
      "grad_norm": 1.4453125,
      "learning_rate": 1.958922362095239e-05,
      "loss": 1.0027,
      "step": 52922
    },
    {
      "epoch": 4.553299492385786,
      "grad_norm": 1.2734375,
      "learning_rate": 1.9581729504558054e-05,
      "loss": 1.8674,
      "step": 52923
    },
    {
      "epoch": 4.553385528693108,
      "grad_norm": 0.78125,
      "learning_rate": 1.9574236793293564e-05,
      "loss": 0.6886,
      "step": 52924
    },
    {
      "epoch": 4.5534715650004305,
      "grad_norm": 0.7265625,
      "learning_rate": 1.9566745487181127e-05,
      "loss": 0.6883,
      "step": 52925
    },
    {
      "epoch": 4.553557601307752,
      "grad_norm": 0.9140625,
      "learning_rate": 1.955925558624244e-05,
      "loss": 0.9861,
      "step": 52926
    },
    {
      "epoch": 4.553643637615074,
      "grad_norm": 1.1015625,
      "learning_rate": 1.955176709049955e-05,
      "loss": 1.2404,
      "step": 52927
    },
    {
      "epoch": 4.553729673922395,
      "grad_norm": 1.1953125,
      "learning_rate": 1.9544279999974268e-05,
      "loss": 1.2681,
      "step": 52928
    },
    {
      "epoch": 4.553815710229717,
      "grad_norm": 1.1640625,
      "learning_rate": 1.953679431468852e-05,
      "loss": 1.0843,
      "step": 52929
    },
    {
      "epoch": 4.553901746537038,
      "grad_norm": 1.109375,
      "learning_rate": 1.9529310034664237e-05,
      "loss": 1.3004,
      "step": 52930
    },
    {
      "epoch": 4.55398778284436,
      "grad_norm": 1.28125,
      "learning_rate": 1.952182715992329e-05,
      "loss": 1.8222,
      "step": 52931
    },
    {
      "epoch": 4.554073819151682,
      "grad_norm": 33.5,
      "learning_rate": 1.951434569048749e-05,
      "loss": 1.1731,
      "step": 52932
    },
    {
      "epoch": 4.5541598554590035,
      "grad_norm": 2.453125,
      "learning_rate": 1.9506865626378766e-05,
      "loss": 1.2209,
      "step": 52933
    },
    {
      "epoch": 4.5542458917663255,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9499386967619102e-05,
      "loss": 0.8625,
      "step": 52934
    },
    {
      "epoch": 4.554331928073647,
      "grad_norm": 1.2421875,
      "learning_rate": 1.9491909714230206e-05,
      "loss": 1.4529,
      "step": 52935
    },
    {
      "epoch": 4.554417964380969,
      "grad_norm": 1.25,
      "learning_rate": 1.948443386623411e-05,
      "loss": 1.4359,
      "step": 52936
    },
    {
      "epoch": 4.55450400068829,
      "grad_norm": 35.5,
      "learning_rate": 1.9476959423652464e-05,
      "loss": 1.1391,
      "step": 52937
    },
    {
      "epoch": 4.554590036995612,
      "grad_norm": 104.0,
      "learning_rate": 1.9469486386507306e-05,
      "loss": 0.5834,
      "step": 52938
    },
    {
      "epoch": 4.554676073302934,
      "grad_norm": 3.21875,
      "learning_rate": 1.946201475482051e-05,
      "loss": 0.8166,
      "step": 52939
    },
    {
      "epoch": 4.554762109610255,
      "grad_norm": 1.34375,
      "learning_rate": 1.9454544528613783e-05,
      "loss": 1.0607,
      "step": 52940
    },
    {
      "epoch": 4.554848145917577,
      "grad_norm": 1.34375,
      "learning_rate": 1.9447075707909047e-05,
      "loss": 1.0919,
      "step": 52941
    },
    {
      "epoch": 4.5549341822248985,
      "grad_norm": 1.2421875,
      "learning_rate": 1.943960829272823e-05,
      "loss": 0.8223,
      "step": 52942
    },
    {
      "epoch": 4.555020218532221,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9432142283092978e-05,
      "loss": 1.1527,
      "step": 52943
    },
    {
      "epoch": 4.555106254839542,
      "grad_norm": 0.9453125,
      "learning_rate": 1.942467767902528e-05,
      "loss": 1.0033,
      "step": 52944
    },
    {
      "epoch": 4.555192291146864,
      "grad_norm": 0.890625,
      "learning_rate": 1.9417214480547008e-05,
      "loss": 0.471,
      "step": 52945
    },
    {
      "epoch": 4.555278327454186,
      "grad_norm": 0.94140625,
      "learning_rate": 1.940975268767986e-05,
      "loss": 1.33,
      "step": 52946
    },
    {
      "epoch": 4.555364363761507,
      "grad_norm": 1.1953125,
      "learning_rate": 1.940229230044571e-05,
      "loss": 1.6346,
      "step": 52947
    },
    {
      "epoch": 4.555450400068829,
      "grad_norm": 92.0,
      "learning_rate": 1.9394833318866434e-05,
      "loss": 0.988,
      "step": 52948
    },
    {
      "epoch": 4.55553643637615,
      "grad_norm": 1.1015625,
      "learning_rate": 1.938737574296373e-05,
      "loss": 0.9651,
      "step": 52949
    },
    {
      "epoch": 4.555622472683472,
      "grad_norm": 1.8359375,
      "learning_rate": 1.9379919572759585e-05,
      "loss": 1.1549,
      "step": 52950
    },
    {
      "epoch": 4.5557085089907945,
      "grad_norm": 1.078125,
      "learning_rate": 1.9372464808275592e-05,
      "loss": 1.555,
      "step": 52951
    },
    {
      "epoch": 4.555794545298116,
      "grad_norm": 1.140625,
      "learning_rate": 1.9365011449533675e-05,
      "loss": 1.2033,
      "step": 52952
    },
    {
      "epoch": 4.555880581605438,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9357559496555654e-05,
      "loss": 1.3979,
      "step": 52953
    },
    {
      "epoch": 4.555966617912759,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9350108949363287e-05,
      "loss": 0.9501,
      "step": 52954
    },
    {
      "epoch": 4.556052654220081,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9342659807978337e-05,
      "loss": 1.7708,
      "step": 52955
    },
    {
      "epoch": 4.556138690527403,
      "grad_norm": 1.1171875,
      "learning_rate": 1.933521207242267e-05,
      "loss": 0.7212,
      "step": 52956
    },
    {
      "epoch": 4.556224726834724,
      "grad_norm": 5.5,
      "learning_rate": 1.9327765742717996e-05,
      "loss": 1.0617,
      "step": 52957
    },
    {
      "epoch": 4.556310763142046,
      "grad_norm": 1.5625,
      "learning_rate": 1.9320320818886127e-05,
      "loss": 1.2008,
      "step": 52958
    },
    {
      "epoch": 4.5563967994493675,
      "grad_norm": 1.140625,
      "learning_rate": 1.9312877300948885e-05,
      "loss": 1.1633,
      "step": 52959
    },
    {
      "epoch": 4.5564828357566896,
      "grad_norm": 1.3203125,
      "learning_rate": 1.9305435188927913e-05,
      "loss": 0.5851,
      "step": 52960
    },
    {
      "epoch": 4.556568872064011,
      "grad_norm": 40.25,
      "learning_rate": 1.9297994482845027e-05,
      "loss": 0.6965,
      "step": 52961
    },
    {
      "epoch": 4.556654908371333,
      "grad_norm": 1.453125,
      "learning_rate": 1.9290555182722047e-05,
      "loss": 0.4953,
      "step": 52962
    },
    {
      "epoch": 4.556740944678655,
      "grad_norm": 15.125,
      "learning_rate": 1.9283117288580732e-05,
      "loss": 1.7097,
      "step": 52963
    },
    {
      "epoch": 4.556826980985976,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9275680800442785e-05,
      "loss": 0.9006,
      "step": 52964
    },
    {
      "epoch": 4.556913017293298,
      "grad_norm": 1.5234375,
      "learning_rate": 1.926824571832997e-05,
      "loss": 1.1123,
      "step": 52965
    },
    {
      "epoch": 4.556999053600619,
      "grad_norm": 5.0625,
      "learning_rate": 1.9260812042264098e-05,
      "loss": 1.1699,
      "step": 52966
    },
    {
      "epoch": 4.557085089907941,
      "grad_norm": 1.28125,
      "learning_rate": 1.925337977226682e-05,
      "loss": 1.3247,
      "step": 52967
    },
    {
      "epoch": 4.5571711262152625,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9245948908359846e-05,
      "loss": 0.89,
      "step": 52968
    },
    {
      "epoch": 4.557257162522585,
      "grad_norm": 1.3125,
      "learning_rate": 1.923851945056504e-05,
      "loss": 0.6887,
      "step": 52969
    },
    {
      "epoch": 4.557343198829907,
      "grad_norm": 1.0,
      "learning_rate": 1.9231091398904056e-05,
      "loss": 0.5417,
      "step": 52970
    },
    {
      "epoch": 4.557429235137228,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9223664753398597e-05,
      "loss": 1.586,
      "step": 52971
    },
    {
      "epoch": 4.55751527144455,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9216239514070422e-05,
      "loss": 1.1603,
      "step": 52972
    },
    {
      "epoch": 4.557601307751871,
      "grad_norm": 17.75,
      "learning_rate": 1.9208815680941237e-05,
      "loss": 1.0442,
      "step": 52973
    },
    {
      "epoch": 4.557687344059193,
      "grad_norm": 2.375,
      "learning_rate": 1.920139325403275e-05,
      "loss": 1.1833,
      "step": 52974
    },
    {
      "epoch": 4.557773380366514,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9193972233366718e-05,
      "loss": 0.6514,
      "step": 52975
    },
    {
      "epoch": 4.557859416673836,
      "grad_norm": 1.0,
      "learning_rate": 1.9186552618964738e-05,
      "loss": 0.7129,
      "step": 52976
    },
    {
      "epoch": 4.5579454529811585,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9179134410848676e-05,
      "loss": 1.3752,
      "step": 52977
    },
    {
      "epoch": 4.55803148928848,
      "grad_norm": 1.6484375,
      "learning_rate": 1.9171717609040075e-05,
      "loss": 1.4026,
      "step": 52978
    },
    {
      "epoch": 4.558117525595802,
      "grad_norm": 64.0,
      "learning_rate": 1.9164302213560636e-05,
      "loss": 0.6212,
      "step": 52979
    },
    {
      "epoch": 4.558203561903123,
      "grad_norm": 1.2109375,
      "learning_rate": 1.9156888224432177e-05,
      "loss": 1.1942,
      "step": 52980
    },
    {
      "epoch": 4.558289598210445,
      "grad_norm": 1.484375,
      "learning_rate": 1.914947564167624e-05,
      "loss": 1.5806,
      "step": 52981
    },
    {
      "epoch": 4.558375634517766,
      "grad_norm": 1.53125,
      "learning_rate": 1.914206446531458e-05,
      "loss": 1.5147,
      "step": 52982
    },
    {
      "epoch": 4.558461670825088,
      "grad_norm": 1.078125,
      "learning_rate": 1.9134654695368902e-05,
      "loss": 1.9259,
      "step": 52983
    },
    {
      "epoch": 4.55854770713241,
      "grad_norm": 1.125,
      "learning_rate": 1.9127246331860805e-05,
      "loss": 1.0446,
      "step": 52984
    },
    {
      "epoch": 4.5586337434397315,
      "grad_norm": 0.875,
      "learning_rate": 1.9119839374811933e-05,
      "loss": 0.6392,
      "step": 52985
    },
    {
      "epoch": 4.558719779747054,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9112433824244102e-05,
      "loss": 1.3778,
      "step": 52986
    },
    {
      "epoch": 4.558805816054375,
      "grad_norm": 14.0,
      "learning_rate": 1.9105029680178855e-05,
      "loss": 0.8126,
      "step": 52987
    },
    {
      "epoch": 4.558891852361697,
      "grad_norm": 1.625,
      "learning_rate": 1.909762694263778e-05,
      "loss": 1.11,
      "step": 52988
    },
    {
      "epoch": 4.558977888669018,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9090225611642753e-05,
      "loss": 0.684,
      "step": 52989
    },
    {
      "epoch": 4.55906392497634,
      "grad_norm": 33.75,
      "learning_rate": 1.90828256872152e-05,
      "loss": 0.9721,
      "step": 52990
    },
    {
      "epoch": 4.559149961283662,
      "grad_norm": 0.8828125,
      "learning_rate": 1.907542716937688e-05,
      "loss": 0.8373,
      "step": 52991
    },
    {
      "epoch": 4.559235997590983,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9068030058149388e-05,
      "loss": 1.1675,
      "step": 52992
    },
    {
      "epoch": 4.559322033898305,
      "grad_norm": 21.0,
      "learning_rate": 1.906063435355432e-05,
      "loss": 1.6274,
      "step": 52993
    },
    {
      "epoch": 4.5594080702056266,
      "grad_norm": 1.109375,
      "learning_rate": 1.905324005561343e-05,
      "loss": 1.8361,
      "step": 52994
    },
    {
      "epoch": 4.559494106512949,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9045847164348263e-05,
      "loss": 1.282,
      "step": 52995
    },
    {
      "epoch": 4.55958014282027,
      "grad_norm": 1.4609375,
      "learning_rate": 1.9038455679780408e-05,
      "loss": 1.0782,
      "step": 52996
    },
    {
      "epoch": 4.559666179127592,
      "grad_norm": 1.265625,
      "learning_rate": 1.9031065601931573e-05,
      "loss": 1.1477,
      "step": 52997
    },
    {
      "epoch": 4.559752215434914,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9023676930823242e-05,
      "loss": 1.5214,
      "step": 52998
    },
    {
      "epoch": 4.559838251742235,
      "grad_norm": 6.34375,
      "learning_rate": 1.9016289666477225e-05,
      "loss": 1.296,
      "step": 52999
    },
    {
      "epoch": 4.559924288049557,
      "grad_norm": 24.125,
      "learning_rate": 1.9008903808914955e-05,
      "loss": 0.6681,
      "step": 53000
    },
    {
      "epoch": 4.560010324356878,
      "grad_norm": 21.5,
      "learning_rate": 1.9001519358158083e-05,
      "loss": 0.7813,
      "step": 53001
    },
    {
      "epoch": 4.5600963606642,
      "grad_norm": 1.0390625,
      "learning_rate": 1.8994136314228305e-05,
      "loss": 1.4038,
      "step": 53002
    },
    {
      "epoch": 4.560182396971522,
      "grad_norm": 1.1171875,
      "learning_rate": 1.8986754677147056e-05,
      "loss": 1.0753,
      "step": 53003
    },
    {
      "epoch": 4.560268433278844,
      "grad_norm": 1.2890625,
      "learning_rate": 1.897937444693598e-05,
      "loss": 0.876,
      "step": 53004
    },
    {
      "epoch": 4.560354469586166,
      "grad_norm": 1.1875,
      "learning_rate": 1.8971995623616732e-05,
      "loss": 0.5967,
      "step": 53005
    },
    {
      "epoch": 4.560440505893487,
      "grad_norm": 1.0078125,
      "learning_rate": 1.8964618207210792e-05,
      "loss": 0.8063,
      "step": 53006
    },
    {
      "epoch": 4.560526542200809,
      "grad_norm": 18.0,
      "learning_rate": 1.8957242197739866e-05,
      "loss": 1.0649,
      "step": 53007
    },
    {
      "epoch": 4.56061257850813,
      "grad_norm": 0.94921875,
      "learning_rate": 1.894986759522538e-05,
      "loss": 1.4303,
      "step": 53008
    },
    {
      "epoch": 4.560698614815452,
      "grad_norm": 1.0703125,
      "learning_rate": 1.894249439968898e-05,
      "loss": 1.2574,
      "step": 53009
    },
    {
      "epoch": 4.560784651122773,
      "grad_norm": 1.078125,
      "learning_rate": 1.8935122611152267e-05,
      "loss": 0.8469,
      "step": 53010
    },
    {
      "epoch": 4.5608706874300955,
      "grad_norm": 1.203125,
      "learning_rate": 1.8927752229636718e-05,
      "loss": 1.1747,
      "step": 53011
    },
    {
      "epoch": 4.560956723737418,
      "grad_norm": 1.203125,
      "learning_rate": 1.892038325516393e-05,
      "loss": 1.805,
      "step": 53012
    },
    {
      "epoch": 4.561042760044739,
      "grad_norm": 1.1171875,
      "learning_rate": 1.8913015687755497e-05,
      "loss": 0.8834,
      "step": 53013
    },
    {
      "epoch": 4.561128796352061,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8905649527432845e-05,
      "loss": 0.6764,
      "step": 53014
    },
    {
      "epoch": 4.561214832659382,
      "grad_norm": 1.03125,
      "learning_rate": 1.8898284774217676e-05,
      "loss": 0.7664,
      "step": 53015
    },
    {
      "epoch": 4.561300868966704,
      "grad_norm": 1.234375,
      "learning_rate": 1.8890921428131424e-05,
      "loss": 0.9127,
      "step": 53016
    },
    {
      "epoch": 4.561386905274025,
      "grad_norm": 1.0,
      "learning_rate": 1.8883559489195677e-05,
      "loss": 1.176,
      "step": 53017
    },
    {
      "epoch": 4.561472941581347,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8876198957431978e-05,
      "loss": 0.874,
      "step": 53018
    },
    {
      "epoch": 4.561558977888669,
      "grad_norm": 1.1953125,
      "learning_rate": 1.8868839832861752e-05,
      "loss": 1.3865,
      "step": 53019
    },
    {
      "epoch": 4.5616450141959906,
      "grad_norm": 1.125,
      "learning_rate": 1.886148211550659e-05,
      "loss": 0.8104,
      "step": 53020
    },
    {
      "epoch": 4.561731050503313,
      "grad_norm": 1.1796875,
      "learning_rate": 1.885412580538809e-05,
      "loss": 0.589,
      "step": 53021
    },
    {
      "epoch": 4.561817086810634,
      "grad_norm": 2.921875,
      "learning_rate": 1.884677090252762e-05,
      "loss": 0.6498,
      "step": 53022
    },
    {
      "epoch": 4.561903123117956,
      "grad_norm": 1.09375,
      "learning_rate": 1.883941740694678e-05,
      "loss": 0.9901,
      "step": 53023
    },
    {
      "epoch": 4.561989159425277,
      "grad_norm": 1.09375,
      "learning_rate": 1.8832065318667103e-05,
      "loss": 1.1691,
      "step": 53024
    },
    {
      "epoch": 4.562075195732599,
      "grad_norm": 1.984375,
      "learning_rate": 1.8824714637710016e-05,
      "loss": 1.3133,
      "step": 53025
    },
    {
      "epoch": 4.562161232039921,
      "grad_norm": 2.15625,
      "learning_rate": 1.881736536409706e-05,
      "loss": 1.2777,
      "step": 53026
    },
    {
      "epoch": 4.562247268347242,
      "grad_norm": 1.3203125,
      "learning_rate": 1.8810017497849775e-05,
      "loss": 1.2378,
      "step": 53027
    },
    {
      "epoch": 4.562333304654564,
      "grad_norm": 1.1171875,
      "learning_rate": 1.8802671038989526e-05,
      "loss": 1.3695,
      "step": 53028
    },
    {
      "epoch": 4.562419340961886,
      "grad_norm": 1.625,
      "learning_rate": 1.8795325987537913e-05,
      "loss": 0.808,
      "step": 53029
    },
    {
      "epoch": 4.562505377269208,
      "grad_norm": 187.0,
      "learning_rate": 1.878798234351642e-05,
      "loss": 0.9267,
      "step": 53030
    },
    {
      "epoch": 4.562591413576529,
      "grad_norm": 0.98828125,
      "learning_rate": 1.878064010694641e-05,
      "loss": 0.8268,
      "step": 53031
    },
    {
      "epoch": 4.562677449883851,
      "grad_norm": 18.625,
      "learning_rate": 1.8773299277849544e-05,
      "loss": 1.2945,
      "step": 53032
    },
    {
      "epoch": 4.562763486191173,
      "grad_norm": 5.59375,
      "learning_rate": 1.876595985624707e-05,
      "loss": 1.2306,
      "step": 53033
    },
    {
      "epoch": 4.562849522498494,
      "grad_norm": 0.98828125,
      "learning_rate": 1.875862184216054e-05,
      "loss": 0.6846,
      "step": 53034
    },
    {
      "epoch": 4.562935558805816,
      "grad_norm": 1.203125,
      "learning_rate": 1.875128523561154e-05,
      "loss": 0.8246,
      "step": 53035
    },
    {
      "epoch": 4.563021595113137,
      "grad_norm": 0.828125,
      "learning_rate": 1.8743950036621383e-05,
      "loss": 1.1335,
      "step": 53036
    },
    {
      "epoch": 4.5631076314204595,
      "grad_norm": 2.75,
      "learning_rate": 1.8736616245211667e-05,
      "loss": 1.1063,
      "step": 53037
    },
    {
      "epoch": 4.563193667727781,
      "grad_norm": 1.0078125,
      "learning_rate": 1.8729283861403658e-05,
      "loss": 0.7782,
      "step": 53038
    },
    {
      "epoch": 4.563279704035103,
      "grad_norm": 6.34375,
      "learning_rate": 1.8721952885218883e-05,
      "loss": 0.8566,
      "step": 53039
    },
    {
      "epoch": 4.563365740342425,
      "grad_norm": 1.3671875,
      "learning_rate": 1.871462331667889e-05,
      "loss": 1.438,
      "step": 53040
    },
    {
      "epoch": 4.563451776649746,
      "grad_norm": 30.625,
      "learning_rate": 1.8707295155804937e-05,
      "loss": 1.2295,
      "step": 53041
    },
    {
      "epoch": 4.563537812957068,
      "grad_norm": 0.98828125,
      "learning_rate": 1.8699968402618507e-05,
      "loss": 0.7333,
      "step": 53042
    },
    {
      "epoch": 4.563623849264389,
      "grad_norm": 5.1875,
      "learning_rate": 1.869264305714119e-05,
      "loss": 0.7353,
      "step": 53043
    },
    {
      "epoch": 4.563709885571711,
      "grad_norm": 1.4765625,
      "learning_rate": 1.868531911939414e-05,
      "loss": 1.4878,
      "step": 53044
    },
    {
      "epoch": 4.5637959218790325,
      "grad_norm": 1.15625,
      "learning_rate": 1.8677996589399005e-05,
      "loss": 0.7239,
      "step": 53045
    },
    {
      "epoch": 4.563881958186355,
      "grad_norm": 1.2890625,
      "learning_rate": 1.86706754671771e-05,
      "loss": 1.2646,
      "step": 53046
    },
    {
      "epoch": 4.563967994493677,
      "grad_norm": 0.93359375,
      "learning_rate": 1.8663355752749854e-05,
      "loss": 1.7312,
      "step": 53047
    },
    {
      "epoch": 4.564054030800998,
      "grad_norm": 1.515625,
      "learning_rate": 1.8656037446138752e-05,
      "loss": 0.9666,
      "step": 53048
    },
    {
      "epoch": 4.56414006710832,
      "grad_norm": 1.125,
      "learning_rate": 1.8648720547364994e-05,
      "loss": 0.9706,
      "step": 53049
    },
    {
      "epoch": 4.564226103415641,
      "grad_norm": 1.2890625,
      "learning_rate": 1.8641405056450176e-05,
      "loss": 1.1364,
      "step": 53050
    },
    {
      "epoch": 4.564312139722963,
      "grad_norm": 1.3203125,
      "learning_rate": 1.8634090973415617e-05,
      "loss": 1.7286,
      "step": 53051
    },
    {
      "epoch": 4.564398176030284,
      "grad_norm": 0.95703125,
      "learning_rate": 1.8626778298282743e-05,
      "loss": 1.4612,
      "step": 53052
    },
    {
      "epoch": 4.564484212337606,
      "grad_norm": 1.4921875,
      "learning_rate": 1.8619467031072868e-05,
      "loss": 1.6763,
      "step": 53053
    },
    {
      "epoch": 4.564570248644928,
      "grad_norm": 2.484375,
      "learning_rate": 1.861215717180753e-05,
      "loss": 1.143,
      "step": 53054
    },
    {
      "epoch": 4.56465628495225,
      "grad_norm": 1.109375,
      "learning_rate": 1.860484872050788e-05,
      "loss": 0.5879,
      "step": 53055
    },
    {
      "epoch": 4.564742321259572,
      "grad_norm": 1.3046875,
      "learning_rate": 1.8597541677195464e-05,
      "loss": 0.7824,
      "step": 53056
    },
    {
      "epoch": 4.564828357566893,
      "grad_norm": 1.390625,
      "learning_rate": 1.8590236041891694e-05,
      "loss": 1.5479,
      "step": 53057
    },
    {
      "epoch": 4.564914393874215,
      "grad_norm": 1.5703125,
      "learning_rate": 1.858293181461779e-05,
      "loss": 0.8059,
      "step": 53058
    },
    {
      "epoch": 4.565000430181536,
      "grad_norm": 0.96875,
      "learning_rate": 1.8575628995395168e-05,
      "loss": 0.4879,
      "step": 53059
    },
    {
      "epoch": 4.565086466488858,
      "grad_norm": 1.1640625,
      "learning_rate": 1.8568327584245204e-05,
      "loss": 0.8558,
      "step": 53060
    },
    {
      "epoch": 4.56517250279618,
      "grad_norm": 1.1796875,
      "learning_rate": 1.8561027581189273e-05,
      "loss": 1.45,
      "step": 53061
    },
    {
      "epoch": 4.565258539103501,
      "grad_norm": 0.9921875,
      "learning_rate": 1.8553728986248685e-05,
      "loss": 0.7225,
      "step": 53062
    },
    {
      "epoch": 4.5653445754108235,
      "grad_norm": 0.97265625,
      "learning_rate": 1.8546431799444817e-05,
      "loss": 1.235,
      "step": 53063
    },
    {
      "epoch": 4.565430611718145,
      "grad_norm": 1.0078125,
      "learning_rate": 1.853913602079893e-05,
      "loss": 0.7621,
      "step": 53064
    },
    {
      "epoch": 4.565516648025467,
      "grad_norm": 1.015625,
      "learning_rate": 1.853184165033256e-05,
      "loss": 1.1177,
      "step": 53065
    },
    {
      "epoch": 4.565602684332789,
      "grad_norm": 7.46875,
      "learning_rate": 1.85245486880668e-05,
      "loss": 1.2176,
      "step": 53066
    },
    {
      "epoch": 4.56568872064011,
      "grad_norm": 1.0546875,
      "learning_rate": 1.851725713402308e-05,
      "loss": 1.5813,
      "step": 53067
    },
    {
      "epoch": 4.565774756947432,
      "grad_norm": 1.4296875,
      "learning_rate": 1.850996698822288e-05,
      "loss": 1.0074,
      "step": 53068
    },
    {
      "epoch": 4.565860793254753,
      "grad_norm": 24.125,
      "learning_rate": 1.8502678250687242e-05,
      "loss": 1.3044,
      "step": 53069
    },
    {
      "epoch": 4.565946829562075,
      "grad_norm": 16.5,
      "learning_rate": 1.8495390921437648e-05,
      "loss": 0.6956,
      "step": 53070
    },
    {
      "epoch": 4.566032865869397,
      "grad_norm": 5.1875,
      "learning_rate": 1.8488105000495413e-05,
      "loss": 1.2602,
      "step": 53071
    },
    {
      "epoch": 4.566118902176719,
      "grad_norm": 1.3046875,
      "learning_rate": 1.8480820487881798e-05,
      "loss": 1.059,
      "step": 53072
    },
    {
      "epoch": 4.566204938484041,
      "grad_norm": 1.03125,
      "learning_rate": 1.847353738361818e-05,
      "loss": 0.9178,
      "step": 53073
    },
    {
      "epoch": 4.566290974791362,
      "grad_norm": 1.0703125,
      "learning_rate": 1.846625568772581e-05,
      "loss": 0.9074,
      "step": 53074
    },
    {
      "epoch": 4.566377011098684,
      "grad_norm": 1.0234375,
      "learning_rate": 1.8458975400225953e-05,
      "loss": 1.3664,
      "step": 53075
    },
    {
      "epoch": 4.566463047406005,
      "grad_norm": 0.93359375,
      "learning_rate": 1.8451696521139983e-05,
      "loss": 1.3719,
      "step": 53076
    },
    {
      "epoch": 4.566549083713327,
      "grad_norm": 1.375,
      "learning_rate": 1.8444419050489104e-05,
      "loss": 0.747,
      "step": 53077
    },
    {
      "epoch": 4.566635120020649,
      "grad_norm": 0.984375,
      "learning_rate": 1.8437142988294687e-05,
      "loss": 0.9099,
      "step": 53078
    },
    {
      "epoch": 4.56672115632797,
      "grad_norm": 1.1171875,
      "learning_rate": 1.8429868334577937e-05,
      "loss": 0.9115,
      "step": 53079
    },
    {
      "epoch": 4.5668071926352924,
      "grad_norm": 2.046875,
      "learning_rate": 1.8422595089360118e-05,
      "loss": 0.7733,
      "step": 53080
    },
    {
      "epoch": 4.566893228942614,
      "grad_norm": 1.0703125,
      "learning_rate": 1.8415323252662653e-05,
      "loss": 1.5856,
      "step": 53081
    },
    {
      "epoch": 4.566979265249936,
      "grad_norm": 1.8984375,
      "learning_rate": 1.8408052824506583e-05,
      "loss": 0.9451,
      "step": 53082
    },
    {
      "epoch": 4.567065301557257,
      "grad_norm": 1.21875,
      "learning_rate": 1.8400783804913335e-05,
      "loss": 1.2104,
      "step": 53083
    },
    {
      "epoch": 4.567151337864579,
      "grad_norm": 1.1015625,
      "learning_rate": 1.8393516193904225e-05,
      "loss": 1.6137,
      "step": 53084
    },
    {
      "epoch": 4.567237374171901,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8386249991500292e-05,
      "loss": 0.604,
      "step": 53085
    },
    {
      "epoch": 4.567323410479222,
      "grad_norm": 1.109375,
      "learning_rate": 1.8378985197722962e-05,
      "loss": 0.776,
      "step": 53086
    },
    {
      "epoch": 4.567409446786544,
      "grad_norm": 1.078125,
      "learning_rate": 1.8371721812593445e-05,
      "loss": 1.4239,
      "step": 53087
    },
    {
      "epoch": 4.567495483093865,
      "grad_norm": 6.8125,
      "learning_rate": 1.8364459836132942e-05,
      "loss": 1.0072,
      "step": 53088
    },
    {
      "epoch": 4.5675815194011875,
      "grad_norm": 0.953125,
      "learning_rate": 1.8357199268362767e-05,
      "loss": 0.9344,
      "step": 53089
    },
    {
      "epoch": 4.567667555708509,
      "grad_norm": 1.3828125,
      "learning_rate": 1.834994010930402e-05,
      "loss": 1.2325,
      "step": 53090
    },
    {
      "epoch": 4.567753592015831,
      "grad_norm": 1.0078125,
      "learning_rate": 1.834268235897807e-05,
      "loss": 0.6337,
      "step": 53091
    },
    {
      "epoch": 4.567839628323153,
      "grad_norm": 1.1484375,
      "learning_rate": 1.833542601740612e-05,
      "loss": 0.8647,
      "step": 53092
    },
    {
      "epoch": 4.567925664630474,
      "grad_norm": 1.0859375,
      "learning_rate": 1.8328171084609324e-05,
      "loss": 1.4439,
      "step": 53093
    },
    {
      "epoch": 4.568011700937796,
      "grad_norm": 0.91796875,
      "learning_rate": 1.832091756060894e-05,
      "loss": 1.3274,
      "step": 53094
    },
    {
      "epoch": 4.568097737245117,
      "grad_norm": 1.0,
      "learning_rate": 1.8313665445426287e-05,
      "loss": 0.7124,
      "step": 53095
    },
    {
      "epoch": 4.568183773552439,
      "grad_norm": 0.8125,
      "learning_rate": 1.83064147390824e-05,
      "loss": 0.4991,
      "step": 53096
    },
    {
      "epoch": 4.5682698098597605,
      "grad_norm": 1.0625,
      "learning_rate": 1.8299165441598544e-05,
      "loss": 1.0745,
      "step": 53097
    },
    {
      "epoch": 4.568355846167083,
      "grad_norm": 0.76953125,
      "learning_rate": 1.829191755299603e-05,
      "loss": 0.7468,
      "step": 53098
    },
    {
      "epoch": 4.568441882474405,
      "grad_norm": 1.1015625,
      "learning_rate": 1.82846710732959e-05,
      "loss": 1.1332,
      "step": 53099
    },
    {
      "epoch": 4.568527918781726,
      "grad_norm": 0.93359375,
      "learning_rate": 1.8277426002519415e-05,
      "loss": 1.2471,
      "step": 53100
    },
    {
      "epoch": 4.568613955089048,
      "grad_norm": 1.6875,
      "learning_rate": 1.8270182340687836e-05,
      "loss": 0.9042,
      "step": 53101
    },
    {
      "epoch": 4.568699991396369,
      "grad_norm": 1.375,
      "learning_rate": 1.82629400878222e-05,
      "loss": 1.6808,
      "step": 53102
    },
    {
      "epoch": 4.568786027703691,
      "grad_norm": 1.1796875,
      "learning_rate": 1.825569924394388e-05,
      "loss": 1.1333,
      "step": 53103
    },
    {
      "epoch": 4.568872064011012,
      "grad_norm": 0.98046875,
      "learning_rate": 1.8248459809073858e-05,
      "loss": 1.1197,
      "step": 53104
    },
    {
      "epoch": 4.568958100318334,
      "grad_norm": 1.15625,
      "learning_rate": 1.8241221783233343e-05,
      "loss": 0.8503,
      "step": 53105
    },
    {
      "epoch": 4.5690441366256564,
      "grad_norm": 1.4921875,
      "learning_rate": 1.82339851664437e-05,
      "loss": 1.1286,
      "step": 53106
    },
    {
      "epoch": 4.569130172932978,
      "grad_norm": 1.125,
      "learning_rate": 1.8226749958725863e-05,
      "loss": 0.504,
      "step": 53107
    },
    {
      "epoch": 4.5692162092403,
      "grad_norm": 1.2109375,
      "learning_rate": 1.821951616010109e-05,
      "loss": 1.5431,
      "step": 53108
    },
    {
      "epoch": 4.569302245547621,
      "grad_norm": 0.9921875,
      "learning_rate": 1.8212283770590475e-05,
      "loss": 0.6038,
      "step": 53109
    },
    {
      "epoch": 4.569388281854943,
      "grad_norm": 1.2734375,
      "learning_rate": 1.8205052790215283e-05,
      "loss": 0.7252,
      "step": 53110
    },
    {
      "epoch": 4.569474318162264,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8197823218996655e-05,
      "loss": 1.0344,
      "step": 53111
    },
    {
      "epoch": 4.569560354469586,
      "grad_norm": 0.9609375,
      "learning_rate": 1.819059505695564e-05,
      "loss": 1.0422,
      "step": 53112
    },
    {
      "epoch": 4.569646390776908,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8183368304113436e-05,
      "loss": 0.8357,
      "step": 53113
    },
    {
      "epoch": 4.569732427084229,
      "grad_norm": 3.578125,
      "learning_rate": 1.8176142960491193e-05,
      "loss": 1.2488,
      "step": 53114
    },
    {
      "epoch": 4.5698184633915515,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8168919026110008e-05,
      "loss": 0.9341,
      "step": 53115
    },
    {
      "epoch": 4.569904499698873,
      "grad_norm": 1.1875,
      "learning_rate": 1.8161696500990977e-05,
      "loss": 1.2758,
      "step": 53116
    },
    {
      "epoch": 4.569990536006195,
      "grad_norm": 1.078125,
      "learning_rate": 1.8154475385155356e-05,
      "loss": 1.6946,
      "step": 53117
    },
    {
      "epoch": 4.570076572313516,
      "grad_norm": 0.87890625,
      "learning_rate": 1.814725567862413e-05,
      "loss": 0.6236,
      "step": 53118
    },
    {
      "epoch": 4.570162608620838,
      "grad_norm": 1.3125,
      "learning_rate": 1.8140037381418506e-05,
      "loss": 1.7209,
      "step": 53119
    },
    {
      "epoch": 4.57024864492816,
      "grad_norm": 0.76171875,
      "learning_rate": 1.813282049355952e-05,
      "loss": 1.2574,
      "step": 53120
    },
    {
      "epoch": 4.570334681235481,
      "grad_norm": 0.98046875,
      "learning_rate": 1.8125605015068325e-05,
      "loss": 0.7765,
      "step": 53121
    },
    {
      "epoch": 4.570420717542803,
      "grad_norm": 95.0,
      "learning_rate": 1.811839094596601e-05,
      "loss": 0.9996,
      "step": 53122
    },
    {
      "epoch": 4.5705067538501245,
      "grad_norm": 14.9375,
      "learning_rate": 1.8111178286273677e-05,
      "loss": 1.1609,
      "step": 53123
    },
    {
      "epoch": 4.570592790157447,
      "grad_norm": 1.1171875,
      "learning_rate": 1.8103967036012413e-05,
      "loss": 0.865,
      "step": 53124
    },
    {
      "epoch": 4.570678826464768,
      "grad_norm": 1.078125,
      "learning_rate": 1.8096757195203373e-05,
      "loss": 0.7308,
      "step": 53125
    },
    {
      "epoch": 4.57076486277209,
      "grad_norm": 1.265625,
      "learning_rate": 1.8089548763867537e-05,
      "loss": 1.1492,
      "step": 53126
    },
    {
      "epoch": 4.570850899079412,
      "grad_norm": 1.34375,
      "learning_rate": 1.8082341742026053e-05,
      "loss": 2.1191,
      "step": 53127
    },
    {
      "epoch": 4.570936935386733,
      "grad_norm": 1.0234375,
      "learning_rate": 1.8075136129700074e-05,
      "loss": 0.8747,
      "step": 53128
    },
    {
      "epoch": 4.571022971694055,
      "grad_norm": 2.734375,
      "learning_rate": 1.8067931926910473e-05,
      "loss": 1.418,
      "step": 53129
    },
    {
      "epoch": 4.571109008001376,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8060729133678568e-05,
      "loss": 0.9227,
      "step": 53130
    },
    {
      "epoch": 4.571195044308698,
      "grad_norm": 1.7265625,
      "learning_rate": 1.8053527750025167e-05,
      "loss": 0.7135,
      "step": 53131
    },
    {
      "epoch": 4.57128108061602,
      "grad_norm": 1.1484375,
      "learning_rate": 1.804632777597148e-05,
      "loss": 1.6246,
      "step": 53132
    },
    {
      "epoch": 4.571367116923342,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8039129211538663e-05,
      "loss": 0.5744,
      "step": 53133
    },
    {
      "epoch": 4.571453153230664,
      "grad_norm": 1.109375,
      "learning_rate": 1.803193205674758e-05,
      "loss": 1.4128,
      "step": 53134
    },
    {
      "epoch": 4.571539189537985,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8024736311619327e-05,
      "loss": 0.8816,
      "step": 53135
    },
    {
      "epoch": 4.571625225845307,
      "grad_norm": 1.2734375,
      "learning_rate": 1.8017541976175057e-05,
      "loss": 1.5132,
      "step": 53136
    },
    {
      "epoch": 4.571711262152628,
      "grad_norm": 1.1796875,
      "learning_rate": 1.801034905043564e-05,
      "loss": 1.0703,
      "step": 53137
    },
    {
      "epoch": 4.57179729845995,
      "grad_norm": 1.140625,
      "learning_rate": 1.8003157534422275e-05,
      "loss": 0.709,
      "step": 53138
    },
    {
      "epoch": 4.571883334767271,
      "grad_norm": 1.1484375,
      "learning_rate": 1.7995967428155956e-05,
      "loss": 0.9168,
      "step": 53139
    },
    {
      "epoch": 4.5719693710745934,
      "grad_norm": 1.3125,
      "learning_rate": 1.7988778731657606e-05,
      "loss": 1.1417,
      "step": 53140
    },
    {
      "epoch": 4.5720554073819155,
      "grad_norm": 1.0078125,
      "learning_rate": 1.7981591444948375e-05,
      "loss": 1.5821,
      "step": 53141
    },
    {
      "epoch": 4.572141443689237,
      "grad_norm": 1.015625,
      "learning_rate": 1.7974405568049302e-05,
      "loss": 1.4892,
      "step": 53142
    },
    {
      "epoch": 4.572227479996559,
      "grad_norm": 14.5625,
      "learning_rate": 1.7967221100981256e-05,
      "loss": 0.5727,
      "step": 53143
    },
    {
      "epoch": 4.57231351630388,
      "grad_norm": 1.15625,
      "learning_rate": 1.796003804376539e-05,
      "loss": 1.0747,
      "step": 53144
    },
    {
      "epoch": 4.572399552611202,
      "grad_norm": 4.375,
      "learning_rate": 1.7952856396422578e-05,
      "loss": 0.7614,
      "step": 53145
    },
    {
      "epoch": 4.572485588918523,
      "grad_norm": 1.0,
      "learning_rate": 1.794567615897402e-05,
      "loss": 0.5146,
      "step": 53146
    },
    {
      "epoch": 4.572571625225845,
      "grad_norm": 0.95703125,
      "learning_rate": 1.793849733144054e-05,
      "loss": 0.9285,
      "step": 53147
    },
    {
      "epoch": 4.572657661533167,
      "grad_norm": 1.2421875,
      "learning_rate": 1.793131991384317e-05,
      "loss": 0.7872,
      "step": 53148
    },
    {
      "epoch": 4.5727436978404885,
      "grad_norm": 1.2578125,
      "learning_rate": 1.792414390620306e-05,
      "loss": 1.3152,
      "step": 53149
    },
    {
      "epoch": 4.572829734147811,
      "grad_norm": 1.921875,
      "learning_rate": 1.791696930854092e-05,
      "loss": 0.7268,
      "step": 53150
    },
    {
      "epoch": 4.572915770455132,
      "grad_norm": 1.0078125,
      "learning_rate": 1.7909796120877953e-05,
      "loss": 0.7128,
      "step": 53151
    },
    {
      "epoch": 4.573001806762454,
      "grad_norm": 1.0390625,
      "learning_rate": 1.7902624343235087e-05,
      "loss": 1.1909,
      "step": 53152
    },
    {
      "epoch": 4.573087843069775,
      "grad_norm": 1.125,
      "learning_rate": 1.7895453975633247e-05,
      "loss": 0.7475,
      "step": 53153
    },
    {
      "epoch": 4.573173879377097,
      "grad_norm": 1.5234375,
      "learning_rate": 1.7888285018093476e-05,
      "loss": 1.1102,
      "step": 53154
    },
    {
      "epoch": 4.573259915684419,
      "grad_norm": 1.2109375,
      "learning_rate": 1.78811174706367e-05,
      "loss": 0.6489,
      "step": 53155
    },
    {
      "epoch": 4.57334595199174,
      "grad_norm": 1.09375,
      "learning_rate": 1.78739513332839e-05,
      "loss": 1.2575,
      "step": 53156
    },
    {
      "epoch": 4.573431988299062,
      "grad_norm": 1.265625,
      "learning_rate": 1.786678660605595e-05,
      "loss": 1.5093,
      "step": 53157
    },
    {
      "epoch": 4.573518024606384,
      "grad_norm": 1.3359375,
      "learning_rate": 1.7859623288973947e-05,
      "loss": 1.6877,
      "step": 53158
    },
    {
      "epoch": 4.573604060913706,
      "grad_norm": 1.1875,
      "learning_rate": 1.785246138205876e-05,
      "loss": 1.0206,
      "step": 53159
    },
    {
      "epoch": 4.573690097221027,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7845300885331373e-05,
      "loss": 0.3864,
      "step": 53160
    },
    {
      "epoch": 4.573776133528349,
      "grad_norm": 1.078125,
      "learning_rate": 1.7838141798812656e-05,
      "loss": 0.5984,
      "step": 53161
    },
    {
      "epoch": 4.573862169835671,
      "grad_norm": 0.95703125,
      "learning_rate": 1.7830984122523653e-05,
      "loss": 0.3472,
      "step": 53162
    },
    {
      "epoch": 4.573948206142992,
      "grad_norm": 1.25,
      "learning_rate": 1.782382785648523e-05,
      "loss": 1.4593,
      "step": 53163
    },
    {
      "epoch": 4.574034242450314,
      "grad_norm": 1.5078125,
      "learning_rate": 1.781667300071832e-05,
      "loss": 0.8448,
      "step": 53164
    },
    {
      "epoch": 4.574120278757635,
      "grad_norm": 1.0703125,
      "learning_rate": 1.7809519555243847e-05,
      "loss": 0.6459,
      "step": 53165
    },
    {
      "epoch": 4.5742063150649575,
      "grad_norm": 1.0859375,
      "learning_rate": 1.7802367520082797e-05,
      "loss": 1.6812,
      "step": 53166
    },
    {
      "epoch": 4.574292351372279,
      "grad_norm": 1.078125,
      "learning_rate": 1.7795216895255982e-05,
      "loss": 1.0568,
      "step": 53167
    },
    {
      "epoch": 4.574378387679601,
      "grad_norm": 1.2734375,
      "learning_rate": 1.778806768078439e-05,
      "loss": 1.5879,
      "step": 53168
    },
    {
      "epoch": 4.574464423986923,
      "grad_norm": 40.5,
      "learning_rate": 1.778091987668895e-05,
      "loss": 0.6777,
      "step": 53169
    },
    {
      "epoch": 4.574550460294244,
      "grad_norm": 1.78125,
      "learning_rate": 1.7773773482990474e-05,
      "loss": 1.3429,
      "step": 53170
    },
    {
      "epoch": 4.574636496601566,
      "grad_norm": 1.125,
      "learning_rate": 1.776662849971e-05,
      "loss": 1.0953,
      "step": 53171
    },
    {
      "epoch": 4.574722532908887,
      "grad_norm": 0.97265625,
      "learning_rate": 1.7759484926868297e-05,
      "loss": 0.6828,
      "step": 53172
    },
    {
      "epoch": 4.574808569216209,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7752342764486284e-05,
      "loss": 1.2088,
      "step": 53173
    },
    {
      "epoch": 4.5748946055235304,
      "grad_norm": 1.203125,
      "learning_rate": 1.7745202012584947e-05,
      "loss": 1.095,
      "step": 53174
    },
    {
      "epoch": 4.5749806418308525,
      "grad_norm": 98.5,
      "learning_rate": 1.773806267118505e-05,
      "loss": 1.232,
      "step": 53175
    },
    {
      "epoch": 4.575066678138175,
      "grad_norm": 1.796875,
      "learning_rate": 1.773092474030752e-05,
      "loss": 1.434,
      "step": 53176
    },
    {
      "epoch": 4.575152714445496,
      "grad_norm": 0.9375,
      "learning_rate": 1.772378821997328e-05,
      "loss": 0.9033,
      "step": 53177
    },
    {
      "epoch": 4.575238750752818,
      "grad_norm": 1.3359375,
      "learning_rate": 1.771665311020315e-05,
      "loss": 1.1068,
      "step": 53178
    },
    {
      "epoch": 4.575324787060139,
      "grad_norm": 1.078125,
      "learning_rate": 1.770951941101795e-05,
      "loss": 0.8239,
      "step": 53179
    },
    {
      "epoch": 4.575410823367461,
      "grad_norm": 1.171875,
      "learning_rate": 1.7702387122438714e-05,
      "loss": 1.4174,
      "step": 53180
    },
    {
      "epoch": 4.575496859674783,
      "grad_norm": 1.2421875,
      "learning_rate": 1.7695256244486092e-05,
      "loss": 1.9695,
      "step": 53181
    },
    {
      "epoch": 4.575582895982104,
      "grad_norm": 1.2578125,
      "learning_rate": 1.768812677718107e-05,
      "loss": 1.193,
      "step": 53182
    },
    {
      "epoch": 4.575668932289426,
      "grad_norm": 1.1484375,
      "learning_rate": 1.768099872054446e-05,
      "loss": 1.1256,
      "step": 53183
    },
    {
      "epoch": 4.575754968596748,
      "grad_norm": 1.2109375,
      "learning_rate": 1.7673872074597142e-05,
      "loss": 1.4028,
      "step": 53184
    },
    {
      "epoch": 4.57584100490407,
      "grad_norm": 1.2578125,
      "learning_rate": 1.766674683935998e-05,
      "loss": 0.6524,
      "step": 53185
    },
    {
      "epoch": 4.575927041211391,
      "grad_norm": 1.265625,
      "learning_rate": 1.765962301485374e-05,
      "loss": 1.0509,
      "step": 53186
    },
    {
      "epoch": 4.576013077518713,
      "grad_norm": 1.453125,
      "learning_rate": 1.7652500601099297e-05,
      "loss": 1.1144,
      "step": 53187
    },
    {
      "epoch": 4.576099113826035,
      "grad_norm": 1.28125,
      "learning_rate": 1.7645379598117518e-05,
      "loss": 0.7228,
      "step": 53188
    },
    {
      "epoch": 4.576185150133356,
      "grad_norm": 0.87109375,
      "learning_rate": 1.7638260005929106e-05,
      "loss": 0.8349,
      "step": 53189
    },
    {
      "epoch": 4.576271186440678,
      "grad_norm": 1.03125,
      "learning_rate": 1.763114182455505e-05,
      "loss": 1.6949,
      "step": 53190
    },
    {
      "epoch": 4.576357222747999,
      "grad_norm": 1.078125,
      "learning_rate": 1.7624025054016058e-05,
      "loss": 1.1241,
      "step": 53191
    },
    {
      "epoch": 4.5764432590553215,
      "grad_norm": 0.94140625,
      "learning_rate": 1.761690969433294e-05,
      "loss": 1.8419,
      "step": 53192
    },
    {
      "epoch": 4.5765292953626435,
      "grad_norm": 1.53125,
      "learning_rate": 1.7609795745526625e-05,
      "loss": 1.0352,
      "step": 53193
    },
    {
      "epoch": 4.576615331669965,
      "grad_norm": 1.34375,
      "learning_rate": 1.7602683207617764e-05,
      "loss": 1.4264,
      "step": 53194
    },
    {
      "epoch": 4.576701367977287,
      "grad_norm": 1.640625,
      "learning_rate": 1.759557208062723e-05,
      "loss": 1.7092,
      "step": 53195
    },
    {
      "epoch": 4.576787404284608,
      "grad_norm": 33.5,
      "learning_rate": 1.7588462364575898e-05,
      "loss": 0.8666,
      "step": 53196
    },
    {
      "epoch": 4.57687344059193,
      "grad_norm": 1.15625,
      "learning_rate": 1.758135405948441e-05,
      "loss": 1.1802,
      "step": 53197
    },
    {
      "epoch": 4.576959476899251,
      "grad_norm": 1.2734375,
      "learning_rate": 1.7574247165373646e-05,
      "loss": 0.7508,
      "step": 53198
    },
    {
      "epoch": 4.577045513206573,
      "grad_norm": 1.1484375,
      "learning_rate": 1.7567141682264475e-05,
      "loss": 0.8235,
      "step": 53199
    },
    {
      "epoch": 4.577131549513895,
      "grad_norm": 0.796875,
      "learning_rate": 1.7560037610177493e-05,
      "loss": 0.8844,
      "step": 53200
    },
    {
      "epoch": 4.5772175858212165,
      "grad_norm": 73.0,
      "learning_rate": 1.7552934949133627e-05,
      "loss": 1.2131,
      "step": 53201
    },
    {
      "epoch": 4.577303622128539,
      "grad_norm": 1.0625,
      "learning_rate": 1.7545833699153525e-05,
      "loss": 1.4833,
      "step": 53202
    },
    {
      "epoch": 4.57738965843586,
      "grad_norm": 4.09375,
      "learning_rate": 1.753873386025806e-05,
      "loss": 1.3695,
      "step": 53203
    },
    {
      "epoch": 4.577475694743182,
      "grad_norm": 1.1953125,
      "learning_rate": 1.7531635432467995e-05,
      "loss": 1.1806,
      "step": 53204
    },
    {
      "epoch": 4.577561731050503,
      "grad_norm": 1.0390625,
      "learning_rate": 1.7524538415804037e-05,
      "loss": 1.3624,
      "step": 53205
    },
    {
      "epoch": 4.577647767357825,
      "grad_norm": 1.0546875,
      "learning_rate": 1.7517442810286945e-05,
      "loss": 1.6772,
      "step": 53206
    },
    {
      "epoch": 4.577733803665147,
      "grad_norm": 1.34375,
      "learning_rate": 1.7510348615937533e-05,
      "loss": 1.1672,
      "step": 53207
    },
    {
      "epoch": 4.577819839972468,
      "grad_norm": 50.25,
      "learning_rate": 1.750325583277651e-05,
      "loss": 1.3301,
      "step": 53208
    },
    {
      "epoch": 4.57790587627979,
      "grad_norm": 1.078125,
      "learning_rate": 1.749616446082458e-05,
      "loss": 1.109,
      "step": 53209
    },
    {
      "epoch": 4.577991912587112,
      "grad_norm": 1.0234375,
      "learning_rate": 1.7489074500102618e-05,
      "loss": 1.7462,
      "step": 53210
    },
    {
      "epoch": 4.578077948894434,
      "grad_norm": 0.87890625,
      "learning_rate": 1.7481985950631164e-05,
      "loss": 0.8628,
      "step": 53211
    },
    {
      "epoch": 4.578163985201755,
      "grad_norm": 1.015625,
      "learning_rate": 1.747489881243114e-05,
      "loss": 1.1991,
      "step": 53212
    },
    {
      "epoch": 4.578250021509077,
      "grad_norm": 10.3125,
      "learning_rate": 1.7467813085523143e-05,
      "loss": 1.2642,
      "step": 53213
    },
    {
      "epoch": 4.578336057816399,
      "grad_norm": 1.0,
      "learning_rate": 1.7460728769927935e-05,
      "loss": 0.8421,
      "step": 53214
    },
    {
      "epoch": 4.57842209412372,
      "grad_norm": 1.0703125,
      "learning_rate": 1.7453645865666335e-05,
      "loss": 1.0923,
      "step": 53215
    },
    {
      "epoch": 4.578508130431042,
      "grad_norm": 1.0703125,
      "learning_rate": 1.7446564372758876e-05,
      "loss": 0.757,
      "step": 53216
    },
    {
      "epoch": 4.578594166738363,
      "grad_norm": 0.95703125,
      "learning_rate": 1.743948429122638e-05,
      "loss": 0.7298,
      "step": 53217
    },
    {
      "epoch": 4.5786802030456855,
      "grad_norm": 1.140625,
      "learning_rate": 1.7432405621089554e-05,
      "loss": 1.1374,
      "step": 53218
    },
    {
      "epoch": 4.578766239353007,
      "grad_norm": 1.0703125,
      "learning_rate": 1.742532836236904e-05,
      "loss": 0.7421,
      "step": 53219
    },
    {
      "epoch": 4.578852275660329,
      "grad_norm": 1.1171875,
      "learning_rate": 1.7418252515085663e-05,
      "loss": 0.9791,
      "step": 53220
    },
    {
      "epoch": 4.578938311967651,
      "grad_norm": 1.140625,
      "learning_rate": 1.741117807925996e-05,
      "loss": 1.4817,
      "step": 53221
    },
    {
      "epoch": 4.579024348274972,
      "grad_norm": 1.1015625,
      "learning_rate": 1.7404105054912745e-05,
      "loss": 0.9084,
      "step": 53222
    },
    {
      "epoch": 4.579110384582294,
      "grad_norm": 1.0234375,
      "learning_rate": 1.7397033442064723e-05,
      "loss": 0.8513,
      "step": 53223
    },
    {
      "epoch": 4.579196420889615,
      "grad_norm": 0.80078125,
      "learning_rate": 1.738996324073644e-05,
      "loss": 0.6009,
      "step": 53224
    },
    {
      "epoch": 4.579282457196937,
      "grad_norm": 1.0078125,
      "learning_rate": 1.7382894450948648e-05,
      "loss": 0.9474,
      "step": 53225
    },
    {
      "epoch": 4.5793684935042585,
      "grad_norm": 1.2421875,
      "learning_rate": 1.7375827072722116e-05,
      "loss": 1.2241,
      "step": 53226
    },
    {
      "epoch": 4.5794545298115805,
      "grad_norm": 1.0546875,
      "learning_rate": 1.7368761106077323e-05,
      "loss": 1.3532,
      "step": 53227
    },
    {
      "epoch": 4.579540566118903,
      "grad_norm": 1.1015625,
      "learning_rate": 1.7361696551035033e-05,
      "loss": 1.3766,
      "step": 53228
    },
    {
      "epoch": 4.579626602426224,
      "grad_norm": 1.3125,
      "learning_rate": 1.7354633407616003e-05,
      "loss": 0.8616,
      "step": 53229
    },
    {
      "epoch": 4.579712638733546,
      "grad_norm": 1.1640625,
      "learning_rate": 1.7347571675840725e-05,
      "loss": 1.3137,
      "step": 53230
    },
    {
      "epoch": 4.579798675040867,
      "grad_norm": 45.0,
      "learning_rate": 1.7340511355730004e-05,
      "loss": 0.5731,
      "step": 53231
    },
    {
      "epoch": 4.579884711348189,
      "grad_norm": 0.875,
      "learning_rate": 1.733345244730439e-05,
      "loss": 1.2021,
      "step": 53232
    },
    {
      "epoch": 4.57997074765551,
      "grad_norm": 0.9921875,
      "learning_rate": 1.7326394950584467e-05,
      "loss": 0.7567,
      "step": 53233
    },
    {
      "epoch": 4.580056783962832,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7319338865591117e-05,
      "loss": 0.6469,
      "step": 53234
    },
    {
      "epoch": 4.580142820270154,
      "grad_norm": 1.40625,
      "learning_rate": 1.7312284192344706e-05,
      "loss": 1.0814,
      "step": 53235
    },
    {
      "epoch": 4.580228856577476,
      "grad_norm": 10.875,
      "learning_rate": 1.7305230930866e-05,
      "loss": 1.2107,
      "step": 53236
    },
    {
      "epoch": 4.580314892884798,
      "grad_norm": 1.0859375,
      "learning_rate": 1.7298179081175645e-05,
      "loss": 1.7787,
      "step": 53237
    },
    {
      "epoch": 4.580400929192119,
      "grad_norm": 1.0234375,
      "learning_rate": 1.729112864329424e-05,
      "loss": 1.3748,
      "step": 53238
    },
    {
      "epoch": 4.580486965499441,
      "grad_norm": 1.015625,
      "learning_rate": 1.728407961724232e-05,
      "loss": 1.0865,
      "step": 53239
    },
    {
      "epoch": 4.580573001806762,
      "grad_norm": 1.078125,
      "learning_rate": 1.7277032003040704e-05,
      "loss": 0.5442,
      "step": 53240
    },
    {
      "epoch": 4.580659038114084,
      "grad_norm": 1.5078125,
      "learning_rate": 1.726998580070982e-05,
      "loss": 1.5876,
      "step": 53241
    },
    {
      "epoch": 4.580745074421406,
      "grad_norm": 0.91796875,
      "learning_rate": 1.7262941010270372e-05,
      "loss": 1.0061,
      "step": 53242
    },
    {
      "epoch": 4.580831110728727,
      "grad_norm": 0.93359375,
      "learning_rate": 1.7255897631742902e-05,
      "loss": 0.6359,
      "step": 53243
    },
    {
      "epoch": 4.5809171470360495,
      "grad_norm": 1.046875,
      "learning_rate": 1.7248855665148055e-05,
      "loss": 0.924,
      "step": 53244
    },
    {
      "epoch": 4.581003183343371,
      "grad_norm": 1.109375,
      "learning_rate": 1.724181511050643e-05,
      "loss": 1.651,
      "step": 53245
    },
    {
      "epoch": 4.581089219650693,
      "grad_norm": 121.5,
      "learning_rate": 1.7234775967838566e-05,
      "loss": 1.1739,
      "step": 53246
    },
    {
      "epoch": 4.581175255958014,
      "grad_norm": 0.99609375,
      "learning_rate": 1.7227738237165114e-05,
      "loss": 0.6453,
      "step": 53247
    },
    {
      "epoch": 4.581261292265336,
      "grad_norm": 3.78125,
      "learning_rate": 1.7220701918506663e-05,
      "loss": 1.0936,
      "step": 53248
    },
    {
      "epoch": 4.581347328572658,
      "grad_norm": 1.09375,
      "learning_rate": 1.7213667011883703e-05,
      "loss": 0.5339,
      "step": 53249
    },
    {
      "epoch": 4.581433364879979,
      "grad_norm": 1.2265625,
      "learning_rate": 1.7206633517316884e-05,
      "loss": 0.8419,
      "step": 53250
    },
    {
      "epoch": 4.581519401187301,
      "grad_norm": 1.0,
      "learning_rate": 1.719960143482685e-05,
      "loss": 0.3666,
      "step": 53251
    },
    {
      "epoch": 4.5816054374946225,
      "grad_norm": 0.9375,
      "learning_rate": 1.719257076443398e-05,
      "loss": 0.7613,
      "step": 53252
    },
    {
      "epoch": 4.5816914738019445,
      "grad_norm": 0.88671875,
      "learning_rate": 1.7185541506158973e-05,
      "loss": 0.6273,
      "step": 53253
    },
    {
      "epoch": 4.581777510109266,
      "grad_norm": 1.0625,
      "learning_rate": 1.7178513660022378e-05,
      "loss": 0.7406,
      "step": 53254
    },
    {
      "epoch": 4.581863546416588,
      "grad_norm": 0.98046875,
      "learning_rate": 1.7171487226044668e-05,
      "loss": 0.8744,
      "step": 53255
    },
    {
      "epoch": 4.58194958272391,
      "grad_norm": 1.265625,
      "learning_rate": 1.71644622042465e-05,
      "loss": 0.8648,
      "step": 53256
    },
    {
      "epoch": 4.582035619031231,
      "grad_norm": 1.0546875,
      "learning_rate": 1.71574385946483e-05,
      "loss": 1.3019,
      "step": 53257
    },
    {
      "epoch": 4.582121655338553,
      "grad_norm": 1.3359375,
      "learning_rate": 1.715041639727083e-05,
      "loss": 1.0066,
      "step": 53258
    },
    {
      "epoch": 4.582207691645874,
      "grad_norm": 1.546875,
      "learning_rate": 1.714339561213435e-05,
      "loss": 1.5862,
      "step": 53259
    },
    {
      "epoch": 4.582293727953196,
      "grad_norm": 1.765625,
      "learning_rate": 1.7136376239259565e-05,
      "loss": 0.7696,
      "step": 53260
    },
    {
      "epoch": 4.5823797642605175,
      "grad_norm": 1.265625,
      "learning_rate": 1.7129358278667017e-05,
      "loss": 1.2595,
      "step": 53261
    },
    {
      "epoch": 4.58246580056784,
      "grad_norm": 1.53125,
      "learning_rate": 1.7122341730377133e-05,
      "loss": 1.0229,
      "step": 53262
    },
    {
      "epoch": 4.582551836875162,
      "grad_norm": 3.328125,
      "learning_rate": 1.7115326594410507e-05,
      "loss": 1.3474,
      "step": 53263
    },
    {
      "epoch": 4.582637873182483,
      "grad_norm": 1.3671875,
      "learning_rate": 1.7108312870787623e-05,
      "loss": 1.386,
      "step": 53264
    },
    {
      "epoch": 4.582723909489805,
      "grad_norm": 0.91015625,
      "learning_rate": 1.710130055952902e-05,
      "loss": 1.5095,
      "step": 53265
    },
    {
      "epoch": 4.582809945797126,
      "grad_norm": 1.2265625,
      "learning_rate": 1.7094289660655128e-05,
      "loss": 0.9136,
      "step": 53266
    },
    {
      "epoch": 4.582895982104448,
      "grad_norm": 1.1015625,
      "learning_rate": 1.7087280174186647e-05,
      "loss": 1.2102,
      "step": 53267
    },
    {
      "epoch": 4.582982018411769,
      "grad_norm": 0.9609375,
      "learning_rate": 1.7080272100143845e-05,
      "loss": 0.8383,
      "step": 53268
    },
    {
      "epoch": 4.583068054719091,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7073265438547313e-05,
      "loss": 0.7753,
      "step": 53269
    },
    {
      "epoch": 4.5831540910264135,
      "grad_norm": 1.0625,
      "learning_rate": 1.7066260189417648e-05,
      "loss": 1.7033,
      "step": 53270
    },
    {
      "epoch": 4.583240127333735,
      "grad_norm": 1.28125,
      "learning_rate": 1.705925635277522e-05,
      "loss": 1.3747,
      "step": 53271
    },
    {
      "epoch": 4.583326163641057,
      "grad_norm": 0.8671875,
      "learning_rate": 1.7052253928640517e-05,
      "loss": 0.7753,
      "step": 53272
    },
    {
      "epoch": 4.583412199948378,
      "grad_norm": 1.3671875,
      "learning_rate": 1.7045252917034016e-05,
      "loss": 1.9122,
      "step": 53273
    },
    {
      "epoch": 4.5834982362557,
      "grad_norm": 1.3984375,
      "learning_rate": 1.703825331797626e-05,
      "loss": 0.9813,
      "step": 53274
    },
    {
      "epoch": 4.583584272563021,
      "grad_norm": 1.265625,
      "learning_rate": 1.7031255131487677e-05,
      "loss": 1.7135,
      "step": 53275
    },
    {
      "epoch": 4.583670308870343,
      "grad_norm": 0.98828125,
      "learning_rate": 1.7024258357588753e-05,
      "loss": 0.7087,
      "step": 53276
    },
    {
      "epoch": 4.583756345177665,
      "grad_norm": 1.109375,
      "learning_rate": 1.701726299629991e-05,
      "loss": 1.2522,
      "step": 53277
    },
    {
      "epoch": 4.5838423814849865,
      "grad_norm": 1.078125,
      "learning_rate": 1.701026904764169e-05,
      "loss": 0.5718,
      "step": 53278
    },
    {
      "epoch": 4.5839284177923085,
      "grad_norm": 1.0078125,
      "learning_rate": 1.7003276511634414e-05,
      "loss": 0.926,
      "step": 53279
    },
    {
      "epoch": 4.58401445409963,
      "grad_norm": 1.1796875,
      "learning_rate": 1.699628538829867e-05,
      "loss": 1.0638,
      "step": 53280
    },
    {
      "epoch": 4.584100490406952,
      "grad_norm": 0.96875,
      "learning_rate": 1.6989295677654838e-05,
      "loss": 1.2638,
      "step": 53281
    },
    {
      "epoch": 4.584186526714273,
      "grad_norm": 1.046875,
      "learning_rate": 1.6982307379723394e-05,
      "loss": 1.0646,
      "step": 53282
    },
    {
      "epoch": 4.584272563021595,
      "grad_norm": 93.0,
      "learning_rate": 1.697532049452477e-05,
      "loss": 1.6979,
      "step": 53283
    },
    {
      "epoch": 4.584358599328917,
      "grad_norm": 1.0859375,
      "learning_rate": 1.696833502207934e-05,
      "loss": 1.2618,
      "step": 53284
    },
    {
      "epoch": 4.584444635636238,
      "grad_norm": 1.125,
      "learning_rate": 1.6961350962407586e-05,
      "loss": 1.1772,
      "step": 53285
    },
    {
      "epoch": 4.58453067194356,
      "grad_norm": 31.125,
      "learning_rate": 1.695436831552999e-05,
      "loss": 1.1349,
      "step": 53286
    },
    {
      "epoch": 4.5846167082508815,
      "grad_norm": 1.234375,
      "learning_rate": 1.694738708146687e-05,
      "loss": 1.1284,
      "step": 53287
    },
    {
      "epoch": 4.584702744558204,
      "grad_norm": 1.109375,
      "learning_rate": 1.694040726023871e-05,
      "loss": 1.2192,
      "step": 53288
    },
    {
      "epoch": 4.584788780865525,
      "grad_norm": 1.3125,
      "learning_rate": 1.693342885186594e-05,
      "loss": 1.6163,
      "step": 53289
    },
    {
      "epoch": 4.584874817172847,
      "grad_norm": 1.109375,
      "learning_rate": 1.6926451856368874e-05,
      "loss": 1.1611,
      "step": 53290
    },
    {
      "epoch": 4.584960853480169,
      "grad_norm": 1.28125,
      "learning_rate": 1.6919476273767998e-05,
      "loss": 0.8961,
      "step": 53291
    },
    {
      "epoch": 4.58504688978749,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6912502104083795e-05,
      "loss": 0.7619,
      "step": 53292
    },
    {
      "epoch": 4.585132926094812,
      "grad_norm": 1.1796875,
      "learning_rate": 1.6905529347336413e-05,
      "loss": 1.5012,
      "step": 53293
    },
    {
      "epoch": 4.585218962402133,
      "grad_norm": 0.94140625,
      "learning_rate": 1.6898558003546506e-05,
      "loss": 0.7686,
      "step": 53294
    },
    {
      "epoch": 4.585304998709455,
      "grad_norm": 22.875,
      "learning_rate": 1.6891588072734333e-05,
      "loss": 1.4258,
      "step": 53295
    },
    {
      "epoch": 4.585391035016777,
      "grad_norm": 1.2734375,
      "learning_rate": 1.6884619554920265e-05,
      "loss": 1.3082,
      "step": 53296
    },
    {
      "epoch": 4.585477071324099,
      "grad_norm": 0.9609375,
      "learning_rate": 1.6877652450124846e-05,
      "loss": 1.6011,
      "step": 53297
    },
    {
      "epoch": 4.585563107631421,
      "grad_norm": 1.171875,
      "learning_rate": 1.6870686758368224e-05,
      "loss": 1.7128,
      "step": 53298
    },
    {
      "epoch": 4.585649143938742,
      "grad_norm": 1.15625,
      "learning_rate": 1.6863722479670885e-05,
      "loss": 0.9611,
      "step": 53299
    },
    {
      "epoch": 4.585735180246064,
      "grad_norm": 1.3359375,
      "learning_rate": 1.685675961405325e-05,
      "loss": 1.4422,
      "step": 53300
    },
    {
      "epoch": 4.585821216553385,
      "grad_norm": 1.328125,
      "learning_rate": 1.684979816153559e-05,
      "loss": 1.1889,
      "step": 53301
    },
    {
      "epoch": 4.585907252860707,
      "grad_norm": 1.96875,
      "learning_rate": 1.6842838122138327e-05,
      "loss": 1.1154,
      "step": 53302
    },
    {
      "epoch": 4.585993289168029,
      "grad_norm": 1.0703125,
      "learning_rate": 1.683587949588178e-05,
      "loss": 1.5961,
      "step": 53303
    },
    {
      "epoch": 4.5860793254753505,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6828922282786263e-05,
      "loss": 0.4962,
      "step": 53304
    },
    {
      "epoch": 4.5861653617826725,
      "grad_norm": 1.015625,
      "learning_rate": 1.6821966482872263e-05,
      "loss": 1.2239,
      "step": 53305
    },
    {
      "epoch": 4.586251398089994,
      "grad_norm": 0.9375,
      "learning_rate": 1.6815012096159988e-05,
      "loss": 0.6281,
      "step": 53306
    },
    {
      "epoch": 4.586337434397316,
      "grad_norm": 1.3125,
      "learning_rate": 1.6808059122669805e-05,
      "loss": 1.1992,
      "step": 53307
    },
    {
      "epoch": 4.586423470704638,
      "grad_norm": 100.5,
      "learning_rate": 1.6801107562422145e-05,
      "loss": 1.302,
      "step": 53308
    },
    {
      "epoch": 4.586509507011959,
      "grad_norm": 1.3984375,
      "learning_rate": 1.6794157415437218e-05,
      "loss": 0.9654,
      "step": 53309
    },
    {
      "epoch": 4.586595543319281,
      "grad_norm": 1.5234375,
      "learning_rate": 1.678720868173539e-05,
      "loss": 1.2112,
      "step": 53310
    },
    {
      "epoch": 4.586681579626602,
      "grad_norm": 1.2890625,
      "learning_rate": 1.678026136133709e-05,
      "loss": 1.6228,
      "step": 53311
    },
    {
      "epoch": 4.586767615933924,
      "grad_norm": 1.171875,
      "learning_rate": 1.6773315454262416e-05,
      "loss": 1.1131,
      "step": 53312
    },
    {
      "epoch": 4.5868536522412455,
      "grad_norm": 2.109375,
      "learning_rate": 1.6766370960531907e-05,
      "loss": 0.668,
      "step": 53313
    },
    {
      "epoch": 4.586939688548568,
      "grad_norm": 1.5625,
      "learning_rate": 1.6759427880165767e-05,
      "loss": 1.0384,
      "step": 53314
    },
    {
      "epoch": 4.58702572485589,
      "grad_norm": 0.9375,
      "learning_rate": 1.675248621318426e-05,
      "loss": 1.134,
      "step": 53315
    },
    {
      "epoch": 4.587111761163211,
      "grad_norm": 1.203125,
      "learning_rate": 1.674554595960781e-05,
      "loss": 0.88,
      "step": 53316
    },
    {
      "epoch": 4.587197797470533,
      "grad_norm": 1.96875,
      "learning_rate": 1.6738607119456574e-05,
      "loss": 1.2435,
      "step": 53317
    },
    {
      "epoch": 4.587283833777854,
      "grad_norm": 57.0,
      "learning_rate": 1.673166969275097e-05,
      "loss": 1.5602,
      "step": 53318
    },
    {
      "epoch": 4.587369870085176,
      "grad_norm": 1.125,
      "learning_rate": 1.6724733679511273e-05,
      "loss": 1.3715,
      "step": 53319
    },
    {
      "epoch": 4.587455906392497,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6717799079757678e-05,
      "loss": 0.7093,
      "step": 53320
    },
    {
      "epoch": 4.587541942699819,
      "grad_norm": 1.078125,
      "learning_rate": 1.6710865893510506e-05,
      "loss": 1.4703,
      "step": 53321
    },
    {
      "epoch": 4.5876279790071415,
      "grad_norm": 1.078125,
      "learning_rate": 1.670393412079019e-05,
      "loss": 1.0567,
      "step": 53322
    },
    {
      "epoch": 4.587714015314463,
      "grad_norm": 1.0859375,
      "learning_rate": 1.6697003761616757e-05,
      "loss": 0.975,
      "step": 53323
    },
    {
      "epoch": 4.587800051621785,
      "grad_norm": 1.3359375,
      "learning_rate": 1.6690074816010647e-05,
      "loss": 1.0687,
      "step": 53324
    },
    {
      "epoch": 4.587886087929106,
      "grad_norm": 1.0,
      "learning_rate": 1.6683147283992005e-05,
      "loss": 1.1376,
      "step": 53325
    },
    {
      "epoch": 4.587972124236428,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6676221165581208e-05,
      "loss": 0.4173,
      "step": 53326
    },
    {
      "epoch": 4.588058160543749,
      "grad_norm": 1.0234375,
      "learning_rate": 1.6669296460798455e-05,
      "loss": 0.6885,
      "step": 53327
    },
    {
      "epoch": 4.588144196851071,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6662373169663956e-05,
      "loss": 0.9346,
      "step": 53328
    },
    {
      "epoch": 4.588230233158393,
      "grad_norm": 1.1640625,
      "learning_rate": 1.665545129219809e-05,
      "loss": 1.2502,
      "step": 53329
    },
    {
      "epoch": 4.5883162694657145,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6648530828420937e-05,
      "loss": 0.9844,
      "step": 53330
    },
    {
      "epoch": 4.5884023057730365,
      "grad_norm": 1.2578125,
      "learning_rate": 1.6641611778352883e-05,
      "loss": 1.0486,
      "step": 53331
    },
    {
      "epoch": 4.588488342080358,
      "grad_norm": 1.1796875,
      "learning_rate": 1.663469414201413e-05,
      "loss": 0.8062,
      "step": 53332
    },
    {
      "epoch": 4.58857437838768,
      "grad_norm": 1264.0,
      "learning_rate": 1.6627777919424825e-05,
      "loss": 1.1827,
      "step": 53333
    },
    {
      "epoch": 4.588660414695001,
      "grad_norm": 1.296875,
      "learning_rate": 1.662086311060529e-05,
      "loss": 1.2486,
      "step": 53334
    },
    {
      "epoch": 4.588746451002323,
      "grad_norm": 1.140625,
      "learning_rate": 1.6613949715575725e-05,
      "loss": 1.2583,
      "step": 53335
    },
    {
      "epoch": 4.588832487309645,
      "grad_norm": 1.421875,
      "learning_rate": 1.6607037734356345e-05,
      "loss": 0.7105,
      "step": 53336
    },
    {
      "epoch": 4.588918523616966,
      "grad_norm": 1.3828125,
      "learning_rate": 1.660012716696735e-05,
      "loss": 0.8965,
      "step": 53337
    },
    {
      "epoch": 4.589004559924288,
      "grad_norm": 2.3125,
      "learning_rate": 1.6593218013429e-05,
      "loss": 0.9812,
      "step": 53338
    },
    {
      "epoch": 4.5890905962316095,
      "grad_norm": 0.8671875,
      "learning_rate": 1.6586310273761396e-05,
      "loss": 0.5059,
      "step": 53339
    },
    {
      "epoch": 4.589176632538932,
      "grad_norm": 0.98828125,
      "learning_rate": 1.657940394798485e-05,
      "loss": 0.6269,
      "step": 53340
    },
    {
      "epoch": 4.589262668846253,
      "grad_norm": 1.3671875,
      "learning_rate": 1.657249903611957e-05,
      "loss": 1.1639,
      "step": 53341
    },
    {
      "epoch": 4.589348705153575,
      "grad_norm": 1.0625,
      "learning_rate": 1.6565595538185706e-05,
      "loss": 1.027,
      "step": 53342
    },
    {
      "epoch": 4.589434741460897,
      "grad_norm": 1.1875,
      "learning_rate": 1.6558693454203466e-05,
      "loss": 0.959,
      "step": 53343
    },
    {
      "epoch": 4.589520777768218,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6551792784192943e-05,
      "loss": 0.634,
      "step": 53344
    },
    {
      "epoch": 4.58960681407554,
      "grad_norm": 5.625,
      "learning_rate": 1.65448935281744e-05,
      "loss": 0.558,
      "step": 53345
    },
    {
      "epoch": 4.589692850382861,
      "grad_norm": 1.46875,
      "learning_rate": 1.6537995686168095e-05,
      "loss": 1.052,
      "step": 53346
    },
    {
      "epoch": 4.589778886690183,
      "grad_norm": 1.3359375,
      "learning_rate": 1.6531099258194073e-05,
      "loss": 1.2061,
      "step": 53347
    },
    {
      "epoch": 4.589864922997505,
      "grad_norm": 1.34375,
      "learning_rate": 1.652420424427259e-05,
      "loss": 0.7724,
      "step": 53348
    },
    {
      "epoch": 4.589950959304827,
      "grad_norm": 0.953125,
      "learning_rate": 1.65173106444238e-05,
      "loss": 0.6217,
      "step": 53349
    },
    {
      "epoch": 4.590036995612149,
      "grad_norm": 3.875,
      "learning_rate": 1.651041845866774e-05,
      "loss": 1.4398,
      "step": 53350
    },
    {
      "epoch": 4.59012303191947,
      "grad_norm": 1.109375,
      "learning_rate": 1.6503527687024733e-05,
      "loss": 1.1571,
      "step": 53351
    },
    {
      "epoch": 4.590209068226792,
      "grad_norm": 1.0859375,
      "learning_rate": 1.6496638329514923e-05,
      "loss": 0.8047,
      "step": 53352
    },
    {
      "epoch": 4.590295104534113,
      "grad_norm": 1.421875,
      "learning_rate": 1.6489750386158352e-05,
      "loss": 1.2766,
      "step": 53353
    },
    {
      "epoch": 4.590381140841435,
      "grad_norm": 1.7890625,
      "learning_rate": 1.6482863856975284e-05,
      "loss": 1.0637,
      "step": 53354
    },
    {
      "epoch": 4.590467177148756,
      "grad_norm": 1.21875,
      "learning_rate": 1.64759787419857e-05,
      "loss": 1.3098,
      "step": 53355
    },
    {
      "epoch": 4.5905532134560785,
      "grad_norm": 4.125,
      "learning_rate": 1.6469095041209858e-05,
      "loss": 1.0289,
      "step": 53356
    },
    {
      "epoch": 4.5906392497634005,
      "grad_norm": 0.98828125,
      "learning_rate": 1.646221275466797e-05,
      "loss": 0.494,
      "step": 53357
    },
    {
      "epoch": 4.590725286070722,
      "grad_norm": 0.97265625,
      "learning_rate": 1.6455331882379964e-05,
      "loss": 0.7763,
      "step": 53358
    },
    {
      "epoch": 4.590811322378044,
      "grad_norm": 1.140625,
      "learning_rate": 1.64484524243661e-05,
      "loss": 1.5603,
      "step": 53359
    },
    {
      "epoch": 4.590897358685365,
      "grad_norm": 0.8515625,
      "learning_rate": 1.6441574380646528e-05,
      "loss": 1.6779,
      "step": 53360
    },
    {
      "epoch": 4.590983394992687,
      "grad_norm": 328.0,
      "learning_rate": 1.643469775124118e-05,
      "loss": 1.0631,
      "step": 53361
    },
    {
      "epoch": 4.591069431300008,
      "grad_norm": 1.703125,
      "learning_rate": 1.6427822536170367e-05,
      "loss": 0.9413,
      "step": 53362
    },
    {
      "epoch": 4.59115546760733,
      "grad_norm": 1.03125,
      "learning_rate": 1.6420948735454133e-05,
      "loss": 1.0269,
      "step": 53363
    },
    {
      "epoch": 4.591241503914652,
      "grad_norm": 1.1015625,
      "learning_rate": 1.641407634911257e-05,
      "loss": 1.1436,
      "step": 53364
    },
    {
      "epoch": 4.5913275402219735,
      "grad_norm": 0.890625,
      "learning_rate": 1.6407205377165724e-05,
      "loss": 0.7984,
      "step": 53365
    },
    {
      "epoch": 4.591413576529296,
      "grad_norm": 11.625,
      "learning_rate": 1.640033581963385e-05,
      "loss": 0.939,
      "step": 53366
    },
    {
      "epoch": 4.591499612836617,
      "grad_norm": 1.171875,
      "learning_rate": 1.6393467676536766e-05,
      "loss": 1.279,
      "step": 53367
    },
    {
      "epoch": 4.591585649143939,
      "grad_norm": 1.25,
      "learning_rate": 1.6386600947894903e-05,
      "loss": 0.7627,
      "step": 53368
    },
    {
      "epoch": 4.59167168545126,
      "grad_norm": 1.421875,
      "learning_rate": 1.637973563372813e-05,
      "loss": 1.2085,
      "step": 53369
    },
    {
      "epoch": 4.591757721758582,
      "grad_norm": 0.93359375,
      "learning_rate": 1.6372871734056548e-05,
      "loss": 1.3998,
      "step": 53370
    },
    {
      "epoch": 4.591843758065904,
      "grad_norm": 18.0,
      "learning_rate": 1.6366009248900303e-05,
      "loss": 1.3419,
      "step": 53371
    },
    {
      "epoch": 4.591929794373225,
      "grad_norm": 1.125,
      "learning_rate": 1.6359148178279326e-05,
      "loss": 1.137,
      "step": 53372
    },
    {
      "epoch": 4.592015830680547,
      "grad_norm": 0.91015625,
      "learning_rate": 1.6352288522213877e-05,
      "loss": 0.756,
      "step": 53373
    },
    {
      "epoch": 4.592101866987869,
      "grad_norm": 1.0859375,
      "learning_rate": 1.6345430280723883e-05,
      "loss": 0.5794,
      "step": 53374
    },
    {
      "epoch": 4.592187903295191,
      "grad_norm": 1.2109375,
      "learning_rate": 1.6338573453829387e-05,
      "loss": 1.4487,
      "step": 53375
    },
    {
      "epoch": 4.592273939602512,
      "grad_norm": 1.1640625,
      "learning_rate": 1.6331718041550592e-05,
      "loss": 1.0669,
      "step": 53376
    },
    {
      "epoch": 4.592359975909834,
      "grad_norm": 2.515625,
      "learning_rate": 1.6324864043907374e-05,
      "loss": 1.4874,
      "step": 53377
    },
    {
      "epoch": 4.592446012217156,
      "grad_norm": 0.92578125,
      "learning_rate": 1.631801146091988e-05,
      "loss": 0.5641,
      "step": 53378
    },
    {
      "epoch": 4.592532048524477,
      "grad_norm": 0.984375,
      "learning_rate": 1.6311160292608207e-05,
      "loss": 0.4667,
      "step": 53379
    },
    {
      "epoch": 4.592618084831799,
      "grad_norm": 7.46875,
      "learning_rate": 1.6304310538992228e-05,
      "loss": 0.9704,
      "step": 53380
    },
    {
      "epoch": 4.59270412113912,
      "grad_norm": 1.0625,
      "learning_rate": 1.6297462200092093e-05,
      "loss": 0.7663,
      "step": 53381
    },
    {
      "epoch": 4.5927901574464425,
      "grad_norm": 1.2578125,
      "learning_rate": 1.6290615275927844e-05,
      "loss": 1.4915,
      "step": 53382
    },
    {
      "epoch": 4.592876193753764,
      "grad_norm": 1.0546875,
      "learning_rate": 1.6283769766519406e-05,
      "loss": 0.9945,
      "step": 53383
    },
    {
      "epoch": 4.592962230061086,
      "grad_norm": 1.0625,
      "learning_rate": 1.6276925671886932e-05,
      "loss": 1.3441,
      "step": 53384
    },
    {
      "epoch": 4.593048266368408,
      "grad_norm": 0.98046875,
      "learning_rate": 1.6270082992050293e-05,
      "loss": 0.473,
      "step": 53385
    },
    {
      "epoch": 4.593134302675729,
      "grad_norm": 1.0703125,
      "learning_rate": 1.6263241727029588e-05,
      "loss": 0.9168,
      "step": 53386
    },
    {
      "epoch": 4.593220338983051,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6256401876844907e-05,
      "loss": 1.0081,
      "step": 53387
    },
    {
      "epoch": 4.593306375290372,
      "grad_norm": 1.1015625,
      "learning_rate": 1.624956344151607e-05,
      "loss": 0.7755,
      "step": 53388
    },
    {
      "epoch": 4.593392411597694,
      "grad_norm": 2.34375,
      "learning_rate": 1.6242726421063226e-05,
      "loss": 1.3379,
      "step": 53389
    },
    {
      "epoch": 4.5934784479050155,
      "grad_norm": 318.0,
      "learning_rate": 1.6235890815506304e-05,
      "loss": 0.8731,
      "step": 53390
    },
    {
      "epoch": 4.5935644842123375,
      "grad_norm": 1.1953125,
      "learning_rate": 1.6229056624865346e-05,
      "loss": 0.8431,
      "step": 53391
    },
    {
      "epoch": 4.59365052051966,
      "grad_norm": 1.34375,
      "learning_rate": 1.622222384916022e-05,
      "loss": 0.9005,
      "step": 53392
    },
    {
      "epoch": 4.593736556826981,
      "grad_norm": 1.296875,
      "learning_rate": 1.621539248841114e-05,
      "loss": 1.3667,
      "step": 53393
    },
    {
      "epoch": 4.593822593134303,
      "grad_norm": 0.7265625,
      "learning_rate": 1.620856254263786e-05,
      "loss": 0.6969,
      "step": 53394
    },
    {
      "epoch": 4.593908629441624,
      "grad_norm": 1.1640625,
      "learning_rate": 1.620173401186048e-05,
      "loss": 1.5176,
      "step": 53395
    },
    {
      "epoch": 4.593994665748946,
      "grad_norm": 1.0703125,
      "learning_rate": 1.6194906896098872e-05,
      "loss": 0.775,
      "step": 53396
    },
    {
      "epoch": 4.594080702056267,
      "grad_norm": 15.125,
      "learning_rate": 1.6188081195373072e-05,
      "loss": 1.5444,
      "step": 53397
    },
    {
      "epoch": 4.594166738363589,
      "grad_norm": 0.8828125,
      "learning_rate": 1.6181256909703123e-05,
      "loss": 0.7367,
      "step": 53398
    },
    {
      "epoch": 4.594252774670911,
      "grad_norm": 1.2421875,
      "learning_rate": 1.6174434039108845e-05,
      "loss": 2.0772,
      "step": 53399
    },
    {
      "epoch": 4.594338810978233,
      "grad_norm": 1.2265625,
      "learning_rate": 1.6167612583610214e-05,
      "loss": 1.441,
      "step": 53400
    },
    {
      "epoch": 4.594424847285555,
      "grad_norm": 0.78125,
      "learning_rate": 1.6160792543227275e-05,
      "loss": 0.7163,
      "step": 53401
    },
    {
      "epoch": 4.594510883592876,
      "grad_norm": 184.0,
      "learning_rate": 1.615397391797985e-05,
      "loss": 1.019,
      "step": 53402
    },
    {
      "epoch": 4.594596919900198,
      "grad_norm": 1.2109375,
      "learning_rate": 1.6147156707887968e-05,
      "loss": 0.8853,
      "step": 53403
    },
    {
      "epoch": 4.594682956207519,
      "grad_norm": 1.5625,
      "learning_rate": 1.6140340912971563e-05,
      "loss": 1.5876,
      "step": 53404
    },
    {
      "epoch": 4.594768992514841,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6133526533250563e-05,
      "loss": 0.4017,
      "step": 53405
    },
    {
      "epoch": 4.594855028822163,
      "grad_norm": 1.421875,
      "learning_rate": 1.61267135687449e-05,
      "loss": 1.2651,
      "step": 53406
    },
    {
      "epoch": 4.594941065129484,
      "grad_norm": 1.1171875,
      "learning_rate": 1.6119902019474497e-05,
      "loss": 1.0543,
      "step": 53407
    },
    {
      "epoch": 4.5950271014368065,
      "grad_norm": 0.92578125,
      "learning_rate": 1.611309188545923e-05,
      "loss": 1.3146,
      "step": 53408
    },
    {
      "epoch": 4.595113137744128,
      "grad_norm": 0.8046875,
      "learning_rate": 1.610628316671914e-05,
      "loss": 0.4165,
      "step": 53409
    },
    {
      "epoch": 4.59519917405145,
      "grad_norm": 1.09375,
      "learning_rate": 1.6099475863273983e-05,
      "loss": 0.8825,
      "step": 53410
    },
    {
      "epoch": 4.595285210358771,
      "grad_norm": 0.8515625,
      "learning_rate": 1.6092669975143802e-05,
      "loss": 0.6451,
      "step": 53411
    },
    {
      "epoch": 4.595371246666093,
      "grad_norm": 1.078125,
      "learning_rate": 1.6085865502348362e-05,
      "loss": 0.8517,
      "step": 53412
    },
    {
      "epoch": 4.595457282973415,
      "grad_norm": 0.96875,
      "learning_rate": 1.60790624449077e-05,
      "loss": 0.7205,
      "step": 53413
    },
    {
      "epoch": 4.595543319280736,
      "grad_norm": 1.1953125,
      "learning_rate": 1.6072260802841686e-05,
      "loss": 1.5846,
      "step": 53414
    },
    {
      "epoch": 4.595629355588058,
      "grad_norm": 1.125,
      "learning_rate": 1.6065460576170142e-05,
      "loss": 1.3778,
      "step": 53415
    },
    {
      "epoch": 4.5957153918953795,
      "grad_norm": 236.0,
      "learning_rate": 1.6058661764912997e-05,
      "loss": 1.3036,
      "step": 53416
    },
    {
      "epoch": 4.5958014282027015,
      "grad_norm": 1.5,
      "learning_rate": 1.605186436909023e-05,
      "loss": 0.9616,
      "step": 53417
    },
    {
      "epoch": 4.595887464510024,
      "grad_norm": 1.0703125,
      "learning_rate": 1.6045068388721552e-05,
      "loss": 1.2508,
      "step": 53418
    },
    {
      "epoch": 4.595973500817345,
      "grad_norm": 1.390625,
      "learning_rate": 1.6038273823826943e-05,
      "loss": 1.0985,
      "step": 53419
    },
    {
      "epoch": 4.596059537124667,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6031480674426334e-05,
      "loss": 0.4183,
      "step": 53420
    },
    {
      "epoch": 4.596145573431988,
      "grad_norm": 1.375,
      "learning_rate": 1.6024688940539435e-05,
      "loss": 1.1558,
      "step": 53421
    },
    {
      "epoch": 4.59623160973931,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6017898622186167e-05,
      "loss": 0.4366,
      "step": 53422
    },
    {
      "epoch": 4.596317646046632,
      "grad_norm": 1.0078125,
      "learning_rate": 1.601110971938652e-05,
      "loss": 1.2273,
      "step": 53423
    },
    {
      "epoch": 4.596403682353953,
      "grad_norm": 196.0,
      "learning_rate": 1.6004322232160196e-05,
      "loss": 0.5422,
      "step": 53424
    },
    {
      "epoch": 4.596489718661275,
      "grad_norm": 1.421875,
      "learning_rate": 1.5997536160527125e-05,
      "loss": 1.0166,
      "step": 53425
    },
    {
      "epoch": 4.596575754968597,
      "grad_norm": 1.359375,
      "learning_rate": 1.5990751504507073e-05,
      "loss": 1.1104,
      "step": 53426
    },
    {
      "epoch": 4.596661791275919,
      "grad_norm": 1.078125,
      "learning_rate": 1.5983968264119907e-05,
      "loss": 0.8558,
      "step": 53427
    },
    {
      "epoch": 4.59674782758324,
      "grad_norm": 117.0,
      "learning_rate": 1.5977186439385617e-05,
      "loss": 0.7898,
      "step": 53428
    },
    {
      "epoch": 4.596833863890562,
      "grad_norm": 1.5,
      "learning_rate": 1.5970406030323846e-05,
      "loss": 0.4719,
      "step": 53429
    },
    {
      "epoch": 4.596919900197884,
      "grad_norm": 1.15625,
      "learning_rate": 1.5963627036954477e-05,
      "loss": 1.3738,
      "step": 53430
    },
    {
      "epoch": 4.597005936505205,
      "grad_norm": 1.3359375,
      "learning_rate": 1.595684945929743e-05,
      "loss": 0.7359,
      "step": 53431
    },
    {
      "epoch": 4.597091972812527,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5950073297372357e-05,
      "loss": 0.7261,
      "step": 53432
    },
    {
      "epoch": 4.597178009119848,
      "grad_norm": 1.421875,
      "learning_rate": 1.5943298551199248e-05,
      "loss": 1.4529,
      "step": 53433
    },
    {
      "epoch": 4.5972640454271705,
      "grad_norm": 1.0703125,
      "learning_rate": 1.5936525220797858e-05,
      "loss": 1.1444,
      "step": 53434
    },
    {
      "epoch": 4.597350081734492,
      "grad_norm": 1.015625,
      "learning_rate": 1.5929753306187956e-05,
      "loss": 1.4082,
      "step": 53435
    },
    {
      "epoch": 4.597436118041814,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5922982807389407e-05,
      "loss": 0.7047,
      "step": 53436
    },
    {
      "epoch": 4.597522154349136,
      "grad_norm": 1.9765625,
      "learning_rate": 1.5916213724421924e-05,
      "loss": 1.1402,
      "step": 53437
    },
    {
      "epoch": 4.597608190656457,
      "grad_norm": 27.25,
      "learning_rate": 1.590944605730543e-05,
      "loss": 0.7455,
      "step": 53438
    },
    {
      "epoch": 4.597694226963779,
      "grad_norm": 0.8515625,
      "learning_rate": 1.5902679806059638e-05,
      "loss": 0.932,
      "step": 53439
    },
    {
      "epoch": 4.5977802632711,
      "grad_norm": 7040.0,
      "learning_rate": 1.5895914970704362e-05,
      "loss": 1.1029,
      "step": 53440
    },
    {
      "epoch": 4.597866299578422,
      "grad_norm": 209.0,
      "learning_rate": 1.5889151551259308e-05,
      "loss": 1.2419,
      "step": 53441
    },
    {
      "epoch": 4.5979523358857435,
      "grad_norm": 1.1328125,
      "learning_rate": 1.58823895477444e-05,
      "loss": 1.0248,
      "step": 53442
    },
    {
      "epoch": 4.5980383721930655,
      "grad_norm": 1.265625,
      "learning_rate": 1.58756289601793e-05,
      "loss": 0.9771,
      "step": 53443
    },
    {
      "epoch": 4.598124408500388,
      "grad_norm": 1.0078125,
      "learning_rate": 1.586886978858393e-05,
      "loss": 0.9331,
      "step": 53444
    },
    {
      "epoch": 4.598210444807709,
      "grad_norm": 1.1875,
      "learning_rate": 1.586211203297788e-05,
      "loss": 0.8543,
      "step": 53445
    },
    {
      "epoch": 4.598296481115031,
      "grad_norm": 0.90625,
      "learning_rate": 1.585535569338098e-05,
      "loss": 1.3168,
      "step": 53446
    },
    {
      "epoch": 4.598382517422352,
      "grad_norm": 1.0703125,
      "learning_rate": 1.5848600769813093e-05,
      "loss": 0.7289,
      "step": 53447
    },
    {
      "epoch": 4.598468553729674,
      "grad_norm": 0.921875,
      "learning_rate": 1.5841847262293764e-05,
      "loss": 0.9215,
      "step": 53448
    },
    {
      "epoch": 4.598554590036995,
      "grad_norm": 1.2421875,
      "learning_rate": 1.5835095170842917e-05,
      "loss": 1.0531,
      "step": 53449
    },
    {
      "epoch": 4.598640626344317,
      "grad_norm": 1.2421875,
      "learning_rate": 1.5828344495480318e-05,
      "loss": 1.0229,
      "step": 53450
    },
    {
      "epoch": 4.598726662651639,
      "grad_norm": 1.0703125,
      "learning_rate": 1.582159523622556e-05,
      "loss": 1.0691,
      "step": 53451
    },
    {
      "epoch": 4.598812698958961,
      "grad_norm": 1.1171875,
      "learning_rate": 1.5814847393098465e-05,
      "loss": 0.7072,
      "step": 53452
    },
    {
      "epoch": 4.598898735266283,
      "grad_norm": 1.046875,
      "learning_rate": 1.5808100966118843e-05,
      "loss": 1.1233,
      "step": 53453
    },
    {
      "epoch": 4.598984771573604,
      "grad_norm": 1.1796875,
      "learning_rate": 1.580135595530635e-05,
      "loss": 1.3247,
      "step": 53454
    },
    {
      "epoch": 4.599070807880926,
      "grad_norm": 4224.0,
      "learning_rate": 1.5794612360680692e-05,
      "loss": 1.1198,
      "step": 53455
    },
    {
      "epoch": 4.599156844188247,
      "grad_norm": 3552.0,
      "learning_rate": 1.5787870182261632e-05,
      "loss": 1.4199,
      "step": 53456
    },
    {
      "epoch": 4.599242880495569,
      "grad_norm": 0.91015625,
      "learning_rate": 1.578112942006882e-05,
      "loss": 0.8269,
      "step": 53457
    },
    {
      "epoch": 4.599328916802891,
      "grad_norm": 1.4453125,
      "learning_rate": 1.5774390074122124e-05,
      "loss": 1.8872,
      "step": 53458
    },
    {
      "epoch": 4.599414953110212,
      "grad_norm": 0.97265625,
      "learning_rate": 1.576765214444109e-05,
      "loss": 1.2294,
      "step": 53459
    },
    {
      "epoch": 4.5995009894175345,
      "grad_norm": 0.984375,
      "learning_rate": 1.576091563104548e-05,
      "loss": 0.6938,
      "step": 53460
    },
    {
      "epoch": 4.599587025724856,
      "grad_norm": 0.93359375,
      "learning_rate": 1.575418053395511e-05,
      "loss": 0.6286,
      "step": 53461
    },
    {
      "epoch": 4.599673062032178,
      "grad_norm": 1.125,
      "learning_rate": 1.574744685318946e-05,
      "loss": 0.757,
      "step": 53462
    },
    {
      "epoch": 4.599759098339499,
      "grad_norm": 1.4140625,
      "learning_rate": 1.5740714588768412e-05,
      "loss": 1.0307,
      "step": 53463
    },
    {
      "epoch": 4.599845134646821,
      "grad_norm": 0.8125,
      "learning_rate": 1.573398374071161e-05,
      "loss": 0.7056,
      "step": 53464
    },
    {
      "epoch": 4.599931170954143,
      "grad_norm": 1.0234375,
      "learning_rate": 1.572725430903865e-05,
      "loss": 0.92,
      "step": 53465
    },
    {
      "epoch": 4.600017207261464,
      "grad_norm": 0.9921875,
      "learning_rate": 1.572052629376941e-05,
      "loss": 0.6537,
      "step": 53466
    },
    {
      "epoch": 4.600103243568786,
      "grad_norm": 141.0,
      "learning_rate": 1.5713799694923315e-05,
      "loss": 0.6323,
      "step": 53467
    },
    {
      "epoch": 4.6001892798761075,
      "grad_norm": 1.03125,
      "learning_rate": 1.5707074512520236e-05,
      "loss": 0.7437,
      "step": 53468
    },
    {
      "epoch": 4.6002753161834296,
      "grad_norm": 1.2421875,
      "learning_rate": 1.5700350746579772e-05,
      "loss": 0.8459,
      "step": 53469
    },
    {
      "epoch": 4.600361352490751,
      "grad_norm": 2.921875,
      "learning_rate": 1.5693628397121573e-05,
      "loss": 1.4316,
      "step": 53470
    },
    {
      "epoch": 4.600447388798073,
      "grad_norm": 4928.0,
      "learning_rate": 1.568690746416529e-05,
      "loss": 1.3105,
      "step": 53471
    },
    {
      "epoch": 4.600533425105395,
      "grad_norm": 1.1484375,
      "learning_rate": 1.5680187947730684e-05,
      "loss": 1.1039,
      "step": 53472
    },
    {
      "epoch": 4.600619461412716,
      "grad_norm": 0.8828125,
      "learning_rate": 1.567346984783724e-05,
      "loss": 0.7873,
      "step": 53473
    },
    {
      "epoch": 4.600705497720038,
      "grad_norm": 1.4765625,
      "learning_rate": 1.566675316450472e-05,
      "loss": 1.4246,
      "step": 53474
    },
    {
      "epoch": 4.600791534027359,
      "grad_norm": 1.0703125,
      "learning_rate": 1.5660037897752833e-05,
      "loss": 1.6255,
      "step": 53475
    },
    {
      "epoch": 4.600877570334681,
      "grad_norm": 44.75,
      "learning_rate": 1.565332404760106e-05,
      "loss": 1.2978,
      "step": 53476
    },
    {
      "epoch": 4.6009636066420025,
      "grad_norm": 1.078125,
      "learning_rate": 1.5646611614069163e-05,
      "loss": 0.9886,
      "step": 53477
    },
    {
      "epoch": 4.601049642949325,
      "grad_norm": 0.9765625,
      "learning_rate": 1.5639900597176624e-05,
      "loss": 0.9597,
      "step": 53478
    },
    {
      "epoch": 4.601135679256647,
      "grad_norm": 1.3671875,
      "learning_rate": 1.5633190996943214e-05,
      "loss": 1.0623,
      "step": 53479
    },
    {
      "epoch": 4.601221715563968,
      "grad_norm": 0.98828125,
      "learning_rate": 1.5626482813388576e-05,
      "loss": 1.6399,
      "step": 53480
    },
    {
      "epoch": 4.60130775187129,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5619776046532253e-05,
      "loss": 1.1919,
      "step": 53481
    },
    {
      "epoch": 4.601393788178611,
      "grad_norm": 0.84765625,
      "learning_rate": 1.5613070696393893e-05,
      "loss": 0.9009,
      "step": 53482
    },
    {
      "epoch": 4.601479824485933,
      "grad_norm": 1.0234375,
      "learning_rate": 1.560636676299304e-05,
      "loss": 0.8877,
      "step": 53483
    },
    {
      "epoch": 4.601565860793254,
      "grad_norm": 1.40625,
      "learning_rate": 1.5599664246349342e-05,
      "loss": 1.0041,
      "step": 53484
    },
    {
      "epoch": 4.601651897100576,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5592963146482507e-05,
      "loss": 0.4331,
      "step": 53485
    },
    {
      "epoch": 4.6017379334078985,
      "grad_norm": 1.140625,
      "learning_rate": 1.5586263463411965e-05,
      "loss": 1.1164,
      "step": 53486
    },
    {
      "epoch": 4.60182396971522,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5579565197157418e-05,
      "loss": 0.5156,
      "step": 53487
    },
    {
      "epoch": 4.601910006022542,
      "grad_norm": 0.95703125,
      "learning_rate": 1.5572868347738467e-05,
      "loss": 0.8374,
      "step": 53488
    },
    {
      "epoch": 4.601996042329863,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5566172915174593e-05,
      "loss": 0.5334,
      "step": 53489
    },
    {
      "epoch": 4.602082078637185,
      "grad_norm": 1.5625,
      "learning_rate": 1.5559478899485448e-05,
      "loss": 1.0806,
      "step": 53490
    },
    {
      "epoch": 4.602168114944506,
      "grad_norm": 1.015625,
      "learning_rate": 1.555278630069068e-05,
      "loss": 0.9204,
      "step": 53491
    },
    {
      "epoch": 4.602254151251828,
      "grad_norm": 1.2890625,
      "learning_rate": 1.5546095118809723e-05,
      "loss": 0.8386,
      "step": 53492
    },
    {
      "epoch": 4.60234018755915,
      "grad_norm": 1.21875,
      "learning_rate": 1.5539405353862223e-05,
      "loss": 0.506,
      "step": 53493
    },
    {
      "epoch": 4.6024262238664715,
      "grad_norm": 0.921875,
      "learning_rate": 1.553271700586778e-05,
      "loss": 0.7002,
      "step": 53494
    },
    {
      "epoch": 4.602512260173794,
      "grad_norm": 1.171875,
      "learning_rate": 1.5526030074845876e-05,
      "loss": 1.2729,
      "step": 53495
    },
    {
      "epoch": 4.602598296481115,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5519344560816163e-05,
      "loss": 0.4998,
      "step": 53496
    },
    {
      "epoch": 4.602684332788437,
      "grad_norm": 1.1171875,
      "learning_rate": 1.5512660463798066e-05,
      "loss": 1.5774,
      "step": 53497
    },
    {
      "epoch": 4.602770369095758,
      "grad_norm": 1.078125,
      "learning_rate": 1.5505977783811242e-05,
      "loss": 1.2544,
      "step": 53498
    },
    {
      "epoch": 4.60285640540308,
      "grad_norm": 1.171875,
      "learning_rate": 1.5499296520875228e-05,
      "loss": 1.0148,
      "step": 53499
    },
    {
      "epoch": 4.602942441710402,
      "grad_norm": 0.953125,
      "learning_rate": 1.549261667500951e-05,
      "loss": 0.7069,
      "step": 53500
    },
    {
      "epoch": 4.603028478017723,
      "grad_norm": 2.09375,
      "learning_rate": 1.5485938246233678e-05,
      "loss": 1.0468,
      "step": 53501
    },
    {
      "epoch": 4.603114514325045,
      "grad_norm": 0.9765625,
      "learning_rate": 1.5479261234567278e-05,
      "loss": 0.9518,
      "step": 53502
    },
    {
      "epoch": 4.6032005506323666,
      "grad_norm": 692.0,
      "learning_rate": 1.5472585640029734e-05,
      "loss": 1.5401,
      "step": 53503
    },
    {
      "epoch": 4.603286586939689,
      "grad_norm": 0.95703125,
      "learning_rate": 1.5465911462640648e-05,
      "loss": 0.9246,
      "step": 53504
    },
    {
      "epoch": 4.60337262324701,
      "grad_norm": 1.0234375,
      "learning_rate": 1.5459238702419554e-05,
      "loss": 1.2585,
      "step": 53505
    },
    {
      "epoch": 4.603458659554332,
      "grad_norm": 1.171875,
      "learning_rate": 1.5452567359385937e-05,
      "loss": 1.5158,
      "step": 53506
    },
    {
      "epoch": 4.603544695861654,
      "grad_norm": 1.21875,
      "learning_rate": 1.5445897433559397e-05,
      "loss": 0.8865,
      "step": 53507
    },
    {
      "epoch": 4.603630732168975,
      "grad_norm": 1.6171875,
      "learning_rate": 1.5439228924959304e-05,
      "loss": 0.8442,
      "step": 53508
    },
    {
      "epoch": 4.603716768476297,
      "grad_norm": 1.0390625,
      "learning_rate": 1.5432561833605197e-05,
      "loss": 0.5164,
      "step": 53509
    },
    {
      "epoch": 4.603802804783618,
      "grad_norm": 1.4765625,
      "learning_rate": 1.5425896159516673e-05,
      "loss": 1.1778,
      "step": 53510
    },
    {
      "epoch": 4.60388884109094,
      "grad_norm": 1.0078125,
      "learning_rate": 1.541923190271305e-05,
      "loss": 1.4191,
      "step": 53511
    },
    {
      "epoch": 4.603974877398262,
      "grad_norm": 1.4609375,
      "learning_rate": 1.5412569063213978e-05,
      "loss": 1.1438,
      "step": 53512
    },
    {
      "epoch": 4.604060913705584,
      "grad_norm": 2.0625,
      "learning_rate": 1.5405907641038943e-05,
      "loss": 0.9576,
      "step": 53513
    },
    {
      "epoch": 4.604146950012906,
      "grad_norm": 0.984375,
      "learning_rate": 1.539924763620737e-05,
      "loss": 1.1377,
      "step": 53514
    },
    {
      "epoch": 4.604232986320227,
      "grad_norm": 1.2109375,
      "learning_rate": 1.539258904873869e-05,
      "loss": 1.0174,
      "step": 53515
    },
    {
      "epoch": 4.604319022627549,
      "grad_norm": 1.0859375,
      "learning_rate": 1.538593187865245e-05,
      "loss": 0.8733,
      "step": 53516
    },
    {
      "epoch": 4.60440505893487,
      "grad_norm": 1.359375,
      "learning_rate": 1.5379276125968122e-05,
      "loss": 1.0402,
      "step": 53517
    },
    {
      "epoch": 4.604491095242192,
      "grad_norm": 1.09375,
      "learning_rate": 1.53726217907052e-05,
      "loss": 0.7675,
      "step": 53518
    },
    {
      "epoch": 4.604577131549513,
      "grad_norm": 61.25,
      "learning_rate": 1.5365968872883052e-05,
      "loss": 1.0092,
      "step": 53519
    },
    {
      "epoch": 4.6046631678568355,
      "grad_norm": 0.8984375,
      "learning_rate": 1.5359317372521164e-05,
      "loss": 1.0304,
      "step": 53520
    },
    {
      "epoch": 4.604749204164158,
      "grad_norm": 0.8515625,
      "learning_rate": 1.5352667289639134e-05,
      "loss": 0.4798,
      "step": 53521
    },
    {
      "epoch": 4.604835240471479,
      "grad_norm": 1.109375,
      "learning_rate": 1.5346018624256165e-05,
      "loss": 1.144,
      "step": 53522
    },
    {
      "epoch": 4.604921276778801,
      "grad_norm": 1.6015625,
      "learning_rate": 1.533937137639191e-05,
      "loss": 0.9085,
      "step": 53523
    },
    {
      "epoch": 4.605007313086122,
      "grad_norm": 1.171875,
      "learning_rate": 1.5332725546065683e-05,
      "loss": 1.466,
      "step": 53524
    },
    {
      "epoch": 4.605093349393444,
      "grad_norm": 1.9609375,
      "learning_rate": 1.5326081133296975e-05,
      "loss": 1.1246,
      "step": 53525
    },
    {
      "epoch": 4.605179385700765,
      "grad_norm": 1.1015625,
      "learning_rate": 1.5319438138105268e-05,
      "loss": 1.1569,
      "step": 53526
    },
    {
      "epoch": 4.605265422008087,
      "grad_norm": 1.046875,
      "learning_rate": 1.5312796560509934e-05,
      "loss": 0.8107,
      "step": 53527
    },
    {
      "epoch": 4.605351458315409,
      "grad_norm": 0.91796875,
      "learning_rate": 1.5306156400530346e-05,
      "loss": 1.094,
      "step": 53528
    },
    {
      "epoch": 4.6054374946227306,
      "grad_norm": 1.3203125,
      "learning_rate": 1.5299517658186045e-05,
      "loss": 0.9294,
      "step": 53529
    },
    {
      "epoch": 4.605523530930053,
      "grad_norm": 1.046875,
      "learning_rate": 1.5292880333496295e-05,
      "loss": 1.3071,
      "step": 53530
    },
    {
      "epoch": 4.605609567237374,
      "grad_norm": 1.265625,
      "learning_rate": 1.528624442648069e-05,
      "loss": 1.2847,
      "step": 53531
    },
    {
      "epoch": 4.605695603544696,
      "grad_norm": 0.96484375,
      "learning_rate": 1.5279609937158544e-05,
      "loss": 0.6155,
      "step": 53532
    },
    {
      "epoch": 4.605781639852018,
      "grad_norm": 0.8671875,
      "learning_rate": 1.527297686554918e-05,
      "loss": 0.6012,
      "step": 53533
    },
    {
      "epoch": 4.605867676159339,
      "grad_norm": 3856.0,
      "learning_rate": 1.526634521167214e-05,
      "loss": 0.802,
      "step": 53534
    },
    {
      "epoch": 4.605953712466661,
      "grad_norm": 1.1796875,
      "learning_rate": 1.525971497554679e-05,
      "loss": 1.1068,
      "step": 53535
    },
    {
      "epoch": 4.606039748773982,
      "grad_norm": 69.5,
      "learning_rate": 1.5253086157192452e-05,
      "loss": 0.4245,
      "step": 53536
    },
    {
      "epoch": 4.606125785081304,
      "grad_norm": 5.5625,
      "learning_rate": 1.5246458756628667e-05,
      "loss": 1.3348,
      "step": 53537
    },
    {
      "epoch": 4.606211821388626,
      "grad_norm": 233.0,
      "learning_rate": 1.5239832773874585e-05,
      "loss": 0.6822,
      "step": 53538
    },
    {
      "epoch": 4.606297857695948,
      "grad_norm": 0.87109375,
      "learning_rate": 1.5233208208949745e-05,
      "loss": 0.7509,
      "step": 53539
    },
    {
      "epoch": 4.60638389400327,
      "grad_norm": 10.9375,
      "learning_rate": 1.5226585061873522e-05,
      "loss": 1.6312,
      "step": 53540
    },
    {
      "epoch": 4.606469930310591,
      "grad_norm": 1.1484375,
      "learning_rate": 1.5219963332665231e-05,
      "loss": 1.3036,
      "step": 53541
    },
    {
      "epoch": 4.606555966617913,
      "grad_norm": 1.3046875,
      "learning_rate": 1.5213343021344195e-05,
      "loss": 1.7644,
      "step": 53542
    },
    {
      "epoch": 4.606642002925234,
      "grad_norm": 1.1484375,
      "learning_rate": 1.520672412792995e-05,
      "loss": 0.6399,
      "step": 53543
    },
    {
      "epoch": 4.606728039232556,
      "grad_norm": 1.484375,
      "learning_rate": 1.5200106652441703e-05,
      "loss": 1.5554,
      "step": 53544
    },
    {
      "epoch": 4.606814075539878,
      "grad_norm": 0.96875,
      "learning_rate": 1.5193490594898829e-05,
      "loss": 1.0345,
      "step": 53545
    },
    {
      "epoch": 4.6069001118471995,
      "grad_norm": 0.9609375,
      "learning_rate": 1.518687595532081e-05,
      "loss": 0.8323,
      "step": 53546
    },
    {
      "epoch": 4.606986148154522,
      "grad_norm": 1.03125,
      "learning_rate": 1.5180262733726802e-05,
      "loss": 1.5109,
      "step": 53547
    },
    {
      "epoch": 4.607072184461843,
      "grad_norm": 5.65625,
      "learning_rate": 1.5173650930136285e-05,
      "loss": 1.0224,
      "step": 53548
    },
    {
      "epoch": 4.607158220769165,
      "grad_norm": 0.98828125,
      "learning_rate": 1.5167040544568466e-05,
      "loss": 1.4048,
      "step": 53549
    },
    {
      "epoch": 4.607244257076486,
      "grad_norm": 2.546875,
      "learning_rate": 1.5160431577042777e-05,
      "loss": 1.0813,
      "step": 53550
    },
    {
      "epoch": 4.607330293383808,
      "grad_norm": 1.3203125,
      "learning_rate": 1.515382402757859e-05,
      "loss": 1.2949,
      "step": 53551
    },
    {
      "epoch": 4.60741632969113,
      "grad_norm": 1.6015625,
      "learning_rate": 1.5147217896195054e-05,
      "loss": 1.639,
      "step": 53552
    },
    {
      "epoch": 4.607502365998451,
      "grad_norm": 40.75,
      "learning_rate": 1.514061318291171e-05,
      "loss": 0.7161,
      "step": 53553
    },
    {
      "epoch": 4.607588402305773,
      "grad_norm": 0.96484375,
      "learning_rate": 1.5134009887747769e-05,
      "loss": 0.4772,
      "step": 53554
    },
    {
      "epoch": 4.607674438613095,
      "grad_norm": 350.0,
      "learning_rate": 1.5127408010722487e-05,
      "loss": 0.7121,
      "step": 53555
    },
    {
      "epoch": 4.607760474920417,
      "grad_norm": 1.078125,
      "learning_rate": 1.5120807551855298e-05,
      "loss": 0.9496,
      "step": 53556
    },
    {
      "epoch": 4.607846511227738,
      "grad_norm": 0.92578125,
      "learning_rate": 1.5114208511165405e-05,
      "loss": 0.4723,
      "step": 53557
    },
    {
      "epoch": 4.60793254753506,
      "grad_norm": 0.89453125,
      "learning_rate": 1.5107610888672186e-05,
      "loss": 0.8009,
      "step": 53558
    },
    {
      "epoch": 4.608018583842382,
      "grad_norm": 1.15625,
      "learning_rate": 1.5101014684394898e-05,
      "loss": 1.0033,
      "step": 53559
    },
    {
      "epoch": 4.608104620149703,
      "grad_norm": 1.3984375,
      "learning_rate": 1.5094419898352808e-05,
      "loss": 1.3552,
      "step": 53560
    },
    {
      "epoch": 4.608190656457025,
      "grad_norm": 0.92578125,
      "learning_rate": 1.5087826530565174e-05,
      "loss": 0.8796,
      "step": 53561
    },
    {
      "epoch": 4.608276692764346,
      "grad_norm": 1.3984375,
      "learning_rate": 1.5081234581051483e-05,
      "loss": 0.7184,
      "step": 53562
    },
    {
      "epoch": 4.608362729071668,
      "grad_norm": 2.125,
      "learning_rate": 1.5074644049830721e-05,
      "loss": 1.6092,
      "step": 53563
    },
    {
      "epoch": 4.60844876537899,
      "grad_norm": 1.328125,
      "learning_rate": 1.5068054936922426e-05,
      "loss": 1.3966,
      "step": 53564
    },
    {
      "epoch": 4.608534801686312,
      "grad_norm": 1.4453125,
      "learning_rate": 1.5061467242345694e-05,
      "loss": 1.9535,
      "step": 53565
    },
    {
      "epoch": 4.608620837993634,
      "grad_norm": 1.390625,
      "learning_rate": 1.5054880966119844e-05,
      "loss": 0.7351,
      "step": 53566
    },
    {
      "epoch": 4.608706874300955,
      "grad_norm": 1.0546875,
      "learning_rate": 1.5048296108264192e-05,
      "loss": 1.1095,
      "step": 53567
    },
    {
      "epoch": 4.608792910608277,
      "grad_norm": 1.1171875,
      "learning_rate": 1.5041712668797947e-05,
      "loss": 1.5869,
      "step": 53568
    },
    {
      "epoch": 4.608878946915598,
      "grad_norm": 1.015625,
      "learning_rate": 1.5035130647740315e-05,
      "loss": 0.7465,
      "step": 53569
    },
    {
      "epoch": 4.60896498322292,
      "grad_norm": 1.0625,
      "learning_rate": 1.5028550045110722e-05,
      "loss": 1.1269,
      "step": 53570
    },
    {
      "epoch": 4.609051019530241,
      "grad_norm": 1.3671875,
      "learning_rate": 1.5021970860928214e-05,
      "loss": 1.0072,
      "step": 53571
    },
    {
      "epoch": 4.6091370558375635,
      "grad_norm": 1.203125,
      "learning_rate": 1.5015393095212105e-05,
      "loss": 0.7509,
      "step": 53572
    },
    {
      "epoch": 4.609223092144886,
      "grad_norm": 4.71875,
      "learning_rate": 1.5008816747981712e-05,
      "loss": 1.2521,
      "step": 53573
    },
    {
      "epoch": 4.609309128452207,
      "grad_norm": 1.171875,
      "learning_rate": 1.5002241819256135e-05,
      "loss": 1.2921,
      "step": 53574
    },
    {
      "epoch": 4.609395164759529,
      "grad_norm": 1.296875,
      "learning_rate": 1.4995668309054689e-05,
      "loss": 1.0016,
      "step": 53575
    },
    {
      "epoch": 4.60948120106685,
      "grad_norm": 0.8984375,
      "learning_rate": 1.4989096217396636e-05,
      "loss": 0.4842,
      "step": 53576
    },
    {
      "epoch": 4.609567237374172,
      "grad_norm": 1.1171875,
      "learning_rate": 1.4982525544301128e-05,
      "loss": 1.0011,
      "step": 53577
    },
    {
      "epoch": 4.609653273681493,
      "grad_norm": 1.171875,
      "learning_rate": 1.4975956289787429e-05,
      "loss": 0.6895,
      "step": 53578
    },
    {
      "epoch": 4.609739309988815,
      "grad_norm": 1.453125,
      "learning_rate": 1.4969388453874632e-05,
      "loss": 1.3644,
      "step": 53579
    },
    {
      "epoch": 4.609825346296137,
      "grad_norm": 1.109375,
      "learning_rate": 1.4962822036582113e-05,
      "loss": 1.0599,
      "step": 53580
    },
    {
      "epoch": 4.609911382603459,
      "grad_norm": 1.1796875,
      "learning_rate": 1.4956257037929023e-05,
      "loss": 1.129,
      "step": 53581
    },
    {
      "epoch": 4.609997418910781,
      "grad_norm": 0.95703125,
      "learning_rate": 1.4949693457934454e-05,
      "loss": 1.1159,
      "step": 53582
    },
    {
      "epoch": 4.610083455218102,
      "grad_norm": 1.2109375,
      "learning_rate": 1.4943131296617729e-05,
      "loss": 1.7384,
      "step": 53583
    },
    {
      "epoch": 4.610169491525424,
      "grad_norm": 1.078125,
      "learning_rate": 1.4936570553998108e-05,
      "loss": 0.9587,
      "step": 53584
    },
    {
      "epoch": 4.610255527832745,
      "grad_norm": 0.96484375,
      "learning_rate": 1.4930011230094575e-05,
      "loss": 1.1249,
      "step": 53585
    },
    {
      "epoch": 4.610341564140067,
      "grad_norm": 1.5625,
      "learning_rate": 1.4923453324926394e-05,
      "loss": 1.3958,
      "step": 53586
    },
    {
      "epoch": 4.610427600447389,
      "grad_norm": 38.0,
      "learning_rate": 1.4916896838512884e-05,
      "loss": 0.8363,
      "step": 53587
    },
    {
      "epoch": 4.61051363675471,
      "grad_norm": 736.0,
      "learning_rate": 1.4910341770873026e-05,
      "loss": 1.0879,
      "step": 53588
    },
    {
      "epoch": 4.6105996730620324,
      "grad_norm": 2.28125,
      "learning_rate": 1.490378812202603e-05,
      "loss": 1.4537,
      "step": 53589
    },
    {
      "epoch": 4.610685709369354,
      "grad_norm": 24.0,
      "learning_rate": 1.4897235891991157e-05,
      "loss": 1.3679,
      "step": 53590
    },
    {
      "epoch": 4.610771745676676,
      "grad_norm": 1.1171875,
      "learning_rate": 1.4890685080787502e-05,
      "loss": 1.2211,
      "step": 53591
    },
    {
      "epoch": 4.610857781983997,
      "grad_norm": 1.015625,
      "learning_rate": 1.488413568843433e-05,
      "loss": 0.9679,
      "step": 53592
    },
    {
      "epoch": 4.610943818291319,
      "grad_norm": 3.296875,
      "learning_rate": 1.4877587714950625e-05,
      "loss": 0.9974,
      "step": 53593
    },
    {
      "epoch": 4.611029854598641,
      "grad_norm": 10.25,
      "learning_rate": 1.4871041160355703e-05,
      "loss": 1.5294,
      "step": 53594
    },
    {
      "epoch": 4.611115890905962,
      "grad_norm": 5.53125,
      "learning_rate": 1.4864496024668551e-05,
      "loss": 1.7866,
      "step": 53595
    },
    {
      "epoch": 4.611201927213284,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4857952307908429e-05,
      "loss": 1.3991,
      "step": 53596
    },
    {
      "epoch": 4.611287963520605,
      "grad_norm": 1.1953125,
      "learning_rate": 1.485141001009449e-05,
      "loss": 1.0368,
      "step": 53597
    },
    {
      "epoch": 4.6113739998279275,
      "grad_norm": 1.40625,
      "learning_rate": 1.484486913124572e-05,
      "loss": 1.1715,
      "step": 53598
    },
    {
      "epoch": 4.611460036135249,
      "grad_norm": 1.046875,
      "learning_rate": 1.4838329671381378e-05,
      "loss": 1.0663,
      "step": 53599
    },
    {
      "epoch": 4.611546072442571,
      "grad_norm": 1.203125,
      "learning_rate": 1.4831791630520619e-05,
      "loss": 1.2133,
      "step": 53600
    },
    {
      "epoch": 4.611632108749893,
      "grad_norm": 1.0546875,
      "learning_rate": 1.4825255008682481e-05,
      "loss": 1.62,
      "step": 53601
    },
    {
      "epoch": 4.611718145057214,
      "grad_norm": 1.0625,
      "learning_rate": 1.481871980588606e-05,
      "loss": 1.1727,
      "step": 53602
    },
    {
      "epoch": 4.611804181364536,
      "grad_norm": 0.984375,
      "learning_rate": 1.481218602215062e-05,
      "loss": 1.0197,
      "step": 53603
    },
    {
      "epoch": 4.611890217671857,
      "grad_norm": 1.1328125,
      "learning_rate": 1.4805653657495088e-05,
      "loss": 1.3938,
      "step": 53604
    },
    {
      "epoch": 4.611976253979179,
      "grad_norm": 1.265625,
      "learning_rate": 1.4799122711938674e-05,
      "loss": 1.3264,
      "step": 53605
    },
    {
      "epoch": 4.6120622902865005,
      "grad_norm": 0.96875,
      "learning_rate": 1.4792593185500524e-05,
      "loss": 0.9318,
      "step": 53606
    },
    {
      "epoch": 4.612148326593823,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4786065078199572e-05,
      "loss": 1.2473,
      "step": 53607
    },
    {
      "epoch": 4.612234362901145,
      "grad_norm": 1.2109375,
      "learning_rate": 1.4779538390055135e-05,
      "loss": 0.9369,
      "step": 53608
    },
    {
      "epoch": 4.612320399208466,
      "grad_norm": 1.1484375,
      "learning_rate": 1.4773013121086087e-05,
      "loss": 1.8232,
      "step": 53609
    },
    {
      "epoch": 4.612406435515788,
      "grad_norm": 0.9609375,
      "learning_rate": 1.4766489271311578e-05,
      "loss": 0.4591,
      "step": 53610
    },
    {
      "epoch": 4.612492471823109,
      "grad_norm": 0.984375,
      "learning_rate": 1.4759966840750815e-05,
      "loss": 0.6353,
      "step": 53611
    },
    {
      "epoch": 4.612578508130431,
      "grad_norm": 0.8984375,
      "learning_rate": 1.4753445829422673e-05,
      "loss": 0.7759,
      "step": 53612
    },
    {
      "epoch": 4.612664544437752,
      "grad_norm": 1.09375,
      "learning_rate": 1.4746926237346359e-05,
      "loss": 1.6342,
      "step": 53613
    },
    {
      "epoch": 4.612750580745074,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4740408064540967e-05,
      "loss": 0.9782,
      "step": 53614
    },
    {
      "epoch": 4.6128366170523964,
      "grad_norm": 1.2109375,
      "learning_rate": 1.4733891311025426e-05,
      "loss": 1.3907,
      "step": 53615
    },
    {
      "epoch": 4.612922653359718,
      "grad_norm": 1.9375,
      "learning_rate": 1.472737597681889e-05,
      "loss": 0.9641,
      "step": 53616
    },
    {
      "epoch": 4.61300868966704,
      "grad_norm": 1.9765625,
      "learning_rate": 1.4720862061940454e-05,
      "loss": 1.6318,
      "step": 53617
    },
    {
      "epoch": 4.613094725974361,
      "grad_norm": 1.46875,
      "learning_rate": 1.4714349566409046e-05,
      "loss": 0.7168,
      "step": 53618
    },
    {
      "epoch": 4.613180762281683,
      "grad_norm": 8.0625,
      "learning_rate": 1.4707838490243819e-05,
      "loss": 1.7696,
      "step": 53619
    },
    {
      "epoch": 4.613266798589004,
      "grad_norm": 2.453125,
      "learning_rate": 1.4701328833463757e-05,
      "loss": 1.2153,
      "step": 53620
    },
    {
      "epoch": 4.613352834896326,
      "grad_norm": 1.1953125,
      "learning_rate": 1.4694820596087954e-05,
      "loss": 0.8998,
      "step": 53621
    },
    {
      "epoch": 4.613438871203648,
      "grad_norm": 1.359375,
      "learning_rate": 1.4688313778135398e-05,
      "loss": 1.6964,
      "step": 53622
    },
    {
      "epoch": 4.613524907510969,
      "grad_norm": 1.203125,
      "learning_rate": 1.4681808379625128e-05,
      "loss": 1.0983,
      "step": 53623
    },
    {
      "epoch": 4.6136109438182915,
      "grad_norm": 0.9375,
      "learning_rate": 1.4675304400576183e-05,
      "loss": 0.635,
      "step": 53624
    },
    {
      "epoch": 4.613696980125613,
      "grad_norm": 2.890625,
      "learning_rate": 1.4668801841007606e-05,
      "loss": 1.0361,
      "step": 53625
    },
    {
      "epoch": 4.613783016432935,
      "grad_norm": 1.328125,
      "learning_rate": 1.4662300700938324e-05,
      "loss": 1.1178,
      "step": 53626
    },
    {
      "epoch": 4.613869052740256,
      "grad_norm": 2.03125,
      "learning_rate": 1.465580098038749e-05,
      "loss": 0.7543,
      "step": 53627
    },
    {
      "epoch": 4.613955089047578,
      "grad_norm": 1.4140625,
      "learning_rate": 1.4649302679373977e-05,
      "loss": 1.254,
      "step": 53628
    },
    {
      "epoch": 4.6140411253549,
      "grad_norm": 3.515625,
      "learning_rate": 1.464280579791688e-05,
      "loss": 1.0527,
      "step": 53629
    },
    {
      "epoch": 4.614127161662221,
      "grad_norm": 14784.0,
      "learning_rate": 1.4636310336035241e-05,
      "loss": 1.0158,
      "step": 53630
    },
    {
      "epoch": 4.614213197969543,
      "grad_norm": 1.671875,
      "learning_rate": 1.4629816293747933e-05,
      "loss": 1.1113,
      "step": 53631
    },
    {
      "epoch": 4.6142992342768645,
      "grad_norm": 1.1796875,
      "learning_rate": 1.4623323671074052e-05,
      "loss": 1.1414,
      "step": 53632
    },
    {
      "epoch": 4.614385270584187,
      "grad_norm": 1.1640625,
      "learning_rate": 1.4616832468032581e-05,
      "loss": 0.8661,
      "step": 53633
    },
    {
      "epoch": 4.614471306891508,
      "grad_norm": 1.0703125,
      "learning_rate": 1.4610342684642397e-05,
      "loss": 0.798,
      "step": 53634
    },
    {
      "epoch": 4.61455734319883,
      "grad_norm": 33.25,
      "learning_rate": 1.4603854320922595e-05,
      "loss": 1.1316,
      "step": 53635
    },
    {
      "epoch": 4.614643379506152,
      "grad_norm": 1.0859375,
      "learning_rate": 1.45973673768921e-05,
      "loss": 1.0482,
      "step": 53636
    },
    {
      "epoch": 4.614729415813473,
      "grad_norm": 1.4296875,
      "learning_rate": 1.4590881852569904e-05,
      "loss": 1.9371,
      "step": 53637
    },
    {
      "epoch": 4.614815452120795,
      "grad_norm": 1.03125,
      "learning_rate": 1.4584397747974987e-05,
      "loss": 1.715,
      "step": 53638
    },
    {
      "epoch": 4.614901488428116,
      "grad_norm": 1.09375,
      "learning_rate": 1.457791506312628e-05,
      "loss": 1.1398,
      "step": 53639
    },
    {
      "epoch": 4.614987524735438,
      "grad_norm": 0.98828125,
      "learning_rate": 1.4571433798042767e-05,
      "loss": 0.7426,
      "step": 53640
    },
    {
      "epoch": 4.61507356104276,
      "grad_norm": 21.625,
      "learning_rate": 1.4564953952743432e-05,
      "loss": 0.6104,
      "step": 53641
    },
    {
      "epoch": 4.615159597350082,
      "grad_norm": 0.8515625,
      "learning_rate": 1.455847552724715e-05,
      "loss": 0.2674,
      "step": 53642
    },
    {
      "epoch": 4.615245633657404,
      "grad_norm": 12.25,
      "learning_rate": 1.4551998521572907e-05,
      "loss": 1.1801,
      "step": 53643
    },
    {
      "epoch": 4.615331669964725,
      "grad_norm": 1.140625,
      "learning_rate": 1.4545522935739686e-05,
      "loss": 1.5543,
      "step": 53644
    },
    {
      "epoch": 4.615417706272047,
      "grad_norm": 103.0,
      "learning_rate": 1.453904876976636e-05,
      "loss": 0.8826,
      "step": 53645
    },
    {
      "epoch": 4.615503742579368,
      "grad_norm": 0.90625,
      "learning_rate": 1.4532576023671918e-05,
      "loss": 1.2157,
      "step": 53646
    },
    {
      "epoch": 4.61558977888669,
      "grad_norm": 0.8515625,
      "learning_rate": 1.4526104697475285e-05,
      "loss": 1.0783,
      "step": 53647
    },
    {
      "epoch": 4.615675815194011,
      "grad_norm": 3.328125,
      "learning_rate": 1.451963479119539e-05,
      "loss": 1.1931,
      "step": 53648
    },
    {
      "epoch": 4.6157618515013334,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4513166304851166e-05,
      "loss": 0.5509,
      "step": 53649
    },
    {
      "epoch": 4.6158478878086555,
      "grad_norm": 516.0,
      "learning_rate": 1.450669923846143e-05,
      "loss": 1.4414,
      "step": 53650
    },
    {
      "epoch": 4.615933924115977,
      "grad_norm": 1.2265625,
      "learning_rate": 1.4500233592045165e-05,
      "loss": 1.5837,
      "step": 53651
    },
    {
      "epoch": 4.616019960423299,
      "grad_norm": 1.0078125,
      "learning_rate": 1.4493769365621413e-05,
      "loss": 1.4757,
      "step": 53652
    },
    {
      "epoch": 4.61610599673062,
      "grad_norm": 1.359375,
      "learning_rate": 1.4487306559208823e-05,
      "loss": 1.1058,
      "step": 53653
    },
    {
      "epoch": 4.616192033037942,
      "grad_norm": 1.078125,
      "learning_rate": 1.4480845172826495e-05,
      "loss": 0.8646,
      "step": 53654
    },
    {
      "epoch": 4.616278069345264,
      "grad_norm": 1.265625,
      "learning_rate": 1.44743852064933e-05,
      "loss": 0.7974,
      "step": 53655
    },
    {
      "epoch": 4.616364105652585,
      "grad_norm": 1.4765625,
      "learning_rate": 1.4467926660228059e-05,
      "loss": 1.1188,
      "step": 53656
    },
    {
      "epoch": 4.616450141959907,
      "grad_norm": 8.8125,
      "learning_rate": 1.4461469534049698e-05,
      "loss": 1.2316,
      "step": 53657
    },
    {
      "epoch": 4.6165361782672285,
      "grad_norm": 1.15625,
      "learning_rate": 1.4455013827977147e-05,
      "loss": 1.3285,
      "step": 53658
    },
    {
      "epoch": 4.616622214574551,
      "grad_norm": 1.3515625,
      "learning_rate": 1.4448559542029172e-05,
      "loss": 1.1104,
      "step": 53659
    },
    {
      "epoch": 4.616708250881873,
      "grad_norm": 1.265625,
      "learning_rate": 1.444210667622481e-05,
      "loss": 1.8901,
      "step": 53660
    },
    {
      "epoch": 4.616794287189194,
      "grad_norm": 1.1328125,
      "learning_rate": 1.4435655230582822e-05,
      "loss": 1.6727,
      "step": 53661
    },
    {
      "epoch": 4.616880323496516,
      "grad_norm": 1.0703125,
      "learning_rate": 1.4429205205122031e-05,
      "loss": 1.5232,
      "step": 53662
    },
    {
      "epoch": 4.616966359803837,
      "grad_norm": 36.75,
      "learning_rate": 1.442275659986142e-05,
      "loss": 0.9734,
      "step": 53663
    },
    {
      "epoch": 4.617052396111159,
      "grad_norm": 1.2109375,
      "learning_rate": 1.4416309414819806e-05,
      "loss": 0.8091,
      "step": 53664
    },
    {
      "epoch": 4.61713843241848,
      "grad_norm": 1.28125,
      "learning_rate": 1.4409863650016119e-05,
      "loss": 0.5014,
      "step": 53665
    },
    {
      "epoch": 4.617224468725802,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4403419305469068e-05,
      "loss": 1.079,
      "step": 53666
    },
    {
      "epoch": 4.6173105050331245,
      "grad_norm": 1.1796875,
      "learning_rate": 1.439697638119758e-05,
      "loss": 0.6217,
      "step": 53667
    },
    {
      "epoch": 4.617396541340446,
      "grad_norm": 0.99609375,
      "learning_rate": 1.439053487722053e-05,
      "loss": 0.531,
      "step": 53668
    },
    {
      "epoch": 4.617482577647768,
      "grad_norm": 1.3671875,
      "learning_rate": 1.438409479355668e-05,
      "loss": 0.9799,
      "step": 53669
    },
    {
      "epoch": 4.617568613955089,
      "grad_norm": 0.9453125,
      "learning_rate": 1.4377656130224903e-05,
      "loss": 0.6268,
      "step": 53670
    },
    {
      "epoch": 4.617654650262411,
      "grad_norm": 1.1171875,
      "learning_rate": 1.4371218887244074e-05,
      "loss": 0.5062,
      "step": 53671
    },
    {
      "epoch": 4.617740686569732,
      "grad_norm": 0.625,
      "learning_rate": 1.4364783064632958e-05,
      "loss": 0.7681,
      "step": 53672
    },
    {
      "epoch": 4.617826722877054,
      "grad_norm": 0.80078125,
      "learning_rate": 1.435834866241037e-05,
      "loss": 0.6256,
      "step": 53673
    },
    {
      "epoch": 4.617912759184376,
      "grad_norm": 1.1328125,
      "learning_rate": 1.4351915680595185e-05,
      "loss": 0.8394,
      "step": 53674
    },
    {
      "epoch": 4.6179987954916975,
      "grad_norm": 1.0859375,
      "learning_rate": 1.4345484119206221e-05,
      "loss": 1.6378,
      "step": 53675
    },
    {
      "epoch": 4.6180848317990195,
      "grad_norm": 1.265625,
      "learning_rate": 1.433905397826224e-05,
      "loss": 1.6497,
      "step": 53676
    },
    {
      "epoch": 4.618170868106341,
      "grad_norm": 3.203125,
      "learning_rate": 1.433262525778206e-05,
      "loss": 0.9606,
      "step": 53677
    },
    {
      "epoch": 4.618256904413663,
      "grad_norm": 1.1640625,
      "learning_rate": 1.4326197957784504e-05,
      "loss": 1.0762,
      "step": 53678
    },
    {
      "epoch": 4.618342940720984,
      "grad_norm": 5.59375,
      "learning_rate": 1.4319772078288384e-05,
      "loss": 1.0723,
      "step": 53679
    },
    {
      "epoch": 4.618428977028306,
      "grad_norm": 1.4921875,
      "learning_rate": 1.4313347619312411e-05,
      "loss": 1.6907,
      "step": 53680
    },
    {
      "epoch": 4.618515013335628,
      "grad_norm": 6.875,
      "learning_rate": 1.4306924580875459e-05,
      "loss": 1.1155,
      "step": 53681
    },
    {
      "epoch": 4.618601049642949,
      "grad_norm": 820.0,
      "learning_rate": 1.4300502962996342e-05,
      "loss": 0.8189,
      "step": 53682
    },
    {
      "epoch": 4.618687085950271,
      "grad_norm": 0.9453125,
      "learning_rate": 1.4294082765693717e-05,
      "loss": 1.3091,
      "step": 53683
    },
    {
      "epoch": 4.6187731222575925,
      "grad_norm": 0.78125,
      "learning_rate": 1.4287663988986454e-05,
      "loss": 0.7466,
      "step": 53684
    },
    {
      "epoch": 4.618859158564915,
      "grad_norm": 1.0,
      "learning_rate": 1.4281246632893319e-05,
      "loss": 0.9344,
      "step": 53685
    },
    {
      "epoch": 4.618945194872236,
      "grad_norm": 41.25,
      "learning_rate": 1.4274830697433073e-05,
      "loss": 1.5136,
      "step": 53686
    },
    {
      "epoch": 4.619031231179558,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4268416182624477e-05,
      "loss": 0.8359,
      "step": 53687
    },
    {
      "epoch": 4.61911726748688,
      "grad_norm": 0.91015625,
      "learning_rate": 1.4262003088486298e-05,
      "loss": 0.7161,
      "step": 53688
    },
    {
      "epoch": 4.619203303794201,
      "grad_norm": 0.94140625,
      "learning_rate": 1.4255591415037294e-05,
      "loss": 0.7352,
      "step": 53689
    },
    {
      "epoch": 4.619289340101523,
      "grad_norm": 1.2421875,
      "learning_rate": 1.4249181162296233e-05,
      "loss": 0.5884,
      "step": 53690
    },
    {
      "epoch": 4.619375376408844,
      "grad_norm": 1.578125,
      "learning_rate": 1.4242772330281818e-05,
      "loss": 0.9088,
      "step": 53691
    },
    {
      "epoch": 4.619461412716166,
      "grad_norm": 1.0234375,
      "learning_rate": 1.4236364919012756e-05,
      "loss": 0.9158,
      "step": 53692
    },
    {
      "epoch": 4.619547449023488,
      "grad_norm": 4.5,
      "learning_rate": 1.422995892850798e-05,
      "loss": 0.9783,
      "step": 53693
    },
    {
      "epoch": 4.61963348533081,
      "grad_norm": 1.46875,
      "learning_rate": 1.4223554358785973e-05,
      "loss": 1.7673,
      "step": 53694
    },
    {
      "epoch": 4.619719521638132,
      "grad_norm": 0.9453125,
      "learning_rate": 1.4217151209865665e-05,
      "loss": 1.3743,
      "step": 53695
    },
    {
      "epoch": 4.619805557945453,
      "grad_norm": 0.97265625,
      "learning_rate": 1.4210749481765706e-05,
      "loss": 0.8005,
      "step": 53696
    },
    {
      "epoch": 4.619891594252775,
      "grad_norm": 0.9296875,
      "learning_rate": 1.4204349174504804e-05,
      "loss": 1.0991,
      "step": 53697
    },
    {
      "epoch": 4.619977630560096,
      "grad_norm": 1.234375,
      "learning_rate": 1.4197950288101669e-05,
      "loss": 1.5749,
      "step": 53698
    },
    {
      "epoch": 4.620063666867418,
      "grad_norm": 1.03125,
      "learning_rate": 1.4191552822575115e-05,
      "loss": 1.1558,
      "step": 53699
    },
    {
      "epoch": 4.620149703174739,
      "grad_norm": 1.234375,
      "learning_rate": 1.4185156777943686e-05,
      "loss": 0.7622,
      "step": 53700
    },
    {
      "epoch": 4.6202357394820615,
      "grad_norm": 0.83203125,
      "learning_rate": 1.417876215422631e-05,
      "loss": 0.7786,
      "step": 53701
    },
    {
      "epoch": 4.6203217757893835,
      "grad_norm": 1.203125,
      "learning_rate": 1.4172368951441472e-05,
      "loss": 1.2556,
      "step": 53702
    },
    {
      "epoch": 4.620407812096705,
      "grad_norm": 1.1328125,
      "learning_rate": 1.416597716960799e-05,
      "loss": 1.2405,
      "step": 53703
    },
    {
      "epoch": 4.620493848404027,
      "grad_norm": 1.2890625,
      "learning_rate": 1.4159586808744574e-05,
      "loss": 1.5446,
      "step": 53704
    },
    {
      "epoch": 4.620579884711348,
      "grad_norm": 179.0,
      "learning_rate": 1.4153197868869872e-05,
      "loss": 1.2594,
      "step": 53705
    },
    {
      "epoch": 4.62066592101867,
      "grad_norm": 1.078125,
      "learning_rate": 1.4146810350002593e-05,
      "loss": 0.8593,
      "step": 53706
    },
    {
      "epoch": 4.620751957325991,
      "grad_norm": 1.2578125,
      "learning_rate": 1.4140424252161332e-05,
      "loss": 1.3768,
      "step": 53707
    },
    {
      "epoch": 4.620837993633313,
      "grad_norm": 1.109375,
      "learning_rate": 1.4134039575364855e-05,
      "loss": 0.5523,
      "step": 53708
    },
    {
      "epoch": 4.620924029940635,
      "grad_norm": 1.21875,
      "learning_rate": 1.4127656319631865e-05,
      "loss": 0.9152,
      "step": 53709
    },
    {
      "epoch": 4.6210100662479565,
      "grad_norm": 1.3125,
      "learning_rate": 1.4121274484980906e-05,
      "loss": 0.925,
      "step": 53710
    },
    {
      "epoch": 4.621096102555279,
      "grad_norm": 1.015625,
      "learning_rate": 1.4114894071430739e-05,
      "loss": 1.2607,
      "step": 53711
    },
    {
      "epoch": 4.6211821388626,
      "grad_norm": 0.859375,
      "learning_rate": 1.4108515079000073e-05,
      "loss": 0.4421,
      "step": 53712
    },
    {
      "epoch": 4.621268175169922,
      "grad_norm": 0.94140625,
      "learning_rate": 1.4102137507707447e-05,
      "loss": 0.7521,
      "step": 53713
    },
    {
      "epoch": 4.621354211477243,
      "grad_norm": 3.53125,
      "learning_rate": 1.4095761357571569e-05,
      "loss": 0.5505,
      "step": 53714
    },
    {
      "epoch": 4.621440247784565,
      "grad_norm": 1.03125,
      "learning_rate": 1.408938662861109e-05,
      "loss": 0.9787,
      "step": 53715
    },
    {
      "epoch": 4.621526284091887,
      "grad_norm": 1.015625,
      "learning_rate": 1.4083013320844662e-05,
      "loss": 1.5747,
      "step": 53716
    },
    {
      "epoch": 4.621612320399208,
      "grad_norm": 1.078125,
      "learning_rate": 1.4076641434290882e-05,
      "loss": 1.207,
      "step": 53717
    },
    {
      "epoch": 4.62169835670653,
      "grad_norm": 1.0078125,
      "learning_rate": 1.4070270968968457e-05,
      "loss": 0.5796,
      "step": 53718
    },
    {
      "epoch": 4.621784393013852,
      "grad_norm": 474.0,
      "learning_rate": 1.4063901924895984e-05,
      "loss": 0.827,
      "step": 53719
    },
    {
      "epoch": 4.621870429321174,
      "grad_norm": 1.3203125,
      "learning_rate": 1.4057534302092112e-05,
      "loss": 1.4251,
      "step": 53720
    },
    {
      "epoch": 4.621956465628495,
      "grad_norm": 0.875,
      "learning_rate": 1.4051168100575384e-05,
      "loss": 0.525,
      "step": 53721
    },
    {
      "epoch": 4.622042501935817,
      "grad_norm": 1.3671875,
      "learning_rate": 1.4044803320364452e-05,
      "loss": 2.0407,
      "step": 53722
    },
    {
      "epoch": 4.622128538243139,
      "grad_norm": 1.3515625,
      "learning_rate": 1.4038439961478023e-05,
      "loss": 1.6749,
      "step": 53723
    },
    {
      "epoch": 4.62221457455046,
      "grad_norm": 1.359375,
      "learning_rate": 1.4032078023934636e-05,
      "loss": 0.9544,
      "step": 53724
    },
    {
      "epoch": 4.622300610857782,
      "grad_norm": 0.96875,
      "learning_rate": 1.4025717507752833e-05,
      "loss": 1.0843,
      "step": 53725
    },
    {
      "epoch": 4.622386647165103,
      "grad_norm": 1.390625,
      "learning_rate": 1.4019358412951378e-05,
      "loss": 1.4363,
      "step": 53726
    },
    {
      "epoch": 4.6224726834724255,
      "grad_norm": 1.1328125,
      "learning_rate": 1.4013000739548753e-05,
      "loss": 0.8274,
      "step": 53727
    },
    {
      "epoch": 4.622558719779747,
      "grad_norm": 1.09375,
      "learning_rate": 1.4006644487563558e-05,
      "loss": 0.528,
      "step": 53728
    },
    {
      "epoch": 4.622644756087069,
      "grad_norm": 0.8828125,
      "learning_rate": 1.4000289657014442e-05,
      "loss": 0.8748,
      "step": 53729
    },
    {
      "epoch": 4.622730792394391,
      "grad_norm": 1.75,
      "learning_rate": 1.3993936247919947e-05,
      "loss": 1.5912,
      "step": 53730
    },
    {
      "epoch": 4.622816828701712,
      "grad_norm": 1.3515625,
      "learning_rate": 1.3987584260298669e-05,
      "loss": 0.6161,
      "step": 53731
    },
    {
      "epoch": 4.622902865009034,
      "grad_norm": 1.5625,
      "learning_rate": 1.3981233694169204e-05,
      "loss": 1.2132,
      "step": 53732
    },
    {
      "epoch": 4.622988901316355,
      "grad_norm": 1.2734375,
      "learning_rate": 1.3974884549550038e-05,
      "loss": 1.567,
      "step": 53733
    },
    {
      "epoch": 4.623074937623677,
      "grad_norm": 1.34375,
      "learning_rate": 1.3968536826459876e-05,
      "loss": 1.0941,
      "step": 53734
    },
    {
      "epoch": 4.6231609739309985,
      "grad_norm": 1.1171875,
      "learning_rate": 1.396219052491715e-05,
      "loss": 1.5843,
      "step": 53735
    },
    {
      "epoch": 4.6232470102383205,
      "grad_norm": 1.140625,
      "learning_rate": 1.3955845644940513e-05,
      "loss": 0.9026,
      "step": 53736
    },
    {
      "epoch": 4.623333046545643,
      "grad_norm": 1.15625,
      "learning_rate": 1.3949502186548501e-05,
      "loss": 0.5551,
      "step": 53737
    },
    {
      "epoch": 4.623419082852964,
      "grad_norm": 44.75,
      "learning_rate": 1.394316014975966e-05,
      "loss": 0.8238,
      "step": 53738
    },
    {
      "epoch": 4.623505119160286,
      "grad_norm": 2.234375,
      "learning_rate": 1.3936819534592582e-05,
      "loss": 0.7686,
      "step": 53739
    },
    {
      "epoch": 4.623591155467607,
      "grad_norm": 1.265625,
      "learning_rate": 1.3930480341065699e-05,
      "loss": 0.7063,
      "step": 53740
    },
    {
      "epoch": 4.623677191774929,
      "grad_norm": 1.3984375,
      "learning_rate": 1.3924142569197663e-05,
      "loss": 0.6194,
      "step": 53741
    },
    {
      "epoch": 4.62376322808225,
      "grad_norm": 1.28125,
      "learning_rate": 1.3917806219007012e-05,
      "loss": 1.4529,
      "step": 53742
    },
    {
      "epoch": 4.623849264389572,
      "grad_norm": 1.1484375,
      "learning_rate": 1.3911471290512178e-05,
      "loss": 0.5724,
      "step": 53743
    },
    {
      "epoch": 4.623935300696894,
      "grad_norm": 34.5,
      "learning_rate": 1.3905137783731813e-05,
      "loss": 0.9342,
      "step": 53744
    },
    {
      "epoch": 4.624021337004216,
      "grad_norm": 1.140625,
      "learning_rate": 1.3898805698684348e-05,
      "loss": 0.5934,
      "step": 53745
    },
    {
      "epoch": 4.624107373311538,
      "grad_norm": 1.28125,
      "learning_rate": 1.389247503538832e-05,
      "loss": 0.8362,
      "step": 53746
    },
    {
      "epoch": 4.624193409618859,
      "grad_norm": 1.21875,
      "learning_rate": 1.3886145793862326e-05,
      "loss": 0.9508,
      "step": 53747
    },
    {
      "epoch": 4.624279445926181,
      "grad_norm": 1.4140625,
      "learning_rate": 1.3879817974124742e-05,
      "loss": 0.932,
      "step": 53748
    },
    {
      "epoch": 4.624365482233502,
      "grad_norm": 1.265625,
      "learning_rate": 1.3873491576194163e-05,
      "loss": 1.046,
      "step": 53749
    },
    {
      "epoch": 4.624451518540824,
      "grad_norm": 0.95703125,
      "learning_rate": 1.3867166600089076e-05,
      "loss": 0.6444,
      "step": 53750
    },
    {
      "epoch": 4.624537554848146,
      "grad_norm": 1.375,
      "learning_rate": 1.3860843045827964e-05,
      "loss": 1.2676,
      "step": 53751
    },
    {
      "epoch": 4.624623591155467,
      "grad_norm": 0.96484375,
      "learning_rate": 1.3854520913429369e-05,
      "loss": 1.3158,
      "step": 53752
    },
    {
      "epoch": 4.6247096274627895,
      "grad_norm": 1.0625,
      "learning_rate": 1.384820020291172e-05,
      "loss": 1.4189,
      "step": 53753
    },
    {
      "epoch": 4.624795663770111,
      "grad_norm": 1.140625,
      "learning_rate": 1.384188091429356e-05,
      "loss": 1.0808,
      "step": 53754
    },
    {
      "epoch": 4.624881700077433,
      "grad_norm": 1.2109375,
      "learning_rate": 1.3835563047593313e-05,
      "loss": 0.9685,
      "step": 53755
    },
    {
      "epoch": 4.624967736384754,
      "grad_norm": 1.1796875,
      "learning_rate": 1.3829246602829526e-05,
      "loss": 1.191,
      "step": 53756
    },
    {
      "epoch": 4.625053772692076,
      "grad_norm": 0.984375,
      "learning_rate": 1.3822931580020626e-05,
      "loss": 0.9511,
      "step": 53757
    },
    {
      "epoch": 4.625139808999398,
      "grad_norm": 1.2421875,
      "learning_rate": 1.3816617979185097e-05,
      "loss": 1.4035,
      "step": 53758
    },
    {
      "epoch": 4.625225845306719,
      "grad_norm": 2.84375,
      "learning_rate": 1.3810305800341428e-05,
      "loss": 0.8316,
      "step": 53759
    },
    {
      "epoch": 4.625311881614041,
      "grad_norm": 1.015625,
      "learning_rate": 1.3803995043507988e-05,
      "loss": 1.1069,
      "step": 53760
    },
    {
      "epoch": 4.6253979179213625,
      "grad_norm": 1.0234375,
      "learning_rate": 1.3797685708703377e-05,
      "loss": 0.8414,
      "step": 53761
    },
    {
      "epoch": 4.6254839542286845,
      "grad_norm": 1.0625,
      "learning_rate": 1.3791377795945914e-05,
      "loss": 1.0758,
      "step": 53762
    },
    {
      "epoch": 4.625569990536006,
      "grad_norm": 0.91015625,
      "learning_rate": 1.3785071305254137e-05,
      "loss": 0.6261,
      "step": 53763
    },
    {
      "epoch": 4.625656026843328,
      "grad_norm": 0.98828125,
      "learning_rate": 1.3778766236646478e-05,
      "loss": 1.2359,
      "step": 53764
    },
    {
      "epoch": 4.62574206315065,
      "grad_norm": 1.6484375,
      "learning_rate": 1.377246259014131e-05,
      "loss": 0.7531,
      "step": 53765
    },
    {
      "epoch": 4.625828099457971,
      "grad_norm": 0.92578125,
      "learning_rate": 1.3766160365757174e-05,
      "loss": 1.8184,
      "step": 53766
    },
    {
      "epoch": 4.625914135765293,
      "grad_norm": 0.8671875,
      "learning_rate": 1.3759859563512444e-05,
      "loss": 1.4596,
      "step": 53767
    },
    {
      "epoch": 4.626000172072614,
      "grad_norm": 0.91796875,
      "learning_rate": 1.3753560183425551e-05,
      "loss": 0.8986,
      "step": 53768
    },
    {
      "epoch": 4.626086208379936,
      "grad_norm": 1.140625,
      "learning_rate": 1.3747262225514867e-05,
      "loss": 0.9531,
      "step": 53769
    },
    {
      "epoch": 4.626172244687258,
      "grad_norm": 1.125,
      "learning_rate": 1.374096568979899e-05,
      "loss": 1.5582,
      "step": 53770
    },
    {
      "epoch": 4.62625828099458,
      "grad_norm": 0.90234375,
      "learning_rate": 1.3734670576296127e-05,
      "loss": 0.5885,
      "step": 53771
    },
    {
      "epoch": 4.626344317301902,
      "grad_norm": 0.96875,
      "learning_rate": 1.3728376885024818e-05,
      "loss": 0.893,
      "step": 53772
    },
    {
      "epoch": 4.626430353609223,
      "grad_norm": 1.1875,
      "learning_rate": 1.3722084616003383e-05,
      "loss": 0.8527,
      "step": 53773
    },
    {
      "epoch": 4.626516389916545,
      "grad_norm": 1.0234375,
      "learning_rate": 1.3715793769250251e-05,
      "loss": 0.8816,
      "step": 53774
    },
    {
      "epoch": 4.626602426223867,
      "grad_norm": 1.4375,
      "learning_rate": 1.3709504344783908e-05,
      "loss": 1.6557,
      "step": 53775
    },
    {
      "epoch": 4.626688462531188,
      "grad_norm": 1.1796875,
      "learning_rate": 1.370321634262267e-05,
      "loss": 0.8684,
      "step": 53776
    },
    {
      "epoch": 4.62677449883851,
      "grad_norm": 458.0,
      "learning_rate": 1.3696929762785026e-05,
      "loss": 1.5637,
      "step": 53777
    },
    {
      "epoch": 4.626860535145831,
      "grad_norm": 1.1875,
      "learning_rate": 1.3690644605289183e-05,
      "loss": 1.1552,
      "step": 53778
    },
    {
      "epoch": 4.6269465714531535,
      "grad_norm": 1.140625,
      "learning_rate": 1.3684360870153622e-05,
      "loss": 0.869,
      "step": 53779
    },
    {
      "epoch": 4.627032607760475,
      "grad_norm": 1.328125,
      "learning_rate": 1.3678078557396779e-05,
      "loss": 1.0825,
      "step": 53780
    },
    {
      "epoch": 4.627118644067797,
      "grad_norm": 1.109375,
      "learning_rate": 1.3671797667036911e-05,
      "loss": 1.1413,
      "step": 53781
    },
    {
      "epoch": 4.627204680375119,
      "grad_norm": 78.0,
      "learning_rate": 1.3665518199092452e-05,
      "loss": 0.9753,
      "step": 53782
    },
    {
      "epoch": 4.62729071668244,
      "grad_norm": 1.171875,
      "learning_rate": 1.3659240153581831e-05,
      "loss": 0.716,
      "step": 53783
    },
    {
      "epoch": 4.627376752989762,
      "grad_norm": 0.8515625,
      "learning_rate": 1.365296353052331e-05,
      "loss": 0.8188,
      "step": 53784
    },
    {
      "epoch": 4.627462789297083,
      "grad_norm": 1.1953125,
      "learning_rate": 1.3646688329935264e-05,
      "loss": 0.9172,
      "step": 53785
    },
    {
      "epoch": 4.627548825604405,
      "grad_norm": 204.0,
      "learning_rate": 1.3640414551836123e-05,
      "loss": 1.5227,
      "step": 53786
    },
    {
      "epoch": 4.6276348619117265,
      "grad_norm": 1.234375,
      "learning_rate": 1.3634142196244092e-05,
      "loss": 1.2745,
      "step": 53787
    },
    {
      "epoch": 4.6277208982190485,
      "grad_norm": 1.6484375,
      "learning_rate": 1.3627871263177715e-05,
      "loss": 0.893,
      "step": 53788
    },
    {
      "epoch": 4.627806934526371,
      "grad_norm": 1.125,
      "learning_rate": 1.3621601752655089e-05,
      "loss": 0.9018,
      "step": 53789
    },
    {
      "epoch": 4.627892970833692,
      "grad_norm": 1.0,
      "learning_rate": 1.3615333664694752e-05,
      "loss": 2.1061,
      "step": 53790
    },
    {
      "epoch": 4.627979007141014,
      "grad_norm": 0.921875,
      "learning_rate": 1.3609066999314967e-05,
      "loss": 0.4934,
      "step": 53791
    },
    {
      "epoch": 4.628065043448335,
      "grad_norm": 31.875,
      "learning_rate": 1.3602801756534056e-05,
      "loss": 0.5409,
      "step": 53792
    },
    {
      "epoch": 4.628151079755657,
      "grad_norm": 480.0,
      "learning_rate": 1.3596537936370334e-05,
      "loss": 1.1627,
      "step": 53793
    },
    {
      "epoch": 4.628237116062978,
      "grad_norm": 1.171875,
      "learning_rate": 1.3590275538842179e-05,
      "loss": 1.2063,
      "step": 53794
    },
    {
      "epoch": 4.6283231523703,
      "grad_norm": 0.88671875,
      "learning_rate": 1.358401456396785e-05,
      "loss": 0.75,
      "step": 53795
    },
    {
      "epoch": 4.628409188677622,
      "grad_norm": 1.3515625,
      "learning_rate": 1.3577755011765614e-05,
      "loss": 1.5162,
      "step": 53796
    },
    {
      "epoch": 4.628495224984944,
      "grad_norm": 1.171875,
      "learning_rate": 1.3571496882253952e-05,
      "loss": 1.1954,
      "step": 53797
    },
    {
      "epoch": 4.628581261292266,
      "grad_norm": 1.25,
      "learning_rate": 1.3565240175450965e-05,
      "loss": 0.9953,
      "step": 53798
    },
    {
      "epoch": 4.628667297599587,
      "grad_norm": 1.0390625,
      "learning_rate": 1.3558984891375081e-05,
      "loss": 0.6467,
      "step": 53799
    },
    {
      "epoch": 4.628753333906909,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3552731030044563e-05,
      "loss": 1.6951,
      "step": 53800
    },
    {
      "epoch": 4.62883937021423,
      "grad_norm": 0.92578125,
      "learning_rate": 1.3546478591477674e-05,
      "loss": 0.6681,
      "step": 53801
    },
    {
      "epoch": 4.628925406521552,
      "grad_norm": 1.0078125,
      "learning_rate": 1.3540227575692787e-05,
      "loss": 1.2294,
      "step": 53802
    },
    {
      "epoch": 4.629011442828874,
      "grad_norm": 1.390625,
      "learning_rate": 1.3533977982708057e-05,
      "loss": 0.8709,
      "step": 53803
    },
    {
      "epoch": 4.629097479136195,
      "grad_norm": 0.9296875,
      "learning_rate": 1.3527729812541856e-05,
      "loss": 0.8271,
      "step": 53804
    },
    {
      "epoch": 4.6291835154435175,
      "grad_norm": 1.1484375,
      "learning_rate": 1.3521483065212448e-05,
      "loss": 1.1618,
      "step": 53805
    },
    {
      "epoch": 4.629269551750839,
      "grad_norm": 1.078125,
      "learning_rate": 1.351523774073804e-05,
      "loss": 0.9736,
      "step": 53806
    },
    {
      "epoch": 4.629355588058161,
      "grad_norm": 1.90625,
      "learning_rate": 1.3508993839136952e-05,
      "loss": 1.0347,
      "step": 53807
    },
    {
      "epoch": 4.629441624365482,
      "grad_norm": 1.3203125,
      "learning_rate": 1.3502751360427501e-05,
      "loss": 1.0414,
      "step": 53808
    },
    {
      "epoch": 4.629527660672804,
      "grad_norm": 27.75,
      "learning_rate": 1.349651030462784e-05,
      "loss": 1.1204,
      "step": 53809
    },
    {
      "epoch": 4.629613696980126,
      "grad_norm": 28.75,
      "learning_rate": 1.3490270671756233e-05,
      "loss": 1.9003,
      "step": 53810
    },
    {
      "epoch": 4.629699733287447,
      "grad_norm": 5.9375,
      "learning_rate": 1.3484032461830997e-05,
      "loss": 0.9794,
      "step": 53811
    },
    {
      "epoch": 4.629785769594769,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3477795674870285e-05,
      "loss": 0.7237,
      "step": 53812
    },
    {
      "epoch": 4.6298718059020905,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3471560310892527e-05,
      "loss": 1.4241,
      "step": 53813
    },
    {
      "epoch": 4.6299578422094125,
      "grad_norm": 0.8515625,
      "learning_rate": 1.346532636991571e-05,
      "loss": 0.5032,
      "step": 53814
    },
    {
      "epoch": 4.630043878516734,
      "grad_norm": 9.6875,
      "learning_rate": 1.345909385195826e-05,
      "loss": 0.6966,
      "step": 53815
    },
    {
      "epoch": 4.630129914824056,
      "grad_norm": 1.03125,
      "learning_rate": 1.3452862757038332e-05,
      "loss": 0.9795,
      "step": 53816
    },
    {
      "epoch": 4.630215951131378,
      "grad_norm": 1.1171875,
      "learning_rate": 1.3446633085174131e-05,
      "loss": 0.8875,
      "step": 53817
    },
    {
      "epoch": 4.630301987438699,
      "grad_norm": 1.6953125,
      "learning_rate": 1.3440404836383924e-05,
      "loss": 0.9397,
      "step": 53818
    },
    {
      "epoch": 4.630388023746021,
      "grad_norm": 1.5625,
      "learning_rate": 1.343417801068586e-05,
      "loss": 1.1424,
      "step": 53819
    },
    {
      "epoch": 4.630474060053342,
      "grad_norm": 1.421875,
      "learning_rate": 1.3427952608098203e-05,
      "loss": 0.8034,
      "step": 53820
    },
    {
      "epoch": 4.630560096360664,
      "grad_norm": 1.4140625,
      "learning_rate": 1.3421728628639163e-05,
      "loss": 0.8297,
      "step": 53821
    },
    {
      "epoch": 4.6306461326679855,
      "grad_norm": 0.921875,
      "learning_rate": 1.3415506072326944e-05,
      "loss": 1.1389,
      "step": 53822
    },
    {
      "epoch": 4.630732168975308,
      "grad_norm": 0.97265625,
      "learning_rate": 1.34092849391797e-05,
      "loss": 0.6941,
      "step": 53823
    },
    {
      "epoch": 4.63081820528263,
      "grad_norm": 1.296875,
      "learning_rate": 1.3403065229215694e-05,
      "loss": 1.0415,
      "step": 53824
    },
    {
      "epoch": 4.630904241589951,
      "grad_norm": 0.9375,
      "learning_rate": 1.3396846942453078e-05,
      "loss": 1.4288,
      "step": 53825
    },
    {
      "epoch": 4.630990277897273,
      "grad_norm": 1.046875,
      "learning_rate": 1.3390630078910005e-05,
      "loss": 0.7826,
      "step": 53826
    },
    {
      "epoch": 4.631076314204594,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3384414638604736e-05,
      "loss": 0.8076,
      "step": 53827
    },
    {
      "epoch": 4.631162350511916,
      "grad_norm": 1.5078125,
      "learning_rate": 1.3378200621555425e-05,
      "loss": 1.5705,
      "step": 53828
    },
    {
      "epoch": 4.631248386819237,
      "grad_norm": 0.94921875,
      "learning_rate": 1.3371988027780225e-05,
      "loss": 0.7866,
      "step": 53829
    },
    {
      "epoch": 4.631334423126559,
      "grad_norm": 1.46875,
      "learning_rate": 1.3365776857297285e-05,
      "loss": 0.6782,
      "step": 53830
    },
    {
      "epoch": 4.6314204594338815,
      "grad_norm": 1.5546875,
      "learning_rate": 1.3359567110124815e-05,
      "loss": 0.921,
      "step": 53831
    },
    {
      "epoch": 4.631506495741203,
      "grad_norm": 1.3203125,
      "learning_rate": 1.3353358786281022e-05,
      "loss": 0.6669,
      "step": 53832
    },
    {
      "epoch": 4.631592532048525,
      "grad_norm": 0.984375,
      "learning_rate": 1.3347151885783948e-05,
      "loss": 1.1092,
      "step": 53833
    },
    {
      "epoch": 4.631678568355846,
      "grad_norm": 0.8984375,
      "learning_rate": 1.33409464086518e-05,
      "loss": 0.5803,
      "step": 53834
    },
    {
      "epoch": 4.631764604663168,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3334742354902784e-05,
      "loss": 1.1868,
      "step": 53835
    },
    {
      "epoch": 4.631850640970489,
      "grad_norm": 0.98046875,
      "learning_rate": 1.3328539724554944e-05,
      "loss": 0.951,
      "step": 53836
    },
    {
      "epoch": 4.631936677277811,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3322338517626486e-05,
      "loss": 0.941,
      "step": 53837
    },
    {
      "epoch": 4.632022713585133,
      "grad_norm": 1.234375,
      "learning_rate": 1.3316138734135563e-05,
      "loss": 0.7839,
      "step": 53838
    },
    {
      "epoch": 4.6321087498924545,
      "grad_norm": 0.90625,
      "learning_rate": 1.330994037410027e-05,
      "loss": 1.2095,
      "step": 53839
    },
    {
      "epoch": 4.6321947861997765,
      "grad_norm": 0.91796875,
      "learning_rate": 1.3303743437538706e-05,
      "loss": 1.1798,
      "step": 53840
    },
    {
      "epoch": 4.632280822507098,
      "grad_norm": 20.375,
      "learning_rate": 1.3297547924469078e-05,
      "loss": 1.0205,
      "step": 53841
    },
    {
      "epoch": 4.63236685881442,
      "grad_norm": 1.4375,
      "learning_rate": 1.329135383490948e-05,
      "loss": 1.039,
      "step": 53842
    },
    {
      "epoch": 4.632452895121741,
      "grad_norm": 0.93359375,
      "learning_rate": 1.3285161168878013e-05,
      "loss": 1.1763,
      "step": 53843
    },
    {
      "epoch": 4.632538931429063,
      "grad_norm": 1.0625,
      "learning_rate": 1.3278969926392714e-05,
      "loss": 1.0071,
      "step": 53844
    },
    {
      "epoch": 4.632624967736385,
      "grad_norm": 1.078125,
      "learning_rate": 1.3272780107471794e-05,
      "loss": 0.9039,
      "step": 53845
    },
    {
      "epoch": 4.632711004043706,
      "grad_norm": 1.0390625,
      "learning_rate": 1.3266591712133402e-05,
      "loss": 0.8901,
      "step": 53846
    },
    {
      "epoch": 4.632797040351028,
      "grad_norm": 6.90625,
      "learning_rate": 1.3260404740395416e-05,
      "loss": 1.3024,
      "step": 53847
    },
    {
      "epoch": 4.6328830766583495,
      "grad_norm": 1.34375,
      "learning_rate": 1.3254219192276208e-05,
      "loss": 1.2629,
      "step": 53848
    },
    {
      "epoch": 4.632969112965672,
      "grad_norm": 2.53125,
      "learning_rate": 1.3248035067793707e-05,
      "loss": 1.3275,
      "step": 53849
    },
    {
      "epoch": 4.633055149272993,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3241852366966011e-05,
      "loss": 0.7728,
      "step": 53850
    },
    {
      "epoch": 4.633141185580315,
      "grad_norm": 1.109375,
      "learning_rate": 1.3235671089811274e-05,
      "loss": 1.7187,
      "step": 53851
    },
    {
      "epoch": 4.633227221887637,
      "grad_norm": 0.8203125,
      "learning_rate": 1.322949123634748e-05,
      "loss": 0.4872,
      "step": 53852
    },
    {
      "epoch": 4.633313258194958,
      "grad_norm": 1.625,
      "learning_rate": 1.322331280659278e-05,
      "loss": 1.7612,
      "step": 53853
    },
    {
      "epoch": 4.63339929450228,
      "grad_norm": 1.0078125,
      "learning_rate": 1.3217135800565216e-05,
      "loss": 0.661,
      "step": 53854
    },
    {
      "epoch": 4.633485330809601,
      "grad_norm": 1.078125,
      "learning_rate": 1.3210960218282886e-05,
      "loss": 0.8498,
      "step": 53855
    },
    {
      "epoch": 4.633571367116923,
      "grad_norm": 0.9453125,
      "learning_rate": 1.3204786059763774e-05,
      "loss": 1.1731,
      "step": 53856
    },
    {
      "epoch": 4.633657403424245,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3198613325026032e-05,
      "loss": 1.0093,
      "step": 53857
    },
    {
      "epoch": 4.633743439731567,
      "grad_norm": 9.1875,
      "learning_rate": 1.3192442014087592e-05,
      "loss": 1.0984,
      "step": 53858
    },
    {
      "epoch": 4.633829476038889,
      "grad_norm": 0.90234375,
      "learning_rate": 1.3186272126966659e-05,
      "loss": 0.9023,
      "step": 53859
    },
    {
      "epoch": 4.63391551234621,
      "grad_norm": 1.0625,
      "learning_rate": 1.3180103663681164e-05,
      "loss": 1.3482,
      "step": 53860
    },
    {
      "epoch": 4.634001548653532,
      "grad_norm": 88.0,
      "learning_rate": 1.317393662424915e-05,
      "loss": 1.0069,
      "step": 53861
    },
    {
      "epoch": 4.634087584960853,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3167771008688768e-05,
      "loss": 0.738,
      "step": 53862
    },
    {
      "epoch": 4.634173621268175,
      "grad_norm": 1.2578125,
      "learning_rate": 1.3161606817017946e-05,
      "loss": 1.3046,
      "step": 53863
    },
    {
      "epoch": 4.634259657575496,
      "grad_norm": 0.875,
      "learning_rate": 1.3155444049254729e-05,
      "loss": 0.9431,
      "step": 53864
    },
    {
      "epoch": 4.6343456938828185,
      "grad_norm": 1.1796875,
      "learning_rate": 1.3149282705417153e-05,
      "loss": 0.9257,
      "step": 53865
    },
    {
      "epoch": 4.6344317301901405,
      "grad_norm": 1.28125,
      "learning_rate": 1.314312278552321e-05,
      "loss": 1.5869,
      "step": 53866
    },
    {
      "epoch": 4.634517766497462,
      "grad_norm": 1.015625,
      "learning_rate": 1.3136964289590935e-05,
      "loss": 1.0717,
      "step": 53867
    },
    {
      "epoch": 4.634603802804784,
      "grad_norm": 0.93359375,
      "learning_rate": 1.313080721763843e-05,
      "loss": 0.8447,
      "step": 53868
    },
    {
      "epoch": 4.634689839112105,
      "grad_norm": 32.25,
      "learning_rate": 1.3124651569683565e-05,
      "loss": 0.5756,
      "step": 53869
    },
    {
      "epoch": 4.634775875419427,
      "grad_norm": 1.5,
      "learning_rate": 1.3118497345744385e-05,
      "loss": 1.0521,
      "step": 53870
    },
    {
      "epoch": 4.634861911726748,
      "grad_norm": 52.25,
      "learning_rate": 1.3112344545838984e-05,
      "loss": 0.6513,
      "step": 53871
    },
    {
      "epoch": 4.63494794803407,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3106193169985236e-05,
      "loss": 1.103,
      "step": 53872
    },
    {
      "epoch": 4.635033984341392,
      "grad_norm": 6.28125,
      "learning_rate": 1.3100043218201185e-05,
      "loss": 1.3538,
      "step": 53873
    },
    {
      "epoch": 4.6351200206487135,
      "grad_norm": 1.796875,
      "learning_rate": 1.3093894690504815e-05,
      "loss": 1.3691,
      "step": 53874
    },
    {
      "epoch": 4.635206056956036,
      "grad_norm": 1.0625,
      "learning_rate": 1.3087747586914056e-05,
      "loss": 0.9435,
      "step": 53875
    },
    {
      "epoch": 4.635292093263357,
      "grad_norm": 1.2734375,
      "learning_rate": 1.3081601907447004e-05,
      "loss": 1.3752,
      "step": 53876
    },
    {
      "epoch": 4.635378129570679,
      "grad_norm": 1.875,
      "learning_rate": 1.307545765212148e-05,
      "loss": 1.0843,
      "step": 53877
    },
    {
      "epoch": 4.635464165878,
      "grad_norm": 7.375,
      "learning_rate": 1.306931482095558e-05,
      "loss": 1.1685,
      "step": 53878
    },
    {
      "epoch": 4.635550202185322,
      "grad_norm": 0.98046875,
      "learning_rate": 1.3063173413967288e-05,
      "loss": 1.2571,
      "step": 53879
    },
    {
      "epoch": 4.635636238492644,
      "grad_norm": 468.0,
      "learning_rate": 1.305703343117448e-05,
      "loss": 1.1556,
      "step": 53880
    },
    {
      "epoch": 4.635722274799965,
      "grad_norm": 1.25,
      "learning_rate": 1.3050894872595088e-05,
      "loss": 0.7761,
      "step": 53881
    },
    {
      "epoch": 4.635808311107287,
      "grad_norm": 1.8984375,
      "learning_rate": 1.3044757738247203e-05,
      "loss": 1.3479,
      "step": 53882
    },
    {
      "epoch": 4.635894347414609,
      "grad_norm": 1.234375,
      "learning_rate": 1.303862202814865e-05,
      "loss": 1.5776,
      "step": 53883
    },
    {
      "epoch": 4.635980383721931,
      "grad_norm": 0.9140625,
      "learning_rate": 1.3032487742317412e-05,
      "loss": 0.5923,
      "step": 53884
    },
    {
      "epoch": 4.636066420029253,
      "grad_norm": 3.6875,
      "learning_rate": 1.3026354880771418e-05,
      "loss": 1.504,
      "step": 53885
    },
    {
      "epoch": 4.636152456336574,
      "grad_norm": 0.88671875,
      "learning_rate": 1.3020223443528656e-05,
      "loss": 1.0271,
      "step": 53886
    },
    {
      "epoch": 4.636238492643896,
      "grad_norm": 1.234375,
      "learning_rate": 1.3014093430607055e-05,
      "loss": 0.7019,
      "step": 53887
    },
    {
      "epoch": 4.636324528951217,
      "grad_norm": 0.953125,
      "learning_rate": 1.3007964842024489e-05,
      "loss": 1.0343,
      "step": 53888
    },
    {
      "epoch": 4.636410565258539,
      "grad_norm": 0.86328125,
      "learning_rate": 1.3001837677798945e-05,
      "loss": 0.4163,
      "step": 53889
    },
    {
      "epoch": 4.63649660156586,
      "grad_norm": 0.93359375,
      "learning_rate": 1.2995711937948241e-05,
      "loss": 0.8895,
      "step": 53890
    },
    {
      "epoch": 4.6365826378731825,
      "grad_norm": 1.578125,
      "learning_rate": 1.2989587622490362e-05,
      "loss": 1.3697,
      "step": 53891
    },
    {
      "epoch": 4.6366686741805045,
      "grad_norm": 1.015625,
      "learning_rate": 1.2983464731443296e-05,
      "loss": 1.2659,
      "step": 53892
    },
    {
      "epoch": 4.636754710487826,
      "grad_norm": 1.3359375,
      "learning_rate": 1.2977343264824859e-05,
      "loss": 1.593,
      "step": 53893
    },
    {
      "epoch": 4.636840746795148,
      "grad_norm": 1.09375,
      "learning_rate": 1.2971223222652929e-05,
      "loss": 0.9739,
      "step": 53894
    },
    {
      "epoch": 4.636926783102469,
      "grad_norm": 1.4375,
      "learning_rate": 1.2965104604945488e-05,
      "loss": 1.7509,
      "step": 53895
    },
    {
      "epoch": 4.637012819409791,
      "grad_norm": 1.390625,
      "learning_rate": 1.2958987411720358e-05,
      "loss": 1.6602,
      "step": 53896
    },
    {
      "epoch": 4.637098855717113,
      "grad_norm": 2.015625,
      "learning_rate": 1.2952871642995468e-05,
      "loss": 0.9399,
      "step": 53897
    },
    {
      "epoch": 4.637184892024434,
      "grad_norm": 1.2265625,
      "learning_rate": 1.2946757298788748e-05,
      "loss": 1.2763,
      "step": 53898
    },
    {
      "epoch": 4.637270928331756,
      "grad_norm": 1.1328125,
      "learning_rate": 1.2940644379118017e-05,
      "loss": 0.9983,
      "step": 53899
    },
    {
      "epoch": 4.6373569646390775,
      "grad_norm": 1.6796875,
      "learning_rate": 1.2934532884001148e-05,
      "loss": 1.0709,
      "step": 53900
    },
    {
      "epoch": 4.6374430009464,
      "grad_norm": 1.3828125,
      "learning_rate": 1.2928422813456075e-05,
      "loss": 1.579,
      "step": 53901
    },
    {
      "epoch": 4.637529037253721,
      "grad_norm": 556.0,
      "learning_rate": 1.2922314167500614e-05,
      "loss": 0.8744,
      "step": 53902
    },
    {
      "epoch": 4.637615073561043,
      "grad_norm": 1.234375,
      "learning_rate": 1.2916206946152643e-05,
      "loss": 0.8855,
      "step": 53903
    },
    {
      "epoch": 4.637701109868365,
      "grad_norm": 1.0625,
      "learning_rate": 1.2910101149430032e-05,
      "loss": 0.5378,
      "step": 53904
    },
    {
      "epoch": 4.637787146175686,
      "grad_norm": 1.265625,
      "learning_rate": 1.2903996777350656e-05,
      "loss": 1.6476,
      "step": 53905
    },
    {
      "epoch": 4.637873182483008,
      "grad_norm": 4.6875,
      "learning_rate": 1.2897893829932394e-05,
      "loss": 1.3214,
      "step": 53906
    },
    {
      "epoch": 4.637959218790329,
      "grad_norm": 1.0625,
      "learning_rate": 1.2891792307193006e-05,
      "loss": 1.3708,
      "step": 53907
    },
    {
      "epoch": 4.638045255097651,
      "grad_norm": 1.71875,
      "learning_rate": 1.2885692209150369e-05,
      "loss": 0.997,
      "step": 53908
    },
    {
      "epoch": 4.638131291404973,
      "grad_norm": 1.125,
      "learning_rate": 1.2879593535822409e-05,
      "loss": 0.7808,
      "step": 53909
    },
    {
      "epoch": 4.638217327712295,
      "grad_norm": 1.0078125,
      "learning_rate": 1.2873496287226838e-05,
      "loss": 1.038,
      "step": 53910
    },
    {
      "epoch": 4.638303364019617,
      "grad_norm": 1.078125,
      "learning_rate": 1.2867400463381584e-05,
      "loss": 0.931,
      "step": 53911
    },
    {
      "epoch": 4.638389400326938,
      "grad_norm": 1.6015625,
      "learning_rate": 1.2861306064304468e-05,
      "loss": 1.18,
      "step": 53912
    },
    {
      "epoch": 4.63847543663426,
      "grad_norm": 118.5,
      "learning_rate": 1.2855213090013252e-05,
      "loss": 1.4599,
      "step": 53913
    },
    {
      "epoch": 4.638561472941581,
      "grad_norm": 20.375,
      "learning_rate": 1.284912154052581e-05,
      "loss": 1.2096,
      "step": 53914
    },
    {
      "epoch": 4.638647509248903,
      "grad_norm": 0.875,
      "learning_rate": 1.2843031415859908e-05,
      "loss": 0.8135,
      "step": 53915
    },
    {
      "epoch": 4.638733545556224,
      "grad_norm": 1.765625,
      "learning_rate": 1.2836942716033417e-05,
      "loss": 0.9133,
      "step": 53916
    },
    {
      "epoch": 4.6388195818635465,
      "grad_norm": 6.5,
      "learning_rate": 1.2830855441064104e-05,
      "loss": 1.0995,
      "step": 53917
    },
    {
      "epoch": 4.6389056181708686,
      "grad_norm": 1.40625,
      "learning_rate": 1.2824769590969786e-05,
      "loss": 0.7908,
      "step": 53918
    },
    {
      "epoch": 4.63899165447819,
      "grad_norm": 1.0703125,
      "learning_rate": 1.2818685165768285e-05,
      "loss": 0.723,
      "step": 53919
    },
    {
      "epoch": 4.639077690785512,
      "grad_norm": 1.65625,
      "learning_rate": 1.2812602165477416e-05,
      "loss": 1.574,
      "step": 53920
    },
    {
      "epoch": 4.639163727092833,
      "grad_norm": 1.2421875,
      "learning_rate": 1.2806520590114834e-05,
      "loss": 0.9609,
      "step": 53921
    },
    {
      "epoch": 4.639249763400155,
      "grad_norm": 1.1796875,
      "learning_rate": 1.2800440439698469e-05,
      "loss": 0.9975,
      "step": 53922
    },
    {
      "epoch": 4.639335799707476,
      "grad_norm": 1.453125,
      "learning_rate": 1.2794361714246083e-05,
      "loss": 1.5201,
      "step": 53923
    },
    {
      "epoch": 4.639421836014798,
      "grad_norm": 1.0078125,
      "learning_rate": 1.2788284413775387e-05,
      "loss": 1.4606,
      "step": 53924
    },
    {
      "epoch": 4.63950787232212,
      "grad_norm": 1.1875,
      "learning_rate": 1.2782208538304308e-05,
      "loss": 1.6424,
      "step": 53925
    },
    {
      "epoch": 4.6395939086294415,
      "grad_norm": 1.125,
      "learning_rate": 1.2776134087850389e-05,
      "loss": 0.9476,
      "step": 53926
    },
    {
      "epoch": 4.639679944936764,
      "grad_norm": 1.2421875,
      "learning_rate": 1.277006106243156e-05,
      "loss": 0.9773,
      "step": 53927
    },
    {
      "epoch": 4.639765981244085,
      "grad_norm": 7.375,
      "learning_rate": 1.2763989462065583e-05,
      "loss": 0.6821,
      "step": 53928
    },
    {
      "epoch": 4.639852017551407,
      "grad_norm": 62.0,
      "learning_rate": 1.2757919286770115e-05,
      "loss": 1.2112,
      "step": 53929
    },
    {
      "epoch": 4.639938053858728,
      "grad_norm": 1.015625,
      "learning_rate": 1.275185053656308e-05,
      "loss": 0.7195,
      "step": 53930
    },
    {
      "epoch": 4.64002409016605,
      "grad_norm": 1.0625,
      "learning_rate": 1.274578321146197e-05,
      "loss": 0.5294,
      "step": 53931
    },
    {
      "epoch": 4.640110126473372,
      "grad_norm": 1.265625,
      "learning_rate": 1.2739717311484766e-05,
      "loss": 0.8314,
      "step": 53932
    },
    {
      "epoch": 4.640196162780693,
      "grad_norm": 0.84765625,
      "learning_rate": 1.2733652836649123e-05,
      "loss": 0.7863,
      "step": 53933
    },
    {
      "epoch": 4.640282199088015,
      "grad_norm": 1.0,
      "learning_rate": 1.2727589786972748e-05,
      "loss": 0.7162,
      "step": 53934
    },
    {
      "epoch": 4.640368235395337,
      "grad_norm": 1.1171875,
      "learning_rate": 1.2721528162473406e-05,
      "loss": 0.9835,
      "step": 53935
    },
    {
      "epoch": 4.640454271702659,
      "grad_norm": 1.4140625,
      "learning_rate": 1.2715467963168859e-05,
      "loss": 1.4613,
      "step": 53936
    },
    {
      "epoch": 4.64054030800998,
      "grad_norm": 4.75,
      "learning_rate": 1.2709409189076704e-05,
      "loss": 1.0383,
      "step": 53937
    },
    {
      "epoch": 4.640626344317302,
      "grad_norm": 1.8984375,
      "learning_rate": 1.2703351840214817e-05,
      "loss": 0.9334,
      "step": 53938
    },
    {
      "epoch": 4.640712380624624,
      "grad_norm": 1.1953125,
      "learning_rate": 1.2697295916600904e-05,
      "loss": 1.0378,
      "step": 53939
    },
    {
      "epoch": 4.640798416931945,
      "grad_norm": 1.25,
      "learning_rate": 1.2691241418252564e-05,
      "loss": 0.8091,
      "step": 53940
    },
    {
      "epoch": 4.640884453239267,
      "grad_norm": 95.0,
      "learning_rate": 1.268518834518756e-05,
      "loss": 0.9055,
      "step": 53941
    },
    {
      "epoch": 4.640970489546588,
      "grad_norm": 1.2734375,
      "learning_rate": 1.2679136697423599e-05,
      "loss": 0.9898,
      "step": 53942
    },
    {
      "epoch": 4.6410565258539105,
      "grad_norm": 0.87109375,
      "learning_rate": 1.267308647497839e-05,
      "loss": 0.6228,
      "step": 53943
    },
    {
      "epoch": 4.641142562161232,
      "grad_norm": 0.99609375,
      "learning_rate": 1.2667037677869697e-05,
      "loss": 1.227,
      "step": 53944
    },
    {
      "epoch": 4.641228598468554,
      "grad_norm": 1.953125,
      "learning_rate": 1.2660990306115062e-05,
      "loss": 0.8474,
      "step": 53945
    },
    {
      "epoch": 4.641314634775876,
      "grad_norm": 1.3515625,
      "learning_rate": 1.2654944359732245e-05,
      "loss": 1.4969,
      "step": 53946
    },
    {
      "epoch": 4.641400671083197,
      "grad_norm": 1.328125,
      "learning_rate": 1.2648899838738959e-05,
      "loss": 0.914,
      "step": 53947
    },
    {
      "epoch": 4.641486707390519,
      "grad_norm": 1.125,
      "learning_rate": 1.2642856743152852e-05,
      "loss": 1.0003,
      "step": 53948
    },
    {
      "epoch": 4.64157274369784,
      "grad_norm": 0.98046875,
      "learning_rate": 1.263681507299158e-05,
      "loss": 1.0369,
      "step": 53949
    },
    {
      "epoch": 4.641658780005162,
      "grad_norm": 0.9375,
      "learning_rate": 1.2630774828272906e-05,
      "loss": 0.609,
      "step": 53950
    },
    {
      "epoch": 4.6417448163124835,
      "grad_norm": 1.1328125,
      "learning_rate": 1.2624736009014314e-05,
      "loss": 1.5227,
      "step": 53951
    },
    {
      "epoch": 4.6418308526198055,
      "grad_norm": 1.4609375,
      "learning_rate": 1.2618698615233626e-05,
      "loss": 0.9351,
      "step": 53952
    },
    {
      "epoch": 4.641916888927128,
      "grad_norm": 1.0078125,
      "learning_rate": 1.2612662646948547e-05,
      "loss": 0.8378,
      "step": 53953
    },
    {
      "epoch": 4.642002925234449,
      "grad_norm": 0.80859375,
      "learning_rate": 1.260662810417651e-05,
      "loss": 0.6053,
      "step": 53954
    },
    {
      "epoch": 4.642088961541771,
      "grad_norm": 1.2265625,
      "learning_rate": 1.2600594986935388e-05,
      "loss": 0.936,
      "step": 53955
    },
    {
      "epoch": 4.642174997849092,
      "grad_norm": 1.0625,
      "learning_rate": 1.2594563295242666e-05,
      "loss": 1.6129,
      "step": 53956
    },
    {
      "epoch": 4.642261034156414,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2588533029116056e-05,
      "loss": 0.7495,
      "step": 53957
    },
    {
      "epoch": 4.642347070463735,
      "grad_norm": 1.390625,
      "learning_rate": 1.2582504188573206e-05,
      "loss": 0.9411,
      "step": 53958
    },
    {
      "epoch": 4.642433106771057,
      "grad_norm": 1.390625,
      "learning_rate": 1.2576476773631718e-05,
      "loss": 1.2414,
      "step": 53959
    },
    {
      "epoch": 4.642519143078379,
      "grad_norm": 0.9140625,
      "learning_rate": 1.2570450784309294e-05,
      "loss": 0.5658,
      "step": 53960
    },
    {
      "epoch": 4.642605179385701,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2564426220623427e-05,
      "loss": 0.4893,
      "step": 53961
    },
    {
      "epoch": 4.642691215693023,
      "grad_norm": 1.5546875,
      "learning_rate": 1.2558403082591873e-05,
      "loss": 1.436,
      "step": 53962
    },
    {
      "epoch": 4.642777252000344,
      "grad_norm": 0.8046875,
      "learning_rate": 1.255238137023218e-05,
      "loss": 0.8237,
      "step": 53963
    },
    {
      "epoch": 4.642863288307666,
      "grad_norm": 1.0078125,
      "learning_rate": 1.254636108356194e-05,
      "loss": 1.0204,
      "step": 53964
    },
    {
      "epoch": 4.642949324614987,
      "grad_norm": 948.0,
      "learning_rate": 1.2540342222598755e-05,
      "loss": 1.1273,
      "step": 53965
    },
    {
      "epoch": 4.643035360922309,
      "grad_norm": 0.91015625,
      "learning_rate": 1.2534324787360384e-05,
      "loss": 0.7816,
      "step": 53966
    },
    {
      "epoch": 4.643121397229631,
      "grad_norm": 1.203125,
      "learning_rate": 1.2528308777864206e-05,
      "loss": 1.1458,
      "step": 53967
    },
    {
      "epoch": 4.643207433536952,
      "grad_norm": 1.09375,
      "learning_rate": 1.2522294194127926e-05,
      "loss": 1.2456,
      "step": 53968
    },
    {
      "epoch": 4.6432934698442745,
      "grad_norm": 1.5390625,
      "learning_rate": 1.251628103616914e-05,
      "loss": 1.0874,
      "step": 53969
    },
    {
      "epoch": 4.643379506151596,
      "grad_norm": 9.8125,
      "learning_rate": 1.251026930400545e-05,
      "loss": 0.6463,
      "step": 53970
    },
    {
      "epoch": 4.643465542458918,
      "grad_norm": 1.5234375,
      "learning_rate": 1.2504258997654394e-05,
      "loss": 1.6553,
      "step": 53971
    },
    {
      "epoch": 4.643551578766239,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2498250117133569e-05,
      "loss": 0.616,
      "step": 53972
    },
    {
      "epoch": 4.643637615073561,
      "grad_norm": 0.875,
      "learning_rate": 1.2492242662460518e-05,
      "loss": 0.8643,
      "step": 53973
    },
    {
      "epoch": 4.643723651380883,
      "grad_norm": 1.2109375,
      "learning_rate": 1.2486236633652892e-05,
      "loss": 0.9578,
      "step": 53974
    },
    {
      "epoch": 4.643809687688204,
      "grad_norm": 1.703125,
      "learning_rate": 1.2480232030728179e-05,
      "loss": 1.3062,
      "step": 53975
    },
    {
      "epoch": 4.643895723995526,
      "grad_norm": 1.046875,
      "learning_rate": 1.2474228853703918e-05,
      "loss": 1.2847,
      "step": 53976
    },
    {
      "epoch": 4.6439817603028475,
      "grad_norm": 1.21875,
      "learning_rate": 1.2468227102597817e-05,
      "loss": 0.8119,
      "step": 53977
    },
    {
      "epoch": 4.6440677966101696,
      "grad_norm": 1.6484375,
      "learning_rate": 1.2462226777427254e-05,
      "loss": 0.6957,
      "step": 53978
    },
    {
      "epoch": 4.644153832917491,
      "grad_norm": 1.1953125,
      "learning_rate": 1.245622787820988e-05,
      "loss": 0.6779,
      "step": 53979
    },
    {
      "epoch": 4.644239869224813,
      "grad_norm": 1.515625,
      "learning_rate": 1.2450230404963237e-05,
      "loss": 1.1821,
      "step": 53980
    },
    {
      "epoch": 4.644325905532135,
      "grad_norm": 1.09375,
      "learning_rate": 1.2444234357704865e-05,
      "loss": 1.3161,
      "step": 53981
    },
    {
      "epoch": 4.644411941839456,
      "grad_norm": 1.046875,
      "learning_rate": 1.2438239736452251e-05,
      "loss": 1.5981,
      "step": 53982
    },
    {
      "epoch": 4.644497978146778,
      "grad_norm": 1.59375,
      "learning_rate": 1.2432246541222935e-05,
      "loss": 0.6352,
      "step": 53983
    },
    {
      "epoch": 4.644584014454099,
      "grad_norm": 1.515625,
      "learning_rate": 1.2426254772034518e-05,
      "loss": 1.4617,
      "step": 53984
    },
    {
      "epoch": 4.644670050761421,
      "grad_norm": 1.109375,
      "learning_rate": 1.2420264428904483e-05,
      "loss": 1.5381,
      "step": 53985
    },
    {
      "epoch": 4.6447560870687425,
      "grad_norm": 0.9296875,
      "learning_rate": 1.2414275511850315e-05,
      "loss": 1.3247,
      "step": 53986
    },
    {
      "epoch": 4.644842123376065,
      "grad_norm": 2.046875,
      "learning_rate": 1.2408288020889557e-05,
      "loss": 1.0262,
      "step": 53987
    },
    {
      "epoch": 4.644928159683387,
      "grad_norm": 620.0,
      "learning_rate": 1.2402301956039753e-05,
      "loss": 1.3481,
      "step": 53988
    },
    {
      "epoch": 4.645014195990708,
      "grad_norm": 1.046875,
      "learning_rate": 1.2396317317318329e-05,
      "loss": 0.5984,
      "step": 53989
    },
    {
      "epoch": 4.64510023229803,
      "grad_norm": 1.34375,
      "learning_rate": 1.2390334104742885e-05,
      "loss": 1.3599,
      "step": 53990
    },
    {
      "epoch": 4.645186268605351,
      "grad_norm": 1.6015625,
      "learning_rate": 1.2384352318330904e-05,
      "loss": 1.3751,
      "step": 53991
    },
    {
      "epoch": 4.645272304912673,
      "grad_norm": 1.3125,
      "learning_rate": 1.2378371958099821e-05,
      "loss": 1.007,
      "step": 53992
    },
    {
      "epoch": 4.645358341219994,
      "grad_norm": 0.8984375,
      "learning_rate": 1.237239302406712e-05,
      "loss": 0.5836,
      "step": 53993
    },
    {
      "epoch": 4.645444377527316,
      "grad_norm": 1.0859375,
      "learning_rate": 1.236641551625045e-05,
      "loss": 0.7727,
      "step": 53994
    },
    {
      "epoch": 4.6455304138346385,
      "grad_norm": 1.2265625,
      "learning_rate": 1.2360439434667025e-05,
      "loss": 1.5194,
      "step": 53995
    },
    {
      "epoch": 4.64561645014196,
      "grad_norm": 1.2265625,
      "learning_rate": 1.2354464779334606e-05,
      "loss": 1.3446,
      "step": 53996
    },
    {
      "epoch": 4.645702486449282,
      "grad_norm": 1.4765625,
      "learning_rate": 1.2348491550270457e-05,
      "loss": 1.3237,
      "step": 53997
    },
    {
      "epoch": 4.645788522756603,
      "grad_norm": 0.85546875,
      "learning_rate": 1.234251974749212e-05,
      "loss": 0.5494,
      "step": 53998
    },
    {
      "epoch": 4.645874559063925,
      "grad_norm": 1.03125,
      "learning_rate": 1.2336549371017136e-05,
      "loss": 0.9267,
      "step": 53999
    },
    {
      "epoch": 4.645960595371246,
      "grad_norm": 1.3359375,
      "learning_rate": 1.233058042086288e-05,
      "loss": 1.2448,
      "step": 54000
    },
    {
      "epoch": 4.646046631678568,
      "grad_norm": 1.171875,
      "learning_rate": 1.2324612897046838e-05,
      "loss": 0.8104,
      "step": 54001
    },
    {
      "epoch": 4.64613266798589,
      "grad_norm": 0.87890625,
      "learning_rate": 1.2318646799586387e-05,
      "loss": 0.6782,
      "step": 54002
    },
    {
      "epoch": 4.6462187042932115,
      "grad_norm": 1.5,
      "learning_rate": 1.231268212849912e-05,
      "loss": 1.8723,
      "step": 54003
    },
    {
      "epoch": 4.646304740600534,
      "grad_norm": 1.375,
      "learning_rate": 1.2306718883802415e-05,
      "loss": 0.751,
      "step": 54004
    },
    {
      "epoch": 4.646390776907855,
      "grad_norm": 1.28125,
      "learning_rate": 1.2300757065513646e-05,
      "loss": 1.6182,
      "step": 54005
    },
    {
      "epoch": 4.646476813215177,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2294796673650299e-05,
      "loss": 0.6465,
      "step": 54006
    },
    {
      "epoch": 4.646562849522499,
      "grad_norm": 1.1953125,
      "learning_rate": 1.2288837708229917e-05,
      "loss": 0.4745,
      "step": 54007
    },
    {
      "epoch": 4.64664888582982,
      "grad_norm": 0.95703125,
      "learning_rate": 1.2282880169269706e-05,
      "loss": 0.5042,
      "step": 54008
    },
    {
      "epoch": 4.646734922137142,
      "grad_norm": 1.1015625,
      "learning_rate": 1.2276924056787264e-05,
      "loss": 1.36,
      "step": 54009
    },
    {
      "epoch": 4.646820958444463,
      "grad_norm": 35.75,
      "learning_rate": 1.2270969370800023e-05,
      "loss": 1.2567,
      "step": 54010
    },
    {
      "epoch": 4.646906994751785,
      "grad_norm": 0.9609375,
      "learning_rate": 1.22650161113253e-05,
      "loss": 1.376,
      "step": 54011
    },
    {
      "epoch": 4.646993031059107,
      "grad_norm": 1.2890625,
      "learning_rate": 1.2259064278380526e-05,
      "loss": 0.7674,
      "step": 54012
    },
    {
      "epoch": 4.647079067366429,
      "grad_norm": 1.015625,
      "learning_rate": 1.2253113871983135e-05,
      "loss": 1.1187,
      "step": 54013
    },
    {
      "epoch": 4.647165103673751,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2247164892150498e-05,
      "loss": 0.7936,
      "step": 54014
    },
    {
      "epoch": 4.647251139981072,
      "grad_norm": 0.98828125,
      "learning_rate": 1.2241217338900102e-05,
      "loss": 1.1119,
      "step": 54015
    },
    {
      "epoch": 4.647337176288394,
      "grad_norm": 125.0,
      "learning_rate": 1.2235271212249265e-05,
      "loss": 1.09,
      "step": 54016
    },
    {
      "epoch": 4.647423212595715,
      "grad_norm": 1.328125,
      "learning_rate": 1.2229326512215366e-05,
      "loss": 1.5471,
      "step": 54017
    },
    {
      "epoch": 4.647509248903037,
      "grad_norm": 1.0,
      "learning_rate": 1.2223383238815833e-05,
      "loss": 1.0039,
      "step": 54018
    },
    {
      "epoch": 4.647595285210359,
      "grad_norm": 1.0625,
      "learning_rate": 1.221744139206804e-05,
      "loss": 1.4426,
      "step": 54019
    },
    {
      "epoch": 4.64768132151768,
      "grad_norm": 0.91796875,
      "learning_rate": 1.2211500971989365e-05,
      "loss": 0.4449,
      "step": 54020
    },
    {
      "epoch": 4.6477673578250025,
      "grad_norm": 1.2265625,
      "learning_rate": 1.220556197859718e-05,
      "loss": 0.7201,
      "step": 54021
    },
    {
      "epoch": 4.647853394132324,
      "grad_norm": 1.7890625,
      "learning_rate": 1.2199624411908861e-05,
      "loss": 1.2252,
      "step": 54022
    },
    {
      "epoch": 4.647939430439646,
      "grad_norm": 1.53125,
      "learning_rate": 1.2193688271941727e-05,
      "loss": 0.833,
      "step": 54023
    },
    {
      "epoch": 4.648025466746967,
      "grad_norm": 0.84375,
      "learning_rate": 1.2187753558713266e-05,
      "loss": 1.0742,
      "step": 54024
    },
    {
      "epoch": 4.648111503054289,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2181820272240684e-05,
      "loss": 1.0835,
      "step": 54025
    },
    {
      "epoch": 4.648197539361611,
      "grad_norm": 1.2109375,
      "learning_rate": 1.2175888412541414e-05,
      "loss": 1.4413,
      "step": 54026
    },
    {
      "epoch": 4.648283575668932,
      "grad_norm": 0.8828125,
      "learning_rate": 1.2169957979632829e-05,
      "loss": 0.7495,
      "step": 54027
    },
    {
      "epoch": 4.648369611976254,
      "grad_norm": 1.296875,
      "learning_rate": 1.2164028973532192e-05,
      "loss": 1.1125,
      "step": 54028
    },
    {
      "epoch": 4.6484556482835755,
      "grad_norm": 1.09375,
      "learning_rate": 1.2158101394256938e-05,
      "loss": 1.2645,
      "step": 54029
    },
    {
      "epoch": 4.648541684590898,
      "grad_norm": 1.2734375,
      "learning_rate": 1.215217524182427e-05,
      "loss": 0.6076,
      "step": 54030
    },
    {
      "epoch": 4.648627720898219,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2146250516251678e-05,
      "loss": 0.7717,
      "step": 54031
    },
    {
      "epoch": 4.648713757205541,
      "grad_norm": 1.1640625,
      "learning_rate": 1.2140327217556369e-05,
      "loss": 0.9136,
      "step": 54032
    },
    {
      "epoch": 4.648799793512863,
      "grad_norm": 1.1953125,
      "learning_rate": 1.2134405345755772e-05,
      "loss": 0.8706,
      "step": 54033
    },
    {
      "epoch": 4.648885829820184,
      "grad_norm": 1112.0,
      "learning_rate": 1.2128484900867153e-05,
      "loss": 1.1678,
      "step": 54034
    },
    {
      "epoch": 4.648971866127506,
      "grad_norm": 1.4453125,
      "learning_rate": 1.2122565882907832e-05,
      "loss": 0.8405,
      "step": 54035
    },
    {
      "epoch": 4.649057902434827,
      "grad_norm": 0.88671875,
      "learning_rate": 1.2116648291895072e-05,
      "loss": 0.9317,
      "step": 54036
    },
    {
      "epoch": 4.649143938742149,
      "grad_norm": 0.9375,
      "learning_rate": 1.2110732127846303e-05,
      "loss": 0.9335,
      "step": 54037
    },
    {
      "epoch": 4.6492299750494706,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2104817390778677e-05,
      "loss": 0.6104,
      "step": 54038
    },
    {
      "epoch": 4.649316011356793,
      "grad_norm": 1.015625,
      "learning_rate": 1.2098904080709571e-05,
      "loss": 1.2241,
      "step": 54039
    },
    {
      "epoch": 4.649402047664115,
      "grad_norm": 0.98828125,
      "learning_rate": 1.209299219765636e-05,
      "loss": 0.6886,
      "step": 54040
    },
    {
      "epoch": 4.649488083971436,
      "grad_norm": 1.25,
      "learning_rate": 1.2087081741636197e-05,
      "loss": 1.2567,
      "step": 54041
    },
    {
      "epoch": 4.649574120278758,
      "grad_norm": 1.3515625,
      "learning_rate": 1.2081172712666455e-05,
      "loss": 1.0991,
      "step": 54042
    },
    {
      "epoch": 4.649660156586079,
      "grad_norm": 0.98828125,
      "learning_rate": 1.2075265110764344e-05,
      "loss": 1.1042,
      "step": 54043
    },
    {
      "epoch": 4.649746192893401,
      "grad_norm": 12.625,
      "learning_rate": 1.2069358935947183e-05,
      "loss": 1.0474,
      "step": 54044
    },
    {
      "epoch": 4.649832229200722,
      "grad_norm": 0.984375,
      "learning_rate": 1.2063454188232347e-05,
      "loss": 0.7442,
      "step": 54045
    },
    {
      "epoch": 4.649918265508044,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2057550867636935e-05,
      "loss": 1.3793,
      "step": 54046
    },
    {
      "epoch": 4.6500043018153665,
      "grad_norm": 1.046875,
      "learning_rate": 1.2051648974178264e-05,
      "loss": 0.5119,
      "step": 54047
    },
    {
      "epoch": 4.650090338122688,
      "grad_norm": 1.890625,
      "learning_rate": 1.2045748507873656e-05,
      "loss": 0.6673,
      "step": 54048
    },
    {
      "epoch": 4.65017637443001,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2039849468740317e-05,
      "loss": 0.6009,
      "step": 54049
    },
    {
      "epoch": 4.650262410737331,
      "grad_norm": 6.0625,
      "learning_rate": 1.2033951856795511e-05,
      "loss": 0.7512,
      "step": 54050
    },
    {
      "epoch": 4.650348447044653,
      "grad_norm": 1.1796875,
      "learning_rate": 1.2028055672056504e-05,
      "loss": 1.5634,
      "step": 54051
    },
    {
      "epoch": 4.650434483351974,
      "grad_norm": 1.4609375,
      "learning_rate": 1.2022160914540559e-05,
      "loss": 1.9383,
      "step": 54052
    },
    {
      "epoch": 4.650520519659296,
      "grad_norm": 1.0546875,
      "learning_rate": 1.2016267584264884e-05,
      "loss": 1.3795,
      "step": 54053
    },
    {
      "epoch": 4.650606555966618,
      "grad_norm": 1.4140625,
      "learning_rate": 1.2010375681246632e-05,
      "loss": 1.0199,
      "step": 54054
    },
    {
      "epoch": 4.6506925922739395,
      "grad_norm": 1.15625,
      "learning_rate": 1.2004485205503179e-05,
      "loss": 1.2879,
      "step": 54055
    },
    {
      "epoch": 4.650778628581262,
      "grad_norm": 0.9765625,
      "learning_rate": 1.1998596157051733e-05,
      "loss": 1.0852,
      "step": 54056
    },
    {
      "epoch": 4.650864664888583,
      "grad_norm": 1.65625,
      "learning_rate": 1.1992708535909446e-05,
      "loss": 1.3851,
      "step": 54057
    },
    {
      "epoch": 4.650950701195905,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1986822342093528e-05,
      "loss": 0.3804,
      "step": 54058
    },
    {
      "epoch": 4.651036737503226,
      "grad_norm": 0.91015625,
      "learning_rate": 1.1980937575621353e-05,
      "loss": 1.116,
      "step": 54059
    },
    {
      "epoch": 4.651122773810548,
      "grad_norm": 1.3671875,
      "learning_rate": 1.1975054236509907e-05,
      "loss": 0.9717,
      "step": 54060
    },
    {
      "epoch": 4.65120881011787,
      "grad_norm": 1.109375,
      "learning_rate": 1.1969172324776568e-05,
      "loss": 1.1124,
      "step": 54061
    },
    {
      "epoch": 4.651294846425191,
      "grad_norm": 1.375,
      "learning_rate": 1.1963291840438484e-05,
      "loss": 1.2989,
      "step": 54062
    },
    {
      "epoch": 4.651380882732513,
      "grad_norm": 1.2578125,
      "learning_rate": 1.1957412783512867e-05,
      "loss": 1.3425,
      "step": 54063
    },
    {
      "epoch": 4.651466919039835,
      "grad_norm": 0.87109375,
      "learning_rate": 1.1951535154016868e-05,
      "loss": 0.5334,
      "step": 54064
    },
    {
      "epoch": 4.651552955347157,
      "grad_norm": 0.8671875,
      "learning_rate": 1.1945658951967753e-05,
      "loss": 0.9506,
      "step": 54065
    },
    {
      "epoch": 4.651638991654478,
      "grad_norm": 3.328125,
      "learning_rate": 1.1939784177382562e-05,
      "loss": 0.577,
      "step": 54066
    },
    {
      "epoch": 4.6517250279618,
      "grad_norm": 0.9296875,
      "learning_rate": 1.1933910830278727e-05,
      "loss": 0.8316,
      "step": 54067
    },
    {
      "epoch": 4.651811064269122,
      "grad_norm": 0.9453125,
      "learning_rate": 1.1928038910673178e-05,
      "loss": 0.9137,
      "step": 54068
    },
    {
      "epoch": 4.651897100576443,
      "grad_norm": 1.7890625,
      "learning_rate": 1.1922168418583123e-05,
      "loss": 1.0831,
      "step": 54069
    },
    {
      "epoch": 4.651983136883765,
      "grad_norm": 0.8828125,
      "learning_rate": 1.1916299354025938e-05,
      "loss": 0.9464,
      "step": 54070
    },
    {
      "epoch": 4.652069173191086,
      "grad_norm": 0.953125,
      "learning_rate": 1.1910431717018555e-05,
      "loss": 1.0197,
      "step": 54071
    },
    {
      "epoch": 4.652155209498408,
      "grad_norm": 1.1171875,
      "learning_rate": 1.1904565507578346e-05,
      "loss": 0.9525,
      "step": 54072
    },
    {
      "epoch": 4.65224124580573,
      "grad_norm": 0.94140625,
      "learning_rate": 1.1898700725722244e-05,
      "loss": 0.8364,
      "step": 54073
    },
    {
      "epoch": 4.652327282113052,
      "grad_norm": 1.0234375,
      "learning_rate": 1.1892837371467568e-05,
      "loss": 1.0095,
      "step": 54074
    },
    {
      "epoch": 4.652413318420374,
      "grad_norm": 1.1328125,
      "learning_rate": 1.1886975444831416e-05,
      "loss": 1.3458,
      "step": 54075
    },
    {
      "epoch": 4.652499354727695,
      "grad_norm": 1.4140625,
      "learning_rate": 1.1881114945830884e-05,
      "loss": 1.1699,
      "step": 54076
    },
    {
      "epoch": 4.652585391035017,
      "grad_norm": 1.03125,
      "learning_rate": 1.1875255874483183e-05,
      "loss": 0.4339,
      "step": 54077
    },
    {
      "epoch": 4.652671427342338,
      "grad_norm": 1.1640625,
      "learning_rate": 1.1869398230805405e-05,
      "loss": 1.0848,
      "step": 54078
    },
    {
      "epoch": 4.65275746364966,
      "grad_norm": 1.1484375,
      "learning_rate": 1.186354201481471e-05,
      "loss": 0.8646,
      "step": 54079
    },
    {
      "epoch": 4.652843499956981,
      "grad_norm": 0.8671875,
      "learning_rate": 1.1857687226528246e-05,
      "loss": 0.9898,
      "step": 54080
    },
    {
      "epoch": 4.6529295362643035,
      "grad_norm": 1.0390625,
      "learning_rate": 1.1851833865963113e-05,
      "loss": 1.6754,
      "step": 54081
    },
    {
      "epoch": 4.653015572571626,
      "grad_norm": 1.4140625,
      "learning_rate": 1.1845981933136351e-05,
      "loss": 0.7925,
      "step": 54082
    },
    {
      "epoch": 4.653101608878947,
      "grad_norm": 1.125,
      "learning_rate": 1.1840131428065225e-05,
      "loss": 0.9007,
      "step": 54083
    },
    {
      "epoch": 4.653187645186269,
      "grad_norm": 0.90234375,
      "learning_rate": 1.183428235076678e-05,
      "loss": 0.5485,
      "step": 54084
    },
    {
      "epoch": 4.65327368149359,
      "grad_norm": 10.5,
      "learning_rate": 1.182843470125805e-05,
      "loss": 1.0027,
      "step": 54085
    },
    {
      "epoch": 4.653359717800912,
      "grad_norm": 1.28125,
      "learning_rate": 1.1822588479556251e-05,
      "loss": 1.8554,
      "step": 54086
    },
    {
      "epoch": 4.653445754108233,
      "grad_norm": 1.3359375,
      "learning_rate": 1.1816743685678422e-05,
      "loss": 1.7001,
      "step": 54087
    },
    {
      "epoch": 4.653531790415555,
      "grad_norm": 56.0,
      "learning_rate": 1.181090031964166e-05,
      "loss": 0.8231,
      "step": 54088
    },
    {
      "epoch": 4.653617826722877,
      "grad_norm": 1.2421875,
      "learning_rate": 1.1805058381463119e-05,
      "loss": 1.4046,
      "step": 54089
    },
    {
      "epoch": 4.653703863030199,
      "grad_norm": 1.140625,
      "learning_rate": 1.1799217871159785e-05,
      "loss": 1.5904,
      "step": 54090
    },
    {
      "epoch": 4.653789899337521,
      "grad_norm": 1.203125,
      "learning_rate": 1.1793378788748754e-05,
      "loss": 1.6332,
      "step": 54091
    },
    {
      "epoch": 4.653875935644842,
      "grad_norm": 0.90625,
      "learning_rate": 1.1787541134247237e-05,
      "loss": 0.7312,
      "step": 54092
    },
    {
      "epoch": 4.653961971952164,
      "grad_norm": 2.859375,
      "learning_rate": 1.1781704907672109e-05,
      "loss": 0.6268,
      "step": 54093
    },
    {
      "epoch": 4.654048008259485,
      "grad_norm": 1.328125,
      "learning_rate": 1.1775870109040631e-05,
      "loss": 1.0484,
      "step": 54094
    },
    {
      "epoch": 4.654134044566807,
      "grad_norm": 1.984375,
      "learning_rate": 1.1770036738369738e-05,
      "loss": 1.4726,
      "step": 54095
    },
    {
      "epoch": 4.654220080874129,
      "grad_norm": 1.046875,
      "learning_rate": 1.1764204795676526e-05,
      "loss": 1.3034,
      "step": 54096
    },
    {
      "epoch": 4.65430611718145,
      "grad_norm": 1.1953125,
      "learning_rate": 1.175837428097809e-05,
      "loss": 0.7014,
      "step": 54097
    },
    {
      "epoch": 4.654392153488772,
      "grad_norm": 1.078125,
      "learning_rate": 1.1752545194291365e-05,
      "loss": 0.8805,
      "step": 54098
    },
    {
      "epoch": 4.654478189796094,
      "grad_norm": 1.546875,
      "learning_rate": 1.1746717535633555e-05,
      "loss": 1.6275,
      "step": 54099
    },
    {
      "epoch": 4.654564226103416,
      "grad_norm": 1.7265625,
      "learning_rate": 1.174089130502165e-05,
      "loss": 1.8733,
      "step": 54100
    },
    {
      "epoch": 4.654650262410737,
      "grad_norm": 0.828125,
      "learning_rate": 1.1735066502472635e-05,
      "loss": 0.8142,
      "step": 54101
    },
    {
      "epoch": 4.654736298718059,
      "grad_norm": 0.99609375,
      "learning_rate": 1.1729243128003608e-05,
      "loss": 0.8191,
      "step": 54102
    },
    {
      "epoch": 4.654822335025381,
      "grad_norm": 1.125,
      "learning_rate": 1.172342118163161e-05,
      "loss": 0.873,
      "step": 54103
    },
    {
      "epoch": 4.654908371332702,
      "grad_norm": 3.328125,
      "learning_rate": 1.1717600663373573e-05,
      "loss": 1.3026,
      "step": 54104
    },
    {
      "epoch": 4.654994407640024,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1711781573246594e-05,
      "loss": 0.3386,
      "step": 54105
    },
    {
      "epoch": 4.655080443947345,
      "grad_norm": 1.109375,
      "learning_rate": 1.1705963911267714e-05,
      "loss": 0.696,
      "step": 54106
    },
    {
      "epoch": 4.6551664802546675,
      "grad_norm": 322.0,
      "learning_rate": 1.170014767745392e-05,
      "loss": 0.6405,
      "step": 54107
    },
    {
      "epoch": 4.655252516561989,
      "grad_norm": 1.0546875,
      "learning_rate": 1.1694332871822256e-05,
      "loss": 1.0222,
      "step": 54108
    },
    {
      "epoch": 4.655338552869311,
      "grad_norm": 1.484375,
      "learning_rate": 1.1688519494389648e-05,
      "loss": 1.5636,
      "step": 54109
    },
    {
      "epoch": 4.655424589176633,
      "grad_norm": 0.96875,
      "learning_rate": 1.1682707545173198e-05,
      "loss": 0.4727,
      "step": 54110
    },
    {
      "epoch": 4.655510625483954,
      "grad_norm": 0.9296875,
      "learning_rate": 1.1676897024189836e-05,
      "loss": 1.1636,
      "step": 54111
    },
    {
      "epoch": 4.655596661791276,
      "grad_norm": 1.3984375,
      "learning_rate": 1.1671087931456547e-05,
      "loss": 0.893,
      "step": 54112
    },
    {
      "epoch": 4.655682698098597,
      "grad_norm": 1.1875,
      "learning_rate": 1.166528026699043e-05,
      "loss": 0.7887,
      "step": 54113
    },
    {
      "epoch": 4.655768734405919,
      "grad_norm": 1.3671875,
      "learning_rate": 1.1659474030808304e-05,
      "loss": 1.0947,
      "step": 54114
    },
    {
      "epoch": 4.6558547707132405,
      "grad_norm": 1.2109375,
      "learning_rate": 1.1653669222927266e-05,
      "loss": 1.6216,
      "step": 54115
    },
    {
      "epoch": 4.655940807020563,
      "grad_norm": 1.0234375,
      "learning_rate": 1.1647865843364302e-05,
      "loss": 0.9198,
      "step": 54116
    },
    {
      "epoch": 4.656026843327885,
      "grad_norm": 1.328125,
      "learning_rate": 1.1642063892136346e-05,
      "loss": 0.8329,
      "step": 54117
    },
    {
      "epoch": 4.656112879635206,
      "grad_norm": 1.1953125,
      "learning_rate": 1.1636263369260325e-05,
      "loss": 1.2218,
      "step": 54118
    },
    {
      "epoch": 4.656198915942528,
      "grad_norm": 0.99609375,
      "learning_rate": 1.1630464274753283e-05,
      "loss": 1.7336,
      "step": 54119
    },
    {
      "epoch": 4.656284952249849,
      "grad_norm": 0.8671875,
      "learning_rate": 1.162466660863215e-05,
      "loss": 1.3373,
      "step": 54120
    },
    {
      "epoch": 4.656370988557171,
      "grad_norm": 1.0703125,
      "learning_rate": 1.1618870370913858e-05,
      "loss": 1.0064,
      "step": 54121
    },
    {
      "epoch": 4.656457024864493,
      "grad_norm": 1.2265625,
      "learning_rate": 1.1613075561615449e-05,
      "loss": 0.9161,
      "step": 54122
    },
    {
      "epoch": 4.656543061171814,
      "grad_norm": 15.375,
      "learning_rate": 1.1607282180753742e-05,
      "loss": 1.4179,
      "step": 54123
    },
    {
      "epoch": 4.6566290974791364,
      "grad_norm": 1.359375,
      "learning_rate": 1.1601490228345834e-05,
      "loss": 1.1815,
      "step": 54124
    },
    {
      "epoch": 4.656715133786458,
      "grad_norm": 1.328125,
      "learning_rate": 1.1595699704408491e-05,
      "loss": 0.8122,
      "step": 54125
    },
    {
      "epoch": 4.65680117009378,
      "grad_norm": 1.21875,
      "learning_rate": 1.1589910608958753e-05,
      "loss": 0.8513,
      "step": 54126
    },
    {
      "epoch": 4.656887206401102,
      "grad_norm": 1.1015625,
      "learning_rate": 1.1584122942013554e-05,
      "loss": 1.2455,
      "step": 54127
    },
    {
      "epoch": 4.656973242708423,
      "grad_norm": 16.25,
      "learning_rate": 1.1578336703589765e-05,
      "loss": 1.3816,
      "step": 54128
    },
    {
      "epoch": 4.657059279015745,
      "grad_norm": 296.0,
      "learning_rate": 1.1572551893704375e-05,
      "loss": 1.2652,
      "step": 54129
    },
    {
      "epoch": 4.657145315323066,
      "grad_norm": 1.4921875,
      "learning_rate": 1.1566768512374314e-05,
      "loss": 1.5028,
      "step": 54130
    },
    {
      "epoch": 4.657231351630388,
      "grad_norm": 1.1640625,
      "learning_rate": 1.1560986559616404e-05,
      "loss": 1.5447,
      "step": 54131
    },
    {
      "epoch": 4.657317387937709,
      "grad_norm": 1.0625,
      "learning_rate": 1.155520603544763e-05,
      "loss": 1.2527,
      "step": 54132
    },
    {
      "epoch": 4.6574034242450315,
      "grad_norm": 0.78125,
      "learning_rate": 1.1549426939884922e-05,
      "loss": 0.9227,
      "step": 54133
    },
    {
      "epoch": 4.657489460552354,
      "grad_norm": 4.96875,
      "learning_rate": 1.15436492729451e-05,
      "loss": 0.4341,
      "step": 54134
    },
    {
      "epoch": 4.657575496859675,
      "grad_norm": 0.85546875,
      "learning_rate": 1.1537873034645097e-05,
      "loss": 0.5352,
      "step": 54135
    },
    {
      "epoch": 4.657661533166997,
      "grad_norm": 3.765625,
      "learning_rate": 1.1532098225001841e-05,
      "loss": 0.6812,
      "step": 54136
    },
    {
      "epoch": 4.657747569474318,
      "grad_norm": 1.2578125,
      "learning_rate": 1.1526324844032154e-05,
      "loss": 1.5226,
      "step": 54137
    },
    {
      "epoch": 4.65783360578164,
      "grad_norm": 1.171875,
      "learning_rate": 1.1520552891753022e-05,
      "loss": 1.3805,
      "step": 54138
    },
    {
      "epoch": 4.657919642088961,
      "grad_norm": 0.8125,
      "learning_rate": 1.1514782368181264e-05,
      "loss": 0.8462,
      "step": 54139
    },
    {
      "epoch": 4.658005678396283,
      "grad_norm": 1.046875,
      "learning_rate": 1.1509013273333702e-05,
      "loss": 0.6368,
      "step": 54140
    },
    {
      "epoch": 4.658091714703605,
      "grad_norm": 5.34375,
      "learning_rate": 1.150324560722732e-05,
      "loss": 0.8911,
      "step": 54141
    },
    {
      "epoch": 4.658177751010927,
      "grad_norm": 1.390625,
      "learning_rate": 1.1497479369878938e-05,
      "loss": 1.217,
      "step": 54142
    },
    {
      "epoch": 4.658263787318249,
      "grad_norm": 1.40625,
      "learning_rate": 1.1491714561305377e-05,
      "loss": 1.1992,
      "step": 54143
    },
    {
      "epoch": 4.65834982362557,
      "grad_norm": 0.953125,
      "learning_rate": 1.1485951181523568e-05,
      "loss": 0.968,
      "step": 54144
    },
    {
      "epoch": 4.658435859932892,
      "grad_norm": 1.109375,
      "learning_rate": 1.148018923055033e-05,
      "loss": 1.0425,
      "step": 54145
    },
    {
      "epoch": 4.658521896240213,
      "grad_norm": 0.890625,
      "learning_rate": 1.147442870840254e-05,
      "loss": 0.885,
      "step": 54146
    },
    {
      "epoch": 4.658607932547535,
      "grad_norm": 1.3359375,
      "learning_rate": 1.1468669615097016e-05,
      "loss": 1.0495,
      "step": 54147
    },
    {
      "epoch": 4.658693968854857,
      "grad_norm": 1.1875,
      "learning_rate": 1.1462911950650633e-05,
      "loss": 1.465,
      "step": 54148
    },
    {
      "epoch": 4.658780005162178,
      "grad_norm": 1.171875,
      "learning_rate": 1.145715571508027e-05,
      "loss": 2.003,
      "step": 54149
    },
    {
      "epoch": 4.6588660414695005,
      "grad_norm": 1.140625,
      "learning_rate": 1.145140090840263e-05,
      "loss": 0.5781,
      "step": 54150
    },
    {
      "epoch": 4.658952077776822,
      "grad_norm": 0.9140625,
      "learning_rate": 1.144564753063465e-05,
      "loss": 0.6741,
      "step": 54151
    },
    {
      "epoch": 4.659038114084144,
      "grad_norm": 0.89453125,
      "learning_rate": 1.1439895581793147e-05,
      "loss": 0.9598,
      "step": 54152
    },
    {
      "epoch": 4.659124150391465,
      "grad_norm": 1.359375,
      "learning_rate": 1.1434145061894941e-05,
      "loss": 1.3959,
      "step": 54153
    },
    {
      "epoch": 4.659210186698787,
      "grad_norm": 0.83203125,
      "learning_rate": 1.1428395970956851e-05,
      "loss": 0.4528,
      "step": 54154
    },
    {
      "epoch": 4.659296223006109,
      "grad_norm": 1.3984375,
      "learning_rate": 1.1422648308995642e-05,
      "loss": 0.8787,
      "step": 54155
    },
    {
      "epoch": 4.65938225931343,
      "grad_norm": 1.1875,
      "learning_rate": 1.1416902076028135e-05,
      "loss": 1.1333,
      "step": 54156
    },
    {
      "epoch": 4.659468295620752,
      "grad_norm": 1.2109375,
      "learning_rate": 1.141115727207126e-05,
      "loss": 1.5264,
      "step": 54157
    },
    {
      "epoch": 4.6595543319280734,
      "grad_norm": 2.734375,
      "learning_rate": 1.1405413897141614e-05,
      "loss": 1.2291,
      "step": 54158
    },
    {
      "epoch": 4.6596403682353955,
      "grad_norm": 0.91015625,
      "learning_rate": 1.1399671951256129e-05,
      "loss": 0.8634,
      "step": 54159
    },
    {
      "epoch": 4.659726404542717,
      "grad_norm": 1.3359375,
      "learning_rate": 1.1393931434431682e-05,
      "loss": 1.4516,
      "step": 54160
    },
    {
      "epoch": 4.659812440850039,
      "grad_norm": 2.984375,
      "learning_rate": 1.1388192346684866e-05,
      "loss": 1.2421,
      "step": 54161
    },
    {
      "epoch": 4.659898477157361,
      "grad_norm": 1.2265625,
      "learning_rate": 1.1382454688032506e-05,
      "loss": 0.9869,
      "step": 54162
    },
    {
      "epoch": 4.659984513464682,
      "grad_norm": 1.171875,
      "learning_rate": 1.1376718458491585e-05,
      "loss": 1.1242,
      "step": 54163
    },
    {
      "epoch": 4.660070549772004,
      "grad_norm": 1.1640625,
      "learning_rate": 1.1370983658078593e-05,
      "loss": 0.9069,
      "step": 54164
    },
    {
      "epoch": 4.660156586079325,
      "grad_norm": 0.859375,
      "learning_rate": 1.1365250286810514e-05,
      "loss": 0.7929,
      "step": 54165
    },
    {
      "epoch": 4.660242622386647,
      "grad_norm": 0.9921875,
      "learning_rate": 1.1359518344704055e-05,
      "loss": 0.6752,
      "step": 54166
    },
    {
      "epoch": 4.6603286586939685,
      "grad_norm": 1.3125,
      "learning_rate": 1.1353787831775875e-05,
      "loss": 1.693,
      "step": 54167
    },
    {
      "epoch": 4.660414695001291,
      "grad_norm": 1.1015625,
      "learning_rate": 1.1348058748042955e-05,
      "loss": 0.9899,
      "step": 54168
    },
    {
      "epoch": 4.660500731308613,
      "grad_norm": 1.4140625,
      "learning_rate": 1.134233109352184e-05,
      "loss": 1.2053,
      "step": 54169
    },
    {
      "epoch": 4.660586767615934,
      "grad_norm": 1.0859375,
      "learning_rate": 1.1336604868229405e-05,
      "loss": 1.0309,
      "step": 54170
    },
    {
      "epoch": 4.660672803923256,
      "grad_norm": 55.75,
      "learning_rate": 1.133088007218236e-05,
      "loss": 1.0245,
      "step": 54171
    },
    {
      "epoch": 4.660758840230577,
      "grad_norm": 1.734375,
      "learning_rate": 1.132515670539741e-05,
      "loss": 1.2978,
      "step": 54172
    },
    {
      "epoch": 4.660844876537899,
      "grad_norm": 0.96484375,
      "learning_rate": 1.131943476789138e-05,
      "loss": 1.0929,
      "step": 54173
    },
    {
      "epoch": 4.66093091284522,
      "grad_norm": 0.8828125,
      "learning_rate": 1.1313714259680918e-05,
      "loss": 1.1524,
      "step": 54174
    },
    {
      "epoch": 4.661016949152542,
      "grad_norm": 1.2890625,
      "learning_rate": 1.1307995180782849e-05,
      "loss": 1.7013,
      "step": 54175
    },
    {
      "epoch": 4.6611029854598645,
      "grad_norm": 1.1796875,
      "learning_rate": 1.1302277531213768e-05,
      "loss": 1.2362,
      "step": 54176
    },
    {
      "epoch": 4.661189021767186,
      "grad_norm": 1.25,
      "learning_rate": 1.1296561310990606e-05,
      "loss": 1.4046,
      "step": 54177
    },
    {
      "epoch": 4.661275058074508,
      "grad_norm": 21.375,
      "learning_rate": 1.1290846520129849e-05,
      "loss": 1.3937,
      "step": 54178
    },
    {
      "epoch": 4.661361094381829,
      "grad_norm": 1.1484375,
      "learning_rate": 1.1285133158648375e-05,
      "loss": 1.102,
      "step": 54179
    },
    {
      "epoch": 4.661447130689151,
      "grad_norm": 1.359375,
      "learning_rate": 1.1279421226562724e-05,
      "loss": 1.8661,
      "step": 54180
    },
    {
      "epoch": 4.661533166996472,
      "grad_norm": 1.2421875,
      "learning_rate": 1.1273710723889885e-05,
      "loss": 1.0932,
      "step": 54181
    },
    {
      "epoch": 4.661619203303794,
      "grad_norm": 1.0625,
      "learning_rate": 1.1268001650646287e-05,
      "loss": 0.9493,
      "step": 54182
    },
    {
      "epoch": 4.661705239611116,
      "grad_norm": 1.1796875,
      "learning_rate": 1.1262294006848694e-05,
      "loss": 0.9388,
      "step": 54183
    },
    {
      "epoch": 4.6617912759184374,
      "grad_norm": 1.1796875,
      "learning_rate": 1.1256587792513928e-05,
      "loss": 1.4386,
      "step": 54184
    },
    {
      "epoch": 4.6618773122257595,
      "grad_norm": 1.4375,
      "learning_rate": 1.125088300765853e-05,
      "loss": 1.0505,
      "step": 54185
    },
    {
      "epoch": 4.661963348533081,
      "grad_norm": 25.375,
      "learning_rate": 1.1245179652299264e-05,
      "loss": 1.6227,
      "step": 54186
    },
    {
      "epoch": 4.662049384840403,
      "grad_norm": 40.25,
      "learning_rate": 1.123947772645284e-05,
      "loss": 1.371,
      "step": 54187
    },
    {
      "epoch": 4.662135421147724,
      "grad_norm": 1.2109375,
      "learning_rate": 1.12337772301358e-05,
      "loss": 0.7149,
      "step": 54188
    },
    {
      "epoch": 4.662221457455046,
      "grad_norm": 1.1796875,
      "learning_rate": 1.1228078163364963e-05,
      "loss": 1.4641,
      "step": 54189
    },
    {
      "epoch": 4.662307493762368,
      "grad_norm": 0.8671875,
      "learning_rate": 1.1222380526156927e-05,
      "loss": 0.9732,
      "step": 54190
    },
    {
      "epoch": 4.662393530069689,
      "grad_norm": 1.7578125,
      "learning_rate": 1.1216684318528348e-05,
      "loss": 1.2569,
      "step": 54191
    },
    {
      "epoch": 4.662479566377011,
      "grad_norm": 0.8203125,
      "learning_rate": 1.1210989540495876e-05,
      "loss": 0.6875,
      "step": 54192
    },
    {
      "epoch": 4.6625656026843325,
      "grad_norm": 1.1328125,
      "learning_rate": 1.1205296192076276e-05,
      "loss": 0.5448,
      "step": 54193
    },
    {
      "epoch": 4.662651638991655,
      "grad_norm": 1.265625,
      "learning_rate": 1.1199604273286035e-05,
      "loss": 1.6027,
      "step": 54194
    },
    {
      "epoch": 4.662737675298976,
      "grad_norm": 1.2109375,
      "learning_rate": 1.1193913784141974e-05,
      "loss": 1.3488,
      "step": 54195
    },
    {
      "epoch": 4.662823711606298,
      "grad_norm": 1.8671875,
      "learning_rate": 1.118822472466058e-05,
      "loss": 1.2513,
      "step": 54196
    },
    {
      "epoch": 4.66290974791362,
      "grad_norm": 8.25,
      "learning_rate": 1.1182537094858559e-05,
      "loss": 0.9807,
      "step": 54197
    },
    {
      "epoch": 4.662995784220941,
      "grad_norm": 1.21875,
      "learning_rate": 1.1176850894752566e-05,
      "loss": 1.298,
      "step": 54198
    },
    {
      "epoch": 4.663081820528263,
      "grad_norm": 0.97265625,
      "learning_rate": 1.11711661243592e-05,
      "loss": 1.1235,
      "step": 54199
    },
    {
      "epoch": 4.663167856835584,
      "grad_norm": 1.0546875,
      "learning_rate": 1.1165482783695114e-05,
      "loss": 1.8127,
      "step": 54200
    },
    {
      "epoch": 4.663253893142906,
      "grad_norm": 1.28125,
      "learning_rate": 1.1159800872776904e-05,
      "loss": 1.5138,
      "step": 54201
    },
    {
      "epoch": 4.663339929450228,
      "grad_norm": 1.2109375,
      "learning_rate": 1.1154120391621225e-05,
      "loss": 1.223,
      "step": 54202
    },
    {
      "epoch": 4.66342596575755,
      "grad_norm": 1.4140625,
      "learning_rate": 1.114844134024462e-05,
      "loss": 1.6578,
      "step": 54203
    },
    {
      "epoch": 4.663512002064872,
      "grad_norm": 1.140625,
      "learning_rate": 1.114276371866385e-05,
      "loss": 1.1506,
      "step": 54204
    },
    {
      "epoch": 4.663598038372193,
      "grad_norm": 1.3046875,
      "learning_rate": 1.1137087526895295e-05,
      "loss": 1.3498,
      "step": 54205
    },
    {
      "epoch": 4.663684074679515,
      "grad_norm": 0.78125,
      "learning_rate": 1.1131412764955772e-05,
      "loss": 0.9823,
      "step": 54206
    },
    {
      "epoch": 4.663770110986836,
      "grad_norm": 1.59375,
      "learning_rate": 1.112573943286177e-05,
      "loss": 1.8657,
      "step": 54207
    },
    {
      "epoch": 4.663856147294158,
      "grad_norm": 0.8359375,
      "learning_rate": 1.112006753062983e-05,
      "loss": 0.7921,
      "step": 54208
    },
    {
      "epoch": 4.663942183601479,
      "grad_norm": 1.109375,
      "learning_rate": 1.1114397058276715e-05,
      "loss": 1.3016,
      "step": 54209
    },
    {
      "epoch": 4.6640282199088015,
      "grad_norm": 1.140625,
      "learning_rate": 1.1108728015818858e-05,
      "loss": 0.6213,
      "step": 54210
    },
    {
      "epoch": 4.6641142562161235,
      "grad_norm": 1.171875,
      "learning_rate": 1.1103060403272858e-05,
      "loss": 1.4951,
      "step": 54211
    },
    {
      "epoch": 4.664200292523445,
      "grad_norm": 1.3984375,
      "learning_rate": 1.1097394220655366e-05,
      "loss": 1.8363,
      "step": 54212
    },
    {
      "epoch": 4.664286328830767,
      "grad_norm": 0.85546875,
      "learning_rate": 1.109172946798287e-05,
      "loss": 0.8659,
      "step": 54213
    },
    {
      "epoch": 4.664372365138088,
      "grad_norm": 6.90625,
      "learning_rate": 1.1086066145271967e-05,
      "loss": 0.6443,
      "step": 54214
    },
    {
      "epoch": 4.66445840144541,
      "grad_norm": 12.9375,
      "learning_rate": 1.1080404252539311e-05,
      "loss": 1.5488,
      "step": 54215
    },
    {
      "epoch": 4.664544437752731,
      "grad_norm": 1.1953125,
      "learning_rate": 1.1074743789801333e-05,
      "loss": 1.6391,
      "step": 54216
    },
    {
      "epoch": 4.664630474060053,
      "grad_norm": 1.1953125,
      "learning_rate": 1.1069084757074633e-05,
      "loss": 1.0674,
      "step": 54217
    },
    {
      "epoch": 4.664716510367375,
      "grad_norm": 1.1328125,
      "learning_rate": 1.106342715437575e-05,
      "loss": 1.0775,
      "step": 54218
    },
    {
      "epoch": 4.6648025466746965,
      "grad_norm": 2.015625,
      "learning_rate": 1.1057770981721227e-05,
      "loss": 1.1455,
      "step": 54219
    },
    {
      "epoch": 4.664888582982019,
      "grad_norm": 1.3046875,
      "learning_rate": 1.1052116239127719e-05,
      "loss": 1.7099,
      "step": 54220
    },
    {
      "epoch": 4.66497461928934,
      "grad_norm": 1.3515625,
      "learning_rate": 1.1046462926611656e-05,
      "loss": 1.2908,
      "step": 54221
    },
    {
      "epoch": 4.665060655596662,
      "grad_norm": 1.328125,
      "learning_rate": 1.1040811044189525e-05,
      "loss": 1.5525,
      "step": 54222
    },
    {
      "epoch": 4.665146691903983,
      "grad_norm": 0.99609375,
      "learning_rate": 1.1035160591878036e-05,
      "loss": 0.6699,
      "step": 54223
    },
    {
      "epoch": 4.665232728211305,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1029511569693506e-05,
      "loss": 0.5796,
      "step": 54224
    },
    {
      "epoch": 4.665318764518627,
      "grad_norm": 1.1875,
      "learning_rate": 1.1023863977652594e-05,
      "loss": 0.9269,
      "step": 54225
    },
    {
      "epoch": 4.665404800825948,
      "grad_norm": 1.03125,
      "learning_rate": 1.101821781577178e-05,
      "loss": 0.8582,
      "step": 54226
    },
    {
      "epoch": 4.66549083713327,
      "grad_norm": 1.4765625,
      "learning_rate": 1.1012573084067557e-05,
      "loss": 0.8072,
      "step": 54227
    },
    {
      "epoch": 4.665576873440592,
      "grad_norm": 1.3125,
      "learning_rate": 1.1006929782556519e-05,
      "loss": 0.5777,
      "step": 54228
    },
    {
      "epoch": 4.665662909747914,
      "grad_norm": 1.03125,
      "learning_rate": 1.1001287911255042e-05,
      "loss": 0.7143,
      "step": 54229
    },
    {
      "epoch": 4.665748946055235,
      "grad_norm": 1.171875,
      "learning_rate": 1.099564747017967e-05,
      "loss": 0.7888,
      "step": 54230
    },
    {
      "epoch": 4.665834982362557,
      "grad_norm": 1.078125,
      "learning_rate": 1.0990008459347e-05,
      "loss": 0.7632,
      "step": 54231
    },
    {
      "epoch": 4.665921018669879,
      "grad_norm": 1.359375,
      "learning_rate": 1.0984370878773409e-05,
      "loss": 1.1731,
      "step": 54232
    },
    {
      "epoch": 4.6660070549772,
      "grad_norm": 1.0546875,
      "learning_rate": 1.097873472847538e-05,
      "loss": 1.3652,
      "step": 54233
    },
    {
      "epoch": 4.666093091284522,
      "grad_norm": 0.98046875,
      "learning_rate": 1.0973100008469516e-05,
      "loss": 0.7557,
      "step": 54234
    },
    {
      "epoch": 4.666179127591843,
      "grad_norm": 2.15625,
      "learning_rate": 1.096746671877219e-05,
      "loss": 1.3975,
      "step": 54235
    },
    {
      "epoch": 4.6662651638991655,
      "grad_norm": 1.203125,
      "learning_rate": 1.0961834859399944e-05,
      "loss": 1.0725,
      "step": 54236
    },
    {
      "epoch": 4.6663512002064875,
      "grad_norm": 1.453125,
      "learning_rate": 1.095620443036921e-05,
      "loss": 1.0589,
      "step": 54237
    },
    {
      "epoch": 4.666437236513809,
      "grad_norm": 27.25,
      "learning_rate": 1.095057543169642e-05,
      "loss": 0.8921,
      "step": 54238
    },
    {
      "epoch": 4.666523272821131,
      "grad_norm": 0.87890625,
      "learning_rate": 1.0944947863398114e-05,
      "loss": 0.9659,
      "step": 54239
    },
    {
      "epoch": 4.666609309128452,
      "grad_norm": 2.328125,
      "learning_rate": 1.0939321725490726e-05,
      "loss": 0.8512,
      "step": 54240
    },
    {
      "epoch": 4.666695345435774,
      "grad_norm": 1.2890625,
      "learning_rate": 1.0933697017990685e-05,
      "loss": 1.6266,
      "step": 54241
    },
    {
      "epoch": 4.666781381743095,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0928073740914478e-05,
      "loss": 1.3587,
      "step": 54242
    },
    {
      "epoch": 4.666867418050417,
      "grad_norm": 1.3671875,
      "learning_rate": 1.092245189427854e-05,
      "loss": 1.3732,
      "step": 54243
    },
    {
      "epoch": 4.666953454357739,
      "grad_norm": 1.4609375,
      "learning_rate": 1.0916831478099243e-05,
      "loss": 1.6139,
      "step": 54244
    },
    {
      "epoch": 4.6670394906650605,
      "grad_norm": 1.09375,
      "learning_rate": 1.0911212492393186e-05,
      "loss": 1.0534,
      "step": 54245
    },
    {
      "epoch": 4.667125526972383,
      "grad_norm": 2.375,
      "learning_rate": 1.0905594937176689e-05,
      "loss": 1.116,
      "step": 54246
    },
    {
      "epoch": 4.667211563279704,
      "grad_norm": 1.359375,
      "learning_rate": 1.0899978812466182e-05,
      "loss": 0.8203,
      "step": 54247
    },
    {
      "epoch": 4.667297599587026,
      "grad_norm": 10.0625,
      "learning_rate": 1.08943641182781e-05,
      "loss": 1.4759,
      "step": 54248
    },
    {
      "epoch": 4.667383635894348,
      "grad_norm": 1.3359375,
      "learning_rate": 1.088875085462887e-05,
      "loss": 1.104,
      "step": 54249
    },
    {
      "epoch": 4.667469672201669,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0883139021534982e-05,
      "loss": 0.7544,
      "step": 54250
    },
    {
      "epoch": 4.667555708508991,
      "grad_norm": 1.109375,
      "learning_rate": 1.0877528619012756e-05,
      "loss": 1.1894,
      "step": 54251
    },
    {
      "epoch": 4.667641744816312,
      "grad_norm": 1.4140625,
      "learning_rate": 1.0871919647078566e-05,
      "loss": 1.5591,
      "step": 54252
    },
    {
      "epoch": 4.667727781123634,
      "grad_norm": 0.9296875,
      "learning_rate": 1.0866312105748954e-05,
      "loss": 1.2495,
      "step": 54253
    },
    {
      "epoch": 4.667813817430956,
      "grad_norm": 1.28125,
      "learning_rate": 1.0860705995040187e-05,
      "loss": 1.7196,
      "step": 54254
    },
    {
      "epoch": 4.667899853738278,
      "grad_norm": 1.1640625,
      "learning_rate": 1.0855101314968807e-05,
      "loss": 0.9609,
      "step": 54255
    },
    {
      "epoch": 4.6679858900456,
      "grad_norm": 0.9921875,
      "learning_rate": 1.0849498065551078e-05,
      "loss": 0.9438,
      "step": 54256
    },
    {
      "epoch": 4.668071926352921,
      "grad_norm": 1.1484375,
      "learning_rate": 1.0843896246803375e-05,
      "loss": 0.9966,
      "step": 54257
    },
    {
      "epoch": 4.668157962660243,
      "grad_norm": 0.93359375,
      "learning_rate": 1.0838295858742241e-05,
      "loss": 1.1346,
      "step": 54258
    },
    {
      "epoch": 4.668243998967564,
      "grad_norm": 1.2890625,
      "learning_rate": 1.0832696901383886e-05,
      "loss": 1.4322,
      "step": 54259
    },
    {
      "epoch": 4.668330035274886,
      "grad_norm": 1.1875,
      "learning_rate": 1.0827099374744798e-05,
      "loss": 0.9422,
      "step": 54260
    },
    {
      "epoch": 4.668416071582207,
      "grad_norm": 0.90234375,
      "learning_rate": 1.0821503278841294e-05,
      "loss": 0.4469,
      "step": 54261
    },
    {
      "epoch": 4.6685021078895295,
      "grad_norm": 1.2265625,
      "learning_rate": 1.081590861368975e-05,
      "loss": 1.3615,
      "step": 54262
    },
    {
      "epoch": 4.6685881441968515,
      "grad_norm": 67.0,
      "learning_rate": 1.081031537930649e-05,
      "loss": 1.2195,
      "step": 54263
    },
    {
      "epoch": 4.668674180504173,
      "grad_norm": 1.2109375,
      "learning_rate": 1.0804723575707997e-05,
      "loss": 1.4375,
      "step": 54264
    },
    {
      "epoch": 4.668760216811495,
      "grad_norm": 1.0078125,
      "learning_rate": 1.0799133202910483e-05,
      "loss": 0.6612,
      "step": 54265
    },
    {
      "epoch": 4.668846253118816,
      "grad_norm": 1.4609375,
      "learning_rate": 1.0793544260930378e-05,
      "loss": 0.8891,
      "step": 54266
    },
    {
      "epoch": 4.668932289426138,
      "grad_norm": 0.97265625,
      "learning_rate": 1.0787956749784e-05,
      "loss": 0.8373,
      "step": 54267
    },
    {
      "epoch": 4.669018325733459,
      "grad_norm": 1.0,
      "learning_rate": 1.0782370669487728e-05,
      "loss": 1.5259,
      "step": 54268
    },
    {
      "epoch": 4.669104362040781,
      "grad_norm": 1.4453125,
      "learning_rate": 1.0776786020057882e-05,
      "loss": 1.0245,
      "step": 54269
    },
    {
      "epoch": 4.669190398348103,
      "grad_norm": 1.1484375,
      "learning_rate": 1.0771202801510727e-05,
      "loss": 1.2091,
      "step": 54270
    },
    {
      "epoch": 4.6692764346554245,
      "grad_norm": 1.828125,
      "learning_rate": 1.0765621013862692e-05,
      "loss": 1.3644,
      "step": 54271
    },
    {
      "epoch": 4.669362470962747,
      "grad_norm": 1.21875,
      "learning_rate": 1.0760040657130098e-05,
      "loss": 1.4897,
      "step": 54272
    },
    {
      "epoch": 4.669448507270068,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0754461731329156e-05,
      "loss": 0.6928,
      "step": 54273
    },
    {
      "epoch": 4.66953454357739,
      "grad_norm": 0.890625,
      "learning_rate": 1.0748884236476297e-05,
      "loss": 0.5973,
      "step": 54274
    },
    {
      "epoch": 4.669620579884711,
      "grad_norm": 0.9375,
      "learning_rate": 1.074330817258784e-05,
      "loss": 1.0842,
      "step": 54275
    },
    {
      "epoch": 4.669706616192033,
      "grad_norm": 3.109375,
      "learning_rate": 1.0737733539679994e-05,
      "loss": 1.2711,
      "step": 54276
    },
    {
      "epoch": 4.669792652499355,
      "grad_norm": 31.625,
      "learning_rate": 1.0732160337769193e-05,
      "loss": 1.2777,
      "step": 54277
    },
    {
      "epoch": 4.669878688806676,
      "grad_norm": 1.1484375,
      "learning_rate": 1.0726588566871586e-05,
      "loss": 0.8258,
      "step": 54278
    },
    {
      "epoch": 4.669964725113998,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0721018227003555e-05,
      "loss": 0.617,
      "step": 54279
    },
    {
      "epoch": 4.67005076142132,
      "grad_norm": 1.125,
      "learning_rate": 1.0715449318181469e-05,
      "loss": 1.001,
      "step": 54280
    },
    {
      "epoch": 4.670136797728642,
      "grad_norm": 1.0078125,
      "learning_rate": 1.0709881840421432e-05,
      "loss": 0.879,
      "step": 54281
    },
    {
      "epoch": 4.670222834035963,
      "grad_norm": 1.09375,
      "learning_rate": 1.0704315793739872e-05,
      "loss": 0.9839,
      "step": 54282
    },
    {
      "epoch": 4.670308870343285,
      "grad_norm": 1.25,
      "learning_rate": 1.0698751178153054e-05,
      "loss": 0.9723,
      "step": 54283
    },
    {
      "epoch": 4.670394906650607,
      "grad_norm": 1.1875,
      "learning_rate": 1.0693187993677189e-05,
      "loss": 0.7316,
      "step": 54284
    },
    {
      "epoch": 4.670480942957928,
      "grad_norm": 1.171875,
      "learning_rate": 1.0687626240328541e-05,
      "loss": 1.493,
      "step": 54285
    },
    {
      "epoch": 4.67056697926525,
      "grad_norm": 1.609375,
      "learning_rate": 1.068206591812354e-05,
      "loss": 1.1953,
      "step": 54286
    },
    {
      "epoch": 4.670653015572571,
      "grad_norm": 1.4296875,
      "learning_rate": 1.067650702707823e-05,
      "loss": 1.3534,
      "step": 54287
    },
    {
      "epoch": 4.6707390518798935,
      "grad_norm": 4.125,
      "learning_rate": 1.0670949567208988e-05,
      "loss": 1.0197,
      "step": 54288
    },
    {
      "epoch": 4.670825088187215,
      "grad_norm": 0.9765625,
      "learning_rate": 1.0665393538532076e-05,
      "loss": 0.8286,
      "step": 54289
    },
    {
      "epoch": 4.670911124494537,
      "grad_norm": 5.59375,
      "learning_rate": 1.0659838941063704e-05,
      "loss": 1.494,
      "step": 54290
    },
    {
      "epoch": 4.670997160801859,
      "grad_norm": 0.859375,
      "learning_rate": 1.0654285774820083e-05,
      "loss": 0.6055,
      "step": 54291
    },
    {
      "epoch": 4.67108319710918,
      "grad_norm": 1.1640625,
      "learning_rate": 1.0648734039817531e-05,
      "loss": 1.5064,
      "step": 54292
    },
    {
      "epoch": 4.671169233416502,
      "grad_norm": 16.875,
      "learning_rate": 1.0643183736072316e-05,
      "loss": 0.9718,
      "step": 54293
    },
    {
      "epoch": 4.671255269723823,
      "grad_norm": 1.3125,
      "learning_rate": 1.0637634863600587e-05,
      "loss": 1.0013,
      "step": 54294
    },
    {
      "epoch": 4.671341306031145,
      "grad_norm": 6.875,
      "learning_rate": 1.0632087422418557e-05,
      "loss": 1.0566,
      "step": 54295
    },
    {
      "epoch": 4.6714273423384665,
      "grad_norm": 1.1015625,
      "learning_rate": 1.0626541412542545e-05,
      "loss": 1.2244,
      "step": 54296
    },
    {
      "epoch": 4.6715133786457885,
      "grad_norm": 1.2578125,
      "learning_rate": 1.0620996833988705e-05,
      "loss": 0.852,
      "step": 54297
    },
    {
      "epoch": 4.671599414953111,
      "grad_norm": 4960.0,
      "learning_rate": 1.0615453686773246e-05,
      "loss": 0.8333,
      "step": 54298
    },
    {
      "epoch": 4.671685451260432,
      "grad_norm": 0.984375,
      "learning_rate": 1.0609911970912434e-05,
      "loss": 0.6085,
      "step": 54299
    },
    {
      "epoch": 4.671771487567754,
      "grad_norm": 12.4375,
      "learning_rate": 1.0604371686422421e-05,
      "loss": 0.7655,
      "step": 54300
    },
    {
      "epoch": 4.671857523875075,
      "grad_norm": 1.171875,
      "learning_rate": 1.0598832833319472e-05,
      "loss": 1.1334,
      "step": 54301
    },
    {
      "epoch": 4.671943560182397,
      "grad_norm": 0.87890625,
      "learning_rate": 1.059329541161974e-05,
      "loss": 1.0017,
      "step": 54302
    },
    {
      "epoch": 4.672029596489718,
      "grad_norm": 0.9375,
      "learning_rate": 1.0587759421339438e-05,
      "loss": 1.1997,
      "step": 54303
    },
    {
      "epoch": 4.67211563279704,
      "grad_norm": 0.9921875,
      "learning_rate": 1.0582224862494716e-05,
      "loss": 1.0045,
      "step": 54304
    },
    {
      "epoch": 4.672201669104362,
      "grad_norm": 1.1484375,
      "learning_rate": 1.0576691735101896e-05,
      "loss": 1.1686,
      "step": 54305
    },
    {
      "epoch": 4.672287705411684,
      "grad_norm": 1.3984375,
      "learning_rate": 1.0571160039176963e-05,
      "loss": 1.0646,
      "step": 54306
    },
    {
      "epoch": 4.672373741719006,
      "grad_norm": 1.953125,
      "learning_rate": 1.0565629774736297e-05,
      "loss": 1.196,
      "step": 54307
    },
    {
      "epoch": 4.672459778026327,
      "grad_norm": 0.97265625,
      "learning_rate": 1.056010094179588e-05,
      "loss": 0.8098,
      "step": 54308
    },
    {
      "epoch": 4.672545814333649,
      "grad_norm": 1.15625,
      "learning_rate": 1.055457354037198e-05,
      "loss": 1.1592,
      "step": 54309
    },
    {
      "epoch": 4.67263185064097,
      "grad_norm": 1.03125,
      "learning_rate": 1.0549047570480863e-05,
      "loss": 0.7875,
      "step": 54310
    },
    {
      "epoch": 4.672717886948292,
      "grad_norm": 1.4609375,
      "learning_rate": 1.0543523032138457e-05,
      "loss": 1.038,
      "step": 54311
    },
    {
      "epoch": 4.672803923255614,
      "grad_norm": 0.98046875,
      "learning_rate": 1.0537999925361142e-05,
      "loss": 0.8377,
      "step": 54312
    },
    {
      "epoch": 4.672889959562935,
      "grad_norm": 102.0,
      "learning_rate": 1.0532478250164956e-05,
      "loss": 1.1063,
      "step": 54313
    },
    {
      "epoch": 4.6729759958702575,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0526958006566057e-05,
      "loss": 0.909,
      "step": 54314
    },
    {
      "epoch": 4.673062032177579,
      "grad_norm": 1.1875,
      "learning_rate": 1.0521439194580652e-05,
      "loss": 1.6232,
      "step": 54315
    },
    {
      "epoch": 4.673148068484901,
      "grad_norm": 1.1796875,
      "learning_rate": 1.0515921814224838e-05,
      "loss": 0.8577,
      "step": 54316
    },
    {
      "epoch": 4.673234104792222,
      "grad_norm": 1.2734375,
      "learning_rate": 1.0510405865514715e-05,
      "loss": 1.665,
      "step": 54317
    },
    {
      "epoch": 4.673320141099544,
      "grad_norm": 2.234375,
      "learning_rate": 1.0504891348466494e-05,
      "loss": 1.2481,
      "step": 54318
    },
    {
      "epoch": 4.673406177406866,
      "grad_norm": 68.0,
      "learning_rate": 1.0499378263096216e-05,
      "loss": 1.0971,
      "step": 54319
    },
    {
      "epoch": 4.673492213714187,
      "grad_norm": 1.03125,
      "learning_rate": 1.0493866609420033e-05,
      "loss": 0.8983,
      "step": 54320
    },
    {
      "epoch": 4.673578250021509,
      "grad_norm": 0.99609375,
      "learning_rate": 1.0488356387454157e-05,
      "loss": 1.9521,
      "step": 54321
    },
    {
      "epoch": 4.6736642863288305,
      "grad_norm": 1.109375,
      "learning_rate": 1.048284759721463e-05,
      "loss": 1.4432,
      "step": 54322
    },
    {
      "epoch": 4.6737503226361525,
      "grad_norm": 144.0,
      "learning_rate": 1.0477340238717493e-05,
      "loss": 1.2989,
      "step": 54323
    },
    {
      "epoch": 4.673836358943474,
      "grad_norm": 1.5703125,
      "learning_rate": 1.0471834311979012e-05,
      "loss": 1.0869,
      "step": 54324
    },
    {
      "epoch": 4.673922395250796,
      "grad_norm": 1.2734375,
      "learning_rate": 1.0466329817015175e-05,
      "loss": 1.2488,
      "step": 54325
    },
    {
      "epoch": 4.674008431558118,
      "grad_norm": 1.1328125,
      "learning_rate": 1.0460826753842078e-05,
      "loss": 1.0558,
      "step": 54326
    },
    {
      "epoch": 4.674094467865439,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0455325122475934e-05,
      "loss": 1.2991,
      "step": 54327
    },
    {
      "epoch": 4.674180504172761,
      "grad_norm": 41.75,
      "learning_rate": 1.0449824922932615e-05,
      "loss": 1.1585,
      "step": 54328
    },
    {
      "epoch": 4.674266540480082,
      "grad_norm": 18.0,
      "learning_rate": 1.04443261552285e-05,
      "loss": 1.2554,
      "step": 54329
    },
    {
      "epoch": 4.674352576787404,
      "grad_norm": 1.0,
      "learning_rate": 1.0438828819379465e-05,
      "loss": 0.9292,
      "step": 54330
    },
    {
      "epoch": 4.6744386130947255,
      "grad_norm": 0.99609375,
      "learning_rate": 1.0433332915401606e-05,
      "loss": 1.3026,
      "step": 54331
    },
    {
      "epoch": 4.674524649402048,
      "grad_norm": 0.88671875,
      "learning_rate": 1.0427838443311077e-05,
      "loss": 1.5253,
      "step": 54332
    },
    {
      "epoch": 4.67461068570937,
      "grad_norm": 412.0,
      "learning_rate": 1.0422345403123867e-05,
      "loss": 0.9203,
      "step": 54333
    },
    {
      "epoch": 4.674696722016691,
      "grad_norm": 1.0546875,
      "learning_rate": 1.041685379485613e-05,
      "loss": 1.2174,
      "step": 54334
    },
    {
      "epoch": 4.674782758324013,
      "grad_norm": 0.8046875,
      "learning_rate": 1.041136361852385e-05,
      "loss": 0.781,
      "step": 54335
    },
    {
      "epoch": 4.674868794631334,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0405874874143074e-05,
      "loss": 1.5299,
      "step": 54336
    },
    {
      "epoch": 4.674954830938656,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0400387561729952e-05,
      "loss": 0.5865,
      "step": 54337
    },
    {
      "epoch": 4.675040867245977,
      "grad_norm": 1.140625,
      "learning_rate": 1.0394901681300418e-05,
      "loss": 1.2292,
      "step": 54338
    },
    {
      "epoch": 4.675126903553299,
      "grad_norm": 1.0078125,
      "learning_rate": 1.0389417232870623e-05,
      "loss": 0.9072,
      "step": 54339
    },
    {
      "epoch": 4.6752129398606215,
      "grad_norm": 1.2578125,
      "learning_rate": 1.038393421645656e-05,
      "loss": 1.0589,
      "step": 54340
    },
    {
      "epoch": 4.675298976167943,
      "grad_norm": 1.0546875,
      "learning_rate": 1.037845263207421e-05,
      "loss": 1.1085,
      "step": 54341
    },
    {
      "epoch": 4.675385012475265,
      "grad_norm": 2.53125,
      "learning_rate": 1.0372972479739673e-05,
      "loss": 0.9603,
      "step": 54342
    },
    {
      "epoch": 4.675471048782586,
      "grad_norm": 0.98046875,
      "learning_rate": 1.0367493759469049e-05,
      "loss": 0.8208,
      "step": 54343
    },
    {
      "epoch": 4.675557085089908,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0362016471278212e-05,
      "loss": 1.135,
      "step": 54344
    },
    {
      "epoch": 4.675643121397229,
      "grad_norm": 3.859375,
      "learning_rate": 1.0356540615183207e-05,
      "loss": 1.135,
      "step": 54345
    },
    {
      "epoch": 4.675729157704551,
      "grad_norm": 1.4140625,
      "learning_rate": 1.0351066191200187e-05,
      "loss": 2.0712,
      "step": 54346
    },
    {
      "epoch": 4.675815194011873,
      "grad_norm": 1.171875,
      "learning_rate": 1.0345593199344971e-05,
      "loss": 0.9448,
      "step": 54347
    },
    {
      "epoch": 4.6759012303191945,
      "grad_norm": 0.97265625,
      "learning_rate": 1.0340121639633715e-05,
      "loss": 0.7606,
      "step": 54348
    },
    {
      "epoch": 4.6759872666265165,
      "grad_norm": 1.125,
      "learning_rate": 1.033465151208235e-05,
      "loss": 0.5101,
      "step": 54349
    },
    {
      "epoch": 4.676073302933838,
      "grad_norm": 1.21875,
      "learning_rate": 1.0329182816706916e-05,
      "loss": 1.9281,
      "step": 54350
    },
    {
      "epoch": 4.67615933924116,
      "grad_norm": 1.0546875,
      "learning_rate": 1.0323715553523405e-05,
      "loss": 0.8093,
      "step": 54351
    },
    {
      "epoch": 4.676245375548481,
      "grad_norm": 1.1015625,
      "learning_rate": 1.0318249722547745e-05,
      "loss": 1.0074,
      "step": 54352
    },
    {
      "epoch": 4.676331411855803,
      "grad_norm": 0.828125,
      "learning_rate": 1.031278532379598e-05,
      "loss": 0.5028,
      "step": 54353
    },
    {
      "epoch": 4.676417448163125,
      "grad_norm": 1.234375,
      "learning_rate": 1.0307322357284098e-05,
      "loss": 1.7799,
      "step": 54354
    },
    {
      "epoch": 4.676503484470446,
      "grad_norm": 1.2734375,
      "learning_rate": 1.0301860823028031e-05,
      "loss": 1.4903,
      "step": 54355
    },
    {
      "epoch": 4.676589520777768,
      "grad_norm": 28.0,
      "learning_rate": 1.0296400721043764e-05,
      "loss": 1.0685,
      "step": 54356
    },
    {
      "epoch": 4.6766755570850895,
      "grad_norm": 0.77734375,
      "learning_rate": 1.029094205134734e-05,
      "loss": 0.4864,
      "step": 54357
    },
    {
      "epoch": 4.676761593392412,
      "grad_norm": 1.5,
      "learning_rate": 1.0285484813954582e-05,
      "loss": 1.1529,
      "step": 54358
    },
    {
      "epoch": 4.676847629699734,
      "grad_norm": 0.87109375,
      "learning_rate": 1.0280029008881641e-05,
      "loss": 0.6584,
      "step": 54359
    },
    {
      "epoch": 4.676933666007055,
      "grad_norm": 1.3203125,
      "learning_rate": 1.0274574636144285e-05,
      "loss": 1.7874,
      "step": 54360
    },
    {
      "epoch": 4.677019702314377,
      "grad_norm": 0.8828125,
      "learning_rate": 1.026912169575861e-05,
      "loss": 0.7442,
      "step": 54361
    },
    {
      "epoch": 4.677105738621698,
      "grad_norm": 1.2734375,
      "learning_rate": 1.0263670187740493e-05,
      "loss": 0.7604,
      "step": 54362
    },
    {
      "epoch": 4.67719177492902,
      "grad_norm": 0.953125,
      "learning_rate": 1.0258220112105865e-05,
      "loss": 0.5925,
      "step": 54363
    },
    {
      "epoch": 4.677277811236342,
      "grad_norm": 0.83984375,
      "learning_rate": 1.0252771468870714e-05,
      "loss": 0.556,
      "step": 54364
    },
    {
      "epoch": 4.677363847543663,
      "grad_norm": 1.09375,
      "learning_rate": 1.0247324258050917e-05,
      "loss": 1.0981,
      "step": 54365
    },
    {
      "epoch": 4.6774498838509855,
      "grad_norm": 1.296875,
      "learning_rate": 1.0241878479662458e-05,
      "loss": 1.1508,
      "step": 54366
    },
    {
      "epoch": 4.677535920158307,
      "grad_norm": 1.1328125,
      "learning_rate": 1.0236434133721329e-05,
      "loss": 1.6248,
      "step": 54367
    },
    {
      "epoch": 4.677621956465629,
      "grad_norm": 1.09375,
      "learning_rate": 1.023099122024329e-05,
      "loss": 1.4219,
      "step": 54368
    },
    {
      "epoch": 4.67770799277295,
      "grad_norm": 1.2578125,
      "learning_rate": 1.022554973924439e-05,
      "loss": 1.082,
      "step": 54369
    },
    {
      "epoch": 4.677794029080272,
      "grad_norm": 0.9140625,
      "learning_rate": 1.0220109690740498e-05,
      "loss": 0.6071,
      "step": 54370
    },
    {
      "epoch": 4.677880065387594,
      "grad_norm": 1.3359375,
      "learning_rate": 1.0214671074747494e-05,
      "loss": 1.6912,
      "step": 54371
    },
    {
      "epoch": 4.677966101694915,
      "grad_norm": 2.890625,
      "learning_rate": 1.020923389128131e-05,
      "loss": 1.0766,
      "step": 54372
    },
    {
      "epoch": 4.678052138002237,
      "grad_norm": 1.6796875,
      "learning_rate": 1.0203798140357933e-05,
      "loss": 1.3607,
      "step": 54373
    },
    {
      "epoch": 4.6781381743095585,
      "grad_norm": 13.9375,
      "learning_rate": 1.0198363821993128e-05,
      "loss": 0.4789,
      "step": 54374
    },
    {
      "epoch": 4.6782242106168805,
      "grad_norm": 11.75,
      "learning_rate": 1.0192930936202883e-05,
      "loss": 1.7529,
      "step": 54375
    },
    {
      "epoch": 4.678310246924202,
      "grad_norm": 1.359375,
      "learning_rate": 1.0187499483003072e-05,
      "loss": 0.6608,
      "step": 54376
    },
    {
      "epoch": 4.678396283231524,
      "grad_norm": 0.9765625,
      "learning_rate": 1.018206946240946e-05,
      "loss": 0.7129,
      "step": 54377
    },
    {
      "epoch": 4.678482319538846,
      "grad_norm": 1.3125,
      "learning_rate": 1.0176640874438147e-05,
      "loss": 1.2349,
      "step": 54378
    },
    {
      "epoch": 4.678568355846167,
      "grad_norm": 1.265625,
      "learning_rate": 1.0171213719104844e-05,
      "loss": 1.9953,
      "step": 54379
    },
    {
      "epoch": 4.678654392153489,
      "grad_norm": 1.1171875,
      "learning_rate": 1.0165787996425423e-05,
      "loss": 0.9631,
      "step": 54380
    },
    {
      "epoch": 4.67874042846081,
      "grad_norm": 1.140625,
      "learning_rate": 1.0160363706415931e-05,
      "loss": 0.5996,
      "step": 54381
    },
    {
      "epoch": 4.678826464768132,
      "grad_norm": 1.328125,
      "learning_rate": 1.0154940849091965e-05,
      "loss": 1.3844,
      "step": 54382
    },
    {
      "epoch": 4.6789125010754535,
      "grad_norm": 1.1796875,
      "learning_rate": 1.0149519424469621e-05,
      "loss": 0.6489,
      "step": 54383
    },
    {
      "epoch": 4.678998537382776,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0144099432564668e-05,
      "loss": 1.0688,
      "step": 54384
    },
    {
      "epoch": 4.679084573690098,
      "grad_norm": 0.88671875,
      "learning_rate": 1.0138680873392869e-05,
      "loss": 0.8256,
      "step": 54385
    },
    {
      "epoch": 4.679170609997419,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0133263746970211e-05,
      "loss": 0.6231,
      "step": 54386
    },
    {
      "epoch": 4.679256646304741,
      "grad_norm": 1.296875,
      "learning_rate": 1.0127848053312516e-05,
      "loss": 1.3158,
      "step": 54387
    },
    {
      "epoch": 4.679342682612062,
      "grad_norm": 0.90625,
      "learning_rate": 1.0122433792435492e-05,
      "loss": 0.7576,
      "step": 54388
    },
    {
      "epoch": 4.679428718919384,
      "grad_norm": 1.625,
      "learning_rate": 1.0117020964355184e-05,
      "loss": 1.7437,
      "step": 54389
    },
    {
      "epoch": 4.679514755226705,
      "grad_norm": 0.98046875,
      "learning_rate": 1.0111609569087244e-05,
      "loss": 0.5608,
      "step": 54390
    },
    {
      "epoch": 4.679600791534027,
      "grad_norm": 0.8828125,
      "learning_rate": 1.0106199606647549e-05,
      "loss": 0.7536,
      "step": 54391
    },
    {
      "epoch": 4.6796868278413495,
      "grad_norm": 1.3671875,
      "learning_rate": 1.010079107705203e-05,
      "loss": 0.895,
      "step": 54392
    },
    {
      "epoch": 4.679772864148671,
      "grad_norm": 0.96875,
      "learning_rate": 1.0095383980316341e-05,
      "loss": 1.2836,
      "step": 54393
    },
    {
      "epoch": 4.679858900455993,
      "grad_norm": 33.75,
      "learning_rate": 1.008997831645636e-05,
      "loss": 0.6405,
      "step": 54394
    },
    {
      "epoch": 4.679944936763314,
      "grad_norm": 1.03125,
      "learning_rate": 1.0084574085488019e-05,
      "loss": 1.1651,
      "step": 54395
    },
    {
      "epoch": 4.680030973070636,
      "grad_norm": 2.234375,
      "learning_rate": 1.0079171287426914e-05,
      "loss": 1.4688,
      "step": 54396
    },
    {
      "epoch": 4.680117009377957,
      "grad_norm": 1.015625,
      "learning_rate": 1.0073769922288977e-05,
      "loss": 1.2244,
      "step": 54397
    },
    {
      "epoch": 4.680203045685279,
      "grad_norm": 1.046875,
      "learning_rate": 1.0068369990089977e-05,
      "loss": 0.9905,
      "step": 54398
    },
    {
      "epoch": 4.680289081992601,
      "grad_norm": 0.921875,
      "learning_rate": 1.0062971490845729e-05,
      "loss": 0.9186,
      "step": 54399
    },
    {
      "epoch": 4.6803751182999225,
      "grad_norm": 1.078125,
      "learning_rate": 1.0057574424572003e-05,
      "loss": 0.5136,
      "step": 54400
    },
    {
      "epoch": 4.6804611546072445,
      "grad_norm": 1.015625,
      "learning_rate": 1.005217879128456e-05,
      "loss": 0.6846,
      "step": 54401
    },
    {
      "epoch": 4.680547190914566,
      "grad_norm": 1.2265625,
      "learning_rate": 1.0046784590999114e-05,
      "loss": 1.8773,
      "step": 54402
    },
    {
      "epoch": 4.680633227221888,
      "grad_norm": 0.98046875,
      "learning_rate": 1.0041391823731649e-05,
      "loss": 1.1827,
      "step": 54403
    },
    {
      "epoch": 4.680719263529209,
      "grad_norm": 1.2734375,
      "learning_rate": 1.0036000489497821e-05,
      "loss": 1.5019,
      "step": 54404
    },
    {
      "epoch": 4.680805299836531,
      "grad_norm": 0.88671875,
      "learning_rate": 1.0030610588313394e-05,
      "loss": 1.05,
      "step": 54405
    },
    {
      "epoch": 4.680891336143853,
      "grad_norm": 1.2734375,
      "learning_rate": 1.0025222120194077e-05,
      "loss": 1.3247,
      "step": 54406
    },
    {
      "epoch": 4.680977372451174,
      "grad_norm": 1.265625,
      "learning_rate": 1.0019835085155748e-05,
      "loss": 1.1697,
      "step": 54407
    },
    {
      "epoch": 4.681063408758496,
      "grad_norm": 1.21875,
      "learning_rate": 1.0014449483214117e-05,
      "loss": 0.8116,
      "step": 54408
    },
    {
      "epoch": 4.6811494450658175,
      "grad_norm": 1.015625,
      "learning_rate": 1.000906531438489e-05,
      "loss": 1.0539,
      "step": 54409
    },
    {
      "epoch": 4.68123548137314,
      "grad_norm": 3.4375,
      "learning_rate": 1.0003682578683837e-05,
      "loss": 1.4505,
      "step": 54410
    },
    {
      "epoch": 4.681321517680461,
      "grad_norm": 0.78515625,
      "learning_rate": 9.99830127612672e-06,
      "loss": 0.7103,
      "step": 54411
    },
    {
      "epoch": 4.681407553987783,
      "grad_norm": 4.1875,
      "learning_rate": 9.992921406729305e-06,
      "loss": 0.662,
      "step": 54412
    },
    {
      "epoch": 4.681493590295105,
      "grad_norm": 0.984375,
      "learning_rate": 9.987542970507246e-06,
      "loss": 0.637,
      "step": 54413
    },
    {
      "epoch": 4.681579626602426,
      "grad_norm": 1.0546875,
      "learning_rate": 9.982165967476364e-06,
      "loss": 1.2899,
      "step": 54414
    },
    {
      "epoch": 4.681665662909748,
      "grad_norm": 1.2890625,
      "learning_rate": 9.976790397652312e-06,
      "loss": 1.7203,
      "step": 54415
    },
    {
      "epoch": 4.681751699217069,
      "grad_norm": 0.9765625,
      "learning_rate": 9.971416261050858e-06,
      "loss": 0.7551,
      "step": 54416
    },
    {
      "epoch": 4.681837735524391,
      "grad_norm": 62.25,
      "learning_rate": 9.966043557687709e-06,
      "loss": 1.3288,
      "step": 54417
    },
    {
      "epoch": 4.681923771831713,
      "grad_norm": 1.03125,
      "learning_rate": 9.960672287578576e-06,
      "loss": 1.4058,
      "step": 54418
    },
    {
      "epoch": 4.682009808139035,
      "grad_norm": 1.078125,
      "learning_rate": 9.955302450739223e-06,
      "loss": 1.0671,
      "step": 54419
    },
    {
      "epoch": 4.682095844446357,
      "grad_norm": 55.5,
      "learning_rate": 9.949934047185194e-06,
      "loss": 1.0292,
      "step": 54420
    },
    {
      "epoch": 4.682181880753678,
      "grad_norm": 0.89453125,
      "learning_rate": 9.944567076932365e-06,
      "loss": 0.706,
      "step": 54421
    },
    {
      "epoch": 4.682267917061,
      "grad_norm": 0.6796875,
      "learning_rate": 9.93920153999639e-06,
      "loss": 0.5561,
      "step": 54422
    },
    {
      "epoch": 4.682353953368321,
      "grad_norm": 38.25,
      "learning_rate": 9.933837436392868e-06,
      "loss": 1.3621,
      "step": 54423
    },
    {
      "epoch": 4.682439989675643,
      "grad_norm": 1.5234375,
      "learning_rate": 9.928474766137562e-06,
      "loss": 1.1508,
      "step": 54424
    },
    {
      "epoch": 4.682526025982964,
      "grad_norm": 1.2578125,
      "learning_rate": 9.923113529246242e-06,
      "loss": 1.2998,
      "step": 54425
    },
    {
      "epoch": 4.6826120622902865,
      "grad_norm": 0.6796875,
      "learning_rate": 9.917753725734446e-06,
      "loss": 0.2801,
      "step": 54426
    },
    {
      "epoch": 4.6826980985976085,
      "grad_norm": 1.453125,
      "learning_rate": 9.912395355617887e-06,
      "loss": 0.9236,
      "step": 54427
    },
    {
      "epoch": 4.68278413490493,
      "grad_norm": 1.2734375,
      "learning_rate": 9.907038418912273e-06,
      "loss": 0.9298,
      "step": 54428
    },
    {
      "epoch": 4.682870171212252,
      "grad_norm": 1.0546875,
      "learning_rate": 9.90168291563326e-06,
      "loss": 1.2649,
      "step": 54429
    },
    {
      "epoch": 4.682956207519573,
      "grad_norm": 0.86328125,
      "learning_rate": 9.896328845796498e-06,
      "loss": 0.6968,
      "step": 54430
    },
    {
      "epoch": 4.683042243826895,
      "grad_norm": 0.91796875,
      "learning_rate": 9.890976209417646e-06,
      "loss": 0.9176,
      "step": 54431
    },
    {
      "epoch": 4.683128280134216,
      "grad_norm": 1.34375,
      "learning_rate": 9.885625006512356e-06,
      "loss": 1.2006,
      "step": 54432
    },
    {
      "epoch": 4.683214316441538,
      "grad_norm": 24.375,
      "learning_rate": 9.880275237096337e-06,
      "loss": 0.9481,
      "step": 54433
    },
    {
      "epoch": 4.68330035274886,
      "grad_norm": 1.1953125,
      "learning_rate": 9.874926901185133e-06,
      "loss": 1.3822,
      "step": 54434
    },
    {
      "epoch": 4.6833863890561815,
      "grad_norm": 1.1875,
      "learning_rate": 9.869579998794453e-06,
      "loss": 1.2595,
      "step": 54435
    },
    {
      "epoch": 4.683472425363504,
      "grad_norm": 1.828125,
      "learning_rate": 9.864234529940008e-06,
      "loss": 1.5677,
      "step": 54436
    },
    {
      "epoch": 4.683558461670825,
      "grad_norm": 1.203125,
      "learning_rate": 9.858890494637284e-06,
      "loss": 1.3525,
      "step": 54437
    },
    {
      "epoch": 4.683644497978147,
      "grad_norm": 1.28125,
      "learning_rate": 9.853547892901938e-06,
      "loss": 1.921,
      "step": 54438
    },
    {
      "epoch": 4.683730534285468,
      "grad_norm": 1.0546875,
      "learning_rate": 9.848206724749675e-06,
      "loss": 0.7555,
      "step": 54439
    },
    {
      "epoch": 4.68381657059279,
      "grad_norm": 1.1953125,
      "learning_rate": 9.842866990196098e-06,
      "loss": 1.548,
      "step": 54440
    },
    {
      "epoch": 4.683902606900112,
      "grad_norm": 0.98828125,
      "learning_rate": 9.837528689256803e-06,
      "loss": 0.8116,
      "step": 54441
    },
    {
      "epoch": 4.683988643207433,
      "grad_norm": 1.4765625,
      "learning_rate": 9.83219182194739e-06,
      "loss": 1.5311,
      "step": 54442
    },
    {
      "epoch": 4.684074679514755,
      "grad_norm": 1.2265625,
      "learning_rate": 9.826856388283456e-06,
      "loss": 1.5066,
      "step": 54443
    },
    {
      "epoch": 4.684160715822077,
      "grad_norm": 1.0859375,
      "learning_rate": 9.821522388280712e-06,
      "loss": 1.003,
      "step": 54444
    },
    {
      "epoch": 4.684246752129399,
      "grad_norm": 1.5234375,
      "learning_rate": 9.816189821954591e-06,
      "loss": 1.1867,
      "step": 54445
    },
    {
      "epoch": 4.68433278843672,
      "grad_norm": 0.8828125,
      "learning_rate": 9.810858689320857e-06,
      "loss": 1.1882,
      "step": 54446
    },
    {
      "epoch": 4.684418824744042,
      "grad_norm": 1.21875,
      "learning_rate": 9.805528990394996e-06,
      "loss": 0.8291,
      "step": 54447
    },
    {
      "epoch": 4.684504861051364,
      "grad_norm": 0.91796875,
      "learning_rate": 9.800200725192553e-06,
      "loss": 1.37,
      "step": 54448
    },
    {
      "epoch": 4.684590897358685,
      "grad_norm": 1.1796875,
      "learning_rate": 9.794873893729295e-06,
      "loss": 0.7771,
      "step": 54449
    },
    {
      "epoch": 4.684676933666007,
      "grad_norm": 0.8515625,
      "learning_rate": 9.78954849602065e-06,
      "loss": 1.4332,
      "step": 54450
    },
    {
      "epoch": 4.684762969973328,
      "grad_norm": 0.99609375,
      "learning_rate": 9.78422453208222e-06,
      "loss": 0.6495,
      "step": 54451
    },
    {
      "epoch": 4.6848490062806505,
      "grad_norm": 0.82421875,
      "learning_rate": 9.778902001929601e-06,
      "loss": 0.5283,
      "step": 54452
    },
    {
      "epoch": 4.684935042587972,
      "grad_norm": 1.1875,
      "learning_rate": 9.77358090557834e-06,
      "loss": 1.223,
      "step": 54453
    },
    {
      "epoch": 4.685021078895294,
      "grad_norm": 360.0,
      "learning_rate": 9.76826124304403e-06,
      "loss": 1.3101,
      "step": 54454
    },
    {
      "epoch": 4.685107115202616,
      "grad_norm": 0.921875,
      "learning_rate": 9.762943014342218e-06,
      "loss": 0.5567,
      "step": 54455
    },
    {
      "epoch": 4.685193151509937,
      "grad_norm": 1.21875,
      "learning_rate": 9.757626219488447e-06,
      "loss": 1.2845,
      "step": 54456
    },
    {
      "epoch": 4.685279187817259,
      "grad_norm": 0.8515625,
      "learning_rate": 9.752310858498259e-06,
      "loss": 0.9654,
      "step": 54457
    },
    {
      "epoch": 4.68536522412458,
      "grad_norm": 1.1875,
      "learning_rate": 9.746996931387253e-06,
      "loss": 0.7734,
      "step": 54458
    },
    {
      "epoch": 4.685451260431902,
      "grad_norm": 1.1015625,
      "learning_rate": 9.741684438170862e-06,
      "loss": 1.3899,
      "step": 54459
    },
    {
      "epoch": 4.6855372967392235,
      "grad_norm": 1.84375,
      "learning_rate": 9.736373378864739e-06,
      "loss": 1.6999,
      "step": 54460
    },
    {
      "epoch": 4.6856233330465455,
      "grad_norm": 2.671875,
      "learning_rate": 9.73106375348437e-06,
      "loss": 1.2861,
      "step": 54461
    },
    {
      "epoch": 4.685709369353868,
      "grad_norm": 3.828125,
      "learning_rate": 9.725755562045247e-06,
      "loss": 1.5741,
      "step": 54462
    },
    {
      "epoch": 4.685795405661189,
      "grad_norm": 1.171875,
      "learning_rate": 9.72044880456302e-06,
      "loss": 1.1137,
      "step": 54463
    },
    {
      "epoch": 4.685881441968511,
      "grad_norm": 1.4296875,
      "learning_rate": 9.715143481053067e-06,
      "loss": 1.4004,
      "step": 54464
    },
    {
      "epoch": 4.685967478275832,
      "grad_norm": 1.125,
      "learning_rate": 9.709839591530934e-06,
      "loss": 0.8626,
      "step": 54465
    },
    {
      "epoch": 4.686053514583154,
      "grad_norm": 1.0546875,
      "learning_rate": 9.704537136012271e-06,
      "loss": 1.3948,
      "step": 54466
    },
    {
      "epoch": 4.686139550890475,
      "grad_norm": 1.125,
      "learning_rate": 9.699236114512345e-06,
      "loss": 0.8241,
      "step": 54467
    },
    {
      "epoch": 4.686225587197797,
      "grad_norm": 1.3984375,
      "learning_rate": 9.693936527046865e-06,
      "loss": 1.4936,
      "step": 54468
    },
    {
      "epoch": 4.686311623505119,
      "grad_norm": 1.59375,
      "learning_rate": 9.68863837363132e-06,
      "loss": 0.9288,
      "step": 54469
    },
    {
      "epoch": 4.686397659812441,
      "grad_norm": 1.0390625,
      "learning_rate": 9.68334165428103e-06,
      "loss": 0.7936,
      "step": 54470
    },
    {
      "epoch": 4.686483696119763,
      "grad_norm": 1.25,
      "learning_rate": 9.678046369011706e-06,
      "loss": 1.3426,
      "step": 54471
    },
    {
      "epoch": 4.686569732427084,
      "grad_norm": 344.0,
      "learning_rate": 9.672752517838668e-06,
      "loss": 1.3098,
      "step": 54472
    },
    {
      "epoch": 4.686655768734406,
      "grad_norm": 1.046875,
      "learning_rate": 9.667460100777458e-06,
      "loss": 0.5683,
      "step": 54473
    },
    {
      "epoch": 4.686741805041728,
      "grad_norm": 1.0390625,
      "learning_rate": 9.662169117843622e-06,
      "loss": 1.1296,
      "step": 54474
    },
    {
      "epoch": 4.686827841349049,
      "grad_norm": 1.1640625,
      "learning_rate": 9.656879569052534e-06,
      "loss": 0.7405,
      "step": 54475
    },
    {
      "epoch": 4.686913877656371,
      "grad_norm": 1.0,
      "learning_rate": 9.651591454419738e-06,
      "loss": 0.7081,
      "step": 54476
    },
    {
      "epoch": 4.686999913963692,
      "grad_norm": 1.15625,
      "learning_rate": 9.646304773960613e-06,
      "loss": 0.8057,
      "step": 54477
    },
    {
      "epoch": 4.6870859502710145,
      "grad_norm": 1.140625,
      "learning_rate": 9.641019527690698e-06,
      "loss": 1.8204,
      "step": 54478
    },
    {
      "epoch": 4.687171986578336,
      "grad_norm": 1.171875,
      "learning_rate": 9.635735715625481e-06,
      "loss": 1.1905,
      "step": 54479
    },
    {
      "epoch": 4.687258022885658,
      "grad_norm": 11.0625,
      "learning_rate": 9.630453337780343e-06,
      "loss": 1.3474,
      "step": 54480
    },
    {
      "epoch": 4.68734405919298,
      "grad_norm": 1.4375,
      "learning_rate": 9.625172394170711e-06,
      "loss": 1.9707,
      "step": 54481
    },
    {
      "epoch": 4.687430095500301,
      "grad_norm": 1.1796875,
      "learning_rate": 9.619892884812132e-06,
      "loss": 1.4127,
      "step": 54482
    },
    {
      "epoch": 4.687516131807623,
      "grad_norm": 1.1875,
      "learning_rate": 9.61461480971998e-06,
      "loss": 0.6972,
      "step": 54483
    },
    {
      "epoch": 4.687602168114944,
      "grad_norm": 1.1328125,
      "learning_rate": 9.609338168909686e-06,
      "loss": 0.7261,
      "step": 54484
    },
    {
      "epoch": 4.687688204422266,
      "grad_norm": 1.359375,
      "learning_rate": 9.604062962396742e-06,
      "loss": 1.2831,
      "step": 54485
    },
    {
      "epoch": 4.687774240729588,
      "grad_norm": 0.953125,
      "learning_rate": 9.598789190196466e-06,
      "loss": 0.4533,
      "step": 54486
    },
    {
      "epoch": 4.6878602770369096,
      "grad_norm": 2.71875,
      "learning_rate": 9.593516852324402e-06,
      "loss": 1.572,
      "step": 54487
    },
    {
      "epoch": 4.687946313344232,
      "grad_norm": 0.88671875,
      "learning_rate": 9.588245948795927e-06,
      "loss": 0.4117,
      "step": 54488
    },
    {
      "epoch": 4.688032349651553,
      "grad_norm": 0.8046875,
      "learning_rate": 9.582976479626471e-06,
      "loss": 0.5324,
      "step": 54489
    },
    {
      "epoch": 4.688118385958875,
      "grad_norm": 1.2265625,
      "learning_rate": 9.577708444831412e-06,
      "loss": 0.8326,
      "step": 54490
    },
    {
      "epoch": 4.688204422266196,
      "grad_norm": 0.92578125,
      "learning_rate": 9.572441844426128e-06,
      "loss": 1.0088,
      "step": 54491
    },
    {
      "epoch": 4.688290458573518,
      "grad_norm": 1.0,
      "learning_rate": 9.567176678426104e-06,
      "loss": 1.0674,
      "step": 54492
    },
    {
      "epoch": 4.68837649488084,
      "grad_norm": 1.2421875,
      "learning_rate": 9.561912946846719e-06,
      "loss": 0.8162,
      "step": 54493
    },
    {
      "epoch": 4.688462531188161,
      "grad_norm": 1.0859375,
      "learning_rate": 9.556650649703347e-06,
      "loss": 1.0079,
      "step": 54494
    },
    {
      "epoch": 4.688548567495483,
      "grad_norm": 1.3828125,
      "learning_rate": 9.551389787011367e-06,
      "loss": 1.2641,
      "step": 54495
    },
    {
      "epoch": 4.688634603802805,
      "grad_norm": 1.21875,
      "learning_rate": 9.54613035878621e-06,
      "loss": 1.0111,
      "step": 54496
    },
    {
      "epoch": 4.688720640110127,
      "grad_norm": 37.5,
      "learning_rate": 9.540872365043196e-06,
      "loss": 1.1912,
      "step": 54497
    },
    {
      "epoch": 4.688806676417448,
      "grad_norm": 1.484375,
      "learning_rate": 9.535615805797759e-06,
      "loss": 2.0562,
      "step": 54498
    },
    {
      "epoch": 4.68889271272477,
      "grad_norm": 1.234375,
      "learning_rate": 9.530360681065276e-06,
      "loss": 1.4146,
      "step": 54499
    },
    {
      "epoch": 4.688978749032092,
      "grad_norm": 1.09375,
      "learning_rate": 9.525106990861066e-06,
      "loss": 1.0169,
      "step": 54500
    },
    {
      "epoch": 4.689064785339413,
      "grad_norm": 1.28125,
      "learning_rate": 9.519854735200616e-06,
      "loss": 0.9981,
      "step": 54501
    },
    {
      "epoch": 4.689150821646735,
      "grad_norm": 1.046875,
      "learning_rate": 9.514603914099085e-06,
      "loss": 1.3876,
      "step": 54502
    },
    {
      "epoch": 4.689236857954056,
      "grad_norm": 1.5,
      "learning_rate": 9.509354527571955e-06,
      "loss": 1.3687,
      "step": 54503
    },
    {
      "epoch": 4.6893228942613785,
      "grad_norm": 1.5234375,
      "learning_rate": 9.504106575634664e-06,
      "loss": 1.5447,
      "step": 54504
    },
    {
      "epoch": 4.6894089305687,
      "grad_norm": 1.0625,
      "learning_rate": 9.49886005830236e-06,
      "loss": 0.9084,
      "step": 54505
    },
    {
      "epoch": 4.689494966876022,
      "grad_norm": 0.98046875,
      "learning_rate": 9.493614975590536e-06,
      "loss": 1.3363,
      "step": 54506
    },
    {
      "epoch": 4.689581003183344,
      "grad_norm": 1.4453125,
      "learning_rate": 9.488371327514511e-06,
      "loss": 1.3801,
      "step": 54507
    },
    {
      "epoch": 4.689667039490665,
      "grad_norm": 1.046875,
      "learning_rate": 9.483129114089551e-06,
      "loss": 0.7254,
      "step": 54508
    },
    {
      "epoch": 4.689753075797987,
      "grad_norm": 1.0,
      "learning_rate": 9.477888335331032e-06,
      "loss": 0.7593,
      "step": 54509
    },
    {
      "epoch": 4.689839112105308,
      "grad_norm": 536.0,
      "learning_rate": 9.47264899125433e-06,
      "loss": 1.5357,
      "step": 54510
    },
    {
      "epoch": 4.68992514841263,
      "grad_norm": 1.0625,
      "learning_rate": 9.46741108187471e-06,
      "loss": 0.9966,
      "step": 54511
    },
    {
      "epoch": 4.6900111847199515,
      "grad_norm": 1.0625,
      "learning_rate": 9.462174607207552e-06,
      "loss": 0.906,
      "step": 54512
    },
    {
      "epoch": 4.6900972210272736,
      "grad_norm": 1.3828125,
      "learning_rate": 9.456939567268008e-06,
      "loss": 0.9274,
      "step": 54513
    },
    {
      "epoch": 4.690183257334596,
      "grad_norm": 1.703125,
      "learning_rate": 9.451705962071622e-06,
      "loss": 1.7645,
      "step": 54514
    },
    {
      "epoch": 4.690269293641917,
      "grad_norm": 1.0625,
      "learning_rate": 9.446473791633548e-06,
      "loss": 0.7651,
      "step": 54515
    },
    {
      "epoch": 4.690355329949239,
      "grad_norm": 3.796875,
      "learning_rate": 9.441243055969162e-06,
      "loss": 1.089,
      "step": 54516
    },
    {
      "epoch": 4.69044136625656,
      "grad_norm": 0.91796875,
      "learning_rate": 9.436013755093731e-06,
      "loss": 0.766,
      "step": 54517
    },
    {
      "epoch": 4.690527402563882,
      "grad_norm": 1.484375,
      "learning_rate": 9.430785889022519e-06,
      "loss": 1.2164,
      "step": 54518
    },
    {
      "epoch": 4.690613438871203,
      "grad_norm": 1.3828125,
      "learning_rate": 9.42555945777085e-06,
      "loss": 0.9287,
      "step": 54519
    },
    {
      "epoch": 4.690699475178525,
      "grad_norm": 0.9296875,
      "learning_rate": 9.42033446135404e-06,
      "loss": 1.1799,
      "step": 54520
    },
    {
      "epoch": 4.690785511485847,
      "grad_norm": 1.296875,
      "learning_rate": 9.415110899787304e-06,
      "loss": 1.4657,
      "step": 54521
    },
    {
      "epoch": 4.690871547793169,
      "grad_norm": 0.953125,
      "learning_rate": 9.409888773085961e-06,
      "loss": 1.6921,
      "step": 54522
    },
    {
      "epoch": 4.690957584100491,
      "grad_norm": 0.98046875,
      "learning_rate": 9.404668081265333e-06,
      "loss": 0.8471,
      "step": 54523
    },
    {
      "epoch": 4.691043620407812,
      "grad_norm": 1.015625,
      "learning_rate": 9.399448824340574e-06,
      "loss": 1.2595,
      "step": 54524
    },
    {
      "epoch": 4.691129656715134,
      "grad_norm": 0.80859375,
      "learning_rate": 9.394231002327003e-06,
      "loss": 1.0881,
      "step": 54525
    },
    {
      "epoch": 4.691215693022455,
      "grad_norm": 3.71875,
      "learning_rate": 9.389014615239943e-06,
      "loss": 1.4887,
      "step": 54526
    },
    {
      "epoch": 4.691301729329777,
      "grad_norm": 1.5859375,
      "learning_rate": 9.38379966309455e-06,
      "loss": 1.0012,
      "step": 54527
    },
    {
      "epoch": 4.691387765637099,
      "grad_norm": 0.99609375,
      "learning_rate": 9.378586145906143e-06,
      "loss": 0.8194,
      "step": 54528
    },
    {
      "epoch": 4.69147380194442,
      "grad_norm": 1.390625,
      "learning_rate": 9.373374063689932e-06,
      "loss": 1.0732,
      "step": 54529
    },
    {
      "epoch": 4.6915598382517425,
      "grad_norm": 1.3515625,
      "learning_rate": 9.368163416461184e-06,
      "loss": 1.1572,
      "step": 54530
    },
    {
      "epoch": 4.691645874559064,
      "grad_norm": 1.4140625,
      "learning_rate": 9.362954204235164e-06,
      "loss": 1.0374,
      "step": 54531
    },
    {
      "epoch": 4.691731910866386,
      "grad_norm": 1.0,
      "learning_rate": 9.35774642702708e-06,
      "loss": 0.719,
      "step": 54532
    },
    {
      "epoch": 4.691817947173707,
      "grad_norm": 1.140625,
      "learning_rate": 9.352540084852146e-06,
      "loss": 0.7056,
      "step": 54533
    },
    {
      "epoch": 4.691903983481029,
      "grad_norm": 1.0078125,
      "learning_rate": 9.347335177725624e-06,
      "loss": 0.9832,
      "step": 54534
    },
    {
      "epoch": 4.691990019788351,
      "grad_norm": 2.125,
      "learning_rate": 9.34213170566267e-06,
      "loss": 1.0473,
      "step": 54535
    },
    {
      "epoch": 4.692076056095672,
      "grad_norm": 3.75,
      "learning_rate": 9.336929668678606e-06,
      "loss": 0.743,
      "step": 54536
    },
    {
      "epoch": 4.692162092402994,
      "grad_norm": 1.1953125,
      "learning_rate": 9.331729066788585e-06,
      "loss": 0.9121,
      "step": 54537
    },
    {
      "epoch": 4.6922481287103155,
      "grad_norm": 1.296875,
      "learning_rate": 9.326529900007762e-06,
      "loss": 1.1674,
      "step": 54538
    },
    {
      "epoch": 4.692334165017638,
      "grad_norm": 26.75,
      "learning_rate": 9.32133216835146e-06,
      "loss": 1.2846,
      "step": 54539
    },
    {
      "epoch": 4.692420201324959,
      "grad_norm": 0.8125,
      "learning_rate": 9.316135871834885e-06,
      "loss": 0.5395,
      "step": 54540
    },
    {
      "epoch": 4.692506237632281,
      "grad_norm": 1.1796875,
      "learning_rate": 9.310941010473084e-06,
      "loss": 0.8853,
      "step": 54541
    },
    {
      "epoch": 4.692592273939603,
      "grad_norm": 1.265625,
      "learning_rate": 9.305747584281431e-06,
      "loss": 1.662,
      "step": 54542
    },
    {
      "epoch": 4.692678310246924,
      "grad_norm": 1.046875,
      "learning_rate": 9.300555593274974e-06,
      "loss": 1.7065,
      "step": 54543
    },
    {
      "epoch": 4.692764346554246,
      "grad_norm": 1.0859375,
      "learning_rate": 9.295365037468972e-06,
      "loss": 0.9764,
      "step": 54544
    },
    {
      "epoch": 4.692850382861567,
      "grad_norm": 0.94140625,
      "learning_rate": 9.290175916878586e-06,
      "loss": 1.5618,
      "step": 54545
    },
    {
      "epoch": 4.692936419168889,
      "grad_norm": 15.75,
      "learning_rate": 9.28498823151902e-06,
      "loss": 0.7018,
      "step": 54546
    },
    {
      "epoch": 4.6930224554762106,
      "grad_norm": 3.375,
      "learning_rate": 9.279801981405378e-06,
      "loss": 1.1799,
      "step": 54547
    },
    {
      "epoch": 4.693108491783533,
      "grad_norm": 1.0390625,
      "learning_rate": 9.274617166552923e-06,
      "loss": 0.9265,
      "step": 54548
    },
    {
      "epoch": 4.693194528090855,
      "grad_norm": 5.75,
      "learning_rate": 9.269433786976756e-06,
      "loss": 1.1405,
      "step": 54549
    },
    {
      "epoch": 4.693280564398176,
      "grad_norm": 904.0,
      "learning_rate": 9.264251842692028e-06,
      "loss": 0.99,
      "step": 54550
    },
    {
      "epoch": 4.693366600705498,
      "grad_norm": 1.046875,
      "learning_rate": 9.259071333713954e-06,
      "loss": 1.3789,
      "step": 54551
    },
    {
      "epoch": 4.693452637012819,
      "grad_norm": 1.0703125,
      "learning_rate": 9.253892260057629e-06,
      "loss": 1.1375,
      "step": 54552
    },
    {
      "epoch": 4.693538673320141,
      "grad_norm": 1.328125,
      "learning_rate": 9.248714621738263e-06,
      "loss": 0.6591,
      "step": 54553
    },
    {
      "epoch": 4.693624709627462,
      "grad_norm": 1.1875,
      "learning_rate": 9.243538418770903e-06,
      "loss": 0.7026,
      "step": 54554
    },
    {
      "epoch": 4.693710745934784,
      "grad_norm": 1.203125,
      "learning_rate": 9.238363651170756e-06,
      "loss": 1.6338,
      "step": 54555
    },
    {
      "epoch": 4.6937967822421065,
      "grad_norm": 1.15625,
      "learning_rate": 9.233190318952978e-06,
      "loss": 0.8833,
      "step": 54556
    },
    {
      "epoch": 4.693882818549428,
      "grad_norm": 0.98046875,
      "learning_rate": 9.228018422132667e-06,
      "loss": 1.1551,
      "step": 54557
    },
    {
      "epoch": 4.69396885485675,
      "grad_norm": 1.1875,
      "learning_rate": 9.222847960724978e-06,
      "loss": 1.8295,
      "step": 54558
    },
    {
      "epoch": 4.694054891164071,
      "grad_norm": 256.0,
      "learning_rate": 9.217678934744955e-06,
      "loss": 0.8669,
      "step": 54559
    },
    {
      "epoch": 4.694140927471393,
      "grad_norm": 1.1484375,
      "learning_rate": 9.212511344207753e-06,
      "loss": 1.3528,
      "step": 54560
    },
    {
      "epoch": 4.694226963778714,
      "grad_norm": 0.76171875,
      "learning_rate": 9.20734518912858e-06,
      "loss": 0.724,
      "step": 54561
    },
    {
      "epoch": 4.694313000086036,
      "grad_norm": 2.484375,
      "learning_rate": 9.202180469522426e-06,
      "loss": 1.6287,
      "step": 54562
    },
    {
      "epoch": 4.694399036393358,
      "grad_norm": 0.796875,
      "learning_rate": 9.197017185404443e-06,
      "loss": 0.8814,
      "step": 54563
    },
    {
      "epoch": 4.6944850727006795,
      "grad_norm": 1.3828125,
      "learning_rate": 9.191855336789734e-06,
      "loss": 1.4012,
      "step": 54564
    },
    {
      "epoch": 4.694571109008002,
      "grad_norm": 0.890625,
      "learning_rate": 9.186694923693395e-06,
      "loss": 0.7462,
      "step": 54565
    },
    {
      "epoch": 4.694657145315323,
      "grad_norm": 0.79296875,
      "learning_rate": 9.18153594613047e-06,
      "loss": 0.6138,
      "step": 54566
    },
    {
      "epoch": 4.694743181622645,
      "grad_norm": 5.1875,
      "learning_rate": 9.17637840411617e-06,
      "loss": 1.6075,
      "step": 54567
    },
    {
      "epoch": 4.694829217929966,
      "grad_norm": 0.96484375,
      "learning_rate": 9.171222297665428e-06,
      "loss": 0.5656,
      "step": 54568
    },
    {
      "epoch": 4.694915254237288,
      "grad_norm": 70.0,
      "learning_rate": 9.16606762679345e-06,
      "loss": 1.0957,
      "step": 54569
    },
    {
      "epoch": 4.69500129054461,
      "grad_norm": 6.875,
      "learning_rate": 9.160914391515229e-06,
      "loss": 1.5509,
      "step": 54570
    },
    {
      "epoch": 4.695087326851931,
      "grad_norm": 0.828125,
      "learning_rate": 9.155762591845917e-06,
      "loss": 0.5201,
      "step": 54571
    },
    {
      "epoch": 4.695173363159253,
      "grad_norm": 2.015625,
      "learning_rate": 9.150612227800503e-06,
      "loss": 1.0655,
      "step": 54572
    },
    {
      "epoch": 4.695259399466575,
      "grad_norm": 11.75,
      "learning_rate": 9.145463299394086e-06,
      "loss": 1.0363,
      "step": 54573
    },
    {
      "epoch": 4.695345435773897,
      "grad_norm": 1.3671875,
      "learning_rate": 9.140315806641707e-06,
      "loss": 1.1917,
      "step": 54574
    },
    {
      "epoch": 4.695431472081218,
      "grad_norm": 0.8046875,
      "learning_rate": 9.13516974955847e-06,
      "loss": 0.5879,
      "step": 54575
    },
    {
      "epoch": 4.69551750838854,
      "grad_norm": 1.484375,
      "learning_rate": 9.130025128159359e-06,
      "loss": 1.8813,
      "step": 54576
    },
    {
      "epoch": 4.695603544695862,
      "grad_norm": 1.0078125,
      "learning_rate": 9.124881942459473e-06,
      "loss": 1.0957,
      "step": 54577
    },
    {
      "epoch": 4.695689581003183,
      "grad_norm": 1.21875,
      "learning_rate": 9.119740192473858e-06,
      "loss": 2.0334,
      "step": 54578
    },
    {
      "epoch": 4.695775617310505,
      "grad_norm": 2976.0,
      "learning_rate": 9.1145998782175e-06,
      "loss": 0.6899,
      "step": 54579
    },
    {
      "epoch": 4.695861653617826,
      "grad_norm": 1.1015625,
      "learning_rate": 9.109460999705444e-06,
      "loss": 1.4552,
      "step": 54580
    },
    {
      "epoch": 4.695947689925148,
      "grad_norm": 1.0078125,
      "learning_rate": 9.104323556952843e-06,
      "loss": 0.8018,
      "step": 54581
    },
    {
      "epoch": 4.69603372623247,
      "grad_norm": 3.53125,
      "learning_rate": 9.099187549974519e-06,
      "loss": 0.7626,
      "step": 54582
    },
    {
      "epoch": 4.696119762539792,
      "grad_norm": 1.046875,
      "learning_rate": 9.094052978785627e-06,
      "loss": 0.599,
      "step": 54583
    },
    {
      "epoch": 4.696205798847114,
      "grad_norm": 1.09375,
      "learning_rate": 9.088919843401155e-06,
      "loss": 1.2417,
      "step": 54584
    },
    {
      "epoch": 4.696291835154435,
      "grad_norm": 0.9140625,
      "learning_rate": 9.083788143836147e-06,
      "loss": 0.6632,
      "step": 54585
    },
    {
      "epoch": 4.696377871461757,
      "grad_norm": 1.7109375,
      "learning_rate": 9.078657880105534e-06,
      "loss": 0.9068,
      "step": 54586
    },
    {
      "epoch": 4.696463907769078,
      "grad_norm": 3.25,
      "learning_rate": 9.07352905222436e-06,
      "loss": 1.3396,
      "step": 54587
    },
    {
      "epoch": 4.6965499440764,
      "grad_norm": 0.77734375,
      "learning_rate": 9.06840166020767e-06,
      "loss": 0.7651,
      "step": 54588
    },
    {
      "epoch": 4.696635980383721,
      "grad_norm": 4352.0,
      "learning_rate": 9.063275704070394e-06,
      "loss": 0.9726,
      "step": 54589
    },
    {
      "epoch": 4.6967220166910435,
      "grad_norm": 1.3828125,
      "learning_rate": 9.05815118382758e-06,
      "loss": 0.5303,
      "step": 54590
    },
    {
      "epoch": 4.696808052998366,
      "grad_norm": 1.546875,
      "learning_rate": 9.053028099494209e-06,
      "loss": 1.5388,
      "step": 54591
    },
    {
      "epoch": 4.696894089305687,
      "grad_norm": 1.4375,
      "learning_rate": 9.04790645108522e-06,
      "loss": 1.2731,
      "step": 54592
    },
    {
      "epoch": 4.696980125613009,
      "grad_norm": 3.953125,
      "learning_rate": 9.042786238615598e-06,
      "loss": 0.9841,
      "step": 54593
    },
    {
      "epoch": 4.69706616192033,
      "grad_norm": 0.8828125,
      "learning_rate": 9.037667462100386e-06,
      "loss": 1.0778,
      "step": 54594
    },
    {
      "epoch": 4.697152198227652,
      "grad_norm": 1.4609375,
      "learning_rate": 9.032550121554461e-06,
      "loss": 1.1169,
      "step": 54595
    },
    {
      "epoch": 4.697238234534974,
      "grad_norm": 26.5,
      "learning_rate": 9.027434216992869e-06,
      "loss": 1.324,
      "step": 54596
    },
    {
      "epoch": 4.697324270842295,
      "grad_norm": 1.8203125,
      "learning_rate": 9.022319748430596e-06,
      "loss": 0.9234,
      "step": 54597
    },
    {
      "epoch": 4.697410307149617,
      "grad_norm": 0.87890625,
      "learning_rate": 9.017206715882464e-06,
      "loss": 0.6839,
      "step": 54598
    },
    {
      "epoch": 4.697496343456939,
      "grad_norm": 1.3671875,
      "learning_rate": 9.012095119363572e-06,
      "loss": 0.7049,
      "step": 54599
    },
    {
      "epoch": 4.697582379764261,
      "grad_norm": 17.75,
      "learning_rate": 9.006984958888742e-06,
      "loss": 1.242,
      "step": 54600
    },
    {
      "epoch": 4.697668416071583,
      "grad_norm": 1.1953125,
      "learning_rate": 9.001876234473016e-06,
      "loss": 0.7823,
      "step": 54601
    },
    {
      "epoch": 4.697754452378904,
      "grad_norm": 1.34375,
      "learning_rate": 8.996768946131329e-06,
      "loss": 1.2727,
      "step": 54602
    },
    {
      "epoch": 4.697840488686226,
      "grad_norm": 1.3671875,
      "learning_rate": 8.99166309387861e-06,
      "loss": 1.3698,
      "step": 54603
    },
    {
      "epoch": 4.697926524993547,
      "grad_norm": 1.078125,
      "learning_rate": 8.98655867772974e-06,
      "loss": 1.2857,
      "step": 54604
    },
    {
      "epoch": 4.698012561300869,
      "grad_norm": 3.34375,
      "learning_rate": 8.981455697699703e-06,
      "loss": 0.6476,
      "step": 54605
    },
    {
      "epoch": 4.69809859760819,
      "grad_norm": 1.0,
      "learning_rate": 8.976354153803435e-06,
      "loss": 1.9133,
      "step": 54606
    },
    {
      "epoch": 4.698184633915512,
      "grad_norm": 1.0546875,
      "learning_rate": 8.971254046055754e-06,
      "loss": 1.54,
      "step": 54607
    },
    {
      "epoch": 4.6982706702228345,
      "grad_norm": 4.65625,
      "learning_rate": 8.966155374471763e-06,
      "loss": 1.3176,
      "step": 54608
    },
    {
      "epoch": 4.698356706530156,
      "grad_norm": 1.34375,
      "learning_rate": 8.961058139066169e-06,
      "loss": 0.9482,
      "step": 54609
    },
    {
      "epoch": 4.698442742837478,
      "grad_norm": 1.296875,
      "learning_rate": 8.955962339854018e-06,
      "loss": 1.6647,
      "step": 54610
    },
    {
      "epoch": 4.698528779144799,
      "grad_norm": 1.65625,
      "learning_rate": 8.95086797685024e-06,
      "loss": 1.1397,
      "step": 54611
    },
    {
      "epoch": 4.698614815452121,
      "grad_norm": 1.34375,
      "learning_rate": 8.945775050069605e-06,
      "loss": 1.3975,
      "step": 54612
    },
    {
      "epoch": 4.698700851759442,
      "grad_norm": 1.2109375,
      "learning_rate": 8.940683559527096e-06,
      "loss": 0.9657,
      "step": 54613
    },
    {
      "epoch": 4.698786888066764,
      "grad_norm": 0.73828125,
      "learning_rate": 8.935593505237539e-06,
      "loss": 0.3586,
      "step": 54614
    },
    {
      "epoch": 4.698872924374086,
      "grad_norm": 1.21875,
      "learning_rate": 8.930504887215863e-06,
      "loss": 1.7579,
      "step": 54615
    },
    {
      "epoch": 4.6989589606814075,
      "grad_norm": 1.3828125,
      "learning_rate": 8.925417705477002e-06,
      "loss": 0.7881,
      "step": 54616
    },
    {
      "epoch": 4.69904499698873,
      "grad_norm": 0.9375,
      "learning_rate": 8.920331960035777e-06,
      "loss": 1.2469,
      "step": 54617
    },
    {
      "epoch": 4.699131033296051,
      "grad_norm": 0.8359375,
      "learning_rate": 8.915247650907011e-06,
      "loss": 1.2103,
      "step": 54618
    },
    {
      "epoch": 4.699217069603373,
      "grad_norm": 1.203125,
      "learning_rate": 8.910164778105744e-06,
      "loss": 0.9496,
      "step": 54619
    },
    {
      "epoch": 4.699303105910694,
      "grad_norm": 1.7265625,
      "learning_rate": 8.905083341646635e-06,
      "loss": 1.5974,
      "step": 54620
    },
    {
      "epoch": 4.699389142218016,
      "grad_norm": 1.1953125,
      "learning_rate": 8.900003341544671e-06,
      "loss": 2.1129,
      "step": 54621
    },
    {
      "epoch": 4.699475178525338,
      "grad_norm": 1.25,
      "learning_rate": 8.894924777814729e-06,
      "loss": 1.2749,
      "step": 54622
    },
    {
      "epoch": 4.699561214832659,
      "grad_norm": 1.1015625,
      "learning_rate": 8.889847650471516e-06,
      "loss": 1.3385,
      "step": 54623
    },
    {
      "epoch": 4.699647251139981,
      "grad_norm": 1.171875,
      "learning_rate": 8.88477195953008e-06,
      "loss": 0.9009,
      "step": 54624
    },
    {
      "epoch": 4.699733287447303,
      "grad_norm": 1.25,
      "learning_rate": 8.87969770500513e-06,
      "loss": 1.5523,
      "step": 54625
    },
    {
      "epoch": 4.699819323754625,
      "grad_norm": 1.125,
      "learning_rate": 8.874624886911542e-06,
      "loss": 0.8565,
      "step": 54626
    },
    {
      "epoch": 4.699905360061946,
      "grad_norm": 58.25,
      "learning_rate": 8.869553505264195e-06,
      "loss": 0.9971,
      "step": 54627
    },
    {
      "epoch": 4.699991396369268,
      "grad_norm": 0.98046875,
      "learning_rate": 8.864483560077797e-06,
      "loss": 0.745,
      "step": 54628
    },
    {
      "epoch": 4.70007743267659,
      "grad_norm": 0.828125,
      "learning_rate": 8.859415051367393e-06,
      "loss": 0.5947,
      "step": 54629
    },
    {
      "epoch": 4.700163468983911,
      "grad_norm": 1.109375,
      "learning_rate": 8.85434797914758e-06,
      "loss": 1.0151,
      "step": 54630
    },
    {
      "epoch": 4.700249505291233,
      "grad_norm": 1.109375,
      "learning_rate": 8.849282343433296e-06,
      "loss": 1.0912,
      "step": 54631
    },
    {
      "epoch": 4.700335541598554,
      "grad_norm": 1.0390625,
      "learning_rate": 8.844218144239359e-06,
      "loss": 0.7952,
      "step": 54632
    },
    {
      "epoch": 4.7004215779058764,
      "grad_norm": 1.015625,
      "learning_rate": 8.839155381580533e-06,
      "loss": 0.9448,
      "step": 54633
    },
    {
      "epoch": 4.700507614213198,
      "grad_norm": 1.046875,
      "learning_rate": 8.834094055471697e-06,
      "loss": 0.7696,
      "step": 54634
    },
    {
      "epoch": 4.70059365052052,
      "grad_norm": 1.28125,
      "learning_rate": 8.829034165927564e-06,
      "loss": 0.8746,
      "step": 54635
    },
    {
      "epoch": 4.700679686827842,
      "grad_norm": 1.09375,
      "learning_rate": 8.823975712963006e-06,
      "loss": 0.9172,
      "step": 54636
    },
    {
      "epoch": 4.700765723135163,
      "grad_norm": 1.1171875,
      "learning_rate": 8.818918696592737e-06,
      "loss": 1.3278,
      "step": 54637
    },
    {
      "epoch": 4.700851759442485,
      "grad_norm": 1.3046875,
      "learning_rate": 8.813863116831689e-06,
      "loss": 1.1772,
      "step": 54638
    },
    {
      "epoch": 4.700937795749806,
      "grad_norm": 1.2421875,
      "learning_rate": 8.808808973694515e-06,
      "loss": 1.0669,
      "step": 54639
    },
    {
      "epoch": 4.701023832057128,
      "grad_norm": 0.66796875,
      "learning_rate": 8.80375626719604e-06,
      "loss": 0.5992,
      "step": 54640
    },
    {
      "epoch": 4.701109868364449,
      "grad_norm": 1.046875,
      "learning_rate": 8.798704997351082e-06,
      "loss": 0.6901,
      "step": 54641
    },
    {
      "epoch": 4.7011959046717715,
      "grad_norm": 0.8046875,
      "learning_rate": 8.793655164174296e-06,
      "loss": 0.483,
      "step": 54642
    },
    {
      "epoch": 4.701281940979094,
      "grad_norm": 2.296875,
      "learning_rate": 8.788606767680618e-06,
      "loss": 1.2986,
      "step": 54643
    },
    {
      "epoch": 4.701367977286415,
      "grad_norm": 1.0703125,
      "learning_rate": 8.7835598078847e-06,
      "loss": 1.3771,
      "step": 54644
    },
    {
      "epoch": 4.701454013593737,
      "grad_norm": 0.875,
      "learning_rate": 8.778514284801309e-06,
      "loss": 0.6603,
      "step": 54645
    },
    {
      "epoch": 4.701540049901058,
      "grad_norm": 1.0703125,
      "learning_rate": 8.773470198445265e-06,
      "loss": 0.648,
      "step": 54646
    },
    {
      "epoch": 4.70162608620838,
      "grad_norm": 1.1953125,
      "learning_rate": 8.768427548831281e-06,
      "loss": 1.0869,
      "step": 54647
    },
    {
      "epoch": 4.701712122515701,
      "grad_norm": 1.1328125,
      "learning_rate": 8.763386335974067e-06,
      "loss": 1.0526,
      "step": 54648
    },
    {
      "epoch": 4.701798158823023,
      "grad_norm": 1.15625,
      "learning_rate": 8.758346559888442e-06,
      "loss": 0.9471,
      "step": 54649
    },
    {
      "epoch": 4.701884195130345,
      "grad_norm": 1.0546875,
      "learning_rate": 8.753308220589118e-06,
      "loss": 0.811,
      "step": 54650
    },
    {
      "epoch": 4.701970231437667,
      "grad_norm": 1.1875,
      "learning_rate": 8.74827131809075e-06,
      "loss": 1.1074,
      "step": 54651
    },
    {
      "epoch": 4.702056267744989,
      "grad_norm": 1.265625,
      "learning_rate": 8.743235852408271e-06,
      "loss": 1.0681,
      "step": 54652
    },
    {
      "epoch": 4.70214230405231,
      "grad_norm": 1.171875,
      "learning_rate": 8.738201823556169e-06,
      "loss": 1.0268,
      "step": 54653
    },
    {
      "epoch": 4.702228340359632,
      "grad_norm": 8.125,
      "learning_rate": 8.733169231549377e-06,
      "loss": 0.9465,
      "step": 54654
    },
    {
      "epoch": 4.702314376666953,
      "grad_norm": 0.91015625,
      "learning_rate": 8.728138076402437e-06,
      "loss": 0.9881,
      "step": 54655
    },
    {
      "epoch": 4.702400412974275,
      "grad_norm": 1.1953125,
      "learning_rate": 8.723108358130172e-06,
      "loss": 1.7781,
      "step": 54656
    },
    {
      "epoch": 4.702486449281597,
      "grad_norm": 1.0234375,
      "learning_rate": 8.718080076747293e-06,
      "loss": 1.3934,
      "step": 54657
    },
    {
      "epoch": 4.702572485588918,
      "grad_norm": 0.984375,
      "learning_rate": 8.713053232268453e-06,
      "loss": 1.0553,
      "step": 54658
    },
    {
      "epoch": 4.7026585218962405,
      "grad_norm": 0.96484375,
      "learning_rate": 8.70802782470842e-06,
      "loss": 1.0426,
      "step": 54659
    },
    {
      "epoch": 4.702744558203562,
      "grad_norm": 1.546875,
      "learning_rate": 8.703003854081792e-06,
      "loss": 1.7167,
      "step": 54660
    },
    {
      "epoch": 4.702830594510884,
      "grad_norm": 1.1875,
      "learning_rate": 8.697981320403336e-06,
      "loss": 0.5961,
      "step": 54661
    },
    {
      "epoch": 4.702916630818205,
      "grad_norm": 13.875,
      "learning_rate": 8.692960223687763e-06,
      "loss": 1.0758,
      "step": 54662
    },
    {
      "epoch": 4.703002667125527,
      "grad_norm": 0.921875,
      "learning_rate": 8.687940563949725e-06,
      "loss": 0.8371,
      "step": 54663
    },
    {
      "epoch": 4.703088703432849,
      "grad_norm": 0.83984375,
      "learning_rate": 8.68292234120388e-06,
      "loss": 1.6042,
      "step": 54664
    },
    {
      "epoch": 4.70317473974017,
      "grad_norm": 1.203125,
      "learning_rate": 8.677905555464993e-06,
      "loss": 1.1692,
      "step": 54665
    },
    {
      "epoch": 4.703260776047492,
      "grad_norm": 6.8125,
      "learning_rate": 8.672890206747607e-06,
      "loss": 0.9466,
      "step": 54666
    },
    {
      "epoch": 4.7033468123548134,
      "grad_norm": 0.8828125,
      "learning_rate": 8.66787629506649e-06,
      "loss": 0.7564,
      "step": 54667
    },
    {
      "epoch": 4.7034328486621355,
      "grad_norm": 1.078125,
      "learning_rate": 8.662863820436296e-06,
      "loss": 1.2253,
      "step": 54668
    },
    {
      "epoch": 4.703518884969457,
      "grad_norm": 1.2265625,
      "learning_rate": 8.657852782871622e-06,
      "loss": 0.9829,
      "step": 54669
    },
    {
      "epoch": 4.703604921276779,
      "grad_norm": 12.3125,
      "learning_rate": 8.652843182387238e-06,
      "loss": 1.5397,
      "step": 54670
    },
    {
      "epoch": 4.703690957584101,
      "grad_norm": 0.9609375,
      "learning_rate": 8.647835018997629e-06,
      "loss": 1.8963,
      "step": 54671
    },
    {
      "epoch": 4.703776993891422,
      "grad_norm": 1.2734375,
      "learning_rate": 8.64282829271762e-06,
      "loss": 0.7431,
      "step": 54672
    },
    {
      "epoch": 4.703863030198744,
      "grad_norm": 1.609375,
      "learning_rate": 8.637823003561752e-06,
      "loss": 1.033,
      "step": 54673
    },
    {
      "epoch": 4.703949066506065,
      "grad_norm": 1.1640625,
      "learning_rate": 8.632819151544679e-06,
      "loss": 1.2384,
      "step": 54674
    },
    {
      "epoch": 4.704035102813387,
      "grad_norm": 1.3046875,
      "learning_rate": 8.627816736681005e-06,
      "loss": 0.5275,
      "step": 54675
    },
    {
      "epoch": 4.7041211391207085,
      "grad_norm": 7.40625,
      "learning_rate": 8.622815758985491e-06,
      "loss": 0.9593,
      "step": 54676
    },
    {
      "epoch": 4.704207175428031,
      "grad_norm": 1.078125,
      "learning_rate": 8.617816218472573e-06,
      "loss": 1.1542,
      "step": 54677
    },
    {
      "epoch": 4.704293211735353,
      "grad_norm": 0.92578125,
      "learning_rate": 8.612818115157017e-06,
      "loss": 0.7141,
      "step": 54678
    },
    {
      "epoch": 4.704379248042674,
      "grad_norm": 0.94921875,
      "learning_rate": 8.607821449053421e-06,
      "loss": 0.6198,
      "step": 54679
    },
    {
      "epoch": 4.704465284349996,
      "grad_norm": 0.828125,
      "learning_rate": 8.60282622017633e-06,
      "loss": 0.7049,
      "step": 54680
    },
    {
      "epoch": 4.704551320657317,
      "grad_norm": 0.9765625,
      "learning_rate": 8.597832428540453e-06,
      "loss": 0.4853,
      "step": 54681
    },
    {
      "epoch": 4.704637356964639,
      "grad_norm": 1.0859375,
      "learning_rate": 8.592840074160334e-06,
      "loss": 1.0022,
      "step": 54682
    },
    {
      "epoch": 4.70472339327196,
      "grad_norm": 1.25,
      "learning_rate": 8.587849157050576e-06,
      "loss": 1.3332,
      "step": 54683
    },
    {
      "epoch": 4.704809429579282,
      "grad_norm": 6.21875,
      "learning_rate": 8.582859677225774e-06,
      "loss": 0.5505,
      "step": 54684
    },
    {
      "epoch": 4.7048954658866045,
      "grad_norm": 1.1640625,
      "learning_rate": 8.57787163470053e-06,
      "loss": 1.9613,
      "step": 54685
    },
    {
      "epoch": 4.704981502193926,
      "grad_norm": 0.90625,
      "learning_rate": 8.572885029489441e-06,
      "loss": 0.6462,
      "step": 54686
    },
    {
      "epoch": 4.705067538501248,
      "grad_norm": 1.3359375,
      "learning_rate": 8.567899861607108e-06,
      "loss": 1.7698,
      "step": 54687
    },
    {
      "epoch": 4.705153574808569,
      "grad_norm": 0.9765625,
      "learning_rate": 8.562916131068077e-06,
      "loss": 1.0852,
      "step": 54688
    },
    {
      "epoch": 4.705239611115891,
      "grad_norm": 1.015625,
      "learning_rate": 8.557933837886889e-06,
      "loss": 1.2092,
      "step": 54689
    },
    {
      "epoch": 4.705325647423212,
      "grad_norm": 1.5,
      "learning_rate": 8.552952982078254e-06,
      "loss": 1.3888,
      "step": 54690
    },
    {
      "epoch": 4.705411683730534,
      "grad_norm": 1.09375,
      "learning_rate": 8.547973563656553e-06,
      "loss": 1.0688,
      "step": 54691
    },
    {
      "epoch": 4.705497720037856,
      "grad_norm": 1.140625,
      "learning_rate": 8.542995582636492e-06,
      "loss": 0.6325,
      "step": 54692
    },
    {
      "epoch": 4.7055837563451774,
      "grad_norm": 1.1953125,
      "learning_rate": 8.538019039032619e-06,
      "loss": 0.9276,
      "step": 54693
    },
    {
      "epoch": 4.7056697926524995,
      "grad_norm": 1.1171875,
      "learning_rate": 8.533043932859418e-06,
      "loss": 1.4367,
      "step": 54694
    },
    {
      "epoch": 4.705755828959821,
      "grad_norm": 1.3125,
      "learning_rate": 8.528070264131494e-06,
      "loss": 1.1116,
      "step": 54695
    },
    {
      "epoch": 4.705841865267143,
      "grad_norm": 1.34375,
      "learning_rate": 8.523098032863385e-06,
      "loss": 0.8901,
      "step": 54696
    },
    {
      "epoch": 4.705927901574464,
      "grad_norm": 1.265625,
      "learning_rate": 8.518127239069528e-06,
      "loss": 1.735,
      "step": 54697
    },
    {
      "epoch": 4.706013937881786,
      "grad_norm": 13.0,
      "learning_rate": 8.513157882764689e-06,
      "loss": 0.4453,
      "step": 54698
    },
    {
      "epoch": 4.706099974189108,
      "grad_norm": 0.78515625,
      "learning_rate": 8.508189963963243e-06,
      "loss": 0.4891,
      "step": 54699
    },
    {
      "epoch": 4.706186010496429,
      "grad_norm": 6.4375,
      "learning_rate": 8.503223482679734e-06,
      "loss": 0.8237,
      "step": 54700
    },
    {
      "epoch": 4.706272046803751,
      "grad_norm": 1.1953125,
      "learning_rate": 8.498258438928707e-06,
      "loss": 0.8084,
      "step": 54701
    },
    {
      "epoch": 4.7063580831110725,
      "grad_norm": 1.2578125,
      "learning_rate": 8.493294832724707e-06,
      "loss": 1.2481,
      "step": 54702
    },
    {
      "epoch": 4.706444119418395,
      "grad_norm": 0.92578125,
      "learning_rate": 8.488332664082222e-06,
      "loss": 0.8262,
      "step": 54703
    },
    {
      "epoch": 4.706530155725716,
      "grad_norm": 1.484375,
      "learning_rate": 8.483371933015738e-06,
      "loss": 1.2718,
      "step": 54704
    },
    {
      "epoch": 4.706616192033038,
      "grad_norm": 1.109375,
      "learning_rate": 8.4784126395398e-06,
      "loss": 0.994,
      "step": 54705
    },
    {
      "epoch": 4.70670222834036,
      "grad_norm": 0.9140625,
      "learning_rate": 8.473454783668954e-06,
      "loss": 0.5631,
      "step": 54706
    },
    {
      "epoch": 4.706788264647681,
      "grad_norm": 1.296875,
      "learning_rate": 8.468498365417576e-06,
      "loss": 1.683,
      "step": 54707
    },
    {
      "epoch": 4.706874300955003,
      "grad_norm": 1.5703125,
      "learning_rate": 8.463543384800321e-06,
      "loss": 0.8561,
      "step": 54708
    },
    {
      "epoch": 4.706960337262324,
      "grad_norm": 16.75,
      "learning_rate": 8.458589841831566e-06,
      "loss": 0.6926,
      "step": 54709
    },
    {
      "epoch": 4.707046373569646,
      "grad_norm": 1.0390625,
      "learning_rate": 8.453637736525854e-06,
      "loss": 0.7636,
      "step": 54710
    },
    {
      "epoch": 4.7071324098769685,
      "grad_norm": 390.0,
      "learning_rate": 8.448687068897676e-06,
      "loss": 0.9103,
      "step": 54711
    },
    {
      "epoch": 4.70721844618429,
      "grad_norm": 0.98046875,
      "learning_rate": 8.443737838961407e-06,
      "loss": 1.2061,
      "step": 54712
    },
    {
      "epoch": 4.707304482491612,
      "grad_norm": 1.4453125,
      "learning_rate": 8.438790046731649e-06,
      "loss": 1.153,
      "step": 54713
    },
    {
      "epoch": 4.707390518798933,
      "grad_norm": 1.0703125,
      "learning_rate": 8.43384369222283e-06,
      "loss": 0.6959,
      "step": 54714
    },
    {
      "epoch": 4.707476555106255,
      "grad_norm": 0.9453125,
      "learning_rate": 8.42889877544939e-06,
      "loss": 1.2064,
      "step": 54715
    },
    {
      "epoch": 4.707562591413577,
      "grad_norm": 1.71875,
      "learning_rate": 8.423955296425868e-06,
      "loss": 1.2756,
      "step": 54716
    },
    {
      "epoch": 4.707648627720898,
      "grad_norm": 0.984375,
      "learning_rate": 8.419013255166642e-06,
      "loss": 0.7684,
      "step": 54717
    },
    {
      "epoch": 4.70773466402822,
      "grad_norm": 1.1796875,
      "learning_rate": 8.4140726516862e-06,
      "loss": 1.0629,
      "step": 54718
    },
    {
      "epoch": 4.7078207003355415,
      "grad_norm": 1.59375,
      "learning_rate": 8.40913348599892e-06,
      "loss": 1.1675,
      "step": 54719
    },
    {
      "epoch": 4.7079067366428635,
      "grad_norm": 0.87890625,
      "learning_rate": 8.404195758119404e-06,
      "loss": 0.7728,
      "step": 54720
    },
    {
      "epoch": 4.707992772950185,
      "grad_norm": 4.3125,
      "learning_rate": 8.39925946806197e-06,
      "loss": 1.2571,
      "step": 54721
    },
    {
      "epoch": 4.708078809257507,
      "grad_norm": 2.859375,
      "learning_rate": 8.394324615841109e-06,
      "loss": 0.9811,
      "step": 54722
    },
    {
      "epoch": 4.708164845564829,
      "grad_norm": 1.2265625,
      "learning_rate": 8.389391201471252e-06,
      "loss": 1.2757,
      "step": 54723
    },
    {
      "epoch": 4.70825088187215,
      "grad_norm": 1.5703125,
      "learning_rate": 8.384459224966723e-06,
      "loss": 1.6381,
      "step": 54724
    },
    {
      "epoch": 4.708336918179472,
      "grad_norm": 1.0703125,
      "learning_rate": 8.379528686342174e-06,
      "loss": 1.2463,
      "step": 54725
    },
    {
      "epoch": 4.708422954486793,
      "grad_norm": 1.2265625,
      "learning_rate": 8.374599585611764e-06,
      "loss": 1.9153,
      "step": 54726
    },
    {
      "epoch": 4.708508990794115,
      "grad_norm": 1.328125,
      "learning_rate": 8.36967192279009e-06,
      "loss": 0.955,
      "step": 54727
    },
    {
      "epoch": 4.7085950271014365,
      "grad_norm": 1.0703125,
      "learning_rate": 8.364745697891529e-06,
      "loss": 1.383,
      "step": 54728
    },
    {
      "epoch": 4.708681063408759,
      "grad_norm": 1.4453125,
      "learning_rate": 8.359820910930405e-06,
      "loss": 1.0951,
      "step": 54729
    },
    {
      "epoch": 4.708767099716081,
      "grad_norm": 1.0,
      "learning_rate": 8.354897561921204e-06,
      "loss": 0.6232,
      "step": 54730
    },
    {
      "epoch": 4.708853136023402,
      "grad_norm": 52.75,
      "learning_rate": 8.349975650878361e-06,
      "loss": 1.334,
      "step": 54731
    },
    {
      "epoch": 4.708939172330724,
      "grad_norm": 0.87109375,
      "learning_rate": 8.345055177816196e-06,
      "loss": 0.5078,
      "step": 54732
    },
    {
      "epoch": 4.709025208638045,
      "grad_norm": 0.98828125,
      "learning_rate": 8.340136142749089e-06,
      "loss": 0.521,
      "step": 54733
    },
    {
      "epoch": 4.709111244945367,
      "grad_norm": 0.83984375,
      "learning_rate": 8.33521854569147e-06,
      "loss": 0.5174,
      "step": 54734
    },
    {
      "epoch": 4.709197281252688,
      "grad_norm": 1.171875,
      "learning_rate": 8.330302386657717e-06,
      "loss": 1.125,
      "step": 54735
    },
    {
      "epoch": 4.70928331756001,
      "grad_norm": 1.3984375,
      "learning_rate": 8.32538766566221e-06,
      "loss": 2.0738,
      "step": 54736
    },
    {
      "epoch": 4.7093693538673325,
      "grad_norm": 1.2265625,
      "learning_rate": 8.320474382719323e-06,
      "loss": 1.1167,
      "step": 54737
    },
    {
      "epoch": 4.709455390174654,
      "grad_norm": 1.046875,
      "learning_rate": 8.315562537843436e-06,
      "loss": 0.7886,
      "step": 54738
    },
    {
      "epoch": 4.709541426481976,
      "grad_norm": 1.0234375,
      "learning_rate": 8.310652131048869e-06,
      "loss": 0.668,
      "step": 54739
    },
    {
      "epoch": 4.709627462789297,
      "grad_norm": 0.98046875,
      "learning_rate": 8.305743162350055e-06,
      "loss": 1.3737,
      "step": 54740
    },
    {
      "epoch": 4.709713499096619,
      "grad_norm": 1.046875,
      "learning_rate": 8.300835631761317e-06,
      "loss": 0.8566,
      "step": 54741
    },
    {
      "epoch": 4.70979953540394,
      "grad_norm": 3.421875,
      "learning_rate": 8.295929539296976e-06,
      "loss": 0.6793,
      "step": 54742
    },
    {
      "epoch": 4.709885571711262,
      "grad_norm": 1.421875,
      "learning_rate": 8.29102488497141e-06,
      "loss": 1.4549,
      "step": 54743
    },
    {
      "epoch": 4.709971608018584,
      "grad_norm": 0.87109375,
      "learning_rate": 8.28612166879894e-06,
      "loss": 1.1711,
      "step": 54744
    },
    {
      "epoch": 4.7100576443259055,
      "grad_norm": 1.015625,
      "learning_rate": 8.281219890794001e-06,
      "loss": 1.2212,
      "step": 54745
    },
    {
      "epoch": 4.7101436806332275,
      "grad_norm": 1.6640625,
      "learning_rate": 8.276319550970801e-06,
      "loss": 1.2589,
      "step": 54746
    },
    {
      "epoch": 4.710229716940549,
      "grad_norm": 0.83203125,
      "learning_rate": 8.271420649343775e-06,
      "loss": 0.5733,
      "step": 54747
    },
    {
      "epoch": 4.710315753247871,
      "grad_norm": 1.2421875,
      "learning_rate": 8.266523185927133e-06,
      "loss": 1.9387,
      "step": 54748
    },
    {
      "epoch": 4.710401789555192,
      "grad_norm": 1.4765625,
      "learning_rate": 8.261627160735308e-06,
      "loss": 0.9295,
      "step": 54749
    },
    {
      "epoch": 4.710487825862514,
      "grad_norm": 1.546875,
      "learning_rate": 8.256732573782622e-06,
      "loss": 0.8536,
      "step": 54750
    },
    {
      "epoch": 4.710573862169836,
      "grad_norm": 0.8984375,
      "learning_rate": 8.25183942508334e-06,
      "loss": 0.7103,
      "step": 54751
    },
    {
      "epoch": 4.710659898477157,
      "grad_norm": 1.828125,
      "learning_rate": 8.246947714651787e-06,
      "loss": 0.9086,
      "step": 54752
    },
    {
      "epoch": 4.710745934784479,
      "grad_norm": 1.0078125,
      "learning_rate": 8.242057442502283e-06,
      "loss": 1.2003,
      "step": 54753
    },
    {
      "epoch": 4.7108319710918005,
      "grad_norm": 1.0546875,
      "learning_rate": 8.237168608649093e-06,
      "loss": 1.5198,
      "step": 54754
    },
    {
      "epoch": 4.710918007399123,
      "grad_norm": 1.171875,
      "learning_rate": 8.232281213106541e-06,
      "loss": 0.473,
      "step": 54755
    },
    {
      "epoch": 4.711004043706444,
      "grad_norm": 1.984375,
      "learning_rate": 8.227395255888948e-06,
      "loss": 1.6421,
      "step": 54756
    },
    {
      "epoch": 4.711090080013766,
      "grad_norm": 0.82421875,
      "learning_rate": 8.222510737010525e-06,
      "loss": 0.7807,
      "step": 54757
    },
    {
      "epoch": 4.711176116321088,
      "grad_norm": 0.90234375,
      "learning_rate": 8.217627656485704e-06,
      "loss": 1.6018,
      "step": 54758
    },
    {
      "epoch": 4.711262152628409,
      "grad_norm": 1.0703125,
      "learning_rate": 8.212746014328587e-06,
      "loss": 1.5064,
      "step": 54759
    },
    {
      "epoch": 4.711348188935731,
      "grad_norm": 0.91015625,
      "learning_rate": 8.20786581055355e-06,
      "loss": 0.8679,
      "step": 54760
    },
    {
      "epoch": 4.711434225243052,
      "grad_norm": 1.53125,
      "learning_rate": 8.202987045174915e-06,
      "loss": 1.325,
      "step": 54761
    },
    {
      "epoch": 4.711520261550374,
      "grad_norm": 1.1640625,
      "learning_rate": 8.198109718206891e-06,
      "loss": 0.725,
      "step": 54762
    },
    {
      "epoch": 4.711606297857696,
      "grad_norm": 1.5859375,
      "learning_rate": 8.193233829663693e-06,
      "loss": 1.3473,
      "step": 54763
    },
    {
      "epoch": 4.711692334165018,
      "grad_norm": 6.1875,
      "learning_rate": 8.188359379559696e-06,
      "loss": 1.2177,
      "step": 54764
    },
    {
      "epoch": 4.71177837047234,
      "grad_norm": 1.9765625,
      "learning_rate": 8.183486367908998e-06,
      "loss": 1.0732,
      "step": 54765
    },
    {
      "epoch": 4.711864406779661,
      "grad_norm": 0.9296875,
      "learning_rate": 8.178614794726035e-06,
      "loss": 0.4451,
      "step": 54766
    },
    {
      "epoch": 4.711950443086983,
      "grad_norm": 1.28125,
      "learning_rate": 8.173744660024962e-06,
      "loss": 1.6214,
      "step": 54767
    },
    {
      "epoch": 4.712036479394304,
      "grad_norm": 1.0390625,
      "learning_rate": 8.168875963819989e-06,
      "loss": 0.9365,
      "step": 54768
    },
    {
      "epoch": 4.712122515701626,
      "grad_norm": 0.9453125,
      "learning_rate": 8.164008706125437e-06,
      "loss": 0.9414,
      "step": 54769
    },
    {
      "epoch": 4.712208552008947,
      "grad_norm": 1.2578125,
      "learning_rate": 8.159142886955517e-06,
      "loss": 0.8197,
      "step": 54770
    },
    {
      "epoch": 4.7122945883162695,
      "grad_norm": 1.2109375,
      "learning_rate": 8.154278506324387e-06,
      "loss": 1.5867,
      "step": 54771
    },
    {
      "epoch": 4.7123806246235915,
      "grad_norm": 1.046875,
      "learning_rate": 8.149415564246366e-06,
      "loss": 0.7589,
      "step": 54772
    },
    {
      "epoch": 4.712466660930913,
      "grad_norm": 1.8046875,
      "learning_rate": 8.144554060735609e-06,
      "loss": 1.2834,
      "step": 54773
    },
    {
      "epoch": 4.712552697238235,
      "grad_norm": 1.140625,
      "learning_rate": 8.139693995806441e-06,
      "loss": 1.0061,
      "step": 54774
    },
    {
      "epoch": 4.712638733545556,
      "grad_norm": 1.15625,
      "learning_rate": 8.134835369472959e-06,
      "loss": 0.9191,
      "step": 54775
    },
    {
      "epoch": 4.712724769852878,
      "grad_norm": 1.453125,
      "learning_rate": 8.129978181749431e-06,
      "loss": 1.8661,
      "step": 54776
    },
    {
      "epoch": 4.712810806160199,
      "grad_norm": 0.9296875,
      "learning_rate": 8.125122432650067e-06,
      "loss": 0.9164,
      "step": 54777
    },
    {
      "epoch": 4.712896842467521,
      "grad_norm": 1.2578125,
      "learning_rate": 8.120268122189022e-06,
      "loss": 1.1424,
      "step": 54778
    },
    {
      "epoch": 4.712982878774843,
      "grad_norm": 0.72265625,
      "learning_rate": 8.115415250380564e-06,
      "loss": 0.3855,
      "step": 54779
    },
    {
      "epoch": 4.7130689150821645,
      "grad_norm": 1.3828125,
      "learning_rate": 8.110563817238847e-06,
      "loss": 1.1615,
      "step": 54780
    },
    {
      "epoch": 4.713154951389487,
      "grad_norm": 0.875,
      "learning_rate": 8.105713822778027e-06,
      "loss": 0.8348,
      "step": 54781
    },
    {
      "epoch": 4.713240987696808,
      "grad_norm": 0.91015625,
      "learning_rate": 8.10086526701237e-06,
      "loss": 0.8728,
      "step": 54782
    },
    {
      "epoch": 4.71332702400413,
      "grad_norm": 1.265625,
      "learning_rate": 8.096018149955975e-06,
      "loss": 1.6054,
      "step": 54783
    },
    {
      "epoch": 4.713413060311451,
      "grad_norm": 1.2734375,
      "learning_rate": 8.091172471623054e-06,
      "loss": 1.4253,
      "step": 54784
    },
    {
      "epoch": 4.713499096618773,
      "grad_norm": 1.1171875,
      "learning_rate": 8.086328232027873e-06,
      "loss": 1.0564,
      "step": 54785
    },
    {
      "epoch": 4.713585132926095,
      "grad_norm": 1.015625,
      "learning_rate": 8.081485431184422e-06,
      "loss": 0.6274,
      "step": 54786
    },
    {
      "epoch": 4.713671169233416,
      "grad_norm": 0.9453125,
      "learning_rate": 8.076644069106964e-06,
      "loss": 0.7236,
      "step": 54787
    },
    {
      "epoch": 4.713757205540738,
      "grad_norm": 0.99609375,
      "learning_rate": 8.071804145809658e-06,
      "loss": 0.4855,
      "step": 54788
    },
    {
      "epoch": 4.71384324184806,
      "grad_norm": 169.0,
      "learning_rate": 8.066965661306657e-06,
      "loss": 1.2225,
      "step": 54789
    },
    {
      "epoch": 4.713929278155382,
      "grad_norm": 1.25,
      "learning_rate": 8.062128615612064e-06,
      "loss": 0.7334,
      "step": 54790
    },
    {
      "epoch": 4.714015314462703,
      "grad_norm": 0.91796875,
      "learning_rate": 8.05729300874014e-06,
      "loss": 0.7683,
      "step": 54791
    },
    {
      "epoch": 4.714101350770025,
      "grad_norm": 0.8671875,
      "learning_rate": 8.052458840704879e-06,
      "loss": 0.6169,
      "step": 54792
    },
    {
      "epoch": 4.714187387077347,
      "grad_norm": 1.40625,
      "learning_rate": 8.047626111520544e-06,
      "loss": 1.9291,
      "step": 54793
    },
    {
      "epoch": 4.714273423384668,
      "grad_norm": 1.203125,
      "learning_rate": 8.042794821201238e-06,
      "loss": 1.3211,
      "step": 54794
    },
    {
      "epoch": 4.71435945969199,
      "grad_norm": 0.98046875,
      "learning_rate": 8.037964969761059e-06,
      "loss": 0.7773,
      "step": 54795
    },
    {
      "epoch": 4.714445495999311,
      "grad_norm": 1.4375,
      "learning_rate": 8.03313655721416e-06,
      "loss": 1.255,
      "step": 54796
    },
    {
      "epoch": 4.7145315323066335,
      "grad_norm": 1.3125,
      "learning_rate": 8.028309583574644e-06,
      "loss": 0.7852,
      "step": 54797
    },
    {
      "epoch": 4.714617568613955,
      "grad_norm": 1.0078125,
      "learning_rate": 8.023484048856611e-06,
      "loss": 1.0072,
      "step": 54798
    },
    {
      "epoch": 4.714703604921277,
      "grad_norm": 1.1875,
      "learning_rate": 8.01865995307427e-06,
      "loss": 1.6366,
      "step": 54799
    },
    {
      "epoch": 4.714789641228599,
      "grad_norm": 1.3125,
      "learning_rate": 8.01383729624161e-06,
      "loss": 0.8833,
      "step": 54800
    },
    {
      "epoch": 4.71487567753592,
      "grad_norm": 1.3671875,
      "learning_rate": 8.009016078372788e-06,
      "loss": 0.9404,
      "step": 54801
    },
    {
      "epoch": 4.714961713843242,
      "grad_norm": 0.73046875,
      "learning_rate": 8.004196299481903e-06,
      "loss": 0.9892,
      "step": 54802
    },
    {
      "epoch": 4.715047750150563,
      "grad_norm": 2.03125,
      "learning_rate": 7.99937795958311e-06,
      "loss": 0.8371,
      "step": 54803
    },
    {
      "epoch": 4.715133786457885,
      "grad_norm": 1.3828125,
      "learning_rate": 7.994561058690398e-06,
      "loss": 1.3078,
      "step": 54804
    },
    {
      "epoch": 4.7152198227652065,
      "grad_norm": 13.0,
      "learning_rate": 7.989745596817921e-06,
      "loss": 1.7278,
      "step": 54805
    },
    {
      "epoch": 4.7153058590725285,
      "grad_norm": 1.0703125,
      "learning_rate": 7.984931573979725e-06,
      "loss": 0.7962,
      "step": 54806
    },
    {
      "epoch": 4.715391895379851,
      "grad_norm": 1.0859375,
      "learning_rate": 7.980118990189966e-06,
      "loss": 0.8748,
      "step": 54807
    },
    {
      "epoch": 4.715477931687172,
      "grad_norm": 20.875,
      "learning_rate": 7.975307845462576e-06,
      "loss": 0.8273,
      "step": 54808
    },
    {
      "epoch": 4.715563967994494,
      "grad_norm": 11.4375,
      "learning_rate": 7.970498139811765e-06,
      "loss": 0.7057,
      "step": 54809
    },
    {
      "epoch": 4.715650004301815,
      "grad_norm": 1.1875,
      "learning_rate": 7.96568987325158e-06,
      "loss": 1.5285,
      "step": 54810
    },
    {
      "epoch": 4.715736040609137,
      "grad_norm": 1.2421875,
      "learning_rate": 7.960883045796063e-06,
      "loss": 0.9289,
      "step": 54811
    },
    {
      "epoch": 4.715822076916458,
      "grad_norm": 1.5859375,
      "learning_rate": 7.95607765745926e-06,
      "loss": 1.0168,
      "step": 54812
    },
    {
      "epoch": 4.71590811322378,
      "grad_norm": 0.87890625,
      "learning_rate": 7.951273708255158e-06,
      "loss": 0.7094,
      "step": 54813
    },
    {
      "epoch": 4.715994149531102,
      "grad_norm": 2.9375,
      "learning_rate": 7.94647119819797e-06,
      "loss": 1.3014,
      "step": 54814
    },
    {
      "epoch": 4.716080185838424,
      "grad_norm": 1.0703125,
      "learning_rate": 7.941670127301625e-06,
      "loss": 1.4493,
      "step": 54815
    },
    {
      "epoch": 4.716166222145746,
      "grad_norm": 1.296875,
      "learning_rate": 7.936870495580173e-06,
      "loss": 1.4389,
      "step": 54816
    },
    {
      "epoch": 4.716252258453067,
      "grad_norm": 2.609375,
      "learning_rate": 7.93207230304771e-06,
      "loss": 0.4926,
      "step": 54817
    },
    {
      "epoch": 4.716338294760389,
      "grad_norm": 2.46875,
      "learning_rate": 7.927275549718228e-06,
      "loss": 0.8189,
      "step": 54818
    },
    {
      "epoch": 4.71642433106771,
      "grad_norm": 1.109375,
      "learning_rate": 7.922480235605766e-06,
      "loss": 1.4561,
      "step": 54819
    },
    {
      "epoch": 4.716510367375032,
      "grad_norm": 0.87890625,
      "learning_rate": 7.917686360724319e-06,
      "loss": 0.5792,
      "step": 54820
    },
    {
      "epoch": 4.716596403682354,
      "grad_norm": 15.125,
      "learning_rate": 7.912893925087927e-06,
      "loss": 0.6647,
      "step": 54821
    },
    {
      "epoch": 4.716682439989675,
      "grad_norm": 1.0859375,
      "learning_rate": 7.908102928710637e-06,
      "loss": 1.1023,
      "step": 54822
    },
    {
      "epoch": 4.7167684762969975,
      "grad_norm": 1.3671875,
      "learning_rate": 7.903313371606491e-06,
      "loss": 1.2967,
      "step": 54823
    },
    {
      "epoch": 4.716854512604319,
      "grad_norm": 2.703125,
      "learning_rate": 7.898525253789368e-06,
      "loss": 1.275,
      "step": 54824
    },
    {
      "epoch": 4.716940548911641,
      "grad_norm": 1.3203125,
      "learning_rate": 7.893738575273367e-06,
      "loss": 1.3359,
      "step": 54825
    },
    {
      "epoch": 4.717026585218963,
      "grad_norm": 4.96875,
      "learning_rate": 7.88895333607248e-06,
      "loss": 1.0702,
      "step": 54826
    },
    {
      "epoch": 4.717112621526284,
      "grad_norm": 1.4609375,
      "learning_rate": 7.884169536200692e-06,
      "loss": 1.5899,
      "step": 54827
    },
    {
      "epoch": 4.717198657833606,
      "grad_norm": 0.86328125,
      "learning_rate": 7.879387175671993e-06,
      "loss": 0.8476,
      "step": 54828
    },
    {
      "epoch": 4.717284694140927,
      "grad_norm": 1.6171875,
      "learning_rate": 7.874606254500372e-06,
      "loss": 1.6537,
      "step": 54829
    },
    {
      "epoch": 4.717370730448249,
      "grad_norm": 1.234375,
      "learning_rate": 7.869826772699818e-06,
      "loss": 0.8292,
      "step": 54830
    },
    {
      "epoch": 4.7174567667555705,
      "grad_norm": 1.3671875,
      "learning_rate": 7.865048730284319e-06,
      "loss": 0.7147,
      "step": 54831
    },
    {
      "epoch": 4.7175428030628925,
      "grad_norm": 3.5625,
      "learning_rate": 7.860272127267864e-06,
      "loss": 1.5445,
      "step": 54832
    },
    {
      "epoch": 4.717628839370215,
      "grad_norm": 0.90625,
      "learning_rate": 7.855496963664333e-06,
      "loss": 0.8193,
      "step": 54833
    },
    {
      "epoch": 4.717714875677536,
      "grad_norm": 1.125,
      "learning_rate": 7.850723239487767e-06,
      "loss": 1.1386,
      "step": 54834
    },
    {
      "epoch": 4.717800911984858,
      "grad_norm": 0.984375,
      "learning_rate": 7.845950954752157e-06,
      "loss": 0.6973,
      "step": 54835
    },
    {
      "epoch": 4.717886948292179,
      "grad_norm": 0.953125,
      "learning_rate": 7.841180109471379e-06,
      "loss": 0.7467,
      "step": 54836
    },
    {
      "epoch": 4.717972984599501,
      "grad_norm": 1.0625,
      "learning_rate": 7.83641070365948e-06,
      "loss": 1.5888,
      "step": 54837
    },
    {
      "epoch": 4.718059020906823,
      "grad_norm": 1.2734375,
      "learning_rate": 7.831642737330335e-06,
      "loss": 1.0608,
      "step": 54838
    },
    {
      "epoch": 4.718145057214144,
      "grad_norm": 0.9453125,
      "learning_rate": 7.82687621049788e-06,
      "loss": 1.0625,
      "step": 54839
    },
    {
      "epoch": 4.718231093521466,
      "grad_norm": 1.0703125,
      "learning_rate": 7.822111123176157e-06,
      "loss": 0.7665,
      "step": 54840
    },
    {
      "epoch": 4.718317129828788,
      "grad_norm": 0.7421875,
      "learning_rate": 7.817347475378989e-06,
      "loss": 0.4896,
      "step": 54841
    },
    {
      "epoch": 4.71840316613611,
      "grad_norm": 2.890625,
      "learning_rate": 7.812585267120364e-06,
      "loss": 0.9858,
      "step": 54842
    },
    {
      "epoch": 4.718489202443431,
      "grad_norm": 1.6015625,
      "learning_rate": 7.807824498414217e-06,
      "loss": 1.5713,
      "step": 54843
    },
    {
      "epoch": 4.718575238750753,
      "grad_norm": 0.8359375,
      "learning_rate": 7.803065169274426e-06,
      "loss": 0.7926,
      "step": 54844
    },
    {
      "epoch": 4.718661275058075,
      "grad_norm": 0.9140625,
      "learning_rate": 7.798307279714923e-06,
      "loss": 0.7738,
      "step": 54845
    },
    {
      "epoch": 4.718747311365396,
      "grad_norm": 1.015625,
      "learning_rate": 7.793550829749697e-06,
      "loss": 1.0493,
      "step": 54846
    },
    {
      "epoch": 4.718833347672718,
      "grad_norm": 0.98828125,
      "learning_rate": 7.788795819392568e-06,
      "loss": 0.8721,
      "step": 54847
    },
    {
      "epoch": 4.718919383980039,
      "grad_norm": 1.5546875,
      "learning_rate": 7.78404224865753e-06,
      "loss": 1.0879,
      "step": 54848
    },
    {
      "epoch": 4.7190054202873615,
      "grad_norm": 4.1875,
      "learning_rate": 7.779290117558401e-06,
      "loss": 1.2227,
      "step": 54849
    },
    {
      "epoch": 4.719091456594683,
      "grad_norm": 1.234375,
      "learning_rate": 7.774539426109118e-06,
      "loss": 1.1018,
      "step": 54850
    },
    {
      "epoch": 4.719177492902005,
      "grad_norm": 1.3515625,
      "learning_rate": 7.76979017432361e-06,
      "loss": 1.8554,
      "step": 54851
    },
    {
      "epoch": 4.719263529209327,
      "grad_norm": 0.953125,
      "learning_rate": 7.765042362215646e-06,
      "loss": 1.1183,
      "step": 54852
    },
    {
      "epoch": 4.719349565516648,
      "grad_norm": 1.21875,
      "learning_rate": 7.760295989799326e-06,
      "loss": 1.6302,
      "step": 54853
    },
    {
      "epoch": 4.71943560182397,
      "grad_norm": 1.2734375,
      "learning_rate": 7.755551057088306e-06,
      "loss": 1.4816,
      "step": 54854
    },
    {
      "epoch": 4.719521638131291,
      "grad_norm": 1.46875,
      "learning_rate": 7.750807564096573e-06,
      "loss": 1.3597,
      "step": 54855
    },
    {
      "epoch": 4.719607674438613,
      "grad_norm": 0.87890625,
      "learning_rate": 7.746065510838064e-06,
      "loss": 1.2203,
      "step": 54856
    },
    {
      "epoch": 4.7196937107459345,
      "grad_norm": 0.79296875,
      "learning_rate": 7.741324897326485e-06,
      "loss": 0.7007,
      "step": 54857
    },
    {
      "epoch": 4.7197797470532565,
      "grad_norm": 0.91796875,
      "learning_rate": 7.736585723575829e-06,
      "loss": 1.277,
      "step": 54858
    },
    {
      "epoch": 4.719865783360579,
      "grad_norm": 11.6875,
      "learning_rate": 7.731847989599917e-06,
      "loss": 0.9499,
      "step": 54859
    },
    {
      "epoch": 4.7199518196679,
      "grad_norm": 1.25,
      "learning_rate": 7.727111695412626e-06,
      "loss": 1.2671,
      "step": 54860
    },
    {
      "epoch": 4.720037855975222,
      "grad_norm": 1.078125,
      "learning_rate": 7.722376841027779e-06,
      "loss": 0.7931,
      "step": 54861
    },
    {
      "epoch": 4.720123892282543,
      "grad_norm": 1.1328125,
      "learning_rate": 7.717643426459254e-06,
      "loss": 1.0214,
      "step": 54862
    },
    {
      "epoch": 4.720209928589865,
      "grad_norm": 1.2578125,
      "learning_rate": 7.712911451720816e-06,
      "loss": 0.8891,
      "step": 54863
    },
    {
      "epoch": 4.720295964897186,
      "grad_norm": 1.140625,
      "learning_rate": 7.708180916826457e-06,
      "loss": 1.2049,
      "step": 54864
    },
    {
      "epoch": 4.720382001204508,
      "grad_norm": 1.2578125,
      "learning_rate": 7.703451821789832e-06,
      "loss": 1.4144,
      "step": 54865
    },
    {
      "epoch": 4.72046803751183,
      "grad_norm": 1.140625,
      "learning_rate": 7.698724166624927e-06,
      "loss": 1.0415,
      "step": 54866
    },
    {
      "epoch": 4.720554073819152,
      "grad_norm": 1.2890625,
      "learning_rate": 7.69399795134551e-06,
      "loss": 1.6441,
      "step": 54867
    },
    {
      "epoch": 4.720640110126474,
      "grad_norm": 8.125,
      "learning_rate": 7.68927317596535e-06,
      "loss": 0.8464,
      "step": 54868
    },
    {
      "epoch": 4.720726146433795,
      "grad_norm": 12.25,
      "learning_rate": 7.684549840498323e-06,
      "loss": 1.0842,
      "step": 54869
    },
    {
      "epoch": 4.720812182741117,
      "grad_norm": 0.921875,
      "learning_rate": 7.679827944958307e-06,
      "loss": 1.0225,
      "step": 54870
    },
    {
      "epoch": 4.720898219048438,
      "grad_norm": 1.2578125,
      "learning_rate": 7.675107489358956e-06,
      "loss": 1.0605,
      "step": 54871
    },
    {
      "epoch": 4.72098425535576,
      "grad_norm": 0.76171875,
      "learning_rate": 7.670388473714207e-06,
      "loss": 0.6472,
      "step": 54872
    },
    {
      "epoch": 4.721070291663082,
      "grad_norm": 1.4375,
      "learning_rate": 7.665670898037824e-06,
      "loss": 0.716,
      "step": 54873
    },
    {
      "epoch": 4.721156327970403,
      "grad_norm": 1.1484375,
      "learning_rate": 7.660954762343631e-06,
      "loss": 0.7295,
      "step": 54874
    },
    {
      "epoch": 4.7212423642777255,
      "grad_norm": 113.0,
      "learning_rate": 7.656240066645337e-06,
      "loss": 0.7881,
      "step": 54875
    },
    {
      "epoch": 4.721328400585047,
      "grad_norm": 1.53125,
      "learning_rate": 7.651526810956822e-06,
      "loss": 1.4405,
      "step": 54876
    },
    {
      "epoch": 4.721414436892369,
      "grad_norm": 1.4453125,
      "learning_rate": 7.646814995291795e-06,
      "loss": 1.0847,
      "step": 54877
    },
    {
      "epoch": 4.72150047319969,
      "grad_norm": 1.0625,
      "learning_rate": 7.642104619664081e-06,
      "loss": 0.6545,
      "step": 54878
    },
    {
      "epoch": 4.721586509507012,
      "grad_norm": 1.0546875,
      "learning_rate": 7.6373956840875e-06,
      "loss": 0.9511,
      "step": 54879
    },
    {
      "epoch": 4.721672545814334,
      "grad_norm": 0.9609375,
      "learning_rate": 7.632688188575709e-06,
      "loss": 1.5036,
      "step": 54880
    },
    {
      "epoch": 4.721758582121655,
      "grad_norm": 1.125,
      "learning_rate": 7.627982133142586e-06,
      "loss": 1.1456,
      "step": 54881
    },
    {
      "epoch": 4.721844618428977,
      "grad_norm": 2.65625,
      "learning_rate": 7.623277517801785e-06,
      "loss": 1.4411,
      "step": 54882
    },
    {
      "epoch": 4.7219306547362985,
      "grad_norm": 44.25,
      "learning_rate": 7.6185743425672415e-06,
      "loss": 0.9106,
      "step": 54883
    },
    {
      "epoch": 4.7220166910436205,
      "grad_norm": 1.015625,
      "learning_rate": 7.613872607452555e-06,
      "loss": 0.5505,
      "step": 54884
    },
    {
      "epoch": 4.722102727350942,
      "grad_norm": 0.953125,
      "learning_rate": 7.609172312471546e-06,
      "loss": 0.591,
      "step": 54885
    },
    {
      "epoch": 4.722188763658264,
      "grad_norm": 1.0390625,
      "learning_rate": 7.604473457637928e-06,
      "loss": 1.1585,
      "step": 54886
    },
    {
      "epoch": 4.722274799965586,
      "grad_norm": 19.875,
      "learning_rate": 7.599776042965467e-06,
      "loss": 1.0695,
      "step": 54887
    },
    {
      "epoch": 4.722360836272907,
      "grad_norm": 0.75,
      "learning_rate": 7.595080068467874e-06,
      "loss": 0.7463,
      "step": 54888
    },
    {
      "epoch": 4.722446872580229,
      "grad_norm": 9.3125,
      "learning_rate": 7.5903855341589165e-06,
      "loss": 0.9275,
      "step": 54889
    },
    {
      "epoch": 4.72253290888755,
      "grad_norm": 1.03125,
      "learning_rate": 7.585692440052305e-06,
      "loss": 0.8768,
      "step": 54890
    },
    {
      "epoch": 4.722618945194872,
      "grad_norm": 1.359375,
      "learning_rate": 7.581000786161807e-06,
      "loss": 1.4807,
      "step": 54891
    },
    {
      "epoch": 4.7227049815021935,
      "grad_norm": 1.125,
      "learning_rate": 7.576310572501077e-06,
      "loss": 0.9936,
      "step": 54892
    },
    {
      "epoch": 4.722791017809516,
      "grad_norm": 1.09375,
      "learning_rate": 7.571621799083883e-06,
      "loss": 0.7948,
      "step": 54893
    },
    {
      "epoch": 4.722877054116838,
      "grad_norm": 1.0078125,
      "learning_rate": 7.566934465923936e-06,
      "loss": 0.8226,
      "step": 54894
    },
    {
      "epoch": 4.722963090424159,
      "grad_norm": 0.95703125,
      "learning_rate": 7.562248573034947e-06,
      "loss": 0.6053,
      "step": 54895
    },
    {
      "epoch": 4.723049126731481,
      "grad_norm": 1.109375,
      "learning_rate": 7.557564120430571e-06,
      "loss": 1.7156,
      "step": 54896
    },
    {
      "epoch": 4.723135163038802,
      "grad_norm": 0.98828125,
      "learning_rate": 7.552881108124576e-06,
      "loss": 0.9452,
      "step": 54897
    },
    {
      "epoch": 4.723221199346124,
      "grad_norm": 1.03125,
      "learning_rate": 7.5481995361305625e-06,
      "loss": 1.3872,
      "step": 54898
    },
    {
      "epoch": 4.723307235653445,
      "grad_norm": 1.296875,
      "learning_rate": 7.543519404462296e-06,
      "loss": 1.3533,
      "step": 54899
    },
    {
      "epoch": 4.723393271960767,
      "grad_norm": 1.015625,
      "learning_rate": 7.5388407131335436e-06,
      "loss": 1.4058,
      "step": 54900
    },
    {
      "epoch": 4.7234793082680895,
      "grad_norm": 8.3125,
      "learning_rate": 7.53416346215785e-06,
      "loss": 1.1059,
      "step": 54901
    },
    {
      "epoch": 4.723565344575411,
      "grad_norm": 1.234375,
      "learning_rate": 7.529487651548927e-06,
      "loss": 1.0187,
      "step": 54902
    },
    {
      "epoch": 4.723651380882733,
      "grad_norm": 1.21875,
      "learning_rate": 7.5248132813205416e-06,
      "loss": 0.8112,
      "step": 54903
    },
    {
      "epoch": 4.723737417190054,
      "grad_norm": 1.1328125,
      "learning_rate": 7.520140351486237e-06,
      "loss": 1.3135,
      "step": 54904
    },
    {
      "epoch": 4.723823453497376,
      "grad_norm": 1.296875,
      "learning_rate": 7.515468862059782e-06,
      "loss": 0.7621,
      "step": 54905
    },
    {
      "epoch": 4.723909489804697,
      "grad_norm": 0.87890625,
      "learning_rate": 7.51079881305472e-06,
      "loss": 0.6092,
      "step": 54906
    },
    {
      "epoch": 4.723995526112019,
      "grad_norm": 1.3046875,
      "learning_rate": 7.506130204484818e-06,
      "loss": 0.9852,
      "step": 54907
    },
    {
      "epoch": 4.724081562419341,
      "grad_norm": 0.9765625,
      "learning_rate": 7.501463036363787e-06,
      "loss": 0.7827,
      "step": 54908
    },
    {
      "epoch": 4.7241675987266625,
      "grad_norm": 0.87109375,
      "learning_rate": 7.496797308705117e-06,
      "loss": 0.7822,
      "step": 54909
    },
    {
      "epoch": 4.7242536350339845,
      "grad_norm": 1.1640625,
      "learning_rate": 7.492133021522518e-06,
      "loss": 1.1292,
      "step": 54910
    },
    {
      "epoch": 4.724339671341306,
      "grad_norm": 1.21875,
      "learning_rate": 7.487470174829647e-06,
      "loss": 1.094,
      "step": 54911
    },
    {
      "epoch": 4.724425707648628,
      "grad_norm": 1.0625,
      "learning_rate": 7.482808768640159e-06,
      "loss": 0.3784,
      "step": 54912
    },
    {
      "epoch": 4.724511743955949,
      "grad_norm": 1.46875,
      "learning_rate": 7.478148802967655e-06,
      "loss": 1.4669,
      "step": 54913
    },
    {
      "epoch": 4.724597780263271,
      "grad_norm": 1.109375,
      "learning_rate": 7.473490277825789e-06,
      "loss": 1.1209,
      "step": 54914
    },
    {
      "epoch": 4.724683816570593,
      "grad_norm": 0.98828125,
      "learning_rate": 7.4688331932281636e-06,
      "loss": 1.28,
      "step": 54915
    },
    {
      "epoch": 4.724769852877914,
      "grad_norm": 1.6328125,
      "learning_rate": 7.464177549188378e-06,
      "loss": 0.9884,
      "step": 54916
    },
    {
      "epoch": 4.724855889185236,
      "grad_norm": 40.5,
      "learning_rate": 7.459523345720143e-06,
      "loss": 0.9725,
      "step": 54917
    },
    {
      "epoch": 4.7249419254925575,
      "grad_norm": 1.03125,
      "learning_rate": 7.454870582836948e-06,
      "loss": 0.9282,
      "step": 54918
    },
    {
      "epoch": 4.72502796179988,
      "grad_norm": 1.0,
      "learning_rate": 7.450219260552505e-06,
      "loss": 1.6403,
      "step": 54919
    },
    {
      "epoch": 4.725113998107201,
      "grad_norm": 1.7890625,
      "learning_rate": 7.445569378880301e-06,
      "loss": 1.0172,
      "step": 54920
    },
    {
      "epoch": 4.725200034414523,
      "grad_norm": 2.125,
      "learning_rate": 7.440920937834051e-06,
      "loss": 1.067,
      "step": 54921
    },
    {
      "epoch": 4.725286070721845,
      "grad_norm": 1.15625,
      "learning_rate": 7.436273937427351e-06,
      "loss": 1.7996,
      "step": 54922
    },
    {
      "epoch": 4.725372107029166,
      "grad_norm": 1.0625,
      "learning_rate": 7.4316283776737495e-06,
      "loss": 1.4594,
      "step": 54923
    },
    {
      "epoch": 4.725458143336488,
      "grad_norm": 1.1015625,
      "learning_rate": 7.426984258586844e-06,
      "loss": 1.2116,
      "step": 54924
    },
    {
      "epoch": 4.725544179643809,
      "grad_norm": 1.7109375,
      "learning_rate": 7.4223415801801255e-06,
      "loss": 1.121,
      "step": 54925
    },
    {
      "epoch": 4.725630215951131,
      "grad_norm": 1.046875,
      "learning_rate": 7.417700342467304e-06,
      "loss": 1.3847,
      "step": 54926
    },
    {
      "epoch": 4.725716252258453,
      "grad_norm": 1.0390625,
      "learning_rate": 7.413060545461981e-06,
      "loss": 1.6719,
      "step": 54927
    },
    {
      "epoch": 4.725802288565775,
      "grad_norm": 1.546875,
      "learning_rate": 7.408422189177533e-06,
      "loss": 0.8419,
      "step": 54928
    },
    {
      "epoch": 4.725888324873097,
      "grad_norm": 1.4765625,
      "learning_rate": 7.403785273627672e-06,
      "loss": 1.1337,
      "step": 54929
    },
    {
      "epoch": 4.725974361180418,
      "grad_norm": 8.0,
      "learning_rate": 7.399149798825999e-06,
      "loss": 0.9588,
      "step": 54930
    },
    {
      "epoch": 4.72606039748774,
      "grad_norm": 1.265625,
      "learning_rate": 7.394515764785948e-06,
      "loss": 1.0425,
      "step": 54931
    },
    {
      "epoch": 4.726146433795061,
      "grad_norm": 1.0078125,
      "learning_rate": 7.389883171521172e-06,
      "loss": 0.683,
      "step": 54932
    },
    {
      "epoch": 4.726232470102383,
      "grad_norm": 0.96875,
      "learning_rate": 7.385252019045163e-06,
      "loss": 1.4233,
      "step": 54933
    },
    {
      "epoch": 4.726318506409704,
      "grad_norm": 1.140625,
      "learning_rate": 7.380622307371465e-06,
      "loss": 0.9277,
      "step": 54934
    },
    {
      "epoch": 4.7264045427170265,
      "grad_norm": 2.984375,
      "learning_rate": 7.375994036513678e-06,
      "loss": 1.159,
      "step": 54935
    },
    {
      "epoch": 4.7264905790243485,
      "grad_norm": 1.0390625,
      "learning_rate": 7.371367206485291e-06,
      "loss": 1.2862,
      "step": 54936
    },
    {
      "epoch": 4.72657661533167,
      "grad_norm": 1.0390625,
      "learning_rate": 7.3667418172997934e-06,
      "loss": 1.2341,
      "step": 54937
    },
    {
      "epoch": 4.726662651638992,
      "grad_norm": 1.015625,
      "learning_rate": 7.3621178689708414e-06,
      "loss": 0.9081,
      "step": 54938
    },
    {
      "epoch": 4.726748687946313,
      "grad_norm": 18.625,
      "learning_rate": 7.357495361511812e-06,
      "loss": 1.5224,
      "step": 54939
    },
    {
      "epoch": 4.726834724253635,
      "grad_norm": 1.3828125,
      "learning_rate": 7.352874294936307e-06,
      "loss": 1.6472,
      "step": 54940
    },
    {
      "epoch": 4.726920760560956,
      "grad_norm": 1.2265625,
      "learning_rate": 7.34825466925787e-06,
      "loss": 1.5667,
      "step": 54941
    },
    {
      "epoch": 4.727006796868278,
      "grad_norm": 1.1171875,
      "learning_rate": 7.343636484489935e-06,
      "loss": 1.2681,
      "step": 54942
    },
    {
      "epoch": 4.7270928331756,
      "grad_norm": 3.828125,
      "learning_rate": 7.339019740646047e-06,
      "loss": 1.0474,
      "step": 54943
    },
    {
      "epoch": 4.7271788694829215,
      "grad_norm": 1.5234375,
      "learning_rate": 7.33440443773975e-06,
      "loss": 0.805,
      "step": 54944
    },
    {
      "epoch": 4.727264905790244,
      "grad_norm": 1.2265625,
      "learning_rate": 7.329790575784479e-06,
      "loss": 2.1923,
      "step": 54945
    },
    {
      "epoch": 4.727350942097565,
      "grad_norm": 1.3359375,
      "learning_rate": 7.3251781547937215e-06,
      "loss": 1.6515,
      "step": 54946
    },
    {
      "epoch": 4.727436978404887,
      "grad_norm": 0.77734375,
      "learning_rate": 7.320567174781023e-06,
      "loss": 1.0553,
      "step": 54947
    },
    {
      "epoch": 4.727523014712209,
      "grad_norm": 111.0,
      "learning_rate": 7.315957635759874e-06,
      "loss": 1.5721,
      "step": 54948
    },
    {
      "epoch": 4.72760905101953,
      "grad_norm": 1.34375,
      "learning_rate": 7.311349537743705e-06,
      "loss": 1.2128,
      "step": 54949
    },
    {
      "epoch": 4.727695087326852,
      "grad_norm": 1.2890625,
      "learning_rate": 7.306742880746009e-06,
      "loss": 0.9567,
      "step": 54950
    },
    {
      "epoch": 4.727781123634173,
      "grad_norm": 1.171875,
      "learning_rate": 7.3021376647802175e-06,
      "loss": 0.9129,
      "step": 54951
    },
    {
      "epoch": 4.727867159941495,
      "grad_norm": 14.3125,
      "learning_rate": 7.297533889859931e-06,
      "loss": 0.773,
      "step": 54952
    },
    {
      "epoch": 4.7279531962488175,
      "grad_norm": 1.3984375,
      "learning_rate": 7.292931555998472e-06,
      "loss": 1.3444,
      "step": 54953
    },
    {
      "epoch": 4.728039232556139,
      "grad_norm": 2.40625,
      "learning_rate": 7.288330663209386e-06,
      "loss": 0.7915,
      "step": 54954
    },
    {
      "epoch": 4.728125268863461,
      "grad_norm": 1.125,
      "learning_rate": 7.2837312115061615e-06,
      "loss": 0.9146,
      "step": 54955
    },
    {
      "epoch": 4.728211305170782,
      "grad_norm": 1.2421875,
      "learning_rate": 7.279133200902066e-06,
      "loss": 0.9834,
      "step": 54956
    },
    {
      "epoch": 4.728297341478104,
      "grad_norm": 0.9609375,
      "learning_rate": 7.274536631410811e-06,
      "loss": 0.9914,
      "step": 54957
    },
    {
      "epoch": 4.728383377785425,
      "grad_norm": 1.5703125,
      "learning_rate": 7.269941503045608e-06,
      "loss": 1.9561,
      "step": 54958
    },
    {
      "epoch": 4.728469414092747,
      "grad_norm": 1.046875,
      "learning_rate": 7.265347815820001e-06,
      "loss": 0.9325,
      "step": 54959
    },
    {
      "epoch": 4.728555450400069,
      "grad_norm": 7.125,
      "learning_rate": 7.2607555697474796e-06,
      "loss": 1.3606,
      "step": 54960
    },
    {
      "epoch": 4.7286414867073905,
      "grad_norm": 1.03125,
      "learning_rate": 7.256164764841422e-06,
      "loss": 0.8752,
      "step": 54961
    },
    {
      "epoch": 4.7287275230147126,
      "grad_norm": 1.0234375,
      "learning_rate": 7.251575401115207e-06,
      "loss": 0.6538,
      "step": 54962
    },
    {
      "epoch": 4.728813559322034,
      "grad_norm": 1.078125,
      "learning_rate": 7.246987478582323e-06,
      "loss": 1.1636,
      "step": 54963
    },
    {
      "epoch": 4.728899595629356,
      "grad_norm": 0.80859375,
      "learning_rate": 7.242400997256149e-06,
      "loss": 0.6359,
      "step": 54964
    },
    {
      "epoch": 4.728985631936677,
      "grad_norm": 1.34375,
      "learning_rate": 7.237815957150118e-06,
      "loss": 1.4042,
      "step": 54965
    },
    {
      "epoch": 4.729071668243999,
      "grad_norm": 1.78125,
      "learning_rate": 7.233232358277664e-06,
      "loss": 1.2309,
      "step": 54966
    },
    {
      "epoch": 4.729157704551321,
      "grad_norm": 1.2890625,
      "learning_rate": 7.228650200652109e-06,
      "loss": 1.0936,
      "step": 54967
    },
    {
      "epoch": 4.729243740858642,
      "grad_norm": 1.203125,
      "learning_rate": 7.224069484286943e-06,
      "loss": 1.2668,
      "step": 54968
    },
    {
      "epoch": 4.729329777165964,
      "grad_norm": 10.25,
      "learning_rate": 7.219490209195545e-06,
      "loss": 1.7331,
      "step": 54969
    },
    {
      "epoch": 4.7294158134732855,
      "grad_norm": 1.1796875,
      "learning_rate": 7.214912375391292e-06,
      "loss": 0.8498,
      "step": 54970
    },
    {
      "epoch": 4.729501849780608,
      "grad_norm": 1.640625,
      "learning_rate": 7.2103359828876165e-06,
      "loss": 1.3578,
      "step": 54971
    },
    {
      "epoch": 4.729587886087929,
      "grad_norm": 7.625,
      "learning_rate": 7.205761031697789e-06,
      "loss": 1.0801,
      "step": 54972
    },
    {
      "epoch": 4.729673922395251,
      "grad_norm": 0.98828125,
      "learning_rate": 7.201187521835295e-06,
      "loss": 0.7886,
      "step": 54973
    },
    {
      "epoch": 4.729759958702573,
      "grad_norm": 446.0,
      "learning_rate": 7.196615453313516e-06,
      "loss": 0.6994,
      "step": 54974
    },
    {
      "epoch": 4.729845995009894,
      "grad_norm": 0.94921875,
      "learning_rate": 7.1920448261457715e-06,
      "loss": 0.7661,
      "step": 54975
    },
    {
      "epoch": 4.729932031317216,
      "grad_norm": 2.78125,
      "learning_rate": 7.187475640345498e-06,
      "loss": 0.6305,
      "step": 54976
    },
    {
      "epoch": 4.730018067624537,
      "grad_norm": 1.1171875,
      "learning_rate": 7.182907895925905e-06,
      "loss": 1.6162,
      "step": 54977
    },
    {
      "epoch": 4.730104103931859,
      "grad_norm": 0.9453125,
      "learning_rate": 7.178341592900539e-06,
      "loss": 1.3513,
      "step": 54978
    },
    {
      "epoch": 4.730190140239181,
      "grad_norm": 1.125,
      "learning_rate": 7.173776731282666e-06,
      "loss": 1.6961,
      "step": 54979
    },
    {
      "epoch": 4.730276176546503,
      "grad_norm": 0.96484375,
      "learning_rate": 7.169213311085609e-06,
      "loss": 1.3264,
      "step": 54980
    },
    {
      "epoch": 4.730362212853825,
      "grad_norm": 0.9921875,
      "learning_rate": 7.164651332322803e-06,
      "loss": 0.9117,
      "step": 54981
    },
    {
      "epoch": 4.730448249161146,
      "grad_norm": 1.1484375,
      "learning_rate": 7.160090795007512e-06,
      "loss": 1.3778,
      "step": 54982
    },
    {
      "epoch": 4.730534285468468,
      "grad_norm": 1.1640625,
      "learning_rate": 7.1555316991531176e-06,
      "loss": 1.2912,
      "step": 54983
    },
    {
      "epoch": 4.730620321775789,
      "grad_norm": 1.0390625,
      "learning_rate": 7.150974044772885e-06,
      "loss": 1.0224,
      "step": 54984
    },
    {
      "epoch": 4.730706358083111,
      "grad_norm": 1.25,
      "learning_rate": 7.1464178318803034e-06,
      "loss": 1.998,
      "step": 54985
    },
    {
      "epoch": 4.730792394390432,
      "grad_norm": 0.97265625,
      "learning_rate": 7.141863060488529e-06,
      "loss": 0.7187,
      "step": 54986
    },
    {
      "epoch": 4.7308784306977545,
      "grad_norm": 0.83984375,
      "learning_rate": 7.137309730610942e-06,
      "loss": 0.8297,
      "step": 54987
    },
    {
      "epoch": 4.730964467005077,
      "grad_norm": 1.09375,
      "learning_rate": 7.1327578422608615e-06,
      "loss": 1.3167,
      "step": 54988
    },
    {
      "epoch": 4.731050503312398,
      "grad_norm": 1.3671875,
      "learning_rate": 7.128207395451614e-06,
      "loss": 1.447,
      "step": 54989
    },
    {
      "epoch": 4.73113653961972,
      "grad_norm": 1.328125,
      "learning_rate": 7.1236583901965745e-06,
      "loss": 0.9183,
      "step": 54990
    },
    {
      "epoch": 4.731222575927041,
      "grad_norm": 1.0859375,
      "learning_rate": 7.119110826508901e-06,
      "loss": 0.8017,
      "step": 54991
    },
    {
      "epoch": 4.731308612234363,
      "grad_norm": 0.71484375,
      "learning_rate": 7.114564704401971e-06,
      "loss": 0.5956,
      "step": 54992
    },
    {
      "epoch": 4.731394648541684,
      "grad_norm": 1.8046875,
      "learning_rate": 7.1100200238890526e-06,
      "loss": 0.9937,
      "step": 54993
    },
    {
      "epoch": 4.731480684849006,
      "grad_norm": 1.6484375,
      "learning_rate": 7.105476784983523e-06,
      "loss": 1.6497,
      "step": 54994
    },
    {
      "epoch": 4.731566721156328,
      "grad_norm": 0.92578125,
      "learning_rate": 7.100934987698593e-06,
      "loss": 1.0241,
      "step": 54995
    },
    {
      "epoch": 4.7316527574636496,
      "grad_norm": 1.2109375,
      "learning_rate": 7.096394632047532e-06,
      "loss": 1.2637,
      "step": 54996
    },
    {
      "epoch": 4.731738793770972,
      "grad_norm": 1.2265625,
      "learning_rate": 7.091855718043661e-06,
      "loss": 1.2807,
      "step": 54997
    },
    {
      "epoch": 4.731824830078293,
      "grad_norm": 1.0859375,
      "learning_rate": 7.087318245700303e-06,
      "loss": 0.4821,
      "step": 54998
    },
    {
      "epoch": 4.731910866385615,
      "grad_norm": 1.0234375,
      "learning_rate": 7.082782215030614e-06,
      "loss": 0.9501,
      "step": 54999
    },
    {
      "epoch": 4.731996902692936,
      "grad_norm": 1.1796875,
      "learning_rate": 7.0782476260479734e-06,
      "loss": 1.2608,
      "step": 55000
    },
    {
      "epoch": 4.732082939000258,
      "grad_norm": 1.0234375,
      "learning_rate": 7.0737144787655364e-06,
      "loss": 0.8887,
      "step": 55001
    },
    {
      "epoch": 4.73216897530758,
      "grad_norm": 1.09375,
      "learning_rate": 7.069182773196625e-06,
      "loss": 1.1943,
      "step": 55002
    },
    {
      "epoch": 4.732255011614901,
      "grad_norm": 1.125,
      "learning_rate": 7.064652509354508e-06,
      "loss": 0.684,
      "step": 55003
    },
    {
      "epoch": 4.732341047922223,
      "grad_norm": 1.0078125,
      "learning_rate": 7.060123687252451e-06,
      "loss": 0.827,
      "step": 55004
    },
    {
      "epoch": 4.732427084229545,
      "grad_norm": 1.5390625,
      "learning_rate": 7.055596306903555e-06,
      "loss": 1.1785,
      "step": 55005
    },
    {
      "epoch": 4.732513120536867,
      "grad_norm": 4.90625,
      "learning_rate": 7.051070368321255e-06,
      "loss": 1.3035,
      "step": 55006
    },
    {
      "epoch": 4.732599156844188,
      "grad_norm": 1.0,
      "learning_rate": 7.04654587151865e-06,
      "loss": 1.1117,
      "step": 55007
    },
    {
      "epoch": 4.73268519315151,
      "grad_norm": 0.984375,
      "learning_rate": 7.042022816509064e-06,
      "loss": 0.8105,
      "step": 55008
    },
    {
      "epoch": 4.732771229458832,
      "grad_norm": 48.25,
      "learning_rate": 7.037501203305707e-06,
      "loss": 0.8012,
      "step": 55009
    },
    {
      "epoch": 4.732857265766153,
      "grad_norm": 1.0390625,
      "learning_rate": 7.032981031921682e-06,
      "loss": 1.1079,
      "step": 55010
    },
    {
      "epoch": 4.732943302073475,
      "grad_norm": 1.25,
      "learning_rate": 7.028462302370364e-06,
      "loss": 1.1083,
      "step": 55011
    },
    {
      "epoch": 4.733029338380796,
      "grad_norm": 1.09375,
      "learning_rate": 7.023945014664967e-06,
      "loss": 1.1866,
      "step": 55012
    },
    {
      "epoch": 4.7331153746881185,
      "grad_norm": 1.359375,
      "learning_rate": 7.019429168818592e-06,
      "loss": 1.1062,
      "step": 55013
    },
    {
      "epoch": 4.73320141099544,
      "grad_norm": 0.9609375,
      "learning_rate": 7.014914764844505e-06,
      "loss": 0.7569,
      "step": 55014
    },
    {
      "epoch": 4.733287447302762,
      "grad_norm": 1.2734375,
      "learning_rate": 7.010401802755972e-06,
      "loss": 1.4122,
      "step": 55015
    },
    {
      "epoch": 4.733373483610084,
      "grad_norm": 0.984375,
      "learning_rate": 7.005890282566096e-06,
      "loss": 0.8019,
      "step": 55016
    },
    {
      "epoch": 4.733459519917405,
      "grad_norm": 1.515625,
      "learning_rate": 7.001380204288088e-06,
      "loss": 0.8657,
      "step": 55017
    },
    {
      "epoch": 4.733545556224727,
      "grad_norm": 0.953125,
      "learning_rate": 6.996871567935159e-06,
      "loss": 0.5898,
      "step": 55018
    },
    {
      "epoch": 4.733631592532048,
      "grad_norm": 198.0,
      "learning_rate": 6.992364373520521e-06,
      "loss": 0.8356,
      "step": 55019
    },
    {
      "epoch": 4.73371762883937,
      "grad_norm": 1.2890625,
      "learning_rate": 6.98785862105733e-06,
      "loss": 0.7569,
      "step": 55020
    },
    {
      "epoch": 4.7338036651466915,
      "grad_norm": 0.9296875,
      "learning_rate": 6.983354310558798e-06,
      "loss": 1.1266,
      "step": 55021
    },
    {
      "epoch": 4.7338897014540136,
      "grad_norm": 1.3671875,
      "learning_rate": 6.978851442038026e-06,
      "loss": 0.846,
      "step": 55022
    },
    {
      "epoch": 4.733975737761336,
      "grad_norm": 1.125,
      "learning_rate": 6.974350015508224e-06,
      "loss": 1.281,
      "step": 55023
    },
    {
      "epoch": 4.734061774068657,
      "grad_norm": 0.91796875,
      "learning_rate": 6.96985003098255e-06,
      "loss": 1.1701,
      "step": 55024
    },
    {
      "epoch": 4.734147810375979,
      "grad_norm": 1.6171875,
      "learning_rate": 6.965351488474214e-06,
      "loss": 1.4864,
      "step": 55025
    },
    {
      "epoch": 4.7342338466833,
      "grad_norm": 195.0,
      "learning_rate": 6.960854387996318e-06,
      "loss": 0.9653,
      "step": 55026
    },
    {
      "epoch": 4.734319882990622,
      "grad_norm": 1.3828125,
      "learning_rate": 6.956358729562018e-06,
      "loss": 1.3294,
      "step": 55027
    },
    {
      "epoch": 4.734405919297943,
      "grad_norm": 1.21875,
      "learning_rate": 6.951864513184469e-06,
      "loss": 1.8233,
      "step": 55028
    },
    {
      "epoch": 4.734491955605265,
      "grad_norm": 16.0,
      "learning_rate": 6.9473717388768844e-06,
      "loss": 1.3416,
      "step": 55029
    },
    {
      "epoch": 4.734577991912587,
      "grad_norm": 2.6875,
      "learning_rate": 6.942880406652252e-06,
      "loss": 1.5885,
      "step": 55030
    },
    {
      "epoch": 4.734664028219909,
      "grad_norm": 13.0625,
      "learning_rate": 6.938390516523896e-06,
      "loss": 0.9487,
      "step": 55031
    },
    {
      "epoch": 4.734750064527231,
      "grad_norm": 1.1015625,
      "learning_rate": 6.9339020685047496e-06,
      "loss": 0.8715,
      "step": 55032
    },
    {
      "epoch": 4.734836100834552,
      "grad_norm": 0.8828125,
      "learning_rate": 6.929415062608136e-06,
      "loss": 0.7584,
      "step": 55033
    },
    {
      "epoch": 4.734922137141874,
      "grad_norm": 1.2421875,
      "learning_rate": 6.9249294988470436e-06,
      "loss": 1.7255,
      "step": 55034
    },
    {
      "epoch": 4.735008173449195,
      "grad_norm": 1.5234375,
      "learning_rate": 6.92044537723463e-06,
      "loss": 1.3177,
      "step": 55035
    },
    {
      "epoch": 4.735094209756517,
      "grad_norm": 1.078125,
      "learning_rate": 6.915962697784051e-06,
      "loss": 0.9394,
      "step": 55036
    },
    {
      "epoch": 4.735180246063839,
      "grad_norm": 1.203125,
      "learning_rate": 6.911481460508351e-06,
      "loss": 1.0844,
      "step": 55037
    },
    {
      "epoch": 4.73526628237116,
      "grad_norm": 4.03125,
      "learning_rate": 6.907001665420687e-06,
      "loss": 0.9733,
      "step": 55038
    },
    {
      "epoch": 4.7353523186784825,
      "grad_norm": 0.88671875,
      "learning_rate": 6.902523312534159e-06,
      "loss": 0.8451,
      "step": 55039
    },
    {
      "epoch": 4.735438354985804,
      "grad_norm": 0.890625,
      "learning_rate": 6.898046401861813e-06,
      "loss": 0.9082,
      "step": 55040
    },
    {
      "epoch": 4.735524391293126,
      "grad_norm": 1.125,
      "learning_rate": 6.8935709334167485e-06,
      "loss": 0.6944,
      "step": 55041
    },
    {
      "epoch": 4.735610427600447,
      "grad_norm": 1.3203125,
      "learning_rate": 6.889096907212122e-06,
      "loss": 0.7371,
      "step": 55042
    },
    {
      "epoch": 4.735696463907769,
      "grad_norm": 0.859375,
      "learning_rate": 6.88462432326098e-06,
      "loss": 1.2725,
      "step": 55043
    },
    {
      "epoch": 4.735782500215091,
      "grad_norm": 2.59375,
      "learning_rate": 6.88015318157642e-06,
      "loss": 1.4667,
      "step": 55044
    },
    {
      "epoch": 4.735868536522412,
      "grad_norm": 1.109375,
      "learning_rate": 6.875683482171491e-06,
      "loss": 0.5586,
      "step": 55045
    },
    {
      "epoch": 4.735954572829734,
      "grad_norm": 1.015625,
      "learning_rate": 6.871215225059291e-06,
      "loss": 0.6689,
      "step": 55046
    },
    {
      "epoch": 4.7360406091370555,
      "grad_norm": 0.953125,
      "learning_rate": 6.8667484102528655e-06,
      "loss": 0.4526,
      "step": 55047
    },
    {
      "epoch": 4.736126645444378,
      "grad_norm": 1.1015625,
      "learning_rate": 6.862283037765316e-06,
      "loss": 1.5752,
      "step": 55048
    },
    {
      "epoch": 4.736212681751699,
      "grad_norm": 1.15625,
      "learning_rate": 6.8578191076096305e-06,
      "loss": 1.4319,
      "step": 55049
    },
    {
      "epoch": 4.736298718059021,
      "grad_norm": 0.96484375,
      "learning_rate": 6.853356619798967e-06,
      "loss": 1.1839,
      "step": 55050
    },
    {
      "epoch": 4.736384754366343,
      "grad_norm": 1.078125,
      "learning_rate": 6.8488955743463145e-06,
      "loss": 0.8053,
      "step": 55051
    },
    {
      "epoch": 4.736470790673664,
      "grad_norm": 0.984375,
      "learning_rate": 6.844435971264662e-06,
      "loss": 1.2467,
      "step": 55052
    },
    {
      "epoch": 4.736556826980986,
      "grad_norm": 0.953125,
      "learning_rate": 6.839977810567222e-06,
      "loss": 0.5144,
      "step": 55053
    },
    {
      "epoch": 4.736642863288307,
      "grad_norm": 1.234375,
      "learning_rate": 6.835521092266872e-06,
      "loss": 0.8994,
      "step": 55054
    },
    {
      "epoch": 4.736728899595629,
      "grad_norm": 0.84765625,
      "learning_rate": 6.831065816376714e-06,
      "loss": 0.9222,
      "step": 55055
    },
    {
      "epoch": 4.7368149359029506,
      "grad_norm": 1.0625,
      "learning_rate": 6.826611982909792e-06,
      "loss": 1.2848,
      "step": 55056
    },
    {
      "epoch": 4.736900972210273,
      "grad_norm": 11.25,
      "learning_rate": 6.822159591879096e-06,
      "loss": 1.425,
      "step": 55057
    },
    {
      "epoch": 4.736987008517595,
      "grad_norm": 1.015625,
      "learning_rate": 6.817708643297615e-06,
      "loss": 0.7675,
      "step": 55058
    },
    {
      "epoch": 4.737073044824916,
      "grad_norm": 1.0078125,
      "learning_rate": 6.813259137178507e-06,
      "loss": 1.1123,
      "step": 55059
    },
    {
      "epoch": 4.737159081132238,
      "grad_norm": 1.2109375,
      "learning_rate": 6.808811073534649e-06,
      "loss": 1.4902,
      "step": 55060
    },
    {
      "epoch": 4.737245117439559,
      "grad_norm": 1.0703125,
      "learning_rate": 6.804364452379141e-06,
      "loss": 1.4829,
      "step": 55061
    },
    {
      "epoch": 4.737331153746881,
      "grad_norm": 3.484375,
      "learning_rate": 6.799919273724919e-06,
      "loss": 0.748,
      "step": 55062
    },
    {
      "epoch": 4.737417190054203,
      "grad_norm": 1.2421875,
      "learning_rate": 6.795475537584972e-06,
      "loss": 1.7266,
      "step": 55063
    },
    {
      "epoch": 4.737503226361524,
      "grad_norm": 1.1953125,
      "learning_rate": 6.7910332439724e-06,
      "loss": 0.8723,
      "step": 55064
    },
    {
      "epoch": 4.7375892626688465,
      "grad_norm": 1.046875,
      "learning_rate": 6.786592392900082e-06,
      "loss": 1.5202,
      "step": 55065
    },
    {
      "epoch": 4.737675298976168,
      "grad_norm": 1.078125,
      "learning_rate": 6.782152984381063e-06,
      "loss": 0.9908,
      "step": 55066
    },
    {
      "epoch": 4.73776133528349,
      "grad_norm": 0.828125,
      "learning_rate": 6.777715018428332e-06,
      "loss": 0.8635,
      "step": 55067
    },
    {
      "epoch": 4.737847371590812,
      "grad_norm": 0.97265625,
      "learning_rate": 6.773278495054825e-06,
      "loss": 1.273,
      "step": 55068
    },
    {
      "epoch": 4.737933407898133,
      "grad_norm": 1.078125,
      "learning_rate": 6.7688434142736395e-06,
      "loss": 0.848,
      "step": 55069
    },
    {
      "epoch": 4.738019444205455,
      "grad_norm": 1.359375,
      "learning_rate": 6.764409776097602e-06,
      "loss": 1.8851,
      "step": 55070
    },
    {
      "epoch": 4.738105480512776,
      "grad_norm": 1.0234375,
      "learning_rate": 6.7599775805397e-06,
      "loss": 0.8089,
      "step": 55071
    },
    {
      "epoch": 4.738191516820098,
      "grad_norm": 21.75,
      "learning_rate": 6.755546827613035e-06,
      "loss": 2.0004,
      "step": 55072
    },
    {
      "epoch": 4.7382775531274195,
      "grad_norm": 1.234375,
      "learning_rate": 6.751117517330374e-06,
      "loss": 1.2917,
      "step": 55073
    },
    {
      "epoch": 4.738363589434742,
      "grad_norm": 1.2578125,
      "learning_rate": 6.746689649704763e-06,
      "loss": 0.8,
      "step": 55074
    },
    {
      "epoch": 4.738449625742064,
      "grad_norm": 1.4140625,
      "learning_rate": 6.742263224749246e-06,
      "loss": 1.2891,
      "step": 55075
    },
    {
      "epoch": 4.738535662049385,
      "grad_norm": 1.203125,
      "learning_rate": 6.73783824247659e-06,
      "loss": 1.2024,
      "step": 55076
    },
    {
      "epoch": 4.738621698356707,
      "grad_norm": 1.296875,
      "learning_rate": 6.733414702899843e-06,
      "loss": 2.0111,
      "step": 55077
    },
    {
      "epoch": 4.738707734664028,
      "grad_norm": 0.9921875,
      "learning_rate": 6.7289926060319365e-06,
      "loss": 1.1248,
      "step": 55078
    },
    {
      "epoch": 4.73879377097135,
      "grad_norm": 1.2890625,
      "learning_rate": 6.7245719518857495e-06,
      "loss": 1.2504,
      "step": 55079
    },
    {
      "epoch": 4.738879807278671,
      "grad_norm": 1.140625,
      "learning_rate": 6.720152740474328e-06,
      "loss": 1.6479,
      "step": 55080
    },
    {
      "epoch": 4.738965843585993,
      "grad_norm": 0.8515625,
      "learning_rate": 6.715734971810438e-06,
      "loss": 1.0822,
      "step": 55081
    },
    {
      "epoch": 4.7390518798933154,
      "grad_norm": 0.9765625,
      "learning_rate": 6.711318645907128e-06,
      "loss": 0.856,
      "step": 55082
    },
    {
      "epoch": 4.739137916200637,
      "grad_norm": 2.234375,
      "learning_rate": 6.706903762777272e-06,
      "loss": 0.8058,
      "step": 55083
    },
    {
      "epoch": 4.739223952507959,
      "grad_norm": 0.9375,
      "learning_rate": 6.702490322433752e-06,
      "loss": 0.772,
      "step": 55084
    },
    {
      "epoch": 4.73930998881528,
      "grad_norm": 1.09375,
      "learning_rate": 6.698078324889556e-06,
      "loss": 0.797,
      "step": 55085
    },
    {
      "epoch": 4.739396025122602,
      "grad_norm": 308.0,
      "learning_rate": 6.693667770157508e-06,
      "loss": 0.9644,
      "step": 55086
    },
    {
      "epoch": 4.739482061429923,
      "grad_norm": 1.234375,
      "learning_rate": 6.689258658250541e-06,
      "loss": 0.6672,
      "step": 55087
    },
    {
      "epoch": 4.739568097737245,
      "grad_norm": 1.265625,
      "learning_rate": 6.6848509891815904e-06,
      "loss": 1.7454,
      "step": 55088
    },
    {
      "epoch": 4.739654134044567,
      "grad_norm": 1.21875,
      "learning_rate": 6.680444762963478e-06,
      "loss": 1.6676,
      "step": 55089
    },
    {
      "epoch": 4.739740170351888,
      "grad_norm": 1.1171875,
      "learning_rate": 6.676039979609083e-06,
      "loss": 0.3966,
      "step": 55090
    },
    {
      "epoch": 4.7398262066592105,
      "grad_norm": 1.1015625,
      "learning_rate": 6.671636639131395e-06,
      "loss": 0.7041,
      "step": 55091
    },
    {
      "epoch": 4.739912242966532,
      "grad_norm": 0.7578125,
      "learning_rate": 6.667234741543127e-06,
      "loss": 0.9544,
      "step": 55092
    },
    {
      "epoch": 4.739998279273854,
      "grad_norm": 1.328125,
      "learning_rate": 6.662834286857322e-06,
      "loss": 1.1882,
      "step": 55093
    },
    {
      "epoch": 4.740084315581175,
      "grad_norm": 1.2890625,
      "learning_rate": 6.6584352750867495e-06,
      "loss": 0.9832,
      "step": 55094
    },
    {
      "epoch": 4.740170351888497,
      "grad_norm": 1.453125,
      "learning_rate": 6.654037706244342e-06,
      "loss": 1.828,
      "step": 55095
    },
    {
      "epoch": 4.740256388195819,
      "grad_norm": 0.90625,
      "learning_rate": 6.649641580342869e-06,
      "loss": 0.6176,
      "step": 55096
    },
    {
      "epoch": 4.74034242450314,
      "grad_norm": 1.09375,
      "learning_rate": 6.6452468973952625e-06,
      "loss": 1.0962,
      "step": 55097
    },
    {
      "epoch": 4.740428460810462,
      "grad_norm": 12.5,
      "learning_rate": 6.6408536574143475e-06,
      "loss": 0.7705,
      "step": 55098
    },
    {
      "epoch": 4.7405144971177835,
      "grad_norm": 1.53125,
      "learning_rate": 6.6364618604130014e-06,
      "loss": 1.2666,
      "step": 55099
    },
    {
      "epoch": 4.740600533425106,
      "grad_norm": 1.1640625,
      "learning_rate": 6.632071506404103e-06,
      "loss": 1.1672,
      "step": 55100
    },
    {
      "epoch": 4.740686569732427,
      "grad_norm": 1.0625,
      "learning_rate": 6.627682595400364e-06,
      "loss": 0.6566,
      "step": 55101
    },
    {
      "epoch": 4.740772606039749,
      "grad_norm": 1.09375,
      "learning_rate": 6.623295127414719e-06,
      "loss": 0.8873,
      "step": 55102
    },
    {
      "epoch": 4.740858642347071,
      "grad_norm": 1.3046875,
      "learning_rate": 6.618909102459991e-06,
      "loss": 1.742,
      "step": 55103
    },
    {
      "epoch": 4.740944678654392,
      "grad_norm": 1.3046875,
      "learning_rate": 6.614524520548948e-06,
      "loss": 1.378,
      "step": 55104
    },
    {
      "epoch": 4.741030714961714,
      "grad_norm": 1.1796875,
      "learning_rate": 6.6101413816945235e-06,
      "loss": 1.0201,
      "step": 55105
    },
    {
      "epoch": 4.741116751269035,
      "grad_norm": 1.3671875,
      "learning_rate": 6.605759685909429e-06,
      "loss": 1.3737,
      "step": 55106
    },
    {
      "epoch": 4.741202787576357,
      "grad_norm": 1.015625,
      "learning_rate": 6.601379433206545e-06,
      "loss": 0.7742,
      "step": 55107
    },
    {
      "epoch": 4.741288823883679,
      "grad_norm": 1.125,
      "learning_rate": 6.597000623598692e-06,
      "loss": 0.8252,
      "step": 55108
    },
    {
      "epoch": 4.741374860191001,
      "grad_norm": 84.5,
      "learning_rate": 6.592623257098585e-06,
      "loss": 1.3986,
      "step": 55109
    },
    {
      "epoch": 4.741460896498323,
      "grad_norm": 0.80859375,
      "learning_rate": 6.588247333719155e-06,
      "loss": 0.5461,
      "step": 55110
    },
    {
      "epoch": 4.741546932805644,
      "grad_norm": 28.25,
      "learning_rate": 6.5838728534731165e-06,
      "loss": 1.21,
      "step": 55111
    },
    {
      "epoch": 4.741632969112966,
      "grad_norm": 1.109375,
      "learning_rate": 6.579499816373291e-06,
      "loss": 0.9546,
      "step": 55112
    },
    {
      "epoch": 4.741719005420287,
      "grad_norm": 1.140625,
      "learning_rate": 6.575128222432503e-06,
      "loss": 1.1389,
      "step": 55113
    },
    {
      "epoch": 4.741805041727609,
      "grad_norm": 1.703125,
      "learning_rate": 6.570758071663463e-06,
      "loss": 1.5252,
      "step": 55114
    },
    {
      "epoch": 4.74189107803493,
      "grad_norm": 1.109375,
      "learning_rate": 6.5663893640789395e-06,
      "loss": 0.9898,
      "step": 55115
    },
    {
      "epoch": 4.741977114342252,
      "grad_norm": 1.3359375,
      "learning_rate": 6.5620220996918114e-06,
      "loss": 1.1486,
      "step": 55116
    },
    {
      "epoch": 4.7420631506495745,
      "grad_norm": 100.5,
      "learning_rate": 6.55765627851479e-06,
      "loss": 0.6679,
      "step": 55117
    },
    {
      "epoch": 4.742149186956896,
      "grad_norm": 0.98046875,
      "learning_rate": 6.553291900560698e-06,
      "loss": 0.6935,
      "step": 55118
    },
    {
      "epoch": 4.742235223264218,
      "grad_norm": 0.9375,
      "learning_rate": 6.548928965842194e-06,
      "loss": 0.5743,
      "step": 55119
    },
    {
      "epoch": 4.742321259571539,
      "grad_norm": 1.9765625,
      "learning_rate": 6.544567474372154e-06,
      "loss": 0.7705,
      "step": 55120
    },
    {
      "epoch": 4.742407295878861,
      "grad_norm": 132.0,
      "learning_rate": 6.540207426163236e-06,
      "loss": 0.8878,
      "step": 55121
    },
    {
      "epoch": 4.742493332186182,
      "grad_norm": 1.1640625,
      "learning_rate": 6.535848821228263e-06,
      "loss": 1.8531,
      "step": 55122
    },
    {
      "epoch": 4.742579368493504,
      "grad_norm": 1.09375,
      "learning_rate": 6.531491659579947e-06,
      "loss": 0.9615,
      "step": 55123
    },
    {
      "epoch": 4.742665404800826,
      "grad_norm": 1.0625,
      "learning_rate": 6.527135941231055e-06,
      "loss": 0.7077,
      "step": 55124
    },
    {
      "epoch": 4.7427514411081475,
      "grad_norm": 2.125,
      "learning_rate": 6.5227816661943e-06,
      "loss": 0.797,
      "step": 55125
    },
    {
      "epoch": 4.74283747741547,
      "grad_norm": 1.0078125,
      "learning_rate": 6.518428834482448e-06,
      "loss": 1.3524,
      "step": 55126
    },
    {
      "epoch": 4.742923513722791,
      "grad_norm": 1.1015625,
      "learning_rate": 6.514077446108213e-06,
      "loss": 1.291,
      "step": 55127
    },
    {
      "epoch": 4.743009550030113,
      "grad_norm": 2.21875,
      "learning_rate": 6.509727501084306e-06,
      "loss": 0.9317,
      "step": 55128
    },
    {
      "epoch": 4.743095586337434,
      "grad_norm": 1.203125,
      "learning_rate": 6.505378999423495e-06,
      "loss": 1.1399,
      "step": 55129
    },
    {
      "epoch": 4.743181622644756,
      "grad_norm": 1.234375,
      "learning_rate": 6.5010319411384355e-06,
      "loss": 0.7398,
      "step": 55130
    },
    {
      "epoch": 4.743267658952078,
      "grad_norm": 1.1015625,
      "learning_rate": 6.496686326241896e-06,
      "loss": 1.0638,
      "step": 55131
    },
    {
      "epoch": 4.743353695259399,
      "grad_norm": 1.09375,
      "learning_rate": 6.492342154746589e-06,
      "loss": 0.7095,
      "step": 55132
    },
    {
      "epoch": 4.743439731566721,
      "grad_norm": 1.546875,
      "learning_rate": 6.48799942666517e-06,
      "loss": 0.9898,
      "step": 55133
    },
    {
      "epoch": 4.743525767874043,
      "grad_norm": 0.95703125,
      "learning_rate": 6.483658142010407e-06,
      "loss": 1.0485,
      "step": 55134
    },
    {
      "epoch": 4.743611804181365,
      "grad_norm": 1.265625,
      "learning_rate": 6.479318300794956e-06,
      "loss": 0.6045,
      "step": 55135
    },
    {
      "epoch": 4.743697840488686,
      "grad_norm": 3.140625,
      "learning_rate": 6.474979903031475e-06,
      "loss": 0.9064,
      "step": 55136
    },
    {
      "epoch": 4.743783876796008,
      "grad_norm": 1.3203125,
      "learning_rate": 6.470642948732675e-06,
      "loss": 1.3428,
      "step": 55137
    },
    {
      "epoch": 4.74386991310333,
      "grad_norm": 1.359375,
      "learning_rate": 6.466307437911323e-06,
      "loss": 1.3231,
      "step": 55138
    },
    {
      "epoch": 4.743955949410651,
      "grad_norm": 0.84375,
      "learning_rate": 6.461973370580021e-06,
      "loss": 0.8795,
      "step": 55139
    },
    {
      "epoch": 4.744041985717973,
      "grad_norm": 3.21875,
      "learning_rate": 6.457640746751425e-06,
      "loss": 0.9058,
      "step": 55140
    },
    {
      "epoch": 4.744128022025294,
      "grad_norm": 1.140625,
      "learning_rate": 6.453309566438248e-06,
      "loss": 0.587,
      "step": 55141
    },
    {
      "epoch": 4.7442140583326164,
      "grad_norm": 1.21875,
      "learning_rate": 6.448979829653145e-06,
      "loss": 1.3011,
      "step": 55142
    },
    {
      "epoch": 4.744300094639938,
      "grad_norm": 8.25,
      "learning_rate": 6.44465153640883e-06,
      "loss": 0.9159,
      "step": 55143
    },
    {
      "epoch": 4.74438613094726,
      "grad_norm": 0.70703125,
      "learning_rate": 6.440324686717847e-06,
      "loss": 0.5529,
      "step": 55144
    },
    {
      "epoch": 4.744472167254582,
      "grad_norm": 1.3046875,
      "learning_rate": 6.4359992805929635e-06,
      "loss": 1.2869,
      "step": 55145
    },
    {
      "epoch": 4.744558203561903,
      "grad_norm": 4.4375,
      "learning_rate": 6.431675318046837e-06,
      "loss": 0.9147,
      "step": 55146
    },
    {
      "epoch": 4.744644239869225,
      "grad_norm": 1.40625,
      "learning_rate": 6.427352799092012e-06,
      "loss": 1.3188,
      "step": 55147
    },
    {
      "epoch": 4.744730276176546,
      "grad_norm": 1.3125,
      "learning_rate": 6.423031723741201e-06,
      "loss": 0.9081,
      "step": 55148
    },
    {
      "epoch": 4.744816312483868,
      "grad_norm": 0.98046875,
      "learning_rate": 6.418712092007006e-06,
      "loss": 0.4325,
      "step": 55149
    },
    {
      "epoch": 4.744902348791189,
      "grad_norm": 1.3671875,
      "learning_rate": 6.414393903902083e-06,
      "loss": 1.1012,
      "step": 55150
    },
    {
      "epoch": 4.7449883850985115,
      "grad_norm": 4.375,
      "learning_rate": 6.410077159439087e-06,
      "loss": 0.9731,
      "step": 55151
    },
    {
      "epoch": 4.745074421405834,
      "grad_norm": 1.859375,
      "learning_rate": 6.405761858630621e-06,
      "loss": 0.7999,
      "step": 55152
    },
    {
      "epoch": 4.745160457713155,
      "grad_norm": 1.1484375,
      "learning_rate": 6.401448001489285e-06,
      "loss": 1.5582,
      "step": 55153
    },
    {
      "epoch": 4.745246494020477,
      "grad_norm": 1.0234375,
      "learning_rate": 6.397135588027736e-06,
      "loss": 1.3635,
      "step": 55154
    },
    {
      "epoch": 4.745332530327798,
      "grad_norm": 1.03125,
      "learning_rate": 6.392824618258519e-06,
      "loss": 1.7701,
      "step": 55155
    },
    {
      "epoch": 4.74541856663512,
      "grad_norm": 1.578125,
      "learning_rate": 6.388515092194292e-06,
      "loss": 1.2324,
      "step": 55156
    },
    {
      "epoch": 4.745504602942441,
      "grad_norm": 1.0703125,
      "learning_rate": 6.384207009847709e-06,
      "loss": 1.0546,
      "step": 55157
    },
    {
      "epoch": 4.745590639249763,
      "grad_norm": 1.203125,
      "learning_rate": 6.379900371231262e-06,
      "loss": 0.7451,
      "step": 55158
    },
    {
      "epoch": 4.745676675557085,
      "grad_norm": 1.0703125,
      "learning_rate": 6.375595176357663e-06,
      "loss": 0.6771,
      "step": 55159
    },
    {
      "epoch": 4.745762711864407,
      "grad_norm": 0.9453125,
      "learning_rate": 6.3712914252394e-06,
      "loss": 1.0958,
      "step": 55160
    },
    {
      "epoch": 4.745848748171729,
      "grad_norm": 1.015625,
      "learning_rate": 6.366989117889077e-06,
      "loss": 1.3424,
      "step": 55161
    },
    {
      "epoch": 4.74593478447905,
      "grad_norm": 0.8671875,
      "learning_rate": 6.362688254319404e-06,
      "loss": 1.3243,
      "step": 55162
    },
    {
      "epoch": 4.746020820786372,
      "grad_norm": 1.3203125,
      "learning_rate": 6.358388834542761e-06,
      "loss": 1.1712,
      "step": 55163
    },
    {
      "epoch": 4.746106857093693,
      "grad_norm": 1.171875,
      "learning_rate": 6.354090858571859e-06,
      "loss": 1.1129,
      "step": 55164
    },
    {
      "epoch": 4.746192893401015,
      "grad_norm": 1.0390625,
      "learning_rate": 6.349794326419245e-06,
      "loss": 0.6352,
      "step": 55165
    },
    {
      "epoch": 4.746278929708337,
      "grad_norm": 1.21875,
      "learning_rate": 6.345499238097463e-06,
      "loss": 1.6864,
      "step": 55166
    },
    {
      "epoch": 4.746364966015658,
      "grad_norm": 1.109375,
      "learning_rate": 6.34120559361906e-06,
      "loss": 0.9705,
      "step": 55167
    },
    {
      "epoch": 4.7464510023229805,
      "grad_norm": 1.2578125,
      "learning_rate": 6.336913392996635e-06,
      "loss": 1.2285,
      "step": 55168
    },
    {
      "epoch": 4.746537038630302,
      "grad_norm": 1.1328125,
      "learning_rate": 6.332622636242736e-06,
      "loss": 1.2777,
      "step": 55169
    },
    {
      "epoch": 4.746623074937624,
      "grad_norm": 0.984375,
      "learning_rate": 6.328333323369961e-06,
      "loss": 1.2376,
      "step": 55170
    },
    {
      "epoch": 4.746709111244945,
      "grad_norm": 60.0,
      "learning_rate": 6.324045454390692e-06,
      "loss": 1.8213,
      "step": 55171
    },
    {
      "epoch": 4.746795147552267,
      "grad_norm": 776.0,
      "learning_rate": 6.319759029317584e-06,
      "loss": 1.2697,
      "step": 55172
    },
    {
      "epoch": 4.746881183859589,
      "grad_norm": 0.921875,
      "learning_rate": 6.315474048163239e-06,
      "loss": 1.2038,
      "step": 55173
    },
    {
      "epoch": 4.74696722016691,
      "grad_norm": 2.203125,
      "learning_rate": 6.31119051094009e-06,
      "loss": 0.9923,
      "step": 55174
    },
    {
      "epoch": 4.747053256474232,
      "grad_norm": 69.0,
      "learning_rate": 6.306908417660628e-06,
      "loss": 0.8158,
      "step": 55175
    },
    {
      "epoch": 4.747139292781553,
      "grad_norm": 1.046875,
      "learning_rate": 6.302627768337566e-06,
      "loss": 0.9603,
      "step": 55176
    },
    {
      "epoch": 4.7472253290888755,
      "grad_norm": 1.28125,
      "learning_rate": 6.298348562983169e-06,
      "loss": 1.5368,
      "step": 55177
    },
    {
      "epoch": 4.747311365396198,
      "grad_norm": 1.8984375,
      "learning_rate": 6.2940708016101524e-06,
      "loss": 1.274,
      "step": 55178
    },
    {
      "epoch": 4.747397401703519,
      "grad_norm": 64.5,
      "learning_rate": 6.289794484230949e-06,
      "loss": 1.2229,
      "step": 55179
    },
    {
      "epoch": 4.747483438010841,
      "grad_norm": 0.9453125,
      "learning_rate": 6.285519610858048e-06,
      "loss": 1.2704,
      "step": 55180
    },
    {
      "epoch": 4.747569474318162,
      "grad_norm": 1.2265625,
      "learning_rate": 6.281246181503997e-06,
      "loss": 1.4502,
      "step": 55181
    },
    {
      "epoch": 4.747655510625484,
      "grad_norm": 1.2734375,
      "learning_rate": 6.276974196181284e-06,
      "loss": 1.3054,
      "step": 55182
    },
    {
      "epoch": 4.747741546932805,
      "grad_norm": 1.1328125,
      "learning_rate": 6.272703654902401e-06,
      "loss": 1.4732,
      "step": 55183
    },
    {
      "epoch": 4.747827583240127,
      "grad_norm": 1.0234375,
      "learning_rate": 6.268434557679836e-06,
      "loss": 0.9578,
      "step": 55184
    },
    {
      "epoch": 4.747913619547449,
      "grad_norm": 1.1640625,
      "learning_rate": 6.264166904526081e-06,
      "loss": 1.4982,
      "step": 55185
    },
    {
      "epoch": 4.747999655854771,
      "grad_norm": 1.2578125,
      "learning_rate": 6.259900695453569e-06,
      "loss": 1.5963,
      "step": 55186
    },
    {
      "epoch": 4.748085692162093,
      "grad_norm": 664.0,
      "learning_rate": 6.255635930474901e-06,
      "loss": 1.3791,
      "step": 55187
    },
    {
      "epoch": 4.748171728469414,
      "grad_norm": 1.3203125,
      "learning_rate": 6.251372609602401e-06,
      "loss": 1.1287,
      "step": 55188
    },
    {
      "epoch": 4.748257764776736,
      "grad_norm": 1.5546875,
      "learning_rate": 6.2471107328486155e-06,
      "loss": 1.0718,
      "step": 55189
    },
    {
      "epoch": 4.748343801084058,
      "grad_norm": 0.9296875,
      "learning_rate": 6.242850300226033e-06,
      "loss": 1.2326,
      "step": 55190
    },
    {
      "epoch": 4.748429837391379,
      "grad_norm": 1.265625,
      "learning_rate": 6.2385913117470325e-06,
      "loss": 1.1363,
      "step": 55191
    },
    {
      "epoch": 4.748515873698701,
      "grad_norm": 1.2734375,
      "learning_rate": 6.234333767424161e-06,
      "loss": 1.1252,
      "step": 55192
    },
    {
      "epoch": 4.748601910006022,
      "grad_norm": 1.015625,
      "learning_rate": 6.230077667269851e-06,
      "loss": 1.5514,
      "step": 55193
    },
    {
      "epoch": 4.7486879463133445,
      "grad_norm": 948.0,
      "learning_rate": 6.225823011296483e-06,
      "loss": 1.4678,
      "step": 55194
    },
    {
      "epoch": 4.748773982620666,
      "grad_norm": 59.25,
      "learning_rate": 6.221569799516602e-06,
      "loss": 0.8002,
      "step": 55195
    },
    {
      "epoch": 4.748860018927988,
      "grad_norm": 1.1953125,
      "learning_rate": 6.217318031942531e-06,
      "loss": 1.2561,
      "step": 55196
    },
    {
      "epoch": 4.74894605523531,
      "grad_norm": 1.2890625,
      "learning_rate": 6.213067708586762e-06,
      "loss": 1.3649,
      "step": 55197
    },
    {
      "epoch": 4.749032091542631,
      "grad_norm": 2.078125,
      "learning_rate": 6.208818829461837e-06,
      "loss": 1.5783,
      "step": 55198
    },
    {
      "epoch": 4.749118127849953,
      "grad_norm": 1.0703125,
      "learning_rate": 6.204571394579972e-06,
      "loss": 0.8639,
      "step": 55199
    },
    {
      "epoch": 4.749204164157274,
      "grad_norm": 1.203125,
      "learning_rate": 6.200325403953766e-06,
      "loss": 1.1288,
      "step": 55200
    },
    {
      "epoch": 4.749290200464596,
      "grad_norm": 1.3125,
      "learning_rate": 6.196080857595543e-06,
      "loss": 0.9052,
      "step": 55201
    },
    {
      "epoch": 4.7493762367719174,
      "grad_norm": 1.1875,
      "learning_rate": 6.1918377555177375e-06,
      "loss": 1.1908,
      "step": 55202
    },
    {
      "epoch": 4.7494622730792395,
      "grad_norm": 1.4609375,
      "learning_rate": 6.187596097732784e-06,
      "loss": 1.5057,
      "step": 55203
    },
    {
      "epoch": 4.749548309386562,
      "grad_norm": 0.78125,
      "learning_rate": 6.183355884253062e-06,
      "loss": 0.5749,
      "step": 55204
    },
    {
      "epoch": 4.749634345693883,
      "grad_norm": 1.234375,
      "learning_rate": 6.179117115090949e-06,
      "loss": 1.2653,
      "step": 55205
    },
    {
      "epoch": 4.749720382001205,
      "grad_norm": 16.875,
      "learning_rate": 6.174879790258936e-06,
      "loss": 1.2416,
      "step": 55206
    },
    {
      "epoch": 4.749806418308526,
      "grad_norm": 0.84765625,
      "learning_rate": 6.170643909769291e-06,
      "loss": 1.2392,
      "step": 55207
    },
    {
      "epoch": 4.749892454615848,
      "grad_norm": 1.28125,
      "learning_rate": 6.166409473634505e-06,
      "loss": 0.9336,
      "step": 55208
    },
    {
      "epoch": 4.749978490923169,
      "grad_norm": 0.84765625,
      "learning_rate": 6.162176481866955e-06,
      "loss": 0.9505,
      "step": 55209
    },
    {
      "epoch": 4.750064527230491,
      "grad_norm": 1.5703125,
      "learning_rate": 6.1579449344789096e-06,
      "loss": 1.4865,
      "step": 55210
    },
    {
      "epoch": 4.750150563537813,
      "grad_norm": 1.703125,
      "learning_rate": 6.153714831482915e-06,
      "loss": 1.5701,
      "step": 55211
    },
    {
      "epoch": 4.750236599845135,
      "grad_norm": 1.1015625,
      "learning_rate": 6.149486172891239e-06,
      "loss": 1.3364,
      "step": 55212
    },
    {
      "epoch": 4.750322636152457,
      "grad_norm": 0.91015625,
      "learning_rate": 6.145258958716204e-06,
      "loss": 1.1599,
      "step": 55213
    },
    {
      "epoch": 4.750408672459778,
      "grad_norm": 1.34375,
      "learning_rate": 6.141033188970302e-06,
      "loss": 1.6581,
      "step": 55214
    },
    {
      "epoch": 4.750408672459778,
      "eval_loss": 1.06758451461792,
      "eval_runtime": 388.7278,
      "eval_samples_per_second": 11.96,
      "eval_steps_per_second": 1.497,
      "step": 55214
    },
    {
      "epoch": 4.7504947087671,
      "grad_norm": 4.4375,
      "learning_rate": 6.136808863665799e-06,
      "loss": 0.6036,
      "step": 55215
    },
    {
      "epoch": 4.750580745074421,
      "grad_norm": 1.3359375,
      "learning_rate": 6.132585982815076e-06,
      "loss": 0.675,
      "step": 55216
    },
    {
      "epoch": 4.750666781381743,
      "grad_norm": 1.3515625,
      "learning_rate": 6.128364546430509e-06,
      "loss": 1.1081,
      "step": 55217
    },
    {
      "epoch": 4.750752817689065,
      "grad_norm": 1.6015625,
      "learning_rate": 6.124144554524424e-06,
      "loss": 1.0574,
      "step": 55218
    },
    {
      "epoch": 4.750838853996386,
      "grad_norm": 1.09375,
      "learning_rate": 6.119926007109145e-06,
      "loss": 0.7763,
      "step": 55219
    },
    {
      "epoch": 4.7509248903037085,
      "grad_norm": 0.96484375,
      "learning_rate": 6.1157089041970485e-06,
      "loss": 0.6606,
      "step": 55220
    },
    {
      "epoch": 4.75101092661103,
      "grad_norm": 1.2890625,
      "learning_rate": 6.111493245800404e-06,
      "loss": 1.1187,
      "step": 55221
    },
    {
      "epoch": 4.751096962918352,
      "grad_norm": 1.15625,
      "learning_rate": 6.107279031931645e-06,
      "loss": 0.5707,
      "step": 55222
    },
    {
      "epoch": 4.751182999225673,
      "grad_norm": 1.6640625,
      "learning_rate": 6.103066262602986e-06,
      "loss": 1.4553,
      "step": 55223
    },
    {
      "epoch": 4.751269035532995,
      "grad_norm": 8.75,
      "learning_rate": 6.098854937826803e-06,
      "loss": 2.0094,
      "step": 55224
    },
    {
      "epoch": 4.751355071840317,
      "grad_norm": 0.87109375,
      "learning_rate": 6.094645057615477e-06,
      "loss": 1.2226,
      "step": 55225
    },
    {
      "epoch": 4.751441108147638,
      "grad_norm": 1.078125,
      "learning_rate": 6.0904366219812215e-06,
      "loss": 1.0902,
      "step": 55226
    },
    {
      "epoch": 4.75152714445496,
      "grad_norm": 1.1953125,
      "learning_rate": 6.086229630936357e-06,
      "loss": 0.6439,
      "step": 55227
    },
    {
      "epoch": 4.7516131807622815,
      "grad_norm": 3.9375,
      "learning_rate": 6.08202408449321e-06,
      "loss": 1.0469,
      "step": 55228
    },
    {
      "epoch": 4.7516992170696035,
      "grad_norm": 0.91015625,
      "learning_rate": 6.077819982664101e-06,
      "loss": 1.0172,
      "step": 55229
    },
    {
      "epoch": 4.751785253376925,
      "grad_norm": 0.97265625,
      "learning_rate": 6.0736173254613e-06,
      "loss": 1.3529,
      "step": 55230
    },
    {
      "epoch": 4.751871289684247,
      "grad_norm": 1.078125,
      "learning_rate": 6.069416112897075e-06,
      "loss": 1.3759,
      "step": 55231
    },
    {
      "epoch": 4.751957325991569,
      "grad_norm": 1.46875,
      "learning_rate": 6.065216344983804e-06,
      "loss": 1.2436,
      "step": 55232
    },
    {
      "epoch": 4.75204336229889,
      "grad_norm": 1.0234375,
      "learning_rate": 6.0610180217337e-06,
      "loss": 1.0345,
      "step": 55233
    },
    {
      "epoch": 4.752129398606212,
      "grad_norm": 1.4609375,
      "learning_rate": 6.056821143158975e-06,
      "loss": 1.6622,
      "step": 55234
    },
    {
      "epoch": 4.752215434913533,
      "grad_norm": 1.125,
      "learning_rate": 6.052625709272064e-06,
      "loss": 1.3542,
      "step": 55235
    },
    {
      "epoch": 4.752301471220855,
      "grad_norm": 1.2265625,
      "learning_rate": 6.048431720085124e-06,
      "loss": 1.2641,
      "step": 55236
    },
    {
      "epoch": 4.7523875075281765,
      "grad_norm": 1.0703125,
      "learning_rate": 6.044239175610422e-06,
      "loss": 0.783,
      "step": 55237
    },
    {
      "epoch": 4.752473543835499,
      "grad_norm": 1.0,
      "learning_rate": 6.040048075860283e-06,
      "loss": 1.064,
      "step": 55238
    },
    {
      "epoch": 4.752559580142821,
      "grad_norm": 1.234375,
      "learning_rate": 6.035858420846973e-06,
      "loss": 0.7564,
      "step": 55239
    },
    {
      "epoch": 4.752645616450142,
      "grad_norm": 0.890625,
      "learning_rate": 6.0316702105826514e-06,
      "loss": 0.7437,
      "step": 55240
    },
    {
      "epoch": 4.752731652757464,
      "grad_norm": 0.91015625,
      "learning_rate": 6.02748344507964e-06,
      "loss": 0.4478,
      "step": 55241
    },
    {
      "epoch": 4.752817689064785,
      "grad_norm": 0.89453125,
      "learning_rate": 6.0232981243502074e-06,
      "loss": 1.0656,
      "step": 55242
    },
    {
      "epoch": 4.752903725372107,
      "grad_norm": 1.1484375,
      "learning_rate": 6.019114248406454e-06,
      "loss": 1.7208,
      "step": 55243
    },
    {
      "epoch": 4.752989761679428,
      "grad_norm": 8.875,
      "learning_rate": 6.014931817260816e-06,
      "loss": 1.4168,
      "step": 55244
    },
    {
      "epoch": 4.75307579798675,
      "grad_norm": 1.0078125,
      "learning_rate": 6.0107508309253935e-06,
      "loss": 0.652,
      "step": 55245
    },
    {
      "epoch": 4.7531618342940725,
      "grad_norm": 1.265625,
      "learning_rate": 6.006571289412399e-06,
      "loss": 1.3645,
      "step": 55246
    },
    {
      "epoch": 4.753247870601394,
      "grad_norm": 1.03125,
      "learning_rate": 6.002393192734213e-06,
      "loss": 0.9395,
      "step": 55247
    },
    {
      "epoch": 4.753333906908716,
      "grad_norm": 1.1328125,
      "learning_rate": 5.9982165409028786e-06,
      "loss": 1.6433,
      "step": 55248
    },
    {
      "epoch": 4.753419943216037,
      "grad_norm": 0.87109375,
      "learning_rate": 5.994041333930667e-06,
      "loss": 0.5085,
      "step": 55249
    },
    {
      "epoch": 4.753505979523359,
      "grad_norm": 1.796875,
      "learning_rate": 5.9898675718298435e-06,
      "loss": 0.9031,
      "step": 55250
    },
    {
      "epoch": 4.75359201583068,
      "grad_norm": 3.953125,
      "learning_rate": 5.985695254612567e-06,
      "loss": 1.3994,
      "step": 55251
    },
    {
      "epoch": 4.753678052138002,
      "grad_norm": 1.3984375,
      "learning_rate": 5.981524382291048e-06,
      "loss": 1.7156,
      "step": 55252
    },
    {
      "epoch": 4.753764088445324,
      "grad_norm": 13.5625,
      "learning_rate": 5.977354954877501e-06,
      "loss": 1.2513,
      "step": 55253
    },
    {
      "epoch": 4.7538501247526455,
      "grad_norm": 2304.0,
      "learning_rate": 5.973186972384082e-06,
      "loss": 1.4111,
      "step": 55254
    },
    {
      "epoch": 4.7539361610599675,
      "grad_norm": 0.76953125,
      "learning_rate": 5.969020434823058e-06,
      "loss": 0.7296,
      "step": 55255
    },
    {
      "epoch": 4.754022197367289,
      "grad_norm": 1.1171875,
      "learning_rate": 5.964855342206476e-06,
      "loss": 0.9663,
      "step": 55256
    },
    {
      "epoch": 4.754108233674611,
      "grad_norm": 1.1171875,
      "learning_rate": 5.960691694546716e-06,
      "loss": 0.7171,
      "step": 55257
    },
    {
      "epoch": 4.754194269981932,
      "grad_norm": 1.4375,
      "learning_rate": 5.956529491855767e-06,
      "loss": 0.8371,
      "step": 55258
    },
    {
      "epoch": 4.754280306289254,
      "grad_norm": 2.203125,
      "learning_rate": 5.952368734145897e-06,
      "loss": 1.1465,
      "step": 55259
    },
    {
      "epoch": 4.754366342596576,
      "grad_norm": 1.390625,
      "learning_rate": 5.948209421429318e-06,
      "loss": 1.072,
      "step": 55260
    },
    {
      "epoch": 4.754452378903897,
      "grad_norm": 1.0,
      "learning_rate": 5.9440515537180776e-06,
      "loss": 1.4277,
      "step": 55261
    },
    {
      "epoch": 4.754538415211219,
      "grad_norm": 83.5,
      "learning_rate": 5.939895131024442e-06,
      "loss": 1.5443,
      "step": 55262
    },
    {
      "epoch": 4.7546244515185405,
      "grad_norm": 0.9765625,
      "learning_rate": 5.935740153360513e-06,
      "loss": 1.0744,
      "step": 55263
    },
    {
      "epoch": 4.754710487825863,
      "grad_norm": 1.1328125,
      "learning_rate": 5.931586620738394e-06,
      "loss": 0.7753,
      "step": 55264
    },
    {
      "epoch": 4.754796524133184,
      "grad_norm": 1.1875,
      "learning_rate": 5.92743453317035e-06,
      "loss": 0.6377,
      "step": 55265
    },
    {
      "epoch": 4.754882560440506,
      "grad_norm": 0.9453125,
      "learning_rate": 5.923283890668485e-06,
      "loss": 0.7568,
      "step": 55266
    },
    {
      "epoch": 4.754968596747828,
      "grad_norm": 39.5,
      "learning_rate": 5.9191346932448986e-06,
      "loss": 0.911,
      "step": 55267
    },
    {
      "epoch": 4.755054633055149,
      "grad_norm": 0.92578125,
      "learning_rate": 5.914986940911693e-06,
      "loss": 0.7667,
      "step": 55268
    },
    {
      "epoch": 4.755140669362471,
      "grad_norm": 1.0703125,
      "learning_rate": 5.910840633681136e-06,
      "loss": 0.7905,
      "step": 55269
    },
    {
      "epoch": 4.755226705669792,
      "grad_norm": 1.1796875,
      "learning_rate": 5.906695771565274e-06,
      "loss": 0.9646,
      "step": 55270
    },
    {
      "epoch": 4.755312741977114,
      "grad_norm": 0.91015625,
      "learning_rate": 5.9025523545762075e-06,
      "loss": 0.7377,
      "step": 55271
    },
    {
      "epoch": 4.755398778284436,
      "grad_norm": 1.1328125,
      "learning_rate": 5.8984103827260935e-06,
      "loss": 0.7524,
      "step": 55272
    },
    {
      "epoch": 4.755484814591758,
      "grad_norm": 0.9140625,
      "learning_rate": 5.894269856026979e-06,
      "loss": 0.6213,
      "step": 55273
    },
    {
      "epoch": 4.75557085089908,
      "grad_norm": 1.09375,
      "learning_rate": 5.890130774491076e-06,
      "loss": 1.1726,
      "step": 55274
    },
    {
      "epoch": 4.755656887206401,
      "grad_norm": 1.359375,
      "learning_rate": 5.88599313813043e-06,
      "loss": 0.632,
      "step": 55275
    },
    {
      "epoch": 4.755742923513723,
      "grad_norm": 1.2578125,
      "learning_rate": 5.881856946957143e-06,
      "loss": 1.3165,
      "step": 55276
    },
    {
      "epoch": 4.755828959821044,
      "grad_norm": 1.140625,
      "learning_rate": 5.877722200983371e-06,
      "loss": 1.224,
      "step": 55277
    },
    {
      "epoch": 4.755914996128366,
      "grad_norm": 1.171875,
      "learning_rate": 5.873588900221105e-06,
      "loss": 1.5041,
      "step": 55278
    },
    {
      "epoch": 4.756001032435687,
      "grad_norm": 1.3671875,
      "learning_rate": 5.869457044682503e-06,
      "loss": 1.565,
      "step": 55279
    },
    {
      "epoch": 4.7560870687430095,
      "grad_norm": 1.0390625,
      "learning_rate": 5.865326634379664e-06,
      "loss": 0.9079,
      "step": 55280
    },
    {
      "epoch": 4.7561731050503315,
      "grad_norm": 1.046875,
      "learning_rate": 5.86119766932458e-06,
      "loss": 0.85,
      "step": 55281
    },
    {
      "epoch": 4.756259141357653,
      "grad_norm": 1.265625,
      "learning_rate": 5.8570701495294644e-06,
      "loss": 0.6175,
      "step": 55282
    },
    {
      "epoch": 4.756345177664975,
      "grad_norm": 1.078125,
      "learning_rate": 5.8529440750062505e-06,
      "loss": 1.7399,
      "step": 55283
    },
    {
      "epoch": 4.756431213972296,
      "grad_norm": 1.0703125,
      "learning_rate": 5.848819445767095e-06,
      "loss": 1.1568,
      "step": 55284
    },
    {
      "epoch": 4.756517250279618,
      "grad_norm": 1.4140625,
      "learning_rate": 5.844696261824045e-06,
      "loss": 1.0281,
      "step": 55285
    },
    {
      "epoch": 4.756603286586939,
      "grad_norm": 1.1171875,
      "learning_rate": 5.840574523189146e-06,
      "loss": 0.7797,
      "step": 55286
    },
    {
      "epoch": 4.756689322894261,
      "grad_norm": 1.25,
      "learning_rate": 5.836454229874444e-06,
      "loss": 1.7033,
      "step": 55287
    },
    {
      "epoch": 4.756775359201583,
      "grad_norm": 1.03125,
      "learning_rate": 5.83233538189204e-06,
      "loss": 0.9871,
      "step": 55288
    },
    {
      "epoch": 4.7568613955089045,
      "grad_norm": 1.2734375,
      "learning_rate": 5.828217979253869e-06,
      "loss": 0.6596,
      "step": 55289
    },
    {
      "epoch": 4.756947431816227,
      "grad_norm": 5.0,
      "learning_rate": 5.824102021972088e-06,
      "loss": 1.8743,
      "step": 55290
    },
    {
      "epoch": 4.757033468123548,
      "grad_norm": 0.98828125,
      "learning_rate": 5.819987510058688e-06,
      "loss": 0.5775,
      "step": 55291
    },
    {
      "epoch": 4.75711950443087,
      "grad_norm": 1.015625,
      "learning_rate": 5.815874443525715e-06,
      "loss": 0.7073,
      "step": 55292
    },
    {
      "epoch": 4.757205540738191,
      "grad_norm": 0.96484375,
      "learning_rate": 5.8117628223852696e-06,
      "loss": 1.4919,
      "step": 55293
    },
    {
      "epoch": 4.757291577045513,
      "grad_norm": 1.125,
      "learning_rate": 5.807652646649231e-06,
      "loss": 1.196,
      "step": 55294
    },
    {
      "epoch": 4.757377613352835,
      "grad_norm": 1.359375,
      "learning_rate": 5.803543916329646e-06,
      "loss": 1.4619,
      "step": 55295
    },
    {
      "epoch": 4.757463649660156,
      "grad_norm": 0.88671875,
      "learning_rate": 5.79943663143867e-06,
      "loss": 1.1559,
      "step": 55296
    },
    {
      "epoch": 4.757549685967478,
      "grad_norm": 0.73828125,
      "learning_rate": 5.795330791988185e-06,
      "loss": 0.7244,
      "step": 55297
    },
    {
      "epoch": 4.7576357222748,
      "grad_norm": 1.0078125,
      "learning_rate": 5.7912263979902345e-06,
      "loss": 0.7455,
      "step": 55298
    },
    {
      "epoch": 4.757721758582122,
      "grad_norm": 1.21875,
      "learning_rate": 5.787123449456866e-06,
      "loss": 1.1655,
      "step": 55299
    },
    {
      "epoch": 4.757807794889444,
      "grad_norm": 1.2265625,
      "learning_rate": 5.783021946400013e-06,
      "loss": 1.4439,
      "step": 55300
    },
    {
      "epoch": 4.757893831196765,
      "grad_norm": 0.99609375,
      "learning_rate": 5.778921888831723e-06,
      "loss": 1.5454,
      "step": 55301
    },
    {
      "epoch": 4.757979867504087,
      "grad_norm": 1.0859375,
      "learning_rate": 5.77482327676393e-06,
      "loss": 1.364,
      "step": 55302
    },
    {
      "epoch": 4.758065903811408,
      "grad_norm": 1.53125,
      "learning_rate": 5.77072611020868e-06,
      "loss": 2.1383,
      "step": 55303
    },
    {
      "epoch": 4.75815194011873,
      "grad_norm": 0.81640625,
      "learning_rate": 5.7666303891779625e-06,
      "loss": 1.1083,
      "step": 55304
    },
    {
      "epoch": 4.758237976426052,
      "grad_norm": 2.40625,
      "learning_rate": 5.762536113683658e-06,
      "loss": 0.9094,
      "step": 55305
    },
    {
      "epoch": 4.7583240127333735,
      "grad_norm": 1.1171875,
      "learning_rate": 5.758443283737868e-06,
      "loss": 1.0259,
      "step": 55306
    },
    {
      "epoch": 4.7584100490406955,
      "grad_norm": 1.1015625,
      "learning_rate": 5.754351899352528e-06,
      "loss": 0.8765,
      "step": 55307
    },
    {
      "epoch": 4.758496085348017,
      "grad_norm": 1.2265625,
      "learning_rate": 5.750261960539516e-06,
      "loss": 1.3308,
      "step": 55308
    },
    {
      "epoch": 4.758582121655339,
      "grad_norm": 1.0078125,
      "learning_rate": 5.746173467310878e-06,
      "loss": 1.4204,
      "step": 55309
    },
    {
      "epoch": 4.75866815796266,
      "grad_norm": 1.625,
      "learning_rate": 5.742086419678605e-06,
      "loss": 1.436,
      "step": 55310
    },
    {
      "epoch": 4.758754194269982,
      "grad_norm": 0.875,
      "learning_rate": 5.738000817654521e-06,
      "loss": 0.4874,
      "step": 55311
    },
    {
      "epoch": 4.758840230577304,
      "grad_norm": 1.1640625,
      "learning_rate": 5.7339166612507265e-06,
      "loss": 1.9786,
      "step": 55312
    },
    {
      "epoch": 4.758926266884625,
      "grad_norm": 1.015625,
      "learning_rate": 5.7298339504790466e-06,
      "loss": 0.735,
      "step": 55313
    },
    {
      "epoch": 4.759012303191947,
      "grad_norm": 2.53125,
      "learning_rate": 5.725752685351471e-06,
      "loss": 1.5345,
      "step": 55314
    },
    {
      "epoch": 4.7590983394992685,
      "grad_norm": 1.078125,
      "learning_rate": 5.72167286587999e-06,
      "loss": 1.5579,
      "step": 55315
    },
    {
      "epoch": 4.759184375806591,
      "grad_norm": 1.234375,
      "learning_rate": 5.717594492076484e-06,
      "loss": 1.0396,
      "step": 55316
    },
    {
      "epoch": 4.759270412113912,
      "grad_norm": 1.21875,
      "learning_rate": 5.713517563952831e-06,
      "loss": 1.3939,
      "step": 55317
    },
    {
      "epoch": 4.759356448421234,
      "grad_norm": 0.96484375,
      "learning_rate": 5.709442081521021e-06,
      "loss": 0.7198,
      "step": 55318
    },
    {
      "epoch": 4.759442484728556,
      "grad_norm": 1.1875,
      "learning_rate": 5.7053680447929916e-06,
      "loss": 0.8451,
      "step": 55319
    },
    {
      "epoch": 4.759528521035877,
      "grad_norm": 0.95703125,
      "learning_rate": 5.701295453780564e-06,
      "loss": 0.7016,
      "step": 55320
    },
    {
      "epoch": 4.759614557343199,
      "grad_norm": 1.3203125,
      "learning_rate": 5.697224308495785e-06,
      "loss": 1.4745,
      "step": 55321
    },
    {
      "epoch": 4.75970059365052,
      "grad_norm": 0.94140625,
      "learning_rate": 5.693154608950424e-06,
      "loss": 0.7596,
      "step": 55322
    },
    {
      "epoch": 4.759786629957842,
      "grad_norm": 0.859375,
      "learning_rate": 5.68908635515647e-06,
      "loss": 0.9171,
      "step": 55323
    },
    {
      "epoch": 4.759872666265164,
      "grad_norm": 1.3984375,
      "learning_rate": 5.685019547125803e-06,
      "loss": 1.6901,
      "step": 55324
    },
    {
      "epoch": 4.759958702572486,
      "grad_norm": 0.95703125,
      "learning_rate": 5.680954184870302e-06,
      "loss": 0.7466,
      "step": 55325
    },
    {
      "epoch": 4.760044738879808,
      "grad_norm": 3.6875,
      "learning_rate": 5.6768902684019015e-06,
      "loss": 0.6109,
      "step": 55326
    },
    {
      "epoch": 4.760130775187129,
      "grad_norm": 1.4921875,
      "learning_rate": 5.672827797732427e-06,
      "loss": 1.8386,
      "step": 55327
    },
    {
      "epoch": 4.760216811494451,
      "grad_norm": 1.2734375,
      "learning_rate": 5.668766772873813e-06,
      "loss": 0.548,
      "step": 55328
    },
    {
      "epoch": 4.760302847801772,
      "grad_norm": 1.3046875,
      "learning_rate": 5.664707193837881e-06,
      "loss": 1.4742,
      "step": 55329
    },
    {
      "epoch": 4.760388884109094,
      "grad_norm": 0.9765625,
      "learning_rate": 5.660649060636569e-06,
      "loss": 0.7893,
      "step": 55330
    },
    {
      "epoch": 4.760474920416415,
      "grad_norm": 0.88671875,
      "learning_rate": 5.656592373281699e-06,
      "loss": 0.9218,
      "step": 55331
    },
    {
      "epoch": 4.7605609567237375,
      "grad_norm": 1.1640625,
      "learning_rate": 5.652537131785152e-06,
      "loss": 0.9385,
      "step": 55332
    },
    {
      "epoch": 4.7606469930310595,
      "grad_norm": 0.984375,
      "learning_rate": 5.648483336158805e-06,
      "loss": 0.7724,
      "step": 55333
    },
    {
      "epoch": 4.760733029338381,
      "grad_norm": 1.34375,
      "learning_rate": 5.644430986414484e-06,
      "loss": 1.1587,
      "step": 55334
    },
    {
      "epoch": 4.760819065645703,
      "grad_norm": 1.4609375,
      "learning_rate": 5.640380082564067e-06,
      "loss": 1.5263,
      "step": 55335
    },
    {
      "epoch": 4.760905101953024,
      "grad_norm": 1.296875,
      "learning_rate": 5.636330624619379e-06,
      "loss": 1.8786,
      "step": 55336
    },
    {
      "epoch": 4.760991138260346,
      "grad_norm": 80.0,
      "learning_rate": 5.632282612592299e-06,
      "loss": 1.0004,
      "step": 55337
    },
    {
      "epoch": 4.761077174567667,
      "grad_norm": 0.6875,
      "learning_rate": 5.628236046494595e-06,
      "loss": 0.6096,
      "step": 55338
    },
    {
      "epoch": 4.761163210874989,
      "grad_norm": 1.0078125,
      "learning_rate": 5.624190926338202e-06,
      "loss": 0.9222,
      "step": 55339
    },
    {
      "epoch": 4.761249247182311,
      "grad_norm": 1.0390625,
      "learning_rate": 5.620147252134889e-06,
      "loss": 0.5269,
      "step": 55340
    },
    {
      "epoch": 4.7613352834896325,
      "grad_norm": 1.3359375,
      "learning_rate": 5.616105023896478e-06,
      "loss": 2.1353,
      "step": 55341
    },
    {
      "epoch": 4.761421319796955,
      "grad_norm": 1.3203125,
      "learning_rate": 5.6120642416347955e-06,
      "loss": 1.0171,
      "step": 55342
    },
    {
      "epoch": 4.761507356104276,
      "grad_norm": 0.99609375,
      "learning_rate": 5.608024905361664e-06,
      "loss": 1.4982,
      "step": 55343
    },
    {
      "epoch": 4.761593392411598,
      "grad_norm": 1.2109375,
      "learning_rate": 5.603987015088963e-06,
      "loss": 1.9379,
      "step": 55344
    },
    {
      "epoch": 4.761679428718919,
      "grad_norm": 45.25,
      "learning_rate": 5.599950570828405e-06,
      "loss": 1.0684,
      "step": 55345
    },
    {
      "epoch": 4.761765465026241,
      "grad_norm": 0.765625,
      "learning_rate": 5.59591557259187e-06,
      "loss": 0.5932,
      "step": 55346
    },
    {
      "epoch": 4.761851501333563,
      "grad_norm": 0.9375,
      "learning_rate": 5.59188202039107e-06,
      "loss": 1.1886,
      "step": 55347
    },
    {
      "epoch": 4.761937537640884,
      "grad_norm": 6.25,
      "learning_rate": 5.5878499142378855e-06,
      "loss": 0.7402,
      "step": 55348
    },
    {
      "epoch": 4.762023573948206,
      "grad_norm": 1.0078125,
      "learning_rate": 5.583819254144085e-06,
      "loss": 1.2689,
      "step": 55349
    },
    {
      "epoch": 4.762109610255528,
      "grad_norm": 2.140625,
      "learning_rate": 5.5797900401214355e-06,
      "loss": 0.9643,
      "step": 55350
    },
    {
      "epoch": 4.76219564656285,
      "grad_norm": 1.328125,
      "learning_rate": 5.575762272181761e-06,
      "loss": 1.2322,
      "step": 55351
    },
    {
      "epoch": 4.762281682870171,
      "grad_norm": 1.25,
      "learning_rate": 5.571735950336832e-06,
      "loss": 1.1702,
      "step": 55352
    },
    {
      "epoch": 4.762367719177493,
      "grad_norm": 1.1015625,
      "learning_rate": 5.567711074598414e-06,
      "loss": 0.9971,
      "step": 55353
    },
    {
      "epoch": 4.762453755484815,
      "grad_norm": 584.0,
      "learning_rate": 5.563687644978277e-06,
      "loss": 0.7673,
      "step": 55354
    },
    {
      "epoch": 4.762539791792136,
      "grad_norm": 1.0390625,
      "learning_rate": 5.559665661488133e-06,
      "loss": 0.6335,
      "step": 55355
    },
    {
      "epoch": 4.762625828099458,
      "grad_norm": 0.99609375,
      "learning_rate": 5.555645124139918e-06,
      "loss": 1.1695,
      "step": 55356
    },
    {
      "epoch": 4.762711864406779,
      "grad_norm": 1.140625,
      "learning_rate": 5.5516260329451786e-06,
      "loss": 1.1357,
      "step": 55357
    },
    {
      "epoch": 4.7627979007141015,
      "grad_norm": 0.984375,
      "learning_rate": 5.547608387915793e-06,
      "loss": 0.5799,
      "step": 55358
    },
    {
      "epoch": 4.762883937021423,
      "grad_norm": 44.0,
      "learning_rate": 5.543592189063528e-06,
      "loss": 0.5955,
      "step": 55359
    },
    {
      "epoch": 4.762969973328745,
      "grad_norm": 1.9296875,
      "learning_rate": 5.539577436400045e-06,
      "loss": 1.2685,
      "step": 55360
    },
    {
      "epoch": 4.763056009636067,
      "grad_norm": 36.75,
      "learning_rate": 5.53556412993711e-06,
      "loss": 0.781,
      "step": 55361
    },
    {
      "epoch": 4.763142045943388,
      "grad_norm": 1.0234375,
      "learning_rate": 5.5315522696865465e-06,
      "loss": 1.1364,
      "step": 55362
    },
    {
      "epoch": 4.76322808225071,
      "grad_norm": 2.78125,
      "learning_rate": 5.527541855659956e-06,
      "loss": 1.5194,
      "step": 55363
    },
    {
      "epoch": 4.763314118558031,
      "grad_norm": 0.93359375,
      "learning_rate": 5.523532887869165e-06,
      "loss": 0.6848,
      "step": 55364
    },
    {
      "epoch": 4.763400154865353,
      "grad_norm": 1.078125,
      "learning_rate": 5.519525366325883e-06,
      "loss": 0.8464,
      "step": 55365
    },
    {
      "epoch": 4.7634861911726745,
      "grad_norm": 0.97265625,
      "learning_rate": 5.51551929104177e-06,
      "loss": 1.0504,
      "step": 55366
    },
    {
      "epoch": 4.7635722274799965,
      "grad_norm": 0.921875,
      "learning_rate": 5.51151466202865e-06,
      "loss": 1.0135,
      "step": 55367
    },
    {
      "epoch": 4.763658263787319,
      "grad_norm": 7.625,
      "learning_rate": 5.507511479298178e-06,
      "loss": 1.3338,
      "step": 55368
    },
    {
      "epoch": 4.76374430009464,
      "grad_norm": 0.921875,
      "learning_rate": 5.503509742862068e-06,
      "loss": 0.7673,
      "step": 55369
    },
    {
      "epoch": 4.763830336401962,
      "grad_norm": 0.91796875,
      "learning_rate": 5.499509452731977e-06,
      "loss": 0.9803,
      "step": 55370
    },
    {
      "epoch": 4.763916372709283,
      "grad_norm": 1.390625,
      "learning_rate": 5.495510608919618e-06,
      "loss": 1.4593,
      "step": 55371
    },
    {
      "epoch": 4.764002409016605,
      "grad_norm": 1.0703125,
      "learning_rate": 5.491513211436816e-06,
      "loss": 0.7526,
      "step": 55372
    },
    {
      "epoch": 4.764088445323926,
      "grad_norm": 0.87890625,
      "learning_rate": 5.4875172602950604e-06,
      "loss": 0.8366,
      "step": 55373
    },
    {
      "epoch": 4.764174481631248,
      "grad_norm": 0.95703125,
      "learning_rate": 5.4835227555061765e-06,
      "loss": 0.6048,
      "step": 55374
    },
    {
      "epoch": 4.76426051793857,
      "grad_norm": 1.125,
      "learning_rate": 5.479529697081875e-06,
      "loss": 1.1203,
      "step": 55375
    },
    {
      "epoch": 4.764346554245892,
      "grad_norm": 1.234375,
      "learning_rate": 5.475538085033649e-06,
      "loss": 1.1303,
      "step": 55376
    },
    {
      "epoch": 4.764432590553214,
      "grad_norm": 1.0234375,
      "learning_rate": 5.471547919373376e-06,
      "loss": 0.9249,
      "step": 55377
    },
    {
      "epoch": 4.764518626860535,
      "grad_norm": 1.078125,
      "learning_rate": 5.46755920011266e-06,
      "loss": 0.8538,
      "step": 55378
    },
    {
      "epoch": 4.764604663167857,
      "grad_norm": 1.0390625,
      "learning_rate": 5.4635719272631e-06,
      "loss": 1.1463,
      "step": 55379
    },
    {
      "epoch": 4.764690699475178,
      "grad_norm": 1.3984375,
      "learning_rate": 5.459586100836411e-06,
      "loss": 1.659,
      "step": 55380
    },
    {
      "epoch": 4.7647767357825,
      "grad_norm": 1.7890625,
      "learning_rate": 5.455601720844306e-06,
      "loss": 1.3277,
      "step": 55381
    },
    {
      "epoch": 4.764862772089822,
      "grad_norm": 1.0625,
      "learning_rate": 5.451618787298329e-06,
      "loss": 0.9375,
      "step": 55382
    },
    {
      "epoch": 4.764948808397143,
      "grad_norm": 0.87109375,
      "learning_rate": 5.447637300210195e-06,
      "loss": 0.473,
      "step": 55383
    },
    {
      "epoch": 4.7650348447044655,
      "grad_norm": 1.09375,
      "learning_rate": 5.443657259591506e-06,
      "loss": 0.8407,
      "step": 55384
    },
    {
      "epoch": 4.765120881011787,
      "grad_norm": 0.89453125,
      "learning_rate": 5.439678665453973e-06,
      "loss": 0.8515,
      "step": 55385
    },
    {
      "epoch": 4.765206917319109,
      "grad_norm": 1.171875,
      "learning_rate": 5.435701517809199e-06,
      "loss": 1.5195,
      "step": 55386
    },
    {
      "epoch": 4.76529295362643,
      "grad_norm": 1.0546875,
      "learning_rate": 5.431725816668786e-06,
      "loss": 1.0406,
      "step": 55387
    },
    {
      "epoch": 4.765378989933752,
      "grad_norm": 1.3046875,
      "learning_rate": 5.4277515620443915e-06,
      "loss": 1.8897,
      "step": 55388
    },
    {
      "epoch": 4.765465026241074,
      "grad_norm": 0.96484375,
      "learning_rate": 5.423778753947672e-06,
      "loss": 0.7692,
      "step": 55389
    },
    {
      "epoch": 4.765551062548395,
      "grad_norm": 2.109375,
      "learning_rate": 5.419807392390175e-06,
      "loss": 1.2969,
      "step": 55390
    },
    {
      "epoch": 4.765637098855717,
      "grad_norm": 1.3984375,
      "learning_rate": 5.415837477383556e-06,
      "loss": 0.929,
      "step": 55391
    },
    {
      "epoch": 4.7657231351630385,
      "grad_norm": 0.80859375,
      "learning_rate": 5.411869008939419e-06,
      "loss": 0.525,
      "step": 55392
    },
    {
      "epoch": 4.7658091714703605,
      "grad_norm": 237.0,
      "learning_rate": 5.4079019870693636e-06,
      "loss": 0.9729,
      "step": 55393
    },
    {
      "epoch": 4.765895207777682,
      "grad_norm": 0.88671875,
      "learning_rate": 5.403936411785049e-06,
      "loss": 0.9124,
      "step": 55394
    },
    {
      "epoch": 4.765981244085004,
      "grad_norm": 1.3984375,
      "learning_rate": 5.399972283097965e-06,
      "loss": 0.9525,
      "step": 55395
    },
    {
      "epoch": 4.766067280392326,
      "grad_norm": 1.3046875,
      "learning_rate": 5.396009601019769e-06,
      "loss": 1.6523,
      "step": 55396
    },
    {
      "epoch": 4.766153316699647,
      "grad_norm": 1.109375,
      "learning_rate": 5.392048365562119e-06,
      "loss": 1.1939,
      "step": 55397
    },
    {
      "epoch": 4.766239353006969,
      "grad_norm": 1.0078125,
      "learning_rate": 5.38808857673645e-06,
      "loss": 1.0942,
      "step": 55398
    },
    {
      "epoch": 4.76632538931429,
      "grad_norm": 1.125,
      "learning_rate": 5.384130234554474e-06,
      "loss": 0.7829,
      "step": 55399
    },
    {
      "epoch": 4.766411425621612,
      "grad_norm": 6.65625,
      "learning_rate": 5.380173339027683e-06,
      "loss": 1.4961,
      "step": 55400
    },
    {
      "epoch": 4.7664974619289335,
      "grad_norm": 1.6484375,
      "learning_rate": 5.376217890167734e-06,
      "loss": 0.9165,
      "step": 55401
    },
    {
      "epoch": 4.766583498236256,
      "grad_norm": 1.9453125,
      "learning_rate": 5.372263887986118e-06,
      "loss": 1.3251,
      "step": 55402
    },
    {
      "epoch": 4.766669534543578,
      "grad_norm": 10.4375,
      "learning_rate": 5.368311332494435e-06,
      "loss": 1.1148,
      "step": 55403
    },
    {
      "epoch": 4.766755570850899,
      "grad_norm": 0.984375,
      "learning_rate": 5.364360223704234e-06,
      "loss": 0.9067,
      "step": 55404
    },
    {
      "epoch": 4.766841607158221,
      "grad_norm": 1.53125,
      "learning_rate": 5.360410561627116e-06,
      "loss": 1.3705,
      "step": 55405
    },
    {
      "epoch": 4.766927643465542,
      "grad_norm": 1.1328125,
      "learning_rate": 5.356462346274571e-06,
      "loss": 1.6745,
      "step": 55406
    },
    {
      "epoch": 4.767013679772864,
      "grad_norm": 1.296875,
      "learning_rate": 5.3525155776581455e-06,
      "loss": 0.9298,
      "step": 55407
    },
    {
      "epoch": 4.767099716080185,
      "grad_norm": 1.09375,
      "learning_rate": 5.3485702557894425e-06,
      "loss": 0.5712,
      "step": 55408
    },
    {
      "epoch": 4.767185752387507,
      "grad_norm": 0.9140625,
      "learning_rate": 5.3446263806799515e-06,
      "loss": 0.7345,
      "step": 55409
    },
    {
      "epoch": 4.7672717886948295,
      "grad_norm": 1.4921875,
      "learning_rate": 5.3406839523412185e-06,
      "loss": 0.9959,
      "step": 55410
    },
    {
      "epoch": 4.767357825002151,
      "grad_norm": 1.109375,
      "learning_rate": 5.336742970784791e-06,
      "loss": 0.4669,
      "step": 55411
    },
    {
      "epoch": 4.767443861309473,
      "grad_norm": 1.0859375,
      "learning_rate": 5.33280343602216e-06,
      "loss": 1.2513,
      "step": 55412
    },
    {
      "epoch": 4.767529897616794,
      "grad_norm": 1.015625,
      "learning_rate": 5.328865348064926e-06,
      "loss": 0.8064,
      "step": 55413
    },
    {
      "epoch": 4.767615933924116,
      "grad_norm": 0.9765625,
      "learning_rate": 5.324928706924526e-06,
      "loss": 0.8003,
      "step": 55414
    },
    {
      "epoch": 4.767701970231438,
      "grad_norm": 1.640625,
      "learning_rate": 5.320993512612449e-06,
      "loss": 1.2067,
      "step": 55415
    },
    {
      "epoch": 4.767788006538759,
      "grad_norm": 1.0,
      "learning_rate": 5.317059765140353e-06,
      "loss": 0.7266,
      "step": 55416
    },
    {
      "epoch": 4.767874042846081,
      "grad_norm": 1.1875,
      "learning_rate": 5.313127464519563e-06,
      "loss": 1.0565,
      "step": 55417
    },
    {
      "epoch": 4.7679600791534025,
      "grad_norm": 0.97265625,
      "learning_rate": 5.309196610761679e-06,
      "loss": 1.1374,
      "step": 55418
    },
    {
      "epoch": 4.7680461154607245,
      "grad_norm": 1.0390625,
      "learning_rate": 5.305267203878195e-06,
      "loss": 0.7087,
      "step": 55419
    },
    {
      "epoch": 4.768132151768047,
      "grad_norm": 15.8125,
      "learning_rate": 5.301339243880599e-06,
      "loss": 1.0335,
      "step": 55420
    },
    {
      "epoch": 4.768218188075368,
      "grad_norm": 4.90625,
      "learning_rate": 5.297412730780382e-06,
      "loss": 1.1948,
      "step": 55421
    },
    {
      "epoch": 4.76830422438269,
      "grad_norm": 1.140625,
      "learning_rate": 5.2934876645889805e-06,
      "loss": 1.5291,
      "step": 55422
    },
    {
      "epoch": 4.768390260690011,
      "grad_norm": 1.234375,
      "learning_rate": 5.289564045317941e-06,
      "loss": 1.9094,
      "step": 55423
    },
    {
      "epoch": 4.768476296997333,
      "grad_norm": 14.875,
      "learning_rate": 5.285641872978753e-06,
      "loss": 0.7331,
      "step": 55424
    },
    {
      "epoch": 4.768562333304654,
      "grad_norm": 1.015625,
      "learning_rate": 5.281721147582796e-06,
      "loss": 0.9867,
      "step": 55425
    },
    {
      "epoch": 4.768648369611976,
      "grad_norm": 1.3203125,
      "learning_rate": 5.277801869141563e-06,
      "loss": 2.253,
      "step": 55426
    },
    {
      "epoch": 4.768734405919298,
      "grad_norm": 1.046875,
      "learning_rate": 5.273884037666599e-06,
      "loss": 0.7429,
      "step": 55427
    },
    {
      "epoch": 4.76882044222662,
      "grad_norm": 1.3046875,
      "learning_rate": 5.269967653169228e-06,
      "loss": 1.8703,
      "step": 55428
    },
    {
      "epoch": 4.768906478533942,
      "grad_norm": 1.0,
      "learning_rate": 5.266052715661052e-06,
      "loss": 0.741,
      "step": 55429
    },
    {
      "epoch": 4.768992514841263,
      "grad_norm": 1.1171875,
      "learning_rate": 5.2621392251534505e-06,
      "loss": 1.0693,
      "step": 55430
    },
    {
      "epoch": 4.769078551148585,
      "grad_norm": 0.953125,
      "learning_rate": 5.258227181657804e-06,
      "loss": 0.5299,
      "step": 55431
    },
    {
      "epoch": 4.769164587455906,
      "grad_norm": 1.296875,
      "learning_rate": 5.254316585185659e-06,
      "loss": 1.9273,
      "step": 55432
    },
    {
      "epoch": 4.769250623763228,
      "grad_norm": 1.59375,
      "learning_rate": 5.25040743574845e-06,
      "loss": 1.2848,
      "step": 55433
    },
    {
      "epoch": 4.76933666007055,
      "grad_norm": 18.625,
      "learning_rate": 5.2464997333575574e-06,
      "loss": 0.9767,
      "step": 55434
    },
    {
      "epoch": 4.769422696377871,
      "grad_norm": 87.0,
      "learning_rate": 5.2425934780244156e-06,
      "loss": 1.4335,
      "step": 55435
    },
    {
      "epoch": 4.7695087326851935,
      "grad_norm": 1.171875,
      "learning_rate": 5.238688669760405e-06,
      "loss": 0.8059,
      "step": 55436
    },
    {
      "epoch": 4.769594768992515,
      "grad_norm": 1.2265625,
      "learning_rate": 5.2347853085770724e-06,
      "loss": 1.1416,
      "step": 55437
    },
    {
      "epoch": 4.769680805299837,
      "grad_norm": 1.3671875,
      "learning_rate": 5.230883394485797e-06,
      "loss": 1.3836,
      "step": 55438
    },
    {
      "epoch": 4.769766841607158,
      "grad_norm": 1.3828125,
      "learning_rate": 5.2269829274979034e-06,
      "loss": 1.5569,
      "step": 55439
    },
    {
      "epoch": 4.76985287791448,
      "grad_norm": 98816.0,
      "learning_rate": 5.223083907624882e-06,
      "loss": 0.9217,
      "step": 55440
    },
    {
      "epoch": 4.769938914221802,
      "grad_norm": 0.94140625,
      "learning_rate": 5.219186334878112e-06,
      "loss": 1.0697,
      "step": 55441
    },
    {
      "epoch": 4.770024950529123,
      "grad_norm": 1.015625,
      "learning_rate": 5.215290209268975e-06,
      "loss": 0.8028,
      "step": 55442
    },
    {
      "epoch": 4.770110986836445,
      "grad_norm": 0.8671875,
      "learning_rate": 5.211395530808905e-06,
      "loss": 0.6482,
      "step": 55443
    },
    {
      "epoch": 4.7701970231437665,
      "grad_norm": 1.3515625,
      "learning_rate": 5.207502299509282e-06,
      "loss": 1.4534,
      "step": 55444
    },
    {
      "epoch": 4.7702830594510885,
      "grad_norm": 0.94921875,
      "learning_rate": 5.20361051538143e-06,
      "loss": 1.2306,
      "step": 55445
    },
    {
      "epoch": 4.77036909575841,
      "grad_norm": 0.9609375,
      "learning_rate": 5.19972017843684e-06,
      "loss": 0.553,
      "step": 55446
    },
    {
      "epoch": 4.770455132065732,
      "grad_norm": 1.140625,
      "learning_rate": 5.1958312886867805e-06,
      "loss": 0.8138,
      "step": 55447
    },
    {
      "epoch": 4.770541168373054,
      "grad_norm": 1.7890625,
      "learning_rate": 5.191943846142744e-06,
      "loss": 1.5718,
      "step": 55448
    },
    {
      "epoch": 4.770627204680375,
      "grad_norm": 0.84765625,
      "learning_rate": 5.188057850815997e-06,
      "loss": 0.6392,
      "step": 55449
    },
    {
      "epoch": 4.770713240987697,
      "grad_norm": 0.88671875,
      "learning_rate": 5.184173302717921e-06,
      "loss": 0.8278,
      "step": 55450
    },
    {
      "epoch": 4.770799277295018,
      "grad_norm": 1.640625,
      "learning_rate": 5.18029020185995e-06,
      "loss": 0.7032,
      "step": 55451
    },
    {
      "epoch": 4.77088531360234,
      "grad_norm": 0.99609375,
      "learning_rate": 5.176408548253409e-06,
      "loss": 0.639,
      "step": 55452
    },
    {
      "epoch": 4.7709713499096615,
      "grad_norm": 1.0703125,
      "learning_rate": 5.172528341909566e-06,
      "loss": 1.5531,
      "step": 55453
    },
    {
      "epoch": 4.771057386216984,
      "grad_norm": 57.25,
      "learning_rate": 5.168649582839912e-06,
      "loss": 1.6422,
      "step": 55454
    },
    {
      "epoch": 4.771143422524306,
      "grad_norm": 1.2578125,
      "learning_rate": 5.164772271055662e-06,
      "loss": 1.0904,
      "step": 55455
    },
    {
      "epoch": 4.771229458831627,
      "grad_norm": 1.0859375,
      "learning_rate": 5.16089640656825e-06,
      "loss": 1.6915,
      "step": 55456
    },
    {
      "epoch": 4.771315495138949,
      "grad_norm": 1.3515625,
      "learning_rate": 5.157021989388944e-06,
      "loss": 1.6361,
      "step": 55457
    },
    {
      "epoch": 4.77140153144627,
      "grad_norm": 0.93359375,
      "learning_rate": 5.1531490195291244e-06,
      "loss": 1.0825,
      "step": 55458
    },
    {
      "epoch": 4.771487567753592,
      "grad_norm": 1.0703125,
      "learning_rate": 5.149277497000115e-06,
      "loss": 1.4485,
      "step": 55459
    },
    {
      "epoch": 4.771573604060913,
      "grad_norm": 1.1875,
      "learning_rate": 5.145407421813186e-06,
      "loss": 1.9393,
      "step": 55460
    },
    {
      "epoch": 4.771659640368235,
      "grad_norm": 1.28125,
      "learning_rate": 5.141538793979717e-06,
      "loss": 0.942,
      "step": 55461
    },
    {
      "epoch": 4.7717456766755575,
      "grad_norm": 0.984375,
      "learning_rate": 5.1376716135110304e-06,
      "loss": 1.1003,
      "step": 55462
    },
    {
      "epoch": 4.771831712982879,
      "grad_norm": 1.5625,
      "learning_rate": 5.133805880418396e-06,
      "loss": 1.0759,
      "step": 55463
    },
    {
      "epoch": 4.771917749290201,
      "grad_norm": 17.25,
      "learning_rate": 5.129941594713084e-06,
      "loss": 0.9416,
      "step": 55464
    },
    {
      "epoch": 4.772003785597522,
      "grad_norm": 49.0,
      "learning_rate": 5.126078756406527e-06,
      "loss": 1.5248,
      "step": 55465
    },
    {
      "epoch": 4.772089821904844,
      "grad_norm": 1.203125,
      "learning_rate": 5.122217365509885e-06,
      "loss": 0.9472,
      "step": 55466
    },
    {
      "epoch": 4.772175858212165,
      "grad_norm": 1.0859375,
      "learning_rate": 5.118357422034536e-06,
      "loss": 0.912,
      "step": 55467
    },
    {
      "epoch": 4.772261894519487,
      "grad_norm": 1.1015625,
      "learning_rate": 5.11449892599175e-06,
      "loss": 1.2083,
      "step": 55468
    },
    {
      "epoch": 4.772347930826809,
      "grad_norm": 1.3671875,
      "learning_rate": 5.110641877392741e-06,
      "loss": 1.0432,
      "step": 55469
    },
    {
      "epoch": 4.7724339671341305,
      "grad_norm": 1.1953125,
      "learning_rate": 5.106786276248887e-06,
      "loss": 1.3597,
      "step": 55470
    },
    {
      "epoch": 4.7725200034414526,
      "grad_norm": 1.078125,
      "learning_rate": 5.102932122571458e-06,
      "loss": 0.8819,
      "step": 55471
    },
    {
      "epoch": 4.772606039748774,
      "grad_norm": 1.359375,
      "learning_rate": 5.099079416371666e-06,
      "loss": 2.2447,
      "step": 55472
    },
    {
      "epoch": 4.772692076056096,
      "grad_norm": 3.984375,
      "learning_rate": 5.095228157660836e-06,
      "loss": 1.4823,
      "step": 55473
    },
    {
      "epoch": 4.772778112363417,
      "grad_norm": 0.97265625,
      "learning_rate": 5.091378346450182e-06,
      "loss": 0.6648,
      "step": 55474
    },
    {
      "epoch": 4.772864148670739,
      "grad_norm": 0.8359375,
      "learning_rate": 5.0875299827510265e-06,
      "loss": 0.804,
      "step": 55475
    },
    {
      "epoch": 4.772950184978061,
      "grad_norm": 1.1875,
      "learning_rate": 5.083683066574585e-06,
      "loss": 1.4163,
      "step": 55476
    },
    {
      "epoch": 4.773036221285382,
      "grad_norm": 0.859375,
      "learning_rate": 5.079837597932124e-06,
      "loss": 1.6992,
      "step": 55477
    },
    {
      "epoch": 4.773122257592704,
      "grad_norm": 1.125,
      "learning_rate": 5.075993576834858e-06,
      "loss": 1.1807,
      "step": 55478
    },
    {
      "epoch": 4.7732082939000255,
      "grad_norm": 1.375,
      "learning_rate": 5.0721510032940565e-06,
      "loss": 0.85,
      "step": 55479
    },
    {
      "epoch": 4.773294330207348,
      "grad_norm": 1.0625,
      "learning_rate": 5.068309877320931e-06,
      "loss": 1.0396,
      "step": 55480
    },
    {
      "epoch": 4.773380366514669,
      "grad_norm": 1.09375,
      "learning_rate": 5.064470198926807e-06,
      "loss": 1.2122,
      "step": 55481
    },
    {
      "epoch": 4.773466402821991,
      "grad_norm": 1.140625,
      "learning_rate": 5.060631968122786e-06,
      "loss": 1.0229,
      "step": 55482
    },
    {
      "epoch": 4.773552439129313,
      "grad_norm": 10.25,
      "learning_rate": 5.056795184920193e-06,
      "loss": 1.7019,
      "step": 55483
    },
    {
      "epoch": 4.773638475436634,
      "grad_norm": 1.109375,
      "learning_rate": 5.052959849330241e-06,
      "loss": 1.1003,
      "step": 55484
    },
    {
      "epoch": 4.773724511743956,
      "grad_norm": 1.390625,
      "learning_rate": 5.049125961364032e-06,
      "loss": 0.9615,
      "step": 55485
    },
    {
      "epoch": 4.773810548051277,
      "grad_norm": 1.0390625,
      "learning_rate": 5.04529352103289e-06,
      "loss": 1.3966,
      "step": 55486
    },
    {
      "epoch": 4.773896584358599,
      "grad_norm": 1.1953125,
      "learning_rate": 5.041462528348084e-06,
      "loss": 1.223,
      "step": 55487
    },
    {
      "epoch": 4.773982620665921,
      "grad_norm": 1.3828125,
      "learning_rate": 5.037632983320661e-06,
      "loss": 0.6513,
      "step": 55488
    },
    {
      "epoch": 4.774068656973243,
      "grad_norm": 1.3046875,
      "learning_rate": 5.0338048859618904e-06,
      "loss": 1.0834,
      "step": 55489
    },
    {
      "epoch": 4.774154693280565,
      "grad_norm": 1.2890625,
      "learning_rate": 5.029978236283039e-06,
      "loss": 1.0324,
      "step": 55490
    },
    {
      "epoch": 4.774240729587886,
      "grad_norm": 1.203125,
      "learning_rate": 5.026153034295155e-06,
      "loss": 0.7754,
      "step": 55491
    },
    {
      "epoch": 4.774326765895208,
      "grad_norm": 3.6875,
      "learning_rate": 5.0223292800095635e-06,
      "loss": 1.3199,
      "step": 55492
    },
    {
      "epoch": 4.774412802202529,
      "grad_norm": 1.3515625,
      "learning_rate": 5.018506973437365e-06,
      "loss": 1.6571,
      "step": 55493
    },
    {
      "epoch": 4.774498838509851,
      "grad_norm": 6.84375,
      "learning_rate": 5.014686114589773e-06,
      "loss": 0.982,
      "step": 55494
    },
    {
      "epoch": 4.774584874817172,
      "grad_norm": 270336.0,
      "learning_rate": 5.010866703478001e-06,
      "loss": 0.7773,
      "step": 55495
    },
    {
      "epoch": 4.7746709111244945,
      "grad_norm": 1.015625,
      "learning_rate": 5.007048740113096e-06,
      "loss": 1.3267,
      "step": 55496
    },
    {
      "epoch": 4.774756947431817,
      "grad_norm": 1.3046875,
      "learning_rate": 5.0032322245063265e-06,
      "loss": 1.1572,
      "step": 55497
    },
    {
      "epoch": 4.774842983739138,
      "grad_norm": 1.0390625,
      "learning_rate": 4.999417156668851e-06,
      "loss": 0.8118,
      "step": 55498
    },
    {
      "epoch": 4.77492902004646,
      "grad_norm": 1.0703125,
      "learning_rate": 4.995603536611771e-06,
      "loss": 1.4536,
      "step": 55499
    },
    {
      "epoch": 4.775015056353781,
      "grad_norm": 1.46875,
      "learning_rate": 4.991791364346299e-06,
      "loss": 1.0352,
      "step": 55500
    },
    {
      "epoch": 4.775101092661103,
      "grad_norm": 0.92578125,
      "learning_rate": 4.9879806398835935e-06,
      "loss": 1.1947,
      "step": 55501
    },
    {
      "epoch": 4.775187128968424,
      "grad_norm": 26.75,
      "learning_rate": 4.984171363234702e-06,
      "loss": 1.0116,
      "step": 55502
    },
    {
      "epoch": 4.775273165275746,
      "grad_norm": 3.515625,
      "learning_rate": 4.980363534410892e-06,
      "loss": 1.4606,
      "step": 55503
    },
    {
      "epoch": 4.775359201583068,
      "grad_norm": 1.03125,
      "learning_rate": 4.976557153423266e-06,
      "loss": 0.9259,
      "step": 55504
    },
    {
      "epoch": 4.7754452378903895,
      "grad_norm": 1.03125,
      "learning_rate": 4.9727522202828725e-06,
      "loss": 0.8862,
      "step": 55505
    },
    {
      "epoch": 4.775531274197712,
      "grad_norm": 0.9140625,
      "learning_rate": 4.968948735000922e-06,
      "loss": 1.0789,
      "step": 55506
    },
    {
      "epoch": 4.775617310505033,
      "grad_norm": 37.5,
      "learning_rate": 4.965146697588518e-06,
      "loss": 1.1311,
      "step": 55507
    },
    {
      "epoch": 4.775703346812355,
      "grad_norm": 1.515625,
      "learning_rate": 4.961346108056763e-06,
      "loss": 1.0238,
      "step": 55508
    },
    {
      "epoch": 4.775789383119676,
      "grad_norm": 1.34375,
      "learning_rate": 4.957546966416815e-06,
      "loss": 1.9347,
      "step": 55509
    },
    {
      "epoch": 4.775875419426998,
      "grad_norm": 1.203125,
      "learning_rate": 4.95374927267972e-06,
      "loss": 1.3718,
      "step": 55510
    },
    {
      "epoch": 4.77596145573432,
      "grad_norm": 4.125,
      "learning_rate": 4.9499530268566905e-06,
      "loss": 0.9663,
      "step": 55511
    },
    {
      "epoch": 4.776047492041641,
      "grad_norm": 1.0,
      "learning_rate": 4.946158228958719e-06,
      "loss": 0.8933,
      "step": 55512
    },
    {
      "epoch": 4.776133528348963,
      "grad_norm": 1.5859375,
      "learning_rate": 4.942364878996963e-06,
      "loss": 1.6706,
      "step": 55513
    },
    {
      "epoch": 4.776219564656285,
      "grad_norm": 1.15625,
      "learning_rate": 4.93857297698258e-06,
      "loss": 1.2562,
      "step": 55514
    },
    {
      "epoch": 4.776305600963607,
      "grad_norm": 1.15625,
      "learning_rate": 4.934782522926506e-06,
      "loss": 1.1646,
      "step": 55515
    },
    {
      "epoch": 4.776391637270928,
      "grad_norm": 3.828125,
      "learning_rate": 4.930993516839899e-06,
      "loss": 1.4984,
      "step": 55516
    },
    {
      "epoch": 4.77647767357825,
      "grad_norm": 0.9765625,
      "learning_rate": 4.927205958733916e-06,
      "loss": 0.684,
      "step": 55517
    },
    {
      "epoch": 4.776563709885572,
      "grad_norm": 0.89453125,
      "learning_rate": 4.923419848619493e-06,
      "loss": 0.8469,
      "step": 55518
    },
    {
      "epoch": 4.776649746192893,
      "grad_norm": 1.3203125,
      "learning_rate": 4.919635186507787e-06,
      "loss": 1.747,
      "step": 55519
    },
    {
      "epoch": 4.776735782500215,
      "grad_norm": 1.453125,
      "learning_rate": 4.915851972409957e-06,
      "loss": 1.2426,
      "step": 55520
    },
    {
      "epoch": 4.776821818807536,
      "grad_norm": 1.328125,
      "learning_rate": 4.9120702063368824e-06,
      "loss": 1.3052,
      "step": 55521
    },
    {
      "epoch": 4.7769078551148585,
      "grad_norm": 1.140625,
      "learning_rate": 4.908289888299722e-06,
      "loss": 1.276,
      "step": 55522
    },
    {
      "epoch": 4.77699389142218,
      "grad_norm": 1.1875,
      "learning_rate": 4.904511018309521e-06,
      "loss": 1.0997,
      "step": 55523
    },
    {
      "epoch": 4.777079927729502,
      "grad_norm": 0.953125,
      "learning_rate": 4.900733596377383e-06,
      "loss": 0.4164,
      "step": 55524
    },
    {
      "epoch": 4.777165964036824,
      "grad_norm": 0.8359375,
      "learning_rate": 4.896957622514298e-06,
      "loss": 0.5429,
      "step": 55525
    },
    {
      "epoch": 4.777252000344145,
      "grad_norm": 1.0625,
      "learning_rate": 4.893183096731258e-06,
      "loss": 0.7712,
      "step": 55526
    },
    {
      "epoch": 4.777338036651467,
      "grad_norm": 1.1328125,
      "learning_rate": 4.889410019039419e-06,
      "loss": 1.0599,
      "step": 55527
    },
    {
      "epoch": 4.777424072958788,
      "grad_norm": 1.2734375,
      "learning_rate": 4.88563838944972e-06,
      "loss": 0.7547,
      "step": 55528
    },
    {
      "epoch": 4.77751010926611,
      "grad_norm": 1.1484375,
      "learning_rate": 4.881868207973261e-06,
      "loss": 0.8836,
      "step": 55529
    },
    {
      "epoch": 4.777596145573432,
      "grad_norm": 1.4296875,
      "learning_rate": 4.878099474621034e-06,
      "loss": 1.2995,
      "step": 55530
    },
    {
      "epoch": 4.7776821818807536,
      "grad_norm": 1.3046875,
      "learning_rate": 4.8743321894040845e-06,
      "loss": 1.3537,
      "step": 55531
    },
    {
      "epoch": 4.777768218188076,
      "grad_norm": 2.859375,
      "learning_rate": 4.870566352333405e-06,
      "loss": 1.167,
      "step": 55532
    },
    {
      "epoch": 4.777854254495397,
      "grad_norm": 2.265625,
      "learning_rate": 4.866801963420042e-06,
      "loss": 0.8942,
      "step": 55533
    },
    {
      "epoch": 4.777940290802719,
      "grad_norm": 0.90625,
      "learning_rate": 4.863039022674986e-06,
      "loss": 1.3309,
      "step": 55534
    },
    {
      "epoch": 4.77802632711004,
      "grad_norm": 0.98046875,
      "learning_rate": 4.85927753010923e-06,
      "loss": 1.282,
      "step": 55535
    },
    {
      "epoch": 4.778112363417362,
      "grad_norm": 1.0078125,
      "learning_rate": 4.855517485733763e-06,
      "loss": 0.8262,
      "step": 55536
    },
    {
      "epoch": 4.778198399724684,
      "grad_norm": 3.515625,
      "learning_rate": 4.851758889559632e-06,
      "loss": 1.2161,
      "step": 55537
    },
    {
      "epoch": 4.778284436032005,
      "grad_norm": 1.15625,
      "learning_rate": 4.848001741597774e-06,
      "loss": 1.3826,
      "step": 55538
    },
    {
      "epoch": 4.778370472339327,
      "grad_norm": 59.25,
      "learning_rate": 4.84424604185929e-06,
      "loss": 1.3841,
      "step": 55539
    },
    {
      "epoch": 4.778456508646649,
      "grad_norm": 0.84765625,
      "learning_rate": 4.8404917903550065e-06,
      "loss": 0.5185,
      "step": 55540
    },
    {
      "epoch": 4.778542544953971,
      "grad_norm": 1.765625,
      "learning_rate": 4.836738987096023e-06,
      "loss": 1.5092,
      "step": 55541
    },
    {
      "epoch": 4.778628581261293,
      "grad_norm": 1.0,
      "learning_rate": 4.832987632093277e-06,
      "loss": 1.0155,
      "step": 55542
    },
    {
      "epoch": 4.778714617568614,
      "grad_norm": 0.98828125,
      "learning_rate": 4.829237725357704e-06,
      "loss": 0.6137,
      "step": 55543
    },
    {
      "epoch": 4.778800653875936,
      "grad_norm": 6.09375,
      "learning_rate": 4.825489266900296e-06,
      "loss": 0.8534,
      "step": 55544
    },
    {
      "epoch": 4.778886690183257,
      "grad_norm": 1.2734375,
      "learning_rate": 4.821742256732042e-06,
      "loss": 1.6813,
      "step": 55545
    },
    {
      "epoch": 4.778972726490579,
      "grad_norm": 1.1796875,
      "learning_rate": 4.81799669486388e-06,
      "loss": 1.268,
      "step": 55546
    },
    {
      "epoch": 4.7790587627979,
      "grad_norm": 104.0,
      "learning_rate": 4.814252581306799e-06,
      "loss": 0.9253,
      "step": 55547
    },
    {
      "epoch": 4.7791447991052225,
      "grad_norm": 1.1875,
      "learning_rate": 4.810509916071682e-06,
      "loss": 0.9386,
      "step": 55548
    },
    {
      "epoch": 4.779230835412545,
      "grad_norm": 1.078125,
      "learning_rate": 4.806768699169517e-06,
      "loss": 0.6402,
      "step": 55549
    },
    {
      "epoch": 4.779316871719866,
      "grad_norm": 1.2421875,
      "learning_rate": 4.803028930611297e-06,
      "loss": 1.1646,
      "step": 55550
    },
    {
      "epoch": 4.779402908027188,
      "grad_norm": 0.90234375,
      "learning_rate": 4.799290610407847e-06,
      "loss": 1.2043,
      "step": 55551
    },
    {
      "epoch": 4.779488944334509,
      "grad_norm": 0.89453125,
      "learning_rate": 4.795553738570213e-06,
      "loss": 1.0872,
      "step": 55552
    },
    {
      "epoch": 4.779574980641831,
      "grad_norm": 1.640625,
      "learning_rate": 4.79181831510922e-06,
      "loss": 1.146,
      "step": 55553
    },
    {
      "epoch": 4.779661016949152,
      "grad_norm": 1.2109375,
      "learning_rate": 4.788084340035859e-06,
      "loss": 0.7215,
      "step": 55554
    },
    {
      "epoch": 4.779747053256474,
      "grad_norm": 1.203125,
      "learning_rate": 4.7843518133610655e-06,
      "loss": 1.3445,
      "step": 55555
    },
    {
      "epoch": 4.779833089563796,
      "grad_norm": 1.015625,
      "learning_rate": 4.78062073509572e-06,
      "loss": 0.3519,
      "step": 55556
    },
    {
      "epoch": 4.779919125871118,
      "grad_norm": 0.8671875,
      "learning_rate": 4.7768911052507025e-06,
      "loss": 1.0689,
      "step": 55557
    },
    {
      "epoch": 4.78000516217844,
      "grad_norm": 23.125,
      "learning_rate": 4.77316292383706e-06,
      "loss": 2.2124,
      "step": 55558
    },
    {
      "epoch": 4.780091198485761,
      "grad_norm": 1.3515625,
      "learning_rate": 4.769436190865506e-06,
      "loss": 0.8395,
      "step": 55559
    },
    {
      "epoch": 4.780177234793083,
      "grad_norm": 5.375,
      "learning_rate": 4.7657109063470314e-06,
      "loss": 0.9585,
      "step": 55560
    },
    {
      "epoch": 4.780263271100404,
      "grad_norm": 1.5078125,
      "learning_rate": 4.761987070292628e-06,
      "loss": 0.984,
      "step": 55561
    },
    {
      "epoch": 4.780349307407726,
      "grad_norm": 324.0,
      "learning_rate": 4.75826468271301e-06,
      "loss": 2.0293,
      "step": 55562
    },
    {
      "epoch": 4.780435343715048,
      "grad_norm": 2.234375,
      "learning_rate": 4.754543743619167e-06,
      "loss": 1.5748,
      "step": 55563
    },
    {
      "epoch": 4.780521380022369,
      "grad_norm": 0.796875,
      "learning_rate": 4.750824253021979e-06,
      "loss": 0.3653,
      "step": 55564
    },
    {
      "epoch": 4.780607416329691,
      "grad_norm": 1.3125,
      "learning_rate": 4.7471062109322725e-06,
      "loss": 1.3231,
      "step": 55565
    },
    {
      "epoch": 4.780693452637013,
      "grad_norm": 1.3984375,
      "learning_rate": 4.743389617361038e-06,
      "loss": 1.0817,
      "step": 55566
    },
    {
      "epoch": 4.780779488944335,
      "grad_norm": 1.0390625,
      "learning_rate": 4.739674472318989e-06,
      "loss": 0.9545,
      "step": 55567
    },
    {
      "epoch": 4.780865525251656,
      "grad_norm": 1.25,
      "learning_rate": 4.735960775817061e-06,
      "loss": 1.3831,
      "step": 55568
    },
    {
      "epoch": 4.780951561558978,
      "grad_norm": 1.296875,
      "learning_rate": 4.73224852786619e-06,
      "loss": 1.3304,
      "step": 55569
    },
    {
      "epoch": 4.7810375978663,
      "grad_norm": 1.0625,
      "learning_rate": 4.728537728477145e-06,
      "loss": 1.2275,
      "step": 55570
    },
    {
      "epoch": 4.781123634173621,
      "grad_norm": 1.0625,
      "learning_rate": 4.724828377660806e-06,
      "loss": 1.6024,
      "step": 55571
    },
    {
      "epoch": 4.781209670480943,
      "grad_norm": 0.98046875,
      "learning_rate": 4.7211204754279975e-06,
      "loss": 0.7067,
      "step": 55572
    },
    {
      "epoch": 4.781295706788264,
      "grad_norm": 0.9921875,
      "learning_rate": 4.7174140217896014e-06,
      "loss": 0.6906,
      "step": 55573
    },
    {
      "epoch": 4.7813817430955865,
      "grad_norm": 1.03125,
      "learning_rate": 4.713709016756384e-06,
      "loss": 0.674,
      "step": 55574
    },
    {
      "epoch": 4.781467779402908,
      "grad_norm": 1.6875,
      "learning_rate": 4.71000546033934e-06,
      "loss": 0.8378,
      "step": 55575
    },
    {
      "epoch": 4.78155381571023,
      "grad_norm": 194.0,
      "learning_rate": 4.706303352549124e-06,
      "loss": 1.3937,
      "step": 55576
    },
    {
      "epoch": 4.781639852017552,
      "grad_norm": 1.2578125,
      "learning_rate": 4.702602693396674e-06,
      "loss": 1.2271,
      "step": 55577
    },
    {
      "epoch": 4.781725888324873,
      "grad_norm": 1.21875,
      "learning_rate": 4.698903482892813e-06,
      "loss": 0.6286,
      "step": 55578
    },
    {
      "epoch": 4.781811924632195,
      "grad_norm": 8576.0,
      "learning_rate": 4.695205721048257e-06,
      "loss": 0.6304,
      "step": 55579
    },
    {
      "epoch": 4.781897960939516,
      "grad_norm": 0.9453125,
      "learning_rate": 4.69150940787394e-06,
      "loss": 0.8618,
      "step": 55580
    },
    {
      "epoch": 4.781983997246838,
      "grad_norm": 1.09375,
      "learning_rate": 4.6878145433806305e-06,
      "loss": 0.9063,
      "step": 55581
    },
    {
      "epoch": 4.7820700335541595,
      "grad_norm": 1.0859375,
      "learning_rate": 4.684121127579099e-06,
      "loss": 0.694,
      "step": 55582
    },
    {
      "epoch": 4.782156069861482,
      "grad_norm": 1.1484375,
      "learning_rate": 4.680429160480226e-06,
      "loss": 1.7953,
      "step": 55583
    },
    {
      "epoch": 4.782242106168804,
      "grad_norm": 1.203125,
      "learning_rate": 4.676738642094669e-06,
      "loss": 1.4093,
      "step": 55584
    },
    {
      "epoch": 4.782328142476125,
      "grad_norm": 2992.0,
      "learning_rate": 4.673049572433419e-06,
      "loss": 1.5028,
      "step": 55585
    },
    {
      "epoch": 4.782414178783447,
      "grad_norm": 40.0,
      "learning_rate": 4.669361951507078e-06,
      "loss": 1.6392,
      "step": 55586
    },
    {
      "epoch": 4.782500215090768,
      "grad_norm": 1.0390625,
      "learning_rate": 4.665675779326583e-06,
      "loss": 0.9214,
      "step": 55587
    },
    {
      "epoch": 4.78258625139809,
      "grad_norm": 27.375,
      "learning_rate": 4.661991055902648e-06,
      "loss": 1.5317,
      "step": 55588
    },
    {
      "epoch": 4.782672287705411,
      "grad_norm": 2.328125,
      "learning_rate": 4.658307781245985e-06,
      "loss": 1.0224,
      "step": 55589
    },
    {
      "epoch": 4.782758324012733,
      "grad_norm": 1.3671875,
      "learning_rate": 4.654625955367476e-06,
      "loss": 1.0461,
      "step": 55590
    },
    {
      "epoch": 4.782844360320055,
      "grad_norm": 1.0703125,
      "learning_rate": 4.6509455782778875e-06,
      "loss": 0.8848,
      "step": 55591
    },
    {
      "epoch": 4.782930396627377,
      "grad_norm": 0.99609375,
      "learning_rate": 4.64726664998788e-06,
      "loss": 0.5591,
      "step": 55592
    },
    {
      "epoch": 4.783016432934699,
      "grad_norm": 1.2265625,
      "learning_rate": 4.6435891705083335e-06,
      "loss": 0.9435,
      "step": 55593
    },
    {
      "epoch": 4.78310246924202,
      "grad_norm": 1.0703125,
      "learning_rate": 4.639913139849905e-06,
      "loss": 0.7637,
      "step": 55594
    },
    {
      "epoch": 4.783188505549342,
      "grad_norm": 1.4921875,
      "learning_rate": 4.636238558023365e-06,
      "loss": 2.1383,
      "step": 55595
    },
    {
      "epoch": 4.783274541856663,
      "grad_norm": 1.2265625,
      "learning_rate": 4.632565425039592e-06,
      "loss": 1.9231,
      "step": 55596
    },
    {
      "epoch": 4.783360578163985,
      "grad_norm": 6.46875,
      "learning_rate": 4.6288937409091344e-06,
      "loss": 1.8475,
      "step": 55597
    },
    {
      "epoch": 4.783446614471307,
      "grad_norm": 1.25,
      "learning_rate": 4.625223505642817e-06,
      "loss": 0.9805,
      "step": 55598
    },
    {
      "epoch": 4.783532650778628,
      "grad_norm": 1.3125,
      "learning_rate": 4.621554719251408e-06,
      "loss": 1.505,
      "step": 55599
    },
    {
      "epoch": 4.7836186870859505,
      "grad_norm": 46.0,
      "learning_rate": 4.617887381745567e-06,
      "loss": 0.8663,
      "step": 55600
    },
    {
      "epoch": 4.783704723393272,
      "grad_norm": 0.81640625,
      "learning_rate": 4.614221493136061e-06,
      "loss": 0.5248,
      "step": 55601
    },
    {
      "epoch": 4.783790759700594,
      "grad_norm": 0.94921875,
      "learning_rate": 4.610557053433662e-06,
      "loss": 0.7487,
      "step": 55602
    },
    {
      "epoch": 4.783876796007915,
      "grad_norm": 1.0625,
      "learning_rate": 4.60689406264897e-06,
      "loss": 1.3747,
      "step": 55603
    },
    {
      "epoch": 4.783962832315237,
      "grad_norm": 2736.0,
      "learning_rate": 4.603232520792811e-06,
      "loss": 1.0765,
      "step": 55604
    },
    {
      "epoch": 4.784048868622559,
      "grad_norm": 0.8828125,
      "learning_rate": 4.599572427875842e-06,
      "loss": 0.8361,
      "step": 55605
    },
    {
      "epoch": 4.78413490492988,
      "grad_norm": 8.3125,
      "learning_rate": 4.595913783908778e-06,
      "loss": 1.7702,
      "step": 55606
    },
    {
      "epoch": 4.784220941237202,
      "grad_norm": 194.0,
      "learning_rate": 4.592256588902333e-06,
      "loss": 0.5675,
      "step": 55607
    },
    {
      "epoch": 4.7843069775445235,
      "grad_norm": 1.3203125,
      "learning_rate": 4.5886008428671075e-06,
      "loss": 0.9151,
      "step": 55608
    },
    {
      "epoch": 4.784393013851846,
      "grad_norm": 1.3046875,
      "learning_rate": 4.584946545813928e-06,
      "loss": 0.9433,
      "step": 55609
    },
    {
      "epoch": 4.784479050159167,
      "grad_norm": 2.734375,
      "learning_rate": 4.581293697753453e-06,
      "loss": 1.5428,
      "step": 55610
    },
    {
      "epoch": 4.784565086466489,
      "grad_norm": 0.98828125,
      "learning_rate": 4.577642298696283e-06,
      "loss": 0.5329,
      "step": 55611
    },
    {
      "epoch": 4.784651122773811,
      "grad_norm": 1.234375,
      "learning_rate": 4.573992348653189e-06,
      "loss": 1.3387,
      "step": 55612
    },
    {
      "epoch": 4.784737159081132,
      "grad_norm": 1.3671875,
      "learning_rate": 4.570343847634828e-06,
      "loss": 1.3931,
      "step": 55613
    },
    {
      "epoch": 4.784823195388454,
      "grad_norm": 2.828125,
      "learning_rate": 4.566696795651804e-06,
      "loss": 1.0232,
      "step": 55614
    },
    {
      "epoch": 4.784909231695775,
      "grad_norm": 1.0390625,
      "learning_rate": 4.5630511927148284e-06,
      "loss": 1.2076,
      "step": 55615
    },
    {
      "epoch": 4.784995268003097,
      "grad_norm": 7.1875,
      "learning_rate": 4.559407038834617e-06,
      "loss": 1.1316,
      "step": 55616
    },
    {
      "epoch": 4.785081304310419,
      "grad_norm": 1.046875,
      "learning_rate": 4.555764334021772e-06,
      "loss": 0.6933,
      "step": 55617
    },
    {
      "epoch": 4.785167340617741,
      "grad_norm": 1.015625,
      "learning_rate": 4.552123078287007e-06,
      "loss": 1.2618,
      "step": 55618
    },
    {
      "epoch": 4.785253376925063,
      "grad_norm": 1.28125,
      "learning_rate": 4.548483271640813e-06,
      "loss": 1.0998,
      "step": 55619
    },
    {
      "epoch": 4.785339413232384,
      "grad_norm": 0.94140625,
      "learning_rate": 4.544844914094015e-06,
      "loss": 0.8419,
      "step": 55620
    },
    {
      "epoch": 4.785425449539706,
      "grad_norm": 2.078125,
      "learning_rate": 4.54120800565716e-06,
      "loss": 1.9067,
      "step": 55621
    },
    {
      "epoch": 4.785511485847027,
      "grad_norm": 3.734375,
      "learning_rate": 4.537572546340907e-06,
      "loss": 0.7636,
      "step": 55622
    },
    {
      "epoch": 4.785597522154349,
      "grad_norm": 2.1875,
      "learning_rate": 4.5339385361559124e-06,
      "loss": 0.7661,
      "step": 55623
    },
    {
      "epoch": 4.78568355846167,
      "grad_norm": 1.015625,
      "learning_rate": 4.5303059751127806e-06,
      "loss": 0.68,
      "step": 55624
    },
    {
      "epoch": 4.785769594768992,
      "grad_norm": 1.2734375,
      "learning_rate": 4.5266748632221685e-06,
      "loss": 2.3214,
      "step": 55625
    },
    {
      "epoch": 4.7858556310763145,
      "grad_norm": 1.8046875,
      "learning_rate": 4.523045200494624e-06,
      "loss": 0.5849,
      "step": 55626
    },
    {
      "epoch": 4.785941667383636,
      "grad_norm": 1.1953125,
      "learning_rate": 4.519416986940805e-06,
      "loss": 1.0523,
      "step": 55627
    },
    {
      "epoch": 4.786027703690958,
      "grad_norm": 3.421875,
      "learning_rate": 4.51579022257137e-06,
      "loss": 0.767,
      "step": 55628
    },
    {
      "epoch": 4.786113739998279,
      "grad_norm": 0.9921875,
      "learning_rate": 4.512164907396865e-06,
      "loss": 1.8098,
      "step": 55629
    },
    {
      "epoch": 4.786199776305601,
      "grad_norm": 1.1796875,
      "learning_rate": 4.508541041427894e-06,
      "loss": 1.289,
      "step": 55630
    },
    {
      "epoch": 4.786285812612922,
      "grad_norm": 0.86328125,
      "learning_rate": 4.504918624675059e-06,
      "loss": 0.6271,
      "step": 55631
    },
    {
      "epoch": 4.786371848920244,
      "grad_norm": 3.40625,
      "learning_rate": 4.5012976571490175e-06,
      "loss": 1.3639,
      "step": 55632
    },
    {
      "epoch": 4.786457885227566,
      "grad_norm": 1.21875,
      "learning_rate": 4.497678138860261e-06,
      "loss": 1.0083,
      "step": 55633
    },
    {
      "epoch": 4.7865439215348875,
      "grad_norm": 0.890625,
      "learning_rate": 4.494060069819506e-06,
      "loss": 0.5638,
      "step": 55634
    },
    {
      "epoch": 4.78662995784221,
      "grad_norm": 1.140625,
      "learning_rate": 4.490443450037185e-06,
      "loss": 1.2977,
      "step": 55635
    },
    {
      "epoch": 4.786715994149531,
      "grad_norm": 1.3828125,
      "learning_rate": 4.486828279523902e-06,
      "loss": 1.1213,
      "step": 55636
    },
    {
      "epoch": 4.786802030456853,
      "grad_norm": 1.2109375,
      "learning_rate": 4.4832145582903716e-06,
      "loss": 1.1558,
      "step": 55637
    },
    {
      "epoch": 4.786888066764174,
      "grad_norm": 1.625,
      "learning_rate": 4.479602286347029e-06,
      "loss": 1.6894,
      "step": 55638
    },
    {
      "epoch": 4.786974103071496,
      "grad_norm": 0.91796875,
      "learning_rate": 4.4759914637044765e-06,
      "loss": 0.5876,
      "step": 55639
    },
    {
      "epoch": 4.787060139378818,
      "grad_norm": 1.109375,
      "learning_rate": 4.472382090373262e-06,
      "loss": 1.341,
      "step": 55640
    },
    {
      "epoch": 4.787146175686139,
      "grad_norm": 0.84375,
      "learning_rate": 4.468774166363987e-06,
      "loss": 0.5875,
      "step": 55641
    },
    {
      "epoch": 4.787232211993461,
      "grad_norm": 4.09375,
      "learning_rate": 4.465167691687088e-06,
      "loss": 0.8239,
      "step": 55642
    },
    {
      "epoch": 4.787318248300783,
      "grad_norm": 19.625,
      "learning_rate": 4.46156266635328e-06,
      "loss": 1.2118,
      "step": 55643
    },
    {
      "epoch": 4.787404284608105,
      "grad_norm": 0.86328125,
      "learning_rate": 4.4579590903729985e-06,
      "loss": 0.7997,
      "step": 55644
    },
    {
      "epoch": 4.787490320915426,
      "grad_norm": 0.84375,
      "learning_rate": 4.454356963756845e-06,
      "loss": 0.5879,
      "step": 55645
    },
    {
      "epoch": 4.787576357222748,
      "grad_norm": 1.1875,
      "learning_rate": 4.450756286515256e-06,
      "loss": 0.749,
      "step": 55646
    },
    {
      "epoch": 4.78766239353007,
      "grad_norm": 0.9140625,
      "learning_rate": 4.44715705865889e-06,
      "loss": 1.8407,
      "step": 55647
    },
    {
      "epoch": 4.787748429837391,
      "grad_norm": 0.75,
      "learning_rate": 4.443559280198184e-06,
      "loss": 0.6807,
      "step": 55648
    },
    {
      "epoch": 4.787834466144713,
      "grad_norm": 0.92578125,
      "learning_rate": 4.439962951143684e-06,
      "loss": 0.8901,
      "step": 55649
    },
    {
      "epoch": 4.787920502452034,
      "grad_norm": 1.015625,
      "learning_rate": 4.436368071505881e-06,
      "loss": 0.7853,
      "step": 55650
    },
    {
      "epoch": 4.7880065387593564,
      "grad_norm": 0.96484375,
      "learning_rate": 4.432774641295378e-06,
      "loss": 0.7682,
      "step": 55651
    },
    {
      "epoch": 4.7880925750666785,
      "grad_norm": 1.109375,
      "learning_rate": 4.429182660522557e-06,
      "loss": 1.3376,
      "step": 55652
    },
    {
      "epoch": 4.788178611374,
      "grad_norm": 3472.0,
      "learning_rate": 4.4255921291980195e-06,
      "loss": 1.3702,
      "step": 55653
    },
    {
      "epoch": 4.788264647681322,
      "grad_norm": 38.5,
      "learning_rate": 4.422003047332312e-06,
      "loss": 0.9772,
      "step": 55654
    },
    {
      "epoch": 4.788350683988643,
      "grad_norm": 1.3671875,
      "learning_rate": 4.4184154149357615e-06,
      "loss": 1.8151,
      "step": 55655
    },
    {
      "epoch": 4.788436720295965,
      "grad_norm": 1.1171875,
      "learning_rate": 4.414829232018969e-06,
      "loss": 1.4736,
      "step": 55656
    },
    {
      "epoch": 4.788522756603287,
      "grad_norm": 1.1640625,
      "learning_rate": 4.411244498592481e-06,
      "loss": 1.2502,
      "step": 55657
    },
    {
      "epoch": 4.788608792910608,
      "grad_norm": 1.1953125,
      "learning_rate": 4.4076612146666804e-06,
      "loss": 0.7841,
      "step": 55658
    },
    {
      "epoch": 4.78869482921793,
      "grad_norm": 1.1953125,
      "learning_rate": 4.404079380252058e-06,
      "loss": 1.2063,
      "step": 55659
    },
    {
      "epoch": 4.7887808655252515,
      "grad_norm": 1.0859375,
      "learning_rate": 4.400498995359159e-06,
      "loss": 1.2594,
      "step": 55660
    },
    {
      "epoch": 4.788866901832574,
      "grad_norm": 1.1875,
      "learning_rate": 4.396920059998366e-06,
      "loss": 1.4502,
      "step": 55661
    },
    {
      "epoch": 4.788952938139895,
      "grad_norm": 1.0,
      "learning_rate": 4.393342574180281e-06,
      "loss": 1.5091,
      "step": 55662
    },
    {
      "epoch": 4.789038974447217,
      "grad_norm": 19.5,
      "learning_rate": 4.389766537915174e-06,
      "loss": 1.1459,
      "step": 55663
    },
    {
      "epoch": 4.789125010754539,
      "grad_norm": 1.4140625,
      "learning_rate": 4.386191951213703e-06,
      "loss": 1.2224,
      "step": 55664
    },
    {
      "epoch": 4.78921104706186,
      "grad_norm": 1.203125,
      "learning_rate": 4.3826188140861365e-06,
      "loss": 0.7786,
      "step": 55665
    },
    {
      "epoch": 4.789297083369182,
      "grad_norm": 5.1875,
      "learning_rate": 4.3790471265430785e-06,
      "loss": 1.1571,
      "step": 55666
    },
    {
      "epoch": 4.789383119676503,
      "grad_norm": 2.0,
      "learning_rate": 4.375476888594909e-06,
      "loss": 0.9656,
      "step": 55667
    },
    {
      "epoch": 4.789469155983825,
      "grad_norm": 1.234375,
      "learning_rate": 4.371908100252064e-06,
      "loss": 0.8823,
      "step": 55668
    },
    {
      "epoch": 4.789555192291147,
      "grad_norm": 1.390625,
      "learning_rate": 4.36834076152498e-06,
      "loss": 1.276,
      "step": 55669
    },
    {
      "epoch": 4.789641228598469,
      "grad_norm": 1.7421875,
      "learning_rate": 4.364774872424149e-06,
      "loss": 0.7611,
      "step": 55670
    },
    {
      "epoch": 4.789727264905791,
      "grad_norm": 0.89453125,
      "learning_rate": 4.361210432959895e-06,
      "loss": 0.8977,
      "step": 55671
    },
    {
      "epoch": 4.789813301213112,
      "grad_norm": 1.0625,
      "learning_rate": 4.35764744314271e-06,
      "loss": 1.383,
      "step": 55672
    },
    {
      "epoch": 4.789899337520434,
      "grad_norm": 1.4609375,
      "learning_rate": 4.3540859029830845e-06,
      "loss": 0.8038,
      "step": 55673
    },
    {
      "epoch": 4.789985373827755,
      "grad_norm": 1.03125,
      "learning_rate": 4.35052581249129e-06,
      "loss": 0.631,
      "step": 55674
    },
    {
      "epoch": 4.790071410135077,
      "grad_norm": 2.234375,
      "learning_rate": 4.346967171677818e-06,
      "loss": 1.2608,
      "step": 55675
    },
    {
      "epoch": 4.790157446442398,
      "grad_norm": 1.03125,
      "learning_rate": 4.343409980553103e-06,
      "loss": 0.7054,
      "step": 55676
    },
    {
      "epoch": 4.7902434827497204,
      "grad_norm": 1.71875,
      "learning_rate": 4.339854239127472e-06,
      "loss": 1.4289,
      "step": 55677
    },
    {
      "epoch": 4.7903295190570425,
      "grad_norm": 2.5,
      "learning_rate": 4.336299947411415e-06,
      "loss": 1.1901,
      "step": 55678
    },
    {
      "epoch": 4.790415555364364,
      "grad_norm": 1.078125,
      "learning_rate": 4.332747105415202e-06,
      "loss": 0.9948,
      "step": 55679
    },
    {
      "epoch": 4.790501591671686,
      "grad_norm": 1.1328125,
      "learning_rate": 4.329195713149325e-06,
      "loss": 0.9258,
      "step": 55680
    },
    {
      "epoch": 4.790587627979007,
      "grad_norm": 1.109375,
      "learning_rate": 4.3256457706242204e-06,
      "loss": 1.104,
      "step": 55681
    },
    {
      "epoch": 4.790673664286329,
      "grad_norm": 0.92578125,
      "learning_rate": 4.322097277850157e-06,
      "loss": 0.8205,
      "step": 55682
    },
    {
      "epoch": 4.79075970059365,
      "grad_norm": 1.1015625,
      "learning_rate": 4.318550234837515e-06,
      "loss": 1.0108,
      "step": 55683
    },
    {
      "epoch": 4.790845736900972,
      "grad_norm": 7.4375,
      "learning_rate": 4.315004641596732e-06,
      "loss": 1.5585,
      "step": 55684
    },
    {
      "epoch": 4.790931773208294,
      "grad_norm": 1.296875,
      "learning_rate": 4.3114604981381864e-06,
      "loss": 1.9172,
      "step": 55685
    },
    {
      "epoch": 4.7910178095156155,
      "grad_norm": 1.2421875,
      "learning_rate": 4.30791780447215e-06,
      "loss": 0.7332,
      "step": 55686
    },
    {
      "epoch": 4.791103845822938,
      "grad_norm": 1.0859375,
      "learning_rate": 4.304376560609114e-06,
      "loss": 1.1617,
      "step": 55687
    },
    {
      "epoch": 4.791189882130259,
      "grad_norm": 1.078125,
      "learning_rate": 4.300836766559346e-06,
      "loss": 1.5452,
      "step": 55688
    },
    {
      "epoch": 4.791275918437581,
      "grad_norm": 1.046875,
      "learning_rate": 4.297298422333229e-06,
      "loss": 1.0185,
      "step": 55689
    },
    {
      "epoch": 4.791361954744902,
      "grad_norm": 1.0078125,
      "learning_rate": 4.293761527941087e-06,
      "loss": 1.2603,
      "step": 55690
    },
    {
      "epoch": 4.791447991052224,
      "grad_norm": 1.5546875,
      "learning_rate": 4.2902260833933e-06,
      "loss": 1.2417,
      "step": 55691
    },
    {
      "epoch": 4.791534027359546,
      "grad_norm": 5.59375,
      "learning_rate": 4.286692088700195e-06,
      "loss": 0.941,
      "step": 55692
    },
    {
      "epoch": 4.791620063666867,
      "grad_norm": 1.1640625,
      "learning_rate": 4.283159543872095e-06,
      "loss": 0.7785,
      "step": 55693
    },
    {
      "epoch": 4.791706099974189,
      "grad_norm": 1.234375,
      "learning_rate": 4.279628448919326e-06,
      "loss": 1.231,
      "step": 55694
    },
    {
      "epoch": 4.791792136281511,
      "grad_norm": 1.0625,
      "learning_rate": 4.276098803852213e-06,
      "loss": 1.6828,
      "step": 55695
    },
    {
      "epoch": 4.791878172588833,
      "grad_norm": 1.3203125,
      "learning_rate": 4.272570608681137e-06,
      "loss": 0.9058,
      "step": 55696
    },
    {
      "epoch": 4.791964208896154,
      "grad_norm": 1.0234375,
      "learning_rate": 4.269043863416366e-06,
      "loss": 0.7987,
      "step": 55697
    },
    {
      "epoch": 4.792050245203476,
      "grad_norm": 1.125,
      "learning_rate": 4.265518568068227e-06,
      "loss": 1.1748,
      "step": 55698
    },
    {
      "epoch": 4.792136281510798,
      "grad_norm": 14.5625,
      "learning_rate": 4.261994722647044e-06,
      "loss": 0.749,
      "step": 55699
    },
    {
      "epoch": 4.792222317818119,
      "grad_norm": 7.375,
      "learning_rate": 4.258472327163143e-06,
      "loss": 0.6033,
      "step": 55700
    },
    {
      "epoch": 4.792308354125441,
      "grad_norm": 0.82421875,
      "learning_rate": 4.254951381626737e-06,
      "loss": 0.3287,
      "step": 55701
    },
    {
      "epoch": 4.792394390432762,
      "grad_norm": 0.6171875,
      "learning_rate": 4.251431886048152e-06,
      "loss": 0.3304,
      "step": 55702
    },
    {
      "epoch": 4.7924804267400845,
      "grad_norm": 1.1328125,
      "learning_rate": 4.247913840437767e-06,
      "loss": 1.8771,
      "step": 55703
    },
    {
      "epoch": 4.792566463047406,
      "grad_norm": 3.71875,
      "learning_rate": 4.244397244805798e-06,
      "loss": 0.9453,
      "step": 55704
    },
    {
      "epoch": 4.792652499354728,
      "grad_norm": 1.1171875,
      "learning_rate": 4.240882099162624e-06,
      "loss": 1.0718,
      "step": 55705
    },
    {
      "epoch": 4.79273853566205,
      "grad_norm": 1.1171875,
      "learning_rate": 4.237368403518349e-06,
      "loss": 1.2359,
      "step": 55706
    },
    {
      "epoch": 4.792824571969371,
      "grad_norm": 0.921875,
      "learning_rate": 4.233856157883353e-06,
      "loss": 0.8153,
      "step": 55707
    },
    {
      "epoch": 4.792910608276693,
      "grad_norm": 1.21875,
      "learning_rate": 4.230345362267962e-06,
      "loss": 0.927,
      "step": 55708
    },
    {
      "epoch": 4.792996644584014,
      "grad_norm": 1.015625,
      "learning_rate": 4.226836016682389e-06,
      "loss": 1.2335,
      "step": 55709
    },
    {
      "epoch": 4.793082680891336,
      "grad_norm": 1.484375,
      "learning_rate": 4.223328121136849e-06,
      "loss": 0.957,
      "step": 55710
    },
    {
      "epoch": 4.7931687171986574,
      "grad_norm": 0.9609375,
      "learning_rate": 4.219821675641666e-06,
      "loss": 0.9256,
      "step": 55711
    },
    {
      "epoch": 4.7932547535059795,
      "grad_norm": 1.296875,
      "learning_rate": 4.216316680207111e-06,
      "loss": 1.179,
      "step": 55712
    },
    {
      "epoch": 4.793340789813302,
      "grad_norm": 0.90234375,
      "learning_rate": 4.212813134843341e-06,
      "loss": 0.6794,
      "step": 55713
    },
    {
      "epoch": 4.793426826120623,
      "grad_norm": 3.078125,
      "learning_rate": 4.209311039560737e-06,
      "loss": 0.8885,
      "step": 55714
    },
    {
      "epoch": 4.793512862427945,
      "grad_norm": 1.2109375,
      "learning_rate": 4.205810394369458e-06,
      "loss": 1.3654,
      "step": 55715
    },
    {
      "epoch": 4.793598898735266,
      "grad_norm": 1.2734375,
      "learning_rate": 4.202311199279718e-06,
      "loss": 1.0615,
      "step": 55716
    },
    {
      "epoch": 4.793684935042588,
      "grad_norm": 0.9921875,
      "learning_rate": 4.198813454301842e-06,
      "loss": 1.3387,
      "step": 55717
    },
    {
      "epoch": 4.793770971349909,
      "grad_norm": 1.1015625,
      "learning_rate": 4.195317159445989e-06,
      "loss": 0.8247,
      "step": 55718
    },
    {
      "epoch": 4.793857007657231,
      "grad_norm": 0.83984375,
      "learning_rate": 4.1918223147224264e-06,
      "loss": 0.7469,
      "step": 55719
    },
    {
      "epoch": 4.793943043964553,
      "grad_norm": 1.0859375,
      "learning_rate": 4.188328920141316e-06,
      "loss": 3.1251,
      "step": 55720
    },
    {
      "epoch": 4.794029080271875,
      "grad_norm": 1.3515625,
      "learning_rate": 4.184836975712925e-06,
      "loss": 1.3534,
      "step": 55721
    },
    {
      "epoch": 4.794115116579197,
      "grad_norm": 0.95703125,
      "learning_rate": 4.181346481447468e-06,
      "loss": 1.6767,
      "step": 55722
    },
    {
      "epoch": 4.794201152886518,
      "grad_norm": 1.0078125,
      "learning_rate": 4.177857437355104e-06,
      "loss": 1.5827,
      "step": 55723
    },
    {
      "epoch": 4.79428718919384,
      "grad_norm": 1.015625,
      "learning_rate": 4.174369843446102e-06,
      "loss": 0.7797,
      "step": 55724
    },
    {
      "epoch": 4.794373225501161,
      "grad_norm": 10.8125,
      "learning_rate": 4.170883699730621e-06,
      "loss": 0.9818,
      "step": 55725
    },
    {
      "epoch": 4.794459261808483,
      "grad_norm": 2.078125,
      "learning_rate": 4.167399006218875e-06,
      "loss": 0.7889,
      "step": 55726
    },
    {
      "epoch": 4.794545298115805,
      "grad_norm": 0.93359375,
      "learning_rate": 4.1639157629210775e-06,
      "loss": 0.7373,
      "step": 55727
    },
    {
      "epoch": 4.794631334423126,
      "grad_norm": 24.625,
      "learning_rate": 4.160433969847388e-06,
      "loss": 0.6308,
      "step": 55728
    },
    {
      "epoch": 4.7947173707304485,
      "grad_norm": 1.0078125,
      "learning_rate": 4.1569536270079644e-06,
      "loss": 0.529,
      "step": 55729
    },
    {
      "epoch": 4.79480340703777,
      "grad_norm": 1.4609375,
      "learning_rate": 4.153474734413021e-06,
      "loss": 1.4033,
      "step": 55730
    },
    {
      "epoch": 4.794889443345092,
      "grad_norm": 1.4921875,
      "learning_rate": 4.149997292072771e-06,
      "loss": 0.9331,
      "step": 55731
    },
    {
      "epoch": 4.794975479652413,
      "grad_norm": 1.390625,
      "learning_rate": 4.146521299997264e-06,
      "loss": 0.8795,
      "step": 55732
    },
    {
      "epoch": 4.795061515959735,
      "grad_norm": 1.4296875,
      "learning_rate": 4.143046758196822e-06,
      "loss": 0.8283,
      "step": 55733
    },
    {
      "epoch": 4.795147552267057,
      "grad_norm": 1.2421875,
      "learning_rate": 4.13957366668144e-06,
      "loss": 0.8548,
      "step": 55734
    },
    {
      "epoch": 4.795233588574378,
      "grad_norm": 4.75,
      "learning_rate": 4.136102025461441e-06,
      "loss": 1.1136,
      "step": 55735
    },
    {
      "epoch": 4.7953196248817,
      "grad_norm": 2.34375,
      "learning_rate": 4.132631834546874e-06,
      "loss": 1.0538,
      "step": 55736
    },
    {
      "epoch": 4.7954056611890215,
      "grad_norm": 1.1328125,
      "learning_rate": 4.129163093947952e-06,
      "loss": 0.8278,
      "step": 55737
    },
    {
      "epoch": 4.7954916974963435,
      "grad_norm": 0.99609375,
      "learning_rate": 4.125695803674778e-06,
      "loss": 0.5143,
      "step": 55738
    },
    {
      "epoch": 4.795577733803665,
      "grad_norm": 1.4140625,
      "learning_rate": 4.122229963737456e-06,
      "loss": 1.2219,
      "step": 55739
    },
    {
      "epoch": 4.795663770110987,
      "grad_norm": 3.515625,
      "learning_rate": 4.1187655741462e-06,
      "loss": 1.4069,
      "step": 55740
    },
    {
      "epoch": 4.795749806418309,
      "grad_norm": 1.4375,
      "learning_rate": 4.115302634911111e-06,
      "loss": 0.9227,
      "step": 55741
    },
    {
      "epoch": 4.79583584272563,
      "grad_norm": 1.0390625,
      "learning_rate": 4.111841146042294e-06,
      "loss": 0.7106,
      "step": 55742
    },
    {
      "epoch": 4.795921879032952,
      "grad_norm": 1.0703125,
      "learning_rate": 4.108381107549908e-06,
      "loss": 1.3248,
      "step": 55743
    },
    {
      "epoch": 4.796007915340273,
      "grad_norm": 0.90234375,
      "learning_rate": 4.10492251944411e-06,
      "loss": 0.8837,
      "step": 55744
    },
    {
      "epoch": 4.796093951647595,
      "grad_norm": 18.625,
      "learning_rate": 4.101465381734948e-06,
      "loss": 1.4603,
      "step": 55745
    },
    {
      "epoch": 4.7961799879549165,
      "grad_norm": 8.125,
      "learning_rate": 4.098009694432525e-06,
      "loss": 1.332,
      "step": 55746
    },
    {
      "epoch": 4.796266024262239,
      "grad_norm": 0.953125,
      "learning_rate": 4.094555457547e-06,
      "loss": 0.8954,
      "step": 55747
    },
    {
      "epoch": 4.796352060569561,
      "grad_norm": 356.0,
      "learning_rate": 4.091102671088421e-06,
      "loss": 1.013,
      "step": 55748
    },
    {
      "epoch": 4.796438096876882,
      "grad_norm": 1.0078125,
      "learning_rate": 4.0876513350669444e-06,
      "loss": 1.3691,
      "step": 55749
    },
    {
      "epoch": 4.796524133184204,
      "grad_norm": 255.0,
      "learning_rate": 4.084201449492619e-06,
      "loss": 0.6026,
      "step": 55750
    },
    {
      "epoch": 4.796610169491525,
      "grad_norm": 2.953125,
      "learning_rate": 4.080753014375549e-06,
      "loss": 0.9676,
      "step": 55751
    },
    {
      "epoch": 4.796696205798847,
      "grad_norm": 1.140625,
      "learning_rate": 4.077306029725836e-06,
      "loss": 1.0811,
      "step": 55752
    },
    {
      "epoch": 4.796782242106168,
      "grad_norm": 1.375,
      "learning_rate": 4.073860495553583e-06,
      "loss": 1.2802,
      "step": 55753
    },
    {
      "epoch": 4.79686827841349,
      "grad_norm": 1.2265625,
      "learning_rate": 4.070416411868783e-06,
      "loss": 2.1596,
      "step": 55754
    },
    {
      "epoch": 4.7969543147208125,
      "grad_norm": 0.68359375,
      "learning_rate": 4.066973778681593e-06,
      "loss": 0.3909,
      "step": 55755
    },
    {
      "epoch": 4.797040351028134,
      "grad_norm": 1.1484375,
      "learning_rate": 4.063532596002062e-06,
      "loss": 0.6374,
      "step": 55756
    },
    {
      "epoch": 4.797126387335456,
      "grad_norm": 1.234375,
      "learning_rate": 4.060092863840182e-06,
      "loss": 1.0486,
      "step": 55757
    },
    {
      "epoch": 4.797212423642777,
      "grad_norm": 1.171875,
      "learning_rate": 4.0566545822061655e-06,
      "loss": 0.6283,
      "step": 55758
    },
    {
      "epoch": 4.797298459950099,
      "grad_norm": 1.1484375,
      "learning_rate": 4.053217751109894e-06,
      "loss": 0.5365,
      "step": 55759
    },
    {
      "epoch": 4.79738449625742,
      "grad_norm": 0.953125,
      "learning_rate": 4.049782370561584e-06,
      "loss": 0.7407,
      "step": 55760
    },
    {
      "epoch": 4.797470532564742,
      "grad_norm": 1.390625,
      "learning_rate": 4.046348440571168e-06,
      "loss": 0.7435,
      "step": 55761
    },
    {
      "epoch": 4.797556568872064,
      "grad_norm": 0.80859375,
      "learning_rate": 4.042915961148696e-06,
      "loss": 0.6481,
      "step": 55762
    },
    {
      "epoch": 4.7976426051793855,
      "grad_norm": 1.2265625,
      "learning_rate": 4.039484932304271e-06,
      "loss": 1.0262,
      "step": 55763
    },
    {
      "epoch": 4.7977286414867075,
      "grad_norm": 1.140625,
      "learning_rate": 4.036055354047885e-06,
      "loss": 1.0001,
      "step": 55764
    },
    {
      "epoch": 4.797814677794029,
      "grad_norm": 4.03125,
      "learning_rate": 4.032627226389585e-06,
      "loss": 1.1666,
      "step": 55765
    },
    {
      "epoch": 4.797900714101351,
      "grad_norm": 10.75,
      "learning_rate": 4.0292005493393626e-06,
      "loss": 1.0063,
      "step": 55766
    },
    {
      "epoch": 4.797986750408673,
      "grad_norm": 4.625,
      "learning_rate": 4.025775322907266e-06,
      "loss": 1.087,
      "step": 55767
    },
    {
      "epoch": 4.798072786715994,
      "grad_norm": 1.09375,
      "learning_rate": 4.022351547103343e-06,
      "loss": 0.8046,
      "step": 55768
    },
    {
      "epoch": 4.798158823023316,
      "grad_norm": 1.46875,
      "learning_rate": 4.018929221937584e-06,
      "loss": 1.3159,
      "step": 55769
    },
    {
      "epoch": 4.798244859330637,
      "grad_norm": 1.28125,
      "learning_rate": 4.015508347419927e-06,
      "loss": 0.9459,
      "step": 55770
    },
    {
      "epoch": 4.798330895637959,
      "grad_norm": 3.75,
      "learning_rate": 4.012088923560531e-06,
      "loss": 1.4678,
      "step": 55771
    },
    {
      "epoch": 4.798416931945281,
      "grad_norm": 1.109375,
      "learning_rate": 4.008670950369275e-06,
      "loss": 1.422,
      "step": 55772
    },
    {
      "epoch": 4.798502968252603,
      "grad_norm": 1.203125,
      "learning_rate": 4.005254427856208e-06,
      "loss": 1.2739,
      "step": 55773
    },
    {
      "epoch": 4.798589004559925,
      "grad_norm": 0.96484375,
      "learning_rate": 4.001839356031267e-06,
      "loss": 1.1005,
      "step": 55774
    },
    {
      "epoch": 4.798675040867246,
      "grad_norm": 0.77734375,
      "learning_rate": 3.998425734904498e-06,
      "loss": 0.7897,
      "step": 55775
    },
    {
      "epoch": 4.798761077174568,
      "grad_norm": 1.1015625,
      "learning_rate": 3.995013564485894e-06,
      "loss": 1.2619,
      "step": 55776
    },
    {
      "epoch": 4.798847113481889,
      "grad_norm": 1.1328125,
      "learning_rate": 3.991602844785392e-06,
      "loss": 1.9899,
      "step": 55777
    },
    {
      "epoch": 4.798933149789211,
      "grad_norm": 1.0703125,
      "learning_rate": 3.988193575812982e-06,
      "loss": 1.4884,
      "step": 55778
    },
    {
      "epoch": 4.799019186096533,
      "grad_norm": 1.234375,
      "learning_rate": 3.984785757578658e-06,
      "loss": 1.0184,
      "step": 55779
    },
    {
      "epoch": 4.799105222403854,
      "grad_norm": 1.1875,
      "learning_rate": 3.981379390092299e-06,
      "loss": 0.9906,
      "step": 55780
    },
    {
      "epoch": 4.7991912587111765,
      "grad_norm": 1.3046875,
      "learning_rate": 3.97797447336401e-06,
      "loss": 1.5441,
      "step": 55781
    },
    {
      "epoch": 4.799277295018498,
      "grad_norm": 0.8046875,
      "learning_rate": 3.974571007403616e-06,
      "loss": 0.6842,
      "step": 55782
    },
    {
      "epoch": 4.79936333132582,
      "grad_norm": 1.109375,
      "learning_rate": 3.971168992221163e-06,
      "loss": 0.8261,
      "step": 55783
    },
    {
      "epoch": 4.799449367633141,
      "grad_norm": 1.2734375,
      "learning_rate": 3.96776842782659e-06,
      "loss": 0.9592,
      "step": 55784
    },
    {
      "epoch": 4.799535403940463,
      "grad_norm": 0.9453125,
      "learning_rate": 3.964369314229776e-06,
      "loss": 1.022,
      "step": 55785
    },
    {
      "epoch": 4.799621440247785,
      "grad_norm": 0.97265625,
      "learning_rate": 3.960971651440714e-06,
      "loss": 0.9058,
      "step": 55786
    },
    {
      "epoch": 4.799707476555106,
      "grad_norm": 1.0703125,
      "learning_rate": 3.957575439469396e-06,
      "loss": 0.7341,
      "step": 55787
    },
    {
      "epoch": 4.799793512862428,
      "grad_norm": 2.203125,
      "learning_rate": 3.9541806783256475e-06,
      "loss": 1.177,
      "step": 55788
    },
    {
      "epoch": 4.7998795491697495,
      "grad_norm": 1.140625,
      "learning_rate": 3.950787368019404e-06,
      "loss": 1.6465,
      "step": 55789
    },
    {
      "epoch": 4.7999655854770715,
      "grad_norm": 1.1796875,
      "learning_rate": 3.947395508560714e-06,
      "loss": 0.9079,
      "step": 55790
    },
    {
      "epoch": 4.800051621784393,
      "grad_norm": 548.0,
      "learning_rate": 3.944005099959347e-06,
      "loss": 1.6629,
      "step": 55791
    },
    {
      "epoch": 4.800137658091715,
      "grad_norm": 16.0,
      "learning_rate": 3.940616142225295e-06,
      "loss": 0.5766,
      "step": 55792
    },
    {
      "epoch": 4.800223694399037,
      "grad_norm": 0.859375,
      "learning_rate": 3.937228635368495e-06,
      "loss": 0.9346,
      "step": 55793
    },
    {
      "epoch": 4.800309730706358,
      "grad_norm": 0.89453125,
      "learning_rate": 3.933842579398828e-06,
      "loss": 0.97,
      "step": 55794
    },
    {
      "epoch": 4.80039576701368,
      "grad_norm": 1.578125,
      "learning_rate": 3.930457974326118e-06,
      "loss": 1.2658,
      "step": 55795
    },
    {
      "epoch": 4.800481803321001,
      "grad_norm": 1.0390625,
      "learning_rate": 3.927074820160414e-06,
      "loss": 0.96,
      "step": 55796
    },
    {
      "epoch": 4.800567839628323,
      "grad_norm": 1.2578125,
      "learning_rate": 3.923693116911542e-06,
      "loss": 0.6772,
      "step": 55797
    },
    {
      "epoch": 4.8006538759356445,
      "grad_norm": 0.9765625,
      "learning_rate": 3.920312864589326e-06,
      "loss": 1.2831,
      "step": 55798
    },
    {
      "epoch": 4.800739912242967,
      "grad_norm": 1.046875,
      "learning_rate": 3.916934063203759e-06,
      "loss": 1.3144,
      "step": 55799
    },
    {
      "epoch": 4.800825948550289,
      "grad_norm": 4.875,
      "learning_rate": 3.9135567127646655e-06,
      "loss": 0.8911,
      "step": 55800
    },
    {
      "epoch": 4.80091198485761,
      "grad_norm": 0.99609375,
      "learning_rate": 3.910180813281927e-06,
      "loss": 0.976,
      "step": 55801
    },
    {
      "epoch": 4.800998021164932,
      "grad_norm": 1.5859375,
      "learning_rate": 3.906806364765425e-06,
      "loss": 1.0933,
      "step": 55802
    },
    {
      "epoch": 4.801084057472253,
      "grad_norm": 1.046875,
      "learning_rate": 3.90343336722504e-06,
      "loss": 1.3949,
      "step": 55803
    },
    {
      "epoch": 4.801170093779575,
      "grad_norm": 1.0625,
      "learning_rate": 3.900061820670653e-06,
      "loss": 1.016,
      "step": 55804
    },
    {
      "epoch": 4.801256130086896,
      "grad_norm": 1.46875,
      "learning_rate": 3.8966917251120335e-06,
      "loss": 1.0908,
      "step": 55805
    },
    {
      "epoch": 4.801342166394218,
      "grad_norm": 70.0,
      "learning_rate": 3.893323080559119e-06,
      "loss": 1.1495,
      "step": 55806
    },
    {
      "epoch": 4.8014282027015405,
      "grad_norm": 1.6640625,
      "learning_rate": 3.88995588702179e-06,
      "loss": 1.4917,
      "step": 55807
    },
    {
      "epoch": 4.801514239008862,
      "grad_norm": 1.515625,
      "learning_rate": 3.8865901445098164e-06,
      "loss": 1.4606,
      "step": 55808
    },
    {
      "epoch": 4.801600275316184,
      "grad_norm": 1.609375,
      "learning_rate": 3.883225853033079e-06,
      "loss": 2.0169,
      "step": 55809
    },
    {
      "epoch": 4.801686311623505,
      "grad_norm": 1.328125,
      "learning_rate": 3.879863012601459e-06,
      "loss": 0.8669,
      "step": 55810
    },
    {
      "epoch": 4.801772347930827,
      "grad_norm": 33.0,
      "learning_rate": 3.876501623224671e-06,
      "loss": 0.7594,
      "step": 55811
    },
    {
      "epoch": 4.801858384238148,
      "grad_norm": 1.1171875,
      "learning_rate": 3.873141684912706e-06,
      "loss": 1.2857,
      "step": 55812
    },
    {
      "epoch": 4.80194442054547,
      "grad_norm": 1.328125,
      "learning_rate": 3.869783197675281e-06,
      "loss": 0.9521,
      "step": 55813
    },
    {
      "epoch": 4.802030456852792,
      "grad_norm": 1.15625,
      "learning_rate": 3.866426161522218e-06,
      "loss": 0.8426,
      "step": 55814
    },
    {
      "epoch": 4.8021164931601135,
      "grad_norm": 0.828125,
      "learning_rate": 3.863070576463401e-06,
      "loss": 0.7449,
      "step": 55815
    },
    {
      "epoch": 4.8022025294674355,
      "grad_norm": 1.0234375,
      "learning_rate": 3.859716442508598e-06,
      "loss": 1.2059,
      "step": 55816
    },
    {
      "epoch": 4.802288565774757,
      "grad_norm": 7.125,
      "learning_rate": 3.856363759667636e-06,
      "loss": 1.3804,
      "step": 55817
    },
    {
      "epoch": 4.802374602082079,
      "grad_norm": 0.8671875,
      "learning_rate": 3.853012527950339e-06,
      "loss": 0.5975,
      "step": 55818
    },
    {
      "epoch": 4.8024606383894,
      "grad_norm": 2.046875,
      "learning_rate": 3.8496627473664225e-06,
      "loss": 1.8894,
      "step": 55819
    },
    {
      "epoch": 4.802546674696722,
      "grad_norm": 1.375,
      "learning_rate": 3.846314417925822e-06,
      "loss": 0.8505,
      "step": 55820
    },
    {
      "epoch": 4.802632711004044,
      "grad_norm": 0.9609375,
      "learning_rate": 3.842967539638198e-06,
      "loss": 0.9601,
      "step": 55821
    },
    {
      "epoch": 4.802718747311365,
      "grad_norm": 0.98046875,
      "learning_rate": 3.839622112513374e-06,
      "loss": 1.2837,
      "step": 55822
    },
    {
      "epoch": 4.802804783618687,
      "grad_norm": 1.046875,
      "learning_rate": 3.836278136561233e-06,
      "loss": 1.4337,
      "step": 55823
    },
    {
      "epoch": 4.8028908199260085,
      "grad_norm": 0.8671875,
      "learning_rate": 3.832935611791433e-06,
      "loss": 0.6013,
      "step": 55824
    },
    {
      "epoch": 4.802976856233331,
      "grad_norm": 0.9765625,
      "learning_rate": 3.829594538213799e-06,
      "loss": 1.0474,
      "step": 55825
    },
    {
      "epoch": 4.803062892540652,
      "grad_norm": 1.1796875,
      "learning_rate": 3.826254915838156e-06,
      "loss": 0.8498,
      "step": 55826
    },
    {
      "epoch": 4.803148928847974,
      "grad_norm": 6.15625,
      "learning_rate": 3.822916744674165e-06,
      "loss": 0.8127,
      "step": 55827
    },
    {
      "epoch": 4.803234965155296,
      "grad_norm": 1.765625,
      "learning_rate": 3.819580024731594e-06,
      "loss": 0.9453,
      "step": 55828
    },
    {
      "epoch": 4.803321001462617,
      "grad_norm": 1.4609375,
      "learning_rate": 3.816244756020326e-06,
      "loss": 1.3133,
      "step": 55829
    },
    {
      "epoch": 4.803407037769939,
      "grad_norm": 0.671875,
      "learning_rate": 3.8129109385500183e-06,
      "loss": 0.506,
      "step": 55830
    },
    {
      "epoch": 4.80349307407726,
      "grad_norm": 4.625,
      "learning_rate": 3.8095785723304967e-06,
      "loss": 1.4596,
      "step": 55831
    },
    {
      "epoch": 4.803579110384582,
      "grad_norm": 1.3828125,
      "learning_rate": 3.8062476573713643e-06,
      "loss": 1.5154,
      "step": 55832
    },
    {
      "epoch": 4.803665146691904,
      "grad_norm": 5.625,
      "learning_rate": 3.802918193682503e-06,
      "loss": 1.3442,
      "step": 55833
    },
    {
      "epoch": 4.803751182999226,
      "grad_norm": 0.703125,
      "learning_rate": 3.799590181273571e-06,
      "loss": 0.6052,
      "step": 55834
    },
    {
      "epoch": 4.803837219306548,
      "grad_norm": 0.83984375,
      "learning_rate": 3.796263620154339e-06,
      "loss": 1.0284,
      "step": 55835
    },
    {
      "epoch": 4.803923255613869,
      "grad_norm": 1.203125,
      "learning_rate": 3.7929385103345203e-06,
      "loss": 1.126,
      "step": 55836
    },
    {
      "epoch": 4.804009291921191,
      "grad_norm": 1.109375,
      "learning_rate": 3.789614851823886e-06,
      "loss": 1.3783,
      "step": 55837
    },
    {
      "epoch": 4.804095328228512,
      "grad_norm": 1.171875,
      "learning_rate": 3.7862926446320946e-06,
      "loss": 1.0362,
      "step": 55838
    },
    {
      "epoch": 4.804181364535834,
      "grad_norm": 21.875,
      "learning_rate": 3.782971888768916e-06,
      "loss": 1.2366,
      "step": 55839
    },
    {
      "epoch": 4.804267400843155,
      "grad_norm": 1.1640625,
      "learning_rate": 3.779652584244009e-06,
      "loss": 1.5424,
      "step": 55840
    },
    {
      "epoch": 4.8043534371504775,
      "grad_norm": 1.296875,
      "learning_rate": 3.7763347310671435e-06,
      "loss": 1.1639,
      "step": 55841
    },
    {
      "epoch": 4.8044394734577995,
      "grad_norm": 1.0546875,
      "learning_rate": 3.773018329247979e-06,
      "loss": 1.2125,
      "step": 55842
    },
    {
      "epoch": 4.804525509765121,
      "grad_norm": 1.3359375,
      "learning_rate": 3.7697033787961744e-06,
      "loss": 1.2945,
      "step": 55843
    },
    {
      "epoch": 4.804611546072443,
      "grad_norm": 0.9609375,
      "learning_rate": 3.766389879721499e-06,
      "loss": 0.5995,
      "step": 55844
    },
    {
      "epoch": 4.804697582379764,
      "grad_norm": 1.0625,
      "learning_rate": 3.7630778320336125e-06,
      "loss": 1.3259,
      "step": 55845
    },
    {
      "epoch": 4.804783618687086,
      "grad_norm": 0.8828125,
      "learning_rate": 3.7597672357422285e-06,
      "loss": 0.514,
      "step": 55846
    },
    {
      "epoch": 4.804869654994407,
      "grad_norm": 1.6171875,
      "learning_rate": 3.756458090857007e-06,
      "loss": 1.2781,
      "step": 55847
    },
    {
      "epoch": 4.804955691301729,
      "grad_norm": 1.3671875,
      "learning_rate": 3.7531503973876614e-06,
      "loss": 1.0905,
      "step": 55848
    },
    {
      "epoch": 4.805041727609051,
      "grad_norm": 1.046875,
      "learning_rate": 3.7498441553437957e-06,
      "loss": 1.3177,
      "step": 55849
    },
    {
      "epoch": 4.8051277639163725,
      "grad_norm": 0.84375,
      "learning_rate": 3.74653936473518e-06,
      "loss": 0.8185,
      "step": 55850
    },
    {
      "epoch": 4.805213800223695,
      "grad_norm": 1.359375,
      "learning_rate": 3.7432360255713616e-06,
      "loss": 0.7797,
      "step": 55851
    },
    {
      "epoch": 4.805299836531016,
      "grad_norm": 0.97265625,
      "learning_rate": 3.7399341378620555e-06,
      "loss": 0.481,
      "step": 55852
    },
    {
      "epoch": 4.805385872838338,
      "grad_norm": 1.0390625,
      "learning_rate": 3.736633701616976e-06,
      "loss": 0.9239,
      "step": 55853
    },
    {
      "epoch": 4.805471909145659,
      "grad_norm": 0.86328125,
      "learning_rate": 3.733334716845671e-06,
      "loss": 1.0458,
      "step": 55854
    },
    {
      "epoch": 4.805557945452981,
      "grad_norm": 1.7734375,
      "learning_rate": 3.7300371835578552e-06,
      "loss": 0.6825,
      "step": 55855
    },
    {
      "epoch": 4.805643981760303,
      "grad_norm": 6.46875,
      "learning_rate": 3.726741101763187e-06,
      "loss": 0.4605,
      "step": 55856
    },
    {
      "epoch": 4.805730018067624,
      "grad_norm": 0.8828125,
      "learning_rate": 3.72344647147127e-06,
      "loss": 0.4932,
      "step": 55857
    },
    {
      "epoch": 4.805816054374946,
      "grad_norm": 1.03125,
      "learning_rate": 3.7201532926917637e-06,
      "loss": 1.345,
      "step": 55858
    },
    {
      "epoch": 4.805902090682268,
      "grad_norm": 0.94140625,
      "learning_rate": 3.716861565434271e-06,
      "loss": 1.2544,
      "step": 55859
    },
    {
      "epoch": 4.80598812698959,
      "grad_norm": 1.015625,
      "learning_rate": 3.713571289708395e-06,
      "loss": 1.8436,
      "step": 55860
    },
    {
      "epoch": 4.806074163296911,
      "grad_norm": 1.0546875,
      "learning_rate": 3.7102824655239065e-06,
      "loss": 1.1381,
      "step": 55861
    },
    {
      "epoch": 4.806160199604233,
      "grad_norm": 1.0546875,
      "learning_rate": 3.706995092890242e-06,
      "loss": 1.7745,
      "step": 55862
    },
    {
      "epoch": 4.806246235911555,
      "grad_norm": 1336.0,
      "learning_rate": 3.7037091718171155e-06,
      "loss": 1.1179,
      "step": 55863
    },
    {
      "epoch": 4.806332272218876,
      "grad_norm": 1.0703125,
      "learning_rate": 3.700424702314076e-06,
      "loss": 1.4658,
      "step": 55864
    },
    {
      "epoch": 4.806418308526198,
      "grad_norm": 1.4375,
      "learning_rate": 3.697141684390837e-06,
      "loss": 1.9393,
      "step": 55865
    },
    {
      "epoch": 4.806504344833519,
      "grad_norm": 0.92578125,
      "learning_rate": 3.693860118056891e-06,
      "loss": 1.1374,
      "step": 55866
    },
    {
      "epoch": 4.8065903811408415,
      "grad_norm": 0.96484375,
      "learning_rate": 3.6905800033218974e-06,
      "loss": 1.29,
      "step": 55867
    },
    {
      "epoch": 4.806676417448163,
      "grad_norm": 142.0,
      "learning_rate": 3.6873013401954038e-06,
      "loss": 1.4644,
      "step": 55868
    },
    {
      "epoch": 4.806762453755485,
      "grad_norm": 7.03125,
      "learning_rate": 3.6840241286870137e-06,
      "loss": 0.6954,
      "step": 55869
    },
    {
      "epoch": 4.806848490062807,
      "grad_norm": 0.98828125,
      "learning_rate": 3.680748368806386e-06,
      "loss": 1.2697,
      "step": 55870
    },
    {
      "epoch": 4.806934526370128,
      "grad_norm": 1.25,
      "learning_rate": 3.677474060563013e-06,
      "loss": 1.3575,
      "step": 55871
    },
    {
      "epoch": 4.80702056267745,
      "grad_norm": 12.625,
      "learning_rate": 3.6742012039664986e-06,
      "loss": 0.8103,
      "step": 55872
    },
    {
      "epoch": 4.807106598984771,
      "grad_norm": 1.609375,
      "learning_rate": 3.6709297990263902e-06,
      "loss": 0.9441,
      "step": 55873
    },
    {
      "epoch": 4.807192635292093,
      "grad_norm": 1.1328125,
      "learning_rate": 3.6676598457522914e-06,
      "loss": 1.8893,
      "step": 55874
    },
    {
      "epoch": 4.8072786715994145,
      "grad_norm": 1.5703125,
      "learning_rate": 3.664391344153806e-06,
      "loss": 0.8934,
      "step": 55875
    },
    {
      "epoch": 4.8073647079067365,
      "grad_norm": 1.0703125,
      "learning_rate": 3.66112429424037e-06,
      "loss": 1.4351,
      "step": 55876
    },
    {
      "epoch": 4.807450744214059,
      "grad_norm": 0.83203125,
      "learning_rate": 3.6578586960216432e-06,
      "loss": 0.9463,
      "step": 55877
    },
    {
      "epoch": 4.80753678052138,
      "grad_norm": 2.640625,
      "learning_rate": 3.654594549507173e-06,
      "loss": 0.975,
      "step": 55878
    },
    {
      "epoch": 4.807622816828702,
      "grad_norm": 68.5,
      "learning_rate": 3.6513318547064524e-06,
      "loss": 0.9062,
      "step": 55879
    },
    {
      "epoch": 4.807708853136023,
      "grad_norm": 1.0703125,
      "learning_rate": 3.6480706116290283e-06,
      "loss": 1.0068,
      "step": 55880
    },
    {
      "epoch": 4.807794889443345,
      "grad_norm": 1.2890625,
      "learning_rate": 3.6448108202845053e-06,
      "loss": 0.8529,
      "step": 55881
    },
    {
      "epoch": 4.807880925750667,
      "grad_norm": 1.421875,
      "learning_rate": 3.6415524806823198e-06,
      "loss": 1.7749,
      "step": 55882
    },
    {
      "epoch": 4.807966962057988,
      "grad_norm": 1.0703125,
      "learning_rate": 3.6382955928321303e-06,
      "loss": 0.6397,
      "step": 55883
    },
    {
      "epoch": 4.80805299836531,
      "grad_norm": 1.1484375,
      "learning_rate": 3.6350401567433187e-06,
      "loss": 1.2288,
      "step": 55884
    },
    {
      "epoch": 4.808139034672632,
      "grad_norm": 1.71875,
      "learning_rate": 3.631786172425544e-06,
      "loss": 1.8206,
      "step": 55885
    },
    {
      "epoch": 4.808225070979954,
      "grad_norm": 0.9609375,
      "learning_rate": 3.628533639888243e-06,
      "loss": 1.4281,
      "step": 55886
    },
    {
      "epoch": 4.808311107287275,
      "grad_norm": 1.296875,
      "learning_rate": 3.625282559140908e-06,
      "loss": 0.9418,
      "step": 55887
    },
    {
      "epoch": 4.808397143594597,
      "grad_norm": 1.0703125,
      "learning_rate": 3.6220329301930867e-06,
      "loss": 1.3171,
      "step": 55888
    },
    {
      "epoch": 4.808483179901919,
      "grad_norm": 0.9453125,
      "learning_rate": 3.618784753054327e-06,
      "loss": 0.852,
      "step": 55889
    },
    {
      "epoch": 4.80856921620924,
      "grad_norm": 1.2578125,
      "learning_rate": 3.615538027734011e-06,
      "loss": 0.9184,
      "step": 55890
    },
    {
      "epoch": 4.808655252516562,
      "grad_norm": 1.46875,
      "learning_rate": 3.612292754241742e-06,
      "loss": 1.5229,
      "step": 55891
    },
    {
      "epoch": 4.808741288823883,
      "grad_norm": 1.1640625,
      "learning_rate": 3.609048932586956e-06,
      "loss": 0.5287,
      "step": 55892
    },
    {
      "epoch": 4.8088273251312055,
      "grad_norm": 1.046875,
      "learning_rate": 3.6058065627792014e-06,
      "loss": 1.3139,
      "step": 55893
    },
    {
      "epoch": 4.8089133614385275,
      "grad_norm": 1.2109375,
      "learning_rate": 3.6025656448279153e-06,
      "loss": 1.4248,
      "step": 55894
    },
    {
      "epoch": 4.808999397745849,
      "grad_norm": 0.80859375,
      "learning_rate": 3.599326178742535e-06,
      "loss": 0.9999,
      "step": 55895
    },
    {
      "epoch": 4.809085434053171,
      "grad_norm": 1.0390625,
      "learning_rate": 3.596088164532607e-06,
      "loss": 0.9878,
      "step": 55896
    },
    {
      "epoch": 4.809171470360492,
      "grad_norm": 1.1484375,
      "learning_rate": 3.5928516022075696e-06,
      "loss": 0.7024,
      "step": 55897
    },
    {
      "epoch": 4.809257506667814,
      "grad_norm": 1.1015625,
      "learning_rate": 3.589616491776915e-06,
      "loss": 1.1084,
      "step": 55898
    },
    {
      "epoch": 4.809343542975135,
      "grad_norm": 1.7421875,
      "learning_rate": 3.5863828332500792e-06,
      "loss": 1.6168,
      "step": 55899
    },
    {
      "epoch": 4.809429579282457,
      "grad_norm": 17.625,
      "learning_rate": 3.5831506266365e-06,
      "loss": 1.3031,
      "step": 55900
    },
    {
      "epoch": 4.809515615589779,
      "grad_norm": 0.98828125,
      "learning_rate": 3.5799198719456695e-06,
      "loss": 0.7368,
      "step": 55901
    },
    {
      "epoch": 4.8096016518971005,
      "grad_norm": 1.046875,
      "learning_rate": 3.5766905691870244e-06,
      "loss": 0.9702,
      "step": 55902
    },
    {
      "epoch": 4.809687688204423,
      "grad_norm": 1.0703125,
      "learning_rate": 3.5734627183700575e-06,
      "loss": 0.7826,
      "step": 55903
    },
    {
      "epoch": 4.809773724511744,
      "grad_norm": 0.98046875,
      "learning_rate": 3.570236319504094e-06,
      "loss": 1.3092,
      "step": 55904
    },
    {
      "epoch": 4.809859760819066,
      "grad_norm": 0.9453125,
      "learning_rate": 3.567011372598683e-06,
      "loss": 0.7832,
      "step": 55905
    },
    {
      "epoch": 4.809945797126387,
      "grad_norm": 0.87890625,
      "learning_rate": 3.563787877663205e-06,
      "loss": 1.0762,
      "step": 55906
    },
    {
      "epoch": 4.810031833433709,
      "grad_norm": 36.75,
      "learning_rate": 3.5605658347070415e-06,
      "loss": 0.8768,
      "step": 55907
    },
    {
      "epoch": 4.810117869741031,
      "grad_norm": 1.0234375,
      "learning_rate": 3.5573452437397403e-06,
      "loss": 1.0256,
      "step": 55908
    },
    {
      "epoch": 4.810203906048352,
      "grad_norm": 1.3984375,
      "learning_rate": 3.5541261047706274e-06,
      "loss": 0.9224,
      "step": 55909
    },
    {
      "epoch": 4.810289942355674,
      "grad_norm": 1.2109375,
      "learning_rate": 3.55090841780914e-06,
      "loss": 1.2709,
      "step": 55910
    },
    {
      "epoch": 4.810375978662996,
      "grad_norm": 1.109375,
      "learning_rate": 3.5476921828647145e-06,
      "loss": 1.308,
      "step": 55911
    },
    {
      "epoch": 4.810462014970318,
      "grad_norm": 0.95703125,
      "learning_rate": 3.5444773999466775e-06,
      "loss": 1.1921,
      "step": 55912
    },
    {
      "epoch": 4.810548051277639,
      "grad_norm": 1.1796875,
      "learning_rate": 3.541264069064576e-06,
      "loss": 1.5941,
      "step": 55913
    },
    {
      "epoch": 4.810634087584961,
      "grad_norm": 2.59375,
      "learning_rate": 3.5380521902276253e-06,
      "loss": 1.0703,
      "step": 55914
    },
    {
      "epoch": 4.810720123892283,
      "grad_norm": 1.2265625,
      "learning_rate": 3.5348417634453733e-06,
      "loss": 1.2171,
      "step": 55915
    },
    {
      "epoch": 4.810806160199604,
      "grad_norm": 47.75,
      "learning_rate": 3.531632788727146e-06,
      "loss": 1.1432,
      "step": 55916
    },
    {
      "epoch": 4.810892196506926,
      "grad_norm": 1.3125,
      "learning_rate": 3.5284252660823248e-06,
      "loss": 1.8416,
      "step": 55917
    },
    {
      "epoch": 4.810978232814247,
      "grad_norm": 30.0,
      "learning_rate": 3.5252191955202904e-06,
      "loss": 0.8371,
      "step": 55918
    },
    {
      "epoch": 4.8110642691215695,
      "grad_norm": 1.015625,
      "learning_rate": 3.5220145770504808e-06,
      "loss": 1.2822,
      "step": 55919
    },
    {
      "epoch": 4.811150305428891,
      "grad_norm": 59.75,
      "learning_rate": 3.518811410682166e-06,
      "loss": 1.1229,
      "step": 55920
    },
    {
      "epoch": 4.811236341736213,
      "grad_norm": 2.328125,
      "learning_rate": 3.515609696424782e-06,
      "loss": 0.7889,
      "step": 55921
    },
    {
      "epoch": 4.811322378043535,
      "grad_norm": 1.0,
      "learning_rate": 3.5124094342876555e-06,
      "loss": 0.7999,
      "step": 55922
    },
    {
      "epoch": 4.811408414350856,
      "grad_norm": 1.03125,
      "learning_rate": 3.5092106242802237e-06,
      "loss": 0.8465,
      "step": 55923
    },
    {
      "epoch": 4.811494450658178,
      "grad_norm": 1.5078125,
      "learning_rate": 3.5060132664117563e-06,
      "loss": 0.6864,
      "step": 55924
    },
    {
      "epoch": 4.811580486965499,
      "grad_norm": 1.34375,
      "learning_rate": 3.5028173606916346e-06,
      "loss": 2.1945,
      "step": 55925
    },
    {
      "epoch": 4.811666523272821,
      "grad_norm": 1.203125,
      "learning_rate": 3.499622907129241e-06,
      "loss": 1.0649,
      "step": 55926
    },
    {
      "epoch": 4.8117525595801425,
      "grad_norm": 1.3125,
      "learning_rate": 3.4964299057338444e-06,
      "loss": 0.8247,
      "step": 55927
    },
    {
      "epoch": 4.8118385958874645,
      "grad_norm": 0.92578125,
      "learning_rate": 3.4932383565148276e-06,
      "loss": 0.8783,
      "step": 55928
    },
    {
      "epoch": 4.811924632194787,
      "grad_norm": 1.2734375,
      "learning_rate": 3.4900482594815707e-06,
      "loss": 1.4893,
      "step": 55929
    },
    {
      "epoch": 4.812010668502108,
      "grad_norm": 1.015625,
      "learning_rate": 3.486859614643345e-06,
      "loss": 1.3001,
      "step": 55930
    },
    {
      "epoch": 4.81209670480943,
      "grad_norm": 1.125,
      "learning_rate": 3.4836724220094206e-06,
      "loss": 1.3161,
      "step": 55931
    },
    {
      "epoch": 4.812182741116751,
      "grad_norm": 1.3046875,
      "learning_rate": 3.48048668158929e-06,
      "loss": 1.5292,
      "step": 55932
    },
    {
      "epoch": 4.812268777424073,
      "grad_norm": 1.0625,
      "learning_rate": 3.477302393392112e-06,
      "loss": 0.6938,
      "step": 55933
    },
    {
      "epoch": 4.812354813731394,
      "grad_norm": 8.8125,
      "learning_rate": 3.4741195574272134e-06,
      "loss": 1.2147,
      "step": 55934
    },
    {
      "epoch": 4.812440850038716,
      "grad_norm": 466.0,
      "learning_rate": 3.47093817370403e-06,
      "loss": 0.8755,
      "step": 55935
    },
    {
      "epoch": 4.812526886346038,
      "grad_norm": 1.109375,
      "learning_rate": 3.4677582422317224e-06,
      "loss": 1.2705,
      "step": 55936
    },
    {
      "epoch": 4.81261292265336,
      "grad_norm": 1.6015625,
      "learning_rate": 3.464579763019671e-06,
      "loss": 1.4495,
      "step": 55937
    },
    {
      "epoch": 4.812698958960682,
      "grad_norm": 1.0625,
      "learning_rate": 3.4614027360771463e-06,
      "loss": 0.8027,
      "step": 55938
    },
    {
      "epoch": 4.812784995268003,
      "grad_norm": 1.3671875,
      "learning_rate": 3.458227161413474e-06,
      "loss": 0.8874,
      "step": 55939
    },
    {
      "epoch": 4.812871031575325,
      "grad_norm": 0.89453125,
      "learning_rate": 3.4550530390379252e-06,
      "loss": 0.7506,
      "step": 55940
    },
    {
      "epoch": 4.812957067882646,
      "grad_norm": 1.1875,
      "learning_rate": 3.4518803689597145e-06,
      "loss": 0.9717,
      "step": 55941
    },
    {
      "epoch": 4.813043104189968,
      "grad_norm": 1.2265625,
      "learning_rate": 3.448709151188223e-06,
      "loss": 1.1548,
      "step": 55942
    },
    {
      "epoch": 4.81312914049729,
      "grad_norm": 1.015625,
      "learning_rate": 3.4455393857326656e-06,
      "loss": 0.7701,
      "step": 55943
    },
    {
      "epoch": 4.813215176804611,
      "grad_norm": 1.1484375,
      "learning_rate": 3.4423710726023683e-06,
      "loss": 1.3118,
      "step": 55944
    },
    {
      "epoch": 4.8133012131119335,
      "grad_norm": 0.8828125,
      "learning_rate": 3.4392042118064903e-06,
      "loss": 0.7157,
      "step": 55945
    },
    {
      "epoch": 4.813387249419255,
      "grad_norm": 1.125,
      "learning_rate": 3.436038803354413e-06,
      "loss": 1.4363,
      "step": 55946
    },
    {
      "epoch": 4.813473285726577,
      "grad_norm": 0.9140625,
      "learning_rate": 3.4328748472553518e-06,
      "loss": 0.7967,
      "step": 55947
    },
    {
      "epoch": 4.813559322033898,
      "grad_norm": 1.09375,
      "learning_rate": 3.4297123435185205e-06,
      "loss": 0.5953,
      "step": 55948
    },
    {
      "epoch": 4.81364535834122,
      "grad_norm": 185.0,
      "learning_rate": 3.426551292153246e-06,
      "loss": 1.5653,
      "step": 55949
    },
    {
      "epoch": 4.813731394648542,
      "grad_norm": 0.83984375,
      "learning_rate": 3.423391693168687e-06,
      "loss": 0.5074,
      "step": 55950
    },
    {
      "epoch": 4.813817430955863,
      "grad_norm": 8.5,
      "learning_rate": 3.4202335465741696e-06,
      "loss": 1.2063,
      "step": 55951
    },
    {
      "epoch": 4.813903467263185,
      "grad_norm": 1.71875,
      "learning_rate": 3.417076852378853e-06,
      "loss": 0.6667,
      "step": 55952
    },
    {
      "epoch": 4.8139895035705065,
      "grad_norm": 1.28125,
      "learning_rate": 3.4139216105920077e-06,
      "loss": 1.9263,
      "step": 55953
    },
    {
      "epoch": 4.8140755398778285,
      "grad_norm": 0.92578125,
      "learning_rate": 3.4107678212229044e-06,
      "loss": 0.858,
      "step": 55954
    },
    {
      "epoch": 4.81416157618515,
      "grad_norm": 1.0390625,
      "learning_rate": 3.407615484280646e-06,
      "loss": 1.2179,
      "step": 55955
    },
    {
      "epoch": 4.814247612492472,
      "grad_norm": 5.34375,
      "learning_rate": 3.4044645997745593e-06,
      "loss": 1.1026,
      "step": 55956
    },
    {
      "epoch": 4.814333648799794,
      "grad_norm": 1.09375,
      "learning_rate": 3.401315167713914e-06,
      "loss": 0.6036,
      "step": 55957
    },
    {
      "epoch": 4.814419685107115,
      "grad_norm": 1.2734375,
      "learning_rate": 3.398167188107759e-06,
      "loss": 0.9036,
      "step": 55958
    },
    {
      "epoch": 4.814505721414437,
      "grad_norm": 0.96484375,
      "learning_rate": 3.39502066096542e-06,
      "loss": 0.9389,
      "step": 55959
    },
    {
      "epoch": 4.814591757721758,
      "grad_norm": 1.109375,
      "learning_rate": 3.3918755862960006e-06,
      "loss": 0.9987,
      "step": 55960
    },
    {
      "epoch": 4.81467779402908,
      "grad_norm": 1.3359375,
      "learning_rate": 3.3887319641088265e-06,
      "loss": 1.5558,
      "step": 55961
    },
    {
      "epoch": 4.8147638303364015,
      "grad_norm": 1.2578125,
      "learning_rate": 3.385589794413002e-06,
      "loss": 1.3562,
      "step": 55962
    },
    {
      "epoch": 4.814849866643724,
      "grad_norm": 1.46875,
      "learning_rate": 3.3824490772177417e-06,
      "loss": 0.8598,
      "step": 55963
    },
    {
      "epoch": 4.814935902951046,
      "grad_norm": 0.875,
      "learning_rate": 3.379309812532205e-06,
      "loss": 0.4343,
      "step": 55964
    },
    {
      "epoch": 4.815021939258367,
      "grad_norm": 1.296875,
      "learning_rate": 3.376172000365607e-06,
      "loss": 1.4455,
      "step": 55965
    },
    {
      "epoch": 4.815107975565689,
      "grad_norm": 1.390625,
      "learning_rate": 3.3730356407271624e-06,
      "loss": 0.9369,
      "step": 55966
    },
    {
      "epoch": 4.81519401187301,
      "grad_norm": 28.5,
      "learning_rate": 3.369900733625919e-06,
      "loss": 1.0869,
      "step": 55967
    },
    {
      "epoch": 4.815280048180332,
      "grad_norm": 0.80078125,
      "learning_rate": 3.3667672790712034e-06,
      "loss": 0.8597,
      "step": 55968
    },
    {
      "epoch": 4.815366084487653,
      "grad_norm": 1.0859375,
      "learning_rate": 3.3636352770720637e-06,
      "loss": 0.5797,
      "step": 55969
    },
    {
      "epoch": 4.815452120794975,
      "grad_norm": 0.73046875,
      "learning_rate": 3.36050472763777e-06,
      "loss": 0.4208,
      "step": 55970
    },
    {
      "epoch": 4.8155381571022975,
      "grad_norm": 1.0390625,
      "learning_rate": 3.357375630777315e-06,
      "loss": 1.0201,
      "step": 55971
    },
    {
      "epoch": 4.815624193409619,
      "grad_norm": 1.1953125,
      "learning_rate": 3.35424798649997e-06,
      "loss": 2.119,
      "step": 55972
    },
    {
      "epoch": 4.815710229716941,
      "grad_norm": 1.1328125,
      "learning_rate": 3.351121794814893e-06,
      "loss": 1.3876,
      "step": 55973
    },
    {
      "epoch": 4.815796266024262,
      "grad_norm": 2.828125,
      "learning_rate": 3.347997055731189e-06,
      "loss": 1.3571,
      "step": 55974
    },
    {
      "epoch": 4.815882302331584,
      "grad_norm": 1.15625,
      "learning_rate": 3.3448737692579613e-06,
      "loss": 1.3185,
      "step": 55975
    },
    {
      "epoch": 4.815968338638905,
      "grad_norm": 0.984375,
      "learning_rate": 3.3417519354044245e-06,
      "loss": 0.7142,
      "step": 55976
    },
    {
      "epoch": 4.816054374946227,
      "grad_norm": 1.09375,
      "learning_rate": 3.3386315541796273e-06,
      "loss": 1.5722,
      "step": 55977
    },
    {
      "epoch": 4.816140411253549,
      "grad_norm": 1.6171875,
      "learning_rate": 3.335512625592729e-06,
      "loss": 1.2868,
      "step": 55978
    },
    {
      "epoch": 4.8162264475608705,
      "grad_norm": 0.97265625,
      "learning_rate": 3.332395149652945e-06,
      "loss": 1.23,
      "step": 55979
    },
    {
      "epoch": 4.8163124838681926,
      "grad_norm": 16.625,
      "learning_rate": 3.3292791263692114e-06,
      "loss": 0.7634,
      "step": 55980
    },
    {
      "epoch": 4.816398520175514,
      "grad_norm": 1.2890625,
      "learning_rate": 3.3261645557507993e-06,
      "loss": 1.1151,
      "step": 55981
    },
    {
      "epoch": 4.816484556482836,
      "grad_norm": 0.9375,
      "learning_rate": 3.323051437806757e-06,
      "loss": 1.6415,
      "step": 55982
    },
    {
      "epoch": 4.816570592790157,
      "grad_norm": 0.99609375,
      "learning_rate": 3.319939772546132e-06,
      "loss": 0.5556,
      "step": 55983
    },
    {
      "epoch": 4.816656629097479,
      "grad_norm": 4.15625,
      "learning_rate": 3.3168295599781404e-06,
      "loss": 1.0084,
      "step": 55984
    },
    {
      "epoch": 4.816742665404801,
      "grad_norm": 1.25,
      "learning_rate": 3.31372080011183e-06,
      "loss": 0.8092,
      "step": 55985
    },
    {
      "epoch": 4.816828701712122,
      "grad_norm": 1.0078125,
      "learning_rate": 3.3106134929562493e-06,
      "loss": 0.7791,
      "step": 55986
    },
    {
      "epoch": 4.816914738019444,
      "grad_norm": 1.4609375,
      "learning_rate": 3.3075076385205573e-06,
      "loss": 1.5039,
      "step": 55987
    },
    {
      "epoch": 4.8170007743267655,
      "grad_norm": 59.75,
      "learning_rate": 3.304403236813802e-06,
      "loss": 1.3171,
      "step": 55988
    },
    {
      "epoch": 4.817086810634088,
      "grad_norm": 0.8828125,
      "learning_rate": 3.3013002878450326e-06,
      "loss": 1.4127,
      "step": 55989
    },
    {
      "epoch": 4.817172846941409,
      "grad_norm": 1.3515625,
      "learning_rate": 3.2981987916233523e-06,
      "loss": 0.7761,
      "step": 55990
    },
    {
      "epoch": 4.817258883248731,
      "grad_norm": 19.75,
      "learning_rate": 3.2950987481578655e-06,
      "loss": 0.7534,
      "step": 55991
    },
    {
      "epoch": 4.817344919556053,
      "grad_norm": 1.0546875,
      "learning_rate": 3.2920001574576197e-06,
      "loss": 1.4025,
      "step": 55992
    },
    {
      "epoch": 4.817430955863374,
      "grad_norm": 49.75,
      "learning_rate": 3.2889030195316085e-06,
      "loss": 1.2703,
      "step": 55993
    },
    {
      "epoch": 4.817516992170696,
      "grad_norm": 1.2734375,
      "learning_rate": 3.2858073343889906e-06,
      "loss": 1.4385,
      "step": 55994
    },
    {
      "epoch": 4.817603028478017,
      "grad_norm": 1.0546875,
      "learning_rate": 3.2827131020388147e-06,
      "loss": 1.4771,
      "step": 55995
    },
    {
      "epoch": 4.817689064785339,
      "grad_norm": 1.046875,
      "learning_rate": 3.2796203224900735e-06,
      "loss": 0.9204,
      "step": 55996
    },
    {
      "epoch": 4.817775101092661,
      "grad_norm": 1.3828125,
      "learning_rate": 3.276528995751815e-06,
      "loss": 1.4498,
      "step": 55997
    },
    {
      "epoch": 4.817861137399983,
      "grad_norm": 1.6484375,
      "learning_rate": 3.273439121833144e-06,
      "loss": 0.9227,
      "step": 55998
    },
    {
      "epoch": 4.817947173707305,
      "grad_norm": 1.1328125,
      "learning_rate": 3.2703507007429965e-06,
      "loss": 1.2438,
      "step": 55999
    },
    {
      "epoch": 4.818033210014626,
      "grad_norm": 1.2890625,
      "learning_rate": 3.267263732490533e-06,
      "loss": 1.2452,
      "step": 56000
    },
    {
      "epoch": 4.818119246321948,
      "grad_norm": 0.8984375,
      "learning_rate": 3.2641782170846348e-06,
      "loss": 1.0827,
      "step": 56001
    },
    {
      "epoch": 4.818205282629269,
      "grad_norm": 1.0625,
      "learning_rate": 3.2610941545344606e-06,
      "loss": 1.0725,
      "step": 56002
    },
    {
      "epoch": 4.818291318936591,
      "grad_norm": 1.3515625,
      "learning_rate": 3.2580115448490043e-06,
      "loss": 0.8324,
      "step": 56003
    },
    {
      "epoch": 4.818377355243913,
      "grad_norm": 0.9609375,
      "learning_rate": 3.2549303880372026e-06,
      "loss": 1.1863,
      "step": 56004
    },
    {
      "epoch": 4.8184633915512345,
      "grad_norm": 1.4140625,
      "learning_rate": 3.2518506841081043e-06,
      "loss": 1.1118,
      "step": 56005
    },
    {
      "epoch": 4.8185494278585566,
      "grad_norm": 1.171875,
      "learning_rate": 3.248772433070757e-06,
      "loss": 1.0743,
      "step": 56006
    },
    {
      "epoch": 4.818635464165878,
      "grad_norm": 0.80859375,
      "learning_rate": 3.245695634934154e-06,
      "loss": 0.6557,
      "step": 56007
    },
    {
      "epoch": 4.8187215004732,
      "grad_norm": 0.90625,
      "learning_rate": 3.242620289707232e-06,
      "loss": 0.6566,
      "step": 56008
    },
    {
      "epoch": 4.818807536780522,
      "grad_norm": 1.0390625,
      "learning_rate": 3.239546397399096e-06,
      "loss": 1.0865,
      "step": 56009
    },
    {
      "epoch": 4.818893573087843,
      "grad_norm": 0.90625,
      "learning_rate": 3.2364739580186263e-06,
      "loss": 0.5717,
      "step": 56010
    },
    {
      "epoch": 4.818979609395165,
      "grad_norm": 28.875,
      "learning_rate": 3.2334029715748725e-06,
      "loss": 0.5746,
      "step": 56011
    },
    {
      "epoch": 4.819065645702486,
      "grad_norm": 1.4296875,
      "learning_rate": 3.2303334380767714e-06,
      "loss": 1.2891,
      "step": 56012
    },
    {
      "epoch": 4.819151682009808,
      "grad_norm": 1.0859375,
      "learning_rate": 3.2272653575333157e-06,
      "loss": 0.7355,
      "step": 56013
    },
    {
      "epoch": 4.8192377183171295,
      "grad_norm": 1.34375,
      "learning_rate": 3.2241987299535536e-06,
      "loss": 1.0178,
      "step": 56014
    },
    {
      "epoch": 4.819323754624452,
      "grad_norm": 11.5,
      "learning_rate": 3.2211335553463673e-06,
      "loss": 0.6469,
      "step": 56015
    },
    {
      "epoch": 4.819409790931774,
      "grad_norm": 1.0546875,
      "learning_rate": 3.218069833720749e-06,
      "loss": 1.4528,
      "step": 56016
    },
    {
      "epoch": 4.819495827239095,
      "grad_norm": 1.171875,
      "learning_rate": 3.215007565085637e-06,
      "loss": 1.4307,
      "step": 56017
    },
    {
      "epoch": 4.819581863546417,
      "grad_norm": 0.90234375,
      "learning_rate": 3.2119467494500233e-06,
      "loss": 1.229,
      "step": 56018
    },
    {
      "epoch": 4.819667899853738,
      "grad_norm": 120.5,
      "learning_rate": 3.208887386822845e-06,
      "loss": 1.4065,
      "step": 56019
    },
    {
      "epoch": 4.81975393616106,
      "grad_norm": 1.3125,
      "learning_rate": 3.20582947721304e-06,
      "loss": 0.7653,
      "step": 56020
    },
    {
      "epoch": 4.819839972468381,
      "grad_norm": 1.1484375,
      "learning_rate": 3.2027730206296013e-06,
      "loss": 0.828,
      "step": 56021
    },
    {
      "epoch": 4.819926008775703,
      "grad_norm": 1.2265625,
      "learning_rate": 3.1997180170813546e-06,
      "loss": 0.7785,
      "step": 56022
    },
    {
      "epoch": 4.8200120450830255,
      "grad_norm": 0.86328125,
      "learning_rate": 3.1966644665774036e-06,
      "loss": 0.6391,
      "step": 56023
    },
    {
      "epoch": 4.820098081390347,
      "grad_norm": 0.77734375,
      "learning_rate": 3.1936123691265197e-06,
      "loss": 1.0008,
      "step": 56024
    },
    {
      "epoch": 4.820184117697669,
      "grad_norm": 1.09375,
      "learning_rate": 3.1905617247377505e-06,
      "loss": 0.6454,
      "step": 56025
    },
    {
      "epoch": 4.82027015400499,
      "grad_norm": 17.625,
      "learning_rate": 3.187512533419923e-06,
      "loss": 1.1709,
      "step": 56026
    },
    {
      "epoch": 4.820356190312312,
      "grad_norm": 1.0078125,
      "learning_rate": 3.1844647951819736e-06,
      "loss": 0.5126,
      "step": 56027
    },
    {
      "epoch": 4.820442226619633,
      "grad_norm": 1.1015625,
      "learning_rate": 3.181418510032841e-06,
      "loss": 0.6568,
      "step": 56028
    },
    {
      "epoch": 4.820528262926955,
      "grad_norm": 7.1875,
      "learning_rate": 3.178373677981461e-06,
      "loss": 1.6636,
      "step": 56029
    },
    {
      "epoch": 4.820614299234277,
      "grad_norm": 1.234375,
      "learning_rate": 3.1753302990367716e-06,
      "loss": 0.9235,
      "step": 56030
    },
    {
      "epoch": 4.8207003355415985,
      "grad_norm": 1.0390625,
      "learning_rate": 3.1722883732075436e-06,
      "loss": 0.7674,
      "step": 56031
    },
    {
      "epoch": 4.820786371848921,
      "grad_norm": 1.0234375,
      "learning_rate": 3.1692479005027143e-06,
      "loss": 0.5422,
      "step": 56032
    },
    {
      "epoch": 4.820872408156242,
      "grad_norm": 0.8515625,
      "learning_rate": 3.166208880931276e-06,
      "loss": 1.0188,
      "step": 56033
    },
    {
      "epoch": 4.820958444463564,
      "grad_norm": 0.98828125,
      "learning_rate": 3.1631713145020004e-06,
      "loss": 1.4532,
      "step": 56034
    },
    {
      "epoch": 4.821044480770885,
      "grad_norm": 1.3359375,
      "learning_rate": 3.160135201223824e-06,
      "loss": 1.5766,
      "step": 56035
    },
    {
      "epoch": 4.821130517078207,
      "grad_norm": 378.0,
      "learning_rate": 3.1571005411056843e-06,
      "loss": 1.5474,
      "step": 56036
    },
    {
      "epoch": 4.821216553385529,
      "grad_norm": 1.34375,
      "learning_rate": 3.154067334156352e-06,
      "loss": 0.6945,
      "step": 56037
    },
    {
      "epoch": 4.82130258969285,
      "grad_norm": 1.1484375,
      "learning_rate": 3.151035580384709e-06,
      "loss": 0.9588,
      "step": 56038
    },
    {
      "epoch": 4.821388626000172,
      "grad_norm": 0.98046875,
      "learning_rate": 3.1480052797996927e-06,
      "loss": 1.107,
      "step": 56039
    },
    {
      "epoch": 4.8214746623074936,
      "grad_norm": 1.015625,
      "learning_rate": 3.144976432410129e-06,
      "loss": 1.112,
      "step": 56040
    },
    {
      "epoch": 4.821560698614816,
      "grad_norm": 1.1953125,
      "learning_rate": 3.141949038224845e-06,
      "loss": 1.7698,
      "step": 56041
    },
    {
      "epoch": 4.821646734922137,
      "grad_norm": 1.078125,
      "learning_rate": 3.138923097252777e-06,
      "loss": 1.6348,
      "step": 56042
    },
    {
      "epoch": 4.821732771229459,
      "grad_norm": 0.94921875,
      "learning_rate": 3.1358986095026965e-06,
      "loss": 0.9128,
      "step": 56043
    },
    {
      "epoch": 4.821818807536781,
      "grad_norm": 1.203125,
      "learning_rate": 3.1328755749834847e-06,
      "loss": 1.4061,
      "step": 56044
    },
    {
      "epoch": 4.821904843844102,
      "grad_norm": 0.82421875,
      "learning_rate": 3.1298539937039684e-06,
      "loss": 0.5931,
      "step": 56045
    },
    {
      "epoch": 4.821990880151424,
      "grad_norm": 1.046875,
      "learning_rate": 3.1268338656729734e-06,
      "loss": 0.8834,
      "step": 56046
    },
    {
      "epoch": 4.822076916458745,
      "grad_norm": 1.3125,
      "learning_rate": 3.1238151908993816e-06,
      "loss": 1.3356,
      "step": 56047
    },
    {
      "epoch": 4.822162952766067,
      "grad_norm": 1.1171875,
      "learning_rate": 3.120797969391964e-06,
      "loss": 1.1085,
      "step": 56048
    },
    {
      "epoch": 4.822248989073389,
      "grad_norm": 7.53125,
      "learning_rate": 3.1177822011596023e-06,
      "loss": 0.9625,
      "step": 56049
    },
    {
      "epoch": 4.822335025380711,
      "grad_norm": 20.0,
      "learning_rate": 3.114767886211123e-06,
      "loss": 0.73,
      "step": 56050
    },
    {
      "epoch": 4.822421061688033,
      "grad_norm": 1.7421875,
      "learning_rate": 3.1117550245552406e-06,
      "loss": 1.0034,
      "step": 56051
    },
    {
      "epoch": 4.822507097995354,
      "grad_norm": 0.984375,
      "learning_rate": 3.108743616200893e-06,
      "loss": 1.2419,
      "step": 56052
    },
    {
      "epoch": 4.822593134302676,
      "grad_norm": 1.4921875,
      "learning_rate": 3.1057336611567953e-06,
      "loss": 1.3259,
      "step": 56053
    },
    {
      "epoch": 4.822679170609997,
      "grad_norm": 1.78125,
      "learning_rate": 3.102725159431774e-06,
      "loss": 0.5273,
      "step": 56054
    },
    {
      "epoch": 4.822765206917319,
      "grad_norm": 1.4375,
      "learning_rate": 3.099718111034655e-06,
      "loss": 0.4763,
      "step": 56055
    },
    {
      "epoch": 4.82285124322464,
      "grad_norm": 0.89453125,
      "learning_rate": 3.096712515974209e-06,
      "loss": 0.6854,
      "step": 56056
    },
    {
      "epoch": 4.8229372795319625,
      "grad_norm": 133.0,
      "learning_rate": 3.093708374259263e-06,
      "loss": 0.7579,
      "step": 56057
    },
    {
      "epoch": 4.823023315839285,
      "grad_norm": 1.3125,
      "learning_rate": 3.0907056858985873e-06,
      "loss": 1.4833,
      "step": 56058
    },
    {
      "epoch": 4.823109352146606,
      "grad_norm": 256.0,
      "learning_rate": 3.0877044509008966e-06,
      "loss": 1.2417,
      "step": 56059
    },
    {
      "epoch": 4.823195388453928,
      "grad_norm": 1.2578125,
      "learning_rate": 3.0847046692750737e-06,
      "loss": 1.3028,
      "step": 56060
    },
    {
      "epoch": 4.823281424761249,
      "grad_norm": 1.140625,
      "learning_rate": 3.081706341029833e-06,
      "loss": 1.8781,
      "step": 56061
    },
    {
      "epoch": 4.823367461068571,
      "grad_norm": 0.921875,
      "learning_rate": 3.0787094661739455e-06,
      "loss": 1.2706,
      "step": 56062
    },
    {
      "epoch": 4.823453497375892,
      "grad_norm": 980.0,
      "learning_rate": 3.0757140447161826e-06,
      "loss": 1.4436,
      "step": 56063
    },
    {
      "epoch": 4.823539533683214,
      "grad_norm": 1.21875,
      "learning_rate": 3.0727200766653696e-06,
      "loss": 1.0098,
      "step": 56064
    },
    {
      "epoch": 4.823625569990536,
      "grad_norm": 1.0078125,
      "learning_rate": 3.0697275620301114e-06,
      "loss": 0.7573,
      "step": 56065
    },
    {
      "epoch": 4.823711606297858,
      "grad_norm": 1.15625,
      "learning_rate": 3.0667365008192893e-06,
      "loss": 0.8706,
      "step": 56066
    },
    {
      "epoch": 4.82379764260518,
      "grad_norm": 1.0859375,
      "learning_rate": 3.063746893041619e-06,
      "loss": 1.2852,
      "step": 56067
    },
    {
      "epoch": 4.823883678912501,
      "grad_norm": 1.0703125,
      "learning_rate": 3.0607587387058156e-06,
      "loss": 0.5971,
      "step": 56068
    },
    {
      "epoch": 4.823969715219823,
      "grad_norm": 1.0859375,
      "learning_rate": 3.0577720378207053e-06,
      "loss": 1.0645,
      "step": 56069
    },
    {
      "epoch": 4.824055751527144,
      "grad_norm": 1.265625,
      "learning_rate": 3.054786790394948e-06,
      "loss": 0.6055,
      "step": 56070
    },
    {
      "epoch": 4.824141787834466,
      "grad_norm": 1.34375,
      "learning_rate": 3.0518029964372583e-06,
      "loss": 1.0517,
      "step": 56071
    },
    {
      "epoch": 4.824227824141788,
      "grad_norm": 1.078125,
      "learning_rate": 3.048820655956408e-06,
      "loss": 0.5824,
      "step": 56072
    },
    {
      "epoch": 4.824313860449109,
      "grad_norm": 1.21875,
      "learning_rate": 3.0458397689611117e-06,
      "loss": 1.4025,
      "step": 56073
    },
    {
      "epoch": 4.824399896756431,
      "grad_norm": 1.0390625,
      "learning_rate": 3.042860335460085e-06,
      "loss": 1.1985,
      "step": 56074
    },
    {
      "epoch": 4.824485933063753,
      "grad_norm": 1.0078125,
      "learning_rate": 3.039882355461987e-06,
      "loss": 0.6364,
      "step": 56075
    },
    {
      "epoch": 4.824571969371075,
      "grad_norm": 212.0,
      "learning_rate": 3.036905828975645e-06,
      "loss": 1.2408,
      "step": 56076
    },
    {
      "epoch": 4.824658005678396,
      "grad_norm": 1.1484375,
      "learning_rate": 3.0339307560096617e-06,
      "loss": 0.9416,
      "step": 56077
    },
    {
      "epoch": 4.824744041985718,
      "grad_norm": 1.5,
      "learning_rate": 3.0309571365728096e-06,
      "loss": 1.6734,
      "step": 56078
    },
    {
      "epoch": 4.82483007829304,
      "grad_norm": 1.078125,
      "learning_rate": 3.027984970673692e-06,
      "loss": 0.8525,
      "step": 56079
    },
    {
      "epoch": 4.824916114600361,
      "grad_norm": 1.0546875,
      "learning_rate": 3.025014258321135e-06,
      "loss": 1.4846,
      "step": 56080
    },
    {
      "epoch": 4.825002150907683,
      "grad_norm": 0.99609375,
      "learning_rate": 3.0220449995236878e-06,
      "loss": 0.8049,
      "step": 56081
    },
    {
      "epoch": 4.825088187215004,
      "grad_norm": 0.9609375,
      "learning_rate": 3.0190771942901762e-06,
      "loss": 1.0241,
      "step": 56082
    },
    {
      "epoch": 4.8251742235223265,
      "grad_norm": 1.0625,
      "learning_rate": 3.0161108426291493e-06,
      "loss": 1.9155,
      "step": 56083
    },
    {
      "epoch": 4.825260259829648,
      "grad_norm": 1.015625,
      "learning_rate": 3.013145944549378e-06,
      "loss": 0.9929,
      "step": 56084
    },
    {
      "epoch": 4.82534629613697,
      "grad_norm": 356.0,
      "learning_rate": 3.0101825000594664e-06,
      "loss": 0.8658,
      "step": 56085
    },
    {
      "epoch": 4.825432332444292,
      "grad_norm": 1.5234375,
      "learning_rate": 3.007220509168129e-06,
      "loss": 1.4682,
      "step": 56086
    },
    {
      "epoch": 4.825518368751613,
      "grad_norm": 1.359375,
      "learning_rate": 3.0042599718839716e-06,
      "loss": 1.8313,
      "step": 56087
    },
    {
      "epoch": 4.825604405058935,
      "grad_norm": 1.1171875,
      "learning_rate": 3.0013008882157632e-06,
      "loss": 1.6489,
      "step": 56088
    },
    {
      "epoch": 4.825690441366256,
      "grad_norm": 1.109375,
      "learning_rate": 2.9983432581720537e-06,
      "loss": 1.3646,
      "step": 56089
    },
    {
      "epoch": 4.825776477673578,
      "grad_norm": 1.1328125,
      "learning_rate": 2.9953870817615027e-06,
      "loss": 1.0237,
      "step": 56090
    },
    {
      "epoch": 4.8258625139808995,
      "grad_norm": 3.46875,
      "learning_rate": 2.992432358992825e-06,
      "loss": 0.8509,
      "step": 56091
    },
    {
      "epoch": 4.825948550288222,
      "grad_norm": 8.75,
      "learning_rate": 2.9894790898745693e-06,
      "loss": 1.0441,
      "step": 56092
    },
    {
      "epoch": 4.826034586595544,
      "grad_norm": 1.6328125,
      "learning_rate": 2.986527274415507e-06,
      "loss": 1.3168,
      "step": 56093
    },
    {
      "epoch": 4.826120622902865,
      "grad_norm": 0.94921875,
      "learning_rate": 2.983576912624131e-06,
      "loss": 0.541,
      "step": 56094
    },
    {
      "epoch": 4.826206659210187,
      "grad_norm": 1.484375,
      "learning_rate": 2.9806280045091005e-06,
      "loss": 1.4754,
      "step": 56095
    },
    {
      "epoch": 4.826292695517508,
      "grad_norm": 0.87109375,
      "learning_rate": 2.9776805500791314e-06,
      "loss": 0.6898,
      "step": 56096
    },
    {
      "epoch": 4.82637873182483,
      "grad_norm": 1.140625,
      "learning_rate": 2.9747345493427726e-06,
      "loss": 1.3224,
      "step": 56097
    },
    {
      "epoch": 4.826464768132151,
      "grad_norm": 1.1875,
      "learning_rate": 2.971790002308572e-06,
      "loss": 1.3927,
      "step": 56098
    },
    {
      "epoch": 4.826550804439473,
      "grad_norm": 1.109375,
      "learning_rate": 2.968846908985301e-06,
      "loss": 0.7312,
      "step": 56099
    },
    {
      "epoch": 4.826636840746795,
      "grad_norm": 1.2265625,
      "learning_rate": 2.9659052693814526e-06,
      "loss": 0.8622,
      "step": 56100
    },
    {
      "epoch": 4.826722877054117,
      "grad_norm": 0.71875,
      "learning_rate": 2.9629650835056866e-06,
      "loss": 0.5629,
      "step": 56101
    },
    {
      "epoch": 4.826808913361439,
      "grad_norm": 1.03125,
      "learning_rate": 2.9600263513665514e-06,
      "loss": 0.5906,
      "step": 56102
    },
    {
      "epoch": 4.82689494966876,
      "grad_norm": 1.0078125,
      "learning_rate": 2.957089072972652e-06,
      "loss": 1.1119,
      "step": 56103
    },
    {
      "epoch": 4.826980985976082,
      "grad_norm": 1.21875,
      "learning_rate": 2.954153248332647e-06,
      "loss": 0.9657,
      "step": 56104
    },
    {
      "epoch": 4.827067022283403,
      "grad_norm": 0.953125,
      "learning_rate": 2.9512188774550862e-06,
      "loss": 0.9793,
      "step": 56105
    },
    {
      "epoch": 4.827153058590725,
      "grad_norm": 1.09375,
      "learning_rate": 2.9482859603484623e-06,
      "loss": 1.1063,
      "step": 56106
    },
    {
      "epoch": 4.827239094898047,
      "grad_norm": 1.1796875,
      "learning_rate": 2.9453544970214907e-06,
      "loss": 0.8373,
      "step": 56107
    },
    {
      "epoch": 4.827325131205368,
      "grad_norm": 1.265625,
      "learning_rate": 2.9424244874826643e-06,
      "loss": 1.5859,
      "step": 56108
    },
    {
      "epoch": 4.8274111675126905,
      "grad_norm": 1.03125,
      "learning_rate": 2.9394959317405877e-06,
      "loss": 1.0707,
      "step": 56109
    },
    {
      "epoch": 4.827497203820012,
      "grad_norm": 1.34375,
      "learning_rate": 2.9365688298038096e-06,
      "loss": 0.8041,
      "step": 56110
    },
    {
      "epoch": 4.827583240127334,
      "grad_norm": 0.734375,
      "learning_rate": 2.933643181680934e-06,
      "loss": 0.4868,
      "step": 56111
    },
    {
      "epoch": 4.827669276434655,
      "grad_norm": 1.1171875,
      "learning_rate": 2.9307189873803984e-06,
      "loss": 0.5976,
      "step": 56112
    },
    {
      "epoch": 4.827755312741977,
      "grad_norm": 1.0859375,
      "learning_rate": 2.927796246910863e-06,
      "loss": 0.9883,
      "step": 56113
    },
    {
      "epoch": 4.827841349049299,
      "grad_norm": 1.1484375,
      "learning_rate": 2.9248749602808765e-06,
      "loss": 1.131,
      "step": 56114
    },
    {
      "epoch": 4.82792738535662,
      "grad_norm": 4.75,
      "learning_rate": 2.921955127498932e-06,
      "loss": 1.6448,
      "step": 56115
    },
    {
      "epoch": 4.828013421663942,
      "grad_norm": 1.078125,
      "learning_rate": 2.9190367485736334e-06,
      "loss": 1.0648,
      "step": 56116
    },
    {
      "epoch": 4.8280994579712635,
      "grad_norm": 0.86328125,
      "learning_rate": 2.9161198235134188e-06,
      "loss": 0.9497,
      "step": 56117
    },
    {
      "epoch": 4.828185494278586,
      "grad_norm": 0.93359375,
      "learning_rate": 2.9132043523268925e-06,
      "loss": 0.9962,
      "step": 56118
    },
    {
      "epoch": 4.828271530585908,
      "grad_norm": 1.8828125,
      "learning_rate": 2.9102903350225473e-06,
      "loss": 1.139,
      "step": 56119
    },
    {
      "epoch": 4.828357566893229,
      "grad_norm": 1.0234375,
      "learning_rate": 2.9073777716089322e-06,
      "loss": 1.4142,
      "step": 56120
    },
    {
      "epoch": 4.828443603200551,
      "grad_norm": 286.0,
      "learning_rate": 2.9044666620945957e-06,
      "loss": 1.0763,
      "step": 56121
    },
    {
      "epoch": 4.828529639507872,
      "grad_norm": 9.4375,
      "learning_rate": 2.901557006487976e-06,
      "loss": 0.8696,
      "step": 56122
    },
    {
      "epoch": 4.828615675815194,
      "grad_norm": 1.328125,
      "learning_rate": 2.898648804797677e-06,
      "loss": 1.1997,
      "step": 56123
    },
    {
      "epoch": 4.828701712122516,
      "grad_norm": 1.3984375,
      "learning_rate": 2.89574205703208e-06,
      "loss": 1.5447,
      "step": 56124
    },
    {
      "epoch": 4.828787748429837,
      "grad_norm": 0.984375,
      "learning_rate": 2.892836763199791e-06,
      "loss": 0.6568,
      "step": 56125
    },
    {
      "epoch": 4.8288737847371594,
      "grad_norm": 1.1796875,
      "learning_rate": 2.889932923309302e-06,
      "loss": 0.8558,
      "step": 56126
    },
    {
      "epoch": 4.828959821044481,
      "grad_norm": 1.09375,
      "learning_rate": 2.8870305373689953e-06,
      "loss": 1.0449,
      "step": 56127
    },
    {
      "epoch": 4.829045857351803,
      "grad_norm": 1.1328125,
      "learning_rate": 2.8841296053874754e-06,
      "loss": 1.3273,
      "step": 56128
    },
    {
      "epoch": 4.829131893659124,
      "grad_norm": 4.28125,
      "learning_rate": 2.8812301273732355e-06,
      "loss": 1.2156,
      "step": 56129
    },
    {
      "epoch": 4.829217929966446,
      "grad_norm": 1.171875,
      "learning_rate": 2.8783321033346578e-06,
      "loss": 0.9383,
      "step": 56130
    },
    {
      "epoch": 4.829303966273768,
      "grad_norm": 0.8125,
      "learning_rate": 2.8754355332802905e-06,
      "loss": 0.6798,
      "step": 56131
    },
    {
      "epoch": 4.829390002581089,
      "grad_norm": 0.8671875,
      "learning_rate": 2.8725404172186277e-06,
      "loss": 1.0968,
      "step": 56132
    },
    {
      "epoch": 4.829476038888411,
      "grad_norm": 1.359375,
      "learning_rate": 2.8696467551580507e-06,
      "loss": 1.0362,
      "step": 56133
    },
    {
      "epoch": 4.829562075195732,
      "grad_norm": 1.015625,
      "learning_rate": 2.866754547107109e-06,
      "loss": 1.4945,
      "step": 56134
    },
    {
      "epoch": 4.8296481115030545,
      "grad_norm": 1.328125,
      "learning_rate": 2.863863793074184e-06,
      "loss": 1.6843,
      "step": 56135
    },
    {
      "epoch": 4.829734147810376,
      "grad_norm": 8.875,
      "learning_rate": 2.8609744930677693e-06,
      "loss": 0.9751,
      "step": 56136
    },
    {
      "epoch": 4.829820184117698,
      "grad_norm": 1.453125,
      "learning_rate": 2.8580866470963586e-06,
      "loss": 0.8066,
      "step": 56137
    },
    {
      "epoch": 4.82990622042502,
      "grad_norm": 1.03125,
      "learning_rate": 2.855200255168333e-06,
      "loss": 1.5103,
      "step": 56138
    },
    {
      "epoch": 4.829992256732341,
      "grad_norm": 1.3203125,
      "learning_rate": 2.8523153172921313e-06,
      "loss": 1.3473,
      "step": 56139
    },
    {
      "epoch": 4.830078293039663,
      "grad_norm": 1.453125,
      "learning_rate": 2.849431833476246e-06,
      "loss": 1.3717,
      "step": 56140
    },
    {
      "epoch": 4.830164329346984,
      "grad_norm": 1.84375,
      "learning_rate": 2.846549803729059e-06,
      "loss": 1.0569,
      "step": 56141
    },
    {
      "epoch": 4.830250365654306,
      "grad_norm": 1.53125,
      "learning_rate": 2.8436692280590646e-06,
      "loss": 1.9893,
      "step": 56142
    },
    {
      "epoch": 4.8303364019616275,
      "grad_norm": 197.0,
      "learning_rate": 2.8407901064746445e-06,
      "loss": 1.4582,
      "step": 56143
    },
    {
      "epoch": 4.83042243826895,
      "grad_norm": 1.3828125,
      "learning_rate": 2.8379124389841803e-06,
      "loss": 1.4173,
      "step": 56144
    },
    {
      "epoch": 4.830508474576272,
      "grad_norm": 0.8046875,
      "learning_rate": 2.835036225596166e-06,
      "loss": 0.8292,
      "step": 56145
    },
    {
      "epoch": 4.830594510883593,
      "grad_norm": 1.265625,
      "learning_rate": 2.8321614663189278e-06,
      "loss": 1.072,
      "step": 56146
    },
    {
      "epoch": 4.830680547190915,
      "grad_norm": 1.1015625,
      "learning_rate": 2.829288161160959e-06,
      "loss": 1.531,
      "step": 56147
    },
    {
      "epoch": 4.830766583498236,
      "grad_norm": 1.2734375,
      "learning_rate": 2.8264163101306415e-06,
      "loss": 0.7541,
      "step": 56148
    },
    {
      "epoch": 4.830852619805558,
      "grad_norm": 1.1015625,
      "learning_rate": 2.8235459132363583e-06,
      "loss": 0.9187,
      "step": 56149
    },
    {
      "epoch": 4.830938656112879,
      "grad_norm": 1.2734375,
      "learning_rate": 2.8206769704864354e-06,
      "loss": 0.7981,
      "step": 56150
    },
    {
      "epoch": 4.831024692420201,
      "grad_norm": 1.21875,
      "learning_rate": 2.8178094818894218e-06,
      "loss": 1.3915,
      "step": 56151
    },
    {
      "epoch": 4.8311107287275235,
      "grad_norm": 2.453125,
      "learning_rate": 2.8149434474535884e-06,
      "loss": 1.4407,
      "step": 56152
    },
    {
      "epoch": 4.831196765034845,
      "grad_norm": 1.03125,
      "learning_rate": 2.8120788671873733e-06,
      "loss": 0.5605,
      "step": 56153
    },
    {
      "epoch": 4.831282801342167,
      "grad_norm": 1.3046875,
      "learning_rate": 2.809215741099103e-06,
      "loss": 0.8938,
      "step": 56154
    },
    {
      "epoch": 4.831368837649488,
      "grad_norm": 1.03125,
      "learning_rate": 2.806354069197159e-06,
      "loss": 0.8685,
      "step": 56155
    },
    {
      "epoch": 4.83145487395681,
      "grad_norm": 1.203125,
      "learning_rate": 2.80349385148998e-06,
      "loss": 1.2681,
      "step": 56156
    },
    {
      "epoch": 4.831540910264131,
      "grad_norm": 1.390625,
      "learning_rate": 2.8006350879858365e-06,
      "loss": 0.918,
      "step": 56157
    },
    {
      "epoch": 4.831626946571453,
      "grad_norm": 1.0390625,
      "learning_rate": 2.7977777786931666e-06,
      "loss": 1.2087,
      "step": 56158
    },
    {
      "epoch": 4.831712982878775,
      "grad_norm": 1.2890625,
      "learning_rate": 2.794921923620297e-06,
      "loss": 1.5279,
      "step": 56159
    },
    {
      "epoch": 4.8317990191860964,
      "grad_norm": 264.0,
      "learning_rate": 2.7920675227755544e-06,
      "loss": 0.736,
      "step": 56160
    },
    {
      "epoch": 4.8318850554934185,
      "grad_norm": 113.5,
      "learning_rate": 2.7892145761673206e-06,
      "loss": 1.2674,
      "step": 56161
    },
    {
      "epoch": 4.83197109180074,
      "grad_norm": 1.03125,
      "learning_rate": 2.7863630838039776e-06,
      "loss": 0.9815,
      "step": 56162
    },
    {
      "epoch": 4.832057128108062,
      "grad_norm": 0.8046875,
      "learning_rate": 2.783513045693742e-06,
      "loss": 0.632,
      "step": 56163
    },
    {
      "epoch": 4.832143164415383,
      "grad_norm": 1.1171875,
      "learning_rate": 2.780664461845106e-06,
      "loss": 1.0395,
      "step": 56164
    },
    {
      "epoch": 4.832229200722705,
      "grad_norm": 1.6015625,
      "learning_rate": 2.7778173322662857e-06,
      "loss": 0.7143,
      "step": 56165
    },
    {
      "epoch": 4.832315237030027,
      "grad_norm": 0.78125,
      "learning_rate": 2.774971656965608e-06,
      "loss": 0.387,
      "step": 56166
    },
    {
      "epoch": 4.832401273337348,
      "grad_norm": 1.0546875,
      "learning_rate": 2.7721274359515104e-06,
      "loss": 0.7881,
      "step": 56167
    },
    {
      "epoch": 4.83248730964467,
      "grad_norm": 1.0625,
      "learning_rate": 2.7692846692321528e-06,
      "loss": 1.4002,
      "step": 56168
    },
    {
      "epoch": 4.8325733459519915,
      "grad_norm": 1.234375,
      "learning_rate": 2.766443356815973e-06,
      "loss": 1.5064,
      "step": 56169
    },
    {
      "epoch": 4.832659382259314,
      "grad_norm": 1.0234375,
      "learning_rate": 2.7636034987112424e-06,
      "loss": 0.6904,
      "step": 56170
    },
    {
      "epoch": 4.832745418566635,
      "grad_norm": 1.1640625,
      "learning_rate": 2.7607650949262874e-06,
      "loss": 1.1862,
      "step": 56171
    },
    {
      "epoch": 4.832831454873957,
      "grad_norm": 0.9609375,
      "learning_rate": 2.757928145469324e-06,
      "loss": 0.5525,
      "step": 56172
    },
    {
      "epoch": 4.832917491181279,
      "grad_norm": 1.03125,
      "learning_rate": 2.7550926503487894e-06,
      "loss": 1.5388,
      "step": 56173
    },
    {
      "epoch": 4.8330035274886,
      "grad_norm": 1.109375,
      "learning_rate": 2.752258609572844e-06,
      "loss": 1.5429,
      "step": 56174
    },
    {
      "epoch": 4.833089563795922,
      "grad_norm": 1.03125,
      "learning_rate": 2.7494260231498147e-06,
      "loss": 0.6331,
      "step": 56175
    },
    {
      "epoch": 4.833175600103243,
      "grad_norm": 1.140625,
      "learning_rate": 2.7465948910880278e-06,
      "loss": 0.9778,
      "step": 56176
    },
    {
      "epoch": 4.833261636410565,
      "grad_norm": 1.125,
      "learning_rate": 2.7437652133957548e-06,
      "loss": 1.0671,
      "step": 56177
    },
    {
      "epoch": 4.833347672717887,
      "grad_norm": 0.95703125,
      "learning_rate": 2.7409369900812665e-06,
      "loss": 0.8898,
      "step": 56178
    },
    {
      "epoch": 4.833433709025209,
      "grad_norm": 8.125,
      "learning_rate": 2.7381102211527787e-06,
      "loss": 0.6184,
      "step": 56179
    },
    {
      "epoch": 4.833519745332531,
      "grad_norm": 1.1796875,
      "learning_rate": 2.7352849066186737e-06,
      "loss": 0.6714,
      "step": 56180
    },
    {
      "epoch": 4.833605781639852,
      "grad_norm": 1.1640625,
      "learning_rate": 2.7324610464871113e-06,
      "loss": 1.3971,
      "step": 56181
    },
    {
      "epoch": 4.833691817947174,
      "grad_norm": 6.59375,
      "learning_rate": 2.7296386407663633e-06,
      "loss": 0.9443,
      "step": 56182
    },
    {
      "epoch": 4.833777854254495,
      "grad_norm": 1.6796875,
      "learning_rate": 2.7268176894647555e-06,
      "loss": 0.5305,
      "step": 56183
    },
    {
      "epoch": 4.833863890561817,
      "grad_norm": 39.0,
      "learning_rate": 2.7239981925904488e-06,
      "loss": 1.1848,
      "step": 56184
    },
    {
      "epoch": 4.833949926869138,
      "grad_norm": 0.95703125,
      "learning_rate": 2.7211801501517698e-06,
      "loss": 1.1387,
      "step": 56185
    },
    {
      "epoch": 4.8340359631764604,
      "grad_norm": 1.015625,
      "learning_rate": 2.7183635621569334e-06,
      "loss": 1.1692,
      "step": 56186
    },
    {
      "epoch": 4.8341219994837825,
      "grad_norm": 5.53125,
      "learning_rate": 2.7155484286141006e-06,
      "loss": 0.9326,
      "step": 56187
    },
    {
      "epoch": 4.834208035791104,
      "grad_norm": 1.3046875,
      "learning_rate": 2.7127347495316534e-06,
      "loss": 1.6761,
      "step": 56188
    },
    {
      "epoch": 4.834294072098426,
      "grad_norm": 14.9375,
      "learning_rate": 2.7099225249176963e-06,
      "loss": 0.9073,
      "step": 56189
    },
    {
      "epoch": 4.834380108405747,
      "grad_norm": 1.0234375,
      "learning_rate": 2.7071117547805e-06,
      "loss": 0.597,
      "step": 56190
    },
    {
      "epoch": 4.834466144713069,
      "grad_norm": 1.5546875,
      "learning_rate": 2.704302439128281e-06,
      "loss": 1.1511,
      "step": 56191
    },
    {
      "epoch": 4.83455218102039,
      "grad_norm": 1.4921875,
      "learning_rate": 2.7014945779692547e-06,
      "loss": 0.9556,
      "step": 56192
    },
    {
      "epoch": 4.834638217327712,
      "grad_norm": 5.25,
      "learning_rate": 2.698688171311636e-06,
      "loss": 1.1448,
      "step": 56193
    },
    {
      "epoch": 4.834724253635034,
      "grad_norm": 5440.0,
      "learning_rate": 2.6958832191636973e-06,
      "loss": 0.8667,
      "step": 56194
    },
    {
      "epoch": 4.8348102899423555,
      "grad_norm": 0.80859375,
      "learning_rate": 2.693079721533487e-06,
      "loss": 0.7895,
      "step": 56195
    },
    {
      "epoch": 4.834896326249678,
      "grad_norm": 1.7734375,
      "learning_rate": 2.6902776784293314e-06,
      "loss": 1.4892,
      "step": 56196
    },
    {
      "epoch": 4.834982362556999,
      "grad_norm": 1.71875,
      "learning_rate": 2.687477089859447e-06,
      "loss": 1.4472,
      "step": 56197
    },
    {
      "epoch": 4.835068398864321,
      "grad_norm": 233.0,
      "learning_rate": 2.684677955831882e-06,
      "loss": 1.0566,
      "step": 56198
    },
    {
      "epoch": 4.835154435171642,
      "grad_norm": 19.375,
      "learning_rate": 2.6818802763549645e-06,
      "loss": 0.8089,
      "step": 56199
    },
    {
      "epoch": 4.835240471478964,
      "grad_norm": 6.625,
      "learning_rate": 2.6790840514367977e-06,
      "loss": 1.295,
      "step": 56200
    },
    {
      "epoch": 4.835326507786286,
      "grad_norm": 1.2421875,
      "learning_rate": 2.676289281085598e-06,
      "loss": 0.9925,
      "step": 56201
    },
    {
      "epoch": 4.835412544093607,
      "grad_norm": 1.484375,
      "learning_rate": 2.673495965309525e-06,
      "loss": 1.7763,
      "step": 56202
    },
    {
      "epoch": 4.835498580400929,
      "grad_norm": 1.0625,
      "learning_rate": 2.6707041041167946e-06,
      "loss": 1.1157,
      "step": 56203
    },
    {
      "epoch": 4.835584616708251,
      "grad_norm": 1.328125,
      "learning_rate": 2.667913697515456e-06,
      "loss": 0.4822,
      "step": 56204
    },
    {
      "epoch": 4.835670653015573,
      "grad_norm": 352.0,
      "learning_rate": 2.665124745513836e-06,
      "loss": 0.8483,
      "step": 56205
    },
    {
      "epoch": 4.835756689322894,
      "grad_norm": 77.5,
      "learning_rate": 2.6623372481199283e-06,
      "loss": 0.804,
      "step": 56206
    },
    {
      "epoch": 4.835842725630216,
      "grad_norm": 1208.0,
      "learning_rate": 2.659551205341948e-06,
      "loss": 0.9337,
      "step": 56207
    },
    {
      "epoch": 4.835928761937538,
      "grad_norm": 1.046875,
      "learning_rate": 2.6567666171881112e-06,
      "loss": 0.7007,
      "step": 56208
    },
    {
      "epoch": 4.836014798244859,
      "grad_norm": 1.1171875,
      "learning_rate": 2.6539834836664665e-06,
      "loss": 0.6081,
      "step": 56209
    },
    {
      "epoch": 4.836100834552181,
      "grad_norm": 1.2734375,
      "learning_rate": 2.651201804785175e-06,
      "loss": 1.0726,
      "step": 56210
    },
    {
      "epoch": 4.836186870859502,
      "grad_norm": 1.2734375,
      "learning_rate": 2.6484215805523958e-06,
      "loss": 1.4654,
      "step": 56211
    },
    {
      "epoch": 4.8362729071668245,
      "grad_norm": 0.98046875,
      "learning_rate": 2.64564281097629e-06,
      "loss": 0.9464,
      "step": 56212
    },
    {
      "epoch": 4.836358943474146,
      "grad_norm": 0.78125,
      "learning_rate": 2.6428654960649055e-06,
      "loss": 0.607,
      "step": 56213
    },
    {
      "epoch": 4.836444979781468,
      "grad_norm": 1.171875,
      "learning_rate": 2.640089635826459e-06,
      "loss": 1.0367,
      "step": 56214
    },
    {
      "epoch": 4.83653101608879,
      "grad_norm": 1.2265625,
      "learning_rate": 2.637315230268944e-06,
      "loss": 1.1965,
      "step": 56215
    },
    {
      "epoch": 4.836617052396111,
      "grad_norm": 1.4921875,
      "learning_rate": 2.6345422794006314e-06,
      "loss": 0.9638,
      "step": 56216
    },
    {
      "epoch": 4.836703088703433,
      "grad_norm": 1.4453125,
      "learning_rate": 2.6317707832295145e-06,
      "loss": 0.5092,
      "step": 56217
    },
    {
      "epoch": 4.836789125010754,
      "grad_norm": 3.609375,
      "learning_rate": 2.6290007417636987e-06,
      "loss": 1.2569,
      "step": 56218
    },
    {
      "epoch": 4.836875161318076,
      "grad_norm": 0.98828125,
      "learning_rate": 2.6262321550113987e-06,
      "loss": 1.6008,
      "step": 56219
    },
    {
      "epoch": 4.8369611976253974,
      "grad_norm": 0.81640625,
      "learning_rate": 2.623465022980609e-06,
      "loss": 0.8831,
      "step": 56220
    },
    {
      "epoch": 4.8370472339327195,
      "grad_norm": 2.6875,
      "learning_rate": 2.6206993456794337e-06,
      "loss": 1.0888,
      "step": 56221
    },
    {
      "epoch": 4.837133270240042,
      "grad_norm": 1.0234375,
      "learning_rate": 2.6179351231159774e-06,
      "loss": 0.9857,
      "step": 56222
    },
    {
      "epoch": 4.837219306547363,
      "grad_norm": 0.91015625,
      "learning_rate": 2.6151723552982896e-06,
      "loss": 1.4148,
      "step": 56223
    },
    {
      "epoch": 4.837305342854685,
      "grad_norm": 1.1328125,
      "learning_rate": 2.612411042234586e-06,
      "loss": 0.6316,
      "step": 56224
    },
    {
      "epoch": 4.837391379162006,
      "grad_norm": 1.359375,
      "learning_rate": 2.6096511839327487e-06,
      "loss": 1.2468,
      "step": 56225
    },
    {
      "epoch": 4.837477415469328,
      "grad_norm": 5.28125,
      "learning_rate": 2.6068927804009935e-06,
      "loss": 1.5007,
      "step": 56226
    },
    {
      "epoch": 4.837563451776649,
      "grad_norm": 1.1484375,
      "learning_rate": 2.604135831647314e-06,
      "loss": 1.1215,
      "step": 56227
    },
    {
      "epoch": 4.837649488083971,
      "grad_norm": 1.15625,
      "learning_rate": 2.601380337679815e-06,
      "loss": 1.1764,
      "step": 56228
    },
    {
      "epoch": 4.837735524391293,
      "grad_norm": 1.1953125,
      "learning_rate": 2.5986262985065455e-06,
      "loss": 1.4389,
      "step": 56229
    },
    {
      "epoch": 4.837821560698615,
      "grad_norm": 1.6953125,
      "learning_rate": 2.5958737141355546e-06,
      "loss": 0.9702,
      "step": 56230
    },
    {
      "epoch": 4.837907597005937,
      "grad_norm": 1.1328125,
      "learning_rate": 2.593122584574892e-06,
      "loss": 0.986,
      "step": 56231
    },
    {
      "epoch": 4.837993633313258,
      "grad_norm": 1.28125,
      "learning_rate": 2.5903729098326055e-06,
      "loss": 1.1626,
      "step": 56232
    },
    {
      "epoch": 4.83807966962058,
      "grad_norm": 0.98046875,
      "learning_rate": 2.5876246899167456e-06,
      "loss": 1.4334,
      "step": 56233
    },
    {
      "epoch": 4.838165705927902,
      "grad_norm": 0.99609375,
      "learning_rate": 2.584877924835305e-06,
      "loss": 0.9308,
      "step": 56234
    },
    {
      "epoch": 4.838251742235223,
      "grad_norm": 1.046875,
      "learning_rate": 2.5821326145963884e-06,
      "loss": 1.1139,
      "step": 56235
    },
    {
      "epoch": 4.838337778542545,
      "grad_norm": 2.015625,
      "learning_rate": 2.57938875920799e-06,
      "loss": 1.139,
      "step": 56236
    },
    {
      "epoch": 4.838423814849866,
      "grad_norm": 37632.0,
      "learning_rate": 2.576646358678103e-06,
      "loss": 0.6314,
      "step": 56237
    },
    {
      "epoch": 4.8385098511571885,
      "grad_norm": 49.25,
      "learning_rate": 2.573905413014832e-06,
      "loss": 0.9758,
      "step": 56238
    },
    {
      "epoch": 4.83859588746451,
      "grad_norm": 1.2109375,
      "learning_rate": 2.57116592222606e-06,
      "loss": 1.8715,
      "step": 56239
    },
    {
      "epoch": 4.838681923771832,
      "grad_norm": 1.0078125,
      "learning_rate": 2.5684278863199464e-06,
      "loss": 0.8179,
      "step": 56240
    },
    {
      "epoch": 4.838767960079154,
      "grad_norm": 1.421875,
      "learning_rate": 2.565691305304485e-06,
      "loss": 1.501,
      "step": 56241
    },
    {
      "epoch": 4.838853996386475,
      "grad_norm": 0.9609375,
      "learning_rate": 2.5629561791875587e-06,
      "loss": 1.4153,
      "step": 56242
    },
    {
      "epoch": 4.838940032693797,
      "grad_norm": 4.84375,
      "learning_rate": 2.560222507977217e-06,
      "loss": 1.0341,
      "step": 56243
    },
    {
      "epoch": 4.839026069001118,
      "grad_norm": 1.03125,
      "learning_rate": 2.5574902916815635e-06,
      "loss": 1.0373,
      "step": 56244
    },
    {
      "epoch": 4.83911210530844,
      "grad_norm": 2.765625,
      "learning_rate": 2.554759530308426e-06,
      "loss": 0.6051,
      "step": 56245
    },
    {
      "epoch": 4.839198141615762,
      "grad_norm": 1.015625,
      "learning_rate": 2.5520302238659086e-06,
      "loss": 1.2811,
      "step": 56246
    },
    {
      "epoch": 4.8392841779230835,
      "grad_norm": 1.15625,
      "learning_rate": 2.5493023723619503e-06,
      "loss": 0.7782,
      "step": 56247
    },
    {
      "epoch": 4.839370214230406,
      "grad_norm": 1.2109375,
      "learning_rate": 2.5465759758045438e-06,
      "loss": 1.0756,
      "step": 56248
    },
    {
      "epoch": 4.839456250537727,
      "grad_norm": 0.96484375,
      "learning_rate": 2.5438510342016272e-06,
      "loss": 0.9735,
      "step": 56249
    },
    {
      "epoch": 4.839542286845049,
      "grad_norm": 1.6484375,
      "learning_rate": 2.541127547561195e-06,
      "loss": 1.8524,
      "step": 56250
    },
    {
      "epoch": 4.83962832315237,
      "grad_norm": 1.125,
      "learning_rate": 2.53840551589124e-06,
      "loss": 1.2292,
      "step": 56251
    },
    {
      "epoch": 4.839714359459692,
      "grad_norm": 1.0234375,
      "learning_rate": 2.535684939199645e-06,
      "loss": 1.592,
      "step": 56252
    },
    {
      "epoch": 4.839800395767014,
      "grad_norm": 1.0390625,
      "learning_rate": 2.532965817494459e-06,
      "loss": 0.9272,
      "step": 56253
    },
    {
      "epoch": 4.839886432074335,
      "grad_norm": 22.125,
      "learning_rate": 2.5302481507836206e-06,
      "loss": 1.0753,
      "step": 56254
    },
    {
      "epoch": 4.839972468381657,
      "grad_norm": 1.8515625,
      "learning_rate": 2.527531939075012e-06,
      "loss": 1.0045,
      "step": 56255
    },
    {
      "epoch": 4.840058504688979,
      "grad_norm": 0.87109375,
      "learning_rate": 2.5248171823766817e-06,
      "loss": 0.6368,
      "step": 56256
    },
    {
      "epoch": 4.840144540996301,
      "grad_norm": 0.9375,
      "learning_rate": 2.522103880696458e-06,
      "loss": 1.2012,
      "step": 56257
    },
    {
      "epoch": 4.840230577303622,
      "grad_norm": 1.78125,
      "learning_rate": 2.519392034042334e-06,
      "loss": 1.4153,
      "step": 56258
    },
    {
      "epoch": 4.840316613610944,
      "grad_norm": 1.359375,
      "learning_rate": 2.516681642422192e-06,
      "loss": 0.738,
      "step": 56259
    },
    {
      "epoch": 4.840402649918266,
      "grad_norm": 1.0078125,
      "learning_rate": 2.513972705844081e-06,
      "loss": 0.6541,
      "step": 56260
    },
    {
      "epoch": 4.840488686225587,
      "grad_norm": 1.2265625,
      "learning_rate": 2.511265224315773e-06,
      "loss": 0.8845,
      "step": 56261
    },
    {
      "epoch": 4.840574722532909,
      "grad_norm": 1.0390625,
      "learning_rate": 2.508559197845317e-06,
      "loss": 1.0439,
      "step": 56262
    },
    {
      "epoch": 4.84066075884023,
      "grad_norm": 1.0703125,
      "learning_rate": 2.50585462644054e-06,
      "loss": 1.1912,
      "step": 56263
    },
    {
      "epoch": 4.8407467951475525,
      "grad_norm": 1.171875,
      "learning_rate": 2.5031515101093804e-06,
      "loss": 1.1871,
      "step": 56264
    },
    {
      "epoch": 4.840832831454874,
      "grad_norm": 1.03125,
      "learning_rate": 2.500449848859776e-06,
      "loss": 0.6908,
      "step": 56265
    },
    {
      "epoch": 4.840918867762196,
      "grad_norm": 45.0,
      "learning_rate": 2.497749642699554e-06,
      "loss": 0.7844,
      "step": 56266
    },
    {
      "epoch": 4.841004904069518,
      "grad_norm": 1.0625,
      "learning_rate": 2.4950508916366523e-06,
      "loss": 0.7902,
      "step": 56267
    },
    {
      "epoch": 4.841090940376839,
      "grad_norm": 0.96484375,
      "learning_rate": 2.4923535956790087e-06,
      "loss": 1.3142,
      "step": 56268
    },
    {
      "epoch": 4.841176976684161,
      "grad_norm": 1.46875,
      "learning_rate": 2.489657754834451e-06,
      "loss": 1.2374,
      "step": 56269
    },
    {
      "epoch": 4.841263012991482,
      "grad_norm": 1.34375,
      "learning_rate": 2.486963369110862e-06,
      "loss": 1.4734,
      "step": 56270
    },
    {
      "epoch": 4.841349049298804,
      "grad_norm": 1.0546875,
      "learning_rate": 2.4842704385161787e-06,
      "loss": 1.4824,
      "step": 56271
    },
    {
      "epoch": 4.8414350856061255,
      "grad_norm": 1.140625,
      "learning_rate": 2.4815789630582286e-06,
      "loss": 1.1776,
      "step": 56272
    },
    {
      "epoch": 4.8415211219134475,
      "grad_norm": 1.296875,
      "learning_rate": 2.4788889427448393e-06,
      "loss": 1.3286,
      "step": 56273
    },
    {
      "epoch": 4.84160715822077,
      "grad_norm": 0.9453125,
      "learning_rate": 2.476200377584059e-06,
      "loss": 0.7526,
      "step": 56274
    },
    {
      "epoch": 4.841693194528091,
      "grad_norm": 1.6015625,
      "learning_rate": 2.473513267583549e-06,
      "loss": 0.93,
      "step": 56275
    },
    {
      "epoch": 4.841779230835413,
      "grad_norm": 0.83984375,
      "learning_rate": 2.4708276127512474e-06,
      "loss": 1.177,
      "step": 56276
    },
    {
      "epoch": 4.841865267142734,
      "grad_norm": 1.3359375,
      "learning_rate": 2.4681434130950366e-06,
      "loss": 1.8675,
      "step": 56277
    },
    {
      "epoch": 4.841951303450056,
      "grad_norm": 1.21875,
      "learning_rate": 2.4654606686227433e-06,
      "loss": 1.1039,
      "step": 56278
    },
    {
      "epoch": 4.842037339757377,
      "grad_norm": 146.0,
      "learning_rate": 2.4627793793421948e-06,
      "loss": 0.7339,
      "step": 56279
    },
    {
      "epoch": 4.842123376064699,
      "grad_norm": 1.1640625,
      "learning_rate": 2.460099545261274e-06,
      "loss": 0.8796,
      "step": 56280
    },
    {
      "epoch": 4.842209412372021,
      "grad_norm": 0.89453125,
      "learning_rate": 2.457421166387752e-06,
      "loss": 1.2752,
      "step": 56281
    },
    {
      "epoch": 4.842295448679343,
      "grad_norm": 1.125,
      "learning_rate": 2.4547442427295674e-06,
      "loss": 1.1791,
      "step": 56282
    },
    {
      "epoch": 4.842381484986665,
      "grad_norm": 0.92578125,
      "learning_rate": 2.452068774294436e-06,
      "loss": 0.7964,
      "step": 56283
    },
    {
      "epoch": 4.842467521293986,
      "grad_norm": 0.93359375,
      "learning_rate": 2.44939476109024e-06,
      "loss": 0.9598,
      "step": 56284
    },
    {
      "epoch": 4.842553557601308,
      "grad_norm": 0.89453125,
      "learning_rate": 2.446722203124863e-06,
      "loss": 0.8017,
      "step": 56285
    },
    {
      "epoch": 4.842639593908629,
      "grad_norm": 1.25,
      "learning_rate": 2.444051100405964e-06,
      "loss": 1.6615,
      "step": 56286
    },
    {
      "epoch": 4.842725630215951,
      "grad_norm": 20.125,
      "learning_rate": 2.4413814529414823e-06,
      "loss": 1.5112,
      "step": 56287
    },
    {
      "epoch": 4.842811666523273,
      "grad_norm": 1.9921875,
      "learning_rate": 2.4387132607391894e-06,
      "loss": 1.5999,
      "step": 56288
    },
    {
      "epoch": 4.842897702830594,
      "grad_norm": 1.375,
      "learning_rate": 2.436046523806912e-06,
      "loss": 1.5824,
      "step": 56289
    },
    {
      "epoch": 4.8429837391379165,
      "grad_norm": 1.234375,
      "learning_rate": 2.4333812421524217e-06,
      "loss": 1.2571,
      "step": 56290
    },
    {
      "epoch": 4.843069775445238,
      "grad_norm": 1.2109375,
      "learning_rate": 2.43071741578349e-06,
      "loss": 0.79,
      "step": 56291
    },
    {
      "epoch": 4.84315581175256,
      "grad_norm": 1.2109375,
      "learning_rate": 2.4280550447079443e-06,
      "loss": 1.4903,
      "step": 56292
    },
    {
      "epoch": 4.843241848059881,
      "grad_norm": 1.171875,
      "learning_rate": 2.425394128933611e-06,
      "loss": 0.7103,
      "step": 56293
    },
    {
      "epoch": 4.843327884367203,
      "grad_norm": 1.1015625,
      "learning_rate": 2.422734668468152e-06,
      "loss": 1.2774,
      "step": 56294
    },
    {
      "epoch": 4.843413920674525,
      "grad_norm": 0.875,
      "learning_rate": 2.420076663319504e-06,
      "loss": 0.7745,
      "step": 56295
    },
    {
      "epoch": 4.843499956981846,
      "grad_norm": 1.0,
      "learning_rate": 2.4174201134952724e-06,
      "loss": 0.9688,
      "step": 56296
    },
    {
      "epoch": 4.843585993289168,
      "grad_norm": 1.3125,
      "learning_rate": 2.41476501900334e-06,
      "loss": 0.8157,
      "step": 56297
    },
    {
      "epoch": 4.8436720295964895,
      "grad_norm": 1.40625,
      "learning_rate": 2.4121113798514782e-06,
      "loss": 0.7568,
      "step": 56298
    },
    {
      "epoch": 4.8437580659038115,
      "grad_norm": 1.1328125,
      "learning_rate": 2.409459196047403e-06,
      "loss": 1.518,
      "step": 56299
    },
    {
      "epoch": 4.843844102211133,
      "grad_norm": 0.98828125,
      "learning_rate": 2.4068084675988865e-06,
      "loss": 1.1044,
      "step": 56300
    },
    {
      "epoch": 4.843930138518455,
      "grad_norm": 1.390625,
      "learning_rate": 2.4041591945137e-06,
      "loss": 0.9288,
      "step": 56301
    },
    {
      "epoch": 4.844016174825777,
      "grad_norm": 1.203125,
      "learning_rate": 2.4015113767995034e-06,
      "loss": 0.665,
      "step": 56302
    },
    {
      "epoch": 4.844102211133098,
      "grad_norm": 1.109375,
      "learning_rate": 2.3988650144641245e-06,
      "loss": 1.0645,
      "step": 56303
    },
    {
      "epoch": 4.84418824744042,
      "grad_norm": 4.375,
      "learning_rate": 2.3962201075153347e-06,
      "loss": 0.8231,
      "step": 56304
    },
    {
      "epoch": 4.844274283747741,
      "grad_norm": 0.9375,
      "learning_rate": 2.393576655960794e-06,
      "loss": 0.4543,
      "step": 56305
    },
    {
      "epoch": 4.844360320055063,
      "grad_norm": 3.421875,
      "learning_rate": 2.3909346598082747e-06,
      "loss": 0.4533,
      "step": 56306
    },
    {
      "epoch": 4.8444463563623845,
      "grad_norm": 0.74609375,
      "learning_rate": 2.3882941190654926e-06,
      "loss": 0.4858,
      "step": 56307
    },
    {
      "epoch": 4.844532392669707,
      "grad_norm": 1.0234375,
      "learning_rate": 2.3856550337401638e-06,
      "loss": 1.3335,
      "step": 56308
    },
    {
      "epoch": 4.844618428977029,
      "grad_norm": 0.80859375,
      "learning_rate": 2.3830174038400043e-06,
      "loss": 0.6181,
      "step": 56309
    },
    {
      "epoch": 4.84470446528435,
      "grad_norm": 0.9140625,
      "learning_rate": 2.38038122937273e-06,
      "loss": 0.7365,
      "step": 56310
    },
    {
      "epoch": 4.844790501591672,
      "grad_norm": 0.98046875,
      "learning_rate": 2.377746510346057e-06,
      "loss": 0.8759,
      "step": 56311
    },
    {
      "epoch": 4.844876537898993,
      "grad_norm": 1.515625,
      "learning_rate": 2.3751132467677573e-06,
      "loss": 1.7388,
      "step": 56312
    },
    {
      "epoch": 4.844962574206315,
      "grad_norm": 1.0234375,
      "learning_rate": 2.3724814386453798e-06,
      "loss": 1.4106,
      "step": 56313
    },
    {
      "epoch": 4.845048610513636,
      "grad_norm": 1.1328125,
      "learning_rate": 2.3698510859867517e-06,
      "loss": 0.9269,
      "step": 56314
    },
    {
      "epoch": 4.845134646820958,
      "grad_norm": 1.0859375,
      "learning_rate": 2.367222188799534e-06,
      "loss": 1.5466,
      "step": 56315
    },
    {
      "epoch": 4.8452206831282805,
      "grad_norm": 0.96484375,
      "learning_rate": 2.3645947470913863e-06,
      "loss": 0.5487,
      "step": 56316
    },
    {
      "epoch": 4.845306719435602,
      "grad_norm": 3.125,
      "learning_rate": 2.3619687608700813e-06,
      "loss": 1.7275,
      "step": 56317
    },
    {
      "epoch": 4.845392755742924,
      "grad_norm": 1.625,
      "learning_rate": 2.3593442301431678e-06,
      "loss": 1.3253,
      "step": 56318
    },
    {
      "epoch": 4.845478792050245,
      "grad_norm": 1.7109375,
      "learning_rate": 2.356721154918362e-06,
      "loss": 0.9874,
      "step": 56319
    },
    {
      "epoch": 4.845564828357567,
      "grad_norm": 9.375,
      "learning_rate": 2.3540995352034354e-06,
      "loss": 1.7824,
      "step": 56320
    },
    {
      "epoch": 4.845650864664888,
      "grad_norm": 1.15625,
      "learning_rate": 2.3514793710059378e-06,
      "loss": 0.9009,
      "step": 56321
    },
    {
      "epoch": 4.84573690097221,
      "grad_norm": 1.015625,
      "learning_rate": 2.3488606623335297e-06,
      "loss": 1.3818,
      "step": 56322
    },
    {
      "epoch": 4.845822937279532,
      "grad_norm": 0.9453125,
      "learning_rate": 2.3462434091939823e-06,
      "loss": 1.0662,
      "step": 56323
    },
    {
      "epoch": 4.8459089735868535,
      "grad_norm": 22.625,
      "learning_rate": 2.343627611594845e-06,
      "loss": 1.4174,
      "step": 56324
    },
    {
      "epoch": 4.8459950098941755,
      "grad_norm": 1.109375,
      "learning_rate": 2.341013269543835e-06,
      "loss": 1.5799,
      "step": 56325
    },
    {
      "epoch": 4.846081046201497,
      "grad_norm": 0.97265625,
      "learning_rate": 2.3384003830485555e-06,
      "loss": 1.2221,
      "step": 56326
    },
    {
      "epoch": 4.846167082508819,
      "grad_norm": 1.0859375,
      "learning_rate": 2.3357889521166685e-06,
      "loss": 0.6379,
      "step": 56327
    },
    {
      "epoch": 4.84625311881614,
      "grad_norm": 1.25,
      "learning_rate": 2.333178976755834e-06,
      "loss": 0.7502,
      "step": 56328
    },
    {
      "epoch": 4.846339155123462,
      "grad_norm": 8.6875,
      "learning_rate": 2.3305704569736573e-06,
      "loss": 1.3874,
      "step": 56329
    },
    {
      "epoch": 4.846425191430784,
      "grad_norm": 4.4375,
      "learning_rate": 2.3279633927777433e-06,
      "loss": 1.4982,
      "step": 56330
    },
    {
      "epoch": 4.846511227738105,
      "grad_norm": 1.109375,
      "learning_rate": 2.325357784175752e-06,
      "loss": 0.7431,
      "step": 56331
    },
    {
      "epoch": 4.846597264045427,
      "grad_norm": 1.0,
      "learning_rate": 2.322753631175345e-06,
      "loss": 1.2162,
      "step": 56332
    },
    {
      "epoch": 4.8466833003527485,
      "grad_norm": 1.0859375,
      "learning_rate": 2.3201509337840154e-06,
      "loss": 1.5755,
      "step": 56333
    },
    {
      "epoch": 4.846769336660071,
      "grad_norm": 1.046875,
      "learning_rate": 2.3175496920094797e-06,
      "loss": 1.2271,
      "step": 56334
    },
    {
      "epoch": 4.846855372967392,
      "grad_norm": 0.82421875,
      "learning_rate": 2.3149499058593426e-06,
      "loss": 0.7902,
      "step": 56335
    },
    {
      "epoch": 4.846941409274714,
      "grad_norm": 1.46875,
      "learning_rate": 2.312351575341154e-06,
      "loss": 1.1883,
      "step": 56336
    },
    {
      "epoch": 4.847027445582036,
      "grad_norm": 1.9609375,
      "learning_rate": 2.309754700462574e-06,
      "loss": 1.3421,
      "step": 56337
    },
    {
      "epoch": 4.847113481889357,
      "grad_norm": 0.7890625,
      "learning_rate": 2.307159281231097e-06,
      "loss": 0.3988,
      "step": 56338
    },
    {
      "epoch": 4.847199518196679,
      "grad_norm": 1.3203125,
      "learning_rate": 2.304565317654439e-06,
      "loss": 1.0109,
      "step": 56339
    },
    {
      "epoch": 4.847285554504,
      "grad_norm": 20.0,
      "learning_rate": 2.3019728097401493e-06,
      "loss": 1.5485,
      "step": 56340
    },
    {
      "epoch": 4.847371590811322,
      "grad_norm": 0.81640625,
      "learning_rate": 2.299381757495722e-06,
      "loss": 0.5658,
      "step": 56341
    },
    {
      "epoch": 4.847457627118644,
      "grad_norm": 1.078125,
      "learning_rate": 2.2967921609288734e-06,
      "loss": 1.3259,
      "step": 56342
    },
    {
      "epoch": 4.847543663425966,
      "grad_norm": 1.046875,
      "learning_rate": 2.294204020047097e-06,
      "loss": 1.7702,
      "step": 56343
    },
    {
      "epoch": 4.847629699733288,
      "grad_norm": 1.4296875,
      "learning_rate": 2.291617334857943e-06,
      "loss": 1.2697,
      "step": 56344
    },
    {
      "epoch": 4.847715736040609,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2890321053690153e-06,
      "loss": 0.6942,
      "step": 56345
    },
    {
      "epoch": 4.847801772347931,
      "grad_norm": 1.2578125,
      "learning_rate": 2.286448331587865e-06,
      "loss": 0.5493,
      "step": 56346
    },
    {
      "epoch": 4.847887808655252,
      "grad_norm": 1.0703125,
      "learning_rate": 2.2838660135220956e-06,
      "loss": 1.3941,
      "step": 56347
    },
    {
      "epoch": 4.847973844962574,
      "grad_norm": 93.0,
      "learning_rate": 2.281285151179202e-06,
      "loss": 1.0802,
      "step": 56348
    },
    {
      "epoch": 4.848059881269895,
      "grad_norm": 1.390625,
      "learning_rate": 2.2787057445667336e-06,
      "loss": 0.7348,
      "step": 56349
    },
    {
      "epoch": 4.8481459175772175,
      "grad_norm": 1.4453125,
      "learning_rate": 2.27612779369224e-06,
      "loss": 0.4711,
      "step": 56350
    },
    {
      "epoch": 4.8482319538845395,
      "grad_norm": 2.515625,
      "learning_rate": 2.2735512985633255e-06,
      "loss": 0.7553,
      "step": 56351
    },
    {
      "epoch": 4.848317990191861,
      "grad_norm": 1.09375,
      "learning_rate": 2.2709762591873738e-06,
      "loss": 1.0836,
      "step": 56352
    },
    {
      "epoch": 4.848404026499183,
      "grad_norm": 0.98046875,
      "learning_rate": 2.268402675572101e-06,
      "loss": 0.6918,
      "step": 56353
    },
    {
      "epoch": 4.848490062806504,
      "grad_norm": 1.25,
      "learning_rate": 2.26583054772489e-06,
      "loss": 1.5183,
      "step": 56354
    },
    {
      "epoch": 4.848576099113826,
      "grad_norm": 1240.0,
      "learning_rate": 2.263259875653345e-06,
      "loss": 1.4528,
      "step": 56355
    },
    {
      "epoch": 4.848662135421148,
      "grad_norm": 0.96484375,
      "learning_rate": 2.260690659365017e-06,
      "loss": 0.9683,
      "step": 56356
    },
    {
      "epoch": 4.848748171728469,
      "grad_norm": 1.53125,
      "learning_rate": 2.2581228988672875e-06,
      "loss": 1.2944,
      "step": 56357
    },
    {
      "epoch": 4.848834208035791,
      "grad_norm": 1.09375,
      "learning_rate": 2.2555565941677624e-06,
      "loss": 1.001,
      "step": 56358
    },
    {
      "epoch": 4.8489202443431125,
      "grad_norm": 1.953125,
      "learning_rate": 2.252991745273936e-06,
      "loss": 1.4901,
      "step": 56359
    },
    {
      "epoch": 4.849006280650435,
      "grad_norm": 1.2734375,
      "learning_rate": 2.2504283521932455e-06,
      "loss": 1.303,
      "step": 56360
    },
    {
      "epoch": 4.849092316957757,
      "grad_norm": 0.87109375,
      "learning_rate": 2.247866414933353e-06,
      "loss": 0.587,
      "step": 56361
    },
    {
      "epoch": 4.849178353265078,
      "grad_norm": 1.1875,
      "learning_rate": 2.2453059335015847e-06,
      "loss": 0.8755,
      "step": 56362
    },
    {
      "epoch": 4.8492643895724,
      "grad_norm": 1.0625,
      "learning_rate": 2.2427469079054907e-06,
      "loss": 1.7154,
      "step": 56363
    },
    {
      "epoch": 4.849350425879721,
      "grad_norm": 2.671875,
      "learning_rate": 2.2401893381525094e-06,
      "loss": 1.4347,
      "step": 56364
    },
    {
      "epoch": 4.849436462187043,
      "grad_norm": 43.5,
      "learning_rate": 2.2376332242502464e-06,
      "loss": 0.8824,
      "step": 56365
    },
    {
      "epoch": 4.849522498494364,
      "grad_norm": 0.953125,
      "learning_rate": 2.2350785662060835e-06,
      "loss": 1.1549,
      "step": 56366
    },
    {
      "epoch": 4.849608534801686,
      "grad_norm": 3.125,
      "learning_rate": 2.2325253640274603e-06,
      "loss": 1.1275,
      "step": 56367
    },
    {
      "epoch": 4.8496945711090085,
      "grad_norm": 1.28125,
      "learning_rate": 2.22997361772187e-06,
      "loss": 1.8964,
      "step": 56368
    },
    {
      "epoch": 4.84978060741633,
      "grad_norm": 1.03125,
      "learning_rate": 2.2274233272968627e-06,
      "loss": 1.5421,
      "step": 56369
    },
    {
      "epoch": 4.849866643723652,
      "grad_norm": 1.6015625,
      "learning_rate": 2.2248744927598206e-06,
      "loss": 1.115,
      "step": 56370
    },
    {
      "epoch": 4.849952680030973,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2223271141181833e-06,
      "loss": 1.0812,
      "step": 56371
    },
    {
      "epoch": 4.850038716338295,
      "grad_norm": 1.03125,
      "learning_rate": 2.219781191379444e-06,
      "loss": 0.7871,
      "step": 56372
    },
    {
      "epoch": 4.850124752645616,
      "grad_norm": 1.0,
      "learning_rate": 2.2172367245510415e-06,
      "loss": 0.7282,
      "step": 56373
    },
    {
      "epoch": 4.850210788952938,
      "grad_norm": 0.97265625,
      "learning_rate": 2.2146937136403587e-06,
      "loss": 0.9868,
      "step": 56374
    },
    {
      "epoch": 4.85029682526026,
      "grad_norm": 1.4453125,
      "learning_rate": 2.2121521586549454e-06,
      "loss": 2.0132,
      "step": 56375
    },
    {
      "epoch": 4.8503828615675815,
      "grad_norm": 1.078125,
      "learning_rate": 2.2096120596021284e-06,
      "loss": 1.2682,
      "step": 56376
    },
    {
      "epoch": 4.8504688978749035,
      "grad_norm": 1.28125,
      "learning_rate": 2.2070734164894022e-06,
      "loss": 1.4015,
      "step": 56377
    },
    {
      "epoch": 4.850554934182225,
      "grad_norm": 0.90234375,
      "learning_rate": 2.20453622932415e-06,
      "loss": 1.0485,
      "step": 56378
    },
    {
      "epoch": 4.850640970489547,
      "grad_norm": 1.09375,
      "learning_rate": 2.202000498113865e-06,
      "loss": 1.0265,
      "step": 56379
    },
    {
      "epoch": 4.850727006796868,
      "grad_norm": 1.109375,
      "learning_rate": 2.1994662228658756e-06,
      "loss": 0.9983,
      "step": 56380
    },
    {
      "epoch": 4.85081304310419,
      "grad_norm": 1.0234375,
      "learning_rate": 2.19693340358762e-06,
      "loss": 1.2361,
      "step": 56381
    },
    {
      "epoch": 4.850899079411512,
      "grad_norm": 1.3671875,
      "learning_rate": 2.1944020402865363e-06,
      "loss": 0.874,
      "step": 56382
    },
    {
      "epoch": 4.850985115718833,
      "grad_norm": 1.0390625,
      "learning_rate": 2.1918721329700076e-06,
      "loss": 1.3562,
      "step": 56383
    },
    {
      "epoch": 4.851071152026155,
      "grad_norm": 1.4609375,
      "learning_rate": 2.189343681645417e-06,
      "loss": 1.9478,
      "step": 56384
    },
    {
      "epoch": 4.8511571883334765,
      "grad_norm": 1.171875,
      "learning_rate": 2.1868166863201477e-06,
      "loss": 0.8756,
      "step": 56385
    },
    {
      "epoch": 4.851243224640799,
      "grad_norm": 1.0546875,
      "learning_rate": 2.1842911470016934e-06,
      "loss": 1.3756,
      "step": 56386
    },
    {
      "epoch": 4.85132926094812,
      "grad_norm": 1.640625,
      "learning_rate": 2.1817670636973263e-06,
      "loss": 1.1716,
      "step": 56387
    },
    {
      "epoch": 4.851415297255442,
      "grad_norm": 1.109375,
      "learning_rate": 2.1792444364144846e-06,
      "loss": 0.9273,
      "step": 56388
    },
    {
      "epoch": 4.851501333562764,
      "grad_norm": 0.99609375,
      "learning_rate": 2.1767232651605516e-06,
      "loss": 0.6956,
      "step": 56389
    },
    {
      "epoch": 4.851587369870085,
      "grad_norm": 1.109375,
      "learning_rate": 2.1742035499428548e-06,
      "loss": 1.2579,
      "step": 56390
    },
    {
      "epoch": 4.851673406177407,
      "grad_norm": 1.421875,
      "learning_rate": 2.171685290768777e-06,
      "loss": 1.1575,
      "step": 56391
    },
    {
      "epoch": 4.851759442484728,
      "grad_norm": 1.1171875,
      "learning_rate": 2.169168487645701e-06,
      "loss": 1.2987,
      "step": 56392
    },
    {
      "epoch": 4.85184547879205,
      "grad_norm": 2.4375,
      "learning_rate": 2.1666531405810098e-06,
      "loss": 1.5603,
      "step": 56393
    },
    {
      "epoch": 4.851931515099372,
      "grad_norm": 0.93359375,
      "learning_rate": 2.164139249582031e-06,
      "loss": 1.3511,
      "step": 56394
    },
    {
      "epoch": 4.852017551406694,
      "grad_norm": 0.953125,
      "learning_rate": 2.1616268146560925e-06,
      "loss": 1.5914,
      "step": 56395
    },
    {
      "epoch": 4.852103587714016,
      "grad_norm": 2.34375,
      "learning_rate": 2.159115835810577e-06,
      "loss": 0.8972,
      "step": 56396
    },
    {
      "epoch": 4.852189624021337,
      "grad_norm": 1.4375,
      "learning_rate": 2.156606313052867e-06,
      "loss": 1.207,
      "step": 56397
    },
    {
      "epoch": 4.852275660328659,
      "grad_norm": 1.171875,
      "learning_rate": 2.1540982463901794e-06,
      "loss": 0.8155,
      "step": 56398
    },
    {
      "epoch": 4.85236169663598,
      "grad_norm": 3.921875,
      "learning_rate": 2.151591635830008e-06,
      "loss": 1.0127,
      "step": 56399
    },
    {
      "epoch": 4.852447732943302,
      "grad_norm": 1.0625,
      "learning_rate": 2.1490864813795697e-06,
      "loss": 1.008,
      "step": 56400
    },
    {
      "epoch": 4.852533769250623,
      "grad_norm": 1.1796875,
      "learning_rate": 2.146582783046247e-06,
      "loss": 0.9707,
      "step": 56401
    },
    {
      "epoch": 4.8526198055579455,
      "grad_norm": 0.9375,
      "learning_rate": 2.144080540837312e-06,
      "loss": 0.9221,
      "step": 56402
    },
    {
      "epoch": 4.8527058418652675,
      "grad_norm": 1.1171875,
      "learning_rate": 2.1415797547601477e-06,
      "loss": 1.5631,
      "step": 56403
    },
    {
      "epoch": 4.852791878172589,
      "grad_norm": 1.28125,
      "learning_rate": 2.1390804248220265e-06,
      "loss": 1.0052,
      "step": 56404
    },
    {
      "epoch": 4.852877914479911,
      "grad_norm": 1.015625,
      "learning_rate": 2.1365825510302195e-06,
      "loss": 1.5873,
      "step": 56405
    },
    {
      "epoch": 4.852963950787232,
      "grad_norm": 1.4765625,
      "learning_rate": 2.13408613339211e-06,
      "loss": 0.882,
      "step": 56406
    },
    {
      "epoch": 4.853049987094554,
      "grad_norm": 1.359375,
      "learning_rate": 2.131591171915026e-06,
      "loss": 0.8108,
      "step": 56407
    },
    {
      "epoch": 4.853136023401875,
      "grad_norm": 0.8203125,
      "learning_rate": 2.1290976666061278e-06,
      "loss": 1.0524,
      "step": 56408
    },
    {
      "epoch": 4.853222059709197,
      "grad_norm": 1.4765625,
      "learning_rate": 2.1266056174727986e-06,
      "loss": 0.8827,
      "step": 56409
    },
    {
      "epoch": 4.853308096016519,
      "grad_norm": 1.4609375,
      "learning_rate": 2.124115024522366e-06,
      "loss": 0.4915,
      "step": 56410
    },
    {
      "epoch": 4.8533941323238405,
      "grad_norm": 1.140625,
      "learning_rate": 2.1216258877620465e-06,
      "loss": 0.718,
      "step": 56411
    },
    {
      "epoch": 4.853480168631163,
      "grad_norm": 2.265625,
      "learning_rate": 2.1191382071991116e-06,
      "loss": 1.0462,
      "step": 56412
    },
    {
      "epoch": 4.853566204938484,
      "grad_norm": 1.0859375,
      "learning_rate": 2.116651982840889e-06,
      "loss": 1.5379,
      "step": 56413
    },
    {
      "epoch": 4.853652241245806,
      "grad_norm": 0.9609375,
      "learning_rate": 2.1141672146945957e-06,
      "loss": 0.769,
      "step": 56414
    },
    {
      "epoch": 4.853738277553127,
      "grad_norm": 1.0078125,
      "learning_rate": 2.1116839027675583e-06,
      "loss": 1.3201,
      "step": 56415
    },
    {
      "epoch": 4.853824313860449,
      "grad_norm": 1.125,
      "learning_rate": 2.1092020470669935e-06,
      "loss": 0.6378,
      "step": 56416
    },
    {
      "epoch": 4.853910350167771,
      "grad_norm": 1.0078125,
      "learning_rate": 2.1067216476002295e-06,
      "loss": 1.285,
      "step": 56417
    },
    {
      "epoch": 4.853996386475092,
      "grad_norm": 0.99609375,
      "learning_rate": 2.1042427043744263e-06,
      "loss": 0.8475,
      "step": 56418
    },
    {
      "epoch": 4.854082422782414,
      "grad_norm": 1.2578125,
      "learning_rate": 2.101765217396856e-06,
      "loss": 0.7272,
      "step": 56419
    },
    {
      "epoch": 4.854168459089736,
      "grad_norm": 3.171875,
      "learning_rate": 2.0992891866748463e-06,
      "loss": 0.7745,
      "step": 56420
    },
    {
      "epoch": 4.854254495397058,
      "grad_norm": 0.9921875,
      "learning_rate": 2.0968146122155585e-06,
      "loss": 1.3275,
      "step": 56421
    },
    {
      "epoch": 4.854340531704379,
      "grad_norm": 1.3828125,
      "learning_rate": 2.0943414940262084e-06,
      "loss": 1.2782,
      "step": 56422
    },
    {
      "epoch": 4.854426568011701,
      "grad_norm": 1.1640625,
      "learning_rate": 2.0918698321140683e-06,
      "loss": 1.2874,
      "step": 56423
    },
    {
      "epoch": 4.854512604319023,
      "grad_norm": 1.1171875,
      "learning_rate": 2.089399626486466e-06,
      "loss": 0.6547,
      "step": 56424
    },
    {
      "epoch": 4.854598640626344,
      "grad_norm": 0.9375,
      "learning_rate": 2.0869308771504504e-06,
      "loss": 0.5005,
      "step": 56425
    },
    {
      "epoch": 4.854684676933666,
      "grad_norm": 1.5390625,
      "learning_rate": 2.0844635841133497e-06,
      "loss": 0.8064,
      "step": 56426
    },
    {
      "epoch": 4.854770713240987,
      "grad_norm": 1.0,
      "learning_rate": 2.081997747382325e-06,
      "loss": 1.644,
      "step": 56427
    },
    {
      "epoch": 4.8548567495483095,
      "grad_norm": 1.1328125,
      "learning_rate": 2.0795333669646476e-06,
      "loss": 1.2122,
      "step": 56428
    },
    {
      "epoch": 4.854942785855631,
      "grad_norm": 38.5,
      "learning_rate": 2.077070442867479e-06,
      "loss": 0.5856,
      "step": 56429
    },
    {
      "epoch": 4.855028822162953,
      "grad_norm": 1.390625,
      "learning_rate": 2.074608975098091e-06,
      "loss": 0.7367,
      "step": 56430
    },
    {
      "epoch": 4.855114858470275,
      "grad_norm": 1.3515625,
      "learning_rate": 2.0721489636635337e-06,
      "loss": 1.7204,
      "step": 56431
    },
    {
      "epoch": 4.855200894777596,
      "grad_norm": 226.0,
      "learning_rate": 2.0696904085711897e-06,
      "loss": 1.4475,
      "step": 56432
    },
    {
      "epoch": 4.855286931084918,
      "grad_norm": 1.0625,
      "learning_rate": 2.0672333098281093e-06,
      "loss": 1.7144,
      "step": 56433
    },
    {
      "epoch": 4.855372967392239,
      "grad_norm": 1.0234375,
      "learning_rate": 2.064777667441509e-06,
      "loss": 0.6859,
      "step": 56434
    },
    {
      "epoch": 4.855459003699561,
      "grad_norm": 1.09375,
      "learning_rate": 2.0623234814186597e-06,
      "loss": 1.5817,
      "step": 56435
    },
    {
      "epoch": 4.8555450400068825,
      "grad_norm": 1.2109375,
      "learning_rate": 2.0598707517666125e-06,
      "loss": 0.9709,
      "step": 56436
    },
    {
      "epoch": 4.8556310763142045,
      "grad_norm": 1.2109375,
      "learning_rate": 2.057419478492639e-06,
      "loss": 1.6612,
      "step": 56437
    },
    {
      "epoch": 4.855717112621527,
      "grad_norm": 1.1796875,
      "learning_rate": 2.054969661603845e-06,
      "loss": 1.4609,
      "step": 56438
    },
    {
      "epoch": 4.855803148928848,
      "grad_norm": 2.671875,
      "learning_rate": 2.0525213011073906e-06,
      "loss": 1.0246,
      "step": 56439
    },
    {
      "epoch": 4.85588918523617,
      "grad_norm": 1.0,
      "learning_rate": 2.0500743970105484e-06,
      "loss": 0.9876,
      "step": 56440
    },
    {
      "epoch": 4.855975221543491,
      "grad_norm": 1.1640625,
      "learning_rate": 2.0476289493203127e-06,
      "loss": 1.0101,
      "step": 56441
    },
    {
      "epoch": 4.856061257850813,
      "grad_norm": 0.890625,
      "learning_rate": 2.0451849580439553e-06,
      "loss": 0.8208,
      "step": 56442
    },
    {
      "epoch": 4.856147294158134,
      "grad_norm": 1.0703125,
      "learning_rate": 2.0427424231885815e-06,
      "loss": 0.8879,
      "step": 56443
    },
    {
      "epoch": 4.856233330465456,
      "grad_norm": 1.0625,
      "learning_rate": 2.0403013447613526e-06,
      "loss": 0.926,
      "step": 56444
    },
    {
      "epoch": 4.856319366772778,
      "grad_norm": 4.84375,
      "learning_rate": 2.0378617227693743e-06,
      "loss": 1.2094,
      "step": 56445
    },
    {
      "epoch": 4.8564054030801,
      "grad_norm": 0.85546875,
      "learning_rate": 2.0354235572198066e-06,
      "loss": 0.8475,
      "step": 56446
    },
    {
      "epoch": 4.856491439387422,
      "grad_norm": 0.98046875,
      "learning_rate": 2.0329868481197556e-06,
      "loss": 1.2693,
      "step": 56447
    },
    {
      "epoch": 4.856577475694743,
      "grad_norm": 0.8515625,
      "learning_rate": 2.0305515954764373e-06,
      "loss": 0.9296,
      "step": 56448
    },
    {
      "epoch": 4.856663512002065,
      "grad_norm": 1.1015625,
      "learning_rate": 2.028117799296847e-06,
      "loss": 1.1953,
      "step": 56449
    },
    {
      "epoch": 4.856749548309386,
      "grad_norm": 1.421875,
      "learning_rate": 2.0256854595881448e-06,
      "loss": 1.0593,
      "step": 56450
    },
    {
      "epoch": 4.856835584616708,
      "grad_norm": 1.453125,
      "learning_rate": 2.0232545763574917e-06,
      "loss": 0.9482,
      "step": 56451
    },
    {
      "epoch": 4.85692162092403,
      "grad_norm": 1.734375,
      "learning_rate": 2.0208251496119934e-06,
      "loss": 1.0854,
      "step": 56452
    },
    {
      "epoch": 4.857007657231351,
      "grad_norm": 1.5234375,
      "learning_rate": 2.0183971793586444e-06,
      "loss": 1.1749,
      "step": 56453
    },
    {
      "epoch": 4.8570936935386735,
      "grad_norm": 0.94921875,
      "learning_rate": 2.015970665604716e-06,
      "loss": 1.1379,
      "step": 56454
    },
    {
      "epoch": 4.857179729845995,
      "grad_norm": 1.71875,
      "learning_rate": 2.0135456083572037e-06,
      "loss": 1.1555,
      "step": 56455
    },
    {
      "epoch": 4.857265766153317,
      "grad_norm": 1.2109375,
      "learning_rate": 2.0111220076232117e-06,
      "loss": 1.2084,
      "step": 56456
    },
    {
      "epoch": 4.857351802460638,
      "grad_norm": 0.90234375,
      "learning_rate": 2.008699863409846e-06,
      "loss": 0.6291,
      "step": 56457
    },
    {
      "epoch": 4.85743783876796,
      "grad_norm": 13.3125,
      "learning_rate": 2.0062791757241574e-06,
      "loss": 1.0333,
      "step": 56458
    },
    {
      "epoch": 4.857523875075282,
      "grad_norm": 0.99609375,
      "learning_rate": 2.00385994457325e-06,
      "loss": 1.7396,
      "step": 56459
    },
    {
      "epoch": 4.857609911382603,
      "grad_norm": 1.46875,
      "learning_rate": 2.0014421699641738e-06,
      "loss": 1.3817,
      "step": 56460
    },
    {
      "epoch": 4.857695947689925,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9990258519040905e-06,
      "loss": 0.6405,
      "step": 56461
    },
    {
      "epoch": 4.8577819839972465,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9966109903999384e-06,
      "loss": 1.3824,
      "step": 56462
    },
    {
      "epoch": 4.8578680203045685,
      "grad_norm": 1.03125,
      "learning_rate": 1.994197585458879e-06,
      "loss": 0.8678,
      "step": 56463
    },
    {
      "epoch": 4.85795405661189,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9917856370879064e-06,
      "loss": 0.8541,
      "step": 56464
    },
    {
      "epoch": 4.858040092919212,
      "grad_norm": 6.5,
      "learning_rate": 1.9893751452941257e-06,
      "loss": 1.0031,
      "step": 56465
    },
    {
      "epoch": 4.858126129226534,
      "grad_norm": 1.015625,
      "learning_rate": 1.986966110084587e-06,
      "loss": 1.1986,
      "step": 56466
    },
    {
      "epoch": 4.858212165533855,
      "grad_norm": 1.21875,
      "learning_rate": 1.9845585314662294e-06,
      "loss": 1.6875,
      "step": 56467
    },
    {
      "epoch": 4.858298201841177,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9821524094462694e-06,
      "loss": 0.511,
      "step": 56468
    },
    {
      "epoch": 4.858384238148498,
      "grad_norm": 0.9375,
      "learning_rate": 1.9797477440315904e-06,
      "loss": 0.6731,
      "step": 56469
    },
    {
      "epoch": 4.85847027445582,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9773445352293528e-06,
      "loss": 0.9397,
      "step": 56470
    },
    {
      "epoch": 4.858556310763142,
      "grad_norm": 1.25,
      "learning_rate": 1.974942783046496e-06,
      "loss": 0.9059,
      "step": 56471
    },
    {
      "epoch": 4.858642347070464,
      "grad_norm": 1.0625,
      "learning_rate": 1.9725424874900143e-06,
      "loss": 1.0202,
      "step": 56472
    },
    {
      "epoch": 4.858728383377786,
      "grad_norm": 1.671875,
      "learning_rate": 1.9701436485670686e-06,
      "loss": 1.4759,
      "step": 56473
    },
    {
      "epoch": 4.858814419685107,
      "grad_norm": 9.8125,
      "learning_rate": 1.9677462662845426e-06,
      "loss": 1.1314,
      "step": 56474
    },
    {
      "epoch": 4.858900455992429,
      "grad_norm": 1.53125,
      "learning_rate": 1.965350340649541e-06,
      "loss": 1.7535,
      "step": 56475
    },
    {
      "epoch": 4.858986492299751,
      "grad_norm": 2.765625,
      "learning_rate": 1.9629558716690034e-06,
      "loss": 0.8854,
      "step": 56476
    },
    {
      "epoch": 4.859072528607072,
      "grad_norm": 1.046875,
      "learning_rate": 1.9605628593499237e-06,
      "loss": 0.7892,
      "step": 56477
    },
    {
      "epoch": 4.859158564914394,
      "grad_norm": 11.5625,
      "learning_rate": 1.9581713036994077e-06,
      "loss": 1.3139,
      "step": 56478
    },
    {
      "epoch": 4.859244601221715,
      "grad_norm": 0.7734375,
      "learning_rate": 1.955781204724283e-06,
      "loss": 0.632,
      "step": 56479
    },
    {
      "epoch": 4.8593306375290375,
      "grad_norm": 4.625,
      "learning_rate": 1.953392562431711e-06,
      "loss": 0.8817,
      "step": 56480
    },
    {
      "epoch": 4.859416673836359,
      "grad_norm": 5.4375,
      "learning_rate": 1.951005376828574e-06,
      "loss": 0.9801,
      "step": 56481
    },
    {
      "epoch": 4.859502710143681,
      "grad_norm": 1.7890625,
      "learning_rate": 1.9486196479218675e-06,
      "loss": 0.9983,
      "step": 56482
    },
    {
      "epoch": 4.859588746451003,
      "grad_norm": 1.3203125,
      "learning_rate": 1.9462353757185857e-06,
      "loss": 1.1987,
      "step": 56483
    },
    {
      "epoch": 4.859674782758324,
      "grad_norm": 1.0859375,
      "learning_rate": 1.943852560225723e-06,
      "loss": 1.1382,
      "step": 56484
    },
    {
      "epoch": 4.859760819065646,
      "grad_norm": 20.875,
      "learning_rate": 1.941471201450218e-06,
      "loss": 1.2745,
      "step": 56485
    },
    {
      "epoch": 4.859846855372967,
      "grad_norm": 1.140625,
      "learning_rate": 1.9390912993990094e-06,
      "loss": 0.9704,
      "step": 56486
    },
    {
      "epoch": 4.859932891680289,
      "grad_norm": 0.921875,
      "learning_rate": 1.936712854079148e-06,
      "loss": 0.8303,
      "step": 56487
    },
    {
      "epoch": 4.8600189279876105,
      "grad_norm": 1.1875,
      "learning_rate": 1.9343358654975165e-06,
      "loss": 1.5962,
      "step": 56488
    },
    {
      "epoch": 4.8601049642949326,
      "grad_norm": 0.984375,
      "learning_rate": 1.9319603336610536e-06,
      "loss": 1.0219,
      "step": 56489
    },
    {
      "epoch": 4.860191000602255,
      "grad_norm": 1.15625,
      "learning_rate": 1.929586258576754e-06,
      "loss": 1.5375,
      "step": 56490
    },
    {
      "epoch": 4.860277036909576,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9272136402515572e-06,
      "loss": 1.5706,
      "step": 56491
    },
    {
      "epoch": 4.860363073216898,
      "grad_norm": 1.25,
      "learning_rate": 1.924842478692401e-06,
      "loss": 1.365,
      "step": 56492
    },
    {
      "epoch": 4.860449109524219,
      "grad_norm": 1.3203125,
      "learning_rate": 1.92247277390617e-06,
      "loss": 0.7369,
      "step": 56493
    },
    {
      "epoch": 4.860535145831541,
      "grad_norm": 2.78125,
      "learning_rate": 1.920104525899857e-06,
      "loss": 0.5336,
      "step": 56494
    },
    {
      "epoch": 4.860621182138862,
      "grad_norm": 98.0,
      "learning_rate": 1.917737734680347e-06,
      "loss": 1.1588,
      "step": 56495
    },
    {
      "epoch": 4.860707218446184,
      "grad_norm": 0.9609375,
      "learning_rate": 1.915372400254578e-06,
      "loss": 1.3782,
      "step": 56496
    },
    {
      "epoch": 4.860793254753506,
      "grad_norm": 140.0,
      "learning_rate": 1.913008522629489e-06,
      "loss": 0.9591,
      "step": 56497
    },
    {
      "epoch": 4.860879291060828,
      "grad_norm": 1.3203125,
      "learning_rate": 1.910646101812019e-06,
      "loss": 0.8058,
      "step": 56498
    },
    {
      "epoch": 4.86096532736815,
      "grad_norm": 7.6875,
      "learning_rate": 1.908285137808996e-06,
      "loss": 0.6687,
      "step": 56499
    },
    {
      "epoch": 4.861051363675471,
      "grad_norm": 191.0,
      "learning_rate": 1.9059256306273586e-06,
      "loss": 0.9945,
      "step": 56500
    },
    {
      "epoch": 4.861137399982793,
      "grad_norm": 1.3203125,
      "learning_rate": 1.903567580274046e-06,
      "loss": 1.4303,
      "step": 56501
    },
    {
      "epoch": 4.861223436290114,
      "grad_norm": 1.265625,
      "learning_rate": 1.9012109867558858e-06,
      "loss": 0.6785,
      "step": 56502
    },
    {
      "epoch": 4.861309472597436,
      "grad_norm": 1.1328125,
      "learning_rate": 1.8988558500798725e-06,
      "loss": 2.0039,
      "step": 56503
    },
    {
      "epoch": 4.861395508904758,
      "grad_norm": 2.609375,
      "learning_rate": 1.8965021702527786e-06,
      "loss": 0.7775,
      "step": 56504
    },
    {
      "epoch": 4.861481545212079,
      "grad_norm": 1.2265625,
      "learning_rate": 1.8941499472815428e-06,
      "loss": 1.1228,
      "step": 56505
    },
    {
      "epoch": 4.8615675815194015,
      "grad_norm": 1.15625,
      "learning_rate": 1.8917991811730483e-06,
      "loss": 1.1192,
      "step": 56506
    },
    {
      "epoch": 4.861653617826723,
      "grad_norm": 1.4140625,
      "learning_rate": 1.8894498719341236e-06,
      "loss": 1.5666,
      "step": 56507
    },
    {
      "epoch": 4.861739654134045,
      "grad_norm": 1.0546875,
      "learning_rate": 1.887102019571707e-06,
      "loss": 1.1396,
      "step": 56508
    },
    {
      "epoch": 4.861825690441366,
      "grad_norm": 1.1328125,
      "learning_rate": 1.8847556240926822e-06,
      "loss": 0.9256,
      "step": 56509
    },
    {
      "epoch": 4.861911726748688,
      "grad_norm": 0.9765625,
      "learning_rate": 1.8824106855038214e-06,
      "loss": 1.0666,
      "step": 56510
    },
    {
      "epoch": 4.86199776305601,
      "grad_norm": 0.9921875,
      "learning_rate": 1.8800672038120636e-06,
      "loss": 1.037,
      "step": 56511
    },
    {
      "epoch": 4.862083799363331,
      "grad_norm": 16.75,
      "learning_rate": 1.877725179024181e-06,
      "loss": 0.8933,
      "step": 56512
    },
    {
      "epoch": 4.862169835670653,
      "grad_norm": 0.99609375,
      "learning_rate": 1.875384611147113e-06,
      "loss": 0.6154,
      "step": 56513
    },
    {
      "epoch": 4.8622558719779745,
      "grad_norm": 1.1171875,
      "learning_rate": 1.8730455001876866e-06,
      "loss": 1.4554,
      "step": 56514
    },
    {
      "epoch": 4.8623419082852966,
      "grad_norm": 1.15625,
      "learning_rate": 1.870707846152675e-06,
      "loss": 0.7368,
      "step": 56515
    },
    {
      "epoch": 4.862427944592618,
      "grad_norm": 1.546875,
      "learning_rate": 1.8683716490490167e-06,
      "loss": 1.4015,
      "step": 56516
    },
    {
      "epoch": 4.86251398089994,
      "grad_norm": 2.5,
      "learning_rate": 1.8660369088834283e-06,
      "loss": 1.3427,
      "step": 56517
    },
    {
      "epoch": 4.862600017207262,
      "grad_norm": 1.1328125,
      "learning_rate": 1.8637036256628492e-06,
      "loss": 0.7953,
      "step": 56518
    },
    {
      "epoch": 4.862686053514583,
      "grad_norm": 1.1640625,
      "learning_rate": 1.8613717993940515e-06,
      "loss": 0.8559,
      "step": 56519
    },
    {
      "epoch": 4.862772089821905,
      "grad_norm": 0.9609375,
      "learning_rate": 1.8590414300838076e-06,
      "loss": 0.67,
      "step": 56520
    },
    {
      "epoch": 4.862858126129226,
      "grad_norm": 1.2421875,
      "learning_rate": 1.856712517739001e-06,
      "loss": 1.3648,
      "step": 56521
    },
    {
      "epoch": 4.862944162436548,
      "grad_norm": 1.640625,
      "learning_rate": 1.8543850623664593e-06,
      "loss": 1.1451,
      "step": 56522
    },
    {
      "epoch": 4.8630301987438695,
      "grad_norm": 2.140625,
      "learning_rate": 1.852059063972955e-06,
      "loss": 1.2135,
      "step": 56523
    },
    {
      "epoch": 4.863116235051192,
      "grad_norm": 1.5625,
      "learning_rate": 1.8497345225652606e-06,
      "loss": 1.4779,
      "step": 56524
    },
    {
      "epoch": 4.863202271358514,
      "grad_norm": 1.3984375,
      "learning_rate": 1.8474114381502593e-06,
      "loss": 0.9462,
      "step": 56525
    },
    {
      "epoch": 4.863288307665835,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8450898107346126e-06,
      "loss": 0.7817,
      "step": 56526
    },
    {
      "epoch": 4.863374343973157,
      "grad_norm": 1.03125,
      "learning_rate": 1.8427696403252591e-06,
      "loss": 0.6,
      "step": 56527
    },
    {
      "epoch": 4.863460380280478,
      "grad_norm": 0.98828125,
      "learning_rate": 1.8404509269289161e-06,
      "loss": 0.9671,
      "step": 56528
    },
    {
      "epoch": 4.8635464165878,
      "grad_norm": 0.9140625,
      "learning_rate": 1.8381336705523555e-06,
      "loss": 1.6746,
      "step": 56529
    },
    {
      "epoch": 4.863632452895121,
      "grad_norm": 245.0,
      "learning_rate": 1.8358178712023498e-06,
      "loss": 0.8684,
      "step": 56530
    },
    {
      "epoch": 4.863718489202443,
      "grad_norm": 1.0390625,
      "learning_rate": 1.833503528885727e-06,
      "loss": 0.6687,
      "step": 56531
    },
    {
      "epoch": 4.8638045255097655,
      "grad_norm": 5.75,
      "learning_rate": 1.8311906436091486e-06,
      "loss": 0.8512,
      "step": 56532
    },
    {
      "epoch": 4.863890561817087,
      "grad_norm": 1.1015625,
      "learning_rate": 1.8288792153794975e-06,
      "loss": 1.6961,
      "step": 56533
    },
    {
      "epoch": 4.863976598124409,
      "grad_norm": 1.1875,
      "learning_rate": 1.8265692442034908e-06,
      "loss": 0.7723,
      "step": 56534
    },
    {
      "epoch": 4.86406263443173,
      "grad_norm": 0.984375,
      "learning_rate": 1.8242607300878455e-06,
      "loss": 1.0179,
      "step": 56535
    },
    {
      "epoch": 4.864148670739052,
      "grad_norm": 15.4375,
      "learning_rate": 1.8219536730393892e-06,
      "loss": 1.4024,
      "step": 56536
    },
    {
      "epoch": 4.864234707046373,
      "grad_norm": 1.0703125,
      "learning_rate": 1.819648073064839e-06,
      "loss": 0.6626,
      "step": 56537
    },
    {
      "epoch": 4.864320743353695,
      "grad_norm": 0.9765625,
      "learning_rate": 1.8173439301708561e-06,
      "loss": 1.0688,
      "step": 56538
    },
    {
      "epoch": 4.864406779661017,
      "grad_norm": 1.296875,
      "learning_rate": 1.8150412443643238e-06,
      "loss": 0.957,
      "step": 56539
    },
    {
      "epoch": 4.8644928159683385,
      "grad_norm": 1.2578125,
      "learning_rate": 1.8127400156519037e-06,
      "loss": 1.0888,
      "step": 56540
    },
    {
      "epoch": 4.864578852275661,
      "grad_norm": 1.4453125,
      "learning_rate": 1.8104402440403124e-06,
      "loss": 0.8053,
      "step": 56541
    },
    {
      "epoch": 4.864664888582982,
      "grad_norm": 1.3984375,
      "learning_rate": 1.8081419295362666e-06,
      "loss": 0.9923,
      "step": 56542
    },
    {
      "epoch": 4.864750924890304,
      "grad_norm": 1.453125,
      "learning_rate": 1.8058450721464837e-06,
      "loss": 1.7718,
      "step": 56543
    },
    {
      "epoch": 4.864836961197625,
      "grad_norm": 6.6875,
      "learning_rate": 1.803549671877791e-06,
      "loss": 1.0057,
      "step": 56544
    },
    {
      "epoch": 4.864922997504947,
      "grad_norm": 1.15625,
      "learning_rate": 1.8012557287367392e-06,
      "loss": 0.7005,
      "step": 56545
    },
    {
      "epoch": 4.865009033812269,
      "grad_norm": 1.0859375,
      "learning_rate": 1.798963242730156e-06,
      "loss": 0.6715,
      "step": 56546
    },
    {
      "epoch": 4.86509507011959,
      "grad_norm": 0.921875,
      "learning_rate": 1.7966722138647584e-06,
      "loss": 0.6049,
      "step": 56547
    },
    {
      "epoch": 4.865181106426912,
      "grad_norm": 1.375,
      "learning_rate": 1.7943826421471521e-06,
      "loss": 1.0712,
      "step": 56548
    },
    {
      "epoch": 4.8652671427342336,
      "grad_norm": 1.53125,
      "learning_rate": 1.792094527584054e-06,
      "loss": 0.9534,
      "step": 56549
    },
    {
      "epoch": 4.865353179041556,
      "grad_norm": 2.546875,
      "learning_rate": 1.7898078701822361e-06,
      "loss": 0.7374,
      "step": 56550
    },
    {
      "epoch": 4.865439215348877,
      "grad_norm": 1.1796875,
      "learning_rate": 1.7875226699483048e-06,
      "loss": 1.3848,
      "step": 56551
    },
    {
      "epoch": 4.865525251656199,
      "grad_norm": 1.40625,
      "learning_rate": 1.7852389268889768e-06,
      "loss": 1.3767,
      "step": 56552
    },
    {
      "epoch": 4.865611287963521,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7829566410109132e-06,
      "loss": 0.6599,
      "step": 56553
    },
    {
      "epoch": 4.865697324270842,
      "grad_norm": 1.734375,
      "learning_rate": 1.7806758123208312e-06,
      "loss": 1.6851,
      "step": 56554
    },
    {
      "epoch": 4.865783360578164,
      "grad_norm": 1.4453125,
      "learning_rate": 1.7783964408253917e-06,
      "loss": 0.8939,
      "step": 56555
    },
    {
      "epoch": 4.865869396885485,
      "grad_norm": 1.171875,
      "learning_rate": 1.7761185265312007e-06,
      "loss": 1.4255,
      "step": 56556
    },
    {
      "epoch": 4.865955433192807,
      "grad_norm": 1.2890625,
      "learning_rate": 1.7738420694450308e-06,
      "loss": 1.0012,
      "step": 56557
    },
    {
      "epoch": 4.866041469500129,
      "grad_norm": 1.1796875,
      "learning_rate": 1.7715670695733765e-06,
      "loss": 0.6689,
      "step": 56558
    },
    {
      "epoch": 4.866127505807451,
      "grad_norm": 1.4609375,
      "learning_rate": 1.769293526923066e-06,
      "loss": 1.4175,
      "step": 56559
    },
    {
      "epoch": 4.866213542114773,
      "grad_norm": 1.25,
      "learning_rate": 1.7670214415006491e-06,
      "loss": 1.1102,
      "step": 56560
    },
    {
      "epoch": 4.866299578422094,
      "grad_norm": 1.0703125,
      "learning_rate": 1.7647508133127322e-06,
      "loss": 0.9128,
      "step": 56561
    },
    {
      "epoch": 4.866385614729416,
      "grad_norm": 1.2265625,
      "learning_rate": 1.7624816423660872e-06,
      "loss": 1.529,
      "step": 56562
    },
    {
      "epoch": 4.866471651036737,
      "grad_norm": 1.09375,
      "learning_rate": 1.7602139286672646e-06,
      "loss": 1.163,
      "step": 56563
    },
    {
      "epoch": 4.866557687344059,
      "grad_norm": 3.65625,
      "learning_rate": 1.757947672222926e-06,
      "loss": 0.9289,
      "step": 56564
    },
    {
      "epoch": 4.86664372365138,
      "grad_norm": 0.93359375,
      "learning_rate": 1.7556828730396767e-06,
      "loss": 0.5885,
      "step": 56565
    },
    {
      "epoch": 4.8667297599587025,
      "grad_norm": 1.0,
      "learning_rate": 1.7534195311241785e-06,
      "loss": 1.0188,
      "step": 56566
    },
    {
      "epoch": 4.866815796266025,
      "grad_norm": 1.1171875,
      "learning_rate": 1.7511576464829815e-06,
      "loss": 0.8475,
      "step": 56567
    },
    {
      "epoch": 4.866901832573346,
      "grad_norm": 1.015625,
      "learning_rate": 1.748897219122747e-06,
      "loss": 1.3084,
      "step": 56568
    },
    {
      "epoch": 4.866987868880668,
      "grad_norm": 1.21875,
      "learning_rate": 1.746638249050081e-06,
      "loss": 1.6321,
      "step": 56569
    },
    {
      "epoch": 4.867073905187989,
      "grad_norm": 5.03125,
      "learning_rate": 1.7443807362716447e-06,
      "loss": 1.4362,
      "step": 56570
    },
    {
      "epoch": 4.867159941495311,
      "grad_norm": 1.09375,
      "learning_rate": 1.7421246807939328e-06,
      "loss": 0.8238,
      "step": 56571
    },
    {
      "epoch": 4.867245977802632,
      "grad_norm": 1.3828125,
      "learning_rate": 1.739870082623607e-06,
      "loss": 1.2115,
      "step": 56572
    },
    {
      "epoch": 4.867332014109954,
      "grad_norm": 1.09375,
      "learning_rate": 1.7376169417672728e-06,
      "loss": 1.3566,
      "step": 56573
    },
    {
      "epoch": 4.867418050417276,
      "grad_norm": 1.1796875,
      "learning_rate": 1.735365258231536e-06,
      "loss": 0.9466,
      "step": 56574
    },
    {
      "epoch": 4.867504086724598,
      "grad_norm": 1.5859375,
      "learning_rate": 1.7331150320228916e-06,
      "loss": 0.8861,
      "step": 56575
    },
    {
      "epoch": 4.86759012303192,
      "grad_norm": 0.8515625,
      "learning_rate": 1.730866263148001e-06,
      "loss": 0.9944,
      "step": 56576
    },
    {
      "epoch": 4.867676159339241,
      "grad_norm": 1.3046875,
      "learning_rate": 1.7286189516134144e-06,
      "loss": 0.802,
      "step": 56577
    },
    {
      "epoch": 4.867762195646563,
      "grad_norm": 1.0390625,
      "learning_rate": 1.7263730974257374e-06,
      "loss": 1.3476,
      "step": 56578
    },
    {
      "epoch": 4.867848231953884,
      "grad_norm": 1.2265625,
      "learning_rate": 1.7241287005914652e-06,
      "loss": 1.0728,
      "step": 56579
    },
    {
      "epoch": 4.867934268261206,
      "grad_norm": 1.2578125,
      "learning_rate": 1.7218857611172034e-06,
      "loss": 0.7647,
      "step": 56580
    },
    {
      "epoch": 4.868020304568528,
      "grad_norm": 1.265625,
      "learning_rate": 1.719644279009558e-06,
      "loss": 0.9063,
      "step": 56581
    },
    {
      "epoch": 4.868106340875849,
      "grad_norm": 6.09375,
      "learning_rate": 1.7174042542750234e-06,
      "loss": 1.335,
      "step": 56582
    },
    {
      "epoch": 4.868192377183171,
      "grad_norm": 2.546875,
      "learning_rate": 1.7151656869201505e-06,
      "loss": 0.817,
      "step": 56583
    },
    {
      "epoch": 4.868278413490493,
      "grad_norm": 1.3046875,
      "learning_rate": 1.7129285769515446e-06,
      "loss": 1.2156,
      "step": 56584
    },
    {
      "epoch": 4.868364449797815,
      "grad_norm": 0.84375,
      "learning_rate": 1.7106929243757008e-06,
      "loss": 0.8202,
      "step": 56585
    },
    {
      "epoch": 4.868450486105137,
      "grad_norm": 1.515625,
      "learning_rate": 1.7084587291991139e-06,
      "loss": 1.9956,
      "step": 56586
    },
    {
      "epoch": 4.868536522412458,
      "grad_norm": 1.2578125,
      "learning_rate": 1.706225991428445e-06,
      "loss": 1.2316,
      "step": 56587
    },
    {
      "epoch": 4.86862255871978,
      "grad_norm": 0.9453125,
      "learning_rate": 1.7039947110701338e-06,
      "loss": 0.9654,
      "step": 56588
    },
    {
      "epoch": 4.868708595027101,
      "grad_norm": 1.109375,
      "learning_rate": 1.70176488813073e-06,
      "loss": 1.1487,
      "step": 56589
    },
    {
      "epoch": 4.868794631334423,
      "grad_norm": 1.40625,
      "learning_rate": 1.6995365226167293e-06,
      "loss": 0.7458,
      "step": 56590
    },
    {
      "epoch": 4.868880667641744,
      "grad_norm": 1.4765625,
      "learning_rate": 1.697309614534681e-06,
      "loss": 1.37,
      "step": 56591
    },
    {
      "epoch": 4.8689667039490665,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6950841638910807e-06,
      "loss": 0.3513,
      "step": 56592
    },
    {
      "epoch": 4.869052740256389,
      "grad_norm": 0.91796875,
      "learning_rate": 1.6928601706924785e-06,
      "loss": 0.6318,
      "step": 56593
    },
    {
      "epoch": 4.86913877656371,
      "grad_norm": 1.4375,
      "learning_rate": 1.6906376349453134e-06,
      "loss": 0.9449,
      "step": 56594
    },
    {
      "epoch": 4.869224812871032,
      "grad_norm": 1.2578125,
      "learning_rate": 1.6884165566560805e-06,
      "loss": 1.6773,
      "step": 56595
    },
    {
      "epoch": 4.869310849178353,
      "grad_norm": 2.3125,
      "learning_rate": 1.6861969358313855e-06,
      "loss": 1.6583,
      "step": 56596
    },
    {
      "epoch": 4.869396885485675,
      "grad_norm": 1.3125,
      "learning_rate": 1.6839787724776124e-06,
      "loss": 1.7694,
      "step": 56597
    },
    {
      "epoch": 4.869482921792997,
      "grad_norm": 1.140625,
      "learning_rate": 1.6817620666012557e-06,
      "loss": 1.1927,
      "step": 56598
    },
    {
      "epoch": 4.869568958100318,
      "grad_norm": 1.203125,
      "learning_rate": 1.6795468182088658e-06,
      "loss": 1.354,
      "step": 56599
    },
    {
      "epoch": 4.86965499440764,
      "grad_norm": 44.5,
      "learning_rate": 1.677333027306882e-06,
      "loss": 1.5323,
      "step": 56600
    },
    {
      "epoch": 4.869741030714962,
      "grad_norm": 25.25,
      "learning_rate": 1.6751206939017994e-06,
      "loss": 1.5425,
      "step": 56601
    },
    {
      "epoch": 4.869827067022284,
      "grad_norm": 0.828125,
      "learning_rate": 1.6729098180000012e-06,
      "loss": 0.5829,
      "step": 56602
    },
    {
      "epoch": 4.869913103329605,
      "grad_norm": 3.03125,
      "learning_rate": 1.6707003996080939e-06,
      "loss": 1.1743,
      "step": 56603
    },
    {
      "epoch": 4.869999139636927,
      "grad_norm": 1.4375,
      "learning_rate": 1.6684924387324606e-06,
      "loss": 1.1608,
      "step": 56604
    },
    {
      "epoch": 4.870085175944249,
      "grad_norm": 1.53125,
      "learning_rate": 1.6662859353795412e-06,
      "loss": 1.4121,
      "step": 56605
    },
    {
      "epoch": 4.87017121225157,
      "grad_norm": 1.03125,
      "learning_rate": 1.6640808895557746e-06,
      "loss": 1.2432,
      "step": 56606
    },
    {
      "epoch": 4.870257248558892,
      "grad_norm": 0.9375,
      "learning_rate": 1.6618773012677114e-06,
      "loss": 0.6663,
      "step": 56607
    },
    {
      "epoch": 4.870343284866213,
      "grad_norm": 94.5,
      "learning_rate": 1.659675170521735e-06,
      "loss": 0.7723,
      "step": 56608
    },
    {
      "epoch": 4.870429321173535,
      "grad_norm": 1.1015625,
      "learning_rate": 1.657474497324285e-06,
      "loss": 1.212,
      "step": 56609
    },
    {
      "epoch": 4.870515357480857,
      "grad_norm": 1.359375,
      "learning_rate": 1.6552752816818006e-06,
      "loss": 1.0825,
      "step": 56610
    },
    {
      "epoch": 4.870601393788179,
      "grad_norm": 1.78125,
      "learning_rate": 1.6530775236006656e-06,
      "loss": 0.7469,
      "step": 56611
    },
    {
      "epoch": 4.870687430095501,
      "grad_norm": 1.390625,
      "learning_rate": 1.6508812230874304e-06,
      "loss": 0.6997,
      "step": 56612
    },
    {
      "epoch": 4.870773466402822,
      "grad_norm": 1.0625,
      "learning_rate": 1.6486863801483675e-06,
      "loss": 0.784,
      "step": 56613
    },
    {
      "epoch": 4.870859502710144,
      "grad_norm": 1.1484375,
      "learning_rate": 1.6464929947900277e-06,
      "loss": 1.2761,
      "step": 56614
    },
    {
      "epoch": 4.870945539017465,
      "grad_norm": 1.1328125,
      "learning_rate": 1.6443010670187385e-06,
      "loss": 1.3673,
      "step": 56615
    },
    {
      "epoch": 4.871031575324787,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6421105968409955e-06,
      "loss": 0.4213,
      "step": 56616
    },
    {
      "epoch": 4.871117611632108,
      "grad_norm": 1.015625,
      "learning_rate": 1.639921584263071e-06,
      "loss": 1.9295,
      "step": 56617
    },
    {
      "epoch": 4.8712036479394305,
      "grad_norm": 1.234375,
      "learning_rate": 1.6377340292915156e-06,
      "loss": 1.7094,
      "step": 56618
    },
    {
      "epoch": 4.871289684246753,
      "grad_norm": 1.21875,
      "learning_rate": 1.6355479319326017e-06,
      "loss": 1.6833,
      "step": 56619
    },
    {
      "epoch": 4.871375720554074,
      "grad_norm": 1.140625,
      "learning_rate": 1.63336329219288e-06,
      "loss": 1.368,
      "step": 56620
    },
    {
      "epoch": 4.871461756861396,
      "grad_norm": 1.40625,
      "learning_rate": 1.6311801100785672e-06,
      "loss": 1.3919,
      "step": 56621
    },
    {
      "epoch": 4.871547793168717,
      "grad_norm": 0.9765625,
      "learning_rate": 1.6289983855961587e-06,
      "loss": 1.2858,
      "step": 56622
    },
    {
      "epoch": 4.871633829476039,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6268181187519826e-06,
      "loss": 0.6512,
      "step": 56623
    },
    {
      "epoch": 4.87171986578336,
      "grad_norm": 0.96484375,
      "learning_rate": 1.6246393095524227e-06,
      "loss": 0.7863,
      "step": 56624
    },
    {
      "epoch": 4.871805902090682,
      "grad_norm": 1.4921875,
      "learning_rate": 1.6224619580038625e-06,
      "loss": 1.862,
      "step": 56625
    },
    {
      "epoch": 4.871891938398004,
      "grad_norm": 29.875,
      "learning_rate": 1.6202860641127414e-06,
      "loss": 0.6124,
      "step": 56626
    },
    {
      "epoch": 4.871977974705326,
      "grad_norm": 1.0390625,
      "learning_rate": 1.618111627885277e-06,
      "loss": 1.4318,
      "step": 56627
    },
    {
      "epoch": 4.872064011012648,
      "grad_norm": 1.03125,
      "learning_rate": 1.6159386493279637e-06,
      "loss": 0.731,
      "step": 56628
    },
    {
      "epoch": 4.872150047319969,
      "grad_norm": 1.1484375,
      "learning_rate": 1.6137671284470745e-06,
      "loss": 1.294,
      "step": 56629
    },
    {
      "epoch": 4.872236083627291,
      "grad_norm": 1.0546875,
      "learning_rate": 1.6115970652489375e-06,
      "loss": 0.7106,
      "step": 56630
    },
    {
      "epoch": 4.872322119934612,
      "grad_norm": 1.390625,
      "learning_rate": 1.6094284597400476e-06,
      "loss": 1.0845,
      "step": 56631
    },
    {
      "epoch": 4.872408156241934,
      "grad_norm": 1.0859375,
      "learning_rate": 1.6072613119265666e-06,
      "loss": 0.7233,
      "step": 56632
    },
    {
      "epoch": 4.872494192549256,
      "grad_norm": 1.0390625,
      "learning_rate": 1.6050956218149338e-06,
      "loss": 1.1568,
      "step": 56633
    },
    {
      "epoch": 4.872580228856577,
      "grad_norm": 1.5078125,
      "learning_rate": 1.6029313894114771e-06,
      "loss": 1.3495,
      "step": 56634
    },
    {
      "epoch": 4.8726662651638994,
      "grad_norm": 0.98828125,
      "learning_rate": 1.6007686147225253e-06,
      "loss": 0.8612,
      "step": 56635
    },
    {
      "epoch": 4.872752301471221,
      "grad_norm": 0.859375,
      "learning_rate": 1.598607297754351e-06,
      "loss": 0.9208,
      "step": 56636
    },
    {
      "epoch": 4.872838337778543,
      "grad_norm": 1.40625,
      "learning_rate": 1.5964474385133375e-06,
      "loss": 1.0568,
      "step": 56637
    },
    {
      "epoch": 4.872924374085864,
      "grad_norm": 1.09375,
      "learning_rate": 1.5942890370058139e-06,
      "loss": 1.3147,
      "step": 56638
    },
    {
      "epoch": 4.873010410393186,
      "grad_norm": 584.0,
      "learning_rate": 1.5921320932379968e-06,
      "loss": 1.33,
      "step": 56639
    },
    {
      "epoch": 4.873096446700508,
      "grad_norm": 1.3515625,
      "learning_rate": 1.5899766072163258e-06,
      "loss": 1.0388,
      "step": 56640
    },
    {
      "epoch": 4.873182483007829,
      "grad_norm": 1.1484375,
      "learning_rate": 1.5878225789469625e-06,
      "loss": 1.5542,
      "step": 56641
    },
    {
      "epoch": 4.873268519315151,
      "grad_norm": 1.3828125,
      "learning_rate": 1.5856700084363462e-06,
      "loss": 1.3989,
      "step": 56642
    },
    {
      "epoch": 4.873354555622472,
      "grad_norm": 1.359375,
      "learning_rate": 1.5835188956906944e-06,
      "loss": 1.6217,
      "step": 56643
    },
    {
      "epoch": 4.8734405919297945,
      "grad_norm": 0.890625,
      "learning_rate": 1.5813692407162795e-06,
      "loss": 0.6211,
      "step": 56644
    },
    {
      "epoch": 4.873526628237116,
      "grad_norm": 1.2421875,
      "learning_rate": 1.5792210435194854e-06,
      "loss": 1.8589,
      "step": 56645
    },
    {
      "epoch": 4.873612664544438,
      "grad_norm": 1.2109375,
      "learning_rate": 1.5770743041064738e-06,
      "loss": 1.519,
      "step": 56646
    },
    {
      "epoch": 4.87369870085176,
      "grad_norm": 75.0,
      "learning_rate": 1.5749290224835732e-06,
      "loss": 1.115,
      "step": 56647
    },
    {
      "epoch": 4.873784737159081,
      "grad_norm": 0.9453125,
      "learning_rate": 1.5727851986571117e-06,
      "loss": 1.2291,
      "step": 56648
    },
    {
      "epoch": 4.873870773466403,
      "grad_norm": 248.0,
      "learning_rate": 1.5706428326333066e-06,
      "loss": 1.0334,
      "step": 56649
    },
    {
      "epoch": 4.873956809773724,
      "grad_norm": 0.9609375,
      "learning_rate": 1.5685019244184306e-06,
      "loss": 1.2527,
      "step": 56650
    },
    {
      "epoch": 4.874042846081046,
      "grad_norm": 1.3984375,
      "learning_rate": 1.5663624740187565e-06,
      "loss": 0.9752,
      "step": 56651
    },
    {
      "epoch": 4.8741288823883675,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5642244814405016e-06,
      "loss": 0.8385,
      "step": 56652
    },
    {
      "epoch": 4.87421491869569,
      "grad_norm": 0.8984375,
      "learning_rate": 1.562087946689994e-06,
      "loss": 1.0674,
      "step": 56653
    },
    {
      "epoch": 4.874300955003012,
      "grad_norm": 1.28125,
      "learning_rate": 1.5599528697733956e-06,
      "loss": 1.3534,
      "step": 56654
    },
    {
      "epoch": 4.874386991310333,
      "grad_norm": 1.203125,
      "learning_rate": 1.557819250696979e-06,
      "loss": 0.9995,
      "step": 56655
    },
    {
      "epoch": 4.874473027617655,
      "grad_norm": 0.96875,
      "learning_rate": 1.5556870894670172e-06,
      "loss": 0.3646,
      "step": 56656
    },
    {
      "epoch": 4.874559063924976,
      "grad_norm": 0.96875,
      "learning_rate": 1.5535563860897273e-06,
      "loss": 1.4282,
      "step": 56657
    },
    {
      "epoch": 4.874645100232298,
      "grad_norm": 2.234375,
      "learning_rate": 1.551427140571382e-06,
      "loss": 0.9632,
      "step": 56658
    },
    {
      "epoch": 4.874731136539619,
      "grad_norm": 0.87890625,
      "learning_rate": 1.5492993529181432e-06,
      "loss": 1.6477,
      "step": 56659
    },
    {
      "epoch": 4.874817172846941,
      "grad_norm": 1.46875,
      "learning_rate": 1.547173023136228e-06,
      "loss": 1.6142,
      "step": 56660
    },
    {
      "epoch": 4.8749032091542635,
      "grad_norm": 1.015625,
      "learning_rate": 1.5450481512319647e-06,
      "loss": 0.6958,
      "step": 56661
    },
    {
      "epoch": 4.874989245461585,
      "grad_norm": 1.703125,
      "learning_rate": 1.542924737211404e-06,
      "loss": 0.9736,
      "step": 56662
    },
    {
      "epoch": 4.875075281768907,
      "grad_norm": 1.2421875,
      "learning_rate": 1.54080278108093e-06,
      "loss": 1.4951,
      "step": 56663
    },
    {
      "epoch": 4.875161318076228,
      "grad_norm": 1.3984375,
      "learning_rate": 1.538682282846593e-06,
      "loss": 0.9628,
      "step": 56664
    },
    {
      "epoch": 4.87524735438355,
      "grad_norm": 0.87109375,
      "learning_rate": 1.5365632425147213e-06,
      "loss": 0.8318,
      "step": 56665
    },
    {
      "epoch": 4.875333390690871,
      "grad_norm": 1.1328125,
      "learning_rate": 1.5344456600914214e-06,
      "loss": 1.6511,
      "step": 56666
    },
    {
      "epoch": 4.875419426998193,
      "grad_norm": 1.1328125,
      "learning_rate": 1.5323295355829103e-06,
      "loss": 1.6899,
      "step": 56667
    },
    {
      "epoch": 4.875505463305515,
      "grad_norm": 1.078125,
      "learning_rate": 1.5302148689954055e-06,
      "loss": 1.843,
      "step": 56668
    },
    {
      "epoch": 4.875591499612836,
      "grad_norm": 1.109375,
      "learning_rate": 1.5281016603350684e-06,
      "loss": 1.168,
      "step": 56669
    },
    {
      "epoch": 4.8756775359201585,
      "grad_norm": 1.015625,
      "learning_rate": 1.5259899096081165e-06,
      "loss": 0.557,
      "step": 56670
    },
    {
      "epoch": 4.87576357222748,
      "grad_norm": 1.2109375,
      "learning_rate": 1.5238796168206558e-06,
      "loss": 1.3277,
      "step": 56671
    },
    {
      "epoch": 4.875849608534802,
      "grad_norm": 1.046875,
      "learning_rate": 1.5217707819789039e-06,
      "loss": 1.6141,
      "step": 56672
    },
    {
      "epoch": 4.875935644842123,
      "grad_norm": 0.9609375,
      "learning_rate": 1.5196634050890223e-06,
      "loss": 1.0139,
      "step": 56673
    },
    {
      "epoch": 4.876021681149445,
      "grad_norm": 0.98828125,
      "learning_rate": 1.5175574861572284e-06,
      "loss": 0.4541,
      "step": 56674
    },
    {
      "epoch": 4.876107717456767,
      "grad_norm": 1.2109375,
      "learning_rate": 1.5154530251895726e-06,
      "loss": 0.986,
      "step": 56675
    },
    {
      "epoch": 4.876193753764088,
      "grad_norm": 1.125,
      "learning_rate": 1.5133500221922724e-06,
      "loss": 1.2992,
      "step": 56676
    },
    {
      "epoch": 4.87627979007141,
      "grad_norm": 0.9375,
      "learning_rate": 1.5112484771714897e-06,
      "loss": 1.2873,
      "step": 56677
    },
    {
      "epoch": 4.8763658263787315,
      "grad_norm": 0.984375,
      "learning_rate": 1.5091483901333857e-06,
      "loss": 0.9084,
      "step": 56678
    },
    {
      "epoch": 4.876451862686054,
      "grad_norm": 32.5,
      "learning_rate": 1.5070497610840116e-06,
      "loss": 1.3181,
      "step": 56679
    },
    {
      "epoch": 4.876537898993375,
      "grad_norm": 17.375,
      "learning_rate": 1.5049525900295847e-06,
      "loss": 1.2535,
      "step": 56680
    },
    {
      "epoch": 4.876623935300697,
      "grad_norm": 282.0,
      "learning_rate": 1.5028568769762109e-06,
      "loss": 0.6317,
      "step": 56681
    },
    {
      "epoch": 4.876709971608019,
      "grad_norm": 0.98046875,
      "learning_rate": 1.5007626219300518e-06,
      "loss": 1.1383,
      "step": 56682
    },
    {
      "epoch": 4.87679600791534,
      "grad_norm": 1.125,
      "learning_rate": 1.498669824897214e-06,
      "loss": 1.1641,
      "step": 56683
    },
    {
      "epoch": 4.876882044222662,
      "grad_norm": 1.46875,
      "learning_rate": 1.496578485883804e-06,
      "loss": 1.1007,
      "step": 56684
    },
    {
      "epoch": 4.876968080529983,
      "grad_norm": 1.21875,
      "learning_rate": 1.4944886048959273e-06,
      "loss": 1.4955,
      "step": 56685
    },
    {
      "epoch": 4.877054116837305,
      "grad_norm": 1.125,
      "learning_rate": 1.492400181939746e-06,
      "loss": 0.637,
      "step": 56686
    },
    {
      "epoch": 4.877140153144627,
      "grad_norm": 0.8671875,
      "learning_rate": 1.4903132170213106e-06,
      "loss": 0.697,
      "step": 56687
    },
    {
      "epoch": 4.877226189451949,
      "grad_norm": 1.1796875,
      "learning_rate": 1.4882277101467833e-06,
      "loss": 0.9026,
      "step": 56688
    },
    {
      "epoch": 4.877312225759271,
      "grad_norm": 1.2890625,
      "learning_rate": 1.4861436613222147e-06,
      "loss": 1.5329,
      "step": 56689
    },
    {
      "epoch": 4.877398262066592,
      "grad_norm": 1.125,
      "learning_rate": 1.4840610705537104e-06,
      "loss": 1.4463,
      "step": 56690
    },
    {
      "epoch": 4.877484298373914,
      "grad_norm": 0.921875,
      "learning_rate": 1.4819799378473776e-06,
      "loss": 1.1793,
      "step": 56691
    },
    {
      "epoch": 4.877570334681235,
      "grad_norm": 1.046875,
      "learning_rate": 1.4799002632093218e-06,
      "loss": 1.3947,
      "step": 56692
    },
    {
      "epoch": 4.877656370988557,
      "grad_norm": 1.09375,
      "learning_rate": 1.4778220466455939e-06,
      "loss": 0.9994,
      "step": 56693
    },
    {
      "epoch": 4.877742407295878,
      "grad_norm": 1.171875,
      "learning_rate": 1.4757452881623e-06,
      "loss": 0.6986,
      "step": 56694
    },
    {
      "epoch": 4.8778284436032004,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4736699877654358e-06,
      "loss": 1.1213,
      "step": 56695
    },
    {
      "epoch": 4.8779144799105225,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4715961454611625e-06,
      "loss": 0.795,
      "step": 56696
    },
    {
      "epoch": 4.878000516217844,
      "grad_norm": 0.953125,
      "learning_rate": 1.4695237612555312e-06,
      "loss": 1.6772,
      "step": 56697
    },
    {
      "epoch": 4.878086552525166,
      "grad_norm": 1.0859375,
      "learning_rate": 1.467452835154537e-06,
      "loss": 1.055,
      "step": 56698
    },
    {
      "epoch": 4.878172588832487,
      "grad_norm": 64.5,
      "learning_rate": 1.4653833671643412e-06,
      "loss": 1.2008,
      "step": 56699
    },
    {
      "epoch": 4.878258625139809,
      "grad_norm": 1.2265625,
      "learning_rate": 1.4633153572908842e-06,
      "loss": 0.4267,
      "step": 56700
    },
    {
      "epoch": 4.87834466144713,
      "grad_norm": 0.88671875,
      "learning_rate": 1.4612488055402717e-06,
      "loss": 1.0583,
      "step": 56701
    },
    {
      "epoch": 4.878430697754452,
      "grad_norm": 1.34375,
      "learning_rate": 1.4591837119186102e-06,
      "loss": 1.394,
      "step": 56702
    },
    {
      "epoch": 4.878516734061774,
      "grad_norm": 0.90625,
      "learning_rate": 1.4571200764318394e-06,
      "loss": 0.809,
      "step": 56703
    },
    {
      "epoch": 4.8786027703690955,
      "grad_norm": 1.40625,
      "learning_rate": 1.4550578990860097e-06,
      "loss": 1.7641,
      "step": 56704
    },
    {
      "epoch": 4.878688806676418,
      "grad_norm": 1.0390625,
      "learning_rate": 1.452997179887172e-06,
      "loss": 0.698,
      "step": 56705
    },
    {
      "epoch": 4.878774842983739,
      "grad_norm": 0.91015625,
      "learning_rate": 1.4509379188413774e-06,
      "loss": 0.6227,
      "step": 56706
    },
    {
      "epoch": 4.878860879291061,
      "grad_norm": 0.96875,
      "learning_rate": 1.4488801159546205e-06,
      "loss": 1.3869,
      "step": 56707
    },
    {
      "epoch": 4.878946915598383,
      "grad_norm": 1.21875,
      "learning_rate": 1.4468237712329525e-06,
      "loss": 1.7927,
      "step": 56708
    },
    {
      "epoch": 4.879032951905704,
      "grad_norm": 1.5,
      "learning_rate": 1.4447688846823127e-06,
      "loss": 1.291,
      "step": 56709
    },
    {
      "epoch": 4.879118988213026,
      "grad_norm": 1.96875,
      "learning_rate": 1.4427154563088074e-06,
      "loss": 1.1667,
      "step": 56710
    },
    {
      "epoch": 4.879205024520347,
      "grad_norm": 1.4453125,
      "learning_rate": 1.4406634861183766e-06,
      "loss": 1.5167,
      "step": 56711
    },
    {
      "epoch": 4.879291060827669,
      "grad_norm": 1.0859375,
      "learning_rate": 1.438612974117015e-06,
      "loss": 1.4852,
      "step": 56712
    },
    {
      "epoch": 4.8793770971349915,
      "grad_norm": 1.078125,
      "learning_rate": 1.436563920310774e-06,
      "loss": 0.7784,
      "step": 56713
    },
    {
      "epoch": 4.879463133442313,
      "grad_norm": 1.09375,
      "learning_rate": 1.4345163247055925e-06,
      "loss": 1.3097,
      "step": 56714
    },
    {
      "epoch": 4.879549169749635,
      "grad_norm": 1.0859375,
      "learning_rate": 1.4324701873075219e-06,
      "loss": 0.8143,
      "step": 56715
    },
    {
      "epoch": 4.879635206056956,
      "grad_norm": 1.0078125,
      "learning_rate": 1.4304255081225016e-06,
      "loss": 0.7201,
      "step": 56716
    },
    {
      "epoch": 4.879721242364278,
      "grad_norm": 1.15625,
      "learning_rate": 1.4283822871565266e-06,
      "loss": 1.4822,
      "step": 56717
    },
    {
      "epoch": 4.879807278671599,
      "grad_norm": 1.234375,
      "learning_rate": 1.426340524415537e-06,
      "loss": 1.3871,
      "step": 56718
    },
    {
      "epoch": 4.879893314978921,
      "grad_norm": 1.25,
      "learning_rate": 1.4243002199055277e-06,
      "loss": 1.5242,
      "step": 56719
    },
    {
      "epoch": 4.879979351286243,
      "grad_norm": 1.3359375,
      "learning_rate": 1.422261373632494e-06,
      "loss": 0.674,
      "step": 56720
    },
    {
      "epoch": 4.8800653875935645,
      "grad_norm": 5.28125,
      "learning_rate": 1.4202239856023758e-06,
      "loss": 1.1063,
      "step": 56721
    },
    {
      "epoch": 4.8801514239008865,
      "grad_norm": 1.125,
      "learning_rate": 1.418188055821168e-06,
      "loss": 0.7448,
      "step": 56722
    },
    {
      "epoch": 4.880237460208208,
      "grad_norm": 12.375,
      "learning_rate": 1.4161535842947548e-06,
      "loss": 0.7898,
      "step": 56723
    },
    {
      "epoch": 4.88032349651553,
      "grad_norm": 1.078125,
      "learning_rate": 1.4141205710291317e-06,
      "loss": 0.7083,
      "step": 56724
    },
    {
      "epoch": 4.880409532822851,
      "grad_norm": 1.0078125,
      "learning_rate": 1.412089016030238e-06,
      "loss": 1.0693,
      "step": 56725
    },
    {
      "epoch": 4.880495569130173,
      "grad_norm": 1.53125,
      "learning_rate": 1.4100589193039582e-06,
      "loss": 1.0594,
      "step": 56726
    },
    {
      "epoch": 4.880581605437495,
      "grad_norm": 1.2734375,
      "learning_rate": 1.4080302808563982e-06,
      "loss": 0.9478,
      "step": 56727
    },
    {
      "epoch": 4.880667641744816,
      "grad_norm": 1.1171875,
      "learning_rate": 1.4060031006932761e-06,
      "loss": 0.805,
      "step": 56728
    },
    {
      "epoch": 4.880753678052138,
      "grad_norm": 1.3515625,
      "learning_rate": 1.4039773788206423e-06,
      "loss": 1.8055,
      "step": 56729
    },
    {
      "epoch": 4.8808397143594595,
      "grad_norm": 0.96875,
      "learning_rate": 1.4019531152444365e-06,
      "loss": 0.6537,
      "step": 56730
    },
    {
      "epoch": 4.880925750666782,
      "grad_norm": 1.546875,
      "learning_rate": 1.399930309970543e-06,
      "loss": 1.7821,
      "step": 56731
    },
    {
      "epoch": 4.881011786974103,
      "grad_norm": 1.3984375,
      "learning_rate": 1.3979089630048459e-06,
      "loss": 1.2778,
      "step": 56732
    },
    {
      "epoch": 4.881097823281425,
      "grad_norm": 1.6015625,
      "learning_rate": 1.3958890743533402e-06,
      "loss": 0.7896,
      "step": 56733
    },
    {
      "epoch": 4.881183859588747,
      "grad_norm": 1.2421875,
      "learning_rate": 1.393870644021855e-06,
      "loss": 1.0577,
      "step": 56734
    },
    {
      "epoch": 4.881269895896068,
      "grad_norm": 9.9375,
      "learning_rate": 1.39185367201633e-06,
      "loss": 0.7344,
      "step": 56735
    },
    {
      "epoch": 4.88135593220339,
      "grad_norm": 0.91796875,
      "learning_rate": 1.389838158342649e-06,
      "loss": 0.8388,
      "step": 56736
    },
    {
      "epoch": 4.881441968510711,
      "grad_norm": 1.1875,
      "learning_rate": 1.3878241030067519e-06,
      "loss": 0.7578,
      "step": 56737
    },
    {
      "epoch": 4.881528004818033,
      "grad_norm": 1.296875,
      "learning_rate": 1.385811506014467e-06,
      "loss": 1.0987,
      "step": 56738
    },
    {
      "epoch": 4.881614041125355,
      "grad_norm": 1.6171875,
      "learning_rate": 1.3838003673717348e-06,
      "loss": 0.7141,
      "step": 56739
    },
    {
      "epoch": 4.881700077432677,
      "grad_norm": 39.5,
      "learning_rate": 1.3817906870843834e-06,
      "loss": 1.2308,
      "step": 56740
    },
    {
      "epoch": 4.881786113739999,
      "grad_norm": 4.125,
      "learning_rate": 1.3797824651583523e-06,
      "loss": 1.0143,
      "step": 56741
    },
    {
      "epoch": 4.88187215004732,
      "grad_norm": 1.140625,
      "learning_rate": 1.3777757015994152e-06,
      "loss": 0.8331,
      "step": 56742
    },
    {
      "epoch": 4.881958186354642,
      "grad_norm": 1.296875,
      "learning_rate": 1.375770396413567e-06,
      "loss": 1.4498,
      "step": 56743
    },
    {
      "epoch": 4.882044222661963,
      "grad_norm": 1.453125,
      "learning_rate": 1.373766549606581e-06,
      "loss": 0.5102,
      "step": 56744
    },
    {
      "epoch": 4.882130258969285,
      "grad_norm": 1.0390625,
      "learning_rate": 1.3717641611843413e-06,
      "loss": 0.6283,
      "step": 56745
    },
    {
      "epoch": 4.882216295276606,
      "grad_norm": 1.0390625,
      "learning_rate": 1.3697632311527875e-06,
      "loss": 1.7459,
      "step": 56746
    },
    {
      "epoch": 4.8823023315839285,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3677637595176373e-06,
      "loss": 0.951,
      "step": 56747
    },
    {
      "epoch": 4.8823883678912505,
      "grad_norm": 12.125,
      "learning_rate": 1.365765746284775e-06,
      "loss": 1.7534,
      "step": 56748
    },
    {
      "epoch": 4.882474404198572,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3637691914601402e-06,
      "loss": 0.8774,
      "step": 56749
    },
    {
      "epoch": 4.882560440505894,
      "grad_norm": 0.97265625,
      "learning_rate": 1.3617740950494506e-06,
      "loss": 0.9227,
      "step": 56750
    },
    {
      "epoch": 4.882646476813215,
      "grad_norm": 0.8125,
      "learning_rate": 1.3597804570585904e-06,
      "loss": 0.9004,
      "step": 56751
    },
    {
      "epoch": 4.882732513120537,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3577882774934436e-06,
      "loss": 1.4654,
      "step": 56752
    },
    {
      "epoch": 4.882818549427858,
      "grad_norm": 1.328125,
      "learning_rate": 1.355797556359728e-06,
      "loss": 1.6157,
      "step": 56753
    },
    {
      "epoch": 4.88290458573518,
      "grad_norm": 78.5,
      "learning_rate": 1.3538082936633278e-06,
      "loss": 1.5401,
      "step": 56754
    },
    {
      "epoch": 4.882990622042502,
      "grad_norm": 1.21875,
      "learning_rate": 1.3518204894100715e-06,
      "loss": 1.2326,
      "step": 56755
    },
    {
      "epoch": 4.8830766583498235,
      "grad_norm": 1.3046875,
      "learning_rate": 1.349834143605788e-06,
      "loss": 1.2059,
      "step": 56756
    },
    {
      "epoch": 4.883162694657146,
      "grad_norm": 1.0546875,
      "learning_rate": 1.347849256256195e-06,
      "loss": 1.6631,
      "step": 56757
    },
    {
      "epoch": 4.883248730964467,
      "grad_norm": 1.1953125,
      "learning_rate": 1.345865827367232e-06,
      "loss": 0.879,
      "step": 56758
    },
    {
      "epoch": 4.883334767271789,
      "grad_norm": 1.359375,
      "learning_rate": 1.343883856944561e-06,
      "loss": 1.6333,
      "step": 56759
    },
    {
      "epoch": 4.88342080357911,
      "grad_norm": 3.0625,
      "learning_rate": 1.3419033449940665e-06,
      "loss": 0.8238,
      "step": 56760
    },
    {
      "epoch": 4.883506839886432,
      "grad_norm": 2.09375,
      "learning_rate": 1.339924291521577e-06,
      "loss": 1.2308,
      "step": 56761
    },
    {
      "epoch": 4.883592876193754,
      "grad_norm": 0.859375,
      "learning_rate": 1.3379466965327546e-06,
      "loss": 0.9873,
      "step": 56762
    },
    {
      "epoch": 4.883678912501075,
      "grad_norm": 1.4296875,
      "learning_rate": 1.3359705600334838e-06,
      "loss": 1.2799,
      "step": 56763
    },
    {
      "epoch": 4.883764948808397,
      "grad_norm": 1.03125,
      "learning_rate": 1.3339958820294818e-06,
      "loss": 0.6104,
      "step": 56764
    },
    {
      "epoch": 4.883850985115719,
      "grad_norm": 0.8125,
      "learning_rate": 1.332022662526633e-06,
      "loss": 0.6995,
      "step": 56765
    },
    {
      "epoch": 4.883937021423041,
      "grad_norm": 1.15625,
      "learning_rate": 1.330050901530544e-06,
      "loss": 1.0851,
      "step": 56766
    },
    {
      "epoch": 4.884023057730362,
      "grad_norm": 1.0546875,
      "learning_rate": 1.328080599047099e-06,
      "loss": 1.2263,
      "step": 56767
    },
    {
      "epoch": 4.884109094037684,
      "grad_norm": 1.1484375,
      "learning_rate": 1.3261117550820712e-06,
      "loss": 1.3315,
      "step": 56768
    },
    {
      "epoch": 4.884195130345006,
      "grad_norm": 1.3125,
      "learning_rate": 1.3241443696411227e-06,
      "loss": 1.6344,
      "step": 56769
    },
    {
      "epoch": 4.884281166652327,
      "grad_norm": 8.375,
      "learning_rate": 1.3221784427301375e-06,
      "loss": 0.716,
      "step": 56770
    },
    {
      "epoch": 4.884367202959649,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3202139743547225e-06,
      "loss": 0.5852,
      "step": 56771
    },
    {
      "epoch": 4.88445323926697,
      "grad_norm": 1.015625,
      "learning_rate": 1.3182509645207065e-06,
      "loss": 0.7913,
      "step": 56772
    },
    {
      "epoch": 4.8845392755742925,
      "grad_norm": 1.265625,
      "learning_rate": 1.316289413233862e-06,
      "loss": 0.9333,
      "step": 56773
    },
    {
      "epoch": 4.884625311881614,
      "grad_norm": 0.85546875,
      "learning_rate": 1.3143293204998518e-06,
      "loss": 0.7403,
      "step": 56774
    },
    {
      "epoch": 4.884711348188936,
      "grad_norm": 0.91015625,
      "learning_rate": 1.3123706863243934e-06,
      "loss": 0.7224,
      "step": 56775
    },
    {
      "epoch": 4.884797384496258,
      "grad_norm": 1.1953125,
      "learning_rate": 1.3104135107133153e-06,
      "loss": 0.6923,
      "step": 56776
    },
    {
      "epoch": 4.884883420803579,
      "grad_norm": 1.234375,
      "learning_rate": 1.3084577936722796e-06,
      "loss": 0.815,
      "step": 56777
    },
    {
      "epoch": 4.884969457110901,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3065035352070042e-06,
      "loss": 1.6176,
      "step": 56778
    },
    {
      "epoch": 4.885055493418222,
      "grad_norm": 1.2421875,
      "learning_rate": 1.3045507353232622e-06,
      "loss": 1.4154,
      "step": 56779
    },
    {
      "epoch": 4.885141529725544,
      "grad_norm": 1.140625,
      "learning_rate": 1.30259939402666e-06,
      "loss": 1.4171,
      "step": 56780
    },
    {
      "epoch": 4.8852275660328655,
      "grad_norm": 1.2265625,
      "learning_rate": 1.3006495113230266e-06,
      "loss": 0.9847,
      "step": 56781
    },
    {
      "epoch": 4.8853136023401875,
      "grad_norm": 1.3203125,
      "learning_rate": 1.2987010872179682e-06,
      "loss": 1.7911,
      "step": 56782
    },
    {
      "epoch": 4.88539963864751,
      "grad_norm": 1.3203125,
      "learning_rate": 1.296754121717203e-06,
      "loss": 1.3372,
      "step": 56783
    },
    {
      "epoch": 4.885485674954831,
      "grad_norm": 1.109375,
      "learning_rate": 1.2948086148264481e-06,
      "loss": 0.6457,
      "step": 56784
    },
    {
      "epoch": 4.885571711262153,
      "grad_norm": 1.25,
      "learning_rate": 1.292864566551366e-06,
      "loss": 1.0197,
      "step": 56785
    },
    {
      "epoch": 4.885657747569474,
      "grad_norm": 1.2890625,
      "learning_rate": 1.2909219768976744e-06,
      "loss": 0.6701,
      "step": 56786
    },
    {
      "epoch": 4.885743783876796,
      "grad_norm": 0.9140625,
      "learning_rate": 1.288980845871035e-06,
      "loss": 0.6626,
      "step": 56787
    },
    {
      "epoch": 4.885829820184117,
      "grad_norm": 1.40625,
      "learning_rate": 1.287041173477166e-06,
      "loss": 0.8218,
      "step": 56788
    },
    {
      "epoch": 4.885915856491439,
      "grad_norm": 0.9296875,
      "learning_rate": 1.2851029597216734e-06,
      "loss": 1.2282,
      "step": 56789
    },
    {
      "epoch": 4.886001892798761,
      "grad_norm": 1.40625,
      "learning_rate": 1.2831662046102754e-06,
      "loss": 0.8593,
      "step": 56790
    },
    {
      "epoch": 4.886087929106083,
      "grad_norm": 0.96875,
      "learning_rate": 1.281230908148634e-06,
      "loss": 0.6825,
      "step": 56791
    },
    {
      "epoch": 4.886173965413405,
      "grad_norm": 1.328125,
      "learning_rate": 1.2792970703423556e-06,
      "loss": 1.2454,
      "step": 56792
    },
    {
      "epoch": 4.886260001720726,
      "grad_norm": 1.140625,
      "learning_rate": 1.2773646911971581e-06,
      "loss": 1.2379,
      "step": 56793
    },
    {
      "epoch": 4.886346038028048,
      "grad_norm": 1.2734375,
      "learning_rate": 1.2754337707187036e-06,
      "loss": 0.6252,
      "step": 56794
    },
    {
      "epoch": 4.886432074335369,
      "grad_norm": 1.28125,
      "learning_rate": 1.2735043089125985e-06,
      "loss": 1.1117,
      "step": 56795
    },
    {
      "epoch": 4.886518110642691,
      "grad_norm": 540.0,
      "learning_rate": 1.2715763057844499e-06,
      "loss": 1.7223,
      "step": 56796
    },
    {
      "epoch": 4.886604146950013,
      "grad_norm": 1.2109375,
      "learning_rate": 1.269649761339975e-06,
      "loss": 0.9751,
      "step": 56797
    },
    {
      "epoch": 4.886690183257334,
      "grad_norm": 1.0625,
      "learning_rate": 1.2677246755847805e-06,
      "loss": 0.7786,
      "step": 56798
    },
    {
      "epoch": 4.8867762195646565,
      "grad_norm": 0.9765625,
      "learning_rate": 1.2658010485245285e-06,
      "loss": 1.3663,
      "step": 56799
    },
    {
      "epoch": 4.886862255871978,
      "grad_norm": 1.6640625,
      "learning_rate": 1.263878880164715e-06,
      "loss": 0.6599,
      "step": 56800
    },
    {
      "epoch": 4.8869482921793,
      "grad_norm": 1.0703125,
      "learning_rate": 1.2619581705111683e-06,
      "loss": 0.5539,
      "step": 56801
    },
    {
      "epoch": 4.887034328486621,
      "grad_norm": 1.0703125,
      "learning_rate": 1.2600389195693284e-06,
      "loss": 0.998,
      "step": 56802
    },
    {
      "epoch": 4.887120364793943,
      "grad_norm": 1.390625,
      "learning_rate": 1.2581211273448578e-06,
      "loss": 1.1811,
      "step": 56803
    },
    {
      "epoch": 4.887206401101265,
      "grad_norm": 4032.0,
      "learning_rate": 1.2562047938434185e-06,
      "loss": 1.6191,
      "step": 56804
    },
    {
      "epoch": 4.887292437408586,
      "grad_norm": 1.25,
      "learning_rate": 1.2542899190705614e-06,
      "loss": 1.2534,
      "step": 56805
    },
    {
      "epoch": 4.887378473715908,
      "grad_norm": 1.171875,
      "learning_rate": 1.2523765030319489e-06,
      "loss": 1.0763,
      "step": 56806
    },
    {
      "epoch": 4.8874645100232295,
      "grad_norm": 1.3203125,
      "learning_rate": 1.2504645457330765e-06,
      "loss": 1.5138,
      "step": 56807
    },
    {
      "epoch": 4.8875505463305515,
      "grad_norm": 0.89453125,
      "learning_rate": 1.2485540471796064e-06,
      "loss": 0.8818,
      "step": 56808
    },
    {
      "epoch": 4.887636582637873,
      "grad_norm": 0.9765625,
      "learning_rate": 1.2466450073771452e-06,
      "loss": 1.2354,
      "step": 56809
    },
    {
      "epoch": 4.887722618945195,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2447374263312438e-06,
      "loss": 0.964,
      "step": 56810
    },
    {
      "epoch": 4.887808655252517,
      "grad_norm": 0.984375,
      "learning_rate": 1.2428313040474538e-06,
      "loss": 0.923,
      "step": 56811
    },
    {
      "epoch": 4.887894691559838,
      "grad_norm": 52.5,
      "learning_rate": 1.2409266405313812e-06,
      "loss": 0.9905,
      "step": 56812
    },
    {
      "epoch": 4.88798072786716,
      "grad_norm": 0.87109375,
      "learning_rate": 1.2390234357885776e-06,
      "loss": 0.8199,
      "step": 56813
    },
    {
      "epoch": 4.888066764174481,
      "grad_norm": 1.34375,
      "learning_rate": 1.2371216898247051e-06,
      "loss": 0.7561,
      "step": 56814
    },
    {
      "epoch": 4.888152800481803,
      "grad_norm": 1.3046875,
      "learning_rate": 1.2352214026452036e-06,
      "loss": 0.9863,
      "step": 56815
    },
    {
      "epoch": 4.8882388367891245,
      "grad_norm": 12.6875,
      "learning_rate": 1.2333225742556242e-06,
      "loss": 0.895,
      "step": 56816
    },
    {
      "epoch": 4.888324873096447,
      "grad_norm": 1.203125,
      "learning_rate": 1.2314252046616292e-06,
      "loss": 1.3326,
      "step": 56817
    },
    {
      "epoch": 4.888410909403769,
      "grad_norm": 12.625,
      "learning_rate": 1.2295292938687141e-06,
      "loss": 1.8962,
      "step": 56818
    },
    {
      "epoch": 4.88849694571109,
      "grad_norm": 1.6484375,
      "learning_rate": 1.2276348418823745e-06,
      "loss": 1.3107,
      "step": 56819
    },
    {
      "epoch": 4.888582982018412,
      "grad_norm": 1.375,
      "learning_rate": 1.2257418487082728e-06,
      "loss": 1.2376,
      "step": 56820
    },
    {
      "epoch": 4.888669018325733,
      "grad_norm": 1.296875,
      "learning_rate": 1.2238503143517931e-06,
      "loss": 1.3148,
      "step": 56821
    },
    {
      "epoch": 4.888755054633055,
      "grad_norm": 1.40625,
      "learning_rate": 1.221960238818598e-06,
      "loss": 1.0888,
      "step": 56822
    },
    {
      "epoch": 4.888841090940377,
      "grad_norm": 1.4609375,
      "learning_rate": 1.2200716221141273e-06,
      "loss": 1.3432,
      "step": 56823
    },
    {
      "epoch": 4.888927127247698,
      "grad_norm": 1.4609375,
      "learning_rate": 1.2181844642439322e-06,
      "loss": 1.1177,
      "step": 56824
    },
    {
      "epoch": 4.8890131635550205,
      "grad_norm": 0.91796875,
      "learning_rate": 1.2162987652135637e-06,
      "loss": 0.9042,
      "step": 56825
    },
    {
      "epoch": 4.889099199862342,
      "grad_norm": 1.046875,
      "learning_rate": 1.2144145250285177e-06,
      "loss": 1.502,
      "step": 56826
    },
    {
      "epoch": 4.889185236169664,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2125317436942895e-06,
      "loss": 0.5147,
      "step": 56827
    },
    {
      "epoch": 4.889271272476986,
      "grad_norm": 1.03125,
      "learning_rate": 1.210650421216375e-06,
      "loss": 0.9094,
      "step": 56828
    },
    {
      "epoch": 4.889357308784307,
      "grad_norm": 1.2734375,
      "learning_rate": 1.2087705576003249e-06,
      "loss": 1.2713,
      "step": 56829
    },
    {
      "epoch": 4.889443345091629,
      "grad_norm": 3.75,
      "learning_rate": 1.2068921528516352e-06,
      "loss": 0.8774,
      "step": 56830
    },
    {
      "epoch": 4.88952938139895,
      "grad_norm": 1.0625,
      "learning_rate": 1.2050152069757457e-06,
      "loss": 2.2901,
      "step": 56831
    },
    {
      "epoch": 4.889615417706272,
      "grad_norm": 1.3359375,
      "learning_rate": 1.2031397199781523e-06,
      "loss": 0.6228,
      "step": 56832
    },
    {
      "epoch": 4.8897014540135935,
      "grad_norm": 1.171875,
      "learning_rate": 1.201265691864406e-06,
      "loss": 0.9838,
      "step": 56833
    },
    {
      "epoch": 4.8897874903209155,
      "grad_norm": 6.875,
      "learning_rate": 1.1993931226399469e-06,
      "loss": 1.6891,
      "step": 56834
    },
    {
      "epoch": 4.889873526628238,
      "grad_norm": 0.78125,
      "learning_rate": 1.1975220123102148e-06,
      "loss": 0.6375,
      "step": 56835
    },
    {
      "epoch": 4.889959562935559,
      "grad_norm": 1.078125,
      "learning_rate": 1.1956523608807612e-06,
      "loss": 0.979,
      "step": 56836
    },
    {
      "epoch": 4.890045599242881,
      "grad_norm": 0.91015625,
      "learning_rate": 1.1937841683570262e-06,
      "loss": 0.7494,
      "step": 56837
    },
    {
      "epoch": 4.890131635550202,
      "grad_norm": 1.171875,
      "learning_rate": 1.191917434744394e-06,
      "loss": 1.1031,
      "step": 56838
    },
    {
      "epoch": 4.890217671857524,
      "grad_norm": 1.046875,
      "learning_rate": 1.190052160048416e-06,
      "loss": 1.6192,
      "step": 56839
    },
    {
      "epoch": 4.890303708164845,
      "grad_norm": 1.2578125,
      "learning_rate": 1.1881883442745322e-06,
      "loss": 1.2949,
      "step": 56840
    },
    {
      "epoch": 4.890389744472167,
      "grad_norm": 1.0390625,
      "learning_rate": 1.186325987428183e-06,
      "loss": 1.326,
      "step": 56841
    },
    {
      "epoch": 4.890475780779489,
      "grad_norm": 0.91015625,
      "learning_rate": 1.184465089514808e-06,
      "loss": 0.4914,
      "step": 56842
    },
    {
      "epoch": 4.890561817086811,
      "grad_norm": 0.75390625,
      "learning_rate": 1.182605650539903e-06,
      "loss": 0.8329,
      "step": 56843
    },
    {
      "epoch": 4.890647853394133,
      "grad_norm": 1.1953125,
      "learning_rate": 1.1807476705087971e-06,
      "loss": 1.238,
      "step": 56844
    },
    {
      "epoch": 4.890733889701454,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1788911494270416e-06,
      "loss": 0.7422,
      "step": 56845
    },
    {
      "epoch": 4.890819926008776,
      "grad_norm": 1.3046875,
      "learning_rate": 1.1770360872999653e-06,
      "loss": 0.7886,
      "step": 56846
    },
    {
      "epoch": 4.890905962316097,
      "grad_norm": 0.99609375,
      "learning_rate": 1.1751824841330638e-06,
      "loss": 0.6615,
      "step": 56847
    },
    {
      "epoch": 4.890991998623419,
      "grad_norm": 1.1015625,
      "learning_rate": 1.1733303399317774e-06,
      "loss": 1.2432,
      "step": 56848
    },
    {
      "epoch": 4.891078034930741,
      "grad_norm": 8.875,
      "learning_rate": 1.171479654701435e-06,
      "loss": 1.483,
      "step": 56849
    },
    {
      "epoch": 4.891164071238062,
      "grad_norm": 2.53125,
      "learning_rate": 1.1696304284474767e-06,
      "loss": 1.1176,
      "step": 56850
    },
    {
      "epoch": 4.8912501075453845,
      "grad_norm": 1.421875,
      "learning_rate": 1.1677826611753982e-06,
      "loss": 1.4658,
      "step": 56851
    },
    {
      "epoch": 4.891336143852706,
      "grad_norm": 684.0,
      "learning_rate": 1.1659363528905286e-06,
      "loss": 1.5061,
      "step": 56852
    },
    {
      "epoch": 4.891422180160028,
      "grad_norm": 680.0,
      "learning_rate": 1.1640915035981969e-06,
      "loss": 1.0758,
      "step": 56853
    },
    {
      "epoch": 4.891508216467349,
      "grad_norm": 1.078125,
      "learning_rate": 1.1622481133039542e-06,
      "loss": 0.6777,
      "step": 56854
    },
    {
      "epoch": 4.891594252774671,
      "grad_norm": 4.09375,
      "learning_rate": 1.1604061820131296e-06,
      "loss": 1.139,
      "step": 56855
    },
    {
      "epoch": 4.891680289081993,
      "grad_norm": 1.6328125,
      "learning_rate": 1.158565709731052e-06,
      "loss": 1.908,
      "step": 56856
    },
    {
      "epoch": 4.891766325389314,
      "grad_norm": 0.90234375,
      "learning_rate": 1.1567266964631618e-06,
      "loss": 0.9013,
      "step": 56857
    },
    {
      "epoch": 4.891852361696636,
      "grad_norm": 1.2265625,
      "learning_rate": 1.1548891422148433e-06,
      "loss": 1.1315,
      "step": 56858
    },
    {
      "epoch": 4.8919383980039575,
      "grad_norm": 1.2265625,
      "learning_rate": 1.1530530469914258e-06,
      "loss": 0.5499,
      "step": 56859
    },
    {
      "epoch": 4.8920244343112795,
      "grad_norm": 0.98828125,
      "learning_rate": 1.1512184107982938e-06,
      "loss": 1.2576,
      "step": 56860
    },
    {
      "epoch": 4.892110470618601,
      "grad_norm": 0.78125,
      "learning_rate": 1.1493852336408872e-06,
      "loss": 0.7751,
      "step": 56861
    },
    {
      "epoch": 4.892196506925923,
      "grad_norm": 1.1328125,
      "learning_rate": 1.14755351552448e-06,
      "loss": 0.7765,
      "step": 56862
    },
    {
      "epoch": 4.892282543233245,
      "grad_norm": 1.1484375,
      "learning_rate": 1.1457232564544562e-06,
      "loss": 1.0695,
      "step": 56863
    },
    {
      "epoch": 4.892368579540566,
      "grad_norm": 24.875,
      "learning_rate": 1.1438944564361453e-06,
      "loss": 1.2853,
      "step": 56864
    },
    {
      "epoch": 4.892454615847888,
      "grad_norm": 1.296875,
      "learning_rate": 1.1420671154749874e-06,
      "loss": 1.1835,
      "step": 56865
    },
    {
      "epoch": 4.892540652155209,
      "grad_norm": 0.9921875,
      "learning_rate": 1.1402412335762002e-06,
      "loss": 0.9011,
      "step": 56866
    },
    {
      "epoch": 4.892626688462531,
      "grad_norm": 1.7890625,
      "learning_rate": 1.1384168107452242e-06,
      "loss": 1.4794,
      "step": 56867
    },
    {
      "epoch": 4.8927127247698525,
      "grad_norm": 1.2421875,
      "learning_rate": 1.1365938469872772e-06,
      "loss": 0.8463,
      "step": 56868
    },
    {
      "epoch": 4.892798761077175,
      "grad_norm": 3.390625,
      "learning_rate": 1.134772342307855e-06,
      "loss": 0.7362,
      "step": 56869
    },
    {
      "epoch": 4.892884797384497,
      "grad_norm": 0.91796875,
      "learning_rate": 1.1329522967121753e-06,
      "loss": 0.7801,
      "step": 56870
    },
    {
      "epoch": 4.892970833691818,
      "grad_norm": 1.4140625,
      "learning_rate": 1.1311337102055674e-06,
      "loss": 1.7958,
      "step": 56871
    },
    {
      "epoch": 4.89305686999914,
      "grad_norm": 1.0546875,
      "learning_rate": 1.129316582793416e-06,
      "loss": 0.6296,
      "step": 56872
    },
    {
      "epoch": 4.893142906306461,
      "grad_norm": 1.1484375,
      "learning_rate": 1.1275009144809389e-06,
      "loss": 0.5492,
      "step": 56873
    },
    {
      "epoch": 4.893228942613783,
      "grad_norm": 0.9453125,
      "learning_rate": 1.125686705273521e-06,
      "loss": 0.6836,
      "step": 56874
    },
    {
      "epoch": 4.893314978921104,
      "grad_norm": 1.7265625,
      "learning_rate": 1.1238739551764355e-06,
      "loss": 1.751,
      "step": 56875
    },
    {
      "epoch": 4.893401015228426,
      "grad_norm": 1.015625,
      "learning_rate": 1.122062664194956e-06,
      "loss": 0.7286,
      "step": 56876
    },
    {
      "epoch": 4.8934870515357485,
      "grad_norm": 1.1484375,
      "learning_rate": 1.1202528323344674e-06,
      "loss": 1.4502,
      "step": 56877
    },
    {
      "epoch": 4.89357308784307,
      "grad_norm": 5.5625,
      "learning_rate": 1.1184444596001874e-06,
      "loss": 1.1969,
      "step": 56878
    },
    {
      "epoch": 4.893659124150392,
      "grad_norm": 0.98828125,
      "learning_rate": 1.1166375459974454e-06,
      "loss": 1.5078,
      "step": 56879
    },
    {
      "epoch": 4.893745160457713,
      "grad_norm": 1.3828125,
      "learning_rate": 1.1148320915315146e-06,
      "loss": 1.3372,
      "step": 56880
    },
    {
      "epoch": 4.893831196765035,
      "grad_norm": 1.1171875,
      "learning_rate": 1.1130280962076688e-06,
      "loss": 0.7442,
      "step": 56881
    },
    {
      "epoch": 4.893917233072356,
      "grad_norm": 1.328125,
      "learning_rate": 1.1112255600311816e-06,
      "loss": 2.113,
      "step": 56882
    },
    {
      "epoch": 4.894003269379678,
      "grad_norm": 1.125,
      "learning_rate": 1.1094244830073263e-06,
      "loss": 1.6683,
      "step": 56883
    },
    {
      "epoch": 4.894089305687,
      "grad_norm": 1.75,
      "learning_rate": 1.1076248651413767e-06,
      "loss": 0.9013,
      "step": 56884
    },
    {
      "epoch": 4.8941753419943215,
      "grad_norm": 1.25,
      "learning_rate": 1.1058267064386063e-06,
      "loss": 1.474,
      "step": 56885
    },
    {
      "epoch": 4.8942613783016435,
      "grad_norm": 2.0625,
      "learning_rate": 1.1040300069042885e-06,
      "loss": 1.2624,
      "step": 56886
    },
    {
      "epoch": 4.894347414608965,
      "grad_norm": 1.3046875,
      "learning_rate": 1.102234766543586e-06,
      "loss": 0.9014,
      "step": 56887
    },
    {
      "epoch": 4.894433450916287,
      "grad_norm": 0.95703125,
      "learning_rate": 1.1004409853618836e-06,
      "loss": 0.6716,
      "step": 56888
    },
    {
      "epoch": 4.894519487223608,
      "grad_norm": 1.1796875,
      "learning_rate": 1.0986486633642879e-06,
      "loss": 0.7376,
      "step": 56889
    },
    {
      "epoch": 4.89460552353093,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0968578005561835e-06,
      "loss": 1.7543,
      "step": 56890
    },
    {
      "epoch": 4.894691559838252,
      "grad_norm": 1.3203125,
      "learning_rate": 1.0950683969427333e-06,
      "loss": 1.0442,
      "step": 56891
    },
    {
      "epoch": 4.894777596145573,
      "grad_norm": 1.0703125,
      "learning_rate": 1.0932804525290996e-06,
      "loss": 1.0857,
      "step": 56892
    },
    {
      "epoch": 4.894863632452895,
      "grad_norm": 0.7734375,
      "learning_rate": 1.091493967320667e-06,
      "loss": 1.2888,
      "step": 56893
    },
    {
      "epoch": 4.8949496687602165,
      "grad_norm": 1.2421875,
      "learning_rate": 1.0897089413225426e-06,
      "loss": 0.8908,
      "step": 56894
    },
    {
      "epoch": 4.895035705067539,
      "grad_norm": 5.78125,
      "learning_rate": 1.0879253745399997e-06,
      "loss": 0.9922,
      "step": 56895
    },
    {
      "epoch": 4.89512174137486,
      "grad_norm": 71.5,
      "learning_rate": 1.0861432669782567e-06,
      "loss": 0.8746,
      "step": 56896
    },
    {
      "epoch": 4.895207777682182,
      "grad_norm": 1.09375,
      "learning_rate": 1.0843626186425315e-06,
      "loss": 0.6432,
      "step": 56897
    },
    {
      "epoch": 4.895293813989504,
      "grad_norm": 7.3125,
      "learning_rate": 1.082583429537931e-06,
      "loss": 0.6894,
      "step": 56898
    },
    {
      "epoch": 4.895379850296825,
      "grad_norm": 0.93359375,
      "learning_rate": 1.0808056996698401e-06,
      "loss": 0.7403,
      "step": 56899
    },
    {
      "epoch": 4.895465886604147,
      "grad_norm": 1.0390625,
      "learning_rate": 1.07902942904331e-06,
      "loss": 1.1894,
      "step": 56900
    },
    {
      "epoch": 4.895551922911468,
      "grad_norm": 1.203125,
      "learning_rate": 1.077254617663559e-06,
      "loss": 1.0209,
      "step": 56901
    },
    {
      "epoch": 4.89563795921879,
      "grad_norm": 1.3515625,
      "learning_rate": 1.075481265535805e-06,
      "loss": 1.6267,
      "step": 56902
    },
    {
      "epoch": 4.895723995526112,
      "grad_norm": 1.390625,
      "learning_rate": 1.0737093726652657e-06,
      "loss": 0.883,
      "step": 56903
    },
    {
      "epoch": 4.895810031833434,
      "grad_norm": 1.2578125,
      "learning_rate": 1.0719389390571044e-06,
      "loss": 0.8871,
      "step": 56904
    },
    {
      "epoch": 4.895896068140756,
      "grad_norm": 1.0,
      "learning_rate": 1.0701699647164275e-06,
      "loss": 0.9659,
      "step": 56905
    },
    {
      "epoch": 4.895982104448077,
      "grad_norm": 1.328125,
      "learning_rate": 1.068402449648509e-06,
      "loss": 2.0214,
      "step": 56906
    },
    {
      "epoch": 4.896068140755399,
      "grad_norm": 0.95703125,
      "learning_rate": 1.0666363938584555e-06,
      "loss": 1.0187,
      "step": 56907
    },
    {
      "epoch": 4.89615417706272,
      "grad_norm": 1.15625,
      "learning_rate": 1.0648717973514855e-06,
      "loss": 0.9283,
      "step": 56908
    },
    {
      "epoch": 4.896240213370042,
      "grad_norm": 1.109375,
      "learning_rate": 1.0631086601327056e-06,
      "loss": 1.2915,
      "step": 56909
    },
    {
      "epoch": 4.896326249677363,
      "grad_norm": 1.203125,
      "learning_rate": 1.0613469822072786e-06,
      "loss": 1.0659,
      "step": 56910
    },
    {
      "epoch": 4.8964122859846855,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0595867635804225e-06,
      "loss": 0.5809,
      "step": 56911
    },
    {
      "epoch": 4.8964983222920075,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0578280042571886e-06,
      "loss": 1.5705,
      "step": 56912
    },
    {
      "epoch": 4.896584358599329,
      "grad_norm": 2.015625,
      "learning_rate": 1.0560707042427952e-06,
      "loss": 1.6986,
      "step": 56913
    },
    {
      "epoch": 4.896670394906651,
      "grad_norm": 7.1875,
      "learning_rate": 1.0543148635423493e-06,
      "loss": 0.9725,
      "step": 56914
    },
    {
      "epoch": 4.896756431213972,
      "grad_norm": 1.2109375,
      "learning_rate": 1.052560482160958e-06,
      "loss": 0.6432,
      "step": 56915
    },
    {
      "epoch": 4.896842467521294,
      "grad_norm": 2.421875,
      "learning_rate": 1.0508075601038391e-06,
      "loss": 1.0984,
      "step": 56916
    },
    {
      "epoch": 4.896928503828615,
      "grad_norm": 205.0,
      "learning_rate": 1.0490560973760444e-06,
      "loss": 1.1597,
      "step": 56917
    },
    {
      "epoch": 4.897014540135937,
      "grad_norm": 1.1953125,
      "learning_rate": 1.0473060939827362e-06,
      "loss": 1.6695,
      "step": 56918
    },
    {
      "epoch": 4.897100576443259,
      "grad_norm": 1.2578125,
      "learning_rate": 1.045557549928966e-06,
      "loss": 0.7118,
      "step": 56919
    },
    {
      "epoch": 4.8971866127505805,
      "grad_norm": 1.1171875,
      "learning_rate": 1.043810465219952e-06,
      "loss": 0.7703,
      "step": 56920
    },
    {
      "epoch": 4.897272649057903,
      "grad_norm": 1.0859375,
      "learning_rate": 1.0420648398607458e-06,
      "loss": 0.8671,
      "step": 56921
    },
    {
      "epoch": 4.897358685365224,
      "grad_norm": 0.97265625,
      "learning_rate": 1.0403206738564542e-06,
      "loss": 1.3944,
      "step": 56922
    },
    {
      "epoch": 4.897444721672546,
      "grad_norm": 1.09375,
      "learning_rate": 1.0385779672121842e-06,
      "loss": 0.9199,
      "step": 56923
    },
    {
      "epoch": 4.897530757979867,
      "grad_norm": 1.3984375,
      "learning_rate": 1.0368367199329876e-06,
      "loss": 0.9519,
      "step": 56924
    },
    {
      "epoch": 4.897616794287189,
      "grad_norm": 1.3046875,
      "learning_rate": 1.0350969320240267e-06,
      "loss": 1.8273,
      "step": 56925
    },
    {
      "epoch": 4.897702830594511,
      "grad_norm": 7.09375,
      "learning_rate": 1.0333586034904086e-06,
      "loss": 1.3913,
      "step": 56926
    },
    {
      "epoch": 4.897788866901832,
      "grad_norm": 1.1171875,
      "learning_rate": 1.0316217343371293e-06,
      "loss": 1.0404,
      "step": 56927
    },
    {
      "epoch": 4.897874903209154,
      "grad_norm": 1.1953125,
      "learning_rate": 1.0298863245692957e-06,
      "loss": 1.2494,
      "step": 56928
    },
    {
      "epoch": 4.897960939516476,
      "grad_norm": 1.1953125,
      "learning_rate": 1.0281523741920706e-06,
      "loss": 0.5747,
      "step": 56929
    },
    {
      "epoch": 4.898046975823798,
      "grad_norm": 1.8125,
      "learning_rate": 1.0264198832103944e-06,
      "loss": 1.1869,
      "step": 56930
    },
    {
      "epoch": 4.898133012131119,
      "grad_norm": 0.73828125,
      "learning_rate": 1.024688851629374e-06,
      "loss": 0.6751,
      "step": 56931
    },
    {
      "epoch": 4.898219048438441,
      "grad_norm": 0.94140625,
      "learning_rate": 1.022959279454172e-06,
      "loss": 1.3295,
      "step": 56932
    },
    {
      "epoch": 4.898305084745763,
      "grad_norm": 1.3203125,
      "learning_rate": 1.021231166689729e-06,
      "loss": 1.6222,
      "step": 56933
    },
    {
      "epoch": 4.898391121053084,
      "grad_norm": 1.515625,
      "learning_rate": 1.0195045133410963e-06,
      "loss": 1.4543,
      "step": 56934
    },
    {
      "epoch": 4.898477157360406,
      "grad_norm": 1.0078125,
      "learning_rate": 1.0177793194134366e-06,
      "loss": 1.0971,
      "step": 56935
    },
    {
      "epoch": 4.898563193667727,
      "grad_norm": 5.28125,
      "learning_rate": 1.0160555849116903e-06,
      "loss": 0.6905,
      "step": 56936
    },
    {
      "epoch": 4.8986492299750495,
      "grad_norm": 1.3828125,
      "learning_rate": 1.0143333098409646e-06,
      "loss": 1.3293,
      "step": 56937
    },
    {
      "epoch": 4.8987352662823715,
      "grad_norm": 0.9375,
      "learning_rate": 1.0126124942061999e-06,
      "loss": 0.9022,
      "step": 56938
    },
    {
      "epoch": 4.898821302589693,
      "grad_norm": 1.578125,
      "learning_rate": 1.0108931380125586e-06,
      "loss": 1.0587,
      "step": 56939
    },
    {
      "epoch": 4.898907338897015,
      "grad_norm": 0.890625,
      "learning_rate": 1.0091752412649812e-06,
      "loss": 0.9229,
      "step": 56940
    },
    {
      "epoch": 4.898993375204336,
      "grad_norm": 2.578125,
      "learning_rate": 1.0074588039685195e-06,
      "loss": 0.7922,
      "step": 56941
    },
    {
      "epoch": 4.899079411511658,
      "grad_norm": 1.1796875,
      "learning_rate": 1.0057438261281693e-06,
      "loss": 1.2618,
      "step": 56942
    },
    {
      "epoch": 4.899165447818979,
      "grad_norm": 1.6171875,
      "learning_rate": 1.0040303077490376e-06,
      "loss": 1.1946,
      "step": 56943
    },
    {
      "epoch": 4.899251484126301,
      "grad_norm": 1.2890625,
      "learning_rate": 1.0023182488360095e-06,
      "loss": 1.1393,
      "step": 56944
    },
    {
      "epoch": 4.899337520433623,
      "grad_norm": 0.97265625,
      "learning_rate": 1.0006076493941362e-06,
      "loss": 0.9588,
      "step": 56945
    },
    {
      "epoch": 4.8994235567409445,
      "grad_norm": 2.234375,
      "learning_rate": 9.988985094284698e-07,
      "loss": 0.6489,
      "step": 56946
    },
    {
      "epoch": 4.899509593048267,
      "grad_norm": 0.7890625,
      "learning_rate": 9.971908289440057e-07,
      "loss": 1.1115,
      "step": 56947
    },
    {
      "epoch": 4.899595629355588,
      "grad_norm": 1.2265625,
      "learning_rate": 9.954846079456848e-07,
      "loss": 0.96,
      "step": 56948
    },
    {
      "epoch": 4.89968166566291,
      "grad_norm": 1.0546875,
      "learning_rate": 9.937798464385029e-07,
      "loss": 1.2106,
      "step": 56949
    },
    {
      "epoch": 4.899767701970232,
      "grad_norm": 1.578125,
      "learning_rate": 9.92076544427456e-07,
      "loss": 1.3279,
      "step": 56950
    },
    {
      "epoch": 4.899853738277553,
      "grad_norm": 1.046875,
      "learning_rate": 9.903747019175403e-07,
      "loss": 0.5244,
      "step": 56951
    },
    {
      "epoch": 4.899939774584875,
      "grad_norm": 3.015625,
      "learning_rate": 9.886743189137516e-07,
      "loss": 0.7816,
      "step": 56952
    },
    {
      "epoch": 4.900025810892196,
      "grad_norm": 1.3203125,
      "learning_rate": 9.869753954210303e-07,
      "loss": 1.1708,
      "step": 56953
    },
    {
      "epoch": 4.900111847199518,
      "grad_norm": 1.078125,
      "learning_rate": 9.852779314443173e-07,
      "loss": 0.8854,
      "step": 56954
    },
    {
      "epoch": 4.90019788350684,
      "grad_norm": 1.171875,
      "learning_rate": 9.835819269886082e-07,
      "loss": 1.0982,
      "step": 56955
    },
    {
      "epoch": 4.900283919814162,
      "grad_norm": 1.40625,
      "learning_rate": 9.818873820589546e-07,
      "loss": 0.985,
      "step": 56956
    },
    {
      "epoch": 4.900369956121484,
      "grad_norm": 1.4375,
      "learning_rate": 9.801942966601307e-07,
      "loss": 1.2373,
      "step": 56957
    },
    {
      "epoch": 4.900455992428805,
      "grad_norm": 1.2578125,
      "learning_rate": 9.785026707972433e-07,
      "loss": 0.8272,
      "step": 56958
    },
    {
      "epoch": 4.900542028736127,
      "grad_norm": 0.828125,
      "learning_rate": 9.768125044751774e-07,
      "loss": 0.5305,
      "step": 56959
    },
    {
      "epoch": 4.900628065043448,
      "grad_norm": 0.9375,
      "learning_rate": 9.75123797698818e-07,
      "loss": 0.7148,
      "step": 56960
    },
    {
      "epoch": 4.90071410135077,
      "grad_norm": 1.1640625,
      "learning_rate": 9.734365504732169e-07,
      "loss": 1.1052,
      "step": 56961
    },
    {
      "epoch": 4.900800137658091,
      "grad_norm": 1.046875,
      "learning_rate": 9.717507628032586e-07,
      "loss": 0.6962,
      "step": 56962
    },
    {
      "epoch": 4.9008861739654135,
      "grad_norm": 1.609375,
      "learning_rate": 9.700664346938837e-07,
      "loss": 1.728,
      "step": 56963
    },
    {
      "epoch": 4.9009722102727356,
      "grad_norm": 1.0234375,
      "learning_rate": 9.683835661500328e-07,
      "loss": 1.1488,
      "step": 56964
    },
    {
      "epoch": 4.901058246580057,
      "grad_norm": 1.015625,
      "learning_rate": 9.667021571765355e-07,
      "loss": 1.2752,
      "step": 56965
    },
    {
      "epoch": 4.901144282887379,
      "grad_norm": 1.234375,
      "learning_rate": 9.650222077783878e-07,
      "loss": 1.354,
      "step": 56966
    },
    {
      "epoch": 4.9012303191947,
      "grad_norm": 0.8828125,
      "learning_rate": 9.633437179605297e-07,
      "loss": 0.6633,
      "step": 56967
    },
    {
      "epoch": 4.901316355502022,
      "grad_norm": 2.90625,
      "learning_rate": 9.616666877278467e-07,
      "loss": 1.7301,
      "step": 56968
    },
    {
      "epoch": 4.901402391809343,
      "grad_norm": 1.109375,
      "learning_rate": 9.599911170851683e-07,
      "loss": 0.5856,
      "step": 56969
    },
    {
      "epoch": 4.901488428116665,
      "grad_norm": 0.79296875,
      "learning_rate": 9.583170060375456e-07,
      "loss": 0.5464,
      "step": 56970
    },
    {
      "epoch": 4.901574464423987,
      "grad_norm": 7.46875,
      "learning_rate": 9.566443545896974e-07,
      "loss": 0.932,
      "step": 56971
    },
    {
      "epoch": 4.9016605007313085,
      "grad_norm": 1.2265625,
      "learning_rate": 9.54973162746675e-07,
      "loss": 0.8048,
      "step": 56972
    },
    {
      "epoch": 4.901746537038631,
      "grad_norm": 2.84375,
      "learning_rate": 9.533034305132526e-07,
      "loss": 1.5715,
      "step": 56973
    },
    {
      "epoch": 4.901832573345952,
      "grad_norm": 1.125,
      "learning_rate": 9.516351578943705e-07,
      "loss": 0.827,
      "step": 56974
    },
    {
      "epoch": 4.901918609653274,
      "grad_norm": 53.0,
      "learning_rate": 9.499683448949137e-07,
      "loss": 0.4938,
      "step": 56975
    },
    {
      "epoch": 4.902004645960595,
      "grad_norm": 1.171875,
      "learning_rate": 9.483029915197117e-07,
      "loss": 1.9216,
      "step": 56976
    },
    {
      "epoch": 4.902090682267917,
      "grad_norm": 0.9296875,
      "learning_rate": 9.466390977736494e-07,
      "loss": 1.2567,
      "step": 56977
    },
    {
      "epoch": 4.902176718575239,
      "grad_norm": 1.09375,
      "learning_rate": 9.449766636616674e-07,
      "loss": 0.8671,
      "step": 56978
    },
    {
      "epoch": 4.90226275488256,
      "grad_norm": 1.03125,
      "learning_rate": 9.433156891885397e-07,
      "loss": 0.8958,
      "step": 56979
    },
    {
      "epoch": 4.902348791189882,
      "grad_norm": 1.0625,
      "learning_rate": 9.416561743590957e-07,
      "loss": 0.8415,
      "step": 56980
    },
    {
      "epoch": 4.902434827497204,
      "grad_norm": 11840.0,
      "learning_rate": 9.399981191783313e-07,
      "loss": 1.0869,
      "step": 56981
    },
    {
      "epoch": 4.902520863804526,
      "grad_norm": 1.171875,
      "learning_rate": 9.383415236509652e-07,
      "loss": 1.2045,
      "step": 56982
    },
    {
      "epoch": 4.902606900111847,
      "grad_norm": 1.2734375,
      "learning_rate": 9.366863877818821e-07,
      "loss": 1.6367,
      "step": 56983
    },
    {
      "epoch": 4.902692936419169,
      "grad_norm": 0.82421875,
      "learning_rate": 9.350327115759672e-07,
      "loss": 0.5469,
      "step": 56984
    },
    {
      "epoch": 4.902778972726491,
      "grad_norm": 1.1875,
      "learning_rate": 9.333804950379943e-07,
      "loss": 1.4523,
      "step": 56985
    },
    {
      "epoch": 4.902865009033812,
      "grad_norm": 1.171875,
      "learning_rate": 9.317297381727929e-07,
      "loss": 1.2862,
      "step": 56986
    },
    {
      "epoch": 4.902951045341134,
      "grad_norm": 0.9453125,
      "learning_rate": 9.300804409853036e-07,
      "loss": 0.4393,
      "step": 56987
    },
    {
      "epoch": 4.903037081648455,
      "grad_norm": 1.390625,
      "learning_rate": 9.284326034801893e-07,
      "loss": 1.1254,
      "step": 56988
    },
    {
      "epoch": 4.9031231179557775,
      "grad_norm": 0.82421875,
      "learning_rate": 9.267862256623349e-07,
      "loss": 0.6456,
      "step": 56989
    },
    {
      "epoch": 4.903209154263099,
      "grad_norm": 0.93359375,
      "learning_rate": 9.251413075366255e-07,
      "loss": 0.7114,
      "step": 56990
    },
    {
      "epoch": 4.903295190570421,
      "grad_norm": 1.125,
      "learning_rate": 9.234978491077795e-07,
      "loss": 1.0677,
      "step": 56991
    },
    {
      "epoch": 4.903381226877743,
      "grad_norm": 1.34375,
      "learning_rate": 9.218558503806263e-07,
      "loss": 1.1452,
      "step": 56992
    },
    {
      "epoch": 4.903467263185064,
      "grad_norm": 0.92578125,
      "learning_rate": 9.202153113599954e-07,
      "loss": 0.6633,
      "step": 56993
    },
    {
      "epoch": 4.903553299492386,
      "grad_norm": 1.125,
      "learning_rate": 9.185762320506608e-07,
      "loss": 1.1131,
      "step": 56994
    },
    {
      "epoch": 4.903639335799707,
      "grad_norm": 1.484375,
      "learning_rate": 9.169386124574519e-07,
      "loss": 1.3597,
      "step": 56995
    },
    {
      "epoch": 4.903725372107029,
      "grad_norm": 1.125,
      "learning_rate": 9.153024525850872e-07,
      "loss": 0.6892,
      "step": 56996
    },
    {
      "epoch": 4.9038114084143505,
      "grad_norm": 0.87890625,
      "learning_rate": 9.136677524384518e-07,
      "loss": 0.6346,
      "step": 56997
    },
    {
      "epoch": 4.9038974447216725,
      "grad_norm": 1.140625,
      "learning_rate": 9.12034512022264e-07,
      "loss": 1.4365,
      "step": 56998
    },
    {
      "epoch": 4.903983481028995,
      "grad_norm": 1.1015625,
      "learning_rate": 9.104027313412422e-07,
      "loss": 1.5977,
      "step": 56999
    },
    {
      "epoch": 4.904069517336316,
      "grad_norm": 1.03125,
      "learning_rate": 9.087724104002715e-07,
      "loss": 0.9931,
      "step": 57000
    },
    {
      "epoch": 4.904155553643638,
      "grad_norm": 0.9140625,
      "learning_rate": 9.071435492040703e-07,
      "loss": 1.1122,
      "step": 57001
    },
    {
      "epoch": 4.904241589950959,
      "grad_norm": 0.8828125,
      "learning_rate": 9.055161477574125e-07,
      "loss": 1.0454,
      "step": 57002
    },
    {
      "epoch": 4.904327626258281,
      "grad_norm": 1.4765625,
      "learning_rate": 9.038902060650722e-07,
      "loss": 0.7926,
      "step": 57003
    },
    {
      "epoch": 4.904413662565602,
      "grad_norm": 0.7421875,
      "learning_rate": 9.022657241317678e-07,
      "loss": 0.7783,
      "step": 57004
    },
    {
      "epoch": 4.904499698872924,
      "grad_norm": 1.0390625,
      "learning_rate": 9.006427019622177e-07,
      "loss": 0.8279,
      "step": 57005
    },
    {
      "epoch": 4.904585735180246,
      "grad_norm": 1.28125,
      "learning_rate": 8.990211395612514e-07,
      "loss": 1.4356,
      "step": 57006
    },
    {
      "epoch": 4.904671771487568,
      "grad_norm": 1.4140625,
      "learning_rate": 8.974010369335872e-07,
      "loss": 0.9881,
      "step": 57007
    },
    {
      "epoch": 4.90475780779489,
      "grad_norm": 1.1796875,
      "learning_rate": 8.957823940839994e-07,
      "loss": 1.1144,
      "step": 57008
    },
    {
      "epoch": 4.904843844102211,
      "grad_norm": 1.4140625,
      "learning_rate": 8.94165211017095e-07,
      "loss": 1.0903,
      "step": 57009
    },
    {
      "epoch": 4.904929880409533,
      "grad_norm": 0.953125,
      "learning_rate": 8.925494877377038e-07,
      "loss": 1.1078,
      "step": 57010
    },
    {
      "epoch": 4.905015916716854,
      "grad_norm": 1.28125,
      "learning_rate": 8.909352242505997e-07,
      "loss": 2.0461,
      "step": 57011
    },
    {
      "epoch": 4.905101953024176,
      "grad_norm": 1.4375,
      "learning_rate": 8.8932242056039e-07,
      "loss": 1.1514,
      "step": 57012
    },
    {
      "epoch": 4.905187989331498,
      "grad_norm": 1.03125,
      "learning_rate": 8.877110766717933e-07,
      "loss": 1.4168,
      "step": 57013
    },
    {
      "epoch": 4.905274025638819,
      "grad_norm": 0.76171875,
      "learning_rate": 8.86101192589639e-07,
      "loss": 0.898,
      "step": 57014
    },
    {
      "epoch": 4.9053600619461415,
      "grad_norm": 0.94921875,
      "learning_rate": 8.844927683185345e-07,
      "loss": 0.7165,
      "step": 57015
    },
    {
      "epoch": 4.905446098253463,
      "grad_norm": 0.86328125,
      "learning_rate": 8.828858038632536e-07,
      "loss": 0.9444,
      "step": 57016
    },
    {
      "epoch": 4.905532134560785,
      "grad_norm": 864.0,
      "learning_rate": 8.812802992284042e-07,
      "loss": 0.9624,
      "step": 57017
    },
    {
      "epoch": 4.905618170868106,
      "grad_norm": 2.390625,
      "learning_rate": 8.796762544187597e-07,
      "loss": 1.0019,
      "step": 57018
    },
    {
      "epoch": 4.905704207175428,
      "grad_norm": 1.21875,
      "learning_rate": 8.78073669439039e-07,
      "loss": 1.5864,
      "step": 57019
    },
    {
      "epoch": 4.90579024348275,
      "grad_norm": 0.8359375,
      "learning_rate": 8.764725442937937e-07,
      "loss": 0.8053,
      "step": 57020
    },
    {
      "epoch": 4.905876279790071,
      "grad_norm": 3.453125,
      "learning_rate": 8.748728789878535e-07,
      "loss": 0.8086,
      "step": 57021
    },
    {
      "epoch": 4.905962316097393,
      "grad_norm": 1.0078125,
      "learning_rate": 8.732746735258257e-07,
      "loss": 0.925,
      "step": 57022
    },
    {
      "epoch": 4.9060483524047145,
      "grad_norm": 193.0,
      "learning_rate": 8.716779279123732e-07,
      "loss": 1.4129,
      "step": 57023
    },
    {
      "epoch": 4.9061343887120366,
      "grad_norm": 1.1640625,
      "learning_rate": 8.700826421522145e-07,
      "loss": 0.8965,
      "step": 57024
    },
    {
      "epoch": 4.906220425019358,
      "grad_norm": 0.8984375,
      "learning_rate": 8.684888162499571e-07,
      "loss": 0.8458,
      "step": 57025
    },
    {
      "epoch": 4.90630646132668,
      "grad_norm": 1.296875,
      "learning_rate": 8.668964502103194e-07,
      "loss": 1.0971,
      "step": 57026
    },
    {
      "epoch": 4.906392497634002,
      "grad_norm": 1.2421875,
      "learning_rate": 8.653055440379087e-07,
      "loss": 1.6768,
      "step": 57027
    },
    {
      "epoch": 4.906478533941323,
      "grad_norm": 1.171875,
      "learning_rate": 8.637160977374437e-07,
      "loss": 1.159,
      "step": 57028
    },
    {
      "epoch": 4.906564570248645,
      "grad_norm": 5.65625,
      "learning_rate": 8.62128111313476e-07,
      "loss": 1.0412,
      "step": 57029
    },
    {
      "epoch": 4.906650606555966,
      "grad_norm": 1.0546875,
      "learning_rate": 8.605415847707243e-07,
      "loss": 1.4059,
      "step": 57030
    },
    {
      "epoch": 4.906736642863288,
      "grad_norm": 1.265625,
      "learning_rate": 8.589565181138515e-07,
      "loss": 0.9085,
      "step": 57031
    },
    {
      "epoch": 4.9068226791706095,
      "grad_norm": 0.91015625,
      "learning_rate": 8.573729113474093e-07,
      "loss": 0.5641,
      "step": 57032
    },
    {
      "epoch": 4.906908715477932,
      "grad_norm": 1.0859375,
      "learning_rate": 8.557907644761165e-07,
      "loss": 1.5479,
      "step": 57033
    },
    {
      "epoch": 4.906994751785254,
      "grad_norm": 0.921875,
      "learning_rate": 8.542100775045247e-07,
      "loss": 1.0176,
      "step": 57034
    },
    {
      "epoch": 4.907080788092575,
      "grad_norm": 1.3515625,
      "learning_rate": 8.526308504372971e-07,
      "loss": 0.949,
      "step": 57035
    },
    {
      "epoch": 4.907166824399897,
      "grad_norm": 1.4609375,
      "learning_rate": 8.510530832790409e-07,
      "loss": 1.5224,
      "step": 57036
    },
    {
      "epoch": 4.907252860707218,
      "grad_norm": 266.0,
      "learning_rate": 8.494767760343636e-07,
      "loss": 1.1353,
      "step": 57037
    },
    {
      "epoch": 4.90733889701454,
      "grad_norm": 1.5859375,
      "learning_rate": 8.479019287079282e-07,
      "loss": 1.5538,
      "step": 57038
    },
    {
      "epoch": 4.907424933321861,
      "grad_norm": 0.9296875,
      "learning_rate": 8.46328541304231e-07,
      "loss": 1.0329,
      "step": 57039
    },
    {
      "epoch": 4.907510969629183,
      "grad_norm": 1.3515625,
      "learning_rate": 8.447566138279905e-07,
      "loss": 1.1168,
      "step": 57040
    },
    {
      "epoch": 4.9075970059365055,
      "grad_norm": 1.109375,
      "learning_rate": 8.431861462837586e-07,
      "loss": 0.9407,
      "step": 57041
    },
    {
      "epoch": 4.907683042243827,
      "grad_norm": 1.3828125,
      "learning_rate": 8.416171386761429e-07,
      "loss": 1.2647,
      "step": 57042
    },
    {
      "epoch": 4.907769078551149,
      "grad_norm": 0.95703125,
      "learning_rate": 8.40049591009695e-07,
      "loss": 0.7604,
      "step": 57043
    },
    {
      "epoch": 4.90785511485847,
      "grad_norm": 1.3828125,
      "learning_rate": 8.384835032890225e-07,
      "loss": 1.5645,
      "step": 57044
    },
    {
      "epoch": 4.907941151165792,
      "grad_norm": 1.1328125,
      "learning_rate": 8.369188755186774e-07,
      "loss": 1.2981,
      "step": 57045
    },
    {
      "epoch": 4.908027187473113,
      "grad_norm": 1.9453125,
      "learning_rate": 8.353557077033225e-07,
      "loss": 0.839,
      "step": 57046
    },
    {
      "epoch": 4.908113223780435,
      "grad_norm": 1.140625,
      "learning_rate": 8.337939998473987e-07,
      "loss": 0.8023,
      "step": 57047
    },
    {
      "epoch": 4.908199260087757,
      "grad_norm": 1.40625,
      "learning_rate": 8.32233751955569e-07,
      "loss": 1.4709,
      "step": 57048
    },
    {
      "epoch": 4.9082852963950785,
      "grad_norm": 1.359375,
      "learning_rate": 8.306749640324407e-07,
      "loss": 1.5511,
      "step": 57049
    },
    {
      "epoch": 4.908371332702401,
      "grad_norm": 1.640625,
      "learning_rate": 8.291176360823993e-07,
      "loss": 1.0351,
      "step": 57050
    },
    {
      "epoch": 4.908457369009722,
      "grad_norm": 0.75,
      "learning_rate": 8.275617681101632e-07,
      "loss": 1.3312,
      "step": 57051
    },
    {
      "epoch": 4.908543405317044,
      "grad_norm": 1.546875,
      "learning_rate": 8.260073601202289e-07,
      "loss": 0.9184,
      "step": 57052
    },
    {
      "epoch": 4.908629441624365,
      "grad_norm": 13.125,
      "learning_rate": 8.244544121171483e-07,
      "loss": 1.141,
      "step": 57053
    },
    {
      "epoch": 4.908715477931687,
      "grad_norm": 81.5,
      "learning_rate": 8.229029241054176e-07,
      "loss": 0.9663,
      "step": 57054
    },
    {
      "epoch": 4.908801514239009,
      "grad_norm": 1.25,
      "learning_rate": 8.213528960896444e-07,
      "loss": 1.7473,
      "step": 57055
    },
    {
      "epoch": 4.90888755054633,
      "grad_norm": 1.015625,
      "learning_rate": 8.198043280742695e-07,
      "loss": 0.9382,
      "step": 57056
    },
    {
      "epoch": 4.908973586853652,
      "grad_norm": 0.70703125,
      "learning_rate": 8.18257220063956e-07,
      "loss": 0.4186,
      "step": 57057
    },
    {
      "epoch": 4.9090596231609736,
      "grad_norm": 1.1875,
      "learning_rate": 8.167115720630891e-07,
      "loss": 0.7754,
      "step": 57058
    },
    {
      "epoch": 4.909145659468296,
      "grad_norm": 1.3515625,
      "learning_rate": 8.151673840763319e-07,
      "loss": 1.6504,
      "step": 57059
    },
    {
      "epoch": 4.909231695775618,
      "grad_norm": 0.87890625,
      "learning_rate": 8.136246561080695e-07,
      "loss": 0.6283,
      "step": 57060
    },
    {
      "epoch": 4.909317732082939,
      "grad_norm": 9.125,
      "learning_rate": 8.120833881628542e-07,
      "loss": 1.6955,
      "step": 57061
    },
    {
      "epoch": 4.909403768390261,
      "grad_norm": 1.2578125,
      "learning_rate": 8.105435802452377e-07,
      "loss": 0.6936,
      "step": 57062
    },
    {
      "epoch": 4.909489804697582,
      "grad_norm": 1.296875,
      "learning_rate": 8.090052323597164e-07,
      "loss": 1.2179,
      "step": 57063
    },
    {
      "epoch": 4.909575841004904,
      "grad_norm": 1.4140625,
      "learning_rate": 8.074683445107312e-07,
      "loss": 1.2489,
      "step": 57064
    },
    {
      "epoch": 4.909661877312226,
      "grad_norm": 0.83203125,
      "learning_rate": 8.05932916702834e-07,
      "loss": 0.5565,
      "step": 57065
    },
    {
      "epoch": 4.909747913619547,
      "grad_norm": 0.71875,
      "learning_rate": 8.043989489404657e-07,
      "loss": 0.8555,
      "step": 57066
    },
    {
      "epoch": 4.9098339499268695,
      "grad_norm": 1.0703125,
      "learning_rate": 8.028664412281783e-07,
      "loss": 1.7308,
      "step": 57067
    },
    {
      "epoch": 4.909919986234191,
      "grad_norm": 1.1640625,
      "learning_rate": 8.013353935704127e-07,
      "loss": 0.6655,
      "step": 57068
    },
    {
      "epoch": 4.910006022541513,
      "grad_norm": 0.8359375,
      "learning_rate": 7.998058059716096e-07,
      "loss": 0.7016,
      "step": 57069
    },
    {
      "epoch": 4.910092058848834,
      "grad_norm": 1.125,
      "learning_rate": 7.98277678436321e-07,
      "loss": 1.051,
      "step": 57070
    },
    {
      "epoch": 4.910178095156156,
      "grad_norm": 0.92578125,
      "learning_rate": 7.967510109689879e-07,
      "loss": 0.7726,
      "step": 57071
    },
    {
      "epoch": 4.910264131463478,
      "grad_norm": 1.390625,
      "learning_rate": 7.952258035740512e-07,
      "loss": 1.5201,
      "step": 57072
    },
    {
      "epoch": 4.910350167770799,
      "grad_norm": 1.9140625,
      "learning_rate": 7.937020562560071e-07,
      "loss": 1.0064,
      "step": 57073
    },
    {
      "epoch": 4.910436204078121,
      "grad_norm": 1.1640625,
      "learning_rate": 7.921797690192412e-07,
      "loss": 0.852,
      "step": 57074
    },
    {
      "epoch": 4.9105222403854425,
      "grad_norm": 1.21875,
      "learning_rate": 7.906589418683053e-07,
      "loss": 1.2849,
      "step": 57075
    },
    {
      "epoch": 4.910608276692765,
      "grad_norm": 1.2734375,
      "learning_rate": 7.891395748076402e-07,
      "loss": 1.2689,
      "step": 57076
    },
    {
      "epoch": 4.910694313000086,
      "grad_norm": 20.625,
      "learning_rate": 7.87621667841576e-07,
      "loss": 1.373,
      "step": 57077
    },
    {
      "epoch": 4.910780349307408,
      "grad_norm": 2.296875,
      "learning_rate": 7.861052209746644e-07,
      "loss": 1.0805,
      "step": 57078
    },
    {
      "epoch": 4.91086638561473,
      "grad_norm": 1.3359375,
      "learning_rate": 7.845902342112909e-07,
      "loss": 1.5953,
      "step": 57079
    },
    {
      "epoch": 4.910952421922051,
      "grad_norm": 1.90625,
      "learning_rate": 7.830767075558965e-07,
      "loss": 0.9912,
      "step": 57080
    },
    {
      "epoch": 4.911038458229373,
      "grad_norm": 1.078125,
      "learning_rate": 7.815646410129218e-07,
      "loss": 1.1991,
      "step": 57081
    },
    {
      "epoch": 4.911124494536694,
      "grad_norm": 0.9453125,
      "learning_rate": 7.800540345868079e-07,
      "loss": 0.6057,
      "step": 57082
    },
    {
      "epoch": 4.911210530844016,
      "grad_norm": 31.75,
      "learning_rate": 7.785448882818847e-07,
      "loss": 0.7918,
      "step": 57083
    },
    {
      "epoch": 4.9112965671513376,
      "grad_norm": 3.859375,
      "learning_rate": 7.770372021026484e-07,
      "loss": 0.9934,
      "step": 57084
    },
    {
      "epoch": 4.91138260345866,
      "grad_norm": 0.953125,
      "learning_rate": 7.755309760534846e-07,
      "loss": 0.6647,
      "step": 57085
    },
    {
      "epoch": 4.911468639765982,
      "grad_norm": 1.5078125,
      "learning_rate": 7.740262101387785e-07,
      "loss": 1.1617,
      "step": 57086
    },
    {
      "epoch": 4.911554676073303,
      "grad_norm": 0.9609375,
      "learning_rate": 7.725229043630267e-07,
      "loss": 0.8041,
      "step": 57087
    },
    {
      "epoch": 4.911640712380625,
      "grad_norm": 11.5625,
      "learning_rate": 7.710210587304478e-07,
      "loss": 1.0041,
      "step": 57088
    },
    {
      "epoch": 4.911726748687946,
      "grad_norm": 1.2109375,
      "learning_rate": 7.69520673245594e-07,
      "loss": 1.9891,
      "step": 57089
    },
    {
      "epoch": 4.911812784995268,
      "grad_norm": 1.1171875,
      "learning_rate": 7.680217479127949e-07,
      "loss": 1.1952,
      "step": 57090
    },
    {
      "epoch": 4.911898821302589,
      "grad_norm": 1.046875,
      "learning_rate": 7.66524282736436e-07,
      "loss": 0.8962,
      "step": 57091
    },
    {
      "epoch": 4.911984857609911,
      "grad_norm": 1.2421875,
      "learning_rate": 7.650282777209028e-07,
      "loss": 1.7354,
      "step": 57092
    },
    {
      "epoch": 4.9120708939172335,
      "grad_norm": 1.671875,
      "learning_rate": 7.635337328705805e-07,
      "loss": 1.7091,
      "step": 57093
    },
    {
      "epoch": 4.912156930224555,
      "grad_norm": 0.8984375,
      "learning_rate": 7.62040648189799e-07,
      "loss": 0.4867,
      "step": 57094
    },
    {
      "epoch": 4.912242966531877,
      "grad_norm": 0.93359375,
      "learning_rate": 7.605490236829437e-07,
      "loss": 1.5706,
      "step": 57095
    },
    {
      "epoch": 4.912329002839198,
      "grad_norm": 11.0625,
      "learning_rate": 7.590588593544001e-07,
      "loss": 1.1237,
      "step": 57096
    },
    {
      "epoch": 4.91241503914652,
      "grad_norm": 0.921875,
      "learning_rate": 7.575701552084979e-07,
      "loss": 1.3103,
      "step": 57097
    },
    {
      "epoch": 4.912501075453841,
      "grad_norm": 1.2734375,
      "learning_rate": 7.560829112496226e-07,
      "loss": 1.7488,
      "step": 57098
    },
    {
      "epoch": 4.912587111761163,
      "grad_norm": 1.2734375,
      "learning_rate": 7.545971274821039e-07,
      "loss": 1.5054,
      "step": 57099
    },
    {
      "epoch": 4.912673148068485,
      "grad_norm": 7.53125,
      "learning_rate": 7.531128039102719e-07,
      "loss": 0.8626,
      "step": 57100
    },
    {
      "epoch": 4.9127591843758065,
      "grad_norm": 1.6953125,
      "learning_rate": 7.516299405385119e-07,
      "loss": 1.3082,
      "step": 57101
    },
    {
      "epoch": 4.912845220683129,
      "grad_norm": 0.8828125,
      "learning_rate": 7.501485373711537e-07,
      "loss": 0.8464,
      "step": 57102
    },
    {
      "epoch": 4.91293125699045,
      "grad_norm": 1.0703125,
      "learning_rate": 7.486685944124716e-07,
      "loss": 1.4083,
      "step": 57103
    },
    {
      "epoch": 4.913017293297772,
      "grad_norm": 15.75,
      "learning_rate": 7.471901116668512e-07,
      "loss": 1.0911,
      "step": 57104
    },
    {
      "epoch": 4.913103329605093,
      "grad_norm": 1.3125,
      "learning_rate": 7.457130891385666e-07,
      "loss": 1.6512,
      "step": 57105
    },
    {
      "epoch": 4.913189365912415,
      "grad_norm": 1.2734375,
      "learning_rate": 7.44237526832059e-07,
      "loss": 1.5529,
      "step": 57106
    },
    {
      "epoch": 4.913275402219737,
      "grad_norm": 4.34375,
      "learning_rate": 7.427634247514914e-07,
      "loss": 1.4017,
      "step": 57107
    },
    {
      "epoch": 4.913361438527058,
      "grad_norm": 1.015625,
      "learning_rate": 7.412907829012494e-07,
      "loss": 1.2113,
      "step": 57108
    },
    {
      "epoch": 4.91344747483438,
      "grad_norm": 0.87890625,
      "learning_rate": 7.398196012856073e-07,
      "loss": 0.6045,
      "step": 57109
    },
    {
      "epoch": 4.913533511141702,
      "grad_norm": 1.3515625,
      "learning_rate": 7.383498799089505e-07,
      "loss": 1.044,
      "step": 57110
    },
    {
      "epoch": 4.913619547449024,
      "grad_norm": 0.97265625,
      "learning_rate": 7.368816187754424e-07,
      "loss": 1.2175,
      "step": 57111
    },
    {
      "epoch": 4.913705583756345,
      "grad_norm": 8.5625,
      "learning_rate": 7.354148178895237e-07,
      "loss": 1.2574,
      "step": 57112
    },
    {
      "epoch": 4.913791620063667,
      "grad_norm": 1.3046875,
      "learning_rate": 7.339494772554133e-07,
      "loss": 0.8797,
      "step": 57113
    },
    {
      "epoch": 4.913877656370989,
      "grad_norm": 1.15625,
      "learning_rate": 7.324855968773303e-07,
      "loss": 1.4063,
      "step": 57114
    },
    {
      "epoch": 4.91396369267831,
      "grad_norm": 1.1953125,
      "learning_rate": 7.310231767597153e-07,
      "loss": 0.8059,
      "step": 57115
    },
    {
      "epoch": 4.914049728985632,
      "grad_norm": 1.6875,
      "learning_rate": 7.295622169066763e-07,
      "loss": 1.3522,
      "step": 57116
    },
    {
      "epoch": 4.914135765292953,
      "grad_norm": 1.0,
      "learning_rate": 7.281027173225985e-07,
      "loss": 0.6458,
      "step": 57117
    },
    {
      "epoch": 4.914221801600275,
      "grad_norm": 1.453125,
      "learning_rate": 7.266446780117564e-07,
      "loss": 1.4314,
      "step": 57118
    },
    {
      "epoch": 4.914307837907597,
      "grad_norm": 1.390625,
      "learning_rate": 7.251880989783133e-07,
      "loss": 0.9801,
      "step": 57119
    },
    {
      "epoch": 4.914393874214919,
      "grad_norm": 1.2109375,
      "learning_rate": 7.23732980226599e-07,
      "loss": 0.7967,
      "step": 57120
    },
    {
      "epoch": 4.914479910522241,
      "grad_norm": 1.359375,
      "learning_rate": 7.222793217608881e-07,
      "loss": 1.8699,
      "step": 57121
    },
    {
      "epoch": 4.914565946829562,
      "grad_norm": 0.92578125,
      "learning_rate": 7.208271235853992e-07,
      "loss": 0.9556,
      "step": 57122
    },
    {
      "epoch": 4.914651983136884,
      "grad_norm": 1.59375,
      "learning_rate": 7.193763857044066e-07,
      "loss": 0.7035,
      "step": 57123
    },
    {
      "epoch": 4.914738019444205,
      "grad_norm": 101.0,
      "learning_rate": 7.17927108122074e-07,
      "loss": 1.0303,
      "step": 57124
    },
    {
      "epoch": 4.914824055751527,
      "grad_norm": 10176.0,
      "learning_rate": 7.164792908427309e-07,
      "loss": 0.9467,
      "step": 57125
    },
    {
      "epoch": 4.914910092058848,
      "grad_norm": 0.9609375,
      "learning_rate": 7.150329338705963e-07,
      "loss": 1.2796,
      "step": 57126
    },
    {
      "epoch": 4.9149961283661705,
      "grad_norm": 1.109375,
      "learning_rate": 7.135880372098891e-07,
      "loss": 1.6252,
      "step": 57127
    },
    {
      "epoch": 4.915082164673493,
      "grad_norm": 1.625,
      "learning_rate": 7.12144600864828e-07,
      "loss": 1.9267,
      "step": 57128
    },
    {
      "epoch": 4.915168200980814,
      "grad_norm": 1.125,
      "learning_rate": 7.10702624839632e-07,
      "loss": 1.0478,
      "step": 57129
    },
    {
      "epoch": 4.915254237288136,
      "grad_norm": 31.375,
      "learning_rate": 7.092621091385199e-07,
      "loss": 0.7382,
      "step": 57130
    },
    {
      "epoch": 4.915340273595457,
      "grad_norm": 1.046875,
      "learning_rate": 7.078230537657105e-07,
      "loss": 1.103,
      "step": 57131
    },
    {
      "epoch": 4.915426309902779,
      "grad_norm": 1.0546875,
      "learning_rate": 7.06385458725367e-07,
      "loss": 1.1445,
      "step": 57132
    },
    {
      "epoch": 4.9155123462101,
      "grad_norm": 1.921875,
      "learning_rate": 7.049493240218197e-07,
      "loss": 1.182,
      "step": 57133
    },
    {
      "epoch": 4.915598382517422,
      "grad_norm": 0.9921875,
      "learning_rate": 7.035146496591205e-07,
      "loss": 0.6431,
      "step": 57134
    },
    {
      "epoch": 4.915684418824744,
      "grad_norm": 1.109375,
      "learning_rate": 7.020814356415994e-07,
      "loss": 1.0667,
      "step": 57135
    },
    {
      "epoch": 4.915770455132066,
      "grad_norm": 1.0625,
      "learning_rate": 7.006496819733089e-07,
      "loss": 1.1433,
      "step": 57136
    },
    {
      "epoch": 4.915856491439388,
      "grad_norm": 0.80859375,
      "learning_rate": 6.992193886585229e-07,
      "loss": 0.781,
      "step": 57137
    },
    {
      "epoch": 4.915942527746709,
      "grad_norm": 1.03125,
      "learning_rate": 6.977905557014608e-07,
      "loss": 1.2857,
      "step": 57138
    },
    {
      "epoch": 4.916028564054031,
      "grad_norm": 0.85546875,
      "learning_rate": 6.963631831062301e-07,
      "loss": 0.4969,
      "step": 57139
    },
    {
      "epoch": 4.916114600361352,
      "grad_norm": 0.8984375,
      "learning_rate": 6.949372708769941e-07,
      "loss": 0.4697,
      "step": 57140
    },
    {
      "epoch": 4.916200636668674,
      "grad_norm": 1.1171875,
      "learning_rate": 6.935128190180273e-07,
      "loss": 0.6078,
      "step": 57141
    },
    {
      "epoch": 4.916286672975996,
      "grad_norm": 0.6953125,
      "learning_rate": 6.92089827533382e-07,
      "loss": 0.3909,
      "step": 57142
    },
    {
      "epoch": 4.916372709283317,
      "grad_norm": 0.9921875,
      "learning_rate": 6.90668296427277e-07,
      "loss": 0.8629,
      "step": 57143
    },
    {
      "epoch": 4.9164587455906394,
      "grad_norm": 380.0,
      "learning_rate": 6.892482257038201e-07,
      "loss": 0.9428,
      "step": 57144
    },
    {
      "epoch": 4.916544781897961,
      "grad_norm": 1.2578125,
      "learning_rate": 6.878296153672304e-07,
      "loss": 1.7368,
      "step": 57145
    },
    {
      "epoch": 4.916630818205283,
      "grad_norm": 1.1484375,
      "learning_rate": 6.864124654216153e-07,
      "loss": 1.4622,
      "step": 57146
    },
    {
      "epoch": 4.916716854512604,
      "grad_norm": 1.8359375,
      "learning_rate": 6.849967758711384e-07,
      "loss": 1.8106,
      "step": 57147
    },
    {
      "epoch": 4.916802890819926,
      "grad_norm": 0.9609375,
      "learning_rate": 6.83582546719963e-07,
      "loss": 1.4181,
      "step": 57148
    },
    {
      "epoch": 4.916888927127248,
      "grad_norm": 1.15625,
      "learning_rate": 6.821697779721414e-07,
      "loss": 1.0315,
      "step": 57149
    },
    {
      "epoch": 4.916974963434569,
      "grad_norm": 0.96484375,
      "learning_rate": 6.80758469631948e-07,
      "loss": 0.8992,
      "step": 57150
    },
    {
      "epoch": 4.917060999741891,
      "grad_norm": 1.3203125,
      "learning_rate": 6.79348621703324e-07,
      "loss": 0.9247,
      "step": 57151
    },
    {
      "epoch": 4.917147036049212,
      "grad_norm": 1.2421875,
      "learning_rate": 6.779402341905438e-07,
      "loss": 1.0593,
      "step": 57152
    },
    {
      "epoch": 4.9172330723565345,
      "grad_norm": 2.109375,
      "learning_rate": 6.765333070976598e-07,
      "loss": 1.3107,
      "step": 57153
    },
    {
      "epoch": 4.917319108663856,
      "grad_norm": 1.1015625,
      "learning_rate": 6.751278404288352e-07,
      "loss": 1.1471,
      "step": 57154
    },
    {
      "epoch": 4.917405144971178,
      "grad_norm": 2.875,
      "learning_rate": 6.737238341881225e-07,
      "loss": 1.0674,
      "step": 57155
    },
    {
      "epoch": 4.9174911812785,
      "grad_norm": 1.4765625,
      "learning_rate": 6.723212883796848e-07,
      "loss": 1.0003,
      "step": 57156
    },
    {
      "epoch": 4.917577217585821,
      "grad_norm": 1.0703125,
      "learning_rate": 6.70920203007519e-07,
      "loss": 1.0752,
      "step": 57157
    },
    {
      "epoch": 4.917663253893143,
      "grad_norm": 1.3515625,
      "learning_rate": 6.695205780758995e-07,
      "loss": 1.8329,
      "step": 57158
    },
    {
      "epoch": 4.917749290200464,
      "grad_norm": 0.94921875,
      "learning_rate": 6.68122413588712e-07,
      "loss": 0.6102,
      "step": 57159
    },
    {
      "epoch": 4.917835326507786,
      "grad_norm": 1.015625,
      "learning_rate": 6.66725709550231e-07,
      "loss": 1.5207,
      "step": 57160
    },
    {
      "epoch": 4.9179213628151075,
      "grad_norm": 124.5,
      "learning_rate": 6.653304659644532e-07,
      "loss": 1.2774,
      "step": 57161
    },
    {
      "epoch": 4.91800739912243,
      "grad_norm": 1.0625,
      "learning_rate": 6.639366828354309e-07,
      "loss": 0.5287,
      "step": 57162
    },
    {
      "epoch": 4.918093435429752,
      "grad_norm": 0.796875,
      "learning_rate": 6.625443601672721e-07,
      "loss": 0.5446,
      "step": 57163
    },
    {
      "epoch": 4.918179471737073,
      "grad_norm": 0.90625,
      "learning_rate": 6.611534979640843e-07,
      "loss": 0.4426,
      "step": 57164
    },
    {
      "epoch": 4.918265508044395,
      "grad_norm": 1.265625,
      "learning_rate": 6.597640962299201e-07,
      "loss": 1.0303,
      "step": 57165
    },
    {
      "epoch": 4.918351544351716,
      "grad_norm": 0.88671875,
      "learning_rate": 6.583761549688316e-07,
      "loss": 0.7404,
      "step": 57166
    },
    {
      "epoch": 4.918437580659038,
      "grad_norm": 0.91796875,
      "learning_rate": 6.569896741848159e-07,
      "loss": 1.28,
      "step": 57167
    },
    {
      "epoch": 4.918523616966359,
      "grad_norm": 1.4140625,
      "learning_rate": 6.556046538820359e-07,
      "loss": 1.347,
      "step": 57168
    },
    {
      "epoch": 4.918609653273681,
      "grad_norm": 11072.0,
      "learning_rate": 6.542210940644888e-07,
      "loss": 1.3313,
      "step": 57169
    },
    {
      "epoch": 4.9186956895810034,
      "grad_norm": 75.5,
      "learning_rate": 6.528389947362268e-07,
      "loss": 0.3676,
      "step": 57170
    },
    {
      "epoch": 4.918781725888325,
      "grad_norm": 1.2109375,
      "learning_rate": 6.514583559013021e-07,
      "loss": 0.7242,
      "step": 57171
    },
    {
      "epoch": 4.918867762195647,
      "grad_norm": 1.40625,
      "learning_rate": 6.500791775637671e-07,
      "loss": 1.5438,
      "step": 57172
    },
    {
      "epoch": 4.918953798502968,
      "grad_norm": 2.046875,
      "learning_rate": 6.487014597275631e-07,
      "loss": 1.4066,
      "step": 57173
    },
    {
      "epoch": 4.91903983481029,
      "grad_norm": 1.0078125,
      "learning_rate": 6.473252023968534e-07,
      "loss": 0.5511,
      "step": 57174
    },
    {
      "epoch": 4.919125871117612,
      "grad_norm": 1.0625,
      "learning_rate": 6.459504055755793e-07,
      "loss": 0.8831,
      "step": 57175
    },
    {
      "epoch": 4.919211907424933,
      "grad_norm": 1.5390625,
      "learning_rate": 6.445770692677378e-07,
      "loss": 1.3814,
      "step": 57176
    },
    {
      "epoch": 4.919297943732255,
      "grad_norm": 1.4296875,
      "learning_rate": 6.432051934774363e-07,
      "loss": 0.675,
      "step": 57177
    },
    {
      "epoch": 4.919383980039576,
      "grad_norm": 0.79296875,
      "learning_rate": 6.418347782086165e-07,
      "loss": 0.7371,
      "step": 57178
    },
    {
      "epoch": 4.9194700163468985,
      "grad_norm": 0.8671875,
      "learning_rate": 6.404658234653304e-07,
      "loss": 0.8151,
      "step": 57179
    },
    {
      "epoch": 4.919556052654221,
      "grad_norm": 1.046875,
      "learning_rate": 6.390983292515751e-07,
      "loss": 0.9019,
      "step": 57180
    },
    {
      "epoch": 4.919642088961542,
      "grad_norm": 6.125,
      "learning_rate": 6.377322955712916e-07,
      "loss": 1.2655,
      "step": 57181
    },
    {
      "epoch": 4.919728125268864,
      "grad_norm": 1.328125,
      "learning_rate": 6.363677224285325e-07,
      "loss": 0.8372,
      "step": 57182
    },
    {
      "epoch": 4.919814161576185,
      "grad_norm": 1.3125,
      "learning_rate": 6.350046098272944e-07,
      "loss": 0.7079,
      "step": 57183
    },
    {
      "epoch": 4.919900197883507,
      "grad_norm": 1.1796875,
      "learning_rate": 6.336429577715741e-07,
      "loss": 0.9698,
      "step": 57184
    },
    {
      "epoch": 4.919986234190828,
      "grad_norm": 1.1015625,
      "learning_rate": 6.322827662652575e-07,
      "loss": 1.0241,
      "step": 57185
    },
    {
      "epoch": 4.92007227049815,
      "grad_norm": 0.86328125,
      "learning_rate": 6.309240353124524e-07,
      "loss": 0.6226,
      "step": 57186
    },
    {
      "epoch": 4.920158306805472,
      "grad_norm": 1.2578125,
      "learning_rate": 6.295667649170445e-07,
      "loss": 1.2346,
      "step": 57187
    },
    {
      "epoch": 4.920244343112794,
      "grad_norm": 182.0,
      "learning_rate": 6.282109550830306e-07,
      "loss": 1.2645,
      "step": 57188
    },
    {
      "epoch": 4.920330379420116,
      "grad_norm": 1.671875,
      "learning_rate": 6.268566058144076e-07,
      "loss": 1.4409,
      "step": 57189
    },
    {
      "epoch": 4.920416415727437,
      "grad_norm": 1.0546875,
      "learning_rate": 6.255037171150613e-07,
      "loss": 0.9933,
      "step": 57190
    },
    {
      "epoch": 4.920502452034759,
      "grad_norm": 1.2265625,
      "learning_rate": 6.241522889890439e-07,
      "loss": 0.8055,
      "step": 57191
    },
    {
      "epoch": 4.92058848834208,
      "grad_norm": 1.2265625,
      "learning_rate": 6.228023214401857e-07,
      "loss": 0.8262,
      "step": 57192
    },
    {
      "epoch": 4.920674524649402,
      "grad_norm": 1.2109375,
      "learning_rate": 6.21453814472539e-07,
      "loss": 1.3424,
      "step": 57193
    },
    {
      "epoch": 4.920760560956724,
      "grad_norm": 1.0234375,
      "learning_rate": 6.201067680900452e-07,
      "loss": 1.1778,
      "step": 57194
    },
    {
      "epoch": 4.920846597264045,
      "grad_norm": 1.1171875,
      "learning_rate": 6.1876118229659e-07,
      "loss": 1.0827,
      "step": 57195
    },
    {
      "epoch": 4.9209326335713675,
      "grad_norm": 1.3359375,
      "learning_rate": 6.174170570961702e-07,
      "loss": 1.2656,
      "step": 57196
    },
    {
      "epoch": 4.921018669878689,
      "grad_norm": 1.359375,
      "learning_rate": 6.16074392492616e-07,
      "loss": 1.3545,
      "step": 57197
    },
    {
      "epoch": 4.921104706186011,
      "grad_norm": 1.0859375,
      "learning_rate": 6.147331884899798e-07,
      "loss": 1.1026,
      "step": 57198
    },
    {
      "epoch": 4.921190742493332,
      "grad_norm": 1.265625,
      "learning_rate": 6.133934450920919e-07,
      "loss": 1.062,
      "step": 57199
    },
    {
      "epoch": 4.921276778800654,
      "grad_norm": 0.87109375,
      "learning_rate": 6.120551623028936e-07,
      "loss": 0.7765,
      "step": 57200
    },
    {
      "epoch": 4.921362815107976,
      "grad_norm": 0.890625,
      "learning_rate": 6.107183401263261e-07,
      "loss": 0.6998,
      "step": 57201
    },
    {
      "epoch": 4.921448851415297,
      "grad_norm": 0.875,
      "learning_rate": 6.093829785662752e-07,
      "loss": 1.2949,
      "step": 57202
    },
    {
      "epoch": 4.921534887722619,
      "grad_norm": 6.9375,
      "learning_rate": 6.080490776266823e-07,
      "loss": 1.5047,
      "step": 57203
    },
    {
      "epoch": 4.9216209240299404,
      "grad_norm": 1.78125,
      "learning_rate": 6.067166373113775e-07,
      "loss": 0.9537,
      "step": 57204
    },
    {
      "epoch": 4.9217069603372625,
      "grad_norm": 0.98828125,
      "learning_rate": 6.053856576243577e-07,
      "loss": 1.7907,
      "step": 57205
    },
    {
      "epoch": 4.921792996644584,
      "grad_norm": 0.8203125,
      "learning_rate": 6.040561385693977e-07,
      "loss": 1.023,
      "step": 57206
    },
    {
      "epoch": 4.921879032951906,
      "grad_norm": 1.2265625,
      "learning_rate": 6.027280801504386e-07,
      "loss": 0.7071,
      "step": 57207
    },
    {
      "epoch": 4.921965069259228,
      "grad_norm": 1.25,
      "learning_rate": 6.01401482371422e-07,
      "loss": 1.7138,
      "step": 57208
    },
    {
      "epoch": 4.922051105566549,
      "grad_norm": 1.09375,
      "learning_rate": 6.000763452361779e-07,
      "loss": 1.9287,
      "step": 57209
    },
    {
      "epoch": 4.922137141873871,
      "grad_norm": 1.1640625,
      "learning_rate": 5.987526687485367e-07,
      "loss": 0.9394,
      "step": 57210
    },
    {
      "epoch": 4.922223178181192,
      "grad_norm": 0.96484375,
      "learning_rate": 5.974304529124397e-07,
      "loss": 0.5396,
      "step": 57211
    },
    {
      "epoch": 4.922309214488514,
      "grad_norm": 1.234375,
      "learning_rate": 5.961096977317171e-07,
      "loss": 1.1247,
      "step": 57212
    },
    {
      "epoch": 4.9223952507958355,
      "grad_norm": 1.21875,
      "learning_rate": 5.947904032102547e-07,
      "loss": 1.4033,
      "step": 57213
    },
    {
      "epoch": 4.922481287103158,
      "grad_norm": 1.0234375,
      "learning_rate": 5.934725693519383e-07,
      "loss": 1.0816,
      "step": 57214
    },
    {
      "epoch": 4.92256732341048,
      "grad_norm": 1.1484375,
      "learning_rate": 5.921561961605426e-07,
      "loss": 0.3933,
      "step": 57215
    },
    {
      "epoch": 4.922653359717801,
      "grad_norm": 0.78515625,
      "learning_rate": 5.908412836399535e-07,
      "loss": 0.6783,
      "step": 57216
    },
    {
      "epoch": 4.922739396025123,
      "grad_norm": 1.1015625,
      "learning_rate": 5.895278317940566e-07,
      "loss": 1.2676,
      "step": 57217
    },
    {
      "epoch": 4.922825432332444,
      "grad_norm": 0.9765625,
      "learning_rate": 5.882158406266269e-07,
      "loss": 1.3412,
      "step": 57218
    },
    {
      "epoch": 4.922911468639766,
      "grad_norm": 1.2109375,
      "learning_rate": 5.8690531014155e-07,
      "loss": 0.5899,
      "step": 57219
    },
    {
      "epoch": 4.922997504947087,
      "grad_norm": 1.15625,
      "learning_rate": 5.855962403426562e-07,
      "loss": 1.6576,
      "step": 57220
    },
    {
      "epoch": 4.923083541254409,
      "grad_norm": 0.8984375,
      "learning_rate": 5.842886312337758e-07,
      "loss": 0.6469,
      "step": 57221
    },
    {
      "epoch": 4.9231695775617315,
      "grad_norm": 1.140625,
      "learning_rate": 5.829824828186835e-07,
      "loss": 0.9148,
      "step": 57222
    },
    {
      "epoch": 4.923255613869053,
      "grad_norm": 0.94140625,
      "learning_rate": 5.816777951012653e-07,
      "loss": 0.7323,
      "step": 57223
    },
    {
      "epoch": 4.923341650176375,
      "grad_norm": 1.3671875,
      "learning_rate": 5.803745680852956e-07,
      "loss": 1.231,
      "step": 57224
    },
    {
      "epoch": 4.923427686483696,
      "grad_norm": 0.91015625,
      "learning_rate": 5.79072801774605e-07,
      "loss": 0.4328,
      "step": 57225
    },
    {
      "epoch": 4.923513722791018,
      "grad_norm": 1.03125,
      "learning_rate": 5.77772496172968e-07,
      "loss": 0.9073,
      "step": 57226
    },
    {
      "epoch": 4.923599759098339,
      "grad_norm": 1.28125,
      "learning_rate": 5.764736512842151e-07,
      "loss": 1.671,
      "step": 57227
    },
    {
      "epoch": 4.923685795405661,
      "grad_norm": 0.76171875,
      "learning_rate": 5.751762671121763e-07,
      "loss": 0.3945,
      "step": 57228
    },
    {
      "epoch": 4.923771831712983,
      "grad_norm": 0.87109375,
      "learning_rate": 5.738803436606266e-07,
      "loss": 1.2787,
      "step": 57229
    },
    {
      "epoch": 4.9238578680203045,
      "grad_norm": 9.0,
      "learning_rate": 5.725858809332851e-07,
      "loss": 0.8648,
      "step": 57230
    },
    {
      "epoch": 4.9239439043276265,
      "grad_norm": 1.2734375,
      "learning_rate": 5.712928789340377e-07,
      "loss": 1.209,
      "step": 57231
    },
    {
      "epoch": 4.924029940634948,
      "grad_norm": 1.375,
      "learning_rate": 5.700013376666035e-07,
      "loss": 0.9967,
      "step": 57232
    },
    {
      "epoch": 4.92411597694227,
      "grad_norm": 1.1953125,
      "learning_rate": 5.687112571348129e-07,
      "loss": 1.0449,
      "step": 57233
    },
    {
      "epoch": 4.924202013249591,
      "grad_norm": 1.078125,
      "learning_rate": 5.674226373423852e-07,
      "loss": 0.8169,
      "step": 57234
    },
    {
      "epoch": 4.924288049556913,
      "grad_norm": 1.3203125,
      "learning_rate": 5.661354782931505e-07,
      "loss": 1.3227,
      "step": 57235
    },
    {
      "epoch": 4.924374085864235,
      "grad_norm": 1.6640625,
      "learning_rate": 5.648497799907726e-07,
      "loss": 1.54,
      "step": 57236
    },
    {
      "epoch": 4.924460122171556,
      "grad_norm": 0.88671875,
      "learning_rate": 5.635655424390818e-07,
      "loss": 0.9184,
      "step": 57237
    },
    {
      "epoch": 4.924546158478878,
      "grad_norm": 1.171875,
      "learning_rate": 5.622827656418527e-07,
      "loss": 1.3303,
      "step": 57238
    },
    {
      "epoch": 4.9246321947861995,
      "grad_norm": 1.0390625,
      "learning_rate": 5.610014496028048e-07,
      "loss": 0.5859,
      "step": 57239
    },
    {
      "epoch": 4.924718231093522,
      "grad_norm": 1.0859375,
      "learning_rate": 5.597215943257128e-07,
      "loss": 0.7955,
      "step": 57240
    },
    {
      "epoch": 4.924804267400843,
      "grad_norm": 1.171875,
      "learning_rate": 5.584431998142958e-07,
      "loss": 1.1139,
      "step": 57241
    },
    {
      "epoch": 4.924890303708165,
      "grad_norm": 0.89453125,
      "learning_rate": 5.571662660722732e-07,
      "loss": 0.7283,
      "step": 57242
    },
    {
      "epoch": 4.924976340015487,
      "grad_norm": 1.109375,
      "learning_rate": 5.558907931034197e-07,
      "loss": 0.8113,
      "step": 57243
    },
    {
      "epoch": 4.925062376322808,
      "grad_norm": 0.75,
      "learning_rate": 5.546167809114544e-07,
      "loss": 0.454,
      "step": 57244
    },
    {
      "epoch": 4.92514841263013,
      "grad_norm": 0.9140625,
      "learning_rate": 5.533442295000412e-07,
      "loss": 0.5722,
      "step": 57245
    },
    {
      "epoch": 4.925234448937451,
      "grad_norm": 1.1328125,
      "learning_rate": 5.520731388730104e-07,
      "loss": 1.496,
      "step": 57246
    },
    {
      "epoch": 4.925320485244773,
      "grad_norm": 1.1875,
      "learning_rate": 5.508035090340257e-07,
      "loss": 1.4875,
      "step": 57247
    },
    {
      "epoch": 4.925406521552095,
      "grad_norm": 3.5,
      "learning_rate": 5.495353399868064e-07,
      "loss": 0.7931,
      "step": 57248
    },
    {
      "epoch": 4.925492557859417,
      "grad_norm": 1.6875,
      "learning_rate": 5.48268631735016e-07,
      "loss": 0.839,
      "step": 57249
    },
    {
      "epoch": 4.925578594166739,
      "grad_norm": 1.7421875,
      "learning_rate": 5.470033842824296e-07,
      "loss": 1.0863,
      "step": 57250
    },
    {
      "epoch": 4.92566463047406,
      "grad_norm": 0.78515625,
      "learning_rate": 5.457395976327661e-07,
      "loss": 0.6778,
      "step": 57251
    },
    {
      "epoch": 4.925750666781382,
      "grad_norm": 0.65234375,
      "learning_rate": 5.44477271789634e-07,
      "loss": 0.5025,
      "step": 57252
    },
    {
      "epoch": 4.925836703088703,
      "grad_norm": 0.90625,
      "learning_rate": 5.432164067567524e-07,
      "loss": 0.8925,
      "step": 57253
    },
    {
      "epoch": 4.925922739396025,
      "grad_norm": 1.0390625,
      "learning_rate": 5.419570025378407e-07,
      "loss": 1.1496,
      "step": 57254
    },
    {
      "epoch": 4.926008775703346,
      "grad_norm": 1.109375,
      "learning_rate": 5.406990591365624e-07,
      "loss": 1.6414,
      "step": 57255
    },
    {
      "epoch": 4.9260948120106685,
      "grad_norm": 1.3984375,
      "learning_rate": 5.394425765566369e-07,
      "loss": 1.4389,
      "step": 57256
    },
    {
      "epoch": 4.9261808483179905,
      "grad_norm": 1.0859375,
      "learning_rate": 5.381875548016724e-07,
      "loss": 0.6255,
      "step": 57257
    },
    {
      "epoch": 4.926266884625312,
      "grad_norm": 1.5390625,
      "learning_rate": 5.369339938753326e-07,
      "loss": 0.985,
      "step": 57258
    },
    {
      "epoch": 4.926352920932634,
      "grad_norm": 1.1328125,
      "learning_rate": 5.356818937813923e-07,
      "loss": 0.7048,
      "step": 57259
    },
    {
      "epoch": 4.926438957239955,
      "grad_norm": 1.40625,
      "learning_rate": 5.344312545233487e-07,
      "loss": 0.748,
      "step": 57260
    },
    {
      "epoch": 4.926524993547277,
      "grad_norm": 1.078125,
      "learning_rate": 5.331820761050321e-07,
      "loss": 0.8679,
      "step": 57261
    },
    {
      "epoch": 4.926611029854598,
      "grad_norm": 1.4140625,
      "learning_rate": 5.319343585299951e-07,
      "loss": 0.6673,
      "step": 57262
    },
    {
      "epoch": 4.92669706616192,
      "grad_norm": 1.2421875,
      "learning_rate": 5.306881018019016e-07,
      "loss": 1.7988,
      "step": 57263
    },
    {
      "epoch": 4.926783102469242,
      "grad_norm": 1.6875,
      "learning_rate": 5.294433059244153e-07,
      "loss": 1.6602,
      "step": 57264
    },
    {
      "epoch": 4.9268691387765635,
      "grad_norm": 1.0703125,
      "learning_rate": 5.281999709011442e-07,
      "loss": 1.0585,
      "step": 57265
    },
    {
      "epoch": 4.926955175083886,
      "grad_norm": 0.9296875,
      "learning_rate": 5.269580967357523e-07,
      "loss": 1.0816,
      "step": 57266
    },
    {
      "epoch": 4.927041211391207,
      "grad_norm": 2.015625,
      "learning_rate": 5.257176834319032e-07,
      "loss": 0.5873,
      "step": 57267
    },
    {
      "epoch": 4.927127247698529,
      "grad_norm": 1.171875,
      "learning_rate": 5.244787309931498e-07,
      "loss": 0.988,
      "step": 57268
    },
    {
      "epoch": 4.92721328400585,
      "grad_norm": 1.3671875,
      "learning_rate": 5.232412394231557e-07,
      "loss": 0.5611,
      "step": 57269
    },
    {
      "epoch": 4.927299320313172,
      "grad_norm": 0.9453125,
      "learning_rate": 5.220052087255289e-07,
      "loss": 0.9241,
      "step": 57270
    },
    {
      "epoch": 4.927385356620494,
      "grad_norm": 0.87890625,
      "learning_rate": 5.207706389039335e-07,
      "loss": 0.6467,
      "step": 57271
    },
    {
      "epoch": 4.927471392927815,
      "grad_norm": 1.15625,
      "learning_rate": 5.195375299619221e-07,
      "loss": 1.1972,
      "step": 57272
    },
    {
      "epoch": 4.927557429235137,
      "grad_norm": 1.515625,
      "learning_rate": 5.183058819031027e-07,
      "loss": 1.8727,
      "step": 57273
    },
    {
      "epoch": 4.927643465542459,
      "grad_norm": 1.1953125,
      "learning_rate": 5.17075694731084e-07,
      "loss": 1.2512,
      "step": 57274
    },
    {
      "epoch": 4.927729501849781,
      "grad_norm": 1.4609375,
      "learning_rate": 5.158469684495293e-07,
      "loss": 1.4698,
      "step": 57275
    },
    {
      "epoch": 4.927815538157102,
      "grad_norm": 1.0625,
      "learning_rate": 5.146197030619359e-07,
      "loss": 1.2153,
      "step": 57276
    },
    {
      "epoch": 4.927901574464424,
      "grad_norm": 1.9765625,
      "learning_rate": 5.133938985719677e-07,
      "loss": 1.0233,
      "step": 57277
    },
    {
      "epoch": 4.927987610771746,
      "grad_norm": 234.0,
      "learning_rate": 5.121695549831217e-07,
      "loss": 1.2529,
      "step": 57278
    },
    {
      "epoch": 4.928073647079067,
      "grad_norm": 1.0703125,
      "learning_rate": 5.109466722990619e-07,
      "loss": 1.0106,
      "step": 57279
    },
    {
      "epoch": 4.928159683386389,
      "grad_norm": 1.1875,
      "learning_rate": 5.097252505233408e-07,
      "loss": 0.7621,
      "step": 57280
    },
    {
      "epoch": 4.92824571969371,
      "grad_norm": 1.1328125,
      "learning_rate": 5.085052896595666e-07,
      "loss": 1.572,
      "step": 57281
    },
    {
      "epoch": 4.9283317560010325,
      "grad_norm": 30.0,
      "learning_rate": 5.072867897112366e-07,
      "loss": 0.998,
      "step": 57282
    },
    {
      "epoch": 4.928417792308354,
      "grad_norm": 54.25,
      "learning_rate": 5.060697506819589e-07,
      "loss": 1.2277,
      "step": 57283
    },
    {
      "epoch": 4.928503828615676,
      "grad_norm": 1.0078125,
      "learning_rate": 5.048541725752309e-07,
      "loss": 0.9609,
      "step": 57284
    },
    {
      "epoch": 4.928589864922998,
      "grad_norm": 1.3359375,
      "learning_rate": 5.036400553947163e-07,
      "loss": 1.052,
      "step": 57285
    },
    {
      "epoch": 4.928675901230319,
      "grad_norm": 1.90625,
      "learning_rate": 5.024273991438566e-07,
      "loss": 0.8151,
      "step": 57286
    },
    {
      "epoch": 4.928761937537641,
      "grad_norm": 1.0234375,
      "learning_rate": 5.012162038262602e-07,
      "loss": 0.9651,
      "step": 57287
    },
    {
      "epoch": 4.928847973844962,
      "grad_norm": 1.03125,
      "learning_rate": 5.000064694454798e-07,
      "loss": 1.6155,
      "step": 57288
    },
    {
      "epoch": 4.928934010152284,
      "grad_norm": 0.83203125,
      "learning_rate": 4.98798196005068e-07,
      "loss": 0.6863,
      "step": 57289
    },
    {
      "epoch": 4.929020046459606,
      "grad_norm": 9.0,
      "learning_rate": 4.975913835084667e-07,
      "loss": 1.6293,
      "step": 57290
    },
    {
      "epoch": 4.9291060827669275,
      "grad_norm": 1.1171875,
      "learning_rate": 4.963860319592839e-07,
      "loss": 1.0417,
      "step": 57291
    },
    {
      "epoch": 4.92919211907425,
      "grad_norm": 1.375,
      "learning_rate": 4.951821413610169e-07,
      "loss": 2.062,
      "step": 57292
    },
    {
      "epoch": 4.929278155381571,
      "grad_norm": 61.5,
      "learning_rate": 4.939797117172184e-07,
      "loss": 0.7532,
      "step": 57293
    },
    {
      "epoch": 4.929364191688893,
      "grad_norm": 1.1328125,
      "learning_rate": 4.927787430313302e-07,
      "loss": 0.8275,
      "step": 57294
    },
    {
      "epoch": 4.929450227996214,
      "grad_norm": 1.2890625,
      "learning_rate": 4.915792353069603e-07,
      "loss": 1.8038,
      "step": 57295
    },
    {
      "epoch": 4.929536264303536,
      "grad_norm": 1.5,
      "learning_rate": 4.903811885475506e-07,
      "loss": 1.3889,
      "step": 57296
    },
    {
      "epoch": 4.929622300610858,
      "grad_norm": 1.3828125,
      "learning_rate": 4.891846027566537e-07,
      "loss": 1.647,
      "step": 57297
    },
    {
      "epoch": 4.929708336918179,
      "grad_norm": 0.92578125,
      "learning_rate": 4.879894779377114e-07,
      "loss": 0.5624,
      "step": 57298
    },
    {
      "epoch": 4.929794373225501,
      "grad_norm": 1.25,
      "learning_rate": 4.867958140942764e-07,
      "loss": 0.7638,
      "step": 57299
    },
    {
      "epoch": 4.929880409532823,
      "grad_norm": 0.71875,
      "learning_rate": 4.856036112297901e-07,
      "loss": 0.5736,
      "step": 57300
    },
    {
      "epoch": 4.929966445840145,
      "grad_norm": 1.3203125,
      "learning_rate": 4.844128693477501e-07,
      "loss": 0.86,
      "step": 57301
    },
    {
      "epoch": 4.930052482147467,
      "grad_norm": 1.0625,
      "learning_rate": 4.832235884517089e-07,
      "loss": 0.5917,
      "step": 57302
    },
    {
      "epoch": 4.930138518454788,
      "grad_norm": 1.0703125,
      "learning_rate": 4.820357685450527e-07,
      "loss": 1.4085,
      "step": 57303
    },
    {
      "epoch": 4.93022455476211,
      "grad_norm": 1.46875,
      "learning_rate": 4.808494096312787e-07,
      "loss": 0.7547,
      "step": 57304
    },
    {
      "epoch": 4.930310591069431,
      "grad_norm": 1.0390625,
      "learning_rate": 4.796645117138843e-07,
      "loss": 1.0458,
      "step": 57305
    },
    {
      "epoch": 4.930396627376753,
      "grad_norm": 1.2734375,
      "learning_rate": 4.784810747963108e-07,
      "loss": 0.9364,
      "step": 57306
    },
    {
      "epoch": 4.930482663684074,
      "grad_norm": 1.1640625,
      "learning_rate": 4.772990988820558e-07,
      "loss": 1.537,
      "step": 57307
    },
    {
      "epoch": 4.9305686999913965,
      "grad_norm": 1.453125,
      "learning_rate": 4.7611858397450526e-07,
      "loss": 1.6974,
      "step": 57308
    },
    {
      "epoch": 4.9306547362987185,
      "grad_norm": 1.859375,
      "learning_rate": 4.749395300772119e-07,
      "loss": 1.4054,
      "step": 57309
    },
    {
      "epoch": 4.93074077260604,
      "grad_norm": 0.765625,
      "learning_rate": 4.737619371935065e-07,
      "loss": 0.4231,
      "step": 57310
    },
    {
      "epoch": 4.930826808913362,
      "grad_norm": 1.1640625,
      "learning_rate": 4.7258580532694164e-07,
      "loss": 0.8203,
      "step": 57311
    },
    {
      "epoch": 4.930912845220683,
      "grad_norm": 1.8671875,
      "learning_rate": 4.7141113448090353e-07,
      "loss": 1.873,
      "step": 57312
    },
    {
      "epoch": 4.930998881528005,
      "grad_norm": 5.6875,
      "learning_rate": 4.7023792465877844e-07,
      "loss": 0.7175,
      "step": 57313
    },
    {
      "epoch": 4.931084917835326,
      "grad_norm": 0.85546875,
      "learning_rate": 4.6906617586411903e-07,
      "loss": 1.0395,
      "step": 57314
    },
    {
      "epoch": 4.931170954142648,
      "grad_norm": 382.0,
      "learning_rate": 4.6789588810025595e-07,
      "loss": 1.0296,
      "step": 57315
    },
    {
      "epoch": 4.93125699044997,
      "grad_norm": 1.015625,
      "learning_rate": 4.667270613706864e-07,
      "loss": 0.9708,
      "step": 57316
    },
    {
      "epoch": 4.9313430267572915,
      "grad_norm": 1.0078125,
      "learning_rate": 4.655596956787411e-07,
      "loss": 1.2826,
      "step": 57317
    },
    {
      "epoch": 4.931429063064614,
      "grad_norm": 1.1328125,
      "learning_rate": 4.6439379102791723e-07,
      "loss": 0.9541,
      "step": 57318
    },
    {
      "epoch": 4.931515099371935,
      "grad_norm": 1.046875,
      "learning_rate": 4.6322934742154544e-07,
      "loss": 1.0205,
      "step": 57319
    },
    {
      "epoch": 4.931601135679257,
      "grad_norm": 1.4296875,
      "learning_rate": 4.620663648631229e-07,
      "loss": 1.0369,
      "step": 57320
    },
    {
      "epoch": 4.931687171986578,
      "grad_norm": 1.4296875,
      "learning_rate": 4.609048433559249e-07,
      "loss": 2.1375,
      "step": 57321
    },
    {
      "epoch": 4.9317732082939,
      "grad_norm": 1.453125,
      "learning_rate": 4.5974478290350394e-07,
      "loss": 1.1443,
      "step": 57322
    },
    {
      "epoch": 4.931859244601222,
      "grad_norm": 0.94921875,
      "learning_rate": 4.585861835091354e-07,
      "loss": 1.3216,
      "step": 57323
    },
    {
      "epoch": 4.931945280908543,
      "grad_norm": 1.0625,
      "learning_rate": 4.5742904517626085e-07,
      "loss": 0.8019,
      "step": 57324
    },
    {
      "epoch": 4.932031317215865,
      "grad_norm": 19.75,
      "learning_rate": 4.5627336790821094e-07,
      "loss": 1.1726,
      "step": 57325
    },
    {
      "epoch": 4.932117353523187,
      "grad_norm": 1.4375,
      "learning_rate": 4.5511915170842745e-07,
      "loss": 1.4358,
      "step": 57326
    },
    {
      "epoch": 4.932203389830509,
      "grad_norm": 1.484375,
      "learning_rate": 4.539663965802965e-07,
      "loss": 1.6522,
      "step": 57327
    },
    {
      "epoch": 4.93228942613783,
      "grad_norm": 1.21875,
      "learning_rate": 4.5281510252714875e-07,
      "loss": 1.642,
      "step": 57328
    },
    {
      "epoch": 4.932375462445152,
      "grad_norm": 0.94140625,
      "learning_rate": 4.516652695523149e-07,
      "loss": 0.9038,
      "step": 57329
    },
    {
      "epoch": 4.932461498752474,
      "grad_norm": 1.5078125,
      "learning_rate": 4.505168976592922e-07,
      "loss": 1.1222,
      "step": 57330
    },
    {
      "epoch": 4.932547535059795,
      "grad_norm": 1.0234375,
      "learning_rate": 4.493699868512446e-07,
      "loss": 1.2907,
      "step": 57331
    },
    {
      "epoch": 4.932633571367117,
      "grad_norm": 1.3125,
      "learning_rate": 4.482245371317251e-07,
      "loss": 1.4611,
      "step": 57332
    },
    {
      "epoch": 4.932719607674438,
      "grad_norm": 1.0,
      "learning_rate": 4.470805485039531e-07,
      "loss": 0.7266,
      "step": 57333
    },
    {
      "epoch": 4.9328056439817605,
      "grad_norm": 0.875,
      "learning_rate": 4.4593802097125933e-07,
      "loss": 0.902,
      "step": 57334
    },
    {
      "epoch": 4.932891680289082,
      "grad_norm": 36.25,
      "learning_rate": 4.4479695453708557e-07,
      "loss": 1.4765,
      "step": 57335
    },
    {
      "epoch": 4.932977716596404,
      "grad_norm": 1.296875,
      "learning_rate": 4.4365734920476244e-07,
      "loss": 0.8858,
      "step": 57336
    },
    {
      "epoch": 4.933063752903726,
      "grad_norm": 0.9453125,
      "learning_rate": 4.425192049775095e-07,
      "loss": 1.1927,
      "step": 57337
    },
    {
      "epoch": 4.933149789211047,
      "grad_norm": 0.94140625,
      "learning_rate": 4.413825218587686e-07,
      "loss": 0.6111,
      "step": 57338
    },
    {
      "epoch": 4.933235825518369,
      "grad_norm": 1.109375,
      "learning_rate": 4.402472998518703e-07,
      "loss": 1.716,
      "step": 57339
    },
    {
      "epoch": 4.93332186182569,
      "grad_norm": 0.9609375,
      "learning_rate": 4.391135389600343e-07,
      "loss": 0.5201,
      "step": 57340
    },
    {
      "epoch": 4.933407898133012,
      "grad_norm": 1.296875,
      "learning_rate": 4.379812391866467e-07,
      "loss": 1.4068,
      "step": 57341
    },
    {
      "epoch": 4.9334939344403335,
      "grad_norm": 1.234375,
      "learning_rate": 4.368504005349827e-07,
      "loss": 1.0246,
      "step": 57342
    },
    {
      "epoch": 4.9335799707476555,
      "grad_norm": 0.99609375,
      "learning_rate": 4.35721023008373e-07,
      "loss": 1.0779,
      "step": 57343
    },
    {
      "epoch": 4.933666007054978,
      "grad_norm": 5600.0,
      "learning_rate": 4.345931066101483e-07,
      "loss": 1.2194,
      "step": 57344
    },
    {
      "epoch": 4.933752043362299,
      "grad_norm": 0.99609375,
      "learning_rate": 4.3346665134352814e-07,
      "loss": 0.9528,
      "step": 57345
    },
    {
      "epoch": 4.933838079669621,
      "grad_norm": 1.921875,
      "learning_rate": 4.323416572118988e-07,
      "loss": 0.9458,
      "step": 57346
    },
    {
      "epoch": 4.933924115976942,
      "grad_norm": 1.453125,
      "learning_rate": 4.3121812421853534e-07,
      "loss": 2.0008,
      "step": 57347
    },
    {
      "epoch": 4.934010152284264,
      "grad_norm": 1.125,
      "learning_rate": 4.3009605236665747e-07,
      "loss": 1.6483,
      "step": 57348
    },
    {
      "epoch": 4.934096188591585,
      "grad_norm": 238.0,
      "learning_rate": 4.289754416595959e-07,
      "loss": 1.3481,
      "step": 57349
    },
    {
      "epoch": 4.934182224898907,
      "grad_norm": 2.65625,
      "learning_rate": 4.2785629210062573e-07,
      "loss": 1.0116,
      "step": 57350
    },
    {
      "epoch": 4.934268261206229,
      "grad_norm": 0.9765625,
      "learning_rate": 4.2673860369302206e-07,
      "loss": 1.1012,
      "step": 57351
    },
    {
      "epoch": 4.934354297513551,
      "grad_norm": 4.0625,
      "learning_rate": 4.2562237644000467e-07,
      "loss": 1.1964,
      "step": 57352
    },
    {
      "epoch": 4.934440333820873,
      "grad_norm": 1.21875,
      "learning_rate": 4.2450761034490413e-07,
      "loss": 1.1808,
      "step": 57353
    },
    {
      "epoch": 4.934526370128194,
      "grad_norm": 1.296875,
      "learning_rate": 4.2339430541099567e-07,
      "loss": 0.5673,
      "step": 57354
    },
    {
      "epoch": 4.934612406435516,
      "grad_norm": 1.3984375,
      "learning_rate": 4.2228246164144336e-07,
      "loss": 1.1678,
      "step": 57355
    },
    {
      "epoch": 4.934698442742837,
      "grad_norm": 0.92578125,
      "learning_rate": 4.211720790395779e-07,
      "loss": 0.6994,
      "step": 57356
    },
    {
      "epoch": 4.934784479050159,
      "grad_norm": 9.5625,
      "learning_rate": 4.200631576086189e-07,
      "loss": 1.2313,
      "step": 57357
    },
    {
      "epoch": 4.934870515357481,
      "grad_norm": 1.0546875,
      "learning_rate": 4.1895569735178607e-07,
      "loss": 1.0698,
      "step": 57358
    },
    {
      "epoch": 4.934956551664802,
      "grad_norm": 1.0546875,
      "learning_rate": 4.178496982723545e-07,
      "loss": 1.3983,
      "step": 57359
    },
    {
      "epoch": 4.9350425879721245,
      "grad_norm": 1.078125,
      "learning_rate": 4.1674516037359943e-07,
      "loss": 1.6629,
      "step": 57360
    },
    {
      "epoch": 4.935128624279446,
      "grad_norm": 1.21875,
      "learning_rate": 4.1564208365862944e-07,
      "loss": 0.5628,
      "step": 57361
    },
    {
      "epoch": 4.935214660586768,
      "grad_norm": 1.0859375,
      "learning_rate": 4.1454046813077517e-07,
      "loss": 1.5785,
      "step": 57362
    },
    {
      "epoch": 4.935300696894089,
      "grad_norm": 172.0,
      "learning_rate": 4.134403137932563e-07,
      "loss": 1.5275,
      "step": 57363
    },
    {
      "epoch": 4.935386733201411,
      "grad_norm": 1.125,
      "learning_rate": 4.12341620649237e-07,
      "loss": 0.8867,
      "step": 57364
    },
    {
      "epoch": 4.935472769508733,
      "grad_norm": 1.65625,
      "learning_rate": 4.112443887019368e-07,
      "loss": 1.3613,
      "step": 57365
    },
    {
      "epoch": 4.935558805816054,
      "grad_norm": 1.1640625,
      "learning_rate": 4.101486179545755e-07,
      "loss": 1.409,
      "step": 57366
    },
    {
      "epoch": 4.935644842123376,
      "grad_norm": 10.5625,
      "learning_rate": 4.090543084103726e-07,
      "loss": 1.1688,
      "step": 57367
    },
    {
      "epoch": 4.9357308784306975,
      "grad_norm": 1.625,
      "learning_rate": 4.0796146007254785e-07,
      "loss": 0.994,
      "step": 57368
    },
    {
      "epoch": 4.9358169147380195,
      "grad_norm": 0.98046875,
      "learning_rate": 4.068700729442654e-07,
      "loss": 1.0564,
      "step": 57369
    },
    {
      "epoch": 4.935902951045341,
      "grad_norm": 1.609375,
      "learning_rate": 4.0578014702868925e-07,
      "loss": 1.6009,
      "step": 57370
    },
    {
      "epoch": 4.935988987352663,
      "grad_norm": 1.1953125,
      "learning_rate": 4.046916823290947e-07,
      "loss": 1.0582,
      "step": 57371
    },
    {
      "epoch": 4.936075023659985,
      "grad_norm": 0.97265625,
      "learning_rate": 4.036046788485348e-07,
      "loss": 1.1756,
      "step": 57372
    },
    {
      "epoch": 4.936161059967306,
      "grad_norm": 1.2734375,
      "learning_rate": 4.025191365903402e-07,
      "loss": 1.5005,
      "step": 57373
    },
    {
      "epoch": 4.936247096274628,
      "grad_norm": 36352.0,
      "learning_rate": 4.014350555575641e-07,
      "loss": 1.1452,
      "step": 57374
    },
    {
      "epoch": 4.936333132581949,
      "grad_norm": 1.078125,
      "learning_rate": 4.0035243575342607e-07,
      "loss": 1.3511,
      "step": 57375
    },
    {
      "epoch": 4.936419168889271,
      "grad_norm": 0.7734375,
      "learning_rate": 3.992712771810902e-07,
      "loss": 0.608,
      "step": 57376
    },
    {
      "epoch": 4.9365052051965925,
      "grad_norm": 1.1015625,
      "learning_rate": 3.981915798437763e-07,
      "loss": 1.1951,
      "step": 57377
    },
    {
      "epoch": 4.936591241503915,
      "grad_norm": 0.95703125,
      "learning_rate": 3.9711334374453735e-07,
      "loss": 1.4395,
      "step": 57378
    },
    {
      "epoch": 4.936677277811237,
      "grad_norm": 1.28125,
      "learning_rate": 3.960365688865375e-07,
      "loss": 1.5846,
      "step": 57379
    },
    {
      "epoch": 4.936763314118558,
      "grad_norm": 1.1171875,
      "learning_rate": 3.949612552730519e-07,
      "loss": 1.1776,
      "step": 57380
    },
    {
      "epoch": 4.93684935042588,
      "grad_norm": 1.2109375,
      "learning_rate": 3.9388740290707825e-07,
      "loss": 1.301,
      "step": 57381
    },
    {
      "epoch": 4.936935386733201,
      "grad_norm": 0.890625,
      "learning_rate": 3.928150117918361e-07,
      "loss": 1.2348,
      "step": 57382
    },
    {
      "epoch": 4.937021423040523,
      "grad_norm": 1.328125,
      "learning_rate": 3.917440819304341e-07,
      "loss": 1.1164,
      "step": 57383
    },
    {
      "epoch": 4.937107459347844,
      "grad_norm": 0.984375,
      "learning_rate": 3.906746133259809e-07,
      "loss": 1.1702,
      "step": 57384
    },
    {
      "epoch": 4.937193495655166,
      "grad_norm": 1.4375,
      "learning_rate": 3.8960660598169607e-07,
      "loss": 1.27,
      "step": 57385
    },
    {
      "epoch": 4.9372795319624885,
      "grad_norm": 6.03125,
      "learning_rate": 3.885400599005773e-07,
      "loss": 0.5925,
      "step": 57386
    },
    {
      "epoch": 4.93736556826981,
      "grad_norm": 3.375,
      "learning_rate": 3.8747497508584415e-07,
      "loss": 1.0247,
      "step": 57387
    },
    {
      "epoch": 4.937451604577132,
      "grad_norm": 2.28125,
      "learning_rate": 3.8641135154060534e-07,
      "loss": 1.4653,
      "step": 57388
    },
    {
      "epoch": 4.937537640884453,
      "grad_norm": 10.625,
      "learning_rate": 3.8534918926791394e-07,
      "loss": 0.9552,
      "step": 57389
    },
    {
      "epoch": 4.937623677191775,
      "grad_norm": 1.0703125,
      "learning_rate": 3.8428848827087857e-07,
      "loss": 1.2288,
      "step": 57390
    },
    {
      "epoch": 4.937709713499096,
      "grad_norm": 0.79296875,
      "learning_rate": 3.8322924855266337e-07,
      "loss": 0.991,
      "step": 57391
    },
    {
      "epoch": 4.937795749806418,
      "grad_norm": 1.2109375,
      "learning_rate": 3.8217147011632147e-07,
      "loss": 1.1958,
      "step": 57392
    },
    {
      "epoch": 4.93788178611374,
      "grad_norm": 1.0703125,
      "learning_rate": 3.8111515296501696e-07,
      "loss": 0.7957,
      "step": 57393
    },
    {
      "epoch": 4.9379678224210615,
      "grad_norm": 1.2890625,
      "learning_rate": 3.8006029710169197e-07,
      "loss": 1.0948,
      "step": 57394
    },
    {
      "epoch": 4.9380538587283835,
      "grad_norm": 1.2109375,
      "learning_rate": 3.7900690252956615e-07,
      "loss": 1.2184,
      "step": 57395
    },
    {
      "epoch": 4.938139895035705,
      "grad_norm": 1.109375,
      "learning_rate": 3.7795496925163707e-07,
      "loss": 1.5444,
      "step": 57396
    },
    {
      "epoch": 4.938225931343027,
      "grad_norm": 0.85546875,
      "learning_rate": 3.769044972710689e-07,
      "loss": 0.4276,
      "step": 57397
    },
    {
      "epoch": 4.938311967650348,
      "grad_norm": 7.875,
      "learning_rate": 3.7585548659085923e-07,
      "loss": 0.9778,
      "step": 57398
    },
    {
      "epoch": 4.93839800395767,
      "grad_norm": 1.0078125,
      "learning_rate": 3.748079372141167e-07,
      "loss": 0.8672,
      "step": 57399
    },
    {
      "epoch": 4.938484040264992,
      "grad_norm": 1.296875,
      "learning_rate": 3.737618491438388e-07,
      "loss": 1.507,
      "step": 57400
    },
    {
      "epoch": 4.938570076572313,
      "grad_norm": 362.0,
      "learning_rate": 3.727172223831898e-07,
      "loss": 0.7348,
      "step": 57401
    },
    {
      "epoch": 4.938656112879635,
      "grad_norm": 29.25,
      "learning_rate": 3.716740569351673e-07,
      "loss": 1.2637,
      "step": 57402
    },
    {
      "epoch": 4.9387421491869565,
      "grad_norm": 1.2109375,
      "learning_rate": 3.7063235280276884e-07,
      "loss": 1.3525,
      "step": 57403
    },
    {
      "epoch": 4.938828185494279,
      "grad_norm": 0.88671875,
      "learning_rate": 3.695921099891586e-07,
      "loss": 1.5034,
      "step": 57404
    },
    {
      "epoch": 4.9389142218016,
      "grad_norm": 0.91796875,
      "learning_rate": 3.6855332849733413e-07,
      "loss": 0.5464,
      "step": 57405
    },
    {
      "epoch": 4.939000258108922,
      "grad_norm": 0.80078125,
      "learning_rate": 3.675160083302376e-07,
      "loss": 0.8271,
      "step": 57406
    },
    {
      "epoch": 4.939086294416244,
      "grad_norm": 20.0,
      "learning_rate": 3.6648014949103304e-07,
      "loss": 0.9361,
      "step": 57407
    },
    {
      "epoch": 4.939172330723565,
      "grad_norm": 5.8125,
      "learning_rate": 3.654457519827181e-07,
      "loss": 0.9532,
      "step": 57408
    },
    {
      "epoch": 4.939258367030887,
      "grad_norm": 1.109375,
      "learning_rate": 3.6441281580829047e-07,
      "loss": 0.7091,
      "step": 57409
    },
    {
      "epoch": 4.939344403338208,
      "grad_norm": 1.078125,
      "learning_rate": 3.633813409707476e-07,
      "loss": 1.5422,
      "step": 57410
    },
    {
      "epoch": 4.93943043964553,
      "grad_norm": 46.25,
      "learning_rate": 3.623513274731982e-07,
      "loss": 0.9443,
      "step": 57411
    },
    {
      "epoch": 4.9395164759528525,
      "grad_norm": 1.34375,
      "learning_rate": 3.613227753185289e-07,
      "loss": 1.0972,
      "step": 57412
    },
    {
      "epoch": 4.939602512260174,
      "grad_norm": 1.15625,
      "learning_rate": 3.6029568450984817e-07,
      "loss": 0.8545,
      "step": 57413
    },
    {
      "epoch": 4.939688548567496,
      "grad_norm": 1.5,
      "learning_rate": 3.592700550501538e-07,
      "loss": 0.9631,
      "step": 57414
    },
    {
      "epoch": 4.939774584874817,
      "grad_norm": 1.21875,
      "learning_rate": 3.582458869423877e-07,
      "loss": 0.8918,
      "step": 57415
    },
    {
      "epoch": 4.939860621182139,
      "grad_norm": 1.0078125,
      "learning_rate": 3.572231801896031e-07,
      "loss": 0.5141,
      "step": 57416
    },
    {
      "epoch": 4.939946657489461,
      "grad_norm": 0.98046875,
      "learning_rate": 3.5620193479474206e-07,
      "loss": 1.224,
      "step": 57417
    },
    {
      "epoch": 4.940032693796782,
      "grad_norm": 2.546875,
      "learning_rate": 3.551821507608022e-07,
      "loss": 0.9152,
      "step": 57418
    },
    {
      "epoch": 4.940118730104104,
      "grad_norm": 25.25,
      "learning_rate": 3.5416382809083656e-07,
      "loss": 0.78,
      "step": 57419
    },
    {
      "epoch": 4.9402047664114255,
      "grad_norm": 1.21875,
      "learning_rate": 3.5314696678767633e-07,
      "loss": 1.0398,
      "step": 57420
    },
    {
      "epoch": 4.9402908027187475,
      "grad_norm": 1.1953125,
      "learning_rate": 3.5213156685448555e-07,
      "loss": 1.708,
      "step": 57421
    },
    {
      "epoch": 4.940376839026069,
      "grad_norm": 1.0390625,
      "learning_rate": 3.511176282940398e-07,
      "loss": 0.6331,
      "step": 57422
    },
    {
      "epoch": 4.940462875333391,
      "grad_norm": 0.73828125,
      "learning_rate": 3.5010515110950324e-07,
      "loss": 0.6699,
      "step": 57423
    },
    {
      "epoch": 4.940548911640713,
      "grad_norm": 1.1484375,
      "learning_rate": 3.4909413530365144e-07,
      "loss": 1.2958,
      "step": 57424
    },
    {
      "epoch": 4.940634947948034,
      "grad_norm": 9.4375,
      "learning_rate": 3.4808458087953743e-07,
      "loss": 1.2438,
      "step": 57425
    },
    {
      "epoch": 4.940720984255356,
      "grad_norm": 1.609375,
      "learning_rate": 3.470764878401034e-07,
      "loss": 1.5068,
      "step": 57426
    },
    {
      "epoch": 4.940807020562677,
      "grad_norm": 0.84375,
      "learning_rate": 3.4606985618829133e-07,
      "loss": 0.7962,
      "step": 57427
    },
    {
      "epoch": 4.940893056869999,
      "grad_norm": 1.0234375,
      "learning_rate": 3.450646859270434e-07,
      "loss": 1.2313,
      "step": 57428
    },
    {
      "epoch": 4.9409790931773205,
      "grad_norm": 3.390625,
      "learning_rate": 3.4406097705935723e-07,
      "loss": 1.1468,
      "step": 57429
    },
    {
      "epoch": 4.941065129484643,
      "grad_norm": 0.73828125,
      "learning_rate": 3.4305872958806384e-07,
      "loss": 0.5858,
      "step": 57430
    },
    {
      "epoch": 4.941151165791965,
      "grad_norm": 1.7109375,
      "learning_rate": 3.420579435161053e-07,
      "loss": 0.9324,
      "step": 57431
    },
    {
      "epoch": 4.941237202099286,
      "grad_norm": 1.140625,
      "learning_rate": 3.4105861884653476e-07,
      "loss": 0.6349,
      "step": 57432
    },
    {
      "epoch": 4.941323238406608,
      "grad_norm": 1.640625,
      "learning_rate": 3.400607555821278e-07,
      "loss": 1.4466,
      "step": 57433
    },
    {
      "epoch": 4.941409274713929,
      "grad_norm": 17.0,
      "learning_rate": 3.3906435372588197e-07,
      "loss": 0.8858,
      "step": 57434
    },
    {
      "epoch": 4.941495311021251,
      "grad_norm": 0.9296875,
      "learning_rate": 3.380694132807394e-07,
      "loss": 0.512,
      "step": 57435
    },
    {
      "epoch": 4.941581347328572,
      "grad_norm": 2752.0,
      "learning_rate": 3.370759342494756e-07,
      "loss": 1.6314,
      "step": 57436
    },
    {
      "epoch": 4.941667383635894,
      "grad_norm": 1.1640625,
      "learning_rate": 3.3608391663514373e-07,
      "loss": 1.1485,
      "step": 57437
    },
    {
      "epoch": 4.9417534199432165,
      "grad_norm": 6.8125,
      "learning_rate": 3.350933604405748e-07,
      "loss": 0.9726,
      "step": 57438
    },
    {
      "epoch": 4.941839456250538,
      "grad_norm": 0.90625,
      "learning_rate": 3.34104265668711e-07,
      "loss": 1.1384,
      "step": 57439
    },
    {
      "epoch": 4.94192549255786,
      "grad_norm": 1.2734375,
      "learning_rate": 3.331166323223833e-07,
      "loss": 0.5926,
      "step": 57440
    },
    {
      "epoch": 4.942011528865181,
      "grad_norm": 1.0078125,
      "learning_rate": 3.3213046040453385e-07,
      "loss": 1.0759,
      "step": 57441
    },
    {
      "epoch": 4.942097565172503,
      "grad_norm": 1.078125,
      "learning_rate": 3.3114574991799375e-07,
      "loss": 0.71,
      "step": 57442
    },
    {
      "epoch": 4.942183601479824,
      "grad_norm": 1.0390625,
      "learning_rate": 3.3016250086564946e-07,
      "loss": 0.6253,
      "step": 57443
    },
    {
      "epoch": 4.942269637787146,
      "grad_norm": 1.640625,
      "learning_rate": 3.2918071325044316e-07,
      "loss": 1.3457,
      "step": 57444
    },
    {
      "epoch": 4.942355674094468,
      "grad_norm": 1.2265625,
      "learning_rate": 3.2820038707520595e-07,
      "loss": 1.8477,
      "step": 57445
    },
    {
      "epoch": 4.9424417104017895,
      "grad_norm": 0.90234375,
      "learning_rate": 3.272215223427688e-07,
      "loss": 0.7731,
      "step": 57446
    },
    {
      "epoch": 4.9425277467091115,
      "grad_norm": 1.2265625,
      "learning_rate": 3.2624411905601837e-07,
      "loss": 1.3371,
      "step": 57447
    },
    {
      "epoch": 4.942613783016433,
      "grad_norm": 0.98828125,
      "learning_rate": 3.252681772178412e-07,
      "loss": 0.5771,
      "step": 57448
    },
    {
      "epoch": 4.942699819323755,
      "grad_norm": 764.0,
      "learning_rate": 3.242936968311239e-07,
      "loss": 1.0204,
      "step": 57449
    },
    {
      "epoch": 4.942785855631076,
      "grad_norm": 0.8671875,
      "learning_rate": 3.233206778985864e-07,
      "loss": 0.8238,
      "step": 57450
    },
    {
      "epoch": 4.942871891938398,
      "grad_norm": 2.59375,
      "learning_rate": 3.223491204232265e-07,
      "loss": 0.9342,
      "step": 57451
    },
    {
      "epoch": 4.94295792824572,
      "grad_norm": 1.0859375,
      "learning_rate": 3.2137902440776414e-07,
      "loss": 0.6277,
      "step": 57452
    },
    {
      "epoch": 4.943043964553041,
      "grad_norm": 1.578125,
      "learning_rate": 3.204103898550859e-07,
      "loss": 1.3909,
      "step": 57453
    },
    {
      "epoch": 4.943130000860363,
      "grad_norm": 1.2890625,
      "learning_rate": 3.194432167680228e-07,
      "loss": 1.6372,
      "step": 57454
    },
    {
      "epoch": 4.9432160371676845,
      "grad_norm": 1.15625,
      "learning_rate": 3.18477505149406e-07,
      "loss": 1.9309,
      "step": 57455
    },
    {
      "epoch": 4.943302073475007,
      "grad_norm": 8.25,
      "learning_rate": 3.17513255002122e-07,
      "loss": 0.9861,
      "step": 57456
    },
    {
      "epoch": 4.943388109782328,
      "grad_norm": 0.96484375,
      "learning_rate": 3.1655046632883546e-07,
      "loss": 0.6721,
      "step": 57457
    },
    {
      "epoch": 4.94347414608965,
      "grad_norm": 1.171875,
      "learning_rate": 3.1558913913254386e-07,
      "loss": 2.1592,
      "step": 57458
    },
    {
      "epoch": 4.943560182396972,
      "grad_norm": 1.2265625,
      "learning_rate": 3.146292734159117e-07,
      "loss": 1.4192,
      "step": 57459
    },
    {
      "epoch": 4.943646218704293,
      "grad_norm": 0.6484375,
      "learning_rate": 3.1367086918182574e-07,
      "loss": 0.6649,
      "step": 57460
    },
    {
      "epoch": 4.943732255011615,
      "grad_norm": 1.2890625,
      "learning_rate": 3.127139264330614e-07,
      "loss": 0.8961,
      "step": 57461
    },
    {
      "epoch": 4.943818291318936,
      "grad_norm": 1.3203125,
      "learning_rate": 3.1175844517244977e-07,
      "loss": 0.8107,
      "step": 57462
    },
    {
      "epoch": 4.943904327626258,
      "grad_norm": 6.6875,
      "learning_rate": 3.108044254027109e-07,
      "loss": 1.0346,
      "step": 57463
    },
    {
      "epoch": 4.94399036393358,
      "grad_norm": 1.4609375,
      "learning_rate": 3.098518671267314e-07,
      "loss": 1.8019,
      "step": 57464
    },
    {
      "epoch": 4.944076400240902,
      "grad_norm": 1.1484375,
      "learning_rate": 3.089007703472313e-07,
      "loss": 1.5283,
      "step": 57465
    },
    {
      "epoch": 4.944162436548224,
      "grad_norm": 1.734375,
      "learning_rate": 3.079511350669861e-07,
      "loss": 1.161,
      "step": 57466
    },
    {
      "epoch": 4.944248472855545,
      "grad_norm": 87.0,
      "learning_rate": 3.07002961288827e-07,
      "loss": 0.7898,
      "step": 57467
    },
    {
      "epoch": 4.944334509162867,
      "grad_norm": 0.91796875,
      "learning_rate": 3.060562490155294e-07,
      "loss": 0.6106,
      "step": 57468
    },
    {
      "epoch": 4.944420545470188,
      "grad_norm": 1.171875,
      "learning_rate": 3.0511099824975797e-07,
      "loss": 1.4845,
      "step": 57469
    },
    {
      "epoch": 4.94450658177751,
      "grad_norm": 0.8984375,
      "learning_rate": 3.041672089943992e-07,
      "loss": 0.7278,
      "step": 57470
    },
    {
      "epoch": 4.944592618084831,
      "grad_norm": 1.40625,
      "learning_rate": 3.032248812521732e-07,
      "loss": 0.7837,
      "step": 57471
    },
    {
      "epoch": 4.9446786543921535,
      "grad_norm": 2.015625,
      "learning_rate": 3.022840150257999e-07,
      "loss": 0.7213,
      "step": 57472
    },
    {
      "epoch": 4.9447646906994756,
      "grad_norm": 1.0546875,
      "learning_rate": 3.0134461031811055e-07,
      "loss": 0.5919,
      "step": 57473
    },
    {
      "epoch": 4.944850727006797,
      "grad_norm": 1.0546875,
      "learning_rate": 3.004066671317696e-07,
      "loss": 1.2799,
      "step": 57474
    },
    {
      "epoch": 4.944936763314119,
      "grad_norm": 1.0546875,
      "learning_rate": 2.994701854695525e-07,
      "loss": 0.8616,
      "step": 57475
    },
    {
      "epoch": 4.94502279962144,
      "grad_norm": 1.0546875,
      "learning_rate": 2.9853516533417944e-07,
      "loss": 0.8055,
      "step": 57476
    },
    {
      "epoch": 4.945108835928762,
      "grad_norm": 1.515625,
      "learning_rate": 2.9760160672842596e-07,
      "loss": 1.2649,
      "step": 57477
    },
    {
      "epoch": 4.945194872236083,
      "grad_norm": 1.1328125,
      "learning_rate": 2.9666950965501205e-07,
      "loss": 1.4207,
      "step": 57478
    },
    {
      "epoch": 4.945280908543405,
      "grad_norm": 1.21875,
      "learning_rate": 2.957388741166578e-07,
      "loss": 1.2084,
      "step": 57479
    },
    {
      "epoch": 4.945366944850727,
      "grad_norm": 1.234375,
      "learning_rate": 2.948097001160277e-07,
      "loss": 1.2415,
      "step": 57480
    },
    {
      "epoch": 4.9454529811580485,
      "grad_norm": 41.5,
      "learning_rate": 2.9388198765595284e-07,
      "loss": 1.1428,
      "step": 57481
    },
    {
      "epoch": 4.945539017465371,
      "grad_norm": 12.0,
      "learning_rate": 2.929557367390423e-07,
      "loss": 1.0922,
      "step": 57482
    },
    {
      "epoch": 4.945625053772692,
      "grad_norm": 1.28125,
      "learning_rate": 2.9203094736807155e-07,
      "loss": 1.4899,
      "step": 57483
    },
    {
      "epoch": 4.945711090080014,
      "grad_norm": 1.3984375,
      "learning_rate": 2.9110761954570517e-07,
      "loss": 0.8924,
      "step": 57484
    },
    {
      "epoch": 4.945797126387335,
      "grad_norm": 1.4921875,
      "learning_rate": 2.901857532746632e-07,
      "loss": 1.6334,
      "step": 57485
    },
    {
      "epoch": 4.945883162694657,
      "grad_norm": 1.1640625,
      "learning_rate": 2.8926534855766575e-07,
      "loss": 0.9052,
      "step": 57486
    },
    {
      "epoch": 4.945969199001979,
      "grad_norm": 584.0,
      "learning_rate": 2.8834640539737724e-07,
      "loss": 0.8804,
      "step": 57487
    },
    {
      "epoch": 4.9460552353093,
      "grad_norm": 1.078125,
      "learning_rate": 2.8742892379646226e-07,
      "loss": 1.1419,
      "step": 57488
    },
    {
      "epoch": 4.946141271616622,
      "grad_norm": 1.0859375,
      "learning_rate": 2.8651290375764083e-07,
      "loss": 0.8234,
      "step": 57489
    },
    {
      "epoch": 4.946227307923944,
      "grad_norm": 0.83984375,
      "learning_rate": 2.855983452835775e-07,
      "loss": 0.8338,
      "step": 57490
    },
    {
      "epoch": 4.946313344231266,
      "grad_norm": 0.75,
      "learning_rate": 2.8468524837693687e-07,
      "loss": 0.8346,
      "step": 57491
    },
    {
      "epoch": 4.946399380538587,
      "grad_norm": 1.578125,
      "learning_rate": 2.837736130404389e-07,
      "loss": 1.1306,
      "step": 57492
    },
    {
      "epoch": 4.946485416845909,
      "grad_norm": 752.0,
      "learning_rate": 2.8286343927669264e-07,
      "loss": 1.1735,
      "step": 57493
    },
    {
      "epoch": 4.946571453153231,
      "grad_norm": 1.0859375,
      "learning_rate": 2.819547270883627e-07,
      "loss": 1.1071,
      "step": 57494
    },
    {
      "epoch": 4.946657489460552,
      "grad_norm": 1.140625,
      "learning_rate": 2.81047476478169e-07,
      "loss": 0.9297,
      "step": 57495
    },
    {
      "epoch": 4.946743525767874,
      "grad_norm": 9.5,
      "learning_rate": 2.801416874486651e-07,
      "loss": 0.933,
      "step": 57496
    },
    {
      "epoch": 4.946829562075195,
      "grad_norm": 17.75,
      "learning_rate": 2.7923736000257107e-07,
      "loss": 1.1348,
      "step": 57497
    },
    {
      "epoch": 4.9469155983825175,
      "grad_norm": 3.40625,
      "learning_rate": 2.783344941425514e-07,
      "loss": 1.3831,
      "step": 57498
    },
    {
      "epoch": 4.947001634689839,
      "grad_norm": 1.0078125,
      "learning_rate": 2.774330898712152e-07,
      "loss": 1.0952,
      "step": 57499
    },
    {
      "epoch": 4.947087670997161,
      "grad_norm": 1.4296875,
      "learning_rate": 2.765331471911714e-07,
      "loss": 1.5412,
      "step": 57500
    },
    {
      "epoch": 4.947173707304483,
      "grad_norm": 1.0859375,
      "learning_rate": 2.756346661050846e-07,
      "loss": 1.029,
      "step": 57501
    },
    {
      "epoch": 4.947259743611804,
      "grad_norm": 1.0,
      "learning_rate": 2.747376466155638e-07,
      "loss": 0.8707,
      "step": 57502
    },
    {
      "epoch": 4.947345779919126,
      "grad_norm": 1.09375,
      "learning_rate": 2.73842088725218e-07,
      "loss": 1.6406,
      "step": 57503
    },
    {
      "epoch": 4.947431816226447,
      "grad_norm": 1.1171875,
      "learning_rate": 2.729479924367673e-07,
      "loss": 0.7906,
      "step": 57504
    },
    {
      "epoch": 4.947517852533769,
      "grad_norm": 1.015625,
      "learning_rate": 2.7205535775270964e-07,
      "loss": 1.055,
      "step": 57505
    },
    {
      "epoch": 4.9476038888410905,
      "grad_norm": 1.03125,
      "learning_rate": 2.711641846756541e-07,
      "loss": 0.72,
      "step": 57506
    },
    {
      "epoch": 4.9476899251484125,
      "grad_norm": 1.4375,
      "learning_rate": 2.702744732083207e-07,
      "loss": 1.594,
      "step": 57507
    },
    {
      "epoch": 4.947775961455735,
      "grad_norm": 1.0234375,
      "learning_rate": 2.693862233532074e-07,
      "loss": 1.6559,
      "step": 57508
    },
    {
      "epoch": 4.947861997763056,
      "grad_norm": 1.34375,
      "learning_rate": 2.684994351129788e-07,
      "loss": 1.3374,
      "step": 57509
    },
    {
      "epoch": 4.947948034070378,
      "grad_norm": 1.296875,
      "learning_rate": 2.676141084901884e-07,
      "loss": 0.9146,
      "step": 57510
    },
    {
      "epoch": 4.948034070377699,
      "grad_norm": 1.203125,
      "learning_rate": 2.6673024348738973e-07,
      "loss": 1.3309,
      "step": 57511
    },
    {
      "epoch": 4.948120106685021,
      "grad_norm": 1.1875,
      "learning_rate": 2.6584784010724725e-07,
      "loss": 1.4499,
      "step": 57512
    },
    {
      "epoch": 4.948206142992342,
      "grad_norm": 0.77734375,
      "learning_rate": 2.6496689835231456e-07,
      "loss": 0.5572,
      "step": 57513
    },
    {
      "epoch": 4.948292179299664,
      "grad_norm": 5152.0,
      "learning_rate": 2.6408741822514515e-07,
      "loss": 0.6987,
      "step": 57514
    },
    {
      "epoch": 4.948378215606986,
      "grad_norm": 1.34375,
      "learning_rate": 2.63209399728348e-07,
      "loss": 0.994,
      "step": 57515
    },
    {
      "epoch": 4.948464251914308,
      "grad_norm": 1.0546875,
      "learning_rate": 2.6233284286447667e-07,
      "loss": 1.1571,
      "step": 57516
    },
    {
      "epoch": 4.94855028822163,
      "grad_norm": 0.953125,
      "learning_rate": 2.6145774763602915e-07,
      "loss": 1.7084,
      "step": 57517
    },
    {
      "epoch": 4.948636324528951,
      "grad_norm": 1.390625,
      "learning_rate": 2.6058411404566997e-07,
      "loss": 1.291,
      "step": 57518
    },
    {
      "epoch": 4.948722360836273,
      "grad_norm": 1.4453125,
      "learning_rate": 2.597119420958971e-07,
      "loss": 0.8799,
      "step": 57519
    },
    {
      "epoch": 4.948808397143594,
      "grad_norm": 1.078125,
      "learning_rate": 2.588412317892641e-07,
      "loss": 1.0384,
      "step": 57520
    },
    {
      "epoch": 4.948894433450916,
      "grad_norm": 1.3046875,
      "learning_rate": 2.5797198312838e-07,
      "loss": 1.1512,
      "step": 57521
    },
    {
      "epoch": 4.948980469758238,
      "grad_norm": 1.1484375,
      "learning_rate": 2.5710419611568725e-07,
      "loss": 1.3496,
      "step": 57522
    },
    {
      "epoch": 4.949066506065559,
      "grad_norm": 11.25,
      "learning_rate": 2.562378707537394e-07,
      "loss": 0.8386,
      "step": 57523
    },
    {
      "epoch": 4.9491525423728815,
      "grad_norm": 1.4453125,
      "learning_rate": 2.553730070451454e-07,
      "loss": 1.1317,
      "step": 57524
    },
    {
      "epoch": 4.949238578680203,
      "grad_norm": 1.1875,
      "learning_rate": 2.545096049924034e-07,
      "loss": 0.7874,
      "step": 57525
    },
    {
      "epoch": 4.949324614987525,
      "grad_norm": 0.90625,
      "learning_rate": 2.5364766459801124e-07,
      "loss": 1.0707,
      "step": 57526
    },
    {
      "epoch": 4.949410651294847,
      "grad_norm": 1.578125,
      "learning_rate": 2.5278718586446704e-07,
      "loss": 0.9101,
      "step": 57527
    },
    {
      "epoch": 4.949496687602168,
      "grad_norm": 272.0,
      "learning_rate": 2.519281687943797e-07,
      "loss": 0.9034,
      "step": 57528
    },
    {
      "epoch": 4.94958272390949,
      "grad_norm": 1.8125,
      "learning_rate": 2.5107061339013637e-07,
      "loss": 1.0912,
      "step": 57529
    },
    {
      "epoch": 4.949668760216811,
      "grad_norm": 1.4453125,
      "learning_rate": 2.5021451965440145e-07,
      "loss": 1.0873,
      "step": 57530
    },
    {
      "epoch": 4.949754796524133,
      "grad_norm": 1.140625,
      "learning_rate": 2.4935988758956195e-07,
      "loss": 0.9183,
      "step": 57531
    },
    {
      "epoch": 4.949840832831455,
      "grad_norm": 44.0,
      "learning_rate": 2.485067171981159e-07,
      "loss": 1.614,
      "step": 57532
    },
    {
      "epoch": 4.9499268691387766,
      "grad_norm": 1.34375,
      "learning_rate": 2.476550084826168e-07,
      "loss": 0.775,
      "step": 57533
    },
    {
      "epoch": 4.950012905446099,
      "grad_norm": 1.0859375,
      "learning_rate": 2.468047614455071e-07,
      "loss": 0.8083,
      "step": 57534
    },
    {
      "epoch": 4.95009894175342,
      "grad_norm": 1.0546875,
      "learning_rate": 2.459559760893404e-07,
      "loss": 1.5545,
      "step": 57535
    },
    {
      "epoch": 4.950184978060742,
      "grad_norm": 3.125,
      "learning_rate": 2.451086524165036e-07,
      "loss": 0.9599,
      "step": 57536
    },
    {
      "epoch": 4.950271014368063,
      "grad_norm": 0.92578125,
      "learning_rate": 2.4426279042960574e-07,
      "loss": 0.903,
      "step": 57537
    },
    {
      "epoch": 4.950357050675385,
      "grad_norm": 1.4296875,
      "learning_rate": 2.434183901309783e-07,
      "loss": 1.6418,
      "step": 57538
    },
    {
      "epoch": 4.950443086982707,
      "grad_norm": 0.8125,
      "learning_rate": 2.4257545152317483e-07,
      "loss": 0.8567,
      "step": 57539
    },
    {
      "epoch": 4.950529123290028,
      "grad_norm": 0.99609375,
      "learning_rate": 2.417339746086378e-07,
      "loss": 1.4724,
      "step": 57540
    },
    {
      "epoch": 4.95061515959735,
      "grad_norm": 1.625,
      "learning_rate": 2.4089395938980964e-07,
      "loss": 0.9048,
      "step": 57541
    },
    {
      "epoch": 4.950701195904672,
      "grad_norm": 1.2890625,
      "learning_rate": 2.400554058691884e-07,
      "loss": 1.4131,
      "step": 57542
    },
    {
      "epoch": 4.950787232211994,
      "grad_norm": 0.94140625,
      "learning_rate": 2.3921831404921655e-07,
      "loss": 0.7059,
      "step": 57543
    },
    {
      "epoch": 4.950873268519315,
      "grad_norm": 1.0546875,
      "learning_rate": 2.3838268393228114e-07,
      "loss": 1.2794,
      "step": 57544
    },
    {
      "epoch": 4.950959304826637,
      "grad_norm": 1.40625,
      "learning_rate": 2.375485155209356e-07,
      "loss": 1.3647,
      "step": 57545
    },
    {
      "epoch": 4.951045341133959,
      "grad_norm": 0.8203125,
      "learning_rate": 2.3671580881756693e-07,
      "loss": 0.6471,
      "step": 57546
    },
    {
      "epoch": 4.95113137744128,
      "grad_norm": 2.578125,
      "learning_rate": 2.3588456382456214e-07,
      "loss": 1.0707,
      "step": 57547
    },
    {
      "epoch": 4.951217413748602,
      "grad_norm": 1.03125,
      "learning_rate": 2.350547805444192e-07,
      "loss": 0.9146,
      "step": 57548
    },
    {
      "epoch": 4.951303450055923,
      "grad_norm": 1.328125,
      "learning_rate": 2.3422645897958062e-07,
      "loss": 1.5889,
      "step": 57549
    },
    {
      "epoch": 4.9513894863632455,
      "grad_norm": 1.03125,
      "learning_rate": 2.3339959913237784e-07,
      "loss": 1.1076,
      "step": 57550
    },
    {
      "epoch": 4.951475522670567,
      "grad_norm": 0.88671875,
      "learning_rate": 2.325742010053089e-07,
      "loss": 0.8721,
      "step": 57551
    },
    {
      "epoch": 4.951561558977889,
      "grad_norm": 1.4140625,
      "learning_rate": 2.3175026460070526e-07,
      "loss": 1.0824,
      "step": 57552
    },
    {
      "epoch": 4.951647595285211,
      "grad_norm": 1.4453125,
      "learning_rate": 2.309277899211204e-07,
      "loss": 0.8451,
      "step": 57553
    },
    {
      "epoch": 4.951733631592532,
      "grad_norm": 1.046875,
      "learning_rate": 2.3010677696883032e-07,
      "loss": 0.8503,
      "step": 57554
    },
    {
      "epoch": 4.951819667899854,
      "grad_norm": 1.328125,
      "learning_rate": 2.2928722574627748e-07,
      "loss": 1.0361,
      "step": 57555
    },
    {
      "epoch": 4.951905704207175,
      "grad_norm": 0.7890625,
      "learning_rate": 2.2846913625584887e-07,
      "loss": 0.8543,
      "step": 57556
    },
    {
      "epoch": 4.951991740514497,
      "grad_norm": 1.3125,
      "learning_rate": 2.2765250849993147e-07,
      "loss": 1.2566,
      "step": 57557
    },
    {
      "epoch": 4.9520777768218185,
      "grad_norm": 1.125,
      "learning_rate": 2.2683734248096776e-07,
      "loss": 1.0537,
      "step": 57558
    },
    {
      "epoch": 4.952163813129141,
      "grad_norm": 0.9453125,
      "learning_rate": 2.2602363820128925e-07,
      "loss": 0.7596,
      "step": 57559
    },
    {
      "epoch": 4.952249849436463,
      "grad_norm": 1.109375,
      "learning_rate": 2.2521139566328286e-07,
      "loss": 1.3001,
      "step": 57560
    },
    {
      "epoch": 4.952335885743784,
      "grad_norm": 1.140625,
      "learning_rate": 2.2440061486933562e-07,
      "loss": 1.1377,
      "step": 57561
    },
    {
      "epoch": 4.952421922051106,
      "grad_norm": 66.5,
      "learning_rate": 2.2359129582183445e-07,
      "loss": 1.2909,
      "step": 57562
    },
    {
      "epoch": 4.952507958358427,
      "grad_norm": 1.3671875,
      "learning_rate": 2.227834385231109e-07,
      "loss": 1.1488,
      "step": 57563
    },
    {
      "epoch": 4.952593994665749,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2197704297555188e-07,
      "loss": 1.2767,
      "step": 57564
    },
    {
      "epoch": 4.95268003097307,
      "grad_norm": 1.8125,
      "learning_rate": 2.211721091815444e-07,
      "loss": 1.5343,
      "step": 57565
    },
    {
      "epoch": 4.952766067280392,
      "grad_norm": 0.9140625,
      "learning_rate": 2.2036863714336442e-07,
      "loss": 0.627,
      "step": 57566
    },
    {
      "epoch": 4.952852103587714,
      "grad_norm": 0.921875,
      "learning_rate": 2.195666268633989e-07,
      "loss": 0.7385,
      "step": 57567
    },
    {
      "epoch": 4.952938139895036,
      "grad_norm": 1.2265625,
      "learning_rate": 2.1876607834397932e-07,
      "loss": 1.7935,
      "step": 57568
    },
    {
      "epoch": 4.953024176202358,
      "grad_norm": 1.484375,
      "learning_rate": 2.1796699158749266e-07,
      "loss": 0.9247,
      "step": 57569
    },
    {
      "epoch": 4.953110212509679,
      "grad_norm": 36.25,
      "learning_rate": 2.171693665962704e-07,
      "loss": 1.2027,
      "step": 57570
    },
    {
      "epoch": 4.953196248817001,
      "grad_norm": 0.9765625,
      "learning_rate": 2.16373203372644e-07,
      "loss": 0.9245,
      "step": 57571
    },
    {
      "epoch": 4.953282285124322,
      "grad_norm": 1.0078125,
      "learning_rate": 2.155785019188894e-07,
      "loss": 1.0361,
      "step": 57572
    },
    {
      "epoch": 4.953368321431644,
      "grad_norm": 0.77734375,
      "learning_rate": 2.1478526223733807e-07,
      "loss": 1.0153,
      "step": 57573
    },
    {
      "epoch": 4.953454357738966,
      "grad_norm": 1.078125,
      "learning_rate": 2.13993484330377e-07,
      "loss": 0.7039,
      "step": 57574
    },
    {
      "epoch": 4.953540394046287,
      "grad_norm": 18.75,
      "learning_rate": 2.1320316820022668e-07,
      "loss": 0.9082,
      "step": 57575
    },
    {
      "epoch": 4.9536264303536095,
      "grad_norm": 0.88671875,
      "learning_rate": 2.1241431384932953e-07,
      "loss": 1.4513,
      "step": 57576
    },
    {
      "epoch": 4.953712466660931,
      "grad_norm": 1.0703125,
      "learning_rate": 2.11626921279906e-07,
      "loss": 0.8637,
      "step": 57577
    },
    {
      "epoch": 4.953798502968253,
      "grad_norm": 1.046875,
      "learning_rate": 2.1084099049423212e-07,
      "loss": 1.0358,
      "step": 57578
    },
    {
      "epoch": 4.953884539275574,
      "grad_norm": 0.953125,
      "learning_rate": 2.1005652149463928e-07,
      "loss": 1.4523,
      "step": 57579
    },
    {
      "epoch": 4.953970575582896,
      "grad_norm": 1.1875,
      "learning_rate": 2.0927351428345898e-07,
      "loss": 1.2085,
      "step": 57580
    },
    {
      "epoch": 4.954056611890218,
      "grad_norm": 0.75,
      "learning_rate": 2.0849196886291167e-07,
      "loss": 0.6378,
      "step": 57581
    },
    {
      "epoch": 4.954142648197539,
      "grad_norm": 1.0234375,
      "learning_rate": 2.0771188523538432e-07,
      "loss": 1.0554,
      "step": 57582
    },
    {
      "epoch": 4.954228684504861,
      "grad_norm": 2.59375,
      "learning_rate": 2.0693326340304185e-07,
      "loss": 0.6323,
      "step": 57583
    },
    {
      "epoch": 4.9543147208121825,
      "grad_norm": 1.234375,
      "learning_rate": 2.0615610336821577e-07,
      "loss": 1.6073,
      "step": 57584
    },
    {
      "epoch": 4.954400757119505,
      "grad_norm": 0.80078125,
      "learning_rate": 2.0538040513323753e-07,
      "loss": 0.5563,
      "step": 57585
    },
    {
      "epoch": 4.954486793426826,
      "grad_norm": 1.1484375,
      "learning_rate": 2.0460616870027205e-07,
      "loss": 0.6178,
      "step": 57586
    },
    {
      "epoch": 4.954572829734148,
      "grad_norm": 1.6953125,
      "learning_rate": 2.038333940715953e-07,
      "loss": 1.0534,
      "step": 57587
    },
    {
      "epoch": 4.95465886604147,
      "grad_norm": 1.0703125,
      "learning_rate": 2.0306208124953874e-07,
      "loss": 0.9685,
      "step": 57588
    },
    {
      "epoch": 4.954744902348791,
      "grad_norm": 1.2890625,
      "learning_rate": 2.022922302363228e-07,
      "loss": 1.0534,
      "step": 57589
    },
    {
      "epoch": 4.954830938656113,
      "grad_norm": 50.5,
      "learning_rate": 2.015238410342235e-07,
      "loss": 0.9142,
      "step": 57590
    },
    {
      "epoch": 4.954916974963434,
      "grad_norm": 1.6796875,
      "learning_rate": 2.007569136454057e-07,
      "loss": 0.7696,
      "step": 57591
    },
    {
      "epoch": 4.955003011270756,
      "grad_norm": 1.359375,
      "learning_rate": 1.9999144807220092e-07,
      "loss": 1.1685,
      "step": 57592
    },
    {
      "epoch": 4.9550890475780776,
      "grad_norm": 1.1640625,
      "learning_rate": 1.992274443168296e-07,
      "loss": 0.8828,
      "step": 57593
    },
    {
      "epoch": 4.9551750838854,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9846490238145664e-07,
      "loss": 0.7051,
      "step": 57594
    },
    {
      "epoch": 4.955261120192722,
      "grad_norm": 1.1953125,
      "learning_rate": 1.9770382226841354e-07,
      "loss": 1.0699,
      "step": 57595
    },
    {
      "epoch": 4.955347156500043,
      "grad_norm": 9.125,
      "learning_rate": 1.9694420397986522e-07,
      "loss": 0.8439,
      "step": 57596
    },
    {
      "epoch": 4.955433192807365,
      "grad_norm": 1.1875,
      "learning_rate": 1.9618604751808767e-07,
      "loss": 0.987,
      "step": 57597
    },
    {
      "epoch": 4.955519229114686,
      "grad_norm": 0.890625,
      "learning_rate": 1.9542935288519025e-07,
      "loss": 0.6675,
      "step": 57598
    },
    {
      "epoch": 4.955605265422008,
      "grad_norm": 0.8828125,
      "learning_rate": 1.946741200835045e-07,
      "loss": 0.8517,
      "step": 57599
    },
    {
      "epoch": 4.955691301729329,
      "grad_norm": 2.1875,
      "learning_rate": 1.939203491151953e-07,
      "loss": 1.207,
      "step": 57600
    },
    {
      "epoch": 4.955777338036651,
      "grad_norm": 1.140625,
      "learning_rate": 1.931680399824276e-07,
      "loss": 0.8051,
      "step": 57601
    },
    {
      "epoch": 4.9558633743439735,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9241719268742185e-07,
      "loss": 1.0533,
      "step": 57602
    },
    {
      "epoch": 4.955949410651295,
      "grad_norm": 1.3125,
      "learning_rate": 1.9166780723245404e-07,
      "loss": 1.4726,
      "step": 57603
    },
    {
      "epoch": 4.956035446958617,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9091988361957802e-07,
      "loss": 0.916,
      "step": 57604
    },
    {
      "epoch": 4.956121483265938,
      "grad_norm": 1.28125,
      "learning_rate": 1.901734218511253e-07,
      "loss": 1.2907,
      "step": 57605
    },
    {
      "epoch": 4.95620751957326,
      "grad_norm": 1.140625,
      "learning_rate": 1.8942842192914979e-07,
      "loss": 0.4392,
      "step": 57606
    },
    {
      "epoch": 4.956293555880581,
      "grad_norm": 1.1953125,
      "learning_rate": 1.8868488385592743e-07,
      "loss": 1.0128,
      "step": 57607
    },
    {
      "epoch": 4.956379592187903,
      "grad_norm": 1.0390625,
      "learning_rate": 1.8794280763356764e-07,
      "loss": 1.1765,
      "step": 57608
    },
    {
      "epoch": 4.956465628495225,
      "grad_norm": 1.9375,
      "learning_rate": 1.872021932642909e-07,
      "loss": 0.798,
      "step": 57609
    },
    {
      "epoch": 4.9565516648025465,
      "grad_norm": 0.98046875,
      "learning_rate": 1.8646304075020658e-07,
      "loss": 1.0091,
      "step": 57610
    },
    {
      "epoch": 4.956637701109869,
      "grad_norm": 1.1328125,
      "learning_rate": 1.857253500935352e-07,
      "loss": 0.7412,
      "step": 57611
    },
    {
      "epoch": 4.95672373741719,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8498912129638612e-07,
      "loss": 0.625,
      "step": 57612
    },
    {
      "epoch": 4.956809773724512,
      "grad_norm": 2.046875,
      "learning_rate": 1.842543543609798e-07,
      "loss": 1.0099,
      "step": 57613
    },
    {
      "epoch": 4.956895810031833,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8352104928937018e-07,
      "loss": 0.9339,
      "step": 57614
    },
    {
      "epoch": 4.956981846339155,
      "grad_norm": 1.1328125,
      "learning_rate": 1.827892060837777e-07,
      "loss": 0.7617,
      "step": 57615
    },
    {
      "epoch": 4.957067882646477,
      "grad_norm": 1.0234375,
      "learning_rate": 1.8205882474631176e-07,
      "loss": 1.7035,
      "step": 57616
    },
    {
      "epoch": 4.957153918953798,
      "grad_norm": 1.140625,
      "learning_rate": 1.813299052791373e-07,
      "loss": 1.1131,
      "step": 57617
    },
    {
      "epoch": 4.95723995526112,
      "grad_norm": 1.1875,
      "learning_rate": 1.8060244768430823e-07,
      "loss": 1.1465,
      "step": 57618
    },
    {
      "epoch": 4.957325991568442,
      "grad_norm": 1.34375,
      "learning_rate": 1.79876451964045e-07,
      "loss": 1.2149,
      "step": 57619
    },
    {
      "epoch": 4.957412027875764,
      "grad_norm": 54.75,
      "learning_rate": 1.7915191812045707e-07,
      "loss": 1.1629,
      "step": 57620
    },
    {
      "epoch": 4.957498064183085,
      "grad_norm": 1.046875,
      "learning_rate": 1.784288461555983e-07,
      "loss": 0.78,
      "step": 57621
    },
    {
      "epoch": 4.957584100490407,
      "grad_norm": 0.9921875,
      "learning_rate": 1.7770723607163363e-07,
      "loss": 0.8412,
      "step": 57622
    },
    {
      "epoch": 4.957670136797729,
      "grad_norm": 1.15625,
      "learning_rate": 1.7698708787067252e-07,
      "loss": 0.3505,
      "step": 57623
    },
    {
      "epoch": 4.95775617310505,
      "grad_norm": 1.875,
      "learning_rate": 1.7626840155482437e-07,
      "loss": 1.413,
      "step": 57624
    },
    {
      "epoch": 4.957842209412372,
      "grad_norm": 1.1875,
      "learning_rate": 1.755511771261986e-07,
      "loss": 1.2107,
      "step": 57625
    },
    {
      "epoch": 4.957928245719693,
      "grad_norm": 1.1484375,
      "learning_rate": 1.7483541458684916e-07,
      "loss": 0.7155,
      "step": 57626
    },
    {
      "epoch": 4.958014282027015,
      "grad_norm": 1.8671875,
      "learning_rate": 1.741211139388854e-07,
      "loss": 1.1065,
      "step": 57627
    },
    {
      "epoch": 4.958100318334337,
      "grad_norm": 1.2890625,
      "learning_rate": 1.734082751844168e-07,
      "loss": 1.5347,
      "step": 57628
    },
    {
      "epoch": 4.958186354641659,
      "grad_norm": 1.1953125,
      "learning_rate": 1.726968983254973e-07,
      "loss": 0.8849,
      "step": 57629
    },
    {
      "epoch": 4.958272390948981,
      "grad_norm": 1.234375,
      "learning_rate": 1.7198698336423625e-07,
      "loss": 0.9733,
      "step": 57630
    },
    {
      "epoch": 4.958358427256302,
      "grad_norm": 1.03125,
      "learning_rate": 1.712785303027431e-07,
      "loss": 0.9834,
      "step": 57631
    },
    {
      "epoch": 4.958444463563624,
      "grad_norm": 1.078125,
      "learning_rate": 1.7057153914301627e-07,
      "loss": 1.0313,
      "step": 57632
    },
    {
      "epoch": 4.958530499870945,
      "grad_norm": 1.0078125,
      "learning_rate": 1.6986600988716516e-07,
      "loss": 0.5081,
      "step": 57633
    },
    {
      "epoch": 4.958616536178267,
      "grad_norm": 1.1484375,
      "learning_rate": 1.6916194253724371e-07,
      "loss": 1.4794,
      "step": 57634
    },
    {
      "epoch": 4.958702572485588,
      "grad_norm": 1592.0,
      "learning_rate": 1.684593370953058e-07,
      "loss": 1.1808,
      "step": 57635
    },
    {
      "epoch": 4.9587886087929105,
      "grad_norm": 1.3515625,
      "learning_rate": 1.677581935634054e-07,
      "loss": 1.5886,
      "step": 57636
    },
    {
      "epoch": 4.958874645100233,
      "grad_norm": 1.09375,
      "learning_rate": 1.6705851194359633e-07,
      "loss": 1.2078,
      "step": 57637
    },
    {
      "epoch": 4.958960681407554,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6636029223793258e-07,
      "loss": 0.4501,
      "step": 57638
    },
    {
      "epoch": 4.959046717714876,
      "grad_norm": 1.59375,
      "learning_rate": 1.6566353444846805e-07,
      "loss": 1.764,
      "step": 57639
    },
    {
      "epoch": 4.959132754022197,
      "grad_norm": 1.3125,
      "learning_rate": 1.6496823857720111e-07,
      "loss": 1.2693,
      "step": 57640
    },
    {
      "epoch": 4.959218790329519,
      "grad_norm": 1.15625,
      "learning_rate": 1.6427440462624123e-07,
      "loss": 1.1263,
      "step": 57641
    },
    {
      "epoch": 4.959304826636841,
      "grad_norm": 0.9375,
      "learning_rate": 1.6358203259753124e-07,
      "loss": 0.5846,
      "step": 57642
    },
    {
      "epoch": 4.959390862944162,
      "grad_norm": 49.75,
      "learning_rate": 1.628911224931251e-07,
      "loss": 1.2347,
      "step": 57643
    },
    {
      "epoch": 4.959476899251484,
      "grad_norm": 1.1953125,
      "learning_rate": 1.622016743150212e-07,
      "loss": 0.9445,
      "step": 57644
    },
    {
      "epoch": 4.959562935558806,
      "grad_norm": 400.0,
      "learning_rate": 1.6151368806532896e-07,
      "loss": 0.6125,
      "step": 57645
    },
    {
      "epoch": 4.959648971866128,
      "grad_norm": 1.0,
      "learning_rate": 1.6082716374593575e-07,
      "loss": 0.8702,
      "step": 57646
    },
    {
      "epoch": 4.959735008173449,
      "grad_norm": 1.296875,
      "learning_rate": 1.6014210135895102e-07,
      "loss": 1.0279,
      "step": 57647
    },
    {
      "epoch": 4.959821044480771,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5945850090631764e-07,
      "loss": 0.5892,
      "step": 57648
    },
    {
      "epoch": 4.959907080788093,
      "grad_norm": 0.91796875,
      "learning_rate": 1.5877636239008952e-07,
      "loss": 0.6727,
      "step": 57649
    },
    {
      "epoch": 4.959993117095414,
      "grad_norm": 1.0859375,
      "learning_rate": 1.5809568581220955e-07,
      "loss": 1.5576,
      "step": 57650
    },
    {
      "epoch": 4.960079153402736,
      "grad_norm": 0.90625,
      "learning_rate": 1.5741647117467617e-07,
      "loss": 1.5031,
      "step": 57651
    },
    {
      "epoch": 4.960165189710057,
      "grad_norm": 1.1640625,
      "learning_rate": 1.5673871847948772e-07,
      "loss": 0.9122,
      "step": 57652
    },
    {
      "epoch": 4.9602512260173794,
      "grad_norm": 1.1171875,
      "learning_rate": 1.5606242772858714e-07,
      "loss": 1.0821,
      "step": 57653
    },
    {
      "epoch": 4.9603372623247015,
      "grad_norm": 1.75,
      "learning_rate": 1.553875989240283e-07,
      "loss": 1.0912,
      "step": 57654
    },
    {
      "epoch": 4.960423298632023,
      "grad_norm": 1.296875,
      "learning_rate": 1.5471423206775415e-07,
      "loss": 1.4128,
      "step": 57655
    },
    {
      "epoch": 4.960509334939345,
      "grad_norm": 1.4765625,
      "learning_rate": 1.5404232716170753e-07,
      "loss": 0.7058,
      "step": 57656
    },
    {
      "epoch": 4.960595371246666,
      "grad_norm": 1.40625,
      "learning_rate": 1.5337188420788683e-07,
      "loss": 1.2933,
      "step": 57657
    },
    {
      "epoch": 4.960681407553988,
      "grad_norm": 1.1875,
      "learning_rate": 1.5270290320823498e-07,
      "loss": 1.0102,
      "step": 57658
    },
    {
      "epoch": 4.960767443861309,
      "grad_norm": 0.95703125,
      "learning_rate": 1.5203538416469488e-07,
      "loss": 0.821,
      "step": 57659
    },
    {
      "epoch": 4.960853480168631,
      "grad_norm": 1.3046875,
      "learning_rate": 1.5136932707926487e-07,
      "loss": 1.2998,
      "step": 57660
    },
    {
      "epoch": 4.960939516475953,
      "grad_norm": 1.3046875,
      "learning_rate": 1.507047319538324e-07,
      "loss": 0.9794,
      "step": 57661
    },
    {
      "epoch": 4.9610255527832745,
      "grad_norm": 1.328125,
      "learning_rate": 1.500415987903958e-07,
      "loss": 0.9506,
      "step": 57662
    },
    {
      "epoch": 4.961111589090597,
      "grad_norm": 1.5,
      "learning_rate": 1.4937992759089802e-07,
      "loss": 1.4986,
      "step": 57663
    },
    {
      "epoch": 4.961197625397918,
      "grad_norm": 1.140625,
      "learning_rate": 1.487197183571709e-07,
      "loss": 1.5071,
      "step": 57664
    },
    {
      "epoch": 4.96128366170524,
      "grad_norm": 0.765625,
      "learning_rate": 1.4806097109126837e-07,
      "loss": 0.529,
      "step": 57665
    },
    {
      "epoch": 4.961369698012561,
      "grad_norm": 1.4296875,
      "learning_rate": 1.4740368579507778e-07,
      "loss": 0.7387,
      "step": 57666
    },
    {
      "epoch": 4.961455734319883,
      "grad_norm": 1.3984375,
      "learning_rate": 1.4674786247054206e-07,
      "loss": 1.2935,
      "step": 57667
    },
    {
      "epoch": 4.961541770627205,
      "grad_norm": 1.40625,
      "learning_rate": 1.4609350111949306e-07,
      "loss": 0.9458,
      "step": 57668
    },
    {
      "epoch": 4.961627806934526,
      "grad_norm": 1.03125,
      "learning_rate": 1.4544060174392915e-07,
      "loss": 1.1616,
      "step": 57669
    },
    {
      "epoch": 4.961713843241848,
      "grad_norm": 1.015625,
      "learning_rate": 1.4478916434573775e-07,
      "loss": 1.1796,
      "step": 57670
    },
    {
      "epoch": 4.96179987954917,
      "grad_norm": 1.265625,
      "learning_rate": 1.4413918892686174e-07,
      "loss": 1.7038,
      "step": 57671
    },
    {
      "epoch": 4.961885915856492,
      "grad_norm": 1.078125,
      "learning_rate": 1.4349067548913298e-07,
      "loss": 1.4352,
      "step": 57672
    },
    {
      "epoch": 4.961971952163813,
      "grad_norm": 0.94140625,
      "learning_rate": 1.4284362403443884e-07,
      "loss": 1.0316,
      "step": 57673
    },
    {
      "epoch": 4.962057988471135,
      "grad_norm": 1.5390625,
      "learning_rate": 1.4219803456472224e-07,
      "loss": 0.9596,
      "step": 57674
    },
    {
      "epoch": 4.962144024778457,
      "grad_norm": 1.2421875,
      "learning_rate": 1.4155390708187056e-07,
      "loss": 0.9433,
      "step": 57675
    },
    {
      "epoch": 4.962230061085778,
      "grad_norm": 1.2421875,
      "learning_rate": 1.4091124158777113e-07,
      "loss": 0.7979,
      "step": 57676
    },
    {
      "epoch": 4.9623160973931,
      "grad_norm": 0.90234375,
      "learning_rate": 1.4027003808431139e-07,
      "loss": 0.5042,
      "step": 57677
    },
    {
      "epoch": 4.962402133700421,
      "grad_norm": 0.96484375,
      "learning_rate": 1.3963029657326764e-07,
      "loss": 1.165,
      "step": 57678
    },
    {
      "epoch": 4.9624881700077434,
      "grad_norm": 1.0078125,
      "learning_rate": 1.3899201705663833e-07,
      "loss": 0.9555,
      "step": 57679
    },
    {
      "epoch": 4.962574206315065,
      "grad_norm": 1.1953125,
      "learning_rate": 1.383551995362553e-07,
      "loss": 0.7934,
      "step": 57680
    },
    {
      "epoch": 4.962660242622387,
      "grad_norm": 1.4296875,
      "learning_rate": 1.3771984401395043e-07,
      "loss": 1.4611,
      "step": 57681
    },
    {
      "epoch": 4.962746278929709,
      "grad_norm": 1.015625,
      "learning_rate": 1.3708595049155558e-07,
      "loss": 1.231,
      "step": 57682
    },
    {
      "epoch": 4.96283231523703,
      "grad_norm": 168.0,
      "learning_rate": 1.3645351897101365e-07,
      "loss": 1.1135,
      "step": 57683
    },
    {
      "epoch": 4.962918351544352,
      "grad_norm": 1.203125,
      "learning_rate": 1.358225494541565e-07,
      "loss": 1.0081,
      "step": 57684
    },
    {
      "epoch": 4.963004387851673,
      "grad_norm": 1.0078125,
      "learning_rate": 1.351930419427605e-07,
      "loss": 0.5167,
      "step": 57685
    },
    {
      "epoch": 4.963090424158995,
      "grad_norm": 1.1953125,
      "learning_rate": 1.345649964386575e-07,
      "loss": 1.3067,
      "step": 57686
    },
    {
      "epoch": 4.963176460466316,
      "grad_norm": 1.0234375,
      "learning_rate": 1.339384129437904e-07,
      "loss": 0.6336,
      "step": 57687
    },
    {
      "epoch": 4.9632624967736385,
      "grad_norm": 0.99609375,
      "learning_rate": 1.3331329145993555e-07,
      "loss": 0.6807,
      "step": 57688
    },
    {
      "epoch": 4.963348533080961,
      "grad_norm": 36.5,
      "learning_rate": 1.326896319889248e-07,
      "loss": 0.743,
      "step": 57689
    },
    {
      "epoch": 4.963434569388282,
      "grad_norm": 1.125,
      "learning_rate": 1.3206743453253456e-07,
      "loss": 1.0435,
      "step": 57690
    },
    {
      "epoch": 4.963520605695604,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3144669909265216e-07,
      "loss": 1.5548,
      "step": 57691
    },
    {
      "epoch": 4.963606642002925,
      "grad_norm": 1.0234375,
      "learning_rate": 1.308274256711095e-07,
      "loss": 1.2377,
      "step": 57692
    },
    {
      "epoch": 4.963692678310247,
      "grad_norm": 1.2578125,
      "learning_rate": 1.302096142696274e-07,
      "loss": 1.1461,
      "step": 57693
    },
    {
      "epoch": 4.963778714617568,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2959326489009326e-07,
      "loss": 1.841,
      "step": 57694
    },
    {
      "epoch": 4.96386475092489,
      "grad_norm": 10.3125,
      "learning_rate": 1.289783775342279e-07,
      "loss": 0.7353,
      "step": 57695
    },
    {
      "epoch": 4.963950787232212,
      "grad_norm": 2.21875,
      "learning_rate": 1.2836495220391874e-07,
      "loss": 1.1588,
      "step": 57696
    },
    {
      "epoch": 4.964036823539534,
      "grad_norm": 0.984375,
      "learning_rate": 1.2775298890094212e-07,
      "loss": 1.355,
      "step": 57697
    },
    {
      "epoch": 4.964122859846856,
      "grad_norm": 1.3125,
      "learning_rate": 1.2714248762707435e-07,
      "loss": 0.8511,
      "step": 57698
    },
    {
      "epoch": 4.964208896154177,
      "grad_norm": 1.234375,
      "learning_rate": 1.2653344838403634e-07,
      "loss": 1.2952,
      "step": 57699
    },
    {
      "epoch": 4.964294932461499,
      "grad_norm": 1.0859375,
      "learning_rate": 1.2592587117371547e-07,
      "loss": 0.9827,
      "step": 57700
    },
    {
      "epoch": 4.96438096876882,
      "grad_norm": 1.640625,
      "learning_rate": 1.2531975599783252e-07,
      "loss": 1.4324,
      "step": 57701
    },
    {
      "epoch": 4.964467005076142,
      "grad_norm": 1.0390625,
      "learning_rate": 1.247151028581639e-07,
      "loss": 1.3516,
      "step": 57702
    },
    {
      "epoch": 4.964553041383464,
      "grad_norm": 6.96875,
      "learning_rate": 1.2411191175648596e-07,
      "loss": 1.4012,
      "step": 57703
    },
    {
      "epoch": 4.964639077690785,
      "grad_norm": 1.234375,
      "learning_rate": 1.2351018269451953e-07,
      "loss": 1.5525,
      "step": 57704
    },
    {
      "epoch": 4.9647251139981075,
      "grad_norm": 1.328125,
      "learning_rate": 1.229099156740965e-07,
      "loss": 1.5871,
      "step": 57705
    },
    {
      "epoch": 4.964811150305429,
      "grad_norm": 0.98828125,
      "learning_rate": 1.2231111069693767e-07,
      "loss": 1.6053,
      "step": 57706
    },
    {
      "epoch": 4.964897186612751,
      "grad_norm": 1.5703125,
      "learning_rate": 1.2171376776481946e-07,
      "loss": 1.4749,
      "step": 57707
    },
    {
      "epoch": 4.964983222920072,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2111788687946269e-07,
      "loss": 0.5501,
      "step": 57708
    },
    {
      "epoch": 4.965069259227394,
      "grad_norm": 1.1328125,
      "learning_rate": 1.2052346804258818e-07,
      "loss": 1.0423,
      "step": 57709
    },
    {
      "epoch": 4.965155295534716,
      "grad_norm": 0.91015625,
      "learning_rate": 1.1993051125602782e-07,
      "loss": 1.0726,
      "step": 57710
    },
    {
      "epoch": 4.965241331842037,
      "grad_norm": 3.6875,
      "learning_rate": 1.1933901652139146e-07,
      "loss": 0.7443,
      "step": 57711
    },
    {
      "epoch": 4.965327368149359,
      "grad_norm": 1.125,
      "learning_rate": 1.1874898384045541e-07,
      "loss": 1.2802,
      "step": 57712
    },
    {
      "epoch": 4.9654134044566804,
      "grad_norm": 1.3671875,
      "learning_rate": 1.1816041321499604e-07,
      "loss": 1.0821,
      "step": 57713
    },
    {
      "epoch": 4.9654994407640025,
      "grad_norm": 1.1328125,
      "learning_rate": 1.175733046466787e-07,
      "loss": 1.7564,
      "step": 57714
    },
    {
      "epoch": 4.965585477071324,
      "grad_norm": 1.6953125,
      "learning_rate": 1.1698765813722423e-07,
      "loss": 1.143,
      "step": 57715
    },
    {
      "epoch": 4.965671513378646,
      "grad_norm": 127.0,
      "learning_rate": 1.1640347368835347e-07,
      "loss": 0.7724,
      "step": 57716
    },
    {
      "epoch": 4.965757549685968,
      "grad_norm": 1.078125,
      "learning_rate": 1.1582075130184277e-07,
      "loss": 1.3569,
      "step": 57717
    },
    {
      "epoch": 4.965843585993289,
      "grad_norm": 1.5859375,
      "learning_rate": 1.1523949097924646e-07,
      "loss": 1.4088,
      "step": 57718
    },
    {
      "epoch": 4.965929622300611,
      "grad_norm": 1.234375,
      "learning_rate": 1.1465969272239641e-07,
      "loss": 1.2234,
      "step": 57719
    },
    {
      "epoch": 4.966015658607932,
      "grad_norm": 1.2734375,
      "learning_rate": 1.1408135653295793e-07,
      "loss": 0.8675,
      "step": 57720
    },
    {
      "epoch": 4.966101694915254,
      "grad_norm": 1.25,
      "learning_rate": 1.1350448241259636e-07,
      "loss": 0.7041,
      "step": 57721
    },
    {
      "epoch": 4.9661877312225755,
      "grad_norm": 1.1640625,
      "learning_rate": 1.1292907036303257e-07,
      "loss": 1.3148,
      "step": 57722
    },
    {
      "epoch": 4.966273767529898,
      "grad_norm": 0.953125,
      "learning_rate": 1.1235512038587637e-07,
      "loss": 1.0934,
      "step": 57723
    },
    {
      "epoch": 4.96635980383722,
      "grad_norm": 1.4609375,
      "learning_rate": 1.117826324829041e-07,
      "loss": 1.0536,
      "step": 57724
    },
    {
      "epoch": 4.966445840144541,
      "grad_norm": 1.0859375,
      "learning_rate": 1.1121160665572561e-07,
      "loss": 0.8336,
      "step": 57725
    },
    {
      "epoch": 4.966531876451863,
      "grad_norm": 0.921875,
      "learning_rate": 1.1064204290606172e-07,
      "loss": 0.6858,
      "step": 57726
    },
    {
      "epoch": 4.966617912759184,
      "grad_norm": 1.1015625,
      "learning_rate": 1.1007394123552228e-07,
      "loss": 1.0332,
      "step": 57727
    },
    {
      "epoch": 4.966703949066506,
      "grad_norm": 0.90625,
      "learning_rate": 1.0950730164577261e-07,
      "loss": 1.0841,
      "step": 57728
    },
    {
      "epoch": 4.966789985373827,
      "grad_norm": 1.2578125,
      "learning_rate": 1.0894212413853355e-07,
      "loss": 1.4825,
      "step": 57729
    },
    {
      "epoch": 4.966876021681149,
      "grad_norm": 0.91796875,
      "learning_rate": 1.0837840871535942e-07,
      "loss": 0.724,
      "step": 57730
    },
    {
      "epoch": 4.9669620579884715,
      "grad_norm": 1.4375,
      "learning_rate": 1.0781615537797107e-07,
      "loss": 1.2513,
      "step": 57731
    },
    {
      "epoch": 4.967048094295793,
      "grad_norm": 1.40625,
      "learning_rate": 1.0725536412803382e-07,
      "loss": 1.1304,
      "step": 57732
    },
    {
      "epoch": 4.967134130603115,
      "grad_norm": 36.75,
      "learning_rate": 1.0669603496710201e-07,
      "loss": 1.2132,
      "step": 57733
    },
    {
      "epoch": 4.967220166910436,
      "grad_norm": 1.078125,
      "learning_rate": 1.0613816789689646e-07,
      "loss": 1.8488,
      "step": 57734
    },
    {
      "epoch": 4.967306203217758,
      "grad_norm": 1.0078125,
      "learning_rate": 1.0558176291902698e-07,
      "loss": 0.7358,
      "step": 57735
    },
    {
      "epoch": 4.967392239525079,
      "grad_norm": 38.75,
      "learning_rate": 1.0502682003504793e-07,
      "loss": 1.4271,
      "step": 57736
    },
    {
      "epoch": 4.967478275832401,
      "grad_norm": 0.99609375,
      "learning_rate": 1.0447333924668012e-07,
      "loss": 0.6921,
      "step": 57737
    },
    {
      "epoch": 4.967564312139723,
      "grad_norm": 1.0703125,
      "learning_rate": 1.0392132055553338e-07,
      "loss": 0.8153,
      "step": 57738
    },
    {
      "epoch": 4.9676503484470445,
      "grad_norm": 1.1015625,
      "learning_rate": 1.0337076396316203e-07,
      "loss": 1.2399,
      "step": 57739
    },
    {
      "epoch": 4.9677363847543665,
      "grad_norm": 0.93359375,
      "learning_rate": 1.0282166947117588e-07,
      "loss": 0.697,
      "step": 57740
    },
    {
      "epoch": 4.967822421061688,
      "grad_norm": 0.89453125,
      "learning_rate": 1.0227403708124028e-07,
      "loss": 0.7435,
      "step": 57741
    },
    {
      "epoch": 4.96790845736901,
      "grad_norm": 1.4140625,
      "learning_rate": 1.0172786679496504e-07,
      "loss": 0.771,
      "step": 57742
    },
    {
      "epoch": 4.967994493676331,
      "grad_norm": 1.1640625,
      "learning_rate": 1.0118315861384896e-07,
      "loss": 1.2153,
      "step": 57743
    },
    {
      "epoch": 4.968080529983653,
      "grad_norm": 1.046875,
      "learning_rate": 1.006399125395574e-07,
      "loss": 1.2899,
      "step": 57744
    },
    {
      "epoch": 4.968166566290975,
      "grad_norm": 21.875,
      "learning_rate": 1.0009812857370016e-07,
      "loss": 1.7697,
      "step": 57745
    },
    {
      "epoch": 4.968252602598296,
      "grad_norm": 0.8515625,
      "learning_rate": 9.955780671783154e-08,
      "loss": 0.5052,
      "step": 57746
    },
    {
      "epoch": 4.968338638905618,
      "grad_norm": 0.9140625,
      "learning_rate": 9.901894697350588e-08,
      "loss": 1.0121,
      "step": 57747
    },
    {
      "epoch": 4.9684246752129395,
      "grad_norm": 1.6796875,
      "learning_rate": 9.8481549342333e-08,
      "loss": 0.8672,
      "step": 57748
    },
    {
      "epoch": 4.968510711520262,
      "grad_norm": 1.078125,
      "learning_rate": 9.79456138258672e-08,
      "loss": 0.7472,
      "step": 57749
    },
    {
      "epoch": 4.968596747827583,
      "grad_norm": 4.28125,
      "learning_rate": 9.74111404257183e-08,
      "loss": 1.0627,
      "step": 57750
    },
    {
      "epoch": 4.968682784134905,
      "grad_norm": 1.359375,
      "learning_rate": 9.687812914338512e-08,
      "loss": 1.1716,
      "step": 57751
    },
    {
      "epoch": 4.968768820442227,
      "grad_norm": 1.4140625,
      "learning_rate": 9.634657998047747e-08,
      "loss": 1.9287,
      "step": 57752
    },
    {
      "epoch": 4.968854856749548,
      "grad_norm": 1.0859375,
      "learning_rate": 9.581649293849415e-08,
      "loss": 0.7155,
      "step": 57753
    },
    {
      "epoch": 4.96894089305687,
      "grad_norm": 1.09375,
      "learning_rate": 9.528786801904499e-08,
      "loss": 1.2058,
      "step": 57754
    },
    {
      "epoch": 4.969026929364191,
      "grad_norm": 1.3046875,
      "learning_rate": 9.476070522362878e-08,
      "loss": 0.9475,
      "step": 57755
    },
    {
      "epoch": 4.969112965671513,
      "grad_norm": 1.3125,
      "learning_rate": 9.423500455385536e-08,
      "loss": 0.9639,
      "step": 57756
    },
    {
      "epoch": 4.969199001978835,
      "grad_norm": 1.328125,
      "learning_rate": 9.371076601116802e-08,
      "loss": 0.5662,
      "step": 57757
    },
    {
      "epoch": 4.969285038286157,
      "grad_norm": 1.2421875,
      "learning_rate": 9.318798959717656e-08,
      "loss": 1.883,
      "step": 57758
    },
    {
      "epoch": 4.969371074593479,
      "grad_norm": 1.3125,
      "learning_rate": 9.266667531343531e-08,
      "loss": 1.3487,
      "step": 57759
    },
    {
      "epoch": 4.9694571109008,
      "grad_norm": 4.125,
      "learning_rate": 9.214682316133205e-08,
      "loss": 1.1555,
      "step": 57760
    },
    {
      "epoch": 4.969543147208122,
      "grad_norm": 1.1640625,
      "learning_rate": 9.162843314253211e-08,
      "loss": 1.3,
      "step": 57761
    },
    {
      "epoch": 4.969629183515443,
      "grad_norm": 1.296875,
      "learning_rate": 9.111150525847878e-08,
      "loss": 0.8552,
      "step": 57762
    },
    {
      "epoch": 4.969715219822765,
      "grad_norm": 1.0703125,
      "learning_rate": 9.059603951067085e-08,
      "loss": 0.7272,
      "step": 57763
    },
    {
      "epoch": 4.969801256130087,
      "grad_norm": 0.83984375,
      "learning_rate": 9.008203590066266e-08,
      "loss": 1.176,
      "step": 57764
    },
    {
      "epoch": 4.9698872924374085,
      "grad_norm": 1.09375,
      "learning_rate": 8.956949442995299e-08,
      "loss": 1.3223,
      "step": 57765
    },
    {
      "epoch": 4.9699733287447305,
      "grad_norm": 2.515625,
      "learning_rate": 8.905841509998513e-08,
      "loss": 0.8029,
      "step": 57766
    },
    {
      "epoch": 4.970059365052052,
      "grad_norm": 1.0703125,
      "learning_rate": 8.85487979123134e-08,
      "loss": 0.7345,
      "step": 57767
    },
    {
      "epoch": 4.970145401359374,
      "grad_norm": 1.0625,
      "learning_rate": 8.804064286843661e-08,
      "loss": 0.7107,
      "step": 57768
    },
    {
      "epoch": 4.970231437666696,
      "grad_norm": 1.1328125,
      "learning_rate": 8.753394996974251e-08,
      "loss": 0.6994,
      "step": 57769
    },
    {
      "epoch": 4.970317473974017,
      "grad_norm": 1.328125,
      "learning_rate": 8.702871921784094e-08,
      "loss": 1.1257,
      "step": 57770
    },
    {
      "epoch": 4.970403510281339,
      "grad_norm": 1.0390625,
      "learning_rate": 8.65249506141752e-08,
      "loss": 1.8761,
      "step": 57771
    },
    {
      "epoch": 4.97048954658866,
      "grad_norm": 0.91015625,
      "learning_rate": 8.602264416013306e-08,
      "loss": 0.6598,
      "step": 57772
    },
    {
      "epoch": 4.970575582895982,
      "grad_norm": 66.5,
      "learning_rate": 8.552179985726883e-08,
      "loss": 1.3797,
      "step": 57773
    },
    {
      "epoch": 4.9706616192033035,
      "grad_norm": 1.34375,
      "learning_rate": 8.50224177070258e-08,
      "loss": 1.0185,
      "step": 57774
    },
    {
      "epoch": 4.970747655510626,
      "grad_norm": 1.0390625,
      "learning_rate": 8.452449771090276e-08,
      "loss": 0.9924,
      "step": 57775
    },
    {
      "epoch": 4.970833691817948,
      "grad_norm": 0.87890625,
      "learning_rate": 8.402803987028751e-08,
      "loss": 0.5672,
      "step": 57776
    },
    {
      "epoch": 4.970919728125269,
      "grad_norm": 1.0234375,
      "learning_rate": 8.353304418662333e-08,
      "loss": 1.2886,
      "step": 57777
    },
    {
      "epoch": 4.971005764432591,
      "grad_norm": 1.0546875,
      "learning_rate": 8.303951066146454e-08,
      "loss": 1.3771,
      "step": 57778
    },
    {
      "epoch": 4.971091800739912,
      "grad_norm": 1.2890625,
      "learning_rate": 8.254743929614339e-08,
      "loss": 1.3446,
      "step": 57779
    },
    {
      "epoch": 4.971177837047234,
      "grad_norm": 1.2421875,
      "learning_rate": 8.205683009215869e-08,
      "loss": 1.8075,
      "step": 57780
    },
    {
      "epoch": 4.971263873354555,
      "grad_norm": 1.375,
      "learning_rate": 8.156768305095375e-08,
      "loss": 1.0883,
      "step": 57781
    },
    {
      "epoch": 4.971349909661877,
      "grad_norm": 1.1328125,
      "learning_rate": 8.10799981738608e-08,
      "loss": 0.948,
      "step": 57782
    },
    {
      "epoch": 4.9714359459691995,
      "grad_norm": 0.91796875,
      "learning_rate": 8.059377546243419e-08,
      "loss": 0.8558,
      "step": 57783
    },
    {
      "epoch": 4.971521982276521,
      "grad_norm": 2.8125,
      "learning_rate": 8.010901491806167e-08,
      "loss": 0.8259,
      "step": 57784
    },
    {
      "epoch": 4.971608018583843,
      "grad_norm": 1.9140625,
      "learning_rate": 7.96257165420755e-08,
      "loss": 0.9343,
      "step": 57785
    },
    {
      "epoch": 4.971694054891164,
      "grad_norm": 1.4609375,
      "learning_rate": 7.914388033603004e-08,
      "loss": 2.0069,
      "step": 57786
    },
    {
      "epoch": 4.971780091198486,
      "grad_norm": 0.7734375,
      "learning_rate": 7.866350630120201e-08,
      "loss": 0.7428,
      "step": 57787
    },
    {
      "epoch": 4.971866127505807,
      "grad_norm": 1.6015625,
      "learning_rate": 7.818459443909021e-08,
      "loss": 1.5397,
      "step": 57788
    },
    {
      "epoch": 4.971952163813129,
      "grad_norm": 1.0390625,
      "learning_rate": 7.770714475102691e-08,
      "loss": 0.7463,
      "step": 57789
    },
    {
      "epoch": 4.972038200120451,
      "grad_norm": 1.2890625,
      "learning_rate": 7.723115723845542e-08,
      "loss": 1.3144,
      "step": 57790
    },
    {
      "epoch": 4.9721242364277725,
      "grad_norm": 324.0,
      "learning_rate": 7.67566319027635e-08,
      "loss": 1.5166,
      "step": 57791
    },
    {
      "epoch": 4.9722102727350945,
      "grad_norm": 1.140625,
      "learning_rate": 7.628356874528341e-08,
      "loss": 0.7903,
      "step": 57792
    },
    {
      "epoch": 4.972296309042416,
      "grad_norm": 0.8828125,
      "learning_rate": 7.581196776751397e-08,
      "loss": 0.7393,
      "step": 57793
    },
    {
      "epoch": 4.972382345349738,
      "grad_norm": 8.25,
      "learning_rate": 7.534182897073194e-08,
      "loss": 1.1525,
      "step": 57794
    },
    {
      "epoch": 4.972468381657059,
      "grad_norm": 111.5,
      "learning_rate": 7.487315235632508e-08,
      "loss": 0.9486,
      "step": 57795
    },
    {
      "epoch": 4.972554417964381,
      "grad_norm": 0.83984375,
      "learning_rate": 7.440593792568118e-08,
      "loss": 0.849,
      "step": 57796
    },
    {
      "epoch": 4.972640454271703,
      "grad_norm": 0.93359375,
      "learning_rate": 7.394018568024353e-08,
      "loss": 0.5947,
      "step": 57797
    },
    {
      "epoch": 4.972726490579024,
      "grad_norm": 0.890625,
      "learning_rate": 7.347589562123336e-08,
      "loss": 0.9434,
      "step": 57798
    },
    {
      "epoch": 4.972812526886346,
      "grad_norm": 0.9921875,
      "learning_rate": 7.301306775009398e-08,
      "loss": 1.2938,
      "step": 57799
    },
    {
      "epoch": 4.9728985631936675,
      "grad_norm": 0.8515625,
      "learning_rate": 7.255170206815765e-08,
      "loss": 0.4802,
      "step": 57800
    },
    {
      "epoch": 4.97298459950099,
      "grad_norm": 1.4609375,
      "learning_rate": 7.209179857675664e-08,
      "loss": 0.7602,
      "step": 57801
    },
    {
      "epoch": 4.973070635808311,
      "grad_norm": 0.90625,
      "learning_rate": 7.163335727727871e-08,
      "loss": 1.1034,
      "step": 57802
    },
    {
      "epoch": 4.973156672115633,
      "grad_norm": 1.4375,
      "learning_rate": 7.117637817105616e-08,
      "loss": 1.569,
      "step": 57803
    },
    {
      "epoch": 4.973242708422955,
      "grad_norm": 0.93359375,
      "learning_rate": 7.072086125936572e-08,
      "loss": 1.6627,
      "step": 57804
    },
    {
      "epoch": 4.973328744730276,
      "grad_norm": 0.75390625,
      "learning_rate": 7.02668065436507e-08,
      "loss": 0.517,
      "step": 57805
    },
    {
      "epoch": 4.973414781037598,
      "grad_norm": 16.25,
      "learning_rate": 6.981421402513233e-08,
      "loss": 1.3352,
      "step": 57806
    },
    {
      "epoch": 4.973500817344919,
      "grad_norm": 1.6015625,
      "learning_rate": 6.93630837051984e-08,
      "loss": 1.2261,
      "step": 57807
    },
    {
      "epoch": 4.973586853652241,
      "grad_norm": 1.2265625,
      "learning_rate": 6.891341558512565e-08,
      "loss": 1.4728,
      "step": 57808
    },
    {
      "epoch": 4.973672889959563,
      "grad_norm": 1.3671875,
      "learning_rate": 6.846520966624637e-08,
      "loss": 1.2587,
      "step": 57809
    },
    {
      "epoch": 4.973758926266885,
      "grad_norm": 0.80078125,
      "learning_rate": 6.80184659498928e-08,
      "loss": 0.4907,
      "step": 57810
    },
    {
      "epoch": 4.973844962574207,
      "grad_norm": 1.2578125,
      "learning_rate": 6.757318443734173e-08,
      "loss": 0.7505,
      "step": 57811
    },
    {
      "epoch": 4.973930998881528,
      "grad_norm": 1.4296875,
      "learning_rate": 6.71293651299254e-08,
      "loss": 1.0241,
      "step": 57812
    },
    {
      "epoch": 4.97401703518885,
      "grad_norm": 1.1015625,
      "learning_rate": 6.668700802886507e-08,
      "loss": 1.0601,
      "step": 57813
    },
    {
      "epoch": 4.974103071496171,
      "grad_norm": 0.97265625,
      "learning_rate": 6.624611313560403e-08,
      "loss": 0.9718,
      "step": 57814
    },
    {
      "epoch": 4.974189107803493,
      "grad_norm": 2.3125,
      "learning_rate": 6.58066804512525e-08,
      "loss": 0.9598,
      "step": 57815
    },
    {
      "epoch": 4.974275144110814,
      "grad_norm": 1.109375,
      "learning_rate": 6.536870997725375e-08,
      "loss": 1.1113,
      "step": 57816
    },
    {
      "epoch": 4.9743611804181365,
      "grad_norm": 1.171875,
      "learning_rate": 6.493220171477355e-08,
      "loss": 1.0718,
      "step": 57817
    },
    {
      "epoch": 4.9744472167254585,
      "grad_norm": 1.2890625,
      "learning_rate": 6.449715566519965e-08,
      "loss": 0.971,
      "step": 57818
    },
    {
      "epoch": 4.97453325303278,
      "grad_norm": 1.2265625,
      "learning_rate": 6.40635718296978e-08,
      "loss": 1.5656,
      "step": 57819
    },
    {
      "epoch": 4.974619289340102,
      "grad_norm": 3.234375,
      "learning_rate": 6.363145020954475e-08,
      "loss": 0.6939,
      "step": 57820
    },
    {
      "epoch": 4.974705325647423,
      "grad_norm": 8.9375,
      "learning_rate": 6.320079080607277e-08,
      "loss": 1.4212,
      "step": 57821
    },
    {
      "epoch": 4.974791361954745,
      "grad_norm": 1.1640625,
      "learning_rate": 6.277159362050311e-08,
      "loss": 0.847,
      "step": 57822
    },
    {
      "epoch": 4.974877398262066,
      "grad_norm": 1.25,
      "learning_rate": 6.234385865411252e-08,
      "loss": 1.409,
      "step": 57823
    },
    {
      "epoch": 4.974963434569388,
      "grad_norm": 1.46875,
      "learning_rate": 6.191758590812224e-08,
      "loss": 1.202,
      "step": 57824
    },
    {
      "epoch": 4.97504947087671,
      "grad_norm": 1.1953125,
      "learning_rate": 6.149277538375353e-08,
      "loss": 1.2951,
      "step": 57825
    },
    {
      "epoch": 4.9751355071840315,
      "grad_norm": 0.91015625,
      "learning_rate": 6.106942708233865e-08,
      "loss": 0.6546,
      "step": 57826
    },
    {
      "epoch": 4.975221543491354,
      "grad_norm": 1.1484375,
      "learning_rate": 6.064754100504333e-08,
      "loss": 1.0144,
      "step": 57827
    },
    {
      "epoch": 4.975307579798675,
      "grad_norm": 1.109375,
      "learning_rate": 6.022711715308882e-08,
      "loss": 1.4065,
      "step": 57828
    },
    {
      "epoch": 4.975393616105997,
      "grad_norm": 1.0234375,
      "learning_rate": 5.980815552775187e-08,
      "loss": 1.1569,
      "step": 57829
    },
    {
      "epoch": 4.975479652413318,
      "grad_norm": 0.7109375,
      "learning_rate": 5.939065613025374e-08,
      "loss": 0.4748,
      "step": 57830
    },
    {
      "epoch": 4.97556568872064,
      "grad_norm": 0.98046875,
      "learning_rate": 5.897461896181566e-08,
      "loss": 0.8195,
      "step": 57831
    },
    {
      "epoch": 4.975651725027962,
      "grad_norm": 1.2578125,
      "learning_rate": 5.8560044023658885e-08,
      "loss": 1.5935,
      "step": 57832
    },
    {
      "epoch": 4.975737761335283,
      "grad_norm": 1.5390625,
      "learning_rate": 5.8146931316949147e-08,
      "loss": 0.7257,
      "step": 57833
    },
    {
      "epoch": 4.975823797642605,
      "grad_norm": 1.421875,
      "learning_rate": 5.773528084290769e-08,
      "loss": 1.4949,
      "step": 57834
    },
    {
      "epoch": 4.975909833949927,
      "grad_norm": 0.8984375,
      "learning_rate": 5.732509260281127e-08,
      "loss": 0.8653,
      "step": 57835
    },
    {
      "epoch": 4.975995870257249,
      "grad_norm": 1.078125,
      "learning_rate": 5.6916366597770106e-08,
      "loss": 0.7187,
      "step": 57836
    },
    {
      "epoch": 4.97608190656457,
      "grad_norm": 0.93359375,
      "learning_rate": 5.650910282900545e-08,
      "loss": 0.7295,
      "step": 57837
    },
    {
      "epoch": 4.976167942871892,
      "grad_norm": 2.65625,
      "learning_rate": 5.610330129773855e-08,
      "loss": 1.274,
      "step": 57838
    },
    {
      "epoch": 4.976253979179214,
      "grad_norm": 1.359375,
      "learning_rate": 5.569896200513513e-08,
      "loss": 1.4117,
      "step": 57839
    },
    {
      "epoch": 4.976340015486535,
      "grad_norm": 1.765625,
      "learning_rate": 5.529608495241645e-08,
      "loss": 1.0349,
      "step": 57840
    },
    {
      "epoch": 4.976426051793857,
      "grad_norm": 10.0625,
      "learning_rate": 5.4894670140637205e-08,
      "loss": 1.286,
      "step": 57841
    },
    {
      "epoch": 4.976512088101178,
      "grad_norm": 1.1796875,
      "learning_rate": 5.449471757112967e-08,
      "loss": 1.0942,
      "step": 57842
    },
    {
      "epoch": 4.9765981244085005,
      "grad_norm": 1.0234375,
      "learning_rate": 5.409622724494856e-08,
      "loss": 1.3721,
      "step": 57843
    },
    {
      "epoch": 4.976684160715822,
      "grad_norm": 1.3515625,
      "learning_rate": 5.3699199163315116e-08,
      "loss": 2.0454,
      "step": 57844
    },
    {
      "epoch": 4.976770197023144,
      "grad_norm": 0.96484375,
      "learning_rate": 5.330363332739507e-08,
      "loss": 0.7956,
      "step": 57845
    },
    {
      "epoch": 4.976856233330466,
      "grad_norm": 1.2265625,
      "learning_rate": 5.2909529738298654e-08,
      "loss": 0.9638,
      "step": 57846
    },
    {
      "epoch": 4.976942269637787,
      "grad_norm": 1.6875,
      "learning_rate": 5.251688839724711e-08,
      "loss": 1.4692,
      "step": 57847
    },
    {
      "epoch": 4.977028305945109,
      "grad_norm": 0.8515625,
      "learning_rate": 5.212570930529514e-08,
      "loss": 0.6743,
      "step": 57848
    },
    {
      "epoch": 4.97711434225243,
      "grad_norm": 1.09375,
      "learning_rate": 5.173599246366401e-08,
      "loss": 0.6618,
      "step": 57849
    },
    {
      "epoch": 4.977200378559752,
      "grad_norm": 1.03125,
      "learning_rate": 5.134773787346392e-08,
      "loss": 1.4344,
      "step": 57850
    },
    {
      "epoch": 4.9772864148670735,
      "grad_norm": 0.8984375,
      "learning_rate": 5.096094553586061e-08,
      "loss": 1.0222,
      "step": 57851
    },
    {
      "epoch": 4.9773724511743955,
      "grad_norm": 1.515625,
      "learning_rate": 5.0575615451964315e-08,
      "loss": 1.0094,
      "step": 57852
    },
    {
      "epoch": 4.977458487481718,
      "grad_norm": 1.125,
      "learning_rate": 5.019174762288525e-08,
      "loss": 1.383,
      "step": 57853
    },
    {
      "epoch": 4.977544523789039,
      "grad_norm": 0.9140625,
      "learning_rate": 4.980934204973364e-08,
      "loss": 0.9342,
      "step": 57854
    },
    {
      "epoch": 4.977630560096361,
      "grad_norm": 13.125,
      "learning_rate": 4.942839873373073e-08,
      "loss": 0.8538,
      "step": 57855
    },
    {
      "epoch": 4.977716596403682,
      "grad_norm": 1.0234375,
      "learning_rate": 4.904891767582021e-08,
      "loss": 0.7052,
      "step": 57856
    },
    {
      "epoch": 4.977802632711004,
      "grad_norm": 1.0390625,
      "learning_rate": 4.8670898877278826e-08,
      "loss": 0.4825,
      "step": 57857
    },
    {
      "epoch": 4.977888669018325,
      "grad_norm": 1.3984375,
      "learning_rate": 4.829434233910579e-08,
      "loss": 0.9641,
      "step": 57858
    },
    {
      "epoch": 4.977974705325647,
      "grad_norm": 106.0,
      "learning_rate": 4.7919248062466834e-08,
      "loss": 1.2291,
      "step": 57859
    },
    {
      "epoch": 4.978060741632969,
      "grad_norm": 1.203125,
      "learning_rate": 4.7545616048361165e-08,
      "loss": 1.0581,
      "step": 57860
    },
    {
      "epoch": 4.978146777940291,
      "grad_norm": 0.796875,
      "learning_rate": 4.717344629801001e-08,
      "loss": 1.241,
      "step": 57861
    },
    {
      "epoch": 4.978232814247613,
      "grad_norm": 1.0234375,
      "learning_rate": 4.6802738812412595e-08,
      "loss": 0.9017,
      "step": 57862
    },
    {
      "epoch": 4.978318850554934,
      "grad_norm": 0.93359375,
      "learning_rate": 4.643349359273463e-08,
      "loss": 1.1252,
      "step": 57863
    },
    {
      "epoch": 4.978404886862256,
      "grad_norm": 1.1953125,
      "learning_rate": 4.606571063991982e-08,
      "loss": 1.3621,
      "step": 57864
    },
    {
      "epoch": 4.978490923169577,
      "grad_norm": 1.03125,
      "learning_rate": 4.5699389955189406e-08,
      "loss": 0.8588,
      "step": 57865
    },
    {
      "epoch": 4.978576959476899,
      "grad_norm": 1.2578125,
      "learning_rate": 4.533453153954259e-08,
      "loss": 0.9325,
      "step": 57866
    },
    {
      "epoch": 4.978662995784221,
      "grad_norm": 1.234375,
      "learning_rate": 4.497113539403408e-08,
      "loss": 1.5986,
      "step": 57867
    },
    {
      "epoch": 4.978749032091542,
      "grad_norm": 0.99609375,
      "learning_rate": 4.46092015197741e-08,
      "loss": 0.724,
      "step": 57868
    },
    {
      "epoch": 4.9788350683988645,
      "grad_norm": 1.328125,
      "learning_rate": 4.424872991776186e-08,
      "loss": 1.2968,
      "step": 57869
    },
    {
      "epoch": 4.978921104706186,
      "grad_norm": 1.4765625,
      "learning_rate": 4.3889720589107565e-08,
      "loss": 1.9107,
      "step": 57870
    },
    {
      "epoch": 4.979007141013508,
      "grad_norm": 1.5390625,
      "learning_rate": 4.353217353486594e-08,
      "loss": 1.237,
      "step": 57871
    },
    {
      "epoch": 4.979093177320829,
      "grad_norm": 2.25,
      "learning_rate": 4.317608875603618e-08,
      "loss": 1.017,
      "step": 57872
    },
    {
      "epoch": 4.979179213628151,
      "grad_norm": 1.109375,
      "learning_rate": 4.2821466253673e-08,
      "loss": 0.9751,
      "step": 57873
    },
    {
      "epoch": 4.979265249935473,
      "grad_norm": 1.515625,
      "learning_rate": 4.2468306028831114e-08,
      "loss": 0.9398,
      "step": 57874
    },
    {
      "epoch": 4.979351286242794,
      "grad_norm": 1.0859375,
      "learning_rate": 4.2116608082509724e-08,
      "loss": 0.7574,
      "step": 57875
    },
    {
      "epoch": 4.979437322550116,
      "grad_norm": 1.0234375,
      "learning_rate": 4.176637241576353e-08,
      "loss": 2.0494,
      "step": 57876
    },
    {
      "epoch": 4.9795233588574375,
      "grad_norm": 1.09375,
      "learning_rate": 4.141759902964726e-08,
      "loss": 1.4228,
      "step": 57877
    },
    {
      "epoch": 4.9796093951647595,
      "grad_norm": 0.7734375,
      "learning_rate": 4.107028792510459e-08,
      "loss": 1.3749,
      "step": 57878
    },
    {
      "epoch": 4.979695431472082,
      "grad_norm": 1.0859375,
      "learning_rate": 4.0724439103245745e-08,
      "loss": 0.904,
      "step": 57879
    },
    {
      "epoch": 4.979781467779403,
      "grad_norm": 14.125,
      "learning_rate": 4.038005256495891e-08,
      "loss": 1.2039,
      "step": 57880
    },
    {
      "epoch": 4.979867504086725,
      "grad_norm": 1.15625,
      "learning_rate": 4.003712831140982e-08,
      "loss": 0.9538,
      "step": 57881
    },
    {
      "epoch": 4.979953540394046,
      "grad_norm": 1.3203125,
      "learning_rate": 3.9695666343431136e-08,
      "loss": 1.0063,
      "step": 57882
    },
    {
      "epoch": 4.980039576701368,
      "grad_norm": 0.92578125,
      "learning_rate": 3.935566666218859e-08,
      "loss": 1.487,
      "step": 57883
    },
    {
      "epoch": 4.98012561300869,
      "grad_norm": 1.1328125,
      "learning_rate": 3.901712926851486e-08,
      "loss": 0.6572,
      "step": 57884
    },
    {
      "epoch": 4.980211649316011,
      "grad_norm": 6.96875,
      "learning_rate": 3.8680054163520164e-08,
      "loss": 0.6179,
      "step": 57885
    },
    {
      "epoch": 4.980297685623333,
      "grad_norm": 0.9765625,
      "learning_rate": 3.8344441348148184e-08,
      "loss": 0.7216,
      "step": 57886
    },
    {
      "epoch": 4.980383721930655,
      "grad_norm": 1.375,
      "learning_rate": 3.801029082339813e-08,
      "loss": 1.7433,
      "step": 57887
    },
    {
      "epoch": 4.980469758237977,
      "grad_norm": 9.375,
      "learning_rate": 3.767760259021369e-08,
      "loss": 0.5142,
      "step": 57888
    },
    {
      "epoch": 4.980555794545298,
      "grad_norm": 1.2109375,
      "learning_rate": 3.7346376649594064e-08,
      "loss": 1.6829,
      "step": 57889
    },
    {
      "epoch": 4.98064183085262,
      "grad_norm": 0.8203125,
      "learning_rate": 3.701661300248294e-08,
      "loss": 0.6791,
      "step": 57890
    },
    {
      "epoch": 4.980727867159942,
      "grad_norm": 1.125,
      "learning_rate": 3.6688311649824004e-08,
      "loss": 1.1884,
      "step": 57891
    },
    {
      "epoch": 4.980813903467263,
      "grad_norm": 1.1796875,
      "learning_rate": 3.636147259267197e-08,
      "loss": 0.5174,
      "step": 57892
    },
    {
      "epoch": 4.980899939774585,
      "grad_norm": 4.09375,
      "learning_rate": 3.603609583191503e-08,
      "loss": 0.8651,
      "step": 57893
    },
    {
      "epoch": 4.980985976081906,
      "grad_norm": 0.984375,
      "learning_rate": 3.571218136849686e-08,
      "loss": 0.6949,
      "step": 57894
    },
    {
      "epoch": 4.9810720123892285,
      "grad_norm": 1.4453125,
      "learning_rate": 3.5389729203416654e-08,
      "loss": 0.6743,
      "step": 57895
    },
    {
      "epoch": 4.98115804869655,
      "grad_norm": 0.953125,
      "learning_rate": 3.506873933756261e-08,
      "loss": 0.4615,
      "step": 57896
    },
    {
      "epoch": 4.981244085003872,
      "grad_norm": 1.6953125,
      "learning_rate": 3.47492117718784e-08,
      "loss": 1.3288,
      "step": 57897
    },
    {
      "epoch": 4.981330121311194,
      "grad_norm": 0.98828125,
      "learning_rate": 3.443114650736323e-08,
      "loss": 0.9589,
      "step": 57898
    },
    {
      "epoch": 4.981416157618515,
      "grad_norm": 1.203125,
      "learning_rate": 3.4114543544849776e-08,
      "loss": 1.1238,
      "step": 57899
    },
    {
      "epoch": 4.981502193925837,
      "grad_norm": 1.3203125,
      "learning_rate": 3.3799402885337226e-08,
      "loss": 0.791,
      "step": 57900
    },
    {
      "epoch": 4.981588230233158,
      "grad_norm": 2.328125,
      "learning_rate": 3.348572452971377e-08,
      "loss": 1.0908,
      "step": 57901
    },
    {
      "epoch": 4.98167426654048,
      "grad_norm": 1.046875,
      "learning_rate": 3.3173508478867574e-08,
      "loss": 0.8813,
      "step": 57902
    },
    {
      "epoch": 4.9817603028478015,
      "grad_norm": 1.4609375,
      "learning_rate": 3.286275473379785e-08,
      "loss": 1.7471,
      "step": 57903
    },
    {
      "epoch": 4.9818463391551235,
      "grad_norm": 191.0,
      "learning_rate": 3.2553463295392774e-08,
      "loss": 0.6339,
      "step": 57904
    },
    {
      "epoch": 4.981932375462446,
      "grad_norm": 1.2578125,
      "learning_rate": 3.2245634164485e-08,
      "loss": 1.8706,
      "step": 57905
    },
    {
      "epoch": 4.982018411769767,
      "grad_norm": 0.82421875,
      "learning_rate": 3.193926734201824e-08,
      "loss": 1.0552,
      "step": 57906
    },
    {
      "epoch": 4.982104448077089,
      "grad_norm": 1.1875,
      "learning_rate": 3.163436282893617e-08,
      "loss": 0.7067,
      "step": 57907
    },
    {
      "epoch": 4.98219048438441,
      "grad_norm": 2.84375,
      "learning_rate": 3.1330920626015944e-08,
      "loss": 1.2444,
      "step": 57908
    },
    {
      "epoch": 4.982276520691732,
      "grad_norm": 0.8671875,
      "learning_rate": 3.102894073425677e-08,
      "loss": 1.1377,
      "step": 57909
    },
    {
      "epoch": 4.982362556999053,
      "grad_norm": 1.40625,
      "learning_rate": 3.072842315454683e-08,
      "loss": 1.5306,
      "step": 57910
    },
    {
      "epoch": 4.982448593306375,
      "grad_norm": 1.046875,
      "learning_rate": 3.042936788766326e-08,
      "loss": 1.0367,
      "step": 57911
    },
    {
      "epoch": 4.982534629613697,
      "grad_norm": 3408.0,
      "learning_rate": 3.013177493449426e-08,
      "loss": 0.4652,
      "step": 57912
    },
    {
      "epoch": 4.982620665921019,
      "grad_norm": 4.96875,
      "learning_rate": 2.983564429603902e-08,
      "loss": 0.6559,
      "step": 57913
    },
    {
      "epoch": 4.982706702228341,
      "grad_norm": 1.3359375,
      "learning_rate": 2.954097597301919e-08,
      "loss": 1.5025,
      "step": 57914
    },
    {
      "epoch": 4.982792738535662,
      "grad_norm": 1.2265625,
      "learning_rate": 2.9247769966378458e-08,
      "loss": 0.8765,
      "step": 57915
    },
    {
      "epoch": 4.982878774842984,
      "grad_norm": 1.5078125,
      "learning_rate": 2.8956026276893975e-08,
      "loss": 1.0944,
      "step": 57916
    },
    {
      "epoch": 4.982964811150305,
      "grad_norm": 1.09375,
      "learning_rate": 2.866574490550944e-08,
      "loss": 0.9099,
      "step": 57917
    },
    {
      "epoch": 4.983050847457627,
      "grad_norm": 1.4609375,
      "learning_rate": 2.837692585305751e-08,
      "loss": 1.2075,
      "step": 57918
    },
    {
      "epoch": 4.983136883764949,
      "grad_norm": 5.84375,
      "learning_rate": 2.808956912037086e-08,
      "loss": 0.808,
      "step": 57919
    },
    {
      "epoch": 4.98322292007227,
      "grad_norm": 1.328125,
      "learning_rate": 2.7803674708282155e-08,
      "loss": 2.1111,
      "step": 57920
    },
    {
      "epoch": 4.9833089563795925,
      "grad_norm": 1.046875,
      "learning_rate": 2.7519242617568553e-08,
      "loss": 1.234,
      "step": 57921
    },
    {
      "epoch": 4.983394992686914,
      "grad_norm": 2.46875,
      "learning_rate": 2.7236272849173738e-08,
      "loss": 1.4582,
      "step": 57922
    },
    {
      "epoch": 4.983481028994236,
      "grad_norm": 1.2890625,
      "learning_rate": 2.695476540387487e-08,
      "loss": 1.5896,
      "step": 57923
    },
    {
      "epoch": 4.983567065301557,
      "grad_norm": 1.015625,
      "learning_rate": 2.6674720282504618e-08,
      "loss": 1.4416,
      "step": 57924
    },
    {
      "epoch": 4.983653101608879,
      "grad_norm": 1.5078125,
      "learning_rate": 2.6396137485840133e-08,
      "loss": 1.1296,
      "step": 57925
    },
    {
      "epoch": 4.983739137916201,
      "grad_norm": 1.6328125,
      "learning_rate": 2.6119017014714087e-08,
      "loss": 1.1712,
      "step": 57926
    },
    {
      "epoch": 4.983825174223522,
      "grad_norm": 1.109375,
      "learning_rate": 2.5843358869959145e-08,
      "loss": 1.9216,
      "step": 57927
    },
    {
      "epoch": 4.983911210530844,
      "grad_norm": 0.9765625,
      "learning_rate": 2.5569163052407974e-08,
      "loss": 1.5457,
      "step": 57928
    },
    {
      "epoch": 4.9839972468381655,
      "grad_norm": 1.203125,
      "learning_rate": 2.529642956283773e-08,
      "loss": 0.8066,
      "step": 57929
    },
    {
      "epoch": 4.9840832831454875,
      "grad_norm": 1.0625,
      "learning_rate": 2.502515840197006e-08,
      "loss": 1.0627,
      "step": 57930
    },
    {
      "epoch": 4.984169319452809,
      "grad_norm": 0.90625,
      "learning_rate": 2.475534957069314e-08,
      "loss": 0.713,
      "step": 57931
    },
    {
      "epoch": 4.984255355760131,
      "grad_norm": 1.0546875,
      "learning_rate": 2.448700306978413e-08,
      "loss": 1.4204,
      "step": 57932
    },
    {
      "epoch": 4.984341392067453,
      "grad_norm": 1.125,
      "learning_rate": 2.4220118900020184e-08,
      "loss": 0.8282,
      "step": 57933
    },
    {
      "epoch": 4.984427428374774,
      "grad_norm": 1.1796875,
      "learning_rate": 2.3954697062122943e-08,
      "loss": 0.4688,
      "step": 57934
    },
    {
      "epoch": 4.984513464682096,
      "grad_norm": 0.93359375,
      "learning_rate": 2.3690737556980592e-08,
      "loss": 0.6795,
      "step": 57935
    },
    {
      "epoch": 4.984599500989417,
      "grad_norm": 1.3515625,
      "learning_rate": 2.3428240385259258e-08,
      "loss": 1.4749,
      "step": 57936
    },
    {
      "epoch": 4.984685537296739,
      "grad_norm": 0.8828125,
      "learning_rate": 2.3167205547791613e-08,
      "loss": 0.5937,
      "step": 57937
    },
    {
      "epoch": 4.9847715736040605,
      "grad_norm": 1.5546875,
      "learning_rate": 2.2907633045299303e-08,
      "loss": 1.4438,
      "step": 57938
    },
    {
      "epoch": 4.984857609911383,
      "grad_norm": 1.28125,
      "learning_rate": 2.264952287855948e-08,
      "loss": 1.38,
      "step": 57939
    },
    {
      "epoch": 4.984943646218705,
      "grad_norm": 1.0390625,
      "learning_rate": 2.2392875048349303e-08,
      "loss": 0.4229,
      "step": 57940
    },
    {
      "epoch": 4.985029682526026,
      "grad_norm": 1.421875,
      "learning_rate": 2.2137689555390417e-08,
      "loss": 1.1902,
      "step": 57941
    },
    {
      "epoch": 4.985115718833348,
      "grad_norm": 1.46875,
      "learning_rate": 2.1883966400404463e-08,
      "loss": 0.9801,
      "step": 57942
    },
    {
      "epoch": 4.985201755140669,
      "grad_norm": 1.2109375,
      "learning_rate": 2.1631705584224114e-08,
      "loss": 1.6864,
      "step": 57943
    },
    {
      "epoch": 4.985287791447991,
      "grad_norm": 1.171875,
      "learning_rate": 2.13809071075155e-08,
      "loss": 1.0635,
      "step": 57944
    },
    {
      "epoch": 4.985373827755312,
      "grad_norm": 0.859375,
      "learning_rate": 2.1131570971000268e-08,
      "loss": 0.6394,
      "step": 57945
    },
    {
      "epoch": 4.985459864062634,
      "grad_norm": 1.09375,
      "learning_rate": 2.0883697175455572e-08,
      "loss": 1.0493,
      "step": 57946
    },
    {
      "epoch": 4.9855459003699565,
      "grad_norm": 1.03125,
      "learning_rate": 2.063728572160306e-08,
      "loss": 1.2271,
      "step": 57947
    },
    {
      "epoch": 4.985631936677278,
      "grad_norm": 5.3125,
      "learning_rate": 2.0392336610108865e-08,
      "loss": 0.9959,
      "step": 57948
    },
    {
      "epoch": 4.9857179729846,
      "grad_norm": 1.234375,
      "learning_rate": 2.0148849841750137e-08,
      "loss": 1.1122,
      "step": 57949
    },
    {
      "epoch": 4.985804009291921,
      "grad_norm": 1.3203125,
      "learning_rate": 1.990682541724853e-08,
      "loss": 0.7384,
      "step": 57950
    },
    {
      "epoch": 4.985890045599243,
      "grad_norm": 1.2578125,
      "learning_rate": 1.9666263337214662e-08,
      "loss": 0.5174,
      "step": 57951
    },
    {
      "epoch": 4.985976081906564,
      "grad_norm": 1.265625,
      "learning_rate": 1.942716360242569e-08,
      "loss": 1.157,
      "step": 57952
    },
    {
      "epoch": 4.986062118213886,
      "grad_norm": 1.75,
      "learning_rate": 1.918952621360326e-08,
      "loss": 1.4133,
      "step": 57953
    },
    {
      "epoch": 4.986148154521208,
      "grad_norm": 1.390625,
      "learning_rate": 1.8953351171413502e-08,
      "loss": 1.5862,
      "step": 57954
    },
    {
      "epoch": 4.9862341908285295,
      "grad_norm": 1.15625,
      "learning_rate": 1.8718638476522552e-08,
      "loss": 0.8014,
      "step": 57955
    },
    {
      "epoch": 4.9863202271358515,
      "grad_norm": 1.09375,
      "learning_rate": 1.848538812965206e-08,
      "loss": 1.0759,
      "step": 57956
    },
    {
      "epoch": 4.986406263443173,
      "grad_norm": 1.6953125,
      "learning_rate": 1.825360013141264e-08,
      "loss": 0.8294,
      "step": 57957
    },
    {
      "epoch": 4.986492299750495,
      "grad_norm": 0.9609375,
      "learning_rate": 1.8023274482636965e-08,
      "loss": 0.9042,
      "step": 57958
    },
    {
      "epoch": 4.986578336057816,
      "grad_norm": 1.015625,
      "learning_rate": 1.779441118382463e-08,
      "loss": 0.8388,
      "step": 57959
    },
    {
      "epoch": 4.986664372365138,
      "grad_norm": 1.21875,
      "learning_rate": 1.7567010235752802e-08,
      "loss": 1.1713,
      "step": 57960
    },
    {
      "epoch": 4.98675040867246,
      "grad_norm": 1.0703125,
      "learning_rate": 1.7341071639087602e-08,
      "loss": 1.0578,
      "step": 57961
    },
    {
      "epoch": 4.986836444979781,
      "grad_norm": 1.21875,
      "learning_rate": 1.711659539438415e-08,
      "loss": 0.4879,
      "step": 57962
    },
    {
      "epoch": 4.986922481287103,
      "grad_norm": 1.1875,
      "learning_rate": 1.6893581502419595e-08,
      "loss": 1.5202,
      "step": 57963
    },
    {
      "epoch": 4.9870085175944245,
      "grad_norm": 1.1796875,
      "learning_rate": 1.667202996380457e-08,
      "loss": 1.2886,
      "step": 57964
    },
    {
      "epoch": 4.987094553901747,
      "grad_norm": 1.1875,
      "learning_rate": 1.6451940779149687e-08,
      "loss": 0.8323,
      "step": 57965
    },
    {
      "epoch": 4.987180590209068,
      "grad_norm": 1.2890625,
      "learning_rate": 1.62333139491766e-08,
      "loss": 1.1764,
      "step": 57966
    },
    {
      "epoch": 4.98726662651639,
      "grad_norm": 1.21875,
      "learning_rate": 1.6016149474440413e-08,
      "loss": 0.7287,
      "step": 57967
    },
    {
      "epoch": 4.987352662823712,
      "grad_norm": 1.03125,
      "learning_rate": 1.580044735566277e-08,
      "loss": 1.3702,
      "step": 57968
    },
    {
      "epoch": 4.987438699131033,
      "grad_norm": 0.671875,
      "learning_rate": 1.5586207593398792e-08,
      "loss": 0.4986,
      "step": 57969
    },
    {
      "epoch": 4.987524735438355,
      "grad_norm": 1.3203125,
      "learning_rate": 1.5373430188314608e-08,
      "loss": 1.0253,
      "step": 57970
    },
    {
      "epoch": 4.987610771745676,
      "grad_norm": 1.1640625,
      "learning_rate": 1.5162115140965328e-08,
      "loss": 0.9583,
      "step": 57971
    },
    {
      "epoch": 4.987696808052998,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4952262452072596e-08,
      "loss": 0.8208,
      "step": 57972
    },
    {
      "epoch": 4.98778284436032,
      "grad_norm": 1.0859375,
      "learning_rate": 1.4743872122247038e-08,
      "loss": 0.9484,
      "step": 57973
    },
    {
      "epoch": 4.987868880667642,
      "grad_norm": 1.2109375,
      "learning_rate": 1.4536944151988252e-08,
      "loss": 0.9469,
      "step": 57974
    },
    {
      "epoch": 4.987954916974964,
      "grad_norm": 0.83984375,
      "learning_rate": 1.4331478541962373e-08,
      "loss": 0.7531,
      "step": 57975
    },
    {
      "epoch": 4.988040953282285,
      "grad_norm": 1.078125,
      "learning_rate": 1.4127475292835535e-08,
      "loss": 1.1337,
      "step": 57976
    },
    {
      "epoch": 4.988126989589607,
      "grad_norm": 0.99609375,
      "learning_rate": 1.3924934405107337e-08,
      "loss": 1.3714,
      "step": 57977
    },
    {
      "epoch": 4.988213025896928,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3723855879388403e-08,
      "loss": 2.2203,
      "step": 57978
    },
    {
      "epoch": 4.98829906220425,
      "grad_norm": 1.4453125,
      "learning_rate": 1.3524239716289355e-08,
      "loss": 1.0246,
      "step": 57979
    },
    {
      "epoch": 4.988385098511571,
      "grad_norm": 0.91015625,
      "learning_rate": 1.3326085916420816e-08,
      "loss": 1.1915,
      "step": 57980
    },
    {
      "epoch": 4.9884711348188935,
      "grad_norm": 0.90625,
      "learning_rate": 1.3129394480282386e-08,
      "loss": 0.7873,
      "step": 57981
    },
    {
      "epoch": 4.9885571711262156,
      "grad_norm": 1.453125,
      "learning_rate": 1.29341654085402e-08,
      "loss": 1.6097,
      "step": 57982
    },
    {
      "epoch": 4.988643207433537,
      "grad_norm": 1.4140625,
      "learning_rate": 1.2740398701749368e-08,
      "loss": 1.069,
      "step": 57983
    },
    {
      "epoch": 4.988729243740859,
      "grad_norm": 1.9921875,
      "learning_rate": 1.254809436040949e-08,
      "loss": 1.6066,
      "step": 57984
    },
    {
      "epoch": 4.98881528004818,
      "grad_norm": 1.2890625,
      "learning_rate": 1.235725238513119e-08,
      "loss": 1.2835,
      "step": 57985
    },
    {
      "epoch": 4.988901316355502,
      "grad_norm": 1.0703125,
      "learning_rate": 1.216787277646958e-08,
      "loss": 0.7951,
      "step": 57986
    },
    {
      "epoch": 4.988987352662823,
      "grad_norm": 16.875,
      "learning_rate": 1.1979955534924258e-08,
      "loss": 1.3886,
      "step": 57987
    },
    {
      "epoch": 4.989073388970145,
      "grad_norm": 1.265625,
      "learning_rate": 1.179350066116136e-08,
      "loss": 1.513,
      "step": 57988
    },
    {
      "epoch": 4.989159425277467,
      "grad_norm": 1.078125,
      "learning_rate": 1.1608508155624975e-08,
      "loss": 1.0682,
      "step": 57989
    },
    {
      "epoch": 4.9892454615847885,
      "grad_norm": 0.87890625,
      "learning_rate": 1.1424978018925725e-08,
      "loss": 1.0815,
      "step": 57990
    },
    {
      "epoch": 4.989331497892111,
      "grad_norm": 1.0546875,
      "learning_rate": 1.124291025156321e-08,
      "loss": 0.9759,
      "step": 57991
    },
    {
      "epoch": 4.989417534199432,
      "grad_norm": 1.6875,
      "learning_rate": 1.106230485403703e-08,
      "loss": 1.5273,
      "step": 57992
    },
    {
      "epoch": 4.989503570506754,
      "grad_norm": 0.94140625,
      "learning_rate": 1.08831618269023e-08,
      "loss": 0.9386,
      "step": 57993
    },
    {
      "epoch": 4.989589606814076,
      "grad_norm": 1.0546875,
      "learning_rate": 1.0705481170714127e-08,
      "loss": 1.382,
      "step": 57994
    },
    {
      "epoch": 4.989675643121397,
      "grad_norm": 1.359375,
      "learning_rate": 1.0529262886027625e-08,
      "loss": 1.2605,
      "step": 57995
    },
    {
      "epoch": 4.989761679428719,
      "grad_norm": 1.4609375,
      "learning_rate": 1.035450697323137e-08,
      "loss": 1.4195,
      "step": 57996
    },
    {
      "epoch": 4.98984771573604,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0181213432935987e-08,
      "loss": 1.4643,
      "step": 57997
    },
    {
      "epoch": 4.989933752043362,
      "grad_norm": 2.9375,
      "learning_rate": 1.0009382265585564e-08,
      "loss": 0.9227,
      "step": 57998
    },
    {
      "epoch": 4.990019788350684,
      "grad_norm": 1.5625,
      "learning_rate": 9.839013471735214e-09,
      "loss": 0.7973,
      "step": 57999
    },
    {
      "epoch": 4.990105824658006,
      "grad_norm": 0.9375,
      "learning_rate": 9.670107051829024e-09,
      "loss": 0.3507,
      "step": 58000
    },
    {
      "epoch": 4.990191860965328,
      "grad_norm": 1.03125,
      "learning_rate": 9.50266300647762e-09,
      "loss": 0.588,
      "step": 58001
    },
    {
      "epoch": 4.990277897272649,
      "grad_norm": 33.0,
      "learning_rate": 9.336681336014063e-09,
      "loss": 0.8855,
      "step": 58002
    },
    {
      "epoch": 4.990363933579971,
      "grad_norm": 1.2578125,
      "learning_rate": 9.172162040993471e-09,
      "loss": 0.9948,
      "step": 58003
    },
    {
      "epoch": 4.990449969887292,
      "grad_norm": 7.625,
      "learning_rate": 9.009105121970951e-09,
      "loss": 1.2802,
      "step": 58004
    },
    {
      "epoch": 4.990536006194614,
      "grad_norm": 0.96875,
      "learning_rate": 8.847510579279571e-09,
      "loss": 0.9241,
      "step": 58005
    },
    {
      "epoch": 4.990622042501936,
      "grad_norm": 1.6015625,
      "learning_rate": 8.687378413529957e-09,
      "loss": 1.0913,
      "step": 58006
    },
    {
      "epoch": 4.9907080788092575,
      "grad_norm": 1.0625,
      "learning_rate": 8.52870862505517e-09,
      "loss": 1.2008,
      "step": 58007
    },
    {
      "epoch": 4.9907941151165796,
      "grad_norm": 1.0859375,
      "learning_rate": 8.371501214465837e-09,
      "loss": 0.5751,
      "step": 58008
    },
    {
      "epoch": 4.990880151423901,
      "grad_norm": 13.4375,
      "learning_rate": 8.215756182095024e-09,
      "loss": 1.06,
      "step": 58009
    },
    {
      "epoch": 4.990966187731223,
      "grad_norm": 0.8359375,
      "learning_rate": 8.061473528497842e-09,
      "loss": 0.5286,
      "step": 58010
    },
    {
      "epoch": 4.991052224038544,
      "grad_norm": 1.1796875,
      "learning_rate": 7.908653254007359e-09,
      "loss": 1.8966,
      "step": 58011
    },
    {
      "epoch": 4.991138260345866,
      "grad_norm": 1.359375,
      "learning_rate": 7.757295359178684e-09,
      "loss": 1.2703,
      "step": 58012
    },
    {
      "epoch": 4.991224296653188,
      "grad_norm": 0.8984375,
      "learning_rate": 7.607399844455909e-09,
      "loss": 0.7497,
      "step": 58013
    },
    {
      "epoch": 4.991310332960509,
      "grad_norm": 0.95703125,
      "learning_rate": 7.4589667101721e-09,
      "loss": 1.351,
      "step": 58014
    },
    {
      "epoch": 4.991396369267831,
      "grad_norm": 1.15625,
      "learning_rate": 7.311995956882367e-09,
      "loss": 1.351,
      "step": 58015
    },
    {
      "epoch": 4.9914824055751525,
      "grad_norm": 1.375,
      "learning_rate": 7.166487584919779e-09,
      "loss": 1.4861,
      "step": 58016
    },
    {
      "epoch": 4.991568441882475,
      "grad_norm": 7.5,
      "learning_rate": 7.022441594783935e-09,
      "loss": 0.5954,
      "step": 58017
    },
    {
      "epoch": 4.991654478189796,
      "grad_norm": 0.9140625,
      "learning_rate": 6.879857986863414e-09,
      "loss": 0.7851,
      "step": 58018
    },
    {
      "epoch": 4.991740514497118,
      "grad_norm": 0.9609375,
      "learning_rate": 6.738736761546793e-09,
      "loss": 0.6711,
      "step": 58019
    },
    {
      "epoch": 4.99182655080444,
      "grad_norm": 1.78125,
      "learning_rate": 6.599077919278162e-09,
      "loss": 1.0215,
      "step": 58020
    },
    {
      "epoch": 4.991912587111761,
      "grad_norm": 1.515625,
      "learning_rate": 6.460881460501611e-09,
      "loss": 1.2147,
      "step": 58021
    },
    {
      "epoch": 4.991998623419083,
      "grad_norm": 1.1640625,
      "learning_rate": 6.324147385550205e-09,
      "loss": 1.0401,
      "step": 58022
    },
    {
      "epoch": 4.992084659726404,
      "grad_norm": 1.4375,
      "learning_rate": 6.188875694868035e-09,
      "loss": 1.2354,
      "step": 58023
    },
    {
      "epoch": 4.992170696033726,
      "grad_norm": 1.4453125,
      "learning_rate": 6.055066388843678e-09,
      "loss": 0.8491,
      "step": 58024
    },
    {
      "epoch": 4.992256732341048,
      "grad_norm": 1.28125,
      "learning_rate": 5.922719467865712e-09,
      "loss": 0.9111,
      "step": 58025
    },
    {
      "epoch": 4.99234276864837,
      "grad_norm": 1.0234375,
      "learning_rate": 5.791834932267203e-09,
      "loss": 1.2942,
      "step": 58026
    },
    {
      "epoch": 4.992428804955692,
      "grad_norm": 1.2734375,
      "learning_rate": 5.662412782547754e-09,
      "loss": 0.7381,
      "step": 58027
    },
    {
      "epoch": 4.992514841263013,
      "grad_norm": 1.0078125,
      "learning_rate": 5.534453019040431e-09,
      "loss": 0.8991,
      "step": 58028
    },
    {
      "epoch": 4.992600877570335,
      "grad_norm": 1.1484375,
      "learning_rate": 5.407955642022788e-09,
      "loss": 1.4687,
      "step": 58029
    },
    {
      "epoch": 4.992686913877656,
      "grad_norm": 0.8515625,
      "learning_rate": 5.282920651994427e-09,
      "loss": 1.1795,
      "step": 58030
    },
    {
      "epoch": 4.992772950184978,
      "grad_norm": 0.765625,
      "learning_rate": 5.159348049288415e-09,
      "loss": 0.6702,
      "step": 58031
    },
    {
      "epoch": 4.992858986492299,
      "grad_norm": 1.1875,
      "learning_rate": 5.037237834182307e-09,
      "loss": 0.8607,
      "step": 58032
    },
    {
      "epoch": 4.9929450227996215,
      "grad_norm": 1.2109375,
      "learning_rate": 4.916590007175703e-09,
      "loss": 1.2184,
      "step": 58033
    },
    {
      "epoch": 4.993031059106944,
      "grad_norm": 1.171875,
      "learning_rate": 4.797404568490649e-09,
      "loss": 1.7788,
      "step": 58034
    },
    {
      "epoch": 4.993117095414265,
      "grad_norm": 0.9453125,
      "learning_rate": 4.679681518571233e-09,
      "loss": 0.4902,
      "step": 58035
    },
    {
      "epoch": 4.993203131721587,
      "grad_norm": 1.1171875,
      "learning_rate": 4.563420857695011e-09,
      "loss": 1.3842,
      "step": 58036
    },
    {
      "epoch": 4.993289168028908,
      "grad_norm": 0.99609375,
      "learning_rate": 4.44862258619505e-09,
      "loss": 0.9189,
      "step": 58037
    },
    {
      "epoch": 4.99337520433623,
      "grad_norm": 1.2109375,
      "learning_rate": 4.335286704459929e-09,
      "loss": 0.9557,
      "step": 58038
    },
    {
      "epoch": 4.993461240643551,
      "grad_norm": 1.7578125,
      "learning_rate": 4.223413212822713e-09,
      "loss": 1.4158,
      "step": 58039
    },
    {
      "epoch": 4.993547276950873,
      "grad_norm": 1.2890625,
      "learning_rate": 4.113002111560959e-09,
      "loss": 1.4751,
      "step": 58040
    },
    {
      "epoch": 4.993633313258195,
      "grad_norm": 60.5,
      "learning_rate": 4.004053401063246e-09,
      "loss": 1.9023,
      "step": 58041
    },
    {
      "epoch": 4.9937193495655166,
      "grad_norm": 1.296875,
      "learning_rate": 3.896567081607127e-09,
      "loss": 0.995,
      "step": 58042
    },
    {
      "epoch": 4.993805385872839,
      "grad_norm": 0.9140625,
      "learning_rate": 3.790543153470161e-09,
      "loss": 1.1972,
      "step": 58043
    },
    {
      "epoch": 4.99389142218016,
      "grad_norm": 1.1796875,
      "learning_rate": 3.6859816170409234e-09,
      "loss": 0.6739,
      "step": 58044
    },
    {
      "epoch": 4.993977458487482,
      "grad_norm": 2.609375,
      "learning_rate": 3.58288247254146e-09,
      "loss": 0.8746,
      "step": 58045
    },
    {
      "epoch": 4.994063494794803,
      "grad_norm": 1.0078125,
      "learning_rate": 3.4812457203603485e-09,
      "loss": 0.9171,
      "step": 58046
    },
    {
      "epoch": 4.994149531102125,
      "grad_norm": 0.9375,
      "learning_rate": 3.3810713607196344e-09,
      "loss": 1.0581,
      "step": 58047
    },
    {
      "epoch": 4.994235567409447,
      "grad_norm": 0.8125,
      "learning_rate": 3.2823593939523833e-09,
      "loss": 0.5619,
      "step": 58048
    },
    {
      "epoch": 4.994321603716768,
      "grad_norm": 0.9609375,
      "learning_rate": 3.1851098203361517e-09,
      "loss": 0.6729,
      "step": 58049
    },
    {
      "epoch": 4.99440764002409,
      "grad_norm": 1.15625,
      "learning_rate": 3.0893226402040063e-09,
      "loss": 1.0914,
      "step": 58050
    },
    {
      "epoch": 4.994493676331412,
      "grad_norm": 0.953125,
      "learning_rate": 2.994997853722481e-09,
      "loss": 0.6457,
      "step": 58051
    },
    {
      "epoch": 4.994579712638734,
      "grad_norm": 1.1328125,
      "learning_rate": 2.902135461280153e-09,
      "loss": 1.8903,
      "step": 58052
    },
    {
      "epoch": 4.994665748946055,
      "grad_norm": 1.28125,
      "learning_rate": 2.810735463099068e-09,
      "loss": 0.7398,
      "step": 58053
    },
    {
      "epoch": 4.994751785253377,
      "grad_norm": 1.4296875,
      "learning_rate": 2.7207978594567805e-09,
      "loss": 1.027,
      "step": 58054
    },
    {
      "epoch": 4.994837821560699,
      "grad_norm": 114.5,
      "learning_rate": 2.632322650630847e-09,
      "loss": 0.8872,
      "step": 58055
    },
    {
      "epoch": 4.99492385786802,
      "grad_norm": 1.0,
      "learning_rate": 2.5453098368433124e-09,
      "loss": 0.7728,
      "step": 58056
    },
    {
      "epoch": 4.995009894175342,
      "grad_norm": 1.5078125,
      "learning_rate": 2.459759418371732e-09,
      "loss": 1.2933,
      "step": 58057
    },
    {
      "epoch": 4.995095930482663,
      "grad_norm": 37.75,
      "learning_rate": 2.37567139543815e-09,
      "loss": 0.9055,
      "step": 58058
    },
    {
      "epoch": 4.9951819667899855,
      "grad_norm": 6912.0,
      "learning_rate": 2.293045768320123e-09,
      "loss": 1.4501,
      "step": 58059
    },
    {
      "epoch": 4.995268003097307,
      "grad_norm": 1.0546875,
      "learning_rate": 2.2118825372396956e-09,
      "loss": 0.5658,
      "step": 58060
    },
    {
      "epoch": 4.995354039404629,
      "grad_norm": 0.96484375,
      "learning_rate": 2.132181702474423e-09,
      "loss": 1.2053,
      "step": 58061
    },
    {
      "epoch": 4.995440075711951,
      "grad_norm": 1.2890625,
      "learning_rate": 2.0539432641908383e-09,
      "loss": 1.6731,
      "step": 58062
    },
    {
      "epoch": 4.995526112019272,
      "grad_norm": 1.171875,
      "learning_rate": 1.9771672227220096e-09,
      "loss": 1.1646,
      "step": 58063
    },
    {
      "epoch": 4.995612148326594,
      "grad_norm": 1.40625,
      "learning_rate": 1.901853578178958e-09,
      "loss": 0.9539,
      "step": 58064
    },
    {
      "epoch": 4.995698184633915,
      "grad_norm": 13.75,
      "learning_rate": 1.828002330783729e-09,
      "loss": 1.6998,
      "step": 58065
    },
    {
      "epoch": 4.995784220941237,
      "grad_norm": 1.0625,
      "learning_rate": 1.755613480869389e-09,
      "loss": 1.0205,
      "step": 58066
    },
    {
      "epoch": 4.9958702572485585,
      "grad_norm": 1.3984375,
      "learning_rate": 1.6846870285469606e-09,
      "loss": 1.0208,
      "step": 58067
    },
    {
      "epoch": 4.995956293555881,
      "grad_norm": 1.2890625,
      "learning_rate": 1.6152229740939995e-09,
      "loss": 1.4155,
      "step": 58068
    },
    {
      "epoch": 4.996042329863203,
      "grad_norm": 1.5390625,
      "learning_rate": 1.5472213176215278e-09,
      "loss": 1.9267,
      "step": 58069
    },
    {
      "epoch": 4.996128366170524,
      "grad_norm": 1.734375,
      "learning_rate": 1.4806820594071013e-09,
      "loss": 1.09,
      "step": 58070
    },
    {
      "epoch": 4.996214402477846,
      "grad_norm": 1.0078125,
      "learning_rate": 1.4156051995617425e-09,
      "loss": 0.3979,
      "step": 58071
    },
    {
      "epoch": 4.996300438785167,
      "grad_norm": 1.0703125,
      "learning_rate": 1.351990738363007e-09,
      "loss": 1.1522,
      "step": 58072
    },
    {
      "epoch": 4.996386475092489,
      "grad_norm": 1.1875,
      "learning_rate": 1.2898386759774284e-09,
      "loss": 1.0929,
      "step": 58073
    },
    {
      "epoch": 4.99647251139981,
      "grad_norm": 2.0625,
      "learning_rate": 1.2291490125715399e-09,
      "loss": 1.2703,
      "step": 58074
    },
    {
      "epoch": 4.996558547707132,
      "grad_norm": 3344.0,
      "learning_rate": 1.1699217483673864e-09,
      "loss": 0.8996,
      "step": 58075
    },
    {
      "epoch": 4.996644584014454,
      "grad_norm": 1.3515625,
      "learning_rate": 1.1121568834204787e-09,
      "loss": 1.4432,
      "step": 58076
    },
    {
      "epoch": 4.996730620321776,
      "grad_norm": 1.2109375,
      "learning_rate": 1.055854418008373e-09,
      "loss": 1.6573,
      "step": 58077
    },
    {
      "epoch": 4.996816656629098,
      "grad_norm": 1.15625,
      "learning_rate": 1.0010143522976023e-09,
      "loss": 0.9357,
      "step": 58078
    },
    {
      "epoch": 4.996902692936419,
      "grad_norm": 0.95703125,
      "learning_rate": 9.476366863991893e-10,
      "loss": 0.5911,
      "step": 58079
    },
    {
      "epoch": 4.996988729243741,
      "grad_norm": 12992.0,
      "learning_rate": 8.95721420479667e-10,
      "loss": 1.3028,
      "step": 58080
    },
    {
      "epoch": 4.997074765551062,
      "grad_norm": 4.0,
      "learning_rate": 8.452685546500582e-10,
      "loss": 1.4307,
      "step": 58081
    },
    {
      "epoch": 4.997160801858384,
      "grad_norm": 1.21875,
      "learning_rate": 7.962780891324073e-10,
      "loss": 1.819,
      "step": 58082
    },
    {
      "epoch": 4.997246838165706,
      "grad_norm": 1.21875,
      "learning_rate": 7.487500240377365e-10,
      "loss": 1.3969,
      "step": 58083
    },
    {
      "epoch": 4.997332874473027,
      "grad_norm": 0.99609375,
      "learning_rate": 7.026843595325793e-10,
      "loss": 1.3732,
      "step": 58084
    },
    {
      "epoch": 4.9974189107803495,
      "grad_norm": 1.0390625,
      "learning_rate": 6.580810957279581e-10,
      "loss": 1.2834,
      "step": 58085
    },
    {
      "epoch": 4.997504947087671,
      "grad_norm": 0.96875,
      "learning_rate": 6.149402327348952e-10,
      "loss": 1.4918,
      "step": 58086
    },
    {
      "epoch": 4.997590983394993,
      "grad_norm": 0.9609375,
      "learning_rate": 5.732617707199239e-10,
      "loss": 1.0816,
      "step": 58087
    },
    {
      "epoch": 4.997677019702314,
      "grad_norm": 1.2734375,
      "learning_rate": 5.330457097385555e-10,
      "loss": 0.7704,
      "step": 58088
    },
    {
      "epoch": 4.997763056009636,
      "grad_norm": 1.0078125,
      "learning_rate": 4.942920500128345e-10,
      "loss": 1.138,
      "step": 58089
    },
    {
      "epoch": 4.997849092316958,
      "grad_norm": 1.0078125,
      "learning_rate": 4.57000791542761e-10,
      "loss": 1.1246,
      "step": 58090
    },
    {
      "epoch": 4.997935128624279,
      "grad_norm": 1.2890625,
      "learning_rate": 4.211719345503795e-10,
      "loss": 0.825,
      "step": 58091
    },
    {
      "epoch": 4.998021164931601,
      "grad_norm": 1.28125,
      "learning_rate": 3.8680547903569007e-10,
      "loss": 0.5776,
      "step": 58092
    },
    {
      "epoch": 4.9981072012389225,
      "grad_norm": 1.0234375,
      "learning_rate": 3.5390142516522615e-10,
      "loss": 1.2695,
      "step": 58093
    },
    {
      "epoch": 4.998193237546245,
      "grad_norm": 1.203125,
      "learning_rate": 3.224597729944989e-10,
      "loss": 1.2583,
      "step": 58094
    },
    {
      "epoch": 4.998279273853566,
      "grad_norm": 1.1171875,
      "learning_rate": 2.9248052263453065e-10,
      "loss": 0.9973,
      "step": 58095
    },
    {
      "epoch": 4.998365310160888,
      "grad_norm": 0.99609375,
      "learning_rate": 2.639636741963436e-10,
      "loss": 1.1919,
      "step": 58096
    },
    {
      "epoch": 4.99845134646821,
      "grad_norm": 1.53125,
      "learning_rate": 2.36909227735449e-10,
      "loss": 0.7852,
      "step": 58097
    },
    {
      "epoch": 4.998537382775531,
      "grad_norm": 1.0859375,
      "learning_rate": 2.1131718330735794e-10,
      "loss": 1.3881,
      "step": 58098
    },
    {
      "epoch": 4.998623419082853,
      "grad_norm": 1.328125,
      "learning_rate": 1.8718754107860392e-10,
      "loss": 0.7725,
      "step": 58099
    },
    {
      "epoch": 4.998709455390174,
      "grad_norm": 1.375,
      "learning_rate": 1.6452030099367577e-10,
      "loss": 1.5415,
      "step": 58100
    },
    {
      "epoch": 4.998795491697496,
      "grad_norm": 1.03125,
      "learning_rate": 1.4331546321910694e-10,
      "loss": 1.6525,
      "step": 58101
    },
    {
      "epoch": 4.9988815280048176,
      "grad_norm": 1.1015625,
      "learning_rate": 1.2357302775489742e-10,
      "loss": 1.2693,
      "step": 58102
    },
    {
      "epoch": 4.99896756431214,
      "grad_norm": 1.21875,
      "learning_rate": 1.0529299471206954e-10,
      "loss": 1.1736,
      "step": 58103
    },
    {
      "epoch": 4.999053600619462,
      "grad_norm": 0.796875,
      "learning_rate": 8.847536409062328e-11,
      "loss": 0.4626,
      "step": 58104
    },
    {
      "epoch": 4.999139636926783,
      "grad_norm": 1.4375,
      "learning_rate": 7.312013600158096e-11,
      "loss": 1.3316,
      "step": 58105
    },
    {
      "epoch": 4.999225673234105,
      "grad_norm": 0.8046875,
      "learning_rate": 5.922731044494257e-11,
      "loss": 0.5899,
      "step": 58106
    },
    {
      "epoch": 4.999311709541426,
      "grad_norm": 1.3671875,
      "learning_rate": 4.679688747621924e-11,
      "loss": 1.4569,
      "step": 58107
    },
    {
      "epoch": 4.999397745848748,
      "grad_norm": 0.80859375,
      "learning_rate": 3.5828867095411e-11,
      "loss": 0.5629,
      "step": 58108
    },
    {
      "epoch": 4.999483782156069,
      "grad_norm": 1.1484375,
      "learning_rate": 2.632324935802899e-11,
      "loss": 1.4671,
      "step": 58109
    },
    {
      "epoch": 4.999569818463391,
      "grad_norm": 1.328125,
      "learning_rate": 1.828003431958436e-11,
      "loss": 1.2309,
      "step": 58110
    },
    {
      "epoch": 4.9996558547707135,
      "grad_norm": 0.85546875,
      "learning_rate": 1.1699221980077113e-11,
      "loss": 0.8052,
      "step": 58111
    },
    {
      "epoch": 4.999741891078035,
      "grad_norm": 508.0,
      "learning_rate": 6.580812395018399e-12,
      "loss": 1.4587,
      "step": 58112
    },
    {
      "epoch": 4.999827927385357,
      "grad_norm": 1.1953125,
      "learning_rate": 2.924805508897066e-12,
      "loss": 1.4389,
      "step": 58113
    },
    {
      "epoch": 4.999913963692678,
      "grad_norm": 1.6796875,
      "learning_rate": 7.312013772242665e-13,
      "loss": 1.6145,
      "step": 58114
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.453125,
      "learning_rate": 0.0,
      "loss": 0.9093,
      "step": 58115
    },
    {
      "epoch": 5.0,
      "step": 58115,
      "total_flos": 1.7116691854509435e+18,
      "train_loss": 1.2267941008374745,
      "train_runtime": 55296.0216,
      "train_samples_per_second": 4.204,
      "train_steps_per_second": 1.051
    }
  ],
  "logging_steps": 1,
  "max_steps": 58115,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 2906,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7116691854509435e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}